ML20214W793
| ML20214W793 | |
| Person / Time | |
|---|---|
| Issue date: | 09/15/1983 |
| From: | Ross D NRC - SARP REVIEW GROUP |
| To: | Bassett O, Long J, Wright R NRC |
| Shared Package | |
| ML20213E209 | List:
|
| References | |
| FOIA-87-113, FOIA-87-60 NUDOCS 8706160188 | |
| Download: ML20214W793 (96) | |
Text
,.
=. w
~ ~ ~
,...-.~
. g*" "Dg#
UNITED STATES f.-
1 NUCLEAR REGULATORY COMMISSION WASHINGTON, D. C. 20555 q
s,
- /
SEP 151983 n
MEMORANDUM FOR: Those'on the Attached List FROM:
Denwood Ross,' Chainnan SARP Senior Review Group
SUBJECT:
SEVERE ACCIDENT ISSUES PAPERS
.In response to a Conunission request the SARP Senior Review Group is
. preparing a Commission paper on Severe Accident Decisions for Existing Nuclear Power Plants. A draft of this paper was presented to the ACRS on September 1 and comented on by the' ACRS. A revision of the paper, taking into account the ACRS views, will be submitted to the Commission in October -
1983.
Among other things the draft paper identified a number of technical issues important to Severe Accident Decisions for existing plants.
We need to prepare NRC staff position papers on each of these issues.
Drafts of these papers will serve as the basis for discussions with the ACRS and IDCOR.
Subsequently, final position papers will be presented -
to the Comission in early 1984. is an organized list of technical issues.
For each principal issue area (e.g.,1.1,1.2, etc.) an NRC leader and a BCL contact have been identified.
For each issue (e.g., 1.1.1, 1.1.2, etc.) NRC staff responsible for drafting the issue paper have been identified.
For each issue the first named staff member is the principal author of the issue paper and is respon-sible for defending this paper in discussion with the ACRS and IDCOR and revising the paper as appropriate based on these discussions.
The principal author for each issue paper should prepare a short draft position paper for this assigned issue and submit to the appropriate NRC issue area leader, the appropriate BCL coordinator, and Jim Malaro, by September 30. should be used as a general model for preparing the position papers.
Be sure that the research programs aimed at resolution of issues and subissues are clearly identified in the paper.
Battelle Columbus Laboratories is independently preparing "strawman" position papers for each of the issues.
Principal authors for each issue paper can coordinate efforts with BCL by calling Rich Denning, BCL at FTS 976-7510.
l I
aup.
8706160188 870610 PDR FOIA SHOLLYS7-60 PDR l
l
_ _, _ _ _ _ _ _.. _ _ _... _ _ _ - ~. _ _.. _
2 SEP 151983 o
BCL will then prepare a coordinated set of issue papers for each issue area and submit it to the NRC issue area leader by October 7.
The issue area leaders will submit the finished product to the Senior SARP review group by October 14.
r.
i
/ I (Benwood Ross, ChainnanL.
SARP Senior Review Group Attachments:
1.
List of Technical Issues 2.
Model for Position Papers 4
e e
4 f
y
.7._,
__._m._m__3
_,r y_,__-,~...,,
p y
__~
..-,.-.m,
___~.___._,.___,_____m_,.,.
,.y SEVERE ACCIDENT ISSUES M
1.0 Severe Accident Phenomenology 1.1 Progression of core melt in the reactor coolant system (O. Bassett)
BCL: Manahan-1.1.1 Reactor coolant system thennal and hydraulic behavior (R. Wright) 1.1.2 -
Rate and magnitude of hydrogen production in the vessel
- and release from reactor coolant system (J. Long) 1.1.3 Fuel debris and in-vessel structure interaction (J. Rosenthal) 1.1.4 Fuel debris and vessel or vessel penetration interation (J. Rosenthal) 1.1.5 Likelihood and magnitude of in-vessel steam explosions (J. Rosenthal) 1.1.6 Recovery potential prior to vessel failure (R. Wright) 1.1.7 Primary system failure from overpressure (R. Wright) 1.2 Loading of the containment. (T. Speis)
BCL: Manahan, Denning
- 1. 2.1 Con and hydraulic behavior (J. Rosenthal, 4 g (J. Larkins, V. Benaroya, J. Long) gas production, ex-te and magnitude of combustible 1.2.2 1.2.3 Distribution of combustible gases (J. Larkins, C. Tinkler) 1.2.4 Cohditions leading to and-resulting from diffusion flames (J. Larkins, C. Tinkler) 1.2.5 Conditions leading to and resulting from deflagration (J. Larkins, C. Tinkler) 1.2.6 Conditions leading to.and resulting from detonation (J. Larkins, C. Tinkler) 1.2.7 Likelihood and magnitude of ex-vessel steam explosions or steam spikes (J. Rosenthal, R. Wright) 1.2.8 Debris coolability in ex-vessel locations (J. Rosenthal, R. Wright 1.2.9 Debris relocation followino vessel failure (J. Rosenthal, R. Wright 1.2.10 Fuel debris-containment shell interactions (J. Rosenthai, R. Wric
_ _ = _ _ _ _ _. _ _ _ _ _
2 1.2.11 Fuel debris-containment floor interactions (J. Rosenthal, R. Wrighti 1.2.12 Interactions between fuel debris and internal containment structures (J. Rosenthal, R. Wright) 1.2.13 Rate and magnitude of non-condensible gas production ex-vessel (J. Larkins, J. Long) 1.3 Response of the containment and other essential equipment (R.' Vollmer)
BCL: Saffell 1.3.1 Characteristics and likelihood of containment leakage from shock loadings (P. Kuo) 1.3.2 Characteristics and likelihood of containment leakage resulting from steam spikes and/or hydrogen burning (J. Costello) 1.3.3 Chiracteristics and likelihood of containment leakage resulting from slow pressurization (P. Kuo) f 1.3.4 Characteristics and likelihood of containment leakage resulting from external events (P. Kuo) 1.3.5 Characteristics and likelihood of containment leakage resulting from thermal loading (W. Fanner) 1.3.6 Characteristics and likelihood of containment leakage resulting from internal missiles (J.Costello) 1.3.7 Potential for basemat penetration (8. Burson) 1.3.8 Reliability of early containment isolation (W. Butler) 1.3.9 Equipment and instrumentation survivability (V.Noonan) 1.4 Fission product release and transport (R.Bernero)
BCL: Cunnane 1.4.1 Rate and magnitude of release of fission products from fuel (in-vessel)
(G. Marino, L. Chan) 1.4.2 Deposition of fission products during in-vessel transport (M. Jankowski) 1.4.3 Rate and magnitude of release of radionuclides from fuel (ex-vessel)
(T. Walker, M. Jankowski) 1.4.4 Deposition of fission products in containment due to natural processes (W. Pasadag) 1.4.5 Effect of engineered safety features on fission product retention (F. Akstulewicz) 1.4.6 Deposition of fission products in other olant buildings (W. Pasadag P. Easir.y) s
3 e
1.5 Ex-containment transport and consequences (R. Bernero)
BCL: Davis 1.5.1 Environmental dispersion (R. Blonct, J. Hulman, I. Sp1ckler)
/
1.5.2 Food chain transport (J. Martini,E.Branagan)-
l.5.3 Dosimetry and health effects (J. Martini E. Branagan) l.5.4 Modeling of emergency response (J. Martin, S. Acharya) 1.5.5 Cost analysis (R. Blond,M. Richter) g 2.0 Safety Assessment 2.1 Characterization of plants and sequences (R. Bernero)
BCL: Davis 2.1.1 Plant categorization (P. Baranowsky, C. Eng)/
'A 2.1.2 Identifi. cation of accident sequences (B.Agrawal) 2.1.3 Quantification of sequence likelihoed (P Bua g (C. {pg)
/
3 2.1.4 Equipment performance and success criteria (B.Agrawal) 2.1.5 Influence of operator action on accident sequence (C. Overby)
BCL: DiSalvo 2.2 Assessment of existing plants (R. Bernero)
BCL: Walters 2.2.1 Response of reference plants to selected severe accidents (B. Agrawal) 2.2.2 Qualitative assessment,of severe accident likelihood (P. Baranowsky, C. Eng-)
2.2.3 Integrated probabilistic risk assessments for reference plants (P. Baranowsky)%
2.2.4 Credibility of PRA techniques (J.Murphyf 2.2.5 Applicability of conclusions concerning existing plants (F. Rowsome, S. Acharya) g 2.2.6 Effects of uncertainties and sensitivitioMthe h
l estimated severe accident consequences ((M Cunninghag 2.2.7 Effects of uncertainties and sensitivities on estimates of severe accident likelihoods (G. Burdick, C. Eng) )
2.2.8 Sabotage (J. Wermiel) e
4 e.
2.3 Assessment of plants with modifications (R. Bernero)
BCL: Walters 2.3.1 Design and operatio.ochang u trp% accidents including sabotage protection (M. Cunningha p 2.3.2 Improvement in severe accident management capability (C. Overby)
(k'p _.__ ___,
\\_
Cunningha## Rosenthal) 2.3.3 Design changes to mitigate accidents 2.3.4 Changes in emergency response capability (L. Soffer) 3.0 Decision Methodology (R. Mattson) 3.1.1 Sele,ction of decision techniques (J. Malaio)
)
3.1.2 Selection of attributes to be used in cost-benefit analysis (J. Malaro) 3.l.3 Selection of weights for cost-benefit analysis (J.Malak) 3.1.4 Role of safety goals (14 Ernst)~ \\
~
3.1.5 Balance of decision acoroaches (J. Malaro) "
^
3.1. 6 Balance of preventive and mitigative measures (J.-Malaro) t4 wh 3 6
[1 2 1.,
JLs av
SEP 151983 List of Addressess for Memorandum Dated:
- 0. 3assett BCL R. Wright C 0enning J. Long M. Manahan J. Rosenthal B. Saffell J. Cunnane T. S
.s M
T. Davis rkins R. DiSalvo V. Benaroya A. Walters C. Tinkler R. Vollmer P. Kuo J. Costello W. Farmer B. Burson W. Butler V. Noonan CC yJ C
'v R. Bernero s s G. Marino L. Chan M. Jankowski T. Walker
~ ~ ~ ~ ~
W. Pasadao F. Akstulewicz P. Easley R. Blond J. Hulman I. Spickler J. Martin.
E. Branagan S. Acharya B. Richter P. Baranowsky C. Eng B. Agrawal C. Overby J. Murphy F. Rowsome G. Burdick J. Wermiel L. Soffer E. Jordan G. Arlotto R. Mattson D. Muller Z. Rosztaczy n/e
3.-
u.a...../. /c;.,;r..y;:sg.Q. :; x g,,z. %.
,; ~. ;
- p,, n.
. n._u.-:. -. c a
_..u
...., u.a:.=
n.-.
. _. r..
.u.;
_ -)t
-i -
w
+
'.:'.}
1-3-84
- ,1 a.
e-4
..re t J C,}
This package contains a 2-page commentary on issue paper format, plus ea
.(.!
some issue papers with marginal notations by DFR.
-t h 'n ep
' ft,'
- '.i r, <
',*f.,
). 'e
.s( s
'.,6
- J Y,?%)
,I:)'4
.,4 1:d f
~ ;).
.[.
s 1
- '1 1
2
- . I
/hk n.. ;a
'..~,j F.i
., y e
% [
(*
'l
+.
'1 y
F.
X '.l
,' I
- [)
.h
[*ji 4
E '. ?
e".
.%.p O
.)
.w
- d
.4
.v -
9
(
l t
i 8 46
.- j
M N.%,%.y.i&c;W-9 QGT6. 7ffiK.Q. % iM-G X A. O2,:.A n %.. e:R'JisD. O CT-g)-lsgm e
~
- m. - - x JY
+
DFROSS: bas FM
~12/30/83 e
w
- e.,
o R,y Is,(
ISSUE PAPERS 4
Basic Assumption: describe technology as of Mar-@rt.84; focus on E[..
what can be done as of then
~
j M24 R
1.
Description of the Issue
- i{.J XQ generally ok ri;d.
..a
.l
.m
- @3 2.
Implications to Regulatory Questions y
. do not try to force a nexus to the Regulatory Quest, ions; p_
do natura11y, TTTiidicated k,d O',
I ',yj 3.
Subissues
}2 g?
genera 11y ok
\\
f
[,p h
t,'t.
Approach toikear "erm Resolution
^
>,..<.:y
+G:
v 'I -
v 4}Gi
. assume a tiine-line coincident with ASTP0 needs /far-Jai 8,4]
,.i 3
s
!?
. continuing resolution in paragraph 7
.\\-
NU
- f. I ' ('
L3J key to NRC efforts (not IDCOR) s.
$pq
.N.,
^-
4 t.
.')
/
9,
'y fy.'5.
Status of the nonding.
s y
q,
. What is known, not so known gi
'x
. _ - -- s W
. IDCOR plus & minus goes here (very specific)
.. ca
,.%?
':f;
. perceived conservatisms, uncertainties --
1g,
.c'.'
'x s
\\
..y
.M f f. NRC Position
.@2
/
. limits on code input & models that must be used on the ASTP0 v
.k,)J time-line
.y. needed additional calculations (based on status of understanding) e:y 2d
,.O'V. when we have completed review of IDCOR or the issue, and we are Pfj ready to take a position (accept, reject, or in-between) put here.
i
\\
a w
.~ l m
QQSV:Efi$Q@fi%&Wf.[h];;i? 'W'-g &f'dy, i?p-?;;Q&.l *. l '
..i '
?
cd
. */f]
4.,
H. ;t; q@E.
n d
-1 Continuing Confimatory Work
~
t ll NRC only - post ASTP0
- g
.vd time line -- reduce il m.JI uncertainties -- best estimate
'?
.i ;.d improvements i:>-
9:a
.cp
.yt specify needed hsearch and proposed end date, 44
.and is it needed N,*
y.
.t.-QM r
- .h.,
3, 3
4 m *
',: )
J
.'T4 S,4 (s. -
...J
( -l 7 F.1 o,t
,.w.;
6,'1
%?*T:I I'h M-
.rw :
f:.*'.4
> "' s-g
,...i
- t.*
E.
I.4
!.<. 5 i,* 4
. C'y+
..*0 f a.:
- f. a N
>??g r.3 Jd
.z
,I
.4 I
b
,1 l
(532
%f %:fM Mji.jQ Q.M Q g p g.gg..
y,.4 4.
~s.u.. w Revised draft b;;u 12/23/C3 9]9
-(
%Y.
ISSUE 1.1.1 m
REACTOR COOLANT SYSTEM THEMAL AND HYDRAILIC BEHAVIOR N
~
@M-Mf Q.
1.
Description of the Issue I.>)
.The issue deals with the impact of the reactor coolant system (RCS) thermal-hydraulic behavior upon the important safety issues including i
J the release of radioactive fission products and hydrogen to the containment
%.h during a severe accident.*
m The thennal-hydraulic behavior of the RCS detennines the temperature g!
rise in the core, which strongly affects the release rate of fission It also detennines the availability of steam / water at the Q
products.
fuel cladding surface as well as the cladding temperature; as a result, se the hydrogen generation rate in a degraded core depends strongly on the h
I RCS thermal-hydraulic behavior. The fission products are transported frcar the core exit through a part of the RCS to the containment by L
/
$ed.2 i (-
following the pathway that is controlled by the RCS thennal-hydraulic 4:;j behavior. Along the pathway, some of the fission products are removed w
due to deposition or chemical reaction with the vessel-internal structures, iN piping walls, and aerosol particles. Such depositions are strongly k
affected by the wall temperatures and flow rates and are, therefore, In conclusion, md critically dependent on the thermal-hydraulic behavior.
3) the RCS thennal-hydraulic behavior plays a major role in detennining the f
release rates of radioactive fission products and hydrogen in the containment d
L q"i during a severe accident by means of its control of core, cladaing, and snm dit RCS temperatures.
Y'i VG The Implications of the Issue to Regulatory Questions h
2.
gh h
Major implications of the issue are the thermal-hydraulics induced h
uncertainties under various accident sequences of (1) the release rate
)
of radioactive fission products and aerosols to the containment, (2) the O
release rate of hydrogen to the containment, and (3) the resulting
[
containment load and equipment survivability. Th= issuch th 3
+h. fattowina reaut a+a-y petMas:
']
+a
.3
- ' NkdbbbG5NNbN/$NbhAM.d[6~;U dji / >Q]3[.j{.j j f
n
. f. ',
- A Tp%q 4 T
//
I
/
3.
How safe are the existing. plants with respect to severe accidents?
\\
${i 3.3 ~ )Alich accidents are to be considered and which ones can be Ne
((f 4
ruled cut?
~
5.
neiat additional research or infonnation is needed?
p 5.1 neiat are-the information gaps bearing on the severe accident a
- W decision?
~
di 4
5.4 Are there specific issues that require more data before they are decided?
h 3.
Subissues di Sq Fuel liquefaction and relocation before and up to fuel melting.
x Effects of multi-dimensional flow patterns in the vessel upon fission 3
product retention and hydrogen generation.
O Core heat generation during fission product release and fuel relocation.
p'
.yy Radiation heat transfer and core debris temperature.
o 3
Surface heating effect on fission product deposition.
.J
%drogen generation affected by blockage fonnation, oxidizable surface
@$p:
area, and hydrogen blanketing effect.
(.
Goolability of degraded core prior to slumping of core below the 6
ifj; lower core support plate.
gp M
Effects of RCS water inventory upon fission product retention.
,i b.
ger-T e
- @1 6P'
[f (9g 6 4.
Approach toAResolution Fit The approach to resolutions is primarily based upon:
h.
M.i 1.
The extension of the existing best-estimate system thennal-M hydraulic codes, RELAPS and TRAC, to include modeling of (a) fission product snd aerosol transport and deposition using existing models from the TRAP / MELT.
P Tj code, and (b) flow and temperature calculations for the primary system gj}
components.under degraded conditions.
,h 5
2.
The development of detailed core behavior codes under degraded (A -
h conditions to replace the core component in RELAPS and TRAC. Two codes under (
h development include SCDAP for calculation of pin behavior up to a total y
q loss of pin-like geometry, and.MELPROG for melt-progression through vessel
[,{
'1 failure after loss of core geometry.
These codes will supply the models
'3 necessary for accurate computation of (a) fission product and aerosol release 7
from the core, (b) Zircaloy oxidation and hydrogen generation, (c) fuel L
disintegration and relocation, and (d) damaged core coolability.
(
s
y $ $ M f9 0 M M Mj{M M & 1 @ M M *py P a M
. O $; 3 d j., / i g
i,.:
@}.
?n fj- (k
.Up 4 D
V 3.
The integration of the above codes via a system of coupled a '---
ng feedback (if necessary) and. loose linking dere possible.
~
i 6
=
ji.p In addition, the MIMAS code d ch as for TMI analysis
'f with simpler modeling than SCDAP an Tiso be usedpD PWR degraded core A$
analysis. The linking and integration of scDAP/MIMAS (modelling degraded 3
core) MELPROG (modelling lower plenum debris interact 11on with structure M
M
%j and water), the extended RELAPS/ TRAC (modelling upper plengt-d = inni jf region of the RCS), and. TRAP-MELT [be madey that a severe accident can
$)
be simulated for the entire RCS to realistically predict the release rates ki, of radioactive fission products and hydrogen in the containment. The k
existing RELAPS and TRAC codes w[be u a provide the initi rq.j conditions prior to the onset of fuel cladding deformation dich generally occurs at around 1000K.
k h(
5.
Status of Understandino tj.j The MARCH code is currently used to model the thennal-hydraulic h
behavior in the reactor core and lower plenum region, and the MERGE code is used to model the rest of the reactor coolant sy' stem including the upper M,
plane of the vessel and the cold and hot legs. A naber of assumptions wy are used in MARCH / MERGE to greatly simplify modelling requirements; as a ki result, the code prediction may become either too conservative or unrealistic h
for some accident sequences. These assumptions include:
(1) one-dimensional g'j
[,
hydrodynamics in the reactor vessel, (2) the shortest pathway for fission
[p:j product transport from the core through the break into the containment.
]),
}
(3) steam and hydrogen only are present in MERGE (i.e., no fission products Q't
\\
W or aerosols), (4) the decay heat generated by fission product deposition on structure surface is negl acted in MERGE, (5) the mass flow rate leaving the system through the break cannot be mechanistically modelled and it is W
either input by the user or calculated by assuming that the flow rate is that required to equalize the reactor coolant system and containment pressures, d,$N(
and (6) the entire core will slump at the time 75 percent (or user input) r.
of the core are molten.
}
L Dccei 3
- 7 3.M af grgf
- y.. A. ;'s%.ry.eg sq 3,7_..;
1, f.
x..e
<..lv_ c _. u m. r
-m a
iu,
-e
,,3, y g..
. ;. g
.j 7 -
W}
5 '. ' %
J.'
@k
-4 k, s
&j Three-dimensional TRAC calculations have been performed for a PWR
- j
[yjij vessel assuming an intact core without.any fuel rod damage. The result Id will be used to assess the first assumption listed above by March 1984.
b.M S
- ^ jd
.9 3
p
,q
..q 6.
NRC Position
\\
Lc
^
Wp t
N/L The SCDAP code repr ts the current state-of-the-art understanding "M
on degraded core be or in tenns of core heating up, hydrogen generation, fit and ft release. Comparisons between SCDAP and the IDCOR's HEATUP W
code ldbemadjeo assess the uncertainties calculated by HEATUP on core tempera r= i.iii.ery and hydrogen generation rate. N9te that he fission product 7,;;
release from a degraded core is ([ot ma g in HEATUP.
l
,s
[.]
Signficant uncertainties are believed.to exist in the MARCH / MERGE
.'e :
calculations of the RCS thennal-hydraulic behavior and therefore in the 1,
rM predicted core temperature and release rates of radioactive fission products il and h drogen to the containment. These uncertainties will be examined in
- i
(
981-80'4 a ". and-i or BWR by comparing the MARCH / MERGE results with the best-estimate mechanistic codes being developed at various nationai N
I laboratories ( REl.AP5/ TRAC, SCDAP/HDiAS, MEl. PROG, and TRAP-MELT).
M' \\'
(
iffj However, this does not necessarily meanbhQnofNj can be made
.IY; until those uncertainties are determined. For inst /nce, if the MARCH / MERGE results supplemented by can'sesifive upper-bound hanf3;alculationsindicate_]g, that thQty' limit,of a plant will not be exce ded during an accident, g /u a decision an probably be made to exempt that p ant from any corrective actions fo that accident sequence.
the othe hand, if the MARCH / MERGE ij results i dicate that the plant saf fT1? sit wil be exceeded to the contrary 3) of licen ee's calculations, be es mate mech istic codes should be used i,i.j to help esolve the issue.
@ h y?hhW
)
- l.y Yb
(
. ch') '
g3 p
/#
i.j 4 e j @p
(
pc g-c g#.
1 is e3 j
,%,..,i b,
.l%,~
N$$.NE5 Nb&{N$N?5$'5 ki$in i% M.- Q ~ t $4. _.'. R.o.,.
e m
- s.,
P-
- "
- t * *.
Revised draft
@;% N c:..
12/23/83 a., %
(
,g d
ISSUE 1.1.2 dl RATE AND NAGNITUDE OF HYDR 0 GEN PRODUCTION IN THE VESSEL AND 3j-l RELEASE FRW THE REACTW C0OLANT SYSTB r
(5f h&
W(%
v.s g;p, I
holications of the Issue to Reculatory Questions h
.. sare are une existing plants?
e.
i
- v..,
- 1. ;
Q Severe accidents involving degraded or melted cores present the j
possibility that hydrogen can be generated at rates that exceed the capacity of ordinary hydrogen recombiners and/or ignitors, h
and in quantities that could jeopardize key accident control M*
equipment, or in some cases, even the containment itself. Thus these rates and wantitities should be understood, or at j
least bounded, so that proper protective measures ca'n be Z
developed.
(Sp# ' p.
hj
)O The probability of such eventsQt' he estimatMn order to V*y, a assess the need for mitigating floatures. However, since an h) event of this sort has happened once, at 1MI the initial perception must be that the probability is high enough to be a significant concern.
- 4 A more detailed estimate of the probability of a severe mN hydrogen burn involves all the many details of event sequences
..d and plant specifics.
It is important to keep in mind that the o.f uncovering of a substantial part of the core (probably greater h
than helf) appears necessary to produce the conditions of high temperature and steaming that lead to cladding oxidation Ii and hydrogen generation. The probability of hydrogen generation is expected to correspond closely to and be bounded b'y the is q
a probability of this type of major core uncovering.
The final judgment on safety in this area, as in others, will td be made deteministically, with the probabilistic estimates h
'1 entering as one factor in the judgment.
@$dbdI$f;jb5$$$(MANESbMN'NY#SDd.MM Di'..bd1 3,.5p.....
~
E g
6
/_.x
.](.
, 23
( !s adDonal protection needed or desirable?
~
9m At the level of this discussion, the question will be
' Ei considered fbr generic classes of reactors, although h
individual reactor features may affect the applications to I
specific plants.
t 7
j h%
y' ur
.h")
/
h The inerting of Mk I and Mk II BWRs (ppears to eliminats e
j the problems of hydrogen combustion.
It is not expected j
that additional features will be needed or desirable (hydrogen as a non-condensible gas will be covered under issure 1.2.9).
g'M Mk III BWRs and Ice Condenser PWRs have en ptrti=1 %teg,- rn Y?
rd1= 'Rt in =i*i =td by distributed ignition systems.
p: Vessel problem is addressed by close.~attenti j'
.n.
in V
};; q
~
n=
Np g(
vessel water level information, multiplicity of water soun:es ?
y g-star training, depressurization ability, etc.
[
M,1 The(fES$kR version of Mk IITpDposes complete reliance on I,
l"&cv operator training and,%1 featugte eliminate the need 7
for a distributed ignition syst. These proposals are under
{Rl s
A4
- ~
k.)
Large dry PWR containments have been considered less
[;j vulnerable tg the hydrogen generated in degraded coresfshort H.J of core melt). The larger quantities of hydrogen associated with complete core melt o6bble bedM gohu;.b l
a problem if burned all at once. We containment aspects
.~
ft are addressed under issues 1.2.
-vessel prg are the same as those of ice condensers, an involve making sure that
@O the operators receive proper in-cori infonnation andD W h' fM
/
trainined in how to use it, as well s availability of
'( f V
multiple soun:es of power and weter.
4)#L 3
CC N l
v "3
f }Ok' u:2hil.:* 1.0$$0S$$5'$?' m.5.l ^'
'. N.: I-i.;;
,;w
'?
L i;
R1
..
- e,. -
rs
%,.,I
'3
.M/ @$$y l.
i Wat criteria should be used to detennine the ne r
^q l*
additional protection?
a 3y y
Protection of containment (issues 1.2)
- Tr.
I 0"
Protection of vital eqJipment 3
Y The 755 active clad reaction used as a critertog
- ,r
y]
protection in degraded core accidents s@ be adyg
/
to some figure probably between 100% and 200% for core a.: 1 6
mel t.
The higher limit would be used if various soun:es of hydrogen other than the active cladding are detennined to be important. Such sources include steel from the j-]~
j upper plenum, additional zirconium from end caps, channel
?
boxes and fittings, and baron carbide from BWR control rods.
'd
(
a lik tho'od of the accident is certainly a criterion
't :
pN to consider.
It will be addressed by a combination of iff
, yj detenninistic methods, and 1MI-2 e 3..
nun w +ha level ^# Mr existing plants be increased?
> [i ca.
()
- j
- Mk I and II.
Inerted.
further protection needed.
7
- Mk III Ice Condensers.
Distributed ignition systems Der offer protection. However, if significantly greater amounts j
of hyd t be considered in a full core melt accident, the ttinks,provided by the ice or suppression pool should
{.),
F) be reexami ed for their adequacy. Attention to the reliab-r
(
111ty of ter and power supplies, including the operator d
functions o interpreting and implementing these supplies, N r.1; are the chie areas where in-vessel protection may be 7.c.
increased.
']
[
Q f
00 W 'Y-L 1
(<c"%
-:N,?d37bN735$$$N.I$13MFNI%$? 4%ydt$-@As 4&W 0...a'f, ' %
5 w_.
g
- 1N 3:3
%q
- e; n..
5 a
]
b GESSAR and large dry PW In-vessel protection same as
- III and Ice Condensers.
Q ikN
$2 eMsradditional research or infomation is needed?
+
A Y,7 Event sequences carried out through the core-melt /
T rubble-bed stage. This may require augmentation of codes k
like MARCH and HEATUP to include a sound appreciation of _ udt
')
if how much steel, boron carbide, and(ex-core zirconfupy hv' g
contribu'te to the reaction.
In addition, the development of best-estimate hanistic codes including SCDAP and MELPROG is needed
- udpN, h.j ti The gestion of how great is the correspondence betwee hydrogen generation in the core and hydrogen release to
(.yj.
the containment from the primary system needs~ further i
.d study. This would probably be addressed through development b!I and use of the mechanistic themal hydraulic codes like
,. t RELAPS and TRAC.
g G.; _e M.
Nl 2.
Subissues 55 m
. _Q:.
mat are the effects of multi-dimensional flow patterns in gf
- q.t the vessel upon hydrogen generation?
q s
h.l Pld.
mat are the effects of blockage fomation, oxidizable e
(.E surface area, and hydrogen blanketing?
Di.
ni Does the core debris remain above the lower core support plate or enter the lower plenum?
W hd mat are the appropriate or bounding rates and magnitudes
$;W of the uter reactions with Zircaloy, steel and boron carbide?
y.j
,$,.1 4
i q
l a-during melting and slumping m,
b-during boiloff in the lower plenum
[ g}i 0n f.]
e~
r--
,____._,_____,m.__,,.
@ @ f @ ff % ec @ -;(.0- g. - @-.p.gg.$ w,,, p.
.,,,, g,
. 7-
~..,.
.g,
3 g
.u -
, W>
m e.4
- ,g,
Is there a one-to-one correspondence (or nearly so) between
~
R sc
'9
" generation" and " release"?
Y e4 V b g'{$
pM 3.
Approach to/ Resolution p
"h The MARCH code should be hed and validated as much as p f].
feasible using the existing experimental results. Realistic upper-c.Xc w
?f!
bound calculations.for various severe accident sequences should be 7h;.
performed for a BWR with Mark III containment and for a PWR with, yc ice-condenser containment.
s,a.,
r
?),,
The development of best-estimate mechanistic codes including SCDAP and f,
Q MEl. PROG should be carried out to reduce the uncertainty in the prediction
'1 I
of hydrogen generation.
{
l
..a p
/
4.
Status of of Understanding 4 (
v.,
%g,~
In the nonnal rod, configuration, the ability to predict the oxidation y
rate of zirconium cladding is believed to be ood.
BWR shroud zirconium has been treated a:Wivalent' d materi in existing, s[
. ps F
[}1,
- 1 Q
r analyses.
.sy,
. )yl n
x at.-
l:'A Oxidation after fuel liquefaction and core slumping is much more
.s:
Q.
uncertain. Model assumptions in different codes treat this effect
.11 differently. Slumping of molten zirconium can decrease the potentia l;4 g for oxidation by allowing it to nJn off to a cooler area or can M
th lif increase the potential by exposing unoxidized surface to steam.
q
{.)
Uncertainties in this aspect of cladding oxidation can lead to a Sff) range of uncertainty of approximat y2 6 percent zirconium
- 9 oxidation during the time in which cor is above the lower core rid support plate.
]ac.;
h J j,edl
(
+
/s9 s
)
'k (Ly#\\t M
M f,3. 8 If:N'!M]5%1M*,7IQ@$p [.]; ~ 'G
> ]i-l. lg
,e s
7 f,j
[ y%s '
q (Ch3
' jj potential for reaction o irconium and steel in the lower d'g '
}?;
lenum of the vessel is' quite uncertain ano depIes on the configuration core debris.
If the core debris fragments upon contact with y,
t.j
( r in th peplenum, a large surface area of reaction could g.,
e3 resuis ror oxidation to occur prior to quenching of the particles.
i$0 Fomation of a debris bed could also enhance oxidation.
Recent IDCOR
[hk analyses, which are under review, assign'a mino to those effects.
Qarrent analyses indicate t ctures above the core q
could reach tamperatures e melti t dich oxidation could be
... d' significant The thenaal-h ng en the core region
(;;y and upper plenum of a PWR is te uncer If large recirculation Nj patterns are possible the poten al for ox tion and hydrogen production could be very Targe.MMo p h.
TRAC calculations have indicated that the rate 6f hyd en release
[
corresponds to the rate of eneration for small break LOCAs that are 2
recovered before sel mel r
h.
Station blackout or vessel N
meltthrough sitta need to be studied to establish this corre-h
/T@M b*
*"'*"c
whM
- )
5.
IstC Position
- d N,-
For PWRs with extreme accident conditions, the NRC staff has considered r
,.Ei; the adoption of 3000-lb hydrogen production as a compromise between,,
hj 100% active cladding ( = 2000-1b hydrogen) and the maximum been estimated for core, upper structures and control s(=6 0-1 ).
$U k'
In somegaa* *aang fbr the Mark III BWR and Ice Co enser PWR,
('M theQapf hydrogen repase to the containment is important in the i
estimation b equipment survival. The models for hydr en generation s(N and release rust be able to predict the rate of hyd n release as i6
(.
/
h g gwM l' d^
t Jm V
,.v
- a..> w&p:.::!s,.ya x, y h.w.:.nwg.pc.,:s! i;w.J. ~.<, ;~;. s. < :.......
x.,
,7. ~.. s,;
- n Wh~.7:;y.,,.__..._
.::y.h :. c.-:1.. ; e.f.
.A>
a.w. y.A. r.%
n w
y~
g.;c,
x;
' c.. ;
- r..
']
se:
Wi
}'pf w.1 w,
7-r,..
9z.t
~-
}"r:+TM',z well as the total release. These predictions must be validated by j
comparisons with experimental data including PBF results.
s k.:$.
i cI
- The IDCOR position on this issue is substantially different and edJ
,y This leading to lower hydrogen production in this time period.
[d..
results in a reduced threat to the containment during this time
'l period in each of the plants examined.
,,:.,. d Ws?
35<
hld.
' {
m[b N
- -l1
'd
' ? 7.5
- f c?,*
%ee
- f
- q
,.a f{.:..n Id!
'?
i.e. }
ElU
- c. -;
'.is,,
.s1 h.5
? ;j v.y I' tl
.b.
,.i t.?.
3 (5.c y'* b
<J
- fu r.s J,* 4 b,$
.,yu.u s
?t ai!
J
.b.'
f
.v :]
'y4 4
}
'N
- Q 9 % h W W % :.6 f Q Q M M y l g n y y g g..g g
,4,7,...
- 7.,
w
%y Revissd draft
's 12/23/83 u1 b
1 U
~
ISSUE 1.2.6
~
h DEBRIS COOLABILITY IN EX-VESSEL LOCATIONS bil g (3 0.9 j g q..
~ 'N -
gp l
1.
Issue Statement Given the severe accident in which a degraded core has failed the reactor 4
j vessel and is d k tad in the reactor cavity,'suc Ge3ul' man' age g of the
,,v ij accident depends upon the ability or containment configuration and plant O!j response, including possible operator ac_tions, to remove the decay heat from y
the degraded core and establish a lo g term stable tituati on which will not v
M],)
challenge the containment.
~ LW (v h
2.
Implication of this Issue to Regulatory Questions
'n.
1,}
After a degraded reactor core has failed the vessel and the core debris is w.
deposited in the reactor cavity, the debris may be cooled by the p
%j of water or by inanersion in water th'at proceeded the melt to the reactor A
]
cavity. The interaction with water may fragment the debris into relatively R'1 small particles. Assuming that an adequate water supply is available for M'
continued cooling, there are two possible severe accident conditions:
y..
- .G 3f.
(1) the debris bed cannot be cooled by water in a stable manner and the threat f(p]1 to the containment integrity is significantly increased, or M
Q ct).h&
n.
khl.
(2) the debris bed is coolable but the steam generation and gas evolution is sufficienttooverpressurizethecontainmentorcreateanenvironmentin e
{y]
jf Mws i
y,j the containment that t'aiTs the containment.
Mt
- d Te If the debris bed is c9olable the core materials cre maintained at relatively kk low temperatures and t 1e ints:raction of the core materials with the basemat is f'
minimized as a potenti 11 failure mode (Issue 1.3.7).
In addition, low debris gud q
J H,.~l
l p M M M M A W ii E W L M yj M I W R,3 G E-W.3 ;i g M M jpp A C M g, 3 M y p[;
=
s 2
JC l},]
temperatures prevent further oxidation of metals and hence limit the generation i
' }j of combustible gases (Issue 1.2.2). Limiting the quantities of combustible
]
gases generated also decreases the likelihood of containment failure from 0
hydrogenburning(Issue 1.3.2). Finally, by limiting core / concrete -
@f interactions, the rate and magnitude of release of radionuclides from the fuel,
.in ex-vessel locations, will also be limited (Issue 1.4.3).
,eg
?$)h This issue therefore is important to several other issues (1.3.7,1.2.2,1.3.2 and 1.4.3) which impact several regulatory questions. The most important
'p
'pA impact of these issues is on Regulatory Question 3 (How safe are the existing h
plants with respect to severe accidents). In addition, if water supply to the h$"
core debris is certain and coolability is assured a potential failure mode
.;j (basemat penetration) can be eliminated. This has important implications for u.
Regulatory Question 4 (How can the level of protection for severe accidents be increased). All of the above issues will provide input to Regulatory 2.
Question 5 regarding the need for additional information or research. Finally, J{
the conditions which will allow for a coolable debris bed in ex-vessel h
locations are plant s e Hie =d gst stron 1 e end on a supply of water.
,7 This implies tha decisionsbaseddnthsissuemus be plant specific 3
(Regulatory Question 2.1).
p, 3.
Subissues M
Bed geometry and size.
$;-s Homogeneity of the bed - stratification of constituents and size and
.)
distribution of bed constituents.
s.:.)
Physical description: melt, solids, gases; fixed or moving; unifom
- 's %I ds or varying with time.
igj Interaction with concrete or other structure; dispersal forces.
P.y Chemical reactions and mechanical actions.
U}t ki radiation, cortduction, convection; direction.
Heat transfer:
g'ij Initial mode of contact between the ' core debris and water.
- M.
Crust fomation, strength and stability.
w n
.,y.
QYL2O;{iLN W W5 M )1[ W 5 W S % K M $6l O O' W :i^ 5 ?:M '- VW a
E.I'$
)
ty.?
g$dt]j -
7 3
f gh h
4.
Status of Understanding u.a Core debris that is deposited in a dry cavity reacts wi the wa r in the j
.y concrete and ablates the concrete with attendant gas an& aerosol production.
sy p
Limited energy is conducted downward into the concrete; the melt continues to Much of the energy is carried off by the 1
heat and reacts with the concrete.
However, if the melt enters the water an[ fragments, the gases and aerosols.
bed may be coolable provided that the particles are not too small and a supply f.
h]d of water continues to be' available.
If the particles are very fine and the bed reheats, the case is similar to the dry case except for the decontamination k
effect of the water.' This non coolable case is similar to cooling water being fl[
introduced on top of a melt in which the water may boil on top of the melt but b
[;;u remov[ teat relatively slowly. If a stable crust forms, little direct cooling g
/
g;r by the water results,and the gases and aerosols again must provide the heat B
transfer mechanism. The aerosols and gases may have fission product h
constituentsandtransformasignificantporgignoftheenergysourcetothe g
upper containment. For all cases, the hydrogei generation, fission product B
release, and debris dispersal need to be known for further accident analysis.
%{.g All of the above scenario components are based on limited experimental results d
with simulated melt constituents.
In FY 83, experimental large melt pours (200 The Core Melt
$i kg) of corium were made at temperatures exceeding 2500*C.
Technology Program will continue this work in FY 84 and provide an experimental mg base for corium pours onto concrete and establish the cooling capability of pre Y}
((f or post introduction of water. The ensuing friability of corium particles and 9
thus the coolability of the core debris will also be determined.
Instrumenta-M) tion for measuring heat flux, upward or downward, and crust strength will be r$;j included as well as aerosol and gas generation.
- ?
m
-9 4;4 Models have been developed and incorporated in codes to predict the coolability y
of debris beds. Verification of these models is based largely on steel melt j
tests.
It appears that the coolability of a postulated debris bed can be g
predicted. However, the prediction of the bed description is uncertain as
- j reflected in the discussion above.
In the event of stable or self-healing g
-Q
Q y y Q '. M y,;: f M Q L:1.._. h '[,Q
&n.L;t,) if T ' Q.f ;#-
)?f 3
J.q a
.O
(
4 c.
m y-9
$jj crusts, the formation of a coolable debris bed may require many hours.
If the
~
)
codes predict the corium test.results reasonably well, minor improvements in 3
heat transfer coefficients, energy partition, and particle size estimates may hl) be all that are necessary for the application of the code to valid reactor
]
plant calculations. The testing is generic, but for low pressure release of core melt, the plant specific effects should be readily calculable. However M
high pressure release of core melt and the subsequent dispersion of the core q-debris.may have a larger dependance on the specific plant geometry. A special s'ij series of high pressure ejection tests has been initiated and will be extended
$1 to one tenth scale to evaluate the dispersion effects of high pressure. This M
latter effect is a major consideration under Issue 1.2.7, Debris Relocatio
-)
,s Following Vessel Failure.
y b hl.4)i $NcL+
~
..-4 5.
Approach to ResoTution The resolution of this issue is b reachediyanexperimenta d
which will validate models' that will be used in system code analyses that will y
Various study full sized plants with applicable reactor and containment types.
,)
Y, experimental programs are underway to obtain numerical values for analysis.
u First, the ability of water to cool particulate debris that is internally
- .s.
h heated by fission products is being studied in the Annular Core Research 5)d Reactor (ACRR) at Sandia. These tests apply to the debris coolability insi
']
and outside of the reactor vessel. Th [ M i mod d of debris coo lity as
[
a fun of particle size, bed porosity, pressure, and decay heat ate is
)
eing ver
'we.
Jh Second, the response of the ed cc is being studied in a series of out-e of-pile tests for two temperatdre regimes in the Core Melt Technology Program:
(1) molten core material inte acting with concrete and (2) solid debris that is hot enough to melt and ablate concrete. Transient tests of molten urania on h
concrete will start in the f rst quarter of FY 84. Thermite melts will be N
inductively sustained as mel s, interacting with concrete under a pool of water v.
1 (FY 84) as a basis for simi ar sustained tests of urania (FY 85). A separate s
l 1
W p
$ @ 2 0 W:M R T5 2 % s %'.9i1.W $ % N kblSL TY 2i& S;hi: $ 5 % : % & n d hid &;L &:= b O 2 5 2 L~2 W Si' y
- Q
- 9..'
5 gEi u.d,'
series of sustained tests utilizing steel hot enough to ablate concrete will be M
W conducted and repeated with 20 to 200 kg masses of UO2 (FY 84).
.c.
?:-
M 6.
NRC Position p
k)
EM For a debris bed of known geometry, composition and particle size, a reasonable estimate of the decay heat fraction in the bed may be made and the dry out
^
A.0 cooling can be calculated with reasonable assurance. The uncertainties in the 1
ijh early transient portion of the accident lead to less reliable predictions of M;
the debris ~ distribution.
a3
$'l j,$j The phenomena involved in cooling a fully molten debris pool with a concrete y
'N boundary are sufficiently well understood for severe accident analysis. As the h
molten pool surface cools, surface crust formation and bulk freezing are more 9-uncertain because crust strength, thermal conductivities and boundary heat h-transfer coefficients are not well known. Thus the overall quantitative y)j determination of the transients during initial melt quench as it reacts with cavity materials is uncertain-Thi uncertainty limits the precision of the i
w:?
V time estimate early containme hk
'1 Cl 7
og P.i wh '
w.,
S:
a;?:
r;j 3;
a v1 del (a$
~
53 h
Ni
")j q
t J-(:. ~. _
- ..~
.._m,
I% $$$$5 NY*AtNEWC2 5 @ MW W W h W S M &S: $ $ l $ N A? Y W:' ?" * %
$i -
m
+s,#
Revised' draft Q j 12/23/83 jy N$r
[h M
$3 ISSUE 1.2.7 A
DEBRIS RELOCATION FOLLOWING VESSEL FAILURE Md Mf, mj 1.
Issue Statement id a-S Given the severe accident in which degraded core material escapes from the 2
y$f3 reactor vessel, the potential for core debris relocation affects the cool-
{p;'i ability of the core and the ability of the plant systems and operator actions 2.?]
to mitigate the threat to containment.
y.'3
.s" OfJ 2.
Implication of this Issue to Regulatory Questions Rii
- 3 For certain core meltdown sequences the primary system may be at high pressure 4
~
p:
just before vessel failure. Consequently, as the core materials penetrate the i
vessel, they will be forcibly ejected from the primary system into the region h)
(f beneath the vessel. If the vessel failure is local, then the dispersal forces
[jf from the primary system depressurization can be very high.
In addition, the f
resulting interactions from molten core materials contacting water can cause significant dispersal of the core debris. This issue is concerned with this to U
g@'?;
potential for significant relocation of the core debris from the region imediately below the reactor vessel due to the above-mentioned dispersal p;
forces.
If the debris from the core is widely distributed; it may be coolable at a steaming rate that will not lead to e
ainment failure
^6 This issue is strongly related to Issue 1.2., which deals with the potential e
interactions of the core debris withthe co ainment shell, floor and internal N
Issue 1.2.8 in turn impacts the response of the containment and
?1 structures.
other essential equipment (Issue 1.3)
Issue 1.2.7 also impacts Issue 1.2.6, vtgid which deals with ex-vessel debris d coolability.
If the core debris is concentrated into a deep bed, th n coolability is more difficult than for a
_~
m?;3 l
da shallow bed. Significant rel ation of debris from the region beneath the ey
,,r o &a g spa '
.g
&;.gd29?5NMEf9:lMQ.MyQWQQ P A%gg.. p. ':-n;..,q;-:-;; 3. t.,,
a;;.y.,
r
~
~.-.. -
].3;
~ S ;-
W.i 4
2 h".b MJ vessel may result in shallow and hence more easily coolable, debris beds. The i
eventual debris location could have a significant affect on the survivability 73
~ -
W of the ESF equipment.
g:L G.
3.
Subissues 3d
]h Dispersal Mechanisms:
Effect of discharge pressure 9
1 Rapid reactions with water Chemical reacti*ons d
Hot liquid drill effect - particularly if the hot melt impinges on a
- c. W wall.
f{]
- h..
Water supply:
y.
(W[
Melt streaming into a water filled cavity Water released to the cavity after some or all of the melt is in the t
cavity.
itj
~
N
.:9 4.Y 4.
Status of Understanding Y[5 et
,33 In the unlikely event of a general core melt accident, PWR reactor vessels with
-:dg bottom entrance instrument tubes may fail while the system is pressurized, possibly at pressures approaching normal operating pressure. Under these f]j$
circumstances, the core materials will be subjected to extensive ex-vessel
@j-dispersal forces. The melt streams may impinge on the cavity walls and act as jh, hot " liquid drills" creating large penetration rates at localized positions or R.
the release may be energetic enough to sweep out through the keyway to the
,a M],
upper containment area.
Ri.
p If a BWR plant were to have a core melt accident, both instrument tubes and Vf control rod drive tubes offer relatively thin metallic surfaces for potential h
melt-through while the system is still pressurized.
In the case of BWR Mark I containments, corium flow across the drywell floor to the liner would melt the liner quickly and fail the primary containment. Mark II containment designs cf 11a
,r
,v--
-,e
,--3--
m.,
+.
-v.,----=-r--v---
[GM[EDNk@@2b($$Nb$8EESNIM:CUN.I[fRjh 4"k' 'i#2;' i $
$h';
- m.*.*
Mj -
9;c E,;
3 W
93 are very plant specific and vary greatly in design, resulting in large
-}
differences in calculated time to failure.
- }
For either the PWR or BWR pla.
O A
specificj;ond4tt6iis must be considered on the basis o specific c c ations while the experimental base g
j; must define mass and heat tran -ar, ch ical and thermal melt / concrete 7;
interactions and provide model tests for chanical transfomation and mass
- o y
]
motion.
pua 5.
Approach to Resoluti'on y
Considerable uncertainty exists as to the operative chemical, physical and h
thermal phenomena that exist as the melt streams from the vessel. Considering h
first the melt discharge from the pressurized vessel, an experimental and (Q
analytical effort has been established to investigate the above uncertainties
.j at pressures to 2000 psi. The experiments utilize thermitic melts with gas q
pressurization. Reactor cavities of 1/20 and 1/10 linear scales are used with hj melts of 10 and 80 kg. Dry and water effect cases will be included. Early 9
results from these tests show that a considerable aerosol source tem results f.,.f fom pressurized ejection. These aerosols are formed by the mechanical process sg of the jet steam impinging on the concrete and have essentially the composition
- ip of the melt. Thus the aerosol particles provide a maximum contamination load h.
on the containment and, since the particles are heat sources, a distributed h
hydrogen ignition source. Equipment survivability within the containment may lj be challenged. Also steam inerting of the containment may be much less effective. These tests utilize model keyways and discharges through the he W
io keyways will be evaluated.
- v m
M 24' m;
7' 6.
NRC Position
....g The plant design and accident scenario must be considered for either PWR or BWR b
plants in evaluating the nature of debris relocation following vessel failure.
k The debris configuration is determined by iterative processes which are d
m c.d (31')
kj
((f,ifyTW92ghi.'ftg,.:j{gygy,y,gqgy,jgA,cg7, ;g$;.p.. g;,.,g;c.~p. 3.m,.
.g g
,.s. _,,
M,y,
- k.\\1 1
- yl
.M:
4 determined by the conservation of momentum and energy. On going research 1?.1, investigating key parameters that govern these dispersal mechanisms such as "l?,
discharge pressure, water / melt interactions, chemical reactions and modeling of
,,3 these processes will reduce the uncertainties for these mechanisms.
~;,
- ..;i n
I h}g s
. d.a (V
lss tu y
$,9 a
W E
h 00 M
M(
- .n'
%sk I' 1.2
.e lb l+I.l D.$
r La kg..f 1
- t.
- N;*
~?>
.. '.G, a
Y.,
l,.
h V r}
(Ljs.
?. h,.
k g, t.
Mi hb h
52 4
o W]
'. j
.i
-}
".1 i
,, u. t.ag::.:G.a,s.a:x. 95%lku t uww.ua -- M, :.--hW.M5p_hiW%%[4lh: ri; ~ f ?.?r.S, ';(.
I iQTWGW
@fi[*,WFM4QW --
s.
w a.
,,, a r_
w c-q:
e, s,
4M Revised draft gig,
- 12/23/83 73
)
ik ISSUE 1.2.8
~
~
(
FUEL DEBRIS - CONTAINMENT SHELL, FLOOR AND i,g INTERNAL STRUCTURE INTERACTIONS
- u l!.2 9.q 3 -]
1.
Issue Statement
-M
'.,? -
Given the improbable severe accident in which degraded core material escapes
($
from the reactor vessel, the interactions of the core debris with the basemat, M
containment shell and containment internal structures must be understood and
$j evaluated in a quantitative sense to provide realistic analyses of potential
[p(.
containment failures.
na
.,r2 g
2.
Implication of the Issue to Regul'atory Questions t.;,n gj}
M This issue considers the interaction of core debris with the concrete floor or yJ;5]
basemat and the containment shell or internal structures. Water may not be
.r d'!i present to the extent that it is a ma,jor reactant except as the water content P"
- i A;.Fi of the concrete.
f.id fS$g Analyses and some experimental work, since the Reactor Safety Study, indicate that the generation of non condensable gases and aerosols from debris / concrete h g
interaction is of greater significance to public safety than basemat pf penetration. The loading of the containment from the debris / concrete inter-({.
action can lead to d conta D d consequently a serious early
~yj external source term, whereas b sema6 penet7aboncausesadelayedsourceterm gj to the environment. The delay ould permit mitigation of societal risks although it could produce an e> pensive recovery process.
b$
J.1
@b 1
Possible interactions of the cc re materials with internal containment structures can have a number of hdverse effects. The interactions could weaken d
3
,. p
.),
c 4
\\22 * "
il W E.
y;gP:6%.~M... Q_ igg. zh:.~fiWWTgMQR.w..:g&;,. n-b.= a.'. h.a Q _;~;
,;rt,., a,l Q~
- :a w f
w c
hL k
j E
O ca W -}'
t
- t. +
a h
structral members that support essential equipment or prov de additional
$[$.
si..
sh to the containment building.
Inpressuresuppresdion, containment
'Cl desfys (BWRs and Ice Condensers), the integrity of the su;pression pool or ice 33g h
compartment is important to reduce containment pressure loads and scrub fission
- 6.. ;
product aerosols.
Interactions between the core debris and t 11s
?&
separating the compartments in these containments d,cause a bypa s of the ad
.a1 pressure suppression function.
In. addition, intera n the core a-d$
materiais and essential equipment (such as fan coolers or spray systems) could fj[
haveveryseriouseffect;(Issue 1.3.9).
Finally, interactions of the core r,n debris with the containment floor can provide a potential containment failure M@:y) mode (Issue 1.3.7).
However, if water can be supplied to the core debris, the
['j potestfat exists to prevent these interactions and attendant failure modes j
(Issue 1.2.6).
- M N.;
This fssue impacts several other issues and they all impact several regulatory questfows. The main impact is on Regulatory Question 3 (How safe are the h.
existing plants with respect to severe accidents). However, they are also
,j important to Regulatory Questions 4 and 5 (how can protection for severe h
accfdents be increased and what additional research is required).
In addition, 6
the potential for debris relocation and interaction with the containment wall, k!
floor w internal structures is very plant specific (Regulatory Question 2.1).
.g:j.
Ei h.j.
3.
Subissues
[f v
Heat conduction direction and amount.
&y Crust formation and stability.
WJ Gas generation and effect on the melt pool.
> ci p@v Water migration in the concrete.
8,U Concrete structural stability.
g, Magnitude and form of the dispersal forces acting on the core debris.
u.
Comunication paths between the reactor cavity and the remainder of 9
-$h the containment building.
J.d
..]
m r.a 1:j j
i, -.
M'N$$325553S$dNsdI bN$NbbNN' ME 5'hh ' '
i.;
61
'.y j -
4.
Status of Understanding e
Qr The hydrogen released in the melt /concret teraction is estimated to equal g{%
{
that of the nominal half core zircolo steam reaction. The noncondensable j
gasesproduced(CO,CO and H )
ld eventually overpressurize the 2
2 containment. The timing an elease rates are not known and must be determined M
on a mechanistic basi permit plant specific analyses.
In the case of BWRs k
the steam sunara nton pool is assumed to provide large source term mitigation factor 10-5 to10-4)
Collapse of th rywell inte th]
' by interaction with s
j the melt
.ld, nowever, create a breach in the drywall
_ to the jjj confinement building bypassing the suppression pool.
- c{jl A substantial body of data on melt / concrete interactions has been assembled
((
since the Reactor Safety Study. These data have been used as the basis for the M
CORCON model of melt / concrete interaction and for the VANESA model of aerosol J,y generation. These models have been based largely on steel melt /concerte g
interaction.
Few tests have been made with prototypic melts involving g.j significant quantities of UO. Furthennore, the models do not describe the 2
El interactions once solidification of the core debris begins.
d J
7'1 Il sT /9
%j V
5.
Approach to Resolution p
%}
}
v:.<
gp An experimental program is under way that utilizes a corium (UO2 - Zr0 ) melt 2
g of about 200 kg to interact with concrete. These tests will provide reliable ij measurements of gas and aerosol releases as well as directional heat fluxes.
f;j Comparsion tests of molten steel will be run to enable :arefui comparsion with pj the rather considerable body of steel / concrete interaction data.
Fission G
product elements will be doped into the corium to study fission product h
retention in the melt or release to the containment. The capability to conduct
- .E large scale UO., melt / concrete tests with sustained heatino is being developed.
~
fM
~
~
Q Sustained tests of hot, solidified, core debris (UO and steel) interacting 2
with concrete are under way. These tests will extend the data base for the y
CORCON model to an extent that the accident termination is predictable. The
.w T:
test variables include power, debris geometry and concrete type.
NL
,4
>d
f.,;ife(J.J.WRe:M E-M f,M. N, u... w
- c. 95Td1.E6. 7N. Ah.;i M. ' llb. A L-%.n.. ;9%-..[.E. ;W#.
U
.w.
.-a c.
-a
%]
4 g.
yy V
,:a 4
.;3 y.)].
M 6.
NRC Position W:
f"'
The interactions of fuel debris with plant structure must be considered on a gj plant specific basis. The downward penetration of the basemat is significant and full penetration cannot be ruled out. Dispersal through the basemat is r*4 slow enough that precautions may be taken that prevent threatening public e
!)
health and safety. Degradation of plant structures must be considered and h.. e failures of structures which in turn fail the containment system are possible.
- b..l1 The degradation of safety equipment, containment penetrations, and pljj instrumentation must be analysed as a function of the specific plant geometry.
gi Recent tests have shown that debris relocation following high pressure ejection med of the melt may be quite extensive and although this aids long term cooling, x.;
)Q.,y.
fuel debris relocation to the containment shell vicinity could result in localized shell failure.
N.R
$l b'd
,.n y
y
.r nj Y'b 3]
I t.[1]
wn I
,m
%1
..A 5::, o
?#=$
- 4 h*,*:)
a-tH h:'b
- y. m WQ
,.e J.M W-5 T-?lj a,.
'1 i$
e tp b
) 't
. 4j
hkb1?$
$$YWSN$hY'??bbWWN 'Y
- a S,.
Revised draft h.,,'*
12/23/S3 m
M M
h_h j:[f!
TECHNICAL ISSUE INFORMATION PAPER
~
4 Issue 1.2.9: ' Rate 'and Magnitude of Noncondensible Gas Production, Ex-vessel by hlh S. B. Burson, RES/DAE, Containment Systems Research Branch By F
1.0 Description of the Issue T.J The production of noncondensible gases during a severe accident follows as a 4g consequence of thermal / chemical interactions netween high-temperature core
- l) materials and concrete, metal-water reactions, or the degradation of contain-h ment building equipment during the accident. The generation of combustible fy components such as H and C0 is a major factor in assessing the containment i
2 ry integrity (Issues 1.2.1,1.2.3 and 1.2.4); moreover, the total rate of non-7 condensible gas generation does contribute significaniily to other issues, such
[
.as containment failure due to overpressurization by steam.
In the long term, y.'-j the noncondensible gas confined in the containment building could become a f
Tt significant factor by itself.
y
- A 2.0 Implications of the Issue to Regulatory Questions 2.1 (Reg. Q. 3) "How safe are the existing plants with respect to severe accidents?"
AakacN
~
Q f
Lv&W bN' If a scribed threshold } r fa re due to overpressurization were exceeded, e.
74
.the plant c
. Se cc...
ered to need additional accident prevention and nig mitigation measures. To make the determination of pressure loading, it is hg necessary to calculate the pressure trace that is expected to characterize any
,Q particular accident scenario. To obtain this information, in addition to the appropriate analytical tools, a knowledge of all of the pressure sources, including noncondensible gases, must be available.
p$3
~
T r1 kj la l-
hijjjyINNYNiMM9Ni[5i[@bk(dNNkgdv.NN6kbhfN
M s
@R '.
Q 2
h;h 2.
(Reg. Q. 4) "How can the level of protection for re accidents be igj increased?" and 13 f
(
"Is additional protection for severe accidents need. Wor rg i
The existing level of protection of a particular plant can be assessed by a pp$
comparison between the specific risks to which it is subjected and its
{
structural (and management) capabilities to survive the accident-imposed j
insult. There are a number of ways in which protection against containment building overpressurization can be increased. The first, and most obvious, is g
(oj' to find methods of reducing the intensity of the sources of overpressurization; gg another is to install systems which could lower the pressure if a given threshed-Ts exceeded.
.3 (Reg. Q. 5) "What additional research or information is needed?"
g$i N-wj tg Compared to the uncertainties associated with many other safety issues, 9
knowledge about the production and behavior of noncondensible gases is i
e.%g relatively advanced. The manner in which the CORCON code treats the production of gases during concrete ablation is
[anist;fc.
For the most part, pg.
the handling of partial pressures of noncon nsible gases in the atmosphere can y) be done with classical thermodynamic metho s.
The CONTAIN code incorporates the contribution by noncondensible gas co ponents in its atmosphere module so p!
that the pressure histories of all of th compartments throughout the Wh containment system can be followed.
If e problem of overpressurization is to I.M be managed by the installation of filte d-vent systems, further developmental
%m research will probably be desirable o esign a eptable ystems to permit m
@j: -
depressurization of the containment bui ih cessive radiological ny
%}l reTeases.
T I
\\h d
a^
m b)
,y/.
N' L
i A
a
- 1
(
? :1 O
f
{'Q:391.@fi%9is,$lig[.M pM(M O D P E dY.' M % ngj2 9 ? N dp M. M M i R 1.']
re vp)1 K
h....N'. -
3
.N
.i 1
'E 3.0 Subissues a
3.1 Thermal-Hydraulic Behavior of Secondary Containment dg TA M
The most significant concern related to the production of noncondensible gases hh is their contribution to the total internal pressure that develops within the containment system as the result of accident conditions. The gases of primary concern are CO, CO and H which are produced by metal-water reactions and as a 2
2
@j consequence of concrete ablation by high-temperature core debris. The primary Ifj source evaluation is derived from an understanding of these processes (treated
$j in the CORCON code) and a knowledge of the details of the accident scenario.
.jj The manner and detail with which the presence of noncondensible gases in the j{gj '
secondary-containment atmospheres differs somewhat among various codes (such as MARCH /NAUA and CONTAIN) so that thermal-hydraulic behavior of containment is id included as an unresolved issue.
N' W
y 3.1.1 Subissue - Hydrogen Production during Debris Bed Fonnation Immediately following_ vessel failure, highly uncertain transient conditions h
exist. If the reactor cavity is flooded with water, molten steel'and unreacted Ek cladding can decompose the water to produce an intense transient hydrogen 2q source. The phenomenon would be enhanced by possible steam explosions. The 41 process is highly dependent on the mode of vessel failure, the structural q
L.q details of the specific plant and the accident scenario hypothesized.
Ng 3.1.2 Subissue - Hydrogen Generation From Core-Concrete Interactions
..y
~ :QQ The primary ex-vessel source of hydrogen is the chemical interaction between
?!j water vapor released from the concrete basemat and high-temperature metallic D
components _(fuel cladding and structural steel) present in the debris pool.
fi The instantaneous composition of the core debris, together with the water g
release rate, thus determine the hydrogen production.
h
~
3.1.3 Subissue - Production of CO and CO during Thermal Ablation of Concrete 2
Q
.y k
-\\ i L
Y$?W5??$$b$$$W?I$NN?5haWNWAS ?D$$bi' N$N$ nil
- .f
~
s
=
' B;.'
4
~
5 n
The production rates of these noncondensible gasses depend upon the heat flux from the overlying debris and the chemical composition of the particular gj concrete under attack. The latter, of course, is a function of the specific plant being considered. The relative abundances of these components'may in turn be influenced by the composition of the debris as the gases bubble through e
it. The CORCON code treats these phenomena mechanistically, but the influence of freezing and crust fonnation is not yet adequately understood.
W 3.2 Influence of Noncond'ensible Gases on Condensation of Water and Heat k,4 Transfer from the Atmosphere to Containment Building Structures e.y;h The magnitude of overpressurization resulting from steam injected into the Q
containment depends sharply upon the effectiveness of the condensation process hj in mitigating the event. Condensation is, in turn, highly dependent upon the Ef.]j percentage of noncondensible gases present in the condensing atmosphere. These Q
effects are included in the modeling of the containment atmosphere in the CONTAIN code.
~
m
- !gj.
3.3 Direct Heating of Containment Atmosphere by Aerosols n.s Q
The high-temperature aerosols generated by the core-concrete interactions g
(including some radioactive fission products) will heat the containment 3
atmosphere and contribute to the threat of overpressurization. The VANESSA code under development at SNL computes the details of aerosol release from a y
molten-debris pool and will contribute to resolution of this issue.
ej
(.j 4.0 Approach to Resolution X
4.1 Near Term Approac W l
- .1 The CORCON-M002 code, which will include the treatment of slurries and crusts 3,
Q]
as well as the effects of an overlying coolant layer will be operational in early CY 1984.
It will provide improved quantitative data on the production NJ rates of all noncondensible gases that result from concrete ablation.
Refinement, testing, and verification will not be complete. The CONTAIN code
'J ghh b
y
QP.g.:f?Otlft%%5%%-QfffMn i%%.Ra S%y:i% {g%;f.MR@..:;...b. ;-Q..,. 3a.},...
95W R
n
~
a aw..-
Gn
~
R1 1
W; 5
- 0..
t 3%
already adequately handles the thermodynamic and thermal-hydraulic behavior of the containment atmosphere. There are a number of possible avenues to mitigation of the threat overpressurization,- e.g.:
reduction of the concrete ablation rate by core dispersal and cooling, providing the containment with a h
suitable filtered-vent system, and the installation of core-retention devices.
Research in these areas is not complete and the evaluation of these options
,f between various combinations of these approaches should be studied.
xi p%
n' j.$ -
4.2 Ultimate Resolution 'of the Issue M
The documented and verified CORCON-H002 code will provide adequat reliable predictions of the production rates of noncondensible gases re ting from y
core-concrete interactions. At present, there is only one her core-concrete f;j' interaction model having a sufficient level of sophist tion to provide d
meaningful comparative calculations in su port of CON verification. The 1
WECHSEL code, under development at the Kern c ungzentrum, D M e, FRG
- $.~
(KfK), when completed and available, will provide a valuable sburce of
{f independent computational predictions for comparison with CORCON results. The
{'%j, large-scale sustained heating experiments at the KfK Beta facility (scheduled N.
to commence in mid CY 1984), together with the large-scale quasi-prototypical hp l
experiments underway at Sandia, will contribute to the data base needed for the M
validation of both CORCON and WECHSEL.
C The details of the initial conditions (dictated by the accident scenario), and ni the structural details of the particulgant must be provid.ed. -Ttil! siaRhs h
for, and evaluation of, hte systems to reduce t inment p #
Hj resulting from accumulated noncondensible gases should ontinue until resolved.
k 5.0 Current Status of Understanding rg Although other possible sources of noncondensible gases have been considered, I
such as the thermal decomposition of organic containment _ building and equipment d'
t.a materials, only the gases produced by the interactions between high-temperature
.q M
core materials and concrete are considered in this treatment.
All concretes
.sj th qy
Qfy:M; Esg@ fly'.QQ%Q%Qj[Dy;Ija[.Uhyg f%[
hY m.
A L.
?%
.=
g
$std -
1 3
N contain a large amount of water; sene is chemically bound in the concrete, a
- 1. ?
-while some is entrained as free water in the interstices. At various 4
temperatures during thermal attack, both types of water are liberated so that a q;3 large amount of water vapor accompanies the thermal attack. The water and j
components of the molten-debris pool react to form hydrogen, carbon dioxide and 2[
carbon monoxide. Depending upon conditions and the availability of interactive
- n 3,ij substances, intact water vapor can also escape. At high temperatures, the w9].
concrete itself decomposes to produce CO and CO. Hydrogen combustion aspects 2
- 3 are treated in detail elsewhere. Hydrogen gas is included here only as a Gj noncondensible component.
p$$
t 5.1 Description of the Reference Case 7
'(
WV W
Conditions used to provide the sample problem in the CORCON-MOD 1 code are used to establish a conceptual frame of reference for further discussion. The case pd}
is selected because, however improbab1'e, it represents ' credible conditions Qj which might be expected after a whole-core meltdown of a large PWR system.
[
Clearly, the problems associated with potential over-pressurization become more g
severe for those plants having smaller total containment volumes and/or lower M
design pressures. At approximately 2.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> after shutdown, the entire molten Xry core of the reactor is assumed to fall on the dry floor of the reactor cavity.
[
The cavity is approximately 6 M in diameter. No water is assumed to reach the g
debris bed. Thermal ablation of the concrete begins immediately and proceeds indefinitely. A detailed discussion of the pool behavior is not included here, since all that is of concern is the rate of evolution of the noncondensible
$j gases which leave the concrete. During the first 2-3 hours of attack,
%j approximately 4000 kg of CO and 2000 kg of CO will be released. The 2
dN corresponding H release is approximately 80 kg (See Issue 1.2).
If the volume 2
d 6
3 i.,
of the containment building is assumed to be approximately 2.5 x 10 ft,
M (which is taken as representative of a large dry PWR containment), the j.
noncondensible gases produced, when corrected to standard temperature and h
pressure, will pressurize an evacuated containment building to one atmosphere il in 3-5 days.
~
n 2' h
..])
i a
y
?h
_N n1
-Qgy?fi6;9':ssyg Q $ gfy@A @ !G@QMO %%i% OajyjONN.
c 0Q.,
~
w,
+
(.9, 7
ab Ms l.;
A
't The gas evolution rate will fall along with the decay heat evolution; however, a
sufficient gas could eventually be produced to threaten most containment
.1 y
structures.
It is clear that the production of noncondensible gases resulting i
from concrete ablation could fail containment unless steps are taken to prevent b
i s,.:
it.
$l$
WA pfy 5.2 Effects of Noncondensible Gas Production M
5.2.1 Interfacial Heat Transfer C.,,
hf[jj
.It has been established experimentally that the bubbling of gas through the interface between two irmiscible fluids enhances the heat transfer between the 6~
1ayers. This results from two distinct effects, 1) distrubance of the surface, j,{q and 2) mass entrainment from the lower layer upward into the upper layer. The 3'
net result of these effects is to improve the transport of heat from the molten (h
fuel to the upper layer of the debris pool. This has the effect of increasing heat transfer to the upper cavity stru~ tures, thus lowering the rate of attack c
Cj on the concrete at the bottom.
y
.it un d.!
5.2.2 Contribution of Potential Containment Overpressurization
?.N.
.z D5 This effect has been covered in the previous paragraphs and must be given M
primary consideration.
q
- w Ud 5.2.3 Influence of Water Condensation Ten M
.}
One of the most significant accident mitigating phenomena is the condensation y
of water upon vapor on the massive structures and equipment housed within the 2
containment such as shielding, primary pumps, overhead crane, etc. The rate at which heat can be transferred to these structures is strongly influenced by the Ed amount of noncondensible gas present in the atmosphere. This especially,is
@~-2 4 -
true for horizontal surfaces. The CONTAIN code incorporates these effects into Yg its treatment of the thermodynamics of the containment atmosphere.
w n
4 a
- 9. d
.J
.h 3:
sh h.]
G
si g (jf.e W.RT.g.l:6 Q O!' W.e.#-A. mt.):M....%24%:. ". %'. -.k?.u:.T-:..1%.:. k.N.- T k
m b$
y:
8 el p$
6.0 NRC Position m.
LG For those accident scenarios which involve molten core / concrete interactions, mt S
noncondensable gases (combined with other sources of pressure) may be produced v
- fl to pose a threat to containment intergrity. The CORCON c, ode provides a method I
for predicting the release rates of non-condensable gases from core-concrete
.[s; interactions which is sufficiently well developed and validated for use in
- l.h severe accident analysis. These accident analyses mube consider specific k,f
- .,j reactor containment designs to draw safety conclusions for core melt accidents.
- e. a,
,I,.r. e tiI '
- f. *}
.n t:fj 3;. wm w
i:
y
(..
d,, '
- ).
- e-
$ ci
&4:1 ivy
'.i
?
p-t*
lb
'$9 h
I..
'R w
[1!
>4s t,i Y
d fj 8:
'1
s APPENDIX A EXAMPLE OF INFORMATION PAPERS ON TECHNICAL ISSUES m...
Rate and Magnitude of Hydrocen Production In-Vessel 1.
Implication of the Issue to Regulatory Questions As for most of the phenomenological issues, the relationship between the issue and the affected regulatory questions is not direct.
The position assumed on the issue affects the results of application studies which have direct input to the regulatory questions.
In the case of in-vessel hydrogen production, the position on the issue can affect the answers to:
How safe are the existing plants with respect to severe accidents? and How can the level of protection for severe accidents be increased?
In Mark I and Mark II BWRs, the rate and magnitude of release can affect the timing of failure due to non-condensible gas buildup with a major effect on consequences.
In Mark III BWR and ice condenser PWR, a high rate of production or release from the reactor vessel could defeat the capability of ignition devices leading to a mechanism for early containment failure. Alternative hydrogen control devices could be preferred.
Total oxidation of zirconium in-vessel could threaten some large dry PWR containments.
2.
Subissues Rate and magnitude of zirconium-water reaction in normal rod con-figuration.
Rate and magnitude of zirconium-water reaction during slumping.
Rate and magnitude of zirconium-water reaction during boil-off in lower plenum.
Rate and magnitude of steel-water reaction during boiloff in lower plenum.
e e/o
t lY' 5-
- \\
. K_
3.
, i:.r:sch.; h i.;tir.
Phase I ? asks.
The.appr:ach-to rescicticn.
e C., is bs+4d on
- the dss ti:;r:..
. c 1;pli:a-i;n 'of :re-.. RJ e t c
. :.i C 3 2.
model under develcpment at SNL to describe the progyccsicn of the-r:.elt in the lower. plenum will be integrated into the SCDAP code *.hich will cover the entire in-vessel phase.1The expected timing for,the integration of K.-e.
"ELFROG with SCCAp will 60t occur in Phsse !!, L:'.::ver.
Very little addi-(r. 2 6.
tional information will be obtained on this issue prior to che.mid-24 de--
e c.
w,. w cision.
Ccmparison calculations between 'SCCAF and MARCH 2 With PSF test SFD 1-1 should prcvide so.:.a ccafir. stien of r.edeling capability at least for the period preceding fuel ' slumping.
Co pari'sens will also be made between MARCH 2 and MAAP. -These comparisons are more likely to identify the magnitude of discrepancies for.different n:cdeling assumptions than to bring resolution.
The potential for extensive oxidation of upper plenum structures
= depends on the thercal-hydraulic coupling between the core and the upper -
plenum.
Model development and application of the TRAC code to examine up-per plenum and core flow patterns is being undertaken in early FY'84 at Los Alamos Phase II Tasks.
MELPROG and SCCAP will be integrated and validated against experiments in PSF, the ACRR and out-of-pile experiments at KfK.
Some accident sequences must be reanalyzed with the validated nodels to confirm the assumptions'made.in Phase I.
Validat' ion experiments covering '
processes-that occur after core slumping begins are:very difficult to parform
.and to instrument.
Large uncertainties will exist in.model predictions even after validation is complete.
I 4.
Status of-Understanding (This will be updated by 3/31/84) t i
~In the normal ' rod configuration, 'the ability to predict the oxida-1 tion rate of zirconium cladding is believed to be geod.
Results of SFD 1-0 were in reasonable' agreement with analysis.
Sh'R shroud zirconium has been treated as equivalent clad material in existing analyses.
A new model is
.,,being incorporated.into MARCH 2 to. treat EWR. shroud exidation.
s A-3
-.e----
Oxidation after slumping begins is much more uncertain.
Model assumptions in different codes treat this effect differently. Slumping of molten zirconium can decrease the potential for oxidation by allowing it to run off to a cooler area or can increase the potential by exposing un-oxidized surface to steam.
Uncertainties in this aspect of cladding oxida-tion can lead to a range of uncertainty of approximately 20-60 percent zirconium oxidation during the time in which the core is above the gridplate.
The potential for reaction of zirconium and steel in the lower plenum of the vessel is quite uncertain and depends on the configuration of the core debris.
If the core debris fragments upon contact with water in the lower plenum, a large surface area for reaction could result for oxidation to occur prior to quenching of the particles.
Formation of a debris bed could also enhance oxidation.
i Currect analyses indicate that some. steel structures above the core could reach temperatures near melting at which oxidation could be significant. The thermal-hydraulic coupling between the core region and upper plenum of a PWR is quite uncertain.
If large recirculation patterns 4
are possible, the potential for oxidation and hydrogen production could be very large.
5.
NRC Position (This would actually be written prior to 3/31/84)
It is recognized that some mechanisms may reduce the potential for zirconium oxidation following melting and slumping of the clad.
Because of the possibility of enhanced oxidation during slumping and subsequent oxidation following fragmentation of fuel debris in the lower plenum, model options have been selected for the analysis that lead to near complete oxidation of the cladding prior to vessel meltthrough.
The IDC00 oosition on this issue is substantially different lead-ing to lower hydrogen production in this time period.
This results in a reduced threat to the containment during this time period in each of the plants examined.
4--+v-Wr v-wmvt i-
+
W--
e
<m--
3---
r--
-v v'-
---wv
-w
l
. r ADDITIONAL GUIDANCE FOR ISSUE PAPERS
- At,,
j 1.
Implications of the Issue to Regulatory Questions In the first paragraph, provide a concise definition or description
- of the issue.
The purpose of this section is ~to describe how different posi-tions on the. issue could influence the answers to. regulatory questions and ultimately lead to diffgrent mid-84 decisions.
For example, consider high a n'd low positions on the magnitude of the source term in severe accidents.
A high position could imply:
reactors are not very safe, proposed backfits appear cost-effective, emergency planning is necessary to large distances.-
A low position could imply:
reactors are safe, backfits are unnecessary or are not cost-effective, minimal off-site planning is necessary.
Remember to include deterministic approaches as well as the PRA approach to regulatory decision-making in your thinking.
Keep it concise (1/2 page).
2.
Subissues Just list the subissues.
They can be in the form of a question or a statement.
Remember that issues are potential areas of dispute.
It is not
' the intent to have the subissues span the entire technical field.
Just identify i
the key areas of uncertainty.
3.
Approach to Resolution ii.
T-DC s ca..
- e. w % s. u.s.
This section should be very task oriented. ' Identify the specific activities that must be undertaken by contractors or by NRC staff to obtain:
(1) A position on the issue by approxic$tely 3/84, I
(2)
Ultimate resolution of the issue, if warranted.
Identify responsible organizations (con. tractors or NRC branches).
The big problem with this section is that what you come up with may not be consistent with tasks or schedules that are in progress.
I believe that you should 4
3
,r--
n--
- - ~ - - -
n
---e n ~ ~ - - - - - - - - - - - - - ~. - - - -
~ ~ ~ - - - - - - - - - - - - ~ - - - - -
.s,_
.-._._,.-.._-m e'
2
.x_
approach this with a fresh viewpoint.
In those cases whereyou are suggest-ing a~ task for.which there is no existing NRC, commitment, you should make an appropriate footnote so that theissue area leader will be made aware of it..
~
~
4.
Status of Uncerstanding
- ' 0 wnhb
. /.
s
. p. a r i-In some cases, this section would be huge if uncontrolled.
Be comprehensive but brief.
Make reference to reports rather than providing extensive descriptions of the results of research programs.
Cover all the subissues. Give an indication of what we-know now versus what additional work can tell us.
This section could be used to determine whether a decision
_ should be deferred until more information is available.
It may also be used to determine whether additional research is warranted.
I think that this section should not exceed 2 pages.
5.
'NRC Position This is the position that the NRC assumes on this issue for the mid-84 decision. We hope that you will be able to formulate a tentative position now, before the IDCOR meetings.
You should assume that a position will have to be established by approximately March of 1984.
Notice a posi-tion does not necessarily imply resolution of an issue.
For example, the
- NRC may want to assume a conservative position on an issue for the mid-84 decision if resolution would require considerable time and expense.
e em 9
s:
a
- ?
- s
,4
ROLE OF SARP IN PROVIDING INFORMATION NEEDS
.t T0.THE MID-84 DECISION i
SEVERE ACCIDENT RESEARCH PROGRAM (SARP)
SARP CONSISTS OF APPROXIMATELY 55 PROJECTS GROUPED UNDER 13 ELEMENTS.
GENERAL RELATIONSHIP TO MID-84 DECISION -- FIGURE 1 TYPES OF PROGRAMS,
g e
MODEL DEVELOPMENT MODEL DEVELOPMENT l'
e EXPERIMENTAL >
AND VALIDATION e
MODEL APPLICATION
.)
ROLE OF EXPERIMENTS SIMILAR TO LOCA-ECC PHILOSOPHY PROOF] TESTS ARE IMPRACTICAL EXPERIMENTS SU'PPORT MODEL DEVELOPMENT I
k l
.I o
e r%
y -
OUTSTANDING SEVERE ACCIDENT ISSUES 5+4..
ISSUE --
~(A)
A POINT OF DISCUSSION, DEBATE OR DISPUTE (B)
A MATTER OF PUBLIC CONCERN (C)
THE ESSENTIAL POINTJ CRUX (D)
A CU'LMINATING-POINT LEADING TO A DECISION GROUPS THAT HAVE IDENTIFIED SEVERE ACCIDENT ISSUES IDCOR ACRS NRC LIST'0F ISSUES -- TABLE 1 SEVERE ACCIDENT PHENOMENOLOGY
- SARP MODEL DEVELOPMENT AND EXPERIMENTAL PROGRAMS SAFETY ASSESSMENT
> SARP APPLICATION STUDIES NRC POLICY OR SARP APPLICA-DECISION METHODOLOGY TION STUDIES e
k g/
,-..e-,_
,c
--.. -.,_...--.,,. - -~ _. -
-r.
.,,,.,4-
,3
-- - - ---ee-,-
-x..
RELATIONSHIP BETWEEN ISSUES AND REGULATORY QUESTIONS ~ ---
LIST OF REGULATORY QUESTIONS -- TABLE 2 STAFF MUST HAVE ANSWERS TO REGULATORY QUESTIONS BY MID-84 INFORMATION NEEDS FOR REGULATORY QUESTIONS -- FIGURE 2 A REGULATORY QUESTION MAY HAVE A NUMBER OF ASSOCIATED ISSUES INPUT TO QUESTIONS IS PROVIDED BY APPLICATIONS STUDIES IN WHICH POSITIONS ON ISSUES ARE HIDDEN AS MODELING ASSUMPTIONS RESOLUTION OF ISSUES SOME SEVERE ACCIDENT ISSUES WILL BE VERY DIFFICULT TO RESOLVE STAFF POSITIONS MUST BE REACHED PRIOR TO MID-84 AND MUST BE FACTORED INTO APPLICATIONS STUDIES DISCUSSIONS WITH IDCOR AND IDENTIFICATION OF ISSUES 6
+
INFORMATION FLOW.TO MID-84 DECISION
-+;_ _
REPORTS ON APPLICATION STUDIES SEVERE ACCIDENT ISSUE PAPERS OUTLINE -- IABLE 3 NEAR TERM OBJECTIVES RELATE ISSUES TO REGULATORY QUESTIONS IDENTIFY SUBISSUES DEVELOP APPROACH TO RESOLUTION, PHASE I AND ll ESTABLISH PRELIMINARY POSITION FOR IDCOR DISCUSSION LONG TERM OBJECTIVES
~ ESTABLISH.NRC POSITION COMPARISON WITH IDCOR REGULATORY QUESTIONS PAPERS OUTLINE -- IABLE 4 NEAR TERM OBJECTIVES IDENTIFY TASKS THAT MUST BE PERFORMED IDENTIFY ASSOCIATED ISSUES LONG TERM OBJECTIVES COMPILE BASIS FOR STAFF POSITION
=
POLICY ISSUES PAPER r
,m
IMPACT ON SARP PROGRAM n=
RATIONALE FOR DEVELOPMENT OF INFORMATION FOR MID-84 DECISION IN SEVERE ACCIDENT RESEARCH PLAN NUREG-0900 IS LARGELY BASED ON PRA INSIGHTS.
CHANGE IN APPROACH TO MORE DETERMINISTIC BASIS NEED ADDITIONAL APPLICATION STUDIES NEED ADDITIONAL POLICY ANALYSIS STUDIES POTENTIAL SMALL IMPACT ON PHENOMEN0 LOGICAL RESEARCH s
ACTIVITIES TO EVALUATE AND REDIRECT SARP 1.
SURVEY OF SARP PROJECT OUTPUTS -- JULY 1983, TABLE 5 2.
EARLY INITIATION OF NEEDED ADDITIONAL STUDIES -- PENDING ACRS COMMENTS TASK FORMULATION (E.G., CASE STUDIES ON EXTENDED DBA)
REGULATORY QUESTIONS AND ISSUES OVERLAY'WITH SARP' PRODUCTS 3.
USE OF HIERARCHIES (AHP) IN RESEARCH PRIORITIZATION, FIGURE 3 e
m
Develop Support Base Evaluate Current Level q~~
of Methods and Data of Plant Safety Elements 1 - 10 Element 11 Accident Consequences and Risk Reevaluation Determine Need for Regulatory
/r Changes Element 13 Regulatory Analysis and Ac nt Evaluate Cost / Benefit Standards Decisions of Possible Plant Development l
Modifications Element 12
/
/
Risk Reduction and
/
Cost Analysis
/j
/
FIGURE 1.
RELATIONSHIP OF SARP ELEMENTS TO SEVERE ACCIDENT DECISIONS 3,-y
~
TABLE 1 SEVERE ACCIDENT ISSUES 1.0 Severe Accident Phenomenology 1.1 Progression of core melt in the reactor coolant system 1.1.1 Reactor coolant system thermal and hydraulic behavior
~'
1.1.2 Rate and magnitude of hydrogen production in the vessel i
and release from reactor coolant system 1.1.3 Fuel debris and in-vessel structure interaction 1.1.4 Fuel debris and vessel or vessel penetration interaction 1.1.5 Likelihood and magnitude of in-vessel steam explosions 1.1.6 Recovery potential prior to vessel failure 1.1.7 Primary system failure from overpressure 1.2 Loading of the containment 1.2.1 Containment thermal and hydraulic behavior
.1.2.2 Rate and magnitude of combustible gas production, ex-vessel 1.2.3 Distribution of combustible gases 1.2.4 Conditions leading to and resulting from diffusion flames
~
1.2.5 Conditions leading to and resulting from deflagration 1.2.6 Conditions leading to' and resulting from detonation 1.2.7 Likelihood and magnitude of ex-vessel steam explosions or steam spikes
~
1.2.8 Debris coolability in ex-vessel locations 1.2.9 Debris relocation following vessel failure 1.2.10 Fuel debris-containment shell interations 1.2.11 Fuel debris-containment floor interactions 1.2.12 Interactions between fuel debris and internal containment structures 1.2.13 Rate and magnitude of non-condensible gas production ex-vessel 1.3 Response of the containment and other essential equipment j
1.3.1 Characteristics and likelihood of containment leakage j
resulting from shock loadings 1.3.2 Characteristics and likelihood.of containment leakage resulting from steam spikes and/or hydrogen burning i
i 1
I
(
~
. TABLE 1 (Continued) 4 1.3.3 Characteristics and likelihood of containment leakage. ~
N-resulting from slow pressurization 1.3.4 Characteristics and likelihood of containment leakag'e resulting from external events 1 *. 3. 5 Characteristics and likelihood of containment leakage resulting from themal loading
' 1.3.6 Characteristics and likelihood of-containment leakage resulting from internal missiles 1.3.7 Potential for basemat penetration 1.3.8 Reliability of early containment isolation 1.3.9 Equipment and instrumentation survivability 1.4 Fission' product release and transport 1.4.1 Rate 'and magnitude of release of fission products from fuel (in-vessel) i -
1.4.2 Deposition of fission products during in-vessel transport 1.4.3 Rate and magnitude of release of radionuclides from fuel (ex-vessel) 1.4.4 Deposition of fission products in containment due to natural processes
- 1.4.5 Effect of engineered safety features on fission product retention 1.4.6 Deposition of fission products in other plant buildings 1.5 Ex-containment transport and consequences 4
1.5.1 Environmental dispersion 1.5.2 Food chain transport 1.5.3 Dosimetry and health effects 1.5.4 Modeling of emergency response 1.5.5 Cost analysis l
2.0 Safety Assessment 2.1 Characterization of plants and sequences 2.1.1 Plant categorization i
2.1.2 Identification of accident sequences 2.1.3 Quantification of sequence likelihood 2.1.4 Equipment performance and success criteria 2.1.5 Influence of operator action on accident sequence 5
t l
+,
,,----.7---.w.
.,,-,,--,w,.,,4.,___
.-_,7,,._m,.w--my._,-,_,
%.,,_m-,.-
..--.e- - -.
'a p
TABLE 1 (Continued) 2.2 Assessment of existing plants 2.2.1 Response of reference plants to selected severe accidents z_
2.2.2 Qualitative assessment of severe accident likelihood
~
2.2.3 Integrated probabilistic risk assessments for reference plants r
2.2.4 Credibility of PRA techniques 2.2.5 Applicability of conclusions concerning existing plants 2.2.6 Effects of uncertainties and sensitivities on the estimated severe accident consequences 2.2.7 Effects of uncertainties and sensitivities on estimates of severe. accident likelihoods 2.3 Assessment of plants with modifications 2.3.1 Design and operation changes to prevent accidents 2.3.2 Improvement in severe accident management capability 2.3.3 Design changes to mitigate accidents 2
2.3.4 Changes in emergency response capability 3
3.0 Decision Methodology 3.1.1 Selection of decision techniques 3.1.2 Selection of attributes to be used in cost-benefit analysis 3.1. 3 Selection of weights for cost-benefit analysis 3.1.4 Role of safety goals 3.1.5 Balance of decision approaches 3.1.6 Balance of preventive and mitigative measures a.
- 'f
,e.
4' e
e
~,
m
.r-
--,.,-,------,,,m
,-g.
--,.,n
,n,,,nn,
,.,,n~m-e,-, - - - --- -,, + - -. - - =,,
m,,.
j 4
. TABLE 2 ORGANIZATION-OF THE REGULATORY QUESTIONS 4
1.
Is additional protect' ion for severe accidents needed or desirable?
1.1 What are the'possible reasons?
~
1.2 What considerations apply to them?
1.3. What is the likelihood that performance could be improved by the alternatives available?
1.4 Are the likely, improvements worth the costs?'
2.
How safe are the existing plants with respect to severe accidents?
2.1 What should be considered in this. measurement of safety?
2.2 How do the terms' of measurement compare, including uncertainties?
2.3 Which accidents are to be considered and which onces can be ruled out?
2.4 Using these measurements, how safe are the existing plants?
- 3..How can the level of protection for severe accidents be increased?
3.1 What types of improvements are available?
3.2 How effective are they?
3.3 What are their costs and side effects?
i-4.
What additional research or information is needed?
'4.1 What are the information gaps bearing on the severe accident decision?
4.2 Are data necessary for implementation of the decision?
4.3 Are data necessary for confirmation of the decision?
e 4.4 Are there specific issues that require more data before they are decided?
j L
5.
How should the Commission decide the severe accident question?
5.1 When will sufficient information be available?
t 5.2 Should the decisions be generic or plant specific?
5.3 Should other regulatory issues and programs be integrated or excluded from the severe accident decision?
5.4 How will we assure that the requisite level of protection for severe accidents, once decided upon, is maintained throughout t
the life of a plant?
E 5.5 Should rulemaking be used or some other method for obtaining i
public comment?
1 8
,,8v,
..m.c
, - ~
-~
a
~
REGULATORY DECISIONS A
Staff Recommendations (Policy Issues Paper)
INTEGRATION A
~
Positions on Regulatory Questions (Information Papers and Briefings)
REGULATORY QUESTIONS l
Positions on Issues l
(Information Papers)
I
~
l Information Needs (Reports on ISSUES i
Applications Studies) i Research I
Results 3
i RC/IDCOR l
Discussions OTHER PROGRAMS SARP PROGRAM IDCOR PROGRAM W
l L______________
Figure 2.
Flow of Infonnation to the Regulatory Decision e
em 6
s. - - -..
a
+ -.,
.-------,,--.n-,---,,,,.,.,e-----w----.-,,,,--,-,-e,..,-,
.,.,.-,-~.,-------,----n-,--
-~,-,-,,--m--
,.e--.-
TABLE 3.
ISSUES PAPEP.S OUTLINE f
(1) The implications of the issue to regulatory questions
+
(2) Subissues l
(3) Approach tp Resolution Phase I tasks Phase II tasks IDCOR exchange (4) Status of Understanding 10/1/83 Initial 3/31/84 Final (5)
NRC Position Comparison with IDCOR O
,----n
,n---,.---,-----,-,.-----ve-,-,
,-,,.,n-,,--
r--,---
n.,--
n.-
3..
TABLE 1 Organization of the Reaulatory Ouestions g_ _
l 1.
What criteria should be used to determine whether additional protection for severe accidents is needed or desirable?
What are the possible reasons for additional protection? -
1.1 1.2 What co'nsiderations apply to them?
1.3 What criteria can be formulated?
1.4 What should be the role of the Commission's Safety Goal?
l 2.
How should the Comission decide the severe accident question?
2.1 Should the dectsions be generic or plant specific?
2.2 Should other regulatory issues and programs be integrated or excluded from the severe accident decision?
I 2.3 How will we assure that the requisite level of protection for severe accidents, once decided upon, is maintained throughout the life of a plant?
2.4 Should rulemaking be used or some other method for obtaining public consent?
l 3.
Now safe are the existing plants with respect to severe accidents?
3.1 What should be considered in this measurement of safety?
3.2 How do the terms of measurement compare, including uncertainties?
3.3 Which accidents are to be considered and which ones can be rule out?
3.4 Using these measurements, how safe are the existing plants?
4.
How can the level of protection for severe accidents be increased?
4.1 What types of improvements are available?
4.2 How effective are they?
4.3 What are their costs and side effects?
5.
What additional research or information is needed?
5.1 What are the information gaps bearing on the severe accident decision?
5.2 Are data necessary for implementation of the decision?
5.3 Are data necessary for confirmation of the decision?
,5.4 Are there specific issues.that require more data before they are decided?
4.
Is additiondl protection for severe accidents needed or desirable?
6.1 What conclusions do we reach when we assess available information in light of severe accident decision criteria (from question 1)?
5.2 What is the likelihcod that performance could be improved by the alternatives available?
6.3 Are the likely improvements worth the cost?
6.4 What is an appropriate balance of prevention and mitigation measures?
2-3
-w,
~
ss.
t ISSUE 1.1.2 N
RATE AND MGNITUDE OF HYDROGEN PRODUCTION IN-VESSEL
' ~
~ -
\\
- l 1.
Implication of the Issue to Regulatory Questions As for most of the phenomenological issues, the relationship between the issue and the affected regulatory questions is not direct.
The position assumed on the issue affects the results of application studies which have direct input to the regul,atory questions.
In the case of in-vessel hydrogen production, the position of the issue can affect the answers to:
2.
How safe are existing plants with respect to severe accidents?
2.3.
Which accidents are to be considered and which ones can be ruled out?
2.4.
Using these measurements, how safe are existing plants?
3.
How can the level of protection for severe accidents be increased?
3.1.
What types of improvements are available?
3.2.
How effective are they?
In Mark I and Mark II BWRs, the rate and magnitude of release can affect the timing of failure due to non-condensible gas buildup with a major effect on consequences.
In Mark III BWRs and ice condenser PWRs, a high rate of production or release from the reactor vessel could defeat the capability of ignition devices leading to a mechanism for early containment failure.
Alternative hydrogen control devices could be preferred.
Total oxidation of zirconium and steel in-vessel.could threaten some large dry PWR containments.
2.
Subissues Rate and magnitude of zirconium-water and steel-water reaction during slumping.
Rate and magnitude of zirconium-water and steel-water reaction during boil-off in lower plenum.
l
Rate and magnitude of ansessamenwe4em-end steel-water reaction in the upper plenum.
._~~
3.
Approach to Resolution Phase I Tasks.
The approach to resolution is primarily based on the development of more mechanistic models.
The largest uncertainty is due to a lack of knowledge of the oxidation kinetics during fuel relocation and in the debris bed in-vessel.
As a result, it is important to understand and adequately model fuel slumping.
Also, most of the data on oxidation rate was obtained for solid materials.
Therefore, there is a large uncertainty regarding the oxidation behavior of molten zircaloy and steel, particularly during flow.
The ELPROG model under development at SIL to describe the progression of the melt in the lower plenum will be integrated into the SCDAP code which will cover the entire in-vessel phase.
The expected timing for the integration of MELPROG with SCDAP will not occur in Phase I, however. Very little additional information will be obtained on this issue prior'to the mid-84 decision. ' Comparison calculations between SCDAP and MRCH 2 with PBF test SFD l-1 should provide some confirmation of modeling capability at least for the period preceding fuel slumping.
Comparisons will also be made between MARCH 2 and MAAP.
These comparisons are more likely to identify the magnitude of discrepancies for different modeling assumptions than to bring resolution.
The potential for extensive oxidation of upper plenum structures depends on the thermal-hydraulic coupling between the core and the upper plenum. Model development and application of the TRAC code to examine upper plenum a core flow patterns is being undertaken.in earl FY S4 at Los 7NL U kL'A AIN N - M d M M d O+-h}
A1amos.
As part of the molten core coolant interaction ~ program, steam as hydrogen generation rates under explosive and/or rapid rate conditions are being investigated in the FITS-series of experiments.
These tests are currently scheduled for completion in December,1983, and some of the results may be used in Phase I.
Also, the application codes WISCI and CSQ-CMCI are available now in reduced scale with limited validation.
The completion date for these codes is unknow1, but it is unlikely that they will impact Phase I.
1
~
The Severe Accident Sensitivity Studies program is using the PEPOD, i-PMDFR and ELPROG codes'to perform sensitivity studies which provide.
9_ -
perspective.in relating the quantity and rate of hydrogen production to h accident.g consequences. The completion date is uncertain and depends on the completion of IELPROG.
Therefore, the impact on Phase I is expected to be minimal.
The Damage Formation and Relocation Experiment Program is designed to provide a separate effects data base for LR fuel damage formation / relocation and reflood under a range of controlled accident sequences.
Initial tests in the DF series will provide high resolution data on damage formation phenomena in rod bundles for typical small and large break accident conditions.
These experiments are scheduled for completion in 1983 j
and will impact Phase I.
The remaining separate effects tests in the DF l
series and the fuel quench tests (DQ series) will not impact Phase,I. g p
Phase II Tasks.
MELPROG and SCDAP will be integrated and validated j
against experiments in PBF, the ACRR and out-of-pile experiments at KfK.
Some l
accident sequences must be reanalyzed with the validated models to confirm the assumptions made in Phase I.
Validation experiments covering processes that occur after core slunping begins are very difficult to perform and to instrunent.
Large uncertainties will exist in model predictions even af ter -
validation is complete.
4.
Status of Understandina (This will be updated by 3/31/84)
In the normal rod configuration, the ability to predict the oxidation rate of zirconium cladding is believed to be good.
Results of SFD 1-0 were in reasonable agreement with analysis.
BWR shroud zirconiun has been treated as equivalent clad material in existing analyses.
A new model is being incorporated into PMRCH 2 to treat BW shroud oxidation.
0xidation af ter slunping begins is much more uncertain.
Model assumptions in different codes treat this effect differently.
Slumping of molten zirconium can decrease the potential for oxidation by allowing it to run off to a cooler area or can increase the potential by exposing unoxidized l
surf ace to steam.
Uncertainties in this aspect of cladding oxidation can lead to a range of uncertainty of approximately 20-60 percent zircqnium oxidation l.
during the time in which the core is above the gridplate.
I
~
+
?.
The potential for reaction of zirconium and steel in the lower 2
plenum of the vessel is quite uncertain and depends on the configuration-of
_4 the core debris.
If the core debris fragments upon contact with water in the lower plenum, a large. surface area for reaction could result for oxidation to
~
occur prior to quenching of the particles.
Formation of a debris bed could also enhance oxidation.
Current analyses indicate that some steel. structures above the core could reach temperatures near melting at which oxidation could be significant.
The thermal-hydralic coupling between the core region and upper plenum of PWI is quite uncertain.
If large recirculation patterns are possible, the potential for oxidation and hydrogen production could be very large.
5.
NRC Position It is recognized that some mechanisms may reduce the potential for zirconium oxidation following melting and slumping of the clad.
Because of the possibility of enhanced oxidation during sitsnping and subsequent oxidation following fragnentation of fuel debris in the lower plenum, model options have l
been selected for the analysis that leads to near complete oxidation of the cladding prior to vessel meltthrough.
The IDCOR position on this issue is substaa+4 !!; d'Timent. seading to lower hydrogen prodor+4x :n unis time period. This results in a reduced a
threat containment during this time period in each of the plants examined.
l f& E i
o
' tu
- . dv i
T L e o ik v d ' d
/
l i
I l
i
_____,_..-..,_,,,_.._...,_...____m___.._.
c-.. -. -. _, _. _,,,. -. _ -
e
~~
RESPONSES TO QUESTIONNAIRE
-~~
MODEL SUPPORT & VALIDATION PROGRAMS This paper contains selected responses from a questionnaire Dr.- Ross sent to SARP contractors in July.
Specifically, it contains responses to questions about the development and validation of computer models.
The paper is organized by SARP program elements.
Each Fin Number is listed under the appropriate Program Element.
Under each Fin No. there are one or two subheadings:
" Experimental Tasks" and/or "Model Development Tasks".
Responses to questions under " Experimental Tasks" give information
~
on experimental data to be used for model development and validation.
1 Responses to questions under "Model Development Tasks" will provide infor-mation on model development and validation plans.
The questions whose responses fall under.each subheading are presented below.
Experimental Tasks Question 8 By whom and for what purpose will the results of the experiments i
be used?
(For example, will they be used to develop or validate computer l
codes?
If so, which codes or modules under which SARP projects?)
j Question 9 When will results of the experiments be available or the uses you listed under Item 87 Can results of the earlier experiments in this task be used before all of the experiments are completed?
If so, what intermediate results will be available and when?
y eF
?
=
2 Question 10
-#=-
Are you providing information to or receiving information from any other SARP program on an informal basis? If so, what kind of information is ~
being exchanged with which SARP projects?
Model Development Tasks Question 4 For each code or module listed above, please indicate what input, if any, is needed from other SARP programs for development of the code; which program is expected to provide each piece of information; when you will need i t.
Furthermore, if your code requires input data that must be obtained from other models or computer codes, are the models available or under development?
Input Needed Source Date Needed W
f Question 5 How will each code be validated (i.e., compared with experiments)?
Please list the experimental data needed to validate each code, the source of that data and when you need it.
Validation Plans and Experimental Data Needed Source Date Needed l
l l
Question 6 Over what range of temperatures, pressures, or other conditions will l
each code be validated?
Parameter Range i
,-.-.,-..-,-.._--___,.y,,,,y,,-,..,._,,-m__,.-
,,_,-,--.,_.._,r._,-
..y
,,.._ -.,,.__ ~
RESPONSES TO QUESTIONNAIRE
.._:3_,.
COMMENTS AND EXPERIMENTAL PROGRAM INFORMATION NEEDS The questionnaire sent to SARP contractors by Denny Ross in July contained two questions that asked about input contractors needed to complete their experimental tasks and one page'for general comments on the SARP program.
Responses to the two questions and general comments on SARP are compiled in this paper.
The questions were numbers 11 and 12 in the Experimental Tasks section.
The questions and the format for their answers are given below.
s Question 11.
Is the completion of your experiments and final report contingent upon the receipt of information from any other project?
If so, what input do you need, which projects do you expect to provide it, and when do you need it?
Information Needed Source Date Needed Question 12.
Do you need more information on the purposes of your experi-nents and uses of your results to help focus your work?
If so, what specific information do you need; when do you need it; and from whom would you expect to receive it?
Information Needed Source Date Needed e
F w
2 6-L, _
Comments were made in response to the following statement:
If you have any comments rega'rding this program [the one -~~
for which the questionnaire was completed] or its relation-ship to other SARP programs, please make them here.
In this paper, comments on SARP programs and answers to questions 11 and 12 are organized by program element.
Each response is preceded by the Fin No. and title of the program for which the questionnaire was com-plated.
If a Fin No. and program title are followed by "None", a question-naire was received for the program,, but there were neither comments nor response to question 11 and question 12.
If a Fin No. is omitted, no
- questionnaire was completed for the program.
d O
'l e
e
=
.=
~
I
' RESPONSES TO QUESTIONNAIRE
-5,_
PROGRAM DESCRIPTIONS 1
In July, Dr. Ross sent a questionnaire to all contractors per-forming tasks-under the Severe Accident Research. Program (SARP).
The tasks under SARP genera,lly fall into one of three categories:
(1) e.nperi-mental, (2) model development, or (3) model application.
The questionnaire was, accordingly, divided into three sections with one section addressing each kind of task.
Some SARP programs consist of several tasks, and in those cases, contractors completed a questionnaire for each task.
Responses to the questionnaire are now being compiled in a series of papers. This paper contains responses to questions designed to give an overview of each task.
{
The paper is organized on the basis of the thirteen SARP program elements and the Fin Numbers under those elements.
For example, suppose
" Program Element A" is the title of one section of the paper, and " Fin No. B" is a subheading.
Under " Fin No. B", there can be up to three divisions j
(" Experimental Tasks", "Model Development Tasks", and "Model Application Tasks") depending upon the type of work being done under Fin No. B.
Under each division heading there is information on the objective of the task i-and the scope of the work being performed.
j Questions whose responses fall under each division are listed l
below.
4 EXPERIMENTAL TASKS Question El.
What is the task title?
)
Question E2. What are the task's objectives?
i i,
k
+
..~
_._-,__..-__,,_.,___.___..___..-_.._L_...-__,._._________-_.._.,_.__~.__.____
1 1
2 Question E3.
List the experiments to be conducted'under this task.
Fo'r -
eir :
each experiment, please give the name and/or' number which identifies the experiment, a brief summary of its objectives, and the date it is expected to be completed.
If your program involves a number of small experiments, you may list these as groups or series.
Completion -
Experiment
' Objective Date l
Question E4.
For each experiment or group of experiments, list important parameters being controlled, key output parameters measured and the ranges these parameters covered.
Experiment Parameter Range Question E5.
Experiments often provide qualitative information.
Do you plan to make qualitative observations duri~ng these experiments?
If so, please list the qualitative information you expect to glean from each experiment or group of experiments.
Question EG.
In some experimental programs, the raw data themselves are of primary interest while in others the data are used to develop correlations or to calculate parameters of interest.
Please list the quantitative information you expect to have at the end of the experimental program.
This list should be very similar to a list of titles of tables, graphs, and charts from your final report.
Question E7.
For each of the pieces of information listed after question 6, please estimate the uncertainty.
The estimate can be written next to the parameter under question 6.
e h
h
I.
3
-en~ ' '
MODEL DEVELOPMENT TASKS Question M1. What is the task title?
I Question M2.
What are the task's objectives?
Question M3.
List codes being developed and give a brief description of each code's capabilities.
If a code contains several modules, please list and describe the capabilities of each major module also.
Code Name Code Capabilities Question M7. When will writing of each code be completed and the code verified?
When will the validation of each code be complete? When will documentation be complete and the code made available to NRC contractors?
4 Completion Date Completion Date Code (Writing and Verification)
(Validation)
Availability Date l
MODEL APPLICATION TASKS i
Question Al.
What is the task title?
Question A2.
What are the task's objectives?
i a
e
~
i
.pu-!
RESPONSES TO QUESTIONNAIRE MODEL APPLICATION STUDIES The questionnaire sent to SARP contractors by Denny Ross in July was divided into three sections:
one for experimental tasks; one for model development tasks and one for model application tasks.
This paper contains responses to questions in the model applications section.
The questions are written below:
Question 8.
What regulatory issues does this task address?
Question 3.
Please list the codes that will be used in this task.
For each one indicate the research project and laboratory that will provide the code and the date you expect to receive it.
Code Source Date Expected Question 4.
What other information do you need to complete your task?
When do you need it and who can provide it?
Information Needed Source Date Needed Question 6.
What will be the sources of significant uncertainty, if any, in your final results.
, e,.
~
.....g.
2 Question 7.
How and by'whom will the results of your model application.
work be used in the Severe Accident Research Program.
Question 5.
When do you expect this model application task to begin_and end?
Start Date:
Finish Date:
Responses to.the model application questions are organized by program element.
For each model application task under~a particular program element the Fin Not and program title are given, and the responses to the six questions listed above follow.
Some SARP programs have more than one model application task.
In those cases, each task is treated separately.
Each task title is given and followed by answers to the six questions.
For other programs there were no responses in the model applications section of the questionnaire.
In those cases', the Fin No. and program title are tested and "None" is written under them.
If a completed questionnaire was not received for a particular program, that program's Fin No. and title are not listed.
l l
i SEP 191983 List of Addressest for Memorandum Dated:
- ai _ !
O. Bassett R. Wright J. Long J. Rosenthal T. Speis p' U V. Benaroya C. Tinkler R. Vollmer P. Kuo J. Costello W. Farmer B. Burson W. Butler
)
V. Noonan R. Bernero G. Marino L. Chan M. Jankowski T. Walker W. Pasadag F. Akstulewicz P. Easley R. Blond J. Hulman I. Soickler J. Martin, E. Branagan S. Acharya B. Richter P. Baranowsky C. Eng B. Agrawal C. Overby J. Murphy F. Rowsome G. Burdick J. Wermiel L. Soffer E. Jordan-G. Arlotto R. Mattson D. Muller Z. Rosztoczy L. Shotkin l
C. Serpan J. Telford R. Curtis
g'" ""%
umiso stares NUCLEAR REGULATORY COMMISSION wAsmwarou. o. c. 2 ossa SEP 19 1983 MEMORANDUM FOR: Those on Attached List FROM:
James C. Malaro, Chief Regulatory Analysis Branch Division of Risk Analysis, RES
SUBJECT:
SEVERE ACCIDENT ISSUE PAPER Rich Denning and Ray DiSalvo from BCL will be at the NRC from September 21-23 to work with NRC staff assigned to prepare Severe Accident Issue Papers (September 15 Memo from D. Ross). There will be kick-off meetings at 1:00 p.m. on Wednesday, September 21 in Room 019 NL, 9:00 a.m., Thursday, September 22 in Room P-422 and 2:00 p.m. on Thursday, September 22 in Room s
106 W111ste Building to discuss help which will be provided by BCL. The meetings will last approximately two hours. My understanding is that BCL g,~
has already prepared strawman issue papers for over 50% of the assigned issues.
w ('
am9s C. Malaro, Chief Regolatory Analy is Branch Division of Risk Analysis, RES
}.
e es E
4 t'1'>
.L4
,y n
s M' 9hi' 4
4
l' 100483 Issue 2.1.1 Plant Categorization
-Q_
l.
Implication of the Issue to Reculatory Questions
'The ability to develop plant classes has great effect on the NRC reg atory decision process.
Since it is desired to make severe accident decisions by mid-1984 on a's many existing LWRs as practical, the issue 'of developing plant classes by evaluating the results of the existing PRAs (12) is highly important.
~
l There is a direct relationship between the issue and the affected regulatory questions.
The position of the issue can effect the answers to:
2.
How should the Commission decide the severe accident question?
2.1 Should the decisions be generic or plant specific?
3.
How safe are the existing plants with respect to severe accidents?
3.3 Which accicents are to be considered and which ones can be ruled out?
3.4 Using these measurements, how safe are the existing plants?
4.
How can the level of protection for severe accidents be increased?
4.2 How effective are they?
- 5. 4 What additonal research or infonnation is needed?
F i
5.1 What are the information gaps bearing on the se<cre t
=
accident decision?
., = -
5.2 Are data necessary for implementation of the decision?
.r
. 5.3 Are data necessary for conformation of the decision?
5.4 Are there specific issues that require more data before they are decided?
~~
i 6.
Is additional protection for severe accidents needed or desirable?
6.2 What is the likelihood that performance could be improved by the alternatives available?
6.3 Are the likely improvements worth the cost?
2.
Subissues l
It is possible to group plants having similar dominant accident j
sequence characteristics?
Do the existing PRAs (12) provide enough information to make inferences on all existing plants?
Can insights drawn from plant categorization be used in severe accident rulemaking?
If so, how should uncertainties be considered?
Given the associated uncertainties, can plant categorization be used to identify weaknesses in plant design?
What level of information is appropriate for plant categorization--
function, systems, or component level?
Can event trees for plant classes be formulated?
If so, how useful are they?
.~
. Plants are categorized based on dominant accident sequences and 4
plant systems needed to mitigate the accidents, consequences resulted l
~!
from these incidents are not considered.
Does this have any impact on regulatory decisions?
3.
Approach to Resolution The Accident Sequence Efaluation Program (ASEP) is the primary NRC research to support the issue.
ASEP is funded and managed by the Division of Risk Analysis of the NRC and the primary contractors are SNL and INEL.
The NRC approach uses the twelve existing PRAs as the starting point to develop plant classes.
Dominant accident sequence classes are fonnulated based on the "like" characteristics of initiating events and system failures of the PRA dominant accident sequences. The plant systems and support systems needed to mitigate. the dominant accident sequences are identified and a plant survey is undertaken to obtain accident mitigation fluid and electrical system drawings for as many plants as possible for all systems and support systems.
Concurrent to'the plant survey, the plants are grouped into appropriate containment types.
After the system drawings are collected, simplified drawings including the important components and dependencies are drawn.
The simplication process is
^
based on simplifying Piping and Instrumentation Drawings (P&ID's) and Electrical One Line Drawings.
Once the systems have been simplified, comparisons are made of the systems (e.g., compare all AFW systems) considering significant differences in the redundancy, diversity, or support system dependencies for each system of
~
interest.
4 :
--.n,
.wr.-
i 1
. For each of the selected accident sequences to be examined, system designs
~g.,
for systems used to mitigate the accident in each plant are grouped, depending on similar system characteristics.
These groups are used to form combinations of known system configurations that are possible for mitigation of the selected accident sequences. These combinations are defined as initial plant classes for the particular accident sequences of interest.
For example, suppose all Brev/
Aupt systems could be grouped into three configurations;'and AFW 3.
In addition, suppose all high pressure injection systems (HPIS) could be grouped into two categories; HPIS1 and HPIS2, Furthermore, suppose that the known combinations of these system groups are only; AFWl/HPIS1, AFW2/HPIS2, and AFW3/HPIS2.
- Then, for calculating the sequence frequency for a sequence involving the failure gof these,two systems (e.g., TMLU sequence), three calculations are made.
One calculation for each of the above known combinations, which are now defined as initial plant classes for this sequence.
After the deductive modeling, quanti-fication, sensitivity and uncertainty analyses tasks, the next step is to use the quantitative results and important insights to collapse the initial plant classes to the final plant classes.
The approaches on modeling and quanti-fication are covered under Issue 2.1.3, Quantification of Sequence Likelihood.
Up to now, the results are presented in the following heirarchial format; by containment type; withi.n containment type, by sequence class; and within sequence, by initial plant class.
Figure 1 is a illustration of the ASEP.
approach to plant categorization.
The next step is to collapse the initial plant classes, where possible, based on similarities in the dominant con-tributors, the sequence frequency and its bound for each plant class.
Next, the results are arranged in a different hierachial format with the
[
containment type as the first level.
The second level the plant class
.n._
APPROACil T0 PLAHT CATEGORIZATIONS Sequence Sequence information Classes Initial Collapsed ** Final ***
Under Final Frequency Upper and Dominant Containment
- Sequence Plant Plant Plant Plant
-(Best
. Lower Dominant
. Type Classes Classes Classes Classes Classes Es tima te)
Bounds Contibutors "A"-2 Trains 3 E -4 of AFWS
\\
T*PCS-AFWS 5 E -5 Hardware /T & M w/o F & B 1 E -7 Failures of AFWS (Sequoyah, h Watts Barl M
"F"-2 or 3
/
Trains of T=PCS.AFMS AFHS w/o F&B
\\
(Sequoyah, "B"-3 Trainst4atts Bar, of AFWS ricQuire) w/o F & B (McQuire)
"G" 2 or 3 Trains of Ic2 Condenser--
AFWS w/o Free Standing F & 8 and Still ains of SLOCA.HPIS
\\
8 E -4 SLOCA HPIS 1 E -6 Operator Procedures 1 E -7 for identifying SLOCA and "C"-3 Trains "C"-3 Initiating HPIS of HPIS Trains of (Sequoyah, ifPIS l
Uatts Bar, (Sequoyah, McQuire)
Watts Bar, McGuire)
- This is an illustration only
- cCollapse classes by examining sequence frequencies, their upper and lower bounds, and thetr dominant contributors.
- Redefine collapsed plant classes by examininq specific plants.
5-
. 1 and the third level is 'the sequence class. By the reformatting the plant.
. g_
classes are redefined.
The' collasped plant classes intrinsically contain specific plants within their description, by venture of the specific _ plant's related system characteristics.
After all sequences within contained typed are identified as continuing similar combinations of plants for those sequences, these similar combinations define the final plant classes and their associated characteriestics.
For each final plant class, the dominant accident sequence, the associated likelihoods, the related range of values and the dominant contributors to the sequence likelihood are given.
4.
Status of Understanding ( As of 9/30/83)
A better understanding of the status of plant categorization can not be made at the time.
Initial plant class identification is progressing and will be completed by mid October,1983.
The number initial plant classes will be more than anticipated due to the large variability in plant design.
5.
NRC POSITION Given the large variability in plant design and the number of attributes that makes the accident sequences important, which include hardware characteristics, functional capabilities (e.g., for feed and bleed), and special vulnerabilities to common cuase failures, the NRC believes the potentially large number of plant classes can be reduced to a more manageable size by two means.
First, the NRC expects that some plant classes that differ with respect to less significant attributes can be consolidated.
Further, and more importantly, the NRC believes that plant classes can be combined if regulatory action
~
n - -..
,w-
,v-+
r
-m.-,
m
-+r pm-w r-
e can be proposed to eliminate their differneces in attributes.
That is,.,
plant classes that are different because of, for example, a particular HPIS dependency on component cooling water, could be merged if one SARP regulatory recommendation is the elimination of such dependencies.
Such regulatory actions could be relatively narrow (for a specific problem or a specific set of plants) or more broad (e.g., requiring all plants to perform reliability anal,yses).
Through the development of relatively detailed initial plant classes and their subsequent collasping and the associated identification of regualtory recommendations, the NRC believes that a defensible and manageable set of plant classes can be developed.
l
-i
6-can be proposed to eliminate their differneces in attributes.
That is, plant classes that are different because of, for example, a particular ^
-7 HPIS dependency on component cooling water, could be merged if one SARP regulatory recommendation is the elimination of such dependencies.' 3dch regulatory actions could be relatively narrow (for a specific problem or a specific set of plants) or more broad (e.g., requiring all plants to perform reliability analyses).
Through the development of relatively detailed initial plant c' lasses and their subsequent collasping and the associated identification of regualtory recommendations, the NRC believes that a defensible and manageable set of plant classes can be developed.
)
S
- e
',},
e
> gr,*, '
g g
e 4:
J E
3---
v, 7-,,
,.,,v.
.,-w,
..,,--.,----y,-
,c.4
-,v-,,..
,m.-
,_.c-.--..--,-
. - - -.w-.--,.
(*
- i 9
WHITE PAPER m
CONDITIONS LEADING TO AND RESULTING FROM DIFFUSION FLAMES y.
r..
A SUBISSUE STATEMENT The presence of deliberate distributed ignition systems in some contairunents has led to thq following questions concerning hydrogen diffusion flames.
Under what postulated accident conditions will a continuously burning hydrogen diffusion flame exist in' the containment?
In particular, will a cteady diffusion flame be established for a specified combination of hydrogen cnd steam release rates, igniter locations, accumulated containment gas concentrations, and water spray fluxes?
Will the diffusion flame be attached to the igniter, the release site, or some intermediary location at which cquipment or containment structures act as flame-holders?
If a diffusion flame is established, how large will the flame be and when will it be extinguished via steam inerting and/or oxygen depletion?
What are the corresponding thermal loads on the containment and key equipment in the vicinity of the diffusion flame?
B IDCOR POSITION AND BASIS Diffusion flames can occur during, accident scenarios involving pre-activated igniters placed reasonably close to the hydrogen release site in c large open section of the reactor containment building.
Stable, standing flames are expected with relatively large hydrogen release velocities (compared to background velocities and local disturbances; e.g.
suppression pool wave motion).
Detached, irregular flames are expected with large steam / hydrogen injection ratins and/or very small release velocities.
Flame heights and heat fluxes can be calculated now for buoyancy-dominated diffusion flames and for momentum-dominated flames, providing there is unobstructed air
[
cccess to the flames.
Effects of steam acetanulation (below the combustion limit concentration of about 5 vol %) and obstructed / vitiated air flow are
/3 P
~.. ~.....
6 s
being investigated in current test programs.
As of now, tests involving continuous hydrogen injection / combustion into a
~
closed volme have been conducted in. four different test vessels.
In two vessels (the 5 cu a Fenwal sphere and the 18 cu m EPRI/Acurex cylinder) the k.
continuous injection tests primarily resulted in a series of deflagrative burns involving flame propagation and associated combustion overpressures.
In the other two vessels (the 2,100 cu a Nevada dewar and the 1/20-scale BWR Mark III) continuous injection tests primarily resulted in continuously burning diffusion flames with only, minor pressure increases.
Although there are no quantitative criteria to delineate these two combustion modes in closed vessels, unconfined diffusion finne stability data correlations indicate that Eteady standing flames should be expected at low steam / hydrogen injection ratios, aource diameters on the order of a few em or larger (depending on steam mass fraction) and for release velocities that are small compared to backsound cross-flow velocities (associated, for example, with fans and sprays).
The location of the diffusion flame depends on the hydrogen / steam / oxygen concentration distribution (including accmulated background concenti ations),
the velocity distribution in the plume, and the presence of structures that can act as flame-holders.
Observations from the Nevada tests indicate that the flame often migrates in stieps back and forth from the igniter to the release site as acetanulated background gas concentrations change during the test.
Large steam / hydrogen injection ratios promote detached flames located between the release site and the top of the vessel.
l Diffusion flame heights and heat fluxes can be predicted for scenarios in which the existing data correlations for buoyant diffusion flames or turbulent jet diffusion flames are applicable.
Buoyant diffusion flame correlations apply to small release velocities, relatively open areas of the containment, negligible acetanulations of steam, and background oxygen concentrations approximately equal to those in ambient air.
Turbulent jet diffusion flame data correlations refer to high release velocities and negligible background
[
sources of manentum compared to jet meenttan.
There are relatively few data for flame heights and heat fluxes when both buoyancy and jet momentum are f
w w
--v-s,wvv v-w
-e--we a-
-,-m-r-e+w rmm-,
-.ew wwc w-- - - -
e,~c
-n-we v
m-r~w
---,-w-----
,-w.,vv--
w--
~ -
6 s
s important, or when enclosure obstructions distort the flee shape.
Flame extinguishment due to oxygen depletion can be predicted, but flame heights and heat fluxes in partially vitiated air are highly uncertain.
Experiments to fill in these data gaps are now being conducted in the Nevada dewar and the Sandia jet flame facility, and tests in a 1/4-scale WR Mark III containment g,
e..
vessel are scheduled for mid 1984.
There is currently an absence of computer models to provide realistic calculations of thermal loads produced by diffusion flames in containment buildings.
This need is evidenced by the fact that pre-test computer predictions (based on the assumption of repetive deflagrations) have not compared favorably with temperat'ures and pressures measured in the Nevada dewar continuous injection tests. Currently available enclosure fire models do not account for effects of obstructed air flows, enclosure pressure increases, and spray cooling, all of which are significant in reactor accident scenarios.
C DIFFERENCES FROM NRC POSITION The only known difference between industry and NRC positions on this issue is the treatment of the effects of diffusion flames on equipment survivability for some containment configurations and accident scenarios.
D RESOLUTION OF DIFFERENCES l
Since diffusion flames only threaten containment integrity through possible thermal damage to key equipment, strategies to resolve the diffusion flame issue should depend on the results of equipment survivability tests in l
the Nevada dewar and tests to determine heat flux distributions in the 1/4-scale WR Mark III.
Analyses of data from these tests will determine whether there is a need for new computer codes or additional testing.
R. Zalosh 11/22/83
- ~
-n~a
- u..a :
ww
--ww -
~
s 4
WHITE PAPER CONDITIONS LEADING TO AND RESULTING FROM DEFLAGRATIONS p.
A.
SUBISSUE STATEMENT Delayed ignition of gn acetanulated hydrogen-air-steam mixture usually results in deflagrative combust. ion; i.e. subsonic flame propagation away from the ignition site.
Analyses 'of accident scenarios in which hydrogen deflagrations are anticipated entail answering the following questions.
Will a particular ignition source located within a hydrogen-air-steam mixture of specified composition and fluid dynamic state trigger a deflagration?
If so,
~
what fraction of the hydrogen will actually burn?
How rapidly will the flame propagate through the mixture?
What are the resulting deflagration temperatures and pressures and how quickly will they decay?
B. IDCOR POSITION AND BASIS A substantial data base and an assortment of computer codes are now available for answering the preceding questions.
With regard to ignition criteria, gas concentrations and surface temperatures required for deliberate ignition of static and turbulent mixtures have been determined for glow plug and for hot coil igniters.
The only combustion uncertainties remaining in the design of these igniters are: 1) igniter effectiveness in the presence of dense water sprays; and 2) ignition temperatures for quatenary mixtures formed by carbon monoxide or carbon dioxide addition, to hydrogen-air-steam.
With regard to hydrogen burn fractions for hydrogen concentrations in the range 4-8 vol%, phenanenological computer models are best suited for evaluating the subtle balance between burning velocity, buoyancy, and heat losses that determines the extent of flame propagation.
With regard to flame speeds.
[
opproximate estimates for single enclosures can be made based on data compilations and comparisons of pressure rise data with calculations using
_ na
- ~ -
~~m n m
- u. w w. a. = -
4 flame speed as a " calibration" parameter. The only remaining gap in the flame speed data base is the-lack of relevant large-scale data for flame cecelerations associated with flame propagation through a multiple compartment
' ~
enclosure.
With regard to pressure rise and decay, reasonably accurate determinations can now be made using best-estimate values of relevant g,
- e..
turbulence parameters.
The hydrogen deflagration data base includes data from test vessels ranging in size from 17 1 to 2100 cu m.
Tests in the smaller vessels have been devoted primarily to the determination of flammability and ignitability limits for hydrogen-air-steam mixtures.
Tests in the 2100 cu m Nevada Test Site Dewar have been geared toward verifying scaling considerations in applying the extensive data base on overpressures, flame speeds, etc. obtained from hundreds of tests in medium-scale ( 0.3 - 25.6 cu m) vessels.
Deflagration data for mixture compositions only slightly within the flammability envelope indicate relatively low pressures and flame velocities.
Flame shapes and speeds are strongly influenced by buoyancy for these weak nixtures with hydrogen concentrations in the range 4-8 vol5.
The fraction of hydrogen burned and the peak pressure are strongly influenced by igniter location and the level of fan-or spray-induced turbulence as well as the nixture composition.
In the absence of fan-or spray-induced turb ulence, hydrogen burn fractions and burning velocities are quite small and peak pressures are well under the calculated adiabatic, constant volune combustion values for this concentration range.
As hydrogen concentrations reach and oxceed about 8 vol5, hydrogen burn fractions, flame speeds, and peak pressures increase sharply even for initially quiescent mixtures.
Steam cddition has been shown to be a mitigating influence in the sense that it reduces flame speeds and peak pressures.
The only deflagration question which still remains in applying the data base from the medium-scale test facilities is the possible effect of equipment, structures, and multiple compartments on augmenting turbulent flame speeds.
However, some limited data from tests in the AECL 6.3 cu m sphere
}
indicate that grating obstructions serve more as a heat sink than as a turbulent flame accelerater.
In fact, current evidence suggests that water
,s spray induced turbulence is more important than turbulence induced by weak mixture flame propagation around obstructions.
This is reassuring because water ' spray effects do not present the same scale-up uncertainties as flame prongation around obstructions.
Empiricabob'servations on weak mixture hydrogen deflagrations h e been used in the development of deflagration computer models.
Three different types of models have been used for test simulations and containment burn predict' ions.
The simplest type of model is the empirical model in which there is no description of flame propagation; merely a uniform deposition of combustion energy at an empirically prescribed rate and hydrogen burn fraction.
Phenomenological models employ mechanistic simulations of flame propagation based on experimental data for flame shapes and burning velocities.
The most complex type of deflagration model is one in which chemical kinetics is used to model combustion rates and flame dynamics effects tre calculated by solving the Navier-Stokes formulations of the conservation cquations.
The empirical models have been primarily used for containment accident scenaario calculations and
, arameter sensitivity studies.
The p
phenomenological models have been primarily used for pre-test and post-test simulations of experimental results.
The detailed hydrodynamics and chemistry codels have been primarily limited to separate effects simuletions such as turbulent flame propagation around obstacles..
l e
w
---t
-w 9
.m 3-iy--
---9
-=y wy
-m
-r
--+-.-------..-------------.i.-iw-
,w-------t-'-=-F-ww----~,-um-m--sr
-.i-- g m-
-7ry
- - =
,a w
s White Paper
~
R. Zalosh CONDITIONS LEADING TO AND RESULTING FROM DETONATIONS p.
~
v.
A.
SUBISSUE STATD4ENT The detonation subissue covers the following questions.
What combinat-ions of hydrogen-air-steam mixture composition, mixture volume, containment configuration, and ignition s~ource characteristics are required ' to produce detonatiot.s?
Can detonations occur when only a small portion of the contain-ment volume contains a nominally detonable gas mixture? What is the transient
' detonative loading of the containment structure in these postulated deto-nations?
The question of structural response is also relevant to the detona-tion subissue but will not be addressed in this summary of the combustion aspects of detonations.
B.
10COR UNDERSTANDING AND BASIS Detonation initiation criteria can be divided into two categories.
The first category is direct initiation of detonations, and the second category is deflagration-to-detonation, transition in which flame propagation velocities escalate from subsonic to supersonic levels.
Recent fundamental studies of direct ini tiation of detonation have produced definitive initiation criteria on ignition source strength mixture composition, and containment configuration requirements.
These studies have shown that the most easily detonable composition (approximately 30 vo1%
hydrogen in dry air) requires an ignition energy of 4100 joules for direct initiation of a spherically expanding Chapman-Jouguet detonation wave.
This required ignition energy increase rapidly as hydrogen concentrations decrease from this worst-case value (e.g. 200 megajoules for 15 vol hydrogen), or as steam is added to the gas mixture.
The minimum detonation initiation energy is eight orders-of-magnitude larger than that required to initiate deflagra-tion combustion, and is at least several orders-of-magnitude greater than the energies associated with typical accidental ignition sources such as electri-
cal sparks.
Thus, directly initiated detonations,can be eliminated as a plausible combustion mode in containment burn scenarios.
The data base for deflagration-to-detonation transition is far less def-initive.
Mpt of the tests in which transition to detonation was observed involved pipe-like or duct-like vessel geometries with length / diameter ratios greater than 30:1 or with specially designed flame acceleration obstacles.
Elongated enclosures do exist in many containment buildings in the, form of ventilation ducting and (to a lesser extent) enclosed stairways.
However, it is difficult to visualize hydrogen concentrations reaching the detonable threshold (15-20 vol: depending on enclosure length and cross-section) in these enclosures; especially in containments with deliberate ignition systems.
If transi tion-to-detonation is hypothesized to occur in one of those elongated enclosures, there are data and analyses to determine whether the detonation wave can propagate from the enclosure into the bulk volume of the containment.
These data correlations and analyses involve measurements and calculations of detonation cell sizes a'nd the number of detonation cells re-quired for detonation propagation from the elongated enclosure into the bulk volume.
Results indicate, for example, that a 15 vol: hydrogen-air mixture would require detona tion propagation from a duct with a characteristic diameter of at least one meter.
Although detonations are highly unlikely in postulated containment burns.
transient pressure loadings from hypothesized detonations can and have been calculated using computer codes with two-dimensional Eulerian or Lagrangian formulations of the governing equations.
Calculated impulses (areas under the pressure-versus-time curves) are quite sensitive to the details of the con-i tainment and gas mixture configurations since they involve complicated shock and detonation wave interactions and reflections off containment walls.
One important limitation of these computer codes is that they cannot calculate deflagra tion-to-de tonation transitions unless they are imposed by the code user.
=
h
,.,,-..-_.,.,-,.----,,,.-.,..--,-.,,,..._,--.,,..,,,...,.-,,_.,,,.,,.,n-
-.,,.,__.-,.,,,_.-._,,,.-_c,
..en- - -,, -, _,,, - - -.,,.,
... _ ~ -.
I C.
DIFFERENCES FROM NRC VIEWS Major differences are in the conditions from transition to detenation and in applications plant-specific geometries, in particular the condenser plants.
D.
RESOLUff0N 0F DIFFERENCES Al though the deflagration-to-detonation transition question is not as neatly resolved as the direct initiation question,.the preponderance of evi-dence indicates that detonations are virtually inconceivable in containments with effective deliberate. ignition systems.
The plausibility of deflagration-to-detonation transiton in accidental ignition scenarios depends upon the hypothesized gas mixture composition and the details of the containment con-figuration, including the presence of any elongated enclosures.
Since it is not feasible to conduct a large-scale test program to determine possible deflagration-to-detonation transitions in these containment configurations, it would be more appropriate that the remaining questions in the detonation sub-issue be tackled with conservative computer code simulation of hypothesized detonations.
These simulations would first require agreement on plausible gas mixture compositons and a survey of 'large-dry containment features promoting deflagration-to-detonation transition.
The detonation computer code calcu-lated transient pressure loadings would then have to be input into dynamic structural response codes to detarmine the resulting threat to containment integrity.
e AO O
s White Paper G.R. Thomas LWR Core Heatup Phenomena p.
~
u..
A.
Issue Statement The core undercooling and heatup process occurring in the initial uncovering of a LWR core is a precursor to core melting and geometry disruption.
The heatup pro-cess results in production of the hydrogen and early release of volatile fission products.
Phenomenological models are required to describe the state of the LWR core during this heatup process.
These models must recognize PWRs and BWRs separ-ately and calculate the rate and amount of hydrogen production and the thermal-hydraulic conditions affecting early fission product release and transport from the core.
Additionally, these models should provide an estimate of heat losses from the core and the resulting impact on surrounding reactor vessel internal components such as the PWR core barrel and upper plenum assembly.
They should also provide a first order estimate of the ti;ne when core debris may become a potential challenge to the integrity of the lower core support structure, and the reactor pressure vessel (RPV).
B.
IDCOR Understanding and Basis Core overheating caused by a sustained undercooling condition, such as for a loss-of-coolant ~ accident (LOCA) with no safety' injection, typically can be divided into three major phases:
1)
Early heatup phase:
the core is partially uncovered, but overheated conditions may cause Zircaloy cladding failures.
2)
Mid-range heatup phase:
the partially uncovered core has locally accelerated Zircaloy oxidation, with its attendant energy and hydrogen production.
During this heatup range the local oxygen and hydrogen uptake may become greater than that necessary for Zircaloy embrittlement.
1 1
3),
Late heatup phase: core heatup has continued until regions are liquified or molten, and major core slumping is proceeding.
The early and mid-range core heatup phases are relatively prolonged in time ana
~
hesielly involve inadequate heat removal from the core.
The decay heat. generated fn. the uncoverdd core regions above the two-phase level
- raises temperatures in those regions.
As the boil-off proceeds more of the core is exposed.
The upper part of the core reaches temperatures at which the fuel rod Zircaloy cladding coul'd' fati (~1000 K) and the exothermic Zircaloy reaction with the steam becomes a signtffcant energy source (~1300 K).
Zi rcaloy oxidation rate is temperature dependent and can reach high-values-e.g., at ~1800 K, local Zircaloy oxidation may generate 10% of local energy at fQll power.
This reaction also generates hydrogen fn proportion to the oxidation rate.
In the increasingly limited steam supply of a progressively degrading core, the hydrogen generated can substantially dilute the-steam and may affect local oxidation reactions.
In practice, it was found that hydrogen ' blanketing' may affect spatial distribution of local peak core tem-peratures, however, the total hydroger, production may change slightly.
There is also an apparent upper temperature limit for Zircaloy oxidation of 2400-2500K, based on experiments, beyond which the oxidation process is substantially re-duced.
This condition is referred to as the Zircaloy oxidation cutoff tempera-ture If an adequate flow of cooling water is not restored, core temperatures continue to increase until the core progressively melts (late heatup phase).
The PWR and BWR core heatup codes developed for IDCOR** are designed to predict the above behavior for both MR or BWR cores with assumed intact core
" Two-phase level" is the local. bailed-up or dryout level below which tne core is j
essentially at or below saturation ' temperature and above which the core is at higher temperatures.
"Two sets of core heatup ccies have been developed for IDCOR:
I)'
stand-alone, highly-nodalized PWR and BWR versions, requiring tabular definition of primary system variables such as RPV coolant in-flow, i
temperature, and pressure, developed by EPRI, with Jaycor (PWR), and with S. Levy. Inc. (BWR),
i i
2) less nodalized (f aster running) PWR and BWR versions incorporated in
~l the IDCOR Modular Accident Analysis Program (MAAP), developed by I
Fauske and Associates, Inc.
i 2
geometries.
The heatup modeling is based on core and RPV thermal hydrauli.cs, core nuclear or decay power, and the oxidizing interaction of core materials such as Zircaloy with steam, resulting in the generation of chemical energy and hydro-s_ __ :
gan.
The codes also include a model for the retardation of Zircaloy oxidation caused by the increasing presence of the hydrogen.
~
K.
The less nodalized PWR and BWR codes incorporated in the MAAP code can optionally account for blockages encountered in the late heatup phase.
Based on studies with this option plus the concept that an intact core geometry promotes more rapid oxi-dation, the intact geometry assumption should provide conservative results for hydrogen production and core' temperatures.
The main outputs from the PWR and BWR core heatup codes are:
(1) the rate and total quantity of hydrogen generated; (ii) the thermal hydraulic and thermochemi-cal input required to predict the release rate and the chemical and physical form of volatile fission products and vaporized core materials at the top core bound-ary; and (iii) the spatial extent of the initial core melt-zone within the orig-inal core boundary.
The generated hydrogen can challenge containment integrity and the volatile fission products released from the core are a threat to public safety.
If the flow of coolant water is restored, the codes can predict the quenching of an overheated core, as either a,one-dimensional bottom-flood quench or a top-flood quench.
The latter results from upper head injection in selected PWRs and upper plenum core spray in BWRs.
Virtually all of the technology egloyed in the modeling of the early and mid-range phases of LWR core heatup and degradation is based on standard treatment of 4
heat transfer and high-tegerature oxidation occuring in a relatively stable geom-etry.
The significant processes are conductive and boiling heat transfer, gaseous convective and radiative heat transfer, and oxidation kinetics.
The data employed for the oxidation kinetics is that of Cathcart-Pawel for temperature below 1850K and that of Baker-Just for temperatures above 1850K.
l
~
3
C.
Differences f rom NRC Views The models employed in the IDCOR codes are quite similar to those employed in the BOIL subroutine of the MARCH code and basically there should.not be any signifi-cant differences between the predictions of the PWR heatup code and the MARCH code.
The BWR heatup code employs substantially more detailed geo. metric and thermal-hydraakic'models than those for a BWR in the B0ll code.
e e
4