469th Advisory Committee on Reactor Safeguards (ACRS) - February 3, 2000
UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
MEETING: 469TH ADVISORY COMMITTEE ON
REACTOR SAFEGUARDS (ACRS)
Two Flint Flint North, Room T2-B3
11545 Rockville Pike
Thursday, Februay 3, 2000
The committee met, pursuant to notice, 8:31 a.m.
DANA POWERS, Chairman, ACRS
THOMAS KRESS, Member, ACRS
GEORGE APOSTOLAKIS, ACRS Member
MARIO BONACA, ACRS Member
JOHN BARTON, ACRS Member
JOHN D. SIEBER, ACRS Member
ROBERT SEALE, ACRS Member
WILLIAM SHACK, ACRS Member
ROBERT UHRIG, ACRS Member
GRAHAM WALLIS, ACRS Member
. P R O C E E D I N G S
CHAIRMAN POWERS: The meeting will now come to order.
This is the first day of the 469th meeting of the Advisory
Committee on Reactor Safeguards.
During today's meeting, the committee will consider
technical aspects associated with the revised reactor oversight process
and related matters, proposed final amendment to 10 CFR 50.72 and 50.73,
proposed regulatory guide and associated NEI document 96-07 guidelines
for 10 CFR 50.59, safety evaluations, proposed revision of the
Commission's safety goal policy statement for reactors, proposed ACRS
The meeting is being conducted in accordance with the
provisions of the Federal Advisory Committee Act. Dr. John T. Larkins
is the designated Federal official for the initial portion of the
We have received no written statements from members of the
public regarding today's session.
We have received a request from a representative of NEI for
time to make oral statements regarding proposed revision of the safety
goal policy statement.
A transcript of portions of the meeting is being kept, and
it is requested that speakers use one of the microphones, identify
themselves, and speak with sufficient clarity and volume so they can be
I want to begin the meeting by calling the members'
attention to their items of interest.
The first item in this summary should be of particular
interest, a congratulatory memorandum from the Chairman.
The members are also directed to the last page of the
package on items of interest, which brings to their attention the
Regulatory Information Conference, which many members have found to be
of use in the past.
I also want to alert members to the fact that we have a
large number of issues to examine in the reconciliation of comments.
That's going to be distributed to you today, fairly early, earlier than
usual, and you should examine it and be prepared to discuss them
I also want to call members' attention to the schedule for
the March meeting. We had agreed that, on March 1st, in the morning, we
would take training in NRC's new ADAMS program and that we would start
the full meeting that afternoon.
Also, the Planning and Procedures Subcommittee is planning
to meet in the morning on the 29th of February and, in the afternoon of
the 29th of February, work on what we need to do in preparation for a
meeting with the Commission, and we're inviting other interested members
to attend that session.
MR. BARTON: That's the afternoon of the 29th?
CHAIRMAN POWERS: Afternoon of the 29th.
And finally, I'd like to welcome our new large member, Mario
Bonaca. Members have been curious on whether you're just gaining weight
or you've escaped from something.
DR. BONACA: Gained weight.
CHAIRMAN POWERS: I see.
Are there any opening comments other members would like to
make on today's session?
CHAIRMAN POWERS: Seeing none, I will turn to the first item
of business, which is technical aspects associated with the revised
reactor oversight process and related matters.
John Barton, I think you're going to lead us through this --
MR. BARTON: I'll try, Chairman.
CHAIRMAN POWERS: -- important area.
MR. BARTON: The purpose of the meeting is to review the
technical components of the reactor oversight process, including
significant determination process and performance indicators.
In a letter dated November 23, 1999, NRR Director requested
the committee to review technical components of the reactor oversight
In particular, we were asked to review the updated
significance determination process and plant performance indicators.
In an SRM dated December 17th, the Commission requested ACRS
review the technical adequacy of the performance indicators, current and
proposed, for the new reactor oversight process, which includes
assessment to the extent to which the performance indicators
collectively provide meaningful insights to those areas of plant
operation that are most important to safety.
The plant operation subcommittee met with the staff and NEI
on January 20, 2000, to discuss these issues.
The subcommittee, at that time, formulated a set of
questions which were transmitted to the staff, and the staff was
requested to respond to those issues at today's session.
At this time, I will turn the meeting over to the NRC staff,
and Frank Gillespie, you have the lead here.
MR. GILLESPIE: I think Bill Dean is going to take the lead.
MR. DEAN: Good morning, Dr. Powers and committee members.
My name is Bill Dean. I'm the Chief of the Inspection
Program Branch in NRR. Under my auspices are the development and
implementation of the new oversight process which we're here to talk to
you about this morning.
With me today, I've got Mike Johnson, who is a Section Chief
in my branch for performance assessment, and Alan Madison, who has been
the task lead for the implementation of the new pilot program and
revised oversight process.
We also have with us today a number of our staff members
that have been key in the development and implementation of this
process, and there may be some opportunity this morning to have some of
them weigh in and provide some additional information as we go through
What we intend to do this morning, in our two hours, is to
provide a brief review of the pilot program results and the readiness
for start of implementation, our feedback in that regard, to cover some
of the defining principles and assumptions.
We think this is important before we get into the actual
detailed discussions of the performance indicators and the significance
determination process, which are the two technical issues that we have
brought forward to the committee for their consideration, that it would
be important to go over some of the defining principles and assumptions
of the whole oversight process.
Alan and Gareth Parry will provide some discussion on the
performance indicators and the significance determination process.
Mike Johnson will then talk about the assessment process,
where we pull together the results of the performance indicators and the
significance determination process, and of course, we'll answer any --
hopefully, try to answer any questions that you have.
We met with the subcommittee on January the 20th. Out of
that subcommittee meeting, there was a number of questions that we have,
and we believe that we've integrated the responses to those questions
within our presentation. So, hopefully, we'll be able to address all of
those here today.
Lastly, I guess this is a -- I don't know what number, but
there's been an ongoing series of briefings for the committee on the
process, and I believe the last time that we met with the full committee
was in June of last year, which was right about the time that we were
starting the pilot program.
So, here we are now. The pilot program is over, and we're
looking at preparing for initial implementation, so it's a good time to
meet with you again.
The pilot program was a six-month program. We performed
this program between the months of June and November of last year. It's
important to note that we're still executing this process at those 9
sites at which we did the pilot program, and so, we're still gaining
information and lessons learned, albeit at a much more discrete and
subtle level than we did earlier in the pilot program.
I think the most important characterization of this new
process that we developed as a result of our pilot program is that the
performance indicators and the baseline inspection do provide a sound
framework for providing oversight of licensee performance and to assure
that reactor safety is maintained.
Now, am I confident in saying that we've had enough time
with this pilot program to prove this premise? The answer to that would
be no, and that could probably take us years to actually prove the
premise that this program will provide reasonable assurance for reactor
But have we had enough time and have we gained enough
lessons learned to demonstrate that the process has demonstrated that we
can have some confidence, a good level of confidence that this process
will provide reasonable assurance and that it's at a point that we can
expand this process beyond the pilot program? I think the answer to
that is yes.
DR. APOSTOLAKIS: I have a question on that.
DR. SHACK: Yes, sir.
DR. APOSTOLAKIS: The committee has been struggling with the
objectives, understanding the objectives of the program, and what you
just said reminded me of that. What exactly is the objective of the
oversight process? To assure safety? And we have to elaborate on that,
what it means. Or to make sure that the plant is operated as licensed?
MR. DEAN: Well, I think you have to actually go back to the
actual safety mission of the agency, and that's to assure reasonable
protection of public health and safety from the operation of nuclear
power plants. I mean that's our overall mission.
DR. APOSTOLAKIS: What does that mean? For example, it
could mean that you have some safety goals, and as long as the goals are
met, you're providing reasonable assurance.
On the other hand, when we were reviewing 50.59, we were
told that the staff wanted to maintain the licensing basis. So, all
changes were evaluated in that context.
We believe that this is a very key element here to
understanding what you're doing.
MS. MADISON: Well, I think we've described that before,
George, when we talked about the basis of the program.
The cornerstone diagram shows as the top item on there is
our basic mission of the agency, and part of the cornerstone of safety
that we developed for this oversight process is the protection of the
public health and safety due to the operation of commercial nuclear
power, and underlying that, in the strategic performance area, are the
goals you speak of. By achieving those goals, we feel we've met our
mission of protecting the public health and safety, and so, the
cornerstones, then, have objectives that are directed at achieving those
goals in the strategic performance areas.
DR. APOSTOLAKIS: Yes, but -- so, let's take the case of a
plant that has a very low core damage frequency, has highly redundant
systems. So, it's maybe -- core damage frequency, say, is 15 times
smaller than the goal.
That means, of course, that there are system
unavailabilities that are lower than the average and maybe the rate of
occurrence of some initiators is lower than the average and so on.
If this process is to assure adequate protection, then in
principle, you could allow this plant to raise the unavailability of
MS. MADISON: In principle, you're right.
DR. APOSTOLAKIS: Whereas if your objective was to make sure
that the status of that plant, the risk profile, remains the same as it
was last time you checked, then you would not allow it to increase, and
that is a major difference in the objectives of the problem.
MS. MADISON: In principle, you're right, George, but you
have a conflict.
The rules and the regulations and the las are still on the
books, and as long as they are, we also have an obligation to make sure
that they're maintained, as does the licensee who signed on the license,
but it probably would make a case for risk-informing those regulations
or risk-informing the license that the licensee has and coming in for
some changes based upon the risk characterization.
DR. APOSTOLAKIS: But what you're say, then, is that the
objective of the program is to make sure that the risk profile -- and
risk profile doesn't mean only the quantitative part -- I mean the whole
thing, the way that it's licensed -- remains the same, as we think it
is. That's what you're saying, because if they want to change it, they
have to follow, for example, Regulatory Guide 1.174 and come in with a
So, that view would be consistent with the 50.59 revision,
with all the regulations we have.
DR. KRESS: Suppose they came in with a change request, an
exemption, and it was significant enough that they did it to 1.174 but
it did change their risk status.
DR. APOSTOLAKIS: Yes.
DR. KRESS: They increased it. Would you do anything to the
performance indicators for that particular plant? The performance
indicators would stay the same as they are now.
DR. APOSTOLAKIS: No, because it would have to be plant
DR. KRESS: I know, but they're not.
DR. BONACA: But in this process, that will not change.
DR. APOSTOLAKIS: Why not? The process allows for change.
DR. BONACA: You have certain values set for unavailability,
etcetera, which are really coming from simply a threshold that you set.
DR. APOSTOLAKIS: Yes, but that's the whole point of raising
the issue, because if that is the view, then the performance indicators
would have to be plant-specific.
So, if you changed the licensing basis of the plant, you
would have to change some of the performance indicators.
DR. BONACA: It seems to me that the only thing that the
process has set up right now, we identify developing adverse trends.
That's really what it does, okay? I don't see that it can quantify
safety. I mean it will identify a trend if something degrades.
DR. KRESS: So, you would see the objective as being to
provide a consistency --
DR. BONACA: Absolutely.
DR. KRESS: -- in the performance and not really to achieve
a level of risk status that's equivalent to what was licensed.
DR. APOSTOLAKIS: It's a very key question. Maybe we are
surprising you with this.
MR. GILLESPIE: You've actually hit the right principle for
this program. This isn't a licensing program. What we're looking at is
the delta change from the condition at the facility.
We, in fact, are kind of -- although I hesitate to use the
word "risk profile," because people jump immediately to quantitative,
you know, but in the global picture, what we're looking at is departures
from that kind of established norm, and that's when we get more engaged.
Departing is not the end of the world. It just means we
have to understand why you're departing.
DR. APOSTOLAKIS: Departing from the license -- from the
profile -- the risk profile that was there when the plant was licensed.
License means, you know, including the amendments and everything.
Right? So, that's consistent, then, with the spirit of 50.59.
MR. GILLESPIE: So, we're looking at a delta. We're looking
at basically kind of -- you know, the surrogate is a delta CDF from
whatever is allowed at that facility, and whatever the allowance is for
that facility could be different from place to place, and we know it's
DR. APOSTOLAKIS: Right.
DR. BONACA: Then there is an expectation that the
indicators will be capable of identifying adverse trends. This is the
definition that we are going to use, and I completely agree with that.
That's all that the process can do, can identify adverse
trends, from one inspection to the next, something is degrading. Okay.
Then, also, I would like to say that the thresholds, then,
are such that they should be able to identify the adverse trends.
DR. APOSTOLAKIS: This was just an issue of objectives.
DR. BONACA: But you see how important it becomes.
DR. APOSTOLAKIS: So, Dr. Kress, do we agree, then, on which
the objective is of the five you've identified?
DR. KRESS: I'm still not sure. We hear that it's to
identify adverse trends, which would be plant-specific, also, but then
we hear it's to maintain the licensing basis as it was, as licensed. I
think those are two different things.
DR. APOSTOLAKIS: They are two different things, but the
objective, though, is what the staff just said.
Now, the issue of trends and so -- I would say that's
implementation and what you want to -- you know, information you want to
get and so on, but the fundamental objective is to maintain the
MR. DEAN: Well, I would say the fundamental objective of
the program is to maintain the level of safety that exists in nuclear
power plants today.
DR. APOSTOLAKIS: Not nuclear power plants, at that plant.
There's a difference. This is the key difference. If you say at
nuclear power plants, you are making it generic, and what I'm saying is
no, to maintain the level of safety at that plant.
DR. KRESS: At that plant.
DR. APOSTOLAKIS: At that plant.
MR. JOHNSON: George, we're not surprised by the question,
because we have discussed it many times before, and we've not satisfied
you, obviously, but you know, I think maybe we think about this more
simplistically than you do.
This is an oversight process, oversight meaning, you know,
going back to our early words to you on what the process was trying to
There's a lot to be worried about.
We have a licensing process to control -- in which we try to
control the licensing basis, and changes that the license tries to make,
we want to make sure that we maintain that licensing basis.
As Frank said, we have various regulatory programs, and this
process doesn't change those programs. What this process does is steps
back to say, on any given day, a licensee may or may not be in
compliance, full compliance with their technical specifications, they
may have things that happen, you know, expected things that go on at a
plant, and so, the role of the regulator and the role of this process is
to step back and look at those things and changes in those types of
things that happen at plants, to ask ourselves, is it okay, is it some
nominal deviation from what is normally expected in terms of the
performance of the plant, or do we need to go further and dig down and
check, for example, to make sure that, with respect to issue A, they're
in compliance with their licensing basis.
DR. APOSTOLAKIS: Right.
MR. JOHNSON: So, it's an oversight process.
DR. APOSTOLAKIS: Sure. But you said of that plant. These
are key words. The whole process is focused on that plant, and if you
do that, you are consistent with the body of regulations.
See, we can take an extreme case and say, okay, as long as
the core damage frequency is less than 10 to minus 4 -- let's limit
ourselves to that -- the oversight process says it's okay.
Now, we know there are many plants whose CDF is less than
that, much less than that. You wouldn't let them raise the CDF up to
the goal just because they keep being below the goal.
This is not the role of this, because then why do we have
Regulatory Guide 1.174? Why do we have all the other regulations?
So, it's really a plant-specific process to make sure that
the level of safety at that plant is maintained, and if there is any
change, you would like to know it, adverse change.
I think we agree, actually.
MR. JOHNSON: Yes, I think we agree.
DR. APOSTOLAKIS: But this is so fundamental, because it
then tells us how we should treat the thresholds, performance
indicators, although we should make a distinction between the two.
MS. MADISON: I think you have to be careful with the term
"plant-specific." It is a program that looks at specific plants and
looks at individual plants, but it is not -- does not carry
DR. APOSTOLAKIS: Sure.
MS. MADISON: There are some plant-specific indicators -- or
type indicators, not necessarily plant-specific indicators. There are
And the same with the inspection program. The inspection
program may be tailored somewhat to the plant, but is ia fairly generic
DR. APOSTOLAKIS: But that's why we're having this
discussion, because we really have to agree on clear objectives and then
discuss the implications of the objectives, because if the objective is
to maintain the level of safety at that plant, then the thresholds must
That doesn't mean you have necessarily a different number
for each plant, but you start with that premise, and then you may decide
that, for certain performance indicators, you can live with more
generic-type thresholds, but this is really key.
We've been discussing this and we're trying to understand
what's going on.
MR. DEAN: Let me cover some other objectives, though, that
I think are important to make sure that we understand, you know, why is
it that we even entered into an effort to try and revise the oversight
process, and certainly, we've gotten some clear direction from the
Commission based on feedback from a number of stakeholders, external
stakeholders, both industry and public stakeholders, that there were
some concerns and problems with our existing oversight process, and the
Commission asked us to develop a process that was more risk-informed, a
process that was more objective, more predictable as to what actions
that the NRC would take for given performance declines, and something
that was more understandable to the public and more scrutable, and so,
that has been a lot of our defining principles as to how we're trying to
revise this process.
We have a focus on risk-significant issues, and I think that
the early returns from the pilot program is that -- from a licensee's
perspective -- is that we have been able to successfully focus not only
our attention but the licensee's attention on those issues that are the
most risk-significant at that plant, and that should be the appropriate
allocation of our efforts and resources, to focus on those things that
are most risk-significant at the plant.
With respect to the oversight process and is it adequate to
support initial implementation at all plants, as I mentioned earlier, I
think that we've gotten diminishing returns from the pilot program.
Like I said, we're still executing the process at all the
pilot sites, and we are still getting some indications of issues that
need refinement, but we're talking about much more subtle and discrete
issues and not major changes that we made early in the pilot program,
where we made substantial changes to the performance indicator program,
to the significance determination process, and key elements like that.
So, we believe we're at a point where we need to increase
the volume and the scope in order to fully exercise the process and gain
additional lessons learned so that we can further define and refine the
DR. APOSTOLAKIS: Do you have an estimate of the reduction
in unnecessary burden?
MR. DEAN: Do I have an estimate? That would be something
that I think would probably be better left to industry to provide some
comment on that.
MR. GILLESPIE: I don't want the staff to get put in a box,
so I'm going to jump in here.
Reduction in regulatory burden, in the case of this program,
can be viewed in different lights. It could be viewed in fewer
inspection hours, which in general the pilot says didn't happen.
Good performers are still going to get inspected in the
future, probably as much as good performers did in the past.
One of the things industry very much wanted out of a new
system was stability and predictability, and one of the things this new
process builds is stability and predictability.
Utilities wanted to say we know where we stand without
waiting every 18 months for a SALP report. What is the value in
regulatory burden to a stable and predictable system on Wall Street to a
utility? Only they can predict that.
But they were very vehement in the beginning that that was
one of the most, if not the most important objective to where they were
So, it isn't a question of, you know, is it 10 less
inspection hours or are we doing this much less or do they get a
licensing action through faster.
The question on regulatory burden is truly one of what's the
value of a stable predictable system where everyone knows the ground
rules, and that's more of a social value, but they can turn it into
dollars and sense on their end.
DR. BONACA: The only objective portion of the process is
the performance indicators. I mean you have not established a pass/fail
system or the baseline inspection, nor have you established how baseline
inspections and performance indicators will be integrated into an
overall cornerstone assessment.
So, I'm just saying yes, you have a more objective set, but
the only objective set is the indicators.
MR. DEAN: That's not totally true. I believe that we have
objectivity that's imbued in various elements. A significance
determination process is an objective look based on the principles of
Reg. Guide 1.174 in terms of ascertaining risk characterization of our
inspection finding, is an attempt to try and make those inspection
findings more objective in nature and being able to convey to the
licensees and to the public what is it about this issue that is of risk
DR. BONACA: I'm only saying that, you know, Wall Street was
mentioned, and they're not going to look at the safety significance.
They're going to look at greens, and if you have all indicators in the
initiators are green, that's a lot of statement coming from the
performance indicators, and there isn't a process that says it's green
but it's not really green because, if you average it and integrate it
with this other information, it should be, really, a yellow or
MR. GILLESPIE: From a safety kind of perspective, one of
the nice parts about this process was we don't try to aggregate it into
a single score, and in fact, that's what a lot of our public groups
really kind of like, because it's a profile, so that you don't get --
and one of the -- maybe one of the deficiencies in what AEOD was doing
earlier on was they were trying to deal with LER's, enforcement items,
and aggregate it all, but they weren't mutually exclusive, and so, one
could outweigh the other. In fact, you could show good performance on
the aggregate, even though the agency is very worried over here.
So, we have deliberately left this as a profile, but people
can see both whites in PI's and in inspections. Inspections are like
PI's. They're divided into cornerstones, and now they're graded also as
MR. DEAN: We'll talk about that in a minute.
The last bullet on this slide in terms of implementing an
ongoing self-assessment process -- you know, are we done making changes?
No. Obviously, we've made notable improvements to address the concerns
raised by the Commission.
We have made a process that's more objective and scrutable
and understandable and risk-informed, and there's still been a lot of
what I would consider to be appropriate stakeholder skepticism, both
internal and external, with respect to the long-term efficacy of this
process, and we have to make sure that we address that skepticism, and
we believe the way to do that is to expand this program to get more
input and more experience on a broader scale, and so, that's why we
believe -- and we've changed our -- I think, if you go back six or seven
months ago, we talked about the next phase of this process,
implementation, would be full implementation, and that's really not the
right connotation, and we've changed that to say the next phase really
is initial implementation.
We've tested out the principles and the major processes
through the pilot program.
Now we're ready to move to an initial implementation phase
where we recognize that we're going to gain lessons and that we need to
come back and revisit this process after we gain about a year's worth of
experience and go through a significant assessment as to what has this
year's worth of information told us about implementing this process at
So, we think we're ready to move into something called
initial implementation but not full implementation where you would have
the concept that this process is now a rigid, etched-in-granite process,
okay? There's still some dynamics that are going to be involved here,
and we have to make sure that we continue to provide an appropriate
self-assessment of this process.
I just wanted to spend a few minutes revisiting some of the
defining principles and assumptions, and one of them gets to this
discussion that we've already had, George, and that is that thresholds
-- you know, the whole concept of thresholds, okay?
This program establishes thresholds both in performance
indicator space and inspection space that, below which, only minimal NRC
interaction is warranted, in effect that when you have plants that have
green performance indicators and green inspection findings, that the
appropriate level of NRC regulatory interaction is the execution of our
baseline inspection program, okay?
So, what does that mean?
Does green mean good? Green does not mean good, and it
shouldn't be equated to good.
What green means is that performance, as determined by the
indicators, performance indicators, inspection findings, is acceptable
to the extent that our regulatory oversight of a baseline inspection
program is the appropriate regulatory oversight.
MR. BARTON: Bill, is that defined someplace? Will I find
those words, green means just what you said? Somewhere in this process
MR. DEAN: Yes. If you go all the way back to the technical
framework of this process back in 99-007 --
MR. BARTON: All right.
MR. DEAN: We can help define where that is.
MR. JOHNSON: That will be in the program implementation
For example, it will be in the SDP manual chapter that you
haven't seen -- or you may have seen. I guess that version is out. It
will be in the new performance indicator manual chapter. We're very
clear about what those terms mean.
MR. BARTON: Okay. Thank you, Mike.
MR. DEAN: This is a clear paradigm shift. That is an area
that our inspectors still feel some discomfort with, that there is,
within the process, what we call a licensee response band where issues
that emerge within this band of performance are issues that are best
turned over to the licensee, they're of very low risk significance or
below, that these are issues that should be entered in a licensee's
corrective action program and dealt with in construct with all the other
issues that licensees themselves identify and put in their corrective
action program, and that the NRC should not be driving resolution of
these issues just because they're issued identified by the NRC.
MR. BARTON: A key part of the new process is reliance on
the old violations being put into the licensee's corrective action
process and that process being an effective means to get to the root
cause and fix them. Where in this new oversight process are we doing an
assessment of the licensee's corrective action programs?
MR. DEAN: We'll get to that. That's a good question, and
we'll build to that.
DR. APOSTOLAKIS: Now, regarding the thresholds, first of
all, I think we have to distinguish between establishing the performance
indicators, the establishment of performance indicators and the
establishment of the thresholds.
Perhaps the indicators can be generic, but with the
thresholds, again I have a problem, because as I recall, you looked at
data over the five -- past five years for a particular indicator and
then you plotted them and you took the 95th percentile, the highest
value of the -- that performance indicator over plan, so you took the
95th percentile as a threshold.
Now, coming back to the objective, if the objective of the
process is to make sure that the safety level at plant X is maintained,
then if that plant X happened to be very good with respect to this
indicator -- say it was down to the 10th percentile of that curve -- by
establishing a threshold at the 95th percentile, aren't you, in effect,
allowing that plant to raise that indicator all the way and then it will
still be green, and then how is that consistent with the notion that I'm
trying to oversee -- that I'm trying to convince myself that the safety
level of that plant has not changed?
See, this is where my problem -- the conceptual problem is.
MS. MADISON: But are you saying, George, that if a plant is
performing in the top 10 percentile, that we should never let them slip
below that, that for some reason our regulations should be written such
that they can't be anything less than in the top 10 percentile once
they've established themselves there?
Because by establishing a threshold -- site-specific
threshold based upon their top 10 percent performance during that period
of time, that's what you're saying, that we would take action if they
slipped below --
DR. APOSTOLAKIS: Yes. If your objective is that the safety
level is maintained, you shouldn't allow them to slip.
MS. MADISON: But in a generic sense is our objective, and
that's why the four outcome measures were meant in a generic sense, that
an industry-wide, industry performance should be maintained in a safe
manner, the maintenance of safety industry-wide, and I don't think we
have the regulations to say that a licensee must perform in the top 10
percent or an excellent manner.
Our regulations all lead to licensees performing in an
adequate, in a safe enough manner.
MR. BARTON: George, i think there is a difference between
the old process and new process as a licensee would perceive it.
In the old process, there was incentives to improve
performance and raise standards. Whether anybody wants to admit to that
or not, I think the SALP process had that ingrained in it.
I think the new process takes away those incentives to
increase performance, to be an excellent performer.
Jack, do you agree?
DR. APOSTOLAKIS: Maybe you're saying the same thing with
I'm not picking one side, not yet. All I'm saying is your
thresholds should be consistent -- the establishment of the thresholds
should be consistent with your objectives.
So, if we agree that the objective is to make sure that the
level of safety at that plant is maintained, then the thresholds have to
be plant-specific. There's no way around it.
DR. KRESS: There is one way around it, George.
DR. APOSTOLAKIS: If, on the other hand, Alan is right and
you want to look at the population of plants and make sure that things
don't change, then again -- then the question would be different. Why
do you rely only on the 95th percentile?
DR. KRESS: Let me throw out a suggestion, George.
Let's presume that what we're talking about is the
derivative of a PI. We want to know whether it's increasing and whether
that increase is such that we begin to be concerned about it.
Now, let's take your really good plant, at the 10th
Now, let's say it goes through a derivative; it's degrading
in performance for, say, one or more of the indicators.
Now, how can we look and see whether that derivative is of
concern to us?
Well, it depends on the performance indicator.
If that derivative is such that it extends in time so it
crosses some threshold, then you have a measure that this derivative --
a threshold away from its base case -- you have a measure of this
derivative, because you know it crossed the threshold.
That means it increased a certain amount over a given amount
So, the question is now would you have the same derivative
measure if you put that threshold higher and higher and higher and
In fact, you could put it all the way up to the 95
percentile, and it depends on whether the degraded performance has an
effect on this derivative sufficiently to drive it all the way up to the
Now, that's the issue, to me.
If a degrading performance that is of concern to me drives
that derivative so that the value gets above the 95, then I've got the
derivative for all plants, and I can use a plant-wide set of thresholds
and not be plant-specific.
If that derivative is not sufficient to hit my concern level
before it gets up to that 95, then I have a problem. Then I need
Do you understand the difference?
DR. APOSTOLAKIS: I still don't know why the 95th percentile
should play such a major role.
DR. KRESS: I could have picked any. That's arbitrary. I
could have picked any threshold, is my point.
DR. APOSTOLAKIS: But this is industry-wide.
DR. KRESS: Yes.
DR. APOSTOLAKIS: And my objective was stated as one of
maintaining the level of safety at that plant.
DR. KRESS: Suppose we were interested in the derivative and
that a degraded performance, whatever caused this performance indicator
to go, actually puts it way beyond the 95, you know, triples it.
DR. APOSTOLAKIS: Sure, then bells will ring.
DR. KRESS: Well, that's what I'm saying. It depends on the
magnitude of the derivative and how far it will go, and I'm not sure we
They have an implied assumption that, if it trips this
threshold that we have set, that that is -- that you will find the
derivative for that particular plant. Even though it started real low
or even if it started high, you'll still get the derivative.
Now, I don't know if that's true or not, because I don't
know enough about the relationship between our concern level and the
thresholds and the derivative, but it's possible that you could have a
set of thresholds for all plants and not have them plant-specific,
although you begin to get a little concerned about that.
DR. APOSTOLAKIS: I'm still not convinced.
MR. DEAN: I'd like to share on insight with you, George,
that may or may not help give you a little bit of a sense of confidence,
but you know, the fact that -- an outgrowth of the fact that we are
publishing on our web-site these performance indicators on a quarterly
basis and it's there for God and country to see, you know, whether a
plant is the green band or the white or the yellow has provided a
tremendous incentive for licensees to assure that their performance is
such that they do not have indicators trip thresholds, okay?
They do not want to be seen as an outlier, and so, what a
number of licensees have done within the pilot process is, within that
green band, have established their own thresholds for performance, as
they train within the green band.
Now, we're not training within the green band, okay? We
have an objective threshold, green/white, that we judge to be an
appropriate threshold for which we change our level of engagement in
regulatory oversight, but licensees are tracking and trending within
those bands and are responding when they start to see thresholds creep
up, to maintain themselves, and not to go up and ride along that 95th
percentile performance level.
DR. APOSTOLAKIS: I guess what I'm saying is that maybe we
ought to be doing something like that, not the licensees, leave it up to
the licensees, I mean just as a matter of consistency.
DR. BONACA: Well, the licensees have been doing this for a
long time, because I mean many of these indicators are the INPO
indicators that were -- and they didn't go through, you know, a very
elaborate derivation of it, but they were very similar.
First of all, I support the perspective that Dr. Kress is
pointing out. I mean I do believe the point he's making is correct.
The concern I have is that thresholds may be high enough
that it will be a long way before you get there, and so, therefore, you
will not be able to see much, particularly because, already, for 10, 15
years, the licensees have been looking at the INPO, and therefore, they
are striving to be well below values which are below that, which says,
then, the threshold may be inscrutable, inscrutable in the sense that
they don't provide you a way, really, of seeing, but I'm sure we'll talk
about that at some point.
MR. DEAN: Yes, we will.
DR. BONACA: Because what is being published in internet,
you're saying, really is only the performance indicators and not the
cornerstone performance indicators.
MS. MADISON: We're publishing the performance indicators,
as well as the inspection findings, which cover the whole cornerstone.
DR. BONACA: So, you publish that, too.
MS. MADISON: Yes.
DR. BONACA: Now, here you're talking about an SDP green.
We haven't seen that. I don't understand exactly how that works.
DR. APOSTOLAKIS: Before we leave the thresholds, one last
Why, then, if this is the thinking, did the staff feel that
it was necessary in establishing the threshold between green and white,
that you had to distinguish between some plant types? In the electric
power, I think you had something there. I don't remember now which one
MS. MADISON: We had to distinguish between plant types
because of the safety systems involved, because BWRs and PWRs don't
necessarily have the same safety system. INPO did the same thing in
their indicators, and we mimicked that to have the same four safety
systems at each plant type.
DR. APOSTOLAKIS: Wasn't there also a distinction between
plants with different numbers of diesels?
MS. MADISON: Yes.
DR. APOSTOLAKIS: So, different kinds of redundancy, then.
So, why would that apply to a threshold between green and
white and not -- well, a higher threshold and not at the baseline? What
is the logic? Why are we departing from the idea of a generic threshold
at that level, but at the lower level we don't?
MR. PARRY: This is Gareth Parry from the staff.
The reason we made that distinction or the reason we did it
for the green/white threshold is because of the way we established the
thresholds, which was to use historical data to determine that
threshold, as you've describe it, and that's based on a single-train
This is going to be part, I think, of a somewhat longer
discussion later, I guess.
DR. APOSTOLAKIS: Okay.
MR. PARRY: Let's come back to this.
MS. MADISON: We will come back to this.
DR. APOSTOLAKIS: Okay.
MR. DEAN: Another principle I wanted to discuss real
briefly was the fact that, to obtain a level of adequate assurance of
performance, that we need both the performance indicators and the
When we go out and make presentations to the public or to
other stakeholders about this process, there's a tendency to latch onto
the performance indicators as being the end-all and be-all, and they're
They're a complementary set of indicators, information by
which we need both of those to be able to judge -- adequately judge
performance at a plant.
The revised oversight process, in utilizing these
performance indicators and these inspection findings, has developed a
process whereby our assessment of license performance is more of a
continual and ongoing assessment process, as opposed to -- for example,
we mentioned earlier about the SALP process, where maybe every 18 or 24
months, you would get a package that gave you an assessment of plant
So, we have embodied in this new process a much more
continuous and ongoing assessment whereby every quarter, as we get new
performance indicator information and as we update our inspection
finding plant issues matrix, that you get an additional set of
information by which you can add that onto your previous information and
use that to judge on a more continuous basis licensee performance.
The performance indicators obviously have a much more major
role into this process than they did in the past. Performance
indicators in the past were really used more to perhaps provide a level
of support or a confirmatory tool, as you will, for decisions when we
got into the senior management meeting process.
We would look at, well, what do the performance indicators
say and do they jibe with what our inspection findings told us, which is
really what we based our assessment on licensee performance on, really
was inspection findings.
So, now we have integrated performance indicators to provide
some at least more objective tools in that area.
The issue of cross-cutting areas -- and this gets back to
the earlier question about performance -- problem identification and
Within this process, I think as you're all aware, that we've
identified three areas that we consider to be cross-cutting areas, that
they find their way into all the cornerstones of safety in terms of
contributing to the attributes, and that would performance -- problem
identification resolution, human performance, and safety conscious work
environment, and it's important to note that, in the revised oversight
process, we're assessing performance in the cornerstones.
I've heard mentioned a couple times an overall assessment of
the cornerstones. We're not providing an overall assessment of the
cornerstones like we did with an overall assessment in the SALP process
of a functional area, okay?
What we're doing is we're identifying issues within a safety
cornerstone, assessing that issue more discretely or assessing that
performance indicator, which is an indicator of performance within that
cornerstone, and dealing with those issues on a more discrete basis, and
as those issues emerge with either a higher threshold being crossed or
as you get more issues within that cornerstone, then what you see is an
analogous NRC regulatory response -- a greater level of inspection,
supplemental inspection, more focused team inspections, as you see
higher thresholds being crossed or as you see more thresholds being
crossed within the cornerstone, but we are not, in this process, trying
to, quote/unquote, assess a cornerstone like we did assess a functional
area with our more subjective process in the past.
DR. BONACA: Let me just ask a question.
There is clearly a perception on the part of the industry
that -- I quote here a statement in the NEI 99-02, a draft of it,
regulatory assessment performance indicator guideline, where it says
that a green performer from performance indicators only -- a green
performer will be allowed to identify and correct perceived problems,
which means essentially that the NRC action or interaction or
intervention is going to be determined by the performance indicators.
MR. DEAN: No. The interaction is determined, as we
mentioned earlier, on the completely integrated set of performance
indicators and inspection.
DR. BONACA: Well, I think we will have to ask the industry
later on if it is the same conclusion they have documented here in this
draft, because when I read that, it says that the performance indicators
being in the green may, in fact, be an impediment to the staff to look
at other things or to take action based on cross-cutting issues.
MS. MADISON: It's always been advertised that the
performance indicators, from the beginning of developing this program in
SECY 99-007, that the performance indicators could not stand alone, that
the had to be supported and supplemented by baseline inspection program
and that just because performance indicators are indicating good
performance did not mean that we wouldn't react or wouldn't take action
based upon inspection findings.
DR. BONACA: Even if everything was green.
MS. MADISON: Even if everything is greener than green in
the performance indicators, if there are indications in the inspection
program, then we'll take action based upon that.
DR. SHACK: Are they weighted the same? That is, if you go
through an inspection and you go through an SDP and you come up with a
white, is that a white like a performance indicator white?
MS. MADISON: Yes, that's the purpose, and I'm going to try
to explain a little bit of that during the SDP, and Mike will go into it
more in the assessment program.
MR. DEAN: The intent was to try and brace our thresholds on
the guidance that's contained within Reg. Guide 1.174 and try and make
the performance indicator thresholds analogous to the inspection finding
Now, is it exact across the board? You know, obviously not,
but I think that we've come pretty close in trying to make them similar
so that a white here and a white here are equivalent.
MR. BARTON: That's an important point, because under the
current process, you could have good PI's and still be in trouble.
MR. DEAN: Oh, yes. Matter of fact, I'll give you a good
example. This came up, matter of fact, in a discussion last night.
I was up in New Jersey last night, matter of fact, speaking
to the public on the new oversight process, and the issue came up about
the complementary nature of inspections and PI's and could something be
evaluated as green in PI's and potentially mask a potential problem, and
in fact, in New Jersey, we've had recent incidents where, in the
emergency preparedness area, the performance indicator has been green,
it's shown good performance over the last year in terms of EP
performance, but that there have been several actual events at Salem
where you have had some problems in --
MR. BARTON: -- misclassification.
MR. DEAN: -- misclassification of events, and that was
evaluated through our inspection program and determined to be a white
issue, even though the green performance indicator in EP would show that
-- you know, give you an indication that performance in that area was
MR. BARTON: So, what does that tell me?
MR. DEAN: So, what's that telling you, is that that's a
good example of where the PI's and the inspection process are
complementary in nature, the fact that the performance indicator is not
the overall indicator of performance in that area, it's an indicator of
performance with a specific aspect within that cornerstone but that our
inspection program is complementary or supplementary to what we get from
the performance indicators and that we may have issues emerge that a
performance indicator doesn't give us the same information that our
MR. BARTON: What does the public see in that case? What's
on the internet?
MR. DEAN: What they would see is they would see, underneath
that cornerstone, okay, if you're familiar with our web-site, you know,
the single page, you have the cornerstones and the PI's underneath that
cornerstone, and then, below that are the inspection findings, and what
they would see is, under that inspection finding, the block for that
current quarter, when that inspection finding emerged, would be colored
white, and then they could click onto that box and it would take them
right to that description of what that inspection finding was as to why
we characterized it as a white issue.
DR. BONACA: The performance indicator was green.
MR. DEAN: That's right.
DR. BONACA: So, you would have not only a white, you would
have a performance indicator of green and then you would have an
MR. DEAN: An inspection finding of white, that's correct.
MS. MADISON: And they would both be inputs into the
assessment program, as Michael described, and the same action would be
taken for a white inspection finding as a white performance indicator.
MR. DEAN: Before I move off the slide, I want to make one
other point, and that has to do with the problem identification and
We recognize that, in establishing this band of performance
and backing away a little bit, as you will, from focusing on these
low-level issues and trying to drive their resolution, that we have to
rely on a licensee's ability to identify and resolve their problems more
substantially than we have in the past.
In order to provide us with some level of assurance that a
licensee does have an effective problem identification and resolution
process, we have embedded in every inspectable area a portion of that
inspection procedure has to focus on problem identification and
resolution activities associated with that inspectable area, and that's
a substantial change from our previous inspection program, where we may
do, every couple of years, perhaps, a programmatic review of a
licensee's problem identification and resolution or their corrective
We have now embedded that in each and every inspection
procedure, as well as having a periodic annual inspection that looks at
problem identification and resolution from a broader perspective.
So, we are spending a lot of our inspection resources and
effort to look at problem identification and resolution, much more than
we did in the past.
DR. BONACA: In your guidance to the resident inspectors,
you specify that, if you have a number of misclassifications, that would
correspond to a white? Is there a criterion for determining that?
MS. MADISON: It's in the significance determination process
for emergency preparedness.
DR. BONACA: Well, that still leaves it to the -- you don't
have a head count. I'm trying to understand how objective that process
MR. DEAN: We're going to talk a little bit about the
significance determination process later. So, hopefully, we'll be able
to address that.
The last point I want to make before we start talking about
some of the technical aspects of the program is that the oversight
process is intended to be indicative within the licensee response.
I think we've talked about that already several times, that
we are backing away from having a more diagnostic approach for those
very low-level, low significant issues, and that's a purpose of our
risk-informed baseline inspection program.
It's intended to be indicative, are we getting indications
of problems whereby, if we do see issues that are crossing
risk-significant thresholds, that would then engender additional or
supplemental inspection, which is designed to be more diagnostic in
nature, it's intended to be looking at what is the root cause that the
licensee has conducted say about that issue, what have they done in
terms of looking at extent of condition.
And as you see more thresholds being crossed or higher
thresholds being crossed, that supplemental inspection becomes much more
independent in terms of its level of diagnostics, and the oversight
that's based on our action matrix -- as I have stated several times, the
action matrix is one of the tools that we have in place to help make our
process more predictable and understandable as to why it is we're taking
the actions that we are taking and that a licensee or the member of the
public can predict and understand why it is we're doing whatever sort of
inspection or regulatory response, whether it be a 50.54(f) letter or an
order -- they can understand what performance issues have led to us
taking that action.
So, that's one of the major intents of the action matrix.
DR. APOSTOLAKIS: Can you explain the first sentence? I
don't understand it. "The oversight process will be indicative within
the licensee response band." What does that mean?
MR. DEAN: I guess what that's referring to is that the
performance indicators, okay, provide indications of performance, are
not measuring performance, but they provide you indicators.
The inspection program is designed to identify indications
of potentially poor performance that have some risk significance, and
so, the process, as long as a licensee is within the green band, their
performance indicators and the inspection findings are characterized as
green, then our process in that realm is more of an indicative process.
We're looking for indications of potential poor performance.
Once you emerge from the green band, you cross a threshold,
whether it's a performance indicator or whether it's an inspection
Our process now is designed to be more diagnostic with
respect to that issue or with respect to that cornerstone, if you have a
degraded cornerstone, if you have several issues within a cornerstone
that cross thresholds.
So, now we move into more of a diagnostic, trying to
understand why is this happening, why did you have issues that caused
you to cross this PI threshold or cross this risk significance threshold
for the inspection findings?
So, there's a shift in our focus of what we're trying to
understand about licensee performance.
DR. APOSTOLAKIS: By the way, again, for my education, when
the inspectors perform the inspection, are they using generic criteria
or plant-specific criteria?
MR. DEAN: In terms of assessing the significance of the
DR. APOSTOLAKIS: Yes.
MR. DEAN: We're going to talk about the significance
determination process, but there is, I think, a -- you know, your
concern about generic thresholds and so on.
I believe that our significance determination process and
the tools that we have for the inspectors to use are much more
plant-specific in terms of looking at, you know, what mitigating systems
are available and so on and so forth.
DR. APOSTOLAKIS: Right. So, the ADP is plant-specific.
MR. DEAN: Yes.
MR. BARTON: Yes, it is, George.
CHAIRMAN POWERS: We've asked in that regard why it is --
it's plant-specific, but it appears to me that they have gotten that
plant specificity by looking at the IPE submittals.
DR. APOSTOLAKIS: That's right.
CHAIRMAN POWERS: And those IPE submittals are now, what,
eight years old?
At the time they came out, the committee was acquainted with
some substantial concerns on whether the analyses in the submittal
represented a complete set of accidents and whether the IPE was, indeed,
faithful to the plant design.
Since that time, anecdotal accounts suggest that several of
How do you correct for that?
MR. DEAN: Alan is going to specifically address that issue
and the concern, and I think we're probably ready to get into Alan's
We'll start with the PI's first, right, Alan?
DR. APOSTOLAKIS: I don't think I got an answer to my
During the inspection, in an inspectable area, the inspector
has industry-wide criteria in his mind or the history of this plant and
how things were done --
MS. MADISON: The simple answer to your question is yes,
both. They're going to have to use some industry-wide guidance.
There's industry-wide standards that they'll be looking at,
but there are plant-specific implementation standards that they'll also
be concerned with and plant-specific design characteristics that they'll
be looking at when they're doing their inspection.
So, the inspection program has both elements in it, both the
generic, both industry-wide-type concerns, as well as a plant-specific
DR. APOSTOLAKIS: So, in my mind, then, the only part of the
whole process that uses generic numbers is the thresholds for the
performance indicators. Everything else is more or less plant-specific.
It doesn't mean you ignore the industry, the rest of the industry.
MR. DEAN: In a general sense, that's accurate. I will say,
for example, in the significance determination process, for example,
initiating event frequencies are basically generic, industry-wide
initiating event frequencies, and a specific plant may have a different
factor built into their IPE that may emerge as you get further into the
risk analysis of an issue, but you know, there's generic aspects to the
significance determination process, although that process, I believe, is
much more aligned towards the plant-specific design.
DR. SHACK: The inspector will be looking for all,
essentially, violations of the licensing basis, just the way he does
now. It's the SDP that suddenly becomes different.
MR. DEAN: Yes, what do you do with those findings and
issues. Do we have something -- a compliance issue that is significant?
If it's not a significant issue, we turn that over to the licensee, they
are still required to comply with the regulations.
It's just that we will not expend a lot of our effort to
drive resolution of that. We'll come back and revisit it as part of our
corrective action program reviews, but it's not --
DR. APOSTOLAKIS: So, it confirms what I said. It is
DR. SHACK: But what he's looking for is essentially a
violation of the licensing, which I guess is plant-specific, yes.
MS. MADISON: Well, we've changed the focus a little bit,
and we're trying to focus them on risk significance rather than
violations, and in fact, that has occurred during the pilot program.
Some of the issues -- some of the significant issues that
have been raised have not been necessarily violations of regulations,
but they have risen to a level of significance that we were concerned --
and the licensee was concerned with the issue.
MR. DEAN: Alan?
MS. MADISON: We're going to talk first about performance
indicators and then about the significance determination process, and
we're going to try to address a couple of the questions that you had in
A little bit later, Gareth Parry -- in fact, in a few
moments, I hope, Gareth Parry is going to be talking about George's
specific issue on plant-specific thresholds.
DR. APOSTOLAKIS: I'm not trying to be a bad guy.
MS. MADISON: No, no, we're trying to address your
MR. BARTON: When did you change, George?
DR. APOSTOLAKIS: I'm really troubled by this.
MS. MADISON: I just wanted to highlight a couple of things
about the performance indicators and the thresholds.
The purpose of that green/white threshold was to indicate or
identify licensee performance below which we needed to start getting
involved as an agency, we needed to start getting, as Bill said, more
diagnostic rather than indicative, and instead of turning the problems
back over to the licensee, focusing on them ourselves and trying to
determine more of the why's.
We're not -- within that green band, as long as it's above
that green/white threshold, we're not ranking, we're not trending within
that green band, although some licensees are, and in fact, in some of
the performance indicators, we don't think it's appropriate,
necessarily, to trend, especially like in the barrier areas, because
they're more data point-type indicators.
Again, one of the other things to focus on is they are
indicators, they're not measures, and in some cases, we don't have a
real clear tie to risk some of these indicators. So, we're not looking
at them as a straight-line-type measurement of performance.
The green/white threshold, as we've been talking about -- we
initially set that trying to come up with like a 95-percent performance
area, but it's really focused on -- with the concept that we have looked
at industry as a whole and feel that industry as a whole is performing
Now we're looking for outliers, folks that are outlying from
nominal performance, and the development of that threshold, then, was
based upon this 95-97 history, saying if that's a safe history, then
where were the outliers in that time period and where would we establish
a threshold to capture those outliers in the future?
CHAIRMAN POWERS: If I have a plant that, because of a
design characteristic, some peculiar feature of it causes me to be in
this upper 5 percent, and there's nothing I can do about it, it's a
design feature, it has been accommodated and corrected with some
compensatory action, presumably, in the licensing process and it's fully
documented, everybody understands that, do I still end up getting a
MS. MADISON: It's a good point. We haven't seen that,
actually, in the reactor safety areas, but we're likely to see that in
some of the non-reactor areas, and we're addressing that by -- our
proposal for addressing that is we recognize that performance.
For example, in the security area, where comp measures may
account and provide backup for some security equipment, but we do need
to, for the public's scrutiny and to maintain a stable program, we will
identify that as a white issue or a white performance indicator, but
we'll note what actions are being taken by the licensee and by the
agency to address that issue.
CHAIRMAN POWERS: I think you're inviting difficulties here.
You have set your thresholds for green/white so high, 95th percentile
high, that you've given white, which on reading the words is not
particularly bad -- it's only requiring some additional attention,
whatnot, it has not impacted the public's health and safety the least
little bit -- you are drawing attention to it.
A white in a field of green stands out, especially since
there's no gradation in the greens, and I think you invite trouble if
you ask people to look at the asterisk that said this is okay.
I don't think it will be captured. I think you invite
difficulty to that plant.
MS. MADISON: And we'll have to look at that, Dana.
MR. GILLESPIE: I think one of the important aspects is what
Alan said, is none of the existing plants seem to have the problem. So,
we want to be careful that we don't try to fix something that's not a
Now, if someone builds a new plant and does it, well, that's
okay, but we've got a number of years to deal with that, quite honestly.
So, you know, we're trying to get a process in place, and
this really hasn't become an issue, and even in security, we're
re-examining the threshold itself to ensure it's not an issue.
MS. MADISON: I was just going to mention that. We have --
one of the bullets on here says we will re-evaluate those. We are
re-evaluating those thresholds based upon the historical data that the
licensees gave us on the 21st of January.
In looking at -- we're considering raising the threshold --
or lowering, actually, the threshold on the security equipment
performance index, but we're still going to -- we still identify some
outliers, and that's the purpose of the index.
I think it's about seven or eight plants that we think will
be identified based upon that, and in talking to our security folks,
they're considered true outliers in performance in the industry.
It may be because of some design concerns that they have on
their security equipment, but their security equipment is considered as
an outlier in performance in the industry.
We haven't seen anything in these performance indicators
that would say otherwise so far.
DR. WALLIS: How many PI's are there?
MS. MADISON: There are 19.
DR. WALLIS: Nineteen. So, it's conceivable that the 95
percentile will identify 50 percent of the plants but only in one PI.
MS. MADISON: We're doing 95 percentile per performance
DR. WALLIS: That's right. So, it could be that this field
of green, every plant could have a white on something.
MS. MADISON: Yes.
DR. WALLIS: This isn't 5 percent of plants in that regard.
MS. MADISON: Per indicator, that's correct.
MR. JOHNSON: But there is no denying the point that Dana
makes, that the relative rarity of whites makes the pucker factor for
when you get a white very high, and that's something we've seen in the
pilot program, and I think Alan's right.
DR. BONACA: I do believe one of the reasons why you don't
see more of these whites that Dana is talking about is because they are
I mean they are so high, the thresholds, in my judgement,
that the issue of 5 percent of the plants for some indicators -- like,
for example, barrier performance. I don't know where you have one of
I mean 50 percent of your tech spec value on containment
leakage, on fuel activity -- I mean you could have bundles of fuel
leaking to get those kind of values.
Again, I think it will go down to the last bullet and it
will talk about then the objective shouldn't be that you change a
threshold when you have an increasing risk. We already said you're not
measuring the risk.
The objective should be that you have a sensitive enough
indicator that it will tell you something.
MS. MADISON: There are some exclusions to that, and in the
barrier indicators, we did not choose on the 95 percentile. We chose
based upon tech spec limits, and if you look at the tech spec limits,
they are really a very small percentage of the Part 100 limits in the
barriers. So, the impact on true risk to the public is very, very
small, even at 50 percent of tech spec.
DR. BONACA: If the objective was purely the one of looking
at increasing risk, but I thought the objective was the one of being
able to see, I mean to have a sensitive indicator that will tell you
this -- there is a trend.
MS. MADISON: And we're looking at those thresholds. We're
also looking at those indicators to determine whether or not we keep
those indicators, because of that very concern.
I think I've talked about that pretty much.
As I mentioned, we are probably making a change to the
security equipment performance index. We're looking at all the other
thresholds in the performance indicators.
As you see on the last bullet, we're talking about the -- at
least for the initiating events and mitigating systems cornerstones, the
yellow and the red thresholds did have a direct tie to risk, in our
estimation, as we developed those thresholds.
In those performance indicators that do not -- for example,
the safety system functional failures do not show a direct correlation
to risk -- we chose not to have a red threshold. We chose just to have
the lower thresholds, because the action taken based upon the action
matrix would be sufficient to get to the root cause of problems in those
DR. APOSTOLAKIS: If one indicator is yellow, then I have a
delta CDF of about 10 to the minus 5. If two of them are yellow, what
happens? Two times 10 to the minus 5?
MS. MADISON: It's strictly on one performance indicator at
a time, but in the action matrix, we try to then add those issues
together to accelerate our action taken to address the problems.
At this point, if there's no other clarification questions,
I'd like Gareth --
CHAIRMAN POWERS: I have a clarification question.
MS. MADISON: Oh, I'm sorry, Dana.
CHAIRMAN POWERS: And it's in this last one, and it comes to
this red corresponds to about 10 to the -- a delta CDF of 10 to the
Maybe we take something that everybody seems to focus on,
scrams, and I look at the information used to come up with that, and I
guess I don't understand exactly how you got the number you did and why
it's pertinent, because I certainly see plants that get about a 10 to
the minus 4 with scrams much lower -- that get scrams much lower than
your yellow-to-red threshold, and I see others where the number of
scrams has to be much higher to get about a 10 to the minus 4.
When I try to say, okay, this top 5 percent of those, I
don't find that in -- I mean just going through the numbers, I don't get
that same number.
MS. MADISON: If the explanation in SECY 99-007 was
inadequate, I'll get someone to -- I would ask Gareth if he would add
some detail to the discussion on the scrams. Gareth, along with several
others, helped develop the thresholds on that performance indicator.
MR. PARRY: I'm not really sure I understood your comment
CHAIRMAN POWERS: I guess what I'm asking is really the
mechanics of deriving the threshold values.
MR. PARRY: The white/yellow and the yellow/red thresholds.
CHAIRMAN POWERS: Any one of them would probably help me,
but I focused here just because the yellow-to-red has some
quantifications with it, so I could go back and check.
MR. PARRY: Right.
Well, the way that was done was to take the parameter in the
suite of PRA models that we used and varied it from the base that was in
the model until we got a delta CDF of 10 to the minus 5 or 10 to the
minus 4, and you'll see that there's a significant variation between
plants, but for most of them, the simple reactor trip, which is the
parameter we used, is not a major contributor to risk, and that's why
you see these rather large numbers associated with the thresholds.
CHAIRMAN POWERS: I believe the number you came up with
there -- and correct me if I'm wrong -- is about 50.
MS. MADISON: No, 25.
MR. BARTON: There is a 50, I think, at one time.
MR. PARRY: For one of the plants, maybe.
CHAIRMAN POWERS: Okay.
If I used the criterion, most of the plants -- then most of
the plants in the tables would be 100.
MR. PARRY: A lot of them would be, that's true, but that's
just a reflection of the fact that simple reactor trips are not major
contributors to risk.
CHAIRMAN POWERS: Whatever they are, they are a performance
indicator, they have a threshold, and I'm interested in how the
threshold was found. Someone has asked me what is the technical
foundation for these thresholds, and I've got to answer him, because he
has a higher pay grade than I do.
MR. PARRY: I've just tried to explain how we did it.
CHAIRMAN POWERS: And I understand, but when I try to go
back and look at the numbers and re-do it myself, I don't come up with
MR. PARRY: How can you re-do the numbers without having the
CHAIRMAN POWERS: Well, I've got these tables.
MR. PARRY: Okay. I see what you're saying. In a sense,
what we did, I think, to come up with the final number which we used was
-- well, we just said it was greater than 25. It's just large. It's
not a very useful threshold in that sense, because it's so large.
CHAIRMAN POWERS: I have one particular plant where, in your
tabulation, it says, gee, if they have more than seven, they've got a
delta CDF of 10 to the minus 4.
MR. PARRY: Okay. That's probably a SPAR model, right?
CHAIRMAN POWERS: And then all of the others -- I mean they
can go up to 100. Here's one with 35. Here's another one that says
greater than 50.
It's not apparent to me how the number was actually arrived
MR. PARRY: Where is the seven? Which plant is this?
CHAIRMAN POWERS: If you look in Appendix H, page H-9, Table
2. Maybe I'm misinterpreting the numbers.
MR. PARRY: Okay. These are the risk significance scrams
that you're talking about.
CHAIRMAN POWERS: Yes.
MR. PARRY: Okay.
MS. MADISON: That's a different scram.
MR. PARRY: These are essentially losses of feedwater.
MS. MADISON: That's a different indicator, though, Dana.
CHAIRMAN POWERS: Yes, I understand it's a different
indicator. Many of your thresholds are very, very subjective, by your
own admission, because you have no quantitative tool to deal with them.
A couple of them you have quantitative tools to deal with.
I'm just trying to understand how you got the actual numbers
in a way that I can go back and reproduce and say, oh, yes, had I
accepted all your assumptions, all your predications, which I'm willing
to do, I, too, would have come up with this number.
MR. PARRY: This is over a year ago now.
CHAIRMAN POWERS: Well, maybe you can give that some
MR. PARRY: There is a discussion of that particular plant,
which is Palo Verde, and it's a design-specific feature, I think, of
that plant, which is the reason why that one comes out a little low, and
I think the exception is that we're going to set it at 10 for the
white/yellow except for those plants where feed-and-bleed is not an
option, which Palo Verde is one of them, I think, and it says that this
plant will be treated in a design-specific way.
CHAIRMAN POWERS: Okay. But you see what my problem is.
You set the number at 10. All the other plants -- I mean they get
numbers of 46, greater than 100, 34, 21.
MR. PARRY: Well, you're mixing two tables there.
CHAIRMAN POWERS: Because I'm just asking a question.
Now, if you want to get specific on one, I'm perfectly
willing to do it. It sounds like you don't have a facile answer to my
MR. PARRY: I think the simple answer to your question is we
looked at the results, we chose the lowest of the set of those results
and chose that as the threshold, unless there was a reason for an
exception, and in this case, for the risk significance scrams, that was
true, because Palo Verde does not have the feed-and-bleed option.
CHAIRMAN POWERS: Okay. So, I can go back and reproduce
your numbers by looking at these tables and come up with exactly that
MR. PARRY: I think you should be able to understand where
the numbers came from. You might come up with a slightly different
perspective, because we've probably done some rounding-off here, but
yes, you should be able to read Appendix H and come up with those
CHAIRMAN POWERS: It will stun me if I do.
MS. MADISON: I'm not laying it all on Gareth's plate
either. Gareth worked with several other folks in industry as well as
NRC, and their discussions, which he probably can't remember now, after
over a year, led to that type of decision.
CHAIRMAN POWERS: I guess I think this is bad practice to
establish thresholds and not have a good documentation on exactly where
those numbers came from, because sooner or later, at some time in the
future, perhaps after Dr. Parry has left the agency for greener pastures
or more delightful pursuits, somebody's going to want to change those
DR. APOSTOLAKIS: Want to define greener? What's the
CHAIRMAN POWERS: Well, it's not white. He's got white here
today. He doesn't need that anymore.
MR. DEAN: Dr. Powers, you make a good point, and one of the
things that we intend to do once we can get out of the developmental
phase and get into a more stable implementation phase is to go back and
develop what we are going to call a basis document that will help do
exactly what you describe, what was the basis for all these decisions
that led to the thresholds, and collect that all in one document so that
there is, indeed, not the reliance on more, okay, but there is a
documented basis that we have in one place.
It's in a number of different places, 99-07, 07-alpha.
There's a lot of places.
MS. MADISON: Yes, we've taken that on as a task. It's kind
of the never-ending-job part of the process.
DR. APOSTOLAKIS: I have another question. Again, it's
Let's take two plants. One, as the IP's have found, is from
the ones that have a core damage frequency greater than 10 to the minus
4, let's say 5 10 to the minus 4, 19 PWR units were found to have that,
and then the other one has a core damage frequency of 3 10 to the minus
5, so big difference.
There will be random changes in performance, right? I mean
it's not always consistent.
Wouldn't it be easier, due to random causes, for the plant
that is already at 5 10 to the minus 4, to have a delta CDF of 10 to the
minus 5 or more, easier than the plant that's already down to the 10 to
minus 5, because that plant has to double its CDF.
So, for the plant that is already at the 5 10 to the minus
4, would I expect it to be in the yellow region a lot of the time,
whereas the other one would not?
MR. PARRY: I don't think that's necessarily the case,
because you are talking about -- you have to decompose what goes into
that 10 to the minus 4, and if the thing that you're changing is in a
very low cut-set, the delta might be the same for both plants.
DR. APOSTOLAKIS: Yes, but it seems to me, if I'm already at
5 10 to the minus 4 --
MR. PARRY: But if we're working on deltas --
DR. APOSTOLAKIS: -- changes there on the order of 10 to the
minus 5 would not be something that would surprise me.
It would even be sensitive to the way I calculate things,
because that's not a first decimal place, where the other one is way
down there, something really drastic has to happen for 10 to the minus 5
So, the question is, should they prepare to see more yellows
for the high core damage frequency plant, and what does that say about
the process? I don't know.
MS. MADISON: Well, there's two answers, I think, to that,
and Gareth started with one of them. That's not necessarily the case,
just because there's a greater risk overall at that plant, that the
change will be greater based upon an equipment failure.
The second is, you know, if part of the purpose of the
program is to focus our resources more effectively where the risk to the
public is greater and if the risk at that plant to the public is greater
and they have more problems and they do go into the white or the yellow
more often, that's where we should be focusing our resources.
DR. APOSTOLAKIS: But the question is whether getting into
the delta CDF of 10 to the minus 5 is something that's sort of expected
due to random causes for that plant. So, there is no reason for alarm,
whereas for the other plant there should be.
I don't know the answer myself.
MS. MADISON: We'll have to watch that during the
implementation phase. That's, again, another question that we'll have
to try to answer during initial implementation.
MR. DEAN: And one of the other things is that, within our
program, part of our inspection procedures is an event response element
which is defined to allow the agency to react appropriately to issues
that cross thresholds but to look at other performance attributes of
that that have to be evaluated.
So, in other words, you may have an event that, because of
the very nature of the event, has a certain risk significance to it and
that we would want to respond with a certain inspection reaction, but
that may not, it and of itself, be a relationship to a performance
issue. It may be something that's related to the actual risk
characteristics of the plant.
MR. GILLESPIE: George, is your fundamental question, if
someone's got a plant that's designed with more redundancy in certain
functional areas, do they have an advantage? The answer is yes. The
answer to that question is yes.
DR. APOSTOLAKIS: I guess I was coming from another point of
If already the CDF is high, then we anticipate random
changes with time around the baseline value, which is an average value.
So, if I'm already high, a delta CDF of 10 to the minus 5
should be something that I should see very often in my plant, because
that means small variations with respect to my baseline CDF.
MR. GILLESPIE: Yes. Now you're exactly where the staff was
in wrestling with thresholds, because up until this point, the criticism
was the thresholds are too tight, and this argument could be used for
saying the thresholds go the other way, and that was precisely the
problem in being risk-informed, by the way, not risk-based, that we
needed to wrestle with.
Now, what the industry data is showing us -- and the team
just got all the industry data in on the PI's -- they have to step back
and look, does that profile look the same as the pilot plants and what
was anticipated, and they're still in the process of kind of doing that,
but we're not necessarily anticipating, I don't think, a lot of whites
by design, if you would.
MS. MADISON: And we have not seen a lot of problems with
the historical data submittal that we would necessarily need to change
thresholds dramatically, although we're still -- it's under review. We
think there are some changes to be made, but we're still looking at it.
Did you want Gareth to talk about the plant-specific issue,
because he has some information he'd like to share.
DR. WALLIS: I have a follow-up question to George's
question about plants with a large CDF.
Now, you get into red by doing something which is increasing
your CDF by 10 to the minus 4.
Can you get out of it by fixing something else which has
something to do with something completely different from what got you
into this red, because you already have a large CDF to play with, so you
can fiddle something else to get you a negative delta CDF to cancel out
the one you've just gotten.
MS. MADISON: It's issue-specific.
So, if you have a piece of equipment failure -- pardon me --
a performance indicator that causes an availability of that piece of
equipment, the emergency diesel generator, to be out for that period of
time, that is an unavailability number that will cause you to cross a
There's not another piece of equipment, as far as that's
concerned, you can throw that back.
DR. WALLIS: You can get back, though, by -- in that
specific -- delta CDF of 10 minus 4 -- by getting back part-way, till
it's a half 10 to the minus 4, then you go back to yellow, or do you
have to fix the whole thing?
I mean you could get out of this state in the same way you
got into it, by reversing exactly the same thing that you got into --
MS. MADISON: By reversing performance.
For example, the scrams -- once the scram crosses the
threshold, that number stays there for a certain period of time.
DR. WALLIS: You can't cancel that out.
MS. MADISON: Well, you can't cancel out unavailability of a
piece of equipment either.
DR. WALLIS: You're bound to stay red for a long time if you
have a lot of scrams, no matter what you do?
MS. MADISON: As your critical hours increases, as your
denominator increases, that number will go down.
MR. BARTON: That's no different than what industry does
not. You cross the threshold, it just stays in there for a few years.
DR. WALLIS: You could also cross the threshold by making
some error which you could fix.
MR. SIEBER: You can't.
DR. WALLIS: You can't?
MS. MADISON: There is an issue in unavailability with fault
exposure hours if you find a design problem which you might consider an
error that's been around for 20 years. An aggressive program on the
part of the licensee, going out to look for design issues, they find
this issue, and when looking at it, it says that piece of equipment
would have been unavailable because of that design issue.
Now, we've tried to accommodate that in the process by
saying that could probably cause you to stay white, yellow, or red for a
long period of time.
If that issue is corrected, if we have reviewed and found
that issue -- the correction to be adequate and we've documented that in
a report, after four quarters, we'll remove that fault exposure hours to
take that off of the books, number one, to say -- you know, to
compensate, you know, it was not necessarily a performance issue on your
part, it was something we needed to focus on, we needed to apply some
resources, and number two, we don't want to mask any future issues that
may crop up, because you have this large number of fault exposure hours
due to a design issue.
We're looking for that type of issue if it comes up in other
performance indicators, and we may need to make similar type of
DR. KRESS: Where in the performance indicators do you
incorporate this time element? If a performance indicators jumps above
the threshold, do you say it has to reside there a certain amount of
time before you trigger some sort of action?
MS. MADISON: No.
DR. KRESS: If you could have a time element, it could take
care of George's problem of randomness --
DR. APOSTOLAKIS: Yes.
DR. KRESS: -- because it wouldn't be there long if it was
random, and if it were a real performance degradation, it probably stays
there a long time.
MR. JOHNSON: Well, remember what we do with all of these
things in terms of the action matrix -- and all of this is driving to
get us to a point where we can decide what the regulatory response
should be and what the licensee response should be, and in fact, the
consequence of, you know, a spike above a threshold, for example, the
consequence -- the ultimate consequence for a white is that we go do
some additional inspection and do some diagnostic look, and in fact, the
result of that inspection could indicate that this was random.
DR. KRESS: Okay. That would be another way to deal with
DR. APOSTOLAKIS: You had how many, six pilots, six pilot
MS. MADISON: Nine.
DR. APOSTOLAKIS: Are the baseline core damage frequencies
for these plants available easily?
MS. MADISON: I'm sure they are.
DR. APOSTOLAKIS: And the question is really did you check
whether there was any correlation between the findings and the baseline
MS. MADISON: Frankly, we didn't have enough findings
greater than white to draw any kind of conclusions in that area. We'll
have to look at that closer during initial implementation.
DR. APOSTOLAKIS: Okay.
MS. MADISON: Gareth?
At this time, I'd like Gareth to address the issue of
MR. PARRY: Let me see if I get this straight. You'd like
to see plant-specific thresholds. Is that right, George?
MR. PARRY: Okay.
DR. APOSTOLAKIS: All this, you know, everything we do
depends fundamentally on what the process is designed to achieve, and it
seems to me -- so, the first thing is we have to have consistency
between the objectives of the process and the way it's implemented, and
then we have the second issue, do you agree with the objectives?
So, from the discussion today, and other discussions, I get
the impression that the process is really designed to maintain or to
alert the staff that the level of safety at that plant has changed in
the wrong direction, because if it changed in a good direction, we
really don't care.
So, if you start with that premise as an objective, then
everything else has to be plant-specific, and that would be consistent
with what we do in other parts of the regulations -- as I mentioned,
I mean we agonized over what is negligible, minimal,
whatever other terms we used, but we said we really want to maintain the
licensing basis, and a lot of other things.
Now, if you start from that point of view, then you're
saying, well, gee, you know, I want a plant-specific set of, say,
performance indicators, but then you may decide that the performance
indicators really should be the same for all plants because of certain
reasons, but you started with the idea that you would try to define it
on a plant-specific basis.
Then the thresholds, which is a separate issue, you know --
I may decide that the PI's are generic, but then the thresholds -- and I
think that's where my disagreement is -- again, I start with that
premise, they have to be plant-specific.
Now, for certain things, I may decide that, you know, the
number I'm using for plant X really should be applied to a whole class
of plants or maybe all the plants. That's fine, too, but you started
again from the fundamental premise that it has to be plant-specific.
If you start with generic, then there are all sorts of
problems with inconsistency and so on, and this is really my fundamental
problem, the consistency with the objectives of the process and then
what are the right objectives of the process.
MR. PARRY: Okay. And I think the staff's on record as
saying that, certainly in the ideal world, we would like plant-specific
thresholds, and I'm trying to think of the way that you'd set this up.
Now, presumably, if you had a good PRA model for each plant,
you could extract from that the appropriate parameter value that would
give you what the long-term expected value of the particular PI would be
that would -- that gives you that level of risk, and you could use that
as the current status of the plant, if you like. Okay?
So, you'd have a target, much like the maintenance rule for
setting their goals on reliability and availability.
Now, what that represents, though, is a long-term average
about which we're going to have statistical fluctuations. Let's get to
that in a minute.
First of all, we're making an assumption here that that
value of the PI is going to be dramatically different from plant to
plant -- at least it's going to be different from plant to plant and
that the variability has a direct correlation with the level of CDF.
I think if you look at the particular parameters that we're
dealing with, which are diesel generator unavailability, HPCI
unavailability -- if you look at those from plant to plant, from the
plant's own assessments and also from the AEOD assessments, you're not
going to find that that varies tremendously, and perhaps it's more a
function of the fact -- and the argument's a little easier to make, I
think, for the HPCI pumps, where they tend to be fairly uniform design
-- that perhaps what we're seeing is more a fundamental limitation on
the way they can be maintained and operated rather than a conscious
decision that, in some plants, we have to really look at this carefully
to maintain the level of risk, and I think, if you look at that
variability, you're not going to see a great variability, which argues,
I think, for the fact that, at the level of the indicators that we're
talking about, that the generic types of values and thresholds are, in
fact, not such a bad approximation.
Now, the other thing is, I keep hearing that you think that
the green/white threshold is high.
DR. APOSTOLAKIS: I think a lot of people think that.
MR. PARRY: Yes. I'm not just saying you. We keep hearing
that it is high.
DR. APOSTOLAKIS: Yes.
MR. PARRY: If you look at the thresholds, in fact, they are
not so very high compared to typical unavailabilities that you see in
The other thing I would point out, too, is that we did look
at this in terms of sensitivity studies, and if you look at the back of
Appendix H, you see, for a couple of the plants, what we did was we took
all the PRA's for the reactor, for the initiating events and mitigating
systems, we bumped them all to the green/white threshold, okay, so
they're all at the top, and in putting all of those at that value, we
still didn't generate a delta CDF that was 10 to the minus 6, and it's
because you can't say that there's one plant that contributes at the
highest level to each of those indicators, but I think these are
plausibility arguments, I think, to suggest that the thresholds we've
chosen are adequate for the purpose that we need them for and that they
are not -- it's not necessary to have plant-specific thresholds.
DR. APOSTOLAKIS: No, but it seems to me -- you see, I think
we have an issue of presentation here, because the logic that you
followed is exactly the logic I would follow.
Now, the last sentence I disagree with, because if you did
all that, then it is plant-specific.
Plant-specific does not mean that the number is different
for each plant. The numbers may turn out to be very close, and you say,
well, I'll pick this number, but this is not what's written here.
The second question -- so, you really follow the logic that
I'm advocating, because your argument was that we really looked at
plant-specific data, but we concluded that there was not really
variability, and instead of producing 103 numbers that are within the
noise of each other, we said, well, go with this, which I think makes
perfect sense, but that's not what it says here.
Second, though, you said that you looked at the data and you
didn't see variability.
Now, I think that deserves some discussion, because if I
look at page H-18 and H-19 and I read in the text that the bars that I
see here are the highest values per year over five years -- these are
not just the actual numbers, these are the highest numbers.
I don't know that if I look, say, at Figure H.1 on page
H-18, that I can conclude that there is no variability. I mean on what
basis are we deciding that this is statistically insignificant?
On what basis are we deciding on the next figure for BWR
high-pressure injection system unavailability, that for example -- you
see if I look at plant 54, it has an unavailability of perhaps .005 or
something, or 6, and then I have other plants that have, you know, .04
and even higher.
So, the argument that we looked at the plant-specific data
and we didn't observe any significant change, I'm not sure how you can
justify that in light of this evidence, because these are the highest
numbers. These are not the actual numbers.
MR. PARRY: I don't want to mislead you, George. We didn't
look at this to see the plant-to-plant variability. What I'm saying is
that I looked at various sources like IP's, the AEOD studies, okay, and
in those studies, the long-term average of the unavailabilities that
people use for these systems are not that variable. That's what I'm
Sure, we expect to see variability because of the
statistical variation, because what we're dealing with is relatively
DR. APOSTOLAKIS: But Gareth, if this is the highest over
five years, the highest per year over a period of five years --
MR. PARRY: It's the highest three-year rolling average over
five years, I believe, is the way it is.
DR. APOSTOLAKIS: Then that should be a robust indicator.
I mean that shouldn't change that much, and if I have these
differences that I see in these pictures, it seems to me I should worry
about this variability, and maybe that should be an input to the studies
that the AEOD has done and so on, but it's not something that
immediately convinces me that the generic value would be good enough,
because for example, for the high-pressure injection system
unavailability, for plants 51, 54, 55, 62, the threshold is way too
high, because it was driven by other plants.
Now, again, should you, within the time constraints you had,
have developed plant-specific values for all these? Probably not. But
in the maintenance rule, you ask the licensees to do it for you. Why
don't you do the same thing here? Spread the work.
There are several issues here -- the conceptual issues, the
practical issues. You know, I do appreciate that you're under
tremendous time pressure, but --
DR. BONACA: I think you have a problem, also, with the fact
that thresholds are being set at a value where safety is not degraded.
So, there is a presumption of risk information, and we are
talking about whether it's legitimate, but I thought the more important
issue we discussed at the beginning of the meeting was that you should
be able to trend performance from what it was last month or last quarter
and see trends.
I mean the importance of the indicators is trending, it
seems to me, and again, I do believe that, by using the criterion of
safety is not being degraded, you're making these indicators too
insensitive to changes, and I still think there is a confusion between
what objectives you have --
MS. MADISON: I think the safety degraded issue is a
measurement sense of how degraded is it. Any error, any industrial
error that's made out at the plant could be construed to be a reduction
in safety or an increase in risk.
We're saying that, at the green/white threshold, it is a
very small increase -- it's a small increase in the risk.
DR. BONACA: I understand that, but let me just tell you why
I have just a very practical concern. I want to get away from -- the
practical concern is, if I get all greens -- assume we get all greens --
and that's probably going to be true for many plants -- then the
indicators are irrelevant.
I am going back to everything else you've got in the
inspection program, the baseline program, being the fundamental element
that you're looking at.
However, you still have a statement of all green, and it may
be an impediment or a reduced ability on your part to rely on --
MS. MADISON: What it means, being all green, is that we
will do the baseline inspection program, and the baseline inspection
program can still identify problems.
DR. BONACA: I understand that, but I'm saying that if, in
fact, it doesn't give you an ability to discriminate and it's generally
green, why do you have it at all? Just throw it away.
MS. MADISON: Because that's one of the basic premises of
the program, that there is a level of performance at the licensee below
which we don't need to get more involved other than the baseline
That is the basic premise of the program, that says if they
are green in those indicators or if they are green in their performance,
that the maximum level of involvement we need to have at that point is
the baseline inspection program, and if that's true for all plants, then
MR. DEAN: And remember -- I mean our inspection process is
a sampling process.
I mean you very well know that we don't look at every aspect
of every plant operation or activity we sample, and that's one of the
reasons why we have a continuous inspection program and why we go back
and do the same inspections, looking at the same things, because maybe
this one time we don't find that performance issue but maybe another
time we will, and then we can pull that thread and maybe uncover --
DR. APOSTOLAKIS: Is plant 54 on two different figures the
MR. PARRY: I think it is, yes.
DR. APOSTOLAKIS: Okay.
So, if I look at Figure H-2, 54 is a very good performer,
page 19, H-19.
The BWR HP injection system unavailability is very small.
Then I go on to the next figure, which is emergency AC power
system unavailability; 54 is again among the best performers.
Then I go to the next figure, BWR RHR system unavailability
-- 54 is doing very well.
Three indicators and 54 is one of the best plants. So, the
threshold is higher than the performance of this plant in all three
So, this plant now would be allowed to deteriorate its
performance on three key indicators and it would still be in green,
because the threshold is determined by the performance of other plants,
and is that something that an oversight process ought to allow?
MS. MADISON: On an individual basis -- that's a question
that we're going to have to answer for the program sometime after
One of the processes that Bill was going to get into at the
end was we need to develop a oversight process for the oversight process
-- in other words, a quality assurance process for this where we look
back at the program, at industry as a whole, in the macro sense, and say
have we maintained safety in the industry, or because of what you're
talking about, George, has safety in an industry-wide sense decreased,
and that's a question we're going to have to answer, and we've committed
to answer that question in June of 2001 based upon our review of initial
Now, that's going to be a process that's going to look at
some industry-wide indicators that we may already have, some yet to be
developed, and it's also going to look at, programmatically, how it's
been implemented and some of the lessons we've learned out of the
program, but that is a question we're going to have to answer.
DR. APOSTOLAKIS: Now I think it's clear what my fundamental
problem is, that there is this plant 54 that is doing very well, and now
you're setting the threshold so high that this plant, on three key
indicators, can -- not that they will try to do it, but if its
performance deteriorates on all three, the agency will do nothing.
MR. GILLESPIE: Inherent in that is they're still in
compliance with their license.
CHAIRMAN POWERS: Suppose that it is true that this plant is
doing very well because of heroic efforts on its part.
It has concentrated resources, it has focused its
engineering department, it has focused its training program. It does
very well on these things, at the expense of substantial investment on
Why shouldn't it be acceptable for this plant to relax on
that because it is going -- it's over-devoting resources in those areas
and should be devoting in the areas such as fire protection or emergency
preparedness, where maybe it doesn't do so well?
I mean why isn't that one of the objectives of having a
risk-informed regulation, that we want resources to go to the areas
where they're needed, as opposed to areas that are thought to be
DR. APOSTOLAKIS: Well, that argument, by itself, could have
some merit, although we would have to look more carefully.
CHAIRMAN POWERS: You can understand why sometimes our
planning and procedures meetings go long, with this kind of support from
DR. APOSTOLAKIS: But it would be absurd for the same plant,
if they want to change some minor thing, to have to argue that they are
within the requirements of 50.59. Either we change all the regulations,
then, to say, you know, go ahead and change things and we'll look only
at what's important or try to be consistent.
MS. MADISON: A lot of those are consistent in that they
utilize Reg. Guide 1.174 as kind of a basis for deciding whether or not
to make that change, and we've tried to rely on Reg. Guide 1.174 to also
help us define what is the significance, and 1 times 10 the minus 6th or
1 times 10 to the minus 5th fits into Reg. Guide 1.174 and the backfit
rule as far as what is considered a significant enough issue that we
need to get involved or we need to review that closely.
MR. BARTON: I think the bottom line here -- and you can
change indicators -- and George's concern is, well, this plant can now
be allowed to slip, and Dana's saying, well, why not, because they're
focused on other areas. Fine.
But isn't the real key here how is the new process going to
allow you to identify problem plants in a more timely basis than the old
process, which was a criticism of the old process?
MS. MADISON: Exactly.
MR. BARTON: Now, how are you going to be able to do that
with all these fluctuations and this, well, it's got a little bit of
risk, but it's still in the licensing basis, and there's this new
process to allow you to identify the problem plants on a more timely
MR. DEAN: The combination of having a more frequent
submittal of information that reflects on plant performance, that we're
looking at this information on a quarterly basis, that we have in place
a predictable process by which we can react to performance issues, that
we will, as performance degrades, apply a greater amount of resources
and focus on those plants to better understand the issues.
I think just the very fact that we have a periodic updating,
a public display of what the overall performance assessment is of the
plant is, in and of itself, a substantial driver of trying to enhance
consistent safe performance of the plants, because there's going to be a
lot more public pressure, as you will, to maintain indicators within
appropriate bands of performance, so that the NRC is not engaging --
MS. MADISON: I think there's two ways.
Number one, by the significance determination process and
the indicators, helping us identify those areas that aren't as
risk-significant, we can stop looking at those areas and spending
resources there, freeing up those resources to look at more
risk-significant areas, where we can identify problems, will help us
identify problem plants in a more timely manner, and as Bill mentioned,
the more frequent information coming up, but frankly, it's the idea that
we put out a system that establishes -- that says, if we have these
inputs, this is what we're going to do. We're no longer a black box.
You know, the Arthur Anderson study -- one of the things --
in '96 -- one of the things they said is we're an agency that had more
information available to us than any other regulatory agency they had
seen. It wasn't that we lacked information at the facilities. It's
just we may have reacted slowly to it.
What this process does is it puts out in front and
advertises this is what we're going to do if we get this information,
and to react differently, we're going to have to justify that. That's
new. That's something we haven't had in the past, and I think, frankly,
it will force us to react in a timely manner.
DR. APOSTOLAKIS: I have a second thought.
DR. KRESS: I do, too.
CHAIRMAN POWERS: Let's let Dr. Kress introduce his
DR. KRESS: As you said, how you implement this system
should start from your fundamental objective. It looks to me like the
fundamental objective is to keep the performance below an acceptable
level -- I mean above an acceptable level.
CHAIRMAN POWERS: Up and down is going to be a problem.
DR. KRESS: But if you viewed that as the objective, then
all this makes sense. The acceptable levels are the thresholds, and it
doesn't matter how far below you are, as long as you're below it.
DR. APOSTOLAKIS: I would answer you the same way I would
answer Dana, that this is a noble objective, rearranging resources and
so on, but I don't think it's the job of this particular regulation to
do that. That's why we have 1.174.
If the licensee feels they can spend the resources in a
better way, they can always come to us and argue to change -- you know,
to request a change in their licensing basis, and we have other
regulations that deal with that. The job of the oversight process is to
make sure that what we approved remains the way we approved it.
MS. MADISON: We would disagree.
DR. APOSTOLAKIS: It's not the job of this regulation to
CHAIRMAN POWERS: I think we've got to move along.
MS. MADISON: I guess we want to ask you at this point --
where do you want us to go? Because I think we've reached our 10:30 --
MR. BARTON: Well, NEI does not have a 15-minute
presentation. As I understand, they just want to make some comments.
Is that true? Is NEI here?
MR. HOUGHTON: Tom Houghton, NEI.
We didn't have a prepared presentation. I think we laid out
our issues at the last meeting.
Industry has presented their data to NRC. We're satisfied
with the program as it is right now and ready to move ahead, and we
believe that there are a number of issues that require looking at the
thresholds, and that will continue in the public venue.
Preliminary look at the data that's been submitted, although
it hasn't all been verified yet, would show that there are a number of
plants which have exceeded the threshold, and so, it is not a program
that will result in all greens for everybody.
CHAIRMAN POWERS: Well, can you speak to the issue of
someone exceeding the green-to-white threshold, for example, not because
of any poor performance on their part but, rather, because the way the
threshold was chosen is not consistent with the kind of design they
I mean they are forced into exceeding this threshold by
design, even though the plant has, throughout the licensing process,
been found certainly safe enough, maybe even exemplarily safe.
MR. HOUGHTON: There were a number of plants that thought
that the thresholds would unfairly treat them. Those issues are being
looked at. However, the preliminary data shows that it hasn't
The plants with the whites that we see so far -- their data
shows that they've had unavailabilities which would show up in the data
and for which they want to make correction.
CHAIRMAN POWERS: And that's kind of the same answer that
you had found, but again, we go from pilots to more extensive, we need
to be alert to that, and we'll have to figure out some way to handle it,
because I think it does not serve any of us well to have a plant
highlighted for no reason.
MS. MADISON: No.
MR. DEAN: I will share one of the things that we would have
gotten to if we had continued the presentation --
MR. BARTON: I think you need to continue to make the points
you want to make on PI's and then jump into the SDP, because we haven't
even talked about that yet.
MS. MADISON: Do you want us to continue on, then?
MR. BARTON: Yes, I think so.
MS. MADISON: Really, the only point I wanted to make on the
next slide, really focus on, was kind of respond to your question you
had on the SSA, why it was not used, although there are some other
issues as far as how we developed these performance indicators.
I just want to remind you there was a rigorous process that
-- we thought a fairly rigorous process -- to go through and select
performance indicators and drive down through what we called the
football diagrams to look for important attributes, important areas to
measure, and see if there was a performance indicator available.
Now, I'll refer you to SECY 99-007 on page I-11, and we
responded to this question over a year ago about the SSA, and our answer
was the SSA indicator proposed by NEI did not differentiate between
plants or add any new information.
Only one plant, a declining trend plant, was in the white
band, and it was also in the white band for transients.
Lowering the threshold by one would capture two average
plants and three watch list plants, all of which were identified by
In addition, the SSA indicator did not show a strong
correlation to the discussion plants in the Arthur Anderson analysis.
For these reasons, we did not include the SSA.
And I guess the two points there I want to highlight are,
you know, the information was provided -- that the SSA provides is also
provided by other indicators. So, we have bounded the SSA.
There's no new information provided by the SSA, and we felt
the other indicators actually were better indicators.
MR. BARTON: I think that was part of a larger question
which said are you satisfied that you've got enough indicators to be
able to assess performance?
MS. MADISON: In connection with the baseline inspection
program, again, you know, being aware that it's not just a performance
indicator program. It's an oversight program that includes performance
indicators and inspection.
MR. DEAN: Could we have better indicators or indicators
that would give us a more comprehensive view of plant performance?
MS. MADISON: Yes. And we're looking at those, and that's
the next slide, the ongoing work.
You had some questions about the other long-term issues that
were out there.
We're continuing to look at the consistency of PI
We feel we're consistent right now, agency-wide, with the
availability definition, and that includes the maintenance rule folks,
because we worked with them to help develop this definition.
Industry, NEI has agreed that this definition is correct.
INPO has agreed that the definition is correct. They're trying to work
with WANO to get them on board.
We finally found an agency that's slower to react than we
DR. APOSTOLAKIS: Would you add the two issues that we
raised today to the ongoing work, please? The definition of the
objectives for the program and the issue of plant-specific information.
Or is that something that you will do in the future?
MS. MADISON: We'll take those questions back and look at
them again. I'm not sure we'll add them to our workload. We'll have to
look at those two questions, though, George. I've noted them in my
DR. APOSTOLAKIS: With the industry and WANO, since the
fundamental objection here seems to be that the thresholds are too high,
it's not really surprising that the industry supports this, is it?
MR. BARTON: Not to me.
DR. APOSTOLAKIS: So, that doesn't really mean much. You
are giving them more than they have now, so why should they object?
MS. MADISON: We're also satisfied. We feel that the level
DR. APOSTOLAKIS: I understand that.
MS. MADISON: -- noted by the performance indicators allow
us enough opportunity to get involved and to do our inspection
activities and identify our concerns before there is unsafe performance
at a plant.
MR. BARTON: George, what I'm hearing loud and clear today
-- maybe we didn't focus on it or absorb it in the past -- is I keep
hearing the inspection program really is what they're relying on, since
the PI's have got, you know, these concerns that we've been talking
about. I think the key here is how good is the inspection program and
the SDP process.
DR. APOSTOLAKIS: And it's plant-specific, so I'm happy.
MS. MADISON: Do you have any other questions on what we
have -- the last two bullets up here are really kind of what I had
mentioned earlier as far as developing an oversight program for the
oversight program, looking at those long-term, a self-assessment
program, looking at industry-wide performance, and we're developing --
as we said, we're continuing to look at additional indicators, better
The research has an effort right now ongoing for us to look
at risk-based indicators, and we're going to consider those in the
DR. BONACA: One last comment I'll make is that I still have
a problem in answering the question about the technical adequacy of the
performance indicators, and the reason is, anytime I raise an objection
to them, or to the threshold set for those, I get statement that says
but we have the baseline inspections and we have the significance
I mean -- and it leads me into limbo about, you know, the
significance of those indicators. That's a problem I'm having.
And I understand the program. Actually, you know, I'm more
impressed today because I heard that you're going to have some kind of
gradation you are going to make, also, on your baseline inspections, so
therefore you have some ways of balancing the full green from the
indicators or something else, but still, I've got a problem in
addressing the question regarding technical adequacy, because anytime I
find some problem with it, I get an answer that says but there is
MS. MADISON: That's one of the problems we had in
developing the program.
The first proposal by NEI was a program that relied entirely
on performance indicators, and when we looked at the performance
indicators that were available, that we had, industry had, or that we
could devise quickly, we couldn't find any that we could just rely
strictly or solely on performance indicators.
We had to devise a program that was supplemented and
complemented by inspection.
We're going to continue to look at the performance
indicators and try to come up with better ones, and more technically
adequate, but remember, again, they're not measures, they're just
indicators, and even if -- and Arthur Anderson said this -- if you add
enough numbers together and you can show a correlation to performance,
by looking backwards, then you probably have an indicator of some worth,
and we've proven that with the safety system functional failure
DR. BONACA: And yet, they will be questioned and judged
independently as a set. Independent of all the considerations we are
making here, there is also the baseline inspection, because that's the
way -- how things happen.
You have a matrix there, and people are going to ask
questions specifically about those.
MR. DEAN: You're absolutely right, and we've struggled with
that in every meeting, whatever venue we have, that there tends to be a
focus on the performance indicators as being a complete, comprehensive
set of information, and they're not.
MS. MADISON: I want to move on now to the significance
determination process, kind of focus a little bit on the basics of this,
and I understand Mr. Bonaca has not read SECY 99-007A that describes --
one of the appendices to that describes the basis for the significance
What we wanted to do and devise was a simple tool for
inspectors to use to characterize inspection findings, and what we
wanted to do was make sure that the output of the significance
determination process correlated closely to the output of the
performance indicator process, colors with the same relative risk
It's an approximation within an order of magnitude,
hopefully a conservative approximation, but it's an approximation.
We're not trying to draw any bright lines between
performance. We have numbers associated with the thresholds, but
they're approximate numbers.
So, in determining the characterization of the significance
of an inspection finding, there's no difference, in our minds, between
.8 and 1.1. They're the same. It's a fuzzy line, in other words.
The SDP process goes through -- I'm just going to quickly
First of all, the input to the significance determination
process is the output of one of our documents, the manual chapter 0610,
which says that the basement of issues or the threshold of issues to be
discussed in the inspection report is right about the minor violation
threshold, and that's true for issues that aren't necessarily
They have the same relative risk significance
characterization of issues that are not violations that you would
discuss in an inspection report.
That's where we define what we call a finding.
Those issues, then, can be put into a significance
determination process, and it's not just one significant -- the
significance determination process, there's multiple processes.
We have one for the reactor side of the house, but it
doesn't have, right now, the issues of containment or shutdown involved
in it. We're still developing those.
There are other processes for the non-reactor side, for EP,
safeguards, and there's actually a couple of processes in the health
physics area, but they're tools. Again, they're simple tools for
inspectors to identify the relative risk significance.
The phase one part of the process is a screening process
that, on a conservative nature, says does this inspection finding have
any likelihood of being greater than green, and if it doesn't pass that
screen, it is a green finding, it should be turned over to the licensee
for evaluation and correction.
If it has any likelihood of being greater than green, it
goes to phase two.
Phase two involves, then, the site-specific work-sheets that
have -- we've gone back and looked at initially the IPE's, the
information that we had available on the docket from the licensees,
developed the site-specific work-sheets, and then went out to the sites
and looked at, site-specifically, what issues, what changes to the
sequences should be made, what changes to the event frequencies should
be made, and what other mitigating systems should be considered within
that phase two screening.
That phase two screening is, again, more site-specific, more
involved, but it, again, is a conservative screen, and it's an
approximation screen in orders of magnitude.
That is an initial determination of the relative risk
significance of the inspection finding.
The phase three review -- what that phase two review, then,
does is say this is definitely greater than green, this inspection
finding is definitely greater than green, and it should be considered by
a -- in a more rigorous manner, and we throw this into the SRA's, the
Senior Risk Analysis in the region, as well as in headquarters, who then
look at this issue more closely and determine its actual risk
So, the phase three is more detailed and would use more
discriminating tools, more definite risk models than the significance
determination process, to come to a final determination.
I see some questions.
CHAIRMAN POWERS: I guess that the phase three is a
problematic area in your first attempts to do it? I get the impression
that phase three may be a time-consuming activity done largely outside
the realm of public scrutiny?
MS. MADISON: That's a definite perception. There's a
couple of reasons for that, we think, that we have tried to address.
The first phase three review that was attempted to be done
was we had not awoken to the fact that site-specific phase two
documents, work-sheets were necessary. We were still under the
misconception that we could do this generically, and that happened at
the -- Prairie Island raised an issue that definitely needed
site-specific information on, and because of that new knowledge, it took
us an inordinate amount of time to come to conclusion on that inspection
The phase three review that was done at the Sequoyah issue
was more of a -- involving how much is enough due process allowed to the
licensee, how much information should we be gathering from the licensee,
how much input do we need to have from the licensee before we come to a
We discussed this at the lessons learned workshop, and one
of the conclusions that we came to is that, in agreement with industry,
because of the public perception issue, when we make the initial
determination that this is a risk-significant issue, that is has
potential for being white or greater, we should document that in a
report; the public needs to have notification of that immediately, and
that's what the new process should have.
So, when the initial phase two review has had some screening
by management, some oversight, that will be documented in an inspection
Now, after that point, we do need to allow -- we may need
more additional information from the licensee, more technical
information from them to complete our review, and we do need to allow
them some sort of appeal process, but that will be further structured
within the process.
MR. JOHNSON: It's a good question. It's not a new issue.
It's, in fact, an issue that we've dealt with for a long time in the
enforcement program, as you're well aware. Escalated enforcement
actions have taken time to resolve.
We've got some challenges. We need to be open to the
public, and we're sensitive to that.
We also need to have a process that allows the licensees to
respond to us.
And so, it's working out how we're going to do that with
this new process that we've run into some challenges and we're putting
in place some fixes.
MR. DEAN: Yes, but I think it's important to emphasize,
just like our current process, if there's an operability issue, that's
dealt with in an immediate nature.
So, there's no change in the fact that, if we've got a
concern about operability of a piece of safety equipment, that's going
to be dealt with in an immediate fashion.
MS. MADISON: But the other part of your question, Dana, is
the -- there is more time required to review the issue, once it's raised
to the level of significance, there is more demand for technical
knowledge in the area of risk, there's more demand on the senior risk
analysis with this process, the analysts, than in the past. We
recognize that that may be an impact on our resources that we're going
to have to address.
CHAIRMAN POWERS: I think we're quickly running out of time
in an area that still is fertile for discussion. I personally have
quite a few questions on the significance determination process, not so
much in those that are clearly treatable with risk analysis tools or
those that are clearly un-treatable by risk analysis tools but those
that lie in they should be able to treat with risk analysis tools, and I
can see you're still struggling with some of those, and I also think
this phase three needs to be looked at in a lot more detail as we gain
MR. BARTON: We've got a Commission paper on this subject
the middle of February. We'll have these people before us again in the
March meeting, and our letter to the Commission is due in March.
Would it be appropriate to get into further discussion on
the SDP in the March meeting, or is that not timely enough for you?
CHAIRMAN POWERS: I guess we're going to have to discuss
that. I don't know what we'd do given our constraints of schedule and
whatnot, because I think the SDP discussion is protracted. I think we
have a number of them that we need to walk through to understand why it
is not capricious and arbitrary.
MR. BARTON: Right. I understand that.
DR. APOSTOLAKIS: Are you -- we have already mentioned
problems with the third bullet, you know, the availability of PRA's and
the IP's, the problems they have, but even if one had a good PRA, a lot
of the inspections deal with issues that are details, the noise of the
PRA, so you would have to be a little creative to see how this finding
affects the PRA.
But in light of all these issues, are you prepared to tell
the Commission that there is a need for research in this area, that this
process will not work very well until we have reasonably good PRA's that
can be used in phase three, and possibly in other phrases, to determine
the significance of issues?
MR. DEAN: I don't think so. I think that, once again, what
we've tried to develop is a risk-informed and not a risk-based process,
and one of the challenges that we have as a staff is trying to make sure
that the significance determination process is as Alan described, that
it's a usable and relatively simple tool that an inspector can use to
provide some risk characterization to his inspection finding that can be
easily communicated to the public and to his management as to why we
believe this issue is important and needs to be dealt with, and we have
to be very careful that we don't fall onto the side of trying risk-base
our process where now we find ourselves into this realm of PRA's with a
lot of uncertainties, and I think that's all recognized, that risk
analysis is still an uncertain proposition in a lot of respects and that
the assumptions that are made, you know, have to be considered.
MS. MADISON: It's really lessons learned out of the
Sequoyah issue. One of the reasons why it took so long is the licensee
kept trying to provide additional information to have us cross this --
what in their minds was a line that we had to cross to get them below
green or below white, but we told them that, because of the
uncertainties, there is no fine line and we didn't consider the
information was enough to cause us to change our opinion of what the
characterization of that issue was.
DR. APOSTOLAKIS: I think, although I appreciate your point,
there is an unintended consequence which is not insignificant for this
Because the staff is reluctant to say when they are dealing
with specific problems and to tell the Commission that there is a need
for research in certain areas, the Office of Research is viewed as
almost unnecessary, and some Commissioners, in public speeches, have
expressed doubts about the need for any further research, and it's
understandable, because the staff never comes back to them to say, gee,
we really can't do this very well unless certain issues are resolved
which are properly within the domain of the Office of Research.
So, I don't know how we can face this, because you know, how
can you do bullet number three there if you have an IPE which had a
different objective, you know, looking for vulnerabilities and so on,
and the Commissioners are not aware of it?
If we don't tell them that, for some cases, the tools are
not there, why should they know? They're not going to go a conference
and read the papers.
So, it seems to me there are conflicting interests here.
On the one hand, of course, you don't want to say, gee, we
can't do this because we don't have perfect tools, but on the other
hand, it seems to me that that attitude, for a long time, has created
the impression on the decision-makers that the Office of Research is not
MS. MADISON: We think with this process, number one, as far
as the SDP process of phase two, we do not need a perfect tool.
DR. APOSTOLAKIS: It doesn't have to be perfect, Alan.
MS. MADISON: We're looking for something that is close
enough, that gives us a characterization of the finding within a band,
and there are uncertainties to it, but there are uncertainties to the
models, to the SPAR models, to the other models that we have. There are
uncertainties there, as well.
DR. BONACA: I would like to ask a question regarding the --
this assessment program.
The question is this:
You have an event -- for example, a misalignment, which may
be significant, and you have a process now by which you're going to
determine the significance of that, and I could go right through it, and
you can come up and say that it was not safety significant, and that's
the conclusion of that.
What if you have a situation where there are multiple
misalignments taking place in a given period of time, okay? Is the
significance process going to be applied to that condition, and how
would it be treated?
I didn't understand by reading that document how that would
For example, you may get lucky and you may have 10
misalignments, and none of them is safety significant, yet the fact
itself that you are having these multiple repeats --
MR. BARTON: It's a programmatic problem.
MR. DEAN: What you're getting, Mario, is something that's
been at the core of a lot of concerns on the part of our inspectors, is
that what do I do with that situation where I have green issues --
here's a green issue, here's another green issue -- I never tripped that
significance threshold, but I'm seeing a pattern and a trend that I
believe is indicative of a potential programmatic problem, and the
Commission, if you back to the SRM that the Commission gave the staff,
after we briefed them on the pilot program preparations, is that they
told us they did not want us aggregating green issues to try and come up
with a different risk significance number, but on the same hand, they
told us to make sure that this program was robust enough to detect
programmatic breakdowns. So, that puts us in a tough situation.
I think what you've seen -- and I discussed earlier about
cross-cutting issues. I think where we see that type of performance
having an impact is in cross-cutting areas.
A number of human performance issues occur over a period of
time, problems in not identifying problems or recurrence of problems
that you thought you resolved, and so, one of the things that we have
included into the program that will be part of the ongoing structure is
to allow our inspectors to be able to weigh in on those situations where
the issue, in and of itself, may not have caused a SDP threshold to be
crossed but that they have seen over a period of time a collection of
these issues and that we want to make sure that we raise the forward in
the inspection report and in the assessment process to make the licensee
aware of the fact that we've seen this pattern or trend, you ought to
DR. BONACA: Just a comment about the process. In fact, I
would have liked to see a question, this is event and there are these
boxes that throw you to a green or send you further in the process.
What would be important to us is the question, are similar
events occurring? Are events with similar characteristics and so on and
so forth -- I mean I think some improvements can be made in the
determination process. I understand where the Commission is going, but
that's an important issue.
MS. MADISON: Well, as far as concurrent issues, if you're
talking about concurrent failures --
DR. BONACA: No, not concurrent, just saying, hey, is
something else happening of a similar type that tells me there is a
That thinking process doesn't address the specific
significance of the event, but it tells me, in fact, if I had a
programmatic breakdown or at least I should be looking into it, and I
think that that would be an important part of the -- because otherwise,
to say the safety determination -- it's almost like a hand waver at
times in other plants to say, oh, but that wasn't safety significant,
you know, so that's no problem, no issue.
Well, there are issues which are important just because they
happen on a certain frequency.
DR. APOSTOLAKIS: I think we should wrap up now, because
there are many important issues.
DR. BONACA: I understand, but I think this is a very
MR. BARTON: I think we need to continue this discussion the
next time we meet with the staff.
DR. APOSTOLAKIS: So, if you had one minute, how would you
wrap up the presentation? No more transparencies.
MR. DEAN: I would wrap up the presentation by leaving the
message that we think that the process that we have designed, the
revised reactor oversight process, is, on a broad number of measures and
given the direction given to us by the Commission, is a substantial
improvement in terms of its structure and its framework as to how we go
about the business of overseeing nuclear power plant activities and
I think the pilot program has given us a substantial set of
information and lessons learned that we have made revisions to and are
working on, making refinements to this process to prepare us for
implementation of this program at all sites.
MR. BARTON: Initial implementation.
MR. DEAN: Initial implementation at all sites, so that we
can utilize the increased scope and breadth of information and
experiences to really fully flesh out the process and be able to address
some of the underlying concerns that not only our internal inspectors
have but a number of external stakeholders about we just aren't
convinced, we're not sure that the pilot program has told us enough, and
we agree with that, and we think that we need to expand the process to
be able to hopefully build the confidence in our inspectors and in our
public stakeholders that, indeed, we have established a good framework
and a good process for a reasonable assurance of plant safety.
DR. APOSTOLAKIS: John, anything else?
MR. BARTON: I don't have anything else.
DR. APOSTOLAKIS: Okay.
We'll recess until 10 minutes after 11.
DR. APOSTOLAKIS: The next subject is proposed final
amendment to 10 CFR 50.72 and 50.73.
Dr. Bonaca is the cognizant member.
DR. BONACA: The staff plans to present a proposed final
amendment to 10 CFR 50.72 and 50.73.
The objectives of the proposed amendment, just to remind,
include to better align reporting requirements to the NRC's reporting
needs, to reduce the reporting burden consistent with the NRC's
reporting needs, and to clarify the reporting requirements where needed.
The staff has met with the industry and other stakeholders
during several workshops and meetings to discuss the proposed
We, the ACRS, reviewed the proposed amendment in March 1999
and issued a letter which included a number of conclusions and
recommendations that I will read here, restate.
One, issue the proposed amendment for public comment.
Two, eliminate the requirement for reporting late
surveillance tests by amending the rule and not by revising the
associated regulatory guide.
Three, the staff should comprehensively examine the NRC
reporting requirements to assure no duplication or inconsistencies.
And four, plant-specific lists of risk-significant systems
should be developed, and they should not be included in the rule.
NEI is concerned with the addition of the requirement for
reporting components. It believes that the requirement lacks clarity,
is ambiguous, and does not warrant backfit, and they are here to, I
believe, provide us with a presentation of that.
We would like the staff, during its presentation, to
specifically address their recommendations concerning plant-specific
lists of risk-significant systems and NEI's concern with added
I would like just to add one question, which is, over the
past month, I have received two drafts of this proposed amendment with
In fact, the last one I received was last Monday and had
significant changes from the December 30th, and I just really wonder if
we are ready to have a final amendment because of that.
I would like you to explain how these issues, which are not
unimportant -- like, for example, the systems which are listed in the
rule, which were taken out in December and now are put back in -- you
know, if we have now a final position on that.
With that, I'll let the staff go to its presentation.
MS. MALLOY: Thank you.
I am Melinda Malloy. I am the Section Chief in the
Rulemaking Group within NRR. The branch that we reside in is the
Generic Issues, Environmental, Financial, and Rulemaking Branch, in the
Division of Regulatory Improvement Programs.
We are, I believe, prepared to address your concerns that
you've raised, to answer the questions that you would like us to
address, and we'll get to them throughout the presentation.
As you know, the proposed rule was published for public
comment back in July of '99. We received 27 letters of public comment,
mostly coming from the industry, and they were critical of the couple of
the areas that you've mentioned.
The staff has worked very hard over the last few months to
take the public comments to heart and to develop revisions at the
rulemaking that we feel are responsive to the public comments but, at
the same time, preserving the staff's need for information.
We have undergone extensive internal reviews over the last
two months, and that's probably the main reason for the revisions that
you've seen, but I think we can say with great confidence that we are at
a point in time where the staff -- and we are talking not just NRR
staff, but we have coordinated extensively with IRO, as well as Research
and other interested parties, to come up with workable requirements for
the rule, and so, with that, I would like to introduce the other folks
that are here to support this briefing.
To my immediate left is our Deputy Division Director, Scott
Newberry, who I think you've seen from time to time, and to his left is
Denny Allison, who is the Task Leader for this particular rulemaking,
and Denny will be giving a presentation for us.
We also have in the audience some key members of our
internal stakeholders that are here to help support us during this
So, go ahead, Denny.
MR. ALLISON: Dr. Bonaca, thank you for the introduction.
As far as -- I'll deal explicitly with the ACRS's
recommendation about the list of systems in the presentation, as well as
with NEI's concern about which -- their biggest concern, of course, is
with the proposed new criterion that was in the proposed rule.
With regard to whether we're ready, I think we have a
position that will be the staff's position. I think it's final.
I'm waiting yet for Brian Sheron's side of NRR to wade in
formally, but we've met twice with all the division directors in NRR,
and the first time we agreed on what to say, in general, and then the
second meeting, with all these same guys, was about how to --
specifically how to say it, because we had some problems with the words.
So, I think that the Federal Register notice that I've
provided to you is the staff's position. I hope so.
MR. NEWBERRY: Well, let me clarify, Mr. Chairman. We're at
the point where what you'll see here today is the proposed position.
We're at the point of filling in aspects of the Federal Register notice
and perhaps some examples in the NUREG.
So, I would request that, you know, what you see here we
work with and that being the proposal in front of the committee.
MR. ALLISON: Now, the objectives of this rulemaking I think
most people subscribe to. That's to clarify the requirements, where
that's needed, to reduce unnecessary burden, not to reduce worthwhile
burden but unnecessary burden, and use risk-informed thinking.
You know, I wouldn't call the whole rule risk-informed, but
we've got some risk-informed thinking in the changes we're making, and
to be consistent with the NRC's new programs, and particularly the new
oversight programs, and in a nutshell, that means don't get rid of
things we need for that program.
DR. WALLIS: Immediate?
MR. ALLISON: I'm sorry?
DR. WALLIS: Immediate means in the blink of an eye or what?
MR. ALLISON: Where does "immediate" --
DR. WALLIS: "Immediate" is key in the first two --
MR. ALLISON: Oh. Yes, sir. That's the title of 50.72, and
all of the requirements in 50.72 are stated that way. Declaration of an
emergency class is to be reported immediately after the state is called.
DR. WALLIS: Well, I was sort of intrigued by the term
"immediate NRC action." How fast can the NRC do anything?
MR. ALLISON: Well --
DR. WALLIS: This means within a day or something?
MR. ALLISON: No. Immediate --
DR. WALLIS: Fifteen minutes? So, it's less than an hour,
MR. ALLISON: Yes, sir, although there are four-hour and
eight-hour reporting requirements, but those are also stated as as soon
as practical and in all cases within four hours.
The principle changes that we're making are we're deleting
outside the design basis of the plant, and you'll see another slide in a
minute as to how we're doing that and what will stand in its stead to
ensure that we don't miss events that we need to know about; the system
actuations, which I'll get into, and that was a specific ACRS comment,
but we're proposing a list that will make things more consistent and, on
balance, a small reduction in the number of reports.
Invalid actuations -- most commenters object to any report
-- any reporting of invalid actuations, because invalid actuations
involve conditions -- pardon me -- do not involve conditions -- plant
conditions that require the actuation, like low reactor coolant system
pressure or something which would turn on the ECCS system.
So, they're for some other reason, usually a dropped jumper
or something, and so, we're going to reduce the burden of those reports
by a good bit by turning them into telephone calls rather than LERs, but
there is a reason, and when we get there, I'll explain it, why we still
The required initial reporting times are being relaxed to
greater or less degree depending on the reporting requirement.
The reporting of emergency conditions is, of course, not
being relaxed. That's still immediately after calling the state.
One of the things -- and it's not the principle comment that
the ACRS had, but I remember Dr. Powers wanting us to go back and look
at these times again.
We were going with a rather simple approach of everything's
in one hour or eight hours or 60 days, and we have done that, and we've
put in a few more shades of gray based on experience and the perceived
The reporting of historical problems -- we're excluding
reports of things that happened more than three years ago and no longer
exist, that haven't existed for the past three years, and that's not a
big problem, but it just eliminates some unnecessary work in searching
old logs and things like that.
Finally, the late surveillance test is the biggest example
of a reduction in reporting burden. It's going to get rid of about 200
LERs per year, and those are simply cases where a surveillance test was
performed late but the system passed anyway, and of course, that doesn't
have much significance, because the system was working all along, and
so, we're getting rid of those LERs.
So, with regard to outside the design basis, in the proposed
rule we have proposed to eliminate that requirement, and we described
how events that we need to know about, events that are significant,
would -- are still captured by these criteria, including this proposed
new one, and that is the one where we got a lot of comment.
Basically, the intent was just to try to make sure we didn't
throw out the baby with the bath-water, but the commenters essentially
were saying that we missed the boat and this would be vague and it would
require a lot of additional reporting. So, we've changed it
In the draft final rule, we're still removing the
requirement to report a condition outside the design basis of the plant,
because that requirement is vague or unclear in its application, and the
other side of the same coin to that is that it requires reporting of
events that are not very significant, depending on how you read the
So, the new criterion we've modified, and what it requires
now is reporting any event or condition that requires corrective action
for a single cause or condition in order to ensure the availability of
multiple trains or channels to perform the required safety function.
This is -- the idea here that we're trying to capture is
this would be an event -- it may not qualify as a common cause failure
-- that is, it may not make independent trains inoperable at the same
time, but it's getting close to it, and it's things like you've
discovered gummed-up solenoid valves due to some common cause and you
have to go and replace a bunch of them and clean out the air system,
that sort of thing, and that sort of thing is the kind of thing that the
NRC needs to consider taking some action to make sure it's addressed.
DR. BONACA: I have two questions.
One, you changed the definition of the new criterion from
December to this, and I don't understand what the intent of the change
And the second question I have -- some of the examples
provided -- to me, they would be reportable under Part 21 -- for
example, the, you know, stem of an MOV that is made of the wrong
material and therefore is subject to certain cracking -- or to other
reporting requirements anyway.
So, the question I'm asking is are you sure you're
fulfilling the objective of assuring that what is being reported under
some other means of reporting is not duplicated here?
MR. ALLISON: Yes, sir.
As to the Part 21, that stem might be reportable, maybe, by
a vendor, if it was discovered by the vendor, but it's -- certainly, it
wouldn't be reportable by a reactor licensee.
The threshold in Part 21 is very high. It's a major
reduction in the level of safety of the plant. That's what a
substantial safety hazard is, and it corresponds more or less with --
you pretty much have to have an abnormal occurrence.
DR. BONACA: But if I found a stem that is cracking and
that's because the material is old or stems in MOVs in other
applications, that's a substantial safety hazard in my mind.
MR. ALLISON: I don't believe that would be reported under
Part 21 as a rule.
DR. BONACA: Okay.
MR. ALLISON: I do remember a case, just to give you a quick
example, when we ran this through the Part 21 process and found that it
was not reportable.
It happened at McGuire, I think, a test of a spare scram
breaker, and it didn't work, because a plastic part was cracking, opened
up the other scram breakers at McGuire, several of them were cracking,
but they hadn't failed, went over to Catawba, same kind of breakers,
cracking, some of them maybe didn't work, not reportable under Part 21.
That's the kind of a threshold that Part 21 has.
Now, maybe that should have been, but -- and it was
reported, of course, under 50.72 and 73.
DR. BONACA: Okay.
So far as the two definitions, could you explain what the
logic was in changing the definition?
MR. ALLISON: Yes, sir.
The December package --
DR. BONACA: You have that at the bottom of page three of
MR. ALLISON: Okay.
Well, the commenters -- we have some problems with this
I would say the first one is a vague point where we say
"could reasonably be expected to apply to other similar components in
Now, the objective is the same here, of course, is to get
something that has some significant generic implications, but the
commenters said that, as soon as something fails, you know, in many,
many cases, they're going to end up in an argument with the inspectors,
then, about whether that same failure mechanism could reasonably be
expected to apply to everything else.
The other one is, of course, there was -- the word
"significant" is in there, "significantly degraded," and by that, we
meant on the verge of failure, not failed. So, we're talking about
substantially -- or greatly reduced margins, but that's hard to define
DR. BONACA: So, you went to this new criteria which you
have now at page four.
MR. ALLISON: Yes, sir, and I think this can be objective,
because it -- it can be a lot more objective, certainly, because it's
going to have to be a change -- first the licensee has to determine
corrective action is necessary, so we're not arguing about someone's
perception, it's a determination that will be made, and it's got to be
necessary for that reason, not for instance, to meet the EQ rule, but to
make the system perform its safety function, and so -- and you don't
have to review every failure, you only have to look at your corrective
actions programs, and the next slide, under the guidance, you see time
is allowed there.
Licensees are given time to decide whether the corrective
action is needed and what it's needed for.
DR. WALLIS: It seems to me there's some vagueness.
I mean if I have a valve which is supposed to open fully and
let in some emergency coolant or something and it turns out that valve
travel in some way is not 100 percent, so it opens 90 percent of the
way, it's just sort of iffy about whether this is significant or not.
MR. ALLISON: Well, the term of art that's in here is the
ability to perform the specified safety function, and that really means
operable, and that's a determination the licensee is going to have to
Operability is a determination the licensee has to make one
way or the other, and the NRC knows what this determination is. If we
disagree with it, we can raise it with the licensee. The inspectors
look at these things. But that's the definition of operability, able to
perform its specified safety function.
Now, something could be operable today but getting worse,
and you have to take corrective action. That would be reportable. But
if something is just operable for the -- will remain so indefinitely,
that would not be under this criterion.
MR. NEWBERRY: Dennis, while we're on that point, thinking
back to Dr. Bonaca's opening remark, there's many comments on that
proposed criteria, as you can well imagine, and it wasn't until recently
that the staff came to this proposal and let, you know, the different
things that the committee may oversee.
I think, in looking at it within the last few days, this is
going to be the first time that many people see this new criterion, but
we're approaching the final, you know, draft rule point.
So, our thought is -- and I think you're going to hear about
this later today -- that we really think it's in everyone's best
interest to have a public meeting, announce a public meeting on,
certainly, this part of the rule, I don't imagine others, but the intent
of the meeting would not be to negotiate a position -- I mean we're in a
rulemaking process here, but certainly for the staff to explain to
anyone who would be interested the rationale for the position, answer
questions of clarification on the position.
I think that would be reasonable to do before we go up to
MR. ALLISON: These are just some of the additional guidance
that you find in that Federal Register notice that I've sent you.
The principle one is that it is -- you screen what your
corrective action program comes up with instead of every failure, you
screen the corrective actions, and you have the time to do that. The
reporting clock doesn't start until you've made that decision.
DR. WALLIS: So, you can dilly-dally in making up your mind?
There ought to be some incentive to determine this, whether a corrective
action is needed or not, pretty quickly.
MR. ALLISON: Well, there is. We have guidance in Generic
Letter 91-18 that requires licensees to make operability determinations
on a time scale that's commensurate with the risk importance, the safety
significance of the issue, and so on, and so, they will make that
determination pretty quickly.
MR. SIEBER: I think the other aspect of that is tech specs,
typically, for systems important to safety, will force you to correct a
non-conforming or inoperable condition within a certain amount of time.
So, that forces the clock to start on LER issuance, correct?
MR. ALLISON: Yes. That's right. But you can't really tell
whether something is truly reportable under this criterion until you
decide what the corrective action is.
MR. SIEBER: That's correct.
MR. ALLISON: That was my presentation on this criterion.
The next one is of lesser importance, but it was the number
two issue, I guess.
DR. BONACA: And we will hear the industry's perspective
MR. ALLISON: Yes, sir.
The number two is system actuation, and in the proposed
rule, we proposed a list of systems and so on.
The ACRS, among others -- well, the industry opposed the
list of systems. They wanted to use the list that's in their FSAR,
which varies from plant to plant. The ACRS commented that this list
shouldn't be in the rule but should be developed.
In the final rule, what we're saying is to go ahead and
impose the list.
Now, the list has been changed in response to specific
So, we've gotten rid of some things that the industry
pointed out didn't really need to be on the list or weren't appropriate,
but this will be, on balance, a small net reduction in reporting, it
will be consistent, and one of the things with regard to the ACRS
recommendation -- the industry commenters said I don't think we're
really to the point where we have good criteria developed that we can
develop a plant-specific list of systems.
Now, that's supposed to come in the future, in the
risk-informing of Part 50, but it's not here right now, so why don't we
do it then? That was their idea, and we basically agreed with it.
So, rather than try to solve the problem of how to define
risk significance in terms of systems in the context of this rule, we're
putting it off, but the things that are on that list, I would say, are
always risk significant. We don't have things on that list that are
going to be insignificant at any plants.
DR. UHRIG: I have a question on that, however. Is there
inconsistency in some FSAR's between the list that you have in this rule
and what they call --
MR. ALLISON: -- ESF's.
DR. UHRIG: There is inconsistency in the FSAR's. One of
the issues was that some of the FSAR's would not recognize, for example,
auxiliary feedwater or emergency power as one of the systems.
MR. ALLISON: That's correct, yes.
DR. UHRIG: And you were trying to resolve that issue.
MR. ALLISON: Some plants don't classify auxiliary
feedwater, for instance, as an ESF. So, they wouldn't be bound to
report it as long as they're using the list in their FSAR.
DR. UHRIG: But with this change in this rule, they would be
bound to report it.
MR. ALLISON: Yes, they would, and so, that would lead to a
few more reports here and there, but the list also eliminates some
reports, about twice as many as it adds, but both of them are small
DR. UHRIG: Does it represent, this change, a backfit in the
MR. ALLISON: I'm sorry.
DR. UHRIG: Does it represent some change also in their
MR. ALLISON: No, it doesn't, because we're not -- this
change does not say these systems are ESF's.
DR. UHRIG: Okay.
MR. ALLISON: It says report the actuation of the following,
and the numbers are small. I think we would require about eight reports
a year that wouldn't be made under the current regimen, but we'll
eliminate about 16.
The next point is invalid system actuations. In the
proposed rule, we recognized that there was no need to pick up the phone
and call us in four hours or eight hours about these, because the plant
conditions that require actuation aren't there in this case.
Licensees objected to any reporting, but we -- and this was
an issue that had been gone through at the advanced notice of proposed
rulemaking stage, as well.
We need those for reliability estimates and things like that
to help us to move towards risk-informed regulation, and in fact, we had
some years ago proposed a data rule to get that information, and the
industry proposed a voluntary alternative, and we accepted it, and one
of the bases for accepting the voluntary alternative was having these
So, in the final rule, what we've done is we're keeping the
reports, but we're changing them to a 60-day phone call under 50.73, and
in the guidance, we specify just what needs to be in the call. It's not
a lot of information, but we have to specify it.
This reduces the burden drastically for those events that
are only spurious actuations, and those are not going to be considered
LER's. The guidance will state this is not considered an LER, but it's
like a factor of 50 reduction in burden for a given event, and this is
maybe 60 events a year.
DR. WALLIS: Are there no spurious actuations which actually
compromise the system's operation later on?
MR. ALLISON: I can't think of any. I mean you could have a
spurious actuation where the system fails to work, that reveals a
failure of some kind, but I can't think of spurious actuations that
really create problems other than possible failures.
DR. WALLIS: Unless it put a plant through a transient that
did some damage.
MR. ALLISON: Well, that will certainly be reportable,
though. If you get a transient, you'll have valid actuations occurring.
The next thing was required initial reporting times, and
rather than the one-hour, eight-hour, and 60-day approach that was in
the proposed rule, in the final rule we're saying one hour and four
hours some events that are of a little more urgency.
One of them is press releases, because -- and the reason for
that report is not the urgency of taking action but it's in responding
to public concern.
The other one is unplanned transients, like valid ECCS
injections, shutdowns required by the technical specifications, and so
on, and then eight-hour reporting for other events under 50.72.
We're also deleting three redundant criteria from 50.72.
Those are actual threats and radiation releases, and the reason is that
those are captured -- under 72, they're captured by other criteria.
DR. WALLIS: What is the concern about reporting to other
MR. ALLISON: Well, the -- that's to respond -- again, going
back to the objectives -- respond to heightened public concern.
If the state gets a report and if they're concerned about it
and they want to call the NRC, we want to know about the event.
DR. WALLIS: It could really be generalized to a plant
notification of any other party.
MR. ALLISON: It could be, but it's -- there is a
difference. That is, if they notify a consultant or their board of
directors, that's not required under the rule. It's only another
government agency or a press release.
Nobody's complained that we need to generalize it further.
Historical problems -- in the proposed rule, we recommended
limiting these reports for just two specific types of events.
In the draft final rule, we're expanding it to all events
reportable under 50.72 and 50.73. That was actually -- I guess that
suggestion really came from the Commissioners in the SRM on the proposed
rule, and we asked for comments specifically, and everybody supported
expanding it to all kinds.
The final change in my list of principle changes is late
surveillance tests, and I discussed that with the first slide. These
events don't involve an impact on the ability to perform a safety
function, and therefore, they're not very important to us.
My last slide is the schedule, and we're going to brief the
CRGR next week, and we're due to provide this package to the
Commissioners on the 10th of March, which means to the EDO a week before
that, and so on.
So, we're getting close to the date, and we're going to have
to hold the meeting that Scott mentioned a minute ago sometime within
the next month.
MR. SIEBER: I guess -- and I want to pick on a specific
phrase that you used, but I've heard it over and over again when we talk
about risk-informed regulation and enforcement and so forth. The phrase
is, well, this is not very important.
To me, that has a bad connotation to people who work in
power plants, and maybe the plant manager, the vice president, or SRO's
can make that differentiation, but everybody else says, well, this isn't
very important and so my attention need not be as high at performing
surveillance tests on time or doing any other thing on time, since it's
not very important, and it would be better if we could use another
phrase than that, because I think it puts a negative motivation into
power plants and workers.
MR. ALLISON: I agree.
MR. SIEBER: All right.
MR. ALLISON: It was a bad term to use.
DR. SEALE: There's, if you will, almost an industry that's
grown up within the Commission and within other groups that are
concerned with the operation of power plants, and that is that group of
people who essentially mine such reports to extract from them useful
data on causes, consequences, remedial interventions, and so on, the
kind of thing that AEOD did in the old days, the kind of things that the
people in INPO and WANO do in their independent realms on events.
In modifying these reporting requirements, have you checked
with those people to be sure that you haven't reduced the usefulness of
these data for the people who are using it with the greatest
MR. ALLISON: Well, we've coordinated with the NRC staff
organizations, and our assertion is that we're not eliminating any
reports that we need, and we asked the public specifically in the
Federal Register notice, if you can identify an example of something
that's needed that would be eliminated, please tell us, and of course,
none were, and so, I think, yes, we've coordinated with everybody.
DR. SEALE: Perhaps I'll want to ask the person from NEI
later whether or not they've inquired in a similar vein with the people
DR. BONACA: Any other questions.
DR. BONACA: I think we'd like to thank you for the
MR. ALLISON: Thank you.
DR. BONACA: We'd like to hear from NEI.
MR. DAVIS: Good morning. Jim Davis, Director of Operations
at Nuclear Energy Institute.
Looks like I have the unenviable position of being between
you and lunch.
I've got a number of slides here, but I've only got two
points to make.
DR. BONACA: Take your time. I really want to hear about
MR. DAVIS: One, when we briefed you last year, in March,
one of the things we said, we thought the rulemaking process embarked on
in this area was very good. We got the ANPR that laid it out in some
We got an opportunity to interact in that arena, and
throughout the process, there were a number of interactions between the
staff, the regional examiners and inspectors that have to enforce this,
and the operators that have to make it work at the plant, and through a
bunch of workshops, tabletop exercises and so forth, there was a lot of
effort put on solving this problem that we've gone through for the last
eight years, so everybody understood exactly what the requirements were
and what the rule said.
In many cases, we found that the intent was clear. We all
knew what we wanted to do, but the perspective from the three visions
didn't quite fit, and there was a lot of time and attention put on that
particular aspect of it, and we told you last year that we were very
satisfied with the process.
Then I come to the point that I will tell you -- and in that
process -- I'm sorry, I moved my slides around -- operability is a key
aspect of it.
At every meeting, operability, operability determinations,
and how we do those are the things that we all understand, and as you
see, we move very quickly to a process of how do you figure out whether
a report is required?
Operability is a key issue. We do the operability
determination. It's a very clear process. It's an expectable
processing. It does involve some risk insights, where it's appropriate.
That's a key to the entire business.
My second point: The draft rule comes out, and as far as
the industry is concerned, the rule should not have gone forward. We
could not support the rule as written. It didn't meet the three
criteria that the staff had put out, it was not clear, it did not reduce
any burden, and the industry was ready to go to the Commission to say
don't put this rule, we'll solve the problem in harmonizing Part 50.
You already know what the problem is. It's 50.73(a)(2)(ii)(C), the
reporting of degraded components. It's related to my first point.
This showed up for the first time in the final Federal
Register notice that came out for public comment. It did not go through
the process that all the other elements in this rule went through.
I have no comments on the list of what systems will be
reported on. We went through the process. We made our comments. We
gave the staff our best input, and they've got to make a decision,
because that's what rulemaking does. This particular element didn't go
through the process.
To address your issue, sir, we don't think that this is a
data collection rule. We've been through the data collection rule.
We've been through the discussion. There are opportunities for the
staff to get the data they need from other arenas.
INPO's database has been made available to the staff and
they're working in that area, and we are really concerned that, in one
case, we say we no longer require the reporting of design basis events
and turn right around in the Federal Register notice, we point out
design basis -- the purpose of this particular section, this data
collection element, was to ensure we continue to collect design basis
information, so we clearly didn't meet the requirements.
I will tell you -- I'll skip a slide. The examples in NUREG
1022 made no sense. It was a very important part of the process that we
bring the implementing NUREG along at the same time we were developing
So, we had the rule, we had the NUREG, we could look at them
both simultaneously, and when this came out and we looked at the
examples, we could make no sense of the examples in the NUREG, and the
further we've gotten into it since the workshop we had last year and the
closure of the public comment, the more confused we've gotten, and we
put forth a big effort to make absolutely sure the staff understood
where we were in that particular arena.
DR. WALLIS: Did this get resolved?
MR. DAVIS: I'll get to that in just a second.
DR. WALLIS: Okay.
MR. DAVIS: If you removed -- I want to make sure I make
If you remove that one small section on reporting degraded
components, we feel that the draft rule does improve the clarity of
reporting in all other areas, does provide a clear focus and a nexus to
safety, and I think that's one of the things that we were trying to
achieve using the operability determination process.
We have one we can understand, one the inspectors can
understand, and we think one that provides the headquarters and the
people upstairs and the operations center the information they need to
make timely decisions, would eliminate the unnecessary reports that
don't help anybody, and would be a great conclusion to eight years'
worth of effort.
Coming into the exercise, what were our recommendations?
If you want to go forward to the rule, eliminate the
degraded component reporting or separate it out and do the backfit
analysis that we think would be required to support that level of
reporting, or if we can't come to agreement in that area, just let's
stop the whole process and let's harmonize this rule as we go through
the Part 50 process.
Looking at what was proposed in the briefing today, it
obviously moves in the right direction.
It gets us back to a discussion of operability and what's in
that area, and our intent is to reinforce what we just heard, as we will
provide a request to the staff that they, one, give us the language and
the examples for NUREG 1022 in advance and that we have a workshop and
follow the same process that we followed on the other pieces of it for
this narrow thing.
You know, I don't want to open the whole rule again, or
we'll spend eons arguing, but we've had a significant enough change and
this is an important enough issue that we need to get it right, and I
think we need to have an opportunity for the industry, other
stakeholders, and the regional people to look at it, discuss it, and
make absolutely sure we understand what the words mean, get the right
words in there in that particular area, and also ensure that we've got
the right examples in NUREG 1022 as we go forward on this, and presuming
that's about to occur, I think we'll have achieved our purpose, but the
nexus is process.
We had a good process, and the one piece of the rule that
becomes the major contention is the one that didn't follow the -- didn't
run through that process. So, I emphasize that, because it's sort of a
more global issue there.
DR. KRESS: Your objection to that part of the rule, I
gather, is more than just it didn't go through the process.
MR. DAVIS: That's absolutely correct.
DR. KRESS: You say it would increase burden significantly.
MR. DAVIS: Yes. One plant looked at it over a period of
nine months, and it would have required them to evaluate a significant
number of items in their plant. It didn't generate a report for every
one of those, but every time you have a component with an abnormality,
you suddenly have to go through this evaluation of if and whether
reasonable could, significantly, and all these other vague words to try
to come up with a engineering determination of whether it fits in that
DR. SEALE: In other words, this is reporting of degraded
but operable components.
DR. BONACA: You seem to have made a distinction there, at
the beginning, in your second overhead, or third, regarding operability.
So, are you saying that the operability determination
process is sufficient to deal with the significant issues on degraded
components without the necessity of reporting? Are you saying that?
MR. DAVIS: Let me answer it this way. We found that
operability determinations that are required in the rest of this revised
rule work. The words that I look at appear to tie this to the same
operability to process for the component that we're looking at. It is
in the operability. It impacts the operability of the system we're
If that is truly what we're saying, I suspect that will go a
long way to solve the problem. That's why we'd like to make sure -- you
know, have the discussion to make sure that's what we really mean in
DR. BONACA: The question I have for the staff is, is this
an event or condition of a single cause? Is this a component which is
operable but degraded? I would like to understand --
MR. ALLISON: It could be.
DR. BONACA: -- how the issue of operability addresses this
or doesn't address this.
MR. ALLISON: It could be, but it would have to be something
that pointed out to the licensee that he has to take corrective action
on multiple trains to ensure that they remain operable. So, that would
be -- so, it could be degraded but operable, but it has to fulfill those
DR. BONACA: It seems to me that there hasn't been
sufficient communication of this issue, and you're talking about, in
fact, a public workshop or something, maybe, under which that could be
MR. ALLISON: Yes. Mr. Davis is in a bad position as far as
commenting on this criterion, since he's just seeing it, but as he said,
it goes a ways towards resolving the comments, and we will schedule a
meeting between now and when we send this paper to the Commission.
MR. NEWBERRY: I'd like to offer another comment. This is a
good discussion a very difficult issue. I can think of a number of
I guess it should be no surprise on the inconsistent views
given the term "design basis," which we're working on to clarify in
another area we've talked about with the committee, but one of the
points here I'd like to emphasize is the inclusion of the term -- the
notion of corrective action.
When I talked to folks in industry or where they came up to
me at every opportunity in the last few months on this issue, said, you
know, we have a process, an Appendix B process, we have a corrective
action program at the facility to handle these issues.
If a degraded condition is identified, we put it into our
process, we evaluate it for operability, we evaluate it also for the
need to take corrective action under Appendix B, and so, it was that
line of commenting and thought that led us to this criterion, to say,
well, we inspect that program, we oversee that program, do we need a
report for all the data that goes into the program?
We concluded no, but when the evaluation is completed and
the utility determines that action is necessary at that plant that could
also occur at another plant, we said, okay -- we looked at the
objectives of the rule. We said, okay, we should have a report for
Now, maybe there are some areas there we would need to
explore further and get some dialogue going with, you know, the
industry, but that was the thought process, was to try to credit
further, as we are in other areas, the programs at the plant.
DR. BONACA: Thank you. Now I understand why it got in
there. All right. I didn't understand it before. I understand it.
MR. SIEBER: On the other hand, the staff has been aware of
NEI's position on this, I presume?
MR. DAVIS: Yes. I mean we were -- the comments were very
clear and very detailed on this. There's no question that the staff
understood exactly where the industry stood and why we had difficulties
with the wording that was in there the first time around.
MR. NEWBERRY: Yes. It was clear we needed to rethink
totally what we had proposed, and that's why it's no longer being
proposed. We came up with the new criteria which Mr. Davis is saying he
thinks is headed in the right direction but we need to talk about
DR. BONACA: My feeling is that we are not ready to write a
letter on this. I mean clearly this is an open issue, in my mind.
Even if the staff has resolved that they want to proceed
with this to the degree to which you're going to have a public meeting
in which there is going to be exchange of information, things may
I would like to have your comments on that, Jack.
MR. SIEBER: Well, I just want to agree with you that, until
the staff resolves this one way or another and takes a position, I don't
think there's anything that we can do to endorse or not endorse where
the staff is at this point in time.
DR. BONACA: It is going to be, you know, a burden.
Clearly, you know how much time is being spent on operability
determinations. I mean it's very time-consuming, and this is going to
So, there has to be a real buying-in from the stakeholders
that this is a necessary thing to do, and communication is important.
MR. DAVIS: I must also admit that we would like to see the
rule change completed in a timely manner. There are other parts of it
that have some benefit to the industry.
So, if we can -- you know, if closure on this one issue can
be achieved quickly, we would support moving forward with the rest of
Even though I haven't seen the rest of it, you know, I got
some insights into it. You've got to have some faith in the process and
the opportunity to share information, that that information will be
used, and so, we're really focused -- I mean this is really a very
I don't want to open the whole rule and go through the whole
process again. I'm just narrowly focused on this one issue that I think
needs some additional thought on our part.
MR. NEWBERRY: Mr. Chairman, I would propose that, you know,
consistent with Mr. Davis' comment on the need for dialogue, we talk
with them and then talk to your staff about a process that we could use
to satisfy the objective of timely implementation of the rule but also
get the committee the information that they would need to inform them so
that you could write a letter on a timeframe to support the rulemaking.
So, we'll take that as an action and get back to you.
DR. BONACA: We will support you promptly, but I think that,
at this stage, with this issue open, that's a major comment that we need
to address, and we really can't right now.
DR. SEALE: That's a reasonable position, yes.
DR. BONACA: With that, I thank you for the presentation.
MR. DAVIS: Thank you very much for the opportunity.
DR. BONACA: Mr. Chairman?
DR. APOSTOLAKIS: Thank you, Mario.
Recess until 10 after one.
[Whereupon, at 12:10 p.m., the meeting was recessed, to
reconvene at 1:10 p.m., this same day.]. A F T E R N O O N S E S S I O N
CHAIRMAN POWERS: Let's come back into session. I think we
are going to discuss a new and different topic that we are relatively
unfamiliar with --
CHAIRMAN POWERS: -- and so I expect all members to pay
close and keen attention as I ask -- Jack, you are going to help us
explore this untrammeled territory?
MR. SIEBER: Yes, sir.
CHAIRMAN POWERS: Okay.
MR. SIEBER: The purpose of this afternoon's session is to
hear a briefing by the NRC Staff and the Nuclear Energy Institute and
hold discussions with them regarding the status of a proposed Regulatory
Guide which, if it is issued, will endorse the guidance of NEI 96-07,
associated with the implementation of the revised 10 CFR 50.59 process
-- 10 CFR 50.59 is a keystone regulation probably used more extensively
by licensees than any other, and it allows under certain controlled
circumstances changes in the plant and also tests and experiments.
The current version 50.59, in force, has been in force for
about 30 years. During this last summer there was a new rule issues
which changes the three criteria in the old 50.59 to eight criteria,
clarifies a number of aspects of the process and was issued as a final
The implementation of the final rule occurs 90 days after
the associated Regulatory Guide is issued and that has not been issued.
It was contemplated by the Staff that the Regulatory Guide would endorse
NEI 96-07 and all of us have received a copy of that and I am sure
reviewed it. At the time I reviewed it there were, it seemed to me, 24
outstanding items based on the matrix that was sent along with it.
I understand also and have received a copy of a final draft
of 96-07, which I got this morning and the Staff got Monday in spite of
the snow and it seems to me that the Regulatory Process is such that a
final determination as to the acceptability of the changes and the
resolution the remaining open items couldn't be done between last Monday
So what we will hear about today is a status report on the
issues of the Regulatory Guide and NEI 97-07. It would be good if we
could talk a little bit about the outstanding items and those which
remain outstanding which, by my count, should be six, unless others have
developed in the meantime -- if we could hear a little bit about that
and what the problems are.
In addition to that presentation, I expect the
representatives from NEI to avail themselves of time during this
What I would like to do now is introduce Eileen McKenna, who
is responsible for this presentation. Eileen?
MS. McKENNA: Okay, thank you very much, members of the
committee. I might also mention Scott Newberry, our Deputy Division
Director, is here at the table for any questions, and I was going to
suggest if the committee had no objection that Mr. Bell sit at the table
here, so we can discuss this. I think it may be helpful to turn it over
at a certain point and then turn back over, to kind of give an idea of
what the discussions have been between us.
I think in terms of my first slide, your introduction pretty
much covered the information I have there that indicated the rule was
issued in October, the 90-day timeframe after the approval of the
guidance for implementation of the rule, and that we are looking to try
to endorse an industry document through a Reg Guide as the regulatory
guidance for the rulemaking.
I want to make a couple comments about why this is a status
briefing rather than coming to you with the draft Reg Guide. I think
our original plan and schedule would have called for us to be ready to
give you a draft Reg Guide to present at this meeting and get a letter,
but because of some of the open issues that you alluded to, we are not
quite to that point in time and I will come back to the question of
where we are on schedule and some of the reasons therefore in a little
while, but that is why it is status report to discuss the distance we
have come but some of the issues that are still remaining, and then
where we expect to be going to.
CHAIRMAN POWERS: I am curious. Is there some absolute need
that you need a letter from us?
MS. McKENNA: Well, I think there are some options for the
committee. You know, we are going out for a draft Reg Guide and then we
will have a public comment period, to be followed by a final Reg Guide.
This is an item of high Commission interest, so we do have Commission
due dates and attention and we are working with that.
I think our view at this point is that we would not be ready
to come to the committee in March, because since that is only a month
away, we would have to have our final, have our Reg Guide put together
in about two weeks, and I don't see that happening, so obviously the
next window after that is April, which is perhaps a little later than we
had hoped to publish in order to meet some of our other objections on
the schedule but we are trying to work with that.
CHAIRMAN POWERS: Well, let me ask you this. If you have
got outstanding items with NEI do you think the ACRS can help resolve
those issues? I mean I am not asking you to overestimate or
underestimate. I am asking for a prognostication on your part about
your abilities. Do you think you'll get it all sorted out and be happy
MS. McKENNA: Well, I think we are going to come to a
resolution. Obviously we may or may not be able to agree with what they
propose and they may not agree with where we come out on some of these
things, and if that is the case then what we may have to do is have a
Reg Guide that at least at the draft stage has some clarifications or
exceptions, if that is where it comes out.
Some of the issues I think the committee may be able to help
us on. There are a few that we are wrestling with that are perhaps more
in the "how do they fit with the regulations and the process" and the
committee may feel less comfortable providing their input in that area,
but we will try to cover what the different ones are.
CHAIRMAN POWERS: I think we need to think carefully about
the value added at this late stage of the process where we are really
working on implementation, I am wondering if it is really not necessary
to have them get a blessing, because that is all we would be doing is
just passing judgment over something that we have seen many times and
they are down into the implementation stage and I am not sure -- I think
we may need to use our judgment about whether we -- we all know and love
Eileen, but she is free to visit us without 50.59.
MS. McKENNA: I hope so.
MS. McKENNA: And as I said, this is the draft stage and we
would have a final stage and there would be another opportunity at that
point for the committee to revisit and after we have the benefit of
comments. I think that is an option that the committee may want to
MR. SIEBER: Yes. Actually, the SRM that controls your due
dates says May, 2000.
MS. McKENNA: Yes, we have approached the Commission about
an extension on that, because obviously if we are here in February and
don't have a draft Reg Guide, we are not going to be at the Commission
in May with a final, and we have made an approach to move that date out
a few months in recognition of those facts, yes, but I think anything
the committee can do to help us with the schedule we would appreciate.
MR. SIEBER: I think that we can certainly try but I don't
think there is value added in rushing through something and perhaps
missing it because once the final document is issues it stays there for
a long time before anybody has an opportunity to change it.
MS. McKENNA: Yes.
MR. SIEBER: So we ought to get it right the first time.
MS. McKENNA: Okay. I thought at this point might be a good
opportunity to ask Mr. Bell to talk a little bit about the development
of 96-07, since what we are trying to do is endorse that document, so I
felt this might be a good opportunity to let me make some presentation
and then I will come back and talk about where we see the status on some
of these issues.
MR. SIEBER: Did you want to do these first?
MS. McKENNA: Well, I suspect we have some overlap in our
slides since we really didn't try to coordinate in any detail. I think
I have covered some of these bullets. We have had some draft
interactions and I can go into this in more detail.
The document that you mentioned, presented on January 18th
for our consideration, we are in the process, there are some questions
and issues we have, and we are in the process of getting a letter back
out that we hope to get out this week, but it is not yet out of what
some of those remaining issues that we are still working on. We have
scheduled a meeting next week to talk about what those issues are and
what we are going to do about them, and, as you may be aware, there is a
Commission briefing scheduled on February 29th on this subject and we
will be probably covering very similar territory with the Commission, as
you will be hearing.
It's rather than where we hope to be, perhaps, in terms of
February. At this point we are looking at perhaps a two month -- it
might be a little shorter if we don't come back for a letter on the Reg
Guide. We may be able to shorten that by a couple of weeks, but it is
that kind of timeframe we see at this point for publishing the Reg
MS. McKENNA: I'll move over to the other side, how's that?
MR. BELL: Thank you and good afternoon. I am Russell Bell,
with the Nuclear Energy Institute. I am the Project Manager on the
50.59 issue. I had a nice cover slide. I think everybody has my copy
of these. I appreciate somebody -- Dr. Seale, was it you? -- who likes
"what's past is prologue"?
DR. SEALE: Yes.
MR. BELL: I had a draft slide that just said "Background"
up there and I thought that this might be an audience that might
appreciate something else.
DR. SEALE: Quicker than most.
MR. BELL: There's certainly an overlap with some of the
things Eileen just said, but just suffice to say this has been a story
in the making for some time, probably the seed were laid for where we
are today back in 1989, when the industry produced the first guideline
document on 50.59, and the consciously or unconsciously a decision was
made not to go the extra mile and get a Reg Guide, you know, NRC
endorsement of the thing.
The rest of these events here that are kind of captured are
almost predictable based on that early direction chosen, so here we are
The most recent thrust/assault at this issue began I guess
three years ago now, say, and we wrote this objective and I thought I
would trot it back out because I think it still holds, so this was an
observed, very extensively used regulation where there is
misunderstanding about its requirements and expectations, there is
regulatory instability, and we certainly experienced that, so we set out
to resolve that.
Indeed, the rulemaking, which was completed last summer but
is not yet effective, went a long way towards resolving the regulatory
instability. It removed the so-called "zero standard" that was
reflected in the original rule. It established key definitions where
there were really no commonly understood definitions before. The margin
of safety criterion was somewhat problematic and that has been replaced,
we think improved. So these kinds of things were accomplished, have
been accomplished already in the rule, and what is left to us, and it is
not necessarily the easy part, is to then translate that into
CHAIRMAN POWERS: Well, you have done a lot in that
direction, but I can't help but think a little bit about our initial
discussions of what to do with 50.59 that came about, I think, because
our ability to quantify some of the questions in the original 50.59 has
just improved so much over the years that what in the past was
indistinguishable from zero suddenly became distinguishable from zero.
We talked about, gee, let's think about doing a
risk-informed or maybe even a risk-based 50.59, but in the interim we
have to do something and get this out of the way quickly and the
operative phrase was "quickly" but now I want to turn to the
Having gone through this, do you and your fellows within the
nuclear industry see advantages to now launching forth on a
MR. BELL: The Staff may be able to update farther, but my
understanding is that we are proposing certain things as a part of
another issue that I think the committee will hear about in this
meeting, the risk-informing of Part 50; 50.59 is certainly a part of
that, so I think the answer is yes, we still see a benefit.
DR. WALLIS: I noticed a lack of enthusiasm. I was
expecting you to come back and say "Yes!"
DR. APOSTOLAKIS: Russ doesn't do things like that.
CHAIRMAN POWERS: You are looking at battle-scarred veterans
DR. APOSTOLAKIS: Well, I think though we have to make clear
what we mean by risk-informing 50.59, because I think a lot of people
think that the objectives would be the same. You would just be using
risk information, and at least some of us are thinking about it in a
different way. Perhaps the benefits of risk-informing 50.59 or the
process of allowing changes without review are not very clear to a lot
of the industry.
If I told you right now that I was advocating a 50.59-like
process that would have as a sole criterion that the core damage
frequency doesn't go about 10 to the minus four, would you say yes, the
way Dr. Wallis wants it?
I would allow you to do anything you want except exceed 10
to the minus four core damage frequency.
That is an extreme, of course, but, you know, the benefits
of a new process have not been articulated very well.
MR. BELL: The other way to come at that is to somehow risk
inform the scope of matters that 50.59 would be applied to. I believe
they are looking at both approaches in terms of out to improve things.
DR. APOSTOLAKIS: But coming back to the issue of quick fix
that Dr. Powers mentioned, is this quick?
MR. BELL: I have seen things quicker and things take
DR. APOSTOLAKIS: This is pretty good?
DR. BONACA: No.
DR. APOSTOLAKIS: Are we going to go into details of this?
DR. SEALE: A blink of El Nino's eye.
DR. APOSTOLAKIS: Because I have two questions I want to
MR. BELL: Well, I might identify -- I guess my purpose is
to, in the middle of Eileen's status report, just to provide you some
context by going through providing an outline of the document and some
of its key aspects. I would try to do that as quickly as possible,
although we have a day and a half workshop devoted to this document
planned in April and so it is quite a challenge to cover that material
in just a few minutes, so I am willing to try though to again provide
DR. APOSTOLAKIS: Can I ask my questions now? I like
plant-specific, document-specific questions.
I noticed that in 96-07 there is actually a quantitative
criterion for the increase in frequency of occurrence of an accident
which is not the way I understand the document to be used only when one
chooses to use a PRA.
This is on page 39 of the document. It says --
MR. BELL: Section 4.3.1 --
DR. APOSTOLAKIS: It's in the book.
MR. BELL: That is the old one.
DR. APOSTOLAKIS: What I have is the old?
MR. BELL: Yes.
MS. McKENNA: I don't think that page changed very much, so
it should be about the same place.
MR. SIEBER: It might not be the right page.
MS. McKENNA: That's possible.
DR. SEALE: What is the section number?
MS. McKENNA: The section is 4.3.1.
DR. APOSTOLAKIS: You have a different one?
MR. BELL: Yes, this is the latest and greatest.
DR. APOSTOLAKIS: Well, I am going with what we have in the
book. So it is 4.3.1 -- so it says, "If the proposed activity affects
the overall system performance in a manner that could cause an accident
previously evaluated to shift to the higher frequency category or result
in a calculated frequency increase to be 10 percent or greater, then the
proposed activity would be more than minimally increased."
Now "or result in a calculated frequency" -- now I can
choose not to calculate the frequency, the change in the frequency?
MS. McKENNA: Yes. I think you phrased it a little
differently than I might have phrased it in terms of the usage of this
criteria, that the criteria is trying to cover both the cases where a
licensee chooses to do a qualitative assessment of a particular change
against this criterion, and also the cases where a licensee chooses to
do some kind of quantitative assessment, whether that is PRA or it is
some other way of approaching it but with some kind of quantification
involved, and that this part of the guidance would apply where that kind
of quantification, numerical usage, comes into play.
DR. APOSTOLAKIS: The distinction is made much more clear
later on on the next section, 4.3.2, where you talk about the equipment
malfunction, where there is a list of eight levels of performance, and
then with boldface letters it says, "Number 8. For use where the change
in likelihood of a malfunction is calculated."
The distinction is much clearer here, whereas there it is
buried in that "or" --
MR. BELL: Clearer that it is optional.
DR. APOSTOLAKIS: Yes, clearer that it is optional that you
don't have to do it.
MR. BELL: Sounds like a --
DR. APOSTOLAKIS: I think it is --
MR. BELL: -- fair comment.
DR. APOSTOLAKIS: -- in the frequency as well, but here,
now, I have a problem with this paragraph. Essentially if you read it
MR. BELL: Which paragraph?
DR. APOSTOLAKIS: Eight.
MR. BELL: Okay.
DR. APOSTOLAKIS: Number 8, C-8.
MR. BELL: I think before you -- I think we addressed that
problem. In fact, that paragraph has basically gone away. We call that
an elegant solution, when you just eliminate things.
DR. APOSTOLAKIS: The intent of this, though --
MR. BELL: Yes --
DR. APOSTOLAKIS: -- was really use this criterion only if
you are sure that you will be below a factor of two, because if you are
above, it is inconclusive. You can still use qualitative arguments to
argue, which seems to me like a cyclical argument because in order for
me to conclude that the change in the probability is greater by more
than a factor of two -- that the change in the likelihood of an
occurrence of malfunction is increased by more than a factor of two I
will have to use qualitative arguments and engineering judgment, so how
then after I conclude it is three I can use qualitative arguments and
engineering judgment to knock it down?
MR. BELL: In fact, the Staff identified that to us.
DR. APOSTOLAKIS: That is why it is eliminated.
MR. BELL: On further thought, that whole thought has been
eliminated. What you have there I think is a December revision.
DR. APOSTOLAKIS: December 20th.
MR. BELL: The latest one is January 18th.
DR. APOSTOLAKIS: So there is no paragraph like that?
Absolutely nothing? A factor of two?
MS. McKENNA: It just has -- there is a CH that says
"increasing the likelihood of a malfunction" -- excuse me, "of
occurrence of malfunction by more than a factor of two."
MR. BELL: And then the note equivalent to the bold --
MR. BARTON: There is a footnote, George.
DR. APOSTOLAKIS: I see the footnote. Okay. So the
self-consistency of this paragraph then goes away.
Now is there any way we can avoid presenting this kind of
criterion or analysis as an additional analysis? In other words, it
looks like, if you read it now, that one would still have to satisfy one
through seven and then as an addition to eight, because perhaps it is
easier to show that it doesn't increase by more than two, but if one
goes through the expense and effort of quantifying these probabilities,
shouldn't that person get some relief from the other requirements?
MR. BELL: In fact, this list of A, B, C, and then there are
subconsiderations under each, are a list of considerations, and in fact
the intent would not be that you have to check all those off.
In fact, many may not apply to a particular activity that
you are trying to evaluate, but it does represent a list of things that
you ought to consider to the extent they are applicable, and that goes
for the last one as well, the one you are talking about, in the case
where you are practically able to or able as a practical matter to
DR. APOSTOLAKIS: I guess my -- not objection, really, but
something that does not excite me too much is this idea that Number 8 is
in addition, that if someone really spends money to do a PRA, then, you
know, still have to do the other stuff. Reducing system redundancy,
diversity or independence -- I mean I can argue qualitatively now that
there is a minimal change or I can go ahead and quantify and then get
Now this factor of two or greater refers to the mean value
of the frequency of failure? Because there is a distribution there. I
have a PRA, I get a distribution. I don't get a number, so it refers to
which number, the mean?
MR. BELL: Do you recall if that was a mean number from --
we took that number from --
MS. McKENNA: I don't think it was specified in the other
DR. APOSTOLAKIS: Because doubling the mean is not really a
minimal change. Most likely what you are going to see is a distortion
of the shape of the epistemic distribution of the failure rate, but the
mean will not jump up by a factor of two. That would have to be a
significant change, so maybe you can eliminate the sentence that
survived under 8 and don't say anything, because I don't think that
sufficient thinking has gone into this, what it means -- unless you want
to do that.
See, if I have a distribution, to move the mean up by a
factor of two is not -- you have to do something significant on the high
You were thinking probably in terms of point estimates,
which are not well defined anyway.
MR. BELL: I think you are probably right.
Well, I think that is a point well taken.
You know, by the way, comments such as that or others the
ACES has -- that subject came up earlier -- to the extent they are known
at the same time that the public comment period is taking place, I would
think that would be the timely way to capture some of that.
DR. APOSTOLAKIS: Let me put the question a different way.
I realize that a lot of these changes are difficult to evaluate
quantitatively, because either the equipment does not appear in the PRA
at all, which is very common, or the change is of such a nature that you
say, my god, how does that affect anything?
Would it be useful to make a distinction between components
that are in the PRA and components that are not and reserve all this
qualitative discussion for the ones that are not, but for the ones that
are in the PRA you must look at the distribution of the failure rate.
It is not an option -- because you are going to get that question
anyway. It's similar to this thing that you will have a two-tier
regulatory system, one risk-informed and one, the traditional system,
which I think is an illusion, because you are going to get the question
of what happens to the core damage frequency anyway.
So if it is in the PRA, please provide arguments. The
arguments can be qualitative, but look at the distribution and tell me
what you think happened. That would make it cleaner. There will be no
ambiguity, at least in the guide. The guy who is doing it, of course,
is going to have a problem, because you can't really say I am doing
something that may affect, you know, the function of a major pump, of a
safety system and then say, "Well, qualitatively I conclude."
I mean the question what happened to the distribution that
everybody else is using will come up. They may still argue that it
doesn't change much.
MR. BELL: The longstanding, I guess, posture on this is
that these are qualitative guidelines and --
DR. APOSTOLAKIS: They will be qualitative.
MR. BELL: -- and the intent with this document was to stick
with that, and not in any case really compel folks to do a quantitative
or probabilistic --
DR. APOSTOLAKIS: But what I am saying is that as a
practical matter, if there is -- if these components are used routinely
in the PRA and there are distributions for the failure rate, I can't
imagine that the reviewer would not go and say, gee, this is the number,
what do you think?
MR. BELL: I agree with you. I would be very surprise if
they had that tool and didn't avail themselves of it.
DR. APOSTOLAKIS: The argument will have to be qualitative,
but at least the issue will be addressed. Maybe we should recognize
that. Just a thought.
DR. BONACA: I would like to ask one question. As you move
through the presentations today, I would appreciate if you could, you
know, emphasize the changes that you made since we met previously when
we reviewed this in detail, first, and second, how you addressed the
comments of the ACES.
We had a number of detailed comments. I think it would
worthwhile for us to know how they were addressed.
MR. BELL: We might be able to do that.
DR. BONACA: I don't mean to disrupt your presentation, just
simply, you know, I looked at it and a lot of this seems to be some
review of things we already reviewed before and I would like to know
what changes took place between the industry and the NRC since that
MR. BELL: I sure hope that is the case, because this is an
implementation document that really implements a final rule, and as you
say, we have been through --
MS. McKENNA: Maybe it would helpful if I went back just
briefly on one of my slides, which was kind of what changed in the rule,
just in case -- it's been several months for some of you and some may be
new members who aren't familiar with all the changes that were made.
There were some organizational changes. A major change was
adding definitions in terms of what "change" means, what facility is
described, what are procedures. In terms of the way those definitions
are applied, it allows some degree of screening as to whether something
is a change for the facility as described, and I think a number of
changes on the evaluation criteria that were alluded to, the concept of
the minimal increases in the likelihoods of failure and in consequences,
not much change with respect to the criteria of new or different
accidents and new or different malfunction, and removing the old margin
of safety and using two other criteria, one on design basis limits,
fission product barriers, and one on methods of evaluation.
Those were the things that were in rule and in the statement
of considerations, which was what the committee had reviewed.
Now at that time of course, 96-07 had been drafted more
along the lines of what the existing rule reflected, and therefore there
were a number of changes that were necessary, and I think you saw they
showed up first in the September version of the document that was
provided to the Staff.
Then, as was mentioned, the Staff provided some comments,
and then in December NEI responded with I think the matrix that you
mentioned of how they responded to the questions we had asked at that
The December version we had another meeting and we had some
additional discussions and there were some relatively small -- I guess I
would characterize them as changes -- in the January versions, and as I
mentioned, there still are some issues that we are wrestling with,
trying to get to agreement among all the parties within the Staff, and
we are trying to put those down on paper to let NEI know what those are
and where we have perhaps some hard spots with the guidance that is
One comment I think in terms of what is different. I think
it is recognized that a lot of the stuff that we saw in September was
kind of carrying forward from what was in the rule and putting that down
on paper. I look at it as there were some additional additions or
extensions or however you want to characterize it of taking the thoughts
a little further that were offered in December, and I think it is
primarily in those areas where material is a little bit new to us that
we are having these discussions, not so much on minimal increases in
consequences and things like that, where I think we are pretty much on
the same page, but when we get to some of the open issues.
As one example, we have had a lot of discussion about the
question of methods, criteria and methods and when is a licensee
changing a method sufficiently that an NRC review would be appropriate,
and one of the areas where we have had a lot of discussion is the extent
to which a method that was approved on a plant-specific, individual
basis can be then used by another licensee as a rationale that that
methodology is acceptable to NRC without further review, and that is an
area where there is some additional information in the guidance that we
have been discussing and we are making progress but we are not totally
in agreement with at this point a couple of other ones that are kind of
in that new bin as well, so I think that is -- some of the old issues
about what was in 96-07 may not really fit on the table anymore.
You know, we are into these areas where things were changed
and then they perhaps, to conform with the rule, and whether that has
all been taken care of and then perhaps in these areas get pushed a
So that is where I see it in terms of where the changes are
DR. WALLIS: Are we going to talk about some of these, like
MR. BELL: Yes, a little bit.
DR. WALLIS: I don't know quite where to interject a
question, because I don't want to interrupt but I do have a question
about that at some time.
MR. BELL: Let's try and move on. That was -- Eileen
highlighted a number of the key changes to the rule itself, okay, and
that part of the process is done and we're quite satisfied with the way
that turned out, but it is bigger than a bread box to take it the rest
of the way and translate that into clear guidance.
That is where we are now. The clarity of the guidance is
one of our objectives, comprehensive in the sense that we have been
looking back at past generic communications, notices, letters and so
forth that have touched on 50.59 and tried to be sure that the guidance
that we are preparing now deals with those issues and if need be
clarifies those kinds of things, so we are trying to have a one-stop
shop for folks on 50.59 implementation.
We think that the result will be more consistent and
effective implementation, owing largely to following through and getting
the NRC endorsement of it, and I feel that we are on track with that.
The status is, as Eileen mentioned, there was an iteration
in here that I have left off my slide, the December version, but we are
now at the point where we have what we consider to be a pretty good
draft subject to a few remaining issues that Eileen has identified.
So how do you implement this process? At a certain
level it boils down to this -- does the rule apply or is some other
process more geared towards governing changes, like in the area of EP,
emergency planning, security. There are change processes set for that.
Tech spec changes, that's another process.
It might surprise you to know that some utilities in the
past have done 50.59s for all of those, duplicate kinds of evaluations
and reportings and so forth. One of the important things this rule
clarifies is that's not necessary. Just do the evaluation where it
makes most sense and follow that set of guidance and you don't have to
do it more than once, so that's important.
Secondly, and it probably should say "must" -- must the
activity you are proposing to do be subject to the eight questions?
Somebody mentioned earlier that we went from three to eight questions,
or three criteria to eight criteria, and this middle step we call the
screening process, and I'll have a little more to say about that.
Finally and more to the point, once you get to the
evaluation criteria, there is NRC approval.
DR. APOSTOLAKIS: Isn't that question the same as the first?
MR. BELL: This one --
DR. APOSTOLAKIS: The first and third bullets, aren't they
the same thing?
MR. BELL: Well, 50.59 -- does it apply would mean do I even
need to do this screening step, or because it is an emergency planning
program change I have a separate criterion for that, in 50.54(q) or --
maybe it is -- so that is what this question means. This is the
screening step and the evaluation step and some of that might be a
Now if you skip a page in your package, I think you will
find a copy of this diagram.
MS. McKENNA: It is in the 96-07.
MR. BELL: That is Figure 1 from our document. This is
basically the applicability question, the step we just talked about.
This is the screening step, the evaluation and then implementation.
Over here I listed a number of the other regulatory
processes that might be more appropriate or are more appropriate for
certain changes. I mentioned EP, security. There are Part 20 kinds of
changes on effluents and things like that. One of the more interesting
ones is the maintenance rule, one of the areas, maintenance rule
guidance related to the new (a)(4) provision on the risk impact
assessments. Well, that, it would seem to me, it would seem to us if
you did a maintenance rule assessment under (a)(4) that you wouldn't
also need to do a 50.59 evaluation that duplicates that assessment, and
so fortunately both guidance documents were in play at the same
timeframe, and with the NRC we have been trying to get the guidance to
dovetail, again to avoid the duplicate or overlapping, I should say,
So that is one of the things that didn't settle down until I
guess the December version and there may be more clarification of that
that is needed.
MR. SIEBER: Maybe I can jump in and ask a question here.
One of the exemptions under the maintenance portion of this guidance is
the hanging of lead from pipes, and it says you don't have to do a 50.59
to hang lead.
I remember always doing that because you don't necessarily
hang lead on the system that is out of service for maintenance. You may
hang it on an active system. You need to know about whether you are
increasing stress in the pipe or stressing a hangar, bending something,
so maybe you can clarify to me exactly what it is you are doing when you
are talking about hanging lead.
MR. BELL: As I say, we are trying to make the two
guidelines dovetail. One of the things that the (a)(4) will say is
that, hey, if you do something like that under the maintenance -- for
ostensibly maintenance purposes you need to consider the effects of
those kinds of activities on other plant systems and if that is a new
addition to the most recent revision of that (a)(4) guidance, it is
intended to get exactly at that kind of question.
MS. McKENNA: This is one of those that I mentioned we are
kind of in this -- it is somewhat technical and also somewhat process
questions that we are wrestling with, because the kind of thing you are
talking about you could look at and say, yes, this is a change because
it is changing the piping or whatever I am hanging it on. The purpose I
am doing it may be because I want to do maintenance on something, and
what are the right kinds of assessments and processes that should be
looking at those changes, and can you truly be under one or the other or
are there overlaps, and that is still something I think we are dealing
with is it is not always easy to tell that it is just maintenance
because it is only working on the thing that you are doing maintenance
on, or it is 50.59 because you are hanging the lead on something else or
you are moving the equipment by something else, or there are other
configurations you can be in.
MR. SIEBER: You're setting up scaffolding -- there are all
kinds of things --
MS. McKENNA: Absolutely.
MR. SIEBER: -- having an impact on other systems and
hopefully the maintenance activity, the assessment that occurs because
of that covers all these other systems, as opposed to somebody putting
blinders on and saying the box I am working in is the piece of equipment
that I am working on and what I do around it, which might have a seismic
impact, a fire impact, change the loading on a system, is somehow or
other not included in that assessment -- just so that's clear.
MR. BELL: That's first an issue for the guidance and then
it is really a training and follow-through issue and we have a
maintenance rule workshop scheduled in March, 50.59 in April, and more
after that in terms of getting this kind of awareness --
MR. SIEBER: Part of that is organizational, because
typically 50.59s are done by the Engineering Department or the Safety
Department, whereas maintenance assessments are done by maintenance
MR. BELL: True.
MR. SIEBER: -- who ordinarily don't do 50.59s.
MS. McKENNA: I think because both of these, the (a)(4) is
in the process, it is kind of similar to 50.59 in a way in that it has
this when the guidance is ready then it becomes effective, and that
hasn't kicked in yet, that we don't really know exactly how it is going
to work yet, and therefore trying to -- we have two moving targets, so
to try to nail down one and then see how it impacts the other is
something we are having some difficulty with.
MR. SIEBER: Well, I would like to see them dovetail in a
way that there are no open holes between the two.
MS. McKENNA: That is what we are trying to look at. I
think NEI is trying to make a proposal that they are separate and --
some of the parts may be separate, parts may overlap and we try to see
where that overlap is, and, as you say, make sure that if we think it is
over there that it really is over there and that it is just not there.
MR. SIEBER: My picture of the process is that it is
interlocking, that it has to be comprehensive enough and everything has
to be covered someplace, otherwise you are going to have some unanalyzed
safety condition out there, which I think is unacceptable.
MR. BELL: In each case where there is perhaps more
appropriate or more specific regulatory process to evaluate the change
the guidance makes the point that, hey, there may be aspects of that
activity that affect both your emergency planning -- maybe it is a
change to your, what do they call it? -- facility --
MR. SIEBER: The EOF?
MR. BELL: Yes, the EOF, that affects one of the SSCs
credited in the safety analyses or designs so we are very careful I
think to identify that in the guidance and then there will be a training
and awareness issue in terms of the follow-through, so you could have to
do both, but where it is a purely -- clearly maintenance, clearly
emergency planning, then there are better rules than in the general
change rules to apply. The point is well taken.
Just a little more on the screening process --
DR. WALLIS: I guess I have a question. I'm sorry. This
big diagram that you showed us, really you need another set of diagrams.
"Perform 10 CFR 50.59 Evaluation" is just one blob on this. That
involves a lot of steps and I think you need to provide a framework for
how you do that.
MR. BELL: We certainly could --
DR. WALLIS: Not just words, but some sort of a diagram --
do this, then this, this, ask these questions.
MR. BELL: That clearly alludes to that section of the
document. It's a lot of words. There are some further documents that
are going to generically implement this on a plant-specific basis --
generically on a plant-specific basis?
MR. BELL: Generic procedures, forms and so forth, for
implementing this thing are -- we are working with some utilities to
develop those. That might be a place for additional pictures.
MR. SIEBER: Well, I agree with Dr. Wallis that it would be
very helpful in this document. It's the process under Section
188.8.131.52.4.3 -- it is difficult to follow unless you almost make a
DR. WALLIS: You have to make your own diagram.
MR. SIEBER: Yes, you have to make the diagram, whether you
do it or NEI does or somebody does it in order to understand it.
DR. SEALE: The hard part is knowing when to quit.
When we started on this, Dr. Powers mentioned the fact that
it was in fact the ability to quantify risk and to come up with
numerical values for changes that are purported to result from some
particular action that to our dismay, I guess, quantified zero, and made
us accept the fact that zero was no longer a neighborhood but was in
fact a point on the line, but there is another aspect to that.
We have mentioned it before here. That is, sometimes when
you make a change and the immediate impact of that change is perhaps a
slight increase in the risk, there are attendant modifications which
reduce the risk and so on balance the effect of -- and I will hesitate
to use the word "everything" -- is a negative.
The question is how far do you go before you declare that
you have got everything, because, you know, clearly it is the old
question of completeness that we face in any kind of evaluation like
this. It is still out there with this. Presumably what you are doing
here is coming up with this list of the regulations you want to look at
and so forth and somehow that tells you when you have done everything in
the context of the regulatory process to evaluate all the changes, but
it is still kind of an open issue, isn't it?
MR. BELL: In the context of 50.59, the guidance is that you
really need to take every change and look at it unto itself. Now you
can link certain other things if they are a direct result or a necessity
of the primary change.
DR. SEALE: Or a direct consequence of the change, yes.
MR. BELL: But there is essentially a prohibition against
drawing that envelope wider and wider until we -- we find, lo, we really
did improve our risk profile.
DR. SEALE: Yes.
MR. BELL: That doesn't sound risk-informed. That sounds
like we may be doing perhaps more than we need, nonetheless, that has
been the state of affairs and this document maintains that.
MS. McKENNA: I think that is the way the process is
structured, that you are looking at the individual changes and you try
to keep each of those minimal as opposed to perhaps a different
framework that was put all the changes together and as long as you net
has not gone more than whatever the number or is a net change of zero,
but the difficulty you have is how you put them together in those kinds
DR. SEALE: So the process known as bundling doesn't apply
MS. McKENNA: That's correct.
DR. SEALE: That is an interesting point.
MR. BELL: It is intuitive.
CHAIRMAN POWERS: To create a risk-informed 10 CFR 50.59,
wouldn't bundling ipso facto be used?
MS. McKENNA: Yes, that's my personal -- because you are
looking at things in a different way -- but you need some method of
looking at them together and if you are doing these individual changes
to different things, saying this one is a little bit here, this one's a
little bit there, that is kind of what the process does now.
You would have to have some different tool to do it in an
across-the-board type of sense.
CHAIRMAN POWERS: I'll be honest with you. I am using this
briefing more to think about going to a risk-informed 50.59 than I am
about the details, because I have a feeling that you can worry about
them enough for both of us, to be quite honest with you.
MS. McKENNA: Yes, I think sometimes down in -- I was going
to say the nitty-gritty but to a certain degree of actually certain
things crack through the system.
CHAIRMAN POWERS: I don't look at it just as being down in
the nitty-gritty. I just think it takes more knowledge about the length
and the breadth of it than I am able to assimilate.
MS. McKENNA: Yes, I think that is fair.
CHAIRMAN POWERS: You have been living with it, and I
haven't -- although I sometimes feel like I have been.
MS. McKENNA: Yes.
CHAIRMAN POWERS: But I do think it deserves to pull out of
this exercise that you are going through lessons that might be
applicable to going to risk informing, because I have a feeling that
people who are thinking about that may not have been living that either,
and they may not be down in the details of knowing what constrains you
and what constraints you want to carry forward and what you would like a
risk-informed 50.59 to get rid of.
I am sure you have run across constraints and said, yes, it
exists. It's because of the way people put the words together in the
past, and if we had to do it all over again we would never have written
the words that way.
I think the risk-informed is essentially a chance to rewrite
DR. BONACA: Although of course this is, what I want to
point out is a tremendous benefit to the industry that finally there is
a convergence of agreement. This is like, you know, this game has been
played in the field for 40 years and the referees have used rules which
are different from ones of the players. That is the fundamental
problem, so the players believe that they can do something. They make
some motion there and they get bolted for it. They get penalties -- and
this is the first time there is an agreement among referees and players
on what rules to play, so in and of itself it is tremendous progress,
the fact that finally they can speak the same language.
The reason why I am bringing this up, mostly to know when is
this going to be done? When is it going to be finished? I am sure that
the industry is pretty anxious to see it done.
MS. McKENNA: I'm sure they are. We certainly are too. As
I think I indicated, we are working to get these issues that I mentioned
that we are still discussing settled in some fashion to be able to put
the draft Reg Guide on the street. We're looking I'll say in the April
timeframe to do that and have a public comment process, then we resolve
and consider the comments and then kind of take it back through, as a
final Reg Guide, to the Commission for their approval, which would then
start the 90-day clock on the rule, so I think we are kind of looking
realistically, if we go out, say, in April and public comment ends some
time in June, get a package back together, it's probably towards the end
of the summer before it is back with the Commission and then however
long it takes from that point in time.
DR. BONACA: In the year 2000?
MS. McKENNA: It will probably be end of the year 2000 or
early 2001 at our current estimate, yes.
DR. WALLIS: Are you going to move to the next slide?
MR. SIEBER: I'm pretty sure that we have all had an
opportunity and have read 96-07. Maybe we could just move through that
DR. WALLIS: I would like to go to the next slide, because I
a very specific question about the next viewgraph.
MR. BELL: I was going to suggest that, you know, some of
this does smack as implementation detail of key issues that have been --
DR. WALLIS: I have a fundamental question, which is not
MS. McKENNA: Let's have your question.
MR. SIEBER: We'll put the next slide up and then we can --
DR. WALLIS: Next slide and page 37. This is a question of
determining whether or not there is a minimal increase -- the next one
-- whether or not there -- minimal permeates this whole document.
MR. BELL: Yes, that's right.
DR. WALLIS: And when I look on page 37, this is how you
determine whether or not you have a minimal increase in frequency of
occurrence, it says, "Normally the determination of a frequency increase
is based upon a qualitative assessment using engineering evaluations,
however the plant-specific frequency in PRA may be used."
Now this seems to me to be going the wrong way altogether.
It ought to --
DR. SEALE: Backwards.
DR. WALLIS: It ought to say normally PRA is the best method
of determining whether or not the frequency has been increased within
allowable limits. Occasionally it may be possible to make a qualitative
assessment which is acceptable.
But to put qualitative assessment as the norm seems to me
very strange. You can waffle about it -- that is the norm, the easy way
-- who is going to ever want to do the proper assessment involving a
It's backwards. Does the Staff really approve this
MS. McKENNA: I'm sorry, go ahead.
DR. SEALE: As a matter of fact, if I were asked to
characterize the relationship, I would say that a quantitative document
like a PRA to the extent that it is would be associated with frequency.
A qualitative assessment would be related to likelihood.
DR. WALLIS: It's the same thing.
DR. SEALE: Well, except likelihood is lesser degree of
DR. WALLIS: Qualitative is associated more with estimate or
DR. SEALE: Yes, right.
DR. WALLIS: And likelihood has a real meaning, like
probability. I am really concerned with putting this back to the sort
of wishy-washy language as the primary approach. Qualitatively is sort
of to be preferred and surely, if possible, you should use a
MR. BELL: Well, you are not misreading the intent. The
intent is to keep with longstanding practice and the utilities feel this
way, too, because they are very comfortable with the way they have done
this in the past, to use PRA in a support mode, not as the primary so
there are a number of considerations of a qualitative nature.
DR. WALLIS: Then how do you accept -- I don't understand
that acceptance criteria for a qualitative assessment. We're very
specific here about PRAs and a change of 10 percent and -- I understand
those, but qualitative seems to leave it all up to argument and
personality and persuasiveness.
MR. SIEBER: I think it is even worse than that, Dr. Wallis,
because the whole idea of going through this was to take the zero sum
game out of it and to be able to use a quantitative measure so that you
could have some leeway above zero change, because the old rule with the
three criteria really said no change.
MS. McKENNA: Right, may be increased.
MR. SIEBER: You had to get better or zero. You could not
tolerate any change, no matter how minimal it was, so this change in the
rule was to get us beyond that and we ought to be using that tool.
CHAIRMAN POWERS: Remember, I'm thinking about the old rule,
when they said zero they meant really indistinguishable from zero and at
the time when the resolution was by decades, what happened is we perhaps
had fooled ourselves or perhaps because of improved technology, now we
don't think that three times 10 to the minus four is the same as 10 to
the minus five because we have much higher resolution in risk and
suddenly what before were so small as to be essentially zero change now
became pretty substantial changes, actually.
They are no longer indistinguishable from zero and the
difficulty a lot of people had was that zero to them did mean
indistinguishable from zero and these numbers that we have now weren't.
DR. WALLIS: If this were risk-informed then PRA would be
the way to go, and then there might be another way, which would be
qualitative. The language sort of puts down PRA, makes it more
difficult, favors the qualitative approach and so you are moving away
MR. BELL: I think -- again, I think that is the intent. I
think it is recognized that if we want to make that lead to the more
effective tool that the time to do that would be when this rule is
risk-informed, as we talked earlier.
This is not the risk-informing of 50.59 and so the emphasis
on qualitative assessment you still see here.
DR. WALLIS: So why would a utility ever use anything other
than qualitative assessment?
DR. APOSTOLAKIS: That has been my problem too. If I were a
manager, I would never touch a PRA, especially if they have a comment
that Dr. Bonaca brought to my attention, Section 4.3.1 -- "It should be
emphasized that PRAs are just one of the tools for evaluating the impact
of proposed activities and their use is not required."
It used to be just a tool. Now it is just one of the tools.
It is a level lower.
MR. SIEBER: It's equal to intuition.
DR. APOSTOLAKIS: I come back to my earlier --
CHAIRMAN POWERS: Haven't we crossed this debate once
MS. McKENNA: I think we have in terms of --
CHAIRMAN POWERS: When we said that if we try to go
risk-informed now we simply delay a process, that it really is quite
MS. McKENNA: Yes.
CHAIRMAN POWERS: -- urgent I would say, to get some
stability in the field on this and although we cannot achieve perhaps
the quintessence of perfection, we can get something that is functional,
was functional, and will be functional in the future.
This has not been an area of abuse by anyone.
MS. McKENNA: Right, and I think the other point is that we
have to consider the wide spectrum of changes that a licensee may be
making, and some of them are going to be more amenable to a quantitative
assessment than others, and so I think that there's a lot of those kinds
of procedural things or I am doing something --
DR. WALLIS: But then you should say -- excuse me -- if it
is amenable to qualitative assessment then it is encouraged that they
use it, you know?
MS. McKENNA: That may be a fair comment.
DR. WALLIS: There may be some cases where qualitative is
DR. APOSTOLAKIS: Which is related to my comment earlier. I
mean if there is a frequency for the failure rate in the common PRAs, it
seems to me it should be a requirement to look at it. Yes, you can
argue qualitatively that it doesn't affect it much, but I just don't see
how a reviewer can ignore that. It can't be an option. That makes the
document much cleaner than it is now.
It is the same thing power uprate. The licensee did not
choose to go the risk-informed approach. Five seconds into the
presentation Dr. Kress -- "What did that do to the CDF?" The licensee,
"We'll tell you what it did. We did it." They had the answer because
they knew the question was coming and it will be the same thing here.
I mean you can't ignore reality that there are PRAs out
DR. KRESS: Unless, George, the qualitative assessments have
already been looked at and said we know that if we meet those if we look
at a typical PRA it's not going to affect it much.
MS. McKENNA: Right.
DR. KRESS: But I don't know that that has been done but I
think in essence that was in the thinking.
DR. APOSTOLAKIS: But it is not in the document.
DR. KRESS: It's not. It's not explicit, that's right.
DR. APOSTOLAKIS: When I say look at the PRA, that is what I
expect 99 percent of the arguments are going to be of that nature.
DR. KRESS: Yes.
DR. APOSTOLAKIS: I don't expect that one would do
calculations because I know these changes are not really the kinds of
changes that you quantify.
MR. BELL: Right.
DR. APOSTOLAKIS: But you may argue a little bit, you know,
and look at the distribution if there is a distribution.
CHAIRMAN POWERS: How many PRAs do you know that have
DR. APOSTOLAKIS: Well, you see, that's another thing.
CHAIRMAN POWERS: I didn't expect an answer.
DR. BONACA: One comment we made in our review, which
relates to this point is that so many of the decision points on
probabilities are hard-wired on defense-in-depth concepts.
For example, if you change something which affects the
diversity, it's by definition an increase in probability of a
MS. McKENNA: Right.
DR. BONACA: Therefore, you have that process that, you
know, almost pushes by definition the use of judgment rather than PRA
and it is all over the place.
I mean in my experience, and I have seen literally thousands
of safety evaluations -- literally -- most of these judgments are based
on that kind of cause, and most of them are hard-wired to criteria that
you have in the general design criteria or somewhere else that tells
you, yes, this is an increasing probability whether it is or not.
The other thing is I think the founding fathers when they
wrote 50.59 were thinking really about the fact that you put accidents
into four different categories, and I believe still today that all they
were worried about was that an increasing probability that shifted an
accident from one category to the next, because you have different
expectations for that.
Then with time we have taken probability to mean any
increase in probability and that is how we go into this bind, but again,
I mean, you know, the engineering judgment, it is so entrenched into the
use of 50.59 that it will be a big shift to go to a frequency category
DR. WALLIS: You mentioned the term frequency category?
What is a frequency category? I don't understand that.
CHAIRMAN POWERS: Page 46.
MS. McKENNA: There you go.
CHAIRMAN POWERS: At the time of the original 50.59 there
was a categorization of accidents in which they did talk about risk.
They did talk about accident frequencies, but the frequencies were
basically, if my memory serves me, expected at a facility every year,
expected in the lifetime of the facility, unlikely to occur, in the
lifetime of the facility, and extremely unlikely to occur -- and they
basically corresponded to something between a frequency of one in 100,
between 10 to the three and 10 to the four, and between 10 to the five
and 10 to the sixth.
DR. BONACA: And then for each of them you had different
MS. McKENNA: Correct, yes.
DR. BONACA: You could not have fuel failures if there were
highly probable or expected during the life, or you could have some fuel
damage if they were infrequent and you could have specifically, you
know, for LOCA you could have some amount of fuel damage, so they were
important for a reason. You didn't want to shift because you had
different expectations of the frequencies, but again the judgment was so
vague that judgment was sufficient for that perspective.
MR. SIEBER: Let me ask a question just to make sure I
When the rulemaking for 50.59 was initiated, was it intended
to make the new 50.59 risk-informed?
MR. BARTON: No.
DR. APOSTOLAKIS: It was explicitly stated no.
MR. SIEBER: Okay, and therefore the guidance probably
shouldn't be risk-informed either, right?
MS. McKENNA: I think the guidance can't be more
risk-informed than the rule is is the way I would characterize it.e
MR. SIEBER: The question is when you set about to
risk-inform Part 50, all of the Part 50, will 50.59 be included in that?
MS. McKENNA: It is one of the rules that is under
consideration. I don't know -- I wasn't that involved in the specifics
of how it might be done.
MR. SIEBER: So today the issue of whether this is
risk-informed or not risk-informed or tends toward deterministic and
qualitative as opposed to quantitative is consistent with the intention
that the rule was formulated in the first place?
DR. WALLIS: I find that entirely incongruous though, that
at a time when we are trying to move toward risk-informed regulation and
this is the propaganda, that there is a conscious effort to go away from
it in this particular change.
DR. APOSTOLAKIS: The rule has been approved though.
DR. WALLIS: I know, it's all right.
DR. APOSTOLAKIS: This is just the Regulatory Guide.
CHAIRMAN POWERS: I mean the decision was made consciously
that there was such a step going to have to be made to go to a
risk-informed 50.59 and not the least was the concern that a
standardization of PRA techniques would have to be in place, that it
would cause an unwarranted delay in the process and so this is viewed as
an interim measure, and I think that was not a bad decision.
I think we did not anticipate it would take this much to get
where we are now, but that is probably because we did not recognize how
pervasive the use of the 50.59 process is, even though we all said it
was pervasive, nor did we understand how intertwined the language is
with itself in the process and so you have to be very careful about
DR. APOSTOLAKIS: In many respects though -- first of all,
this is not risk-informed. The fact that you are just referring to
frequencies of malfunctions doesn't make it risk-informed.
MS. McKENNA: Right.
DR. APOSTOLAKIS: Okay. We are not using any risk insights,
but I think the use of this is very similar, the issue that is being
raised is very similar to this two-tier system that I mentioned earlier.
The fact that there is a PRA out there forces you to do
certain things regardless of the regulatory system.
In this particular instance the fact that there are
distributions for the failure rates for certain equipment is a fact of
life and you can't -- what if the reviewer says I would like -- it's
boring -- for this pump I have --
CHAIRMAN POWERS: George, let's make very clear who the
reviewer is in this case.
DR. APOSTOLAKIS: Whoever goes to inspect the records.
CHAIRMAN POWERS: This is something that occurs on a perhaps
DR. APOSTOLAKIS: So the probability of having this is low,
you are saying?
CHAIRMAN POWERS: It happens once in awhile. I mean it does
happen once in awhile.
MS. McKENNA: You are talking about the inspection?
CHAIRMAN POWERS: I mean a 50.59 determination is not
trotted in to the NRC Staff and they say here is what we did, did we do
it right? That is not done. A review is done at the plant.
MR. BARTON: An annual review of safety evaluations might
pick one of these things up.
DR. APOSTOLAKIS: So the PSA group at the plant may not
raise this question, you don't think?
MR. SIEBER: You probably will do three or four or five of
these a day. I mean you do them. That's the way of life.
DR. BONACA: You probably have 1000 or hundred per plant and
they have maybe three, four thousand issues they are screened for.
CHAIRMAN POWERS: That's right.
MR. GILLESPIE: Right.
CHAIRMAN POWERS: That is not to say they do not get
reviewed. When you do one, it gets reviewed at the plant.
MR. SIEBER: It is at the plant that it gets reviewed.
DR. WALLIS: It's a management decision when the PSA group
comes up and says actually we have an increase which is more than 10
percent, and yet the qualitative determination people say it's fine.
DR. APOSTOLAKIS: And I will tell you something else. In
Section 4.3.1, the sentence I read before, it really should be deleted
because PRA is not a tool for evaluating the impact of proposed
activities. We just agreed that this is not risk-informed. It's a
MS. McKENNA: The point was that when you are looking at,
the licensee is looking at the change and trying to decide whether that
is a good change to make or what are the ramifications of the change
that it may be helpful to them to understand exactly what you were
talking about and does it change their PRA in any sense, but that is I
think the intent.
DR. APOSTOLAKIS: It doesn't make sense --
DR. BONACA: Well, let me just say, it allows at least for
it to be considered. We get to the point where we used to use PRA to
make, disclose a probability, and every time we did, we got in trouble,
because if you make a qualitative call there's no problem. No one
questions it. If you have a quantitative evaluation, everybody
questions it and then there is very much insight -- are you affecting a
DR. APOSTOLAKIS: But I am not allowed to argue on the basis
of risk insights. If I go to the criteria later, it says deleting or
modifying system protection features or equipment protection features.
Can I come back and say, yes, I deleted these protection features, but
look, this component has a risk importance of 10 to the minus 100? I am
not allowed to say that.
What I have to do is argue that the probability of
malfunction of this component regardless of how important it is
negligible -- is minimal, so there is no PRA at all. Just because you
use a failure rate, you can't say -- so it seems to me that PRAs are not
in fact invited to participate in this, so why -- I mean that's a fact.
All the criteria you have later have nothing to do with risk insight.
CHAIRMAN POWERS: I think the discussion has gone to the
minutiae here. Can we progress ahead?
MR. BELL: And I suggest I give it back to Eileen.
MS. McKENNA: Yes, we are running short on time. Yes, go
DR. BONACA: I have one specific question regarding
something we communicated to the Staff and it has been misinterpreted.
MS. McKENNA: Okay.
DR. BONACA: And that is on the second slide from NEI
regarding fission product barriers exceeded or altered and I believe
that the ACRS for the specific case made an example that you have a
design change. The design limit hasn't been changed.
MS. McKENNA: Right.
DR. BONACA: What you have done, you have diminished the
capability of the barrier because you have put ruptured disk right above
the design limit, and so you can -- in fact, it can alter the capability
of the barrier without affecting its design limit from inside pressure,
and here the guideline uses the word "altered" in a different sense. It
doesn't address that and I just wanted to point that out. It has
defined a new definition, which is interesting -- because we wrote it
down and the word was taken and it was altered. The word "altered" was
DR. BONACA: I don't think it is a major issue, just simply
that a point that we made we think has some merit because at the design
level you must still have a clad that you design to have a certain
capability of withstanding internal pressures, but you may decide to
have it do certain things and you are essentially reducing the margin
that you have in the barrier.
I mean all I am talking about is the margin. I don't think
that the margin should be reduced in any way or form without NRC review.
MR. SIEBER: Well, you know the example that it gave, which
is corrosion of a containment liner, you know the 50.59 screening would
say that has to be reviewed by the Commission, okay, because now you
have basically lowered the design strength and the capability and we
know about cladding. They change cladding from one form of zircaloy to
another throughout the years and every one of those you had to go back
and get a license amendment for it, so neither one of those -- all those
cases would screen through 50.59, either the old rule or the new rule,
to go to the Commission and get a license amendment.
DR. BONACA: And the word "alter" really meant to control
that capability of barriers, realizing that for example in containment
we are counting on 130 psi rate because although it is not in the design
basis, we have by now taken credit for it in severe accident and we like
to have that margin, and so, you know, this in the guideline has really
misinterpreted what we meant.
MS. McKENNA: Okay.
MR. SIEBER: What I would like to do, we are past our normal
time, but I would like to get an opportunity to look at what are the
outstanding items right now.
MS. McKENNA: Okay. I did have it in the packet and I think
we have touched on a number of these through the discussion.
The first one, in a way it is similar to some of the
discussion on maintenance, where the question of changes to fire
protection programs, which are programs that are in the FSAR or
referenced in the FSAR, and the proposal was that most plants have this
license condition and the proposal is evaluate against the license
condition, not against 50.59.
This is again one of these where the question of whether
there is truly complete overlap versus there is a partial overlap is
what we are looking at. I think our view is that it is not a complete
overlap between what the license condition provides and 50.59, but that
is one of those kind of regulatory process questions as well.
MR. SIEBER: Okay. You are worried more about the records
and the bases upon which you would do it rather than the fact that it is
being reviewed properly or not, right?
MS. McKENNA: Well, I mean, yes, kind of what is the --
because the license condition says you can make changes as long as you
don't adversely affect the capability of safe shutdown.
MR. SIEBER: So there is no record of it other than the
MS. McKENNA: Right. There is no requirement for that
record and I think 50.59 would say if you are making a change to the
FSAR, keep records and explain why it is okay and keep the records and
all that kind of stuff, so there are issues with that.
MR. SIEBER: Okay.
MS. McKENNA: I think I have mentioned already the question
on methods. The second one is kind of the plant-specific versus the
generic. The first one is just in terms of how if you are looking at
changes to pieces of the method and one of the other things is that as
long as your answers come out about the same, you haven't really changed
anything, but again just kind of one of the things we are looking at is
it's not necessarily did your peak number come out the same, but that
you have kind of the same shape and response, and that is one of the
clarifications I think we are looking to make with the guidance.
One of the issues that we are looking at is whether for
instance for fuel one of the things that is in there, typically we look
at something like a DNBR as to how you would judge your performance of
the fuel and what was proposed in there as the design basis limit was
basically the 95/95 confidence level.
I think the Staff's view is that it should be the particular
value for that fuel that is the limit, not the confidence level. We
didn't really get into the screening question. You may have seen in the
document some discussion about screening with respect to whether you are
affecting functions. We are looking at this and part of that, there is
some guidance definition, if you will, of what is meant by design
functions, and we are looking at that as to whether is sufficiently
comprehensive at the screening level to make sure that changes move
forward for evaluation.
MR. SIEBER: Could we go back to the third bullet there?
MS. McKENNA: Yes.
MR. SIEBER: Why isn't the Staff comfortable with the 95/95
MS. McKENNA: I think I would have to call on one of our
reactor systems staff who I think we have in the audience to -- Chris,
do you feel comfortable answering that? Would you come to the mike?
MR. SIEBER: Go to the microphone, please.
MR. JACKSON: I am a little bit uncomfortable here.
The limit for fission product departure from nucleate
boiling is the ratio at which you would lose confidence so that you
might experience a department from nucleate boiling. The 95/95 is the
acceptance criteria. That is just 95 percent probability with 95
percent confidence, so we would see that as the confidence bounds of the
acceptance criteria for whatever limit you came up with, so that is the
MR. SIEBER: So you are satisfied with that or not? You
want a specific number?
MR. JACKSON: I want the value, the ratio.
MR. SIEBER: That would just give you more information --
how confident you are that that number is a good number?
MR. JACKSON: The value that they come up with would be NRC
reviewed and approved.
MR. SIEBER: Right.
MR. JACKSON: And they would demonstrate through analysis
that they have achieved the 95/95, but the limit is the ratio -- is the
critical heat flux. I mean that is the value that they would use to
calculate at the plant.
MS. McKENNA: That is Chris Jackson, Reactor Systems Branch.
We have just one more slide, just a couple more, actually I
think the other slide, we talked the numerical values. I think in
general we are comfortable, with some clarifications we were interested
in. I think the committee indicated some clarifications that we might
want to consider with respect to the numerical values that we see in
The last one was this maintenance rule assessment I think
that we have already talked about, so those are the areas where we still
have some questions and we will be sending that letter in the very near
MR. SIEBER: Okay. Let me ask a final question.
Does either the Staff or NEI feel that any one of these
issues is unresolvable in a reasonable amount of time?
MS. McKENNA: No.
MR. BELL: No.
MS. McKENNA: As I said, I think the nature of the
resolution may vary. NEI may agree to make some adjustments. We may
agree that we are just going to disagree and we'll take exception to
certain things but I think we can resolve them. It's just what kind of
resolution we come to on the particular issues we are dealing with.
MR. SIEBER: Thank you. Does anybody else have any
questions that they would like to ask at this time?
DR. SEALE: When is this industry workshop that you are
going to have?
MR. BELL: April 10th and 11th, the Clearwater Beach Hilton,
Florida, good place.
MR. SIEBER: Any other questions?
MR. SIEBER: If there are no other questions, Mr. Chairman,
I return the meeting to you.
CHAIRMAN POWERS: Thank you, John. I think we need to
struggle internally to come up with a strategy on this, to minimize any
impediments the Staff has in moving forward. I will articulate my own
feelings here that we have a low level of value added at this point of
getting to the implementation.
I think our time might be better spent on discussing the
generation going to risk-inform 50.59 rather than rehashing old issues,
but I certainly think we need to discuss it with the committee and get
information back to the Staff as quickly as possible, so they can set
their own schedules in this regard.
With that, Chairman's privilege, I will recess us till five
CHAIRMAN POWERS: Let's come back to into session. Our
final presentation today deals with a topic that can't possibly have any
controversy associated with it.
CHAIRMAN POWERS: And I am sure the presentation will go
very smoothly since there will probably not be a single question. Most
of the questions we find have procreated dramatically.
Dr. Kress, I think you are going to lead us through this
DR. KRESS: I don't know if that is the proper words or not.
CHAIRMAN POWERS: Maybe introduce it.
DR. KRESS: Introduce it maybe -- yes. This is the session
on proposed and potential revisions to the Commission's Safety Goal
As you know, we have had meetings on this before and the
Staff has told us, at least identified the issues they are looking at
that might be part of a revised policy statement. At this point I think
we are going to get a kind of status report on where they stand on, what
sort of position they are going to take on these various -- I think
there was about eight -- issues that they are looking at and we
commented before we thought these were a comprehensive set of issues and
the right things to look at and see whether or not they ought to be in
the policy statement.
I think today we are going to hear what the Staff about each
of these and with that, George, you have anything you want to say before
we get started?
I would like you guys to listen carefully on the Staff's
positions on each of these issues, and then see what you think because
we will be writing a letter, probably not this time, but at least at the
next meeting, the March meeting, so with that I will turn the floor over
to I guess Joe Murphy or -- Joe? Okay.
MR. MURPHY: Thank you, Mr. Chairman. As you said, we have
discussed this subject with the committee on several occasions over the
last several years. I would like to back up in the history a little bit
more than is indicated on the slide and just remind you that the policy
statement itself was issued by the Commission in 1986.
There was a very important Staff Requirements Memorandum
that was issued June 15, 1990, which I will reference, that I think
explains the policy statement significantly, and one of the things we
are doing is trying to incorporate some of the messages from that SRM
into the policy statement.
Finally, I want to point out that the Commission in its
initial SRM that authorizes us to go forward with considering this in
the SRM on SECY 98-101 stated that the safety goal should remain a high
level document describing the principles consistent with the
Commission's views on how safe is safe enough, and then told us the
Staff should be mindful not to include too many quantitative guidelines
which would make it overly prescriptive. I think that guidance is
With that, I will go to what I have up here. In SECY 99-191
we informed the Commission of the progress in developing recommendations
and we made a recommendation for a study of the feasibility of
overarching safety principles. As you are aware the related SRM told us
to proceed with the recommendation to modify the policy statement but
disapproved the study of the feasibility of the overarching safety
principles and of course that is reflected in this presentation.
I know that time is short, but I would like to run through
just briefly the relationship between the safety goals and the
regulations so you can see where this fits in the whole picture.
Basically the regulations establish the requirements that
enable us to do our job. The policy statements provide a high level
expression of the safety philosophy of the Commission and the
expectations of the Agency.
DR. KRESS: How does that influence what you do in the --
MR. MURPHY: Well, I'll give you an example. After the
Safety Goal Policy Statement was issued in '86 or as it was being
issued, being developed, we issued the regulatory analysis guidelines,
the thoughts in terms of what is acceptable and what isn't in terms of
limits on core damage frequency and large early release and this sort of
thing, translated into the regulatory analysis guidelines which gave us
an indication of how far we should go in looking for regulations.
DR. KRESS: Do you think now -- that was one area where it
did influence it, and I'll agree, it was a big influence. Do you think
now in -- I'll call it the era of risk-informed regulation the Safety
Goal Policy Statement would influence the direction that would go in?
MR. MURPHY: I think if we -- yes and no, and the reason
it's such an answer is right now a lot of what is going on in
risk-informing Part 50 draws on the principle of Reg Guide 1.174. One
of the things I am talking about is taking much of what is in that
document, which is addressed towards changes in licensing design basis,
and is in the form of a Regulatory Guide, and putting it up into a
Commission policy statement.
Once that information is in the policy statement, yes, then
the policy statement will affect what is going on. Right now this is
going on in parallel. We are bringing the Safety Goal Policy Statement
up to the current practice at the same time we are going forward in
DR. KRESS: Would you proceed to risk inform the regulations
in the same way, even whether or not the policy on safety goal policy
was changed or not? Is it necessary to have a policy statement?
MR. MURPHY: It's not necessary to have a policy statement.
The policy statement does provide better guidance to the Staff in terms
of high level philosophical guidance.
DR. KRESS: The reason I am asking those questions, I have
some very distinct opinions on things that are needed, that are
policy-like things to properly risk inform the regulations. I just
don't know whether they need to be in the policy statement or not or
whether you could proceed with them, as long as they are not too
inconsistent with the statement as it now exists.
MR. MURPHY: I think the policy statement should remain a
high level document, so it covers basic philosophy, if you will, of
safety, as opposed to getting into great specifics.
A lot of the things that take the guidance the Commission
has given in the statement and put it into regulation, there has to be a
lot of flexibility in doing that, and I think that is what the
Commission meant when they gave us the warning that we had --
DR. KRESS: About not being too quantitative?
MR. MURPHY: Yes.
DR. WALLIS: Well, Joe, the policy statement could serve the
role of sort of a constitution to which you appeal when you have to make
a decision and it is difficult to decide which way to go on some
regulatory matters and you go back and appeal to certain items in the
policy statement in order to make a rational decision based upon some
larger principle or assertion.
I haven't seen that happen. The Safety Goal Policy
Statement seems to be out here somewhere, where we are always down in
the details and very rarely does anyone say we can resolve our
controversy by appealing to the policy statement.
DR. KRESS: Tom has a point, Tom King from the Staff.
MR. KING: That is not totally true. When we put together
1.174 we went back to the policy statement to find the words that were
in there about assessing total risk, about using mean values, about
defense-in-depth, treatment of uncertainties. We went back and used the
policy statement and I think as we go forward in risk-informing Part 50
we may come back there again and see what does it say about certain
issues, so it has been useful.
DR. WALLIS: 1.174 is often cited as being a successful
story and I am glad to see it's done that here, but it doesn't seem to
happen in other contexts very much.
MR. HOLAHAN: This is Gary Holahan of the Staff.
I would use a little bit different analogy. It seems to me
the safety goals are more like the Declaration of Independence, which is
an expression of desires and expectations but in fact has no legal
status at all, and I think that is what the safety goal is. It is not
the Atomic Energy Act. It is not the regulations. But it gives you
some idea about what you are trying to achieve.
DR. WALLIS: Well, the Declaration of Independence was an
excuse for performing an illegal act at the time.
MR. HOLAHAN: I believe that matter has been settled.
DR. SEALE: Somehow I knew that was going to come up.
MR. MURPHY: The point I wanted to make, I think, Gary said
very well, and that was the safety goal is not a legal requirement but
they are guidance for the Staff as to how to develop regulations.
As to the Safety Goal Policy Statement being used in all our
regulations, it would clearly not because many of them, most of them
were developed before the policy statement.
Has it influenced those that have been developed since the
policy statement was issued? It has through the regulatory analysis
guidelines primarily. One of the reasons for putting everything in one
place is to have this high level guidance for when we go forward with
risk-informing the rest of the regulations and have it in a place that
expresses Commission policy, perhaps in a more logical way or more
visible way than buried in a Regulatory Guide, but we have the
Regulatory Guide. We are going forward.
This is not stopping our progress in going forward, so it is
In terms of the changes to reflect current policy, I have
already talked about some of these. The five principles in Reg Guide
1.174 give us the principles for integrated risk-informed
decision-making. We think they should be generalized, and I will show
you those in a minute, to reflect the broader usage, rather than just
for design basis changes.
And put into the implementation of the policy statement --
DR. WALLIS: Let me ask you something: Are these goals
being met today? Is there a measure of how well the safety goals are
MR. MURPHY: We have been -- the policy statement, as it
applies right now, applies to the industry as a whole, rather than
We have results of all the IPEs, which we can compare
against at least the subsidiary on core damage frequency. Some meet it;
some are a little bit above it, based on the analyses that were done
some time ago.
DR. WALLIS: But understand --
MR. MURPHY: But understand that the purpose of the goal --
and it is an important message out of the 1990 SRM -- that is, the goal
is something you strive to meet.
DR. WALLIS: Yes, but then --
MR. MURPHY: In striving to meet it, you use cost/benefit
DR. WALLIS: But it is the primary statement about how safe
is safe enough. Then you ought to have an unequivocal answer as to how
well are we doing on this measure.
MR. KING: This is Tom King again. After the safety goal
implementing guidance came out in the early 90s, there was a request
from the Commission for the staff to go assess how well plants stack up
against the safety goals.
To do that right required not only risk assessment at full
power, but at shutdown and external events, maybe not for every plant,
but certainly representative of the types of plants that are out there.
And we put together the price tag of doing that. It never
made it through the budget process to really get funded.
So the best we have now is, we took the IPE results. As Joe
said, there is a section in the IPE insights report on -- I forget the
official title, but it is basically a comparison against safety goals.
And it takes the CDF measures that were reported in all the
IPEs, and it tries to, based upon NUREG 1150 results, extrapolate those
to what they mean in terms of the QHOs.
And it says basically that given that information, most of
the plants meet the safety goals. There are probably a dozen or so that
you could argue are questionable, but we didn't carry it any further to
do any specific calculations to say definitely yes or no for that dozen
And that's about as far as we've gone so far.
DR. KRESS: Clearly, the NUREG 1150 plants meet it.
MR. KING: Yes.
DR. KRESS: But they're not --
MR. KING: Again, that NUREG 1150 analysis is limited, too.
It's not a shutdown risk.
DR. SHACK: Again, though, if this is a statement of how
safe is safe enough, how does this jibe with the expectation you have
for the new reactors, where, for example, the core damage frequency
would be a factor much, much lower?
DR. KRESS: Ten lower, a factor of ten lower.
DR. SHACK: Well, I guess that was the expectation. I'm not
even sure that if you walked in with a 10(-5) that you would have been
told to go back and look at that again.
MR. KING: That question was put before the Commission back
in 1990 when we proposed implementing guidance for the safety goals.
And the question was, should future plants have a 10(-5)th
CDF versus current plants' 10(-4)th. The Commission said no; 10(-4)th for
And even though advanced plant designers are coming in,
EPRI, through their utility requirements document, the ALWRs having
their own goals that exceed what the safety goals put forth, that's not
DR. SHACK: Isn't there an expectation statement, though?
MR. KING: Yes, there is an advanced reactor policy
statement that was issued back in '86 also that said we, the Commission,
expects future plants to do better.
But they didn't say what "do better" means. They just said,
we expect you to do better.
DR. SHACK: So, although they're safe enough, you expect
them to be a factor of ten safety?
MR. KING: They didn't say a factor of ten; they said, you
know, consider passive safety features, and, you know, others. Less
reliance on human actions and other things that would improve safety,
but they didn't say a factor of ten.
MR. HOLAHAN: But I think in implementing that later on in
the process, a factor of ten was, in fact, used as a way of judging
whether those expectations were being met.
DR. WALLIS: I have a problem as a member of the public,
understanding why safety enough is something you strive for. I would
think that safe enough is the minimal standard, and more safe than safe
DR. KRESS: Well, safe enough has Joe's qualifying phrase on
the end of it; safe enough in the sense that to get there, you have to
MR. MURPHY: Yes.
DR. KRESS: That's a qualifier.
MR. MURPHY: Perhaps in terms of this, later on in the
discussion, there is a discussion of the structure of the safety goals
that derives from comments that the Committee made.
They deal with the question of a three-region approach that
has an area where you -- the risk is too high and you must do something;
an area where you employ cost/benefit to decide whether you do
something; and then an area where you have reached the level low enough
where the risk is low enough, and you would not enforce any more
I'll get to that in a minute.
DR. APOSTOLAKIS: But let me understand this, though. We
apply now, cost/benefit analysis, even to plants that have a core damage
frequency and LERF below the goal; is that true?
MR. MURPHY: We apply cost/benefit analysis most times in
terms of the more generic things, in terms of doing rulemaking. In
terms of plant-specific backfits, I have to ask Gary, but I don't
believe we've had many in the last few years that have been justified on
the basis of the backfit rule and the cost/benefit analysis.
MR. HOLAHAN: Well, we have not had many, and I think that
the judgment about how to do it wouldn't be based on whether they were
above or below 10(-4).
DR. APOSTOLAKIS: My point then is that it's really -- I
mean, when you say it's safe enough, that means something specific here.
It's not safe enough so that the regulatory agency does not concern
itself with this plant anymore.
MR. HOLAHAN: No.
DR. APOSTOLAKIS: Because in the three-region approach, the
way the British have published it, it's show the bottom region -- my
understanding is they would not even consider cost/benefit analysis.
It's so low that we would just leave it alone.
Cost/benefit is between -- in the middle region. So if
something is safe enough, why would you do cost/benefit analysis?
DR. SHACK: I don't think you would here.
MR. MURPHY: No, you wouldn't. Let me -- remember, these
things are structured more --
DR. APOSTOLAKIS: Even generically, though. This always
comes back to the issue of plant-specific goals.
MR. MURPHY: Now, the structure that was mentioned is very
similar to what's in the backfit rule. Backfit requires the necessary
to achieve adequate protection.
Backfits are allowed if there is a substantial increase in
the overall protection, and the costs are justified by the degree of
And then finally, backfits are not allowed if they can't
pass the backfit rule. The backfit rule essentially establishes the
The safety goals help us define the bottom line.
DR. APOSTOLAKIS: What would the safety goals be in that?
MR. MURPHY: The safety goals are here.
DR. APOSTOLAKIS: There?
MR. MURPHY: At the bottom, yes.
DR. KRESS: The safety goals are lower in adequate
protection and that defines your three regions.
MR. MURPHY: You have an area of adequate protection.
Between, underneath that, we must comply with. Below that, you have an
area in which the backfit rule controls and you do it if it's
DR. APOSTOLAKIS: But right now, we really have not
specified the boundary between the first and second regions?
MR. MURPHY: No.
MR. HOLAHAN: Let's also be careful. The Commission has not
defined that adequate protection equals some numerical value.
DR. APOSTOLAKIS: Yes, I agree, I agree.
MR. HOLAHAN: So it's hard to --
DR. APOSTOLAKIS: My thesis is that the safety goals, as
they are interpreted today, would not define any of the boundaries. But
they are not defined in any boundary.
DR. WALLIS: That's the problem I have. When you talk about
-- when it says the risk to the population should not exceed .1 percent
or something, that's a pretty clear thing, and it says should not
exceed. It's a clear one boundary. It's not two regions, three
regions; it's one criterion.
And you should be above it. I mean, you should not cross
this threshold. It's not a goal to be aimed at; to me, it's a statement
DR. KRESS: It doesn't say "must not."
DR. WALLIS: It's equivocation to say that it's not.
DR. APOSTOLAKIS: But that boundary, though, is not there.
That's another point. It is not there at all.
DR. WALLIS: Well, that's what I have a problem with. When
you say here's this fundamental statement of philosophy, but it really
doesn't matter, because we interpret it some other way.
MR. MURPHY: I think the way you interpret it at the bottom
boundary is -- the way decisions are made is in terms of the rules. The
backfit rule was developed using the safety goal as the basis for
establishing where the breakpoint was cost/benefit analysis.
And to that extent, the safety goal has affected that bottom
DR. APOSTOLAKIS: How has it affected that?
MR. MURPHY: In the developing of a 2,000 per person rem and
how the whole thing was put together. There is in the regulatory
analysis guidelines, there is a plot, a graph, a matrix, that shows you
the relationship between core damage frequency and large early release
frequency that gives you an indication of whether or not there is a
substantial increase in safety involved in what you're proposing, one of
the requirements of the backfit rule.
So it enters in through that mechanism of getting into it.
The safety goals themselves are not part of the regulations, but it
focuses in on that bottom area.
Now, in the top area, I think there is a -- I'm losing my
viewgraphs here -- a very good statement that came out of the SECY
99-246. And that was that risk estimates are important, but they're not
the whole body of things that are considered as you get into this
question of whether or not there's reasonable assurance of adequate
DR. KRESS: Let me ask you a question about that, Joe. I
agree with you on the statement.
But we have in 1.174, a line that represents the upper
boundary and that if the lower boundary is the safety goal, you have a
line that represents the upper boundary.
Is it inconceivable to think that that upper boundary line
could not be incorporated into an enhanced definition of adequate
protection that includes all these things also? Is that beyond the
MR. MURPHY: No, and I think, as you know, we had proposed
looking for such a thing in the overarching safety principles.
And the Commission's guidance came back and said it's
premature at this time. You need to get more experience with what you
And the way I interpreted their SRM, without reading it
literally, you need to get more experience, and you --
DR. KRESS: But your point assumes that there is a
three-region concept that ought to be policy, and there is some line up
there that we're searching for, and whatever the value ought to be,
whether it's the 1.174 value or not, it seems to me like it would be
appropriate to incorporate that as an addition to the definition of
adequate protection, not the sole definition, but in addition to it.
MR. MURPHY: Yes. See, I think what we can say is that that
structure -- we have a structure similar to what the Committee talked
about. We have that already in the backfit rule.
We need to include the position, and perhaps we need to word
it a little differently. Assume the 6/15/90 that basically says the
safety goal is to find the bottom demarkation line between cost/benefit
space and no change necessary.
DR. KRESS: I think we agree on that.
MR. MURPHY: And then finally is to take what the guidance
is that we have from the Commission. As we get more experience, it may
well be appropriate to consider the degree to which we can use risk
analyses and defense-in-depth to better -- to provide a better
definition of the upper limits.
And whether you want to call that upper limit adequate
protection, or you want to say adequate protection is broader than this
upper limit, but we can define the upper limit in a different way, which
DR. KRESS: I would certainly say something broader that
MR. MURPHY: I think I would agree with that. I think the
Commission has given guidance to get more experience with what we're
doing, and to then come back and try to do that.
DR. KRESS: Do you might say eventually you might get there,
but you're just not ready yet?
MR. MURPHY: Yes, I think that was the direction we have --
DR. KRESS: You need to define that upper limit.
MR. MURPHY: At this point --
DR. KRESS: But how can you -- the question I would have is,
how can you proceed to mis-conform the regulations without that upper
limit unless you use some ad hoc value, which I am presuming is going to
be 1.174, because that's the only thing that's around right now.
DR. APOSTOLAKIS: Why are you using 1.174?
DR. KRESS: That's 10(-3), basically, CDF, and then a LERF of
10(-4), I think.
DR. APOSTOLAKIS: That's --
DR. KRESS: I think the line is drawn on one of your charts;
MR. HOLAHAN: No.
DR. APOSTOLAKIS: There is not.
DR. KRESS: There ought to be a line at --
MR. MURPHY: The line on the charts are 1.174.
DR. WALLIS: If I remember correctly, it doesn't go any
further than that.
DR. KRESS: I thought that was a line. It was just the top
of the chart.
MR. HOLAHAN: I think it's falling off the end of the earth.
The map just doesn't go further than that.
MR. MURPHY: The numbers are basically used for the
demarcation line, 1.174, roughly akin to the safety goals.
DR. KRESS: There ought to be an upper line.
DR. APOSTOLAKIS: But I think it depends --
DR. KRESS: I think you have to have limits.
DR. APOSTOLAKIS: Yes.
DR. KRESS: In order to do a proper, risk-informed
DR. APOSTOLAKIS: And I think it is not inconsistent with
what you're saying, Joe, to change the approach a little.
Instead of agonizing over what is adequate protection,
which, of course, is what he just showed us, what does that mean?
Logically, it means, yes, risk insights, and
defense-in-depth, and safety margins and whatever else you need.
But by the very logical method, I can have definitions of
inadequate protection. If any of these Ns is above a certain limit,
then I'm sure I don't have adequate protection.
And if you think that way, then you are not inconsistent
with the Commission's SRM; you've satisfied Dr. Kress because there is
an unacceptable level of core damage frequency the way we calculate it
now, for which, if you exceed -- I don't care what else you do --
adequate protection is not there.
And you can say the same thing about defense-in-depth. We
have been told many times that I don't care what the risk number is; if
you don't have redundancy in this place, we're not going to accept it,
and we try to justify that.
DR. WALLIS: You're talking to this community, and I think
the first bullet up there really talks to the public. You've got to be
able to tell the people what kind of adequate protection they're
getting, why you think it's adequate, and what assurance you have.
Well, all our arguments here seem to be internal on how does
sort of a bureaucracy make decisions. But, surely, the first question
is, are you fulfilling your public trust to make number one happen.
And if you can't provide a measure of it, how do they know
you're doing your job?
DR. APOSTOLAKIS: Again, these are two distinct questions,
in my view. And the center of -- or studies of what strategic and
international studies also ask the NRC to define numerically, and what
is adequate protection.
I think it would be very difficult right now to define it
numerically. However, it will not be as difficult --
DR. KRESS: To do what I said.
DR. APOSTOLAKIS: To define inadequate protection.
DR. KRESS: Yes.
DR. APOSTOLAKIS: Because that I can do in terms of each
measure, not the combination.
DR. KRESS: Yes.
DR. APOSTOLAKIS: And that will help me with risk-informing
Is the airline industry, for example, using as a sole
criteria of adequate protection, the probability of death per passenger
mile? Probably not. It's a collection of things they are doing to make
sure that flying is safe.
DR. KRESS: Absolutely.
DR. APOSTOLAKIS: So the lack of a numerical measure is not
something unique to us.
DR. WALLIS: What do you tell them when they ask this
straightforward question? Tell us why you have reasonable assurance of
DR. KRESS: You tell them we do this --
DR. WALLIS: In two sentences.
DR. KRESS: You tell them we mean all these regulations, we
do all this training.
DR. WALLIS: Yeah, but that is a circular argument.
DR. KRESS: I know, but then --
DR. WALLIS: Anything you do is okay.
DR. KRESS: Then you also tell them we keep the CDF below
this number, and we keep the LERF below this number, and that is what I
DR. APOSTOLAKIS: Other things, safety margins. Then your
criteria have nothing to do with the real failures, the design criteria.
And all these things, you have multiple, successive barriers that a
DR. KRESS: You say all those things.
DR. APOSTOLAKIS: Yeah.
DR. KRESS: It is all adequate protection.
DR. WALLIS: They are means to an end. They are means to an
end. What is the end?
DR. APOSTOLAKIS: Adequate protection.
DR. WALLIS: And how do you know you have got that? You
know what I mean.
MR. MURPHY: Our regulations are not geared to just being
adequate protection, because virtually every regulation I can think of
has been more than that. It has been justified using the backfit rule
and cost benefit range, which means it has been cost beneficial to take
it further on down, if you will, in this three regions than just
whatever that list limit that you were just talking about that might be
part of an adequate protection definition. So mostly we are below that.
This really is an indication as to where to stop on the
DR. APOSTOLAKIS: See, if you follow, though, it just
occurred to me, if you follow my line of thinking, then it seems to me
you can define a limit above which --
DR. KRESS: Much closer to the macro.
DR. APOSTOLAKIS: Above which -- sorry -- there is
inadequate protection, but you cannot use CDF and LERF to define how
safe is safe enough. Because the mere fact that the CDF is maybe 9 --
10 to the minus 5 does not guarantee that this agency will say this is
safe enough, because there are other things that the agency is looking
DR. KRESS: Well, you can define it as being a region below
which you no longer have to do cost benefit. No longer do I have to do
DR. APOSTOLAKIS: If your CDF was all inclusive, right now
we know it isn't, --
DR. KRESS: I would use CDF and LERF. I would use CDF and
LERF. And I would also have in the policy statement that policy is to
have a balance between those two, and I would actually have that as part
of the policy statement. You know, rather than as part of the
subsidiary objective, I would actually incorporate both of those in
there and say there is a policy that we will balance these. Balance, of
course, not being equal, it is being some value of each to meet the
DR. APOSTOLAKIS: I think it is the value of the CDF and
LERF, plus a whole host of other regulations.
DR. KRESS: The presumption is always there that you meet
all the regulations. That presumption is always in there, even with the
safety goals, and that you do all the training and the inspection and
all the other things. You always have that, I have always that
presumption in there.
DR. APOSTOLAKIS: I mean if --
DR. KRESS: If you don't meet those, why you are going to
get -- you are going to get --
DR. APOSTOLAKIS: Right. That doesn't help very much in the
sense that if we have a policy statement that goes along the lines we
are discussing, then the staff would want to use the statement.
DR. KRESS: Oh, I would have no objection to having those
DR. APOSTOLAKIS: But if the other regulations are part of
the statement, a cyclical argument again. You are not supposed to touch
those. And if you want to eliminate some of them, --
DR. KRESS: Yeah, I think that does give you a problem,
DR. APOSTOLAKIS: That gives you --
DR. KRESS: Yeah.
DR. APOSTOLAKIS: It seems to me that inadequate protection
in terms of individual metrics would be easier to define.
DR. KRESS: It would certainly help process a lot.
DR. APOSTOLAKIS: And it would help in risk-informing the
DR. WALLIS: Who is getting the assurance? The assurance is
being given to whoever is being protected. So it seems to me that that
person has to have some say in what is reasonable.
MR. MURPHY: Well, as this is used, this is the legal
requirement, the finding that is made when we license a plant, that
there is reasonable assurance, there is no undue risk to the health and
safety of the public. But the reasonable assurance here is by the
person in the NRC making the decision to take a licensing or regulatory
DR. APOSTOLAKIS: We, ourselves, wrote a letter agreeing
with the certification of AP600. We had reasonable assurance, I
suppose. We had better. In fact, --
DR. WALLIS: Well, it is also a moving target. I mean as
society gets safer, as the other accidents become less likely.
DR. APOSTOLAKIS: That's right.
DR. WALLIS: It is generally happening. Aircraft are safer
and so on, then maybe this is a moving goal.
MR. MURPHY: The safety goal policy statement stated that in
a qualitative term that there should be minimum impact on life and
health, I think. That is interpreted as a tenth of a percent.
DR. KRESS: Unfortunately, that tenth of a percent is a
moving target because both the accident rates are changing and the
cancer rates are changing.
DR. APOSTOLAKIS: See, I just remembered something. We are
arguing here in terms of the three regions, Joe, thinking about having
in mind CDF and LERF and so on. Maybe that is not the right context,
because now I remember when the U.K. Health & Safety Executive published
their report last year, which I don't know whether it has been adopted,
they gave three regions for a quantity that was independent of the
They said the individual risk from any hazardous activity in
the United Kingdom should be, if it is greater than 10 to the minus 4
for the general public, for a member of the general public, it is
unacceptable. And if it is less than 10 to the minus 6, it is in a
region where it is broadly acceptable, and in between you have this cost
So if you define now the high level goals like individual
probability of death or some societal measure, I think they use 50 or
more deaths, you free yourself from issues of adequate protection is the
combination of all the regulations we have, plus core damage frequency
and so on, because this is now a very high level policy statement, it
refers to individual risk.
DR. KRESS: I think we have that already in the qualitative.
DR. APOSTOLAKIS: But it is, again, a goal. It is a goal
and a single value. It doesn't tell you what is acceptable or
unacceptable, clearly unacceptable.
DR. KRESS: Well, we have it as a goal, we don't have it as
a limit yet.
DR. APOSTOLAKIS: But I think we are downplaying that
because we are not really -- we are going in a direction that does not
utilize the health effects. We are using CDF and LERF.
DR. KRESS: Well, I think -- you and I disagree a little
there. I think LERF is a good surrogate for health effects. I have no
problem with that as a surrogate.
DR. APOSTOLAKIS: Any kind of health effects?
DR. KRESS: The one we have is a good surrogate for early
fatalities. It is not a good surrogate yet for land contamination.
DR. APOSTOLAKIS: Right.
DR. KRESS: It is not a good surrogate for -- it can be, but
just have a different value.
DR. APOSTOLAKIS: That's right.
DR. KRESS: So it is a good surrogate, and I have no problem
using -- in fact, I think it is a good thing to use because it focuses
on design issues as opposed to siting issues, which you can deal with
elsewhere, and then emergency response issues. But I do --
DR. APOSTOLAKIS: But LERF can be a policy statement, can't
DR. KRESS: I think a policy statement --
DR. APOSTOLAKIS: A surrogate.
DR. KRESS: I have no problem with using the high level
goals as they are, and then, as a subsidiary, saying that these high
level goals can be achieved by a proper balance between LERF and CDF,
where LERF is a surrogate for them. I have no problem with doing that
way, as long as it is in there somewhere as guidance.
DR. APOSTOLAKIS: But when you say appropriate balance, then
how would you define the three regions -- the two values?
DR. KRESS: I would have three regions on LERF and three
regions on CDF, each of them -- each of them would have a policy
objective associated with it.
DR. APOSTOLAKIS: So each one would have an unacceptable
DR. KRESS: Yeah, and they would be consistent. You start
with LERF, CDF is incorporated in LERF. You put three regions on LERF
and then you say, what balance do I want to have now between CDF and
LERF? And you start with one and then you make the other regions
consistent with it. But it is perfectly reasonable to do it that way.
DR. APOSTOLAKIS: I understand. At least we can try.
DR. KRESS: And, in fact, you would tie that then to your --
this is, in essence, a definition of defense-in-depth with respect to
quantifiable uncertainty. And I like the way they presented, I think
Tom King presented a look at this balance, plus looking at individual
sequences to see if there was a balance there. And that, to me, ought
to be a regulatory policy, a regulatory objective, and it ought to be
part of the policy statement. Then you have something to work to.
DR. APOSTOLAKIS: Well, maybe that is what Joe means over
there, as experience is gained. Maybe after the initial --
DR. KRESS: That may be. Yeah, I am assuming that is what
that means. So I don't know whether the time is ripe now, or they need
to think about it some more and do it later or not.
MR. MURPHY: I think the point that Dr. Apostolakis made is
very good, that by recognizing that adequate protection or reasonable
assurance and adequate protection has many more things to be considered
that just quantitative risk analysis. Can we use our experience, as we
try to risk-inform things, as we look at the past analyses that have
been done, can we use this in some way to come to a better definition of
at least a portion of what contributes to that big thing called
reasonable assurance of adequate protection? I can see some merit in
doing something like that.
I have a feeling this presentation is getting away from me.
DR. KRESS: Sorry about that.
MR. MURPHY: What I had on a slide that I don't want to talk
about where somebody says something fast is generalized versions of the
five principles. I think they flow -- no, they don't flow very --
DR. APOSTOLAKIS: Generalized versions means wordsmithing
MR. MURPHY: I mean these words.
DR. APOSTOLAKIS: Okay. Good. So we may want to change
some of the words. Now, plant performance should be monitored. Is that
what the oversight process is supposed to do?
MR. MURPHY: Yeah.
DR. APOSTOLAKIS: Do you have any opinion as to what the
objective of this oversight process is that we are monitoring? What are
we trying to preserve?
MR. MURPHY: Trying to preserve? I think you are trying to
find out what is happening.
DR. APOSTOLAKIS: I mean why are we --
MR. MURPHY: I think you are trying to find out what is
DR. APOSTOLAKIS: What is happening meaning?
MR. MURPHY: When you go to use -- what this says is that if
it is possible to come up with a rule and make this a part of the safety
goal policy statement, that if you state something, you are going to use
this policy statement to develop a regulation, to handle an area, you
would prefer to have it such that there is a way of tracking performance
against that rule. It is performance-based regulation.
DR. WALLIS: It is a reality check.
MR. MURPHY: Yeah. It is just a call for performance-based
MR. HOLAHAN: I think the other thing that we sort of
skipped of is these are generalized principles, you know, from the
versions expressed in Reg. Guide 1.174. In the Reg. Guide the call for
performance monitoring is for the licensee to do the monitoring. Okay.
The purpose of the licensee doing the monitoring is to assure that the
assumptions made in the analysis are still, you know, verified to the
extent that you can. And then the staff's oversight process is to see,
in fact, that those things are taking place.
But most of the monitoring that we think about is the things
that the licensee does, not the things that the staff does.
DR. KRESS: I don't -- yeah, I don't see Principle Number 4
as being the same animal as the other principles. It is a different
MR. MURPHY: No. It is.
DR. KRESS: And I wouldn't have it in my principles. I
would have something like an acceptable level of risk will be
maintained, and an acceptable balance will be maintained between
prevention and mitigation. Those are principles that, you know, I would
MR. MURPHY: That is a good suggestion.
DR. KRESS: Yeah, and I get rid of Number 4.
DR. APOSTOLAKIS: These are principles for changes.
DR. KRESS: Yeah.
MR. MURPHY: Yeah.
MR. HOLAHAN: It is a principle for change. As a matter of
fact, in Reg. Guide 1.174, part of the argument about why we should
control, you know, the size of changes is that you want to maintain
some, you know, some balance, that you don't want the whole 99 percent
of the risk to be associated with one kind of issue.
DR. KRESS: And I would have something -- words in there.
MR. MURPHY: Yeah, I think that is a good suggestion.
DR. WALLIS: You have principles and you have regulations,
so that they should be enforced, that is -- this doesn't have to be a
principle, it is just --
MR. MURPHY: No, again, these are principles in a policy
statement that is intended to set up --
DR. WALLIS: You don't have to have a surrogate to say we
will have regulations and we will make sure they are enforced, that is
DR. APOSTOLAKIS: But this tells you, though, Graham, that
you have no defense-in-depth, for instance.
DR. WALLIS: Well, that is all right. But this other thing
about the balance between regulations and the last one is that you check
that they really do it, that is so obvious. Otherwise, that is
implementation of a principle, it is not a principle.
DR. KRESS: These are principles of appropriate regulation
or something. I don't know what the title.
DR. WALLIS: It is something else.
DR. KRESS: What these principles are.
DR. WALLIS: It is way far from a safety goal.
DR. APOSTOLAKIS: I understand the staff is revising 1.174.
Are you revising, updating 1.174?
MR. HOLAHAN: There are a couple of areas in which we have
committed to update 1.174, but they are not major changes. Although, I
can see you are tempted to wordsmith the document.
DR. APOSTOLAKIS: No, but the first --
MR. HOLAHAN: Can I quote you on that?
DR. APOSTOLAKIS: The first principle, though, Gary, I sort
of agree with Dr. Wallis, it is kind of an obvious to say it is my
principle that the licensees will comply with the regulation.
MR. HOLAHAN: Well, if you quote the whole principle as it
is in 1.174, what it says is you either meet the regulation or you use a
DR. APOSTOLAKIS: Yeah, right.
MR. HOLAHAN: -- like the exemption process or a proposed
rule change in order to assure that wherever you are going, you will
continue to meet the regulations.
DR. KRESS: Yeah, but that doesn't --
DR. APOSTOLAKIS: That is sort of --
DR. KRESS: Yeah, that doesn't translate well to the
DR. APOSTOLAKIS: We discussed that in the past.
MR. MURPHY: Well, let me get to what I thought was going to
be the controversial part.
DR. KRESS: It probably will be.
MR. MURPHY: The treatment of core damage frequency is a
fundamental goal. In your May 11, '98 letter, the ACRS recommended that
-- I thought you had recommended that core damage frequency be elevated
as a fundamental goal, but when I went back and read your letter
carefully, I found that your recommendation --
DR. APOSTOLAKIS: Good idea.
MR. MURPHY: Yes, it is. Was that the elevation as a
fundamental goal be scrutinized.
DR. APOSTOLAKIS: You know how many hours we spend here over
MR. MURPHY: Yes. And I think the difference between those
words is significant.
DR. APOSTOLAKIS: You should come here on Saturday.
MR. MURPHY: So I think we have scrutinized it. We also
have done something else that I would recommend, and let's go back and
read the '86 policy statement. It is an excellent piece of work. It
has a lot of things in it. It is very forward-thinking for its time,
amazingly so when you look back at it from this standpoint.
It has the following statement in it in terms of core damage
frequency, that the Commission has as its objective providing reasonable
assurance while giving appropriate consideration to the uncertainties
involved that a severe core damage accident will not occur at a U.S.
nuclear power plant.
Rather than try to raise a frequency as a fundamental goal,
I think it would be better to take this word, with some editing to get
the words so that they fit into the body of the text better, but get
this thought as a qualitative goal, and retain the 10 to the minus 4 CDF
as a subsidiary objective.
DR. KRESS: I wouldn't object to that, except I still think
you need, in a risk-informed world, limits. And when it becomes a goal,
it is a type of limit, but it is not the type of limit I think you need.
MR. MURPHY: No, I --
DR. KRESS: So I think you need to say 10 to the minus 4 is
the goal. The limit is, even as a subsidiary, I don't mind where it
shows up, as long as it shows up somewhere, as a limit you have some
other number which --
MR. MURPHY: My feeling is -- we don't disagree in
principle. My feeling is that we need the goal right now, the lower
line, if you will. Do we need a limit? Yes. But I personally think it
is premature to do it.
DR. APOSTOLAKIS: But what you are saying, I thought, Joe,
was that you don't want the number to be in the policy statement. Can
we accommodate what Dr. Kress wants by putting it in a lower level
DR. KRESS: Reg. Guide or something? In a Reg. Guide.
DR. APOSTOLAKIS: Yeah. Because then you can change that
DR. KRESS: That is why I was asking you about the influence
of policy statements before. I think as long as it has the force of
guiding the regulations, I don't care whether it is a policy statement
DR. APOSTOLAKIS: Yeah. But I tend to agree with you. I
don't -- I think you miscalculated, this is not a controversial issue.
I mean if the Commission has a statement, which I must admit I don't
remember, maybe changing a few words would probably satisfy the original
intent. But we can also state some -- give some numbers somewhere else.
DR. KRESS: But that statement is not in there as a primary
goal, it is still a subsidiary, even the qualitative one.
MR. MURPHY: Well, as it is in the goal now, it is a
paragraph, in the writing it is not called a goal or anything,
basically, all it does is elevate that.
DR. KRESS: Yeah, these guys are proposing to elevate that
statement, which would -- to me, is probably a good source.
DR. APOSTOLAKIS: I think it is fine.
MR. MURPHY: And then keep the 10 to minus 4 as a --
DR. KRESS: As a subsidiary.
MR. MURPHY: As I will get to later, to do this, I believe
that it has to be coupled with a subsidiary goal in LERF.
DR. APOSTOLAKIS: Yes.
MR. MURPHY: And I will get to that in a minute.
DR. WALLIS: I rather like this qualitative goal, too. It
goes back to what Gary was saying, you know, I don't think the question
of independence is quite right, but you can make statements which are
qualitative, which then have to be interpreted, and that interpretation
may vary from year to year as you know more.
MR. MURPHY: Yes.
DR. WALLIS: So you can change the lower level stuff. But
you are still meeting your goal because it is still valid.
MR. MURPHY: Yeah. The treatment of uncertainties.
Uncertainties are right now discussed at some length in the policy
statement. It is more than most of us remember, I think, where we
thought that it was a discussion that said use mean values and that was
it. In fact, there is much more than that in the policy statement, but
I think it needs to be updated to include the discussion of
uncertainties that are in the guidance there provided in Reg. Guide
1.174, effectively bring the discussion of uncertainties up to the state
of the art.
DR. APOSTOLAKIS: Now, an interesting question here is when
the Commission selected this approach of 1/10th of 1 percent, why did
they do that? Did they do it -- first of all, I think that is true that
they wanted the contribution to risk from nuclear power to be small, but
small may mean, you know, 1/10th, not necessarily 1/10th of 1 percent.
Was the reason they chose that 1/10th of 1 percent, interesting enough,
is a number that appears in the policy statement? I thought we tried to
avoid numbers, but this is a number.
DR. KRESS: That is the number in there, yeah.
DR. APOSTOLAKIS: But, anyway, is the reason why they chose
such a small number, I guess 1/10th of 1 percent sounds better than
one-thousandth, because they knew that there were a lot of uncertainties
on the assessment side?
This has nothing to do with our ability to estimate core
damage frequency -- I mean how do we know that, is it stated somewhere?
DR. SEALE: It has to do with the fact that to one
significant figure a person lives to be a hundred years old and then he
DR. APOSTOLAKIS: Yeah.
DR. SEALE: And that the risk from nuclear power should be
about 10 percent of the cumulative risks from everything else.
DR. WALLIS: 1/10th of a percent.
DR. SEALE: 10 percent.
DR. WALLIS: Oh, you mean taking a hundred. 10 percent is a
MR. MURPHY: There was a study --
DR. SEALE: But 10 percent of --
DR. APOSTOLAKIS: Wait a minute, if it is 1/10th of 1
percent per year, why is it more for a hundred years?
DR. WALLIS: No, that is a bogus argument. This is from
accidents, too. I mean you die from old age, that is not an accident.
DR. SEALE: I know, but it is essentially the risk of
nuclear power is --
DR. APOSTOLAKIS: 1/10th of 1 percent.
DR. SEALE: 1/10th of 1 percent.
DR. WALLIS: I think this is a political, I think OSHA does
the same -- I think OSHA has a tenth of 1 percent. It is a political
thing. OSHA has the same --
DR. APOSTOLAKIS: When they selected that number, were they
DR. WALLIS: It is politically acceptable.
DR. APOSTOLAKIS: -- going to allow for the fact that there
are uncertainties in the assessment.
DR. WALLIS: No, it is politically acceptable is what it is.
DR. APOSTOLAKIS: How do you know?
MR. SIEBER: I think there is some substance to that. There
was a paper written in the 1970s, a doctoral thesis at MIT, which you
may be able to find, that establishes that number for risks incurred
that come from outside forces where the participant can't see or
anticipate it, it is one in a thousand. But it is a good paper and it
has some basis.
CHAIRMAN POWERS: From MIT and it is a good paper.
MR. BARTON: That is not an oxymoron.
CHAIRMAN POWERS: I didn't say that.
DR. APOSTOLAKIS: A very pleasant surprise to see that some
people do read those papers.
DR. KRESS: But, basically, it was a consensus agreement
that that is --
DR. SEALE: Yeah.
DR. APOSTOLAKIS: You guys still don't understand -- answer
my question. I understand it is consensus. But is it -- I mean if the
Commission and the community were convinced that the estimates of health
effects from PRAs were with high confidence, would they still choose
1/10th of 1 percent? This is critical.
CHAIRMAN POWERS: It is certainly my understanding that the
health effects from PRAs are trips and falls, because of the large mass.
DR. KRESS: I am glad you showed up, Dana.
DR. SEALE: In principle, George, you don't want to start
arguing about whether it is a factor of 3 or a factor of 2 or whatever,
it is 10 percent or a factor of 10.
DR. APOSTOLAKIS: But the reason why -- I understand that.
The reason why I am raising the issue because if -- is that if the
1/10th of 1 percent was based simply on political reasons and did not
include anything about the assessment, then the whole burden on
quantifying uncertainties is on the assessor.
DR. SEALE: Sure.
DR. APOSTOLAKIS: Because the regulator, the policy maker
has not given me any -- I don't know, relaxation there.
DR. KRESS: I think you have got a legitimate point there,
George, and I think it is a good question. My own personal opinion is
they intended that to be a mean value given what they knew about the
ability to assess the risk. Now, that is an opinion.
DR. APOSTOLAKIS: Oh, that is a very different
DR. KRESS: Yeah, that is an opinion.
MR. MURPHY: Let me try to share a couple of thoughts with
DR. APOSTOLAKIS: Okay, yeah.
MR. MURPHY: It derives from a statement that is in the
policy statement, the real safety goal, the qualitative safe goal, the
Commission's first qualitative safety goal is that the risk from nuclear
power plant operation should not be a significant contributor to a
person's risk of accidental death or injury. I think that is a
statement where uncertainty did not enter into it.
DR. KRESS: Well, you could think uncertainty there, and
what you would do is just say, what is the uncertainty in the average --
in the death rate, normal death rate?
MR. MURPHY: I think -- yeah, but I think when you came down
-- that qualitative statement did not consider uncertainty. When you
pick a number to go with the quantitative health objective, and, yeah,
uncertainty enters into that. And remember that this policy statement
was begun I believe around '76 -- it was '77. It was published as a
draft for comment in '83 and got issued in '86.
There were many, many debates as to whether that meant 1
percent, or a tenth of a percent. I don't remember any other numbers
being debated, but I remember those two numbers being debated at length.
DR. KRESS: On the treatment of uncertainty, your proposal
is to update the discussion that is in 1.174 and make it --
MR. MURPHY: Yeah, what the discussion says now is that it
is important that you understand the uncertainties. That is in the
existing policy statement. It says to use the mean value for a
comparison, but you should calculate a distribution. You should
recognize there are things we have that are not in the distribution, and
where you believe those things are important, you should do sensitivity
studies to try to get some handle in your own mind as to what those
importances are and factor this into the decision process.
And that is, I think there are better ways of getting the
message across now. There is a nice discussion in 1.174 that can be
converted at a high level and put into the policy statement. But I
don't think it will, you know, it is not anything fundamentally
different than what you have heard before, it is just updates. What is
there is actually pretty good.
I almost hate to put this one up.
DR. KRESS: Yeah, because -- okay, go ahead.
MR. MURPHY: Defense-in-depth. The current policy
statement, again, addresses this in some detail. It talks about the
mandate, and that is the word that is in the policy statement, of
maintaining both prevention and mitigation. It is, defense-in-depth is
one of the five principles from Reg. Guide 1.174 that we have talked
about earlier, so we are already talking about that. And, of course,
they note there are ongoing discussions on the subject.
What I propose to do at this point, and this could change
again, depending on whatever the ACRS does in its discussions on
defense-in-depth, is to incorporate the statement on defense-in-depth
that is in the Commission's White Paper into the policy statement or
some, perhaps a shortened version of it. And if you don't remember what
is in the White Paper, that is it.
DR. KRESS: Yeah, I --
MR. MURPHY: This is a direct quote.
DR. KRESS: Yeah.
DR. WALLIS: I like this because it gives you much more of
an idea of how much defense-in-depth you might have if you could
MR. MURPHY: Yeah.
DR. WALLIS: So you are beginning to evaluate it rather than
just making it some kind of a principle.
DR. KRESS: I thought you were going to incorporate the
definition of defense-in-depth that is in the White Paper also.
MR. MURPHY: The definition actually is a footnote.
DR. KRESS: I know, it was a footnote.
MR. MURPHY: I don't know how to make footnotes in
viewgraphs. I'm sorry, that is the problem. But, yeah, I would take
along with it the definition from the White Paper.
DR. KRESS: Okay.
CHAIRMAN POWERS: Unlike the rest of the panel, I have no
enthusiasm for this whatsoever, because I think it does not make clear
in its presentation that a major thinking in the defense-in-depth is
addressing the questions of things that are not in the PRA, and the
possibility that the PRA is itself completely incorrect.
DR. APOSTOLAKIS: Well, to the extent practicable it says.
CHAIRMAN POWERS: But, you see, if I am operating on the
basis of I like defense-in-depth because it is a way of defending myself
against the hubris that I might actually be able to calculate something
real, and then justifying it based on the calculation is undoing me.
DR. APOSTOLAKIS: And I take the opposite view. I think
this sides with you, because it starts with the premise that the
structuralist approach is the one we take and then we use risk to
evaluate some of the elements of the defense-in-depth, and go the other
DR. KRESS: I have a view that combines both your views. I
think there are two kinds of defense-in-depth.
DR. APOSTOLAKIS: What kind?
DR. KRESS: There is the defense-in-depth that one does when
one expresses a regulatory objective that I want balance between
prevention and mitigation.
DR. APOSTOLAKIS: Right.
DR. KRESS: And balance in terms of the contribution to risk
of the various sequences and balance to the uncertainties in various
sequences. We heard that with Tom King the other day. That is one kind
of defense-in-depth and it deals with what you can quantify with a PRA
and it is quantifiable uncertainties and so forth.
Then I think there is another kind of defense-in-depth which
is called -- I don't know what the uncertainties are, or they are
unquantifiable, or they are very -- or they are too big for -- too big
to be acceptable. Then I would put sufficient attention to preventing
initiation, to intervention before things go too far, to providing
diagnosis, and to mitigate the hazard vector, whatever it is. That is
another kind of -- you put attention on all those and that is where you
can't quantify the uncertainty. I think both of those are elements of
defense-in-depth and they ought to be both be part incorporated in the
policy statement, and it handles both your problems if you deal with it
as two things instead of one.
DR. APOSTOLAKIS: It sounds promising but I will have to
think a little bit more about it.
DR. WALLIS: Well, it is difficult, though, because we
always get the question, how much defense-in-depth is enough?
DR. APOSTOLAKIS: See, you don't get that question if you
follow Dana's approach.
DR. KRESS: You don't get with the first part.
DR. APOSTOLAKIS: There is no how much.
DR. WALLIS: I mean you have an infinite amount.
DR. APOSTOLAKIS: Yes.
DR. WALLIS: You can't -- you have got to stop somewhere.
DR. APOSTOLAKIS: Yes.
MR. MURPHY: Do we have enough on defense-in-depth or do you
want to discuss it further?
DR. APOSTOLAKIS: No, I don't think we need to discuss it
DR. KRESS: That is a subject we are going to talk about
MR. MURPHY: Okay. Frequency of a large release of
radioactive material. In the policy statement, the '86 policy
statement, there was a charge from the staff to consider a general
performance guideline of 10 to the minus 6 per reactor year for a large
early -- for a large release of radioactive material, and asked us to
define what that large release was.
We tried several definitions over time, and in 1993 we came
to the conclusion that we were unable to develop an adequate definition
that would fit with the 10 to the minus 6 guideline.
At that time we requested permission from the Commission to
terminate such activities and that permission was granted. However, in
looking at it, as I said, if you are going to have a subsidiary to go on
core damage frequency, it seems that you need a subsidiary goal on LERF
to balance, for defense-in-depth purposes to balance the two. And as
the ACRS has noted, a LERF of 10 to the minus 5 is consistent with the
QHO. It is also consistent with the regulatory analysis guidelines and
with Reg. Guide 1.174.
DR. WALLIS: This is a QHO which is not site-specific, it is
the same factors in the middle of a city or out in the prairie
MR. MURPHY: Well, it is individual risk of 5 times 10 to
the minus 7.
DR. WALLIS: But if someone happens to be on the borders of
the plant or something?
DR. KRESS: No.
DR. WALLIS: Plants have more people on the borders.
MR. MURPHY: It is the -- for the individual risk it is
specified in the policy statement as being the average individual within
one mile of the plant.
DR. WALLIS: One person?
MR. MURPHY: The average individual, yes.
DR. WALLIS: Does it say how many people are there?
MR. MURPHY: It does not talk about societal risk. It is
average, it is individual risk.
DR. WALLIS: Clearly, this is --
DR. KRESS: You calculate the total number of deaths within
one mile and divide by the number of people living in one mile.
DR. APOSTOLAKIS: You are not allowed to say that there are
no people within one mile, so it is really individual risk.
MR. MURPHY: Yeah.
DR. APOSTOLAKIS: It is the same thing as assuming that
there is a guy there all the time.
DR. KRESS: Yeah, absolutely.
DR. WALLIS: It is very different from the goal.
DR. KRESS: It is a little different than saying it is a guy
there all the time. It is saying there is a guy, part of him is here,
and part of him is here, and part of him is here.
MR. MURPHY: Yeah. It is said in terms of the --
DR. KRESS: It is true, because you would calculate it by
the wind rows.
DR. APOSTOLAKIS: Is there a document where the way LERF is
calculated is clearly described?
DR. KRESS: Yes.
DR. APOSTOLAKIS: Which one?
DR. KRESS: Gosh, I forget what the document was. They had
-- I think it was the Brookhaven document, where they calculate LERF.
CHAIRMAN POWERS: Yeah, it is an appendix in 1.174.
DR. KRESS: An appendix.
DR. APOSTOLAKIS: Oh, it is an appendix.
MR. KING: No, it is a reference in 1.174, it is a
reference. There is a NUREG/CR on it.
DR. KRESS: I would not disagree with this position except I
still think eventually you have got to have limits as well as goals for
DR. APOSTOLAKIS: Well, the three region thing is up in the
air, I don't think we agreed on it.
DR. KRESS: The three region, yeah.
MR. MURPHY: Yeah, and clearly all I am talking about right
now is the lower boundary line.
DR. KRESS: The lower, yeah, right.
MR. MURPHY: And, yeah, I think we have talked about the
upper warning, you know, how I feel about it. I think it is a good
idea, I still think it is premature, but we have beat that one to death.
DR. APOSTOLAKIS: It is still not clear to me, Joe, that
these goals are really the boundary, I mean the lower limit, the three
region approach. You may be right but I am not sure, I am convinced.
But my specifying a goal, --
DR. KRESS: I think they certainly are the lower boundary.
I am not sure we arrived at the appropriate and right values for the
lower boundary because I don't --
DR. APOSTOLAKIS: Yeah, 10 to the minus 4 is too high.
DR. KRESS: It may be too high.
DR. APOSTOLAKIS: It is too high.
DR. KRESS: Yeah, I mean, but that is what is in the safety
DR. APOSTOLAKIS: I will tell you what the limit is, the
upper limit is 10 to the minus 3 and the lower 10 to the minus 5.
DR. KRESS: It could very well be.
DR. APOSTOLAKIS: For CDF.
DR. KRESS: I mean I think both of them are open to
DR. APOSTOLAKIS: And they are not risk limits.
MR. MURPHY: Okay. What we are relying on mostly is the
guidance that came out of this June 15, 1990 SRM in terms of how we
would define the use of the existing safety goals. And we are just
trying to take that guidance and put it back into them.
DR. WALLIS: It is interesting to me that all the numbers
you have quoted throughout have always been rounded off to a factor of
MR. MURPHY: I think it is safe to say in most applications
with PRA, my own view is you should --
DR. WALLIS: Also, there is .1 percent, all the numbers seem
MR. MURPHY: All the things should be in the order of --
DR. WALLIS: If we had 11 fingers, it would be different.
MR. MURPHY: Why would that be?
DR. APOSTOLAKIS: I think the world would be different, so
CHAIRMAN POWERS: Actually, I believe that the virtues of
the Babylonian system, a base 60 system, have been frequently cited.
DR. WALLIS: Binary, because then you could be much more
CHAIRMAN POWERS: I think the belief is the base 60 system
is there are so many even divisors in it.
DR. WALLIS: I am not being facetious, it gives us an idea
of the beast we are dealing with, and we are making decisions on a
factor of 10. That is a pretty gross type of factor.
MR. MURPHY: The uncertainties we have and our ability to do
the risk analysis, I don't think a factor of 10 is --
DR. WALLIS: Ten miles, too. I mean --
DR. KRESS: There is a technical basis for the 10 miles,
believe or not, even though it is a rounded off number.
MR. MURPHY: Let me move on to societal risk.
DR. WALLIS: What happens if you go metric, does it become
MR. HOLAHAN: 6.23 kilometers, or is it 16 -- 16, I guess.
MR. MURPHY: The qualitative latent cancer safety goals and
the QHOs are expressed in terms of a fractional impact. It considers
the population within 10 miles of the plant. Initially, that started
out as 50 miles and after public comment on the '83 version of the
safety goals, it was changed to 10 miles. The regulatory analysis
guidelines, on the other hand, considered integrated dose up to 50
The reason for the choice of the 10 miles was that it
focuses attention on the area where the dose is usually the highest. I
am not a health physicist, but I think they use the phrase "critical
population," and so I think this is appropriate.
DR. WALLIS: Well, 10 is really a surrogate for all the
people who were irradiated within a thousand miles of Chernobyl. It
doesn't -- there is no implication that 10 miles is a limit, it is
simply a surrogate for all distances.
MR. MURPHY: Yeah. But, see, what we have done is, in
studies like NUREG-1150, we have looked at the risk as a function of
distance. And there is a knee in the curve that starts at around 8
miles and ends at around 12, if anyone would like --
DR. KRESS: That was a technical basis, I was told.
MR. MURPHY: Yeah.
DR. SEALE: And you have to get to 10 miles before you can
effectively mount any kind of evacuation or before you can do anything.
MR. MURPHY: The regulatory analysis guidelines use 50
miles, those results have a large uncertainty and we are required as
part of the regulatory analysis to consider what the impact of that
I will talk more about this, but the main point I want to
make is that I see no reason to change either one of these two
documents, even though they are not totally consistent in the distance.
The 10 mile zone seems to be appropriate for the safety goals. The
qualitative goal states that the societal risk to life and health should
be no more than -- should not be a significant addition to other
societal risk, so its percentage is roughly appropriate in terms of the
QHO. However, what is left out of that is an overall societal impact.
And we need to consider --
DR. KRESS: Like total number of deaths, for example would
be your measure of --
MR. MURPHY: Person-rem deaths.
DR. KRESS: Yeah, or person-rem.
MR. MURPHY: The overall impact is. But that raises its own
questions. And what we find when we try to think of how to set a
reasonable goal on that is that a significant proportion of the
population dose is calculated, it comes not from cloud patches, but from
ground shine and ingestion.
DR. KRESS: Assuming you don't evacuate. Assuming you don't
MR. MURPHY: Well, that is assuming some portion of the
population evacuates. As I will get to, the calculations, and I am
talking now specifically on NUREG-1150, are based on the EPA protective
action guides. They assume that a significant part of population
evacuates, 99 percent, that those that evacuate, evacuate at a given
speed. It is based on analysis of other evacuations. The 99 percent
itself is based on an analysis of evacuations.
DR. KRESS: Now, those people, the dose to those is from --
MR. MURPHY: Primarily from cloud.
DR. KRESS: Is primarily cloud.
MR. MURPHY: Well, it depends on when they left and when
they didn't. Some of them are able to outrun the cloud and they don't
get anything, if they evacuate early enough.
DR. KRESS: Then they don't get anything.
MR. MURPHY: Right.
DR. KRESS: All right. But once again --
MR. MURPHY: In others, there is a distribution in terms of
who leaves when.
DR. KRESS: It is if you come back. It is if you come back
and don't relocate that you get this ground shine.
MR. MURPHY: Right. Now, in terms of relocation, the
assumption is that if you get a dose, and this is based on the EPA
protective action guidelines, if you get a dose -- if your first year
dose would exceed 2 rem, or any succeeding year would exceed half a rem,
that you would be relocated. That was the assumption that was built in
and that is the assumption in the protective action guidelines.
The key thing about evacuation and relocation is they are
both totally outside the control of the NRC. Those are not NRC
functions, they are functions primarily of the state governments in most
We have an additional problem that the current level 3 PRA
tools have significant weaknesses that limit their utility of
predictions at significant distances from the plant.
DR. APOSTOLAKIS: So now you are allowing the ability of the
assessment tool to do something to have an impact on your goal.
MR. MURPHY: No. I am saying --
DR. APOSTOLAKIS: Well, you are saying I can't calculate it,
therefore I don't need a land contamination.
MR. MURPHY: That is not the conclusion I want you to draw.
DR. APOSTOLAKIS: But why do have the third bullet then?
MR. MURPHY: I want you to understand that the present
techniques are weak. That does not mean do it or don't it.
DR. APOSTOLAKIS: That it the problem, the rule should be
independent of that, should it not?
MR. MURPHY: Yeah. The safety goal, for instance, if you
read the '86 statement, it is clear, as I interpret it, at least, it
applies -- it applies to shutdown conditions. It applies to all -- it
just talks about overall risk. Whether or not we could calculate it at
the time or didn't, didn't matter. It sets a limit. It sets a goal
that we should shoot for.
The same thing in terms of land contamination, it also,
whether we need it or not, should not be particularly affected by the
fact that we -- weak tools. But if we have weak tools, we need to do
something about it, if we think this is important. And so that reason
it is here.
DR. WALLIS: Having weak tools is the biggest justification
for doing research, because if you need those tools, you don't have
MR. MURPHY: You got it. You look at the next viewgraph.
What I want to do, we have considered how to handle this. In light of
the way the safety goal policy statement is structured, in light of the
fact that we derive most of our authorization from the Atomic Energy
Act, which really doesn't address the environment, we would like to add
-- but there are other laws that, of course, do, that influence our
We are recommending that we add a qualitative goal for
protecting the environment.
DR. KRESS: Do you have any idea what that might be at the
MR. MURPHY: I haven't come up with words yet. It would be
not much more than that statement alone. It would be at a very high
level, something like what is in the strategic plan.
DR. KRESS: Now, one of our concerns about societal impact
had to do with the fact that the two goals as they exist now are both
individual risk goals.
MR. MURPHY: Yeah.
DR. KRESS: In the implementation. We were concerned that
there ought to be a goal on either total deaths or land contamination,
one or the other of those. And we considered whether or not total
deaths were incorporated in the regulations anywhere, and they are, of
course, in the siting rules, for one place. The siting rules limit the
population densities and things like that.
But if that is a regulatory objective, and it does show up
in our regulations in a number of places, limiting the total number of
deaths, shouldn't it be in the policy statement as one of the
Commission's policies, to limit the number of total deaths? That could
be a qualitative statement also.
MR. MURPHY: Yeah.
DR. KRESS: But, you know, I was of the feeling you might
want to -- in terms of protecting the environment, that is one thing,
that is a land -- to me, that is a land contamination. I think you
might want to think about a qualitative statement on total deaths also.
MR. MURPHY: I don't have any major objection to it. What I
am concerned about is, do I really want to get siting policy in a safety
goal policy statement?
DR. KRESS: That is a legitimate question.
MR. MURPHY: And that is perhaps the thing that troubles me
the most as I think about it. But should I have -- you know, the
overall impact is something worth considering. As I say, I have a
problem right now, I have a double problem. One is the tools I have are
very weak. Those of you who are familiar with it, the assumption in
NUREG-1150 was that when a puff release occurs, it goes in one direction
DR. KRESS: Absolutely.
MR. MURPHY: The overall impact of that is hard to discuss,
but I know it doesn't represent reality. I know that the wind
persistence data from the United States indicates that there is almost
no place in the United States where the wind persistence in one
direction for six hours is greater than 50 percent. And I know that in
valley sites and river sites, and ocean sites, there tends to be a
predominant flow either up and down a valley or in and out the sea. And
so the wind rows is very particularized in which way it goes. So this
may -- so the plume, instead of going long-way this way, may be going
back and forth. And what the overall effect of that is, whether there
is conservative or non-conservative, quite frankly, I don't know. But I
do know it doesn't model reality.
DR. KRESS: It depends on the wind rows and the population
MR. MURPHY: So we need to do more analysis. And what we
are suggesting is that we do more analysis and we do develop improved
tools, but that has to be done in consideration of the regular
prioritization process we have in our planning and budgeting process.
Beyond that, we can say that we have land contamination
considered already in the regulatory analysis guidelines. That is based
on NUREG-1150, and, as I said, we think those things are -- they are the
best we have, but they are weak.
Overall societal impact, the only question you have is, do
you want to limit it somewhere? And I will give you an example of what
I mean. In NUREG-1150, we have two sets of numbers. We have considered
population dose person-rem out to 50 miles. We have also considered it
to 500 miles. Now, with this meteorological model, I am not sure I
believe anything with that 500 miles. I think the weaknesses in that
are extremely great. But, in fact, half the dose came from greater than
50 miles when you did that calculation.
Now, what does this come from?
DR. KRESS: That dose is not a lot.
MR. MURPHY: Yeah, but what this came from was giving a
large number of people extremely small doses. And, you know, whether
you should credit something like that, or even consider something like
that is something that we need to decide. I think it takes a little
careful determination as to what an appropriate distance for
consideration is, what the critical population is, what you should be
worried about. And so to the extent that you can, although I would
agree with George, a societal question does not derive from the tools
that calculated it, but when you try to set a limit, it seems you would
want to set some -- or a goal, you ought to have a goal that you have
some capability of trying to determine whether you meet it or not.
DR. KRESS: My view of that, Joe, is that NRC should ask
itself the question, should I be concerned about giving a large number
of people a small dose? And small being enough to do some damage, but
maybe not kill them. Of course, they ought to be concerned with that.
The question is, can you develop a LERF, for example, that deals with
early fatalities that already incorporates that goal, how small it has
to be and how many people? Maybe you have already bounded it with the
LERF you have.
MR. MURPHY: Okay. You may need, without thinking this
thing through, you may need, for want of a better word, an ERF.
DR. KRESS: Yeah, an ERF.
MR. MURPHY: Or, you know, at least get the "early" out of
DR. KRESS: Yeah. And my feeling there --
MR. MURPHY: So can consider late releases.
DR. KRESS: My feeling there is in order to judge whether
the LERF you have deals appropriately with things like early deaths,
land contamination, total person-rem out to real far or not, you need
some common metric to compare how much the regulatory agency values not
having those things happen.
You need a loss function for each of them expressed in
dollars some way. It is not easy to do. And loss functions are
generally very subjective things. But you need some way to compare each
of them and say, well, I value this land thing more than I do this,
therefore, it ought to be our LERF goal. Or I value this early
fatalities more and it ought to be our LERF goal.
I suspect when you did that, you would come up with a LERF
on early fatalities as being the one that controls, but I don't know
that because I have never seen the exercise done.
MR. MURPHY: The other thing that -- and we have not
discussed this in-house, so it is a personal opinion, I would like to
see the things we have expressed in things that are under the control of
the NRC. And when I get into all the emergency actions, protective
actions, you know, I am getting beyond it, and that leads me back to the
LERF or whatever, some sort of release guideline, so I tend to agree
with you very much on that.
DR. KRESS: I would put it all in terms of LERF because it
is under your control.
CHAIRMAN POWERS: Can I ask you a question in another
context? NRC has suggested that doses, the cleanup of sites to dose
levels on the order of 20 millirem, all pathways, all sources is
adequate. Can't that give you a good capping on how far out to carry to
your dose dispersion calculations?
MR. MURPHY: I don't think I can't answer your question,
CHAIRMAN POWERS: In another context, Commissioner Diaz has
acquainted me with one of his own assessments and that is that at doses
below 100 millirem, we simply can't distinguish the effects from natural
effects, and that might give you even another capping on how far you
disperse, you carry your dispersal calculations.
MR. MURPHY: So the calculations that we did in 1150 are
based on EPA protective action guides which allow -- say, you would
relocate if you would get less than 2 rem in the first year, and a half
rem per year thereafter, or a half rem in any year thereafter. These
are quite different than the numbers you were just quoting.
CHAIRMAN POWERS: Yes.
MR. MURPHY: What would actually happen --
CHAIRMAN POWERS: I think I am looking at a different
MR. MURPHY: I think you have a different question, but in
terms of what, it would -- how it enters into this, I just had the one
set of data that was calculated using one set of assumptions. Now,
obviously, it is like a yo-yo, you know, if I push down and get low
dose, I get large land contamination. If I get low land contamination,
I get large dose.
CHAIRMAN POWERS: I guess I didn't follow that at all.
MR. MURPHY: What I am saying is if I allow people in -- if
I want to minimize the amount of land that is interdicted by setting a
high goal for that, then I get a higher population dose. If I get a
lower population dose, then the amount of land will go up. We have
picked a point that is based on the EPA protective action guides as we
did our calculation. I don't know what would happen in a real accident.
CHAIRMAN POWERS: Well, I think you are addressing a
different question than I was --
MR. MURPHY: Okay, maybe I didn't understand your question.
CHAIRMAN POWERS: I was really coming to this, do we go out
to 50 miles or 500 miles? And when do we stop, and when do we quit
giving large populations minuscule doses and then imputing from the
linear hypothesis some health hazard? And it seems to me that if you
said I carry it out until I fall down to 25 millirem from all sources --
DR. KRESS: Which may be plant and site-specific. Well,
since it is a dose, it would be, depending on the wind rows and the
CHAIRMAN POWERS: Percent always depends on that.
DR. KRESS: So it wouldn't be one fixed number, it would
depend on the site.
CHAIRMAN POWERS: True. I mean I think that -- certainly,
if I lived next to a plant, I would be happiest if you took your
analysis and considered my site and not somebody else's site, and
whatnot. And that is a way of capping it.
DR. WALLIS: The reality, it seems to me, if you look at the
Chernobyl experience, you can get some evidence for what actually
happened in terms of land contamination, and how many -- for how many
years the sheep in Scotland could not be eaten and things like -- this
is actually a matter of record, not hypothesis. You might use this
reality to get you some kind of a basis for decision making.
CHAIRMAN POWERS: Well, unfortunately, what you have to do
is go back and actually look at the contamination, because European
countries have interdictions of the food supplies on a more restrictive
basis than the NRC has ever considered.
DR. WALLIS: But you could probably translate to the United
CHAIRMAN POWERS: In which case, the sheep would never have
been interdicted in Britain.
DR. KRESS: This would help you get the loss function I was
talking about, you know, how much does it cost you?
CHAIRMAN POWERS: Joe, can you complete this in the next
MR. MURPHY: If I can complete it -- well, it depends how
many questions I get, but --
DR. KRESS: Well, I have got at least one on this one.
MR. MURPHY: Yeah. Temporary changes in risk, the existing
safety goal. This is out of the '86 policy. The statement I quoted
earlier, the Commission's first qualitative safety goal is the risk from
nuclear power plant operation should not be a significant contributor to
a person's risk of accidental death or injury.
We raised a question earlier whether -- how we should
consider temporary changes in risk, as changes from configuration
control and that sort of thing. I think, if we are looking at that
qualitative statement, I think in principle the temporary risks are
DR. KRESS: In principle, but that principle doesn't
translate into anything useful in this case.
MR. MURPHY: Now, taking it from there and trying to get
that into an implementation is going to take some time.
DR. KRESS: Yeah. And in order to do it, I think you need a
cap on the temporary risk, and I will tell you why, even though you have
a statement in there. The total CDF, as you note, is an annualized
average over the lifetime of the plant.
A temporary change is a here and now thing that certainly
adds into that, as you say, but you cannot account for it in your
calculations as CDF because you don't know, it is never accounted for
because you don't know how big it is going to be, how long it is going
to be, or how many of these you are going to have. And the idea would
be, with a cap, is to say, well, I don't want -- I have a got a CDF
calculation that doesn't include it, I don't want these things to add
more than, say, 10 percent more to my CDF. Pick out a number, 10
percent would be a good guess.
Then I look at historical records and maybe if I just look
at how many shutdowns I have and say, I cannot have more than X number,
N number of these temporary spikes because I have only two of them each
shutdown or something. This is just experience. Therefore, I have a
number for how many spikes I expect. I have a CDF for the plant and I
don't want these spikes to add more than 10 percent more to the CDF.
That gives you an integral of the cap DT that you cannot exceed as a
temporary risk, and it is a cap. And I think that is a reasonable way
to approach this, and I think you do need a cap on temporary risk in
order to incorporate it properly into the risk-informed system.
MR. MURPHY: I suspect several of us have various reactions.
Let me try one first and then ask Gary if he has one.
When you do what you said, I don't disagree in principle
with what you said, but recognize that all the spikes aren't up, some of
the spikes are down.
DR. KRESS: I would ignore the down ones.
MR. MURPHY: I wouldn't. I would take the integral and say
if the day to day variation in risk, how well does that -- as actually
happens by looking at the configuration controls, how does that compare
with my average? Then I would look at that and say, are any spikes high
enough that they raise this question that the risk was a significant
contributor to a person's risk as he goes about his daily life? So this
considers the variation of the risk at the plant. The hardest part of
it may be consideration that the individual's risk from other causes
changes on a daily basis, too, and how you factor that kind of thing in.
MR. HOLAHAN: Yeah, I would like to say I agreed with some
of what I heard, but I am not sure I agreed with any of it. I am not
very enthusiastic about having any sorts of limits or goals on temporary
risks. I think that the spikes, ups and downs, need to be included in
the analysis. Okay. To a certain extent we do that now. We include,
you know, unreliability and unavailability of equipment, you know, it is
averaged in the PRA.
The difficulty I see is there is a temptation to take, you
know, the highest spike and compare it to some goal. But I think Joe
said it correctly, you know, remember the safety goal is derived from,
you know, 1/10th of 1 percent of accidents. But the risks of accidents
go up and down. As a matter of fact, the accident risk is dominated by
automobile accidents, automobile fatalities, and those definitely go up
As a matter of fact, right now, sitting on the fourth floor
of this building, I suspect our automobile risk is exceedingly low.
Okay. But it snows sometimes and you go out on the road, obviously, the
risks go up and down. And if you want to control the peaks, you have
got to compare peaks to peaks, okay, and not peaks to averages. I think
it is meaningless to say at one point in time the reactor risk peaked
up, you know, by a factor of 10 and that it would be compared to
something. Well, should it be compared to drunk driving or driving
while you are talking on the cell phone? What do you compare it to?
If you start comparing it to the averaged automobile
fatalities, I think you have -- all of a sudden, you know, doing the
wrong arithmetic. So I think you should put it in the analysis,
calculate the mean values and compare mean values to mean values. And I
think that is taking care of the arithmetic all right.
MR. KING: I kind of like the idea of a cap on risk, but I
don't think you need to change the policy statement to implement such a
thing in a Reg. Guide or anyplace else. So I agree with Joe's.
DR. KRESS: Yeah.
MR. KING: And at that, I don't think we have settled
internally exactly how we are going to deal with changes and risks, but
I do agree, we don't need to do anything to the policy statement to let
us do that.
DR. KRESS: I think this is an issue having to do with risk
management in outages. I think that is where it belongs, in some sort
of rule there. And I agree, it shouldn't -- it doesn't belong in a
MR. MURPHY: Let me share one --
MR. HOLAHAN: I would like to correct my statement. We are
sitting on the second floor, but the automobile risk isn't any higher on
the second floor than it was on the fourth.
MR. MURPHY: Let me just mention, at least three or four
years ago OECD did a study of the use of, for want of a better word,
risk meters, or that type of device in the U.K., and a report was
published. And as that report recalled the results of that,
particularly for the Torness Plant in Scotland, they used a philosophy
that basically said the instantaneous spike that you are talking about,
and then comparing that to the width, that if the spike was a factor of
3, you could stay there one-third of the year. If the spike was a
factor of 10, the maximum time you could stay there was 1/10th of a
year, or 30 days. And if the spike was a factor of a hundred, you could
stay there for no more than three days and they set a limit on a spike
of a hundred.
DR. KRESS: That is almost kind of like -- that is almost
what I was saying.
MR. MURPHY: Yeah. They also set a limit that said here is
your instantaneous PRA -- I mean here is your average PRA, your annual
average, and you take all the spikes, you record all the changes in the
plant as you go along, as you run this device, and at the end you
integrate it, and the integration has to be within a factor of 2 of the
annual average, or a factor of 3 -- a factor of X, I forget the number.
And that was the way that they used it is in terms of setting a goal for
how you would use this system.
And with that, I think I am done, Dana.
DR. KRESS: Well, we thank you, Joe. Unless there are more
DR. WALLIS: I want to know what happens next.
DR. KRESS: Well, we -- our plans I think at this time are
to possibly write a letter in March on this proposal and just basically
tell them what we think about their positions on each one of these
issues. You know, we have expressed some opinions here. We have to
discuss it among ourselves and come to some committee.
DR. WALLIS: Is this something that goes to the Commission
and the Commission will make a decision?
DR. KRESS: It is going to the Commission at the end of
March, I understand.
DR. WALLIS: Does it go out to the public?
MR. MURPHY: We have to give the Commission a paper on
modifications of the safety goal policy statement by the end of March.
DR. WALLIS: Does it go to the public?
DR. KRESS: No, it is going to the Commission.
DR. SEALE: Not at this time.
DR. WALLIS: The Commission will make a decision of what
they think is in the public interest without consulting the public.
MR. MURPHY: We will get the Commission's advice. What we
are calling for is that after we get permission to go forward, we go
change the policy statement. And that draft then would circulate for
DR. WALLIS: It would?
MR. MURPHY: Yes.
DR. SEALE: It is in the Federal Register.
DR. WALLIS: I thought it was.
MR. MURPHY: I would think there would be more than the
Federal Register, we would probably need to have a workshop or two on
DR. SEALE: The first decision is whether or not you want to
open a can of worms.
CHAIRMAN POWERS: At this point I think I am going to bring
this session to a close and we can go off the record.
[Whereupon, at 4:37 p.m., the meeting was recessed to
reconvene at 8:30 a.m., Friday, February 4, 2000.]
Page Last Reviewed/Updated Tuesday, July 12, 2016