ML20147B883
| ML20147B883 | |
| Person / Time | |
|---|---|
| Site: | Zion File:ZionSolutions icon.png |
| Issue date: | 02/26/1988 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML20147B879 | List: |
| References | |
| NUDOCS 8803020234 | |
| Download: ML20147B883 (6) | |
Text
.
ENCLOSURE 1 SAFETY EVALUATION REK,*ti BY THE OFFICE OF NUCLEAR REACTOR REGULATION COMMONWEALTH EDISON COMPANY ZION UNITS 1 AND 2 DOCKET NOS. 50-295/304 MATERIAL PROPERTIES FOR FRACTURE TOUGHNESS REQUIREMENTS FOR PROTECTION AGAINST PRESSURIZED THERMAL SHOCK EVENTS
1.0 BACKGROUND
The PTS submittal for Zion 1 and 2 submitted by the Itcensee, Comonwealth Edison Co. (CECO), on January 17, 1986 was based on a report by Westinghouse Electric Corporation, WCAP-10962. The controlling reactor vessel material from the standpoint of pressurized thermal shock evaluations was identified as the circumferential beltline weld in the Zion 1 reactor vessel and the lower shell longitudinal welds in Zion 2.
Both critical welds were made with weld wire heat number 72105 and weld flux 8669 and are designated WF-70 by Babcock and Wilcox (B&W), the vessel manufacturer. Weld wire 72105 also had been used with weld flux 8773 to make a number of surveillance welds for B&W vessels including Zion 1 and 2.
All were designated WF-209-1. Having no evidence that the weld flux lot affects the copper content of welds, the staff accepts the inclusion of measurements from both WF-70 and WF-209-1 in the same data base.
To arrive at the best estimate copper and nickel contents for these vessel welds, Westinghouse averaged 87 measurements that had been reported for several weldsents made with wire 72105. All but 30 of these came from studies made by B&W and reported in BAW 1799.
The unusually large number of measurements of copper and nickel was the result of some special studies conducted by B&W to detemine throughwall variability in a nozzle dropout that contained WF-70 weld material and other studies on pieces of WF-209-1 material from the archives.
The average chemical composition given in BAW 1799 was 0.35% Cu and 0.59% N1.
However, in WCAP-10962, Westinghouse added about 30 measurements taken mainly from two surveillance reports by Southwest Research Institute (SWRI) which gave the results of X-ray fluorescence measurements on irradiated broken Charpy bars. The measurements of copper content were significantly lower than the values given in BAW 1799. The average of 87 values was 0.32% Cu and 0.56% N1. The differences in the averages seem small when one considers, that the standard deviations reported in BAW 1799 were about 0.05% Cu and, about 0.01% N1. However, the difference is crucial with regard to meeting the screening criterion at the end of licensed life for Zion 1.
In its review of the first Ceco submittal for Zion 1, dated June 24, 1986, the staff concluded that many of the 30 measurements added to the list in WCAP-10962 were not credible, and the copper and nickel contents should be 0.35% and 0.59%, respectively, as reported in BAW 1799. The finding was 8803020234 880226 PDR ADOCK 05000295 S
PDR S N._ i*
hi
l
. transmitted to Ceco by letter of August 14, 1986. At a meeting on October 3, 1986, representatives of CECO and Westinghouse presented additional data and arguments in support of their position, but their arguments were not persuasive. The CECO position was put in writing in another submittal dated which referenced WCAP-11350, a compilation of the technical December 29, 19863 infomation presented at the meeting. The staff again rejected the Ceco position in an SER transmitted to CECO on May 7, 1987. A third submittal by CECO dated September 18,1983 criticized our SER and restated the contents of their December 29, 1986 submittal in the form of attachments to the letter.
This third SER is a review of the September 18, 1987 submittal.
1.1 Review of the September 18, 1987 Submittal It is important to begin this review of the Ceco effort to determine best estimate values for the copper and nickel contents of weld WF-70 in the Zion 1 circumferential beltline weld and the Zion 2 lower shell longitudirial welds by noting that Ceco has submitted no measurements taken on the actual vessel welds. None were expected, because sampling the beltline welds in a credible way would be a major undertaking and would leave stress raisers or repair welds.
The alternative that has been followed in reviewing all other submittals is to estimate the chemical composition of the beltline welds from an analysis of data from other weldments made with the same heat of weld wire, i.e., from weldments that can be readily sampled as described below. Traceability of weld wire is by heat number - American Chain and Cable Company heat number 72105 in this case.
Most of the copper in the weld comes from the plating that was applied to improve weld cleanliness. Some details of the processing are as follows, quoting from BAW 1799:
"The weld wire vendor typically produces the plated product in the following sequence:
1.
Redraw rod is normally (0.25 inch diameter) (as) received from the material supplier.
2.
The redraw rod is subjected to batch electroplating (copper sulfate solution) after surface preparation.
3.
The rod is then redrawn to the desired wire size -- 0.125 inch in the case of the 177-FA reactor vessel welding."
Lt.r# 6.*.;
- 7. y9.-._
i.W, 'k ?
5
E 3-Ib Obviously, plating thickness variability, especially between batches.
contributes to variability in copper content between weldments. Nickel.
=-
on the other hand, is present as an alloying element, in heat 72105, and t
the varability of nickel content is very small, as discussed below.
?
Response to Comments in the Ceco Sept. 18, 1987 Submittal f
[
The approach used in both this safety enluation and the previous one for h
arriving at a best estimate of the chemistry is to first average the w
.-I measurements made on each individur.1 weldsent.
In the September 18, 1987
?:
(
Ceco submittal, an objection is raised to this procedure. Quoting from page 3
2 of Attachment 8, to that submittal:
R E
4
-f
?
'Given the basic assumption that all weldsents have the same copper L
content, it is irrelevant whether ten measurements were made on one
=
weldsent or one measurement was made on each of ten weldments."
ir.
E Table I, which is taken from the Ceco submittal, Attachment D. Table I, d
illustrates the problem. There are 13 weldments represented. The number of E-copper measurements on each weldment ranges from 1 to 26. Inspection of Table 1 1
shows that the variation of copper content within weldments is smaller than the L
variation from weldment to weldment. This is to be expected, considering the way the copper plating is applied to batches of draw rod, and batch-to-batch 1
m variation is more likely to show up between weldments. Also, in many cases the 2
8 measurements for a single weldment may be made on samples taken side by side.
7$
7 By first averaging the results for each weldment, the potential for bias F
resulting from unequal numbers of measurements on each weldment is eliminated, g.
r I
Based on these considerations, the staff considers the Zion 1 weld, for aample, to be simply another weldsent for which we must estimate the copper r
content by reviewing the averages for the 13 weldments, eliminating any that
_t ere not credible values, and averaging the remainder to obtain the best estimate for the vessel weld.
e The staff does not include in the average the measurements shcun for weldments E
5, 9 and 11, because one of the two measurements in each case was discarded by B&W in their treatment of the data in BAW 1799. Their reason was that some K
early measurements on weld metal qualification weldments were found to be r
significantly lower (0.07% Cu on average) than the results of the re-analysis.
.:r Contrary to assertions made by Ceco in their September 18, 1987 submittal,(p.
4 of attachment C) the authors of BAW 1799 state:
"These comparisons indicate that original test results on WF series welds yielded Cu and Ni concentrations of 0.07 and 0.06 wt %, respectively, lower than the actual values."
(underlines a E )
J M
m E
I b
[..% f
[
i 4
Moreover, the B&W authors applied these "deltas" in estimating copper and nickel contents for some weld metals for which only early measurements were available.
If the staff were to use that technique in evaluating the results in Table 1 for weldments 5, 9 and 11, their averages would increase by 0.035%
Cu each and the grand average would increase about 0.01% Cu. Rather than attempt to "correct" the averages given in Table 1 for weldments 5, 9 and 11, the staff has omitted them, there being a number of other weldments to provide data.
The staff does not include in the average the results for weldments 3 and 4 in Table 1 because most of the 10 measurements for weldsent 3 and 13 measurements for weldment 4 were by X-ray fluorescence (XRF) on irradiated Charpy bars.
The staff believes these measurements lack credibility, because the author of the report doubted the accuracy of his X-ray fluorescence measurements on irradiated Charpy bars, and because the nickel contents measured by this technique lacked the expected consistency. To elaborate on the latter point first:
the XRF measurements on irradiated Charpy bars given in Table 1 for weldments 3 and 4 show that the nickel contents range from 0.47% to 0.57% Ni, whereas measurements by emission spectrographic analysis ranged from 0.57% to 0.62%, with 44 of 57 values being either 0.58% or 0.59% N1. The Ni content measurements using XRF are significantly more variable and lower than the Ni content measurements using emission spectrographic analysis.
Based on the 1
wire processing techniques used to fabricate the weld wire the Ni content should not be as variable as reported in the XRF measurements.
This is an indication that there were difficulties with the X-ray fluorescence technique when applied to irradiated broken Charpy bars. The Ceco submittal, Attachment t
E, attacks this point by showing one pair of measurements where the value obtained by XRF is higher than that obtained by ICP (Inductively Coupled
=
Plasma) and 4 other pairs of measurements where there is good agreement between XRF and ATA (atomic absorption) methods. Nevertheless, the staff believes the bulk of the evidence on nickel measurements supports its conclusion that the XRF measurements on irradiated Charpy bars reported for weldments 3 and 4 lack credibility. The second reason for the staff's position is the surveillance reports by Southwest Research Institute plus private conversations with the author, the late E. B. "Woody" Norris about his XRF measurements on irradiated Charpy bars in connection with a plant other than Zion 1 and 2.
He explained that low values of Cu content could be caused by the fact that the window on his XRF equipment was so wide that about 2/3 of the Charpy broken half was exposed. Thus, if the weld specimen contained some base metal beyond the notch, which is pemitted by ASTM E-185, and if the base metal has lower copper than the weld, as would probably the case for heat 72105, the measurement would be low.
In addition, he explained that the calibration of his XRF equipment was difficult when using irradiated broken Charpy halves partly because of their small size and partly because there is gama radiation from the irradiation Charpy halves.
s Mb
.I w
.L.dh.
o
5-The Ceco submittal, attachment C, attacks the staff's safety evaluation for allegedly throwing out the averages for weldments for which there were fewer than 4 measurements. This is a misunderstanding of our safety evaluation which itsts two approaches: a grand average for 11 weldsents, yielding 0.348% copper, and an alternative result which gave a grand average considering only the 6 weldments that had 4 or more measurements, yielding 0.359% copper.
Both approaches support the staff conclusion.
The CECO submittal also criticizes the safety evaluation for omitting,a few other data points without reason. The staff accepts their criticism 'to avoid further error or lack of clarity, the averages given in Table 1 from the CECO submittal are used, with the exceptions discussed above.
1.2 Staff Position Having deleted the averages for weldments 3, 4, 5, 9 and 11 for the reasons discussed in the above response to Ceco conenents, the staff then averaged the eight values remaining to obtain 0.35% Cu and 0.581 Ni for the best-estimate values. The eight values themselves represent averages of 4 to 26 measurements, with the exception of weldsents 1 and 2 which are single values.
If only weldments 6, 7, 8, 10, 12 and 13 are included, the result is 0.36% Cu and 0.59% Nf. Stated another way, for those 6 weldr ents for which multiple credible measurements are available, the range of copper content is 0.301%
to 0.419% and only one has a copper content below 0.350%. The range of nickel content is 0.5821 to 0.605%. These results clearly justify the position taken by the staff in its safety evaluations of August 14 of 1986, and May 7, 1987, that the best-estimate values for the Zion 1 and 2 critical welds are 0.35% Cu and 0.591 Nf.
Without in any way diminishing the conclusion stated above, a broader view of the issue must be recognized. The difference between the Ceco estimate (0.32%
copper) and the staff estimate (0.35% copper) is 0.03% copper, which is small compared to the uncertainly caused by the spread in measurements. A conservative measure of uncertainty can be found in the report on B&W weld chemistries (BAW 1799). For heat 72105, one standard deviation was 0.05% Cu, In fact, Heat 72105 has the largest stanoard deviation of any of the B&W welds except for one other.
If all of the measurements reported by CECO were judged to be credible, the spread would be even greater. Because the difference between the staff and the licensee amounts to one-half of one standard deviation for the data base, the staff feels there is r.athing to be gained by further statistical arguments with the licensee nor by the third-party review requested by CECO.
YtL? M tL L d25.'.C:hid.3a.' ~ ~ % --
f C
O TABLE t
- DATA SOATED BY vetLDenINT RES100AL MAf tRLAL CMMISTRY 0ATA FOR WELO Wt At HEAF 72106 es teni es tee e.ies ese=
sua te she cessou adoes CV Ni u
meanos em as secu ese e
a use uw 1
.360 J?0 se measH em en encu ame e
a use use 2
.200 360 m
ownsi em en sus see e
s uw we a
w mi em se o==
ses v
s ans a se er ame e
a o ne use se.su i em se w am.Di.
e see T
s 4 see sees sa em se sem mee v
e a me snee 3
251 S42 sits se m.e. s la sa sa w sai ens se meu nas e
a aus em sa me -sat i s?fs se sens ses e
g am q ue sa maan ons en one see v
s use saae n
w son sus en name o
e v
arm s ue se as asu erra se maae o
v v
asse saes ens se see.
ses v
6 s tas een e.s. pos i u
mi ens se e==
see e
s e sse een a
e mi em se esse aan v
s saw use a
se e mi ens se seem see v
e o ne s ame a
pe ros i em en s==
ame y
a asw aus 4
.241
.524 a
se poi s em se s==
see v
s 484 t aas een e
e e sse am m.
n u m, em em em u ne e
em e ve u
m, ens se w
w mi em se swa see e
a ssm asm sa w mi ens en som see v
s ass asse so w sui em se saae ata v
v ano asm n
= imi ens se eaae ata e
v esw eem a
w son om se ease ata v
v use use sa se seu ens sese een ata e
v use asse 5
.350
.535 ao m aan ene enson see esa e
e use use 8 8'e E
s.
see sta 5
5 345 e.s see.
es sua
- e. m i sm e as esa
=
am u
eu e
e use s ese 8
.355
.542 m
em es
%.c w m.i wn ens so we su a
e ame sem ee.
osa e
em use i.:
w mi sm en.
m mi em e
eso esa
=
e e se e9 l
4 we ass i u n's se see sea e
a E34 Ess ena e
use esw f
.362 m
I.;
ee ma e
e a2 =
s een d631 ee.n
= su em se x.c
~
w re i am se
.. m i em os one op e
e use een l
w amo sm se see uma e
a um s ans
.c
= seu one en see una e
e use e sas ac w sei em me een ma e
e uw use
= un i am en saa ma e
e um e see esa e
e e sse e sse
.c w em i em to een.
esa e
s e sse s aae w mi em se
=
w nei ens en see esa e
e um s een
= un om en eu e
e uw am
.ac
.se.e m sa i em se osa a
e uw e sas su a
e am s ee ac w smi sus se m.e.
e ana e
e anw sem
=
m ani em en o
w mi ens se me sea e
e s ese e ens
.c c
w me ens so we eu a
a use ame Hec w ami ens os ese eu e
e use s ees 8
.301
.542 w asu ens as ese ena a
e em ses w mn ens se see saa e
o amo e ase
.c w sei ens en see un a
e a sie s een m smi ens en ese on a
e use s ese w mu em en see ma i
s we s een
.ac c
w mi em en see sa e
e un asse
.c w iei em os ese ma e
e e sse um ac se mi em se es e ma e
use asse ac w mi em se new ou e
e o ne asse w seu ens se ese saa e
asse anse
.ac w MEi ans se see ma e
9 eJee t ess
.c m'
em as ans naa e
a a.mst e ase e m' '3 eD.a a se l 5
Fe eBe b
5 am T4 sr
- M :
du w w.
=
we es i
em e ue m
em en een ma a
e use e ase l
- o. m.
u spD
- sm Da hast maa y
a am a.asa I is
- x
- re i s,
w use una
=
a a mo eni ee es se ese meno een ata e
v we
- Jee y
g es m se sem umson see ena e
a use use j
i o.
w we eine see see esa e
e e.se s ese me em en see me e
e sem e ss l
anse
=eo een esa a
s s ame asse e.s sc
. se anse
=eo oss ma e
e asse saa u
eo ese esa e
a em anse o
w is em n.e see saa a
e une a ese u
e se eene w
== ne mee ee een ena e
e asse ame
.s w se mee see ese ana e
a me e ase 10
.419
.583 o
e..se asse neo een ma a
e am a sse w
as se ese mee ese saa e
e tem sans
=
mee mee ese ma e
e eam use w
.em. se asse mee ene een a
e ese as4 we es a men mee see una e
e asse use sum mee see ma e
a een ese
==.se w
es ie s=,
e,ee see ma a
e a ese use m
es
== n-asse sese one ata a
v we ame n
M
.000 i
,, ns som wane one ma e
um w.
l e
u,n
. m.
i
.x, 30
.no ma e
a.s um is
.30 m
.es s e.ie ma u.n
.m.
ae es me 1
Reprinted from Attachment D to the Ceco sutaittal of Sept.18s 1987.
.ey_...
.. _ _.,