ML20085H260

Assessment of Desulfurization Procedures for TMI-1 Rcs

ML20085H260
Person / Time
Site:	Three Mile Island
Issue date:	04/14/1983
From:	Macdonald D AFFILIATION NOT ASSIGNED
To:	NRC
Shared Package
ML20083L846	List: ML20083L842 ML20084E011 ML20084E014 ML20085H248 ML20085H260 ML20085H269
References
NUDOCS 8309070455
Download: ML20085H260 (35)

Text

'.. .

~

• Attachm:nt 4

~ T.' .

- ~ ': . -. ,_

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

TECHNICAL' EVALUATION REPORT Assessment of Desulfurization Procedures for TMI-1 Reactor Coolant System By Dr. Digby D. Macdonald -

1149 Regency Dr.

Columbus, OH 43220 Submitted to:

Mr. C. McCracken .

Chemical Engineering Branch "

Nuclear Regulatory Consnission

! Washington, D.C. 20555 1 . - . _

Apri1 14,1983 I. ! #

8309070455 830825 PDR ADOCK 05000289

• P- - _  :.z; _ ,  : PDR. .

. - . , - _w _ ~ y -

e. .

.,~.

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

NOTICE Dr. D.D. Macdonald makes no warranty or representation, expressed or implied relative to the accuracy, completeness, or usefulness of the information contained in this report.

. or that the use of any information apparatus, method, or process

. disclosed in this report may not infringe privately owned rights.

Dr. 0.0. Macdonald assumes no liability relative to the use of or

! for damages resulting from the use of any information, apparatus, method, or process disclosed in this report.

No part of this report is to be reproduced for advertising without the expressed written consent of the author.

h. ,-

9 T

?

e I

e d. -* .

rd' - - -

.-1 y -v---

, t .- ~ ': ..;.. . ~

~ .

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

TABLE OF CONTENTS i

I. INTRODUCTION 1 II. ASSESSMENT OF THE PROBLEM 1 III. DESULFURIZATION PROCEDURES 9

l. Proposed RCS Cleaning Conditions 10

. 2. Consequences of Steam Generator Repaie Procedures 11

3. Performance of Simulated Desulfurization Procedures 14 IV. PROBABILITY OF FURTHER CORROSION DAMAGE DUE TO' l DESULFURIZATION OF THE RCS 15 V.

SUMMARY

ANDCONCLUSIONS 26.'

l 4

e l .

• ~

~' .

s . , ,

. . < ~-

i..- . . . . . .

I. INTRODUCTION The following Technical Evaluation Report (TER) provides an assessment of the desulfurization procedures proposed by GPU Nuclear to clean up the Reactor Coolant System (RCS) at Three Mile Island Unit 1 (TMI-1) prior to return to normal service. Emphasis in this review is placed on assessing whether or not any aspect of the

~

desulfurization procedures proposed will increase the risk of further corrosion damage .to the RCS, in particular to the Once Throuc' Steam Generators (OTSGs). The technical basis of the review reflects material that has been submitted by GPU and its representatives to NRC at various .

I meetings wnich have been held at'NRC, Bethesda, MD, and at Battelle Columbus Laboratories, Columbus, Ohio. Material submitted by GPU Nuclear in response to requests for information by this and other consultants on behalf of NRC has also been taken into consider'ation.

Finally, areas of existing concern are identified and, where possible,

, are addressed with information submitted by GPU Nuclear at the last ,

review meeting held at NRC, Bethesda, MD, April 5,1983.

II. ASSESSENT OF THE PROBLEM 4

• Late in November of 1981, extensive primary-to-secondary system leakage in both OTSGs at TMI-1 was detected during hot functional

- testing of the RCS in the reactor shutdown mode. Extensive examination of the OTSGs revealed the existence of rr.any thousands of defective tubes _

in both steam generators., .The defects detected had the following characteristics:

1 .

1 . -

. ' ~': .w..'~ . . . .

• E

.?.~. .. .. ...a.,~. .

(i) They t: d-d r.o be distributed towards the circumference of the OTSGs.

(ii) Greater than 95% were located within a few inches of the 24-inch thick upper tube sheet (UTS).

(iii) Failures initiated on the ID surface of the tubes and l propagated outwards towards the secondary-to-primary pressure boundary.

(iv) The failures occurred in the form of circumferential or part-i circumferential cracks normal to the tube dxis.

(v) The cracks exhibited an intergranular morphology which is characteristic of stress assisted intergranular attack. Although tensile st.ress has not been unequivocally identified as a necessary condition for failure to occur, the distribution of failures and their circumferential orientation normal to the tube axis strongly suggests that crack -

propagation is stress-assisted. Since the tubes are in tension nly during shutdown, it is apparent that crack propagation occurred prin-cipally dur'ing told lay-up.

(vi) Chemical and electro-optical [ Scanning Electron Microscopy (SEM)/ Energy Dispersive Analysis by X-rays (EDAX)(R), Auger Electron Spectroscopy (AES), and. Electron Spectroscopy for Chemical Analysis (ESCA)] analyses have shown the presence of sulfur-containing species l

on the tube surfaces and within the confines of cracks in the Inconel '

600(R) tube walls.

(vii) Meta 11ographic examination and laboratory tests have shown the Inconel 600(N) tube material to be in a highly sensitized state due

~

to extensive precipitation of chromium carbides along the grain boundaries.

Such microstructures are kncwn to be highly susceptible to intergranular 9

9* . . . .

3 , .

a -.  : .- .

3

:b~.

attack by partia11y' reduced sulfur species, such as thiosulfate 2

(520f) and polythionates (5,0 -) in aqueous solution under near ambient conditions. Laboratory studies on actual Inconel 600(R) tube, samples from the TMI-1 OTSGs have confinned that the tube material is susceptible to intergranular stress-assisted cracking in thiosulfate environments.

(viii) Development of a 8x1 Eddy Current testing probe has greatly enhanced the ability to detect defects over the 0.540 S.D. probe that was currently available. Fracture mechanics analyses indicate that those defects not detected by the 8x1 probe will not result in a guillotine break by mec'hanical fatigue crack growth over the life time r

. of the plant. For those defects which will propagate, crack growth is I predicted to first grow radially prior to propagating circumferential1y to a critical (unstable) size. Thus, such defects will result in a leak-before-break scenario which can be detected by chemical and radiochemical analysis of the secondary side.

Evaluation of the plant history during the shutdown period prior to

!. November,1981, has indicated that sulfur-containing species were .

l probably introduced into the RCS on more than one occasion. The three most probable sources of sulfur species are considered to be: .

(i) Sodium thiosulfate from the containment spray system. This .

l . compound was to be used to "fix" iodine (reduce it to water-soluble 131 g 133 iodide) in the event of the release of radioactive iodine (1 , ,

I135) into the containment 1 area. The dissolved iodide would then be removed by ion exchan'ge. -

(ii) Sulfur-containing oil through the reactor coolant bleed Tanks after they were contaminated by overflow from the miscellaneous waste t

I storage tank.

-. .%. . . e .. . , , . . , . , . ,, , .,---.w.w....,~.,,..-,.-m-..-y_, s,.-7 - - . . , , -- ..-w__ _ . _ , , , , , - - ..~..,.,,i

• * ~

~. .. ..- .z. . .. ~ 4-(iii) Sulfuric acid by inadvertent addition via the RCS chemical addition system.

Review of the operating history of the plant during 1979-1981 (1) indicates that some oil was introduced into the RCS in March,1979, and that rulfuric acid was added in October of the same year. However, sodium t'hiosulfate apparently accumulated in the building spray piping during 1979-1981 as the result of valve leakage. Operation of the spray pumps in June, August, and Septen6er,1981, added sodium thiosulfate to the Berated Water Storage Tank (BWST) and injection of thiosulfate into 1

the RCS occurred at least once during the September,1981, cooldown.

Although it is possible that all three sources of sulfur-containing species contributed to the existence of aggressive partially-reduced polysulfur entities in the RCS, the first two are highly unlikely for the following reasons:

l (i) Sulfur in oil exists in the form of organo-sulfur compounds, i such as thiols (RSH) and mercaptans (RSR'), where R and R' represent carbon-based organic groups. The sulfur in these compounds exist in the lowest oxidation state available (-2) and hence these compounds cannot .

contribute elemental sul' fur to a corrosion site without first being partially oxidized. Because these compounds are normally very stable, .

except under high temp'erature strongly oxidizing conditions, it is unlikely that they made any significant contribution to the polysulfur species inventory in the RCS. Furthermore, in my opinion no convincing

~

case has been put forward to show that sufficient oil was introduced to 5 account for the total amount of sulfur that is considered to exist in the l l

e

- . , .- u--- - - - - - , - - - - e,- -

.e - - -- -. .

?* . . .

..:.'.. <~ . . .

~. ,

-. .: . . . .~ . . - 5 -

. RCS--note that oil normally contains only a few tenths of one percent by I weight of total sulfur.

(ii) Sulfur in sulfuric acid exists in the highest oxidation state available (+6), and hence cannot act as a source of elemental sulfur without prior reduction to an intermediate polysulfur form. While such processes are thermodynamically possible, particularly under c'onditions which exist during hot functional testing, the reactions are extremely slow and it is unlikely that they could have proceeded to any signifi-cant extent under the conditions that existed in the RCS. Furthermore, no clear evidence is available to show that sulfate itself at l' east at the levels that could have existed in the RCS, is capable of inducing intergranular stress-assisted cracking of sensitized Inconel 600(R) ,

even in the highly sensitized state.

In view of the arguments presented above, it is apparent that the most probable source of polysulfur species was the injection of thiosulfate into the RCS which occurred at least in September,1981.

Some work (2,3) has been carried out on the ' stress corrosion -

cracking of austenitic stainless, steels, including Inconel 600(R) , in ,

thiosulfate solutions under near-ambient conditions. The findings of these studies may be sumarized as follows: -

(i) A sensitized microstructure, due to precipitation of chromium carbides along grain boundaries, is a necessary condition for the

~

propagation of intergranular cracks.

(ii) A minimum c"oncentration of thiosulfate is required for '

intergranular cracking toYesult. In the case of heavily sensitized Type 304 Stainless Steel, this concentration is approximately

. . .-l i . l

• ~

, T. .- 't . . ..- . . . .

? * *

. .~ . . , 6,,

1 x 10-6 moles /1, The equivalent minimum concentration for sensitized Inconel 600(R) is approximately 1 x 10-5 moles /1, as indicated by the U-bend tests of Newman (2).

(iii) A minimum electrochemical potential exists (s-0.3 V vs the Standard Hydrogen Electrode) below which intergranular cracking is not observed, at least in the case of Type 304 SS.

(jv) A maximum electrochemical potential exists (s+0.3 V vs SHE) above which intergranular cracking is not observed, again at least in i

the case of Type 304 SS.

(v) Crack propagation, once started, proceeds at an extremely high rate (2). Wort. perfonned at the Brookhaven National Laboratory using constant extension rate tests (CERTs) on sensitized Inconel 600(R)

(solution annealed 1135'C for 30 minutes, sensitized 621*C for 18 hours2.083333e-4 days <br />0.005 hours <br />2.97619e-5 weeks <br />6.849e-6 months <br />,

[C] = 0.04%) in. air-saturated 1.3% H3 B0 3 at pH = 4.6 containing 0.7 ppm S as Na223 3 0 has indicated a mean crack velocity of s 0.5 m/ day at

'40

• C . The 'mean crack velocity was found to increase with temperature up to 80*C, but a further increase .in temperature to 95'C resulted in a decrease in the average crack velocity.

(vi) Some evidence . exists that the addition of lithium ions to thiosulfate-containing boric acid solution of the type which is be-lieved to have existed,in the RCS, may inhibit the growth of cracks in' sensitized Inconel 600(R) . However, the mechanism by which this occurs is not fully understood, nor is the phenomenon firmly established.

It is believed that the role of thios'ulfate in inducing inter- ,

granular stress assisted cracking in sensitized Inconel 600(R) is its abil.ity to donate, by dissociation, highly active monoatomic sulfur to

(. .

s. . .-

• *

• 7 -

?~. -. . ..~... '

the chromium-depleted zones adjacent to the grain boundaries. This phenomenon is sost simply written as follows:

^ -

2 2 50-2

+ S0 - , [3) , ())

where [5] represents highly active monoatomic sulfur. The monoatomic sulfur which is deposited at the crack tip can then react with iron

- and nickel to form the respective (non-protective) sulfides Fe + [S] (2)

+ FeS Ni + [5] + NiS .

(3)

These processes effectively depassivate the crack tip surfaces, thereby

.! leading to rapid propagation of the crack. The high reactivity of elemental sulfur towards iron in aqueous solutions has been demonstrat'ed by Macdonald et al. (4), who showed that under certain conditions the reaction is autocatalytic in nature; i.e., the corrosion rate accelerates with time rather than decrecsing as is characteristics of self-passivating systems. Such a phenomenon may well account for the ,

extremely high crack velocities observed in sensitized Inconel 600(R) in thiosulfate-containing boric acid. However, it is also to be noted that reactions 2 and 3 may also proceed on surfaces exterior to the crack, in which case metal sulfides will form in the corrosion product layer.

- Extensive evidence exists that iron sulfides, and probably nickel sulfide also, &re easily oxidized in aqueous environments to form the corresponding oxides (or hydroxide's and oxy-hydroxides) and soluble species containing sulfur of varying oxidation state. One s

G r _

, ~-

< \

t.  !

7 ,

. . , . 8.,

. , . . .. z . .

example of this type of reaction can be written as follows:

2 2M5 + H2 O + 20 2MO + S 0 + 2H+ (4) 2+ 2 l

Accordingly, it is apparent that the introduction of oxygen into the environment can lead to regeneration of thiosulfate, and other equally' aggressive polythionic species, which in turn may cause the propagation

, of cracks. In other words, the cyclic sequence of reactions represented below .

2 2 -

S0-2

+

[5]+S0- (5)

[5]+M _

+ MS *(6)

+ 2

.. MS + 1/2H2 O + 0 2 MO + 2H+ + 1/252 0 - (7) 2 2 S0 (B) 2 + M + 1/22H O + O2+ M0+2H++S0+1/2S0f~ 2 results in the net production of one half of one mole of thiosulfate ion per mole of. thiosulfate consumed in the stoichiometric c.onversion of metal (M) into its oxide MO. Of greater impbrtance, however, is the fact that reactions 5 and 6 and reaction 7 are likely to be separated in time according to the availability of oxygen. Thus, during periods of low oxygen levels reactions 5 and 6, which lead to rapid crack propagation, will occur whereas introduction of oxygen into the environment, for* example during wet lay-up, may result in a burst in the thiosulfate concentration, which in turn could induce cracking

, of the sensitized OTS,G tubes. Of course, the overall stoichiometry of the cyclic process, as represented by reaction 8, indicates an eventual

. reduction in the total thiosulfate concentration in the system, but

• ~ ~ =

l

z. . .- = .

. 9

.~.., * .

~. ..

it is probable that many cycles will be required in order that the

~

thiosu? fate level will be reduced to below the critical level for

. intergranular stress-assisted cracking.

The cyclic chain of events depicted above can be broken by converting the sulfur species which exists in any one of the steps into

[ an innocuous form, such as sulfate (SOf"), which can then be removed from the system using standard ion-exchange techniques. The most effective method for accomplishing this goal is to oxidize the metal sulfide and any polythionic species (including thiosulfate) up to the +6 sulfate level by the addition of a solution-phase oxidizing agent of s'ufficient oxidizing power so as to minimize the fonnation of sulfur species of

}

intermediate oxidation state (s), which themsel'ves may induce cracking of the OTS tubes. This, then, is the technical justification for .

~

developing and implementing desulfurization techniques for the clean-up of the TMI-1 Reactor Coolant System.

II I,. DESULFUR'IZATION PROCEDURES

~

The desulfurization procedure veing considered for clean-up of the TMI-1 RCS involves the addition of hydrogen peroxide to the coolant system so as to oxidize metal sulfides and any residual thiosulfate to sulfate which will then be removed by ion-exchange. Specific details of the technique have been developed and tested by a GPU Nuclear con-tractor, and these have been presented to NRC and its consultants during several meetings over the past six months. The most comprehen- ,

sive plan for desulfurizati.on was presented to NRC personnel and consultants at a meeting in Bethesda, MD, on March 10, 1983. Also

--,% ee e

,w

,- w' v r-- -- - - * --v 'w-w ww,,wv- - - - e- - -

v e- *v

~~

~

r .. : ..

1 - . .

.10 ,. .

presented at this meeting was a report en a Third Party Review of the ~

TMI-l Steam Generator Repair activities. This third party was charged by GPU with assessing various aspects of the repair and clean-up procedures.

(1) Proposed RCS Cleaning Conditions The environment proposed by GPU Nuclear to desulfurize the TMI-l RCS consists of the following:

Boron (boricacid) 2300 ppm pH (ambient temperature) 8.0-8.2 H0 22concentration 15-20 ppm Temperature 130*F Cover Gas N 2

Lithium ion concentration 2-2 .2 ppm The pH of the system will~ be maintained within the desired range by the addition of amonium hydroxide,'and the level of hydrogen peroxide will be kept constant by the injection o.f conce.ntration H 220 into the RCS via positive dis' placement pumps. The extent'of clean-up sill be assessed on a continuous basis by analyzing the RCS environment for .

sulfate; it is estimated that complete oxidation of the sulfur in the RCS will result in a concentration of sulfate of the order of 2-5 ppm. ,

The basis for the' parameters listed above can be sumarized as follows:

(1) Oxidation of reduced sulfur species to sulfate occurs most

~

rapidly in alkaline solutions. Accordingly, a pH of 8.0 to 8.2 was 5 select'ed as being most appropriate. Hydrogen peroxide was considered t- ,,, , wi- rw,w-y-- .~. .-.

~

l.

i . .

1 l- '

t . . *

.:..'~ . . . .

.: .~:

' ~

.h. .. .. . .,

to be the best oxidant available because of its previously-demonstrated effectiveness in oxidizing reduced forms of sulfur, the lack of solid

. decomposition products, the ease with which its concentration can be controlled, and the ease with which it can be removed or neutralized once desulfurization is accomplished (e.g...by the addition of hydrogen or hydrazine or by thermal decomposition followed by normal degassing).

(ii) Hydrogen peroxide is always formed in situ during normai cool-down/ aeration of PWR plants, so that its presence is not considered to' be foreign to .the RCS.

(iii) Hydrogen peroxide is routinely added to other (We,stinghouse)

PWRs as a means of solubulizing radioactive species. Accordingly,

! extensive experience is available on the behavior of hydrogen peroxide in PWR primary systems, albeit not for Babcock and Wilcox designs. .

Nevertheless, the fact that H 0 has been used in PWRs at approximately 22

~

the same concentration levels as that proposed for the clean 'up of the TMI-1 RCS provides confidence that the cleaning procedures will not have unexpected deleterious effects on the materials of construction .

in the TMI primary circuit. This question is addressed in more detail later in this report.

(2) Consequence of Steam Generator Repair Procedures -

Although no attempt is made here to review the steam generator repair procedures being carried out by GPU Nuclear on the TMI-1 OTSGs, certain aspects require comment because they have a bearing on the possible effectiveness of the RCS clean-up activities.

T . . - .

i: . - -

.5~.

-. - . . .~.... 12 .

. Repair of those tubes in which defects are located within the

, crevices of the upper tube sheet are being accomplished by explosive expansion of the tubes to form a new tube--tube sheet seal. Detona-tion of the charge results in two phenomena of direct relevance to this analysis:

(1) Deposition of a layer of organic material on the tube ID sur- ,

face which may hinder desulfurization of the underlying corrosion product film.

(ii) Creatio'n of a plastica 11y deformed region (including transi-tion zone) in which high residual stresses will exist. It is important to note, however, that stress is only one of the three components I

that are normally considered to be necessary for stress corrosion

~

cracking to occur; the others being a sensitized microstructure and an aggressive environment. Clearly, a sensitized microstructure already exists because the original tubes are still in place in the OTSGs at TMI-1. Therefore, it is necessary to remove the aggressive environment in order to prevent the reoccurrence of intergranular stress-assisted -

cracking. ,

Recent work by a GPU Nuclear contractor on tube specimens that were expanded under conditions which very closely simulate those which ,

exist in the OTSGs (principally the same surface to gas volume ratio) shows that a polypropylene film is deposited on the metal surface as

~

i the resuit.of kinetic (explosive) expansion of the tubes. In simu-lated clean-up tests," employing realistic hydrogen peroxide levels in +

solution, it has been found that in excess of 350 hours0.00405 days <br />0.0972 hours <br />5.787037e-4 weeks <br />1.33175e-4 months <br /> of exposure will be required to remove the deposited film. This time is well in 9

-v -

r - -- , n e , - .

lt .

• ~. ... ; ,, -

I ?*,~.

.: ..- .z. . ~ . .

, ' 13 .

excess of that which is probably required to desulfurize an organic film-free surface, and hence will determine the maximum exposure time required to effectively desulfurize the system. It is important to recognize, however, that the converse may also be argued to hold true;

. that is, the polypropylene surface film may inhibit the release of deleterious sulfur species from the underlying corrosion product film l and hence that these surfaces will contribute less rapidly to the build-up of polythionic species in solution upon exposure of the system to oxygen.

It has been postulated that the thin polypropylene film which exists on the tube surfaces will undergo thermal degradation as the temperature is raised during hot functional testing of the RCS.

Seymour (5) lists the maximum resistance to continuous heat for bulk' polypropylene (unfilled) as being 100*C and that f6r tale-filled polypropylene as being 120*C. These values are nonna11y a few degrees lower than the I1 eat deflection temperatures under a flexural load of

- 264 psi, as defined by the ASTM Test D-668 which 1.s a useful measure of the tendency of a plastic to sag under load. In the case of thin ,

films, however, the maximum temperature that can be endured is expected

~

to be somewhat lower because of the absence of an extended three- ,

dimensional structure, particularly in the dimension normal to the surface. Because polypropylene is a hydrocarbonrit 'is also hydrophobic.

Accordingly, exposure of the film to high temperature aqueous environ-ments (probably in ex~ cess of 100*C) should result in the transfonnation -

of the film into three-dimensional globules; a process which will be

'drivien by the thermodynamic tendency of the hydrophobic film to reduce 1 - . .

L .

l . , . .- .. _. .

7

~

, T.

• ~. -. ..- .. . .

14 -

. its surface area-to-volume ratio. Thus, a possible means of degrading the protective properties of the deposited polypropylene surface film would be to briefly expose it to high temperatures before attempting desulfurization with hydrogen peroxide. This, of course, should be eveheated experimentally before any attempt is made to institute this procedure for clean-up of the RCS.

(3) Performance of Simulated Desulfurization Procedureq, The ext'ensive laboratory work that has been carried out by a GPU Nuclear contractor has provided considerable data on the effectiyeness

of the desulfurization procedure, at least under simulated RCS ctsnditio- .

} The principal findings of this work can be sumarized as follows:

(i) Complete, or rearly complete, oxidation of sulfides to sulfate is possible.

(ii) Dissolution of sulfur species on the surface of tubes taken from the TMI-l GTSGs has been demonstrated.

(iii) No sulfur compounds of intermediate oxidation state [i.e.,

between -2(sulfide) and +6(sulfate)] have been detected by hexane

~

extraction /UV-Visible absorption studies. Although the sensitivity of this technique is of the order of 2 ppm the lack of positive results indicates that intermediate sulfur species, once formed, are rapidly l ~

oxidized to sulfate. Accordingly, the mean-life-time of any polythionic species which could lead to intergranular cracking of sensitized Inconel 600(R) in the s'ystem ,is presumabl) very short, and hence would f.ot pose _

a significant threat to the . integrity of the steam generator tubes.

, v v *

+

s 6

a

1. s - x t

l

.- L:~. . . . ,

(,

Although the tests performed so far are most promising in terms of the efficiency of the procedure for desulfurizing the RCS, a number of shortcomings have been identified. These include:

(1) An apparent failure to recognize the importance of sulfur pick-up from the atmosphere, at least in early experiments. Thus, absorption and oxidation of sulfur-containing species from the atmos-

- phere (most probebly 50 2

) represents a significant fraction of the total sulfate levels observe'd in thc Jaboratory tests, and hence the actual efficiency for removal of sulfur from the steam generator tubes remains in scce doubt. 'It is our understanding that many of these tests are being repeated in order to exclude sulfur pick-up from the atmosphere.

(ii) To date it is apparent that the laborabory studies have been carried out on THI-1 OTSG tubes that are not heavily contaminated with sulfur. However, heavily sulfur contaminated tubes which realistically simulate those in the OTSGs (i.e. sulfur contamination by exposure to s

thiosulfate') have been tested, and the cleaning technique has been found to be effective in removing the sulfur, contaminant.

n IV. PROBABILITY OF FURTHER CORROSION DAMAGE x :DUE TO DESULFURIZATION OF THE RCS A question of major concern in the clean-up activities'is whether ,

or not hydrogen peroxide desulfurization may result in further corrosion damage to various components in the Reactor Coolant System. Any answer V .

to this question must be given in light of the probability of corrosion 4 '

> damage continuing to occur due to sulfur which already exists in the RCS

.x c in the c.yent that desulfurization procedures are not ir.stituted. Becauso

- - . ~

i T

• 4

~ ~

N ,

k , s t .

~

T.

.. i.. . -

16

~. .

the proposed desulfurization activities are unique to TMI-1, it is not possible to ' provide an unequivocal answer to the first question at this l time. Instead, the opinion expressed below is based upon previous experience with H22 0 additions to PWR primary systems, and on experi-mental work that is currently being carried out by GPU Nuclear.

The pertinent findings to date with regard to the possibility of corrosion damage being indyced by the cleaning procedure itself may be summarized as follows:

(i) Hydrogen peroxide is always fonned in PWR plants during cooldown and aeration. .

(ii) Hydrogen peroxide is routinely added to other PWRs ,

} (Westinghouse) at similar levels to that proposed for desulfurization of TMI-1 as a means of solubulizing activity during shutdown.

, (iii) No corrosion problems associated with H 022 additions to these PWR primary systems apparently have been identified.

(iv) Simulated desulfurization procedures in the laboratory have apparently not produced any corrosion on sensitized Type 304 SS or Incorel 600(R) .

-(v) Hydrogen peroxide cleaning tests carried out in the laboratory on THI-1 OTSG tube samples apparently have not produced any signs of .

corrosion damage.

l- .

Although additional work is now being carried out by GPU Nuclear on other TMI-1 primary system materials, including Zircaloy-4, Type 410 SS, 17-4 PH SS Inconel X.-750(E} ,

Inconel 82 Weld Metal (R) , and Type 308 SS ,

Weld Metal, the findings to date indicate that the probability of i

P l .

i -

2. . .

17. .

..z., ._

corrosion damage resulting from the cleaning procedures themselves must be rated as being very low.

With regard to the possibility that further corrosion damage will occur due to sulfur species which already exist in the RCS, in the event that desulfurization is not carried out, it is important to recognize the following:

(1) The oxidation of metal sulfides to fonn polythionic species in solution under near-ambient conditions is well recognized in inorganic chemistry, and is known to occur in other industria1' systems (mining.

oil and gas processing). , .

(ii) Laboratory tests and previous TMI-1 experience have demonstrated that intergranular stress assisted cracking of TMI-1 OTSG tubes can .

occur in the presence of only very small concentrations of sodium thiosulfate.

(iii) Laboratory tests on sensitized Type 304 SS in t.i osulfate-containing ' boric acid have shown that this material is also extremely,

. susceptible to intergranular stress, assisted cr'acking at very low .

thiosulfate concentrations (%10-6 moles /t). Sinc.e the RCS contains Type 304 SS, some of which is believed to be sensitized (see below), a significant risk of failure probably exists if the concentration of ,

polythionic species is, ellowed to increase above a critical level. ~

(iv) Extensive examinations of the RCS and related systems for evidence of further sulfur attack have now been reported by GPU Nuclear (April 5,1983, meeting, NRC, Bethesda, MD). A review of operating .

history of cracking of the Spent Fuel Pool Cooling Pipe at weld Heat-Affected-Zones (HAZ) revealed that cracks occurred in stagnant, borated

%= . .. .. ..

. c: .,

I . .- - -

.- . .- .e ..,, . .

18 -

. and oggenated environments at near ambient temperature. Although

, partially reduced sulfur species were not specifically identified as being the cause of the failure in this case, attempts to crack heavily sensitized Type 304 SS in uncontaminated spent fuel pool environments at near ambient temperatures.have been unsuccessful. Since the pool medium is the same as that which exists in the RCS, it is possible that thiosulfate was introduced *into the system at times earlier than the September 1981 cool down referred to previously in this report. The cracked sections 'of pipe were replaced with Type 304L; a grade which is not susceptible to sensitization on welding.

(v) Intergranular cracking of Waste Disposal Gas (WDG) piping.was l }

detected in 1982. The cracks apparently were associated only with weld heat-affected zones, and the cause of the failures was attributed to

" localized sulfur assisted IGSCC in HAZ."

i (vi) Corrosion damage to the PORY (Power Operating Relief Valve) has also been noted. Thus, although no corrosion damage was noted during a 1979 refurbishment 'of this component, examination in 1982 revealed pitting corrosion on various martensitic (Type 410 SS) and .

Inconel X-750(R) components. The valve was replaced in April 1981, and the replacement valve was examined in February 1983. Again general and -

pitting corrosion of Type 410 SS and Inconel X-750(R) parts were

, observed and, significantly, pure crystalline sulfur 'and " sulfur compounds" were found on the PORV body and parts. High sulfur deposits were also found on the block va'1ve, but no unusual' corrosion damage was noted, '

possibly because of the absence of components manufactured from Type 410 SS or Inconcel X-750(R) . The block valve was cleaned, inspected, m

T

' ~ -

_. ..- .. . ..- 19 .

and then reinstalled, as was the PORV, except that in this case the internal parts were replaced. GPU Nuclear concluded that the observed damage occurred after 1979 but prior to or during the 1981 hot func-tional testing. There can be little doubt that the primary cause of failure was the presence of partially reduced forms of sulfur in the system.

(vii) Examinations were also conducted by GPU Nuclear of the pressurizer valve area. Valve RC-VI, which serves the pressurizer spray, was found to contain sulfur deposits, but did not exhibit any obvious corrosion. The pressurizer vent (Valve RC-V17) exhibite'd minor pitting, as did Valve RC-RVB (Safety Valve) and Valve WDG-V1 (RCDT relief valve). Safety valve RC-RV1A did not show any corrosion damage.

I'n all cases, the corrosion damage was conside ed to be minor and not'to affect the valve integrity or function. However, the presence or elemental sulfur, at least on one of the valves, strongly implies that the cause of t$e corrosion damage was th,e prese,nce of partially reduced forms of sulfur in the' local environmerit.

.(viii) Concern had been expressed at the March 10, 1983, meeting ,

between NRC staff and co'nsultants and GPU Nuclear personnel that damage l might have resulted to the control rod drive (CRD) mechanism (s). ,

Consequently GPU Nucle'ar personnel examined as many components as was practical, including the leadscrew,the motor tube (Type 304 SS), and the and cap (also Type 304 SS) in order to determine the extent of any damage. The first two components were apparently examined visually,

~

whereas samples were taken from the end cap for a more thoro. ugh .

l meta'llographic examination. No evidence of corrosion was found on the

< ~-

[, " ~ - 0 leadscrew, but sulfur deposits of unknown fom, were observed. No evi-dance of corrosion or cracking was found on the motor tube or on the end cap. The fact that sulfur deposits were found in .the CRD, albeit of unknown fom, indicates that partially reduced sulfur species pene-trated the CR0 mechanism sometime during the period of contamination of i

l the RCS with thiosulfate.

(ix) During the past two months, extensive examinations have'also 1

been carried out of other RCS components, as summarized in Table 1.

The techniques used in these inspections included dye penetrant testing, wipe sampling,, radiography, ultrasonic surveying, visual or, video j examination, functional testing, and/or metallography, in addition to I eddy current testing as described previously. No indications of damage were found on any of the components listed in Table 1.

(x) Although the examinations conducted so far of the RCS must be considered extensive and comprehensive, GPU Nuclear have indicated that the inspection activities are continuing. The components which are now being examined include the Pressurizer internals (spray pipe and ,

nozzle, shell, ladder welds, and heater bundle), RCS piping (hot leg .

vent,' pressurizer vent), and auxillary systems including the make-up 4

tank relief valve and nozzle. These components, together with those ,

which have been examined to date, are shown schematically in Figure 1.

It is interesting to note that sulfur and sulfur-induced corrosion damage has been observed 'in regions of the RCS which have not been 5

exposed to a liquid environment (e.g., general corrosion and pitting in the PORV and cracking in the WDG piping). This indicates that besides i

l .

SODIUJ s' ; i d

IH3 -

WMD -

@s'SSSS%SSj

~

g AAAAAA //f f ,

SWST - !

j [ =^

d,/, WASTE GAS DUllDING SPAAY j

/

{- n PUMPS  ; f

$$$. *%'. 's's's'i <d Y $ sis $

s l (vtst non) ,

, - , ' " ". ' ,{ .

0 ( ) M 6'fA VA W A W C VAO '?

~

MISC WASTE I I I **I TAN'K Bt610 TANKS j .!

e; n ' , ~ 'h.' ;l

@(

{ DeMin )- 'o S:

f . __

A

.b,

~

I

{ orMiu)_

MU1 '.

-j i vV I f ,RCP HPI f# 4, 4 -

s u a n

++

a v I

i ll) rf i,a

-b/VV! >+

MH,%%%'%%%

& S j4) v hf, ~b^^'+ -n .

D

- .,449/

H mx '

i m g> g M M 3]l ""'"""8 g ,,- -; - -

l t t 'czzszzzzzszzszzszczzzzzzeeeeeeeeeeecess

~

.! !s 1 j w v e' g!g'j .

Inspections O TO BEINSPECTED .

/ /

INSPECTED .

I PUMPS .N Figure 1. RCS and related components which have been (shaded) or have not been (unshaded) examined for ,

sulfur-induced corrosion damage. .

$ 8 . e 9

et

a*

• f 1 . . < ~-

[. ~ -

. 23 .

thiosulfate, which can only exist in the dissolved state (note that it isanion), a volatile polysulfur species must be present in the system.

The most likely species .in this regard are the hydrogen polysulfides, H32 x (x=2, ... 5). These species are slightly stronger acids than hydrogen sulfide K)

H3 2x HS~ x

+ H+ -(9)

K

' ~

2 HS' x

g S-x

+ H+ (10) 1 with K) % 10-5 5 vs l d for2 H 5. Accordingly, at ]H- values of 'six and below, which is characteristic of boric acid, a significant (and even -

dominant fraction) of the total polysulfide content of the aqueous phase will exist in the fully protonated form H 32x . Like the first member of the homologous series (hydrogen sulfide, x=1), the hydrogen polysulfides are volatile and may readily enter the gas phase from the solution phase.

Another characteristic of the hydrogen polysulfides is that they are.

A very labile compounds; that is they tend to undergo rearrangement to '

other foms. The most important reactions, as far as the present discus-sion is concerned, involve the reversible pick-up and deposition of .

elemental sulfur

. Deposition

. H3-2x H3 2 x-1 + S(elemental)' (11)

Pick-up, Thus, in many ways the polysulfides (including the anions HS, and 2

S -) mimic the role that has been attributed to thiosulfate in the l

RCS/0TSG corrosion phenomena,; i.e., they act as a source of highly l

G

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

c ... .

l- . . .

< ~': ... . .. . - - . .

1. ' - ' 3A

.{.~. . . . - ...;...~ .

active monoatomic sulfur which induces intergranular stress assisted cracking in appropriately-sensitized microstructures. Therefore it is interesting to conjecture that much of the reduced sulfur species that existed in the RCS may in fact have been the hydrogen polysulfides.

A variety of mechanistic paths are available for the formation of polysulfides from thiosulfate. The most probable paths include the heterogeneous reduction of thiosulfate with subsequent reaction with a second thiosulfate ion in a step-wise fashion to form various members o.f the polysulfide series:

2 2 S0 2

+ M + MS + S0 - (12)

MS + HO 2

+ MO + H S 2

(13)

+3

• 2 H5 2

2 H32 2 + SO - (34) 2 H3 22 + 30 23

+ H323+SOf (15)

I .

2 H324 + 30 23

+ HI2 .i + SOf (16)

Alternatively, homogeneous reduction of thiosulfate in the high temper-ature hydrogen-rich environment that existed during hot fdnctional testing would also appear to result in the formation of the hydrogen polysulfides:

e

- g e S

..~ .

'. ~

t . .- ~-

25 2

S0- + H + H5 + SOf (17)

, 2 2. 2 2

H3 +' + (18) 2 32 0f~ H32 2 + S03 l, l 2

H3 24

+ 30~2

+ H3 25 +SOf" (19) or, if the reducing conditions were sufficiently strong, ,

52 0f + 2H+ + 4H (20)

+

2 2H 32 2 + 3H 2 O 2

H322 + 30- 2

+ H3 23+SOf" (21)

! It is important to note that the first mechanisms does not require ,

i the presence of mol'ecular hydrogen in the system, so that polysulfides could form during cold shut-down conditions. However, because it is ,

necessary to produce sulfut in the -2 oxidation state as a precursor to polysulfide femation, and because the reactions are expected to pro-caed more rapidly at higher temperatures, the formation of the poly-sulfide species during hot functional testing appears to be an attractive .

possibility. ,

From the above, it is apparent that damage has already resulted to components in the RCS other than the OTSGs due to residual sulfur contamination. Because the sulfur contaminants still remain in the system, we believe it to be prudent to assume that corrosion will continue. Accordingly, it is necessary to clean (i.e., desulfurize) the reaction coolant. system. -

t 8 9

2.

'1

~

.~;. <

' 26 -

?' ~. -. ..- . . .

V. Supe %RY AND CONCLUSIONS Based upon the information that has been made available to this consultant up to the time of preparation of this report, the following conclusions can be drawn.

- (i) Intergranular stress assisted cracking of the TMI-1 OTSG tubes most probably resulted from contamination of the RCS with sodium thiosulfate.

(ii) The steam generator tubes, and possibly other components in the RCS, are in a highly sensitized metallurgical state and are therefore predisposed to cracking in the presence of low levels of poly-thionic species. .

(iii) Post-failure analyses have demonstrated that the steam generator tube surfaces are contaminated with sulfur of various oxida-tion states.

(iv) The observation of sulfur deposits and sulfur-induced corro-sion damage on components which have not been e,xposed to the aqueous thiosulfate contaminated environment indicates that a volatile poly-sulfide entity is present in the system. This entity is most likely one ,

or more of the hydrogen 'polysulfides (H2x b ) which may form from thiosul-fate by direct reaction with metal (e.g., Fe or Ni) or by reduction by ,

hydrogen during hot functional testing. The volatile hydrogen polysulfides are known to be highly labile, which would account for the formation of deposits of elemental sulfur.

(v) A significan't risk exists.that.if desulfurization is not ,

carr'ied out, metal sulfides and possibly other sulfur-containing species

'i n t'he corrosion product film will undergo partial oxidation during wet i

ed

~

l.- j ,- .. , ,

~

Il

$~. ..

lay-up or shutdown conditions to fonn polythionic species in solution. I 1

Because these species may lead to a reoccurrence of intergranular stress assisted cracking of sensitized materials in the RCS, it is recommended that procedures be adopted to desulfurize the reactor coolant system.

! (vi) Preliminary experimental data indicate that alkaline peroxide is an effective medium for desulfurizing TMI-1 OTSG tubes. Recent work i

has shown that this cleaning technique is capable of removing sulfur from corrosion product films which are covered with organic films as a result of the steam generator repair activities. Some concern still exists as to the viability of the technique for removing sulfur which is deeply embedded in the corrosion product film, since specimens of this .

I type apparently have not been evaluated in laboratory tests at this time.

However, the technique has been shown to be effective for cleaning I Inconel 600(R) specimens that were heavily contaminated in simulated thiosulfate contaminated SG environments.

(vii) Available data indicate that the desulfurization procedure itself does not pose a significant risk with regard to corrosion of ,

materials in the RCS. However, the various materials that are specific to the TMI-1 RCS are now being tested in simulated desulfurization ,

environments, so that an unequivocal statement in this regard must .

! await completion of the experimental work by GpU Nuclear.

l 1

_ _ - - - - - + - - - - - , _ _ , - y .,.--.-..<-.%. . - - - . - + .- , , , - ,,,c. ---, , ,.,-,--,.v.,_m--.-- - . . _ - , , . - - - . _ . , , __,..,-e,.- ,.

. l

. c. ., .

T. .-  : -

.. ... ..c .& .

. ..- . 28 '

REFERENCES

1. Submission by GPU Nuclear to ACRS Subcomittee, "TMI-l Steam Generator Status," June 7,1982,

2. R. Newman, Brookhaven National Laboratory, Data presented to NRC Consultants,1982.

3. D.D. Macdonald and G.A. Cragnolino, " Corrosion and Corrosion Cracking of Materials for Water Cooled Reactors," EPRI Progress Report, Project RP1166-1, FCC 7806 July 1-Dec. 30,1980.

4. D.D. iiacdonald, B. Roberts, and J.B. Hine, Corrosion Sci,18,, 411 (1978). .

5. Raymond B. Seymour, Plastics vs. Corrosives, SPE Monographs, John .

Wiley & Sons, New York (1982).

~

l l

l .

l __

v Attachmtnt 5

! INTRA LABORATORY CORRESPONDENCE OAK RIDGE NATICN AL L ABORATORY 4

August 17, 1982 To: R. N. McClung From: C. V. Dodd d . V. Oeeld

Subject:

Travel to Harrisburg, Pennsylvania, August 8-9, 1982 e

On Monday, August 8, I met with General Public Utilities (GPU) and NRC personnel to discuss the eddy-current inspection of the Three-1111e Island Unit i steam generators. A partial list of attendees is attached.

Nick Kazanas of GPU gave a presentation of the development and qualification program for the eddy-current inspection of the steam generators. A set of inside diameter calibration standards with circumferential notch lengths of'O.060, 0.100, and 0.187 in., notch depths i of 20, 40, 60, and 80%, and widths of typically 0.004 to 0.005 in. were

, constructed. In addition, an axial notch standard with 0.060 in. long notches was constructed.

4 These standards were used to test circumferential differential probes of

) 0.510 and 0.540 in., and an eight-coil array with 0.187 in. outside diameter pancake coils. Various gain and other conditions were run. The best combination for the differential coil system was the 0.540 in.

outside diameter probe wich a permanent magnet saturating, core and a gain setting of 60. The permanent magnet should not have increased the signal any but only reduce the noise a smal) mount. I examined a probe and its field did not seem strong enough to saturate any ferromagnetism associated with Inconel (about 0.3 T is usually needed). However, it did seem to be more carefully made than the regular 0.540 in. probe. An additional mix of the 200 and 400 kHz mixed signal with an 800 kHz signal reduced the noise due to probe, chatter and pilgering.

2 The 8X1 absolute probe showed more signal to the small defects at a gain of 53, but a fairly large lif t-off signal was also present. The 8X1 probe array is being revised to increase the probe body from 0.520 to 0.540 in, which should reduce ,ti)e lif.t-off , problem.

A correlation of the defects seen by the pancake coils with the defects seen with the 0.540 in. outside diameter differential probe showed that of 3233 defects detected by the pancake coil, 3216 were also detected by the t

differential probe. This number was improved to 3229 by using the mix to reduce the inside diameter noise. This shows an excellent match and also showed that the 0.540 in. dif ferential probe, operated under these conditions, can reliably detect the same type of circumferential crack.

In order to directly apply the results obtained from the electrodischarge ,

~

machined standards, an " effective axial crack width" for. these intergranular stress-corrosion cracks aust be determined.

/

l r?";'.'f. '

t R. W. 'kClung Page Two August 17, 1982 On Ibnday af ternoon, John Janiszewski of GPU Cave a presentation of the results of the metallography of the cracks. Some cracks appeared to have regions of bulk intergranular attack associated with them, and some appeared to be very narrow with very little branching and axial component.

However, the crack would only need a few branches to effectively disrupt the flow of eddy currents.

-o John Janiszewski Tdll generate an " effective axial component" by reviewing the results of the previous metallography and furnish it to us . This number will show how applicable the calibration results from electrodischarge. machined notches is for the circumferential cracks, and furnish an independent verification of the ability to detect the defects with a different probe. The results of a dimensiocal analysis experiment using large scale models at ORNL will be used to correct the sensitivity at one length to the sensitivity at another. It was stated that the region near the crack was depleted in chrome, but no estimate was made of the bulk electrical and magnetic properties of the region. A total of 19 ft of " good" tubing has been examined by metallography, with an additional 6 ft to be examined. No defects were detected by this test that were missed by the eddy-current test. Some of the defects detected by the eddy currents were not found by the metallographic examination, prubably due to the way the samples were cut. Some of the eddy-current signals turned out to be due to manufacturing, handling, and assembly artif acts, and would not be detrimental to the service of the tube.

On Tuesday I visited the data reduction site at the Host Inn near the plant. I reviewed the results of the scan on tube A71-126. This tube was pulled and a section sent to ORNL for examination. A through-wall defect was detected using a high-frequency (5 MHz) scan with a small (0.020 in.

mean radius) probe from the outside. The defect was then etched and showed a 0.005 to 0.010-in.-wide affected region on the outer surface. It is not known if this entire region appears as a low conductivity region or not. The defect was recorded as ,80% through-wall by ConAm, and a blind remeasurement of the defect from tape showed 84%. The magnitude of the -

signal was 1/2 V at a gain of 34, and an 0.510 in. outside diameter differential probe was used. .

I also looked at runs using' the 0.540 in. outside diameter dif ferential probe with a gain of 60. The practical noise level to get a reasonable measurement of the defect depth appears to be around 0.5 V, although in many cases smaller defects can be measured. Based on the 0.005 in wide standards, this falls in the range of a 0.060 in. long defect, 40%

through-wall. Depth measurements on smaller defects will probably be very inaccurate.

A 100% inspection of the full length of all the tubes is being perf ormed -

using the 0.540 in. probe with a gain of 60. The number of indications is about threa times as many as were observed with the 0.510 in, dif ferential probe. Tube B10-48 showed 16 inside diameter defects, all about i V in amplitude, between 30 and g0% of the wall. The lowest defect was near the third support plate. The high sensitivity of this new inspection is also

l -

, l t #. ,

R. W. !!cClung Page Three August 17, 1982 picking up a number of outside diameter signals from the manufacturing l process, which are not detrimental to the service and can be ignored. The tubes that exhibit the inside diameter signals should be rescanned with l

the 8X1 pencake coil array, and the tubes with defects greater than the l plugging limit plugged. The tubes with defects below the plugging limit i

can be reexamined at later intervals to monitor growth of this type of defect.

There appears to be a drif t and a quality assurance problem with the probes. This problem doesn't affect the accuracy of the test, but results in frequent probe changes. The inclusion of several of these types of defects in an in-line standard (for future tests) should be considered.

The instrument gain in the field is set to give a repeatable voltage amplitude from a drilled hole flaw, rather than an absolute number.

The pancake coil array is operated at a single frequency and much more susceptible to different types of noise than the differential probe. This array also requires much more equipment than the differential probe.

Zetec is working on a more compact system, but no estimate of the availability of this instrumentation was given.

The results that I saw on Honday and Tuesday answered all the previous questions that Emmett Murphy had submitted in his letter of April 12, 1982, except the one on safety evaluation. Some attempt should be made ,co determine how large a defect would have to be before it would present a safety problem. I feel that the defects that can now be reliably detected are much smaller than those that would present a hazard, but have no information to back this up.

The study, done by the utility, their contractors, and the EPRI NDE Center to determine their sensitivity limits, was. outstanding.

CVD:jlb -

cc: R. Barley, CPU J. H. DeVan -

L. Frank, NRC/ .

J. C. Criess ,

F. J. Homan N. Kazanas, GPU A. L. Lotts C. NbCracken, NRC J. Ibscara, NRC P. Patriarca C. M. Slaughter J. H. Smith -

P. Wu, NRC.

C. V. Dodd/ File

/

v

Attachment 6 dia .

Burns and Roe,Inc. f 800 Kmderkamack Road e Oradelt. NewJersey o7649 m Tel. N.J.(201) 263-2000 Teles *130353 eCable BURNS ROE ORA i

May 16, 1983 .

Mr. R. F. Wilson Vice President, Technical Functions GPU Nuclear 100 Interpace Parkway Parsippany, NJ 07054

Dear Mr. Wilson:

Attached is the supplementary report of the Third Party Review of the TMI-l Steam Generator Repair Program.

This Review was requested originally by your letter of April 12, 1982 and has been conducted in accordance with the Charter you provided. An interim report was issued on September 27,1982, and an intended final report was issued on February 18, 1983. The February 18, 1983 report contained several connents and reconnendations. You subsequently asked that the Review Group continue its review of the Repair Program and address specifically the GPU Nuclear responses to the February 18 report.

The Review Group continued its review in meetings on April 12 and 13 and follow-up discussions with GPU Nuclear. The results are contained in the attached supplementary report dated May 16,1982.

We believe all of our connents and reconnendations relating to safety of the steam generator repair have been satisfactorily resolved by GPU Nuclear. We therefore consider this Review complete.

Sincerely, 7, t. * ' - '

ClE.'J.}t Wagner

. dA. .-

~

EJW:jb

Attachment:

" Report of Third Party Review of Three Mile Island, Unit 1, Steam Generator Repair", Supplement 1, dated May 16, 1983

. cc: P. R. Clark, GPU Nuclear -

J. Wetmore, GPU Nuclear i Third Party Review Members i

. - - - . , , , - , - . . . - , - . . , _ _ _ _ ...r . ,. - , _ - , _ . , - , - _ _ . _