ML20249B432

Responds to 980608 Telcon Requesting Copies of Available Documents Re Evaluations of Cracks in Vertical Welds of Core Shroud at Nine Mile Point Nuclear Station,Unit 1.Documents Listed in Encl 1 Relevant to Request

ML20249B432
Person / Time
Site:	Nine Mile Point
Issue date:	06/18/1998
From:	Hood D NRC (Affiliation Not Assigned)
To:	Swift E AFFILIATION NOT ASSIGNED
References
TAC-M99720, NUDOCS 9806230092
Download: ML20249B432 (5)

Text

. _ _ _ _ . __. _ . _ _ _ _ _ _ _ . _ - - - - _ _ _ - _ - _ . - _ _ _ - _ - - - - - _ _ _ _ _

, s g ea **e og 6~O-220 g 1 UNITED STATES r S NUCLEAR REGULATORY COMMISSION WASHINGTON, D.C. 20006-4001

%, . . . . . [f June 18, 1998 Dr. Edward A. Swift 5173 Skyline Drive Syracuse, NY 13215

SUBJECT:

DOCUMENTS REGARDING EVALUATION OF CORE SHROUD WELDS AT NINE MILE POINT NUCLEAR STATION, UNIT NO.1 (TAC NO. M99720)

Dear Dr. Swift:

This will respond to your telephone call to me on June 8,1998, to request copies of available documents regarding the evaluations of the cracks in the vertical welds of the core shroud at Nine Mile Point Nuclear Station, Unit 1. You requested documents that (1) have been issued since the U.S. Nuclear Regulatory Commission's (NRC's) two meetings (one with Niagara Mohawk Power Corporation and one with the public) on April 14,1997, that you attended, and (2) are relevant to the NRC's ongoing review of the request by Niagara Mohawk Power Corporation that the vertical weld inspection be extended from 10,600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> of hot operation to 14,500 hours0.00579 days <br />0.139 hours <br />8.267196e-4 weeks <br />1.9025e-4 months <br /> of hot operation.

Documents relevant to your request are listed in Enclosure 1, along with their accession numbers. Non-proprietary versions of these documents are available in your local public document room (PDR) located at the Reference and Documents Department of the Pennfield Library, State University of New York, Oswego, NY 13126, (315) 341-3563/3564. The accession ]

numbers will assist the local librarian in locating the documents. )

In addition to documents on the Nine Mile Point docket, the NRC staff has reviewed information on vessel intemals that is generic to boiling water reactors (BWRs), including several topical reports submitted by the BWR Vessel and Intemals Project (BWRVIP). One such report of I interest to shroud inspection intervals, recently approved by the NRC, is BWRVIP Report i BWRVIP-14, titled "BWR Vessel and Intemals Project, Evaluation and Crack Growth in BWR Stainless SteelIntemals (BWRVIP-14)," dated March 1996. Because the NRC's approvalletter (accession number 9806110271) of June 8,1998, is recent and has not yet reached your PDR, I ,l am enclosing a copy with this letter as Enclosure 2. j e

4 .,

9806230092 900618 PDR ADOCK 05000220 1 H PDR J J

I

( s-1 V

June 18, 1998 E. A. Swift I trust you will find the local PDR convenient for your review purposes. If not, you are encouraged to provide your comments to the local librarian (Ms. Mary Bennett) or to the NRC's PDR staff (Ms. Jona Souder, LPDR Program Manager; or Ms. Teresa Union, LPDR Information Services Ubrarian) at 1-800-397-4209.

Sincerely.

Original Signed by:

Darl Hood, Senior Project Manager Project Directorate 1-1 Division of Reactor Projects- 1/11 Office of Nuclear Reactor Regulation Docket No. 50-220

Enclosures:

1. Ust of Documents

2. June 8,1998, letter cc w/encis: See next page DISTRIBUTION:

<tDocket Files

• PUBLIC PDI-1 Reading SBajwa SUttle DHood OGC ACRS ESullivan RHermann KWichman WKoo KKavanagh DScrenci VDricks LDoerflein, RI DOCUMENT NAME: G;\NMP1\ SWIFT.LTR To receive a copy of this document, Indicate in the box: "C" = Copy without attachment / enclosure *E" = Copy with rit:chment/ enclosure "N" = No copy ,

OFFICE PM P01 1 ,lE LA:PDI 1s Qjd. l D PDI-1 ,,,-- l l l NAME DHood/Lec .E ) N SLittle N SBsjWe f/7 DATE 06/,(./98 06/ 0 98 06/ ///9'8 Official. Record-Copy t

.. ; o . s12 4

i

_ __ ___ _ _j

I s

June 18,1998 E. A. Swift I trust you will find the local PDR convenient for your review purposes. If not, you are encouraged to provide your comments to the local librarian (Ms. Mary Bennett) or to the NRC's PDR staff (Ms. Jona Souder, LPDR Program Manager; or Ms. Teresa Linton, LPDR Information Services Librarian) at 1-800-397-4209.

Sincerely, Original Signed by:

Darl Hood, Senior Project Manager Project Directorate I-1 Division of Reactor Projects - 1/11 Office of Nuclear Reactor Regulation Docket No. 50-220

Enclosures:

1. List of Documents

2. June 8,1998, letter cc w/encts: See next page DISTRIBUTION:

Docket File PUBLIC PDi-1 Reading SBajwa SLittle DHood OGC ACRS ESullivan RHermann KWichman WKoo KKavanagh DScrenci '

VDricks LDoerflein, RI DOCUMENT NAME: G:\NMP1\ SWIFT.LTR To receive a copy of this document, Indicate in the box: "C" = Copy without attachment / enclosure "E" = Copy with rit:chment/ enclosure "N" = No copy ,

OFFICE PM:PDI 1 ,lE LAPDI1,QJi l D:PDI 1 ,,, l l  !

want DNood/tec .P 3 N sttet:7 seawe 4/r oatE 06/, u9a 06/ 0 9a 06/ <//9's Official Record Copy l

l

s E. A. Swift I trust you will find the local PDR convenient for your review purposes. If not, you are encouraged to provide your comments to the local librarian (Ms. Mary Bennett) or to the NRC's PDR staff (Ms. Jona Souder, LPDR Program Manager; or Ms. Teresa Linton, LPDR Information Services Librarian) at 1-800-397-4209.

Sincerely, Badh/M Dari Hood, Senior Project Manager Project Directorate 1-1 Division of Reactor Projects - t/II

' Office of Nuclear Reactor Regulation Docket No. 50-220

Enclosures:

1. List of Documents

2. June 8,1998, letter cc w/encls: See next page

. I l'

l 1

f.

s I

( John H. Mueller Nine Mile Point Nuclear Staten Niagara Mohawk Power Corporation Unit No.1 cc:

Regional Admidstrator, Region i U.S. Nuclear Regulatory Commission 475 Allendale Road King of Prussia, PA 19406 Resident inspector U.S. Nuclear Regulatory Commission P.O. Box 126 Lycoming, NY 13093 Charles Donaldson, Esquire Assistant Attomey General New York Department of Law 120 Broadway New York, NY - 10271 Mr.' Paul D. Eddy State of New York Department of Public Servie,e Power Division, System Operations 3 Empire State Plaza Albany, NY 12223 Mr. F. William Valentino, President New York State Energy, Research, and Development Authority Corporate Plaza West 286 Washington Avenue Extension Albany, NY 12203-6399 Mark J. Wetterhahn, Esquire Winston & Strawn 1400 L Street, NW Washington, DC 20005-3502 -

Gary D. Wilson, Esquire Niagara Mohawk Powsr Corporation 300 Erie Boulevard West Syracuse, NY 13202 Supervisor Town of Scriba Route 8, Box 382 Oswego, NY 13126

LIST OF DOCUMENTS REGARDING CORE SHROUD WELD CRACKING AT NINE MILE POINT NUCLEAR STATION, UNIT 1

1. NRC letter to Niagara Mohawk Power Corporation (NMPC) dated May 8,1997, titled

" Modification to Core Shroud Stabilizer Lower Wedge Retaining Clip and Evaluation of Shroud Vertical Wold Cracking." Accession No. 9705160106 2, NMPC letter to NRC dated September 30,1997, titled " Analyses and Evaluation of Core Shroud Boat Samples." Accession No. 9710070370

3. NMPC letter to NRC dated November 20,1997, title " Delay in Additional Tests of Core Shroud Boat Samples." Accession No. 9712020027

4. NMPC letter to NRC dated January 30,1998, titled " Core Shroud Vertical Weld Fluence Measurements." Accession No. 9802110082

5. NMPC letter to NitC i dated February 27,1998, titled " Generic Letter 94-03, integranular Stress Corrosion Cracking (IGSCC) in Boiling Water Reactors." Accession No. 9803090214

6. NMPC letter to NRC dated March 31,1998, titled "10CFR50, Appendix H, Reactor Vessel Material Surveillance Program Requirements, Report of Test Results." Accession No.

9804210352

7. NMPC letter to NRC dated April 16,1998, titled " Background information on Nine Mile Point Unit 1 Core Shroud Boat Sample Evaluation." Accession No. 9804240338

8. NMPC letter to NRC dated April 30,1998, titled " Generic Letter 94-03, Integranular Stress Corrosion Cracking of Core Shrouds in Boiling Water Reactors." Accession No. 9805140024 P

Enclosure 1

-)

. pf ~%,t p UNITED STATES

!' u j NUCLEAR RE2ULATORY COMMISSION l*

$ g c'ASHIN 3 ton. C.'3. 3008H001

y. . . . . j \

June 8,1998 CarlTerry, BWRVIP Chairman Niagara Mohawk Power Company Post Office Box 63 Lycoming,NY 13093

SUBJECT:

SAFETY EVALUATION OF THE BWR VESSEL AND INTERNALS PROJECT BWRVlP-14 REPORT (TAC NO. M94975)

Dear Mr. Terry,

By letter dated March 18,1996, as supplemented by letter dated July 28,1997, the Bol!!ng Water Reactor Vessel and intemals Project (BWRVIP) submitted the Electric Power Research Institute (EPRI) Proprietary Report TR 105873, "BWR Vessel and Intemals Project, Evaluation of Crack Growth in BWR Stainless Steel intemals (BWRVIP 14)," dated March 1996, for staff review and approval. The staff requested additionalinformation (RAI)in c letter dated December 9,1996, and the BWRVIP provided its response, Structural Integrity Associates .

Report No. SIR 97-025, Revision 3," Responses to NRC Request for Additional information on BWRVIP Report BWRVIP-14 on Evaluation of Crack Growth in RPV intemals," dated June 1997, by letter dated July 28,1997.

The BWRVIP-14 report provides a methodology for assessment of crack growth in BWR stainless steel shrouds and other stainless steelintemals components. The assessment was limited to the circumferential welds of the shroud where most of the reported intergranular stress corrosion cracking (IGSCC) has occurred to date. The methodology was developed specifically for crack growth in the radial (through thickness direction). Residual and applied stresses and stress intensity factors (K) have been developed for crack propagation in this direction.

The NRC staff has reviewed the BWRVIP 14 report and finds,in the enclosed Safety Evaluation (SE), that the three approaches described in the BWRVIP 14 report for the evaluation of crack growth are acceptable for use except where the staff's conclusions differ from the BWRVIP's, as discussed in the enclosed SE. This finding is based on information submitted both originally and in response to the staff's request for additionalinformation (RAI), dated December 9,1996, that clarified the guidance in BWRVIP 14. The staff requests that the BWRVIP review and resolve the issues raised in the enclosed SE, and incorporate the staff's conclusions into a revised BWRVIP 14 report. Please inform the staff within 90 days of the date of this letter as to your proposed actions and schedule for such a revision. *,

1 l

l i

i I

l aamuy ae- 2

CarlTony 2-Please contact C. E. (Gene) Carpenter, Jrl, of my staff at (001) 415-216g if you have any further questions regarding this subject.

Sincerely, N

Gus C. Laines, Acting Director Division of Engineering ONice of Nuclear Reactor Regulation

Enclosure:

As stated cc: See next page .

e 9e 4

O

l CO:

l George Jones, Executive Chairman Dana Covill, Technical Chairman l BWRVIP Assessment Task BWRVIP Assessment Task l Pennsylvania Power & Light GPU Nuclear A6-1 1 Upper Pond Road Two North Ninth Street Parsippany, NJ 07054 Allentown, PA 18101 l

Joe Hagan, Executive Chairman Carl Larsen, Technical Chairman BWRVIP inspection Task BWRVIP inspection Task .

Entergy Yankee Atomic P. O. Box 756 580 Main Street Waterloo Road Bolton, MA 01740 Port Gibson, MS 39150 Paul Semis, Executive Chairman Vaughn Wagoner, Technical Chairman BWRVIP Integration Task BWRVIP Integration Task Washington Public Power Supply System Carolina Power & Light Company P. O. Box 968 One Hannover Square SC1 l

North Power Plant Loop P.O. Box 1551 .

I Richland, WA 99352-0968 Raleigh, NC 27612 Lewis Sumner, Executive Chairman John Wilson, Technical Chairman i BWRVIP Mitigation Task BWRVIP Mitigation Task Southern Nuclear Operating Co. Public Service Electric & Gas Co.

40 inverness Center Parkway N51 -

Birmingham, AL 35201 Post Office Box 236 I Hancocks Bridge, NJ 08038 John Blomgren, Executive Chairman Bruce McLeod, Technical Chairman BWRVIP Repair Task BWRVIP Repair Task

, Commonwealth Edison Co. Southern Nuclear Operating Co.

l 1400 Opus Place, Suite 600 Post Office Box 1295 Downers Grove, IL 60515 5701 40 inverness Center Parkway l

Birmingham, AL 35201 Bill Campbell, BWRVIP Vice Chairman Warren Bilanin, EPRI BWRVIP Caro!ina Power & Light Integr&tlon Manager P. O. Box 1551 Joe Gilman, EPRI BWRVlP Raleigh, NC 27612 Mitigation Manager Ken Wolfe, EPRI BWRVIP Robert Carter, EPRI BWRVIP Repair Manager Assessment Manager Electric Power Research Institute Greg Selby, EPRI BWRVIP P. O. Box 10412 Inspection Manager 3412 Hillview Ave.

EPRI NDE Center Palo Alto, CA B4303 P. O. Box 217097 1300 W. T. Harris Blvd.

Charlotte, NC 28221 l

E 1

U.S. NUCLEAR REGULATORY COMMISSION OFFICE OF NUCLEAR REACTOR REGULATION DIVISION OF ENGINEERING SAFETY EVALUATION OF EPRI TOPICAL REPORT TR-105873

• BWR VESSEL AND INTERNALS PROJECT. EVALUATION OF CRACK GROWTH IN BWR STAINLESS STEEL INTERNALS. (BWRVIP-14)"

1.0 INTRODUCTION

By letter dated March 18,1996, as supplemented by letter dated July 28,1997, the Boiling l

Water Reactor Vessel and infomats Project (BWRVIP) submitted the Electric Power Research l' Institute (EPRI) Proprietary Report TR 105873, "BWR Vessel and intamals Project, Evaluation of Crack Growth la BWR Stainless Steel Intemals (BWRVIP-14)," dated March 1996, for staff review and approval. The staff requested additionalinformation (RAl)in a letter dated December 9,1996, and the BWRVIP provided its response, Structural Integrity Associates Report No. SIR 97-025, Revision 3," Responses to NRC Request for Additional informa. tion on

• I BWRVIP Report BWRVIP 14 on Evaluation of Crack Growth in RPVIntemels," dated June 1997, by letter dated July 28,1997.

The BWRVIP-14 report provides a methodology for assessment of crack growth in BWR stainless steel shrouds and other stainless steel intemals components. The assessment was limited to the circumferential welds of the shroud where most of the reported intergrariular stress corrosion cracking (IGSCC) has occurred to date. The methodology was developed specifically for crack growth in the radial (through thickness direction). Residual and applied stresses and stress intensity factors (K) have been developed for crack propagation in this direction.

1.1 Background

The cracking of stainless steel BWR core shrouds was identified as a significant issue beginning in 1993 and 1994. To investigate the issues of core shroud integrity and other BWR Intemals, the BWR utilities formed the Bolling Water Reactor Vessel and Intemals Project (BWRVIP) to address service related degradation of BWR vessels and infomals. One of the major issues involved is the reinspection intervals. The current methodology is to determine the inspection interval based on characterizing all observed cracks as being through the wal! and propagating the cracks 4

around the circumference of the core shroud (assuming a crack growth rate of 5 x 10 in/hr). This crack growth rate (CGR) is based on a stainless steel crack growth rate correlation provided in the NRC Technical Report NUREG 0313, Revision 2,' Technical Report on Material Selection and Processing Guidelines for BWR Coolant Pressure Boundary Piping,"

issued February 1984.

The NRC staff, with assistance from the Argonne National Laboratory (ANL), has assessed the -

BWRVIP's submittals in this safety evaluation (SE). The Argonne Technical Evaluation Report, ENCLOSURE

.J

%&&f O 'I/p

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

2

" Technical Evaluation Report on EPRI TR-105873, BWR Vessels and intomais Project,

' Evaluation of Crack Growth in BWR Stainless Steel RPV Intemais,"is encioned.

1.2 SER Organization This SER contains a brief summary of the general contents of the BWRVIP 14 report and a summary of the ANL Technical Evaluation Report, followed by the staff's evaluation and conclusions.

2.0

SUMMARY

OF BWRVIP-14 TOPICAL REPORT .

The BWRVIP methodology involved the development of an empirical model to account for the variability of important IGSCC parameters in providing an assessment of the crack growth rate in BWR stainless steel components. The BWRVIP 14 report includes a data base of unirradiated stainless steel crack growth rates, which is limited to data that had defined environmental -)

conditions and well characterized important crack growth parameters. This database was used to dortve an empirical crack growth rate law to account for stress intensity (K), and

- environmental conditions such as conductivity, o!actrochemical potential (ECP), and temperature. .

The BWRVIP 14 report also discusses the weld residual stresses developed during fabrication of the weld, in which both experimental measurements and analytical techniques were used. In general, the results on residual stress distributions were similar to that recommended in NUREG-0313, Rev. 2, for large diameter stainless steel pipe. The BWRVIP 14 report wams that variability - due to fit up stresses, weld joint geometry, and welding parameters auch as heat input, weld sequence, weld starts and stops and repairs - can affect local residual stress distribution.

Further, the report estimates the through-wall stress intensity factor distributions for the assumed

{

ideal residual stress profiles using fracture mechanics models, and it outlines three parallel '

attemate approaches to an evaluation methodology for determining crack growth rates (CGRs) in BWR shrouds. In the first approach, a K-independent crack growth rate of 2.2 x 104 in/hr is ,

recommended. This CGR represents the highest CGR - with a stress intensity factor of 25 ksi j

((in) and the 95th percentile curve of the model with conductivity of 0.15 pS/cm and ECP of 200 l mV(SHE) - which are typical for BWR plants operating under moderate hydrogen water -

chemistry conditions. The second approach involves the use of the 95th percentile curve of the model with conductivity of 0.15 pS/cm and ECP OF 200 mV(SHE) but with K-dependence. The third approach involves using all the variables in the crack growth model to establish the 95th percentile crack growth curve. An exampic problem representing actual BWR shroud conditions is also presented. Appendices to the BWRVlP 14 report present operating plant ultrasonic data that indicate that the present NRC (NUREG-0313) CGR is conservative. The report proposes that the BWRVIP's crack growth methodology be used for crack evaluation.

3.0

SUMMARY

OF ARGONNE TECHNICAL EVALUATION REPORT

. It should be noted that the attached Argonne National Laboratory (ANL) technical evaluation report (TER) does not necessarily represent approved staff positions, but is one of the sources that the staff has consulted in the formulation of its conclusions in this SER.

3 ANL, in their TER, found that the BWRVIP's first calculational approach (e.g., using a CGR of 2.2 x 104 in/hr) can be used in welds with well characterized res: dual stress profiles. ANL's TEP, points out that the BWRVIP CGR model does not depend on stress intensity t ' lly; rather, it was developed from the BWR 95th percentile curve whose stress intensity and chemistry values are given in the above summary. Therefore, it is applicable only to weldment geometries for which the constant value of K represents a conservative estimate of the actual variable value.

The staff concludes this approach can be used if the value of K is explicitly determined to be less than 25 kal/in, for weldment geometries to which that rate would be applied. Further, the licensee would need to certify that the components are operated in accordance with the EPRI j BWR Water Chemistry Guidelines.

One limitation of the BWRVIP's data base correlation is that ii consists solely of unirradiated base materials. Weldments and their associated fusion heat affected zones could have somewhat higher impurity levels due to flux / copper contamination or impurfty diffusion than the

• wrought base metals that are solely considered in the data base correlation. It is well known that changes in irradiated materials properties are a strong function of certain impurities. In addition, there are few, if any, valid crack growth measurements for irradiated metals. Benchmark comparisons with ultrasonic measurements made on actualirradiated core shroud cracks may

~

not be valid because the uncertainties in the ultrasonic measurements could mask any variations.

In the crack growth rate correlation.

Therefore, in order to use a K-based crack growth model such as the BWRVIP's correlation, careful estimates of the residual stresses, as well as the applied loads, are required. The prediction of the CGR is a strong function of the assumed residual stress distribution.

Unfortunately, when it comes to the welds in actual fabricated core support structures, only in the case of Double V cylinder to-cylinder welds (such as the H-4 welds) have the residual stresses been adequately characterized by measurements and by calculations. Other weld geometries may have deleterious residual stress configurations. For example, one calculation for Nine Mile Point Nuclear Plant on the H-8 weld geometry that joins the conical shroud support to the shroud cylinder revertled a very unfavorable stress distribution. In thst case, rather than decreasing as the crack grows, the K values associated with this weld increased. in this example, having an environment with a conductivity of 0.15 mS/cm and an ECP of 200 MeV, after the crack has grown about a third through wall, the predicted crack growth rate is in excess of the current NRC l accepted bounding crack growth rate.

4.0 STAFF EVALUATION l The staff accepts the recommendation in the ANL TER that dny use of the correlation or of any data derived from unirradiated specimens should be limited to materials with fluences in the range less than 5 x 10" n/cm' (E>1MEV), unless further technical justification or data are provided for a case by case review. The BWRVIP 14 report should be revised to more fully address the applicability of its model to irradiated material.

in those cases where it can be demonstrated that the electrochemical pctantial (ECP) and conductivity are reduced (e.g., such as by direct ECP measurements o; by methods demonstrated to achieve lower ECP such a hydrogen injection), the Maff may consider a reduction in CGR on a case by case basis. )

The discussion in the BWRVIP-14 report on the effects of weld repairs arid other residual stress distribution uncertainties (e.g., fit-up stresses, weld sequence, weld starts and stops) stated that the BWRVlP can give little guidance as to how to address the individual parameters for each

r . . - _ . __ _ .. _ _ . . . __ _ . _ .._

4 I individual BWR shroud. The ANL TER recommends that any calculations should be performed using a more conservative estimate of tlw stress profile (i.e., by adding a uniform 10 kal stress to the profile). Based on the ANL TER, the staff concludes that specific calculations of the stress intensity profile for each weldment geometry is necessary. Th!s needs to be done also for the CGR whenever K is not explicitly addressed. Wald repairs have a sign;ficant effect on residual stress distributions. Therefore, the staff expects repairs to be documented, their effect on the residual stress distribution evaluated, and be subject to staff approval on an application specific basis.

5.0 CONCLUSION

S The three approaches for the evaluation of crack growth given in the BWRVIP-14 report have i been found acceptable for use subject to staff review and the fo lowing conditions: {

1. the first approach (constant K and a CGR of 2.2 x 104 in/hr) may be used provided that repairs, etc., are considered in evaluating the residual stresses;

2. that the component is operated in accordance with the EPMI BWR Water Chemistry Guidelines; and, . 1

3. the stress intensity factor K is explicitly determined to be less than 25 ks!/in.

However, for each of the approaches, in addition to me consideration of repairs and water chemistry considerations, it will be necessary for the t.taff to wview the licensee's crack growth evaluation, including an evaluation of the applied and residua! stresses to determine the acceptability of the assumed crack growth rate. Residual stress determinations niust include repairs and any other relevant factors.

Cracking in weldments that have been irradiated by fluences in the range greater than 5 x 10

n/cm'(E > 1 MEV) is outside of the secpe of this SER and would require review on a case basis that is supported with relevant data.

Enclosure:

Technical Evaluation Report

r 1

{

Technical Evaluation Report on I EPRI TR-105873, BWR Vessels and l

Internals Project, Evaluation of Crack Growth \

in BWR Stainless Steel RPV Internal \

l 1 l

January 27,1998 ,

Prepared by W. J. Shack Argonne National Laboratory 9700 South Cass Avenue Argonne. IL 60439 l

Technical Monitor Herbert F. Conrad USNRC Omce of Nudear Reactor Regulation s l

1 i

. . .. ._3 #

Introduction his technical evaluation report reviews EPRI TR-105873. BWR Vassels and Internals Prqfect. D=f=% qf Crack Grav& in BWR Stainless Steel RPV Internals, which presents three approaches for the evaluation of crack growth in stainless steel RPV internals and the technical bases for these approaches. De overall conclusions and recommendations are presented Arst. nis is followed by a general discussion of the crack growth model developed in EPRI TR-105873 and compenson of the predictions of the model with a much larger dataset than that contained in EPRI M-105873. De calculations of the residual stresses associated with core shroud large<11ameter cylinder-to-cylinder

• double V" welds presented in EPRI TR-105873 are compared with independent calculations of these types of stresses done by Battelle Columbus Laboratodes under subcontract to Argonne National laboratory as part of an NRC sponsored sesearch program on erwironmentally assisted cracking. The responses to six Requests for Additional

{ infonnation (RAls) on EPRI TR-105873 a e also discussed.

l Concluelons and Recommendations p _ The BWRVIP crack growth model appears to provide a good correlation with a large database of experimental measurements of CGRs for unfrradiated materials under a range of .

! water chemistry conditions that bound those expected in-reactor. Because the model depends on the stress intensity factor (K), pracucal applicadon of the model also requires knowledge of the residual and applied stresses for the components of interest. As formulated in EPRI n-105873, it apphes only to situadons involving constant loads.

3ree approaches for the evaluation of crack growth are given in EPRI TRt108873: (1) use of a stress intensitj- factor (X) independent CGR of 2.2 x 10-5 in/h (1.6 x 10-30 m/s): (2) use of a K dependent CGR with a constant of propordonahty corresponding to bounding l values the conductivity (0.15 pS/cm) and ECP (200 mV): and (3) use of a K dependent CGR with plant specific data for conducuvity and ECP. Although the first approach does not explicitly depend on the stress intensity, it is derived from the BWRVIP 95% percentile correlation for a stress intensity of 27.5 Mpaml/2 (25 kalint /2) and a conductivity of  ;

0.15 pS/cm and ECP of 200 .mV. It is thus applicable only to weldment geometdes for which a constant value of 27.5 Mpa ml/2 represents a conservative esumate of K. Similarly applicadon of approaches (2) and (3) is appropriate only for situadons in which the residual and applied stresses have been suitably characterized. Computing the contribution to K of the applied stresses is usually not too difficult. Detenr with a specific weldment geometry is more complex. q the residual stresses associated Because the crack growth modelis based on data for unirradiated matedals, the use of the of the crack growth model and the three approaches for the evaluation of crack growth f given in EPRI TR-105873 should be '.imited to materials with fluence < 5 x 10*dn/cm2 i (E > 1 MeV) until additional evidence is presented to support the argument that thi data on the unirradiated matedals is conservative for irradiated matedals. For fluences less than this, it can be argued that the primary change to the material from the irradiation that 4 affects the CGR is radiation hardening, and that this would lead in a decrease in the crack up deformauon rate for a given load and crack geometry. nis suggests that the a correlauon based on unirradiated materials is likely to be conservative for such materials.

The limited data that are available from the tests in Halden are consistent with these

r assumption. At higher Suences more changes are occurring in the material and additional data are needed to va9date the use of any correlation for such materials.

Weldment geometries for which the residual stresses would be consistent with the assumptions implicit in the constant crack growth rate approach proposed in EPRI TR-105873, i.e., approach 1, are cylinder-to-cylinder " double V* welds (e.g., H4 core shroud welds), plate-to-plate " double V" welds (vertical core shroud welds), and butt welds in large-diameter (212 in.) piping. 'Ihe use of the second and third approaches proposed in EPRI TR-105873 requires throughwall distributions of the residual stress. The throughwall residual stress profile given in Table 4-5 of EPRI TR-105873 is a reasonable representation of the residual stresses associated with the " double-V" plate-to-plate shroud welds (horizontal or vertical core shroud welds). To mitigate any uncertainties associated with variations in weld procedures, etc., however, the analyses should be performed using a more conservative estimate of the stress profile obtained by adding a uniform 10 kai (11 MPa) stress to the profile in Talale 4-5. For

• single V* butt welds in large diameter piping, the distribution given in NUREG-0313 Rev. 2, provides a suitably conservative representauon of the residual stress profile for such geometries. For other weld geometries (cylinder to support ring welds, etc.) the residual stress and corresponding K analysis are mare complex and may lead to less favorable stress attuations.

In such cases, either detailed analyses must be performed to determine the distribution of -

residual stress and K or the more conservative NRR CGR of 5 x 10-5 in/h (3.5 x 10-10 m/s) should be used.

RAls 2-6 were addressed very well, and by and large we have few comments on them.

We do not agree with all the elements of the response to RAI 1. concerning the 11gh3 crack growth rates (CGRs) observed in the Halden tests, but we do agree wf!h the main conclusion of the response that the specimens are being used so far out of any fracture mechanics validity range that it is meaningless to compare the CGRs in the test with the predicuons of the BWRVIP correlation.

Genersi Discussion

'Ihe proposed BWRVIP model is limited to constant loads. In order to be able to compare the model to a larger database,it was extended to include cyclic loading, but the dependence on temperature, electrochemical potenual (ECP), conductivity, and K for constant loads is the same as the original BWRVIP correlation.

'Ihe BWRVIP model is of the form:

a = C K",

(1) where C is function of conductivity, temperature, and electrochemical potendal and K is the fracture mechanics stress intensity factor. However, under static loading. K is widely considered to be a surrogate for a more fundamental parameter, namely, the crack tip strain rate, i .etP. Ford has proposed an empirical relation between &ct and K of the form:

let = B K4 ,

(2)

r I..- .

1 where B is a constant. De BWRVIP model can then be expressed in terms of det and new constants D and 6 = m/4 A

& = 6&ct . (3)

For cyclic loading there is an additional contribution to the crack tip strain rate. Shoji has argued that the crack tip strain rate is proportional to the CGR under cyclic loading in an inert en *-onment (air).3 hat is iet = &aar = . (4) where it is assumed that all of the crack growth during a cycle occurs during the rising load portien of the cycle of duration (n. Two expressions for da/dN were evamined. One is that used by Ford et al.2 in EPRI NP-50645: the other is that developed by L. James and D. Jones, which is the basis for the ASME Section XI fatigue crack growth cuives.

Differences between the two models are fairly small at low R. but quite large for high R

(>0.8). Both models were examined for use with the BWRVIP model. The James and Jones correlauon seems to give somewhat better results and has been used for the current studies. The proportionality factor in Eq. (4) can be inferred from some of Shoji's finite

• element results or can be treated as a fitting parameter. We have chosen the value proposed by Ford et al.2 let = .

(5) where a is in em, t in sec.

The BWRVIP model modified to include cyclic loading can then be written as

&=d +BK .

(6)

The modified BWRVIP crack growth model appears to provide good correlation with available crack growth rate data as shown in F1gs. la and Ib. The distribution of the data about the model is approximately log-normal as assumed in EPRI TR-105873. The results shown in Fig. I are for the "best fit" correlationwssentially the median of the log-normal distribution. For dispostuon the BWRVIP proposes to use the 95th percentile distribution of the fitted log-normal distribuuon, which for the constant load case corresponds to an increase in the calculated CGR by a factor of 10.8 and results in a correlation that actually bounds =97% of the data. De data base used in this comparison consists of 122 data points primartly from GE and ABB, which was used to develop the original constant-load correlation, and an additional data base of 152 data points from ANL. The statistics of the combined GE, ABB, and ANL data bases are about the same as those estimated by BWRVIP based only on the GE and ABB.

l l

~

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

,}

• 4 10* ,.

I 1

3 .

g ie* r k h=a"a*Ea=* '

. .. 3  ;

] .r .. r . .

.si .

,  : )u g

/ .

I* ; . :.e a )  :

E ,0- I ..g,e:* ,

.., yg ..

ss. ..s.1 a

), .-s-

-- ~ r ae-* -

.e n.os sa -

10

• e to* to* to* so* 1d' est e.1 1 to 100 1800 con sustant s sier.e (a) (b) ngure 1.(a) Comparison qf experimental and predicted CGRsfor the BWRVIP model modffled to accountfor cyclic loading: (b) Cumulative distribution qf the tutto qf the i.qa,-;,mr.tal to the predicted CGRs, 'Best Estimate" assumes no prior distribution when ~1~'n*\g the cumulative distribution. " log-normal" assumes a log-normal distnbution De data used for the comparison covers a wide range of material condicons (non-senMtized, sensitized by a variety of heat treatments. Type 304 SS. Type 316NG SSs, cast stainless stee's), solution conductivities, and electrochemical potentials. No obvious trends in the errors associated with the model are observable over the range of ECPs and conducuvides in the data base as shown in Figs 2a and 2b. Because it is an hmpirical correladon, it can predict some "unphysical" results when extrapolated outside of the datahame. For example, it is well established that at low ECPs, impunty addiuons (conducuvity increases) are less deletenous than at high ECPs. In the BWRVIP model the ECP and conductMty behavior are uncoupled. However, over the range of water chemistry condicons in the database, which bound those expected in operating reactors, the simple BWRVIP model seems to give adequate results. Because the database includes data from three different organizauons (-1/4 of the data is from GE.1/4 from ABB, and 1/2 from ANL), the

• errors" in the model also incorporate uncertainues in experimental procedures and measurements of quanuues such as ECP. No obvious " lab trend

• in errors is observed.

One somewhat surprising feature of the BWRVIP correlation is that it predicts that CGRs are independent of the degree of sensitization,' at least as measured by the electrochemical potenuokineuc rescuvadon (EPR) test. Although the BWRVIP correlation was developed using only data on sensitized Type 304 SS, the database used for the present comparison includes -65 data points for Types 316NG and 347 SSs and CF-3 and CF-8 and BM cast SSs. De correladon seems to predict CGRs for the sensitized and nonsensitized matenals equally well. De lack of dependence on sensitization seems countenntuitive, and does not necessanly imply that sensitization is not imponant. It could play a educal role in determining crack inittauon and the value of the threshold stress intensity needed to sustain a signi8 cant COR. even ifit has little effect on the OGR once the crack is acuvely growing. Mechanssucally this could occur, for example, if the rate limiting process for crack propagation ns cathodic reduction at surface of the matenal, as Kassner and MacDonald have suggested, rather than anodic dissolution at the crack tip. De BWRVIP l

, .s a 5

1. .. , ,,..., W ,.

,,,.y ...., .... ,, ,,..,y

" < r:

r  : i i o.ot b  !

ess rf'd 4 J )*

p osoig ..f. .f...l, .g om , -(

, ,(- ,( ,( , ,',,

sCP (mV/sNs) N"I l (a) (b)

F1gure 2.(N vadauon in the ratio gthe predicted CGR to the e+,e .cr.taUy observed CGR as afunction dECP; (b) Vadation in the ratio gthe predicted CGR to the expenmentaUy observed CGR as afunction gco@"M; model does not attempt to account for or predict any threshold behavior, which might indeed be dependent on EPR. Hence, it does tend to give more conservauve predictions for lower stress intensity values that might be below threshold levels.

One hmitauon of the database is that it consists solely of unirradiated, base materials.

The heat affected zones of weldments could have somewhat higher impunty levels than these base materials and neutron fluence will produce a number of changes in the material.

There are very few (if any) valid crack growth measurements for irradiated materials. The ,

discussion in EPRI TR-105873 on the applicability of the model to irradiated matedals is limited. 'Ihe benchmark compadsons with ultrasonic (UT) measurements on actual core shroud cracks presented in the report are not very conclusive because the uncertainues in the UT measurements and the residual stresses could mask eas0y vadauons in the CGR correlation.

Irradiation will produce redistribution of alloying elements, i.e., chromium, and impurity elements in the matedal. However, the relauvely weak dependence observed for the unirradiated materials on sensitization, which also results in a redistribuuon of alloying and impurity elements, suggests that although the re& distribution of elements under irradiation probably plays a cdtical role in initinuon and the determination of thre hold stress intensity values for significant CGRs. it may not be the rate limiting process for active crack growth. In addiuon to the redistdbution of alloying and impurity elements, irradiation also markedly affects the mechanical properties of the material. It appears hkely that radiation hardening would lead in a decrease in the crack tip deformadon rate for a given load and crack geometry, ahhough this could be offset by the introduction of altemate mechanisms for dislocation motion. This suggests that the a correladon based on unirradiated matedals could be conservative for such materials.1he Dmited data that are avadable are consistent with these assumption. Becsuse high levels of radiatien produce many changes in microstructure and properties which are not well understood, however, use of the of the correlation should be hmited to materials with fluence < 5x 10 20n/cm2

- _ _ _ _ - _ _ . _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ i

.,s s

m. .

e _....., ........... .. . . .,, . ..

= F 4 . F -j 4 m - 2 k s s -

5 l

g  :

ww  : g .

• • • j 10 .-

.: 5 -

e h j m b i

,,,f ,,,# ,,,,,, ,,,,: ,s

- ' ' '- >'- '- 8,

, ,o- sa .3 ,, ,, , e e.'i es e4 e4 e.e e4 eer e4 sn *"

(a) (b)

F1gure 3.la) Normalised K as afunction qf depthfor an H4 weldpom a BatteUejinite element solution; (b) Normaused K as afunction qf depthfor an H8 weldjoining a conical support to the shroud cyhnder '

i l

(E > 1 M eV). For fluences less than this, the redistdbution of alloying elements and , ,

impurides is quite limited, it can be argued that the primary change to the matedal from I the irradlauon is radlauon hardening, which as argued previously is likely, for a given set of '

environmental condluons and loads, to reduce the CGR compared with an unirradiated matedal.

In order to use a K-based CGR model such as the BWRVIP correlauon, estimates of the residual stresses as well as the applied loads are required . De prediction of CGRs in components depends strongly on both the CGR curve used and the residual stress distribudon. Finite element calculauons and experimental measurements of the residual stresses in core shroud weldments are presented in EPRI TR-105873. Recent calculadons done by Battelle under a subcontract to ANL give residual stress distobutions for the "H4-type' cylinder-to-cylinder weld that are in good agreement with those reported in EPRI TR-105873. The throughwall distdbuuon of the normalised K from the Battelle calculauons for an H4 weld shown in Fig. Sa can be compared with the throughwall distdbuuon given in Fig. 5-3 of EPRI TR-105873. The peak value in the Battelle distribuuon is somewhat higher (=30%) than the BWRVIP distdbuuon, but falls off faster and predicts arrest at a shallower depth. Calculations by M. Manahan for Nine Mile 1 give similar results. Addluonal expedmental results on a ' core shroud weldment from a canceled plant are also consistent with these results.3 For such distdbudons the choice of a CGR correlauon is almost unimportant--cracks will arrest =0.45 of the way through the wall with tt.e BWRVIP CGR correlauon,10 times the BWRVIP correladon, or 100 times the BWRVIP correlauon. For other cid geometdes the resvits are not so favorable. Results from the Battelle calculation for @e H8 weld are shown in Fig. 3b. His weld geometry, which may be unique to Nine Mile 1. joins the conical shroud support to the shroud cylinder. Rather than decreasing as the crack grows, the K values for this weld increase, and with a conductivity of 0.15 pS/cm and an ECP 200 MeV the predicted CGR after the crack has grown =0.3 throughwall is predicted to be 5.2 x 10 8 in/h (3.7 x 10-30 m/s).

Bligh0y in excess of the current NRR bounding CGR.

l

,,v s 7 Sample calculations of throughwall crack growth are shown in F1gs. 4 and 5. In Fig. 4a calculations with the BWRVIP 95th percentile cune a conductivity of 0.15 pS/cm and an ECP of 200 mV and the BWRVIP throughwall stress pronle are compared with the throughwall crack growth obtained assuming a Axed CGR of 2.2 x 10-5 in/h. With this throughwall stress distribution the cracks would never grow throughwall for any K-dependent CGR model and the constant CGR model is clearly consenative. Corresponding calculations with a more conservauve throughwall residual stress distnbution obtained by adding a uniform 10 kai stress to BWRVIF distnbution are shown in Fig. 4b. For this consenadve estimate of the throughwa!! residual stress proAle, the Axed CGR model and the K dependent model give almost the same predictions. 'Ibe cracks can grow throughwall, but the throughwall growth is greatly reduced for an ECP = -200 mV such as might be produced by a hydrogen water chemistry. As shown in F1g. 5, the proposed constant CGR of 2.2 x 10-8 in/h is also consistent with the crack growth expected in piping welds when the BWRVIP 95th percentile is used with the NRR 0313 residual stress distdbuuon and a typical applied stress of 15 ksi. The calculations show that the fixed crack growth rate model proposed in EPRI TR-105873 is conservative for " double V" plate-to-plate welds and butt piping welds.

The proposed BWRVIP correlation for a normal BWR water chemistry is close to the depostuon curve used by the Swedish regulatory authortues (SKI) as shown in Fig. 6. For -

normal water chemistries both the BWRVIP and SKI cunes predict CGRs somewhat lower than those obtained from the NUREG-0313 Rev 2 correladon. The BWRVIP correlauon for hydrogen water chemistry condicons is quite consenauve compared with curve adopted by SK! and the predictions of other models such as the GE Pledge code.

1 ..y ,,,,q ,,

..) , , ,

n es. - .

Asenhal+ asphoo +10 -e=

/ "

OCPea00 mV, ,

f Cenemo.16 us/em s ,

/ .

s /

/ ,!

l .

lK .

/

_ ..r,,socom n..,w + .ppi.a

'C"8" *V-C*nemo.is useem

' - > *1 J - -'

o', . . . .i ... a ,. ._, et o.o, o,i , so ,,, ,,,,

m 0.1 1 to too sono nme ( years) ' "" ( '")

(a) (b)

F1gure 4.(a) 1hroughwall cmck growth with aJUced rate of2.2 x 10~0 in/h and with the BWRVIP 95th percentile CGR curve and the expected residual and applied stresses (b)

Throughwall crack growth with the BWRVIP 95th percentile CGR curvefor normal water chemistry lECP = 200 mV) and a hydrogen water chemistry (ECP =-200 mV) and a conservattue estimate qfresidual and apptfed stresses where the residual stresses haue been increased by a un(form 10 kat

I~ ,, y

• r 8

(

\ *' 1 .., .., .., ,

o

/  :

/ .

ruures.

I ,

7hroughwallcrackgrowth uth ajbesd l rate g2.2 x W 5in/h and uMh the

. BWRVIP 95th percentile CGR cwve and l the NRR 0313 residual stress distribution j -.."PeSn

• and an applied stress (15 kst -

0.1

' ' '"A ' "A "A '

l 0.01 0.1 1 10 100 1000 l Time ( years) 10*

7 j

10*

to* rigune 6 l & <* '

g ,,. I e.,.: s

.-}-

Comparison gthe BWRVIP 95th percentile cwves with the NUREG 0313 0 s .. _1'. sewe 1 and S10 CGR curves to* -- a

... ,,.- , , , , , $8818-.- q

- _ . sw=n e.c  :

- . sww.4oo u.v -

to e ao so so  ;

K (IAPe m")

References 1.

T. Shoji, Quantitative Prediction (Enutronmentally Assisted Cracking Based on Crack 7Yp Strain Rafe, Proc. Conf. on Predictive Capabilities in Environmentally-Assisted Cracking. R. Rungta, ed., PVP Vol. 99. Amedcan Society of Mechanical Engineers, NY, pp.127-142 (1985).

2.

F. P. Ford. D. F. Taylor. P. L Andresen, and R Ballinger, Corrosion Assisted Cracking of Stainless and low-alloy Steels, EPRI NP.5064s Electric Power Research Institute, Palo Alto. CA (Febnaary 1987).

3.

E. A. Paysant. S. Spooner, X Zhu, C. R. Hubbard, S. T. Rosinski, and J. Dowickt.

Experimental Determination dResidualStress by Neutron _Digraction in a Bouing Water Reactor Core Shroud. Proc. of ASME PVP Conf., Vol. 322, American Society of Mechanical Engineers, New York, pp. 55 61 (1996).

{

"[" 9 RAI No.1 to SWRYlP

'Ibe 95th percentile crack growth curve presented in EPRI TR-105873 appears to give an acceptable description of CGRs in stainless steels under constant loads as a function of electrochemical potential, conductivity, stress intensity, and temperature based on laboratory data and in-reactor tests under conditions in which radiation levels are low recirculation piping). However, as indicated in the report (Fig. C-9) and as determined in recent tests in the Halden reactor, CGRs under high radiation condicons typical of the core (although the fluence levels of the materials in some cases are 1-4 orders of magnitude lower than those usuaDy considered necessary to produce IASCC) are often an order of magnitude or more greater than those predicted by the 95th percentile crack growth curve presented in EPRI TR-105873. The figure below compares the predicuons of the 95th percentile crack growth curve with data on sensitized materials with relauvely low fluence obtained in the Halden tests. The ~

model predicuons appear very unconservative. None of the proposed approaches for treating crack growth in the report (page 7-3) would appear to give reasonable results even if the observed CGRs were reduced by a factor of 4 to account for the high K values at which they were performed. Present arguments, analyses, and/or addiuonal data to explain these differences and justify the applicadon of the proposed crack growth curve to the analysis of crack growth in core shrouds. -

'!he major points in the BWRVIP response to Quesdon 1 are that (1) the Halden data are much higher than the predicuons of the BWRVIP model [by a factor of about 500 (2 and half orders of magnitude)) ,

(ii) the specimen geometries and loading conditions in the Halden tests are well outside the range required to obtain valid CGR measurements according to all applicable AS'IM standards; the tests were intended to produce accelerated CGRs to evaluate the effecuveness of reducing electrochemical potential (ECP) to mitigate cracking of core components

(!!!) the predicuons of the model are in reasonable agreement with the estimates of crack growth obtained from repeated inspecuons of cracked core shrouds.

ANL Conclusions and Recommendations s

We agree (and have always agreed) that the Halden tests are well outside the validity range for CGR measurements. We also believe that it is unlikely that there are mysterious new phenomena associated with core shroud cracking that would lead to much higher CGRs than have been observed in laboratory testing. However, we viewed the magnitude of the differences between the test results and the model predictions as larger than would be expected even though the results were not strictly valid.

Reviewing the design of the Halden specimen again, it is clear that it is a very unusual design. Although its designers went to a good deal of trouble to make a constant K design, on reflection, it is a little hard to understand just what K means for such a design, and it is difficult to come to a definitive answer overjust how far off an invalid test can be. Our Anal

,, c.

• 10 I

e o.3 .,;... . oJ ..

,,, .. ., 6 l

i u om L s

• =

as ?

t i e.ts } c.ts ,,

c.1 f . ,,j o.1 , _f. -j j 4 s.

1_o ,>. .. -

4 L

4 1 _. L. .

i AN a) .5 .o.os b "I i g e.* .t. .l. .t. ,t. .

4g i b- - if. f ii I = >1 i o a.es o. e.ts os oas e.s o oss o.i o.ss os oss o.s Creek Stewth, an. (sWRVIP ss%) Crack Stewth, in. (sWRYlP b.e.)

c.s  :.,. ... ... ...

i FM 7 3 Comparisons of 4dtrasonic measuromones of crack

$ 02 ;

, -l growth h core shrouds weh crook growth predictions for

$ o.,s .

(a) the BWRWP 95% confidence curve, tb) the BWRWP .

, 'best estimate *curn, and(c) a CGM curn 5 times that of a the *best estimate

• Curve. The bands indicate 1* $, *

~

5 5 uncertainty estimaw kr the UTmeasurements.

5 o .* .

j 4os F #8 i

,, o1* .. .i. ... ... i -

c.os o.1 c.ts o. oJs o.:

creet Growth, in. (swRviP s*s .)

conclusion is that the BWRVIP is correct and the Halden data cannot be used to judge the adequacy of the BWRVIP correlations. It is an

• apples

• and

• oranges

• comparison.

Discussion The BWRVIP response also makes the point that the most relevant CGRs to compare against are the field expedence. However, the data presented in Appendix J of BWRVIP-14 and Fig.1-5 of the response are insufficient to judge whether such compadsons actually support the BWRVIP CGR correlation. De calculations in Appendix J of BWRVIP-14 use the 95% confidence curve. This is appropdate for a dispostuon analysis, but it is inappropdate when comparing predictions to observations to judge the " goodness' of a model. If the "best estimate

• model is used, the predicted increments of crack growth should be reduced by a factor of 10.3. We presume that the calculations in Fig.1-5 of the response are abo based on the 95% confidence curve, which again is inappropdate if one is trying tojudge the goodness of the model. Also if one assumes that the uncertainty in the ultrasonic measurement of the crack depth is to.1-in., then the *ineertainty in the increment of crack growth will be larger (variances add). The data in F1g.1-5 of the response have been replotted in Fig. 7a: the corresponding results for the "best estimate

• CGR curve and CGRs 5 times greater than the "best estimate

• curve are shown in Fig. 7b

o , *- 11 and 7c. As might be expected, almost all of the measurements are in the " noise". It is impossible to make any judgment of which of the CGR curvo agree better with the " data."

RAI No. 2 to SWRVIP The procedures desenbed in Chapters 5 and 6 to compute the stress intensity K associated with a crack in a residual stress Seld assume that the redistnbution of the stresses as the crack propagates is completely clastic. As noted on page H-ll and in Fig. H-14, the finite element solutions show that as the crack grows, the stresses redistribute such that the stresses around the crack tip remain positive even for a crack with a/t = 3/4. De implication is that the crack can continue to grow. The stress intensity due to residual stresses proposed for the weldment (Table 5-1. F1g. 5-l 3)is negative at a/t = 0.45. De conclusion is that the crack would anest at a/t =0.45 Is there a real inconsistency here? Demonstrate by comparison to the finite element solutions that the proposed method of calculating K as the crack grows adequately accounts for the redistnbution of the residual stresses.

1 ANL Conclusions and Recommendations The revised text described in Appendix A of the response clartfles the apparent -

inconsistency. There is good agreement between the BWRVIP calculations for residual stress and the corresponding stress intensity factors and the results of the independent calculadons by Battelle. There is also sausfactory agreement between the approximate calculadons of the stress intensity factor using simphfied methods and-more rigorous finite-element calculadons.

The revision to Section 5-2 of BWRVIP-14 proposed in Appendix B appears to provide an improved method for calculating K in cylindrical structures. Previous solutions typically used either results for R/h=10, which is somewhat nonconservauve for higher R/h ratios.

or the edge crack soludon R/h==, which is very conservauve for most practical structures.

RAI No. 3 to BWRVIP Predicuons of crack growth depend very 'strongly on the assumed throughwall distribuuon of residual stress. Although several sets of expedmental measurements of surface stresses are presented, only one set of measurements of throughwall stresses are presented. How sensitive are these measurements to uncenainties in the postuon and values of the strain changes recorded on the surface as material is removed and on uncertainties in the values of the crack compliance functions C:(a) defined in Eq. (2) of Appendix F.

ANL Conclusions and Recommendations he discussion addresses the issue satisfactorily. As is noted in the response to RAI No. 4 there are now available addtuonal experimental measurements obtained by neutron diffraction that suppon the previous experimental work and the analytical estimates of the residual stresses.

I

p  ;

e**

12 l

RAI No. 4 to BWRVIP lt is difBeult to judge the adequacy of the proposed " generic" residual stress based solely on compansons between the residual stress obtained from finite element calculation and measurements and the proposed distdbution, because the controlling parameter is really the stress intensity. Compare the stress intensity computed from the proposed residual stress distnbution with those computed from the measured and finite element solutions to assure that the preycxd distnbution is conservative.

ANL Conclusions and Recommendations >

1he intent of the question was to focus on the comparison of the approximate K to the more exact K solution rather than the companson of the ayy.- : ate residual stress distdbution with the more exact distdbution. The response covers more ground, but it does address the quesuon antisfactorily. At least for cylinder-to-cyclinder welds, the residual stress and fracture mechanics analyses,i.e., K calculations, seem to be well done with good agreement between different analytical approaches and mignineant experimental confirmauon of the results.

RAI No. 8 to BWRVIP -

The example in Chapter 6 and the comparison with field experience in Appendix I focus on throughwall growth. Since the overall integrity of the structure depends primarily on the net cross section area, it is important to examine also the rate at which cracks extend around the circumference of the shroud. Comparisons of the CGRs with those observed experimentally could also aid in the assessme*nt of the adequacy of proposed CGR curves. Extend the analysis Section 6.2 to include circumferendal crack growth.

ANL Conclusions and Recommandstions For disposinon purposes BWRVIP proposes to use a limiting CGR of 5 x 10-5 in/h to predict the increase in crack length. This is acceptable. However, as noted in our discussion of the response to RAI No.1, a "best estimate" calculation of the increase in crack length provides a useful" reality check" on the overall model. As noted there, the "best esumate* ofincrease in crack length based on the BWRVIP CGR correlation is less than 1 mm/y. ,

RAI No. S to BWRVIP It is suggested that the stress intensity factor for a non-uniform distnbution of stress is computed using the method of Buchalet and Bamford, which is stdctly applicable only to full 367 cracks, that the stress intensity corresponding to a crack which is less than 360* in extent can be conservatively estimated by using the corresponding reduction in the M factor of Zahoor (Eq. 5.-2) which is dedved for a uniform stmas.

What is the basis for the assertion that this procedure is conservative?

(

pM k

, 13 ANL Conclusions and Recommendations h respsnse provides a very full and coue..cing answer.

h a

4, e

t 4

1 J