ML031220526

Forthcoming Meeting with Pacific Gas & Electric Co. on 04/15/2003

ML031220526
Person / Time
Site:	Diablo Canyon
Issue date:	04/15/2003
From:	Pacific Gas & Electric Co
To:	Office of Nuclear Reactor Regulation
References
Download: ML031220526 (75)

Text

DCPP Unit 2 Steam Generators Operational Assessment for Voltage-Based Alternate Repair Criteria April 15, 2003 1

Introduction

• PG&E identified an unexpected ODSCC voltage growth during our last U2 refueling outage.

• We took conservative actions by plugging down to 1.2 volts bobbin and Plus Point inspecting everything above 1 volt bobbin.

• PG&E is convinced that full cycle operation of U2 cycle 12 is safe, and safer than performing a mid-cycle inspection.

• PG&E will demonstrate this conclusion in this meeting.

2

Agenda

• Background

• Pulled Tube Results

• Destructive Exam

• Effect on ARC Correlations

• POPCD

• General Methodology

• RAI Discussions

• ARC Method Conservatisms

• Methods to Determine Safe Cycle 12 Operation

• Applicability of ARC Correlation to DCPP Unit 2

• Conclusion 3

Background - NRC Interface

• 0.01 POB reporting threshold was exceeded for any operating period for Unit 2 Cycle 12 using a 0.6 POD

• POPCD for U2C12 requested in PG&E letter dated February 24, 2003

• POD of 1.0 for R44C45 requested in PG&E letter dated March 3, 2003 (LAR 03-04)

• March 4, 2003 NRC meeting Run time of 0.4 EFPY (140 days) was justified based on POD of 1.0 for R44C45, use of VDG, and assumed 2400 psi burst for R44C45 Low safety significance of exceeding 0.01 POB based on risk assessment

• NRC letter to PG&E dated March 10, 2003 stated that proposed amendment for POD of 1.0 for R44C45 is not needed for restart

• Unit 2 entered Mode 4 on March 23, started clock for 90 day report submittal 4

Background - 2R11 Plus Point Inspections and Tube Plugging

• Prior to March 4 meeting, Plus Point inspected 100% >1 volt bobbin indications

• After March 4 meeting, Plus Point inspected 100% <1 volt bobbin indications in SG 2-1 No outliers found that required additional plugging, where outlier is defined as large Plus Point to bobbin voltage ratio POPCD updated, insignificant affect

• 340 tubes plugged in 2R1 1 (269 TSP ODSCC)

• 1 12 tubes with TSP ODSCC > 2 volts bobbin (100%)

156 tubes with TSP ODSCC <= 2 and > 1.2 volt bobbin (100%)

1 tube with TSP ODSCC < 1.2 volts bobbin

• Total Unit 2 SG plugging is now at 5.7%

• Worst SG 2-4 is 9.3% plugged 15% plugging limit not exceeded W

5

Pulled Tube Destructive Examination

• R44C45, 21.5 Bobbin Volts

• R35C57, 5.09 Bobbin Volts Examination Procedure o Lab NDE o RT Leak Tests Performed with In-situ Equipment due to Potential of High Leak Rate o Test Plan included Testing at NOP and SLB Equivalent Conditions o Post-Leak Test Plus-point o Oxidation of Crack Faces via Furnace Heat Treatment o RT Burst Testing Performed w/ Bladder/Foil Installed o Tensile Testing of FS Areas for Material Properties 0 SEM of Crack Burst Faces 6

Results of Pulled Tube Destructive Examination

• R44C U

,45 3LB Freespan Leak Rate Testing o Leaked at NOP and SLB Equivalent AP's o SLB Leak Rate Measured in Lab: 0.88 gpm o Adjusted to 2405 for Database: 62.9 Iph (0.277 gpm) o Actual Leak Rate Added to As-Found Calculation for SG 2-4 Total Leak Rate RT Burst Testing o Actual: 4214 psi o Adjusted for Temp/Material Properties: 3841 psi o Normalized to ARC Database: 3757 psi o Satisfies 1.4xSLB Deterministic Criteria (3667 psi) o Removed from Final As-Found POB Calculation (Actual Burst > 2405 psi)

E F 7

Results of Pulled Tube Destructive Examination

• R35C57 SLB Freespan Leak Rate Testing o Started Leaking at -2200 psi and Continued at SLB Equivalent AP o SLB Leak Rate Measured in Lab: 0.0368 gpm o Adjusted to 2405 for Database: 0.54 Iph (0.0024 gpm)

RT Burst Testing o Actual Result: 5950 psi o Adjusted for Temp/Material Properties: 5423 psi o Normalized to ARC Database: 5006 psi o Also Satisfies 1.4xSLB Deterministic Evaluation No Credit Given in Final As-Found Calculations (Negligible Differences) 8

Condition Monitoring Update Based on Pulled Tube Results

• Condition Monitoring satisfied for Unit 2 Cycle 11 Addendum 5 correlations, APSLB = 2405 psi Pulled tube SG 2-4 R44C45 2H indication deleted from as-found leak and burst distribution because of known leak and burst results (0.28 gpm SLB leak rate and no burst at SLB)

Pulled tube SG 2-4 R35C57 2H indication retained in distributions SG2-4 o

POB: 3.84 x 10-3, less than 1% threshold value o

Leak Rate: 3.49 gpm, less than 10.5 gpm acceptance limit 3.21 gpm from Monte Carlo plus 0.28 gpm for R44C45

• Other SGs for Comparison SG 2-1 o

POB: 1.18x10-3 o

Leak Rate: 0.682 gpm SG2-2 o

POB: 5.66 x 10-4 o

Leak Rate: 0.362 gpm SG2-3 o

POB: 1.58 x 104 0

Leak Rate: 0.211 gpm 9

R44C45 Post-Leak Test O

s A

t S

10

Burst Test Appearance, Tube 44-45 11

Burst Test Appearance, Tube 35-57 12

SEM Fractographic Montages 35-57 Top, 44-45 Bottom P"-,

.oft-,

- I.

13

Measured Depth Versus Length Profiles Tube 35-57 0.

0 0

C-0 01 02 03 04 05 06 0.7 Axial Length, inches 08 Tube 4445 0.!

a U

-.)

100 90 80 70 60 50 40 30 20 10 0

0 0.1 02 03 04 05 06 07 Axial Length, inches 08 I IVICII I 14

Comparable Degradation From Bobbin ARC Database 4£

/

ay,,4

,(ao SK dr 7Yr,^XSJj'-%

y ~ r 4,.

,44 4

4 e£

W f

4i-4<

z 4;

44 44%'

At a

p ;,

t

_~,,

451 S

t I 4 4 4

4 4

"44 15

Tube Pull Effect on Correlations - Introduction

• Destructive examination of two (2) pulled tube sections from Diablo Canyon 2 results in data for the ODSCC ARC database.

• One indication at 21.5 V and one at 5.09 V.

• Burst Pressure, Probability of Leak, and Leak Rate as a function of the bobbin amplitude.

• Burst Pressure -

No significant effect.

• Burst pressure of freespan sections about 12,000 psi.

• Probability of Leak -

No significant effect.

• Leak Rate -

No significant effect.

• The p-value is < 5% for a AP of 2560 psi.

• The p-value is 1 % for a AP of 2405 psi.

16

Burst Pressure -Tolerance Bounds Burst Pressure vs. Volts for 7/8" Alloy 600 SG Tubes Additional Data, Reference ajf = 68.8 ksi @ 650'F 12 10

-4 co P-4 4Co Coa)4 Co 8

6 4

2 0

0.1 1

10 Bobbin Amplitude (Volts) 100 17

Burst Pressure -

Effect on Regression Burst Pressure vs Volts for 7/8" OD Alloy 600 SG Tubes Reference Database, Reference af = 68.8 ksi @ 650'F 12.0 10.0 Co co Co C)

C4 8.0 6.0 4.0 2.0 0 0 0.1 1

10 100 Bobbin Amplitude (Volts) 18

Burst Pressure -

Effect on POB 1E+00 lE-01 1E-02 1E-03

$-4

z 0

1E-04 1E-05 1E-06 1E-07 Probability of Burst of 7/8" SG Tubes During SLB, Effect of Additional Data, AP = 2405 psi

--- PoB(2405 psi) Addendum 5

-PoB(2405psi)w/New Data

_ Structural Limit There is no meaningful change in the POB at any voltage level.

I -

I____

_____1 1

1 1E-08 1E-09 lE-10 lE-1l 0.1 1

10 100 Bobbin Amplitude (Volts) 19

Burst Pressure - Regression Verification Burst Pressure for 7/8' Diameter Tubes Residual Analysis of Plant Data

• The scatter plot of the data continue to show no dependence of the residuals on the predicted values.

a

'4 M

W a

0~

a E

4) 01 S

0.

a Burst Pressure for 7/8" Diameter Tubes Residual Analysis of Burst Pressure Data 32 2 0 3 Burst Pressure Data 0

I-Normal Distribution 1 5 1 0 0 5

0.

05 -

1 0

-2 0 30 30 35 40 45 50 55 60 66 70 75 80 85 90 95 100 106 Predicted Burst Pressure (ksi)

• There is no apparent trend in the data to indicate a departure from normality of the residuals.

• Numerical evaluation of the EdF data is not necessary, there will be no change in the conclusion.

30 25

-20 1 5

-10

-05 00 05 10 15 20 25

.30 Observed Residual Burst Pressure Values (ksi) 20

Probability of Leak -

Regression Curve POL=99%

Probability of Leak vs. Bobbin Amplitude

/

7/8" x 0.050" SG Tubes, Effect of Inclusion of New Data

/

100%

• 1

,-1II-4j-L zinn En KN fl I21 fl I

I I 111111 I II1111 F
;
i

I VTI M l I A

90% -

80% -

70% -

I I

I I lI I I n Add. 5 Data A New Data Addendum 5 Addendum 5+

I T 1111 I

I I II I I I l Ii I

I I

I 1 1 1 1 '

I OL I

I I

I I

I I

1 1111 X

I I

I i

i al l

I I I I I I I A

I POL = 52%

I r 1 1.

I Both indications leaked.

I l

I I I 11111 I T 1111T 14M 0

~0

~Q s0 i

I 1I i

-- 1tl -- I f I--1-tt1 I

I I lI I 11 I I I lI II III 60% -

50% -

40% -

30% -

I I

I I

I I

' I



I I I I I III I I I II I I"

The POL of the indications is not at variance with the expectations from the ARC curve.

I I

I I

I I I

I I

I I I 1lI I

I IIII I

I I

I It 20%

10%

. I.

I.

I I I I I I.

I I I 1 1 1 1 I

I I 1 1 I

- I I 1 1 n O4.1 I

I I

I I

I~

JJ-rhr.A U

I

.. i.~i

~ ~

f*

%.WV

{

d 0.0 0.01 0.1 A_.

Ha 10 1

Bobbin Amplitude (Volts) 100 21

POL -

Effect of Added Data

• There is a slight increase in the POL for indications greater than 0.25 Volts.

Ratio of POL Predictions, w/ to w/o New Data 7/8" x 0.050" Alloy 600 MA SG Tubes O

• Peak increase of 8%

at about 2.5 V.

• Increase of about 1 %

at 10 V.

• Negligible effect at 20 V.

-0 0R 0

I RatIofIoI 11

-I I

-Mii I I I HI 01 I

10 100 Bobbin Amplitude (Volts) 22

Leak Rate -

Effect on Regression SLB Leak Rate (2405 psi) vs Bobbin Amplitude 7/8" x 0.050" Alloy 600 MA Tubes Data 1000 100

=,

10 P4 ct 1

0.1 0.01 1

10 100 Bobbin Amplitude, V (Volts) 23

Leak Rate -

Overall Effect SLB Leak Rate (2405 psi) vs Bobbin Amplitude 7/8" x 0.050" Alloy 600 MA Tubes Data 1000 The Diablo Canyon data strengthen the database.

0 Add. 5 Database A

Added Data Add. 5 Expected Add. 5+ Expected

_=

=

1

_ _I I

I

-t 4--+-+-

+

4 4

4--4--4----4----4--

100 -

U I

-r lici$M

(3

6

_4_

C

W4 14 C

a)
4 10 1

The Leak rate predictions

___ldecrease for < 9 V indications.

I_

_l

/ O

/

__________3_

0.1 0.01 1

10 Bobbin Amplitude, V (Volts) 100 24

Tube Pull Effect on Correlations - Conclusions

• The Diablo Canyon 2 data are in accord with expectations relative the statistics of the database.

• Burst Pressure

• The data do not depart significantly from the database.

• The regression curve is not significantly affected.

• There is no change in the conclusion regarding EdF data.

• Probability of Leak

• The data agree with expectations.

• Leak Rate

• The data strengthen the database.

• The expected leak rate from ODSCC indications decreases when the Diablo data are included in the database.

• The p-value is < 5% for a AP of 2560 psi and 1% for a AP of 2405 psi.

25

POPCD Assessment for Cycle 12 Operation

• DCPP POPCD supports full cycle operation

• POB = 0.00745 and <0.01 reporting threshold

• SLB leak rate = 3.06 gpm

• Analysis conditions

• Addendum 5+ database with DCPP 2R1 1 pulled tube results included in database

• Voltage dependent growth with binning based on statistical methodology presented at March NRC meeting APSLB = 2405 psi

• GL 95-05 methods with non-VDG growth and POD = 0.6 yield burst probability of 0.040 and SLB leak rate of 5.54 gpm 26

POPCD Assessment for Full Cycle Operation Case APSLB Growth POD EFPY POB

<1% POB for LR full cycle?..tgpm) 1 2405 VDG POPCD 1.54 7.45 E-03 Yes 3.06 2

2405 VDG 1.0 for 1.54 7.94 E-02 No, 0.53 9.6 R44045 EFPY for 1 %

3 2405 VDG 0.6 1.54 9.79 E-02 No 10.02 4

2405 Non VDG 0.6 1.54 4.02 E-02 No 5.54 I

27

POPCD Discussion Topics

• POPCD Description

• Updated DCPP POPCD for Final Inspection Results

• Responses to NRC POPCD Questions

• Added Conservatisms in ARC Analyses

• Offset excessive and arbitrary conservatism of POD = 0.6

• Voltage dependent growth when need indicated

• Potential industry methodology to provide allowance for large growth rates o Applicable only with generic POPCD - not appropriate in combination with conservatism of POD=0.6 o Not applicable to Unit 2 Cycle 12 because outlier present in existing analysis 28

POPCD Description

• POPCD is calculated as the ratio of indications reported at the prior inspection to the total indications found at the subsequent inspection (all indications reported in the prior cycle plus new indications).

• POPCD for a given voltage bin at the prior end of cycle inspection (EOCn) is defined as:

EOCn+1 RPC confirmed plus not RPC inspected and detected at EOCn EOCn RPC confirmed

+ plus not RPC inspected and repaired at EOCn POPCD =

Numerator New EOCn+1 RPC

+

confirmed plus not detected indications (i.e.,

not detected at EOCn) 29

POPCD Description

• POPCD based totally on field ARC inspection experience

• POPCD based on premise that all structurally and leakage significant indications are detected at EOCn+1

• Successive ARC inspections support this conclusion

• All indications historically found to challenge tube integrity were reported in a prior inspection

• All new indications assumed to be undetected at prior inspection

• POPCD conservatively accounts for newly initiated indications (difference from traditional POD) in operational assessments as well as previously undetected indications

• Prior cycle voltage for new indications based on reanalysis of prior inspection data with the knowledge that an indication is present 30

Final DCPP POPCD Results

• Minor update to DCPP POPCD data based on extending SG 21 +Point inspection to lower bobbin voltages

• No discernable change in POPCD distribution

• DCPP POPCD data (4419 indications) available from 5 inspections including 2R1 1

• DCPP POPCD obtained for loglogistic fit including lower 90% confidence distribution

• Comparison of DCPP and Industry POPCD

• DCPP POPCD a little higher above 1 volt and moderately lower below 1 volt

• POD about 0.99 at 4 volts I

31

DCPP Final POPCD Data

-U

'tnp nj~nfr'r ri-i-ji~lA lull i )~

a

-DA n,

Ioi 1 nflt,u I aule L. Indrr r iial r

-,Lu II, I III.-

U. II<.

v.

I New Indications Bobbin Call in Both Prior Insp.

POPCD InsNd ntions RPC Confirmed RPC Confirmed + Not Inspected Voltage RPC RPC RPC.

Bin RPC Confirmed RPC Confirmed Confirmed Adjusted Confirmed Plus Not Confirmed Plus Not and Frac.

Count Frac.

Count (50%)

Median Inspected Inspected Plugged (50 1) 0.01-0 20 110 517 16 59 3

0 147 19/129 0 107 62/579 0.106 0106 0.21-0 40 232 1123 120 594 20 0.376 140/372 0.353 614/1737 0.353 0 353 0 41-0 60 105 410 160 616 15 0 625 175/280 0 606 631/1041 0.606 0 606 0 61-0 80 48 138 131 366 7

0.742 138/186 0.730 373/511 0.729 0 729 0.81-1 00 20 39 97 203 3

0 833 100/120 0.841 206/245 0.838 0.838 1.01-1 20 9

18 53 91 3

0 862 56/65 0.839 94/112 0 834 0.838 1.21-1 40 4

9 51 78 0

0927 51/55 0.897 78/87 0 889 0.889 1.41-1 60 1

1 22 33 1

0 958 23/24 0.971 34/35 0 953 0.953 1.61-1.80 0

0 19 21 0

1.000 19/19 1 000 21/21 0 968 0 968 1.81-2.00 0

0 18 19 0

1 000 18/18 1 000 19/19 0964 0968 2.01-2.20 0

0 0

0 5

1.000 5/5 1.000 5/5 0 871 0.968 2.21-2 50 0

0 0

0 5

1 000 5/5 1.000 5/5 0.871 0.968 2.51-3.20 0

0 0

0 10 1.000 10/10 1 000 10/10 0933 0968 13.21-3 50 0

0 0

0 3

1.000 3/3 1.000 3/3 0.794 0 968 3.51-5 00 0

0 0

0 6

1 000 6/6 1.000 6/6 0 891 0 968

>5 00 0

0 0

0 3

1 000 3/3 1 000 3/3 0 794 0.968 Sum (2) 0 0

18 19 32 1.000 50/50 1.000 51/51 0.987 0.987 Total 529 2255 687 2080 84 Notes 1: Median adjusted to maintain POD constant or increasing based on expected POD behavior rather than statistical lower bound 2' Averapoe volts = 2.65 based on mid-bin volta2e weighted bv number of indications per bin 5.25 volts assiged to > 5 volt bin 32

DCPP Comparison of Final and NRC Submittal POPCD DCPP POPCD Distribution - Comparison of Final POPCD with NRC Submittal Loglogistic Fit to Adjusted Median Data of Final POPCD Data (3/21/03) 1.0 0.9 U

00 0

04

.0 0

c).

cw 9o 0

t, 0.8 0.7 0.6 0.5 0.4 0.3 0.2

,/P I

I Im~

4

+DCPP POPCD Final Nominal_

= If-_

... _.. DC PP POPCD NRC Nom inal-

=

I I I X~~~~

==I

==a 0.1 0.0 0.1 1.0 Bobbin Coil Volts 33

Industry POPCD Benchmarking

• Response to Question 9 of 1/99 NRC RAls

• 32 benchmarking cases plus 17 sensitivity cases

• Compared OA projections with CM results for EOC voltages, burst probabilities and leak rates

• For a leak rate acceptance criterion of 0.25 gpm (generally <5% of acceptance limit), POPCD projections in agreement with CM for 31 of 32 SGs

• Exception was a high voltage growth indication for a previously detected indication

• For a burst probability acceptance criterion of 5E-04 (5%

of limit), projections agree with CM for 30 of 32 cases

• Exceptions were the above high growth indication and hiWh arowth rates found for delumed tubes WI I r7r0 34

DCPP Benchmarking Analyses OA projections for Cycle 1 1 compared to CM analyses for all 4 SGs

• Applied Cycle 1 1 voltage dependent growth rates to assess POPCD vs POD of 0.6 rather than growth

• POPCD projected leakage > CM for all SGs and 4% > CM for limiting SG 24 compared to POD = 0.6 over-estimate by about 70%

• Burst probability > CM for 3 SGs and underestimated by 0.0019 (8%) for limiting SG 24 compared to 0.6 overestimate by about 60%

o POPCD underestimation can be eliminated by applying conservative 5% uncertainty to 21.5 volt flaw instead of ARC 1 0%

35

POPCD Benchmarking Results

I SG Calc Description POB Leak Rate EOC 11 As-Found 1.18E-03 0.682 SG 2-1 Projected EOC 11 with SG 2-1 Cycle 11 VDG 1.15E-03 0.763 2R1 1 As-Found 5.66E-04 0.362 SG 2-2 Projected EOC 11 with SG 2-2 Cycle 11 VDG 6.04E-04 0.361 2R1 1 As-Found 1.58E-04 0.211 SG 2-3 Projected EOC 11 with SG 2-3 Cycle 11 VDGI 1.95E-04 0.261 2R1 1 As-Found 2.38E-02 3.72 SG 24 Projected EOC 11 with SG 2-4 Cycle 11 VDG 2.19E-02 3.87 Projected EOC 11, POD=0.6, Prel. VDG 3.96E-02 6.42 There were no SG 2-3 indications in the upper VDG bin (>1.66v). Therefore, the upper bin used a composite of all SGs, except for R44C45 in SG 2-4.

M nU 11781 36

POPCD Conclusions

• POD of 0.6 results in overly conservative estimates of number and severity of non-detected, large indications Severity of conservatism increases progressively as indication voltage increases above 1 volt

• All POPCD and POD assessments for axial ODSCC support a high probability of detection above 1 volt and approach unity near 4 volts

• DCPP POPCD results are supported with minor differences between PODs by EPRI and ANL, as well as the industry POPCD analyses 37

POPCD RAI Discussions 38

ARC Analysis Conservatisms for POPCD Applications Objective

• Replace POD of 0.6 which does not adequately address growth issues

• Provide analysis methods to accurately reflect the fact that growth rate issues dominate under predicting burst pressures and leak rates, rather than POD effects

• Proposed Conservatisms in ARC Analyses

• Voltage dependent growth when need indicated by growth rate data

• Potential industry methodology to provide allowance for large growth rates o Applicable only with generic POPCD - not appropriate in combination with conservatism of POD=0.6 o Not applicable to Unit 2 Cycle 12 because outlier present in existing VDG analysis 39

Conservatism Added with VDG

• Voltage dependent growth rates increase burst probability and leak rates above the GL 95-05 guidelines for voltage independent growth rates Consistent results when growth found to show dependence on BOC volts

• Voltage dependent growth provides technically correct conservatism to ARC analyses that offsets arbitrary conservatism of POD = 0.6 for plants with large indications that could potentially challenge tube integrity 40

Comparisons of EOC Projections with and without VDG (POD = 0.6)

Plant/SG Cycle Growth POB SLB LR (gpm)

DCPP-2/SG 24 12 VDG 9.79E-02 10.0 Non VDG 4.02E-02 5.54 AA-1/SG C 6

VDG NA(locked) 10.6 Non VDG 7.8 AB-1/SG C 8

VDG NA(locked) 36.4 Non VDG 19.0 A-1/SG C 15 VDG 1.37E-02 8.3 NonVDG 8.02E-03 7.8 I

41

Introduction - Extreme Values of ODSCC ARC Amplitudes & Growth

• ARC have been in use for ODSCC indications in SG tubes since the early 1990's.

• One ARC reporting threshold is that the POB of all of the TSP ODSCC indications in a single SG should not exceed 0.01 during a SLB event.

• A 21.5 V indication at Diablo Canyon 2 resulted from growth during the previous plant operating cycle of 11.9 V/EFPY.

• The presence of this indication led to the conclusion that the POB for one or more indications was greater than 0.01.

• The next largest voltage growth was found to be 3.0 V/EFPY.

• The magnitude of the maximum observed growth is relatively rare and was not predicted from the prior cycle analysis.

42

Phenomenological Reason

• The SG 2-4 R44C45-02H indication was throughwall or nearly so at the prior inspection.

• Voltage increases exponentially with depth, but is approximately linear for shallow depths.

• The exponential nature of the dependence accelerates with decreasing ligament thickness.

• Voltage increases exponentially with throughwall length after 100% depth is achieved.

• Information presented and discussed in detail at the last meeting.

43

Examples of Growth Dependence Bobbin Volts vs. Depth for EPRI ODSCC Database 10t I

i I

I 3/4-Diameter Tubes A 7/8" Diameter Tubas

... Expon (3/4" Diameter Tubes)-

Expon (7/8" Diameter Tubes) lT --

l-I I

I I

I Iii.

• Exponential dependence of volts on depth illustrated at left.

• R44C45 was about 1 00%

deep at the last inspection.

tH o

a

.2I SI1 l I I

A A A A A

-AK--

I 0%

10%

20%

o%

40%

50%

60%

70%

Destructive Examination Maximum Depth (%)

80%

90%

10%

t0 3C

• Exponential dependence of volts on 100% throughwall length is illustrated at right.

iZ coI

• R44C45 had a 100% TW length on the order of 0.5".

000 0 10 020 030 040 Destructive Examination 100% TW Length 0 50 0 60 44

Performance of Predictive Methods

• Why are we having this discussion?

None of the methodologies employed for predicting the growth of indications remaining in service led to the expectation of a value on the order of that observed.

o Singularly large growth values have been observed at several plants during the 10-year application history of the ODSCC ARC.

o While the observation at Diablo Canyon 2 would be considered to be rare, it is not unique.

• The historical approach to compensating for the lack of a direct prediction model for large growth rates was to use a uniform POD of 0.6 per GL 95-05 to leave larger indications in service.

• There is a desire to quantify the likelihood of outliers.

• Singularly large values can be termed as extremes of growth or as the appearance of outliers.

45

Maximum Voltage Growths per EFPY

• In over 35 applications of the ODSCC ARC been few extremes or outliers.

there have Volts / EFPY Tube Size Plant 7.0 7/8" Al 8.1 3/4" AAI 8.1 3/4" AB1 8.6 3/4" AC1 9.6 3/4" AAI 11.9 7/8" Y2 46

Illustration of Growth 25 20 1 ----- --

The challenge is to develop a methodology for predicting the extremes or outliers of the growth distribution for amplitudes that are less than 2 Volts.

a1.

0 C,

15 10 l

[

9

• A WI4 4

4 4

,I* 4 4

4 4

4 4- -

5 0

.5 1

I W

I.

I 0

2 4

6 a

10 BOC Voltage, Volts Note: Voltages Corrected for Different Calibration Procedures and Growth Rates Corrected for temperature 12 18,

-I 47

Illustration of Growth 8

0

-SG-C (All Indications) 0

• The distribution of growth 7

Largest Growth in SG-A at each level of initial (3rd Largest in All SGs) amplitude is significantly 6

0 skewed.

0 5

0 Skew is expected.

0 0

The lower bound to 0

0 0

0 0

NDE error.

Tail length weakly 2-0 o 00 dependent on initial 0

0 o ) OoO o 0 Ooc lamplitude.

0-,

• The chance of extreme 0 00 0O 0 growth is remote.

00

-I 0

05 1

15 2

BOC Voltage 48

Growth of Indications in DCPP SG 2-4 Growth of ODSCC ARC Indications in SG 2-4 Diablo Canyon 2R11 12.0-11.0 a SG 2-4 Data 10.0.

Z'9.0 8.0..

Overall, the growth of the 7.0-indications is not characterized 6.0 -

by anything remarkable.

5.0 R44C45 is an exception.

4.0 3.0 -

2.0 -

3 1 1.0 -I L

01.0~

0.0 0.5 1.0 1.5 2.0 2.5 Previous Amplitude (Volts)

I 49

Diablo Canyon SG 2-4 Growth Distribution

• The distribution of growth values for indications that exhibited positive growth fits a lognormal distribution.

Distribution of SG 2-4 Voltage Growth for Indications that Exhibited Positive Growth During the Cycle 20%

I IT A_

I I

l I

I I

I I

The probability of occurrence of a growth of 11.9V is 2.8E-5. For a sample size of 900 positive growth values, the probability of observing growth of that magnitude is 2.5%.

I The chart demonstrates that a Lognormal distribution may be used L to characterize the probability of occurrence of larger growth values.

100%

90%

15% I-U)M 0

.Z:

O 10%

CZ co c0 0

r.)

-44 5%

0%

J =~

Observed PDF

-Observed CDF LogNormal T I I

-80%

- 70%

-60%

- 50%

-40%

- 30%

- 20%

a 0

r.)Z

.0 t.4 a)

CZ Htt fi 1In1H1 10%

0%

I...

I..

i I. "

I.

I I I' I I ' I I I

'I I

.I Ci C"

1-0 0O c c'i

.4 CD In

_40 00 t

co o

d o

Co o

o 0

CD 00 0

Common LOG of Voltage Growth per EFPY 50

Diablo Canyon SG 2-4 Growth Statistics

• The probability of a growth of 11.9 V is 2.8 10-5 for a single indication.

• For 900 indications in the SG the probability of at least one indication exhibiting a growth of > 11.9 V is 2.5%

• Probability of observing an indication with this level of growth in four SGs with a similar number of indications is 10%.

• Expect such a growth to occur about once in every five or so inspections.

• The industry experience has been to observe an outlying growth on the order of once in six inspections.

I 51

Multifaceted Methodology

• Start with a good predictor of the number and bulk of the distribution of the indications.

• An industry methodology exists for predicting the number of indications and their severity based on using information available at the current inspection to update the POD at the previous inspection.

• This is then used predict the number to be found in the following inspection.

• POPCD has been demonstrated to agree well with inspection data.

• Add an independent methodology to deal with the likelihood of extreme or outlying growth values.

b2

Introduction to Extreme Values

• Explain extreme values arising in observed samples and to forecast extremes that may be expected to occur.

• An extreme is a maximum or a minimum value.

• May also deal with 2nd or 3rd largest or smallest, etc.

• Exceedances arise when dealing with design requirements.

• Specification of design requirement, e.g., seawall height.

Predictions made of cycles between exceedance values.

Probabilities calculated regarding the number of cycles before repeating an exceedance.

• Distributions of maxima or minima of a distribution are not the same as the distribution itself.

Four basic extreme value distributions, characterized by bounding conditions, are possible.

Extreme values are associated with specific data distributions.

53

Contributing Factors

• There are a number of potential contributing factors to be considered in the development of the method.

• Geometry is a factor in looking at the historical database because of different tube sizes.

• Number of indications in an inspection increases the chances of seeing a large growth value.

• Operating conditions may influence the growth of the indications.

o Length of operating cycle o Operating temperature o Number of shut down / start up cycles between inspections.

• The total time of operation and the age of the indications in the SG tubes.

I i

54

One Approach Using Extreme Values

• Accumulate data from ARC applications

• Segregate the 5 largest voltages and growths from each

• Normalize growth data to a reference temperature

• Calculate the extreme statistics.

• Simulate historical data using extreme value statistics.

• Determine the predicted number of outlying observations

• Modify as necessary to account for other variations

• Integrate probabilities of outlying observations into SG Monte Carlo simulations.

• POB less than 1.10-2 may be expected for any one cycle.

• Predicted POB and leak rates will increase slightly due to the inclusion of outlying growth rates.

55

Summary -

The Industry Approach is TBD Analysis of the statistics and apply extreme value distributions.

Simulate return periods for exceedances above a specified level.

Perform Grubb's test for outliers for approximately normal data.

max log(Vi ) - log(V)

N - 1 t(al[2N], N-2) s W

N

+ (a/l[2N],N-2)

Growth data may be approximately lognormally distributed for indications exhibiting growth.

R44C45 is at outlier at a 95% confidence level.

Perhaps a Grubb's statistic can be simulated.

DCPP 2 Cycle 12 VDG includes its own extreme value; approval of industry method not required for Cycle 12 56

Alternate Methods for Demonstrating Safe Cycle 12 Operation

• Issue: Demonstration of acceptable EOC-12 burst probability is an analysis issue and not a real safety issue

• Plates locked due to denting and packed crevices prevent burst although not implementing mechanical locking until 2R12

• Application of a realistic POD, such as POPCD, demonstrates acceptability of full cycle operation without credit for locking

• Reduced potential for large growth by plugging at 1.2 bobbin volts

• Risk analyses show full cycle operation is low risk significant and lower risk than full cycle operation with a mid-cycle inspection 57

Alternate Methods for Demonstrating Safe Cycle 12 Operation

• Alternate methods assessed for Cycle 12 operation

• POPCD o Recommended method

• POD = 1.0 for > 3 volt indications o Interim alternative to POPCD

• Exclusion of R44C45 growth rate from analysis o Preventive repair to 1.2 bobbin volts substantially reduces likelihood of large growth rate

• Operator actions to lower APSLB per Par. 6.a.3 of GL 95-05

• Risk analysis to supplement methods if resulting burst probability above NRC reporting threshold of 0.01 I O&1 I 58

SG 2-4 OA Sensitivity Results with Updated Correlations (All Results)

Case APSLB Growth POD EFPY POB

<1% POB for EFPY for LR full cycle?

> 1% POB (jpm) 1 2405 VDG POPCD 1.54 7.45 E-03 Yes 3.06 2

2405 VDG 1.0 for R44C45 1.54 7.94 E-02 No 0.53 9.6 3

2405 VDG 0.6 1.54 9.79 E-02 No NR 10.02 4

2405 Non VDG 0.6 1.54 4.02 E-02 No NR 5.54 5

2405 VDG 1.0 for >3 volts 1.54 4.82 E-02 No NR 6.94 6

2405 Non VDG 1.0 for >3 volts NR NR No 1.01 NR 7

2405 VDGw/o 1.0 for R44C45 1.54 2.06 E-02 No NR 8.11 R44C45 8

2405 VDG w/o 1.0 for >3 volts 1.54 9.58 E-03 Yes 5.90 R44C45 9

1900 VDG 1.0 for R44C45 1.54 1.74 E-02 No NR NR 10 1900 VDG 1.0 for >3 volts 1.54 1.4E-02 No NR NR 11 2150 VDG 1.0 for R44C45 1.54 3.81 E-02 No NR NR 12 2150 VDG 1.0 for >3 volts 1.54 2.28 E-02 No NR NR 59

Bases for Alternate Assumptions

• Application of POD = 1.0 above 3 volts supported by No missed indications greater than 3.2 volts in industry No missed indications greater than 1.6 volts at DCPP

• Bases for exclusion of R44C45 growth in OA analyses

• R44C45 had a 2.0 bobbin volt and 2.97 +Point volt indication at BOC-1 1 o Conditions excluded by preventive plugging to 1.2 volts and extensive +Point inspections

• Very low probability of another very large growth indication o Largest growth rate observed for 1.2 volt flaw was 2.48 volVtEFPY 60

Bases for Alternate Assumptions

• Operator actions can potentially prevent primary pressure from reaching PORV setpoint

• Permitted in ARC assessments per Par. 6.a.3 of GL 95-05 for probabilistic assessment of operator actions

• Sensitivity analyses performed for assumed SLB pressure differentials of 1900 and 2150 psid

• Potentially capable of controlling APSLB <1900 psid based on recent simulator run

• Requires timely manual control of intact SG ADV

• 1900 psid APSLB reduces POB by a factor of 4

• PG&E evaluating EOP changes for optimum response to all events 61

Cycle 12 OA Conclusions

• DCPP POPCD supports full cycle operation

• NRC approval requested for 2R12 application

• Risk assessment not required

• POD of 1.0 above 3 volts plus exclusion of R44C45 growth also supports full cycle operation

• All other scenarios yield EOC POB>1 % and require risk assessment for balance of cycle 12 62

Risk Assessment

• Objective

• Assessment of risk significance for operating a full cycle following 2R1 1, leaving in service SG tubes with ODSCC indications <1.2 bobbin volts

• Assessment of "Full Cycle Operation without a Mid-Cycle Inspection" vs. "Full Cycle Operation with a Mid-Cycle Inspection"

• Why?

• To satisfy GL 95-05, Section 6.a.3 and our Tech.

Spec. Requirements.

• In relative terms, PG&E believes that full cycle operation without a mid-cycle inspection is the safest option based on a scoping risk assessment.

63

Risk Assessment

• Relative Risk Significance Evaluation is

• Consistent with the NRC's consideration of Safety for Notice of Enforcement Discretion (NOED) Requests.

• Consistent with NUREG-1 024, "Technical Specifications -

Enhancing the Safety Impact," concern that:

0 Allowable outage times that are too short will subject the plant to unnecessary trips, transients, and fatigue cycling.

• Consistent with RG 1.177, "An Approach for Plant-Specific Risk-Informed Decision making: Technical Specifications,"

which requires evaluation of a proposed change on the total risk.

• Risk Significance Criteria-Using NUREG 1.174 Criteria

• Large Early Release Frequency (LERF) is used to determine absolute risk significance.

Core Damage Frequency (CDF) is used to determine relative risk significance.

64

Risk Assessment

• Results of Scoping Risk Assessment Investigation For "full cycle operation without a mid-cycle inspection" LERF and CDF are in 1.OE-9/y range (Based on earlier reported PG&E evaluation)

For "full cycle operation with a mid-cycle inspection":

o For the first 9 months (mid-cycle), CDF is in 5.OE-10/y range o CDF for transition to shutdown for inspection is in 5E-8/y range (Using WCAP-14333-P-A and DCPP specific CCDP) o Mid-loop CDF is in 1.OE-7/y to 5.OE-8/y range (Using ORAM model) o For the second 9 months, CDF is in 1.OE-1 O/y range.

o Total CDF is in the range of 1.OE-7/y to 2.OE-7/y 65

Risk Assessment

• Conclusions Both "full cycle operation without a mid-cycle inspection" and "full cycle operation with a mid-cycle inspection" ARE low risk significance.

• Relatively, "full cycle operation without a mid-cycle inspection" is safer than "full cycle operation with a mid-cycle inspection."

• "Full cycle operation without a mid-cycle inspection" provides the additional safety benefit of averting the transition risk and fatigue cycling associated with shutting down the unit.

I 66

Demonstration of Applicability of ARC Bobbin Voltage Correlation to DCPP Unit 2

• Tube pulls resulted in no significant changes to burst and leak rate correlations

• Degradation morphology at DCPP Unit 2 consistent with bobbin ARC database

• RPC confirmation rates of bobbin ARC indications at DCPP Unit 2 = 90%

• RPC confirmation rate of bobbin ARC database indications

= 96%

• Crack profiles of highest Plus point voltage indications left in service evaluated to assure that wall penetration and associated high bobbin voltages will not develop during next cycle of operation

• Burst pressures calculated from these NDE profiles led to further evaluations by multiple approaches

• Burst pressure evaluation demonstrates applicability of bobbin voltage ARC to DCPP Unit 2 67

Demonstration of Applicability of ARC Bobbin Voltage Correlation to DCPP Unit 2

• Largest Plus point voltage indications were profiled and RTS for Cycle 12 (Top 20 overall and top 5 in each SG, -total 28)

• OA Analyses presented at March 4th meeting

• Additional ODSCC indications profiled, including:

• All indications with >1.5 volts Plus point

• All repeat indications

• Additional indications in SG 2-1 (data was taken after March 4 meeting)

• Most of these profiled indications have a large Plus point to bobbin voltage ratio (175)

• Wanted to further assure that no deep indications were being RTS for Cycle 12

• Additional sizing performed for small Plus point to bobbin voltage ratio (159) 68

Plus Point to Bobbin Voltage Comparison for ODSCC at TSPs DCPP-2 All SGs 2R11 6

5 -

4.

a) 0 c3

.5 0

1 0

0 1

2 3

4 5

6 Bobbin Voltage

-7 I kral; 69

Plus Point to Bobbin Voltage Comparison for ODSCC at TSPs 2R11 Indications Sized by Plus Point 6

5 4

0 tM 0

.5 3 IL to 2

1 0

0 1

2 3

4 5

6 Bobbin Voltage I!&

70

Development of BP Correlation Using Length and Plus-point Voltage

• Addendum 5 database contains 18 flaws that have Plus point length/voltage available

• BP correlation developed based on length and voltage

• Very good agreement with actual BP of pulled tubes

• Exactly predicts 2R1 1 BPs for pulled tubes

• BPs can now be predicted for DCPP 2R1 1 axial indications with length and voltage available

• BP can then be plotted against the bobbin voltage for same indication

• Profiled flaws fell below the mean Addendum 5 regression line

• Resulted from profiling only those flaws that had higher Plus point to bobbin voltage ratios (single dominant type flaws)

Needed additional sizing for smaller ratios where multiple shallow cracks are represented by a high bobbin voltage 71

Fit vs. Measured BP for Addendum 5 Database Flaws Having Plus Point Data BP = f(+Pt Volts and Length) 10000

• Fitted versus Adjusted Burst Pressure 9000 -..

Ideal Prediction 8000 X DCPP-2 2R1 1 Pulled Tubes 7000 -

6000 -

y2 5000 -

2 4000 -

3000 -

Standard Error of Estimate = 572 psi 2000 -

1000 -

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 100 Fitted Burst Pressure, psi 72

Calculated Burst Pressures from Plus point Volts/Length Correlation vs. Known Bobbin Voltage for Same Indication Comparison of Addendum 5 Database to DCPP Specific Data Burst Pressure vs. Volts for 7/8" Alloy 600 SG Tubes Reference Database, Reference Sy+S = 137.56 ksi 12 0 ARCDatabase 7/8" ARC A

DC2NDD 10.0

-0 DC 1&2 Pulled U.*

DC 2Rll Pxofiled Flaws 0

DC 2Rll Iength Measured 800 8 0 -

U 4.0 -

2.0-0.0 0.1 1

10 100 Bobbin Amphtude (Volts)

I kr~l q 73

POPCD Approval Options

• Option 1 -

• NRC approval of POPCD for Unit 2 Cycle 12 by June 2003, in advance of 90 day report submittal

• No need for NRC approval of POD of 1.0 for R44C45

• Option 2 -

• NRC approval of POD of 1.0 for R44C45 by June 2003, in advance of 90 day report submittal

• NRC approval of POPCD for Unit 2 Cycle 12 by September 2003 74

Future PG&E Actions

• Submit DCPP POPCD LAR

• Submit 90 day report by end of June with:

• Destructive exam results

• Updated correlations

• Updated Monte Carlo CM and OA

• Risk assessment

• Chemical Cleaning (1R12, 2R12)

• Reviewing feasibility of TSP locking in 1 R1 2 and 2R12

• License amendment needed to change ARC repair limit and ARC leak and burst methods

• SG replacements (1R15, 2R15) -;oo 75