ML20135D259

Rev 0 to ANO Unit 2 Projected End of Cycle 12 Circumferential Crack Population

ML20135D259
Person / Time
Site:	Arkansas Nuclear
Issue date:	10/16/1996
From:	AFFILIATION NOT ASSIGNED
To:
Shared Package
ML20135D238	List: 2CAN029707, Forwards non-proprietary Repts Re ANO-2 SG Cycle 12 Operational Assessment ML20135D243 ML20135D248 ML20135D255 ML20135D259 ML20135D262
References
TR-96-021(NP), TR-96-021(NP)-R00, TR-96-21(NP), TR-96-21(NP)-R, NUDOCS 9703050156
Download: ML20135D259 (26)

Text

. _. _ _ _ _ -.-....-

i l

l

}

i l

l TR-96-021(NP), Rev. 0 ANO Unit 2 j

Projected End of Cycle 12 l

Circumferential Crack Population l

i l

i l

Prepared For Entergy Operations, Inc.

By Tetra Engineering Group, Inc.

l October 16,1996 i

i I

i i

9703050156 970226 i

PDR ADOCK 05000368 l

p PDR W Tetra Engineering Group, Inc.

l USA: 110 Hopmeadow Street, Suite 800, Westogue, CT, 06089 (1).860.651.4622 France:Immeuble Petra B, B.P. 272, 06905 SOPHIA ANTIPOLIS (33).92.96.92.54 i

i

Tetra Engineenng Group, Inc.

@ 1996 AllRmhts R1 served.

l i

2 1

l l

1 l.

Non-Proprietary This is a non-proprietary version of a Tetra Engineering Group, Inc. report. Proprietary information has been l

removed at locations indicated by a heavy vertical bar in the right margin. It is submitted in confidence and is to be used solely for the purpose for which it is furnished. This report, parts thereof, or the information contained within, may not be transmitted, disclosed, or reproduced in any form without the written permission of Tetra Engineering Group, Inc.

@ Copyright,1996 Tetra Engineering Group, Inc.

Copyright under International Copyright Conventions and under PAN AMERICAN Conventions.

i I

ALL RIGHTS RESERVED i

i I

i TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Contents e i

I j

Tttt: Engineenng Group. Inc.

C 1996 AllRights Reserved l

Contents l

Introduction 1

l Historical Review 2

Inspection Results..

.2 Inspection Distributions _

.2 Comparison of Largest Flaws

.3 Projection of Number of Flaws 6

General Approach..

.6 Discussion of Weighted POD._

.6 Application to ANO-2.

7 Cold Leg Cracking..

.(0 Current Cycle Predictions.-

.. I 1 NDE Sizing Uncertainty 12 Plus Point RPC Probe.

13 Probe Comparison -

. 13 Plus Point Probe Bias..

.15 Apparent Growth 17 Monte Carlo EOC Distribution 20 Monte Carlo Process..

-20 Monte Carlo Simulation Results..

.20 Conclusion 22 References 23 i

l l

l L

l TR-96421 ANO Unit 2 EOC 12 Circumferential Flaw Population Contents e il s.

l L r,,,. e,,,

, c,

, c.

.,,,s,,,,,,,,. n.

a.

1 i

a a

Introduction i

4 The purpose of this report is to provide an assessment of the circumferential flaw i

population in the'ANO Unit 2 steam generator tubes. Circumferential flaws have 1

occurred in the ANO Unit 2 steam generator tubes as a result of outer diameter stress corrosion cracking (ODSCC) at the top of the tubesheet. These flaws are identified by NDE eddy current inspection techniques during inspection outages.

i A projection of the population of circumferential flaws as a function of percent degraded area (PDA) expected at EOC 12 is provided. This projection considers l

NDE uncertainty and potential growth of existing flaws.

t J.

Steam generator tubes with identified circumferential stress corrosion cracks are currently repaired on detection without regard to the size of the flaw. New flaws j

may occur over the operating period or flaws may be inadvertently left in service j

due to NDE detection limitations.

Eddy Current probes of various designs have been used. Recent advances in eddy current test equipment and techniques have allowed for more accurate detection i

and sizing of flaws. Most recently, Plus Point rotating pancake probes and techniques which allow depth sizing of circumferential flaws over 10 degree are segments have been used. An assessment of the accuracy of the current NDE technology is provided.

6 i.

1 i-l l

1 i

i a

i i

TR-96 021 ANO Unit 2 EOC 12 Circumferential Flaw Population Introduction = 1 d

.-vve, - - - -..,

, - ~ - -

a e,--

m

Tetro Engineering Group, Inc.

c 1996 AllRights Reserved H. torical Rev.iew is Inspection Results Circumferential flaws were first identified in the ANO Unit 2 steam generator j

tubes in 1992. Table 1 provides a historical review of the number of flaws and the outage that they were identified, reference 1. Staic ef the art NDE equipment and techniques were employed at each outage. All tubes identified circumferential flaws were removed from service.

Table 1 Circumferential Cracks In ANO Unit 2 Steam Generator Tubes Circumferential Cracks Outage EFPY Probe Type SG A SG B Both 2F92 8.51 0.080 RPC 208 11 219 2R9 8.85 0.080 RPC 17 8

25 2P93 9.36 0.080 RPC 45 3

48 2R10 10.16 0.Il5 RPC 147 23 170 2P95 10.85 0.I15 RPC 203 80 283 2Rll 11.46 Plus Point 523 215 738 Total 1143 340 1483 Inspection Distributions Figure 1 shows the distribution of average depths for each inspection. Average depth is defined as the maximum depth times the are length divided by 360 degrees. Average depth provides a conservative representation of the degraded area of the flaw without the costly requirement of a detailed 10 degree increment analysis for every flaw.

The observed distributions of average depths can all be model by a gamma distribution. Inspections 2F92 through 2P95 are statistically indistinguishable.

The 2R11 inspection peaks at a lower size, showing the influence of the improved detection capability of the Plus Point RPC probe. The difTerences are statistically significant. When adjusted for the expected MRPC POD the 2P95 (and earlier) distributions are similar to the 2R11 distributions.

TR-96421 ANO Unit 2 EOC 12 Circumferential Flaw Population Historical Review e 2

Titts Engineering Group, Inc.

C 1996 AllRights Reserved The repeatability of the distribution shape will be used to model the flaws projected to be present at the end of a future operating period. As demonstrated in figure 1, the shape of the distribution has remained constant regardless of the period of operation and regardless of the number of flaws detected.

Figure 1 ANO Unit 2 SG A Circumferential Cracks Av6rege Depth Gamma Distribution 0.080 0.070 p. _..

g 5...

,,....a...

.. 2R9 0.060

. + 2PB3I 3., n 0.050

^*

2R10'

' m-2P95

$ 0.040

' 0.030 1.

4

• 0.020 N

0.010 s

.a.

...a j..m 0.000 m

0 10 20 30 40 50 60 70 80 90 100 Average Depth Comparison of Largest Flaws Because the probability of tube burst and potential leakage values are controlled by the largest flaws present in the steam generator tubes, the tail region of each inspection was compared. Boxplots are used to provide this comparison. Figures 2-4 show Tukey Boxplots where the central " box" represents the 25,50 (median) and 75 percentiles of the data. The bars represent the 10% and 90% values (outliers are shown by individual circles and stars). figure 2 shows a comparison the largest twenty flaws at each inspection. Because of the increasing number of flaws, it is expected that more flaws will occur in the tail region of the distribution and therefore the trend in the largest twenty flaws should be increasing. A better comparison would be to examine the largestpercentage of flaws present at each inspection. Figure 3 and 4 show the largest 20% and 10% of flaws at each inspection respectively. In these figures it is clear that the trend in the largest flaws or tail region of the distribution has remain stable or decreased with each inspection. Especially note that the median values for the latest inspections are essentially the same.

TR-96 021 ANO Unit 2 EOC 12 Circumferential Flaw Population Historical Review e 3

Tetra Engineering Group. Inc.

C 1996 All Rights Reserved.

i i

Figure 2 Twenty Largest Flaws at Each Outage i

100 m,

90 80 M1 70 mi 60-va l

l E

4 0

N=

20 20 20 20 20 20 i

2F92 2R9 2P93 2R10 2P95 2R11 Outage Figure 3 Largest 20% of Flaws per Outage 100 90 g

80 0

3

[ 70 6 60 g

v 50-l 1

O 1

@ 40 30 20 10 0

N=

41 10 10 M

57 148 2F92 2R9 2P93 2R10 2P95 2R11 Outage i

i l

1 i

i TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Historical Review e 4

Tetra Engineenng Group, Inc.

C 1996 AIIRights Rsssrved l

)

i Figure 4 l

Largest 10% of Flaws per Outage i

100 90-g I

80 0

oT 70 g

O

~

30-20 I

10

)

i O'

I N=

21 5

5 17 28 74 l

2F92 2R9 2P93 2R10 2P95 2R11 i

l Outage l

TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Historical Review e 5

Titt3 Engoneenng Group, Inc.

C 1996 AllRights R1 served Projection of Number of Flaws General Approach The number of circumferential cracks of all sizes expected in a future cycle is estimated using plant specific Weibull models. The general Weibull formulation is a three parameter distribution, where the cumulative form is given by; 4un-af P(EFPY) = 1 - e where; 1

P(EFPY) = Cumulative percentage cracked as a function of total EFPY p = Weibull Shape Parameter (dimensionless) c = Weibull Location Parameter in EFPY 2 = Weibull Scale Parameter in EFPY, Separate models are used for each steam generator and for the hot and cold legs within each steam generator.

At each inspection, the number of total flaws is estimated based on the number found and the " weighted POD" for that inspection. Since all discovered flaws are repaired, they are removed from the population and the number of remaining 1

(mostly small) flaws are estimated. These are carried forward to the next inspection. In this way a more accurate projection of"new flaws" can be made.

These adjusted results are used to determine the Weibull Parameters. The Weibull Location Parameter is taken to be zero EFPY for hot leg cracks and is estimated from all available data for cold leg cracks. The Weibull shape and scale parameters are estimated using the Levenberg Marquart non-linear solution algorithm.

D!scussion of Weighted POD The Probability of Detection (POD) of the eddy current in-service inspection of steam generator tubes is a function of the size (described by the flaw length, depth, or degraded area). Equation 1 gives the general form; POD = P(CrackDetectedlC, = c,)

where:

(1)

C, = Size of actual flaw TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Projection of Number of Flaws e 6

3 Tttre Engineenng Group, Inc.

01996 AllRights Reserved.

4 To estimate the number of flaws expected in the next operating cycle, it is necessary to account for the varying POD as a function of flaw size in producing l

an overall or weighted POD. This is given in equation 2.

l Co mEE WeightedPOD = lP(CrackDetectedlC, = c,f(c,)de,

...e.

where:

(2) l Thus the Weighted POD represents the probability of detecting a flaw of any size drawn from the flaw population present in the steam generator. The' distribution fo represents the actual distribution of flaws, not the distribution as detected.

The goal in this procedure is to determine if changes in the number and sizes of flaws found in ANO Unit 2 2R11 inspection were due to improved inspection technology (Plus Point RPC Probe).

The Weighted POD used for computation was 60%. This value has been shown to be typical of eddy-current SG inspections for a variety of flaw morphologies.

1 In the Surry Study involving the original Surry Steam Generators, reference 3, 60% was established as reasonable for pitting and wastage. Recent results for the PISC III study, reference 4, show at least 55% or more for all types of defects.

The number of circumferential cracks was limited, but available results on POD as a function of through-wall depth do not differ significantly from the other types of defects in the study. A recent French (EdF) study of pulled tubes, reference 5, reported a 56% POD for circumferential flaws of 5% to 78% PDA. These studies support the use of a 60% Weighted POD for use in projecting flaw populations of j

circumferential flaws.

Application to ANO-2 i

Figure 5 below shows the cumulative as-found circumferential cracks (SG A, Hot Leg), the best estimates of all cracks, and the projections based on each data set up to 2P95. For all inspections up to 2P95 MRPC was used for TTS inspections.

A weighted POD value of 60% is used for these inspections. The 2R11 inspection 1

was used with Plus Point for TFS inspection. No equivalent weighted POD is -

available for the Plus Point, but all available information indicates it is better than MRPC. Using the updated values to perform a best-estimate Weibull projection, the best-estimate of SG A Hot Leg flaws is 538 compared with the 473 found by inspection. This gives a point estimate of the 2R11 POD as 87%.

TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Projection of Number of Flaws e 7

1 Titt: Engineenng Group. Inc.

@ 1996 AIIRighis Reserved.

Figure 5 4

Projection of Circumferential Cracks SG A 1200 t

1000

- - - ~.

p 800 s-,,

2

% 600

- ~. " - -

j

)

o 2

400,,

t~

1m I

200j 0

8.51 8.85 9.36 10.16 10.85 11.4 EFPY L-__QWeibull Projection up to 2P95

_-M-Adjusted Projection The Weibull fit to the Best Estimate crack numbers is shown below:

Table 2 SG A Weibull Parameters Parameter Estimate Asymptotic Lower 95%

Upper 9?%

Standard Error Confidence Confideht Band Band Beta 4.63 0.2688 3.78 5.49 Lambda 17.2I 0.534 15.52 18.91 The fit is very good with a R squared value of 99%.

l The SG B Hot leg data is more limited, but greater numbers are now beginning to be found. The Figure 6 shows the as-found and best estimate numbers.

)

TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Projection of Number of Flaws e 8

a Titra Engineering Group, Inc.

C 1996 AIIRights Reserved.

Figure 6 SG B Hot Leg 0.08 1 -M-- Found % -

i.-+-BEst Estimate' % i

~

E-e-Projicted AsSFou$d -

0.06

-i MProjehted Beat Estimate ~

);lh 8

s 0.03 0.02 0.01 9

8.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 EFPY The rate ofincrease is quite large and may be subject to adjustment when more inspections are performed.

Weibull Parameters are shown below.

Table 3 SG B Weibull Parameters Best-Estimate (2F92 through 2P95)

Parameter Estimate Asymptotic Lower 95%

Upper 95%

Standard Error Confidence Confidence Band Band Beta 11.07 0.89 8.23 13.9 Lambda 14.37 0.352 13.25 15.488 The high beta values are consistent with the rapid increase in number but should not be considered reliable due to the intrinsic difficulty in fitting exponential type life distributions such as Weibull's to small tail data. A more realistic assumption may be to assume that the SG B Hot Leg will show the same time behavior as SG A but with a delay in initiation. A review of the Millstone Point Unit II data shows a similar delay (roughly 1.2 EFPY). Under this assumption a three parameter Weibull is postulated with the following parameters.

TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Popult. tion Projection of Number of Flaws e 9

Tztc Engineering Group, Inc.

C 1996 All Rights Reserved.

Table 4 i

SG B Weibull Parameters Parameter Estimate Asymptotic Lower 95%

Upper 95%

Standard Error Confidence Confidence Band Band Beta 4.63 NA NA NA Lambda 17.21 NA NA NA Epsilon 2(EFPY)

NA NA NA The Figure 7 shows the wide range of projections under these assumptions; Figure 7 SG B Hot Leg 0.3 l

0.25

'-+-Found %

1

. - G-Projected,Be'sI Estimate u 0.2

+3 Pirim'eter Weibull:

Ee T 0.15

.9 Ti a

0.1

-+

O.05 O

I-I 8

8.5 9

9.5 10 10.5 11 11.5 12 12.5 13 EFPY l

Cold Leg Cracking The first indications of cold leg cracking were found in 2R11 in SG A. Since no prior indications were available, the Millstone Point Unit 11 data was reviewed to i

determine a reasonable relationship between hot and cold leg trends. Since more cold leg cracking occurred at MPIl it was possible to examine both trends. It was determined that a three-parameter Weibull could be used to represent trends in both legs with a change in offset. For MPII this was 0.5 EFPY. To apply this to

)

ANO-2, the observed number of cold leg cracks was adjusted by the estimated 2R11 POD (87%). This number defective was then used to compute the approximate offset parameter (c) in the three-parameter Weibull model. For SG A this value is approximately 6 EFPY Since no Cold Leg indications were found in TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Projection of Number of Flaws.10

Tetr3 Engineenng Group, Inc.

O 1996 AIIRights Reserved.

SG B, the assumption was made that it is similar to the SG A Cold Leg but an additional 2 EPFY offset exists.

Current Cycle Predictions The most reasonable projections of numbers of circumferential cracks for a current cycle beginning at 11.4 EFPY and ending at 13 EFPY are:

Table 5 EOC 12 Projected Number of Circumferential Cracks Assumption SG A Hot Leg SG B Hot Leg SG A Cold Leg SG B Cold Leg 845 497 90 20 l

l TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Projection of Number of Flaws e 11

T;tra Engineenng Group, Ir:c.

C 1996 AIIRights R1 served.

NDE Sizing Uncertainty Uncertainty in the sizing of flaws is determined by comparing the results of the NDE tests with metallurgical exam results. The most preferable comparison is between NDE tests in an operating steam generator and then subsequent removal of the tube for destructive examination. Currently the destructive examination results from thirteen tubes removed from the ANO Unit 2 and similar style steam generators are available for comparison with the NDE results. Because of the rapid advancement of NDE technology, tubes that were removed from the steam generator only a few years ago were not tested with the techniques currently being used in the field today. This necessitates the use of some laboratory generated flaws with extrapolation to the field conditions.

Figure 8 shows the comparison of ECT measured PDA with metallographically determined PDA for the currently available data base of steam generator tubes.

Included in the data base are tube with laboratory generated flaws and tubes removed from operating steam generators. The 95%/95% confidence interval is also shown. This data includes Plus Point. 0.115 RPC and 0.080 RPC measurements.

Figure 8 Comparison of ECT PDA with Met PDA All Tubes All Probes 120 100 80

/

60

/

i

.. ~

.a.

20 I

fa 0

s 0

20 40 60 80 100

)

ECT PDA TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population NDE Sizing Uncertainty.12

T;tra Engineering Group, Inc.

O 1J96 AH Rights Reserved.

~

Plus Point RPC Probe The Plus Point RPC probe was used for the latest ANO Unit 2 steam generator tube inspection. It is also the probe intended to be useil for the next inspection.

Therefore the uncertainty associated with this probe sh >uld be used in any flaw size calculations. Figure 9 shows the comparison of Plus Point ECT PDA and Met PDA for all available data. The data includes both pulled tubes and laboratory generated specimens. The 95%/95% confidence interval and a 15%

1 PDA bound are also shown. A uniform distribution 15% PDA bound as selected for simplicity in Monte Carlo calculations.

Figure 9 Comparison of ECT PDA with Met PDA Plus Point Probes-Laboratory and Pulled Tube Data 90 n=MiintData -

80 15%.RM Bound ;

. l.kmar Regression'-

~

70

- 95% W & L.ow er Unt '

y-60

$ 50 f"

++

g p

e 40

/.

g,_,_,

r

=

- N 30

7 7.'

.,f

,9

,g, W 0 3 0

20 40 60 80 100 ECT PDA Probe Comparison It would be desirable to use only pulled tube data for determination of the Plus Point ECT uncertainty. However, because the technology is new, insufficient data currently exists. Figure 10 shows the correlation of ECT PDA with Met PDA for pulled tubes only, Various probe types and techniques were used to determine the ECT PDA. Most of the ECT data shown in Figure 10 was 0.115 an 0.080 RPC data, with only a small number of Plus Point values.

4 TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population NDE Sizing Uncertainty.13

Titra Engineonng Gs'oup. Inc.

C 1996 All Rights Reserved.

F6gwe 10 Comparwon of ECT PDA with Met PDA Puned Tube Data - AN Probe Types if f

/

E0 /

/

p E

<0 L.

A

/

/--

30 l

'o C-p

/j j.

0 10 20 30 40 50 60 70 80 90 100 ECT PDA Plus Point is known to have an improved detection and sizing ability for TFS circumferential flaws. To investigate this effect, all available data from laboratory specimens and pulled tubes (Maine Yankee) tested with both the 0.115 RPC and the Plus Point RPC were examined. The data shows a large bias between the 0.115 RPC and the Plus Point RPC. The 0.115 RPC tends to undercall the size while the Plus Point tends to overcall the size, especially for large flaws.

Figure 11 0.115 RPC and Plus Point RPC Paired Data Comparison 100 p'

90

/-

/

/

< 60 O 50

/

i a

g 7.

3 40 30 o 131 44=

t

. Plus W 20

,-. Adjusted 0.115 -

• ** l Plus W 8. Adjusted 0.115 10 a

0.115 0. --/ --

0 20 40 60 80 100 ECT PDA t

TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population NDE Sizing Uncertainty 14

1 Titra Engineering Group, Inc.

01996 AllRights Reserved.

A linear regression of each set of data was determined. The probe bias as a function of PDA was then calculated as the difference between the two regression lines. The 0.115 data could then be adjusted by the calculateo probe bias and a new regression of the adjusted data determined. The adjusted regression falls on top of the Plus Point data regression. This is shown in Figure 11.

Figure 12 shows a comparison between the pulled tube data (Figure 10) adjusted to account for the 0.115 / Plus Point probe bias and a regression of data from Plus point examinations only. The regression of the adjusted pulled tube data is nearly identical to the regression of the Plus Point only data.

Figure 12 Comparison of Plus Point ECT Correlation with Pulled Tube Data Adjusted for Plus Point Probe Bias

)

100 M.c '

-x FGleifTubiiiRegression

,j 90 Pused Tube Data.

-e1 p:

y

,, ' Adjusted Puned Tube Values 80 Rus W BCToorre -

M Adj Puged Tube Regression

/.-

l

~

50 7

e E 40

>j 30 20

,,,,x e -. -

10 0

0 20 40 60 80 1@

ECT PDA This effectively demonstrates that the Plus Point ECT sizing data acquired from laboratory generated crack specimens accurately represents the data acquired in the field when adjusted for probe type and technique. The Plus Point sizing correlation and uncertainty shown in Figure 8 c.an be expected to accurately represent the experience in ANO 2R11 inspction.

Plus Point Probe Bias Figure 9 above shows the comparison between metallographic determined flaw size and the Plus Point ECT determination. For large flaws, the regression of the data shows a tendency of the NDE to overcall the actual flaw size. Because of

. This is shown TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population NDE Sizing Uncertainty 15

T tra Engineering Group, Inc.

@ 1996 AllRights Reserved, in Figure 13. This is done to avoid potential inn-conservatism in the projection of highly degraded tubes.

Figure 13 Plus Point Probe Regression Fixed Through 100%

.... l -.

l.

I 7

-- l 90 e. Plus Pant Probe Data -

< -Regresa5xiFixed through 100%

{-

80

-~415% Erroi Band 1 ~ "

~-

q j

70 60 N

b 50

" 40 30

~~-

20,

10 0

10 20 30 40 50 60 70 80 90 100 i

ECT PDA TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population NDE Sizing Uncertainty.16

Tetra Engineenng Group, Inc.

C 1996 AllRights Reserved.

l Apparent Growth The apparent growth of circumferential cracks in steam generator tubes in the j

ANO unit 2 steam generators was determined by a comparison of the eddy current detemlined PDA from successive inspections. This comparison was made possible by the identification of a number of flaws that were inadvertently left in j

service due to limitations of the non-destructive test techniques and equipment at the time of the inspection.

Tables 6 and 7 provide the initial ECT PDA size and the apparent growth per EFPY for the 2P93 to 2R10 and the 2R10 to 2P95 the growth comparisons. The ECT probes used were the 0.080 pancake for the 2P93 inspection and 0.115 pancake for the other two inspections. The operation period between 2P93 and 2R10 was 0.8 EFPY. The operating period between 2R10 and 2P95 was 0.7 EFPY The apparent growth per EFPY was calculated as the difference between the two sizes nonnalized by dividing by the appropriate operating period.

Table 6 Apparent Growth per EFPY 2P93 to 2R10 Tube Initial Size Apparent Tube Initial Size Apparent Number (PDA)

Growth Number (PDA)

Growth (PDA/EFPY)

(PDA/EFPY) j R37L43 2.931 5.731 R31L39 23.587 13.767 R54L106 5.991 8.018 R70L106 23.717 7.980 R21L137 6.726 1.261 R24L130 23.909

-2.829 R17L139 7.187 32.463 R36L128 24.958

-12.244 R94LB4 10.026 1.757 R96L52 27.097 11.988 R68L104 10.620 21.557 R32L126 31.388 25.205 R75L81 13.317 9.979 R48L50 34.472 0.308 R19L53 13.473 7.148 R24L132 35.078

-17.947 R104L84 14.551

-2.230 R67L61 36.132 5.397 R64L60 15.533 7.874 R93L85 37.596

-13.512 R49L107 16.401 29.134 R82L82 39.620

-35.136 R77L129 18.221

-0.622 R111L73 48.318

-4.285 TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Apparent Growth = 17

Tetro Engineenng Group, Inc.

01996 AIIRights Reserved Table 7 Apparent Growth per EFPY 2R10 to 2P95 Tube Initial Size Apparent Tube Initial Size Apparent Number (PDA)

Growth Number (PDA)

Growth (PDA/EFPY)

(PDA/EFPY)

R78L76 1.408 0.792 R53L63 15.797

-0.047 R97L77 3.977 3.968 R23L35 17.561 18.121 R106L72 4.504 7.780 R27L53 21.561

-0575 R12L18 6.847 16.063 R66L64 22.320 2.790 R16L54 12.309 3.437 R14L26 26.010 1.299 R23L33 12.619 1.223 R14L30 36.725 5.235 In a number of cases in the 2P93 to 2R10 data set the apparent growth is a large negative number. At face value this would imply the crack is decreasing in size.

Since this can not occur physically, the negative apparent growth must be asweiated with NDE inspection technique. The change in probe types from the 0.080 RPC of the 2P93 inspection to the 0.115 probe of the 2R10 inspection is suspected as the cause. These large negative numbers are not seen in the 2R10 to 2P95 data set, where only the 0.115 RPC probe was used.

Figure 14 show the apparent growth data plotted as a function ofinitial PDA size.

A linear regression for each paired data set is shown.

Figure 14 Apparent Growth per EFPY Comparison of Inspections 40 2P93W2Rio.

{30 I 2R10 to 2P95.

(

a.

- - Unear (2P93 to 2R10)'

p 20 Lineer(2R10 to 2F95):

g Mg

..(,

.g I

i :,

~

.s f0

'h,.q.."pp.

j.

. 60 :

180-100

=

.1_

1 0

' ' 20 '

' 40 g -10 0

'N 1! -20

' -y g

g -30 s

s

-40 1

Initial Size (PDA)

TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Apparent Growth.18 i

I

m

_m Tctra Engineering Group, Inc.

O 1996 AllR4 hts Reserved.

Again, the 2P93 to 2R10 data is suspect. The increased sensitivity of the 0.115 probe would explain the large apparent growth for small initial size flaws. This effect would be seen even if there was no physical change in the flaw.

l The 2R10 to 2P95 data set provides a reasonable and realistic answer. A few flaws show 10% to 20% growth in PDA while the remainder show little or no growth. There are no large negative values. The 2R10 to 2P95 data set is shown in figure 15. Also show is a uniform 15% PDA error band.

This regression and error band is used in the Monte Carlo calculations.

Figure 15 Apparrent Growth per EFPY as a Function of Initial Size 2R10 to 2P95 Operating Period 25

AMo,6ime.fitRidicareer s

P mas.eoe nhperaFPY -

20.5 %

1s%PDA Unform Enor Band unh 0 LouerLimit g%

g g

$15 D--

~

j10 E5 N

,-h 4,

0

~

= - -- _ p = ---

0

.201

- 40/

80' 12;

-100

.j ;

. [-

Initial Size (PDA)

TR.96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Apparent Growth 19 i

T tra Engineenng Group, Inc.

C 1996 AllRights Reserved.

Monte Carlo EOC Distribution The projected EOC circumferential flaw dir sution is determined by a Monte Carlo combination of uncertainties in NDE sizing and flaw growth, reference 6.

Monte Carlo Process The Monte Carlo calculation process is used to extrapolate known flaw distributions obtained from short cycle lengths to the extended cycle length (16 EFPM). The as found distribution from the most recent inspection of the limiting steam generator is used as a starting point or base distribution. This distribution (SG A from 2R11) represents the known distribution of flaw sizes that have occurred for cycle lengths of 0.6 EFPY. This is then adjusted for NDE sizing uncertainty. This is done by randomly sampling flaw PDA from the base distribution. This represents an individual inspection result. NDE uncertainty and growth is then applied to obtain the EOC "true" size. This process is then repeated for a large number (>10') of trials and an Probability density function (PDF) representing the actual PDA distribution is obtained.

This distribution is then adjusted to account for potential growth in flaw sizes for j

the extrapolated cycle length. The applied growth term is based on the growth correlation shown in the previous chapter plus a randomly selected distribution of scatter about the correlation.

Monte Carlo Simulation Results The output of the Monte Carlo simulation is an essentially continuous normalized distribution. This distribution is multiplied by the number of flaws projected to be present at the end of the operating period. Figure 16 shows the Monte Carlo generated EOC distribution for ANO Unit 2 EOC 12.

TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Monte Carlo EOC Distribution 20

_. _... _ - _ - _ _ _. _ _. _ _.. _. _.. _ _ _ _ _.... _. ~ _ _ _..... _ _ _ _.. _.. _. ~. _. _ _. _.... _. _ _.. _.

)

j 20 Tetra Engineering Group. Inc.

@ 1996 AIIRights Reserved.

5 1

j Figure 16 j

Projected EOC 12 Circumferential Crack Distribution j

30 7

1 l

i I

i 25 j._..

-_i l

t

}

20 i

E l

8 15

_. _ _ __ _ -..------- - _ _ _ -.. _ _ _ _ l 4

E i

' 10 j I

2

?.

5

_J t

t 0:

M,,,,,m -,.-,.,,,.,,om,j j

0 5101520253035404550556065707580859095 j

PDA i

4 l

I i

A 1

I 1

4 t

j

.l 4

i i

f 1

4 1

l 1

TR.96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Monte Carlo EOC Distribution 21 l

Tetra Eryneering Group, Inc.

@ 1996 All Rights Reserved.

Conclusion i

The number and PDA distribution of circumferential cracks has been determined for EOC 12 for the ANO Unit 2 steam generators. The Monte Carlo simulation technique was used to account for known uncertainties in NDE detection and sizing, and apparent growth of flaws. This projected distribution will be used as an input to EOC 12 probability of burst and leakage calculations.

I

(

t l

I l

i i

J

?

i l

TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population Conclusion 22

=.

i N ntra Engineenng Group, Inc.

C 1996 AIIRights Reserwd.

4 1

References 4

1 1.

2. "Probabilistic Updating of Flaw Information", W. H. Tang, Journal ofTesting and Evaluation, Vol.1, November 1973, pp 459-467.

3. R. J. Kurtz, et. Al., " Steam Generator Tube Integrity Program / Steam Generator Group Project - Final Project Summary Report", NUREG /

CR-5117. May,1990.

4. " Main Results of the PISC III Exercise on Steam Generator Tubes Inspection", M. Bieth, C. Birac, R. Comby, G. Maciga,15th Annual Steam Generator NDE Workshop, Long Beach California, July 28-31,1996.

5. " Contribution of Pulled Tube Examination on NDE Techniques Evaluation",

R. Comby,15th Annual Steam Generator NDE Workshop, Long Beach California, July 28-31,1996.

6.

r i

TR-96-021 ANO Unit 2 EOC 12 Circumferential Flaw Population References e 23 4

m d

4 0

4

l 8

ATTACIIMENT 7 l

TR-95-025 i

NON-PROPRIETARY a

1 l

1 i

t e

l t

i t

4 1

1 i

1 i

1