463th Meeting - June 3, 1999
UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
MEETING: 463RD ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
U.S. Nuclear Regulatory Commission
11545 Rockville Pike
Conference Room 2B3
Two White Flint North
Thursday, June 3, 1999
The subcommittee met, pursuant to notice, at 8:30 a.m.
DANA A. POWERS, Chairman, ACRS
GEORGE APOSTOLAKIS, Member, ACRS
ROBERT L. SEALE, Member, ACRS
GRAHAM B. WALLIS, Member, ACRS
THOMAS S. KRESS, Member, ACRS
MARIO V. BONACA, Member, ACRS
ROBERT E. UHRIG, Member, ACRS
WILLIAM J. SHACK, Member, ACRS
P R O C E E D I N G S
DR. POWERS: The meeting will now come to order.
This is the second day of the 463rd meeting of the Advisory
Committee on Reactor Safeguards. During today's meeting the Committee
will consider use of averted onsite costs and voluntary initiatives in
regulatory analyses; development of a low-power and shutdown risk
program; strategies for ACRS review of license renewal activities;
options for crediting existing programs for license renewal; proposed
resolution of generic safety issue GSI-165, spring-actuated safety and
relief valve reliability; a report on the joint ACRS/ACNW working group;
proposed ACRS reports.
The meeting is being conducted in accordance with the
provisions of the Federal Advisory Committee Act. Dr. Richard P. Savio
is the designated Federal official for the initial portion of the
We have received no written statements or requests for time
to make oral statements from members of the public regarding today's
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 that they
can be readily heard.
Do Members have any comments they want to make during the
opening part of this meeting?
DR. KRESS: I think our designated Federal official must be
DR. POWERS: We have changed our designated Federal official
from Richard P. Savio to Sam Duraiswamy, and we have -- I'd like to
introduce to the Members Edith Barbley, who is from NMSS, a group that
we have limited interactions with, so she can acquaint us with that
organization. She is on rotational assignment to the ACRS/ACNW staff,
and she's going to be here with us for two months.
Well, Edith, approximately.
MS. BARBLEY: They said 45 to --
DR. POWERS: Forty-five to 60 days. Well, you're very
welcome, and we look forward working with you. And so when Members see
her wandering the halls, don't ask to see her badge.
Edith, you'll be working across from Sam?
MS. BARBLEY: Yes, and David. Thank you.
DR. POWERS: And David. Well, you're very welcome, and I
hope you enjoy your stay here with the ACRS.
We're much better than the ACNW.
With that, I think I'll turn to the first topic on our
agenda, which is the use of averted onsite costs and the issue of
voluntary initiatives in regulatory analysis. Averted onsite cost has
been a theme of interest to the ACRS for some time, I believe.
DR. KRESS: It has. We've talked about it before.
DR. POWERS: Okay. Dr. Kress, if you'll take over this
portion of the meeting, I appreciate it.
DR. KRESS: Okay. Just a little bit of background. When
the Agency does backfit analyses, which it has to do whenever it's
imposing new requirements that are not compliance issues or such, they
do a backfit, and this involves determining the costs to the -- the
total costs of the thing to both NRC and the licensee, as well as the
benefits, and the costs involve -- they've traditionally included what's
called averted onsite costs, and those are costs that mostly accrue to
the licensee for, you know, if you have a -- if you have an accident
because you didn't have this rule or requirement, then you're going to
crap up your system and you have to clean it up and fix it, and you've
got lost power and lost time and down time, and all those things cost.
And NRC has traditionally included those in their backfit analysis for
The benefits include man rems saved by the rule as converted
into dollars, and one of the issues associated with that has been what
do you -- when you talk about calculating the change in release of
fission products as a change in man rems, do you give credit for
voluntary actions that are not under NRC's regulatory control? And so
we'll also hear about that. So it's those two parts of the regulatory
analysis process for backfits we're going to hear.
With that as an introduction, I'll turn it over --
DR. POWERS: Let me ask a couple of quick questions, Tom.
Now the regulation itself speaks to societal costs.
DR. KRESS: Um-hum.
DR. POWERS: And a lot of the debate on averted onsite costs
has been personalized by the licensees. But it really is societal costs
and societal benefits.
DR. KRESS: That's the debate. The licensee claims that
these are costs borne by the licensee himself that are not really
societal, where NRC disputes that, and the licensees also say that this
is inappropriate meddling of NRC in the management issues. These are --
and that besides that they have insurance to cover these, and --
DR. POWERS: But in a -- I mean, I guess there are a couple
of issues that come to mind there. When we do the cost of a backfit we
include in that cost the cost of downtime -- replacement power that you
have to make in downtime.
DR. KRESS: Um-hum.
DR. POWERS: Okay. But now we don't want to include it as a
consideration in the benefits that you derive here?
DR. KRESS: It is included. I mean, traditionally --
DR. POWERS: Traditionally.
DR. KRESS: But that's the dispute.
DR. POWERS: The contention.
DR. KRESS: Yes, the contention.
DR. POWERS: The other thing that I guess I don't understand
very well is that when we compute the benefits in terms of man rems
saved, we monetize that by $2,000 per man rem -- averted.
DR. KRESS: Yes.
DR. POWERS: And that $2,000 per man rem comes I think from
a consideration of fatalities.
DR. KRESS: It does.
DR. POWERS: And --
DR. KRESS: It's derived from the cost of a life.
DR. POWERS: But the societal costs of an accident would
include not only fatalities but the burdens imposed by latent injuries.
DR. KRESS: Yes. That is a subject we ought to take up at
some time. It won't be part of this one, but that is a subject of
interest, where does the $2,000 per man rem come from and does it
actually cover those things as well as -- at one time it was considered
to be already discounted at that value.
DR. POWERS: Um-hum.
DR. KRESS: For future occurrences. But that's changed.
They no longer consider that a discounted value. But that is an issue,
what does the cost actually cover.
DR. POWERS: The other point that comes to mind is this
issue of voluntary initiatives is not confined just to regulatory
analyses. We have considerations, a broader interest in voluntary
DR. KRESS: That's an excellent point. You know, anytime we
do a risk-informed activity, the question's going to come up. So this
has much broader implications, that part of it, than does just for
regulatory analysis. So it's well worth thinking about in those terms.
DR. POWERS: Well, thank you.
DR. KRESS: Those were very good comments, Dana.
DR. POWERS: Just trying to get myself squared away here.
DR. KRESS: You know, my original concern about averted
onsite costs was both whether or not you should include them, but it was
how you included them, because you're talking about if you have an
accident sometime in the future, you may have this cost as some
probability. So you have a future probabilistic cost that you're mixing
up with if you pass this rule it's sure --
DR. POWERS: Yes, there's some --
DR. KRESS: Have a cost of implementing --
DR. POWERS: There are hard dollars and soft dollars.
DR. KRESS: So it's hard dollars and soft dollars, and you
have to handle those correctly. And that was one of my concerns about
this at first. But with that, we'll just turn it over to let you
educate us on this issue.
MR. FELD: I might just mention before I get into the heart
of my presentation that the staff had issued a NUREG that basically
dealt with the reassessment of the dollars per person rem value, and
when we recalculated that value, which is now $2,000, we did take into
account not only fatalities but also injuries and other health effects,
latent health effects. Basically the approach was to try and weight the
-- to apply the conversion factors to a death might be a factor of 1,
but to a fatality it might -- there was a decision made that it was
equivalent to about some fraction of a death, and therefore it was
weighted in that fashion.
DR. POWERS: Okay. Now, I am familiar with a Brookhaven
report that wrestles with the cost of a fatality. Maybe I'm not
familiar with your NUREG.
MR. FELD: Yes. The Brookhaven report was a contractor
study that supported our NUREG. Unfortunately I don't have the NUREG
number right in front of me, but it was published about two years ago,
three years ago.
DR. POWERS: If you could remind me of it, whenever we have
a change, that might be useful for me to look at.
DR. SEALE: So what you're saying is that when you take the
$2,000-per-fatality number, there's actually subsumed into that some
proportionate number of injuries of varying severity, and the offset
costs for those are also included in the $2,000.
MR. FELD: Right. To be more precise, the $2,000 is based
on the -- on a value of life, but when we looked at the expected
consequences of a person rem in terms of health effects --
DR. SEALE: Yes.
MR. FELD: We didn't just account for deaths, we accounted
for latent health effects as well as injuries.
My intent this morning is to discuss two issues, both
pertaining to the NRC's regulatory analysis methodology. The first
relates to our treatment of averted onsite costs in regulatory analyses
and the second concerns the treatment of voluntary initiatives.
I would first like to turn to the treatment of averted
onsite cots. The staff recently prepared a Commission paper which was a
response to a Nuclear Energy Institute letter which requested that the
Commission reassess its treatment of averted onsite costs in the
regulatory analysis. A copy of that Commission paper, a draft was
submitted to this committee about two weeks ago. And, basically, the
conclusion of that paper was that we did not recommend any changes in
I think it would be helpful initially to define our terms.
Averted onsite costs are basically the consequences of averting an
accident that typically accrued directly to the licensee. There are a
number of elements that go into what makes up averted onsite costs.
Typically, it would include replacement power, it would include
decontamination and cleanup costs. It might include certain repairs and
modifications if, in fact, the accident were not sufficiently severe
that there was a likelihood that the reactor could be restarted. It
also could include economic penalties associated with early
decommissioning. And, lastly, it could involve certain administrative
and legal costs that might be associated with a whole series of
litigations or lawsuits that the licensee would face as a result of an
DR. KRESS: I notice with the decommissioning costs, you put
the words "early decommissioning" on there because that is the point.
MR. FELD: You are looking at the differential between what
are the costs if you decommissioned --
DR. KRESS: At the end of life.
MR. FELD: -- it under the normal -- if it had operated to
its completion versus what it would cost if you had to decommission it
DR. KRESS: And that is an essential difference.
MR. FELD: There is a differential there that would be of
DR. KRESS: Is it true that these AOSCs are quite often the
predominant costs in the regulatory analysis?
MR. FELD: It can very well be. Certainly the TMI
experience would suggest that it is. They are trying to place an
economic value on averted onsite costs in some general sense. It is a
little bit difficult, it is going to depend on the severity of the
accident and, even more importantly, it is going to depend on the point
in time at which that accident occurs. The later it occurs in the life
of the reactor, for example, the smaller would be the replacement power
penalty, the smaller would be the early decommissioning penalty, and so
DR. SEALE: In the case of TMI 2, how long after the
accident was the cost of replacement power taken off the table?
MR. FELD: In theory, it should have been taken off the
table -- it should be only accounted for during the period of time that
the TMI reactor was assumed to -- would be assumed to have been
DR. SEALE: Yes.
MR. FELD: In other words, if you look at -- you say if the
TMI unit had 40 years of useful life to it, and the accident occurred in
its 30th year, for example, then the replacement power cost should be --
should capture the incremental costs associated with those 10 years for
which the plant was not available.
DR. SEALE: So it should, in the case of TMI 2, run for 39
MR. FELD: Correct.
DR. SEALE: Wow.
DR. KRESS: And there is still consideration of license
renewal in that?
MR. FELD: And we do account for -- we have looked at the
effects of license renewal on that result. It turns out that those
additional years, because of discounting and present worth
considerations don't weight all that heavily.
DR. APOSTOLAKIS: Now, why do you limit the costs to onsite?
I mean TMI showed that the whole --
DR. KRESS: Oh, it is not limited, but this, the offsites
are not in dispute, it is just the onsite.
DR. APOSTOLAKIS: Oh.
MR. FELD: If you averted an accident, the staff calculates
the reduction in person rem, it calculates the reduction in offsite
property damage, and we say it should also account for the averted
onsite property damage costs as well.
DR. APOSTOLAKIS: No, no, what I meant was that -- I mean
you may have an accident in New York and San Onofre is shut down. Is
that part of --
DR. KRESS: That is not part of it.
DR. POWERS: No.
DR. APOSTOLAKIS: I mean that is what is going to happen.
DR. KRESS: That is an interesting thought.
DR. APOSTOLAKIS: That is exactly what is going to happen.
DR. SEALE: Well, it did happen. The B&W accident, with B&W
reactors that were similar, TMI was shut down for some period of time.
DR. UHRIG: What about the decommissioning costs, does that
take into account the money that the utility does not collect into its
decommissioning fund because it didn't operate the last 10 or 20 years?
Because they would normally charge those, the customers the additional
MR. FELD: As was noted earlier, the focus in a regulatory
analysis is to look at the societal consequences. And because we are
looking at societal consequences, we are not all that interested in who
is actually bearing the cost, whether it is coming from the utility or
from its ratepayers. The reason there is a differential is because,
based on present worth considerations, and perhaps based on the
expectation of whether these costs are going to increase in real terms
or not, there is a differential, when you calculate what the cost, the
actual costs are to decommission in 1999 versus in the year 2009, you
find that there is the difference in the present worth value of that
DR. UHRIG: But there is inflation in those years probably,
and there may be additional regulatory requirements in 2010 versus
DR. KRESS: Yeah, but you can't anticipate those. You just
have to go with what you have today.
DR. UHRIG: I wouldn't bet against them.
DR. KRESS: I don't. You have to go with what you have,
what you know today.
DR. UHRIG: What the existing rules are.
DR. WALLIS: There is this focus on person rem, but, in
fact, at TMI the biggest societal cost was all the disruption by
evacuations and orders that changed, and people didn't know whether they
should leave or not, and there was considerable societal cost associated
with that, without any radiation exposure at all.
MR. FELD: Yes, and that would be captured under an estimate
of the offsite property damages, other offsite damages, yes, that are
non-radiological. And in response to the question that was posed early
about -- what about the fact that other reactors might have to be shut
down because of the accident at a different reactor, that was also an
issue that was looked at. And the position that the staff took based on
an OGC comment was that that would be viewed as speculative and,
therefore, we were directed not to account for those possible secondary
or tertiary effects.
MR. SIEBER: Would you tell us what you used for cost of
MR. FELD: Well, again, the cost of replacement power can
vary depending on the reactor in question and where it is located. But
the staff has expended a great deal of funds in developing estimates of
replacement energy. We have at Oak Ridge developed a very sophisticated
economic dispatch model that looks at the production costs on a system
basis, where it is looking at what costs would be with a particular
reactor operational versus what those costs would be without it being
operational. And the differential between those two cases is what would
basically be the replacement energy cost, and it is based on the actual
fuel, the incremental fuel and operational costs that are associated
with all of the generating units in the system, and it looks at the
economic dispatch of those units as they would occur logically based on
MR. SIEBER: So you do not consider the wholesale power
market, which is generally where replacement power comes from?
MR. FELD: That would be factored. In fact, there may be
certain utility areas, power pools, for which the expectation would be
that they would have to make outside purchases to provide that energy or
that power. In those instances, those would be the basis for the cost
estimate. In other words, we are looking at what is going to provide
the power in lieu of this nuclear unit if it lost, and it depends on the
amount of reserves available in that service area. For some service
areas, you may be expecting greater reliance on outside purchases,
unless you would expect less efficient units to pick up the slack.
DR. KRESS: By the way, as part of the handout package we
received ahead of time, there is handbook on making regulatory analysis.
I forget the NUREG number, but it has a lot of this in it. It is very
DR. BONACA: So also what is allowed to be in the rate
bases, because if you are not in the rate base, of course you have a
direct loss on the initial investment to the capital.
DR. KRESS: It is NUREG/BR-0184.
DR. POWERS: I might just comment, backfit analysis is
sufficiently an arcane subject and comes up so frequently in this
committee that members may want to retain that handbook.
DR. KRESS: Yes, that is why I brought it up.
DR. POWERS: It is very useful.
DR. UHRIG: What is the number again?
DR. KRESS: It is NUREG/BR-0184.
DR. POWERS: When you get home you will find it.
MR. FELD: It is entitled, "Technical Evaluation" --
"Technology Evaluation Handbook."
DR. KRESS: It is "Regulatory Analysis, Technical Evaluation
DR. BONACA: The recovery from rate base, however, there is
plant-specific and depends on -- so you are including those, but you do
not, you can't use a normal computer program to figure this out?
MR. FELD: I am not quite clear on -- if by looking at the
effects on the rate base of the replacement power, if that is the
question, I again would argue that that is not the controlling
consideration in calculating this cost differential, because that,
again, is focusing on who is going to bear the cost, which is not the
fundamental issue from a societal perspective.
DR. BONACA: Okay. Thank you. I understand.
DR. APOSTOLAKIS: So is this a document that the members
carry with them?
DR. SEALE: Some of them, whenever they are in a bad --
DR. APOSTOLAKIS: Everybody seems to have a copy.
DR. KRESS: You got one as part of your review package for
DR. APOSTOLAKIS: So we spent half an hour on this
MR. FELD: All right. Just to finish off on this, on the
averted onsite costs, recognizing that it can vary substantially, I
think it is fair to say that in general we are talking about averted
onsite cost being on the order of several billions of dollars.
With respect to the current policy, the current policy
concerning the treatment of averted onsite costs appears in the
NUREG/BR-0058, which is the NRC's -- the Regulatory Analysis Guidelines
of the U.S. Nuclear Regulatory Commission. These guidelines were
revised and updated in 1995 and in that analysis, in that document,
basically, the position that is taken is that one should include averted
onsite costs in the cost-benefit calculation. However, it goes
on to say that the inclusion of averted on-site costs were to result in
a significantly different overall cost benefit result than the analyst
is obligated to also display the cost benefit results without averted
on-site costs included. In this way, the decision maker is made aware
of the sensitivity of these results to this particular attribute.
And, the decision maker always has the option to attach
whatever weight it deems appropriate to this particular consequence.
DR. SEALE: I don't want to interrupt your train of thought
right now, but as you go along and if you even think about it, I would
be interested in knowing of some specific cases where successful that
term begs definition but successful cost benefit analyses were
conducted, and also a case where, when you conducted that analysis,
there wasn't a significant change in the answer if you didn't include
averted off-site costs. Just think about it, and maybe at the end you
could, all right
MR. FELD: Okay. A fundamental question then is, given this
policy, what is the rationale or the basis for the NRC having adopted
this position. I think there are several important considerations.
The first, and what I believe to be the most important, goes
to the very essence of what a cost benefit analysis is supposed to be.
And I feel that by excluding a particular attribute, one is in direct
conflict with the ultimate objective of a cost benefit analysis. Cost
benefit analysis is designed, by its nature, to require, to bring
together all of the consequences of an action, regardless of how
disparate those consequences might be.
Every effort should be made to identify all of the
consequences, to express them in commensurate units, so the decision
maker can actually compare and understand what the differences are
between these attributes, and to allow the decision maker to make a
decision based on complete knowledge.
The thought that we would ostensibly, arbitrarily,
selectively exclude a specific direct consequence of an action, just, in
my mind, is totally inconsistent with the essence of what a cost benefit
analysis should be.
DR. APOSTOLAKIS: But then I don't understand what
"speculative" means. I mean, the Office of General Counsel says these
costs are "speculative". We are dealing with "speculative" of course,
here, I mean in the sense that there is uncertainty, right?
MR. FELD: That's correct.
DR. APOSTOLAKIS: So it seems that those societal effects
should be included, except those that are declared by the lawyers as not
DR. POWERS: Well I think I mean, I'm not absolutely
positive about this, but I think there's an accounting rule that
specifically addresses this. Collateral legal implications are not
allowed in making economic analyses, usually. That is, third parties
affected by your decision, through no fault of your own, control of your
own you just don't put them in to the accounting equations.
DR. APOSTOLAKIS: This may apply when you are responsible
for what happens. But here, we're talking about societal costs. So
there's a difference.
The NRC claims that all these costs are borne by society,
that it's irrelevant whether Commonwealth Edison pays or the City of
Chicago. So to me, shutting down a hundred nuclear plants is a major
DR. POWERS: But that would be a decision that the local
society made, not since you have no control over whether they make a
rational decision or not, you can't factor on a rational basis.
DR. APOSTOLAKIS: Well, I don't understand the local
society. I thought the NRC would be able to shut down at least until
the Agency finds out what happened. Right?
DR. POWERS: Well, I don't think that you can argue that the
NRC shut down all the hundred nuclear plants in response to the TMI
incident, or the Brown's Ferry incident.
DR. APOSTOLAKIS: But this time there will.
DR. POWERS: You think so?
DR. APOSTOLAKIS: Yeah, I think that is a good --
DR. POWERS: Well you have a confidence --
DR. APOSTOLAKIS: -- at least for a few days.
DR. KRESS: So can you put a probability on it?
DR. POWERS: Yeah, we have to put a probability on that?
DR. KRESS: Because that you have to include the
DR. APOSTOLAKIS: So there's some sort of decision somewhere
that we can put probabilities on certain things and not on others.
DR. KRESS: If you can establish the probability of that,
well maybe it's worthwhile.
DR. POWERS: We could.
DR. BONACA: I think it's pretty high.
DR. APOSTOLAKIS: If it's possible to shut down. I mean,
TMI cost many, many millions of dollars to the whole industry, in part
in necessary backfits, but also in reaction to the event.
DR. BONACA: Sure. Exactly.
DR. APOSTOLAKIS: So there was a huge --
DR. BONACA: It's not just a shutdown, yeah.
DR. APOSTOLAKIS: Oh yeah.
DR. BONACA: Yeah.
DR. SHACK: That probability's very different from the
probability that an accident will occur, that comes out of the PRA.
DR. KRESS: Yes. It's a different --
DR. SHACK: You know, it's a different quality of a
DR. APOSTOLAKIS: Okay, let's take what Mario just said
then. I think you can put some probability on NRC actions and generic
requirements after the accident.
DR. KRESS: But see, you can put a degree of belief
DR. APOSTOLAKIS: Well, it's always a degree of belief for
me, so I have no problem with that.
DR. KRESS: Some of them aren't.
DR. POWERS: You Bayesians. I swear.
DR. APOSTOLAKIS: I just mentioned a shutdown as an example,
but I think Mario's example is very appropriate.
DR. KRESS: George, there is one practical implication of
what you're saying. If you did do that, the backfit room might as well
not be there, because every room you ever want to pass or every
requirement is going to pass it.
DR. APOSTOLAKIS: That would depend on --
DR. KRESS: Pass that part, pass the cost benefit part.
DR. APOSTOLAKIS: It depends on the severity and the impact
on core damage frequency, of the proposed measure. See, that's what
keeps it rational: the probability that you will do, you prevent
something is very low. But if you start putting billions of dollars,
DR. KRESS: Well, you will have to put some probabilities on
those, that's for sure. I don't know how you put those on there.
DR. WALLIS: I think I'd like to get at a much more basic
thing. There are two players here there's the utility and society.
When the society is put at risk by having a nuclear reactor, there are
costs that are borne by that society because the reactor is there. And
the benefit apparently accrues to the utilities.
There are other situations where the cost are borne by the
utility and the benefit accrues to society. I don't know how you make
this balance, but when there are two players there and each side really
needs to make an independent cost benefit analysis -
DR. BONACA: But isn't it true that the genetic implication
of an issue will also have some relevance? What I mean is that TMI
brought up a number of genetic issues that caused, then, the necessary
response to be so expensive to the whole industry. Okay? If the issue
was so specific to the plant, then you would tend to think that you
would only look at the on-site, the regular costs.
But the if the issue was so clearly broadly generic, broadly
generic, then wouldn't you look at, you know, the implication of the
impact? That would make a difference, it seems to me. What I mean is,
again, TMI identified genetic issues and the result of the response was
huge and impacted the whole industry.
DR. KRESS: I think his contention is right. You really
want to look at all societal costs. And those are societal costs,
whether they're internalized. The tradition has been, you do consider
private and internalized costs. I mean, it's, we've that's been
decided over and over and over again, I think, by different agencies.
MR. FELD: I guess my response to that comment would be that
I take exception that there are two bodies in this that there's a
utility and that there's society. There's really only one; it's
society. And utility is a component or an element of that society.
When we capture the societal consequences, we are accounting for
consequences to all parties in society: the rate payers, the public,
DR. WALLIS: So you're lumping it all together, but someone
in the utility trying to make a decision has to make a cost benefit
analysis from their point of view.
MR. FELD: That's exactly exactly. And that, I think, is
really the basic, fundamental conflict between the utility's position
and the NRC's position.
The utility is right. If they say "from our perspective,
that's not a cost or that's not a benefit," well that's true. But the
NRC is not responsible for developing a decision framework that's going
to decide whether it makes sense for a utility to do something or not.
We're concerned with whether it makes sense for society to
do something or not. Our perspective is different. And the fact that
it may not be a cost to the utility is not really that relevant. The
question is, is it a cost to some component of society?
DR. WALLIS: Yeah, but the way to do that, it seems to me,
always, it ought to be reflected in the costs to the utility. If the
utility having a plant there is imposing a risk cost on society outside,
then that should be charged to the utility. The way it's done now, it's
not. It's done through some regulatory lumping of all the costs and
DR. SEALE: Well, it's also rather difficult to calculate
the benefit, when you consider that the real benefit occurred before,
and whether or not the accident ever happened.
DR. WALLIS: The benefit is simply the provision of power
which is brought by the people.
DR. SEALE: Made the power available, that's right.
DR. WALLIS: That's a commercial thing; it has nothing to do
DR. SEALE: But that's the benefit as far as society is
concerned. It had the power.
DR. BONACA: Well, when you do backfits on cars, where you
now install safety belts on all cars, you're not looking at the I'm
just wondering. I mean, there are some parallels there. You're not
looking at the implications of the averted costs on a particular
accident in a car; you're looking at implications of the same issue in
all of the cars.
DR. APOSTOLAKIS: It's a generic thing.
DR. SHACK: But this is generic, too. I mean, when they
look at a regulation, they're looking at the costs on the whole
industry. You know, if it's a BWR, you consider all BWRs, so it is
generic in that sense. You know, your particular case is a little
different one. You know, there's this shutdown. And that would
certainly seem to me to be speculative. Again, you're also constrained
by whatever the lawyers tell you you can do.
MR. FELD: I'd like to comment on the question of shutdown
of other reactors. You know, probably in theory, one could make a case
that one should look at all the consequences of an action, and those
consequences can be direct, or they can be indirect, secondary, tertiary
effects. But the complexity of the analysis really grows when you start
to consider all of these secondary and tertiary effects.
And simply for the economics of doing the cost benefit
analysis, there has to be some cut-off. I think the approach that has
been taken here is to limit ourselves to the direct consequences. So,
for example, when I indicated that I had a problem with excluding a
particular consequence my comment was, the difficulty of excluding a
direct consequence of an action is to me far more egregious than
ignoring tertiary or fourth effects that might occur down the line.
DR. KRESS: You guys have to remember, all this is just
aimed at whether or not NRC should require a new, make a new
requirement. It's not like we're going to basically, that's it. And
there are other things involved in that regulatory analysis.
First there has to be a substantial improvement in safety.
It has to pass that screen. And then there's a safety goal screen that
it has to pass. And if we've got this business of adequate protection
goals, and if it's required for adequate protection, then there's no
cost benefit involved at all. They'd pass it. But if it's already
meeting adequate protection, you're in the range where you don't want to
impose new requirements unless it's cost beneficial.
We're in that narrow range between adequate protection and
safety goals and we're just making decisions this is a decision
process as to whether to pass a new requirement. So it's not that
earth-shaking a problem as to whether to include all these costs; it's
now are we going to have a practical system to screen out or not screen
out all these requirements. That's basically all it is.
DR. WALLIS: I think it's really fundamental. If you looked
at the cost benefit analysis for those sounders for AP-600, where it's
claimed that the cost to society of an AP-600 reactor is $7 a year,
something is obviously wacky. You know, if cost benefit analysis where,
on the one side, there are millions at stake, and the other side there's
a benefit of $7 to society, obviously something is wrong. So something
is wrong with the calculation of the cost, based solely on person rem.
MR. FELD: Shall I continue?
DR. KRESS: Yes.
MR. FELD: So in spite of the fact that I feel it's totally
inconsistent with the cost benefit analysis, we do recognize that this
still remains a controversial issue. As a result of that, as a part of
this exercise, this latest review, the Staff contacted the Office of
Management and Budget to ask its views on this issue.
Basically, the OMB is the Federal agency that is, has direct
responsibility for developing regulatory analysis guidance, and has
oversight and review responsibilities with respect to the development of
analyses by other agencies.
And we basically put the question directly to: what do you
feel regarding averted on-site costs? Their reaction basically is that,
what you're really talking about here are internalized or private
benefits. By that, they meant that these are benefits that are accruing
to the same party that is effectively incurring the costs of that action
or that regulation.
In their view, there's nothing controversial about including
these benefits. In their view, it's standard practice on the part of
cost benefit practitioners to include them. And to be perfectly honest,
I think they were somewhat surprised that this was an issue that was of
concern to the NRC. I think the language they used was, this is
something that goes without saying.
DR. WALLIS: So if you had perfect containment and there's
never any emission of radioactivity whatsoever --
DR. KRESS: It's all on-site.
DR. WALLIS: It's all averted on-site costs, the NRC would
be still regulating, or to try to save the utility from the disaster of
MR. FELD: Well, again, I think as the point was made
earlier, this agency can only impose an incremental burden, a new
requirement on the licensee if it's not involving adequate protection,
if we can demonstrate that there's a substantial improvement in public
health and safety.
That test has to be made before we can even consider costs
and benefit. If it does not contribute to a substantial improvement in
health and safety, then we say the consideration of that action ceases.
We don't get into costs and benefits. We look at that very early on in
the regulatory analysis.
DR. KRESS: In this case, it would never pass that screen.
MR. FELD: Right.
DR. WALLIS: But if the containment is very good and the
cost to society of an accident is very small which it's claimed to be
for AB-600 then it seems to me that all the costs are on-site costs.
I you why you bother to regulate at that point, at all. It's all up to
MR. FELD: Well, I think I'm saying that, you know, we would
find that that would be a regulation that this agency could impose
because it doesn't pass the substantial test.
DR. KRESS: You wouldn't regulate that.
MR. FELD: Right.
DR. KRESS: Wouldn't. Can't. The backfit regulatory
analysis wouldn't let you regulate it.
DR. SEALE: My definition of perfect nuclear industry is one
that doesn't require an NRC.
DR. WALLIS: Well one which has claimed to have a cost of
$7 a year of risk, that's as imperfect as you can get.
DR. APOSTOLAKIS: You have FAA there. What do they do if
there is a generic problem somewhere? Don't they order all the planes
of the same vintage shut down and grounded?
DR. KRESS: Yeah.
DR. APOSTOLAKIS: Is that similar to our shutting down
DR. KRESS: Well, not exactly.
DR. APOSTOLAKIS: Not exactly.
MR. FELD: I guess the case can be made that our reactors
are not --
DR. KRESS: It's more like a compliance issue.
MR. FELD: Each reactor is a unique entity. They have been
designed differently, so maybe the case can be made that where airplanes
are identical; that's not the case with a power reactor.
But basically, we then followed up our discussions with
other agencies by contacting the Federal Aviation Administration,
because of our feeling that they were dealing with a situation that was
highly analogous to ours.
The NRC is responsible for primarily protecting public
health and safety; the Federal Aviation Administration is essentially
responsible for protecting passenger safety. And just as NRC's
regulations can result in a reduction in the damage or loss of property
to the licensees in this case, a power reactor an FAA regulation can
also result in a reduction of loss of damage to its licensees' property
in that case, the airplanes themselves.
So we asked the FAA, what is their position regarding the
loss of airplanes in their calculations of costs and benefits. And once
again, they made the case that this is something that is clearly
considered in their regulatory analyses. It's not controversial. It's
In fact, in their regulatory analysis guidance, they
indicate that there are basically three principal safety benefits that
they must address in their analyses: (1) death; (2) injuries; and (3)
property damage. So, it's an integral part of their analysis, as well.
Lastly, we felt that one of the reasons why we would argue
for averted on-site costs is we feel that its exclusion produces what we
would view as inconsistent or illogical results. That can best be seen
through an illustrative example.
Consider for the moment that the NRC is considering two
different regulatory fixes. One is a mitigated fix, which essentially
is focusing on improving the containment or emergency planning. The
other is a preventive fix that is focusing on reducing the probability
of an accident.
Well, the mitigated fix, by its very nature, is only capable
of reducing the person rem exposure or the off-site property damage. It
cannot effect the on-site costs because it doesn't effect the
probability of that accident occurring.
The preventive fix, on the other hand, can effect the person
rem exposure, the off-site property damage, and the on-site property
So, if we said that the Agency were now, had a policy where
it was not giving any consideration to averted on-site costs, and it
just happened that the costs of these two fixes were identical, and the
person rem saved and the off-site property damaged saved were identical,
then the conclusion that one must draw from this comparison is that this
Agency should be indifferent between these two alternatives.
In my mind, that's clearly an illogical conclusion,
recognizing the fact that by adopting the preventive fix, society also
stands to avert billions of dollars in property damage losses. This is
also our basis or rationale for our maintaining the position we have.
DR. WALLIS: Earlier you said that these averted on-site
costs were the dominant costs, so this would mean that reducing CDF is
far more important on a cost benefit basis than improving a containment,
in the argument you just gave. They have the same effect on public
safety, but they have an enormous effect on averted on-site costs.
Reducing CDF reduces averted on-site costs, where the most cost is.
Therefore, that's what we should do.
MR. FELD: Except not it's not always the case that these
averted on-site costs are going to be dominant. I think there is, you
DR. WALLIS: They seem to be at the moment.
MR. FELD: Well, again it depends on the severity of the
accident; it would depend on when that accident actually occurred over
the life of that reactor, as to how important those consequences would
be, in fact.
MR. ROSENTHAL: Now let me just interject for a moment
please. My name is Jack Rosenthal, and I'm the Branch Chief for
Regulatory Effectiveness Analysis and Fact Branch.
Sid Feld is an economist on the NRC staff, and a member of
that branch, and has long been involved in developing regulatory
guidance. When Sid and I first discussed responding to the paper, we
said, you know, there was this danger that, that it would be a question
of Sid's contention versus any NEI's contention. And most of those
ideas are in fact Sid's.
The point was and I heard somebody use the word
"contention" earlier. The first point is that it's standard economic
practice to consider all the costs and all the benefits. And then cost
benefit starts out with the President's Council of Economic Advisors,
which is that regulation should be net beneficial. It doesn't give you
very much more, but stems from that.
So we went to OMB, we went to FAA, and we tried to develop a
model. So what we're trying to convey is that I believe that we're
providing standard government methodology of how we think the analysis
should be done, you know, consistent with Council of Economic Advisors,
OMB, etc., and not simply a question of one person's views versus
another's. So this slide becomes very important to me.
Okay, while I have interrupted, and it's the ACRS's choice
we're somewhat more than halfway done time-wise and about a third of the
way through the presentation. The second issue is in fact, I think,
more contentious than the first --
DR. KRESS: I'm glad you mentioned that because I think my
preference would be to skip straight to the voluntary initiative part at
this point, and cover it first, and maybe return to this if we have
time. I think there's broader implications of voluntary actions
MR. ROSENTHAL: Perhaps we could Sid, did you have just
some summary statement that you wanted to make? I mean, I think if they
understand on Slide 3, clearly, that NEI came in with a different view.
But do you have some summary statement that you want to make on AOC and
then move on?
MR. FELD: Well, I think that in summary what I would say is
that the industry has consistently maintained that this is not an
appropriate policy. And I think that they provided two essential
One is that it's allowing the Agency to impose regulations
that do not contribute to the public health and safety, but it's
involving the NRC in issues that are internal and concern their own
personal investment and operational decisions, and it's clearly beyond
the NRC's purview.
I think, reading the paper and perhaps even this slide would
suggest that the Staff has problems with both of those arguments for a
number of reasons that I have tried to identify. So in balance, we
don't find any compelling reason to change our policy in response to the
industry's concerns. That would be my conclusion, I guess.
At this point, I'll turn to the discussion of voluntary
initiatives. Here, too, the NRC has been asked to prepare a Commission
paper in response to a Commission SRM, which essentially directed the
Staff to reconsider its position with respect to the treatment of
voluntary initiatives in regulatory analyses, and to identify options or
alternative policies for the Commission's consideration.
A couple of weeks ago, we provided the Committee with a
draft of this Commission paper and I just want to indicate to you that
that paper has undergone a number of revisions. There may be some
inconsistencies between what that paper says and what this discussion
may contain, so I just want to alert you to that.
Once again, it's appropriate to start with a definition.
Basically, voluntary initiatives --
DR. KRESS: Excuse me.
MR. FELD: Sure.
DR. KRESS: If you hit on one of those inconsistencies, you
might want to flag it for us to be sure we recognize it.
MR. FELD: One of those inconsistencies relates to the
well, first of all, we've --
DR. KRESS: You can wait until you get to it if it's in your
MR. FELD: All right. Definitionally, I think basically,
voluntary initiatives are actions that are performed by licensees that,
although not required by NRC regulation, clearly complement the NRC's
The current policy on voluntary initiatives also appears in
the regulatory analysis guidelines in Revision 2. Essentially, what it
states is that for base case calculations, no credit should be given for
voluntary initiatives. However, for sensitivity analysis purposes, full
credit should be provided for voluntary initiatives. That way, the
decision maker, again, is afforded the full sensitivity of these cost
benefit results with respect to this particular attribute.
DR. WALLIS: So credit is in the cost benefit sense.
MR. FELD: Credit is in a cost benefit sense and in the
calculation of risk.
DR. KRESS: The implications of sensitivity analysis in
these is that the decision maker can use that as input and make
judgments as to whether or not, or now much credit to give.
MR. FELD: The decision maker essentially is being told that
as a result of our assumption regarding the future role of voluntary
initiatives, here's what the cost benefit results can be, what the range
of results can be. And the decision maker can --
DR. KRESS: Does he have --
MR. FELD: -- decide, you know, how much weight he thinks
it's appropriate to give to those, in effect.
DR. KRESS: Does he have any guidance on --
MR. FELD: Currently, he does not.
DR. KRESS: When we talk about the decision maker --
MR. FELD: I think we're talking here about the Commission.
DR. KRESS: The Commission itself?
DR. POWERS: Well, I mean, there are different kinds of
decisions, some of which can be made by NRR. I mean, they have to
acknowledge that other people can make decisions here
DR. KRESS: They would be either the EDO or the Commission.
DR. POWERS: Well, actually the director of NRR has certain
MR. FELD: In this description of our policy, I've included
although this does not appear in our current policy I've included
the words "to the extent practicable." And this is a change that we made
in the paper itself.
There was concern expressed that although we say that we
should not give credit to voluntary initiatives in our base case
calculations, we recognize that in certain instances, it may not be
practical to actually not give credit. And those instances would be,
for example, where the risk calculations and cost and benefit
calculations are based on PRA results.
The PRA, which is looking at the as-is state, effectively
includes voluntary initiatives when it calculates the risk. Therefore,
in many instances it may not be possible to modify those PRA results; in
fact, it may be very hard or impossible. And as a result, we may be
unable to show what the risks would be without those voluntary actions.
So although we're saying we're providing the results with no
credit, in certain instances we may not be able to actually do so, and
one should be aware of that.
DR. WALLIS: I don't understand this voluntary bit. If
there's a limit that you're not allowed to go above here, if the
licensee deliberately maintains a margin by staying here, that's
MR. FELD: Maybe it would be helpful to try and explain how
voluntary initiatives are important in our calculation of risk in cost
and benefits. When we calculate the costs and benefits of a regulatory
action, what we're really looking at is the differential in costs and
benefits between two cases.
The first case is, what are the costs and benefits to the
utility? What are the costs and benefits if the regulatory action is
adopted? What are the costs and benefits in the future going to be?
The second state we'd look at is, what would the cost and
benefits be in the future if the regulatory action were not adopted?
And it's the differential between those two states that is the
incremental, or delta, cost and benefit that we're trying to capture in
our in our regulatory analysis. Therefore, what we characterize as that
baseline state, what the cost and benefits would be without that action,
becomes critical in our determination of what the risks are, what risk
benefits are in adopting this action, and what the cost and benefits are
of adopting that action.
So, for example, if we say we're not going to give any
credit to the voluntary initiative, what that effectively means is that
in the base case, in the baseline, the risks are greater and the risks
of adopting the requirement would be greater and the benefits would be
greater, because the baseline has been reduced. We're saying, we're not
assuming they're going to have those --
DR. WALLIS: You're saying I mean, suppose that all the
licensees have adopted some initiative, which then becomes required by a
regulation. According to you, the regulation benefit is zero because
it's already been adopted. Is that right? Or it is huge you give
great credit for a regulation which has absolutely no effect whatsoever?
MR. FELD: Is we're saying we're not going to give credit to
the voluntary initiative, then the benefit of adopting the regulation is
DR. WALLIS: Although it has no effect whatsoever?
MR. FELD: Well, we're saying it does have an effect because
we don't have an assurance that that voluntary initiative will be in
place in the future.
DR. WALLIS: Ah hah.
MR. FELD: It may be in place right now, but what we're
looking at is the cost and benefits in the future.
DR. WALLIS: Okay.
MR. FELD: And if we don't have an assurance that those
actions are going to be available in the future, we're saying we
shouldn't give credit to them.
DR. KRESS: And there may be some inconsistencies in how
it's, a voluntary action is the scope and effectiveness across the --
MR. FELD: Right, and there are a number of problems that
we've identified with voluntary initiatives that we feel make it
compelling to not give credit to them.
DR. POWERS: Do you explicitly or implicitly give a value to
the ability to enforce?
MR. FELD: Enforcement is clearly a consideration in our
determination that voluntary initiatives may not be of equal weight to
regulatory action. And if we can't enforce them, we would argue that
that suggests that maybe it's not appropriate to assume that those
programs will be available in the future.
DR. POWERS: So it's really an on and off situation.
MR. FELD: Right.
DR. POWERS: There's --
DR. KRESS: Normally, what you're looking at is pardon me.
MR. FELD: Sure.
DR. KRESS: You're making a difference. You're calculating
benefits and costs and you're subtracting one from the other to see if
you get a net benefit.
MR. FELD: Right.
DR. KRESS: Do you do an uncertainty analysis on each of
these, or sensitivity at least?
MR. FELD: Yes. As in the case of averted on-site costs,
and as in the case of our treatment of voluntary initiatives, when we an
assumption that we feel that there's a great deal of uncertainty to --
DR. KRESS: You do limit the uncertainties depending on
questionable parts of it.
MR. FELD: Right.
DR. KRESS: Is there a danger of subtracting too-big numbers
to get a small one here, getting that net? The benefits are big; the
costs are big; the difference is maybe small?
DR. POWERS: It's not that they're big; it's that they're
big and uncertain.
DR. KRESS: Yeah, big and uncertain is my point.
MR. ROSENTHAL: We are recommending a change, somewhat of a
change in current policy, which I think you'll find more satisfying. So
if you could get to that, I think you'll be pleased.
MR. FELD: I'll cover this one very quickly.
When we adop4ed the policy we acknowledged that it was
controversial. We went to the ACRS. We went to the CRGR. We discussed
it with the Commission. And these discussions took place in 1992 to
1995 timeframe, when we were developing draft guidelines and a final
And we found that in each case, we found that the Committees
and the Commission supported the position that we were taking with
respect to voluntary initiatives.
DR. KRESS: I recognize those words, the first quote.
DR. APOSTOLAKIS: And there is a comma that's missing there.
DR. KRESS: The word on the first bullet obviously were
DR. WALLIS: That's such a strange word to me. IT seems to
me if the risk to society of CDF is so much that a utility can
voluntarily reduce that CDF by a factor of 10, then there ought to be a
way of getting credit for that. Something has happened which has
reduced the risk to society.
DR. KRESS: Wait for his bottom line.
DR. WALLIS: There ought to be a charge for a CDF, which he
can avert by doing better.
MR. SIEBER: The utility could make it a part of the design
basis of the plant, and under those circumstances I think credit would
MR. FELD: Clearly, I think there are different degrees of
voluntary initiatives -- different characteristics and traits to these
voluntary initiatives that could influence our decision as to whether or
not to give credit or not.
MR. SIEBER: Right.
MR. FELD: And that, in fact, is the thrust of the new
position, or policy that we're recommending to the Commission at this
point in time.
This slide indicates that we made this decision to give
greater weight to this scenario where we weren't going to give credit to
the voluntary initiatives. We did so because we recognize that there
were a number of issues or problems associated with a large number of
Basically, they all came down to the fact that we didn't
feel that we could have confidence or a great assurance that these
voluntary programs would necessarily be available in the future. We
felt that we needed assurance regarding the scope, the duration, the
level of ethic, that these voluntary programs were going to have in the
future, before we could really assume that they were of equal footing to
This basically identifies a number of those concerns with
voluntary initiatives. They're highly discretionary. The nature of the
program can be very vague.
There's clearly non-uniformity across licensees. Some can
be very aggressive in a voluntary program; others can be very lax. Some
licensees might not even be a party to a voluntary initiatives. They
lack enforcement, which was an issue that was an issue that was
identified by one of the Committee members just a moment ago. The fact
that they could be dissipated by the licensee, even without the NRC's
knowledge, was of concern to many.
And then there's the issue of backsliding, which I believe
in the environment of deregulation that's coming upon the electric
utilities, a concern which should be of increasing importance. As cost
competition becomes a more critical consideration, licensees may give
much greater consideration to reducing their commitments to voluntary
actions in order to save money.
Well, then what are the concerns that we have with the
current policy? The first concern is that it appears to run counter to
our direction-setting issue, DSI-13, which the Commission has basically
adopted a policy to promote and encourage the use of voluntary
initiatives in lieu of regulation.
And there's also a concern that it doesn't provide
sufficient incentive to the licensees to adopt voluntary initiatives.
If we're not going to give any credit for them in the regulatory
analysis, why should a licensee even bother to propose or utilize a
A second concern that exists is with respect to the risk
calculation itself. The NRC calculates risks in a wide number of
venues. It calculates risks in PRAs. When it looks at GSIs and USIs.
When it prioritizes generic issues. In all of these venues, we are
effectively giving credit to voluntary initiatives. These programs are
basically looking at the risks in an as-is state.
In a regulatory analysis, we're arguing that we don't want
-- for base-case calculations, we're not going to give credit to these
issues. And therefore, there's an inconsistency in how we're
calculating risk. For some, this is a concern.
The third concern, which I think is a change in the paper
from what you saw, is a recognition that just recently, in fact on May
27 the Commission issued a new SRM to the Staff concerning the use by
industry of voluntary initiatives in the regulatory process.
This SRM is basically a direction to the Staff to begin to
implement and put in place the objectives of DSI-13. It calls upon the
Staff to develop processes and guidelines to enable the Staff, the NRC
to have greater assurance in relying on voluntary initiatives as a
substitute for regulations.
Clearly, to the extent that these guidelines are implemented
and put in place in the future, the position that we currently have
where we're not giving credit to voluntary initiatives becomes a much
more untenable policy. If, in fact, we have guidelines that are going
to give the assurance that these programs are going to be available in
the future, and we can have that greater assurance, then it doesn't make
sense that on the other hand we're not going to give credit for these
As a result of this review, the Staff has identified three
operations, three alternative policies to the current policy.
DR. KRESS: Of papers --
MR. FELD: Yes, and this is a change.
DR. KRESS: You've subsumed two of them into B?
MR. FELD: Actually, the first option that was identified
before has been dropped completely
DR. KRESS: The A option before has been dropped.
MR. FELD: Dropped. And B, C, and D of before, have been
moved up to A, B, and C.
DR. KRESS: I see.
MR. FELD: The first option basically says that we're going
to continue to calculate the costs and benefits based on a no-credit and
full-credit scenario. And effectively then, we're going to try to
capture the full level of uncertainty associated with our assumption
regarding the future role of voluntary initiatives.
But in addition to that, we're going to attempt to develop a
best-estimate scenario. We're going to look at the voluntary
initiatives in question, we're going to look at the specific voluntary
initiatives, and based on their characteristics and traits, we're going
to make a reasoned judgment as to how much weight we should give to
those programs being available in the future.
DR. KRESS: Will there be guidance?
MR. FELD: There will be guidance available. We've
identified, for example, in the paper a number of features that we think
are relevant in this decision. For example, if the cost of that
voluntary initiative are primarily up-front costs that have already been
incurred, and the operating or to-go costs are very small, it stands to
reason that there's a greater likelihood that those voluntary programs
If the voluntary program is relatively non-controversial and
is standard practice on the part of industry and by that, we mean that
it's been in place for a very long period of time and it's fully
supported by all of the licensees again, we think there's a greater
likelihood of that program continuing in the future.
If there are written commitments attached to the voluntary
program, we feel that we may have some leverage to provide enforcement
control over that voluntary initiative. Again, we might give greater
credit to it being available in the future.
Finally, we believe that in the future when the Agency has
developed these guidelines that the Commission has asked for, to give it
the assurance that these programs will be effective and available, to
the extent that the voluntary initiative in question has guidelines that
are applicable to it, again we feel that that would provide the
assurance for us to give the consideration to it being available in the
DR. KRESS: Do you see this guidance having a quantitative
nature in the sense you want to give anywhere from zero to one credit,
and you have a matrix or a decision chart that has these attributes in
MR. FELD: I'm not sure how quantitative we can be, but I
think the intent clearly is that based on the characteristics of that
voluntary initiative, based on guidelines that may be developed in the
future, our feeling is that this best estimate will either be very close
to the zero credit scenario, or very close to the, you know, the 100%
credit scenario, or be somewhere exactly in the middle.
DR. KRESS: It's more like a none-or-all.
MR. FELD: Right.
MR. ROSENTHAL: I'm sorry again, this is very important,
at least in my mind. That is, you're not just simply providing three
numbers, you know, the two bounds and the best estimate, but that the
words that will go with that to the decision maker will likely be more
important than the numbers. That's the attributes that Sid was just
talking about. You know, is a capital expense? Is it ASME, is it an
NEI commitment? Etc., and that in fact, at least in my mind, it's the
text that will carry the day for the decision maker in terms of how much
credit to give.
DR. POWERS: Option B is a little more than what it says
As a decision maker, I would have in front of me a no-credit
case, a full-credit case, and a best-estimate case with a text on where
this best estimate came from.
MR. FELD: Right.
MR. ROSENTHAL: Right.
DR. POWERS: So option B a little richer than it says on
MR. ROSENTHAL: Right.
MR. FELD: And that the zero- and full-credit cases are not
really options that the Staff is proposing as likely. I think it's just
a way of bracketing or enveloping the full range of uncertainty that's
reflective of this consideration.
DR. WALLIS: This credit's a one-shot thing. I mean, you
make a decision and give credit, and then if they take the voluntary
initiative away, you don't have any way of taking the credit away with
MR. ROSENTHAL: Actually I do, and that is for example, if
there's a generic issue on air-operated valves, and we come to some
decision, and sometime later there is operational experience to change
our view. We could reopen that generic issue, go back through the loop
of saying, is there a substantial improvement in safety? Is it an
adequate safety issue or a safety enhancement?
Go right back through the loop, re-do the cost benefit
analysis, and then re-emerge out. That would be driven by operating
DR. WALLIS: Well, the credit could be condition on the
voluntary initiative maintained.
MR. ROSENTHAL: Yes.
DR. WALLIS: So you do have some leverage to see is not just
put in, in order to get some credit and then removed afterward.
MR. ROSENTHAL: Uh huh.
DR. SHACK: But he still has to go through a full regulatory
DR. KRESS: That's right.
DR. SHACK: If you haven't passed a new rule and you decide
you need a new rule, you still have to go through everything you need to
do to get the rule. I mean, it's not as though the credit goes away
pops into place.
DR. KRESS: That's right. You either have a rule or you
DR. SHACK: Right.
DR. WALLIS: But these are bureaucratic details which are
tiresome if all you're looking at is the overall effect. If there's a
voluntary initiative which is good, then, you know, it should happen.
DR. KRESS: I need to look at it. If you're a regulatory
agency, you really want to be sure that everything is in place and
DR. SHACK: I mean, the good thing is that many of these
voluntary initiatives are clearly in everybody's best interests.
DR. KRESS: Yeah, it seems to be.
DR. SHACK: I mean, the NRC has almost given up regulating
water chemistry because the industry standards are so stringent because
it's obviously in their best interests.
DR. KRESS: Yeah, that's the nice thing about that. It's in
everybody's interest. They've got the right idea if they can assure
themselves that it's always consistent --
DR. SHACK: I mean, that's a case where clearly the full
credit is --
DR. KRESS: Yeah, is probably what you would give a full
credit for the assessment.
DR. APOSTOLAKIS: I'm still not clear what "selectively"
means. What between A and B. Selectively including scenario.
MR. FELD: Option B, then, is basically, for all practical
purposes, very similar to Option A. But it states that, effectively, if
we find that by doing the zero-percent and hundred-percent scenarios
that the bottom-line cost benefit conclusion is not changed between that
wide range of uncertainty, then there's no need to do the best-estimate
case because it's obviously not going to effect the bottom-line
DR. APOSTOLAKIS: It's not cost-beneficial to do it.
MR. FELD: So what we're saying is, basically, Option B
appears to be a more efficient way of using NRC's resources in doing
these regulatory analyses.
DR. APOSTOLAKIS: And what is measured credit?
MR. FELD: I think what I mean by measured credit is that
we're looking at the specific voluntary actions that are in question
here, and we're trying to decide how much credit to give to them. And
it's a measured credit. It's based on our understanding of the
characteristics and traits of those specific voluntary actions that are
under consideration. So it's a measured credit; some conscious thought
has been given to how much credit to give to it.
MR. ROSENTHAL: Yeah, apparently at one time people didn't
want to do this because it gives a lot of judgment to the analyst in
terms of how much credit or not to give, although intellectually the
best estimate just seems the best way to go. And that's why I got back
to the question of, what would the text be that accompanied the numbers.
And it's not just a question of the upper and lower bound,
and if the cost benefit changes, then you do a best estimate. But you
need all this text about, is an ASME standard? Is it a capital
expenditure? Etc., to remove the subjectivity from the analyst and give
the decision maker, then, some hard facts to drive the decision.
DR. APOSTOLAKIS: So the decision maker is, as you said, the
commission most of the time?
MR. ROSENTHAL: Uh hmm.
MR. FELD: I'm thinking in terms of rule making and so
DR. APOSTOLAKIS: So they will have, then, to weigh the
three different calculations as to what extent they should decision
MR. FELD: Although the Staff's recommendation would be that
which is consistent with the best estimate. I mean, the best estimate
means to the Staff that it's the most likely result.
MR. ROSENTHAL: Just to be perfectly clear, I mean we think
it's the Staff's obligation to make a recommendation to the Commission.
We just wouldn't simply pass the numbers out.
DR. APOSTOLAKIS: Right. Now, what you said though, Jack,
regarding the voluntary measures, does it come from a commitment to a
society code isn't that part of the measured credit? Or the words
were simply elaborated on? It would give credit to a commitment that
comes from the cytomechanical engineers, then say something that the
utility's doing to encourage, you know, to improve the morality ploys,
which may disappear tomorrow.
By the way, speaking of that, Southern California's use of
the risk monitor has improved the safety culture of the plant. Is that
something that you can include somewhere here?
MR. ROSENTHAL: No.
DR. APOSTOLAKIS: I don't think so, and yet that may have
more significant impact --
DR. POWERS: Understand that this is --
DR. KRESS: -- evaluate safety culture.
DR. POWERS: Understand that these things as I understand
it are very much for generic issues. So specific plant capabilities
are not --
DR. APOSTOLAKIS: So, if the whole industry installed these
monitors, that probably would --
DR. POWERS: Then you might want to start thinking about it.
DR. APOSTOLAKIS: It sounds good to me.
DR. WALLIS: In a way it's experimental. Try something and
see if it works out. And then you can be less or more selective,
depending on your experience. That seems a reasonable step to take. If
it doesn't work out, you can always go back to the old policy.
DR. POWERS: Are you ready to move on?
DR. KRESS: I think we have expired our time. So, you know,
I thank the speaker you could tell, we're very interested in this
subject and it was a very good presentation. We appreciate that.
With that, I think I'll turn it back to you, Mr. Chairman.
DR. POWERS: I think you have something from NEI?
DR. KRESS: Oh, I'm sorry. Did you wish to --
DR. SEALE: Marion.
DR. KRESS: Marion. Thanks for pointing that out to me.
MR. MARION: Good morning. My name is Alex Marion. I'm the
director of programs department. I appreciate the opportunity to chat
with you briefly about the industry's concerns with this longstanding,
The industry and the NRC has been interacting on this for
about the last fifteen, sixteen years. Someone made the comment earlier
about the controversy involved in this and NEI's position and Sid Feld's
position. I never met Sid until today; he seems to be a very nice,
MR. MARION: He is an economist; I am engineer. So there
are obviously going to be some things that we disagree on. But anyway,
be that as it may, I'd like to make just a couple points on both topics.
In terms of regulatory decision making that's called for by
the backfitting rule, 10 C.F.R. 50.109, the costs, the evaluation of the
costs, should focus on the implementation of that regulatory
requirement. And the costs are the direct and indirect costs associated
with licensees implementing the regulatory requirement.
The reason that the NRC has gotten into the more
comprehensive cost benefit analysis, stems from an executive order that
was issued in 1993 calling for all regulatory agencies to do a more
comprehensive economic analysis of their regulations.
NRC as an independent Federal agency was excluded from that
Executive Order. However, there was some communication later on from
the Office of Management and Budget that recommended Federal agencies
I'm sorry, independent Federal agencies to I'm thinking of complying,
because I represent a regulated industry --
DR. SEALE: Comply is all right.
MR. MARION: -- to meet the intent, or something.
DR. POWERS: That's an NRC policy that --
MR. MARION: Do something consistent with the direction
provided in the executive order.
DR. POWERS: The Commission has always had a policy that
says that, unless there's a compelling reason not to comply with these
administrative directives, that they will do so. I mean, they do pretty
MR. MARION: Yes, I know they do. I just wanted to give you
I mentioned that it's been a long-standing controversy. We
fundamentally believe that the NRC regulatory decision making associated
with rulemaking or other regulatory actions should be primarily based on
public health and safety considerations, not financial investments of
utilities or its investors, or the associated economic risks for
operating and maintaining a nuclear power plant.
DR. POWERS: Now, since the first test that they use on any
backfit that they propose is a substantial improvement of safety, it
would seem that they do exactly what you ask for.
MR. MARION: Then why are we even bothering to talk about a
comprehensive cost benefit analysis?
DR. POWERS: Because they have a second test.
MR. MARION: Well, I was interested during Mr. Feld's to
find out if he would identify any specific examples that have been done
recently where the cost benefit analysis has been conducted to support a
regulatory action by the NRC. I'd like to post that question, if I can.
MR. FELD: We've gone back and looked at the effectiveness
One that I was involved with most recently concerned the
station blackout rule. One of the commissioners asked us to go back and
see what the licensees had done, what the actual costs were, how they
compared to what we had estimated, and to try and look at what the
benefits of that action were.
The Commission in that instance was very helpful, in that
they identified what they felt was the change in risk associated with
station blackout. Utilizing our $2,000 person rem value, when we went
back and looked at the cost and benefits that were actually, what we
expected to be resulting from that rule we found that, in effect, it was
cost effective. It did pass that cost benefit test in most instances.
MR. MARION: What would have happened if it didn't make the
MR. FELD: What?
MR. MARION: I'm just posing a rhetorical question.
Okay, fundamentally, regulatory decision making should not
consider economic risk factors associated with the postulated transient
at a plant, or an accident at a plant. That's fundamentally our bottom
line, and I guess we will continue to disagree.
We feel that the Commission should reconsider their policy
on this, and we've glad to hear, very pleased to hear that there's a
Staff paper before the Commission on this. And we're hoping that the
Commission continues to focus on the first criteria, or the first
threshold, which is adequate protection of public health and safety.
One other aspect of this that I kind of feel compelled to
identify is, in light of the utility licensee's use of probabilistic
safety assessment, it seems to me that there's going to be greater
concentration on that first threshold of quantifying the impact of a
regulatory action, whether it be through rulemaking or the imposition of
some regulatory requirement and a generic communication.
But there is a methodology for quantifying the impact in
terms of core damage frequency. If you make that test, then I would
say, why are we continuing to argue and create all this hate and
discontent about comprehensive cost benefit analysis.
Now I would like to move on briefly to the other topic of
voluntary initiatives credited or not credited --
DR. WALLIS: Well, now wait a minute this probabilistic
safety analysis. My claim all along is that everything is cost benefit,
and if you do a public PRA, get a CDF, really there should be some cost
associated with that. I mean, there should be some benefit with
changing CDF. There obviously is a benefit to sum up, and that should
sum into the cost benefit. Everything --
MR. MARION: Right.
DR. WALLIS: Safety is not an independent thing. Safety is
really is cost benefit and really should be interpreted in that way,
somehow. That's my personal view.
MR. MARION: I'm not disagreeing with what you're saying.
I'm just suggesting that there's a better tool available to everybody to
really quantify the relationship of actions to core damage frequency.
And the issue of what cost you put on that is a separate question.
DR. WALLIS: The reason sometimes the cost benefit
comparisons look silly is because one hasn't properly evaluated the
MR. MARION: Right, and, you know, our basic point is that
it's not the NRC's statutory responsibility to look at economic risk
DR. POWERS: I continue to disagree on that. If I
understand your words, you're saying, you want to use PRA and decide
whether something needs to be done or not. And if it does, we'll do so
regardless of cost. Is that what you're saying?
MR. MARION: If it reduces core damage frequency and
advances public health and safety.
DR. POWERS: So you, so you would have absolutely no
difficulty at all with the NRC coming in and saying, we're going to
regulate shutdown activities that we have not regulated in the past
because we know that by doing so and ensuring that safety systems are
available, we're going to reduce the core damage frequency by a bunch?
MR. MARION: I'm not familiar with what action the NRC is
proposing in that particular area so I hesitate to give you a yes or no
DR. POWERS: Oh, they wrote a shutdown rule they wrote a
proposed shutdown rule a couple years ago that was pretty extensive.
DR. SEALE: At-random was pretty good too.
DR. POWERS: I think you mean, regret a position that's just
use the PRA.
MR. MARION: No, I'm just saying there's a tool that helps
the NRC make a regulatory decision on that first threshold of
demonstrating adequate protection of public health and safety. We
didn't have that tool ten, fifteen years ago in wide use, as we have it
Voluntary industry initiatives. We think NRC should
encourage licensee actions that enhance safety. Again, we maintain our
firm belief that NRC's primary responsibility's ensuring adequate
Safety enhancements, just as a term, essentially go beyond
the level of protection provided by the current body of regulations.
You need to maintain that distinction.
Utility licensees need the flexibility to make a
determination of where they're going to apply their resources to improve
plant operations, etc. When they do so, and the NRC decides that they
want to pursue a regulatory action that achieves the same objective
that's already been accomplished by the utility licensee, fundamentally
we think credit should be given for that.
Fundamentally, if it's already in if a program, if you
will, is already in place and the results of that program is being
successfully implemented to the point that the objectives are the same,
then why do you have to regulate to it?
What we heard during the discussion is, the answer to the
"why" deals with control, deals with enforcement, deals with backsliding
I was surprised cherrypicking didn't come up today because that always
occurs when we talk about voluntary industry initiatives, whether NRC
takes credit or whether the industry wants to implement it.
But fundamentally, it's kind of troubling that we're looking
at these decision making processing and trying to come up with some kind
of calculus that captures every element of that decision making so that
everybody knows what everyone is doing and why everyone is doing it. As
I'm sure all of you can recognize, there is no such calculus that would
help in dealing with some of the decisions that licensees have to come
to grips with in operating nuclear power plants safely. And there is no
calculus that's going to capture decision making of NRC to support their
role of regulating, inspecting, and enforcing.
But one thing that does help bring the two together is
communication to develop a common understanding of what people are
doing, why they're doing it, and fundamentally what the benefit, what
the objective, what the result is from that action that's being pursued,
either by the utility or by the NRC. And we need to talk about these
more and more.
I was kind of surprised to hear that there's an SRM from the
Commission on the DSI-13 SECY paper. We were asked by the Staff to
review that SECY and meet with the NRC. And that had not occurred.
I'm really troubled about that, especially since NEI is, I
think, the organization in the industry that deals with voluntary
initiatives more so than anyone else, any other entity that I'm aware
of. That's a topic that we're going to have to discuss with the Staff
in further detail, and probably have some discussions with the
Commission. I don't know.
It's troubling that we keep finding this cookbook process,
this effort to continually find a calculus, a methodology that captures
the complexity of decision making. And I submit, you'll never get
Anyway, that completes my comments. Does anyone have any
questions about the two topics I discussed? Okay. Thank you.
DR. KRESS: Thank you.
DR. POWERS: We have covered this topic?
DR. KRESS: I think we've completed it.
DR. POWERS: I thank all the speakers. And I will recess
until 20 after the hour.
DR. POWERS: Let's come back into session.
The particular topic we're looking at now is one near and
dear to the hearts of the Committee, that is, the assessment of risk
associated with low-power and shutdown programs. The Committee has
written to the Commission a couple of times on this subject.
Dr. Kress, you are our cognizant Member in this particular
DR. KRESS: Yes.
DR. POWERS: And so I'll turn the meeting to you.
DR. KRESS: Okay. Thank you.
Well, it turns out that -- I don't know if we had anything
to do with it or not, but they are undertaking a program to look at the
low-power and shutdown risk, and I think what we're going to hear today
is merely a status report on the plans for that program and what they've
done so far in the way of a workshop and getting together some
Is that right, Mark?
MR. CUNNINGHAM: That's correct.
DR. KRESS: Okay. I'll turn it over to you, Mark.
MR. CUNNINGHAM: Okay.
Good morning. I'm Mark Cunningham from the Office of
Research at NRC. As Dr. Kress indicated, I'm going to make a
presentation, kind of a status report on the program that we're working
into now on low-power and shutdown research. I should note that I've
got some help from people over here to the left of me: Tom King, the
director of our division; Gareth Parry from NRR; and Erasmia Lois from
the Research staff, who are more familiar with some of the details of
some of the -- in particular the workshop that went on.
Two of the other key players in this are Mary Drouin and
Nathan Siu. Mary is on vacation this week, and Nathan is on travel, so
unfortunately they couldn't be here to help me out for the tough
DR. POWERS: Now Mary is off in the South of France; Nathan
drew a shorter straw.
MR. CUNNINGHAM: Yes. Well, he's in I believe Diablo
Canyon, so that's not necessarily --
DR. POWERS: A very much shorter straw.
MR. CUNNINGHAM: Talking about fire standards, fire PRA
DR. SEALE: Gee, I wonder why.
DR. POWERS: Interesting stuff.
MR. CUNNINGHAM: Anyway, the presentation we've got today
has some background historical information, and then try to give you an
idea of what's in the staff shutdown program, at least in its initial
stages now, summarize some of our current understanding on shutdown
risk, summarize the results of a workshop that we held last -- I guess
in late April, a public workshop about trying to gather information on
what's going on in shutdown risk assessment, give you an idea then of
what we plan to do over the next six months or so in terms of following
up on that.
As many of you are quite aware, previous NRC studies,
previous studies from around the world have indicated that under some
circumstances low-power and shutdown risk can be comparable to
full-power risk. We had sponsored some studies a number of years ago
looking at Surry and Grand Gulf. I think they showed these results, and
many of the results we've seen since that time would tend to provide the
DR. KRESS: If the CDF risk were 1 or 2 times 10 to the 4,
then it's likely when you included shutdown in that also it would be 4
times 10 to the 4?
MR. CUNNINGHAM: We'll come back with some examples, but
that's the type of thing we're talking about; yes. It could be of the
same order, so it would tend to have the effect of doubling, if you
will, in some cases.
DR. WALLIS: You meant 10 to the minus 4?
MR. CUNNINGHAM: Yes, I did.
DR. KRESS: The minus has to be there.
MR. CUNNINGHAM: Yes.
DR. POWERS: I don't know. On some of the localized things
we've got numbers up to 1.
DR. KRESS: You know, I've heard George Apostolakis say
there's no difference between 2 times 10 to the minus 4 and 4 times 10
to the minus 4. Is that right, George?
DR. POWERS: Well, I've certainly heard Professor
Apostolakis argue that there is a certain amount of uncertainty in these
numbers, but I think we make an error in looking at the issues of
shutdown risk strictly in terms of the bottom-line number.
I think that our regulatory implementation of the
risk-informed approach has up till now focused very heavily on
evaluating systems and their importance to risk, and I think that you're
much better off thinking in terms of how does the inclusion of
quantitative assessments of risk during low-power and shutdown
operations affect your classification of systems as either
risk-significant -- high risk significance or low risk significance than
simply asking does it double or -- where does it move you on the
horizontal scale on that marvelous fuzzy line to plot that appeared in
DR. APOSTOLAKIS: Have you heard Professor Apostolakis say
anything you want to say, Mark?
MR. CUNNINGHAM: Not right now.
However, the Committee in general has made at least one
recommendation if not several to the Commission that there's a
substantial research program needed to have a better understanding of
shutdown risk. In response, we have a what I would call at the moment
an approved budget that's of what I call moderate size. The Commission
approved that as part of the FY '99, and we're now looking at the future
years to have such a budget.
The Commission introduced one constraint on that. Back in
the days of the policy issues of Reg Guide 1.174, we raised the issue
that there may be additional guidelines, acceptance guidelines, the
fuzzy lines and things, for shutdown conditions. The Commission told us
that that should not be part of this program, but otherwise did not
particularly constrain the program. They also asked, however, for a
status report on what we're doing in the program. There's a paper
that's going up in parallel with this meeting that provides a similar
perspective in written form to what we're going to talk about today. So
that'll be going up to the Commission here in the next few weeks.
So at any rate, as Dr. Kress noted, we have a beginning or
are developing a low-power and shutdown research program. The overall
objective is to develop an understanding of shutdown risk sufficient to
support the risk-informed decision making that the Agency is making.
Another way to put it is, if you will, is what's needed to make at the
very least an incremental step in terms of guidance and methods that
could help us take a step forward in the context of 1.174 or other
DR. POWERS: When I look at an objective like that, I have
mixed feelings about it, because I say on the one hand if I had to do
the job I love objectives like that, because it looks to me like
anything I do meets the objective. On the other hand, I'm a bit
frustrated in that it does not give me any design-to guidance.
MR. CUNNINGHAM: Yes.
DR. POWERS: And it does not say, for instance, develop an
understanding of LPSD sufficient to find out where you are on the
horizontal axis of the marvelous fuzzy-lined plot in 1.174, which would
be something I could design to.
DR. KRESS: That is the way I would interpret that sentence.
DR. SEALE: It's easy to score, but it's hard to win.
DR. POWERS: Well, yes, I mean, it seems to me that the day
before my performance review I could sit down and do something and come
in and say I satisfy this one.
DR. SEALE: Yes.
DR. POWERS: I may not have done it well, but I've certainly
met the requirements.
Why don't we have objectives that we can design to?
DR. SEALE: Again, I think implicit in this are some of the
things that you've been talking about. The sufficiency issue is, as you
say, it could be -- if we did six months' worth of work, we could make
an incremental improvement over what's in 1.174. If we did six years'
worth, it would be a much larger step beyond 1.174.
What we're struggling with now is our understanding of
shutdown risk is a little out of date, the staff's understanding, and
we're struggling now to see what improvements have been made in the rest
of the world, including the U.S., to give us a better idea of how far we
can reasonably go.
DR. POWERS: Certainly one of the ACRS' comments in their
letter to the Commission was to complain that the scoping studies that
have been done in the past were now a bit out of date.
MR. CUNNINGHAM: Yes.
DR. POWERS: And that the plants were doing different
MR. CUNNINGHAM: Yes. And in a fair measure that's what
we're trying to do now is to sort out what advances have been made
elsewhere in the time since those studies were completed. So that's why
at this point we're a little fuzzy, because we don't know how much may
have been done someplace else that we can take advantage of.
DR. WALLIS: By understanding you mean measures and data,
you don't mean a fuzzy understanding, you mean actually coming up with
numbers that describe this worst -- so we can see what it is and we can
make comparisons with other risks.
MR. CUNNINGHAM: Yes, that's correct.
DR. WALLIS: Scale and so on.
MR. CUNNINGHAM: That's right. I think of this again in the
context of 1.174, where you have a horizontal-axis question and a
vertical-axis question that's fairly quantitative. As we talk about the
-- when we get into the workshop results there's people have suggested
there may be more qualitative ways that we can handle this, but that
remains to be seen --
DR. KRESS: An objective I think this Committee might be
more pleased would be one like to develop the capability to routinely
include low-power and shutdown in the risk assessments for individual
plants and the uncertainty analysis, which would include calculating
importance factors, LERF, CDF, delta CDF, delta LERF. But, you know,
that would be a design-to objective.
MR. CUNNINGHAM: Yes. That's certainly not different than
what we're talking about here, but it is much more crisp.
In order to work towards this objective, we've got a number
of things we have under way. One is that we're trying to assess, as I
said a little bit earlier, both domestic and international information
on what's been done in low-power and shutdown risk analysis and what the
concerns are today. Given that, we intend to perform research
activities, methods development, additional analyses, that sort of
thing, to get at these concerns.
DR. POWERS: In what part of the NRC's activities does
someone come in and say now what is the technology that NRC needs to do
its job the way we want it to do its job? With respect to low-power
MR. CUNNINGHAM: I'm sorry, I missed the first of what you
DR. POWERS: I mean, this program seems to be saying okay,
what is the current state of the art out in the world, the state of
MR. CUNNINGHAM: Um-hum.
DR. POWERS: And is that different than the understanding
that we have now? Better, worse, or something like that?
What I am looking for is who is it that goes out and says,
now, if I want NRC to do its risk-informed regulatory job the way I
would like it to do its job with respect to low power and shutdown, what
is it that it needs to do it?
MR. CUNNINGHAM: Okay.
DR. POWERS: Where is that said, or is that the next step
that I am just not seeing?
MR. CUNNINGHAM: Well, I think that is kind of what would
come out of the second bullet here in terms of the research work. Part
of what you have to do is -- well, what do we have?, is one question,
and then another question is, what do we need?
DR. POWERS: Yes.
MR. CUNNINGHAM: Okay. And I think part of what we are
intending to do here is the needs statement of what is needed, and I
guess, again, we would put it in the context of, what do we need to go
into the next step or the next update, or supplement to 1.174 and say
this is how you can better address the issues of shutdown risk.
DR. POWERS: Okay. So it is just coming.
MR. CUNNINGHAM: Yes.
DR. POWERS: I have just got to be patient here.
MR. CUNNINGHAM: Okay. One of the things we are doing as
part of our research and information gathering phase of this is we have
an international cooperative research group called COOPRA, whatever that
stands for. It is intended to NRC's mechanism for cooperating
throughout the world with people doing PRA research.
There are four working groups to COOPRA. One of them is
focused on shutdown risk. We have a program that Mary is chairing, an
effort -- back up. Mary chairs a working group. It has membership,
from what count was, of like 14 countries trying to look at what have
each of these countries done, what are their plans for using shutdown
risk and doing research.
The idea is that we can go through an identify collaborative
research, if you will, or -- either in the sense of, you know, if
country X does this piece, then we will do one other piece, or more
jointly funded programs or that sort of thing.
But this working group had its first real meeting in January
and they have got another one coming up in the fall. But I expect that
will be a good mechanism for us to have a very good understanding from
the international set of what is going on and what research they are
DR. POWERS: Is the technology for PRA internationalized,
that is, does a Japanese PRA and a Finnish look kind of the same?
MR. CUNNINGHAM: At first glance, yes. Yes. Again, as we
proceed with the program, our idea is that --
DR. APOSTOLAKIS: Mark, isn't it also true, though, that we
tend to pay more attention to uncertainty analysis than some of our
MR. CUNNINGHAM: We, --
DR. APOSTOLAKIS: In the United States.
MR. CUNNINGHAM: People in this room, I think yes.
DR. POWERS: Deftly said, sir.
DR. SEALE: Very, very perceptive.
MR. CUNNINGHAM: I am not sure we are consistent throughout
the United States.
DR. SEALE: Very perceptive.
MR. CUNNINGHAM: Internationally, it depends. Individual
countries do things very differently. Certainly, you see in a number of
countries more of a consideration of the parameter uncertainty and that
sort of thing without pursuing very far the modeling uncertainty, and
that sort of issue.
DR. APOSTOLAKIS: Now, that is going a little beyond the
topic, but are they receptive to risk-informed regulation? I mean how
is 1.174 received by these groups?
MR. CUNNINGHAM: Again, it varies. You see countries that
are developing similar types of approaches. I believe Canada, for
example, has a similar type of process underway. Others are not so
interested in being quantitative and very explicit about it, they are
more qualitative. Some of it is driven by the internal mechanisms of
the country as of how the regulator and the licensee interact with each
other. In many countries it is not as formalized, if you will, as it is
in the U.S. So that to some degree also drives how they see these
DR. APOSTOLAKIS: Right.
MR. CUNNINGHAM: But, certainly, the work we have been doing
over the last few years on 1.174 has been a subject that they have shown
a great -- internationally, they have shown a great deal of interest in.
At any rate, so we see it towards the end of this program
that one of the goals, if you will, would be to develop guidance or see
an update to 1.174 where we can say better how we should treat shutdown,
and maybe the right way to say it is a routine type of thing, routinely
you we would -- you could incorporate shutdown risk by these acceptable
Also, in parallel, we are considering development of a
consensus standard on shutdown risk.
DR. SEALE: That is sort of phase B or C of the present
MR. CUNNINGHAM: Yes. Yes, that's correct.
DR. POWERS: How do you decide when an aspect of PRA
technology has reached sufficient maturity that it is appropriate to
ossify it by creating a standard?
MR. CUNNINGHAM: Well, ossify is the intriguing word in
there, obviously. One of the things that we --
DR. POWERS: It was deliberate, by the way.
MR. CUNNINGHAM: Yes, I am quite sure. One of the concerns
that we have had in development of the ASME standard is that if you
become very prescriptive, you do just that, that you prevent or preclude
-- you show -- there is no incentive at that point for innovation.
DR. SEALE: Discriminating against it.
MR. CUNNINGHAM: Yes, you almost -- you have a bias against
it. Certainly the people working on the ASME standard were aware of
that and were trying to write it not at such a prescriptive level that
it would preclude those things. You know, how you can preclude that
other than trying to set out, if you will, functional requirements for
an analysis and not be very prescriptive is perhaps a good step.
DR. POWERS: I mean the question comes up without the
pejorative "ossify," because I think you did a very good job and the
people doing this standard on operational PRA did a good job in trying
to skirt that ossification, while at the same time setting down what are
One of the dangers, it seems to me, in cooperative groups a
la your COOPRA, is that groups like that are usually more effective at
writing standards than they are at doing research itself. Research is
best done in individual institutions where people work on things. And
so there is a drive to write a standard in an area where I think there
is a limited amount of experience and insight, at least in looking at
the literature in this field. So things haven't -- the pot hasn't been
stirred and inspected to the extent that maybe merits even that limited
kind of standard writing that you have done to date at this point, I
MR. CUNNINGHAM: Yes. Yes.
DR. POWERS: But I don't know what point. It does get to
the point that you say, okay, we are really confident that these few
things you absolutely have to do and we will set those. I don't know
when you can say that.
MR. CUNNINGHAM: You know, we will come back to a little bit
in terms of the discussion from the workshop, because that was a topic
of the workshop, of the need for a standard. So it might be a better
place to talk about this.
But just to be clear, by the way, COOPRA is a research
organization, an organization of researchers, so it is not writing
standards at this point. It is more looking at trying to develop
collaborative, international collaborative research efforts.
DR. APOSTOLAKIS: It is way of exchanging information.
MR. CUNNINGHAM: That's correct, yes. And that is being
worked, and one of the goals of the COOPRA effort is to get the people
talking who were the researchers, as opposed to the managers and other
things. You get it down to the technical level to get people to
communicate at that level and plan at that level.
The next few slides talk about our understanding based on
the Grand Gulf and Surry PRAs a few years ago, and our initial
literature search on shutdown risk to see where we think things are
today. And again a fair amount of the information we've got here is
from our COOPRA partners.
Basically you see a pattern that's emerged time and time
again that core damage frequency is comparable to full-power operation
for some operating states and shutdown conditions.
DR. APOSTOLAKIS: Mark, do you have a mechanism through
which you can get the LPSD PRAs that the industry has done in this
country? This international group doesn't help you with that.
MR. CUNNINGHAM: It doesn't help. That's correct, it
doesn't. Yes, we have mechanisms. Again, I come back to that a little
bit later. We have some plans over the next few months to sit down with
utilities who have done PRAs, shutdown PRAs, and sit down and understand
what they've done, that sort of thing. That's --
DR. APOSTOLAKIS: Okay.
MR. CUNNINGHAM: A future activity as opposed to a past, if
DR. UHRIG: What do you mean by operating states? The
containment is open versus closed?
MR. CUNNINGHAM: Operating state in this context can be
low-power conditions --
DR. UHRIG: Mode 5, Mode 6, Mode 4?
MR. CUNNINGHAM: Yes, that's right, except in PRAs what
we've found to do is a Mode 5 designation isn't necessarily precise
enough because things change even within Mode 5, so instead of, you
know, Modes 1 through 6, you might have low power and shutdown ten or 15
operating states. You know, it varies going down in power, coming up in
power, cold shutdown. Mid-loop operation would be an operating state,
that sort of thing.
DR. UHRIG: Containment open.
MR. CUNNINGHAM: Yes. Well, in each of those you have -- I
think the operating states are more defined, at least in our PRAs in the
past by system configurations on core cooling and inventory control.
Then with each of those you have the question of well, what's the status
of the containment given that, in that mode, and what you see, and I'm
going to talk about some of the results, is the ones that tend to be the
most important operating states from our studies have been ones where
the containment has been open during those modes. I'll come back to
that in a minute.
So at any rate, we have seen that the, if you will, the
annual average core damage frequency is comparable for some parts of
shutdown conditions, as with full power. However, the core damage
frequency varies considerably across operating states, so that mid-loop
operation in PWRs tends to be a fairly high CDF state. Refueling
operations tend to be a very much lower -- have a very much lower CDF
And then the contributors, what's causing the core damage
event, the initiators and that sort of thing, can be quite different.
You see different types of human performance at shutdown, because
there's far fewer procedures, if you will. The redundancy at full power
is lost in many cases in a number of situations in other operating
states, so single failures can be much more important in shutdown
condition than in power operations.
The next two slides show some of the results, just to
elaborate on this point a little bit, from our Grand Gulf and Surry
studies. I'm going to focus on a few of the points in this basically.
What I show here are core damage frequency results and risk
results for early fatality risk and latent cancer fatality risk. This
one is Grand Gulf, and the next one will be Surry. The values in here
are from internal events that are shown on the slide. I want to focus
more on the mean values, which is over here, and make again a couple of
Just focusing on those things for the moment, basically you
see in this case the core damage frequency, the mean core damage
frequency from Grand Gulf for POS 5, which is cold shutdown, is similar
to the core damage frequency estimated in 1150 for full-power
operations. It's a fairly small number for a lot of reasons at Grand
Gulf, but again they're similar numbers. The early fatality and latent
fatality risks actually -- they're all small numbers at Grand Gulf again
for a variety of reasons that are kind of unique to Grand Gulf, but the
POS 5 risk can be somewhat higher than the full-power risk.
Basically this deals with the issues that we were just
talking about on containment status. As POS 5 was analyzed in the Grand
Gulf study a few years ago, when they were in this mode, the suppression
pool was drained and the containment was open. So at least two of the
barriers that you would normally expect at least in full-power operation
to be there to mitigate releases are not there. So you could have
greater consequences, if you will, from a core damage accident.
I'll skip over now to Surry, a similar plot. Again this is
for internal events. I should note that -- let me do this again.
The core damage frequency that I've circled up here, in this
case it shows that for the internal events the core damage frequency is
a good bit lower. I should note that for fire, though, these numbers
don't show fire. The core damage frequency for mid-loop from fire was
about 2 times 10 to the minus 5. So that brings it up to be -- again it
would be comparable, if you will, to full-power operations. Again, you
also tend to see in some cases the somewhat higher risks from mid-loop
Again in this case Surry was analyzed four or five years
ago. The containment in mid-loop operations was not necessarily
buttoned up. So you didn't necessarily have the -- I believe it was the
equipment hatch was the issue. It wasn't necessarily in place during
mid-loop operation. So you had a potential there for having a bypass,
if you will, of the containment.
Again, based on what we've seen from international studies
and other reviews, you see a similar type of pattern for shutdown risk
To turn now to the public workshop that we held to basically
gather information to help us understand where people were in terms of
assessing shutdown risk, what tools they have, what results they have.
So we had three points that we wanted to cover in that workshop that we
held -- it was a public workshop held on April 27.
One is share results on what's been going on in terms of
risk analyses, what information and methods. What information and
methods do we need in order to be able to use this in decision making?
And again, what's an acceptable approach or structure for a standard in
DR. KRESS: What's a reasonable estimate for fraction of a
year that you're in shutdown?
MR. CUNNINGHAM: I'm sorry, what's a --
DR. KRESS: A reasonable estimate of the time, fraction of a
year, that you're in shutdown mode.
MR. CUNNINGHAM: It varies a good bit, I think, especially
going to shorter outages and things, you people strive for 30-day
outages and things like that. So I don't --
DR. KRESS: Could be on the order of one-tenth?
MR. CUNNINGHAM: My first answer would probably be like 10
percent or something like that, and that's planned outages. You also
have unplanned outages and that sort of thing.
DR. KRESS: Do you reduce the CDF full power by one-tenth
when you add in the --
MR. CUNNINGHAM: We should. I don't think we ever do.
DR. KRESS: It's not worth it. It's not that significant.
MR. CUNNINGHAM: Professor Apostolakis has noted that 2 and
4 times 10 to the minus 6 are the same number, so we'll -- in that same
DR. BONACA: Just one point. I mean, typically the highest
risk is in mid-loop operations, so you are looking for refueling outages
MR. CUNNINGHAM: Yes.
DR. BONACA: As the one of concern.
MR. CUNNINGHAM: That's correct.
DR. BONACA: And second, typically you don't refuel every
year, you refuel every --
MR. CUNNINGHAM: Yes.
DR. BONACA: Eighteen months or even 24 months.
MR. CUNNINGHAM: Yes.
DR. BONACA: So I would say it might be 5, 6 percent at the
MR. CUNNINGHAM: Certainly for the PEs where you see
mid-loop is the thing that sticks out.
DR. KRESS: Well within the uncertainty, though.
DR. POWERS: But also understand that the numbers that Mark
showed on those previous charts are all annualized.
MR. CUNNINGHAM: Those are annual averages; that's right.
DR. KRESS: So if I want to know what the risk is during the
period that I'm actually shut down, multiply the numbers.
Yes. I was just saying you're double-counting, though.
When you add shutdown in with the full power, you're double-counting
full power --
DR. POWERS: Not very much.
DR. KRESS: I know. It wasn't really a serious comment.
DR. APOSTOLAKIS: Gareth wants to say something.
MR. PARRY: Actually I don't -- this is Gareth Parry from
the staff. You don't in fact double-count, because the initiating event
frequencies that you calculate are on a per-calendar-year basis and you
can't, for example, scram a reactor if it's not at power. So you get
the right numbers.
DR. WALLIS: You showed us these CDF early fatality risk
tables and so on.
MR. CUNNINGHAM: Yes.
DR. WALLIS: Do you need to do a lot more work before
actually making some preliminary decisions about what to do about these
sorts of numbers? You sort of imply that you need to do more research
to gather more information and so on, make better PRAs. But you've
already used some information in generating these tables.
MR. CUNNINGHAM: That's correct. And the question is -- I
guess there are several questions. There was a particular approach that
was used that -- in those shutdown PRAs that was screened out had a
first filter, if you will, to screen out a large -- the vast majority of
the sequences and plant operating states and things like that. One of
the issues raised is is that process sufficient, that screening process
sufficiently robust --
DR. WALLIS: Aha, so --
MR. CUNNINGHAM: That you're not missing something.
DR. WALLIS: Right. So you're just suspicious the numbers
will not be good enough --
MR. CUNNINGHAM: Yes.
DR. WALLIS: To guide decisions.
MR. CUNNINGHAM: That's correct. And another point is the
process that was used, even with this screening, at least for the NRC
studies, it was a very resource-intensive piece of work to get to those
Another question is, are there more practical ways today to
provide robust estimates without having to do several years' worth of
DR. SEALE: All of the completeness arguments apply in
spades, it seems to me.
MR. CUNNINGHAM: Yes, that's right.
DR. POWERS: Just to remind Members, the screening process
took out the mode of operation which the Wolf Creek shutdown event was
MR. CUNNINGHAM: Yes, that's correct. And that was going to
get -- kind of related to the point I was going to make that we had
human performance as an issue in shutdown conditions that is more
variable and more of a concern than in full power because of the Wolf
Creek example, if you will. There are more opportunities to do things,
if you will, without procedures and without the redundancy of the
DR. POWERS: Mark, I might just suggest that you move on to
the workshop --
MR. CUNNINGHAM: Okay.
DR. POWERS: In order to not do too much damage to our
Well, we're doing the damage but we blame you.
MR. CUNNINGHAM: That's quite all right.
DR. POWERS: That's kind of the rules here.
DR. WALLIS: Well, we could shorten our shutdown periods and
MR. CUNNINGHAM: The workshop was a one day workshop. The
morning was spent having -- we opened it up offered outside folks to
make presentations on what they have done, so we had a fairly large
number of industry presentations, be it from vendors or from consulting
engineering groups or from individual utilities.
The afternoon was spent trying to glean from what we heard
in the morning and from our own knowledge of shutdown risk what we could
learn about the four topics there at the bottom -- what do we see about
results, the scope, present methods, and the need for a standard.
MR. CUNNINGHAM: The title of this slide is Views Expressed
by Workshop Attendees.
Again, this is a snapshot -- the workshop was a snapshot of
remarks and things like that from a number of attendees, a number of
them from the industry but different segments of the industry, if you
will. It doesn't mean that the Staff necessarily agrees with all these
positions, but again this is just a characterization of what was said at
We also don't have at this point a lot of backup to say,
well, you have the question, well, why did they say that LERF was not a
useful metric, for example, we don't have a lot of background as to --
there wasn't a lot of discussion in this meeting as to the details of
DR. SEALE: I have a lot of prejudice about the value of the
falseness of that conclusion though.
MR. CUNNINGHAM: Okay.
DR. SEALE: Well, for one thing, there are all these
questions of the containment being open, so that the ratio of LERF to
CDF is not just the characteristic of the closed containment, whatever
kind it might be, but more than that, the driving term for release,
namely the energetics if a release occurred, are quite different for a
shut-down system. You don't have necessarily or you may depending on
what the shutdown mode is, you don't have the pool of pressurized hot
water to act as a driver or anything else, or those things, so it seems
to me the CDF to LERF transition is one of the most uncertain aspects of
the whole process.
MR. CUNNINGHAM: Yes. I think that was what was intended
here as a point, that the 1.174 concept of LERF doesn't fit well in this
circumstance at all.
DR. SEALE: Well, except that it still is a risk.
MR. CUNNINGHAM: Yes, it is. The question comes back to is
there a better metric you could use --
DR. SEALE: Yes.
MR. CUNNINGHAM: -- instead of LERF, which is based on an
early containment failure, as you say --
DR. SEALE: And it is our risk metric, really.
DR. KRESS: I think the concept applies. You just may have
to define how you determine what an early failure is.
DR. SEALE: That's right.
DR. KRESS: You know, if the containment is already open,
that's an automatic early failure.
DR. SEALE: Yes.
DR. KRESS: And the point about the driving forces is not
necessarily the case. You boil down the core and strike off the steam
zirc reaction, you have got just about equivalent driving forces there
with equivalent energies that you have in normal full power.
The timing may be different.
DR. SEALE: Yes, the timing would --
MR. CUNNINGHAM: That's right.
DR. WALLIS: Would it be true to say that if CDF is
comparable, LERF is probably bigger?
DR. SEALE: Oh, yes.
DR. KRESS: More than likely.
DR. WALLIS: So it becomes a more useful metric, important
MR. CUNNINGHAM: It just may be not a very -- the definition
that we use for LERF --
DR. WALLIS: The way it's defined --
MR. CUNNINGHAM: The way it is defined is not
DR. WALLIS: The actual release is a more --
MR. CUNNINGHAM: Yes, the conditional risk if you will may
be, can be for some POS's higher than what we see for --
DR. BONACA: One thing that I just wanted to point out is
that again we were talking about 5 percent of the time, whatever. In
reality, I mean the risk in the true time where you are at risk, which
is really mid-loop operation, is extremely high, I mean because the time
is extremely short.
MR. CUNNINGHAM: Yes.
DR. KRESS: It's got to be factored in.
DR. BONACA: Yes, so it is much shorter than 5 percent. I
MR. CUNNINGHAM: Yes.
DR. BONACA: -- the time in which you are at risk, you know?
MR. CUNNINGHAM: That's right, and --
DR. BONACA: To the point where that should provoke a number
of questions, like, you know, can you really avert -- avoid that risk?
MR. CUNNINGHAM: Yes.
DR. SEALE: Not only is the containment open but the primary
loop is open.
DR. BONACA: I mean we used to do this refueling in mid-loop
operation but there are other ways to do it. It's just I wanted to
throw in just a little, so --
MR. CUNNINGHAM: That's right. We are touching on a number
of points on this slide, as you say, that the instantaneous CDF can be
higher for some of the POS's like mid-loop for a short amount of time.
DR. BONACA: Short amount of time.
MR. CUNNINGHAM: That's right, so again -- maybe I will just
skip this slide and go on to the next on, where it talks -- perhaps what
you can do about some of these things.
Again, some of the views from the workshop is in a sense
focused on Dr. Bonaca's point of go in and if mid-loop is the real
issue, then focus your risk analysis and your work on risk significant
configurations and not go spend the resources to go across all POS's.
Other views offered at the workshop is the transition risk
which has always been an unknown in shutdown risk analysis. Arguments
were made that because during transitions there are more controls put
on, that maybe that is not as much of an issue as while you are in
No real agreement on whether or not fires and floods and
seismic should be included, at least from our perspective we did see
place in Surry where the fire CDF was a fairly -- was the dominant
contributor to mid-loop operation CDF.
Fuel handling, fuel pool cooling didn't seem to be big
issues. The issue raised of whether or not unplanned outages had been
adequately assessed in risk analyses.
MR. MARKLEY: Mark, I guess I have a question on the
transition and why they feel like there is no need for stuff. Because
most of the human errors, I mean at least in my experience, a large
number of them come during the transition periods when you are moving
stuff around and changing, and human error would seem to be larger.
MR. CUNNINGHAM: Again, I am in a situation where I don't
know much of the details of why people said these things. But it is
something we can pursue, if you will.
DR. WALLIS: This isn't really the basis of research, this
is all just hearsay.
MR. CUNNINGHAM: This is, yeah, anecdotal information at
this point, yes.
DR. SEALE: Yeah, but the fact that people had the
perception that the refueling risk was the shutdown risk, and that
seemed to be the case for a while, --
MR. CUNNINGHAM: Yes.
DR. SEALE: -- and that is no longer apparent once you have
looked at this, tells you something about the quality of that earlier
MR. CUNNINGHAM: Yeah, that's right. Early on I don't think
there was a recognition by anybody of the occurrences during mid-loop
operation, for example,.
DR. SEALE: Yes.
MR. CUNNINGHAM: They are just not recognized. And today I
think there is a very general recognition of the importance of
controlling configuration and managing your risk during mid-loop.
Talking about methods that would be needed to supplement
1.174, if you will, there was some discussion within the workshop that
the qualitative arguments may be adequate to supplement 1.174, and it
doesn't necessarily need a quantitative PRA approach.
DR. KRESS: Let me ask you a question about that one, Mark.
It seems to me like there are two basic applications for shutdown risk
assessments. One of them is what I would call risk management, to
control your outage, to plan your outage and to be sure you don't get
into too risky a configuration. That takes one kind of PRA and it may
very well be you could do this with defense-in-depth and that sort of
But the other need is the regulatory need, where you need to
include risk assessments not for a specific outage, but for all future
outages that go into your risk assessment to assess the risk status of a
plant now and in the future for regulatory decision purposes, for
risk-informing regulations. Now, that is a different kind of PRA. And
defense-in-depth doesn't even help you, I mean doesn't -- I mean it
influences the numbers because you have defense-in-depth, but it doesn't
help you control or do anything.
So it seems to me like there is basically two types of PRA
issue needed, and you need one type, the utilities need another type.
MR. CUNNINGHAM: That is a good point, and it may just -- in
a sense, I suspect the people making these arguments, and there have
been people who, by and large, the industry has been focused on.
DR. KRESS: Focusing on risk management.
MR. CUNNINGHAM: Risk management, that is correct. And even
within that, you will see within the industry some that are much more
qualitative and some that are very quantitative.
DR. KRESS: Which I think is all right.
MR. CUNNINGHAM: Yes. Yes.
DR. KRESS: That is good, and that is needed. But I think
you have a different need.
MR. CUNNINGHAM: Yes, that's right. That is a different
need than 1.174.
DR. APOSTOLAKIS: I am not even sure that this statement is
DR. WALLIS: Oh, I think it is wrong.
DR. APOSTOLAKIS: The qualitative defense-in-depth concepts
are adequate for RG 1.174. The whole idea there is --
DR. KRESS: I think it is adequate for risk management. I
think 1.174 is more of the regulatory need, and I would agree with you.
DR. WALLIS: Isn't it wrong when you have lost much of your
DR. SHACK: If you are Grand Gulf and you can bound it, then
it is adequate for 1.174. You know, it may be it is adequate for some
people and not for others.
DR. KRESS: Well, you know, you would always make the
DR. SHACK: Yeah.
DR. APOSTOLAKIS: Well, but you don't know. If you don't
have a detailed PRA somewhere, you really don't know whether you are
DR. KRESS: That is my opinion.
DR. APOSTOLAKIS: Even for risk management.
DR. WALLIS: It seems to me this is the wrong statement. It
is the wrong statement.
MR. KING: This is Tom King. What South Texas told us at
the workshop was they used detailed PRA to identify those states where
they needed to do risk management.
DR. APOSTOLAKIS: Yes.
MR. KING: Where they stop, when they get into those high
risk configurations, they stop other activities and they put dedicated
people to watch RHR, watch inventory and so forth. So they were doing
both, in effect.
DR. KRESS: Even there, that is a different kind of PRA than
what you need.
DR. APOSTOLAKIS: Yeah, that is very different.
MR. CUNNINGHAM: And part of that, too, is getting at the
barriers that we talked about, is you see some of the risky operations
come, then you make sure the containment is buttoned up and that sort of
DR. APOSTOLAKIS: But I disagree with the first bullet,
though, I mean make it clear, even for regulatory purposes, I am not
sure that -- I mean you are undermining 1.174, I think if you do much of
DR. POWERS: Let's make it clear that the speaker doesn't
vouch for this.
DR. APOSTOLAKIS: No, I understand.
MR. CUNNINGHAM: I am not an advocate of this position.
DR. SEALE: You are a reporter.
MR. CUNNINGHAM: Yes, that's right.
DR. APOSTOLAKIS: I understand that.
MR. CUNNINGHAM: That's correct.
DR. APOSTOLAKIS: It also amazes me, though --
DR. WALLIS: You mention the importance of human factors.
Isn't this a case where you say you dedicated people to watch the RHR
and all sorts -- this depends very much on those people assessing
probably what is going on, and most of the events seem to be based on
human misunderstanding or omission.
MR. CUNNINGHAM: That's right. Yeah, you can't -- one of
the issues that you see is poor instrumentation or lack of
instrumentation, so in that sense --
DR. WALLIS: Interpreting instrumentation they have got.
MR. CUNNINGHAM: Yes. Yes, that's right.
Two or three topics that were mentioned as being quite
different for shutdown, that may require additional research, are the
success criteria, how much flow, how much injection do you really need,
failure data for equipment, and then the source term issue.
DR. KRESS: The source terms wouldn't be very much
different, mostly just in the timing of the approach.
MR. CUNNINGHAM: That's right.
DR. POWERS: I think in my limited thinking on this subject,
I think that if you talk about things that go completely into core
degradation -- you are right, I mean there is this problem of higher
oxygen potentials and like that that complicate things and can have some
fairly dramatic effects on the elemental distribution of your source
term to consume air, high oxidation potentials exist.
And those kinds of things we can handle with
existing kinds of models, but I think there's a difference that you have
to recognize for shutdown and it comes about because of how you define
what are success criteria.
If we uncover a core during power operation and subsequently
flood it, we say we have been successful. I think when you are open in
a shutdown situation you uncover a core, discover that you have done so,
flood it, shatter the fuel, then you have not ended the accident there,
because water will leach iodine out of the fuel and iodine can partition
into the atmosphere, and you have sort of thing, and that is not
different than iodine leaching, partitioning out of containment sumps
and what-not like that, but it is something that we have not considered
that kind of a source term in our power operations and yet it could
leave you in a situation where you can't do anything about the plant.
It is too radioactive to approach, so you can't seal it up.
DR. KRESS: And I presume higher ingress by being a higher
probability shutdown. I am not sure.
DR. POWERS: Well, in whatever limited work has been done
largely in connection with the PHEBUS program, I think it is an open
issue and I think it is an open issue because people have a hard time
dealing with one of your favorite subjects, which is parallel flow
So I don't think we know whether it is more probable.
Clearly it is probable to put air into the system if you go into a full
core melt, but is it more probable at part way through and things like
that? I think it rests upon, like I say, your favorite topic, which is
parallel flow instability -- and we're using up your time.
MR. CUNNINGHAM: That's okay. The last topic of the
workshop was the need for a standard in shutdown risk. I think there
was general agreement that there was a need for a standard, but it broke
in a sense into two camps. One was -- in the sense of timing relative
to the present work and one opinion was that we should wait on this to
better understand the implications of what is -- if you will, the
lessons learned from the present ASME standards work, and try not to
avoid perhaps some of the dead ends that were hit in that activity.
Another side of the coin was that there's so much work going
out by individual licensees out there that the sooner that we get some
sort of a standardization of this, the better it would be for individual
licensees. It would reduce the variation or the variance of the cost to
The second point there is that we shouldn't assume that the
shutdown risk analysis would be kind of a cookie cutter of the full
power risk analysis. The standard then may have to be shaped quite
differently because of the way risk would be assessed in shutdown.
Basically, our overall goal is that over the next about six
months we need to come up with a statement that is a perspectives
report, if you will, on shutdown risk summarizing what our understanding
is and what we believe is needed in terms of future research.
In order to get there, we have had a couple of things
already We are going to continue with work over the next few months. We
are going to go out and become more familiar with some of the industry
shutdown PRAs including South Texas, Grand Gulf and some others.
We want to better understand the more commonly used risk
management methods like ORAM and EOOS and that sort of thing so we are
going to be on the, some of the Staff is going to be on a travel
circuit, if you will, to get a better understanding of these things.
DR. POWERS: We know what ORAM is, but what is --
MR. CUNNINGHAM: Pardon me?
DR. POWERS: ORAM I am familiar with. I can find it on the
web. Can I find EOOS on the web if I can't find anything else about it?
EOOS -- I just don't know what it is.
MR. CUNNINGHAM: I am not sure -- I am looking at Gareth or
Erasmia. Do you know?
MR. PARRY: Yes. It's either EPRI or SAIC and I think they
would -- it's EPRI, okay. Then you would be able to find it from them.
MR. CUNNINGHAM: We are going to continue a review of what
we can learn about domestic studies and through COOPRA and just continue
the work looking internationally.
DR. SEALE: Mark, you had a list of three or four utilities
and then SAI and other people who were involved directly in your
MR. CUNNINGHAM: Yes.
DR. SEALE: How long is the list of utilities who have
shutdown PRAs that would be appropriate candidates for inclusion in this
MR. CUNNINGHAM: The list -- we are going to see four at
DR. SEALE: Yes.
MR. CUNNINGHAM: I don't have a sense of how many others
there are. The four were kind of -- you know, we got one that's very
detailed, one pretty simple and two in the middle. I have heard, again
anecdotally, that at least half the utilities have shutdown PRAs. I
have heard that. Whether that is again how sophisticated they are,
whether they are being used more for qualitative risk management
purposes and things I don't know.
DR. WALLIS: That final report, that perspectives report --
MR. CUNNINGHAM: Yes?
DR. WALLIS: I would like to go back to what Dana said
earlier. I think that it would be good if you spelled out more clearly
what it is specifically that is needed by the NRC.
MR. CUNNINGHAM: Yes.
DR. WALLIS: Right now, so that you -- what we are going to
wind up with at the end is a measure of A, B, C, D. We need those to
make sudden decisions, whatever they are -- X, Y, Z -- and focus on
those things. Do we have enough information to give them that, so that
they can then decide what to do.
DR. KRESS: Are you thinking about, Mark, the needs for an
additional database for shutdown risk? I have in mind, you know, you
need to -- there's probably a need to survey the whole industry to find
out what frequency different systems and components are at, what is the
durations they are out, an average type thing, and then -- and the
average time of shutdowns for the industry. These are data that you are
going to need --
MR. CUNNINGHAM: Yes.
DR. KRESS: When you make a risk assessment and then I don't
know if it has all been collected very well, because they are not
normally in a PRA database.
MR. CUNNINGHAM: That's correct. You could see that
evolving, that we are starting to do a better job of collecting power
operation data on equipment failure rates and that sort of thing.
DR. KRESS: That's not going to -- it's a different
MR. CUNNINGHAM: That's right.
DR. SEALE: You need to know about refueling machines and
cranes and things like that.
MR. CUNNINGHAM: And you need to know times of, durations of
different configurations and that sort of thing.
DR. KRESS: They're different and whether there's
correlations. You know, you take this one out but you won't take this
MR. CUNNINGHAM: Yes.
DR. KRESS: Or these two will come out at the same time or
something like that.
MR. CUNNINGHAM: That's right.
DR. KRESS: So you need to worry about correlations.
DR. BONACA: The other thing about it, I think there is a
lot of information available because I believe all utilities must
DR. KRESS: I think it is available. You just have to go
get it and you have to figure out what you want and what --
DR. BONACA: They have the shutdown management programs in
MR. CUNNINGHAM: Yes.
DR. BONACA: Whereby they have procedures and also they have
very clearly identified whether or not they are developing either
quantitative or qualitative PRA profiles, okay? It qualitatively means
that it is not fully quantified but you have clear identification of the
delta between certain conditions you are addressing, so through that
information I think there is a lot of information available to see where
the status of the industry is just by those programs.
DR. KRESS: I don't know how you retrieve that data. That's
the only -- put in the form you need it.
MR. CUNNINGHAM: That's right, and we have talked mostly
today about the issue of how you would treat it in the context of
license amendments and that sort of thing, considering it in the 1.174
You get into the new risk-informed oversight program, the
inspectors are going to be faced with the question of how do you --
what, if anything, do you need to inspect in terms of shutdown, the
programs, the results or that sort of thing, so there is another
interest here, if you will.
Last slide, basically over the next few months we are going
to be visiting the sites and other engineering firms and that sort of
thing. As it's laid out now, we would envision having a draft report in
the October time frame and coming to the ACRS perhaps for a subcommittee
in November to talk to them -- talk to you some more. If you like, we
can perhaps add a subcommittee in before that where once we've gone on
the plant visits and the reviews, and perhaps in the September time
frame or something if you're interested we could give you more of a
perspective at that point of what we've seen out in the industry.
That's a decision for obviously --
DR. WALLIS: Well, I have a sense of urgency about this that
I don't see reflected. What your numbers told us was that the CDF is
comparable to power operation, that the actual cancer fatality risks are
higher because the things are open and so on. So it would seem that the
risk to the public of shutdown is maybe greater than the power operation
on which NRC's been focusing so much attention. It would seem that this
would be a really high-priority item to deal with.
MR. CUNNINGHAM: I think it is a high-priority -- one of the
things that's not reflected very well in the results I showed was again
those are about five years old, and there's been a lot of work in the
industry that's been done since then to manage their outages better, to
reduce the duration, to -- someone talked about the idea of just
avoiding mid-loop operation. Now after this work we did for Surry,
that's one of the things they went back to do. They said we're just
going to do everything we can to not go --
DR. WALLIS: So it looks like a place where the NRC could
have a great influence on public safety, and we get a lot of stuff to
look at where the payoff to public safety may be small. I mean, it's
just detail stuff. But this looks like an item where you have a chance
to make a real contribution.
MR. CUNNINGHAM: Yes.
DR. WALLIS: This is quite measurable.
DR. KRESS: Let me ask you another question about planning
and research. As I said earlier, I think a PRA for shutdown looks a lot
different than a PRA for full power, and it looks a lot different for
your needs than it does for risk management. It seems to me like
there's going to be a need to develop approved PRA methodology.
Actually you're not going to be able to take your standard full-power
PRA in and wedge shutdown into it. So there's a need to develop PRA
methodology for doing shutdown risk. Now whose job is that? Are you
guys going to do that in Research or is that --
MR. CUNNINGHAM: That's one of the --
DR. KRESS: I can't see industry doing that, because it's
your need. You need this kind of PRA to do your risk-informed
MR. CUNNINGHAM: Yes.
DR. KRESS: So it's a need you have.
MR. CUNNINGHAM: Yes. In that sense --
DR. KRESS: Do you envision this leading to a research
MR. CUNNINGHAM: Yes.
DR. KRESS: Perhaps --
MR. CUNNINGHAM: Yes, that's correct. You know, one of the
things we have to deal with, is the real need methods development or is
it more applications of existing methods or what. But all of that would
be in the scope of this program, if you will.
DR. KRESS: That's part of the scope.
DR. POWERS: Let me make sure you understand, Graham, those
numbers Mark put up there are part of a scoping study that they did with
the ground rules that are fairly typical of PRAs. That is, actions that
are proceduralized were allowed; unproceduralized or heroic actions were
not considered in getting those risk numbers. Well -- and that's a
well-founded process in power operation PRA. But it kind of flies in
the face of sensibilities when it comes to shutdown PRAs, the idea that
you could have a core merrily boiling its coolant off and -- I don't
have a procedure, so I just can't do anything about that darn thing.
That's a little bit unfair, in that there are lots of
accidents that can occur in shutdown where there isn't time to respond.
MR. CUNNINGHAM: Yes.
DR. POWERS: But the fact is that the recovery options
available to the operations staff are richer than probably are reflected
in those numbers.
DR. SEALE: That's right.
DR. POWERS: And that's why some, including Members of the
Commission, have questioned the accuracy of those things. And I think
that's the kind of information you need to go out and ferret out because
you presume that a plant is not so rigid in its rules on how it does its
So those scoping numbers, you know, they always excite me,
and I rant and rave. On the other hand, I also try to temper my
comments by saying I also know where those numbers came from, and so
that's one of the reasons the ACRS in its letter to the Commission
didn't say do something about shutdown risk. Now we said go find out
about shutdown risk, because there's more to it than I think the scoping
studies have said.
DR. WALLIS: Well, not only are there opportunities for
heroic action that solves the problems, probably for heroic action that
makes it worse.
DR. POWERS: Yes. Right now what -- I mean, what you have
is an empirical data base that says something like 58 percent of the
augmented inspection teams that the NRC sends out are associated with
shutdown or low-power events. That's one data point.
The other data point I get is Mr. Pietrangelo comes in front
of me and says yes, that's all true, but it's ancient history. But if
you look at the modern day, and someday he's promised me he's going to
show me this plot that shows that so many improvements have been made in
the way we do outage management nowadays that those numbers are way
down. Okay. So it's a dynamic period.
DR. WALLIS: That may be true.
I have another question for you. We talk about PRA
methodology, and we sort of think about probabilistic things, but I
wonder how much we really understand these sort of technical, physical,
chemical scenarios of uncovering the core or failing to cool it properly
or the various things that can happen when the top's off. But you spend
hundreds of millions on LOCAs and things which are very much less
likely. Do you have the basic technical research to know what happens
in the event of --
MR. CUNNINGHAM: No, I think that's -- within the context of
the research program we could be talking about here in six months that's
fair game as well as the probabilistic parts.
DR. WALLIS: Do you have any idea of how much that might
MR. CUNNINGHAM: In terms of?
DR. WALLIS: Of the needs for research in that sort of area
and the technical details.
MR. CUNNINGHAM: I don't have a good sense now. The part --
DR. WALLIS: What is the basis now? Are there calculations
made of all kinds of scenarios that could occur in event of shutdown --
a failure to cool the core during shutdown?
MR. CUNNINGHAM: Yes, there are thermohydraulic calculations
done, there are source term calculations done that would --
DR. WALLIS: Are they of the quality that we've seen for
LOCAs and things like that?
MR. CUNNINGHAM: I suspect --
DR. POWERS: The source term calculations of course are
superb and without fault. But the thermohydraulics --
DR. KRESS: Pretty superficial.
DR. POWERS: Well, there's been --
DR. WALLIS: Can we answer that without, you know -- in a
more professional way?
MR. KING: I don't know of any experimental work that we've
funded that's looked at the shutdown condition with the head off the
vessel and thermohydraulically run a calculation and compared it to
DR. WALLIS: Well, what happens?
DR. POWERS: Well, what I know is that in connection with
PHEBUS feasibility -- there's a PHEBUS test in France planned to look at
the issue of air intrusion on the behavior of the fuel. And in
connection with the development of the feasibility for that,
calculations of a variety of natures were run, and they varied from lump
node codes trying to jury rig the air intrusion accident to fairly
sophisticated codes that Mr. Patankar I believe develops on natural
circulation kinds of phenomena and what not, but they ran into the
problem that this -- when you've got the head off, literally the head
off, that you've got this huge big-diameter vessel with steam coming up.
Well, that sounds all very good, and you're very confident
that the steam keeps the air out, except Dr. Kress points out accurately
that this is inherently unstable, and you get these instabilities. And
the problem is it's irreversible. Once air starts going down into the
core, it reacts, and the oxygen's removed out of that, and so it creates
a vacuum, and you never get out of that instability, and so you can get
air intrusion. And that seems to be a very difficult thing to handle
computationally right now.
DR. WALLIS: So, you see, there are lots of lacunae in the
DR. POWERS: Yes. Now, the source term itself, there is
some confidence that if we knew exactly what happened to the clad, we
would know how to calculate it, because the only thing that really
changes is the oxygen potential, and we think we know what oxygen
DR. KRESS: I don't think we really handle heatup rates very
DR. POWERS: Well, the real --
DR. KRESS: The heatup rates around the change.
DR. POWERS: The real challenge comes up is if you heat very
vigorously, you drive the clad to melting and now it can drain away and
leave the fuel exposed, and that is all the difference in the world. If
you expose the fuel to oxygen, there have been tests now done at the
esteemed institution in the southeast of the United States on air
affects fuel and a really extensive set of experiments --
DR. KRESS: Is that Savannah River?
DR. POWERS: No. No, much more esteemed than Savannah
River. You may have heard of it, it is very near where you live. And
there has been some work in Canada that pretty clearly shows that if you
expose this fuel to air, you are going to get a ruthenium source that
you had not anticipated.
Now, you say, well, ruthenium has been more a fission
product. Ruthenium is the element that I characterize as being a
disaster that they make movies out of, because it has the prompt
fatality consequences of iodine and the latent fatality consequences of
cesium. You do not want to get large releases of ruthenium.
DR. WALLIS: Well, there is also a possibility of this
reactor gone critical with the head off. I believe that has happened in
at least one case.
MR. CUNNINGHAM: SO 1.
DR. WALLIS: So I guess that you might find out in your
studies that there is a great need for research on what actually happens
during the --
MR. CUNNINGHAM: The physical process.
DR. WALLIS: Within something the Commission would have to
face up to.
MR. CUNNINGHAM: Yes. I have run well over my time.
DR. APOSTOLAKIS: Do you think we should have a subcommittee
DR. POWERS: I think it is essential. But I think it is
essential in the November timeframe. I am not sure that Mark's offer to
have an interim briefing in there is so crucial to us. That is my
DR. APOSTOLAKIS: No, I think that is --
DR. SEALE: If you find some surprises, --
MR. CUNNINGHAM: Yeah, maybe that is it.
DR. SEALE: -- that may be appropriate. But you are
probably a better judge of whether that is opportune.
MR. CUNNINGHAM: Okay. What we can do is we can go back and
after we get through the next few months, we can see if there would be
something that we could -- that would merit a few hours or half a day or
something like that.
DR. SEALE: This does look like a candidate for having some
DR. WALLIS: Well, is this final report going to be of the
form "more work is needed," or is it going to be of the form "there
isn't a problem," or is it going -- or you have a great problem here and
you have got face up to it? I hope it is going to really face up to the
problem and not end up saying, well, work is needed and action,
something will come --
MR. CUNNINGHAM: I think this is a planning effort, so one
of the outcomes of the report will be this is -- specifically, this is
what we need to do over the next few years to -- given the money we have
for low power and shutdown research.
DR. WALLIS: I think what the Commission probably needs to
know is it looks as if, from some of the indications, this might turn
out to be a real sort of significant problem that would shake the
agency, in which case they need to know.
MR. CUNNINGHAM: Well, again, we are seeing the confirmation
over and over again that the shutdown risks are similar. The Commission
is aware of that, but perhaps this will just reinforce, but I don't
DR. KRESS: Well, lest you put too much weight on your
workshop opinions, I suggest you start from top level principles which
might include stuff like we want a realistic estimate of the risk. We
want to include most of the POSs because they will add some in there and
we need it even though I wouldn't -- I wouldn't say we don't want
scoping analysis that are bounding in this case, because we are talking
about the needs for risk-informed regulation.
MR. CUNNINGHAM: Yes.
DR. KRESS: So, you know, I wouldn't put too much weight on
those workshop things.
MR. CUNNINGHAM: Yeah.
DR. KRESS: And I think you need to start thinking about
this database I mentioned, and you need to start thinking about how --
what is a PRA going to look like that will do a shutdown risk? And it
is not an ordinary PRA. I think your problem there is going to be, for
-- you have got to deal with the individual plants, PRAs for individual
plants, and individual plants' future shutdown configurations are
unknown to you. But you have got an industry-wide database that tells
you what shutdowns generally might look like on a probabilistic sense.
So you are going to have to think about likely about
probabilistic sense, future configurations for a given plant, and how
you deal with that in a PRA. How do I incorporate likely future
configurations in a PRA? And you will have to use this database for
industry-wide, because you don't have a database for an individual
plant. You will have to use a database industry-wide to get likely
future conditions and have the PRA configured in such a way that it can
select internally those likely conditions.
MR. CUNNINGHAM: Yeah.
DR. KRESS: That is a different -- that is why I am saying
it is a different PRA.
MR. CUNNINGHAM: Yes.
DR. POWERS: I am going to have to intercede because I have
got -- I am going to run into a scheduling problem here.
DR. KRESS: I think we are basically finished.
DR. BONACA: I have just one comment if I could, just one
simple comment I had. The observation I made before, clearly, I am not
-- I understand the impact on the industry, but cutoff load, which some
plants do, eliminates much of the risk, which is really mid-loop
oftentimes. I am only saying that there may be even an analysis done at
some point to understand, okay, what -- in the aggregate, by using the
number of plants you have got, the number of refuelings you have per
year, and so on and so forth, to look at the individual issue of
mid-loop operation versus full offload, just to get an understanding.
DR. KRESS: I think Graham would say there ought to be a
cost-benefit assessment of that, because that is going to cost you a lot
of full core offload.
DR. BONACA: I understand that, and I am not proposing it
right now. I am only saying that one should understand the dimension of
that issue alone, because there is one action there that certainly may
not be desirable but is feasible. That's all.
MR. MARKLEY: There certainly won't be 17 day outages.
DR. KRESS: Yes.
DR. BONACA: Yeah, I understand that. I am not proposing it
DR. KRESS: Thank you, Mark.
MR. CUNNINGHAM: Thank you for the comments.
DR. KRESS: It was very, very interesting. We will look
forward to continuing our interactions on this issue.
MR. CUNNINGHAM: Thank you, all.
DR. KRESS: I will turn the floor back to you, Mr. Chairman.
DR. POWERS: Okay. At this point I want to go off the
[Whereupon, at 11:40 p.m., the meeting was recessed, to
reconvene at 12:47 p.m., this same day.]. A F T E R N O O N S E S S I O N
DR. POWERS: Let's come back into session. We are moving
now to the topic of options for crediting existing programs for license
Bob, I think you're in charge of it.
DR. SEALE: Yes. I'm working sort of as a vice-chairman,
along with Mario, on this particular Subcommittee on License Renewal.
And the question of credit for existing programs is going to be looked
at as a specific issue as we go forward.
You remember the last time we talked about license renewal,
among other things we were exposed to the statistics that there are
about 400-odd programs of one sort or another in the existing license
basis for BGE that have been judged to be sufficient unto themselves to
provide the kind of monitoring and oversight of those activities to
carry forth into the renewed license. There are between ten and 20 of
those programs that were found to be slightly deficient in a few areas,
and there were only 25 new programs -- or about 25 new programs that
were identified as being required in order to meet at least the first
cut at the full requirements of license renewal.
Clearly this affords an opportunity to use considerable --
or to realize considerable resource conservation if we are able to carry
these programs forward without serious change.
This is an issue that has been the subject of discussions
between people from the Nuclear Energy Institute and the Commissioners,
and we understand that the staff is in the process of preparing a draft
Commission paper on the subject. That paper is not fixed yet or not
finished, so we're not able to get the full story at this point.
Mr. Grimes has offered to give us an update and a status
report and so on, and so this is a good time for us to get a heads-up on
what we can look forward to.
Chris, I'd like you to go ahead and do that, but I'd also
want you to -- or would appreciate it if you could tell us when do you
need something from us, and what that something might be.
MR. GRIMES: Yes, Dr. Seale, I'd be happy to.
My name is Chris Grimes. I'm chief of the License Renewal
and Standardization Branch.
As Dr. Seale mentioned, there's an issue that has existed
since the license renewal rule, Part 54, was amended in 1995, and it
concerns how existing programs are credited for the purpose of a license
renewal application. Recognizing that the vast majority of aging
management programs that utilities would refer to in order to address
the finding that we would need to make for a renewed license are
existing programs, and then, as Dr. Powers pointed out, but it gets back
to a basic question about what mission did the Commission intend to send
us on when they said to go review aging management programs for passive,
long-lived systems, structures, and components, in order to determine
that those programs are effective at managing aging effects, what does
"demonstrate the effectiveness of" mean in a regulatory review context?
We have a Commission paper that we had hoped to have
completed by now. We had targeted completion for May 28. I apologize.
It was probably too optimistic to think that after we'd met all our
milestones for the first two applications that we meet the milestone for
that Commission paper.
But I can tell you that it has been a very difficult issue
for the NRC and the industry to come to grips with in terms of what is
it that we're arguing about. And that has led us to have the Commission
paper go through a series of evolutions including feedback from the
License Renewal Steering Committee and the EDO's office. And so we have
shared with the Committee the latest draft, which does include changes
that -- the feedback that we got from the Steering Committee and our
management and the EDO's office, but because the EDO has not yet signed
the paper out, we've shared it with you as predecisional information
because there's still a chance that the nature of the options, the
characterization of the impact of those options, might change. So we've
given it to you in draft form.
We are hopeful that the Commission paper will be completed
shortly. As a matter of fact, they are so hopeful in the EDO's office
that there is a Commission briefing tentatively scheduled on the subject
for the week of July 12. And we're going to talk specifically in that
briefing that the staff is going to present its views about what the
Commission intended, referring to language in the statements of
consideration, for the 1995 amendment to Part 54.
NEI will be there to present their views about how they read
that same language from the statements of consideration and what they
had thought the nature of the license renewal would consist of. And I
think that we may find that after all of our effort to try and
characterize the nature of this controversy and what its real impact is
relative to a license renewal application, that we'll find that we're
not so different, but that it gets back to and how is stability and
predictability ensured in a license renewal review process.
And so we will continue to keep the Committee posted on this
issue. It is a fundamental policy issue, the question about what
mission did the Commission intend for us to do. And so it's very
important for us to get that mission statement clarified and to resolve
any differences about what the nature of the job is.
DR. POWERS: When you look at the statements of
consideration and the language that was bandied around during the
development of the license renewal rule, do you find suggestions that
the word "to demonstrate" was in any way different than its engineering
MR. GRIMES: No, we don't feel that -- we feel that
"demonstrate" means something substantial. We also felt that the
language in the statements of consideration challenges us to challenge
the existing programs and to determine what additional requirements
should be imposed to manage aging effects for the period of extended
The difficulty comes about in the language in the statements
of consideration that says but the current licensing basis is still
adequate and will carry forward. And using, you know, referring to that
language, we also understand the industry's views that the threshold for
challenging the adequacy of existing programs should meet some
And then it gets down to a very practical and pragmatic
question about well, what's the appropriate threshold vis-a-vis this
question about adequate protection. If the program is adequate for the
first 40 years, why isn't it adequate for the next 20? That, you know,
I think that's a fair question. And we view "demonstrate" to mean well,
it may be okay for the first 40, but we have a higher expectation to
manage aging effects through a period of extended operation.
DR. POWERS: I guess it was not my impression that the -- I
had gotten the impression, rightly or wrongly, that things that were
adequate for the first 40 years did not have embedded in them the
concept of aging that went beyond that, and that what you were looking
for was equivalency, but that it was not ipso facto true that because it
was good for 40 that it was good for 60.
DR. SEALE: Or fully sufficient.
DR. POWERS: Fully sufficient.
The particular issue used as an example for challenging you
is one that intrigues me, because -- the smaller-than-four-inch-pipe
issue, where it seems that it's argued that it's adequate to infer
rather than to demonstrate, it seems to me that's contrary to the
MR. GRIMES: We would agree, but NEI would disagree. And I
think that trying to capture the nature of that disagreement has been a
challenge for us for this Commission paper. We also fully expect to
address that example along with others.
The first example that we exposed you to was this question
about is the equipment qualification process under 50.49 adequate to
manage aging effects. And we went probing and poking and stumbling
through how 50.49 is managed today. And the industry press reported the
staff's efforts as being rereviewing compliance with those regulatory
requirements, while we tried to characterize the nature of our inquiry
as a review of the adequacy of the aging management aspects of EQ.
And so there is, you know, we're set up in a situation where
asking questions about the adequacy of programs to manage aging effects
infers a challenge to the current licensing basis. And I think that the
Commission intended that we should, you know, challenge these existing
programs. But then you can also read the language in the statements and
consideration about the adequacy of the COB and its ability to carry
forward as one about not duplicating compliance reviews, not challenging
And at the risk of -- I don't want to contradict Dr. Seale,
but the numbers that BG&E reported are counts of individual procedures,
activities, and programs. So they get a very large number.
DR. SEALE: Yes, that's right.
MR. GRIMES: For statistical purposes. Three hundred
twenty-nine procedures that are relied upon for aging management were
deemed adequate as is; 101 procedures or programs were modified in some
shape, way, or form, in order to address aging effects for all of the
applicable passive long-lived systems, structures, and components. And
16 programs were new. And so if the intent of our mission --
DR. SEALE: Something was incorrect. That's easy.
MR. GRIMES: For me the dialogue has now gone on for four
months, and they're burned into my brain.
DR. SEALE: I bet they are.
MR. GRIMES: There is a different set of statistics for
Oconee, which counts things in program measures that get up into the
couple dozen. But we're still talking about an impact to the license
renewal that affects 25 to 30 percent of activities, whatever those
things are, and trying to characterize that in terms of what should the
threshold be in order to effect a change, and to what extent should the
NRC staff review challenge those decisions is the fundamental policy
issue that we need to rectify.
At the same time, as Dr. Apostolakis points out, there is
this concern about if we push too hard, if we say that the '95 amendment
to the rule was a mistake in some way, if we are not careful how we
characterize the nature of the resolution of this controversy, then
there is a real fear that there are utilities out there who are sitting
on the fence that might not pursue license renewal because of a concern
about lack of stability and regulatory control that might otherwise
continued to operate their plants, and that is a real concern, so we are
not trying to overblow this issue. At the same time, we are not trying
to underestimate its importance.
My view is that we will continue to keep the committee
informed, but I think that the Staff and NEI first need to present the
issue to the Commission and let the Commission help us sort out what is
the fundamental question and the policy issue that needs to be addressed
and then thereafter we will come back and ask the ACRS to review the
case as it was presented by the Staff and NEI, and then you can present
your views to the Commission and we will let them know that you are
doing that in such a way that their decision is timed to consider your
DR. SEALE: Okay. You are talking about this being on the
12th of July?
MR. GRIMES: The week of the 12th of July.
DR. SEALE: In your schedule --
MR. GRIMES: That's correct.
DR. SEALE: Well, it turns out that is the week of our July
meeting and then we don't have a meeting in August, so if you wanted to
come to see us in the latter part of July, then it would be the first of
September, roughly, when we could give you some feedback.
Is that sequence doable?
MR. GRIMES: It seems doable to me. We would probably argue
before the Commission that this is a moot issue for Calvert Cliffs and
Oconee because we did the job we felt we needed to do and we exercised
that term "demonstrate" to its fullest extent, and we hope that you will
agree with us after you have reviewed the Oconee safety evaluation,
unless this is an issue that really applies to the next two applicants
in terms of both the scope and depth of the material that they would
cover in their application and the scope and depth of our review, but we
will -- you know, we will relay the scheduling aspects to the Office of
the Secretary and then let them sort out the schedule with the
DR. SEALE: Well, I would assume that -- well, let me ask
Do you believe that it would be appropriate for us to try to
have a subcommittee meeting to get the overall full and fruity feedback
review on, the story on what you have presented to the Commissioners and
any at least initial feedback as to what their comments might have been,
and then plan to have a presentation to the committee in the September
MR. GRIMES: I think that would be a useful thing to do.
As I said, it would depend in part on how the meeting with
the Commission goes, because it could be that we -- that that dialogue
may end up fully reviewing what the nature of this disagreement is and
we find we are not as we thought we were, but on the other hand, it
would have been nice for me -- my druther would have been wish we could
have practiced on you first before we go to the Commission.
MR. GRIMES: But it doesn't look like it is going to work
out that way, but I think it would be useful to schedule the
subcommittee and if it turns out we don't have as much to talk about on
this particular issue then we could carry it forward and talk about the
ramifications to improvements in the Standard Review Plan. You have
lots of opportunity here.
DR. POWERS: You don't want to try to do something actually
in the July meeting then?
DR. SEALE: It doesn't appear that you are going to be in a
position to give us much to do. If they came up with something that
would give us a July meeting goal that could really satisfy their needs,
I think we could certainly work towards that.
DR. POWERS: Yes. I was just going to say that as you move
forward, then you find that it would be convenient for you to do
something in July rather than waiting for September, we will maintain
schedule flexibility on this issue to meet your needs as well.
DR. SEALE: Yes.
MR. GRIMES: I appreciate that.
DR. POWERS: It is a thorny issue and it has -- we have a
little different charter in this regard than maybe the Commission does,
charged for looking at technical adequacy on this, in our statutory
response. We have another charter to advise the Commission, and we have
to address both of them in regard to this issue because I think it is
MR. GRIMES: We agree. We appreciate that, and we think
that the ACRS views on this controversy are important and can assist us.
Obviously, we think you will agree with us, but at the same time you
have pointed out that there are probably areas of the Staff review that
are not worth the effort and there are areas of the Staff review that
probably aren't deep enough, and so this gets right to the heart of that
question because it is basically a question about the scope and depth of
the Staff evaluation.
DR. POWERS: Sure.
MR. GRIMES: At least now I finally came to understand how
Corbin McNeil could figure six months to complete a licence renewal
review after I read the March letter from NEI -- because if you take
NEI's explanation about carrying the current licensing basis forward to
an extreme, then using the statistics from BG&E we only look at 16
things out of 436, which makes the review very simple.
But in fairness, I think NEI needs an opportunity to clarify
what they meant in their letters, because we find some inconsistencies
in their explanation about what they viewed as the issue and it's hard
for us to understand.
Before I forget again, Doug Walters expressed his regrets.
There is an NEI workshop on design basis that is being held right now,
and they weren't able to attend this session and they fully expect that
there would be more dialogue on this topic, and so I am trying to
represent NEI's perspective as best I can.
DR. POWERS: That just adds fuel to your idea of a
subcommittee, it seems to me, because we're certainly going to
understand all the nuances, and that takes more time.
DR. SEALE: Yes. We certainly need to offer them the
opportunity to bring their own soapbox.
Any other comments? Mario, would you have any comments?
DR. BONACA: I just had a question regarding the March 24th
letter from Bill Travers on this issue.
There is an example there being provided regarding the Staff
found that the ASME Section 11 Code doesn't address certain areas such
as cracking of less than four inch diameter reactor coolant system
piping and so on. Okay.
Well, that is important data to cover but wouldn't that
consideration be applied also for operating plants irrespective of the
license renewal process?
MR. GRIMES: And in fact there is an explanation about how
that matter is being pursued for operating plants and it gets back to an
interesting regulatory process concern, and that is for these issues,
these questions that have been raised that have aging management
implications for which we have no fully developed generic safety issue
or task action plan that plots a future course, if there is a question
that is lingering out there that has an aging management implication, is
it fair for the NRC Staff to raise that question as an adequacy of aging
management for the purpose of license renewal when we still haven't
decided how we are going to proceed to address the issue for the current
That is the small bore piping issue that we refer to as here
is an example of where First Forty hasn't even gotten the issue off the
ground yet for which there is an aging management issue. You have heard
the other extreme of that example, which is the fully-developed generic
safety issue 190 on fatigue, where we manage to come up with a
plant-specific way to address the issue pending a generic resolution of
the GSI and then we have a series of examples of everything in between
EQ is an example of one that is beyond. We found nothing
needed to be added to the program. But then there are other questions
about things like monitoring buried piping for which it's not even an
issue for current operating plants, but it is an aging management
So we fully expect that at both the Commission briefing and
the ACRS Subcommittee meeting both the staff and NEI would be prepared
with their examples to illustrate the two sides.
DR. SEALE: And I can certainly anticipate a lot of
questions from the committee on exactly these issues.
Any other comments? Bill, do you have any?
DR. SHACK: No. Thank you.
DR. SEALE: George.
DR. APOSTOLAKIS: No, I have some crazy thoughts. Could one
use -- this is an entirely different subject. Could one, in the name of
regulatory stability, if you don't want to touch 54, I wonder whether
one could use 1.174, before the 40 years expire, to change the licensing
basis and them demonstrate according to 54 that you meet it for the next
MR. GRIMES: Sure.
DR. POWERS: I don't see why not.
MR. GRIMES: We fully --
DR. APOSTOLAKIS: But how would you do that? I mean --
MR. GRIMES: Well, we fully expect that plants -- the later
plants, as a matter of fact, Arkansas, which will be in the next license
renewal application, is going to refer to their ISI program for a lot of
the aging management.
DR. SEALE: It is a pilot.
MR. GRIMES: And they will have --
DR. APOSTOLAKIS: Components though. No, ISI, I'm sorry.
MR. GRIMES: ISI applies -- ISI is relied on extensively for
aging management and piping and pressure boundary. And they will have a
risk-informed ISI program, and then we will be asked to judge the
adequacy of aging management for all of the -- for the whole scope of
pressure boundary that is covered by ISI.
DR. APOSTOLAKIS: So risk information can be injected.
MR. GRIMES: Sure.
DR. KRESS: But your idea was to fix all the current plants
before they get to license renewal and just keep the license renewal
stay rule as it is.
DR. APOSTOLAKIS: Well, yeah, 54 refers to the current
DR. KRESS: That is an intriguing thought, yeah. Right.
Let's make them change the current --
DR. APOSTOLAKIS: Change the current licensing basis.
DR. KRESS: That is an intriguing thought.
DR. APOSTOLAKIS: In other words, it comes back to something
that is similar to the 50.59 debate.
MR. GRIMES: Sure.
DR. KRESS: We would still have to --
DR. APOSTOLAKIS: Can you go and argue on the basis of 1.174
that you can reduce your safety margin, you know, some components,
passive components and so on, by some delta X?
DR. SEALE: Small delta.
DR. APOSTOLAKIS: Even though you have not reduced it yet.
And then come the license renewal and say, look, this aging mechanism
will deteriorate this thing and at it the end of 20 years, it will still
be within the delta X you have already approved.
DR. KRESS: The problem is 1.174 does not give you title to
do that. It is strictly written for licensees coming in with requested
changes to their licensing basis. It doesn't give an RC the power to
make any requested changes.
DR. SEALE: You don't get a blank check.
DR. KRESS: You don't get a blank check with it. But the
DR. APOSTOLAKIS: But it will be specific.
DR. KRESS: Yeah, but it is not -- there is no licensee who
is going to come in with this request. It is going to originate from
the NRC and they can't originate it.
DR. SEALE: Obviously, we need a subcommittee meeting.
DR. KRESS: Yes.
DR. SEALE: I am sure you have heard most of these arguments
DR. APOSTOLAKIS: I was hoping they were original.
DR. SEALE: But thank you very much. Is there anything else
you want to tell us at this point?
MR. GRIMES: I can tell you that amongst our inventory of
generic renewal issues, which you can certainly contribute to, because
everybody gets to give us their good ideas, one of those issues is how
to do risk-informed license renewal, you know, with a vision towards how
does -- what role does the PRA play in identifying either scoping, the
plausibility of aging effects, or the effectiveness of aging management.
DR. APOSTOLAKIS: Is there room, though, in the current
regulation for this? And I will tell you why I am asking. I have been
talking for the last maybe couple of years to people, you know, it was
not a systematic survey, but around the world about -- and also in this
country, about probabilistic methods for aging mechanisms and including
them aging assessments and so on, PRA, including them in PRA. I would
say that all, most if not all of the people I talked to said, yeah, but
there is no reason to do that because the license renewal rule is not
going to use any of that.
So that, the existence of the present rule has discouraged
the development of methods of the thinking as to how risk information
could be used in that, because right now people are not spending money
on research just to advance the state of the art, they want to see a
goal. So perhaps we are creating a chicken and an egg here, you know.
The rule says don't use PRA, or it does not exclude -- it doesn't say
that, but, you know, it does not encourage it. Therefore, there is no
work on PRA and how it could be used for these kinds of issues.
Then we come back to the rule and say, well, gee, how can we
use risk information? We look at the state of the art, it is not very
good. And, you know, that is not a healthy situation.
MR. GRIMES: I am sure that if I am not -- if I don't do
this correctly, I am going to get hit in the back of the head with a
DR. POWERS: I am kind of hoping you do it poorly, I would
like to see this.
MR. GRIMES: I noticed Mark was sitting behind me. In fact,
at the same time the Commission acknowledged that the license renewal
rule is predominantly deterministic and is founded on the concepts of
defense-in-depth, there is language in the Statements of Consideration
that says risk insights can be useful to making determinations about the
effectiveness of aging management. So it contemplated it.
I would assume from the relative timing of the construction
of the Statements of Consideration in 1995 relative to the struggle that
we were going through at that point to try and develop Reg. Guide 1.174,
its predecessors, and at that I was working on risk-informed tech specs,
I imagine that we didn't want to go too far in speculating about how
that would be done until we had a better about integrated risk control,
which is still an issue, you know, in Reg. Guide 1.174 space as to how
many little bites can you take before you have gone too far.
And so I think that we have not tried to discourage the use
of risk insights, we have just been reluctant to extend them too far
into the design basis. And I think that we can envision a rule change,
after we have gotten through a few of the license renewals, when both
the aging management aspects and research in that area, along with the
potential for developing PRAs in the future can actually measure risk of
You are correct, there is a reluctance to try and spend
money on developing those tools if there isn't somebody that says how
they are going to get a benefit from that.
DR. SEALE: But your perception is that the existing rule is
permissive rather than prohibitive?
MR. GRIMES: That's correct. Because when we look at the
language that says demonstrated effectiveness, and you look in the
Statements of Consideration, and it says risk insights can be used to
make those judgments, we view that as an area that could be explored on
an effect-specific, component-specific basis, not, as the language in
the Statements of Consideration points out, not on wholesale change
DR. APOSTOLAKIS: So you may not even then have to change
the rule, because if the rule is permissive, you can issue a Regulatory
Guide that says, well, here is a way to do it.
MR. GRIMES: That's correct.
DR. APOSTOLAKIS: And that is much easier.
DR. POWERS: I am willing to bet that after you have gone
through a few of these and get a feeling on how license renewal input,
that you will indeed get as many applications as some of the upper
estimates. But if you don't go risk-informed, the agency won't be able
to respond quite as fast.
DR. SEALE: It will drive you out of your mind.
DR. POWERS: Well, it will just swamp the system.
DR. SEALE: Sure, that is what I mean.
DR. POWERS: That, in fact, if the agency doesn't focus its
attention in some way, and the only rational way I know of is based on
risk, they won't be able to process all of them. And I think it is
inevitable that we go risk-informed here, just because the workload is
MR. GRIMES: No, we agree because we see tremendous
efficiencies that can be gained by making smart samples in the review
process. And at the same time we expect that you are going to be
challenging us to have a fully developed technical basis to defend the
findings and that is where we will have an interchange that I think will
be very constructive in terms of where do you think the staff should be
focusing its efforts, and where can we be using more generic findings
relative to the adequacy of aging management programs.
DR. POWERS: I think you certainly saw in the before lunch
session that our lists are going in that direction, but I think, in
contrast to what I thought going into this, it is going to be as a
result of the initial applications rather than prior to them. We are
learning as much from the initial applications as you are, as the
licensees are, as the public is. I think it is going to be coming out
of that. We had all better learn some lessons or we will drown in this.
MR. GRIMES: That's correct.
DR. SEALE: Well, I must say that while I am sorry you
weren't necessarily able to give us the full story, we certainly
appreciate your update. I think we are in a lot better position to
understand exactly where you stand, and also to anticipate what it is
that we are going to have to do in order to be responsive. We agree
that this is a critical item and we don't want to be on the critical
So anything you can do to give us a hi-sign to get up into
the process earlier than otherwise thought, let us know. Of course,
Noel is your contact on that.
And with that, Mr. Chairman, I'll hand it back to you and we
saved some time.
DR. POWERS: I'll just remind Chris that we are committed to
this process and that you can count on us for flexibility in
accommodating your needs. So don't feel like you're imposing on us.
MR. GRIMES: And I appreciate that. And as soon as we get
feedback from -- we're also asking the Commission to approve immediate
release of the paper so that whatever other constructive dialogue we can
accomplish before the Commission meeting, we'll try to --
DR. POWERS: Very good.
MR. GRIMES: We'll try to do that, and we'll work through
Noel to make sure that we make the most effective use of your time and
DR. POWERS: That's very good, and we'll be as flexible as
I want to go off the record now.
[Discussion off the record.]
DR. POWERS: Let's get come back into session and continue
with the Mario Bonaca show. What we have now is the proposed resolution
of Generic Safety Issue 165 that has to do with the spring-actuated
safety and relief valve reliability, and Mario, you are the cognizant
member of the committee, and I guess if your voice holds out, you will
lead us in this discussion.
DR. BONACA: I think my voice will do.
We have now the Staff here to -- you may remember the last
meeting a question was raised regarding the applicability of the sample
used in the INEL study to all the other plants in the U.S. and the Staff
went back and performed a significant review of a large number of units,
and they are ready here to report on that specific question. I believe
that was the main question that the ACRS raised at that time, so I will
let the Staff go to their presentation.
DR. POWERS: Well, I certainly hope we will discuss the
whole resolution. The sampling was simply the Achilles heel that I saw
in the draft proposed resolution, so there's more.
DR. BONACA: That's right, and I understand the presentation
is covering the whole issue.
MR. GORMLEY: Okay. We have the viewgraph package assembled
to discuss the whole issue, so if that's all right, we can proceed in
that thing, and Mr. Cherny will --
MR. CRAIG: Before we get started -- John Craig, the
Director of the Division of Engineering Technology. Owen has been the
Project Manager on this issue for awhile. Frank Cherny has some
expertise in the ASME code and we have a number of support help here
from NRR also that we have been working with on this issue that involves
four-inch spring-operated valves.
When the issue was identified as a result of an event at
Shearon-Harris, and we will talk about that a little bit, the Staff did
a prioritization and the resolution of the issue at that time was going
to be some changes in surveillance, and what you will hear in a few
minutes are that the surveillance that the Staff thought would be
appropriate to address this issue has been incorporated in the ASME code
and it has been implemented in all but, I believe, seven plants, and it
will be implemented in the remaining seven plants in I believe the next
refueling outage, and you will hear a little bit more about that.
Had the ASME not revised the code and the plants not
implemented those changes, the Staff would have gone to great lengths
to, I assume, support a cost beneficial safety enhancement that would
have in fact gone forward to require exactly or in large part what the
ASME code has already done.
In addition to that, we will have part of the presentation,
and Owen will give some insights that we got from looking at different
plant types, et cetera, to look at the valves, some LER searches and
other things, to talk a little bit about the performance of the
With that, I will turn it over to Frank Cherny, who will
start the presentation. Thank you.
DR. POWERS: The bottom line is we essentially -- this issue
is basically resolved through an ASME modification.
MR. CRAIG: Yes.
DR. POWERS: Okay. Good.
MR. CHERNY: Good afternoon. I am going to talk a little
bit about the scope of GSI-165, why we got into prioritizing it in the
first place, a little bit of background on the so-called Shearon-Harris
event, which I guess was the primary motivator for prioritizing the
issue, talk a little bit about the ASME code surveillance requirements
which are gradually being implemented on all the plants in the country,
which we have finally concluded are adequate resolution for this issue.
With that brief introduction, before we ever heard of
GSI-165 the Staff had known that there had been numerous problems over
the years with set point drift on safety relief valves leakage problems,
occasionally a stuck-open valve, and there had been a number of generic
activities and generic issues over the years that have addressed those
types of concerns on the large primary system safety and safety relief
valves and the main steam safety valves on PWRs, so this issue does not
have within its scope any of those.
John mentioned the four inch size. That is kind of a
nominal cut-off size. What we are saying is four inches and less, and
those are, generally speaking, the relief valves that are installed
primarily for overpressure protection purposes on safety-related support
The reason we got involved in prioritizing GSI-165 in the
first place, the so-called Shearon-Harris event was kind of the catalyst
for that, I guess. We completed the prioritization of this issue in
November of 1993 but the Shearon-Harris event took place around April of
1991 and involved severely degraded pressure relief valves which if they
had been asked to perform their function at the wrong point in time a
significant amount of high-pressure injection flow would have bypassed
the reactor vessel and gone out these relief valves.
I guess we have a viewgraph here of the Shearon-Harris
configuration. Let me put that up.
I'm not going to try and pretend to be an expert on the
safety injection system at Shearon Harris, but just to set the stage for
how we got into this situation in terms of the prioritization.
The Shearon Harris folks and a few other plants installed
what is termed this configuration down here, which is an ultimate
minimum flow configuration, which was installed to prevent the three
convection problems, which are the same as the charging problems at most
Westinghouse three-loop plants, from going into a deadhead operation
situation on certain kinds of small-break LOCA situations. During
normal plant operation this valve and this valve are open and provide
mini-flow function for the pumps during normal charging service. When
you have a safety injection system, those two valves isolate, and these
two valves down here open and provide access to these two relief valves.
Now I should say that that's the way the Shearon Harris
configuration used to be. In April of 1991 there was a failure of a
leakoff line right next to one of those two relief valves, the
three-quarter-inch line. Subsequent plant inspections found that not
only was there a failure in that line, but that these two relief valves
were seriously degraded, so much so that if the system had been called
upon for a certain type of event, the system would not have been able to
provide anywhere near its FSAR required design --
DR. WALLIS: You mean it was stuck? What does "degraded"
MR. CHERNY: One valve had a severely cracked spring in it
and was -- both valves had been obviously subjected to --
DR. WALLIS: So they would have opened at too low a pressure
MR. CHERNY: They both would have opened at too low a
pressure, and they would have, you know, the pump flow would have gone
up the valves rather than into the core. That's the basic message.
NRR subsequently investigated and found that there were
about six other plants in the country that had similar configurations.
DR. WALLIS: There's no way the operators would know that
and close those other valves in series? ICS 752.
MR. CHERNY: Well, I guess there are certain indications in
the control room that would give you an indication. The two relief
valves in question relieved to the refueling water storage tank, okay?
That's where the flow would go out.
DR. WALLIS: I think there should be some indication.
MR. CRAIG: There's certainly level indication and alarms in
the reactor water storage tank, and as it got high, they would certainly
get that. There are also some charging flow lines that give you
charging flow rates, and I think it's pretty straightforward that they
would be able to tell the water was being diverted.
MR. CHERNY: Subsequently what happened at Shearon Harris
was the two relief valves that are shown here were replaced with flow
orifice limiting devices rather than valves. The other plants that had
similar configurations for one thing had much smaller relief valves than
what were being used on Shearon Harris, and through a combination of
increased surveillance and inspections, they were able to resolve any
remaining concerns on those. So this particular configuration that
started the concerns for GSI-165 really has been eliminated, is not a
concern specifically on any of the plants.
However, the concern about diversion of flow through these
auxiliary-type relief valves motivated the prioritization of GSI 165 in
the first place. As far as the -- I can put this up. I think we've
talked about most of this already. That's the wrong viewgraph; this
MR. GORMLEY: Which one? This one?
MR. CHERNY: No, where's the Shearon Harris viewgraph?
MR. GORMLEY: That's the first one.
MR. CHERNY: There was an information notice that was sent
out in August of 1992, Information Notice 92-61, and subsequently there
was a supplement to that information notice which was sent out in
November of 1992. The supplement did in fact show the modified
configuration with flow orifice devices on it. And as I said before,
the plants that had similar configurations have implemented various
kinds of changes to correct those concerns.
That's all I really intend to say about Shearon Harris per
Just repeating slightly, the relief valves thus that are in
the scope of GSI 165 are the small spring-operated relief valves less
than four inches in size. When we prioritized this issue, there were
two key assumptions that resulted in this issue coming out as a
high-priority issue. There was one assumption that was made that the
relief valves -- that 10 percent of the relief valves installed on
safety-related support systems had the relief capability to fail their
trans. That's a fairly high number. That's what was assumed. That was
a judgment call. A fair number of systems and PRA people were involved
in those discussions, but at the time of the prioritization, a lot of
resources weren't spent on studying P&IDs and FSARs and things like
that. So we made that 10 percent assumption.
The other thing that was done was we made an assumption that
an increased testing frequency somewhat in excess of what was about to
be implemented in 10 CFR 50.55(a), that slightly increased testing
frequency for testing that could be easily and economically performed at
the plant sites could be performed on a small set of
high-safety-significant relief valves to resolve this issue.
Now just about at the time that we were sitting down to
prioritize the issue, there was a revision made to 10 CFR 50.55(a). I
think it became effective in September of 1992. And it for the first
time required through the 120-month update process, something that you
all are rather familiar with from a briefing not too long ago, required
that gradually the plants implement new ASME code testing requirements
for these types of valves. That revision of 50.55(a) endorsed for the
first time in the 1989 edition of ASME section 11, that revision of
section 11 endorsed an ASME O&M testing standard, Part 1 of ANSI ASME OM
1987, which for the first time has testing requirements in it for these
types of valves.
And just to describe briefly what they involve, it requires
you to test as a minimum set pressure test, leakage tests, and good
visual inspection a minimum of once every ten years, which isn't a
fairly high frequency, but there's an additional requirement. You're
supposed to group these valves in accordance with manufacturer and model
number types, and you're supposed to test a minimum of 20 percent of
valves in such a grouping every four years. So if you have -- what
really happens is if you have small numbers of valves in the groups,
they really get tested more often than once every ten years when you
actually try and apply the code.
What we had assumed when we did the prioritization was we
said well, they've got this once-every-ten-year test requirement that's
just coming into play. Supposing we have a couple really
risk-significant relief valves that we find as we go about doing our
study, what would be a good testing frequency. And I think what we used
at that time was we said well, we'll go from once every ten years to
once a year. So that's sort of what was used as the model for the
DR. WALLIS: Well, the original -- excuse me, the original
problem at Shearon Harris was thought to be the result of ECCS testing.
MR. CHERNY: That's correct.
DR. WALLIS: Testing actually was the cause of degrading the
MR. CHERNY: In a sense. They had a very strange
configuration there, though, too. They had a situation where they had
-- they were subject to a lot of water hammers when those valves
actuated, they had trapped air pockets in downstream and upstream
DR. WALLIS: So it was --
MR. CHERNY: But you're right, the testing actually
instigated those failures, but so did the design.
DR. WALLIS: So this is another case of water hammer.
MR. CHERNY: That was -- at Shearon Harris it was a water
hammer, severe water hammer problem.
The ASME code requirements --
DR. WALLIS: So the real cure is to prevent the water
hammers. The first action would be if you're going to have valves like
this, make sure that there aren't any potentials for water hammer in the
MR. CHERNY: Um-hum. There are other valves that over the
years have -- relief valves that have been prone to water hammer
problems. I think the most famous one is the case of some of the PWR
pressurizer safety valves that have purposely installed water slugs in
the inlet piping to the valves. And as a result of a TMI action plan
item, the design of all of those configurations was revisited right
after the TMI 2 accident, and additional piping supports were put in,
valve qualifications were performed, and those were -- believe me, those
were fixed for a lot of money, but they're fixed.
With regard to the ASME code requirements that became
effective in that rulemaking in September of 1992, we have gone through
a number of years now and, as it turns out, those code requirements have
been implemented in most of the plants in the country. There are about
seven plants left that are due to implement those new -- those
requirements at their next refueling outages. After --
DR. WALLIS: I am still puzzled, I'm sorry. Why do you test
the valves when the cause of the problem is the waterhammer?
MR. CHERNY: Well, we are testing -- we are testing the
whole population of valves, not just the Shearon Harris.
DR. WALLIS: The cause of the problem at Shearon Harris was
MR. CHERNY: Right. And they did a design change.
DR. WALLIS: Why did this lead to paying attention to the
valves? The valves could have been 100 percent perfect in every way.
MR. CHERNY: Most of the valves are not installed in those
kind of configurations. That was a very strange configuration that they
had there. In addition, that was not --
DR. WALLIS: It would seem to me the generic safety issue is
waterhammer, not valves.
MR. CHERNY: Now, as I said before, the Shearon Harris
problem was fixed before we ever really started doing anything on the
GSI. The concern about the GSI was whether the failure of other relief
valves and support systems could defeat something like ECCS functions.
DR. WALLIS: Okay. So it is not just the valve, it is the
whole system and the way it behaves that you are worried about .
MR. CHERNY: Right. Right. But the focus was on the
contribution from the relief valves, for this issue, that is the scope
of this issue. Okay.
DR. WALLIS: So there wasn't an issue on the waterhammer?
MR. CHERNY: There was a separate waterhammer issue some
years ago, but that was not the scope of this issue.
DR. WALLIS: I am still not quite clear why this became a
GSI at all, apart from the waterhammer.
MR. CHERNY: Shearon Harris did not become a GSI. Shearon
Harris was just a motivator to take a look at relief valves and support
DR. WALLIS: Why?
MR. CHERNY: We do not need generic issue for Shearon
Harris. Shearon Harris was --
DR. WALLIS: So there was some other reason why the valves
became a generic safety issue?
MR. CHERNY: People were concerned about the number of LERs
reporting all kinds of different kinds of valve problems.
DR. WALLIS: Other problems with valves.
MR. CHERNY: Release valves, ring adjustment problems, set
point drift problems, leakage problems.
DR. WALLIS: Other problems with valves.
MR. CHERNY: Premature actuations of valves that had nothing
to do with waterhammers.
DR. WALLIS: Okay.
MR. CHERNY: And that is why the focus, too, was on the ASME
code testing for these kind of valves. Before the September 1992
rulemaking, there was no requirements to test these kind of valves at
all, nobody had to do any tests on them. With the putting in place of
that rulemaking, surveillance requirements for the first time on these
kinds of valves were put into place.
And as I was saying before, there are seven plants left in
the country that at their next refueling outages will be implementing
these requirements for the first time. After those seven plants
implement those requirements, all the plants in the country will be
doing regular code required surveillance testing on those types of
What we have done as part of the resolution process we have
done some PRA work and some additional deterministic work which has led
to the conclusion that the additional increase in frequency of testing
that we postulated when we did the original prioritization is no longer
necessary, and we have concluded that the surveillance tests that are
already in place with the September 1992 rulemaking are adequate to
address the GSI 165 concern.
And Owen is going to talk now a little bit about some of
that additional work that was done to allow us to reach that conclusion.
MR. GORMLEY: Okay. Thank you, Frank.
DR. WALLIS: Well, one way is to fix the valves and make
sure they are reliable. The other way is to make sure that if they do
fail, the operators know it and they have an action which they can take
to close some other valve or something.
MR. CHERNY: He is going to talk about some of that.
DR. WALLIS: About some of that.
MR. GORMLEY: I am not going to talk about that. I think
most of the systems have some kind of a flow measuring device, an
orifice or have a tailpipe temperature that tells them that the valve is
passing flow. Now, my recollection is that the tailpipe temperatures
are not alarmed, but we are relying on the operator to --
DR. WALLIS: Believe it. Was it TMI it had been leaking?
MR. GORMLEY: Yeah, believe it or look for it.
DR. WALLIS: Believe it.
MR. GORMLEY: Yeah. Let's see, a bad thing happened here.
As you recall from our discussion last month, we initially
looked at five plants and found that in most of the plants there were no
valves that were of any concern at all. We looked at one of the
Combustion Engineering plants and found a valve in it that seemed to be
bigger than what would be required for simple thermal relief -- thermal
relief valves which are in there to comply with ASME code requirements
to protect dead legs of the plant, by far the majority of the valves
that are used for this purpose, and so we analyzed that additionally.
What we found is that the valve has the capability to direct
somewhere between 30 and 50 percent of the available flow, depending on
what kind of assumptions you make about how far the valve opens, what
degree of overpressure or accumulation you allow.
It isn't clear that that does not actually fail the train
and that is something that we need to keep in mind as we go through
this. In PRA space you only have the option of it is failed or it isn't
failed, but in fact in a case that we just examined today if two valves
fail, one in each train, the combined trains still provide enough flow
to keep the clad temperatures within acceptable limits.
DR. WALLIS: These are high pressure injection?
MR. GORMLEY: Yes. Yes, so having the valve fail completely
open is not -- doesn't actually constitute failure of the train. We
examined that train or we examined that plant with a Sapphire analysis,
and we found that having that valve fail with an 8.6 percent failure
rate produces an acceptable CDF.
DR. WALLIS: That is 8.6 percent of the times when it is
called upon to open --
MR. GORMLEY: No.
DR. WALLIS: -- or it is challenged --
MR. GORMLEY: That it fails -- prematurely opens 8.6 percent
of the time after the pumps start.
DR. WALLIS: You took that calculation from the acceptable
increase to get 8.6?
MR. GORMLEY: No. This is a failure rate calculated from
DR. WALLIS: You had reasons to believe that 8.6 was the
right number to use?
MR. GORMLEY: No, 8.6 is a number that we can support
technically. It is wildly over-conservative.
For these valves to have any effect on the diversion of
flow, they have to fail wide open. The failure rate data contains
results of all of these ASME tests where all you do is approach the set
point and see if the valve cracks open.
Where we looked at the LER data, we'll come to that but when
we looked at the LER data, we didn't find any circumstances where the
valve springs broke -- also valve springs would have to break in order
for the valve to fly wide open.
MR. CHERNY: Except for Shearon-Harris.
MR. GORMLEY: Well, yes, that's right.
DR. WALLIS: Was there a water hammer in these lines in this
MR. GORMLEY: Well, we didn't find any examples when we
looked at the LERs.
DR. BONACA: So that CDF, really it is --
MR. GORMLEY: Probably two orders of magnitude too high.
DR. BONACA: Well, that means that typically you do a PRA to
look at best estimate or as realistic as you can, but this is really --
there is an assumption there that is far from realistic.
MR. GORMLEY: And that is the problem. It is not an
assumption. It's a calculated value.
DR. BONACA: That's because you haven't set a rate for
frequency of failure wide-open, therefore you assume the very
MR. GORMLEY: It's the calculated rate that we have data to
support and unfortunately the data contains all of these minor failures
to meet or set-point drift failures.
DR. WALLIS: What is the acceptable increase in CDF?
MR. GORMLEY: We have been going with 1.5 times 10 to the
minus 5 -- sorry, one times 10 to the minus 5.
DR. WALLIS: So if one had a price on this increase in CDF
instead of just saying it is acceptable, it might well be that we were
spending the money to prevent it happening.
MR. GORMLEY: We believe that the money is being spent to
prevent it from happening.
DR. WALLIS: Though it's acceptable, you are still spending
MR. GORMLEY: Because ASME has invoked the testing.
DR. BONACA: Still it gives me concern with using, you know,
that kind of failure rate when it is not really applicable to the
condition you are examining, because assume that rather than getting six
in 10 to the minus 6 you calculated two in 10 to the minus 5, and you
may conclude that you have a generic issue while you don't have it.
MR. GORMLEY: Yes. Yes. Except in this case we concluded
even though we were using the high rate, we concluded we did not have a
generic issue. If we had concluded that we had a generic issue, then we
would go back to work. Probably the first place would be to go to work
on the failure data. I think that is the most productive area, but it
is an intangible that I wanted to bring to your attention.
DR. BONACA: How far does the NPRDS database go? I am
trying to understand this.
MR. GORMLEY: Oh, we went back to, I believe, 1990 or 1989
when the additional testing requirements began to be imposed and the
data in the database improved.
DR. BONACA: Is there a case you may remember when one of
these failures occurred?
MR. GORMLEY: Well, that is the problem. You can't find
out. What we wanted to do was weed out all of these minor set-point or
seat leakage problems. There is not enough information in there that
allowed us to do that, so --
DR. BONACA: So you are telling me that you cannot find --
you cannot really quantify a failure rate for the wide-open failure of
these, so the failure rate is low as a minimum --
MR. GORMLEY: It is very high. It is a very high failure
rate that we have used.
MR. CHERNY: Yes, but the failure rate he is asking about is
what is the failure rate for a seriously failed condition --
DR. BONACA: Yes.
MR. CHERNY: And that is a number we don't have.
MR. GORMLEY: We have gone about that in a different way.
We tried to find valve spring failures and we didn't find any. I asked
the manufacturer, a fellow who told me he worked there for 20 years and
had not seen any valve spring failures.
DR. BONACA: Okay.
MR. GORMLEY: There was a Braidwood LER that came in after
we had done this work. The incident occurred I think last year where
they had an air cushioned water hammer and the valve was only a 20
gallon a minute valve, but the incident was of interest to me because
there was a deformation of the spring that took place. In that
particular instance, it lowered the set-point pressure by about 10
percent, 5 to 10 percent actually.
In the case of these Combustion Engineering plants, the
valves are set for about 2500 psi, and the pump shutoff head for the
HPSI is down in the 1200 to 1400 psi range, so there would have to be a
really significant failure of the spring in order for the valve to fail
DR. WALLIS: How do we get assurance that there isn't going
to be a waterhammer?
MR. GORMLEY: I don't know what the answer to that question
is. You can be sure that it won't happen again at Braidwood. They
expended a great deal of energy, the whole thing was very mysterious to
MR. IMBRO: Let me try and a little bit. I think that this
is -- my understanding with the Shearon Harris, the waterhammer was
really a result of the valve chattering. And as Frank Cherny pointed
out, the design at Shearon Harris was very unusual, where they actually
use these relief valves in place of orifices, which most plants use for
the mini-flow, for the pump mini-flow. So these valves basically were
always called upon when the pump started to lift somehow.
So, but what happened at Shearon Harris, as I understand, is
that they kind of set up a cyclical problem, where as the pressure
increased, the valves opened up, and then as soon as the pressure
dropped, valves would close again, and that was within the operating
range of the pump. So you set up kind of a chattering effect on this
valve, and you had the fluid column moving back and forth which caused
So, again, the Shearon Harris is a very special, special
case because of the somewhat unusual design of the system. So we would
not really expect waterhammers to occur. Most of these valves we are
talking about here are relief valves for overpressure protection from
thermal expansion and that type of thing, so, really -- and they set
usually 110 percent of the design pressure of the piping to protect the
piping, and you never really see a situation with pump operation where
you would lift these valves, or very infrequently anyway, I will say
MR. SIEBER: Isn't it true, though, that when you have a
configuration with a motor-operating valve followed by a relief valve,
that while the motor-operated valve is opening, you create an orifice
effect which makes the relief valve chatter? I know of applications --
MR. IMBRO: It is possible, I am sure.
MR. SIEBER: -- at some plants where it actually will bend
the stem, and then the valve with stick because the stem is bent. And
that is a design problem in configuration.
MR. IMBRO: Right.
MR. SIEBER: The way the pipe is laid out. And that sounds
like what this is.
MR. IMBRO: Yeah. Again, you know, the Shearon Harris thing
was very unusual. I have never really seen any other examples of that
type of situation, although they say there are a few others out there
like that. But, again, that was the focus of the generic issue was not
so much the issue as to Shearon Harris, but basically, given that we had
a valve -- a failure of a relief valve, was to look more broadly at
relief valves used in systems to determine whether or not a failure, any
type -- you know, caused by anything, waterhammer or whatever, would
really result in the ECCS system or a system not being able to perform a
That was really -- the focus wasn't on the waterhammer or
the specific event that Shearon Harris, but was really on the focus of
let's look at the safety valves less than four inches and see whether or
not -- if one of those would fail or could be somehow degraded, would
this create a situation where the ECCS, or whatever system it is in,
would not perform its safety function. So that is really the focus on
MR. SIEBER: Well, the only reason why I mentioned that is
that if the root cause is really configuration related causing valve
chatter, then testing will never solve it.
MR. IMBRO: Yeah, I agree, certainly.
DR. WALLIS: How about orifices, just a hole, instead of a
valve, you have a whole?
MR. GORMLEY: Yes.
DR. WALLIS: So what happens when there is no -- when they
are not functioning? What is there, just air or something? What is in
MR. GORMLEY: Water, it is still filled with water. Yeah.
DR. WALLIS: Going nowhere, just sitting there?
MR. GORMLEY: Yeah.
DR. WALLIS: Into the atmosphere?
MR. GORMLEY: Yes.
MR. SIEBER: Well, no, it goes to the RWST.
MR. IMBRO: It goes to the RWST. Right.
MR. SIEBER: Comes from and goes to, so there is no change
DR. WALLIS: You have got to keep that orifice covered with
water, or otherwise water backs back down the pipe, and you turn on the
MR. GORMLEY: I see, yes.
DR. WALLIS: -- it will blow out the orifice with a
MR. SIEBER: If the levels are high.
DR. WALLIS: It seems to be very critical.
MR. GORMLEY: Yes. You are talking about flow measurement
DR. WALLIS: I am talking about keeping the orifice full of
MR. GORMLEY: In which orifice? At Shearon Harris?
DR. WALLIS: If you replace the safety valve with an
orifice, right to the outside world, you have got to be sure that it is
covered with water, otherwise when you turn the pump on, you get
waterhammer -- the orifice.
MR. GORMLEY: I believe that the whole system is filled with
water at all times, and when it goes back to the reactor -- refueling
water storage tank.
DR. WALLIS: It's all full, all the way down.
MR. GORMLEY: Yeah.
MR. SIEBER: Right.
MR. IMBRO: Yes.
DR. WALLIS: Okay.
MR. GORMLEY: The problem at Braidwood was that they weren't
getting the air out of the valve when they replaced it, and so they
immediately found it during post-maintenance testing. And then they put
on another valve, which they also didn't get the air out of, and they
did that about five times before they found it, and it was very
mysterious to them, which is why I say it is not going to happen there
DR. WALLIS: They really know now.
MR. GORMLEY: Yes, they do.
DR. POWERS: Replicate testing, replicate testing. I keep
telling you, this is very important. Sometimes it's carried to
extremes, but --
MR. GORMLEY: And now the question of the day. The last
time you said gee, only five plants, so I went back and I looked at a
number of others. I went to the, you know, those charts in the
contractor's report where it said which the more important systems were,
and I picked out first of all the plants that had high numbers, and also
from inspection I concluded that the HPSI, LPSI, and in the GE plants
the core spray were the important systems, even though there seemed to
be no systems in the GE plants that really were as important as in the
So I looked at -- the other thing I noted was that the LPSI
in the B&W plant seemed to be quite important. So I looked at almost
all of the B&W plants. I looked at a large fraction of the CE plants,
and four other plants. And what I found was that there are no valves in
those plants that will cause the system to bypass a significant amount
In the process I did find one plant that had a large,
oversize relief valve on each HPSI loop, but that's only a single plant,
and really doesn't factor into the generic -- into the generic issue
resolution. Even in that plant, you remember I was saying earlier, in
that plant even if both valves do fail, you still get enough flow into
the core, you get enough partial flow from each loop into the core to
meet the clad temperature restrictions.
I'm hoping that that's going to be enough -- looking at
enough plants to --
DR. POWERS: I think the most important thing in your
sampling here is that you, by doing 24, that you have gotten away from
those plants that have been so thoroughly examined --
MR. GORMLEY: Yes.
DR. POWERS: That it never seems to happen, but they scrub
out so many things, and we're always worried about the ones that haven't
had that intensive examination. I congratulate you on 24. That wasn't
as necessary as getting away from those five representative plants, or
at least arguing why they were indeed representative. I think moving
off those constitutes better sampling than that. Just because everybody
and his dog has done Surry.
MR. GORMLEY: Well, the thing that drives you there is if
you're going to have --
DR. POWERS: Got the information.
MR. GORMLEY: PRA. Yes.
DR. POWERS: Yes. You've got the information there. And
that's a real good indicator, except it's not representative of all the
MR. GORMLEY: Yes.
DR. POWERS: And now you've taken -- it seems to me you've
solved this problem six ways from Sunday now.
MR. GORMLEY: Great.
DR. POWERS: Looked at a bunch of plants, you've got an ASME
code testing requirement. I mean, this has really solved it. But for
the sampling issue, when it comes up elsewhere, I think the issue is how
representative are those plants that we called representative back in
'86 when we did 1150. They aren't anymore, because they've been looked
In the case of Surry, they were looked at in WASH-1400.
They were looked at in NUREG-1150. They were looked at in the IPEs.
They've been looked at so many times and so scrubbed that they just
really aren't representative of the plant population anymore.
DR. WALLIS: So 24 is enough. It always puzzles me, what's
an adequate sample size when you're resolving these issues?
DR. KRESS: Twenty-three is not.
DR. POWERS: Well, you can actually -- you can actually sit
down and come up with some rational view on that if you believe that
these things are random. And I think sampling roughly a fifth of the
plants gives you a fairly high probability, especially when it's more
than a fifth of the plants, because he left out the sister plants. He
didn't need to look at them again.
DR. WALLIS: As long as the sisters had the same
architect-engineers or something, because, you know, the details of each
plant are different.
DR. POWERS: Well, the sisters are going to be fairly close
on these issues.
I mean, in the case of this issue, it didn't matter. All of
them are doing the testing. And the only question you're addressing
here is is that testing in the SME sufficient for regulatory purposes.
DR. WALLIS: I guess what we're saying is you don't need
assurance that there's no plant out there, all you need is assurance
that this is not a generic issue.
MR. GORMLEY: Yes.
DR. WALLIS: For the generic issue, this is probably a
perfectly good sample size.
MR. GORMLEY: And we talked last time we looked at the LER
data base and didn't come up with anything. We also touched on the
question of the always meeting the single-failure criterion no matter
what the situation is with the valve.
DR. POWERS: It seems to me your second bullet there is yet
more information, that is an inferential test.
MR. GORMLEY: Yes.
DR. POWERS: So you have got this explicit testing at fairly
infrequent intervals, and then you have this very frequent inferential
MR. GORMLEY: And there is also an annual -- no, a once a
refueling test of the ECCS system as a whole.
DR. POWERS: Sure.
MR. GORMLEY: Let's see, I don't know, do we want to go over
again the -- go back over the details of the examination of the five
plants. We ran the PRA, looked for the important systems, got that --
found that large valve on the CE plant and calculated an acceptable CDF.
Was there anything else we needed to talk about or not?
And the last viewgraph in your package is just what it was
that I did to select the 24 plants, the additions to the 24 plants. If
you don't have any questions on that, it is sort of irrelevant at this
DR. POWERS: Not on this particular GSI, but as a generic
question? I hesitate to use the word -- you come up with a delta CDF of
6 times 10 to the minus 6th and you are looking at a threshold of 1
times 10 to the minus 5th, and both numbers are uncertain by factors of
3. Have you set that threshold of 1 times 10 to the minus 5th through
some recognition of the uncertainties in the numbers?
MR. GORMLEY: Harold. Harold Vander Molen.
MR. VANDER MOLEN: I would like to find the person that put
these TV sets here. Can I have that question repeated once again? The
TV distracted me.
DR. POWERS: He looked at his analysis and he said, gee, the
delta CDF is 6 times 10 to the minus 6th. I am looking for a threshold
of 1 times 10 to the minus 5th. Those CDF numbers when they are
calculated have uncertainties that are arguably are perhaps a factor of
MR. VANDER MOLEN: Probably larger than that.
DR. POWERS: Maybe larger than that. And so I am looking at
a delta and I see it is below my threshold, and that's fine if you have
chosen the threshold such that you have recognized the large
uncertainties in the numbers in this bracket.
MR. VANDER MOLEN: This is a discussion that goes all the
way back in the safety goal days, and I can't say that I remember all of
it off the top of my head, but the goals and these criteria were set on
the mean of the distribution intentionally. And the reasons for it, we
would have to go back into some of the studies that have gone on in
decision theory. But, yes, the uncertainties were consciously taken
into consideration when these things were set up.
In the case of generic issues, the thresholds were adjusted
a number of years ago, but even the thresholds we used, even in the
prioritization case, consciously included some allowance for the
uncertainty in the numbers.
DR. POWERS: But it seems to me, if I subtract two means and
I get a number of -- that are uncertain by a factor of 2, I get a number
that the uncertainty, the difference of those two means is roughly a
factor of -- 5? Okay. So, and then that is fine. So what you are
saying is that delta in the mean, I have to know to a confidence level
that is very small, in the sense of I am only 50 percent confident that
the mean is that small.
MR. VANDER MOLEN: Well, if I am understanding your
statement correctly, you are really saying that you are taking the
difference between two fairly large numbers, each with an uncertainty,
and of course the delta between them is going to be fairly small, but
the uncertainties are going to be just as large as they ever were, so
that the relative uncertainty of the difference is going to be fairly
high, am I characterizing it correctly?
DR. POWERS: That's right. That's right.
MR. VANDER MOLEN: It is a little bit more complicated than
that. We can -- and are looking into ways actually of calculating an
uncertainty distribution for the delta. Now, it is not easy to do with
our codes. What you really have to do is -- it is not built into our
current PRA codes at all. The closest thing we can come to is to
generate a series of calculations for the base case and another one for
the adjusted case, with and without your proposed fix for the issue, and
save each calculation, starting them both with the same random number
seed and then form a distribution for the delta. Hopefully, and, mind
you, we are still looking into this, -- this is not an easy thing to do,
-- hopefully, we will be able to form an uncertainty analysis directly
for the difference.
DR. POWERS: And what you end up saying in this particular
case is at some confidence level, you are -- you can assure that the
delta CDF was less than the 1 times 10 to the minus 5th threshold, and
that confidence level is not very high, because the uncertainties are
MR. VANDER MOLEN: Yes, but they are always are in PRA. I
wish they were not, but that's the best of what we have.
DR. POWERS: But we are stuck with that.
MR. VANDER MOLEN: Yes.
DR. POWERS: My question, when you find the delta CDF is
close but not above the 1 times 10 to the minus 5th threshold, how close
does it have to be before you say, well, it really probably is, there is
a high probability that the delta exceeds my threshold?
MR. VANDER MOLEN: We don't have strict guidelines for that.
We do -- if you will look at our policy papers, we do leave ourselves a
little bit of judgmental wiggle room. If we really think, based on our
engineering judgment, that there is a real possibility of a safety
problem and that things might be a little higher, we will continue to
take action or investigate further trying to bring the uncertainty
bounds down a little bit. Short of doing vast improvements of PRA, I
don't think we can do too much more.
DR. POWERS: It is just merely having a confidence level
that you command.
MR. VANDER MOLEN: Yes. In many cases we look at a lot of
these issues, and we have been through over 400 of them so far. In many
cases, it is not near the threshold, it is fairly -- yes, exactly. And
that is pretty clear cut.
But we do have the occasional issue where it is close, there
is no question about that.
MR. GORMLEY: Well, the other thing is that the calculations
are done in a fairly consistent and routinized manner. And you can make
the argument that while the results themselves aren't very good, perhaps
the difference between them might be better.
MR. VANDER MOLEN: Actually they often are.
DR. POWERS: I have heard people make that --
MR. VANDER MOLEN: If you have an issue -- I'm sorry.
DR. POWERS: And I have never yet seen a case where that is
DR. POWERS: Where it is verified. I mean I have heard the
argument that changes in a PRA are more accurately known than in one
single one, but every time I do an uncertainty analysis, the deltas
always have bigger relative errors than --
MR. GORMLEY: It sounds to me like a correlation of the --
if you can't be right, be consistent theory.
DR. POWERS: Yeah. That was just strictly for my
understanding, it didn't have anything to do with this issue. This
issue is, to my mind at least, effectively and very thoroughly resolved
DR. WALLIS: Well, I am trying to follow the logic here and
this CDF change is less than some criterion for one plant, therefore,
there isn't a generic issue. And even if it were one plant, that still
doesn't make it a generic issue. It just means you have got to fix that
DR. BONACA: In fact, it is the case here. Imagine that
there would be communication to the CE plant that they have this
configuration, or is there? What action -- you mentioned at some point
MR. GORMLEY: Yes. We will be consulting with NRR and they
will be deciding what action to take relative to the plant with the two
valves. With the plant with the one valve, we just used as the model
for the calculation, it was a surrogate for all plants, it was like the
worst case. So the fact -- our approach was if we can show that this
plant with the large valve in it meets the CDF criteria, then we can be
sure that all the rest do.
DR. POWERS: Ted, suppose that calculation had come out 1.6
times 10 to the minus fifth, suppose it had come out 5 times 10 to the
minus fifth, what conceivable else could you have done?
MR. GORMLEY: Oh, I would have gone right after the failure
DR. POWERS: Well, you would have refined the CDF
calculation, but suppose that after you had done that it still didn't
work? Then I seems to me that you have resolved the issue because you
are testing it directly. You are testing it inferentially and there is
not much else you can do on that thing.
MR. GORMLEY: Right. Well, that's right. First of all, the
testing improves the failure rate for you, but we would be trying to
find some way to beg the issue on the failure rates.
For example, we calculated 8.6 times 10 to the minus 2. The
overall failure rate in the San Onofre plant for relief valves is 8.5
times 10 to the minus 3. There is an order of magnitude right there.
The difference between premature opening and failure to
close is the difference between 8.6 times 10 to the minus 2 and 1.5
times 10 to the minus 2, so there are things that we could have done to
sharpen the pencil on failure rates but the more we deviate from the
hard data the more difficult it is to sell the result.
MR. CRAIG: But the point is for this issue, even if we had
come up with a large enough CDF and a low enough cost, the action that
we would have put forward is the action that has already been addressed
by the ASME and it has already been implemented in the plants, so I am
sure the ASME will find some satisfaction that we have done -- we would
have done some analysis that might support from a cost benefit
standpoint changes they have made, but the changes have been made and
put in, so it saves us from going forward.
DR. POWERS: All the probabilistic analysis did was give you
comfort that you didn't need any additional testing --
MR. CRAIG: Right.
DR. POWERS: But in fact you already have additional testing
in an inferential nature.
MR. CRAIG: That's right.
DR. POWERS: Instead of a direct nature, so you have
belt-and-suspenders resolution to this issue, it seems to me.
MR. CRAIG: At least, yes.
DR. POWERS: A belt, suspenders and whatever else -- maybe a
MR. CHERNY: I might just add as far as the code testing is
concerned that when we first started working on this issue I did talk to
the chairman of the ASME committee that has written the code
requirements that we have been talking about, and I told him that there
was a possibility that as we went through the resolution of this issue
that we might identify a couple risk-significant relief valves that may
be, you know, we might be able to provide some justification for more
frequent surveillance testing than the once every 10 year thing that we
had already put in the code, and he was very receptive to receiving that
kind of information and to adjusting the code requirements accordingly
if we could provide some basis for so doing.
Well, as it turns out, we haven't been able to find any
basis for that.
DR. WALLIS: So this happened in -- when was this called a
GSI, in '92?
MR. CHERNY: We finished the prioritization I think in
November of '93 if I have the date right.
DR. WALLIS: So there were six years where no one quite knew
whether this was important or not?
MR. GORMLEY: No. We have been working on it for three
years and everything that we look at comes up as not being meaningful in
any way, so there was -- you know, when you discover that, there's no
rush to completion, and frankly, we thought that since there's some
uncertainty about it we thought that it would be a good idea to do all
of these other investigations.
We started out we were only going to do two plants, and we
ended up, you see, with 24.
DR. WALLIS: So an external perception, it looks good to
dispose of these GSIs as soon as possible?
MR. GORMLEY: Yes, indeed.
DR. WALLIS: Avoid even hanging around -- the more there's
the suspicion there might be something to them and that early resolution
does have a benefit.
MR. IMBRO: Let me just add -- this is Gene Imbro again with
the Staff -- with regard to the issue on the relief valves for the CE
plants, I just want to point out, you know, our management meets
periodically with the CE Owners Group, and this certainly is an issue
that we can converse with them and try to achieve some kind of -- get
some industry action on this.
DR. WALLIS: This Slide 9 says testing of most pumps. Do
you mean testing of the valves or --
MR. GORMLEY: No, actually pump testing --
DR. WALLIS: System testing, isn't it?
MR. GORMLEY: Well, yes, it basically is system testing.
DR. WALLIS: No way of testing a pump as --
MR. CHERNY: No, there is a quarterly ASME code pump test
that is performed and you have to measure the pump flow when you do that
test and it is confusing.
DR. WALLIS: With the valve test at the same time?
MR. CHERNY: In the sense that if the valve lifts and you
don't get your full pump flow, you are going to start looking why don't
you have it.
DR. WALLIS: So everything is fine?
MR. GORMLEY: Yes, sir.
DR. BONACA: Any additional questions?
MR. CHERNY: No. As I said earlier, we have had other
generic activities that have looked hard at those bigger valves over the
DR. UHRIG: Not a problem?
MR. CHERNY: There were problems. There aren't anymore.
There were a lot of problems in the early '80s with some of those valves
which have since been fixed. There were ring adjustment problems, there
were chattering problems, there were materials problems. There was a
whole flock of problems with pressurizer safety valves right after the
TMI-2 accident that were fixed.
DR. WALLIS: Let's go back to what came up in the discussion
though. It seems as if the failures or the events that concern people
had something to do with water hammer rather than just a valve and the
water hammer does come up an awful lot of the time.
MR. CHERNY: Could I try that one? I have been involved in
ASME code activities for a number of years and the practice -- two
comments on that.
The practice that all the engineering firms follow for water
hammer is as much as possible to design not to have water hammers, okay?
That is the first thing.
The other thing is we did have a USI, an Unresolved Safety
Issue, back in the early to mid-eighties. I have forgotten the number
of it. I think it was A40 something-or-other, which we studied water
hammer on these plants extensively, and we finally, even though it was
classified as a USI we came out, as I recall, not being able to justify
any new requirements, and that was extensive studies of that all over
the place on that issue.
DR. WALLIS: So there isn't a requirement then? We have this
other case of the fire system had a water hammer and it seemed that all
the fire codes had to do with fires and didn't say anything about
designing the system so it doesn't have a water hammer, which is an
MR. CHERNY: If a system has a water hammer problem, the
water hammer problem should be fixed.
It is like anything else. You try your best to design those
things out of the system. Every now and then one slips through, like
the Shearon-Harris example we were talking about a few minutes ago.
DR. BONACA: Does anybody else have any other comments?
DR. BONACA: If none, I will turn it over to you, Mr.
DR. POWERS: Thank you very much. I certainly thank you for
the presentation. It was again, to my mind, it sounds like a
belt-and-suspenders type of resolution to me.
MR. CRAIG: Good. Thanks. We hope everybody agrees with
that. Thank you.
DR. POWERS: At this point I think we can go off the
[Whereupon, at 3:35 p.m., the meeting was recessed, to
reconvene at 8:30 a.m., Friday, June 4, 1999.]
Page Last Reviewed/Updated Tuesday, July 12, 2016