ML20211N263
| ML20211N263 | |
| Person / Time | |
|---|---|
| Issue date: | 02/10/1987 |
| From: | Speis T Office of Nuclear Reactor Regulation |
| To: | Harold Denton Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML20209C630 | List: |
| References | |
| FOIA-87-10, RTR-NUREG-1150 NUDOCS 8703020029 | |
| Download: ML20211N263 (7) | |
Text
-
e e
n
[o
- r %,%, UNITED STATES l' ._ '% NUCLEAR REGULATORY COMMISSION 2 . --
i e,as e. ore s o e resss M, /, L ,
j.
a.g wyj ; . -
.(..
..... ~
N I 0 007 c Ps ea,+4 h;.c .-
PEYP A C JP FCE: Farold R. Denton, Director /
Of fice cf fiuclear Reactor Regulation ~ ~ ' ~ ' '
FEC+': Theris P. Sreis, Director '^*
Divisien of Safety Peview & Oversight Of fice of Nuclear Reactor Pegulation St??ECT: !PrACT CF N'JFEG-1150 RESULis rer year recuest of Jaruary 30, 1437, this reccrandur crevides cur arswers tc the questiens yev raised cercerning the impect of the ht; REG-l!50 results er GE restersitilities.
- 1. Are the ccre da age frecercy esticates in NU:EG ll!O reascrable?
ite authers of h)EG-ll!C ha c identified scre liritations of the study:
- a. The ccetrol circuitry was not analp ed in detail,
- b. The U EG !!!C aan1/ ses rely more upon gaaeric data thar de ful'-
scope EPAs.
- c. TFe assess'ert of ocerster arrors was si. plified, except fer Peach Sette , whera a c0*;rehersive, detailad hean rellatility aralysis was fer# cered for c;'ereter actio'is durirg an A?n$ scenacie,
- d. !rters.,ste- de;erdecies were explicitly redelled in tre event trees ard 'ault trees. Scarches were rade for selected subtle interacticrs identified by rast FRLs ard by plant specific experierce. Hcwever, tre cuantification for ccmon cause everts was hiably acreexirate ard generic.
The core darage frecuercy estirates in W CEG-!!!0 reasc' ably rarre-sert the rarge en acted. Differences between plant CCFs are due te dif fererces in the calculations (assurptiens, data base, details, brealedge of plant) of well as actual dif'ereaces between the clarts.
Tre licitations listed centribute sera ir;recisions to the numerical results. Thus, tre '.;5EG l!50 core darega frequercies s>culd not te interpreted as esac!. w cre discus 11cn is provided in tre arswer to questien 5.
- 2. Are N;;EG !!!O piart desiars and certilrrents rerresentative ef tre rafer classes of plants?
F rca' the f erscactive c' chor$1rg a gerefal set of plants with varyirg design ard contair ent types, the plants Ct0sen Jre certainly appropriate
'cr esa*inir Pl arts with rajor and distf rCtive dif ferences. Iuture plies fer NL}EG !!!0 f eclude assess crts of CC and eu olants ard !! Par 6 ccriairear,t wrica .111 ag ert the set n'th ad?t ticnal das f gr v ariatiers.
- ~m g ,, , p 3 /-/0 m a m, l 'lx A , M/.)7 u - .
~
t o
l Harold R. Centen 2 However, no set cf plants can be chosen to be representative of broad classes of designs or types, especially in the assessrent of vulnera-bilities with regard to core darage frequency. The features of plant design and operation that dominate core damage frequency estimates tend I to vary greatly even within a fairly narrowly defined class of plants.
These features irclude balance-of-plant design (e.g., support systets),
operating and erercency procedures, training, maintenance and testing.
As the authors of NUREG-1150 have clearly stated, core rnelt frecuency esticates canret be directly extrapolated frcrn one plant to arether.
Also, reviews have shewn that P:A applicatices are not yet sufficiertly unifore to allcw corparisen among PRA studies.
Any overall perspective on the core damage frequency risk of nuclear pcwer plants shculd include a cortparisen of the NUPEG-IISO results with these frcr the irdustry scresored FRAs which have been subritted and reviewedbyhy/RYto1x10(SeeTablelj3 rarge (4 x 10" /RY) dueResults of those to differences in PRAs vary over a wide plant features :
and P:A rethodology. In additien, it is irrrertant to rote that for sore of these plants, tredifications rade during the course of the FRA cr as a result of prelirinary firdirgs are already ircorporated in the core trelt frecuency esticate.
f In surrary, there is n.o scund basis for estrapolatirs the NU:EG-1150 I
results to otter plants def f red to be within the narre class. Further, it should ret be inferred from the disparity of the BWR and FWP results displayed in NU:EG-!!!O that 8,ios have a lower protability of core trelt than Fn:5.
$1-flarly for certlinreat tehavior, the exact failure redes and causes, I
ard their relative likelihoods, vary considerably amorg plants. The cuartitathe infor atier on the type of contairrent fattures and their relative likelihood and irrortance that will be previded in NUSE3 !!50 for the reference plants shculd rot be interpreted as applicable to all plarts withfr a Certaineert Class.
. 3. Oces NUsEG ll!O present an accropriate picture of containment failure prctatility in severe accidents?
0u0 to the strong deperdence of containrent perforrance on processes tnat still suf fer from uncertainties and the limitation of the anplytical rodels to accurately predict centainrent failure pressure ard timing, we cannot disagree with the approach and results of NUPEG-!!!0. Although we cannot verify tha actual nurerical esticates ard uncertainty ranges in NU:EG il50. we believe that the centairrent analysis reflects, in a gereral serse, the current state of browledge of accident phercrenology and scurce tere behavice. The large scurces of urcertainty which esist can only te elimirated by further research, evaluation, ard possible fines (for eaa ple, depressuritation of the prirary $/ stem could eliminof e the cencern about direct centairtant heatirg).
i.
l i
l Harold R. Den *cn a 3 l
Eecause of the large rarges of uncertainty ascribed to the ccrditional ,
pretatility of early contairment failure, all of the plant types appear i
to be equivelent in this respect. This apparent equivalence is not rearirgful. The irrortant cortribution of NUREG-ll!0 is to identify
, the dominant phencrena and failure ecdes for each plant.
4 Few have the perceptions of containrent performance changed since Wa$H-14CO?
Previously the Reactor Safety Study (R$$) fcund that the possibility of avoidirg centairment failure in severe accident was ouite small. The most likely dry-containment failure modes identified in R$$ were (1)baseret relt through which is characterized by low atnespheric releases and (2) an early failure by overpressure resulting from various ner condensitie gases released within the containment. Another failure rode was due to a large steam esplosion that cculd rupture the RPV and containrent. However, the R$$ considered this failure rede to ce irgrcbable. The R$$ concluded that if an ex-vessel steam esp 1csien were to occur, it would not rupture a large contairrent.
1 Fcr !W: ccetainrents, the R$$ found that the most likely containeent failure redes are ex vessel stear emplotten and overpressure fatture resultirg frer various ncn condensible gases released within the containrent.
Recently. tre NUDEG-l!$0 sucrerting analyses have provided evidence that additional certainrent failure redes are possible. For era ple, large dry centairrents right fall due to direct centaincent heating and Mark I and ctrer free sttrdirg steel coetairrents right fail due to direct certact of tre steel certainment stell with the core detris. These containrent failure r des were not idertified in tre R$$. Since they rignt cause an early failure, rassive releases of radioactive raterials to the atmosphere cculd result. While there are sore changes in our perception of how and why ccrtairments fail, the ran;e of pretablitties for early failure rode is sti11 very large and erco casses the R$$ values. The greatest differerces fecc tre RS$ arise from cur understanding of severe accident phenorereicgy ard reccgnitien of ptercrenalogical uncertainties that were beycnd the a
j carattlity of the R$$ state of the art. Accordingfy, the present state of kncwledge previde no assurance that the likelihood of containrent failure is significantly Icwee than in the R$$.
- 5. New dcas NUREG ll!O compare with our previous cerceptions of risk?
The NUREG ll!O estimates of relic health risk are displayed in teres of total early ard latent fatalities and individual early ard latent fatalttles fcr all five plants. Cerparisoes between the NUR[G ll$0 risk L calculation and those in the R$$ reveals that the R$$ values lie in the i upper certien of the AUR[G ll!O risk rarge. This appears to be driver by tre Icaer severe core damage frequency esticate, the range ef early centainrent failure IOelihocd and the use of revised health effect medels in Ml REG 11!0 results. The latent fatality risk estimate of Milestcre 3 PRA is comparable to the NV:fG 1150 estfrated values for ;
Surry.
I
)
l l
- Pa-old A. Certen 4 j The display of individual early fatalities, with a corparison to the current sa'ety goal is sh(wn in figu e 1. This display gives tFe ferressier that Fesch Betten is approxicately an order of ragnitude safer than all other plants in the study. This result differs from our perceptiers that EWR Mark ~1 plants are accidents tran large dry containrents (generally more vulnerable to severe Zion).
4 The differences in NUREG 1150 results are dependent upon the Icw estimate of core damage frequency for Peach Botter and the emptasis on I pFencrenological uncertainties for the large dry containrant performance.
The Peach Ectton core relt frequency erbodied in these risk estimates is about 40 times lower than that for lien. The indivicual risk estimates are rearly proportional to core relt frequency. As discussed in cuestion 2 ateve, it is ret clear that the core relt frequency dif fererce betaten Peach Ectter and Zion represents a real difference bet een these plant types.
Fre<ious estessments of early containment failure pretability have been ,
retsurably Icwer for large dry containments than for Park 1 EW:s. The
- Tasis of htEEG l!!O cr bourding values of early failure probability due te direct certaineert Featirg has resulted in relatively ecual rargas of *arly f ailure pretabilities for large dry subatnspheric ard Park I ccatainreats.
P*cerdlass of the Cortaire t failure probability ard failure rode, the i fir 51 facter that effects the urcertainties en risk perception is the l FMellirs of fissicn product tahavior followirg early contairrent failure fcr the various plants. TPe ragritudes of tFele fission predwCt i re'*tses, ard the way in which they are assured to te reduced by
- su::eef s'en pcols, sprays, ard secondary buildings, right have an l 1P;;rtart ir set on these risk cceparisors. Our bre ledge of tcw fissien i pr: duct tehavior was redelled in NUREG ll50 is currently not sufficient j to assess its impact en the risk estimates.
1
, 6. R at V C actiens do the ht:EG ll!0 result suggest.
UEG ll!0 reaf fires the basic finding of the reacter safety study, that !
3 the rish associated with operation of nuclear pcwer platns is relathely trall. Thus, coetirued operation of nuclear pcaer plants is ret in Questier.
F:aever, more reliable, safer operation of these plants is possible.
M8EG !!!0 setts to irdicate that there is ro rajor fin that could drastically tr:rc<e the performance of existirg plants in case of severe accidents. At tP* sare tire aralyses ef the refererce plants provide insights for each plant ccrsidered with respect to irportart contributers to severe accidents.
TFe retort idertifies the rain causes of core daraga as well as the ressers leading to large radioacthe raterial releases. Tre risk reduction part of the repert discusses potential irgrova" Pets. Sore of these irgrove-ents crald to arplicible to f* ore tPan tre plant. Cortination of potential fines 1
a
l l
)
i Harold R. Centon 5
]
identified for a given plant could result in significant improvements in terms of accident prevention and accident mitigation. NPC should implement the Commissions Severe Accident Policy which trandates examination of individual plants and implementation of cost effective improvements.
The report also indicates that some of the most cbvious fixes like flooding k the core debris or protecting the containmer.t wall from core debris can not D be ace;/ clished easily due to restrictions inherent in the current designs.
Ihe da .ns of these plants were not required to consider severe accidents.
/)
/
Changir4 the design new is either not possible or is very expensive. NRC l should develop and publish severe accident requirements for future plants, now. These requirements will help the designers of ne.< plants to account for j
severe accident considerations in the design state, thus avoid scre of the I difficulties the existing plants are facing.
Pelative to our understarding of physical phencrena involved in the assess.
rent of severe accidents and severe accident consequences NUREG.1150 indicates a need for additienal research activities in selected areJs, like Core melt
- progression and direct heating of containments, twp(G.ll50 also provides reans for a systematic review of severe accident related research needs including identification of activities that have the best potential to tepreve the safety of existing ard future plants. Improvctents in aralytical tools could also be significant in specific areas Itke the effect of seccndary centainments en release to the envircreent. hDC shculd use the inferration new available to revise and reprioritize the severe accident researr.h program, i finilly, ore rust note that htFEG.1150 cersidered only internal events. ,
ERAS typically predict that 20 to 50 percent of the core darage frequency '
i is due to esterral events. Tre contribution could be even higher for plants which have a low core damage frequency due to internal events, fer esample Feach Botton, h3C should proceed expeditiously with assessment of enternal events for the reference plant, thus complete the plants rish profile. It would be helpful if the treatunt of exterral events
. =culd be perforced in a censistent rnanner with the internal event
- aralyses remitting reaningful allocation of risk between intenrat and esternal events.
A, f (C h M v Themis P. $pois, Director Olvision of Safety Review & Oversight Office of fluclear Feacter Regulatten j CC' R. Eerrero
- f. liiraglia I . 'h v a k l
l
_ _ . . - - --. =.--_-_ = . - - - . - . . -_-_. -. _ .-__
- b TABLE 1 t
RESULTS ON CORE MELT FRE0VENCY t
ESTIMATED CORE MELT FREQUEN0Y AS REPORTED IN PRA (THESE val.UES 00 REVIEW RESULTS PLANT NOT INCLUDE PLANT EST!KATE0 CORE NAME TYPE SPCNSOR MOO!FICATIONS) COMMENTS MELT FREQUEN0Y IN0!AN POINT PR INOUSTRY s5 x 10 i/RY Internal and s4 x 10 4/RY 2 External Events (NUR!G/CR 2934) -
!NDIAN POINT PWR IN0U$TRY s2 x 10 */RY Internal and s4 x 10 */Rf 3 External Events (NUREG/CR 2934)
LINERICK 1 EWR INOVSTRY s5 x 10 8/RY Internal and s1 x 10 */RY t External Events (NUREG 1063)
GE$$AR SWR IN0USTRY $4 x 10 8/RY Internal and NA (Standard External Events i Casign)
HILLSTON! 3 PWR INOUSTRY s7 x 10 5/RY Internal and s2 x 10 */RY External Events (NUREG/CR4142)
SHOREHAM lh0V$ FRY s1 x 10 8/RY Internal and GWR s2 x 10 */RY Enternal Events (NU8!G/CR4050) ;
M10LANO FvR ISOUSTRY s3 x 10 4/Rf Internal and Not ReRu n tified External Events .
SEA 3 ROCK PR ISOU$TRY $2 x 10 */RY Internal and N t Requsntified titernal Events iAMEE 40wE Fwa ISOUSTRY s2 x 10 8/RY Internal (vents Review On;oing l 000 NEE 3 PWR INOUSTRY s) x 10 4/RY' Internal and s2 x 10**/RY !
Enternal Events (NUR!G/CR 4374)
BIG 200K BWR Ih0USTRY *1 x 10 3/RY Internal and PO!NT Not Retuantifled Etteenal Events 2!0N Fwa !N,UST8Y 7 x 10 8/RY Internal and s2 x 10 */Rf ;
External Events (Staff Revice &
NJRE0/0R 3300)
MILLSTC'i! 1 6WR !s0VSTRY se x 10 4/RY Internal (vents Not ReRVintif f e:
(!!AP)
HA00AM NECK Na th0U57RY s$ = 10 */RY !nternal Events *$ s 10 */RY
86/3080)
"This value J:es include so e design and operational esdifications r ade durin3 the course of tt,e P;4,
T
~'
> e== =~ . . - -.
I e 1
/ I
.c.
9 4$
d o
+ a l"iv i-HHH D C' Hf S H-}ti&H!HF .,
a 9
- ,. e*
. g ed O
C
+t- +Ht ' +h'rGt HI "O M F C- M it -
%d 0 9 8 4#
A S
g h
-t- tri F,".ir.'itl @ W 8 c
k
+
" % e y b 8 cc Q 8 > c
- u
+ + Ginistif f fH2inttMfH% S
~
t O.
.M
.% =
.n (J %
9
~
c Y *t* itI-ithI t I*.M ILD UihNt-tt-i Uo 2:
Q e-
$ gM- - -
(.
- f * *- fh if ff ikkf*h *f f L) h h f fNh
- i* it" h .s m h 0 h, b
y .
E ~><. - luu + - h 1u.t z ..laa:_; u..
o e- e e o I i e -
ul Q) tul
- U.t ,
et 5
0 e 4
A_