473rd Advisory Committee on Reactor Safeguards - June 7, 2000
UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
***
MEETING: 473RD ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
Two White Flint North, Room T2-B3
11545 Rockville Pike
Rockville, MD
Wednesday, June 7, 2000
The committee met, pursuant to notice, at 8:30
a.m.
MEMBERS PRESENT:
DANA A. POWERS, Chairman
GEORGE APOSTOLAKIS, Vice-Chairman
JOHN J. BARTON
MARIO V. BONACA
THOMAS S. KRESS
ROBERT L. SEALE
WILLIAM J. SHACK
JOHN D. SIEBER
ROBERT E. UHRIG
GRAHAM B. WALLIS. C O N T E N T S
ATTACHMENTS PAGE
Introductory Statement 3
Agenda 3
Briefing to the ACRS on the
GSI-173A 12
Draft Report - Regulatory Effectiveness of the
SBO Rule 69
Final Regulatory Guide 1.183 (DG-1081) 122
Response to SRM on PRA Quality 154
. P R O C E E D I N G S
[8:30 a.m.]
CHAIRMAN POWERS: The meeting will come to order.
This is the first day of the 473rd meeting of the Advisory
Committee on Reactor Safeguards.
During today's meeting, the Committee will
consider the following: Proposed Resolution of Generic
Safety Issue 173-A, Spent Fuel Storage Pool for Operation
Facilities. I'm sure this will prove to be totally non-
controversial; Regulatory Effectiveness of the Station
Blackout Rule; Proposed Final Standard Review Plan Section
and Regulatory Guide Associated with the Revised Source Term
Rule; and Assessment of the Quality of Probabilistic Risk
Assessments.
The meeting is being conducted in accordance with
the provisions of the Federal Advisory Committee Act. Dr.
John D. Larkins in the Designated Federal Official for the
initial portion of the meeting.
We have received no written statements or requests
for time to make oral statements from members of the public
regarding today's session.
The transcript of portions of the meeting is being
kept, and it is requested that the speakers use one of the
microphones, identify themselves, and speak with sufficient
clarity and volume so they can be readily heard.
I want to begin with a few items of current
interest. First, members will notice in their package that
Mr. McGaffigan has been reappointed to the Commission.
And members will also notice that Oconee got its
license extension. However, there is one item of interest
that is of special importance to this Committee, and that is
the NRC has seen fit to award Mr. Theron Brown a Meritorious
Service Award, so I think he deserves a round of applause
for this.
[Applause.]
CHAIRMAN POWERS: This is a well-deserved award,
and we have benefitted greatly from this.
Are there any items that members want to raise
before we begin the scheduled proceedings?
[No response.]
CHAIRMAN POWERS: If not, we'll turn to the first
item of business, which is the spent fuel storage pool for
operating facilities, and Dr. Kress, I think you are taking
the lead on this?
DR. KRESS: Yes, this is Generic Safety Issue 173-
A. It's been with us awhile.
The issues are the potential to either lose
inventory out of the spent fuel pool due to some sort of
leak, may be driven by seismic or otherwise, or to lose
cooling to the pool so that eventually it heats up, and
boils off.
This is for operating reactors, and recently we
had a similar condition we were looking at for
decommissioning plants. Basically, the issues are the same,
only with operating plants, the pools is there all the time
and the heat load may even be greater, because it's
decreasing with the decommissioning plan.
So the issues are virtually the same, in my mind.
There have been events, and in response to some of
these events, this GSI was initiated, and in 1996, we were
briefed, the Staff was briefed, briefed us on an action that
was developed, and the action plan, I think, basically had
three components:
One was a plant-specific evaluation or regulatory
analysis for safety enhancement backfits. They determined
that this was not a compliance issue. It was a safety
enhancement issue.
Then they were going to implement rulemaking as
part of the shutdown rule to deal with this particular
issue. And then they were going to revise the Staff
Guidance. That's the Standard Review Plan, Regulatory Guide
1.13.
What we have is SRP and the Reg Guide to look at.
When we reviewed this in 1996, I went back and dug out our
letters and looked at them. Apparently there were no
identified outstanding technical issues at that time.
And the staff did not ask for us a letter at that
time, so we didn't produce one.
We did produce a letter in 1998, which endorsed a
high priority ranking for this particular GSI. And
currently, the Staff is in the process of revising its
guidance, and I think the intention is to work with the
industry to revise an ANSI ANS standard and use it.
I think the objective of the presentations today
is to see if they could get our endorsement to close the
issue. And so with that as a really abbreviated
introduction, I'll turn it over to George Hubbard, I guess.
MR. HUBBARD: I'm George Hubbard, the Acting
Branch Chief for Plant Systems Branch. I think Dr. Kress
has pretty much summarized the history of where we've been
on this issue, and what we're looking for today is to get
your agreement with regard to our proposed action on the
GSI-173.
With that, I'll introduce Chris Gratton, who is in
the Plant Systems Branch, and he will walk us through some
of the history and bring us to where we are today and deal
with that.
Chris?
MR. GRATTON: Thank you. Also here today is Tim
Collins from the Plant Systems Branch. That was a very good
lead-in to the situation that we're in right now.
We initiated this issue back in 1992, and haven't
really addressed, other than trying to revise those -- the
guidance documents that Dr. Kress was talking about, in the
last two or three years.
So as we go along through this, the background and
the history, if there are any questions on where we are or
what it was that we were attempting to do, please stop, and
I'll try and clarify what where we're going.
The purpose of the presentation is to update the
ACRS on the status of the GSI-173, and to inform you that
based on the results of the reviews and the evaluations,
that we intend to close the GSI.
We do not have any additional actions or
recommendations, and we intend to seek your concurrence that
the GSI can be closed.
As Dr. Kress said, the last time the Staff
presented to the ACRS on this topic was in August of 1996,
after completing our review of the technical issues.
Since that time, the Staff evaluated certain
design features to determine whether plant-specific safety
enhancement backfits could be justified at those plants.
The Staff completed that review and published our findings
in a report dated September 30, 1997.
At the time, the ACRS did not request a
presentation on the followup activities for GSI-173-A.
Slide 4. The Staff developed and implemented a
generic action plan for ensuring the safety of spent fuel
storage pools in response to two postulated event sequences
at two separate plants.
The principal safety concern addressed by the
action plan involved the potential for sustained loss of
cooling and the potential for a substantial loss of coolant
inventory that could expose irradiated fuel.
The latter concern, Part B to the action plan, was
previously resolved and will not be discussed here today.
The first postulated event sequence was reported to the NRC
in November of 1992.
In the report, it was contended that the design of
the Susquehanna Station failed to meet regulatory
requirements with respect to sustained loss of cooling
function to the spent fuel pool that could result from the
loss of offsite power or a LOCA.
The heat and water vapor added to the reactor
building atmosphere by the subsequent spent fuel pool
boiling could cause the failure of accident mitigation or
other safety equipment, and the associated increase in
consequence -- and have an associated increase in
consequences from the initiating event.
Using probabilistic and deterministic methods, the
Staff evaluated these issues and determined that public
health and safety were adequately protected for Susquehanna.
However, the Staff also concluded that a broader
evaluation of the potential for this type of event to occur
at other facilities was justified.
In addition to reviewing the Susquehanna safety
issue, the action plan also called for a review of the
events related to wet storage of spent fuel. From these
reviews, the Staff identified areas to evaluate for further
regulatory actions.
Design information in support of the evaluation
was developed through four site visits, reviews of plant
SERs, and a Staff survey of plant operating procedures and
designs that was completed in May of 1996.
The Staff published its findings in a report dated
July 26th, 1996. During the development of the action plan,
the Staff met with the ACRS several times.
The Staff received and implemented recommendations
from the ACRS that aided in the Staff's coming to a
resolution on these issues.
Over the next few slides, I'll discuss the focus
of the action plan, and the findings and the followup
actions.
Concurrent with the action plan, the Executive
Director for Operations directed the Office for Analysis and
Evaluation of Operational Data to perform an independent
study of the likelihood and consequences of an extended loss
of spent fuel pool cooling.
NRR reviewed the AEOD report upon its completion,
found the results were consistent with the findings from the
action plan, and modified our action plan to include certain
insights from the AEOD study.
DR. KRESS: That AEOD study, my impression is that
they really didn't quantify the consequences; they just said
they were severe or something like that?
MR. GRATTON: I don't know the exact answer to
that question. Jose Ibara from AEOD is here and may be able
to shed more insight on that, on the exact --
MR. IBARA: Yes, Jose Ibara from Research. We did
not quantify it, but we did have data as to how many degrees
events occurred, and so forth. It was not quantified.
DR. KRESS: Thank you.
DR. WALLIS: I'm not sure now. You're talking
about the consequences here?
MR. GRATTON: For the AEOD study?
DR. WALLIS: The consequences are released to the
environment of radioactive material. That was not
investigated?
MR. GRATTON: Not in the AEOD study.
DR. WALLIS: But it was part of your thought
process? Somebody must have thought about it.
MR. HUBBARD: This is George Hubbard with the
Plant Systems Branch. We did not carry it to the
consequences. What we looked at in the Susquehanna study in
looking at this action plan, we were looking at the low
probability of these events or the sequences that we looked
at, and we didn't carry it forward to -- I believe we felt
the possibilities of these events was low enough that we
didn't carry it through to the consequences.
DR. KRESS: My impression, Graham, is that this is
like looking at a LERF, where you don't really do the
consequences, but you have something that's relating to the
consequences.
DR. WALLIS: I'm just asking because your
transparency says they evaluated the consequences, and it
appears that they didn't. So, that's the only reason I'm
following up on this.
MR. GRATTON: The request from the EDO said the
likelihood of consequences, but what they actually looked
into, I don't believe, went into the consequence stage.
CHAIRMAN POWERS: I guess there is the question of
suppose that I come along and say, gee, the probability of
an accident is very, very low? But I'm supposed to be risk-
informed. Don't I have to multiply that by the consequences
before I start taking it off my list?
DR. KRESS: The risk ought to have consequences
built into it. The risk is the product of the probability
of the consequences.
So you can't say the risk is very low until you do
the consequence part, in my opinion.
DR. WALLIS: So does that mean that we will not
know what the risk is today?
DR. KRESS: Well, once again, if you have -- if
you calculate a LERF for an operating reactor, say, and then
you basically have a measure of consequences built into your
LERF.
I think the appropriate question is what value of
LERF is an acceptable value? I think you have a -- you
don't have a measure of the consequences if you're using a
LERF that is for one situation, and you're trying to apply
it to another, I don't think.
DR. WALLIS: At least you have an order of
magnitude.
DR. KRESS: Yes, it may be conservative and ought
to be.
MR. GRATTON: To identify the spent fuel storage
issues and evaluate concerns for identified spent fuel
storage issues, we focused the evaluation on design features
and safety functions of the spent fuel storage system.
Coolant inventory, coolant temperature and fuel
reactivity were areas evaluated for identifying and
evaluating spent fuel storage issues.
Coolant inventory affects the capability to cool
the stored fuel. It provides radiation shielding from the
stored fuel, and mitigates the effects of fuel handing
accidents.
We found common design features that reduce the
potential for loss of coolant inventory.
A reinforced, seismically designed structured
capable of retaining its function following a design basis
event was found at each facility, welded, leak-tight liners
with leak detection piping, anti-siphon measures on piping
entering the pool, and alarms and indications relevant to
coolant inventory that alert operators to level decreases.
DR. KRESS: I'm sorry. My impression is that you
looked at every plant?
MR. GRATTON: Yes.
DR. KRESS: The design of every plant?
MR. GRATTON: Right, exactly.
CHAIRMAN POWERS: That includes Sharon Harris?
MR. GRATTON: Yes, it did. If you remember in the
beginning of the slides there were four site visits. They
were all not visited, but information about the design of
the spent fuel cooling system was collected on each site.
CHAIRMAN POWERS: I have a letter from Mr.
Thompson concerning the spent fuel storage facilities at
Sharon Harris, and it calls attention to some peculiarities
of that design. Have you looked at that?
MR. GRATTON: What peculiarities are you referring
to, in particular?
CHAIRMAN POWERS: Availability of makeup water
supply, the ability to drain the pool down into lower
regions of the plant and things like that. He has an
extensive list of things.
DR. BONACA: Yes, also the use of cooling water
from the cooling system from one power plant to multiple
pools, and the original design was intended to have power
and cooling coming from different units to the different
pools. Therefore, you have certainly a higher potential for
common cause consequences from failures of the cooling
system or electrical system.
I mean, I believe that there was a USQ, and in the
report we received, there was no explanation of how the USQ
was resolved.
MR. HUBBARD: This is George Hubbard again. On
the Sharon Harris, the -- you know, we are aware of it, but
due to the hearings that are going on, we don't feel that
it's appropriate to get into a deep discussion, you know, of
the Sharon Harris situation.
DR. KRESS: In addition, that's just one plant.
The other plants don't share these peculiarities, I presume?
MR. GRATTON: That's correct that Sharon Harris is
a unique site.
DR. WALLIS: You mentioned seismic and design
basis. Now, there's always a probability of seismic events
with exceed the design basis and do actually rupture a
liner. Does that figure into this?
MR. GRATTON: The liner is not the leak-tight or
the structure that provides the assurance that the inventory
will remain there. It protects the concrete behind there.
The liner is leak-tight, but it is not a design
feature that maintains the coolant.
DR. WALLIS: But to get back to seismic, there is
a seismic event with some probability which will rupture the
reinforced concrete. Does that figure in these
calculations, or is it just the design basis?
MR. GRATTON: It's just the design basis.
DR. WALLIS: So this is not really risk-informed
then?
DR. KRESS: If you do a risk analysis, you have to
include the seismic.
DR. WALLIS: Yes, you would have to. But that's
not included your assessments?
MR. GRATTON: No, we did not consider beyond
design basis seismic events. From coolant inventory, we
identified five categories of plants that had design
features contrary to the design guidance that the Staff
uses.
Four plants lacked passive safety, anti-siphon
devices on piping that extended below the top of the stored
fuel. Five plants had spent fuel pool transfer tubes
entering the spent fuel pool below the level of the fuel,
and were not separated from the fuel by a Wier or other
passive device.
Three plants in the category above with the
transfer tubes have interfacing systems connected to those
transfer tubes. These are all at Oconee.
Six plants have indirect spent fuel pool level
indication. Four plants do not have isolation capabilities
for liner leak-off systems.
These were the design features that were
identified relative to inventory, while we were going
through and picking up and performing our review of the
individual plants.
DR. KRESS: But these don't constitute compliance
issues?
MR. GRATTON: Exactly. These were not compliance
issues, but they were different from the other plants. The
majority of the plants had certain features. These had
these unique features.
The Staff also concluded that temperature had a
less direct effort on safe storage of fuel compared with
inventory. Coolant temperature was limited by evaporative
cooling and the rack design ensures a subcooled environment
surrounding the fuel. As a result, forced cooling of the
pool is not required to protect cladding integrity when
adequate level is maintained in the spent fuel pool.
However, temperature does have an effect on
structural loads, the purification system operation,
operator performance and the environment surrounding the
pool.
Normal operation of the spent fuel pool cooling
system keeps the pool temperature low enough to prevent
exceeding the acceptance standards. Short term exposure
under abnormal conditions to temperatures above 150 degrees,
which could be experienced during a temporary power outage,
should not affect the pool structure, the large thermal
capacity --
DR. KRESS: I take it that's Fahrenheit?
MR. GRATTON: Yes -- the large thermal capacity.
While the purification system performs no safety
related function it does keep the pool's activity low,
reduces corrosion and keeps the water clear to aid the
operators during refueling operation. Coolant temperature
also affects fuel handling operations. High temperatures
result in operator heat stress and can hamper operations by
fogging.
DR. WALLIS: You talked about the temperature of
the pool. You talked about subcooling in the racks.
How good is the analysis for temperature
distributions throughout all this pool? It is not a uniform
temperature.
MR. GRATTON: It is not a uniform temperature but
they have natural circulation through the pool.
In the previous analysis we did not actually do a
thermal hydraulic analysis of the pool as part of this
study, but what we had found was that the licensees'
calculations indicated that the region and the rack stayed
subcooled even under boiling conditions so the racks, which
are 40 feet below the surface, stayed subcooled.
DR. WALLIS: So someone reviewed the licensees'
analyses and said they were okay?
MR. GRATTON: As part of licensing action when you
do the original design work that is looked at.
Latent heat and vapor are added to the surrounded
buildings at very high spent fuel pool temperatures.
Subject to the ventilation system, this could affect the
operability of equipment sharing spaces through condensation
and operating temperature.
This issue was extensively evaluated for
Susquehanna. Three features defined this issue --
multiunits, open paths from the spent fuel pool to the
safety-related equipment, and a short heatup time.
The Staff identified seven sites that have this
configuration.
Because of a wide variety of cooling system
designs at operating plants the Staff reviewed the
capability and reliability of each. The Staff noticed that
some reactors lacked the design capability to supply onsite
power to a system capable of cooling the spent fuel pool.
That was identified at seven sites.
Some spent fuel storage systems have low primary
cooling capability relative to the potential decay heat load
in the spent fuel pool. That was at four reactors.
Some reactors rely on infrequently used backup
systems to address Loss of Offsite Power Events and
mechanical failures. That was at 10 reactor sites.
DR. KRESS: Were these judgments made before or
after the NRC started allowing higher density storage in the
pool and more fuel in the pool?
MR. GRATTON: What decisions are you -- are you
referring to the allowing of this configuration in the pool?
DR. KRESS: Things like the subcooled boiling.
Did that include the higher density fuel?
MR. GRATTON: That was an original, the way I
understand it, that was an original concept but every time a
plant goes through a relicensing --
DR. KRESS: -- they have to relook at those
things.
MR. GRATTON: That is a reconsideration.
What I want to do right now is to let you
understand that we are looking at the -- this is a portion
of the presentation that has already been given in 1996. I
am just sort of recapping what the Staff presented
previously, so on the previous slide where the coolant
inventory issues that the Staff identified that looked into
it and the results of it, these were the coolant temperature
issues or the areas that the Staff identified when reviewing
features about coolant temperature.
On the next slide --
DR. WALLIS: Can I ask you about the temperature
distribution? Is it all theory or have there been events in
pools where there have been temperature transients which in
some way confirm that the analysis was okay or is it all a
matter of someone looks at theory and says I don't see any
mistakes, looks reasonable, must be okay?
MR. GRATTON: Are you referring to heatup rates or
the distribution of the --
DR. WALLIS: All these questions about subcooled
boiling and temperature distribution and what happens in a
transient and how hot does it get.
MR. GRATTON: I am not familiar with any
studies --
DR. WALLIS: Was it all theoretical studies or are
there some evidence from real pools?
MR. HUBBARD: This is George Hubbard again.
In some cases in particular we had one plant that
came in for a rerack within the last three or four years.
We went to Research and had them run some CFD codes to
determine the heat distribution throughout the pool because
there were some concerns with the circulation because they
were putting racks or asking for permission to put racks in
the cask pit for a period of time and so with NRR and our
people there and Research folks we ran some CFD calculations
to determine what is the good distribution.
From that effort I believe we found not a big
variation in the temperature from the bulk temperature that
we consider when we normally do the rerack calculations. I
would say it was maybe on the order of 10 or 15 degrees
difference with that particular analysis that we did there.
MR. GRATTON: What George is referring to is
another analytical type thing.
Plants have instrumented -- I don't want to
speculate, but I know plants have instrumented their pools
and have calculated things like decay heatup rates. For a
particular decay heat rate, the pool heatup rates, but off
the top of my head I do not remember if they have
instrumented the pool such that they have gotten a
temperature gradient from the top to the bottom to verify --
DR. WALLIS: It just seems to me that when so much
depends on codes and analysis in all these nuclear plants,
have we any evidence at all, like some event, that you could
use to check those? It would be very useful.
MR. GRATTON: I mean other than a loss of cooling
event where the pool heats up slowly over a couple of hours,
that would be the only thing that I could think of, off the
top of my head, where you could go back and use the analysis
to go back and verify that the system, the natural
circulation system, is performing properly.
DR. BONACA: One interesting point is that
typically these pools run well below the 150 degrees. I mean
they run around at least 50 degrees below that, something on
that order, so it gives some confidence that if you have
variations there will be --
DR. KRESS: That is if your cooling system works.
DR. BONACA: Absolutely. I agree with that.
I am saying that the design limit of the pool
under normal conditions is a way from --
DR. KRESS: I don't think from a risk perspective
you worry very much about these temperature distributions
and subcooled boiling, because all you are doing is possibly
damaging individual fuel -- what you really worry about is
if you lose all the inventory and lead this thing into a
meltdown type situation.
DR. BONACA: That's true.
DR. KRESS: And that is a completely different
question.
MR. HUBBARD: George Hubbard again. I believe
relative to AEOD when they took a look at it, and I don't
remember the numbers right offhand, but they looked at
events that led to heatup and how frequent that was and if
you want a little more detail on it, Jose maybe can give us
the actual numbers, but I think they divided it up on how
often you had a heatup of 10 degrees or 20 degrees and it
was very infrequent.
Do you have the numbers there, Jose?
MR. IBARRA: Yes. We did look at actual events
and there were a few but we did look at events in which the
temperature rose like 20 degrees, and if I remember, that
occurs only in 3 in 1000 reactor years.
DR. KRESS: There is certainly not enough
information to validate the code.
MR. GRATTON: Right. On Slide 7 it summarizes the
Staff's review of the fuel reactivity portion of the spent
fuel pool action plan.
The Staff reviewed the design of the spent fuel
pool storage, structures and components which control the
stored fuel including the use of solid insoluble boron and
did not identify any issues relative to the spent fuel pool
reactivity control and therefore did not include any
followup actions for fuel reactivity.
DR. KRESS: But what happens to soluble boron as
you boil away water? Does it concentrate or does it go off
with the steam?
MR. GRATTON: No, it concentrates.
MR. BARTON: Is there a saturation limit on it
where it would precipitate out at some --
CHAIRMAN POWERS: Isn't there vapor pressure?
MR. GRATTON: I don't know.
CHAIRMAN POWERS: I mean I think there is a vapor
pressure for boric acid.
DR. KRESS: So it would leave with the vapor?
CHAIRMAN POWERS: I don't know. It may depend on
the partition.
DR. KRESS: Yes, it depends on the partition
coefficient.
CHAIRMAN POWERS: Do you know what that is?
DR. KRESS: Well, I think at low pressure it is
such that it would concentrate. At high pressure it is such
that it would dilute. But these are low pressure, so I
suspect he is right. It concentrates.
MR. GRATTON: At the completion of this portion of
the review the Staff concluded that existing systems,
structures and components related to the storage of
irradiated fuel met the regulations.
Protection was provided by several layers of
defense. The Staff also concluded that because of the
design and operational factors associated with spent fuel
pools they constituted only a small fraction of the overall
risk of operating a nuclear power plant.
DR. KRESS: Most of these pools are located
outside of containment?
MR. GRATTON: That's correct.
DR. KRESS: Was there any defense-in-depth
thinking going into this?
MR. GRATTON: Let me correct something. The BWRs
are in the secondary containment. Is that what you are
referring to? Are you talking about the primary
containment?
DR. KRESS: I am talking about primary
containment. BWRs are in secondary.
MR. GRATTON: There was only one that was in
primary containment, I believe.
DR. KRESS: Yes, one of the MARK IIIs, I think,
would be --
MR. GRATTON: Well, I think it was Big Rock Point.
Was that the only one that was in containment, George?
I don't remember. Go ahead.
DR. KRESS: But anyway, I was thinking here's fuel
that we are talking about some sort of an accident
condition. It is basically outside containment.
MR. GRATTON: Right. It is either in the fuel
handling building or in the secondary containment.
MR. HUBBARD: George Hubbard again. I believe you
were mentioning did we look at defense-in-depth and that was
considered with regard to multiple systems for providing
makeup and that was looked at as part of this, in coming up
with these conclusions. We did look at that.
DR. BONACA: I have a question on that, because in
the original report -- he talks about several layers of
defense and refers specifically to prevention, mitigation
and radiation protection, and when it talks about prevention
it specifically states quality control and design,
construction and operation.
Now several of the older plants do not have in
fact a quality requirement imposed on the cooling systems
and we have reviewed one just recently and it didn't have
it, so is it that old plants have these quality requirements
or controls or is it just that you have a few that do not
have them?
I mean this is a very generic claim made in the
front of the report to say it is not a generic concern and
prevention is identified as purely quality controls.
MR. GRATTON: I think the intention of the
statement was that consistently across all the plants that
there was a defense-in-depth applied.
If it was not a quality control like let's say a
seismically qualified spent fuel pool cooling system that
they had a seismically qualified pool and seismically
qualified makeup systems, redundant makeup systems, that
could provide water to the pool such that evaporative
cooling would be available to cool the pool in the event
that a seismic event took the cooling system out, so it is
very hard to make a general statement about all the pools
since they are all so different but since we looked at them
all, I think that this general statement was made in the
effect that at each -- if you took an individual slide of
plant there was a defense-in-depth at that plant that was
noted.
DR. BONACA: I mean one, they were reviewed,
claimed, and it was accepted that they do not have to
monitor aging of the system because it's not part of the
design basis and is not part of quality commitments, so I
know for one that it is not the only pool out there, so I am
saying that the statement is general to me and I immediately
had some examples that don't meet this.
DR. KRESS: If you entered into one of these loss
of cooling or loss of inventory sequences, the response is
all operator action, I presume?
MR. GRATTON: There is no automatic response,
that's correct.
DR. KRESS: And he is told what to do in some sort
of procedures, operating procedures?
MR. GRATTON: Yes. Notwithstanding the Staff
reviewed each plant against the criteria in identified areas
where potential safety enhancements could be investigated.
Ten design features, five inventory related and
five related to spent fuel pool decay heat reliability, were
identified for further evaluation by the Staff to see
whether safety enhancement backfits could be justified.
We added an additional design feature to the
review based on the results of the AEOD study; 48 plants
had one or more of the design features of concern.
The Staff also planned other actions as a result
of the study. Rulemaking, which was previously mentioned,
was in progress for the shutdown rule at the time and the
Staff plans to incorporate lessons learned into the review
guidance documents that the Staff uses to review changes to
spent fuel storage designs.
DR. KRESS: Is that still the plan, to have the
shutdown include --
MR. GRATTON: No.
DR. KRESS: No?
MR. GRATTON: No, that has been tabled from the
time that this was originally issued.
The Staff presented the results of the spent fuel
pool action plan, which is the previous four, five slides,
to the Commission and to the ACRS in July and August of
1996, respectively. At the time the committee was satisfied
with the performance of the review and the Staff did not
feel it was necessary to obtain a closure letter on this
issue from the ACRS at the time.
Where we are is that was the spent fuel pool
action plan -- about 1996 in August. We have just finished
the presentation on the Staff's review, identification and
review of the spent fuel storage issues.
From that we identified 11 followup activities
that in 1997 we went out and evaluated.
The Staff planned to address the followup issues
from the spent fuel pool task action plan either by
performing a regulatory analysis to determine whether a
safety enhancement could be justified or by gathering
additional information to augment the information during the
spent fuel pool task action plan.
The Staff determined that seven design criteria
warranted regulatory analysis. For the other four issues
some evaluation was warranted. This could be an evaluation
of the administrative controls, the capability to align and
operate a backup cooling system, or a review of the actual
design of the component of concern.
For each of the issues requiring regulatory
analysis a probabilistic analysis was first performed as a
screening criteria to determine the likelihood of obtaining
a given endstate. The Staff visited seven plants to gather
information about five of the seven issues requiring a
regulatory analysis. With the other two issues the Staff
addressed the issues either through a voluntary action by
all of the licensees in that group or by using information
already available to the Staff here at headquarters.
Conservative endstates were chosen for these
evaluations. One was for the inventory issues an endstate
of one foot above the top of the fuel was chosen and for the
loss of decay heat removal eight hours of sustained boiling
in the spent fuel pool was chosen.
DR. KRESS: Why was eight hours decided? I figure
that's the amount of time that anybody could recognize what
is going on and make corrective actions, or something?
MR. GRATTON: No. It really had to do with the
capability of equipment, the vapor to transport to areas
where equipment, safety equipment, was located.
DR. KRESS: Oh, you were worried about the effect
of the steam on other equipment?
MR. GRATTON: Yes. It is not the boiling itself,
it's the shared systems and structures, components of the
design features, and I believe the eight hours was a
conclusion in the Susquehanna safety evaluation report.
Is that correct, George, or Sam, do you remember
that -- where the eight hours came from?
MR. LEE: The eight hours is in addition to about
14 hours of time that is heatup time already, so we are
looking at about 22 hours of time starting from the loss.
MR. GRATTON: The Staff used the following
screening criteria with these endstates, which parallels the
Staff's guidance on regulatory analysis.
If the frequency that was calculated was less than
10 to the minus 6 per reactor year, the probability was low
enough so that a safety enhancement backfit would not be
justified.
If the frequency was greater than one times 10 to
the minus 5th the Staff would do further evaluation on the
design feature to determine whether or not a safety
enhancement backfit was justified.
Between those two limits, 10 to the minus 6th to
10 to the minus 5th engineering judgment was used
considering the available margin to determine whether
further analysis was justified.
DR. KRESS: Now these criteria are basically the
same as the safety goal screening criteria of regulatory
analysis? They are based on it.
MR. GRATTON: Right. We try and base it on it.
DR. KRESS: You have to dig it out of that matrix
but they are basically the same, which brings to mind a
number of questions --
MR. GRATTON: -- that I hope I can answer.
DR. KRESS: Okay. Number one, I presume none of
these -- the reason you didn't proceed with any of these
backfits is that none of them passed the screening criteria?
MR. GRATTON: Well, as we go through, one group
did meet this -- I will call it meeting the screening
criteria if it exceeded 10 to the minus 6th --
DR. KRESS: Okay.
MR. GRATTON: -- and I will tell you how we
addressed it on those. The rest of them, as you will see,
did not contain a frequency of greater than 10 to the minus
6th.
DR. KRESS: Now let me see if I can express my
question in a way that it is understandable.
The safety goal screening criteria is basically a
LERF. It is a CDF and a conditional containment failure
probability, but if one looks at it properly it is basically
like a LERF, and the numbers, the values, the limiting
values or acceptance values or the values that pass the
screen come out of the prompt fatality safety goal. They are
derived from it.
It is like the LERF in 1.174 is derived from the
prompt fatality safety goal. That is why it is called a
safety goal screen.
Now for a given LERF value, or a given combination
of CDF and containment failure probability, to meet the
prompt fatality safety goal it relies on information or
knowledge about what fission products are released, how many
and what mix of isotopes there are because those are what
cause these prompt fatalities, plus it also depends on the
atmosphere transport things but those can be dealt with.
But the problem, the question I have, is these
particular safety goal screening criteria that are part of
the regulatory analysis are based on the standard source
term for an operating reactor if an operating reactor
undergoes a core melt accident.
Now there is some question as to whether a spent
fuel pool accident produces the same mix of fission
products. The quantities may be more or may be less,
depending on when and how much fuel is in there, but it is
the mix of fission products that is of concern to me,
particularly if the accident leads to the final conclusion,
which is a higher oxidation of zirconium driven accident.
The question that this brings to mind is are the
safety goal screening criteria in the regulatory analysis
appropriate to use for a spent fuel pool accident, and that
is the question, and if it is not appropriate, what
acceptance criteria or what screening criteria should be
used?
Do you understand my question?
MR. GRATTON: I understand it but I am not -- my
legion of experts over there can also assist me in this
response, but I am not sure there is a one for one
correlation.
The endstates that we chose are not accident
conditions per se.
DR. KRESS: Of course you have got some
conservatism there in the endstates.
MR. GRATTON: We had a large conservatism in both
of these things, plus the progress of the accident beyond
that point is really not very well understood or researched.
DR. KRESS: But to have an acceptance criteria you
need to know something about that additional progress and
you need to know how conservative your endpoints are.
MR. GRATTON: What we tried to select were
endpoints that were conservative enough that we felt if we
fell in the 10 to the minus 6th range that it was an
extremely conservative --
DR. KRESS: Well, that is the part that bothers me
because those are intuitive judgments.
MR. GRATTON: Right.
DR. KRESS: And sometimes we found out in the
severe accident business that our intuition tells us the
wrong things.
I worry about using intuition to determine precise
acceptance criteria for something like this.
MR. GRATTON: Sure.
DR. KRESS: And that is the part that bothers me
about the whole study, I think.
DR. BONACA: I had one other questions, by the
way. You referred to these probabilistic evaluation or
analysis but to what extent has there been an evaluation of
both potential initiators, because we talk about not looking
at source term because of very unlikely events but I am not
aware of any thorough, systematic assessment of the
potential initiators to be considered.
For example, you are more focusing on certain
criteria that you have to meet and how unlikely it is, but
it is also, it seems to me, surmised that it is unlikely to
get there.
For example, I would suspect that sabotage wasn't
considered as a possible initiator.
MR. GRATTON: No, it wasn't.
DR. BONACA: And just making an example here, and
typically if you want to have some credible probabilistic
assessment you would do some systematic assessment of the
potential initiators and you would find it varies
significantly from plant to plant, and I don't think that
was done, was it?
MR. GRATTON: Yes, we did.
DR. BONACA: You did?
MR. GRATTON: For each plant, and Sam can give a
more detailed answer to this, but for each plant we selected
lead plants which were representative of the issue for, like
say for the shared systems and structures issue.
There were 13 plants in it and I believe they were
at four or five different sites so obviously they were all
multiunit sites, but we went in and -- Sam, I will let you
describe how we described the initiating events and
evaluated the probability or the frequencies.
MR. LEE: This is Sam Lee. First of all, I would
just like to make a minor correction for the record -- the
criteria that we had used was 10 to the minus 5. I think
there was a mixup. It was 10 to the minus 6.
As far as the initiating events go, there were
about five initiating events that we looked at, which
include loss of offsite power, loss of spent fuel pool
cooling system, and loss of spent fuel pool inventory, and
we even considered earthquake as well in the analysis, so
those all factored into the analysis.
When we looked at the results, depending on the
plant, for one specific plant like Hatch the dominant
initiating event frequency that contributed most to the
total estimate was from the loss of offsite power sequence
and second to that was the earthquake initiating event, so
we did look at it pretty comprehensively.
DR. WALLIS: May I ask a much simpler question?
MR. GRATTON: Sure.
DR. WALLIS: You have all this instrumentation
here. Are other people around in this building? If the
pool is boiling, is there someone in there to see it is
boiling?
MR. GRATTON: Yes.
DR. WALLIS: All the time?
MR. GRATTON: Not all the time, but there are
operators --
DR. WALLIS: So every eight hours maybe someone is
around that's supposed to look? Maybe that is where the
eight hours comes from -- every eight hours someone is going
to be around the building, and if it is boiling it is going
to be obvious.
MR. HUBBARD: This is George Hubbard. I think
particularly for an operating plant someone is going to be
around, particularly in the high heat load situations during
refueling. Those situations are going to have people
definitely there and they are going to recognize it is
getting awful warm before it ever gets to the boiling state,
so for the operating plant I think there's going to be
people there to realize that something is different.
MR. GRATTON: I still believe though, and I can't
find it right now, it's in the Susquehanna SER, that there
was a concern about how long safety-related equipment could
last in an environment where the spent fuel pool was
boiling.
There was a conclusion made, based on expert
judgment I believe or whatever the conclusion was, was that
for boiling of eight hours the safety-related equipment was
robust enough that they felt it could continue to operate in
the environment in a pool and adjacent space boiling. I
think that is what we used for the basis of it but I can --
it is in the Susquehanna safety evaluation report.
So the first group of plants -- I'm sorry?
DR. KRESS: How long does it take for a typical
pool with its loading to uncover the fuel?
MR. GRATTON: To uncover the fuel?
DR. KRESS: If it went into boiling.
MR. GRATTON: A typical pool -- none of them are
typical but if you use a boiloff rate of 50 gallons per
minute it takes a significant amount of time for a 400,000
gallon pool or 350,000 gallon pool to boil off.
As a rule of thumb there is about 200,000 gallons
of water above the racks -- above the top of the racks -- on
a typical pool.
In the followup actions, in the first group, these
are the ones that we plan on doing the risk assessment for
the probabilistic analysis on. There was another group that
required additional evaluation when the Staff gathered
information via the site visits that we went on and we also
had teleconferences with licensees and reviewed material at
the NRC to determine the need for further regulatory action.
DR. KRESS: You actually didn't proceed to any
kind of cost benefit analysis?
MR. GRATTON: No, we did not.
DR. KRESS: Didn't get that far in the
regulatory --
MR. GRATTON: The seven design features -- we'll
go back to that, number 9. The seven design features was
the plant performed probabilistic analysis are the absence
of passive anti-siphoning devices, the transfer tubes that I
talked about with Oconee, piping entering the spent fuel
pool below the fuel -- and this is at the SSF at Oconee has
piping actually attached to the transfer tube in a
configuration where the transfer tube isolation valve is
operated with it open when the plant is operating, limited
instrumentation for loss of coolant events, the effects of
adverse environments on the multiunit plants with shared
systems and structures -- this was the issue that came out
of the Susquehanna review, the absence of onsite power for
spent fuel pool cooling systems and limited instrumentation
for a loss of cooling event.
These were the issues that we did regulatory
analysis on.
The four design features which the Staff gathered
additional information on are shown on Slide 10.
And that is the absence of a liner leak detection
or isolation system limited to K heat removal capability for
the systems that supply cooling to the spent fuel pool,
infrequently used backup systems for spent fuel pool cooling
and the issue that came out of the AEOD study, which was the
influence of reactor cavity seals on inventory losses in the
spent fuel pool. We focused that review on the seals with
pneumatic components.
Giving an overview of the results of the 11 design
features that were evaluated up in the follow-up actions,
five design features were reviewed using probabilistic
analysis and did not meet the screening criteria for further
evaluation, so they were screened out.
One design feature category was eliminated from
evaluation when all the licensees in the category took
voluntary actions to address the concern.
DR. KRESS: Which one was that?
MR. GRAFTON: I will go over that. But that was
-- the design feature was the anti-siphon devices. And I
believe there was four plants that had deep-running pipes in
their spent fuel pools, and they either committed to cut the
pipe or to provide certain administrative controls that the
staff reviewed and found acceptable to address the issue.
One design feature did meet the screening criteria
for this design feature, which was the shared systems and
structures at multi-unit sites. The staff performed
additional analysis to determine whether the safety
enhancement backfit was justified.
For the design features where the staff needed
additional information to determine whether to perform
further analysis, the staff gathered the necessary
information during site visits and by reviewing in-house
material. And after evaluating the information, determined
that none of the four requirement further review. And I
will go into a little bit more detail on each one of those,
just so you get an idea of which ones we looked at.
DR. KRESS: That is sort of like George's three
region approach.
MR. GRAFTON: For each of the five design features
listed on this slide, the staff collected plant-specific
information and performed probabilistic evaluations to
determine the frequency of each end state. In each case,
the frequency of the end state occurrence was low enough
that the staff concluded that no further evaluation was
warranted. The first issue was draining the spent fuel pool
through the fuel transfer system. That was evaluated at
Oconee.
Draining through an interfacing system, again, but
I repeated myself, but it is Oconee because of the SSF
system. An absence of a direct low level alarm and limited
instrumentation for loss of cooling, the last one, were both
evaluated at Hatch and Dresden. And the absence of on-site
power for spent fuel pool cooling system, in this issue
there were four sites that had this. Two plants took
voluntary actions to supply emergency power to their spent
fuel pool cooling system, and the other two were evaluated
and found to have a low frequency of the end state.
One of those plants, which was ANO-2, their backup
cooling method is actually evaporative cooling, so, they
don't rely on a backup system to cool their spent fuel pool.
DR. KRESS: But do they have to make-up water
then?
MR. GRAFTON: Yes. There is a seismic
qualified --
DR. KRESS: They have a good make-up, seismic
qualified make-up system.
MR. GRAFTON: Event frequency of one design
feature exceeded the screening criteria. This design
feature was the shared systems and structures. Thirteen
plants share this design feature, they are Calvert Cliffs,
D.C. Cook, Dresden, Hatch, LaSalle, Quad Cities and Point
Beach. You notice there is 13 of them, which doesn't make
much sense, but the design of the Hatch plant isolates the
Unit 2 safety-related equipment from the spent fuel pool
area, so only Unit 1 is susceptible to this condition.
The plants that we used to evaluate this condition
were Dresden and Hatch. The staff estimated the frequency
of sustained boiling events at both of these plants using
plant-specific information obtained from the site. The
results indicated a low likelihood of the events resulting
in sustained boiling, but one that exceeded the screening
criteria. As a result, the staff performed the following
additional evaluations.
The staff reviewed an evaluation by the Dresden
licensee on the effects of sustained boiling event on the
safety-related equipment in the adjacent spaces. The staff
agreed that, considering the low event frequency, the
multiple reliable cooling systems, plus the low LOCA event
frequency, makes the likelihood of a sustained boiling event
that affects safety-related equipment in the reactor
building a low frequency event. No backfits were justified.
At this plant they have two off-site power supplies and five
on-site power supplies.
DR. WALLIS: What do you mean by low frequency?
What is the number?
MR. GRAFTON: The number for Dresden, I believe
was 4.3 times 10 to the minus 6th. Is that correct, Sam?
Do you have that?
MR. LEE: You are looking at the total number,
yes, that is correct.
MR. GRAFTON: That was the total number for all
event sequences.
MR. LEE: That's right.
MR. GRAFTON: Hatch had a higher event frequency
for sustained boiling, I believe theirs was, in an operating
condition, theirs was 4.4 times 10 to the minus 4, okay, for
operating. All right. However, no credit was given for
contingent actions to restore cooling by operator staff,
either by supplying temporary power to the spent fuel pool
cooling system or obtaining a diesel for their auxiliary
decay heat removal system. During refueling, the frequency
for Hatch was calculated -- was actually recalculated for 9
times 10 to the minus 6th, which was more in line with other
plants that we looked at.
Hatch has a unique design. They have a normal
spent fuel pool cooling system which is not safety-related
and does not have emergency power to it, but they have an
alternate decay heat removal system that, when they do
refueling, they bring in an auxiliary diesel generator to
provide emergency power to the system, so it is a very
reliable system, plus they use RHR in the spent fuel pool
cooling assist mode, which is also a safety-related system
that is supplied from an on-site power supply, as another
method for backup cooling.
So, if you were to look at Hatch when it is
operating, the diesel for the ADHR system is not required to
be on-site because you do not have the high heat load in the
pool. Well, without a high heat load in the pool, you don't
have a rapid heat-up rate and boiling rate, you know, but,
you know, you do have both plants operating. So, even
though they have a high event frequency when the plant was
operating, there is a low decay heat load in the spent fuel
pool at that time. And, conversely, when you off-loaded the
pool, that is when the frequency went way down because you
have a much more reliable system.
One other thing to note about the Hatch system was
because of the size of the piping, they are not able to use
the DHR system -- excuse me, the RHR system in spent fuel
pool cooling mode when the plant is operating, because they
have to have the top of the plant off and use a circulation
path that goes from the reactor vessel back to spent fuel
pool, otherwise, it will cavitate the large RHR pump if they
just did it recirculating the spent fuel pool.
So, those individual plant-specific factors are
what dropped the frequency from the higher rate down to the
10 to the minus 6 range.
DR. POWERS: The auxiliary diesel at Hatch is
located on-site someplace?
MR. GRAFTON: Yeah. They truck it in. It
actually sits on a truck and it is pigtailed out, I believe
it was outside of the building. Sam, do you remember that?
MR. LEE: That's correct.
MR. GRAFTON: Yeah, they just, they back it up to
the outside of the spent fuel pool cooling building and the
rigs sits outside the spent fuel pool cooling building and
they just pigtail it to the cooling system.
Regardless, the staff reviewed the licensee's
evaluation of a sustained boiling event at Hatch and agreed
that no safety-related equipment would be affected by the
relatively mild environment created by the sustained boiling
event. The staff also concluded that because of the
differences in plant design, all plants in this group should
be reviewed for sustained boiling.
So, we looked at the frequency of it and the
frequency said that it was a low probability event, but we
also went in and said, okay, what happens if it does boil?
Let's look at where the safety-related equipment is located.
Let's look at happens if it were to flood and make an
evaluation on that. And we found that the safety-related
equipment was located far enough away and in an area where
it would not flood, that we did not feel that there was a
problem at that plant.
But because the sites were so unique, when looking
at Dresden and at Hatch, we said we need to go back and look
at the other plants that were in this group of 13. So, the
staff conducted reviews on Hatch and Dresden. One of the
conclusions from the reviews was that on-site power to the
spent fuel pool cooling system resulted in low frequencies
of sustained boiling. So we focused in on the factor that
brought the frequency down, and that was the factor, that
they had emergency power to the spent fuel pool cooling
pumps, it brought the frequency of boiling way down, because
the systems were very reliable.
So, when we looked at the other 13 in that group,
we eliminated or we screened out those that had emergency
power to the spent fuel pool cooling pumps, and the only
plant that remained was LaSalle. And LaSalle had a unique
configuration that left it vulnerable to a grid-centered
loss of off-site power.
After evaluating the design weaknesses, the
licensee took voluntary actions that were acceptable to
staff to address the concern. They went back in and they
modified their procedures to ensure that 120 volt power was
resupplied to two valves in their system that would have
lost -- they were not supplied with on-site power, and even
though the pumps were available to operate, these two valves
would have isolated and stopped cooling to the spent fuel
pool. So, the LaSalle issue went away.
In addition to performing probabilistic
evaluations, the staff gathered and evaluated further
information about certain design features to determine
whether further regulatory actions were necessary. This is
the second group, the group of four that we are collecting
additional information on. The staff looked at all the
plants with liner leak-offs to make sure that the make-up
capabilities exceeded the leak-off rate should a tear
develop in the liner that maximized the leak rate through
these leak-off lines. And we found that all of them did
have make-up that exceeded that capacity.
The staff reviewed plants in the group for limited
decay heat removal to determine whether any additional
administrative controls were warranted. The staff found
that licensees had procedures to take early actions on the
loss of decay heat removal, to isolate the purification
system and to line up make-up early. They also took actions
to refuel in months where their ultimate heat sink
temperature was lower and, thus, the cooling systems that
reject heat to the ultimate heat sink were lower.
We found that they did not have a problem, even
though when we looked at their design numbers, and when you
take into consideration the design of these plants, you look
at the maximum heat load and the minimum capability to cool
it, so the highest ultimate heat sink temperature, what
their actual practices were, they would tend to refuel in
off months and they were able to keep their temperatures
well below even though their low temperature alarm of 125
degrees.
For the refueling cavity seal issue, AEOD
identified a loss of the seal during refueling could
dramatically lower the spent fuel pool level. The staff
previously reviewed this issue in response to an event at
Haddam Neck in 1994. The staff found that no plant was
vulnerable to this type of failure, to the type of failure
experienced at Haddam Neck.
Design changes included those that employ a solid
wedge type primary seal to reduce the probability of
significant leakage through the seal. Other seals employ
similar design features or ones that act to reduce flow from
the reactor cavity seal should a leak occur.
There was -- I believe there was four plants with
this design that had pneumatic components to them. We
looked at all of them and we felt confident that the work
that was done in response to the bulletin in 1994 would
limit the flow out of a cavity seal such that the spent fuel
pool could be isolated if one of these minor leaks were to
occur.
DR. WALLIS: What is the life of these seals?
MR. GRAFTON: That I don't know.
DR. SIEBER: I think that the probability of a
malfunction during installation or removal, or the failure
of the air supply pressure to the seal itself is more common
than the seal just aging and falling apart.
DR. WALLIS: Because it does eventually age,
doesn't it?
MR. GRAFTON: It does age, but they are all tested
prior to having them, you know, filled and aligned. They
put them in place and then with the spent fuel pool
isolated, they fill the area around the seal and check for
leakage. So, it is tested in place. If it were degrading,
it would --
DR. WALLIS: Then replacing a seal is not -- to
what activity?
MR. GRAFTON: I would I think it would be
extremely large activity.
DR. WALLIS: I think it would be, wouldn't it?
MR. GRAFTON: Yes.
DR. KRESS: I recall Sandia did some aging studies
on those seals, and there are numbers for how long they
last. I don't recall what the results were, but there were
values available for that.
MR. GRAFTON: I am not familiar with their
inspection activities on the seals, so I can't comment on
that.
DR. SIEBER: What is your modification to the
gates that close the gap between the wall of the pool and
the edge of the gate, so as to minimize the flow through a
failed seal?
MR. BARTON: That sounds familiar.
MR. GRAFTON: Are you talking about spent fuel
pool gates?
DR. SIEBER: Gates, where the seals are.
MR. GRAFTON: There was an event, and I believe it
was at Hatch, where they had a gate that had a double seal
but it was supplied from the same air supply.
DR. SIEBER: Right.
MR. GRAFTON: And the air supply failed and caused
a leak through that gate. And they came back and they split
the air supplies such that, you know, half the seals were
provided from one unit's air supply, the other one from the
other air supply.
DR. SIEBER: I seem to remember some kind of a
physical barrier that they installed to help close that gap.
MR. HUBBARD: This is George Hubbard. Jose Ibarra
just told me, I guess from the AEOD study they looked at,
and the cavity seals are periodically replaced.
MR. GRAFTON: For the infrequently used backup
system, the staff verified that the backup systems are
aligned and tested before they are put in use, and there are
administrative controls to ensure that these systems are
operated properly.
One other note was that we found that the
licensees employed outage safety assessments in a manner
consistent with NUMARC 91-06, Guidelines for Industry
Actions, which gives outage safety guidelines for ensuring
that there is adequate core in spent fuel pool cooling at
each site. So, for the sites that we looked at infrequently
used backup systems and the limited decay heat removal
capability, this -- each one of those sites had guidance
documents that ensured that the systems were available,
i.e., they had all their maintenance performed on them
before the outage would start, and they were tested in place
before they were needed to be used during the refuelings.
To sum up what's been presented, the technical
issues that were identified by the Part 20 report, and the
issues identified during the Generic Spent Fuel Pool Task
Action Plan, have been evaluated by the Staff, and found
that the plants are in compliance with the current
regulations.
The Staff's identified design features as a result
of the review that we felt warranted further review and
evaluation to determine whether the safety enhancement
backfits were warranted.
Some design features received regulatory analysis;
others, the Staff gathered information on about the design
feature. As a result, the Staff could not justify safety
enhancement backfits at any plants.
The Staff completed their review of all technical
issues, and we plan to close GSI-173 on this basis.
DR. KRESS: Let me ask you a question about the
backfit procedures, rules, backfit rule: You have a safety
goal screen, and then you look to see if there is
substantial increase in protection of the health and safety,
and substantial decrease in the risk.
And then you look at cost/benefit. Where is the
substantial stuff? Is that in there before the safety goal
screen or after it, or is it a part of the safety goal
screen somehow?
I didn't see any here where you looked at delta
changes and asked if this was a substantial or not, and I'm
not sure where it fits into the regulatory analysis.
MR. LEE: This is Sam Lee. We didn't look at a
delta, per se. When these plants were identified as having
unique features, we wanted to take a look at what the -- I
don't want to say the word, risk, because the end state that
we evaluated were far, far -- were pretty conservative.
MR. COLLINS: This is Tim Collins. Dr. Kress, in
regard to your general question. Substantial would come
after the screen. First, you get passed the screen, and you
don't even look at substantial.
Then you look at substantial, just in case. You
may get a very small benefit, which is still cost-
beneficial, because of the very small cost associated.
Okay, so first you've got to get by the screen, and then it
has to be substantial enough to make it worthwhile at all.
Those are really the steps: Screen, substantial,
cost-benefit, right.
MR. GRATTON: Anyway, that's the end of my
presentation.
DR. KRESS: I have no more questions.
DR. WALLIS: Did someone on this Committee look at
this report?
DR. KRESS: I did.
DR. WALLIS: So the report has been -- I want to
make sure it isn't an oral evaluation because --
DR. KRESS: We had intended to have some
committee meeting on this, but we decided we could look at
the reports and handle it all in one full Committee meeting.
Are there any other questions?
CHAIRMAN POWERS: I'm still a little uncertain on
where we stand with these spent fuel pools as a risk
contributor. It looks to me like this analysis looks very
much at the boil off scenarios.
And the question comes about, what about the drain
off scenarios initiated by the seismic event? It looks to
me like it looked at primarily the facilities susceptible to
design basis earthquake damage, and they're not -- they're
design to meet that earthquake.
DR. KRESS: Yes.
CHAIRMAN POWERS: And so we come back to Dr.
Wallis's question. Surely there is some earthquake that
will fail those, and with some repeat frequency.
And so what is that risk? What risk do they pose
when you consider those earthquakes? I guess I don't have a
real good feel for where we stand there.
MR. HUBBARD: Let me refer you back to the GI-82,
which took a look at whether the zirconium fire at an
operating plant -- and that was done back in, oh, mid-80s.
And based on that study, it determined that the
seismic risk was the significant contributor to it, the
spent fuel pool zirconium fire.
But from that study, we were not able to justify
backfitting that on industry, as the licensing basis or a
design basis requirement.
And it may be that you want to go back and take a
look at the GI-82, which I think probably more addresses
your question.
DR. KRESS: What was the basis for the lack of
justification? Was it also a safety goal screen, or was it
a cost/benefit?
MR. HUBBARD: I believe the answer is, yes, it do
a cost/benefit. It was a safety goal screen, yes. So I
think that for that issue, the whole regulatory process was
followed for, you know, justifying the backfit.
And we were unable to justify the backfitting.
DR. KRESS: Does the -- how -- this is a
regulatory process question, I guess. If you read the
backfit rule, it doesn't -- it's silent about safety goal
screens.
But if you look at your Regulatory Analysis Guides
for rulemaking and backfitting, that's where you see the
safety goal screen brought up.
And it tells how to do it, and gives acceptance
criteria, or gives a matrix for what you do, depending on
various values.
The Regulatory Guide, does that bind you to
actually go by the results that one gets from following the
Regulatory Guide, or is that just -- or do you have a lot of
flexibility in looking at that and saying, oh, maybe that's
not quite appropriate for the given case or something? How
binding is that on your ability to do a backfit or
something?
MR. COLLINS: It's not binding. You can make
recommendations which take into account, other
considerations which you can't quantify. So the Guide is
really a guide, yes.
DR. BONACA: I had just a question. As we look at
these power plants, most of them were designed with spent
fuel pools, and there was an understanding that by sometime
in the early 80s, fuel would be taken out.
Most of them were not ever filled their pools. In
fact, many of them right now around with full spent fuel
pools, and they're putting additional fuel in dry
containment storage and so on and so forth.
I think there has been some kind of true shift
here in the scenarios we're looking at insofar as spent fuel
at sites. Have you thought at all as you were reviewing
this issue of the spent fuel pool, about the fact that there
has been truly a shift.
These facilities now have, if you go and travel
around, you see that they have spent fuel pool full, some of
them are expanding, like the Sharon Harris facility where
they will have literally each pool 4,000 assemblies of
different types, loaded and supported by the same systems.
And many of these sites have also now dry storage
facilities there. I mean, you know, it's a different
scenario from what we envisioned in the early 80s.
Have you given any thought about how that would
affect to some degree, this issue of spent fuel storage?
MR. HUBBARD: I think the answer is, yes, we did
put some consideration to it. And, in particular, when a
utility comes in and looks at the -- you know, they ask for
a re-rack or increase in their capacity, you know, we look
at it from the thermal hydraulic standpoint, the
criticality, the structural standpoint.
The Division of Engineering gets involved in
looking at these issues. The people, Chris and the others,
Steve Jones, who was involved in this very deeply, was very
well aware of the fact that now we're dealing with more fuel
in the spent fuel pools. We're packing them in tighter, and
that was part of the consideration.
As you go through this, it's concern with having
the more fuel there is, you know, the temperature control.
You know, what is the --
DR. BONACA: Well, I think there is a rising
public expectation for this final repository somewhere. I
will expect that we will see a rising public interest on
these repositories that are not anymore temporary. In fact,
in some cases, they may become permanent.
MR. HUBBARD: Yes.
DR. BONACA: There is a real shift there, and I
notice also a shift in public interest.
MR. HUBBARD: And I think definitely there is the
shift. As you mentioned Harris, you know, license amendment
that's before a hearing panel, Millstone III also is in for
a re-rack.
There is public interest there, so, one of the
issues that got us even more focused on this was the
Millstone I and the Time Magazine article.
And so we were quite aware of the increased
storage, and, you know, the public interest. And it's --
you know, yes, people are concerned, and we are aware of
that.
CHAIRMAN POWERS: It looks to me like the
resolution of this issue is focused very much on full
probabilities of occurrence.
And I don't have a good feeling for the actual
risk, because as a strong believer in the structural
component of defense-in-depth, I said, well, you can prevent
accidents only so far.
I have qualms about or arguments about those to
ten to the minus fifth or ten to the minus sixths kinds of
levels. What about consequences of these accidents?
I don't have a good feel for what that product of
probability times consequence is.
DR. WALLIS: I think there is also the question of
how risk averse you are; that these things get less likely
but have bigger consequences. It's not necessarily a linear
process that you go through in evaluating your criteria for
decisionmaking.
DR. KRESS: Well, I certainly share Dana's
feeling.
CHAIRMAN POWERS: You don't believe in prevention
lower than ten to the minus fifth or ten to the minus sixth,
and, you know, there are not too many options here, unless
you're going to build a containment over it.
DR. KRESS: You almost have to rely entirely on
prevention, and some mitigation, based on operation action.
But the question is, have we provided enough
prevention based on the real risk? And that involves
knowing what the consequences actually are.
DR. SHACK: Just coming back to Dana's question
again, when we went through for the decommissioning plants,
I thought the seismic was really the thing that left you out
there.
And what was the fraction of plants when you -- I
mean, there they looked at the severe earthquakes, and there
was a -- what was the fraction of plants that didn't have
the high confidence?
DR. KRESS: I don't recall, but the problem is
that with decommissioning plants, there are a limited number
of those at any given time, and a limited amount of time
they're at risk.
You know, the risk goes away after a certain
amount of time.
DR. SHACK: After a couple of months.
DR. KRESS: Here we're talking about operating
plants where the risk is there all the time. And it's all
of them. So it's quite a different situation.
DR. SHACK: You don't know how often the pool is
going to get busted up.
DR. KRESS: Well, that should be part of a risk
analysis, and that's what Dana, I think, is partly asking.
Do we have a good feel for what the full number of -- the
full probability is and the full consequences, which we need
to know for the full risk equation?
And whatever that value is, which maybe we have to
go to this GI-82 to find out something about it.
Then the question is, is that an acceptable number
to us, even though there are no risk acceptance criteria out
there that are binding by law, other than the safety goals.
And so, I think that's the question that bothers
Dan and bothers me also, is, I don't have a good feel for
what the risk actually is, because I haven't seen a good
convolution of the frequency times the consequences. And
that's the thing that bothers us.
CHAIRMAN POWERS: The question comes up that
you've got some coupling between these systems and the
systems of an operating plant.
And so when I come in to use Reg Guide 1.174 and I
want to locate myself on a horizontal axis, is this
probability big enough that I need to take it into account?
If I locate myself on that horizontal axis, and is the
change that I'm proposing to the plant causing a delta in
this risk that I need to consider, along with the delta on
the normal operation risk?
I don't have a good feel for that.
DR. KRESS: I don't either, and it's a good
question, because I don't think 1.174 deals much with spent
fuel pools.
CHAIRMAN POWERS: I think it's a stepchild of all
of them.
DR. KRESS: A stepchild of all them.
CHAIRMAN POWERS: One of the things that people
struggle with, I think, when we ask these questions, typical
PRA really doesn't make the rest.
DR. KRESS: For example, how would a PRA determine
the effects of steam on safety equipment and a change in
reliability of that, given the number of hours of exposure
or something? You couldn't deal with it very well.
CHAIRMAN POWERS: I'm not sure we have good
database, deterministic number on that.
DR. KRESS: I don't think we have, either. So,
it's a question that is another one of those model
uncertainties that you have to deal with with defense-in-
depth, I think. How much defense-in-depth do you put on it?
I don't know.
Well, you can see what's bothering us.
MR. HUBBARD: I tried, in looking at it, is that,
you know, we did the GI-82, and determined that we couldn't
backfit the zirconium fire. We had these issues that came
up with regard to Susquehanna.
We took a look and addressed the Susquehanna
issue. We then took the bigger generic look, taking to see
did the plants -- do they meet our regulations? Is there a
big concern there?
We identified that the plants were meeting our
requirements or guidance. Except in some cases, we went and
made plant-specific backfits, and so, you know, we came to
this conclusion, and as you probably have recognized, we
probably haven't been looking at operating plants.
We probably should have asked you, when you asked
us whether we wanted a letter from you back in '96, we
should have said, yes.
We didn't, so, the thing that I would like to also
mention is the fact that the operating plant still has its
full EP, you know. That's there.
You've got the full complement of safety and non-
safety systems that, you know, we've had all along. And I
think the biggest part of risk at the operating plant is the
reactor itself.
And, yes, we didn't go through and carry this
through to give you other than the GI-82. We didn't go
through in this effort that we've been doing in the 90s to
carry it through to the risk which includes consequences.
And that's your concern.
CHAIRMAN POWERS: Maybe the situation here is that
-- maybe you've got a specific question that's addressed by
this GSI. Maybe this is resolved, this specific question,
but we've got this larger question that Art Bonaca
mentioned. It's a permanent feature that's not going to
change until the Department of Energy gets its act together.
Spent fuel storage --
VOICE: I can't hear you.
CHAIRMAN POWERS: Storage -- and maybe we need to
look at it in a more holistic fashion, independent of this
specific question, because this is a fairly specific
question.
DR. KRESS: What is the specific question being
asked of this GSI? I think it is, are there particular
vulnerabilities to loss of spent fuel pool cooling and loss
of inventory.
And I think you have to think the risk issue is
tied into the answer to that question, because that's what
you mean by vulnerabilities.
I don't think we can divorce this particular issue
from the question of risk.
DR. BONACA: The reason I raised that issue before
is that it ties into the issue of consequences. I mean, the
size of, the amount of spent fuel in the pools, as well as
in the dry storage, et cetera, has created really a
different kind of scenario from what we saw just a limited
number of batches discharged and they're removed from the
sites.
So, to some degree, it ties into the issue of
understanding the potential.
DR. KRESS: Every couple of years, you get a fresh
load of fuel.
DR. BONACA: That's right, and you add up, and you
have thousands of assemblies now sitting there. So, it ties
into the consequences.
DR. KRESS: I'm reluctant to divorce this issue
from the risk.
Any further comments?
[No response.]
CHAIRMAN POWERS: Thank you very much. I think
that was a fine briefing that you gave us. I very much
enjoyed it.
We'll recess now until 17 after the hour.
[Recess.]
CHAIRMAN POWERS: Let's come back into session.
Our next topic is one I am really looking forward
to. We have assumed that in the move for greater use of risk
that indeed we would be able to have a more quantitative and
better understanding of regulatory effectiveness and as a
first step in that area we are moving toward -- we are going
to hear a briefing on the regulatory effectiveness of the
station blackout rule.
Mario, I think this is a topic you are going to
lead us through?
DR. BONACA: Yes, Mr. Chairman.
We have a report in front of us that shows that
the station blackout rule has brought significant risk
reduction on the one hand, and it will be interesting to the
committee to see the specifics of it.
It also shows that some of the benefits obtained
by the implementation of the rule are somewhat being eroded
by some conflicting guidance and regulation and there are
opportunities there for reducing this erosion by some
clarification, so I think it will be an interesting thing
for the committee to look at how the special blackout rule
was implemented and the benefits brought about and the
opportunities there.
With that I will let the presenter go ahead.
MR. ROSENTHAL: I am Jack Rosenthal. I am the
Branch Chief of the Regulatory Effectiveness Assessment and
Human Factors Branch in the Office of Research.
Our office director, Ashok Thadani, asked me to
make some introductory comments.
We try to relate our work to the Agency's goals
and we have this initiative to look at the effectiveness of
major rules, which is clearly related to the maintain safety
goal or factor, and you will hear about that.
Equally or perhaps more important, throughout the
Agency, another one of our goals is to make things more
efficient and effective and we used this as almost a
buzzword and it means different things to different people
throughout the agency, and sometimes, often we are talking
about it in terms of our internal processes, so here is an
example now in our branch where we are trying to look
external to the Agency and to say, okay, has this Agency
been effective in achieving its desired outcomes external to
the Agency. I think it is an important piece of work.
We are budgeted to do about two rules a year. The
first one out of the chute is the blackout rule. Shortly
thereafter we will look at, we will publish a draft report
on ATWS, then we are going to look at Appendix J, Option B,
for which we put in a lot of work and want to see what it
would get for us. After that, we are going to take a look
at the resolution of A-45, decay heat removal, which did not
result in a rule but there was considerable action, and then
we will march out in time from there and there will be
progressively more external impact on how we pick our rules.
The particular report now is a draft report and we
decided that it would be better to come to the ACRS as a
draft where you have an opportunity to influence the product
rather than to come with a final and say it's done, so that
is the mode we would like to be in.
But it is a draft report and we have already
gotten comments internal to the Staff that have been
incorporated in the report and we have provided it to Union
of Concerned Scientists, Public Citizen, EPRI, NEI, INPO, et
cetera, for public comments. I think now is an ideal time
to publicly discuss what is in the report.
With that, I am going to turn it over to Bill.
MR. RAUGHLEY: What we are talking about is the
draft report, Regulatory Effectiveness of the Station
Blackout Rule. As Jack briefly mentioned, this report
reflects comments made by NRR and the Regions and that we
have asked the industry for their comments, specifically the
reasonableness of the approach, the appropriateness of the
conclusions, and what other rules, Reg Guides, inspections
procedures that we should address as part of this regulatory
effectiveness process.
I am addressing this in the context of you were
asked to be informed about the paper -- so I'll pick certain
highlights to speak to.
As some background, the report will provide a
basis to respond to the Commission. We expect to write a
SECY after revising the report to address the industry
comments. The Commission first asked the question after
hearing about the NUREG-1560, which provides the
perspectives on reactor safety, and that was back in 1997.
At that time that NUREG showed a station blackout
and ATWS to be dominant contributors to core melt and the
Commission asked what about the effectiveness of the SBO and
ATWS rules in view of that, and then this became an action
item in the PRA implementation plan.
Station blackout is defined in 10 CFR 50.2. It is
a complete loss of offsite and onsite AC emergency power and
a turbine trip. The risks were first known in WASH-1400,
1975 report that highlighted the station blackout could be
dominant, and if you look at the 1560 which summarizes the
IPE PRAs for all the plants in the industry you would
conclude it is a dominant or the dominant contributor to
core melt.
As far as some historical highlights before the
station blackout rule was passed there were requirements to
address offsite and onsite power. Specifically we are in
the 1974 to 1977 timeframe.
We had Reg Guide 1.93, which addressed or required
LCOs, limiting conditions of operation, when you have less
than the required number of either offsite or onsite power
supplies and if those problems persist for a certain amount
of time they are required to go to hot shutdown and
shutdown.
There is Reg Guide 1.108, which has since been
cancelled but that required tests to demonstrate the
reliability, and those of you familiar with the topic may be
familiar that plants were required to do 69 tests per site,
23 per plant and to demonstrate diesel reliability with no
failures.
That Reg Guide also speaks to EIS CB-2. It is
Branch Technical Position from the Electrical and Control
Branch, which is part of the Standard Review Plan and that
establishes that these are reliability goals, 99 percent at
a 50 percent confidence level, so it is a point estimate.
Reg Guide 1.9 has been revised several times but
that basically addresses start and load testing as EQ
testing of the diesel when it was originally purchased, and
you may recall that was the famous 300 start/stop test with
no failures.
Some of the evolution to the SBO rule, the
Commission, A-44, is an unresolved safety issue. The
Commission thought that in view of events in the late '70s
dealing with loops and diesel unavailability that this
should be a generic issue. Just some of the scores at the
time -- there were 11 of 78 plants surveyed had less than 95
percent diesel reliability, two of 78 had less than 90
percent reliability, only seven of 57 sites monitored diesel
reliability and only three of 56 sites surveyed kept records
of diesel reliability.
In response to A-44, the Staff issued NUREG-1032,
and this integrates the results of several engineering
analyses that were completed.
DR. WALLIS: If they didn't keep records of
reliability, how did you know that they were 90 percent
reliable?
MR. RAUGHLEY: The NRC had sent out an information
notice and asked people for bean counts.
DR. WALLIS: They hadn't been keeping records so
maybe this was an assessment other than based on tests?
MR. RAUGHLEY: There was a Generic Letter, an IN
sent out, and I guess they provided them with the
information.
NUREG-1032 pulled together several engineering
studies that were done by contractors for the NRC to address
different aspects of station blackout. That report provided
four findings or results that provided the basis for the
station blackout rule and the accompanying Reg Guide 1.15.
That is that the station blackout was highly dependent on
the redundancy and reliability of the EDGs and it was also
highly redundant on the frequency and duration of the loops.
The NRC also performed a regulatory analysis which
provides the cost benefit analysis of implementing the
station blackout rule and that was documented in NUREG-1109.
The station blackout rule itself, there's three
basic parts -- (a) requires that the licensees be able to
withstand the station blackout for a certain amount of time
and recover from that event, and that the duration of
station blackout is based on the four factors I mentioned
from NUREG 1032, and it requires licensees to do a coping
analysis to demonstrate the degree to which the support
systems will support a station blackout, and lastly it asks
licensees to specifically supply the coping duration based
on the plant design factors they picked, procedures offsite
and procedures to cope with an SBO specifically in terms of
recovery from loss of either or both of onsite or offsite
power and to make any modifications necessary to achieve
desired coping times.
Other documents that are related to the rule are
highly important in our assessment.
Reg Guide 1.155 was the Reg Guide that accompanied
the station blackout rule and that establishes specific
diesel reliability requirements, that they be .95 or .975
reliable. It requires or provides guidance for reliability
programs. It provides guidance for procedures to restore
offsite and onsite power and it provides the tables --
there's eight or ten tables -- that they go through to
determine the coping capability based on severe weather
condition category, offsite power configuration, onsite
power configuration, et cetera.
DR. WALLIS: If 95 percent is the number and the
reliability of a truck diesel is 99.999 and they always
start -- why is this not the same?
MR. RAUGHLEY: I don't know.
CHAIRMAN POWERS: Big diesels?
MR. BARTON: Big diesels that were designed to run
all the time but they sit there.
DR. WALLIS: They sit there. That's the problem.
DR. KRESS: And they have to have a signal to
start them up.
MR. ROSENTHAL: Let me also point out that what
you are talking about is the emergency start of a standby
electrical power station, and that the focus has been on the
diesel engine but it includes the diesel engine, the
generator, the voltage regulator, the output breakers, the
support systems, air support systems dependency, service
water dependencies, et cetera, et cetera, so you are really
starting, emergency starting a small power station and when
we go today and when we look at the reliability of the
equipment, which we will hear about later, it is meeting the
goals but the problems involve not so much the engine as the
output breaker or some service water dependency or the
voltage exciter, et cetera, et cetera tends to be a lot of
the peripheral equipment.
DR. KRESS: You are going to tell us how you
verify that the reliability has been met?
MR. RAUGHLEY: Yes.
DR. KRESS: Okay.
MR. RAUGHLEY: Two slides up.
The two documents in parentheses, the first one,
NUSTACK-108 is an EPRI document that the Reg Guide 1.15
relies on for the definition of valid starts and stops and
the Reg Guide also offers a NUMARC document for an alternate
means of compliance and that closely parallels the Reg
Guide.
The station blackout rule resolution specifically
referred the details of the reliability program to the
resolution of Generic Issue B-56, which is diesel
reliability and that resolution tied the reliability program
to the maintenance rule, so we get into a lot of maintenance
rule documents -- Reg Guide 1.160, the corresponding NUMARC
document 9301, which references another NUMARC document,
which refers to Rev. 1 of NUMARC-87-00. We have the
inspection procedure and there is a maintenance rule
handbook so that there's a number of documents.
As far as the assessment, the definition we picked
for regulatory effectiveness was that a regulation is
effective if the expectations are being met.
DR. WALLIS: I am not sure that many regulations
are too clear about the expectations --
MR. RAUGHLEY: I'm sorry?
DR. WALLIS: Just I am not sure that expectations
are specified in a way that enables measurement of them but
I see regulations, so I would like to see a much clearer
statement of what an expectation is from them in the
statement of considerations or something, and it often isn't
there, so I wonder how you determine what the expectations
really are.
MR. RAUGHLEY: It's on the next --
DR. WALLIS: Maybe the next station blackout rule
has a good job done on expectations.
MR. RAUGHLEY: There could be very objective
expectations in terms of diesel reliability, costs, risk
reductions, coping time. This is one of the more --
DR. WALLIS: Explicit --
MR. RAUGHLEY: -- performance-based, risk informed
rules that we have, I think. It is quite quantitative.
I'm doing the ATWS rule also, but that's not the
case. But the deregulation within the regulation would
include the rule, the accompanying Reg Guide, and the
accompanying inspection documents.
But the scope of the assessment was to determine
if the rule is effective and if there are any areas that
need attention.
I didn't try to second-guess. I worked only from
documentation. I worked only from publicly available
information. I didn't try to second-guess or read anything
into it.
I didn't rely on what people said they did, just
tried to keep it objective.
DR. WALLIS: Now that you've set the standard, we
can use these when we look at others.
MR. RAUGHLEY: That was the point, we would
address something that was a little cleaner than others and
maybe have a template to address other things that maybe
aren't as effective.
I did not address plant-specific problems. One of
the comments from the internal review I did have some plant-
specific problems in there, and the consensus was it drew
too much away from the station blackout rule itself, and
drew too much focus to plant-specific problems.
And consistent with the original station blackout
rule, which didn't address seal failure, we did not either,
however, that was resolved by RES on 11/99.
DR. SIEBER: Isn't the seal failure for a number
of plants, though, the dominant pathway to core damage? So
if you don't address that --
MR. RAUGHLEY: That was addressed as a separate
generic issue.
DR. SIEBER: But you did put a number in there
that says the risk, the overall risk of SBO includes
potential for seal failure in certain plants, and that
dominates the risk?
MR. RAUGHLEY: That number is in the risk numbers
in the appendix of the report. Again, I did what the
station blackout did, so it deferred the resolution and the
discussion. That's all another subject.
DR. SIEBER: Okay.
MR. RAUGHLEY: But as a matter of interest, I did
point out in the report that there's a paragraph in there
that discusses an 11/99 resolution if GSI-23. The method
was, we compared the expectations to the outcomes in areas
where we could find objective measure being risk, value
impact, coping time, and reliability.
And we used operating experience trends to look at
the loop frequency and duration. They were readily
available in NUREG 54.96.
The data, I used publicly available data. We
obtained the expectations from NRC documents.
There's an FRN which is very detailed, that issued
the station blackout rule as the statement of considerations
of all the relevant documents and summarizes the relevant
numbers.
For the outcomes, I used the NRC databases of the
IPE. I developed from the IPE, LERs, the station blackout
rule safety evaluations completed by NRR and several NRC EDG
reliability studies that were completed and a few in
progress.
DR. WALLIS: Are you going to go through the
outcomes?
MR. RAUGHLEY: Yes. The areas that we addressed,
as I mentioned, were risk, value impact, coping time, and
EDG reliability. In the area of risk, the expectation was
that the mean industry SPO CDF would be reduced by 2.6 E to
the minus five.
And we actually achieved a reduction of 3.2 E to
the five. And what the rule did was, it had five or six
licensee-specific IPE PRAs. I believe the NUREG 1150 plants
had models of its own, and from that it was able to
determine that the average mean SBO CDF was 4.6 or 4.2 E to
the minus five, and it set a goal to reduce that to 1.6 E to
the minus five.
In fact, the industry average is 1.0 E to the
minus five.
As it turns out, what I did is, I rank-ordered the
plants from the highest to lowest loop initiating
frequencies, so those -- there were 21 plants that had a
loop initiating or loop frequency of .1 or greater.
And 15 of the 21 plants had an alternate AC power
supply. Nineteen of these 21 plants had reduced their
station blackout CDF to less than ten to the minus six.
Eight plants that had the most severe weather
category, ESW-5, and the ten plants that had the eight-hour
coping time, also had alternate access power supplies.
What we did find is that the Reg Guide 1.93, which
I mentioned before, requires plant shutdown with less that
the full complement of power supplies, and in the context of
station blackout, that would add risk, so we address that in
the conclusions.
In the area of value impact, that was completed in
NUREG 1150, and it's dependent on the risk and estimates of
the mods that were expected from the station blackout rule.
And what the NRC did is, they issued that as a
draft, initially, and significant revisions were made as a
result of industry comments that our costs were too low, so
we ended up using their numbers.
So the basis was to establish them on our risk
number and their cost estimates. And it was done on averted
rem and it gave a range of values, and the result is that we
are just within the range of values.
But there are two reasons for that. One is the
original expected that 39 plants would make modifications.
And at least 72 made modifications.
These ranged from adding diesels, to making cross
ties, to changing, upgrading battery sizes. There were
several hardware modifications made. This just wasn't a
paper exercise.
DR. WALLIS: Why didn't the averted rem go up?
MR. RAUGHLEY: The averted rem stay the same, but
the costs --
DR. WALLIS: Why didn't they go up?
MR. RAUGHLEY: The original cost was based on 39
plants making modifications, and actually 72 did.
DR. WALLIS: I can see why the cost goes up, but
surely the averted rem would also go up; wouldn't it? I
don't know why it stayed the same.
MR. RAUGHLEY: Well, it went down, the risk went
down.
DR. WALLIS: The person-rem seemed to stay the
same. I would expect it to have changed, to have gone up,
actually. You've done a better job so you've averted more
rem; haven't you?
MR. RAUGHLEY: But you spent significantly more --
four times more money.
DR. WALLIS: I understand that. The money, I
understand, but the rem, shouldn't the rem change, too?
You've changed the CDF.
MR. RAUGHLEY: The rem would go down in
proportion.
DR. WALLIS: But it stays the same, the same
number of averted rem.
MR. RAUGHLEY: The arithmetic worked out --
DR. WALLIS: It puzzles me. I think that would
change.
MR. RAUGHLEY: The cost that was expected that
they would spend $60 million, and they actually spent $200.
DR. WALLIS: So this averted person-rem is
calculated, though, but --
MR. ROSENTHAL: Let me try again. If there were
plans that chose to add additional equipment, such as chose
to add a diesel, and they may have made that choice for a
combination of risk and also operational convenience, and
there are circumstances where they are incurring the costs
or racking the cost of that additional diesel against the
blackout rule.
But in the risk analysis, there isn't a
corresponding --
MR. RAUGHLEY: Reduction.
MR. ROSENTHAL: Reduction of risk, but rather that
diesel was added because they thought it was proper to do,
or it gave them operational flexibility, et cetera.
And that's why --
DR. WALLIS: You've reduced CDF by this three to
minus five, which sounds great. Why isn't there some change
in your expected averted rem because you've done better on
CDF?
DR. KRESS: Well, it's not much better. It's
about the same, and so you get about the same averted rem as
you were expecting, I think.
DR. WALLIS: I was just puzzled that someone had
carried the same calculation over without changing it.
DR. SHACK: It's the same rounded off.
DR. KRESS: I think it's roundoff of significant
figures or something, 2.6 and 3.2, in my mind, are the same
thing, so you get about the same.
DR. WALLIS: Well, the 145 and 145 are exactly the
same.
DR. KRESS: Yes, that is peculiar to some extent,
yes.
MR. RAUGHLEY: I will get back to you with the
specific arithmetic.
DR. WALLIS: It will look better if you do.
MR. RAUGHLEY: Pardon me?
DR. WALLIS: If you check it, it might look
better.
MR. RAUGHLEY: Well, what I did there was, I made
a list of the mods which are an appendix to the report from
the licensee submittals. I used the industry estimated
costs.
Licensees also added power supplies. There were
19 diesel generators added.
Davis Besse was the only licensee that supplied a
cost estimate for the diesel additions, so we used that for
the others. They said they spent $9.07 million.
They had a non-safety diesel, and they received $5
million. They estimated $5 million in cost benefit from
reduced outage time and increased additional -- not having
to replace power.
In the area of coping time, the Reg Guide provided
for licensees to select two-, four-, eight-, or 16-hour
coping times. And most of the licensees ended up in the
four- and eight-hour range. That's documented in an
appendix in the back of the report.
In the area of diesel reliability, the Reg Guide
required that the licensees establish a goal of 95, based on
individual diesel reliability.
We have a report, INEL report that was done in
1995 for diesel operating experience between '87 up through
'93. I used a draft report that's out for comments, which
updates the operating experience from '87 through '98, so I
relied on that because it has more of the post-station
blackout rule operating experience.
But that report uses the unit average train
performance. And so you sort of are comparing an apple and
an orange here, but you have to recognize that if the
individual reliabilities -- well, if the safety performance
is better than 95, then the individual reliabilities would
have to be that.
It would be conservative. The INEL report is
conservative in that it may not show all the diesels that
have less than 95 percent reliability, but nonetheless,
that's what we had to use.
What was good about the INEL report was that it
developed a standard model for diesel reliability. So all
the licensees have different ways they've modeled the
diesel, so what it did is, it created its own model, and
then it dissected each PRA to take the inputs and run it
through their model. So we have a consistent comparison
across the industry, and I thought that was really good to
get away from all the diversity between the different IPE
PRAs.
The results were that .95 target reliabilities
were generally met with and without MOOS, which is
maintenance and testing out of service while at power. And
what this accounts for is, if a licensee takes a diesel out
of service while the unit is running, and then has a loop,
loses voltage to the bus and it's called into service, then
that would be counted as a demand and that would count
against them. We'll get into that a little more pretty
quick here.
And the .975 target reliabilities, the MOOS had
risen to levels that many or most of the licensees were not
able to meet the .975 target considering MOOS.
We've done a -- there is another NUREG available
which looked at the risk significance of MOOS, and basically
it was based on six IPE PRAs, a sample.
And it shows that less than two percent MOOS is
likely not to have much risk impact but values between 02 to
04 could be risk significant in that they would cause
changes on the order of ten to the minus five to the risk.
What I also did was, I took the three lowest
values from the INEL report and did a followup at the plant
sites to find out that when we think or we're saying the
reliability is less than .95, they think it's .99, and I
identified some differences in the scorekeeping here that we
need to clarify.
Basically, this report, the INEL report, is based
solely on actual safety demands of the diesel while in
service, and the annual tests where you actually run the
diesel through its full pace.
So it's based on those demands and starts. And
you would expect that if -- well, you would expect the plant
site numbers to be equal to those numbers, and there are
some different methods of scorekeeping going on that need to
be addressed.
The INEL report also shows that many licensees
have achieved higher diesel reliabilities than used in their
IPE PRA, so if they were to plug those numbers into their
IPEs, they would obtain additional risk benefits. There is
some unclaimed risk benefit from the station blackout rule
that we have yet to -- you have to recognize that it is
there, but it hasn't been claimed yet.
In the comment cycle for the internal comment
review, there was considerable discussion on what was the
diesel performance basis. So what I did was to address all
those concerns. There is two sections of the report that
pretty closely plagiarized what is in the Reg. Guides, with
specific reference to the section, so there is no question
about what is said. And then I go on to show, to point out
the conflicts between those individual sections.
You have NUREG-1032, which is the technical basis
of the station blackout rule, and that established that MOOS
was small, both at power and non-power, it was .006. It
established -- it used the diesel, or the emergency power
system boundary to include the load sequencer and the bus,
and it only used -- and it used actual tests and unplanned
demands to count valid start load runs.
Then we get to Reg. Guide 1.15, establish the
target reliabilities, and, specifically, in the discussion
it excludes MOOS, but it points out that it is small and it
used .007. And it specifically addresses that this is small
compared to the reliabilities expected in the regulatory
position. It says, however, this can be significant. It
goes on to say the contribution must be kept low. And then
they had the vision to say that as long as the
unavailability due to testing and maintenance, that is MOOS,
that is my words, that is MOOS, is not excessive, the
maximum EDG failure rates for each diesel specified would
result in overall acceptable reliability. And the fact of
the matter is, it is not -- the amount of it is not small.
I would think we appropriately considered it in the
reliability assessment.
And then the Reg. Guide also uses NUMARC 87.01,
which requires or stipulates that the licensees monitor EDG
unavailability versus the industry, and they have been doing
that since 1989, and that is available as an industry PI.
Then we get into the station blackout rule, which
gives you several alternatives. The point being there that,
with the exception of the Reg. Guide, target reliability,
all those other alternatives could be non-conservative with
respect to risk, and you could erode the risk benefits
obtained from the station blackout rule. So, there is
nothing in there to cap how much unavailability you can
have, there is nothing in those Reg. Guides that say how do
you balance availability and reliability and unavailability.
And then maintenance preventable failures is a different
score.
It also endorses --
DR. BONACA: Could you explain this balance
reliability, the unavailability?
MR. RAUGHLEY: In the discussion, the maintenance
rule has three basic elements. It has got A-2, which is
performance criteria; A-1, which is if you don't meet the
performance criteria, you go and establish goals under A-1
to get your performance back up. And then A-3 says that you
should balance reliability and unavailability. And there is
not much detail on how to do that.
So we know from 1032 and the licensee IPE PRAs,
you know, the decreased reliability or increased
unavailability erode the risk benefits. So, you have got,
if you are going to give up one, then you have got to raise
the other. And that detail is not there, which could cause
licensees to erode the diesel reliabilities. So that is the
point of that slide.
And then there is additional requirements. Reg.
Guide 1.93 was revised as part of GSI 56 resolution, and it
says the start and load run failures should include
conditional failures from maintenance, and nobody has -- you
go out and do maintenance, and you find the problem, you
don't generally count that as a failure. That is why you
are doing the maintenance and testing is to identify the
problems and fix them before they happen in service.
And the system boundary there excludes the load
sequencer and the bus. That was a point of discussion in
the review, that the INEL study included the load sequencer
and some people believe that it shouldn't. However, the
load sequencer, you need the load sequencer for the diesel
to perform its safety function. It starts and load the
diesel, keeps it from being overrated.
DR. POWERS: It seems like we have had recent
events where the load sequencer --
MR. RAUGHLEY: Yeah, there is six events, I
believe, in the INEL report where the load sequencer failed,
and that influences the reliability.
And then the last, but not least, is that the SBO
and maintenance rule inspection documents introduced another
performance standard in that they required inspectors to go
out and verify compliance to the target reliability used in
trigger values. And that was a considerable -- statistical
wars I guess went on between the ACRS and the industry from
1988 through 1992.
DR. SEALE: That is Hal Lewis.
MR. RAUGHLEY: Yes. And Med helped me dig out the
correspondence from that and Appendix D provides that the --
from what we can make there. The ACRS was using the
binomial theorem and coming up with failures and demands to
assure with high confidence that the reliability goals were
met. And the industry's position, or the industry used the
binomial theorem to develop failures and demands to assure
with high confidence that it didn't meet the goal. They
didn't think --
DR. POWERS: I would love to tell Ray, tell
that --
MR. RAUGHLEY: Yeah. So, finally, the EDO issued
a memo and that specifically says the triggers do not in any
statistical fashion demonstrate the target reliabilities.
However, unfortunately, they crept back into the inspection
procedure, so that is what we have been inspecting to.
The other point of the ACRS at the time was that
if you used the industry values, you would be waiting for
long periods of time before you identified problems, and
that is not consistent with maintaining high reliability.
If you think you have got a problem and you investigate it
and you find out it was a no-never-mind, that is the way you
assure reliability, you know, and go about it in a timely
fashion rather than waiting for the score to go bad.
And then if you also, if you use those trigger
values and superimpose those on the information in the INEL
reports, you would clearly be outside the IPE PRA bounds for
the six and eight hour -- the diesels with six and eight
hour emission times. So you would clearly erode the risk
benefits using those.
DR. BONACA: Just before we leave that point, I
mean you had two slides there, and essentially, it seems to
me that meeting the reliability targets is really central to
the station blackout success.
MR. RAUGHLEY: Yes.
DR. BONACA: And what you are saying, you have the
guidance resulting from, you know, multiple Reg. Guides and
rules, et cetera, it is confusing enough that it is not
clear what the licensees are measuring to demonstrate they
are meeting their reliability targets.
MR. RAUGHLEY: Yes. Yeah, we have got a half a
dozen different documents, all addressing diesel
reliability. They need to revised in a consistent manner.
DR. BONACA: And you are going to give use an
estimate of the amount of erosion of the 3.2 and 10 to the
minus 5 that you expect to happen from this variation,
variance.
MR. RAUGHLEY: I put two tables in the report.
DR. BONACA: That's right. Yes. Okay.
MR. RAUGHLEY: To show how much it would erode.
DR. BONACA: And I would appreciate it if you can
tell us how do you estimate that erosion. Could it be more
than what you are looking at here?
MR. RAUGHLEY: Yeah, it would be more or less,
depending on what your case CDF is. If you are SBO CDF is
10 to the minus 6 or 10 to the minus 7, it is not likely
that the diesel reliability is going to make a whole lot of
difference, but there are a number of plants, I believe, --
well, there is about half that have SBO CDFs in the 10 to
the minus 5 range and changing the diesel reliability would
affect those. And what I did was I extracted information
from NUREG-1032, went through the different permutations and
combinations of things, Table 3 or 4 in the report. It
shows you at what point it could become important.
DR. BONACA: All right. Thank you.
MR. ROSENTHAL: Let me just interject, because I
want to make sure that we, at least in my mind, that we have
right tonal quality. From the reliability studies, much of
that comes out of Pat Barnowski's branch, that you have been
briefed on separately, those are studies where you drop a
wrench on an electrical bus in the middle of the night, and
you ask, do things really start and load and power-actuated
equipment? You know, not the monthly stylized tests. So I
think that that is real good data industry-wide on how good
the diesel system is. Of course, the data density is
sparse.
DR. BONACA: Right.
MR. ROSENTHAL: Okay. But from that work, it says
that, at least on an industry-wide basis, including MOOS, we
are meeting the .95. So I don't want to be -- from a safety
standpoint, we think that you are meeting the safety
objective. There is some question on the 975. But we are
meeting the 95. And we have good, experimental basis for
saying that on an industry-wide average basis. So we don't
want to be overly shrill. Safety-wise, I think we are okay.
But what we are saying is, wait a minute, there is
all this guidance out and here is an opportunity to clear
up, to clean up the regulatory bases documents consistent
with the principles of good regulation. But I don't want to
leave the impression that we are sitting here saying that we
have an imminent safety problem, because we don't.
DR. BONACA: Yeah. No, I didn't mean that. It is
just simply that the impression I got is, given the fact
that this was a costly rule, in fact it cost four times as
much as was supposed to be invested, I mean, you know, you
certainly don't want to squander this other benefit from
this costly rule implementation, invaluable rule
implementation, by having confusing guidance. And we have
noted that issue on confusing definitions of availability
and reliabilities in other reviews. And, certainly, I just
wanted to ask that question to make sure that was the
message you were giving us.
MR. RAUGHLEY: Yes, sir. Some of the insights
from the operating experience review is some of the
modifications from the SBO rule have been used specifically.
I put an example in there on Turkey Point. Originally, they
had two diesels, two safety-related diesels and five non-
safety diesels shared between two plants. As a result of
the station blackout rule, they added two safety diesels.
During an event subsequent to the addition of those diesels,
they lost all five non-safety diesels and one of the two
original diesels. So, had they not made the mod, they would
have been reduced to one diesel.
The information in the NRC inspection report shows
that the load running on the remaining diesel would have
been about 20 percent more than its rating. We could debate
whether it would have burnt up or not, but likely it would
have. And you realize in that condition, all the protective
features are bypassed because the philosophy is you run the
diesel till it does burn up. So, fortunately, so I think
you can hold this up as a shining example of a case where
the SBO rule did a lot of good.
In addition, their initial analysis, Turkey
Point's initial analysis showed that the SBO CDF was on the
order of 10 to the minus 4 and, as a result of adding these
diesels, they were able to reduce it to 10 to the minus 6.
And the proof was in the pudding.
And several licensees, there were 19 non-safety
diesels added, they have provided many licensees with
increased allowed outage times and replacement power
benefits. So there has been some economic benefit, if that
is important.
As far as the LOOP, there are favorable trends on
the LOOP frequency and duration. They are documented in
NUREG-5496, which looks at the operating experience from
1980 through 1996, and there haven't been many LOOPs since
then either. But I think the important point there is -- I
looked at all the events where restoration took more than
four hours. For the plant events, we noted in a report that
the number of plant events more than four hours was up from
zero to four, but the recovery time stayed about the same,
about 20 minutes. For the grid events, the recovery times
increased from 36 to 140 minutes. And for the weather-
related events, the recovery times decreased from 4-1/2
hours to 2-1/2 hours.
But I think the point of all the LOOP, what we see
there is that the diesels worked when they were actually
needed.
The other insights, I think the station blackout
rule provides some defense-in-depth to deregulation of the
electric power system. We don't have a whole lot of control
or say on how that goes, but should the LOOP initiating
frequency increase, or take longer to restore power, there
are additional benefits there from the hardware that was
added.
And then one thing we did find in looking at the
events, we identified four events where there is the
potential unavailability of the alternate access power
supply when it was needed. There were AC and DC
dependencies on the unit that was down. There was an IN
issued on it, but there have been events subsequent to that.
So maybe we need some things in the inspection, attributes
in the inspection procedure to address common mode failure.
The NUMARC document that I mentioned earlier, 87001, has
specific guidance that could be used as inspection
attributes in that area.
So, our overall conclusion was that the rule was
effective and the costs were reasonable because the
reliabilities were achieved, the risk reduction was
achieved, most of the licensees -- or all the licensees
picked four hour or eight hour coping times. But there are
opportunities to improve the clarity of the Reg. Guides, and
this would be consistent with the principles of good
regulation. In the area of reliability, there is the Reg.
Guides that we went through, but they need to be revised in
a consistent manner so we put the reliability terms on an
equal footing.
Just to recap those, you have MOOS, you need to
include the load sequencer in the boundaries, address how to
balance reliability and availability, and establish common
start load run criteria. The inspection documents, you need
to delete the use of the trigger values and provide some
guidance on the common mode failure inspections. And in the
Reg. Guide 1.93, which addresses the -- requires shut down
with less than the number of power supplies, there may be a
better way to go about that. For example, you might want to
check the availability of your coping systems before you do
something like that. You might want to check that the grid,
you know, that when you shut the unit down, that the grid
are able supply sufficient capacity shutdown loads. Maybe
shutdown is not the right spot, maybe how shutdown is the
right spot, but I think we can improve that.
As a Lesson Learned, I think as a result of seeing
how the station blackout diesel reliability, addressing it
in the station blackout rule and the maintenance rule, what
it could have been done back then is when we issued the
maintenance rule, when we issued one piece of regulatory
information maybe go back through to make sure everything is
consistent. It doesn't appear that there is a good process
to do that, but we are about to go through some major
changes to the regulatory documents to make them risk
performance based and to the extent that we do that we need
to be sure that we used terms, goals, criteria and
measurements consistent with it we are going to have to go
back, I think it would be wise to go back and make sure we
don't undo anything like the station blackout rule as we
revise on future revisions to the documents.
Your response?
MR. SIEBER: I have a question that is sort of a
summary question. When you look at expectations and cost,
you found that the actual risk reduction achieved was
slightly better than the expectation that the rule
envisioned when it was issued, but the cost -- NRC's
original estimate was $60 million and the actual cost to the
industry was $230 million, which is almost a factor of four.
MR. RAUGHLEY: $175 or $180 million to the 19
power supplies. We did not anticipate that at all.
MR. SIEBER: But nonetheless the SBO rule prompted
licensees to change those power supplies. Is that correct?
MR. RAUGHLEY: Yes.
MR. SIEBER: Okay. Does the utility industry
agree that the costs for the SBO rule were reasonable?
MR. RAUGHLEY: It is out for their comment.
MR. SIEBER: Okay.
MR. RAUGHLEY: You could debate that.
MR. SIEBER: I would prefer not to debate that
here, but I just wanted to know whether, since you state
that it was reasonable, if the people who spent the money
think it was reasonable too, and I guess the answer is you
won't know until somebody tells you.
MR. ROSENTHAL: But we specifically sent it to
NEI, EPRI, INPO, et cetera, asking for their comments.
DR. BONACA: The other thing is in the evaluation
we have to recognize some licensees chose to buy diesels
because they were useful for other purposes too. They could
have chosen to address the rule without purchasing that, so
in the cost benefit you will have to eliminate those made
their own choice, address it that way.
I know of some sites that had options --
MR. SIEBER: That might have happened in a couple
of cases but I remember our budgets didn't allow for those
kinds of things.
MR. RAUGHLEY: We're trying to look at things,
simple things we can do to make us better regulators in the
Lessons Learned, and belaboring sunk costs in the future is
kind of -- I don't know the relevance of that.
CHAIRMAN POWERS: Well, the only relevance that I
can think of for future would be is there something
inherently flawed in your estimation, cost estimation
process? The only thing that comes to mind is that there
are going to be ancillary thinking about what is
purchased -- (a) with the appearance of cost of a specific
rule up and then how you split those out if you are going to
do a cost benefit backward analysis on this is kind of a
difficult thing.
You say the guy could have bought a wagon and
instead he chose to buy a railroad car, for other reasons,
how much of that railroad car actually applies to the rule?
That is a tough one to figure out.
MR. SIEBER: I guess the only reason I bring it up
is I recall other cost benefit analysis that sort of had the
same result. NRC Staff sometimes comes up with a lower
number than the actual cost to the utility and so the
question is is your estimation process effective in some
ways?
DR. WALLIS: In the introduction Jack talked about
the goals of the Agency, one of which I believe is to
maintain safety. It seems here that you have actually
improved safety, which is what the Europeans prefer to use
as a goal, rather than maintaining safety.
In the present climate the move seems to be to
maintain safety and reduce the burden. Here you have
actually improved safety and you have actually increased the
burden, but you still have done a good job, so it just seems
to me that, I am just pointing out that it seems to me a
slight difference between taking literally the goals and the
actual implementation here.
CHAIRMAN POWERS: I think that here's one where a
risk analysis had clearly pointed to a vulnerability the
plants had and it passed the backfit argument. I mean it is
a real regulatory success basically.
DR. WALLIS: Oh, it is. It's fine, but some
lawyer might say, hey, you are only supposed to maintain
safety, not improve it, what are you doing?
CHAIRMAN POWERS: I think we would probably ask
that lawyer to go look at the Atomic Energy Act.
DR. WALLIS: I am in favor of that --
[Laughter.]
DR. BONACA: I have a question regarding this is a
draft document. Will the report include a recommendation
regarding cleaning up guidance documents like Reg Guides and
so on and so forth, or will it just be moot and say there
are opportunities there and let's face them.
MR. ROSENTHAL: Of course, we are receptive to
your input. The current plan would be to stop the report at
the conclusions and then to have a cover letter or
transmittal memo which would contain the recommendations and
it's not clear to me -- the current plan is that Ashok would
be sending it to in this case NRR. The Commissioners,
single Commissioners have expressed an interest in this
report as a prototypical document, so I am sure that we are
also going to be communicating with them.
We intend to have recommendations and intend to
follow up on the conclusions but I think that it will be in
the transmittal memos.
DR. BONACA: The next question I have, regarding
that, is that if you had to provide a recommendation for
providing consistent guidance to all licensees on how to
meet their reliability target or how to measure in order to
demonstrate meeting reliability, would you recommend to have
a new document do that or just simply going back and making
all the previous -- the question I am asking is because you
have all these Reg Guides and NUREG and rules with
inconsistent data.
I have no appreciation for what it takes to change
all these rules and that may be -- what would you recommend?
MR. ROSENTHAL: We haven't thought through the
mechanics. The goal would be to go to, and the thing that
makes most sense to us is to have reliability goals where we
have good definitions for reliability. It is consistent
with the way the Agency is going. We clean up the issue of
maintenance out of service in that reliability goal. You
know, RES is pushing to go from performance indicators that
look at unavailability, which is what we are doing now, to
looking at reliability.
If we could get agreement that what we ought to be
looking for is the reliability and assure that reliability
is consistent with what is in, presumed in the IPEs, then we
have to go back and clean up the documents.
How we would clean up the documents we have not
thought through.
DR. BONACA: The easiest way would be to have
white paper that says in meeting these targets this is --
and really that could quote the work done by E&L as a
means -- it would provide a standard for everybody to
follow. I believe the industry would welcome a standard
other than branching out and trying to figure out what they
have got to do and then coming up with numbers which we
claim are inconsistent with that.
MR. ROSENTHAL: But at a minimum Hal Lewis does
deserve at least a footnote that says, hey, he really was
right, and we --
DR. KRESS: Don't say that too loud.
MR. ROSENTHAL: -- and to change what we tell our
inspectors to look at.
DR. BONACA: And it is a problem a white paper
would not resolve, because still there is a guidance there
in the field that says --
MR. ROSENTHAL: Right, so we are arguing that we
ought to, that this is an opportunity to clean up the
regulatory process and make it more coherent and consistent.
DR. BONACA: Another question is is the issue
significant enough that it deserves that kind of action, and
I don't know.
I have two opinions. They are personal opinions.
One is even if quantitatively there may not be a
significant impact, qualitatively it is a problem to have
incoherent items in documentation. So I just wanted to get
a feeling from you and then from other members regarding
that issue, because then we have to make a decision here,
which is are we going to write a letter report on this, and
I would like to hear also from Mr. Rosenthal what you expect
from us.
MR. ROSENTHAL: We are conferring as we speak
here.
DR. BONACA: Okay.
MR. ROSENTHAL: Officially we came to provide it
for information to the ACRS so the choice would be yours.
My new Acting Division Director and I were
whispering at each other letters are always welcome or
useful.
[Laughter.]
DR. SEALE: And that is a sage hand that made that
comment.
Could I make one other point? In your initial
comments you made the observation that the station blackout
rule had very in many cases plant-specific impacts. That
is, all plants were not the same, that the influence of the
station blackout rule varied considerably from one plant to
another.
One of the concerns I have is that when you get
your response back to be specific to just make the point,
are you going to get a Florida Power response or are you
going to get a -- what is the other one? -- St. Lucie or are
you going to get a Turkey Point response?
The Turkey Point response based on what you said
is going to likely say some very specific things that might
not be said if you had a homogenized response. I hope it
will and it strikes me that it would be very worthwhile in
your review of the comments that you could somehow bring our
attention to those facts or those things which were
different from plant to plant and see whether or not the
comments you get reflect those differences because, you
know, if you make the room dark enough all cats are gray,
and I don't think that this is a homogenous issue for all
plants.
DR. UHRIG: It is also related to the era in which
they were built.
DR. SEALE: Of course.
DR. UHRIG: Turkey Point was built in an era --
DR. SEALE: The regulatory climate was very
different.
DR. UHRIG: Later on they would never been allowed
to have two diesels for two plants.
DR. SEALE: That's correct, and that is one of the
problems with the station blackout rule. That is, you have
to make this homogenous assessment which doesn't mean the
same thing for every plant.
MR. RAUGHLEY: In the Reg Guide that accompanies
the station blackout rule it has got eight tables, one
recognizing five different severe weather categories, so a
plant located on the East Coast would be driven to pay more
attention to this than a plant in Arizona because of the
severe weather conditions.
It recognized the different power system
configurations, where there are two lines, one line. You
know, how are you counting the offsite power supplies that
you have? Some people count the unit output.
DR. SEALE: On the other hand, the Secretary of
Energy has just come out this last week with a warning that
the plants, particularly in the Southwest and the Northeast,
are going to be particularly vulnerable to grid system
unreliabilities because of high heat loads this summer, and
that is not in there I don't think.
MR. SIEBER: On your last slide where you talked
about Lessons Learned, I think you have one there that you
have listed, but I think an important one was one that was
said right in the beginning, which is the analysis, the
regulatory analysis was simplified because the expectations
were explicitly and clearly stated when the rule was
promulgated.
I would think that a Lesson Learned would be in
new rulemakings to specifically identify and state what the
expectations are for that rule so that you can do an
analysis and determine after it has been in effect for
awhile whether you have achieved the expectation or not, and
that to me is another part of performance-basing.
DR. SEALE: Gets you out of the feel good
business.
MR. ROSENTHAL: I have to actually compliment
Bill. He has been a regulatory archeologist where you go to
the rule. If it is not in the rule you go to the statement
of considerations. If it isn't there you go back to a
supporting NUREG or the NUREG to the NUREG and then the
NUREG CR that backed that up, and always looking for
something that is public and documented, but it has been
quite a dig, and as he said earlier, ATWS is not as clean a
story.
I think we probably chose the cleanest one.
MR. SIEBER: I was just remarking that I think
that the fact that that is a Lesson Learned and it is
something we should be doing in the future.
DR. BONACA: Any other questions?
[No response.]
DR. BONACA: Before we adjourn, I would like to go
around the table, since we had no --
CHAIRMAN POWERS: Want to do that off the record?
DR. BONACA: All right.
CHAIRMAN POWERS: I'll take us off the record.
DR. BONACA: They I will pass it off to you, Mr.
Chairman.
CHAIRMAN POWERS: Any other comments people want
to make about this presentation?
[No response.]
CHAIRMAN POWERS: Fine. At this point we can stop
the transcription until this afternoon, and I want to go and
discuss the first two presentations we have had and receive
any comments the members would like to make as guidance to
authors or potential authors.
Thank you very much.
MR. ROSENTHAL: Thank you.
[Whereupon, at 11:30 a.m., the meeting was
recessed, to reconvene at 1:00 p.m., this same day.]. A F T E R N O O N S E S S I O N
[1:05 p.m.]
CHAIRMAN POWERS: Let's come back into session.
Members have before them, a list of potential issues that
might be associated with review of an AP-1000 design for
certification.
I hope you will take the chance to look this over,
make any comments on the list, and get it back to us. We
are obligated to provide the staff with some indication of
the kind of topics that we would want to look into if this
certification does come to pass.
And I'd like to get that letter over to the Staff
this week.
Okay, at this point we'll turn to the next item of
business, which is the Regulatory Guide and whatnot
associated with the revised source term rule. And, Dr.
Kress, I guess this, too, is one of your areas of expertise.
DR. KRESS: Yes. You will recall that previously
the Staff worked on a rule to allow the voluntary use of
alternative source terms based on all the information we
knew, and the NUREG 1465 source terms that were developed as
a result of this new information.
We wrote a letter -- I forget the timing on that,
but it wasn't too long ago -- commending the Staff on what a
good job they are doing in determining all the ramifications
of the use of the new source terms, and what risk
implications there might be and things.
Well, they now have issued the rule for public
comment, and they have also developed a Reg Guide and a
Standard Review Plan to go along with the rule, and have
issued it for public comments and received the comments
back.
And we saw a draft of that, but we didn't review
it, I don't think. I don't recall whether we reviewed the
earlier draft or not. But at any rate, what we have before
us is the final version, basically, of the Reg Guide and the
Standard Review Plan.
Before they go up to the Commission and say we
want to issue this in final now, they'd like our comments
and perhaps a letter as to what we think about it.
They did make some changes over the draft that we
may have seen, and some of these changes are noted in your
package. They have to do with gap fractions, chemical forms
of fission products. There is a list of them here that you
can read. There is no use in my going over them, because I
think the Staff intends to pretty much highlight what these
changes were over what we've seen before.
So, with that as a really quick introduction, I'll
just turn it over to the Staff.
MR. LaVIE: Good afternoon. I'm Steve LaVie with
the NRR, Probabilistic Safety Analysis Branch, which, as you
may be aware, also picked up the dose assessment people in a
recent reorganization.
As Dr. Kress pointed out, I'm going to present to
you, the changes that have occurred in the guidance
documents we prepared for the implementation of the
alternate source term.
Dr. Kress gave you a little bit of the background,
and let me touch a couple of the points again. You may
recall that we've been before the Committee several times.
Back in 1998, we prepared a rulemaking plan which you folks
provided us comments on.
And we also at that time presented the results of
the baselining study. As a result of that rulemaking plan,
what we recommended to the Commission and that the
Commission endorsed, was to go prepare a rule change and
also a new Regulatory Guide, and, of course, a Standard
Review Plan section to support that Regulatory Guide.
This is the path we've taken. We haven't deviated
from that too far. In March of 1999, we published the draft
rule in the Federal Register for public comment.
In December of 199, we published the final rule in
the Federal Register. That rule became effective in January
of 2000.
At that time we included an announcement of a
public comment on the draft guide. We did provide the draft
guide as part of the final rule package to the Committee.
That would have been in the September/October timeframe in
1999.
As I pointed out, the final rule became effective
on the 24th of January, and the public comment period that
ran for 75 days, ended at the end of March.
We received numerous comments, six official
letters from the Nuclear Energy Institute, the Nuclear
Energy Environmental Qualification Group, from Duke Energy,
Virginia Power, South Texas Project Nuclear Operating
Company, and the Florida Power Corporation.
We also received numerous informal comments by e-
mail and other approaches. People came up to us and so
forth and made a comment, and we attempted to address all of
these.
We also addressed the ACRS recommendations that
you provided to us in the October 1999 letter. There were a
total of 138 comments, several of which were redundant, but
nevertheless, we had some things to look at.
Now, we dispositioned all those comments we
received. I provided in the package, and you should have,
the disposition of the comments we did receive.
What I'm going to discuss this afternoon is the
significant changes that we made that would change the
technical content and a couple of policy items.
These are categorized, basically, into the areas
of fuel gap fraction; fuel handing accident chemical form
for the release; selective implementation; 50.59 guidance.
As you may be aware, 50.59 guidance was being prepared in
parallel with this effort. We have had to go back and
adjust a little bit for what's proceeded in that process.
And there were several other smaller technical
changes which we'll go over quickly.
Perhaps the most significant change we had, had to
do with the fuel gap fractions. In the draft guide, we had,
in essence, stayed with the traditional values we had been
using in licensing for several years.
The industry provided us with several comments on
this, both formally and informally, to the effect that we
needed to do something different.
And the industry suggestions can be summarized as
that they suggested that we have gap fractions that would
vary from three percent at 50 gigawatt days per metric ton
uranium, to 9.3 percent at 75 gigawatt days per metric ton
of uranium.
DR. KRESS: Is that based on the experimental
data?
MR. LaVIE: It was based on some experimental
data, results of fuel sipping on fuel that has been
irradiated in plants to date. It was largely based on an
EPRI report, that EPRI had looked at this issue.
The industry had extrapolated that data to make
the recommendation.
DR. KRESS: Extrapolated it to 75?
MR. LaVIE: Yes. They had no experimental data
above 65.
DR. KRESS: That's interesting that you end up
with 9.3, which is four significant figures because it's a
percent.
MR. LaVIE: Right.
DR. KRESS: For something that's extrapolated.
MR. LaVIE: There is a large amount of uncertainty
here, yes.
They also suggested that we allow the licensee to
vary the gap fraction across the core. This hadn't been
something that we had given a whole lot of thought to
before, but we were able to come to some agreement on that
for some accidents.
And they also suggested that we address the fuel
heatup impact separately. This would largely address
accidents such as the reactivity insertion accident.
Now, when the Staff looked at this comment -- of
course, we had to consider it even before we published the
draft -- is, we just felt there was insufficient data to
support iodine gap fractions above 65 gigawatt day metric
ton uranium. There was insufficient data above that point.
Now, the industry data they presented to us was
largely the result of low burnup data that had been
collected over fuel that's been burned to date. The
majority of the data was much less than 50,000 and very few
points above.
There are large amounts of uncertainty in these
gap fractions, as you may be aware of. The data they
presented was based on actual fuel, and as such, they did
not have the operational transients that could occur during
operations and still be within the fuel limits.
The current fuel management that's being used
today in power plants is far more aggressive than that under
which the data was collected. A lot of this data was
historical and, say, maybe five-ten years old. We have much
more aggressive burnup regimes right now, so there was
uncertainty involved with that.
The biggest problem we had is, in the majority of
experiments that had been done to date, iodine has not been
measured directly. Iodine has always been inferred from
some other measurement.
DR. KRESS: You don't get a gamma.
MR. LaVIE: Right. It's been done through
thermodynamic correlations, ratio of the diffusivity of
noble gases to iodine, which is the basis of the ANS 5.4
method, with a great deal of uncertainty.
So, we went into this comment with the idea of
what can we do about this? While we did see all these
uncertainties, we also had this feeling that based on some
of the data we were looking at, that perhaps we were high on
gap fraction.
DR. KRESS: The gap fractions you had been using,
taking actual fuel and cutting it open --
MR. LaVIE: Right, back in 1960 or sometime. They
were first documented in Reg Guide 125 for fuel handling
accidents, based largely on work done by Westinghouse.
So we needed to come up with an approach for this.
Recognizing this is a deterministic design basis proceeding
with this Regulatory Guide, we decided we would use the
NUREG 1465 data for the LOCA, for the gap fraction part of
the LOCA, gap phase, and we would allow that to be used for
ranges zero to 62 gigawatt day per metric ton uranium.
DR. KRESS: Let me -- associated with that, when I
read the Reg Guide, you also decided for the gap fraction,
that you would use the NUREG 1465 speciation for iodine.
MR. LaVIE: That's correct.
DR. KRESS: That strikes me as a little strange.
You're saying that in the gap, the iodine consists of 95
something percent of cesium iodide, a certain fraction of
elemental iodine, and a certain fraction of oriatic iodine.
But the NUREG 1465 values came out of chemical
effects after the stuff got out of the fuel, and it
certainly did not imply that gap had these, and, in fact,
all the data I know of about gap fractions chose the cesium
iodide as primarily -- I mean, the iodine is primarily
cesium iodide.
MR. LaVIE: That's correct.
DR. KRESS: And so I don't think it makes much
difference in what you apply or use the gap fractions for,
and it doesn't change the gap fraction; it changes the
speciation.
MR. LaVIE: Right.
DR. KRESS: And it's such a small amount of
iodine, I don't think it makes much difference, but it
seemed a little strange and inconsistent with the technical
basis.
MR. LaVIE: True. I don't disagree. For the
LOCA, as you point out, it's really not a large consequence,
because you're looking at five percent of the total core
inventory, which is immediately going to be followed by the
other 95 percent of it.
So it's not going to make a lot of difference for
the LOCA. Now, when we went to the other accidents, there
was no data for what was in the gap.
However, we looked at the NUREG -- I want to say
737, but that's not correct. NUREG 0772 had the data that
showed that what was in the gap was predominantly cesium
iodide.
DR. KRESS: That's the only evidence.
MR. LaVIE: So we felt, you know, throwing in
elemental made it a little bit more conservative, because we
are in deterministic space here.
DR. KRESS: Yes, it certainly wouldn't hurt.
MR. LaVIE: It wouldn't hurt, and then the
conversion gets us the organic. We used the same conversion
ratio that was specified in 1465. We believe that's
conservative and reasonable for this deterministic approach.
DR. KRESS: You're just saying that once you use
this fraction for whatever you're using it for, you probably
ought to be conservative and assume some of it got to be --
so it doesn't --
MR. LaVIE: Our intent was to be conservative,
because we really -- there is -- all the research and data
has been done on the LOCAs. Now, the risk-informing --
DR. KRESS: Now, the real problem is that those
values for the LOCA, for the early in-vessel release, come
out of very specific considerations.
MR. LaVIE: Right.
DR. KRESS: And what they were was, you look at
the whole spectrum of accident sequences, and what happens
is, you release the cesium and the iodine and the chemical
reactions in the gas phase create cesium iodide, but part of
the -- during part of the sequence, you're steam-starved and
you're hydrogen-rich, and during that part, you get a
certain amount of hydrogen iodide, which we called something
different than a cesium iodide.
And then there was a certain amount of organic
iodine that always gets produced because there is some
organics in there, too.
MR. LaVIE: Right.
DR. KRESS: And these were the three values, the
maximums in any of the sequences, and they used them in the
NUREG 1465 as bounding values. And it all happened because
the chemistry effects in the downstream of the core after it
got released, none of those chemical effects are going to
happen when the gap gets released.
The concentrations are different, the temperatures
are different, and they're always steam-rich when the gap is
there. So there's no reason to expect the NUREG 1465 in-
vessel release speciation to be the same in the gap.
And I would have gone -- like you say, it's
conservative and doesn't make any difference, probably, in
what you do, but I would, just for consistency's sake, not
had that kind of -- you know, it just looks funny in Reg
Guide to have something that doesn't have any technical
basis at all.
And so that's my concern. My concern is not that
it's there and will cause anybody any problem, because I
don't think it will, but it's just a coherence, consistency
type of problem.
MR. LaVIE: Right. Am interpreting what you are
saying correctly, that you would prefer to have seen us
specify that it was all cesium iodide?
DR. KRESS: I would have just kept it the way it
was in the old Reg Guide, frankly, for the gap. You had a
speciation in there, and it was 99 --
MR. LaVIE: 99 percent elemental, I think.
DR. KRESS: I'm sorry, I would have just called it
cesium iodide.
MR. LaVIE: Okay, see, in the original draft, we
called it elemental, 99.75 percent elemental.
DR. KRESS: I wouldn't have done that, because the
evidence is that it's cesium iodide in the gap, and I would
use that.
MR. LaVIE: And we thought this was a reasonably
conservative compromise with considering the deterministic
design basis approach we're dealing with here and would be
appropriate.
DR. KRESS: I would agree that it is, but it just
looks funny when you use an inconsistent technical basis to
establish your conservatism. But I agree that it's not a
big deal.
DR. LAVIE: We continued with the protocol we had
in the draft guide, using the Reg. Guide 1.77 data for
reactivity insertion accidents. As you're all aware,
there's a great deal of work going on right now about the
reactivity insertion accidents because of the Cabris results
and the lack of beta data to change. We're holding the
status quo on the reactivity insertion accidents, so they'll
continue to use the gap fractions from the old data.
Now, we had some work done by PNNL to address the
environmental impact of the fuel burn-up from 60 to 62
gigawatt days per metric ton uranium. And that data became
available during this period. It is being documented as an
update to the NUREG CR-5009.
Now the PNNL analyses were done for core average
and the peak rod average at 35, 60 and 65 gigawatt days
metric ton uranium. The analyses were done using the
FRAPCON-3 code, with using the Missah release model. It was
a best-estimate approach. There were no operational
transients addressed.
DR. POWERS: Do I understand what the Missah
release model is? Do I know what the Missah release model
is?
DR. LAVIE: The FRAPCON model, the FRAPCON code
allows the user to choose, I think it's three different
release models --
DR. POWERS: What are those from?
DR. LAVIE: It has the traditional ANS 5.7
approach -- excuse me, 5.4 approach, which our contractor
believes is extremely over-conservative. And based on his
considerations, he decided to this of the three options he
had available.
He also calculated the core inventories up to 75
megawatt day metric-ton uranium, although they won't be
useful to us because we don't have the gap fraction data
that high. The staff decided to use the PNNL data with some
adjustments. In our approach, the bottom line here was to
balance the uncertainty in the gap fractions with other
analysis conservatisms. For example, in the fuel handling
accident, we always considered the rod to be damages to be
the peak burn-up rod, and it's also in the peak power
position. With fuel management the way it's performed, this
is an impossibility. The rod, the high burn-up rods would
not be anywhere near peak-power position.
So what we ended up putting into the draft
guidance -- and for comparison, I put the previous numbers
from draft guide 1081 in there. For the local, we decided
to stick with the .005 for the gap fraction and use that all
the way up to 62 gigawatt days metric-ton uranium. This was
consistent with some of the PNNL data that showed this would
be appropriate. If you'll recall, 1465 did have some
language that if you had long-term cooling, you could use
.03. However, 1465 also said that the data was potentially
suspect above 45 gigawatt day metric-ton uranium. So
combined with that information and what we had from PNNL, we
decided to settle on .005 for the entire range zero to 62.
Now for the norm locus, fuel handling lock rotor
accident, we decided to split it into two regions. For that
fuel which is lower than 40 gigawatt days metric-ton
uranium, those would be the fractions we would suggest to
use. It would be acceptable to the staff. And above 40,000
but no higher than 62, we would use the fractions shown
there.
For the reactivity insertion accidents, as I
mentioned before, we stayed with the old data until we have
a basis for changing it. Now --
MR. KRESS: Is the expectation that when they
look at these other accidents, that they will make an
assessment of what burn-up --
DR. LAVIE: I'll be getting to that --
MR. KRESS: -- is -- oh, you're gonna get to
that.
DR. LAVIE: I'm gonna get to that. One step --
MR. KRESS: Okay.
DR. LAVIE: What we are putting in the guidance
is for the default approach, acceptable to the staff, is
that the gap fractions associated with the peak burn-up, rod
burn-up, in the core would be used with the rod inventory
adjusted for the maximum radial peaking factor. So they've
run the origin code, they've come up with a core inventory;
they multiply for the maximum peaking factor for the entire
core, out of the COLA. So this'll change with each core
upload. And that'll give them the inventory in the entire
rod. They then will pick up the maximum burn-up in the
core, okay, and use that for the gap fraction. So the
fraction times the inventory gives them the release.
Now that's -- for the fuel-handling accident, we
are insisting on that approach because you really don't know
which element you're gonna drop in the fuel-handling
accident. Recently, at one time reactors used to shuffle
one-third of the core, and that way only a third of the core
moved. However, nowadays, for various reasons, we are
seeing a lot more full core offloads. So there's an equal
probability that any element could be dropped. So for the
fuel-handling accident, we're retaining the conservatism
that they must use the maximum COLA and the maximum burn-up.
Now, for the other accidents that get fuel damage
in some designs -- a main steam line breaks, lock rotor
accidents, steam tube ruptures -- if the licensee can
demonstrate to us with reasonable certainty that he knows
where the damaged fuel is in the core, then we will allow
him to use the gap fraction appropriate for that element
along with the radial peaking factor for that element.
However, for maintaining some degree of conservatism because
of the uncertainty of the gap fractions, we are not letting
them use a radial peaking factor less than one.
In some of these three-burn cores, the radial
peaking factor can get down to .6 and .7 in that third
region. However, to maintain a degree of conservatism, we
will not let them go less than one. Now this is totally an
optional method. It will not apply to the fuel-handling
accident.
I need to point out that very late in the process,
after this document was distributed for review is -- in the
last week, as a matter of fact -- we have gotten the
technical comment on this table. And the NRR is working to
resolve this technical comment. We expect to be able to
resolve it and retain what we have here. If we find out
that we cannot retain what we have here and we have
significant changes, I will arrange to come back to the
Committee and explain why we're changing it. But this is
what we expect to publish.
This is the iodine spiking species, which we've
jumped ahead a little bit. For the fuel-handling accident
we had previously specified 99.75 percent elemental and .25
percent organic -- recognizably conservative. The industry
suggests that we use the NUREG 1465 species. Now the staff,
when they first got this comment, gave some thought to this
because we do have a very, very low pH in the spent fuel
pool, typically 4 or 5. And if you'll recall, 1465, with
the iodine species specified there, if you maintain the
sump pH grade of the 7.
We also had some concern about transport data.
The transport data we have for pool DF came out of that
original Westinghouse work back in 1959. Cold water -- now
there's been an awful lot of work done on transport through
pools, but it's typically not the, the stagnant pool such as
a spent-fuel pool.
So for the final guide, we decided that we would
adopt the fractions in 1465. However, from a release
standpoint, to the environment, we decided that the cesium
iodide completely dissociates in the pool water and that
because of the pH, it will re-evolve as elemental iodine.
Now, yes, this is a mechanistic process that would over
time. However, for the purposes of deterministic
calculation, we have taken this assumption. We are using a
pool DF of 200, effective because it varies for the
different isotopes, different species. And with a pool DF
of 200, the release from the pool ends up being 57 percent
elemental and 43 percent organic.
MR. KRESS: Now the 200 comes out of the DF for
elemental iodine -- that's pretty high.
DR. LAVIE: And organic.
MR. KRESS: Unless it goes into solution.
DR. LAVIE: Right.
MR. KRESS: And then it has to dissipate. And
the organic is essentially not captured.
DR. LAVIE: Not captured at all. The organic is
what controls. The organic is what controls -- this is why
the organic jumps so high.
MR. KRESS: So basically, everything that comes
out -- well, it's about half and half when it gets there.
DR. LAVIE: Right.
MR. KRESS: But it's --
DR. LAVIE: We enter it here --
MR. KRESS: -- organic is so small that what
you're doing is just reducing the elemental down --
DR. LAVIE: Right, and the organic becomes much
more controlling.
MR. KRESS: Yes.
DR. LAVIE: So from an analysis standpoint to the
environment, effectively we're back to what we had in the
original. Now we do point out in the guide, however, if
the applicant, or licensee wants to come in with a
justifiable mechanistic treatment for pool treatment that
looks actually at actual iodine species and so forth, it
will be considered on a case-by-case basis. But the default
deterministic method will, is what's hear.
MR. KRESS: The problem I have with that is -- I
like to do a concept of the cesium iodide and go into the
pool water with elemental iodine. It won't go in as the
iodine, but you know, if it's gonna get released, it'd be
released as the iodine.
DR. LAVIE: Right.
MR. KRESS: But the problem I have with this is
the assumption that, that the gap has .15 percent organic.
/v
DR. LAVIE: Right.
MR. KRESS: There's no basis for that choice at
all. It might very well be .25, as is in the original DT
1081. I don't know what it is inside the gap, but there is
no evidence for something like .15 percent organic in the
gap at all. And to put it in here and then say, now you've
got 43 percent of what's released is organic is a little
strange to me. It's such a, such a small amount of stuff,
all it does is govern things like the required closure time
--
DR. LAVIE: Filters.
MR. KRESS: -- mostly. It doesn't have real risk
implications as much; it has some exposure and it affects
things you do. So I'm not really concerned. But here you
have a whole set of deterministic regulations. They're
basically based on this .15 percent organic, which has no
basis at all. You know, it just shows up. And that's the
kind of thing that bothers me.
DR. LAVIE: Okay.
MR. KRESS: You know, it's not a big deal, but it
just bothers me to have regulations like that.
DR. LAVIE: The um, the Westinghouse data that
was used as the basis of the original did point out that
there was organic due to manufacturing processes.
MR. KRESS: Manufacturing --
DR. LAVIE: -- and, so that -- but they projected
.25 percent.
MR. KRESS: Yeah, and I think that's a better
number to use because it's the only data we have.
DR. POWERS: In this stage of the calculation,
you're really hypothesizing, things go into the pool and
then come back out.
MR. KRESS: Yeah, and in that case --
DR. LAVIE: Okay, that's a good point, because I
want to point out that the way these analyses are currently
done, it's considered to be an instantaneous release from
the pool. Now if somebody wants to come back and
mechanistically -- one of the things we have to stop
worrying about if we start to say that some of this is gonna
be captured in the pool and released over time is now we
have to start looking, now the market changes the way this
is modeled, to pick up the continuous release. Right now,
we do a puff release and it's gone.
This is a simplification, granted. We think it's
conservative. We think it's conservative. But if they want
to come in with a mechanistic treatment, it will be
considered.
DR. POWERS: Well, what I was going to point out
is that we do have the results of the Febus experiments.
They're a little different; they're looking at core
degradation. But they have the iodine coming into the
containment model, closing to the sump, and then they get a
repartitioning out -- their consensus is for round-number
purposes, it's about 50/50 elemental-organic, maybe actually
a little higher in organic than elemental at various stages.
MR. KRESS: And that's because the iodine
converts to organic in the pool, and they know that here.
DR. POWERS: That's right. And I'm wondering if
-- the first line under final guide may well be wrong, but
the third line ends up being right, perhaps for the wrong
reasons.
MR. KRESS: It's probably right, but it will have
the wrong quantities in it because of the DFs that are --
DR. POWERS: Right.
MR. KRESS: It'll have much lower quantities, but
once again, they got the conservatism that they're assuming
it comes in instantaneously, when actually it takes a
considerable amount of time for this stuff to come out.
DR. LAVIE: So one of the things we recognize
with this particular calculation, as you pointed out, the
risk basis of this is maybe, as the risk-informed Part 50
process continues, this may be an accident we'll stop
worrying about. So this particular accident is not worth an
awful lot of modeling --
MR. KRESS: It's not, it's not worth a lot,
that's right.
[LAUGHTER]
MR. KRESS: The other problem is that the DF
factors are, are a little strange because they're based on,
they're based on suppression pool data, where the gas that
carries this stuff is a steam, it's a condensing gas.
DR. LAVIE: Okay --
MR. KRESS: And what you have here is, you're
gonna have cold, relatively cool nova gases that carry up
these things, and I don't think you'll get the same DF.
DR. LAVIE: Okay, the DFs we used here, Dr.
Kress, we came out of the Westinghouse work back in 1959.
Westinghouse did a series of experiments, small-scale
experiments where they entrained iodine in a carbon dioxide
carrier --
MR. KRESS: Oh, it was a carbon dioxide -- okay.
DR. LAVIE: Okay, and it was cooled, the water
cooled and the pH was controlled in order to do that. They
then took that data, recorded it, graphed it, what have you.
They then went and got a 14x14 assembly, sheered it off,
connected a gas volume below it, put it at the bottom of a
23-foot pool. But they didn't use the iodine; they used a
carbon dioxide, they just used the carbon dioxide carrier.
Okay, and then they measured the rate and the rise of the
bubbles and how much bubbles actually had rose to the
surface.
MR. KRESS: Okay, well that sounds like --
DR. LAVIE: And then they then correlated it --
MR. KRESS: -- pretty good chance of being
appropriate.
DR. LAVIE: The staff took the Westinghouse data
and then massaged it a little bit further, because the staff
at that time was concerned that the Westinghouse model
didn't match some formulas the staff already had on mass
transfer, even though Westinghouse had actually measured it.
Okay, the staff tried to backfit a formula into the data.
So the Westinghouse originally came up with DFs for
elemental iodine as high as 800, and then the staff, because
of its manipulations, pushed it down to 100. We have
brought it back somewhat.
MR. KRESS: Well, I tell you why I'm still in
this discussion with this relatively minor thing, and that
is that one of the purposes for redoing the source term
itself was to get a little more realism into it. Now we've
introduced completely unrealistic --
DR. POWERS: All this conservatism here.
MR. KRESS: Yeah, parts -- completely
unrealistic, no basis parts for part of the thing that has
little relevance or little impact in terms of things, but
it's still, it's still going away from the intent of the new
source term to put a little realism into it. So that's,
that's all that's bothering me. It's not that I'm concerned
about this point here or any kind of safety impact it'd
have.
DR. LAVIE: We did recognize in conservatism is
this is a decrease by a factor of two. Plants typically
come in -- the limit for this particular accident is 75 rem
thyroid, and they typically come in anywhere from 25 to 30.
I don't think of any plant that's had a fuel-handling
accident be eliminated. This accident's primarily done to
ensure that the filters and systems involved with the spent
fuel handling areas are adequate.
MR. KRESS: I guess the question is will this
serve any purpose? I guess it will. I'll have to think
about it.
DR. LAVIE: We'll consider your suggestion about
sticking with the 100 percent cesium iodide --
MR. KRESS: You're gonna get basically the same -
-
DR. LAVIE: Right.
MR. KRESS: -- when you do that.
DR. POWERS: I guess I wonder -- you know, if I'm
sitting around, trying to figure out how I'm going to
respond to an accident, and you tell me, okay, you've got a
puff release and everything's over. And I pick up one set
of actions. If instead you're telling me I've got a
protracted release --
MR. KRESS: Over a long period of time.
DR. POWERS: -- over days and days and days, I
think I'd come up with a different set of actions.
MR. KRESS: Yeah, and that's the other thing that
bothers me. Part of the source term specification is the
time. And here we've gone back to the puff release, just
for convenience, when we know it's not a puff release, but
we're saying it's conservative and I'm not sure it is,
because you, you have one set of actions versus another and
I'm not sure which is the right things.
DR. LAVIE: I can try to address that. Is that -
- when we model this accident, was assume partially at the
release of the gap activity that has collectively up to the
point where the fuel was removed from the core, so what
we're releasing is in essence the gas that's in the rod;
it's gonna come out. The rod is at 800, 1600 pounds of
pressure, depending on its burn-up, and even at the 23-feet
pool depth, you're only looking at about 30 pounds of
pressure, so the gas release from the fuel element will be
very, very rapid.
If the filter systems in the fuel handling
building are capable of handling a puff release, they're
capable of handling a protracted release.
MR. KRESS: Well, your puff release is gonna be,
the amount of iodine that's in it, it's gonna be about one
percent of the total that's release from the gap. The rest
of it ends up in the pool, and what we're concerned about is
that other 99 percent is not dealt with here at all, because
it's gonna start coming out also.
DR. LAVIE: That's correct.
MR. KRESS: And it should be dealt with in terms
of fuel handling accidents somewhat, and it may come out for
a long period of time and it may be a lot more than the --
this thing may be designed only to handle at one percent,
and here you've got 99 percent more of it coming out. But I
think that's the concern.
DR. LAVIE: That's the basis of our second
bullet. We're assuming that it completely dissociates --
MR. KRESS: Yeah, we're saying it will re-evolve.
DR. LAVIE: Instantaneously. Of course it won't,
but we're modeling as it does, so we are capturing that
release.
MR. KRESS: You're already adding that in as a
puff release.
DR. LAVIE: Right.
MR. KRESS: Okay, so we're saying if it's
designed to handle that as a puff release --
DR. LAVIE: Right.
MR. KRESS: But it may not.
DR. LAVIE: But if they want to come back and
look at it more deterministically, more mechanistically,
then we certainly will consider it.
MR. KRESS: Yeah, it comes out in more protracted
time versus a puff release -- with a puff release, you've
got competition between where that iodine goes. Does it all
go into the -- if you're assumption is it all has to go
through whatever the clean-up system is, then you may be
right. But if the assumption is that that cesium iodide
leaks out the containment, goes through some other sort of
chemical reaction with the sprays or whatever, gets removed
-- but you may be right. I guess I missed that statement
through. You are using all of the cesium --
DR. LAVIE: Right. Right, yeah. Most licensees
when they do this particular calculation, do this with a
spreadsheet. You know, a times b times c. It's really --
most of them do not do it on a time-dependent basis. It's
the total quantity released. They get a total curies
released; they convert that right to rem. Very few people
actually model this in any degree of --
MR. KRESS: I think I may have been tempted to go
back and say, you know, the DG 1081 says you get 99 percent
elemental, .75 and .25 organic, and it all goes into the
containment instantaneously and deal with it, because you
end up at the same place.
DR. LAVIE: Exactly the same place, but for a
licensee who wants to do it mechanistically --
MR. KRESS: You can still say, all right, if you
don't agree with this tell me, justify some other. So I
think I'd, rather than have these arbitrary things in there
that really, really raise questions -- particular the first
line raising this question -- I would have just gone back to
the DG 1081 and said it all goes into the containment
immediately.
DR. LAVIE: Okay.
MR. KRESS: Or, as an alternative, I would have I
would have said, .25 percent of it goes in immediately and
only a DF of 500 of the element goes in immediately, and the
rest of it comes out protracted over time. And I would have
made a calculation for what that protracted time release is.
But that's not an easy calculation to make.
DR. LAVIE: No. It may be very plant-specific.
MR. KRESS: It may be plant-specific, but you
could, you could deal with the plant-specific issues in a
general way.
DR. LAVIE: Actually we're hoping this one goes
away when we risk-inform Part 50. We may be down to just a
loca.
On the second limitation, just to refresh you, as
you recall from our original discussions last fall, is we
have two different ways a licensee can get into the
alternate source term is a full limitation, in which they
come in and do it as a minimum of the full-fledged, full-
blown loca analysis, and then we would grant them a broad
scope approval that this is now in your design basis; you
can then use the alternate source in TEDE for all future
radiological analyses.
Realizing that not all licensees may want to go
that route, and with deference to the Commission direction,
we also provide a means to do it selectively, that they
could pick small little applications and apply that small
application. When we wrote the draft guidance, it was our
intent that, those selective limitation licensees, if they
want to use the alternate source term for some other
application that they had to come back and talk to us. This
was one of the benefits of going to the full.
We got several comments on that, and when we
started looking back at it, we had realized that we had
probably gone a step too far. And what we finally decided
to do and change the guidance to is, recognizing that when
we gave them the approval for the selective implementation,
we approved some characteristic of the alternative source
term and the use of TEDE if they did something that required
a dose calculation. And we put that in the design basis.
Now once that's in the design basis, that's part of their
design basis, and we really don't need to worry about that
again. So if the licensee wants to make a subsequent
modification using those characteristics that we have
already approved, they can go ahead and do it assuming they
can pass the 50.59 criteria.
However, the staff review will be required under
50.67, the regulation for the alternate source term, if they
decide to use an alternate source term criteria
characteristic, or the dose criteria, which is not already
in their design basis. So a licensee who may come in and do
a timing-only application could not subsequently go off
without coming back to us and do a calculation that involved
the other characteristics. Or, if the licensee decided he
had new data on gap fractions and wanted to change his gap
fraction, he could not do that without staff approval
because that's what's currently required under 50.67.
The revised position here is consistent with the
50.59 guidance and 50.67. So it would give them a little
bit more flexibility than what we had previously.
MR. KRESS: I particularly thought this was a
very insightful part of your change. And it addressed the
earlier question, one of the earlier questions we had --
DR. LAVIE: Right.
MR. KRESS: -- and I thought this was very good
way of handling it.
DR. LAVIE: I believe this was always our intent,
but when we wrote the language it didn't come across that
way.
MR. KRESS: It just never came true that way,
yeah.
DR. LAVIE: One of the things that came up -- out
of all the 134 comments we received, and some of them were
very, very good comments, this one was one of the ones that
really blew us away because it was one of the last sets of
comments that came in, and I think this was one of the best
ones we got. The individual pointed out, under the
alternative source term and the way that we wrote the draft
guidance is that it's possible for a licensee to have
analyses on his books that are not based on TEDE and are not
based on the alternative source term. If he could show that
they were bounding, he could let them stand. But our
guidance said that in the future, if you have any reason to
resolve those calculations for any reason, use the, what's
now in your design basis, the alternate source term and
TEDE.
This gave a problem with a guide to 50.59 because
as you call the new 50.59 guide in determining what's a
minimal increase in consequences has you comparing prior to
after. Okay, so we had all these calculations out there on
whole-body and thyroid, and the new doses were going to be
in TEDE.
Working with Alan McKenna and the other folks on
50.59, we decided the approach to do is that for this
particular regulatory guide, we would put in an equation and
some guidance on how to convert that prior value, and it's
nothing magic to this. This is the waiting factor for the
thyroid, which goes into constituting TEDE. So what the
guide will say is that if you have one of these situations,
before making the 50.59 comparison, convert your result. It
was a very good catch.
MR. KRESS: It was a very good guidance.
DR. LAVIE: We didn't want to go into too much
detail in here because we prefer that people go to the 50.59
reg. guide and the industry document that it endorses,
rather than going into all sorts of detail, what constitutes
minimum and all that stuff. So all we did here is that,
when you get ready to do the comparison, this is how to get
the prior result.
There was a lot of other technical changes, not as
major as the ones we've just gone over. With the guide to
EQ, as you recall, we have a generic safety issue in
progress resolving the cesium in the sump water. This has
not yet been resolved, so because we're going final in the
guide, we have added text to the guide that talks about the
GSI and also specifies that until the GSI's resolved, the
licensees are allowed to use TID 1484 or the AST.
MR. KRESS: When that GSI gets revolved, will you
go back and redo the guide and the review plan, or just --
DR. LAVIE: We expect that we'll be revising this
guide in probably the next two-year time frame because of
what's going on in risk-informing Part 50, and at that time.
DR. SEIBER: I see. But you'll have the
opportunity to pick it up.
DR. LAVIE: Of course, as guidance, the licensee
could point out to us in their submittal that the GSI had
now required them to do something different or said that
they could use the TID 1484 all the time. This is only for
re-analysis required under the guide.
It was pointed out that the appendix I we had put
in, though it discussed all the good guidance on the EQ
doses inside containment, we really silent with regard to
doses outside containment, so there was some general
guidance added there, on that topic.
The steam generator iodine transport -- we
corrected an error regarding the decontamination credit when
the tubes are uncovered. Although we had described very
clearly the model for flashing and non-flashing and
scrubbing, we negated all of it in the final paragraph that
says, if your tubes are uncovered, you get no partitioning
credit. That was technically wrong. That's been corrected.
They now go back to the flashing fraction. If it's, if it
is not flashed and stayed in the bulk water, then the
partitioning would apply regardless of where the water was.
However, for the fraction that did flash, then they would
have to adjust the scrubbing fraction. So we corrected that
error.
One the spray DF, although we would expect a lot
of the licensees to use the rad trap models, which don't
have a spray DF limitation per se, recognize that some
people will stick to the models in the standard review plan,
for which there is a spray DF maximum. We had put in some
guidance there as to when you take the total volume, because
recognizing the new source term enters the activity over
time instead of all the time equals zero. However, we made
a mistake when we put that in, and we have since revised
that to correct the guidance.
We were asked to consider allowing building mixing
credit for the fuel handling accident. This is something we
have typically not done, but we decided that if a licensee
can justify it, it ought to be allowed.
MR. KRESS: I'm not quite sure I understand that.
DR. LAVIE: Okay. In the fuel handling accident,
traditionally we have assumed that that puff of gas that
leaves the fuel goes straight into the ventilation plenum,
no mixing in the building. And in most cases, that's the
way it's going to happen. Now there are some designs,
however, where there may be a potential for mixing, and if
they can justify it, it'll be allowed.
MR. KRESS: So you have a mix-in and it goes into
the ventilation, but it does it --
DR. LAVIE: But a, much delayed period of time.
DR. SEIBER: But in the aggregate, it's the same as
a puff release.
DR. LAVIE: Exactly. Yeah, what comes out of the
water is still a puff, but it mixes in the building and can
be released over the next two hours.
DR. SEIBER: But it's still considered puff.
DR. LAVIE: Right. They get some dilution. The
concentration's a little lower.
We have also decided -- there was a requirement
that was added on the loss-of-coolant accident for the ECCS
system leakage outside the containment. There was a
requirement added several years ago, which was largely
intended to force the installation of filters, that if you
did not have an engineered safeguard filter system in those
areas, then you analyze a 50-gallon-per-minute leak. That
requirement has been deleted.
Dr. Kress pointed out, the ACRS gave us a letter
last October. There were three recommendations. Two of the
recommendations we had, part of our response to table action
until we got to the guide, we we're going to discuss what we
did to resolve them, now there we're in the guide stage.
The ACRS had recommended the removal, the
requirement to have prior NRC approval for changes resulting
in reduction of safety margins should be re-evaluated in
light of analytical assessments performed by research and
the results of the pilots. And the discussion also
identified using 50.59 as an alternative. The Staff
committed in the response to look at this requirement during
the public comment period. As a result of our review, we've
decided to retain the language in the guide, for two
reasons.
One is, this particular guide has to apply to the
initial implementation of the alternate source term for
which 50.59 would not be applicable. The re-baselining in
the pilots provided a lot of good insights, but it was based
on a limited sample of plants, and we don't feel that that
limited sample provides an a priori basis to summarily
disposition all potential plant-specific and modification-
specific impacts. So we retained the requirement that the
licensee will need to consider that.
We did, however, add language referencing 50.59
for the subsequent modifications.
MR. KRESS: But I think that's basically what we
had in mind anyway with this, was the subsequent
modification. So I think this is a pretty good response.
DR. LAVIE: The other recommendation that was
carried over to the guide was the recommendation we should
modify the proposed redefinition of the source term to
eliminate the connotation that the release is necessary to
the containment, but should retain the wording "released
from the RCS." At the time we responded, we pointed out
that we weren't going to change the rule language, but that
we were committed to reviewing the reg. guide to ensure that
the description was appropriate, that it would not cause
this confusion. The reason we did this is that the 50.2
definition had to address accidents other than the loca,
since the reg. guide and the alternate source term does.
And those accidents may not involve the RCS or containment.
So we decided that we'd best leave the definition in the
50.2 alone and make sure that it was clear in the guide.
The accident-specific appendices in the draft
guide and the final guide provide the guidance in what
constitutes the source term. We don't expect a licensee to
go back to 1465 and interpret it. The guidance is in this
reg. guide.
MR. KRESS: As long as it's clear.
DR. LAVIE: believe it is, Dr. Kress. The final
guide is a stand-alone document. We don't expect licensees
to refer back to 1465. However, we did add clarifications
to the final guide to ensure that the release from the loca
is consistent with the definition of 1465.
MR. KRESS: I think that fixes that problem.
DR. LAVIE: Okay. Those were the major changes
made. As I pointed out earlier on the gap fractions, we
expected to resolve those and be able to use the numbers we
currently have there. If we don't, I will get back to the
Committee and will decide the approach we want to take at
that time. Am I able to answer any more questions you might
have?
MR. KRESS: I thought the reg. guide were pretty
well-conceived documents that dealt with this issue very
nicely. I had a lot of minor problems with it, some of
which I've already --
DR. LAVIE: Okay.
MR. KRESS: -- with things like speciation and
gap release. I had a bunch of other little things that I'd
like to bring to your attention, and I don't think it's
worthy of putting it in a letter, and I don't even know if
it's worth wasting our time on here. A lot of them are
editorial comments. Some of them are things like --
DR. POWERS: There are a few things that they've,
you might help to go through that.
MR. KRESS: Well, okay. Some of them --
DR. POWERS: I mean, for editorial purposes, the
draft guide does have quite a few --
MR. KRESS: Yeah, and I could give those to him
separately. But there are things like, they've retained
this business of requiring the power to be 1.02, the rating
power, which in view of all the other uncertainties, that
seems a little strange.
They justify the breathing rates when they
calculate the doses, to three significant figures, which
seems a little strange also.
MR. ESTRADA: Good catch.
MR. KRESS: They still are talking about some
magic thing called an initiation temperature for, ignition,
ignition temperature. And there's gap release speciation;
it's not clear always in the reg. guide as to when the clock
starts on these timing things. So there's a lot of little
bitty, a few little bitty things in there that I don't think
make a hill of beans in difference in the outcome, but these
are mostly editorially things that I can probably write down
as a list and give them to him and send -- but I don't think
they're worthy of putting in a letter.
DR. POWERS: Have you received things like that?
DR. LAVIE: Yes, certainly.
MR. KRESS: I haven't done it yet, but I --
DR. LAVIE: Whatever we can do to improve the
document is certainly appreciated.
MR. KRESS: You can look at them and dispose of
them as you see fit. I don't see that it makes much
difference.
The other thing, I guess, is -- if there are no
more questions from other members, I think we, we have an
NEI reg here, and you don't care to comment? Okay, so with
that --
DR. POWERS: Sounds like the industry's happy.
MR. KRESS: Well, I think the industry's pretty
well pleased with the outcome of this. It looks like a
pretty good set of guidance to me.
DR. POWERS: The important thing is that we're
injecting some of the products of the substantial research
effort into the regulations.
MR. KRESS: And getting a little more realism
into it.
DR. POWERS: I guess I need to cogitate more
about this checkered approach toward defining some of the
criteria here. I hesitate only because if the accident
disappears from consideration, how much effort do I want to
invest in an accident that's never limiting anyway?
MR. KRESS: That's always a good question.
DR. POWERS: Members have any other comments
they'd like to make on this?
DR. BARTON: Nah, it's a pretty good piece of
work.
DR. POWERS: Okay, well thank you.
DR. LAVIE: Thank you.
DR. POWERS: And we will -- I can't start until
the Federal Register says I can start, recess until 2:30.
[Recess.]
DR. POWERS: Let's go back into session. We have
got to welcome Professor Apostolakis here. I hope the
graduation celebrations went well.
DR. APOSTOLAKIS: They went very well.
DR. POWERS: Members have before them a Document
14, Reconciliation of ACRS Comments and Recommendations.
They should examine these and if they have any comments.
I am tempted to say that we are now going to
explore an issue that involves an oxymoron, but I am going
to avoid that, and turn to the issue of quality in PRA. And
Professor Apostolakis, I think you are our leader in this
area.
DR. APOSTOLAKIS: Yes. Thank you, Mr. Chairman.
The Commission directed the staff to develop some
recommendations regarding the judgment of how good a PRA is
in the absence of the ASME and ANS standards, which, as we
all know, are being developed right now. This is a very
recent SRM, April 18, 2000. And your response is expected
by the end of this month, I understand. June 30th, is that
correct?
MR. MARKLEY: Yes.
DR. APOSTOLAKIS: We have not received the
document from the staff, understandably so, but since the
Commission expects it by the end of the month, I suppose we
will also see it then. And we will have to discuss at the
end of this whether we want to write a letter. If we do,
that will be in July.
The ASME standard, though, has been promised to us
sometime in June. Right, Mike?
MR. MARKLEY: Yes, June 14th.
DR. APOSTOLAKIS: June 14th. And we have, in
fact, scheduled a subcommittee meeting June 28th to discuss
that. So I don't know now whether the work you are doing
now will really ever be used, unless the ASME standard turns
out not to be acceptable, in which case, of course, what you
are doing now will be very, very valuable. But these are
perhaps questions whose answers are coming.
MR. MARKLEY: Yes, that's correct.
DR. APOSTOLAKIS: Okay. So without further ado, I
guess Mr. Cunningham has the floor.
DR. POWERS: Let me, before we get into this, the
Commission asked you to address the issue of PRA quality.
Unpleasant experiences that have occurred throughout my
professional career have taught me that the definition of
quality is a variable thing. And can you tell me what the
Commission had in mind when they used this word "quality"?
MR. CUNNINGHAM: We will try, try to lay the
context, anyway, of the question.
DR. APOSTOLAKIS: Okay. Go ahead.
MR. CUNNINGHAM: Thank you. My name is Mark
Cunningham, I am with the PRA branch in the Office of
Research. We have got four of us up here today. To my
right is Mary Drouin, also with the Office of Research;
Gareth Parry, from the Office of Nuclear Reactor Regulation;
and Richard Barrett, the chief of the PRA branch in the
Office of Nuclear Reactor Regulation.
I am going to start this out, but all of us are
going to hop in at various points and talk about what we
have got.
DR. APOSTOLAKIS: So there is much to be said
about PRA quality then?
MR. CUNNINGHAM: Pardon?
DR. APOSTOLAKIS: Since all four of you will be
contributing.
DR. POWERS: It takes that many people to search
for PRA quality.
DR. APOSTOLAKIS: In the absence of tools.
[Laughter.]
DR. POWERS: Because they are being out-gunned.
MR. CUNNINGHAM: Okay. Well, let's see, there is
four parts to our presentation today. I am going to provide
some background of what is going on right now in terms of
the SRM and other activities and give you a first, kind of a
general idea of how we intend to respond to the SRM, at
least today. Part of that response is going to be what we
call an attachment to the SRM. We are writing a document
that Mary and Rich and Gareth will talk to you about that
contains some -- the substance of the presentation, and then
I will come back at the end and talk a little bit about what
we have to do over the next month or so.
DR. APOSTOLAKIS: Over the next month?
MR. CUNNINGHAM: In responding to the SRM.
DR. APOSTOLAKIS: But the paper is due the 30th.
MR. CUNNINGHAM: Yes. Well, this month. What we
will be doing this month.
DR. APOSTOLAKIS: Okay.
DR. POWERS: This month is young yet, George.
DR. KRESS: Next month in the staff means the one
coming up, not the one --
MR. CUNNINGHAM: Over the next 30 days is what I
meant.
DR. APOSTOLAKIS: The next 30 days takes you into
July.
DR. POWERS: Yes, but all these people are working
16 hours a day, so they actually get two months for every
calendar month.
DR. APOSTOLAKIS: Maybe we can start to talk about
serious matters.
MR. CUNNINGHAM: Anyway, right now the staff has
got four issues related to the general issue of PRA quality
facing it, that it will be facing over the next few months.
The first is the response to the SRM that Professor
Apostolakis talked about. This came about in a briefing
that the staff had on the risk-informed regulation
implementation plan. And in that briefing, the staff talked
about some of its concerns, that the schedule for the ASME
standard was slipping, or appeared to be slipping.
The Commission made comments, various
Commissioners at that time made comments about the general
question of, how are we going to deal with the fact that the
standard is slipping in time, and what are we doing to make
sure that the PRAs that we are reviewing and we are using
are good enough for the task? So, in a sense, that is what
I think the intention was in terms of the quality. Are
these -- are the PRAs that we have available to us, or the
licensees are using, good enough? And how are we confident
that they are good enough to be used in the applications
that we have got in front of us today?
So, I think that was, as I recall, kind of the
general context of the quality question, and it got kind of
condensed down to a definition of PRA quality.
So, anyway, right now we have this, we owe at the
end of the month a response to that SRM. In parallel, we
have got a couple of other things going on. The Nuclear
Energy Institute has submitted a certification document, one
of many documents that it is submitting in the context of
possible use in the Option 2 analysis of Part 50.
DR. APOSTOLAKIS: I am not sure that we have
really investigated or discussed the certification process.
We had a small presentation, as I recall vaguely, but I
think --
DR. SEALE: We had a presentation from the GE
Owners Group, as I recall.
DR. APOSTOLAKIS: Right. But it was --
DR. SEALE: On their certification process.
DR. APOSTOLAKIS: It didn't go into detail,
though. It was more a high level description. In fact,
what I am saying is I am wondering whether the members would
benefit by reading this document if you can give it to us.
MR. CUNNINGHAM: Yes, certainly in the context of
how the staff is proceeding and things, and how --
DR. APOSTOLAKIS: Yeah, because we really have to
understand.
MR. CUNNINGHAM: It is an important background
document to the Option 2 work that you will be reviewing.
DR. APOSTOLAKIS: We have to understand the
certification process. And I mean just to have high level
discussions, as you know, we look at this and that, it
doesn't mean anything to me. I really have to see the
details.
MR. BARRETT: The schedule for the staff review of
that document hasn't been firmed up yet, but it is going to
be -- I believe it is going to be a schedule sometime toward
the end of this year. So, in your thinking about what you
want to review and when you want to review it, you might
keep in mind that we will certainly be coming to talk to you
about the staff review.
DR. APOSTOLAKIS: I would like to have the
document itself, you know, to start educating myself, you
know, in anticipation of your visit here, Rich.
DR. SHACK: Is that the NEI document that will
also be used to classify components under the Option 2?
That is a different document?
MR. BARRETT: Those are separate documents, but we
are going to be reviewing them in tandem.
MR. MARKLEY: But there is also four separate
certification processes, right, for each one of the Owners
Groups, that they are not altogether that linked? I mean
there are similarities, but they are different. Is that
correct?
MR. CUNNINGHAM: That is correct.
MR. BARRETT: There is one NEI submittal, NEI
0002, but there are the sub-tier criteria that are used by
the various groups are different, because of the differences
in the reactors. Maybe Mike Cheok could --
MR. MARKLEY: That's fine.
MR. BARRETT: Okay.
DR. APOSTOLAKIS: So we will get a copy of this?
MS. DROUIN: Yes.
MR. CUNNINGHAM: Yes. As Rich was kind of
alluding to, over the next few months or so, we will be --
the staff will be reviewing the information NEI submitted in
the context of its possible use in the Option 2 work.
DR. APOSTOLAKIS: But what if the ASME standard is
approved, is accepted by the staff, then what happens to the
NEI document?
MR. CUNNINGHAM: I will come back to that.
DR. APOSTOLAKIS: Okay.
MR. CUNNINGHAM: The next slide, slide 4 is the
other things that are happening kind of in parallel in time
with the NEI information and the Commission SRM is that we
expect to see the next version of the ASME standard for
staff review, if you will, about the end of this month. So
this is Rev. 12 of the ASME, proposed ASME standard.
DR. APOSTOLAKIS: Out of how many, do you think?
MR. CUNNINGHAM: Out of 12.
DR. APOSTOLAKIS: Okay.
MR. CUNNINGHAM: The goal of ASME is to have it
out the latter part of this month. There is a public
workshop on it June 27th, that sort of thing, to tell
people. And they are asking, they are soliciting comment
over a 60 day period for that. So, in the July-August
timeframe, the staff expects to be looking at the ASME
document.
In about the same timeframe, ANS expects to issue
its draft standard on external hazards.
DR. APOSTOLAKIS: It says only seismic.
MR. CUNNINGHAM: It says seismic on the slide and
I have to apologize. The slide -- the standard covers -- is
mostly seismic, but it covers also things such as external
flood and high winds and that sort of thing, how to analyze
them.
DR. APOSTOLAKIS: Fires?
MR. CUNNINGHAM: Not internal fires.
DR. APOSTOLAKIS: Oh. Who is covering that?
MR. CUNNINGHAM: The National Fire Protection.
DR. APOSTOLAKIS: 805?
MR. CUNNINGHAM: Yes, the 805 at this point is the
standard that is under review. So, ANS will be issuing its
-- or has a goal of issuing its external hazards standard
for public comment, again, around the end of this month or
in early July, again, for a 60 day period or review. Just
for what it is worth, that covers, in the seismic area, that
covers both the seismic PRA and the seismic margins
approach.
So, again, the staff is expecting that in the
July-August timeframe, we will be looking at those documents
and commenting on them.
The goal of ASME right now is to have the final
version of their standard out in January of next year. The
goal for the seismic or the external hazards work by ANS is
to have it done in September. I should also note there, ANS
is also working on a standard for low power and shutdown
analysis. That right now is scheduled to be out in
September and finalized in December, but we just have a
feeling that that is not going to happen that quickly. That
one is a little further behind that the seismic work.
So, at any rate, we have got three or four things
in front of us.
DR. APOSTOLAKIS: The ANS low power shutdown work
is the standard on how to do a risk -- a PRA for those
modes. It is not how to manage risk during those modes.
MR. CUNNINGHAM: Correct. Correct. And it has a
quantitative approach to assessing the risk and a more
qualitative approach to assessing the risk. It is intended
to be part of standard. But you are right, it is not how to
manage an outage, if you will.
So, again, the staff has three or four things in
front of us, all of which are related to the issue of PRA
quality. So what we intended to do is --
DR. APOSTOLAKIS: Actually, you know, that is
confusing. You can qualitatively manage risk, but you
cannot qualitatively assess risk. The only way to assess it
is quantitatively.
MS. DROUIN: What was proposed by the project team
from the qualitative approach was to create a benchmark, and
then you would compare your plant against that benchmark to
see where you fell. And to try and establish a benchmark to
the level where it didn't contribute on a relative basis to
the full power, and as long as you met that benchmark or you
were below it, then you were okay. I mean that is just kind
of a quick summary of what the qualitative approach.
DR. APOSTOLAKIS: Which is really managing. Which
is really managing rather than assessing.
MS. DROUIN: In a sense, yes.
DR. KRESS: When you said benchmark, you mean a
benchmark PRA?
MS. DROUIN: A benchmark simplified, I should say
simplified PRA.
DR. KRESS: Well, that is not managing risk at
all, it is just a qualitative way to compare your PRA with
something that has a known --
MS. DROUIN: Risk associated with it.
DR. KRESS: A known uncertainty.
MS. DROUIN: But it does have a management part,
because you would manage, --
DR. APOSTOLAKIS: Yeah, because the management, it
is really management.
MS. DROUIN: -- you know, your configuration to
meet or stay below that benchmark.
DR. APOSTOLAKIS: What you do. Again, it is not
really a simplified PRA. A limited scope PRA. You simply
something --
MS. DROUIN: That is probably a better
characterization, yes.
MR. CUNNINGHAM: Slide 5 then provides basically
an outline of what we intend to -- of what the structure
will look like of the response to the SRM, and that is going
to lay out what we intend to do over the next few months to
deal with this, how to deal in kind of a more integral way,
each of these individual issues I have talked about before.
What that means is the SRM is going to first
summarize what the Staff is now doing and what that means is
again in the context of the Commission briefing is in the
context of the applications that the Staff is now using, in
the places the Staff is now using PRA what is the Staff
doing to ensure the appropriate quality and scope of the PRA
for that application.
One example of that really is in the license
amendment context. Today we use Reg Guide 1.174 and SRP
Chapter 19 to guide us through how we ensure that the PRA is
adequate for the intended use. It is general guidance at
this point but it served us well I think in terms of license
amendments. That we are going to try to summarize to the
Commission how that is being done today, and Rich will come
back a little bit later to talk about how we are going to
elaborate on what is in the existing Reg Guide and SRP.
We are going to then propose or recommend to the
Commission that the Staff or inform the Commission that the
Staff has some other things we are going to do to integrate
all these pieces together.
One is we are in the process of writing a couple
of things that will help better lay out how we use PRA in
the Staff reviews and then what is needed in that PRA, in
those PRAs for those applications.
The idea is to draft a document and attach it to
the Commission paper and summarize it in the paper itself.
DR. APOSTOLAKIS: I must say, Mark, I am a little
bit confused. Given the time scale on which the Agency
operates, if the ASME standard is any good you don't need to
do this because by the time you are done with this, the
standard will be out, so do you guys know something we don't
know?
MR. CUNNINGHAM: No, not well.
DR. APOSTOLAKIS: Are you preparing yourselves now
independently to judge the last bullet there --
MR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: -- to be able to review the
standard and --
MR. CUNNINGHAM: Yes, that is what we are doing.
DR. APOSTOLAKIS: So you are positioning
yourselves?
MR. CUNNINGHAM: We are positioning ourselves. We
are doing our homework or whatever to say -- and these
Staff documents were intended to lay out what are we going
to use as the basis to review the ASME standard when it
comes in next month and what are we going to use as the
basis for reviewing the certification document.
We think it needs to be one document that spells
out how we are going to do that for each of those things.
To get back at a point that I believe you asked
about earlier, I think we go back to Reg Guide 1.174. We
talked about either a consensus standard or certification
process could be found to be acceptable in ensuring needed
PRA quality.
DR. APOSTOLAKIS: For some applications.
MR. CUNNINGHAM: For some applications, and I
think that is the context we are still working in. We are
not presuming that one or the other would obviate the need
for the other, that we can see that either of these
documents, the ASME standard, ANS standard or the
certification process could be acceptable so we are
expecting it. We are not prejudging at this point that
either will replace -- one that will replace the other.
DR. APOSTOLAKIS: So you are not really sending a
message to the ASME that the standard may not be acceptable?
MR. CUNNINGHAM: No, I don't think we are
intending to send that message at all but we need to do our
homework and set out what is it that we want to establish is
what we want, if you will, and Mary and Gareth and Rich will
talk about this in a little while.
DR. APOSTOLAKIS: Is Mary the only one who
participates in the ASME activities --
MS. DROUIN: Yes.
DR. APOSTOLAKIS: -- from the Staff, I mean from
you four?
MR. CUNNINGHAM: Mary is the representative of the
NRC on a committee. All of us are involved in reviewing the
material.
DR. APOSTOLAKIS: Oh, okay.
MR. CUNNINGHAM: But Mary is the official point of
contact, if you will.
At any rate, we will talk about a little bit later
these documents. We think it may be appropriate to update
Reg Guide 1.174 and/or the Chapter 19 to reflect what we
write in these documents and just somewhat coincidentally we
are going through the process to come up with the next
update of the Reg Guide and the SRP, so the timing may be
right for that.
We intend to review the submittals, either the
certification documents and the ASME standard draft, against
what we are writing down, basically, provide comments back
to the appropriate people, either ASME or ANS or NEI, and
then when we get down the road and get the final versions of
those documents we would be prepared to review them, again
against what we have written down and endorse them or
endorse them with exceptions.
At any rate, the remainder of the presentation is
basically what is going to be in these couple of documents
that we're talking about.
Mary is going to take it from here to talk about
in general what will be in this attachment. That will be a
mixture of Mary and Rich and Gareth talking about the
details.
MS. DROUIN: Mary Drouin, Office of Research.
This attachment that is going to be to the SECY
basically has three parts to it.
The first part is laying out what are those risk-
informed activities where we feel PRA quality is an issue
and needs to be addressed, because one of the things we want
to do is come up with an integrated and uniform, consistent
definition there that is going to be applied across these.
Now we do recognize that PRA quality is variable
with application but when we define what a PRA is, that
definition of a PRA should be uniform and consistent, so we
first are going to talk about what are those activities that
this will be addressing and then Rich will get into NRC's
decisionmaking process, how PRA and the quality of the PRA
folds into our decisionmaking process and we'll get into
that next.
The last part of the attachment then gets into the
details of what we consider to be technically acceptable in
a PRA and then it has in there a discussion of what the PRA
scope and level of analysis needs to be, the elements and
characteristics of peer review, because that is one way to
get to your technical acceptability.
The next one is we call it the PRA application
process and that is intending to lay out the characteristics
and attributes of a decision process in looking at what
scope and elements you need for a specific application.
The last one, laying out the attributes and
characteristics of an expert panel, because in many cases
you might be using an expert panel to supplement your PRA if
your PRA doesn't cover the necessary scope or elements.
DR. APOSTOLAKIS: Could we attempt to define what
a good enough PRA is for a particular application by saying
that we make some decisions using that PRA in a particular
context. If one did a more detailed analysis the decision
would not change. In other words, that the decision is
robust.
Shouldn't that be the ultimate criterion, because
when you are dealing with these things you can't experiment.
You can't go and blow up things and see what happens. The
only thing that makes a connection between risk assessment
and whatever, with real life, the physical world, is the
decisions you make, so that should be the ultimate criteria,
that if somebody came back with a ten volume PRA for this
issue, which you handled with three pages, the decision
would not change.
DR. KRESS: It is like proving the negative,
George.
MR. CUNNINGHAM: The decision might change.
DR. KRESS: Yes, you don't know whether --
DR. APOSTOLAKIS: Then it is not good enough.
DR. KRESS: You don't know whether the decision
would change and so you have a never-ending set of things to
worry about.
DR. APOSTOLAKIS: But this is the ultimate
criterion though.
MS. DROUIN: I think there's a different way --
DR. APOSTOLAKIS: This is the ultimate criterion.
MS. DROUIN: I think there is a different way to
look at it, George, in that you could lay out your minimum
requirements for a PRA. I think that would be very
difficult. That would cover every application --
DR. APOSTOLAKIS: Sure.
MS. DROUIN: -- because you could have an
application that comes in that has nothing to do, where for
example earthquakes have no effect --
DR. APOSTOLAKIS: That's right.
MS. DROUIN: -- and you might have a PRA where
either it either didn't cover earthquakes or they did a
lousy job.
DR. APOSTOLAKIS: That's what I mean, that they
should really start from the endpoint, the decisionmaking
process rather than starting by defining --
MS. DROUIN: And that is what we've done.
DR. APOSTOLAKIS: I don't dispute that. I am just
saying that --
MS. DROUIN: That is why that shows up first.
DR. KRESS: I think you ask yourself on the
application do you have some PRA need for this application,
an output of some kind, and then you have to ask yourself
based on the decision I want to make how good do I need to
know that number.
DR. APOSTOLAKIS: Right.
DR. KRESS: And if I only know it this well then I
make one kind of decision, and if I know it this well make
another one, and then you ask yourself the secondary
question how good does my PRA give me so I can tie that to
my need in the decision.
I don't think you do what you do. I think you
look at them both in that kind of context.
MS. DROUIN: I think we are going to answer your
question.
MR. CUNNINGHAM: Yes, you are jumping into Rich's
presentation already basically, because he gets at many of
the points that you raise, Dr. Kress.
DR. WALLIS: What I would like to do is make a
distinction between going through the motions and the
quality of the work. This comes up in, say, codes, thermal
hydraulics and so on. You can make a structure where you
have all the scope and level and elements, you have all the
right things in there but then when you are actually
modeling something you have to have equations, you have to
have coefficients, and they have to come from somewhere, and
the weak point of a lot of these things is they look good,
but there is very little guidance on what the coefficient
should be, where you could get them from, how you know if
they are good enough, so the devil is in those sorts of
details, not in the fact that the scope looks good and
everything. It's a different level.
Do you have something to say about that level in
PRA?
DR. KRESS: I think that is in your PRA elements
and characterization.
MS. DROUIN: If you bear with us, we are going to
get to that.
DR. WALLIS: That is part of the elements and
characteristics?
MS. DROUIN: Yes.
DR. WALLIS: Because you could have an element and
it can be lousy.
MS. DROUIN: It could. They could do it lousy.
DR. WALLIS: Okay. You are going to tell us.
MS. DROUIN: So the point is that the paper has
three parts. They have been done sequentially on purpose,
starting off with here are the activities, here is the
decisionmaking process of how the PRA plays into that, and
then it gets into the technical acceptability.
So the first one is at a high level we view these
more as areas where we feel that PRA quality is an issue and
that the rest of the document would play into is risk
informing 10 CFR Part 50, the plant oversight process, our
operating events assessment and our license amendments, and
all the different activities would fit into one of these
categories where we think the PRA quality needs to be
addressed.
On that, I am going to turn it over to Rich to get
into the decisionmaking process.
MR. BARRETT: I am Rich Barrett with Nuclear
Reactor Regulation.
I think a number of you have hit very early on
what we think is a key point of all this paper and that is
PRA quality is not something you can look at in isolation.
It is, as some have pointed out, dependent on what
application you are looking at.
I think more importantly it depends on your entire
decisionmaking process and so what we intend to do in the
paper is to address the decisionmaking process, the
decisionmaking process that a licensee goes through to come
to the application that they have submitted and the review
process that the NRC goes through and the factors that we
take into account in making those decisions.
Ultimately what we are trying to do here is not to
get quality PRAs. Quality PRAs are a step toward getting to
NRC reviews which allow us to make a finding, to make a
finding that the risk is acceptable and that we understand
the risk well enough that the risk can be bounded.
What we have here is perhaps a seemingly complex
spider chart which is intended to talk about some of the
factors that we take into account in making a decision and
in reviewing and accepting a licensee application, whether
it is a license amendment or some larger application.
If you go back and you read through Reg Guide
1.174, and SRP Chapter 19, as I have done in the last month.
You will find that there is an impressive amount
of information in there that would enlighten us in that
particular area. The people who drafted that document, and
I was not one of them, did a very nice job of thinking this
process through very carefully and I think maybe some of us
have forgotten a lot of that information.
What it basically says is that the Staff will
review an application from the licensees. We are not going
to have cases where the existence of a peer review or the
existence of a quality PRA will obviate the need for a Staff
review.
The question is what will the Staff review consist
of, how deep will it be, how resource intensive, and where
will it be focused. In making those decisions where we want
to start with is the question what is it that can give us
assurance that the risk is in an acceptable range and that
we understand the risk well enough to approve this
application.
In a simplified way there are really three factors
that I think come into play.
One of them is what I call, for lack of a better
term, risk limitation. There are cases where you can
examine an application and in a qualitative way after you
have looked at it come to the conclusion that your exposure
to risk is limited, and that is an important piece of
information to have before you ever go in and start looking
at the PRA results with the quantification or anything else.
The risk might be limited because of the nature of
the change that is being made. It might be limited because
of the extent of the changes being made. It could very well
be that there are controls that have been placed on this
application which limit. There could be backstops which
again limit the risk exposure, and you need to understand
those from the very start because that helps you not only
decide how much analysis you have to do, but it can help you
to focus where in the analysis you need to look.
Even if the risk is limited, there might be some
risk and you can understand where that risk is. You know
where to delve into the analysis.
We think it is very important that that is the
first place you look when you look at an application.
CHAIRMAN POWERS: Earlier today we discussed an
issue in which a persuasive case was made that the
probabilities of an event were very low, but that didn't
deter us because we had -- there was not a characterization
of the consequences and we had one of our members portray
sufficient of a nightmare to us that it looked like those
consequences were very big.
When you use this word "risk" you really are
looking at the products of these or are you only looking at
the probabilities of an event?
MR. BARRETT: I think that if you make a
qualitative judgment, I think my favorite example was that
two years ago, the ACRS rightfully asked us to look at the
potential risk significance of powerup rates in the BWRs.
There were five-eight percent powerup rates.
And what we did was, we went in and we asked
ourselves, what could be the impact of this on risk? And we
looked at both the probabilities and the potential impacts
on consequences, things such as different amounts of fission
products, perhaps the different effects on containment
response.
We looked at the probabilistic aspects, less time,
perhaps, for operator reactions, maybe different success
criteria for systems and operator actions.
And we did look at a couple of plant-specific
applications. But by and large, we came to, I think, a
pretty robust and qualitative conclusion that the risk from
these small powerup rates was really quite limited.
DR. APOSTOLAKIS: So you looked at both,
consequences and probability?
MR. BARRETT: Yes, you would have to look at
consequences. You can't allow yourself to get seduced into
just looking at CDF and LERF, and perhaps you miss some of
the other factors that impact risk.
DR. APOSTOLAKIS: Maybe you need a better word
than risk limitation.
MR. BARRETT: I would welcome a better term.
DR. APOSTOLAKIS: Risk implications or --
MR. BARRETT: Risk vulnerability.
DR. APOSTOLAKIS: Something, yes, because my mind
went immediately to limitations of PRA, but that's not what
you mean.
MR. BARRETT: Yes.
DR. APOSTOLAKIS: Implications? I don't know. Do
you plan to talk about this more, or do you want me to make
comments?
MR. BARRETT: I welcome your comments.
DR. APOSTOLAKIS: I would change this a little
bit, and maybe combine the boxes, non-PRA insights and
deterministic analysis into one and call it non-PRA
insights, or traditional analyses. I would avoid the word,
deterministic, because a lot of it goes into PRA as well.
What you mean is the traditional engineering
analysis that the Agency is used to.
And then I don't know why PRA has this honor of
being peer-reviewed and the other stuff does not. I would
delete the boxes that says peer-reviewed, or everything is
peer-reviewed.
MR. BARRETT: Everything, indeed, is peer-
reviewed. I mean, anything that is submitted to the NRC
goes through some sort of a quality check. What I was
trying to highlight here is the role of PRA quality.
DR. APOSTOLAKIS: Okay.
MR. BARRETT: And there we're talking about the
peer review which we're calling the certification process.
DR. APOSTOLAKIS: Yes, but you don't want to send
the wrong message that you are imposing extra requirements
on the PRA. I mean, the other stuff will be -- but do you
agree that perhaps the two boxes should be combined into
one?
If not, I would change the word, deterministic, to
traditional engineering analysis.
MR. BARRETT: Well, I think -- which two boxes are
you talking about?
DR. APOSTOLAKIS: Non-PRA insights and
deterministic analyses.
MR. BARRETT: No, those are two different things,
really.
DR. APOSTOLAKIS: They are two different things?
MR. BARRETT: Yes, those really are. I think in
the case of the traditional deterministic analyses, what
we're talking about there are questions of margin and
defense-in-depth, and some of the traditional types of
licensing questions that are issues that we bring into --
the things that make this risk-informed instead of risk-
based.
Non-PRA insights really is non-PRA. It should
really say unquantified risk insights.
DR. APOSTOLAKIS: Oh.
MR. BARRETT: That's what it really should say.
DR. APOSTOLAKIS: Okay, so why don't we say that?
MR. BARRETT: We will.
DR. APOSTOLAKIS: And you agree to change the
other one to traditional engineering analysis?
MR. BARRETT: I understand that.
DR. APOSTOLAKIS: We had the same battle when you
guys put together the diagram in 1.174. Deterministic
analysis feeds a lot into PRA.
DR. WALLIS: It will become a traditional PRA
analysis soon when it becomes a tradition.
DR. APOSTOLAKIS: Engineering? I don't know.
CHAIRMAN POWERS: It's hard to find a valid --
DR. SHACK: PRA is not engineering?
[Laughter.]
CHAIRMAN POWERS: You're going to get him upset.
DR. APOSTOLAKIS: I don't know what deterministic
analysis is. You mean the traditional licensing kind of
analysis, but you don't want to call it that?
MR. BARRETT: Yes, I don't necessarily want to
call it that, because we don't necessarily do it in the same
way as we do licensing analysis when we're doing this kind
of thing. We may, in fact, do more best estimate than we
would do in a licensing, in a case that was wholly based on
licensing on deterministic design basis type events.
But let me try to find better terms.
DR. APOSTOLAKIS: Yes.
MR. BARRETT: Putting out a chart like this is
difficult. It's difficult to find the right words. But I
like unquantified risk insights. I could just put down
defense-in-depth in margins, how about that?
DR. APOSTOLAKIS: If you want to.
MR. BARRETT: Okay.
CHAIRMAN POWERS: Good.
DR. APOSTOLAKIS: Well, actually, no, unquantified
risk insights is better.
MR. BARRETT: No, I meant for the deterministic
analysis.
DR. APOSTOLAKIS: Oh, yes, yes, that would be
fine.
MR. BARRETT: Okay. Actually in the more complete
version of this chart, that's what's in there.
DR. APOSTOLAKIS: Defense in depth and safety
margin considerations.
MR. BARRETT: Okay, so we have PRA, unquantified
risk insights, and defense-in-depth and margin
considerations.
DR. APOSTOLAKIS: Right, and the peer-review is
only for PRA, or do all three feed into a peer review?
MR. BARRETT: All three would fit into a peer
review. There would always be a quality review internal to
a licensee before they submitted it, but the peer review
process we're talking about here -- and maybe what I would
do is just put in NEI 002 as meaning that this is what that
box means.
That box means that these plants, these PRAs have
been subjected to a peer review of that type.
The second consideration before we get to the
analysis part, is on the far right side here, which is
performance monitoring.
We frequently find that one of the best ways of
limiting risk and assuring ourselves of safety is if we can
find that the application has a good way of defining
measures and criteria that can be monitored, real-time, to
assure that the decision we've made does not lead to
unacceptable consequences.
Now, we've had a number of discussions as to what
constitutes an acceptable performance monitoring program.
You have to have relevant measures; you have to have good
criteria; and you have to be sure that you can detect
unacceptable performance in a timely way before it becomes a
risk and a public safety issue.
But we do have examples where this is important,
and so we need to look at this before we delve into the PRA
as well. So, having said all of that, and we then look at
the analyses and we see that the PRA, the quantified PRA, is
an important part of the analysis but it's not the only part
of the analysis.
The Reg Guide points us to other things such as
unquantified risk insights and defense-in-depth, and margin,
and other factors, operational experience and a whole wealth
of information that can be brought to bear to give you
assurance.
DR. APOSTOLAKIS: Should the box currently labeled
risk limitation, also feed into the analysis? I thought
what you said earlier was -- well, maybe you didn't say it,
but if you do this assessment of the significance of the
issue you are dealing with, that will certainly affect your
analysis.
And right now it appears that it goes straight to
the integrated decisionmaking, and the analysis is done
independently of that. In reality, it won't be.
MR. BARRETT: I think you could probably draw
lines from all three of these in various directions.
DR. APOSTOLAKIS: Maybe that's a shell of the
issue that you are dealing with, and everything else is
within that. You know, if you -- for example, if
earthquakes is an issue that is irrelevant to the issue at
hand, you're not going to ask a PRA to do an earthquake
analysis.
MR. BARRETT: Right. The idea of this is to look
at the whole process from the bottom up, as to say the Staff
-- you're the Staffer and you're reviewing the application.
What is it that you're going to be looking for? And it
doesn't get so much into the internal workings of the
licensee and how they went about making their decisions.
DR. APOSTOLAKIS: And the other point I want to
make is that I congratulate you on spelling decisionmaking
process correctly, as opposed to the ACRS Staff that makes
it one word. It drives me up the wall.
MR. BARRETT: We'll accept any comment.
[Laughter.]
DR. APOSTOLAKIS: Even at the expense of your
colleagues here.
[Laughter.]
DR. SIEBER: They actually have it both ways on
that chart, so you can use.
DR. WALLIS: Which one do you think is correct,
George?
DR. APOSTOLAKIS: They hyphenated one.
DR. KRESS: George, I will offer an opinion on
this weighty subject that both ways are correct, and they
used them correctly both there. In the integrated
decisionmaking, it is a noun; it is a thing. Up there, it's
an adjective talking about the process, and it's an
adjective, so they've used it correctly.
DR. APOSTOLAKIS: These guys used it.
DR. KRESS: These guys used it --
DR. APOSTOLAKIS: But in our records, we don't.
DR. KRESS: Maybe not, but here it's used
correctly both times.
DR. APOSTOLAKIS: Decisionmaking process in our
letters is one word, decisionmaking.
MR. MARKLEY: George, we do it both ways also.
CHAIRMAN POWERS: I think we can move on.
DR. KRESS: Before you move on, though, I would
like to be more substantive. I would like to see that
other.
Now, what I envisioned, Rich, is that you have
some sort of decision to make about some change, and whether
or not its acceptable.
And you're going to have some criteria to guide
you on whether it's acceptable or not. And some of that
criteria may be, does it meet certain risk levels. And if
it doesn't, you'll have to have a PRA or some sort of way to
judge what the risk implications are.
But your decision as to whether it's acceptable or
not, may depend on what quality of PRA you have in
determining that, and it may or may not be acceptable, but
if the quality is not very good of the PRA, you could offset
that by having performance monitoring, more defense in depth
and bigger margins, so that the integrated decisionmaking
and these other things are all tied to the quality of the
PRA.
If it's a poor quality, you have to have more
performance, you have to have more defense-in-depth, better
margins, so that the way I view it is that you have a set of
basically a matrix of criteria that, depending on what
quality of PRA you have, you will have an acceptance
criteria that depends -- that varies these other things. Is
that a way to look at this?
MR. BARRETT: If I was a licensee -- well, as a
regulator, I think I look at these three areas as being
tradeoffs, making tradeoffs.
If I was a licensee, I would think in terms of
iterating. That is to say, if I could not, myself, make the
arguments that risk is limited, or that I could perform as
monitor, and yet I didn't think that I had the sufficient
quality of PRA, I might go back and put controls on this
thing or backstops on the applications so that I could
further limit the risk, or I might go back and look harder
for some way to monitor performance.
DR. KRESS: Or put more defense-in-depth in so
that the actual risk numbers you get are coming down or
something like that.
DR. SIEBER: If you were writing risk-informed
rules, this would be the model that you would want to use to
set up how those rules and all the supporting documents like
Reg Guides and so forth would be, because it should, in my
view, have a risk expectation associated with the rule, and
a way to monitor performance to make sure that the input
assumptions to the risk analysis are correct.
And so this then becomes the model.
DR. APOSTOLAKIS: That's what 1.174 does.
DR. SIEBER: Well, this is the model for risk-
informed rulemaking, which is one of the tasks that's ahead
of us in the near future.
MR. PARRY: I think this gets back to the question
that you raised earlier; that if you had a -- if you could
define a quality of PRA, then you could -- I don't know the
way you phrased it, exactly.
But you'd say this is the -- once you've got that
and you've got a robust decision.
I think you can look at it a different way, which
we haven't quite addressed yet. And again, it's to do with
tradeoffs.
The more confidence you have in PRA results, so
whatever that role -- whatever the role of the PRA is in the
decisionmaking process and it's balanced on these three legs
that Rich has got in this diagram, the more confidence you
have in the PRA results, then perhaps the less conservative
you can make your decisions.
So I think that if you look at it in terms of
Option 2, for example, the more confidence you have in your
PRA -- in the PRA results that you're using, then perhaps
you can shift more components into one box rather than
another. So it's not a matter of making a robust decision
at that level; the decision may change.
DR. APOSTOLAKIS: I still think I can place what
you said in the context of my interpretation, in the sense
that the decision is not robust if based on this current
state of knowledge, somebody feels that you have made a
conservative decision, and by doing more analysis, that
person feels that he can convince you that that's the case.
So then your PRA is not adequate for the
application. I think ultimately what drives it is the
decision.
MR. PARRY: Yes, but I think it depends on what
you want to get out of it.
DR. APOSTOLAKIS: I think we understand that. In
fact, that's the point of Rich's comment when he addressed
the risk limitation there. You know, let's not forget where
we are and what kind of decision we're about to make, and
then we jump into the analysis.
MR. BARRETT: I guess the last point I want to
make on this slide is the importance of the staff's review
of NEI 0002, and the allied process for risk categorization,
because, in reality, we are going to be faced with PRAs that
have been subjected to this peer review. That is going to
be our tie to what is actually in the industry. And, so, it
is very important, as Mary said earlier, that we have a
common set of expectations that we apply to this peer review
process and that we apply to our review of the ASME
standard.
DR. KRESS: And when you say PRA here, this could
be a PRA that has associated with it a fairly robust
uncertainty analysis, or it could be one of these nominal
PRAs that we call best estimate, for whatever that means,
without uncertainty.
DR. APOSTOLAKIS: Could be.
DR. KRESS: Are we including both of those as the
definition of what a PRA is?
MR. BARRETT: I think the logic that we are going
to apply is that we are going to, and Mary is going to talk
about this in just a minute, we are going to lay down our
expectation of what is a PRA. And then the second question
we are going to ask is, does the -- to what extent does the
NEI peer review process give us the assurance that a PRA
meets that standard? And it may give us 90 percent
confidence, it may give us 100 percent confidence. We need
to understand how much confidence it gives us and we need to
understand where the deltas are. And then with that, armed
with that knowledge, we can go forward to specific
applications.
DR. APOSTOLAKIS: I would say, Tom, that there are
many cases where you really don't need an uncertainty
analysis. It really is very problem-specific. You don't
need an explicit quantitative analysis, you always do it in
your mind, of course. But that is a lesson that I think one
can learn from actual applications, that you don't always
need it. But, anyway, we will see when it comes.
DR. POWERS: George, do we have enough cases
before us where people have done explicit uncertainty
analyses that we are justified in drawing conclusions on
when and where you don't need uncertainty analyses?
DR. APOSTOLAKIS: First of all, it depends on what
we mean by uncertainty analysis. What I mean is the
traditional failure rate type of thing. Yes, when people
realized that jumping into a full scope, complete
uncertainty analysis PRA was too expensive, they started
doing it in phases.
The first phase was, you know, using insights from
other PRAs and rough point estimates just to rank things.
Very rarely, the insights regarding what is important from
this point estimate calculation were upset by a detailed
uncertainty analysis, I will say. You pretty much had the
good grasp of the major accident sequences most of the time.
And then you refine it and you refine it, you go down to
more detail and so on. So, there are insights, important
insights you can gain without going through the whole
exercise.
DR. POWERS: But I guess when I think about
detailed uncertainty analysis and PRAs, and where I have
seen them, they come up with very few.
DR. APOSTOLAKIS: Very few what?
DR. POWERS: Very few examples.
DR. APOSTOLAKIS: Of complete uncertainty
analysis?
DR. POWERS: Any kind of uncertainty analysis at
all.
DR. APOSTOLAKIS: Well, the PLG, PRAs are full
uncertainty analysis. How many have they done? I don't
know. I am sure others have done it, too.
MR. PARRY: I think, though, where the uncertainty
analysis is most important is when you are making the
decision. And I think even if you look at -- it needn't be,
as you say, a full quantitative uncertainty analysis. If
you can understand what the sources of the uncertainties
are, this is what Reg. Guide 1.174 says, understand the
sources of uncertainty and see how they impact the decision.
In fact, if you look at the NEI guidance on
categorization, you will see elements of that in that
process. An example of what they do is for -- they will do
the categorization according to Rohr and Fussell-Vesselly.
But they will also ask you to do sensitivity studies to vary
some of the parameters that are perhaps the more
controversial parameters, and then use the results of those
to adjust the categorization. And I think that is an
appropriate use of uncertainty analysis in the context of
your decision. I don't think you have to have a PRA that
has fully quantified everything to play the game for many of
these things.
DR. POWERS: Well, I guess what I am asking is,
you somehow have reached this judgment that these
sensitivity studies and varying some of the parameters is
somewhat adequate for the --
MR. PARRY: It is adequate.
DR. POWERS: That it is potentially adequate.
MR. PARRY: Right.
DR. POWERS: And what I am asking is, have we had
enough people do something that is akin to what they tried
to do in 1150? Has that been done often enough that I can
use a judgmental, that I can develop some judgment on when
something much less than that is adequate. And George says
that -- he is a bright guy, and he has seen a lot of these
things, and he says, gee, I can get all the insights I want
with very little.
DR. APOSTOLAKIS: Not all.
DR. POWERS: I am not nearly so bright, but I do
get to do a lot of uncertainty analyses on deterministic
models, and I am always stunned at what I find to be the
influential parameters. I am 0 for 10 in outguessing the
system.
MR. PARRY: But those are very nonlinear systems.
DR. APOSTOLAKIS: Very nonlinear systems, plus
what you say about the number of studies. Yes, in level 1
PRAs for internal events, there are many of them, both
nationally, internationally, so things are beginning to
converge to a certain picture. You know, you sort of expect
to see certain things for PWRs. We had this wonderful
compilation of insights from IPEs that Mary put together.
Yes, there is a wealth of experience there. You are not
going to be surprised.
DR. POWERS: Well, I think of this wonderful
compilation that Mary put together, an outstanding piece of
work, that I continue to refer to, but, unfortunately, some
of its most influential graphs which show the range of
results obtained from a variety of analyses do not have
uncertainty bars on them. And it makes it very difficult
for me to interpret the significance of the range of core
damage frequencies among PWRs if I don't have those
uncertainty bars there.
DR. APOSTOLAKIS: And that may be an instance
where you have to do uncertainty analysis. All I said was
that sometimes you get very useful insights without an
uncertainty analysis.
DR. POWERS: Well, what I end up doing --
DR. APOSTOLAKIS: And this is the major
contributors to risk. I don't think that if you do the
point estimate and then you go and round your uncertainty
calculations, you are going to upset the order that much.
DR. POWERS: See, what I end up doing is I take
Mary's plots and I use them as an ensemble. And I say --
and, so, somebody does a -- gives me a result from a PWR, I
plot it, put it on Mary's plot, and I say, now that is the
uncertainty that I am going to have to make my decisions in,
because it is the ensemble. It is the only thing I have got
to go by.
DR. APOSTOLAKIS: I don't know what to say to
that. I mean you need uncertainty analysis there, what can
I tell you. All I said was there are instances where you
don't. You can have 80 percent of your insights by doing a
very quick calculation with point estimates, that is all.
MS. DROUIN: And I think if you also go to those
wonderful plots, and you look at the outliers that are
forming the bands, you may not have uncertainty analysis on
them, but on those few outliers you tend to have sensitivity
studies and know what the effect of those things are, there
is sensitivity analysis.
DR. POWERS: Well, you try to communicate it in
your document.
MS. DROUIN: And you know why they are outliers.
DR. POWERS: Why they are there.
MS. DROUIN: So you are getting the same
information. You are getting the information that you need
to know.
DR. APOSTOLAKIS: Are you going to discuss
uncertainty soon, or this is it?
MS. DROUIN: I'm sorry?
DR. APOSTOLAKIS: Uncertainty analysis. Is it
going to be discussed separately today? Or we are
discussing it now? I mean is there a slide?
MS. DROUIN: Not as a specific topic, as a slide,
no. You might see one.
DR. APOSTOLAKIS: I think -- then let me make one
last comment.
MS. DROUIN: But I will show --
DR. APOSTOLAKIS: If you have later, that is fine,
I will wait. Do you want me to do it now or later?
MS. DROUIN: Do it later.
SPEAKER: Much later.
[Laughter.]
DR. POWERS: Let's go on. I think we have spent
enough time on it.
MS. DROUIN: How about when we get into the slide
where it will be addressed, even though it is not a specific
topic.
MR. BARRETT: Let's skip -- I want to skip to page
12. The intervening pages have all been discussing.
DR. APOSTOLAKIS: Oh, wonderful. You should come
here more often, Rich.
MR. BARRETT: Doesn't that feel good when you get
to skip over some items. What I have tried to do is just
work through three examples. What these examples are, what
I have tried to do is pick three examples which are kind of
extremes. One in which you are principally relying on the
analysis for your assurance. One in which you are
principally relying on your qualitative judgment that the
risk is low. And one in which you are relying primarily on
performance monitoring.
And, unfortunately, in our last minute rush to
make this slide, we didn't quite get it right. So, I am
going to -- what I want you to do is, on that -- what I have
done here is basically say, if you see the word "high," what
that means is that there is high reliance on that factor.
So risk limitation, high reliance that there is a limited
risk. High reliance on the analysis, or high reliance on a
good performance monitoring program.
The intent of the first example was to refer to
the risk-informed standard tech specs, or, in general, the
use of risk for configuration management. Now, you could
call the use of a risk monitor, or you could use the example
of the industry's proposal to take A-4 of the maintenance
rule and the tech specs and rationalize those into a single
set of requirements that are risk-informed, or you could
refer to element 4 of the risk-informed standard tech specs,
which is a proposal for licensees in a -- who reach a LCO, a
limiting condition for operation, to use their PRA or use
their risk insights to decide what their end state will be.
Any of these examples of using your risk methods
to determine the configuration of the plant, we would say
that in those cases, the risk limitation is low. You
really, you can't say that it is very limited, because the
possibility is that you could stray quite a bit from a good
configuration if you don't have a good PRA model. So we
would say this is an example where you really have to have
high consequences -- high confidence, rather, in your
analysis methodologies. You cannot limit -- you can limit
the risk, but it is very difficult to limit the risk, and it
is very difficult to monitor the performance in a timely
fashion. So that is an example where you really have to
have consequence -- high confidence, I keep saying high
consequence, high confidence in your methods. And not only
your PRA, but your risk management method.
The second example is risk-informed ISI. Having
done the pilots and having done the topicals, the
Westinghouse Owners Group topical and the EPRI topical, we
have come to the conclusion that the exposure, the risk
exposure of this type of an application is not that high.
What you are basically doing is you are using risk as one of
two criteria for sampling the piping that you are going to
look at to try to determine if you have a problem with the
welds. And it is good to choose piping in high risk areas,
it is also good to use the other criterion, which is to
sample piping in areas where you expect to see degradation,
but you are not very vulnerable to any error that you might
make in selecting those areas. So that is --
DR. APOSTOLAKIS: Why is it high?
MR. BARRETT: We are saying that there is a high
confidence that we can accept this application based on the
low risk exposure. Okay. And we are not --
DR. APOSTOLAKIS: It means that you have high
under tech spec and high under risk-informed ISI.
MR. BARRETT: Yeah, unfortunately, what that
should say in the first line is "low."
DR. APOSTOLAKIS: Okay.
MR. BARRETT: All right. Sorry. In our haste to
make the slide, we got it wrong.
DR. APOSTOLAKIS: No, that is fine, as long as you
tell us what it should say.
MR. BARRETT: It should say "low," "high" --
DR. APOSTOLAKIS: Okay.
That's a case where you are defending -- well, we
have made that conclusion and we are not going to be asking
licensees for tons and tons on --
DR. APOSTOLAKIS: As I recall the analysis,
actually, if you put there "nonexistent" that would be okay
too.
MR. BARRETT: And the third example are the steam
generator tubes where you know that you cannot say that you
have high confidence that the risk is low because you are
talking about something that represents two of the barriers.
The potential is there for high risk if you don't
have a good handle on steam generator tube degradation so
you are either going to have to have very high confidence in
your analysis methodologies or you are going to have to
place confidence in your ability to monitor performance in a
timely way.
DR. APOSTOLAKIS: So tell us what the correct
words are. Risk limitation is what?
MR. BARRETT: The risk limitation is low.
DR. APOSTOLAKIS: Analysis, medium?
MR. BARRETT: We think that in this particular
case, if you have been following, and I know you have, NEI
97-06, we are putting our eggs in the basket of performance
monitoring. We think that if you understand what the
licensee finds at the end of a cycle, and then there's
timely feedback into their ability to predict what is going
to happen in the next cycle that that is where you are
getting the risk limitation.
Without going into a great deal of detail, what we
have here is three different examples of risk-informed
applications where the confidence that you gain is from
three different parts of the triad but in most cases we
tried to work about six or seven other examples and none of
us could agree where to put the highs and lows and mediums,
so some of the other applications are not quite so simple.
The reason they are not so simple is because they
can interact, because you can, if your PRA isn't good enough
to do what you want to do then you will limit what you want
to do, so there are tradeoffs.
CHAIRMAN POWERS: I find these examples
illuminating. I hope in your document you are going to have
several others in there, and it is illuminating in that it
tells me the kinds of applications where you think you are
going to be getting risk based applications in and it adds a
little more meat to the concept of integrated decisionmaking
that you are not just relying on the PRA to give you an
answer, so I find this a useful slide.
MR. BARRETT: Thank you.
DR. APOSTOLAKIS: A low risk limitation means that
the risk is potentially high?
MR. BARRETT: Right.
DR. APOSTOLAKIS: Okay.
[Laughter.]
CHAIRMAN POWERS: You might work on your risk
communications a little bit.
[Laughter.]
MR. BARRETT: I have got to work on these two
slides. That's all.
DR. KRESS: I think we had something like this,
George, in our joint letter where we said that if you have
an application where the inherent hazard or risk is fairly
low that you don't have to have that good of a PRA and you
could rely mostly on just performance and looking at how it
goes, but if you had inherently a pretty high risk then you
better have a pretty good risk analysis and your performance
monitoring might --
CHAIRMAN POWERS: I am still struggling on when
can I look at just the frequencies of accidents and I don't
have to look at the product of those frequencies and
consequences.
DR. APOSTOLAKIS: No, you always look at both.
CHAIRMAN POWERS: There must be some probability
that I don't have to --
DR. KRESS: I am sure there is, Dana.
DR. APOSTOLAKIS: The consequences are huge -- no,
no, the whole business here is called low probability, high
consequence events. I mean you have to look at both.
DR. SEALE: PTS comes close to that, Dana. You
said that it is so low that you don't --
CHAIRMAN POWERS: That's a good point --
DR. APOSTOLAKIS: But you know the consequences.
DR. SEALE: Not in their glory, you don't.
CHAIRMAN POWERS: You have a worst case.
DR. SEALE: Yes, that's right.
DR. WALLIS: The problem I have, Rich, with this
very nice matrix, nine entries in it, is what does this tell
me about the PRA quality subject, and I thought it was in
the analysis somewhere, but that is only part of the
analysis. I don't know what this tells me about PRA
quality.
MR. BARRETT: What this tries to do is to remind
everyone that when you look at PRA quality, and Mary is
going to talk to you in a lot more detail about PRA quality
itself, but you need to look at it in the context of other
factors. That's basically what it does.
DR. WALLIS: Does it mean that if I have a low in
analyses I don't need a good PRA?
MR. BARRETT: It may very well be that you don't
need that -- well, from my perspective as a manager in NRR,
it may well be that I don't have to concentrate the kind of
resources on the review of that PRA that I might otherwise
have to do.
DR. APOSTOLAKIS: Maybe using the word "quality"
is not appropriate here. It seems to me we are talking
about scope. You can't mean that you will accept something
of poor quality because its significance is not high. That
doesn't make sense.
You are accepting something that is of smaller
scope. Maybe they don't do uncertainty analysis.
MS. DROUIN: Right.
DR. APOSTOLAKIS: Maybe they don't do the human
error very well, quantification, otherwise I agree with Dr.
Wallis.
MS. DROUIN: We always expect them when they --
DR. APOSTOLAKIS: I mean you can't say the quality
is poor but I accept it anyway.
MS. DROUIN: We expect them to do their arithmetic
correctly.
DR. APOSTOLAKIS: Of course. It is the scope that
is different.
MS. DROUIN: Correct.
MR. PARRY: Quality has to be appropriate for the
use that is made of it in this decision.
DR. APOSTOLAKIS: But when you say quality you
really mean scope, I think. You can't mean that they are
allowed to miscalculate things --
MR. PARRY: No, they have to follow certain basic
rules but it's scope, level of detail, level of
approximation.
DR. APOSTOLAKIS: That is the scope.
MS. DROUIN: For the scope that applies to the
decision being made --
DR. APOSTOLAKIS: Right.
MS. DROUIN: -- the scope has to be done, the
technical analysis for that scope has to be done correctly.
DR. APOSTOLAKIS: So the quality is still good.
MS. DROUIN: The quality is still good.
DR. APOSTOLAKIS: It is just a different scope.
MS. DROUIN: For the scope that doesn't apply --
DR. SEALE: You can't get blood out of a turnip.
If the information is not in the analysis you
can't make it be there. I mean if you don't know enough to
give yourself useful numbers from a PRA point of view about
steam generator tubes, you can grind those numbers till the
cows come home and you are still not going to get any more
than what you got.
DR. APOSTOLAKIS: I really think that using the
word "quality" the way we have been using it has created
some confusion.
MR. CUNNINGHAM: We agree. I think it is a
shorthand that we use.
DR. APOSTOLAKIS: Scope and detail is really the
appropriate --
MS. DROUIN: As we go forward, you will see that
we have tried not to use the word "quality."
MR. BARRETT: But there are gradations, even in
terms of data. How recent is your data? How plant-specific
is your data? There are gradations in quality and you --
DR. APOSTOLAKIS: That is a borderline issue,
you're right, but I can still call it the scope and detail.
MS. DROUIN: Okay.
DR. APOSTOLAKIS: Ohhhh -- raise it up a little
bit, otherwise we'll gave Gareth's head blocking everything.
MS. DROUIN: I'm sorry.
MR. PARRY: You need to raise the slide a little
bit.
MS. DROUIN: Better?
DR. APOSTOLAKIS: Yes.
MS. DROUIN: Okay. In looking at the technical
acceptability of the PRA that is going to support your
decisionmaking process, I think there is a process for
determining what that technical acceptability is. That is
what is shown here in that shaded box.
You are trying to get risk insights out of the
PRA, this process to support your decision, and there's
steps here going through starting off with what your PRA is
and then looking at it and determining what is the
appropriate scope that you need to support those risk
insights.
You are either going to be in scope or you are
going to be out of scope. If you are in scope, then what
are the elements in the characteristics that are needed to
support that scope and then imposing a peer review to
confirm that technical acceptability, and then coming out of
the peer review and you look at the results from that if
they are acceptable you are going over to an expert panel to
integrate your risk insights.
If it is not acceptable then you are going to go
through some decision process on what to do with this
unacceptability and the unacceptability might be an
insufficiency or it might be a difference or it might be
something missing, because you can also get there because
you are out of scope, but you would have some kind of expert
panel input there too, and then you would go also back to
the expert panel to integrate your risk to come through --
to generate your risk insights.
I kind of glossed over this quickly but I am going
to go through each one of these boxes. I am just kind of
laying the stage here to show you how it all fits together
and hopefully as we go through these individually the flow
chart will start to gel and make some sense.
So if we start at the beginning we first lay down
independent of the application at this point, what is the
scope and level of analysis for a PRA. You want to call it
a PRA. We're saying here is what the scope and level of
analysis is and starting off with your plant operating
states the PRA is going to look at your full and low power
and your cold and hot shutdown. You want to capture the
entire risk.
On your initiating events, you want to consider
both internal and external events.
DR. APOSTOLAKIS: Let me understand this. All the
time you have to do this?
MS. DROUIN: No.
DR. APOSTOLAKIS: Oh.
MS. DROUIN: No.
DR. APOSTOLAKIS: Okay.
MS. DROUIN: We are going to get back to that. We
are just saying right now if you call something a PRA this
is what we are calling a PRA and the PRA is going to
characterize your risk and we are looking at Level 1, core
damage frequency, Level 2, LERF and including late
containment failures, and we are not including Level 3 in
defining what we mean by a PRA.
DR. APOSTOLAKIS: So you are beginning now to put
fires under internal -- that's very good.
MS. DROUIN: We have been doing that for a long
time now, George.
DR. APOSTOLAKIS: Really?
MS. DROUIN: Yes.
DR. APOSTOLAKIS: But you kept it a secret from
me.
MS. DROUIN: Well, I didn't mean to.
DR. SEALE: Earlier today we'd been talking about
our frustration about not having the consequences included
in the result of an assessment or that we didn't carry the
so-called risk determination to the point of assessing
consequences.
It would appear that you are adopting that
practice as standard.
MS. DROUIN: I will say what you are seeing now is
preliminary, you know, and that is why you do not have, you
know, the attachment to the SECY, because we are still
formulating, we are still going through and discussing among
ourselves.
DR. SEALE: Yes. The reason I mentioned it is, as
I said earlier, we had some frustration with the fact that
consequences were not a product of some of these so-called
risk assessments and so this may very well be a concern we
would have with this definition of what you have indicated
on the risk characterization. You really haven't come up
with consequences.
MS. DROUIN: Correct.
MR. BARRETT: If I could say a word about that,
this is something that we are grappling with right now in
the context of Option 2 and in the context of the pilot for
Option 2.
That is, how do you categorize equipment that has
no impact on CDF or LERF, equipment such as containment
sprays, containment of fan coolers, filter systems,
ventilation systems, and as you get to Option 3 things like
containment leakage requirements and in fact we have already
been dealing with these in license amendments.
I think I would agree with Mary. I don't think
that the right way to do that is to go to a full Level 3
PRA. That is a lot of effort and expense to try to deal
with that issue, but we are trying to figure out a way of
dealing with that issue.
DR. SEALE: But you can't ignore it though.
MR. BARRETT: We are trying to figure out a way of
not ignoring it but at the same time not placing a burden on
the PRAs that I don't think they can meet right now.
MR. SIEBER: Those are really defense-in-depth
kind of features for the most part.
MR. BARRETT: Yes, they are, but if you play the
defense-in-depth card and you say because of defense-in-
depth I need all of this stuff and I need all of it to be
gold-plated, you are not putting it on the same footing as
everything else. We are trying to come up with a way to put
it on a similar footing so that you have some criterion that
you can use that is comparable to CDF and LERF to make these
decisions.
MR. SIEBER: Would you postulate that perhaps some
defense-in-depth features would then disappear because they
don't have an impact?
MR. BARRETT: I think whatever decision you made
you would want to keep into account defense-in-depth.
For instance, suppose a licensee came in and said,
well, you know, we think we can live with higher
consequences offsite from a TMI type of accident. What
would be the considerations?
The considerations might be reductions in your
requirements on mitigating systems like sprays. It might be
a reduction in your requirements on the leak tightness of
the containment, or it might be a reduction in some of the
gas treatment systems. In a sense, you might want to make a
decision like that considering the defense-in-depth, so you
might not say, well, I am not going to take away this or I
am not going to take away that, but I might allow you to
relax some of those, so you still have defense-in-depth but
you have used some criterion to accept some relaxation.
MR. SIEBER: But in the case of containment
leakage the probability doesn't change. The consequence
changes.
MR. BARRETT: Exactly.
MR. SIEBER: And so if you don't look at Level 3
in the consequences, it doesn't show up.
MR. BARRETT: We try to deal with it qualitatively
or at least --
MS. DROUIN: This was a very poor choice of words.
When we were saying not required, it did not mean to imply
that we weren't concerned about the consequences.
All we were trying to day by that is that we were
not going to require a Level 3 PRA, so that was very
misleading there.
MR. SIEBER: But the consequence to the public
doesn't appear in either CDF or LERF either.
DR. SEALE: Have you entertained the idea of a so-
called standard cite?
MR. BARRETT: He has gone that far. You know in
past applications, let's take the example of the
Decommissioning Technical Report. What we tried to do there
was to look at a couple of types of sites, a high population
site. We picked a specific site. We postulated a site where
the uniform density -- you know, we tried to do it on a
generic basis without going to the extent of asking every
licensee that decommissions to have a level three PRA.
MR. WALLIS: Richard, your subject is PRA
qualitative. If I look at this, I say, "Well, how does this
help me access PRA qualities?" I look at this thing and I
say "Well, this PRA that I have here doesn't have
(inaudible) in it and I think it should have, therefore, it
is of low quality" just to live a check list. If I need to
evaluate for full quality purposes or how does it --
MS. DROVIN: It depends on the application.
MR. WALLIS: How does it affect my decision about
whether or not this PRA is of sufficient quality.
MR. CUNNINGHAM: In a sense yes, though. It is
kind of a first screen. You are expecting to see initiators
of these types and if you get a PRA submittal and a missing
logo or something like that. It's again --
MR. WALLIS: Then I would want to know why.
MR. CUNNINGHAM: And is it important to the
application.
MR. APOSTOLAKIS: One question.
MS. DORVIN: We're going to have a slide on that.
MR. APOSTOLAKIS: Is level two really this
definition? I thought level two was the source there;
wasn't it?
MR. CUNNINGHAM: Yes, that's right.
MR. APOSTOLAKIS: So this is a variation.
MR. KRESS: That's a one plus.
MR. CUNNINGHAM: This is Level one and a half --
MR. APOSTOLAKIS: This is a level one plus.
MR. KRESS: That is what we call it. Yes.
MR. APOSTOLAKIS: Well, the Germans call something
else level -- This is new level two perhaps, I don't know.
MR. KRESS: (Inaudible), maybe.
MR. APOSTOLAKIS: Level two minds.
MR. KRESS: It looks at containment cellular
level.
MR. APOSTOLAKIS: Just containment, not the actual
(inaudible).
MR. KRESS: It is not exactly true that this
doesn't have consequences in it because LERF is a measure of
consequences.
MR. CUNNINGHAM: LERF. Sure.
MR. KRESS: If you have it there.
MR. PARRY: Can I just add something to respond to
Dr. Wallis. I think the way you have to look at this. This
is really just a bit of semantics. It helps you define what
your PRA contains. So when you come to judge the quality of
the PRA, I think you have to judge it in terms of the
quality given the scope of the PRA and given the results
that it is providing for the decision making. So this is
just really -- you don't say that a PRA is of low quality
because it has low power and shut down, because it doesn't
have low power and shut down in it. If you recognize that
one, you're making a decision, and you compensate for that
then you have used the results appropriately.
MR. WALLIS: So it doesn't meet the
specifications. I would put it that way. Do you need the
PRA's for some purpose?
MR. PARRY: I think what the decision, like for
example, if we were doing something according to REGI 1174,
what it says is that you look at all contributions to risk.
Okay, low power and shut down on internal and external
events. If you happen to have a PRA that doesn't cover all
those things, then you would have to limit the role that the
PRA analysis plays in the decision making. But within that
context, you can still ask the question of the PRA for the
use that it is being made of in the decision. It's no
reflection on the quality of the PRA that it does not have a
full power -- a low power and shut down portion to it.
MR. SEALE: That's why it's --
MR. POWERS: It seems to me --
MR. PARRY: That why I think we've talked about we
don't want to (inaudible) quality in a bad way.
MR. POWERS: It seems to me that you are working
here at a fairly high level and I might, as a commissioner,
say it doesn't really matter about this stuff, none of the
PRA's have most of the things in them anyway. Tell me about
the ones that we actually have which are full power PRAs and
what you are requiring there. And I might ask questions
like: Is it all right for you guys to get a PRA that works
only at the system level, or does it have to go to the
training level or does it have to go to the component level?
Are you going to address those kinds of questions?
MS. DROVIN: Bear with us.
MR. BARRETT: Yes.
MS. DROVIN: I mean we are just starting at the
beginning first saying --
MR. POWERS: I understand.
MS. DROVIN: -- saying what the scope of the PRA
and again, we are not saying that for every application the
scope needs to be this. Remember I said this is independent
of the application. We're just saying --
MR. POWERS: This is my checklist on what it's
gotten.
MS. DROVIN: For a PRA this is the scope we are
looking at. And we wanted to define --
MR. BONACA: I have a question about-- I am sorry.
MS. DROVIN: -- the details for this scope. Yes.
MR. BONACA: Do you consider (inaudible) accidents?
MS. DROVIN: Yes. But we're not covered here. If
you went back, we're just talking about reactor activities.
MR. BONACA: But you are going to have that
included?
MS. DROVIN: At some point, yes. Right now, I
mean, Tom if you wanted to address that --
MR. KRESS: No. No.
MR. BONACA: No, no. I just am saying that, you
know, I mean, I know that a lot of PRAs do not include
(inaudible) pull accidents, since we've been talking about
it.
MR. POWERS: I don't know of too many that do.
MR. KRESS: That's right.
MS. DROVIN: No. We're strictly focused on the
reactor right now.
MR. PARRY: Right now this is the reactor.
MR. BONACA: And since you are looking at full low
power and shut down, and you have operations there that
include transfer to the pool, things of that kind, I would
expect that would be part of the scope of a complete PRA.
MS. DROVIN: Ultimately, yes. But right now we're
just looking at the reactor part.
So once you've established the scope, given that
scope, we go to the next level and looking at the results
that we are trying to use in the decision making process
defining what those -- what are the results that we want to
get out of this scope of a PRA such things as a core damage
frequency; large early release; identifying what our
dominant accident sequence is; having an understanding of
these. So once you have laid out what are these results
that you are trying to get out then to specify for the PRA
what would be those elements that you need and then what
would be the characteristics and attributes of those
elements.
So this is the kind of level of detail. We gave
you some examples here of what we are going to be getting
into. Where we will lay out all the elements and for each of
those elements what are the characteristics.
MR. POWERS: Let me ask you a question, and maybe
it is just poor understanding on my part. Come along and
you say, our initiating events and it says, "It has to be a
thorough identification of the initiating events." I presume
that's what you mean. And, I think it is pretty clear that
none of us could make a thorough identification of the
initiating events, or said more fairly, that if you came in
with a listing of initiating events, I could always define
an initiating event that you had not included in your list.
MS. DROVIN: Okay.
MR. POWERS: And so it wouldn't be thorough. There
must be a different definition of thorough here.
MS. DROVIN: Again, I am not trying to say that
this is what you need for every application. We are just
laying out for a PRA given that scope, given that you want
to produce a core damage frequency and you want to produce,
you know, what your dominant accident sequences are, I mean,
I just gave you some examples of the results, these are the
characteristics and attributes that you would want for these
elements.
MR. BONACA: I am having trouble with this in terms
of quality.
MR. POWERS: So what you are saying really is not
thorough but inadequate characterization of the initiating
event.
MS. DROVIN: Probably a better choice of words.
MR. POWERS: Yeah.
MS. DROVIN: So then when, for example, when we go
to review the certification process or we go to review the
ASME standard we would be looking at did they achieve this;
does the standard, when we look at the requirements that are
in the standard will it give us an adequate identification
and characterization of the initiators.
MR. POWERS: Suppose I come into you with my PRA
and I say, "Look, I can't get a best estimate analysis code
through Graham Wallis' committee. He doesn't like anything
I produce. So all I did was come in with bounding
conservative estimates on the success criteria." I know I'm
conservative on these things, they're not best estimate. I
mean, is there anything wrong with that? He's penalizing
himself. Didn't have to be best estimate.
MR. CUNNINGHAM: It can be because depending on the
application what is conservative and bounding in one
application may not be in other application.
MR. POWERS: He has been very careful. He is a very
careful guy. I mean he's smart enough to figure out what is
conservative and what is not conservative on the outcomes.
MR. CUNNINGHAM: He's a very bright person if he
can do that.
MR. POWERS: He's not dumb. He knows he can't get
things through Graham.
MR. BARRETT: But if it materially effects the base
line core damage frequency and lower frequency, it could put
him outside of the acceptable ranges in REGI 1.174. So at
some point that licensee is going to end up of putting
himself out of the range where we could approve anything.
MR. POWERS: Might want to come in if he done
that, I suppose. I mean --
MR. APOSTOLAKIS: Yes, bounding on our leases
would probably play a role in some place here (inaudible).
MR. PARRY: I think part of the answer to that is
recognizing that it is a bounding analysis and it comes into
deciding how you cope with that uncertainty in the decision
making. I mean we can't say it as a general rule but I
think that is where you would have to address these issues.
I mean in principle there is nothing wrong with it, but as
long as you recognize what it is and what it is doing to the
rest of your decision.
As Mark said what it could do is it could obscure
some other things by raising --
MR. POWERS: Very often true.
MR. PARRY: -- raising certain frequencies of
sequences that, you know, obscure others.
MR. SEALE: More than that if it is a bounding
calculation, you have given up the capability in your PRA to
make a value judgment between --
Mr. SHACK: Discrimination.
MR. SEALE: -- two alternative ways of doing
things in terms of their risk consequences.
MR. BARRETT: Exactly.
MR. SEALE: Which is the problem we have had from
the beginning.
MR. APOSTOLAKIS: This slide does not say anything
about uncertainty or not. Is that some where else?
MS. DROVIN: What I was saying its buried in here
in the sense that we haven't -- this table here is about
three or four pages long.
MR. APOSTOLAKIS: Oh, you are just showing part of
it?
MS. DROVIN: I'm sorry, I should have pulled out
the one that showed the uncertainty analysis.
MR. APOSTOLAKIS: Okay. Now --
MS. DROVIN: I am just showing you some examples
here.
MR. APOSTOLAKIS: Yeah, but all the examples really
refer to the construction of the (inaudible) sequences. That
is why I was wondering. You are talking about adequate
identification of initiators put in the hardware but I --
MS. DROVIN: But all of the elements for --
MR. APOSTOLAKIS: I think it is time that we --
MS. DROVIN: If I go back here --
MR. APOSTOLAKIS: No. I believe you.
MS. DROVIN: I mean we have elements that cover all
of this.
MR. APOSTOLAKIS: I believe you, Mary.
MS. DROVIN: You will see.
MR. APOSTOLAKIS: I want to make another point
here.
MS. DROVIN: Okay.
MR. APOSTOLAKIS: That it seems to me that the
uncertainty analysis that most PRA do, on federal rates and
things like that, is really the easy part. The most
difficult one is the modulate certainty issue.
MS. DROVIN: Absolutely.
MR. APOSTOLAKIS: And maybe it is time that we
started
emphasizing that. And at the beginning there will be some, you
know, crude approaches perhaps, qualitative waving your arms
and so on, but I think it is time we started emphasizing
this that unless, in fact, coming back to the area of
discussion, you know, can you really do without uncertainty
analysis; yes. Now what is it going to opposite the order
ranking of the accident sequences; not a different value of
the failure rate. It is a model thing. If you have missed
something; if you have mismodeled something. So is it a
word like mismodeled?
MR. POWERS: If there isn't there ought to be one.
MR. APOSTOLAKIS: If there isn't there should be
one.
MS. DROVIN: One of the elements that we have,
George --
MR. APOSTOLAKIS: But I really think we ought to
start emphasizing that. Recognizing that, you know, the
methods, right now to handle it may be are not the best they
need some proven. But we really have to do this.
MS. DROVIN: I agree. When you see the full
document what you will see is one element that we call
interpretation of results. And it get into what you do with
the uncertainties.
MR. APOSTOLAKIS: Good. So put it up front.
MR. PARRY: Can I come back to this quality thing.
I am still struggling with quality. Because it seems to me
you can have all these list of things you have in these
slides here, and then if I -- let me give an example, I only
think in terms of examples. If I'm thinking of say the
quality of an automobile, there is brown A and there is
Brown B, not to mention Rolls Royce and Ford and all those
sort of things. And they all have steering wheels; they all
have wheels; they all have brakes. They all have these
things all these things you listed here all PRAs have. What
is it that makes one better quality than another?
MS. DROVIN: Okay, it is not our intent to write in
essence a standard. What we are trying to do here is at a
high level put the frame work down and we were going to rely
on either the standard or the certification to come in and
tell us how have they done these things. The detail should
be there. We are using --
DR. WALLIS: But how do you judge quality if your
quality -- that's the thing -- At the very beginning the
question was asked, "how do you define quality." I still
don't see it.
MR. APOSTOLAKIS: I think you would go a long way
towards answering that question if you selected a few PRAs
or IPEs from your wonderful report covering a spectrum of
quality as it was judged at that time and go through this
process and say, "Look, this is what we decide that this
particular PRA or in this particular issue they did the
whole job." Because then you will start answering Dr.
Wallis' --
DR. WALLIS: And then these sorts of things answer
the questions.
MR. APOSTOLAKIS: This is too high level.
MS. DROVIN: And it is meant to be high level.
MR. APOSTOLAKIS: No, I understand. But I think
giving a few case -- like Rich did earlier with risk
conforming, risk conforming situations. I think you had some
good examples there for human liability, for examples, "boy,
these guys really did a lossy job." If you can tell us why
you judged that and take out the criteria you used and put
them in here I think that would go a long way towards
answering these questions.
And then, "the other guy did an excellent job,"
well, why?
Mr. POWER: If I could just parathentically say
when you look at the NEI 0002 document, you're going to find
high level questions like this analogous to this. But you
are also going to see sub-tier criteria of things that get
down deeper and ask more detailed questions. If you are
reviewing an application, as I said earlier, the first
decision you have to make is how deeply are you going to go
in a particular area in the review of the particular
question.
MR. APOSTOLAKIS: Just a minute.
MR. Power: -- of the type that Mary has here. And
that is going to depend on what you see also.
MR. PARRY: I think quality maybe defined in terms
of what this garry will do, not what is in it. That goes
back to the very beginning.
MR. APOSTOLAKIS: To the decision, yeah.
MS. DROVIN: And we're going to get to that. But I
am saying if you just had the PRA and you're trying to judge
it based on this thing here and let's use the ASME standard
as an example. And if I go in and I'm looking at the
initiating event and that there is no requirements in there
that tell me how they do an adequate identification and
characterization of initiators, they are going to flunk
that.
MR. APOSTOLAKIS: Right. At this level, yes.
MS. DROVIN: Because I need that --
MR. APOSTOLAKIS: You're right.
MS. DROVIN: -- to get a technically acceptable
PRA. So, I am going to be looking for requirements and all
those requirements of sufficient detail to assure that.
Mr. APOSTOLAKIS: I guess Dr. Wallis' question,
some of mine go beyond this. You are absolutely right.
Now this is page 15 of 19 and it is 4:05, for
twenty-five minutes.
MS. DROVIN: We can do it. Trust me.
MR. POWERS: The question is whether we can do it,
either.
MS. DROVIN: I'll talk about us here.
But the next slides, you will see kind of a
similar format. Once we have established, you know, we say
we have this scope that we have established and these
characteristics and attributes that we are going to look at,
how do you go about confirming that you've achieved those?
One way is through appear/review process. You can use that
to confirm your technical acceptability. So now if someone
chooses to go that route, so when we look at the NEI
certification there will be certain attributes just to that
process that we would be looking for such that that process
is an adequate mechanism for insuring your technical.
So again, here are some examples where the
elements we would be looking for better talked about the
team qualifications; the peer review process; the
documentation; and then for each of those laying out what we
would think would be the characteristics and --
MR. APOSTOLAKIS: You know team qualification is
something that I think is utterly misleading.
MS. DROVIN: I'm sorry.
MR. APOSTOLAKIS: Usually, the team qualification,
usually it is in terms of how many years you've been in the
business, right. You could be in the business -- you could
be for twenty-years and be consistently wrong.
MS. DROVIN: I don't think (inaudible) is.
MR. APOSTOLAKIS: I mean it's not a problem. I
think the process it really what matters in qualifications.
Conflicts of interest, I mean, yeah, right.
MR. PARRY: That may exclude anybody that knows
anything valuable.
MR. APOSTOLAKIS: By the way you don't have
twenty-five minutes because Mr. Bradley will have a few
minutes.
MS. DROVIN: Okay.
MR. APOSTOLAKIS: Okay, we understand where you are
coming from.
MS. DROVIN: The next one gets into the heart, in
some sense, because we do recognize that not every
application needs to meet that scope or all of those
elements and attributes. So how do you go about deciding
what needs to be in scope; what needs to be out of scope
such that you have technically acceptable PRA and you can
have confidence in the results you are going to be using to
generate your risk insights.
MR. SEALE: So you're going to housesit us?
MS. DROVIN: So there is some, a decision process,
that you can go through. We've tried to layout what we
think are the necessary elements of this process and then
for each of the elements what would be the characteristics
and attributes for each element.
Then the last piece is getting to the expert
panel. The expert panel review has a potential to be used in
two ways. First taking your results to integrate them into
the decision process, integrate your risk insights. The
other thing is that when you look at your PRA and when you
are out of scope or say you're in scope and you haven't done
your arithmetic correctly, or you have some difference or
some deficiency and you are going to use your extra panel to
make up for that.
MR. APOSTOLAKIS: You know we are doing a lot of
that, and I don't like it.
MS. DROVIN: Excuse me.
MR. APOSTOLAKIS: Every time a method is found to
have problems we say, " Why don't the expert finally take
care of that?" So I would be very cautious --
MS. DROVIN: Oh, I agree.
MR. APOSTOLAKIS: -- using that kind of argument.
I mean, if the importance measures have a problem why don't
the expert find out what to do about it. We miss something,
ah, let the expert find it.
MR. SEALE: By the time you get the well developed
PRA, you may even have some expert panel members that are
truly qualified.
Mr. Shack: Well, you might.
MR. SEALE: And their related process. They are an
inverse process.
MS. DROVIN: So again, given that those are the two
ways that an expert panel can be used, we are trying to
identify what would be the necessary elements of an expert
panel considering those two applications and then what would
be, again, the characteristics and attributes, and again,
these are just examples this is not complete just to give
you a flavor of where we are going. And those are all the
pieces that fill the (inaudible)
MR. APOSTOLAKIS: I think it would help you a lot
and us for sure, if you actually -- if you would set an
example from the ideas that you are very familiar with. Go
through them; apply your criteria here and say, "Well, gee,
my decision at the time was that this was a poor analysis."
"Can I conclude that by strictly applying what I have here
or do I have to expand this?" I think you are going to
learn a hell of a lot by doing that, and then you are also
going to convince others. Because I remember there were
some outlines, right?
MS. DROVIN: Oh, yes. We can walk you through some
examples.
MR. APOSTOLAKIS: You came in here you gave us
reasons and so on. Are these judgments, you used at the
time, do these judgments flow naturally from what you have
here or do you have to add something; and then you will be
able to answer Dr. Wallis' concerns, I think, much much
better. But also you will have to select some good ones.
MR. CUNNINGHAM: Yes, that's right. You'd have to
pick some --
MR. APOSTOLAKIS: Show that, you know, yeah, we
decided that this was a very good one because this and that.
MR. CUNNINGHAM: That's right.
MR. APOSTOLAKIS: Because the problem is the
(inaudible) presentation not the high level stuff. So I
would like to see that and it shouldn't be hard for you to
do especially, Mary.
MS. DROVIN: Yes, we've gone through several so we
can do that.
MR. APOSTOLAKIS: Okay.
MS. DROVIN: Our last slide is (inaudible).
MR. APOSTOLAKIS: It's so sketch.
MR. SEALE: It's sketch.
MR. APOSTOLAKIS: I can't believe that you are
talking about the weeks, usually here it is years.
MR. CUNNINGHAM: Well, sometimes you have
schedules opposed upon you. At any rate, the bottom line on
this is that we got two or three items facing us right now
that are all related to this issue of PRA quality, if you
will. How are we going to review the ASME; how we're going
to review ANS; how are we going to review the NE document,
any "I" documents in the context of option two.
What you heard about is a brief description of how
we are going to take that on in a more integral way. Our
goal is in the next couple of weeks to put together a
commission paper that lays this out. We are going to be
discussing with the PRE Steering Committee in the middle of
June. The idea is to have the paper to the DDO on June 27th.
MR. APOSTOLAKIS: What is RILP, R-I-L-P?
MR. CUNNINGHAM: Risk Informed Licensing Panel.
That is a panel of, basically, division directors but some
others. Tom is the vice-chair, Gary Hollihand, is the
chairman. It is used to air issues among the various
organizations in NRA on risk informed issues, basically.
MR. APOSTOLAKIS: So they are risked informed in
the since that they understand PRE?
MR. CUNNINGHAM: They bring together all of the
disciplines that are (inaudible), yes. Believers and
skeptics all brought together to hammer out the issues.
MR. APOSTOLAKIS: Are you asking us to write
anything?
MR. CUNNINGHAM: No, we are not asking you to
write a letter at this point. That you have nothing, as you
said at the beginning --
MR. APOSTOLAKIS: That's right.
MR. CUNNINGHAM: We have nothing -- we've provided
you nothing so it wouldn't be fair to ask to write a letter.
MR. APOSTOLAKIS: Would you like us to write one in
July; or you really don't care?
Mr. King: Yes, we would like you to write
something in July. This is Tom King.
MR. APOSTOLAKIS: Oh, you would?
Mr. King: Yes.
MR. CUNNINGHAM: Yes.
Mr. King: It is unfortunate do to the schedule we
didn't have time to get you something in advance.
MR. APOSTOLAKIS: That's all right.
Mr. King: But, you know, consider this in effect
what we are writing is a high level standard review plan
that the staff is going to use to assess the various things
that are on our plate that deal with PRE quality. And I
think when you --
MR. APOSTOLAKIS: That is all we will have to
review then (inaudible) schedule for (inaudible).
Mr. King: I think a letter in July would be very
useful.
MR. APOSTOLAKIS: Anything else, lady and
gentlemen?
Mr. King: Thank you for your time.
MR. APOSTOLAKIS: Well, maybe we can give Mr.
Bradley a few minutes and thank you very much. Very cordial
as usual.
Mr. King: Thank you.
BIFF BRADLEY
MR. APOSTOLAKIS: Welcome back, Mr. Bradley,
MR. BRADLEY: Back again. I am Biff Bradley. I am
with the Indiana staff in their Regulatory Reform area. I
think that I have more time than I need. I don't have a
lot.
Mr APOSTOLAKIS: You don't have to use all of it.
MR. BRADLEY: Good. Well, you know bereavety is the
sole of wit. I concur with, I think, everything I heard the
staff just say and I think, you know, the philosophy that
was laid out, there is no fundamental disagreement with
that.
Mr. KRESS: Is it a lot like the certification
process, the NEI certification process? What they laid out?
MR. BRADLEY: Well, I think what they laid out is
one, it has to be -- as an acknowledgment there will have to
be some NRC review for any application and I view the
standard or certification process as a means to streamline
and expedite that review and whichever one you use.
I also concur that it is like a, you know, trying
to define truth or duty -- trying to define quality in the
abstract of -- it is not a -- I think has created a lot of
problems. In my view, quality has to be defined in the
contents to the application. Maybe suitability is a better
word to be using. And I think, you know, for option two, as
we move along and we understand what the application is and
what is involved, you know, we came to the conclusion that
we could use this certification process for that particular
application and that is what we submitted to NRC with
respect to review for that purpose.
So, we look forward to working with the staff and
working through the issues with the certification process. I
think we've, you know, struggled a lot with the ASME
standard and it has been in the context of trying to write a
standard that would envelope all the different applications
that are going on and that is a very difficult thing to do,
and it is much easier to address a specific application and
come to grips of what you need to do that.
The one thing I would mention, there was a fair
discussion, not a discussion, of IPEs. I think the industry,
we are going to have to take some initiative to update the
data bank that is out there with regard to what has been
done. IPEs are becoming, you know, extremely obsolete.
MR. APOSTOLAKIS: Becoming, Biff. Becoming.
MR. BRADLEY: Well, they have and we moved way
beyond that. But I think it tends to be, that what is sort
of in the public record, you know, and that's what -- we
need to establish a new record that gets us --
MR. POWERS: Well, the IPEs are the sole virtue
and up to datedness compare the IP tripilees.
MR. BRADLEY: Yeah.
MR. APOSTOLAKIS: No. But the reason that he
brought up the IPEs is not to use them in any particular
way, but to actually show what good or bad is in that
context of that particular analysis for plan A. I mean we
are not going to sing loud to anybody.
MR. BRADLEY: Right. So, you know, generally, I
look forward to seeing the commission paper and we are, you
know, facing a lot of challenges in regulatory reform and
despite the amount of publicity that this PRA quality issue
gets, I think there are other challenges in regulatory
reform that are going to be more difficult in having to do
with how you reform the regulatory treatment in light of
these insights and how you measure the impact of that. And
so, I view the categorization of something we can
successfully accomplish.
I think the harder part is going to be dealing
with non-internal events risk because generally you are
going to be using some qualitative, or semi-quantitative
methods there. If it is something we are going to do in the
next, you know, reasonable five years or so.
So, you know, please to hear what the staff had to
say and I think the industry concurs with the approach that
was laid out.
MR. APOSTOLAKIS: I think you would help the whole
thing along if you and the staff agree to drop the word
"quality". Supergroup, perhaps, is a good word. Quality is
really bad.
MR. POWERS: Unfortunately, it is the commission
that is the one that has used the word.
MR. APOSTOLAKIS: Sometimes the commission uses
staff's words. I mean, if the staff comes back and it says
this is the reason. I think the commission will be willing
to listen. You always want quality.
MR. BRADLEY: Right.
MR. APOSTOLAKIS: Quality is always there.
MR. POWERS: Well, I will -- many of you have not
enjoyed the things that Dr. KRESS and I enjoyed during our
careers at the National Laboratory where we learned that
quality doesn't always mean what you think the dictionary
says quality is.
MR. APOSTOLAKIS: [Laughter].
MR. POWERS: We learned definitions like,
satisfying requirements of the customer and any number of
other things that have --
MR. APOSTOLAKIS: To avoid fault.
MR. POWERS: -- a unique definition to the word
quality. So --
MR. APOSTOLAKIS: You satisfies
DR. SEALE: Whimsy is the characteristic of many
of these words.
DR. POWERS: That is why one of the reasons I asked
for the definition of quality at the beginning.
DR. SEALE: Right.
MR. POWERS: Unfortunately, I had been brainwashed
to learn --
MR. SEALE: MR. BRADLEY --
MR. POWERS: -- to give up my knowledge of the
English language and adopt a bureaucratic knowledge of the
word quality.
MR. APOSTOLAKIS: Dr. Seale.
DR. SEALE: You have had input to this process,
clearly, in much the same way the NRC staff has in working
with the ASME and also, I guess, the ANS people who are
working on the other two parts of it. And presumably, but
presumably, in addition to those inputs as you have gone
along, you are going to be doing a similar review of the
final document to, if you will, paint your imprint on this
overall standard in much the same way that the NRC staff has
done it.
Are you going to steal their outline and follow it
when you go home, or if you got another outline of how you
are going to go -- through process you are going to go
through in reviewing the ASME standard?
MR. BRADLEY: If I understand your question
correctly, I think much of that has already transpired in
the writing of the standard. There is a history going back a
couple of years and at various times I think the staff has,
you know, weighed in with their position and then the
industry has weighed in. I think our general sense has been
that a standard should reflect to the degree possible and
appropriate existing practice. And we at the existing
practice in this case is the peer review process that has
been developed and it has been applied. So our efforts as an
industry has been to make the standard comport with that
process to the degree we can and I think it has come a long
way. We are looking at Rev. 12 now, and it will be out for
public comment. And our view, I think a success in this is
if we have a standard and ASME standard that is reasonably
consistent with our process of (inaudible).
MR. SEALE: Thank you. That is a very fair
characterization of your position. I appreciate it.
MR. APOSTOLAKIS: Okay, anything else? Any other
comments?
Thank you, Biff.
MR. BRADLEY: Sure.
MR. APOSTOLAKIS: Back to you, Mr. Chairman
MR. POWERS: Thank you very much.
At this point we can dispense with the
transcription.
[Whereupon, the recorded portion of the meeting
was recessed, to reconvene at 8:30 a.m., Thursday, June 8,
2000.]
Page Last Reviewed/Updated Tuesday, July 12, 2016