ML20079G627
| ML20079G627 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 09/10/1982 |
| From: | Mccracken C Office of Nuclear Reactor Regulation |
| To: | Benaroya V Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML20079G498 | List:
|
| References | |
| FOIA-83-243, FOIA-83-A-18 NUDOCS 8209280423 | |
| Download: ML20079G627 (7) | |
Text
.
...;...,g...,.
'o L.
f 6
- ll.
JISTRIp!E10R:
% ntral File CM5B' Reading File TMI-1 File W l0 IEI-n>
MEMORANOW FOR: Victor Bemarsys. Chief Chemical Engineering Branch Fa0N:
Ceared McCracksa. Sect 16e Leader Cheetcal Teressley Section smJECT:
STA155 Up0ER FOR TRI HO.1 STIAM GEERATOR REPAIRS Three meettags ears held since the last update was issued on Augu:;t 17, 1982.
p' Ampest 25. T M. Perstppesy: Independent third party review of th8 8888IP pf988e8. [y.3..
SER revitar, fatipe n
4..
-~
Sepeseter 1.198t. Persippoort prettel tem of preye ravfee, teinary states of
.t..
septesher 2.19st. FiOpesy uttaess leak tests of goalifi-entfem blocks;-
Igi: -
Us,.. -
tased as discussless et thees meettags, the 'currest schedele is as follows:
WJL Septesher 15. Iget. Oggtsees: WUR Fresentation to staff and FRC to $sstify lett.iatiaat o.f. repairs.
c Septesher 17. 1982: $$tff'fssue repair SER.
Fellowing this espose11a seetry. ECT and other appropriate tests will he perferued to verify acceptable espansions. Assuming acceptable ren'alts from the first expension, production expansion will start a few days hter.
- 0ctober 13, 1982 Setheeds: GPUN presentation on status of all progrees afflecting recovery to staff and staff consultants.
- 0ctober it.1982. Bethesda: Staff and staff consultants meet to discuss CJN's recovery pv43rass and areas of concern.
- 1tovember 1982: GPUN suturits safety evaluation addressing all aspects of recovery from the OTSG corrosion problem.
- Docamber 1962: Staff issue restart SER.
l i
>p,. w gg..,gg,g.sub34n~ W ehunis". M... hpy.Ha
-o
<,... (6owso3@
M
l
~
Victor Benaroya gp 7 g g 3
[.
Meeting Sumasries 1
'y 08/25/82 :
Independent third party review (!TPR) meeting on repair process;
,)l the prelimina7y findin@ of the ITPR are to be issued by 9/10/82.
l The staff will be prov' ded with this information by the Sept.15 j
i l
meeting in Bethesda so that any ITPR concerns for the repair pro-r
};
cess can be factored into our evaluation. No concerns were raised at the meeting that up41d delay repair initiation.
l i'
09/01/82:
A.
GPUN has initiated an catensive prmgram to:
1.
Develop and prove a method for desulfurization.
['
2.
Assess the risk if desulfurization is not performed.
L I
It is anticipated that preliminary results from this program will be available 'fkr et by the October 13, 1982 meeting.
8.
A program is bei completed which addresses the rate of I
creek gati on various starting crack sizes.
l
~
7tel nary fesults ladicate that oefects which are too ame11 fler detect b ECT (see Attachment 1. Trip Report
(
by C. V. Redd, will not propagate to failure due to 1(
Mgh casle fhti Ma anticipated operating cycles.
A subetential.
n.eafsts between the ECT level 0
of detest 4M11ty
' defect sinas which are predicted to propagate, l
e l
t l
09/02/82 : preliminary leak rete' testing of 10 tube qualification blocks j
i at Foster heeler hage shown leakages approaching the program 7
goals of less than 1 a[10 4 #
One block was found to be j
I leeking at approatmat 1y.2 x f/kr, but the welding con-1 docted on the block after assably may have affected the expsnsion
(
seal. At any rate, these leakage rats goals are approximately 3
three or6ars of magnttade below the plant Technical Specification 1
limit af lia GpM.
l A
f General Comments by Dr. D. MsCdonald have been received pertaining to BW's evaluation l
of tubing recoved from TM! ho.1 ( Attachment 2).
Dr. Macdonald's comments focus primarily on the rule of thiosulfate in the corrosion process and the necessity for desulfurization prior to restart. These coments will be pro-vided to GPUN and they will be requested to respond during the October 13, y
1342 meeting.
The 100". ECT utilizing the 0.540 probe for both steam generators has revealed j
some defects outside of tx tubesheet which are attributable to 10.
A m ber of these defects have very s. all E(.T si;rnals which indicate eitner that they
- b
i
{
a victor senareya SEP 101982 are circumferential1y small or a're not far thrwegh wall. Preliminary analysis indicates that these defects are unlikely to raptd1y propagata as a consequece of vibration. I have discussed with GPWI the possibility of leaving a
[
statistically significant number of these tuk in service. The advantage to leaving uma ef these tubes in service is thet they can be monitored i
frequently emeegh to provide rate f afbreation for corrosion propagation of this type of defhet. The startup program will include ECT's after:
1.
,% critical het fhattianals.
)
2.
Apprenfestely 30 EPPD.
3.
Apprestastely 90,addittenal EPPO.
These pertedts faspectfens will fhmetion to mootter f6r initiatten of new deftsts and with anglegged "menttergble defects
- tubes establish corrosion propsestien rete >la t estles. De the seest significant propagation is detected after progettical het fhastfees, the tubes ces then be removed from servlee. This'eessert will he discussed ta more detail with other staff l
morters 'eediesseeltests.at the October 13 meettag..
r
-, v y., -
N'opdate'siridte..the Octsher,i3 anottog'will be issued if there are
[
av sigetftsamt iteus to report.
t
.c.
9 i
Caered McCracken. Section Leader Chesiest Tectnelogy section Chemissi Engineering tranch Attachments: As stated cca POR 30-280 R. Delleer D. Eisenhet W. Johnston T. Rosak S. Lataas J. Ibijght J. Stela S. Powitkki T.SmTlivan M. Csere4 W. Nazelten C. Cheng E. Murphy TMI consultants P. the
- 5. Young W1111mia Seagreves. FRC R. Jacobs Dr. Macdonald. Ohio State P. Grant Robert Dillon, PHL b'
-(b'*, -
I.
l J. Rajan l
'e>;;.
E
. L..Frnah - - -- -- - F
-C. - Gedd, OWL--
l --
..C...NcM 'm
. t......... a,p m u j..,..,...,..
1 I
l I 9LaR2.
I F-
w.ter uo. um-h-u._
is y c. w w, a, j INTRA LABORATDRY CORRESPONDENCE OAK RIDGE NATICN AL L ABOR ATORY August 17, 1982 To:
R. W. ItClung From:
C. V. Dodd C. v. OM
Subject:
Travel to Harrisburg, Pennsylvania, August S-9, 1982 On Monday, August 8, I met with General Public Utilities (GPU) and NRC personnel to discuts the eddy-current inspection of the~Three-Mile Island Unit 1 steam generators.
A partial list of attendees is attached.
Nick Kazanas of GPU gave a presentation of the development and qualification program for the eddy-current inspection of the steam generators. A set of inside diameter calibration standards with circunferential notch lengths of 0.060, 0.100, and 0.187 in., notch depths of 20, 40, 60, and 80%, and widths of typically 0.004 to 0.005 in, were constructed.
In addition, an axial notch standard with 0.060 in. leng notches was constructed.
These standards were used to test circu=ferential differential probes of 0.510 and 0.540 in., and an eight-coil array with 0.187 in. outside diameter pancake coils.
Various gain and other conditions were run.
The best cocbination for the differential coil system was the 0.540 in.
outside diameter probe with a permanent ma gnet saturating core and e gain setting of 60.
The permanent cagnet should not have increased the ignal any but only reduce the noise a small amount.
I examined a probe a; d its field did not seem strong enough to saturate any ferro=agnetism assc:iated with Inconel (about 0.3 T is usually needed).
However, it did sees to be core carefully =ade than the regular 0.540 in. probe.
An additional =ix of the 200 and 400 kH: nixed signal with an 800 kHz signal reduced the noise due to probe chatter and pilgering.
l The 8X1 absolute probe showed sore signal to the scall defects at a gain of 53, but a fairly large lif t-off signal was also present.
The 8X1 probe array is being revised to increase the probe body from 0.520 to 0.540 in, which should reduce the lif t-of f problem.
A correlation of the defects seen by the pancake c ils with the defects l
seen with the 0.540 in. outside diameter differential probe showed that of 3233 defects detected by the pancake coil, 3216 were also detected by the differential probe.
This number was i= proved to 3229 by using the mix to reduce the inside diameter noise.
This shows an excellent catch and also showed that the '0.540 in. differential probe, operated under these conditions, can reliably detect the sace type of circumf erential crack.
~
In order to directly apply the results obtained from the electredischarge machined standards, an "eff ective axial crack width" fo'r these intergranular stress-corrosion cracks cust be dete rmined.
1
R. W. IkClung Page Two August 17, 1982 On Monday af ternoon, John Janiszewski of GPU gave a presentation of the results of the metallography of the cracks.
Some cracks appeared to,have regions of bulk intergranular attack associated with them, and so=e i
appeared to be very narrow with very little branching and axial component.
However, the crack Lould only need a few branches to effectively disrupt ~
the flow of eddy currents.
John Janis:ewski will generate an " effective axial component" by reviewing the results of the previous metallography and furnish it to us.
This number will show how applicable the calibration results from electrodischarge machined notches is for the circumferential cracks,' and furnish an independent verification of the ability to detect the defects wich a different prebe.
11e results of a dimensional analysis experiment using large scale models at ORNL will be used to correct the sensitivity at one length to the sensitivity at another.
It was stated that the region near the crack was depleted in chrome, but no estimate was made of the bulk electrical and magnetic properties of the region.
A total of 19 ft of " good" tubing has been examined by =etallography, with an additional 6 ft to be examined.
No defects were detected by this test that were =issed by the eddy-current test.
Some of the defects detected by the eddy currents were not found by the metallographic examination, probably due to the way the samples were cut. Some of the eddy-current signals turned out to be - due to manufacturing, handling, and assembly artifacts, and would not be detrimental to the service of the tube.
On ~ esday I visited the data reduction site at the Host inn near the plam.- 1 reviewed the results of the scan on tube A71-126.
This tube was pulled and a section sent to ORNL for examination. A through-wall defect was detected using a high-frequency (5 IE:) scan with a small (0.020 in.
mean radius) probe from the outside.
The defect was then etched and i
_ showed a 0.005 to 0.010-in.-wide af fected region on the outer surface.
It is not known-if this entire region appears as a low conductivity region or not.
The defect was recorded as 80% through-wall by ConAm, and a blind I
remeasurement of the defect from tape showed 84%. The magnitude of the signal was 1/2 V at a gain of 34, and an 0.510 in. outside diameter i
differential probe was used.
I also looked at runs using thE 0.540 in, outside diameter dif ferential probe with a gain of 60.
The practical noise level to get a reasonable measurement of the defect depth appears to be around 0.5 V, although in many cases smaller defects can be measured'.
Based on the 0.005 in. vide standards, this falls in the range of a 0.060 in. long defect, 40%
through-wall.
Depth measurements on smaller defects will probably be very inaccurate.
A 100% inspection of the full length of all the tubes is being perf ormed using the 0.540 in. probe with a gain of 60.
The number cf indications is about three times as many as were observed with the 0.510 in. differential probe. Tube B10-48 showed 16 inside diameter defects, all about 1 V in amplitude, between 30 and 60% of the wall.
The lowest def ect was near the third support plate.
The high sensitivity of this new inspection is also
l R. W. McClung Page Ihree August 17, 1982 picking up a'nunber of outside diameter signals from the canufacturing process, which are not detrimental to the service and can be ignored.
The tubes that exhibit ene inside dia=eter signals should be rescanned with the 8X1 pancake coil array, and the tubes with defects greater than the plugging limit plugged.
The tubes with defects below the plugging limit can be reexamined at later intervals to monitor growth of this type of defect.
There appears to be a drif t and a quality assurance problem with the probes.
This proble= doesn't' affect the accuracy of the test, but results in frequent probe changes.
The inclusion of several of these types of defects in an in-line standard (for future tests) should be considered.
The instrument gain in the field is set to give a repeatable voltage amplitude from a drilled hole flaw, rather than an absolute number.
The pancake coil array is cperated at a single frequency and much more susceptible to different types of noise than the dif f erential probe.
This array also requires cuch more equipment than the differential probe.
Zetec is workin; cn a more compact system, but no estimate of the availability of this instrumentation was given.
The results that I saw on Ibnday and Tuesday answered all the previous questions that E==ett Murphy had submitted in his letter of April 12, 1982, except'the one on safety evaluation.
Some attempt should be made to determine how large a defect would have to be before it would present a safety probler I feel that the defects that can now be reliably detected are such small : than those that would present a hazard, but have no information te back this up.
The study, done by the utility, their contractors, and the EPRI NDE Center to determine their sensitivity limits, was outstanding.
I CVD:jlb I
cc:
R. Earley, CPU j
J. H. DeVan L. Frank,. NRC J. C. Griess F. J. Hocan N. Kazanas, CPU A. L. Lotts C. McCracken, NRC/
J. Muscara, NRC P. Patriarca l
G. M. Slaughter J. H. Smith P. Wu, NRC C. V. Dodd/ File
(
/
_,D'fr7J-
_fbcny. _,_
~
TJ% ham m:,,,
.s, -, i cpna goi-m-gg[
e m i V n..,,y n,,4 7 i._, [.
E
~
WS 79/---f2A 7
...-q ', n
~ AN t 5 2 Eb3 9% V'lk-l.o..,)... G96 p) =* =
J. H.
D [5 +< 7 1 -s I C,
. ( p,:.
.s y.\\)s o...tos.
.s
..a.- 7-- 9 Re. e
, n 9gg._ g
=
=
. G a.
g. n.,...,.. ; -. m.-aa n o.m=, c.u..+
,, m. _,,,
7
. ~
Fn an u.
6..&.
1 -
Aia. c pi yg2 9y
=
e-h $h e m
if isLa
-<cu>mn nn
[.2. Swlb1.14l 1.-: #Putu' -
nn -9y2 ggua.
j
^
m.w
,.[D. An cewn c
.. - GPva
, q q a. 8. w xom
..y,,..~~~
~1 y*F-x
-.. g -..
~-f.C~J 4 Cot.rT.~-
. T@ 0 sq.
p g g _ g gj "7-k 0 h d I /* k h/
4J i- $ C3. H _g t f
h f
y a
a.
Qi.. 6
~
- y. r',*
9, -..
b e
d 98 l
Y s
6
.Q
-~
~
A 17s c*c n'k T~ ^ % L 1149 Regency Dr.
Columbus, Ohio 43220 August 30, 1982 Mr. Victor Benaroya, Chief Chemical Engineering Branch '
Division of Engineering United States Nuclear Reglatory Commission Washington, D.C.
20555
Subject:
Review of B & W Document No. 77-1135317, " Evaluation of Tube Samples from TMI-1."
Dear Mr. Benaroya,
I have now had an opportunity to review the recent report from B & W referred to above. My review and comments are given below.
In general, the report describes a thorough and comprehensive evaluation cf Inconel-600 tubes removed from the TMI-l steam generators subsequent.to detection of extensive primary-to-secondary leaks in September 1981.
- However, there are a number of features of the report that deserve comment, particularly with regard to findings which may have implicatitns for future activities by GPU and B & W.in repairing the damaged steam generators. The more important of these are as follows:
~ ge 11 - Several references have been made to " greenish" and " greenish-yell: "
deposits in areas which gave eddy current indications. However, n:
positive identification of this deposit has been made. Compounds having this characteristic color include copper oxyhydroxy chlorides, nickel hydroxide, and various sulfur compcunds such as elemental sulfur, iron disulfide (FeS, pyrite), and possibly some nickel 2
sulfides (as indicated in the B & W report).
Although electron diffraction studies were performed in an effort to identify some of the deposits observed, these efforts were only partly successful.
'I believe that efforts shuuld continue to identify the deposits, because positive identification may provide unequivocal evidence for any given failure mechanism,
(
l Page 26 - The report states the 26-3/4" location did not show any significant sulfur peak, and that other spectra, not recorded, also did not show the presence of sulfur.
However, it is difficult to s'ipport this statement from the spectrum shown in Fig. 2.6.6, because of the large background signal in the region where sulfur is evident.
It appears to me that sulfur species could very well be present, in which case there is nothing unique about these particular cases.
Page 35 - The authors have made extensive use of the sodium azide test for sulfur species. However, it should be emphasized that,although the nan 3 test detects " reduced" forms of sulfur, it is not very
. r
-r nn n, n, e ygW Jz A G V Y l
M
~
. 3 2
specific as to the form of the sulfur-containing species.
From a chemistry viewpoint, the most important characteristic of the
" reduced" forms of sulfur is that they contain elemental sulfur, which can act as a powerful oxidant, particularly when deposited at the tip of a stress corrosion crack.
In my opinion, a better method would have been to extract the zero-valent sulfur from the surface using cyclohexane followed by UV ' absorption spectrometry to determine the concentration of sulfur.
I believe that this technique, which can also be applied to bulk water samples, would provide more reliable measures of the local concentrations of aggressive sulfur.
Section 2.8.2 - The Auger depth profile-data contained within this section are difficult to understand on a quantitative basis, since the data indicate that the films (and possibly the alloy after sufficient sputtering) contain up to 10% carbon. d of sputtering For example, in Fig. 2.8.10 the oxygen signal is lost after sl400 indicating that the surface oxide film is completely removed at this point.
However, the carbon content remains constant from 4001 onwards at about 9%.
Even taking into account hydrocarbon contamin-ation, and the fact that Ar+ sputtering could result in " scrambling" of the underlying atomic layers, this figure appears to be far too high when compared to the bulk alloy carbon concentration of 0.041%.
The alogorithm used for computing elemental composition was not described in detail, and hence it is not possible to comment on the viability of the particular method used. The data do, however, demonstrate the very serious problems that can arise when using Auger Spectroscopy for quantitative or semi-quantitative surface analysis.
Similar comments may be levelled against the ESCA (or XPS) data, i.e., it is freeuently difficult to give an unambigeous interpreta-tion of the data without independent chemical evidence.
Thus, on page 55 it is stated that the hydrocarbon is doubly and singly bonded to oxygen, presumeably as ketone (R C=0), aldehyde (RCHu,,
2 or acid (RC00H) groups and alcohol (R0H) functions, respectively.
The authors also state that carbon-nitrogen bonds were detected; presumeably these indicate the existence of primary (RNH ).
2 secondary (RR'NH), or tertiary (RR'R"N) amines or amides (including substituted amides of the type RCONR'R").
It was st:ted at a meeting of the Task Group at GPU that the primary circuit may have been contaminated with an oil of some type.
If so, the oil may provide an explanation for the ESCA observation, but this has not been demonstrated by the authors.
It is also claimed in this section, and repeated elsewhere in the report, that minor portions of the sulfur peak are associated with "503-" -- I cannot find any evidence that a sulfur (+2) species of this type exists under any conditions which are relevant to TMI-1.
Page 55 - The SIMs results which show that significant amounts of hydrogen are present in the alloy, par ticularly in the region of the crack tip, are very interesting. Tne authors have been very careful not to attach toom much significance to these findings, and at no time imply that a " hydrogen embrittlement" phenomenon is involved in the failure process.
However, it has been claimed by others on numerous occasions
1_. _
that partially reduced sulfur species act to poison the hydrogen recombination reaction which occurs at the metal surface during free corrosion, therby accelerating penetration of hydrogen into the bulk alloy phase.
It is emphasized that this is an area of great controversy in corrosion science (SCC vs HE), so that it would be premature to attach too much significance to this single observation.
Both the ESCA (page 55) and the electron diffraction (page 57) studies indicate that metal sulfides (e.g. FeS and "Fe8Ni 58 16"?) are present on the metal surface in addition to various oxides (e.g., NiO, Fe3 4, 0
00H", "Fe2 Nic.6 4").
It should be recalled that
" Fe0.7Ni0 0
thermodyn.3amic agrumen.4ts cresented earlier indicated that the product of the reaction between " reduced forms of sul fur" (e.g., thiosulfate) and Ni-Cr Fe ' alloys, such as Inconel-600, would be FeS (or other iron sulfides) and possibly nickel sulfide.
The latter has not been detected, although evidence is cited for the mixed sulfide FesNigS s".
i It was also argued in an earlier report by this consultant that these' active sulfides would spontaneously oxidize in aqueous solution upon exposure to oxygen.
Accordingly, the fact that metal sulfides have been detected on surfaces that existed for a significant time,
outside of the reactor indicates that the amount of metal sulfide analyzed is in fact a lower limit.
Interestingly, one.cf the oxidation reactions which can occur may be written in the form MS + 3/2 0
> M0 + S0 2
2 in which the sulfide anic. is converted to 502 This species may the oxide surface, thereby explaining subsequently chemisorb or to$ ".Alternatively, the sulfide may oxidire the ESCA' observation of ':0 as prev ously proposed tc form polythionic acids (H Sx0s) which i
2 presumeably could also ex-lain the ESCA results.
In short, the electrooptical analyses reported by B & W lend credence to the hypothesis that polysulfur oxyanions may be regenerated by the oxidation of metal sulfides on the Incorie-600 surface.
If so, the danger of cracking :t a later date should be considered to be very real.
The extensive metallographic examinations reported by B & W clearly show a sensitized microstructure, in spite of the fact that the bulk carbon concentration is quite low for Inconel-600, particularly for that of the vintage used in TMI. The fact that Inconel-600 is much more susceptible to carbide-induced sensitization that is Type 304SS for equivalent carbon concentrations can be attributed to the much lower carbon solubility in the former compared with the latter at the same temperature.
Indeed, as a point for future consideration, sensiti-I zation-immune low carbon grades of Inconel-600 may not be practical in the sense that the low carbon grades of austenitic stainless l
steels are (.e.g., Type 304L).
l l
Page 77 - Some effort was invested in this study to determine whether or not slow crack growth would occur in the tubes under static, axial, tensile loading. The environments to which the inside surfaces of the tube were exposed included laboratory air, moist laboratory air, l
l and primary coolant.
Apparently no attempt was made to determine i
l
-.--.x
- ~
4_
whether or not crack extension would occur if the inside surfaces were exposed to thiosulfate (or polythionic acid)-contaminated environments, prepared either ex situ by the addition of the
~
contaminant to the coolant, or In situ by the oxidation of metal
. sulfides on the-met'al surface. The metal sulfides could easily have been prepared by prior exposuYe of the tube to thiospifate solutions in the absence of oxygen.
I be'lieve tF.at this would have been a crucial experiment, since it may have indicated whether or not the fears of subsequent cracking ini PWR primary systems which have not been desulfurized are well-founded.
At this point, I wish to reiterate that the B & W study represents a thorough and comprehensive analysis of the failed TMI-l SG tubes, particularly in view of the very severe time contraints under which the authors had to operate.
However,- I believe that a central issue in the whole matter of the TMI failures has not been addressed: that is, what was the precise role of thiosulfate and will the failures continue to occur if the primary circuits are put back into operation without adequate desulfurization measures being takens2;$ am sure that you will agree that these are very impor+. ant questions indee'd.
o I look forward to your response and those of other interested parties.
Yours sincerely, i
ccg g s
D.D. Macdonald DDM:ab 1.
i i
i i
l l
l I
.-