ML20043H220

Seminar on Use of Unusual Event Repts for Improving Nuclear Power Plant Safety.

ML20043H220
Person / Time
Site:	05200002
Issue date:	05/18/1990
From:	INTERNATIONAL ATOMIC ENERGY AGENCY
To:
Shared Package
ML20043H221	List: ML20043H220 ML20046B362
References
IAEA-SR-169, NUDOCS 9006220253
Download: ML20043H220 (53)

Text

r 3 CG lb 5 i

DGA.KFT 2 016 e4116 33 15:14 TAI -31 70 3334016

- 1 q INTERNATIONAL ATCWC ENERGY AGENCY f

h &

oECD NUCLEAR ENERGY AGENCY in cooperet:0n with the i WORLD ASSOCIATION OF NUCLEAR OPERATOR $

SEMINAR ON THE USE OF UNUSUAL EVENT. REPORTS FOR IMPROVING NUCLEAR POWER PLANT SAFETY '

Vienna, Austna,14-18 May 1990  :

t IAEA SR 169/ 66 - l Lessons Learned from Unusual Event Analysis i to It:preve Measures to Cope with Steam  ;

F Generator Tube Ruptures t

r I

Dr. K. Kotthoff M. Simon GesellsChaft fG Reaktorsicherheit (GRS) mbH f Cologne, FRG I

1 i

f l

e f

l j

- - ON i O D lo 2 2 D '2 6 3  !

- That 43 e preprint of a papet intended T0f presehtat:0n et e scaentific rneettrq. Secogee of the Dr0 visional nature Of ftt Scif tant and 4.nce trisnpas of sutistance of Clytail r-wy have to be fnsos before pubhcotton,the proprent is fnano switsB!e 06 the i u ncerataTr?Ig that At swall not be Cated sh De litereture of in 42V WeV be f tprDdwCed Art its present form. The views empresseI 1*4 statemta f*d3e refnsin the fesponsbuity of the named e.athof fs); the seed do act necesaartly foilect th084 Of the 90=em.

t'ent Of the Ostgfettrg Member Statelt} Of Of the CeSigf.418MB orr,anizationt:1. In particular, ne.eer See IAEA aor sar *M*' ',

09pana80t*Dn O!bOdd 4909)&Ofeng thtJ meetstag 540 be bed 99100f13ble 107 eny Mletetsaf toprDduceg ch thd pow 9hvFt..

~1 I

-l Vessel Head Penetration Cracking l in Nuclear Reactors s ~!

i e -

Nuclear Vessel Head Penetration Cracking in France Part i l

'!- By Mycle Schneider  !

WISE Pans France e -

Part 11 The intomational Perspecthe By Thomas Panten and Helmut Hirsch Gruppe Okologs.Hannover Germany Norbert Meyer artj Detlef Rieck '

Gredswald 6

h I

ww.4 w a,oane.aos i.w.-s-w sad Greenpomos Sweden g March 1993

l.  !

l 1

n ,

~

l i

.a . . . . . . . , ,

l

e. l l

. CONTENTS .j I l

- 1.1. Introduction  ;

3  ;

1.2 The Vessel Head Penetration Crocking Phenomenon 4 3 -l 1.2.1 The History of Detectionin France '

4 l 1.2.1.1.THE 900 MW REACTORS '

7 l 1.2.1.2.THE 1300 MW REACTORS l 7  ;

1.2.1.2.1 Vessel Head Penetrauons  !

7 1.2.' 12. Pressurizer lastrumentauon Nazzles 9

l.3 Technical description of the Vessel Crocks and their Origin '

9 13.1. Vessel Head and Penesration Design 9 l 13.2 Charactersstics and Origin of Identined Cracks 12 -l l.4 Safety implications of the Vessel Head Crocking The French Official Analysis q

13 j I.5 . ounterMeasures Adopted C

tag =*=

13 j L5.1 13 ,

1.5.2 Laak Naww= Symuns .

14 l 1.5.3 ControlRod Anti E decuan Symon 14 . l 1.5.4 Tesaperata l.owerag/Dowaranag 14  :

1.5.5 Penemnon RY- and Rapsirs 14  :

Vessel Head P1-'

1.5.6 l 15 Conclusion 1.6 l l

i

.  ?

l L

I h

3

. t I

a _ - . _,..

SUMMARY

Part I la September 1991, dunng routine esting on a French 900 MW nuclear reacser, a leak was detected reactor vessel bead. The cause was idendried as a cncked vessel bead penetracon. These devices allow the-control rods, the crveral device for safely shutung down the reactor, to manoeuvre into the pressure vesset The rupture of one or several vessel bead penetra60ns could therefore lead not only to a loss o(ccolant ac also a severe reduction of reactor control The pbeoomenon was immedimly taken senously by the French operaer Dectricite De France and th authorides. Inspecuous have also been camed out at other reactors, cocGruung the genenc r.haracter of i problem. By the end of 1992, it was clear that potconaDy an 53 operstmg French pressurued waner r (PWR) could be affected. Nevenbeless, by February 1993, only 10 reactors had all their penetranons inem and crachng was detected at 8 of them. Of particular significance could be the fact that at least I reacror (Gravelines-4) loaded 'with plutonium tzranium fuel (MOX) is affecsed. The lack of automauc inspection devices, high doses ioduced by manual inspection and high 6 p ~5 ~ on soclear pmwr in France hindered the safety autbonues to impose a stncter inspecuan Mia The performance c(inearh tectak t ues, leak Wa instrt=~=* and control rod anti-efeerm devices reams imetent. The problem has been costly in terms of rad 6 oon exposure, and in& and repair. "Ite most expensive economk burden stems from elecmetry reptment costs which can be valued at snare billion Freoch franca so far.

l Part II I h

Reports on vessel bead penetracon crachng (VHPC) ftodings have not been confined to France, but fu incidences have nw in Sweden (Rhingals), Switratand (Bernau) and Belgium (7thange).

Altbougb plant operators and regulatory bodies from other countries have repxtedly been following th issue, no inspection efforts were imm-Aietely initiated; on the coctrary, the safety implicadocs have been played down and the appleabtlity no reactors outs >de of France neglected.

It is noteworthy tha the undersranding of the VHPC pwm has imprwed linie over the last year, despite the conciscalincrease in namher of cases. Fouowing the first i ntranm of cruiang, the nuclear scalysts bas-tened n give an answer to wby tbc cracks had developed. This explanauan had to be moddied sep by step according to new evidence showag up in otber reacsers. This socnewhat unorganised approach, bowever, not been successful in amving at a thoroogb understanding of the penetrauco cracking phenomenon.

1 I

t1mma esas pourfmaties emarma a usoJ.as atace

? ,

n ,

With the exclusion of the possibility of circumferencu! crrbng, the risk of serious accidents has delibersteJy ,

c; been ruled out, despite reported cracks indicating that such an exclusion is more than doubtful. Apart frtun the fact that toegitudmal crebag also poses the threat of unnoticed advanced cornmoo of the VHPs and even the vessel head, a rupture of a VHP initiating from circumferential :: ricking is posable and wi!! lead to an ontso-E 1 stable leak in the pnmary circuit, which could be the precursor to a core-melt _h' 1

There are further possible his resukeg from VHPC wkh the potential of large radaoactive reimes in

1. coctrast to the official statements on the risk of VHPC that suggest the worst potential consequence would be corrosion of the outer surface of the vessel head.The hazard of the potential impatrment of the reactor control e system has completely been neglected despite its vital importance for the reactor to be controllable.

W .

The development of the VHPC issac exhibits a sorry. state of awareness of both nuclear operwors and regulato.

i p ry bodies. What should prudently be done from a safety point of view are ham inW g=riem into the fail-are emne and kinetics of VHPC to arrive at a thorough anderemeng of tbc pbecomeann With innmente

' and highly reliable ine at all reactors employing inconej 600 VHPs, and inclusion of the VHPs into in-serwce inspecuoo programs in all 5ther reactors.

k Unless these menwes have been WW'M mo upelse and we!!-founded ref== program can be nr*M which of course has e be based on a thorough knowledge of both failurt mode and exaent. Rescsors k a sb. - d e c.

in - v H P C b.s - - .

bD i

b 4

4

PART 1: Nuclear Vossol Hoad Ponctration cracntng iii ria. uc j

t.1. kttroduction In Man:b 1991. EDF made the fo!)owing statement on the'sute of beslth* Moeir 900 MW rexsors :

results of controls and tests codemken in the framewort of the seven decennuj irgw already camed demonstrate the excellent resistance charactenstk of the matertal used with time".'

Six months later, in September 1991, a routine Wit bydrotest on the prut.ary rukg system of a Frenc l

900 MW reactor led to the discovery of what semor EDF Wrat ofncial Jean-Pierre Meroer called 'th most senous' nuclear power plant problem the French operator had yet to face.' Substantial cracks covered in the reactor vessel bead penetrances Since then and up until mid-February 1993, other cracts have been detusM in at least a dozen more rea: t After seam genenfor tube crachng. brtken control rod guide tube split l pins, i pressaruer instrumen tration cracking, badly installed controlled venting sand filters, fauhy coctrot ~ ==M cab es, ma n stca:n line cracting (even in the co-bresi zooe), msufGenent maint- procedures, etc. yc1 another geocn lem troubles the standardized French reactor program.

Nucleorsies Week has eneM : 'Irocically, the peactraboes had been 4amed la a Frmr-2' erne s locomel-600 reactor w5-: n as one of the places most saaceptible e stress corrosion crack:ng reactors *. While EDF was preparing *no take a look" at VHPs ckring planned outages,

• events ov  ;

utility" when cracts were discovered at Bugey 3 ' EDF scales that it was aware of the locooel

~

• an analysis of the suspicaous paru had led to the vesse! beads being act rhPed as amoegs tive compooects; this IIlcstraces the tarh*mf dttncn!ry of a priority analysis *,*

One year aAer the first dtscovery of the mael head penetradoo cracking (VHPC) problem l only a etrations had been Med' Only five reactor vesac! heads had had ad penetraucos checiev acd seven l par-tial inspecuocs had been camed out. At least 24 plants (6 a 9CO MW sod 18 x 1.300 MW) see '

more threatened than the rest of the plaats becacse of higher temperacares under the vesse) bead.Th some of the 1.300 MW rrrinr have been put into serwce fairly recently suggesu the Manhewt of a low cracking probcbility. One second generaboo CPY 900 MW reactor c'MM did oot present any TMs seemed to confirm the thecry that CPY reactors are less suscept.ible to VHPC because of the low perature eder the vessel bead.

~

s Two menth later, in November 1991 the bad surprue came with the ima<% of the VHPs at Blayai a CPY reactor. Three of the 65 penetrances were affeaed. " Cracks bave been detected stere o expect them*, wrote EDFs chief safety inspector Pierre Tangry in his last inrmi repcrt' It

' the phenomenon pe**11y affects all 53 French operacng PWRa.

1 EDF.Baan de sazr.d des matra.lcs anddage 900 MW,Marca 21.1991 2 Nxleades Week. November 21.1991 3 N-wes Wect, boary 2.1992 4

EDF,"Sibczd :adlaue 1997* Rapport de 1*f*y~~ Gdae:n! pour la strati Ned6 aire 6angvy Report), has 1 $ sacring to kan-Pient Mertaer. EDF. gaated in4~-a Ne6,=

Weck. a Wect. 57* 24.1992 See 24,1992 6 8erey.3.-4 a$d 5.F'wnasvme-1. Paleel 1 accad ag to N 7

EDF. $ facti medLa:,rc 1992". Rapport de I*laspassar Genen! par la $4 red Noddare Gampy Rapart). h 8 D5IN.MAGNtX".Febrnary 15 - 21,1993

~

In December 1992 probably the most significant and the worst mformanoo concermng s.afety implicaboas had e be digested by the EDF management. Tbc kahog VHP of Bugey 3 wheb Grst drew acenoon to the crack-ing probkm wu extracted and examined in detail. Metallurgists in the laboratory observed

• incipient circum-fertatial cruiing*, cootradictmg the re:ults of Mffing and expert pr*3nat'This type of anchog can mored lead to rupture without any previous kabag and thus witbout warmng. The rupusre of a VHP and the subsequent eyecuon of a control rod drive m~+=am (CRDM) wouM crease a breach in the primary emhag sys:cn whacb couW enemally trigger a core melt

• It should be notad that it took 15 Incoths before the circumferential crack indicatiocs were Menti 5ed after the dascovery of the leahng VHP and they went anno- i ticed with the current in-service inspecuco and actKksmacinear9matm t-+ap l By the mMdle of February 1993, only 20% of EIWs reacters had utdergone n n,rd te in-service inm-@ of l their VHPs; eigbt of these were affcced by the eacbeg Wanrt Five VHPs mere identi5ed cracted in eight partal remanr inererW By this time vessel head twd-r had been W for at least a dazen reacsors. Around 500 million French francs have been spect on the problem du:rmg 1992.

The different tirds o(inspecuoo w'niques used for the VHPs of the vanous reacsors have not been spectGed.

j 11is obvious that visual inspection, for exampk. is nuset ks: reliabk than ultrasound or liquid penetrant exam-ination. DSIN refused to transmit the data e the aatbar on the grounds that it wouM be 'clearty wutna the operator's errmetwe* no supply the informanoo and DSLVs role is to *cx;resa its point of view" on the posi-tion taken by EDF. DSIN had in faa ahtady asked EDF by te = wie of nu-h- 1992 to supply the data e the author, but the aatbar has not. so fr, had any positive reply truca EDF.'

The difnculty of rapid in depth h;+T-x of all the pt=~inny damated vessel heads shows just how problem-stic EDFs very high level of arv+=r <Legm4rer is. Whtle this is not the only safcry related issue. EDF is forced to admit that prodoaioc capaaty repinermrar is act ocJy costly but somerrmr5 bardly feasibk. (see

- Lt. for a detalkd outage eerutm) This is a!! the more drearv hu- of wdespread cleanc space- '

hestmg and extensive electnccy exports wbcb su+ad Oday to the prodocuan of about 10 reaaort Under these circumstraces, vbo conM possibly take the polzscal decsxc c( shumeg dawn mort than 20 reacton at the same ume in winter 7 i l

i I

9 lanen to DSIN dmed. December 12.1992 and Fetrary 24,1993, pernoed coma ==e=rv= wt:1 Veidh b e DSIN.

Feernary 25,1993 and necer fices Df!N. dased Fetriary 26,1993. Sewn! rm;ssats for h pur forvei dinczj to y the EDF d.d us =~e=" any pocave r= ~

I

,, . n. , . ~. - a + ,,. . , ~ a m.

I 1.2. THE VESSEL HEAD PENETRATION CRACKING PHENOMENON 2

1.2.1. The history of detection in France l

Between September 1991 and the maile of February 1993, about 5% of the ca. 700 kW VHPs (20% of the tout) were found to be affected by the cradang W_ k& +R+s were carned out on 18 renesors[a third of the currently operatmg ptacta (six CPO 900 MW reactors three CPY 900 MW reactors and nine 1.360 f MW reactors). Only 10 of the 18 re*2ars had all their VHPs inspeced. (see foUowing table for further dentis) .

DSIN is cunently exmaning EDf's proposed kLep~ te program for 1993.

1 Table 1.1: Published Inspection Results in French Reactors by March ist 1993 ,

Capacity Program Hours Temperature Cracked VHPs" Reactor l MW gross life time *C under VH l I

i Complete inspection Bugey-2 957 CPO 75615 '315A 6 Bugey-4 937 CPO- 75554 3154 8 Bugey-3 957 CPO 74330 3154 .2 l Gravelines B4 957 CPY 72d98 289.1 5 -

Blaytis 1 957 CPY 70914 289.1 3 Tricastm-4 957 CPY 70400 289.1 1 Paluel-1 1345 N 46957 313.7 .

O Paluel 2 1345 N 45265 313.7 0 l Flamanville-1 1345 P4- 38186 313.7 1 St.Alban/St.Maurice 2 1348 P4 30%:7 313.7 1 .

PartialInspection i l Fessenbeim-2 930 CPO 87303 313A 0 l

F e

• im 1 930 CPO 84320 313A 1 Bugey-5 937 CPO 78548 315h 2 Paluel 3 1345 P4 43604 313.7 0 Paluel-4 1345 P4 39199 313.7 5 St. Alban/St.Maonce-1 1348 P4 35935 . 313.7 2 Fla:nanvine 2 1345 P4 ~ 35501 313.7 0 Nel 1345 P'4 28975 313.7 1 Scaras : CEA.EDF Nedancs West. DSIN l

l 10 - Uscil the end d 1992. a:xxrdag to Nh Weet February 11,1993 11 Nesber d VHP: wth idese15ed cracta l

l e l M  ;

i 1

l I

l.2.1.1. THE 900 MW REACTORS l l

On September 23,1991, the prunary system of the Bugey 3 rerwr, a 900 MW PWR, was sutycted so a hydruest. Such a test is pan of the decennial maintenance program for a!! French PWRs.The test is performad at 207 bars after imioaing the fue! ehn The operwant design pressure of these reac- l iors is 155 ban. An acoustic system. specially instalkd during the hydrocest detected the leak at the Bugey-3 reactor According to Franatome, after shutdown of the prunary pumps at 166 bars, the acoustic sensors pro- ,

duced an increasmg noise inscanng a pressure increase." A leak of about 1 litre per hour was identified j

on one (n *54 on the pertphery of the vessel head) of the 65 tenael.600 vessel head penetratins. The

~

CRDM and tLermocouple cooduits enter the core through these penetrabons. The Fessenbeun and Begey reactors were the fkst 900 MW reactors of the so ca!1ed CPO (Contrat de Programme - 0) senes of Freoch reacsors.

By the middle of October 1991 investigations hai! revealed !! loogirmimmt cracks of 15 to 73 mm length occumng in two groups ao sbc inner wall of the incrWema~1 VHP of the Bugey-3 reactor. Four external t acks indicaed penetracng cracks of wtuct only two were thought to have caused the lest in mMem the sube itself had hamme somewhat oval. -

Discussions beeween EDF and abe safety ausbartties DSIN led so the daneba to nam +ar all the other i

penetrances of the vessel head before the end of 1991. At the same time it was decided no extend the nammarions to aboct 40% to 50% of the VHP of two other 900 MW reactors wtuct were shut down j

for refue!!!ng at the same time (Begey-4 and Fa==>ahalm 1). Bat accanhng to a Fr=mmnme document by the end of November cely 12 other VHPs at Bogey-3 ' at Bogey 4 and 12 at th*-1 had been

,,. m Of greater impcnance. bowever. was the fact that the VHP: namined those not housang any CRDMs.

altbough the leaking VHP (n*54) at Bugey 3 did h fact boose a CRDM. In fact, in the case.of the 6 reactors of the CPO senes the vessej bead has 65 CRDM and thermmple sbatta.*Itere are 48 coottnu-oosly fuxnoaing coctrol rod mar +*nmt 5 rarhAnnt *= Met 8 a$didocal penetrat>ons exist for MOX cse bot are oc!y +W when MOX fuel is @fe1 None of the CPO reactors are currently beded with MOX nor will be in the foreseeable future.

Up and! the eed of November 1991.ocly one more VHP equ:pped with a CRDM had been eta =uned (a*65 at the edge of the vessel head at Bugey 4). It was aho found severtly dsmaged with at least 8 crack Er#v arvwn!

The main latpee6aa echmques used wert televtscal and eddy cirrtet mentremenn in excepdonal cases, the more predse Squid penennt and ukrasound mr*ait (imptw"hle in the case of VHPs with t

thermal sleeves) were used. Televiscal observa6oo was the ocly noe mannst echn2que used in these 12 6--- 1sner to Frmamome owner Group (FROG) mem6ers, daad Ocsche 22.1991 and tatks dmad NA 27.1991

- l

s e m - m ie .c-. u w sec.m < - ~

o I ,

scspecuons. DSN stated that "the way in which controts are performed today presect senous problems"f Cert.un control practices require the " des:ructive dismantling of intatal mechsaic*ts of the sleeves". j And as DSN points out. *only a limited number of reptwment compoocots are currently avadable'.

Arother significant problem is "tbe high cost in dose of the **= min =r%s*.* Indeed EDF *eent>d that the complete inspection of the Buf y.3 vessel bead and perual inspectie ts at two other plants wouU ajone cost 1 man-Sv (100 man-rem)." By the end of 1992, the collective dose resafting frocn inspecuan I and maintenance s.ssoctmed with the VHPC probacm was evaluated eo be about 8 man Sv (800 man-rem)." 1 I,

}

]

Automated inspection technology is expected to replace etisting techrx>Iogy by and by. (see chapter 1.5.1.) New atenals and coctrol mesbods had not been ulequately quanfied and eva!uned by the a:ad-l j

die of February 1993.

The first follow-up senes of etsmutations revealed more cracts of the same type around nocher VHP at Bugey 3 stound eight others at Bugey-4 and arou:x! coe at Fa'* 1.

e On December 30.1991. EDF decided to perform cuernal televisual **=miaarims ou the three ocher resciars at Bugey and Fa<=anheim and to lestalllest decemm syrems ce att six reacsors. A complete 1 i i esamination of the '/HPs at Bugey 5 was to be carned out during ibe planned refuciling outage which I begun at the end of April 1992. Televtau.nl erammation of the VHPs did not reveal any a-tin

- However. DSN requesaed further inverig-tma On Decemb:r 31,1991, DSN asked EDF to carry out =M+ml televtsual *" min =* on the interior l surface of the Bogey 2 veuel head and the two Fa=Wm rmvrs during their planned outage berto-l ning at the end of January 1992. According to EDF these controls dad not reveal any ano taly. Othe control techniques revealed tboogt six affected VHPs at Bogey-2 aner mmp2 <> inspection and two g cracked VHPs at Bctey 5 after partial hspecuon. As by the saddle of February 1993 the two

,E Fessenbei:n rem-rers as well as the Berey 5 resca:r were'sts act e*h in-e**d By es.rly 1992, the Engey 3 reactor had anderhoce more thorough errminations. The inspection a:ch:uques nacd indM

e. internal televtsu.! observanoo of all VHPL

. toternal eddy current 1-mm of all VHP :

4

• intanal uhrasound in= par'im of the our.cr suTaces of all VHPs.

. bqcnd peactrant enminitiocs of weMs and the inner surfaa of the lacrime ='ad VHP n*54;

. e'= min

• tion by rephca of a liquid ym ; him on the weld of VHP n' 54;

. general televtsaal er ammation of tbc insade surfact of the vesac! bead;

. a print was taken oc VHPs c'54 and 57.

13 DSD4.N<ms d'1akrmem= Paris. December 2.199I 14 - Net--a West. Jaamary,2nd.1992 I 15 EDF.-Samheu=n 1992" 7.appan & 1%r-*-- CMedal paw h 54reWicum (Tangryf.cpam. laamry 190 I

. - - -- _ . _ _ 1 l

t l

- The folkmng uatement appears in bold leners in a 45 page report on the VHPC problem" : "No cr-cumferential crack initisbon has been i$entined on the inner or outer surface *. Like several other

• facts" presented in this document. this sta' ment had to be corrected later. In December 1992 detailed ana!ysis l camed out on the estracted Bugey 3 VHP n' 54 revealed " incipient circumferential creeking in the weld sooe oc the ostside of the penetrat$on". (bo&d lencri from the author). DSIN mnt% that the escks are 'probably due so the fact that the space between the penetraboc and the vessel head was )

maintamed in the presena of;rt: nary wwer, followmg a throughwall noegitudinal crack *? EDF com- .

mented that "if you leave (a longnudmally crxied) penetrapon in there long enough*,it could begin to l

crack from the outside " Commoc tracks on the vessel head indicated in fact that the crack had been throughws!! and was ahtady leakmg before the hydrotest.  ;

1 The recent identificaton of incapient carcxanferential eackmg around the leumg Bugey 3 VHP was a also a slap in the face for the reliability of in servu:e iW techniques. In fact. probably no other VHP of the French nuclear progran was cummed as carefully as n*54 of Bugey 3.

I In November 1992 Mx:r bad news came for the CPY sta: sors thought to be snart corrosion rmunat

- beeuw of lower c estures imder the vessel head. 'Ibree of the 65 VHPs of the Blayais 1 reactor were found cruew fbe cracks reached a depe of 10 man? Later on, two enore CPY staen:rs wert found affected by VHPC (Tricastin.4 one penetrabon was cracked. Gravelines-4 Ove more VHPs were crxied').

~

The Gravelmes-4 case is of perucular significance, atace k is the orst rencaor loaded with 30% aranium-plutonium mixed calde fuel (MOX) stert VHPC has been hf=4 Because of higher residual reac-tivtty, the vessel head is W with eight mManant CRDMs.'the tracnon of delayed neuaues ts lower for plutociam 239 than for armssa 235. la other words, a MOX loaded reactor reacts faster than a uranitun cort. This is of perucular rnnerrn la the case of specs 5c amamt noensrios. Whatever the  !

probability asse**=>at might be, it is evuSeat that VHPC tacreases the rhi of CRDM ejection i

Accorthng to DSIN, so this case *a bcal lacrease in power could lead to a prunary em man bodmg cruis Os crise debo111 tion) and a migere rise in feel power output wi2 the risk of etmAtng and pe!!ct dam- l age'. Even if safety linWs see not reached 'e=L ='wh< brve shown a rMan in the (safety) mar-

' gins when swtsching from a arne==m th a (plssoc. turn-eracium) inized cort"? In oder words, it is coort difficult to operase a reactor wie a MOX man wr2 a trannan core.

l i

l 16 EDF.*1 cc ds rimetem",an&ead

)

17 D5Di. MAGNUC. February 15 - 21.1993 15 Me=cs Week. Ianmary 21.1993 19 D1Di. MAGNUC.Decessher 7 13.1992 20 ba=W de 1*1mdesme.'1tacyciage du planameus Ames,nes h & ens sous prendos",Jemmary 31.1999 i

... ~ . . .

i i

l 1.2.1.2.THE 1.300 MW REACTORS l 1.9

• S C Vessel Head Penettstions la December 1991 DSIN asked EDF to psfairm the relevant control operations on the Gru W d

- 1.300 MW reacsors. la February 1992 some of the vessel bead peserataans of Paluel 3 were czaamed.

i No cracks were identined. '

l Up until the cad of nanhar 1992 only two 1.300 MW vessel heads, Flamanville-1 and Paluel 1 ha been completely e==taad One crack was found at Flamanvine-l some a Paluel-1. Parnal examma- l does were camed out on Sve other reactors. No cencks have been identined yet a Flamanville 2 and '

Paluel 3. One penetration was found craded a Casamon 1, two at Saint-Alban 1 and Eve a PalueM.

Up until tbc end of February 1993, only two mort 1.300'MW reacsors had had all dieir VHPs inep.aa j

(Palvel 2 with no creci indication and Saini-Alban 1 with the two crachad VHPs).

Saint Alban 1 was shot down on May 22,1992 and was not estaried until December 16,1992. This l was due to an incredible discovery of more than 200 cracks is the four mais sacan lines of the reac:or.  ;

DSIN requested repair of at least some of these cracks before it graased the restart beence. ,

L2.1.2.2. Pressurizer lastrumentation Nasales l As pointed out eerber, the vesact head peastrance problem is act the only one hakad to the see ofj

!acceel 600. Even before VHF cracking was discovered, Framasonw identlGod about a dore, places 1

where the material could cause trouble. Steam generssor tubes were obvicusly maMm but also VhPs.

i la 1989 another spot proved tremhlemanw : the pressuriser astrumentasson accales (or c1aneceons) of the -

~

l

- 1.300 MW macscrs. The tass ====taa aosden of du 900 MW rescarts are made of naalessj Similar to the VHP lustory, the probacm was firm ideadGod detag 2(T1 kr hydmeests. Problems were I

identified at Cattenom-2 and Nogent 1 dunag eie trut Sve year inaparma kg.a on these reactors which were put into service in August and Sepueber 1937 respecdvely.1he preemsuer posasenes 11 mstnamentation norr.les with a dameser of 30 asm a the pressunaar (five at the up and sin a the bot-som). These nozzles allow conteuously measurements of tempernsure, pnssure sad waar level.The con- ,

nections are expanded and welded com the stamiens sesel ciaddag of the inner side of the pressuriser. f At Canenom 2 a welding defect was identified and at Nogent 1 a longitudinal crack was detected l

between the weld and the edge of the expanded part as well as as oudsaed zone. Stress corrosion crad-hig was idannrwd as being responsible for 112 crachang. It was decaded to replace the faulty connec.

tions. The repair was quahfied 'delicase* by the safety authorum:s.1be pmcedures had tol j

beforehand. It was'also doculed to carry out visual (external) and liquid penetrant (internal) inspection '

j on other 1.300 MW reacsors during their refuelhag or other planned outage. "  :

l i

. l

.21 Enimers de I'ladamas. 5enssim sur la $ssent des lamansmans Nacianusa, s*00. es=enas Mayoues Ital 7

- em i. w ,v w .e. w .,cmnrou

. 'the inspecuons camed out over the followmg two months revealed longitudmal and circumferential

- cracks around 40% of the inspected conneamas. A carcumferential cract was, for ,ata. ident(m' at Belleville l. The safety authorities stated that this crack "confinns the **imaar of defecss which can czone the rupture of the connection"." !adeed, contrary to the longitudinal credts, whid are enasdred to leak-before break, carcianferential cracks can lead'to the rupture of a tube without any "wannog leak *. At Saint-Alban 1 crzks were found on five oat'of the eleven acazles. .

The ia=aartiaa program has toen extended to all the 1.300 MW stacsors la opersoon and under con- ,

struenon (GolfecM. Al'. "e incnminated instrumentation noszies bad to be repaired within two years. In the meanume an antiejecnon devicz will be lawaw la die case of die reacsors under coamrucuan

' ~

cracked oozzles had to be replaced.

e 6

e I 9 1

I l

e e

4 9

e

- l 22 as.=a 4. n , s.tw. .w i. w i - n- m. m.m wAas.= ise

l.3. TECHNICAL DESCRIPTION OF THE VESSEL HEAD CRACKS AND THEIR ORIGIN.

1.3.1. Vessel Head and Penetration Design l

De CPO reactor vessel head is made out of ferritic steel (16 MND 5) and has an inner mainta=* stel cladding. j It has 65 VHPs (or adapten). De CRDMS are made of stamleu steel (Z2 CN 19.10). De upper pan of the VHP is also made of stainless saael which is aaaa~wt through a bunetalbc weld to the lower part made of .f

. 4 inconel 600 (NC 15 Fe). The VHPs are shrunk fined and welded Oncocel Ig2) onto the lower part of the vessel ,

head. heir external diameter is about 100 mm and the wall thickness is about 15 mm. On the inner side they' have thermal sleeves. De gap between the VHP and the thennat sleeve is normally about 3mm." .

-i IJ.2. Charactettstics and Origin of identifled Cracks The cracks identified until the end of 1992 were all longitudmal ones varyms in size and eumber with the dif, i

ferent VHPs. Circumferential cracks. as Frematome pomis out. "might be mese desnmental from a safety point of view".* Dey would be dermitely more dangerous because they are less likafy to leak befon: ruplure. _

1 The most 1!kely mekaaam to have caused the W is insergranular seress corrosion creding OG5CC). '

Framaiome suggests that a combeanon of the following specific candmaas could instisse the

-M= =-

-f l

+ high residual stresses indwwi by weld shrinkage ovahzados 1

+ annealing sensitization of the snaterial

-l

. operating temperansts under the vessel head of above 300* C j

• VHP n*54 geometry and Mh'anaa at the edge of the vessei head ,

~

• EDF and DSIN seemed at the thee (earty 1992) to agree with this analysis and the pra+=ninara role of sempers.

i tures. The temperature under the CPO vessel head is 315T whereas that temperanse in the CPY (CPI l

reactors (the 28 other 900 MW reamors in Pana) is 290"C. De 20 French 1.300 MW reacsorl 1.400 MW reacnors also opersee at shove 300*C (P4 and P'4 inshally 6-vv=ad at 319'C md the operand ar less than 315*C: N4 a 319'C) under te vessei bead and are therefore also likely to be subject to acceleratedl

~

IGSCC. nis dess not mesa that the ph====a=aa does not occur whos temperature ander the vessel hea is lower, but as Frunaan=* poissed out, 'cred initianon times are longer *. EDF used to maintain that thf lower temperature of te CFY stamars reenks in a four fond incnase in crad initiaison times. AAcr the det of a lary number of VHPCs wish CPY macaers, EDF was obliged to reconsider ks peulosophy on this m b '

-l .

o Mh'a*= cf cenas penetra6 ass also appears to be of smear Pw So far most of the severely cracted l

. -l

.)

.ABB RaaknarGmbH g-M s the annua! l

' 23 G. Bam and F.D'Asseca. 5pemene FragemeDenges att adP and ihrt L*-

j esaferease ef the German Asamforum.' ';W Sayammber tarts.tW2 j

'.;a Fmmmons. lenar to Femmasone Owear Oseap (Pit 00) members dead Osadier 22. IMt

( -

..= . . . . . . - . - . . . . . . - - .

1 1

I

. j VHPs have been identified at the edge of the vessel head. Esumations of te sacss on the VHPs have been too  !

low, according to EDF.* These esumations have considered the stress indooed by the weldmg seems but have f

- f not included the deformations produced during the welding itself.

. l The 1991 Annual Report of EDFs safety inspectorme gives an overview of the state of analpis of the cracking l mechanism by January 1992. Corrosion initianon in incon 1400 matenals was shought to be relatively weu understood hemw of EDFs expenence with sicam genera. ubes. EDF uses the foucmag formula to calcu.

late crack initiation tunes:

Tc = FcFm s.4 s EA/RT .  ;

Fe depends on chemical conditions (pH and hydrogen pressure) and is assumed to vary very Dale.

Fm is linked to the structure and composition of the alloy. Accordaag to EDF Sis factor can vary ,

considerably between 2.108 and 10' and "structurtl examination is accessary for a more accurate detersunation". -

+

t 8 su.M to the maneum stress on the surface.

T is the absolute temperature of the & , , . [

. EA is the activauon energy, according to EDF "most commonly evaluated to be about 183 kJ/ Mole which leads to about a doubling of the initiation time for a deasese in susperanse of 10*C".

Deformation parsmeiers enabled EDF to calculate the :.:reas at the crack to be 30% to 40% above the conven- l tional clasucity limit. .  ;

The initiation phase is defined as the time taken for a crack to develop to a point where its rase of furter for-maten can be considered stabic. In the present came te dept is considered to be about 100 micreas. But EDF also affirms that "the propagation speed of the crack is not aconsely known" and that "the e5ect of tempera-ture on propagation speed could be annBar so that on initiation *. !Jberatory tests have shows his speed to be !  ;

micron per hour exposure to 31TC and that

• corresponds to a throughwall pesecration of De VHP in about two '

cycicIs** A large degree of uncertamty subsisa, craci propagation speed vanes between 0.1 and 4 mm:rons per bour. In other words, the ersek could propagate auf5dently feat for a thr9ghwau ,w s i of the VHP in less than six months (obviously ha adag on the ked facaer of the rescaer).

EDF also mentions " doubts about the resistance

• of lacanel lg2 and incond-82 support saasenals. Laboratory l  !

- tests have revealed cracking under primary annimas conditions, whereas "this pheaameaaa has, so far, not been _

_ f observed la precs6ce'. ,

~

By January 1993, EDF had abandoned the 1 des of the pr= lamia =ar sarl- of temperstare under the vesset head on crack initiation Ihnes. la its report covering te year 1991. EDf4 chief inspecsor for safety published a uble of calculated risk facsors based on temperature and equivalent operatag hours for vmoss French and for- [

f eign reactors in ecer wtaan with the Bogey 3 reecree,(see able H.1.1.) EDFs own taaparw= results over abe 9

+ .  !

25- EDF."$6md NocMam 199F.Rappan de rinspeauer General peer is Samut Mac66mmJanuary 1932 ~

as ' unser am nem r.o at=ca's cydes,ihas anos eso yuan.nst is as== 5.5 == paressoas is espe per yer M MarTDM

-. _ . . -.. .= . - .

i year 1992 strongly contradict with those assumpnons. In the next report, covenng the year 1992, thelI inspeciar adouts that, he had assumed that 'the stress level and the made of the maaerial'wer idemoca least for the Freed reacsors. De results of the iay%s camed out in 1992 anfortunately show that these l j

- last two parameters play a much more determining role in the initiauon and progresson of the M.an than operating time and tempwature*. i Table 1.1. indeed clearly shows that opwating temperature under the vessel head and operanag life can hardly j

be considered to be the major facsors triggenng the VHPC phenameaaa Nevertheless. EDF considers that "although it is art possible at present to evaluate the thee necessary for the i onset of a crack, k can be rumsonably assumed that the crack propaganos speed resnains seesibly consant*. On this basis, EDF has evaluased. *with a high degree of certainty not to go wrong *, the propaganos speed ofideo-tified cracks (0.3 to 0.5 suamos per hour or 16 & 4,4 mm per year). Past expenesce soggests that this "high degree of cenamty* is very open to criticisst l

A major problem is that manufacsunng promdures for VHPs vary considerably. As poseed out in as EDF doc.

I ument : "Anention is drawn Wally to the fact that the welding conditions wert aerther specified nor docu-mented and no special anention was given to the deformanna of the ad5mers"." k is therefore higtdy unhkely '

that any pertacular aspect of the u-%g process will ever be abic so be hakad to the prohaklity o

~ -

occurrenceof aVHPCprobissa.  ;

. \'

l

. 1 I

4 C

j

~

.i I

3 27 .v.

• 5 anni secMars 1992*.Rappsst de I'lasposeer Otedral pour is 5ared Nedeaut (Tangry Raporth Jassary 1993 28 EL ,*14 cave de seems.r".undsend

_ _._ucac~r:ma

w u ,.a m w a ao. , m an um m e e .

I.4. SAFETY IMPLICATIONS OF THE VESSEL HEAD CRACKING -

THE FRENCH OFFICIAL ANALYSIS When VINC was first discovered. EDF was not unduly pe*W la its December 1991 issue the internal mag-azine le Mr Efterrique declared : *One thing is cenmn : the cracks do not threasen the safety of the inmrt2 nons", And the top technical manager Jean-Pient Mercxr is qucned as saying: *Since the matcrity c(the cracts i

' situated at the intenor limit of the wisej bead and since they are bngitudinal, they can never lead to tube rupuh The uncorftesed devebprnent of caracnferential crt:ks could bower lead to rupnare*? Just before Chrucnas 1991 Mercier old reperiers: "We can say in the clearest and firmest way that there is no safety problem *?

.EDFs chief nackar safety inspector's annual report 1991 is more prudent in its final paragraph ce th

'It is too cerly to mm-Inde on the extent to which the safety of our plants is being affected try what is the mo imporunt event of the year 1991 for our reacsors*.

The safety law considered that "cra:iteg could eventually result in the cyme of a control rod mecha-nism. Tbc penetrauon which is welded ecno the claddtag, could then become &tW from the vesac! bea Control rod ejecoon is obvously considered a very serious risk.

l EDF safety analysis is based "on the low prnhaMty of dangerous cracking and oc earty leak Mina prior to any adaptor ejection risk. De analysis of the conaequences of the ejecdos of a rod is thus '~*paauhle al long as the anti ejecuoo devices iare act nn'W4 os all the rerinr1 presenting a potential nsk. This an

' should be serminated in Damnhar 1991*. Nothing has been p#d_w s o far on the results <*_M ,

ne 1992 annual EDF safety laer"*nrate report merely states that control rod ejection is a design basis acedent. Otmously everyttung should be done to prevent it. *Arn4*ar prevention is essentially based on the f i

understandings of crack prnpagerun speede

• A sure ngcrous acedent prevection struegy is ara 4ed I

VHP leaks also have c4ber implurhe. As pointed out by the EDF citief safety inspeczorate in last years annual report. such a leak *can lead to the depoart of tvweatrated boric acid around the cracked penetra and to the very rapid corrosion of the vessel bend steel *. In fact the outer surface c(the vessel head is not cov-cred with steel. Tests and emperscal data from the Amencan Turtey Point 4 and Salem reactors in 1987 rev corrosion speeds of between 5 and 12 cui per year! Under these csraunstaoces reliable leak desecuoc sys-tems are otmously absolueeJy --aral- .

l The safety authorities (DSTN) mantain that VHP cra:iing "does not )perdize the mer+anical res: sta the affeced penetradon's'. DSIN also posats out that *cven in the hyp*at case of one c(the defects lea e a coolant leak in the pnmary carcuit on the outer side of the penetration, the operabocal procedures w albw operascrs to detect k and to take counter-tnessures without tbc safety of the plant being affe l

The DSIN 1991 annual report contains hardly spore than a page of text on this issuet other than the up general informauco en its Michel electrook server, little further pMhad informanoo has been m 29 IDF.I.aVw Decmque.D--=hr 1991 '

3o Nedeonics Week January 2.1992 31 EDF.56tei Neckart 1991.Rappan de rimer-* - Otaeni per la Sarnt Nacideus 32 D5IN.MAGNUC. ILit.91 at 21.9J2.as ==D as spdated verwoes ame3 Marca Ia 1993.

I i

I.5. COUNTER-MEASURES ADOPTED -

Rcmedial Measures Currently Taken or Planned in France 1.5.1. Inspect!on Tbc main elements of the inspecoon program have already been prescoted in chapter 2.1. Afner the deseaco of VHP cracbng in a 1.300 MW reacar it was dec>ded o exeM the inspecuan program to the older 1.300 MW reactors. The surprise discovery of VHPC in a CPY rexsor led no a compkse revisicxi of the inspecnoo program for 1993, the new program was still being enmmed by the safety authoruaes at the end of February 1973.

The suMe: demand for scenere VHP inspecoon capacity hrewe of the time conurmmg and dose costly characenstics of manual mspectico prem+ res, bas led e a rush e develop apprognue robot Wogy.

ABB Reaktor GmbH eveloped in 1992 an inspea>on robot for the Swiss utility NOK (Nor+whweuensebe Kraftwerte AG) which operates the two Beznso reacsors. By the nuMle of September 1992, one yer after the detecuon of the problem at Bogey. ABB had ahtady used the equipment co sit French reacsors, two S=us plants, two Ringhals rexsors to Sweden and the German Otxigheim PWR IXirms October 1992, au the VHPs of the Belgian Tihenge 1 remor were ineersad According to the punt macatement coe penpberal penetrauon (n*56) had a crack of about 5 mm neogth and 1 mm deep? AnoGer foar renacr1 were iW by ABB until the end of 1992. (see part D for furtter desads) l According to an overview established by Nucleccies Week Framatome camed out its Grst inspecuoo in September 1992 on Bugey 2 wxb its ow1: rotxx. EDF used a robot at Begey 5 developed together with ACB (Ateliers et Chantiers de Bretagne). Westingbouse aM Labcrelec have just put ooto tb market a fourth new l inspection rotxx.

l I

1.5.2. Leak Detection Systems In January 1992 a specul leak detecuan system was inetw fcr the first time on the reactcr vessel bred of Bugey 2 and later on the two Feuenheim rextms. However,sabsequent qualincs.uco procedures showed that  ;

the performance of the system was lower than expeced. As a resch. DSIN 1:msted the resort and operaung permit for Bugey 2 to three tocoms.

On May 7,1992, a new neak Me symem was iav>w Accordmg to DSIN, performance was up to capec-tsoon. During the same mooth the new system was ine M on Bugey-3 sext luer oo Bugey-1 and 4, as well as on MM-2. F W1 was to be equzpped dunnt shat down in autumn 1992.The reaaor was shut down on Ocsober 31 for 3 days for as temariw tonerveooco on the leak earra rynem. Btsyais-1

• as equipped with special separate leak WM systems for each of its three cracked VHPs. 'Ibc restar actually got its liceoce e restart and operate for four monos with tbc three cracted VHPs.

33 Pocmal ccee acacce, Titage pmr plas 4.re:urst, Nnaber 24,1992 a

?

~

, {

... . . . . . . . . . ~ , . . - .

i The same systan as in the case of Bugey and Femenhese is to ine instaued dwing refielling or otty planned outages on the 1.300 MW reactors Canence-l'. Flamanville 1, Paluel-2 and Saint-Alban-2. Another type of leal' detecuon system has been installed at Paluel 1 and .4. Qualificsoon and iesuas of the system is under  !

way.The new system has also been installed on Saint Alban l. l 1.F,3. Control Rod Anti-Qection Systeen ,

The six Bugey and Pe<waheim reactors are the only reactors in Fraam wtuch were not originally equipped '

l with a control rod and-ejecuon systan. Accorthag to the DSIN, sack syssens are now " progressively

• being installed on all reacsors. His is father proof that the risk of control rod specnon is being takes senously by

~ ~

EDF and the safety authannes.kThe taliabdity of such systems. lasaun$ sher the original demgn stage of the ,

reactors, is questionabic. No further.nformanon has beca avalablesa abe nachaw=1 fessures and renabday o the systems.  ;

i 1.5.4. Temperature Lowering / Downreting The temperature under the vessel head of a about half of me 1.300 MW rencaors had been lowered by the mid-die of Sepiceber 1992 froen 313.7 to 309.4* C. Accc ':ag to EDF, only about two days work is involved." The ,

loss in power output is dif5cah 'to ascertain, but prob.uly of the order of"a few MW".

I.5.5. Penetration Replacesnent.and Repairs ,

f

. By early November 1992. the VHP which caused the leak 'at die Rosey 3 rescaor had been replaced. The {

extraction was qualified by EDF tocksical staff as very delicane", becanae they shnak Atting had been l employed and the process had to be reversed to exuacs the tuben. Also 59 of the 65 VHP have thenal sleeves )

which had to be extracted before the penaratice inett could be resnoved. Four VHPs have been repared at l Bog. 4 and one at Palmel-4. EDF has also asked Framstonne to replace Incomel 600 penetratim with .;

lamael 690 coes in the vessel heads under construcace for Choar-B1 and 52 *and to bear all the responsabil- J ity, including fianaeum1"."

. . j 1.5.0. Vesset Head Replacement EDF will be replacing the vessel heads of all sis Fesseabeim and Bugey reactors within three years.

ProvisionaaDy the vessel head of Regey J *ias been seplaced with an identical vessel head made for the Spamsk f munned reassar which has arvr es cosapleand.The first og rapi=~====r vessel head is to be estaDed by spring 194.The old vessel heads are no be mored on sine for at least 10 years to cool off. before being " sacred la containers * - whatever that meana ' At least another six enspecafied reacscr vessel heads are to be replaced. EDF has also already placed as order wish Franssone for the consenscnon of a 1.300 MW vessel head.

34 D$IN. MAGNUC. 4 ' -28.1992 35 W Week. Sepamber 24.1992 36 Nucleoecs Week.Jammary 2.1992 37 EDF.1.as Coevertise de Ceous. published in Emerpresse. March 30.1992 1

l P

i r

. I.6. CONCLUSION I

In September 1991 the vessel bead penetraboo cracking (VHPC) phenomenon was identified for the first time on a French reaciar (Bugey 3). The problem was immediately taken aenously by the operator EDF and the French safety autbonties (DSN). It was clear fnxn the beginning that the rupture of a VHP would lead to a loss of coolant accident. which, even if contained withm the design basis of the plants, could have senous con-sequences for tbc safety of the reactor. Inspectioc was soon extended to other reactors of the same type (Fruenheim and Bugey) and laser to the 1.300 MW reactors. At fnt,1stle anention was paid to the second senes of 900 MW reaciers (CPY), since they were thought to be less sensitive to the pheanmenon because of tbc lower temperature under the vessel head.

By the end of 1992, as a result of severe VHPC oc a CPY reasar, it naally bacame clear that all the French I FWRs were potentially affected by tbc problem. Wnb the recent Mecana in Deccznber 1992 of inapient ar-cumferennal cracking oc the first iramemad VHP extracted frocn Bugey 3, it also hemme clear that the t problem is not neeenarily hmned o locgitudinal (or axial) crackL Cirnunferential crackang could lead to the sudden rupture of a VHP wnhout pnar leaking. dereas longitudinal cracks are casacr to idennfy, since they usually leak before rupaart.

Analysis of the VHPC phenomenon and the amtude of the opersor and the safecy authority soggest that :

EDF and DSN were uupsi.ed to barxDe the VHPC problem, in spite of the faa that VHPs had already beca identifed before by the rextor constructor, Framatome, as one of the places most likeJy to be affected by stress corrosion crackmg.

+ The early inspecuons were limited to easy acern VHPs. During the first taooths follonics the discovery of l VHPC, inspecions were limited almost eaclusively to VHPs non equipped with control rods, altbougb the tnost severely damaged VHPs identifed did in fzt house control rods.

• Only 10 out of the 53 parad="y affeced rescaors have been mbyect to complete VHP inspecnons talmost one and a half years after the int VHPC dar><nna). VHPC was identified on 8 of these reactors. The actual I tesung techn> ques empioyed for ext indivwiual textor are still uccicar.

i

• locipient carannferecca! cracking was only ideoufied around one VHP after it had been extraesed. Extreme murr= sbound therefore be caernzed in interpreung the resuhs of in service-inspecuoc

. VHPC has been identifed on a reador (Gravelines-O inaded with plutonium-armnuun mixed oxide fuel (MOXj, = tact is of parucular signtficance, since the safety margins nunciatad with control rod ejecuoc are even narrower than those Ecepted for uranium fuel cores.

• EDFs crack initiation and propassoon analysis proved to be wroeg. By the end of 1992, it had heerry encar that the temperature un&r the vessel head and the operstag life-time art not the most critical factors. Stress I and the physical state of the vessel matenal seern e Anmime.The VHPC phaanmeana is, as yet. not well understood.

1 -

i

.. .. - . . . . . . . . . . . ~ . . . ..

• Perfonnance and reliability of controt tod anti-cjection devices and leak dewem symans are difTsuk to evalusse. Vessel head replacernent is a lengey process and k is not known whether VHP repairs are rehable.  ;

t

= Reacsor estages due to VHPC have reached the equivalent of ihree cmunulated reacsor operadas years (1,209

. days of elecencity generation). The power replacement costs can be evaluned at more around 8 bdhos French francs. (see annex 1.1.) '

i t

la view of the lack of detailed knowledge available on the VHPC W and the fact that existieg inspecuan and leak desecmon devices have not bees quahned and are of enknown rehabGity, k is surpnang .

I abat French safety authorities grant operadag linmans for reactors for which VHPC has been identined and where nocrimmaamd VHPs have not been repaired or vessel heads replaced. [

French safety sothondes have bowever sequessed the reactor bauder Pennename to replace the tecorel 600 VHPs r3 reacsors under construcmon with !acceel 690 VHPa.

It is now c$ ear that VHPC aarmws safety margins to a point which is act conskland P for new reac.-

ears. Beanng France's increasing ' ;='== on nuclear power for elecencity prodscaion in mind, one wonders j what level of genene safety problem, if any, could lead to the decisica shaidag down reecaers (temporar0y) on -  ;

a large scale? What level of freedom c(h'a is lea to the had of safety amiharines la this country? . .

2 6

^

I e

i

. . .j t

i i

. l l

. t f

i i

. +

)

s

- . , _ - . - .. m _ .. .. _ . . - . . , - _ _ _ . . - . . .-- ._ .-.i

. . em ... -.a-.w..n --,

e

- l i

ANNEX Li.

Table I.2. : Reactor Outages due to Vessel Head Penetration Cfecking in France ~

(by the end of 1992)  ;

dates shut down time (in days)

Reactor ,

Jan. 4 - Feb.1.1992 27 7 Bugey-2 Bugey-3 Aug. 5.1991 - Aug. 281992 (ext.) act 241 l Oct. 5,1991 - Oct 8,1992 (ut.) met 305 Bugey-4 .

Bugey-5 Apr. 25 - Sep. 22,1992 (ext) act 87-  !

Aug. 8 - Nov. 3,1992 (ext.) met, 24 Cattenom 1 Fessenheim 1 Aug.2,1991 Apr.11,1892 (ext.) act 126 Oct 31 Nov. 3,1992 3 i

Feeeenheim 2 Feb.2 Mar.14,1992 40 Flamanville 1 Aug. 8 - Nov. 25,1992 (cat.) -act - 47 Apr.18 - June 29,1992 (ext.) act 9 >

Flamanville 2 Oct. 31,1992 - 7 7 j Gravelines-B4

' Paluel 1 June 6- Ang.29,1992(eit) act , 21 j 4

Sep.12,1992 - ? - 7 ,

Paluel- 2 '

Paluel 3 Dec. 21','1991 - Apr. 4,1992 set 42 Paluel-4 Mar.21 - Sep. 6,1992 (ext.) . act 92

^

Saint Alban 1 May 22 - Dec.16,1992 (ext) act 145 Saintl Alban 2 Sep.19,1992 - 7 7  ;

Trica. tin 4 Nov.21,1992 - 7 7 Total met > 1.209 l I

l Nones : art. eunds for erwaded planned emage; ens est Egwe servesponds so es ember's emmanas of k om inesass is maage emus for l ime aa. repas. . ,' r werk en sussel heads. Pa61 a are based en ens fe5sweeg amenderwuses : Aiwage dessnaani anese-l Laca cperauons a eswassed try EDF : 21 weeks; Aswage afunfEag and pumal espasses : 9 weeks.

1 Assuming power replacement costs of roughly 0.25 French frtacs per kWit (based on the reference cost calev-lations of the French Ministry for Industry)", she total loss due to VHPC related ouages until the end of 1992 ,

- is already at least around 8 talhon French imacs.

f

.h

• i k

i I

' ~

38 Referesa masu for operunas and feel enty for a 600 MW son! And power plass, aparedag 2.000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br /> per year were per er 26.3 {

to31.7 assumes per kWh. Minusess de I* lad === *Reppen ser les estu de sedisumes de la psodmance (W (engme ust.

~

- migos", Seynamber 1986.

~

i PART 11: Tho intornational t'orspscuvo  ;

II.1. INTRO: FRENCH DISCOVERY OF VESSEL HEAD PENETRATION  ;

CRACKING (VHPC)

On Sepu:mber 23,1991, the French Bugey 3 pressunzed sst:r textor (955 MWe) underwent a hydrorest, a test to be camed oct every ten years accordmg to French regutarinn With the reacsor pressurued to 207 bar, a leakage of about 11/bour was detected by acoustic incans in the vessel head, and subsequent ernmmarion revealed a through-wall crack in coe of the vessel head penetrations (VHPa) that al*ow passage of the control rods into the reactor.

Sooo after, with the discovery of further cracked VHP in two other French PWRs (Bugey-4 and F~Mm.

1), utility EDF rntned it would have to face ano6cr geocric probiczn of senous th% doe to the stan-dardizanoo of its nucacar poner plant popula ion. Boch EDF and Fr-ry cbc vessel manufacturtr, hastened to have their tectrucal teams analyze the problem, and an expensive MW and repair pro 5 ram was initiat-ed. A detailed descripdon of the development of tbc w:ase! bead penetranoc crachng issue in France is given in part 1 of this report. .

By the cod of 1991, EDF fawnr General for Nucicar Safety Pierre Taaguy presented his annual report on Nuclear Safety for 1991', inciading a first explanation of the observed cra:t W as satas.correcco cracims (SCC) of Incocel 600, s' high-nickel alloy. Aho beluded was the derivaban of the relative risk u develop umitar VHP crach for a number of PWRs outsade of France, being based solely on operadng time and ,

vessel under-bead temperature-Table 11.1.1: EDF's ' Risk Analysis' for fors!gn rsactoes from late 1991 Equivalerrt Hours Temperature Tamparature Risk compared Ptarrt (to 3uty '91) *C Factor to Bugey.3 Yankee Rowe (USA) 189.670 270.4 0.033 0.09 Ckooee1(USA) 112,800 308 0.63 1.01 Tihange-1 (Belgium) 111,968 312 1.34 Ohi-2 Capeo) 74.300 316 1.06 1.15 73.900 3183 1.24 1.33 Nonh Anna-1(USA)

Trojan (USA) 67,890 316 1.06 1.05 Doel-4 (Belgxas) 44.156 3213 1.48 0.97 Tihange-3 (Belgnan) 47,089 3213 1.48 1.02 j Altbough world. wide the number of rescaxs wie VHPs made of Incooel 600 even is far higher, no immrente accon oatsde Fraoce was triggered. On the cactrary, pcblicanon of the EDF results by Nucleocics Weel' caused cocstemation smoog unlities whose pr. ants were named in the EDF report,' denying the eussence of an

• Ma* tafety cooctrn. ,

i I

1. EDF. Storud NocMarr 199 t. Rapport de 1 Inspease General pow la Servat Neckar

2. Nede:=:s Wed Jasonry 23,1992

3. ' Nh Wed. Jaanary 30.1992 l

n .

3__

e.,a m. mwns pr.un.*

II,2. DEVELOPMENT OF THE PROBLEM IN OTHER COUNTRIES'

!!.2.1. Sweden In Sme&n t!ree pressurued wza n:acton are currently in & at Rmghals, aD of etich are of 3-bop Westingboese design, and all of stict empby Inconel 600 for the vessei bend penetranons. bghals 2 (880 i MWe) started operation in August 1974, Ringhals 3 (960 MWe) and -4 (960 MWe)in hymher 1980 and June 1082, respecsively.

Accordmg to the Swedish Nuclear Power famererste (SKI), Ringhals<perasar Vanenfall did " intensively s:udy tbc problems s$entified in France with crachng in the VHPs',' FoGowing the French study assoctaung the nsk to develop penetrauco crachng with opermang time and anda-bend temperature of the reactor, the respective da:2 for hghals 2 (111.799 bours to July 1991, about 320 *C) suggested a high ausoeptibihty to VHPC and it was coocluded that crachog could not be **ew j VHP inspections, bowever sere not camed out until tbc routme ountge of bghals-2 beginning May 7,1992.

The inspecuoc of twelve unsleeved penetrabons out of a total of 65 penetrarma revealed four cract indica-  ;

tions *ooe crack of 16 miIlimetres in length, ooe 4 mm deep, and one 2 mm deep. Cracbog discovered in a j fourth penetration couldn't be properly sneasured",'sa Nthen Week reponed. ,

SKI's requirement e inspect the skeved penetranons as well could - due to ina$ equate mspection equipment -

not be satisfied until the beginning of July, the reacser remaining 6 awn that tkne. FinaDy. oc errnpimon of the Ringhals-2 inspection, one further crack WW in a sleeved gs .Jos was decerned Of the five cracks j

. found only two were repured by crouve removal, the r=&* not being ermeriered as safety significant and j i

thus left witbout further acoon * .

On July 19,1992, SKI gave permtsuoo to restart Ringtals-2. The operator Vartmfaf1 was required to u*rnit a i plan for fuure inspections of penetraoons at Ringhals-2,-3 and -4, together wkb an a: count of a!! reactor coco- !

ponents employing Innmel 600 and a safety analysis, not.lara than by 31 Wher 1992.' l An inspectico of the hghals 3 penetrations was scheduled daring Itr routme annual outage from June 5 July

11. Only coe crack indication was found, which was no longer drun=hte after pohshing down 0.3 millimetres.

hgbals spcumin Goes:a Larsson was goosed by NtM* Week to nwnment 'they believe now there i was no crack. that it was something else".'

The inspectico of hghals-4 vessel bead petnetrabons was not camed out notil its planned rouune a-"ul maine outage from September 1 e September 30. Two crack indicaoocs were detected; both, bost .a.

were locaEzed at a 6ee below the weld jointng the pencersoons to the vesaci bead and SKI accepr.ed these e be left without further measures antil oczt year's in& ' .

4. SKI Kvanata Rappan. Forza kvinnast 1992

5. Nha Wer.k. May 21,1992

6. Nedcar News. Sepmaber 1992 I

7. SKI Kranats Rapport. Andra och tnd,a kvartalet 1992 L N'h Wed. Jane 25.1992

9. SKI Kvarub Rappon. Andra ed tr%a kvanaist 1992

.- ~j l

l b

~'

By February 1993, the Van = fall report on the Rinhals findings was subained," sassag that there had been four cracked penetrations left in Ringhals-2 and two in Ringhals 4 with the biggest crack Da penetraban no 68 l of Ringhals-2) having a crack length of 21 mm and a depth of 10.5 man. Based on acMalled conservative l assumpoons regarthog crack growth it was condaded that leurvention would not be escassary before 1994 at

_ {

i two of the cracked penetrapons of Ringhals-2, and .4 respecarvely. Why these assmaphons cannot reasonably be called corservative is included la chapter 11.4.1.1. t 11.2.1. Switzertand ^

l i

- Of the three pressurued water reactors curitetty la operation in Switzerland, the two 2-loop onics Bezaau l (364 i MWe) and Remam2 (364 MWe) were built by Wesangboose a main contracsor, and bot have heir consul rod vessel penetrances made frons laconsI 600.Dey went imo opuntion in Jely 1969 and Ociober 1971. napetzavely. , l In R y 1992, stility Nordostschwenartsche KraR=eske 040K) asked ABB Reaktor from Germany to ' ,

l develop an autnaarM inspecuon strategy for the Beznau vessel heads." On Jane 1 Bezano dsector Hans ,

Wenger announced that "we found no abaannalines" aner taaparaa== on the Bezasu 2 vessel head had been ,

1 carned out. Only 75% of all penestuoms had been inspeasd, indsding te analeeved and putpheral positsons, according to Nucleonics Week.58 However, ao safety si=**= was recognised by HSK, he Swim Naclear Sarery Inspectorme, ane peraunion given to nstart the reacsor aner its planned outage time. ..

Ya*a* of the Bezano 1 vessel head took place dunag its ansaal revianos,trous Jaly 3 to September 1.The aum==sM eddyw:urriet test (see chapest 3.2.4) sevealed two ancks: *cae messering 23 minnastria long and I aun deep, the other 3 man long and less than i aan deep, amar es wekt of te poetramos"." heet ancks were reponedly repasred (probably by erosive removal) and the ask want back ime operanca on September 1.

When questioned on 'the fia&ags la the Besano 1 vessel peansrahams in October 1992. HSK officials stated l that further analysis of the problem was necessary and naaripanad but at the same time ao umante safety smoru.nnna could be idennried =

i 1

11.2.3. USA It is of perucular interest that op to now Gaa.1993) ao O for VHPC have been reponed from the 1 USA, altbough the maionry of PWRs Ekely to develop VHPC accordag to the canent stase of knowledge is l U.S. based and as carty as 1989 tannmal 600 pressurizar aosalsa in US-PWRs had been fomed to exhsbu seess '!

-I conosion aatting. Narianair Week reponed in September 1989: PSG&E's Don Graf, proyect leader on the pressuriser insee, said some lacanel 400 penserances, sech as la<cre instrumentation p.; css and control l rod drive penetrations have &ameters of four to six-inches or langer. Circumfercanal breaks of the lococel.

600 penetradoes in those " big holes" would indeed lead to leaks that would be beyond the make-up capacity of the charging pampa, Gsaf self'."

- 10. - vansaran tashnis 2 ess 4 Rankaarunkar Spockees i 4% + .

• GEK M3. cam m

11. Baro G.d'Associ F: SpeneDe FragemeDenges der stP and ihn LAsenges. Hartenegang Dt. Anoedonna, Kangswimar,14/15.

hgn = h- 1992

12. Nh week.Jens 4,1992 ,

13. N.,a=== Week 0maher 8,1992 .

14 Peruanal communicsson.Pener Messinger.GP Swsanstand

-15. Nedensees week. Sepasaber 2a.1999 90&E is Sainsmass th=AN== Ca 4

21 assa smaa sewin.,is ennemme a esmana saws es

C Gen e. We W.C. Gr.e *t's.1ecpwe .

Following the nra reports on VHPC in France, 'for the most part. U.S. reaction to the pretuninary informaoon  !

on the Frisch development has been muter". The U.S. ooclear regulatory . ~ .. . (NRC) took a sandar  !

starse, spessing its belief that the cracking was not only h= bat on operating hours and ander head temper-ature, but also on other parameters like stress, manenal conditions, and manadarsuring processes, so that the P French risk cxxnparuon in EDF's 1991 annual report wm not drectly appbcable. Fanbermore, NRC's Thomas Morley said that "a double. ended failure leading to an unnotable loss.of-coolant is unhkely"." The possibdity of cracks in the VHPs going unnoticed during a bydrotest according to ASME apecincations (which only require a 110% overpressure, nahke the 125% in Frana) was acknowledged, but no safety concere seen.

Even after further dtscovenes of crackesi penetrations a ' European PWRs no concern was espressed by U.S.

utilities, these following the prob!cm via the Westiagboose Owners Group (WOG). However, by May Westingbouse analy.ts had found out that "we do have ===dar masenals and aanDar operating conditions, but there are a lot of paramesers".*

In September 1992. Nx% Week reponed that the NRC had SaaDy come to the erwhimics that the crack-ing could occur in the U S.-plants, but at the same time the NRC mm% the problem of low anfety signin.

czace, because . .

. of the fact that no U.S. plants have reponed cracks orleakage; (

. no leakage was foand during the opersnan of foreign plaass; '

. if crn&ing has occurred here it is probably axial and not die more marious circumferential:

.- leaks would probably be hd before a large faDure of a W A l (and fina!!y. diat) t

. even if a failure occurred, the amaD break loss.cf coolant accklent that would result is wuhan the design basis capabdity of the plaats."'

• A discussion on the appropnateness of these assamptions is incioded is chapter 4: Discussion of Safety Implicadons.

U.S. vendors expect the NRC to reqmre subnes to inspect their plants for VHPC in the future and thus are busy to have equipment available. Babcock &Wilcos Noclear Services (BWNS) manager George Beam expec:s nrsti aa~a% in "cae or two lead plants" to be carned out in spring 1993.*

II.2.4.- Other countries Given a high saaceptibidry ofInconel 600 to cceromon cracting as the root cause of the probiesns with cracked VHPs a vast sambor of PWRa world-wide would appear to be at high risk. However, re* ema of the interna-tional noclear community has generaDy been hule more than showtag laterest.

16. F-rhcs Week. Ncmober 21.1992

17. N=r4-naa Wed. Jassary 30.1992

18. N=riar='= Week, May 28.1992 ,

19. NacJeoescs Wed. **t==h- 24.1992

20. M h Week.Omaber 8.1992

On October 211991, air informatico note oc the VHPC at Bugey 3, France, was circulated within the Fr-arme Owners Group (FROG)" to Belgian Dectrabel. Soud Afrkan ESKOM. Chmese GNPJVC and Japanese KEPCO. Up to now, no informauon has been made available on whe6er inspecuaes or any action have been carried out or are being planned at the Soud Afncan Koeberg-! and 2, the Chinese Guangdoog.1 and -2 (under construcuoo) or the Korean Ukhm-1 and -2 units (wtuch were also addre the mformation note). .

Out of the Bej g iao Pw7, populauca, units Tihange.3 and Doci4 ocre i==% both of wted have beco in openocx) for relatively li::le ame (since June 1985 and A;ril 1985, respectively), but are runneg with sel undertead tem;uh (the outlet tempersure is 330.3 "C for boe units). In EDF's " risk analysis" for French mod scsne foreign units as wc!! (see interrhm), PWR Tihange-1 was lisaed as having a risk of develop Ing VHPC higber than Bogey 3 by a fusor of 1.34. f=.-; n' -w howcwr, were not w+=4M before the October 1992, and according to plaot management one crack d 1 mm depd1 and 5 mm leng6 in a pertpbe penetrar. ion could be identified." LWals of moo 6er four Belgian plants' vessel head penetrabocs w design are in operauoo).

be planned bercre the end of 1992 (in Belgium a tacal of seven PWRs of W~hh Inciasion of the Japanese Otti.2 unit in the EDF tist of rencsors at risk did not trigh bsp onit; r 6er, a sphan of Japan's KEPCO was gooted in Netmaics Week as saymg "we have ao plan to

• launch addithat investigations".* In September 1991 bowever, an ABB nF='al reponed that "the company has also sold a complete set of CRDM in& h %m ==f to a Japanese firm, wide has since used it to essmw penetrations oc a Japeoese PWR"." The spec Sc she was not W

• Sis of the seven Spanish PWRs are of Westingbouse make (Almarez 1, first power: S/81; Almarcz 2,10/

^

Asco.1,843; Asco-1 IQf85; Jose Cabrers,7#28; Var *fh-187). A Itmited inspecoon at the Asco-2 and inspectiocs at the Asco-1 and Almartz-1 ocits appear to be scheduled for their toutme outages in Fe fuse and September 1993, respectively; dy* mag on the rescits of these enmkhoes, inspecuans at the other units are said to be cocs>dered as wetL* An ofreial of the Spanish safety satbority (CSN) was quoted saymg "We doo*t see any reasco to advsoce NRC on this hos** in January 1993, with regard to VH In Germany, Incooel 600 is employed for the VHP: in two PWRs W Obrigheim 057 MWe; start of opera in October 1968) and Malbetm-Khiich (1302 MWe, March 1986), with the facer currectly being sbut-down due to licensmg dencuencaea. By January 1993 it was reponed that two han rescars had been inspec VHPC;* the resalta, however, were not made public. Past expcnence wie VHPC in Germacy is desenbe below (chapw 2 5.).

h 22,1991

21. Frm . Frassarme Chrem Group takrmne= en Bapry3 IW

22. Persona! cornmarmeeve Myde h. Paas

23. New 5'eci.Jaaaary 30,1992 24 Nedeoca Weel. WM 24,1992

25. Encrprase No 5763,16.2.93

26. Nedeocics Week. Jaanary 21,1993 27.

Eero G.D' Assean F.Rytander L: Meeces the apper.vesacneed chassage wick eddy came and 11T; N

! 23 vines. .u. = cum. c massa mancv

% :a. anna n. en u i

Bruam has no PWR in operabon, but Sizewell *B' (1258 MWe) is 'under consduction and expected to deliver  !

frst power in 1994. The stacur vessel has been manufnamed between August 1982 and h 1990 by Framatome in its Chalon shop, the shop that made the rescer vessels for the French nuclear programme as -

well. Although incorporaboa,of steam generator tubes made from fannaef@0 lassend of 400 was reported for Sizewell 'B', it appears likely that the VHPs have been manufacared bom locome! 600, as sa EDF official in January 1992 was quoted saying that quahficaboo of welds herseen penetrations of famael@0 and vesse! ,

heads had not yet been carned out. rna.% that a high reactor oudet temperature of 324.8 *C is planrad  ;

for Sizewen B' operation, this reactor might be at high risk ddevelopag VHPC m wen.

The following table gives an account of the occurrences of VHPC that have been dessved and made publicly

,. 'available. It must be n,W so further increase over the fouowing ana#** and years, a inspecsion sche- p evolve and data from reaacrs outande of Europe win probably also hannme avaRahie-

. .t Mest power Met output eartiet Number of WPs with

, (MWe) temperatuse erseks detected Sweden:

bghals-2 5/1975 860 320 4(4!)

hghals-3 9/1981 915 (1) bghals-4 11/1983 915 2 .

Belgium:

Tsange-1 .3/1975 870 312 1 Switzariand: -

Beznau 1 7/1969 350 . 2 aw ends os Fmwa VHPC asastreman, ses shie !.I)  ;

P IL2J. Former M Expertence The French trypochens that anployment of f annael 600, together wft a high under vessel tempersare, is the ,

main cause of VHPC is not anade*mt with the expenence thei has been made with VHPC is the seventaes with other reacscrs. i VHPC has been found, amacag others, la the Scmet VVDt rescaer design, ed. in 1975 and 1984 VHPC was .

observed in she VVER 2 sencaor at Fwaah*'s. famner GDR, wist some cracts being through-waB.The dam-aged material was not Incomel 600 bot included' weld and base matenal of massealtic assialess steels.'

e e

2a. hs Bneisk FWit. Nedser Eagleserias lamrassanal $ penal h

. .. .,w - .. -.. ....

(0X18H22B2T2 and OX18H9, Sonet anmm.clanre). It was not nyeumute moodstructiw IHerva mapec. i tkm bat rather an increase af mdmmvity above the reactor vessel during operation that indicaed the damage." l

?

Sirmlar cracks had been obserwd before 1972 at the two VVER onits Novovoronezh-1 and 2." and the prob-lems of VHPC in other reactors of VVER type were sobrect of ruwndve invesdgadoes in lhe faaber Soviet Union and Fntw is ,

In the caperuratal bodmg water reactor at Kahl, Germany (Grst power in 19615 a leek in the prun was found during operabon that could be traced back to a strough. wall arcumferential anck in a control rod penetration. The latpecoons of all 21 penetraucas took several months and revealed indie of 27 axial and i

2 circumferential cracks."

],

c s..

f i

s i

Cd  ;

i i

j

~

i i

1 l

1 1

1

29. MaDer F (1976): Schadensbancan SUS-Sundrahrs %"nTR.2, KKWR.WPC 2206. NPP e w q

30. hid.

31. Tempmarry resis i datacmenom % perveds SUS blokov VVER (Tempanmus aghet and eyasudag Itis of the cleaness of VHP d VVE3t units);Tredy 2KT (Im) vyp.153

---_---w Kahl,

32. Ehsaur. Kinas (Imy Rapsesear an den Doesrsahdarcht.hnamesa des Rambordnsckgefases in Y

~

Renbanaging 1973 .

. 25 v amm = n . ma me n ise.

.cr.. a. a . - ,, _ ,

II.3. DESCRIPTION OF VESSEL HEAD PENETRATIONS AND THEIR INSPECTION .  ;

IlJ.1. Mr Design .

In a pressurssed water reassor (PWR), the chain reactice taidag' piam in the core is bodi modernaed and cooled -

by water, his is pressunzed so that it is not boiling although temperatures are typically around 300.*C, i Therefort, the core has to be placed ine a pressure vesset At the anske thme, the chain reacean needs to be l

contro!!able, which' is accomplished by so caDed control rods that can be laserted into the core. hey are of vital imponance for a safe operanon of the noclear reacsor.

Vessel head penetrations (VHPs), also called adaptors or control rod ditwe martamiam (CRDM) aostles, are 'l Sned into the twent pressare vessel head to aDow insernos and withdrawal of the control rods. Figure 11.1 shows the basic configwation of a penetrance thrtegh the vessel head.

De wall thickness of a PWR*: vessel head typically is on the order of aromed 200 mm and made from ferritie  :

steel, of which the surfaces in cootact with the enalant are cladded wkh semialema seest to beaer withesand j conossen. .

De ;;a .eces rypicaDy have an oster diam ***r of about 100 aun and their well We la shoot 15 me, in most PWRs they are made froen laccee! 600", a high eicket. alloy widt arosed 756 anchel contest. Daae se

, shrunk-St lato the vessel head and then manuaBy weidad to the bosomt side d the hand. his weld bears a5 horteostal and axla! loads.De lower and of the pressure housing of a CRDM is threaded and seal-welded coto ,

the upper end of a VHP, from the lower weld upwards the VHPs thus fann part of the reacsor coolant pressure l boundary. . .

As the nozzle-to-vessel head weld has to foDow die curvasore of the vessel head, these eahibit se acomotricity. -

the degree of which being Q- T-f =t ce location and she weld seem tachaarnaa relative to the nozzle axis being greatest for the aannes fanbest off center,"The weld produces aseves speases on these outer aan1** beadmg them slightly and defornung them to a slighdy oval shape"." -

i Most penetrations have thermal sle' eves 'made from stataless sanet to redace the thermal stresses that would be -

imposed on the penetrations by fast thennal transients.The sleeves are Sned to die CRDM housings and do not bear the pnmary circuit pressure. De radial gap between peneestion and therinal sleeve is about 3 mm: this ,

I width, however, can very as mock as i 30% due to the ovanty of the penetradoes." Furthermore, some sleeves .

are air anered due to the slight bend %elow the weld.De fact that most poseentmas hold slurinal sleeves has sinah inflanece on tie penetrances' lavaku y (see r aest cdiapeer).

)

)

1

33. Gerians renesar vendar KWU desse weih ths ==g==a= of the % spemar Un beve ymmetrutuus made hou famnc seel which are prisacsad by as imme.r samaalens sand tube and screwed isio she seemer vesad head

34. $sby EP. Bands WE: CRDhd PNsle i==r eda = Nedser Flam la====t Newessber@emassbar IM2

35. - ud. , J 1

m__ - _ _ _ __ .- ___.---~_____________.m _ _ -.., I

IL3.2. Methods to Detect VHPC 11.3.2.1. Leak Detection Ocariy, from a safety point of view, leak detection systems ce their own are not sufficient to tadde the VHPC problem. For leak detectaan systems to be effecslve is guaranteems that no lastanema== failure can occur. the leak before break (LBB) ooocept weald have to be valid under au possible carcann-stances, and this is not the case with VHPs, as further detaGod is. chapter 4.1.1' Ferdurmare, assurnace.

~

would be needed that leakages could not damage other components befort besag stopped, and no other hazardous contamination being caused. .

The strategy of relying solely on defecxive penetranons showag up with leaks before ruptmag has also

' bee.n rejeaed by nuclear edustry officials, though not adentsag the enfety risk, as "industnaDy enac-ceptable because of both the ascertainty involved and the long outage thnes it weald haply *,G.e.. -

replacing the flawed penetrations).

Nevertheless, leak detection devices should la any case be taes=HM as a - i'-- vy safety measure. l After first anempts of EDF to installleak enwaan sysseums in their older reassors failed due to non- ^

approval by~ DSIN, EDF developed a acw system based on analysing the altroge 13 onament la the atmospbert around the reacsor vessel head that was accepted by French regalators. la Sepamober 1992, l k was reported that EDF was planalag to have 22 of these units instaDed whble a year,but only at $5% l

. of its 900 and 1300 MW PWR unit ===P to the penetranca cra&ing probiam'? It was not dis . -

closed why and which 15% are lea wkhout these leak A*w*= systems. , ,

1 11.3.2.2. Hydrotests As taentioned cariar, the crack Metad la a VHP at Bagey-8. Trance, was detecsed durmg a hydroicat. '

French hydrotests at camed out a about 125% overpressert, wheetas the ASME(American Socacty of Mechanical Engineers) code only requires 110% overpressare, a code applied to most of the Westinghouse design reecear pressart vessels.

It has been argued that she' leak Maread at Bugey 3 could caly develop because of the empressure caus-ing a probably small crack grow to throngb-wad dimaaelam and that Freed overpressere regaraments l' abould therefort be telaxed. Sock proposals, howewr, have been put off by the French safety authorities.

It is likely that VHPC might them go ==aawad and thes an *tsaportant edacanon of prunary cirtait

- integrity

• be gives sp. la any case having a leak develop during a hydrosest under controued conditions  !

is preferabic to the aaaaa8*"y dasastrous consequences of such a thing happemag in operaban.

WhDe k is fue that hydrotests may reveal some flawed P in the prunary cdromit, and a higher ovupresaart wHI be mort parveen%l In doing so, passing a hydrotest data by a0 masas give assurance of the integrity of abe primary cartett. A Swiss expert groep assessed the ability of hydrotesa, ccanpie ~

E Neclecens Week. November 21.1991

37. N-h Week. Sepenher 17,1992 St. Nh Week.Futramry 20,19P2 .,.

27 seesus =em m.e annmaeas e

c un ransw orw.c..

~

mected by techniques to analyze a::oustic c nissions frtxn cracks under stress, as ~si#cantly worse" than other inspectaac Mniques in detecting flaws."

11.3.2.3. Visual inspection, Uguld Tw L.61 Tasting, and Uttrasonic Testing Visual inspection of the res:sor vessel head during refuelhng is mmmm practice. This, however, does not mean that visual inspecuan is a Mnque suitable for crack #~+h rather, k is osed because of' its simplicity (and low price). Ihrect visual iva of the inner urface of Ibe vessel head penetra-'

tions is obviously impossible, and automatic inspect >on with TV camera suffers frtxn the resolution of these devices being too poor to identify surface cracts, their openings are in the range of pm.

Tubing surfaces of the primary circuit, incloding the VHPs, are all scss often covered with deposks con-cealing potential cracts. bh-ore, deseccon of ancis on a video screen is an extraorthnarily diffi-cult task. pucing highest demands both on 'w vigilance and expericace. A!! this adds up to tbc murument that it cannot be assumed that visualinspecuco is suitable to Meet cracks".

Urdt in-service inspecdoo conditions, k is also oflimited use to apply liquid penetrant iesting. stere a chemical is appued coto the inspected surface and np-vd to nec=mniate in surface cracks. After washing, sarface cracts would then be enhand for visual lae-+h A pr* mad % of stus techanque sa de effecdve, however, is that the area onder inve<rigerk genernRy has to be prepared la advanas. i.e.

surf *:e depasits be rernowd. Experieoce shows that ch=te=1 removal of deposits often fails due to n

their strocg adberence," and =~ hie *f removal appears to be latpractical for am-*g long or many pepes, the traer being the case with VHPs. As measurwnente on the prkaary ciremt bsw to be fully autocuted and regnotely-controGed, few ryssems have been 4.M c

'Ibe standard votametric in-service bi+t-3 mehavpe of primary cfrtzsit s - ,-=nts, ultrasonic test.

ing, cocM princtpally be used ce the VHPs as wcII; however, altr**mb inspenice of the VHPs from the outer surface wouM - espectaIly in the weld regscxzs and above - be geomeencally very difficuk. and especuoo from the inner surface is made impossible by the thermal sleeve.

'Ibe tacter of course also appbes to penetrant tesdag. Whereas both penetrant and ultrasonic techniques thus esnw be used for screenmg nu VHPs of the vessel (unless all thermal sleeves would be removed).

they can be of some ese .<-@=-3 to some A,occh e My in analyzing a fisw once it has been Meesed and toenth4 TWm like reinoving thermal sleeves and probably some surface treet=f as weII would be memury.

In Swo6en, a reinosely controGed fi-m penetrant syssem was used in such a man- to venty the Crack fladings of eddy currtet tests in the Ringhats-2 pennetrarren.* 'Ibeir results (see chapter 4.1.) are c5carty of bener quaiky than the eddy currtet results of the same cra1 stereas the eddy current tests

39. Prand 0; w edehohe Dnickprstang and schau-a-gang . se einer ook.euensches i

E3Pcw. DGZ5', DAOl4ahrestagnag 6.-L Mai 1991.1.saurs

40. Oppermar.e W. Kar.kel G: Anpassung sad Erprobung eines Rotr;mneopetfrystems far des regcInhiges Easatz bei Rol l

in W"W Sy=emen Bb(U-t990 257

41. Dad.

42- Restate f op pesectmat synea sheds arv light os czhcking;f4eci Eag Isr Jos 1993

,,,s., or. e . s. .w.=

28

a .r.

o  %..n - cw.

~ ,

~

'merely showed one in& cation, the liquid penetrant inspecuon could identify tids as a crack Geld, resolv. l

'ag the single cracks. ,

'l  ;

~

Ultrasound has been used in several plants to analyze VHPC f'adings as well, in particular to deseraune  !

crack depths. However, estabhstung anck depths from ultrasome snessurements is known to be inaccu. ,

rate and difficult, with errors op to more than 100% even for more "maphiarleasa#* sechniques, e.g. l TOFD- (time.of flight diffracoco-) tedniques.* Unformaaefy, it is the best technique avadable for  !

9 J volumetne crack evaluanon. I I . .

11.3.2.4. Eddy Cunent Testing l The eddy cuneet teethod" appears to hamme the maw most widely adopted for VHP ta=aaraaa as  !

by now eqmpmear has been developed that aDowsnaaparnan of the inner surface of the peaconnon r with eddy cunents without the need to remove the thermal aleeven.

When anendon was direaed to VHP ianparmaa by the crad findaags at Bogey-1 ao appropriate in-ear.-

p .

vice ta=aae+iaa swehnds were aveltahie. There are two specs 8ca that an= rime the Wweaselesad challenge":' Srst,15e high radiance levels reqare a resnesely controDod Maatw to be chosen, and k turns out to be dif5cak to apply robouc equipment u te VHP's geometry; and socned, most of the pan-eerations are need with thenna! sleews, thos makag the loner peaceance surface laane,,ande go most of the asardarti surface ta=aaraaa * :--- (e,g. Inquid penecast testing).

However, by June it was reponed that at Besano-1 Switartand, for the Srst tinu nunose lampamna of VHP: had been =we==fnDy completed (akhough caly 75% of the pesetances had beca inspecsed"'),

and the same equipment of ABB Reaktor was soon after seed to inspect the sleeved positicas at

)

Ringhals-1 where a sysaem jointly developed by Westingboose, Laborelee and Jeanoot Schneider, as wen as eqmpment developed by Rahrarer AWucos Nuclear Service (BWNS), failed to work property.

By Septesnber, ABB amdals reponed that they had la=amad VHPs in sx Freed, two Soedish, two Swiss and one German plant;" la January 1993 mese aanbers had reponedly grown to twelve French,

. three Swedish, two Swiss, two German and one Belgian.*

The ABB robouc equipment can be Seed with two kinds of inepaman probes: a rotaung probe, which fks inao the sleeved or analeeved penetradoes and scans the inner surface by a combinanon of rotanon f

43. -s. Wilkas AI,4- FV. Kascunas EK: Aamsary d thraunc Rs= ssang Tschaques tr Ramssor Pressee Venads.EPfu.

NP 6273, Mant 1999 R -fly,she eddy cerveur method is bened as suoardag the mapase dahr specians endar : . ;

'~

whes q"try as ehermanas magande Asid endamd tus a suR. Der to the vadery d perumsars (perment@ry, sondmalvtry, esmans, tegemo-cy ) Saw desecnoe is a speames senasty espeeds as =@=a= d the tasages at e _i Idmunal her Sewisms mock <sps

. 45. Baro G D'Annacci F. Rylander L Mesmas the apper venssMead chaBeast wah addy osuussa and tJT;Ned Eas las. 3ammary 1993 4 Nedeonics Week. June 4,1992

47. Baro G d'Aresci F: $pcmens Fragesacangen der adP and thre r- --~ Hartmaagung DL Aaanicrum. N J.-.14115.

Seynamber 1992 ,

44. Baro G,D'Annecai F.Rytander b Mandag the appo.v-anHand cha5secc with addy maream and UT; Ned Eag ist.Jamanry 1993

====newsame, nan.m a m. m==an. a, 29 w

s .u n oe w.w .r u,w .n e

and vertical movement. 'and a scs:alled " gap 4ennaer", winch fits into the gap between thennal sleeve and penetration.

Tbc rotatmg probe is equtpped with an eddy current probe desgned to be senaisive to both axial and cir.

cumfercatial crack orientations, with an ultrasonic longiandmal wave probe, and with an altrasonic probr. with normal meadent waves, this laner merely to desernume the probe position by &%g the seid seam. *Ihe nitrasonic probes are to be appbed on unsleeved penetratsons only, beoot they are of  ;

use only for defect evaluation if a flaw les been h beforehand. ],

. . i The eddy cairrent probe is said to aDow ia*a*h of the peestradon's laner surface trough the sleeve.

]

Tbia, however, necessitates the use of low L% =h so have the magnede Seid peacerale through the )

stain! css smel sleeve and the sleeve penetration gap, with a corresponding loss la sensitivity.

Lhast., off censered or aged sleeves further -a8Na the inspeedom. ,

. i Therefore a gap scanner was developed, equipped wish an eddy cornet probe measunng only 1.5 mm in ]

width io fit inio the sleeve-penetrabon gap. M=yi!*wm isso his gap is ernmphahai by Sting the  :

probe to se end of a flexible stainless sieel artp. This allows bending around she cankal guide Anad to me lower end of the eermal sleeve (see chapier 3.1, naure n.1) and mas =aavitarwof the probe in  ;

the vertical duccsion. Croumferential movennent is shaply made possbie by moundag the linear drive for the metal strip como a rotanas table.. ,

As gap scanner manipolados is "soch that the gay probe is laserted alcag te sleeve lato the gap **. it is $

kaprnhahle that tids system rehably cercummets Ibc problems saemming 6cm variances la deeve-pen-etradoo gap width due to acede ovahry, so scene sleeves benag of. center, and to operanomal degrada-  ;

tion. After Westiagboose's fallare to perform inspections on the Ringhals 2 VHPs, a Ringhals  !

=w**-a was goosed saying. "The praetrabans at Ringhals-2 are slightly deformed aner all the years ,

of operation. The,smeasuresnents don't match what's la the laboratory"." Another inspection team recently reported that "N6azie eccentricity' anti bandag may crease gap resenctacas that prevent blade probe IW of some peceerations, so thermative inspoemaa techniques are necessary"" (a blade probe is a gap probch j i

As me.nuoned above, most of the VHP inspecnons so far have been carned out with the ABB system.

and as rescacn have gone back isso operanon it is highly kapartant to know the flaw.desaccon reliabili-ty of the systema.

- ABB officials r'eport that with the rota les probe, laboratory testa "abowed that axial as well as  !

circumferential inner defects,with crack depths of aboet 3 saa on the inside of the sozde were i desenshie tiirough the thermal sleeve. In the absence of a thermal sleeve, defoco with a depth size less'- [

than I sun were dearnehte at the inside surface.Ousar defeas were desmmhle if they were over 50%  !

49. ibed. l

30. Week, Jane 4.1992 i

S t. Selby SP Brooks WE: OtDM Noezie ' . '=: Nedear Plaar Jourant. Mmember.h 1992 r

M M MM M E M ME t 2-

w311 degraiauon, Wt on tbc wad thickness (For the penetrauons this hi@ to a crack dep(h of about 8 mm).The crack wkhb was insignif" ant. The ge- probe revealed iare% of amficsal defects in starderd test blocks with evec emsfler crm:i depths"." Invetrir*e% were also coo-doczed on a few real penetrations and defects to further qualify the techmque.

+

i However, the important quesdon from the safety point-of view is not: -How small a crack is i

detectableT. but rather How large a cract can go r=adaa~*d7" This is a question of probabdines, and it is unhkely that so called POD- (prohahrbry c( dat~e_k=-) unrves haw: been determined for this eddy curTent nechnique. It is wcD-known that cracts snhersntially deeper than those estabhsbed as detectable crvJts during qualirath of an impar *= w*mque can quite easily be mi=*ed " Further tabng into account that not &Il vessel heads have undergone a ervanlaaa VHP inanacew= (e.g. 75% at Bernau-2L that it is unclear to what extent the gap seanner is actuaI!y in use and W_h not only performed  ;

with the less sensitive teatting probe, and that W the gap scanner a nozzle is seanned within the rele-vtat esta of the seld seam" only, altbough defects could occur over the whole of the penetrauco (defects stre reported :) be wc!! below the weld seem in Ringhals-4), there can be no guarannee that tbose reactors that have been W already do not have VHP: contauung cradts of coes>derable dimension. ,

.c

52. PAro G. D' Annecca F. Rytander L Masc.ag the apper. esad4 cad eknDenge widt eddy currests and (JT, Ned Eag tar Jaauary 1993

$3. This was ooc d the rest.!u d the noeg4asang inww PiSC (Prograsume kr the Inspecdos d Snect Cg-) enar:s.

pkues I and II; eddy currear testieg d scara peecrasar tutsag has been indesed isso its soon to be f!= M phase ID SC Baro G. D'Annecci F, Rytander L Meencg the apper-weemd4eed d=% widt oddy carrauts :si tJT; Ned Eag tar. Jassary 1993 vissa eu. ,m neea.. - ,,..,....mac,e.e 31

_- - ~ .- . _ . ._

+

Sead'===* * *%==c- i' i

1 II.4. DISCUS $10N OF SAFETY IMPUCATIONS l

. . l M.4.1. Origin and Developmoett of VHPC

~

11.4.1.1. Fellure Mechanism of inoonel 400 VHPs An analysis of the failurewh*aina of VHPC accessitmes extensive ed=ia*th< of the Dawed paru ,

of the cornponent. In addition to a detailed demaipoon of the findaag speaal masenal investiganon tech. _

naques have to be susployed, like maao and miaafracsography, muaDography, and exrosmalysis of l i

corromon prodoca at the crack front. Detaled resales of such invesigaticas have act yet beca made avadable; therefore, the fouowag Am==laa is based on the hypothesis of imergranular setas corrosson f cracking GGSCC) of Inconci 600, as adopend by EDF."

j Damage to fama 1600 la the case of stress corramos anctag and low. cycle fangue is aknast endu-sively interayanif we, in contrast to austenitic seels, and the amceptibillry of Inconel 600 to IGSCC is

-l

~

• known from the fiadags in mese generator (SG) subing. The IGSCC-bypothesis, however, does act cover all speas of VHPC. l Besides the genera! ====*iMity of the maaerial, several other conditions must be fulfiled to cause IGSCC,includag: ,

1. Accumulatice of kapurities at the surface (and later at te crack tip)

Tensile stresses over the whole crack reglos. .;

2.

F.Jws. ore, mMirhat ,*a,it%eing is cased is some cases by overtemung during the weidag process. l 1

i In the SG tubes, heat transfer mamelana prevail that lead so an ar==alariaa. of ;r.y. ids on the sur- .

j face or at the crack tip, similar impurity accumulauan is also found la crevices in other locations.

Although the VHP geometry exhibits a gap between VHP and thenna! slee,ve, this is open at its lower j end and thus does not lead to deposits. An eramntarian due to bodang processes is improbable, as spe-cial heat tmosfer effects like those in SG subsag do not occur.

Sensitizanon as a result of the welding process that could possibly lead to IGSCC ander average condi- -l tions of the pnmary cirasit's water cheunistry might be capecand over the whole cercamference above  !'

and below the weld.The fact that only cretsia segnons of &c welds are affecsed (0*- and 180*-regmas in ingare IL2) does act support his W - '=_ in particular because the residual sensees from the = cad-ins process se present over the satire ciraunference. These m generally higher nas opersnag stess-es, in the range of 0.51.0 times the yield stress (La. that stress eat can be solerased by the masenal witheatirreversible serala).

Crack locanon and crack onectanon are n ar==~ar wie the hypoeems of IGSCC caused by rendaal weld stresses or masenal sensitivarirm la figure 11.2, damage is shown la the 180*-position est lies cosnpiciely outside the weld regson.

55. sanni NecWars 1991. Rappan as rinspeaser Gendal pour la Sased Nadaart I

$ 4

• 4 As the remdual weld stresses cannot be the cause of VHPC, other loads have to be considered. No inae-rior pressure can lead to insile stresses in the affected regmas, as in the lowerpart of te VHP: the pri '

mary circuit pressure is present both in and outside the penetrauons. Strais la the apper part of the peo-etrations is restricted by the vessel head.

Residual stresses caused by ovahaw= of the penetranons as a result of the conaaracxion and wekkas l process have been a*=waM to be a ccocibuta ao VHPC as ocu.The nahW c(crack Sadings smahhehad by Fr=mnme? howewr, do not exhibit a sypernare link between ovalasanon and damage tre=Ma It can be conc!vded that though there is scoe probability of IGSCC being a esilure =~hamen of VHPC, this can only be shown merlaarvely by ietsasive investigations and aimalanon experuncats.

Further syna gistic effects also have to be taken isso accoast is this respect The descriptions of findings are not mantient in the avadable sources. In a Pramastume source? groups of single cracks of 15-75 nun length are mentioned. The ar**=are 5gures in that report abow clearly confined and inalmM strictly vertical cracks, some of thesa through-wall 1he sambar and examat of the cracks indicates a high degree of damage, and thus the danger o(pecesration ined=hries The desenption and the resalts of a liquid penetrant enemmariam perfonned on VHP: at Ringhab 2

- show a (- =i48y different appearance;" the damaged area on te laser surface is charassertsad by a

, Geld of mainly verucally oriented cracts. The closely spaced axial cuads me ===ar*M by cracks run-ning at tilt angles of up to 45*. Fecen these flgares as laneestve damage of the antarial over a wide see has to be a*===M Damage appears e be far advanced. AMeaantly,it can be coactaded est a cucuse-

~

fatetial rupeare along the cracks of the neid that are tilted to the vertical direction is poemble.

^

EDF's J.-P. Mercier describes the develaa=aar of VHPC as foGows: "Ibe crads that have been found are all loopadaal startag from the laser ammaser of the lower part of the adaptor, ander the weld, and propagating peits towards the exneraal part of the pencarances"."la fiar==har 1992, however, the beginning of circunderential crackzag was observed when investiganas the through-wall crad of the damaged penetraboe at Bugey 3.* ,

In figure 112, the longstadmal crada do initiane la a region where there is no daffarence between outer and inner pressure in the penetranoa. This region is about 100 mm above the lower sed of the penetra-tion, but stiB ander the weld. There are two regions of dessage, he bigger cae at the 180*-position, and a =nari,e one at the 0* position. The 0*-ag>on is cioner ao the weid, etiere ancks developing upwards are reading the primary cirtmit's pressure boundary. A through-wad crack above this weld constiaries a leak of the primary citamit, although the leaking medium Grut has to peactrue Ibc gap of aboui 200 mm lengt'h between the VHP and the vessel head perforapon. ,

50 Fr=== Framacene Owners Group - Idarnunas as Begry-3 landsm.Onder 22.1991 S7. aind. -

SL Ramose 8moreness puestrasr rysans dads are light es meding;Ned Eng lus,les 19F3

59. Merunst JP (1992) How EDF has eoped with vessel heed peastrenos credias; Asca. Maydene 1992 aa Massac, Jan 21.1993 emen..s.eamarrem - . .,ma a 33

..n . mm 2 o . ..... ,

b

, A deformation of the VHP is impossible under normal conditions: either the damage occurs in the lower  :

part W the penetration without pressure, or - in the apper part - he vessel head reatncts atrain in the-radial duection and thus a considerable crack opealog according to plastic collapse of longitudinal  ;

cracks is not ennenvable. .

. i

~

This situation, however, implies that intensive damage must have deveJoped before it is detectable from outside the reactor. A local loss of stability that leads to a constriaion of six penetrabon, or a laseral l bend, can lead to malfuncooning of the control rods. Due to the anlal cracks possibly growing together ,

in the circumfercatial duccoon, rupture of a VHP cannot be *=riadat It is particular1y amous that the lealdag medians is trapped la te annow gap of about 200 mm length between VHP and vessel head perforanon. Due so primary waar relaxation au les solved ingredients (espeaa!!y boric acid) crymaffiw. and the'foGowing fadere =a*aa3=== possibly in comh'aanna can evolve: .

I

1. By crystalhzation of the lagmdicats and formadca of carrastos producza, astovaEastion of the VHP is .

possible. .

2.- The cryentinad ingredients and tis conosion products are temporarily closing the leak, thus conceslag l the progressag damagt. ,

I

3. The highly aggressive medium forming h the gap preferaldy attacks the femtic steel of the vessel head, l

la the neighbourtioadof me VHP as latensin damage das a comana cormaion la as aggnamin medi-em has to be --w*d Even unds seriaal weer demisty raadinnas without impurity acasmalanna l

- . high velocities of contam corrosion have been observed (310 mm per load cycle).The cracts penetrate the wall right above the weld la the 0*.positica 'see

( figure II.2): here, te load-bearing laterface between VHP and vessel head is attacked, posing te treat of loss of stabinty at his larmann of he VHP as well. The damage e the vessel head has to be considered more critical than hat to the VHP.

frota a safay viewpoint. Strong and uncontroDed corronica of the vessel head can put the matakty of the _j vessel head la the p.60s regions at risti.

An estimme of the faBure kinetics is very difficadt. EDF has misd.to dertw sad an estimane for differ-eat reactors" from formulas that describe the crack leenhanan period for IOSCC of laceae.1600. la this

-[

comparison, the Alarnemina of the iaflesece of Bader4eed tegBperature leads to the feauk that higher temperunres are lacrosslag te risk. This, however, h based ce the hypothesis that IGSCC is the pre- {

i damiaam fallare ===* mal == The parameter stress is not lactaded la te analysis, and no emptanswm for te naarmagelaa of higher naar=gehiliry of the peripheral VHPs is given.PineDy, a derivanon of the  !

risk to develop YHPC is given, reistive to the Begey 3 nacar (see table in depaar 1). For a number of runsaas these estissacs abould act be adopsad sacriticaDy:  ;

1. The failure mode has not been clarsfied unambiguously.

2. Even if IGSCC is assemed, important parameters like stress or primary waaer chemistry are not aimm.m.d

3. The failwe dev-laamaw does act necessar9y lead to a drere=hle kak.The occurnace sad detecubdsty

- of 4 icek is subyed to a aussber of eveabsalities (crati poordon b henght and carounference, weld seam ,

61. Strud Ned6aut 1991. Rappet de l'inspammerrw youris Sased Nestemre ,

ee - - - - .. - . -e-y- .

- i geonesy, Ly--y doeure of the leak by compressive satas and deposits). Mas, relaung te risk to Bugey 3 is prahl--* and more severe damageis possible in VHPs where ao leaks have been descsad.

I l ,

A detalM analysis of the failure kinetics is very dafGcult and would require a thorough enammanaa of

. a large number of Sadings and sufGcient statistacs.

1 French and Swedish analyses of earty 1993* do not include new assumpuces or considerances on me I VHPC causes and sw*mattmt. Aner more thorough en amma'inae of she cracked VHPs of Bugey 3 and l

-4, in France the hypochess of satts corrosson cracksag under higL residen! seresses, stemming from manufacsure, as the cause c(creding is still adspeed. Theene=maiaa above eus renales valid..

. .j The establishment of crack growth rates eg O.3 0.5 macreadour at 315* C' in the Annual Report on Noclear Safety for 1992 senoaaly soffers from the sein act known exacs esilars =*eha-dm (the non-conelation between peneastion ovahsanos and crack fladings eg resnans to be explanad). ,

The same bonds true for the results Vanesfan ahraiad with regart! to the cracks at the Ringhals seits,"

the calculated anck growths have been derived from data on steam generator minag tailure machs-nisms, where diffwent parameters apply.The French VHPC Sadings at considerably lower teasperr aares were not take into accoast la these calculancas. .

- Currently, elasso plastic saias analyses are curried out in France for the VHPs, with modeRing bot the manufacxunng proces and opeational loads. la part, resalm for these at~inda a ramas ope, aner sunpler linear-clastic calmfariana were not successful. .

These recent analyses caneatially give tracesre -w*aaleaf endmans of the residas!!ife times of craded VHPs. De analysis of the Radings at the Ringhals reacsors is done fbr axial ancks only, although the development of cucumferential cracks from the closely araced axialcracks is a~-W The calmtariaae do not include any correction of the stresa intensity factors to a: coast for the tafluence of adjacent cracks, leedtog to an overistheados of the VHP's resistance against crack propaganon.Only seess car.

rosion cracilag is assumed as the cease of crack growth, the possible omabiaabon with low cycle fatigue processes is not taken into account.

Devite these ova =='i==b a residual liie time of only about 15,000 boers natil a cntacal ornck length f

would be reached Oe antil possbie rupture) was catalaiad for the 10.5 mm oraci.is VHP no 68 of Ringhals 2. This corresponds to about two years of opwsbca and thw is shortar than typecal inspecsaoo cydes, so that this cal-I=ad residuallifetime woeld act aDow further opmanon of the plant accordmg to convenelaaal regneartaan Therefore, for this finding a " realistic, beat estnesse" . analysis with a crack propassalon speed lowered by 'a facsor of 2.3 was cumed out, leading tr. a =le=tanat reandual life tune of 35.000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br />.

0- ,

62. . EDF "la Cswe as Raenser", andand; Vaassfall *Ringhals 2 och 4. Raskrartaakar SpeahSMar i 3,_ , --"anagar*

OEK 4/93.02/02,93)

63. EDF:$er,ed Nacidaue 1992" Rapport de linspesser Osmeral pour is Sassed Neddaist. 3esesry 1995.

64. Vanssians *Ranghals 2 est 4. Rankaarsaker SpeicisQlvest 1 i 3 ..; .- . ' ' . " M E 493 G3Al2/R3 2

m m.,,,a. 35 yesa sens m i. .

. _, -_= . . . .

..c. ..n. m -a. u ., _

The critical wack lengths a8valaart in France (eg 350 mm fo~ r axial cracb) and the derived residual life times appear 80 be too high and not conservatrve, also in compenson with the Swedish nauks.

Therefore, a detained analyais of the failure bnetics is not given here, as k would requem a thorough examination of a large acaber of findings and suffi:ient statisaca.

11.4.1.1 Other Reactor Designs The damage already mentioned above in the VHPs of the VVER 2 at h6 3. through-wall axial '

l and far advanced circumferential cracks - at fast sight also exhibited the characnenstics c( stress corro.

I sion crachng (transcryuartine breached cracks. 'tish eyes' la the microfracsographic image).*

- It is noteworthy that the Rheianberg VHPs are not made ofInocal 600 ba,r of a Sonet austeni6c stain- )

less steel with different toughness and corrosioo charact.mstxs (a high alloy mostenitic Cr Ni sasel with addition of W ned T1), and that the outlet temperat1re of this PWR which has been shot < lown by ,

now, was well below 300 *C 061 *C). Funber damage analysis led to the hypothesis of low. cycle fatigue, coupled with accelerated crack growth due to the primary water cheunistry anoditions The cyclic land of the weld wm camed by thermopulsanons and bem secanes.-

First occurrences of VHPC were found in 1974/1975, intiating as lamparwm of an VHPs and replace-

• meet of thcae with la&adana Perthennere, technologgal measures thought to conoseract VHPC were d

meroduced;la panicular, indecsion aam'hama at the VHP that canned ther=T aadr=* were changed f and the VHP welds were replaced by Sanges to adece speassa.  ;

However, la 1934 VHPC occurred again, widch was detected as a leak during operanon, although the VHPs had beco annuaBy learwead by addy-current echsignes until 1981. Again only the cracked  ;

VHPs were replaced, notil in' 1986fl937 aR YHPs were exchanged with reenage of the upper and lower  ;

Danges."  ;

VHPC at the units Novovoronezh-1 and 2 in the fonner Sonet Uenon was also enributed to thermopul- ,

• ~ annam doe to injection of e#rtional cold feed water. In contrast to omtrent PWR designs, both reactors sere not eqmpped with a xmancal reactor vessel head but with a plais apper reacsor plase wnb holes for the VHPs. 'Itas, asyuumetric stresses in the wekis of the VHPs that lead to VHP ovahssoon were ,

not prescat.

VHPC at the experuneatal reacsor at Kahl, FRG, (which la a boiling water reassor, is contrast to a!!

cher rescurs armandened here so far) ocnarred la most cases in Ibe p-@.1 VHPs. Both stress corro-anos craching and seress aar41tanam due to aunperature varianons were considered as cannes of crack-ing. Both snar*malana were act sufficient to explain ec damage exuet, and reamtoal stresses probably present were also takes imeo comederance *

65. Mauer F (1976). N SUS-Sussek:WWER-2.KKWR WPC 23% NPP ewahmy ,

l f6. Energwwerts Mont: Kernkraftnesk thsdesburs.h'*** ad 23 Jats: 5 :,'_h; N S41.1991

67. Chenier K (1990): 24 Jahre Karakraftwerk Rheisters Scandreness sum V . -

• Karusessgu 33 (19a0)3 m=* i- ( '" in VM-W Kahl.

as. n==ar. Ki.e.(tm):n r==-an ese mennessemechfaariness as Ranksanagung 1973 .

MeDar F(1976): W am h WWER-2. KKWR WPC 22% WP Rhensharg j

]

1

== ==aa a===== a====== = =maaa ana

. 36 ,

.s. .. ... .. . . ....

11.4.1.3. Conclus3 orts on Failurs Mechanism and Demiopment The foGowsag conclus>ons may be drawn from the currently avadable ilata, winch mainly rder to the l

VHP design of Westingbouse rextors tut also explam the sanilarities with the above mentioned occur.

rences at other reacsors:

. The failure mehaakw curreedy cannot be considered as being conclusively analyzed. !ctasive dam-age in defined regions of tbc VHPs is occumag. This darnage does not aucewily lead e a leak that is defectable from outside the reactor, before break occurs. If it does, h will be at its Saal stage of development caly.

Bob axial and circumferectial cracking is possible. Cracks may be isolated or part of a crack 6ekt.

the lacer being an indicaban of complex load structures.

• A descripdon of Ibc failure kinetics and anumava of failrut growth rates and falkre probahlity is

- impossible due to the lasufDcient knowledge c(the' failure ==*=i=m'

. A loss of stability can occur both as impairment of the gmdare of rtacsor control rods (constriction.

bending of VHP) and as rupture of the VHP (crcumfereocal cracks developag between axial cracks, l f h comnive damage to the n=a~'4ag weld).'

. A loss of strength of the rextor vessel head cannot be nh in t>e came of esseasive =n="otled f corrosion starung in the gap between VHP and vesac! head.

. Due to the lack of knowledge on the canaes of VHPC, hs avondance by sce.evGive anMar techoo.

logical measures cannot be guarsoiced.

IRA.2. Description of possible accidesit scenarios '

11.4.2.1. General sapocts .

The damage raehanierna descr%ed above wi!! now be aamanad according to possible reactor safety conseq=we lhe causes and Le devat==rt of VHPC, and thos also the possible catent it can reach.

have not yet been clartSed in detail. Therefore, only goa5tadve naala ations regardmg an *at sec.-

narios are possible.

Dectrsiid de France nraany maa*rs two severe v=i=== reautting from a VHP leak a possible:

• corrosion of the exterior surface of the vessel head, which has no claddsag, due to leakage of przmary coonaat contasang baric acad.

=

efechoc c( a control rod drive ma+=akm (CRDM)

• CRDM eyectico is regarded as a possible conseqococe of a VHP break due to anMen grosth of a cir-comferennal crack.

However. it is thought that the atnence of cucumferenaa! cracking can be explained by the orientadon of the stresses in the crack region'. Therefore, -OtDM esecdon is umawred to be imy=**~." This premature exclusion of the possibility of crcumferennal cra:iing has already been M* ad in the pre-vious chaper.

~

wo.r a timx m- a>F ha cop 4 =* ===1 had s. - ,: Ama. w ro=== im

+9.

7o. aw 37 vena.u c= - . anae==

. .-- - .-. - - . - .- - . -. . ~

, t Pet O. The rusmasonas Fernascwe ,

4

. I

i. .

(

GeneraDy, longitudmal cracks are regarded to be of littk coocera freen a safety viewpoint." As will be shown in the followir2g, this evaluataan, too, is not correct.

Plant operators ha~ v e not itscussed the possible consequences of a fallare of one or more CRDM because of VHF cracks anWor ovah*=naa of VHP tubes. Furthenaart, he formance of a smallleak in l the pnmary circuit ckse to one or more breaks of VHPs can lead to a sevat accident. This appbes even . j if there is no CRDM ejecnon. ,

In practice, it has to be assumed that.several of the potable consequences m.ahaamt will oczar, which  :

can be causally combeed. It is coeceivable that break of a VHP tabe leads to failure of the CRDM  :

F anest system. E# ecnon of the CRDM which is poshed upwards by te prbaary pnasure in the reactor '!

vessel may be kladered by additional systems now being inst'lled a la the latest French stactors. l Nevertheless, a leak at the VHP wiB occur and CRDM operabdity wlII at least be seriously reduced.  !

},

11.4.2.2. taak in the pdmary eitouR _

A leak can occur due to the fannanon of a through-waH crask la the VHP abow the said liaking it to the essel head. This was me case, for *-aala wtus the VHP crack was desecand at the French NPP  ;

Bogey 3 (the leakage these was observed dunag pressure tests, and not during operamon). [

Qacks through the VHP war may not only occur during presamt enas, but also bocasse of vadacons in l pnmary presset donag operation. Quk grove sensaDy prooseds la tree pheasa: Onck inhianon, see  !

ble crack growth, and immable anck propagados (brask). The trut two phones ma occer during longer l' penods of time and amain local ph -<== a. seasta te imaasehne asishbastood of te crack caly.

In the tird phase, however, te anci propagman approximueely wit GIe speed ef sosed'and penetrues l the wall of the m'apanaar To trigger this phase, a short seems plak esceneng a critical vidue is suffi-cient.'!his critical value depends on shape and sine of the crack. la te case of ektmenfetatial cracks, the '

tube then asuaBy breaks isso two parat IAngrediaal cruis'can seer open over longer da- la any l case, the break can be ar==p=alat by comoderable plasse deformanons (backling of the abe). j I

Fulniment of the leak.before-break (125.) crtierton beplies est te escond phone of anck propaganen l contaoues until the cack has penetrued the tube war compienely, without occurrence d a break. iA. witb- l cut the third amenhie phase being inst =sant The leak thus gaurasad wlR very Rely be sigmficandy small- j er than te opsting created by a bnak which can be as large as te auss secnon of he tube. it is assmand 1

that these aman leaks wtB be sknely demeand by leek densceos symmas and do not lead to sevat accadeen.

Plastic defonnathms of the tabe accoripanying the propagation of a crack right through the wall increase the size of te leak. Furthenacre, the leak wiB remain opes even sher pnmary pnssure has

. been redoczal.They can also impair te operability c( the mooisonag tube or(,,WM la te VHP.

71. Manner JP (1992): How EDF has soyed week summel head paastremos escenas; Anom. HeyGame 1992

. Penserenos cancknas foemd at RJaghals.2; Asen MeySune 1992 v =1h de ctsapainst4 ,NessesrEnsi dagbasense.o 1,Mytve2 38 .

• • '"*-"***'========""'""'

l l

j

\

In the case of VHPC, NPP operators sssame that there are no crcumferential ancks, and that the leak- j before-break pnoople is dermitely fulfiled. Both as=" marks are not cessm.tive and have already {

been div=d in the pevious sections. .

Therefore, a conservative assumpooo of the leak size in case of a VHP break would be the size of the interior VHP cross section (oremmt width 70 mm). Leaks of this or even consaderably smafier size (frocn a width of about 15 mm upwards)in tbc primary circuit can lead to severe accidents.

Furthermore the same stress peak which rm"m failure of ooc VHF can also initiate unstable crack growth in other VHPs, leadmg to a leak size wi@g to several VHP aoss seasons.

he occurrence of ooc (or several) VHP leaks durirg reacscr operation w1A to the amdent case-gories unnfl or intermediate leak in the pnmary circuit. Although thesc mer*ar* are design basis acci-d:nts, due to unforeseen but not ignorable malfmema of safety systems or funbcr <-v=t failures the pnsW1ity of a subsequent core melt does exist.

He coune of events following a small or intermediate prtmary circuit leak is ema;4et depends on many parameters and can vary from plant to plant. Therefore, it will be @rmsed in a short and anmmary rme<m only.

Incressmg radaoactivity r="t* the pnnury cremt and/or deceasing pressure la the prunary carcud will signal that a leakage of pnmary errdann has occurred. The reacsor will be scr=med foGowed by acti-vaooo of the emergency core cooling sysaems wtuch serve to resnove the decay beat 6'ag prr*va-even after the chain reacteo has stopped.

Tbc ardien ingress of large aaw=ra of cold water (wie ternperatures below 60* Q into the reasor pressure vessel **ia! ot water b (above 250* Q at high pressure results la se unstable regime with steam and water in the pnmary caremt. D=4N on leak size, pnmary pressore may remsm high for a longer penod of time (HP-pa&), or may decrease rapidly (LP-pae).

The ws!1s c( the pnmary carcus, partienlarly the reactor pressure vessel waQ, are subsecs to loads gener, ated by internal pressure as wil as by nernperanre grhrt (thermal stress). These loads can lead to failure of fade pnmary caremt r~nenarats, followed by huge radion:tive rete ." from abe pnmary dreuit and core mek.

It has beco acempud to detamine the probsbditics of primary czrcuit leaks as we0 as the conditiocal probabGities of a subaegacot core met with the akd of prnh=MiA risk analyses (PRAs) for nuclear power planta. Soch PRAs have beco perfcrmed in several a3cntnes in the last two demin.

The results of an American, a French and a Gertnan' risk study, perfonned 1989/90, have been dis-cussed and rampared by Werner." Reference plant for the French study *' was a 900 MWe 2nd geocn-tion unit (series CP2).

72. Den PMe Karmbaftwats Phase B.GR.S.A.160th C" A* thrrN KAtademi 1999

73. Werner W(1991): Abacfk Erybasse sm WWh Sc6erheftsanalysca.OR.5-Fedgesprsch 1990,ww, Mars 1991

74. Erade Prn6m=* de $4:sd des Rmcieuri 4 Esa sees Pramos de Palist 903 Mws, Rag 9 art de synn IPsN Apra 1990 ve==a na .rmn . umma .uc, 39

• l i

~

l An investigation of the merits and shortcomungs of probabilistic risk analyses lies outade the scope of the present study " as does a comparuon of the desagn of the various stfertace plants. la what foUows, results concerning primary circuit leakages only will be refermd to and deceased.

- Cort melt accidents foDowag the LP-path can lead toiste failure of te cooralamnar and canyaratively '

low releases to the environment. Machaa===< leading to earlier comramwnear failut and higher rescases i (for example, hydrogen explosion), however, cannot be excinded.

i i

The HP-path occurs if the size of the prunary leak is not sufficient e reduce preamst. HP see mek will ,

probably lead to earty contalamaat failurt and rapid, high releases of radioactivity to the environment.

Thus, the bazards ==earssent with a prunary circuit leak do not decrease wie decrummag isek slac. l' l

Measures of accident managesneat (l.a. laserventaos of operadas personnel into the courne of te aces-l dect) are planned to prevent core melt or, if this is not possitne, to at least transform HP sequences isso sequences with low primary pressure. Those core melt sequences have been desigassed IJ**. {

t Regarding the transformanos of HP- into LP*-sequences, it is'assmaad that the probabdity of correcl meervention by the operating personnel (La. all necessary measures, and so evaag mensma me teksa) :

l is as high as 99%. (In some special cases, probatmlines of 90 to 975 are assumed.) Becense of es last -

of demiled inveangstions, hootver, tone prnhabl**= me based ce prelladaary, sough coanderadoes.

Thaely intervention la case c(pruassy circuit leaks la most cases h laterveanon wimm 30 to l a

135 minutes after occurvesce of the leak. Canne and likely dew agummer of te acc6 dent have to be analysed within this short parle ' of time, and the laservandon urategy has to be emiscasd and plann Geners!!y human arar is a risk facsor act adequenely takes inao accoast la PRAs.The three risk sembes analysed in Werner's contribanos" la most cases inciede arrors of asaksiae caly when estuaanng acc dent probabdities; errors of trua=l*< ion are usually not takaa lam account. " Anal)ies ofincidents in th

• last years regarthog the laDuence of banan errors la5icas, however, that wrong Ah of a senous nature in the comuni room, which can get the plant iam,a dangerous maar., can have a probabdity of l 1

occurrence u.s, .AA to the probabdity of initiators of aerious eschaical failerta"."The PRAs men . '

tioned above therefore may sigmficandy undem acadest prnhah* i i

According to the German Risk Study, Phase B," the overall core meh probabdity is 2.9 x 10 5 per teac-

- tar year. Without accident managaneet, the tractaan of the HP path is 975. If aerwaaar managa!

is redoced to 3.6 x 104 per reac.

lar4=w la the study, as r==liaad above, overall core melt t accidents.

tar year, with Hp-sequences accountag for only 12,5% c(all cort l

i 71 see, e.g.: Hamh M. Lafah T. $danacher O.Thomques O (1999y 1AEA Safety Tarysis sad Psotmhihane RJak Asseu l

Raport pregated for Grosepasse lasarmanamat.Oruppe Otmige.Hammeur Aagem 1999 i

% Dessache Rasikosruhe Karnkraftwerks Phase 3; ORS A-It00,wA8t str Rastsarachsst=t; Kala:Jami 1989 ehh&*_ ,- (RS Factenspeed 1990,erw. Mars 1991 l

77. Warner W(1991) Aktes5e F ,y-- a pr

, 78. l Asid.  :

aea=adse Karakraftwarbs Phasr B';OIts A-140&,WA8t far Rasknorsichertsk: Kas: hei 1989

79. Deutsche l

i

=um. saae seurmw== .assuma = ==maan saae=s

40 i

(  !

4 l

The Readi PRA" shows an overall core mek probabdity - for operados, and iDdeding acQdent man. l

. agament - of 3.2 x 10 5 per reaaor year, the fracoon of the HP-padi being 185. Witboat acodent man .  !

asement. core meh probetality is higher by a faczar of 18,* sad thus exceeds the lhak of 10 4 proposed I by 1AEA for current reacsors. .

Small and intermediate-size leaks in the primary circuit (~'da: presserum and esens geacrator .l leakages) according to the derman nsk study" contribute 16% to the expected cort melt frequency  !

(without amiear management). If amient management is included, this i actice is lamensed to abce 45% - accordingly, almost half of all cort melts to be expected would be initiated by such a prumary cir.

cuit leak. -. -!

Accordtog to the French study,' 28 % of all core melt events are landstad by small leaks in ene prumary circuit (including accadent management).

f i

The probabdity estumates given here are beset with many amoertainties and at of very lamused rele- l

. vance. The insertstag point, however, lies in ibe fact that akhoegh amadent management is cimmed to i reduce overaH cort mek probabdities, the probabdsey of a cost melt due to a prknary cocuit leak is not f

reduced io the same eneet. ne chmace e coment pnamy drcuit lents trongh aeddest management is -j

. Ibus smaDer than in the case of other accadest categones;te eSacsiveness of the measures is louer is j the case of kasef-coolant ==t a'a r." ]

II.4.2.3. Impairmant of the saaster anstal syntese l

'!he consequences of a VHP leak can be enhanced if manhoring' tabes or CRDM ta te VHP asbes art l

. ampared (comunassasse falhas).The cannot rods are te comani part of te sencer proucaon symem. {

In most pressurued water reacnors, a sapid power decrease er amani cuia be adnieved only by immoduc. .

tion of ibe control rods isso the ruecear core. The ' mcsion of die CRDW is imperadve la cans of an acci- ]

dent. If it is impasred, new, naewpenad acciden t sequences will develop. Beat producsson in the core l can be hagher than expecsad, and the possitalities for inservendom of the operating pereoamel will be fur. l aber reduced. This is of perucular nievance for those resomrs bevlag a high proportaos of MOX fuel l (i.e. aused oxide feel contanang bodi plinomaan and =rma==n), becamme sedi a core leaves only sessncs.  ;

. ed margias for reaczar consol anyway.

A r**amaa of the abihty a regulase the control rods, and has mactor power, can also resak from local . 1 detonaanon of VHP: by corvosaca prodecss between vessel head and VHP, as described above. This l wiB act me**marily be amunpamaad by a pnmary carcum leakage, but can, however, lead to a severt j redaccom of the mobilary of the control rod. It is posable that this inapearment of reactor consol will at j

)

. I to. Erade PrM== de Sarund des Rammeurs & Ems namn Premmes em Peber 900 MWs.Itapport ds Sysshess, PSN Apal 1990 )

si. au .  !

32. n xima d x.,*.a rs.Pi .ons.4.i.av - n,a -xa.a.mna,

83. Arade Prahmhuime de Strud des Rancesurs & Eas som Psumman de Palist 900 MWs. Rapport de Syssimme.FSN AprQ1990 at w. mar w(1991y Aksumus Eryanname as c ~ " Echuhanaamsiyess;0ELS.Postemmysach 1990, sw. Man 19H summa, sans pensvannen annamma a sumana smassess 41

-' . ..r... .n.  : . .. .. .. -)

i a

first 30 ===5 by the reactor opentors, and will become apparent only as soon as changes in the reactor's operating regune h-w memasary (for example &anas orthaary sert-ep or shot down proce.

dann, or when rescear scrum is initiated for some reason widct may have mothing to do with prunary j coolant leakage). It is possible that a hadden CRDM failure will then have catastrophic consequences for i the further course a(events. ,

j

. l Coocerrung reactor scram, an that is said in the Genaan risk stady is: "The eseca of rencsor acru'm l failure have not been investigated _. Such segmences art, from the viewpoint of their frequency, )

insignificant, they se pessimistnaUy sessed as core melt cases."

]

~

II.4.2.4. Raaetor vesset head failure. .

j As alttedy Ame=*aad la the sectaos on origa and development of VHPC, there may be cormaon of the I faritie vessel head matersal la the vichky of one or several VHPs with through-going cracks wishout j detectaon by the leakage locanon systems. The resolang loss, of suength of the masanal can proceed l until the stabdity of the VHP moemang (wou plus squeezing of te VHP inao the vessel head)is andas-  !

gered. At the same time, ovahsation'of VHP above the weld an occur, leading e a local wideamg of - l the VHPs. hc== of the weakened mounting and the deformations of VHPs, spectaan of a CRDM can- l not be exduded even la case at longina&aal cracks, particularly since plastic collapse of longitadamal cracks can lead to fureer VHP defenaanos. j Since it is possible that several VHP: wiR hsw throngb-war aracks, conosion of the venact head will act necessar0y be lhaland a te vidairy of a edagle VHP. th sjecdos of seveal CRDM, foi- l lowed by global fallare of the vessel head, has a te considered as he possible worst case. It has to be l kept in mind that the perforations of the vesset head and the aqueedag-is of the VHPs cause a very . .j complex load strwart in the vessel head nearial i l

11.4.2.5. Further hasards It is possible that pans of broken VHPs will fall iam he reaspor vennel, damaging es feel elements la j f

the care area and Ibes the kategrity of the first proescalve barvtar, te fuel rod claddlag.This can lead to an uncontroued, significant release of radioacxiviry Ana the' fuel rods lado the pnmary cuous. Fauing -l down of lower VHP pens is also possible la case of a bnak beeow the weld - a region where cracks  ;

have already been found is. Such breaks do not coastmae leaks of the pnmary cucat. Apart from the ,

possible locnase of radioactrvity in the pnaary circuit, however, they could tartmaly lead to impair- l mest of OLDM. ,

It casant be excluded that VHP break with a pnmary circuit leak is coupled with faring of VHP parts  :

into es cart la this case, EM first prosecuve bemer (fbel elesment Ma4 and te second prosactive .j bemer (envelope of the primary circalt) fail aimmle==aly.The last naamaing bemer agamst radioad l tive releases into the savtromment would then be the r=r=-u=r .!

I 1

85. th* RJalkaswho KarakraRwasks Phase B; CRS-A.lsoOr.f"MAA far Ramananmaartmic Estademi 1989 f unia sans ==vamn= sens=== = ==maaa saamas -

f

. 42 -

- -m - -v- ,,4+ - _ -- . - - - - -

s >

. In order to connieract CRDM 'cjection, addirinant support systems are at presamt being installed in-French rescan. h is act clear, however,if such symems are also ptmanad in an posentiauy endangered j

. reactors in other countries. Regarding the VVER-plants in Eastern Ewope, Psalmad and the former  !

Soviet Union, and the Westinghouse plant in Krska Slownsa, for *% no informanon has yet been l pubbshed on the plans ofinstalbog additional CRDM support sysaema. f

. CRDM ejemna aher VHP treak or failure of vessel head sabday cannot be ,-- My excluded.11is ,

accident sequence would combtne all e case assumpocos 4*eid so far. Furthermore, if the eject-ed CRDM is not rehably held back by the anti-miemte bemer what many plants have te=M it may damage the ermrammear by its dynanue unpact, thus impaanng the integrity of the last prosective bemer right at the depanms of the moaent ~r-e t

f I

6 e . P e

0 l

. l l

1 l

i l

l l

1 O

O i

.a.. . ... .-.

1

)

I 11.5. COUNTERMEASURES ADOPTED

~

R.5.1. Measures to repair VHPC j One of the repair methods adopted so far has beca keminat crosive renoval of the crack affected masenal, as l reponed for Ringhals-2, Ringhals 3, Beznau-l. Due to the high worter doses ths.1 result fran maanal reper, i aeveral vendors have staand that they are developeg remotely controued repar equyment eat could deliver repaar tooling to the crack locanons.

- i

- Erosive removal, however, is always dec casing the waB stuckness of the asserkt, tus weakensag strength -

reserves in the case of repairing amaD defects and being impossible for larger ones. Repair weidaag of VHPC i has not been reponed, t is would probably lead to funber weld residual stresses of encontrollabic pansras j intrateat into the VKPs. * '

Replacing of peactrations that have been identified as being' flawed does act solve the probless either. ,

Extracoon of a penetratson out of its vessel head fixture is *a very dehcane insk, perdy because the cylinders .l were origiaa!!y shrunk-fit loto place before weldag, and the process most be reversed to remove .thesn. .  ;

Moreover, $ of the 65 penetracons have thermal sleeves, which most be removed before the paastrances .

shemerives can be extracsar'" { quotation concerns Bogey 3, Pasce). .

l The altnacon again is further counpbcated by the high rasation doses la manual work on the vessel  !

i

. head, and also by the problem of laseruan and melding in of a are peaceration. Dinamilar wekhag (i.e. weid-ing of two different materials) of a penetration to the aged and posaihty conoded vessel head perforance is a

• delicate task" as we!!. In particolar because using laccesi 600 pensamicas again would make Unie sense (a 1

.' defecuve pe'octration at Bogey 3 was reponed to have been replaced by a new one? seost probably made from f amael 600).

Consequently, EDF has announced k is plannang to replace a5 ee rescaer vessel heads of its older 6 CPO reac- i nors and at least seven snore, whidi appears to be the only optica, especaUy is the light of the above measures being based 'on the quesponable assumption that all cra&s of concern are husad durhg ta& The pen-strations are to be made from laccoc! d90 wtuch is believed to be less nWhle to stress.corrosace cracking j (SCC). As laag as the VHPC falkre muhm=== is not bemer undermood, however, these coedy replacements with Incomel 690 may turn out to be imistake. ',  !

. . , i

}

1LE.2. Measures to prevent VHPC  ;

i

. 1 Measures to prevent VHPCs have been dertved from te parameters that are thought to contnbute to the forms-

~

tion of VHPC.

• i

86. Nedeoence Week.14maber 21.1991 i

87. Naciscacs Week. h3 ==h=e to.1992 l

4 -

- - -.-- l

-~ , - - . - .

_pn in. ,umuu -rm ..

I o .;'

i One messere that has actually been implemented in some of the French 1300 MWe-PWRs and winch was dis- l cussed at some stage in Sweden as well" is to lower the seder head temperature in the reactor vennel. '

Accorthng to a formula the Reach analysis came up with, crack propagation speed woeld lacrease Ibirfold with end 10* C rise in temperature." By 6iverting a pan of the ooid leg lalet coolant into the apper riacsar. it is possible to increase the tanperature difference between ondeg and ender head and thereby redace he under-head temperature by 2$* C in the French 1300 MWe<escsors (im from 315' C down to 290* C).This, howev- l er, is acmmpanied by downrating the urst. accorthng to a EDF official. "by a few messwans"

• I.awaing the l under-bead temperature, wtsch involves only two days modificanon work for the French plants, has no far not been reported from other countries than France.

• i j

Another parameter abooght to have played a role in the dewtarmaar of VHIC are acesses tatroduced into the i

sh by the welding process and the oval geomery of the weld seam. Reducing these setsses by shot-peening the penetradoes would prsacapelly be possible and he asnee te&nique has been seed already on seem generator tubes." However, robotic eqmpment would have to be heyM for tids tactaque to be apptsad onder the vessel head due to the high ra&nnna level, and the extent m whsch messes are affecsed art unimown j

' for abe penetrabons. So far shot-peening does not apper to have been applied in any plant j 4

1 h  :

For both temperature lowering and stress relieving messores k inest be s=T aal=d that the detailed fadere math ==ne of VHPC have not yet been earmhliehad anmtscady, and thes it does act follow &at teir impie- l

=*aranna would e=**=rily prevent VHPC. Eg. the fladings at he Frund Blayals reactor indscene that ten -

paramre lowering is probably noi a suitable coanamnesamel -  :

j As k is -mly assumed that the sasceptibility of lacasel400 to VHPC (be it seems corrosion cradang..

some other aging effecz. or a comhiaarina) is a mala cause of the probians wbb VHPC, measures like plating j

~ i the famae1400 penetradons (e.g. whh coppery' or sisevtag them have been proposed to separane laconel 600 '

from the prunary coolant. This, however, would require casensive spatiftentina and installaban work if it were to be exercued on all lacomel 600 VHP to tackle the VHPC probian, and again the anoertamnes due to lack of failure understandag would renam.  !

According to corrent knowledge of the prnhtem the most prudent way to marv==nar the VHPC problem there- \

fore appears to be replacing c( the vessel heads dg g lacomel 600 peaeeranons by heads that have their l penettstions made from a presumably scre resistant inamerial A premnditinn to do so, however, would be the  !

I clanfk: anon of the VHPC failure taode to establish and qualify noch a masenal i

h i 4

1

]

i

~

an- Nacisar New. Gr--*- 1992

99. Nh Week.Jaanary 23.1992 to. Modeomse Wesk.Jasmary 2.1992

91. ma

92. Nedsmuss Week.May 7.1992 r

v s.,

e

- - - - - - - - ~ *

1. . i a

a 11.6. CONCLUSION

• b When first repons on the occurrence of VHPC a the Bugey-3 reacsor mere cuculand in ime 1991, the inerna- ,

isonal nuclear comununity &sapproved of any safety imprime;aae for their reacsors, and only when it became [

apparent after further findings that another genene flaw of a PWR deman had been delecend by' chance did the .;

poblem raise concern in asher' countries. The findings a the Tvesch Blayals.!'anit la lane 1992 dramancaDy  !

mahanced the VHPC probleen, as they give clear endence that up e now the anderlying failure =a<*=a>=nt of l VHPC have not been undersacod, and further inndeam are likely to occur.

Ahhough aD Sn&ngs reponed in the wake of Ibe Bagey 3 incadent have ao far be restricsed to tone reaaar designs employing Incoac! 600 as the VHP base amerial and ibis amaserial appears en be, highly aunceptible to VHPC, probians with oder masenals and reacnor designs have cocurred in the past and doe to the lack of i knowledge of the fadure maramaten cannot canegorically be ruled cal A ammber of, highly necessary measures  ;

must be derived froen the endence gathered so far on VHPC and its safety haphredame-l'. NtadM ena=inatinae of the flawe ; vunpnaents, analyses ofloads,operanag and residual stresses and simula. -

tion experunents have to be carned out to arrive at an anderstanding of failure causes and kinetics of VHPC. )

2. Nosslestniczive assang of all VHPs and the vessel heads of all renesars wie penesrations made' from lacomel ,

i 400 has to be perfanned h==MImaly,using highly rehabic evaination whaiq=as (e.g. a further developed and j

=ma==M liquid penecant test adapted to the penetrance desga).

3

3. In-service lampa<*= of VHPs has to be regularly performed is aB cther seacaer designs as well and thus (

Madad isso their151programa. -

As long as these neamros have act beenn=plamaand and a tho' rough analysis of te failee mode been estab. )

lished, measurea Ike vessel head War m amenal aaharadr= or amaperature lowering by down.rssing ]

3 cannot solve the problem sind guarantee safe opersboa. This also means est te reassors at risk have to bc l j shot.down sotil measures based an~an iederstandag of the failee ==<*+ai=a have been derived. )

l Consequently, the Preach approach to inckle 9e ;Mee of VHPC by soplacing the vessel heads at their older j units may well tara out to be too short.aghaer.. bante adasanaaa* cosa have boca estunated for these sees.

sures, and with the dauy cost of as amplanned twtw teang 100 so 200 million Prances to IDF,it is cieur why EDF is no longer trying to understand the (vmntion cracksag) W la desar' and prefers to put its efforts isso estunanog crack propagaboa timee r haw ranczors run even with proves VHPC.This approach has i also been fouowed by Swediah Ringhals. ope nor Vanear=11 and there can be linie doubt that maular stances wn! be adopsed in other countries when the prit.ca has been reshand widely. -

]

VHPC is another eaample of the growing piews of mawial problemas that the modear indasuy has been and still is fadas, e.g. seen generssor problems and rpi- sanssive boiling weer reaciarJGSCC, and also most recently intensive cracking in araMivad annenitic stainless aseel ambias is Gennan reassors. Nome of these hazardoos problems were fonnees but rather naled out, whidt act only shows that amerial and compo.

aeot charactensacs for nuclear power plant appbcations have been groesly overestimated, but alst reveals the highly dangerous sine of healib the modear reactor populistion has reached.

93. And.

46 . v sum a = . a. .w. ,

I

it. FIGURES Figure R.1: Vessel Head Penetration Design .

.~. .

a l .

Upper dn.ye . ,

mechanism N r,, . .

b

' ' ~

! .d .

s l e $<

M L fa 1 II[9

' r

..* 2 s.

', h

.<[f:fd3f'.h' Thermalsieeve ,

. l  : ~x.:g.: .

~

. . , ;, 9. z. :.. L . , . . . ~

' j '

' yessel hesd dome

,,c .

~

carbon steel

=

7' l~ .

51a,inless steel cladding

.!:l.'?E'!I' , ,

i inconel 182 buttering.

. f .. l b. . .

. ~. ' ; :.,' ' .( 4- I l -

Inconet 182 weld ','

L' J Inconel ve'ssel penetrat. ion 4

0 '. j f

n - -

la l

T. .; ;

(

' Connecting rod ,

3 7 Conicalguide L- -J l

I .

94. Annet. Msyamee IM2 .

I

~

47

.nu n mm eman.s a muun amena I

s 1

Rgure it.2: Findings of VHPC at Bugey 3"  ;

.. ,._ , . ~ . .  ;

s . ,. ., , ,

~. .v#. * >

.sy* *.

w . ,. ,/ e

. s .

e. .n . .

.. '..* .*s -

. . c. . ., .

. . s. , .

. ,/ .

• . . j*p l .

• *  ?

.,, /. ...

t

.c s .

.. i,

. s<...., .-. . .. - .,

s

e. . . . ,T . .,. . . .

r .

s

,/

.,.a...

>* .. ... s. . -

s- a .. .

.: .: 't.. l

. . .,/.

, a .

,s . - agn .

L ~

r,, . . .. .

t, u .

c f.', . % / s .: . e - .---- - -- ;

,ft * . . .)

~ . . ,

a,

, - .s . .

s . .

04n . . . .

r...

~

m. .

.l.

r

. 1

• ok . ..

gu .

~

' .. ". . . ~t.tquite Yalbt A*'T **

I 9$. W W Owusa Grosy - h as Sw3 ladness.Omdar22. tML

. l

.I l

1

.- l l