ML20084P905

Provides Recommendations Re Repair of safe-ends on Reactor Vessel.Recommendations Contingent Upon Type of Failure AEC Wants to Prevent

ML20084P905
Person / Time
Site:	Nine Mile Point
Issue date:	05/18/1970
From:	Staehle R OHIO STATE UNIV., COLUMBUS, OH
To:	Harold Denton US ATOMIC ENERGY COMMISSION (AEC)
Shared Package
ML19289B377	List: ML20084P627 ML20084P712 ML20084P719 ML20084P726 ML20084P905 ML20084P951 ML20084P959
References
NUDOCS 8306130068
Download: ML20084P905 (6)

Text

r

. ; _ f, . .

f. , ,. _ u'.- '

']

/

v ..

R..W. Stachle 506311avens Corners Road G.dunna, Ohio reply to: Dr. R. W. Staehle The Ohio State Univ.

Dept. of Met. Engr.

116 W. 19th Avenue Columbus, Ohio 43210 18 m y, 1970 t l

Mr. IIarold R. Denton, Chief Tmhnical Support Branch U. S. Atcutic Energy Comtission Division of Canpliance mshington, D. C. 20545  ;

6A

Subject:

Course -of Action on Repairs for Dresden -2 at Nine Mile Point -  !

Dear Ibrold:

On May 14 you requested my recammen3ations for a course of action to be taken for the repair of safe ends on the reactor vessels of  :

Dresden -2 and Nine Mile Point. The purpose of this letter is to transntit these reammen3ations together with their justification. In addition to these reccanendations I have imluded suggestions for measures to be taken to reduce the likelihood of spurious cracking.

Prinnry RecaumeIxlations The rccarmendations here are contingent upon the kird of " failure" '

which the AEC wishes to avoid:

1. For example, if it is desircd to avoid the incideme of any .

transgranular stress corrosion cracking, then it is necessary to replace all stainless steel piping with either a higher nickel alloy, a ferritic stainless steel or a duplex alloy.

In my opinion this is an unroasomble criterion in view of the economic implications. However, it must be kept in mind that incidences of transgranular cracking are always likely in austenitic stainless steel systons ard extrane care must P

• o b

r

• t

, ".. p) (]

'.' '~

q  %)

be exerciscd to avoid this phenanenon.

2. If it is desired to avoid only the type of cracking cbserved in the first failure of the safe end at Nine Mile Point, then a different course of action is appropriate. It is my operating assumption, ard I think it is the only reacorable one, tlnt the cracking which ms chserved could have lead to a large rupture durim a thornnl transient. Ibwever, it is not appropriate hare to coment on the consequeir;es of the large rupture. Your interpretation of this is much nore expert than mine. It is my contention tint such a failure can result anl that there is a reasomble possibility its initial stages muld not b2 detectcd. Thus, it nny not leak sufficiently for it to be detceted before the major failure occurs.

If the AEC considers that such a failure must be avoided, then I recomterd tlut:

a. The intergranular made of sensitized TC b2 considered as leading to the extensively penetratcd cracking of the type observal at NG anl IACBPT(.

b. The MC should require that all safe ends b2 replaced either with solid scctions or wc3d cuerlayed inside ani outside.

c. The MC should require that a cordition for replacanent is tlut the licensee denonstrate (by test results) the adequacy of the replacanent material ard prccedure through agread upon tests. Incidentally, sane of these tests my be alrcady empleted. These tests should ,

demonstrate (l) that the new nuterial will resist ECC '

in the foua used in the replaccment ard (2) that the configuration usal will not be subject to premature failures as a result of thernnl shock, fatigue, or other appropriate n,cchanical requiranents.

d. The most reasomble replacanent nuterial is probably Type 303L weld deposited nnterial since this has had wide use as weld overlay nuterial. Acceptable nuterials muld also include Ircoloy-800 ard Inconel-600 providing that other defects are not intrcduccd. The 308L Ins the disadvantage relative to the other tm nuterials of a ready susceptibility to transgranular SCC in chloride environnents. The uncertainties in the latter tm naterials involve their behavior in the weld sensitized cordition with respcct to intergranular ECC in the mter

-Og emironncnts.

I D

r-

.. -3, s

~

e. Follo dg i

the repairs the systan mist ) (m be hydrostatically testcd.

l 1 believe t!nt in its re;ulatory capacity the MC las no choice but to rcqaire repair of all cafo ends. The Iuttern of occurrerce of failures contains elenents of frequelcy aid unpredictability which cannot be denicd. Regard] css of the detailed raticialization of the crackirg an1 attorpts by various inrties to obfuscate the issues, tle fact rcnnins that this crackiln is o'curriig with grcnt frequercy (K@ , IACIr.G, Oyster Creek) in sensitizol stainless steel, tlut the crackiny is extensive, airl tlut it was not predictcd. Fran a nuruganent point of view one nust conclude tlnt, regardless of the stress analysis or any other intermaliately amcliorating fixes or calculations, there is a high probability of failures occurring again; further, there is the reasonable prob:tbility that one of the failures nny be the open ended type.

Seconiary Rccom:nesdations There are a rurrber of other reconnen3ations which should be considered at this time. vTnile they are not as .inunliately critical as the prinnry reconnerdations atove, I believe tlnt they should be seriously considerod and chculd fonn the Insis for action taken in the near future.

1. Consideration shoald be given to the developnent of improval alloja far :_c:_vice in rc2ctc.r 7= a = =rd pipirg. Th co presently available continue to exhibit problons. Two areas in which I work might be u;rlertaken are the duplex alloys and the high l purity ferritic alloys.

2. Consideration should be given to shot peening inside arrl outside surfaces of stainless steel. This would prevent the initiation  !

cracking. hhile there are almys questions about coverage of shot peening the probability of SCC is greatly lowered.

3. Consideration should be given to an exterior coatirg which would inhibit SCC fran the outside in.

4. The criteria for leaks of valves, fittirgs, etc. in operating plants should be reviewed ard tightened. In nrf opinion there sinuld be no drippirn water of any kird anywirre ncnr the reactor or its ccmponents.

4-5.

o o The AFC shouki conluct a }rcxjram to deterntim the offect of crack extent (depth + circtrifercnLial distribation) on the role of n'echnical failure unlec aIprcyrjato loTdirg schwes.

This is a 1resscd to the gaestion: "Ibw aid un:ler wint coalitions would tha fEC crack pwpjate u2 der mcelunical loads c:qteted during service."

6. There is a reasomble imentive to determine the extent of the tran; granular !EC on the interml surface observcd by ID7L in exantinity tia interior curface of i ha secon! core spray nnzzle.

CcaTnents anl Justification in Sugxrt of Prinnry Reca'TuoMation

1. The first qacation in my niild is wbather the Alr will consider that a canplete Lrcok of any safe eM is sufficiently serious to pose a throat to safcLy. If there is no thrcut, then there is no point in pronulgating any rcqairo'l repair action.

2. I believe t]nt thare is a reasomble possibility tlnt the intergranular cracking of the type cbservcd at N@ ard UsCEtR can load to a ecmplete scuerame or substantial yaw n ing of the pipe. Such an assertion is h2scd on the fact tlut in both cases the cracks were penetratcd 50--100? of the wl1 over 1/3 of the circumfereme. Tint such a defect can occur before leaks are fcuM was a great surprise ard also a cause for substantial comern.

J. The critical questions hre relative to tha causative parameters are the lc/cls of oxygen ard stress v.nich are critical. For l example there is the reasonable possibility tlnt the cracking '

may Invc occurred in tJ:e gas pluse where radiolytically rroducal oxygen lnd accumulatul a2d which wuld be naturatcd with water. I Ilosever, for various reasons I believe this to be an unreasonable I assumption. For exanple, the deposition of copper in the IACBVR case suggests tlat vater vns in the crevice for a protracttd period of time. I suggest tlut it is, practically speaking, unreasonable to associate these failures czclusively with these gasifica regions.

With respect to the variable of stress; I believe, along with faith others, that a higher stress nny be requircd to propagate the intergranular cracks than in the case of transgranular cracks.

Ilowever, I l>2lieve it is unreasonable to cut titis effect too fine with respect to dccidizg which safe erd should be repaird. Despite the fact that stress analyses are being coMuctd on as-bailt pipes there is no clear certainty that these analyses are sufficiently necurate with respect to w1nt really cxists.

S 9

E _________--_z_----__

.O o + o j Further, there is no basis for confiderce which suggests that 4

!i these stresses will stay the same during operation. They nuy 1; either imrcase oc decroane.  ;

4 A posibly causative Dictcr mr/ be considered th  ;<

Despi.te the FIrr by Mrd, T'athis and nrjself s!S,o iryfluoride ion.

the viru]ence  !

of this contamimnt in causing SCC of sensitirol netorial, therc {

is no insis for sugqcating tlat it was uniquely caumtive in i the vide range of cases obserwd. It nust be concludul for the I -

present tlnt flueride (as well as chloride) is urdesirable but tJut it is not uniqacly causative for cracking of the haavily sensitizcd nutorial. ...

5. There is the additioml issue concerning whether the snaller cracks will propagate later. This nutter is conjcctural.

I?cwever, from the point of viea of public safety it is reasomble to assu.na tlat c<isting cracks will propagate, but at a slcraer i rate. This is the apprcach often usod in the chantical iMustry I when schaSuling the reordering of replaccment nutorials. Naturally, it is toipting to conclude otherwise depending on ones aconantic prejudices or on cne's aversion for politicalinfighting.

There were two instarces at NMP where specimens were taken -

to assess the effects of stress: The flarca region of tha first safe end ard the seccal safe cal. In both cases cracks were nh m~1 Tp

_ e n& = c'g. Let sa w cle rly cf Llu pIcpagatir.g type as opposc3 to the " disk geometry" associated with the exterml

" strawberry" examinad. Thus there was na evideme for tMsc additioral cracks at the other two locations to be " running out of gas." The possibility tlat the cracks in the flaroi region of the first safe ord were related positively to oxygen comentration must be rejectcd in vica of the narrowness of the crevice, i.e.

the high surface to volume ratio for the solution would permit the oxygen to deplete rapidly.

The observation of the lorgitudiml crack in the seconi core spray nozzle frcm the GE investigation suggest that the residual stresses nay be unprcdictable and significant.

6. The strong incentive for repair at the present time is relatcd also to the fact that repairs later will be incrmsirgly difficult cwing to the accumulation of radioactivity. A failure several years herce will be increasirgly difficult to repair and will result in groater risks to personnel tlun perfonning the repair inrr.odiately. Knile this point is obvious, it should be considered relative to the very high probability that failures will occur unless repairs are nude.

__ 1 a---'e=' E '

.t

".. - r, -

o o

7. I do not believe Llut a p cgrm of interinn inspcction 1 is sufficient to prevent the o;carre2ce of a scrious j crack. The initial cracks obscrved at IM occurral within three months. It is nusomb!c tlut a cracking system v;hich .

starts sloaly ard ruuin urdcto:tcd airly could, within a j; short tim, begin pcepgating at a very rapid rate. i l

8 There is no lucis for the fact tlnt the reactor syston must l be dymmic (i.e. Irating, conting, or otherwice contortim) l for crach to promote. The caupanents can be essentially still, and the residual stre sea 1.ould be sufficient for cracks to initiate anl prolmgale. j

9. The significance of the transgranular cracks startirn fran the l i

inside of the cafa e'd of the rwon1 core spray nozzle at  ;

IGP is rot clear. They were citarly caused by simultaneous l presence of chlorid: ions a:d c:;ygen at the 1-10 ppn level. l Ilowever, it is not clmr at dat tilue chloride was present. l The mtural yaestion here is schy vns more clchensive transgrarclar

~

not observcd in otLE parts of the systcm, i.e. in the othar core spray nozzle. This proMea could be rcasonably associated with only this seccad core spray line. Ibwever, tlus issue l

l is rnt clcar. There is an inantive for pursuirg this natter further.

10. r'amn oc thn y rtra H rn 7 u tM yen: im ef trangranular cracks frcm the outside despite the reasonability of occasiona1 contamination proLably results from the fact that t]n metal surfaces when hot are dry ard when ecol and wet are not sufficiently varm for cracks to propagate.

Concluding Rainrks The stress corrosion problea is characterizcd 17f its incrcdibility.

After all, the Imterial is " stainless" ard does not corrog,gtgay  !

occur once but after. all,an irdivigaljailure;i does not 'rchp7.4t an j

epidonic.* This subconcious belief has anesthetizcd both technical perspcctive og expert- and the practical judgment of engineering mamgcment. I belicve it is time for the AEC to rublish a decisive instruction tlut all sensitizcd

^

stainless steel must be clilain;tod. Unless such action is taken stmunrily, the roactor in3ustry jeopardizes itself in terms of a nnjor failure. That such a failure v.ould be grevious not only to the A1C tut also the country s already deficient electrical pcwer goes without saying.

While tha above pragraph as well as the entire letter is not constructcd in gaantitative terns and is rhetorical in place'it portrays the seriousness with which the problon must be considercd acted uoon.

It is never possible to inve anywhere naar sufficiently quantitative infornntion to unke such ergineering decisions.

,I urge you to iInplanent those rccaimenhtions without delay.

S erely,

• This is in jest but represents, unfortumtely gq3 lah R. . . tachle prevailing opinion.

_ _ _ _ _ _ _ _ _ _ _ _ Consultant

__