461st Meeting - April 9, 1999
UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
461ST ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
11545 Rockville Pike, Room T-2B3
Friday, April 9, 1999
The subcommittee met pursuant to notice, at 10:43 a.m.
DANA POWERS, Chairman, ACRS
GEORGE APOSTOLAKIS, Member, ACRS
JOHN BARTON, Member, ACRS
MARIO FONTANA, Member, ACRS
THOMAS KRESS, Member, ACRS
DON MILLER, Member, ACRS
ROBERT SEALE, Member, ACRS
WILLIAM SHACK, Member, ACRS
GRAHAM WALLIS, Member, ACRS
MARIO V. BONACA, Member, ACRS
ROBERT E. UHRIG, Member, ACRS. P R O C E E D I N G S
DR. POWERS: Let's come into session. This is the third day
of the 461st meeting of the Advisory Committee on Reactor Safeguards.
During today's meeting the committee will consider the following:
proposed ACRS reports; impact of the use of high burnup or oxide fuel on
the revised source term; relationship and balance between PRA results
and defense-in-depth; reconciliation of ACRS comments and
recommendations; report of the Planning and Procedures Subcommittee;
future ACRS activities; and we shall work on reports.
A portion of today's meeting may be closed to discuss
organizational personnel matters that relate solely to the internal
personnel rules and practices of this advisory committee in matters the
release of which would constitute clearly unwarranted invasion of
The meeting is being conducted in accordance with the
provisions of the Federal Advisory Committee Act. Mr. Richard P. Savio
is the Designated Federal Official for the initial portion of the
We have received no written statements or requests for time
to make oral statements from members of the public regarding today's
A transcript of portions of the meeting is being kept and it
is requested that speakers use one of the microphones, identify
themselves, and speak with sufficient clarity and volume so they can be
I think the first item on today's agenda is to discuss an
SRM that we received from the Commission concerning high burnup fuel and
MOX. I have asked Mr. Tinkler from the RES staff -- and is Ralph going
to be here?
MR. TINKLER: I understood he was.
DR. POWERS: I have asked them to attend in much the role of
invited experts to assist us in developing our response to this SRM, so
feel free to come up and join us. I think everybody knows Mr.
Tinkler -- Charlie. I guess I'll have to be wired in or something.
Dr. Meyer has arrived.
You are here in the role of invited experts, so you should
feel free to pipe in, both of you, at any time something moves you to
This is a quotation out of the SRM -- "Consider the impact
on the revised source term if high burnup or mixed oxide fuel were used
in place of standard uranium fuel."
The committee has indeed written on this subject of high
burnup fuel, both on its reactor accidents implications and its source
term implications. So much of the focus of this discussion is going to
be on the subject of MOX fuel I think most of the members are aware that
the Department of Energy is contemplating the disposition of weapons
grade plutonium by converting it to a mixed oxide fuel. Most of the
members I think are aware that the use of mixed oxide fuel is not done
in the United States now, but it is done in Europe and certainly
contemplated in Japan.
It would be a substantial change in the way we have
manipulated fuel in this country to go to mixed oxide fuels. The face
of it, it's not such a heroic change from a technical point of view with
respect to things like neutronics and fuel behavior because in fact fuel
over the normal course of burnup becomes mixed oxide in the sense that
plutonium is generated.
However, it would be a mistake to think that the experiences
with higher burnup fuel that results in the generation of plutonium
yields a product that is identical to mixed oxide fuel.
Many of the concerns associated with mixed oxide fuel arise
out of a concern over proliferation and an independent group called The
Nuclear Control Institute has been in the process of looking in a good
deal of depth at some of the problems that might arise in connection
with the use of mixed oxide fuel and a fellow named Dr. Lyman has done
some analyses to attempt to assess risk.
I caution you that he has not undertaken what I would call a
formal risk assessment. He has gone through and done what I would call
a scoping assessment of risk in a fairly technically sophisticated
fashion. He has certainly used origin codes to look at the inventories
that you would have in fuel and he has used the max code to look at the
consequences should you release radionuclides from this fuel in an
The bridge between inventories and consequences has not been
done in a mechanistic fashion. Rather he has looked at publications
produced by the NRC, at source terms from plants similar to those that
would use the mixed oxide fuel, selected release categories out of
those, and used release fractions applied to the revised inventories.
The changes in inventories that occur in mixed oxide fuel --
clearly there is more plutonium. Plutonium as we know is not high on
the list of radionuclides that pose consequences, but because there is
plutonium in the fuel and some amount of it one does get other
actinides, and the biggest differences that one sees in the inventories,
the ones that are striking, really have to do with actinides, the
actinide americium and curium.
These radionuclides do cause pause. When one goes through
and looks at the consequences of various release fractions of the
actinides, one finds they can be extremely potent if they are released
in any abundance. Typically they are not found to be released in
abundance from fuel, but should they be they have substantial potential
The truth of the matter is we don't know much about the
release of the actinides. They get mobilized from the fuel in a way
that can result in release to the outside poorly, only at the highest of
temperatures or by dispersal events. The NUREG 1150 or the NUREG 1465
type model of the source term really admits to a substantial uncertainty
concerning the release fractions of these.
If people recall the history of that document, release
fractions were specified for the actinides. Based on a lot of
discussion they were subsequently reduced, so there are low release
fractions here but there is a substantial uncertainty.
When one things about actinide releases the experience of
Chernobyl comes to mind. Many people will remind you that something on
the order of three and a half percent of the actinides were released in
the Chernobyl accident. One has to remember, however, that the
Chernobyl accident was a unique experience, a different type of
accident, one that involved a substantial dispersal of fuel, actual fuel
particles mechanically released from the plant. As there were
explosions, things threw the fuel out.
In fact, in a recent review that went back and looked at all
the data that had been collected since the accident it was found that in
every case the actinide release and the fuel release were identical.
That is, the actinides were being released not by heating them up and
vaporizing them but rather they were being released because you were
dispersing fuel, and when I say identical, of course there is some
uncertainty bound but this three and a half percent release means that
only some very small fraction, perhaps a tenth of a percent, could
possibly have been released by a vaporization process that separated the
actinides from the fuel. The NUREG 1465 source term would say the same
thing. It's hard to mobilize these things. They are refractory in
their nature. They don't vaporize very well -- and they did not in the
Chernobyl released those radionuclides because of the
dispersal event. That does not eliminate the --
DR. WALLIS: Would you try and put that in perspective with
the releases of some of the other --
DR. POWERS: Oh, certainly. The releases of the noble gases
of course were virtually quantitative. Releases of iodine, a volatile
radionuclide, was on the order of 67 percent.
DR. WALLIS: So it was a lot.
DR. POWERS: Releases of cesium were on the order of 50
percent. Releases of thorium appeared to be very nearly quantitative.
Release of ruthenium, a metallic fission product that ordinarily is
thought to have a very low volatility, were on the order of 13 percent.
Again this is sometimes attributed to a unique feature of
the accident in that it burst the core fault and exposed this fuel to an
oxidizing environment. I, myself, don't think that is peculiar to the
Chernobyl accident. I, myself, believe that in fact the progression of
the severe accident would expose fuel frequently to strongly oxidizing
environments, but you can see those release fractions, those notable
radionuclides are all much higher than this three and a half percent.
Just because the fuel -- the actinides were released by a
dispersal event does not eliminate them from concern. There are
dispersal type events hypothesized for severe accidents at conventional
In truth, ice condensers may be particularly susceptible to
dispersal events. It is not that you just have a dispersal. You must
have a dispersal and a containment failure in not necessarily
simultaneously but close enough in time that the dispersed material does
not have an opportunity to settle out within the plant before you have
the failure of the containment.
Ice condensers have weak containments. Some of the kinds of
dispersal events that could involve releases of the radionuclides as a
dispersion rather than a vaporization process -- certainly ex-vessel
steam explosions, high pressure melt explosion events are dispersal
DR. FONTANA: Is the concern -- there are several things
that happen in an ice condenser that kind of work against each other.
The ice bays themselves could act as shock absorbers with respect to the
steam explosion and so on, but the high pressure melt ejection or steam
explosion could blow up the ice so you don't have much decay -- what is
that they are concerned about?
DR. POWERS: Typically by the time any of these dispersal
events will occur, the ice is largely melted and you have a large volume
of water down below the reactor vessel.
DR. FONTANA: Okay.
DR. POWERS: Now that of course is good and bad. Water is a
coolant but it is also the medium for steam explosion and it also acts
in a slug capacity.
I think, and maybe one of our invited guests would care to
comment because I think he's actively involved in looking at ice
condensers and dispersal events, some of the concerns of course are the
direct loading you get from events, but I think a lot of the concerns
have come from dispersal of fuel and material directly against the steel
MR. TINKLER: Yes. We are in the process right now of
addressing direct containment heating and high pressure melt ejection
for ice condenser plants.
It turns out that the entrainment of debris into the bulk
containment atmosphere isn't so much a concern for the ice condenser
plants as it is the accompanying hydrogen combustion, which is
affiliated with many of these scenarios because there's station blackout
and there is no control for hydrogen during those events.
The dispersal in our analyses, which is largely a heat
transfer issue, isn't the principal mechanism that is causing the
challenges -- primarily the hydrogen.
Now there is another issue associated with the debris
transport to the containment shell or liner because the refueling --
excuse me, the in-core instrument tunnel exits into an instrument room
which is adjacent to the shell, but that has been considered and there
is some data from past DCH tests on that.
I would also comment somewhat on this general issue because
I do have a view on this and it is that the fragmentation of molten fuel
associated with dispersal into fine particulation it seems to me is a
stronger effect from strong steam explosions than from the high pressure
melt ejection, which is largely an entrainment process. We do measure
fragmentation of debris in the DCH tests and we typically see particle
sizes which on a weighted basis, mass weighted basis, obviously are
You know, we see a millimeter or larger particles. We do
see part of the distribution of smaller particles but we have trouble
measuring particles much smaller than about 50 microns.
The steam explosion tests which have been done in the Faro
and Krotos Program, they look at the post-steam explosion particle sizes
and they do see a distribution of small particles but even in the
stronger steam explosions it is a relatively modest amount of the debris
which is fragmented to the sub-20 micron size.
DR. POWERS: I think it is fair to say that -- well,
historical background, WASH-1400 credited steam explosions to be an
enormously important source term event. They envisioned the steam
explosion producing a lot of very, very fine material that remains
suspended for long periods of time.
Subsequently examinations of that said well, there is a
peculiarity of steam explosions. Yes, they can produce fine material
but they also produce fine water droplets and fine water droplets have
the capacity to sweep fine particles out of the atmosphere, and there
has been a general trend to discount steam explosions as an important
source of radionuclide release in event of a reactor accident even
though ex-vessel possibilities of steam explosions probably have
actually gone up in people's minds. There's just not a source term
High pressure melt ejection -- there's been a limited
examination in the source term characteristics of it, but it is evident
that there are multiple processes going on. There are dispersals that
yield the millimeter size particle events. There are other mechanical
processes going on like the effervescence of dissolved gas out of melts
that create another size particles, and then there is a certain amount
of vaporization processes and entrainment processes.
My only point in this slide is that, yes, you have to worry
about dispersal events when you think about actinide releases.
The Commission's SRM asked us to look at the impact of
higher burnup fuel and MOX. With respect to the revised accident source
term, NUREG 1465 I think a lot of this is just background, that was a
revised source term intended to replace TID14844 for licensing and
regulatory processes. It is to describe a source term for a relatively
It goes through and utilizes the results of a lot of
research sponsored by the NRC and release fractions and release rates
for release from the fuel cladding gap, release during the core
degradation process that is called in-vessel release, release from the
ex-vessel processes, which is largely release during melt-concrete
interactions and something that is called late in-vessel release which
people have generally considered to be the revaporization of
radionuclides piping systems back into the containment atmosphere.
These release rates and release fractions were derived from
a variety of analyses done many of them in connection with NUREG-1150
that are not exclusively for this purpose. Most of those analyses were
done with something called the source term code package, which is a
technology that now is 15, 20 years ago old and is really not used to
any significant extent.
These calculational results were modified by the process of
expert elicitation to develop refinements on the code predictions, but
all of the predictions and all of the analyses were predicated on the
existence of conventional urania fuel and what I would call moderate
I don't think that this is ever specified in any great
extent but perhaps less than 40 gigawatt days per ton were what were
envisioned at the time the source term code package and the expert
elicitations were done.
NUREG-1465 of course does get used in the regulatory
process. It is used in connection with design basis accidents and the
evaluation of engineering safety systems and AP-600 was kind of the test
case for one of the major uses of it. I think it was also used for
System 80-Plus. What they focus on is not the entire release scenario
but rather the release, the gap release and the in-vessel releases.
These are used to assess what the worst two hours, as far as
containment concentrations, are of suspended radionuclides.
What this tells you is it's these early phases that are
going to be important for the applications of NUREG-1150 and that rates
count. Frequently when one things about severe accident source terms,
one says, well, you release the material and I don't really care how
fast it comes out -- it comes out to some extent and that is the number
I need to work with.
Well, rates do count because of this worst two hours.
The question I think the Commission is posing to us are the
extensive release, that is the release fractions specified in NUREG-1465
and the release rates still appropriate for MOX and high burnup fuel.
Again, we have written on this subject of high burnup fuel and I think
our focus is on the MOX fuel.
I have provided you with a variety of documents related both
to high burnup and MOX fuel. I can't claim that this is an
authoritative examination of the literature or even the most salient
papers out of the literature. What I can tell you is this is a
cross-section of documents that I myself found useful in developing my
thinking on the subject of MOX fuel and you have been provided those for
your benefit to examine.
When you think about release from fuel you think in terms of
mechanistic processes now. Fission products get borne in grains of
fuel. They must diffuse through the grain of fuel to a boundary of the
grain. That diffusion process is affected clearly by the size of the
grain. Interestingly, it is also affected by the oxygen potential.
That is, fuel with higher oxygen potentials has higher diffusion
coefficients, and it is very understandable.
It says oxygen goes into the lattice, causes the lattice to
expand, creates a little more room among the atoms so that fission
products can diffuse.
Once a fission product has reached the boundary of a grain,
it has to diffuse to a grain edge and the reason that it has to go to a
grain edge is because it is at the edge that it begins to communicate
with the open porosity in the fuel. How much edge you have and how open
that is to the diffusion paths do depend upon burnup, do depend on
The burnup effects have been considered for a long time in
modern fission product transport. In fact, the old ANS standard on
fission product -- fission gas release from fuel includes a burnup
correction factor, and it is a relatively potent factor. Well, once a
fission product reaches the grain edge it now has to diffuse through the
pore structure. The magnitude of that pore structure is a function of
burnup. The ability of the fission product to diffuse through it
depends on what chemical species the fission product adopts. Is it a
volatile species, for instance a cesium atom, or is it a relatively
nonvolatile species like a cesium uranate?
Once you have gone through the pore structure you reach the
fuel cladding gap and you flow to a breach in the cladding. The
cladding at some point in the accident has actually gone and actually
getting through the pore structure is sufficient to reach the breach and
enter into the bulk flow.
Fission product entering into the bulk flow then goes
through a thermal gradient. It can nucleate to form particles or it can
condense on surfaces.
NUREG-1465, when it specifies release from the fuel -- or
its release fractions, is not telling you how much got released from the
fuel. It is telling you how much got released from the fuel and
successfully negotiated these pathways through the reactor coolant
system to get into the containment, okay, so you cannot compare release
fractions that you measure with fuel in experiments against the 1465
numbers. You have to compare release fractions and transport fractions
to get to those numbers.
These processes of nucleation and deposition depend on lots
of variables, temperatures, chemical speciation, oxygen potentials and
I will attempt to analyze these now in a fairly mechanistic
fashion. There are tools out there for doing this. Those tools by and
large were never developed for heroic changes in the type of fuel we
look at. In general, there was a stock kind of fuel. In general, the
codes have been developed for PWR fuels. They even make scant
recognition that PWR and BWR fuels are different.
They tend not to take into account things like heterogeneity
of fuel due to the existence of plutonium or localized burnup effects or
oxygen potential effects. They tend to be much more correlations of
Sometimes that correlation of experimental data is well
hidden. That is, there is a lot of mechanistic dressing around the
correlation of experimental data. That means that the capabilities to
deal with high burnup are modest. Some of the codes do in fact employ a
burnup correction that looks much like that that is built into the NS
5.4 standard. Some have somewhat more mechanistic descriptions. Most
of those focus exclusively on the fuel porosity -- that is, how much
open space is there among the grains for fission products that succeed
in getting to the grain surface to vaporize and flow through to reach
the outside world.
I don't know of any codes that recognize that burnup affects
oxygen potential. There are some efforts to recognize that interstitial
oxygen affects diffusion coefficients and some modest effort to look at
the effect on chemical speciation as you go to higher burnups.
There have been attempts in the rest of the world to improve
upon these codes, particularly with respect to fission gas and volatile
fission product release. The Transuranium Institute and the Canadians
have been working on fuels that or models that are extremely mechanistic
for the release of these highly volatile materials.
I am aware of nothing going on to improve our models for the
more refractory fission products, certainly not for the actinides, but
not even for things like strontium and barium that we consider to be
Codes have been developed for conventional urania fuel at
fairly moderate burnups. High burnup fuel is just different. I think
that is the thing that caught us by surprise. Maybe Dr. Meyer would
like to comment also on this area.
The high burnup fuel has differences because of
microstructure and oxygen potential. Once you reach sufficient burnup
you develop something that is called a rim region. It is a
microstructurally different region. Whereas normal fuel has grains
typically on the order of 12 to 15 microns in size, the rim region has
grains on the order of 1 micron in size. It is highly porous, a large
amount of porosity into it, and you can see what all this is leading to.
You have got very tiny grains, so fission products borne in
tiny grains have a small distance to move, large amounts of surface area
enter into a highly porous region. They can release quickly from that
rim region into the fuel cladding gap.
DR. MILLER: Dana, what is the physical reason for the
dramatic change in grain size, or is that beyond the scope of this
DR. POWERS: More importantly, Don, it's beyond the scope of
There are attempts to model the development of this rim
region. I have seen some modelled. It is a little difficult to do,
because you can see it is running against surface energy. Surface
energy would make these grains big.
DR. MILLER: Does this go through this transition rather
quickly with burnup or does it kind of move slowly?
DR. POWERS: Why do you ask all hard questions, Don?
DR. MILLER: I guess you are doing your Master's exam here.
I was giving you a quiz.
DR. POWERS: I think that it is fair to say that most of the
thinking is that the development of this rim region is occurring from
the day that you start irradiation of the fuel. The appearance of it,
so that you can detect it microstructurally, is a fairly abrupt event
occurring -- and this has a lot to do with many, many variables -- but I
would say somewhere around maybe 45-50 gigawatt days per ton you start
seeing a clear rim region and it just starts growing on you. These are
kind of round numbers and what-not, because it is a microstructural
DR. MILLER: So if gathered something out of here, at 50
megawatts per metric ton, things are changing dramatically.
DR. POWERS: I think if you will recall some of Dr. Meyer's
presentations before the committee and the subcommittee, he too would
say that somewhere in that region there seem to be some differences
occurring in lots of things.
MR. MEYER: Yes.
DR. POWERS: Feel free to disagree and what-not. It's an
area of some debate.
MR. CRONENBERG: But isn't that pretty small too, Dana? I
mean we are talking about a rim, we are talking about 5 percent of fuel
DR. POWERS: I mean yes. The rim grows and if I get up to
eight -- eighty gigawatt days fuel, the rim is a huge volume. The 50
megawatt day fuel, it's a tiny volume. It's on the outside so it -- as
it grows in width it gets pretty significant.
DR. MILLER: Well, how wide is this at say 50?
DR. POWERS: At 50? You barely see it in a microscope.
DR. MILLER: It does affect the surface dramatically.
DR. POWERS: That's right.
DR. MILLER: That's a key issue.
DR. POWERS: Well, it affects things so you get somewhat
higher fission gas release in that region.
DR. SEALE: Dana --
MR. MEYER: Could I jump in at this point and say that in
addition to this visible rim effect, I think that even in the bulk of
the fuel that you have this growing accumulation of fission gases which
are accumulating in bubbles on the grain boundaries so I don't think
that the rim is responsible for all of the high burnup effects that you
You are beginning to get a microstructure that is almost
linked together by pores on boundaries, even in the bulk, inside of the
DR. MILLER: So suffice to say the diffusion effects, or
diffusion coefficients on fission products changes substantially
MR. MEYER: Yes. If you have a model that characterizes
release by a diffusion coefficient then the effect of diffusion
coefficient is changed substantially.
DR. SEALE: Dana, I seem to remember back in the deep dark
misty past of fast reactor fuel concerns where high burnups were the
rigor rather than the potential for the future that there was a
microstructure re: orientation of the fuel with radially aligned grains
that were zone-refined into the fuel over operation that essentially the
long axis was in the direction of the temperature gradient and all those
other good things.
Is this the prenatal or is the beginning of that?
DR. POWERS: No, I think this an entirely different
phenomena. We have a noted expert on exactly that with us. Ralph Meyer
has cut his technical teeth on some of those issues. He might want to
comment, but I think that really is a temperature issue and this is
really an irradiation issue here.
DR. SEALE: Okay.
DR. POWERS: Dr. Meyer, would you care to comment?
MR. MEYER: No that is enough on that. The LMFBR fuels
operate at higher center line temperatures. So that is --
DR. SEALE: Yes. Yes. Thank you. Because you gave me a
rational basis to abandon my predispositions.
DR. POWERS: Yes, we specifically -- light water reactors,
they specifically demand that be no center line melting.
DR. SEALE: Below temperature.
DR. POWERS: Mixed oxide fuel is a real mystery. Plutonium
dioxide is complete admissible in uranium dioxide. They make up powders
that press them together. They center them, they don't get homogeneity.
They get heterogeneous regions of locally high plutonium. Plutonium
burns preferentially in the irradiation. You get locally high porosity.
There are some interesting neutronic effects. Again, the point is MOX
is just different.
DR. MILLER: So I will find the reasons for all that
embedded in these papers.
DR. POWERS: Dozens of them.
DR. MILLER: So I do not have any questions then.
DR. POWERS: Dozens of them. Well, chemistry, unlike
thermal-hydraulics, is not a predictive science, and we rely heavily on
experimental data. And our models are built, as I said, in general, all
release models somewhere, somehow have at their core a correlation of
This other question is -- What have we got for high burnup
fuel? What have we got for MOX fuel? There are some tests going on
there, some tests in France called VERCOURS. Also, in France, there is
the PHEBUS program. VERCOURS tests tend to be small tests where they
are heating pellets in furnaces under somewhat controlled conditions,
and they have done this with MOX and high burnup fuel.
We contemplate in this country having yet a different type
of fuel with yet different types of programs, and that is fuel with
burnable poisons. There are neutronic effects when you put plutonium in
fuel and you need to have burnable poisons. And some proposals are to
include coatings of a zirconium boride in the pellets. Well, the French
don't do that and they haven't really experimented with that.
The PHEBUS program is conducting a test specifically to look
at the release of refractory radionuclides, including actinides. It is
an attempt to get fuel to a very high temperature in an advanced state
of degradation. And this particular test is scheduled for the current
I think when we were at the Quadripartite meeting we heard
some words about some proposed Japanese tests. Myself, I don't know
anything about this, but I think they harken back to the kinds of
experiments on fission product release that one of our esteemed
colleagues got to oversee at Oak Ridge. I think the will be very
similar. And I think the intention is to look at some high burnup fuel
DR. WALLIS: Dana, you keep talking about low volatility.
Well, that is all right if their chemistry is all right. They could be
combined with something else which makes them gaseous, if they are in
the right chemical form.
DR. POWERS: We are working in an environment of extremely
high temperatures here and chemistry does simplify. We have not
identified for most fission products, certainly, in most of the
actinides, anything that is extremely volatile. We have identified
volatile forms. The actinides do tend to go -- for instance, americium
goes from an AM-203 to product a volatile species which is AMO, but
these are fairly high temperature phenomena.
We don't have anything with the actinides that is the
equivalent of ruthenium going to the tetroxide, which is the difference
between a material that is virtually not volatile to one that will
vaporize in a warm hand.
Well, the data that come from these tests is scant and all I
have been able to do is intercept some e-mail traffic to show you some
results that are coming apparently out of the VERCOURS program. They
are looking at the effects of burnup and MOX. I have listed there a
couple of tests that are getting talked about the Internet now,
literally over the Internet. So I would hesitate to say these are
validated and peer reviewed data.
Here are a couple of attempts to look at relatively high
burnup fuel. This they call their high burnup, this they call up low
burnup test. And what they have are the cesium release fractions here.
This is a test with MOX fuel. Now, the first thing that you
notice from this, these experimental results, you can't tell a damn
thing from it. There is no measure of experimental error. There is no
test that is replicated. So when we look at this high burnup, low
burnup, and that is a very modest difference in burnup there, and we see
26 to 18 percent, I cannot attest to you that there is any difference
between those numbers at all. If they redid this test six times, they
may have gotten numbers up in the 26 percent.
DR. FONTANA: There is just one test each?
DR. POWERS: That's right. That's where we stand right now
in the database. And, in fact, if you look at the release rates here
for these experiments, yes, you see some differences, but I can't attest
to you that it is -- there are huge numbers of variables that affect
release. They can control a fraction of those. There are limits in
your ability to control things and tests at high temperatures never go
the way you anticipate. That is a fundamental high temperature
chemistry rule. And so there are minor variations here. But on the
fact of it, it certainly suggests that there is a burnup effect.
This burnup effect disappears if I take this fuel up to
complete melting. You get -- for cesium, you get to 100 percent
release, you aren't go to release any more. The rate effect persists.
The difference we see for MOX is impressive. It is
completely different. It starts releasing cesium earlier. It releases
it at a faster rate, and it releases at a greater extent throughout the
experiment. It would surprise me if we could attribute that strictly to
a stochastic variability in the test. But, again, I have not a
demonstration of that.
DR. FONTANA: The statement that you made a couple of
sentences ago, that the effect disappears when you take fuel to melting,
is that for the two rows or all three rows?
DR. POWERS: In fact, Mario, is it is true for everything.
DR. FONTANA: Okay. You would think so.
DR. POWERS: For the volatiles, if I melt it, I tend to
release all of the volatiles, okay, 90 percent of the volatiles and it
is all gone. That is not true of the actinides and you will notice I
have no data on actinides here.
There are a few data points floating around on ruthenium,
but, basically, the French have simply not reduced all the data. And
they do these things with gamma scans, it takes forever to reduce some
of the data. I think Dr. Kress can attest to you at length about both
the cost and the time required to reduce data from some of these tests.
Plus, after you do that, the data reduction, there is an elaborate
effort required to make sure that you have not lost something and that
you are properly interpreting it. A lot of things called fudge factors
come in here magically.
DR. WALLIS: When you compare these, I mean are they in the
same form, or do you say fuel, whether they are exactly the same --
DR. POWERS: Pellets.
DR. WALLIS: -- geometry and history and so on?
DR. POWERS: Yes.
DR. WALLIS: So there is no reason to assume that they were
made differently or anything? There are all kinds of variables.
DR. POWERS: There are a huge number of variables and the
same for the fact that they look at pellets. To say that those
variables have been controlled, no, not a chance. These guys, when they
do irradiated release experiments, you get what is made available to
you. And there is a problem. I mean if somebody gives you a fuel rod
and you pick a pellet out of that, and you want to do a replicate test,
the next pellet down is different, it has a different irradiation
And people take steps to try to correct for that. They used
paired rods, half pellets. There are a lot of steps. You never control
everything. And, in fact, if you look at some of the databases that
were prepared that lie at the heart of one of our release models called
COURSOR, and I do invite members, after drinking a little bit, to
interrogate Dr. Kress on the history of the development of the COURSOR
model, because it is an amazing description of the realities of science.
DR. WALLIS: There is one pellet in each of these? I don't
DR. POWERS: I think these -- well --
DR. WALLIS: How is it possible to run a program of any
integrity whatsoever with one pellet? That doesn't make any sense.
DR. POWERS: The experiments are a little more elaborate
than what I have described, but I think you have a write-up on them.
You can see the details.
DR. WALLIS: It just seems incredible.
DR. POWERS: What you have to understand is that these are
hot pellets, and there are --
DR. WALLIS: It takes an act of Congress to get one pellet,
is that what that is?
DR. POWERS: Well, I understand it takes a hot cell and a
high temperature furnace. Now, hot cells you have got, and high
temperature furnaces you have got, but now you have got to have a hot
cell with a hot temperature furnace. Those are rare things.
DR. WALLIS: There is not a shortage of pellets?
DR. POWERS: It is a shortage of experimental techniques and
these are expensive to do. This is an expensive, expensive process.
DR. WALLIS: Once anyone has done one experiment, he doesn't
want to do it again.
DR. POWERS: You can't afford to, usually.
DR. KRESS: We would typically get three experiments a year,
DR. POWERS: On a good year. On a good year.
I just wanted to put up and show you some of the
complexities and the chemistry involved when you start talking about MOX
fuel and especially MOX fuel with burnable poisons. Chemical phenomena
are driven by this solid state reaction. Hexavalent plutonium just
loves to go into solution in urania. It has -- it would be by itself
happier at high temperatures to be trivalent. That is not so happy in
urania. But when it occurs driven largely by entropic considerations,
it yields oxygen. That gives you a higher oxygen potential. It also
gives you vacancies. Vacancies accumulate and create porosity.
DR. FONTANA: I hate to show my ignorance, but what is O sub
DR. POWERS: Oxygen on a oxygen site.
DR. FONTANA: Oh, okay.
DR. POWERS: This is all done in what is called quarter
notation. If you are not familiar with it, I will be glad to explain it
to you offline. Here I put it up to emphasize the complexity here, that
these are solid state and gas state and heterogeneous reactions.
If we have a coating, oxygen can be consumed by a burnable
poison coating, to put zirconium into the lattice and create boric
oxide. Boric oxide can react with fission products. This reaction
actually is driven by reducing the cesium pressure and reducing its
volatility substantially at the cost of increasing the iodide
My point here is there is a lot of complexities and most of
these complexities were not considered in formulating the NUREG-1465
model of the source term. What was considered in formulating 1465 was
to attempt to get a relative bounding, but not outrageously bounding
estimate on the fission product releases that could occur in a reactor
So, in formulating our advice to the Commission on this
matter, I think we have to bear in mind, that, yes, this kind of detail
was not recognized. The real question, is 1465 bounding or not? So I
have prepared a draft letter that I think you have in front of you in
which I have taken -- made two suggestions.
The first is that when we go in to look at applications to
use MOX, when we go in to look at even high burnup fuel, that really we
ought to do this on a risk-informed basis, that we ought not try to
adhere to the old deterministic regulations. I think they serve us
poorly in that we cannot access the improved understanding and
technologies we have available if we were to live with the TID14844
source term. That, in fact, we should require the use of something like
the NUREG 1465 revised source term, because that gives us access to the
And, second, we need to have the analytic tools and the
experimental data to answer the question -- Does the revised source term
still pose an adequate bound for the purposes that we do safety analyses
in connection with design basis accidents or in connection with the
assessment of the adequacy of engineered safety devices? Or do we need
to put corrections in to accommodate some of this new experimental data
-- experimental analytic data? I don't know the answer to that. I only
know that things are different than what they were when the thing --
when the model was set up.
DR. WALLIS: Was the model confirmed by TMI or something,
where there was just some --
DR. POWERS: It has been looked at with TMI and it is
another one of those things that you can say the model is bounded, okay,
that the releases probably -- were definitely as bad at TMI as the model
would predict. The model has been -- the model is inherently based on
the experimental data. In fact, I think it is a truism there were no
experiments that did not at one time or another get incorporated into
DR. WALLIS: But what experiments? I mean full scale
releases in a real plant is not --
DR. POWERS: Is not in the offing, it just did not occur.
But these varied between fairly large PBF tests, fairly small, but
somewhat integral tests conducted at Oak Ridge Laboratory, scale
experiments conducted at Batelle, in pile tests done at Sandia.
DR. WALLIS: And they all fit with some model?
DR. POWERS: As a correlation.
DR. KRESS: And data out of a test in Germany where it used
very small, not even real fuel, just simulated fuel.
DR. POWERS: That's right. Yes. Sasha experiments.
DR. FONTANA: Backing up a little bit. Does using MOX fuel
go along with high burnup necessarily?
DR. POWERS: No. No. And, in fact, in France they put
burnup restrictions, tighter burnup restrictions on MOX than they do on
Now, I do not understand all the rationale for that, but it
may be simply a lack of experience with the fuels. There are some
differences in the fuel and fuel clad behavior, but most of them seem to
accrue to the benefit. I mean MOX is what is called a softer fuel than
DR. WALLIS: It seems to me all MOXs aren't equal, you can
DR. POWERS: Yes. Absolutely.
DR. WALLIS: You can put them in different ways together.
DR. POWERS: Absolutely.
DR. WALLIS: Sandwiches of MOX, all kinds of stuff, a big
DR. POWERS: Absolutely. And it is interesting, you will
read in some of the papers that the kind of mix that we are
contemplating for the DOE dispersal is one of the worst from a fission
product release standpoint.
I turn to our invited experts and ask if they have any
comments that would help the Committee.
MR. MEYER: I would like to comment on the temperature
ranges here and your concern with the higher releases of volatile
radionuclides at low temperatures. There are low temperatures and there
are low temperatures. If we confine our attention for a moment to fuel
temperatures during normal operation then there is not much of an effect
of mixed oxide on observed fission gas release other than the indirect
effect of the higher temperatures at a given power rating because of the
reduced thermal conductivity of the mixed oxide fuel. And in a Halden
report as recent as 1995, Tony Turnbull has concluded that he can't find
any difference in the release of fission gases in this temperature range
between MOX fuel and UO2 fuel.
In 1995 when we at the NRC first introduced a high burnup
correction factor for fission gas release, and then high burnup was
gigawatt days per ton in our minds, this correction factor was in fact
based on data from mixed oxide fuel, because we didn't have LWR type
fuel data with MOX, so we relied on LMFBR data. Also, the ANS 5.4
standard for fission gas release is -- uses data from, as I recall, from
19 gigawatt days per ton up to the mid-50s, all of which are from 25
percent plutonium mixed oxide fuels.
So I completely agree with all of the reason to expect some
differences, but, as a practical matter, there doesn't seem to be much
difference up to burnups around 45 or maybe 50 gigawatt days per ton in
the normal operating range.
Now, the relevance of this to NUREG-1465 is that the gap
release in 1465 is used for the fuel handling accident and the fuel
handling accident is an accident that doesn't involve any kind of
transient or high temperature operation of the fuel.
And so if you include in the gap release, data from -- in
this intermediate temperature range, above the normal operating range
and up into this range of the VERCOURS testing, or the HI or the VI
testing, then you are putting in there data from a temperature range
that is not applicable to the fuel handling accident.
The other applications for source term in licensing are --
another one is the reactivity initiated accident, the rod drop accident
specifically in a PWR. And NUREG-1465 says it doesn't apply to that
accident because there is something unique in the transient itself that
seems to shake loose a lot of fission gas. And so there you are, I
think, dependent on a totally different database, that is one from the
pulse reactor test programs rather than the high temperature laboratory
And the third application of NUREG-1465 or of a source term,
then, is to the loss of coolant accident. And of the loss of coolant
accident, according -- the way we analyze it according to Part 100 of
the regulation, requires the assumption of a substantial core melt
involved in producing the source term that you use for this event, even
though the event itself is analyzed according to 50.46 and Appendix K
doesn't -- doesn't result in core melt. And so in that case, you are
now into the early containment release for the source term.
So I just want to express a little concern about this
intermediate temperature range because I don't disagree for a minute
with the technical discussion that has been presented, but this
intermediate temperature range doesn't have very much of an application
in regulatory analysis.
DR. POWERS: It was certainly a focus of an awful lot of
discussion before this Committee in connection with AP600, Ralph.
MR. MEYER: Okay.
DR. POWERS: So now why is that a small concern? Was it
just because it was new?
MR. MEYER: I guess I am not aware of those discussions.
Can you help, Charlie?
MR. TINKLER: No, actually, and I was going to comment with
respect to the significance of the data. While it is interesting data,
the VERCOURS data, 1780 degrees Kelvin, from a risk perspective, when
you are doing these severe accident calculations, once the core gets to
1780 degrees Kelvin, it is on the way up. It doesn't hang around at
1780 degrees Kelvin for very long because you are deep into the
oxidation escalation part of the sequence. So, I guess --
DR. KRESS: You have already set it off.
MR. TINKLER: You have set it off and you are on your way
up, and you are not stopping at 1800 degrees Kelvin, and you are not
going to hang around there very long either. So --
DR. POWERS: I would think that in looking at what little I
know about these tests, that actually I would interpret them as having
been through the oxidation transient and now on the recovery.
MR. TINKLER: Oh.
DR. POWERS: I mean they are more suited as a description of
the recovery phase. Now, they artificially held it, and it won't held.
And you are absolutely correct, the temperature escalation will wipe out
some apparent chemical effects, because, I mean you get on a steep
slope, you can't tell the difference between slow and fast in some
But I think the preponderance of the literature, even for
the low temperature range where the Halden apparently can't see
anything, other people see things. And they, of course, have to be
given some credence because their papers are getting into the public
MR. MEYER: Well, you have to get the temperature right and
that's -- I mean, clearly, the accumulation of fission gas in a mixed
oxide fuel rod is going to be higher because it is going to run a little
higher temperature and that is going to pump the gas out. But we are
accustomed to talking about release fractions as a function of
temperature, and when you get it in those terms is when it is difficult
to see the difference in the low temperature range.
DR. KRESS: Yes, the problem with those, Ralph, is that
release fractions as a function of temperature were all developed by
tests like Dana mentioned where they were held at temperature.
MR. MEYER: Right.
DR. KRESS: If one looks at most of the release models, they
are diffusive like in nature, and it depends on how long you are at
temperature, how much you get released.
MR. MEYER: Yes.
DR. KRESS: If it translates the results into a diffusive
like model, which may, like the ones in Victoria, and then superimpose
on that a temperature transient, you get an entirely different basis
with respect from I guess the COURSOR like model.
DR. POWERS: Let me interrupt and, first of all, ask if Dr.
Kress can continue the chairing of the meeting because George and I have
to excuse ourselves.
DR. KRESS: Yes, sir.
DR. POWERS: And, also, to pose the question, I don't think
that we want to plunge into a great deal of detail in our letter to the
Commission, technical detail. I think we do need to get the advice from
our invited experts on what -- how they would care to respond to the
question that has been posed. The question that has been posed is
fairly bare, in fact, I quoted the entirety of the question. And I
believe the invited experts have reviewed the transcript, they will know
that that is all we have. So, if you can pursue that as well as some of
the technical details that would be useful. I think we have to go.
DR. KRESS: Well, with that, I will just at this point turn
it over to the invited experts and say, what advice do you have for us?
DR. WALLIS: Well, could I ask something before we do that?
DR. KRESS: Yes.
DR. WALLIS: Someone behind me mentioned LOCA, Appendix K,
something which I have asked in my naive way about. I mean this strange
regulatory world where you are required to calculate transients and
nothing ever melts, and then you are suddenly supposed to turn around
and assume a substantial melt always seemed to me preposterous and
ridiculous. And it doesn't matter whether it is MOX or not. You can
still make the same preposterous and ridiculous assumption. Who cares?
If that is the regulatory world, who cares what the fuel is?
MR. MEYER: Well, it is a design basis accident.
DR. WALLIS: So if that is the law, who cares about what the
MR. MEYER: Well, the name of the game is to postulate one
of the worst credible accidents that you can think about, the loss of
coolant accident, and at the same time, sort of the worst source term
that you can think about, and patch those together in order to achieve
this level of protection. This is historic, almost prehistoric. We
have always done it this way. It doesn't seem ridiculous to us, but it
is not mechanistic.
DR. WALLIS: But there is a huge gap. I mean, obviously,
there is a huge logical gap between assuming -- you have to calculate it
is a nothing melt and then make this large leap to assuming a huge melt.
That is such a ridiculous --
MR. MEYER: Well, --
DR. WALLIS: It needs a different adjective. It is such a
remarkable leap that it doesn't matter what the fuel is. If you are
going to make that kind of gross assumption, all the details are
MR. MEYER: The design basis accident, in the first place,
is there to establish adequacy of emergency core cooling systems. And
so you have these mechanistic requirements on limited damage to the core
in order to give as much guarantee as you can that the emergency core
cooling systems are going to function properly. And then in our
defense-in-depth kind of attitude, then we leap ahead and say, okay, --
DR. WALLIS: Are you invoking something which we don't
DR. KRESS: You are saying why should the containment --
DR. WALLIS: So you are appealing to some higher power like
MR. TINKLER: It is the design basis conditions for
successive layers of defense-in-depth. There is a DBA calculation for
the containment. There is one for the core in the ECCS and there is yet
another one for off-site dose.
DR. WALLIS: You are justifying what is. But if you are
going to worry about what the fuel really is, then you are going to have
to go to a different level of logical consistency.
DR. FONTANA: In other words, it is not exactly
risk-informed. The detail that you are asking for with respect to the
fuel behavior may be much finer than what is required on the basis of
the accident best defined.
MR. TINKLER: Well, you could distinguish between what is an
appropriate source term for design basis and what should be done as part
of an integrated risk assessment. You can make that distinction, and
the Committee can make that distinction. There is some subjectivity to
boundary conditions imposed for the design basis.
And I would like to just clarify something as far as the
staff's interpretation of 1465, because Dana did use the word "bounding"
a few times. And if you look at the language of 1465, it is pretty
clear it is representative. It is not intended to be bounded --
bounding and says that explicitly. There are aspects of it that you
might say are conservative because it focuses on certain kinds of
sequences, but it is not intended to be bounding. And in that respect,
it is meant to be a substantial challenge to the mitigation of off-site
does. And, really, that is about what you should draw from its
DR. WALLIS: If you knew it was MOX, you wouldn't make --
why would you make any other guess?
DR. KRESS: I will tell why. The reason you do these things
that they talk about is because that was a reasonable way to develop a
set of regulations and a set of criteria on which to judge the designs
in the absence of real information which would be an integrated risk
analysis with full knowledge and characterization of the uncertainties.
If you go use this type of approach, or an LWR, with the
standard fuel that we have and the standard design elements, then later
come back and do a risk assessment, the risk assessment is the proof of
the pudding that this process works, because you send up with a set of
reactors that meet your safety goal criteria, your risk criteria. So it
is an indirect validation that this process works.
And now we are asking, will this process also work for a
different kind of fuel, a different kind of burnup? And the question is
moot, I mean is not answerable by saying, well, if you just throw in the
same criteria in the design basis, are you still going to bound the
DR. WALLIS: So you are saying this very arbitrary
assumption that was made about releases turned out --
DR. KRESS: Was based on knowledge --
DR. WALLIS: -- when you did some more complicated analyses
to be okay?
DR. KRESS: Yeah. But it was based on, these were based on
judgments and good thinking and conservatism that relied on experience
with the light water reactor without MOX fuel, without high burnup.
DR. WALLIS: You are validating those judgments and leaps of
DR. KRESS: Faith. By the actual PRA, because it is the
only way you can do it.
DR. WALLIS: By the PRA, which you also believe, I mean it
is a much better tool.
DR. KRESS: Well, that is the other question, how much are
you going to believe the PRA, and what is its uncertainty?
DR. WALLIS: So you really saying for PRA type world you
need to be much more sophisticated in how you model the releases?
DR. KRESS: That's right.
DR. WALLIS: That's why you have to worry about it.
DR. KRESS: That's right. And that is where the real risks
DR. WALLIS: That if you stick with the Appendix K world,
there probably would never be a reason to do that.
DR. SHACK: We don't plan to design a reactor for MOX, the
reactor is there. You are going to put the MOX in it.
DR. KRESS: That's right. But the question is --
DR. SHACK: Now you have to figure out --
DR. KRESS: Well, the question is, should we allow that?
DR. SHACK: Right. I mean what is the delta risk in doing
DR. KRESS: Yes. And the only --
DR. SHACK: We are past design basis.
DR. KRESS: And there is no way to answer that by going to
the design basis.
DR. SHACK: That's right.
DR. KRESS: You have to rely on something else.
DR. WALLIS: Well, you could. You could legally say design
basis is still valid, forget it.
DR. KRESS: Yeah, you could, if you had the database. And
that is the other part of this discussion, does the design basis still
bound -- that was the question, does it still bound the utilization of
MOX in high burnup fuel? That is the question that we have been asked.
DR. WALLIS: I guess in a legal sense it might today.
DR. KRESS: Might. I don't -- I mean I am not sure.
DR. SHACK: I think the answer probably is it does. Right?
At least that is certainly the answer you had better get.
MR. MEYER: If you go all the way back to the general design
criteria, it probably does. Some of the implementations of those
criteria into specific numerical regulations may have to be tinkered
DR. UHRIG: Are there not some experiments being carried
out, individual pins and assemblies in Watts Bar?
MR. BARTON: No, that's tritium. That's tritium, yeah.
DR. SHACK: Besides, nobody plans to melt the core in a
reactor to find out what happens.
DR. SEALE: That's right.
DR. KRESS: You are not going to get any data out of that.
DR. SEALE: But as long as it is still water moderated, it
looks like a light water reactor design, the general design criteria
DR. KRESS: But I think the real way I have view this is to
say, is to run a risk assessment with the fuel that we now have, in the
particular plants we are going to stick things in, it is plant-specific,
and say, all right, does it make any difference or does it make an
acceptable difference if I stick in MOX and high burnup fuel? And the
only way I can do is by calculation, and the only way I can do it by
calculation is to have enough data to validate the fission product
release in the transport models that are in the code.
DR. SHACK: Just coming back to acceptance criteria, suppose
you could do that and you found that you got a delta risk at this plant,
but it was still within the range of risks at all the other plants?
DR. KRESS: Then you make your judgment. Is this okay?
DR. FONTANA: Yeah, but those plants weren't licensed on
those kind of calculations. You have to start over.
DR. KRESS: It is like a change to the licensing basis.
DR. FONTANA: Yeah.
DR. KRESS: It is almost like that.
DR. UHRIG: Tom, you threw high burnup fuel and MOX
together. They are not necessarily together. You are not going to run
MOX up to a high burnup fuel, are you?
DR. KRESS: It is two separate issues.
DR. UHRIG: Okay. I was confused.
DR. FONTANA: I think it depends on a strategy that you want
to do. If you are just trying to take as much weapons fuel and make it
as unusable as possible, what I would do is not burn it up very much,
and just push a lot more through the system. If you are trying to
change the isotopic ratio, I guess --
DR. SEALE: You want to spread those neutrons around.
DR. FONTANA: Yeah, that is what you want to do. I don't
know how they are going to do it. They will probably -- most
economically, I guess.
DR. UHRIG: What you are advocating is a low burnup?
DR. FONTANA: Well, if you try to push as much weapons fuel
through the system and make it unusable, then you would want to do that.
DR. KRESS: And the other part --
DR. UHRIG: It is recovered from the spent fuel.
DR. KRESS: The other part of this, the other consideration
is how much of the total core is going to be MOX, and how much of the
total core is going to go to these high burnups? I mean that is -- you
are talking about a fraction of the core is what you are dealing with.
MR. TINKLER: That is the point that we feel should be
considered. That we have data to 47 gigawatt days or a 62 watt day per
ton limit, three cycle core. At the beginning of the cycle, none of the
fuel is about 47, at the end of cycle, only one-third of the core is
over 47, two-thirds of the core is under that burnup limit, so --
DR. KRESS: It is not always kosher to assume you are even
at the end of the cycle when you are going to have the accident.
MR. TINKLER: But, you know, it is -- I guess we would say
for those kinds of loads, we don't think for either the DBA or the
severe accident you would ever be able to see the difference above the
kinds of releases we have for the data that we have.
Now, we think there is some -- we are not sure --
notwithstanding the VERCOURS data, we are not sure that the effect is
clear from the existing data on burnup. We are willing to concede that
it is possible at high burnups it could exist. But even from the risk
perspective, with the severe accident, we have -- a total 1465, a total
75 percent release of the iodine, 75 percent release of the cesium. It
is invessel, ex-vessel.
DR. KRESS: Yeah, but you only count the invessel.
MR. TINKLER: Well, that is for the DBA.
DR. WALLIS: Who cares if you are going to release so much?
MR. TINKLER: But for the severe accident risk assessment,
you would consider all of it. So you would be up to 75 percent of the
DR. KRESS: That is releases.
DR. WALLIS: Within uncertainties, the difference between
50-75 percent is nothing, within uncertainty about what is going on.
MR. TINKLER: Well, you are arguing that the design basis
accident is pretty severe and I would agree.
DR. WALLIS: Well, it is big accident and that is it.
That's what this Committee keeps telling me, don't worry about the
details. A LERF is a LERF and it doesn't matter whether it is a little
LERF or a big LERF.
MR. TINKLER: Well, without adequate dose mitigation, the
DBA looks an awful lot like a severe accident. It is pretty
DR. KRESS: That's exactly right. So, you know, those are
the issues that we have to debate, and they are legitimate.
MR. TINKLER: Without beating the obvious, though, for
plants now allowed to go up to 62, current approach is to allow TID14844
for adjustment, it is hard to see that that is a better approach as
opposed, you know, an impediment to the use of 1465. We do see
substantial benefits for the use of 1465 in other areas, safety
enhancement, as well. The most recent case with Perry, the buffering of
the sump, we consider it a good improvement.
DR. MILLER: You said the fuel was one-third MOX, is that on
a -- basis?
MR. TINKLER: I'm sorry, I was referring to one-third high
burnup. One-third high burnup.
DR. MILLER: Okay. I thought you said MOX.
MR. MEYER: But the third also applies to the MOX. In fact,
what we expect --
DR. MILLER: That is on a fuel basis? Is that regional or
MR. MEYER: No. When the weapons material is designed to be
put into a PWR, we expect the design to be no more than a third of the
DR. MILLER: Initially?
MR. MEYER: At any time. At any time.
DR. MILLER: A third of the assemblies will be MOX?
MR. MEYER: Yes.
DR. SEALE: Each load would be one-third MOX.
MR. MEYER: Yes.
DR. SEALE: At most.
MR. MEYER: Yes.
DR. FONTANA: Would these be restricted to the lower burnup
MR. MEYER: Not necessarily. Not necessarily.
DR. MILLER: Just in the normal rotation?
DR. SEALE: If it is the shuffle.
MR. MEYER: There would be some consideration of the vessel
effluents because of the harder neutron spectrum from the MOX
assemblies, so you might not load them out of the periphery.
DR. UHRIG: I probably asked the wrong question here. Would
they be restricted to the total burnup of something less than say 47 or
MR. MEYER: I can't say at this time. Just offhand, I don't
see why they would have to be, provided that the database were available
to justify it. And I think this is exactly where the French are going.
They have this notion of MOX parity, they call it. They want the same
burnup limit for their MOX fuel as they have for UO2 fuel, and according
to their concerns, the missing data are for the reactivity initiated
accident. The Cabri data, Cabri just hasn't tested high enough burnup
MOX fuel rods yet.
DR. FONTANA: Is that because of smaller beta?
MR. MEYER: There are a lot of things. If we get into this,
it will get confusing.
Could we go back to your draft letter?
DR. FONTANA: Okay.
MR. MEYER: I have one concern, and I think maybe Charlie
has a comment about the letter, and it is a concern that is reflective
of the comments that I have already made. Where it says, "We are aware
of experimental studies that show there to be enhanced release of
fission gases to the fuel cladding gap."
DR. KRESS: What page are you on?
MR. MEYER: I don't know.
DR. KRESS: Oh, it is on -- in our letter it is line 45.
DR. SEALE: Yes.
MR. MEYER: That would be it, yes. My only concern is that
someone who might not be familiar with the details would conclude that
the release rate for MOX fuel was higher than for UO2 fuel because we
use release rates, I mean these fractional values to quantify this.
Well, let me back up and try and say that over again. I
will start over again.
DR. KRESS: Okay.
MR. MEYER: During normal operation, which is being
discussed in this sentence, the increased release to the fuel cladding
gap appears to be determined by the temperature effects rather than by
changes in the release rates.
DR. KRESS: Because it tends to reach equilibrium.
MR. MEYER: That is because the thermal conductivity of MOX
is lower, that at a given power level it runs at a higher temperature,
and because of the higher temperature --
DR. KRESS: It reaches equilibrium.
DR. WALLIS: If you run it at the same power you could get
more releases because it would be a higher temperature.
MR. MEYER: That is correct.
DR. WALLIS: So it is true that normal operation, if it is
at the same power level, there would be --
MR. MEYER: Yes. Yes. You know, maybe it wouldn't mislead
anybody. It just -- that is my only comment.
DR. WALLIS: Why don't they operate at lower power level?
DR. KRESS: But your concern is this --
MR. MEYER: You might -- what if, you know, if you stuck in
a phrase like because of its somewhat higher temperatures? Just stuck
DR. KRESS: We will take that under consideration.
MR. MEYER: Okay. Thank you. Charlie may have had a
MR. TINKLER: Well, I already made my comment. With respect
to the issue of fuel dispersal versus vaporization, while it is true
they are fuel dispersal events in a general sense, I think there is -- I
make a distinction between steam explosions which involve a strong
shockwave -- the potential for fragmentation versus fluid entrainment
process, a little different from the ability to finely fragment molten
UO2 with a fluid entrainment process, as opposed to a strong shockwave
passing through molten material, are two different processes.
DR. KRESS: I agree with you completely.
MR. TINKLER: And it is true that when we measure debris, we
get distribution, but some of that stuff we measure in a DCH test is
concrete dust that we create when we blow stuff out of the vessel. It
is not UO2, it is not actinides, it is dust.
DR. KRESS: Now, at one time there was a consideration that
the fuel actually had to dissolve to a gas state.
MR. TINKLER: Right.
DR. KRESS: And that as you dispersed it at high pressure
and low pressure, that that gas goes out of solution and that is part of
fragmentation, as opposed to an entrainment process. Has that been --
MR. TINKLER: Its ability to be -- it is hard to tell
because the thermite we use in these tests is pretty effervescent, okay.
DR. KRESS: Is effervescent.
MR. TINKLER: Yeah. So if you are drawing your conclusions
about the effervescence from thermite, you might be --
DR. KRESS: That is what I was counting, in other words, you
are calling it the thermite really.
MR. TINKLER: And if you look at the Argonne test where they
used U02 results, just a few noted reactor melts, it is a pretty small
percentage, down in the 20 microns or less. But 20 microns is still
pretty big particles for a dispersal off-site.
DR. KRESS: Especially, you know, those settle out pretty
quick, 20 microns. So, you are right, those are big particles.
MR. TINKLER: Those are big particles.
DR. KRESS: I don't recall what we are asking in this
letter, because I forgot -- I haven't read it in a couple of days.
Well, we appreciate the input. We will -- when we are
scheduled to --
MR. SINGH: I think sometime this afternoon.
DR. KRESS: This afternoon.
MR. SINGH: Possibly, or maybe tomorrow morning, I am not
DR. KRESS: So I guess at this time we are scheduled to
break for lunch. Okay.
Well, I appreciate the valued experts and we will take your
advice into consideration. Thank you very much.
At this time I am going to declare a recess for lunch.
[Whereupon, at 12:10 p.m., the meeting was recessed, to
reconvene at 1:25 p.m., this same day.]. A F T E R N O O N S E S S I O N
DR. KRESS: Can I call the meeting to order, please? Dana
and George are indisposed at the moment.
MR. BARTON: Are you sure that that is all that is wrong
DR. KRESS: No.
DR. SEALE: They may even be impaled by now.
DR. KRESS: The next item on the agenda was to talk about
the little working group's product on the role of defense-in-depth in a
risk-informed regulatory system. It started out being a hard paper that
is being developed by our imminent fellow, Jack Sorenson, and then we
worked on it and produced a paper jointly authored by Jack and George
and Dana and I to be presented at some conference or other, I have
forgotten which conference.
And then I was tasked with pulling together the ideas in the
paper and carrying it to the extent of a possible letter on the role of
defense-in-depth in a risk-informed regulatory system. George and I and
Dana were supposed to develop letter jointly and put it forth as a
possible Committee position. Well, things didn't quite work that
I developed the first draft, nobody liked it. I developed a
second draft, nobody liked it. We have gone back and forth. Dana and
George aren't here, but you can -- your handout has attached to it some
of the e-mails we have traded back and forth between George, Jack and me
and Dana. And that can give you a flavor of the problems and the areas
What I suggest we do now is read the letter and read the
attachments so you know what the problem is. I don't have much hopes,
really, of getting a letter out this time. Maybe we can get a start and
get a feel from the rest of the Committee on what ought to go in it.
Dana tells me he has some compromised positions that he
thinks may reconcile most of the problems. I haven't seen that yet, but
when he gets it, I would like to see it.
The things I would like to point out to you mostly that are
areas of contention are, number 1, I have taken the foolhardy or bold
step, depending on your viewpoint, of defining yet again another --
producing another definition of defense-in-depth. That is wrought,
fraught with land mines and all sort of things, so that is one area of
The other area of contention is -- it is a novel definition
that allows you put necessary and sufficiency limits on defense-in-depth
and I have outlined to do that, and there is a lot of contention as to
whether that is practical, workable, doable, those sort of things. And
I have even put into the letter an example of how to apply this, and the
example has to do with the balance between CDF and containment failure
as a defense-in-depth concept. And that also is one of the areas of
So having flagged some of the points, there are others, but
those are the major ones, I think, maybe you could spend the time now
while we are waiting for Dana and George, because I don't think we can
advance very far without them. Maybe you could read this whole package
and then we would be prepared to discuss it further.
Does that sound like a good way to --
DR. WALLIS: It is not very different from draft number 4
dated March 29th.
DR. KRESS: No, it is not.
DR. WALLIS: I had some comments on that which I haven't
shared with you yet.
DR. KRESS: Okay. Well, we would be pleased to have those,
DR. WALLIS: I notice some of the points have disappeared.
DR. KRESS: Well, that may have happened, yeah. Maybe other
people had the same problems.
DR. SEALE: One other comment here. If anybody has any
other examples --
DR. KRESS: That is the other point, yes.
DR. SEALE: On page 2, I think it be worthwhile to try to
DR. KRESS: It would be nice to have those, yes.
DR. SHACK: Again, if anybody could explain the examples
that are already there.
DR. SEALE: That would help, too.
DR. SHACK: That would help, too.
DR. KRESS: Yeah, that is why we need --
DR. SEALE: I am not sure I read that first one the way --
as being correct.
DR. SHACK: I am pretty sure I read it as being incorrect.
DR. SEALE: Yeah, I think that is --
DR. KRESS: The original drafts had no examples.
DR. SEALE: Yeah, but that is not the steam generator.
MR. BONACA: What example, you are talking about the one in
DR. WALLIS: Well, the 50.59 --
MR. BARTON: Yeah, these two here. These two, these
DR. SEALE: The first example.
DR. KRESS: Well, I can make a stab at it. I didn't put
these examples in, they were supplied to me. The problem is that is an
DR. SHACK: What is an arbitrary choice?
DR. KRESS: Not to choose risk-informed limits on the
DR. SHACK: No, the viewgraph we saw said you should define
the accident leakage limits based on risk information. It was directly
contrary to the statement.
DR. KRESS: Oh. Well, we will wait till Dana comes back. I
didn't like either one of the examples, to tell you the truth.
DR. WALLIS: You keep saying that defense-in-depth
compensates for uncertainty, controls risk -- all that. That is just an
assumption, no one has ever proved it to me.
DR. KRESS: Oh, I fully admit --
DR. WALLIS: Intuitively, it is reasonable.
DR. KRESS: I fully admit that what I have done is concocted
a concept of defense-in-depth and concocted a definition that may or may
not have any basis in anybody else's definition there.
MR. BONACA: Well, when I received your letter, okay, first
of all, I thought that it was important, because what you did -- I mean
I really believe that defense-in-depth had to be a concept that we
should espouse to some degree, some level, okay. Now, what we have in
place today is so fundamentally deterministic that I have got a problem
with it, too.
DR. KRESS: Yeah, that is --
MR. BONACA: And I can understand that the commitments to
the current defense-in-depth are an impediment to the application of
PRA, and I totally agree with that.
On the other hand, I think what we are trying to do to
establish new criteria for defense-in-depth, which would allow, in fact,
a better technology to be applied, and then to articulate a better --
DR. KRESS: That is more or less the motivation.
MR. BONACA: Yes, and I agree with that entirely. I mean
DR. KRESS: You have put your finger on exactly the problem
and the motivation.
DR. SEALE: The camel that we call defense-in-depth today is
a product of the time in which the only guidance we had for designing
reactors were the general design criteria and we did not have -- we did
not rely on much more than bounding -- I won't even say calculations --
evaluations, bounding evaluations of plant accident behavior. And that
is a very different breed of cat, or it ought to be, from what
defense-in-depth would be today.
MR. BONACA: Yes. See, the issues, for example, are like,
you know, common mode failure. We all agree that it is a fundamental
problem. Therefore, in the past, not having a better technology, we
devised diversity, redundancy, et cetera, to deal with those issues in a
very structured way, that was the best we knew. Now, PRA is much more
capable of identifying those linkages, so we would propose different
means of addressing the same concern. But you would still want to
design through those kinds of new technology against common mode
failure. So, the objective -- so, you know, I thought that your paper
was on target. Okay.
I don't know how it has evolved now, but --
DR. KRESS: It still has that in it. And so, you know, I
have had to fight to keep it in.
MR. BONACA: Those are the examples, however, I would
suggest we put in. Like, for example --
DR. KRESS: That would be a good one.
MR. BONACA: Issue, you know, common mode failure is a
concern. In the past, deterministic has been using these, you know,
precepts because there was nothing better, but today you have a
technology that we know much more capability, that can be applied, and,
you know, that is an example that we could make of how you would build
this new -- again, so, anyway.
DR. KRESS: There is also --
DR. SHACK: What would we have to know to apply your concept
to the decision on a containment spray for an AP600?
DR. KRESS: You would have to know -- you would have to have
two things, and these don't exist. This is part of the problem. You
would have to know what is an acceptable value of, say, LERF. You would
have to have a value that the Commission says this is the acceptable
LERF. You are above, it is not acceptable; you are below it, you are
okay. That it item number 1.
DR. SHACK: Okay, I give you that number.
DR. KRESS: Yes, okay. There is something else --
DR. SHACK: Numbers I can supply to you easily.
DR. KRESS: There is something else you have to have to
apply my concept and that is, we all know that when you calculate that
LERF, because it is a calculated value, what you have to go through to
get it, that there is a relatively large uncertainty in it. And when
you -- the thing you have to have in addition to the number is an
acceptable value on that uncertainty, how big it can be. So that is
another policy issue.
I don't want to know the number just in the LERF, I want to
know how much uncertainty you are willing to accept. That's two things
you have to have. You give me those two things, and then you have to do
enough defense-in-depth in order to meet both of them. And that is
enough information to fix all the defense-in-depth you need.
Now, the point is, that I am making here, is the value that
is acceptable for uncertainty may very well depend on the absolute value
that you have for LERF, or may even depend on the absolute value you
have for CDF, which is part of LERF. So there is not necessarily one
value, it may be variable. The bigger those are, the less uncertainty
you are willing to tolerate. The lower they are, maybe it is all right.
So this is an unknown function that somebody has to develop
as a policy statement. It is not easy to come by. Policy things are
never. But if you had those two things, then my concept deals with
using defense-in-depth to both realize a risk objective, which is the
LERF, and to develop your level of confidence with the uncertainty, to
deal with both of those things, which is what defense-in-depth is
supposed to do, it compensates for --
DR. WALLIS: Well, I agree with what you say, but your
write-up mixes up these concepts. I mean acceptable risk and acceptable
level of uncertainty are two concepts.
DR. KRESS: Two separate --
DR. WALLIS: At some level, those are essential design
DR. KRESS: They would have to be design criteria.
DR. WALLIS: Defense-in-depth with multiple layers of
protection is one way to meet those criteria. It is a different thing,
it is a different animal completely from acceptable risk.
DR. KRESS: I agree. I may have --
DR. WALLIS: You put them all in the same sentence.
DR. KRESS: Well, I may have phrased it wrong and it may
need some clarification. But you are right, they are separate things.
The defense-in-depth provisions you do are a way to meet these things.
DR. WALLIS: And as far as we have an intuitive sense that
that is the best way to meet it. But I haven't seen the existence proof
that it is the best.
DR. KRESS: Well, I do make a point in there.
DR. WALLIS: There are other ways perhaps.
DR. KRESS: I do make a point in there that the concept is
that if you have these two numbers that I talked about, uncertainty
acceptance and LERF acceptance, --
DR. SHACK: And you are confident that you model uncertainty
is small enough that the uncertainty that you can compute and measure is
DR. KRESS: You have to do something about that. You know,
there's lots of issues here. But my point was that the reason you put
more and more defense-in-depth on a design is because you are
uncomfortable with the uncertainty in your risk numbers. Therefore,
your comfort level goes up when you put more and more defense-in-depth.
I reinterpreted that to say there must then be an inverse
correlation between the amount of uncertainty and the amount of
defense-in-depth. The more defense-in-depth you have, the less
uncertainty in your risk metric. That is an assumption in here.
DR. WALLIS: We are also comfortable --
DR. KRESS: And that hasn't been proven. That is an
inference on my part. And if that is not true, why you might as well
just throw this in the garbage, too. So there is a huge leap of faith
there that I mean to meant to point out to you.
DR. SEALE: Tom, I take issue one of your comments here.
You know, earlier, George made the point as he went through his
cornerstones argument that the further you went from the initiating
event in a sequence, through to the consequences, the greater the
uncertainty was because you were propagating more and more uncertainties
through each other.
DR. KRESS: That's true. That's true.
DR. SEALE: Okay. You said just now that the more
defense-in-depth you had, the less uncertainty you had.
DR. WALLIS: No.
DR. KRESS: Yes, I did. I said that.
DR. SEALE: And I would say the more comfort you have, you
don't have less uncertainty, it is just that you are more and more
comfortable that the real world is somewhere on the good side of where
you are because of your defense-in-depth.
DR. WALLIS: I think I know why are more comfortable, that
was what I was going to say, is that you don't think in terms of
probabilistic things and uncertainties. Even if a mathematician could
prove to you that with three layers, there is less uncertainty than with
six, with proper design.
DR. SEALE: Yes.
DR. WALLIS: You tend to think in terms of either it fails
or it doesn't. You tend to think in terms of yes-no. And so you much
more comfortable to have six yes-noes than three, because that is the
way you are sort of constrained to think. You don't think in terms of
DR. KRESS: I will put forth a hypothesis. That hypothesis
is the more defense-in-depth you apply, in the traditional sense of
defense-in-depth, the less uncertain is the final number you have on the
risk metric. That is a hypothesis that I thoroughly believe and am
putting forward. I know how to prove it, you can prove it with a PRA.
DR. WALLIS: That is not true at all.
DR. KRESS: Well, that is a hypothesis.
DR. WALLIS: Not true at all. If you had two layers and you
tested them ad infinitum, till you were absolutely how they would
behave, that would give you less uncertainty than if you had six layers
that you were uncertain about every stage, obviously.
DR. KRESS: No, no. No, no. No.
DR. WALLIS: Having more layers doesn't make you any more
DR. KRESS: Yes, it does.
DR. WALLIS: No.
DR. KRESS: If each layer has its own level of uncertainty
DR. WALLIS: But they may have greater levels of uncertainty
than two layers which you really design properly.
DR. SHACK: I think he is adding a third layer to the two
DR. KRESS: That's exactly right.
DR. WALLIS: No, no, no, no, no.
DR. KRESS: That is what I am doing. That is exactly what I
DR. SHACK: He is not substituting three imperfect layers
DR. KRESS: No, I am not.
DR. WALLIS: You aren't allowed to that. You aren't allowed
to do that.
DR. KRESS: Oh, yes, I am.
DR. SHACK: That's what defense-in-depth means.
DR. WALLIS: Oh, well, then --
DR. SHACK: That is alternate design procedure, to use three
imperfect layers in place of two perfect.
DR. KRESS: Yeah, but that is something else altogether.
DR. SHACK: That is a different beast. Defense-in-depth
says I do the best I can with two.
DR. KRESS: That's right. And then --
DR. SHACK: If that is not enough, I had a third.
DR. WALLIS: No, but I don't think that is the case. I
don't think that is the case.
DR. SEALE: You may have more comfort, but you don't have
more -- less uncertainty.
DR. SHACK: I sure as hell do.
DR. KRESS: I do. Yes, I am glad I have got one guy
agreeing with me on that one.
DR. SEALE: No, no.
DR. KRESS: I can prove it. This is a hypothesis that can
DR. WALLIS: No.
DR. KRESS: And I can prove it using a PRA.
DR. WALLIS: Gee whiz.
DR. MILLER: I think Bob is right, but the probability of
having a problem is less. You have more uncertainty on a less -- lower
DR. WALLIS: Yes, that's right. I agree with that.
DR. MILLER: Uncertainties are added in quadriture and the
probability is a product.
DR. KRESS: You are right, but you are talking about a
DR. WALLIS: You are mixing up uncertainty and probability.
DR. KRESS: You are mixing -- well, you guys are mixing it
DR. WALLIS: No, no, no, no. You get a C for that.
DR. KRESS: You wait till I write my --
DR. WALLIS: You can say anything you like about me.
DR. KRESS: I am going to. You are guys are wrong about
Now, the mathematician here is going write something down
for us. Now, the quadriture doesn't work here, John. You have got to
apply the same quadriture to two separate situations, though. You can't
use it for a static thing. I am adding more defense-in-depth.
DR. WALLIS: You have reduced the risk.
DR. MILLER: You have reduced the risk by adding three
DR. WALLIS: You haven't reduced the uncertainty.
DR. MILLER: And you reduced -- the uncertainty, in your
mind, is reduced, I agree with that. But the uncertainty in the end
result is actually --
DR. WALLIS: Well, I guess you have to define uncertainty --
DR. KRESS: How can you guys say that?
DR. SHACK: If you are saying sigma isn't smaller, that is
one thing. If you are saying the uncertainty on whether it fails or it
doesn't fail, is a different beast. We are not talking about whether
sigma is larger or smaller. We are asking whether the uncertainty in
the failure probability is larger or smaller, and that is different.
DR. SEALE: Well, that is what I was calling the comfort.
DR. SHACK: Well, that is the uncertainty we are talking
about here. You have got to define what your objective is, and that is
to prevent failure.
DR. SEALE: That's right.
DR. SHACK: And then the question, what is the uncertainty
on the failure?
DR. WALLIS: Well, I guess we have to define what we mean by
uncertainty more carefully.
DR. SEALE: That's right.
DR. WALLIS: Maybe that's why --
DR. SEALE: That is the case.
DR. WALLIS: If Tom defines it his way, he may be able to
DR. KRESS: Well, I have to admit it, I defined it as sigma.
Sigma on the LERF, in terms of sigma, you know, it may be not be
DR. WALLIS: Sigma.
DR. KRESS: Sigma. But it is the same thing as some
multiple of sigma on the LERF.
DR. WALLIS: Now, you could put in another layer which is
extraordinarily uncertain and not very effective, I don't think that
helps you very much in terms of certainty.
DR. MILLER: No.
DR. SHACK: No, but it doesn't hurt you.
DR. WALLIS: Well, in terms of risk, it doesn't. But in
terms of uncertainty, --
DR. MILLER: It doesn't hurt you either. Relative
uncertainty still goes down.
DR. WALLIS: It is relative.
DR. KRESS: Yeah. That is what I was trying to say.
DR. MILLER: Well, then we agree.
DR. KRESS: The relative uncertainty gets smaller, you have
DR. WALLIS: Okay. So your revised paper is probably more
acceptable than the first one. Maybe I -- maybe the C was in my mind.
DR. KRESS: I don't know. But, anyway, this is --
DR. MILLER: Now, let's read the paper.
DR. SHACK: I think, you know, your approach is a useful way
to think about defense-in-depth.
DR. KRESS: Yes, I know, whether you can actually quantify
it and put it --
DR. SHACK: Right. Now, whether it provides a useful way to
make policy and decisions, I think is a very --
DR. KRESS: And that is one of the points of contention
between one, too.
DR. SEALE: That is a real different question.
DR. KRESS: Dana particularly thinks it is not useful in
that sense. And that may -- that is a legitimate.
DR. SHACK: You would have -- you know, I mean unless you
like create debating societies.
DR. KRESS: That is one of the points, yeah.
DR. SHACK: Sooner or later, somebody is going to have to
make a judgment.
DR. KRESS: But, you know, the way to start with something
that is as difficult as this is to get an overriding principle that you
DR. SHACK: Uh-oh.
DR. KRESS: And then work backwards and say, what can we do
with this to make it practically implementable? That, I am asking -- I
am not going to do that. All I want to do is say this is the principle,
let the staff work on how to implement it.
DR. WALLIS: Well, I don't think you can ever justify it as
a principle. You can say that acceptable risk and acceptable levels of
uncertainty are the basic design goals and they, you could say, were
principles. Defense-in-depth is one approach to meeting them. The
other one is to do a much more thorough design and testing of fewer
elements and have less defense-in-depth but more substance to the actual
defense at each layer. That meets the design objectives.
DR. KRESS: Yes, well, that is almost -- that is actually
covered in there, because what you are doing is you are either reducing
the uncertainty or reducing the risk metric, one or the other.
DR. WALLIS: So that is what you have to show when you
involve defense-in-depth. You have to say that with this
defense-in-depth I get a better -- a better something.
DR. KRESS: Well, that is the hypothesis.
DR. WALLIS: But you can't appeal to defense-in-depth by
itself without a measure of reduced uncertainty.
DR. SHACK: No, but this definition covers that.
DR. KRESS: I thought I covered that.
DR. SHACK: If you make the first two barriers so reliable
and so uncertain, then when you compute to see if you have got livable
uncertainty, you say --
DR. WALLIS: That is not the way it is invoked, though. It
is invoked saying we have got these various things, we have got to
maintain defense-in-depth. It doesn't say anything the certainty of the
barriers. It says that more barriers is good. That is the --
DR. KRESS: I did amend -- I hope that wasn't in there.
DR. SEALE: Part of the reason you do defense-in-depth,
though, is that you don't know for sure whether you are playing with a
full deck. I mean --
DR. KRESS: That is one of Dana's.
DR. SEALE: You have got a completeness problem.
DR. APOSTOLAKIS: You are chairing. You don't like it?
DR. KRESS: Not in this. This is -- I have got too much of
a conflict of interest to chair this. But I guess Dana has, too.
DR. SHACK: Yeah, like I say, who are we going to hand it to
DR. APOSTOLAKIS: Well, you want to give it to a person who
DR. KRESS: George would be more neutral.
DR. APOSTOLAKIS: I am neutral.
DR. KRESS: Yeah, let's turn this over to George. You
haven't been in on this debate we have already had, so we would be
interested in your comments that are coming in cold.
DR. SEALE: Why don't you stay there for five minutes so
George knows where we are, rather than having to start over.
DR. KRESS: I would rather have him start over.
MR. BARTON: That is probably why we never got this thing
DR. APOSTOLAKIS: Did you actually read the letter?
DR. SHACK: Don't worry, if we want to start over again, we
DR. APOSTOLAKIS: What is the major objection?
DR. KRESS: To the letter? It appears to be --
DR. APOSTOLAKIS: Let me start from here. Have you guys
read the paper? Are there any objections to the paper?
MR. BARTON: Jack's paper?
DR. APOSTOLAKIS: Yes. There are.
DR. SHACK: Well, I don't have any objections, I have read
MR. BARTON: It is a good paper.
DR. APOSTOLAKIS: Anybody has a problem with the paper?
DR. WALLIS: We are talking about detail.
DR. APOSTOLAKIS: Of the paper or the letter?
DR. KRESS: They are confused. They don't realize that
there is a paper and a letter.
DR. APOSTOLAKIS: No, everything knows there is a paper.
DR. KRESS: Well, some of them --
DR. SHACK: Yes, well, --
DR. APOSTOLAKIS: Okay. Now, I will take -- I will put on
the table an extreme view, okay, which Tom may not like. I would be
very happy, I think it would be a useful contribution to the ongoing
debate among the Commissioners and senior staff to send a paper with a
very short letter that says here is our -- I like the opening thing.
Oh, you took that away?
As part of our continuing --
DR. KRESS: Dana hated that.
DR. APOSTOLAKIS: What did you like, what did you put there?
DR. KRESS: I don't remember. Dana hated it.
DR. APOSTOLAKIS: As part of our continuing interest in
contributing to issues, --
DR. KRESS: Yes, something like that.
DR. APOSTOLAKIS: -- please see the attached paper. Thank
you very much. Dana Powers, Chairman of the Committee. One sentence.
DR. FONTANA: I move we change chairman again.
DR. APOSTOLAKIS: Or maybe two.
DR. WALLIS: The attached paper --
DR. APOSTOLAKIS: In other words, this is something that
will go upstairs very quickly, will make a contribution, and we can
address Tom's other ideas, which I think, you know, are a significant
advance, but they require debate.
DR. SHACK: The paper leaves you with the structionalist
approach and the rationalist approach. Tom has come down with both
DR. KRESS: Yeah. Gotten down out of the clouds to a lower
level of clouds.
DR. SEALE: Yes.
DR. APOSTOLAKIS: So I would like to see whether people are
willing to go along with that. And we are not rejecting this letter.
DR. KRESS: It is a rationalist case.
DR. APOSTOLAKIS: We are not rejecting it.
DR. SHACK: We know which camp you are in.
DR. KRESS: Okay.
MR. BARTON: What are we doing with it, if we are not
rejecting? What is your plan for this letter?
DR. APOSTOLAKIS: Well, this letter can go up in June or
July after the members express concerns via e-mail, you know, so that
Tom will have input to think about.
DR. SEALE: Not only that, but you are also doing a little
bit of evangelistic work on the Commissioners from a less partisan
perspective if they see the paper, have a chance to digest it and so
forth, and then get exposed to coming down with both feet in one camp or
DR. APOSTOLAKIS: Right.
MR. SAVIO: We have an SRM to respond to May 17th.
DR. APOSTOLAKIS: Well, the paper is probably good enough
DR. KRESS: April could be the response.
MR. BARTON: A response for the SRM.
DR. APOSTOLAKIS: Yes, the SRM doesn't say much.
MR. MARKLEY: Does it answer the question in the SRM,
DR. APOSTOLAKIS: Which is?
MR. MARKLEY: The balance between PRA and defense-in-depth.
DR. APOSTOLAKIS: Well, option two, that is what it is. It
says use structuralist at the very top and rationalist below. I am
willing to defend it.
DR. SEALE: It is not a simple question.
DR. APOSTOLAKIS: It is not a simple question.
DR. SEALE: If you send something like the paper to the
Commissioners and say we are thinking about it, this is the way the
problem is being structured. We are going to examine the issues in
greater detail, and we will get back to you when we have more to say.
DR. APOSTOLAKIS: That's right. I think option two
addresses the concern of the Commission. And we will come back with a
DR. SEALE: Yeah.
DR. WALLIS: I just want to make sure that we do come back,
that it not one of these things we complain about.
DR. APOSTOLAKIS: This letter will not die, Graham. This
letter has a lot of sweat -- what is it, blood, sweat and tears. And I
don't think it should die either. I mean --
DR. SEALE: There is even some consternation in it.
DR. APOSTOLAKIS: Maybe we can impose a deadline on
ourselves, that maybe by the July meeting at the latest, this goes up.
Dana, I just proposed, but evidently there is a lot of
objection to this letter. That we send the paper which our colleagues
do not object to, with a very short cover letter to respond to the SRM
we have, and then inform them that we are continuing to work on the
subject and, you know, something is coming up soon.
DR. POWERS: Can you give me a thumbnail description of the
nature of the objections to the letter?
DR. KRESS: They are a lot like some of yours.
DR. POWERS: Obviously sage objections then.
DR. KRESS: So if you remember what yours are, there is a
flavor of a lot of that in the objections I am getting here, plus some
others like I may put forth a hypothesis that some of the people don't
even think is true, the basis for the concept, in other words. That
hypothesis is that the more defense-in-depth you have, the less
uncertainty you have in the final risk -- with the hypothesis. There
are a number of people here that are telling me that that is not true,
it just the opposite.
DR. POWERS: I thought that the nature of your hypothesis
was in reality that is what should be true.
DR. KRESS: It should be true.
DR. POWERS: And that -- the situation is, in fact, that
defense-in-depth as originally formulated is an archaic concept.
DR. SEALE: That's right.
DR. POWERS: And that it has to evolve in the face of this
new technology. And I thought that the --
DR. KRESS: That a better representation of what I was
trying to say.
DR. POWERS: That the hypothesis that is articulated in here
is that in a perfect world, there would be an inverse correlation.
DR. KRESS: Absolutely.
DR. POWERS: But that when we found that there was not such
an inverse correlation, that was a point at which we were misapplying
DR. KRESS: That is a wonderful -- a better explanation of
DR. MILLER: Could you explain what the imperfect parts of
the world are that --
DR. POWERS: Well, I think we see them when we speak to
applying defense-in-depth to the safety analysis of digital electronic
DR. MILLER: That is because you have potential for common
mode failures, is that it?
DR. POWERS: I think we see arbitrary appeals to
defense-in-depth there. I think your potential for common mode failures
and their impact is something that lies legitimately in the domain of
PRA to assess and evaluate for you and make decisions on whether you
need redundant and diverse systems. To come back and say I need
redundancy, diversity or something else in the name of defense-in-depth
is a misapplication of the modern concepts of defense-in-depth.
DR. SEALE: It was appropriate in the --
DR. POWERS: It was appropriate when we lacked the tools.
DR. SEALE: That's right.
DR. POWERS: It is inappropriate now. That is the concept.
Now, Tom and I have had a running dialogue on this now for three months
with some of the more vitriolic e-mails passing back and forth.
DR. KRESS: Those are -- some of them are in here.
DR. POWERS: He does go on and offer -- Tom is interested in
seeking resolution, a definitive path through the morass in this letter,
and I have been a reluctant dragon on this. I see the virtue of trying
to cut through this. But I think it requires more thought, and where we
last left it was that the letter necessarily has to be longer. But I
think I would not be in favor of producing a letter that simply said
here is our paper and we are working on this more. I think this concept
of inverse correlation has to appear.
DR. KRESS: I agree.
DR. POWERS: Now, much of the rest of it on the definition
of acceptance criteria and the need for those, I think that has to
DR. KRESS: I agree.
DR. POWERS: Now, the definition of new risk measures and
the call for a definition of acceptance criteria there --
DR. KRESS: It doesn't have to --
DR. POWERS: I think is something that deserves longer-term
DR. KRESS: Yes, I agree with that, too, Dana.
DR. POWERS: So my feeling is much more optimistic maybe,
that we can produce a letter that I think expresses truisms and facts,
and alert the Commission to this pitfall that they have on the road to
risk-informed regulation. And, in fact, I have suggested there are
three others that need to be attended to. One of those three others is
in fact subsumed in the existing letter, and that it is these risk
measures and their acceptance criteria, I think is one that needs to.
We have just gotten out discussions where this precise topic
came up, and it is evident that we need not only to square our risk
measures, but to square our thinking about acceptance criteria with
international acceptance criteria.
DR. KRESS: Yes, that's true.
DR. POWERS: And at least say why --
DR. KRESS: Why they are different.
DR. POWERS: Why we differ and why we should not go to the
most restrictive of acceptance criteria that you find in the field.
So I guess my advice to you, Tom, is no, let's not go for a
minimalist letter. Let's take the salient points and those that are
more speculative about cutting through the morass. Sure, we can work on
those and set a deadline. I suspect it is the fall, but I don't think
it should be beyond the fall.
DR. APOSTOLAKIS: Is the letter going out at this time,
DR. POWERS: The letter is scheduled to go up next month.
DR. FONTANA: Next month?
DR. POWERS: In May.
DR. APOSTOLAKIS: If we already know that we agree on
certain things and we want to put them in the letter, why don't we send
it at this time. They asked us, we know that they are concerned about
it. We don't have to go by the deadline of the SRM.
DR. SHACK: No, I think he was just giving you the maximum
DR. APOSTOLAKIS: No, but I think we should -- I mean, Tom,
do you feel you can draft a letter based on this discussion?
DR. KRESS: I think with Dana's help.
DR. APOSTOLAKIS: Cutting and pasting.
DR. KRESS: I think with Dana's help. Not cutting and
pasting. With Dana's help.
DR. APOSTOLAKIS: A new letter.
DR. KRESS: Well, it would be -- I think it would have just
a bit of a different flavor, and it would be -- it wouldn't be so
dogmatic in the sense as this is what defense-in-depth is, this is the
way you have to handle it. It would be more like these are the
attributes you should have in a proper defense-in-depth concept. And
cast it in that way. And then you would say, it would be no longer a
hypothesis that says if you apply defense-in-depth properly, the more
you put on, the more the uncertainty should be reduced. It would have
just a little different flavor to it. It would preserve much of the
stuff that is in there. It wouldn't come down as this is the way it is,
the way it has to be done type thing. It would be when you develop a
defense-in-depth policy, it needs to have these kinds of characteristics
DR. WALLIS: I am concerned though because I think this has
potential to be a very significant letter, and I am concerned about
rushing into a letter the way we do, rushing into a consensus, which we
do wonderfully on some things, so important, without time to mull it
over and say, well, do we really do mean that, or is there a better way
to do it? And I am sorry it has taken some time already. But I would
feel uncomfortably, personally, about endorsing something which is sort
of a compromise of things, various things people think at the moment,
rather than something which is really solid.
DR. KRESS: You would like to spend more time.
DR. WALLIS: Yes, I think so.
DR. APOSTOLAKIS: So you are arguing for May.
DR. WALLIS: I don't want to put it off much at all, but I
do think --
DR. APOSTOLAKIS: No, May, we have to send something.
DR. WALLIS: I am nervous about rushing into a decision in a
DR. APOSTOLAKIS: In May we have to do it.
DR. WALLIS: That's fine, we will definitely do that.
DR. APOSTOLAKIS: Well, all right.
DR. POWERS: Well, let's turn charter Tom and Dana to
struggle with some words and see what we can come up with Saturday
DR. SEALE: I sure would like to hear Dana's evolution of
the relationship between layers and uncertainty, because Tom's was a
DR. POWERS: Well, I think you look at the paper, you see
that this distills fairly logically in the paper that there has to be a
change and that there are two categorizations of change, two
categorizations of defense-in-depth that appear. One is the dogmatic
overall and the other is a consciously applied based on quantitative
analysis, which tends to be on the microscopic side.
I think we can -- I mean I think that is the inverse
correlation area. Tom has a more mathematical definition than appears
in the paper, and it his attempt to define necessity and sufficiency.
And necessity and sufficiency I think is the thing that we don't see in
the arbitrary appeals to defense-in-depth.
DR. APOSTOLAKIS: I think in my mind it is very simple. The
only reason defense-in-depth deserves a name in the new era is that many
times in the past we have been wrong, and I would like to have that
feeling that I have something to protect me.
DR. WALLIS: That is the yes-no thing.
DR. APOSTOLAKIS: Now, the rest --
DR. SEALE: That is a completeness question.
DR. APOSTOLAKIS: Yes, exactly, it is the completeness
question. Now, we can debate it for a long time, you know, where we can
come up with an exact definition and so on. But if that was not a
factor, I would argue for the rationalist approach all the way.
DR. WALLIS: Even then the question of how many layers you
need depends on how comfortable you feel about your likelihood of being
DR. APOSTOLAKIS: Sure.
DR. WALLIS: So you face the same problem.
DR. KRESS: It's the same.
DR. WALLIS: And you can always say adding another layer is
better, but that is never any good, because you can add infinite number
DR. APOSTOLAKIS: That's why we have a specific proposal in
DR. SHACK: No, but if you have no model uncertainty, this
does become a mathematical problem. I can do the analysis if I don't
have to worry about model uncertainty.
DR. KRESS: You can do it precisely.
DR. SHACK: I can, yes.
DR. POWERS: And what we are saying is that we have become
sophisticated in small areas. We are sufficient that the residual
uncertainty of completeness is small and we are not even going to allow
DR. FONTANA: Let me come up with an old example here and
see how this fits. Design the reactor, put a containment around it. We
have got defense-in-depth. Then somebody does a WASH 740 and finds a V
MR. BARTON: Find what?
DR. FONTANA: A V sequence, a bypass sequence. The
defense-in-depth that you thought you had wasn't there, to the extent
that you --
DR. APOSTOLAKIS: Well, that's right.
DR. POWERS: That's true. Keep going.
DR. FONTANA: All right. So if you were trying to do a
mathematic thing, you would be wrong because you forgot something, or
you didn't know.
DR. SHACK: But if I had no incompleteness --
DR. WALLIS: It is like steam generator tube ruptures. You
can put another box around the whole thing to catch whatever comes out
the steam generator tube, or you can make better steam generators.
DR. KRESS: Absolutely.
DR. WALLIS: Now, you have got a choice. It is not clear
defense-in-depth is the obvious way to go.
DR. FONTANA: According to Tom's definition, I think you are
not allowed to make a crappy steam generator because you put a box
DR. SEALE: We made the comment before you guys got back
that the present, in-place, defense-in-depth process is appropriate to
the general design criteria as being the way we do the business. And
that is not a bad comment to make actually. And that as we have
developed now the models, the understanding of the pieces and so on of
the process, we can attack this problem more mathematically and begin to
get to the kind of trade that you are talking about.
To the extent that we have residual concerns about
completeness, you know, maybe there is a version of the Y2K bug that
bites fuel elements or something like that --
DR. APOSTOLAKIS: But completeness --
DR. SEALE: Completeness in the sense that there is an
accident out there that we haven't experienced yet. To that extent we
still feel that we need defense-in-depth.
DR. POWERS: No, I think he has got -- I think he has upon
an important point that we want to make, Tom, is that at no time were we
speaking to the GDCs. That as a generalized good engineering practice
-- in fact, we touch upon in the paper, defense-in-depth is a good idea.
It builds into your plant the kind of redundancy that leads to low core
DR. SEALE: Right.
DR. POWERS: We are speaking primarily to the issues of the
safety regulation of plants where you have to make decisions. And you
want decisions to be necessary. And having made those decisions, you
want to make them sufficient. And that is where we are having troubles
because we think we have a tool that for many, many kinds of decisions
is sufficient. And we find these arbitrary appeals undermining its
MR. BONACA: Can we talk about the issue of -- for example,
I mean I was mentioning how I liked this paper, personally, because it
leaves defense-in-depth as a concept, but at such a high level, and then
it allows for new technology to work itself out. An example that should
be put in, for example, is protection for common mode. We didn't know
any better than looking at diversity, redundancies, but today, if you
look at the PRAs, they are a much more powerful tool to use.
I think an example like that could help point out exactly
what we talk about in the structure of this. Anyway.
DR. POWERS: Good.
DR. FONTANA: Can we get another copy of that paper? I
didn't bring mine with me. Thank you.
DR. APOSTOLAKIS: So what do we do now? I am really anxious
to go through the first reading of the letter on the Safety Goal Policy
Statement because I want to get your input.
DR. POWERS: I guess we can accommodate that. I can take
five minutes to just ask Tom the outcome of the discussion on MOX.
DR. KRESS: Yes, let me see if I can recall that. When we
called upon staff, or the invited experts to give us whatever advice
they would like on the ultimate question, we got two kinds of advice.
One of them had to do with going into the MOX letter and carefully
looking at the paragraph starting on line 45 and be sure that that
clearly says what we want it to.
And the other type of advice we had was when we think about
this issue, that we need to keep in mind that the actual implementation
of higher burnup fuel and MOX, two separate things, they are two
separate things, is going to be in such a way that there will be only a
limited fraction of the core in those states. And that if one is asking
the question of whether it is permissible or not, you need to keep in
mind that concept that there is actually a limited amount. And it turns
out that that limited amount is pretty big, it is like one-third of the
core will be MOX fuel.
DR. POWERS: Well, if DOE has its way, it is even bigger
DR. KRESS: Okay. And so, you know, that may not be as much
of a consideration as one might think. But it reduces things -- it
reduces the increase by two-thirds if you are thinking about the whole
core. I don't know how much it reduces it, but it reduces it.
DR. POWERS: What it means is that the increase is only a
third a big as what you thought.
DR. KRESS: Might have thought.
DR. POWERS: If it is uniform. Now, unfortunately, it is
also the hot portions of the core.
DR. KRESS: And the same way with burnup, all the fuel won't
go to 60.
DR. POWERS: There it is really a third.
DR. KRESS: Yes. So that was the other thing that they --
and they still believe 1465 source term has sufficient boundingness in
it that they feel comfortable with using MOX and stuff, but that left me
a little cold because I don't know where they derive their comfort from.
DR. POWERS: Yeah, I guess the question is, how insistent
are that they be in a position to demonstrate that? Because what we
have seen out of that particular group of experts is a lot of pencil
whipping and not much data.
DR. KRESS: And my problem, we got into this long discussion
between the dichotomy between design basis and reality, real risk, and
my concept on how to show this is, in a perfect world, you would just
run your risk analysis with these phenomena in it is and see how much
delta risk you are going to get.
We don't have that because you don't have the models to
account for these effects and there is no data, not enough database to
develop empirical models. The only recourse one has in the absence of
such a database and models is to try to see if you can bound the
problem. And one way to bound it would be to go into a risk assessment
using a melt core or something and take the third of the core that is
MOX and the third of the core that is high burnup and bound the releases
from those by some high number, just from that part of it. Run your
risk sequences and see what that did to you in delta risk.
That would be one thing that they haven't done. I don't
know what that would do to you in delta risk, but they imply, the two
guys imply that that probably would be an acceptable delta risk. But to
me that would be plant-specific and it depends on what type you are
going to put it in.
DR. POWERS: I guess, I mean what it does is it presumes
that phenomenology remains the same.
DR. KRESS: Yes.
DR. POWERS: And the experience with the high burn, of
course, is that at least in one area it isn't the same.
DR. KRESS: It presumes that you could do something to the
releases from the high burnup or the MOX fuels, like arbitrarily
increase them to some high level, and that the rest of the fuel would
not be affected, the rest of the core would not be affected, you treat
it like it normally is. That is an assumption that, you know, is not
But, anyway, short of that demonstration, I can't buy off on
it unless we have got data or models to show me that the risk increase
DR. POWERS: Yes, I guess the question, I think the question
is, if we need to include a recommendation in the letter that says we
think that this is an issue that is going to demand experimental
validated conclusions and not just an analyses.
DR. KRESS: I think that is true.
DR. POWERS: Okay. George, let's turn to your pink pussycat
DR. APOSTOLAKIS: Okay. What I tried to do with -- well,
Mike and I worked on this, so what we tried to do was to put the letter
together in such a way that all the views that were expressed here
yesterday are here. So you see I am saying at some point that we are
unable to reach a conclusion.
Now, yesterday, though, it is true that all the members
except three thought that the idea of an over-arching policy was very
good, was excellent. So if today the Committee feels that this should
be reflected in the letter, I can easily take the same letter and give
it a different flavor and make it a conclusion, in which case, of
course, there should be a minority view. Now, I shouldn't really read
it, should I? Or just go over it paragraph by paragraph? Or maybe just
go ahead and read it?
DR. POWERS: Why don't you go ahead read it.
[Whereupon, at 2:22 p.m., the recorded portion of the
meeting was concluded.]
Page Last Reviewed/Updated Tuesday, July 12, 2016