ML040980585

To SIR-04-032, Probabilistic Fracture Mechanics Analysis of CRDM Inspection Alternatives at Point Beach, Unit 1

ML040980585
Person / Time
Site:	Point Beach
Issue date:	03/16/2004
From:	Gustin H, Riccardella P Structural Integrity Associates
To:	Office of Nuclear Reactor Regulation
References
FOIA/PA-2004-0282 SIR-04-032, Rev 0
Download: ML040980585 (22)

Text

-Report No.: SIR-04-032 Revision No.: 0 ProjectNo.: PBCH-09Q

-File No.PBCH-09Q-401 March, 2004 Probabilistic Fracture Mechanics Analysis of CRDM Inspection Alternatives at Point Beach Unit 1 Preparedfor:

Nuclear Management Company Two Rivers, WI Contract 30000002, Rev. 2 to Line 6 (1/27/04)

Prepared by:

Structural Integrity Associates, Inc.

Denver, CO Prepared by:

Reviewed by:

Approved by:

62/4<

P. C. Riccardella H. L. Gustin H

- -J/'g°Lt --

H. L. Gustin Date:

3/16/04 Date:

3/16/04 Date:

3/16/04

REVISION CONTROL SHEET Document Number:

SIR-04-032

'Title: PProbabilistic Fracture Mechanics Analysis of CRDM Inspection Alternatives at Point Beach Unit.

'Client:

.' ' ' Ncl'air Mafiagemernt Cornphr i..

I....

SI Project.Number:... PBCH-09Q ky

.1 i

.,"J I

Section I

Pages

Revision l Date I

Comments 1.0 1-L-.1.-2.

.... 03/16/04

. Initial Issue :.

2.0 2 2-5 3.0 3-1 5 4.0

., 4-1 l 5.0 5-1 Appendix A A-i - A-3

Table of Contents Section Pageg

1.0 INTRODUCTION

2.0 OVERVIEW OF MRP PFM TOOL...........................

2-1 3.0 APPLICATION TO POINT BEACH UNIT I........................

3~~~~~~~.0 PLCTO OPON EC NT1...............

7................... 3-1; 3.1 -

Inspection Coverage Assumptionsf................................

3-1i 3.2 Analyses and Results.....................

/'..........:;

i

3. 2 4.0_

C O

N C

L S IO N

6.......

4 1

4.0 CONCRLUSIONS..............

4-1

. *:s s

-a

- r

5.0 REFERENCES

'-;';;;;5-Il I

'APPENDIX POINT BEACH UNIT 1 RPV TOP HEAD NOZZLE INSPECTION COVERAGE.

A-1' I

I II I

i SIR-04-032, Rev. 0

List of Tables Table Page Table 2-1. Benchmarking of PFM Model with Respect to Nozzles with Circumferential Cracks2-3 Table 3-1., Detailed MRPERCRD Input Parameters for Point Beach Unit 1 Paitial

.Inspection Coyerage Analysis.................................................................................. 3-2 Table 4-1.. Results ofPFM Evaluation 4-1

a.

I.

a, V

SIR-04-032, Rev. 0 lV

List of Figures Figure.,

Page Figure 2-1.Flow Chaot of PFM Meth6dolgo 2 -4 Figure 2-2. Comparison of POD curve used for DE with Vendor Dedi o6nstration Programs'.;. 2'5 Figure 2-3. Benchmarking of PFM Model with respect to Plhints withiN6zilb Leaksinf 'Cracking2-5 Figure 3-1. Compairison of Probability 'of Leakage' for Point iBd'cai Unit1'f To'PHead Inspection's under Full and Partial Inspection Coverage Assumptions......................................... 3-4 Figure 3-2. Comparison of Probability of Leakage for Point Beach Unit l Top Head Inspections under Full and Partial Inspection Coyerage Assumptions.

3-5

,.i I

' I I.

m.

t a

I, -:

f f

\\

A; i

v 1;,

q

• 4 P

J.,

A-,

\\

f z--

-/

A.

i OF, P

s A;

F I

SIR-04-032, Rev. 0 v

I.. i. I -

I, ; ' 1

1.0 INTRODUCTION

NMC experienced some limitations in inspection coverage in implementing RPV top head inspections during the Fall 2002 refueling outage (RFO) at Point Beach Unit 1. The steep curvature of the head at the outer row of CRDM nozzles resulted in an inability to obtain complete coverage with the UT blade probe used to perform nozzle inspections between the thermal sleeves and nozzles. In some nozzles, thermal sleeves were removed to allow full access to the tube material for inspection with a rotating UT probe instead of the blade probe. Removal and eventual re-welding of thermal sleeves is a time consuming and dose intensive operation. In some outer row nozzles, for which the blade probe could access at least 50% of the nozzle circumference for inspection, NRC agreement to not remove the thermal sleeves was obtained.

Subsequent to the RFO, it was discovered that there had been some slippage of the rotating probe UT system, such that 100% coverage was not achieved, even for the nozzles in which thermal sleeves were removed. It is conservatively estimated in this report that only §2O of the total nozzle circumferential lengths were completely inspected in the Fall 2002 inspections.

Although the UT blade probe has been improved, and subsequent inspections of the Point Beach Unit 2 CRDM nozzles achieved full coverage without removal of sleeves, it is anticipated that unit-specific differences may result in a repeat of the inability to achieve full inspection coverage in the upcoming Spring 2004 inspections of Point Beach Unit 1. Because of the high dose nature of the thermal sleeve removal and re-welding process (> 3 REM/nozzle in the Fall 2002 inspections), NMC has authorized this analysis to determine the expected safety impact of not removing thermal sleeves, and thus performing less than 100% inspections, should that eventuality arise. This option would require relaxation to NRC Order EA-03-09 [1].

The analysis utilizes a probabilistic fracture mechanics (PFM) tool developed over the past two years by the PWR Materials Reliability Program (MRP) [2]. The PFM tool incorporates the following major elements:

o computation of applied stress intensity factors for circumferential cracks in nozzles of various angles in the head as a function of crack length, o determination of critical circumferential flaw sizes for nozzle failure, o an empirical (Weibull) analysis of the probability of nozzle cracking or leakage as a function of operating time and temperature of the RPV head, o statistical analysis of PWSCC crack growth rates in the PWR primary water environment as a function of applied stress intensity factor and service temperature, and o determination of the effects of inspections (inspection type, frequency, coverage and effectiveness),

to compute plant specific probabilities of nozzle leakage and failure (i.e. nozzle ejection from the head) as a function of operating time and head temperature under various head inspection scenarios.

The MRP tool (MRPERCRD) has been applied to the Point Beach Unit I head inspections considering plant specific parameters such as head geometry, number of nozzles, head temperature, plant operating time and various levels of inspection coverage, to determine SIR-04-032, Rev. 0 1-1

whether a significant improvement in plant safety (in terms of roibability of n6zzle leakage of failure) would be achieved by increased levels of nozzle inspection coverage.

ICa.

• i]

f..: +..J ;; '

a t

J ;

5

• ;z.:

t*.

4,.

I.

S

• 

I I

I



l*I*I..

a

'It IL

'.

I'

.

'j*j

r.

4 v;;:1.

4

,....,,,

,

• • • 1'*.

• 'I*'J

I/I

,I.

I' a,

'*..

a' P

1'* 

'f'4JJ*rId!

'1 I.



f



X s

!),

I&t j{,

jr i',

w l j

'S

@ 14 P s

+

Nl l;

1; r

f ; t S t -

• '* i,1 X * @, J; ; t; i ! s s 4 -

i l 1. ! X I............

t,

- t * -

R R, ",

P 6.

s

@E a.f i.%?,. ;

i M.S,.

i m

• sWs R-tZ Z

b ;

• ^;

• *w; sH s

I I

SIR-04-032, Rev. 0

!1..

2.0 OVERVIEW OF NIRP PFM TOOL Figure 2-1 presents a flow chart of the probabilistic fracture mechanics methodology developed for the MRP' [2]. The methodology has been implemented in the computer program MRPERCRD. The methodology imfplements a tifiie-dependent Monte Carlo analysis scheme whicl'predicts the probability of lA&kage and noz'zle ejection versus time for a specific sei of top head'parameters.' Determinis ic parameters'specific to the top head being analyzed include' number of nozzles, the angle6'&of each 'no'bile Withr'rspect' to the head, no'zzie diameter aid wall thickness, number of heats of nozzle material, and identification of which nozzles are from which heat. 'Anoiher'plant-specific 'inpit consists of K-matrices for eachlof several n6zzle angles. Th s&e'are m'ati"'es of stress intensity factor vers'u's-crack length for several characteristic nozzle angies (usually four) into ivhicfil'the nozi';ls are lumped based on their angle. The K-matrices are'obtairied'fro'i

'determ'l'hiistidf ratrenmechahics analyses of the specific head geometry and may ihclude strerTinistencity' factor data fdr'ranges of nozzle yield strengths and nozzle-to-vessel interfernfce fits 'fo'o cracks centered at botth'e uphill and downhill sides of the nozzles.

Statistical par'ameters (random variablesJ utilized in the Monte Carlo analysis include:

o head operating temperature o yield strengths for each heat of nozzle material o nozzle interferences (or gaps) ;

'o number of assuned cra'cks per nozzle (for' 6DE 'detection) o initial crack size (for NDE detection) o distribution of crack locations (uphill or dowvnhill) o Weibull distribution of time to leakage or cracking (dependent on plant operating time and head temperature) '

o stress corrosion crack growth law

o correlation tactor Ibe~tieen time' to crack initiation and crack giowth, and o critical crack size for' eacfh characteristic nozzle angle.

The statstical parameters 'are' injut ype isorrnal triangular, log -normal; log-triangular, Poisson, Weiblull,etc.), iean d standard devriation or raii"e. As illustrated in Figure 2-1, the analysis algo~iifim con1sists'of twv 'esfe'd Monte Catlo simulation loops, which step through time' for each nozzle in a head,'and then for hte' total number of head simulations specified. For each'nozzle simulatioi, a tine tlkage (oii ackingr is predicted based on the Weibull distribution. ' When 1eakag'e is pr'edifted,' acircuniferential through-wall crack equal to 300 ofnozzle circuniference is eonservati('eA: assii ed to exist The assumed circumferential crack is then grown based 'on the'nozzie-spe'ific stress intensity factor and' a'stress corrosion crack growvth law obtained from andon ingo thestatistical ci'aick groNth law distribution.

The crack growth anailysis for'dadhnozzle conitinbe's until either the'end ofth'eevaluatiorifperiod,'

or until the crack length reache.s the' critical flawv sizie'for that nozzle (established based on random'sampling of the critical crack size distrin1utiofi). The analysisis'repeated for each nozzle in the head, and then for the total iumber of to'p head'sirnulatioiis'specified by the user. The-software records the total number of top heads predicted to experience at least one nozzle leak or failure as a function of operating time, as well as the total number of nozzles with predicted leaks or failures versus time. The probability of a nozzle leak or failure at a given time is the ratio of SIR-04-032, Rev. 0 2-1

the number oftop heads predicted to have leaks-or failures divided'by th 'total number oftop heads simulated.

A correlation factor betwveen crack initiation and crack growthis Included as a user input,,which.

allows one to simulate an inter-relationship betwyeen the time to" iniitiation and the ctr'ack grwth rate for each nozzle. A-high negative correfation factor (-0;9 or -1) ir lies tht a mdte*al heat`.

that tends' to be ba rii j

n9r~)ipiista w

ioteia hat bad from the perspective of crack initiatid '(i.e. leaks earlin life) would also s

have a high crack growth 'rae. A correatiQn factorcQfremplilesnocouplation. ;.

The program also permits the user. tospecify inspections performed at various times within the.

analysis interval..Either visual inspections (for leakige),or non-dekructive examinations (f6r, cracking) or combinations the b

The user also sp'ifiesjinspection coverage

(% of nozzles inspected) and reliability for each knspeet'ioidPqoabilit' 6fdetecting a l6'k if it exists in a visual examination or probability of detection versus creick depth for a noi-destructive examination).When inspections ace'performed' criacks in noz'zes that are predicted to be detected are removed from the simulation, and are no longer considered thretts togrov to leakage or failure. One can thus perform multiple analyses, with and without various forms of inspection, at various intervals, to compare theprobabilities of leakage and failure, and thus evaluate the effectiveness of different inspection options.

Figure 2-2 illustrates the probability'of detection (POD) curve used for'top head ultrasonic inspections, and a comparison to vendor demonstration pr~ograms onicracked nozzles. The POD.

curve, obtained from Ref. [3],jindicates a very low probability of detection fo'.'ery small cracks, for which the tiwoNDE vendors missed about half of-the cracks in the demonstration 'samples.

As the crack dppths get larger (greater than 0.I 5").however, the vendlors detected 'all cracks in the samples. For this crack size, the POD curve predicts about a 75% probabilit 'of ditection`. -The POD curve continues to increase with increasing crack depth,' approabhing a rmaximium POD of 95% for'crack-depths of 0.35.", about half of the typical CRDM'nozzle wvall thickness.' (The program assumes at least 50/o chance of missing a crack'no matter'how deep The MRPERCRD methodology has been beirichmarked and calibirted with respect to' field inspection results, both in tems of predicted nozzle eakg dthe'occurrenc'6f ' -'

circumferential cracks. Figure 2-3 presents a comnparisr pof.MPRC9RDpredictions wvith-:

plants that have observed leakage'or significant ciigck in,,C-0,M nozzle. -The data poins represent the time to first leakage or.rrapking dlstribqpfor the fourteen U.S. plants that have experienced leaking or cracking. The solid Curves r

.h R.E. kd Monte Carlo' prediction of time to.first'leakage for,the ba'se~ca'se'pqrwqt'e3. Th h a

e on this plot is EFPYs at 6000F, or, EP S, which is onsist nt witht data siice they too are, normalized to 600°F.(EDy).JThe agreement squtg dashed curve in gure23 is.

the leakage prediction corresponding to a set of benchmarkedparameters developed from the circumferential crack comparison (see discussion below). itisseen that switching from baase case to benchmarked parameters doesn't have a strong effect' on the'leakage predictions, so the dashed curve is a little mnore coniservative, but still in reasonable agreement wiihthe field,feakage data.

I A

A11 l

CI;;

,-AA I

-A ii, A'

'~.

il31rX-U-UJV3, XKVV. U 4_4

Table 2-1 contains data from the eleven nozzles in U.S. plants that were found to have circumferential cracking, sorted in order of increasing crack length. The nozzles were sorted into bins of 300 crack length increments, and the bins were summed as shown in column 4 of the table. There were 11 total cracks of length greater than 300, seven of length greater than 600, and so on, down to two of length greater than 1500.

Table 2-1. Benchmarking of PFM Model with Respect to Nozzles with Circumferential Cracks Plant Data; MRPERCRD Results Cumulative#

Conservative

. Ultimate Circ. Crack of nozzles w/Crack Frequency Base Case

. Correlation and Corr.

Sensitivity Lengths (0)

Nozzles Length Greater TfAn' (881 Inspected)

Factor =-1 Factor =

Case 30-60 4

30 ill 1.25x10-2 2.24x10-2 2.24x10 2

!>>1317xO-2 1.01x10 60-90

.1 60 7

7.95x10 3 2.94x10-3 3.94x10- 3 6.40x104; 9.19x10-3 90-120 3

90-6 6.81x10 3 1.62x103 2.30x103

-3.87xl0-5.14x10 3 120-150 1

120.

3 3.41x10 3

.1.19x10 3 1.66xio 3 '

2.8ix103

3.49x103 150-180 2

150 2

2.27x10 8.98x10 1.36xl0 3 2.25x10.

2.74x1 0' Predicted Collapse:

0.5 5.68x104 3.97x10 4 6.54x104 I l 1.09x10-4 1.23x10- 3 The group of plants containing these circumferential cracks had an average age of-19.75 EDYs at the time of inspection (very close to, 20 EDYs). A total of 881 nozzles in plants of this age have been inspected. Thus 881 was used as a denominator to compute frequency of occurrence of circumferential cracks exceeding the various crack lengths in column 5 of Table 2-1. Finally, columns 6-9 of the table present MRPERCRD predictions of cumulative probabilities of cracking at twenty years, computed on a per nozzle (rather than a per head) basis. It is seen from these results that, except for the probability of a 300 crack, the base case consistently under predicts the probabilities of large circumferential cracks by about a factor of 2 to 4. The remaining columns contain MRPERCRD analysis results with increasingly conservative input parameters. The first step was to change the crack growth to initiation correlation factor to -1.0. This increased the circumferential crack probabilities somewhat, but they still under-predict the cumulative distribution from the plant data. In the next column, a combination of correlation factor =-I plus a more conservative Weibull 0 distribution was assumed (triangular with 0-mean = 15.2, +/-6.5).

This gave the best general comparison of circumferential crack probabilities, over-predicting at some crack lengths, under-predicting atf'thers, and agreeing almost exactly at the largest crack length (>1500). This case was therefore designated as the 'benchmarked" parameters, and that column of Table 2-1 is highlighted. These benchmarked'piaineters were used for the plant specific evaluation of Point Beach Unit 1, discussed in the following section.

SIR-04-032, Rev. 0 2-3

Random:

1. of Cra

• Locatior Determnii Initial D All cracks Initializ StarNew Determmsic TopHead Top Head Vaniabe..*,

Simulation,......*;;.

,,A et,.

..,..i

Lcks, U r

, Individual n;.(UpHillorDownlill)

Nozzle,;;*

r mepth & Length Yes Nozl

\\

care OD don

\\

ek

\\

C G

nnet ranorc,..

. fT'.

Sta~C each crack

?

t

=

~(YS Fit da/s; t'

1--

Crntical Crack--

r

.Size.etc)

K matrices indexed to:

= tlcakA^

Next Crack Interference Fit, Yield Strerigth; UpHill, DotvnHill Interpolate.

Center, Middle &..

.K.l.

Outer Nozzle SCZ G -t

1. t" ;

FailuIre l41+

speck9./

Tor i

N-for leakage.

Numbe'r for.

inspection % -.-

-i

, t.bNo,,.

adrc?//'-............ c....

Failure Modes:

es,

'p*

FailureCounter:

N:.o',;.

No Yes.

LJP

-st

No Yes

/

L~ast 1_/

Nozle7 A -
All VL1l1_:_1. i -

.1, j -,% 1

t m

I,

(

Figure 2-1. Flow Chart of PFM Methodology SIR-04-032, Rev. 0 2:4

Probability of Detection Curve Used In MRPER Algorithm 90%

75%

C.

0 60%

a1 10

.0 45%

30%

15%

.0% I--

0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 Flaw Size (In)

Figure 2-2.! Comparison of POD curve used wir NDE with Vendor Demonstration Programs 0.7 is

.0 0.

Z. 0.6 Z2 045 E

3 Figure 2-3. Benchmarking of PFM Model with respect to Plants with Nozzle Leaks or Cracking SIR-04-032, Rev. 0 2-5

3.0 APPLICATION TO POINT BEACH UNIT 1I -.

3.1 Inspection Coverage Assumptions Nozzle by nozzle inspection results from the Fall 2002 inspection of the Point Beach Unit I top head were reviewed to estimate the approximate'percent coverage achieVed in that inspection (see Appendix A). The majority of coverage problems were experienced idi the outermost row of nozzles with thermal sleeves (nozzles 26 through 33). Four of these nozzles were inspected with the blade probe, achieving less than full inspection coverage (but greater than 5 0%). Three of the' nozzles in this group required thernidl sleeVes to be removed for inspection' with the rotating probe,- and the stall problem was then encountered. Nozzle I (top dead center) also required thermal sleeve removal, but didn't experience stall problems with the rotating'probe. 'Several other nozzles that did not have thermal sleeves present to start with were also inspected with the rotating probe, five of which experienced stall problems.

A total of seven transducers were mounted on the rotating probe, and plots of transducer' coverage versus azimuthal angle were reviewed [4] to determine which portions of the nozzles were scanned by each transducer. For purposes of this evaluation, it was conservatively assumed that any portion of the nozzle circumference'thait was niot'scanned by all seven transducers was un-inspected. The resulting coverage fractions for each nozzle are listed in Appendix A. The total inspection coverage achieved by both 1he'blade probe and rotating probe is summed on page A-3, and amounts to a total of 82% of the circumferential lengths of all nozzles. Finally, in keeping with the assumptions in Ref. [2]; NDE of the nozzles (with no weld inspection) was specified as 80% coverage, under the assumption that 20% of the chance of leakage is due to weld cracking.- Thus the resulting percent coverage input to MRPERCRD for the Fall 2002 inspections was 65%. All inspections applied the FULLV POD curve illustrated in Figure 2-2.

For the upcoming Spring 2004 inspections, assumptions were made regarding which nozzles might encounter inspection coverage limitations if the thermal sleeves are not removed, and what percentage the limitations would be.--As seen in Appendix A, it was assumed that all eight,

nozzles in the outermost row with thermals sleeves, plus the'top dead center nozzle; might-.

experience coverage limitations up to 50% of the nozzle-circumiferences. (Note, if inspection limitations are actually encountered during the outage, the-analysis~will be redone considering' the actual nozzles and coverage levels.) -As indicated on page A-3; the sum of the assumed inspection coverage' for the Spring 2004 inspection amounts to 90.8%, and applying the 80%

factor for lack 6f weld inspections, the resulting percent coverage input to MRPERCRD for the Spring 2004 inspections was 73%.

As a base case for comparison, a second analysis was performed of the theoretical case of.:

complete inspection coverage, in accordance with the NRC Order,- in both the Fall 2002 and -

Spring 2004 inspections. ' Accounting for no weld examinations, this results in percent coverage input to MRPERCRD of 80% for both inspections. --

S.

.0 SLR-04-032, Rev.

0

3-1i-

I 3.2 Analyses and Results A tabulation of the detailed input parameters used in the Point Beach Unit 1 analyses is presented in Table 3-1. The results are presented graphically in Figures 3-1 and 3-2.

Table 3-1. - Detailed MRPERCRD input Parameters for Point Beach Unit 1 Partial Inspection.Coverage Analysis Item.

Variable.

ProbablUlty Mean Standard Lower Upper Distribution Value Deviation Bound Bound Units Number of Nozzles in the Top Head 49 nozzles Nozzle ivall thickness 0.625 inches Maximum angle for center Nozzle 5degrees 1.Constant*

NA NA NA ders Maximum angle for middleinner' nozzle.

18 degrees Maximum angle for riiddle oute; riozzle.

30 degrees Maximu!m angle to, outer nozz1.

I.

43.5 degrees 2

Number of Heats of Material '

-- Constant 5

NA NA NA NA 3

Reference Temperature for Leakage Constant 600 NA NA NA OF 4

Reference Temperature for Crack Constant 617 NA NA NA OF Inspection Plan r ;'

,v

• . "rC

a e

...r d~.tl 'd rage NumberofNDElnspecti6hs Conltat'ii I

2 NA 5

Percentage of Nozzles Inspected Consbanif

'80%

i 65%,73%

NA

' Ispecti6n technique Constant t FULLV' FULLV NA NA lIspection Time 1 Constant 177828.

177828 Hvus Inspectidn Yime 2 Constant 190311 190311 Hours 6

Input Nozzle Temperatute Normal 501.6 0.001 NA

'NA Normal 60' 1.5 NA NA KSI Normal 40.5-1.5 NA NA KSI 7

Material Yield Strength Normal 46.5 1.5 NA NA KSI lNcrmalI'

47 1.5' NA NA KSI

• .*'-Normal 43 1.5;'"

NA NA KSI

'Probabi;ityofLeakage f

Weibull 8

-:Weibul)tH'eta 5.2 NA'.

NA NA EDYs Weibullba

'..3; NA

'NA NA

• ActiV tior n rgy fdr~

la'ak3e,,

50

'NA NA, NA kcalmol

• LocalIVariab~lity in Lemakage applied t6'E s

9 W'eibfcrl alpha,.

i..

0 I

NA

-6.5 6.5 EDYs 10 Constant term in crack growth law Lop Triangle

-15.2485 NA

.035.. -17.461 NA 11 Exponential term in crack growth law Normal 1.16 0

NA NA NA 12 Local variability ir crack growtpateh I, log Tringle

0

NA

.1 1.6 NA 12 term by matenial heat?

L_________-NA-1.

N 13 Crack growvth threlhold

N'ormal

'8.188 0.001 NA ksi-in^.5 14 Crack growth activation energy '

N o m aI 31.05 0.001 NA NA kcalmol 14 threshold 3105_____ANA kcl/_l 15 Correlation coefficient between SCC_

osat..

AN AN 15 initiation and SCC crack growth '-'

Constant

• 1.0 NA NA NA NA 16 Initial shrink fit for the center no~ile

Normal

'-0.00032 0.00083 NA NA inches SIR-04-032, Rev. 0 3-2

1, I

.,I ;

ItmVariable Probability Mean Standard Lower Upperl Unit Distribution Value Deviation Bound Bound Units 17 Initial shrink fit for the middle Inner

-Normal

-0.00005

,o.oui7`

NA, NA Inches.

nozzle 1-1 Initial shrink fit for the middle outer i

18Inozzle Normal'

-'.00020*

0.00158-;

NA

NA4, Inches'~

19 Initial shrink fit for the outer nozzle Normal

-0.00030 0.00138 NA NA Inches 20 Number of Cracks per Nozzle Poisson 2

NA NA NA NA 21 Initial Crack Depth-Normal:.' -

0.05; v

NA6

'inches 22 Tsar

,,, Constant..,,.Q5..l NA NA.,,

,NA NA Critical Crack Length E'*i

, I

.i I.

.. nter nozzle..

Normial.:.

.10.01 0.01...

NA.

.,-NA inches 23 Middle inner nozzle Normal,,

10.18 0.01 NA NA.

'inches Middle outer nozzle Norirnal 10.92 0.01_

NA NA Inches Outer nozzle' Normil 12.24, 0.01. '

N NA

7. Inches 24 Number of equivalent full power hours Constarnt 224694 NA NA'

,': NA-

' ' Hours 25 Number of time Intervals Consktnt 27 NA NA

- NA ;

NA 26 Number of Monte Carlo Simulations Constant 500,000.

NA NA '

NA NA

-11I IJ

I

.i J

I I-

'-';X 5

From Figure 3-1,it is seen that th6 predicted probability'of leakage (POLK for Point BeachiUnit I reached approximately 10% prior,tolihe initial inspections performed in Fall, 2002. Following

-that inspection, and the subsequent inspections planned for Spring 2004; the;POL drops dramatically, to -1% at the'time of head replacemnent in Fall 2005,'under either~inspection coverage assumption.; 'The difference between the theoretical case of full nozzle inspections in' accordance with the NRC Order and the-artial coverage case described,above is illustrated by the ied and green lines in the figure: Had full coverage been-achieved inboth inspections, the POL-at head drplachement'Vould ha've been'0.8%. -With the conservative estimates of partial coverage levels for the Fall 2002 and the Spring 2004 inspections; the resulting probability of -

leakage at the time of head replacement Would be' 1.5%.,.

S i m i l a r l y, f r o F i u r 2 -, i t i s s e e t h ) t h Similarl, 'from Figure 2-2, it'is'sd-n that the difference in'the predicted probability of nozzle ejection is also quite small.' Had-full'c6verage be'en achievedin both inspections, the probability of failure at the time of head rep'lacemfent would hav6bii 1;'8'x' 104.- With th& conservative estimates of partial coverage leviels for 'the 'Fall 2002 and the Spring 2004'inspections, the '

resulting probability of failure at the time of head'replacenment hi ould.be 4.3x 104. These:

differences are considered 'o be quite srn~all, ad calearlyinot'wortfi the.acditional nain-REM exposure that would be incurred if therinal sleeves arXiri

~it6-be removed to' achieve full inspection coverage. -The probabilities also xemain well below generally accepted limit& (POL <

5% and POF < iX.10 3)'i r the initial ba6fiih.6 inatib7ns were performed.

the, m as

i.

One final point should be noted regarding the assumptions made irk this PFM analysis. Although the predicted probabilities of leakage and failure ire a function of the-many input variables assumed in the analysis, the specific set of variables used to compare inspection programs have been benchmarked and calibrated with respect to field experience. Also,.changes.to these variables would affect the analyses of the full coverage case as well a's the partial'coverage case in approximately the samemanner. -Thu's the comparison and conclusions of this' study are expected to remain the same for realistic ranges of these input variables.

SIR-04-032, Rev. 0 3

-3

I 2.001-al 1.80E-01 J

• No Insp.

1.60E-01 Part.lnsp.Coverage Insp. Per NRC Order

.S 1.40E-01

=

1.20E-01 1.OOE-01 10, f

8.00E-02 Planned H I

Reol. F-05 0 6OOE-02.

4.00E-02 F-02.

2.00E-02

____5 S-04 O.OOE+00 0

5 10 15 20 EFPYs Figure 3-1. Comparison of Probability of Leakage for Point Beach Unit 1 Top Head Inspections under Full and Partial Inspection Coverage Assumptions 25 SIR-04-032, Rev. 0 3-4

6.50E-03 6.0,.,

Part.lnsp.CoCerage 6.00E,-0,3 i

.{.

.. tInsp. Per NRC Order

/

5.50E-03

-5.00E-0D3

-4.50)E-03 cm 4.00E-03 c3.50E-03,

/

z 3.00E-03_:__

u-2.50E-03 Planned Hea J 0L-0 2.OOE-03 f

1.50E-03 Pn Head 1.00E-03 I

5.00E-04 S.04 j 0.o002+00 0

5 10 15 20 25 EFPYs l,

• 1

~~,

I.'ts*,*,1a Figure 3-2. Comparison of Probability of Leakage for Po5int Beach Unitf 1 Top 'Head Inspections under Full and Partial Inspection Coverage Assumptions SIR-04-032, Rev. 0 3 5

4.0 CONCLUSION

S Conservative evaluation of partial inspection coverage of the RPV top head nozzles at Point Beach Unit I has been conducted using a generic MRP probabilistic fracture mechanics tool [2].

The tool has been discussed. extensively with NR-C / ACRS, and has been benchmarked and calibrated with respect to a large body of inspection results in U.S. PWVRs. The results indicate only a small differenc& in the probabilities of leakage and failure compared to full coverage inspections (see Table 4-l)': The differences do not appear to warrant the additional man-REM exposure that would be incurred to remove thermal sleeves in order to achieve full inspection coverage, should limitations be encountered. The resulting probabilities of leakage and failure are well within generally accepted li mits urider full or partial inspection coverage assumptions.

Table 4-1. Results of PFM Evaluation Probabilities at Head Re placement Full Coverage Generally (per NRC Partial Accepted Order) coverage Limits POL 0.80%

1.50%

5%

POF 1.8 x 104 4.3 x 104 1 x lo 3 SIR-04-032, Rev. 0 4-1

-;~

.d

5.0 REFERENCES

1.

U.S. NRC Order EA-03-009, "Interim Inspection Requirements for Reactor Pressure"

-Vessel Heads at Pressurized Water Reactors"; issued on February I 1, 2003 ;* :

• )

o 4

4 1

• i; sty

2.

'Materials Reliability Program, 'Probabilistic Fracture Mechanics Analysis of PWR.,

..Reactor Pr.essure Vessel Top. Head Nozzle Cracking," MIRP-105., March 2004. EPRI:.

Proprietary.r.

• A s

i;::

.,4 a

2;;

v

{l 3.-

,.-.Dimitrijevic, V.,and Ammirato, F., !se of Nondestructive EvaluationDatato Improv~e Analysis of Reactor Pressure Vessel Integrity," EPRI Report TR-102074, Yankee Atomic Electric Co. March 1993

4.

Framatome NCR #6028873, Rev. I_,and D9s/ositi 21, 91/2003.

S 0..2 4

4 SIR-04-032, Rev. 0 5-'1---

APPENDIX A POINT BEACH UNIT 1 RPV TOP HEAD NOZZLE INSPECTION COVERAGE SIR-04-032, Rev. 0 A-1

Fall 2002 Nozzle Angle (actual)

- 2 3

45 6

7 8

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 0

0 19.5 19.5 19.5 19.5 13.6 13.6 13.6 13.6 28.1 28.1 28.1 28.1 31.8 31.8 31.8 31.8 30 30 30 30 31.8 31.8 31.8 31.8 37 37 37 37 37

.1 Rotating Removed TS & Re-welded I

Blade I

Blade 1 Blade I

Blade 1 Rotating I

Rotating.

1,,

0.375 Rdtating ' Stall" 1 Rotating I Blade Blade 1 Blade 1 Blade 1 Blade I

Blade I

Blade 1 Blade I

Blade I

Blade 1 Blade 1 Blade 1 Blade I

Blade I

Blade 1 Blade 0.78 Blade 0.66 Blade 0.52 Blade 1 Blade 0.5 Blade Spring 2004 (assumed) 0.5 I. 1 1

I 1 1

1 1

1 1

1 1

1 1

1 1

1 1

1 1

1I I

I 0.5 0.5 0.5

.0.5 0.5 0.5 0.5 0.5 1

1 1

1 1

1 1

1 I

Removed TS & Re-welded, 31 37 0.42 Rotating Stall Removed TS & Re-welded, 32 37 0.25 Rotating Stall Removed TS & Re-welded, 33 37 0.17 Rotating Stall 34 35 36 37 38 39 40 41 42 43 44 43.5 43.5 43.5 43.5 9.7 9.7 9.7 9.7 22 22 22 I

0.24 0.17 1

1 1

1 0.49 1

0.39 Rotating Rotating Rotating Rotating Blade Blade Blade Blade Rotating Rotating Rotating Stall Stall Stall Stall SIR-04-032, Rev. 0

'A-"2

I 45 46 47 48 49 22 22 22 22 22 0.44 Rotating Stall 0.25 Rotating Stall 0.33 Rotating Stall 1

Rotating 1

Rotatina 1

1 1

1 1

Totals 0.8 Factor for r weld insp.

49 0.816 49 12 0.908 0.653 0.727 SIR-04-032, Rev. 0 A-3