Text

Draft TMI-1 Once-Through Steam Generator Failure Analysis Rept, Technical Data Rept
	ML20064M520
Person / Time
Site:	Crane
Issue date:	06/25/1982
From:	Jackie Jones, Rosalyn Jones, Olszewski J; BABCOCK & WILCOX CO., ELECTRIC POWER RESEARCH INSTITUTE, GENERAL PUBLIC UTILITIES CORP.
To:	;
Shared Package
ML20063K013	List: ML20063K020; ML20063K042; ML20064M520;
References
	TDR-341-DRFT, TRD-341-DRFT, NUDOCS 8209030170
	Download: ML20064M520 (116)
	v • d • e

{{#Wiki_filter:_ _ - - _ _ _ _. l DRAFT ONLY - NOT FOR RELEASE gggg TDR NO. 341 REVISION NO. O BUDGET TECHNICAL DATA REPORT ACTIVITY NO. 120010 PAGE OF PROJECT: DEPARTMENT /SECTION OTSG Failure Analysis TMI-1 RELEASE DATE REVISION DATE DOCUMENT TITLE: TMI-l OTSG Failure Analysis Report ORIGINATOR SIGNATURE DATE APPROVALIS) etGNATURE DATE J. D. Jones R. L. Long R. L. Jones (EPRI) D. G. Slear J. S. Olszewski (B&W) I APPROVAL FOR EXTERNAL DISTRIBUTION DATE R. F. Wilson o DISTRIBUTION ABSTRACT: Statement of Problem R. L. Long I J. D. Jones Investigate the failure of the OTSG tubes to R. L. Jones identify the damage mechanism. J. S. Olszewski I R. F. Wilson Key Results D. G. Slear F. S. Giacobbe 1. OTSG tube cracks were I.D. initiated. M. J. Graham 2. Sulfur was present on crack surfaces. I 3. Analysis of primary coolant revealed the presence of sulfur. 4. Sulfur has been shown to promote inter-I granular stress assisted cracking (IGSAC) i in OTSG tube material. I 5. Aggressive sulfur species formed subse-quent to hot functional testing and cracked the OTSG tubes during cooldown or cold shutdown. I DRAFT ONLY - NOT FOR RELEASE I oCOVER PAGE ONLY 8209030170 020625 PDR ADOCK 05000289 40000030 7 81 P PDR

TABLE OF CONTENTS I EXECUTIVE

SUMMARY

II DESCRIPTION OF OTSGS AND CRACKING PROBLEM III OTSG OPERATING HISTORY IV OTSG CHDilSTRY VI METALLURGICAL EXAMINATIONS VIII OTSG TUBE STRESS ANALYSIS i IX CRACKING TESTING X PROBABLE CAUSE OF FAILURE l !I i lI lI lI 'I I I I

I I-1 I. EXECUTIVE

SUMMARY

A. Failure Analysis Task Group The Failure Analysis Task Group vas organized as Task 1 of the Once Through Steam Generator Project Team. The principal purpose of Task 1 was to identify the cause of the TMI-1 Once Through Steam Generator I (OTSG) tube damage mechanism. The group was then to develop recommenda-- tions for passivation and cleanup and prevention of the damage mechanism occurring again. The Failure Analysis Task Group was organized under the direction of R. L. Long and included individuals from a variety of organizations within GPUN. Table I-1 indicates the key responsible individuals f rom I G PUN. A similar Failure Analysis Task Group was organized at Babcock & Wilcox I under the direction of J. S. Olszewski and key individuals in this group are shown in Table I-2. As noted in Table I-1, Dr. Robin Jones of the Electric Power Research I Institute, Systems & Materials Group, also was an active participant in the Failure Analysis Task Group. B. Repo rt Purpose and Organization The purpose of this report is to present the findings of the Failure I Analysis Task Group. Section II of the report presents a brief descrip-tion of the Once Through Steam Generator (OTSG) as a basis for subse-quent discussions and a summary description of the intergranular stress assisted attack phenomena observed in the OTSG's, following the dis-I covery of primary to secondary leakage in November, 1981. Sections III and IV present Operational and Chemistry history, respec-I tively. The results of a fairly extensive sampling program implemented af ter the November 1981 discovery is included in Section V. In Section VI, the results of the detailed metallurgical examinations of tube samples pulled f rom the TMI-l OTSG's are described. Section VII presents a summary of the detailed information that has been gathered regarding the f abrication history of the OTSG's. In Section I VIII the results of tube stress analyses, performed primarily by MPR, Incorporated, are presented. Section IX describes the results of cracking testing which has been underway, attempting (1) to demonstrate the conditions which caused the attack, (2) to verify that cracks are no I longer propagating and (3) to test the validity of the passivation and cleanup proposals. I Finally, Section X delineates the postulated cause of failure and the recommendations coming out of the Failure Analysis Group. I I

I I I-2 C. Findings 1. History a. Fabrication History The f abrication history of the TMI-l OTSG's is typical of such I Babcox and Wilcox unit s. There is no indication that tube material, f abrication or installation in the OTSG's was in any way extraordinary. The heat treatment of the whole OTSG I f ollowing assembly has the effect of sensitizing the tube material to possible intergranular attack. This is common to al10TSG's manufactured by this process. Reference 2 reports the details of the fabrication history, b. Operational History -I The operational history of the TMI-1 OTSG's reveals no subjec-tion of the tubes to excessive stress during the time when damage occured. Operations did, however, have a significant impact on the chemical environment of the OTSG tube. There were five (5) identifiable instances of probable intrusion of chemical contam-I inants 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 I through the vent header. Some oil may subsequently have found its way into the RCS. In October 1979, sulfuric acid was injected into the reactor coolant makeup system. Although attempts were made to prevent the acid f rom reaching the RCS, I chemistry results indicate some contamination of the RCS occurred. In July 1980, May 1981, and September 1981, there are indications that sodium thiosulfate from the building spray system found its way into the RCS. Significant to the localization of the attack was the history of the water level on the primary side of the OSTG. Figure I-l shows water level as a function of time. It is noteworthy that following the hot functional testing in September 1981, water level was promptly lowered then slowly raised over several l days. This allowed a drying then rewetting of the tubes in the 5 upper portion of the steam generator. Reference 1 reports the details of the operational history. 2. Metallurgical Results a. Analysis Program A multi-task program was planned to provide inf ormation related to the steam generator tube damage problem. This program con-tained the following analyses / examinations: I

I I-3 1) Visual Examination

2) Eddy-Current Examination

3) Radiography

4) Sectioning and Bending

5) Scanning Electron Microscopy (SEM) and Energy Dispersive X-Ray Analysis (EDAX)

6) Auger Elec tron Spectroscopy ( AES)

7) Electron Spectroscopy f or Chemical Analysis (ESCA)

8) Sodium Azide Spot Test

9) Metallography-Microstructural Analysis

10) STEM, EPR and Huey Testing I

11) Residual Stress and Plastic Strain

12) Tension Testing

13) Hardness Testing

14) Dimensional Measurements b.

Test Program Re sults/ Conclusions The detailed test results are presented in Section VI. The f ollowing summarizes those results and sets forth some con-clusions:

1) The tubing f ailed due to stress assisted intergranular attack. This lead to through wall penetrations and circum-I ferentially oriented cracks.

In all cases, cracks were initiated on the inside surface.

2) The intergranular morphology has been confirmed by both I

Metallography and Elec tron Microscopy.

3) Transmission Electron Microscopy has also confirmed that no I

secondary modes of failure are associated with the inter-granular corrosion, that is, no evidence of any low or high cycle fatique was observed on these f racture surfaces. 4) In conjunction with the cracking, there has also been general intergranular attack observed. In general, this has been characterized as severe by Battelle Columbus and has I

I I 1-4 been characterized as not severe by B&W. The differences between the two laboratories is most likely associated with the extent of severity of the cracks observed at the loca-I tions where intergranular attack was noted. More severe cracking, in general, relates to more severe intergranular attack.

5) Analysis of surf ace films on fracture surf aces and on the I.D. surface of the tubing indicate that sulfur is present up to levels of eight (8) atomic percent, and that it is 8

this sulf ur that is most likely responsible for the cracking phenomena. Although the form of sulf ur has not been defini-tively determined, it is believed to be either in the form of a nickel sulfide, Ni2 S3 or some other reduced form I of sulfur. It is believed that the presence of the reduced sulfur form is responsible for the cracking mechanism and without such a contaminant the attack would not have occurred.

6) Microstructural evaluation of the tubing f rom numerous loca-tions, has indicated that the structure is representative of I

that normally expected for steam generator tubing. However, tests have concluded that the material is in a sensitized condition, and hence, is expected to be susceptible to intergranular attack. 7) Visual, metallographic or bend specimens perf ormed in clean I regions of the tubing, that is regions where the standard differential eddy current probe oparates and no eddy current indications were recorded, all show the tubing to be f ree of defect. In addition, all examinations perf ormed at loca-tions where absolute eddy-current probing indicated a nearly through wall or through wall defect has confirmed the presence of cracking in those regions. Comparisons of I metallographic and eddy-current results have generally shown a one-to-one correlation. In the seal weld Heat Affected Zone (HAZ) where due to tube end effects the standard differential probe is ineffective, defects were first I detected visually, then confirmed by absolute probe.

8) Although there is excellent correlation between eddy-current I

indications and metallography, it has also been learned that eddy-current signals of 80% to 90% through wall, 'a most instances actually represent a 100% through wall defect.

9) In all cases to date, cracks which have been examined either by metallography or by bend testing, have shown the defects to be at least 90% and generally 100% through wall in

<l penetration. No defects in the range of 10% to 50% have l5 been observed. This would suggest that crack growth rates !'I il

I I-5 are rapid and that defects, once initiated, propagate through wall. Some eddy current indications in the 50% - 90% through wall range have been observed. I

10) Auger and ESCA analysis have shown the presence of carbon, nickel, chromium, oxygen and sulfur on the f racture sur-I faces.

In addition, normal trace quantities of fission products have also been observed. The important results from this analysis have shown that sulfur concentrations along the I. D. surf ace of the tubing down to the 27 inch I point, are generally uniform with perhaps a slightly decreasing level as you go down the tube length. In addi-tion, the large presence of oxygen also confirms that oxygen I was definitely available for the cracking reactions and very possibly played a major roll for the f ailures observed.

11) The consistent circumferential orientation of the cracks I

indicates that a longitudinal stress is part of the cracking mechanism. Residual stresses in the roll alone were not responsible for the cracking. 3. Chemistry Results Generally, the reactor coolant system remained within specifications I for those parameters for which an analysis requirement existed for the period extending f rom April 1979 through November 1981. Two incidents of intrusion of ionic substances not accounted for by I specific analyses have been identified. Sodium thiosulfate at levels of 4-5 ppm as thiosulfate is considered to be the most likely contaminent. The ionic species f rom the first contamination I incident in July 1980 were removed f rom the bulk liquid by deminer-alization in August 1980. The ionic species from the second contam-ination 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 ppm thiosulfate residual could have still been present at the start of September 1981. The operational history shows that addi-tional sodium thiosulfate in the RCS may have resulted f rom injec-I tions of Borated Water Storaga Tank (BWST) contents during cooldown from hot functional testing. The quantity was not sufficient to be detectable by conductivity. Induction of an unidentified organic substance, probably oil or grease, into RCS auxiliary systems I occurred in mid-March 1979. The exact quantity of this substance or its potential role in the failure mechanism has not been estab-lished. It has been determined that this substance could not have I introduced sulfur into the RCS in the quantities observed. Details are provided in Reference 1. I I I

I I I-6 4. Distribution of Damage In-situ eddy-current results exhibit tube wall defect indications I at varied rates distributed both axially and radially in both OTSG ' A ' a nd ' B ' tube bundles. The majority of the defect indications are in the upper tubesheet (UTS) region and particularly confined to the tube roll transition zone. Ninety-five (95) percent of all I tubes with defect indications are contained in the top seven (7) inches of UTS with less than 200 tubes containing defect indications below the UTS (80 percent of those defects are above the top support plate). Radial distribution of tubes with def ect indications in 'A' OTSG shows a high percentage in the UTS periphery with the defect rate decreasing rapidly as you move toward the center of the I bundle. In the 'B' OTSG, however, half of the UTS exhibits an approximate 20 percent defect rate with defect rates to 90 percent in a broader peripheral area of the other half. Reference 9 docu-ments defect rates in detail. 5. Damage Scenario I The occurrence of stress assisted cracking requires that three (3) conditions be satisfied simultaneously: o a sufficiently high tensile stress I o a susceptible material microstructure o an aggressive environment I The information presented in subsequent sections of this report relating to these three (3) f actors is summarized below: a. Tensile Stress I Section VIII presents inf ormation about OTSG tubing stresses. Since the cracks are oriented circumferential1y in the tubes, axial tensile stresses are of principal interest. Both operating and residual stresses must be considered since both can play a role in IGSCc /for example, both categories of stress are involved in IGSCC of BWR stainless steel piping). I Cracking must have occurred in a situation in which the sum of the operating and residual stresses in the axial direction was greater than that in the hoop direction, otherwise the crack I orientation would have been axial. The Section VIII analyses indicate that this condition is satisfied during cooldown and cold shutdown. Highlights of the stress analysis f rom the f ailure scenario viewpoint are:

1) Tubing axial tensile stresses are largest during cooldown when they may approach the yield stress.

2) Significant axial tensile stresses also exist during cold shutdown.

I I I-7

3) Locally high axial tensile stresses are possible in the seal weld heat affected zone and in the vicinity of the roll transitio n.

4) Under heatup and at full operating temperature the hoop stress generally is larger than the axial stress.

5) The axial stresses are generally larger at the periphery than in the center of the tube bundle.

Thus the stress analysis results suggest that the cracking must have occurred during cooldown or during cold shutdown. The stress analysis also explains why the seal weld heat affected I zone and the roll transition region should be particularly prone to cracking and why more cracking occured in the periphery than in the center of the tube bundle. b. Susceptible Material Microstructure The OTSG fabrication history is documented in Section VII. The fabrication history puts the tubing into service in the mill annealed plus stress relieved condition which is expected to be heavily sensitized (i.e., grain boundary chromium content less I than 10%). Metallurgical examination has confirmed that the expected microstructure is present. c. Aggressive Environment The results presented in Section V indicate that sulfur was I present in the primary system water and three (3) possible sources of sulfur have been identified from the OTSG chemistry history (Section IV). I If SO4 and S 02 3 were introduced to the primary water as the OTSG operatir.g and chemistry histories suggest, they would be expected to persist as long as the water was at room tempera-I ture even if the oxygen content of the water was reduced by hydrazine additions. However, hydrogenating and heating the water to perform a hot functional would be expected to result in the generation of S-, possibly accompanied by S and other intermediate species. Subsequent cooling to room temperature and oxygenating following the hot functional would rapidly oxidize S-~ to S and could also result in the appearance of I significant concentrations of other species of higher oxidation states. Although it is not possible to predict either the identities or the concentrations of the sulfur species present f ollowing the hot functional test, it is clear that this tran-I sient is likely to have greatly affected the aggressiveness of the environment with regard to low temperature sulfur-induced attack of the OTSG tubing. I I

I I I-8 6. Proposed Failure Scenario The discussions presented above suggest that the probable cause of failure was as follows: During lay up the primary system was contaminated with sulfur by a. the accidental introduction of sulfuric acid, sodium thiosul-I f ate, and possibly o sulf ur containing oil. The amount of sulfur present may have reached several ppm, but the contami-nated water was not aggressive enough to crack mill annealed I plus stress relieved Alloy 600. Cracking tests confirm that cracking would not have been expected to occur at this stage. I b. The temperature and oxidation potential transient associated 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 water was injected during the tests of the high pressure and low pressure injection systems. I When the water level in the OTSGs was lowered following the hot c. functional test, high concentrations of aggressive metastable sulfur species developed in the dry out region at the top of the generators due to the combined effects of solution concentration I 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 aggressise sulfur species. d. Sulfur-induced IGSCC of the Alloy 600 tubing occurred rapidly in the dry out zone with preferential attack at high stress loca-I tions. Little or no cracking occurred below the water line because the bulk solution was less aggressive. I Cracking terminated either because continued chemistry changes e. 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 I dropped again, the chemical state of the sulfur in the primary water was suf ficiently different from its state immediately af ter the hot functional to prevent a full scale recurrence of steps c. and d. in the new dry-out zone. f. Cracking was discovered when the OTSG's were pressurized. This scenario is consistent with all of the observed features of the cracking phenomenon (with the possible exception of the nonaxisym-metric radial distribution of cracking in OTSG-B) and is also con-I sistent with the timing of the cracking and the results of the metallurgical examinations and corrosion tests. I I

I I-9 1 7. Cracking Testing Cracking testing to confirm the above scenario is currently being conducted. Preliminary observations are: i a. M2320 Archive Material appears less susceptible to cracking than actual generator tube material. b. The presence of an inert atmosphere reduces cracking tendency. c. Sulfate at low temperature will not cause cracking under oxidizing conditions. I d. Thiosulfate at low temperature will cause cracking under oxidizing conditions. I e. Archive Material in the mill annealed and stress relieved con-dition has cracked in 5 ppm thiosulfate, but not in 1 ppm thio-sulfate, indicating a threshold concentration requirement of greater than 1 ppm. I I I I , I I I I lI I

I E I - 10 TABLE I-l GPUN FAILURE ANALYSIS TASK GROUP I RESPONSIBLE GPUN ACTIVITY INDIVIDUAL ORGANIZATION Group Leader R. Long T&E Met & Cracking Test S. Giacobbe Matls. Tech SG Fabrication History J. Wildermuth QA Chemistry History K. Frederick Systems Lab. SG Operating History J. Jones T&E Chem. and Swipe Samples M. Campagna/J. Tangen Proj. Engr./ Chem. Engr. Industry Experience & Research R. Jones (EPRI) Full Time Task Support J. Paules TMI-l STA H. Crawf ord TMI-l STA J. Tangen Chem. Engr. Assigned Task Support J. Sipp Chem. Eng r. lI R. Demuth Matls. TF H. Shipman Plant Engr. i G. Reed Plant Chem. A. Clemons T&E J. Moore Tech. Functions Independent Metallurgical J. Janiszewski Consultant Stress Analysis S. Weems (MPR) Metallurgical Failure Analysis A. Agrawal ( Battelle) I I

I l I I - 11 TABLE I-2 I B&W FAILURE ANALYSIS TASK GROUP I RESPONSIBLE G PUN ACTIVITY INDIVIDUAL ORGANIZATION Group Leader J. Olszewski Mech. Eng r. Met & Cracking Test G. Clevinger/G. Theus LRC/ ARC SG Fabrication History R. Boberg Mech. Engr. Chemistry History M. Bell Engr. Chem. and Swipe Samples M. Bell Eng r. Full Time Task Support J. Smith Mech. Engr. R. Carey Mech. Engr. M. Rigdon LRC Assigned Task Support R. Post Mech. Engr. R. Pelger ARC E. Pa rdue LRC T. Ha rd t LRC

I

,I I I .I

I

!I

I I OTSG PRIMARY SIDE LEVEL IN FEET ELEVATION I EE s s =, ; = =

=

= = m i i i i I -1 3 I e_ C c-I -o Q 8 V . I c,, 32 n m> m o, g r-- I g m< m m m c m a m o g 7 c. m s c, es 33 !I >0 C m -wa 0 m 32 ~ W o m m m 3 m > C2 m O b a -- C m aa m q en E mm i Ea gg W ~, c a ee o ~. m m lg 2== T n v3 -n en m -~~. O r a T ~ mo ( O I "m ~ C O I G my ma \\ 3-r- - m-mg <m \\ C" CD mC A f "U "i r= H f t m ga __,m e________-_ I h N,- _a m t,,J M< C H c I Q

M n

Z T H (M 'C

==h m C C ~ C 5 a m "U -4 I C C "U 33 cc 32 ~ -4 m m 32

O

~U t.f) U CA r= "C C ri I m 32 - --( m 'Qp m -4 2 mm C3 C2 C I l ......4 7 r= 32 h I o Cka I

I I II - 1 I II. DESCRIPTION OF OTSG'S AND CRACKING PROBLEM A. Summary Description of OTSGs The OTSGs are vertical straight tube and shell, once through heat exchangers (See Figure II-1) with shell side boiling that produces super-heated steam at constant pressure at the turbine throttle (about 925 psia and 570 to 5900 F) over the power range. Primary coolant f rom the reactor enters the steam generator through a nozzle at the top, flows down-I ward through more than 15,000 Alloy 600 (Ni, Cr, and Fe) tubes, is col-lected in the bottom head, and exits through two outlet nozzles. On the secondary side, the subcooled f eedwater is sprayed downward into a I steamfilled annulus between the shell and tube-bundle shroud where it is heated to saturation temperature by direct condensation of steam on the water droplet s. The saturated feedwater then enters the tube bundle at the lower tubesheet where nucleate boiling begins. Af ter reaching 100% I quality, the steam becomes superheated, leaves the tube bundle above the 15th tube support plate (TSP), flows downward through a steam annulus, and exits through two steam outlet nozzles. I hube support plates maintain the tubes in a uniform pattern along their le ng th. A unique feature of the design is the broached tube support plate concept ( Figure II - 2). These tube support plates are f abricated f rom I 1 1/2 inches thick carbon steel, drilled, and broached at three points spaced 1200 apart. The broached design effectively eliminates stagnant areas where solids can concentrate by creating large openings for the flow of water and steam. Another f eature in the OTSGs is an untubed inspection lane (Figure 11-3). I This lane is for=ed by omission of a row of tubes halfway across the tube bundle to f acilitate inspection of the tube bundle and possible chemical cleaning at some time later in the operating life of the equipment. The attached exhibits provide additional general OTSG data: Table II General OTSG Data I Figure II OTSG Longitudinal Section Elevations (Typ.) Figure II TMI-l OTSG Upper Tube Sheet Detail (Typ.) Figure II OTSG MFW, AEW, and Steam Penetrations B. OTSG Heat Transfer The characteristics of the f our heat transfer regions that exist in the OTSG as feedwater is converted to superheated steam are described further: 1. Feedwater Heating. Feedwater is heated to saturation temperature by direct contact heat exchange. The feedwater entering the unit is sprayed into the downcomer annulus f ormed by the I

I I II - 2 I shell and the cylindrical baf fle around the tube bundle (see Figure 11-4). Steam is drawn by aspiration into the downcomer and heats the feedwater to saturation temperature. The saturated water level in the downcomer provides a static head to balance the static head in the nucleate boiling section, and the I required head to overcome pressure drop in the circuit formed by the downcomer, the boiling sections and the bypass steam flow to the feedwater heating region. The downcomer water level varies with steam flow f rom 15 to 1007. load. A constant minimum level I is held below 15% load. 2. Nucleate Boiling. The saturated water enters the tube bundle just above the lower tubesheet ani the steam water mixture flows upward on the I outside of the tubes countercurrent to the reactor coolant flow. The vapor content of the mixture increases almost uniformly u..til DNB is reached, and then film boiling and super-heating occurs. 3. Film Boiling. I Dry saturated steam is produced in the film boiling region of the tube bundle. 4 Superheated S team. Saturated steam is raised to final temperature in the super-heater region. The amount of surface available for superheat varies inversely with load. As load decreases the superheat section gains surface f rom the nucleate and film boiling re-gions. Mass inventory in the steam generator increases with load as the length of the heat transf er regions vary. Changes I in temperature, pressure and load conditions cause an adjustment in the length of the individual heat transfer regions and result in a change in the inventory requirements. If the inventory is I greater than that required, the pressure increases. Invento ry is controlled automatically as a function of load by the feed-water controls in the integrated control system. Primary and secondary side OTSG operating temperatures vary over the power range. Typical operating termperatures are presented in Figure II - 7. In accition, Figure II - 8 is attached to I illus trate prima ry fluid, seconda ry fluid, and tube wall tem-peratures over the length of the OTSG at 100% power. C. Description of Intergranular Attack 1. Basic Phenomenon The f ailure mechanism responsible f or the crackirg of the D11 Unit 1 steam generator tubing is stress assisted intergranular I

I I II - 3 I attack. This phenomenon preferentially attacks the material grain boundaries by elec trochemical dissolution, then under the influence of stress, forms a continuous crack network perpendicular to the applied stress direction. In the case of the TMI-l steam generator tubes the cracks were circumferential I in orientation and the corrodant responsible for the attack is believed to be a reduced sulfur species. 2. Results of Eddy-Current Examinations In situ eddy current results exhibits tube wall defect indica-tions at varied rates distributed both axially and radially in I both OTSG 'A' and 'B' tube bundles. The majority of the def ect indications are in the upper tubesheet (UTS) region and particu-larly confined in the tube roll transition zone. Ninety-five (95) percent of all tubes with defect indications are contained in the top cc ven (7) inches of UTS with less than 200 tubes containing defect indications below the UTS (80 percent of those defects are above the top support plate). Radial distribution of tubes with defect indications in 'A' OTSG shows a high per-centage in the UTS periphery with the defect rate decreasing rapidly as you move toward the center of the bundle. In the 'B' I OTSG, however, half of the UTS exhibits an approximate 20 per-cent defect rate with defect rates to 90 percent in a broader peripheral area of the other half. Reference 9 documents defect rates in detail. I I I !I i

I II - 4 I TABLE II -1 GENERAL OTSG DATA Bottom of Support Skirt to Top of Inlet Nozzle 73 ft., 2 1/2 in. Wei gh t s : Shipping . 70 ton s Flooded 7c 0 ton s Operating (15% lo ad) 62i cons Operating (100% load) 63 7 to.t c Pressures: Design Full Load Operating Hydrotest Prima ry 2500 psig 2200 psia 3125 psig Steam Outlet 1050 psig 925 psia 1312.5 ps ig Feedwater 1100 psig Primary Drop 33 psi Te mpe ra ture s (OF) Design Full Load Operating Hydrotest (min) I ( Shel1 Temp. ) Primary 650 602 Inlet 100 Secondary 600 595 100 Feedwater 600 470 Operating Limits and Precautions I Minimum Pressurization Temperature 100F Maximum Heatup/Cooldown Rate 100 F/hr Minimum Annulus Water Temp'erature 10 F below saturation Operating Conditions (Full Load) I Hea t Transfer 4.35 x 109 BTU /hr Primary Fluid Flow 68.94 x 106 lb/hr Stem Flow 5.30 x 106 lb/hr Heat Transfer Tubes Numbe r 15,531 I Material Inconel alloy 600 Size 0.625 0.D., 0.034 minimum wall, 5 6 f t., 2-3/8 in. long, with 52 ft. 1-3/8 in. heating length Total Heating Are a 132,436 f t. 2 I I

i I i FIGURE II-l Once-Through Steam Generator i l Reactor Coolant inlet Manway I Auxihary Feedwater / Nozzle 15th I o{, Steam Annulus % 14th n, -13th ,y l Steam Outlet N, H' b i d - 10th Feedwater inlet - 9th Feedwater y' ' 1 I t 7'.'-Tube Support Pfate Heating Annu!us II 1 '!

4.. I.

g.s v Handhole M*"**V I Reactor l Coolant Outlet ll l-1 I I FIGURE II - 3 E FIGURE II - 2 ' g Tube Bundle Cross OTSG Broach.ed Tube Support Plate Cection Chowing, Lane W m es-Minimum msr g g, e -. Mf-fgg!'Z35 ?f.+ * 'I i@ - outsee tube radius l .3125snches

.g f.

g 7 =.... ..._. ;:g_- ~ I '^ Q~ 1 ' y.2 =?~ -~:. w e ::-:.::,gy =,- - . r c... s .:kk-{l? $ n: ' y'. ;-. ,.l l .r r. Minimum I _:A ah'- ' l..

j H 4 7:

4-. - 4.i ' 7 ? '#^ '" g i I ..4 Y , N.1) I dnli radius -:4. r, 4.,... -.. l': .320 inches ".M-e- w Note: Plates fabncated ~ b

'.)

'IkN [., e 7- @_e? TiRT -PT N -- ~ ' V from 1%-inen thich plate. I YM[. - - e u. .a g. F r *W$'". df. s - d :... c:.... ' N-% e- -i t,* p ' .W l ~ n b k~ b o Tube Y i I

I

g rrouas 11

-4 I OTSG Longitudinal Section Elevations (Typ.) I PRIMARY SIDE (INSIDE TUBES) - I UPPER ELEVATION TUBESHEET (UTS) FEET I - 346'6" AUXILIARY N .mmimmnummoia FEEDWATER L-i Jppgilil! gig I N0ZZLES W --- 3 42' 8" (AFW) b!bl![f 'lI @j!!,; [ l 339'9" ), I'l.1..idlhl I 'S 336'5" .l 3'@. i STEAM OUTLET -333'8" 330'6" N0ZZLES (MFW)g] lll y!al Bhl!!!i"' j MAIN FEEDWATER 327'4" lg.ngunin

r 324'4" l Ill,..af l

L di SECONDARY SIDE - 321'1" (EXTERNAL TO TUBES) % [g;d j ,i77" 317'9" ) ,,,,o 1 *, - 314'6" l Im 311'8" I* - 308'5" TUBE ~ f m - 305'1" SUPPORT PLATES 'I" ' = 3 01'10" l LOWER 1 00" SECONDARY ~ LOWER HANDHOLE SECONDARY M ANWAY - 294'5" j LOWER I TUBESHEET (LTS) l t Il 1 REACTOR BLDG. FLOOR 281'0" r

I I F I G U RE II - 5 TMI-L OTSG Upper Tubesheet Detail (.T y p ). .051" MIN. ? .635" REF. I r-.051" MIN. l .187 q Y V l A/ /g ^t I /

L
L

/ / / / / / I N y .035 5/16" 1" MIN. EXPANSION MIN. CLADDING l l l / / / / ll N' V 24" - 5/16" .875" .olo-l l 4> .240 TYP. l fl\\ f\\ I 1 i + .7578 l i r w ,A 9 I \\ .y l TYP. TUBE PATTERfi .625" TUBE 0.D. X.034" 3 Minimum Wall 4 y MIN.003 (Di'ector) M AX.016 ' r i a.' c t e r ) .635" DIA. HOLE s

F I G U R E II - 6 OTSG, MFW, AFW, and Steam Penetrations l AFW STEAM >g AFW \\ AFW 'W u ~ / I ,g l E -Lane t in .urw + s.- x ggo7 >g ~ 11* urw e l 27* jl 45* uru RECIRC 32 xm RETURN PENETRATIONS / \\ (TYPICAL) AFW ii AFW

I

\\ STEAM l. a i AFW S tc a.: 2-21' Stcan Gutlet ArW: 6-2 Inlets, (Originally 7.Z-Axis plugged in 1980.) Located a 36" below lower edge of UTS W MW: 3'.-3" Inlets, spaced a 11 around OTSG. Recirc Return: Returns through UTS vent and level sensing connection. j N 14" from lower edge of UTS. 1" connection. i

E F.I C U R E II _ 7 OTsG Primary and seconday Temperatures versus Load 620 I i Reactor Outlet Temperature 600 p Steam Temp. 580 { Reactor Average ~ Temperature 560 Reactor inlet Temperature !I 5% I Tsat AT 925 PSIA I 520 i 0 20 40 60 80 100 Reactor Po w r, f, I I II lI

I . i I .I I I I I I I I !I 4 ~ l s. I am ED I O 8 i i. E 5 s e" -m E E W s g 5 -g 's w R =x [ E O n ~ ? =5 -M s E s i W = 4 MS g W CM \\ 3 -O O E3 .O u a MJ Ma I mh g O o O go - i. g Og i \\ gg s E o g WJ E E "w \\ a.J e t 34 wa g>3 Cm X s =E I \\ \\ ~ \\ \\ 1 \\ = I 5 i i i i i bbbbi 1 kWE.

N 6 1

l

I E 111-1 J l III. OTSG OPERATING HISTORY A. Purpose and Key Points 1. Statement of Purpose The failure of the OTSG tubes by intergranular stress assisted cracking leads us to examine the environment to which the tubes have been I subjected. The purpose of this analysis of the operating history is to provide information on those environmental parameters which might rea so n-ably have had some impact on the tube failures. Included as elements of I the operating history are those incidents in which the introduction of chemical contaminants into the reactor coolant system is considered like-ly. 2. Summary of Key Points The water surf ace in the primary side of the OTSG's during the a. I layup period since Feburary 1979 has been in the upper tube sheet for a minimum of 31 days and a maximum of 243 days. The large uncertainty is due to the lack of good data on reactor coolant drain tank level. b. It is probable that oil was introduced into the reactor coolant system in early March 1979. c. It is probable that sulfuric acid was introduced into the reac-tor coolant system in mid-October 1979. d. It is probable that sodium thiosulfate was introduced into the reactor coolant system in July 1980, May 1981 and early Septem-B. Methods i In collecting the operational data, a review was made of operator logs, shif t foremans' logs, and automated printout s. In addition, inter-views were conducted with plant operators and supervisotf personnel. Two TMI-1 shift technical advisors were assigned to this task full time. I I I 1

I E III-2 I W C. Results 1. Pressure, Temperature and Level Plot The environmental parameters of prvssure, temperature, and level in the primary side of the OTSG are plotted over the period February 1979 to I November 1981. Reference 1. contains these plot s. The re is a t t ime s an uncertainty of about 55 inches in the OTSG primary side level due to the venting arrangement of the reactor coolant hot legs and the pressurizers. I The graphs show pressurizer level and include the uncertainty in OTSG 1evel as a reference line below the level of te OTSG upper tube sheet. Thus, when the plot of pressurizer level is above the reference line, 8 OTSG level may lie in the upper tube sheet. Figure III-l shows sche-matica11y the venting arrangement which gives rise to the uncertainty. The difference in level between the pressurizer and the OTSG is the level of water in the reactor coolant drain tank. This tank level is I not always recorded. When the drain tank level was recorded, both pres-surizer and OTSG 1evel were plotted. The OTSG level is always the higher level. Figure I-l summarizes OTSG level behavior for the period July - November 1981. 2. Contamination Incidents I Among the significant events considered to have an impact on the OTSG environment were incidents of possible chemical contamination of the Reactor Coolant System (RCS). Three such incidents have been identified. The possibility that resin breakdown contributed chemical contaminants was considered and found not to be likely. Section IV. B. 7 of this report refers. a. Sodium Thiosulfate The possibility of contamination of the RCS by sodium thio-I sulfate was investigated and found to be probable. This investigation is reported in Reference 1. A summary of the results of that investigation is provided here. It has been determined that liquid f rom the sodium thiosulfate tank reached the suction piping of the Reactor Building (RB) spray pumps by two possible flow paths. The first is leakage l through the BS-V-4A/4B valve s. The second is past the BS-V-4A/ E 4B valves when cycled during the valve surveillance section of S.P. 1300-3A. Chemistry results of samples taken in the RB spray suction and discharge piping in January 1982 show high I concentrations of sulfates (14 ppm). It is probable that small quantities of sodium thiosulfate reached the RB spray piping throughout plant life. Prior to the long lay up following the i TMI-2 accident, frequent cycling of the RB spray pumps inter-spersed with Borated Water Storage Tank (BWST) cleanup periods prevented dangerous accumulation of the chemical. 1 I

I E III-3 i When the RB spray pump surveillance S.P. 1300-3A is performed, each spray pump is placed in the recirculation mode taking I suction f rom the BWST and discharging back into the BJST. This would pump sodium thiosulfate in the RB spray piping into the BWST. This method was used on 6/25/81, 8/30/81 and 9/3/81 for a total of 12.2 hours in the time period since the last re-I fueling. Analysis of the contents of the BWST in January 1982 showed .1 ppm of sulfate s. However, the BWST was placed on cleanup through a precoat filter between the last surveillance I of the RB spray pumps and the time the sample was taken. This leads to the potentist of much higher sulfur contents in the BWST prior to cleanup, possibly as high as 850 ppb. Sodium thiosulf ate in the BWST could be injected into the RCS by using the BWST for normal RCS makeup, by performing the re-fueling interval surveillances S.P. 130 3-11. 54, S.P. 1303-11.8, I or by leakage. In May of 1981 the BWST was used to add approxi-mately 2000 gallons of water to the RCS. This was prior to the time that the RB spray pumps were run for the surveillance. During September of 1981, near the end of hot functional testing I periods and af ter the RB spray pump surveillance was run, water was injected into the RCS from the BWST while performing T.P. 665/1 and S.P. 1303-11.8 High Pressure Injection (HPI) and S.P. I 1301-11.54 Low Pressure Injection (LPI) functional testing. These tests would have injected sodium thiosulfate into the RCS at various temperatures. The last ficw path from the BWST to i the RCS is leakage past DH-V-5A/5B. These valves have had a past history of leakage. At least twice they have been repaired to try to obtain a tight seat, and leakage through these valves was observed in the period of interest (during the integrated 1 leak rate test). Pressurizer level increased from 208" to 376" between 9/10/81 and 9/27/81 due to this leakage. g Chemistry results have been reviewed which indicate unexplained 3 increases in RCS conductivity and pH between 7/18/80 and 7/25/80 and again between 5/11/81 and 5/18/81. The s e t ime frame s coin-cide with the addition of water f rom the BWST to the RCS. The I operational history does not support the proposition that the BWST was contaminated by sodium thiosulf ate prior to those addi-tions; or that the piping involved in the water transfer was I contaminated. The possibility that unrecorded plant operations occurred which cross contaminated the RCS from the RB spray system can not be discounted, given the limited information in plant logs. To preclude future problems with sodium thiosulfate contamina-tion and since the system is no longer required for its origi-I nally intended purpose, the sodium thiosulfare tank and asso-ciated piping has been drained and flushed. I i

I E III-4 b. Sulfuric Acid I The possibility of contamination of the RCS by sulfuric acid was investigated and found to be probable. This investigation is reported in Reference 1. The results are summarized here. On 9 October 1979, 3000 milliliters of sulfuric acid (98%) were added to approximately 37 to 49 gallons of water in the lithium hydroxide mix tank (CA-T3). This solution was pumped into the I makeup system using the lithium hydroxide pump (CA-P2). The tank was filled again with approximately 25 to 37 gallons of reclaimed water which was subsequently pumped into the makeup system. (The acid was to have been added to the neutralizer mix tank (WDL-T8) in the liouid waste disposal system via the caustic mix tank (CA-T2). The inadvertent addition resulted from using the lithium hydroxide mix tank instead of the caustic mix tank.) The error was not recognized until October 13, 1979. Samples were taken at several locations in the makeup system on October I 14, 1979, in an attempt to determine the location of the acid and establish boundaries for a system flush. A special operating procedure was prepared to flush the makeup system. I The lines were flushed with reclaimed water until the pH was greater than 5. A second flush with borated water was performed to flush the lines to the RCS. The flushes were performed from October 14 through October 25, 1979. In order to verify that the acid did not reach the RG, a sample was taken on 1 November 1979 and analyzed by B&W. The analysis I indicated less than 0.66 ppm sulfur (less than minimum detectable with the analysis used). I Although preventive actions were taken, it appears that some of the acid may have reached the RCS. The initial sample taken at MU-V-207, which ties in just upstream of MU-V-18, indicated a pH of 1. This indicates flow through this portion of the system. I The position of MU-V-18 at the time of the acid addition is unc e rt ain. The last valve lineup that was signed off on 9/7/79 indicates that MU-V-18 should have been open, however, it was I not signed of f; a "CR" was placed next to the position in-dicating control room control of valve position. The shift foreman's log indicates that the positions of MU-V-78 and MU-V-47 we re cha nge d in o rd e r t o ad d wat e r f rom "B" re ac t o r i coolant bleed rank to raise pressurizer level on 9/7/79. This operation would have required MU-V-18 to have been opened. No guidance is given to the operator as to what to do with MU-V-18 af ter the operation is complete, he may shut it or leave it I i

I I III-5 open. MU-V-78 and MU-V-79 were left in their fill lineup posi-tions which prevented the acid from going directly to the makeup I rank. The initial flush procedure does not require operation of MU-V-18 for the first flush path. This flush path required the flush water to flow through MU-V-18, implying that MU-V-18 was open. If the valve was open, a continuous flow path to the RCS I existed. If the valve was shut, a restricted flow path existed due to leakage past MU-V-18. I Samples for pH were also taken at MU-V-143B and MU-V-159B which are downstream of MU-V-18. MU-V-143B had a pH=5. MU-V-159B had a pH=6. If check valves MU-V-107B and MU-V-94 did not leak, I then the flow of water for these samples would have been from the makeup system. Since the water had a pH=1 at MU-V-207, this flow would draw the low pH water closer to the RCS and possibly past MU-V-143B when sampling MU-V-159B. If the check valves did i leak, the low pH water may or may not have been drawn into this line. I The first flush path of the initial flush procedure supplied reclaimed water at MU-V-143B and drained through MU-V-207. If low fii Water was in the line between MU-V-143B and MU-V-159B, t he pressure of the reclaimed water would have been sufficient to drive this water into the RCS. pH samples taken at a drain on each of the makeur pumps also I indicated pH=1, indicating flow through this portion of the w system. Since flow from the lithium hydroxide pump is one-sixth of a gallon per minute, there is little driving force to seat check valves in the system. The acid would have had to have passed through MU-PlA and MU-PlB discharge lines to the pumps. 14akage past the bearing seals nuld account for some flow and MU-V-12 possibly being open may..ve accounted f or additional I flow. MU-V-76A, MU-V-768, MU-V-69A and MU-V-69B are normally locked closed and there is no evidence of changes in these valve positions. Therefore, for acid to reach MU-Plc would require flow through the pumps' recirculation lines and/or through I leaking valves. Some flow of the acid through the makeup punps may also have occurred during the flushes on other parts of the system. Review of the RCS chemistry for this period indicates a signifi-cant change in pH and conductivity with essentially no change in I boron concentration. Calculations were performed to estimate the volume of acid necessary to cause the change in pit. The volume of acid was relatively small, being approximately 0.4 liters of the concentrated acid. ll

I I III-6 Calculations predict a sulfur concentration that should have been detectable by the BW analysis that was performed. How-ever, that sample was obtained on 1 November 1979 which was af ter approximately 15,000 gallons of water was added to the Reactor Coolant System. This addition of water would have re-i duced the sulfur concentration to a value below the minimum detectable concentration. As suming that sulfuric acid in the amount indicated did find its I way into the RCS, it is unlikely that the attack on the OTSG tubes resulted directly. The sulfur in sulf uric acid is in the form of sulfate which is not an aggressive species at ambient i temperature in the reactor coolant environment. There is a possibility that it could be aggressive at elevated tempera-tures. Also, in the presence of sodium thiosulf ate, a reduction in pH could activate the corrosion mechanism. c. Oil The possibility of contamination of the RCS by oil has been investigated. Findings indicate that such contamination is likely to have occurred in March 1979. Probable overfilling of the miscellaneous waste storage tank is a likely mechanism by which oil might be introduced into a reactor coolant bleed tank and subsequently into the RCS, I Another mechanism considered was the possible contamination of the. nitrogen cover gas by oil. This has been found not to be the case. The nitrogen supplied to TMI-l is not produced in such a way as to allow oil contamination. If in f act oil were introduced into the RCS in the amounts pos-sible, it is unlikely that it would by itself result in the observed attack on the tubes. The quantity of sulfur available is too low. Reference 1. details these findings. 3. Timing of the Tube Failures The time of the OTSG tube f ailures may be bracketed based on opera-tiona naside ra tion s. During hot f unctional testing on 9/4/81 at 0637, the leak rate of the RCS at full pressure was measured and f ound to be within specifications at 0.5 gallons per minute. On 11/ 21/ 81 with the RCS at about 40 PSI pressure leakage through the OTSG tubes were observed. 8, One <ither incident which might bear on the timing of the tube f ailures was an abnormal depletion of hydrazine on the secondary side of

he OTSG's.

In the A OTSG between 9/9/S1 and 9/16/81 hydrazine decreased !I ~ f ron 92 ppm to 16 ppm. In the B OTSG between 9/17/81 and 9/23/81 hydra-ziae decreased f rom 68.75 ppm to 0.012 ppm. These decreases were at-l tribu teri to wort on the OTSG AEW piping which was going on at the time. l i

I E III-7 A possible scenario is that failed OTSG rubes allowed negative pressure on the primary side containment to draw air into the secondary side despite the normal nitrogen purge. 4. Consideration of Asymmetric Failure Patterns The radial distribution of the failures in each OTSG has some similarity. Consideration was given to the fact that the distribution of Auxiliary Feedwater nozzles around the periphery of the OTSG roughly I corresponds to the distribution of the failures. The Auxiliary Feedwater nozzles at TMI-l were used only once. During the loss of power testing in 1974 flow was initiated through this path when the primary temperature was about 550*F. The nozzles have not been required to be used since. I It is unlikely that the location of these nozzles had any operational effect on the tubes. I The radial distribution of the failures also displays some dis-similarit ie s. The failures in the B OTSG occurred predominantly in the side opposite the recirculation return nozzle on the secondary side. Consideration was given to the possibility that recirculation of the I secondary side may have cooled the tubes near the return line causing condensation and dilution of the agressive environment on the primary side. The A OTSG however was recirculated first following the cooldown I f rom Hot Testing. Additionally, the recirculation flow is not cooled so it does not appear likely that condensation on the primary side would re su lt. The failures in the A OTSG occurred predominantly in the I periphery where the tensile stress is higher. In the absence of any clear difference in the material and environment between the A and B OTSG and in the presence of the stress dependence in the A OTSG, it may be inferred that distribution of failures in the B OTSG is stress related. The fabrication history of the B OTSG dcsc :.ot s plain the apparent asymetry in tensile stress in the tubes. One possibility is that the I whole steam generator is slightly bowed. Another is that the upper and lower tube sheets are not precisely paralle l. In an attempt to confirm asymetry of tube tensile stress, a study is being conducted by Babcock and Wilcox to determine the feasibility of measuring the relative tensile I stress in the tubes by measuring the tube's fundamental f requency of transverse vibration. D. Conc lu sions Consideration of the operational history supports the hypothesis that the attack on the OTSG tubes was promoted by sulfur introduced into the RCS as I sodium thiosulfate. The probable earlier introduction of oil could not have produced the levels of sulfur contamination observed. The sulfuric acid introduced sulfur in its fully oxidized state which does not promote I intergranular stress cracking at ambient temperature in the reactor coolant e nvironme nt. Clean up was conducted prior to heat up. lI J

I I g Figure 111-1 R.C. System Drain Arrangement I NITROGEN , RC-V28 3

R.C. DRAIN TANK p

) a a n RC V18 RC V19 ' RC-V14A ' RC V14A I X RC VISA RC V158 RC V17 Y WDG V71 1 I FLEX 1BLE VENT -- LINE E em A s B V e e O O I R,C V21, x RC V6B RC V6A y RC-V60 RC-V6C RC-V20m . RC V7B X RC V7A f RC V7D3 RC-V7C r P P e e Q X WDL-V307 I WDG V2 hQ a,) g 374 W n y WDL-P16 WDL V304 WDL-V56 I AIN TO MISC. WASTE TANK STORAGE TANK " WD V303 TO REACTOR N WOL V161 COOLANT BLEED WDL-V302 WDL V300 TANKS l J WOL V301 I e I '~ @ WDL V305 ^ WOL-P8 5 .t i

I E IV - 1 I IV. OTSG CHEMISTRY A. Introduction Investigation into the chemistry history surrounding the failure of I OTSG tubes concentrated upon reactor coolant chemistry in the period between the cooldown of the system in early April 1979, and identification of primary to seconda ry leakage in November 1981. The ef f o rt s we re f u rthe r focused upon parameters and situations which might provide indications of I introduction of sulfur or sulfur containing substances into the system. Justification of this direction for intial efforts was based upon initial results from the metallurgical examinations. The first influencing factor I was that the failures were initiated from the ID of the tubes. The second f actor was that sulfur was identified as a major contaminant present on f racture surf aces. Although lower levels of chlorine were also identified, Inconels are virtually immune to chloride ion stress corrosion cracking. B. Methods and Results 1. Comparison to Specifications Initial review involved a simple comparison of historical data with I specifications. In general, compliance with the specifications was good. During the entire period, departure f rom specification was noted only in the chloride parameter with the highest recorded value being 0.5 ppm which was noted on 6/18/79. Values of 0.01 ppm and 0.06 ppm were recorded on 6/15/79 and 6/22/79 respectively. Since no treatraent was undertaken during this period, it appears that a contaminated sample or analytical interference was responsible. Several of the I other chloride values in the range of 0.1 ppm to 0.5 ppm were sur-rounded by similar cocditions. In any event, industry experience has demonstrated that these levels of chlorides, even if actually present, would not produce chloride stress corrosion cracking when present for I only a f ew days at ambient temperatures. 2. Historical Sulfur Measurements It became apparent during the review of the historical data that no direct measurements of sulfur or sulfur compounds in the reactor I coolant system were available on a regular basis. Some measurements were performed on samples from various points within the spent fuel and rad waste systems which included two samples f rom the decay heat system and one from the makeup system, in support of the investigation of I cracking in spent f uel cooling system piping in late 1979. In addi-tion, a few determinations were made in conjunction with the sulfuric acid induction incident. All of these measurements appear to have been performed by a gravimetric technique with no preconcentration step and, as such, results below 1 to 2 ppm are questionable. I I

I IV - 2 3. Indirect Indications of Sulfur Contamination Efforts turned to other parameters which might provide indirect evidence of introduction of ionic species which could not be accounted I for by the specific analyses which were performed. Specific con-ductivity appeared to be the best such indicator. Boric acid and lithium hydroxide are known major contributors to the system's specific I conductivity and are added intentionally. Other contributors for which monitoring is conducted are sodium and chlorides. While an exact prediction of specfic conductivity resulting from a given combination of these substances requires calculation of the complex chembal I equilibrium which results, a few of the values recorded in the period were sufficiently high to require contribution from unmeasured ionic species. Such unexplained increases occurred in October 1979, July 5 1980, May 1981 and November 1981. The October 1979 increase can be attributed to the sulfuric acid induction incident and the November 1981 increase to contamination from inleakage of secondary fluid I through cracked OTSG tubes. The increases in July 1980 and May 1981 both involved transfer of wattr from the BWST to one of the decay heat systems. In January 1982, significant levels of sulfur were found in piping which interconnects with this flow path. In addition, the analyses have indicated that a major portion of this sulfur was in the form of the thiosulfate ion. Figures IV-1 and IV-2 show some of the key chemical parameters for these two periods. 4. July 1980 Incident In Figure IV-1, the pronounced conductivity and pH increases I between 7/18/80 and 7/25/80, while other parameters exhibit little or no change, are significant. These changes can be explained by the addition of small quantities of the mixture of sodium thiosulf ate, I sotiua hydroxide and boric acid which was contained in the sodium thiosu2 fate tank. All parameters had returned to normal by 8/8/81. The reductions in specific conductivity and pH can be attributed to processing of the decay heat system through a resin coated precoat I filter and <aixed bed demineralizer from 8/4/80 to 8/5/80 and through a resin coated precoat filter from 8/5/80 to 8/11/80. Removal from the bulk water of virtually all of the intruding ionic species is indicated. 5. May 1981 Incident Similar increases were noted between 5/11/81 and 5/18/81 and are shown in Figure IV-2. This incident was not, however, immediately followed by a cleanup of the decay heat system and it appears that part I of the intruding ionic species remained in the system for an extended period. A resin coated precoat filter was employed from 8/9/81 to 8/11/81 and resulting drop in specific conductivity indicates removal I of approximately 60% of the intruding ionic species. A subsequent test conducted at GPU System Laboratory to determine what levels of con-tamination by the mixture from the thiosulfate tank would produce a specific conductivity change of 11 umho in a simulated reactor coolant solution indicated that 6-7 ppm as sodium thiosulfate would have been I

I IV - 3 re quired. This corresponds to 4-5 ppm as thiosulfate ion. Assuming this intial level, approximately 1-2 ppm as thiosulfate would have I remained in the system until the fill and vent process at the beginning o f RFT. 6. Oil Introduction An additional incident which has a potential for inducing sub-stances which contain sulfur into the system occurred a few days before I the TMI-2 accident. Oil or oil-like substances were discovered in auxiliary systems which have direct communications with the reactor coolant system. Extensive documentation of this incident is not avail- ,g able, probably because of the shif t in priorities following the acci- 'E dent. One value of 2.8 ppm as oil and grease was recorded in the BWST bulk water and a footnote in a chemistry log book indicates the presence of oil in a sample from a RCET which also appeared to have a very high crud content. a. Crud Measurements Selected crud values from Reactor Coolant Bleed Tanks A, B, and C are presented in Table IV-1. It should be noted that values signi-ficantly greater than the normal _ 0.1 ppm were recorded as late as I 9/5/81. The analytical technique employed for crud analysis is a gravimetric one and the filters are dried at approximately 105'C, a temperature which would volatilize little, if any, of the hydro-carbons in oil or grease. b. Recent Oil and Grease Results In addition, oil and grease analyses were perfomed en a number of samples taken from the decay heat and associated primary systems during March and April of 1982. The results are presented in Table 'I IV-2. Significant levels of oil and grease were recorded in RCBT A. RCBT B, and the BWST. These levels were observed in samples dipped from the surface and should be regarded only as quali6ative I evidence of the presence of oil or organics and not as a quantita-tive measure of these substances in the bulk liquid. 7. Resins as a Source of Sulfur Consideration was given to the possiblity that breakdown of plant demineralizer resins might release sulfur to the RCS. hio concepts I were examined. First, the resins in the demineralizer were con-s ide red. Only the cation form may release sulfur due to radiation damage and it is very hardy, requiring greater than 108 rad to break down. Thus the resin in the demineralizer did not break down from known radiation ef fect s. Secondly, consideration was given to the possibility that resin escaped into the RCS where high temperatures could nave caused breakdown. There is no indication that this oc-I curred. In particular, crud measurements on the RCS did not reflect the presence of resin or resin fines. I a m m

I I IV - 4 i C. Conc lusions The following conclusions may be drawn: 1. Generally, the reactor coolant system remained within specifica-tions for those parameters for which an analysis requirement I existed for the period extending from April 1979 through Ibvember, 1981. I 2. Two incidents of intrusion of ionic substances not accounted for by specific analyses have been identified. Sodium thiosulfate at levels of 4-5 ppm as thiosulfate is considered to be the most likely contaminent. 3. The ionic species from the first contamination incident in July, 1980, were removed frc.n the bulk liquid by demineralization in August, 1980. 4. The ionic species from the second contamination incident in May, i 1981, appear to have been only partly removed by processing through a resin water precoat filter in August, 1981. A 1-2 ppm thiosul-fate residual could have still been present at the start of fill and vent prior to hot functional testing in September, 1981. 5. Induction of an unidentified organic substance, probably oil or grease, into RCS auxiliary systems occurred in mid-March, 1979. I The exact quantity of this substance or its potential role in any failure mechanism has not been established. Reference 1. documents this incident. D. Chemistry of Sulfur in Water Since the matallurgical results indicate that sulfur was involved in I the attack on the OTSG tubes, a brief discussion of sulfur chemistry is provided here. Because there is a large number of species that can form in the S-O-H system in aqueous solution it is necessary to briefly explain the chemistry of the sulfur-water system to provide a basis for assessing I whether or not aggressive species were likely to be present. ( A full dis-cussion of aqueous sulfur chemistry is presented in reference 4.) Table IV-1 summarizes the names, structures, and sulfur oxidation numbers of .l various species that can fonn in the S-0-Il systems. In general, the higher 5 oxidation number species are stable at the more positive (anodic) elec tro-chemical potentials, and the lower oxidation number species at the more negative (cathodic) potentials, but many species are metastable. . I I 5 lI

I IV - 5 SO -~ is the equilibrium species at room temperature in 4 oxygenated water at pH-5 but H S (S--) is the stable species 2 if the water is hydrogenated. At intermediate potentials there is I a small region where elemental sulfur is the stable form. This region is believed to disappear at higher temperatures but the equilibrium thermodynamics are much less well established at 300*C than at 25'C. Although S0[ and S-~ are generally the dominant equilibrium species within the pH and temperature ranges of interest, small equilibrium concentrations of other species I species are HS0 '_HS 0}, also exist. Among these less dominant 3 2 a nd HS. There are also numerous polysulfide species (S 2 2' S ', S~~ etc.) which become more important if the sulfur electrochemical 3 activity is high and the potential is low. In addition to equilibrium species, metastable forms also have to be considered. In the pH I range of interest SOcan exist at high potentials and S 0}^ at lower 2 potentials. I Determining the amounts of these species likely to be present in solution is difficult because of lack of knowledge about the kinetics of the various oxidation, red uc t rui, and disproportion-I ate reactions involved. At low temperature, the rate of reduction of SO'[ at low potentials (where it should go to S~~) is so slow that S0[ seems to persist indefinitely. Solutions of I S0 and other metastable oxyanions are also quite stable at room 23 temperature although from equilibrium considerations they should disproportionate to S and S0[. Also, oxidation of sulfide (S~~) to sulfur is more rapid than f urther oxidation to oxyanions I like S03 and SO4 at room temperature. Kinetic inf ormation is sparse at higher temperatures but the data of De (reference 6) seem to S023 is extremely elec troactive above 150*C and indicate that that kinetics of the reactions between the various oxidation states I can be described as follows: I fast fast fast slow slow S(-2) Z S(0) Z S(+2) Z S(+2. 5) -+- S( F4) --> S(+6) fast fast fast I I I

TABLE IV-I AQUE005 SULFUrt $F. iES Sulfur Formula Str;cture 0xidacion Number

Name H S or 3

-2 sulfide 2 I H3'3 S-S -1 22 2 HS'3 S-S-S -2/3 polysulfides = -= 23 3 ,= E H3, S-S-- -2/x 2x 5 S rings 0 sulfur 8 I 9 ~ S0 0-5-S +2 thiosulfate = 4 p3 0 I ~ = 0 S0 0-46 0 I 0 S0 0-$-0 +4 sulfite 3 (sulfurous acid) S0 +4 sulfur dioxide 2 0Q S0 0- - -0 +5 dithionate 26 I SO " 0- -0 +6 sulfate 4 lI

0xidation number is the fcm=1 electrical charge assigned to the sulfur on the assumption that H is +1 and 0 is -2 in these compounds.

'I I

W W W W W W W W W W W W W M M M W W W NO. 341 10 DIETZGEN GWAPH PAPE R DI ETZ G E N C O N PO R AT8 D N 10 M 10 PE R 8NCH seams m es. o. a. 3000 N _ -t v 2000 ~~~~-~~~ 3I f FIGURE IV-1 [ .o oa Selected RCS ] O 1.0- ---- Chemistry Data 0 O O 01 .m . ~._ g a () _ Q..___ q. . cp g) <r .- (> .o 0.2-i). O o." M'--

]

a o ,m.s N ?. ..N L_._ o /. N D s N -.x f y . N 0.0 tr---- o m- -~~* 25 o ./ O 15 ap s A d t> ~---., 6.0 5g

o. -.-.-+__.._-4.

- q . %-~,__ PH .-4> y - ~'~g -'~ -'~~ .~ ~ ~ ~-~ "~~' ~~ ~ '~~ '~' ~

4. 0_

7/11 7/15 7/18 7/25 7/28 8/1 8/4 8/5 8/8 8/11 .lul y, 1980 August, 1980

NO. 341 - f D DIETZGEN GRAPH PAPER DIE TZ G E N C O RPOR ATIO N 80 X 3D PER INEH esaos He u. e a. l FIGURE IV-2 Selected RCS Chemistry Data 3000 ~ .tn

-+ -.--_____..a g

m. g.. 2000 0.1 .n g g g._._ -.o.._. g s .._q) 0.1 0 N o N H D' g j;) 2 m M_r. ~' s p / \\ / 0 ~~~~ ~'~ V 25 -3,.- 0 g_-}: _ p w .O., g a ._ g g 6.0 o 5 .g._ n

O

_..g . ____+__.___ ~ , ___.-q, _.-o ._4> 5/1 5/4 5/11 5/18 5/22 5/29 6/1 6 / 5' 6/3 6/17- !tay, 1981 .h nm,1981

I I V-1 V. OTSG AND PRIMARY SYSTEM CHEMISTRY SAMPLING A. Wipe Sampling Program 1. Introduction I A wipe sampling program was organized to aid in determining con-taminant source (s) and concentrations, as well as, to identify potential damage mechanism (s) in the TMI-l steam generators. All originally scheduled wipe sampling is complete. Analysis of wipe I samples collected has been completed for sulf ur and chlorine con-tent, and for spec trographic analysis. 2. Wipe Sample Analysis A wipe sampling program was initiated to establish the contaminant levels in the OTSG's (both primary and secondary side). Pa rt of each sample collected was analyzed f or chlorine using ASTM D-512-67 Me thod C. These same samples were also analyzed for I sulfur. Analytic procedures used were such that the sulfur was first oxidized to sulfate resulting in a measurement of total sulfur present. ASTM D-516-68 Method B Modified was then followed to determine sulf ur content. Finally, some of the collected samples were selected by GPUN/B&W for spectrographic semi quantitative analysis. 3. Results and Conclusions Results of wipe sample analysis are shown in Tables V-1 and V-2. Examination of the results of analysis resulted in the following conclusions: a. Sulfur Concentrations I 1) Tube ID wipe analysis revealed slightly higher con-centrations of sulfur in OTSG B than in OTSG A. 2) Tube ID wipes disclosed primary side sulf ate concen-I trations at the UTS were at least two times as high as concentrations at the LTS, in both generators. 8 3) Primary side sulf ate concentrations, based on OTSG upper plenum tube ID samples ranged f rom 970 to 3600 ug/f t Concentrations of sulf ate at the primary UTS surf aces, with only one UTS wipe sample from each generator in-I cluded in the data base for subsequent comparison, I

I I V-2 indicated concentrations of 770 and 930 ug/ft2 This 2 measured on compares with levels less than 220 ug/f t the tubes during f abrication. 4) Data indicates presence of sulfur at lower levels on the primary side tube surf aces in the OTSG's, thus there is potential for a sulfur residue on other surf aces in the I RCS system. Reference 13 reports the results of an inspection of the RCS. 5) Secondary side sulfur concentrations were generally higher than on the original secondary side tube surfaces af ter manuf acture but before gervice ranging f rom less than detectable to 3100 ug/f t See Table V-3 f or I OTSG-B tube hole ID's data and Table V-1 for OTSG-A data. b. Chlorides Present 2 Chloride levels were well below the 1000 ug/f t maximum limit established in the B&W cleanind specifications for RCS su rf ace s. c. Additional Sampling The f ollowing wipe samples are presently at LC in the process of being analyzed. I 6 CRDM Leadscrew samples 4 RC Bleedtank samples 1 Vent header sample 4 Retainer (taken at LRC) wipe samples I 1 Retainer (taken at Site) sample 1 clean wipe (Blank) 2 RV 0 ring swipes Preliminary results indicate that sulfur levels sitilar to these in the OTSG tubes exist in other areas of the RCS. B. Water Sampling Program 1. Introduction As a result of the tube damage in the TMI-l steam generators, a water sampling program was established to provide input to damage I mechanism identification and to aid in determining the contaminant sou me. All currently scheduled water sampling and analysis is completed. I I I

I I V-3 2. Water Sample Analysis Analyses perf ormed on f racture surf aces of pulled tubes revealed I the presence of sulfur. A water sampling program was initiated to provide data on sulfur in the RCS and auxiliary systems. The method of analysis for sulfur used by B&W on the water sample is B&W procedure #718T-75, which is modified ASTM D-516-80 Method B, I performed af ter preconcentration to improve sensitivity. This procedure is turbidometric determination of total sulfur af ter oxidation with bromine water. The sulfate form of the sulfur I present may be determined by this method without the oxidation step if the concentration of sulfur in the unconcentrated sample is higher than about 500 ppb. Reduced sulfur may also be deter-mined with a non-specific titration using a standard iodine solu-I tion. Analysis performed by Westinghouse was by ion chromo-tography. Minimum sensitivity of this Westinghouse technique is about 50 ppb. I 3. Results and Conclusions Water sampling ef forts began in December when two secondary side I OTSG samples and one decay heat sample were taken and analyzed. The results of these analyses are given in Table V-4. High decay heat sample sulfur levels in the December decay heat sample (730 I ppb), plus indications of sulfur in areas of OTSG primary side tube defects prompted taking additional RCS/ decay heat water samples. Twenty-one more water samples were taken in January and early February from various locations in the RCS and analyzed by I B&W. Table V-5 contains the analysis results from these water samples. Total sulfur is reported for each sample. When appre-ciable levels of total sulfur were reported, (samples from the I Reactor Building Spray System), reduced sulfur is also reported-given as sulfate. I In March and April, samples were analyzed by Westinghouse. Table V-6 contains the analysis result s. There are a number of comments to be made based on sample analysis data: a. Water samples f rom the decay heat system showed sulfur level I reduction between December and February (.7 ppm - Dec.,.4 ppm - Jan., .1 ppm - Feb.). There is no way to indicate what peak RCS sulfur levels were reached prior to sampling. Reduction of RCS decay heat sulfur level is attributed to (1) I flushing of the system line s, (2) dilution as a result of leakage from the BWST ware.r into the system and (3) system cleanup. I I I

I I V-4 I b. RB Spray System sample points all indicated elevated sulfur concentrations (14 ppm to 176 ppm as SO ) much higher than 4 other sample point s. Solution pH reading were also taken from four spray system samples as shown in Table V-8. These four samples (Table V-6, #7 through 10) were analyzed for reduced sulfur. Significant amounts of reduced sulfur I were shown in the RB Spray System interconnect lines to Decay Heat System (161 ppm - DilA, 4 ppm DHB). Since the RB Spray System showed by far the greatest sulfur concentrations of I samples taken, it is identified as a possible sulfur intro-duction point into the Decay llear System. Also the presence of appreciable quantities of reduced sulfur ties in with a damage mechanism of sulfur attack at lower temperature s. I c. The caustic mix tank water sample showed 7 ppm sulfare. Due to the high ratio of sodium to sulfur, which prevents the I formation of an acid environment, and due to the dilution factors involved, the caustic mix tank proves an unlikely source of sulfur contamination to the RCS/ Decay Heat system, d. The ion chromotography results were subject t o int e rf e re nc e and generally yielded more erratic results than the turbidi-metric technique. One set of samples in early March was I contaminated by the lids of the scintillation borries and made unusable. e. For comparison betweer the TMI-l result s and industry ex-I perience, some RCS/ Decay Heat water samples were taken from other B6W operating units (see Table V-7). These showed levels of sulfur below minimum detectable concentrations. I I I I I I I

m e m e e m e e M M M M M M M M TABl.E V-I RPE SAMPLE ANALYSIS Date Tubesheet Sample Analysis Sample # Taken OTSG Tube I.D. Tube llole I.D. General Area Location CL (pg/f tz) 2 SO (pg/f tj 4 1 1/13/82 A R151T7 Upper Plenum 470 3000 2 1/13/82 A R4T2 3 1/13/82 A RIT5 110 970 4 1/13/82 A R148T24 5 1/13/82 A R6T14 (70 2500 6 A R139T15 4 70 1100 7

==-NOT TAKEN 8 1/14/82 B R3T23 Upper Plenum ( 70 2400 9 1/14/82 B R72T2 10 1/14/82 B R8T22 4.70 3600 11 1/14/82 B R2T16 12 1/14/82 B R3T11 c 70 3000 13 B R7ST3 C-70 2300 14 1/14/82 B W-axis 56 930 15 1/15/82 A Near W-axis Lower 4 70 5230 16 1/15/82 A 17 1/15/82 A A 70 .,230 18 1/15/82 A Y-axis t 70 1500

WIPE SAMPl.E ANALYSIS Date Tubesheet Sample Analysis Sample # Taken OTSG Tube I.D. Tube Hole I.D. General Area I.ocation Cl.(ug/ftZ) S0,(ug/ft2L 19 1/15/82 A Near Y-Axis Lower Plenum 20 A <.70 580 21 1/15/82 A W-Axis 22 1/15/82 A Y-Axis 10 250 23 1/15/82 B Near W-Axis ( 70 c280 24 1/15/82 B 25 1/15/82 B 26 1/15/82 B %70 1700 27 1/15/82 B Z-Axis (70 <.280 28 1/15/82 B 470 470 29 1/15/82 B W-Axis 30 1/15/82 B Z-Axis 10 220 31 1/15/82 A 71-126 Upper Plenuta 4 60 540 32 1/20/82 A 112-7 350 1100 33 1/20/82 A 11-66 34 1/20/82 A 16-69 350 820 35 1/20/82 A 12-62 36 1/20/82 A 13-63 (60 <.240 37 1/20/82 A 112-5

g g g g e e e e e M M M M M S3 W TABI.E V-1 (cont'd) 111PE SAMPI.E ANALYSIS Date Tubesheet '; ample Analysis Sjugp l e" Taken OTSG Tube I.D. Tube llole I.D. General Area Location CI.(ug/f t2) S0 (ug/ft2) t 38 1/20/82 A 112-9 Upper Plemm 350 2200 39 1/20/82 A 68-7 40 1/20/82 A 88-11 180 3100 41 1/20/81 A 62-8 42 1/20/82 A 133-74 82 (240 43 1/20/82 A 146-8 44 1/20/82 A 146-6 380 650 45 1/20/82 A Y-Axis 63 770 46 1/20/82 A W-Axis 47 1/20/82 A 23-93 l I 6

I Table V-2 EMISSION SPEC RESULTS ,I BLANK

2
9 Ash Weight (mg), blanked

.9 4.9 3.8 I Area Wiped (ft2) .145 .145

I Emmission Spectroscopic Results, %

1, 13 Compound

Blank

2
9 Zno 2 wt %

.5 wt % .5 wt % Na 0

1. 0

.3 .3 2 Cu0 .1 .1 .2 Mo0 <.03 .2 .3 3 V0 .03 (.03 <.03 25 Ca0 714 4 5 I TiO .07 .2 .3 2 Al 0 23 coo 4.03 .03 .03 Zr0 .1 .7 .7 2 Fe 0 1.5 13 17 23 Sn0 .08 .1 .1 2 .m Ni0 4.03 20 23 l l Cr 0 (.03 15 10 23 l Mn0 .3 .2 .3 2 Pb0 .03 4 03 <. 03 Mg0 > 25 1.8 3 l SiO > 25 4 5 2

Compounds listed are based on standards used.

" e form of the elements in the samples may be dif ferent. i 1I l

I Table V-2 (Continued) EMISSION SPEC RESULTS BLANK

19
24
33
43 Ash Weight (mg), blanked 1.0 4.2 1.4 37.4 11.5 Area wiped (ft2)

.145.145 .170 .170 I Compound

Blank

19
24
33
43 Zn0 3 wt %

.8 wt % .8 wt % <. 2 wt % .2 wt % Na 0 .8 .5 .4 <.3 .4 2 Cu0 .2 .1 .2 .03 .08 Mo0 <.03 .5 .4 .2 .5 3 v0 <. 0 3 .05 <.03 <.03 <.03 25 Ca 0 >14 4 6 .4 1.8 TiO .07 .9 .5 1.9 .3 I 2 A1 0 1.2 .5 .6 .1 .7 23 Co 0 <.03 .06 .05 4 03 (.03 Zr0 .1 3 1.3 <.03 4 03 2 Fe 0 1.5 >25 20 >25 > 25 23 Sn0 .09 .2 .2 <. 0 3 <. 03 2 Ni0 <.03 >25 11 8 10 Cr 0 <.03 25 11 2 ? 23 Mn0 .3 .5 .4 .8 .7 2 Pb0 .03 <.03 .03 <.03 .03 Mg0 >25 2 6 3 1.3 SiO 25 4 9 .5 3 2

Compounds listed are based on standards used.

The form of the elements in the samples may be different, i

I

Table V-3 lg TMI -1 WIPES (DECEMBER 1981) la TUBE HOLE ID'S AT UTS l OTSC B 1 i SAMPLE TUBE DEPTH WIPE TAKEN SO4 (ug/ft ) 1. 8-25 3 in. 2800 2. 8-25 8 in. 3300 3. 10-25 3 in. 1900 4. 11-23 3 in. 2800 5. 11-23 8 in. 2000 1 I 6. 33-30 8 in. 2000 i !I i i a i !I l l l 1 I ) i 1

!I i i Table V-4 TMI-l WATER SAMPLES FOR TOTAL SULFUR ANALYTICAL RESULTS ) j TOTAL SULFUR SAMPLE AS SULFATE (ppm) e i RCS ,7 i OTSG-A .3 I OTSG-B .3 4 iiI iiI i i,s SAMPLES TAKEN 12/01/81 i I!I (I .I I I

Table V-5 TMI-l WATER SAMPLES ANALYTICAL RESULTS TOTAL SULFUR (ppm) REDUCED SULFUR (ppm) CAMPLE DATE TAKEN (AS SULFATE) (AS SULFATE)

1. Decay Heat A 1/18/82

.4 PPM

2. Decay Heat B 1/18/82

.4 PPM

3. Core Flood A 1/19/82 4.1 PPM

4. Core Flood B 1/19/82

<.1 PPM

5. Spent Fuel Pool 1/18/82

.4 PPM

6. Borated Water Recir.

Pump Discharge 1/20/82 .3 PPM

7. R.B. Spray Pump Inlet 1/20/82 14 PPM 41

8. R.B. Spray Pump Outlet 1/20/82 15 PPM 1

9. R.B. Spray System Inter-Co nnec t to D.H.A.

1/20/82 176 PPM 161

10. R.B. Spray System Inter-Connec t to D.H.B.

1/20/82 14 PPM 4

11. Make-Up Pump Outlet A 1/20/82

<,.1 PPM

12. Make-Up Pump outlet B 1/20/82

.3 PPM

13. Caustic Mix Tank 1/20/82 7.0 PPM

14. Lithium Hydroxide Mix Tank 1/20/82

(.1 PPM

15. Boric Acid Mix Tank 1/19/82

(.1 PPM

16. Borated Water Storage Tank 1/20/82

<.1 PPM

17. Demin. Water Storage Tank 1/19/82

<.1 PPM

18. Reclaimed Water Storage Tank 1/19/82

(.1 PPM

19. Bleed Hold-Up Tank B 1/19/82

(.1 PPM

20. Bleed Hold-Up Tank C 1/20/82 4 1 PPM

21. Decay Heat Removal A 2/4/82 f.1 PPM

22. Control Sample 4 1 PPM

Not Analyzed Due To Low Concentration.
- Not Analyzed Due To High Caustic Concentration.

Table V-6 ION CHROMOTOGRAPHY RESULTS SAMPLE POINT DATE SO4 ppb REMARKS Decay Heat A 3/18 123 IWT Mixed Bed 3/17 246 Precoat A inlet 3/17 246 Precoat A outlet 3/17 (Interference Peak) Bldg. Spray Pump A inlet 3/17 2465 Pre-Flush Bldg. Spray Pump B inlet 3/17 (Interference Peak) Pre-Flush Bldg. Spray /BWST Line 3/18 822 Pre-Flush Bldg. Spray Pump A outlet 3/13 2493 Pre-Flush Bldg. Spray Pump B outlet 3/17 2876 Pre-Flush I Bldg. Spray Pump A inlet 3/20 752 Post-Flush (2.5 Volumes) Bldg. Spray Pump B inlet 3/20 656 Post-Flush (2.5 Volumes) Bldg. Spray /BWST line 3/20 64 Post-Flush (2.5 Volumes) Bldg. Spray Pump A outlet 3/20 546 Post-Flush (2.5 Volumes) Bldg. Spray Pump B outlet 3/20 764 Post-Flush I (2.5 Volumes) Pre-coat A inlet 3/20 149 end of third run Pre-coat A outlet 3/20 74 end of third run I BWST 1/20 257 repeat on sample analyzed by B&W Spent Fuel pool 1/18 559 repeat on sample analyzed by B&W Decay Heat A 2/4 211 repeat on sample analyzed by B&W Precoat A inlet 3/23 161 Fourth run I

I Table V-6 (Continued) I ION CHROMOTOGRAPHY RESULTS SAMPLE POINT DATE SO4 ppb REMARKS I Precoat B outlet 3/23 416 Fourth run I Bldg. Spray Pump A 4/1 369 Post-Flush (6 Volumes) l I Bldg. Spray Pump B 4/1 269 Post-Flush (6 Volumes) Bldg. Spray Pump A inlet 4/1 1400 Post-Flush (6 Volumes) Bldg. Spray Pump B inlet 4/1 276 Post-Flush (6 Volumes) Bldg. Spray /BWST line 4/1 269 Post-Flush (6 Volumes) I I I I B I I I I

IlE I Table V-7 WATER SAMPLES j TOTAL SULFUR ANALYTICAL RESULTS ] TOTAL SULFATE j SAMPLE AS SULFATE (ppm) Oconee 1 RCS (1) {.1 Oconee 3 RCS (.1 Another B&W Unit:(2) f RCS 4.1 i RCS 41 Make-up (.1 Make-up 41 .E i Decay Heat Cooler Outlet (.1 i

I i

j (1) Analysis completed March 12, 1982 l (2) Analysis completed March 3, 1982 lI e I I I I I

I I VI - 1 I VI METALLURGICAL EXAMINATIONS A. Introduction In order to positively identify the cause of steam generator tube I f ailure, extensive metallurgical examination of tube samples removed f rom both OTSG's was conducted by Babcock & Wilcox and Battelle Columbus La bo ra to rie s. The laboratory investigation was intended to identify the I f ailure mechanism, extent of tube damage, and possible environmental factors contributing to the f ailures. The metallurgical examinations together with the results of the primary system water chemistry analysis (Section V) suggest that the OTSG tube cracking initiated f rom the prima ry side and is a manifestation of sulfur-induced intergranular stress assisted cracking (IGSAC). Accordingly, a brief review of current knowledge relating to sulfur-induced IGSAC is presented here to provide a basis for the subsequent discussion. B. Attack of Inconel 600 by Water Containing Dissolved Sulfur Species 1. Industry Experience A state of-the art review of relevant industry experience of attack I of Alloy 600 by aqueous environments containing dissolved sulfur species has been perf ormed f or EPRI by several EPRI contractors (Reference 4). A number of nuclear plants have experienced steam generator tube cracking problems attributed to the introduction of sulfur species into the I secondary system; also tube pitting, intergranular attack, and wastage have been observed in pot and model boiler tests in which sulfur species were present in the form of ion exchange resins. IGSAC of steam generator tubes initiating f rom the primary side has also been observed in several N plants but these have been attributed to pure water cracking ("Coriou cracking") rather than to sulfur-induced attack. The only instances of IGSAC due to sulfur contaminated primary-type water uncovered by the EPRI I review were in sensitized Type 304 stainless steel rather than Inconel 600 -- both TMI-l and ANO-1 have experienced ICSAC in their spent fuel pools and associated borated water piping and these cases of IGSAC have I been attributed to contamination by sodium thiosulfate. 2. Laboratory Studies Laboratory studies have shown that Inconel 600 can be very susceptible to IGSAC at ambient temperatures (25-1000C) in the presence of metastable sulfur oxyanions such as tetrathionate or thiosulphate. The i susceptibility increases with increa ed sensitization (i.e., decreased grair boundary chromium content), increased temperature (at least up to 800C), and decreased solution pH. For highly sensitized material the I rack growth rate can be extremely fast even in dilute solutions a rate of about imm/ day has been reported at 800C in a solution containing 0.7 ppm S as ~ thiosulf ate (Re f erence 5). The mechanism of cracking is not completely understood but susceptibility appears to be restricted to a narrow range of electrochemical potentials that correspond to the region i

I VI - 2 of thermodynamic stability of atomic sulfur (see Section IV). This suggests that the mechanism of cracking may involve interaction between atomic sulfur and the bare metal at the crack tip. Under f reely corroding I conditions near room temperature, aeration of the solution is required to raise the electro chemical potential into the cracking range. However, it is not clear that aeration would be required for cracking at higher I temperatures where it has been observed that thiosulfate is highly electroactive and that elemental sulfur can be produced af a significant rate by disproportionation (Ref erence 6). There is also evidence of a synergistic ef fect on cracking susceptibility for the copre,ence of chloride ions and sulfur oxyanions (Ref erence 7) and very recent result s s (Reference 5) suggest that the presence of lithuim hydroixide may inhibit cracking in borated water environments. 3. Acidic Sulfate IGSAC of Inconel 600 can also occur at high (greater than 3000C) 1 temperatures in acidic sulfate environments. The mechanism in this case seems to be related to that of IGSAC in pure water and caustic environments in that susceptibility to cracking is not asacciated with the degree of sensitization of the Inconel. In slightly acidic or neutral solutions, high potentials are required to induce IGSAC in sulf ate solutions. These potentials are unlikely to be attained in the reducint; i environment present in the primary system of a PWR at high temperatures so it is not probable that acid sulfate cracking would be the cause of tube f ailures initiating f rom the primary side. 4. Key Observations The key observations identified by the EPRI review can be sumarized as follows: a. IGSAC of Alloy 600 in sulf ate containing water has beea observed at operating temperatures (3000C) in laboratory tests bnc is unlikely I to occur under PWR prima ry system conditions. b. IGSAC of Alloy 600 in water containing sulfur oxyanions such as I thiosulfate has been observed at low temperatures (25-1000C) in laboratory tests under conditions that are more likely to occur in the PWR prima ry system. Cracking is very rapid and susceptibility depends on degree of sensitization, pH, temperature, and I electro chemical potential. c. Plant and model boiler experience is entirely related to secondary I side problems. None of the previous plant occurrences of primary side ICSAC of Alloy 600 tubing have been attributed to attack by sulfur species. I I I

~ . [ l \\ VI - 3 I C. Description of Samples Taken i Af ter ideatification of the le.aking OTSG tubes by Nitrogen bubble s testing, it was decided that in ordar to detemine the cause of f ailure, tube samples would need to be removed f rom the steam generators for anilysis. The initial selection of tube samples was made af ter eddy h current testing had been commenced and the choices were made based on N naxinizing the number of defect indications in each tube and providing an a&quate cample of eddy-current signals for eddy current qualification. Four tubes were initially selected from the "B" generator. One (1) tube was a known leaker f rom the bubble test results (Bil-23), the other three tubes contained eddy current indications with greater than 80% I through vall penetratien. Three of these samples were twelve (12) inches in lengtl. with the tcp 5/16". removed during the pulling operation. These samples are de signated as short pulls. In addition to these three short I pulls, a sample approximately sixty (60) inches in length was also . pulled. In this case, the top two (2) inches were removed during the tube pulling operation. This sample and others which were in excess of twelve (12) inches were designated long pulls. Af ter the initial samples had been removed, it was later discovered that eddy. current signal anomalies were showing up at the roll transition I redio n. In order to detemine the disposition of these tubes, additional tube samples were selected for removal which contained these eddy-current signals. This time, fif teen (15) tubes were removed from the "A" g ene ra to r. Three (3) tubes were long pulls and twelve (12) were short pulls. The' designation of those tubes is contained in Table Vl-1, along with the laboratory which perf ormed the failure analysis on each tube. .D. Failure Analysis Program 36W Lynchburg Research Center and Battelle Columbus Laboratories were assigned a multi-task program to provide information related to the steam generator tube damage problem. This program contained the following analyse s /examnina tions : 1. Visual Examination This included an initial on site visual examination of the first four (4) tubes pulled as well as a detailed visual examination at the I ' feedback as to the extent of damage and characterization of the la bo ra to rie s. The on site examination was designed te provide quick f ailure mechanism. k 2. Edd y-Cur rent Examinatio n Eddy current exaninations were perf omed af ter tube removal, I both on site and at the laboratorie s. These examinations were conducted to detemine if any signal changes occurred due to the tube being out of the tubasheet, and also to more accurately locate the defects.

I I VI - 4 3. Radiography Radiography was perf omed on the tube samples to determine if incipient defects could be detected which may have been missed by eddy.cu rrent. 4 Sectioning and Bending In order to locate cracks which could not be seen visually, sections of tubes containing eddy-current si nals were split into 8 longitudinal segments and bent around a mandrel to open up defects. 5. Scanning Electron Microscopy (SEM) and Energy Dispersive X ray Analysis (EDAX) SEM analysis was conducted on the f racture surf aces to provide details on fracture topography at magnification up to 5,000 X. EDAX I analysis, which is conducted in conjunction with the SEM analysis, is used to determine the qualitative elemental composition of deposits observed on the tube surf aces. 6. Auger Electron Spectroscopy ( AES) AES analysis was perf omed on surf ace films to determine the I quantitative elemental composition of these films at various depths through the films. 7. Electron Spectroscopy f or Chemical Analysis (ESCA) This surface film analysis technique was used to determine the I compound form of various contaminants and metallic elements present on the tubing. 8. Electron Diffraction Bulk aurf ace analysis techniques f or compound identification. 9. Sodium Azide Spot Test This spot test provided a qualitative assessment as to whether reduced sulfur species were present at the tube surf aces. I. 10. Metallography Microstructural Analysis I This analysis was used to assess crack morphology as well as define the grain structure and carbide distribution. 11. STEM, EPR and Huey Testing These techniques were utilized to assess the degree of chromium depletion in the grain boundaries and hence, the degree of I-sensitization as it relates to material susceptability to intergranular corrosion. I

I I VI - 5 8 12. Residual Stress and Plastic Strain The X rcy technique was utilized to provide a stress profile in I the roll and roll transition region. 13 Tension Testing 14. Hardness Testing 15. Dimensional Measurements E. Eddy-Current Verification I Nineteen tubes have been removed f rom the two OTSG's to date. These tubes were chosen to contain a complete sampling of the types of defects being found by eddy current and as such are not a representative sample of I the entire tube population. That is, the samples were all located in the outer periphery of the tube bundles and all contained multiple defects when in reality, the majority of defective tubes have cracks only at the roll transition or in the top 1/4" of the tube. The campled tubes contained 97 NDE indications as shown on Table VI-2 (the top and roll transition of one tube was removed prior to pulling). Defects in the 0-1/4" location were detected either visually or by radiography, as the eddy current technique utilized c.t that time was was not capable of detecting defects at the end of the tube in rolled and I welded regions. Subsequent developments in eddy current, howeve r, later permitted examination of that region. I Defects in the other locations were detected by eddy current and are considered verified when a crack is detected by either visual examination or laboratory bending of a strip with the I.D. in tension. All eddy current indications tested in the laboratory were found to be associated with a tube anomaly, with the vast majority being cracks. I In the roll transition region, 8 out of 9 indications tested in the laboratory were found to be cracks; one was located at a pite. Below the roll transition, 26 of 29 indications were found to be cracks; details of the other tube anomalies are noted in Table VI-2. The nineteen samples consist of approximately 38 feet of tubing. Samples removed for metallurgical testing to the present time represent I approximately 14 feet of this tubing. No instances were noted in this tubing in which a defect was identified which was not previously noted during eddy current testing. Since a one-to one correspondence between eddy current indications and tube wall defects has been demonstrated in the laboratory, it may be concluded that eddy current results are a f air representation of the I

I I VI - 6 distribution of tube defects below the top 1/4" of the tube. This conclusion must be qualified in that eddy current signals of less than about 1 volt are only about 75% repeatable. Reference 9 reports the details of eddy current testing. F. Crack Characterization The intergranular stress corrosion cracks have been observed to proceed from the I.D. of the tube, assuming a " thumbnail" shape with an aspect ratio of 2 to 3 (See Figure VI-1). The cracks appear to be I oriented in a circumferential plane. Some of the larger cracks tend to have a slight spiral as they propagate around the tube. In several instances, multiple circumferential cracks have been found at the same I axial location on the tube. The detailed characterization of all the examined cracks is reported in Reference 10. In general, the cracks were extremely tight and not visible either with light microscopy under low power magnification (30X) or high power magnification (500X) in the SEM. Also, in some cases, cracks were associated with visible reddish brown surface dcposits while in other I cases there were cracks present with no visible surf ace deposits other than the overall surfacc oxide film. Of the 30 verified cracks not in the top edge or roll transition I a rea s, 8 were visually detected on 0.D. of the tube, suggesting that stable crack growth can occur past the point of wall penetration. Even beyond wall penetration, the crack tends to maintain the slant sided I configuration; both field experience and experimental work tend to support this view. Table VI-3 compares eddy-current results with actual wall penetra-I tions for the defects in the sampled tubes. The vast majority of the defects have proceeded completely through the wall - this ranges f rom just barely through (not visible on the tube outside surface) to a visible I outside surf ace crack extending 90* or more. The corresponding eddy current depth call is generally less on the average. This suggests that an eddy current indication showing significant depth should be considered a crack that is already through wall. The unverified eddy current indications have similar indicated through wall penetrations to those already verified. These indications I are also likely to be cracks that are or completely or nearly through the tube wall. I Circumferential cracks were f ound within 1/4" of the top of the tube in 8 of the tube s. These cracks were not detectable with the eddy current technique originally used, and were found in the laboratory by bending or radiography. Eddy-cu r re nt techniques have since been developed to inspect I the top 1/4" of the tube and detect these defects. The circumferential cracks ranged f rom 60* to 360* in extent. In two case s, small axial cracks linked the circumferential cracks to the tube ends. I I

3 I I VI - 7 G. Metallography All metallographic examinations performed on tube samples removed I f rom TMI steam generators have shown the microstructure to be consistent along the length of the tube and on different tube samples (see Table VI-4 for locations). The microstructure consists of discrete chromium carbide I in the grain boundaries with a moderate dispersion of carbide within the grains ( Figure VI-2). This microstructure is typical for tubing material used in B&W steam generators and is representative of their mill anreal I plus stress relief heat treatments. The microstructure, however, does suggest that the material could be sensitized. Metallography in regions showing eddy current defects has provided I some interesting data. In all cases, microspecimena prepared in regions where eddy current indications were found, have in fact, uncove red cracks. In most cases cracks were not visible to the naked eye or visible I under scanning elec tron microscopy, although there were cases where cracking was severe enough that it was visibly observable. In these two dif ferent circumstances it appears that the degree of intergranular attack associated with the cracks is different. That is to say, f or major cracks where crack extent was significant and cracks were open, there generally tended to be regions of intergranular attack on either side of the crack 10 to 20 grains wide and through the wall in penetration. In most cases, I however, where cracking was very tight and very limited in extent, the degree of intergranular corrosion was significantly less. In addition, metallographic specimens made in areas away f rom cracks where deposits have been observed in a few cases, showed some very small areas of I localized IGA. In general, these areas did not exceed 5 mils in depth and no cracks were associated with these areas. Regardless of the area examined, no cracks were ever detected where eddy current signals were not p re se nt. Variations in grain size do exist, however the distribution is random throughout the microstructure and there were no changes detected through I the tube wall thickness. There is typically a 0.8 to 1. 2 mil grain site. The only exception to this is in the rolled regions where the grains on the inside surf ace did show deformation f rom the rolling process but only to the limited depth expected. The carbide morphology and grain size suggest that the tube I approached 1900*F during the final mill anneal. At this temperature, significant quantities of carbon are dissolved and available for precipitation at the grain boundaries during the final stress relief. H. Sensitization As previously noted, the carbide morphology in the microstructure I suggests that the tube material could be in the sensitized condition. Several dif ferent techniques were used to check f or sensitization and quantif y its extent. I I

e I t VI - 8 Samples from one tube were subjected to the modified Huey test f requently used on Inconel. Attack at the grain boundaries was so severe that a valid weight loss measurement could not be made. Qualitatively, I however, the material was highly sensitized. Electrokinetic Potentiostatic Reactivation (EPR) testing performed on I samples f rom three of the tubes gave peak potentials of 110 to 125 mV, indicating a severe degree of sensitization. Scanning Transmission Electron Microscopy (STEM) was used to measure I the chromium level at the grain boundaries of samples from actual tubing, archive material in the mill annealed condition, and archive material that had received a simulated stress relief treatment. Both the actual tubing I and the stress relieved archive material exhibited severe sensitization. Chromium levels were 5-8% near the grain boundary. Levels below approximately 10% render the material susceptable to intergranular attack as suggested by G. P. Airey (Ref erence 8). Archive material in the mill I annealed condition, however was not significantly sensitized and exhibited chromium levels above 10%. I. Surf ace Analysis Following bending at the eddy current locations (putting the I.D. in I tension) in order to open up the cracks for observation, energy dispersive X ray analysis (EDAX) and scanning electron microscopy was performed on the f racture surf aces of many samples. In addition to confirming the intergranular nature of the cracks, surf ace chemistry was also analyzed to I determine the presence of any corrodant. To date, EDAX analysis has shown sulfur to be the primary contaminant I present on the f racture surf aces. Chlorine has also been observed in random locations but some of this is believed to be handling contamination. In addition to looking at the f racture surf aces with EDAX, the I.D. I surf aces of the tubes were also examined by this technique. This examination showed sulfur to be one of the contaminants present on the surf ace. The sulfur was not uniformly distributed but in general it was present as a major contaminant. In addition to the EDAX analysis, Auger analysis of surf ace films was also performed. This analysis again found sulfur to be one of the major I contaminants, however, carbon was present in large quantities (50-90%) on the as received surface. This carbon is reported to be in several forms either as a hydrocarbon, a carbonate or graphite carbon. The analyses I indicated, however, that the carbonate was present mostly on the surface and as you sputtered to greater depths through the oxide layer the carbon deeper in the oxide appeared to be present as a hydrocarbon. I I I

I I VI - 9 Sulfur levels measured by Auger analysis range f rom approximately two atomic percent to eight atomic percent. This sulfur was identified in the oxide up to 4,500 A in depth, at which time the percentage had decreased I to around 3.2 atomic percent. The analysis for chloride typically indicated chloride was present as less than 1 atomic percent and generally in the range of.2 to.6 atomic percent. In addition to these contaminants, typical metallic elements expected such as nickel, chromium, iron, were also observed. Nomal trace amounts of fission products such as cesium, be ry llium, boron and zirconium were I also observed (Table VI-5). In order to determine the f om of some of the metallic elements on I the f racture surface, ESCA analysis was utilized to identify some of the compounds pre sent. Typically, iron was present as an iron hydroxide Fe00H, chronium as the oxide Cr2 3, and nickel has been observed as a 0 nickel oxide Nio, nickel hydroxide Ni(OH)2, and a nickel sulfide i N1 S. The latter suggests at least some of the sulfur is combined 23 with nickel, possibly as a complex nickel sulfide. At the surf ace of the film, significant quantities of nickel sulfate have been detected; deeper into the film, the nickel sulfur compound is primarily nickel sulfide. Significant variation exists in the levels of sulfur f rom point to point, even at the same elevation on the same tube. Ne ve rt hele ss, it I appears that the sulfur contamination is present in significant quantities in several possible foms. They a re : I o sulfate as nickel sulfate o sulfide with nickel, possibly as a reduced form of nickel sulfide o in a spinel Fe0 structure with S substituted for 0 and Ni for Fe in significant quantities. In order to detemine the distribution of contaminants along the length of the tube and the upper tubesheet region, discrete analyses were perf omed on the I.D. surf ace. In the top five (5) inches of the tube, samples were taken every half inch and analyzed by both Auger and ESCA. In addition, samples were then analyzed from a location approximately 10 C( inches and 25 inches f rom the top of the tube. This analysis showed that the sulfur concentration along the length of the tube generally remained constant. There may have been a slight decrease with distance, however, from the top of the tube depending on the accuracy of the data (Figure I VI-3). There were dif ferences noted in the level of concentration of sulf ur on I.D. surf aces between dif f erent tubes. In addtion, the Auger analysis showed that there was a general increase in nickel concentration I and chromium concentration with distance from the top of the tube. This increase in nickel and chromium would suggest a decreasing film thickness with depth. All surf ace analysis data to date, therefore, suggests that sulfur is the prima ry cracking agent and that it is necessary and responsible for the intergranular attack. The question as to whether or not the mechanism I

I I VI - 10 can be reactivated still is unanswered. However, cracking testing does suggest that at this tin it is not active. J. Mechanical Testing and Residual Stress Measurements Hardness traverses made in the roll regions indicated a maximum I hardness on the order of 270 DPH (Tables VI-6 and VI-7). It was interesting to note, however, that this maximum hardness value typically occurred 4 mils in f rom the I.D. surf ace and was in f act a peak hardness with values dropping of f rapidly as you go further into the tube wall. I Ha rdness midway through the wall would generally be in the order of 190 to 200 DPH which is typical for mill annealed and stress relieved material. This is also the range of hardness which was measured below the roll .I t ra nsitio n. %r comparison purposes, a fully hard Inconel tube would have a haniness somewhere in the range of 300 to 310 DPH. The data tells us that although the surface may be heavily cold worked, there is metal at I mid-thickness that is at lower hardness values and consequently possesses original ductility. Although cracking has most definitely been observed in the roll I region, it does not appear that the rolling process produces any preferential attack. This f act is somehwat confirmed by the results of residual stress measurements made in the roll regions. There was I generally a relatively unifonn level of plastic strain and residual stress on the inner surface of the tube and there were no stress peaks observed which would correlate directly with cracking locations (Figures VI-4 and VI-5). I Wall thickness measurements were also made and these indicate that wall thinning was on the order of one (1) mil (Table VI-8). Three tube tensile tests were run f rom material f rom two clean areas of tubing and one on a tube containing a defect. The results of these tensile tests indicated that the material f rom the clean area still meets I original meciunical property specifications and that there was no evidence of incipient defects associated with those tested specimens (Table VI-9). The tensile test on the defective tube shows that with a defect through I wall f or one quarter of the circumference, the tube can still withstand design basis loads. K. Material Chemistry Analysis Two pieces f rom sampled tubes were analyzed f or bulk chemical composition. Results of this analysis, shown in Table VI-10, indicate I that the material meets the chemical specifications in ASTM SB-163 for Inconel 600 type material. 'I I I

I I VI - 11 I L. Effects of Heat and Heat Treatment Analysis of computer plots of total and defective tube locations on a I per-heat basis led to the conclusion that the overall def ect level of a heat of OTSC tubing is detemined by the location of the tubes in the gene rat o r. Heats showing good overall performance had a large number of tubes in areas with f ew defects; poor heats were concentrated in bad areas. The effect of heat chemistry was analyzed using a calculated residual I chromium parameter which represented the available chromium in the matrix af ter complete carbide precipitation. A small secondary effect of chemistry on performance was observed; it is, howe ve r, so small that it is of little practical significance. Analysis of the full vessel stress relief records indicate that all the tubing in both OTSGs would be expected to be significantly sensitized. M. Co nclusions 1. The tubing has f ailed due to stress assisted intergranular attack. I This has led to through wall penetratiors and circumferential1y oriented cracks. In all cases, cracks have initiated on the inside surf ace as supported by the following observations. a. Longer crack length on the inside surf ace than the outside surface when cracks are completely through the wall. b. Non-through wall cracks are visible on inside surf aces but not on ouside surfaces. I c. Bend samples with the inside surf ace in tension have opened up cracks while bend samples with the outside surface in tension have not opened cracks. d. Eddy-current indicates the cracks are originating on the inside su rf ace. I e. The general crack shape is thumbnail shaped with the apex at the 0.D., showing definitely that the cracks have originated on the I.D. 2. The intergranular cracking morphology has been confirmed by both Metallography and Elec tron Microscopy. I 3. Transmission Electron Microscopy has also confirmed that no secondary modes of failure is associated with the intergranular corrosion, that is, no evidence of any low or high cycle f atigue was observed on these f racture surf ace s. I I

I I 1 I VI - 12 4 Some general intergranular attacks not associated with I.D. cracking has been observed. These " islands" of IGA are in general associated with I.D. deposits. IGA found at crack locations tend to involve I more penetration in tems of grains (depth) compared to the 6 grains of penetration typical of the IGA " islands." I 5. Analysis of surface films on f racture surfaces and on the I.D. surf ace of the tubing indicste that sulfur is present up to levels of eight atomic percent. The form of sulfur is believed to be either in I the fom of a nickel sulfide, Ni S2 3 or some other reduced form of sulf u r. It is the presence of the reduced sulfur form that is responsible for the cracking mechanism and without such a contami-nant the cracking would have not occurred. 6. Microstructural evaluation of the tubing f rom numerous locations, has indicated that the structure is representative of that normally I expected for steam generator tubing. However, tests have concluded that the material is in a sensitized condition and hence is expected to be susceptible to intergranular attack. 7. To date, all visual, metallographic or bend specimens performed in clean regions oi the tubing, that is regions where no eddy current indications were recorded, all show the tubing to be f ree of defects I in those areas. In addition, all examinations perf ormed at locations where absolute eddy current probing indicated a nearly through will or through wall defect has confimed the presence of cracking in those regions. To date, the results of correlations between I-metallegraphic and eddy current results have shown a one to one correlation. I 8. Although there is excellent correlation between eddy current indications and metallography, it has also been learned that eddy current signals of 80 to 9% through wall, in most instances, actually represent 100% through wall defects. 9. In all cases to date, cracks which have been examined either by metallography or by bend testing, have shown the defects to be at I least 9% and generally 100% through wall in penetration. This would suggest that crack growth rates are rapid and that defects, once initiated, propagate to through wall. 10. Auger and ESCA analysis have shown the presence of carbon, sulfur, nicxel, chromium, oxygen and sulfur on the fracture surf aces. In addition, normal trace amounts of fission products have been I observed. The important results from this analysis have shown that sulfur concen-trations along the I.D. surf ace of the tubing down to the 27 inch point, are generally uniform with a slightly decreasing I level as you go down the tube sample. The significant presence of oxygen also confims that oxygen was definitely available for the corrosion reactions and that very possibly played a major role f or the f ailures observed. I I

I I VI - 13 11. The consistent circumferential orientation of the cracks indicates that a longitudinal stress is part of the cracking mechanism. I Residual stresses in the roll alone were not responsible for the cracking. The ref o re, the fact that the cracks occurred when the tube was under applied axial tension rather than a hoop stress, suggest the cracks formed during cooldown or in the cold shutdown condition. I 12. The observed patterns of f ailure within each OTSG cannot be explained on the basis of tube heat chemistry or differences in sensitization I due to full vessel stress relief; all tube heats are susceptible in areas of significant failure, and all saw essentially the same sensitization conditions during stress relief. I I I lI 'I ,!I I I I I I I I

I I TABLE VI - 1 TUBE SAMPLES EXAMINED "A" OTSG INVEF1IGATING TUBE NUMBER HEAT NO. SAMPLE LENGTH y JRATORY W 13-63 M2408 Long B&W 71-126 M1671 Long Battelle 133-74 M2408 Long B&W 112-7 M2560 Short B&W 88-11 M2626 Short Battelle 112-9 M2626 Short B&W 146-8 M2404 Short Battelle I 112-5 M2560 Short Battelle 16-69 M2345 Short B&W 12-62 M2345 Short B&W 4 62-8 M2560 Short BGW l 88-7 M2560 Short B&W 11-66 C2690 Short B&W 146-6 M2560 Short Battelle 23-93 M2409 Short Battelle I "B" OTSG 33-30 M2867 Long B&W 8-25 M2709 Short Battelle 11-23 M2320 Short Battelle I I l I l

'I I TABLE VI - 2 DEFECT

SUMMARY

MAY 11, 1982 ,I LOCATION OF TUBES NDE INDICATIONS CRACKS NO CRACKS DEFECT EXAMINED INDICATIONS EVALUATED VERIFIED FOUND 0-1/4" LOCATION 18 13 8 8 0 I ROLL TRANS. CRACK 18 10 9 8 1 (NOTE 1) CIRCUMFERENTIAL .I CRACK NOT IN TOP EDGE, ROLL TRANS. WITHIN UTS 19 68 29 26 3 (NOTE 2) CIRCUMFERENTIAL CRACKS BELOW 4 6 4 4 0 NOTE 1 - 1 AREA - ID PITS NOTE 2 - 1 AREA 0F IGA W/0 CRACK 1 10-30% E.C. INDICATION - NO VISUAL DEFECT 1 MECHANICAL DAMAGE I lI lI 'I !I I I

~ i I TABLE VI - 3 DEFECT DEPTH Visual / Bend Corresponding Untested EC I Verified EC Data Indications Location No. Avg. Pe ne t. Avg. Penet. Ra ng e o f Pe ne t. I;a. Range of Penet. 0-1/4" 9 99% 1 95% Roll Region 2 100% 95% 95% 10 60-95% Roll Transition 7 100% 90% 65-95% 2 95% Roll Transition - 12 in. 10 100% 93% 50*-100% 29 59-95% 12-16 in. 3 80-100% 16-24 in. 4 90-95% Below UTS lower surface 3 100% 92% 90 iiS% 1 90% I I

- Standard differential probe data Other EC data 4 x 1 absolute I

9 I I I I I I I

a I .I TABLE VI - 4 METALLOGRAPHY X - EXAMINED 0 - TO BE EXAMINED l AXIAL LOCATION FROM TOP OF TUBESIIEET A146-8 A133-74 A62-8 A71-126 A13-63 A146-6 B8-25 B33-30 B11-23 0-2 X X X X X 2-4 X X X 4!I 4-8 X i 8-24 X X X X X i 24-50 X X X j 50-70 X X I

I

!I

.I d

i t 1 i,l ! I !I I

I I 1 TABLE VI - 5 RESULTS OF AUGER /ESCA ANALYSIS SPECIMEN 8-25-3 CRACK AND FRACTL11E I ESCA Analysis, % l 600 A 1500 A Element As Is Sputtered Sputtered l Cu(a) 4.6 6.6 Ni 3.0 3.4 1 Fe 1.7 2.3 g 4 i W Cr 1.1 1.5 g 0 -10 11.1 11.8 g 4 I Ru 0.9 I Ag 2.7 0.3 C >90 75.8 64.8 3 0.4 Cs 0.4 P 0.5 i Be 7.2 i

I (a) Cu mainly from copper holder i

i I I

I

,; I

I TABLE VI - 6 I HARDNESS MEASUREMENTS NEAR TOP OF B-GENERATOR SAMPLE 10-29 DISTANCE l-1/2 MILS 1-1/2 MILS I FROM TOP. FRCM INSIDE MID-FROM OUTSIDE INCHES SURFACE WALL SURFACE I -0.02 275 -0.01 224 0 211 202 I 0.01 279 214 0.05 237 I 0.09 267 188 0.13 283 I 1 0.17 292 201 0.21 284 0.25 286 208 0.29 290 0.33 277 207 0.37 290 O.41 273 198 0.45 267 0.49 239 207 0.53 252 4 I 0.57 247 205 0.61 226 0.65 252 192 221 l i 0.67 302 193 0.75 285 183 0.83 250 210 0.91 295 242 0.99 263 212 I 1.07 176 185 1.15 201 202 1.23 187 183 203 I 1.24 204 1.39 189 166 1.59 178 185 j 1.83 196 181 2.07 188 I I I I

.u I TABLE VI HARDNESS MEASUREMENTS - BATTELLi. 280 1 260 Location 1 Location 5 240 220 200 I 180 l I I I I I 1 l 1 I 160 280 260 Location 2 Location 6 240 220 f ~ M I i : ?M ,0 I } 160 l I I I f I f f 3 5 200 I E j 260 Location 3 Location 7 E 240 220 200 180 160 I f f f I f I I I I I I I I 280 260 Location 4 Location D 240 220 200 m 1 60 I I I 1GO 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 m- --n -y 00 10 00 ID POSITION ACROSS T}(E TUCS ?!ALL TUDE A 14S 8

TABLE VI - 7 (CONTINUED) = 220 I 260 Location 1 Location 5 240 220 200 180 100 I I f f f I f f f f f i I f 280 260 Location 2 Location G 240 I c 220 t I 200 180 I 3 160 I I I f I f f I f I 1 e E 5 280 I U j 260 Location 3 Location 7 E$, 240 220 I 200 180 / I 100 f f I f f I I i 1 I 1 1 I t 280 260 Location 4 Location 8 240 220 I 20n w 100 100 I f f I f I f f f I f f 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 I OD ID OD ID POSITION ACROS5 THE TUrE WALL 2 I TUBE A 14GG

I I 4 j TABLE VI - 8 i 1 WALL THICKNESS MEASUREMENTS i 4 l TUBE A-11-66 DISTANCE FROM TOP OF TUBE WALL THICKNESS - INCHES l 3/32" .0367 j 7/32" .0355 i 3/8" .0355 l l 1/2" .0352 9/16" .0350 i j 5/8" .0353 I 7/8" .0366 i 1" .0363 l l 1-1/8" .0363 l 1-1/4" .0360 1-3/8" .0363 lI 1-7/8" .0363 i 'l I I I I I I

I I TABLE VI - 9 FAILURE ANALYSIS RESULTS MECHANICAL TESTING l TUBE NO. A-71-126 B-33-30 A-13-63(1) ASTM i AXIAL LOCATION 60 7/16-68 11/16 33 1/4-41 3/4 2 5/16-10 7/16 B-163 HEAT No. M 1671 M 2867 M 2408 T.S., PSI 101,000 101,000 44,500(2) 59,200(3) 80,000 Y.S., PSI 53,000 PSI 53,000 PSI 38,000(2) 50,000(3) 35,000 d ELONG. 33% 30% i (1) Eddy Current Defect 90-100% 'IV at 7 1/8" 1 (2) Based on nominal cross section (3) Based on actual cross section - 75% of nominal I i I 1 ) lI I I I I

II TABLE VI - 10 1 CHEMISTRY DATA l Tube No. A-71-126 B-33-30 Heat No. MI671 M2867 Analysis Heat Check Analysis Heat Check C .034 .027 0.30 .041 .031 .033 Cr 15.3 14.94 15.53 15.66 15.15 15.22 Ni 74.38 74.66 73.47 74.90 76.35 76.22 Fe 9.6 9.30 9.80 7.76 7.37 7.40 Mn .36 .30 .32 .40 .34 .34 I S <.01 .002 .0003 <.001 .0018 .0030 s 1 l P .01 .011 Si .33 .35 .35 .28 .29 I Ti .20 .19 .24 .28 .20 .21 l N .013 .013 B .0027 .003 .008 .0030 .0028 cc .1 .01 .01 .075 .030 .030 g Al <.1 .13 .16 .17 .16 .; g Mg .019 .021 Cu <.01 .01 .015 .03 .03 Mo <.1 Cb <.1 !ll i 4 II I I i ) ll

I FIGURE VI - 1 I SKETCH OF TYPICAL CRACK SHAPES ~ t i p,<. # + M_w? 'N,, jc.s_,w,p(4 %- / D i 'N N 'N a) Advancing crack not yet through wall. l yr,y -,Qin$"5~'% ~...x' n h 5 [( %%h, ~n j s l l .? I b) 100% through-wall crack continuing to grow. I I l l l l l I

I FIGURE VI - 2 i y'd. r-( <~; e .r\\ '< ,,\\,. t

4..'-

/ /, wk ..N,.".'."..\\ .\\. /. .. Q '. &...

p.

..V. *. j ~.s,., 's ...u w

n 1

4... s,.. x;t. f. ! s i 9 o ' J. ' -~..C,,- t .t N .\\. I ./ \\ L j / .f.. ,\\. y yl -L..irN l .,.. f. / s.- ty. s I' **, I i M* /;- s ,g i .,-f, /- *. d.- ik' I. /.4, h.. ( -}s<.., '. :~.. h.,/ :... Y. g . l. u/ ( s .t .~ s .s . s,. >,,, p. .. x,se.... .s /.n. ',.. X.' '

i i ~ %;

-y_._.): i s...,, p; %..,..,/.: 1 ./. v7 f. ... n.7 y.. /.. '...< /,, l i ~ - ~ -:s:.s / w. '. 7 i, s s. ..s \\.. /. ,..., ~ 3 t ). ' t 1/,,Y. , /,, ) .., s.r A, A, ( ,m .L.;..T I r (, A i i en 1 ) 1 I i f i 1 1 I I, m.-y,,-y--._.- p ,--y.

m. =5;'w S a M 7 l 2 b3 M fo fc ~ / 3 / / 6 M ) )(* (D 2 / e co S l h ,C l m l r / M 0 0 j1 m -o l [ -i n n o I i t m i e s s l. o 5 y P na sd S l ue m sv C r f ri o ee Vc p e o tR T a nes a cA f-n. r r e F {, P e e c m l ^ c i na o t t e 3 s A i i3 D m 1 t \\ 1 2 I m g 1 f I g I j s g 1 f 4 5 3 2 1 g . E s ' ; E 3.'

ea man me uma e sus e uma uma aus a sus aus e sus eem PLASTIC STRAIN VS POSITION TMI-1 TUDE SAMPLE Ali-66 i..D. .:1..:l:.i : :* E 25

=a INSIDE SURFACE C4) a:

OUTSIDE SURFACE Cx) c'5 3 r 20 x 2 / q o o x 15 o a: t-a o e' o o m o m 10 t-m 4 _J Q, 5 o o o o o o o o o o o o o oo o o o o o oo o o 0 ^ ^ O 1 2 3 4 5 DISTANCE FROM TOP GF SAMPLE, Inches

~ em m h m e m' W ma mmW Eus e em m m,m RESIDUAL STRESS VS POSITION TMI-t TUBE SAMPLE A11-66

l. : ' ': ' '. 3 :

1 100 INSIDE SURFACE C O s LO OUTSIDE SURFACE Cx) W 80 ItJ Cf. s 1-e O o 1 60 5 p. ? >~ O 5 = H o F1 mM n tj 40 ge X o 2 Lt j>j o H o W 20 m o ! L:J o o o o o n: o O o o oo o 4n. o o o o oo o Yi o o o O .o 0 1 2 3 4 5 DIS f ANCE rTROM TOP Or' SAMPLE, Inches ^

I VII - 1 VII. TMI-l OTSG F ABRICATION HIS" DRY A chronological listing of the f ollowing key events in the tube fabri-cation history is provided in Figure VII - 1.

Tube Hollow Round Manuf acturing
Tube Making Process
Tube Installation
OTSG Stress Relief (Tubes Installed in Vessel)
OTSG Site Storage The detailed f abrication history is presented in reference 2.

I: A. Tube Raw Material Melting e The raw material f or OTSG inconel tubing was melted by B&W - Tubular Products Division (TPD).* Individual elements were first I purchased f rom material suppliers (i.e., nickel. cnrome, iron, etc.). Ordering specifications for these elements placed require-ments f or purity and limitations on undesirable trace elements. I Proportional amounts of each element were electric are melted. Spot checks of the molten metal were made and additional elements added as required to achieve the SB-163 Ni-Cr-Fe chemistry. The official ladle analysis was made just prior to pouring into ingot s. A final check analysis was made of the ingot af ter cooling. B. Hollow Rounds Fabrication Hollow rounds f abrication was perf ormed by B&W-TPD. Inconel ingots were first hot formed into blooms. The blooms were hot rolled into I 8" diameter solid bars. A hot extending operation next f ormed 4" hollows from the bars. The 4" hollows were then cold worked to 2" O.D.,. 3 8 8" wall. The cold working operation was perf ormed in two I passe s. An open fired annealing was performed af ter each cold working operation. C. Tube Fabrication All tubes used in the TMI-l OTSG's were manufactured by Pacific Tube Company, Los Angele s, CA. Although PATCO has no records of I the tubing supplied for the TMI-l generators, interviews of PATCO personnel involved in the TMI-l tube manufacturing process and review of B&W correspondence during this time period provided a relatively complete description. The tube manufacturing f acility at PATCO was designed around the order requirements for nuclear grade tubing. 1

0rganization name has subsequently been changed to Tubular I

Products Group (TPG).

I VII - 2 Care was taken in the manuf acturing of the tubes to prevent contam-ination such as sulfur f rom coming in contact with tubes. All = handling of ttms was closely controlled (gloves were worn by I personnel and tube holding devices were made of Inconel). It was noted that if any contaminants were on tubes during the drawing p roces s, the tubes would have a discoloration af ter final anneal-ing. Tube. discoloration was cause for rejection of a tube. I A brieE description of the tube making process is provided below: ~ tube hollows received by PATCO - 2" 0.D.,.188" wall - tube hollow reduced to 1 1/4" 0.D., 080" wall (rocker die - 1 draw) d ~[- - cold drawing to.6 25" 0.D.,.034" wass (floating mandral - 4 draws) F - bright annealed - straighten (rotary-six roll straightener) - centerless ground An intermediate anneal was perf ormed af ter each drawing operation and tubes were cleaned prior to each annealing. Inte rmediate anneals were perf ormed in a bright annealing f urnace at 1750*F. The final tube annealing was performed in a two zone continuous feed hydrogen furnace. The environment temperatures were 1800*F h for the first zone and 2000*F for the second zone. The tubes were g processed through the f urnace at a rate of 50 inches / minute. Based on the temperature of the furnace and the feed rate, actual tube metal temperature was between 19 50-2 000 *F. i 1 Prior to shipment to B&W, the tubes received the following NED tests: Ultrasonic Test, Liquid Penetrant Test, Eddy Current, Hydrostatic Test and Metal Comparator Check. D. Tube Installation The OTSG Inconel tubes were received by B&W-NED (Barberton, Ohio) f rom the PATCO individually wrapped in paper and packaged several hundred to a lot in long wooden boxes. The boxes were loaded onto tubing racks and elevated to the required height of area of tube sheet being tubed. 1 All tube installation steps were performed with the generator in the horizontal position with "Z" axis up. ( See Figure VII-2. ) Normal tubing procedure was to begin at the top of the tubesheet and thread tubes into the vessel in horizontal rows until tubing 3 was completed at bottom of the tubesheet. The tubing operation was performed one tube at a time, except for the manway area, which was the final area to be tubed utilizing a cluster of some 750 tubes.

I I VII - 3 During the tubing operation, the tubes were removed f rom a box one at a time, paper wrapping removed, handled with clean white cloth gloves, and immediately hand inserted into the vessel. During the I actual insertion of each tube, each was wiped with an acetone wetted clean cloth and visually inspected. For the manway area, the tubes were also individually unwrapped, cleaned, inspected and I placed into a tubing fixture, protected f rom damage and cleanliness maintained, and tubed as a group. Following the completion of vessel tubing, all tubes at the lower I tubesheet end were roller expanded iato their respective tubesheet holes. The expanding operation itself was perf ormed by shop personnel under direction of shcp supervision. A written procedure I was issued by Tool Design which described expanding operation to be followed and dictated periodic cleaning, roll replacement and inspection. Once the tube ends were 1007. expanded on the lower end of vessel, the tubes were pretensioned. This was accomplished by premeasuring tube ends on the upper end and air heating tubes to obtain required growth, thus obtaining pretension on cooling af ter rolling. Using the same procedure as used in the lower end, the tubes were roller expanded in their respective upper tubesheet holes. The tube to tubesheet welding was perf ormed following roll expan-sion of the tubes by an automatic procedure. The welds were sub-I sequently visually and dye penetrant inspected followed by helium and hydrostatic leak tests (hydrotest performed af ter post weld OTSG stress relief). Leaks at the tube to tubesheet weld dis-covered during the hydrotest were repaired by the following method:

Remove portion of weld in area of leak
Expand tube to 2 5/16" (maximum) depth overlapping original expanded area by 1/8"
Dry out tubesheet crevice area
Re weld area removed around tube leak E. OTSG S tress Relief The thermal cycle applied to the completed vessels was to perform I

the ASME Code post-weld heat treatment to all welds which had not previously received a stress relief. Welds either side of both tubesheets were in this category. B&W Engineering recognized the i need for supplemental Code PWHT limits to protect the integrity of the vessels, therefore, a set of structural limitations were also imposed. The required Code cycle was to heat to 1100-ll50*F, hold I the welds requiring PWHT for a minimum of one hour per inch of thickness in the 1100-ll50*F range, furnace cool to below 600*F. The maximum permitted ASME Code heating and cooling rates were I 100 *F/ hou r.

E I VII - 4 i Due to the B&W imposed structural limitations, and the thermal attributes of the installed furnace power, furnace insulation, vessel mass, artificial cooling devices, etc., heating rates in I excess of 20*F/ hour were not achieved above 600*F, and cooling rates above 600*F were probably never greater than 15'F/ hour. Also, because of structural limitations, controlled heating and cooling above 200*F was utilized. I Because of this slow process, Unit-1 (A-0TSG) took about 113 hours from start to beginning of cool, plus 134 hours of cooling to below l 200*F. Uni t-2 (B-0TSG) took about 84 hours f rom sta rt to beginning 5 of cool, plus 130 hours of cooling. Tubes on Unit-1 were above 1100*F (average of all tube temperature s) for 18 hours while Unit-2 tube average temperature was above 1100*F for 13 hours. The following is a brief description of the stress relief operation. c The f urnace consisted of six lengthwise electric zones of approxi-I mately 14'2" each of 85' ental, by 18' wide x 18" high. Heating elements were installed in the support car floor, the furnace roof and each end wall. One end wall was the furnace door. The vessel I was set on the car to a prescribed location and oriented with "Z" axis down rotated 12' clockwise of f of the major axis (See Figure VII-3). Stainless steel duct piping was welded to the top vessel nozzle, one bottom nozzle and to two secondary side openings. These connections were flow loops which were installed outside the I furnace proper, in the adjacent equipment room. Each flow loop consisted of one or two axial flow f ans, an electric heater and a cooler. Each loop was built to be evacuated af ter welding the vessel into them, and backfilled with argon gas. Prior to the above steps of vessel loading and connection to the ductc, numerous sheath thermocouples were placed in the tubes at I prescribed locations and on the I.D. surface of the primary welds (heads to tubesheets). Thermocouples were installed at prescribed locations on the vessel 0.D. also. Extensive care was taken to assure that all themocouples were connected to a prescribed, pre-assigned recorder or controller I channel. Data sheets were designed to be filled out hourly. The data was reviewed and reduced to pertinent summary conditions to compare with the permitted structural limitations. The entire operation was la rgely manual in many respects, regard-less of the use of the automatic program contollers, as many manual operations were required due to the hourly reviews of the process I status. Heating rates on the order of 20*F/ hour at the beginning but as low as 5'F/ hour near hold range were common. I I

I I VII - 5 .I In general cooling rates of no more than 15*F/ hour were used. Actual metal temperature changes probably did not exceed 20*F/ hour over extended lengths of time. Overall cooling rates (1100*F to 200*F) were on the order of 7'F/ hour f or the TMI-l units. Plots of Unit-1 and 2 post-weld heat treatment (average) shell/ tube themocouple readings are provided in Figure VII-4. These plots I depict the gradual heatup and cooldown ramps above 200*F. The close tie between tube and shell temperature over the entire range is also identified. A detailed review of individual hourly thermocouple readings indi-cates tube temperatures varied less than 40*F radially across the I upper tubesheets. This data also indicates the highest themocouple readings were located at the center of the tube bundle during heatup and at the outer tubes near the "X" axis during cooldown. The lowest individual thermocouple reading varied at different I locations on the outmost tubes during the entire cycle. It is common B&W practice to obtain surf ace samples of the OTSG 1 secondary side tubes af ter all manufacturing operations are com-pleted. One of the elements evaluated in these samples is sulfur. The TMI-l units were f ound to have an average sulf ur level of 162 2 I and 216 ugms/f t on the A and B tube surf aces, respectively. This compares to an observed range for other operating plants of 42-557 ugms/f t2, F. OTSG Shipping / Site S torage The steam generators were transported to the site with shipping l covers over all openings and desiccant for humidity control. At a the site the vessels were stored outside with a nitrogen purge on the primary and secondary side. The above outside storage lasted one year to 18 months. At the end of this time period the vessels I were moved into the containment building and upended. The exact method of storage f or the f ollowing 21/2 years before commercial operation is not readily available and not a subject of this report. G. Conclusions I The review of the TMI-l OTSG tubing records indicate that accept-able engineering practices and manuf acturing techniques were used throughout the manuf acturing process. Furthermore, no f abrication information has been found which can be related to the tube eddy I current defect patterns as noted in the A and B-0TSG. I I I

l E I = M bE a= = o I o U a ".hd ene N N E Z 5"S E S F. I O(> 5"r a2E a d i a O O app M N g "=- =x y.

::: E a

M=a i = o o I O

D.:.::. E_ -

s f'1 E -d _ MM E 5

== 2 a W O ~N .o._- _e Nv g o

== VI se E I bo m 2* . U. N M I o -D M. 9'.'$ M I i r.EE

M O

O I m. [ o e N M Qu L-s- [ Och W-c- z ~- M D u m s tas Q O EQ O 883 w EMMm [4 I ~ dilE% M [ $Z W no _o I ' E o-g,, O-O- g e.- E" "M o-MW zm_ tm8 a D En==

- M X

w OOw = O* Z m e.m. 4 r

E I FIGURE VII-2 I I l 1 Z OPEN LANE l 8 g 3 I I y I I I I X X ORIENTATION OF GENERATORS DURING TUSING OPERATION I g i i

I FIGURE VII-3 I x i t i I i I I \\ t \\ I i g N +I x. 3 1 i l \\ l i 1 i i i \\ t 1 mM uNE i QRIENTATION OF GENERATOR DURING POST WELD STRESS 8

l l l l I // i l 'd' 1 8 l l i vrr T- - 2 gg l l ,M#, l gd* ,'V- - '. I l 1. -.. I -.. i L l l I i l I /[I I I i j 2 I i l I // I i i I i I i i i i I // l I I y, a 2 l i l !// I I I I I N ej i i // I I I I I 3 2 i i // I I i ae '- ~ E 3 E I //l i I n 4 e l l // l l U l l // l l h l I }}' I I 2 8 "5 i i/ i 5 l 2 I / I I c p, I /i l I I "d " d If I I l l l l I 'd' / I I i l I I 5 ,f l l I I I I I n,. = a / I I I .l l l 1 m 7 /l l l l l l l E ~ E$ / l l l l l l 52 l l l l l I ~ l KYL 6 _J w= ( d t h l I ( l l l l l l 5 \\ I I I l l l l l l i \\l l I I I I i I l 2 b \\\\ l l l l l l l I \\\\ l i I I i na \\\\ l l I I I I I I I \\\\ l I I I I I ns ; l \\\\ l I I I I l I W I I I I I 'd' I l lt-J l l l l l 7 I l hl I I I I . 3 'E l l l kg l l l l I I l l l \\\\ l I I Nd!! I I l 1 l %' l l l l i l \\\\ l I I ,g, l l l l I ik\\ I l = l I I I I I Si I I I I l l l k._ l I I l l l l l I l \\ l hd5 l l l l l l I I { l A g o, o o o o o I o o o J. '32317d3dh31

I I i, ., m, I .1 ,Y uc _ l m

n/.

a I o l I l dV L - i i i i I ?! l g f 3 l I // l dd G 9 l l l fi 5 l 1 l ,/l l l l I 5 I i l l if I l l ^7 ll e 2 j l l ., e E h[5 ff h nI g ~ 3 b' E9E I // 5 k;; I \\// dd C 1 lm 5 I // 1 2 -~~ ~ g [ l I // l "S 2 // l g l 3 ut 5 d .J B i / aS g g 7 5

e.

l/ {g f na ti z e se e C dd I g -I I I l l k a< : = r i i i W i l ~-~ aa 4 g g( i l m i c -\\- nL m ,a k i l l l I l l l i H 11 i s i i e I ! \\\\ l l l l l "I m I I l %sl l l l l "d c t i I I i i i I I I INXi I i 2 I I I I 'Al j "S I I 1 Twi i dd I I l l l l Kl I I I I I I i l l

l t

ne : I I I I I I l l l l c l l 1 I I I I I l l dd I! S I I i il i I i a

== 5 ~~ = = = = = = ~ ~ 3.* 3831783dd 31

I VIII - 1 I VIII. OTSG TUBE STRESS ANALYSIS A. Introduction Intergranular stress cracking defects originating at the tube inside diameter and oriented circumf erentially have been found on a substantial number of the TMI-l steam generator tubes. The cracking distribution I has been divided for analysis into four locations: the heat affected zone of the seal weld, the roll, the roll transition, and the free standing tube. Accordingly, stresses (residual and applied), in I particular, axial tensile stresses have been estimated and computed at these four locations to determine how well these stresses can be correlated with the actual defects f ound. B. Stress Calculation Results (See Reference 3 for more details) c 1. Residual Stress Table VIII-l shows best estimate residual tensile stresses (in-cluding those due to tube preload) at the inside diameter of the tube f or the steam generators as now installed at TMI-1. Figure VIII-l shows the tube /tubesheet joint configuration at TMI-l and our estimate of the residual stress distribution in this I joint. Figure VIII-2 shows a plot of the tube wall thicknesses as measured f rom an actual tube sample removed f rom one of the TMI-l steam generators. This figure indicates that there was a roll I transition (between rolled and unrolled portions of the tube) abeve as well as below the rolled joint. Accordingly, substantial axial tensile stresses are expected (due to rolling) above as well as below the roll area of the tube. Figures VIII-3 and VIII-4, which are from a Huntington Alloys Tech-nical Bulletin for Inconel 600 show that residual stresses re-I maining after ll500F heat treatment reduce to above 22,000 psi. These figures, which are for cold-drawn 20% and cold-drawn annealed rod, are considered to be reasonably representative of the Inconel I 600 tubes at TMI-l since the tubes also had some cold work in the areas of interest. The maximum stresses in Table VIII-l exceed the 22,000 psi value by I 4,000 psi which is our best estimate value for tube preload af ter the ll500F heat treatment of the steam generators. The 4,000 psi preload stress value was calculated based on a 1981 fiber optics I inspection, which showed tube B-22-30 at TMI-l to be parted and separated by about 0.09 inches. Because of differential thermal expansion (tubes expand slightly greater than shell) between the tubes and the shell of steam generator when the steam generator was I at 11500F during heat treatment, this 4,000 psi stress was not subject to s tress relief. Therefore, the 4,000 psi value was added in Table VIII-l to the pertinent 22,000 psi residual stresses men-tioned above. 5

I VIII - 2 I It should be noted that the stresses in Table VIII-1 are sub-stantially in excess of the stresses reported by EPRI by their February 24, 1982 letter to GPUN. EPRI reported stresses between 20,000 and 5,000 psi for an 11500F for 10 hours heat t re atme nt. 2. Applied Load Stresses As indicated in other sections of this report, the intergranular stress cracking found at TMI-l probably occurred during the time period following the cooldown transient of September 6,1981. I Accordingly, the applied load stresses for the cooldown transient discussed below may not have existed while the intergranular stress cracking was occurring. Howeve r, there is a possibility that these I or other stresses from loads applied in the past still existed in c some parts of the tube /tubesheet joint s. Specificially, we consider it reasonable that applied loads may have resulted in locked up axial stresses in both the rolled portion of the tube and in the weld HAZ of the tube. For this condition to have existed, it is only necessary that some very I small amount of sliding occur between the tube and the tubesheet in the rolled joint. If the load capability of the rolled joint were about equal to an applied load, then some sliding could occur. If the applied load were then removed and no sliding to the original I position occurred, then axial tensile stresses would be locked into the tube within the roll and weld HAZ regions. Accordingly, we conclude that applied load stresses may have been a contributor to the intergranular stress cracking at TMI-1 and our best estimate values for the September 6, 1981 cooldown transient I are presented in Table VIII-2. See Figure VIII-5 which pertains to this cooldown transient. The axial stress values in Table VIII-2 are about half those for a I design basis cooldown. As can be seen from Figure VIII-5, the September 6, 1981 cooldown was relatively gentle regarding tube loads even though some short step change s in temperature did I occur. Tube loading due to this transient is mainly proportional 0 to the maximum mismatch in temperature (90 F) between the steam generator shell and the tubes; also, pressures in the steam gener-ator have some effect. The applied load stresses agree reasonably well with those devel-oped by EPRI in their Report NP-2146, which was based on TMI-2 tube I loads measured by strain gages. Specifically, a maximum axial Icid stress of 10,185 psi was calculated for a tube near the periphery during a cooldown transient which was more severe than I 5

1 I VIII - 3 the September 6, 1981 transient at TMI-1. This 10,185 psi EPRI I value compares favorably with the 11,000 psi value in Table VIII-2. I The axial stresses in Table VIII-2 are noted to be higher in the peripheral tubes than in the center tubes of the tube bundle. This is calculated to occur primarily as a result of elastic defomation of the tubesheets. 3. Conclu si.e s I Table VIII-3 summarizes the combined maximum residual, preload and applied stresses which are estimated to have occurred in the tubes at both top and bottom tubesheets at TMI-l on September 6, 1981. Based on these stresses, we would conclude the following: c Stresses in the outer tubes are slightly higher than in the a. center bundle tubes. This is generally consistent with the I actual defect radial distribution f ound' at TMI-1. b. Substantial tensile stresses are considered to have probably I been present in the four locations where defects have been found axially within the tubes as indicated in Table VIII-3. I The stresses discussed herein are applicable to both the upper c. and lower tubesheets because the weight of a tube results in a negligible diffe'rence in stress for the bottom versus the top tubesheet. This and the fact that no substantial defects have I been found in the lower tubesheet area indicate that some factor other than stresses has played an imponant role in the cracking of the tubes at the upper tubesheet. I I I I I I I 5

I I TABLE VIII-1 TUBE RESIDUAL TENSILE STRESSES PSI AT FOUR LOCATIONS OF INTEREST Location Axial Circumf e rential Weld HAZ 22,000 -22,000 Roll -10,000 -22,000 Roll Transition 26,000(2) 22,000 i Tube (between tubesheets) 0 to 26,000( 2),(3) 0 to 22,000(3) i NOTES (1) The stresses in this table are for the case where the rolled joinc. is B tightly cons trained by the tubesheet. If the rolled tube were loose in the tubesheet then circumferential stress in the loose roll area would approach zero and axial s tress in the loose roll area would aproach about 4,000 psi. Axial stress in the weld HAZ would approach I 26,000 psi. ( 2) This stress includes 4,000 psi due to tube preload af ter heat t re a t-I ment. Such preload stresses are expected to be higher for the periphery of the tube bundle than for tubes located in the center of the bundle. I (3) Tube stresses between tubesheets are primarily due to the tube straight-ening manuf acturing process and may vary considerably depending on the location within the tube. I I I I

I I TABLE VIII-2 MAXIMUM APPL'ED LOAD TUBE STRESSES (ps i) FOR THE SEPTEMBER 6, 1981 COOLDOWN TRANSIENT I Axial ( 2) Location Center Tube Outer Tube Circumferential(3) Weld HAZ 0 to 6,000 0 to 11,000 10,000 Roll (l) 0 to 6,000 0 to 11,000 0 Roll Transition 6,000 11,000 10,000 Tube (between tube-sheets) 6,000 11,000 10,000 NOTES (1) The axial s tresses in these locations can range from zero to the maximum values indicated depending on the actual tightness of the rolled joint. (2) Axial stres ses for outer bundle tubes are hig! er than for center bundle tubes because of the elastic deflec tion of the h5esheets. Tube preload I stresses are not included in this t able. (3) These value s are estimated stresses due to pressure loads, accurate I values can be calculated, if the de tailed configuration of the jcint can be determined. 15 I I I I I I 5

TABLE VIII-3 SLHMARY OF MAXIMUM ESTIMATED STRESSES CONSIDERING MAXIMUM APPLIED DURING SEPTEMBER 6, 1981 COOLDOWN, I PRELOAD AND RESIDUAL STRESSES (psi) Axial (1) Location Center Tube Outer Tube Circumferential Weld HAZ 31,000 39,000 -12,000 Ro11(2) 9,000 17,000 -22,000 Roll Transition 31,000 39,000 32,000 Tube (between tube-sheets) (3) 5,000 to 31,000 13,000 to 39,000 10,000 to 32,000 I NOTES (1) Axial stresses include estimated preload stresses of 3,000 psi for a center tube and 6,000 psi for an outer tube because of elasticity of the tubesheets. (2) Axial stresses in the roll area are based on the assumption that the initial residual compressive stresses have been relieved and changed to tensile stresses due to preload and applied loads as a result of some I sliding of the rolled joint. (3) Tube stresses between tubesheets are primarily due to the tube I straightening manufacturing process and may vary considerably depending on the location within the tube. I I I I I I I I

I I CLADDING B WELD ROLLED AREA (WITH WALL THINNING) TUBESHEET 4 \\ g TUBE UBE HE TS) TUBE ROLL TRANSACTIONS WELD HAZ } g AXIAL \\ l b \\ 1 w 1 STRESS $ WITHIN o O TUBE ] l CIRCUMFERENTIAL l E I u J .s I POSITION (SEE ABOVE SKETCH) I TUBE /TUBESHEET ROLLED / WELDED JOINT ESTIMATED DISTRIBUTION OF TUBE RESIDUAL STRESSES FIGURE Vill-1 5

I l8 I TUBE A 11-66 WALLTHICKNESS I O O l 5 ~ ^^ ^ O [NON-ROLLED g .es, e EOR 1,0N,0,Tvee PERCENT WALL REDUCTION IN ROLL [ AREA = (.0363.0353) (100) = 280 ( (.0363) I j ~ I LENGTH OF ROLLIS ABOUT 0.6 INCH .035 l ROLLED PORTION OF TUBE I T O 1 2 DISTANCE FROM TOP OF TUBE SPECIMEN, INCH (ZERO DISTANCE IS ESTIMATED TO BE ABOUT I 1/16 INCH BELOW TOP OF TUBESHEET) I FIGURE Vill-2 I I I I

I 5 30 - ) 0 0 4 I 900*r 25 l 004 18 00*r j W 1200 r N 20 -E b ~ 35 c ~ saco r 5 Ii " to... stoo r s i naco r o. I 1 0 g 1 2 a 4 Time, hr I EFFECT OF HEATING TIME AND TEMPERATURE ON RESIDUAL STRESS OF COLD-DRAWN (200) ROD FIGURE VIII-3 I I I I

[ 3 I 30 V -A i i 1 V 900*F 1800*F l 1000*F ^ E 12CC*T 20 1 o I E [.15

?G I

l2 1300*F " to I. 1SUO*F 5 5 1600*F o f f f I O 1 2 3 4 Time, hr I I EFFECT OF HEATING TIME AND TEMPERATURE ON RESIDUAL STRESS OF COLD-DRAWN, ANNEALED ROD FIGURE VIII-4 I I I

I I I I TMI-1 COOLDOWN TRANSIENT 600 5 %'N SHELL TEMPERATURE (ESTIMATE) s N.-. - -.s 4 TUBE TEMPERATURE

............... 3. -.

(TC MEASURED) .,,Ns 500

...,N E

s n 140 F [ gs,* 90 F 400 N..,, ...y e v ly H B& W CALCULATED 300 SHELL TEMPERATURE IE2 B&W CALCULATED !y TUBE TEMPERATURE 200 I (B&W CALCULATED VALUES ARE FOR DESIGN BASIS COOLDOWN) 100 I l l l l l O 1 2 3 4 5 6 I TIME (HOURS) I I T,., s m e e N e R A T O R T e P e R A T U R e s DURING COOLDOWN TRANSIENT OF 9/6/81 FIGURE Vill-5 I

I IX - 1 IX. CRACKING TESTING A. Test Program A cracking testing program has been developed and tests are underway at the B&W Alliance Research Center, Alliance, OH, Oak Ridge National Labora-tories and Battelle Columbus Laboratories. The programs consist of five I phases:

1) reproduce intergranular attack (IGA) in primary water environ-ments, 2) arrest the attack, 3) verify the cracking scenario, 4) determine that the attack will not resta rt upon plant start up, and 5) assess effect i I of tube repair techniques on material resistance to IGA.

Results from this test program will be recorded in a separate report, Reference 11.

1. Reproduce IGA a.

Pu rpose The purpose is to produce IGA in OTSG tubing in primary water and contaminated primary water, thus identifying the probable causa-tive chemical (s).

b. Test Mate rial The test material is Alloy 600 OTSC tubing with a heat treatment i

equivalent to TMI-1 (i.e. mill annealed plus ll50*F/18 hours). Some tests will also be done with solution annealed plus stress relieved and actual TMI-l tubing. I

c. Test Specimens I

Test specimens are stressed longitudinal strips of tubing. Un-s tressed rings of tubing.

d. Test Methods Electrochemical corrosion tests are being used to screen test environments until causative environments are identified.

I Conventional exposure tests are being used to verify results obtained f rom the accelerated electrochemical corrosion test s.

e. Test

1. Baseline Boric Acid Environments 5000,

Hf03, 13,000 ppm H B03 3

2. Thiosulf ate Contamination a) Borated water + 100,10,1 ppm Na2 2 3 S0 b) Borated water + 200 ppm N H2 4 and thiosulf ate

3. Sulfate Contamination I

a) Borated water + 100,10, 1 ppm Na2SO4 b) Borated water + 100 ppm N H2 4 and sulfate I

I I IX - 2

f. Test Temperatures Test temperatures of 1000F, 1300F, 170 F and 5500F are being used.

2. Arrest the Attack

a. Pu rpose Once the aggressive environments are determined, tests will be I

conducted to determine if various proposed changes will prevent the attack f rom proceeding.

b. Te s t Ma te rial I

c The test material is Alloy 600 OTSG tubing with a heat treatment equivalent to TMI-1 (i.e., mill annealed plus ll50'F/18 hours). Some tests will also be done with solution annealed plus stress relieved and actual TMI-1 tubing.

c. Test Specimens Stressed tubular tensile specimens.

d. Test Method The causative environment will be used to partially crack the I

Alloy 600 test specimens which are loaded in tension a load cell is used to reflect cracking. Several possible fixes will then be tried to determine if cracking can be stopped f rom con-tinuing; this would be indicated by no f urther drop off in load, e. Possible Fixes

1) Dry layup with dry air

2) Add chemicals to boric acid solution that may react with the sulfur

3) Add LiOH2 to boric acid

4) Gaseous ammonia in dry air 5)

Sodium tetraborate (soak) I 6) Heat up to 250*F to oxidize S - SO4 in presence of primary water 7) Heat up to 250*F to oxidize S - SO4 in presence of dry air I I

I I IX - 3

3. Verif y the Cracking Scenario a.

Pu rp ose As set down in Section X of this report, a scenario defining the probable cause of f ailure has been established. In order to verify this scenario, testing is being perf ormed to duplicate those circumstances,

b. Te st Material The material is actual TMI tubing and Inconel 600 archive tubing with a heat treatment equivalent to TMI-1 (i.e., mill annealed plus 1150*F/18 hours).

I

c. Test Methods I

Utilizing C rings, bent longitudinal tube strips and full section tubing, the actual hot functional thermal cycle is being dupli-cated with the exact primary system water chemistry at that time. To this chemistry is added 1 ppm sodium thiosulf ate I contamination up to a maximum of 10 ppm. In addition, tests will be conducted with sodium sulf ate contamination in the 1-10 ppm range.

4. Determine That Cracking Will Not Restart a.

Pu rpose The purpose is to establish if the sulfur film on tubing surf aces or in intergranular attacked areas can produce additional damage I under aqueous shutdown or operating conditions.

b. Test Material Actual TMI tubing samples with and without sulfur removal.

c. Test Method Tests will be conducted using C ring and bent beam speci-mens placed in primary coolant with Boron concentrations repre-sentative of variations which occur during core life. Diluted sulf ate and chloride contaminants will also be introduced into some test environments.

These tests will be conducted at 170*F and 550*F. I I I

I IX - 4 i h

5. Assess Ef fects of Tube Repair Techniques on Material Resistance to g

IGA l a. Pu rp ose I Repair techniques which involve expansion and/or rolling of the tubing within the tubesheet will introduce additional residual I stresses into the tube. Tests will be conducted to assess if the final repair configurations will af fect tube performance f rom l corrosion standpoint.

b. Test Mate rial The test material is actual TMI tubing material with and without intergranular cracks and with and without I.D. surf ace cleaning.

c. Tes t Me thod g

3 Tests will be conducted on specimens which mock up the repair configuration and subject the sample to typical operational temperatures and environments. Additional environments will I include secondary side chemistry with caustic and primary side chemistry with diluted sulf ate and chloride contamination. B. Conclusions

1. To date, early cracking tests have shown that the cracking mechanism is not currently active in the cleaned up primary coolant.

2. The presence of an inert atmosphere reduces cracking tendency.

3. Sulfate at low temperature will not cause cracking under oxidizing conditions.

I

4. Thiosulf ate at low temperature will cause cracking under oxidizing conditions.

I

5. Archive Material in the mill annealed and stress relieved condition has cracked in 5 ppm thiosulfcte, but not in 1 ppm thiosulfate, indicating a threshold concentration requirement greater than 1 ppm.

I I I I I

I X-1 X PROBABLE CAUSE OF FAILURES t A. In t roduc t ion In this section of the report an attempt will be made to combine the results reported in the other sections into a coherent failu re I scenario which can explain the major features of the OTSG tube cracking phenomenon and is consistent with what is currently known about the OTSG fabrication and operating histories. Additionally, the implica-tions of the failure scenario with regard to plant recovery activities I are noted and recommendations based on the conclusions in this report are presented. I B. Development of a Cracking Scenario The occurrence of intergranular stress assisted cracking (IGSAC) I requires that three conditions be satisfied simultaneously: a sufficiently high tensile stress I a susceptible material microstructure an aggresive environment. I The information presented in the other report sections relating to these three factors is summarized below. 1. Tensile Stress I Section VIII presents information about OTSG tubing s t re s s e s. Since the cracks are oriented circumferentially in the I tubes, axial tensile stresses are of principal interest. Both operating and residual stresses must be considered since both can play a role in IGSAC (for example, both categories of stress are I involved in IGSAC of BWR stainless steel piping). Cracking must have occured in a situation in which the sum of the operating and residual stresses in the axial direction was greater than that in the hoop direction otherwise the crack orientation would have been I axial. The Section VIII analyses indicate that this condition is satisfied during cooldown and cold shutdown. Highlight s of the stress analysis from the failure scenario viewpoint are:

a. Tubing axial tensile stresses are largest during cooldown when they may approach the yield stress.

b. Significant axial tensile stresses also exist during cold shutdown.

I I I 1

I X-2

c. Locally high axial tensile stresses are possible in the seal weld heat affected zone and in the vicinity of the roll t rans it io n.

I

d. Under heatup and at full operating temperature the hoop stress generally is larger than the axial stress.

I

e. The axial stresses are generally larger at the periphery than in the center of the tube bundle.

Thus, the stress analysis results suggest that the cracking must have occurred during cooldown or during coli shutdown. The s t re s s analysis also explains why the seal weld heat af fected zone and the roll transition region should be particularly prone to I cracking and why more cracking occured in the periphery than in the center of the tube bundle. 2. Susceptible Material Microstructure The OTSG fabrication history is documented in Section VII. The fabrication history puts the tubing into service in the mill I annealed plus stress relieved condition which is expected to be heavily sensitized (i.e., low grain boundary chromium content). Metallurgical examination has confimed that the expectec micro-I structure is present. Thus, the fabrication history results indicate that the tubing microstructure was highly sensitized, and therefore susceptible to attack by water cont aining sulfur oxyanions. 3. Aggressive Environmer.t l The results presented in Section V indicate that sulfur was ! E present in the primary system water and three possible sources of sulfur have been identified f rom the OTSG chemistry history (Section IV). As indicated in the discussion presented in Section I IV, if SO{ and S 02 3 were introduced to the primary vater as the OTSG operating and chemistry histories suggest, they would be expected to persist as long as the water was at room tempera-ture even if the oxygen content of the water was reduced by hydra-zine additions. However, hydrogenating and heating the water to perform a hot functional would be expected to result in the I generation of S-, possibly accompanied by S and other inter-mediate species. Subsequent cooling to room temperature and oxygenating following the hot functional would rapidly oxidize S-to S and could also result in the appearance of significant I concentrations of other species of higher oxidation states. Although it is not possible to predict either the identities or I I I

X-3 the concentrations of the sulfur species present following the hot' functional, it is clear that this transient is likely to have greatly af fected the aggressiveness of the environment with regard to low temperature sulfur-induced attack of the OTSG tubing. 4. Proposed Failure Scenario Tne discussions presented above suggests that the probable I cause of failure was as follows: a. During layup the primary system was contaminated with sulfur I by the accidental introduction of sulfuric acid, sodium thio-sulf ate, and possibly a sulfur-containing oil. The anount of sulfur present may have reached several ppm, but the con-I taminated water was not aggressive enough to crack mill annealed plus stress relieved Alloy 600. The cracking test s confirm that cracking would not have been expected to occur at this stage. b. The temperature and oxidation potential transient associated with the hot functional test resulted in a change in the I types and concentrations of sulfur species present in the prima ry wat e r. Further changes occured when thiosulfate-contaminated water was injected during the rests of the HPI and LPI systems, c. When the water level in.he OTSGs was lowered following the hot functional, high concentrations of aggressive metastable I sulfur species developed in the dry-out region at the top of the generators due to the combined effects of solution con-centration by evaporation and the comparatively high avail-I ability of oxygen. Changes in the sulfur spe(les 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 tubing occurred rapidly in the dry-out zone with preferential attack at high stress locations. Little or no cracking occurred below the water line because the bulk solution was less aggressive, Cracking terminated either because continued chemistry e. 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 suf ficiently dif f erent from its state immediately af ter the hot functional to prevent a full scale recurrence of steps e and d in the new dry-out

zone, I

I X-4 f. Cracking was discovered when the OISGs were pressurized. 5. Eddy-Cu rre nt Indications Below the Surf ace I The recent eddy-current indications of defects below the initial water surf ace requires special mention. I The fact that eddy-current indication at about the 1 volt level was observable at the 3rd Tube Support Plate (TSP) +3" in tube A 149-34 on 5/7/82, but not at that location on 4/17/82, implies that either cracking is still taking place or that earlier I eddy-current examination did not detect this particular defect. The flaw growth program indicates that detection repeatibility of 1 volt and less signals is about 75% ( Re f e re nc e 12). I c The possibility therefore exists that the indicated defect existed but went undetected on the 4/17/82 examination. Th e f ac t I that all known I.D. defects as called by eddy-current testing are greater than 50% through wall with the majority nearly through wall and that the flaw growth program revealed no new defects with signals greater than 1 volt, indicate that cracking is not an on-going process. As indicated in the above f ailure scenario, the sulfur pro-I moted intergranular stress assisted cracking responsible for the observed defects in the OTSG tube depends on a concentration of aggressive sulfur species, which appears to exist only temporarily during a reduction-oxidation cycle of the solution present in the reactor coolant system during hot functional testing. It is un-likely that this concentration of aggressive sulfur species would persist some months af ter the tubes were exposed to atmospheric I oxygen and then suddenly initiate cracking. The OTSG tubes below the 7th TSP were re-wet on 4/6/82 with water which was analyzed on 4/26/82 to have 24 ppb sulfate and 1.06 ppm lithium. That solution I remained at that level until 5/9/82, when the OTSG was again drained. In the absence of another redcction-oxidation cycle and in the presence of the benign solution described above, it is probably that the defect was not initiated subsequent to 4/17/82. The question remains as to how a crack might have been initiated below the initial dryout zone. The region of the OTSG tubes above the water level immediately following hot functional testing (about 9/8/81) was down to about the 13th tube support plate. The vast majority of defects are in this region. Water level slowly increased until it was above the upper tube sheet on 9/27/81. I I

X-5 It is supposed, as indicated in the f ailure scenario, that rapid oxidation of the solution lef t on the exposed tubes led to significant concentrations of aggressive sulfur species and hence the wide spread attack on the tube s. Oxidation of the bulk solution I would have proceeded much more slowly, yielding a less aggressive environment. In fact, it now appears that oxidation of the bulk solution may have been even slower than at first thought, such that some reduced sulfur species persisted until water level reductions I on 10/6/81 and finally draining of the OTSG in early December 1981 exposed a thin film of the bulk solution to an oxidizing environ-ment. The resulting concentration of metastable sulfur species, while much less damaging than the initial solution, could cause isolated attack on the OTSG tubes. Alt e rnat ively, the bulk solution may have oxidized more rapidly than originally thought, again resulting in a marginally aggressive solution below the water surface, decreasing in aggres-siveness with depth such that the high stress regions associated I with the lower tube sheet were protected. The proposed f ailure scenario does not require that there be I no defects below the initial water surface. In fact, the total absence of such defects would require a rather special oxidation rate of the bulk solution, as suggested by the foregoing dis-I cussion. The existence of the special oxidation rate is plausible but its absence is not surprising. 6. Scenario Summary This scenario is consistent with all of the observed features of the cracking phenomenon (with the possible exception of the non-axisymmetric radial distribution of cracking in OTSG-B) and is also consistent with the timing of the cracking and the results of the metallurgical examinations and corrosion tests. Clearly the key feature of the scenario is the generation of a highly aggres-sive, transient environment following the hot functional test. 7. Implications with Respect to Plant Recovery The scenario proposed above suggests that the cracking process could be reactivated by the temperature and oxidation potential I transients that are likely following repair during preparations for plant start-up unless the inventory of sulfur in the primary system is reduced to a low level and positive steps are taken to prevent recont aminat ion. Accordingly, an attempt should be made to remove I sulfur-containing surf ace films f rom the OTSG tubing and other primary system surface s. Oxidation to generate soluble sulfur 4 species followed by removal via demineralization is thought to be the most promising approach. I I

X-6 D. Conclusion I A scenario based on IGSAC by metastable sulfur species in the period immediately following the hot functional test can explain the I major features of the OTSG cracking phenomenon and is consistent with the findings of the various activitie s undertaken by the Failure Analysis Task Group. E. Unresolved Questions Raised by Failure Analysis 1. What caused the distribution of damage in the B OTSG7 a. The damage is probably stress related.

1) There is no clear diff erence in environment.

2) There is no clear dif ference in material.

3) There is clear stress dependence in A OSTG.

b. The detailed answer to the questions will not impact: 1) Plant rec ove ry.

2) Subsequent operations.

Further study of this question will be carried out following c. the closeout of the Failure Analysis Task, and the Final I Repo rt. 2. What role, if any, did oil play in the OTSG damage? a. Oil was not the sole source of the damaging sulfur. In fact, it is unlikely that any sulfur was released from the oil available for introdution into the Reactor Coolant System ( RCS). b. One cannot say with certainty what effect if any that oil, if I pre sen t in the (RCS), had on the OTSG. The detailed answer to the question will not impact: c. 1) Plant rec ove ry.

2) Subsequent operations.

J. Further study of this question will be carried out following the closeout of the Failure Analysis Task, and the Final Re po rt. I

I X-7 F. Recommendations 1. Additional water samples should be taken and analyzed for sulfur levels below.1 ppm, to better quantify existing sulfur levels. 2. A cleaning program should be conducted to remove sulfur con-tamination in the RCS and related systems. Cleanup is necessary to I preve nt subsequent damage to equipment resulting from sulfur at-tack. A sampling program is recommended to verify the effective-ness of the cleaning process relative to sample results from 1. above. 3. Repair procedures and programs should be based on the consideration that currently known eddy-current defects are 100% through wall or I will be through wall af ter several thermal cycles due to fatigue of c the remaining ligament. I 4. Based on the sulfur species present, in addition to the system cleanup of step 2. above it may be necessary to take steps to remove the surface films if they are in a form which could reactivate and become corrosive. Additional cracking tests I scheduled will aid in making this decision. 5. Materials utilized in the repairs of the steam generators, need to I be carefully specified to assure a material which would have ade-quate immunity to future corrosion problems. In addition, these materials need to be qualified by acceptable techniques to assess the degree of sensitization. 6. Plant operational chemistry specificatio:;s should be revised to address control of sulfur in the primary and secondary coolant systems. 7. Steps must be taken to preclude the introduction of chemical con-taminants into the RCS and its support systems. I I

I I REFERENCES 1. Jo ne s, J. (1982), J.D. Jones, OTSG Failure Analysis Operational History Final Report, GPUN TDR #336, May 12, 19 82. 2. Boberg (19 82), R.L. Boberg, D11-1 OTSG Fabrication History, Babcock and Wilcox, May 10, 1982. 3. Moore (19 82), J.P. Moore, DiI-l OTSG Tube S tress Analysis, GPUN TDR undated. 4 Jo ne s, R. (1982), R.L. Jones, EPRI Activities in Support of Dil-1 Steam Generator Recovery, EPRI Memorandum Report to GPUN, April 1982. 5. Newman (1982), R.C. Newman, Unpublished research presented at NRC Failure Analysis Team Meeting GPUN Headquarters, Ma rch 9, 19 82. 6. De (1975), P.K. De, Ph.D thesis, Ohio State University 1975. 7. Berge and Noel (1982), P. Berge and D.D. Noel, Presentation at EPRI Materials and Corrosion Committee Meeting, EPRI, Palo Alto, January 19-20, 1982. 8. Airey et. al. (19 81), G.P. Airey, A.R. Vaia, N. Pessal and R.G. Aspden, Detecting Grain Boundary Chromium Depletion in Inconel 600, Journal of Metals, Vol. 33, No. 11, 1981. 9. Cowfer (1982 A), C.D. Cowf er, Report on Dil-1 OTSG Tubing I Damage Fourth Quarter 1981 Failure Mode, GPUN Technical Functions Division Report, undated. 10. Rigdon and Pa rdue (198 2), M. A. Rigdon and E.B.S. Pa rdue, Evaluation of Tube Samples f rom TMI-1, Babcock and Wilcox Resea rch and Development Division LRC, undated. 11. Gsacobbe (1982), F.S. Giacobbe, DiI-l OTSG Cracking Test Program, GPUN System Laboratory Report, undated. 12. Cowf er (1982 B), C.D. Cowf er, Dil-1 OTSG Task 4 - Eddy-Current Defect Data Evaluation - Indication at TSP 03 + 03 Inches - Tube A 149-34, GPUN Memorandum MT/2043, by 12, 198 2 to D.G. Slear. 13. Kazanas (1982), N. Kazanas, Results of Inspection of Dil-1 RCS. GPUN TDR, undated. I}}

ML20064M520

Contents

Text

SUMMARY

SUMMARY

=

SUMMARY

Navigation menu

ML20064M520

Text

SUMMARY

SUMMARY

=

SUMMARY

Navigation menu

Search