ML20155B809
| ML20155B809 | |
| Person / Time | |
|---|---|
| Issue date: | 10/22/1998 |
| From: | Wen P NRC (Affiliation Not Assigned) |
| To: | Essig T NRC (Affiliation Not Assigned) |
| References | |
| NUDOCS 9810300300 | |
| Download: ML20155B809 (46) | |
Text
. _ _
f =Ec u
y%
4 UNITED STATES g
j NUCLEAR REGULATORY COMMISSION t
WASHINGTON, D.C. 20666-0001
/
+ ***
October 22, 1998 MEMORANDUM TO: Thomas H. Essig, Acting Chief Generic issues and Environmental Projects Branch Division of Reactor Program Management Office of Nuclear Reactor Regulation FROM:
Peter C. Wen, Project Mana-eM C.
W Generic Issues and Environme J Projects Branch Division of Reactor Program Management Office of Nuclear Reactor Regulation
SUBJECT:
SUMMARY
OF OCTOBER 5,1998, MEETING WITH WESTINGHOUSE OWNERS GROUP (WOG) REGARDING PART-LENGTH CONTROL ROD DRIVE MECHANISM (CRDM) HOUSING ISSUES On October 5,1998, a public meeting was held at the U.S. Nuclear Regulatory Commission's (NRC's) offices in R]ckville, Maryland, between members of the WOG, Westinghouse, and NRC staff. Attachment i lists attendees at the meeting and Attachment 2 contains a copy of the material presented at the meeting.
This meeting followed-up on a previous NRC/WOG meeting held on June 11,1998. Since then, the NRC evaluated WOG's proposed resolution of the subject issues and sent a letter to WOG on August 11,1998. In that letter, the staff raised concerns regarding the WOG's proposeo sarr.p:ing approach to the part-length CRDM weld inspections. The purpose of the October 5,1998 meeting was to continue discussion of WOG's activities on the issues and to address the concern raised by the NRC letter of August 11,1998.
The WOG representative started the meeting by presenting an overview of the WOG regulatory response group (RRG) activities on the part-length CRDM housing issues and the status of the WOG UJbgroup program and actions performed since the June 11,1998, meeting. Since the part-length CRDM housing leak event occurred at Prairie Island Unit 2 on January 23,1998, WOG has inspected 144 welds on part-length CHOMs; these include 102 of Type 309 stainless steel (309-SS) welds and 42 of alloy-82 welds. Other than the weld that failed at Prairie Island Unit 2, no flaw was identified during these inspections.
The WOG also presented the results from its review of fabrication records. The WOG indicated that its review was concentrated on the heat numbers, welders and inspectors involved with the flawed part-length CRDM housing weld installed in Prairie Island Unit 2, and identified that Welder 4 and Inspectors 6 and 2 were involved in manufacturing of the flawed weld. On the basis of this record review and weld inspections performed so far, the WOG concluded that the Prairie Island flaw appears to be isolated. The staff stated that it did not agree that sufficient I
information was presented to reach a conclusion that the failure at Prairie Island Unit 2 was an
,t isolated event rather than a breakdown in the quality assurance program at the vendor's shop, b U 4
9810300300 981022 0
3' PDR TOPRP ENVWEST 464 C
PDR y c s, n.. %, u k o
5 2
E10 fiM @IdSSi$ ON
g.
T. Essig October 22, 1998 The staff asked what fraction of the welds associated with the Welder 4 and Inspectors 2 and 6 had been included in the WOG inspections. The information sought by the staff was not readily available during the meeting, but a WOG representative stated that the information is available and will be included in the formal WOG's response.
Lastly, a Westinghouse representative presented the statistical and risk evaluations based on the revised number of weld inspections. His conclusions on the statistical evaluation are as follows:
Target level:
"309-SS"
" Alloy-82" Sub-Lot (N=216)
Sub-Lot (N=120)
No more than Max. # of
- of
- of welds Max. # of
- of
- of welds defective samples already defective samples already defective welds in the required to inspected welds in the required to inspected welds in the uninspected achieve by WOG uninspected achieve by WOG uninspected welds the target welds the target welds level level 0.5 %
1 216 102 0
120 42 2.5 %
5 42 102 3
34 42 5.0 %
10 0
102 6
0 42 Note: N = Number of uninspected welds The preceding table is based on a confidence level of 95% and a " single sample" plan with a "zero acceptance" criterion (i.e., if WOG found one sample with defective weld, it would have to inspect all of the samples.) It is also based on two assumptions: (1) the probability of detection =
0.9 and (2) the binomial parameter (the probability of a single weld having a defect) = 0.0147 for "309-SS" and = 0.0153 for " Alloy-82," respectively. The table indicates that the WOG's sampling to date is sufficient to demonstrate with 95% confidence that no more than 2.5% to 5% of the remaining uninspected population would contain defective welds.
During the meeting, the staff pointed out that the WOG's handout contained an error in page 19 under the title of" MINIMUM SAMPLE SIZE REQUIRED IS A FUNCTION OF;"(see Attachment 2). The word " minimum"in the first bullet should be corrected with the word of" maximum." The WOG representative acknowledged this typo, but stated that the WOG's conclusion was actually based on the word " maximum."
The WOG concludes that the target level of no more than 2.5% to 5% defective welds in the uninspected welds does not constitute a significant threat to plant safety goals. This conclusion is derived from WOG's following risk assessment results:
y-T. Essig
-3 October 22, 1998 r
Target Level:
"309-SS" Sub-Lot
" Alloy-82" Sub-Lot No more than incremental incremental incremental incremental l
CDF for all CDF per CDF for all CDF per defective '
plants plant plants plant welds in the uninspected welds 0.5 %
5.0 %
3E-5 1 E-6 2E-5 2E-6 The risk assessment indicates that the expected incremental core-damage frequency (CDF) for 5% defective welds or less meets the guideline provided in RG1.174, "An Approach for Using Probabilistic Risk Assessment in Risk-Informed Decision on Plant-Specific Changes to the Current Licensing Basis". The staff made the following comments regarding the risk evaluations during the meeting:
The WOG estimated the initiating event frequency (i.e., the frequency of a load sufficient to cause the flaw to rupture) to be between 1E-5 to 1E-3 events / reactor year. A uniform distribution on load frequency (i.e., any value between these two numbers is equally likely)is assumed. This has the effect of using a mean value of SE-4 svents/ reactor year.
The staff suggested that the WOG refine its estimation of these values on an individual plant basis since the initiating event frequency has a great impact Sn the risk evaluation outcome.
The current PRA medium LOCA frequency is on the order of 1E-3/ reactor year. The staff suggested that a comparison could be made between the current PRA madium LOCA frequency and the frequency of a medium LOCA event postulated from the failure of a part-length CRDM housing weld.
On the basis of the preceding technical, statistical, and risk assessment, and the deterministic evaluation discussed at the June 11,1998, meeting, the WOG representative concluded that
- inspections completed to date are sufficient to resolve the part-length CRDM housing issue. The WOG is in the process of writing a formal response to NRC's August 11,1998, letter; it will be transmitted by October 15,1998. The staff replied that it will review the WOG's formal response and will brief senior NRR management.
Attachments: As stated cc w/atts: See next page
e e
cc w/atts: See next page DISTRIBUTION: See attached page T. Essig On the basis of the preceding technical, statistical, and risk assessment, and the deterministic evaluation discussed at the June 11,1998, meeting, the WOG representative concluded that inspections completed to date are sufficient to resolve the part-length CRDM housing issue.
WOG is in the process of writing a formal response to NRC's August 11,1998, letter; it will be transmitted by October 15,1998. The staff replied that it will review the WOG's formal response and will brief senior NRR management.
Attachments: As stated j
cc w/atts: See next page Document Name: g:\\pxq\\msum1005.wog
- See Previous Concurrence OFFICE PM:PGEB PRAB:RES SPSB:DSSA NAME PWen*:sw LAbramson*
SLong*
DATE 10/14 /98 10/19/98 10/21 /98
/
1
/
E,MfB
,(A}BhMfB SC:PGEh OFFICE g,
NAME NYrb TM k ['FAkstUiewlcz DATE 10Q//98 10/D//98 10/1p/98 OFFICIAL OFFICE COPY 1
l l
1 l
l
_ - - _ _. _. _. = _,.
's/
l l
Distribution: Mtg. Summary w/ WOG Re Part Length CRDM issues Dated October 22, 1998 Hard Cooy o
Docket File PUBLIC PGEB R/F OGC l
ACRS PWen Frail l
SCollins/FMiraglia BSheron RZimmerman BBoger JRoe DMatthews TEssig FAkstulewicz
- GHolahan RBarrett
- AEl-Bassioni SLong i
GLainas
- JStrosnider EJSullivan RHermann DHood LAbramson GTracey, EDO l
l 1
(
i i-i t.-
f^
l
\\
4
=
l l
NRC/WOG MEETING ON PART-LENGTH CRDM HOUSING ISSUES LIST OF ATTENDEES October 5,1998 NAME ORGANIZATION Gary Holahan NRR/DSSA Adel El-Bassioni NRR/DSSA/SPSB Steve Long NRR/DSSA/SPSB Jack Strosnider NRR/DE Ted Sullivan NRR/DE/EMCB Robert Hermann NRR/DE/EMCB Darl Hood NRR/DRPE/PDI-1 Peter Wen NRR/DRPM/PGEB Lee Abramson RES/ DST /PRAB Ted Huminski CP&L Roger Newton WEPCO R. Scott Boggs FP&L Dick Labott PSE&G Phil Lindberg WPSC Karl Jacobs NY Power Authority David Cambell Westinghouse Nancy Closky Westinghouse Warren Bamford Westinghouse Robert Perdue
. Westinghouse David Boyle Westinghouse l
-r
i i
l WOGSG ON PLCRDM i
i I
NRC MEETING i
i October 5,1998 4
1 i
1 1
I 1
oNoga\\smt\\smiletters\\ WOO PIIRDM Subgroup 10/198 i
AGENDA i
INTRODUCTION MEETING PURPOSE RRG PERSPECTIVE RESPONSE TO QUESTIONS RAISED BY 8/11/98 NRC LETTER UPDATE OF WOGSG EFFORTS SINCE LAST MEETING (6/11/98)
- Latest inspection Results
- Statistical and Risk Assessment OVERALL CONCLUSIONS NRC AND WOGSG DISCUSSION oNogs\\smfsmtletters\\ WOO PIIRDM Subgroup 10/5718 M'-)
i i
i INTRODUCTION I
i j
e Since the leak at Prairie Island Unit 2, utilities have addressed the issue with a high level of concern.
1
{
- Failure analysis and condition assessment.
j
- Inspections exceeding Section XI Code inspection criteria.
l
- Manufacturing records including material, weiders, and inspectors.
I l
i 3
eNoss\\smfamstetters\\W00 PlfRDM Subgroup 10/5,98
INTRODUCTION Since the June 11 meeting, the risk assessment has been updated to include all the inspections and records data.
The results of these activities were consistent and the final report meets the key principles of risk-informed regulation.
Today's presentations are consistent with the final report.
4 c:\\ logs \\smt\\smiletter \\ WOO PifRDat Subpoup 10/3/98 A
i
=
a l
INTRODUCTION l
Formal response to the August 11 letter is in process and will be transmitted by October 15.
The results of these evaluations lead us to conclude that no j
additional inspections are necessary.
l The WOGSG has arranged to have the technical people present today to address any remaining concerns.
i 5
o%gstamt\\smiletters\\ WOO PLCRDM Subgroup 10/3r>3
MEETING PURPOSE Update of WOGSG program results and activities.
Discussion of updated statistical and risk evaluation for PLCRDM housing issue.
Response to NRC questions and concerns.
Resolution of the issue.
6 oNogs\\smosmtletters\\W00 PLCRDM Subgroup 10/3nt I
I i
WOG IRG/RRG PAST ACTIVITIES
- 1. Developed an integrated plan for resolution of and responses to generic NRC questions.
- 2. (Feb.-March) Maintained NRC/ Industry interface discussions.
l
- 3. RRG closed issue; transitioned to Subgroup (April 7,1998).
- 4. RRG issued recommendations to utilities.
i i
7 of. logs \\smt\\smiletters\\ WOO PLCRDM Subgroup 10/5/98
WOG RRG RECOMMENDATIONS FOR VOLUNTARY ACTIONS For operating plants
- Docket plan to address issue within 30 days of receipt of letter For plants in refueling / cold shutdown
- Docket plan to address issue within 30 days or prior to startup Compensatory actions until resolution of issue
- Increase RCS leakage monitoring awareness
- Include this in their individual response letter 8
o%gs\\smesmticticru\\WOG PlfRDM Subgroup 10/3/98
.?
i i
l RRG IDENTIFIED OPTIONS FOR RESOLUTION TO WOG MEMBERS WITH PLCRDMs 1
Remove the housing and cap the reactor head penetrations t
Perform non-destructive examinations to confirm the absence of i
l cracking 4
Perform additional records search to better identify applicability and obtain other data to confirm the absence of any cracking l
Address the capability of using RCS leakage monitoring i
awareness while the issue is being resolved Form a subgroup to develop a potentialintegrated plan for resolution 9
oNogs\\smfamiletten\\ WOO PLCRDM Subgroup 10/3/98 i
__._____._____.._.__..______________._.,-,e._
.ea.
jew.a e..umm._,e
--(..
E.4 e M.u 4a4->=34.h4244M4 6 4,e4.M_4w 4
_aAma_*,-4__4a4,4-
.-m_--dJm_5ea&s4=m-m. jaw.a,4 w Anh a maM s4.*44M-MT
-" h O
4 e
f S
si 8
E B::
5 o
O CnO ITl m
TOD
-1 i
00 s
)
o i
WOGSG PLCRDM PROGRAM
- ELEMENTS 1.
Statistical Evaluation i
2.
Quantitative Risk Evaluation i
3.
Completion of Fabrication Records Review i
i j
4.
Inspection Data Base of Part Length CRDM i
i j
5.
Inspection Process Guidelines
]
6.
Support utility metallographic investigations 7.
White paper report - Technical assessment of PLCRDM i
Issue 11 1
o:\\ logs \\smt\\smtktters\\WOG PLCRDM Subgroup 10/3/98
WOGSG RESPONSE TO QUESTIONS RAISED BY NRC LETTER OF 8/11/98 Section XI Margins issues at Prairie Island Unit 2
- We agree with the comments raised Statistical and Risk Evaluation
- Comments are based on previous evaluations
- The evaluation has been revised to address the concerns raised on this subject Responses will be included in WOGSG letter 12 o:\\ logs \\smt'emtkuer \\ WOO Pl.CRDM Subgroup 10/57)1
PLANT WELD INSPECTION DATA TYPE 309 SS ONLY j
PLANTWELD INSPECTION DATA TYPE 309 SS ONLY PLANT NUMBER OF WELDS NO.OF ASSEMBLIES INSPECTED INDICATIONS Prairie Island 2 (partial)*
5 10 1
i Diablo Canyon 2 7
14 0
BeaverValley 1 5
10 0
North Anna 2 5
10 0
D.C. Cook 2 9
18 (spare incl.)
0 D.C. Cook 1 8
16 0
Indian Point 2 **
8 16 0
i Point Beach 1 **
4 8
0 TOTAL 51 102 1
Inspections in hot cell facilities at Westinghouse.
" Inspections in hot cell facilities at Westinghouse, August 1998 13 obgs\\smtismtletteru\\ WOO PLCRDM Subgroup 10/3/93
i 4
i l
PLANT WELD INSPECTION DATA i
TYPE ALLOY 82 ONLY PLANTWELD INSPECTION DATA TYPE ALLOY 82 ONLY*
PLANT NUMBER OF WELDS NO.OF ASSEMBLIES INSPECTED INDICATIONS BeaverValley 2 4
8 0
Watts Bar 2 8
16 0
McGuire1 8
16 0
Sequoyah 1 1
2(spare) 0 i
TOTAL 21 42 0
Totalinspected as of August 1998.
14 o%gs\\smt\\smtletter \\ WOO P1fRDM Subenmp 10/198
,_,E
-- m--
. -m
l CONCLUSIONS FROM FABRICATION RECORDS REVIEW l
Relationship between heat numbers and manufacturing dates is l
random Welder number 4 was the most active welder (35%)
Inspectors 6 and 2 were most active (31% between them)
Welder 4 and Inspectors 6 and 2 combined on more than 30%
of all buttering welds l
Inspections' have concentrated on the heat & welder & inspector (Welder 4, inspector 6 and 2) involved with the Prairie Island l
crack Results to date support the conclusion that the Prairie Island i
Unit 2 flaw was an isolated event 15 o%gs\\s:nt\\smtlettm\\ WOO PIERDM Subgroup 10/3/98 1.
Sta":istical & Ris< Evalua1: ion 16 o$ togs \\smtismiletters\\ WOO PLCRDM Subgroup 10/5/93
OBJECTIVES OF THE STATISTICAL & RISK EVALUATIONS Statistical Evaluation: Determine a sample size that is appropriate in the sense of demonstrating, with an acceptable level of confidence, that the number of defects left in the population is no more than a specified acceptable number.
Risk Evaluation: Determine t7e " acceptable number" of defectives.
obgsismt\\smiletters\\WOG PLCRDM Subgroup 10/5/93
I i
TEST PLAN & SUB-LOT DEFINITIONS i
l Test Plan:
- The acceptable level of confidence is 95%.
l
- A " single sample" plan with "zero acceptance" cnterion.
i The relevant sub lots are:
- Alloy 82 we.ds ("82" sub-lot).
- Type 309/308/403 welds ("309" sub-lot).
4 I
18 obgsismthmtletten\\ WOO PLCRDM Subgroup 10/5718
MINIMUM SAMPLE SIZE REQUIRED IS A FUNCTION OF:
maxwawv Tle-minimum number of defects in the population that will be tolerated.
The evic ence gat 7ered from past inspections (and elsewhere) that can be used in evaluating the test plan:
- No Evidence
- WOG Only inspection Evidence
- WOG + B&W Inspection Evidence i
19 obgs\\smt\\smtletter \\ WOO PIIRDM Subgroup IW3r>t
"NO EVIDENCE" SAMPLE SIZE CALCULATIONS Let H(x, n, /N, N) represent the cumulative hypergeometric distribution for the probability of getting x or fewer defects in a sample of size n for a population of size N with fraction f defectives. Setting x = 0 and noting that H(0, n, /N, N) equals h(0, n, /N, N), where the latter is the probability of getting exactly zero defects, the probability of accepting a lot with a single sample of size n in which the target fraction defective is f* is (1) h(0, n, /*N, N) = B Where B = 1-the desired confidence = " consumer risk" =5%
Solve (1) for sample size n given alternative target fraction defective /* to get the sample sizes in the "None" rows of Tables A and B.
20 e%gs%mfamiletteWWOO Pl CRDhl Subpoup 10/3/98
X Mkruses Excel - Appendix C les CN-RRA-038-98rev.xt: '
~
~
Egx b.,@diEN.MM.INi/55NINAMN_d..~@-.@MO.YSN...~[.fNNMY$kNcDNN41'AM/II[MhdM83N8 Ud p.
~ ~.
.m
.~._
n..._
I s
A Kffiils Alli!LyJ)WAIMIFfEHiil5l#RIKIMITilli5lifdilhi5ll 1
C31 i,vj l =HYPGEOMDIST(0.031 A31.5C511)
==u* A > Cly B a.e l W C t-d l S chWa vt D v kMW MFAWAMs-D-mm u.;l w F,u ld G.* j p= Hh u le.
- l. :J
- l - Ke
, - l.,
-~4 i
'.4r
{
l
- Y
}
((
= smM4 5% aAtquation gj.g0 ng N] = Bet.a 45' I'i4+ad3NM calc @ ions f.9 scMas!<I
_._I _.
_. g t
4'T 4 fN.is theIatget X defective i
i_
- _... _ a _
B+ta is i+5s thaa 5%
r8' *"s i
l
.V
~].
i
[
i
- S*
Nis the population 1
Table C.
l
~
~
~
.._._..i.._....:_.__..,I._._
._..? i
+11
~103 Velds '
N:
_ 2._1_S i....
I i - ~ ~[ ~~~
'~
~ ~i[
~~'
nes samv.e sue gg 7,,,,, x g,,,
Defectsin N Defective
(< 5%)
{
g
'/k pa 6 % hel t
t
. _,y 43 IC9trfgt $y_npi.She Bold d1 Comments
$14 0.5% goal i
i 6
(15 1
05%
04C%
2I5 Essentlang the solelot[216).
I^
T I
U~~
~ lh.
I 16' 1
0.5%
2.78 %
210
[
l y
b_...._.
h
!',I E
J f- - d
[
- U I 0*d.90S
-i i
- 20 1.5% goal l
==r 2f' 3
1.4 %
4.96x 13S
}
j
y
,-.-7-.-_....
'23-2n goaa 1
d.
h._
'h.
I0%
Ib%
_t, 24t 4
1.9%
482%
114
_. 9 l_
4 SX 5.01%
H3
_, _,,, _ j (27 5
2.3x '
488%
37
~~ I
{~~
~"~~ ~ ' I --
~~~~
~gl
~
~
l 128-5 23%
510%
SG l~ ~ '
__]((~ ~
[~] _].j
- 23-5% n.
T30 11 5.1 x I 47e<
51 31' 11 5.f *4 l 5 10% H 50
___ Q,
<32+
l t
r, Nl N Ifil ~sh S ample Sizes / phatYjfgfigh$pggjggyggytjyffffpggjg$fj 7lMnif#NM MM!9ESRJ 4f;WW/?WX4 p')
MeTgdM$jMgMllM))MI)[Qg@@)dll)M))}[i$dgiS40%{rgglifM@tJgd{ifsij{dggQ 21 0:Jegsinnfamtlettern\\ WOO PLCRDM Subgroup 10/5M
l 1
l i
I INCORPORATING EXPE'llENCE: AN OVERVIEW i
i i
Estimate the value of"phat"= probability of a single l
l weld having a defect using alternative experience j
bases:
- plat for WOG Only experience.
l
- phat for WOG + B&W experience.
i h
Encode each phat into the Perdue-Abramson Model i
and solve for the sample size that just yields a confidence of 95% (= consumer risk of 5%) for Target
% Defective.
i l
22 c:\\lossimmt\\smiletters\\'VG') PIIRDM Subgroup 10/3/98
' ~
t i
EVIDENCE USED l
1 defect in 68 welds sampled for WOG "309" welds.
1 defect in 68 + 97 = 165 WOG "309" + B&W Type C-i l
"B&WOG Activities in Response to the Prairie Island Unit 2 Partial Length CRDM Motor Tube Leakage," received April 24,1998, page 1).
O defects in 8 WOG "82" welds.
O defects in 8 WOG and ' 8 B&W Type A/B welds.
23 oNoss\\smosmtictiers\\ WOO PLCRDM Sabgroup 10/5/98 i
i,.e, 1m..-.
,3,-
,,..,w,
ESTIMATION OF phat Given a random sample of size n with number of defects x, the standard
" maximum likelihood" estimator of p = the binomial parameter representing the probability of a defect in any one weld is (2) phat = x / n with standard error =sqrt[phat*(1-phat)/n].
The estimator phat is the expected value of the binomial parameter p and phat*N = expected value or mean of the binomial distribution on the number of defects in a lot of size N.
24 obgs\\smt\\smiletters\\ WOO FLCRDU Subgroup 10/3/98
i i
)
l USING "309" DATA TO OVERCOME SKIMPY i
EXPERIENCE ON "82" WELDS Table A-2: Bayes' Theorem Calculations for "82" Binomial Par.ameter SVOG only evidence on "309" as prior)
I l
(1)
(2)
(3)
(4)
(5)
(6)
Percentile phat Prior Likelihood Priorx Posterior Probability
@i""18)
Likelihood Probability 05th 0.0000 0.185 1.000 0.185 0.210 50th 0.0147 0.630 0.888 0.559 0.636 95th 0.0387 0.185 0.729 0.135 0.154 25 obgs\\smthmtletters\\W X) PIERDM Subgroup 10/3/98
ESTIMATES OF phat t'< BINOMIAL PARAMETER;> FOR ALTERNATIVE EVIDENCE DtaUed Estimteof BinnialPaander "307'(WXiaiyevidmce) 0.0147 "307'(WXi+B&Wevidmce) 0.0061 "82"(WXiaivevidmce) 0.0153 "8Z'(WX3+B&Wevidence) 0.0062 26 i
e%gi\\smtemtletter \\W00 PIIRDM Subyoup 10/5/93
EXAMPLE: PERDUE-ABRAMSON MODEL OF CONSUMER RISK X Af# cross # E mcel - 309 WOG xis
~
- .9 E gEdd M EdligstweT@ iridestRFamat WIoots it]2ata WWindb4$ tie 4($yp35)Ps;#Hf(#3t/p$h.WW/WM#tlWWM M H i t ' &Kl#_.125.]
h lidhlNlINlblMlkNl NilkP$l*MlhlhMlNlfNldl[OslhNl]75*
3,,,,j[Ql Wl; yk}
k%M{ g!,y b;g@
~alio g@2irsjhreniyanstlWinsiqgsqwmydh%
T20 F*]
l =1-T21
[.~-~'C--
e
& 4:U e 4yIw + I4 *
- m m < Lu
'f*>
T i
s
...A~
u.u p&
4 je ge mq j
y m
+2 jUser bput
['
<3 Ptwa
'309" VOO
[jh
-N m
s.
e, <oo, e 2-
,k, 1 -Q 4 5 ': fAsnber of veidsI 216
- wst be > = 4 for doutd, s ample, plan wkh 2 weldsesanW Pa I
1.4 7E-C2 I
(t h}{ ~
df 1.8.i h
$ (,.
fO a7:
s=rodas.we of De4+ction D.S tske 0 4. POO < e 0 gg h
- s-Cond Prob.of Leak ivsfweM t.00E.00 l
)
.s-sge Svvvie Sue 42 Mak a s ample size <
- Number of VeWs* Si 4 10 y
1 l
M, 1o Tuget teat rme s esweid a 50c o2 4
t v
3 Tuget teek ene
,t ot 5 40E-00
[Calculatedl ooutde sampains ans For 1 e. 2.esos in eaoh sart,ie. Aee pt e O tw cum Reieet a. 2. Poo. ces c7
~
j* -
'3
<ss se aw sampana Pian We.ept 0.neiece a.1. Assumes POD.100X or een CF as identdied t
.F*
ir b
<<t 285 i
L_. _.
I i
k,, -[.^.: w u]
f [
f32 ne..a s s.--.r,
! 30s von I
i
-J an
~
ese o
in 1
i.
s>
1 g
'tn j
j M2 Doubie oo Aie sampie
?? - N_ + > * *' '
(/
k Sample Plan Plan (Ea6
$bgle Sample
(
h 1,td
$D Pre-isitte No (Each Surve Sunple S
! Single Sa. pie Plan (POO.
,as' isnt sire. it 21
! =santPOn.it ces c71
\\
y 20 Consumer Rest (psoh. leak eate/yllot > ta' Set]
10.100%
10.103X 10 Otl4X 4472%
ll 4 97RM j
-]
A Confidenoe [ prob. leak tmedysilot e ta' Set [
89 8924
$$$97X 89.916x 35628%
95 074x
+22 e Mean teaklyelot l 3.18E.00 3.17E.00 3.17E.00 2.5&E.00 2 62E.00
- F3 Variance [t eakfyrtlot) 3.13E.00 3.13E.00 313E.00 252E.00 2 58E_-00
'24 -- Prob. of Sampang 100% of Lot 1.75E-04 8 61E-04 4 63E-01 4 28E41
- d Sum o# BinomialProb.
1.00000
+2Se t
j, 4"30s* VOG I
1 -J am i
8 12 7 Consernes Hisk Table l
D f
C l
D if I
t-i i
T I
sklilii ria5% defective sire ik2.5% defective size -1 M4 defective'sirey:/agsfjjMiTbEWWfm*NMr9*aNW4MP////c.t' 6
J2 a
i e-
.9.
. bhb N
/
J 27 c:\\ logs \\smt\\smtletteru\\WOG PifRDM Subgroup 10/3/98
TABLE A REQUIRED SAMPLE SIZES FOR "309" SUB-LOT FOR ALTERNATIVE TARGET % DEFECTIVES (N=216)
Target % Defective Evidence 0.5%
1.0%
1.5%
2.0%
2.5%
5.0%
Assumed:
None 216 168 136 114 97 51 WOG only 216 178 137 91 42 0
WOG+
176 91 0
0 0
0 B&W l
28 e:\\l<,gs\\smeandicuern\\ WOO PifRDkl Subgroup 10/5S8
e l
TABLE B REQUIRED SAMPLE SIZES FOR "82" SUB-LOT FOR ALTERNATIVE TARGET % DEFECTIVES (N=120?
Target % Defective 1
i Evidence 0.5%
1.0%
1.5%
2.0%
2.5%
5.0%
Assumed:
None 120 114 i
114 93 76 47 WOG 120 108 108 74 34 0
only WOG+
120 70 70 0
0 0
B&W i
29 o$ logs \\smeandleucts\\ WOO PlfRDM Subgroup 10/5S8
~ ~.
=.
l
SUMMARY
OF STATISTICAL RESULTS An additional 34 welds have been inspected in each sub-lot (total of 68) with no defectives found.
Using "WOG-Only" evidence and Tables A & B, this is:
- sufficient to demonstrate no more than 2.5% (3 welds) defective in the "82" sub-lot.
" roughly" sufficient to demonstrate no more than 2.5% (6 welds) defective in the "309" sub-lot (a sample of 42 was required).
eMogs\\swt\\sndlcuen\\ WOO PIERDkt Subgroup 10/3/93 e
7 o
i i
l RISK ASSESSMENT i
I i
The objective is to calculate the incremental core
[
damage risk for postulated alternative numbers of l
flawed welds corresponding to the Target %
Defectives in the statistical evaluation.
i
[
Comparing the corresponding incremental core damage risk per plant to a key criteria of DG-1061 (RG-1.174) criterion (" proposed increase in risk is l'
estimated to be small") will then support a ris < -
i informed statement about which target is " safe i
enough".
^
]
31 o:\\lossismosmiletters\\ WOO PLCRDM Subg,,* y 10/5S8
OVERVIEW OF THE RISK ASSESSMENT The first step is to calculate the probability distribution on the conditional CDF given the existence of a flawed weld.
The mean of this distribu n is used as the conditional probability that a postulated flaw would propagate to core damage.
Given this Conditional Probability of Core Damage and an array of postulated numbers of flaus corresponding to the percentages of Target Defectives, the binomial distribution can be used to calculate the probability of 1 or more core damage events per year for the sub-lot for each of the postulated number of sub-lot flaws.
The implied CDF risk per plant is then compared to the the DG-1061 (RG-1.174) criterion (where "small enough" is interpreted as approximately 1E-6 events / year) to provide a quantitative assessment of the acceptable Target % Defective in terms ofincremental CDF risk.
32 o%gs\\smt\\smtlettm\\ WOO PLERDM 5uhgroup 10/5/98
CALCULATION OF THE CONDITIONAL CDF RISK The " Likelihood of a load sufficient to cause the flaw to rupture" is estimated to be between 1E-5 to 1E-3 events per year. Assume these values describe the parameters of a uniform distribution on Load Frequency (i.e., any value between these two numbers is equally likely).
The calculated " probability of core damage given a rupture resulting in a Medium LOCA"is 1E-4 to 1E-2. Assume these two numbers are parameters for a uniform distribution on MLOCA frequency.
The product of the Load Frequency and MLOCA frequency distributions is the distribution describing our uncertainty on core damage risk given a single flawed weld.
33 oflogs\\smt'amtletters\\ WOO PLCRDM Subgroup 10/5/93
~
- 1 EXAMPLE CALCULATIONS The calculated conditional mean CDF Risk is 2.5E-6/ reactor year.
(In EXCEL), the probability of seeing 1 or more core damage events in a year given 11 flaws (= 5% of 216 remaining welds) in the '309' population is (1-BINOMDIST(0,11,2.5E-6, False)) = 3E-5 for the whole industry.
Given that there are 22 affected plants with '309' welds, the 1
corresponding expected core damage risk per plant (per year) when as many as 5% of the '309' weids are defective is 3E-5/22 = 1 E-6.
This is then compared to the DG-1061 (RG-1.174) criterion of an incremental plant CDF risk of approximately 1E-6.
34 oNogs\\smt\\smticuers\\ WOO PIERDM Subgroup 10/37)$
e
m RISK ASSESSMENT RESULTS: !NCREMENTAL CDF RISK PER PLANT FOR EACH SUB-LOT Table C Incremental Core Damage Frequency for Alternative Target % defective in the "309" Sub Imt (216 weld remaining) 0.5%
1%
1.5%
2%
2.5%
5%
All Plants 3F4 5Fr6 8FA 1F,5 1Fr5 3Fr5 Per Plant 1F,7._
1F,7 3F,7 I
5F,7 6E-7 1FA Table D Incremental Core Damage Frequency for Alternative Target % defective in the "82" Sub Lot (120 weld remaining) 0.5%
1%
1.5%
2%
2.5%
5%
All Plants 3F,6 3F4 5F,6 5Fr6 8Fr6 2F,5 PerPlanf 3F,7 3F,7 6F,7 6F,7 9F,7 2FA 35 o \\ logs \\sm0amtletters\\ WOO PIIRDM Subgrosv 10/3/98
SUMMARY
OF RISK ASSESSMENT RESULTS 4
For both sub-lots, the 5% target defective meets the DG-1061 (RG-1.174) incremental risk criterion.
A target defective of 2.5% results in an order of magnitude less risk thari the criterion.
All other " key principles" of the Regulatory Guide DG-1061 (RG-1.174) are met.
The risk assessment supports a choice of either the 5% or a
" defense in depth" 2.5% Target Defective for the statistical analysis.
36 o:\\ logs \\smt\\smiletters\\W(Mi Ti.CRDM Subgroup 10/198 i
I c..
SUMMARY
FOR THE STATISTICAL AND RISK EVALUATION 4
i The sampling done to date by WOGSG members is sufficient to i
demonstrate with 95% confidence that the remaining defects are no more than 2.5% to 5% of the remaining population.
4 i
The risk assessment indicates that the expected incremental i
CDF for 5% defective or less meets the DG-1061 (RG 1.174) risk criterion.
1 37 o%gs\\smt\\smitetters\\ WOO FLCEDM Subgroup 10/5/98
OVERALL CONCLUSIONS Cracking at Prairie Island 2 PLCRDM G-9 is not service induced.
144 Total welds inspected. No additional indications have been found.
- 102 Type 309 welds 42 Type Alloy 82 welds Records search combined with inspection results support the conclusion that the Prairie Island Unit 2 flaw was an isolated event and not a systematic breakdown of quality assurance 38 o:\\ logs \\smtismtleiters\\ WOO I*IfR DAI Subgroup 10,5S3
OVERALL CONCLUSIONS [ CONTINUED?
Sampling to date is sufficient to demonstrate with 95%
confidence that the remaining defects are no more than 2.5% to 5% of the remaining population.
This range of defects does not constitute a significant threat to plant safety goals.
Although the inspection data was not random, the inspections were focused toward variables that could be common mode failure (i.e. Weld Heat 1419U309 and Welder Number 4).
Technical, statistical and risk analyses have shown that inspections completed to date are sufficient to resolve the issue.
39 o%gsisatamiletters\\WOG PLCRDM Subpoup 10/3/98
~
=.
6 A
Westinghouse Owners Group cc:
Mr. Nicholas Liparuto, Manager Regulatory end Engineering Networks Westinghouse Electric Corporation Mail Stop ECE 4-15 P.O. Box 355 Pittsburgh, PA 15230-0355 Mr. Andrew Drake, Project Manager Westinghouse, Owners Group Westinghouse Electric Corporation Mail Stop ECE 5-16 P.O. Box 355 Pittsburgh, PA 15230-0355
. Jack Bastin, Director Westinghouse Electric Corporation 11921 Rockville Pike, Suite 107 Rockville, MD 20852 Mr. Hank Sepp, Manager Regulatory and Licensing Engineenng Westinghouse Electric Corporation PO Box 355 Pittsburgh, PA 15230-0355 4
1 J
,-