ML20085H248
| ML20085H248 | |
| Person / Time | |
|---|---|
| Site: | Three Mile Island  |
| Issue date: | 04/15/1983 |
| From: | Dillon R Battelle Memorial Institute, PACIFIC NORTHWEST NATION |
| To: | NRC |
| Shared Package | |
| ML20083L846 | List: |
| References | |
| NUDOCS 8309070453 | |
| Download: ML20085H248 (15) | |
Text
__
t t ,
Attacha:nt 3 l CONSULTANT REPORT TO NRC, PAUL WU, ON TMI-1 RESTART R. L. Dillon Pacific Northwest Laboratory e Richland, Washington 99352 April 15,1983 INTRODUCTION This is an analysis of written and oral presentations to NRC-NRR by GPU and its consultants.,on the circumstances and corrosion effects of sulfur incursions into the TMI-1 primary coolant system. The analysis primarily concerns itself with the readiness of TMI-1 for restart, with l
some possible corrosion consequences of sulfur removal, or non-removal.
Salient points relating to the GPU investigation and proposed actions have been reviewed and commentary prepared on still unresolved questions.
In review of the GPU program, the following reference materials have been used: ,
- 1. Assessment of TMI-1 Plant Safety for Return to Service After SG Repair, Topical Report 008. Rev. 2, March 29,1983.
- 2. Safety Evalu~ation of TMI-1 RCS Cleaning, Topical Report TR-010, .
Rev. O, March 3. 1983.
- 3. GPU Nuclear - TMI-1 OTGS Repair and Return to Service, NRC presentation, i April 5,1983.
- 4. Various draft consultant reports. .
- 5. R. C. Newman, et al. , " Evaluation of SCC Test Methods for Inconel 600 in Low Temperature Aqueous Solutions," Symposium held at NBS, Gaithersburg, MD, April 26-2S,1982.
- 6. Various texts on inorganic chemistry.
- 7. Personal notes summarizing discussions at NRC scheduled hearings; also '
phone conversations with GPU staff to request clarification of technical points. ,
i.
- ' 8309070453 830825
- PDR ADOCK 05000289 L: F 9DR
..-. . .= - -- . .. - .. .
l
. . l l
1 SULFUR INTRUSION AND ITS CONSEOUENCES The i'ntroduction of sulfur, species to the TMI-1 -primary systen has apparently occurred several times. Circumstances were unfavorable to steam l generator attack until the period August - November,1981. At that time, a 1-2 ppm thiosulfate residual from an. earlier contamination incident of
! May,1981, possibly aggravated by additional thiosulfate infections from the Borated Water Storage Tank, was present in the primary coolant. Follow-ing a hot functional test a fluctuating water level was established in the ,
upper steam generator region. Both enhanced sulfur deposition and contact with oxypen occurred in the regions.of the tube, later identified with cir- .
ctanferential IGSCC.
Tube samples suspected from EC testing to contain cracks were removed
! . from the OTSG for metallurgical examination. Tubing failures are due to IGSCC circumferential cracks. Cracks are initiated from the ID. No secondary mode of failure has been observed. There are, however, " islands" of IGA .
1.5-3 mils deep associated with ID deposits. Severe cracking is usually related to more severe IGA. Most of the cracks examined by metallography or bend tests have been through wall. Grain boundary carbon morphology and chromium depletion fully confirm the sensitization expected from 'the SG fabri-cation history. . SEM inspection confirmed the cracking; ID tube surfaces and crack walls were'shown by the EDAX, Auge and ESCA analysis to be contaminated with carbon species and sulfur. The carbon, in the form of carbonates (exter- -
nal surface) and hydrocarbons, was present in substantial amounts, apparently resultino from an overflow of oil from the miscellaneous weste storage tank into the reactor coolant bleed tank, thus into the RCS. Having duly noted the carbonaceous surface contamination and its source, the oil contamination T-is ' disregarded in further analysis of the corrosion events. A significant fraction of the carbon deposition may have occurred subsequent to the failure.
).
. Examining the sulfur contaminated surfaces in some detail has established ,
the sulfur to exist superficially as 50,". Deeper within the oxide the sulfur exists as metallic sulfide or some other reduced sulfur species. The concen-
)
tration of the sulfur in the' crack prone tube elevations ranges up to 8 atm
~
4 w - , . - - - - . ----.y- 7 -,,-r,m-n 9p.-- .,,,--,w --
, , , - . . - .--.--,w ------.-y,-m,g. 9 ,, 9 94-r,e: --we - - .
3 percent. Sulfur concentrations are roughly constant down to the 9th tube support plate, decreasing in lower parts of the tube and the rest of the primary system. There is some uncertainty in superficial sulfur concentra- l tion, except where confirmed by Auger /ESCA methods or by chemical analysis.
Quoted sulfur concentrations are generally based on swipe sample chemical analysis. Swipe sample sulfur is suggested to be about 10% of the total sulfur present, but the relationship is somewhat shaky.
l Laboratory tests confirm the cracking susceptibility of TMI-1 tube samples to very low concentrations of thiosulfate. In the absence of thio-sulfate, no cracking has been produced on Inconel samples in primary system water chemistry. In thiosulfate solutions and the absence of oxygen, no cracking has been observed in laboratory. Crack initiation and growth are temperature dependent. For susceptible material, time to crack initiation is minimized and crack growth rate increased at 170*F.
These observations reported by GPU sumarize the basic infomation -
relating to the condition of OTSG tubes and the water chemistry during the I critical post, HFT time period, where the cracking reactions are likely to haveoccuhed. The reactor observatioris seem internally consistent and com-patible with confimatory experimentation.
4 DISTRIBUTION OF IGSCC TUBE FAILURES
~
The physical description of the OTSG is satisfactory in accounting for
. the strong association of circumferential cracks with peripheral tube locations. ,
Localization of the crack indications in the peripheral tubes and the random ,
occurrence among the many Inconel heats represented in the peripheral tube regions emphasizes the importance of axial stress and the absence of important heat to heat variations in susceptibility. The piesience of longitudinal cracks in the tubes below the weld and above the axpanded tube region are due to local stress conditions, not of major concern either in assessing proposed remedial I
actions or the' question of sulfur removal from the primary system.
The thiosulfate induced IGSCC is a phenomenon of sensitized Inconel 600.
The relatively high temperature annealing temperature applied in the fabrication
. g O
of the TMI-l SG leads to a very high sensitivity to oxidizing sulfur species.
l Lot-to-lot variations in tube composition and metallurgical conditions are clearly indicated as extraneous ' issues.
The NDE methods have been shown to reliably identify partial wall pene-trations . This includes tube cracks within the tube sheet. A good correletion has been demonstrated between EC indications and subsequent metallographic examination of recovered tube sections. This observer was left with the feel-ing that the GPU EC effort was State-of-the-Art, with some very useful innovations in probe design to share with the industry. The numbers of cracks and their vertical displacement along the length of the tube are consistent with established stress distribution in the tube and reconstruction of the sulfur contamination event (s) previously considered in this comentary.
l CORROSION OF THE BALANCE OF THE RCS The extent and intensity of sulfur induced IGSCC in the steam generators raises serious concerns about the presence of similar corrosion in t's balance of the plant wetted by the recirculating primary coolant. The utility has undertaken a comprehensive examination of the balance of primary system com-ponents. The focus has been on those materials known to be susceptible to sulfur induced IGSCC, particulary those components in a highty stressed condi-tion. Rationalizing from the OTSG experience, attention has been paid to coolant-air interfaces and wet-dry areas. The salient features of the inspection are these:
- Thousands of observations have been made.
e Safety critical items have been identified for careful examination.
. e Comoonents representing combinations of material, metallurgical condition and stress enviroment which favor IGSCC have been inspected.
. e A variety of NDE methods have been used suitable to the component and ,.
its location: EC, UT, . radiography, dye penetrant, visual observation.
- NDE methods have been supplemented by metallography and microscopic inspection where it is feasible to take samples.
l l No evidence of IGSCC has been.found.
{
'. ~ .
f 7
While I am not qualified in NDE and not prepared by intimate knowledge of the primary system comconents to judge actual testing locations, the plan itself is persua ,ive. I am left with the impression that the diversity of NDE methods snployed, the variety of materiais and components tested, the selection of highly stressed components and the sheer number of observations are assurance that there will be no epidemic of component failures on start up.
Supporting System Inspections undertaken in addition to the RCS include the Spent Fuel System, the decay heat removal system and the building spray system. These inspections were part of a three year program initiated as a result of cracked recirculation piping in the' fuel storage system. Follow of known indications (presumably in the fuel storage system) show no evidence of crack growth. The affected section of 304 Fuel Storage Recirculation pipe i has been replaced with 304t..
More recently cracks have been identified in the Waste Disposal Gas Piping.
Crackinp is associated with sulfur contamination, presumably there as a r'esult of a. steam / condensate transport process. Affected 304 piping has been replaced by.304L.
Inspection of PORV's has shown one of them to be sulfur contaminated.
-PORV bodies have been cleaned and inspected, internals have been replaced. As a general comment relating to the failures of the piping and PORV corrosion, replacementynf damaged material is feasible and has been done. No basis is
~
apparent for recontamination and reestablishment of corrosion attack.
CLEANUP OF SULFUR FROM THE'RCS .
GPU proposes to remove sulfur residuals in the corrosion product film.
Sulfur is present on the corrosion product surface as sulfate. Within the film the sulfur is present in metal sulfides. The removal process consists l
of oxidizing sulfur compounds to soluble sulfate. The solution to be cir-culated in the RCS, purification system and decay heat removal system, will -
consist of borated, s 2000 ppm Baron, and litheated water, s 2 ppm Li, adjusted to pH 8.0-8.2 with 30 wt/o amonia and containing 15-20 ppm of l
4.
t hydrogen peroxide (added as 30 w/o product); peroxide additions to be made as necessary to maintain concentration. Solution temperature will be main-tained at 130*F with nump heat. '
The cleaning process is expected to take 2-3 weeks, and will be con-cluded when additional 5 is no longer disolved. It is anticipated that t
50-80% of the sulfur present will be removed.
Sulfur cleanup it. Justified on the assumption that sulfur dissolved and removed from the RCS is eliminated once for all as a future corrosion threat. The question remains unanswered as to whether the 20-50% of the sulfur expected to survive th'e cleahing procass .is a different, lesser
problem. It is claimed and probably true that the unremoved sulfur is inaccessible for one reason or another, and would be unlikely to' participate in a reacti"ated sulfur initiated corrosion process.
Convincing argument that any special measures need to be taken to remove superficial sulfur is more difficult. If the sulfur put in solution by tihe peroxide cleanina process is largely sulfate, not much is achieved. The sul- .
fate is not an SCC initiatcr of the power of more reduced fonns like thiosulfate, tetrathionate or perhaps polyscifides. It is true, however, that there are well authenticated cases of SCC in dilute, < 1 ppm S0 g", oxygen containing sulfate solutions.. If. it.has not been shown that SCC occurs in low temperature solutions, neither has it been shown that it does not. It should be borne in mind that a considerable inventory of sulfur is associated with the primary system surfaces. ,
Rough approximations suggest a number as high as 3 Kg. In solution this -
amount of sulfate approximates 10 ppm. The amount of surface sulfur is esti-mated largely from swipe analysis. It is assumed that the swipe picks up 10%
!~ of the sulfur present.. Estimating 2x10 5 ft2 for the primary system pressure l 3
- boundary surface area and an upper limit of 15000 un S (as sulfate) per ft ,
l the . inventory of S (SO S ") in'the primary system is 3 Kg.
If'the 3 Kg of sulfur ,
(as sulfate) were dissolved in the estimated 3x105 liters of primary coolant, the concentration would be 10 ppm.
l
,. y Such concentrations of sulfate in the presence of oxygen could be a' matter of concern to sensitized 304 stainless steel. This matter will be examined in the Lead Corrosion Tests Loop 3, in which Na5 will be added in order to build up sulfate in the recirculating solution to levels anticipated during the reat: tor cleaning operation (by GPU; this writer suspects levels mightbehigher). Sensitized and stressed 304 specimens should be included in the complement of corrosion specimens.
GPU has committed itself to provide information on the level and distri-bution of' residual S in the cleaned tube specimens.
s EVIDENCE ON THE CESSATION OF SULFUR INDUCED IGSCC -
- Laboratory tests have proven unable to initiate cracking or cause crack growth in tube specimens exposed to solution taken from the decay heat coolingsystem(0.5ppmS0[].
- Periodic monitoring of tube crack indications in reactor shows no change with time.
- ' Cracking stopped either because of continuing chemistry changes or because sulfur concentration in the dryout region was diluted back to average S concentrations.
- If cracking mechanism should reactivate during start up or cooldown, the precritical testing sequence should produce ter^ mtion through wall which can be detected by leak surveillance.
- Lower rates of calibrated crack growth should be detectable by ECT after .
the precritical test.
I have a concern in regard to tube rupture / collapse for circumferential cracks.
In the case of longitudinal cracks in steam generator tubes, calculations of residual properties are supported by a substantial experimental program. The significance of that program is such that a major NRC interest in the Surry 2A e steam generator (now at Hanford) was as a source of SG tube artifacts for study.
I understand the fracture mechanics calculations of residual tube properties in circumferentially cracked tubes are presently unsupported by experimental data.
g l
ACTIVITIES TO PREVENT RECURRENCE OF COOLANT CONTAMINATION AND CONSEQUENT CORROSION The thiosulfate tank has been removed from the system. An analytical procedure has been selected for sulfur and a sampling schedule established.
A chemical cleaning program (discussed elsewhere) will be conducted to remove sulfur residual on wetted primary system surfaces. Administrative controls have been put in force that are expected to practically eliminate inadvertent entry of other chemicals to the RCS. A consistency check comparing. solution conductivity to pH, boric acid and Li+ base line conductivity will identify significant chenical intrusfons. .
~
A very useful observation from BWL (Roger Newman) has shown that movement of sulfur compound anions to crack tips which is a necessary condition for f sulfur induced IGSCC can be prevented by the presence of competing inactive anions in sufficient concentration. Newman actually specifies Li+ concentra-tion 2 than 10 times the sulfur species anion concentration. The Li+ is in reality a measure of the boric acid anion concentration which is the species actually competing with sulfur ion.
[This principle is poorly described'in the GpU handout. No relationship is indicated between Li* and the associated boric acid anion
. LiOH + H 3B03 + LiH 2803 + HO 2
-Li+ + H 2803" I think it is important to be meticulous in discussing the physical principles '
involved in any of the recovery processes of TMI-1. : 1 _,,_.......-...r l
' e= w ' m ur,,,4 M
_-________7-___.
A question relating to the [Li+] to [S] concentrations is the ratio exist-ing at the time of the cracking incident. I don't know that the subject has been discussed. I would guess [Li]/[S] .somewhat less than 10 but probably greater than 1. This relationship needs consideration in the NRC review.
- * * /0 l
THE CORROSION TEST PROGRAM )
Objectives: .
1 l
- Determine conditions under which corrosion occurred and how arrested.
l l
- Verify proposed corrosion scenario.
- Relative ' sensitivity of kinetically expanded tube to corrosion.
- Verify H2 0: sulfur removal will not cause corrosion.
, The assumption that cracking in the steam generator has been arreste' d is based on: (1) stressed tube, sample filled with coolant taken from decay ,
heat removal system. (2) Eddy Current examinations of 100 tubes exposed l to primary coolant over a period of several months. No evidence of crack growth .has been observed.
The conditions under which IGSCC will cccur were explained in the .
following inquiries: (1) Electrochemical potentials regimes for cracking.
(2) Concentration of S 230 " in boric acid required to produce crackSg in sensitized archive tubes (5 ppm S230 "). (3) Tube specimens cut from TMI-1 OTSG cracked at 1 ppi 23 S 0 " for reasons not clearly established. (4) Tube samples exposed in barated water at the oxidation potentials of (3) above, but without S230 " present, did not crack. (5) Samples in deaerated horated water with ihiosulfate present did not crack. '
Scenario verification studies in loops at ORNL attempted to reconstruct the post' hot functional test cooldown condition in which tube cracking pre-sumably occurred. Cracking occurred in 5 ppm S 230 " s'olutions; no cracking occurred in 1 ppm S23 0 " or 30 ppm 50g".
To test the sensitivity of kinetically expanded tube sections, two tests were applied. The first in 10% caustic solution electrochemical test was known to cause SCC in highly stressed roll transition metal. Tb second, an
. accelerated test of highly stressed specimens in thiosulfate-contan nated T
boric acid. No evidences of cracking were observed in either test.
e s
- * ~
- ...~. . . . .
r The [Li+] : [S anion] ratio is also of interest in evaluating the effect of S removal from the primary system surfaces. Prudence might suggest the
- sulfur species not be allowed to exceed 0.1 [Li+].
l TUBE REPAIR Since the preponderance of tube cracking is in the upper few inches of the tube sheet, most of the tubes can be reexpanded and resealed into the tube sheet. The requirement is that at least 8" of sound tube is available to form the new seal. All tubes which remain in service will be kinetically i
expanded into the tube sheet. Those tubes which have crack indications in the new expansion zone or below (with certain exceptions) will be plugged rather than repaired. Those unrepairable tubes (plugged), for which a hazard of tube separation is identified, will be stabilized as well as plugged. That is, a red will be attached to the top plug of a length to extend into the region of maximun cross flow in the steaming generator.
The explosive expansion process has been experimentally evalua':d for any direct effects on the tube ID corrosion and for any hannful interactions
! with residual sulfur present in the corrosion product layer.
- The policy which dictates kinetic expansion and resealing of qualified l tubes and plugoing' of those tubes not susceptible to successful resealing .
l, infers a crack detection system of considerable reliability. The presence of even a few tubes capable of open-end breaks is a matter of serious safety concern. Thus it becomes a matter of establishing that the ECT is capable of de'tecting tube cracks of dimensions liable to plastic tearing or ligament _
necking. We are assured that nominally such cracks are within the Eddy Current detection range. In case unstable cracks are somehow missed, they will be exercised during the scheduled preoperational test programt "If the cracks are 1005 through wall and of a size to propagate to failure under loading, they will be detected by the leakage monitoring program."
e 4
, e
. 6
Because the two accelerated tests of kinetically expanded tube sections were crack initiation tests, a series of long term tests are in prcgress to address the problem of delayed crack initiation. These have been designated
" lead" tests. These tests will ant'icipate projected start up by 4 months or more and run about 13 months. They will include load cycling, thermal cyclir.g.
tube samples with and without crack indications, reactor cycle chemistry changes.
The test will approximately simulate operating conditions. Three loops are involved in the test. A two loop test (in progress) will explore the effects of sulfur c'ontamination as sulfate and as thiosulfate. One loop will use unexpanded TMI-1 tube samples precoated with imunol. A third loop will simulate the hydrogen peroxide cleaning process; no known defects will be included.
Generally, the laboratory tests seem well designed to test the questions asked as wall as others of equal urgency but not specifically asked. I believe the EC monitoring of 100 tubes conclusively establishes the cracking process is dead or dormant in the present reactor coolant chemistry. It is not informa-tive about the effect of the reducing water chemistry during operation. However, I these conditions are examined ir. the " lead" tests though with some loss in simulative accura'cy and with less statistical significance. I am prepared, how-ever, to accept the evidence presented that the sulfur induced IGSCC is arrested.
I' think the reconstructed picture of the tube cracking circumstances, pp. 13-14, is probably close enough to the actual to be acceptable. I _ - .- s.r.
- cerned about apparent contradictions regarding the cracking solution chemistry.
Pages 13-14 of the GPU Assessment document indicate reduction of 23 S 0 " in the ,
hydrogenated units of the hot functional test, followed by partial oxidation of .
the sulfide after cooling and contact with oxygen containing cover gas Hydrazine, S0" 23 S" + HO 2
. S- . e, . pa,tia,,, oxidizee fo,.s
~ ' ' '
~
_ . , J. . . .__.2. .. . ..:. ... ma Pp. 22-23 same document, laboratory tests produce cracking in 5 23 0 " + oxygen.
I don't know that the apparent 1nconsistencies in describing the cracking environment are important to the reactor recovery operation, but they certainly invite questions. ,
The lead tests evaluation of the cracking susceptibility of the k'netically
. expanded tubes are generally satisfactory. I would like to be assured that on conclusion of cleaning simlation a corrosion test (SCC) in cold, high oxygen, 10 ppm sulfate will Ne schedu.'ed. This exposure should equal or exceed the duration of the post cleaning lay up of' the reactor preceding the hot functional test.
- C0miENTS ON CLEANUP AND RESTART i
- Before dealing directly with any conclusions regarding restart, I should summarize the significant observations from the foregoing analysis. The I
nature of the cracking attack, its distribution and time of occurrence are adequately explained by reference to the OTSG design characteristics and the operational history. The extent of attack is defined by a careful examination t of the primary coolant system. The examination of the RCS system and support systems beyond the OTSG's has, with two exceptions, indicated the primary system components to be free of corrosion. The decision on startup seems i to hinge on the post repair steam generator condition and the possible
- reestablishment of an aggressive coolant chemistry. Corrosion is presently dormant, but the presence of a considerable inventory of sulfur compounds attached to oxide surfaces and admixed with oxide layers, creates the possi-bility of a reactivated corrosion process. Consequently, a cleaning process is proposed which will solubilize a large part of the sulfur so that it can be removed from the system. There are risks associated with the sulfur oxidation and removal as well as the alternative of living with a large S inventory in the system. ,
The risk of cleaning is that a relatively large inventory of sulfur com-l pounds (5 ppm or more) will be put into solution. Some significant fraction of the sulfur attached to prima'ry system surfaces is in a reduced ' form--sulfide g-- -m.
i . .
a
. z. -
or free sulfur. During the oxidation of the sulfur / sulfide to sulfate, inter-mediate oxidation states will be involved, some of which are likely to be crack inducing species. Axial stress levels in the tubes will be high at the proposed cleaning process temperature. The conditions for IGSCC are present if the necessary sulfur species are present. Confirmatory tests are run at sulfur levels (Na5 additions) that I am not certain are conservative. The level of metastable sulfur compounds in solution has been and remains a question.
Some comfort is provided by the lack of polysulfides, tetrathionates and the like in hexane extracts of cleaning solutions.
The cleaning process will proceed for 2-3 weeks building up to some unspecified level of sulfate in solution. The level could reach 5-10 ppm of sulfate. In the presence of oxygen and of high temperatures, this is a good recipe for SCC of. sensitized stressed stainless steel. At 130*F 'there is no SCC data known to me. I can site, however, the cracking of the TMI-1 fuel pcol recirculation pipe and the association of sulfur with the cracks.
Nobody knows if thicsulfate or perhaps sulfate ion was the specific ion resp'onsible, but most people would probably bet on thiosulfate.
The buildup of high sulfur levels in the cleaning solution is based on no solution cleanup during the process. The argument against cleanup is understandable but arbitr.ary; removal of S0 g" by ion exchange could be done.
The reasons for foregoing cleanup are explained'by GpU as the following:
- Follow of S0g" buildup i:; a means of determining when the cleaning process is complete. ,
- Sulfate ion concentration will establish the approximate levels of S in the system.
- Some breakdown of the IX resins will be caused by the -peroxide solution.
These considerations must be balanced against the virtues of removing sulfate from solution as low probability cracking agent. -
- There are possible hazards in leaving sulfur in the oxides. Some coolant chemistry condition could rapidly and unexpectedly put sulfur in solution where it might escape early detection. Given the proposed administrative l
i .. . ....b * .. 72*
. Md..a, . _. . .. .- .
- - ~
1 .
w_r._ .
.., _____z..__.. . _ _ _ _ . _ . .
l changes requiring sulfur analysis, this possibly is remote. The sulfur might undergo redox reactions putting the adsorped (or admixed) sulfur in an aggressive state. I don't much fear this situation for several reasons. The sulfur is pretty well fixed in the oxide at this time. Fonnation of new oxide will occur, possibly separating the sulfur from the oxide metal interface. Deeply penetra-ting sulfur species are likely to be relatively ininune from solution potential effects. A slow removal of sulfur can be expected with time as the normal corrosion process occurs during operation. These kinds of conjectures could be studied from " Lead" corrosion test specimens--and should be.
In reference to the cleaning decision, I am not strongly pro or con.
I tend toward not cleaning, but I would like the " Lead" test corrosion data to establish what happens to 5 presently incorporated in the corrosion product oxides. ,
I believe TMI-1 restart is appropriate. This view is confined to considera -
tion of . corrosion related factors. The likelihood of reactivation of IGSCC based on some manipulation of 'the sulfur inventor; now fixed in or on corrosion product surfaces is small . Release of sulfur to solution from the corrosion product is slow, amounting to days or weeks even for the cleaning process. The metastable species that appear capable of initiating or sustaining cracking reactions are rapidly oxidized to relatively inert species, with the result ,
that they can. only be present in the most minute quantities (ppb's or less as a guess)--a very different situation from the transient condition where 3-5 ppm of dissolved sulfide was suddenly oxidized during the crack initiating event.
The repetition of the sulfur contamination incident is precluded physically arid administratively.
The proposed preoperational high temperature exercise of the reactor is an effective proof test of the cleaned, repaired system. Most of the existing uncertainties on the reactor condition--especially mechanical factors--can be resolved by successful performance in the precritical test. ,
I am reassured on the startup by another consideration. I think the GPU investigation has been very wel.1 handled. By, and large I have been impressed 9
e h
.~.
. - . . . . , = . ._.
, ir e m;. , . .. h _, ,_
..3 with their representations in meetings and in reports. Reactor startup is not infrequently a chaotic and confusing affair. TMI-1 restart will involve experienced people with new but improved procedures.
g 6
.g O
O 4
9 e
E 9
9 O
O e
e e
o O
e t
4 e
_ _ _ _ _ _ _ . _ _ _ _ _ _ _ _