ML20129C364

Rev 3 to Assessment of TMI-1 Plant Safety for Return to Svc After Steam Generator Repair

ML20129C364
Person / Time
Site:	Three Mile Island
Issue date:	09/04/1983
From:	Moran T, Wald P, Wilson R GENERAL PUBLIC UTILITIES CORP.
To:
Shared Package
ML20127A342	List: IA-84-780, Forwards Listing of Util Responses to Generic Ltr 82-22 Re Economic & Worker Safety Aspects of Steam Generator Tube Degradation Problem.Supporting Documentation Encl ML20129C199 ML20129C244 ML20129C285 ML20129C343 ML20129C364 ML20129C398
References
FOIA-84-780 NUDOCS 8506050485
Download: ML20129C364 (165)

Text

1 '

4 *,

< T~n;

,Y~ f$$ /,. / N4UAAY dCff 4/d y H. Silver ASSESSMENT OF TMI-l PLANT SAFErf FCR RETURN TO SERVICE AFTER STEAM GENERATOR REPAIR TOPICAL REPORT 008 REV. 3 PROJECT NO: 5000 51712 T. M. MORAN AUGUST 19, 1983 APPROVALS:

bN  ? J'b Manager, Plant Analysis Data he?h!w." k - fI-L' --[ b Di-=ctor, ysgems Engineering Dace t

s V -

Y .i h 93 D Vice P:ssiden: DAte Technical Functions 0506050485 841217 PDR FOIA DETJEN84-780 PDR

/d

. I, _F , _

fg j..: ,- ,

't -h

. . s

- Table of Contents

'I. INTRODUCTION. .

A. Purpose T.. Background

. C. Steam Generator Repair Program D. Safety Evaluation Logic

~

II. FAILURE ANALYSIS A. Operational History B. Metallurgical Test Program C. Corrosion Test Program D. Damage Scenario E. Distribution of Damage III. CORROSION TEST PROGRAM A. Introduction

5. Corrosion Mechanism Determination Tests C. Corrosion Scenario : Verification Tests D. Repaired Tubing Corrosion Tests E.= Conclusions -

IV. - PREVENTION OF RECURRENCE A. Introduction

3. Prevention of Future Chemical Contamination -

C. Changes in Operating Chemistry -

D. Cleanup of Sulfur Erem Tubes E. Conclusions V. KINETIC EXPANSION REPAIR DESCRIPTION

SUMMARY

A. Description of Process and Geometry B. Design Bases of Kinetic Joint C. Qualification Program D. Repair Testing E. Post-Repair Testing F. Conclusions VI. E7?ICTS OF EXPANSION REPAIR A. Possible Introduction of Chemical Impurities

5. Possible Effects on OTSG Structure C. Corrosion D. Effects of Expansions on Existing Plugs E. Relief of Preload

~ F. Conclusions VII. PLEGOING REPAIR DESCRIPTION

SUMMARY

A. Introduction B. Plug Types C. Plugging and Stabilization Criteria _,

D. Post Repair Testing E. Conclusions

~

i

. c . ,3 . , o,.

e a Table of Contents (cont'd)

VIII. COMPARISON OF TUBE PLUGGING WITH DESIGN BASES A.- Introduction B. Operational Performance C. Accident and Transient Performance D. Moisture Carryover Considerations E. . Con lusions IX. UNREPAIRED PORTION OF TUBES A. New Damage Not Occurring

5. Defect Detectability C. Undetected Defects D. OTSG Tube Failure Analysis for Unplugged Tubes E. Conclusions X .~ OPERATIONAL CONSIDERATIONS A. Primary to Secondary Leakage B. Radiological Concerns C. Secondary Side Chemistry .

D .' Development of Procedural Guidelines for Steam Generator Tube Rupture E. Conclusions -

XI. ENVIRONMENTAL IMPACT A. Introduction .

3. Of fsite Dose Estimates .

C. Exposure Estimates D. Sampling and Monitoring E. Conclusions XII. TECHNICAL SPECIFICATION COMPLIANCE XIII.

SUMMARY

AND CONCLUSIONS ,

APPENDICES A. Precritical and Post-Critical Test Programs j B. Response to Questions REFERENCES .

FIGURES Figure I-I TMI-1 Steam Generator

. Figure I-2 OTSG Task Organization -

Figure 1-3 Disposition of Tubes in TMI-1 Steam Generators figure I-4 Kinetic Expansion Process

_gure I-5 TMI-1 Steam Generator Typical Cracks Figure I-6 Kinetic Expansion Length Figure I-7 Plant Return to Service Safety Evaluation Overview ~

figure II-1 Number of Tubes with Defects, vs. SG Elevation Figure II-I Three Mile Island Steam Generator B Figure.II-3 Three Mile Island Steam Generator A Figure VII-1 Outline of Basic Tube Plugging /Stabili:ation Plan Figure VII-I Lane / Wedge of Tubes to be Stabilized P W e i

j

, + r

..y . * .,

c. , , o.

-s v Table of Contents (cont'd)

FIGURES - (cont'd) -

Figure VIII-l Reduction in RC Flow Rate vs. Number of Tubes Plugged per Steam Generator Figure VIII-2 Comparison of FSAR Flow Coastdown to Flow Coastdown with 1500 Tube Plugged Figure VIII-3 Effect of Tube Plugging on Natural Circulation - THot Loop 1 vs. Time Figure VIII-4 Effect of Tube Plugging on Natural Circulation - Total Core Flow vs. Time Figure 1X-1 ECT Calibration

• Figure IX-2 Critical Crack Sizes .

Figure IX-3 OTSG Loading Cycle for Tube Mechanical Evaluation Figure IX-4 da/dn vs I for Inconel 600 Figure IX-5 OTSG Leak Race as a Function of Crack Len;;th and Tube Tensile Load Figure IX-6 TMI-l OTSG Tube Leakrate During Plant Cooldown i Figure X-1 Tube Rupture Guidelines Figure A-1 TMI-l Researt Test Program Iacluding OTSG Repair Figure A-2 TMI-1 OTSG Tube Repair P.recritical Test Program TABLES Table IV-1 Three Mile Island Unic 1 Administrative Controls Prfmar/ Water Chemistry .

Table'X-1 Leakage Detection Me.thods Sn- ary Table Table II-1 Laboratory Induced Cracks E/C Correlation Table IX-2 Critical Crack Sizes and Leakage Table XI-1 Maximum Hypothetical Off-Site Doses for 1 lbm/hr. and 6 G7H Primary to Secondary Leak Table XI-2 Exposures from OTSG Program - -

Table XI-3 Radiation Fields at TMI-l Table A-1 Post-Repair ECT Inspection Table A-2 Post-Repair ECT Preservice Baseline Inspection Results Table A-3 Hot Testing Leakage Monitoring Program Table A-4 Steam Line Fittings Inspection i

,g e

- iii -

l

• o;p .

0.

s- .)

j I. INTRODUCTION A. - Purpose.

In November 1981 primary to secondary side leaks were discovered -1 t

in both TMI-1 Once Through Steam Generators '(OTSC). Subsequent detailed failure analysis showed that extensive circumferential cracking had occurred in the 0TSG tubes. This safety evaluation

• describes the results of the failure analysis, the evaluation of .

the methods of repair, and. the operational, safety and environ- -

mental impact of operating the repaired generators.

! -5. Background

.TMI-1 is a 776 MWe pressurized water reactor having two verti-cal, straight tube and shell once-through-steam generators (OTSGs). Each OTSG contains 15,531 Inconel-600 cubes, 0.625 in.

OD, .034 in. wall, 56 ft. 2-3/8 in. long, rolled and sealed-

' welded into 24 in thick carbon steel tube sheets at the top and bottom of the OTSGs. (See figure I.1)

. The plant was shut down_ early in 1979 for refueling and has remained in'the cold shutdown conditien since the TMI-2 accident-at the direction of the NRC. In anticipation of bringing the unit critical and returning to service, hot functional casts were performed in August-September 1981 and did.not indicate any problems with the OTSGs. However, in November 1981, during pressurization for additional tests, primary to secondary leaks were detected in the OTSGs.

As soon as GPU Nuclear Corporation realized the excent of damage to the TMI-l steam generators in.early December,1981, a dedi-cated CTSG cask organization was established to coordinate the

• repairs of the steen generators. The structure 'of this task organization is shown on Figure I-2. The scope of the task organization included determining the cause of damage to the steam generators, defining the status of the steam generators in terms of what type of damage and at what locations, evaluating-the numerous repair options and implementing the one chosen,

' evaluating the effect of the repair en both OTSG and plant per-formance, and establishing whether or not additional TMI-1 com-ponents had been damaged by the aggressive environment which was

~

apparently created in the once-through steam generators. An internal safety evaluation was performed which included these areas. Throughout the entire OTSG repair program, GPU Nuclear Corporation made every effort to obtain the advice and counsel-of experts throughout the utility, manufacturing, and research-communities. As can be seen by reviewing the task organization on Figure I-2, the organizations and companies involved in de-fining the status of the' steam generators and assisting in their -

repair cover a broad range of expertise. ,

?

9 me

o . .

r* ' -

l 1

FIGURE I.1 .

TMI-1 Steam Generator .

ELEVATION CROSS SECTION

~

~

PRIMARY SIDE  ::r;.

.' r r u ,

(INSIDE TUSES) m-M'*,.-N; M" *l 5.,. .". . -

..,.r.-..-...'e- --

UPPER .. . . ,, . . . s p . . %.-^

TUBESHEET (UTS) ..-. .

tw  : ::;< ;. bg:.',-@5p >.O.

AUXILIARY W5W*A f ~gg~~ .e $.? & ~;$f b ik y

}- M.7 Qa Fs:!DWATER O,mtw g q,g ,wr.-.-. v.. .a - gt,,

e < -'

NOZZLE 5 1

~

• ' WW * **

w wi 1 u#. Q % 4 - 'C i Erem:W::,2*T.Sz.W.T,*-'hW

• Pgfyg.r i .g -- n d *

~

- =ves .a.@:.,y+; --

'd .4pg .: ,'*'" O*s'-. .

..-- ... .r

- w w.pt nh :.,*:.>.7a.,s

4. t.

. ..... h=-==.y, usa - ~. ., ..; .--~t * ;. J.:  ::.:w Aqd. g;4.<.<..e-b O""--- I' Yktl5;b$k?UDLN.y STEAM OUTLET

%l, j /M e 1

. :#A:n'04*#A' r J d>' N ':'*r {' g '

I  !'* - , LANE W --

MAIN FEEDWATER 3 d.e iew W .

N0ZZLES (MFW) --- - C TUSE 3 4. . y e

6. ..

m. . .

I C y.g ,: D .

SECONDARY SIDE < 8.* -

7 -

(EXTERNAL TO TUBES)N%.ud -

nm -

l N. *b

.a o ,

im -

Weight operating. . . . ... . . . . . . . . 637 tons i

, j. H eig ht. . . . . . . . . . . . . .'. . . . . . . . . 7 3 f e et rT " Primary flow. . . . . . . . .'. . . . . . . . . 6 9X 106 #/hr.

SUP  ? us Steam flow . . . . . . . . . . . . . : . . . . . 6.1 X 106 #/hr.

. PLATES *

- :-. Number tubes . . . . . . . . . . . . . . . . . .' 15 5 31 1, s Tube size. material . . . . . . . . . . . . . 0.625" od. 034 wall LOWER ' 7 "T

~

iE ~ inconc! 600

• SECONDARY

• ~

• s Manufacture date . . . . . . . . . . . . . . 5/G9 to 11/70 HAN0 HOLD MifJhtfid f ~

l . .

59 O

! t j OTSG TASK ORG ANIZATION i

P INDEPENDENT TillRD - VICE PRES.TECitN' ICAL VICE PRES. TMI.1 PARTY HEVIEW FUNCilONS P

j 1 .

1 S&W PROJECT MGR. OTSG PROJECT MGR. .......'

4 C R NATO 3 a

C 3

fut

r n

..a. . .~.

.~. .... ....>

1. Alt ung 015G AfrAla 085G Ecr . 0150 g ,, ,, PaeMAny pgggy ggg,gg A N ALV 515 OrtsONS BfPAla E M AMaf A AT60N EI"I A E cal *:4an E .

gp g y, S T,t,M gg MOolfsCATIONS C4 E ANUP I

t saw see saw .sw eaw saw 87= caw sw im.aa pea saw Gas etw e 6108888

• sera smas ff SPm hl e45 00888 I O

• FiwellAIGadmW liee sesaw e

1 l .

j' f

o + .

In order to provide added assurance that the TMI-1 OTSG repair was conducted in a prudent, safe and technically correct manner, an independent third party review was established made up of .

experts from throughout the utility and research industries.

This independent third party reported directly co the Vice-President of Technical Functions and was tasked to provide an independent and objective safety evaluation of the failure anal-ysis program, eddy current examination program, OTSG performance evaluation, OTSG repair criteria, and the overall OTSG repair program. The advice and recommendations provided by this third party review have proven very beneficial. Their participation provides added assurance that the OTSG repair activities both confo.m co the NRC rules and regulations governing the operation of TMI-1 and assurance that'the adequacy of the steam generator repair program allows safe operation of the TMI-1-nuclear unit.

C. Steam Generator Renair Program The soproach taken to restore the Steam Generators to service as to evaluate the condition of each tube with eddy current techn. ques developed specifically for the geometry of this-corrosion mechanism. Fotlowing ECT the status of each tube was evaluated and one of the available repair methods was chosen.

Figure I-3 summarizes the disposition of all the tubes in the TMI-1 Steam Generators after repairs have been com'pleted. Th'is figure indicates the four methods of disposition,' the basis for

.. selecting those methods and some other concerns that were con-sidered and resolved in selecting those methods.

The first category includes the tubes removed from service prior to the repair. These are tubes that have been previously ,

plugged due to indications of defects from ECT inspections from previous operating cycles. Also included in this casagory are those tubes which had sections removed from the steam generator for metallurgical examination and those cubes which indicated leakage during the initial tests after damage was discovered.

l The second category is the primary repair method for the steam generators. This repair method for the TMI-1 OTSGs involves l expanding and resealing the existing tube walls within the upper  !

tubesheet at points below where the cracking of the tubes oc-curred. - The expansion closes the gap between the cubes and the '

tubesheet. The expansion is done kinetically using explosives ,

- (deconating cord) encased in a polyethylene insert (see Figure '

I-4). The insert transmits the explosive energy to the tube wall causing an interference pressure between the tube and the l tubesheet. l l

The tube expansion repair method is feasible because of the

specific nature and location of the cracking in the TMI-l Steam Generator tubes. The majority of the cracking is located in the a

-w. , . -. , ,, .m.. ~ . - - ~ ._.-,# , . , , - , . . , - - . - - - - . - , , - - - . - - - . .

DISPOSITION OF TURES IN TMI-1 STEAM GENERATORS o!.

m ,' -

l 31,062 ,,_;

TUBES TOTAL - .

31 -

gj

,' ' y3 ji ?

V i 'V V 3 - u  ?.I .

~'

347 [., *,

28775 857 , u,, # / 83 l

. essend is.w sweise whb o. be.we del == 6.iew whb des.su b.l.w us er coAs indiasises < se*, < 4e%

psier se espais US + 8 sepaised by espassies sepoised by piogging IW below US + 5 apaised by apanies t

. Dasis iof '**'*"P'"8*"' '"'**

• 3** I*3*' '* *""*d *88'*8 ****d' **b' willset paspegate by lespectises os desige basis qualified Dispositiott .,, ,,,,gg , ,,,,gg,,,g,,g . ,,,g gg g g..g. g,,g,g

, eneminatise se ECT did see desoci ssesbs -safety sosivsla lmeellies speseties below US S S opesations whb op se 1588 leshad deslag lakielleab plegged subes -ECT salibeatlee psegsaw

$ tests domessesales desesenitier g

. 48% IW e R- '

-ECT salsbeaties peessam elluss se EfW New and -ssesbs Resolteilen

• g,,,,,,,ased desessibisie, sanwal sisalesies willeet taildesieg esste sa of allief of < 48% 1W seashs evaleased as adequase safety -bighly sedsslag seedleises

-ssacks belaw 44% IW will at essapessesse paeseet concerns ses psepagate by sbeseiselpsepagatise t enesbasisalleads

-aey absough wall seashs below sepais which wpse seissad desieg lespessies psossam wellleak desieg test psessam

-Isog and shast teses sessesies esses demeesssated ebat lesel l IGA is met a seesess and that sessh s will met paapagate by sessesloe soechanises

-psessueless eates which well pseveet new seask .

lesmaties 5 -

O

• d o ,

. n FIGURE 1 4 KINETIC EXPANSION PRO. CESS .

i

- ORDNANCE CORD l

T

' .e-

: D ETON ATING

\-

CAP HOLDER I

j i l.) / / / n I , ,

, \ k .m.m --

! + __v 3

g. y j%g W

\. . 1- W r :2WS[l$  ; *

' E ,.

,,.. ~

L, Q D ETON AT'ING ELECTRICAL

.}f s  ! j ;l CAP sic 3At i

h -

\'f ji jr t .!

t h,; . U .

030 NANCE CCRC

/

, , y =- -

?I '*

fA; D 115 IN

, s00 STER y r X E H

l,.3j V NJe OETOMATi.'ic CORO a 0 1

t g ) -

PCLYt:l H YLENE INS 5H T _

e Tb <e i

g . . . . ...

J B t I 't g paprv.rm(

.., -+-,-*-+-.+~-.~-,.m - -.4 -q.- .- - . . . - - - .m-t

a ~

. - i .

FIGURE I.5 TMI-1 Steam Generator Typical Cracks INCONEL TUBE WELD CLA0 OVERLAY  ? '

p'"% TYPICAL CRACKS ROLL TRANSITION '$ .

A

^

/ STEEL TUBESHEET

/l

/

CRACK CHARACTEiilSTICS: CIRCUMFERENTIAL BELOW FILET WELD -

NOT FULL ARC GENERALLY VERY TIGHT PRIMARY S10E 1NITIATED

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

_, . . . ~ .

o *.

• 2;

~

r' upper ends of the tubes of the two generators, at or near the upper 1 in. to 1.5 in. where the 56 ft. long tubes were mechanically rolled and then seal velded to the tube sheet -

cladding. (see figure I-5). The combination of rolled joint and  ;

seal weld held the tubes tightly in place within the cubesheets. i l

At TMI-l both 17, in. and 22 in. long expansions will be utilized  !

depending on the axial location (within.the upper tubesheet) of i che lowest defect. The expansion length is chosen to provide -

• ' the minimum length necessary between the lowest defect and the bottom of the expansion to serve as the new pressure boundary. .

This expansion length corresponds to eight inches above the lower face of the upper tubesheet (US+8). This length provides

, the dividing line between those tubes with defects which could be repaired by expansion and those that would be removed from service. For the TMI-l OTSG geometry and materials, a 6 in.

long joint below the lowest defect has been shown to provide 4

adequate leak tightness and load carrying capability and is the basis for the joint qualification program. All tubes that remain in service will be kinetically expanded irrespective of whether or not a defect has been detected. (see Figure I-6).

. The third category includes those tubes which cannot be repaired by expansion due to unacceptable defects in the region below eight inches above the lower face of the upper tube sheet. .

These tubes will be removed from service by plugging.

The final category sea chose tubes with ECT indications that are less than 401 through wall. Since analysis indicates that these tubes will not fail by mechanical, thermal and accident loads, they are being -left in service to provide characterization of these indications after they have been exposed to operation.

Leaving this category in service provides information in future .

ECT inspections of the stability of these indications.

D. Safetv Evaluation Logic

' To determine if the plant could be safely returned to service, a program was initiated. co define all the significant effects of operation of the steam generators af ter exposure to the damage mechanism and after the steem generators were repaired. The main product of this program was a logic diagram which defined the major areas that needed to be add'ressed and also defined the detailed cases, inspections and analyses which were performed to support each of these areas. A condensed version of this logic diagram is presented in Figure I-7. This diagram lists the major areas that were considered and references the sections of this report which descri.a the results which support the conclusion that the TMI-l Steam Generators can be operated safely. The ,

rssults of these programs demonstrace the following:

(1) The failure mechanism is understood well enough to define the root cause of the steam generator damage; 3- -

v ,e- n , , -- . ,,.- . .,- .- ,- , - - , , ,,,,.,----,--,r,-,-,,ae-,- .e ,- ,,.------,m,.,-,--,,-.a , , . - . - ,.,,,---n. - v. - - -, - - - - - , - - , ,, n w--

i.

t FIGURE I.6 k

Kinetic Expansion Length -

PRIMARY FACE UTS ,

9//////MS/#4 Expanded only ZO.NE WITH LENGTH OF DEFECTS EXPANSION A< '

A 17 s B 22

~ ~

Expande'd an'd Removed from Service

% .1.,,_

. = ZONE WITH . LENGTH OF s %3 DEFECTS EXPANSION B4 g= ,

p 22 C

16 by,-- = --

0 17

= E 17 C<

lh .

2q _ _ _ _ _ _ =C -

D / 2_2,"_ _ _ _ _ _

W/fMf///A F?&

SECONDARY FACE UTS t Ef :

\ .

• 1STH TSP 6

o

-- --.,s - - - . , ~ - - - - . . - - ,- u , -. . - ,,. -- - - - - . . - -

c j)l1 O.

, z{ _

)

8 I

X

(

E

• S S .

N RNT O EOC l I I VRA XT DI VP l A

0t A

OE NM I NU N EC PL PA AC e -

M )

S A A.

- E S

N N R X E

R l

C O G (

I U

- i O O W OV E

N A

I T

W I T R S D E M S

T E

l f

M! CE Hl A 1

l P T S C A S I

S RE i

l(

V Rl S l

8 B. XA G C. ll E O R

R Gl E

T V OL A Nl OF( V ED Al i

P Xi O( I W

T P OAS N R M I8 0 S(Oi l Ti ON GN()XPlR X ISW E AAA 0I V E R 0 REET CE RSO I ERR E LTA PI R S E 0 S NS P Z ABE0KY GS T 0G N 0N T0 I

V T EE O I

O S ED NP R AG NL TY ER PADTEAA 0CLWlA. AE A XOR l

I E K1 G APU R RIS TPA 0 N PXL APAANP EV SPE 0 O

I EEP R - - tN UGR EACAE(

E CGRN l I C PE

- - - DR E I(A( - -

C T * * * *

• A U

L l N E. O A G O l l

I T

V N I

E G

(

N C E fE T A O E

fI R P

Y AA SR O TYM I T G S T T E P E PF D A

C E

NI N E N P UAN P TS I

- F A G SSU S RM A L 0 P1 & . N O E I

PTl S S S Y S P S E.

E C S C UYl R R S S (l .

C I *

• V R ) )

E E C. T A. Y S N V A V R I

E I

( F ( T O MT E C RR U L S )

T AN TE N 0U S R 0 E

Ul N IM9

/F 0 E V N SER VE 0L l TT UA li( I R E RR NU l i

TN C S U R PU AS T T I

EL T

U S0E C E E E l V V N MVO E A1R l O Ol EVT ART OR H M E

0 M M A.

1 EN PN -

T XE OR R (I EV RO PC MO C I N * * *

• A L S P T S

E S T T M D O L S E

ES O I

A T '

R NT C u A S I

N l R R O l l H E U I l AC D L Si l F E L O N A R A MU T

E )8 R C.

1 Ol [

M (1 C (l ,

l ll!i

a , e. .

. t I

(2) 0ther components in.the RCS and supporting safety systems were not visibly damaged by the failure mechanism; (3)' The plant can be operated such that this failure mechanism .

is arrested and will not recur; (4) The Steam Generators can be repaired' and operated within '

the design basis; (5) The plant can be operated with some tube leakage without adversely impacting the environment.

The remainder of this section provides a brief synopsis of the l

entire report with emphasis on the logic used to determine that l the plant can be safely operated with repaired steam generators.

Reoort Summarv A detailed failure analysis was performed including (1) review of the OTSGs fabrication history, (2) coordination of metal-lurgical examinations of tubes pulled from the OTSGs, (3) review of the OTSG operating plant chenistry histories, (4) coordina-tion of OTSG tube stress analyses, and (5) development of a failure scenario. This failure scenario, which provides a reasonable match between plant conditions and the mechanism which caused the tube cracks, concludes that sulfur contamina.

tion in the presence of sensitized cubing material at the oxygenated, cold conditions existing after hot functional tests led to the observed intergranular stress assisted corrosion.

Section II summarizes the failure analysis. ,

An inspection of additional RCS ccmponents which included non-

' destructive testing was performed to determine if other ccm-ponents sustained similar damage to that found in the OTSG.

Emphasis was placed on materials which were susceptible to attack in components which fulfilled critical functions. 'No damage was found. An inspection of RCS supporting systems has also been conducted. Details can be found in Section II.E.

As shown in Section IV, paths for chemical injection into the RCS and administrative controls on chemicals were examined in an effort to prevent future chemical contaminatics of the RCS.

Additional periodic chemical analyses will be performed during plant operation and some administrative limits for chemical concentrations have been changed. A sulfur conversion and removal process to clean the surfaces of the Ranctor Coolant '

System has also been conducted.

i e m l

a e To determine that the OTSG is operable in accordance with the original design basis, the CTSG was analyzed in two sections:

l the repaired portion and the unrepaired portion. In the re-I paired region, both the expansion repair and tube plugging were '

considered. For the expansion repair the important character-istics were the load carrying capability and leak tightness of the new joint. A 6 in. expansion was qualified as the design basis load carrying joint using mechanical and corrosion tests.

l Details of this program are summarized in Section V. In addi-tion to the qualification program, a process monitoring program l was set up to oversee the expansion process.

Plugging repair is summarized in Section VII. B&W Welded Plugs, B&W explosive plugs and Westinghouse rolled plugs were quali-fled. Analysis verified that adjacent expansions would have no detrimental effect on existing plugs, and analyses documented in Section VIII show that the system will not be adversely affected by either the number or distribution of plugged tubes for normal, accident and transient performance.

In the unrepaired region of the OTSG, various tests and analyses discussed in Section IX have shown that:

(1) Corrosion tests indicate that the tracking mechanism has been arrested and does not reactivate in low sulfur water chemistry. If rapid cracking should reactivate due to a'n unknown mechanism at operating temperatures 'or during heacup and cooldown cycles, it is anticipated that the precritical testing sequence would allow sufficient time for defects to propagate through wall to a si=e that would allow leakage to be detected. Therefore the precritical leakage monitoring during the hot testing will detect crack propagation.

~

(2) Analysis has demonstrated that cracks below a minimum range of length and through wall thickness will not propagate mechanically. Analyses included calculating a minimum size below which a crack will not become unstable due to plastic tearing or ligament necking during a main steam line break (MSL3). This range of crack sizes is detectable by the ECT inspection sys. tem that was used to inspect the steam generators.

(3) Any defects in the detectable range that are undetected du' ring the 100% ECT inspection because of equipment or an-alyst error will be exercised during the test program. If they are 100% through wall and of a size to propagate to failure under loading, they will be detected by leakage monitoring programs.

O w*

o < -

~

Both an ECT flaw growth program which monitored a sample of tubes for new defect indications and corrosion testing on actual defective TMI tubes in the present primary coolant chemistry. .

showed that the damage mechanism had been arrested.

To determine if all unacceptable defects were detected by ECT and those defects not detected would not propagate to failure, an extensive ECT calibration program was devised and the small-est size defect which could be consistently detected by ECT was determined. Comparison of field ECT results to metallurgical examination of tube samples removed from below roll transition in the TMI Steam Generators showed a one to one correlation between actual and ECT predicted defects. Stress analysis showed that cracks of the size that could propagate to failure by combinations of mechanical loads were within the ECT detect-ability limits. Local IGA one to two grains deep was examined during the metallurgical examinatier. program and there was no indication that this effect was related to the failure mechanism.

A precritical testing program has been designed that will pro-vide confirmation of the adequacy of the OT3G repair and OTSG operability. The program tests for leakage in the repaired region using secondary to primary drip and nitrogen bubble

~ tests, and a primary to secondary operational leak test. In the unrepaired region, axial stresses will be placed on the OTSG~

tubes from normal and accelerated cooldown transients. The accelerated cooldown will be at a race larger th'an the normal cooldown race based on past operating experience but will be within the cooldown race limitations of the existing Technical Specifications. A period of hot operation is included which will allow time for defects on the threshold of propagation to

~ propagate or leak. Leakage calculations indicate. chat leakage from tubes with mechanically unacceocable through wall cracks will be detectable during the test period.

Operation with a primary to secondary leak at the repair design goal of 1 lb/hr. and at a more conservative race of 6 gal /hr.

has been evaluated. These leakage rates have been found to pose no threat to the health and safety of the public and allow the plant to operate within existing Appendix I Technical Specifications. Details can be found in Section XI.

This report concludes that TMI Unit I can operate with the re-paired CTSGs without undue risk to the health and safety of the public.

4

. . =

_m m. m . o

8

. .c II .. FAILURE ANALYSIS .

Three' Mile Island Unit I was in cold shutdown from March 1979 .

until September 1981. In September 1981 hot functional testing was performed. The plant was returned to cold shutdown for some I

final modifications prior to startup. The plant was pressurized to about 40 psig in November 1981 and small. leaks from primary to secondary side were detected in the tubes of the once through

. steam generators (OISG's). -

A detailed failure analysis was performed to determine the root cause of the steam generator damage. This analysis included a

-review of the steam generator operational history, a metal-lurgical and corrosion test program, a review of OTSG stresses and fabrication history, and the development of a failure scenario. In addition, the degree of damage both in the OTSG's I and the remainder of the RCS was investigated.

A. Onerational Historv The time of the OTSG tube failures may be bracketed based on operational considerations. During HFT on September 4, 1981 the

• leak rate of the RCS at full pressure was measured and found t,o be within specification at .5 gpm. On November 21,1981 with the RCS at about 40 psi, leaksge through the OTSG Pubes was , '

observed.

A review of operational history of the TMI-l steam generators was performed for the period April,1979, through November,1981 to determine whether instances of chemical contamination er excessive tube stress could be identified as the cause of the l tube failures. A detailod description of OTSG operating history is found in Reference 2 and Reference 22.

The operational history of the TMI-1 GTSG's reveals that the tubes were not subjected to excessive stress, and generally, the reactor coolant system chemistry remained within specifications for the period extending from April 1979 through November 1981.

Certa'in operational events did, however, have a significant i

^

impact on the chemical environment of the OTSG tubes. There were five identifiable instances of probable intrusion of chemical contaminants into the Reactor Coolant System (RCS). In March 1979 oil was introduced into the Reactor Coolant Bleed tanks probably by overflowing the miscellaneous Waste Storage Tank through the vent header. Some oil may subsequently have found its way into the RCS. Tube surface. analysis has shown that carbon was present on the surface in the form of carbonate and hydrocarbons.

7-

-l l

1 v --,y -p e -%W-- yw -w.e---- ..i.e-T - - - -

4

-4

-P 1

In October 1979 sulfuric acid was injected into the Reactor Coolant Makeup System. Although attempts were made to prevent the acid from reaching the RCS, chemistry results indicate some .

contami- nation of the RCS occurred (see Reference 22). In July

~

1980, May 1981 and September 1981, a surveillance test was performed which may have allowed sodium thiosulfate from the Reactor Building 9 pray System to find its way into the RCS.

Sodium thiosulfact at levels of 4-5 ppe as thiosulfate is considered to be the most likely contaminant. The ionic species -

from the first contamination incident in July 1980 were removed from the bulk liquid by domineralization in August 1980. The

1 ionic species from the second contamination incident in May 1981 appear to have been only partly removed by processing through a resin water. precoat filter in August 1981. A 1-2 ppa chiosul-face residual could have still been present at the start of September 1981. Additional sodium thiosulfate in the RCS may have resulted from injections of Borated Water Storage Tank

-(BWST) concents during cooldown from hot functionsi-cesting.

This water had been previously mixed with water from the Reactor Building Spray piping. The quantity of thiosulfate was not I sufficient to be detectable by conductivity.

Significant to the localization of the attack was the history of

.the water level on the primary side of the OTSG. Following the hat functional testing in September 1981, water level was promptly lowered on September 8,1981 then slowly raised over the rest of the month. This allowed a drying th'en rewetting of ,

the tubes in the upper portion of the steam generator, causing chemical concentration in that region.

0xygen introduction is also believed to have played a role in the damage mechanism. There were three occasions when oxygen was introduced into the system. The water ~from'the BWST injected during high pressure injection and icwe pressure testing was prob. ably saturated with oxygen to approximately 8 ppe. When the water level was lowered, the OTSG primary side was vented to the vaste gas system. The maximum oxygen specification in that system is 2*. Thus, oxygen was available

? at the liquid surface while the liquid level was being lowered.

The RCS was vented to atmosphere through a CRDM vent on October  !

7,1981 and remained open until filling in November when the leaks were discovered.

i

~

B. Metallurgical Test Program After identification of the leaking OTSG tubes by nitrogen bubble testing, it was decided that in order to determine the .

cause of failure, cube samples would need to be removed from the I steam generators for analysis. The initial selection of tube samples was made after eddy-current testing had been commenced ~

and the choices were made based on maximizing the number of defect indications in each tube and providing an adequate sample of addy-current signals for eddy-current qualification.

e~

9 s

4 Four tubes were initially selected from the "3" generator. One tube was a known lenker from the bubble test results, .the other three tubes contained eddy-current indications of greater than 80% through wall penetration. ~~

After the initial samples had been removed. .it was confirmed that eddy-current signal anomalies were showing up at the roll transition. region. In order to determine the disposition of these tubes, additional tube samples were selected for removal

~

which contained these' eddy-current signals. This time, fifteen (15) tubes were removed from the "A" generator.

A third set of tube samples were removed which '.ncluded 6 tubes from the "B" generator and 4 tubes from the "A" generator.

These samples were taken to obtain some low level defects frca deep in the steam generator, to sample tubes from specific areas, and obtain tube ends to be characterized (in previous samples the tube ends had been removed during palling).

1. Analysis Program .

A multi-task program was conducted to provide information related to the steam generator tube damage problem. This program contained the following analyses / examinations:

a. Visual Examination -

b. Eddy-Current Examination -

c. Radiography ,

d. Sectioning and Bending

e. Scanning Electron Microscopy (SIM) and Energy Dispersive

~

X-Ray Analysis (EDAX)

f. Auger Electron Spectroscopy (AES)

g. Electron' Spectroscopy for Chemical Analysis (ESCA)"

4 h. Sodium Azide Spot Test-

i. Metallography-Microstructural Analysis

j. Scanning Transmission Electron Microscopy (STEM),

Electrokinetic Potenciostatic Reactivatica (EPR) and Huey Testing -

k. Residual Stress and Plastic Strain

1. Tension Testing.

m. Hardness Testing.

n. . Dimensional Measurements.

2. Test Program Results/ Conclusions The detailed test results are presented in Reference 2. The following summarizes those results and sets forth some con-clusions.

4. The tubing has failed due to intergranular stress as- -

sisted cracking. The intergranular morphology has been

~

9-

e. :o '

confirmed by Metallography and Electron Microscopy.

This has led in many. cases to through wall penetrations and circumferentially oriented cracks. In all cases, cracks have initiated on the primary side surface.- '

b. Microstructural evaluation of the tubing from numerous locations, has indicated that the structure is repre-sentative of that normally expected for steam generator tubing. Tests have concluded that the material is in a _

sensitized condition and hence is expected to be suscep-cible to intergranular attack in oxidizing acids.

c. Trancaission Electron Microscopy has also confirmed that no secondary mode of failure is associated with the intergranular corrosion, that is, no evidence of any low or high cycle' fatigue was observed on the'se fracture surfaces.

d. The consistent circumferential orientation of the cracks below the weld heat affected zone, indicates that an axia'. stress is part of the cracking mechanism.

Residual stresses in the roll alone were not responsible for the cracking. Therefore, the fact that the cracks occurred when the tube was under a. higher applied axial cension stress rather than hoop stress, confirms that the cracks formed during cooldown or cold shutdown. -

e. Axial cracks ,have been observed at the top end of the tubes near the seal weld. Some of these cracks pene-trate 100% chrough the wall but they do not penetrate the weld metal. The axial orientation in this case is expected based on the residual stress distribution in the area of the seal weld. ~

f. Auger analysis of. surface films on fracture surfaces and on the I.D. . surface of the tubing indicates that sulfur i is present up to levels of eight atomic percent. The sulfur concentrations along the I.D. surface of the tubing down to the 9th' cube support place, are generally ,

uniform with perhaps a slightly decreasing level lower l in the tube sample. The form of sulfur is believed to be either in the form of nickel sulfide (Ni 323), or some other reduced form of sulfur. The reduced sulfur form generated from the contaminating species is directly responsible for the cracking mechanism.

Auger analysis also showed that carbon was present at '

levels from 50-90 atomic. percent on firs ~ and second 1-l l

i

-. - . , ~ . - . .

r .

~ . - - _

t

,O: .

1

. i round tubes, but a maximum of 50% on third round tubes. I It is thus inferred that the extensive carbon contamina- 'l

. tion on the first and second round tubes was the result .!

of contamination either during or immediately af ter tube removal.

In addition to sulfur and carbon, the Auger and ESCA '

analysis have shown the presence of nickel, chromium, oxygen and _nonnal trace quantities of fission products -

on the fracture surface.

g. In conjunction with the cracking, there has also been intergranular corrosion observed. . These " islands" of IGA are not always associated with cracking and in general are associated with I.D. deposits. IGA found at crack locations tend to penetrate deeper than the ap-proximately 1.5 to 3 mils of penetration typical of the IGA " islands." Most severe cracking in general rel ces to more severe intergranular corrosion.

h. In 39 out of 42 cases to date, cracks which have been examined either by metallography or by bend testing have shown the defects to be 100% chrough wall. The re-

' maining three cases exhibited penetrations of 66, 70 and 70%.

C. Corrosion Test Program A corrosion test program was put into place and addressed the areas of crack arrest, corrosive species and verification of the corrosion scenario. The corrosion testing program is addressed in detail in Section III of this report. .

The following conclusions can be drawn free corrosion casts which relate to the failure scenario.

a. Thiosulfate can produce cracking similar to that observed in the steam generator cubing.

b. In the absence of thiosulfate no cracking has been produced in the laboratory in primary water chemistry.

c. Tubing removed from the steam generators appears to have a lower thiosulfate concentration threshold for cracking than an equivalent archive tube which has been sensitized.

d. Tubing thermal history is a key parameter in establishing material susceptibility. A threshold level of sensitization must exist. Data suggests higher mill annealing tempera- ,

tures favor cracking in sulfur contaminated primary water.

o'

s: ,

o

e. Crack' initiation and growth rate are temperature dependent.

For susceptible material, crack initiating time will be decreased and crack growth increased by raising camperature .

up to 170*F.

f. An oxidizing potential is required for cracking to occur.

In the absence of oxygen, cracking has not been observed in the laboratory.

g. Crack growth rates appear to be very rapid and can be as high as 1 mm/ day. Lab specimens have exhibited partial through wall penetration in areas of lower stress.

.D. Damate Scenario The conditions needed' for Intergranular Stress Assisted Cracking were evaluated and compared to the conditions in the TMI OTSG's. Based on stress snalysis, fabrication history, the timing of the cracking, metallurgical and corrosion testing and observed features of the cracking phenomena, a failure scenario was proposed.

1. Intertranular Stress Assisted Cracking (IGSAC)

The occurrence of stress assisted cracking requires that, three conditions be satisfied simultaneouslyy o a sufficientLy high tensile stress o a susceptible material microstructure o an aggressive environment The information presented in Reference 2 relating to those three factors is sn=marized below.

a. Tensile Stress Since the cracks are oriented circumferencially in the tubes below the weld heat affected :ene, the sum of the operating and residual stresses in the axial direction was greater than that in the hoop _ direction. Axial tensile strusses are of principal interest. Very lictie tensile scpess is required to crack Inconel that is this susceptible in the presence of reduced forms of sulfur.

~

However, the higher the tensile stress the more rapid the crack propagation and the more cracks that actually occur.

b d

. . - . ~

g3 s a The stress analysis results suggests that the cracking must have occurred during cooldown or during cold shut-

~down because the axial tensile stresses are largest .

during this time. 'The analysis also indicates that the seal weld heat affected zone and the roll transition regions would be.particularly prone co' cracking due to locally high axial tensile stresses which are possible in that region. More cracking occurred in the periphery than in the center of the tube bundle because the axial -

stresses at and below the roll are generally larger at the periphery than in the center of the cube bundle.

b. Suscentible Material Microstructure There is no indication that tube material, fabrication or installation in the OTSG's was in any way extra-ordinary. The heat treatment of the whole OTSG fol-loving assembly puts the tubing into service in the mill annealed plus stress relieved condition which is ex-pected to be hesvily sensitized (i.e., low grain boundary chromium concent less than 10%) thus making it .

more vulnerable to IGSAC . Metallurgical examination has confirmed that the expected microstructure is present.

A large number of heats of Inconal 600 are present in the OTSC's which differ in composition and which may have responded differently to the stress relieving heat treatment. The degree of susceptibility as a function of the tubing heat number could not be established.

c. Aerressive Environment .

As previously stated in Section II.A, the results indi-cate that sulfur was present in the primary system water and three possible sources of sulfur have been identi-fled from the OTSG chemistry history.

If SO4 and S 02 3 were introduced to the primary water as the OTSG operating and chemistry histories suggest, they would be expected to persist as long as the water was at room temperature even if the oxygen content of the water was reduced by hydrazine additions.

However, hydrogenacing and heating the water to perform a hot functional test would be expected to result in the generation of S~~, possibly accompanied by S and other intermediate species. Subsequent cooling to room temperature and oxygenacing following the hot functional O

ma

, & +

l --f casts.cavidly oxidize S~~ to S and could also result in the appearance of significant concentrations of other species of higher oxidation states. 'Although it is not .

possible to predict either the identities or the concen-trations of the sulfur species present following the hot functional test, Lit is clear that this transient is likely to have greatly affected the a'ggressiveness of the environment with regard to low temperature sulfur induced attack of the-0TSG tubing.

2. Procosed tailure Scenario This following scenario is consistent with all. the observed features of the cracking phenomenon, the timing of the

- cracking and the results of the metallurgical examinations and corrosion tests. ,

a. During layup the primary system was contaminated with sulfur by the accidental introduction of sulfuric acid, sodium thiosulfate, and possibly a sulfur-containing-oil. , The amount of sulfur present may have reached .,

several ppm, but the contaminated water was not aggres-sive enough to crack mill annealed plus stress relieved Alloy 600. The corrosion tests confirm that cracking would not have been expected to occur at this stage.,

b.- The temperature and oxidation potential' transient as-sociated with the hot functional test resulted in a change in the types and concentrations of sulfur species present in the primary water. - Further changes occurred when thiosulfate-contaminated oxygenated water was in-jected during the tests of the HPI and LPI systems,

c. When the water level in the OTSG's was lowered following the hot functional test, high concentraciens of aggres-sive metastable sulfur species developed in the dry-out region at the top of the generators due to the combined effects of solution concentration by evaporation and the comparatively high availability of oxygen. Changes in the sulfur species in the more dilute bulk solution proceeded more slowly resulting in lower concentrations of aggressive sulfur species. .

d. Sulfur-induced IGSAC of the Alloy 600 cubing occurred rapidly in the dry-out zone with preferential accack at high stress locations in the most highly sensitized tubes. Cracking occurred to a lesser extent lower in the generator. Statistically this would be expected e-e

, n,. , -- - - - , , , , - - , = - - - , - , , - . - - .

a p . . ~P because the bulk solution was less aggressive than the environment seen by tubes in the dry out zone. Cracks would occur in areas low in the generator which were slightly more susceptible to IGSAC due to surface film -

anomalies or residual stress ancmalies.

e. Cracking terminated either becau ce continued chemistry

, changes resulted in the formation of less aggressive sulfur species or because the environment in the dry-out region was diluted by the slowly-rising bulk solution.

By the time the water level was dropped again, the chemical state of the sulfur in the primary water was sufficiently different from its state Lamediately after the hot functional tests to prevent a recurrence of steps C and D in the new dry-out zone.

f. Cracking was discovered when the OTSG's were pressuri=ed.

E. Distribution of Damage To evaluate the extent of the damage, an eddy current testing (ECT) program was devised to examine the OTSG's. In addition, an inspection of other components in the reactor coolant system (RCS) and supporting systems was cor. ducted to determine if damage similar ,to ch'at found in the OTSG's was evident.

1. OTSG Eddy-current Examinations .

^

Special eddy current techniques were developed and an ex-tensive testing program was established to provide an ac-curate description of actual OTSC cube cracking (Reference 20). In-situ eddy-current results exhibit tube wall defect indications at varied densities distributed both axially and radially in both OTSG 'A' and 'B' tube bundles. The sajority of the defect indications were in the upper tube-sheet (UTS) region and particularly confined in the tube roll transition zone. After an absolute probe inspection of the roll transition and mechanically expanded area of approximately 18,000 cubes, ECT indications were being reported with such frequency that it was decided to affect a kinetic expansion for all tubes in both tube bundles.

Further ECT data was not interpreted above elevation US+14 inches due to the decision to repair the top 17 inches of all the tubes. Figure II-l gives the number of tubes with defects by elevation in each generator. Radial distribution of tubes (as shown in Figure II-2 and II-3) with defect indications requiring plugging in both 'A' and 'B' OTSG shows a hi-her percentage in the periphery with the defect a

"'" mag, w "

FIGUM II.1 d

l V Em' V,\ i j '

ec

'>>=

a<<

9 s} I' yam W w ww E 'e \

azz w E- e -

w=w = #x

+05 E e~e Q a

s s #, a o2 E # =m ,s wo

• LL =h k e\rz =

W< -

? *\* CE O>

y W U J { e\*\ >2 w Wy #s#/

hy -

an~ //A oe J

Jy, bN Y *, e

!s -

With@MW!i:Uh!$cG%vMtrhMicWQRQQf ,\'

l MMS e,\,,

y s=wswggg , s,

,1_----------. xx E~NI@lNf(%%%\\Y$$$$QQQ(Q%%i ,

,T =

A%\

\ \

s i

'<.\@

as s e- s- a e

s o

s e o d #

e x

o e o m a m " #

m o m o w w *Q M N N M -

S3801 dO U38 WON .

aa 4

4 t

9

--- -. . - - - . .wy_ . _ _ , , . - . . _ , . . y-%. ..-.-y..-_ _ . . ,-.y._, , - -, , _ _ . , . .,, . , , , , _ - - - - . , , , , , . , , . , - - - - - , - - -,-,--,__,c , , , , , ,

o n - . -p 4

~

I. s 4

THREE MILE ISUND NUCLEAR GENERATING STATION .

tMIT I P m

STDM ECOATGt 5 ...............

.. . ~............ .~.

a.. .... ..... ....

. .. w ..._ ........................~ ................................... .... .................... ...... ,

.s...

. ........ ....,s.......... .... ........... ........ .... ... ...... . . .............

........ .. .. ~ .. ........ .. . ............... .. ......................................................

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

. .s. . . ..

.. . . .s

..s........

. . . . . . .. .... . . . . . . . . . . .GPU . . . . Nuclear Corporat.ion l*e 18-NOV-62

• Pt.UGGED Tt>8E .

-. x . - - a

....wa , , ,

1 .

4 THtEE HILE 151.#C NUCLEAR GDERATIhG STATION IFIT I 1 p.

STDit (DOAftR A .... . . . . ........ . . .

. . . ..e ... .. .

. ...M.g.-

...........M....... ..g.

...a . .......

..# . . . . .e. . . . . . . . . . . . ...$e....

. - . - . . . . . . . ..** . . . . . .. .. ... . .. . ... ..............e

... . .... . *. *... ..............s.

......e .g f .. ... . . . .

. . . . . .e

.e. ... . ................. . . .. . . . ...........e.... . .. ...... .. ... .... ...... .... . . . .

,,...e.... . . . . . .

.....t".*.z...U.**.'.*.*.*.*.*.'.,.?./*.,*'?*.*..*..*.***.*.**.*?*?'.*?.?*.*.*.*..*.*.*.......*.... ... . . . .. . . ....... .. .. ................. ................. ***t**"..... ..

. . . . . . . * **** *.*.*.**/

.?.%**.*# *.* ." . . */*??

. /. .. .* .*..T...*.*.*.*?.?

• ***^

.. .' . . 71 '" . V.* . *.* f.?.f *.*.*.*T

• ? *I.? ***'.*. *.*.f.?

• .*** . .... . .*.*.**.*??

• t .... .. ... ..................

. . . . . .. . . . . . ...... ......... ... ... . . . .. ....s .... .. . .... . . , .

. e.

.e

.. .e. ..

e .

....s

..4.

e '. ..

.e .

. .e. ..

..u

4. .

. .. e ..... . . . . . .. ... .... .. . .. ..........

. . . ........e..... . e.

...~. .. ... . . .. . . ..

. . ... ...e.

4 .. .. .... e.. .... . . . . . .

. . .... e.

. .. .e... ....

. . ... . . . . . .. . .e ..

. . . . . .... . .. .. ..A

.s

. . . .. . . . ....... . . . . , . .......2...

.. , .. . .. .... ... e.... . . .. . . . . ....... .. .. .. .. .. ... . . .

.. . . , ....v.

. . . . ... . . .. . ........... . . .. . ... ., .. ... ...*. .... . .. .....e. ..... . *. .....

e...

.. .. .. ... . . . . .....e s.. ..

. . ...t..

...........,.. . . . ~... . . ..... .... ..

. . . ~. ..... . .. .

.. ..r..................

. . . . . . . . .. . .. ..... .. . . . ... . .....e....

. .. . .. ... .. . .. . .....a....... . . .... . .. .. . . .... .. . . . .

..... . . . .. . . .. . . .e........... ..........

. .. . . . .... .. . .. ..... . . . . . . . . . . GPU

. . . . . . Nuclear . Corporation l .

18-NOV-82

+ Pt.UGGED TU8E e*

' ~ - - '- -------y,-i.,,,-. .. .,, , _ _ _

^ ' ' ' - - ' , . , , , . , . . _ _ _ - - - - - - - - - . , _ . . , ,,, _ '"F

_.~ .  := ~ .

4

,~ p 4-i . ,' '

t.

rate decreasing as you move toward the center of the bundle. Defects indicated below the upper tubesheet are b located toward the periphery'in the 16th span and were random below the 16th" span. References 20 and 63 give a I detailed description of ECT results.

2. Tube End Damage In the fall of 1982, corrosion and cracking problems were _

~

identified in the steam generator tube ends, where they extend above the seal weld and upper tube sheet. The tube end damage was evident with metallurgical analysis of tube ends removed from the generators with the. Last 10-tube sample. After kinetic expa' nsion, damage was visible. The typical crack above che seal weld is a combination of both axial and circumferential cracking, the pattern depending on stress due to weld shrinkage in the heat affected zone of the seal weld, and on other factors. Meca11urgical_ evidence shows that the weld material arrested the cracks in all samples, although some cracks extend through the tubing material behind the weld to the tubing below. The force of

, , kinetic expansion removed parts of some tube tops where a circumferential crack was located in conjunction with verci-cal cracks. Other tube tops were bulged ou:, where vertical cracks nere through wall but circumferential crac.ks were in regions with ductile material remaining.

• In order to furth,er define the problem, GPUNC removed tube end pieces from the tops of approximately 12 tubes and con-ducted a metallurgical examination in order to define what, if any, ductility remained. The evidence from this examina-tion indicated that about 1/3 of those pieces removed were incargranularly cracked on all sides (both circumferentially and axially). '

Evaluation-of the metallurgical evidence indicated that the weld material arrested the cracks in all cases noted. ( Re f.

57). Additional dye penetrant tests were conducted on seal welds in the upper tubesheet to further confirm that the welds and the heat af fected =one between the tube seal veld and the tubesheet cladding were not cracked. The absence of cracking as noted in these dye penetrant checks provides assurance that the seal welds themselves and the upper tube-sheet cladding were not cracked. Similar examinations of the lower tubesheet welds and tube ends also showed no damage.

9 16 - F

. 4 e

L

s ,

4 & y With the damaged areas defined, GFUNC evaluated the poten-cial for loose pieces from the tube ends both above the seal

)-

weld and in the area behind the weld where vertical and '

.circumferencial cracks existed. This evaluation is docu-s mented separately in Reference 55. It was concluded that tubing below or behind the seal weld was unlikely to be i degraded to the point of loosening under the low loading in these areas. However, tubing above the seal weld was con-sidered to have potential to break loose. Thus, the deci- _

sion was made to remove all tube ends above the seal weld by milling.

J

' 3. RCS Itsoection The sulfur induced attack on the OTSG tube prompted an in-

spectAon of other elements of the Reactor Coolant System, to determine if other components sustained similar damage. An f inspection plan was developed based on a review of the materials involved and the accessibility of the materials within the system. Representative items in the Raaccor Coolat.t System that were most likely to have suf fered attack were selected for examination. The items chosen ~ represented the most susceptible materials and reflected environmental and stress concerns.

• Materials located in either of three environmental condi' -

tions were evaluated. -

~

' a. Primary coola'nt-air interface where most of the defects occurred in the OTSG. '

b. Dry areas since the last refueling, but which have,been previously wet.

c. Wet areas, covered by primary coolant.

Since the known attack had occurred in the OTSG on stress-relieved Inconel 600 cubing material (FWHT) which was under stress in the cold shutdown condition, this same and other sbailar conditions were, therefore, to be suspected in other parts of the RCS. In addition, attention was given to other materials which are known to be susceptible to IGSAC. Other than the OTSG tube preload stress, areas of concern with respect to stress included bolting that has a steady load due to torquaing, residual stresses induced by welding, and force-fit items.

e-

A

} ..

5

. 4 o

^ The plan included tests of sufficient diversity >to reflect the different materials, stresses, and environments that are

_ present in the RCS. The premise for this logic is that generic material groups will behave similarly. There fore , '

heat-to-heat varir.tions. were not considered unless evidence of intergranular attack and stress assisted cracking existed.

The inspection plan was developed to also account for crici-cal functions of the RCS items. The function of the pres- _

sure boundary, core support, and fuel integrity received the most emphasis. This was to determine the general condition

' of the system and, of course, because they are the most directly safety-related.

The non-destructive examination methods used were; ultra-sonic, liquid penetrant, eddy current, radiography, visual, and wipe sampling. Other examinations included functional t

check on equipment and destructive metallurgical examina-tions, both at the TMI-1 site and at B&W Research Laboratory 4

a at Lynchburg. The selection of examinations was governed by factors relating to the type of material, geometry of material, location and accessibility, and radiological con-trol limitations. The following is a summary of methods used and example materials examined by each method.

Ultrasonic Examination Method - this inspection included the pressurizer spray nozzle safe end, CRDM motor tube exten-sions, make up piping nozzles, plenum lifcing lug bolts,

, plenum cover to 'pIlenum cylinder bolts, pressurizer surge nozzle, core barrel bolts and Icw pr' essure injection pipe welds.

~ The ultrasonic method used to examine bolts of the TMI-l core barrel assembly had the capability of detecting indi-

- cations having a depth of 20 percent of the diameter of the bolts. This sensitivity is considered sufficient primarily i bedause a large number of bolts were examined at TMI-1 and j no evidence of intergranular attack or IGSAC was found. For  !

example, 96 of the core barrel assembly Inconel X750 bolts

- were UT inspected; if intergranular attack or IGSAC had j occurred,' it is likely to have been detected in this exten-sive sample.

1 Radiographic Examination Method - This method is a volu- l metric type of examination that produces a visual image of  :

the test specimen. For this reason, this method was chosen i to validate the structural integrity of the thermal sleeves H for the safe end nozzles. The pressurizer spray nozzle and  :

the three make up nozzles were located in a coolant / gas '

interface and the coolant dry area respectively. -

)

' ~

1

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

. ~ . . _ -

s' n.

, _; '. . . '~

+-

Liquid Penetrant Examination - Special consideration was given to the welds of the secondary oversheath to assembly oversheath of the incore detectors. Items examined by this .

method were: Upper OTSG Inconel (tube sheet) and stainless steel weld cladding and the incore detectors closure and' '

sheath, incore detector the dry region portion make up nos-

' sie, lower OTSG cladding surface and incore detector par-tions from the wet regions.

Eddy Current Examination Method - The ID surfaces of the RV

+

vent valve thermocouple and'the CRDM nozzle were the areas of special concerns which required this method of volumetric and surface examination. Both components are located in the area basically -dry of coolant.

V'sual Examination Method - Concern for the fuel integrity was the major reason for incorporating these inspections.

The areas of interest were submerged by the reactor coolant;

. the top of core control components, the baffle placa re*gion and the annulus between CSA and RV. Areas of similtr condi-tions, even though they were dry of reactor coolant, were the ' plenum assembly and the vent valve assembly.

Wipe Sampling Method - Th,is method was performed prior to non-destructive examination other than visual. The samp.les-were chemically analyzed to determine the concentration of any aggressive species.

~

The results of the inspections and tests which involved over a thousand selected components, indicated that there was no

- evidence of a problem similar. co that seen on the OTSG tubes. The functional tests all indicated that the tested assemblies were operational. The destructive examinations revealed that even on a microscopic level, no evidence of intergranular attack could be found. Therefore we concludt that based on this Inspection & Test Plan, the materials in the Reactor Coolant System are re-certified for continued -

safe operation. The details of this inspection are reported in Reference 21. ~l

4. Supporting Systems Inspection An ICSCC problem was originally detected in the Spent Fuel System in 1979, and a three year inspection program was established which was specific to Spent Fuel, Decay Heat and Building Spray Systems. As of June 25, 1982 all required volumetric examinations of the first cycle on the IGSCC sched..e were' completed and no discrepancies were noted. As e

e*

' 4' o. , ,

of August 5,1982, visual examinations were completed for Decay Heat and Building Spray with no additional indications identified. The plotting and trending of the known indi-cations did not reveal evidence of growth. In March 1982, -

cracks which were attributed to IGSCC were found in the Wasta Gas System.

In October 1982 and February '1983, examinations of PORVs which had been removed from service also showed damage attributed to sulfur intrusions. Damage to the PORV internals can be characterized as general corrosion and pitting rather than IGSCC. Valve internals affacted were made of martensicia, stainless steels and inconel. The corrosion apppeared to result from physical / gaseous transport of sulfur species. As a result of these findings, additional supporting systems inspections have been conducted, including inspection of the pressurizer, and a program was implemented to identify sulfur concentrations in plant fluids and on the surfaces of system components.

Results are docume1ted in Reference 64 and were reported in response to LERs 84-02, 82-11, and 83-03. Some minor

- corrosion and surface pitting were identified, but no additional significant damage was found. Damaged components have been repaired or replaced, and the pressuri:er has been

, cleaned by hydrotazing.

m 4

4 e -

e=

20 -

1 -

~

e- ., fa: ,

III.. CORROSION TEST PROGRAM A. Introduction An extensive corrosion testing program was~ initiated in December of 1981 to support the steam generator repair program. The pro-gram in several phases was designed to accomplish the following:

(1) Determine the conditions under which the corrosion mechanism occurred and how it could be arrested, (2) Verify the proposed -

corrosion scenario to provide assurance that the mechanism was understood (3) . Determine whether tubing that has been kinetically expanded would be more susceptible to corrosion in service than other tubing and (4) verify that cleaning using

, hydrogen peroxide will not cause corrosion. The following sections describe the results of this program.

B. Corrosion Mechanism Determination Tests In December 1981, analysis of tube s'amples removed from the TMI-1 "B" Steam Generator identified the corrosion mechanism as stress assisted intergranular cracking. Cracking was circum-farentially orienced and initiated from the primary side surface of the tubing. Analysis of the circumstances which led up to failure indicated that' through wall penetration of cracks oc-curred sometime af ter the hot functional test sequence and prior to the pressurizing of the unit in November of 1981. In view of this fact, a concern existed that the corrosion mechanism might still be active. .

A corrosion test program was immediately put into place to ascertain whether or not significanc corrosion was still occurring. The first of these tests was initiated in February of 1982. In this test, sensitized samples of 304 stainless steel and Inconel 600 were immersed in primary coolant removed from the decay heat loop. This coolant was analyzed and found to contain 350 ppb sulfate. Specimens utilized in this test were bent strip specimens spring loaded to apply constant loads near the yield point of the. material. Tests were conducted for two week periods at 1000y. Specimens were examined periodically for evidence of cracking an ultimately examined metallurgically to assess if any cracking had taken place. The result of this test indicated that the current environment in the primary circuit of the steam generators was not sufficiently aggressive to initiate cracks.

The next concern was whether or not existing incipient defects would, in fact, propagate under the environmental conditions which currently exist in the unit. To thi. end, an actual tube sample removed from the OTSC with a known eddy current defect -

determined to be a crack greater than 90% through wall was ~

casted in primary coolant removed from the decay heat loop.

This would have been a similar solution to that used in the

~ bC

, -w -

g . c .F initial. screening test. This test consisted of filling the tube i specimen with the decay heat solution on the internal surfaces, l then axially loading the specimen to 1100 lbs. at a test ,,

, temperature of 100*F. However, prior to putting the primary coolant ints the tube, the sample was also tested with load in dry air as well as air of.high humidity. In neither case were any cracks observed. After all testing was complaced, the specimen was examined metallurgically to look for signs of growth. There was no obvious extension of the intergranular cracks and no evidence of additional attack in the area of this crack. It thus appeared that crack growth was also arrested and 4

no further tube degradation was expected. This was confirmed by the eddy current examination performed on the 100 cube sample throughout the next several months. No evidence of any growth of known defects or detection of new defects was observed from December 1981 to the termination of program in July 1982.

In March of 1982 after this inititi testing had been completed, indicating that cracks were neither propagating nor initiating, a program was initiated which would define the environmental conditions necessary to produce the type of intergranular corrosion observed in the 'TMI tube samples.

• A number of tests utilizing stressed bent strip specimens were begun at the B&W Alliance Research Laboratory (Referance 27).~ l These tests utilized anodic polarization to accelerate the cracking process and help to define electrochemical potential regimes for this cracking to occur. Solutions of coric acid containing various concentrations of thiosulfate contaminant were tested. Those tests showed that thiosulfate at levels in excess of 5 ppm would cause cracking in sensitized archive tubes provided the degree of sansitization was sufficient. It was also determined that an oxidizing potential in the presence of a reduced sulfur form was required for this cracking.

Specimens made from actual TMI tube samples removed from the steam generscor were tested. These samples appeared to be more sensitive to the cracking phenomena since they cracked at thiosulfate concentrations as low as 1 ppa. This is believed to be due to either a difference in degree of sensitization of material removed from the generator or due to the effects of previous exposure of these samples to the thiosulfate

, contaminant in the primary system.

Samples were also tested in clean borated water during this phase of the corrosion program. It was found that in all cases, even when polarized in the cracking potential range, that in the absence of thiosulfate, specimens would not crack. Crackit. was observed at open circuit potential in an air saturated eviron-ment in thiosulfate contaminated solutions. However, if the

• e=

-. . _ . _ _ _ _ _ _ _ . .- _._ _ . _ _ _ . _ _ - _ _ _ - _ _ _ _ ~ _ _ .-_ _ - _ __

m .

-o: o-solution was deserated and an inere cover gas utilized, cracking was not observed in any specimens. Based on the~results of-approximately 60 tests it appears that thiosulfate or reduced .

metastable sulfur can produce and is a necessary requisite for the cracking observed. Additional results indicated that time to failure decreased as thiosulfate concentration was increased and also as temperature was increased up to 170cF.

- During this same time period testing was also being conducted at -

Brookhaven National Laboratories for the NRC. These tests were constant extention race tests (CERT) utilizing solution annealed and sensitiaed Inconel 600 test specimens. The purposes of these tests were to define the minimum thiosulfate concentration required for cracking as well as to establish the effect of temperature and Lithium Hydroxide concentration on cracking susceptibility. The results of these tests indicated the cracking in the absence of Lithium could be expected in highly

-sensitized material at thiosulfate levels on the order of 70 ppb . - However, in the presence of Lithium it was found that

. cracking would not be. experienced on sensitized materials provided the ratio of lithium to sulfur remained greater than or equal to 10. . Although additional tests are being planned to expand on the knowledge and understanding of the influence of lithium on inhibiting cracking, this data has been utilized in preparing new administrative chemistry guidelines for TMI-l operation. The lower limit on lithium has been raised such 'that at a concentration of 100 ppb sulfate the above lithium to, sulfur ratio is equalled or exceeded.

Brookhaven also conducted a series of costs to establish the influence of temperature on crack growth race. Results of their tests indicated that approximately 170'20F produced the maximum cracking velocity.

At this phase the evaluation had established that:

, o Cracks in the OT3G were not currently propagating o Cracking in non-reduced sulfur contaminated environment was not anticipated o The corrosion appeared to be a low temperature phenomenon o oxidizing conditions were required for cracking e o A highly sensitized microstructure was required o Lithium hydroxide could possibly be an inhibitor of crack 1 initiation or propagation.

From a metallurgical and corrosion viewpoint it therefore ap-pears that a repair process is feasible, that the tubing was not damagpd to the point where it no longer was serviceable; rather -

it exhibited properties of material which are typical for any currently operating generator.

• W .

14 o C. Corrosion Scenario 7erification Tests During the summer of'1982, testing was conducted at Oak Ridge .

National Laboratories in an attempt to verify the proposed sc enario. As defined in the failure analysis,. it was believed that corrosion occurred during the cooldown phase after hot funceional testing. i period in which oxygen was introduced into-the primary system as well as lowering of the water level in the OTSC's. It was felt that the lowering of the water level allowed reduced sulfur species to concentrate in the region above the water line and in the presence of oxygen caused crac4ing.

Although it would not be possible to totally duplicate the cor-rosion scenario, an attempt was made to establish test para-asters which were a close approximation to a, hot functional test sequence. (Ref. 24) This included chemistries similar to that. [

which existed at the time of the hot functional test as well as temperature cycles, plus exposure of test specimens in vapor as well as liquid phase. In addition, in order to account for any influence of tube surface films on the cracking mechanism, all te'st specimens were actual tube samples removed from the TMI steam generators. This allowed an assessment of whether oxidizing /

reducing conditions in the steam generators 'could change sur, face films and form metastable sulfur compounds which could lead to intergranular corrosion. Autoclaves were set ap' to test sulfur contaminated borated . vater solutions with 1 ppe and 5 ppa thiosulfate and 30 ppe sulfate. This latter test assessed if oxidized forms of sulfur of themselves would be aggressive. The test sequence allowed for examination of specimens after an initial exposure at 170*F. In all cases, no cracking was .

observed at that' phase.

The specimens were then put back into the autoclave and the temperature was raised to 500*F. Subsequently during the cool down to ambient phase, air was introduced into the system when the temperature reached 212*F. The specimens were then taken down to 130*F at which time they were held for several days.

Examinations of specimens removed af ter the hot functional sequence showed no cracking for the 1 ppe chiosulfate solution and no cracking for the 30 ppe sodium sulfate solution.

However, cracking was observed on specimens in the liquid phase of the 5 ppa thiosulfate test. No cracking was observed in the vapor phase of any test. This indicated that a threshold concentration of thiosulfate may be required for cracking to occur. What is not known, however, is whether the cracking occurred during heat up or cool down for this particular sequence. It may logically be assumed that cracking occurred during the heacup phase as tests conducted have thus far shown ~

cracking does not occur at elevated temperatures. In addition, tssting conducted in the 1 ppa thiosulfate and the 30 ppe

b. 9

.. .... e sodium sulfate indicates that a threshold level of available reduced sulfur is necessary and that sulfur in the surface film of itself is not sufficient to produce intergranular cracking. .

D. Roosired Tubina Corrosion Tests (Proprietary)

This corrosion data base, supported the conclusion that an ef-factive repair could be made. The choice of explosive expansion

' appeared to be the most technically feasible solutien, however, ,

with the expansion process came a noti geometry which would have a transition from the expanded portion to the- unexpanded portion of the tube which was not stress relieved. An accelerated short term test was conducted to assess the impact of this transition on the tube susceptibility to corrosion (Refarence 29).

Two tests were conducted to study this particular question. One test was a 10% sodium hydroxide electrochemical test which was known' to produce cracking in highly strassed roll transition regions and the second was an accelerated test in chiosulfate contaminated boric acid. Specimens consisted of Inconel single tube /cubesheet mockups with .an actual expanded joint. The caustic cast had previously been correlated with long ' term '

costing results, with approximately 8 1/2 years of service time

, . corresponding to approximately 5 days of casting under acceler-

, ated exposure conditions. Test results shotted no evidence of cracking.

• The second test was conducted using a boric acid solution con--

tain'ing 1 ppa of thio' sulfate and 1 ppm chloride. This level of thiosulfate was utilized because it was shown to produce cracking in highly stressed specimens of actual- TMI . cube samples. It was therefore felt that if the residual stresses were sufficiently high, that cracking could be expected in'.chis environment. The chlorides were added to provide an additional accelerating effect. Tests were conducted at 170*F. If no l cracking was observed in a specimen tested at 170*F, the sample L was recesced at 550*F.. Testing was conducted on ten single tube /cubesheet mockups that had been expanded using 19 gr/f t and i 14 gr/ft double expansion plus various ces binations of Lumunal treatment and H 02 2 cleaning. A complete description of test specimen configuration is given in Reference 29. These casts have shown no evidence *of stress corrosion in the expanded region or in other regions either at 170*F or at 550*F.

Even though all short cars tests were done under accelerated conditions, it was still felt that.a certain time dependency may be required for corrosion to initiate. Long term casts were I

~

i i

. _ . . , ._ . L _ _ *. ._n _

. . ~ . . . ,,

~*.,. .'

developed to address this time dependent parsmecer (Rafer-

'ence 8). These tests have-been scheduled to lead the actual

-performance of the generator and thus provide additional insight as to the expected performance of the tubes. The long ters .

corrosion testing program was developed to assess performance of the tubing both in the unexpanded regions of the generator and t tubing at the new expanded transition region. The test is -

• designed to run for approximately 13 months of operating time and to lead actual operation of the generator by a minimum of one month. As presently scheduled, lead testing will probably .

preceed operation of the generator by a minimum of 4 months.

- These tests will be conducted under simulated ooerational para-meters which will include load cycling _as well as thermal '

-cycling. Chemistry will simulate.that expected under normal reactor operations. This will include decreasing boron levels as well as decreasing lithium levels throughout the test period.

Test samples will be made from actual TMI staan generator tubes. Samples will' be utilized both with known eddy current defects as well as without known eddy current cafects. A mini- ,

mum of 4 different heats of material will be ut ilized with samples from various elevations within the generator.

In the lead test, samples will be 6 inch full section tubes as well as I.D. t. crossed C-rings. This test is intended to

. evaluate cond. cions in the span below the UTS. _The repair test .

will utilize single tube /tubesheet mockups with an expanded

, region in the middle of the tubesheet area. This' test is in-tended to evaluate conditions on the new joint and the transi-tion. Load cycling will subject samples to axial loads of.500 to 1100 lbs. The 1100 lb. load will be in conjunction with all cool downs from 600*7 to ambient. Tests will also include phases where oxygen will be introduced into the system at low . .

comperatures as might be experienced during normal generator shutdown conditions. Chemistry control will include vacuum deseration, addition of hydrazine, and hydrogen'over pressure.

It is believed that this will closely sinulate the reducing conditions that would be expected during normal generator opera-tions.

Samples in the lead test loops . include both defective and non-defective tubing in order to assess both the initiation and propagation phases of intergranular stress assisted cracking.

Tubes will be eddy current tested utilizing the .540" standard differential probe as well as a single coil absolute probe. The size and addy current signature of the currently known defective tubes will be monitored and any changes in crack shape or eddy _

current sit 11 will be closely watched. During the lead test, .

the tubes will be examined at the end of each test cycle (ap-proximately every other month) and assessments will be made as '

to crack initiation or growth at each phase. In addition, at the end of each test cycle C-rings will be removed and destruc-tively evaluated by metallography to assess the initiation of e

. +- . 4 . .. - ~

af .,E' J

any intergranular attack. Through this lead test program, an assessment can be made regarding plant operation in the unlikely event that crack initiation or propagation is observed.

To date, two' lead test loops have been subjected to the hot functional cost und three operating cycle simulations.

Potential sulfur contamination in solution was simulated using L sulfate in one loop and thiosulfate in the second loop. Tube specimens used were taken from unexpanded TMI-l tubing. One '

loop used semples which were precoated with immunol. Both six-inch sec- tions and C-rings have been a=amined. No changes in preexisting eddy current indicatiens were detected, and no intergranular attack was observed. The test loops are continuing through the remainder of the operating cycle simulations.

A third lead test loop simulated the hydrogen peroxide cleaning procesu, then continued through the hot functional tests and the i first operating cycle simulation. Sulfur in solution is I sulfate. All tube specimens used in this loop are six-inch sections of actual TMI cubing which have been Immunol coated and l subjected to expansion process debris. Additional specimens I which are representative of reactor coolant system materials have also been included in this loop. No intergranular attack has been observed. The cast loop is continuing through the remainder of the operating cycle simulations.

• l l

1 f~

The repair test loop,, which uses single tube /cubesheet uock- l ups, is being tested in a similar manner. The test sequence complaced included the H 02 2 cleaning cycle, hot functionals and the first operating cycla.~ All tubes were immunol precoated prior to expansion. After each cast cycle, these tubes will also be eddy current cested on a periodic basis to monitor" for new attack. No known defects were included into the repair test specimens. No intergranular attack has been observed. The repair test loop is now undergoing the second operating cycle simulation.

l1 The long term corrosion test program will provide a means for i

' saking a comprehensive assessment of tube performance in actual generator operation over long periods of time.

E. Conclusions Looking back at what has been learned about the TMI OTSC tube corrosion mechanism some conclusions can be drawn. Test results show that an active reduced sulfur species is required for low temperature damage to t'* Inconal 600 cubes. In the absence of this active species no crack initiation or propagation has been observed. Therefore, in the presence of clean borated water -

during normal

~

e

9 ,

8 .. .'

t operation, one does not expect cracking. If for some unforeseen circumstance cracking would occur by' reduced sulfur af ter return to operation, it would'most undoubtedly occur when the system is open to air at low temperatures. Cracking at elevated tempera- ,

cures under deserated conditions will be governed by those

• mechanisms and parameters which could affect any operating generator in the industry. To further address the low temperature concern regarding the oxidation of reduced sulfur species in the tube surface film, lithium levels will be administratively controlled at highc- concentrations than previously specified. Therefore, in the event there is formation of metastable sulfur species during an oxydizing *

, transient lithitra vill be present to act as an inhibitor.- .!

e 4

e 1

i 9

9 I'

f e

e*

28 -

t I -

. 7 5 P!"

iL IV. PREVENTION OF RECURRENCE "

A. Introduction Steps have been taken to ensure that an aggressive environment L could not be reintroduced into the RCS and cause additional 3 damage. Prevention of direct injection of contaminants will be r accomplished by administrative controls. In addition, the i

^

sodium thiosulfate tank had been removed. Chemistry changes [_ f have been made to include an analysis for sulfur, a conductivity consistency check which will indicate the need for reanalysis of -

samples, and an increase in the lithium concentration

• specification due to its inhibiting effect on crack ..nitiation. [

In addition, to preclude reactivation of the sulfur which in the -

OTSC and RCS, a chemical cleaning program was conducted to E oxidize and remove as much sulfur as possible from ene RCS. -

This section discusses the steps taken to prevent recurrence.

B. Prevention of Future Chemical Contamination Direct injection of foreign chemicals into the RCS during I periods of operation is essentially limited to those substances which are placed into the Lithium Hydroxide Mix Tank or the Boric Acid Mix Tank. Injection through the reaecor coolant bleed tanks during startup must also be considered. .The y probability of injection of a foreign chemical into the RCS from =

these tanks is dependent upon the administrative controls which -

are exercised over ad,ditions to the tanks. -

When the RCS is cold and depressurized, additional paths for -

introduction of foreign chemicals exist. A path from the '_

Caustic Mix Tank to the suction of the Decay Heat Pumps is,cne potential mechanism, and contaminants from the Borated Water -

Storage Tank and its associated piping systems is another. Since the sodium chiosulfate has been eliminated frem the plant, and

• the line to the thiosulfate tank cut, sodium hydroxide is the -

centaminant which could be introduced via either of the de-scribed pathways. In the event that very dilute caustic did -

reach the tubes, damage would not be exps:ted since increase in L pH is toward a more benign condition. Introduction of other 3

chemicals is prevented through administrative controls. -

Administrative controls which are in effect include (1) clear k labeling of tanks in the Chemical Addition Room, (2) locking '

open the breakers to pumps CA-P-2,3, & 4 and placing them under -

e the administrative control of the Locked Valve and Component List and (3) review nf applicable procedures to insure that -

adequate guidance is provided. _

1 E

=

_a

~s 6 Since the range of chemicals which could be injected if adminis-trative controls were to fail is wide, specific chemical analy-ses to detect the presence of the full range are not practical.

However, because of its potentially deleterious effect on the I -

OTSG tubes sulfur (as sulfate) will be sampled daily in the RCS. The other parameters which prove most useful in detecting ingress of un- wanted chemical species are pH and conductivity.

These para- meters vary with lithium hydroxide and boric acid concentrations with possible conductivity values ranging from a low of approx. 2 micromhos/em to a high of nearly 20 micromhos/em and pH values frem 4.6 to 8.5, depending upon operating conditions. Detection of inadvertent additions depends upon changes in either or both of these parameters which do not correspond to known additions, dilutions or creatment to the system. A consistency check on conductivity will be performed five times per week to confirm that the conductivity reading is consistent with the pH, boric acid, lithium hydroxide, and ionic species concentrations being measured.

Specific analyses based upon the conditions under which the changes take place can then further define conditions.

The increased administrative controls, removal of the sodium thiosulfate tank and increased sampling requirements will ensure prevention or quick detection of unwanted chemical contamination of the RCS. .

C. Changes in Oeerating Chemistrv -

Administrative primary water chemistry limits were implemented to prevent recurrence of the damage mechanism. This included an increase in the lower concentration limit for lithium des to its inhibiting ef fect on crack initiation and propagation, and an analysis for sulfur (as sulfate). A consistency check of pH and conductivity will be implemented. The check will improve our ,

ability to detect the presence of potentially harmful ionic species. Table IV-1 shows the changes in the Primary Water Chemistry Administrative Limits.

1. The lower limit for lithium concentration was increased from

.2 ppm to 1.0 ppm. This was done because lithium may have inhibiting effect on crack initiation and propagation.

2. Chloride was changed to meet revised 3&W Water Chemistry Guidelines.

~

O

i. v. *.

TABLE I**-1 THREE MILE ISLAND UNIT 1 PRIMARY b'/.!EE CHEMIST?.Y ADMINISTMTIVE LI:!!T CEANCES OLD E'J .

SAMPLING SA !PLING PARAMETER FREOUENCY FREOUENCY OLD LIMIT NE'J LIMIT -

Lithium

• NONE Daily 0.2 - 2.0 (ppm) 1. 0 - 2. 0 ( p pm)

Varies with baron

-concentration Chloride,l 5X/wk 5X/wk $0.15'(ppm) {0.1 (ppm)

Sulfate, (50 4=) NONE Daily NONE f,100 (ppb)

Sodium NO E OX/wk NO:3 40.1 p;m i

.pH 5X/vk 5X/wk 4.3-8.5 a.6-5.5 Conductivity SX/wk $X/wk NONE Check for Con-sistency '4ich Boric Acid and LiCH Concentrat.on O

4 e

l l

r -

i f g.. ,, n -

p

<' 3. Sulfur (reported as sulface) was added to the Administrative Limits because of its deleterious effects on crack inicia-tion and growth in Alloy 600. -.

4 Because sodium becomes easily activated and is an important contributor to total activity in the RCS, it will be moni-tored.

D. Cleanuo of Sulfur from RCS To preclude corrosion by sulfur contaminants already in the RCS, GPUNC conducted a chemical cleaning program to remove sulfur.

Testing of a steam generator tube showed that near the outer surface of the oxide film, sulfur was predominantly present as sulfate. Further into the surface film, metal sulfides pre-dominated.

The cleaning process was selected to chemically convert the metal sulfides on the stesa generacot tubing into soluble sul-faces. In general, the process used in the plant was as follovs:

The RCS, Makeup and Purificati'on System,' and the Decay Heat Removal Systems were in use.

Nitrogen added to the pret surizer vapor space was used co' increase system pressure to approximately 307 psig ,

Main coolant pungs were operated and cooling water flow through the Decay Heat Removal heat exchangers was adjusted so that the entire system operated at approximately 130*F + ,

, 5'.

~

When cleaning was initiated, the coolant contained a boric acid concentration between 1800 and 2300 ppa boron and  :

lithium concentration of 1.8 - 2.2 ppe.

Concentrated anunonium hydroxide ( 30 we %) was added to increase the reactor coolant pH to 8.0 ,8.5.

3 wt ?. Hydrogen Peroxide was added to the RCS from the 4%

boric acid mix tank via the core flood cank fill line and the pressure test connection on RCP "B" to establish a residual concentration in the range of 15 - 25 ppe.

Since the peroxide decomposes, continuous additions were made as needed throughout the cleaning process to keep the peroxide in specification.

5 o*

32 -

,w - , , - -+-,,cw - -- - --w , -- - - - -,- - r,-- ----n,,------- - -- - --v---

+

e- .

e my

,I'

..j .

• The cleaning solution was continuously circulated. Cleaning took approximately 2-3 weeks. Termination cf the cleaning '

process at 400 hours0.00463 days <br />0.111 hours <br />6.613757e-4 weeks <br />1.522e-4 months <br /> was based on developmental test results. ,i 4

The RCS sulfur cleanup behavior closely followed earlier laboratory simulations, with RCS sulfate concentration reaching a plateau at the end of 400 hours0.00463 days <br />0.111 hours <br />6.613757e-4 weeks <br />1.522e-4 months <br />.

Sulfate concentration af ter 400 hours0.00463 days <br />0.111 hours <br />6.613757e-4 weeks <br />1.522e-4 months <br /> was 0.4 ppe. The _

total sulfate picked up in the coolant was 0.33 lbs.

Both the dissolved sulfate and the ammonia added were

' removed from solution by ion exchange resin in the normal purification systems.

A comprehensive test program was performed to determine the effectiveness of the cleanits process, and to verify that the conditions of cleaning would not introduce a corrosive environ-

' ment. The program and results are discussed in a separate safety evaluation. Hydrogen peroxide appears to be effective in removing sulfur from both tubing surfaces and from inside crevices. Based on testing, 400 hours0.00463 days <br />0.111 hours <br />6.613757e-4 weeks <br />1.522e-4 months <br /> of exposure to the hydrogen peroxide solution is expected to have removed up to 80% of the sulfur which was present in deposits in the OTSCs.

E. Conclusions .

GPUNC has implemented corrective actions in four areas in order

, to prevent recurrence of OTSG corrosion. The sodium thiosulfate tank has been removed from service to prevent additional in-advertent introductions of sulfur, and chemical cleaning was conducted to remove existing sulfur. Stricter administrat,ive controls have been placed on introduction of other potential t-chemical contaminants. Stricter controls have also been'placed on RCS chemistry to maintain a non-corrosive environment. These actions are expected to prevent contamination and corrosion by sulfur and by other chemicals.

t e-4-  !

l

8 .  ? .

t a e ,

, s V. KINETIC EXPANSION REP' AIR DESCRIPTION

SUMMARY

. A. Description of Process and Geometrv 1.' Introduction TMI-1 OTSC cube examinations have revealed a large number of tubes'with defects within the. upper tubesheet. A defect is-  !

defined as any eddy current indication interpreted as greater than 40% through wall. The linics of eddy current detectability are defined in Section IX. The repair approach is to establish a new primary system pressure boundary below these defects. A kinetic expansion of the

, tube within the tubesheet was used to effect this repair. '

All tubes which were not plugged were kinetically expanded irrespective of whether or not they have a defect, and irrespective of whether they were to be plugged following expansion. This repair provides a load carrying and essentially leak-tight joint below known defects. The following sections summarize the repair program. Details can be found in Reference 1 and Reference 23.

2. Kinetic Tube Exeansion (Proprietary)

The process steps which are involved with this repair have the objective of providing a new pressure boundary below" known defects through kinetic (explosive) expansion of cne tube within the c,ubesheet. Extensive testing by Foster Wheeler has indicated that the most effective kinetic ex-pansion technique for this repair is to use a 19 gr/f t ini-tial shot followed by a second shot at 14 gr/ft. The chrege is provided by a detonating cord held in picce by a pl,astic insert. An electrical charge' sets off a cap igniting the transfer cord which in turn fires a booster setting off the detonating cord.

Preliminary testing has determined that a 6" long expansion below the lowqsc defect will provide the desired load carry-ing margins. The expansion serving as the new pressure boundary is the bottom six-inches of a 17 inch expansion extending through the cracked area to the top of the upper tube sheet. Thus all tubes for which the lowest defect is at 11" or ab'ove have been provided with a new six-inch joint. Tubes with defects lower than 11" will be considered individually. Those with the lowest defect between 11" and 16" will be expanded using a 22" expansion. Those with defects lower than 16" below the top of the upper tubesheet will be taken o.. of service.

0

-,- ,,,,m -

'6

• e- ,. .

The TMI-1 OTSG repair process is as follows.

Step Description 1 Flush the secondary side tube to upper tubesheet crevice.

2 .

Heat crevice to drive out moisture (vaporize. water). _

3 Precoat tubes with Immunol 4 Kinetically expand tubes 5 Clean debris from kinetic expansion 6 Mill tube ends 7 Flush OTSG 8 Plug necessary tubes i 9 Clean OTSG with felt plugs.

B. Design b$ses of Kinecie Joinc l The new joint comprises a kinetic expansion of eit.ter 17 or 22 l l

inches which begins just below the upper tubesheet top surface

, 1 in the area of the. original shop roll expansion. The kinetic l expansion will be the pressure boundary and structural attach-ment of the cube to the tubesheet.

The original OTSG design basis is sununarized in Reference 'l.

The following is' a summary of design basis for the new kinetically expanded joint.

1. The repaired tube shall sustain the maximum design basis axial tensile load of 3140 lb. from the generic 177 FA MSL3 accident analysis. Since this is a thermally induced load, satisfying this criteria requires no relative'movemenc (slip-page) between the expanded area and the cubesheet at the axial strain corresponding to .:is load (about .0016 in/in).

2. Thermal / Pressure Cveles The initia1' design life objective for the tube kinetic ex-pansion is 5 years.

Sufficient cyclic testing and/or analysis will be performed during the qualification program to satisfy this objective. -

^

l i

'(. 4' e (I.

A design life of 35 years has been established as a goal.

I]l j

Additional qualification testing has been performed which

'I consisted of leak tasting of specimens which had experienced cycling equivalent to a 15 year life. Specimens used were '

separate from those used to initially qualify the repair for a 5 year life. Assessment of the joint for the full 35 year design life will be performed when data on actual steam generator performance is available to supplement the results of the 5 year and 15 year test programs. .

3. Tube Preload

' The design objective stated in the original steam generator equipment specification for TMI-L OTSCs was that the tubes not be in counression when cold.

3 The repaired tube censile preload shall not be changed by more than + 30 lbf at ambient temparature. This design

, objective is intended to assure that tube preload tension is maintained so that the vibrational characteristics of the~

tube will be unchanged for a preloud change of this magnitude.

4 Residual Stresses 0se design objective is to minimize tensile strass in the transition region between che expanded and unexpanded '

portions of the tube. Analysis shows that an abrupt transi-tion results in higher residual stresses and larger stress concentrations. A transition length between 1/8" and 1/4" has been established as a goal.

An objective of maintaining additional resifual tensife-stresses (both circumferencial and axial) resulting frem kinetic expansion in the transition less than 45% of the .2%

offset yield stress at room temperature has been established.

5. Heat Transfer Requirements No credit is taken for heat transfer within the tubesheet.

6. Pressure Boundarv Leakare The original design basis for steam generator tube leakage was to provide generators with no detectable leaks at ship-ment and to control leakage to an acceptable operating level by monitoring and repair over the 40 year life of the plant.

The kinetically expanded joints used for repair of the TMI-l steam generators are designed to be essencially leak tight. -

The expansion is designed to provide a seal below potential

y = +

^* -*

. ,, . s

. leak paths in all tubes to be repaired. Tubes with unacceptable leakage as indicated by the precritical drip

' and nitrogen bubble tests (see Appendix A) may be roll expanded above the lower 6" to' attempt to seal the leakage.

• If this is unsuccessful the tube will be plugged and/or stabilized if necessary.

For plant operation, primary-to-secondary steam generator leakage limits will continue to-be set by the Technical Specification limit of l' gym. However, in order to control l the amount of waste that requires processing, a design goal of 1 lb/hr projected total leakage from both generators has been set for the qualification program. Subble testing can distinguish a leak that is of the magnitude of 0.1 gallons per day. An engineering evaluation of bubble test results as they relate to expected leakage will be conducted in order to determine'what tubes require plugging. Statistical analysis will be applied to the verification test results.

C. Cualification Program A series of' mechanical cests and'* chemical and corrosion tests were performed to qualify the kinetic expansion, and the kinetic expansion 9rocess to meet the design goals of producing a joint capable of. carrying required loads, providing a leak.cight seal, minimizing rr sidual stress, and tube preload changes. A series e

of prelir.ipary tests was conducted to establish the optimum parameters for a kine, tic expansion process that will yield

, acceptable joints with low residual stresses. Additional tests were conducted on a full size steam generator at B&W's Mt.

Vernon Works. A more detailed description of the tests and results can be found in Reference 23. .

1. . Mechanical Tests

c. Preliminarv Leak and Axial Load Tests (Proprietary)

Kinetic expansions were tested to determine the maximum axial load which could cause the expansion to slip. ~

After a set of expansion parameters were postulated, leak rate and axial load tests were performed to deter-mine whether the expansion would still appear adequate for a corroded tubesheet, afte- thermal and pressure cycling, and after adjacent tubes have been expanded.

The following acceptance goals were applied.

(1) Water leak at a pressure differential of 1275 psig (Primary to Secondary) 3.3 x 10-5 lb/hr per tube.

W 4 es + me

~

e n L

(2) Pullout load consistently above 3140 lb per tube.

(3) Margin in pullout loads and leak race to account for ,

possible deterioration-of joint integrity from thermal cycling and for statistical analysis.

(4) Minimize expansion length.

(5) Minimize longitudinal strain induced in the tube by -

the expansion process.

(6) Minimize in plane deformation of the expanded tube block hole and adjacent. holes.

A 6" expansion using a 19 grain /ft insert followed by a 14 gr/ft insert gave acceptable preliminary results. Pullout loads for high and low strength tubing with the expansion as a function of the number of af ter-hits on uncorroded blocks showed pullout loads consistently in the 4000 to 5000 lb range for up to 4 after hits. Pullout load results for expansion on corroded blocks using corroded high

, and low yield cubing were consistently in the 4500 to 5500 lb range.

The effects of thermal and pressure cycling on ~

pullout load were minimal. '

b. Leak and Axia'l Load Qualification Tests (Proprietary) j i

These tests predicted the leak tightness and confirmed the axial load carrying capability of the chosen eKpan-sion technique, and showed what effect kinetic expansion will have on adjacent repaired tubes and determined the j effect of re-expanding previously expanded tubes.

Seven blocks were charmally cycled as follows:

38 cycles 70*F to 610*F to 70*F -

l One block was then exposed to a series of lead j cycles. These were:  !

h 100 cycles 780 lbs. compression to 1110 lbs. cension 180 cycles 635 lbs. compression to 175 lbs. cension

, 6000 cycles 510 lbs. compression to 125 lbs. compression Cycles selected correspond to the 5 years qualification period.

O V

l l

• 8 .

4 The majority of leak and slip load qualification.

testing was performed at room temperature with the exception of one block loaded at 330*F and one block leak tested at 400*F. -'

Acceptance criteria required that a statistical evalua-tion of the results show a 99% confidence level that 99%

of all tubes expanded would have a pullout load greater than 3140 lbs. A mean leakage race goal of less than 3.2 x 10-5 lbs/hr/ tube was desired.

Results indicate that thermal cycling tends to decrease pullout load, however thermally cycled blocks pulled at

, 70*F gave a 99% confidence level that 99% of the tube expanded will have pullout loads in excess'of 4170 lb.

One block which was pull tested at 330*F gave 99/99 statistical confidence that pullout load would be in excess of 3590 lbs. The goal of 99% confidence that 99%

of the tubes have pullout above 3140 lb is easily met.

In addition, an expansion pull test perfoTmed on a full scala generator at Mt. Vernon showed a load carrying capability of at least 3600 pounds.

Leak rate re:iults af ter thermal cycling equivalent to 5 years a vary from 1. L8 x10-6 to 187.4 x 10-6 lbm/hr/ cube, The average tube leakage was considered to be one-cench of the total test block leakage in each case. Statistics indicate a 99% confidence,that 99% of the normally expanded tubes will have leakage rates no greater then 132.4 x 10-6 lbs/hr/ tube. While this rate exceeds the design objective of 3.2 x 10-5 lbm/hr/ cube, it is still a very low leak race. Results of leak races.after axial loading are found in Reference 23. If every tube in both OTSGs' leaked a'c this maximum race the cumulative leak race would still be less than one-hundredth of the Technical Specifications limit of 1.0 gym. The leak race of the one block showed an increase between 10*F and 400*F (6% of the total range of leak races) leading to the conclusion that the leak race for a tube at operating temperatures would differ only slightly from what it would be at room temperature.

Leak rate results after thermal and load cycling equivalent i to 15 years vary fren 31 x 10-6 lba/hr/ tube to 68 x 10-6 lba/hr/ tube. This 15 year cese data is comparable to ch'e 5 year test data with no indication of joint leakage increasing due to thermal and load cycling.

a. W e

e , .

c. Residual Stress Testing (Proprietary)

C (1) Praliminary Transition Geometric Limitations This test decernined the expansion parameters which would lead to a transition that would minimize the transition residual stress and stress concentration factor. It was concluded that a transition length between .125 and .25 inch would be a goal, vich a

• minimum acceptable transition length of .1-inch. A number of insert shapes were evaluated to determine which provided a smooth transition.

A 30 degree taper on the end of the polyethylene insert was found to provide the optimum transition.

(2) Residual Stress Measurements The actual residual stress was measured in special i' cast blocks using X-ray diffraction and strain gage te hniques to determine post-kinetic expansion tube stresses in the transition area at the bottom of the -

expansion and at a second point near the middle of the expansion. Both hard rolled and kinetically expanded tubes vers examined using high and low yield _ strength materials.

• The'gou,1 ,for this cast was that the additional resi-dual stress in the tuce resulting from kinetic ex-pansion would not exceed 45% of yield strength.

Results are reported in Reference 23.

(3) Comoarison of Einetic and Roller Exnansion Sample Inconel 600 cubes were expanded by rolling and kinetic processes in ordar to compare the re-sulting hardness and microstructure.

Both processes increase the hardness, with the roller expansion showing a greater hardness (100 to 103 RB equivalent) than the kinetic expansion (92 to 93 RB ). The roller expansion tends to harden the inner surface more than the outer, while the kinetic expansion is slightly harder at the outer surface. Prior to expansion the tube was signi-ficantly harder on the ' outer surface. Since the hardening effect in the mechanically expanded tube is more pronounced and less uniform than in the kinetically expanded tube, the kinetic expansion may -

be expected to be less susceptible to stress-corrosien cracking than mechanical expansion. j

_ 40 _

-1

• M

o ..

(4) Corrosion Testing of Transitions

r Accelerated stress corrosion cracking tests wer's t performed on expanded tube /tubesheet mockups. The

• sockups were tested in 10% sodium hydroxide (NaOH)

~

l at constant potential and destructively examined for ,

stress corrosion cracking (SCC) due to residual l stresses from the repair expansion process. The-results were compared to existing data on tubes that have been stress relieved after expanding. Two test ~

configurations were used; the first was .an Inconal 600 cube kinetically . expanded in the tubesheet, the

' second was_an Inconel 600 cube kinetically expanded plus a hard roll in _ the expanded region.

Test results show no evidence of cracking.

d. Induced Strain Tests (Proprietary)

Tests were performed to determine the effects of the expansion on the tube-to-tubesheet velds, and the tube length, and to determine the strain scored in the ex-pansion. A design goal of changing the preload by less than ;,30 lbs due to elongation was applied.

Results show that kinetic expansion has a minimal effect on the overall longitudinal tube strain and as-fabri-cated preload,. . Induced strain measurements taken before and af,ter expansion indicated' maximum longitudinal' strain values of less chan .04 percent. - This relates to a reduction in tube preload of about 16 pounds, which is -

considerably less than the 30 pound design limit.

In some cases, prior to expansion, the degradation of the tube in the area of- the seal weld allowed the tube to . slip down, relieving all or part of the preload.

These tubes were' then expanded.

These tubes have been evaluated to determine the potential effects of loss of preload, and are discussed i in Section VI-E. l

e. Ligament Distortion (Propriocary) l The effects of explosive expansion on the tubesheet ligament were determined. The dimensions of adjacent holes in the tubesheet were measured before and after -

the expaar# in and compared.

Results show that only a minimal effect was noted in the -

diameter of adjacent holes due to tube expansion using charges between 25/14 and 14/14.

• i

s, <

i- a ,

.o Full scale testing in a steam generator at Mt. Vernon using strain gages and profilometry showed no degreda-tion of the tubesheet ligaments. ,

Based on the mechanical qualification tests it can be con-cluded that the kinetic expansion joint will meet the five-year design life objective. The repaired tube will sustain the maximum design bssis axial load of 3140 lbs., residual stresses will be minimized, tube preload will not change -

~

more than '+ 30 lbf, and leakage will be much less than tech-nical specification limits.

2. Chemical and Corrosion-Testing

a. Residue Test (Proprietary)

The amount and type of explosive residue that shoule'. be expected to remain in the steam generator af ter all tube repairs are completed was determined. A satisfactory cleaning method to reduce contaminants to acceptable

, primary system water chemistry levels was identified.

Results show that the chemical composition and amounts of the residue are acceptable. _ The use, of ordinance transfer cord minia.izes the amount of debris-in the upper head. "

Techniques for cleanup of the tubes af ter kinetic ex-

, pansion,will include felt plug wiping on.che inside of the tubes and water flushing the top surface of the upper tubesheet, tubes and the inside of the upper head. To enhance the water flushing, all primary side surfaces exposed to kinetic expansion residue will be coated with a film of water soluble Immunol X-236.

b. Crack Chante Tests (Proprietary)

The effect of kinetic expansion on existing cracks was determined. i Explosive expansion using 25 gr/ft cord was found to a

cause no damage to the base metal, although existing cracks were found to be slightly separated so as to become more visible.

Double expansions using 19/14 were done on several {

samples from TMI-l steam generator and on sulfur induced i laboratory grown IGSAC cr. ks that were less than  !

through wall. Neither through wall nor less than through wall cracks propagate circumferencially or ~

, a m ,. , , , - , , - . ,, - - - , , .. ,, ,. ,-- , - . - - - - - . . -

through wall. They are however slightly widened to become more visible. There is no evidence of any ductile tearing of tube material. -

c. Effects on Residual Sulfur Testing was performed to assess what happens to the sulfur on the surface of steam generator tubes (i.e.

driven into the base metal) af ter kinetic expansion. -

It was determined that the sulfur concentration on the tube I.D. in the area of the kinetic expansion does not change, that it ia not driven further into the base metal, and that the expansion does not significantly alter the grain boundary structure in a way that would trap sulfur.

D. Renair Testint The in process inspection and monitoring program was . designed to verify that the in generator expansions are similar to those obtained in the qualification program. Actual OTSG expansion

• profilometry and ECT results were compared to test program data co verify that the expansions are similar. Data obtained from IMI-l was also compgrad statistically to test program data. .The program consisted of video surveillance, profilometry measure-ments, and eddy current (ECT) examinations.

Video surveillance of operations during the expansion process were conducted where practical to verify that proper procedures were followed and that the correct tubes were expanded or examined. Random out-of generator expansions were also con-ducted to verify that expansion inserts had not chcnged since the qualification program.

Verification sampling was performed on the tubes expanded by the initial charge strength in the first three lots in each OTSG and consisted of ECT and profilometry. ECT using an 8xl probe was performed on almost 100% of the tubes expanded in the first lot in both OTSG's. Profilometry was performed on expanded tubes selected at random from the first three lots.

In addition to verification sampling, random diameter and depth checks sampling were done following initial expansion. The '

l sampling plan can be found in Reference 19.

Results are presented below of both the expansion inagth in-spection program and the program to verify that each tube was expcuded as - required. ,

o , .

~

1. Results' of the Ernansion Length Insoection Program Out-of-generator l expansions indicated that the process

. expansion inserts and detonating materials performed as ~

those used in. che, qualification program. Profilometry and diameter and depth gauge checks showed that the in generator expansions were within the' range of variation of the quali- -

fication program expansions. In addition, eddy current examination using the absolute (8x1) probe was conducted 'for ,

the first lot of kinetically expanded cubes in both steam

, generators. The 8x1 absolute probe'has been chosen for this in process monitoring in the newly expanded area because the coining process of the expansion creates so much background noise that the .540" standard differential probe is not useful following kinetic expansion.- The 8x1 absolute probe provides 360* coverage. A judgement concerning defect are length can be made depending on how many coils of the 8xl probe detect an indication. Laboratory testing has shown that a 1 coil indication can have an are length of 5* to 40*, a 2 coil indication has an are length up to 85*, and a 3 coil indication has an are length up to 130*. Although the 8x1 absolute probe can be used to quantify the circum-ferential extent of a particular indication, it cannot be used to accurately determine the percent thru-wall of the indication. The scope of the examination included 151 cubes in OTSG B and 284 tubes in OTSC A. The eddy current dat's was analyzed from the top of the 6" qualification length for kinetic expansion down through the bottom of the upper tube-sheet. As a result of this data evaluation, 9 tubes in OTSG B and 6 tubes in OTSC A were reported as having indications which had not previously been detected by the .540" OD .

high gain standard differential probe. ,

These nov 8xl ECT indications were evaluated to determine their significance. The evaluatica included 1) Fiberscope examination of selected tube areas where indications were detecte'd, 2) Comparison of the size of any visual indica-tions to the ECT sensitivity curves, and 3) Laboratory metallographic and ECT examination of known cracks which had been expanded. The following was concluded from this evaluation:

1. The only visual evidence seen which correlates with the ECT indications is small pics and a mechanical scratch.

2. Laboratory expansion of known cracks confirms no growth (no ductile tearing) and indicates no change in 8x1 ECT signal from known cracks.

W

i

3. If some of these ECT indications are from small cracks which were previously below the .540 ECT sensitivity within the upper.tubesheet (UTS), their geometry is so '

small that the reliability of the new joint is not affected.

4. The most probable reason for chase new indications is that the 8xl probe is more sensitive than the .540 probe within the UTS. The 8xl probe appears to be sensitive enough to respond to pics which are so small they are of no consequence.

Section II documents extensive work done' to evaluate the maximum size crack which can be left in service for the life of the plant and not cause tube failure under norsal or accident tube loadings. Acceptable circumferencial extent vs. throughwall depth curves for various -loading and analysis conditions in the free span are shown in Figure II.2. The .pic indications found in the area of the joint are smaller than the crack size leading to failure by any mechanical means in the free span. These curves are con-servative for indications in the joint since loads imposed on the tubes are transmitted to the tubesheet in the area of the expansion. Loads on tubing in the area of the defects will be equal to or less than thoste analyzed for,the free-span. Leakage through any small defects which are 100% -

throughwall is also expected.co be less than-or equal to similar cracks in the freespan. Unacceptable leakage will be identified during precritical testing and the tube will-be either plugged or repaired. For these reasons, it is concluded that small pics or undetected cracks in the quali-find area do not affect the reliability of the new joint.

It is expected that additional indications will be ident'i-

, fied during the baseline 8xl eddy current examination of the expanded region to be conducted following the kinetic expan-sion. These indications will be evaluated to confirm that they are acceptable, and will be left in service and re-exmained during the 90 day ECT program.

2. Identification and Expansior. of Misfires (Proprietary)

Inderts which did not fire were generally easily identi-fiable because they remained in the tube. However, during the 19 grain / foot expansions several misfired candles were ejected from the tubesheet. The locations of these we're found using a diameter gage to determine which tube in that '

sequence was not expanded. Because there is no measurable 1

. , - , , , - -, - ,- r , -- ----- -- - -n s.

_ ._ ~ _ - - -

n. .

. . e change in tube diameter caused by the addition of a 14 grain / foot expansion this technique could not be employed during the second round of expansions in the event of an unidentified misfire. The. cause of ejected misfires was ~ '

thcught to be a whipping action of the ordinance transfer cords in which sufficient kinetic energy was imparted to the cords by the rising candles to pull a misfired candle from the tube. Devices were made to lock the transfer cords to the tubesheet in order to prevent the occurrence of

" jumpers." ~

In spite of precautions taken, " jumpers" occurred in several separate instances. In these in1tances the tubes of the previous sequence were reexpanded with 14' grain / foot devices to assure that the " jumpers" had received 14 grain / foot expansions. This was done in such a manner that no tubes ever received more than 4 after hits (the maximum number presently technically justified for normal tubes). However, during one expansion sequence in the "B" generator a mis-fired candle was discovered under some shielding. The sequence to which that candle belonged could not be identi-4 fied and the misfire loe ion could not be defined more 4

precisely than one of thu first 7000 cubes done in that generator.

The concern for a singly expanded tube is that, potentially, being unable to carry sufficient axial load, the tube would fail, slip downward and lock. Therefore, an evaluation was.

performed to determine the likelihood of slippage during operation, and the potential' consequences if the singly expanded tube were plugged, stabilized or lef t in service.

I A statistical

• evaluation of available data for single Z0 grain / foot expansions frem pre qualification testing showed '

?. a 99/97.5 confidence interval that no slippage would occur

,. under normal or accident loading. While the data used is for 8" expansions and for expansions receiving many less after hits than a typical singly expanded tube would receive in the generator, the resulting high confidence is felt to be indicative of what would be expected if a test program using more correct configurations were performed. Actual pullout strength will be dependent on the location and size of cracks in the singly expanded tube. Leakage data for tubes with one expansion is available from the preliminary qualification program for a defect free length'of 8". For this length, leakage is expected to be sufficient to be identified during precritical testing. Similarly, leakage will be greater the shorter the defect free length. In addition, the tight annulus between the tube and tubesheet r

will provide a substantial restriction to any leakage. -'

e

.g_, , , - - . , . , . . _ _ , ,y. -,4-, ,.s,.., _ , . _ _ _ _ , m ..-

t' - .c - , n- - , ,.w . - - = - r,-- _-,m

p. .. -

In the event that the single expanded tube will remain in service, then slip under transion't or accidenc ~ conditions, effects of slippage were evaluated. Assuming a minimum 6" leak length and a worst case 0.001" radial annulus, the leak

• rate past the expansion would be.about 0.25 gym or one quarter of the Technical Spec.ification. limit. This annulus size is considered to be conservative because any actual 1eak. path past the expansion is expected to be a discrete path ~such as a scratch. A leaking tube would be located by conventional means and would be plugged and, if necessary,

~

stabilized. '

Of the approximately 7000 cubes which could have been the single expansion site, only about.100 were to be plugged.

Slippage of a plugged tube could not be identified due to leakage. Th'us if the tube slipped and locked, it could bow or be dynamically unstable in cross flow areas and cause

• wear damage to adjacent tubes. Approximately half of the plugged cubes will be stabilized, which should prevent wear

- damage. The remainder are in areas of minimal cross flow, so dynamic -instability should not be a problem. Even if the single expanded tube were one of these, and it slipped '-

enough to bow under operating conditions, a relatively small point load against the neighboring tubes or tubesheet could force a node at the point of contact and prevent wearing.

Based on these evaluations, it was concluded.that the like-lihood was email that a single tube expanded only once would be located in an~ area which mi;ht result in problems, and would then slip with loading. Therefore, no further actions were taken.

E. Post Reesir Testing -

• Following repair, testing will be performed to verify the acceptability of the joints. Post Repair casts include:

1. 150 peig bubble test. Upper tubesheet plugs, tubing and expansions leak test.

2. 150 psig drip test. Lower tubesheet plugs, cubing and expansions leak test.

3. Hot OTSG Testing Hot OTSG testing will include transients that will place operating loads on the new joint. These transients will include:

~

47 -

l I

+- -

--c. - , - , - - y -y_m -_..,_m,,.,,em,.c 7 --.,_--,,--.-__w .

,, --,c ._.,,,,,y,g - , , , , -----,-,-ww.----

=

r r .,f .

9 .

a. normal cooldown

b. accelerated cooldown

c. normal cooldown using modified tube rupture emergency peccedure { '

Leakage will'be monitored before, during, and after the transients. A 1400 psid operational leak test will also be '

conducted. . A more detailed explanation of the testing programs can be found in Appendix A. ,

The cold testing described in. items 1 and 2 above has been completed. The results are discussed in Appendix A. Based

<nt the drip and bubble tests, and investigative ECT inspec-tion, GPUN concluded the following. With the possible exception of four tubes which bubbled only slightly cud were not investigated further during stopper casting or by ECT, all leakage was found to come from below the top of the qualification length of the kinetic expansion joint. No joints were found to leak.

F. Conclu'sions

3ased on a qualification program, the kinetic joint meets or exceeds the design bases of the original joint, including the followin's factors

,

a. Load-carrying capability. -

b. Tube preload. ,

c. Minimization of residual stresses.

Leakage is projected to be less than one-one hundradth of the technical specification limit of 1 gpm. Kinetic expansion in the upper. cub' e sheet is a safe and reliable method of repair for all tubes that will remain in service in the TMI-l steam genera-tors. The tube joints will remain structurally sound and essen-tially leak-tight during all design conditions over at lesse a five year period.

m

)

~

i 60m e _ , - . , . ,m 7,. ,, , . - . . v_ -- .-, , ._ . - , . . , , - - .

s- . , +: .

VI. EFFECTS OF EXPANSION REPAIR The possible effects of the kinetic expansion process with respect

, to introduction of chemical impurities, the effect on the OTSG -

structure, the effect on tubesheet corrosion characteristics and the effect on existint plugs have been evaluated.

'A. Possible Introduction of Chemical Imourities The. specification for OTSC cube repair addresses the issue of ~

Lapurities in the system. It specifies that the inside of the steen generator will not be exposed to materials containing more than 250 ppa sulfur and 250 ppe total chlorides and fluorides, ,

and'specified detectable amounts of. low melting point metals.

- Required deviations will be addressed on a case basis. A material control program with quality controls was required'for

^ confirmation that material is not introduced inside the steam generator without assurance that its constituents are known and acceptable was implemented.

The large majority of the debris was demonstrated (cesting in mockups and a full-scale OTSG) to be particulate matter. The large particulate debris was removed by manually picking up pieces by hand, and by vacuuming in both the Inwer and upper heads. -Partic.ulate debris in the tubes was removed by forcing felt plugs through each tube. '

In addition to the ea,sily removed particulate debris, laboratory tests showed a thin layer of material was deposited on the ex-posed CTSG surfaces. This film consists primarily of carbon, identified as polypropylene. The amounts of contaminants, sul-fur and other elements, in this layer are low (traces only). In order to minimize the potential for the film interfering with a subsequent sulfur removal treatment, steps were taken to j i minimize film thickness. The GTSG surfaces were coated prior to 1 expansion with a substance which, when flushed with water, l reduced the film thickness significantly.

f This residual film is not uniform over the length of the tubes,  !

being thickest nearest the expansion. The results of tests show  !

a remaining layer, after flushing, which averages 50 Angstroms l thick over the length of the tube. A similar condition should '

exist at the top of the cubesheet surfaces and the inner sur-faces in the done of the upper head.

As a v'orst case, it may be assumed that the 50 A film on the 31,000 cubes uelts as the reactor coolant approaches operating temperature. The polypropylera is not soluble in hot water, but

' ~

d

?

, .~m . + . .

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

3.g,  :

• P -

p

,. the high velocity coolant through the tubes could entrain par-ticles as they sof ten had lose film tension on the tube sur-face. The turbulant reactor coolant flow would carry these. .

minute droplets of molten plastic throughout the reactor coolant system as a very dilute emulsion of about 0.44 ppa polypro-pylene. The demineralizers and filters in the letdown system

): would gradually remove this slight impurity from the reactor

_ coolant. Industry experience shows that there may be a tendency for the so1Len . droplets to collect in relatively stagnant areas -

of the reactor coolant system. The reactor head would be the most likely area of concentration, and during cooldova che poly-propylene may solidify as a film on the under side of the dome.

In the unlii.ely event that the total volume of plastic were to separate out in the head area, the CRDM lead screws would col-lect a negligible amount of residue.

. 'Should thers be a reactor cooldown, any molten droplets re-maining in che coolant would tend to solidify as a film on the coolest surface available. Both the letdown and decay heat systems will be cooling portions of the reactor coolant and the polypropyleue would redeposit as a thin film on the latdown and

decay heat' cooler tubes. The hot surfaces in the core are the '

j least likely places' for the plastic to solidify. There is no reasonable assumption which would indicate the half, cup volume of plastic film could cause a pr'blem o before it is removed from the system. .

B. Possible Effects on GTSG Structure

• It has been postulated that the kinetic expansion may, because i of the large number of tubes involved, have significant effects on the steam generator as a structure as vell as on the indi-

. vidual tubes.

For individual expansion, evaluation of.the tubesheet ligaments has shown no significant effects of expansion. It was postu-7 laced that with multiple kinetic exnansions there could be a shock wave reinforcement such that the sequence of explosions or the length of the prima cord should be controlicd to insure that the tubesheet is not overstressed. -The concern was that the

[ shock wave may travel at about the speed of sound through the material, and if adjacent tubes exploded in a manner such that their shock wave reinforces shock waves from other tubes, there could be a condition where the tubesheet is overstressed.

Testing was performed in the steam generator at Mt. Vernon using strain sages ac.! an accelerometer to demonstrate that the coin-

cident explosioes of the maximum number of tubes 6o be expanded 50 - -

w. w. - - - - * -r --- -' -- r 7 w M -P*e -

--.- w y--we.-y + ,aw-.,m-- am--7g-y ye-- w ew- y, -

• P-T-e---.e 4

1 at any one time was acceptable. This was done by exploding 132 charges in the longest row in the generator. A ma imum stress intensity of 95,000 psi at 800HZ was obtained at the strain gage closest to the expanded row. This compares to a static yield strength of 70,000 psi-for the tubesheet material. Since the yield strength.of steel increases markedly at high strain rates (up to twice static yield) and that no residual strain was measured on the strain gauges following expansion it is. con-eluded that no plastic deformation occurred. The maximum stress intensity recorded for the weld between the tubesheet and shell was less than 10,000 psi. The strain gauge on the tube recorded very low values indicating that no significant excitation of the tube bundle occurred. A fatigue analysis has been performed and the tubesheet at- the periphery was found to be . limiting. The analysis was conservative in that it assumes that the principal stresses occur simultaneously and that a11' blasts yielded the smee peak value. The other strain guage locations clearly show that the stress diminishes as the distance from the expansion increases. The results was a maximum fatigue usage of .12.

From this data it was concluded that the use of up to 137

, charges is acceptable and the total number of separate blasts will not present a fatigue problem.

In addition, Foster Wheeler has kinetically expanded over 2000 feedwater heaters and expanded as many as 5000 cubes.in a teater in~one detonation. They report that they have never expertenced any tubesheet overstressing problems and do not believe this is of concern since the plan is to expand only 132 tubes simul-taneously plus any mi'sfires from the previous row up to a total of 137-total tubes for the TMI-1-Steam Generator repair.

The combination of Foster-Wheeler experience and the strain gage data show the explosive expansion process to have no adver'se effect on the steam generator.

C. Corrosion (Proprietary) .

Several concerns have been addressed relative to the sus-captibility of the repair to corrosion. In tubes with through wall cracks, a leak path for primary system water may still exist even after kinetic expansion over the full 17 inches. The breech exposes the carbon steel tubesheet wall to the corrosive effects of a buffered solution of boric acid, i.e., clean reactor coolant.

As described earlier in this report, ID cube cracking due to the corresive effect of sulfur and/or sulfur containing ions has been identified as a probable contributing cause of the TMI-l 4

,e 4

s . . . - ,--m, - - - . - r -

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

h scesa generator problems. The first concern is therefore resi-dual sulfur deposits in crevices above and below the repair seal '

area-, particularly in pockets that may have resulted from cor- ,

l rosion of the tubesheet wall. Such deposit could cause an attack on the Inconel' tubing. Sulfur attack would be prefer-ential to the Inconel rather than the carbon steel tubesheet. It is presumed that the source of sulfur contamination no longer exists and the RCS is essentially sulfur free. As in the ori-ginal design, a crevice will exist below the seal between the CD -

of the Inconel tube and the carbon steel tubesheet wall. The crevice has been flushed to reduce soluble deposits, par-ticularly in the crevice area below the repair seal.

Several years ago model boiler testing was performed at B&W Alliance Research Center in support of a repair to a Florida Power Corp., Crys:a1 River 3 Steam Generator. While these tests dealt with damaged tube to tubesheet welds rather than inter-granular tule fractures, several conclusions applicable to the TMI-1 repair acogram can be drawn. They are:

1. Iron oxide formation, probably the result of tubesheet cor-

. resian in the tube-cubesheet crevice, plugged the leak path of on intentionally damaged tube. Under ambient conditions,

the cxide was less tightly packed and the leak path remained

-slightly open. Under boiler operating conditions of high temperature and pressure, the oxide became more tightly ~

. packed.and appeared to further plug the leak' path.

~

, 2. There was no perceptible evidence of tubesheet hole cor-rosion in the tubesheet bore area of the removed tube section. (The Mark II model boiler was operated fer 6,448 hours0.00519 days <br />0.124 hours <br />7.407407e-4 weeks <br />1.70464e-4 months <br />.) , ,

The leak paths in the model boiler tes'es were generally larger than those noted.at TMI-1. Therefore, the self-sealing mechanism noted above should be enhanced in the TMI-l genera-tors. This sealing mechanism along with the kinetic expansion tends to severely reduce the possibility of free circulation of reactor coolant in the upper crevice area which could produce an aggressively corrosive environment.

Fibre optics inspection of the TMI-l generator tubesheet (video

, tape is available) showed pitting of the tubesheet vall to be very minimal reducing the likelihood of significant pockets of sulfur deposit being present in the crevice above the seal chat could corrosively degrade the seal repair. The packing property of the iron oxide noted in the model boiler test also works co 1Leit the possibility of existence of an aggressively corrosive -

environment due to the presence of small sulfur deposits. ..

)

e

, - ~ , , , c -

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

-1 o, e ~ .

1 1

j-

i d Based on the test results, tubesheet inspection and planned i

1 crevice flushing, extensive corrosion of the tubesheet by resi-dual sulfur deposits or reactor coolant is considered highly ,

unlikely.

t

~

D. Effects of Expansions on Existing Plu_gs The TMI-1 OTSG tubes have been taken ou: of service by four different procedures:

1. Explosively welded plugs - Plugs inserted into the tube and

, explosively welded in position within the tubesheet.

/

2. Welded plugs - Plugs welded to the tube ends or tubesheet at the top of the upper tubesheet.

3. Hydraulically expanded tubes sealed with a welded plug -

tubes that have been immobilized by expansion after a short section of the tube within the tubesheet was removed. The tubes were then taken out of service bf installing welded plugs in the tubest.eet openings.

4. Mechanically rolled plugs.

B&W has evaluated the effect the forces of the kinetic expan-sions may have on the integrity of the first three of these **

plugs and expansions and has concluded the kinetic expansions will not affect their, mechanical integrity or _ leak tightness.

Tests of the kinetic expansion process in steam generator model test blocks with conditions simulating those in the TMI-1 steam generators show that the kinetic expansion does not produce any permanent tubesheet ligament deformation. This leads to the conclusion that plugged tubes adjacent to kinetic expansions will not be altered by changes in the tubesheet ligament, since no permanent change is noted.

During additional tests on an actual CTSG, B&W examined, by dye penetrant tests, the tube-to-tubesheet welds and tubesheet ligaments of the kinetically expanded tubes and the tube-to-tubesheet welds adjacent to expanded tubes and have not seen any de gradation.

Thirdly, the extensive laboratory and field experience of B&W with explosive plugging tubes in operating steam generators indicates that damage to plugt .1 cubes due to detonation of explosives'in adjacent tubes does not occur.

s

/

e w, -, - - - - - -

^

'~

? . .

A Tests were performed on qualification blockscwith rolled plugs 1; in place and explosive expansions of all adjacent tube loca-

.. tjons. Leak rate and axial load te'sts were done to verify that the rolled plugs . continued to meet the acceptance criteria to

• which they were originally qualified for use.

Lastly, a pre-operational post-kinetic expansion pressure test of each generator was performed to verify. the integrity of the primary-to secondary pressure boundary thus previding added assurance that the plugged tubes have not degraded. All ll unacceptably leaking plugs were repaired or replaced.

'l 4 E. Relief of' Preload In some cases, prior to expansion, the degradation of the tube in the area of the seal veld allowed the tube to slip down.

relieving all or part of the preload. The tubes were then expanded.

GPUN examined the effects of changing preload with respect to the limiting transient and accident loads on the tube.

(Ref. 60) The maximum cenpressive load under FSAR accident conditions is a 620 lbs. load associated with a feedwater line break. The approximately 775 lb.. generic design basis compressive load associated with a 100*F/hr heat-up proves to be the LLaiting case. We understand that B&W generic calculations assume a preload of not more than 100 lbs. cension. Thus, for a tube with no preload, 875 lbs. is the m' ximum a compressive design

, basis load expected under normal, transient or accident conditions. However, measurement of the gap left at' the seal weld following relief of preload confirms up to 280 lbs. of preload may have been lost in some peripheral tubes at TMI-1.

. For conservatism, an evaluation was performed of the abili~ty of a tube to withstand 1025 lbs. of compressive load. This is a conservatively large loading since actual heat ups are conducted at rates well below 100*F/hr and the 1025 lb. load value is

' considered conservative for a 100*F/hr heat-up transient.

Buckling is not expected to occur under this load. Tube bowing is limited by the small clearances of the tube support place holes. Further, the applied load is secondary in nature; it is caused by thermal differencial expansion. As the tube begins to bow under the loading, the magnitude of the load is reduced.

Thus, a non-preloaded tube will not be overstressed by the transient load conditions.

e

~

1? .

. e > e J

GPUN has also examined the magnitude of lateral displacement to be expected in a tube loaded compressively to 1025 lbs. The magnitude of the lateral displacement in the-16th span of the .

tube will be the largest since that span is longest, but all of the spans will contribute in relieving the load. Displacement magnitude will also be dependent in part of the initial curva-ture of the tube. Lateral displacement nominally less chan the dimensions of the gap between tubes, even under transient conditions, is expected as a result of the loss of preload. -

However, even if tubes were to contact each other, no problem is expected. During a heatup transient.. flow rates are very lov and the time duration is relatively short; no significant tu'ce vibration or wear would be expected.

l GPUN has also considered the effects of the change in preload on the natural vibration frequency of a tube. A non preloaded tube is expected to have a natural frequency about 15% lower than one preloaded. The effect of this frequency reduction is considered small. EPRI has reported that other operating plant steam generators have variations in tube trequencies of as much as about 10 to 20% within a single steam generator. In addition, test data reported by EPRI shows that another OTSG plant now operating has tube frequencies about 15% lower than those of the tubes at TMI-2. If, ,as expected, the TMI-1 cubes have approxi-mately the same preload as those as TMI-2, then the cubes at the other plant mentioned above have about the same frequency as the ,

unloaded TMI-l tubes, Accordi'ngly, no problemiis expected at

[ TMI-l due to non praloaded tubes.

Based on the above described evaluations, GPUN has concluded that the relief of preload in some of the steam generator tubes does not render them unacceptable for continued use. .

F. Conclusions p

Based on the above evaluati~on and testing, the kinetic expansion process will have no adverse effect on the OTSG structure, tube-f sheet ccerosion, or plugs previously installed. In addition, a cleaning process has been used to remove the residue from the expansion process. In conclusion, overall there are no adverse

, effects from the kinetic expansion process.

l l

l l

e' i . .. .

w e e

• i VII. PLUCOING REPAIR DESCRIPTION

SUMMARY

A. Introduction Those tubes which have defects below the 16" from the primary surface of the upper tubesheet (US) and could not be recovered and returned to service by the above described Kinetic Expansion repair were removed from service by plugging. A defect is defined as any eddy current indication interpreted as greater than or equal to 40% through wall. For conservatism, any less than 40% I.D. indication identified during the 1982 record examination with a large enough circumferential excent to be }

detected on three or more of the e.Eght coils of the absolute probe was treated as a defect for plugging purposes. The limits of eddy current detectability are defined in Section IX. There are a total of 2f 7 cubes in A and d8 tubes in B OTSG that were plugged with either Westinghouse rolled plugs or B&W welded and explosive p.ags prior to kinetic e.<pansion. An additional 658 tubes in A and 199 cubes in B OTSG have been removed from service by plugging since kinetic expansion. 19 tubes in A and L 10 cubes in B steam generator whic.: have been cut and removed for metallurgical examination were plugged with a B&W welded .

tapered cap on the top and an explosive plug at the bottom cabesheet. Defective tubes in some locations were stabilized as indicated in Table VII-1. 680 cubes were stabilized.. The i purpose of tube stabilization is to minimize the risk due to -

propagation of tube defects located in regions with high potential for flow induced vibration resulting in circumferential tube severence and causing damage to adjacent

~

tubes or creating loose parts. The lower tube end was plugged with an explosive plug or a Westinghouse roll plug. The i following sections givs an evaluation of the methods selected for tube plugging and stabilization, and a description of the types of plugs to be used.

B. Tynes of Plugs

1. B&W Welded Taoered Plua, Welded Cao, Stabilizer and B&W Exelosive Plug B&W Welded Tapered Plug is used to plug the bare upper tube-sheet uole for tubes where the tube end has been removed.

B&W's welded cap is used to seal the upper tube end for those tubes which will be plugged and stabilized. Prior to installing the veld cap, the existing tube end is machined off leaving a portion of the tube end and the existing weld protruding above the tubesheet surface. The weld of the nail cap fuses with the e._ sting seal weld, providing the desired pressure boundary.

W

c.w..-,,n L&WExplosivePlugisusedtoilugLTScubeendswhereB&W plugs are used in UTS. Both B&W plug types MK-1 and MK-3 4 have been qualified for OTSG tube plugging and used in the operating B&W units. ,

B&W standard design stabilizer rods are threaded onto the i welded cap to fans a stabilizer assembly of the desired - '

length. The stabilizer is a multi piece assembly of solid ..

rod made of Inconel SB-166. Joint tightness is maintained by crimping the pieces together beyond the threaded sec-tions. The segmeat length is dependent upon the tube bundle location.

B&W welded tapered plugs, velded cap, explosive plugs and -

stablizer rods have been 1.reviously qualified ses References 1 30 threugh 35. ..

l

2. Westinghouse Rolled Plues Westinghouse plugs were designed for a primary pressure of 2500 psi and 650*F and a secondarf pressure of 1050 psi and 600'F. Cracks in the roll transition or the area of the seal weld do not exclude the. use of Westinghouse rolled plug.

The Westinghouse rolled plug is machined from bar stock that has received a thermal heat creatment which has been dedon-strated by laboratorf testing to have improved resistan:e to .--

intergranular attack in caustic and polytheonic acid envi-ronments, compared to treatments at different temperatures and times.

The Westinghouse Roll Plug Qualification Program for TMI-l has been completed for a 5 year life, and results are docu- ... -

mented in Westinghouse Report WCAP-10084 C. Pluering and Stabilization Criteria

?inal tube plugging and stabilization criteria were selected to minimize the possibility of cracks propagating to a size which could part under stress conditions, either plugged or unplugged tubes. Analyses pre-amted in Section IX show that the vibra-tional effects of cro s flow are not expected to contribute to crack growth in an unplugged tube. However, as a precautionary measure, plugged tubes with defects in the area of highest cross flow, between the 15th support place and US+4, are stabilized.

In addition, stabilizers have been used where eddy current in-dicates a crack of significant size in historical defect areas, '

and below the 15th support place where I.D. .xl indications are greater than 2 coils. Measures to monitor for crack propagation due to flow-induced vibration or other means, including leakage -

8

, V ,

y e

e e

~

VII-1 CUTI.INE OF S458C TUSC PLUGGIEC/5Thef t 3 33WG PIM M IdHC4noa8L sos puman Dustsch li S t. Ettfoth autPEtttod Amy Detectable Pluggable Defect .

- Indication Plueeable* Indication

) 44l TW -

, g f y ,,,,,9 y g l $ a 1 33 Costa.3s

-t

. l m = =

W. =

u m

15th SP 35th SP to L5 1 15th SP to 1.5 - 4 15th SP to L5-4 us e 4 to Any tube spar 15th Se to LS-g to 5*I s 3 Call * *** 8*

• 1 * * * * *

• I 05 e 4 u"tstorical Defect Area (Note el LS -4 to -34 by Dotton 4*

isote Il W ees 2 and el . og gg ,gge ,u" E***O*U*

tstori

,,,,g , cal espansion i-Plugl stabillae i

Flug and stabl!!ae 1

Flug and Stabilise to I

Plug Only I

Plug Caly

. I 716g and Stabilise to Dottom of at least In Span tottom of lith SP to set toe of is $P leth SP of Defect ikote 38 Isole 33

. 1. Includes tube sections from bottom of 15th support plate to 4-inches op into bottom of upper tubesheet. .

t 3. Includes tube section from bottom of lith support piste to t ! aches down isora the top of the lower tubesheet.

3. See Figure Val 2 for tubes In Lane / Wedge area.'

. 4. Sal Is FCT probe with 5 absolute colle (

y -

}

e o-monitoring (Section X) and eddy current inspection for wear l (Appendix A) are discussed separately.

The steam generator has been divided into six areas for purposes ~

of dispositioning tubes for plugging and stabilization. Two areas are within the upper tube sheet: between US+8 and US+4, and between US+4 and US+0, where US+0 is the lower face of the upper tubesheet. Tubes where the lowest defect is above US+8 are repaired by kinetic expansion. Tubing between the tube- _

sheets has been divided into three areas: tubes between the 15th support place and US+4, tubes between the 15th support plate and LS-4 in the lane / wedge area, and tubes between the 15th support place and LS-4 outside the lane / wedge area. The last region considered was that within the lower tubesheet.

Plugging and stabilization criteria are discussed in detail in Reference 25, and summarized below:

03 i

1. Tubes with Defects Between US+4 and US+E" ra Tubes with a defect between US+5 and US+8 may not be effec-l tively repaired by the 22" Kinetic Expanaion. A qualified length of 6" expansion is required to insure a leak-tight, load-carrying joint to assure the OTSG integrity is retained _;

under the most adverse conditions during operation. There-fore, those tubes with defects between US+4" and.+8" were both kinetically expanded to 22" and plugged. Even if the l- existing crack would propagate in the future-and sever at US+5", testing-(IWf. 23) indicates that the expansion joint below the severan~ce would provide enough engagement to main-tain the preload in the tube and carry the loads associated with the most severe transient during normal operation.

Thus tubes with defects in this area are not expected to wear adjacent tubes due en dynamic instability in high' ~

crossflow. There fott. . it was concluded that these tubes need not be stabilized (unless stabilization is required by

, defects of greater than 3 coils between US+4" and US+5").

2. Tubes with Defects Between US+4 and US+0 Even with a 22" expansion, these tubes would not have the 3' kinetic expansion joint below the defect to maintain preload and assure that the tube will not slip under the most severe transient during normal operation (i.e. , 100*F/hr.' cool-down). If the tube ratcheced down during operation, the l potential exists for the tube becoming dynamically unstable or buckle. Therefore, after receiving a 22" expansion, a tube with a defect in this area was plugged and stabilized through the 14th tube suppert place.

l i l

l

?

-m -

3. Tubes with Defects Between 15th Suovert Place to US+0 1

j For conservatism, any eddy current indication identified during the 1982 technical specification record inspection in '

this area, regardless of type, through-wall measurement or .i circumferential extent, was created as a defect. Tubes with l

" defects in. this ' area received a '.7" expansion and were i

stabilized 'to the 14th tube support place, unless a lower '

defect was located in an ares requiring a longer stabilizer. .

This criterion. covers the high cross flow tube span in the top of the steam generator.. Even if a tube were postulated to become severed at this elevation, it could not get free and damage adjacent tubes.

During ~the post-repair eddy current baseline inspection, several less than 40% through wall indications were ident ified in this area but not subsequently plugged or stabilized. All of these less than 40% through wall calls were of. such low voltage and such uncertainty in phase angle that it could not be concluded with confidence whether they represented intergranular.actack or surface anomalies. -It was concluded that they are not of a size to warrant plug-ging. Details of the post repair eddy current baseline

" inspection results are in Appendix A.

.4 Tubes with Defects from 15th Suecore Plate to LS-4 in Lane Wedse Area ,

B&W plants have a history of corrosion and vibration prob-

-lens in the areas of the untubed inspection lane. As a precautionary meaJura, an area of potential problems has I' been defined one row on either side of the lane, widening to

[

a wedge shape as the lane nears the periphery. For tubes in this area, any I.D. eddy current indication -regardless of type, through-wall measurement, or circumferencial extant, was created as a defect. These tubes receivef a 17" expan-sion, then were plugged and stabilized through the 14th support place unless other criteria required stabilization through the defect in the lower spans.

5. Tuoes with Defects Betwuen the 15th Sueoort Plate and LS-4 In the tube span between the 15th support plate and LS-4 outside the lane / wedge area defects were divided into two groups: 1) ECT indications greater than or equal to 40%

f through wall with 8xl indication greater than 2 coils, and 1

2) ECT indications greater than or equal to 40% chrough wall l g with 8x1 indication less than or equal to 2 coils.

~

m o g

• e

- , .wy '

g

. o 8 .* n .

g e S

i i 3 5 3s 2

w

. 18 "

3.2 8i d[ 8 O m i-

!l.

oo 8

o s

& a u m y, < E 11 e s

m 8g 8 s  :

8$ .: W  !

.j j/

o e _

i D

=

6 o e 8c. h i*

o s

+8 .

hS[ ~

h._

O

. o .. a .

After a 17" expansion, tubes with defects greater than or equal to 40% chrough wall and 8x1 greater than 2 coils were stabilized through the span with the lowest defect for that tube. Tubes'with greater than 40% through wall defects and -

8x1 indications on 2 coils or fewer were expanded to 17" or 22" as appropriate and plugged with the Westinghouse rolled plug.

This criterion provides a means for determining the tube are length excent of a defect in order to decide if the cube should be stabilized. The tube would be stabilized at least within the tube span containing the ECT indication if there is any substantial size or are length involved in the ECT indication. If an ECT indication is seen on less chan three coils on the 8xl ECT probe it means that the are length of the degraded area of the tube of a maximum of abouc 0.41 inch long at the inside diameter of the tube. Because of the " thumbnail" shape of the inside diameter cracka found at TMI-I this means that the average are length of the largest two-coil ECT crack would be about 0.26 inch. This size crack is acceptable without stabilizing and is not expected to propagate to failure by mechanical means during opera-tion (Ref. 25).

This criteria for stabilizing, based on are length as measured by the 8:cl probe is not invoked for ECT indications less than 40 percent through wall because such an indication is too small to fail the tube. Even if a tube had a 360*

1 indication, the dube would not fail with less than 40 percent penetration (Ref. 25).

l

6. Tubes with Defects in the Lower Tubesheet Belov LS-6 Tubes with' pluggable defects in the . lower 20" of the LTS vere removed from service using a %estinghouse rolled plug or an explosive plug. A welded plug was used in lieu of mechanical rolled plugs if the defect was in the rolled area.

D. Post Resair Testing The following post repair tests will be performed to verify plug integrity.

1. 150 psig bubble case

2. 150 psig drip test

3. 1400 psid operational leak .

Details c post repair testing can be found in Appendix A. The cold tests described as items 1 and 2 above have been com-placed. All plugs which were identified as leaking unacceptably ..

were repaired or replaced.

.. 60 -

l 9

o .- e .

e e a

E. onclusions The CISG Tube Plugging Plan has restored the pressure boundary integrity of the steam generators by removing defective tubes

• from service for those tubes which have defects below the region available for expansion repair.

The plugs used have been previously qualified'for use. Tubes have been stabilized which have large defects in areas of high crossflow or areas of historic problems wit,h vibration or cor-rosion at BW plants. Thus. plugged tubes with the highest potential for tube severence in the future are prevented from wearing adjacent tubs.s.

+

l l

l l

4

.j i

l

o: P VIII. . COMPARISON OF TUBE PLUGGING WITH DESIGN BASIS A. Introduction- .

This section describes the results of analyses performed to determine if the steam generators could be safely operated with up to 1500 tubes plugged. The first part of this section re-views the operational consideration of operating with tubes removed from service for: reduction in total ficw and margin to .

departure from nucleate boiling effects of asymetric flow dis-tribution, effects on flow coastdown rate, effects on steam '

generator mass inventory and capability of natural circulation.

The second part reviews the effects of removing tubes from ser-vice on smell and large break loss of coolant accidents as well.

as all other accidents and transients. anal fzed in the FSAR. The third part cemiders the effects of plugging on moisture carry-These antlyses have concluded that the margins of safety over.

as defined f.n the Technical Specification will not be reduced by operating the TMI-1 steam generators with up to 1500 cubes re-moved from service.

3. Ooerational Performance

1. RC Flow Rate and Martin 'to Minimum DNBR-

' The calculated RC flow rate for all four RC pumps operac'ing as a function of an equal number of tubes plugged in each '

steam generator is shown in Figure VIII-1. Generally, a reduction in tubes available for RC flow will cause the tube bundle pressure drop to increase. Since the remaining sys-tem pressure losses are about four times greater than the tube bundle pressure losses, only a slight reduction in total RC flow race will result. The total RC core flow for

• 1500 plugged tubes will be the same as the symmetric case, i.e. 750 in each steam generator. From the figure, the reduction in total RC flow will be from 109% of design to 108.2%, a change of 0.8%. (Reference 42)

In order to determine the impact on the existing steady state Departure from Nucleate Boiling Ratio (DNBR) resulting from the RCS flow reduction at steady state, a study was performed to determine the minimum RCS flow rate required to maintain the existing.DNB ratio for the TMI-l licensed power level of 2535 MWt. The existing DNB steady state ratio of 2.0123 was determined at a conservative power level of 2568 MWt and an RCS flow rate of 106.5% of design flow.

~

+

I e

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

~ . .

Epp?"' . ,

The methodology for the analysis was to calculate the hot bundle flow by using a CRATA (Raference 36) core model which took heat balance input from the CIPP code (Reference 37).

By using the hot bundle flow in the computer code TEMP -

(Reference 33), the Minimum DNBR (MDNBR) was calculated for the hot sub channel from the BAW-2 correlation (a cor-relacion.

The results indicate that DNBR value of 2.0123 can be main-tained with an RCS flow rate of 104% of design and a power of 2535 MWt as compared with the original 106.5% flow and 2563 MWT. The minimum calculated RCS flow rate at TMI-l has been 109.5% of design flow. The maximum error on this value is 1.5%. This results in a mininum available flow of 108%

of design. This will be reduced to 107.2% af ter the tubes are plugged. This is substantially greater than the design basis flow race of 106.5% which would be required to main-tain the design basis steady state DNBR value of 2.0123 at 2568 MWt. For the TMI-l power level of 2535 MWT, the 104%

design flow requirement to maintain the same DNBR provides for an even greater margin. It can thus be very conser-vatively concluded that the plant design basis flow race considerations will be preserved with 1500 cubes plugged.

2. Asymmetric RC Looo Flow Dist?ibution The final plugging pattern will be about 6* of the tubes in the A Steam Generator and about 2% of the tubes in the B Steam Generator.~ In order to investigate the asymmetric effect of the RC loop flowrates, an evaluation of an exag-gerated plugging pattern of 1500 cubes in one of the TMI-1 steam generators has been performed. The Loop A flowrate will be approximately 2-1/2% smaller than Loop B. -

1 Field data at TMI-l during the last cycle has shown that the A loop has typically about 3* more flow than the B loop.

The result of mare plugging in the A Steam Generator will chus be a somewhat more balanced loop flow distribution. '

The new flow difference is expected to be approximately 0.5%.

3. RC Flow Coastdown Rate With a significant number of tubes being plugged, the resis-tance factor for the RC flow passing through the CTSG will be increased. This increased resistance may change the flow distribution if one RC pump is tripped while the other pump in the coolant loop is maintained in operation. The com- )

bined core flow during the coastdown may also be different.

l l

~

y ,. 'o .

Further, the minimum margin to DNBR during a loss of flow event is known to be dependent on pump coastdown races. To address chose issues the following analysis was done.

A computer analysis has.been conducted-using the B&W code

" PUMP" (Raference 39) for the TMI-l type reactor coolant

~

system's flow coastdown curves with zero and 1500 cubes plugged.in the 1 Steam Generator. The FSAR analyses served B as the base case for the four pump coastdosn transient.

. Results of the analyses with 1500 tubes pliigged in one steam

. generator show that the FSAR coastdown rate is still bounding.

Figure VIII-2 sununarizes the data obtained from the four .

pump coastdown~ transient performed with the TMI-1 version of PUMP code and compares this data with the FSAR analyzed flow coastdown. -This comparison shows that the flow even with 1500 plugged tubes starts at a higher level chan the flo.w assumed in the FSAR analysis and coasts down at approxi-'

mately the same rate since the flow is at all times greater

. than that assumed'in the ThAR the minimum margin to DNB will not be changed. -

, 4 Steam Generator Water Inventorv and Ooerating Level Indication The water inventory in the steam generator, will increase by

, 'a small amount due to the decrease in average quality in the plugged section. RETRAN-02 (Reference 43) and TRANSG (Reference 44), which are both one dimensional transient

thermal hydraulic computer programs with slip option, have been used. The inventory increase was calculated to be 5*.

, or less, which is less than 2000 pounds with 1500 cubes plugged.

Total secondary side flow will increase onif slightly with

-decreased steam outlet temperature. This would cand to cause a slight increase in pressure drop. This increase will be offset by the reduction in average quality (increase in density). The net effect should be little or no increase in the startup level.

5. canabiliev of Natural circulacion The impact of steam generator tube plugging on the capability of stable transition to natural circulation was examined by using the B&W computer program AUX (Reference 49). Synenetric plugging of 1500 cubes in each side was

\

\

~

i f

+

+ e es.

--, + ---s , .,--,,----,,-w- - , - - , . - +,,-,-,,-a-- -

v--,,, -, , - ,-- , . - . , ,- - - , - - - , - ~ ~ - - - - - - , - - -- --m,-

a . * ..

W FIGURE Vill.1 REDUCTION IN RC FLOW RATE VS7 NUMBER OF TUBES PLUGGED PER STEAM GENERATOR 4

110

. 108 -

tt 3

2 u.

e

_cn .

'g 106 -

c -

o I 104 -

2

a. .

O

m.

• y I 4C

>=

0 102 -

100 I ' ' ' ' ' '

500 1000 1500 2000 2500 3000 3500 4000 Number of Tubes Plugged in OTSG D

W O

,,,,,,,,--_---w -em** -

\

e .

h *

• 6 1

NWI WOW NiT l

s e

b u

T d

e g

g ODN T u D

T SW l

P S AO E 0 AODCT D 0 8

)

s O U 6 d

CPS A T 1 ~

n WM OP U OD CE y o c

e L G R S F RWG A (

b F UOU S e OOLL F FP

. F i m

T N RP0 O

S AM0 I

SU16 RF P AR PS U

ME V O Ot F l O CuS .

CV 0 0 0 O 0 3 2 9 6 1

g ;;eO % # 2U[ kl WE

(.

t 9

b

- -~ ~

i

~

V

~

FIGURE Vill- 3 , ,

e O

E.2 e >3 9lI e ~

aC O

>oO 2

O*L e2 C

C

__. 1 N,

l- _

~

{ C

=

C 2 I h

o 4

a -

3 0 -

O C C

.J

[} m

.gW -

32

>~p 4 -

g t/5 C *

~ o 2

C >3 I -8 . uS

• cc je EN .

~

E

e. .=

cm -

3 a0 0 ~

L .J l W4 -

.m> 0 3C C

>> \

u. -

O

> ~

U W -

W

u. - I W C  !

r= C '

.:. N 7

3 t =

. _g g g l 6 4 I I i I I I I $

4 6 4 4 4

• 3 C C C C C C C C C o w w N O e -

e e

( cas/q1 thold DJ00 letal i

l

I- _ _ . . -

4 e 4

• i.

FIGURE Vill-4 Ee

=

E.E

a aA 4: .

o=$

e o o C

. r n 2 o

$ ) C

> C a

o o

E o

.J z o o *

<w z 2' .

2P $

O us c 0> - o z- e .

cA $ $

o0 -

o so aa AW W wo a=

3> *

> l L _

c C v u

w W

W w

T

~

, C C

Y N i

I I l C

o o . e o o o o # # M *

• w w e e e m .

(3 sooJGoc) 1041 O

e

e . _4 .

4 assumed to be the bounding case. Figures VIII-3 and VIII-4 compare the' analysis results of 1500 cubes plugged to no tubes plugged.- At about 500 seconds after the reactor coolant pumps trip, the stable natural circulation flow with '

1500 cubes plugged is about 8% less than that with no tubes plugged. . At no time is natural circulation lost and the reactor coolant system' remains subcooled. The subcooling margin for 1500 tubes plugged is only about 2*F less chan the case without plugging (about 90*F). Therefore these analyses have shown that natural circulation is still an effective method for decay heat removal.

C. Accihmt and Transient Performance

1. IDCA Analyses The potential effects of SG tube plugging on generic Large Break LOCA (L3LOCA) and Small Break LOCA (SBLOCA) analyses (Reference 40) with 1500 cubes plugged has been exawined.

' With about 6% of the tubes in the A Steam Generator and about 2% tubes in the B Steam Generator being plugged, the generic (2772 MWt) LOCA analyses for S&W 177FA Lowered Loop Plants remain valid for TMI-l' with continued operation"at core-power levels up to the licensed 2535 MWe at the existing LOCA limits. An overview of this examination is -

provided below.

a. SBLOCA Concerns The evaluation models used in the existing SBLOCA analyses (Reference 40) assume equilibrium conditions within the control volumes used to model the SG secon-dary side. For this reason, the localized cooling gf-facts of EFW spray on particular tubes, an'd the effects -!

on this cooling if particular tubes are deactivated, cannot be accurately predicted with these models.

In the application of the revised S3LOCA evaluation model (Reference 41), it was assumed that:

l l

1. The percentage reduction in the number of peripheral tubes removed from service will degrade the EFW l l

spray cooling heat removal capability in a 'L:1 '

relationship.

2. The degradation in heat removal capability from EFW spray cooling translates directly to a reduction in depressurization rate by a 1:1 r d ationship.

l l

6 4 -

e ,. , , , . -,,.-mw-,~ ~ -,.--+--e em+..., - ,e-ne- e--,,- -,,a-, - --,r-e,,-.-.,~-,.,,,---,,,-r-.=,sg g , -,++ , ,,nn

e _ ,. .,

' ' " ~

p- -

u-w J- .a .. a'- .

y ;+; =ws a t

4sJ l In reality. .these relationships are expected to be con-

servative. Since if EFW spray impacts s' deactivated

• /' tube, it will not be heated and/or-' flashed immediately de, ' but will either: -

l. Be redirected onto adjacent active cubes, or;

2. Flow inward into the tube bundle, providing cooling

to active interior tubes, or;

3. Fall into the naturated stess or saturated water i' i region, resulting in increased cooling within these e regions and/or an increase in the fill taca to. the -

4 appropriate level setpoint. '

t A more detailed discussion of EFW spray effectiveness and the TMI-l response to SBLOCA with plugged tubes is given in Ref. 59.

Therefore the water is available in the staa.m generator and will result in greater EFW spray cooling and higher depressurization rate than predicted by the analyses.

In this evsluation (Reference 42), two break cases were considered. The first is the worst case with respect to peak clad t.emperature for a small break LOCA, identified as approximately a 0.07 fg2 cold leg break. The.

second, belongs to the category of breaks in which' SG .

heat removal is needed to help depressurize the RCS. '

The 0.01-ft2 break was analyzed because this was the largest break size which would result in RCS repres-surization.

Plugging 1500 SG tubes was used as a upper cound which represents a deactivation of approximately 5% of TMI-l's total tubes. Also, because substantial tube plugging will be done in the peripheral SG tubes regions, it is estimated that about 18% of TMI-l's total peripheral tubes will be deactivated. -

Worse Case 0.07 ft Cold Les Break with One HPI Train For this break size, the primary system pressure

' decreases below 1000 psi (approximately the secondary side pressure) at about 300 seconds. After this time, SC heat removal is no . longer possible, and the secondary ,;

i side becomes a heat source for the primary system. Core uncovery begins at about 1350 seconds and ends at about 1750 seconds. The maximum time that SG (and EFW spray) cooling can be of benefit during the accident is the '

,. first 300 seconds. This is very short when compared with the time to begin core uncovery.

_ _ e --

u k ..s -. ~-

e

q . .

4 The plugging of 1500 SG cubes will result in a reduction of the initial RCS liquid inventory by about 200 ft3 This results in the core being uncovered about 3 seconds

.{ earlier and in approximately a 10F increase in peak -

cladding temperature (to about 1100*F). This will have minimal impact on the outcome of this accident. It should also be noted that the generic analyses show that, for a 0.07 fg2 break with 2 KPI trains, the core does not uncover and temperature remains below 700*F.

., 0.01 ft 2Cold Lee Break The 0.01,ft 2cold leg break case was evaluated using j

the revised SBLOCA model. All of the original analysis assumptions were preserved, including a,20 minute operator delay in initiating emergency feedwater. Two cases were analyzed considering the effects of tube

! plugging on the decrease in RCS depressurization rate which cocid increase the time to initiate ESFAS and the rate of heat transfer in the boiler condenser mode. It was found that:

With a 1600 psig low RCS pressure ESFAS setpoint, the 0.01 fg2 case will result in ESFAS actuation regard-less of the reduced peripheral and internal SG heat ~

removal caused by the plugging of 1500 SG tubes. Before ESFAS actuation, the RCS was subcooled and either forced or natural circulation existed. Consequently, SG heat removal was found to take place throughout the entire 'SG cube region, not predominantly in the peripheral regions. Therefore, the SG heat removal race is not reduced by more than 52 during the period prior to, ESFAS, and this will delay only slightly the activation -

of ESFAS.

The 0.01 ft2 break case will cause the RCS to enter -

che boilar-condenser (3-C) mode. In this mode, EFW spray cooling of the peripheral tubes is an important factor in the RCS depressurization. Thus, peripheral SG tube plugging could have a more significant effect on this cooling mode. The evaluation showed that, with 182 of all peripheral tubes plugged, sufficient steam generator EFW spray heat removal capability remains so that the rate of RCS depressurization is reduced by only about 12*. Even with this reduction, calculation's with all other original analysis assumptions unchanged show a minimum of five feet of coolant remains above the core throughout the event. Since the 0.01 ft2 break is 9

~

.o'

- . a. .

approximately the largest break which would-result in RCS repressurization, the plugging of 1500 SG tubes is expected to have only minimal effect on SBLOCA transients.

In suemary, these small break size LOCA analysis show that for the previously limiting case of 0.07 ft2 cold leg break with only one HPI train availabis, peak clad temperature increased by only _10*F to about 1100*7. For the 0.01 fg2 cold leg break, the slight delay in ESFAS actuation and the reduced area for EFW cooling have an insignificant effect on the outcome of the transient since a minimum of five feet of coolant remains above

.c the core for both the plugged -tube and unplugged tube cases. Therefore the generic LOCA analyses remain valid for TMI-l even with a small reduction in SG heat removal caused by tube plugging.

b. LELOCA

}-

~

The important parameters for ,che LBLOCA which are

, effected by plugging of tubes are the initial flow and flow coastdown. The effects of the reductien in coolant volume associated with 1500 plugged tubes (200 ft2) are negligible for this event. The plugging.of 1500 SG tubes at TMI-1 will reduce total system flow. Howevet, the reduced flow will still be greater than the flowrate used in the generic LBLOCA analyses. This, coupled with TMI-l's lower core power (2535 MWe vs the generic 2772 MWC) provides margin in initial conditions for TMI-1, relative to the generic analysis. During the early portion of a LBLOCA transient when the reactor coolant pumps are coasting down, the analyzed system flow rate (see VIII.B.1.3) with additional resistance due to plugging will be greater than assumed designed flow race with the case of no plugging.

The reduction of 200 ft3 of primary coolant volume will have licele impact to the consequence of LBLOCA.

Since the OTSG's are unevenly plugged with more tubes plugged in A chan B. If a cold leg break occurs in' the A side, the reduction of RC fluid is part of that blown out of the break, and there will be no impact to the result at all. However, a break in the B side will result in slightly less total fluid passing through the core during the blowdown period. For about 11,000 ft3 total fluid loss within approximately 24 seconds, the '

reduction of 200 ft3 will correspond to 0.4 second

~

e

-me 7 - am-- ye-ige-p-3-gy*gv*F-- y-W*T--"Y--- r"""" ' " ' ' - "

e '

x .

s '. , 'O .

9

. shorter blowdown time and thus a slightly earlier fuel heatup between blowdown and refill. This difference is minimal and therefore, the resulting peak cladding tem -

perature that occurs during this developed reflood stage -

should not be changed. Also, veut flow is conserva-tively neglected during the refill /reflooding phases of LBLOCA analyses for the 177FA Lowered Loop plants. Tube plugging will therefore have no impact on core flooding rates.

2. FSAR Analyses of Other Transients An assessment of the impact of the plugged steam generator tubes on the ability of the NSSS to safely respond to FSAR transient conditions has been performed. The plant is ex-pec.ted to be operated with up to a total of 1500 SG tubes plugged for both steam generators at the licensed rated power level. Each event in the TMI-1 FSAR will be addressed in light of the expected impact of steam generator tube alugging on assumptions used to produce the current FSAR tslysis.

a. Uncomeensated Oeerating Reactivity Changes This event is core bu nup related and is norm, ally com-pensated for by Integrated Control System action over

• the life of the fuel cycle. Steam generator plugging

.w ill not affect the core kinet.ics and thus have no impact on the event.

b. Startue Accident /CRA Withdrawal at Power The CRA Withdrawal from Startup conditions and at Iscwer results in primary system overpressurization. The FSAR prediction of RC pressure and peak thermal power is based on the conservative assumption that all heat pro-duced in the core remains in the primary system, i.e.,

no steam generator heat transfer. The tube plugging will result in a 200 ft3 volume reduction of the primary coolanc ( 2*.). At peak thermal power the reactor coolant pressure increase was 118 psi in the FSAR. With the small volume reduction and consequently slightly higher heacup race the peak pressure may in-crease slightly but will remain well below the 2750 psig limit. Therefore, the FSAR analyses remain bounding with respect to the acceptance criteria on thermal power and system pressure.

.m y ..-- , . , - , . , -

, - - -- . - , - - - - _ ,,- e-- -. - ,. .,#,, , , . - - . - -

4 4

'o .

9 .

c. Moderator Dilution Accident The moderator dilution event is a relatively slow over-pressurization transient due to increased reactivity by ~~

baron dilution.- Change in steam generator plugging will not affect the basic assumptions of this analysis and therefore the FSAR remains bounding.

d. Cold Water Accident (Pumo Startun)

The pump treartup event is ,a small overcooling transient due to an increase in flow from an idle loop. The analysis performed for Section 8.5.3 demonstrated that the pump characteristic curves differences are negli- .

gible for the unplugged and the liCC plugged tube cases. . The reactivscy change will cause a power and RCS pressure increase. The transient will be terminated by either the high reactor pressure trip or the power / flow trip. The FSAR remains bounding.

e. Loss of Coolant Fles See Section VIII C.l.

p- f. Stuck /Drooned Rod Event ,

The FSAR analysis is bounding since SG heat transfer and RCS flow do not effect this event. -

3 Loss of Electric Power The unit will trip on loss of electric power. With the

, loss of the reactor coolant pumps, natural circulation in the primary loop and heat removal by the emergency feedwater system are required. The impact of tube

, plugging on the ability of natural circulation is desce-strated in Section B.5..

h. Steen Line Failure The licensing basis for TMI-l is the double-ended rup-ture. This FSAR analysis is ba' sed on a very conserva-tive prediction of SG secondary inventory. Operation with plugged tubes results in a secondary inventory greater than operation without plugged tubes but it is not as ' great as that considersd for the FSAR analysis.

Secondary inventory is one of the parameters that deter-r 1e the safety considerations of return to criticality, 70 -

e e e,-e -- , , ~ - - -,. w, , , _ > , , - , - ---w---gn,--n.e-,,,w,,,,,--,w-,-,4,,- e,+w- m,-~w--y,m--.--,- , . - .- . , ,- . , , - - . , - -,,,--g-,n, - , , - , , , - - . - - - - + - , - - . - - , , , - - + - - - , ,w

o- ., a. .

and reactor building pressure. The calculated water increase with 1500 plugged cubes is less than 5%, which will result in a maximum steam generator inventory of 42,000 lb. per steam generator in contrast to the FSAR ,

analysis assumption using a steam generator inventory of 55,000 lb per steam generator. It is therefore con-cluded that the FSAR case remains bounding.

i. Steam Generator Tube Failure ,

The sesam generator tube rupture accident is analyzed s - assuming a 435 gym leak from a completely severed OTSG

' tube. The RCS is depressurized and isolated at 34 minutes, at which time leakage from the RCS is assumed

to stop. The reduced RC flow as a result of the plugged Cubes is greater than the RCS flow assumed for this cooldown rate (even the 100% design flow is not required to cool the RCS). Similarly, more than enough OTSG heat ,

transfer area is available to cool the RCS.

• Offsite dose from the Tube Rupture Event will not be affected by plugging 1500 cubes because neither the time

-l required to isolate the OTSG nor the leak rate from the broken tube is affected by the tube plugging.

j. ' Fuel Randline Accident

• f This accident,is assumed to occur during outage a refueling outage while the reactor is shut down. Change 4

in s aam generator plugging pattern has no impact to the assumptions.

4

k. Rod Election Accident

.,i Fast reactivity excursions are not influenced by SG heat removal. The event is an adiabatic heacup. The FSAR analysis remains bounding.

1. Maxi:num Hvoothetical Accident The analysis assumed that a given amount of radio-activity has been released following core exposure and studied the effectiveness of the building spray system and Engineering Safeguard systems leakage on to the environment. The steam generators are not related to the scenario and thus have no~ impact on the conclusion.

l I

1 I

o. '

m. Weste Gas Tank Ruoture The Wasta Gas Tank is located in the Auxiliary Building and the analy is of its rupture is not related to the '

steam generator's function. The event is thus un-affected.

n. Loss of Main Feedwater/Feedwater Line Break A loss of feedwater accident is an event resulting in primary system heatup, increased pressurizar level and 3 pressure, and reactor trip either by anticipatory func-tion (loss of main feedwater pumps) or high.RCS pres-sure. The long term cooling relies on emergency feed-water heat removal through the steam generators. With the plugging of 1500 cubes in the steam generators, the intitial heacup rate will be slightly faster. However, the anticipatory trip on high pressure will shut the reactor down and reduce the heat input to its decay heat level regardless of the minor difference in heacup rate. Emergency feedwater has the flow capability of removing decay heat up to about 7 percent power. This is greater than the decay heat at any time af ter shut-down. In the SBLOCA analysis using the revised LOCA model it was demonstrated that heat transfer. rate is not significantly changed with the amount of plugged tubes.

'Therefore the FSAR analysis of the loss of feedwater accident remains valid.

o. Steam Generator Overfill Steam generator overfill was analyzed as a part of the TMI-l Restart Report, (Reference 50). This analysi's identified that it takes at least 10 to 17 minutes for auxiliary feedwater to overfill to the top of the steam

~

generator's. shroud. Operators are instructed to isolate the feedwater flow path as soon as the OTSG water level reaches 82.5% on the operating range (high level alarm) and to trip or throttle feedwater pumps if the level reaches 90%.

. The impact of up to 1500 plugged tubes in one steam generator will be about 5% inventory increase at about the same level indication. This has been shown in Section B.4. This implies a reduction of the over-filling time by about 60 to 100 seconda. The time for the operator to respond to the high level alarm will be shortened. Moreover since there ~.. still sufficient l

~

, . 3 .e- .

$ .e p

• time and unambiguous symptoms ava'ilable for the opera-tors, their prompt response is expected and thus the overfill would be corrected.- In addition, a stress analysis has-been performed on the consequences of -

flooding the 14I Unit 1 Main Steam line. Ihe results of deadweight internal pressure and thermal expansion analysis show that the mein steam piping can withstand these affects. Therefore' operating the steam generators with 1500 plugged tubes will not present a safety con-cern with respect to steam generator overfill.

D. Moisture Carry-over Considerations Evaluations were performed to determine whether the plugging pattern would allow moisture to be carried up with the steam, i^

causing potential for erosion of components or steam lines.

Calculations of steam conditions at the entrance to the turbine show approximately 33*F of superheat, Indicating no moisture problems in the bulk steam. Evaluations were also done, to determine the potential.for local effe:ts in areas of high plugging before mixing equalizes temperature. Calculations include multidimensional thermal hydraulic simulation of the OTSC with plugged tubes using the THEDA II Code. Results indi-cate that moisture carryover from the clustered plugging of tubes should not be a problem. The effects of the aspirator bleed ports and the geometric design of the 15th tube support place (outermost holes are not broached) work together to sub-stantially reduce moisture carryover. An ISI program will be

- implemented to further assure the integrity of peripheral tubes and downstream components and piping. Peripheral tubes in areas of high plugging are included in the post repair eddy current program. A supplemental ISI program of steam system fittings will be implemented. Both programs are described in Appendix A.

. E. Conclusions Evaluation has shown that up to 1500 plugged cubes per steam generator have no adverse effects on performance of the steam generators. The reductions in flow and heat transfer are not large enough to affect the licensing basis analyses for transients or accidents. In addition, moisture carryover is not expected to cause erosion problems, but monitoring of the steam system fittings will identify erosion should it occur.

e t

a 4

f e

e t

6

,- s-v--- .cwS. .- ,-----,--e.w.--.-"vs - , - , . , , , , - - r... - . - ,-v-+.---- - ,e,,w, m ,.,,-*.,,ey ,,---...-.-....-w3-,. " - '*r-=w--'-w- 7 - ' - r, 7-*

-a o

• l l

II. UNREPAIRED PORTION OF TUBES Reference 2 and Sections IV through VI of this report discusses '

tubes in the area of the kinetic expansion, plugged tubes and how they meet the design basis. This section discusses how tubes in the remainder of the steam generator meet the design basis. The rationale for resuming operation with the existing steam generator tubing is based on these facts: .

1. Corrosion tests indicate that the cracking mechanism has been arrested and will not reactivate in low sulfur primary coolant water chemistry. If the cracking does reactivate due to an unknown mechanism at operating temperatures or during heacup and cooldown cycles, it is anticipated that the precritical testing sequence would allow sufficient time .

for defects to propagate through wall to a size that would allow leakage to be detected.

2. . Analyses have demonstrated that cracks below a minimum range of length and through wall chickness will not propagate to failure by combinations of flow induced vibration, thermal cycles and aschanical loading. Analyses have also cal-culated a minimum size below which a crack will not become unstable due to plastic tearing or ligament necking during a MSLB. The range of crack sizes above this was de.tectable by the ECT inspection program that was used to inspect the ~

steam generators, and were removed or plugged.

3. Any through wall defects that are large enough to propagate unstably and are not picked up during the 100% ECT inspec-tion because of equipment or analyst error will be detected by leakage monitoring programs during the test program.,

A. Nes Damase Not Occurrine The following paragraphs address the subject of new damage not.

occurring in the steam generators. Short term corrosion testing program has provided evidence that the crack mechanism is ar-rested and the long term corrosion tests will act as an antici-patory program for crack initiation. An eddy current flaw growth program has shown that cracks are not initiating or pro-pagating. Defect indications which are less than 40 through wall and less than 90* in circumferential excent will be lef t in service and monitored for crack propagation. The precritical testing program will detect reinitiation of a rapid attack through leakage monitoring. A slewer mec' hanism will be detected at the scheduled eddy current examinations or by the leakage monitoring program.

e'

~

. The corrosion casting program results are described in Section III. Tests on actual and archive Steam Generator tubing to date have established: .

(a) Cracking will not occur unless an active reduced species of sulfur is present and cracks in SG tubing will not propagate in the present chemical environment; .

(b) Sulfur induced cracking requires an oxidizing potential ~

which does not exist under normal hot operating conditions; (c) Lithium hydroxide is an effective inhibitor of the cracking -

mechanism.

Tubing which has undergone the repair and chemical cleaning process has also been tested. Accelerated tests performed on this tubing under severe chemical environments has not produced any cracking. To provide assurance that the mechanism has no long term time dependency, a long term corrosion test program has been initiated to provide an anticipatory assessment of tube performance under actual steam generator operating conditions.

This program will lead plant operation 'and will run for approxi-mately one year.

Since identification of the steam generator problem, cracks in' the generators have been monitored for growth. Eddy current testing of thout 100 cubes in each steam generator was conducted l on a repetitive basis to attempt to ascertain if .the inter-granular attack mechanism was continuing to damage the OTSC tubing during continued dry primary side lay up conditions. The sample selected for this monitoring assumed half tube sheet symmetry and included tubes with no defects, tubes with a variety of defect indications and cubes in periphery and ina l

terior areas of the bundle previously identified as high and low l defect rate areas, respectively. The method used involved a relative comparison of the low gain .510 standard differential  !

prebe eddy current responses from seven repetitive examinations of the same tube population over a period of time extending from j December 1981 through July 1982. The eddy current data was evaluated and compared with previous data for each tube to determine if reported variances from test to test were related

• to variability in the physical repeatability of analysis of threshold-level defects or to the appearance of fresh defects l i

grown in the interim period between tests. In July 1982, a "nev baseline" condition was established with both the .510" std.

sain technique and the .540 high gain technique performed con-secutively (within 3 days of each other). -l l

l l

~

}

l

"-- *-ev - - , . - ,, m, , , , _ _ _ _ _ , ,

-o -6

• The consistent pactorn of the test comparisons indicated that significant growth.of new intergranular cracks was not detected.

Some variability in repeatability of recorded results was ob-served, however careful review and comparison with previous data

• established these as expected variances due to such things as

" probe motion". noise levels, and previous indications inad-

. -vertently not recorded.

The comparison of the July 1982 .540" high gain technique data _

to the July 1982 510" standard gain technique data run 3 days apart showed a 94% (188 of 201 cubes) agreement with a "no-growth" result. Where 6% (13 of 201 cubes) of the tubes had recorded greater defect indications on the .540" high gain tech-nique, these are established as a product of'the higher sensi-tivity of the .540 technique. This result is consistent with other comparisons of these two cachniques. As further confir-macion, in August 1982 about 29 tubes were selected fro's OTSG-1B at-large, where reinspection with .540 high gain techniques had revealed defect indications in addition to those previously identified by the .510" standard gain technique. These 29 tubes were rerun with the .510 standard gain technique and in all cases, a comparison of the two .510 data sets revealed that the cube's condition was unchanged.

It is also noted that these findings are not altered, by the results of the 100% inspection of both OTSG's by the .540" high

~

gain technique (when compared to .510" data from about 6 months earlier). No significant patterus of crack growth . vere apparent in . chia bulk comparison of data.

Within the limits of eddy current cast sensitivity and repeat-ability, no new cracks were formed or developing in the OTSC tubing during the period from December 1981 to August 1982*.

Tubes with ECT indications below US+8" of I.D. 20-40 through wall and verified by the 8xl probe to be actual defects are considered degraded tubes. Depending on their circumferential extent, the tubes will be left in service and monitored on an extended ISI program. The extended ISI program will include 100% reinspection of the 40% and less through wall indications as a separate subset for three continuous refueling outages. If these eddy current examinations show no substantial growth in

• the cracks, they will be left in service. Tubes showing signs of crack propagation will be taken out of service based on normal and accepted criteria. Lack of defect propagation will give additional assurance that the mechanism is arrested in the

. long term.

The degraded tubes in this group were exanined in May 1993 as part of the post-expansion special baseline ECT inspection. -

With one exception, the indications were unchanged from the 76 -

e

, m. _ - - -

' .e *

=. s August 1982 record inspection. On one tube, the phase angle for an OD indication located near a tube support plate shifted such that the peevious 35% chrough wall indication was evaluated as 60%. The Sul data did not change (1 coil,1/2 volt). The phase

• change was apparently due to interference from the tube support plate. This indication, like the other inspected, was evaluated as unchanged. E:vever, due to the uncertainty in through wall extent, the tube was plugged.

As discussed in Appendix A, the post-repair eddy current inspection also identified several small indications not noted as a result of the 1982 record inspection. The 1982, 540 probe record inspect *.on tapes were re-examined, and it was determine that most of the 1983 indications are within.che 1982 tape noise level, but can be considered identifiable given the precise location information from the 1983 results. These indications are considered consistent with the 1982 record inspection, supporting the conclusion that corrosion has not reinitiated since August 1982.

Degraded tubes identified in the i983 post expansion ECT examination have been added to those previously identified for special monitoring at the 90-day and subsequent inspections.

Changes in these indications will be evaluated to determine if

, grcwth of the existing ECT indications is occurring. ,

More rapid propagation if it occurs, will be evident during the procritical and power. ascension test program described in Appendix A. This program will subject the tubes to normal heat- '

up and cooldown stresses and to one accelerated cooldown stress test. These transients are expected to cause defects to open so that any circumferential cracks which are of a size that wculd I

propagate unstably will be readily detectsbie due to leakade.

~

Leakage will be monitored both during cooldown and during steady l

state operations. Between each of these cooldown tests the plant'will be maintained in a hot, pressurized condition to allow time to determine if cracks have propagated to a chrough wall extent and to allow sufficient time to detect any changes in leakage. Cracks which opened during the cooldown transients are expected to continue to leak during steady state operations since plastic deformation at the crack tips and erosion of the crack surfaces result in a permanent crack opening displacement even when loads are reduced. The cast program will be completed by a cooldown from hot conditions to cold shutdown temperatures as a final indication of cube integrity. Laakage monitoring will be continuous during this test.

j.' Laakag monitoring programs during the hot test period are adequate to detect 100% through wall cracks wht**

mechanically fait during a transient or accides tuttion. -

Adequate time for propagation due to intergrania... stress assisted cracking will be allowed during the test pregram.

Since'

~

4 l

l l .

n" " * '

, ., e ,1EEb'"

r

• 6 i

previous experience indicated that the corrosion mechanism I propagates rapidly, a lack of any significant leakage would provide added assurance that cracks are not propagating.

Corrosion testing indicated that the mechanism will not be -

-4ctive in high temperature environments such as chose that will exist during the test period. Eddy current examination is not planned to be conducted at this phase of the test program since leakage detection has higher integral sensitivity and reliability. In addition, opening the steam generator for inspection ,would expose the tubing to an unnecessary oxidizing l i

environment. Goed engineering practice dictates that exposure  !

to air should be minimized.

• Both long and short term corresion tests provide evidence that the crack mechanism is arrested. The flaw growth program has shown that crac' k s are not propagating or initiating within the steam generators. The precritical and power escalation test program will give assurance in the short term, and in the long

-term, eddy current testing, steam generator leakage monitoring, and the long term corrosion test program will give tube in-tegrity data and assurance that cracks are not initiating or propagating during operation. In Addition the monitoring of degraded tubes will give additional assurance that the cracking mechanism is arrested.

5. Defect Detectability -

An eddy current inspection was conducted of 1005 of the in-service tubes in both steam generators for the full length

' below the top ten inches of the upper tube sheet. The system

• that was used for this inspection was a .540 inch diazecer standard differential probe with a effective gain of approxi-mately 60. Any high noise or otherwise difficult to interpret indications were resolved in conjunction with data from an eight coil absolute probe. The selection of this system is documented in detail in Reference 20. The following sn==mrizes the quali-fication process which resulted in determining that this systes demonstrated adequate sensitivity to detect defects that should be removed from service. This section discusses laboratory calibrations, comparison of field data to metallurgical inspec-I tions, measurement of laboratory grown cracks and comparison of differential probe data to absolute probe data.

l 4

The eddy current probe systems were tested against electro-discharge machined notch defects at the EPRI non-destructive examination research center in Charlotte, NC. Circumferential machined notches of .187, .100, and .060 inches were machined on the inside diameter of '"tI archive steam generator tubing. Each length of notch was machined to through wall depths of 20%, 40%,

60% and 80%. Minimum levels of ddcectability were determined by . -

comparing defect signal to a field noise level of .3 volts.

e

E

9. ,

s 4

' The results of these calibrations for a .540 inch diametsr dif- l farential probe with a gain of 60 at a frequency of 400 HZ are  !

shown in Figure IX-1. Cracks with geometries that fall to the right and above the curve are detectable, those to the left and -

below are undetectable. The perfectly horizontal geometry of {

the machined notch is not totally representative of the cracks l in the steam generator, but it provides the least detectable geometry for differential eddy current probes. This geometry, .

therefore, should provide a conservative estimate of defect detectability. Details of this qualification program are con-tained-in Reference 20.

On three separate occasions tubes were removed from both sesam generators for metallurgical examination. Details of these examinations are contained in References 3 and 4 One of the purposes of these examinations was to correlate the ECT signals

, with actual defect geometry. Eddy current reported thru-vall penetrations ranging from 40 to 95%; 39 of 42 cracks investi-gated in the laboratory had 100% wall penetration, and the three cracks were observed to penetrate 66, 70 and 70% through wall.

One possible explanation for the thru-wall discrepancy is taat

. the cracks may not be open in-situ. Although the Intergranular

' attack penetrates the entire wall, sufficient continuity exists across the grain boundaries to give a less than thru-wall eddy current signal. There were no cases below the roll transition area in which a defect had not been detected by ECT.' Details on

, metallurgical correlations are included in Reference 20.

The particular geometry of these defects were tight circum-farencial cracks that in most cases, were undetectable by visual examination under microscope or even by high resolution radio--

graph. The primary means of detecting these cracks is by tube axial sectioning and reverse bendin.g on a 1/4 inch mandrel. to open up the cracks. Tubes were examined by this r.ethod noc only in areas where defects had been detected by ECT but also in good areas of defective tubes and in good areas of tubes taken from low defect areas of the steam generators. In all, 24.6 feet of tubing was examined by this method and no defects were detected except where ECT inspection had indicated a defect. These results increase the confidence level that the sensitivity of the ECT inspection method is sufficient to detect all defects in the steam generator tubing.

To compare the absolute to the .540 high gain standard differen-cial (S.D.), the sample of 3232 cubes pr3viously tested by Absolute ECT (4x1) was recasted using the .540 high gain S.D.

technique. The sample was predominantly from the high and' low reject areas of OTSG "A". The first comm rison using the normal S.D. 540 technique indicated a correlation of 99.5%. The quantity of tubes that did not correlate was 16 low level in- _

dications. With absolute data, it was determined that these

6 .

v .

~ , ~

. - d d deet

= cc

= uet,  %

de =d d e' 4 ac.id t

n f n f n 9

d eC i e yt oo si t oo r o

vt si t 0t l

l A -e ro ec de c l

an uS pG iod t on l r e er e i 6c a co det

• nf i i sl aC r aC r ede t at cde i

el i l

gt r a bteh oA oA t nd end gf bS bS suC juCo um ui aG aG eo/

l l r eif Tw LI LI TbE PbE ro/

0 '

n O

t e o - =

- 4 .

- - 5*

MC - -

0 y 0 a !>!gY"

Q::-

.i~

e. :b

e. ** y. bi:" ggg" g  !!

i;;i-i 1

.,I i,^

,,!i M

?

1 0

9

. B: S S m 5 0T i

f 0 4 U 4i 5. ni N 8 OC '

I A

T S

.T 7;* a

+j; !

AG . :ps

.+ "g"  : 7 MI .. : - " ; :.<

i.

1 -

RR ii; i.*

, p . 2 .

X O O -

e.

f t..,

0 l I F F NY

a* . o l iX - e 6

l a

.. d b.. i+- '

g OT  :

.3-l iea tW  :.

. .!  : W n CIV .f cW h

u L o AT I

C e *.

.i$

mi :agaa

. ;i e 0 5

g u

./ tw o l

F CIS .Ee-, r I d .. .

h M.+

GN

.I:

g;. 0 T RE . - ,Y g!  :

i 4

US yi: s d( w . !."

LT .  :,

LC ..."

, " m *;:.g,.i;"

AE T

i .s

i. $

.., j e 0 3

E F s::

l

<: -+ :p:.;

MO

. . l

.;.. t_3

!:. .i:::1{i. i i: s !!

i::

j,..!

^I

' =

iii!i l3 :*: .

i  !

Yi ji :s:.i r. ! f i  ! .

0 33.

: 2 i

~ -

c'i r  :-

2 f!t::% #3leh.i :w :=..TT.g1 i !f":M

-e  :

V' ! $:. cd  ;~ 5-f -;

c I

-..:.+ ~'5+T:

i 2

n

.. i: l.~ ii" - .

e =i e 0

_Mb .Vp M -- 1

.f-w::.

5 p.

~

.i:: '

m:n:7=.,< S i

~~ . .:

~

• ti ;i t I [:  ! '

T U

5 6 " ,

7 0 5" 8 dse 0" 700 w 7 4 1 7 o 1 5 el 3' 806 1 1 1 8 nr ie 1

1. 0 l

e fd B

• *

• en l 0 0 0 0

du * )

" 0 6 0 4 0 no 0 r o 4 4 3 3 2 1 8 2 1

Ub 6 . f n0 5 s o0 i .

dt(

r ah a ct

. . did f

• ce.5 3C* .]::2 - .o ni alaw i

. t t

suh qc sCt e/o TEN t

a

~* * .

rh-.

, indications were all one coil, suggesting that the circumferen-j tial excent was-relatively small. In reviewing the .540 scans, it also appeared that the 16 indications not detected were consistently in the field of the high noise level. To better -

detect these 16 indications, an I.D. frequency mixing (to remove tube noise / chatter) was added to the S.D. 540 technique. With the added I.D. mixing, S.D. 540 capabilities were enhanced to 100% correlation with the absolute technique as the remaining 16 indications were detected.

This comparison establishes that'the normal S.D. 540 technique is as sensitive a method for flaw detection as the absolute.

B & W -Alliance Research Center conducted a program to artifi-cially induce IGSAC 'in archive Inconel tubing. Following exposure to thiosulfate-bearing solutions, the tube specimens were eddy current tested. Scanning Electron Microscopy clearly showed the intergranular nature of the cracking and confirmed that the laboratory induced cracks reproduced the type of cracking and crack shape found in the service failures. For cracks investigated by successive grinding and polishing, measured axial' extent ranged from .006 .017 inches. This is somewhat larger than that of the EDM notches used during the previous 0.540" probe qualifications tests. This value is also

, ' larger than the .002" minimum seen during failure analysis on i

actual tubing. However, the measurements on service tubes were

, usually estimates made on SEM photos rather than metallographic sections, and would be expected to be lover. The crack axial extent in service and laboratory induced cracking can thus be e

_ concluded to be comparable.

Correlation of eddy current results with metallographic 'observa-tions was performed on samples with the following results.* A summary is shown in Table IX-1.

1. The threshold of detectability appears to be comparable to that determined by the original qualification testing. A crack, .040" x 40%, was below the level previousl~y found detectable and was in fact not de- ,

tected. Cracks of .315" x 38 and 0.140" x 54" were I detectable by 0.540" probe. This is illustrated further by plotting the points on Figure IX-1.

2. Using the G7UN ECT 2 step screening technique, 8 samples were tested. Four samples were dispositioned as accep-  !

table and four samples were dispositioned as having 1 unacceptable defects. When confirmed by metallography '

there was 100% correlation. I

. ~l.

1 e ,. - ..- ,-.- ..,,,,..n . . - , . , .,, ,.- - - - - , . - . - , - = - - - - , - - - - , -

2,..~....w.-..~.. .  :.

]

L Fram these tests it can be concluded that the detectability of laboratory induced cracks confirms the qualification o'E the .540

- differential probe using actual steam generator cubing as well as EDM notches. .Similar qualification programs were conducted

• to determine sensitivity of the 8xl absolute probe, and to cor-relate sensitivities of both probes in the high noise area of the tubesheet.

The ECT system used during the steam generator inspection has the sensitivity to detect crack of the sizes indicated in Figure II-1. .All defects above that size have been identified except e:

' for a small number that may have been missed due to random 4

equipment or interpretation errors.

l; C. Undetected Defects Some number of undetected defects or other tube surface 4

anomalies may remain in service after the repair is complaced.

These defects fall into the following categories:

1) Local Intergranular Attack

2) Below the detectable limits of ECT i

3) Detectable by ECT but missed through random error Specimens of actual OTSG cubing have exhibited areas of general surface IGA one to two grains deep. This local ICA is sbailar to that seen in other Inconel 600 cubes and is generally agreed to be the result of the tube manufacturing process. A few isolated instances of IGA from 6 to 10 grains deep have been found; they are associated closely with visible multiple cracking.

• One use of the long term " lead" corrosion casting program described in Section III will be to show that these phenomena are not contributors to tube failure. Specimens selected for this test program will contain general surface ICA as well as crack indications. IGA islands cannot be specifically included as test specimens because its occurrence is randon, and it can-not be detected other than by destructive examination. However, by bounding this condition with specimens containing surface ICA and actual cracks, the influence of this condition can be assessed especially on consideration of the fact that metallo-graphy has shown that the most extensive ICA is in the vicinity of major cracks. The development of IGA and/or cracks will.also be assessed during the test program 'as specimens will be period-ically removed from the test solutions and metallographically

, evaluated. Both the metallurgical examination program and the long term corrosion testing program provide assurance that the -

steam generators can be operated safely with local ICA on the tubing.

t

~

, 1 0

. ..w- y.. .__.~__...,.r.,,__+_.. ,,,.-.,mm.,.y .,m.._,m, ~ , . , , _ , _ , _ _ _ _ . - - . _ _ _ _ _ . . . _ _ - _ . . . , _ - - _ . - - . , _ _ - - _ _ _ . , . _ . , _ _ . -

TAlli.1: lX-l_ '

i 3.AllDRATOltY UlDilCED CilACKS E/C COHREI.ATION ^

kT

. - j;-

.. l i

4 EnDY CllititENT EMAM '

PHYSICAL .510 .540 HETAI.IA)GRAPilIC CORREI.ATION SAHPI.E ID APPEARANCE 4x1 GPUN 1 CIRC.

& T.W. 't T.W. l COII.S TilRU HIN. AXIA'L DISP. I.ENGTH HALI, t EXTENT i

A - 1.75 DISTORTED 5 20 - OD HE 1 - ID R 0 .~2 "

A - 2.32 50% .014"  !

' DISTORTED 35 - OD HE .

1 - ID R 0.17" 5 634 .017" D - 3.32 DISTORTED 20 - OD 420 - OD 1 - ID A 0.4"-0.5"

! 184 .006" C- .76 ACCEPTABLE 20 - ID <20 - ID HDD A

D - 1.9 SURPACE ANOHALITY ACCEPTABI.E HDD 35 - ID*- HDD A HG VISIBI,E DEFECT I

E - 4.0 ACCEPTABLE HDD

! 65 - ID 1 - ID R .315"H 38% .0065 E - 4.3 ACCEPTABI.E HDD

i. HDD HDD A ,030 25%

}

F - 4.8 ACCEPTABLE 85 - In i 55 - ID 1 - ID i

R .14" p4% .012 1

4 i

1. R = REJECT 3 A = ACCEPT '

i 2.

IIE = HOT I:XAMINED (TUBE III REDilCTION DID 8807 AI.I.ON PASSAGE OP 0.540 PR  ;

3.

IIIGil GAIN PARAHETER SIHill.ATED 0.540 SENSITIVITY '

! 4. It = lillt. TIPI.E .

,{ 5. It is believed the physical OD distortion on the tube has producqd the l , OD differential eddy current response. i 1

e l'

- . = . -

- ~;-+. -- -

~ l

' i I' a

y l

I 4

.i

'l ~In addition to surface IGA, the existence of email cracks below -

-) the threshold of addy current detectability has been con- '

sidered. Corrosion tests have shown that crack propagation by

'l chemical means is unlikely. Stress analyses were conducted to -

determine whether small' cracks could propagate under conditions

. of mechanical loading during normal operating, transient, or-accident conditions. The evaluation performed had determined

! the maximum flaw size which will remain stable under normal operating conditions and transient loading. The acceptability

of small cracks in service is based on demonstration that addy -

current examinations have identified existing cracks of this *

maximum flaw, and that the small cracks will not propagate

.j rapidly to this size during operation.

The tube loads are derived in part from a generic B&W document l (Ref. 52), in part from measurements of the TMI-2 OTsG cubes .

(Ref. 51) and in part from GPUN calculations (Ref. 61).  !

Recourse is made to field measurements because the steam-generator performed better than design assumptiqns predicted.

Twenty degrees more superheat is seasured than predicted. The -

PUN calculations were performed to provide additional infor-nation on how TMI-l tubes are loaded compared to generic cal-l culations, and how calculated loading varies with changes in initial assumptions. These calculations have previded a range '

of potential loadings'for normal and transient conditions. .

The axial load on the tube during anticipated transients, such as heat-ups, power ch,anges, and reactor trips, and steady-state operation are due to:

a. Differences in tube average temperature and the average temperature of the steam generator vessel vall.

b. By virtue of the end fixity of the tube, a longitudinal pressure stress evolves through Poisson's effect. Pressure effects on the shell and tubesheet also contribute to tube axial load.

c. A residual tube axial load component, called cube preload, l exists from fabrication.

d. Tubesheet flexibility mitigates axial load, to a greater

' extent near the OTSC vertical center-line than near the periphery.

Superimposed on the steady axial load is a high cycle, flow induced vibration (FIV) bending load. The frequency and dis-place-int magnitude of FIV was measured at TMI-2 (Ref. 51).

Potential for crack propagation was evaluated in two ways, a . :

fatigue frseture mechanics calculation incorporating FIV and normal operating transients, and solid mechanics calculations of one time transient and accident loads. These calculations were t

9

4 .

.5 .

OTSG Tube Critical Crack Sizes 2.00 -

0.D. MAX ARC-LENGTH.

1 .

\. . .

\. -

1.75

\

\. .

ECT + -

\. \.

\

\

MSLB \. @ - '

. 1.50 -

LINE M. MSLB

• s *

(3140 lbs) \. (1408 lbs) s s

@ \

,3 1.25 -

2 \.

ECT +

2 \* ~

g 1.00 -

E \. \

e N Z .75 -

's,

-f N o 's -

5 s. '

.50 - ~a MAX INITIAL CRACK SIZE TO '

ALLOW 40 YRS. OF STABLE CRACK PROP 0GATION.

AKth=4.0 ,

.25 -

I I I i 0 20 40 60 80 100 DEFECT DEPTH !N % W ALL THICKNESS FIGUREIX 2

_ i

4. ,. '.* .. ,

used to generate curves showing crack depth vs. circumferencial extent for the maximum stable crack configurations. The resu,1cs are shown in Figure II-2.

D. OTSG Tube Failure Analysis for Unplugged Tubes Curve A in Figure IX-2 represents the maximum crack size found.

. co be stable with respect to fati'gue flaw growth over a 40 year-lifetime. The fracture mechanics- model uses a preexisting crack

, and evaluates its propagation under high cycle and low cycle fatigue.

During steady state operation the steam generrtor could have an .

axial tension of up to 500 lb. acting on the tubes. In the analysis, the load cycle imposed on the tubes included mechanical and thermal factors. Low cycle, high amplitude loads were combined with high cycle flow induced vibration i (F.I.V) loading. A graphical representation of the load cycle

}

is shown in J igure IX-3. The analytical model, which used EPRI fracture mechanics code BIGIF, cycled the high cycle FIV load I about 500 lbs. axial can's ion, the steady-statt value calculated from TMI-2 test data. The F.I.V. deflection selected .

corresponds to the largest peak deflection seen at a TMI-2 sensor during a steady state condition. This is 3 mils, peak half-emplituce displacement. The sensor was locaded at a " lane"~

tube which e:.periences higher crossflow than the average tube.

The vibrational load amplitude was selected for conservatism to be the maximum tube d,isplacement seen under stes'd y-state loading.

Combined with high cycle loading was the maximum tension excur-sion represented by the 100*F/hr. cooldown, as calculated by 5&W for the design basis steam generator, which imposes an axial load of 1107 lbs. This peripheral tube axial censile loading is approximately equal to the maximum load calculated by GPUN for any individual TMI-l tube during a normal cooldown with a ILaiting tube-to-shell delta T of 70*F. The FIV ind one cooldown comprise a load block as shown in Figurs IX-3, with six cycle times per year.

Empirically derived values were used to represent FIV cube loading. The value of R (K minimus/K maximus) thus is approximately equivalent to actual conditions.

The fatigue calculation considered the fact that there is a point at which small indwelling cracks have no ef fect on fatigue resistance (the endurance limit). This value of the stress intensity, below which cracks do not propagate, is the stress intensity threshold (delta KTH)*

A modified Paris equation was incorporated in "5IGIF" with the '

feature that if the stress intensity range did not exceed threshold, no growth would occur.

n . -> .

The TMI-1 calculation was performed using delta KTH=4.0 KSI (in) 1/2. This value is based on the empirical data for Inconcel 600 shown in Figure IX-4. The intercept of the abscissa is the threshold for propagation. The relationship has ,

been determined for threshold stress intensity and R values so that data taken at different R values has been used to calculate the appropriate threshold for TMI-1 conditions. Delta KTH

• 4.0 KSI (in) 1/2 is judged a conservative value.

The evaluation has shown that the-high cycle flow induced vibra-tion does not contribute to propagation for ID circumferential cracks. Low cycle leading from the startup and cooldown is the significant contributor to stable crack growth. Curve A '.n Figure IX-2 represents the maximum crack size which can with-stand 40 years of heacup and cooldown cycles without reaching a sian which will propagate rapidly to a through wall defect based on the conservative analytical assumptions made. Since this curve is to the right and above the eddy current detectability curve, cracks of this size will not exist in the free span area of the repaired steam generator.

Solid mechanics models have been used to evaluata the ability of cracked tubing 'to withstand one ti=e transient loads. (Ref. 17 and 61) The maximum accident axial loading on the tubes is during a main steam line break, and is defined as 3140 lbs for peripheral tubes and 14C8 lbs for' core tubes in generic ,

licensing documents. GPUN's TMI-l specific analyses (Ref. 61.)

indicate these values are conservatively high, and that actual MSL3 loading would be'2500 lbs for a peripheral tube and 1000 lbs for a core tube. These axial loads include preload, pressure effects and tube-to-shell temperature differencials.

Figure II-2 shows curves of the derived critical crack sizep in through wall and circumferential extent for'the various load-ings. In the analysis, the fact that a flawed tube will move

. laterally under the axial load is included so that the centroid of the damaged cross section approaches the line of action of the load through the intact tube centerline. With this model, the induced bending moment at the flaw is reduced. Assumptions on the tube stiffness remaining and the manner in which strain is absorbed in the area of the crack have been selected to make the results conservative.

Figure IX-2 shows that the maximum crack sizes for failure under l transient and accident loading is to the right and above ECT I, sensitivity. Thus the probability is very small that cracks of this size will remain in service after the completion of repairs.

The laboratory calibration results and the correlacion of the differential production probe to the absolute probe results provida confidence that all defects above the detectable size -

.. . 3 .

1 mounsix.s -

OTSG i.oading Cycle for Tube Mechanical Evaluation 1100 - - - - - - - - -- - - - - - - - - - -

FIV - 2.4 X 10s CYCLES /YR.

HEATUP j g g q C00LDOWN a 500 ----

wsm 7e*w F V, amt 100 0

TIME +

e FIV ALTERNATING LDAD, REGION I ~ t 550 PSI, .004" MAX, DISP.

e HEATUP/COOLDOWN, S CYCLES /YR.

e 40 YEAR LIFE e MAXIMUM AMPLITUDES

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

s MGURE IX-4 Fatique - Crack Propagation Behavior of Inconel 600 da/dn vs AK forINCO 600 .

10-4 A 75* JOURNAL OF ENGINEERING MATERIALS CURVE O 800* JOURNAL OF ENGINEERING MATERIALS CURVE

' ~ .

G 77* MIT CURVE ,

A O 554' MIT CURVE g'

10-5 -

• 6

_~

e 00 .'

i x ~

o/s ,

.1 6.A ~

4 -

ra 3

• 10-6 --

- A

_ f

~

A 608* P0lNT BORATED WATER

\

10-7 ' ' ' ' ' ' '

1 ~10 100 LOG AK, KS1 vin" STRESS INTENSITY FACTOR RANGE

nounu m.s OTSG Leak Rate as a Function of Crack Length & Tube Tensile Load .

100l .

30 - 1100 # tension /

30 -

70 -

E

a. A E 80 -

I ~:

m 500 # tension G =

m .

M 5

a so -

E g .

n a0 -

E 30 -

g , , ,+

200 # tension 10 f I e i f I O 0.1 0.2 0.3 0.4 0.5 0.8 0.7 0.8 0.3 1.0 1.1 1.2 1.3 .

TUBE CHACK 00 ARC LENGTH (INCHES) m

a . ..~ .

mounsix-s Y MI-1 OYSG -

Tube Lusitrate Durisig i>laut tuoidown Froin MSLB Pro-Critical Crault la One Tube -

at Various Coo! clown Conditionc

, IM 81 AREA A TlVil I SPECIFIC iVISLG

. l l AREA S GENERIC IVISLB CUIWE N O. OhSI3 140 -

.:.,:::.gi  ?.Mi.h:: .g".

q 1- NU TUUE Pile LDAU:

130 - -

COULDOWN W/70*F

. .. SilELL TO TUDE A T 120 _ .

14A LIMIT.

. :: 2- 200# TUDE Pf1E-LOAO:

i10 - ..1 y . i:: :

. . NM)ii.:. ._ ;, cggLgown yj7gup

- .: SilELL TO TUDE A T 100 - i . :::S:: f: ' :i :f: LlMlT.

. v

. .

g ' g.:['? h"i: ' 5 :..

o --  :  :- ': 3- NO TUDE Pi1E LOAO:

jjj 3 .!: g  ::: COOLDOWN AT 100*F/IIR c= g

!l  :!!(Y:::.?:!:  !!.3 W/140*F SilELL TO TUDE

-3 80 - o y.g::  :

'.~ ? A T LIMIT 4o NfW, , !!Yi:. :i

. y*\

• g ~ ... .{(!!l 70 - .'!  :

'Q 4- 200# TUDE PRE-LOAO:

Eo ';! : -

g MU 80 - ii :.: !N. .:.

i:

5. ' '

I I

C00LOOWN AT 100"F/IIR W/140*F SilELL TO TUDE'

!8' +i' 1 g A T LIMIT

@'5E'

.. t:. $

!:: \

W5 3-50 -  ::::! .: .:

' ' . '99

\ 48 5- BSW GENERIC

\

• C00LOOWN TUBE LOADS.

)

40 -. l:  : I 3B DOES NOT INCLUDE RCS &

.)jf

ii ..:.

' 5B V / OTSG PRESSURE LOADS OR

_ M;;y:gg , ,! [f russ nnwsa 28 20 - e ~. \

/},

// 1B y r/ /

10 - /

O i i i 6 0 500 1000 1500 2000 TUBE AXIAL LOAD (LBS.)

(During Cooldown) 4 .

E . l o . .. . '

I

.. \

will be found. However, there is a small probability that some . l large cracks may not have been detected due to problems such as '.'

l high noise levels, probe lift off or chatter, or data analysis errors. There is also a small probability that small cracks

. i

,W*

could grow circumferentially with time. Because of these "'

possibilities, an evaluation was performed to determine if tube cracks of the size that would propagate can be detected due to leakage during the hot testing program and during subsequent ( l,];

operation. ,.

~

Leskage through a 100% through-wall crack of a given length is a _'

function of crack opening displacement (COD). The COD of a '

l circumferential crack is a fune. ion of the axial loading of the tube. As discussed above in the OTSG tube stress analysis, the {f tube axial load is a function of several variables and may have either tensile or compressive values. A calculation was '

performed on the basis of first principles in order to establish the tube loading under various operating conditions. This -

calculation incleded the ef fects of dif ferential thermal growth .

f, between shell asd tube, pressure loadings and tubesheet deflection as well as a value for tube preload. The nominal *

- preload value of 280 pounds was based on gap measurements between parted tube surf aces as measured in the TMI-1 OTSGs.

The minimum preload value was taken to be sero, assuming all, preload had been relieved prior to expansion. The calculated results of tabe loading based on first principles correspond ,

fairly well with the values calculated in the generic evaluation

~

(Ref 52). ,

CODS have been calculated using these tensile tube loads for the .

MgLE through well crack sizes using an NSAC model. After .

determining a conservatively small minimum crack opening displacement, leakage through the opening was esiculated. The, evaluation includes consideration of phase changes and pressure drop as the primary fluid passes through the crack. Figure IX-5 shove the reistionship of tube leakage versus crack are length for various tube loads which control the 000. Figure IX-5 was ,1.

developed by NSAC based on a model for two phase flow through a crack with initially saturated or subcooled fluid (Ref 66). ,

The greater the tube loadsas, the smaller the initial crack size ,

will be to produce an egaivalent leak rate. Because tube S.

loading is greater during cooldown than during any other normal '

operation condition, laakage monitoring will be conducted during .

cooldown. The following table (11-2) shows the cooldown leak rates for peripheral and core tubes through a crack of a size y that will not propagate to the through-wall MSL3 critical crack size during numerous ecoldown/hestup cycles. For purposes of 4, this discussion, this crack is called a "procritical crack."

.b Figure 11-4 shows leakage through this precritical crack for the "*

nominal range of potential tube loadings and locations. e

-o , ..a .

Table IX-2 TMI-1 OTSG Tubes Pre-Critical Crack Sizes and Cooldown Leakrate Tube Location Core Perdpherv Crack size Basis Generic TMI-1 Generic TMI-1 MSLB Specific MSLB Specific MSL5 MSLB MSL5 load (1bs) 3140 2500 1400 1000 Procritical Crack Sise (inches) 1.25 1.5 0.5 0.53 Cooldown Tube Load A

~

5

~

A

~ ~

5 A

~

B A- 3

~

Basis (see note below) ~ ~

Cooldown w/70'7 T Limit (Normal Cooldown)

Cooldown Tube Load (Lbs) 150 400 15 0 400 800 1050 800 1050 I

t.eakrace @ :ooldown (CPH) 8 40 11 52 12 19 29 44 j Cooldown w/140*7 T Limit .

(Hot Testing Cooldown) ,

Cooldown Tube Lead (Lbs)

• 600 800 600 800 1500 1750 1500 1750 14akrace @ cooldown (C7H) 73 110 96 144 34 42 74 92 Notet ,

A= Tubes which have had preload relieved Be Tubes with 280d nominal preload * -

l 6

e t

..A.

1

\

r- , A .

l l

Figure II-6 and Table I1-2 can be used to conclude that pre-critical cracks which may exist in the TMI-1 OTSGs can be detected by leakrate monitoring during the Hot OTSC Test Program. Specifically, one cooldown will be conducted at -

approximately a 90*F/hr rate with a tube-to-shell differential temperature approaching 140*F. Considering the TMI-1 specific MSLB loads, the majority of the OTSC tubes would have leskrates from any existing pre-critical cracks within the graphical area represented between line 3A and line 4A. The minimum leakrate represented by this area is about 75 GPH which is easily ,

detectable.

Any cracks growing circumferentially the pre-critical size will be subject to cooldowns conducted in accordance with GPUNC's normal operating guidelines (70*F shell to tube differential temperature limit). The majority of the OTSG tubes would have leakrates from any existing pre-critical cracks within the graphical area represented between Line LA and Line 2A. The minimum leakrate in this case of CPU is still readily detectable. Because cooldown/heacup thermal load cycling is the only significant contributor to mechanical growth of circumfe-rential cracks, leakage evaluation after cooldown loading is expected to be representative through the following cooldown. ,

s 6

O e

• i e

I

' s' .

The leak rate predicted for cooldowns indicated in Table IX-2 and in Figure IX-6 are above the detectable leakage shown in Section 1.

Several faccors which will increase leakage through a given size crack were not considered in developing the evaluation above.

These include the removal of crack surf ace material from erosion and grain drop-out, and the fact that existing circumferential cracks which have experienced a tensile axial load will retain a residual crack opening displacement. If a crack was opened on .

cooldown, empirical evidence (Ref. 62) indicates that plastic deforma' tion at the crack tip keeps the crack from closing when the cooldown load is relieved.

The OTSG 1eakrate will be monitored during each plant cooldown.

High leskrates would be investigated prior to restart. If the plant is restarted, operation will be subject to an administrative limit on steady state leakage which has been established with sargin below the lowest leakage value in Table IX-2. Laplicit in setting this limit is reliance on a Theresidual crack opening displacement remaining after cooldown.

administrative, leak ta,te limit assumes all increases in leakage to be from one OTSC tube and has a value of 6 sph above the baseline.

Exceeding the administrative ilmit will require bringing the plant to cold shutdown and leak testing the OTSG and repairing' any identifiable leaks. Bubble tests are expected to identify individual leaking tubes. Subble test sensitivity is approx-instely .1 gyd/ tube. Thus,-any cracks in service of a size which is or may propagate circumferentially to a critical size will be identified.

In addition to the steady state leakage limit, normal plant cooldowns will be accomplished at less than 100*F/hr and limit the CISG shell to tube delta T to 70'F. This limit will reduce tube stress which could be lead to mechanical crack propagation but will still provide adequate stress on cooldown to open pre-critical cracks to a large enough residual CCD to enhance leakage detection.

E. Conclusions Evaluations of tubing lef t in service have been performed to verify that they are acceptable for use. The possibility for additional corrosion occurring after return to operation has been considered and found unlikely. In considering existing damage, the ability of cracked cubint to withstand steady state, transient and accident loading has been examined. All cracks of 8

e

i .,

s , .

a size that could be. expected to propagate under loading a're within the range of detectability by eddy ' current testing. If a through well circumferential crack of critical size for growth has been inadvertently left in service will be detectable due to -

leakage during the procritical OTSG test program. Cracks below that size which grow circumferentially under cooldown/heacup loading will also be detectable due to leakage before they reach the cirtical size.

X. OPERATIONAL CONSIDERATIONS

, The operational concerns of primary to secondary leakage were evaluated. Concerns included leakage monitorir.3 during normal operations'in both steaming and nonsteaming conditions, and sampting steps to be taken wh'n e leakage is detected. In addition, a program has been formulated that includes procedure resiew and operator .,

training which will provide bsproved operator guidelines for dealing with tube leakage and tube rupture events.

The operational guidelines discussed in this section are applicable during normal operation with low levels of primary to secondary leakage. A more detailed description of these guidelines can be found in reference 58. For primary to secondary leakage ratio of 50 spe or greater, these guidelines will be superseded by tube rupture guidelines as discussed in Section X.D.

Operational concerns can be grouped into three gener_al aresst

1. Primary to Secondary' Leakage which includes leakage detection methods, and actions required based on primary to secondary leakage. '

2. Radiological concerns which include detection methods, worker protection measures and plant discharge limits.

3. Secondary' side chemistry limits based on baron and ilthium con-centrations.

This section summarises the guidelines for operating with tube l 1eakage.

A. Primarv to Secondarv Leskate l During normal operation the methods which will be used to detect j and monitor leakage are

1. Offgas continuous monitor (RMA-5) ,

2. Tritius samples from the condensate and primary system.

3. Offsas grab samples 9

es

L

.)

.f j.. .., . ;y'.

"fC /;' . d//'r*? , 4 l' l- /..,, / a! /

if . .C tt 5 p ,p. th mt / // 7 v M .* Ts ! .

QPU Nuclear Corporation

. N oute It ScuIn .

Miccletown.Per.nsyivarra

• T057 717 944 762*

, s 4 e-TELEX 84. 2386 Wrtter s Dire:t DialNum:er:

Sectemt:er 30,1963

$211-43-279 Office of Nuclear Reactor Regulation

_ Attn: J. F. Stolz, Chief Ccerating Reacters Branen H Divisien of Licensing U. S. Nuclear Regulatory Ccmmissicn Wasnington, D. C. 20555

Dear Sir:

Three Mile Island NucIsa Station, Unit 1 (TMI-1)

Cperating License No. CPR-50

'Oochet No. 50-289 ..

Comments on NRC SER Conca:ning IP.I-l Steam Generate:s Recai This letter is intanced to convey GPUN's response to you August 25 safety evaluation en TNI-l :sturn te service witn recaired staan generate:s.

Attac. vent 1 tc tnis letter occuments GPUN actions en conf 1=atory items in ycur SER. Attacv.ent 2 is a list of GCUN ccanents en tne Staf f SER.

We hece these items will succc t p scaring tne sensculee succlement to ycur

$!R.

Sincerely,,,,

o

" . j, f ,

H. O. ' il Vice P:ssicent - TNI-1 cc: H. Silver .

b W 7f GPU Nuclear corceratier is a ss: sic s'y cf ine Gent'ai 86:oc Ut... ties cerce at,cs .. ..

a ,. . c. . . j

, u .

.}i ?

Attac. vent 1 .

GPUN Actict's en Ccnfirmatcry Items ,

Section Pace, Paracraon Item /Rescense 4.3.1 39(2nd) Item: ". ..an t.ccated TCR 406 confi: ming the accifica- ..

tiens ciscussac aceva and centaining accectacle cocunentation cf the analytical justifications for tnese mocifications is summittec pric: to '

restart."

Response: TOR 406, Rev. 2 is enclosec. 5ccerting references and p ccecures have been mace availacle to memcers of ycur staff.

4.3.2 42(item 5) Item: Reactor Ccclant P>.:re PPSH fc Emergency Operations, "The licenses sncule insure tnat tae copy fc Cent:ci Room use is clear anc legible".

Rescense: The Cent:clied Cecy of the accrevec anc fc:melly issued 1202-5 has clear anc legible eccles cf the PP5H curves. A copy of.the procedure nas been made available for review by memgers of your staff.

4.3.2 43(item 7) Item: 5/G Isolation / Steaming Criteria, "...tne T licensee will make,a change for SG isolation f cm a dose to cose rate criteria."

Mespense: TheCont:c11ecCcpyoftreaco$vecancfe:mally

, issued 1202-5 nas ecsa :ste. stner tnan ecse isciation criteria. A ccpy of the p;ccacute has been made availacle fc :eview ty mefreers cf ycur staff.

4.3.2 43-44(item 8) Item: P:sssure Cent:ci cf 1sciatec SGs "The licenses states that inst =ctions fc: cont:c111ng crassure in an isclatse SG are plannec fc:

inclusien in D8 1202-5". -

Resodnse: Althougn the sucject of this.,quete is feecing an

• isolatec OT5G fc pressure cent:cl, the SER .

mistakenly refers to steaming" an isolatec SG ,

in the sentence esfers the quote.

~

C82W l

- - - ,-e.. -

,.,v- _ . - .,,_,_.,-,.,_-,___,.-.----_,,--,_,.,,c__mm__,_._._,-_m_,

1

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

y, a -c, .s' ..

j

. . s < n0- l g

ii .Secticn Pace, Paracraon ,

1 -

' Item /Resconse

- TDR 406, Rev. 2 no longer recommencs feecing an isolated OTSG for. pressure control (see last

• item of Sumary of Change, page 9 of 74 TCR 404, Rev. 2). It should be noted, hcwever,- tnat cue

• to an editing errer a reference to this form of

-

• OTSG pressure. control still exists in secticn 3.2.A page 30 of 74 This error will te corrected with the next revision to TOR 406. E2 1202-3 reflects the revised TDR 404, Rev. 2. .

4.3.2 44(item 9) Item: SG Shell to Tube . Differential Tamperature Limit.

"The licensee will be reouired to clarify proce- -

dural action for Celta T's in excess of ICO*F.*

Response:. The Delta T limit in the procedure is 70*F. If this limit is accroached the guicance in 1202-5 .

does require that the cocidewn rate te recuced

- or secured so as not to exceed 7088'. The guidance for a 10U88' Oelta T has been removec from TDR 404, Mov. 2 (see abstract and section 3.2.2.3 of TDR 404, Rev. 2). It snoulc be noted, however, that due to an editing er.mr, section 5.2.7 still contains the ecnfusinC second reference to a 100*F limit. This er:or will be corrected witn tne next :evision to TCR 404.- EP 1202-5 reflects the revised guicance in TDR 404, Rev. 2. ~

e 8 0

0 emp e

s O

s e S O

C82*

e e

% 9

,+ 'r 6 e 9 t s' 0 6

e Section W . 2 e

• 9 9

6 C

3.1 4

9 e

M 6 3.2 9 0

e se e

O 39 13 (

M e

e e>O e

l s e O

. .o . o fj Seccien Pace, Paracrach Item /Cerrent 3.3 15(table) Item: Taele 3.3-1. .

Comment: The nuncers of tutes listec in the last twe -

colums are app;cximations only. In some cases more were done in the baseline and will be

speated after 90 days. In some cases fewer were cone because the population of tubes in .

that category was less than shown in the table cue'to plugging cf adjacent tutes c the tutes locaticn in the perionery. ,

3.3 16 (tcp) Item: "As early as feasible in pcst critical ocera-tien, the licenses shall ccnfirm the baseline primary-to-secencary leakage rate, and estaclish the minimum inc: nase in suen leakage rate whien can : 11acly be measured (expected to be aceut

.lgpm) . If leakage exceecs the casaline leakage rate by that minimum increase, the plant sna11 be shut ccwn anc leak testec". .

Comment: As ciscussed in TR-CC8, GPUN nas establist'ac .1 gpm as an acministrative limit en leakage acave' baseline. This leakage rate is cetectacle.

3.4.1.b 16(bottem) Item: Planned tasting. ,

Ccmment: The second phase cf testng is ccmplete and results nave ::een tesortec in TR-CCS, Rev. 3.

Assessment of the jcint for the full 35 year cesign life will ce performed wnen cata en actual steam generate: perfc mance is availacle )

to sucplement the results of the 5 year anc 15 year test p:cgrams. No accitienal testing is plannec pric: tc tne startup af ter t: e first refueling cutage.

3.4.1.c 17 (tec) Item: "This cojective assures that ece;:ressive Icacs curing cperation anc vibraticnal enaracteristics of tne tute will remain unenanged."

. Comment:

~~

As statec in TR-CC8, enly tne maintenance cf vitrational characteristics .ts an ecjective. It is also a gcal that the tuces not ce in ecepres-sien when celo.

. 6*

O .

C82M

s . .

t f* .7 y

Section Pace, Paracracn Item /C ment .

3.4.2.d -20(3:c) Item: "

. ..the 1025 pcunes necessary to cause tu::s -

bowing." ,

Coment: As discussac in TR-CC8 and cur letter cf A:g'.:st 3, tube Dewing Oegins at app;cximately 8CO lbs, but 1cacs must Isach 1025 lbs befcre the lateral cisplacement of the tube excescs the ncminal -

si::e cf the scace between tubes.

~

3.4.2.c 21(4tn) Item: " . .1.0 ksi inl/2, a

• Comment: Calculatiens were cene at A.0 ksi inl/2, 3.4.2.d 21(cottem) Item: Stacle cracks.

t C:monti This paragra::n mey neec revision based en TR-CCS,. Rev. 3.

3.4.2 23(bett::rn) Item: "The Staff will :enditi:n the license to require submittal cf the ex:encec life cycle p gram qualificati n test :ssults by Startup afts: the first regularly senedulac refueling arts:

restart."

Coment: See cements fer p.16, secti::n 3..a.1.c. -

3.5 28-29 Item: Cleanup of Centaminant Cement: Scme parameters thh::ughcut in this secticn may need tc ce upcated tc reflect actual chemi:a1 cleaning as ciscussed in TR-008, Rev. 3.

Scron (b :10 acid) ISCG-23C0 ppm pH (ambient tem;:erature) 8.0-6.5 H2 0; cencentratien 15-25.p m Temeerature 130*F

• ScF Cover Gas N2(PS)

Litnium ion c ncentrati:n 1.8-2.2 Ouration of Treatment ACC hrs.

• ~~ Testing using samcles prior tc beginning cleaning monit::ec perfc:mance throt.gn SCO n s.

3.6 30(item 2) Item: "All RCS piping..were flusnec..."

Comment: All RCS pioing with a clameter g:eatc; than 1" -

was flushed.

O C82aK

, s ..

o- ,. -.= . .

!.j 7 Section Pace. Paracracn Item /Cc. snenk .

3.6 30(item 4) Ium: "The c=clant will be...menitorec continuously -

fc: pH and concuctivity." ,

Comment: Per TR-008, as recc:ced in you: Tacle 3.6-1, these parameters will ce mcnitoreofm times per week.

3.6 31(tacle) Item: Table 3.6-1.

Comment: There are several cifferences in Table 3.6-1 from our plans as outlinec.in TR-008.

Parameter Old !.imit New Umit Lithium 0.2-2.C(p;rn) 1.0-2.0(p;;rn)

Chierides 0.15(opm) 0.10(ppm)

Sodium Ncne 0.1 (p::m)

. 3.S 37-35 Item: Occupational Ocse Assessement.

Comment: Final man-rem expcsures anc final nutr.::ers of tuces plugged are racc: cec in TR-008, Rev. 3.

4.3.1 39(2nd) Item: "This dose cc::ssponcs to levels prescribed in 10CFR, Ps:t 20..." ,

Consnent: Thiscosecorrescendstcemergencyplanadtion levels. .

4.3.2 41(item 1) Item: "

. ..a 50-gom lesk rate criterien. . .cc:rescencs -

to the cceplete separation of cne tube."

Comment: A 50-gpm leak rate is acp;cximately 105 of tne leakage f:cm a c:za::lete separation (ccuole encec tube rupture) cf one,tuce. The criterien cc: responds to emer;ency plan action levels.

5.2 46 Item: License Ccnditiens 4 anc 5.

Ccmment: See conments for page 16, see!1cn 3.3. anc page 16, secticn 3.4.1.b. . -

e O

082aK ..

w

'. 3 -- , ..' _ , 9~

4. . N-16 activity measurements using portable steam line monitors ,q

~

1..

5. Primary Leak Rate Calculation These leakage detection methods are summarized in Table X-1, as well-as the frequency use during operations and during cooldewn. ,

RMA-5 will be the first indication of increased primary to 7' g, secondary leakage. The monitor will continuously sample the vacuum pump exhaust from the main condenser. Upon a 1 gph in - _

crease in leaktate in 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> or a 6 sph increase during a , ,

cooldown as calculated from RNA-5, offgas grab sample and an RCS sample will be taken and the primary to secondary leak rate will be calcula'ted using total' gas activities. The portable steam [

line monitor will detect N-16 activity and will be used to -

evaluate which steam generator is leaking. ,

Primary leak este calculations which are done daily per .,

. \N '

Technical Specification requirement can also identify increased -

primary to secondary leakage. _ Since the leak race cannot dis- f'GI einguish betwe:n unidentified system leakage and primary to -

secondary leakage, if an unidentified increase in leak race occurs, a critium and offgas grab sample will be taken to allow for an accurate determination of the primary to secondary leak }-

race.

Shutdown limits based on primary to secondary leakage wil con- '"

sist of the Technical Specification limit of 1 sps and an admin- '

istrative limit of 6 sph above a baseline leakage. Baseline isakage will be determined during the procritical hot testing l program. When a steady state leakage increase of 6 gph is I reached the plant will be brought to a cold shutdown, the OTSG will be leak tested and the leaking tubes r,e paired. Tube ,

leakage will be tested by the bubble test method. This method has a sensitivity of approximately .1 gal / day / cube or 4x10-3 l gph/ tube, therefore if no leakage is detected during the bubble '

test it can be assumed that no individual tube has reached the

• critical crack size and primary to secondary leakage may be due I to aggregate tube leakage. The baseline leak rate value will be redetermined based on an evaluation of the OTSG 1eak rate test ./ '

results and operating history after the leak test is performed. 't$a

'u-When primary leakage reaches 6 sph greater than the new .

setablished baseline the plant will again be shutdown and task tested.

When shutdown is required by steam generator tube leakage, the t

plant should be shutdown expaditiously but in a manner to pre- ',

clude reactor trip and subsequent lifting of relief valves or atmospheric dump valves. Cooldown rates will be limited by a ,j '

tube to shell delta T of 70* to reduce tube loading during .'

cooldown. .

• re

(

_. 1

..L':. .

g(. .

l l l

• e mee

7- .

o

_ , ,* ' . ' .

.e .

B. Radiolosical Concerns During normal operation with steam generator tube leakage, radiological concerns arise in the following areas: -

1. General and specific area radiation level

2. Turbine building sump activity (with respect to discharge to the envirotument).

3. Powdex resia and backwash water activity.

2 Specific and General Area radiation limits will be determined and will be based on preventing the turbine building from becoming an RWP area (greater than 5 ar/hr). Limits are needed due to the necessity for easy access into the turbine building during operation. Routine radiation surveys will be taken in the turbine building in the vicinity of the Powdex and Graver System vessel and in other selected areas. These areas will be restricted if necessary to prevent unnecessary exposure to plant personnel. Precautions will also address secondary sida system vent and drain operations. . .

In the Powdex sump, pH and conductivity analysis will decernine if the water which has been processed by the (Ecodyne Graver)

Powdex Backwash Recovery system will be returned to the TMI-I condensate system or to the turbine building sump. Any radio-active powdex will be dewatered in High Integrity Con-tainers / Liners and shipped to commercial low level waste burial sites.

C. Secondarv Side Chemiserv Secondary side chemistry limitations for Baron and Lithium will be based on considerations of chemical introduction into the turbine.

D. Developmene of Procedural Guidelines for Steam Generster Tube

. Ruoture A program has been formulated for providing improved operator guidelines for dealing with tube Leskage and tube rupture events. The guidelines cover two categories of events. The first category addresses tube ruptures for which subcooling margin is maintained. The second category deals with tube ruptures for which subcooling margin is not mintained and would '

include va'rious contingencies including multiple tube ruptures in one or both SC's, loss of react.or coolant pumps and loss of condenser.

91 n

1 a.

-mn_ , nm e n w er-, ~ . --e

-e .

.* . T*

+ a

1. Continsencies for Consideration The following is an ouciine of the programs for developing guidelines for SG tube rupture. -

a. Guidelines for Tube Ruptures for Which Subcooling Margi'n is Maintained The program to develop guidelines for tube ruptures for which subcooling raargin is taaintained will include the following basic assumptions.

e O

5 e

4

+

m O

9 92 -

~f.

,e

^

s, . * .

< *, s 4

4 TABLE X-1 LEAKAGE DETECTION METHDDS

SUMMARY

TABLE Method.

Sensitivity Frequency Special Actions RMA-5 0.48 sph with 3.8 uCi/cc Continuous strip when leak rate and 20 cfm exhaust flow chart reading. calculated by RMA-5 increased by Igph in 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />, or 6 sph during cooldown sample take off gas grab sample and RCS sample l Tritium .3 gym at 02

. uCi/mi H3 after S hours Of fgas Grab .01 gym at 3.8 uC'./cc 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> Frequency increased Sample and 20 cfm exhause flow with known leakage Portable When leakage is Steam Line

~

detected deter- -

Monitor mine which ,

. generator is leaking 4

• 93 -

• me emo

c: ,- .* .

(1) Break size small enough to maintain subcooling 5 margin. -

(2) Oca OTSG affected.

(3) Reactor Coolant Pumps operating. .

(4) Condenser available.

(5) Decay heat removal from the non-affected SG.

(6) SG steemed at 95% operating range level to assure natural circulation.

, Contingency considerations for design basis tube -

Il ruptures include:

(1) PORY unavailable.

(2) Reactor Coolant Pumps unavailable.

(3) No condenser available.

(4) High radiation release considerations. .

(5) Steam line flooding consideration.

(6) Both SG's are affected.

1

b. Guidelines for Tube Ruotures For *4hich Subceoling Martin is Not Maintained

~

4 The program to develop guidelines for tube ruptures for which subcooling margin is not maintained will include

, the following basic assumptions.

(1) Break size from one SG large enough to cause loss of subcooling.

(2) No reacto'r coolant pumps running (since subcooling margin is lost).

(3) Condenser available.

(4) PORV available.

(5) Unaffected SG is steamed. .

j Contingency considerations include:

(1) PORY unavailable.

(2) RCS voiding keeps pressure above SG safety valve setpoint.

(3) Primary feed and bleed heat removal.

1 (a) With PORY available (b) Without PORY available Both the analyses employ the RETRAN code. This code models

, TMI-1 and has been benchmarked from transients on both TMI-l and TMI-2. Use of this code ensures that plant response under various primary-to-secondary leak scenarios is under-stood.

l l

O O d

e . ,* .. .

RdURE X.1

' Tube Rupture Guidelines Primary to ,,

Secondary Laakage

> 50 gpm J

L k Manual Automatic Shutdown Shutdown t  ;

Cooldown 3

v v Forced Natural HPI Circulation Circulation Cooling L J

& vf .

. Decay .

. Heat Removal New Guidance:

- multiple tube ruptures ruptures in both steam generators HPI cooling Secondary water management i

improved guidance r

Minimum subcooling reduced l

RCP trip criteria tube to shell AT ,

steam generator steaming, feeding, flooding

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

s . ,

The guidelines developed from the RETRAN analysis for tube ruptures are summarized below and have been used for writing Operator new procedures and revising old procedures.

training prior to restart includes re;ponse to tube rupture events using new and revised procedures.

4

2. Guideline Summart- -

The symptoms of a tube rupture define It is only autry used when conditions leakag.e for this emergency procedure.When conditions require it (as defined by exceeds 50 gym.

high leakage or significant offsite releases), the plant expeditiously as possible, will be shutdown and cooled a.:

and certain normal plant limits (RCP NPSH, normal tube /shell delta T, and fuel-in-compression limics) are waived.

a. Inumediate Action

• The cube leak in question may not be large enough to cause a reactor trip. In such a case, the operator-begins a load reduction as rapidly asAvoiding possiblea without reactor causing a reactor trip (10%/ min.).

trip prevents lifting of the OTSG safety valves.

b. Follovuo Actions "

(1) Subcooling Maintained and Reactor Coolant Pumps Available -

Once the load reduction is initiated or a reactor trip has occurred, the operator has several major

' goals to achieve while bringing the plant to a cold shutdown condition.

First, he prevents lifting of '

the OTSG safety valves; second, isolates the af-facted OTSC to prevent unnecessary radioactivity releases; third, minimizes primary to secondary leakage by minimizing primary to secondary dif-farential pressure; and, fourth minimize stresses on the OTSC tubes by limiting tube /shell delta T.

Finally, the operator will minimize offsite dose by T allowing the leakage OTSG to flood if offsite doses are large enough (approaching levels at which a Site Emergency would be declared).

The major differences between the previous plant procedure and the new procedure are be the following.

~~

l

~

c~ , -f , .

e s (a) Maintain a Minimum Subcooling Margin Minimum subcooling margin means that primary to secondary differential pressure is minimized. -

Minimum differential pressure mesas that leakage is reduced; thus reducing off site dose and making the event more manageable. In order to maintain the minimum subcooling margin, several plant limits have to be violated: fuel in compression limits and RCP NPSR limits. The '

former is acceptable to violate during emer-gency conditions, while the latter is being reevaluated to determine acceptable emergency operation of the pump.

(b) Steaming / Isolation Criteria for the Affected OTSG The present procedure allows the operator to let the OTSG flood anytime that RCS pressure is below 1000 psig._ The revised procedure has the operator steam the OTSG as necessary for the following purposes. First, to prevent lifting of the OTSG safety valves. As the OTSG 1evel increases, steam generator pressure,in.che isolated generator could increase toward the safety valve setpoints. Pressure should be cont. rolled to prevent a safety valve lift.

The generator is also steamed to prevent it from flooding. Flooding is undesirable because an RCS pressure increase under this condition could cause water relief out of the OTSG safety valves. A flooded OTSG would also act as a

.' second pressurizer and slow depressurization of the RCS (as occurred in the GINNA tube rupture). ,

The OTSG will be isolated under two conditions.

First, if SWST level goes below 21 ft. indi-cated level. At this level, there is still sufficient inventory to fill up both OTSG's to the main steam isolation valves and have 30,000 gallons of water left to go on feed and bleed cooling. A second reason to isolate the OTSG is for radiological considerations. If the Radiological Assessment Coordinator (RAC) determines that offsite doses are approaching

s , .

4 the levels which would require declaration of a Site Emergency, (regardless of cause) the affected setam generator will be isolated.

(c) Tube-to-Shell Delta T Plant administrative limits and precautions will require asintaining the OTSG average tube temperature within 70*F of the average shell-temperature.

{~

(2) Ioss of Subcooling Margin with Natural Circulation

- Cooling-When RCS sabcooling is lost, the operator must treat LOCA, as well as tube rupture symptoms. First he trips RCP's and then verifies RPI and EFW have initiated. lRe is then able to pursue the followup tube rupture actions. All of the guidance for followup actions without loss of subcooling apply, as well as the additional guidance provided below.

The objective in this portion of the procedure is to maintain natural circulation, reestablish subcooling margin, restart a reactor coolant pump, and return to the section of the procedu.e for forced flow -

cooldown. .

~

When subeooling is regained in the RCS, then HPI is throttled, RC7's are started and the operator con-tinues with 1.677/ min cooldown. If subcooling can-not be restored, the operator cools the plant down

. on natural circulation, steaming as necessary to meet the objectives described in the forced flow section.

(3) Loss of Subcooling and Loss of Heat Sink -

Metural circulation cooldown will continue until subcooling is restored or the OTSC heat sink is lost (for example, due en icon of natural circulation in the unaffected loop). With no steam generator heat sink, the operator must put the plant in a feed and bleed cooling mode. Feed and bleed cooling is ini-tiated by isolating the OTSG's, assuring full EPI is operating, and opening the POR7. If RCS pressure remains below 1000 psig, then the operator continues to control secondary side pressure just below RCS pressure. If the OTSG heat sink is rcstored, the

' feed and bleed is terminated and a natural cir- .

culation cooldown is reinitiated.

a .- .e o a e If PCS pressure stays above 1000 psig during feed and bleed cooling (e.g., a head bubble prevents depressurization or the PORY fails closed) then the secondary side safety valves should be protected -

from challenge. The operator controls OTSG pressure ,

with whatever means are available (turbine bypass, ETW or ADV). When the OTSG is about to flood, the operator cpens the ADV and leaves it open. This action minimizes the chances that safety valves will be forced to relieve water and/or steam and fail -

, open. The steaming capacity of an ADV at 1000 psig exceeds decay heat levels within several minutes after reactor trip. HPI captcity exceeds the capacity of one ADV. .Therefore, the RCS pressure can be controlled at 1000 psig in this mode without

, lifting safety valves. Subcooling margin can be regained and the plant cooled down in this mode until an OTSG heat sink can be restored or until the plant can be put on decay heat removal.

A simplified schematic of tha tube rupture guide-l'ines is shown in Figure X-1. A fourth possible scenario exiJes under current procedures which has not been considered a preferred course of action in formulating the guidelines: maintenance of subcooling margin but tripping of reactor coolant pumps on 1600 psi RCS pressure. Pump trip on loss of subcooling margin instead of RCS pressure allows the operator to maintain forced flow for about 3 ruptured tubes - 1600 psig SFAS is much more restrictive. Forced RC flow provides several benefits during a tube ruptura.

1. It minimizes prizary tu secotiary delta P and thus r2 duces tube leckage and tube tensile load.

2. Prevents steam formation in the RCS. (Steam voiding prevents RCS depressurization.)

3. Provides pressurizar spray so that RCS pressure control is not dependent ~ on the PORV or pres-surizer vents.

Therefore, GPUNC is taking action to have the 1600 psi pressure pump trip requirement changed to trip on subccoling margin.

i

~ -

9

,g- = , - . - . , ,r -_- -,wv w-- w - ~ = - se - - - - - w - - w--- *-T 8*"'""""*T-W'7""---* * ~ ' * - - - ' ' '

e- c p ,

E. Conclusions Primary to secondary leakage will be monitored during non-steaming and stesming conditions. Sampling requirements on the- -

detection of a primary to secondary leak have been established, as well as administrative limits on leakage.

The combination of analysis of tube ruptures, procedure improve-inent and training improvement give assurance that operators can _

i safely respond to a primary to secondary leak.

e e

6 O

O

_ 99 _ ,

e s **

+ , . , _ . , . , _ _ _.

e. .. .s . .

ENVIRONMENTAL IMPACT XI.

A. Introduction The impact of operating TMI-l wi.h primary to secondary leakage was evaluated. Offsite dose estimates were determined at several leakage rates, using actual anticipated failed fuel percentages.. These calculated estimates have oeen compared .with Appendix I technical specification requirements. The-effect of leakage on onsite exposure was also considered and found to be small. Exposures associated with steam generator work leading to return to service are also discussed below.

3. Offsite Dose Estimates The maximum primary to secondary leakage rate at which TMI-l might operate can not be determined without operating ex-perience. The of fsite consequences of such operation will be dependent on the failed fuel percentage and actual plant system and environmental conditions. Dose will be determined during operation by monitoring. The technical specifications for IMI-l incorporate the Appendix I offsite dose limitations. If offsite doses approach these limits due to primary to secondary leakage, it will be necessary to shut the plant down to look for leaks. .

For planning purposes, two calculations were performel using .

different hypothetical leaks races,1 lbm/hr and 6 gph. I lbm/hr is the repair leak rate goal. 6 gph was selected as a leak race with which similar plants have operating ex'perience, and which is similar to the leak rate change at which admin-istrative procedures TMI-l would shutdown to look for leaks.

Both calculations assumed 0.03% failed fuel, which was seen at TMI-l at the end of cycle 4. Results of the two estimates are compared in Table XI-l to Appendix I technical specification limits. Source terms and methods for calculation can be found in Raference 11 and 54 o

O s

- 100 -

.e r e .

Table XI-1 Hvoethetical Maximum Individual Offsite Dose (1)

OFFSITE DOSE AFD FRACTION OF APP. I LIMIT 10 CFR 50 Source 1.0 LBM/HR 6 GPH App. I Dose  % of App. Dose  % of App. Limit (ar/yr) I Limit (ar/hr) I Limit (mr/yr)

Iodine & 7.68E-2 0.5 4.61 31 15 Particulates Noble Cases

• Gasmsa 5.68E-2 0.6 2.74 27 10

• Beta 6.9 6E-2 0.4 3.33 17 20 Liouid Effluent

• Whole Body 1.04E-3 0.1 4.88E-2 1.6 3 (adults)

• Liver 1.52E-3 0.1 7.2SE-2 0.7 10 (teens)

(1) Based on 80% of the plant capacity factor.

As can be seen in Table XI-1, offsite doses are not expected to approach Appendix I limits due t'o prinary to secondary leakage.

However, monitoring of actual offsite exposure will be used to set leakage limits which prevent exceeding technical specifi-cation limits. .

C. Exposure Estimates The man-rem exposures measured on self-reading dosimeters for the completed OTSG repair program were 1233 man-rem. Individual activities for the steam generator program are presented in Table XI-2, along with exposures for each task.

O e

9

- 10 1 - -

P m

- -- j

r Table XI-2 Excesures from OTSG Proitram (based on self-reading dosimetry)

I Task Man-Ram

1. RCS inspection 12

2. Eddy Current Testing 39

3. Pre-Repair Testing 5

4. Tube Sample Pulling Plug.-ing 120 and Stabilization

5. Plugging and Stabilizacian

a. W plugs 37

b. Stabilization 138

c. Post-testing plugging, 85 stabilization and plug repair

6. Kinetic Expansion

a. Pre-expansion Preparation 16

b. First Pass Expansion 168

c. First Pass Debris Remcval 132

d. Second Pass Expansion 167 -

e. Second Pass Debris Removal 75

7. End Milling '125

8. Cisan up

a. Flush -

la

b. Soak and clean 5

c. Individual Tube Cleaning 71

9. Testing

a. Drip Tests 12

b. Subble Tests 5

c. Final Inspection and Turnover 3 Total 1233 Man-Rem

~

- 102 - .

O

'c .

,I ' < ,

Table XI-3 Radiation Fields at TMI-1 Location L

Upper and Lower Heads 1.3 R/hr.

4 Maaway 0. 13 R/hr.

Tent 0.01 R/hr.

Low Zo.se 0.001 R/hr. -

' Operating with the limited leakage associated with the repaired ~

joint was also considered. The additional exposure is expected to be low. Normal health physics and ALAR practices will iden-tify any additional radiation areas and minimize related worker exposure in operating with a primary to secondary leak. The '

largest sources of radiation exposure are expected to be the Powdex Demineralizar vessels. Associated contact radiation levels have been calculated for a 6 gph leak to range from 0.7 mr/hr for one day filter operation to 5.7 mr/hr for 15 day operation. In addition to these estimates, experience at sini-lar plants operating with small primary to. secondary leakage has been exposure increases of less than one man-rem per year.

Based on this information, the annual exposure at TMI-l is ex-pected to increase by less than 1% due to leakage.

D. Samulins and Monitoring .

Appropriate monitoring and sampling of all waste streams will be conducted per established Guidelines. Modifications will be

, installed in the Turbine Building to provide radiation and con-i camination control and effluent release control / accountability.

These modifications will consist of Powdex and Turbine Building sump painting, and liquid monitors to measure activity during l operation.

. E. Conclusions The operation of TMI-l with small primary to secondary leakage is not expected to cause offsite doses nearing the Appendix I '

Technical Specification limits. Final verification of estimates will be by monitoring during operation. Should monitoring in-dicate that primary to secondary leakage is causing offsite doses to approach these 1Laits, steps will be taken to reduce the activity contribution. Exposures onsite as a result of primary to secondary leak. age are expected to be minimal compared to prior plant experience. Exposure during the investigation of the steam generator problem, the repair, and testind afterwards,

was measured as 1233 man-rea on self-reading dosimeters.

l i

1

- 103 -

l XII. TECHNICAL SPECIFICATION COMPLIANCE

^

This safety evaluation demonstraces that the TMI-1 OTSGs are operable per T.S. 3.1.1.2, and have met the surveillance conditions -

for operability given in T.S. 4.19, or as defined in the related'

'T.S. change request. Specifically, the eddy current testing required by Table 4.19.2 has been complaced and all tubes found by ECT to have indications below the top of the repair which either

• exceed the repair limit or to contain throughwall cracks have been repaired or plugged. As further described in Appendix A, the bubble test results'in OTSG "B" showed no observable leaks'and those in

, OTSG "A" showed only very small amounts of leakage which we have concluded are too small to declare the tubes unserviceable or to warrant further action prior' to proceeding.

In addition, the following technical specifications were evaluated in light of the selected repairs. Operation with the repairs in place was found to be acceptable in each case.

SER T.S. Subiect Topic for Evaluation Reference 1.5.6 Heat Ba1. Calib. Flow asymmetry Section VIII 1.6 Def'n, Quad. Pwr. Tilt Flow asymmetry Section VIII .i 2.'1 Fig. 2.-1,2.1-3 Flow vs T Flow, flow asymmetry Section VIII 2.3 Fig. 2.3-2, Table 2.3-1 -

Nuclear Overp. Flow . Section VIII ,

3.1.1.1.a Permissible pump. comb. Flow Section VIII 3.1.2 RCS heacup-cooldown

• Stress vs. Temp. change: Section V and assumes vessel as limit II 3.1.4 RCS activity Laakage Section X 3.1.5 RCS chemistry Further attack in conj. Section'IV w/resid. S or following -

S distress 3.1.6 RCS leakage , Leakage Section IX 3.5.2.4 Quad Pwr. Tilt Flow asynnetry Section VIII 3.5.2.5 Quad. 341. Flow asynnetry Section VIII 3.13 Secondary activity Appendix I w/ leakage Sect. ion X 3.22 Appendix I Leakage Section X 3.23 Appendix I Leakage Section X 4.2 RCS:ISI Testing of RCS Components Section II.E 5.3.2.1 RCS code req. Repair qualification Section V O

d

- 104 - -

i.

T_~

. . .js .

XIII. SID0fARY AND CONCLUSIONS The previous twelve sections along with the references and appen- "

dices associated with this safely evaluation provide a broad-ranging E discussion of the adequacy and safety of the TMI-1 OTSG repair and '

the ability of the plant to be safely returned to service. The main ~

points associated with deteenining that the plant is safe to operate can be summarized as follows: ,

1. Knowledge of the failure scenario is sufficient to provide a firm technical basis for OTSG repair decisions, insure that the environment for such a damage mechanism is not established in the future, 'and provide a technical basis for assuring safe performance of the OTSG tubes below the area of the new joint.

2. Evaluation of operation of TMI-1 with small primary-to-secondary leakage has confirmed that Appendix I Technical Specification considerations are satisfied.

' 3. All tubes with no defect indications below an elakation 8 inches above the lower face of the upper tubesheet (UTS+8) have been adequately repaired by the kinetic expansion process. The kinetic expansion process qualification program provides assurance that a load carrying rad leak ilmiting joint

' acceptable for safe operation has been formed.

4 The performance of the CISG/RCS considering the tubes to be plugged is satisfactory and no power limitations are required.

Tubes with defect i.ndications below the UTS+8 elevation will be removed from service by approved plugging methods. The OTSG/RCS performance uith these tubes plugged has been evaluated for both normal operating and emergency conditions.

5. Circumferential defects snailer than the threshold detectability of ECT or less than 40% through wall are acceptable. Frsecure Mechanics Analysis. of circumferential tube defects has been conducted. The analysis identified crack geometries which would propagate from mechanical loads during both normal operating or accident conditions. Geometries which would propagate to a double ended tube rupture during 40 years of operation or during

' an accident were characterized as " unstable." The results have been compared to the ECT sensitivity for various geometries of circumferential defects. Th'is comparison shows that the G7UNC 100% ECT inspection of the TMI-l OTSGs was sensitive enough to find " unstable" defect geometries.

6. The examination of Reactor Coolant System (RCS) has confirmed that c'.. aggressive environment that caused damage to the OTSG tubes and to the PORY did not damage the remainder of the l reactor coolant system. The RCS examination results provide the .

- 105 -

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

_I. - yY' -o ..

• =

t

' basis for concluding that there are no corroded components in service which will preclude the RCS from functioning properly

, and supporting safe operation of TMI-1.

7. Analysis of design basis and higher primary-to-secondary leak rates confirms that the operating and emergency procedures are technically correct. The procedures provide adequate basis for training the operators to respond to normal and emergency primary-to-secondary leakage. The procedures have been modified to improve even further operator guidelines and handle greater i than design basis accidents.

8. Steam generator testing together with long life continuing

'7 , laboratory testing will provide confirmatory data on repair stability and the absence of new high velocity cracking. The steam generator testing will be completed with essentially zero decay heat power and poses no safety risk.

Conclusion

.; _ In conclusion. TMI-1 can be safely returned to service once the repairs and other activities discussed in this safety evaluation report are completed. This conclusion is based on sound analy-tical and empirical data developed by GPU Nuclear Corporation during the OTSC repair program. The scope of technical evalu-acions has been broad based and the involvement of numerous -

l' independent technical experts has been extensive, throughout the TMI-l OfSC repair program. The methodical, technical approach

to evaluating the var'ious aspects of the problem in order to

make the best and safest decisions provides a high degree of

confidence that TMI-1 steam generators can be safely operated. 1 l

i t

9 i

106 - -

• i

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

, , p .s

• a 5

APPENDIX A PRECRITICAL AND POST CRITICAL TEST PROGRAMS I. INIRODUCTION The TMI-1 restart test program has been planned to provide a

. deliberate, methodical, well planned verification of proper

, installation and performance of the steam generator joint-

' modifications, to verify conformance with design and licensing bases. . Cold and het procritical testing have been combined with '

p 'the power escalation program to create a progressive testing 4, program for the OT1Gs. The program includes the following:

}'

j . Verification of the adequacy of the OTSG Tube Repair Program by pre-service leak testing of individual tubes

. Verification of the adequacy of the OTSG Tube Repair

' Program by operational leak testing and on-line tsonitoring throughout the test program 4

. Verification of the adequacy of the repaired OTSG tube joint and tubing in service to carry loads under normal

, operating transient conditions.

1 .

. Verification of acceptable system readiness and plant operation with new and modified plant operating, surveillance, emergency, and abnormal procedures Performance of sufficient modified system / plant steady state and transient operations to provide' operator training and familiarization with modified system / plant response throughout a range that is likely to be experienced during the design life of the plant The scope and chronology of testing planned to meet these goals are l discussed below. The test sequence is susunarized in Figure A-1.

II. PRECRITICAL TESTING i

Both hot and cold (pre-service) testing will be performed prior to criticality.

• The overall objective for the pre-service prograta is to demonstrate the success of the repair by providing adequate assurance of the post-repair primary to secondary structural and leak tightness integrity of the steam generators.

- 107 - -

+

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

FIGURE A.1

• TMI-1 Restart .-

Test Progran! ,

Including OTSG Repair ..

COMP 1ETE OTSG OTSG

'i OTSG DRIP BUSSLE RESTORE FILL & BCS

- ~ -

ECT -

VENT HO l REPAIRS TEST TEST RCS -

RCS CLE P

+'

SEC. PLANT READY TO -

i SUPPORT

! ISEATUP i

4 ESTABLISH OTSG/FW llEATUP OTSG N AH FOR OTSG COOL HOT CllEMISTRY -. _ llOT TEST - FUNCTIONAL MMR DOWN TO SUPPORT llOTTEST AND SOAK TISTINp PilVSICS 45 IIEATUP t .

'3 4 NATURAL POWER RETURN OTSG

• gy CIRCULATION ESCALATSON _ TO 100% EDDY TESTING TESTING

- 90 DAYS TEST

• FormaalManagersat Revlew I I 48 % 76% 100%

4 *

, e

FIGURE A-2 TMI-1 GTSG Tube Repair Procritical -~

Test Program .

COMPLETE PERFORM PERFORtA EVALUATE pgLL Atae

~ ' ~

PA OTSO OTSO ausatE ST

~

TEST - TEST 4, ACS

/

(2 DAYS $ (2 DAYSl (3 DAYS $

CONDUCT RE-ESTASilSH H/W RCS

& ACS H O RCS CONDUCT THERMAL CLEA

_ 532* F/2155# -

OPERATIONAL CHEMISTRY FOR OTSO SOAK '

130*F/300# FOR H/U TUBE TESTS LEAK TESTING 532*F/2 tE5#

12 WKS) 12 WKS) 12 DAYS) (2 DAYS) It WK)

RCS THERMAL RCS m

COOtOOWN is/u RCS COOL DOWN H/U RCS

@ S0*F/HR.

532*F/2155# - 8 FOR 2 HRS. 532 F/ 156#

~

O " III" ~

FOR 2 llRS.

t I

112 HRS) Il1 DAYSI (12 HASI .

THERMAL RCS

. SOAK ~ COOLDOWN TO -

TOlWT E32*F/21s5# 130*F/300, REVIEW k TEST 4NG 111 DAYS) l12 HRSI 5

O

e. ,

1 s

The specific post-repaired features of .the steam generators to be tested include the tubing kinetic expansions, the Westinghouse roll i plug installations, and the B&W weld plug and explosive plug l

installations. In addition, the testing will simulate operational .'

loading of partial through wall defects in tubes remaining in service.

The sequence of cold and hot precritical tests is shown in Figure A-2.

A. Pre-service Test Program i

1. Cold Testing Tests conducted prior to heat up included the following:

a. Drip test, whereby the primary side is drained completely, the secondary side pressurized to 150 psig, and water leakage from tube ends observed in the lower head. This method leak tests plugs installed in the icwer tubesheet, the tubing expansicas, and the full length of tubing remaining in service. .

b. Bubble test, whereby the primary side is drained tc a few inches above the upper tubesheet, secondary side water  !

level is lowered and pressurized to 150 psig.' Kinetic cube expansions, cubing above the lowered water level, and upper tubeshee,c plugs are leak tested by ' visually observing l gas bubbles in the upper head. Based on experience at '

similar plants, this method is expected to have a lower limit of detectability of about 0.1 gal / day per tube (leakage at normal operating pressure and temperature).

c. Baseline eddy current testina, to verify that the kinetic expansion has not changed the condition of tubing, and to provide a baseline for post-critical ECT. Precritical ECT is discussed in greater detail in Section III with the post-critical program.

As a result of findings in the drip, bubble, and ECT tests, some additional repairs were conducted. Following these .

repairs, the drip and bubble tests were repeated.

The results of the cold complete testing program are as follows:

a. Initial Drip Tests To begin the cold testing program, a drip cost was conducted twice in each steam generator. The first test identified 10 cubes in "A" and 21 in "B" as leaking. The ~

second identified 6 in "A" and 5 in "B". 2 in "A" and 5 in "B" leaked in both tests. This inconsistency between the two tests is thought to be due in part to condensation in the upper head, in isolated cases to wet felt plugs held in blocked tubes, and to the difficulty and uncer-tainty in observation. Because drip races were low, in

- 108 -

g n i

sost cases less than i drip / min., identification of the dripping tube was difficult. The. drip test is used only_

-se an indicator of a need for further investigation. The determination of the acceptability of a tube was based on

• the bubble test and ECT results.

Several plugs were identified as dripping. Welded and explosive plugs were repaired or replaced. Mechanical

g. plugs were evaluated and rerolled or replaced as necessary.

f

b. Initial Bubble Tests The bubble test identified 10 cubes in "A" and 13 cubes in "B" as lenkers. Of these 23 tubes, 4 tubes in "A" and 11 tubes in "B" had already been identified foe further in-vestigation by the drip test. By placing a tempcrary stopper in each bubbling tube at the lower face of the upper tube sheet (US+C) and at US+14, the axial location of the leak could be further identified. All of the leak.1 were found to come from below the top of the qualification length of the kinetic expansion joint, with the passible exception.of one tube which bubbled slightly, and was not

. investigated further during stopper testing because the bubbling rate was too low for visual tube identification when the stoppers were placed. No joints were found to leak.

• Several plugs were identified as ' bubbling; Welded plugs were,tepaired or replaced. Mechanical plugs were ,

evaluated and rerolled or replaced as necessary.

c. Eddy Current Testins i

. In addition ~ to the baseline eddy current testing, each of the tubes identified in the drip and bubble test was investigated using eddy current techniques. Using the

.540 differential probe, eddy current signals have been  !

found in the unospanded portion of all but one bubbling tube. This particular tube (OTSG B 80/45) is the tube  ;

discussed above which bubbled slightly but was not '

i

} investigated.further during stopper testing. Of the 10 tubes in "A" and 10 in "B" which dripped but did not b.abble, only 1 tube in "3" has a pluggable eddy current indication (greater than or equal to 40% through wall).

ECT also identified an indication in one dripping tube in "A" of less than 20% through wall at UT3-10.

4 The 23 tubes have siso been examined using the 8 x 1 I absolute ECT probe. A11' signals are on 1 or 2 coils only. '

4 corresponding to a range of are lengths of a minimum ,

.i I

- 109 - ..

4

w 4 .

sensitivity of .024 inches to approximately .413 inches.

The voltage data for most 2 coil defects is not balanced for both coils, suggesting arc, lengths closer to the lower end of the range than the higher. Based on this informa-

• tion, GPUN considers that the indications always existed but are now somewhat more visible.

GPUN has plugged or plugged and stablised the 23 bubbling or dripping tubes with ECT indication greater than or equal to 40% through wall. The tubes with less than 40% ~

ECT indications were left in service and added to the augmented ISI list or reinspection during subsequent.ECT, and will add to the data base. GPUN has evaluated the 23 pluggable ECT indications (in dripping or bubbling tubes) and determined the following: -

1) All leaking defects are between UTS-02 and UTS +07.

2) Four of these defects can be discerned on previous (1982 ECT) inspection records but were missed apparently fue to analyst error.

3) Fifteen of these defects have small amplitude signals which in the 1982 examination were masked by UTS noise.

4) Four of the signals are at the' entry point' of the -

upper tubesheet, and were previously masked by probe saturation on tube sheet entry signal.

In addition to the ECT described above, GPUN has conducted '

L the post-repair pre-service eddy current examination described in Table A-1. The pre-service inspection data

, was obtained as a very sensitive benchmark against which

, to coopers siellar ECT inspections proposed to be obtained *

! after 90 Full Power days. Such comparisons may provide further information confirming arrest of crack growth in an operating generator. The results of these pre-service tests are shown in Table A-2.

O

. tio .

e

e e

Table A-1 ..*

Post Repair ECT Inspection Summary I ,

90 F.p. Days Cycle 6 Scops Fre-Service inspection Inspection Refueling Outage Remarks l

1(a) All tubes 40E (i) 0.540" 8.D.M.C. Repeated Repeated-T.W. below the Full Length ~

qual. length (ii) S a I to confirm S.D.

in both OTSC's Ind. & circianferential entent t

1(b) All adjacent tubes

• Wear baseline exam. In Repeated Repeated *Use ECT technigene proven to 10 selected area of interest. .(using by laboratory testing to  ;

! plugged unsta- ECT probe demonstrated be adequate for wear exam.

l bilised tubes adequate for wear examina-with defect tion in area near defect.) ,

i in 15th, 10th &

let spans (10/Es.-

OTSC) s C 1(c) All adjacent tubes *Emanine for 0.D. wear Repeated Re~peated *Use ECT technique proven to 10 selected baseline in the 16th span by laboratory testing to 1 plugged unsta- (Using ECT probe desen- be adequate for wear exam.

1 bilised tubes strated adequate for j per each wear examination in area OTSC in the near defect.) 7 Periphery l 1(d) 11 tubes had S x 1 absolute for 6" q.L. Repeated Repeated 1 new indications

• l

) in the 6" qual.

j length .

l  !

} 1(e) 50 cubes in the 0.540" 8.p. high gain in- Repeated Repeated l high pluggishg spection for full length - ,

density area -

j per each OTSC i

i o

i 4

Table A-1 (Cont'd)

Post Repair ECT Inspection Summary

, 90 F.p. Days Cycle 6 Scorg Pre-Service Inspection Inspection Refueling Outage Remarks 1(f) 31 of tubes in 3 x ! examination in UTS Repeated TBD addition to 1(d) in each OTSC from top of 6" qual.

length to lower .

surface of UTS 3(g) All adjacent tubes

• Wear baseline examination Repeated Repeated *Use ECT technique proven to 5 plugged at same elevation as defect by laboratorf testing to tubes per each indication in the plugged be adequate for wear exes.

OTSC with 3V tube indications in '

lower part of '

OTSC's e

p2 3% of tubes re- 0.540" S.D. high gain Repeated Repeated

" saining in full length inspection s

service per each See Note 1.

OTSC in addition to above 1(a),

(b). (c) (d), .

(e) & (s)

Note I: 'By doing pre-service examination in above categories 1(a) & 1(e), if ~ no new defect or no indication of defect growth from previous 0.540" high gain data, such data obtained in 1982 may be considered representing condi-tions af ter expansion. No need to perform item (2) for the preservice baseline.

e e

G e

Q

,- , , . 9 .

Table A-2 Post Repair ECT Pre-Service Baseline Insoection Results

- Probe Number of Tubes Scope from Table A-1 Type OTSC A OTSC B

1. 4 80 tubes /0TSG .540 SD 11 1 (items Ib, c, e, g) indications 340% TW *

2. All 440% TW previously .540 SD 1* No change identified to be left in service (item la)

3. "3% of unexpanded tube 8x1 3** O within UTS below 6" joint (verified 2 40%

TW using .540)

(Leems if, d) 4 <3% of 6" joint 8x1 9*** 21*** '

(items if, d)

• On one tube the phase angle for an OD indication at TSP #6 shifted such

. chat the previous 35% through well indication was evaluated as 60%.

The Sul data did not change (1 coil,1/2 volt). The phase shift was apparently due to interference from TSP #6. GPUN evaluated this crack as unchanged, but because of the uncertainty in through wall excent, the tube will be plugged.

• • The 11 indications shown in line 1 can also be detected by the Sx1 but are not included in the table twice.

• • • These low voltage indications are not unanticipated since the 8x1 probe is more sensitive than the .540 probe in the upper tubesheet. Noise levels in the joint area are too high to use the .540 probe for latestigation of through well extent. These tubes were dispositioned for plugging based on bubble test results.

4 4

- 113 -

e, . 9_ <

The tube indications in line 1, of Table A-2 (11 in "A" and one in "B") were evalated in the pre-service inspec-tion as all greater than 40% through wall with predomi- -

nantly_ a low voltage amplitude (equal to or less than one volt). The one indication in OTSG "B" is located axially at UTS +2 and the 11 indications in'OTSG "A" are located between TSP 13 and UTS +6. These 12 indications were

$' further e===Ined by the Sul probe and were all found to be equal to or less than two coils in circumferential extent -

with 6 of the 12 evaluated as one coil. During 1982, every unplugged tube in each steam generator was examined using the sensitive high gain differential probe. ~ We have esamined the 1982 high gain differential probe baseline inspection result tapes and determir.ed most of the pre-service 1983 indications are within the 1982 cape noise level but can be considered identifiable given the precise

'.ocation information from the.1983 pre-service inspection results. Therefore,' the post-repair pre- service ECT

-eeules are considered consistent with the 1982 record i.nspection, and the 1982 inspection is valid as the required Technical Specification 100* inspection. No further ECT was conducted.

In addition to the 12 tubes of greater than or equal to 40% chr9 ugh well indications reported on line 1, a total of 6 tubes were called at less than 40% but at equal to' or greater than 20% through wall in OTSG "A" (one evaluated at 25% TW, five at 20% tif), and one in "B" (20% TW). All of the less than 40% through wall calls are of such low voltage and such uncertainty in phase angle that we cannot now conclude with any confidence whether they represent intergranular attack or surface anomalies. We have con-cluded that they are not of a size to warrant plugging.

With regard to the Sul inspection results identified on line 3 and line 4 of the above table, these are not unanticipated since the Sul probe technique is more sensitive than the standard dif ferent probe in the upper tubesheet region. Prior developmental testing, conducted as part of the espansion joint qualification program and documented in Section V.D.1 did identify that tube expansion in the tubesheet (or close to the expansion region) had the potential for making small defects in the tubes more visible by ECT.

We have plugged all tubes identified on line 1 and line 3 of Table A-2 having eddy current indications from the pre-service inspection of 40% or greater even though we are not sure at this time of the accuracy of the evaluation.

O e

114 -

9

-e .

q ,

6 ~ . m This dispositioning is consistent with the tee'hnical spec-ifications and with the plugging criteria in Section VII.

After the 1982 record ECT Inspection, any indication in the 16th span will be plugged as a precautionary measure.

• GFUW does not plan to plus additional tubes with indica-tions in' the 16th span called as less than 40% through well (six of seven tubes identified above) because of the .

uncertainty associated with interpretin~ g such a small ECT signal. These tubes were added to the augmenced ISI list for reinspection during subsequent ECT, and will add to the data base.

'd . Follow-up Testing Following repair of the residual tubes identified above, additional drip and bubble tests of both OTSG's were conducted on June 17 and June 26. These-tests of OTSG "5" showed no observable leaks. The drip test on OTSG "A" identified five plugs (four rolled plugs'and one explosive plug bottom tube sheet) and one tube with drops of water clinging to the end. No drops were seen to fall free over the 30 second obserration period fran any of these six tubes. This particular tube (unplugged) had been inspec-ted by ECT in the previous month and dispositioned as no detectable defects. This tube was not plugged.. The leak-age from the plugs was extremely small and since futute operation may seal the plug leakage with time, no further repair action regarding the dripping plugs was taken.

The bubble test on OTSG "A" identified seven plugs (one welded and six rolled) and three unplugged tubes from which some gas bubbl,es were detected by visual obser-vetion. The welded plus was repaired. This was the' first bubble test on this welded plus since it had been

, repaired. All of the tubes and rolled plugs passing nitrogen in OTSG "A" had very fine streams of tiny bubbles which did not cause surface disturbances of the water layer. This amount of bubbling is less than that observed in previous testing. Due to the small amount of leakage, no further action was taken regarding the three tubes and six rolled plugs. The results of pre-critical hot func-tional OTSG testing will be used to disposition these tubes prior to critical plant operation.

Throughout the RCS cleanup, we monitored for primary to secondary leakage by examining the secondary side for activity and boron. No crit OTSG. Tritium was 4.4 x 10'jun was in uci/mi detected the RCSinand either has a minimum detectable activity of about 5 x 10-6 uci/al.

If one assumed equal sizing in the OTSG secondary .

~

- 115 - .

e e

e

7.-

s. ., ./ =

side, the leak, if any, at 300 poi differential pressure was less.than 1.1 x 10-3 spa in each generator, based on no detectable tritium. Similarly, boron in the secondary was undetectable, confirming that the leak, if any was -

less than 3.1 x 10-3 gym in each generator. There were several Cesium-137 measurements above the minimum detectable activity of 1 x 10-7 uci/mi. Since Cesium-137 was 1.6 x 10-3 uci/mi in the RCs, they correspond to a primag spa to secondary leak rate of ejust to in OTSG A and 1.9 x 10 or less than 1.4 x 10 _

spa in OTSG B.

Our overall evaluation of the post repair inspection results, including the pre-service inspection baseline ECT testing, is as follows:

1. With the possible exception of four tubes, we have identified no kinetic expansion joints which leisk nitrogen from above the 6" qualifiction length at 150 poi delta P.

2. Existence of small are length 100% through wat'. defects found by bubble testing is consistent with discussion in Section II.

3. Existence of naall are length partial through wall defects -

found by ECT is consistent' with discussion in Section IX.

4 8x1 ECT probe results from within the tubesheet region are consistent with prior developmental results and are, as expected.

5. The total evidence available continues to support a conclusion that cracks are not growing and that new cracks are not occurring.

2. Not Testing The sequence of tests to be performed during the steam generator hot testing period is discussed below. Conventional leakage monitoring techniques planned for post-critical operatione (Section X) would not be effective during this test period. The very low concentration of fission products in the RCS after a 4 year outage are insufficient to provide a seasurable indication on a continous basis for the leak rates to be detected. Therefore a tracer will be used in the primary system. Use of the Kr-85 as a tracer has.been evaluated in a separate safety evaluation (Ref. 65). The het testing leakage monitoring program is summarized in Table A-3.

- 116 - -

~

. e: . .J When the scess generator hot functional test sequence is complete, 'a management review of the results will be conducted. If the results are satisfactory, the plant will proceed with the normal preeritical hot functional program.

and subsequently with critical operation. The test sequence

-is-as follows:

a. Normal Heatup Tb RCS temperature and pressure will be raised co 532*F -

and 21554 in accordance with normal operating procedure.

9 h

9 4

e e

J l

l i

i l

I 117 - -

, s.~ -

, . 4 , ,

. ~

Table A-3 Hot Testing Leakage Monitoring

- )

l Parameter Sjisole Location Frecuenev@ Sensitivity Krypton-85 RMA-5L (offgas monitor) GontinuousY 0.18 1 0.05 GPHY RMA-5L (grab sample) 4 Hours (steady 1+.4GPHP state) 1 Hour (cool-

i. downs)

Boren OTSG , Daily 1.6 GPHM Tritium OTSC and Hoewell Daily 11 GPHY Gamma OTSG and Hoewell Daily 4 Isotopic RMA-5L (Grab sample) 1 Hour (Cooldowns) 4 Analysis RMA-5L (Grab samole) 4 Hour (Steady State) ly RMA-5L (Grab sample) 2 gph increase calculated from RMA-5L readings Iy Notes:

Y Counts / minute to be recorded on a scrip

, chart or logged every 15 minutes during cooldown and every 30 minutes during senady state.

Y Assumes all Scron equally mixed in .

OTSG secondary side water and 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> of constant leskrate.

M Assumes Tritium equally mixed in hoewell, condensate, feed and OTSG.

Based on 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> of constant leakrate.

Y Sensitivity to be determined during testing. RCS cleanup has reduced many isotopes in RCS to less chan or barely detectable levels (including Casium-137).

k Assumes 20 cfm discharge flow from vacuum pumps and 4.5 x 10-2 uei/cc Kr-85. _

Some additional uncertainty in sensitivity expected would be due to gas flow variation in vacuum pumps or 'a lower Kr-85 concentration. , ,

4 Frequencies subject to change based on engineering evaluation of the data during leak race casting.

O

- 118 -

3- .: 34 x j b. Doerational Leak Test This test is required by technical. specifications whenever work has been performed in the reactor coolant system. .

The pressure in the primary system will be raised to approximately 2285#, creating a differential pressure between the primary and secondary ]f approximately 1400d (maximum normal opera.ing differential pressure). This is expected to be the maxi-i= differential pressure experi--

enced by the repaired tubes.

c. First Thermal Soak Conditions will be allowed to equilibrace at 532'F and 21554 for approximately one week, to provide baseline leakage data and to allow monitoring of leakage for trends.

d. Normal Cooldown Transieg A controlled cooldown will be conducted according to normal procedure, at approximately 60*F/hr for approxi-mately three hours to 350*F. Tube to shell delta T will be maintained at less than 70*F to limit the stresses placed on the tube. Laakage will be monitored throughout the transient.

e. Second Themal Soak -

The RCS temperiture and pressure will be returned to 532*F and 21554, and held there for eleven days. Leakage data vill be obtained for comparison with the earlier thermal soak, and to monitor for developing trends.

f. Accelerated Co~oldown A controlled cooldown will be conducted at close to the maxLaum rate permitted by technical specifications, at. l approximately 90*F/hr for approximately two hours. This transient is expected to apply greater loads to the repaired tubes than the earlier cooldown. Tube to shall delta T will be monitored to determine stresses but will

} be permitted to increase as high as 140*F. Leakage will

[ be monitored throughout the transient.

I

g. Third Thermal Soak i The RCS comperature and pressure will be returned to 532*F .

[ and 21554, and held there for approximately eleven days.

l Leakage data will be obtained for comparison with the i

earlier thermal soaks, and to monitor for trending. ,

- 119 - -

-~

I - . . ..

.o . [y

h. Normal Plant Cooldown  %

The plant will be cooled down to 130*T using the tube During this rupture procedure for training purposes. C-cooldown cube to shell delta T 4i11 be limited to 70*F, and fuel pin compression limits will be observed. The

• plant will be maintained at 300#, 130*F pending management _

review of the OTSG hot functional results.

i. Flow Rate Testing _

Because tubes were plugged during the repair process, the RCS flow race must be missured after repair to verify compliance with technical specification. This cast will be performed when the plant is at normal operating temperature and pressuse. l

-III. POST-CRITICAL ptANT TESTING The deliberate, machodical approach to testing will be maintained The normal program has throughout the power ascension 1.rogram.

been lenschened to permit ample time for leakage monitoring and trending, as well as for familiarization with plant performance following the modification. After the power escalation program is complete, and again at the next refueling outage, special inservice inspection programs will be conducted to look for the effects of "

operation after the repair on tubing and components. ,

A. Power Escalation Testing Power escalacion testing is expected to serve two purposes in testing the steam generators. First, the slow progression from power level to power level will, permit monitorian of ,

Second, possibly changing plant condicions such as leakage.

several of the tests already planned for start-up casting will Leakage monitoring apply loads to the steam generator tubes.

before and after the Lenasients will provide informacion on the condition of joints and cubing.

l The power escalation program is discussed chronologically below. Transient tests of interest are noted for each power level.

f i 1. Lower power Testing i

Following hot functional testing and initial criticality, After the normal plant zero power test will be conducted.

this approximately one week program, low power natural circulation testing will be performed.

120 - * -

i a

7

, ,; .. y -

This test verifies the tuning of the Integrated Control j-

' System (ICS) to maintain preset OTSG 1evels under loss of It main feedwater and natural circulation conditions.

also verifies proper response of the EFW system as well as '-

the establishment and maintenancu of natural circulation under varying conditions. Testing will be conducted at approximately 3% of raced thermal power to simulate the decay heat load that would correspond to significant core burnup. EFW initiation is expected to stress the OTSG tubes. This test will also verify that plugged tubes have ~

no effect on establishing natural circulation flow.

A mana. . .ent review will be conducted following natural circulat,8.on testing. Satisfactory OTSG and plant performance will be necessary to increase power.

2. Operation at Less Than 50% Power Power escalation testing will be conducted for several days each at 15%, 25%, and 40-48% power. The 40-48:

At plateau testing will include a loss of feedwater test.

a power level of.approximately 40-48%, both main feedwater pumps will be tripped. All three emergency feedwater

• pumps will start automatically and OTSG level will be I controlled at approximately 30 incher by emergency feed-water. The use of EFW will cool and stress the OTSG cubes.

The RCS Overcooling Control test will demonstrate that the control room operator can properly throttle EFW flow to ,

prevent overcooling of the RCS following a loss of RCP's with OTSC level initially at 30" on the startup range.

The effects of this transient and of operation to date will be monitored during a one month soak at approxima'tely 4

48% power. The power level was selected as the minimum

permitting two main feed pump operation. Some operator Prior to training will also be conducted during the soak.

increasing power, another management review will be held to evaluate test results and plant performance during the soak.

3. Power Escalation to 100% Power Testing at 75% and 100% power will conclude the power escalation program. The 75% plateau will include approximately five days of testing followed by another one month soak to observe plant performance. Power will then be increased to 100% power for approximately one week of testing. Testing of interest at this plateau include the 100% turbine generator trip. ..

1 I

' - 121 - ..

r- .

e. '

h 4 .;

G e A management review of plant performance will then be conducted before the plant procedes with normal power operation.

5. Eddy Current Testing Either 90 calendar days after reaching full power, or 120 calendar days after exceeding 50% power, the plant will be shutdown for the performance of addy current testing. Test results will be compared with a baseline taken prior to restart to verify the lack of defect propagation during normal operation. The baseline, 90 day, and next refueling ECT examinations are discussed in detail in Reference 56 and summarized in Table A-1. Testing listed for the preservice baseline wase performed after repairs were complete to pro-vide evidence that no changes have occurred since the 1982 100% record inspection. Results are dis ussed in Appendix A.II.A. Any new or changed ECT indications found in the subsequent inspections will be evaluated. If evaluation shows they are unacceptable, they will be treated as new defects a

with subsequent actions taken as required by Technical Specifications.

C. Steam Fittings ISI Program GPUNC will inspect the main steam line fittings at IMI-I for potential water droplet erosion. The following criteria were used (in descending order of erosion probability) to select monitoring data points. .

1 - Fittings close or adjacent to the OTSG

' 2 - Fittings of 90* configuration 3 - Fittings from the "A" steam generator

• Since there are more cubes plugged in the "A" OTSG than the "3", the "A" steam lines may have a higher probability of I

erosion. Therefore, inspections will be made of steam lings from the A OTSC. Based on the above the following fittings were inspected before start-up and will be reinspected at the g next refueling outage following restart.

O 6

- 122 - ,

" " ' * ' ' 'Wwawm ,_.-_ , _ _ . _ _ _

o e V- V .

9 o

. c ,

P Table A-4 Steam Line Fittings Inspection "A" OTSC

'Ites Qtv. Description 1 2 First 45* Ells after -

OTSG exit

'This inspection will be done using ultrasonic testing techniques for measuriar pipe (fitting) wall thickness according to an adaptation of ASTM E 797-81 " Standard Practice for Measuring Thickness by Manual Ultrasonic Pulse - Echo Contact Method."

The ultrasonic data density (number of data points) of the pipe wall surface will correspond closely to that already done on the turbine extraction piping at TMI-I.

Future inspection frequency and scope . vill be determined 2 feer the results of the next refueling outage's inspection have been evaluated.

h 6

9 O

D 123 - ..

=

+

m a v .

. e 4

REFERENCES (1) (YfSG Repair Safety Evaluation Report, Aug.1982. .

(2) TMI-1 OTSG Failure Analysis Report, July 1982.

(3) Final Report on. Failure Analysis of Inconcel 600 Tubes from OTSG A and 5 of Three Mile Island Unit 1, June 30,1982 - Battelle - Columbus Laboratories.

(4) Final Report: Evaluation of Tube Samples from TMI-1, July 7, 1982; B&W #RDD: 83:3390-03:01.

(5) Technical Specification for OTSG Tube Plugging with B&W Welded Caps and Stabilizers SP-1101-12-030.

(6) Nuclear Safety / Environmental Impact Evaluation for OTSG Tube Plugging Using B&W Welded Cap with Stabilizer.

(7)' OTSG Tube Plugging Phase I SP-1101-12-029.

(8) GPUN Specification SP-1101-22-008. OTSG Tube Repair-Long Ter:n' Corrosion Testing, Rev. 1. .

(9) R. C. Newman, e t. al. " Evaluation of SCC Test Methods for inconel 600 in Low Temperature Aqueous Solutior.s." Symposium held at Nacional Bureau of Standards, Gaithersburg, Maryland, April 26-28, 1982.

(10) NUREG 0565, " Generic Evaluation of Small Break Loss-of-coolant Accident Behavior of B&W Designed 177FA Operating Plants." January 1980.

(11) MPR Report, "WI-1 OTSC Primary-to-Secondary Leakage," September 13, 1982.

(12) NUREG-0017, " Calculation of Release of Radioactive Materials in Gaseous & Liquid Effluent from PWR," April 1976.

(13) *TMI-1 Plant Technical Specifications.

(14) EPRI NP-2146 Topical Report D,c.1981, " Static Strain Analysis of TMI-2 OTSG Tubes."

(15) ASME Section XI, 1980 Edition.

(16) Westinghouse Electric Corporation Report WCAP-10084, TMI-1 Steam Generscor Tube Rolled Plug Qualification Test Report, April, 1982.

(17) GPUN TDR #388 Mechanical Integrity Analysis of TMI-1 OTSG Tubes.

(18) GPUN Spec SP-1101-22-006, Rev. 5.

- 124 - -

4 9

I ..o .

l

• ~

g y -

. e o 1

REFERENCES - (Contd.)

l?

d fi (19)~ GPUN SP-1101-22-009 OTSG Kinetic Tube Expansion Process Monitoring and

! Inspection, Rev. 5. -

(20) TDR 423 Three Mile Island-1 OTSG Tubing Eddy Current Program Qualification.

(21) GPUN TDR 343 RCS Inspection, 6/14/82. ,

(22) J. D. Jones, OfSG Failure Analysis Operational History Final Report, GPUN TDR. #336 May 12,1982.

(23) Three Mile Island Unit Once Through Steam Generator Repair Kinetic Expansion Technical Report - GPUN-TDR-007 Rev.1, March 1983.

(24) J. C. Griess and J. H. DeVan, Oak Ridge National Laboratory, The Behavior of Inconel 600 in Sulfur Contaminated Boric Acid Solutions, September 29, 1982.

(25) GPUN TDR 421, .TMI-1 OTSG Adequacy of Tube Plugging and. Stabilizing Repair Criteria May 1983.

(26) GPUN TDR 9368. Primary to Secondary Leak and Leak Race Determination Methods. .

(27) J. V. Monter, "TMI-1 OTSG Corrosion Test Program - Pinal Report," B&W Report RDD 83-15433-01-01:01 1543301-01, Itay 9,1983.

(28) GPUN TDR 9359 Evaluation Criteria for a Primary to Secondary Leak.

(29) GPUN SP-1101-22-007, Rev. 2, Short Term Corrosion Testing (30) Stress" Report for OT36 Stabilizer Weid Cap, B&W 33-0231-00 (31) Welded Taper Plug Stress Report, B&W 1002581C-02 (32) Stress Report for MK-1, B&W 32-1127439-00 (33) Stress Report for MK-3, S&W 32-1127439-01 (34) OTSG Stabilizer Design Review, B&W 80-0150-00 (35) B&W Position paper on Use of Tube Stabilizers in IMI-1 OTS6, B&W 51-1132-602-00 IS6) 'CHATA-Core Hydraulics and Thermal Analysis-Revision 4, BAW-230, Rev. 4, Babcock & Wilcox, Jut . 1979.

(37) CIPP-C* DATA Input Processing System .

- 125 - ..

A g

--- -. = -

[ *

• 1 o: e 3

1 i REFERENCES - (Contd.)

Il '

(38) TEMF-Thermal Enthalpy Mixing Program, BAW-321. Rev. 2, Babcock &

Wilcox, June 1979. -

(39) B&W Document No. 32-1135 309-00, "? ump code certification for oconee II" by D. J. Raltenan, July 1982.

-(40) NUREG 0565, " Generic Evaluation of Small Break Loss-of-Coolant Accident Behavior of B&W Designed 177FA Operating Plants." January

~

1980.

,) (41) Topical Report BAW 100927 Rev. 3, October 1982.

'l

.i (42) BAW-177 " Preliminary Calculations of the Effect of Plugged Steam

.} ' Generator Tubes on Plant Performance" March 1982.

(43) RETRAN-02, EFRI NP1850 CCM, May, 1981.

~'

(44) Transient Model of Steam Generator Units in Nuclear Power 11 ants -

TRANSC-01 EFRI NP-1368, March 1980. .

(45) Requests for Information on Steam Generator Feedwater Addition Events Letter from Thomas Cox of USNRC to J. H. Taylor, (3&W) dated June 20, 1982. -

(46) Letter from G. T. Fairburn,(B&W) to J. F. Fritzen (GPU) dated December 11, 1979, TMI-79 201.

(47) B&W Document 32-1138230-001 " Evaluation of Effects of 1500 Plugged Tubes on TMI-1 Post LOCA Core Safety, November 1982.

(48) B&W Document, Engineering Criteria for Tube Repair at TMI-1 *

,. #51-1137529-00.

(49) "AUI. A Fortran Program for Dynamic Simulation of Reactor Coolant System and Emergency Feedwater System." S&W Document NPGD581, August, 1981.

(50) Repore in Response to NRC Staff-Recomended Requirements for Restart of Three Mile Island Nuclear Station Unit One - Amendment 25.

(51) Flow-Induced vibration Analysis of TMI-2 OTSG Tubes, EPRI NP-1876, Vol. 1, Proj. 5140-1, Final Report, June 1981.

(52) Determination of Minimum Required Tube Wall Thickness for 177-FA 7 OTSC's. BAW-10146, October 1980.

(53) Fracture Analysis of Steam Generator Tubes, Part II, Steam Intensity Factor and Crack Opening Displacements (COD), by Prof. F. Erdogan, .

Lehigh Univ., Prepared for CPU Nuclear, 9/15/82.

- 126 -

t e

t -Qr Je

_., 0 U %

• % UNITED STATES f -

f * ". , f }

- NUCLEAR REGULATORY CO$tMISSION WAssiNcTON. O. C. 20555 g S. /, j

,  %, . .c

]

Docket No.: 50-289 r ;,r gg.g ,

MEMORANDUM FOR: Gus C. Lainas for Operating Assistant Reactors, DL Director FROM:

Daniel R. Muller, Assistant Director for Radiation Protection, OSI

SUBJECT:

SAFETY EVALUATION REPORT INPUT FOR RETURN TO SERVICE OF TMI-l STEAM GENERATORS (TAC #47484)

PLANT NAME: Three Mile Island Unit i LICENSING STAGE: OR DOCKET NUMBER: 50-239 RESPONSIBLE BRANCH: ORB #4; J. VanVliet, PM May 9, 1983 REQUESTED COMPLETION DATE:

DESCRIPTION OF RESPONSE: SER Input For Return to Service of TMI-l's OTSG

~

REVIEW STATUS: Complete The Radiation Protection Section of the Radiological Assessment Branch has completed its review of TMI-l's plant safety assessment for return to service after the OTSG repair. Although the final dose estimates for the OTSG repair project are much The higher than the original estimates, they are still within an acceptable range.

increase in onsite annual exposure due to the OTSG repairs is expected to be minimal.

This review was performed by C. Hinson, RPS, RAB.

r 5 Y.. ,4l f'~

g/ c*.

Daniel R. Muller, Assistant Director for Radiation Protection Division of Systems Integration Enclosure SER Input For Return to Service of TMI-l cc: w/ enclosure R. Mattson J. Stol:

J. VanVliet .

! 0. Lynch M. Wangler -

p C. Hinson ip 1 i .

(

//

[

y -

O .

Our initial Safety Evaluation Report input (October 13, 1982 letter from Stol::,

NRC to Hukill, GPU) for the TMI-I OTSG Repair was submitted during the early stages of the OTSG repair program.. At that time, GPU's person-rem exposure estimates for each task were based en time and motion studies done in an actual OTSG at B&W's test facility at Mt. Vernon, Indiana. These estimates included only those items in 'the steam generator program identified as part of the kinetic expansion repair. GPU's estimate for these tasks totalled 268 person-rems. During the course of tube expansion, several problem areas were discovered which resulted in a greater amount of time spent inside the OTSGs.

GPU's current estimate for exposure due to kinetic expansion alone 1s'586 person-rems. This is more than double their original estimate of 268 person-rems.

Several problem areas., including equipment malfunction, contributed to this increased exposure. Video camera problems during tube expansion blasting resulted in having to remove the camera prior to each blast and then reinstall- it i

prior to each personnel entry. Although this resulted in increased personnel l

exposure, it was more person-rem efficient than performing corrective maintenance on the camera inside the generator. Mechanical problems with the debris removal device resulted in much higher exposures than originally estimated during the first pass' debris removal. The licensee redesigned this devicte prior to the second pass debris removal, and the resulting doses were reduced significantly from those experienced on the first pass. The procedure of precoating the generator tubes after the expansion inserts weire in place resulted in an unacceptably high misfire rate due to seepage of the liquid precoat into the inserts. Inserting the inserts after the precoat was applied reduced the number of misfires (and therefore the time spent in the generatbE required to identify and replace misfired inserts),but resulted in higher exposures because twc

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

6 k

generator entries were required for each firing. Another prcblem encountered was misfire jump out. The ' force of the insert detonations caused the detonating cords to pull some of the misfired inse*rts out of the OTSG tubes. The time spent in locating the unexpanded tubes from the misfires resulted in additional exposure. During the second' pass expansion, misfire jump outs were prevented by using hold-dcwn devices and additional care in moving about the head area.

These changes resulted in some increase in the tirre spent in the generator.

The actual time involved in removing debris from the generator following each This, too, detonation turned out to be much longer than origina11y planned for.

resulted in additional unplanned exposure. ' -

Although these unexpected problems resulted in a higher overall exposure than originally estimated for the kinetic expansion process, the licensee incorporated ALARA technique's in resolving these problems". ThelJcenseemonitoredand documented all generator entries to minimi:e exposures and measure worker efficiency. The licensee also minimi:ed exposure for the performance of corrective maintenance. Where equipment redesign was necessary, the licensee incorporated

' lessons learned about equipment reliability into the equipment redesign to minimize equipment downtime and thereby minimize exposure. All workers received training in the job to be performed prior to actual steam generator entry. Prior to initiation of the kinetic tube expansion, the steam generator heads were flushed and shielded to reduce worker exposure. Additional ALARA practices used .

by the licensee during the OTSG repair were described in our previous Safety Evaluation Report (October 13,1983). -

i t

~

~

'~

In addition to the 586 person-rems expended for the kinetic expansion process, the licensee will expend'approximately 697 person-rems for other portions of the program such as preparatory work, tube plugging, end milling, clean-up, and testing. Approximately 331 tubes were plugged before the magnitude of the

' tube cracking problem was discovered and it was decided to repair the tubes using the kinetic expansion process. Approximately 814 additional tubes are being plugged or stabilized as part of the OT5G program. Tube plugging and stabilization will account for approximately two-thirds of the non-kinetic expansion exposure. The balance of this exposure will be due to RCS inspection, eddy current testing, end milling, clean-up, and testing.

The total estimated exposure for the OTSG repair is 1283 person-rems. Additional plugging may result in a slightly higher total for the job. The licensee has performed all steam generator work in accordance with ALARA guidelines and all required radiation protection practices. They have incorporated the ALARA guidance provided in Regulatory Guide 8.8, Rev. 3. "Information Relevant to Ensuring That Occupational Exposures At Nuclear Power Stations Will Be As low As Is Reasonably Achievable" and have maintained individual radiation doses within the limits of 10 CFR Part 20. Although the licensee's final exposure estimate for the expansion process is more than double their original estimate, the final estimate of 1283 man-rems for the entire OTSG repair is comparable to exposures from steam generator repairs at other facilities. Based on the above, the staff .

i finds the occupational dose control aspects of the OT5G repair project, and the final exposure for the project acceptable. ,

l I

L

The licensee has taken several steps to prevent future chemical contamination of the OTSGs and the RCS. These include removal of the sodium thiosulfate tan and implementation of administrative controls to prevent direct injection of contaminants into the RCS. Primary to secondary leakage is not expected to occur following the OTSG repair, but if this does occur, it is expected to increase the annual exposure at TMI-l by less than 1%. This increase is small and is acceptable.

m e

o 6

O o

.