ML20128P693
| ML20128P693 | |
| Person / Time | |
|---|---|
| Issue date: | 01/11/1984 |
| From: | Ryder C NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | |
| Shared Package | |
| ML20127B155 | List: |
| References | |
| FOIA-85-110 NUDOCS 8507130298 | |
| Download: ML20128P693 (74) | |
Text
{{#Wiki_filter:_ s ,
-s, - f.n PEER REVIEW-BCL Report on the Radionuclide Release Under Specific BWR Accident Conditions.
28,29 July 1984 Excerpts from the transcript. Prepared by C. Ryder ASTP0/NRC 1/11/84 8507130298 FOIA 050425 PDR PDR ALVAREZ85-110 di
a ,e ,
- s .
NOTE:.The transcripts can be' misleading-because-they are twice removed from the meeting, once because they are written and again because of the time between the meeting and the- transcript. At best, the transcript giv,es~ general indications of issues. INDEX Page
~
.Calulations based on 100% and 50% zirconium reacted M1-3 Solidifying corium and modelling . .. 265 - 6 Comparison of VANESA aerosol output with experiments '270 Comparison of VANESA isotope output with RSS 271 Condition of corium in a reactory cavity and accident sequences 276 Limitations of CORCON. late into an accident sequence 278
- Influence of CORCON output on VANESA output- 279 - 82 Corium temperature _
281 CORCON and modelling of corium 286 - 7 Conclusions about VANESA 20-1 j* e S
i -a , . ,- i ..
- j. . .
a
"[
p
- 7. ..
~
12 I.want to.begin by just reminding you MR. POWERS:
- 13 how the VANESA code fits into this overall calculational 14-scheme for the source term.
18 VANESA is the model that we are using to calculate 16 the fission. product release from source term that comes about 17 because of the melt interactions in the reactor cavity region. 18 Before we can do calculations with VANESA model, 19 we need information from several of the other codes used in 20 the overall calculation procedure. 21' (Slide.) 22 There are, of course, accident and plant 5 The specifications that are .the definition of the problem. M things that,most concern us in VANESA about these are the 5 reactor cavity geometry and the concrete composition. f l
. 1 x
g ._ _
+,
260
,..n..... 1 We also get from the MARCH code and the'CORSOR code 2 :-
what the coremelt looks like as far as its initial mass and
~3 -its- initial temperature when it comes down into the reactor 4 cavity.
5- We also get the melt composition with respect to' the s .fiss' ion products and whether the zirconium has been 7 completely oxidized or not, whether we have metallic 1-
'8 zirconium coming in from the reactor activity.
t [ 8 The information .from these accident and plant 10 specifications.and the outputs from the MARCH code and the II I. f r.*W1.' CORSOR code are then fit into:the CORCON code, which models i- 12' ~ melt interactions with concrete in.the reactor cavity. 18
! We use this code primarily to determine what the 14 melt temperature is during the course of this interaction, 15 ni_. what the ga's generation:. rate is due to the mol-ten r_ ore debris it: r1 14
, attacking the concrete.. We get-CO and H 2 O release rates.
2 17 We also get the geometric surface area of the melt as it 18 erodes through the base mass, but the surface area keeps-II
. changing ~on us.
-' " Finally, we.ge't the rate at which concrete gets 21
., .- Emelted and: incorporated into the core debris melt. It's quite an important quantity forj us,- because it tends to 8 fdilute the melt and thereby compress the vaporization of g
fission prod..ucts, accentuate the vaporization of nonradioactiv e E
. materials..
f; _
- 7. e.
. q Ie- t 261 a
1 Information is then fed into VANESA model. VANESA 2 .itself calculates large-input quantities. Here's the mass 3 .. rate a$ which aerosols are generated.
=4 Here's the aerosol composition, aerosol particle '
5 size,.and.some of the material properties, notably the 8 . density. 7 We also take information on the gas generation, do a~ chemical-calculations to get the gas flux out of melt and 8 the composition of that gas with respect to hydrogen and carbon lo - monoxide,-as well as CO 2 and H 20. 11~ (Slide.) Ut Obviously, the calculations of VANESA are only as
! ~ 13 good as the information it gets from these~other sources. One
.. 14 would like to know how that information -- how sensitive M
VANESA is to that information from other sources. One would is 'like to know how well the VANESA code compares ~to some 17 experimental data and how sensitive it is to the model of the M
' melt-concrete interactions.
M It's the discussion of these quantities-that I am 30 - going to concentrate on in Ethis morning's talk. 21 (Slide.) E ~ ~ What I want to show you are a series of calculations 23
-- comparison calculations that we have made with VANESA code.
M'
- Calculations we'do with VANESA areffor the Surry plant.
35 Two conditions: One where 'all of' the zirconium has N -$N, I L
. ;.g a - s
. . o. ,
v f- 262 ; 4 I been oxidized to zirconium dioxide in the initial' melt; one I 2
- in which only half of
- the zirconium has been oxidized with 8
-2irconium dioxide.
4-At the time of the reactor safety study, I think .; 5
- there was general perception.thatLzircaloy. reactions would be 8
complete. I - think our modern perception is, certainly the - 7' results that we are getting-in as interim source term study, ' 8 ~ that zirconium has not been completely reactive. We became concerned about- this -because zirconium is
" quite a reactive metal, and it will tend to reduce fission II products and enhance the volatility.
U (Slide.) a-
=b U What I showed-here l's a plot of the~ aerosol genera-I' ~
i tion rate predicted by die- VANESA code as . a function of time 18 ' for the two situations. Surry has none 6f the zirconium { 18 present at the metal for the Surry plant. But you see, in I the calculations, the melt initially comes down quite hot, I u compacts the base mat, and'gets partially quenched. Quenching 1, 19 causes-the aerosol' generation rate to drop quite dramatically. 5
-m'
- Then, as the ' decay heat - and- the heat of oxidizing any II
' . metallic zirconium overcomes this quenching,-temperatures rise.
.g
'we get the melt flipping from the metallic phase being less i SB ~
than'in the oxide phase,.to the other way around. So, we get a peak in the aerosol generation rate'-. 5 Then, as the melt continues to cool.and'the gas [_
, .. .,, ---.--wi.',--.._.,--y,.
g , y --.,.-c .,3._, y-.r6, ,g--.,e ~+3 ,g
.i.e .. ,.
263 1 evolution rate drops, aerosol production rate drops, what you e 2 see is that b'etween the no-zirconium case and the
~
3- -half-zirconium case there is some difference in the aerosol 4 generation in the early time. ~ Eventually, all the zirconium 5 gets oxidized, and the two calculations proceed in the same a fashion. There's not what I would call an order of magnitude 7' -difference in those two. calculations. s We conclu'ded from that whether the zirconium was 8 oxidized or not was not an enormously sensitive variable in 10 our calculation. 11 MR. HILLIARD: Thi.s is Hilliard from Hanford 12 Engineering Development Laboratories. 13 Is that due to hydrogen production, the difference 14 in those two curves? 15 MR. POWERS: Hydrogen, in the general reducing 16 quality of gases coming through the melt,.makes the biggest
~17 single difference.
18
.. In the zirconium there, nearly all the water gets 18 reduced-to hydrogen, nearly all.
# ~
When the zirconium is not present, then it's like 21 80 percent reduction of the. water vapor coming into the E concrete that's turned into hydrogen. It makes a different 8 We.get different volatilities. oxygen potential.
" ;We get some differences in the speciation of the 8 aerosol that's coming in there.
p - 264 3 1 MR. KRESS: Kress, Oak Ridge. 2-Dana, you indicated part of that curve was a 3
-result of the. flipping over of the layers. But you get 4 temperatures out of the CORCON code.
5 MR. POWERS: Yes. 6 MR. KRESS: Does the CORCON code recognize this 7 flipping over and its possible effect -- 8 MR. POWERS: Yes, it does. And it has quite an 8 effect on the temperature when it does the flipping. You - are' losing a heat transfer mechanism, particularly in the 11 oxid'e phase. 12 ' MR. COOPER: Cooper, Harvard. I 13
. With the.very active bubbling, wculd that not 14 serve to keep it.really rather well mixed?
' " MR. POWERS: What we have observed experimentally
" is even at the very highest gas evolution-rates during the
~ II
. initial deposition here is that stratification occurs. We , 18 j get a metallic and oxidic phase, some mixing in the inter-
" facial layers.
But basically,.you have two distinct phases, both 21 of which are reasonably well mixed. That is a strictly
" experimental observation that we have made and fairly large-m-
scale' tests. Now, when you use tests in which you are using simulant melts, like thermitic melts, where you would have ,
),
y - w -
265 1 iron and aluminum oxide, in that case you start off with I 2 metallic phase, more dense, and you don't get this flipping 3
-phenomenon.
4 When we use corium melt, we start off with the 5 exidic phase, more dense than the metallic phase. In that 6 case, what we have observed experimentally is, starting off 7 with the oxide phase, interacting at the bottom, as the 8 interaction proceeds we get into a sandwich kind of structure, with the oxide on the bottom -- heavy oxide on the bottom, 10 metallic phase in between, light oxide on top. II We've never been able to sustain a corium test long 12 enough to go -- completely switched. But I think that's 13 merely a matter of time and duration of the test. 14
- MR. REYNOLDS: Reynolds, University of Virginia.
15 I had a question about the -- you're assuming it is 16 all liquid, and you could have the melt falling into water I down there, for example, and it will solidify, and then it 18 will maybe eventually melt, or you may have some solid mixed I' with -- well, with concrete, which melts at a much lower 20 temperature. 21 Could you have solids there? Could they remain solid for a long period of time?
=
MR. PCHERS: You're absolutely correct. And I'll show you some calculations that show you even the initial 25 melting temperatures thtt we have used in some cases are e i
266 um I sufficiently low that you would expect a lot of solidifica-e 2 tion. You can't model that with the CORCON code. So, we've 3 _ ignored it; we've said it's really a liquid, just has a low 4 temperature. It's an uncertainty area. We have been doing 5 experiments specifically in that area, of looking at what' 6 happens if this core debris is solvent, rather than coming 7 down as a complete liquid. um 8 Two observations I will makes One is that hot, 8 solid core debris attacks concrete, and it attacks it just 10 about as vigorously as does liquid core debris if the tempera-11 tures were the same -- in other words, extrapolating the line, u you really can't tell the difference. ( 18 The other observation is when'we have done 14 laboratory experiments and doped solid UO;-zirconium mixtures 15 with fission products, put them into liquid concrete, we've is found fission products partition into the liquid phase with 17 the solid largely as you would predict, based on structural 18 considerations. Those that adopt cubic structures in the 18 solid phase tend to stay in the solid UO . Those that don't 2 20 tend to go into the partitiontpreferentially, into liquid 21 concrete. How much effect that has on our release calcula-22 tions, I can't really tell you. What I can tell you is that 25 the amount of aerosol generation we get at these relatively 24 low temperatures is predicted by VANESA as low. 25 Consequently, it is not going to be a real dramatic
- f. ' . -
-- 267 l
1 effect either way. It'does not really matter if we drop this I generation rate, which is, here, sitting in 5 grams a second 3 -down to'one gram a second. 4- okay? 8 When we get the very high-generation rates that 8 really have the biggest impact on the containment behavior, 7 we get those only at very high temperatures -- when the core 8 debris'will be liquid. 9 10 11 12 ( H 14 18 16 17 18 19 Et 34 i .. l
.1 e
~. . ..-.~ . . . - - . -- . - . - . - . - - - . - - - ~ -. .
.y : , ,, ;
268 L~ 1 (Slide.) 2 What I would'like to do now is to compare the l 8 . calculations we got with the VANESA code to a correlation of
~
4 experimental data that we have developed sometime ago for the !~ , -5 - Zion-Indian' Point study. What this is is a completely . s- empirical correlation of aerosol generation rate data that j- 7 we collected in experiments of two general categories: one ; d-a category of' experiments with'relatively small-scale corium l i 8 melt interacting with concrete. Corium melt would be~about 4 10 3 kiloorams interacting over a four-inch-diameter circle of
- i. - 11 concrete and some very large-scale experiments where we were .
e 12 pouring 200 kilogram melts of. stainless steel onto 15-inch-18 diameter concrete cavities. (. , 14 ~ (Slide.) i - 18 ' The general form of the. empirical correlation is l le shown here. Aerosol concentrations measured-in these !
'17 experiments were found to have. a ' temperature dependence --
! 18 essentially;an exponential fashion -- and to depend on the 8 . superficial velocity of the gas through'the melt, and what we , 8 modeled as essentially a linear fashion. i 81 with this correlation -- that's all-it-is, is an I
- 8' inspirical correlation -- we found most of our experimental 8 data to be fit this way, with uncertainties of like plus or 8 minus 25 percent or so.- !
1 i .. l 8 MR. SILBERBERG: What-kind of distribution do you
, - - ~ ~ , - - - - , . - . - +,~A. . . ----r %---m---,,,,,,m. ., ..-~ m *r. ,-,,+,.~+---w.,,, y.
'*-7*' ,
.c7
- s. ,
t *
' ~
269- '
,. - 7
~ '
1 '~ - think you have on this? 2 - MR. POWERS: This, in fact -- this slide was made
~
3 _s .
- c. ~
,up sometime ago, Mel. It says 2 microm,eters.<- When we went back and recalculated for Albuquerque atmosphere, we found the size was really,1.3 micrometers, and,the geometric standard e .
i '
, : deviations, if you fis the data to'a 109 normal distritiution, -
7 e varied between about!2 and 2.3. 8 MR. COOPER: Particle size presumably is diameter.-
, ; ( ,
Is it aerodyrmic diameter, crxi.is,it by mass, mr # 9 median aerodynamic -
~ 10 . .
MR. POWERS: Classic cascade factor. You can S.ee 11 what this correlation al.'.ows us to do. We gan get from the y , /- * .- CORCON code melt temperatu'ros and calculate superficial, gas t 13 . velocities, just as we,did'for the VANESA code. 14 - We cea effectiv'cly decouple'tne CORCON code from 15 our comparison,"hnd compare it with just what VANESA 16 r - calculates to this empirical model. -
/
17 - MR. REYNOLDS: Care you repeat' what this is for, what 18 kind of aerosols experiment it was for? y - 19 sj'j - MR. POWERS:, It was for core debris interacting with g ., e concrete. The experiments were two types. Large scale, 200 21 kg, stainless steel melt going on to large concrete crucibles; y
.; i and small experiments, 30 kg, cor'ium compositi.on-melts going a s-into relatively small concrete. '
i N - ,
/ ,,
i - 5
- A. g
/
e-. f . . . . . . .
270 9 e e e 5 (Slide.) - 4 m 8 What I show here is exactly the same calculations 7 as before with the calculation that we get from the empirical 8 correlation shown as the green dotted line. What you see is e early in the calculation the empirical correlation is exactly 10 bracketed by the VANESA calculations. Even late in the 11 accident, where there is.some disagreement between the two
\
ut there is no more than a factor of two difference in the f 13 oredibtions,whichwouldgetfromtheempiricalcorrelation 14 and the more mechanistic model. The biggest difference is 15 a curve very late in time, which is really beyond where an 16 experimental data would be collected. Experimental data tend 17 to be concentrated in this phase of the interaction. 18 Even out here at very late times, where aerosol is generation rates are slow, we don't have a very big 20 difference. This is the difference between 9 grams per 21 second being generated and about 5 grams being generated per zt "second. m 23 What this tells me is that at least the VANESA 24 model is not orders of magnitude in the amount of aerosol I. 25 generation range predicted; that it seemed to agree with the
a ;* - . . . 271 k g empirical experimental data rather-well. 2 Another comparison that is quite interesting to make
'a i
_.s then"to compare what we calculate now with either the
~
4 empirical model or the VANESA model with what one would get 5 from the reactor safety study model. 6' (Slide. ) 7 .The reactor safety study model for the aerosol source a term during melt / concrete interactions was really concentrated o' on just fission product release. They didn't.have descriptions 10 of the nonradioactive contributor to this aerosol, which makes 11 up most of the aerosol. So in absolute value, one really
#~ ~
12 cannot. compare-the reactor safety study model to these more
,i 13 mechanistic and modalistic models. One can compare the timing 14 and the timing that chosen-in the reactor safety study was to 15 have aerosol interactions taking place, first, on an 16- exponentially decaying release rate, and then they got rid of 17 'everything in the last half-hour.
13 So their source term cut-off stopped at about two is hours into the interaction, whereas we predict much more a protracted aerosol generation rate. 21 The differences between what were calculated by the
~
22 r'eactor safety _ study model and what we would calculate now also n- extend into the-fission product releases for individual se isotopes.
. Im ..
25 (Slide.) 4 l
272 B 1 Here I have for those same calculational accidents 2 that I showed before, compared the amount of tellurium 3 entained in the melt as a function of time of the melt / concrete 4 interaction. My zero times here on all th,ese plots respond to 5 when melt comes out of the reactor vessel into the reactor 6 cavity. 7 In the reactor safety study, they assumed all the 8 tellurium would be released so at the end of two hours every-8 thing is gone, nothing is retained in the melt. 10 The VANESA model, on the other hand, calculates that 11 there is some rslease of tellurium but that it would take t2 essentially infinite time to release all of the tellurium out 13 ( of the melt. I4 For this particular calculation, we started off with 15 just a little less than six kg of tellurium and after about 16 seven hours we still had about four kg still in the melt, and 17 based on the curvature here, it's going to be a long time before 18 release is significantly more. 18 (Slide.) 20 On the other hand, f or other isotopes, the VANESA 21 model predicts less release than the reactor safety study, but ZI much more release. 23 Here I have compared the percent of strontium and [ l 24 barium released as a function of time. The lower curve 25 l corresponds to the reactor safety study model, which estimated i NB l
i 273 l l e 1 that 5. percent of barium and strontium inventories would be
' released over the course of two hours.
2
- The VANESA model, on the other hand, predicts the 3 _ ,
4 barium release comes out to be about 13 percent over the same , I 5 time period, and then it continues to release at a rather 6 slow rate thereafter. Strontium release'for these calculations was quite 7 high. We initially got about 18 percent release. That went 8
?? g through another release period, topped out about 27 percent 10 of the inventory was released overall, over this time period, 11 seven hours.
g MR. SILBERBERG: Dan, mechanistically, what are the
; a differences between the fact that the tellurium would have a 14 very low release compared to the reactor safety study and a barium and strontium study comes out high?
MR. POWERS: Two chemical effects are taking place 16 17 here for barium and strontium. We have the larger number of vapor species that we allow. In the reactor safety study, 18 le they considered barium oxide and barium metal. We also 20 considered hydroxides; there's also some strontium mixed 21 compounds. 22 That is what's doing most of the relief here. 23 Another chemical effect is that as you create a - 24 more and more dilute solution of tellurium in metals, they i 25 , just become less and less volatile for -- their chemical
s'.. 274 l fg' activity drops.down.
-I 2 That was an effect that was not recognized at the
- 3. reactor safety study. They took a boundary approximation 4
of essentially.specifying the vapor pressure that persisted for 4 5- all-time. It didn't matter what the dilution was. 8 'MR. REYNOLDS: What about the case with half ' 7 zirconium not-oxidized, versus all of the zirconium oxidized?
-8 L MR. POWERS: That factors this calculation here.
g MR. REYNOLDS: Is which?
. 10 MR. POWERS: This is the half zirconium is
- 11 present. .
n MR. REYNOLDS: It seems the zirconium will have a lot
- 13 of effect on both the barium and the tellurium, and perhaps LJ 14 in opposite directions. If you.think about what the Oak 9_
15 Ridge people reported last time, that the tellurium would 16 stay with the zirconium, whereas the barium -- the zirconium 17 should make the barium release higher. 18 MR. POWERS: We find with tellurium it doesn't make a much difference whether we have zirconium present or not. 30 We get just amazingly consistent tellurium release rates. It 21? 'does-make a difference for. strontium and barium. > -- 3'
~-
MR. REYNOLDS: Where on that picture is all the n- zirconium oxidized, then the other half -- 24 MR.. POWERS: Right..in here.
-m T$eotherareaofsensitivityisthekindof g O t e -
*i+-rv*"-
~
[ 275 1-
'information we get from the -MARCH code. We get from the MARCH
.o code' initial melt temperatures and the time at which the melt 3 _.
_ccams down to the reactor cavity so we can specify the amount 4 of decay heating. 5
, (Slide.)
6 We.had in this last sequence of plans three very 7 interesting problems, because they spanned quite a range, both 8 . initial melt temperatures and the time at which we started the 9 molten core / concrete interaction. 'It varied between a little 10 ' over an hour to start that up to 30 hours or something like 11 that. 12 It'is a long time. So the amount of decay heat we 13 [ had was very much less in the TPI sequence'and'less in this 14 TC sequence. 15
-Initial melt temperatures spanned quite a range.
16 TPI sequence, in fact, had an initial melt sequence specified 17 by MARCH, 1762 Kelvin, which is just sitting out at the 18 liquidis temperature of stainless steel. We couldn't get 19 CORCON to accept such a low' melt temperature, just didn't 20 recognize that anything would be molten at that temperature, 21 so we had to do.this low-temperature case starting with a 3 melt temperature of 1900' Kelvin. 23 What I want to do is show you that this kind of 24 (, information - what effect it has on predicted VANESA release 25 rates.
~276 1-(Slide.)
2
.First I show you how the. core debris temperature 3 . .
[ calculated'by CORCON varies depending on which of these 4 sequences you are calculating.
.5 Essentially, the TQUV and the TC' sequences really 6-
. behave.almost identical. The fact that the TQUV starts at
-7' about 400* higher' initial. melt temperature than does the TC 8
sequence affects things only for a little over an hour. 9 And-thereafter, things behave -- one would have a hard time 10 - seeing if there was-any difference in these.two cases. 11 . TPI sequence, where.the melt temperature is initially 12 quite-low, there is an induction period, runs out to about
.t 2 6000 seconds, in 'which there is a difference in the melt 14 temperature.
15 , y But eventually, things settle down and it, too, is-16 falling into'a' consistent-temperature range with all the other 17 . Essentially, what's happening is the core debris / sequences. 18 concrete interaction is sufficiently vigorous that it's wiping g. out any of the past history of the core melt behavior that a occurs within the vessel.
' 21 It can_take a little while to do that, but essentially CORCON predicts all sequences kind of become the Esame-as far as gas generation rates'and melt temperatures.
M
. MR. SILBERBERG: I. guess what that says is -- and 5
that's~'a pretty:long time -- it says that the-losses just 7
1 277 l l l 1 1 4 really aren't that much greater than the heat _ source. 2 . MR. POWERS: What happens if you get a balancing 3 - _act? You have the decay heat as a source; also you have the 4 . chemical reaction of the zirconium-in here; then you have 5 . the concrete decomposition and gas evolution as a loss; and
-s -
those'two things will come'into balance to give you a 7-consistent temperature.-If you have less decay heat, you 8 get less concrete decomposition. e : 10 (Slide.) 11 MR. POWERS: Again, by looking at the aerosol U~ .. generation' rate for the TQUV and TC sequences, these are the (- 13 .- ones that just seemed amazingly similar. And you can see
~
- 14 that the aerosol evolution rate tends to be the same, as is well.
16 They..are practically indistinguishable between the 17 TQUV.and the TC sequence, both in the amount of mass 18 generated and the actual speciation. 19
-(Slide.)
m The TPI sequence, on the other hand, there _are some 21 - differences, again, in the initial period of time. When melt 25 temperature.is-much lower than the other two sequences, you 23 . get much lower _ gas evolution rates. ife get'much lower 34 . 1 ( aerosol evol,ution rates. But'again, after a period of time, a we come out here again where it's not .over a factor of two
- y . -
- , , , , - - + , , e--- -w- ..c. ,-w
278 1 difference in the aerosol evolution as a function of time, out I here. 2 3 _ , So again, there seems to be a balancing, and we are 4 concluding from this that we are relatively insensitive to 5 the predictions of the MARCH code; that we certainly don't 6 have to have temperatures accurate to more than plus or minus 7 100*; that the timing at which the melt comes down into the a reactor cavity, we don't have to have that especially 9 accurate. m> 10 MR. KUHLMAN: Xuhlman from Battelle. What' s 11 happening at 24,000 stacking to cause that die in the aerosol u generation? I u MR. POWERS: We have reached very much the limits 14 beyond which CORCON shouldn't-be used, which I think is largely 15 responsible for that drop. We'restartingtogetcrosshay(() 16 formation and things like that. The heat transfer mechanisms 17 are changing into a regime that CORCON doesn't really bother 18 with. 19 MR. LEVY: Question. How come you do not see the 20 dip and then the rise back in the aerosol, like you had in the 21 Surry? Can you explain that?
~
22 MR. POWERS: You caught me. This is a different plant n and different concrete, and it doesn't have that sharp rise' 24 that you do for the Surry. 'Surry has a sile'.cious concrete. 25 This is a high CO2 concrete.here, and we don't get that spike. 1
- _-;o*
-s... -
l 279 4 1' The melt temperature: stays ~very consistently high with CO 2 e
, ~2 concrete,-especially this CO 2, which is very refractory, I 3
-whereas with the silacious concrete, ' the fact that you are i
-4 dissolving:and melting. concrete itself is cooling the melt i
-8 y and. keeping.'its temperature down. -So as you change your
- s-3 at transfer, and even though the metallic or oxide phases you
~
7' get pretty fair oscillations.in-the melt concrete-with , .s silaciousiconcrete. l MR. LEVY: So one could argue, probably, the 4 . 10 characteristics of the concrete.-- they're a mite more c 11 sensitive than the MARCH sequences.
~
12 i
. MR. POWERS: -What you can demonstrate, and we have -
-18
~
l' actually done these calculations recently.-- makes a big 14 difference on whether you are in this high CO2, this 80 percent 15 , calcium' carbonate concrete, or in a concrete with a significant is amount of silicon. That>could.be like the other situation, i. II s : I think, of'50 percent calcium carbonate concrete-and
" silacious concrete behave amazingly similar.
~ But this high CO2 concrete behaves different. This is a fairly-common concrete'in the southeastern part of the 21 :
United States. t- pp. E 3 s.
. MR. COOPER:' On page 15 of-your status of VANESA
~8 validatiion, you, mention that VANESA was sensitive ' to the
" CORCON results, and data available~in the literature suggest CORCON underpredicts melt temperature. Then you go on to say 4 'en. = n v -y _n, s -,
e 4 t -we v --r - t - e+ -e=* %-- w r++rw,-1-~<+ w- - - - - - - - = - - - - - - - * - * +-
l 280 m t that that could be very important for aerosol generation. f 2 Now, in your talk, I guess you are saying when you looked 3 _i.nto thfs in more detail, you found the reaching of some sort
~
4 of temperature equilibrium later on was more important than 5 the initial? g MR. POWERS: Yes. And I will go on and explain that 7 a little further here. 8 Recall in the sequences that what happened in the 9 melt / concrete interaction is that all sequences seemed to come 10 out to kind of the same thing. There are differences here. 11 They amount to about 50 degrees, as predicted by CORCON. You 12 have to ask the question: What happens if this kind of
. u consistent temperature is off by 100 or 200* here? What 14 happens?
m
~
15 16 17 18 19 20 21 22 24
, w
.s. . .. 281 m 1 (Glide.) 2 MR. POWERS : VANESSA is very sensitive here. What 3 ~ _'I have talked to is an aerosol evolution rate as a function 4 of that kind of equilibrium temperature out there. You can 5 see it's erssantially an exponential rate, especially if you 6 get up at temperatures where we are fully liquid. 7 Al raises a very good point when he says, hey, it's 8 very likely you can precipitate solids out. When you get into 9 these low temperatures where you're precipitating solids 10 we get lower evolution rates not very sensitive to the change s 11 in tsmperature in this r:gime here. 12 Once we're in the melt regime, whatever temperature
. 13 ws g t from CORCON makes a big difference on our total evolu-14 tion rats, has a similar effect on each isotope release rate.
15 There's another effect that is almost of equivalent importance 16 and that is th: gas _ volution rat? we get from CORCON. 17 Eb (Slide.) 18 Here's a plot in which plotted release rate as a 19 function of that gas flow rate from melt at about 2200 20 d.grecs Kelvin. You can see th:rt's *;ssentially a linear 21 depend:nc in this regime. It kind of levels off at either 22 nd. So it's very sensitive? to what those precise calcula, tion s 23 that coms from CORCON. 24 " MR. GREEN: George Grten, Brookhaven. 26 Ben, did you say that CORCON overDredicts or under-
s * . , , w 282 me 1 predicts melt temperature? 2 MR. POWERS: It's very tough to say exactly what 3
-1t does.
The one comparison I have seen, it was underpredicti ng 4 melt temperaturas. That's a comparison with thesustained steel 5 melt, interacting with the concrete that I was doing the TC and 6 TQ calculations with. In that case, it underpredicted the 7 melt temperature b'y about a hundr?d degrees. 8 MR. GREEN: Would you agree that the correlations 9 in the code would tend or their own merit to indicate that the 10 code is, in f act, overpredicting, or would overpredict, m<. tit 11 temperature just on the basis of the critical examination of i-th? models?
' 13 MR. POWERS : I guess I would not comment. I don't 14 know anough to comment on that.
15 Wr I 16 : , 17 I N d 18 . 19 20 . MR. POWERS: It might even the understanding of the 21 overall accident. I will agr:e to that little carochialism 22 up here on the podium in thinking that the world starts at 23 time the vessel break melts through. Perhaps that would be 24 ( helpful. I,think that's what Ken Lee has done in the documen-25 ting, r: porting th<.- overall accident. He puts thas? in some
283 4 1 sort of context for you. f' 2 MR. LEVY: Could I get you to comment on what would o i-heppen to those answers if you had a layer of water on the ton 4
.of the conerste? The sensitivity of temperaturs and things of 5
that. kind? , 6 MR. POWERS: There are really two things that happe- 1 7 when you put a layer of water over the top of the melt. -It's an 4 area where we have limited exosrimental experience, but it's s growing, the experience base. The first is much lika the 10 supprassion pool problem. You have this water layer you'rs 11 passing bubbles of vapor and anrosol-laden gas through it. You have a certain amount of that aerosol material gets trapped
'r u in'ths bubbles in the water pool. It may be released later on 14 as the pool boils off, or due to bubbles just br2aking at the la surface and flipping the material out.
1 16 The other sffect the water has, of course, is pool 17 - melt down. That's an effect that, right now, we cannot meddle 18 with ths CORCON code. We are trying to get ths CORCON oode so i t 19 will do exactly that, model a pool laytr. We're trying to get 20 the sxperimental data to back it.uo. Right now, we can't do 21 that calculation. at MR. CAMPBELL: Campbell, Oak. Ridge. , 28 On th: previous slide, you made the. comment that 34
-; we get this . result likt it is a reality. Is there .xperim:nta:
SS support for that type of curve?
a ~
., 284
- MR. POWERS : Let me show you the excerimental 0
2 support for the one with -- as a function _ of gas evolution. 3' - ,
~
(Slide.) 4 This is a plot of some experimentally measured s-aerosol concentrations as a function of superficial velocity.
-s At the tima, we~were looking for essentially a linear relation -
7
. ship between thy dots, or the experimkntal. measurements.
a You can see what happens is for-fairly high super-9 ficial gas velocity. There does seem to be a kind of a linaar 10 relationship batween the velocity of a malt, the gases coming : through the mult, and the concentration of aerosols carried 12 out by that gas. Until you g:t down to relatively low super-
' EB ficial velocities, in that it starts coming into a constant.
14 That's exactly what you'd axpect. is There's going to be some svolution rate of aerosols is from a melt if.there's no sparging at all, just due to convec-17 tion. For the tsmperature effect, I don't have a plot with i N I me on that. The data tend to be much soottier. We hava eiths: - 19 very hot melt or very. cold malt and we get, essentially, an 3D - exponential relationship-looking sort of thing. But if that 21 area is clear-cut, it's definitely true. You jack the melt at tamparature up..a little bit,it can make a vary big effect on m. thz amount of acrosol evolution and change the composition of
, that evoluti.on.
m MR. COOPER: On page 16 of your status of VANESA i
- -- . . ~ . . - :- . ~ . . _ _.- . . - . . --
_3 . y. , ( ,
~
285
' validation, you-mantion. proposed tests with molten concreta. 't. ;g.
- Did you mean with molten metal? ' You-want to look at sparging i s' .. l
_of. molten metal? i 4-
- -. MR. POWERS: I'm not sure what tests?
-8
, . MR. COOPER: A series of tests we run with gas
- s. . -
- sparging through molten concrete to determine-ths number of
;7
l aerosol particias --- l 8 .
. 4
. MR. POWERS : We'rs.dsfinitely looking at concrste !
e . . there.- We're speaking thers of a mechanical-relationship. 10 Maybe a little bit of background would help avary-11 body to know what we're talking about here. , u (Slide.)
- G
'The VANESA model has several. mechanisms of aerosol 14 ganaration considered. The dominant one, the one that la attracts a lot of our attention, is what we call the vaporiza-18 tion model. This is just the fact that ths' malt is very hot.
17 Constituents of the msit have a ~ esrtain sparging pressurs, la - sparging gas through it. The vapors go into those gas bubbles 19 carried out the' vapors thart to condense to form aerosol. The 2 other mcchanism, the one we call the mechanical mechanism, is 31 -
.due.to the fact.that when bubbles of gas come to.the surface St -
and br:ak,.they throw off aerosol-sized particles. ., SB . The difference, Eof course, betwssn these two is as "
; tha vaporiza. tion mechanism produces aerosols having a compos-s-
..1 286 1- ition reflective of vapor prsssures at the particular melt I
temperatures, whereas the mechanical release produces aerosol 3 . size hafving the bulk melt composition. But it's the surface,
~
4 the top layer, okay? 8 In the case of melt conerste interactions, tha t'.s 8 going to be molten concrete.' very quickly, it's molten concrete. 7 within minutes it's just a layer of molten concrete fission 8 products partitioned into that molten concrete as those fuels-8 slowly dissolve into it. 10 MR. COOPER: So that't the too layer? 11 MR. POWERS : Yes, becauss it becomes moltan concrst e. U Now, what we had to do on the model was,.we had.to use data fo e
- 18 gas bubbles going through water. Thers was no reason for us 14 to belisvs that the sizs data that we got for water would 2
apply to noiten concr:te, so we have done the experiments. Is And I just happsned to put the slids in. 17 (Slide.) le Here are some photographs. These are photograohs N of aerosol particles produced in exactly that way. Sparge 80
. liquid c6ncrete with the' gas, en inert gas, and look.at the 21 aerosols. You can see what we get little spheres running 88 about a, micron in size, almost exactly what we get from the 8 water data; no we feel relatively good about what VANESA did.
N
+
I You can see-there are cuite a.few bubbles. SS ut MR. HENRY: Henry, Falski Associates, p
o . 287 1 Data on the CORCON that you do for these sequences, 2 does it say th'.* top surface of the pool stay molten through, 3 '
'with the presence of upward radiation, or does it solidify?
4 MR. POWERS : Let ce tell you what I observed in ths 5 . axperiments, Bob. Then, what the code calculates when we have 6 concrets melting and creating a molten layer over the too, it 7 really never crusts. Molten concrets is basically a molten a glass and it naver really solidifies to form a solid crust, 9 even whsn it's very cold. It's quite performable and gas 10 bubbles come right up through it and break through the surfacs 11 for long periods of time. 12 So what CORCON predicts is probably is not as r. 13 germains as how ths material itself behaves. Right now, that' s 14 an area that's fairly tough to model the amount of radiant u heat loss, especially for cavities ~like zion and what not, 16 which we suspcet dots stay molten, because the aerosols being 17 produced tend to reflect that radient energy right back. 18 The gases that are coming off are not transparent 19 to the radiation when th2y have CO But that too, tends to 2 20 k:sp hsat into the melt and not let it radiate out into the 21 fres environment. Even if those wsre out thers, what would 22 happen, it would radiato up to the concrete, drive that ovr;r-23 head concrete up to its melting point, and set. And, of cours :, i 24 I
; if that overhead concretc is molten, than obviously, tha surfa4:: )
25 l of the melt,which is also concrete, would stay molten. l mm
288 MR. HENRY: I understand that. What is the longest I time duration of any of the experiments? 3
~ MR. POWERS: Eighty minutes. But mora importantly, is what happened after we stopped. Then we would turn off 5 -
powe.r to the melt and it would cool.down and temperatures woul i a drop quito a bit. What you saw was, over the course of twentv, 7-or thirty minutes, the surface layer didn't really freeze. It a just got more and mors viscous. It eventually got so viscous 9 that it behaved like a solid. But it was definitsly a glass and not a crystal solid crust. 11 MR. STRATTON: It Sasms to me the importance of 12 a layer of water on top of this nolton fu.s1 really-hasn't been 13 i . taken into account enough. It seems to ma, if thers is a good 14 bit of turbulance in the way a bubble is coming through the la moltan fuel, it's going to open up. The whole thing will be is so turbulent that th; water will bs very effective in cooling 17 the surface of this molten fuel, and it will tend, then, to is gutnch the sparging of the bubbling. Would .you comment on 19 the effect of this? 20 MR. POWERS : We have done a couple of experiments 21 in this area. I think you ars entirely right. If you have -- 2s we take a melt, put it onto* concrete and pump it full of water 23 than we .get closer to what Bob is talking about. You do get 24
- a solid crust over th2 top, because we haven't had snough 26 time to'make this concrets glass. That crust is not gas-imp
- r-
. i.
i.: x.. t l
- o l
289 l
; I 1
meable. We still get the core debris still attacking.
-2 . .
Concrets gases ars.still coming through, okay? 3 , Dut I suspect that that crust would interfere in 4 our aerosol production rate, so if nothing sise, it ought 5 to s. top the mechanical generation. It may even intsrfers in e the vaporization generation because the vapor has to pass
-7 through a cold zone. We simply-havin't done enough experim:nt s in this arsa. They are kind of touchy to do because thers's s
anothsr thing that can happen, of course; that the whols 10 system can explods on you. And so, we approach it with a 11 little bit of trepsdation. 12 In ths nsxt couple of months we're going to be t 13 doing some rathsr well-instrumented tests specifically in 14 support of CORCON and VANESA to look at this coolant layer ! 2 , problem to ses if we can inde2d modsl it, both with aerosol le g:nsration and'the melt-concrsts interaction. 17 MR. DANA: I was und:r the imprsssion that, at leas t ' ' 14 the th2rmal hydraulics of th: layer of water, had b:en looked 19 at in CORCON without worrying about the aerosol? 20 MR. POWERS: CORCON is definitely setting up to 21 handle the coolant lay:r. The problem they have is they' don't h w 2B anything to check the calculation against. Thers's no goo,d 23 data set. 34
; ME. SILDERBERG: I und:rstand, but ws've never 25 been bashful in the past of making a calculation --
i
... .. l 290 l'
(Laughtsr.) I 2 MR. POWERS : In this case we've learned before 3 .
- this august body: be careful about that.
4. MR. GREEN : George Green, Brookhaven. b
. I'd like to address my comment to Bill Stratton 6
and explain something. This coolant layer problem is somsthing 7 that has been looksd at in the last coupla of months and we a expected to find some enhanced mods of boiling, because it 9 would bs a new mode of boiling when the glass flops through. 10 And wo expscted all sorts of things. 11 One of the things we found, if you got into a 12 certain regime, the temperature that you would account for 13 vapor explosions. Rscently, we Spent a week talking with 14 Dana on the CORCON group and some others and after my talk 15 about low temperature sxperiments, they took a high temperatur e 16 of iron aluminum thermite experiments whers they fired it off 17 and after thsy firsd it off -- this is just last weak, by the is way: Tuesday. 19 They poured water on it and they got an extraordina cy so violent vapor explosion. It's one area of core-concrete 21 interaction modelling that may make a second version of this n - study required once we can get a battar understanding of what 23 happens when you drop water on top of moltsn dsbris. 24
* (Slids.)
MR,PFOVW$ 0] l y I just want2d to conclude this portion of my talk n
.. - . . I, 291 9
pr 1 with saying that it looks to us like the mechanistic modslling , 2 in VANESA at least, is in fairly good agreement with emocrica L 3 [ data. We are not off on the wrong tangent on this mechanistic 4 model as far as the overall evolution of aerosols. 5
. We still need to do morb checking of particular 6
isotopic releases, and particular. We are v>.ry anxious to 7 chsck the tellurium ralsase predictions. We think VANESA is fairly inscr_sitive to what we get from the MARCH cods calcu-9 lations. We certainly don't need plus or minus five degree ntxtix rs
, 10 .' Fe don't need plus or minus five minute melt injection.
11 VANESA, of course, is quits ssnsitiva to what wa 12 calculate the course of the melt-concrete interactions. 13 Finally, we ar? gt'tting differsnces in the VANESA calculations 14 with what our pnrc:ptions had been from the reactor safety 15 study and th2y go in both directions. Some isotopes seem to 16 be rsitas?d more efficiently and some much less officiently. 17 . Pages 291 to 302 are about a high pressure ejection of corium from a reactor vessel into a reactor cavity. d - n
-w
f, 4 7 302 4 MR. CAMPBELL: A very different issue, briefly. 4 Once you have this melt, you have two or three or four layers, 8 depen. ding on just when it is. 6 MR. POWERS: Three is the most we have ever 7 observed. 8 MR. CAMPBELL: You may have mortar on top of that.
.9 MR. POWERS: Plus water.
10 MR. CAMPBELL: The different fission products are-in 11. different layers. You're taking that into account as your ut ' source term,.but you're also taking into account the throwing 13 . out of the ones that come from the lower layer, bubble, and . - 14 go through the higher layer. 2 So it works both ways. 16 MR. POWERS: In the VANESA code, we only recognized 17 two layers of melt, and we do take into account of things la moving from one layer to the other layer. 19 MR. CAMPBELL: You take into account scrubbing a material back out? 21 MR. POWERS: That's right. 5 You're going to have things coming out of the a metallic phase and going into the oxidic phase, just moving M i there. And then the release character is a little different a from the oxidic phase than it is from the metallic phase. ( , m - -n- _,- -. _, s. ..g--
<a , g [
,- 303 1
Yes? Roger, Oak Ridge?' 2 MR. ROGER: To follow up on this last question, 3 J$na gave us, earlier, a comparison of your aerosol generation
'4 with experimental data. Then you later gave us calculations 6
for various fission product releases. 4 Do you have any comparison of the experimental 7 release of specific' fission products with your calculation? 8 MR. POWERS: We have a certain amount of data, but 9 we could get done the comparison directly. 10 We just haven't made the comparison yet. 11 MR. ROGER: Obviously, you will. , MR. POWERS: 'Yes. n
- Qualitatively, we stopped working on the VANESA code 14 when things looked qualitatively right, so I expect the 18 comparison to come out all right. The precise flow in 16 temperatures in the experiments are the biggest problem that 17 we have right now.
18
-MR. ROGER: I am not sure that it necessarily follows 19 that because the basic mass of material, the carrier material, so follows, that the specific fission products do, particularly 21 because of the point that Dr. Campbell brought up.
23 Harking back to the work that was done in melt , se refining back at Argonne in the '50s, fission products went M ' into slag layers rather definitively by up to an order of as magnitude.
~~
7 d . ._. SC / 304 g So the trace components may not' behave like the g mass component. 3 _
- MR. POWERS: That's exactly right. When I said i-4 things behave qualitatively, I was speaking on isotopic basis, a
because the bulk material is right, and I have only shown
, you comparisons for bulk material. Doesn't mean that specific 7 isotopes will.
g We have done some comparisons, and they are reported g in the validation study. I have a great. deal more material to on the high pressure ejection tests, but I think I'll stop 11 here. Ut If People are interested, I wculd be glad to talk
. up to them more on the subject. .
14 MR. SILBERBERG: Thank you, Dana. 18 le . 17 18 19 30 21 2B B 5 i : Y u . - - . - - - _ _ - _ _ _ - , - - _ . _ - - - - - . - _ _ - _ . - - _ _ _ - - _ _ - - _ _ _ -
s e . a.
^
TECHNOLOGY for ENERGY CORPORATION TO: Distribution FROM:- E. P. Stroupe, Director [. [.. k fh u National 10COR Program DATE: November 21, 1984 -
SUBJECT:
Replace:: vent tables for Task 18.1 Attached are two replacement tables. The text has not been adjusted to match the table. It will be modified later. cb r cNEENEMQYCENit9 PELLIS$1P*t PARKWAY MNOXVILLE. TN 3/922 ' PHONE 1418) 964 $414 f tLEt $101731770
MNUS
~
,@" - - , =
. CD-CAs-84-147 Project No. 4040
. November 15, 1984 Dr. E. Fuller Electric Power Research Institute 3412 Hillview Avenue Palo Alto, CA 94303
Dear Ed,
I enclose revised Tables 2 and 20 of the IDCOR Subtask.18.1 report. These tables present the revised Peach Bottom source terms and the correspondingly revised results of the CRAC2 . calculations. The text of the Subtask 18.1 report will be revised to take account of these conclusions. However, I do not expect that the conclusions of the current draft report will change significantly. Sincerely yours,
- f. , - H .
f G. D. Kaiser, Ph. D. Manager, Consequence Assessment Department Consulting Divison GDK/bh Enclosure cca 'M. Fontana (TEC) Project File O
/ w-
..4 v
Table 2 PEACH BOT'!Of SOURCE TE30tS . Fraction of Core Inventory Deleased Accident Tr" Td Tw h 6* Sequence (hr) (hr) M M _ Ie-Kr I Cs-Rb Te-Sb Ba-Sr [ 1w (3.1f t ) 2 42 ac h -10t33)IiI 10 10 4 1.0 0.19 0.19 0.11 4E-4I "I 6E-4 1E-6 4 'tw (1.Cf t ) 37 60 5(28)III 10 10 1.0 0.04 0.04 0.06 OE-5 s. 3E-4 IE-6 - 4 TC(V)I I 13 s 50 4- 10 10 1.0 0.03_, 0.03 0.07 58-5 2E-4 1E-6
~ '
4 TC (v.Can) I"I 2 15 1 10 "10 1.0 6E-4 KE-4 4E-4 4E-6 IE-5 IE-6
" ' 10 0.13 . 0.13 0.11 4E-4 TC(NV) 5 50 4 10 1.0 IE-3 IE-6 -
TC(NV SPRAY) (P) 5 10 - 'd- 10 10 4 1.0 0.03 0.03 4E-3 OE-6 , 3E-4 1E-6 ' 4 sD 23 30 21 10 ?10 1.0 0.04 0.04 0.06 15-5 2E-5 IE-6 4 TQvw 19 30 10 10. 10 1.0 0.05 0.05 0.04 iiE-5 2E-5 1E-6, ,
+ .
s .\ s , O Intcrval between start of hypothetical accident and- release of radicactive materi,al do the ? atmosphere. Totc1 time during which the major portion of the radioactive material is releasd to' tim atmosphere. c Intarval between recognition of Impending release (decision to initiate public protective measures) and the' release of radio' . - active materlat to the atmosphetg Height of release. .4 , o Rata of release of heat in calories per second. s. . . , i , s , g ,_,'. y
's W 41 Irnluttec Ru, Rh, Co, Mo, Tc.
- y gm N
s-
* - (*
x . D' g
'N' \ t s;
* , i e
~
*s Table 2 (Continued)
. PEACH BOTTOM SOURCE TERMS ,;
2 Includes Y, La, Er, Nb, Ce, Pr,.Nd, Np, Pu, Am, Car upper bound Reduced to 10 hr. .in CRAC2 analysis, b sequences run with two warning times - long (mechanical failure), short (human error). 4E - 4 = 4.0 x 10
~
1 rc V - ve!. ting is wetwell "CRD - quenching by CitD flow "NV - no venting in wetwell PSpitAY - drywell sprays operating , e
" ~
Table 20 PEACH B0110M - AREAS UNDER CONDITIONAL CCDFs . 2a 1w(0.lf t 3 ,gggg 3 2a ,c (y , ,e 1g(,,, Tg ,,, 1gvw (V,CRD) (NV, SPRAY) Early Fatality 0 0 0 0 0 0 0 0 Early Injury 0.8 0 0 0 6.9 0 0 0 Latint Cancer' 2.1E&3 740 670~ 25 1.8E+3 540 750 880 .F;t011ty Who12 Body 3.0E+7' 1.1E+7 9.6E+6 3.5E+5 2.5E+7 7.9E+6 1.1E+7 1.3+7 . Man-Rem Cff-Site 1.1E+9 1.6E48 1.4E+8 6.8E+6 7.7E*8 1.1E+8 1.6E+8 2.2+8 - ' Costa ' (S)d '#W recults unaffected by different warning times Using the Peach Dottom daytime or night-time specific evacuation scheme described in Section 5.5.1.2.
" Includes thyroid cancer fatalities (about 10% of total) . The cancers would appear spread over a perlod of about 30 years. .d 1980 dollars. *3.0E +7 = 3.0 x 10 .
, (CCffA
. - s. - in , : ,
..........s,,i.e..
~~ %. -
....,,,s..o Y
a I CYCsONE DESIGN: SENSITIVITY. ELASTICITY AND ERROR ANALYSES dot;cl.A$ W. COOMit Enuronmental Health Sciences Department. Harsard School of Pubic Health. Boston. MA 02II5. U.5A iRecentrJ Inr pubiscusson :9 Juir l982) Abstreet-The eff'ecuseness factor y =(-In openetrationi) spressure dropi. as a useful figure ot' ment for _ , cyclone design opumizauon. remaimag the same for a system of senes, partilet elements as it is for the
.ndsvidual elements. assunung they act independently. TI e partial dernatives os a function with respect to its sanables can be used to determine its sensitivity. the chan58 'n the function per unit change in the wartsote' .
elastictty, the fractional change in the function per uma fractional change in the vanatHe: and error propagation, the contribution of the sariable's sanance to the function variance. An analysss of the sensstivity. elasticity and error propagation of each of the independent vanables in the cyclone effectneness factor e
~~
indicates advantages for designs that use an assembly of high e:Tectiseness-factor elements uperated at loser flow rates and.or tower pressure drops than for those designs that use a sangle cyclone. s BNTRODUCT10% : g This attacle presents an analysts of cyclone perform. 2 ance equanons for sensitiuty, elasticity and error s propaganon. Such analyses are useful in explonng optimal and near-optimal designs. t One formulanon of the cyclone design problem for a partcular appicanon is to mmimize penetration 2 w ithin certam constraints, such as total cost or system pressure drop _ Penetration is the tauo of the number of parteles per umt ume flowing from the control : device outlet to the number per unit time entenng at its inlet. Similarly. penetration can be set at a desired !cvel. h Vs and the deugn adjusted to minimizeTost of pressure drop. Formulated m these terms, this is a constramed opumazation problem. typu: ally more difficult to solve ~
" *
- than the unconstramed problem of maximizmg or mintmtzmg some rigure of ment. A figure of ment ihat simpliries the opt:mizanon problem and has useful qualities is the "etTectiveness factor" (qs. the negative I loganthm of the penetration. multiplied by the re-g ciprocal of the pressure drop AP across the cyclone Cooper.19Nic /2 vp3 4 = [-IniPni],JP fil 1 in which the penetration is the penetration of a s 6 :
particular partcle size. Psid,l. The etrectiveness factor ; as an mtenuve rather than extensive vanable and has units which are the inverse of those of SP lt can be Fig.1. EfTectrveness factors so for senes paraitet uewed as the volume of gas cleaned completely per 'yuems. unit o( energy. in units such as m8 J-' if AP is in Nm 2. A system of several identcal deuces actmy m filter's inherent etlicacy for particulate removal m wnes and or m purallel and acting independently companson to its pressure dropiU.S. AEC.19R The would have a system value ofq. the system's etfective- etrectiseness factor has recently been applied to eye-ness factor." which would be the same as the salue f or lones (Cooper.19xit and to the indiudual collectmg exh andividual deuce. unce the deuces
- pressure elements in a packed bed aCooper,19x2a: and a niter drops are .additne and their penetrations multiphca. (Cooper. 19M 2bl. Other things being equal. it is tise. for each particle size. (See Fig. l. This hgure of advantageous in terms of ethcienev'and or pressure ment has bcen used m tiltranon work as a mcasure of a drop to use a wnes of high-q deuces rather than a
. ; 415 -qvfgtb. ~j ,
_m ( 2
s Cyckne deugn: sen>>mit>. elasucay and error anaspes N
- * - - te- showing the strong etTect of volume dow rate and 7 z cyclone duct cut diameter on pressure drop. along
. f with the induence of the inlet area iubt i . eae j Efecrtreness factor. sen.uttrttres, ehtstacittes 7 , ,
- i l The erTecuveness factor can be put mto a power tam
- 1. ' ,l '
! ea form as
\- y = 2tCTl' ';** 2'tabDJt 3Q2p,. etai
,~~~'g ' '
This form suggests that it will be most consentent to determine elasticity. from w hch the parttal densatives
,, can be obtamed easdy.as can their squares. The pow er-law form exponents for several vanables are shown ~==dneaumamen .r. wee : below,for which the Cunningham shp correction has been neglected for partsles of the sizes one typically controls by cyclones:
unable exponent _, , p, t12n-21 4, it:n-21
.7 -
- I #2n - 21 Fig. 2. Cyclone, with dimensions #8 ~'
s Leith and Licht.1972; Leith and . O - 2
- I 828
- 23 Mehta.1973L diameter raised to ahe 0.14 power and is 0.67 for D For a typcal value of n.n = 0.67. I 12n - 21 = 0 3. The
* ""
- E'#***" ' "I'*"***
= ! m at 233 K:11 -ne ts proportional to the absolute from a one.per cent change in the vanables. As the gas temperature raised to the 0.3 power Alexander.1949L volume rate of dow meresses by 1 per cent. the The factor C is a dimensionless combination ot.
cycl ne dimension rata etimenes factq wdidecrey U m cede decreased penetration is more than otiset by mereased C = titD'8 JF)(211 -Df' ilt-a 21-tr Z- pressure drop. Thus. there is an mhetent advantage in
"' ' " P " ' " '
- F M l
- d'
- d*3).3 + 1F -Z'O 2 - S1] 110) the dow be divided among them. This is the rationale m whch the pnmes indicate the quantines have been behind the "muluc!one'. The factors within C carinot o p.~ .+.m r , made dimensionless by dividing them by D:4 = a D. has e their induences seen so readily, so we performed a
, h' = h D. etc. Z' is an estimate of the distance the series of computer simulations for the condinons sortex extends below the gas exit duct Alexander. shown in Table 1. havmg chosen parucles S um in 1949t: diameter havmg the density of water and penetrations 3
near cyc nes wu D = l n
- Z' = 2.JD;(D2 abi' 2 till Table I shows the base conditions for three cyc!ones:
and d is the cone diameter at the length Z iletth.1979t one which 'we -designed" using the optimization These equanons desenbing cyclone penetration procedure of teith and Mehta (1973)and swo staridard made it difficult to infer,without evaluatmg them how . designs.a high-etTiciency design and a general purpose penetrauon will vary when one or more of the cyclone design taken from the monograph by Leith i1979L The dimensmns is varied. rows list the sanables studied: the dimensions of the The equation for predsting pressure drop chosen by eyelone, the particle and the gas vanables. The last nse Leith and . Mehta 189731'was that deseloped by rows show the derned quanunes C. In Pn. Pn. .if and Shepherd and Lapple 11940t v. The clasticities were calculated by takmg the base
, cae values and mcreasmg each in turn by I per eent.
- aP = Kiuh.Djsters 2
. 12) returning it to its onginal salue after the increase.
which Leith and Mehta found to itne almost as good Several conclusions can be drawn from the results
.igreement with expenmental results as did several presented m Table 1:
other more compheated expressions. The salue for K is a l a The cyclone diameter is the s inable to w hich t he 16 for a cyclone with a standard tangential inlet effectiveness factor is most sensstne.
. iShepherd and Lapple. '194ut so this equation s 21 The ratio D, D is one to whoch the design is also becomes: . quite sensiuve. a charactensuc exploited by the AP = 8p,Q2;ah DJ. (131 Leith-Mehta design.
1. e ==; .g L
- -- .--- -3
.o. '
, e, .
Cgione design: senutnaty, clastetty and error anaipes asy should be pret'erable to selectmg only a standard Cooper D. W. ilvs:bi Optimmns t!!ter hber diameter design or to calculatmg an optimal Letth-Mehta 4rme earns Em ramaient 16.1129 If.t.* deugn for esery wries parallel contiguration con. -Gardner R. H O'Neill R. V. Mankan J. B.. Carney J H
*Md 18981: A comparwon of senutmt) analym and error analysts based on a stream ecos9 tem model. Es.=
- Modellume 12. I?) I96
- 4. &=minlucarats-Thrs work has benented from my discus. Leith D. s19791 Cytones. In Huauh=4 4 Easen=mcarui
. wons math solleagues Dr. Dased Leith and John A. Dirgo. Enviarenau iEdited by Pereira N. C. and Wang L K.:
;' Humana Press. Cldton. NJ.
Leith D.and Licht W c197:1 Collecten erticancy or cglone
- REFERE.%CES type partele collectors a new theoretcal approactt.
A1.Ch.E Simp. Sener: Aar-1971. Alesander R. McK. e1949 Fundanwatals o(c)1',one design Leith D.~and Mehta D. 41973) Cyclone performance and and operatson. Pmr. Austrate. Imr. Mna. Metall (New dessgn. Atnunparnr Ensanmaerar 7. 2:1-149. Serw 192-3. 203-223. . Parker R Taen R Calvert 5. Drehmel D. and Abbott J. Cooper D. W. e19*ee Theoretcal compartion of ethewncy . t1981 Particle collection m cyclones at high temperature
.*--- - - -- 4 ad poner for smgle. stage and multipie-stage partculate and high pressure. Eater. $ct. Ter*aos 15.451 4 58.
.- wrubbing. Arairmenerar Entrnmairar 14.1001-1004. Shepherd C. 8. and Lapp 4e C. E. 41940: Flow pattern and Cooper D. W. a1981: Minimisms cretone energy consump- pressure drop m cyclone dust coelectors. lad. Envae Chem.
tion./.A w PWlur. C.merod Ass.Jl. 1893-1894. 32.I:46-1248. Cooper D. W. el98:an Filter bede energy.ethcwns packmg U.S.. Atoms Energy Commission t1950 Nuadh=4 .=
., diamescr. 4 4w P,dlui. Cuatrod Au. 32.20$ 20s Aremals Washington. D C.
em 9
'\
31Ne ^ "-.-_(d 9 m 9 k e
. . pe
... 4.#Q .;
~^-
6
gefd -
* )
s rtM OY RI CTit.INE A R MorlON 49 ' -
' the weight Insufficient attention to these considerations sometimes ! cads to erroneous conclusions. Thus to explain the shape of the path of a stream of tobacco smoke the medium injected horizontally into a smoke chamber Prosad [84]. starting from the assumption !
that smoke particles settle indisidually, had to take the absurd value of 24 for the mean particle radius. In fact the shape of the path was undoubtedly determined by the i13.6) rate of settling of the stream as a whole because its density exceeded that of air on account of the carbon dioxide contained in it. A striking example of rapid settling of a cloud is furnished by the " tire cloud" which descended with tremendous speed from the volcano Mont Pelee in 1902 and turned the town of Saint Pierre into cinders. , Evidently the concentration of the disperse phase (volcanic ash etc.) was so great that il3 D the density of the cloud. despite its high temperature, was much higher than that of air.
. If Stohec A scry complicated system of movement exists in cumulo-nimbus clouds contam. i shole cloud. ing droplets of all sizes from e = 10 to r 3 mm. In this case, under the intluence of the higher temperature of the cloud in comparison with the surrounding air a rapid rise of the whole cloud takes place at a rate up to 10 m sec-L while the
( 13.M) drops of water in it are falling individually at speeds between 0-01 and 8-9 m sec-1 The resultant velocity of some droplets is therefore directed upwards while others gg3,93 move downwards. These phenomena play an essential role in the process of precipita-tion from clouds (see page 319).
"* 'I }* , t 14. THE MOTION OF AN AEROSOL IN A CONFINED SPACE tional to the acts Rn and For aerosols in an enclosure the motion of the particles includes that of the medium ;
caused by convection currents, artificial agitation, etc., as well as their own motion j re!stive to the medium. Just now we are interested only in the latter, and shall examine i a ongin, and it fot particles settling under gravity. If the particles of an aerosoi eccupying a space 7 i
's inside and - e nfined by walls settle with a velocity V the medium moves in the reverse direction s, and hence, with a mean velocity y V, where y is usually a very small fraction of the total volume greater than of the disperse phase. Since the njedium is entrained in the vicinity of the particles. ,
then in the spaces between them the velocity of the countertlow is greater then y V. l der I gmim a Thus the rate of settling of partic!cs in the present case, unlike the motion of a free re ditTerence el ud is less than that ofisolated particles in an infinite volume by the factor ! -,- xy g 2te humidity w here x > !. Grc plays b Acc rding to Cunningham (46] still another factor should be taken into account, , ich is mainly in the derivation of Stokes' formula one of the boundary conditions is that the velocity
- inside and of the medium is zero at an infmitely great distance from a particle. When a cloud of Isarticles settles in a confined space. however, the velocity of the medium is zero at a ution of fuci 'Itance e from the centre of a particle,where 2 e = n-'
- is the mean distance between th the aid of alsacent particles. Thus each particle experiences the same resistance which it would ,
- 4. and so on. N.rience at the centre of a closed spherical vessel of radius 9. According to Cunning. ,
. I bekre 'am s calculati ns this resistance, on a Stokes approximation, is equal to 6:tr Vu ;
ration of the 81 - I 25ra). Following Oscen,the correction becomes less the greater the Reynolds
- cs the cloud number Ve;!y. All other authors occupied with this problem have arrived by way of ,
' because the 8airly complicated, but not rigorous, considerations at a correction factor of I + xp away at the '"I".*5 UI x equal to 5 5 [85], 70 (33] and 4 5 (86]. Rigorous solution of the ,)
7 1 I""blem is obviously extremely difficult.
- t "
l \
-7 I
,d I
' t 50 }
fler MCCHANICs OF Al'RusOLs i l The dirrerence between correction factors of the type 1 + xy (l) and I + x r/a f*5"
= 1 + xef '
i(ll)is ofimportance because at the usual salues of y in acrosois the ; I factor~ ti - x11 , j 11) is practi; 'iy equal to I while the factor 11!) may be a few per cent greater than 1.t$at the velo For the sm il values of j 9 which are of interest this probicm has been investigated g power 3 3 experimentally only by Kermak [85] who measured the rate of settlement in monodis- l The prin l j perse suspensions of various animal erythrocytes with radii 2 4, 3 0. 3 7 and 4 4 ,,, is consec: in water. It turned out that for y < 0 04-0 08 g. r = 1-Si j the experimental results agree well with a correction factor i + xy and x has values lying in the range 4 8-6 9 for partic!cs at : vanous erythrocytes. Unfortunately the settling rate of isolated particles was not coefficient cl
," measured in this work but was determined by means of extrapolation. 4 serosols is t!
Thus from the rather scanty data available it may only be said that in rIthe tion. Un fort setding l of aerowls in a confined space the resistance of the medium at low y is !probably a his enou I equal to 6:gr V9 (1 + xy) and x is close to 5 or 6. is necessar';!
).
The rate of settling of concentrated suspensions has become important recently thuefore, t!' in connection with the fluidization of powders (see page 367). In the fluidized state a concentration of particles for which the settling rate is equal to the flow velocity is automatically established. Experiments on fluidization have led to the formula g 15. M C i V/ = V,(1 - 9 )', (14.1) l where V/ is the settling rate of the entire system of particles and V, is that of an The mot i isolated particle. motion in ti For spherical particles Lewis and Bowerman [87] and Richardson and Zaki field [88]is qE w; of the partiC obtained the same value 4 65 for the coefficient.t. An approximate theoretical calcula-
- tion of the settling rate was made by Richardson and Zaki who started from two
{ models for the distribution of spheres in space; they obtained two curves (V/,y) ; l . one of which expenmental lies curve. about 40 per cent higher and the other 20 per cent lower than the The mo practical at In conclusion a phenomenon will be mentioned which is familiar to everyone earth,s graq working with aerosols. When concentrated aerosols settle the upper boundary [90] is and Eh .
'; usuaily flat attd horizontal, a phenomenort which is exhibited both in the laboratory has played (
i and in natural mists. The explanation is that, for an aerosol density exceeding that of
^*f050I
{ . the gas adjoining it, hydrostatic forces. counteract any disturbance of the horizontal ser plates ar position of the upper boundary of the aerosol by convection,just as in liquids. Such observation ;
, stabilization of the upper surface will be observed only when the particles move asa ahonzonta ;
whole with preceding the medium, which necessitates a sufficiently high concentration (see section). where /7 is j The surface of aerosols dispersed in dense gases like chlorine or carbon dioxide The strengt!j i etc is particularly stable [89}. - 0f a particiq
- simultancote Many theoretical and experimental papers have been devoted lately to the sedimen-i tation of particles in a limited space, or hindered settling. Only equations which referd*P'"di"ECi to very small values of the volume fraction of the disperse phase, y, (the fraction of
*t*"5"F E' determmed +
the total volume which is filled by the disperse phase) will be given here. Following Cunningham's ides (see p.49), but allowing for backwarus flow, Happel [610] and Kuwobara [611] obtained at y - O the formula V//V, = 1 - xgd in some l with
= 1 - 21 x y=and I l 5 [610] and ' 62 [611]. Brinkman [612] deduced the formula Hawksicy [613] V , both above V//V r the expression V//V, = 1 -921 ", //V, = 1 - 4 5 5. Experiments [614] confirmed venical 6ch '
for almost isodisperse liquid suspensions. The following p - '
. m. 4. 3
e, l
-4 9
4TC A DY RIXTILINE A R stoTioN SI p 1 i e x r"e.. results of all other investigations can be expressed by the formula V//V, = 1 - x5
! ;crowls the factor = (1 + x 7)-' with x = 4 0 [615. 616]. 4 5 [617] and 5 4 [618l. The conclusion (see p. 50) ;
nt greater than I. l that the velocity of hindered settling at small concentrations depends on p raised to the been investigated i tirst power seems to be confirmed. but no theoretical basis for this is apparent. l ment in monodis- The principal difficulty encountered in precision measurements of V//V,, at small
. is convection. Wilson [619] using very dilute aqueous suspensions of glass spheres {
FO. 3 7 and 4 4." i results agree wcH of r = l-5/4, found that it was impossible to obtain strictly vertical trajectories of range 4 8-6 9 for partic:es at room temperature, although they were realized at 4'C. when the thermal particles was not coeflicient of water was equal to zero. Only in more concentrated suspensions and hat in the 5:ttling acrosols is the downward gradient of concentration high enough to suppress convec-tion. Unfortunately, this is often ignored. It is difficult to combine the two conditions.
)
i
>w T is probably a high enough weight concentration and a low enough particle concentration, which
~
is necessary for the neglect of coagulation in not very coarse aerosols. It seems. } uportant recently therefore, that much sedimentation analysis of aerosols is erroneous. he fluidized state 1; se flow velocity is the formula 4 l$. MOTION OF PARTICLES IN VERTICAL AND HORIZONTAL ELECTRIC FIELDS. PRACTICAL APPLICATIONS ' ' (14.1 ) V, is that of an The motion of aerosol particles in an electric field is no ditTerent in principle from . motion in the earth's gravitational field. The force acting on a particle in an electric field is qE where g is the charge on the particle and E the field strength. The velocity on and Zaki[88) of the particle given by formula (8.2) is ieoretical calcula-i Ig started from two V = 4EB = qE I + A g 6:rry. (15.1) o curves (V/.5) }
,tt lower than the The movement of particles in a vertical field is very interesting on account of the 4;t practical advantage obtained by the electric field being superimposed upon the j .iliar to everyone earth's gravitational field. The vertical electric field method developed by Millikan h
sper boundary is [90] and Ehrenhaft [91]is one of the most fruitful methods of studying aerosols and j in the laboratory has played a very large role in advancing knowledge in this field.
;3 ,
exceeding that of 3 ACf0503 particles are introduced into a chamber formed by two horizontal conden-of the horizontal >cr plates and having side walls ofinsulating material provided with windows for the '} , ' s in liquids. Such observation, illumination and charging of the particles. Observations are made with trticles move as a 4-a horizontal microscope having an eyepiece graticule. The field strength E = R/h. A ancentration (see where R is the potential differeau and h the distance between the condenser plates. N The strength and sense of the electric field can be varied as desired. The rate of fall ' r carbon dioxide "Ia Particle V, is determined first with the field switched off and then under the ! simultaneous influence of the electric and gravitational fields. V -e V or V, - Ve ly to the sedimen. jh.; depending on the sense of the electric field. Hence V,is found. In addition, the field
.tions which refer "itensity E, which exactly balances the gravitational force on the particle is sometim es . (the fraction of 'tet ermined I '
here. 2 backwards flow. E, = mg/g = 7:rr rg/q. (15.2)
//V, = 1 - g ..
in some chambers provision is made for varying the pressure between wide limits formula V//V, F both above and below stmospheric pressure. The technique of working with the p
. [614] confirmed tertical lic!d method has been weil set out in the literature [53 92); it permits the I;!
.espensions. The "Howing pr bicms to be solved.
.1 i
l 4 l, l e l i f
L,
- 17.
ACCIDENT SE00ENCE LIKELIHOOD INFORMATION
, FOR NUREG-0956
- 1. BACKGROUND:
ASEP supports the NRC source term reassessment and severe accident risk reduction work in three areas. First, it provides ASTP0 with the accident sequence likelihood infonnation for all the source term accident sequences of the 6 reference plants and a limited rebaseline of sequence likelihood for some of the accident sequences. Second, it provides SARRP Phase I by early 1985 with a detailed rebaseline of sequei.ce likelihood of the ASEP identified dominant accident sequences for the 6 reference plants along with the dominant factors that drive the sequence likelihood. Third, ASEP provides SARRP Phase 11 by late 1985 with the identification of generic plant groups and the identi-groups. and description of the dominant accident sequences by the plant fication This paper only identifies the ASTP0 needs and the ASEP approach to et its needs.
- 2. ASTP0 NEEDS:
In November 1983, Bob Bernero, in his meno on " Source Term Report NUREG-0956 " requested from DRA the " estimate of the probabilities corresponding to the sequences analyzed in the Battelle reports (BMI-2104)". ASEP was given the task to support NUREG-0956. Conversations were held between ASEP and ASTP0 on the precise needs of NUREG-0956 from ASEP. ASTP0 did not express the depth and breath of the ASEP input but conveyed that " probability estimate" will play only a minor role in NUREG-0956 since it contains mostly deterninistic analyses. Point estimate on the sequence likelihoods was told to be sufficient since the consequence analysis will probably contain no uncertainty analysis. ASEP input was needed about by mid January 1984; as of now, the NUREG-0956 schedule has slipped nine months.
- 3. SCOPE OF WORK:
ASTP0 has identified 19 accident sequences for 6 reference plants for its source low term reassessment. probability sequences. The 19 accident sequences consist of both high and Most of the dominant accident sequences are rebase-lined by ASEP. Those that were not rebaselined either required entirely new system models or extensive mndification of the original models. Some examples of insights or system changes that were not incorporated in the rebaseline effort are: 1) the modifications of plants in response to ATWS requirements;
- 2) the possibility that the operator can reach cold shutdown before havin go to the recirculation mode of injection in PWR small LOCA accidents; 3)gthe to possibility that the operator can cool the RCS enough to go to low pressure injection before core melt in PWR small LOCA accidents; 4) the fact that Surry now has an AFWS cross-tie between units; and 5) the fact that the procedures for checking the Sequoyah ice condenser drain plugs have been modified.
i 3
., .=,,' * 'l l
The following list contains the ASTP0 accident sequences by plants and their status of rebaseline. SEQUENCE PLANT SEQUENCE LIKELIHOOD UPDATE Grand Gulf TC No TOUV Yes TPQI Yes SE Yes Surry A$ No _Sq D Yes M B' Yes V Yes Peach Bottom TC No TW Yes AE No Sequoyah TMLB' Yes S HF No Zion i k 5D No 3 Yes 2 TMLB' Yes Limerick TC No TOUV No TPE No I Rebaselined based on Zion review. 2 Not rebaselined since it is not a RSSMAP and WASH-1400 plant and it is a recent PRA.
- 4. LIMITATIONS OF REBASELINING OF SEQUENCE LIKELIHOODS:
Due to the time constraint established by ASTP0, the superficial need by NUREG-0956, and the unavailability of the ASEP system models in late 1983 and early 1984, a detailed rebaselining of sequence likelihoods was not performed that would require an extensive re-analysis of the PRA involving modeling and assumption changes, reassessment of data, etc. Therefore, the limited rebaselining information was used from the August 1983 interim report. The following list of general steps describe the limited rebaseline process.
- 1. Review the insights collected from recent PRAs, operating experience.
TMI fixes or any recent plant fixes, and special safety studies. t Detennine if the sequence likelihood of the PRA dominant accident
. sequences should be changed to reflect the new insights.
l l 9 e - w y m---n-,-+w w 4-
( ; .s ,,i f
- 2. If the insights applied to the dominant accident sequences, apply the quantitative' changes to the major cutsets for the failure expressions denoting the sequences. If an insight excludes or includes certain '
' failure modes in the. PRA, delete or add cutset expressions by hand I
i representing those failure modes. Subtract out the cutset ' probabilities for those that are deleted and estimate the additional e L probability for any cutset expression added to the PRA using generic data for that insight. Adjust the sequence likelihood accordingly to yield a limited rebaselined likelihood for all the PRA dominant accident sequences. 3. Review the PRA non-dominant accident sequences and apply the new insights to.them. Detemine if they still remain non-dominant. If they become dominant, add them to the list of dominant accident
, sequences.
In the above rebaselining process, the key steps are ascertaining which in-sights should be applied to the specific plant of interest and what are the ' quantifiable changes.- ASEP has rebaselined the sequence like11 hoods using, for the most part, those insights and quantitative changes agreed to by an author or otherwise knowledgeable person for each PRA. However, additional insights are sometimes. listed in the rebaseline tables in the August 1983 interim - report. These additional insights could be applicable to the plant of interest but an extensive rebaseline effort is required. While the limited rebaseline effort provides a more current estimate of the likelihoods, the other insights. not factored sequence into the limited rebaseline effort could possibly change the likelihoods. j Therefore, the sequence likelihoods from the limited rebaselined efforts only present a "better value" and not the "best value" for 1 the sequences of interest. The attached package is a draft input to NUREG-0956.
- 5. FURTHER ASEP WORK FOR NUREG-0956 i The assumptions made on the rebaselined accident sequences will be verified by the utilities. SNL is presently ' putting together the items for verification.
After the verification, the accident sequences will be rebaselined, if neces-sary, and.will be provided in an appendix to NUREG-0956 by August-September 1984 The verification. items for NUREG-0956 will be organized with the items from the ASEP work for SARRP Phase 1. Its objective is to detemine the current plant risk of the 6 reference plants. ASEP is performing a detailed rebaselining of the selected ' dominant accident sequences for each of the reference plant using the ASEP generic system models and data. ' ASEP is perfoming the initial screening to detemine the major contributors that drive sequence likelihoods; the important items (e.g., hardwares,' assumptions) for each reference plant-t
. will be verified along with the items from the _ limited rebaseline effort. At i
the same time frame, the ASEP. data . base will be reviewed by data analysts for their 1985. applicability. ASEP products for SARRP Phase I will be provided by early
'j,.,
DRAFT APPENDIX C ACCIDENT SEQUENCE LIKEL1H00D INFORMATION This appendix will serve to give the reader some probabilistic perspective on the sequences analyzed in the source tem reassessment studies. Probabilistic infomation will be given for the nineteen sequences that were analyzed in this document. The sequences were chosen from six reference plants with PRAs and were selected either because they were dominant or because they represented a unique phenomenological situation. The infomation compiled in this appendix came from the WASH-1400 PRAs*1 (Peach Bottom and Surry), the RSSMAP PRAs (Sequoyah*2 and Grand Gulf *3), the Limerick PRA*4 the Zion PRA*5 , the Review of the Zion PRA*6
, and the PRA rebaselining report produced by the Accident Sequence Evaluation Program *7 . In many cases the infomation presented here is based on the analysis in the original PRA, but has been modified (rebaselined) to incorporate some new insights that have been gained since the original PRAs were published. These new insights were not incorporated by modification of the original PRA models but by back-of-the-envelope type calculations.
The original PRA values were modified where possible, but there were several cases in where insights or system changes were not incorporated. The changes that were not incorporated. required either an entirely new model or extensive modification of the original model to incorporate them properly. Some examples of insights or system changes that were not incorporated are 1) the modifications of plants in response to the Anticipated Transient Without Scram (ATWS) requirements; 2) the possibility that the operator can reach cold {' shutdown before having to go to the recirculation mode of injection in Pressurized Water Reactor (PWRs) small Loss of Coolant Accident (LOCA) situations; 3) the possibility that the operator can cool the reactor coolant system enough to go to low pressure injection before core melt in PWR small l LOCAs; 4) the fact that Surry now has an auxilliary feedwater cross-tie between units; and 5) the fact that the procedures for checking the Sequoyah ice ! condenser drain plugs have been modified.
4d Because all of: the numbers presented in this appendix are at least
- partially. based 'on pas't PRAs the reader should be careful when comparing the numbers from one plant to the numbers from another. Dif ferent PRAs have ,
different analysts, different methods, different levels of detail and different i perspectives. The WASH-1400 and RSSMAP PRAs were NRC sponsored while Zion and Limerick were industry sponsored. . The RSSMAP models are not as detailed as the I 1 WASH-1400 Zion and Limerick models. The WASH-1400 PRAs are ten years old while the Zion PRA represents the state-of-the-art -methodology. For these reasons.
- and others, the reader should not compare the absolute numbers given in 'this appendix,- but instead should compa're the contributions of the sequences te the total core melt' frequency at the respective plants, r
As pointed out in the previous paragraph, each PRA is different. Each of the PRAs handled uncertainties differently. The relevant WASH-1400 PRAs used Monte Carlo simulations to propagate data uncertainties. The RSSMAP PRAs did
- not consider uncertainties. The Limerick PRA did a brief uncertainty analysis 1
and. the Zion PRA did a detailed Bayesian analysis. Because of these
~
differences, the actual uncertainties were not presented along with the sequence l dGscription. It should be pointed out here that the uncertainties in sequence l frequencies are large, typically on the order of one or two orders of magnitude. i 1 h v e- -----r,--_ ..w,- w- . , . . , --.,,,,-,.7 ,+ .i- ..,..-...~..,.ym+ , -.,.,,.m_w...,,,c...,e,._w.-..,~. , , , , ,
l ' i
'a .
.e. e w . )
l p s i l e s ae fd t a aa r e ye b bd e _ dv e ee mr h t ep _ n l l e o l g f n s i Am Ce d Ot n , I s e p s y e p es g d m u rn ae y p ll r I t e ac v M sj e y L . ea n r vl a u l t o oe vr t . f ne ea f i s sir o s ee 3e' cr rt e n np oi N o e crc O uw go s Rd s.On ws el s e hs t e y.a sss sh c upp . dc pmm e it h nu uu c 1f as 5pp n o 'yf MI5 e
- ro t u
g N 0 e v) e LMM ELL e s l es s w tf p ce a err ns C f ua A t Ki oac oe tR uC e( fff o ff n e t r is ti os eoo t t SD ct e ve oo B e _ ts _ n hww ee TTT h 8 vt . c $ ea y s rr t
. . . a pa l I23 p
) p e e 0 oe g 4
1 0 ts
)opw t e.
ny f h t M I r : . 5 st o s) O o - A ng e )t mI . - n . Sn e e W m en mis Eve tMsL s a 4 E f o t e o tu (n y( C 2 e t e r n I s . e w F ae m e5 n) r l ( l c neS h e 8 on oe tK s y l el tcL tM t v e cmg St ce( Rd t s ej On e 7 f e to jay a c
- os nl nl a .sys n
- a i r ny e r t s lh t p t a n1 os ens us pu r
9 e y e> oi ra p e et Cn ul e y e s
/ m e ser 5 a C
a A C S (e ad t rl ep soo eCC K5 LMsp
. l n
0 t e r A Lm a sa cr p,e e crp u 7 t A t u a rr A 3 e e t C O-nl ea eo Rt weu oss e
- s ee d t
7 4 L mv ee l ss s t C a f oI t t 5 e r f e or ee a $. f fM o I 1 g i0 e f rr op r C 1 ofL o n r u0 sl
< a r r d A : t -
q4 sl ea eww roo uoa i e V e1 r LF ett owe d
- FTf s
a Kl e t l
- e5 hA 8 H l l
y t C u l A 1 a- - .. $ q i 1 F - - t 12 A n i W t o V
- l i O F I t
a c KT N R T N s s t 5t R _. h t t u q V E O C Ti Nn T N i T TS NS 0E Tt EU c A i E EE Equ l E _ a l i E l VL V t VC p e t F MRGM M' l l 0 t A EI A E i g EK Ft 5 t E I E1 En EF E A F$A T _. t K M T A A D l - A E5 5 KIR oL C hM KMI L KMR I N AM f t f I II TM EU l C tTU uCL tt Ut ttA u1V EtE UlI NI N u U T t If Q5 qnI Q5 q5A IV A il G I a NV fl tUA Ef t7n QiI [fR MI t p t S E S N o IE SB 5FF SS S5U $$C OR I C DD
, ;' ,{ !;i ! ! , i , 4 ; $ ,
-- _ _ . - _ _ _ _ _ - _ _ - - . - - _ _ . -. .. . . . . . _ - ~ -. .
a PLANT: Peach gottom $tgutKE
~
SEqutKE: TC M 5ClelfileII: This seguenca is initi;ted by a transient f;19eved by'e1 fallere to achieve reacter Sebcriticality. Af tre f llere . . . ' to achieve subcriticality, the power is empetted to - SiqutIICE Filt0Hf E T: - 7t-6 (From IIASN tete with some revislens) egellibrete et 30s. The heat sent to the seperessten IRITIAtlNG EVENT FREgiftKY: 7 transients / year pool is greater then the residual heet eveeval cepacity. The contalanent is estiasted to fall due to stese ever.- CORE INIMIGE CONTRitWileII: 305 - pressortsation et about fif ty eight sein. The seeeeeecy injettten Is assomed ie falI when the contelament feiIs doe to either defoemetten of the injectlen lines or ca,ltetten of the pumps. IRiflAf flIS - 1 -- Any transient other then a less of offsite power transient that requires a reacter shutdoen. EMNT ' StqutKE This segeence makes several assumptions which may not be valid. .
- Dl50U551011 I? It is assumed that the failure of three adjacent rods alli not shut the reector down. This may be conservative.
2e it is asseried that the High Pressere Coelant lajectles, low Pressere Ceelant injection er tow Pressere Core 5ersy -
- pusys will not fall before containment fallere. However, it is possible that the ponys could fell prior.to contalseent .
fallere due to lack of net pesill.e settles head er high temperatores.
- 3) Traditional procedures were essemed. The new energency procedures may allow the operators to remetW within the beet resevel cepecity of the suppressten peel.
- 4) AlW5 3A laplementation was not considered.
SEQUtKE Reetter subcriticality. FUlecil0NAL l tAILURE I SEQUt KE EVERT -- Reacter Protectlen System (RPS) falls. ST51fM FAILifRES -- The Standby Ligold Control (SLC) system falls. SEQUEKE EVENT C - SPS* SLC ST5IEN C -- IE-6 IntAVAILAtlLITIES ] SEQUt KE EVENT 1. RP5 socceeds and control rods are laserted Into the core OR ST5IEN SUCCESS 2. Decirculetten pusy trly and manual shutdown of the reactor esing SLC or penselly drietag in control rods. l CRIIERIA . DOMINAlli FACTOel -- Fallure of the operator to initiate $tt er to manually insert the control reds. DRiflNG SEQUt K E -- Failure of any three adjacent rods to Insert.
-- Common made faltere of the IIPS logic system resulting from her. rrors in testing and meintenance. l I'
i _ _ _ _ _ _ _ . ___ . _ _ _ _ _ _ _ . - + , . . = .- _. . , . . - - m _m - . __ . - . _ __ _ _ _ _ _ _ _ _ _ _ _ _ _
Planit~ Peach Botten .'.0o 5ttpitut '*
- SitpptIICit TW
- DESCRIPflm: This seguenco is 1:ltleted by a transient (f '
a fallure et the Resideal lleet Removal (IWut))systeef;llowed by 5tquintt FREqut K T: SE-4 (Free Astp Rebaseline) to remove heat from the suppression pool (W). If ~ , the heat rejectlen Is not inttleted within twenty-nine. INiilATilIG EVENT Filf4UEET: 7 transients / year howrs the containment is asseewd to _everpresserlse. and fall at apprealmstely 137 psi. The rapid-CORE DAfglGE CONTRitiffl0H: 305 depresserlastion cowsed by contalament fallere is assweed . to cause the suppresslen peel water to flash, resulting
. In cavitation of the low pressere Energency Core Cooling '
System pumps due to Insofficient net positten section ' head, injection is therefore predicted to fall subsequent to containment fallere. IMlilATING-LVENT 1=ig*T2*I3 wh m Ty 1 Transient due te less of e H s Ne w (0. 0. Transient due to awtematic trly with laterruption of main fee &ater (3/yr). . 7 --
- Transient due to en automatic trip withewt laterruption of main feeheter (4/yr). ,
1 4 3 5tqutIICE
.Bl5CUS$10M 1hevery can fregtency of magnitude orders of this seguence is somewhat depending subjective.
on what recovery Because aedel is osed. ofMost the long time available, the credit given for recovery , fallere forther by opening normal containment vent lines. SWRs may else be able to delay contalernt
, 5f0UtKE -- Less of Reacter Cootent System (KS) lategrity.
FUllC110NAt. -- Containment protectlen free overpressure due to steam. FAILURE -- fallere to replenish lost K 5 Inventory. ! 5tgullett EVENT W --
- SYSTEM FAILURES Tallere mode cooling te recover of thethe IWatPower Conversten system. . System (PC5) er fallere to remove heat from the suppressten peel estag the peel 5l M M E EVENT ig 1
$Y5fth 23
! IfMAVAILABILIIlE5 W -- SE-4 IE-6 SEQUINCE EVENT 1. 5751tM SUCCESS 2. One PC5. IMt train with flow to the heat enchanger of that train and cooling water operettenal to that heat eschenger cet CRittRIA . DOMINANT FACicits -- DettvlNG StquthCE Marhare falleres la the output piping and valves of the fuergency Service Water (t5W) system. Fallere of the operator to start any high pressure t5u pusy within twenty-five hours. I i , 4 l 1 I i i
- --_______-___ _ _ _ - - _ _ _ - , . - _ . - _ - . - .. - - _- .. . ... . . - ~
~., ,
PtANT: Grand Getf - 5tqutKt StqutKE: y3 T C Ot5CRIPflen: This segmence is initiated by a transient felleued by a f allere to achieve reacter subcriticality. After the
- 5tqutKE FRtqutKT: St-6(FeenGrandGulfR55MP) f ailure to achieve subtriticality the power is espected to equilibrate at 165. The heat sent to the soppresslen INITIATINE EVENT FRtqufET: F translents/ year peel is greater than the residval heat esmoval capacity.
the contalament is estlested to fall due to steam ever-tent DAMGt CONTR100110ll: 235 presseriretten at about eighty minutes. The emergency lajectlen is assumed to fall when the containment falls due to either deformellen of the injectlen lines er cavitetten of the punys. INITI Af fles 1 EVENT 23 -- Any transient other than a less of of fsite power transient that regulres a reacter shutdown. SEQUEEE The sequence description assumes several things, any of which if not tree could che Ol5CU5510Il the segmence cessten. 1? It is assumed that the fallure of three adjacent rods will not shot the reacter . This may conservative. 20 14 is assumed that the pumps will not fall before containment fallere. This may not be trees the pumps could fall ! earlier due to lack of net positive sectlen head or high tesperatore.
- 3) It is assumed that the high pressere core spray can prevent core melt. This may be a non-conservative assesytten.
Because from reaching cold water is pet core het en the tog of the core, power lastabilities may ector and upward moving steam may prevent water spots.
- 4) Traditlenal procedures are assemed. The new emergency peacedures may allow the operetten to remain within the beat removal capacity of the soppression pool.
- 5) The Alif5 3A laglementation have not been considered. '
SEQUtlett peacter subtriticality. ' fUNCIlonAl. FAlluRE , 5tguileCE EftNT -- The Reacter protectlen System (RP5) falls.
- ST51tM FAILURES -- The Standby Liguld Centrol (5tt) System falls.
i l i 5EQUENCE EVENT RPS -- St-6 ST5ftM SLC -- Il-1 i i ImeAVAILAtiLITits i SEQUffect EVENT 1. RPS succeeds and control rods are laserted late the core ER
'5T51tM SUCCESS 2. Recirculatten pusy trly and manual shutdown of the reacter estag 5LC er meneally driving In control reds.
j CRittRIA . t I I DOMIMANI f4C10R5 -- Failure of the operator to initiate the 5tt er to manually insert the control reds. l DRif fleG Siquil0CE - Failure of any three adjacent rods to insert. Coomen mode failure of the RPS logic system reselting from human errors in testing and unintenance. ea w- m m emuuhe es
, : I
- d f n n e o r t
neC o e mu r e . el l t s s s f o t sces t gsa ib r e e n e nnl w e jwi esa l e l i aie oevt nan t v aoF r r a n P h oIi ce it vo n i t3 t m eh e g e u gr a et h de nh et ebt d n t vg0 an-7 eess n T it rs e i hiG rti a aoo ei p d 5 s e t pd e e dt n .) nde d r orR n tmrleett)Qssaer ieaAP(lhl e li ro ec 8' a l c e t oU rc cI I i a r . sf p (datl v aaf t e n d ef aee "d n h t e ps o c af e .er tf sh #a pen 5 n rte)semA T , avo U a t s fQeve l n r e $ n - 3T et l e ( e l
)at li .
$ e at e1 ccs .ie . t b rvs m t -
ih e yl wbd a e r n b yst - orse t et et e " wa
'd l
e e stl t s e i e _ ahs t)pnva
- e i
a sea y e m _ d el v s _ e(il cl s oR eat
- e v rlE c l nNnmhi *f a Oe n r t h al Ron v t s a o lef I wf a f i s(ii sa saec "
'i n
m e (rp syo e i m n e f tl ny l st n a t n .
'm e i
aenI e " et r n w . l. 7a l i tl 1 et n
- e)5 v Coptvea reI v *f t
lf veG h t . i a os dt nelPmp f aas e m saew a ph
- o o f h i r i g v(Reuei s h p s n t v s 're o c n r ii sg
- se y o 2snst et em csabet aht su en t
- io n e fdn e
iew d e a n e f ya s . n:rul set aup *t
*p s l rO o d nf u
ert sl yM ha u. e m t n e pe _ mp g anreS l o mt ee cn. "rw Er ad es e rtt i eW g i a s o Ai l n ey e nardbl t " te gu yg0 n m l sF rhyouf e t r* i c neis s n8 ottid esit msce ul em tl
" In r as ea ftd l
o s
. r o s aa ii ,t f rsaslj li a e. o m f i rb hrT naeeesaao "t sel1 c e t e 1P- asvRhafl t el gt8 r u t d 5
'o h he td o
fai9 ott1 l e s y e s n We 5 I N
"I t fm o y e sn. e e
p s l o i a 5u 5d t l "W er p3 y )3 1 e t r c r rc l Ei CF p se t eC hRh h t o p s .t en lIi i r uv ttl ce p ie) e i ER ao n (r g u BnW cS UC 't cc er a .n s cd Q5 c ee . rotMsi rs an5 a5 ni ad Br m ft SO "i
*a t
.t a a
e f ef fl n4o he uw d t e t e( at rnm n es t r "m
*o eh vw t
s l - e3hl OEs1 o i a ti e at it ne
't
*u i
t n
.o yt Ii8oo E - wp c o
t i M msy ac ma
.f
*a co e
t t e so.Gt e G.On s s I I rS o r f
*n
*e jg eu ve R yo Wti a I nf r e f n ot
._ bn gd l cD O n s ui e aa es h dt a
- o. sd t e . wer t ) ea dn t) n ary vl o*
pt i ssW ei
- ea r t e ap l ur f 8
.t p w V o Ol e oo sm
*ue he )st so e iip n )1. uu t p . m Mcc i ld
*dy wd Sen s e ( c s 3. bos e l
/ e Cr e e2e t sv- e
" n-t2 me Rpv r v s y sir e e sh
- eE od sut (rn ei yI EG rK.ae h l
a s vAS it I nt
'i4
- s( si mv eo5 t
etr sIi m v t i l any An i
- n in t C f aed5a f M vib . na
_
- as g smR a nnCr e R adeit
" rC ea yo sdwaPf i
dr ncab
)
'TP cm n Srt f s eO l
e h e t et anro t at e . ef t o lt l al t 6et r t aiSnt o n r ~ -
~
uo q nnl ao amu 1. thea y .f rHl el etW e5t l i a e e ?3 Q es sr lih ot s en( so tB 3. r h t5o eC p hn5d onti5e s a
/
s y 'T sd e oci Cen orsR. evve l prl oo fPe l a r i amnu a s
' ir t e sea osl i b
R e n i t e " ho t y rol erp t pe cpel cm eat r yW t asrree ltt l hr tV oOsosn mMVfV o r s P s
- f r c r ot noo af e O Oi E n N oe at ti .ett f ot e eeMnMt ee w a S a v eno m A
>t r 1 3 y cn Retn a
ei l S. e e e rbDrrr onl op rroool hh wdwu m 5
- na f ie ua ieu eim ttt t c o 3 1 ec u ol r li m ull rt o uac fffif a s l r 2 qy au il o qll 133 sr ooo oc
, F < T er stl aereaa - - oee p
( V re sni eoa FrfRFF 0. E E l pn eeemee F
- f v LCF 0F3 COO nnnunh E
- Kl 0 t
l 3 2 ec o - O O O pO T 2 U l he - - - Q f 1 1 r - - - P QI PQI - - - - - E PQI - - - - - 0
- R 0 f t Y F l l Q R G
o P K t T N i S S SE ll E RC 3 u t 0 TE T l T5 ON d 2 q v E NR N f N5 n 1 E E EU E i tE TE CU a R E G N VL V L VC AQ r F G G N L EI E I EC G : N A 0 A A FE E E l E1 EN S U S I M EF E A E$A i C il t E T A D A l A lI K5 tU 5 COE le IR tTU C I M I EI KMll C e l MR I i nG AN f I iN ttA NI l l U u T E U0 UCL UT UfV ttt 0fi iE Q5 V A L E Q q t I N R O NV IE tl 5O QNI EUA QS EY Q5A Evn 05l tTR IMI OR _ P S S I E SIF SS SSU 5SC DD 4 i' il 4I ] , 4! 4 , : i - 4
d e r l . e r e e e) r 7 s s e t c r a a( 5 e e e r s r p s - rI e. e e oI d r h e p t hsyw e
. er . b h e g r rd) t i i
- b t s h r e( u e eh m e s pt n h s ipei e r t e dt e r oit t f p se c a o e eaj e e d el n r r y h lh i e l e l i sl l h l a cel t a u t a s e f s
. n a
eIf l t y e:o b n m l e a s ess l d e u p t o v s l n epe a e s r - l ml u e p m o - opa. nny s n-u d e e t e s d e eeb l y e ul e r s c gtd a e o e: e . h y r . sl ; n ss t a p
- e mm r sci . ee i s
S p t o . i rI l tt hio l ss Tcf e yy t n e r m e w y ss e o t
- v C s N b nn e y -
0 ee e S ll s r Ef l e tt i e n . p h cc h s em Ki t jj ee t s t e tR e nn n e r tt Ut d Ii as Q5 i i P ry SK E s n ee rr ee e) r . w o f s r eC i ss er l sI t e sC t ss lt r eR u ee rr y
. ac e r -
pp f a pr f t e t l n ee i tr Oe D l hw r n S a go g ae .l t cC e e s il h e t . ay it nh mc ee iP tH . m tj a eh l r oe Rsn me ht o dr _ m t f n g
)t 5n Ce ei rr pe O yI
.S t agn tt oh u
n ii Rv s l na Af o e s ( le es ail e l t u he rl o he t a m T r t ee t r a hg eS (p eC e l t n tC I e yt e i sI wl c el y .d a e ol r co St no rnr pee l a
)
t e no ec t o s et St s lf a oce l e u nl te eee t s n m qe a cr rl r t n i r er lh ee oop a l e a a so os jl Cs rn c oi nl I w oe s e r s Cn i a e o b a / y e ie e f rel tt an ht rl l er e s A t e ep l a soe ri ea R A l t e a pm y C O M nt cr a d sCh e t o r t L L it n eo f n oa rr
$- S di Pef fe A
l l e Rt es t o o e . m 2 5 a mo fe rm hc gae er o E 1 m s ut or ee F ien ra r < s sr sl e lt - Hpo uw l s t ld F ( : A ae sl ay S eey ir T - sr d ea Fs hh n aa g LF TTA FH
- - io - -
- K N - -
t S t U e l $
- 2. ta - - . . . - -
Il - - E E 123 - - Q i E W f
- R 0 _
T F l 1 R Q K T I S 5 t sE s d E, t U Q N E V N O C TE RR T N f f TS' NS eK lt n 5 E tU E i EE tU a r M F G E G N 1 1 0 A L yL tI V E L l VC EC AQ FE A G t N l E1 EN EF t U S t E E - A ESA T K5 OE l 5 t
- E K f A N I
l A lt tU KIR tTu KM tt KMl TEA l C NMR EEt I NG AN NI U0 n pt u t fu I E uCt U1 uT T Utt IV A q T it Q5 qNl Q5 q5A Qsi MI i I R nV E1 tUA E1 tY I N EfR OR _ L t t N O IE SD SIT S5 5S SSC DO P 5 S I C I l > ; ; ;j a (l l - : i 1 l
, g SiquillCE: T gtfV g
DESCRIPfleII: This seguenc; ls initt:ted by o transients due to a less . cf cffsitt power (i )i f:llowed by the enerallettlety of 1-5tqutIICE FREQUEIICV: CI-S(FromAstPRebasillnel the I:r,wer converstes systesi (O), fallero cf the high
- press:ro lxjection systems joy, and f:llwe cf the low . '
; - l#1TIAllINI Evtlli FRtqutNCV ~ 9,1 transients / year peesswee lajection systems uv '. -
CORE DAMAGE CONTRIBUil0N: 175 INITIAllNG Ig -- Reactor shutdown initiated by a less of offsite power. EvtNT i SEQUtilCE There are two 10UV type seguences possible: I Dl5CU55101l I) The early igUy where Nigh pressere Core Spray (NPC5) and Reacter Core Isoletten Coeling (RCIC) are lost tenedtately and
- the low pressure injectlen systems are lost either due to ladependent falleres of _the low Pressere Coolant injection
- 2) A (LPCI) and Low Pressure Core Spray (LPC5) systems or becewse the reacter coolant system is not depresseetted,.
later IQUV edvere all AC power is lost causing eventual test of NPC5 and RCIC systems and lamediate less of the low pressere lajection systems. The results presented here have been rebaselined with the fellowing changes: a generic the long teve IQUV seguence (see seque,nce event system enavallebilities).fregmency of 0.1 Instead of SiquillCE -- Less of Reacter Coolant System (IICS) lategelty. * ] ' FUNCil0NAL -- Fallere to replenish test IICS leventory. FAILURE SEQUENCE EVENT Q -- Failure of the Power Conversion System to provide easemp. SV5f tM FAILURES U -- Failure of the MMS or the 3CIC systems to provide high pressere meteep to the reacter core. V -- Failure of the LPCS or the LPCI to provide low pressure meteep to the reacter core, i SEQUENCE EVENT Q -- 1.8 Sf5ftM Independent fallere -- FE-3 l llNAVAILABILITit$ U[independentfallere)?--4E3 VL Diesel enevallability -- 3E-4 Monrecovery of offsite power one heer -- 0.45 4 i flonrecovery of offstte power la eight heers -- 8.12 flonrecovery of diesel la eight hours -- 9.7 1 UV given no recovery of offsite power er diesels in eight heers -- 1.0 (batteries are assumed to deplete in eight heers) 5tqutilCE EVillt 0 -- One complete condensate and feedwater path. 1 ST5IEM $UCCE55 U -- Elther the NPCS er the IICIC train. CRIllRIA V -- Automatic depresserlastles and any one of the low pressere pusys (LPCI er iPCS). 4 DOMINAIIT FACTOR $ -- One of sin meter operated valves is closed for maintenance er the IICIC puu, is down for anintenance. DRIVING SEQUtitCE -- Diesel #3 falls to start, diesel #1 falls to start. diesel f2 falls to start. ) -- Batteries deplete. ; 4 i f I i a i i
PLANit Lleerick SigutEt . . SEquElett: TC M 5CRIPileN: 1his sequence is faltleted by a transient followed by e . i telleet to achieve subtriticality. The Litwelcs PRA . . SEqutNCE FREQUEK V: Case A -- It.F. Case 8 -- It.6 lacloded thirty nine segmences that fit this descrtption. 1he results are presented in two cases: A) these , IFrom Limerick PRAl ~ segmences in which core melt is ceased by tellere to shot the reacter down, and R) these seguences in d ich the l#lilATlHE EVENT FREQUEKT: SE-5 transients / year centrol rods fell to insert, bet the reacter is shutdeun threegh alternate methods. Core melt eeenteetly eccors CORE OffemGE C9111Ritullful: tese A -- 15. Case 8 -- 95 in the Case R sequences due to falleet of other systees. INITIAf flIG T -- Any transient after which the control rods fall to insert.. EVENT Sf0UEEE --- DISCUSSION 5tqulEE Case A: Fellere to shot reactor down. FUEll0NAL Case B Fallere to mainteln reacter inventory af ter the reeCler was shutdown. I FAILURE 5t0UEEE EVENT Case A: Failure er verless costleettens of the Reacter Protection. Standby Liguld Centrol (5tC). Recleteletten , 515ILM FAILURES Pump TrlP (RPT). Feedseter liv). Reacter Core Isoletten Coeling (RCIC). and Nigh Pressere lajection (MPCI) systems (see success criterie). Case B: Fellare of injection systems or Residual fleet Removal (IBIR) systems af ter the reacter hos been shetdeun by alternate seens. I ' SEQUElICE EVENT Failed Systems or Functions STsitM SUCCESS CRIIERIA 2 SLC e 2 SLC e HPCI M5fv Transient 2 SLC Fu a f SLC e 2 SLC e fu e LEVEL 5 FW LEVEL I Initiator PUMP RCIC 1 IIMIt 2 RNil MPCI TRIP RUNRACK 1 RIP RPT 1URRillt TRIP A A A A A 11 A A N j 161V CLO50st A A A 11 A R A A N A: acceptable LO55 0F 0FF5ITE A A A N A R A A A N: not ec'ceptable POWER i lleADVERTEIII OP(Il A A A N A* R A A A RELitF VALVE ! (10RV) , j lleter the snelysis classifies this tes6inetten of felled systems for en 10RV as being not acceptable. , , The sequence eos determined throwgh onelysis to be acceptable too late to be changed in the snelysis. DUMinAnt Factor 5 rellere of rods to insert, j DRIVIIIG SEQUENCE J I
Ptalli Limerick SiqutKE SiqutKtt.'1gUV K5CRlpfl0N3 The iggV seguence for lleerick actually consists of foer . seguencss: fouv.1905, leUV and it W all foer seyenc:s *
- 5tqutKt FRtqutKT: It-5(FromLimerickPRA) seguences represert tresl%Ients followed by e . sets of Irjec tlen. This type cf seguenc2 inveires's relativelF -
fest core melt la en intact coatelement et low pressore. IN111Af flIE EVENT FRtqutKY: Tg -- 5t 2. Ty - 1.78
.ig -- 3.2 i g -- 3.W. keestents/ year The four seguences are described besefly beten la the table ender seguence discussten.
CONE DAMAGE C0lifRituilell: 361 INITIAllNE I=I s1g e T, 7 e Tg ehere Tg -- toss of Of fsite Power. E VENT Ty -- Meln Steen Isolet ten Velve (ft515) cleswee and less of Fee 6eter. Of fsite reuee. In - Menwel Shutdome. T1 ** I"'bI"' I'IP ' 5t0utKE IqW -- A turbine trip (T ),g e eenwel sheideus (T m) er a less of poser Conversten System (K5). (T,) treestent followed by e 015CU5510N fallere of er a fellere to recover the K5 (q). fellere of high pressere injectien (U) and'fellere of low pressere injection (V). 1 IquI -- Sese as TQUT escept that V becomes It fallere to actuate the Aetematic Depresserirettee System (A05) le e,tleely fashion. Tg W -- A less of offsite peuer transleet (f,) followed by less of all injectlee (W and V). This sessence is the sost probable of the four types and incletes the station blocteet seguence. Tg us -- TEfollowed by less of high pressere lajectlee (ts) and a fellere to acteste the Ass la e ( M y feshlen. All seguences esceptgi W result la early felleres of core lajection le i W seguence, g the high pressvee.lajectlen systems (Migh pressere Conlant injectlee and Reacter Core Isolettee Coeling - IIPCI and KlC) can conceivably last for more thee fear homes. At sese time greater thee four hours, the betterles elli deplete end control of the teettne driven HPCI and KIC
- systems will be lost. Failure of lePCI and MIC is essened at this Else and core seelt will eventeetly ector.
StOutKE -- Less of Reacter Coolset System (K 5) lategrity.
; FUlstil0IIAL -- Fallere to replenish lost K 5 inventory.
FAILUlit a SiqutKE EMNT q -- Fallere of K5 to provide esteep ester. ST5f tM FAILDK5 9 -- Fallere of the IIPCI and the MIC systems to provide high pressere meteep to the reacter core, i T -- Fallere of the low pressere Core Spray (LKS) and the low pressere Coelent lajectlen (tKI) systems to provide low i pressere makeup to the reactor core. I -- f ellere to depressortae the reacter vessel et the appropriate llee (i.e.. fellere of the ADS). Stqut K E EVENT Tg 7, T Tg (Actual T, calculetten teclodes fellere of diesels end F ST51tM neerecovery of diesels and of fsite power).
- UNAVAltaelliffts 4 -- 2E-2 7E-3 2t-l (nearecovery) I
- W -- 4.9t-3 4.9E-3 4.9t-3 4.9t 3 (for fece hrs -- 1.9 after four brs if no K ts avelleble)
V -- 7.7E-5 7.7t-5 7.7t-5 I (if no AC) I -- 2t-3 ft-3 2t-3 -- l 5tgutIICE tytNT 4 - One Feechseter (ful sad one condensate pump (If RE >540 pslo) and one M5tv ceen (se that fit con operate) and low vecem j $151tM SUCCESS Interlocks on M5tvs and en bypess valves overridden ti condenser veceum f alls below 7* bg. , CRittRIA U -- lePIC or KIC.
- y -- One out of foer LKl pesys or one out of fece LK5 pomys.
3 -- Operetten of two est of five A05 volves. DOMINAf1I FAC10R$ -- Operator falls to lettlete A05 in a timely menner. . DetlflNG 5tqut K E -- 1renstent results la e less of ladicotten f or need of A05. { -- Fallere to restore offsite pe=er with thirty sin. Commen ande fellere of diesels.
-- Fallere to recover diesel generator withle thirty min.
j -- leerdware felleres of HPCI turbine.
- -- Hardware f alleres of KIC turbine.
I
PLANT: Limeritt SE9utMt
- 540utME: TM C 5CelPflet: This seguenc2 everesents a treestent (t) cith a stect .
open relist wt e (P) la solch all lejectlen tills (fl. Sgquiert fatgutMt: N-F (from timerict PRA) the let seguents tee timerict else octuelle consists of .. four seguences: IPgut. Irgue, l.Put end I pers. Inese . Infilallas tvinf FMqutMI: i g .. St .t. f -. 1.73 seguences, escept t PUe, all evlelt la eehly fellee, et , I g -- 3.2. 1 g,-- 3.95, treasleets/yeer core lejectles. fof I pww. g high peessere injectlee * (Migh Pressere Coelant lejectles and Beectee (ere COAT BMWGt CINIltitull0N: It lSelettee Coeling - MPCI and DCIC) mer operate estil bettery depletten lpeebe6fy e four heers), less of le-jectlee end subsequent core melt ses essened to acter et that llee, Inlitallus I=Tg + f, e i, e fg where f tithi )g -- toss of offsite Pomer. l l ,, MesseI Sheldene.-- Male stese Isolellen telee (N519) closere end toss of freeseter. Ofistle T,' -- tertine Irlp. SiqutME St%U551918 IPqtiv ..stect A tertlag tripgvetve open relief (I ). eend naamel e fellereshutdeus (f to of Ir a fellere I er e im the eetever of hPCSCeewesten (Of, fellere of ystem MS) Ug) elgtreesteet pressere h feHoued lajectienby e . (U) and fellere of leu pressere lajection (V). IPOWS -- Some es IPquT encept that V becaers si . fashlee. fellere to acteste the Astematic Depresselrsettee System (ADS) la a timely Tg PUT .. A less of of fsfle poner treesleet (ig ) fellemed by a stoch spee tellet velee end a less of all lejectles (4 and 1). i This seguence is the east probable er the foer types and lecludes the stellen blotteet seguence. i gPUE -. T g fellemed by less of hlgh pressere Injectlee (U) and a fellere to eclease the A35 le a tesely fashion. All seguences eaceptg I f9V teselt la early felleres of core injecties. In the f PUT segmente, the high presgere tejettlen systems (WPCl and #CIC) ces conceleobly test for more thee foer beers. elli deplete and control of the turtlee driven MPCI and SClC systems elli be lost.At some flee geester thee foer heers, the betteeles this tlee and core melt will eventually octor. f ellees of MPCI end DCIC is essemurd et SEOutIICE toss of Beector Coelant System (ACS) leventory. fustilenAL -. fellere to repleelsh test DC5 laventory, f48tWRt ' SiqutME IVtNT P feltere to reclose all safety reIIef selves. ST5itM FAlltet5 4 .. feIIere of posee conversloe system to provide esteep. U -. fallere of the NPCI and the Mit systems to provide high presseee esteep to the reactee core. V .. letters of the les Pressere Cece Speer (LPCS) and the tem Pressere teeleet lajectlee (LPCI) systems to provide les pressere esteep to the teetter core. 1 I -. f ellere to depresserlee the reacter ressel et the appropriate time (f.e., felleee of the Ass). i Sigutett EVtNI Tg f,g Ig Stilta I g (Actuel fgcalculetten lacledes fellere of diesels and tasavaltaeltllit5 P-. ft.2 een-rece,tey of diesels sad offsite powerl. Il-f it-f it-P 4-. Pt-2 Ft -3 7t-l (non-recovery) I } W-- 4.9E-3 4.9t-3 i
- 4. 9t - 3 St-3 (f or flest two bes. - 3.0 ef ter tue boots If me I F.Ft-S AC is ovellable). '
V.. 7.Ft -S F.M-S I (If as AC). j 3.. 2t 3 Ft 3 ft-3 --
- Silpptlett tviel P .. Closure of all safety eglief volves. 5 -- HPCt er RCIC.
5151tn SUCCt%5 g .- thee feceseter (f t) and one condensate peep (if St
- See W .. De set of four LPCI pesys or one est of four pumps.
CRittRIA pile) ead one M51T open (se that fu con operate) and 3 .- Operettee of toe out of flee ADS velves. I low seceum Interlocks on N51Ts and on 6ppess velves j overrieden if tendenser vacuse f alls below F bg. i 90Risett F ACIORS -- Operater tells to teltlete ADS In a timely eenaer.
- teiltinG 5tQutleCI, -- leansient results la e less of ladication for need of Aps. !
-- f allece to restore of fsite power with thirty ele.
l -- Commen mode fellere of diesels. a f allere to recover diesel genereter olthle thirty ele.
-- leere. ore felleres of 18PCI feeblee.
j -. Heremere fellere of DCIC teetine I
,1 1
L i - _ . . - - . _ _ _ _ _ - - _ _ _ -
,e o PtAMit Segooyah SigiftuCE .
SElpitlICE: 1 K 23 M50RIPfiluer This seguenc'> ls initi;ted by ear treestent Cscept less - . et of fsite power levolvlag amtssetic reacter trip with es , SEllutKE FilEllutKV: 3E-6(FromSegueyehR55MRp) laterruption of sein seedmeter lajectlen (T er say -
- translentlavetolagestamatictrlywithas$s)la~feedseter .
lillflAllHE event FnEqut K V: 7 transleets/ year laterruption (1 3). This is followed first by fellere to; recover the Power Conversten System (MS) (for i CORE DAftASE CONTRISWil0H: 31 transleets) er by M5 heresere and other system Ialleres for I, (both are referred to es M): and followed, second
' by f aViere of Assillery Feedseter System (AfW5). (t).
1hese falleres result la less of normel and emergency seems of suppfylag water to the steen generators. The l' steam 9enerators bell dry the M5 pressere lacreases. - water is discharged through the safety relief volves and
' the core encovers and selts.
4 listilAillIG T23
- I2*I3 I
'" **'***"* I'IP"
- Evtut - 1 I"' I "'
f3--1reasient due to an estametic trip without laterruption
'"" W "I'" 'I "*'" I"d"* "" I3#FI' of main feedmeter (4/ year),
SEQUtleCE 915095510H t ,.; e may be reduced more then en order of segnitude if feed and bleed credit can be given. It does depend, however. en precedures and specific plant characteristics such as evallability of poltV support systems during feed and bleed operetten. Without such information. this insight cannot be applied at this flee. ' SEllutleCE Failure of core decay heat removal to the ultimate he'et sink. f ulatilonAL fAILINIE t SE'JutNCE EVENT M -- Fellere of the MS. SYSTEM FAltunts L -- Fallere of the AtW5. 4 5tilutleCE EVElli M -- IE-2 5f5ftM L -- 4E-5
- INIAVAILASILiflES SEQUEIICE EVENT M -- M5 Operettenal.
J 5T51fM SUCCE55 L -- 1. One of one steam driven pump or one of tuo electric pomys AND i 'CRittRIA 2. One of five safety volves per generator er two of four steam relief valves. 00MINAIIT FACTORS -- Failure of single lseletten valve between the condensate storage tents and the AfW5 peeps. DalvileG 5tquiteCE -- Tallure of the Essential Raw Coollag Water system. i f J 4 i I . i k_______ _ __ _ _ _-_ _._______ _ . - - . _. _ - - - - - _ _ _ _ _ - . _ - - - . _ . _ _ __ - _ _ _ _ _ _ _ _ _
.l ptAni: Segueyah 5EglItaCE .
StqtitlICE: 5 NF Ot 5LIIIPf leII: This seguem e is initiated by a small 80CA having an 2 egelvalent diameter between one half inch and tuo laches StqutIICE FRtqui1ICT: St.6 (From R$5 MAP PRA) ReElecoletten Systte (tCCRS). (H) and the Centainm INITIAf flIG tytti FRtqutNCT: 7t.3 SLUCAS/ year Spray Recirculatten System (CSR$). (F). 4 CORC DAMAGE C0lliRIOUTION: 51 1
- IntilAlllIG Sy -- Small (LOCA) break of the slae of one half Inch to two inches.
IVENT , StqutlICE i Ol5tU5511NI Falleres of the ECCR5 and the C5R5 are ensinated by a common mode centribetor. Between the opper and lower contalement ' compartments are two drains that must be closed during refueling operstlens. The ICCf6 and C5n5 draw water from the susp la the lower contaisument. the soup for forther recirculatten. The water that the C5R5 sprays late the opper compartment must pass threegh the drales to retern to cespartment to the opper cespartment. Closure of the drains useld cause all water to altleetely be transferred from the lower i Atteepting to draw water from the dry Susy would fall both ECCRS and CSR5. i This s to Rest e could be reduced by an order of segnitude er more if the operator can depresserlee deelne the lajectlen phase at Heat after refeeling Renoval have been leproved. witheet gelag to the recirculatten sede er if the precedures for deoble checking the deeln plegs. (TVA has layreved precedures for checking the drale plegs.) SEQUtleCf -- Fallere of Reacter Coelant System lategelty, funCil0IIAl. -- Core decay heat removal during recircolatten phase. FAlttlRE -- Radleactielty reseval frees contalement during recirculatten phase. 5f0Utlett EVENT H -- Fallere of the ECCRS. - 515ftM fAltuets F -- Fallere of the C5RS. 5tquinCE EVERT leF -- 3t-3 t ) 5YSTEM } tmAVAll ASillif t5 i i 5(IIUtlett EVtWT ST5ftM $UCCESS M- 1. the of two les pressere Recirceletten System (tylts) AND 3 CRIltRIA 2. One of two centrifugal charging penps or one of two safety lajection pengs. AIIIe '
- 3. lue of foer heat enchangers corresponding to the operating tpRS.
,I
] F -- 1. One of two contalement spray pusys. DOMINAIIT FAC10R5 Operator f altere to rconve plegs from intercampartment drains following refuellag futivileG StquffECE 4 4 i 4 J i .
. 1 l
l 2 - s ir . +
.: z .t.g jt$2 EI J
' =
J. lE :
- e ! #
j }*.,l3 Iplf i
- 1. )
if g$s# I. I2 21 ,, I' *-$ ag i j.- 8 1 g{ a sy1 i l3:i l i {. i.I [I }- s g Ms. M:.3 h'l .:
$Ds :: -
1 1: isa!! t . S t s,. ss 5 Ssg i s i s. l sy-; 4 L =Jg: - ii:i 3 9m. islJ 3; 3.
-s t;
u, x, a1 3: ,3 g s;l 5 !3
- {s.
s e, li1 e sI. =.2*g,Ilji 2!. i2 3! J t
*ii1j-l r=:s t:f g!I .: i 2
e
- ee-s; J
3 i gJ .gs lI :1 2 sr-I sjjj-2 -
! i !!
II l} JJfli. J g* ! j,dglillgD: i 11 lig e s ,g l 1
- -s :
l > >,s st..
- - E
.1 ,a .
, l.,
- y =<2>
55,.; 3 = !g ; 2.Itursc35! J-I - i- , .
- 31 . .i rr
- n i
j. E,sii
~'
,, -1 M
s-J 4 ts 22 ea a :=
]
gg 1 :! x v ; I
,3 Je > ; il jj s t t ss i - 51 l 1 1
. 3
}
t Ji,=n
=. 8 ,
'!]s-i ,
[g si Ili - !. s.
.i ,
<s-t a
f;
- ~35 g2 5:
} i 1l 5 5- s-i ,s ;l* ;
$} -
i.
, .: 1; . ->=1 >;g!
1,. 11 o1 j h* I- i ist ,,,1 i -3 s 53 is i 131: si
- i s3 .!';3 r
] '!
priI.,s l;: ! <: =: - . .s n in 1 [ g
- =-3 g r;s 4.l !
13:* ,1 **- =: I..;!ri it
.= !! ! gg (
- I I "'
f*lff {:d' d -**aIj
* -u -
t Ig 3 I f f :..)I; != ju?? o .:, II i s I:: II I
-- 3 #
. 3:'I8 ' ' ::
32[i, 22
-i i pII. I ji-
, i :: Il l r
- I $ h$ l 'f$!!f j'fffj[jf j f!f!jIj , _,
lnl3 1 In
!! 15" !!
,l J
![
f I II
,e Isl I!r III ti le v!l
==
i! lss
.sa a,
[v>
$3 l
l I
a.. d PLAllt: Sorry - SEllutKE . SEllutKE: V(InterfeclagLEA) OtSCRIPileN: treach of high pressere reactor coollag systte bewaeery et its laterface eith the les pressere cooling system.
-SEllUE K E FKflW K v << E-6 (From ASEP Rebesellae).
ItlllAllNG ETEIIT FaEllllEKT: << E 6 It K As/ year
. CENE BAMAGE CeuttitiffilNI:
- 15 j IWlil ATIM t EvtNT V - Fallere of Lee Presswee lajectlen System (LPl5) chett velves separating (Pl5 from Nigh Pressere lajection Sys, ten.
SEllut K E A portion of the Emergency Core Coellag System (ECCS) eses deoble check volves la series as a berrier between the LPl5, 915CU5511NI which is setside the contalasent, and the high pressere Beector Coolant System (KS) Witch is laside the cc-telament. 1 tallere of this barrier would result le reacter coolant discharge to the LPl5 eetside the contalement. The LP15 weeld thee fall due to everpressere and result le a core selt. leien the components of the low pressere lejectlen system are subjected to high primary system pressere, fellere is espected ' to eccee et some pelat le the system. The locatlan and mode of fallere is oncerteln. It is possitte that same delay la core sacovery could be achieved by operatten of the high pressere lajectlen system if this system is operettenal. Because of oncertefaty le the response of the system and the ability of the operator to properly laterpret and respond to the accident, it was essemed la the searce term saalysis that the active components of the ECC5 are laeperable. SEllUEEE -- toss of peacter Coolant System (KS) lategrity. fuMCileNAL -- Fallere to replenish lost K5 laventory. ' i FAILUht ' i SEllutKE ENNT V -- toss of barrier between K 5 and LP15. . Sv5ftM FAltWRES i l SEllUE M E E W NT T -- << E-6 (based on testlag and monitoring LPl5 check valves).
- 5T5IEN
- i UNAVAllABILITIES SEllutNCE EMNT All three LPIS trelas isoletet from K5.
- ST5ftn SUCCE55 CRITERIA i
I 00NINANT FACTORS Failure of toe LP15 check valves in series. 1111IVIN SillutKE , } i
1 . ptAni Sorry StqutKE 5tqutEE: As M5tRIPflen: This seguence Is Inittsted by a large LEA heelag ee egulealentdiametergossofpowerpreeentsthereater less of poner. The thee sin l 5tqutKE TRfqutKV: . e (" t-0). (From AStP pebeseline)
" Englaeerlag Sefetr Festores (Ests) from operettag.
Less of E5fs result la fellere of Emergency Coolant INITIAfl!IE.IVilft FRfqlWNCT: IE-4 ILOCA/ year Injectien (ECI) dich is needed to mittgete the occident. The (Cl functieaal requirements for a large LOCA are the t0Rt DAMAGE CtNIftl9Wil0N: e il accumulators (onef fected) and the law pressere lajection N, system (felled). . 5 f y IntTIATING A -- targe LOCA (pipe breek) > sin inches. ., EVthi , > 5tnutntt .--- 4 DISCUS 5 ION - 4 4 SEQUENCE -- less of Reacter Coolant System (RC5) Integrity. * ' FUNCTIOWL -- Fallere to replenish lost RC5 latentory. ,, FAltURE SEqut KE EVENT B - . Less of offsite and eastte power. N 7 '1 ST5f tM FAltUnt5 r SEQUEE E ETENT toss of offsite power -- It-3 SV5ftM Less of easite power -- 7t-3 [ llhAVAILAttilitt5 8 -- FE-6 ~
~
;_. , +
. ..s l ~ 5fgutKt (VtNT Sufficient AC and M power to emergency buses to ererete the slaisum required 15F se6systens which org:
Sv5ftM SUCCESS ~ i ^ CRlitRIA 1. the of two law Pressere laJectlen System trains AW j _, F. Two of three accumulaters I. passive). ' m ... ~ x s ... 00hlNAMI FACTORS Importance calculatleas were act done for this sequence by A$tP state A5tP did not find it to be dominent. i DRIVING 510UtKE ~' j- g 's
^-
1 1
% A
- g
% . ~ . -N ,- ., o'.h'-'s -
~
e
D PtANT: Sorry 51QUtEt OtSCRIPIl0N: 1his seguence is infilated by a sanall LOCA having an SEQUtKE: 5D 3 equivalent diameter of less than two incks (a practer Coolant Puey (EP) seal LOCA) felleared by a fallere of SEQUENCE FRtqutNCT: 9t-5 (Frts A5tP mebaseline) toergency Coelant lajectlen (tCl) system. INiilAllM EVENT FREQUEEV: II-2 SLOCAs/ year CORE DAMAGE COMTRIBUf10N: 585 4 INITIATING 53 -- Small LOCAs (< 2*) caused by a reacter coolant pway seal fallere. The frequency was based en operational wata. EVENT SEQUtKE 530 is a new sequence. ASEP espanded the sorry 2
$ D (which is defined as < 2") late two break stres: Sy and 5 where DISCUSSION 3
- $2 -- Small LOCA not including RCP seal leak.
1 53 - Small LOCA caused by a EP seal leak. SEQUEEE -- toss of Reactor Coelant System (RCS) lategrity. FUNCTIONAL . -- Failure to replenish the lost RC5 Inventory. FAlttmt SEQUE E E EVENT D -- Fallure of ECI: fallere of high pressere injectlen, gepresseritation of the primary system using the Ausl$lary ST5ftM FAILURES Feedwater System and subsequent low pressure injectlen is not considered. SEQUENCE EVENT 0 -- 9E-3 SYSTEM UNAVAllAtlLITIts
~
i SEQUENCE EVtNT One of three Migh Pressere Injectlen System trains. ST5f tM SUCCL55 CRaitalA DONINANT FACIONS -- Failure in sensing a low temperature conditten of the operating Beren lajectlen Tank (RIT) beater. l t*IvlNG SEQUE E E -- Hardware faults leading failure of alternate means of sensing sli law temperatore. l 1 . i
D qth e . vi* i j. i jsI.sf*liI, -: _ji.i r
.Ia s
i l23
- ,3 Ej s
r 3 :: Al*s:l ::s=I I ! .f. - 1 2 L L!t s[ji,y ig [5 e ev1. 3 :s *il*i ,3 jv :=1 1 [ ,-! .: gIl- .I gr ils us .1 we .[i :-1 [i*12tisil]i agi,I.
- Ss- :: 3:*r J sigi:I;lgh
!= :: l
~
ill [J] I aj s,s TE ' al dd slJ 44 p (.y
** jgizisi . 3:i {I-i s2 ] :t
=
S jj.ft.118!j 3p
- i i 2 ,. 5 2
l!! L 3 3 I :; I.51.2'31'.t J212 .81 g I (l'1 ,I . a.~ I ! 21 2 I
]E11 !s-{lsi l As := l , 2 si
--se i
s 1.1 - -:- :, 1t !, J .
! ! i
,12E2'I ng: .2
-J' Is r-JJ ' I t
- I d l.i 3 t 1 41
- 15I'Il'4: i:
IJI: 3:! :3 5!
=z2 i 15 *, 5 i ,ii i ell * ,,
il21 i 1 3I 5 g - lii'
*3 E i
' 8,;I.!
.9 3 !* 1'1 1 53 l{
It - i ;4 : JJ != it on i i. =- [ 2 ee !!
. I a it
- 1 :11 s i 1. 2
- 1. is i ,.i j;i at-t i [-
-[ l 1 !. !
1 !ji: ~ , . it i. is fs
= <
. b I
- I iii s-lIE -Ia .I 1
5 5 2 j l2 8
-l 15 i Il ! : i,j i s.t.
23
-3 E: !! 2 I!'j s.
l i
-]I!i. I.I i
a 1,
-!! 31 53 sti it i
- 1I si 3
Ii ji l l1} II 8! I ss l ! "3: 1
.i
{f r is:-l=
.lis! :
l } is; 111 :: e. 5* g ,*.1; l:- 5-i
- f 13 jj
- - u 23 U **
- I i
~
s.i3;- s :II gs ilj'j j j !!*:! 1: 63 * [ 8: 2 .J l'{' g - f j un te : 2 :. 1;. I 133 :- 5 2
.. si
..jgf
.
- s.
3;,_gr [ s g.; ; g 13 3:. 3 ],3 ; ; -
; . {}s it *: * $$ .I .
E51.[..i 1
..? ?. .=
? 8* *h u 7: r, i :
- 3 i ,* i g :;,t .
b 3 al 3 g 3 33 3
- s. E am.& J a.h J J J J 41 I a
' .'l s t gi 1 i E E
5 E s! E sti un E8 II
*I I{ f:.=
. v .
I E X a II:siIa 3 35 g 8 g-g
*Ei. !! t b.1-1?>i e
) 8..
[ = =s mEE lva li
~
o PLAIIT: lien 5tquilICE Y. ' 5tquilICE: 50(SEFC-lessofCaspenenttoelingHeter) M5ChlPillNI: The IP55 reelew identified the most duminant segmence as
- 2 ene initiated by a less of the Camponent toeling Water 5tqUtlett FREcutilCT: 7t-4 (From IPS$ Reelew) System (CCW5), melch falls the charging pumps and peactor Coelant pus, (MP) seal thermal barrier cooling. la this INITIAlllIG EVENT FRtipffleCT: 7t-4 SLOCAs/ year sitwatten, the DCP -seals are assumed to fall in thirty-ftoe min. The safety lajectlen pumps will actuate CORE DAMAGE CONTN10Uil0N: $05 en low peacter Coelant System ( ACS) pressore. The Safety lajectlen (51) pumps will fall la stout ten minutes due to the less of CCW5. Primary systems esteep will be lost. Core melt will octor unless coollag to the SI-pas ,s or charging pumps is restored within forty-five minutes.
INITIATING EVENT
$2 -- less of the (CCW5) which leads to a small LOCA.
.5EQUENCE DISCUS $l0lt NopipebreaksmallLOCAinitiatedlessofinjectlenseguences(5f)werefewedtobedominantbyeithert'helien Probabilistic presented here. TheSafety Study assumption t er(IP55) the IPS$ reelew (NUREG/CR-3 jus . Informatten en the less of CCW5 sequence is therefore hat a catastrophic seal fallere ecc)ers due to lost of seal cooling is in this sequence. 1here is informatten that shows that a catastrophic seal LOCA (1200 gym) may not occur la Westinghouse pumps, if no catastrophic seal f ailure does not occur this sequence would become bogus.
5tqutlett -- Failure of 11C5 Integetty. ' FUNCTIONAL -- Fallere to replenish lost 905 Inventory. FAILURE
- 5tqutlett (VENT D -- Failure of $1 pumps caused by failure of CCW5.
ST5f tM f AltUnts ' i SEQUENCE EVENT toss of CCW5: SYSTEM 0--7t-4(lessofCCW5) UNAVAILASILITits a SEQUtleCE EVENT 1. Two of five CCWS pumps AND 1. One of ftve CCWS pus,s AND a ST5itM SUCCESS 2. Two of sin SW5 puses. 2. Ihree of sia 5W5 pusys, i CRIIERIA
' doth success criteria result in steller nus6ers, i
DOMINANT IAC10ll5 -- Pipe rupture in CCWS. DRIVING 5tqutNCC -- Haresare er Maintenance eutages of CCMS poses. 2 -- Failure to recover the CCW5 1 1 9 j i ]
o PLAlli: Ilon . StqutKE . 5tllUEIICE: TRS' (5E - selsele Indeced) M5CRIPfleII: In this tegnence, a selselc event accors that is leege enough to fall both offsite power and the service water-5fgUtleCE fatqutIICT: St-6 (from 2P55 and IP55 Review) pumps. fallere of the service water falls the diesel generaters due te lack of cooling. A less of all AC InlilAffilE EVfili FRtquiliti Different probabilities for different poser (statten blactr.et) ectors which results in a
- magnitude seismic events f allere of peacter Coelant Pes, (aCP) cooling, a Ep
. - LtKA. fallere of safety injection and fattere of the Cent DAMAGE CONTRIBUil0II: 15 contalament systems. Core melt results.
IN111AllllE EVENT i -- The Initiating event is a selsmic event large enough to fall both of fsite poser and the service water. ' SigUENCE 9150U551011 This segmence does not strictly seet the description of a IllLt' sequence. PCS (M) created. er AFU (L) is irrelevant because it is assumed that the MP seals fall catastrophically (1200 gr However we have talen IRS' to represent station blachest seguences. core melt with fallere of the contalousent systems ehlth is that was analysed in the searce ters study.The It assenes that a stspran catastrophic IICP LOCA occers within one half heer af ter less of cooling to the KPs. decrease if the assumptlen that a less of ACP cooling caused a catastropfalt seal fallere mere not mode,The probability of this seg i SfgUtleCE TUleCil0RAL -- faltere of Reacter Coelant System (RCS) lategetty. FAILUltt
' -- fallere to replenish less K 5 leventory.
-- f allere of containment heat removal.
-- Fallere of redlesctivity reeevel.
SEQUENCE EVENT M -- NA 5f51tM FAILINtts L -- NA l'-- Fallere of offsite and onsite amergency AC system leading to failure of safety injection and the contalse SigUtilCE EVENT
$f5ftM 1he the leveldescriptlen of the weavailabilities (campenent fragilities and sognitudes of different selselc initiators) goes bey of this description.
UNAVAltasittilts 5tquinCE EVENT 1. Any AC bus for front Ifne systems Ann ST51tM 5UCCESS 2. Seppert cooling systems regelre two buses between two enits. : CRil[RIA > 1 DonlNANT FAC10R5 -- The seismic faltfater. DRiviteG 5tqUtnCE -- Fragility of service water pony shafts. ' i 1
4.';.-. .
.i APPENDIX'8 TECHNICAL BASES AND USER'S MANUAL FOR SPARC --
A SUPPRE55IGN POOL AEROSOL REM) VAL CODE PC Owc:arski-RI Schreck AK Postaa* Batta11e's Pacific Northwest Laboratory
*Benton City Technology
- ?
T l 1 _ _ _ _ _ _ . .}}