ML20127M172
Text
{{#Wiki_filter:. T /E2 E/5
/ o
~g UNITED STATES
!" o . NUCLEAR REGULATORY COMMISSION 3 i ADVISORY COMMITTEE ON REACTOR SAFEGUARDS OFFICIAL USE ONLY o WASMNGTON, D. C. 20555 PREPARED FOR ACRS USE ONLY Yg ,o*/
FOIA EXEMPTION 5 February 18, 1982 MEMORANDUM FOR: -M. Bender, W. Kerr and D. Okrent, ACRS Memberg
~
FROM: W. L. Baldewicz, ACRS Fellow _ c/c
SUBJECT:
NRC Staff Questions on Severe-Accident Phenomenol of the Zion PRA Background and Introduction At a meeting on January 28 and 29,1982, the NRC Staff and their consultants questioned Commonwealth Edison and their cons 0ltants regarding details of the' Zion probabilistic risk assessment (PRA). (The meeting agenda is attached). The technical content of the discussions is probably of interest to, at least, the ACRS subcommittees on the Zion station, Class-Nine - Accidents and Reliability and Probabilistic Assessment. This memorandum is intended to highlight what seemed to be important issues and questions raised at the meeting. I present these comments with the understanding that they are tentative, since neither the Staff, the ACRS nor I have thoroughly reviewed the Zion study, which is proving to be not easily reviewable. . Summary items The NRC Staff is questioning the Zion-PRA analysis on these items: the amount of hydrogen produced during various core-melt accident scenarios the likelihood and degree of core-melt coherency the reliability of the containment cooling functioning particularly in severe-accident environments the probability of significant failure in the containment isolation function the ultimate failure strength of the containment the probability of safety-valve failures (sticking open) used in transient event trees 8505230199 841002 PDR FOIA SHOLLYS4-243 PDR
.I , . - 0FFICIAL USE ONLY More detaiJed discussion of these and other points is included as attach-ment 2.
-In the next few months, I plan to complete a review of NUREG-0850 Vol.1
(" Preliminary Assessment of Core Melt Accidents at the Zion and Indian Point Nuclear Power Plants and Strategies for Mitigating Their Effects," Nov. 1981). I look forward to your comments and suggestions as to how I, the ACRS and the NRC might maintain a firm grip on a lifeline with reality as we drift further into severe-accident space. cc: ACRS Members ACRS Technical Staff ACRS Fellows - J. Meyer, NRR/RSB (P-1132) S. Newberry, NRR/RRAB (216)
.. t ATTACHMENT 1 SPECIFIC QUESTIONS / ISSUES THAT SHOULD BE
, ADDRESSED AT THE ZION PRO?ABILISTIC SAFETY STUDY MEETING OF I JANUARY 28, 29, 1982-(The numbering of questions here is consistent with the numbering of the agenda items.) III. Plant Description III-1 Provide details of the obstructions and irregularities in the reactor cavity that might affect the proposed sweepout model coolability considerations and Basemat penetration evaluations, such as the sump and the instrument tubes. III-2. Provide best estimates of the water depths in the reactor cavity at vessel failure for the dominant accident sequences.
- III-3. Describe what pieces of equipment or system components, if any, are located within the line-of-sight solid angle from the instrument tunnel.
III-4 Describe the location and relation of fans, filters, cooling coils and ductwork for the containment fan coolers. Will the system continue to function if the filters are clogged? III-5 Describe the geometric features of rooms, cells or general regions into which hydrogen could be introduced. IV. Hydrogen Production : In Vessel IV-1. Discuss hydrogen production from al.1 sources (in-vessel) including the probability of producing those amounts that deviate from your best estimate. IV-2. Concerning the core heatup, dryout, and initial melting stage of the in-vessel accident progression, you assert that the core heatup and melting evolution is incoherent and results in only small amounts of hydrogen Being produced ^. Ba5Ed on our assessment of flow velocities and flow patterns and core heat transfer, we believe that a' much more . coherent heatup, oxidation and meltdown will occur. Considering this, address the following: (a) justify " oxygen starvation and hydrogen blanketing" under the turbulent conditions that will be present in the core region (e.g. you estimate several centimeters per second steam flow (p. 3.1-16) while we estimate'several meters per second). (b) do you consider natural convection processes in the core region which would tend to cool the hotter portions and heatup the cooler and thus tend to produce a more coherent heatup and meltdown.
; '~
'I -
-c-F-
l (c) how do you account .for the heat source from Zr/H2O reaction which could possibly reduce the incoherence by adding heat to the core periphery where the core is not steam starved? IV-3. During core slumping in-vessel we would expect larger amounts of
. steel to be present in the corium mix (lin addition to larger amounts
, of molten fuel). Explain how you determined the amount of steel present in.the mix which enters the lower head?
IV-4 We estimate a large amount of hydrogen. is being produced as the core slumps into the vessel head. A portion of your argr.ent that only small amounts of additional hydrogen are produced is that the particle sizes will Be large and that hydrogen blanketir.g will occur
-- both minimizing hydrogen production. Further justify this assess-ment in light of corium fragmentation experiments (which have yielded very small particles) and the non-prototypic (quiescent) .
characteristics of " hydrogen blanketing" experiments. , IV-5. In your study, the oxidation of zirconium prior to core slumping
- is estimated most probably to fall within the range 20 40". of the ._
core cladding. Sest estimates of the TMI oxidation however range from 40 to 60".. We Believe this latter estimate should be given considerable weight in such estimates. We would like your description of how much weight you have given to these estimates. IV-6. RE: Section 3.1.1.3 In this section you make a case for in-vessel coolability. For the various sequences considered what probability do you assign to the possibility of coolability (containment event tree mode "H") and how does it a ffect final ' conclusions.
F
. 4 V. Hydrocen Production: Ex-Vessel V-1. Discuss hydrogen production from all sources (ex-vessel) including the proBaSility of producing those amounts that deviate from your Best estimate.
V-2 Describe in detail the physical progression and final disposition of the 50% of core raterial not involved in the vessel failure and cavity dynamics described in Section 3. In particular describe the hydrogen production from this mass of core material . V-3. Your description of the effect of fuel ejected 'r:m the vessel assumes that the concrete is a rather benign . e t-tal . On the contrary, experiments indicate a vigorous gas sc:1ution from the concrete which could alter your postulated well ittaved flow and . interaction patterns in the cavity. What is your assessment of the. effect of this concrete outgassing, in particular as it relates to the subsequent accident progression and the hydrogen and carbon monoxide production. . . . V-4. Summarize the amount of meta s (wt % a~nd total nass) in the corium entering the cavity for the various cases considered. What is the effect of the instrument tube and supporting steel
~
V-5 obstructions on the Slow out from the cavity. V-6. In your analysis, vessel penetration times are relatively short. We would assume that because of this, there would be solid components in the ejected corium. Yet your analysis of the removal of material from the cavity assumes a fluid form. Explain this apparent inconsistency. V-7 We note that the models for fuel sweepout from the reactor cavity appear to be Based on failure of the reactor vessel at 2500 psi, as in a TMLB accident. Nevertheless, the sweepout models developed are applied to accidents where the failure occurs as low as a few hundred pounds per square inch. We are told that experi-ments on the Behavior of molten fuel at pressures up to 100 psi have shown no such sweepout. If there is evidence to validate your sweepout models at high pressure or its extrapolation to low pressure . it would be helpful .
~
VI. Hydrogen Burning VI-1. In order to enable the staff to place the issue of hydrogen (and carbon monoxide) burning into an appropriate perspective, provide us with the probabilities you assigned to those classes and bounding cases. that result in vigorous hydrogen burns as listed in the Tables 4.3.1, 2, 3, 4, 5 and 6.
- 3. ' ll .'
- i. -
- ~4 IVI-2. Discuss the flame temperature criterion. In particular explain how the "H2 Burn Pro 5a51-11 ties" and the times above and.5elow the FTC were determin~ed. . Explain your position .that this criterion is conservative since it appears to preclude the accumulation of mixtures that permit detonation phenomena.
VI-3. Discuss how, if at all, you included in the ZPSS the potential for large burns resulting from restoration of containment cooling at a point when large quantiti'es of hydrogen are present in the contain-ment. VI-4 .You use a single-volume containment nodel ar.d assume a homogeneous mix of hydrogen. !!e are concerned that irsufficient attention has
- Seen given in t'.e r analysis to non-unifor . affects including non uniform distributions of hydrogen in the containment which could lead -to the possibility of damaging detcr.ations in local regions .
[ damaging to ESF and to the containment itself). Please elaborate on these matters. Y e e mne e e *
. L. .J' ,': ,' .
j . .. - b VIII. Core /i'ater/ Concrete Interactions
~V III-l Quenching of the debris depends on the ability to supply coolant
-and to extract heat. Most of the analyses in the PRA consider only the limitation of providing coolant (see for example pp. 3.2-13 and 3.2-14). Explain how potential formation of a crust over the debris and the resulting reduction in the ability to extract heat will alter any conclusion on quench rate, steam-inerting of contain-ment, and hydrogen generation.
Further, explain how a potential crust over the debris will alter the arguments for core debris expulsion from the cavity - in particular entrainment and flow up out of the :5vity. on - VIII-2. Analyses of basemat penetration are based penetrating the entire 9 foot ba sema t . Explain how penetration o/ f the two foot thick ' basemat below the reactor cavity sump will be prevented and how earlier _ penetration of this section will alter risk. , . . .
'VIII-3 Gas evolution due to core debris concrete interactions is evaluated only during discharge and some long-term interactions. Wh'at gas evolution would be associated with other steps of the. dynamic dispersal described in the PRA? Examples of these steps are flow of debris up the key-way walls and restriction of concrete attack to twice the area of the jet. Consider also the experimentally
. demonstrated fact that flowing materials erode concrete more rapidly than a nonflowing pool .
VIII-4. Explain why debris bed stratification by particle size and its well-known inhibition of coolability was not considered in assessing the coolability of fragmented core debris. Also, coolability is quite dependent on the packing efficiencies of the particulate. A 60% efficiency is assumed throughout the analysis. It is known- the distributed particle-size masses can pack more efficiently than this. Explain how the conclusion of permanent coolability of fragmented core debris might be changed if packing efficiency were increased to 75%.
4
.c .-l, ,
IX Vessel Failure IX-1 !! hat are the ramifications of multiple instrument. tube failures (say 12 'to 24} on the accident progression? Consider the steam explosion potential when large ar.ounts of water are present in the cavity. IX-2 Be prepared to discuss the ZPSS analysis of the vessel failure mode considering such . items as the uniform temperature assumption, the role of crusts and plugging, r.cr. uniform wall fluxes, steel ablation, and the composition of the corium attacking the vessel . IX-3 Phat probability do you associate with the scenario where the core tarrel fails and the molten p:c', and core support plate fall into the plenum? What are the vessel. faGure implications?
- XI Containment Failure Analysis ..
XI-l In your analysis the section of the cylinder at the junction with - the foundation mat has not Seen considered on the ground that the effects of loading' at this discontinuity is sel f-limiting. Explain what you mean by self-limiting.
! XI-2 If self-limiting involves cracking of concrete, yielding of reinforcing steel and steel liner, how can you know that the integrity, structural as well as leaktightness, 'of the containment is insured without making any analysis of this portion of the containment.
XI-3 Indicate whether the steel liner strain values obtained in experiments are based on' uniaxial or biaxial tension tests. If based on uniaxial tensile tests, how can such test data be applied to the steel liner which is under biaxial tension? f e
.i,,': ATTACHMENT 2 Discussion of Issues Raised in NRC Review of the:
" Zion Probabilistic Safety Study (ZPSS)"
This report discusses severe-accident phenomenology, in particular core-melt behavior and containment responses, and raises related questions applicable to the ZPSS. The report is intended to reflect the discussions which took place at the review meeting of the NRC Staff and Coninonwealth Edison which was held in Bethesda, Md. on January 28 and 29, 1982. It is hoped that this report will be of use to ACRS activities involving Zion, Class-Nine accidents, hydrogen, and probabilistic risk assessment. I. How Cores Melt As you can tell from the meeting agenda (Attachment 1) the discussion was intended to help the NRC reviewers of the Zion PRA understand how that study analyzed core-melt phenomena and related events of severe-accident scenarios ~ related to containment challenge. The meeting was successful in obtaining from Commonwealth Edison (and its consultants) explanations responsive to the questions, but points of contention and doubt still remain. (The Staff's and ACRS reviews may resolve some of what remains.) At the heart of the problem of trying to understand the ZPSS, I think, is an understanding of what was offered as their best-guess core-melt scenario, which will now be described. At some point following an initiating event and failure of sufficient core cooling, the core begins to o'verheat, first in the center. Steam is available and cladding oxidation occurs earlier than fuel melting. In the rod-cluster control assemblies the silver-indium-cadmium alloy, the neutron - absorber material, melts very early and flows out of the stainless-steel guide tubes which fail due to overheating. The silver would probably run down toward the core bottom and solidify (being cooled by water below the core), plugging the fuel-assembly bottom nozzles or core-support plate flow holes. Later, the fuel in the central portion of the core melts and, along with core structural debris, falls or drips or flows down to the solid " silver plug" on the bottom support plate. Eventually the core debris penetrates the support plate and falls i
into the lower vessel head which probably contains water. For core melts
~
with an intact primary coolant system, the ZPSS does not assume that primary system integrity is breached at that point, but instead, it assumes that the melt vaporizes all the water from the lower head and then fails one or more instrumentation penetration holes. The molten debris then spurts downward under considerable pressure through the hole (s) which ablates in the process to a diameter cf about 0.3m. In just a few seconds the molten fuel, liquified debris and any entrained solids splatter off the concrete floor of the reactor cavity several meters below the vessel. With or without there being water in the cavity, a lot of the debris from the vessel is assumed to be energetically " swept out" of the cavity through the instrument tunnel. -- The core debris, now dispersed far and wide -- in the vessel, beneath the vessel, in the tonnel and on the nearby containment floor -- is expected usually to be easily coolable so that the containment basemat would not be penetrated, nor would additional significant quantities of incondensable flammable gases be produced by debris-concrete interactions. During the 4 above core-melt proceedings it is expected that normally about half the fuel would melt by the time of vessel failure. It is predicted that about a third (20 to 40%) of the fuel. cladding would be oxidized prior to the time that the debris melts through the lower core plate. Another 20 or 50% of the remaining unoxidized core metal is assumed to undergo oxidation before the lower vessel head fails. Smaller amounts of hydrogen (and/or carbon dioxide) than those evolved from core metal oxidation are estimated to be produced in core-concrete interactions for the typical core-melt accident scenario. Provisions for alternative core-melt phenomenology are made by assigning lesser probabilities to event-tree branches (" split fractions") than those assigned to the sequence described above. Moreover, the ZPSS authors at the review meeting reassured questioners of their core-melt phenomenology that the specific physical assumptions make little difference anyway in affecting containment-failure probability. Sensitivity studies were employed to reach that conclusion.
- x ~
, 7 .
- + : -
6 (I have not 'yet reviewed the ZPSS sensitivity studies arid cannot say if I agree or disagree' with the conclusion.) II. How Containment Fails A. Steam Explosion: The NRC Staff and the ZPSS authors are in agree-
.~ m ent that containment failure directly due to missiles generated by in-vessel steam explosions is a relatively low-probability event given a core melt. In addition, the Zioa people have studied steam explosion phenomena and concluded that steam explosions would not produce other effects, besides missiles, that could indirectly affect containment reliability. I do not
. . feel confident that anyone knows enough about the details of "how cores melt" and about steam explosions to be able to agree with the ZPSS, particularly
.regarding the indirect effects of steam explosions.
B. Failure of Containment Isolation: A*: one point in the meeting the ZPSS spokespersons stated that the probability of experiencing excessive containment leakage (i.e., greater than design basis which is about 0.1%
- per day) was " epsilon", a very small numter (10-4 per demand?). It would seem that that question needs further study. Perhaps, there is industry data in this area.
C. Hydrogen and Containment Overpressure: The NRC is carefully re-viewing the analysis of hydrogen behavior during severe accidents in the ZPSS. Like the melting of cores the generating of hydrogen, by steam metal oxidation reactions, is difficult to model because the problem is so analytically complicated and there is little data or experience on severe , , accident phenomena. Many questions regarding hydrogen were raised at the ZPSS review meeting. The authors of the study admitted that the ZPSS report itself could be faulted, in a way, for not being clearer on its treatment of hydrogen. This is so because the report contains a lot of words on hydrogen which are somewhat extraneous. This is a result of their attempt 1
., ,i.
i l a , to be.metjculous and detailed at the outset of the study, because the ( dominant factors governing hydrogen generation and behavior were not known at the time. In the end, however, sensitivity studies were co'mpleted and it was found that, indeed, many detailed concerns are not important factors. It was found that the key elements in determining hydrogen production for Zion are:
- 1) the accident sequence itself (i.e., which systems fail and in what order?).
- 2) lower vessel plenum phenomena,
- 3) core-debris / concrete reactions -- maintaining debris, coolability, l 4) containment heat removal systems reliability.
In addition, it was noted at the review meeting that the typical Zion core-melt accident would occur with such speed and incoherency that there would l rarely be anything approaching 80 or 100% cladding-steam reaction prior to vessel failure. The authors of the ZPSS seemed to feel that some accident scenarios, to less ~ than core melting such as that at TMI-2, may be better hydrogen generators than most core melts. This point is related to item l number 1.) above, in that accident sequences with repeated core wetting and dryout cycles favor hydrogen production via metal-steam reaction, but may ultimately result in core coolability. The IPSS emphasized core-melt acci-dents under the assumption that those were the " greatest contributors to risk". D. Safety Valve Failures: According to the staff reviewers of the ZPSS, that study may have underestimated the probability of a safety valve sticking open in'some " transient" sequences (i.e., non-LOCA initiated events). Presumably, this issue will be receiving careful attention during the staff's review. The ACRS review should watch for developments there. i
l I lw
~
g e, 3
.j. a .
9
. W 5 l $
k i: n 3 > d 4
'.. k
; - I f u
s '; ~
<t, .
.=
1 d. k' [..'"
..]A m,.
g *.2*i
.s r :g y s
' . A;.
j $ . I e f : c. . . . i,d r&2 j m
-(.
, :.x f -
?Cs, ,
f . .
> v. l
[ . %. . .. + , s .,w t e ..- d =. u L w e a.. OC s - h.
,,,.'.'J*
s U U
- _" ,gy ' '
( .r, ,,. qg j { , I
- s .U W = Il t o ;<: sk= d.E q .
x s
- - k Jg* t e r4 j . . _. ,1.- . f. ~**-{ Q - b.* ,,
a a
+ ., 3. 2 y_r_,: g i
l C t
/. m , t. q.-
)
t
-Q t c 3, n*,.
~ q e : J -- y i .,
g -- q _ ,[?,1
-ep
- wI L.:
.e 8 1 w- sg 4
g g , .*
.~
a C% % E -
-y :
l 4. ! e s .mgn o I W)Tg nqh I-I; 1 l -:n .<v.-. {-'; $ L V'$ l 3n.=*i
..)l p i
I i
-' ' . . .k.3 a ',
1 4; r, y e. t - . g 19[
^N 4 r -n... ;
1
.m
- -% - u 1l!
4 t.. r+;[- ., sa !. t '* ,) g
,{"!!g '; .
s '; , 7* ,, -, " , *
- tw f
p
);J s .- ;i
. ;!4
-e el ,a d -[,j
- A*! , M -('
M{ C . '.- 1.a W.79 i w. x , ~ m i,
.m
,' ::. ~ .; ~.-.g, r-r ,
;_ 4 :
d Y,j ..
,1s (( .
i, ,-
.m .m- - 2 ' Y ,.v, W .M.,
h & N5': ',
?' .-
r .c ,, w: . e
r
. -) ,
NRR TECHNICAL ASSISTANCE PROGPJJ4
~
PPAS ACTION SU!F.ARY SHEET DATE RECEIVED BY PPAS: /Il/9 FIN (S): ! .' TYPE OF ACTION: - RFP PPAS REMARKS, COMMENTS, IDENTIFICATION OF DEFICIENCIES: 1U bN - Cw W. O tnmo'.hLuc\%'badLkO. Q m&q g&y k&L b %fc ~ ~ S c1$eNy \,'" . ,u (( gb q , C0CUMENT(S) MISSING FROM FILE
'g ,j q 2 V) 1 1
CITE ACTION COMPLETED BY PPAS: \l 7~L OL. ,, ye I j
% a , .-.-.--rm----.--,,--,y , . . . ~ - - , , , , - , , , , ,,}}