ML040420381

Response to RAI Regarding Steam Generator Tube Inspection Report for the 2002 Outage

ML040420381
Person / Time
Site:	Prairie Island
Issue date:	02/04/2004
From:	Solymossy J Nuclear Management Co
To:	Document Control Desk, Office of Nuclear Reactor Regulation
References
L-P1-04-012, TAC MB8804
Download: ML040420381 (12)

Text

Prairie Island Nuclear Generating Plant Committed to Nuclear Exce Operated by Nuclear Management Company, LLC L-PI-04-012 FEB 0 4 2004 U S Nuclear Regulatory Commission ATTN: Document Control Desk Washington, DC 20555 PRAIRIE ISLAND NUCLEAR GENERATING PLANT DOCKET 50-282 LICENSE No. DPR-42 RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION - PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE (TAC NO. MB8804)

By letters dated December 13, 2002, December 26, 2002 and two letters dated March 6, 2003, Nuclear Management Company, LLC, (NMC) submitted the steam generator tube inspection summary reports for the Prairie Island Nuclear Generating Plant, Unit 1, from the Fall 2002 outage. By letter dated December 15, 2003, the Nuclear Regulatory Commission (NRC) submitted a Request for Additional Information (RAI), in which they requested answers to 15 questions associated with the NMC letters. The attachment to this letter includes the NMC answers to the NRC's questions.

This letter contains no new commitments and no revisions to existing commitments.

CTJL 4-eQJ2 4

Q 4 Joseph M. Solymossy Site Vice President, Prairie Island Nuclear Generating Plant CC Regional Administrator, USNRC, Region IlIl Project Manager, Prairie Island Nuclear Generating Plant, USNRC, NRR NRC Resident Inspector-Prairie Island Nuclear Generating Plant Attachment 1717 Wakonade Drive East

• Welch, Minnesota 55089-9642 Telephone: 651.388.1121

ATTACHMENT NUCLEAR MANAGEMENT COMPANY, LLC PRAIRIE ISLAND NUCLEAR GENERATING PLANT DOCKET 50-282 RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT I STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE Page 1 of 11

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE Requested Information Steam Generator Tube Support Plate Voltage Based Repair Criteria 90-Day Report

1. On page 2 of your attachment to your March 6, 2003, letter, you indicate that all distorted signal indications (DSI) were inspected with a rotating probe to identify possible instances of wastage at outside diameter stress corrosion cracking (ODSCC) locations. You further indicated that "no such indications with voltages greater than the 2.0-volt limit were found at EOC 21". Please clarify this last statement. Were indications of wastage found at any tube support location (regardless of voltage)? If so, how were indications attributed to wastage differentiated from indications attributed to closely spaced intergranular attack/ODSCC? Were all volumetric indications at tube supports plugged upon detection?

Response

We found and plugged five Single Volumetric Indications (SVls) at cold leg tube support plates due to the poor signal to noise ratio of the bobbin coil (small voltage and deep as measured depth). We also found and plugged two very similar SVI type indications at hot leg tube support plates. We differentiate between wastage and closely spaced intergranular attack (IGA)/ODSCC by analyzing the result of the pancake coil in addition to the +Pt. coil. From past tube pull results we know that wastage and thinning type indications have +Pt. responses that are volumetric in nature (go both up and down on both legs of the coil) and the pancake response is large in comparison to the +Pt., with a shallow phase and a smooth signal formation.

Also from past tube pull results, we know IGA/ODSCC type indications have +Pt.

responses that are axial in nature (go up on one coil leg and down on the other) and the pancake response is either non-existent or very small in comparison to the +Pt.

with a deep phase and a jagged signal formation. Rotating Pancake Coil (RPC) inspections did not reveal any ODSCC indications at dents, mixed residual signals, indication not reportable (INR) locations or at cold leg thinning locations at cold leg tube support plates (TSPs). Cold leg thinning volumetric indications are left in service if sized at less than 40% through-wall.

2. Per generic letter 95-05, Voltage-Based Repair Criteria for Westinghouse Steam Generator Tubes Affected by Outside Diameter Stress Corrosion Cracking, locations with large mix residuals are to be inspected with a rotating probe.

Please discuss whether any indications were found at locations with large mix residuals and discuss how these tubes were dispositioned (i.e., were indications found at large mix residual locations repaired upon detection)?

2

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE

Response

No ODSCC indications were found at mixed residual signal locations via RPC inspections.

3. On page 3 of your attachment to your March 6, 2003 letter, you indicate the approach for assessing primary-to-secondary leakage under postulated accident conditions was consistent with one presented to the NRC in Reference 9. On page 21, you indicate that the Monte Carlo methodology, "other than for occasional use of uncorrelated leak rate selections" followed standard practice and was benchmarked. Please clarify whether the methodology used for assessing primary-to-secondary leakage under postulated accident conditions at Prairie Island Unit 1 was consistent with the NRC approved methodology discussed in letter dated March 27, 2002 (ML020870777). The staff notes that this is an NRC approved methodology for cases where the p-value exceeds 5%.

If the NRC approved methodology was not used, please provide a detailed description of the statistical analysis supporting the method you did use for assessing leakage under postulated accident conditions. In addition, clarify whether Figure 4.1 was calculated with the NRC approved methodology. Further clarify, how Figure 4.1 was used in the leakage analysis. For example, does Figure 4.1 represent the 95/95 leakage value in those instances where "no leak rate correlation is assumed" If it doesn' and it was used in assessing primary-to-secondary leakage under postulated accident conditions, discuss how the uncertainty in Figure 4.1 was modeled.

Response

Pulled tube leak rate test results have led to an updating of correlation coefficients used in the application of the bobbin voltage based repair criteria. Updated correlations of burst pressure with bobbin voltage have relatively little impact on structural integrity analyses. The probability of leakage at steam line break (SLB) conditions versus bobbin voltage has also changed based on tube pull results over the past year. However, these changes are not dramatic and do not require any changes to the standard calculation methodology. New leak rate measurements at 2560 pounds per square inch (psi) SLB differential pressure conditions have resulted in new correlation coefficients between SLB leak rate and bobbin voltage. While the absolute changes are not substantial, the degree of correlation between SLB leak rate and bobbin voltage, expressed as the "p" value, has exceeded the accepted value of 0.05 needed to apply a linear correlation between the log of the SLB leak rate and the log of the bobbin voltage. The revised value of p is 0.076. The p value for the correlation of log leak rate and log voltage for an SLB differential pressure of 2405 psi remains below 0.05. Hence, by previous standardized procedures, a leak rate correlation could be applied at a differential pressure of 2405 psi but not at 2560 psi. This runs counter to physical behavior and the use of a leak rate correlation for 0.750 inch diameter tubing at either pressure differential. Sampling of an SLB leak 3

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE rate distribution, which is uncorrelated with bobbin voltage, leads to grossly overestimated SLB leak rates. In this approach a very high leak rates would be just as probable for extremely low-level bobbin voltages as for voltages that exceeded the 2.0 volt limit by an order of magnitude.

At a Nuclear Energy Institute (NEI)/NRC meeting in February of 2002, a proposal was advanced to sample the slope of the log leak rate/ log voltage correlation, as is done in the standard calculation methodology, but when the slope is zero or less, no correlation is assumed. When the slope is greater than zero, the correlation is applied. The approach can be considered as sampling of the p value. The net effect is using uncorrelated leak rates a fraction of the time. This fraction is about p.

With the revised database, the value of p is 0.076. Thus, an uncorrelated leak rate effect is included and the basic requirement of not using a leak rate correlation with a p value less than 0.05 is essentially met.

The above procedure was followed in the present analysis for calculations at a differential SLB pressure of 2560 psi. Prairie Island has established the basis to implement an SLB pressure differential of 2405 psi but has not made a submittal to the NRC. SLB leak rate calculations are presented in this report at 2405 psi for information purposes. Use of an SLB leak rate correlation for all calculations at 2405 psi is justified by the revised leak rate correlations with a p value of 0.023.

When Monte Carlo sampling leads to conditions requiring use of an uncorrelated leak rates at 2560 psi, the distribution of leak rates in Figure 4.1 is sampled, regardless of the bobbin voltage. The ordinate is the cumulative distribution function and the abscissa is the log of the SLB leak rate expressed in terms of liters per hours at room temperature density. The logarithm values are to the base 10. The irregular line is the revised 2560 psi SLB leak rate database. The dotted smooth curve is the inferred parent population of SLB leak rates given by a log normal distribution using best estimate mean and standard deviation values from the test database. The smooth solid curve is generated from 100,000 Monte Carlo simulations of SLB leak rates applying Student Mt" statistics to the mean leak rate estimates and Chi Squared statistics to the standard deviation. It is seen that the variation added to the distribution of leak rates from a relatively complete statistical treatment is rather small.

Figure 4.2 contains the latest revised probability of leak and leak rate correlations parameters used for calculations at 2560 psi. The Monte Carlo simulation methodology, other than for occasional use of uncorrelated leak rate selections followed standard practice and was benchmarked versus Bobbin Voltage ARC Round Robin results. The burst pressure analysis methodology followed standard practice per WCAP 14277. The latest burst pressure database correlation coefficients were applied. These parameters are presented in Figure 4.2.

4

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE I

0.9 _

O.8 t-0.7 -

0.6 _

Rankj MQk.O Rankj 0.5 _

I I

I I

I

./

,:I I

I I

I I

I I

I 0.4 -

0.3 _

0.2 -

0.1 _-

I I

0 0

-4

- 4

-3

-2

-1 0

1 2

SORTLQ3j,MQ kI.SORTLQ2j 3

4 5

6 6

Figure 4.1 Distribution of SLB Leak Rates at 2560 psi for Use in Calculations without a Leak Rate Correlation Equation 5

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE r;

7,777

.- 6SGDB Variable-alue:

Burst Intercept 7.4934 Burst Slope

-2.3775 Burst Sig 0.6661 Burst V(OJ 0.009427291 B UFSt Vl I-0.004564115 Burst V(2J 0.015630532 Burst DOF 95 Leak Intercept

-0.0691 Leak Slope 0.717 Leak Sig 0.81 08

• Leak V(OJ 0.310201604

' Leak V(lJ

-0.260564345 Leak V(2J 0.236125369 Leak DIOF 27_I_____

POL Intercept

-5.1017 POL Slope 7.3403_

POL V(YO 1.3742__!

POL V(lJ

-1.7365 POL V(2J 2.6428

- POL DOF 113

~~~....

~.

Figure 4.2 Latest Revised Leak Rate and Burst Pressure Correlation Parameters 6

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE

4. Please provide a copy of Figure 2.8 from this 90-day report. Figure 2.8 was missing from your submittal.

Response

300 -

250-3 SG 11, Composite, Cycle 21 200 -

o0 SG 12, SG Specific, 150 -

° 100 Z~ 50 - _t-;:

cumulative fraction.

5. Please clarify your statement on page 17 of your attachment to your March 6, 2003 letter where you indicate that the "maximum voltage observed in any simulation keeps increasing as the number of simulations increases since the analystuncertainty is unbounded." In particularis this 2statement implying that there is a potential that the maximum voltage can increase as the number of simulations increases or that the maximum voltage always increases.

7

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE

Response

This is a theoretical rather than a practical observation. Monte Carlo simulations involve selections from distributions. If a distribution is unbounded, such as a normal distribution, which is not truncated, then the maximum value that can be selected is unbounded. On theoretical grounds as the number of selections increases, there is a statistically insignificant better chance of selecting a very large value.

6. Table 5.1 in your attachment to your March 6, 2003 letter provides the probability of burst associated with condition monitoring. Please verify the probability of burst values provided in this Table. If they are correct, discuss how the probability of burst at 2405 pounds per square inch (psi) can be greater than the probability of burst at 2560 psi. In addition, given that you projected a higher number and more severe indications for the EOC 21 (made at BOC 21 and reported in your May 29, 2001 submittal) than was actually observed at EOC 21, please clarify why the projected probability of burst (reported in Table 7-2 in your May 29, 2001 submittal) was less than the actual probability of burst (reported in Table 5.1 of your attachment to the March 6, 2003 letter).

Response

There are no significant differences between the reported probability of burst values because of very low probabilities of burst, differences in the number of simulations and use of 95/95 burst probability statistics. At very low probabilities of burst, very large numbers of simulations are required to determine accurate and precise burst probability values. When a calculated bounding 95/95 burst probability value is far removed from the acceptance limit of 0.01, very large number of simulations are not needed to demonstrate acceptable structural integrity. Use of 95/95 statistics leads to effective inclusion of a simulated burst even when no bursts are observed in a given number of simulations. In this case, the burst probability is bounding but not necessarily accurate (i.e. the burst probability cannot be higher but can be lower).

7. On page 28 of your attachment to your March 6, 2003 letter, you indicate that the composite voltage growth rate was -0.13 volts per EFPY for Cycle 21. This does not appear to be consistent with Table 2.2 of this attachment. Please clarify.

Response

The units are not equivalent. Page 5 states that Table 2.2 lists growth rates as a percentage of the BOC average voltage.

8

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE

8. Please clarify why the projected number of indications for EOC 21 provided in your submittal dated May 29, 2001 do not match those provided in your March 6, 2003 submittal (page 5 of Attachment).

Response

The projected number of indications in SG 11 for EOC 21 was 561. A total number observed was 365. Corresponding values for SG 12 are 266 predicted versus 204 observed. At Prairie Island Unit 1, the projected number of indications continues to be conservative compared to actual observations.

9. Please indicate the length of cycle 21.

Response

The length of cycle 21 was approximately 605 days.

Inservice Inspection Summary 90-Day Report

1. In your March 6, 2003 (ML030730056) letteryou indicated that one sleeved tube (R4C76) was not inspected due to an obstruction and was plugged. Please describe the nature of the obstruction and the type of sleeve used.

Response

The sleeve in question is an ABB-Combustion Engineering tungsten inert gas (TIG) welded sleeve. The obstruction is due to a collapse of the sleeve and was caused by a diode effect when secondary side water seeps into the crevice between the parent tube and the sleeve through a parent tube flaw during cold shutdown conditions. During heatup/startup conditions, the parent tube flaw seals and the water in the crevice heats up and expands to exert an inward pressure on the sleeve in excess of its yield strength.

2. In Table II of your March 6, 2003 letter, you provide the location and extent of wall thickness penetration for each indication of an imperfection. Several indications listed in this table were in the freespan (e. g., indications 49, 146, etc.)

Please describe the nature of the eddy current signals at these locations. For example, please discuss whether a flaw was present at this location and if so, provide the size (length, depth, percent degraded area) and nature of the indication (primary water stress corrosion cracking, outside diameter stress corrosion cracking, etc.).

9

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE

Response

The nature of the freespan indications is wear at old AVB intersections. In 1986, the original AVB bars were removed and replaced. Wear indications are sizable with the bobbin probe and the percent through wall depth is reported in the "PERCENT column of the subject report.

3. Table Ill of your report indicates a single volumetric indication was detected in tube R10C69 at the weld centerline. Please discuss the nature and cause of this indication.

Response

Previous inspection of the tube were reviewed to determine if the indication had existed prior to the 2002 inspection - data from 1996, 1997, and 1999 was reviewed and no previous indication were found (R10C69 was not inspected in 2001). The inspection of sleeves was expanded to 100% from 25% - no additional indications were found. The tube and sleeve were in-situ pressure tested to document the structural integrity of the tube - the tube passed the testing. This demonstrated margin to rupture during normal and accident conditions. The tube was repaired by plugging. Therefore the cause was determined to be volumetric (typically wastage),

which had not grown enough over two fuel cycles to impact on the tube structural integrity. This was the only indication on a sleeved tube, and the tube was repaired.

Therefore, any further degradation could not affect the integrity of the steam generator tubes. Since this is the last inspection prior to the replacement of the Unit 1 Steam Generators, additional evaluation was not warranted.

4. Table II of your report indicates various indications located at the first and second tube support plate on the cold leg side (R35C77, R31C82, etc.). Please discuss the nature of these indications (e. g., cold leg thinning, outside diameter stress corrosion cracking). If the degradation was attributed to cold leg thinning, please discuss how cold leg thinning can be differentiated from closely spaced stress corrosion cracking or intergranular attack.

Response

The nature of the indications is thinning at cold leg tube support plates. A tube pulled from steam generator 22 in 1980 confirmed the presence of wall thinning and corrosion. Tubes pulled from steam generator 11 and 12 in 1997 confirmed the presence of old inactive corrosion due to phosphate wastage. These indications can be differentiated from ODSCC/IGA using MRPC probes. Therefore, it is our practice 10

RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION PRAIRIE ISLAND UNIT 1 STEAM GENERATOR TUBE INSPECTION REPORT FOR THEIR 2002 OUTAGE to MRPC test all cold leg thinning indications >40% through wall or <40% through wall and >1.5 Volts on the bobbin coil.

5. You indicated in Table I of your report that only 25% of the free span dents were inspected with a rotating probe. Please discuss the results of your inspection.

Please also discuss how the tubes that were to be examined were determined.

For example, was it random sample or were all dents above 5 volts examined with a rotating probe and the remaining sample was random. If cracks were found during this inspection, please discuss your basis for not expanding the scope of the inspection.

Response

Selection of the 25% freespan dent population >5.0 volts was determined by the vendor's data management personnel. No cracking has ever been identified at any dented location at Pi, which is the basis for no scope expansion.

Steam Generator Inspection Results Day Report

1. In your December 13, 2002 letter, you indicated that several single axial indications (SAD) and multiple axial indications (MAD) were no longer detectable.

Please discuss any insights you may have on why these indications are no longer detectable.

Response

During the January 1997 Unit 2 examination we discovered previous re-roll tubes left in-service with the F* repair criteria exhibited signs of leakage. It was determined that the installation of the hydraulic expansion had a tendency to open up the PWSCC indications that were the cause for installing the re-rolls. To eliminate or reduce the leakage potential we began to install a 3" hard roll over the original equipment manufacturer's 2.75" hard roll to close/seal existing cracks. As a result, a large percentage of the PWSCC indications are no longer detectable after re-rolling.

Primarily for configuration management reasons we report the missing indications as MAD or SAD, which helps account for all known PWSCC indications.

11