ACRS Meeting on the Ad Hoc Subcommittee on Differing Professional Opinion Issues - October 11, 2000
NUCLEAR REGULATORY COMMISSION
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
MEETING ON THE AD HOC SUBCOMMITTEE
ON DIFFERING PROFESSIONAL OPINION ISSUES
Two White Flint North
11545 Rockville Pike
Wednesday, October 11, 2000
The above-entitled meeting commenced, pursuant to
notice, at 8:30 a.m.. MEMBERS PRESENT:
DR. DANA POWERS, Chairman
DR. MARIO FONTANA, ACRS
MR. TOM KRESS, ACRS
MR. JACK SIEBER, ACRS
MR. RON BALLINGER, Consultant
PROFESSOR IVAN CATTON, Consultant
MR. JAMES HIGGINS, Consultant
DR. JOE HOPENFELD
MR. ROBERT SPENCE
MR. SAM DURAISWAMY
MS. UNDINE SHOOP
MR. JACK STROSNIDER
MR. JACK HAYES
MR. KEN KARWOSKI
MR. JOE MUSCARA
MR. STEVE ARNDT
MR. JOE DONOGHUE
MR. STEVE LONG
MS. ANN RAMEY-SMITH
MR. GARETH PARRY
MR. CHARLIE TINKLER
. P R O C E E D I N G S
DR. POWERS: The meeting will now come to order.
This is the second day of the meeting of the Ad Hoc ACRS
Subcommittee on Differing Professional Opinion Issues.
I'm Dana Powers, Chairman of the Subcommittee.
ACRS members in attendance are Dr. Mario Fontana, Tom Kress,
Jack Sieber. Additionally, we will have Ron Ballinger in
attendance as a consultant and a member of this
subcommittee. We also have Professor Ivan Catton, Mr. James
Higgins, as invited independent consultants to the
The purpose of the meeting is for the subcommittee
to review the technical issues contained in the differing
professional opinion on steam generator tube integrity.
This review was requested by the Executive Director for
Operations to assist him with the DPO resolution path.
The subcommittee will gather information, analyze
relevant issues and facts, and formulate proposed
recommendations for the disposition of the technical issues
contained in the DPO, as appropriate, for deliberation by
the full Advisory Committee on Reactor Safeguards.
The subcommittee will hear from Dr. Joe Hopenfeld and Mr.
Robert Spence today.
The meeting is being conducted in accordance with
the provisions of the Federal Advisory Committee Act. Mr.
Sam Duraiswamy is the designated Federal official for this
Ms. Undine Shoop, a staff member who is assisting
the panel, is also present. We have received no written
comments or requests for time to make oral statements from
members of the public.
A transcript of this 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 readily heard.
Do members of the panel have any comments they'd
like to make before we start on the session today?
DR. CATTON: Just one, Dana, particularly for us,
too, because we won't be at your deliberations. What sort
of format do you want the report in? This is something we
talked about Friday.
DR. POWERS: Well, what I wanted to do, Professor
Catton, is that before you leave on Friday, I would like to
get an oral presentation of your initial thoughts, comments
and whatnot. I don't think we'll hold you to those, but at
least what you think at that time.
Then I'd like to get something in writing from you
and I pretty much leave that to your discretion. What we're
interested in is understanding the contentions that exist,
the data and analyses that exist to support those
contentions from the staff and the author of the DPO.
Certainly, the extent that you can put it in a
here's the issue, here's one position, here's the other
position, and here's what is available to support each side
of this, especially when the data and analyses are not
definitive, are the ones that are going to be the ones that
are the most difficult for us to handle.
Some of the issues, I think, will emerge that the
case is relatively clear. There is either no data, no
applicable analyses, or there are data and applicable
analyses in sufficient magnitude that the point is really
I think those will manifest themselves very
clearly. I think the ones where especially decisions have
to be made with a heavy does of engineering judgment is the
ones that we're going to be most interested in what your
Any other questions before we start?
DR. POWERS: At that point, I think I'll turn the
floor to Dr. Hopenfeld. The agenda has various breaks
listed in it, but since you're going to be doing the heavy
lifting today, at any time you think you need to take a
break or think it would be useful to take a break, just sing
out and we'll declare one.
DR. HOPENFELD: Thanks a lot. Good morning. I
would like to thank the members of the panel and their
consultants for agreeing to resolve this differing
professional opinion, also known as DPO.
This DPO has gone unresolved for almost ten years
and it is high time for it to be resolved now.
I would also like to welcome the public for coming
to this meeting to listen to my safety concerns regarding
steam generators. This is the first time that the NRC opens
the door to the DPO process. I welcome this change and hope
that it will become permanent.
I would also like to thank Dr. John Larkins, I
don't see him here, for allowing me six hours for today's
presentation. I requested this much time because there are
many subjects to cover, as Dr. Powers just indicated, and I
want to make absolutely sure that all my concerns are
Tomorrow, you will be able to judge whether the
NRC staff addresses my concerns adequately.
I believe that the likelihood for a catastrophic
accident from defective steam generators is significantly
higher; as a matter of fact, a hundred times larger than
what the NRC predicts.
This is the crux of the DPO. That's all what I'll
be talking today about.
Ten years ago, several plants started exhibiting
severe cases of stress corrosion cracking. This type of
corrosion is nasty, because it's unpredictable.
Standard engineering practice is to select
materials and environments that are not susceptible to
stress corrosion. Nevertheless, when an improper material
selection is made, delaying the proper response is not the
viable option, especially when the component degradation
bears serious safety consequences.
Instead of repairing defective steam generators,
the NRC allowed these units to continue to operate without
adequate safeguards. The recent incident in Indian Point 2
demonstrates that this policy is ill advised.
Safety is very subjective. At one time, I thought
that it was safe to drive over a hundred miles an hour. I
do not believe so now. Maybe a bad example, probably to my
reflexes, but nevertheless.
Because of this, the NRC set a standard which is
based on the proposition that risk to the public must not
exceed ten-to-the-minus-five core melts per reactor year,
roughly once every thousand years.
Since we cannot sense an impeding core melt, we
must rely on inspection and engineering analysis to prevent
such catastrophic events. To be credible, such analysis
must be based on a solid foundation. Unfortunately, in the
past decade, this principle has been replaced at the NRC by
My purpose today is to convince the NRC, with your
help, that plants should not be allowed to operate with
defective steam generators, as prescribed by the so-called
alternate repair criteria, ARC.
I recommend that all plants that currently operate
under this rule be shut down and the standard 40 percent
plugging rule be strictly enforced.
These plants obtained their license to operate
under the conditions that tube deterioration would not
exceed 40 percent of weld thickness. This must remain so.
Toward this end, my job today is to pierce the
veil that masks the alternate repair criteria to show you
that it has no technical merit.
I will be talking for the first hour about the
process and the process, to a large degree, is related to
the technical issues. However, most of the day I will spend
on the technical issues.
In the fall of 1991 -- do you have a pointer, sir?
Thank you. In the fall of 1991, the ACRS sent a letter to
the Commission indicating that the 40 percent criteria for
plugging tubes should be revised. That 40 percent
originally came from waste studies and the committee, ACRS
committee indicated to the Commission that the kind of
phenomena that we see now and that we saw then is different.
It's very shallow, very tight, through the wall or partially
through the wall cracks, and those cracks are so tight that
there is no -- one shouldn't worry about a tube burst
because it really doesn't affect the strength of the
Well, I thought that the ACRS overlooked one
important factor and that was that under accident
conditions, these very tight cracks can open up because of
the various loads that will act on the tube.
So under normal conditions, they're absolutely
right. There's nothing going to happen. That tube is going
to be safe and probably not going to leak.
But these plants were designed for certain what we
call design basis accidents and it became very clear to me
that the load that you're going to have under these
accidents are going to be such that all those little cracks
or partially through the wall cracks are going to open up
and you start losing inventory.
Now, what I mean inventory, for those people who
are not that familiar with the lingo, a reactor is really no
different than a teapot. As long as you've got water in
there, it's not going to burn.
But you start losing water and you uncover the
core, then you get to a more severe situation, you melt the
core. The difference between the teapot and a reactor is
that when the core melts, you can also burn the city.
There's another subtle difference. In the case of
a teapot, you can hear the steam whistling. In the case of
a reactor, you may not, or when you hear the -- when you see
the steam outside the building, it may be too late and the
instrumentation that we have to warn you about the possible
inventory loss is frequently not accurate enough or it could
provide you misleading information.
So I felt that you can have all these cracks
opening up and you may have many, many pinholes or a lot of
-- hundreds of cracks opening up and the total amount of
inventory loss would be equivalent to more than one tube.
Those that have some sea time or have been at sea
heard stories about the chief engineer walking next to old
pipelines with a cane, these little cracks, little jets
emanating from small cracks could be very, very small, you
can't see them, they could be very abrasive. They could cut
So it wasn't only the issue of losing inventory.
It's propagating that accident, and that was my concern at
There was another issue and it was really brought
out by Professor Lewis that the instruments that we have
have a certain limitation. They can detect certain things,
but there are limitations, and I do not think that the ACRS
really emphasized that point. They basically said go on and
reevaluate that 40 percent and come up with something
better, and they were absolutely right, but I thought it was
an appropriate time at this to bring the point that it's not
only the tube burst that is of concern, it's the total
leakage and the other mechanisms to cause that.
So we have, the NRC, a process called the DPV, DPO
process. It's a two-step process. The first part of it,
you bring your concern to the division level and if you are
not satisfied with the reply, then you take it to higher
authorities, the EDO.
Just before I came down here, I read the Inside
NRC, where the EDO is being quoted as saying "Well, that DPO
process is not a resolution, it's sort of a consensus thing.
It's a disclosure."
Well, when I wrote that DPV, my purpose wasn't to
come and just raise flags. I just expected a dialogue. I
was really concerned. I thought we should look at it. I
didn't expect anyone to accept my views. I expected, I
think, really to approach it in a professional manner, take
a look at the issue, and see what -- if the guy is crazy,
just tell him so.
Well, that's not what happened and now I can see
what the EDO says, and I strongly disagree with what he
says. He says all you got to do is just tell us that you
have -- that there is something there and then we'll decide
Now, imagine yourself, you're on the assembly line
somewhere in Akron, Ohio and working on the Firestone tires,
and you find that the epoxy mix is wrong. So you tell your
supervisor and all the supervisor listen to you and you go
back to the assembly line and that's about all that happens.
And that's what the EDO tells you. He tells you
really you just tell us what happens and we'll take care of
Now, I don't know of one case, of my own
knowledge, that a serious safety issue has been resolved to
the satisfaction of the submitter at the NRC. What that
really tells me that what he is saying, well, you just tell
us and that really gives an appearance to the public that
we're taking care of it or we consider what our employees'
Well, most professional people, when they have a
differing professional opinion, they're driven by more than
just presenting it. They're looking for resolution, and
it's a normal thing to do. And what he is saying,
basically, is I want dummies and so they want -- just tell
us what the issue is and go away.
I disagree with that kind of an approach. I hope
we can make it more effective than what the EDO claims.
Well, anyway, going back to this, I wrote the DPV
and submitted it through the channel and the next thing that
happened was that the NRC came back to me and they told me,
look, don't submit a DPO. We ought to make a generic safety
issue out of it.
Well, during that time, though, before the meeting
that we had regarding the DPO, there was another meeting
with Congressman DeFazio, where the NRC management went to
him and told him that we have done a lot of studies and we
are ready to get -- at that time, Trojan was down because of
these cracks and they told him we have evaluated the thing,
there is no problem, we can get it up to power.
Well, what they didn't tell him, they didn't tell
him there was a DPV on the subject. They didn't tell him
there were allegation within Westinghouse that Westinghouse
is providing misleading information to the NRC.
The reactor went on-line, I think, somewhere in
February, beginning of February. When the Research
Division, I think, at that time, the Director was Mr.
Beckjord, he told me come up with a GSI.
Well, as soon as he told me the GSI, my antennas
went up. The GSI is the program that was mandated by
Congress in 1978 up at TMI and the purpose there was to
address safety issues and resolve them promptly.
But that's not what happened. That program turned
out to be entirely different. It's being used, not to solve
problems, to delay problems.
If you look at your appendix, the last page there
has a summary of all the generic safety issues since '83, I
believe it is, where Congress mandated that we should keep
track of them before, but we didn't have to keep track of
What you see here is that it takes four and a half
years to resolve a safety issue. Four and a half years.
Some of those safety issues, which are really high priority,
takes all the way to like 17 years to resolve them.
Now, what does it say? What does it state? It
states really that safety issues are not a priority item at
the NRC, when you can work four and a half years on the
And another thing, and I can't give you the
specifics on that, but you can get it, that many, many of
these issues, the technical work was completed way before
the closure date.
And what does that mean? That means it is being
delayed by management. There's just no other way to read
So what you have, you have this GSI that is not
doing what it was intended to do and if I were a
Congressman, I would really like to know where my money
To summarize this table, this is a reflection of
the safety culture at the NRC. Anyway, being a good
soldier, I went back and I wrote another report summarizing
the various issues as I saw them at the time to get a
generic safety issue initiated.
The issues that I highlighted at that point were
basically that under certain conditions, you will deplete
the inventory or the refueling tank of water if the leakage
from the primary to the secondary is large enough and that
would lead to a core melt.
And I have pointed out various mechanisms. There
was jet corrosion/erosion there. There were vibrations,
there was MSOB loads. Basically, it didn't go in a very
detailed analysis, but highlighted most of the main points.
And I sent this thing to the Division of Research.
Well, they took that document and they set up a committee to
prioritize this activity.
By the way, John, you weren't here. I would like
to thank you for giving me six hours, because I am going and
going and if I'm rambling too much, you are more than
welcome to stop me.
So I went back to the Research people and I gave
them that package and they set up a committee to study this.
The number, the risk that I came up with
originally was ten-to-the-minus-four core melts per reactor
Well, this committee had done a much more thorough
study. It was chaired by a very competent man, I believe
the name was Dr. Burda, and he's not with us anymore, but I
remember that there were a lot of serious discussions.
There was help from PNNL on this. There were additional
calculations. They came up with a number like
I think it's important, I'm pointing this out
because I'm going to come back to this number later on. So
please try to remember this number. Anyway, they also
prioritized this as a high priority.
In September 1992, I provided additional
information. I just didn't have enough time and I started
it and that really relates primarily to severe accidents. I
never thought the severe accidents are as important as the
design basis accidents, but, nevertheless, for completeness,
I have provided that information.
And, again, please, try to remember this, because
I'm going to come back to that September 1992 later on and
it becomes very, very important.
Well, on November 9, Trojan shut down due to a
tube leak. Well, what happened, at that time, the press got
a hold -- and I don't know how, but they got a hold of the
DPV and some additional material, and they really went after
the NRC for not disclosing that information before.
So what happened, at this point, the generic
safety issue, which was already identified as a high
priority, went to NRR and Research asked NRR for comments.
Now, Research doesn't have to go and usually are
not required to get NRR's blessing on this, but they went
and they sent it to NRR and NRR told them drop it.
Now, at this point, for those people who are not
familiar with how the organization or the NRC, I'll be
referring to NRR and RES many times. So I might as well
tell you my perspective of who they are.
NRR is very, very simple. It's the regulatory arm
of the NRC. They basically think of as somebody has a
licensing action, a relaxation or something else, they would
come to NRR and they'll act on it.
They are like the MVA. I mean, say, if you want
to change your license or you're blind in one eye and you
want to still drive, you go to NRR and that's their
Now, Research is a little bit different and the
reason it's different is because the name research is very,
very misleading. You do research in industry to stay ahead
of your competition, and in academia, you do research to do
basic studies and produce Ph.D.s.
Research here doesn't do any of that. Originally,
26 years ago, when it split from the AEC, the intent there
was that they will do independent research. In other words,
when they develop all these computer codes, somebody will
have an opportunity to take a look at the code and say this
is an independent assessment of what the licensee is
submitting to us.
But that's not what happened over the years. Some
of these computer codes that NRC has developed were taken by
the industry and modified here and there and they came back
and resubmitted them and the action was taken appropriately.
So it's not -- the independent assessment becomes
very, very fuzzy. Five or six years ago, in a constant
surge to find the mission, the NRC management dictated the
various divisions at Research that they should produce
papers for review at high, good quality journals and they
have to be peer reviewed.
I don't know how many peer reviewed papers were
produced. I suspect not very many. You can't dictate
overnight and make people world class researchers. Most of
the people that came in have a different background. They
may be expert in many areas.
So the point here is that the function of Research
is really a support group to NRR. Basically, that's what it
is, and I just want to make somebody, especially from the
public, who may feel -- or may get the impression that this
is an independent research group. It is not.
Now, going back to January '93, remember, Trojan
was down. There was a lot of pressure on NRC to explain,
provide justification for getting that plant back on-line,
and I remember the Research Division produced several memos
and none of them really went very far.
Then Mr. Beckjord pulled Mike Mayfield, Mr.
Mayfield from Christmas vacation and, in two weeks, asked
him basically to put an assessment on the Trojan to justify
operation, future operations.
So we have -- here is somebody from the public.
I've been referring to the public several times. I didn't
know how many people were from the public, but we've got
another one there.
Anyway, so Mike came in and in two weeks produced
a very, very impressive package about the analysis of the
cracks in Trojan, and he had concluded that the leakage
would be between 33 to 1350. The risk was acceptable and
the mean here was about 145 gpm.
Now, that went out on the street and one thing
that I think the Research people forgot is that if you have
a mean leakage of 135, there may be no risk to the public,
but you cannot meet Part 100. There is just no way that a
plant can meet a 145. You need maybe somewhere between one
to ten gpm, depending on the site. You may meet Part 100.
But there is no way that you could be within the law and
meet 145 gpm.
The people, at that time, were very, very -- at
NRR -- were very, very concerned about this, but it became,
to some degree, academic, because the Trojan management
decided -- and there were several reasons, because of the
cost of electricity and they were able to buy electricity
from Canada, they decided that NRC poses too many problems
here, there are just too many letters, memos running back
and forth, and there was just too much uncertainty that
businessmen cannot be exposed to, so they shut down the
It wasn't really a technical issue as much as a
straightforward business, and I think these memos going back
and forth didn't help it.
The reason I'm showing you this is because later
on I'm going to come back to this. I'm going to come back
to this number, because as you see, this two-week effort
became later on an advance study.
In February of 1993, we find additional plants are
being allowed to operate with degraded tubes. In April,
Congressman DeFazio was very much upset with NRC that they
didn't tell him originally that there were disagreements.
So he wanted to know what was going on here.
They just misled him, that's basically what he
said. You come in here and tell us that we don't have any
problems. Then I find out that you do have problems. He
didn't think that he looked very good in front of his
So NRC made a presentation and basically told him,
look, we're going to be very tough on the industry. And one
thing I remember Congressman DeFazio said, look, you're
taking everything what Westinghouse tells you, they are
being sued. These steam generators are defective. They're
being sued and all that you're telling me is that you're
doing what Westinghouse tells you to do.
That did bother me. But nevertheless, we walked
away, I happened to be at that meeting and when we got out,
NRC indicated to him that we're going to be very, very rough
on this thing here. We have a lot of uncertainties. One
limit that we're going to set, we're not going to let
anybody exceed one volt, and I'll come back to that one volt
Between February and May, there was a task force
at the NRC and basically what the task force did was really
trying to come up with -- explain away basically the
voltage-based plugging criteria, which was really invented
It was a rationale to allow people to operate
steam generators with defective tubes. That's all it was.
Somewhere around mid-June, that activity was
summarized in a NUREG report, NUREG-1477, which is still,
however, in a draft form. But all that work that was done,
the purpose here was to set the foundation for rulemaking on
defective steam generators.
Now, this is very important, that NUREG report is
very important because I'll be coming back to this, because
it's continuously being used as a justification to indicate
that there is no risk.
So we'll go back to that NUREG-1477 and we'll take
it apart in a technical way.
In 1993, NRR management goes to the Commission and
tells them that we have had hearings with ACRS, we are very
much concerned about the -- so in November 1993, the NRC/NRR
management went to the Commission and told them, look, we
cannot -- we don't have the time, we don't have the
personnel to deal with these steam generator issues on a
case-by-case basis. We want to make a rule and we want to
set a rule on -- we have done our homework and we'll finish
it within a year or year and a half, and the Commission said
And this was a major activity. It was not a
little thing on the side. It was a major activity at the
In 1994, before the rule -- during the rule
activities, NRR decided that we need something as an interim
and as an interim, they took the findings from NUREG-1477
and basically translated it into a generic letter, which
was, at that time, called Generic Letter 95-05, and one
thing that was bothersome about that letter, bothersome to
me, was that suddenly we find more relaxation.
Remember back when they talked to Congressman
DeFazio, they said we're going to go on one vote. Suddenly
we find ourselves two votes, and it looked like the door was
open beyond that.
And there were other technical issues with the
NUREG and with the whole approach, especially in connection
of dose releases to the public violating Part 100.
So I wrote a DPO and the reason is remember when
they originally told me to go and write the generic safety
issue and forget about the DPV, they didn't know how to
dispose of that DPV. So that DPV was still active.
The standard procedures are that when you submit
the DPV, they are supposed to take an action within 30 days
and give you a response. All they told me was go and write
a GSI and you don't know what that meant. That meant let's
not do anything, and that table says that.
So I felt that I should submit a DPO, take it a
notch above the division level, take it to the EDO, and
hopefully that would be addressed.
Well, before the DPO was going to deliberate on
it, he thought that I ought to present this thing to the
ACRS, and let's see what the ACRS has to do -- has to say
So ACRS had a meeting, I believe it was September
1 or somewhere around there, and they had endorsed that GL
95-05 as an interim measure and recommended that the SG,
steam generator issue be addressed via rulemaking.
I'd like to take a little bit of time about this.
Now, the Commission takes very seriously what the ACRS
recommends. They should. They are highly knowledgeable,
but they are limited in the time that they can spend on
So they rely on what the NRR people or Research
people tell them. They take it on face value many times.
They just can't go and look at what's underneath it.
Well, let me just go, just to refresh -- I hope I
can get it right. I think it is right. Can you see it
well? If you can't, I'll have to tell you what it is.
You remember I told you that we spent a month,
Research spent a month coming up with a risk assessment back
in March and they came up with a number like
3.4-times-ten-to-the-minus-four. I came up with
This number really concerned the ACRS very much.
I remember they were really shaking their heads and trying
to find an explanation, what happened here, how do you
justify this. You have a generic safety issue with a high
-- this is more, this is orders of magnitude, order and a
half magnitude, from what the Commission guidelines,
although the Commission guidelines at that time were not
definitive. They were still thinking about it.
Nevertheless, it bothered them. It really --
another thing that bothered them, and I'll show you on the
next page, was how the staff calculates how they meet Part
100. That really concerned them.
Well, so when I'm making this presentation, and I
had only five minutes basically, Mr. Mayfield interrupted
me, Mr. Wong interrupted me and they came in and they said
now, hey, we have done advanced studies. You see, we have
done advanced studies here for 1477, which we'll go back,
again, we'll take those advanced studies into pieces and
we'll show you what advanced studies they mean, and we have
done also very serious studies for Trojan and all those show
that these numbers are way too conservative.
We also believe, and if you can read over the
wording here, it's kind of difficult to exactly understand
what they mean, but you get the impression here that the
ACRS believed that those tubes, also this discussion related
to the outer diameter stress corrosion cracking in the
support plates, and those support -- the cracking is
confined to that region, and they felt that the fact that
those tubes are confined to that support plate, it gives you
In other words, they gave you the impression that
these tubes are really constrained by that support plate.
There is nothing further from the truth. These tubes are
going to go all over the place. The support plates and the
tubes are going to get divulged very fast when you've got
this big blow-down, and we'll discuss that later on.
But nevertheless, that's what the ACRS -- that's
what the impression that, oh, this study was just like done,
some kind of a second cousin kind of approach; well, you
know, did some scoping studies. We've done some very
serious studies here.
Now, why am I telling you all this? I'm telling
you all this because later on, the NRR went to the
Commission and probably to the public, and to the public,
using the ACRS as the justification to operate these
defective steam generators.
In other words, ACRS was the rationale and the
ACRS did provide the rationale, that's true, but I'd like to
remind you -- I'd like to provide you -- I hope I've got
these things right.
I'd like to remind you that the ACRS just did tell
them go ahead and tell the Commission that everything is
okay, go ahead and justify some of these things.
One of the things that really bothered them, and I
believe that Dr. Powers remembered that very well, because
he brought up that point, justified the dose -- how you
calculate the dose releases and he even submitted an
analysis by himself, and I thought it was a good start.
And another thing that you see here, the word
interim, interim approach. It was just not permanent. It
was an interim. I don't know what the word interim means.
It could be between now and eternity, but what it really
meant behind, I believe it meant within the context of what
happened there, context of that time, what it really meant
was that we are working on the rulemaking and we're talking
about a year or two or three.
Well, I'd like to tell you, this thing is a
permanent feature now. There is agreement about to be
signed with NEI and this is part of it. You don't hear any
interim, but you do hear from NRC management that we went
through the public, we got public approval on this.
Yes, what they got public approval on was on an
interim basis, because that's the only thing the public
knows, that there was an interim approach here and they were
working on it.
Well, I'd like to -- since I brought this issue of
the advanced studies that Research has done, and that was
the reason, partially the reason for the ACRS agreeing or
going along with this, this is a memo or a letter from one
of the participants, basically, I believe, it's
Westinghouse, one of the participants of the task force
during the preparation of NUREG-1477, and what the expert
says here, he says that the model or the way the 32, the
thousand gpm was calculated and the risk, this is -- here is
the key word -- an arbitrary estimate.
So here we have an expert, not Hopenfeld, but an
expert tells them that this is not an advanced study. This
is very important, because I feel that the ACRS is, to some
degree, being used here as a tool to go to the public and
say, yeah, we've got ACRS looked at it, but they never tell
you what were the caveats behind it.
Okay. Now, on May 20, 1997, we are jumping three
years. Remember that DPO is still there and the NRR
continuously tells the Commission that we're working on it
as part of that rulemaking activity.
DR. CATTON: Joe?
DR. HOPENFELD: Yes, sir.
DR. CATTON: That excerpt from a letter you put up
there, that was a letter from Westinghouse to who?
DR. HOPENFELD: This was not a -- okay. During
the -- we spent three months preparing that NUREG-1477.
During that time, Westinghouse was making a lot of
presentations regarding what should be or shouldn't be in
that NUREG. One thing, they wanted to keep the voltage very
high, but basically everything what they said was included
in the NUREG.
That letter -- well, as part of this deliberation,
there was discussion regarding the research model and
evidently the Westinghouse people felt very strongly that
that model was just an arbitrary thing.
And my point in bringing this model here was only
to show you that what was referred to in the ACRS letter is
an advanced, a better study compared to what was there on
the record, really wasn't really that, because their own
experts or some other experts really indicated the same.
MR. BALLINGER: Excuse me. Is that Emmett Murphy?
DR. HOPENFELD: Emmett Murphy, yes. It was just
an informal note and my point here really, it's not any
formality of anything, but my point here is just to indicate
to you that I wasn't the only one questioning it.
Westinghouse questioned that, too, that this was not some
kind of an advanced study.
But nevertheless, on the record, and that's why
the ACRS agreed that this was one advanced study and there
was another advanced study that's in the NUREG-1477, and Mr.
Wong was talking about and we'll talk about that later.
So somewhere in 1977, mid-1977, the -- well. I
believe that you have the -- I believe that one of my
transparencies disappeared here, so I'll just talk off the
top of my -- what I remember. I probably misplaced it or
But before -- I'm terribly sorry about this. Let
me go back and I'll go by memory. Oh, thank you very much.
I think page five here. Yes.
Page five, for some reason, came out in my
transparency. On break, I could take it, but you can look
on your page five and we'll go through this. I'm terribly
sorry about that.
On May 20, 1977, the NRR informs the Commission
that they have discovered potential failures during severe
accidents and, therefore, they would like to drop the
rulemaking activities and, instead, go and resort to a
generic safety issue, GL-95.
Before they informed the Commission that they had
problems with the rulemaking, six months earlier, the NRR
management went to the Commission, and I thought I had
another viewgraph to highlight that, maybe I didn't, they
informed the Commission that they are just about to get the
rulemaking out and this is going to be a precedent-setting
rule, it's going to be a backfit.
However, six months later, they go to the
Commission and they said there is no cost-benefit in doing
Now, imagine yourself being a CEO, you're going to
the board and you said I have a new product here and I need
some money to work on it. You work on it for three years
and just about when you finish, you say everything is okay,
the trucks are ready to deliver. Four months later, you're
telling I just found out that the bottom line is not there
and, therefore, I've got to drop all the rule.
In between, while the activities under rulemaking
were going on, there was a resentment on the part of the
industry that they felt that the rulemaking was too -- the
rule itself was too complicated and there was an indication
that they didn't like it.
So we now find that suddenly the rulemaking is
dropped and the rationale that's provided, the rationale is
that we found some new problems with severe accidents. One
of the problems that was alluded to was this jet that I told
you before affecting the adjacent tubes.
So now we find an excuse. Now, mind you, that
back in September 1992, there was -- on that DPO, there was
a discussion on the severe accidents, where I have very
clearly indicted there is a potential problem there.
So here NRR is working for three years on this
major activity and telling the Commission everything is
okay, but then six months later, suddenly disappears. Now
we've got to work on something else. We're going to start
on a generic letter, which is a much lower -- has a much
lower hierarchy in terms of its importance.
So they get an okay to work on the generic letter
and there's a year of activity on that letter and it went
through -- it was so complicated, nobody could understand
and NRR people couldn't figure out what it was.
So finally they decided, well, we're going to drop
that and we're going to a regulatory guide. Now, in June
1999, this regulatory guide, by the way, went for public
comments in the summer of '99 and in June, the industry
requested -- they had a whole list of comments and rationale
that NRC dropped the regulatory guide and the NRC did.
So we have, since 1993 to June 1999, activities
going on relating to all kinds of safety issues regarding
steam generators and what do we have, what's the bottom
line? Nothing. Nothing comes out of all these studies.
Now, we start working together with the industry
to come up with an agreement. Well, the industry didn't
want any of that stuff to begin with.
So in June 1999, we, again, repeating it, we, the
regulatory guide is dropped out and you have a whole new set
of regulations that are being discussed with NEI and I
understand that the resolution or the agreement is planned
for the beginning of next year.
Meanwhile, while all these activities are going
on, nothing really says much about the DPO, except it went
for public comments and when it came back from public
comments, the NRR people made further changes in their
assessment of it and basically that's where it stands today.
Meanwhile, while all these activities go on, we
see there are another 17 reactors, as of June 1999, allowed
to operate with degraded tubes. We've got all these
assessments going on, but the bottom line is 17 reactors
operating not in accordance to the safety rules or the
safety guidelines that the Commission had set.
At this point, where nothing was happening --
well, there was one thing happening which I thought was
very, very significant. Farley came in here for a
relaxation. I think it was to allow them to operate without
a mid-term inspection or mid-cycle inspection.
Now, Farley was very, very significant. It was
the first time, it was the first time that the proposition
was that if they come in for a relaxation, they would have
to do it on the risk -- under the new policy of risk-based
regulation. So they would have to use some kind of a risk
One comment that came from the industry in June
1999 was that we do not want to or we do not know how to
assess severe accidents. Take it out. We don't know how to
Well, the NRR feared very strongly that they
should assess the severe accidents. So they told them if
you want relaxation, you better come back and address the
issue of severe accidents.
So suddenly, three months later, Farley comes back and makes
an assessment regarding the -- or talks about the severe
accidents, and the staff writes a report and says we believe
that there are no problems.
Well, one problem that came up was going back to
this jet issue, the staff found out that if you have very
small cracks, which you cannot detect, they still could
cause you a potential accident propagation during severe
accidents. They neglected that aspect of design basis.
And Farley evidently had a potential for small
cracks. So they said, well, we believe that there is no
Now, think about this for one second. Just think.
The severe accident is a very complex scenario. It's
extremely complex. Most people cannot -- it's being used to
analyze things. You don't design for it.
Well, here the staff tells you that they believe
that there is no problem, and they have experience. They
understand it. But what is your experience to tell somebody
that certain cracks are not going to be there during this
very, very complex scenario?
My answer is none. And this is important, because
we're getting into this phase of risk-informed regulation
which can be very, very subjective and it can be abused, and
I think this is a very good example where you have a
problem, you identify the problem, the people identify a
problem and then when you have to take an action, say, well,
we believe that there is no problem and that's enough, and
that's sufficient to pass and get that plant operational.
Now, I am not arguing about severe accidents. I
personally am not sure that it should be there. But if the
policy is to include severe accidents, then this is a
concoction of a story. This is just -- so I write a letter
to the EDO and the only reason I wrote a letter to the EDO
at this point was because it set the precedent for how we
deal with risk-informed regulation.
If there is no seriousness behind it, then
risk-informed regulation is just a joke. That's all it is,
and that's the reason I thought I would just voice my
The reply I got from the EDO basically said, well,
just we believe that the staff knows what they're doing.
And as the EDO knows, they have the insight and experience
about severe accidents, and I don't know, we haven't had
many, but they know what's going to happen there.
Now, as we go along, I will show you why this is
complete nonsense, because these reactors were not designed
for these conditions.
Okay. I would like to give you my own perspective
on this. I realize that at 40 percent through the wall
criteria, which has been around from day one, imposes heavy
financial burden and we have tried now for ten years to come
up with something better or something different and we
haven't been able to.
So you must conclude that if we haven't come up
with anything better, that we should still go back, that we
should go and retain that 40 percent. It does have some
theoretical basis to it, maybe very little, because we don't
experience -- this is for corrosion, but the probe, the eddy
current probe has a limited sensitivity and some studies
show that it's really limited to 40 percent.
So there is some rationale, whether it's -- it has
served us well. That doesn't mean that it solved the
problem, but it served us well and we cannot go now and
experiment with something that we don't know what we're
So the other thing that you get from this several
years of experience, what you get is that what the industry
wants is that they would like to have an infinite -- not
infinite -- wide margin of freedom as to decide which tubes
to plug, which tubes not to plug, but, at the same time,
they want the NRC approving it.
So when Con Edison got stuck recently, the first
thing, they went and said, well, you know, we were right,
NRC approved it.
So there is an interest on their part to have the
NRC ultimate say-so, yet they want to run the whole thing
and they are.
Now, if you look back, again, the decision of
leaving these degraded tubes in place, allowing these steam
units to operate, continue operating with degraded tubes was
made back in 1992, early 1992.
We didn't have any data at that time. I'll take
it back. There was very little meager data that came from
Westinghouse, very, very little. But there weren't any
different -- I'll give you analogy.
If you're taking Firestone tires and putting them
on a tricycle and look at the data, that's what we had. So
now the management, they grabbed that. It was there. They
took that and they said, okay, we believe this is safe
enough and we'll just let it operate.
And they built a machine, put in place a machine
and they hired people that shared that vision. Most of
those people are not here anymore, but some of them are
here. So the thing is still moving, but there's no
difference as far as this belief that we can operate safely
with those cracks.
Now, again, safety is a very subjective issue and
it could be very well that you can operate this for a
thousand years, but that's not what are our guidelines. We
operate on the basis of risk, and my purpose here today, and
I'll go to the nuts and bolts of this to show you where the
Okay. You have to have some kind of a rationale
when you replace something. You have to say, well, I've got
a better mousetrap. So going back to early '92,
Westinghouse invented what's called voltage-based plugging
criteria. In other words, we're not going to measure the
thickness or not going to base our plugging based on that 40
percent wall thickness indication or degradation.
We are going to base this thing on the basis of a
model. But if you trip down everything what they say, it's
a strictly unproven theoretical model. That's all it is.
There is nothing else behind that. It's an unproven
So you look into that now, what's behind it.
First, it's very nonscientific. This is very easy to --
DR. KRESS: Joe, is it all right to interrupt you?
DR. HOPENFELD: Yes, sir. Please, interrupt me
anytime, because it's easier for me if we talk and if I lost
one slide, I'll go and look for it.
DR. KRESS: When the ACRS reviewed the
voltage-based plugging criteria, I think their view was that
it was strictly an empirical model, without any real
technical basis behind it.
Although they didn't -- the ACRS, as I recall,
didn't have real problems with the empirical model. Their
problem was do you have enough data to support an empirical
model and is that data covering the ranges of things of
interest, such as the pressure difference, the crack size
and the crack characteristics.
So I wouldn't call that a nonscientific model. I
would just call it an empirical model.
DR. HOPENFELD: Let me amplify that, because I
understand that's -- going back to the letter you wrote to
the Commission, that was your view. My view, and I'll
elaborate on that, I'll going detail and elaborate why I
think it is nonscientific and why I don't think it's
DR. KRESS: Okay.
DR. HOPENFELD: It is not empirical. It's purely
theoretical. Now, I'll tell you, to be more exact, you can
-- some people refer to something like in the literature,
you find references to this as semi-empirical. Now, I don't
know what semi-empirical means. It's not -- let me put --
I'll give you -- I'll go back.
Sir, please notice this is my perspective and I
call it analytical and I'm going to back -- I'm going to
tell you why I think it's analytical. It goes back to the
crux of this voltage measurement. Until we get there, it's
difficult, but I understand that was your perspective and I
disagree with that letter.
DR. CATTON: Joe, just to make a comment on this.
A very complicated area is interaction of fields with
heterogeneous media. As soon as this thing has cracks in
it, it's a heterogeneous media, and a real simple example of
how badly you can conclude what's going on is a simple
device, a little heat exchanger.
If you look at the literature on these kinds of
devices for a heat transfer problem, which there it's the
interaction of a temperature field with in-flow and so
forth, you find, in the friction side of it, you'll find
several decades difference, and the primary reason is you
don't have the right variables in the equation.
So what people put on paper is not really an
empirical relationship that's any good for anything other
than where the test was.
So if you don't describe everything, and this
means geometry, what the interfaces look like, everything,
small changes can make huge differences in what you measure.
And I think this is the same.
DR. HOPENFELD: Ivan, I'm really bringing you --
I'll get to the equations of this, so we can see the
parameters that play there, exactly what you're talking
about and any feedback would be greatly appreciated, but
that's exactly the bottom line.
This thing is too complex to call it -- it's not
semi-empirical, but we'll go back. I'll grant you, there is
some empiricism in there, there is some.
I'm sure that you have gone through all the
volumes of material, backup information, and you have to
agree with me that there is a lot of statements there,
assumptions that you don't know where they come from.
The main -- another problem is that all these
things can be easy -- the answer can be easily adjusted and
I will show you that as we go along. And the bothersome
thing, extremely bothersome and makes constraints on us, is
that a lot of that stuff is being stamped proprietary.
Some of that is so obvious. Obviously, there is
no competitor that sits there and they're going to re-derive
F equals MA. But it is stamped proprietary and then you go
back and you're trying to find what the references are and
then you have these agreements and the stuff goes out for
If I was in the public, if I had the time, there
is no way I could follow what it is, because it's all
I am, to some degree, restricted in talking about
some stuff, because I don't want to get in trouble with any
of the lawyers. So I'm restricted on giving you any
numbers, but you have all the information in front of you
and I will try to point out where the problems are.
But some of the things, it's ridiculous to call
Now, at this point, I think I'm going to start
with the technical issue. I mean, enough with the process,
I think. Unless you want to break, that's fine with me.
DR. POWERS: We are scheduled for a break. Maybe
looking ahead at the slides, this would be a good one to
complete and then it looks like there is a nice place for a
break on the next slide.
I think, incidentally, that the committee is
persuaded of your qualifications to address this issue.
DR. HOPENFELD: I'm sorry?
DR. POWERS: The committee is persuaded of your
DR. HOPENFELD: I brought it for a different
reason. I brought it for two reasons. Basically, one, I
listed some in my statements. I forgot to list here that
I'm also on the ten most wanted at the NRC. But that's not
DR. CATTON: The most important is the fourth
DR. HOPENFELD: Yes, that is kind of important.
We're about the same time we got out.
DR. POWERS: I think there's a conflict of
interest here somewhere.
DR. KRESS: What does the UCLA stand for?
DR. POWERS: I thought USBC was getting all the
Nobel Prizes now.
DR. HOPENFELD: I don't know about Nobel Prizes,
but we had a pretty good football team when I was there. I
don't know what they're doing now.
But anyway, going back to that, what I really
wanted to -- the reason I'm really showing this is that I
was personally involved in three different steam generators.
One was the element, the R steam generator, and there was --
you had sodium on one side and you had water on the other
side, and then I was involved in advanced fossil power plant
for many years in the design for a program that I want to
But anyway, it's high temperature, high corrosion
environment, and I was involved in several PWR testing for
steam -- under steam line break and feed line break for the
MB-2 program, which was really a prototypic -- the first
time anybody took a prototypic steam generator, basically,
and sliced from a steam generator and took a look at it.
What I'm trying to tell you here is that these
various steam generators have certain things in common, even
though their operating conditions are entirely different.
One thing, I remember spending years really
arguing what should be the design basis for the LMFBR. It
was a hockey stick type of a steam generator, what should be
the design; should we use one tube, should we use three
tubes, should we use four tubes, and it was going on and on.
You had all kinds of rationale for it and we
finally came up with, I believe, it was like five, there was
one in the center and the forest around it.
About eight years ago, there was an accident in
Dounray, Scotland, and you know what the -- the steam
generator went. Do you know how many tubes ruptured or got
deteriorated? Forty-eight of 50. And now I can take the
slide off, because that really was my main point here.
What I'm trying to say is that when you do the
design stage, you come up with your best estimate. I hope
that later on, I'll ask Mr. Spence to talk about that. He
may mention it. We have a steam line break as a design
basis accident. That doesn't mean that a full steam line
guillotine type break is the worst case. It may not be.
What I'm saying is there are uncertainties here
and we shouldn't worship this. And is that time to take a
DR. POWERS: Yes. We'll recess till five after
DR. POWERS: Let's come back into session, and I
will turn the floor back to Dr. Hopenfeld.
DR. HOPENFELD: Thank you. I would like -- while
I was flipping the pages before, one slide came out and I
didn't notice that. I'm not going to harp on this for too
long. I just want to make one point here.
That there was a meeting of the ACRS in November
1996, and the ACRS was echoing really industry concerns
about this whole rulemaking thing. The industry said we've
got this ASME code that already takes care of it, and they
were concerned about it.
But as far as the DPO, I came back and indicated
to you -- in fact, I even said that what was presented to
you previously was really misleading.
And at that time, I believe you stated to the
Commission, the ACRS stated to the Commission that the NRR
should resolve the DPO and the generic safety issue before
they issue any rulemaking. These should be resolved before
anything else takes place.
And that was the commitment that NRR, the EDO, at
the time, made, that he is going to resolve the DPO and the
GSI and this is going back to 1996. We are four years
later, where do you think the GSI is? We'll tell you later.
DR. POWERS: It goes without saying, the committee
has commented frequently on the pace of resolution of GSIs.
DR. HOPENFELD: Thank you. Yes. That is true. I
was trying to give you that data and I think there is much
more to that data to extract for it and I think we can make
it positive, but I really just wanted to highlight that four
and a half years and the 17 years, which hits everybody,
anybody that looks at it. What's going on here?
I'd like to give you a little bit of feel for the
background, mostly for the public, and I may be boring you a
little bit. But the importance of this accident is really
that you have a heat exchangers, with acres and acres of
land, so there's a lot of surface area. So there's a lot of
problems -- there's a high probability that if you operated
those with a lot of cracks, that something will go wrong.
It's a very important component. It's a
safety-related component, because it -- remember, we have
three barriers there for safety, cladding, coolant and a
containment. In this case, there's no containment. You
bypass the containment. So you're losing a major barrier
And there was an earlier study, I believe it was
in the mid '80s, NUREG-0844, which concluded that really the
steam generator is primarily -- it's a financial concern if
you have -- it's not a safety-related component.
It's primarily what drives the inspection and the
maintenance of that is -- it has no really major safety
The thinking behind that was that if you operate
with good steam generator tubes, there is justification. If
the unit operates all the time, you don't expect it's not --
it wasn't designed to have major disasters, and that was
So the main difference in opinion between myself
and NRR is that they believe that if you operate with
degraded tubes, the risk is acceptable and the DPO position
is not acceptable, as I said before. It's the crux of the
Now, what is this accident? Most of you probably
know. You have, at a given time, at any given time during
the operation, you may have a steam line break and if you
have a steam line break, you depressurize the system. Just
like when you drive your car and the hose breaks and all the
steam is going to come out. You just depressurize all that
And other accidents, we have two separate
accidents which these plants originally were designed to.
They were designed for a steam line break or they were
designed for a tube rupture, but not the two at the same
So you could have a steam line break and you --
there's no radioactivity escaping, except a minor amount,
because the safety valves will isolate, but basically it's a
-- the system is designed for.
You also have a tube rupture, and, again, this is
sealed, so you don't have an accident. It's not part of the
Now, when you operate with thousands of cracks,
what happens to that design accident? Originally, you
assumed that those tubes are perfect just the way they came
from the mill. Well, they're not the same anymore.
So now what happens is when you have that accident
occurring, you depressurize the system and you put different
loads on those tubes and suddenly you have all those cracks
that you think that they are tight and sitting there opening
up, and when they open up and if the operator cannot control
that accident or cannot depressurize the system, then
eventually you run out of water.
This is your refueling tanks which keeps the --
it's a storage tank. Well, you can say, well, so what, big
deal, you know, I ran out of water, I'll just go and get
some few helicopters, I'll pump some more water in there and
eventually I'll do something.
It's not that simple and the reason it's not that
simple is because this water has to be borated and if it's
not borated, you run the chance of recriticality in the
core. So you can't just get any old water. There are
procedures you have to go and make sure it's borated
So that was the issue. You see, it's a risk
issue. And that's what we'll be talking today in more
So basically, originally, we had -- the plants
were designed for a steam line break, main steam line break.
Okay. That's what you design. You don't worry about a tube
rupture or anything happening at the same time.
And these are basically the criteria for what you
designed it to. Then you design the plant to withstand a
tube rupture, and, again, these are the conditions that you
design to. You can withstand a 600 gpm.
However, there are differences. Notice there are
differences in what occurs during the transient. In the
steam line break, the depressurization is fast and,
relatively speaking, the steam generator tube rupture, it is
For one parameter that comes into calculating
off-site doses under requirements of Part 100 is to know
what the iodine spike data is. It's available for SDTR. It
is not available for a main steam line break.
So my first presentation on the subject back in
'92 or whatever it was, '94 maybe, I have indicated to you
that if now we are allowing these plants to operate with all
these cracks, we are not talking about steam line break
anymore, nor are we talking about steam generator tube
rupture. We're talking about a different material. We're
talking about a different accident.
DR. CATTON: What is the MSLB/L?
DR. HOPENFELD: Main steam line break with
leakage. The L stands for leakage. I'm sorry, I didn't
make it clear.
DR. CATTON: You did, I just didn't look at the
DR. HOPENFELD: And I outlined that, I circled
that leakage. So that is really the difference. You have a
fast depressurization and now you have more than 30 gpm
coming out from the primary, and that depends on how many
cracks open up.
I claim I don't know how many. Maybe none. But
there is a strong support for the proposition there would be
many opening up.
But, again, the point here is if the leakage is
fairly small, the operator can take care of it. There is no
problem. I don't want to scare anybody.
But if the leakage is large, there are so many
things happening, and we'll have an operator talking to you
today, so many things happening here that you will -- there
is a high probability that you'll melt the core, and it's a
question of risk, because there is no way of absolutely
saying that this will happen or not happen.
Now, I have relatively an easier job than my
friends at NRR, because I take the position when I don't
know something and public safety is my main concern, I'm
going to be conservative. I'm going to err on the side of
Well, if they don't want to take that position,
which is fine, I think the burden on them to come and
explain to you all their beliefs, all their judgments, where
they come from, who are those people, what's their
background. I think you're entitled to know that, because
Okay. The procedure to justify operation with
degraded tubes is as follows. The main assumption is that
you are safe to operate if the voltage that the probe reads
during the refueling outage, if the voltage reads, the probe
reads voltages less than two volts, one or two volts, or --
and here is the caveat that they have -- or higher, by
special approval, and you know what that means. It can run
to anything, although they have been limiting themselves to
Originally, they started with one, then it went to
two, and now we're at three, but it can be more.
Now, how do you decide what happens next is
because these are the voltages during the outage. Now, you
really want to know what happens during the cycle, so you
have to figure out what is going to be the voltage during
the 18 months period that the steam generator is going to be
in service. That's called end-of-cycle voltage.
Then you have the requirements or the
specification says that you should be limited to 150 gpd,
gallons per day, for operational leakage. You're allowed,
per steam generator, 150 gpd. It used to be 500, now I
believe it's 150.
But all that really says, and that 150 has been
around for many years, it really has nothing to do with
operating defective steam generators. It's something that
you can measure whether you exceed 150 or not.
So it's an operational limit. But then you have a
limit of one gpm under steam line break conditions. What it
says is during a steam line break, you cannot exceed one
gpm. Now, you ask yourself, what kind -- can you measure
that when you have a steam line break? You can't do
I mean, you don't impose on somebody a condition
where you can't measure, you can't control, you can't do
anything with it.
DR. BONACA: Excuse me, a question. So the one
gpm is the one in the tech specs, right?
DR. HOPENFELD: That's correct. But this is
derived for the steam line break.
DR. BONACA: I understand.
DR. HOPENFELD: At that time, going back to my
table, they were thinking in terms of good tubes. They
didn't have these degraded tubes. So that one gpm, you
could say, yes, it's a reasonable number. But now we have
all these thousands of different cracks and how are you
going to dictate to them that they're going to stay with one
gpm just because you want to. That's exactly what they do.
So the bottom line here is that -- now, the one
gpm can even exceed it if the tubes are confined. Remember,
those support plates, the NRC believed -- or the NRR people
who designed this believed these act like O-rings that will
hold the thing. Anybody with any design, has experience
with O-rings, you know that that is not a -- even for a very
simple component, it's not an easy thing to design.
Dirt gets in there and motion, so they believed
they're going to be so tight that the leak is not going to
get out of the support plate region.
So basically, if you look at this, if you really
look at this and please think about this, before we had
degraded tubes, we had basically the same specification as
we have now after a degraded tube, after we allow them to
operate with degraded tubes.
Another thing that's very interesting is note that
in the steam line break, there was one gpm, and I'm not
going to argue whether it's one gpm, ten gpm, whatever that
is, we had one gpm.
Well, let's take a look at what happened at IP-2
or other reactors that had experience with large leakage.
Well, the reactor -- the NRC said they shouldn't exceed one
gpm and that wasn't under steam line break.
If there was a steam line break, obviously, it
would be much more. So this, in a sense, this thing here
has no meaning. You can't measure it. You can't do
anything about it. When that steam line breaks, we're going
to have any leakage that the plant decides is going to
It depends on what the forces are going to act on
the tubes and not some dictation by a regulation.
But anyway, that goes back to this, to the risk
that we're talking about. NRC assumes that if they follow
this procedure, this procedure will keep them from putting
the public at risk at a higher than ten-to-the-minus-five
per reactor year, which is the Commission guidelines for
safety. That's what says it's safe or not safe.
Now, whether it's safe or not safe, I don't know,
but that's the standard we have to live to.
DR. BONACA: Let me just ask a question, Joe. At
one gpm, however, it was -- it is a number which is tied to
the dose at the site boundary. It implies also one percent
failed fuel, I believe, in the reactor.
DR. HOPENFELD: Right.
DR. BONACA: So I'm only saying that I don't think
the goal was just the one gpm. The one gpm is an assumption
made in the tech specs that goes with the assumption of one
percent failed fuel, and typically plants run with one
percent failed fuel.
DR. HOPENFELD: I understand that and I'm going to
discuss this. But this is working, you want to make sure
that you stay within Part 100, and that is true. But so
what? It's still -- what drives this thing is not what you
want, what the SRP wants. What drives this thing is what
DR. BONACA: I understand.
DR. HOPENFELD: So you can say, well, I've got
this Part 100 and it says that I shouldn't exceed five gpm
or whatever, but I'm just showing you that evidently the
system is not interested what the NRC tells them, because if
it was, each time you have a steam generator tube rupture,
you wouldn't see 150 or 200 or I think they've gone as high
as 600 gpm. You wouldn't see that, because this thing
doesn't allow that.
DR. BONACA: That's the question, because thinking
about the --
DR. HOPENFELD: I understand where it comes from.
DR. BONACA: The actual limit is there. So in our
estimating that, we have to take account of leakage,
whatever that is. I agree with you. The one gpm just was
an assumption there.
DR. HOPENFELD: But I understand the assumption
and I think it's okay if you stay within what we were
talking originally, a steam line break. But that's why I
put the thing in the third column there. We are not talking
about steam line break.
Unless you can show that these cracks are going to
stay within that region, and that's probably what they're
trying to say and that's what's going to be -- we can argue
it. That's why I'm bringing all this at this point, so
we'll focus in on that issue.
So there is a need to fix this. If you want to
operate with cracked tubes, you fix this one, because either
you can measure it -- if you can predict it, fine, but the
issue is can you really predict it.
DR. BONACA: I don't want to belabor it, but, for
example, I could say, okay, I am going to fix the amount of
allowable failed fuel not to one percent, but to one per
thousand and then I allow a larger leakage.
DR. HOPENFELD: Right.
DR. BONACA: I'm trying to point out that that
number was part of a product that ended up with the dose
DR. HOPENFELD: You're absolutely right. Look, if
you can have perfect fuel with no cladding cracking, it
doesn't matter. But that's not the real world. There is
It probably, and I don't know, it probably -- the
lawyers probably got in there and it's probably in the
warranties going between the fuel manufacturer and the
supplier and the utility. There's probably some verbiage
there that the lawyers put in, but I'm not looking at it
from that perspective. I'm telling you this number doesn't
mean a thing.
DR. KRESS: I'm not quite sure, Joe, I understand
your last bullet on that slide.
DR. HOPENFELD: Which one, sir?
DR. KRESS: The last one.
DR. HOPENFELD: This one?
DR. KRESS: Yes.
DR. HOPENFELD: Okay. The reason you have all
this procedure of controlling how much leakage you're going
to have or the rationale behind this is you don't want to
exceed the ten-to-the-minus-five core melt per reactor year.
If you were to say, well, if this is going to be,
say, a thousand gpm, it could very well be that you will
exceed the ten-to-the-minus-five, because the operator is
not going to be able to handle it.
DR. KRESS: Okay. There's more to it than just
DR. HOPENFELD: Correct.
DR. KRESS: There's operational procedures and
DR. HOPENFELD: That's correct.
DR. KRESS: So you're making --
DR. HOPENFELD: What frequencies?
DR. KRESS: The frequencies at which this main
steam line break could initiate.
DR. HOPENFELD: Correct. I'll go into that. Yes,
DR. KRESS: So you're saying there's a lot more to
it than just those procedures.
DR. HOPENFELD: Yes, yes, yes. Absolutely,
there's a lot more to it. This is ten-to-the-minus-five.
That's what they are saying. That's their -- you see, the
NRR people said we -- by doing this, we will guarantee the
public that we are going to exceed that ten-to-the-minus
five; we're going to have ten-to-the-minus-seven or whatever
they said, six-times-ten-to-the-minus-six, and we're going
to guarantee that.
Now, how are they going to do that is going to be
a subject we're going to be discussing this. But this is --
I'm trying to focus on the issue. This is the issue of
going back to what's safe is safe and the
ten-to-the-minus-five is a number.
In order to meet that number, they'll give you a
rationale tomorrow why they meet it. And I will give you a
rationale today why they not meet it.
Now, I have an easier job, because I can err on
the side of safety and they don't want to err on the side of
safety. But don't let them out of here and tell them, well,
we believe, because that's what they told you previously.
DR. KRESS: I was assuming that there were two
DR. HOPENFELD: Sir?
DR. KRESS: I was assuming that there were two
relatively independent objectives. One was to assure you
didn't exceed the 10 CFR 100 doses and then there was
another objective of not exceeding that value of risk, which
has a lot of other things associated with it.
DR. HOPENFELD: Right. I do have that in the next
slide, we'll be talking about that.
DR. KRESS: Okay.
DR. HOPENFELD: I don't know whether you can
divorce them. In real life, I don't know how you divorce
DR. KRESS: They are related, of course.
DR. HOPENFELD: They are related. I don't know if
you can say, well, today, we're not going to exceed this,
we'll stop as soon as we -- yes, if I had such a mechanism
there, a shutoff mechanism that cuts me off as soon as I go
over that one gpm, yeah, I'm okay, but I don't think we have
anything like that, because nobody ever invented one yet.
DR. BONACA: I had just one more question, which
is more to help me in the review. You pointed out that
clearly if you have large leakage rates, the success of the
operator is much more questionable.
DR. HOPENFELD: Correct.
DR. BONACA: And it becomes even more questionable
the larger is the leakage rate.
DR. HOPENFELD: Correct.
DR. BONACA: And I've been looking at some of the
sensitivities, again, to make my judgment on, and I've been
looking at this INEL report, that's the 1996 INEL report
with the sensitivities.
DR. HOPENFELD: Right.
DR. BONACA: Is that the right document?
DR. HOPENFELD: I believe it's one, but we're
going to spend a lot of time on the operator action today.
As a matter of fact, since I am not an operator, I asked Mr.
Robert Spence to talk about that aspect of it and he will
answer your question.
I really didn't prepare myself too much about the
DR. BONACA: No, just I'm trying to understand
DR. HOPENFELD: Yes. I'll be glad to -- I'm
familiar with the report, but the detail of operation, I
will just give you just an overview of it.
The main point is, focus their attention, again,
they say, well, we are about ten-to-the-minus-five, and I
say, no, ten-to-the-minus-four and the other research report
also says around ten-to-the-minus-four, and that's what
we're going to be trying to --
DR. KRESS: Joe, not to belabor this too much.
DR. HOPENFELD: Sure. We've got plenty of time,
DR. KRESS: Okay. I understand. What would you
say to a condition where the leakage was such that you
exceeded 10 CFR 100, but the risk was still actually below
ten-to-the-minus-five? I can conceive of that happening,
depending on how --
DR. HOPENFELD: Sure. Sure.
DR. KRESS: Is that acceptable or is it --
DR. HOPENFELD: Okay. I'll answer the question to
you, because I used to drive very fast when I was younger.
But I think all of us drive 80 miles an hour, okay, and
nobody is going to really worry about it. When you go 200,
you start worrying.
So if you go -- and I think the cutoff number,
depending on how you do this, is something like five gpm,
depending on the site, it's a site-specific kind of thing.
DR. KRESS: It doesn't take much leakage.
DR. HOPENFELD: So if we're talking about five or
ten or 50, we go to 145, 150, that's what the Research
people came out with, remember back to '92, Trojan, they
told him it's going to be -- the mean is going to be 144,
and they probably thought, hey, how am I going to meet Part
100 on that.
We will have to do a lot of nobbing to get the
thing down somehow. So to answer the question, yes, but I
-- yes. Maybe you should look into the relative risk of
these two, I don't know. I haven't looked at it.
Just to make it clear as to -- I verbally described to you
what they're doing. It may be easier to describe here.
What they do, they have -- they're getting readings from the
field as to what the voltage distribution is on all the
tubes or sample or sample of the tubes, and then they adjust
the thing by voltage growth to the cycle.
Now, I got to -- I'm going to spend some time
about this, because this is a major assumption and those
people who are stress corrosion experts probably would know
that stress corrosion, there are two parameters that operate
in stress corrosion. One is initiation, another is
And you, as a rule, really cannot say that the
historical data can be projected into the future. You can
maybe say that in fatigue cracks, where you can count the
number of cycles, but when you talk about stress corrosion,
which is a much more complex phenomenon, it depends on the
environment, it depends on the stresses, it depends on the
chemistry, it depends on the material, you have --
The process is so complicated that you cannot say
what happened in the past is going to happen in the future,
but that's exactly what they say, with something that they
cannot even measure.
So the next thing, what you have is you take all
this thing here and you put some uncertainties in it and you
come up with a distribution at the end of the cycle. And
then after you do all that, you still have to tell somebody
what the leakage is. So you take and you say I'll take my
end of the cycle distribution of defects and I will multiply
by something, some factor to say -- to determine whether I
will or will not have a leakage, and you see these are two
And I see one member here from Research that's in
there and maybe he can help me, if he wants to, is that
basically you can provide -- put any distribution between
these points. You can draw anything you want to.
The NRC, and we'll go to this, claims that they
have a distribution log logistic, and I'm not a
statistician, but he can tell you, one of his contentions
was that the log logistic distribution is not conservative.
What does that mean really? If you wish, he can
tell you later on.
The next thing, what is being done, they take this
-- all this distribution, multiplying by that probability of
leakage, and they put -- go to the data from the laboratory
and they come up with some kind of a leakage rate during a
steam line break, and that goes to calculate that Part 100
and that goes to calculate and they put additional
uncertainties on it to come up with and tell you what the
But that, in a nutshell, is illogical. You have
to go item by item and start probing into really what this
means, but that's just the overview of the whole picture,
the way I understand it.
DR. KRESS: Excuse me, Joe. What's the little
dots on the middle curve?
DR. HOPENFELD: Those?
DR. KRESS: Yes. The ones on the --
DR. HOPENFELD: This is the data. This is whether
you leak or not.
DR. KRESS: It's your scale. Okay.
DR. HOPENFELD: This one, yes, this is the
voltage. They take specimens and --
DR. KRESS: That's not a data point.
DR. HOPENFELD: Yes, it is.
DR. KRESS: It's a data point.
DR. HOPENFELD: Could you please it explain it
better? It's a yes or no thing. It's a fail or not fail.
DR. KRESS: So you have a bunch of data points
down there and a bunch of data points here.
DR. HOPENFELD: Right. And this is a logarithmic
scale, so they drive all kind of -- would you like to very
briefly say something about the logistics thing?
DR. POWERS: You need to use a microphone and
identify yourself. It's kind of selfish, but I'd give you a
little break here on my throat.
MR. BUSLICK: Okay. There is no theoretical basis
for using a log logistic curve for this response problem.
So a logical thing to do would be to try to use different
curves, different families of curves, like a Kochi and a lot
Kochi, normal, loss normal, and see how the goodness of fit
for these different curves, families of curves are, when you
use the maximum likelihood estimate of the parameters for
You want to see how good and see what the
differences of results are. I found that, if I recall
correctly, I could give you a reference, it's in the PDR,
that the log logistic was one of the least conservative,
underestimated the leakage.
That all of the families of curves fit about the
same, the goodness of fit characteristics were about the
In the cases that I examined, if I recall
correctly, the changes in the leak -- in the estimated leak
rate for a typical case, typical set of voltages that were
measured in the plant, may have changed the leak rate by a
factor of four or so between a more conservative one, maybe
not the most conservative, and the log logistic.
I have the details, I just don't have them with
me, and that's basically what was done.
DR. HOPENFELD: Thank you.
DR. KRESS: I still would like to have you clarify
that middle curve for me.
DR. HOPENFELD: This curve?
DR. KRESS: How were the data obtained?
DR. HOPENFELD: You mean this curve?
DR. KRESS: Yes.
DR. HOPENFELD: I'll tell you how I think, what I
believe, you take a specimen, you subject it to the
pressure, to whatever the steam line pressure differential
DR. KRESS: The specimen has only the one bobbin
voltage indicator in it.
DR. HOPENFELD: No, I think those -- if I
understand correctly, you take some tubes, which were in the
plant, and you take and you pressurize them.
DR. KRESS: So that tube has a distribution of
readings to it.
DR. HOPENFELD: Yes. The certain -- let's see.
You see whether they leak or not. Go ahead.
DR. POWERS: And please use the microphone and
MR. MUSCARA: Joe Muscara, with NRC Research. It
essentially comes from what Westinghouse conducted to
develop the voltage-based criterion. Many of those points
are from tubes removed from service. There are some data
points developed in an autoclave in the laboratory.
What they've done -- there are two aspects of
this. One, is there is a probability that a cracked tube
will leak and, secondly, if it does leak, how much does it
The middle curve has to do with the probability of
the tube leaking. So they've taken a number of tubes from
the field, they have different voltage response, and tested
And what that graphs shows is whether a tube at a
given voltage responds, leak or doesn't leak.
DR. HOPENFELD: That's what I thought.
MR. MUSCARA: So you have a number of data points
at the bottom, those tubes that did not leak --
DR. KRESS: Let me ask you a question about that.
Does this curve say that a tube with all voltage responses
below that level?
MR. MUSCARA: There's data for the specific tube.
I'm assuming -- I assume they took the highest voltage for
DR. KRESS: That's the highest voltage on the
DR. SHACK: It's the voltage you measure according
to your specification for how you measure the voltage for a
tube. You have a procedure for doing that.
DR. HOPENFELD: One tube, one defect, one
MR. MUSCARA: But there are many voltages along
that crack. So you have to select a voltage from that
DR. KRESS: But that's the point I was trying to
get at. So one tube, one voltage, one crack.
MR. MUSCARA: Right, and I suspect that that's the
highest voltage noticed for that particular crack.
I think that's -- from what I've read, that's what
they do. They take the highest voltage for a given crack.
DR. KRESS: But the point I wanted, wasn't clear
to me, is if one tube, one crack that you're looking at.
MR. MUSCARA: Or a cracked zone.
DR. KRESS: Or a cracked zone. That clarifies it.
MR. MUSCARA: Some tubes at a given voltage leak,
other tubes at the same voltage don't leak.
DR. KRESS: Yes. Okay. Except that almost looks
like a --
DR. CATTON: And these voltages measured in the
steam generator and then the tube is tested later or are
these voltages measured on the pulled tube or what?
MR. MUSCARA: Yes. That's the voltage that was
measured in situ during operation, the in-service
DR. CATTON: So that's what that voltage is.
MR. MUSCARA: That's what that number is. Of
course, they do measure the voltage after the tube is
pulled, but the number that they're providing here is the
voltage response of the tube in-service.
DR. HOPENFELD: Well, let me make a couple of
comments on that. Thank you very much for straightening me
out on this. This is not -- my point here is really there
is a disagreement whether this distribution could or
couldn't be, but the point that I just want to make now,
just in case I forget, you ask yourself what causes
something to leak.
Not the voltage, what causes something to leak is
how deep is the crack and that voltage has nothing to do
with how deep that crack is and what loads are going to be
on the crack.
That goes back to what you said. When they say it's
empirical, it's empirical, but it really doesn't relate to
real world conditions. It's empirical of something, but
that's really was the point.
But exactly how -- there's another point that
wasn't mentioned here. When you -- if you measure that
voltage in the plant and you pull those tubes, many times,
those tubes get damaged, and I don't know whether you tear
ligaments or you fix ligaments or whatever, and when you put
this thing in the test conditions, I don't know what these
points -- what they really represent.
I did go back to the database, I couldn't figure
it out. But it's not really very essential to my points
MR. BALLINGER: As a point of clarification,
you're going to, I'm sure, explain to us your issues with
respect to making the jump to the next step.
DR. HOPENFELD: Right.
MR. BALLINGER: That is to say, voltage to leak
DR. HOPENFELD: I'm going to spend a lot of time,
probably most of the day today on this.
MR. BALLINGER: Sure. But with respect to the
choice of the curve fitting technique that you use to fit
DR. HOPENFELD: First of all, this is not me.
This is NRC. These are these people.
MR. BALLINGER: I'm using the generic you.
DR. HOPENFELD: Okay. Sorry.
MR. BALLINGER: That's based on the so-called
engineering judgment part and there are statistical
techniques which identify the goodness of fit.
DR. HOPENFELD: Right.
MR. BALLINGER: And at some point, it's your
choice, the generic you.
DR. HOPENFELD: Right.
MR. BALLINGER: Of which to use.
DR. HOPENFELD: Right.
MR. BALLINGER: So there may indeed be an
empirical correlation between the parameter that you measure
and the depth of the crack.
DR. HOPENFELD: It could very well be.
MR. BALLINGER: It then becomes your choice.
DR. HOPENFELD: Right.
MR. BALLINGER: On how you fit that data and what
relationship that you use, and that relationship may have
absolutely no connection with -- it's just a strictly
DR. HOPENFELD: I did talk a lot to our
statisticians and I forgot all my statistics, but I remember
the basic concept, and I understand what you're saying, sir.
The point really is that maybe all that thing is
okay within this laboratory that they're testing all these
things, but, now, what that really means later on, a month
and a half, a year and a half later in the plant are two
MR. BALLINGER: But that's a different question.
I mean, in --
DR. HOPENFELD: But there is a different question,
but, you see, that's really what I'm after.
MR. BALLINGER: But there is nothing inherently
bad about making a choice of what you use to fit the data.
DR. HOPENFELD: No.
MR. BALLINGER: It could be empirical. The
relationship is empirical and can't be derived.
DR. HOPENFELD: I'm not questioning that.
DR. KRESS: But I'm presuming that's not the full
database that goes into establishing that particular curve,
because I would have never chosen that one for that
MR. BALLINGER: Nor would I, based on the cartoon.
DR. KRESS: It's just a cartoon, I'm assuming.
DR. HOPENFELD: I'm not questioning this. Go
ahead. I'm getting a break here.
MR. BUSLICK: The point is if there is no
theoretical basis for one curve and for probably a family of
curves for probability of leakage or another, then if you
have several families which have equal goodness of fit, the
real question is why choose the log logistic if it tends to
give one of the lower leakages.
MR. BALLINGER: That's exactly correct. Then you
have to have some other piece of information which may or
may not -- which may be relevant.
DR. HOPENFELD: Really, my main point here was
that -- and I brought it in here really to show you even the
experts, there is disagreement. And it could make a
difference, up to a factor of four, you know, it makes -- it
probably has no meaning as far as the overall risk is
concerned, but when you talk about this legalistic aspect of
Part 100, it may.
DR. CATTON: When you look at this last figure,
what is the range? You don't have any numbers on here from
the bottom of the data to the top of the data. Is that a
factor of four?
DR. HOPENFELD: No. I think what I'm talking
about, if you pick up this distribution --
DR. CATTON: Well, I understand --
DR. HOPENFELD: -- or pick up a different
distribution and multiply by this, you can come up with a
definition factor, a definition leakage.
DR. CATTON: But just on that last figure, where
you have leakage rate is a function of voltage.
DR. HOPENFELD: Right.
DR. CATTON: If you just blindly plot all the data
that you can find, what is the scatter?
DR. HOPENFELD: I'll tell you, I'm glad you're
bringing it. I think you have to go back to this
proprietary information, and that's -- you have all that
data in there.
DR. CATTON: Is it decades?
DR. HOPENFELD: It's several orders magnitude, but
I don't know exactly. To answer your question, okay, can I
come back to this?
DR. CATTON: Sure.
DR. HOPENFELD: I would like to come back to that,
because it will become clear.
DR. CATTON: I just want to raise that issue
because this is a problem of a heterogeneous media and
unless you relate to the proper parameters, you never get it
DR. HOPENFELD: Correct. That's my next slide.
MR. HIGGINS: Could you indicate on there, if it's
possible, the one volt, two volts, three volts that you
talked about before?
DR. HOPENFELD: In here?
MR. HIGGINS: Right.
DR. HOPENFELD: I'll tell you, all the data is in
that proprietary stuff and I -- this came in from many years
ago and since then, they have generated a lot more data over
the years, and I really don't want to quote numbers without
really going back. But all that information is in your
hands and it's all stamped proprietary.
MR. BALLINGER: One last thing, and then I'll drop
the statistical thing.
DR. HOPENFELD: Sure.
MR. BALLINGER: That is, as long as you're doing
interpolation, the goodness of fit works okay. But the
choice of distribution that you use, the choice of
relationship you use makes a big difference when you start
DR. HOPENFELD: Okay.
MR. BALLINGER: That's where it makes a
DR. HOPENFELD: But what makes a difference to us,
really, from my perspective, is whether the number they came
up with to calculate their dose releases, what kind of
uncertainty do you have; do you have a factor of four or are
you conservative, what are you, and that's really why I'm
bringing it out.
We'll have an opportunity to talk about this a
little bit more.
Let me go and, again, outline the differences here
between the NRC approach and what my concerns are.
Basically, the whole philosophy is that we have
this laboratory data which was obtained in simulated
environments under certain conditions and then we have a
specimen, I think these were a tubular specimen.
Anyway, you have some tubes, degraded tubes that
were pulled out from the plant and they were tested at
different pressure rates. When they got up to 2,500, they
observed what the leakage was. Basically, that's all it is.
What my claim is, that the database or all the
database that the industry has generated is irrelevant to
the steam line break accident, because, for one, there is no
physical relation between the voltage and the leakage. And
therefore, laboratory data cannot be used in a different
There is no reason, physical reason or scientific
reason why there should be any relation between the voltage
and the leakage. These are completely two different
phenomena. Let me say why.
That voltage probe that produces the voltage
reading, I don't remember the rule there, but you run a
current through the coil and it produces magnetic field and
you have secondary currents in the material and then you get
a feedback and you read different voltage, that's what you
read, I think.
But that voltage that that probe reads depends on
the volume of the cracks. Really, that's what it depends
on. It depends on the crack orientation. If you have
various different cracks, oriented and the spacing between
the cracks, they're going to affect the voltage that you
The geometry of the probe or the field of view of
the probe and the environment, you have impurities in there
and you have a support plate and you have deposits in there
and their physical characteristics is going to affect what
that probe does, and then that probe, you can get away from
some of that by running the probe at different frequencies,
but you see this is not a straightforward kind of thing.
It's not something that I take a voltmeter and
measure voltage of a clean system. It's not. But now, when
I need leakage, the physical parameters that drive leakage
are different. Okay.
What drives leakage is the loads. If you're a
tube, sitting there, and you have some cracks partially
through the wall, what is going to decide whether that crack
is going to open up is the loads on that act on that crack.
All they simulate in these tests are internal
loads or pressure and nobody has shown me that these
internal loads are really the main loads on that specimen.
But that's all you have.
Implied in this, to do the steam line break, all
you have acting on those tubes are the internal loads.
There are no other phenomena. There's no erosion from these
jets. There is no vibration. There is no bending. All we
have is a nice clean environment where we're testing these.
So that's what the database is.
Plus, and that's another plus, these tests are
conducted in an environment which is entirely different than
the plant environment. They're not testing those
necessarily under the same pressure, same delta P, same
temperature. So what do we have?
We have some conditions that we're simulating,
some, and now we're going to argue whether it's a lot and
semi or part-semi, and we're taking those conditions and
have theoretical models, untested theoretical models. We
apply all that and we come up with a regulatory position.
We say this is safe, and that's what the difference is.
The procedure, practically going back, makes
really no distinction. There is no allotment in here
anywhere in the entire process of this voltage-based
approach, there is nothing in here that really makes a
distinction whether you have a degraded tube or you have a
perfectly good tube.
All you have is some model that tells you, okay,
this model tells you that you're okay, so everything -- it
depends here on the validity of that model.
Again, I'm repeating this, it's the one gpm here
that you really have no control over. You cannot assure
somebody that under steam line break, you can have hundreds
and hundreds of times more flow, more leakage, if the
mechanism is there, than the one gpm.
So if the industry had come with some kind of a
mechanism, some kind of a shut-off valve, that as soon as
you exceed that one gpm, it shuts off the system, then,
yeah, we can operate with any cracks you want.
But furthermore, even if you don't -- you can
operate in any leakage if you have a genius operator that
will control anything you have. So if you have this perfect
operator somewhere that can control no matter what the
reactor does, then it's fine.
MR. HIGGINS: Joe, does the previous curves that
you showed us, with the leak rate derivation, does that
ensure -- the calculations using that ensure that you stay
below the one gpm?
DR. HOPENFELD: Which one, the -- no, it does not.
Absolutely not. No, it doesn't, because that curve, by
itself, is just -- again, it's a theoretical thing obtained
for certain data within a certain environment.
Now, if it stays within that environment, that
curve, it would be okay, but we're not we're not interested
in that environment.
MR. HIGGINS: I mean if you do your analyses with
that assumption that that is the leakage rate, will that
keep you under the one gpm? Because you didn't put any
values on it.
DR. HOPENFELD: If you do -- well, let me -- give
me one second and I'll address that, because I'm going to
break the thing item by item. So come back to me and hit me
with this, because I will come back to this.
DR. BONACA: Before you leave it, because you --
you know, there was a correlation of voltage measurements
and leakage. And the point I'm making is that for a steam
line break, what you should measure is the residual tube
strength to withstand the steam line break.
DR. HOPENFELD: That's correct.
DR. BONACA: That would be --
DR. HOPENFELD: Under steam line break conditions,
under those loads, not loads in the laboratory.
DR. BONACA: That's right. So if you could
measure, by some metrics, the residual tube strength to a
standard steam line break or, let's say --
DR. HOPENFELD: You're correct.
DR. BONACA: -- the damage that would not allow a
tube to withstand a steam line break, that would be a
DR. HOPENFELD: Right.
DR. BONACA: But you're saying that going from
voltage to leakage, you cannot infer an intermediate step --
DR. HOPENFELD: Right.
DR. BONACA: -- that says --
DR. HOPENFELD: Let me put it the other way
around. If you were to take a tube and, say, hundreds of
tubes, they all had some cracks in them. And you put them
in the laboratory and you run tests under bending, you run
tests under torsion, you run tests under vibrations, and you
run tests on all these conditions that you can think that
represent real life, and then you see on all these, I didn't
have any of these things, these are super-duper tubes, that
material is unbelievable, it never breaks.
And then you don't have any leakage and I say,
yeah, that's fine, but that is not what's being done. All
they do is take these samples and they internally pressurize
it and then pressurize it under different pressures,
different temperatures that you have in the plant, and then
they generate this data that I was showing you before, and
that's what's being applied.
And what I'm saying, in all the statistics and all
the methodology is fine, as long as you stay within that
laboratory. You go back to your laboratory, all the
statistical things and all the correlations, that's fine.
But it's an entirely different situation when you're talking
about an environment that really has nothing to do with
this, and that's what I will give you the physics of it, why
it has nothing to do with it.
Okay. To summarize this in a pictorial way, I
realize this is an important thing, so I put a lot of stuff
in here, so we can focus a little bit better on all these
You start, you go to the laboratory and you run a
whole bunch of specimens. Some of them came from U-bends,
some of them may have come from tubes, some may come from
the plant, and from that laboratory, you generate a leakage
versus voltage data.
Now, if you go back to the proprietary
information, it is difficult to understand what's really
behind how the data was generated, because some of those
specimens, especially those that came from the plant, they
See, there's a lot of crud in the system,
especially when you go to shutdown, so some of these things
are plugged. The cracks plug and obviously you don't pull
it at full temperature and full power, so you don't really
know which one is plugged and which one is not.
Many years ago, there were some tests at PNL about
plugging these cracks and the idea there was they were going
to come up with some rationale for leak before break, and
what they found, it was very interesting, you look in one of
those PNL reports, indicating very clearly that this
plugging and non-plugging is a very random, unpredictable
So you don't really know how to interpret that and
whether, looking at the database, and I spent a lot of time
looking to figure out which specimen they're talking about,
and they say this was in there, this wasn't in there, and
you don't know what was included and what wasn't included.
So there's room here to make all kinds of
adjustments about plugging of these specimens.
Then, as I already said before, and Dr. Busnick
discussed it, there is a statistical distribution adjustment
that's to -- which is, again, within this boundary here.
And then you have, which I think is probably more difficult
to interpret, is when you pull those tubes, a lot of them
get damaged. You damage the ligaments.
And I don't know, please go back to the database
and see if you can figure out which was included and which
wasn't included, and, more importantly, if NRC has to have
an audit function, how do you go and audit that stuff?
If a utility comes in here and says, well, here is
our database, we -- how do you know? I mean, you can't
characterize the condition of the tube. You don't even know
whether it's representing the time that you're talking
about, how long it was there, you don't know. You don't
know anything about those cracks, except you know there are
a lot of cracks, but you don't -- can't characterize them.
So what you do, you get a statistical relation or
some kind of regression curve for the leakage versus
voltage, and that's okay. If you are running these reactors
in these environment, that's fine, as long as you do that.
Now, when you go -- and this was the first thing I
learned at school, that if you go to a different environment
and you know that there is no mechanistic explanation to the
phenomenon that you have, you can show, as I did before,
that the parameters that control leakage are different than
the parameters that control -- the parameters that control
leakage is the length to diameter ratio of the crack.
It's the opening area and the pressure drop. And
that's not what controls what the voltage is read by the
probe. These are two different things.
Now, I'm not saying that you couldn't possibly
have some kind of a correlation. You can always get a
statistical correlation, anything you want, and that's fine,
there's nothing wrong with that.
But you have to stick to that environment and
don't go beyond that point.
DR. BONACA: I have a question.
DR. HOPENFELD: Yes.
DR. BONACA: The question I'm asking is, are they
isolated and pressurized internally?
DR. HOPENFELD: From what I understand, and I
don't know the exact setup, the experimental setup, I did at
some time, they take the specimen and they apply pressure at
a certain rate. And recently, somebody brought up the issue
of that even the rate makes a difference on the rupture, but
then it was brought out that that difference only occurs
when the crack is very, very deep, and it's a secondary
In other words, that's not important. What is
important here is that up to ten years of running these, the
industry constantly finds some new phenomena, and you would
expect it. When you run something, you don't know what
You're taking some specimen, you test them, and
then you say that's what we've got. So it's not something
that I would --
DR. BONACA: But I'm saying so, therefore,
although you do not simulate at this location, that the main
steam line break may bring about in the tubes, they do
simulate the delta pressure that the wall may see in a steam
DR. HOPENFELD: Some of them do, some of them
don't. Some of them have to operate on the different
DR. BONACA: Okay.
DR. HOPENFELD: So this is the next thing. So
what you have, you go with this data. You also have, as you
saw before, and that's the reason I brought this procedure
that they use, you have this measured voltage distribution
and then you go -- so all this gives you incorrect leakage.
So far, yes, this is all experimental. But now it's all on
the local beyond that point and, obviously, where is the
weight of this thing, what weighs more, the analytical here
or on the experimental, and I'll show you I believe it's
this part -- the analytical part over-weighs those little
laboratory tests they've been doing.
This is not little, I mean, this is many years of
hard work and I'm not trying to minimize it, it's great and
it's good to have this database, but don't worship it.
If industry wants to use it around those steam
generators, they should give you a little bit more
justification than they've been giving you so far.
So the next thing, and I like to call these knobs,
you have pressure and temperature adjustment, because the
pressure may not be the same and the temperature may not be
the same. Some of these were run at room temperature. Some
were run at different pressures.
So you have pressure and temperature adjustment.
Then you have crack growth adjustment. In other words,
remember, we had these voltage things, you can make a lot of
adjustment there in order because you don't know what the
growth is. So you go into the histogram of the plant and
you adjust it in the adjustment.
Then you have, as we already mentioned before, you
had this probability of leakage adjustment that -- no, I'm
sorry. Then the probability of leakage was already -- then
you have the POD adjustment. So what's the probability
that, since the whole concept is statistical, what is the
probability that you're going to miss some of those cracks?
Well, that is a little bit bothersome and I'm a
little out of my field on this, but if I remember correctly,
the POD, the concept of probability of detection of cracks
really came from single cracks or maybe even from fatigue,
when you were really worried about what the threshold that
you can withstand.
Now, what you have, you don't have a single crack,
really. You have a network of cracks. You have cracks
growing, coalescing, they're linking, they're doing all
kinds of things all together, and I'm not so sure --
eventually you do have one crack that starts, but that
doesn't mean that you don't have the next one.
I'm not absolutely sure that the statistical
concept that was originally intended and was developed over
the many, many years, and nuclear is not the only industry
this is being used. It's being used in the oil industry.
So I'm not so sure that this is strictly
applicable. But having said that, we do have some data in
thereabout you'll really have to look at it, and I'll come
back with some numbers later on.
Now, the main one here is then you have damage
adjustment. As I already mentioned before, you have tubes
that are going to be exposed to different loads. They're
going to be exposed to erosion.
What adjustment is there? It's being ignored.
It's not there. It's just completely forgotten. It's not
coming in. So the adjustment, the knob goes from zero.
And then the adjustment in the chemistry of the
iodine spiking. What you do there, you make an adjustment
to come up with what you want. Well, they finally -- you
meet two criteria here, and I don't know whether they're
independent, how they're being used.
One says I'm going to meet Part 100 and another
says my risk is going to be such that it will stay within
the ten-to-the-minus-five core melts per year.
So you see what the methodology is, is that you
take a very small database and you apply all these little
knobs that you have and you come up with any answer you
want. That's really what it boils down.
MR. HIGGINS: Joe?
DR. HOPENFELD: Yes, sir.
MR. HIGGINS: The adjustments that -- theoretical
adjustments that you're describing there, are you talking
about industry calculations or NRC or both?
DR. HOPENFELD: Okay. I'm going to go through
this. This is just the purpose of this slide, is exactly to
address that issue.
The industry -- and this is proprietary
information -- has developed computer codes to make pressure
and temperature adjustments which are to take that data from
the laboratory and adjust this thing to plant operation, in
other words, that data is being corrected.
Some of those are analytical equations, kind of
straightforward. But if you go back, and I'm going to
discuss that, the validity of them is very questionable.
And the analytical equation, I can't go to that,
because it's in your proprietary information, but there is a
computer code, and I don't know if I'm even allowed to say
what the name is, but talks about flow through cracks and
what it does, there are some basic flaws in that computer
Now, how that computer code is used, it's used to
-- it's used to show the analytical equations that they have
derived agreeable with a more sophisticated tool.
But then you ask yourself the question, that tool,
let's call it computer X, that tool, in order for it to show
that your analytical tools are correct, and in order to
calculate leakage on a computer -- on a real flow model, you
have to know the length of the crack.
You have to know the L-over-D of the crack,
because that determines the nature of the flow. Now, how do
you get that information from the voltage data? And I think
you ought to go back, I was going to spend some time, but
we'll have to close the doors, if you go back to your
proprietary information, there is a description as to how
they're doing all that stuff.
DR. CATTON: This proprietary information you're
talking about, I don't know about the rest of you, but what
I have is just figure titles. So I really don't know
anything about this proprietary data.
DR. HOPENFELD: Well, you may-- I don't know if
they have given you all those -- the database that discusses
how the data is being --
DR. CATTON: They gave me a lot of stuff, but
wherever it referred to proprietary, all I had is a figure
DR. HOPENFELD: I don't know how they're handling
this, but the description of the computer code and the
thermal hydraulics through those cracks is in that code, and
I just don't want to go into that, even though I think it
shouldn't be classified as proprietary, but I thought it was
proprietary and I'm not ready to go into the details of
But I'll tell you one thing, though.
DR. CATTON: Is that the code called Crack Flow or
DR. HOPENFELD: Yes. Well, you said it. I don't
even want to mention it.
DR. CATTON: You mean even the name is
DR. HOPENFELD: No, it's not, but I am -- I am
under tremendous pressure in this area, so I'm trying not to
-- I don't know what -- I'm not a lawyer and I don't know
what proprietary or what's not proprietary.
I wanted to have this meeting open to the public
so I can go through to a more -- be free to explain the
general things that I am going into, the minute two-phase
flow, but although I did mention it.
It's flawed, the model is flawed, and later on,
you will see there is some recently data or another model
developed at Argonne and it's completely different and I
will talk about this a little bit later.
So the point here is that when they tell you that
we have all these computer codes, we have all these tools to
extrapolate, it's not really -- there are a lot of flaws in
One thin that I found was missing there, and I
couldn't find any description of these, is that when you
deal with flow at high pressure, high velocities, it takes
time -- take this again.
See, it takes time for the fluids to flush into
steam. Again, those -- we have liquid under water at 2,500
pounds and you have, on the secondary side, the atmosphere.
So now you have that 2,500 pounds, up to 2,500 pounds,
water, getting out of the tube and flushes into steam.
Now, these are very thin tubes. It's 40 mils. To
give you a feel for mils, a mil is your hair. One hair is
one mil. So 40 mils. These are not big, huge, one-inch
tubes. They're very thin tubes.
So you have residence time. It takes time to
become -- to turn into steam. So the order of magnitude for
this is ten-to-the-minus-four, and for these kind of
experiments, you see the liquid comes out of the pressurized
nozzle and then the two-phased region really develops
further, depending on what the L-over-D on that tube is.
But it's on the order of ten-to-the-minus-four.
So now if we have all these complicated situations you
mentioned, the computer code Crack Flow has a two-phase flow
model in it. They flow -- they have characteristics for the
distinction between flows and an equilibrium flaw, but
there's no distinction between what the stability of the
thing, whether you're going to -- you will have -- depending
on the thickness or the tightness of the crack, the flow in
there is going to be different.
Now, the answer is different, too.
DR. CATTON: What gives you the highest flow rate?
DR. HOPENFELD: I think it gives you one-phase
when you have liquid.
DR. CATTON: Frozen form.
DR. HOPENFELD: Yes, but frozen would be
DR. CATTON: It would go all the way through.
DR. HOPENFELD: But they have it two-phased in
both cases. That's not the way they defined it. They
defined the frozen flow, it's either two-phase. In either
case, it's two-phase, it's not liquid.
DR. CATTON: It's liquid inside the tube, isn't
DR. HOPENFELD: It's liquid inside the tube, yes,
but when they say frozen, they don't mean that you have --
DR. CATTON: Frozen, you maintain --
DR. HOPENFELD: I think frozen --
DR. CATTON: -- the state through the crack.
DR. HOPENFELD: No. I think what they mean in
there, I've tried to figure that out, I think what they
mean, frozen, is you have -- it's just like in a chemical
equation. You can either have equilibrium, if you're
freezing by that composition, but the basic assumption is
the Henry model, using frozen in terms of a two-phase or
it's an equilibrium between the phases.
That's what they're discussing. What I'm saying,
there is no where in there that this criteria of stability
of the thing appears in the code, and you can go look for
yourself. I was looking for it and it's not there.
DR. CATTON: What role does this play?
DR. HOPENFELD: It plays the role --
DR. CATTON: You showed a previous diagram that
says we go from the laboratory test and then some things
that should be done, but a lot are not, to a conclusion.
DR. HOPENFELD: Right.
DR. CATTON: No where in there did I see anything
about the modeling.
DR. HOPENFELD: Right. What roles this play is in
there, when they -- remember back, I told you that there is
a pressure and temperature adjustment. They have a
theoretical model and I'm questioning the validity of the
DR. CATTON: Okay.
DR. HOPENFELD: And what I'm saying, if you have a
theoretical model, that's fine. But first tell somebody
what you have and all they have is a two-phased flow in
Now, then Argonne comes the other day and if you
look in the very recent ones, they just completely don't
have any two-phase flow. They just use -- they don't even
have an LED over there. They're just using plain orifice
flow equation, and they called it a new -- I'll go back to
Anyway, the point is here that there is -- it
takes time to nucleate, although there are probably plenty
of nucleation sites there, it takes time to flush into steam
and that depends on what kind of crack you have.
So if you have a very, very tight crack, it could
very well be that you have a two-phased flow and the
equation, the Henry equation that he developed in 1971, are
But you can't say whether they are applicable or
not unless you can characterize what you have, what kind of
crack you have.
So now, my antennas, my warning signal, hey, why
am I telling you all that.
This is maybe a factor of four and they tell us in
the proprietary information that they have validated all
these theoretical equations and codes, okay. I'm showing
you right now, they didn't validate them according to
physics. There's something somewhere wrong. What they've
validated is questionable.
That goes back into the physics of flow through
cracks and it's a very complicated thing. Again, I'll come
back to it, but Dr. Shrock at Berkeley studied that for many
years and he came up with a correlation showing L-over-D is
an important factor.
Argonne, after two or three months, recently came
up with a model that basically is very similar to what they
did when they designed the aqueducts. They just neglected
all that, they just say that the leakage is simply --
DR. CATTON: Yes, but there's a difference. The
work that Trough did was for thick wall and this is thin
DR. HOPENFELD: It's not the thickness input.
It's the L-over-D. These are very, very tight cracks.
Okay. It's L-over-D that determines it.
DR. CATTON: Residence time is really what --
DR. HOPENFELD: Residence time, right, and the
L-over-D comes into the pressure drop thing. Remember,
these are very, very tight cracks. They may not even leak
under certain conditions.
So unless you can characterize what you have, you
really don't know what you have, and that's fine, too. But
don't go and advertise that we've got all these things,
we're applying all these corrections, and we've got an
answer and it goes back to all these adjustments that they
DR. CATTON: So there are two parts to what you're
telling me. One, you use the laboratory characterization of
DR. HOPENFELD: Right. Not characterization.
Just voltage. They don't characterize the thing.
DR. CATTON: They don't try to relate the --
DR. HOPENFELD: No.
DR. CATTON: They don't.
DR. HOPENFELD: There is no characterization
whatsoever. There is voltage. They may have did some
metallography, but I don't think they've done -- they've
correlated with the voltage.
DR. CATTON: So it's an inadequate
characterization of the relationship between voltage and
DR. HOPENFELD: I would say it's less than
DR. CATTON: Plus leakage measurements and then
some adjustment that's based on what could be a deficient
DR. HOPENFELD: Correct. Not applicable. It
probably is applicable under certain -- it may be -- I don't
want to say deficient. I'm sure that those two-phased flow
equations, they have been used in nozzle -- in industrial
equipment. It was developed by Henry and Fosky in '71. I
think they're applicable in certain areas.
But now that code that you named applied that
thing all over the place and our friends at NRC/NRR don't
question that, say, well, we've gotten this code that's been
And that's the point here. So they've got this,
they have to make -- all those tests were run at pressure,
at 2,500 pounds, and at temperature, which they didn't.
Then you wouldn't have to -- at least that aspect of it, you
don't have to worry about it.
But this is just the first one. Now, the next
thing is the crack growth -- to summarize what I said
before, this is strictly a theoretical model so far.
Now, what number that came in there is a
probability of detection and that came from NUREG-1477.
They have used point six. Again, as I mentioned before, the
POD concept may or may not apply here, but the database --
all the amplitude of the cracks depends on -- I mean, the
voltage depends on the separation, conductivity, the
permeability, the crack volume, the frequency, and the coil
These are the parameters that measure the
amplitude of the voltage that you measure, and I think,
Ivan, isn't that what you are talking about? That the
system would be much more complicated.
DR. CATTON: That's right.
DR. HOPENFELD: That's what you were talking
about, those two phases. I'm sure there is an analogy.
DR. CATTON: You probably don't have all the
DR. HOPENFELD: I'm sure I don't, but those are
the parameters that come in. The frequency is a very
important one and the permeability is very -- and the
conductivity, because if you have some -- like in the case
of Indian Point 2, there was copper got into the system and
they got wrong readings.
All these things sitting here.
DR. CATTON: The crack morphology, surface
morphology is probably --
DR. HOPENFELD: Right, correct.
DR. CATTON: -- a key.
DR. HOPENFELD: Right. This crack volume,
morphology goes into the separation between those cracks.
DR. CATTON: There is another piece of it. Often,
you can pick it up with the permeability, but there are
multiple ways to get the same permeability.
DR. HOPENFELD: Okay.
DR. CATTON: And the behavior may not be the same
as reflected by the voltage.
DR. HOPENFELD: There is probably another
parameter that has to do with -- if I remember my physics,
but it probably has to do with the thickness of the tube,
too. The skin thickness probably affects it, too. It can't
be an infinite thick material and get reading.
So there are a lot of parameters, I think, but the
bottom line of all this, again, I don't know, I'm not sure
about the POD concept, whether it's applicable at all, but
let's assume it is.
The NUREG-2336 indicated the laboratory tests
showing that .27 to .5 is a number that you get from other
tests and from laboratory tests.
And another thing is that that POD concept really
has, in the .6, hasn't been really verified against actual
You can do it in a laboratory, but if it was
verified, I'm not familiar with it.
So there is a question about that .6, whether
DR. POWERS: Joe, I've seen quite an inventory of
DR. HOPENFELD: I'm sorry?
DR. POWERS: I've seen quite an inventory of data
taken from the -- I believe it was a steam generator that
was from Surry.
DR. HOPENFELD: Yes, PNL.
DR. POWERS: And they quote POD plots as a
function of crack size and you see if a particular size, .6
kind of works.
DR. HOPENFELD: Yes, but there is also data to
show from that NUREG, there's .27 to .5 at the intersection.
So all I'm pointing out is that that number is
still hasn't been verified on an actual plant.
DR. POWERS: What I'm asking, I guess, is that
these were data that they collected from tubes in a steam
generator that has seen about six years wroth of service.
DR. HOPENFELD: Correct. Right.
DR. POWERS: Had a substantial amount of flaws and
whatnot on the tubes.
DR. HOPENFELD: Correct.
DR. POWERS: Is there any reason to discount that
as not actual plant --
DR. HOPENFELD: No, I'm not discounting. I'll
tell you what I am discounting, you see, that's one part of
the equation of getting the crack system, but it takes a lot
This is not something that you go in there. It
takes a group of people who are expert in this type of data
and this kind of inspection, NDE inspection, to have them
come up with what they detect.
So I don't know, I'm not worried about the Surry
equipment, it's still is not done at the plant to verify
versus an operating plant. That's what you want to check it
DR. POWERS: But I guess it seems like they take a
steam generator that's been pulled from a plant. They had
multiple teams, round-robin kinds of things.
DR. HOPENFELD: Right.
DR. POWERS: Used a variety of techniques, which
escape my mind right now, and they show these plots --
DR. HOPENFELD: I'm familiar with what you say.
DR. POWERS: It seems like a pretty decent applied
DR. HOPENFELD: Yes, it is. But what I am saying,
you still have to verify -- I mean, there is uncertainty in
it because you see there is an uncertainty in this NUREG
showing that the numbers are different. Tomorrow. So these
numbers are definition.
DR. CATTON: What is a tube-to-tube intersection?
DR. HOPENFELD: Okay. You have -- let me see if I
have it. Okay. The U tubes go to the support plates every
40 inches that hold the tubes together, to prevent the tubes
from -- in view of things like, what was it, 40 feet high or
whatever it is.
So you have support every 40 inches and that's
what the tube-to-tube support plate is.
DR. CATTON: Also, tube support. Tube-to-tube
intersection is this.
DR. HOPENFELD: It's the tube support, I'm sorry.
DR. CATTON: I understand.
DR. HOPENFELD: Let's see. Yes. It's about an
inch and where the tube goes into that support, it's just a
crate basically, that's what I'm calling a tube support
Again, this basically talks about as mini-robin
coil test and it talks about the .25 to .5 and gives you a
reference for this, and provides you, if anybody wants to
look more into that, fine. It gives you more thing that you
That's as much -- then there's a --
MR. SIEBER: Just to clarify this for myself.
DR. HOPENFELD: Yes.
MR. SIEBER: The probability of detection of .6 is
really for characterized flaws that are equivalent to about
40 percent through wall and the larger the flaw, the greater
the probability of detection, as I recall it.
So that if you're 80 percent through wall, .8 and
so forth. So even if you can't detect anything below 40
percent, does it make a big difference?
DR. HOPENFELD: No, it probably doesn't.
MR. SIEBER: Okay.
DR. HOPENFELD: All I'm saying, that the data
that's available and you may want to go back and look into
the validity of it, that the .6 that was picked up from
NUREG-1477, it may or may not be representative to other
Would you like to say something? Mr. Spence has
looked into that and maybe he can make comments on that.
MR. SPENCE: The intersection between the tube and
the tube support plate has metal around it and that's a
solid plate. It is not an egg crate. And the GL-9505 one.
And it also has magnetite and all kinds of metal oxides in
there and that's giving the coils trouble seeing the flaw.
And that's why -- I think that's why you're
getting -- I did the numbers, as a matter of fact, to come
up with the .2 and the .5, and that's only for the crack
I think the rest of the testing, the .6 is
basically free span. But that was, again, a very small
sample of the round-robin data.
DR. HOPENFELD: That's a very good point and my
point is, and I haven't gone as much in detail as I probably
should to all the data at PNL, but there's volumes for that
All I want to point out here that there is
discrepancy and what the reason for those discrepancies, I
don't know, but I strongly feel that there's a lot of human
factors involved, and what you're doing there in the plant
is not exactly what you're doing at PNL. It may be close,
but it's not exactly the same.
MR. SIEBER: Just one final question. The whole
voltage scheme is only for cracks that are at the tube
DR. HOPENFELD: Correct.
MR. SIEBER: So the free span value of probability
of detection of .6 really doesn't apply.
DR. HOPENFELD: That is what was put into 1477.
That's what they are required to do.
MR. SIEBER: Okay.
MR. SPENCE: Could I make one other point? And
that is, the original setup for the coil eddy current
testing was to find dish shapes, wall thinning, corrosion
type things, and for that, it does a little bit better job
than finding the crack itself.
And there is no correlation between crack size and
voltage that I've been able to determine, even between crack
size and dishing.
DR. HOPENFELD: Thank you, Bob.
MR. SPENCE: Yes, sir.
DR. HOPENFELD: And this one is not very clear to
me, but this relates to severe accidents. Originally, the
preparation for NUREG-1477, the NRC didn't say anything
about cracks and erosion of cracks from jets, although that
information was already in that DPO, going back to '92.
They completely ignored that.
Later on, they found that there is a potential for
cracks to cause jet erosion of adjacent tubes and they only
focused their attention on severe accidents. Now, the
condition for a severe accident and design basis are not
that different with respect to that aspect of corrosion, but
I don't know, they, for some reason, they said that it
doesn't exist in the design basis. The erosion only exists
in the severe accident, and maybe tomorrow you will get the
explanation of why.
I don't quite understand, but I did talk about it
in that -- in discussing the various -- what do you call --
differing -- DPO consideration document, because they were
talking about these.
And for some reason, they said that cracks which
are larger than .125, they're going to be written off like
the tube is gone.
But based on data, and I don't know exactly which
data, they haven't shown that, based on data, most of the
cracks are not going to be below .125.
In other words, there will be no through the wall
cracks which are smaller than .125 and presumably, I don't
know how they can show that -- how they can prevent, because
the voltage doesn't tell you what the crack size is, how
they can prevent the larger cracks are not going to be there
and maybe you want to ask them to explain this.
But nevertheless, they said that less than .125,
cracks are not going to exist.
Now, if you look into the basic theories of how a
crack grows and you look into these networks of cracks and
the intensity factor, there is nothing in there that tells
you that you are going to limit how -- what kind of size of
crack is going to get to through the wall. You can have
cracks growing on the order of several grain sizes.
So I don't quite understand that, but what the
practical application was, the summaries in the case of
Farley, there was an indication that you could have small
cracks, but they say, well, we believe it's not there and
there's no problem.
Now, I really don't understand the whole
methodology. I'm just repeating what they say. You may
want to ask them, but evidently .125. As far as I'm
concerned, any crack, if the velocity of the jet is
sufficiently high, will damage the next tube.
The next adjustment, and that goes back to crack
propagation at growth, which, in a classical fracture
mechanics, it's controlled by a K factor and even there, you
can see this is not a simple thing that you can go and you
see various researchers have a spectrum of orders of
magnitude, depending on the pH of these cracks, how they
behave and how they grow, and most of them, I don't know all
of them, but probably a lot of them are single cracks.
Usually you study, in laboratories, single cracks.
Whether there have been studies on network of cracks. When
you have a network of cracks, that crack -- the K factor,
the intensity factor is much more complex because these
cracks get together and they grow or they stop growing.
The propagation of the crack changes and what you
have is entirely different. It's a dynamic thing. It
depends on the pressure, it depends on the load that -- on
the stresses that operate on the tube.
MR. BALLINGER: Excuse me.
DR. HOPENFELD: Yes.
MR. BALLINGER: As a point of clarification.
DR. HOPENFELD: Yes.
MR. BALLINGER: That data was derived from a lot
of different tests, very few of which were actually cracked
tubes, very, very few.
DR. HOPENFELD: Okay.
MR. BALLINGER: In the actual geometry. In fact,
DR. HOPENFELD: Good.
MR. BALLINGER: When you do look at actually
cracked tubes, real cracked tubes, you get a little bit of a
-- you get much better -- you don't get the scatter that you
got there. That scatter is due not to the inherent -- well,
in large extent, not to the inherent problem of an
individual stress corrosion crack growth. It's the test
DR. HOPENFELD: That's exactly what my point is.
MR. BALLINGER: In the attempt to simulate the
DR. HOPENFELD: That was exactly my point.
MR. BALLINGER: And most of those were -- okay.
We can --
DR. HOPENFELD: Really, I brought it in for that
very reason. It's not a study in fracture mechanics.
MR. BALLINGER: What I'm getting at, though, is
that when you actually use actual tubes, real tubes with
real cracks and real geometries, the scatter is a lot
DR. HOPENFELD: But my point here was really that
what I was trying to say is that there is an uncertainty
when you go from one test to another because the environment
is different and that probably is responsible for all of
this variation, orders of magnitude in the crack growth
So even though the pH is the same, there are large
And the next thought that I was going to inject
at, that you have those variations within those ten
thousands of tubes sitting there in support plates with
having pH all over the map and you have all kinds of
stresses. Some of them are being stressed because of the
U-tube moving. Some of them you have the stresses because
you have flow induced vibration.
So you have an entire spectrum of environments.
And all I was saying, if you look at the literature, yes,
there's a range you see for the same condition, for the same
environment, you have a large scatter, who knows, and maybe
-- I haven't looked at each one of them and --
MR. BALLINGER: But let's be careful, again, let's
be careful that that data doesn't represent -- if you were
to take a real tube with a real stress corrosion crack in
prototypic environments and in prototypic pressures and
stuff and run several tests again and again and again, you
would get much, much, much, much, much less scatter than you
That scatter is not due to variations in -- like
DR. HOPENFELD: But still the point is that if you
were taking many different tubes, actual tubes with cracks
in them, and you were running them in all these different
environments, you're going to get different answers.
MR. BALLINGER: I think we're trying to compare
apples and oranges here, and we need to be careful.
DR. HOPENFELD: Okay. Thank you. I appreciate
that. Maybe that's not a right point. The point is that
the scatter of this that you have a very dynamic environment
and you cannot say that because my voltage growth rate was
over a certain period of time, X, that it's going to remain
X for the next 18 months. That really was the point,
because it is not the same environment, and that's the crux
You don't know where the cracks are. All you're
saying, my voltage in the last period was -- had that
distribution and the same distribution is going to be
occurring for the next period.
In other words, just looking at the extreme,
suppose you just started at the beginning of that cycle, you
finish your incubation period and you start into your
So how is that going to come into this? In other
words, the time, the previous time, the year and a half
history is not applicable to extrapolate the kinetics of
cracks, especially in a dynamic situation where these cracks
grow, coalesce and stop and grow and so forth.
Here is some plant data, and I think this came
from Farley, and this shows that the prediction usually are
that the -- remember, you are not supposed to exceed, at the
beginning, you're not supposed to -- originally it was one,
then it went to two volts, but you see you'll find fairly
high volts, you'll find three, you'll find all the way as
So what you predicted before as real life
experience doesn't verify that at all.
DR. BONACA: Explain to me who are A, B and C
DR. HOPENFELD: These are different steam
generators at a given plant. It was Farley and I don't
remember whether it was cycle 14, it was about three years
ago, and this is not restricted to Farley. There are about
-- we'll go back, I think Breakwood and Byron had the same
kind of phenomenon.
So these growth rate -- and Arkansas. All these
growth rates that you see are not really consistent with
this concept of -- that you can take prior voltages and
project them to the next -- let me ask you, sir.
Do you know any industry -- I've looked into the
oil industry and I haven't seen anywhere there where there
is justification of using the data on the cracks and say,
well, since it didn't change in the last ten years,
whatever, we can project for the next ten years.
This is the only industry in the world, I think,
that does that. And I know the Japanese don't do that. I
think they don't allow any cracks. As soon as they see a
crack, any surface crack, it's being plugged. The tube is
So it's a concept, but it has no physical
rationale to it.
DR. POWERS: I guess I don't follow exactly why
this slide speaks to the projection issue. I mean, it looks
like it's a set of data for some particular steam
DR. HOPENFELD: Yes. It was at Farley, right, but
what I'm trying to show you, that you can get very, very
high crack growth, you can get 13.7 volts, which would leak
something on the order of six gpm.
Remember that one gpm limit that they had. What
do you think --
DR. POWERS: I mean, it seems to me that 13.7
volts is substantially beyond even three.
DR. HOPENFELD: Yes. But that's my point. That's
what you find. And if you go to the mechanism, you don't
really need many cracks to cause damage during the steam
DR. POWERS: But if you had an indication of 13.7
volts, wouldn't they plug that tube?
DR. HOPENFELD: Well, they didn't plug it before.
This is what they found during the outage. They must have
plugged it, yes. they pulled it.
DR. POWERS: So that particular tube is not going
to leak anything.
DR. HOPENFELD: No, that's not my point. My point
DR. CATTON: I think the point he's making is that
in the previous time they did it, they didn't have the 13.7.
They were under the three or whatever.
DR. HOPENFELD: Yes.
DR. CATTON: So in one cycle, they went from three
DR. HOPENFELD: Well, one or zero. My point,
either the POD is not worth anything or they get huge growth
rates. I'm not -- now, I don't know whether that represents
100 percent sample of all the tubes in there, but my point
here is that this idea of using prior history to tell you
what kind of voltages you're going to have in the future,
this is flawed.
DR. CATTON: What you're saying is if I had looked
at this same slide taken at the end of cycle 13 --
DR. HOPENFELD: Right.
DR. CATTON: -- I would have found no tubes that
were not plugged that had indications greater than three
DR. HOPENFELD: Right.
DR. CATTON: I'm not sure what they used as the
DR. HOPENFELD: Well, I don't know what they used,
but I'm not even questioning whether they have 2.3 or 2.5.
I'm questioning this concept, can you use this concept where
nobody -- there is no physical reason for it. There is no
theory that can justify that.
MR. BALLINGER: Do we have the data from the end
of cycle 13?
DR. HOPENFELD: I'm sure we do and I don't know if
I brought it with me, but I'm sure they have it.
MR. SIEBER: If I go back to this --
DR. HOPENFELD: Yes.
MR. SIEBER: -- overhead of yours, all that the
Farley data shows me is this is a distribution that tells me
that it's probabilistic in nature and that's the way the
methodology for coming up with a bottom voltage versus
number of indications and then later on the postulated main
steam line leakage.
DR. HOPENFELD: Correct.
MR. SIEBER: It's just a combination of a lot of
probabilities, which define an expectation and the
uncertainty associated with it.
DR. HOPENFELD: Right.
MR. SIEBER: So this is what I would expect to
find for that and that's --
DR. HOPENFELD: Well, I don't know if you would
expect that high, because this is all way, way really here,
you get really very high. I mean, you really -- look, what
you do, you take these numbers, I don't know how many more
of those, if you had -- you take these numbers and multiply
that thing by the number of cracks and you ask yourself,
okay, what is my leakage.
Well, for one thing, I've got this constraint of
Part 100. Well, already you're exceeding with one of them.
I don't know. You have to add all those.
So you see, you're exceeding that. You're
violating the law, for one thing, but never mind the law,
then the next question is, okay, I've got this baby here,
now I hit it with that -- say it was left in service. Now
I've got the steam line break. Now what?
Okay. I was at the tail of this distribution, but
now what am I going to do? I'm going to have one rupture,
ten ruptures? That's really what it is.
DR. CATTON: Joe, just following what John said,
is there some argument somewhere about the number of tubes
that they have to examine at each cycle? Then you can base
it on the statistics and say what I'm allowing is that one
or two tubes escape or three or four tubes escape, on
average. Then you would expect that.
So somewhere there must be a number.
DR. HOPENFELD: I think it varies. I don't know.
Sometimes they're 100 percent, sometimes -- I don't know how
-- what the --
DR. CATTON: If you do 100 percent then this is a
DR. HOPENFELD: I don't know whether -- I don't
know what the size of the sample of this one.
DR. CATTON: If you do 75 percent of the tubes,
there's some probability that some number would escape you.
DR. HOPENFELD: I don't have an answer to that.
MR. BALLINGER: I think you had a sample of 20
MR. SIEBER: It's 20 percent.
DR. HOPENFELD: This was 20 percent.
MR. BALLINGER: That's why I was asking if we had
the previous one.
DR. BONACA: Then if you get more than --
MR. SIEBER: So many indications.
DR. BONACA: -- so many, then you expand it.
MR. BALLINGER: Expand it, yes.
DR. HOPENFELD: Well, I don't know what the basis
of that, they didn't indicate to me, you have to read the
report. I don't know what percentage. Sometimes they go to
100 percent. I just don't know what this one is.
MR. SIEBER: You contract never go 100 percent.
DR. POWERS: We may have some authoritative
information. If you'll use the microphone, identify
yourself, speak with sufficient clarity and volume that you
can be readily heard.
MR. MUSCARA: Joe Muscara, again. This data comes
from the voltage-based criterion. It applies to the support
plate intersections. They're required to do a 100 percent
inspection. So this data is based on 100 percent inspection
of the intersections.
DR. BONACA: So the question I have now is this is
the end of cycle 14.
DR. HOPENFELD: They're in cycle 14.
DR. BONACA: Preparing for cycle 15. What
criteria do they use here to flag tubes? Is there anything
DR. HOPENFELD: Anything above -- I don't know.
It used to be one. Then it went to two and now it went to
three, if you can show that the support plate is not going
And so obviously this one, I don't know what -- I
remember Westinghouse once came and they wanted 20. So I
don't know which criteria you have.
MR. SIEBER: That's just one tube, though, right?
Do we know anything about --
DR. HOPENFELD: It's more than that. You have
MR. SIEBER: Yes, but there's one tube at 13.7.
DR. HOPENFELD: One tube at 13.7. Actually, no,
not really. If you took the POD into that, then you have to
multiply -- divide it by .6, right?
MR. SIEBER: It would seem to me the probability
of detection would be pretty good with a flaw --
DR. HOPENFELD: But you don't know what the flaw
is. Look, you can have that 13.7 here with a network, very,
very, very tight network with a lot of cracks and it's not a
crack going through the wall. You don't know what it is.
That's really the problem. That is exactly the problem.
You don't know where those cracks are.
MR. SIEBER: Well, I don't know what the
characteristics of the flaw is for that tube, either.
DR. HOPENFELD: Right. That's really the reason.
That's the problem with this concept.
DR. POWERS: If we argue, for purposes of
argument, suppose that the probability of detection is 100
percent. This is a 100 percent inspection of the tubes and
the tube support plate and surely this must represent a
failure of the prediction from cycle 13. I presume cycle
13, everything was plugged, such that they would have
expected nothing to go over whatever their voltage limit
was, which I presume was about three volts.
DR. HOPENFELD: I don't know.
DR. POWERS: Joe Muscara told us that it's 100
percent inspection at the tube support plate.
DR. HOPENFELD: He said on the 13 was also 100
MR. MUSCARA: It was 100 percent inspection.
DR. POWERS: So what the slide clearly
demonstrates is that there has been a failure of the
Now, all right, predictive methods have some
probability of failure and presumably the question is
whether this is an excessive one and I think what the
speaker is arguing is yes, it clearly is, because it goes
over the one gallon per minute limit.
DR. HOPENFELD: Another point -- Dr. Powers, there
is one more point here. I didn't bring all of them, because
there is a limit to the time we have, but if you look at
other plants, they're showing the same thing. This is not a
single event, and I will summarize that later on.
It's not a unique event.
DR. BONACA: It would be certainly for the purpose
interesting to see the previous reading for one plant to see
two or three and see how that travels, if it travels.
DR. HOPENFELD: That's correct.
DR. BONACA: And we don't see that here.
DR. HOPENFELD: Right, because the reason is I was
just trying to make my point here that it's not -- the
history is not -- you can't -- it's like the stock market.
You can't look back and say, well, you know, this stock was
doing pretty good, it's going to do the same thing next
year, it's not.
DR. BONACA: I mentioned it because to the extent
there is information you can provide over the next two or
three days, it would be interesting and important.
MR. BALLINGER: In the case of Indian Point Unit
2, the previous cycle inspection was not 100 percent
DR. HOPENFELD: But that wasn't because of -- the
failure was not in the support plate. It was somewhere in
Okay. Let me elaborate a little bit on the field
experience. I'll read this off. The first one, the Trojan
was one, it wasn't -- it was a sleeve. It wasn't a crack,
but it did show that the eddy current -- it was the first
time I think that they have eddy current the sleeve and it
was supposed to work. Well, it didn't.
This is a quotation from Inside NRC that Arkansas
had -- was consistently wrong in its prediction during the
inspection, as to what the predictions as to the voltage and
should be, and I don't know where they got the information.
They must have talked to somebody, but it's a quotation from
I already mentioned Farley. This is a very
interesting one and we're going to spend a lot of time on
this one, Breakwood and Byron, because when they had these
large, observed these large voltage growth, and I think it
was in '95, back in '95 or '96, I don't remember, even
earlier than that, they said they came to the NRC and said
we got this large voltages, would you allow us to fix that
support plate so it won't move. I won't guarantee it won't
move, although a member of the ACRS was under the impression
it's not going to move anyway.
Well, now they're coming in and they say we're
going to fix those things, we'll put some tie rods in there
or plug some of the tubes and those tubes will prevent the
support plate from moving.
And they came in and I don't know whether that's
part of my presentation, this point might come up later,
they came back and provided a rationale why that they had
the capability of tying these support plates so they won't
This is a very important point and I will come
back to it later, but it's lengthy, so I just would like to
leave it for later on.
And then recently, I understand there was some
eddy current at a plant that only a visual observation
showed that it was leaking. It wasn't the eddy current. So
what you see throughout here, throughout a period of eight
years, eddy current is not an absolute thing. It's the best
we have, but it's not absolute.
Okay. Another adjustment that we have, and this
is a very important point, I'd like to spend a lot of time
on this. If you go back to NUREG-1477, all the analysis,
all the studies are based on that pressure differential, the
delta P is the one that controls the damage.
Now, there's a large potential during the
blow-down event for damage of the tubes. You have energy
there that is way above what you need, if you take all the
tubes together, you have energy in there way above to
rupture any tube ten times over.
So the question, what's the efficiency of this
potential stored energy that you have there to break the
tubes, and then you have -- during the event, you have
hydrodynamic loads that bent the -- that could bend the
tubes, and you also have forces due to the tube sheet
moving, moving the tubes.
Now, one important point to make here, and I will
come back to this again, that these motion of the tube sheet
and motion -- the constraint of the support plates were
designed under a condition of 1,500 pounds. That's how the
plant was designed, 1,500 pounds.
But here the forces that we have are all the way
to 2,500 pounds. So the plant wasn't designed to withstand
this kind of environment.
When you go into the laboratory data, and, again,
I don't know why it's proprietary, but I may take a lot of
force to pull some of those tubes. Sometimes it may take --
they will just come out. Some, it takes a lot of force to
pull them out. So now we've got this big, huge, massive
tube sheet pulling those -- pushing on those tubes and you
have thermal expansion in the system, too.
So what do you think is going to happen? Either
the tube is going to give in, probably is going to give in,
bend or collapse, and open up more area for a leakage.
Another mechanism in there that you have is a
potential for vibration. You have -- during the event, you
have a mechanism to set up -- to amplify the natural
frequency of the tubes, and since there is a large amount of
energy, the vibration could bend them and, again, increase
So you have axial forces that can also break the
ligaments and you see that. I mean, you pull those tubes
out of there and they -- the ligaments are torn. It's not
what you have in the -- what you started with before you
pulled the tube.
So then you have excitation frequencies that may
equal the natural frequency, which what it does is really
amplifies the amplitude of the motion of the tube within the
And what I would like to do, and I don't know when
we're going to break for lunch, I would like that after the
break, that Mr. Robert Spence discuss his experience with
Now, if you remember back to the timeline,
Research goes to NRC and to NRR and tells them that we have
never experienced a steam line break in this country. Well,
that's just not true and Mr. Spence will describe his
experience with it.
The most significant part of it is that evidently
there are frequencies of that event, like an earthquake, I
guess, that excite natural frequencies or some components of
the energy sufficiently large that you can get a lot of
So I would ask him, later on -- I don't know.
When are we breaking for lunch?
DR. POWERS: We are past what our scheduled break
time is. I thought we could come to finishing this point
and that would lead naturally into Mr. Spence's
DR. HOPENFELD: Yes. I would like to get into
that --okay. So I would like to spend another five-ten
minutes about this, and then Mr. Spence will take over and
talk in the details about that. I just want to give you a
little bit more my perspective, but he is much more
knowledgeable in that.
But I would like to show you where I come from on
this, because I'm not a vibration expert.
DR. CATTON: Joe, your argument is not that that
statement is incorrect, I hope.
DR. HOPENFELD: Which one?
DR. CATTON: The first one, leakage is the
function of pressure differential only, because it is.
DR. HOPENFELD: No. If you --
DR. CATTON: Let me finish. What NRC misses is
the change in the characteristics before and after the MSLB.
It's not that it is not a function pressure differential
only, because it is. It's just it's changed. I have a new
DR. HOPENFELD: Okay. I am stating what NRC says,
that's all I'm saying.
DR. CATTON: Well, but that's true. That's a true
statement. Yes. But you have to include what happens --
DR. HOPENFELD: Okay. I see your point.
DR. CATTON: -- to the generator.
DR. HOPENFELD: I see your point.
DR. CATTON: You put a statement up like that and
you leave yourself wide open.
DR. HOPENFELD: Okay. I see your point.
DR. CATTON: Because it's a true statement.
DR. HOPENFELD: Okay. It is a true statement,
yes. It is a true statement --
DR. CATTON: However, the following is neglected.
DR. HOPENFELD: Right. You're absolutely right.
I should have said, well, if you -- it's a true statement if
the following doesn't happen.
DR. CATTON: That's right.
DR. HOPENFELD: If you have a pipe and it's
pressurized, the only driver is delta P. But if that pipe
flexes and breaks -- well, what the point is here, Ivan,
that it's not only the -- the bottom line here, it's not
only the pressure, it's also the area, the opening area is
going to be affected by these other forces.
DR. CATTON: It changes.
DR. BONACA: Yes, it changes.
DR. HOPENFELD: I just want to make sure we're all
DR. CATTON: I had trouble following what you were
talking about in your written discussion because of that
kind of a statement.
DR. HOPENFELD: Okay. I'll do better next time.
DR. KRESS: Joe, your second bullet under A, what
does that -- would you explain what the stored mechanical
DR. HOPENFELD: Okay. You have -- you have this
big vessel, ten feet in diameter, 30 feet high. You have
inventory of water in there at 1,000 pounds and at 550
You suddenly, all that energy -- suddenly you open
the cap, all that energy comes out. So if you look at the
enthalpy of this, there is a lot of energy in there at that
DR. KRESS: That's the enthalpy you're calling the
DR. HOPENFELD: That's the thermal energy, right.
DR. KRESS: You're calling enthalpy mechanical
DR. HOPENFELD: Right. The thermal energy. You
say we'll take that and you say, well, that's the first
thing, if it doesn't pencil out, you don't look beyond that.
And this does show you that there is a potential there, and
it doesn't mean that -- I don't know, conversion could be
Mr. Spence evidently had seen one, he had been
next to it, and he'd tell you that it wasn't a little wind
passing by and he'll tell you the thing flew up 150 feet in
So the energy is there. Now, what damage it's
going to do, I don't know. I honestly don't know. But
these people, when they come tomorrow and talk to you about
this and they'll tell you about all the test data they ran
under internal pressure only, I think you ought to ask them
about this and it's the burden on them is to prove to you,
to show you that these delta P's, that's all there is.
They've made the calculations, and I'll go back into that
and discuss this thing after the lunch.
DR. POWERS: I think, at this point, we can take a
recess for lunch until ten after the hour.
[Whereupon, at 12:10 p.m., the meeting was
recessed, to reconvene this same day at 1:10 p.m.]
. AFTERNOON SESSION
DR. POWERS: Let's go back into session, and I
guess, at this point, we'll turn the floor to Mr. Spence.
MR. SPENCE: Thank you.
Dr. Hopenfeld asked me to discuss the resonance
vibrations that I witnessed during a main steam line break
at Turkey Point 3 in 1997, as well as review past operator
experience on steam generator tube ruptures, which I'll do a
little bit later.
I should say that, in so doing, I'll present my
own views only, and experience, in the role of independent
reviewer of U.S. and foreign operating experience that I've
done for AEOD and Research for the last 10 years as a
reactor systems engineer.
To establish my credentials to talk on this
subject, I've been a member of NRC augmented inspection
teams and human performance teams investigating operator
performance during events.
To qualify as a headquarters operations officer,
where I assessed the safety significance of reactor events
in real time to initiate the NRC response, I attended NRC
systems and simulator courses for each type of nuclear
reactor in this country, including Westinghouse.
I've been in charge of the conceptual design of
the nuclear island for a 600-megawatt, barge-mounted
Westinghouse reactor for offshore power systems.
Earlier, as a systems engineer, I worked on heat
exchangers, valves, pipes, piping design.
I worked as a turbine operator during a three-
month strike and a licensed research reactor operator, and
most importantly to this effort, I was a start-up engineer
working for Florida Power and Light in the operations
department during hot functional testing on the two
Westinghouse units at Turkey Point, when the main steam line
break occurred there.
Now, when -- I knew very little about this DPO
until the EDO appointed me at Dr. Hopenfeld's request to
serve on a previous DPO panel.
As Dr. Hopenfeld mentioned, the ACRS approved
Generic Letter 95-05 as an interim measure, which has become
NRR is approving increases to Generic Letter 95-05
alternate repair criteria to 3 volts now.
But while I was on the panel, I started reviewing
stuff and found many questions about the technical basis for
I found unpredictable tube leaks, tube leak
breaks, and little defense-in-depth against the main steam
line break that, in my opinion, will result in steam
generator leaks or ruptures.
Now, I didn't form that opinion while I was on the
panel but only afterwards, when I was researching it in more
detail to appear here.
After a number of Dr. Hopenfeld's safety concerns
were ruled out of scope, I recommended that the EDO dissolve
Now, I cannot adequate address those issues in a
public forum because of the restrictions placed on the use
of proprietary design and test information and the emergency
I did provide you a copy -- a proprietary copy of
my slides for GSI 188.
There were concerns this morning that Dr.
Hopenfeld mentioned that he said he couldn't identify.
Some of those issues, some of the numbers,
etcetera, are in that package that you have.
The heart of this matter is really in the
proprietary information, and I hope that you have access to
that to review it.
I noticed you mentioned that you didn't.
DR. CATTON: I didn't look at the stack of stuff
given to me this morning. Maybe it's in there.
MR. SPENCE: Okay. Only part is, but there's
masses that aren't.
At NRR's recommendation, vibration during a main
steam line break is now being considered as a potential
generic safety issue, and that's what that's all about. We
meet next week.
That identifies technical inconsistencies in the
basis -- in the technical basis for GL 95-05. It also poses
questions about the Robinson 2 and Turkey Point 3 cladding
separation, tube leaks, and main steam line breaks that may
necessitate on-site investigation to answer.
The GSI panel -- it's my understanding the GSI
panel will not be allowed to do that.
I believe that these issues will simply be
incorporated into Dr. Hopenfeld's GSI 163, which has lain
dormant for many years despite its high priority.
The GSI 188 panel can only recommend whether more
study is warranted, but only your panel can recommend
whether Generic Letter 95-05 should be rescinded or can be
I believe there's enough evidence available that
you will be able to make a decision on the fate of GL 95-05
without further research.
In the few minutes Dr. Hopenfeld asked me to talk
in, I can only summarize the lessons that appear to have
been forgotten from the cold hydros and the steam line
breaks at Robinson 2 and Turkey Point 3 without going into
Both experienced cladding separation and tube
leaks as a result of their cold hydros.
You have the information on Robinson 2 in that
That's the Robinson 2 steam generator with the
tube sheet and the divider plate, okay. The next slide's
going to show this area here and what happened to it during
the cold hydro.
There's going to be three areas that, when you
have 2,000 pounds or 2,500 pounds or whatever the pressure
is in the reactor vessel or the safety injection system or
CVCS charging pump pressure in here, and this, during a
steam line break, is going to be open essentially to
atmosphere, assuming that the leak rate is small at that
point, this here will curve up a little bit, in the middle
of the D, and this here -- this also in here will raise, so
you'll have a little bit -- you'll have like that, as well
as superimposed -- I don't know. I didn't do that well.
DR. POWERS: I think we understand.
MR. SPENCE: You got the idea? Okay.
This is -- this came out of one of the Robinson
reports that I believe you have, but it shows the cladding
separation in here, it shows the weld problems here, and
this was on 20 to 30 tubes, and it affected the first row --
80 tubes worth of the first row.
Okay. And they observed cracking in this area
here. This is the divider plate. It had come across the --
this is on the welds of the tubes, okay?
Now, I do not have pictures of the Turkey Point
thing, but at Turkey Point, after my crew was running the
hydrolazer pump, we had trouble maintaining pressure and
barely got done with the inspection of the primary system
before we went down to lower pressure.
Afterwards, I stuck my head in the steam
generator, saw the drips coming from a tube and surrounding
wetness on the tube sheet, okay?
I didn't see any cracks, but for example, if this
was the tube, stuck my finger up inside there and noticed
for sure that it was leaking inside the tube, not at a weld,
and then I noticed water around here that was not leaking,
okay, but it wasn't right next to the tube sheet.
I can't tell you which steam generator it was. I
can see the leak, but I can't tell you what steam generator
Because of -- now, at Robinson 2, they ran the
test pressure basically at 3,000 pounds, 3,100 pounds on the
primary side, and they had the secondary side of the steam
generator open to atmosphere.
Hearing of the results of their cold hydro, Turkey
Point went ahead and pressurized the secondary side of the
steam generator so that the delta P across that tube sheet
would only be a few thousand pounds, perhaps a little bit
larger, 2,000 psi, and I assume it was not a hydro test
pressure but some kind of normal operating pressure, but I
have not been able to find the numbers.
Regardless, at Turkey Point, the cladding
separation was not as severe as experienced at Robinson 2,
but it still occurred at differential pressures across the
tube sheet that could happen in a main steam line break.
Now, despite its 2-foot thickness, the tube sheet
will bow up slightly in the middle of each D, as well as the
divider plate, and it will push up some of the tubes,
probably the ones in here and the ones in here worst.
The tubes are held in place to tube-to-tube-sheet
welds in this area here, okay, and metallic oxides in the
gaps between the tubes and the tube support plates.
If the tube support plates remain in place, then
the tubes have either -- either have to bend or slide
through the tube support plate.
However, parts of the upper and lower tube support
plates are expected to vibrate at their lowest natural
frequencies despite stay rod, spacers, bars, and wedges
welded to the wrapper.
Now, on page 48 of your proprietary hand-out, I
think, based on a tech spec amendment that South Texas
project recently put in, that I believe has not been
approved yet -- is that right?
MR. LYON: That is correct.
MR. SPENCE: Okay. Thank you.
They only asked for a 3-volt alternate repair
criteria increment for tube support plates 3C, Charlie
through M, Mother, because they, quote, did not -- do not
deflect significantly relative to any tube during normal
operation or design basis accidents.
Well, that leaves tube support plates A, D, N, P,
Q, and R.
Now, A is the first tube support plate going
across here, B is an economizer section, and N through R are
all the way up at the top.
DR. SIEBER: What model steam generators are
MR. SPENCE: Warren?
MR. LYON: Warren Lyon.
I believe that's a Model E, if I remember
MR. SPENCE: 44-E?
MR. LYON: All I remember is the E.
MR. SPENCE: Okay.
Now, I have personally seen the resonance
vibrations from the steam line break at Turkey Point set up.
I tried to describe my experience in the May 22nd,
memo that you have copies of, of this year.
Just very briefly, I came out of the control room
a little after seven o'clock in the morning to do a round
I was the shift start-up engineer on, and I saw
this valve with the steam coming out of it. We had already
had leakers on here. So, I went up to investigate it.
I got about to this area, and there was something
in here -- I think it was some type of scaffolding, but I
couldn't see the top of this. But I started hearing
simmering noises, and they were increasing.
I had already been through a steam line break over
in the fossil plant, I knew what it sounded like, etcetera,
and I went from here to here to here to here, and it blew
off the line.
This here cracked, double-ended guillotine, if you
will, right at the full circumferential break here.
These two hit the containment wall and then went
up, and by the time I was here and I heard the noise, I
looked up, I saw the valves, and they were about at the --
towards the top of the containment, and then they fell down
a couple stories down.
The second boom came from here, and this valve
here blew off.
What they later found the problem was is that the
piping system was designed -- was not designed to take the
forces, the reactor forces here from relieving of the
valves. These valves previously had only been pop-tested
One point I will mention, in the proprietary
thing, I've given you my opinion on what would have happened
if these valves had not blown off at that particular time.
This was, again, hot functional testing; there was
no fuel in the pot.
There was no procedures in place to test the
operation of these valves.
The SAR said that they were going to look at doing
a trip test on the turbines as the power level increased,
but the procedures did not call out for that.
This was December 2nd of '71. On November 24th of
'71, the AEC identified that little problem to Florida Power
and Light Company. It would have been a most interesting
There were a couple of causes for this, and that
is, the supports here were designed so that the steam line
had to grow out this way, okay, and it had to move -- these
things here had to move north and south. This is north,
that's south. So, it had to grow east, south, and north at
the same time.
When this valve is shut, okay, this cannot move --
the thermal expansion is such that it's not going to move to
the west, okay?
So, now you're also putting a very high stress on
the saddle here.
I think you'll find that the reports I'll pass on
to you in the next few days do not have a specific number,
okay, on stress level, but to give you an indication,
bending moment here is 188,000 foot pounds.
Now, the results of the steam line break were
The piping here, especially down below, moved six
to eight inches. You could see it on the supports.
The turbine duct here was oscillating maybe a
couple inches up and down. Everything was moving at
Stuff was going to its lowest natural frequency,
When I got into the control room, stuff in there
was moving around, as well.
The noise was such that the operators could -- we
basically could not communicate, okay?
If the safety valves were to have lifted and then
supported properly, etcetera, okay, the safety valves can be
designed to avoid frequencies under 50 hertz. Pipe breaks
cannot be tuned that way.
Now, I've been unable to find data on the lowest
natural frequencies of the tube sheet, the U-tube assembly,
which will have its own individual frequency, or tube
sections between the tube support plates, okay?
The tube support plates and tube sheets interact
in an extremely complicated three-dimensional manner during
a steam line break that, to my knowledge, has not been fully
In fact, the RELAP-5 computer code that the NRC
and Westinghouse relies upon is a one-dimensional thermal
hydraulic code, it does not model two-phase transient
turbulent flow, cross-flow, or time-dependent resonance
vibrations or frequencies.
You've been supplied with a copy of a letter from
Dr. Ward to me on that subject, and Dr. Ward -- I understand
NRR has got him here in the next day or so, so you can ask
him about it, if you'd like.
I believe that resonance number one will be set up
within the steam generator.
If they're strong enough to do what I observed,
they're strong enough to go back through the piping system,
through the shell, through the wrapper, through the tube
sheet, through the hydraulics of the oscillations of the
Now, that's something I didn't tell you about.
Sonic booms occurred at maybe one to three cycles per
second. In an emergency, I don't judge time right, you
know, so don't quote me on that one, but it's boom, boom,
and it was very noticeable.
What was interesting to me is that continued at
the same rate, not like you'd expect with a pressure
decreased in the steam generator but at the same rate
throughout the entire event, until the end, and then, it
became longer and even louder, with some gigantic booms
before it stopped down to quiet, okay?
The reason I started getting into this is because
I think it's NUREG 6365, as well as 1477, have diagrams of
the pressure -- secondary pressure, and they didn't look
right to me, because they didn't match what I had
experienced, and I started talking with Dr. Ward about it.
Now, I believe the resonances in the steam
generator are going to cause tube bending, which is going to
increase the crack growth, as well as movement between the
tubes and the tube support plates.
It's going to increase the crack growth by erosion
of the lengths between the cracks, both micro cracks and
macro cracks, through both wear and ablation, and later,
you'll get some theoretical -- I think it was ASME articles
on the wear evidenced by movement of tubes.
I think it's going to expose some of the cracks in
the intersections that are allowed by GL 95-05 to the
secondary side and, hence, to the atmosphere.
Now, when you have a tube support plate that's
partly moving up here and the next tube support plate is
moving down here because you have different frequencies --
each one of the tube support plates has got different
frequencies, and that's all proprietary information, and
that's about all I can say about that, but take a look at
Now, if there's cracks in there and those cracks
happen to be especially towards the tail-end of the voltage
distribution, those are going to be the biggest cracks, and
those are the ones that are going to open up.
The closer it gets to through-wall, the more
chance of it opening up during a main steam line break.
There's also some proprietary information that,
once a crack gets outside the tube support plate
intersection area, there is a very low correlation between
the exposure of that crack -- let's say, for the sake of
argument, that this is a tube support plate.
If the crack is in here and it comes up in here
versus it comes down to here, there's very little
correlation between the amount of flow coming out of that
crack and the amount of distance that the crack is exposed.
Now, the size of the steam line break and the
operator actions that are taken will have a significant
effect on the amount of time this common mode failure
mechanism is working.
The longer the mechanism is working, the more
crack damage they'll cause.
DR. SIEBER: Question.
MR. SPENCE: Yes, sir.
DR. SIEBER: From your observation at Turkey
Point, once this transient has gone to termination --
MR. SPENCE: Yes, sir.
DR. SIEBER: -- if you inspect the steam
generator, do things come back to their normal position, or
is everything permanently upset or deflected or bent or what
MR. SPENCE: I did not inspect the internals. I
cannot answer that question, but it's a real good question.
I inspected -- a reactor operator and I went out and
inspected the area before we turned the RCP back on, and we
could observe some movement inside the containment, but we
couldn't observe any damage, any movement out -- you know,
that the steam generator area was not designed for.
DR. SIEBER: The reason why I asked the question
is that, if you did have a permanent deflection in a tube
support plate, that would potentially uncover permanently
some of the cracking that would have occurred and leave you
in a more vulnerable situation as far as crack growth and
MR. SPENCE: I will be showing an interesting
slide a little bit later on that subject.
DR. BALLINGER: Can I ask a question?
MR. SPENCE: Yes, sir.
DR. BALLINGER: Was there any estimate of the
amount of flow through that opening, as compared to the
normal 100-percent-power steam flow?
MR. SPENCE: I don't remember seeing a number. I
know, in CE reactors, their restriction orifices on the main
steam lines are 170 percent. I've heard that, in
Westinghouse, it may be less than that. I'm not sure if
that number is proprietary or not.
DR. BALLINGER: You have several relief valves,
and it looks like two of them blew off.
MR. SPENCE: Three of them.
DR. BALLINGER: Out of four?
MR. SPENCE: Yes, sir, on the one line.
DR. BALLINGER: On one line, but there are --
there's another eight.
MR. SPENCE: On the other two steam generators,
DR. BALLINGER: Okay. So, I'm just curious as to
what fraction of the flow you had compared to the 100-
percent power flow from that steam generator.
MR. SPENCE: The report I read said that the --
I'm sure it would have been in excess of 100 percent, but
exactly how much, I don't know.
The report I read said that the flow was within
the restriction orifice capability. In other words, the
restriction orifice was not choking the flow. The choking
was coming at the break area.
DR. SIEBER: Do you have any idea of what
difference in flow there would be with the valves just blown
off, compared to all those valves open, like you would get
on a reactor trip?
DR. BALLINGER: That's what I was getting at.
MR. SPENCE: Okay.
DR. SIEBER: They're, what, two-and-a-half-inch
MR. SPENCE: There's going to be a difference for
the steam generator, because the steam safety valves are
DR. SIEBER: Right.
MR. SPENCE: They're not going to have a 50-cycle;
they're going to have higher, okay?
So, I wouldn't worry about a full thing. What it
is, it's the cycles that the steam line is --
DR. SIEBER: This is the pulsation.
MR. SPENCE: The pulsations from the steam.
DR. SIEBER: Right.
MR. SPENCE: That's what's going to kill it.
DR. SIEBER: Okay.
DR. BONACA: You're not aware if, prior to
operation again, the plant had a major outage on the steam
MR. SPENCE: The plant was -- it was delayed by
more than half-a-year because of this.
I mean it took -- I didn't tell you about all the
damage, but it took -- I remember, I couldn't get back up
here to look at this.
I think this thing was gone.
There was damage to this valve, to a bypass valve.
There was a pressure -- a tap on here that was gone. This
whole -- it was a mess.
DR. BONACA: The question I had was regarding the
steam generator, specifically, the internals. Do you know
if there was major work done on those?
MR. SPENCE: Shortly after the blow-down, the
operations superintendent and I were talking about it, and
we were both concerned about the steam generator internals.
I don't know the full results of that, but I know there was
some investigation done.
The reports show that there was a one-word
sentence that Westinghouse steam generator experts looked at
it and said it was okay.
DR. SIEBER: Does the Model E steam generator have
a pre-heater on it, as I recall, with cross-flow paths?
MR. SPENCE: Yes.
DR. SIEBER: And would that make a difference as
far as the dynamics of the way the steam generator operates?
I think the cross-flow is on the cold leg side.
MR. SPENCE: Yes, it would.
DR. SIEBER: To me, it would change the vibrations
and also change the forces the tube sees due to the flow
while the stuff is rushing out of there.
MR. SPENCE: That's exactly what I was talking
about with the RELAP code, is the cross-flow would be
DR. SIEBER: I don't know how you would model
that, but a Model 51 or 53 doesn't have that feature to it,
MR. SPENCE: I would suspect that it would have
less trouble, and it depends where the break is.
In a steam line, it's one thing. If the break's
in a feedwater line, it's another. So, now you've got
everything going down backwards, too.
DR. SIEBER: Well, on the Model E, the feedwater
came in pretty close to the bottom, as I recall, whereas 47s
and 51s, it came in at the top, went around the sparger ring
with the J-tubes and then down the wrapper, and so, if you
broke the feed line in a Model 51, I think you would get
steam out, as opposed to hot water, because it's high in the
The only thing that goes in at the bundle area is
the aux feed, right? It also goes in pretty high.
DR. HIGGINS: Do you know if, during any of the
testing that the NSSS manufacturers did, they did main steam
line break simulations with, say, scaled models on steam
MR. SPENCE: I have not seen -- they did the MB2,
but that wasn't --
DR. HOPENFELD: No.
MR. SPENCE: No, I know of no such testing.
That's not to say it doesn't exist, you know, but I haven't
DR. HOPENFELD: They may have done it in-house at
Westinghouse, but they haven't provided us the data.
DR. CATTON: Those of us who tried to do
experiments to simulate the GE system ran into all kinds of
problems with this.
What you have is you have a series of contractions
in the flow paths, so you have volumes, and all of these
start to interact.
They choke and un-choke, and each time they choke
and un-choke, you get pressure spikes. It's just really
And then you compound it because of the high flow
There's a critical velocity above which you begin
to get fluid-induced vibrations, whereas a CE generator,
they know what these are, because they measured it.
The Westinghouse, as near as I can tell, guess at
it, and actually, at North Anna, they got in trouble when
they changed the recirculation ratio just a little bit.
If you're anywhere near this, you're going to get
all kinds of tube vibration.
I doubt that it would impact much a new generator,
but if you had one that had cracks that were -- it's
certainly going to loosen everything up, and then the
supports -- at each set of supports -- and I don't know how
many you mentioned, but it sounded like there were a lot.
As you depressurize this system, you are going to
get pressure loads across these supports, and they are going
to move up and down, because they're going to choke and un-
It's a rather chaotic process.
DR. BALLINGER: If you have dented support plates
where they're cracked and you have this kind of thing,
what's likely to happen to those support plates? In some of
these older steam generators, there's a lot of hourglassing
in the flow slots, and there's a lot of cracking in the
So, that's another complicating factor.
MR. SPENCE: Let me throw in some more
complicating factors, and then I'll talk a little bit about
time and break size, too, because a smaller break will take
longer to depressurize the steam generator. That means the
longer steam generator tubes are subjected to these common
mode failure mechanisms.
In Robinson 2 case, they only had a six-inch
break, and it took them an hour to depressurize. In Turkey
Point 3's case, it took only a few minutes. They say they -
- it was about three. I think it was about five, because I
know it went down below -- it still kept -- it kept blowing
after the instrumentation went to zero.
It took three to five minutes to depressurize
during the initial blow-down, but then the steam generator
As I was leaving, about an hour later, it was
blowing out the holes at, I'd estimate, 15 to 20 feet per
Now, if my analysis is correct, one could expect
to see tube leaks or tube wear in grooves where the
magnitude of the differential movement between the tube
support plates was high.
I have no data on Robinson 2, but Turkey Point 3 -
- can you pick out the steam generator that had the problem?
Interesting, right in the groove, and this is the 3A steam
generator where it happened, this is B, and that's C, and I
think there were a couple little things down there, okay?
A and C had the cladding separation during the
cold hydro, B did not. So, you might have some effects from
Now, the numbers that were -- the numbers, I
think, 20 -- I think if you count these up, you got more
Now, this was for the first ISI inspection in '74.
So, that was basically one year's worth of service, okay?
I talked to a metallurgist out at Region III,
said, hey, it takes a few years for the corrosion products
to start doing the kind of stuff that we're talking about,
that would normally take out tubes in the steam generator.
DR. BALLINGER: Let me be clear on this. This
generator is the one that went through the cold -- they all
went through the cold hydro.
MR. SPENCE: They all went through the cold hydro,
but these two had the cladding separation, A and C. B did
not, for whatever reason, okay, and A is the one that went
through the blow-down.
DR. SIEBER: After the blow-down, did they do a
pre-service examination again, or had they done that before
and said we don't need to do one?
MR. SPENCE: I'm going to give you a smart-aleck
answer. I don't know, because I left. I got an office job.
I used up too many of my nine lives there.
Anyway, there here -- now, when I was on the
panel, I suggested we go down to Turkey Point and find out
what caused this, let's find out when the tubes were
plugged, etcetera, let's check the traces on the eddy
current test, and that was ruled out of scope.
I also asked my supervisor here -- excuse me -- my
manager, with respect to the GSI 198, and that's not going
to be done in that one either.
It would be a nice idea to check what happened to
the tubes at Robinson 2 afterwards.
Resonance vibrations and tube-to-tube-support-
plate movement are not modeled, and GL 95-05 required
industry testing of tube samples.
I also noted some other concerns about the
industry testing, but they're proprietary in nature, and
they're in your packet.
Resonance vibrations and relative tube-to-tube-
support-plate movement during main steam line breaks are
common mode failure mechanisms that can drive the issues in
this DPO, and I think that's exactly why he asked me to
serve on the panel, when he heard about my experience down
at Turkey Point.
These common mode failure mechanisms would
invalidate any risk analysis the NRC and industry used to
support GL 95-05 and conclude that the frequency of major --
multiple major tube leaks or ruptures during a design basis
steam like break would be on the order of 6 times 10 to the
minus 6 or 1 times 10 to the minus 7. Those numbers, of
course -- if that were true, then it would not be a problem,
okay? I don't believe it.
Now, a little later, I'll talk about how operator
response to past steam generator tube ruptures may be
related to the risk of the steam line break.
Thank you for your attention and for your
DR. BALLINGER: I have a question.
I'm looking at that figure, and it's fuzzy, but
the legend says that the triangles are plugged due to
thinning and the little dots say they're affected by
wastage, cavitation, and erosion.
Now, that implies, at least, that they know what
those were, not necessarily due to cladding separation.
MR. SPENCE: What's interesting, if you really go
down the line, you won't find any of them that say there was
DR. BALLINGER: That's right.
MR. SPENCE: And even though I'm here to tell you
I've seen it, I've seen at least eight leaks.
Reporting, back in those days, was not quite as
robust as it is now. At least, I hope it's more robust.
DR. CATTON: Robust just means that it happens.
DR. HOPENFELD: Thank you very much, Bob.
I think this presentation puts the idea of POD and
the statistical differences in perspective. I think you can
see there are much larger uncertainties here that we don't
even an ability to cope with.
Let me just briefly summarize basically again.
You put those defective tubes back in service, and
at some time during the zero to 18 months, you have a
rupture. Initially, you have a large energy release which
takes on the order of minutes, and again, it depends on the
size of the break. It may take even longer.
Later on, you get into a longer period of time the
tubes may be exposed to flow vibrations and then also to the
motion of the tubes due to bowing or thermal expansion of
the tube sheet.
Now, it's important at this point to note -- and I
think that's where -- that the plants were not designed for
that kind of a pressure. They were not analyzed for that.
They were designed for 1,550 and 1,600, and all the data
that is being presented to you, even though the severe
accidents that are being conducted at Argonne are based on
2,500 and do not consider the other forces that can come
into play, it's fine to study creep rupture, to study
ligament breakage, and model that under internal pressure,
but that's not really the main issue, unless you can prove
that that is the driving force for the fracture.
So, basically, if you put yourself in a position
of the ligament in a cracked tube, really what makes a
difference is the stress state of that ligament and the
pressure forces acting on it due to the delta P in the pipe,
but there are other forces, as we've heard before.
So, the question is really, what's the largest
driving force for breaking that ligament, and the person
that comes in and shows you all that data -- it behooves him
to tell you that, really, the driving force show you, prove
it to you, that what really drives this thing is the
internal pressure, and they haven't done so, and I hope
that, tomorrow, you will ask him to show you the numbers
where all these other forces have gone, why are they not
Going back to the vibration thing, you can see, in
the typical steam generator, you have a range of -- this is
a simple equation of calculating frequency, depends on the
span length here and the rate use, and this is the
properties of the fluid and the modulus of inertia.
The main point of this graph is that, when you
have rate uses varying between one to two feet, you have a
whole spectrum of possible natural frequencies, and
therefore, a whole bunch of possibilities for exciting some
of the tubes to start vibrating.
That's the whole purpose of this graph.
Now, this is probably known to the Westinghouse
people. They ran a lot of vibration tests in Florida,
especially under heaters, I believe, that they had vibration
problems. So, they realized.
But when they come here with an application to
relax the 2-volt requirement to go to higher voltages, they
provide analysis of what are the forces on the support
plates, and those forces are basically based on delta P
across those plates.
Now, what they have claimed is that they could
take a code like RELAP-5 and benchmark it against some data
that was conducted 15 years ago on the prototypic steam
generator. That was a MB2 program.
Let me tell you briefly a little bit, because it's
very, very important, and the reason it's important, because
it shows that -- in those particular case, it was ComEd that
came in for an application to relax the GL 95-05, and it's
very important to understand how the process works.
They take the code like RELAP-5, modify it, and
they claim they benchmark the thing against some data that
was conducted on the prototypic steam generator, and then
it's being approved by NRR.
That particular test had a slice of a full-size or
95-percent length of U-tubes but was only a slice of the
The tubes were enclosed in a large vessel. Most
of the volume, as you will see later, was really occupied by
The volume ratio, as you can see here, was all
basically that empty space. This was only the volume of the
So, when you benchmark a code against something,
the first thing you do is see what the scaling factors --
whether the scaling factors allow you to do that, and when
we have looked into the scaling of this -- I have a report -
- it wasn't meant to be scaled to study the dynamic aspects
of that kind of a phenomena.
We didn't have any accelerometers on the tubes to
measure any vibrations.
Nevertheless, this experiment is being used as the
justification to ignore vibration, to ignore all the forces
that you have on the steam generator, because it was
benchmarked, so to speak, against prototypic data.
You can see some of the results -- and again,
these came from computer codes -- compare the flow quality
with and without the dead space, and you can see that this
experiment really had nothing to do with the forces that you
will have during a steam line break. Yet it is being used
as a explanation, as a reason why you could operate under GL
95-05 at much higher voltages.
Now if you design a washing machine or something
and you want to go and put it on the market, you go to
Chicago, go to the UL people and tell them I have this
washing machine and I made all these calculations and could
you give me your stamp, because I don't think it vibrates,
I've got all these computer codes, they say it won't
vibrate, they'll throw you out.
They'll say, well, we want to test it. They're
not going to put their stamp on it.
But Westinghouse comes in here, or Com Edison --
the work was done by Westinghouse, and we approved it.
We approved that thing without asking any
questions, and you can go back, and I think you have the
SERs, and not even one question that goes into why can't we
just neglect vibrations?
The next potential damage mechanism is due to
erosion from jets.
You have 2,600 pounds or 2,500 pounds on one side,
and you have zero pounds on the other side. You have a
temperature that varies between 1,700 F under severe
conditions and, I think, 550 under normal conditions, and
you have a whole range of abrasive material present.
Now, if you have any one of those two-phase flows,
could very severely penetrate and damage the next tube.
Just to give you an idea, this is a piece that I got in a
It took a few seconds to make these slots with an
Obviously, it depends on -- this is aluminum and
it depends on the velocity and the pressure and the size of
In machining -- in regular machining, the
pressures are more than an order of magnitude, 15 times as
high as what you get here.
On the other hand, these things take on the order
of a few seconds.
Here we can have minutes or maybe even hours where
that jet could cause the damage, and the main problem here
is that we can't predict -- I believe it's impossible to
predict how much abrasive particles you're going to have.
You have corrosion products on the primary side
and especially during depressurization you have what's known
as particle burst.
Then you have this big sludge pile on the bottom
that you have all kind of material in there.
You basically have the entire periodic table, the
source is there, and you can go ahead and use your
imagination how it's going to be trapped, whether it's from
the sludge pile or from the primary side or for in between
the cracks in the support plate.
So, there is a potential here, and in my mind,
it's almost impossible to predict, but the people that did
the research said that they know how to do it, they've got
these computer codes, VICTORIA, I don't know the various
names they have, and they can predict exactly how many
particles and what their concentration and they're already
So, I'd like to go later on and talk about that a
little bit more.
DR. SIEBER: The sludge pile you're talking about
is not the one that's ordinarily referred to that lies on
top of the tube sheet.
DR. HOPENFELD: Yeah, that's the one I'm talking
DR. SIEBER: That's on the secondary side.
DR. HOPENFELD: Yeah. If you have a jet coming
out somewhere, it will carry some of that particle on the
Now, exactly the mechanism, how it does it, I
don't know, but I'm identifying here sources for particles.
DR. SIEBER: All right.
DR. HOPENFELD: The primary, the secondary, in
between, and who knows where else?
I mean I really don't want to spend my -- I didn't
want to spend the time to get into the detailed mechanism.
I'll leave it to those people who write papers, because you
can come up with an infinite number of mechanisms, and it
depends on your imagination, but the source is there.
Now, the material that you have is basically -- on
the primary side, it's chromium, cobalt, whatever corrosion
products you have, and silver that comes during the severe
accident, I think there's a lot of silver in there, plus you
have all kind of aerosols in there.
Now, originally, when I started calculating this,
I took some equations that came from several power plants on
erosion of blades from droplets from wet steam and I've
calculated erosion rates or penetration rates through the
next tube on the order of -- I believe it was on the order
of minutes, but realizing that there's probably an order of
magnitude, at least an order of magnitude of uncertainty in
these kind of calculations, but it's an indication the
potential is there.
Later on, the NRR people got some data on -- from
a coal gasification program, and they came up with very,
very fast penetration due to these hot jets, and I think
they came up with something on the order of it took 30
seconds or so to penetrate through the wall, on the average.
It depended whether there were particles or not
particles in the stream.
There is an industrial experience especially in
the pulp and paper industry.
In the early '70s and in the '80s, there are a lot
of steam explosions occurring in capped boilers, and the
reason for those, really -- there are many reasons, but one
particular one, or two of them, that I'm personally familiar
with -- they were initiated with a pin-hole leak in one of
the tubes that penetrated the next tube, which was about 1
inch, and all that water was dumped on that pile of green
liquor that sits there on the bottom of that boiler.
It's a big capped boiler. It's a water-cooled
boiler about 30 feet high, and when you damage one of those
water tubes, all that water dumps into the bottom, and you
have a big steam explosion.
So, this is not completely way out. There is a
potential here for damage because of jets, abrasive jets.
It depends what the concentration of the particles is and
what are the particles.
You can also have probably a clean jet. In fact,
they use a water jet, very small, thin water jets to cut
wafers in the electronic industry.
So, it depends on the -- what's going to happen
here, but you cannot ignore it, and that's really my point.
I can't prove that it's there or not, but we've got to
consider that, and I get really kind of very shaky when they
tell me that the RELAP codes and all these things are going
to predict the particle size, and I'll go back into that and
tell you why they cannot.
So, basically, on this subject, you have a crack,
depending on what the pressure is the velocity, somewhere
down you will form a two-phase mixture, drop and solid
particles, and they will impinge on the next tubes.
Now, this is -- because the next tube is already
corroded, the surface is already gone or cracked, you get
into a brittle type of erosion.
You don't need much plastic deformation to cut
through here, and you don't know how much you really need to
damage it, but that's the kind of thing.
If you want to run some tests, you can't just
start with a nice clean piece of metal to run tests on, and
it will probably require many samples.
Now, the research people say that they can get
this information within several months, and NRR is very
happy with that.
DR. POWERS: Joe, I think I understand how a jet
can impact an adjacent tube. What's not clear to me is how
it propagates any further than that.
DR. HOPENFELD: I'm sorry?
DR. POWERS: If I've got two tubes, one of them
leaks, and a jet cuts through the adjacent tube, how does
damage propagate any further?
DR. HOPENFELD: Oh, okay. That was my next slide.
DR. POWERS: Oh, I'm sorry.
DR. HOPENFELD: Usually those jets expand, and it
depends on what is a two-phase, one-phase. If it's one-
phase, just plain water, it's not going to expand. If it's
pure steam, it will expand quite a lot.
So, you have something like 400-feet-per-second
jet hitting it, usually you fan out. As a rough
calculation, you can say that you'll double its initial
size, and then this one will double again, and I don't think
you have to go too many of this.
So, that's the potential mechanism for enlarging
the area between the various jets.
This one will open, and this next one will go, and
you can see that very, very fast. You start with two, then
within -- what do we have here? -- two minutes, you have 16
So, it doesn't take many of those 7.6 gpm cracks
there, the tail of that distribution, to start you going,
and you have -- if you look at the transient, you about an
hour to do this.
Now, what the NRR people have done -- and it's
discussed in my DPO -- they have, after a long time, agreed
that there's a potential problem, we ought to look into --
it's under severe accident condition.
The same thing -- maybe the chemistry of these
things is different, but you have the same potential
mechanism during the design basis accident.
DR. BONACA: Let me ask you a question about that.
DR. HOPENFELD: Yes.
DR. BONACA: You say one hour.
If I have a steam line break --
DR. HOPENFELD: Right.
DR. BONACA: -- my primary side will depressurize
DR. HOPENFELD: Right.
DR. BONACA: -- below the head of the HPCI system,
and then, if I have no steam generator tube rupture, it will
repressurize to the head of the HPCI, say 1,400 psi.
DR. HOPENFELD: Right.
DR. BONACA: If I do have a hole, it will
naturally depressurize to some intermediate level between
the high head of the HPCI and somewhere below, because --
DR. SIEBER: Because of the pump curve.
DR. BONACA: Because of the pump curve. So, the
pressurize to which it is exposed now, the jet, will not be
coming in at the same velocity and the same -- I'm saying
that -- you know, I'm trying to understand the timeframe for
this, and it seems to me that larger is the hole by which
they are pressurized, okay, and more you have
depressurization on the primary side that you can now really
repressurize by itself, because you are leaking out of the
secondary side, so that the phenomenon will be self-
containing a little bit?
DR. HOPENFELD: I don't think so. No, I don't
believe it's going to be containing, because -- well, maybe
the pressure may fall down, but the maximum pressure -- it
goes back up to 2,500, and that's the reason that they are
testing it at 2,500.
DR. BONACA: I'm saying, by the time you have a
hole, say one rupture, it's not going to go back up.
DR. HOPENFELD: It depends on how many do you have
and how does that affect the pump.
I mean if it's very small originally, then you
DR. BONACA: It cannot go beyond the shut-off head
of the high-pressure injection, which is typically 1,400 psi
DR. HOPENFELD: That's right.
DR. BONACA: Okay.
DR. HOPENFELD: But I don't know how long it takes
to get there either.
DR. BONACA: All right. Well, the blow-down
typically takes you below that in seconds.
DR. HOPENFELD: Right, but then it comes back.
DR. BONACA: Yeah, if you have no steam generator
DR. HOPENFELD: It depends on the relatively size.
DR. BONACA: If you have a tube rupture, then even
for one or two, you're going to come back to the shut-off
DR. HOPENFELD: Depending on the relative area --
and I don't know what that is -- it may not be the 2,500,
but it will be below that, 1,000 or whatever, but that's not
what's going to be driving.
I think the biggest uncertainty is really the
abrasive aspect of that jet, and you know, it doesn't have
to be 2,500.
I used the 2,500 because that's what they're using
to test these samples.
DR. BONACA: The reason why I'm making the point
on the 2,500 is that, when I look at some of the studies
being done, for example, by INEL, there is a significant
dependency between the K's they're assuming, like steam
break, and the delta pressure that is pertinent to that,
because that says, although steam line break is less
frequent than a stuck-open valve, the delta pressure is much
more severe, it's 2,500 psi, once you have the break on the
DR. HOPENFELD: But you see, they usually assume a
constant area, which is implicit in those assumptions. This
is not really a constant area here.
DR. BONACA: Okay.
DR. HOPENFELD: Obviously, you can't pull more
than the pump can pull in there, but this is not exactly the
same situation, it's somewhere in between. But your point
is well taken, 2,500 may be too high.
DR. BALLINGER: This also assume a dry steam
DR. HOPENFELD: Yes, it is a dry steam generator.
As soon as you depressurize, the procedures are
that the steam generator does stay dry. You turn off the
After this introduction, after the GL 95-05 was
put into the -- into effect -- and again, I'll remind you
again, originally it was meant to be only an interim basis,
we have erosion of that 1 volt or 2 volts and we're going to
3 volts and we're going above that, and again, the rationale
that is being provided -- and I already discussed that -- is
that we can prove and show you -- that's what the licensee
says -- that we don't have any damage using this MB2 data to
indicate to you the forces on the plate.
They may move a little bit, and if you go back to
your proprietary material, you see they moved with one code
and they don't move with another code, and you can see that
all of it hangs on a computer code that was benchmarked
against the wrong data.
That data was just not applicable, wasn't designed
for that purpose.
But what bothers me is that, when you look into
the SER, we don't even question that, we just accept it.
DR. BONACA: Wouldn't the movement of the plate
have an impact, also, if you stayed with the original
plugging criteria of 40 percent through-wall? Don't you
think it would be much less impact?
DR. HOPENFELD: You would think so, yes.
I think NRR has a very valid point with the
They said we don't think that 40 percent is really
ideal, we would like to do something else, we want to
tighten up our regulations, because that 40 percent came
from some wastage studies, it's not really applicable here,
so we want to improve that, and I think that the intent was
perfect, but as they were going along and the industry came
along and said, hey, we don't need any of that stuff, okay,
we want infinite flexibility to decide what we want to plug
and what we don't want to plug, and that's really what the
Yes, this is not ideal, but the 40 percent served
Nobody tested it, and I think what Bob was telling
you, he had some potential problems where there was no 40
percent, these were brand new tubes, and there was some
DR. BONACA: I guess 40 percent was to give you
some indication of the residual strength of the tube.
DR. HOPENFELD: Yeah, well, 40 percent, you know,
using the ASME code, you just write it off as a corrosion
allowance and that's it, you forget about it, whether it's -
- you don't go and start analyzing whether there are cracks
or not cracks.
Now, if you want to go in the second level of
details, you go into the crack propagation and so forth, but
that's why I think the 40 percent, from what we have, is the
best thing that we can do at the present time.
Now, if you remove the vibration, if you remove
all those loads and the potential of this erosion thing,
yeah, that's fine, then you can just go to whatever you want
to go to.
One thing to explain away the reason why we could
go to higher voltages is research is going to provide us
information on how to -- that would allow us to operate with
higher voltages, and there would be no problem with erosion
from jets and so forth.
I'll give you a few examples of the kind of work
that is now being proposed, that just a couple of weeks ago
was sent to NRR.
It's being proposed as the NRC solution to
operating degraded tubes, a potential solution.
One is ANL has developed -- and I'm just quoting
what one of the latest reports says -- developed a leakage
methodology, and that is the equation for the flow,
obviously it's a function of delta P and area, of flow area.
For some reason, I don't know why, it doesn't have
the L over D ratio, and I think they would be advised to go
and see what Dr. Shrock has done, because it's also an L
I've looked at that report. I think they're using
mostly, but not all, EDM notches. There may be some cracks,
DR. CATTON: It really depends on what you mean by
"A". The geometry could include L over D.
DR. HOPENFELD: No, it did not. It could, but
their evaluation was based on the ligament.
I don't remember the person's name, but his
calculations were based on the strength of this ligament
based on the internal pressure only. Only internal pressure
came in there.
So, if you are working in a laboratory and you are
willing to forget the real world, you have the luxury of
doing this, but it is not appropriate for licensing
Now, I've looked at some of the letters that were
going between NRC and Com Edison after the IP2 incident, and
they were using -- Com Edison were using these kind of
equations to predict what kind of leakage you're going to
have during an accident.
They don't even state the assumption that the
other forces that could act on that ligament, and we have
been through this before, could come into play.
Now, that is the basis for allowing a plant to
operate, or it's being used as a basis to allowing plants to
It's fine to do all that research at Argonne, I
have no problems with that. When you take this thing,
without putting it in the right context, and you start
regulating with it, that's when I have a problem with it.
Another issue that I'd like to discuss at some
length has to do with inlet plenum mixing.
Now, the issue of inlet plenum mixing comes in
during severe accidents. You remember now that we are on
risk-informed regulation, you have to look into severe
Well, if you are doing severe accidents, what you
have -- you have a situation where the driving force is
natural circulation between the reactor vessel and the steam
generator, and the flow goes up, partially mixed here, goes
up and then turns around and comes back.
Now, if you don't mix the flow here, you get --
during this severe accident, you get to creep rupture
problems, because the temperature is very high, and you
rupture the tube before you rupture another component in the
The component that most commonly is talked about
is the surge line.
So, here is the competition here between any one
of those tubes and a component in this part of the system.
If this component breaks first, then you're okay, because
this is within the containment, but if you -- if this one
breaks, you're out into the open. So, it's a competition
that we're talking about.
Well, the easiest way to solve the problem is the
What you do, you say, well, I can lower the
temperature, and I can keep the temperature here relatively
very low by mixing all that flow.
So, back -- remember, going back to the time line
-- that's the reason I put that time line there. Back to
the time line, remember, somewhere in '95, the NRR found out
that they are getting a potential problem here with creep
rupture in severe accidents.
Before that, severe accidents weren't that
important, because they were not part of the risk-informed
regulation, but now that you worry about it, you have to
come up with an explanation of why you're not getting -- why
you're not going to increase your risk of a core melt.
So, one way of solving -- to solve the problem was
to mix this.
So, if you mix these -- I don't remember what the
temperatures are, but there's a very large temperature
differential here, something like 500 or 600 degrees F, but
if you mix this thing perfectly, you lower the temperature,
and remember, from your creep rupture basic curve, the
rupture properties are -- you have more strength at the
So, the uncertainty here is not that much the
creep rupture properties, although they went and built a
very expensive facility at Argonne to find those properties,
but that's not really the major thing.
The major thing is to -- it's not the uncertainty
in the creep rupture property. The uncertainty is what's
going on in here.
DR. HIGGINS: It seems like everything you talked
about up till right now has been associated with the main
steam line break and a subsequent rupture of the tubes.
DR. HOPENFELD: Correct.
DR. HIGGINS: It seems like, in this one, now,
you've jumped to a different type of a scenario.
DR. HOPENFELD: Correct. I should really
DR. HIGGINS: Would you say a couple of words
DR. HOPENFELD: Yes.
DR. HIGGINS: I guess your concerns are broader
than just a main steam line break.
DR. HOPENFELD: Right. I'm sorry.
I titled the previous view-graph "Examples of
Research," how do we resolve -- how do we extend the 95-05,
and in the case of a steam line break or the design basis,
is those loads that I talked about, but now, under risk-
informed regulations, it's not enough just to say here is
95-05, because 95-05 by itself doesn't talk about severe
But now when you go and you ask for relaxation
under that risk-informed regulations, there is somewhere in
the system that tells you, aha, you've got to look into the
severe accident case, too, you have to show us that whatever
you're going to do to the system, whatever you're going to
perturb the thing beyond your present tech specs, whatever
relaxation you're asking, you are not going to affect the
severe accident case.
Does that answer your question?
DR. CATTON: And what it gets down to is that
there is a race to determine which piece of that system will
Now, depending on the assumptions you make, you
can make any part of it go first. If you assume that there
is no mixing in that lower chamber, hands down, it's steam
generator tubes first.
Depending on how much mixing you assume, you bring
the times closer together, and you can even make the surge
line or some small pipe that connects into the hot leg go
DR. HOPENFELD: Correct.
DR. CATTON: Where it really all comes down, as
near as I can tell, is the RELAP-5 code -- and I don't -- I
think you're faulting the -- you're kind of blaming NRC for
doing it deliberately.
I don't think it's deliberate; I just think it's
DR. HOPENFELD: No, I didn't get to the RELAP on
DR. CATTON: Well, you're going to get there. But
that's where the mixing comes from.
DR. HOPENFELD: Yeah, but I didn't get there --
DR. CATTON: The assumption of mixing comes by the
nodalization that's used with RELAP-5. From there on, it's
justification for having done so.
DR. HOPENFELD: Give me a minute.
DR. BONACA: I would like also to ask -- here the
question is -- the issue is steam generator tube rupture
induced by severe accident.
DR. HOPENFELD: Right.
DR. BONACA: Okay.
Now, you also, however -- I wasn't clear whether
you're making a contention that not only this cooling issue
is central to that, but also the damages in the steam
Now, it wasn't clear.
DR. HOPENFELD: Okay. I'll try to make it very
The issue of -- if it was four years ago, I
probably wouldn't even bring it up, or I'd just bring it up
as of just general interest, but under the risk-informed
regulation -- and I understand, at this time -- and I think
Farley was the first one where you have to address the
severe accident issue, and that's why I'm bringing it now.
I wouldn't have brought it out before.
At this point, when we give somebody -- we're
relaxing their technical specifications, we ask them to come
up with a justification that the severe accident is not
going to affect the core melt frequency, and that's the
reason I'm bringing that as another example, but I would
like to get into the technical reasons here, just take it
one step further.
DR. BALLINGER: Now, these are calculations,
DR. HOPENFELD: This is just a schematic.
DR. BALLINGER: No, I'm saying you're going to get
to the calculations.
DR. HOPENFELD: Yeah, I will, right now.
DR. BALLINGER: But my understanding is that there
is a discrepancy between the one-seventh Westinghouse test
and what's been observed at TMI. I don't see that.
DR. HOPENFELD: I'm not going to get there. Let
me tell you where I'll get, and then I'll be able to -- let
me get to my point, what I'm trying to say.
But did I answer your question, why I'm bringing
in that severe accident?
DR. HIGGINS: You did answer it. I guess I'm just
trying to get my arms around the scope of what's included
here, because there are a number of different ways that you
could address the steam generator issue.
One is the core damage-induced steam generator
Another one, the one that you've been talking
about, the one that's induced by a main steam line break.
One we haven't even talked about yet is the
spontaneous steam generator tube rupture at some frequency.
DR. HOPENFELD: Well, they're all really part the
whole picture, and I think this severe accident is part of -
- if you come in and you tell somebody I want to operate at
3 volts, under 95-05, their reply would be, okay, well, you
should look at severe accidents.
Now, industry has said no, we don't want to look
at severe accidents, but the NRC said, yeah, you look at
severe accidents, and in the case of Farley -- and maybe
that didn't ring a bell at the time, when they came in '99,
in September '99, and asked for relaxation in the case of
Farley, the staff did some calculations for them and said,
well, don't worry about this, we don't have any problem, and
that's what I'm going to tell you, why they do have a
So, it's not separate, and you've got to take all
of it together.
That's why I have so many pages.
DR. BONACA: Yeah, but until now, you have spoken
about steam line break --
DR. HOPENFELD: Correct.
DR. BONACA: -- which is in the design basis of
the plant, and our intention is that allowing this kind of
inspection and flagging makes it a different accident than
what is in the FSAR.
DR. HOPENFELD: That's correct.
DR. BONACA: Okay.
You could say that, within 50.59, we have created
a new type of accident, because it results in a leakage
which is much beyond what you'd assume, and in fact, if you
postulate what you're proposing, it's much beyond that, and
you get a combination of steam line break and steam
generator tube rupture, and all those issues come together
into a challenge of the actual design basis of the plant
DR. HOPENFELD: Correct.
DR. BONACA: Now, this is -- you're saying now,
separately from that, you have a concern --
DR. HOPENFELD: Well, it's not separately. It
still has to be addressed. It was separate three years ago,
but now it's not separate anymore.
DR. BONACA: Separately in a sense that you have -
- by other means, you come to core damage.
DR. HOPENFELD: Oh, yeah.
DR. BONACA: That cascades into potential steam
generator tube rupture if the tubes are not in the pristine
conditions and you have all those things.
DR. HOPENFELD: This is a station blackout type of
DR. BONACA: Exactly.
DR. HOPENFELD: It's a TML3 or TMLB or something.
I didn't get into that, but to give you an introduction of
the whole thing, yes, this is a different type of an
accident. This is not the steam line break.
DR. KRESS: There's two or three severe accident
sequences that can do this.
DR. HOPENFELD: An ATWS is even higher than that.
DR. KRESS: No, no, I think the station blackout
is the main one.
DR. HOPENFELD: This is the station blackout that
I'm talking about. I thought the ATWS has pretty high
DR. KRESS: It's pretty high, too. It's up there.
DR. HOPENFELD: I'm sorry. Sometimes I'm going
But let me say again, I'm talking about -- it's
not a steam line break, it's a station blackout, and in the
last two or three years --
Art, can you tell me when we're supposed to
address this thing, if somebody comes with a risk-informed?
We didn't have to do it in the past.
DR. KRESS: I think the ACRS almost forced them to
look at this. There wasn't any regulation that said you had
DR. HOPENFELD: In 1999 is the first time that I
saw -- when Farley came in here -- that it's being
addressed, and the industry was fighting that. They didn't
want to have them do that.
But since it's here, I think I ought to talk to
you about it or explain it to you, what it is, to see how
these things are being approached more than anything else.
DR. KRESS: The concern of ACRS wasn't so much
that this increases the CDF, because if you're into a severe
accident, you've already got a CDF. It was that this
converted it into an increase in the large early release,
because it could go into containment.
DR. CATTON: It has nothing to do with the CDF.
DR. KRESS: Well, a little bit. You can add a
little to the CDF if this happens.
DR. CATTON: You've already had it.
DR. KRESS: You've already had it, yeah.
DR. CATTON: It's on the table.
DR. HOPENFELD: There is a difference, and I think
the difference is that this is called an LERF, not that you
have a containment bypass.
DR. KRESS: That's right.
DR. HOPENFELD: So, you're talking about another
order of magnitude of safety.
So, if you could go and live up to 10 to the minus
4, now you stop at 10 to the minus 5.
DR. KRESS: That's the distinction.
DR. BONACA: That's the distinction, and we have
been confident -- I mean there has been some confidence, I
believe, from 1150, that because of failures of the primary
side, you will not have this bypass of containment in many
sequences of this type, and now, this could create a much
bigger group of sequences that will bypass containment.
DR. CATTON: Actually, this all started when
somebody in Holland got ahold of RELAP-5 and did some
calculations that were absolutely incorrect, but they
concluded that -- actually, they argued that it was the
nozzle on the reactor vessel that would go, and then people
started to look at the problem, and over a period of time,
it involved into this particular configuration, and in the
Westinghouse tests, one-seventh-scale tests were done, but
nobody did flow visualization.
The temperature measurements were pretty good, but
the scaling was improper.
So, as a result, all you know is that that kind of
phenomenon can occur.
DR. POWERS: To be precise, the Westinghouse
experiments did not include the steam generator.
DR. KRESS: They had a simulated steam generator.
DR. CATTON: They had it simulated.
DR. KRESS: It wasn't a steam generator.
DR. POWERS: It does not look like that at all.
DR. KRESS: No, it doesn't look like that at all.
It wasn't scaled very well.
DR. HOPENFELD: Okay.
So, going back to '96, I believe, or '95, when the
NRR felt that they'll -- they got an inkling that they'll
have to address the severe accident issue, they asked
research to look at it, and research solved the thing in a
report called NUREG-1570 that, again, it's being used for
licensing, and basically, the answer was that, if you have
good mixing, there is still a chance that you will rupture a
few tubes, but the probability was low.
So, at that point, I thought it would be useful to
take a look at that mixing assumption.
I had a report, remember, going back to September
1992, when I assumed that there was no mixing.
There was no reason to believe that there would be
So, I assumed that there was no mixing.
Then EPRI had a report on this subject, a very,
very elaborate report, and they assumed that you could have
mixing between 100 and 200 gpm, and there was a very, very
clear effect, and when you have mixing, they assumed that
they had a leakage, primary to secondary leakage of 100 to
200 gpm. It will have a profound effect on the mixing in
the steam generator.
If you then -- what happens is, when you have
mixing in the steam generator and if you have a large flow
due to leakage of the tube, all that leakage will bypass the
plenum, basically, and you're going to get high-temperature
gas or steam in contact with the tubes.
Well, all that was all forgotten, because the
research studies were based on Westinghouse one-seventh-
scale model, which, besides the scaling problems that have
been discussed for a long time, it didn't have any leakage.
So, all that data that RELAP-5 was benchmarked
just wasn't applicable.
Another implication of this is going back to here,
that we have never really looked into that, and it should
be, and that is that these very high -- when the flow rate
here -- the natural circulation flow is very slow. It's
like a couple of feet per second.
When you have large leakage flow, then you really
have some kind of a combined natural forced convection flow
in that pipe, and that by itself is going to affect the
rupture of these components on the primary side.
So, the point here is that all that analysis came
to a criticism, and I think Dr. Catton was involved in this,
and there were a lot of questions, but nothing happens, and
two weeks ago, we get another letter to NRR telling that
we're going to do more of the same.
One of the criticisms that came up during one of
the ACRS meetings -- there were a lot of deposits in here
which were not taken care of.
Actually, what you are really interested is
knowing what the tube-to-tube temperature variations, and
those were not calculated.
So, now, they want to continue this kind of study
to come up with an improved temperature distribution, which
is fine, but the main problem here is that we don't have
data. There is just no data.
It's a three-dimensional kind of thing. RELAP is
There is no data to justify any of that.
DR. BALLINGER: Can I ask a question?
Can you go back to the previous slide, the one
that showed the schematic?
DR. HOPENFELD: This?
DR. BALLINGER: Right.
Now, I guess I understand the argument, but if you
have a single or even two tubes failed, isn't that going to
short-circuit the flow?
DR. HOPENFELD: Sure.
DR. BALLINGER: So, how do you get high
temperatures in the other tubes?
DR. HOPENFELD: Well, if you already have a
failure, a large failure, then, you know, you have the
leakage. It depends on the relative amount of steam you
have. There's a lot of steam there.
DR. BALLINGER: But if you've got flow already out
a leaky tube, that short-circuits the high-temperature flow
through a tube which has already failed, and so, you don't
have to worry about a creep rupture problem. How does it
get the temperature that would result in a creep rupture to
the other tubes?
DR. CATTON: The tube that's broken -- it's going
right out the SRV and into the atmosphere. That's the
DR. BALLINGER: I thought you were arguing that
you get high temperatures in other tubes, therefore you get
rupture of the other tubes, and therefore, you propagate the
DR. HOPENFELD: No, that's already bypassing.
There would be no mixing in here. Whatever temperature
comes in, whatever steam comes in here at the higher flow
rate, it will get out.
DR. BALLINGER: So, you don't propagate the
failure by this mechanism.
DR. HOPENFELD: No, I didn't say it would
propagate the failure.
DR. BONACA: But you said that the tubes fail
first, which I understood the same way, that the tubes would
be exposed to higher temperature.
DR. HOPENFELD: Higher temperature than the surge
DR. BONACA: Why?
DR. BALLINGER: But the tube has already failed
and it's already bypassed.
DR. KRESS: So, what's the consequence?
DR. BONACA: You have to have a mechanism by which
you fail the surge line.
My understanding of your contention was that the
tubes now -- there is some leakage coming through. That
will cause the tube to hit higher temperature than the surge
line, and that will cascade into more rupture.
DR. HOPENFELD: But it's not only the temperature,
it's the size of the thing. The component is of a different
You have to look into the actual component
calculation, the stress calculation, and you'll see it.
I don't have a graph to show you where the cross-
over point is.
DR. BALLINGER: I'm trying to get the scenario
DR. HOPENFELD: Okay.
DR. BALLINGER: You're saying that you have
multiple damage to steam generator tubes, which are leaking
at some --
DR. HOPENFELD: Correct.
DR. BALLINGER: -- small rate.
DR. HOPENFELD: But sufficiently large to affect
the -- I'm not going to have mixing.
DR. BALLINGER: But these tubes would not
necessarily burst during the accident. But when you get
this small amount of leakage, you alter the natural
DR. HOPENFELD: I didn't say small.
It's sufficiently larger than the natural
circulation, because otherwise, natural circulation would
DR. BALLINGER: Well, I'll give you that. But now
the hot gas goes up these tubes which are leaking a little
DR. HOPENFELD: Which are leaking. I don't know
how much they leak.
DR. BALLINGER: Well, cause a higher temperature
in the tube, result in rupture of the tube.
DR. HOPENFELD: Higher temperature relative to the
DR. BALLINGER: I mean higher temperature with
respect to the stress rupture.
DR. HOPENFELD: But you see, you have to go to the
stress calculation -- to the structure calculation of
whatever component -- say, the surge line -- versus the
It's not only the temperature, and if you go and
do that -- which I didn't bring the data with me, but you go
there and look at it, you will see that, if you lower the
temperature, okay, if you lower the temperature of the tubes
or if you allow for mixing here, the surge line will break
DR. BALLINGER: I'm not worried about the surge
I'm trying to reduce this to terms that a
metallurgist can understand.
DR. HOPENFELD: It's not a metallurgy problem.
DR. CATTON: Maybe I can help.
DR. BALLINGER: I'm trying to envision a way to
propagate this so that you get larger release.
DR. BONACA: I had the same understanding as Dr.
I mean my understanding was, if you have this
effect, okay, of circulation, it will provide cooling to the
tubes to the point that the surge line heats up first and
DR. HOPENFELD: It's not necessarily the heating.
It's a combination of the structure and the temperature.
DR. BONACA: Conversely, if you have some leakage
to the tubes, that leakage such that dominates that
recirculation portion of the steam, that cooling is not
happening anymore, and this will result in further increase
of temperature to further failure of the tubes.
DR. CATTON: I don't think a change in flow to the
tube because of a leak impacts the heating rate of the surge
line much at all.
DR. BONACA: I'm talking about the heating rate of
DR. CATTON: If you look at the Westinghouse one-
seventh-scale data carefully -- and they have some
appendices with a whole bunch of temperatures in them -- and
none of their tubes leaked -- what you'll find is that, in
some of the tubes, the temperature at the inlet is almost
the same as the temperature coming out of the model hot leg.
What that says is that it's a rather complex
process that's going on in that chamber, and making the
assumption of .87 mixing really is without basis.
DR. BONACA: It seems to me that one should give
some belief to both possibilities, but that's just a
DR. HOPENFELD: Really, the only way to answer
your question -- if you go back and see the surge line
temperature going up and you see where they cross over, that
temperature makes a difference, but my point is here that
you cannot ignore, because I calculated it, the Japanese
calculated the same thing.
They came up with the conclusion that this is very
marginal if you allow -- if you don't allow mixing.
EPRI calculated it the same way. They had a
model. They had, well, you can't use this Westinghouse
data. So, they had a model which wouldn't allow mixing.
So, there were three models here, okay, all of
them showing there is no mixing.
Now, we have the NRC people going and developing
calculations which are based on perfect mixing without
analyzing -- without really looking for the entire picture,
looking as to what happens to the surge line, how does that
affect it, without coupling the whole issue, and then you
use these results in 1540 to regulate plants.
That's the thing.
It's not all this. I don't mind if you do this
thing until doomsday playing with these models. That's
fine. It's good to present papers.
But when you start using this into the regulatory
arena and you start really licensing plants, you tell them,
well, you can have this inspection, you can't have this
inspection, that's where the concern is.
MR. LONG: This is Steve Long with NRR.
I don't think there's much disagreement here
between the staff and the DPO author on the effects of
leakage, or at least our inability to handle them.
The concern is that you're trying to determine if
the surge line will heat enough to fail first or the tubes
will heat enough to fail first, and there's a lot of
discussion about whether or not the scaling for the one-
seventh-scale test to the prototype, various different shape
prototypes between CE and Westinghouse, really captures the
phenomena correctly about leakage.
When you add the leakage, a whole bunch of
different things happen.
First of all, if you're leaking substantially from
some tubes, the flow doesn't have to come back from the
outlet plenum side to the inlet plenum to let hot fluid come
into the inlet plenum.
So, you really cut down on the mixing that way.
So, you may very well get hot gases going up to a
lot of the tubes.
Then, in addition, if the leakage is high enough,
you'll actually cut down on the cycling or stop the cycling
of the PRV on the top of the pressurizer, so you cut down
the flow through the surge line and you slow the heat-up of
the surge line, at least to the extent that the surge line
doesn't sit at the top of the hot leg and just get the hot
temperature as it goes by it, it's off to the side.
So, there are a bunch of different things that we
don't handle well if you start adding substantial leakage.
One of our concerns has been to try to keep the leakage down
to the approximately 1 gpm that's in the design basis now,
and if we felt we had to make it lower, then we'd have to
come up with enough analysis for the backfit.
DR. KRESS: What is a good rule of thumb for what
you would call substantial leakage?
MR. LONG: We've done some studies that assume a
fixed-size hole in the steam generator tube, and if you size
that hole so that you get approximately 100 gpm leakage
under the design basis conditions, where there's water on
the primary -- and I don't remember what the hole size is --
we can look it up for you -- that hole would stop the
cycling of the pressurizer valves before you got to failure
of the RCS by creep.
I was trying to size the hole so that you could
relate it to the design basis-type limits that we had in
leakage of water.
So, I can't tell you -- that's approximately the
size that seems to -- in the Surry plant model right now.
That would make the effect of preventing the
safety valves on the pressurizer from cycling until the
point of failure.
It would alter the flow path before that through
the surge line.
We don't know how to handle the leakage effects on
So, it may be well below that that the effects on
the mixing occur.
DR. BONACA: Before you move further, Dr.
Hopenfeld, I would like to ask you -- you presented in a
previous slide your scenario -- you presented the
containment bypass frequency of 1.6 times 10 to the minus 5.
How did you get to that number?
DR. HOPENFELD: This has been a long time. I'll
have to recollect how I got the number, but I'll give you
the rationale. I don't remember.
This slide came from a presentation I made to the
ACRS in '98, I think, and I had that number, and I think Dr.
Buslick helped me with that, and maybe he will remember, but
I got those numbers from -- there was a rationale for
getting those, but I just don't recall exactly where it came
DR. BUSLICK: I don't really remember for sure,
but 1.6 times 10 to the minus 5 per year, I think, is the
total station blackout.
DR. HOPENFELD: Yeah, I think that's the answer.
DR. BONACA: Then what you did you assumed the
station blackout and then assumed no conditional
probability. You have a station blackout and that will take
you to a containment bypass. Okay.
DR. HOPENFELD: I am not a PRA man, but I went to
Dr. Buslick and he gave me a number.
DR. BONACA: That's what I saw yesterday from some
papers, but I wanted to confirm it.
DR. HOPENFELD: Thanks a lot, Steve. You couldn't
state it better.
So, we do have an agreement here now.
The next thing -- what kind of got me a little
MR. LYON: Let me raise one more point that Steve
was sort of getting to.
Remember, we're starting with a core damage
So, the fluid back in the reactor vessel is really
up there, you've got the chemical reactions going on, and so
forth, and then, as that fluid flows along the top of the
hot leg, it is interchanging energy with the cooler fluid
flowing back, so we're getting a cooling effect there.
Then, as you get into the steam generator inlet
plenum, you get into the mixing there, both phenomena, by
the way, quite uncertain, from what I have seen, but if you
get into a situation where you have a substantial leak, say
one tube partially breaks, and you set up a mechanism to
take that really hot fluid, say 3,000 Fahrenheit, whatever,
that's back in the vessel, and move that up into the area of
one tube, and if that is then moving out and hitting other
tubes, you may have a propagation mechanism for making the
leak substantially greater and failing a number of tubes.
DR. HOPENFELD: My point really here was that
there is a proposal here to do additional study of this,
doing more analytical study and code calculation. I really
don't think you can do that.
You have to get some data with leakage to
benchmark these codes, and I have nothing against that, but
just to do more of the same that was done before I don't
think is very useful.
DR. BONACA: The last comment I would like to make
about this is, when I compare these containment bypass
frequencies, there is a full agreement on the frequency of
station blackout, and then I believe the DPO takes a
position that there is certainty that, if you have a station
blackout, you have a bypass situation, so conditional
probability is 1, and the other position is surge line fails
first, so there is no bypass, and you know, I wonder if
there was an estimation of somewhere in between, given that
there is significant uncertainty on the phenomenology of
this. We can explore that tomorrow.
MR. BUSLICK: Steve Long corrected me. It's not
the entire station blackout core damage but the high dry
station blackout core damage frequency.
DR. POWERS: Dr. Hopenfeld, are we arrived at a
point that it would be appropriate to take a break?
DR. HOPENFELD: Fine.
DR. POWERS: Why don't we take a break till
quarter after the hour.
CHAIRMAN POWERS: Let's come back into session.
DR. HOPENFELD: Remember, going back to the time
line, the entire NRC justification for operating with
cracked tubes is based on NUREG 1477.
And the assumptions in there are that the primary
and secondary leakage rate is between 480 to 540, and, of
course, this is primarily for the risk assessment. It's not
for the CFR-500.
The crack opening is .576 to .72, and the crack
area does not change once the corrosion products are forced
out. Now, you can see that this is a constant area, a
certain area that was assumed, and a certain flow rate was
assumed, which really neglects all the items that we're
talking about, all the factors that we're talking about
before, the jet and the forces due to the steam line break.
So, when you make these assumptions, sure, the
pump, if you write the basic equations for a pump operation,
obviously there's a certain amount, maximum amount of flow
that you can force through a constant area.
Makeup of water was added to the RWST, and the
main assumption -- and that's the one that we're going to
analyze and look at a little bit more -- is this ten to the
minus three, and that's the one really that bothers me more
than anything else.
Because where does it come from, and what's the
justification for it?
DR. BONACA: This, if I understand it, is probably
the where the steam line break is ten to the minus four, and
operator failure is ten to the minus three?
DR. HOPENFELD: The probability of -- no, on this
one, I believe the probability -- it was not the steam line
break; it was the safety relief valve, and I think that was
ten to the minus three, if I remember correctly. Is that
MR. LONG: Correct.
DR. BONACA: And how do you get ten to the minus
MR. LONG: Let's talk about it tomorrow. I have
to get the book.
DR. BONACA: I'm asking --
DR. HOPENFELD: Well, ten to the minus four times
ten to the minus -- let's see, ten to the minus three times
ten to the minus four is ten to the minus -- this is ten to
the minus six. Where does seven come from? I don't know.
DR. HOPENFELD: I got it from NUREG 1477,and
probably you have the numbers.
DR. BONACA: Yes, there is -- I have reviewed
those documents, too. There is a full range of spectrums,
depending on the transient, and that's why I'm trying to
nail down which accident we're talking about.
DR. HOPENFELD: We're talking about a steam line
break. This is in NUREG 1477.
DR. BONACA: That's a steam line break?
DR. HOPENFELD: It's a steam line break.
DR. BONACA: You said that it's a stuck-open SRV?
DR. HOPENFELD: Well, no. A stuck-open SRV is a
steam line break.
DR. BONACA: Well, the way it's characterized is
DR. HOPENFELD: Right, the frequencies are
different, but originally, actually when I looked at that
thing, remember, I had two months to look into that problem.
I talked to various people, and I came up with the
number of ten to the minus four, and that's why that I stuck
the ten to the minus four in there, and gave the operator
zero credit for it, that he didn't do anything.
Then when the committee was formed and they did
some more studies, they came up with a higher frequency, and
they were talking about the relief valve.
So another way of looking at it, if you want to go
to the ten to the minus three, then you say, well, ten to
the minus three, and we'll give some credit to the operator
that maybe he'll look at it. But anyway, the number is ten
to the minus four, as far as I can see in how you come to
You may come to it from different angles.
MR. HIGGINS: Do you postulate the same drastic
effects in the steam generators from a stuck-open relief
valve on the secondary side as the main steam line break?
DR. HOPENFELD: Well, this was done by NRC. These
are their numbers.
But I believe that, yes, I think you could --
within the uncertainty that you have, there is a limiter
there, but within the uncertainty that you have, it probably
doesn't make that much difference.
DR. BONACA: Yes, but some of the reports,
however, show, depending on the initiators and how
challenging it is, they assign different frequencies for the
initiator, different success criteria, and other things that
come after that.
DR. HOPENFELD: Right.
DR. BONACA: And so --
DR. HOPENFELD: I'm just showing you what 1477
used, and what has been used as a justification for the last
eight years as quoted in the reply to that DPO document.
That's what's being used, and that's what I'm addressing.
I'm just telling you what they are talking about.
DR. BONACA: Okay.
DR. HOPENFELD: That's the number that is in that
NUREG. Now, tomorrow, hopefully you'll ask them where they
got this surface area from, where they got this flow rate,
and they should justify that thing.
And why is the surface area constant, if you have
other mechanisms, loads, and that's really the crux of the
Sure, if you have a constant area, you're limited
by the pump, but that's not real life.
DR. BONACA: Sure.
DR. HOPENFELD: So as I said before, the DPO
approach was, this is too complicated, whatever the
frequency you have, that's it, and the probability that you
would lose the inventory is one, once you get to that point,
if you have cracked tubes.
That was the approach from the beginning. Now,
whether it's ten to the minus three and you give operator
credit or it's ten to the minus four, it's not really the
main point here.
The main point here is that you are ten to the
minus four, which is two orders of magnitude, which is an
order of magnitude higher than what the ten to the minus
five that we were supposed to abide by.
If the Commission tomorrow says, well, ten to the
minus five is not a good number; let's go to the ten to the
minus four, I'll just retire and just forget about what I
But that's what they said, and they set the rule,
and if they set the rule, they would stick to it, otherwise,
this whole risk-based-informed is just one big joke.
And that's really the point. So, the whole thing
is, if you've got some -- I'm sorry, was there a question
that somebody raised his hand for?
So the whole thing could be really explained away
if you say if you have a super-duper operator and he can do
marvels and he can put it down, but it's not a simple thing,
when you have a large leakage, to bring that kind of system
to an ordinary shutdown, because there are conflicting
You have on the one side, you have steam coming
out from an opening in the steam generator, and it goes out
to the site, and the only way you can stop that is to reduce
the pressure on the primary side.
So when the pressures become the same, the leakage
stops. But on the other hand, you can't go too fast,
because if you go too fast, there is the possibility that
you uncover the core, plus, you have limitations of PTS,
pressurized thermal shocks, but that's not the main point
The main point is that you can't go, you're
limited, this is not a simple operation. Now, maybe if
you're running at 100, 200, 300 gpm -- I don't know, because
I'm not an operator -- you probably could handle it.
When you get to larger, some theoretical
predictions can go above 5,000 gpm. But the point is that
it's not a straightforward kind of thing, because some
plants don't have pumps that you can throttle, so you have
to turn off pumps on and off, and some of them were not
designed for that purpose.
You may be overheating them, so if you lose pumps
while you're operating, then the operator has got another
So it's not straightforward, and I'm not an
operator, so Mr. Spence will talk about this a little more.
But the main point here that I'd like to bring to you, is
that if you go back and operating experience, then in
reality, even in IP-2, relatively -- compared to this,
relatively trivial accidents have caused operator problems.
The one that I -- that was brought up to me, to my
attention recently, was the one at, I think, Palo Verde.
They took 28 minutes before there was a recognition there
was a tube rupture.
Now, this is relatively a no-accident; this was --
the plant was designed, compared to what I'm talking about.
So if it takes you 28 minutes, then you can say,
well, this much more severe accident is going to take --
he's going to follow and do all these -- sure, he can do all
of that, if the equipment operates that way.
But I've driven a lot of cars over the years --
it's my hobby -- and things just don't happen that way with
real-life cars. Reactors are different, but nevertheless,
this number here that is being used is ten to the minus
three for operator error, is, I think, very, very
DR. CATTON: Can I ask a question about -- you've
assigned a number of ten to the minus one for operator
DR. BALLINGER: Well, let me go back to that.
What I did originally, I said ten to the minus four, because
that was a frequency given to me for steam line breaks
upstream of the isolation valve.
See, you have an isolation valve and you can
isolate that thing, but there's a section there and it
varies from plant to plant what it is, that independent of
what the isolation valve does, you have many steam line
breaks with a bypass.
DR. BALLINGER: Okay, so this is not operator,
this is valve operator. What's the ten to the minus one,
operator, the guy?
DR. HOPENFELD: Okay, that's an operator.
DR. BALLINGER: Now, can that be affected by
operator training? What if they trained on these kinds of
DR. HOPENFELD: Bob will talk about that. I think
it's a very good question. I've asked it a couple of weeks
ago from -- I asked NRR to provide me statistics of the
operator simulator results on that kind of accident, and
they don't have it. But that is one place to get that
DR. BONACA: Because it's a steam line break,
rather than a tube rupture, so really if you look at the
procedures, they way they were set, it would involve
different procedures, probably.
DR. HOPENFELD: Well, it's a LOCA, basically.
That's all it is.
DR. BONACA: But you don't start with the LOCA;
you start with the rapid depressurization, and you think --
DR. CATTON: To pick it up.
DR. BONACA: To pick it up.
DR. CATTON: But i believe the operator is trained
on a simulator. The simulator is based on RELAP, and we
heard this morning about what you actually will see, and
they're quite different.
DR. HOPENFELD: That's the bottom line here.
DR. BONACA: They use RELAP anyway.
DR. CATTON: Well, whatever codes are used, RELAP
is the one.
DR. BONACA: When you say RELAP, I agree with you,
but some say that insofar as the Palo Verde event, you know,
if you have a straight steam generator tube rupture, and it
is a minute leak, it may make it hard for the operators at
the beginning to --
DR. HOPENFELD: I don't think this was a minute
one, though. I think it was a full rupture.
DR. SIEBER: One tends to mask the other.
DR. HOPENFELD: Well, I think that particular one
DR. BALLINGER: It was 250 gallons a minute, I
DR. HOPENFELD: Is that what it was? Okay, that's
about half; 250 is about half -- 500 for full rupture, and
250 is about half.
DR. BALLINGER: Two-forty.
CHAIRMAN POWERS: I'd like you to be a little more
accurate in your estimates.
DR. HOPENFELD: Well, 240, it sure is higher than
a one-gpm, 500 is, you have to agree with that.
CHAIRMAN POWERS: Yes.
DR. HOPENFELD: We looked at the Indian Point 2
experience and I think there was some problem at controlling
the steam flow to the condenser, and there was a slow
cooldown rate, and there were some other problems that I'm
ont familiar with. Hopefully Bob will talk about that.
Theoretical predictions go to something on the
order of 3,000 to 6,000 gpm, which is the limit that even
their theoretician claims that there is no way of
controlling the accident. It's probably anywhere in
Now, to summarize the severe accident, because it
does fall into all that stuff that I have been talking
about, already Mr. Bosnick mentioned and this comes from the
station blackout scenario.
And the data that was obtained to justify that
this number would -- that this is going to be the number, is
based on the study of Westinghouse with a 1:7 scale model
which did not include the main -- the mixing in the plenum,
and the phenomenon is a three-dimensional phenomenon, not a
one-dimensional phenomenon that's being treated by RELAP.
DR. BALLINGER: Again, I'd like to put things in
perspective, though. Six thousand gallons a minute is about
DR. HOPENFELD: I think so, yes.
DR. BALLINGER: It's about ten tubes, and in the
Indian Point experience, they did get the plant shut down
without any damage.
DR. HOPENFELD: That's 150 gpm. But that's a very
good point. At 150 gpm, the thing is that it's still -- my
point is here is about the operator response, okay?
This was a very mild accident and still the
operator -- what I'm really trying to say is that there is
room here for the operator to make errors, and when I see
ten to the minus three, it's kind of hard to believe.
DR. BONACA: Actually, you know, it's interesting
that in some of the reports like this, the report that we
have, they are analyzing up to 15 tube ruptures, and they
present an interesting perspective in this range in the
middle. It seems to be the least challenge to the operator
because it depressurizes so fast that it brings you down to
no pressure for entry level, so even if you are confused for
a long time about where you are, but you stay low and
leaking low. And the more challenges seem to be the fuel
tube ruptures, because you're staying, you intend to come
back to pressurize, or the very high leakage rate beyond 15
tubes where you cannot make it up. You cannot make it up,
so it's an interesting perspective on that report.
DR. HOPENFELD: Okay, I'd just like to bring to
your attention here -- I'll just summarize it, because I
don't want to harp on this severe accident too much -- that
the Japanese JAERI came up also with a prediction that the
creep rupture, that the tubes would rupture much earlier
than NRC predicts, because NRC said, well, their computer
codes predict that it doesn't, but they didn't say that they
are mixing it, and therefore that's what the difference is.
And, again, one of the discussions in the document
you have that's called a reply to -- I mean, the DPO
consideration document -- talks about that these cracks are
going to be constrained within that support plate.
And the jet coming out is not going to go
anywhere, it's going to be deflected by that support plate.
Now, remember, you have 2500 pounds on the inside
of that tube, and it's kind of very difficult to see how
that support plate is going to do anything, especially going
back, that it's going to move the tube sheet. Remember, it
was only designed for 1500 and not for 2500.
So, to say that this thing is confined within that
support plate and that it is going to prevent the jet from
damaging adjacent tubes, is not very realistic.
At this point, since I didn't talk to much about
the operator action, it's probably the most important thing
in this whole presentation, I asked Robert Spence to talk
about it a little bit.
DR. BONACA: You're going to talk about deflection
DR. HOPENFELD: That's what they're saying. I
don't want to get into that. I mean, you can come up with
200 different scenarios.
DR. BONACA: All right.
MR. SPENCE: For reference to what I'm going to
talk about, you were given a handout this morning, a table
of three or four pages, about operator.
Steam generator tube rupture, operator performance
and NUREG 6365. Now, I put that together based on NUREG
6365, combining basically looking at it from what an
operator did, what worked for him, what didn't work for him,
what problems he had with equipment performance, also a
comparison of radiation releases as well as what kind of
isotopes were released.
These were only steam generator tube ruptures,
basically without main steam line break, et cetera.
I go back to 1975 and all the way up to Indian
Point 2. Where's that pointer? Can I use it?
Okay, again, basically what I'm going to talk
about is that ten to the minus three an appropriate
estimated probability of operator actions?
These numbers are not -- you've got three
different scenarios that will cause the design basis
problem: Main steam line rupture, stuck-open relief valve,
and feedwater line break.
The interesting part about this slide is the human
error contribution to the event. It's very high, and is
probably going to be some of the highest in any accident
MR. HIGGINS: Could you clarify a couple of things
on that? That is, I assume, 1.0 E to the minus three at the
MR. SPENCE: Yes, that's supposed to be ten to the
minus three, yes. The zero doesn't belong there, sorry
MR. HIGGINS: And the seven on it?
MR. SPENCE: Is -- well, refer -- in your
proprietary document, there will be a reference to where
that came from. All these little footnotes are references
DR. KRESS: And your point about the human error
contribution to the CDF per year is that it's high like 93
percent, then that value you get is almost directly
proportional to what you assume for this human error
MR. SPENCE: Yes, sir. And I think these numbers,
if I'm not mistaken, and somebody can check me on what
reference I used, but I think it's 1477, NUREG 1477 numbers.
[Pause to adjust microphones.]
MR. SPENCE: How's that?
MR. SPENCE: Okay. All right, I'm sorry, did
anyone else have other questions?
MR. SPENCE: Okay, now, I've tried to get the
latest -- I've been trying to get the latest Westinghouse
emergency response guidelines from the owners group since
May of this year, and have been unable to do so.
That request has been refused twice that I know
of. So, some of -- so what I tried to do is put together my
own concept of what's important in this faulted or ruptured
The first thing he's got to do is maintain -- get
the reactor subcritical and maintain it that way with some
type of boron addition.
This is the diagnostic step that is the unusual
feature, the newest -- the latest symptom-based procedures,
he really doesn't have to diagnose it.
But this is the unusual feature: He does have to
diagnose it, and it's very difficult for him to determine
the primary system flow rate, when everything is in such
One recommendation might be to try to come up with
in the SPDS system, some type of calculation that might be
able to tell him some kind of rough number of what -- how
much leakage he has.
Now, then this is where your ten to the minus
three comes in to depressurize, cool down the reactor
coolant system. He's got to worry about maintaining
adequate sub-cooled margin, and yet he has to decrease the
reactor coolant system pressure, so he's kind if working
inversely proportional to what he's trained to do.
Your safety injection system is going to come on,
and kick the pressure up, and that's also working against
him because what he really wants to do is take the pressure
It's going to take about two hours to get down at
the best down to RHR cooling. But by the same token, he
doesn't want to cool down too fast, because he's got 100
degrees per hour cooldown rate max, and he's worried about
The hour-long steam line break at Robinson 2, a
cooldown, I think, 213 degrees in one hour. At Turkey
Point, the cooldown rate that I saw was 60 degrees F within
So, he's got all kinds of transients going on that
he's trying to respond to, and he's trying to get down as
soon as he can to RHR to stop the release of radiation to
So those are -- oh, the other thing, the other
important thing that is in the DPO is that he has to refill
the refueling water storage tank. If you look at the
hierarchy of goals, this is very low in what he's trying to
He's got his hands full. So, I went back and just
for the sake of argument, I just took what was in NUREG
6365, and said, okay, how did he meet those goals? What
happened in -- there were ten events there, and I included
IP2, so we've got 11 events, and I think we really didn't
have good data on a foreign event, so let's call it nine out
of ten events.
There was a delay in tripping the reactor. What's
interest, at both Turkey Point and Robinson 2, when the
event occurred -- see, the operator doesn't know what
happened. He doesn't know if he's got a relief valve stuck
open, until he sees a trend, or main steam line, and until
he sees a trend in the pressure -- in the steam generator
level going down.
DR. BALLINGER: Wouldn't you see tail pipe
temperature on the relief valve?
MR. SPENCE: On the main steam relief valve?
DR. BALLINGER: Isn't there a --
MR. SPENCE: I don't know of any. Does anybody
else know of any?
DR. SIEBER: You can hear them.
MR. SPENCE: You're right, but you don't know
whether it's a relief valve and it's going to --
DR. SIEBER: Or a break.
MR. SPENCE: Or a break, that's right.
DR. SIEBER: Relief valves are usually quieter
than a big break?
MR. SPENCE: Yes.
And what the operators are going to do, naturally
-- what are they used to? They're used to working CVCS
pumps, charging pumps. So they're going to go over there
and if they've got one shut down, they're going to start it
In fact, they may even, if they've got boration
going in, they may stop boration, which is what happened at
-- I can't remember which one of the two events, Robinson or
Turkey Point -- which is exactly what you don't want to do
at that point.
Because now you've got to worry about cold water
addition, and re-criticality, and what you need to do is
pump in that nice boron in your refueling motor storage tank
into the core.
Okay, this was an old thing where you could either
use a steam generator tube rupture procedure or a steam
generator leak procedure. Again, I don't know if that's
going to be applicable in today's world or not.
There have been a number of delays in either
keeping feedwater going into the steam generator, which is
just going to exacerbate the continuing oscillations of the
steam break, giving it more fluid, and making the common
mode residence frequencies last longer.
DR. SIEBER: I think that the more difficult
problem for an operator is if he ends up with a main steam
line break somewhere, or a stuck-open safety valve, and then
the steam generator tube rupture occurs because a lot of the
parameters will track one another.
He may make an assumption that he knows what it is
he's got, and start off down that track without picking up
for minutes, perhaps, the fact that he's got two problems
running at the same time. I think that's tough for an
MR. SPENCE: Yes, and he's really relying upon his
radiation monitor. And if he doesn't have that radiation
monitor available --
DR. SIEBER: You're talking about N-16 monitors?
MR. SPENCE: Yes.
DR. SIEBER: All plants have them.
MR. SPENCE: Yes.
DR. SIEBER: And some of them are local readout.
MR. SPENCE: Right.
DR. SIEBER: So both of those are a little bit of
MR. SPENCE: And if you don't have it, he's going
to misdiagnose it, not always, but it has contributed in the
past, the loss of that.
DR. SIEBER: You'll see that everywhere, probably,
though, because you'll pick it up on other area radiation
monitors, the fact that you've got more activity.
MR. SPENCE: That's right, but whether or not and
what the operator attributes that to, I'll talk about some
simulator testing a little bit later that occurred over in
Norway, in which I've got the videotape of.
And the operators talk about how well the
condenser radiation monitor, gas radiation monitor off the
steam generator, et cetera -- well, there's no flow going
through that line, so that's why it's alarming. They could
rationalize it out.
DR. SIEBER: Well, the other problem is that you
may not get a radiation signal, because when you get a trip
like that, well, the safety valve opening is all being
MR. SPENCE: Yes, and there's no radiation
DR. SIEBER: That's outside, and you don't have
anything there to pick it up because it's not a monitor
MR. SPENCE: Right.
DR. BONACA: I hear you talking about current
response using EPGs. I mean, symptom-oriented procedures,
MR. SPENCE: These here are what happened.
DR. BONACA: This is before at Westinghouse.
MR. SPENCE: Some of it is applicable; some of it
may not be because of the change to the symptom-oriented
DR. BONACA: Okay.
MR. SPENCE: And I think one case there was --
they were releasing radiation that they didn't have to,
because the swap over with the lines going back inside the
containment didn't swap.
And I think that was a radiation monitor thing,
MR. SIEBER: That's on the air ejecter?
MR. SPENCE: Yes.
MR. SIEBER: Okay.
MR. SPENCE: The other thing I alluded to before,
when I was talking about the noise-- it affects
communication and operator performance. This was, you know,
I mentioned it at Turkey Point, but it was also mentioned in
the report from Robinson, too.
Okay. The pressurizing cool-down. In these
things, it's -- well, when the safety ejection comes back
in, and whether or not you have pressurizer spray determines
whether or not you can control the pressure. If you -- when
you've got decay heat in the there, and if you isolate a
feed water generator, steam generator, you've got to start
your auxiliary feed water into your good feed water heaters.
DR. CATTON: How could they recognize whether or
not they have a trip level in the head? Oops, sorry. Go
MR. SPENCE: You're going to get a pressurizer
level, basically down to zero, which happened. That -- at
Turkey Point, that lasted I think for 15 minutes. For
Robinson, I think it went 30 minutes. Okay.
And you also have -- I'm sorry -- you also have
temperature -- should have temperature indication up there.
DR. BONACA: This is all for steam generator tube
rupture. I mean, this is not steam line break?
MR. SPENCE: This, this. You're right. This is
steam generator tube rupture.
DR. BONACA: Okay.
MR. SPENCE: Okay. The only thing that wasn't was
my comment about the noise.
DR. BONACA: That's right.
MR. SPENCE: Okay. If you loose the pressurizer
spray, you can also go solid on the primary system for a
long period of time. Several plants overfill their steam
generator. In the Halden experiments, I think it was two,
and please correct me if I'm wrong, Jay. Two operator
simulations basically were going to rupture the steam
generator tube sheet, is that correct? Okay. Two out of
four dealing with that particular scenario.
Power operator relief valve. This happened at
Indian Point Two, because it took them so long. They were
running out of pneumatic supply, and they had to make a
containment entry to put in some more bottles.
Delaying the power operator relief valve, again,
is the operators want to keep the pressure up, and for the
sub-cooled margin, when they actually got to open it up and
get down as soon as they can.
Okay. Delay in initiating RHR happened in Indian
Point Two, where they had a procedure glitch with respect to
I think it was the temperature where they could get it on.
It was changed in one procedure, but not in another. Yes,
MR. HIGGINS: Based on all of these things, you're
saying that the one times ten to the minus three human error
probability is too low, I assume, and are you proposing your
own alternative value or are you just saying it's too low,
you don't like that one?
MR. SPENCE: I'm suggesting it's too low. I don't
like ten to the minus three. That's a 99.9 percent chance
of everything working, including the operator. I am not
proposing a specific number, because these are all problems.
Now, these are all successful events. All I'm saying is
there were a heck of a lot of operator problems involved in
MR. HIGGINS: The typical value for this error in
most of the IPEs in their tube rupture sequence is about
times e to the minus two, would you agree with that one?
MR. SPENCE: It would depend -- I think it would
be on a sliding scale, believe it or not. I think, because
the operator error rate and how it affects a steam generator
tube rupture and main steam line break is going to depend
upon how much time he has to fulfill his functions, and as
his time decreases, the more the probability of him being
unable to fulfill his functions is going to get smaller.
DR. BONACA: I would like to comment on one thing
about that. This scenario probably is one of the most
challenging that the operator has, because right now, the
operator has goals of containing the release within 30
minutes. That's one thing that challenges the operator to
no end, because they get into the event. By the time they
recognize it, they are dealing with all these issues, and so
the issue is not really core damage, but they are focusing
always on stopping the release within 30 minutes, and that's
very challenging. I mean, for some plants, it's very hard
to demonstrate and get to something that they have to do on
the simulators or show they can do it. And so, it puts a
lot of pressure on that. Although the other one is much
more complex steam line break, everything goes in the
direction of taking you down to the RHR. The larger is the
break that you have on the secondary side, even if you don't
recognize it, and you're depressurizing fast, and you're
unable to come back and pressure the primary site because of
So I'm saying that, you know, I'm not questioning
at all that 10 to the minus three might be overly
optimistic. In fact, it may be, but I'm only saying that
this is a different scenario and is one of the most
challenging to the operators, because of the goals that set
on them, which is the 30 minutes to stop the release.. And
that's very hard to do, very hard to do.
MR. SPENCE: And I think that between the maximum
and minimum breaks that we might be talking about here, I
think the worst case is going to be in the middle some
place, because like you say, one, the reactor is going to
take itself down and the other -- and the one in the center
is the one he's really going to work on.
DR. BONACA: And small, too. The small ones may
be the most confusing because he has a steam line break. He
doesn't have much leakage.
MR. SPENCE: Right.
DR. BONACA: Therefore, he may stay -- you know,
the steam generator may not blow down as fast for him to
MR. SPENCE: Right.
DR. BONACA: And he might just do something that
is totally, you know--
MR. SPENCE: But then, he doesn't have to worry
about the refueling water storage tank as much because he
has a longer period of time to respond.
DR. BONACA: That's true. That's true.
MR. SPENCE: Right.
DR. POWERS: Mr. Bonaca, I wonder in some sense if
the event tree doesn't have to be fairly complicated here?
Because an operator can, in the end, do everything
correctly, but if it takes an excessive amount of time, not
so excessive that we would run into the RWST probe; that if
we have damage propagating at a crack growth rates, that
gets you into an irreversible problem, and I'm wondering if
simply debating over 10 to the minus two, 10 to the minus
three in an operator's success or non-success is a
sufficiently sophisticated event tree for this.
MR. HIGGINS: Yeah, in actuality, the dominant
sequence is related to tube rupture in most of the plants is
a failure of HPI, coupled with the steam generator tube
DR. POWERS: Right.
MR. HIGGINS: And then the other item in that cut
set is typically this operator failure to rapidly
depressurize. And it varies from plant to plant, and it
varies depending on the leak size. But often times, they
only have in the neighborhood of 15 minutes to do that for
success so that puts even a tighter on the operator action
in order to have successfully cooled down without core
MR. SPENCE: Yeah, but now, with our symptom-based
procedures, by the time he figures out what he's got, and
gets into the procedures, if it's small, okay, and then he
takes, you know, 20 minutes to get into where his actions
are going, he's in trouble.
I'll try this, and if you want to but in and take
over, go ahead, Jay.
Jay Persinsky is our team leader on human
performance issues and research. And he had a study on
human performance over at Halden, which is in Norway, right?
And they were working with the LEVISA crew out of Finland.
And they were looking at both -- at staffing levels with
respect to the current type of operations on the current
type of plant control rooms versus the new type of CRT
And that you were looking at four operators on the
normal plant, and two operators on the other plant.
The scenario that -- they did a number of
scenarios, but the scenario that was of interest, of course,
to me was a steam generator leakage rate -- leak rate --
which was then followed by an open steam generator safety
valve for an unfiltered release to atmosphere. Like I said,
we've got the tapes, the videotapes of the actual scenarios,
with an English translation somewhat. I looked at it--at
one particular one -- that -- and it was really consistent
of what could be found in an American plant. They were
using some procedures. One of the fellows was trying to
diagnose stuff, but he was not -- they were not diagnosing
the scenario correctly. Okay.
This -- one of two four-man conventional reactor
crews, and one of the advanced reactor crews performed very
poorly, and I've got the tape on this if you'd like to see
it. And the interesting result is that the longer -- this
scenario, too, goes on for a long time, because he's cooling
down for a couple of hours. Someone asked about training.
The trouble with the training is that once -- because the
utilities have got to train on so many different scenarios,
once the operators diagnose it, they say okay let's stop
that, and we'll go on to the next one. Where they're
running into trouble is the long-term cool-down and switch
over on this thing.
DR. POWERS: I just have to ask a question.
MR. SPENCE: Yes, sir.
DR. POWERS: Isn't Holden a boiling water reactor?
MR. SPENCE: Jay?
MR. PERSINSKY: Yup.
MR. SPENCE: Yeah.
MR. PERSINSKY: Jay Persinski, Office of Research.
The Halden reactor is, in fact, a boiling water reactor.
This was done on the simulator. The simulator is, in fact,
a DVBR, which is a type of pressurized water reactor.
DR. POWERS: I guess -- then it comes to mind that
we have a Russian crew working this or a Finnish crew
MR. PERSINSKY: It was a Finnish crew working.
It's a LEVISA crew that typically worked the LEVISA plant.
One of the -- where he talks about the conventional reactor,
that was actually done at the LEVISA simulator.
DR. POWERS: Oh, okay.
MR. PERSINSKY: The advanced crew, or the advanced
reactor was done at the Halden simulator.
DR. POWERS: So we didn't have a problem of
boiling water reactor crews trying to do a PWRC?
MR. PERSINSKY: No, we did not have that problem.
MR. SPENCE: Not only that, but the simulators
were modified somewhat to look like an American reactor.
DR. POWERS: Now, did that introduce any problem.
I mean, I've got a Finnish crew familiar with a BBR Russian
reactor working on an American modified simulator. I could
believe they might have some trouble.
MR. SPENCE: That wasn't the problem.
DR. POWERS: Okay.
MR. SPENCE: The problem was their diagnosing and
getting the correct answer of what was really happening to
the plant, okay. And that was the same type of thing I
would have expected to see in a control room. I saw it in
Turkey Point for quite some time. When I got in there,
about 30 seconds after the line blew off, you know, I'm not
believing that I saw a safety valves go up and fly over the
containment. I didn't want to tell them that. You know, I
says -- they would have thrown me right out, okay. And it
wasn't until, you know, the noise stopped and we could go
out and didn't have to worry about shrapnel and so fort that
we could ascertain what really happened.
Okay. The point being that this takes a long
period of time to cool down and to stop that release and
that the training scenarios probably don't go into that long
thing. And the high work load is going to build up on these
guys, as time goes on. That was an interesting conclusion
They also tried it with three men and what was it,
Jay, on the one out of two? Did they cut down to one or--
MR. PERSINSKY: No, it was two or three in that
MR. SPENCE: It was three here, normally, and then
they cut down to two?
MR. PERSINSKY: Right.
MR. SPENCE: Okay, and this was four, they cut
down to three. Well, that's exactly what's going to happen
in an actual scenario, because either you're turbine
operators are not going to be there, or you're going to send
a reactor operator out to find out what's going on, or
you're going to have fires, and somebody's going to have to
go off into the fire brigade. Or you're going to have to
call the NRC and stay on the line and tell them what the
heck is going on. So you're going to lose somebody.
DR. CATTON: And you also couple this with some
wort of implicit faith in the ability of the simulator to
represent the event, and we know that's not true.
SP Yeah. Yeah.
DR. CATTON: I just thought I'd put that out.
MR. SPENCE: And operate it -- when it really hits
the fan, operators just kind of stop for a minute to try to
think what's going on.
DR. CATTON: An interesting example of this in
some of the testing that was done at the University of
Maryland, where they found that the water in the primary
system was moving from one loop to the other, and all kinds
of strange things were appearing on their instrumentation.
And that doesn't happen with a simulator. And you wonder
what would the operator do if he saw that. It doesn't fit
any of his symptoms.
MR. SPENCE: Yeah. I -- you know, I can tell you
from real life that the minute that happens, the operator
just -- you kind of freeze to try to figure out what's going
DR. CATTON: Not according to NRC. Not according
to NRC. Relap properly reproduces the accident.
DR. POWERS: Well, first -- Catton, I point out to
you that we do have a list of events where they successfully
got the plant down. So.
DR. CATTON: In spite of all these operators.
DR. POWERS: In spite of all these possibilities,
it is possible to get these plants down.
MR. SPENCE: It is possible, okay, and those were
relatively small leaks. Okay. Jay, had--
MR. BALLINGER: I would -- Gannai with 700 gallons
MR. SPENCE: Well, that's also--
MR. BALLINGER: The water was -- my was 700 in
MR. SPENCE: That's the initial, that's the
initial leak rate. And then it decreases from that. So
that's only the top.
MR. BALLINGER: And you would expect that in any
case for a steam generator tube rupture?
MR. SPENCE: Yes, sir. Okay. But now, to reach
your part 100 limits, I believe you're dumping 10 gallons a
minute off site. So -- I think the average is 130 to 135
GPM or something like that of all the steam generator tube
ruptures. So you can put that in -- and if you assume that
your pressure -- that at main steam line break point, you're
going to have X and then you're going to have one-tenth X
later on, you're still up. The NRC assumes a factor of 10
difference in the leakage rate between main steam line break
and normal operating conditions.
DR. HOPENFELD: In the case of Gannai, the tubes
were not -- there was an effect on other tubes, but the
tubes were not defective -- the fuel was brand new.
MR. BALLINGER: It was a wrench or something. A
loose part, right?
DR. HOPENFELD: A loose part, but so that really
-- they did it. I think they took samples -- it was -- they
took samples from the snow in that spot that did exceed, but
it was really a brand new fuel, so there was no really
reason to expect it. But the point is, in my DPO, my
thinking at the time when I reviewed all the data, I thought
after a thousand that when you exceed a thousand, that's
where we really -- that's where we really start worrying.
DR. POWERS: Dr. Kress, while you were out,
professor Catton brought up the experiences at the
University of Maryland, I believe, where because you have
multiple loops you got water transferring from one loop to
another. I think that's done correctly, and I think that's
an issue that you were mentioning to me is the concern.
DR. KRESS: It definitely was.
DR. POWERS: And you might want to pursue that a
little with professor Catton.
DR. KRESS: Yeah. We'll get together.
DR. CATTON: Well, we raised that issue. That was
before I left the Committee when that issue was on that.
And we were told that the simulator -- the simulator
fidelity is proven by comparison to RELAP, and that's the
way it is.
This has been something that has bothered me for
15 to 20 years.
DR. POWERS: RELAP is the ASME standard of--
DR. CATTON: Whatever.
MR. SPENCE: I've got one last point to make on
the operator thing. And that is with respect to risk and
the probability of whether or not the operator is going to
perform his functions. Jay Persinski worked with
Microsinks, is that -- with a -- he got a little contractor
-- Microsink Task Network Model, which was basically a
modeling of operator performance for steam generator tube
rupture and a stuck open relief valve, okay. We could have
had it modified to see what a steam generator -- I'm sorry
-- a main steam line break with a steam generator tube
rupture could have done. This was ruled out of scope in the
previous DPO Panel, and it was also ruled out with respect
to research by our manager. We really thought that that
would have been good, because then we could have got in
there -- it's got the whole analysis of what the operator
actions have to be. You could tweak the times. Try to set
them up. It was already set up with respect to the Halden
experiments. And we could have gone back to fit in some of
the steam generator tube rupture events as well as done some
testing down at TTC to set up some scenarios. But again,
that's out. That would have been a good thing to do to
really find out what the operators might be expected to do.
Jay, do you have any comments on that or -- any other
DR. HOPENFELD: I have two items to cover, and one
of them is fairly simple. The other one is very much more
complicated, and I'd like to go fast so, because I see
already that that I'm starting losing my clients here. So.
DR. POWERS: I think you've got the panel here
with -- in rapt attention.
DR. HOPENFELD: As you see, I have a very, very
lengthy summary about 50 pages, well, maybe not 50, but
about 20 pages of specific questions which really summarize
what I've been talking about, but on a much more specific
level. And I though that because of the time clicking here,
I guess everybody probably want to go home. So I'm going to
just complete this part of the presentation. Go through two
subjects. One is the iodine spiking, which is relatively
easy. And then the other one, which is very difficult for
me to talk to, but I have to, and that's an independent
assessment. But I don't think that's going to take us --
oh, probably, we should be done by 5:00 p.m., and what I'm
going to leave the discussion, the specific questions to NRR
for you to look at, because they go to a lower-level of
specificity and, in the future, if you want to address those
to NRR, I don't know how else you could handle it. If you
had more time, I would have gone through it, but I think --
I'll try to give you the flavor as to what I'm talking
about, and I think that going through these slides is not
going to really add too much to the overall understanding.
It's just another level of specificity.
But let's go to the next item, and that so far
we've been talking primarily on the design basis accident.
We talked a little bit on the severe accidents. The next
one is a legal requirement that you have to meet part 100,
which you have to leave -- I mean 300 REM as the result of
an accident. What the -- what the dose is, is very simple.
The equation for calculating dose is extremely simple. What
it is is the leakage time the spike times the initial
What the spike means is when you go -- disturb the
system or disturb the primary system, due to temperature or
pressure drop or whatever, any crack in the cladding will
flush some of the fission products out of the fuel and that
goes into the coolant. There are also other sources. This
is not the only source. There are corrosion products that
may be laying around, and when you shock the system, you may
get corrosion product coming into the -- that were deposited
on the fuel. So it's a -- the mechanism is not understood,
but I don't know how important it is to understand it. The
calculations that we are doing -- the important thing is
that this is not exact science. People, over the years,
sort of empirically came up with some numbers. And there
was a conservatism, and nobody really tried to quantify that
conservatism. It may not even be necessary.
Iodine chemistry is a very complicated thing,
especially if you can see, though we are talking about very,
very low concentration. We're getting into the region that
maybe the classical chemistry may not even work anymore,
because the mean -- because the molecules already is not
getting out of where you can calculate your equilibrium
factors. So we don't want to get into that, but we don't
want to do things that violate some basic laws.
And that's exactly what NRC does. What happened,
again, going back, remember, we had steam line -- you had
SGTR, and you have a steam line break. Now, we've got a new
phenomenon, so if -- when you come up with larger leakages
that would allow you to meet the 300 REM requirement, and
I'll go back to the 300, too, but what you can do, going
back to the equation, is a very simple mathematical trick is
-- well, you can say, well, this spike here, well, if the
leakage is higher than off the one GPM, what I can do is
just get this one down, and I'm back in business. I'm
sorry, I can get the initial activity down, because I have a
control over that with a clean up system. And most of the
power plants operated at a lower than tech spec activity
anyway, and then I'll be in business.
Well, it turns out that you can't do that; that
it's not that simple. And the reason it's not that simple
because there is data to indicate that if you are -- if you
lower the iodine concentration in the coolant, then it
affects this spike. I don't exactly claim to understand
why, but if you look at the data, it shows that you can't
just arbitrarily -- there is some dependence here between
the spike and the initial concentration of the iodine in the
Now, for years, the plants did not want to lower
that initial concentration. They were happy with the one
microcurie per gram, and I don't know -- it could very well
be because of contractual obligations or the legal mumbo
jumbo that was in the contract between the supplier and the
power plant. But anyway, they didn't want to go to a lower
than one microcurie per gram. Now, suddenly, we find that
NRR says, well, in order for us to meet Part 100, let's
lower that tech spec, allow them to operate at point -- give
credit for operating at point one or whatever. Well, you
can't do that if there's data out there to show that if you
do that, that 500 number, whatever that number is, but it's
sort of a consensus number, you can't take that 500 number
and still lower the concentration at the same time, because
you can have a spike that's just -- that's all the way up to
The point I'm trying to get across -- you can't
just do these things -- adjust these things just to meet --
to get a final answer that you are happy with. You have to
be consistent as to what you're doing, and I think the NRR
people are not consistent with what they're doing. And I
think this was recognized a long time ago. I think in '94,
when I presented this and we discussed that, but nothing has
been done since.
What was done basically, and I think that's what
you'll probably hear tomorrow, NRR provided a table that
shows -- basically agrees with my argument here that if you
lower the concentration and if you put a larger leakage, and
the larger spike, then you would -- you could exceed the SRP
value of 30 REM. Now, in the -- Part 100 calls for 300
REMs. The SRP calls for 30, and I don't know exactly what
caused -- where the 30 number comes from, but again, it sort
of evolved over the years, and it's an empirical number, and
you said, in order to meet that 300, we have to use 30.
Otherwise, why not use 300? So they put in the SRP, and
that's what the licensees are required to meet, is the 30.
So when we're talking about meeting the requirements it's
the 30, and it's not the 300. So you can't just say, well,
that what's was done in the table that they provided us is
that, look, you'll have to have a huge spike in order to
exceed the 300 REM. But it's not the 300; it's the 30. But
even then, the main point is that where you don't have any
data, and one argument was that, well, steam line breaks
don't occur very often. Well, they don't -- we hardly have
any, even though we heard one or two, then why even worry
about the regulations. Just forget about Part 100. Just --
if we don't have to worry about it.
So the argument was -- that's made -- and that's
the crux of this thing is that we don't have data to show
that on the steam line break conditions the depressurization
is so high that the iodine spike is going to be very large,
so don't you just ignore it? And what I'm claiming here is
that you just can't do it in an arbitrary way. And I think
that hasn't been resolved as being recognized as a potential
issue and it's still there.
Now, the thing away. But then, after the comments
on the DPO came from the -- after it came back from public
comments in the summer of 1999, they just took that sentence
of the paragraph which says that they're going to ask the
licensee to come up with a better leakage assessment or put
uncertainties on it, and that's why I say, well, if you do
that, then you better go and look at this iodine spike,
revisit the whole thing again. You have to address this
issue. And they just know -- I don't know how you resolve
it, and I don't know what you do. Now, there's some
suggestion from Dr. Powers several years ago, but I don't
know of anybody who picked up on that.
So, basically, you cannot -- to summarize it, we
cannot arbitrarily to say, well, because you want to meet
the Part 100 requirements, we're just going to lower the
activity, the initial concentration without really looking
at what the data. If you do that, you selectively use
what's available there. What the database is. If you want
to operate in that -- on the basis that you can use
selective reasoning, then it's okay.
Let me go back a little bit now. I'm done with
the iodine. I know you don't want to harp on it, because I
don't know what I can add to it. Let me go back to the time
line, and I stop here, around June or in mid-June, ConEdison
submitted a proposal for to justify -- or the justification
for the next cycle. The public, or some members of the
public, the Union of Concerned Scientists, was very critical
of that, and they have asked me to -- or they asked the NRR
or NRC to allow me to talk about these issues at the public
hearing. And NRR said no. I was kind of a little bit
disturbed about it, because what is it that I could -- why
prevent me? What is it that I could harm anybody by talking
about these issues, basically, would summarize what I told
you all today, I would summarize into a few minutes, so
somebody would get a flavor what different -- perspective on
But that -- that -- really what bothered me about
their reply -- that -- the reason that was given -- the
reason was given that they don't want me to talk about this
at the public hearing was because the DPO issue is the
generic issue, and the IP2 is the specific issue. Now, it's
been now three months, and I've talked to a lot of people in
trying to help me to understand the difference between the
specific and the generic. I don't see how you can separate
the two. But anyway, later on, it occurred to me the reason
that they really didn't want me to -- prevented me from --
to come and talk about it was basically what the IG found
out in two months or three months later. And that is that
we let inexperienced engineers review those actions, which
are very, very important safety actions. They are not
supervised, and they have constraints on them. They're not
allowed to have an open dialogue with the licensees. So you
have an inexperienced person -- reviews an action, and he is
constrained to follow up on that. And what -- the reason
I'm bringing all this is that it calls into question as to
how we do business, and what is really is needed -- what is
really needed is an independent assessment. When a licensee
comes here, and he wants to take a -- ask for relaxation,
whatever the relaxation is, I think we should have a third
party that says, that provides an independent assessment of
what that action is. When you go and buy some instrument
that has to do with monitoring the environment, you can't
put it on the market until you get EPA approval on it that
it was tested by a third party. And the same thing here.
You have to have an independent third party that can step
aside and assess what the licensee is submitting to you,
because you don't have enough checks and balances within
this agency to take care of that.
Now, one of the items that I talked about, and
it's a follow up on actions at the NRC, and it does relate
to the independent assessment. In the -- one of the -- I
believe -- I don't remember the date. I think it was -- I
think it was the '94 ACRS meeting, the ACRS told the staff
that we need more adequate data for empirical correlation.
The empirical correlation was inadequate. I don't see that
we have any -- after six years that we have any data which
is more adequate. We have some little bit more data, but
it's not necessarily more adequate if you consider all these
effects of vibration and forces that you get through a steam
line break. So we don't have any.
Then a more adequate characterization of the
gradation of leak assessment and morphology affect the
morphology on the leak. We don't have that. Then there was
a requirement for NRR to come up and quantify, and you can
look at the letter to the Commission to quantify the
conservatism because they claimed that it's too conservative
independent of how you -- in spite of how you use the iodine
spiking. The request was that they quantify this. There's
nothing on that. Then we have another, and that is the
GSI-163. That 163, which is a high priority, has been in
the works now for nine years, eight years. It's still
fairly young compared to some of the 17 years that you've
seen before on the pump seal. So it's not really that --
it's not desperate yet. It's still got many years to go and
But that 163, the reason for that was given that
it's not being worked on is that first, we got to resolve
the DPO. I submit to you the resolution of this 163 and the
resolution of DPO are completely two different things. The
subjects are the same, but the procedures, how you resolve
GSI is different. You go to a cost benefit. You look into
different design options. It really is not the DPO. It
just talks about the issue. It just briefly looks into
this. I haven't had a chance to go and look into really --
put -- start with a clear -- see what other options are, and
there are options. I may not just give you something off
the top of my head. You can put a double-walled pipe in
there, with leak detector. I'm just talking off the top of
my head. But there are other options.
And that's where the GSI is supposed to look at.
But if nobody wants to work on it, you continuously keep
delaying it because it could very well be that you'll have
to have a back fit. So if you -- well, let me ask you,
right now, the latest thing in the gimmicks -- and this is
the means which we communicate to the public -- it says that
it depends what you people are going to come up with.
That's how you're going to be -- you are the ones who going
to be resolving this GSI-163. I don't think that that's
what you -- you may have not know that, but that's what
you'll be doing. If you say that that DPO doesn't have
enough merit, you also said, well, you might as well close
this GSI-163. And that's exactly what they have done.
Now, there is, at the same time, I told you after
the failure of the rule making, the generic letter, the
regulatory guide, all these were substituted by discussions,
which I haven't been to any of them. I don't know whether
they are open door or closed door. It doesn't matter,
because all the data is all proprietary. So you can even
sit there, but you don't know what they're talking about --
I mean, if you're from the public.
And there's such a huge amount of data that you
have to spend your lifetime to go through there, and it's
very difficult to understand it, and I hope you'll go
through some of it, and you'll see it for yourself. So you
have this discussion going with NEI to come up with an
agreement. And now NRR says, well, the DPO has nothing to
do with this agreement that we're working. In other words,
the DPO has nothing to do with degraded tubes because it's
not related to it. We have something else we're talking. I
hope that tomorrow, they'll tell you what it is that they're
talking to NEI about. They are not talking about any -- I
mean, according to what they stated, and they stated on
several, and I have the thing in writing from the EDO saying
that the discussion with NEI, that really is going to come
up as to how we going to regulate the steam generators for
the next 20 or 30 years really has nothing to do with all
those items that we talked about today. So I don't know
what they're talking about, but I hope you ask them. Ask
them, what is it they're signing. What is it they're
agreeing with it?
About a week ago, the IAG came up with a
recommendation -- finding about the DPO process at the NRC.
Now, why I'm telling you? Why is the DPO related to? If I
am asking you or I'm recommending that you have a function
that does an independent assessment to NRC activity, the
reason for it is that you don't have a check and balance
system within the NRC. There is a system what's called
DPV-DPO, which I briefly talked to you at the beginning of
this meeting, but it's ineffective. So putting all this
together, you have a function at NRC, the regulatory people
do not take your recommendations seriously. They do not
follow up on that. At a meeting in '96, I believe it was,
and I brought it in, you can look up in your time line, you
asked me or you specifically recommended to the Commission
that the NRC -- the NRR staff resolves the DPO and resolve
the GSI before they come up with their rule making. Well,
now, they've substituted the rule making with the agreement
with NEI, and they say, well, the DPO and the GSI are not
related to it. So it's sort of going in circles here.
To summarize, the methodology in GR-95 was adopted
by NRC in its entirety from Westinghouse. Westinghouse had
a very good reason at the time. They were being sued left
and right. And they had a very good reason to come up and
explain away how they can keep these steam generators alive
for a longer period of time before they are replaced. And
we took it, what they recommended, and followed up
completely, and bought it completely. The ACRS was sold on
the 95-05, and I've spent some time before, there was a lot
of information that was provided to you what I believe was
misleading. So when you concluded on that basis that the
risk was really 10 to the minus seven or whatever, that was
on the basis of the information that was provided to you.
And because of that, and the 1570 relates to the
severe accident. That's not really the main thrust of my
presentation, but basically what this -- what this reads to
a conclusion, recommendation, and that's basically the
bottom line to rescind 95-05 and shut down all the plants
that don't meet the 40 percent -- the 40 percent plugging
I think I'm 10 minutes over my time. And I really
appreciate the time that you gave me. We could -- I don't
want to bore you with all the questions I have, and I'm not
going to go through that. I thought that if we had time, I
would, but it doesn't look like it would be a fair thing to
DR. POWERS: Well, I don't want to deter you. If
you think that the questions are self-explanatory
HOP I think so. I think after this presentation,
they are. I believe so.
DR. POWERS: Well, in that case, first, I'd like
to thank you for an outstanding set of presentations. Very
well put together. Very clear. Fast-paced. Went right
through the material in a nice way. Then I'll turn to the
rest of the panel and the consultants, and ask if you have
any questions on the material you'd like to direct to Dr.
Hoppenfeld at this time.
MR. BALLINGER: I have a question of you. If we
read these questions, and then we discover that there is
something we don't understand, can we?
DR. POWERS: We'll figure out some way to handle
that. I think Dr. Hoppenfeld may be away toward the end of
this week, so he may not be directly accessible, but in some
way, we will get a hold of you.
DR. BONACA: I would like to point out -- I
thought the agenda that we had time until 5:30 p.m. for
DR. POWERS: We have plenty of time.
DR. HOPENFELD: I would like to make a comment to
reply to you. My telephone -- you have my telephone number.
I'll be -- and obviously, any question you have, though,
please e-mail it, and I'll e-mail back to you or reply. I'm
going to be out of the country for three weeks as of Friday,
but then I'm going to be back, and I don't know how long I'm
going to be here, but I'm going to be enough to answer any
DR. POWERS: Yeah, the protocol for members of the
panel to communicate with anyone on the staff is go through
Undine. That is, talk to her, and she will get the answer
for you. Right? Of course.
DR. HOPENFELD: I am responsible for all these
questions, and I would answer them very, very -- there were
a couple of things -- I just in passing in mentioning. I
think, when you hear from the staff telling you about their
beliefs or their judgement, or their -- I think you got to
find out what their experience is, what the qualification is
to make these judgements.
DR. CATTON: Could I just try to make sure that I
understand what the primary issues are? After listening to
you all day, I kind of get lost in the detail, but the first
was the meaning of voltage and its relationship to leakage.
That was number one.
Number two was impact of the main steam line break
or other similar kinds of upsets on leakage in overall tube
The third was the severe accident issues as raised
by risk-based regulation.
DR. HOPENFELD: That's right.
DR. CATTON: The fourth is--
DR. HOPENFELD: Because it's raised by risk--
DR. CATTON: I understand. Without risk-based
regulation, you have the deterministic approach and the
issue doesn't come up.
The fourth is the operator performance. And the
fifth really is managerial issues and how DPOs are treated.
DR. HOPENFELD: Process. Process.
DR. CATTON: Okay.
DR. POWERS: Process.
DR. CATTON: Okay. Managerial process.
DR. POWERS: Raising the issue across this reminds
me some have asked about the plans of the subcommittee in
conducting its business. I went through those some
yesterday, but I don't think they got the full exposition.
The schedule that the subcommittee has set up for this week
was intended to allow Dr. Hoppenfeld and the NRC staff to
present their views on the issues at hand. And, in some
sense, the various parties may be surprised by the
respective views, since I'm sure that over the course of
time views have been refined and expanded. And there may be
instances where it would be useful to have a rebuttal of
those views. The meeting this week has not been planned to
accommodate a rebuttal, but the subcommittee would be very
interested in any rebuttal views that people would like to
have and so we have implored the ACRS itself to make
available some time during its November meeting, and again
in its December meeting to allow rebuttals. The ACRS has
graciously consented to do that with a proviso that anyone
wishing to provide a rebuttal of the -- on the views that
are presented today and in the next few days that they
provide in advance a written summary of the rebuttal.
That's some piece of information that people
should have. Are there any other comments that the panel
wants to make?
What I would like people to do is clearly Dr.
Hoppenfeld has provided us a list of questions of some
length and of some interest, and I will hope that the panel
members will take some time to examine these questions and
examine the presentation today to see if they want to refine
their list of contentions that they prepared last night.
And with that, we'll stand in recess until
tomorrow morning at 8:30 a.m. And, again, thank you very
much, Dr. Hoppenfeld. That was very nicely done.
[Whereupon, the meeting was recessed, to reconvene
at 8:30 a.m., October 12, 2000]
Page Last Reviewed/Updated Tuesday, July 12, 2016