ML110030901
| ML110030901 | |
| Person / Time | |
|---|---|
| Site: | Pilgrim |
| Issue date: | 01/03/2011 |
| From: | Jamali K Elsevier, US Dept of Energy, Office of Nuclear Energy |
| To: | Atomic Safety and Licensing Board Panel |
| SECY RAS | |
| Shared Package | |
| ML110030897 | List: |
| References | |
| RAS 19369, 50-293-LR, ASLBP 06-848-02-LR, PWA 00014 | |
| Download: ML110030901 (7) | |
Text
PWA00014 Pilgrim LR Proceeding SO-293-LR, 06-848-02-LR Kamiar Jamaili - PRA's Subject to Uncertainties Co n tent!l-I i'St -g,.ava i I BIble.at Sde.n ceDI ~
ReHabil ity Enginee,.-ng and System Safety j.o u r n a l h o m.e p.ag.e : 'W'Ww.e l:S<evle r.oo m / locat e/ res.s.
Use of r isk measures in design and licensing of fu ure r eactors Kamiar Jamali Unit.a:::l Srars nq,aRmrnt of~. Offor ofllucmr.f.nagy, Jooo J'~
.~ I!; Wammg;DD. DC 2lJ.s.as.,jJ _~
A RT~
C L
E: IN F O ABS T R AC T
....rDd!!- h imlry:
~
1 "
lFcb~T)'
2() 1 ()
~iniJ'eV:isedfonn
- 5 April 2I[J 1U
~ 6 Apri l 2lJ l ()
Av.tah~ m in.c 110 Ap ril :2il l ()
iN1Ud -t!D.T !fIe.KJtm" S.J.:R!tl' P'.rdI:abitistic: risk.z;s~t CPU
)
Sa.fcty ga~b J't.cc.cpt;mCc a.iltcria tNext :gener.artia11 nJdea:r pla:nt
~1 1 rmoduw ~B
!Pire::J~ CY...aJ :rI5 equ~
C1I:n.'II!!
lise o1' I rrt:o. m.ado n aBd j n:sigtn:s fro.m,p nnah lli:5l:k:: lis..l, ase:5f,rnems. ( PIi:A5.) 1111 B...c,~.ar f'eacror :§.'l fery
.i!lpp llcado n:s. h.as. been Ir.cre.a:s:.ing by tt.e nlide.a r lOOliS'tfy aDd! '!:he f1!~I,aU)f's... boI: h do:m.esdc.a Uy aDd!
Ifllter-n.at ~n.a _Uy. T his. ls:.i!I diesJI'.i!lbJe ueBd. a s. PR~ flaw demon:s:tr.ared. ca,p.ablli cy t o. I IDp mve s.a1e'ly aOO o,peratlOI\\aJ t1ex;lb ilJty beYDnd '[ hat provided 'I:hmt,tgtlt derenninlsa: k:.a,pp-m.ad~ aJo...e.. Bu t [I.e re c.a n be pt)'tef1i[.La I,pi rfa lis.. l 111e I im l tatloilS o t' fi s.k ass.es.s.mem tee nBO~ c.ail be m..t thi roughl ap,proadhes tha't A:!,ly l~v£J y 0.,0 q ua rlltl ta[ Ne PItA,re:5U lts (referred! to as r.ts.k,measures,In [ his.Ilaj)ef'). bec.a1Hoe o.'f t,Be LIIR.i!I nblgllous b tU,por.e,lu laJ ty rnis.le.ad ing,mess.ag:e 'Clu t caR be diet lw,red by n s;k;.ba:s.ed rmmbe rs.... T h is Is
,p.ard CliI.i!J Cly u-uefo'r fllrut'e reacro.~ w llle('e 1<RAs..are IiSedI d l1d ngt:t.e deSJg;R.a,OO I ice RS..lB8P roces.ses.. FO' tt.e:se.a,p.P-' =.a'[JOBS".It I:s. lmpor taB't 'to ensure t hr.at 'tile.aoru.a.l ~ de fa cto.. (If' even peroet\\red lISe o:f rt S..K Il'1le.1sures. ln '[I.e co n'tE~x:t o.'t'etciler reglll a'to~ ord:es.Jg;n.acce p't.anoe crtten.a Is..a
'ded.. W h ile '[Be iss-lN!S d iscussed I~ Ie ca n Ilave a sJgn ilicalu iIl'lillle nee on design ce rt i1icatlo.n o.r combined! license.a pp.licad o.BS.
f, flft1ire re.acto,ts.. 'they ca n also,BaW :SeCO.Bd!.a,ry impactS 0.,1} cur re nl1y o,J)ef'.i!lt Ing reactot"S...
1 ~ Intn xl:uct:ion Proba.b ilis t ic ri.sk.a.sse.s.s me:n t (PM ) results.and in.sig h ts have he lped ID improve nud e.a.r po.......er pla nt s a fe ty a n d o pe:ratio naJ n e:xib iJ ity f o r m ore t ha.n.30 yea.~. T hi.s su a:e:s.s has k!d,to i_n c m.a.sed u se o f P RAs by t h e n 'uc:le.a:r i nd ust:Jy a nd regulato ry a u t ho rit ie.s 1NO rldv.rid e.. W hj Ie t.h is t:r1!n d is large ly pcsi tive.. t h e re can be:
pot e n t ia l n eg.a:t ive conseq uences t ha 't ha ve n o t b een
\\<'Vid e ly d i.scus.sed in. re lated Ii terat:u re:. "Wi th :so m,e e::x£I!P'tio ns (e..g... ( 11)"
I t "Was becau se o f t hi.s positive (D J]t ributio n 1D safety that the US N:u c Je.ar Regu la 'ln ty Com m issio n ( NRC ) g radua lly re:fi ned the::i r o r ig inaJ determin_istic-ba.ser::)
J]ud eB,r
.sa..fety
~Jat:io[15 b y i n c o r po rat:i n g 'the u se o f ri.sk in fo rma.'t io n a nd in.sig h ts "Wi't hi n a ri.sk-info rrned fr.a.mev."CJ rk.
Ris k-informed regulation.s for the c urren t f.l.eet of op e-rating l ight -"Wa'lEr reacto rs ( LW Rs) are d e fi ned t hrou g h a
com b inatio n 0 f rule-m a king a nd pub Licat io n o f 10VW!r-t ie r docu ments, su c h as regu la 'lD ry g uid es o r NRC's e nd o rse::m,e n t o f certain nuclear ind us try docu m ents. Thus. in a ri.sk-i nJo rrned fra m ewor k., ri.sk info rm.ad on a n d insig bts su p ~ e m,e n t t h e trad i tio na l dete rm in is t ic a p proache.s a n d fo rm a part o f th e 0 ve:r.a.11.sa..fe:ty ca.se ("Wb.ic h i.s s o m eti m es refe rred ID as the safety ba.s is ) fo r a n 'u d e.ar pla nt. The Com mi.ssio n has aJ.so cal ~
fo r i.nc::re.a.seCI u se o f PRA tech no logy in a ll regu laID ry m..atte:~ in a m a n_n e r t ha.t comple ments NRC's p redo mina nt ly dete rministic:
a p p roach,e.s "Wi th.i n the (D nfin.e.s o f a ri.sk-i n_fo rmed as 0 pposed ID a
£.mail.adclress.: bnxiaT ~..un.aJj/J:I}J.q.doe.g CN
()
l..8:!20tt.. ~
fro m elT.l.litlE!' P:uh1 iish.ed by iI!I~
lA:d..
d eli: 1 U.l () 16/j.Jrcss.2lO l (u)4.001 l:oublls;t.edi by Elsevier LId..
rl.s k-ba.sec::'I regula'lD ry con.struct. Som e 0 f 't b e d is ti ngui.sh_in.g featu res bet:wee:n 'the: 'tv.'O are a l.so d i..scus.sed in 't bi.s pape;L T he: n uc:le.ar industry aJ.so has u.seCI PRA techniq u e.s extE:ns ive:ly "Wit b b en e:fi,d a l resul ts.. i ncl ud ing ion 'the desig n o f advanced o r evol u tio na ry nud ear re.actD~. These ben e fi ts a re:.. in part.. l1!.I.ated ID t h.e f.ar:t 't hat 'these s.arn.e u.se:rs can a l.so con trol a n d lim.it 'the:
infl u e nce o'f the in(D m p lete safety info rmat io n t ha:t i.s p rovided thorou g h the resu Its o f the: P RA a lon e.. Factors 't ha 't are us u a lly no't fully accounted for in a PRA m,ode:1 but are ge rma.ne to 'the:
(D nsid e ra tio n o f adequacy 0 f safety fe.a.tu res fo r a spe cific i.ss\\E or aa::i.d en t scenario may i n d :ude:: m.agn itude:s of re evant.sa..fety ma.rgi n.s. i n CDr poaration 0 f d efe n.se i.n doq::r t h. potent ia l for c o rrec-tive 0 r com pe;nsa'lDry attio n.s. degree o f c o n.se:rvati.s m in a na lysi.s.
a nd m.any o'the r s.. The very sam,e P RA in formatio n.. however. w h,e n used to (D mply "Wi t h -we.11-in'te:nt ion ed l'1!g ulato ry JX>l icie:s a n d a pproac hes can le.ad to som e undesirab le (D nseq u e nces.. Som e o'f the u ndesi_rab Ie (D nsequ e n CE.5 in a pp Ucatio n.s i_nvolv i.ng fu ture re.actD r'S a re a lso d i.scussed 'below ~
PRAs prov iCIe bot h q ua litative a n d q ua n'titative: informa tio n_
Recent: 'tre nd s in t he: de:ve kJ pme nt: o f n,eN ris k-rela ted a p pro ach es..
"Whe:th.er 't hey are per formed b y 'the: regu laID ry staff. nuclear ind ustty~ 0 r o 'the r dom estic o r i_n te:rna tio na l 'bod ies.. are tl:JWards heavie r e mph.a.si.s in use of q ua nt ita:t ive P RA res'ul ts ( in'te n:.h.ange-ably re.ferred 'to as '" r is k m easu res'" i.n t.hi.s paper ).. It i.s ""'e.U-know n tha t q ua n 'titat:ive results o f PRAs, in pa.rt:icul.ar~ a re: su 'bject 'to var io us types o f unce-:rt ain ties.. Examp le.s of 'these 'u nce:rta..i nde.s incl ude prob a b ilistic q u a n't ifi.c.a:tio n o f s ing le a lX1 (D mmon--
ca:u.se h.ar d "Ware o r software fai I u res.. o a:u rre nce o f ~i I]
phy sic.a.1 phe n o m.e:na.. hu m.a n e r ro rs o f 0 m _issio n a nd c o m _m.issio n"
magn itudes 0 f SOU r<l! terms. r.ul ionud ide release and transpert.
atmospheric dispersio n, biological elfucts o f rad iation, dose calcu lations. and many others. Unlike d<!l:erministicuna.rtain t ies related to physical phenomena (e.g.. neu tron ics:, t hermal-hydrau-li cs~ PRII una.rtaint ies aR! not R!.ad ily reducib le in most instances. Uncertainties associatm w ith physical phenomena can 0 FtP.n be redumd by tests:, experi ments:, 0 ""rati ng experience on actual or prototype designs. or improvements: in analytical models or IDmputat ional ca pab ilities. Despi te th.is well-known limitatio n. if quant itative PRA resu lts: are used in the IDntext o f risk.aa:eptan<l! criteria ( ce., w hen t h.ey are com pared agai nst a set of threshold val ues estab I ished by either the industry or the regu laID r). it wou Id be di Fficul t to IDunt er the unambiguous but potentially m.islead i ng or i nID rrocl: message that is de~ vered by such a number-based pro<l!55; i..e.., implying that a design is unaa:eptable or unsafe because it d id not meet a particu lar risk-based numerical t h reshold (labeled as a risk.aa:eptan<l! cri terion).
An i m pertant issue that is outs ide of the SID pe of th is paper.
b ut is wo rthy o f detailed discussions of its own, is t hat the introduction and i m(>act 0 f PRIIs: i nthe design and ~ cens:i ng stilges fo r a future reactor is by and large different from the way t hat risk-informed regulations have been a(>plim to existing reactor *.
Currently oper.lting R!iLClD rs h.ad a demrministically estab ~shed liamsing basis (w hich ind uded t he plant's safety basis) before plant-speci fico r generic risk information and insights: were made avai lable th rough PRIIs. The PRIIs: generally con firmed t hat the origi nal determ.i nistic ap proach to design and Iia.nsi ng was conservative (e.g.. plants couJd respend ID some accident scenarios in manners t hat were not cred ited in the determ.ini£tic analyses) and further identified changes that couJd improve plant design or oper.ltional safely. Meeti ng the deterministic req uire-ments meant t hat implementatio n of their atb!ndant provis ions embodied w it hin the IDnceptso f defense in dept h. safety margi ns.
conservative assum ptio ns and analyses. qual i Iy assurance, and numerous other factors ( many of w hich are not R!.ad ily measur-able w ithin a PRII model) created a safety cushion or marg in t hat proto::ted t hese plants from u ncertainties. including those from
" unknow n unk noWIlS" (fo r w hich a euphemism can be "emerging safety issu es~ as d iscuss:ed in Section 1 ~ On the other hand, P RA models h.ave to R!ly on realistic inputs: to ellSure t hat risk significant insights are not obscu red by arti ficiaUy biased resul ts:
derill'ed from the ap p ~ ca ti o n 0 f uneven ID llServatism5. Themfore, great caR! must be exercised in bringing PRAs into t he design p racess to ensu re t hat t he fu ndamental pi Uars of determ.i ni£tic safety as:su rance process ment ioned above are not u nd u Iy com pro mised. Thus, fo r futu re reactors. 'Use 0 f risk in fo rroation can have a far more sign i ficant im (>iLct on the safety basis of the plant, ind uding the petential ID drive some key design decisio ns.
The int ent of risk-in formed R!gu lations is to ensure t hei r influence is pas itive in safety t radeoff do::isions.
N:RC publ ished t he Safely Goals Pol icy Staternen to n Aug ust B, 1986 [l ~ W hile t he tex t o f t his I'll I icy statl!!Tlent does use the p hlaSe "aca.pt.m Ie risk," the tit le and the rest of the d iscuss io ns were careful to avo id the use of th.e Q,Janti tati,,,, Health Objectives (Q:HOs) o f prompt fatalities ( pr s) and latent cancer fatalities { [Lrs) as R!gu latory risk--.aa::eptan<l! criteria. In o th.er wo ms. the selection of the tl'rm.i no logy of " safely goals-was very delibera1E. An impertant attribu te of t he calcu latio n of plant-specific P Fs and las for IDm paroon w ith the dual QHOs: is t hat both are by ne<l!55ity " integrar" quantit ies t hat are deri,"ed from the ID nt ribu tio ns of all aa:ident scenarios t hat are considered in the plant-specific PRA model.
The Com m i£sion '5 1 005 P RA I'IlI icy Statement 0 n use 0 f PRII metbads in nudear regu laIDry activities [31. w hich was issued in t he aFtP.rmath of the ID mpletio n o f PRIIs for all o""rating nudear plan ts in aa:ordan<J! w i th the Individual Plant Exami nat ions Generic letter l41 states. in (>art:
The use o fPRII technology.ba uld be increased in aU R!gu latory matter. ID the exb!n t sup perl:l!d by t he state-af -the-art in PRII methods and data and in a manner th.at complements: the N:RC's determi n istic ap proach and su!>p<lrts t he N:RC.
tr.ld i tio nal defense-i n-<lepth p hiJosop by.
The Cam mission's safety goals for nuclear power plants and subsidiary numerical obj ectives are to be used w ith appropriate consideration o f uncertain*ties in maki ng regu latory judgments: an the nem for proposing and back-fit ti.ng new generic requj remen~s 0 n n udear pav.",r plant lia.nsees.
Th.e Commission a(>p rm"ed th.e staff's W hite Paper on Risk-rnfo rmed and Performance-Based Regu Iation in March 1900 [51, w hjch provided defini tions of risk-in.formed and risk-based regu Latio ns. It R!it erates t hat the Commissio n does not endorse an a(>p roach t hat is risk-basm, w herein decisio n-making is solely based an the numerical resul ts o f a risk assessment.
Regulatory Guide 1.174 16 1 est;Ib lished the framework for risk-in fo rmed regu lations in applications r<gam ing making plant-specific changes to t he I i<J!osi ng basis. Its approach ensures that numerical PRA resul ts would not form the sole basis for maki ng nudear.afety do::i£ions by Iisti ng five key p ri nc ip les (Le., meeti ng cu rR!l1t regu lations lw hich are primarily deterministiq.
m.eeting defense-i n"4lepth p ri nci pies, main tai ni ng. uffident safely margi n, keepi ng increases in risk. malJ, and performan<l!
m.o nitored) that have ID be met for a risk-i nformed ap proach_
Clearly, curren t regu lations are by and large based 0 n detenn i-njstic req u irements. A key pertion of the section an SID pe (Section 1.4) states:
The N:RC has chosen a mare restrictive pol icy that wou Id permjt o nly small increases in risk, and then only when it is reasanab Iy assu red. among other thj ngs, t hatsuffid entdere:nse in depth and sufficient margi ns are mai ntai ned. Thjs pel.iq is adopted because 0 f uncertai nties and to acIDun t for the fact t hat safety issues continue ID emerge regardi.ng design, construction, and 0 perational matter. notwithstandi ng the maturi ly of t he nudear power industry. Th...... fac1D rs suggest t hat nuclear pcwer R!.aclD rs shou Id operate rou ti nely on Iy at a prudent marg i n above adequate protection. The safety goal subsidiary objectives are used as an examp le of such a prudent margi n.
The clause abou t ID nti nual emergen<l! of safely issues for pLants w ith many years of operating experien<J! is an altern.ative way to state the ID ncern regardi ng u na.rta int ies about the
" unknown unkn owns~ that are a mare significant con<J!rn for future R!.aclD r designs.
One R!iL5on th.at RI!gu latory Gu ide 1.174 h.as worked weU in app~catio n is that it was in tended for operati ng plan ts w it h a primarily determjnistic
~cens i ng basis already in place, w hjch means t hat the plants ",,,,re already deten:nined to be safe: before app lying the resul ts of plant-speci fic PRAs.
ri.nally, N'otel o f Chapter 19 o f the Standam Review Plan (SRP )
17 1 states that t he Q:H:Qwsu rregates 0 f OJ re Damage FR!Q uency (CDf) and large Release Frequency (lRF) are goals and not regu laID ry requ i rements.
Th.e key cond usio n from the: above is t hat the N:RC Commissioners have not endorsed a " risk-based-app rnach to regu latio n because of the u ncertainties in quantitative R!S ul ts: o f
P RAs.. Th.ese u na>rtainties are large for cu rrently 0 perati ng nudear plants, pa:rticu larly in the SCI~ lI ed l.e!J'eI 2 and level 3 PRlls. The fact that t he large una>rtaint ies in the estimal25 of pmbabilities for h ardwa~ failu~ and human erro[S. and understanding and p robabi I istic quanti fic<rtio n of occu rmnce of some physical p he:no mena in PRAs of cu rrently 0 perating reactors seem less so because of repeated reuse shou Id not be O!Ierlooke(L T realment of una>rtaint ies in severe acddent prog~ si on and delineatio n has always 'been 1 i m ited in ris k assessments performed m d<lte, even in t he stoo ies th<1l: went the fu rthest in such. analyses, such as N'UREG-1150 1 8 ~.
Another irn portant consideration, also related to t he general c<rll>gory of u ncertai nties, is t he issue of stab>-O f~the-art: in PRII methods and data. Th is is an issue for risk model i ng of all reactor designs as all uCled to above, and it is esped aUy so for designs t hat pri mari Iy rely on pa:ss ive safety fu nctions performed by safel¥-related Systems. Structures, and Com po nents (SSCs) and d igital systems (e.g.. in instru mentation and control I&<C). The cu r~nt stab>-Of-the;Lrt does not permit a high quality model ing for ~I iabil ity evaluations for these systems. rn particular. there is considerab Ie u nrertai nty w ith respect to t he m ntributio n of software m mmon~us e f ailures (CtJ') 1D digital system reU.a-b i I ity. For the potentially safer and mo ~ passive advanced reactor designs.. it is poss ib Ie that digital systems and human errors of commission (due in partto longer ti meconstan see. e.g.. I.1J1) might have a higher relative risk m nt ribu tio n, a m ntribubon t hat may be d i fficu It to assess w ith any s ig ni Iicant level 0 f con Eidence.
These issues 0 ffer addi tional reasons t o ap p Iy quantit<rtive PRII resul ts j udid ously fo r futu re nuclear plants..
TheComm ission also offered another goal of l E-6Jyr w ithin the Safl!!y Goals PcUqr S1>II:ement for frlX! uenqr of large m1mses to the eOYi 10 oment for fu rther staff exam in.ation.. A CIeEin ition for la'lle reIe.ase was not offered in that document 12~ fn 19l the staff consi~
several optio ns aoo Iinally r ommended that a la'lle relea.E 'be deli ned as a ~Iease th<1l: h.as the potential for causi ng an offsite early fataUty. ~ral other SKY papers (denotes pape-s subm ittul t o the Commissioners
'by the NRC staff}. Staff Requi~mts Memoranda (SRMs~ and Advisory Committee on Rea<:tor Sai>guarCIs (ACRS) letters to the Commission (e.g. llOl) we~
deKItul m this subj ect The Com mission directed the stalf m ensu~
that the r evaluation of la'lle mlease magn itOOe 'be m nsistent v.ith.
ACRS proposed guidel ines I in kirlg the hieraJt:hical lweI!; of the safety goal dbjertives, where the Ia'lle ~I ease gu iIlel.ine was m nsiClerul the t hi rd level objertive (the quaUtative and q uantitative health obj ectives
~
the level one aoo two dbj ectives ~ Amnrdi ng to these guideU nes.
each. subordinate I_
I o f the safl!!y !Jlal dbjocti\\eS should:
- be m nsistent w ith the level above,
- not 'be SCI conservative as t o Cmilte a de facID new pol.icy.
rep ~ent a simpUEicatio n o f the previous level
- prO!lide a basis fo r assuri ng th<rt the Safety Goal Pc Ucy Obj ectives a~ 'be:ing met,
- be delined m have broad generic appUcabil.ity.
- be st<1I:ed in terms t hat are understandable to t he publ ic. and
- generally m mply w it h. cu rrent PRII usage and practia..
fn 'the end. the staff reached the overall cond usion t hat develop ment of a large release deEi n:ition and magn itude, 'beyo nd a si m pie q ual i tative statan ent related to the frequency 0 f 1 E - 6Jyr is neither practical nor required fo r des ign or regu lam ry pu rposes.
fn aCJ(J ition, basul upon th.e wo rk done elaluabng la'lle releases in N'U REG*-l1 50 181 and other related activities. the staff notul that the general *performanCl! gu idel i ne of 1 E -6JyT and the fiF stibsid iary dbjective 0 f 1 E -4Jyr are not m ns is tent w it h. the o riginal QROs 1.111 ( i.e.. they a~ more conservative. and the degree of m nservatism de*pends on the specilic pl.ant ).
[n add ition, t he Commission rejected the use o f lE - 5Jy r of macID r operatio n as a CDP goal for advancul desigru in. SECY-9O~ 16 1.12 ~ and its SlUvt This rejection shou Id be exam i ned m geth.er w it h a series of Co mmiss ion Pc I icy Statemen*ts 0 n
~
ulabon 0 f advan.ced reacID [s. The last in the sedes pub I ished in OcID'ber of 2008 lUI stal25:
The Commission expttt:s. as a minimum, at least the same degree 0 f protection of the I!Ilvi ron ment aoo pub I ic healt h and safety and the m mmon defense and secu rity that is requi red for cu rrent generation lig ht-water reactors. l'urthermore, the Commissio n expects that advan.ced reactors w ill pmvidl!
en hanced margi ns 0 f safety and/or use si m p I iEied. inherent:,
passive. or otherinnO!lative means m acmmplish their safety and secu ri*ty functions. The in.m rpo ration of en.hanced safety margi IlS may hell' offset the effects 0 f added uncertai nties in t he PRII model and/Or in acddent analyses aris ing from the nOllei ty of advanced reac1Dr designs. l Elsewhere other attri-butes of advanced designs are described as: rei iable and less complBt: shutDow n heat remO!lal systems; longer t ime m n-stants and sufooent instrumentatio n; simpliEied safety sys-tems; minimize poten tial fo r severe acddents 'by inm rporating redu ndanqr. diversity. safety syst.em. i ndepende:nce; i nm !'pO-rate defen~in-depth ; e1L ~
The important aspects o f th is Poliqr Stall!ment a~ : (a) it m ntains o nly qualitativ'e but well~p mven prindples fo r en hanced safety of nuclear rea.ctor designs.. and (h ) itspecilically lacks any ds k~based numerical criteria. Because of large uncertaint ies of ds k-based nu merical resu Its. ris k analysts typ ically do not m ns iller vari<1l:ions of less than factors 0 f 10. or 50 in such nu m'bers as meani ngfu I increments. Ris k Bt: perts may convert the abO!le pol.iqr statement int o a m r~pond ing numeric.al crita:io n
'by prO!lid ing an order o f magnitooe as the smaliend iscdminalD r for decid ing how much safer advanced reactors sho uld 'be fro m current reactors. Th is.. however. is a non-sequitur and a prob lem inherent to risk.:tJased calculations. An order of magnitude is a very la'lle i ncmmen t in th.e real w orill. and cur rmt nucJear
~lD rs are already much safer than any other m mpa:rable industrial fad I it ies and hazaroous human activities. Ultra-m n-servatism in design has a price, both @conomically and operationally. As discussed in Sectio n 3. th.e proposed new surrogate numerical risk-based cri teria can be far more restdctive than the QHIOs. They are also quantitatively unprm ictable in
~real risk s pace" and not com parab Ie w ith QlROs as they are no n-i nmg ral measu res of risk. They are more restrictive in the sense t hat a rea.ctor that in a hypo thetical case may fail lD meet some 0 f the new criteda (described in Sectio n 3) can stil l meet the QflOs 'by orders of magn i tude.
fn s pite of the abO!le discussio ns and the b road pol iqr gu iClanCl!
'by th.e NRC Commissioners, this paper's observation is t hat th roughout many ptibliation.s of the natio nal and international
~ ulatory agend es and commercial entities, there is an i ncreasi ng trend mward more prevalent use of risk.based
~ ulatory concepts in general, and the 'use of some fo rm of nu merical risk t h rI!5 ho Ids as il.CD!ptanCl! c riteria vis-a-vis safety goals..
in particular. For example. a nurriber of NRC staff oocu ments (e.g.. 1.14.151). as well as industry aoo internatio n.al pubUcatio ns {e.g.. I.16-23n have emp loyed variou.s types of risk-acceptance criteria (consistent w ith the termi no logy em played w ith in the oocu ments) w h ich. i oyolve some form of a frequenqr versus consequenCl! ( FC) curve, or FC anchor poi n'ts or regions. It can 'be shown that these ap proaches generally establ ish much me re restrictive nu merical th resholds than th.e QiROs, and a~
applied as non-integral quanti ties. W hi le t he intentio ns 'behind th is trend are ndb Ie and mo tivated in part: flO m a desi re t o
. E+O 1 E-1 1E-2
~
~.
- 1 E-3 I
ill:'
£ - 1E-4
"" I" u..
1E-5 5() rom P'lte y em ~ _", '1llr8!lfb' Bill
!!tfocl!l (lftillla r",nllJll_eBrtjllolol1ll
.. "~
.... 'po ** I>,.,
'JO.I E-6 1E*7 1.......----------------!I..-
11001 O.(n 0.1 1.0 10.0 100.0 1000.0 Dose, rom
. ~ 1_ I'requenq> =. C<I11=fU= Gln-e
.
- 62 ) af lJRl!C. 1SO[L analytical models and many other faclDr:s (e.g.. impact of safe:y ma~
ns~ Even then, the use of representative parametl!rs (such as the mean) iI.55iOwb!rl w ith. t he frequencies and m n.sequenms of individual or integra12d aa::ident SlEnarios has I i rnita:ions of its own, as th.e types and w idths of the underlyi ng distributions of the input random variables are generally ass igned by subj ect:ive judgmmt rt is clear t hat these issues bem rTll! more dominant in analyses offutu rqal'lvanCl!!d reactor designs w ith less knowledge about several key aspects of the safety of t he des ig n. such as the fidel ity of analyses i.n thermal ~f1 uids, neutro nics, 'fission product t ransport matl!rial properties at high 'II';m peratures, component reliabilities. and the " unknown u nk nown.s.~
- . Th.e QHOs hae a logical relationship w it h the risk that the members of the pub I ic are otherwise ex posed to as articu lab!rl in the Qual itative hmJt h obj ectives. They establ ish the risks of nudear power plant 0 peratio ns at a sma ll fraction of t he ris ks that t he members of th.e pub lic. not the general publ ic at large, but t hose liv ing in the vid nity of tbe plant are al ready exposed to. A reduction in these ris ks for 'futu re reactors proposed by any stakeholder (w hich would be mnsi.sll';nt w ith th.e stall!!d Qualitative goal of t he Commissio n), should be within. reason and not so d rastic as to deprive the same population from the benefits that t hey may otherwise real ize from operation. of these reactors,
- . Plant~pI!lCi Fic PEs and leEs are calculated for m mpari.son against t he QHOs, Both of t hese, as well as the more widely used surroga'lI'; metrics tD QHOs.. such as OJF and lRF f or lWR ap pI.ications, are inll';gral Quant ities that are derived fro m the contributions of all acdoont scenarios that are m nsioored in the plant~specifi c risk moooL rnteglal risk measures inm rpo-rall! at least th ree i m po rtant properties:
L Defi.n:ition or characterization of individual aa:ioontsamar-ios is dependent on bo th the specific PRA modI:!! (e.g..
large fa'll It trel!!' small event tree versus smaU fault t rt!l!,l'large event tree) and the sped Fic plant design (e,g.. complex w ith more active safe;ty sy.sll';ms __ sus less c.omplex w ith. more passive safety sy.sll';ms~ IntEgra12d risk m easu ~ are not subj ect: to such depenoondes on the calcu lation mo&l or plant design.
o I'twiU be a chaUenge to estab lish critl!ria to ensu ~ t hat ind ividual aa:ident scenarios a~ defined o r character-i:md at the same level 0 f ~ reso.lu tion ~ across d i ffurent pla nt dlesigns and iI.55iOda12d P RA models for use w it h.
this type of FC cu rve constTucL The syst'em would be inherent ly unstable and dependent 0 n subj ective i nter-pretatio.ns by all sides in a d ispute.
2 Relative UIlcertain*ties decrease w h.en t he assodated ra~
dom. variables are summed, and t hey increase w hen the random variab les are mult iplied_ Therefore, the effects of u ncertain*ties are minimized when integrall';d risk measu res are used as opposed to w h.en i.n:II';Tmedia:te and product Quantities, such as frequencies and con.sequenms of individual acddent scenarios are used.
- 3. Com pariso n of any partial level o.f plant risk. such as t h.ose t hat are based on individual aa:ioont SlEnarios, agai nst SOrTll! Quantitative criteria can misinfo rm. or even. m islead.
The poten*tial fo r mis i nfo rmatio.n is large because it wou Id not be known as to w hat fraction ( is it 0.001% or Hal of the overall integ ral risk (wen. w ith i n. the same catl!gory, such. as i nll';rnal eIIoentsl is being com pared agai nst the cri'lI';ria o Thus. the riskof an ind ividual scenario would/should not necessarily be UIlac:ceptable if it faUs in. th.e "unacxejr tab l e~ region of an. FC cu rve. because the QHCE (as safety goals) might still be met w ith large margin o A mnverse corollary is that th.e risk of individual scenarios sh.ou.1Il not necessari Iy be viewed as " ac:r:ejr tilb l e~ in. the other region either. as a pruoont approach
'ID safety as.su rance always seeks tD incorporall! reaso ~
able additio nal contro ls w he~ ever a proper Quali tative engi neeri ng judgrTll!nt 0 r a Quantitative analysis so dictates, FaU i ng w ith i n the aa:eptable reg io n cou Id den.y the designers and others from t horough engineering th.i n ki ng i n.th.e safety design process.
- rf it is assumed t hat a fut ure design. of an. HTG R o r an. SMR meets the FC curve. then t he NRC w ill be on remrtlfor
m rti.fyi ng that t he level of risk -bil!il!ld safe!ty of t hi.s design is "aa:eptilble, - and in (JJ ntrast; any design th.at does not meet th is level of safety. even Fe r a si ngle aa:ident sa mario w i th all the attmdant unm rtainty. is unsafe. The same prob lem is en(JJu nb!red even if t he governing document is from the industry. w hether or not it is I!X plicitly mdorsed by t he N:RC, such as an ASME or ANiS st mdard as in 118 ~ HOw (JJ uld the regulator aa:epta design with one o r rno re accident scenarios in th.e *'u.nacm ptab le-region when t he governing industry standard itself has labeled it as such?
- Some cu rrent L.... 1Rs w ill likely not meet this FC CUIVe.
A misunderstanding o f the intmt of this culVe and t he role that N:UREC reporn play at N:RC (JJuld lead some t o incorrect (JJ ocl usions conm rn i ng the adequacy of safe!ty of cu rrent plants, because t he N:RC m dJo r t he nuclear industry them-selves (as. e.g. in [15,181) hav e labeled plants that do not meet th is cu.rve as " unacmptab Ie.-
- The FC curve is. in fact. i ntroduci ng new and more restrictive aa:eptanm crib!ria than the QHO safety goals as evident by ins pectio n and as ment ioned in 115 ~ in contrad iction ID the ACRS guidanm mentio ned above.
- The combi ned effect 0 fusi ng risk met rics as aa:eptanm cri b!ria and ap p lyi ng them on the level of ind ividual accident scena rios can lead to otherundesi rab Ie out(JJ mes. Future mac1D r designs off~ing lower total ( integrab!d ) risk th.an current operating reactors may be erroneously labeled as u nsafe-and not be pu rsued. 0 r be bu rdened w it h costly and un nemssary design mod ificatio ns.
o An exam p Ie of the above ( i nvo Ivi 'Ill a pob!n tially safer fut ure reactor design) is a reactor aJO lant ~n e break !br a high-12m peratu re gas-coo led reactor (HTGR). [n a by pothe-tical c.ase, it can be assu med that an ap pi iean t calc:u lab!s the frequency and the conseq uences of the scenario in a way that allow s them to show t hat it is "acceptahle.- Anyone ind ined to quest:ionth.e va~d i ty 0 f t he calcu.latio ns can:
(a) lXIint t o the degree of unc.ertainty in the pipe b reak frequency because of very
~ m ited nu mber of years 0 f ope-ati ng experienm w ith these reactors;
) poi ntln (JJnd itions such as high operating temperatures as add i-tional r&so ns fo r much higher fai lure frequency lXl'ten tial th.an in t he l W Rex perienm ; and (c ) challenge t he assumed rad ionud ide airborne fractions produced by unm rtaint ies in so urce t erms {e.g. long-b!rm diffusio n o f rad ionudi des through aJilteC fuel IJ'ilrtid es. ~us pension causul by vibratio n effects, higher t emperatu res. lower plab!out, e1L~ These challenges can lead ID a m nclusio n t hat the
.smnario falls in the "u nacceptable" regio ninstead
- Si mp Ie andJor pas.siv'e reaclm designs wou III have fe!wer number:s of acciden t scenarios t han com pi ex and active des igns at the same Iev'el of accident smnario delinit ion (e.g. sys b!m level) and w ithin the same PRA model.
The diff~ence in the number o f ac:cident scenarios (JJ uld be in multiples of 10 rather than in algebraic fractions. As a hypoth.etic.al examp le. two reactors may have the same ris k pro Ii Ie. bu t the Ii r:st has 10 sequences w ith JO rem at 1 E - 61yr.
and the second has one sequence w ith a consequence o f JO rem at l E-5/Yr. Und er t he FC curve (JJ nstruct, one is deemed acrept.able and the 0 ther is not; w h ich does not make sense in " real risk space.'*
o Thus. the use of risk-based a<E:eptance crit~ ia 0 nthe level 0 f individ ual aa:ident.smnarios (as opposed to integral quan-tities) m<l1( be viewed as penaliz:ing simple and passive designs in favo r of active and (JJm pi ex designs. in vio lation 0 f the Co mmission Policy Statement on Mvanmd Rmctors lUI.
- Again. bo::ause integral mllB.Sures o f risk are not Obtained in this model applications of these smnario*-I..... el and risk-based acm ptanCl! cri teria w ill be variable !br each design, s pec:i fi.c PRA model, and mac1D r silil!. The v a riab i~ty can be substantial in some GlSe.S..
It is i m lXIrta.nt that t he N:RC staff be mg niz.a.nt of t he above issues in complyi ng w it h t he Com m issio n d iroct:io n in b!sti ng the concepts embod ied in N:URE& 1860 in an actual lim n.sing app roval process for a future plant The.<ta.ff should ensure th.at t heir review w i II not deviab! from t h.e 10 ng-standi ng Com mission premaents in estab ~s hing t he many elements of a ris k-informul ap p roach. W hLle th is paper has IDuchul ulXIn on Iy a r..w ID pies.
fum A! I>aper:s can d iscu.ss the use of PRA. includ ing the introduc-t ion o f a proiJ(lSl>d tech nology-ooutral generic risk measure t h.at w i II allow !br cross-com IJ'ilriso n 0 f the level of safe!ty !br d i lferen t plant designs independent of site-speci fic characb!risties; al>"
proach to defe!n.se-in-de:p th; selectio n of t he So-GlUed l.im nsing-basis events; and selertion of safe!ty SSC:s in a risk-i n.formed and per!brm.an<p-based f rame:wo 11<.
It should be added t hat alternative and complementary risk met ries ID QHOs can be usefu l ID a IXItential ap pi icant for a design certi.lication o r (JJ mbined license, fo r examplelD assist in dem min.ation o f h.aving reached a sufficient mix o f preven*tive and mitigative featu res in a new design ( i.e.. safety design t rade-o ff decisionsl arID (JJ mlJ'ilre relative safe!ty of different designs.
The tech nology -neutral generic ris k meas ure mentioned above w ill satisfy t he latter need !br futu re reactor designs !br w hich the CDP and lRF metries may not be ful ly applicable. An example o f an alb!m ative FC curve that can be effe!ctively used for safety design t rade-off decisions is discussed in Sectio n 6.
The imlJ'ilct of the aforementioned issues may not be as great in p ractiCl! w hen t he FC cu rve of N:U RE& 1860 or a si m i lar (JJ nstruct is used o nly by t he designer as opiJ(lSl>d ID the regulator. The designer can use such constructs 0 r con.CI!pts as (JJ m plementary in!brmation in an iterative manner th roughout the design promss.
A prOb lem t hat may be en(JJ un teA!d in that promss is t hat a p roper interpretation 0 f some risk-bil!il!ld (JJ ncepts may not be as int uit ive for the designer. especially for t hose w ho are not PRA ex perts. as it may appear at lirst. rn addit ion, manuals of practice, stICh as standards 0 r guides that aA! developed by the industry may be mdorsed or refel1>nced by the regu laID rs and be usul in way s that produCl! t he uni.n b!nded resu.l ts (e.g. leading to rej ection o f safer des igns ). For t his reaso n, it is suggesb!d th.at t he use 0 f quant itative PRA results in t he (JJ ntext of des ign or regu lalDry risk-accep'tance crib!ria be avoided by all. Insb!ild.
Sectio n 6 provides an alb!rnative (JJ nstruct that ffi<IY be used by t he industry t hat w ill aa:ompl ishthe in tended purpose (design safe!ty trade-off decisio ns) without the negative (JJ nnotatio ns th.at are associated w ith N:U REC-1860s version 0 f an FC cu eve.
S. Inmlpmtati<>n of tho! as :Rem criteri'l>n u....t in lU (FR l00,l5U.34 The 15 rem cri b!rion used in 10 CFR 100 and 10 cm 5034 is o fb!n used as a de facto dose ac:ceptance cri b!rion !br DBAs by the N:RC staff. This *usage is. boweve:r. contrad ictory to actual Commissio n IXII icy and gu idance as described exp l.icitly in N:RC regu latio ns. as d iscussed in t his sectio n. SinCl! a nud ear plant is designed t o adequab!ly respond t o the O<E:U r",nee 0 f Des ig n &a.sis Events { DBE includes Anticipated Operational O<E:u nences and OIesign Basis A<E:i den ts~ t he ex pecta.tionis t h.at the associamd o ffsitecon.sequences w ill be smaU {e.g. fractions of 15 rem TEDEl.
Add iti onalcor'ltrols.
may be con*sldered Ad dltlonall controls need not be considered l og Consequeooe is-2 ~ A Ol~..a:l.JiCci cEt t ~i1.eI1r.J~
f;requ_erq' "eJ'5U!5 amsequ_eTl.a!: Ql:rw!
.h t GIll used by.o;pplic.o.l1I: durin~ dosign procrs..
whether add itional m ntrols shou Id be m ns idered for the s(ll!Ci Fic s<l!.nario.
(v ) The two regions are sepalated bya band 0 f perhaps an 0 l1ler of magnitude variation w ith diffused bou ndaries (such as in Regu latory Gu ide 1.174) on frequency and con.sequence.
lather t h.an Firm boundaries. This is because any single parametl>r o f samario frequency or m nsl'fluence (the mean is typically u..oo fo r all) is itself subj ect to un<l!.rtainty and eJ1suing ch.allenges, as the ranges of variabili ty and the underlying d istributions are generally assignol subjectively.
(vi) The m n.sequenee scale may be related ID app ropriate pub ~ c heal t h measu res and/or cost-beJ1eFit for t he i nel usion of the add itional control under m nsideration..
(vii) Si nce t his cu."'" is u..oo as a desig n aid for t he applicant,
~ Iatory staff would hav-e no position about t he acc:ept-ab i I ity or the lack thereof associated w it h any (>art of its m nstruct. ind udLng the ancbo r poin~s. The regu laID r must use the total ity of the safety in formatio n del iv-ered by the design and t he proposed operatio nal plan t hat ind udes the trad itional deterministic requirements alo ng with the su pplemental PM in formation in m ncJuding that the pro-posed plant is safe.
Note that the bo unda'Y region 0 f essent ially constant risk is on 1'1 m nceptuaL The designer may decide that in certai 0.
sub-Jt>gio ns and bocause o f specific considerations. such as ev-ents w it h particulady high o r low f lElquencies and/ or con.sequenCl!5.
and in t hose areas g.ov-erned by exis ti ng regu lations. devi.ations fro m the boundary region are warranted
- 7. Summary and. co nc:I'usimlS Risk-informed regu latio n is buil t around t he m acept o f usi ng tradi tional detl>rmi.n.is tic techniques of safety assu ranee su p plemen lEd by PM information and insights. Trad itional deb>rminis tic tech niques include concepts such as inm rpolation of red undancy and d iv-ersity. incorpo ration o f safety margins.
ap pi ication 0 f defense in depth. ap pi ication of q ua~ ty assu lance.
et c PRA resul ts should play a ~mited and supportive role in maki ng decisions about adequaq of safety i.n a risk-in.fo tmed regulalD 'Y framework.
Hcwever. re<l!.nt trends i nthe dev-elop ment of new risk-relalEd ap!>rlladles, w hether they are perfo oned by the ind ustry. N:RC staff 0 r other domestic or i nb>m ational bod ies, are towards heavier em pbasis in use of quanti tat iv-e PM "",til ts. These risk measures are somet imes compared to risk t hreshold values t hat hav-e attai ned an actual. o r ev-en a de facto. regulalDry stature o f
" risk accep tance cri teria" i.n CErtain instanC1!5. Such ap pi ications of risk measu"", for a nuclear reactor design 0 r a specific plant are not always in keepi ng w it h the te""ts 0 f risk-i n formed r<'gula-t ions. w h.ic.h caLi for com pari ng (i.ntegral) measu res of the calcu lalEd risk (e.g.. PEs and LCFs or t heir suitable surrogates such as t he COP or the LRP) against QHOs (or their s'urrogab>
targets. e.g.. 1 E - 4Jyr fo r CDr ) on 1'1 as safety goals."
In add itio n, using numerical PRA results. particlllarly those that are not integral quan*ti ties, in a ris k-aa:eptanee m ntext, ev-en by the nuclmr industry (as opposed to the reglIlators) can have numerous u ndesirable con.sequm ces. Examples of these among many d i.scussed in the text i nd ude: the tendency to pena~:z.e simple. passive safety system designs in favor of complex. active designs; and futu re rea::tor des igns 0 fferi ng lower i ntegralEd risk t han t hose 0 f the cu mmt and high 1'1 safe 0 perati '1!l reacID rs may be erroneously labeled as unsafe and not be pursued. or be bu rdened w it h m stly bu t unnecessary des ig n mod i licatio ns.
These issues can lead ID serious un i nb>nded m n.sequenees in licens ing of fut ure rmc:t:ors o r creating new challenges regarding t he safety adequacy of ex isting plants.
The paper also 0 ffered an al ternat iv-e use fo r a frequency v-ersus con.sequencecurv-e as adesign or operational safety optimization too l for use by the reaclD r desig ner o r plant operator.
Diisdaimo<Y The wo rk relalEd to t he development of this (>aper was condUCb>d at t he IElq uest 0 f th.e Oi rector of the Advancol Reactor Programs at t he Office of ~w Reaclllrs (now retired ) in the last quarter 0 flOOS at the US N:RC. w hi Ie t he au tho r was on loan from t he US Dl!partment of E ~.
Neither the author. nor t he United States Government; any agency thereof. 0 r any 0 f thei r em ployees makes any warlanty.
ex pressed or impliol. or assumes any legal liability o r responsi-b ility for the accuracy. m mpleteness. o r any thi rd party's use of t he results of such use o f any in formatio n. plUduct, or process d isclosed. 0 r represents t hat its use would not infri nge privately ow ned rights. Reference herei 0. to any sped lic com mereal product, precess. 0 r service by trade name. tlademark. manu fac-turer. or otherwise. doos not neCl!5sari ly m nstit ute or imply its endorsement; remm mendation. 0 r favo ring by the Uoi ted States Govem ment, 0 r any agency thereoE The views and 0 pi 0. ions of the auth.or exp res sed herein do not necessarily state or reflect t hose of t he UnilEd States g.ov-ernment o r any agency t hereo E A.d latow IedgrDl!n1!5 The aut hor w ishes to thank Or. Do n Dube (US N:RC) w ho was t he Ii rst expert to l1!View the eady versio ns of t his paper and offered his b road and in-depth knowledge in su pport of its dev-elopment. Mr. Alan Kuritz.ky and Drs. Mo hseJ1 Khat ilr-Rahbar and Doug True p rovidol many usefu I insights.
(1 1 ChojllllOn J, H=
RisJc';"_. l:edm:i~41e1Jtr.i ~
.. oo 1ia=i'18 f.,.",.,.".n ilr n.... """""" pi "'tl. In : ANSI'SA~~~~
tJll'SA during /he.... duT ~Km
,ll ~ ffi.~ '~ ll. ~.
121 US
- c. ~1 Regis.... ~1 j;Jt l W2fL s..fety goo.tls for /he oporotio", of mdur pawer plonts" August 2 1. 19!11l
III lIS NJC. le:teQl ll.<8ister.
- q. _
an u.. 01 prdJ.mlis;;c risk
_=ment metI1oo. in mdo.oT regulrnll)' ""~.1lin~1 Micy !ltot~
August 16. vot 60; US). l!l!1!!i. P. 42622- !L 1<11 lIS N:RC CeJl~ Ister CUIIlO2<i. D n:I:Mdu~1 p1~tt..... min-. i" ~
..ment vulner;lbilities,.}lavembor n. 19sa I~ I lIS :RC SOC'MI!l-144. White popor.., ris!c-in:b!:mod ~d pedm:m~ce""'<d regu1oticm. J;>.TI~T)' 22. 1 :Il SQrf ""1uiT<:mO:Ill5 mo:mor.oTldum ~Pl"""ed UoTcl1 1. 199!L 161 lIS JC. ~'b.1:al)' gu.ide 1.1, 4. An ~Jlllroo<il1 fo,... ing prdJ.mlistic risk
_.....,.,nt in risk in:brm<d docisi<Jn. an plo_pocific dh_ lie tile liEJ1sm:!: msis,. R~:ian 1. moem1Jer 2 -
1'71 lIS NIl<: NUR1lC-()fIiliJ. lIS
..a...T liquJ.. t<JI)' Cammission.uoom """;ew p1.. n. limsion l. M ~d1 1 lCh.>pter 191*
lSI lIS
- c. NUJtrC.l UiJ. Severe..nctent rim: ~TI "",,,,,,,,,,,nt for five lIS
_T
_erP:Io_; 0rt.dJer 19 191 lIS JlC. SEc 4!9. 102.Im1'lementa:i.., 0I~ gml p<J!icy;M.>:r.nlll. 199!L 11111 lIS C Ml!S I.<;Jbor I!o iNRCCI""j,:n",n. Ml!S<:<II1llDeI1ts.., ~n ;mpl~
- <JJl p'b.n i
- l1lihe uiot
- r gml policy. IiIb)' 11. 199'.
111 1 lIS JC. SECI'..tJ19fL Stm!s "'JI<>rt on risk';nfmmed dh_ lie tile
~~l ""1ui~ofl ll a;R port 9ll (optian l) w ~
.. ti<Jn.on ris!c. :b!:mod
.n~nges 10 10 a;R port 9a.4<1 (o:nnbustihle g& m_~
tembor 14. 2Il (12 1 lIS -
C SEC/'.
111 ~
utianol)' light wo1:<r.- (U'IIi) ~n iss"", w /heir r&1iomh:ips I!o ~
reguIotOl)' requi.-.men Juno 2Il. 19 1131 lISNJC.Feder.. Rqister. ""t 7:l (99).N 2IlOS-tlZl,. oIicy.... t<ment on regul.. ~.., 01 ~ v"Tla<I ~
0rt.dJer 14. 2IlOS. P. 00612~.
11 41 lIS NU
- c. l nfLD~
pre~p1ic.o.~ <m...rel)'... ~luo.ti.., "'JI<>rtfor /he m>du.IoT hign..~er.o1ltae gos-cooled ~r. Uorcll 19s9.
11:51 lIS NJC. NUJtrC.1S61l. ""","".i1il)' -d)' for..
risk inbmed w I""f<Jrl1l.>TI.....b&ed regul..
~,,)', lructluTe for hi;tu", plott Dicensin<. *....,""""'"
200'.
(161 ~
.. Al<lmia. ToJl"l ~regulot"'}'<riteri.m /he.... Tldm -IHTCR"OO I<<CR-8!5
- Sept"mlJer 199!L (1'1 1lur0000000TI (hmn:rissiom..
~7OJI""'TI...rei)' 'Jlllroo<il1 for mOOuJ.. T Hi' DoaJment na. ~
.-Sf4.l. ~ lri dlod dism:mution. April 1:5" 2OO~
11 ~ I ANS-S.1.2tJrot Nud...., uiot:r <ria.. ~nd... f<ty do:sign p:roc= for madu:lor bolium-<llalod......am ~ts. ~; Jun.e n. 200fL 1191 ~eI)' 1Ii!:p
- 4th in"'n.>tian~l "'Pial meeting.., ~.igh temperalll"" __
- St.off requirements mo:mor.oTldum.. JlP7""ed Septem'bor 10.
- c. SIlCV-iI!Ul(J!;G. SUfi ~roo<il1 ~Tding.. risk.,informod md perfor:n:an...b&ed,.".,..., to p.ort ~Il 01 ;;t1o 10 01 the (hde 01 I'<d_l
- c.
- 0. 2000.