ML19256F101
| ML19256F101 | |
| Person / Time | |
|---|---|
| Issue date: | 11/02/1979 |
| From: | NRC COMMISSION (OCM) |
| To: | |
| References | |
| REF-10CFR9.7 NUDOCS 7911210433 | |
| Download: ML19256F101 (79) | |
Text
I NUCLEAR REGULATORY COMMISSION IN THE MATTER OF:
PUBLIC MEETING BRIEFING ON SWF.1ING OF FUEL CLADDING -
RESULTS_OF RF3EARCH & ACTIONS BEING TAKEN s,,
c.
y,,f,.rW,'(,-nt-O :?
1.,.,.
.s.
-i'
,,m.
_.c-
- - y ~.:p_.; ;_v,.q-w g,. +. ':;-*
~
.2
~.r.
n ;.;;? :- n.., cs_
. / M,,.,...
7 idp$ $d2NU f5
[
l
. g'cfj j
/;,
. -. -m n
~-
Pla.ce -9 Washington, D. C.-
.3, Date -
Friday, 2 November 1979 Pages 1 - 63 e
e m
.;g.,
jj/$
{}j
',f V
~
.?
r
~
=,
Telephone:
(202)347-3700 ACE-FEDERAL REPORTERS,INC.
OfficialReporters L
444 North Capitol Street gej-Washington, D.C. 20001 gg210 NATIONWIDE COVERAGE. DAILY
1 iR8140
(_)
DISCLAIMER This is an unofficial transcript of a meeting of the United States Nuclear Regulatory Commission held on Friday, 2 Noverber 1979 in the Commissions's offices at 1717 H Street, N. W., Washington, D. C.
The meeting was open to public attendance and observation.
This transcript has not been reviewed, corrected, or edited, and it may contain
' inaccuracies, b~
The transcript is intended solely for general informational purposes.
As provided by 10 CFR 9.103, it is not part of the formal or informal record of decision of the matters discussed.
Expressions of opinion in this transcript do not necessarily reflect final determinations or beliefs.
No pleading or other paper may be filed with the Commission in any proceeding as the result of or addressed to any statement or argument contained he~ rein, except as the Commission may authorize.
O e
i 3/S IS4 W
e
e 2
CR8140 1
UNITED STATES OF AMERICA q
/
2 NUCLEAR REGULATORY COMMISSION 3
PUBLIC MEETING
/ ;
4 BRIEFING ON SWELLING OF FUEL CLADDING -
5 RESULTS OF RESEARCH & ACTIONS BEING TAKEN 6
7 Room 1130 1717 H Street, N.W.
8 Washington, D. C.
9 Friday, 2 November 1979 10 The Commission met, pursuant to notice, at 1:40 p.m.
11 BEFORE:
12 DR. JOSEPH M. HENDRIE, Chairman
()
13 VICTOR GILINSKY, Commissioner 14 PETER A. BRADFORD, Commissioner l-15 JOHN F. AHEARNE, Commissioner 16 ALSO PRESENT:
17 Messrs. Eiser. hut, Denton, Meyer, Lauben, Matton, Bickwit, 18 and Levine.
l3/S 1S5 20 21 i
22 i
i 23 24 l
a,ww.i nmorars. tx.
l 25 I
1
CR 8140 3
MELTZER t-1 mte 1 1
PROCEEDINGS v,
2 (1:40 p.m.)
~ ~ '
3 CHAIRMAN HENDRIE:
We are gathered -- the Commission
')
4 please please come to order.
5 We meet this afternoon to discuss with the staff 6
some recent research results on fuel rod behavior in certain 7
accident conditions.
And the first item of business, since 8
this matter has come up rather recently and suddenly, I will 9
ask my colleagues to join me in voting to hold a meeting on 10 less than one week's notice.
Those in favor?
11 COMMISSIONER AHEARNE:
Aye.
12 COMMISSIONER BRADFORD:
Aye.
13 COMMISSIONER GILINSKY:
Aye.
14 CHAIRMAN HENDRIE:
Aye.
So ordered.
15 I see we have the assembled might of the Office of 16 Nuclear Reactor Regulation.
Harold, why don't you go ahead 17 and lay this matter out for us.
18 MR. DENTON:
Let me introduce two people at the 19 table you might not know.
At the end is Norm Lauben from 20 the Analysis Branch; and Ralph Meyer, who is chief of the 21 Fuel Section.
He will carry most of the presentation today.
22 I sent you a memo earlier this week on a potential i
23 deficiency in the ECCS evaluation model.
The reason for the 24 memo was a result of fuel clad burst tests that had been a
wel Reporuts, lm.
25 performed at Oak Ridge over a wide variety of conditions.
i i
i3/S 156
mte 2 '
4 I
When we first compared the results of those tests to the 2
evaluation models which had been approved for use by the 3
vendors, we found some significant differences.
Yesterday we 4
met with all the vendors that sur gly fuel for power reactors, 5
and as of late last night had concluded that the bulk of them 6
demonstrated that in the region of interest their models were 7
sufficiently conservative and did envelope the results that 8
were obtained at Oak Ridge.
9 We obtained further information today from 10 Westinghouse, and I think with possibly one or two exceptions, II which we will explain later, the vendors have satisfied us that 12 all the plants are within -- that the performance of all the 13
{
emergency core cooling systems are within the Appendix K I#
requirements of the Commission.
15 I would lika to have Roger Mattson explain a little 16 bit more about the nature of this problem, and then we will I7 turn down the table for some detailed presentations, to compare 18 the evaluation models of the vendors with the results from I9 the Oak Ridge tests.
20 DR. MATTSON:
I'd like to start by going back with 2I some history a little bit, to 1974, when the Appendix K to 22 Part 50 rule:was issued by the Atomic Energy Commission, You 23 will recall that was a result of some two years' worth of 24 national rulemaking hearings, starting in 1972.
Ao eral Reporters, Inc.
25 One of the issues treated in that rulemaking hearing '
i 1375 157
mte 3 5
I was the question of swelling and rupture of fuel rods as they
)
heat up during the loss of coolant accident.
Thereisaspeci'l 2
3 fic section of the final rule, section Roman numeral I.B, that 4
speaks to how swelling and regture are to be treated in the 5
evaluation models.
Paraphrasing that section, it says that 6
swelling and the incidence of rupture are not to be under-7 estimated for applicable data.
8 Now, at the time of the hearing the Commission 9
identified a number of gaps in knowledge, as they w're called e
10 at that time, that required continuing research to confirm Il that certain bounding assumptions made in the models in the 12 period 1973-74 had in fact been bounding assumptic a.
And 13 one of the areas highlighted for this continuing research was Id clad swelling and rupture.
15 We estimate that since that time there has been 16 about $40 million spent by the NRC and the vendors of fuel 17 researching this area to increase the understanding of swelling I
18 and rupture.
We have been following this research closely 19 since 1974.
20 Back in January of this year, you may recall, we 21 required some changes in the Combustion Engineering evaluation 22 model because the data had clearly indicated to us that by 23 that time that Combustion was underestimating the incidence 24 of swelling and rupture in a range of interest to those plants.!
cr
'wei Reporters, Inc.
i 25 There was a broad notification and some changes in the model l
15/$
158 i
mte 4 6
I and some calculations of peak clad temperature to confirm that 2
they conformed to the 50.46 limit of peak clad temperature 3
of 4200 degrees.
4 Since that time, there have been more data become J
5 available from the Oak Ridge multirod burst tests, which are 6
probably the most definitive tests yet performed in this area.
7 They not only add to the body of knowledge, but they also give 8
you a capability t o interpret the large number of single rod 9
tests that have been done down through the years in this 10 araa.
11 Ralph Meyer from DSS has been in charge of following 12 that information, interpreting that information as the data became available, and has finished a staff report on rod 13 k')
14 swelling and rupture, which is due to be issued in the next 15 week or so.
16 I'd like to turn now first to Norm Lauben to ask 17 him to describe how swelling and rupture fit into the overall 18 calculation of peak clad temperature in loss of coolant 19 accidents.
That is one of several boxes in the large black 20 box of Appendix K evaluation models.
I 21 After that brief description of how it fits all in, 22 then Ralph will summarize for you what we've learned about the 23 data and why the evaluation models for the operating plants l
l 24 fall in a region of the data which leads us to believe there l
me
-wW Remners, ltw.
25 is not a safety problem at this point in time.
mte 5 7
I COMMISSIONER AHEARNE:
For most of the plants?
'O 2
MR. DENTON:
At the end of the meeting we'11 come 3
back to the one or two cases that we're still following up on m
4 e -.
today.
5 6
7 8
!375 160 10 11 12 f,
13 U
14 15 16 17 18 19 20 21 22 23 24 ce nel Reponen. Inc.
25 l
l
40 02 01 8
kap /MM i
MR. LAUBENs Thank you, Roger.
May we have the r
2 first slide, please?
3 (Slide.)
(')
4 This is a further breakdown of the smaller box 5
that Roger was talking about, having to do with the more 6
im por tant parameters involved in the fuel swelling and 7
rupture part of the analysis.
When we talk about swelling 6
we generally mean two things:
strain on the cladding, which 9
is the degree of expans ion in the cladding s and blockage, 10 which is the resultant blockage of the f uel channel for flow 11 during a loss of coolant accident.
12 There are really three things that we have been 13 looking a t.
On your handout they are outlined in heavy 14 black.
Unf ortuna tely on the slide they are not, but the k) 15 first model has to do with pin pre ssure and rupture 16 tem pe ra ture.
That is the relationship between those two 17 parameters.
The second is the strain and the third is the 18 blockage, all three of which are part of the requirements, 19 in appendix K.
20 In determining first whether the cladding will 21 rupture or not, there are listed on the top lef t-hand box 22 six parameters which play the major role in that.
One, of 23 course, cladding temperature.
As the cladding gets ho tter 24 it tends more to want to rupture.
In addition, fuel 25 temperature plays an important role.
Burnup in terms of 1375 161
40 02 02 9
kap /MM i
fission gas, the amount of gas inside the cladding between
(})
2 the f uel and the cladding also is important.
3 In addi tion, the clean dimensions of the fuel pin 4
and the exposed dimensions as the fuel burns up also plays a
-])
5 role.
The temperature of the plenum, which is influenced by 6
the coolant in particular and certain other f actors, is 7
im por tan t, as well as the plasti.c strain.
By plastic strain 8
here, I mean the degree of expansion of the cladding prior 9
to rupture.
10 Obviously, if the cladding expands more that 11 reduces the pre ssure inside the pin.
12 All these, then, go to influence the calculation 13 in each one of the evaluation models of the pin 14 pressure. Then each evaluation model also has a relationship b-15 between pin pressure and rupture temperature.
Now, the 16 recent experiments clearly show an additional f actor 17 involved in whether or not the cladding is going to rupture 18
-- is what we call the ramp rate, over on the right-hand 19 side.
20 By ramp rate, we mean the rate of change in 21 temperature witn time.
If the temperature is rising f aster 22 it's going to have an eff ect on whether or not the cladding 23 ruptures.
And as you can see f rom the arrows, it's going to 24 have an effect on the degree of strain of the cladding, and 25 also degree of the flow area blockage.
i3/S 162
40 02 03 10 kap /MM 1
I've also listed two of the more important
('
2 parameters that influence ramp rate and they are, of course, 3
the power of the particular pin and in addition the heat rg 4
transf ered into and out of the cladding.
As it turns out, 5
then, going back to the. lef t-hand side, the strain that is 6
on the cladding influences the calculation in the following 7
ways the strain quite clearly directly af fects the cladding 6
dimensionst it aff ects the amount of radiation heat Y
transf er into and out of the cladding it affects the 10 amount of conductive transf er in ano out of the cladding, 11 which is also
-- gap heat transf er is part of tne f actors 12 of transfering heat from the fuel to the cladding; and very 13 importantly is a metal / water reaction.
14 Another important aspect of the appendix K is the
()
15 requirement to calculate in a conservative manner the degree 16 of metal / water reaction, both inside and outside of the 17 cladding at the point of rupture.
la Going back to the right-hand side, then, blockage 19 directly aff ects the flow areat in fact, that's exactly 20 w ha t it is.
In addition, each model must have a model for 21 diversion of flow around the blocked areat that is, when 22 you have a blocked region the flow is going to be reduced 23 during the loss of coolant accident, and that must be 24 accounted s at in flow diversion model 25 In addition, any eff ects on heat transfer 1375 163
4002b4 11 kap /MM 1
correlations or heat transf er models must also be taken into 2
a ccount.
Those three asterisks there are only to mention
(
3 t ha t in certain cases, in particular the reflood period of
'T 4
the accident, when the flood'.ng rate is less than one inch 5
per second, there is even f urther definitions of how the 6
analysis should be handled.
7 Those things, then, results of the strain, 8
results of the flow area, flow diversion and heat transfer, 9
are part of what goes into calculating the perf ormance 10 parameters in each model.
And of course the important 11 perf ormance parameters that we are most concerned about in 12 the loss of coolant accident are the peak cladding 13 temperature during the accident and the extent of 14 oxida tion on the cladding.
Both of those will affect
()
15 whether or not and how the fuel will fail.
16 MR. DENTON:
Ralph, why don't you turn to the Oak 17 Ridge test and their interpretation?
le MR. MEYER:
Okay.
First, I want to comment on 19 what has not happened, just to add some emphasis.
We are 20 not in posse ssion of new tests f rom Oak Ridge that show 21 surprising results that are diff erent f rom results that we 22 have seen before.
We simply have more data, better data, 23 than we had before.
We are at a new plateau in our 24 understanding of the behavior of zircalloy during this 25 a ccioent and have, during the past six months or so, been i375 L64
140.02 05 I2 kap /MM i
analyzing those data in order to produce an evaluation this
('
2 summer -- which has been delayed until now --- in order to 3
compare with the current vendor models in the a pproved ECCS 4
codes.
)
5 Furthermore, le t me point out that the graphical 6
correlations which I am about to show are the results of our 7
own independent analysis of these data, and have had very 8
little peer review.
It was, in f act, in the process of 9
ge tting peer review and investigating the importance of some 10 of our inclusions, that we made the public announcements 11 that we have to when we go outside for information.
12 Now, to begin, let me show the first slide.
13 (Slide.)
14 And I'll explain two terms that we will be using 15 for the balance of this discussions that is, strain and 16 blockage.
This is the cross section of one of the Oak Ridge 17 test bundles in a plane where the blockage was rather large 18 and it shows several things.
19 It shows f our rods that actually burst in this 20 planes it shows a number of other rods that didn't burst.
21 The strain that we will talk about is the deformation of the 22 rods at the point of burst, so if you look at one of those 23 ruptured rods, the circumf erence is larger than the original 24 circumference of the two.
And if it's stretched 50 percent, 25 then the strain would be 50 percent.
- 3/$ ins
40 02 06 13 kap /MM 1
The blockage, on the other hand, is a geometrical 2
measure of the spaces in be tween the f uel rods, and how much
(]
3 they havs been closed down by the swelling of the fuel rods 4
during this ballooning and rupture event.
'])
5 The next slide --
6 COMMISSIONER AHEARNE:
The numbers that you have, 7
are they the percentage of the blockage or something like 8
that?
9 MR. MEYER:
Yes.
These numbers are the strain of 10 the rod in this plane.
11 CHAIRMAN HENDRIE:
The lower number, I take it.
12 MR. MEYER:
Yes.
One, two, three and four are 13 just the locations.
35 is the rupture strain for that rod.
14 Look down below at rod number fives that is not the rupture
(
15 strain but t hat is the strain in that plane right there.
16 The rupture strain would be a larger number.
It ruptured 17 but it ruptured at a diff erent elevation.
16 COMMISSIONER AHEARNE:
Oh, I see.
19 MR. MEYER:
And that's an important distinction to 20 make between rupture strain and. blockage.
The relation 21 between rupture strain ana blockage is not one to one.
It's 22 not completely straightforward because the ruptures don't 23 line up in a coplanar manner.
24 COMMISSIONER GILINSKYr What does this represent?
25 I notice it says computer simulation.
I375 i66
14 40 02 07 kap /MM 1
MR. MEYER:
Yes, the Oak Ridge tests were analyzed
~S 2
by -- I think these were three f oot --- Dale, is that right?
(J 3
T hr ee f oo t -- they were bundles about three f eet long.
And 4
after they were tested, they were tested in steam, they had 5
a heating arrangement tha t simulated f uel pellets and the 6
store energy and decay heat power in the rods.
And they 7
were tested in flowing steam under conditions as close as 8
could be mocked up to be expected in reactor conditions Y
and the rods swelled and ruptured.
10 COMMISSIONER BRADFORD:
When was this done?
.11 MR. MEYER:
These kinds of tests have been going 12 on f or years.
This particular bundle, B-1, was done in 13 1978. There have been thr.ee of the multirod bundle te sts at 14 Gak Ridge so far.
This is the first of the three.
h (J
N 15 COMMISSIONER GILINSKY:
Where does the computer 16 simulation come in?
17 MR. MEY ER :
The computer simulation is simply --
16 what they did was, af ter they ran the tests and then did 19 some flow tests on the deformed bundle, they put the whole 20 thing in epoxy and sectioned it at many elevations.
And 21 simply had so much da ta that they ran it into a computer to 22 find the plane of maximum blockage, and other 23 characteristics of the blockage as a f unction of axial 24 po si tion.
And the computer simply drew the pictures.
The 25 photographs aren't as easy to --
13/$
167
40 02 08 15 kap /MM i
COMMISSIONER AHEARNE:
This is supposed to be an 2
accurate representation of the actual experimental data?
(]}
3 MR. MEYER:
That's correc t.
That's correct.
4 COMMISSIONER BRADFORD:
And these were in all
(~}
5 respec.ts equivslent to power reactor f uel --
6 MR. MEYER:
No, but in as many respects as is 7
practical in an out-of-pile test.
There are in-pile tests 8
scheduled.
Research has one program that is planned and 9
funded for the near future.
Another program is planned a 10 li ttle f urther out.
As you go to more and more realism the J1 tests become much more expensive. You heard an estimate of 12 the cost already, just doing mostly out-of-pile tests.
13 So I don't want to imply that the work is all done 14 and that we know all of the final answers.
But we certainly
()
15 know a lot more than we knew five years ago.
16 COMMISSIONER GILINSKY:
What ha ppens to the 17 pellets as these rods balloon?
IS MR. MEYER:
Well, these rods had ceramic-coated 19 heaters inside.
20 COMMISSIONER AHEARNE:
In a real case.
21 COMMISSIONER GILINSKY:
What would ha ppen in a 22 real fuel rod?
23 MR. MEYER:
Well, I know the pellets --
24 COMMISSIONER GILINSKY:
Would they just tumble 25 down?
i3[b lbb
40 02 09 16 kap /MM i
CHAIRMAN HENDRIE:
You might lose a couple of
. s 2
pelle ts at a substantial blowout, like one, four and seven U
3 on this plot.
But most of the rest are clamped in the (T
4 cladding, aren't they, by mechanical direction?
5 MR. MEYER:
The large def ormation with rupture in 6
it is localized over an axial length of just a couple of 7
centime ters.
And so you might lose a f ew pellet f ragments.
8 But by and large --
9 COMMISSIONER GILINSKY:
I was really asking 10 whe ther the heaters are a reasonable mock-up and you're li saying they are ?
12 MR. MEY ER :
They're the best we can do.
You might 13 find some f ault with them, but they do mock up the heat flow 14 characteristics much better than some of the test
(>
15 arrangements, for example, where you would pass an electric 16 current directly through the zircalloy tube in order to --
17 MR. MATTSON:
The subject of the kind of 18 simulation that you use in heating has been roundly 19 discussed for a number of years. There have been a variety 20 of ways, all the way f rom skim heaters to internal heaters 21 to radiant heaters to all kinds of things, with these 22 pressurized rods.
23 COMMISSIONER BRADFORD:
Are the se pre ssurized 24 rods?
25 MR. MATTSON:
Yes,,r:'se are pressurized rods.
i375 169
40 02 10 17 kap /MM 1
That's what makes them --
2 CHAIRMAN HENDRIE:
I t seems to me back in '73 '74, C) 3 when ene was trying to form appendix K and the 4
recomraendation finally came out of the rulemaking that when
{1 5
one looked at wha t experiments there were.then, on rod 6
ballooning and rupture, that it seemed to make a lot of 7
diff erence the way in which the experiment was set up, and 8
that there were a number of experimental results which were 9
judged not to be sufficiently applicable to power reactor 10 conditions, so that you had much competence in using the li results of those particular runs.
12 I take it tha t in the intervening time -- I guess 13 I've lost track -- that we think we understand a little 14 better how to set up test conditions.
I trus t that's the h
15 case.
16 MR. MEYER:
Tha t is true.
In fact, the data base 17 tha t we use to derive the correlations I will describe, were 16 selected with several criteria; one, the test had to be run 19 in an aqueous environment.
Some of the earlier tests were 20 done in inert gases.
The tests had to use internal indirect 21 cladding heaters because the indirect heat flux modes are 22 different if you use other hea ting methods.
So we did use 23 criteria to protect against the causes of these aberrations 24 tha t we have seen in the past.
25 CHAIRMAN HENDRIE:
I can remember external heating 13/$
170
40 02 11 18 kap /MM i
with heat lamps, high intensity heat lamps, and pneumatic 2
loadings.
It was just a wide range of results and it was g
3 clear that not all of those experiments could be used in 4
compiling the correlations.
5 Please go ahead.
6 7
0 i375
!71 10
.11 12 13 14 O
i5 16
,j C'
17 18 lv 20 21 22 23 24 m.
25
40 03 01 19 mgcMM i
MR. MEYER:
The next slide will show --
2 (Slide.)
i
(-
3 This slide shows one of the correlations.
Now (3
4 this is the first strain correlation.
I have to define a 5
few more terms.
Norm mentioned that there was a heating 6
rate eff ect, and there is a heating rate eff ect.
So we have 7
separated the data into two general dif ferent heating rate 8
effects.
9 We call one slow ramp and one f ast ramp.
Slow 10 ramp is generally below about 14 degrees C. per second, and
.11 f ast ramps are higher than that.
12 We also drew a number of curves that we called 13 composite curves which are the envelope of both of those.
14 So on the top of all of these slides you will see either
(~
./
15 composite slow ramp or f ast ramp.
I haven't shown -- I 16 don't have all of the figures in this package because there 17 are about 50 or 60 of them.
I have selected the ones I 16 think are the most critical and illustrate the f eatures, so 19 you will see that word " composite" changed to other things 20 during the presentation.
21 The black line here is the envelope of the two 22 correlations that we recently developed for circumf erential 23 strain, and it's shown here as a function of rupture 24 tem pe ra ture.
And also overlayed on this is a plat of the 25 present Westingnouse strain model that's used in the ECCS i3/S !/2
20 40 03 02 mgcMM 1
analysis.
2 You can see from this figure, if you don't probe 3
any deeper, that there are several regions of a pparent
(~N, 4
substantial underprediction of the Westinghouse curve.
This 5
appearance -- this situation was apparent to us on many or I 6
gue ss I could say most of the vendor models earlier this 7
week even, and I will show you later why some of this 8
apparent discrepancy is not relevant.
9 COMMISSIONER AHEARNE:
To make sure I understand 10 the significance of if that discrepancy were to occur is 11 that in reaching a given temperature, the experimental da ta 12 would say there is a much greater strain, much greater 13 swelling, therefore much greater blockage?
14 MR. MEY ER :
That's correct.
The blockage curve is k-15 on the next slide.
16 (Slide.)
17 Now here we have plotted the blockage curve in 18 terms of something we call an engineering hoop stress, 19 which is simply the measure of the dif f erential pressure 20 across the cladding wall, and we have expressed it in stress 21 so we can eliminate some design variables and compere all of 22 the vendor models on the same basis.
23 We could have shown this as a f unction of 24 temperature because there is, in fact, the rela tionship 25 between rupture temperature and rupture pressure.
Here 1375 173
40 03 03 21 mgcMM i
again, you can see regions of apparent underprediction of 2
the Westinghouse model.
And again, I will come back to the
'~
3 Westinghouse. curves with a later slide and show why that, in 4
the end, is not very importan t.
(~}
5 Now what I plan to do then is to go through, 6
vendor by vendor, and show the vendor strain and rupture 7
curves compared with our strain and rupture curves which we 8
believe are the best representation of the data that will 9
satisfy Appendix K, so it's an indirect comparison of the 10 da ta,
11 (Slide.)
12 This shows in the black line our fast ramp 13 correlation for assembly flow blockage, and the dashed line 14 is the Babcock & Wilcox approved model.
We now know from 15 the discussions with our Analysis Branch and with the vendor 16 yesteroay that for B&W, and in fact for all of the PWR 17 vendors, tha t the region of application of these curves, 18 that is where the burst would occur, and 19 temperature / pressure diff erential is very limited.
And it's 20 not spread out over the whole region f or which we have data.
21 COMMISSIONER AHEARNE:
Could you explain why?
22 CHAIRMAN HENDRIE:
And that's the four to eleven?
23 MR. MEYER:
Okay. I'll try.
And, Norm, you'll 24 have to help me with this explanation.
25 The tem pe ra ture -- there is a correlation between 13/5 174
0 03 04 22 mgcMM i
rupture temperature and rupture pressure.
If you have a 2
certain temperature on a cladding and you reach a certain
[])
3 dif f erential pressure, the cladding will rupture.
It will 4
swell no more af ter it ruptures and vents its gas.
(]}
5 Pressure -- internal rod pressure and temperature 6
are not independent variables.
In f act, tne internal rod 7
pre ssure de pends on the cladding temperature and on the 8
eff ective temperature of the plenum volume in the f uel rod, 9
so there is that f urther rela tion which has the eff ect of --
10 COMMISSIONER AHEARNE:
Does it also depend on how 11 long the temperature has --
12 MR. MEYER:
Has been maintained?
Yes.
But that 13 effect is handled in the ramp rate eff ect which we have 14 taken account of explicitly, although not with great
-)
25 precision.
(
16 DR. MATTSON :
And the internal pre ssure of the 17 fuel rod is a function of the fill pressure at normal 18 operations and changes as the temperature increases.
As the 19 temperature changes, the internal temperature changes, the 20 initial pre ssure at operating conditions is a design f eature 21 of the fuel.
So f or a given f uel design and a given 22 characteristic of a reactor in responding to a loss of 23 coolant accident, there is a narrow range of conditions over 24 which rupture can occur, determined by the characteristics 25 of the design, which I've called the initial conditions.
1375
!75
40 03 05 23 mgcMM 1
COMMISSIONER AHEARNE:
You are not here saying 2
then that even taking into account the cladding shif ting to
,C 3
a plastic range, you had taken that into account.
There 4
isn't a greater plastic rangt than you had expected?
{'
5 MR. MEYER:
I would say that the super-plastic 6
peaks that you ref erred to do appear to be larger now than 7
we thought they were, but the difference is in degree, and 8
it's. night like night and day.
If you were to characterize 9
our new correlations compared with correlations that those 10 of us who did this analysis would have had a f ew years ago, 11 they are larger.
But, in fact, if you compare the 12 correlations that we have recently derived with the earlier 13 rem models, in some cases, the earlier rem moc:als which were 14 derived f rom diff erent data and by diff erent people by
()
15 methods that I don't even -- haven't even f ollowed 16 through -- are quite similar and some times even larger.
17 MR. EISENHUT I think it's important to cealize 18 two things f rom this.
One is, you can see if you compare 19 the curves, there's obviously regions where the, approaches are 20 conservative and non-conservative.
That's the reason, in 21 fact, earlier in the week we in f act considered there was a 22 potentially significant problem.
23 Now we find the more regions of interest, because 24 each -- af ter yesterday's meeting we found the specific 25 value of fission and hoop stress that corresponds to a i375
!76
40 03 06 24 mgc MM i
plant, and each plant has its own specific value, depending 2
on its own physical parame.ters, its operating
(
3 characteristics, et cetera.
4 And we found for all operating plants the values
('
5 f all between four and ele ven.
So f or the B&W situation, the o
B&W curve previously already overpredicted this new --
7 COMMISSIONER AHEARNE:
What is your uncertainty on 8
that?
9 MR. LAUBEN:
I can't say it in terms of stress, 10 but we have -- Exxon is the only one we had doing 11 uncertainty analysis, and I believe the number comes out to 12 be plus or minus about 100 psi, some thing like that, 13 which -- you see, you would multiply 100 by about 9, is that 14 wha t it is?
I think it's 9.
(
15 COMMISSIONER AHEARNE:
What I'm concerned with is 16 you are using tnis -- see, your i ssue is you are now 17 bounding, and you're bounoing it in a certain region, and 18 I'm really interested -- is this four plus or minus two?
lY MR. LAUBEN :
It's about plus or minus one.
As a 20 matter of f act, the Analysis Branch believes now, on the 21 basis of this, that we need to account f or these 22 uncertainties in everybody's analysis.
23 Exxon already does, and we are going to a ssure 24 that either the curves will bound it in a certain way by 25 making sure the valleys are narrow enough, or that they do 1375 177
40 03 0'7 25 mgcMM i
an uncertainty analysis to assure that these other
.s 2
uncertainties, that we are not le tting some outliers k
3 t hrough.
4 DR. MATTSON:
It will depend, I think, on what
~
5 model you eventually decide they ought to use -- whether o
they ought to.go to the new model, as the Fuels Branch has 7
decided in here, or stick with their present model.
If B&W, 8
in this example, stuck with their present model, then they Y
are higher than the new model for significant dif f erences.to 10 the distances to the right or lef t and can handle the 11 uncertainty.
And you will see in some others, that isn't 12 true.
There would be a question that would be addressed.
13 MR. EISENHUT I was going to say, the other part 14 of it is, if in fact the range fell right at the crossover
()
15 point, beyond that there were diversions greatly, you 16 in fact would want to certainly look more.
17 MR. MEYER:
The next slide shows the B&W burst 16 strain curve.
IV (Slide.)
20 And here in f act is a region within the 21 amperature range of interest between about 80'J and 900 22 degrees C. where the present B&W curve appears to 23 underpredict what we believe the data looks like, and the 24 underpredic tion is significant from a strain point of view.
25 However, in the B&W analysis, the peak cladding 1.5/$
i78
40 03 08 26 mgcMM i
temperature is much more sensitive to the blockage than 1t
,3 2
is to the rod strain, and, in fact, if B&W were to take the iJ rod strain from our curve and reduce their blockage down to 3
r"S 4
our curve, their blockage curve is higher, they would lower
,,/
5 peak cladding temperatures.,
6 If they didn't reduce their blockage curve but 7
just increased the strain, they still would be within the 8
50.46 limits.
9 COMMISSIONER AHEARNE:
Why did they end up with a 10 much greater strain but much less blockage?
11 MR. MEYER:
We said earlier that you couldn't get
, [7 12 directly from strain to blockage and, in fact, the early c.
13 modeling a ttempts treated the strain and blockage 14 separately.
We looked at bundles and model blockage, and I
15 you looked at single rods and model strains.
Those were j
16 data with heating techniques that were diff erent than we 17 would like to rely on now, so what we have done is developed 18 a method to g e t -- to actually get from the single rod 19 strains to the blockages based on the small number of good 20 blockage tests that now have been done.
21 COMMISSIONER GILINSKY:
Could you explain once 22 again why strain and blockage are not directly related?
23 24 i375 179 23
27 40 04 01 pv MM 1
MR. MEYER:
They are related, but the relation is
(])
2 not direct, because the rup.ture strains do not all occur in 3
the plane of maximum blockage.
If you run a test bundle 4
under as prototypical conditions as you can, the rods burst
(' )
5 at diff erent elevations: they don't all burst at the same 6
e le va tion s.
You cannot take the burst strain, which is the 7
largest strain at the location of the burst, and you can't 8
just line those up to form the blockage; you have to know 9
some thing about axial distribution.
You can make 10 assumptions, like they are randomly distributed, or you can
.11 make other assumptions, to arrive at a model that will let 12 you convert rod strains into blockages.
13 What we did was to look at the three good bundle 14 tests that have been done at Oak Ridge in the last year and
()
15 a half, and we ratioed the rod burst strain to the blockage 10 in the plane of maximum blockage.
And we found that tha t 17 ratio was nearly constant for the three different bundles 18 even though they ruptured under different condi tion s.
And 19 so we used that ratio as the means f or ge tting f rom burst 20 strain to blockage.
21 It's a fairly complex step, but it's a very 22 reasonable step to take.
23 DR. MATISON:
I think you can also see it on the 24 first page of thi s handou t.
If you will look at the four
(,
25 rods that are burst in this particular plane, the geometry i375
!00 0
40 04 02 28 pv MM 1
of the burst rods are all different.
'N 2
(Slide.)
3 Fur thermore, the strains are slightly differents rN 4
30, 35, 35, and 36.
So, the aucunt of strain at burst and 5
the configuration of the rod at burst are not only a 6
f unc tion of axial location up and down the length of the 7
bundle, as Ralph has explained, but they are also subject to 8
minor variations in even microstructure of the cladding.
9 It's a random process when it gets down to this fine 10 metallurgical structure, and the exact configuration, the
.11 exact amount of strain at burst will vary randomly f rom rod 12 to rod.
13 COMMISSIONER AHEARNE:
I gue ss what I was 14 concerned with is wondering whether B&W's model wasn't
()
15 predicting significant le ss deformation.
16 (Slide.)
17 MR. MEYER:
The strain model does in that one 16 region, yes.
But the peak cladding temperature analysis is 19 not a very strong f unction of the strain; it's a stronger 20 f unction of the blockage.
And the overprediction on the 21 blockage in the B&W analysis seems to clearly dominate the 22 calculation.
23 DR. MAITSON:
That's because the strain influences 24 the heat transf er area predominantly, whereae the blockage 25 influences the channel dimension, hence the flow and flow i375 iRI
40 04 03 29 pv MM i
rate and flow quality.
So, one is a much stronger influence
()
2 than the other, so you account for both of them in the model 3
change, and the expectation would be that the peak cladding
(']\\
4 temperature would come down.
5 MR. EISENHUT:
If you took both of these 6
considerations together.
7 DR. MATTSON:
Yes.
8 MR. MEYER:
Okay, let's move on to the next slide.
9 (Slide.)
10 Now, this is a slide for Combustion Engineering.
11 The dashed curve is a proposed model that Combustion 12 Engineering has before us, along with some analysis they
~
13 have done on that model.
It is a diff erent model than is 14 currently in their approved code, but neverthele ss it's one
(
15 that has been studied and, in the range of interest, is in 16 good agreement with the current curve.
And in fact, the 17 dashed line here is the curve that we prescribed to lo Combustion Engineerine about 18 months ago for the pur po se 19 of that analysis.
So, you can view it as an early version 20 of our present curve.
That's the same curve.
21 (Slide.)
22 The next slide shows the CE blockage.
This is the 23 fast ramp blockage curve.
The hoop stress region of 24 a pplication f or the CE plants is f rom three to nine KPSI and 25 for f ast ramps the CE curve is clearly predicting more than i375 182
40 04 d4 30 pv MM i
the current f ast ramp curve in the range of interest.
p 2
COMMISSIONER AHEARNE:
You mentioned in the 3
previous that you had directed them about 18 months ago --
4 MR. MEY ER :
Yes.
f3
~
5 COMMISSIONER AHEARNE:
And you said it's what we 6
prescribed to them.
Is that then this predictions is that 7
also part of what we had been prescribing to them?
8 MR. MEYER:
Would you repeat that?
9 COMMISSIONER AHEARNE:
What I interpreted you.to 10 say i's we had prescribed to them either data that must be 11 matched or a model that must be used.
12 MR. MEYER:
Yes.
13 COMMISSIONER AHEARNE:
So, now I am asking is that 14 same remark valid for this prediction?
(')
15 DR. MATISON:
Did you prescribe this curve for 16 Combustion Engineering in the same way you had prescribed 17 the curve on the previous slide?
18 MR. MEYER:
Oh, yes, yes, that's correc t.
19 COMMISSIONER AHEARNE:
Okay, then I would get back 20 to whether you did a similar prescription f or Westinghouse 21 at a later stage.
22 MR. MEYER:
Okay, I can answer it now.
We did 23 not, because the only curve -- the only strain and blockage 24 curves that a ppeared to be out of line 18 months ago was the 25 CE model because at tha t point we hadn't refined our own i3/S 183
40 04 05 31 pv MM i
correlation, and the CE curve at that point looked clearly
[])
2 to underpredict any reasonable prediction of -- any 3
reasonable correlation of the data.
And we went ahead at 4
tha t time and required them to do analysis with new curve.
()
5 DR. MATTSON:
As you will see, there is some 6
variation among all of the five fuel vendors and goodness of 7
fit relative to our current data.
At the time we required 8
Combustion Engineering to change, they were the worst of the 9
lot.
Today that isn't true.
Another vendor is the worst of 10 the lot, but still pretty good at this point.
Il COMMISSIONER BRADFORD:
When you change a 12 requirement like that, what does that mean in terms of the 13 operating plants at that time?
Are you simply requiring the 14 changes in the analysis, or does that result in an actual 15 change in fuel rods?
16 MR. MEYER:
In this case, with the analysis with 17 the improved curves, they showed that the met the 16 temperature and oxidation limits of 50.46 and there was no IV change indicated then f or the o perating plants.
20 DR. MATTSON:
That is a question we asked 21 ourselves.
The original change in the model by us to 22 determine conformance with Appendix K -- that is, the point 23 at which we reached a decision we had to require a change 24 based on curve understanding of the inf ormation -- ac tually 25 o ccurred in, I believe, a construction permit, the Perkins i3/5 184
40 04 06 32 pv MM i
case, if my memory is correct.
2 Simultaneously, as we were reaching those 3
decisions and doing those calculations, we asked the c T 4
question of what does it do to the operating Combustion 5
Engineering reactors to make this kind of change and 6
received representation f rom Combustion Engineering that the 7
kind of analysis that they had done bounded those other 8
conditions, so there was no need to move on the opera ting 9
plants.
10 The analysis that was done to demonstrate 11 conformance with A ppendix K incorporated the conditions a t 12 the plants.
It was a question we asked, and the answer was 13 such that we did not have to move on the operating plants.
14 MR. LAUBEN:
I think f t should be made clear about m
(s/
15 that process.
Not only did they change the curve f rom the 16 strain curve that was not shown here, some lower value which 17 we f ound unacceptable, but in addition to that they 18 requested some compensating changes f or us to review and 19 approve.
And we did review those and have asked Combustion 20 Engineering a lot of questions about those compensating 21 c hange s.
And although we haven't completed the paperwork, 22 we are on the verge of approving those compensating 23 c hanges.
24 It's a similar process to what ha ppened to 25 metal / water action in the area of Westinghouse.
When you 13/S i85
40 04 07 33 pv MM i
correct the error, they couldn't meet the requirements of
([
2 50.46.
3 If they had gone only to our recommended strain
}
and blockage curves, Combustion Engineering would not have 4
r(
5 been able to meet the requirements of 50.46.
So, they need 6
to request some compensating model c hanges, which, of 7
course, we had to be sure were all within the scope of 8
Appendix K.
9 I don't know if tha t hel ps --
10 MR. MEYER:
Those model changes were in a steam
.11 cooling model.
12 COMMISSIONER BRADFORD:
Does that mean there was a 13 period of time within which they did not meet the 14 requirements of 50.46 until you had approved the changes
('
15 that they had requested you to consider?
16 MR. LAUBEN They were able to do it f airly 17 rapidly, and I am not sure what time scale might have been 18 in between.
We never saw an analysis with only the strain 19 curves changed.
I can't say for certain how long it took 20 him to do it to make the compensating changes, and I am not 21 even sure whether they made analysis along the strain curves 22 change.
23 DR. MATTSON:
I am going to have to go back and 24 review the documents to make absolutely certain of what I am 25 about to say, bu t i t is my recollection it went something i3/S 186
40 04 08 34 pv MM i
like this You make the change in the swelling and rupture 2
model.
You are told if a compensating, offsetting feature
[])
3 of the model that vill keep it within 50.46, having not 4
reviewed in detail all of the nuance of that compensating
(]}
5 f eature, we were able to reach a technical judgment 6
relatively shortly that it looked like a good f eature, that 7
it looked like we could approve it, that we needed to go 8
through the numbers in some disciplined longer-term f ashion, 9
but that in principle we approved the compensating feature.
10 So that the two were laid on the table side by 11 side, judged to be legitimate under Appendix K, and when the 12 two were combined they produced a peak clad temperature 13 lower th6n 2200 degrees, and so a nonconformance of Appendix 14 K did not exist.
kmJ 15 The full documentation, SER, what have you, on the 16 compensating features is the thing that Norm says is nearly 17 complete, and it's come out the way we expected it to come 18 out, that they did demonstrate af ter a round of questions 19 and answers completely to our satisf action that the f eatures 20 were within our requirements.
21 COMMISSIONER BRADFORD:
Can you explain in 22 layman's terms where these compensating f eatures come f rom?
23 Why are they sort of lying around there just to be drawn on?
24 DR. DENTON:
Let me say this some features we 25 have beer. toa the to change the basic Appendix K evaluation i3/S
!87
40 Q4 09 35 pv MM I
model until more tests were in and we had a peak overview.
2 You may recall we were cited recently by someone
({}
3 for f ailure to adopt the new ANS decay heat curves which 4
s how that the amount of fission heat generated af ter
(])
5 shutdown is actually lower than we assumed it would be.
6 That's one that we have set aside and have not permitted 7
licensee to incorporate because we thought it wasn't saf e 8
until we do a complete modification of Appendix K.
9 But there are some which are more technical that 10 we do permit, and I will let Norm answer those types that 11 might be in people's minds.
12 MR. LAUBEN:
Could you go back to the very first 13 slide again?
14 (Slide.)
()
15 The important parame ters slide.
Thank you.
16 In particular, the ones that Combustion changed 17 were the ones that have stars by them as compensating 18 changes, and in fact the flow area change that they made was 19 to. increase the blockage.
So, in a sense, that was not a 20 compensating change.
21 But the way that they proposed to treat flow 22 diversion, they actually need to increase the amount of flow 23
-- decrease the amount of flow area in the subchannel 24 analysis, so we said, "That's fine.
There's nothing wrong 25 with that."
They did do a more realist.ic calculation of the
- 375 l88
40 04 10 36 pv MM i
flow numbers.
In thermal hydraulics.there are simple ways 2
to do things conservatively.
They did a more realistic
[]}
3 analysis.
Not yet, I would say, a completely realistic 4
analysis.
There is still some conservatism in that flow
()
5 conversion calculation.
6 In addition, they made some modification to the 7
heat tran sf er.
That wasn't as substantial a change as in 8
flow diversion, but there was a slight modification to 9
. increase the amount of heat transfer, which we f elt was in 10 the range of what one would expect the heat transfr.' under 11 those conditions.
So that in the case of Combustion, the 12 particular compensating changes were very closely related to 13 the blockage and stress models.
14 COMMISSIONER GILINSKY:
Do these sorts of things
(_)
15 get run by the ACRS?
16 MR. LAUBEN Yes, they do.
I don't remember when 17 we have last discussed this with the ACRS.
Maybe I had 16 better say that this particular one, we didn't.
Obviously, 19 the metal / water reaction, we did.
20 COMMISSIONER AHEARNE:
When do you intend to 21 discu ss --
22 DR. DENTON:
Normally, these are discussed during 23 the next ACRS meeting.
We have a se ssion whereby we talk 24 about the changes that have gone on in the past mon th.
25 MR. MEYER:
This is scheduled, I hear, this coming 1375 I89
37 40 04 l1 pv MM i
Wednesday morning for ACRS.
2 COMMISSIONER AHEARNE:
Next Wednesday?
3 MR. MEYER:
That's correct.
4
(
5 6
7 8
i3/S 190 9
10 11 12 13
(
14 15 16 17 18 lY 20 21 22 23 24 25
40 05 01 38 gshMMM 1
MR. LAUBEN:
I should point out in all of these
()
2 analyses, I think in particular because of our rather 3
stringent requirements on metal-water reaction, it turns out 4
tha t the analyses and temperature ranges, oh, say between 5
2l00 and 2200 degrees, but above 2000 degrees, are rather 6
sensitive to many changes.
7 And there's no -- well, let me say that there's no t 8
complete unif ormity by any means among the models that each 9
vendor has.
10 They all may have these f eatures, but there's not a great 11 deal of uniformity.
12 For instance, one model may have an extreme sensitivity 13 to flow diversion.
Another model may not.
14 MR. DENTON: These models, I should mention, all m
15 the licensing models -- you know there's another set of 16 models that you hear Research talk about, which I guess are 17 called "best estimate" models.
They are on a track which 16 calculate lower temperatures than through the so-called 19 licensing models.
20 COMMISSIONER BRADFORD:
When the code models are 21 accepted for licensing purposes, you assume that they are 22 accurate, conservative, but at least internally consistent?
23 MR. MATTSON: We don't a ssume its we require them 24 to demonstra te it.
25 COMMISSIONER BRADFORD:
All right.
Now, again,
!3/$ i9i
40 05 0'2 39 gshMMM i
through what proce ss i s i t that these compensating features 2
become available if they were all.right in the first place?
~'
k.-
3 MR. LAUBEN:
Well, let me try to put it this 4
way. Very often, the vendor, I think, understandably, will 7(s 5
want to expend the amount of energy and time in the model 6
that would make it reasonable and acceptable, and still pass 7
the requirements of Appendix K and have his plants still 8
meet the requirements of 5046.
9 If he wants to use a simple model, it's our policy that 10 the model needs to be conservative, demonstrably
.11 conservative.
12 And that's what he'll do.
And it's easy to pass on such 13 models as that.
14 And usually, they are f airly straightforward and not too (j
15 difficult to evaluate if it's that kind of thing.
16 A more difficult part is if he Ls trying to propose a 17 more sophisticated model, in which case we have to spend 16 much more time evaluating sophisticated models because 19 generally, it means that they are closer to reality.
20 And it's quite possible that they still have -- bo th 21 models may have conservatism.
One is going to be closer to 22 best estimate and somewhat more sophisticated and requires 23 more review time.
But it doesn't mean that the new model, 24 albeit, more sophisticatec and not quite as conservative, 25 still doesn't mee t the requirements of the rule or the
!3/S I92
40 05 0'3 40 gsnMMM i
requirements for sufficient conservatism that the staff 2
would regard it formally.
((N 3
MR. MATTSON: There's another reason, and that is 4
science marches on.
With each passing year, you learn a
(
5 little more about thermal hydraulics, about reactors, about 6
Cores.
7 And while there were areas specifically treated in 8
Appendix 1, at the time of rule-making and are specifically 9
constrained by the language of Appendix K, we f eel that 10 we have very little flexibility in what kind of change to 11 allow.
12 An what we're talking about today is a good example.
13 Thou shalt not underestimate the degree of swelling and the 14 incidence of rupture.
It's narrow language that constrains
(/
15 us very narrowly.
16 There are other areas that enter these evaluation models 17 tha t were not so narrowly constrained in Appendix K.
And as 16 Nnrm has explained, one of the reasons those will change is 19 because it's easier to generate and approve a conservative 20 simple moael than it is a non-conservative or realistic 21 complicated model, sophisticated model in those areas.
22 Another reason is that as science marches on in the years 23 since 1974, there may be areas you simply know more about 24 today than when you first put the model together, and you
(
25 hold them.
A vendor would hold them, not meaning to spend 1375 i93
41 40 05 04 gshMMM i
the money to use that advanced information to change his N
2 evaluation model and go through the process of having the t
3 staf f examine it, approve it, and cost him money until he 4
needs it, so to speak.
7 5
So that information goes in a non-conservative direction 6
and he would pull that reserve out of his pocket for new 7
information of a conservative nature and use it.
8 MR. DENTON: This is a subject that we have long 9
debateo internally about revising Appendix K. And I think 10 it's f air to say that the staff has been very chary of 11 granting a pprovals of models that go in the direction to 12 relax things, although they have done it when an applicant 13 has made a good case.
14 MR. MATTSON:
I think we've talked to the
('
15 commission about that bef ore.
The industry objects to the 16 stringency with which we view Appendix K and the language in 17 Appendix K in some areas have changed, the one Harold 18 mentioned being the prime example, and our unwillingne ss to 19 accept new decay information in f ace of a narrow description 20 in Appendix K and how to treat decay heat and our 21 unwillingne ss to change that in Appendix K without thinking 22 of the overall conservatism and all of the elements that 23 make up that conservatism.
24 MR. DENTON:
Let me finish this one.
25 The ANS model almost would be a 20 percent change if we i3/S 194
40 05 05 42 gshMMM i
were to - -10 to 20 percent in the amount of heat generated.
2 We have not incorporated that.
.(
3 CHAIRMAN HENDRIE: Go ahead, Sol, briefly, I trust.
(
4 MR. LEVINE:
I hope so.
I think it's time that we 5
really re-examined Appendiy
- fundamentally., We have over 6
the past 5 years spent multi-millions of dollars on research 7
results which almost unif ormly show very large conservatisms 8
in the evaluation models as opposed to our best estimate 9
models.
10 We recommended this about two years ago to the staff.
.11 There is a program been started which is moving at a truly 12 glacial pace.
13 MP. MATTSON: Before Three Mile Island.
14 MR. LEVINE:
But since then has become zero.
15 (Laughter.)
16 MR. LEVINE:
I think it's unconscionable that we 17 now believe that there is somewhere in the neighborhood of a 18 thousand degrees conservatism in peak clad temperature, as 19 calculated between the two models, and we're doing nothing 20 about it.
21 This whole exercise in which you're struggling with the 22 details of these models is an artifice of Appendix K 23 because the peak clad temperatures we predict now would not 24 get you into the regime of swelling or blockage or anything 25 like tha t.
i375 195
40 05 06 43 gshMMM i
I think we are just -- and we/re alarming the public 2
unnecessarily about these things.
I think it's time that we N.
3 have enough knowledge to start to really in the next year 4
get a new model.
{
5 COMMISSIONER AHEARNE:
You're saying the best 6
estimate models that you have developed give temperatures so 7
much lower than tney're talking about here that this is a 6
non-issue?
Y MR. LEVINE:
I t's a non-i ssue, but the staf f is 10 legally bound to consider it by our regulations, of course, 11 so I'm not saying that they should not consider it.
12 They have to.
But what we have to do is get eff ective 13 action taken to change the regulation in a more realistic 14 direction.
(.
15 MR. MATTSON:
I would.just add to that that the 16 crisis atmosphere with which we seem to approach these 17 repe titive second-order eff ects with Appendix K is in large 18 measure dictated by, of course, the environment in which we 19 operate and the legal constraints upon us.
20 But against a backdrop of what Sol has just described, 21 sometimes it's not worth it to jump through a hoop in 24 22 hours2.546296e-4 days <br />0.00611 hours <br />3.637566e-5 weeks <br />8.371e-6 months <br /> with five major fuel vendors on the conditions of 70 23 operating plants for a second-order eff ect compared to the 24 amount of overall conservatism present there, is highly 25 oisruptive to the peer review process and the legitimate 1375 I96
40 05 d7 44 gshMMM i
source of scientific inquiry and engineering judgment.
2 CHAIRMAN HENDRIE:
I will point out that today's 3
exercise and the events of this week in this ma tter are one 4
of the penalties that one pays for the benefits presumed of
(
5 promptly putting into the open licensing process any piece 6
of new information which comes to the a ttention of the staff 7
ct a point possibly before the staff has had a chance to e
decide whether it is, in fact, a serious matter that needs 9
to enter the formal licensing -- the adjudication process or 10 not.
11 Bu t I must say that the reason we operate in the present 12 mode where these things get sho t to the surface perhaps 13 before they are well understood and then may cause the sort 14 of what I trust will be short-term publicity that's intended (s
15 on this one is becausa at an earlier time, an earlier 16 commission found some things that they thought should have 17 gone forward into the open adjudicatory process more lo promptly than they did.
19 And I think we sort of have to recognize that there 20 probably isn't any ideal way to handle everything.
21 The mode that we are in now, while it does, on occasion, 22 create the sort of enterprise we are now engaged in, does 23 have the benefit of getting those things up rapidly and then 24 if one does turn out to be serious, it at least was noticed 25 all around at an early time.
1375 I97
40 05 08 45 gshMMM 1
But I think the commission recognizes the pros and cons 2
of the situation and whether or not we might want to
, ~
3 reconsider the present mode at some time as something that 4
the commissioners can think about and talk about some time in
(~
5 the future.
6 COMMISSIONER GILINSKYJ I must say as it was 7
described to me yesterday, it didn't sound like a second 8
order effect.
9 CHAIRMAN HENDRIE: Well, I think the reason that it 10 comes up and gets exposed into the process is because it has 11 become clear to one or another element of the staff that it 12 has the potential for not being -- or.the potential f or 13 being a significant matter.
14 And there may be some time of uncertainty.
Then af ter
(
15 that point, before you know whether, yes, it is, or no, it 16 isn't, really.
17 But as I say, we have chosen and are operating now on the 16 basis that those things are to come up as soon as they are 19 recognized.
20 I'm always glad to see circumstances which appear to have 21 been more difficult becoming less difficult as time goes 22 along rather than the reverse.
23 So there's something to be said f or the present situation 24 at t ha t.
25 Are there other questions along the line that we were b ! b k ']0
46 0 05 09 gshMMM i
pursuing here just before this small philosophical recent 2
discussion, and if not, why, I'll ask Ralph to plow on 3
through his other graphs and hope that we can progress t
4 along.
5 6
7 a
1375
!99 v
10 11 l
A 12 83 14
(~/
x_
15 16 17 18 lY 20 21 22 23 24 25
47 While then I would like to turn now briefly to the
- r. 8140 1
1s-1 2
boiling water reactor picture.
-[,
3 (Slide.)
In this case if one forgets that the boiling water
(.
4 S
reactor is really different than the pressurized water 6
reactor, it's possible to get alarmed and think that the curves that GE uses and that Exxon uses for their BWR reloads 7
8 are really out of line.
But in fact the boiling water reactor because the fuel is either unpressurized or pressurized 9
to very low fuel pressures and also because of the nature of 10 11 the ballooning process is slow in that reactor, is a process that produces very small strains and in fact insignificant 12 13 blockages.
So, I've only shown a strain curve here.
(~)
v 14 The ruptures for the boiling water reactors would 15 occur above 960 degrees Cs The slowness of the ramp rate 16 would mean that at these high temperatures the cladding would 17 oxidize before it ruptured and would therefore become brittle I
18 and rupture with smaller strains than the PWR would rupture.
f 19 So, in this range both the GE and the Exxon BWR curves are 20 in fact in good agreement with the strains that we would i
21 predict from our recent review of the data, l
22 I think the next slide is the first of the final 23 series on the Westinghouse -- oh, I am sorry.
We have to go 24 through the Exxon PWR curves briefly.
l m
wel Ruorun, Inc.
l 25 (Slide.)
1 3. / b.
? 0 0
48 I should point out as Norm mentioned earlier, Exxon sis-2 1
is the only vendor for which we have performed a fairly detailed 2
~
3 sensitivity analysis.since they had earlier claimed to be
~
down in the valloy of these strain and blockage curves and we
([
4 5
had lifted that back in late '76 and early '77.
Here is the Exxon curve shown in the dashed line compared with our 6
And"in the 7
fast ramp correlation for blockage on the fast ramp.
range of application in the Exxon PWR reloads the Exxon curve 8
is clearly overpredicting what we think the data are telling 9
10 us.
11 The next curve will be different.
12 (Slide.)
13 This is the blockage curve for slow ramps.
The Exxon curve is the same because these earlier vendor models 14 15 did not recognize explicit ramp rate effects and our recent 16 analysis does.
And it shows a difference in'the location of 17 this shoulder of this peak that is located at around 7 KPSI I
18 in this slide.
i 19 Yesterday when we discussed this with Exxon, they f
20 showed us that one of the convolutions that we had gone l
through in plotting our data actually caused the data points 21 l
22 to shift a little bit.
And when they plotted the data directly in terms of hoop stress on this figure'it showed that the data:
23 did not suggest that the shoulder of our curve was any better 24 ye Reponm inc.
ce 25 than the shoulder of their curve.
and I think that that was a' i375
?0i
49 very fair critique of our modeling of those data and is a 1s-3 1
'h 2
good explanation of why the Exxon curve should not be faulted tv 3
on this account.
(
4 CHAIRMAN HENDRIE:
You can fool with that result 5
to your peer review as to see if they were acute enough to 6
pick it up.
7 MR. MEYER:
The next slide shows a similar 8
discrepancy.
9 (Slide.)
10 At around 950 degrees you can see that the Exxon 11 strain curve is below our strain curve.
Yesterday during the 12 meeting when we in fact laid the data over this we found 13 that the data did in fact go underneath the Exxon curve in CJ 14 that region.
We have simply smoothed out that region by the 15 model fitting that we had selected.
16 COMMISSIONER BRADFORD:
What does that mean?
17 MR. MEYER:
That means on the basis of the new 18 data there is no reason to believe that the Exxon curve is l
I I
19 underpredicting.
l 20 COMMISSIONER BRADFORD: I understand that.
But your 21 curve there is a curve that you use for regulatory --
l 22 MR. MEYER:
No, nc.
This is the curve that has l
i just been developed as a result of our analysis of the world 23 24 supply of data for recent years.
I i
m wsI Reonm, lm.
25 CHAIRMAN HENDRIE:
They have just had pointed out i3/S 102 1
s-4 1
to them that in transfering it --
2 MR. MEYER:
There are certainly small uncertainties
(
3 CHAIRMAN HENDRIE:
-- here what they didn't plot as
(
4 well as they should have.
5 MR. MEYER:
There are certainly small uncertainties 6
and should be revisions made in our curve after we go through a 7
review process.
And in the case of the Exxon curve --
8 MR. MATTSON:
There are an infinite number of ways 9
to overestimate the data said another way.
10 CHAIRMAN HENDRIE:
On to Westinghouse --
Il COMMISSIONER BRADFORD:
On the last two slides 12 then what you're saying it was a result of being in the wrong 13 place?
v I4 MR. MEYER:
It's conservative and I think it over 15 does it.
Yes, I think it may well be in the wrong place.
16 MR. MATTSON:
Although we might choose to leave it 17 there and still allow Exxon to be under it.
The question is j
i 18 according to Appendix K whether the data are underestimated.
l I9 COMMISSIONER BRADFORD:
Yes.
l 20 MR. MEYER:
And even though our curve was in a j
i 21 different place when we laid the data on, their curve doesn't l l
22 appear to underestimate the data.
(
23 Okay, on to Westinghouse.
24 (Slide.)
es eral Reporters, Inc.
25 These are the curves that we started with except now 13/S 203 l
51 1s-5 1
I have laid sn here the window of application for the Westing-2 house curve.
This is the strain curve, the one that in m
s,
3 general is less sensitive performer than the blockage curve, I
4 but curves criss-cross one another and the range of interest 5
deviations here are not the kinds of deviations that we would 6
focus in on as causing significant changes.
Not in the 7
strain curve.
8 CHAIRMAN HENDRIE:
I think the next one is more 9
significant.
10 MR. MEYER:
The blockage curves are more significant.
l 11 (Slide.)
12 And here is where Westinghouse has identified a 13 fairly narrow stress range that is applicable for their m
e
.j I4 plants from five to about eight KPSI and in that range and 15 really only in that range if you look at the high side, do we 16 agree above eight KPSI.
Our interpretation is significantly 17 different from their interpretation.
And on this matter we 18 have a difference of opinion with Westinghouse because they I
I9 said that they didn't care for our model.
I i
20 But nevertheless in the range which appears to f
i
{
21 contain all of the ruptures for the Westinghouse plant, and 22 this is the area that a considerable amount of checking was 23 done overnight, the Westinghouse fast ramp curve overpredicts l
24 or you might say agrees with our curve.
i m
wel Rmorun, Inc.
4 25 COMM.7SSIONER AHEARNE:
How hard have you checked f ) l.')
. UA h
I
1s-6 that narrow range?
In other words, the validity of the 8?
1 2
MR. LAUBEN.
Oh, we have them look at every single V
3 operating plant overnight and they read those numbers over the (X
4 phone to me and they confirm that in their analysis no plant 5
ruptures had a stress greater than 8.2 KPSI.
Is that the 6
question?
7 COMMISSIONER AHEARNE:
Well, the question more is 8
do you do anything to check it?
9 MR. DENTON:
We intend to confirm this.
10 MR. EISENHUT:
That's the same question we asked.
11 MR. MATTSON:
That's the question we were at earlier 12 which is what is the sensitivity of the range?
And it depends 13 how much attention you pay to that depends on what they do out-14 side the range.
In this case with a significant -- where they 15 significantly diverge under us the sensitivity at 8 KPSI is 16 an interesting point which will have to be followed up on.
17 MR. LAUBEN:
I guess I ought to say something about 18 that because it illustrates on different models to make it l
l l
19 different.
We with parts of our own models looked at some of 20 this and the difference between 30 percent blockage and 50 l
21 percent blockage, for instance, gave quite a large difference 22 in the answer of the clad temperature for our models.
Something i
23 like the difference between 2100 and 3300.
That's quite a 24 large difference.
a wat Rmorters, lm.
25 Westinghouse just did a very similar sensitivity 13/S
?05
53 1s-7 1
study when they found that they were in a range for the two
[
2 clads that were in the range of 8 KPSI and they found that the 3
difference between about 40 and 55 percent blockage only
(.
4 amounted to a change in peak clad temperature of about 20 5
degrees.
So, you can see quite a variability in *be sensitivity-6 COMMISSIONER AHEARNE:
If we now went to what Sol 7
was talking about --
8 MR. LhUBEN:
That's correct.
In other words, when 9
we constructed our audit model it was a very easy thing to 10 propose a very conservative flow diversion model and that's 11 indeed what we did.
We didn't even have a mechanistic model 12 as they called it, but rather looked at a large series of 13 calculations with large blockages up to 90 percent and decided s..
14 what we'll just do is bound all those blockages with a single 15 flow diversion curve which resulted in a very substantial 16 reduction of flow, i
17 And this is why our calculation indicated a great i
18 deal more sensitivity than the Westinghouse calculation.
i 19 MR. DENTON:
The two plants at the stream of the l
20 curve I believe were Trojan and Salem.
They were up at the 21 upper extreme of 8.
We looked into those plants in a bit more l 22 detail.
I think we found a peak clad temperature calculated j
23 in the application was on the order of 2150.
24 MR. EISENHUT:
On that order.
=
ni amorms. w 25 MR. DENTON:
So, they had 50 degrees of margin i
13/5
?06
54 sis-8 i
.below the 2200.
(,
2 We also checked on the status of these plants.
3 Trojan is down and will be down for awhile.
And Salem is just i
4 resuming operation.
5 MR. EISENHUT:
It will be down until next Wednesday.
6 MR. DENTON:
It won't be up in full power operation 7
any time soon.
So, just going strictly by the rules of the 8
analysis of the Commission's regulations, neither of them are 9
exceeding the Commission's regulations today and, in fact, Normi 10 estimates I guess that if the actual hoop stress is 8.3 the 11 temperature difference when added to the presently calculated 12 2150 still leaves them below 2200.
13 MR. LAUBEN: 'Right.
It's based on their analysis s
14 that they did last night which says it is only a 20 degree 15 change.
16 MR. DENTON: In order to confirm these we've been j
i 17 dealing with the vendors rather than utilities.
We will issue 18 that as to all the utilities supplying the background informa-l i
19 tion that led us into this concern and they asked them to come i i
20 back and confirm where they are one by one in the hoop stress l
21 and where they fall in these various models.
22 MR. EISENHUT:
Yes.
I think it is important to 23 realize two things.
One is the difference between what we 24 learned and had before us on Monday and what we learned Ct eral Reporters, lac.
25 yesterday.
Was what we really learned -- were the windows of i3/5
?07
interest on the curves.
We didn't'have those windows of sis-9 i
interest on Monday, and therefore you have to say it was a "D
2 VJ 3
potential problem.
()
4 Secondly, if you look at the areas with the wide 5
divergence you might find that you have something all thecway 6
up to 1000 degrees penalty on top of the peak clad temperature.
7 We went through them one by one.
The one that's the most 8
critical because of a very narrow range is Westinghouse.
So, 9
we asked Westinghouse literally to work around the clock to' l
10 make the calculations on the most limiting plants last night.
11 They in fact called their preliminary information in this 12 morning.
As Mr. Denton said there are two plants that are at 13 the 8.2-8.3 range.
Most of the others are down to the order of s
14 about 6.
So, we don't really feel that there is any problem 15 even on the Westinghouse plants.
Particularly, when we consider 16 that the penalty, if there is a penalty, is less than 20 17 degrees up and also recognizing that both plants are shut down 18 and we will see continuing work and also their peak clad i
19 temperature is already calculated to less than 2200.
So, they i
20 could absorb a penalty with essentially no real effect to the i
21 clad.
22 What we have done is we have asked as a result of 23 the meeting yesterday, we have asked each of the vendors to 24 supply to us by the end of the day the information that they o
,eral Reporters, Inc.
25 gave us yesterday.
We ask that it be submitted to us in writing I
i3/S
?08
56 1s-10 and then the next step would be, as we just mentioned, we w uld consider sending letters to licensees saying every
~'
2
\\ J licensee should go through now and reverify that this is in 3
i ].
fact the exact situation in his plant, with any increases of 4
his plant already taken into consideration.
In other words, 5
we should have those letters out by next Monday.
6 So, basically I guess very simply the potential 7
problem that we had and what we saw on Monday and Tuesday we 8
believe is not -- we found unique features in all the plants 9
10 because of the ranges of interest.
That is not a significant 11 problem, with the exception of perhaps two plants that we're i
12 following up on where again we think it is not significant.
nd_t-6 13 v
14 13/5
?09 is 16 l
17 j'
18 f
i 19 20 l
I 21 22 i
23 24 i
,.a_......
25
7 CR 8140 MELTZER t-7 mte 1 1
CHAIRMAN HENDRIE:
Okay.
Let me -- since the
(~'
2 regulatory problem appears to have peaked and receded and 3
become a non-problem, let me inquire, since Sol brought it up,
(['-
4 if we turn to something closer to the real world of reactor 5
safety and not a particular set of regulatory requirements 6
which may be getting a little creaky in their joints as further 7
information is developed over the years -- we do have a sub-8 stantial amount of work going on.
9 There are more LOFT tests to run.
There are sort 10 of separate effect tests still gaing on.
It becomes a 11 question how often one ought to try to go through and refurbish 12 the regulatory side models for these complex accident evalua-13 tions.
(_-
14 I suppose in a perfect world one might do it month 15 by month.
But that changing cf regulations is a fairly 16 complicated and time-consuming process, and one hesitates to 17 do'it very often.
18 Let me ask you, then, do you think what we have in 19 hand out of the works in '74 would justify a rulemaking at 20 this point?
Or should we look for it in a year or two years?
21 That is, is there a sufficient amount of work that ought to 22 be factored into those things to make it worthwhile thinking 23 of that now?
24 MR. LEVINE:
I think that we have enough information m
val Rumners, Inc.
l 25 in hand that if we spent the next year we could formulate a i
l 1r g-ld d
I
mte 2 I
rule that would be an improvement upon this rule, and then
(])
2 have a rulemaking.
We do have to exercise our more realistic 3
codes somewhat better than we now have.
There will be some Q-4 more data coming in from some more LOCA tests involved.
But 3
I think in parallel with that right now we could start tech-6 nical work to formulate a new rule.
It would take about a year to do it, and then we could have a rulemaking.
7 8
CHAIRMAN HENDRIE:
So you think there is enough in 9
hand, certainly, to get started with the formation of a rule?
/
10 DR. MATTSON:
Some time ago we were directed to 11 do that, and some progress has been made in that regard until 12 Three Mile Island.
Since then the Analysis Branch nearly in 13 its entirety has been devoted to post-TMI reevaluation of U,
I#
small break --
15 MR. LEVINE:
I would suggest, because of the staff 16 workload, you might consider another approach, and that is II having the staff direct the work of contractors to make the 18 necessary analyses, to try to gat done faster that way.
19 CHAIRMAN HENDRIE:
That's a possible option.
In any event, I was going to remark that the staff 20 21 is pretty well engaged on Three Mile Island-connected matters, 22 and will be for some time.
So that the efforts of very many 23 staff people along this line are not going to be available, i
certainly in the next few months, at any rate.
So one would i
24 On rel Rmo,wn, im:,
95 be looking down the line next year some time even for a start. !'
l 13/5 H 1
59 mte 3 -
1 Even contractor efforts require a fair amount of oversight on (m
2 a thing like this, where you are going to want to go into a 3
detailed rulemaking, and it is a significant assignment of
(
4 staff resources.
5 DR. MATTSON:
Plus some recently raised uncertainties 6
on human performance and machine performance.
That 1,000 7
degrees of cushion in Appendix K might be a pretty good 8
cushion between now and the time we pick up on some of the 9
human reliability and other reliability factors indicated by 10 Three Mile Island.
II Not only can you not afford it in terms of resources 12 '
and expertise unless you use the laboratories, but I caution 13 they are awfully busy at the moment also.
You might not want k
Id to for the next year.
15 CHAIRMAN HENDRIE:
Well, I thf.nk one doesn't neces-sarily, in effect, hand over the 1,000 degrees to designers 16 17 of plants to raise power densities because one has improved 18 the evaluation models.
It's just that, having what are now I9 h getting to be a bit old and rusty and perhaps a little too 20 rough-cut set of evaluation models and regulations results 21 occasionally in our meeting here.
While it's pleasant to meet 22 on Friday afternoons --
23 DR. MATTSON:
At the risk of belaboring it, there's 24 one more facet of that, and that's this. business of incentive et eral Reporters, Inc.
25 to make productive improvements in things.
And by the agency 137S
?12 i
mte 4 I
remaining inflexible on Appendix K, which I have been arguing 2
for in the last few minutes, I think there is a stifling effect
-s(,'
3 upon improved realism.
That is, improved models for under-4 standing the realistic behavior during transients and accidents.
(
5 And there's a lesson from Three Mile Island about the need for 6
realism and analysis.
7 So there's a balance to be struck there somewhere 8
and it involves all of these factors.
I'm not sure where it 9
comes out.
10 CHAIRMAN HENDRIE:
Well, I certainly am not about 11 to propose to my colleagues that we mandate some vigorous 12 forward initiative on new rulemaking on Appendix K this after-13 noon.
I think the day has ample things to occupy us without
\\s 14 that.
15 Other comments?
Harold?
16 MR. DENTON:
I would like to inject-just one other 17 note, that while I am confident today that we don't have a 18 safety problem in the reactors, based on what we have seen, 19
'they have worked through all this very fast and we do need to 20 confirm it and look very carefully plant by plant at its 21 uncertainties and margins. And keep in mind the speed at which 22 we have had to look today doesn't preclude that we might not 23 find a fuel assembly somewhere that we will want to correct l
l 24 down the line.
I a?.mn nwomn. sne.
25 CHAIRMAN HENDRIE:
Yes, just so.
And I trust you'll i
i i3/S
?l3
mte 5 I
keep us informed, as you have.
2 Peter, I see a comment coming.
'./
3 COMMISSIONER BRADFORD:
Yes.
Can I just get a two-t 4
minute sort of summary on why we would be learning this from 5
these experiments in 1978 and
'9, instead of back at the time 6
of the ECCS controversy or before?
7 MR. LEVINE:
They weren't done.
8 COMMISSIONER BRADFORD:
Okay.
That's the two-second 9
answer.
Why not?
10 MR. LEVINE:
There were programs going on then that were, in my view. underfunded, underplanned, and people were II 12 drawing to just what I would call quick and dirty experiments, 13 which turned out to be inapplicable, as a: matter of fact, as N-14 Ralph pointed out earlier.
It wasn't until the AEC established 15 the Division of Reactor Safety Research in 1973 that we began 16 to formulate any sensible scientific program, that we began 17 to get the experiments going that give us data that is 18 applicable.
19 COMMISSIONER GILINSKY:
When you say applicable, 20 what was then the basis for'the curves in Appendix K?
21 MR. LEVINE:
I guess I don't know the details, 22 because I was not i nvolved.
But the data was -- the data j
from the experiments is not directly applicable, and the 23 i
people in the licensing process made very conservative assump-l 24 rei nemnm. irw.
25 tions to make sure they were bounding reality significantly.
l I
I 13/S
?I4
62
.te 6 4
1 This other people know in more detail than I do.
Roger?
(])
2 DR. MATTSON:
The single rod data was just starting 3
to become available with the kind of aqueous environment that t
4 Ralph described earlier.
But there were data literally all 5
over the map in those days, and there was no common thread by 6
which to correlate them all.
7 CHAIRMAN HENDRIE:
In fact, some stuff came in in 8
very late days of the formulation of the proposed rule, as I 9
remember.
10 DR. MATTSON; Yes, there were people doing experi-11 ments right in parallel with the hearing, as it became obvious 12 that this was one of the more difficult points.
And it was 13 starting at that point and carrying through with these expen-14 sive, complex, multi-rod burst tests at Oak Ridge National 15 Laboratory that this body of knowledge now available began to 16 be generated.
17 COMMISSIONER GILINSKY:
What then formed the basis 18 for Appendix K?
19 DR. MATTSON:
Some of the old not-so-good data, and 20 in fact rejected from the correlation data today, and some of 21 the early data on water -- in water environments, but not 22 multirod bursts, not with well-established heater technology;
(
23 some of it proprietary to the individual vendors; and a fair 1
24 amount of it coming from the initial work at Oak Ridge j
w nel Rmorters, lm.
25 National Laboratory, Phil Rittenhouse and some of the people i
1375
?15 j
I
63 mte 7 1
that worked back in the early 70s, you may recall.
2 CHAIRMAN HENDRIE:
Other comments?
3 (No response.)
~
(
4 CHAIRMAN HENDRIE:
Thank you very much.
e-7 5
(Whereupon, at 3 :05 p.m., the hearing was adjourned )
6 7
13/5
?l6 9
10 11 12 13 C
14 15 16 17 18 19 20 21 22
~
23 l
24 car vel Reporters, Inc.
25 i
l I