Fatigue Reactor Components

EPRI Project ManagerS. RosinskiJ. Carey

EPRI • 3412 Hillview Avenue, Palo Alto, California 94304 • PO Box 10412, Palo Alto, California 94303 • USA800.313.3774 • 650.855.2121 • [email protected] • www.epri.com

Proceedings of the 2000International Conference on Fatigueof Reactor Components (MRP-46)PWR Materials Reliability Program (PWRMRP)

1006070Also referenced as OECD/NEA/CSNI/R (2000) 24

Proceedings, June 2001

Cosponsors

Organisation for Economic Co-operation and Development(OECD) Nuclear Energy Agency/Committee on the Safety ofNuclear Installations (NEA/CSNI)

U.S. Nuclear Regulatory Commission

DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES

THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS ANACCOUNT OF WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER RESEARCHINSTITUTE, INC. (EPRI). NEITHER EPRI, ANY MEMBER OF EPRI, ANY COSPONSOR, THEORGANIZATION(S) BELOW, NOR ANY PERSON ACTING ON BEHALF OF ANY OF THEM:

(A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR IMPLIED, (I)WITH RESPECT TO THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, ORSIMILAR ITEM DISCLOSED IN THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESSFOR A PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON ORINTERFERE WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PARTY'S INTELLECTUALPROPERTY, OR (III) THAT THIS DOCUMENT IS SUITABLE TO ANY PARTICULAR USER'SCIRCUMSTANCE; OR

(B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY WHATSOEVER(INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTATIVEHAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES) RESULTING FROM YOURSELECTION OR USE OF THIS DOCUMENT OR ANY INFORMATION, APPARATUS, METHOD,PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT.

ORGANIZATION(S) THAT PREPARED THIS DOCUMENT

EPRI

ORDERING INFORMATION

Requests for copies of this report should be directed to EPRI Customer Fulfillment, 1355 Willow Way,Suite 278, Concord, CA 94520, (800) 313-3774, press 2.

Electric Power Research Institute and EPRI are registered service marks of the Electric PowerResearch Institute, Inc. EPRI. ELECTRIFY THE WORLD is a service mark of the Electric PowerResearch Institute, Inc.

Copyright © 2001 Electric Power Research Institute, Inc. All rights reserved.

iii

CITATIONS

This proceedings was prepared by

EPRI1300 W.T. Harris Blvd.Charlotte, NC 28262

Principal InvestigatorS. Rosinski

This proceedings contains information presented at the First International Conference on Fatigueof Reactor Components, sponsored by EPRI, OECD/NEA/CSNI, and the U.S. NRC.

The report is a corporate document that should be cited in the literature in the following manner:

Proceedings of the 2000 International Conference on Fatigue of Reactor Components (MRP-46): PWR Materials Reliability Program (PWRMRP), EPRI, Palo Alto, CA, and Organisationfor Economic Co-operation and Development (OECD/NEA/CSNI/R[2000] 24), and the U.S.NRC: 2001. 1006070.

v

REPORT SUMMARY

This report contains information presented at the First International Conference on Fatigue ofReactor Components held July 31–August 2, 2000, in Napa, California. The conference—sponsored by EPRI, Organisation for Economic Co-operation and Development Nuclear EnergyAgency/Committee on the Safety of Nuclear Installations (OECD NEA/CSNI), and the U.S.Nuclear Regulatory Commission (U.S. NRC)—provided a forum for the technical discussion offatigue issues that affect the integrity and operation of light water reactor components.Approximately 90 fatigue experts, representing 12 countries, participated in the conference.Strong representation was shown by nuclear operators, vendors, regulatory agencies, researchand development organizations, and other experts.

BackgroundFatigue is a primary degradation mechanism that affects nuclear power plant componentsworldwide. The effective management of fatigue is important to the continued safe operation ofplant components during present operation and as plants consider long-term operation. The EPRIMaterials Reliability Program (MRP) identified the need to bring together international expertsto discuss significant fatigue issues affecting nuclear plant operations in order to share commonexperiences and identify outstanding technical issues.

Objectivesx To provide a forum for the technical discussion of fatigue issues that affect the integrity and

operation of light water reactor components

x To share common experiences regarding fatigue of reactor components to ensure continuedsafe operation

x To identify common areas of interest in order to foster future international research andcollaboration activities

ApproachThe conference was organized in a series of technical presentations and group discussionsessions focused on the following fatigue-related topics:

x Thermal fatigue

x Environmental fatigue

x Nondestructive evaluation/testing

x Fatigue monitoring/evaluation

x Codes and standards

vi

x Vibration/high cycle fatigue

x The conference was structured to benefit utility and plant managers as well as system,materials, structural integrity, licensing, and maintenance/repair engineers.

ResultsApproximately 90 fatigue experts, representing 12 countries, participated in the conference.Strong representation was shown by nuclear operators, vendors, regulatory agencies, researchand development organizations, and other experts. At the conclusion of the conference, majordiscussion points were summarized. While not intended to represent a comprehensive list ofconference conclusions endorsed by all participants, these points were the subject ofconsiderable discussion during the conference and may be used to foster future internationalresearch and collaboration activities. Based on the degree of technical exchange that occurredand the breadth of information provided during the conference, the participants recommendedthat another conference on this topic be held in 18–24 months.

EPRI PerspectiveFatigue management is an important aspect of the continued safe operation of plant components.Periodic discussion of fatigue-related issues in an international forum allows the sharing ofcommon experiences and fosters international collaboration in the resolution of fatigue issues. Itis anticipated that future conferences in this series will continue to be a major forum for thediscussion of plant component fatigue issues.

KeywordsThermal fatigueEnvironmental fatigueFatigue evaluation

vii

ACKNOWLEDGMENTS

Appreciation is extended to the conference organizers for their contributions:

EPRIS. RosinskiJ. Carey

OECDE. Mathet

U.S. NRCE. Hackett

And to the following individuals for preparing this proceedings volume:C. LaundonL. Perry

ix

CONFERENCE SUMMARY

The primary objective of the conference was to provide a forum for the technical discussion offatigue issues that affect the integrity and operation of light water reactor components.Approximately 90 fatigue experts, representing 12 countries, participated in the conference.Strong representation was made by nuclear operators, vendors, regulatory agencies, research anddevelopment organizations, and other experts.

Technical discussions were focused in the following areas:

Thermal FatigueEnvironmental FatigueNondestructive Evaluation/TestingFatigue Monitoring/EvaluationCodes and StandardsVibration/High Cycle Fatigue

Following the technical presentations, a general discussion was held to summarize major pointsidentified by various speakers during the conference. While not intended to represent acomprehensive list of conference conclusions endorsed by all participants, these points receivedconsiderable discussion during the conference and may be used to foster future internationalresearch/collaboration activities. The major discussion points identified by the participants areprovided below.

x Collaboration and cooperation on an international scale are critical to the success of resolvingfatigue issues, including sharing of data, test programs, and theories.

x It is recognized that conservatism exists in the ASME Code fatigue design procedures.Plant-specific analyses using actual plant operating parameters (transient occurrence andseverity) may significantly reduce the conservatism in overall fatigue usage factordetermination.

x Significant advancements have been made in the international community regarding theeffects of thermal fatigue and reactor water environment. Additional research andinternational collaboration are recommended in these areas in order to resolve technicalissues and utilize the results of these efforts in various operating plant criteria.

x An understanding should be developed between ASME Code analysis and laboratory testingregarding reactor water effects on fatigue life.

x

x The characterization of thermal hydraulic phenomena is complex and an important aspect inthe quantification of thermal fatigue. Additional work regarding the proper characterizationof thermal hydraulic phenomena is recommended.

x A background document regarding the development of implicit fatigue design criteria inB31.1 should be developed.

x The following fatigue design Code changes/improvements were discussed:

– Modification of low cycle fatigue analysis procedures

– Addition of thermal fatigue analysis procedures for Class 2 piping

– Addition of “warnings” in Class 1, 2, and 3 design codes for dead legs/stratificationand mixing tees with corresponding thresholds

– Changes in the existing Code fatigue design S-N curves based on additional data andthe further evaluation of these data

– Determination of updated reduction factors

– Differentiation between thermal and mechanical loads

– Consideration of surface striping/craze cracking

x The development of expert tools is recommended to provide a larger understanding of fatiguedegradation mechanisms.

x Weld overlay repairs were presented as an effective method for repairing leaking standardsocket welds and providing sufficient fatigue resistance to operate to the next outage andbeyond. In addition, welding process and geometry enhancements were reported to improvethe fatigue life of socket welds.

x Instrumentation and monitoring can confirm the existence of high cycle thermal loads, exceptfor high frequency fluctuations.

x Field experience indicates that the relatively limited number of locations that experiencethermal fatigue are due primarily to the following:

– Stratification

– Dead legs and vortex conditions without a leak

– Reversing zones with large 'T

x Risk-informed considerations, including reactor operating experience, should be applied tothe management of fatigue technical issues.

Based on the degree of technical exchange that occurred and the breadth of information providedduring the conference, the participants recommended that another conference on this topic beheld in 18-24 months.

xi

ATTENDEE LIST

Frank AmmiratoEPRI1300 W.T. Harris Blvd.Charlotte, NC 28262704-547-6129 / 704-547-6168 (fax)[email protected]

Robert BainStone & Webster Engineering245 Summer St.Boston, MA 02210617-589-2109 / 617-589-1315 (fax)[email protected]

Gabriel BarslivoSwedish Nuclear Power InspectorateSE-106 58 StockholmSweden46-869-88660 / 46-866-19086 (fax)[email protected]

Alan BilaninContinuum Dynamics, Inc.34 Lexington AveEwing, NJ 08618-2302609-538-0444 x108 / 609-538-0464 (fax)alan@continuum_dynamics.com

Urs R. BlumerCCI AGPO Box8404 WinterthurSwitzerland41-52-262-5936 / 41-52-262-0039 (fax)[email protected]

Scott BondAmeren UEPO Box 620Fulton, MO 65251573-676-8519 / 573-676-4334 (fax)[email protected]

Kevin BratcherFramatome Technologies Inc.3315 Old Forest Rd.Lynchburg, VA 24506804-832-2789 / 804-832-2831 (fax)[email protected]

John CareyEPRI3412 Hillview Ave.Palo Alto, CA 94304-1344650-855-2105 / 650-855-7945 (fax)[email protected]

Bob CarterEPRI1300 W.T. Harris Blvd.Charlotte, NC 28262704-547-6019 / 704-547-6035 (fax)[email protected]

Kenneth ChangAmerican Electric Power500 Circle DriveBuchanan, MI 49107616-697-5525/616-697-5570 (fax)[email protected]

xii

Omesh ChopraArgonne National Laboratory9700 South Cass Ave., Bldg. 212Argonne, IL 60439630-252-5117/630-252-3604 (fax)[email protected]

Roy CorieriNiagara MohawkPO Box 63Lycoming, NY 13093315-349-4051 / 315-349-1581 (fax)[email protected]

Kurt CozensNuclear Energy Institute1776 I Street, NW - Suite 400Washington, DC 20006202-739-8085 / 202-785-1898 (fax)[email protected]

J. Michael DavisDuke Energy Corp.526 South Church Street - M/C EC090Charlotte, NC 28202704-382-1784 / 704-382-3993 (fax)[email protected]

Art DeardorffStructural Integrity Associates3315 Almaden Expwy - Suite 24San Jose, CA 95118408-978-8200 / 408-978-8964 (fax)[email protected]

Shashi DharNiagara MohawkPO Box 63Lycoming, NY 13093315-349-4732 / 315-349-1836 (fax)[email protected]

Hoang DinhDuke Energy Corp.13225 Hagers Ferry RoadHuntersville, NC 28078704-875-5675/704-875-4364 (fax)[email protected]

Frederic DulcereEDF6, Quai Watier78401 ChatouFrance33-130-877806 / 33-130-877323 (fax)[email protected]

Robin DyleSouthern Nuclear Operating Co.40 Inverness Center ParkwayBirmingham, AL 35242205-992-5885 / 205-992-5793 (fax)[email protected]

Claude FaidyEDF12-14, Avenue Dutrievoz69628 Villeurbanne CedexFrance33-472-82-7279 / 33-472-82-7699 (fax)[email protected]

Glenn GardnerNortheast UtilitiesPO Box 128Waterford, CT 06385860-440-0373/860-440-2025 (fax)[email protected]

Yogen GarudAPTECH Engineering Services1282 Reamwood Ave.Sunnyvale, CA 94089408-745-7000 / 408-734-0445 (fax)[email protected]

xiii

Colombe GomaneIPSNBP 692260 Fontenay aux RosesFrance46-54-8887 / 47-46-1014 (fax)[email protected]

Jeff GoodwinSouthern Company ServicesPO Box 2625Birmingham, AL 35202205-992-6962/205-992-0324 (fax)[email protected]

Steve GosselinPacific Northwest National Laboratory902 Battelle Blvd, MS-K526Richland, WA 99352509-375-4463/509-375-6497 (fax)[email protected]

Prasoon GoyalToledo Edison5501 North State Route 2Oak Harbor, OH 43449419-321-7351 / 419-249-2342 (fax)[email protected]

Mark GrayWestinghouse Electric Co.PO Box 355Monroeville, PA 15146412-374-6481 / 412-374-6277 (fax)[email protected]

Michael GuthrieCarolina Power and Light Co.3581 West Entrance RdHartsville, SC 29550843-857-1049 / 843-857-1368 (fax)[email protected]

Makoto HiguchiIHI Co., Ltd.1 Shin-nakahara, Isogo-kuYokohama, 2358501Japan81-45-759-2194 / 81-45-759-2125 (fax)[email protected]

Paul HirschbergStructural Integrity Associates3315 Almaden Expwy - Suite 24San Jose, CA 95118408-978-8200 / 408-978-8964 (fax)[email protected]

Chris HoffmannWestinghouse - CE Nuclear Power2000 Day Hill Rd. - Dept 9483-1903Windsor, CT 06095860-285-4929 / 860-285-4232 (fax)[email protected]

John HoffmanVermont Yankee Nuclear Power Corp546 Governor Hunt RdVernon, VT 05354-9766802-451-3095 / 802-451-3030 (fax)[email protected]

Jon HornbuckleSouthern Nuclear Operating Co.40 Inverness Center ParkwayBirmingham, AL 35201205-992-7974 / 205-992-6108 (fax)[email protected]

Ping HsuSouthern Nuclear Operating Co.42 Inverness ParkwayBirmingham, AL 35242205-992-0989 / 205-992-0234 (fax)[email protected]

xiv

Karl JacobsNew York Power Authority123 Main StreetWhite Plains, NY 10601914-739-5276 / 914-287-3710 (fax)[email protected]

Douglas KalinouskyUS NRCM/S T10 E10Washington, DC 20555301-415-6788 / 301-415-5160 (fax)[email protected]

Dietmar KalkhofPaul Scherrer InstitutCH-5232 Villigen PSISwitzerland41-563-102620 / 41-563-102199 (fax)[email protected]

Ariadni KapsalopoulouNew Jersey Dept. of Environmental Protection33 Arctic Parkway - PO Box 415Trenton, NJ 08625609-984-7539 / 609-984-7513 (fax)[email protected]

Nikolai KarpuninSEC NRS14/23, Avtozavodskaya St.Moscow, 109280Russia795-275-5548 / 795-275-5548 (fax)[email protected]

Greg KesselDominion Generation (Virgina Power)5000 Dominion Blvd. 1NEGlen Allen, VA 23060804-273-3604/804-273-3614 (fax)[email protected]

Suresh KhatriPG&E Co.PO Box 56Avila, CA805-545-4169 / 805-545-6515 (fax)[email protected]

Dae Whan KimKorea Atomic Energy Research InstitutePO Box 105Yusung, Taejon 305-600Korea82-42-868-2046 / 82-42-868-8346 (fax)[email protected]

In Sup KimKorea Advanced Inst. of Science & Technology373-1 Kusong-dongYusong-ku, Tajeon 305-701Korea82-42-869-3815 / 82-42-869-3810 (fax)[email protected]

Wolfgang KornerSwiss Federal Nuclear Safety InspectorateCH-5232 Villigen-HSKSwitzerland41-56-3103960/41-56-3103854 (fax)[email protected]

Jan LagerstromRinghals AB430 22 VarobackaSweden46-640-667624 / 46-340-660186 (fax)[email protected]

Jean-Alain Le DuffFramatomeEE/S - BALF0721 A Tour Framatome92084 ParisFrance33-479-63078 / 33-479-60501 (fax)[email protected]

xv

Sang Gyu LeeKorea Advanced Inst. of Science & Technology373-1 Kusong-dongYusong-ku, TajeonKorea82-42-869-3855 / 82-42-869-3895 (fax)[email protected]

Young Ho LeeKorea Advanced Inst. of Science & Technology373-1 Kusong-dongYusong-ku, TaejonKorea82-42-868-3855 / 82-42-868-3895 (fax)[email protected]

Bob LisowyjOmaha Public Power DistrictPO Box 399Fort Calhoun, NE 68023402-533-6491 / 402-533-7390 (fax)[email protected]

Eduardo ManeschyEletrobras Temonuclear S.A. - EletronuclearRua da Candelaria 65 - 6 AndarRio de Janeiro, RJ 20091-020Brasil55-21-588-7681 / 55-21-588-7260 (fax)[email protected]

Ted MarstonEPRI3412 Hillview Ave.Palo Alto, CA 94304-1344650-855-2997 / 650-855-2213 (fax)[email protected]

Eric MathetOECD Nuclear Energy AgencyLe Seine St-Germain - 12, boulevard des lles92130 Issy-les-MoulineauxFrance33-145-24-1057 / 33-145-24-1110 (fax)[email protected]

Klaus MetznerPreussenElektra Kernkraft5 TresckowstrHannover D 30457Germany49-511-439-4009/49-511-439-4377 (fax)[email protected]

Frank MichelGRSSchwertnergasse 150667 KölnGermany49-221-2068-753 / 49-221-2068-888 (fax)[email protected]

Todd MielkeWisconsin Electric Power Co.6610 Nuclear Rd.Two Rivers, WI 54241920-755-6376 / 920-755-6032 (fax)[email protected]

Itaru MuroyaMitsubishi Heavy Industries2-1-1 Shinhama Arai-cho TakasagoHyogo 676-8686Japan81-794-45-6800 / 81-794-45-6080 (fax)[email protected]

Joseph MuscaraU.S. Nuclear Regulatory CommissionWashington, D.C. 20555301-415-5844 / 301-415-5074 (fax)[email protected]

Jeong NaEnergy Services Group8979 Pocahontas TrailWilliamsburg, VA 23185757-864-8471 / 757-864-4914 (fax)[email protected]

xvi

Takao NakamuraThe Kansai Electric Power Co., Inc.3-3-22, Nakanoshima, kita-kuOsaka 530-8270Japan81-6-6441-8221/81-6-6441-4277 (fax)[email protected]

Gilles NavarroEDF6, Quai Watier78401 ChatouFrance33-130-88571 / 33-130-877323 (fax)[email protected]

Robert NickellApplied Science & Technology16630 Sagewood LanePoway, CA858-485-9024 / 858-485-9024 (fax)[email protected]

Paul NorrisSouthern Nuclear Operating Co.Bin B063, PO Box 1295Birmingham, AL 35242205-992-5718 / 205-992-5793 (fax)[email protected]

Hitoshi OhataJapan Atomic Power Company1-6-1, Ohtemachi, Chiyoda-kuTokyo 100-0004Japan81-3-3201-7087 / 81-3-3215-3930 (fax)[email protected]

Frank OtrembaUniversity of StuttgartPfaffenwaldring 32D-70569 StuttgartGermany49-711-685-2601 / 49-711-685-3053 (fax)[email protected]

Susan Otto-RodgersEPRI1300 W.T. Harris Blvd.Charlotte, NC 28262704-547-6072 / 704-547-6168 (fax)[email protected]

Michael RobinsonDuke Energy Corp.526 South Church Street, EC090Charlotte, NC 28202704-373-3522/704-382-3993 (fax)[email protected]

Eberhard RoosUniversity of StuttgartPfaffenwaldring 32D-70569 StuttgartGermany49-711-685-2604 / 49-711-685-3144 (fax)[email protected]

Stan RosinskiEPRI1300 W.T. Harris Blvd.Charlotte, NC 28262704-547-6123 / 704-547-6035 (fax)[email protected]

Guy RousselAVNAvenue Du Roi 157B-1190 BrusselsBelgium32-25-368-333 / 32-25-368-585 (fax)[email protected]

Suresh SahgalNOK, Nuclear Power Plant BeznauCH-5312 DoetingenSwitzerland41-56-2667-84/41-56-2667702 (fax)[email protected]

xvii

Takeshi SakaiJapan Atomic Power Company1-Banchi, Myojin-cho, Tsuruga-shiFukui-kenJapan81-770-26-1111 / 81-770-26-8081 (fax)[email protected]

Andrew SiemaszkoFirstEnergy Corp.5501 North State Route 2 - M/C DB-1056Oak Harbor, OH 43449419-321-7341 / 419-249-2340 (fax)[email protected]

Fred SimonenPacific Northwest National LaboratoryPO Box 999Richland, WA 99352509-375-2087- / 509-375-6497 (fax)[email protected]

Ram SingalNiagara MohawkPO Box 63 (M/C ESB2)Lycoming, NY 13093315-349-4480 / 315-349-1579 (fax)[email protected]

Larry SmithConstellation Nuclear Services, Inc.1650 Calvert Cliffs ParkwayLusby, MD 20657410-793-3416 / 410-793-3431 (fax)[email protected]

Jussi SolinVTT Manufacturing TechnologyPO Box 170402044 VTTFinland358-9-4566875 / 358-9-4567002 (fax)[email protected]

Les SpainVirginia Power Co.5000 Dominion BlvdGlen Allen, VA 23060804-273-2602 / 804-273-3877 (fax)[email protected]

Jean-Michel StephanEdFRoute De Sens EcuellesF-77818 Moret-Sur-Loing CedexFrance33-1 60 73 60 85 / 33-1 60 73 65 59 (fax)[email protected]

Yu-Shing SunPECO EnergyCB-63B-3 965 Chesterbrook Blvd.Wayne, PA 19087610-640-6137 / 610-640-6582 (fax)[email protected]

Bruce SwoyerPP&L, Inc.2 North Nine Street (GENA62)Allentown, PA 18101-1179610-774-7643 / 610-774-7830 (fax)[email protected]

Mike TestaFirstEnergy Corp.PO Box 4Shippingport, PA 15077724-682-5552 / 724-682-5536 (fax)[email protected]

Raymond ToEntergy OperationsSR 333Russellville, AR 72802501-858-4371 / 501-858-4955 (fax)[email protected]

xviii

Kazuya TsutsumiMitsubishi Heavy Industries2-1-1 Shinhama Arai-cho TakasagoHyogo 676-8686Japan81-794-45-6723 / 81-794-45-6080 (fax)[email protected]

Louis Van Der WielMinistry of EnviornmentPO Box 90801The [email protected]

Bernard Van SantOmaha Public Power DistrictPO Box 399Fort Calhoun, NE 68023402-533-7385 / 402-533-6597 (fax)[email protected]

Edward WaisWais & Associates2475 Spalding Dr.Atlanta, GA 30350770-396-8797 / 770-396-4194 (fax)[email protected]

Mark WalkerDominion Generation5000 Dominion BlvdGlen Allen, VA 23060804-273-2226 / 804-273-3448 (fax)[email protected]

Stan WalkerEPRI1300 W.T. Harris Blvd.Charlotte, NC 28262704-547-6081 / 704-547-6168 (fax)[email protected]

Dan YuanSouthern California Edison14000 Mesa Rd., Bldg. G48BSan Clemente, CA 92674949-368-2478/949-368-2451 (fax)[email protected]

xix

CONTENTS

OPENING SESSION

1 EMBRACING THE FUTURE–INTERNATIONAL CONFERENCE ON FATIGUE OFREACTOR COMPONENTS.................................................................................................... 1-1

2 US UTILITY PROGRAM FOR FATIGUE MANAGEMENT .................................................. 2-1

3 ACTIVITIES OF THE OECD-NEA IN THE AREA OF THERMAL FATIGUE IN LWRPIPING.................................................................................................................................... 3-1

4 FATIGUE OF REACTOR COMPONENTS: NRC ACTIVITIES ............................................ 4-1

5 A REGULATOR'S VIEW ON THE FATIGUE ISSUE ........................................................... 5-1

THERMAL FATIGUE I

6 THERMAL FATIGUE IN FRENCH RHR SYSTEM............................................................... 6-1

7 RESULTS OF THERMAL STRATIFICATION MEASUREMENTS AT NUCLEARPOWER PLANT BEZNAU...................................................................................................... 7-1

8 LEAKAGE FROM CVCS PIPE OF REGENERATIVE HEAT EXCHANGER INDUCEDBY HIGH CYCLE THERMAL FATIGUE AT TSURUGA NUCLEAR POWER STATIONUNIT 2 .................................................................................................................................... 8-1

9 OPERATING EXPERIENCE REGARDING THERMAL FATIGUE OF UNISOLABLEPIPING CONNECTED TO PWR REACTOR COOLANT SYSTEMS ...................................... 9-1

10 AUXILIARY FEEDWATER LINE STRATIFICATION AND COUFAST SIMULATION ..... 10-1

11 FATIGUE EVALUATION IN PIPING CAUSED BY THERMAL STRATIFICATION ......... 11-1

xx

THERMAL FATIGUE II

12 CONSIDERATION OF THERMAL FATIGUE AND CYCLE MONITORING OF AB31.1 PLANT ....................................................................................................................... 12-1

13 MAIN RESULTS OF EDF’S EXPERIENCE ON IN SITU MEASUREMENTSRELATED TO THERMAL FATIGUE ON PWR REACTOR COOLANT PIPING ................... 13-1

14 EVALUATION OF OCONEE-2 HIGH PRESSURE INJECTION/NORMAL MAKEUP(HPI/NMU) LINE WELD FAILURE........................................................................................ 14-1

15 CURRENT ACTIVITIES ON GUIDELINES OF HIGH-CYCLE THERMAL FATIGUEIN JAPAN............................................................................................................................. 15-1

FATIGUE MONITORING/EVALUATION

16 STATUS AND UPDATE OF THE EPRI FATIGUEPRO FATIGUE MONITORINGPROGRAM ........................................................................................................................... 16-1

17 REMARKS TO THE DIFFERENT FACTORS INFLUENCING FATIGUE ANALYSISAND FATIGUE DESIGN CURVES ....................................................................................... 17-1

18 THE RUSSIAN REGULATORY APPROACHES IN THE FATIGUE EVALUATIONOF NPP CONSTRUCTION COMPONENTS......................................................................... 18-1

19 THE EVALUATION SYSTEM OF THERMAL STRATIFICATION STRESS USINGOUTER SURFACE TEMPERATURE ................................................................................... 19-1

NDE/NDT

20 MICROSTRUCTURAL CHANGES OF PRESSURE VESSEL STEEL DURINGFATIGUE IN HIGH TEMPERATURE WATER ENVIRONMENT........................................... 20-1

21 MICROSTRUCTURAL INVESTIGATIONS AND MONITORING OF DEGRADATIONOF LCF DAMAGE IN AUSTENITIC STEEL X6CRNITI 18-10 .............................................. 21-1

22 NDE TECHNOLOGY FOR DETECTION OF THERMAL FATIGUE DAMAGE INPIPING.................................................................................................................................. 22-1

23 NDE METHOD FOR DETECTION OF FATIGUE DAMAGE ............................................ 23-1

xxi

ENVIRONMENTAL FATIGUE I

24 ENVIRONMENTAL EFFECTS ON FATIGUE CRACK INITIATION IN PIPING ANDPRESSURE VESSEL STEELS ............................................................................................ 24-1

25 MECHANISM OF FATIGUE CRACK INITIATION IN LIGHT WATER REACTORCOOLANT ENVIRONMENTS............................................................................................... 25-1

26 ENVIRONMENTAL FATIGUE EVALUATION ON JAPANESE NUCLEAR POWERPLANTS................................................................................................................................ 26-1

27 EVALUATION OF ENVIRONMENTAL EFFECTS ON FATIGUE LIFE OF PIPING......... 27-1

28 FATIGUE LIFE REDUCTION IN PWR WATER ENVIRONMENT FOR STAINLESSSTEELS................................................................................................................................ 28-1

ENVIRONMENTAL FATIGUE II

29 AN APPROACH FOR EVALUATING THE EFFECTS OF REACTOR WATERENVIRONMENTS ON FATIGUE LIFE.................................................................................. 29-1

30 DESIGN BASIS ENVIRONMENTAL FATIGUE EVALUATION AT OCONEE ................. 30-1

31 AN UPDATED METHOD TO EVALUATE REACTOR WATER EFFECTS ONFATIGUE LIFE FOR CARBON AND LOW ALLOY STEELS ............................................... 31-1

32 ENVIRONMENTAL FATIGUE, CRACK GROWTH RATES IN TITANIUMSTABILIZED STAINLESS STEEL........................................................................................ 32-1

CODES AND STANDARDS

33 STRESS INTENSIFICATION FACTORS......................................................................... 33-1

34 FATIGUE EVALUATIONS USING ASME SECTION XI NON-MANDATORYAPPENDIX L ........................................................................................................................ 34-1

35 AN UPDATE ON THE CONSIDERATION OF REACTOR WATER EFFECTS INCODE FATIGUE INITIATION EVALUATIONS FOR PRESSURE VESSELS ANDPIPING.................................................................................................................................. 35-1

36 CODES AND THERMAL FATIGUE: STATUS AND ON-GOING DEVELOPMENT......... 36-1

xxii

VIBRATION/HIGH CYCLE FATIGUE

37 VIBRATION FATIGUE TESTING OF SOCKET WELDS, PHASE II ................................ 37-1

OPENING SESSION

1-1

1 EMBRACING THE FUTURE–INTERNATIONAL CONFERENCE ON FATIGUE OFREACTOR COMPONENTS

Theodore U. Marston, Ph.D.Vice President and Chief Nuclear Officer

30 July 2000Napa, CA

1-3

Theodore U. Marston, Ph.D.Vice President and Chief Nuclear Officer

30 July 2000Napa, CA

-Embracing the Future-

InternationalConference on

Fatigue of ReactorComponents

1-4

Presentation Outline

• Welcome and conference objectives

• US nuclear industry status

• Deregulation and privatization impacts

• EPRI nuclear status and future directions

• Future directions of the US nuclear program

• Role of fatigue in the current and future nuclearpicture

• Concluding remarks

1-5

Current U.S. Industry Status

• 103 plants generating ~25% of the U.S. electricity in 1/00– Improved, risk-informed regulatory process

• Maintain safety, more effective, efficient and realistic

– Objective measures show positive trends

• Capacity factor--up; refueling outage duration--down; forcedoutage rate--down; operating costs--decreasing

– License renewal is a reality

• Process taking less than 2 years and less than $15M

– First renewal granted to Calvert Cliffs end of March 2000

– Oconee granted 23rd of May, cost <$5/Kw(e)

– Public and stakeholder acceptance increasing

• Safety record

• Emission-free energy source and 1400 new Mw(e)

• Energy diversity and security

1-6

Current License Renewal SituationAlready filedAlready filedArkansas Nuclear One Unit 1Hatch 1,2

Already filedAlready filedAlready filedAlready filedArkansas NuclearArkansas Nuclear One Unit 1 One Unit 1Hatch 1,2Hatch 1,2

2000200020002000Turkey Point 3,4Turkey Point 3,4

2001200120012001Catawba 1,2Catawba 1,2McGuire 1,2McGuire 1,2Peach Bottom 2,3Peach Bottom 2,3North Anna 1,2North Anna 1,2Surry 1,2Surry 1,2

2002200220022002St. Lucie 1,2St. Lucie 1,2SummerSummerCrystal River 3Crystal River 3Fort CalhounFort Calhoun

20032003Arkansas Nuclear One Unit 2CooperFarley 1,2Robinson 2

2003200320032003Arkansas NuclearArkansas Nuclear One Unit 2 One Unit 2CooperCooperFarleyFarley 1,2 1,2Robinson 2Robinson 2

ApprovedApprovedCalvert Cliffs 1,2Oconee 1,2,3

ApprovedApprovedApprovedApprovedCalvert Cliffs 1,2Calvert Cliffs 1,2OconeeOconee 1,2,3 1,2,3

1-7

Current U.S. Industry Status (cont.)

• Extensive restructuring of generators– Successful efforts to reduce cost of doing business

• Transfer of assets, through sale and barter

• Consolidation of nuclear operating companies– Alignment of operating companies

• Joint purchasing agreements, some web-based

• Increased pressure on service providers to share risksand rewards, including EPRI

• Consolidation of suppliers and A/Es

• Retail electricity markets transition

1-8

U.S. Nuclear Picture

0

5

10

15

20

25

30

35

40

45

50

1990 2000 2010

Co

mp

an

ies

Active Companies Merged Exited Nuclear

1-9

U.S. Nuclear Plant Transactions

830 21

117136

16

44

378

447

0

100

200

300

400

500

600

Pilgrim TMI 1 Clinton NMP 1 NMP 2 Oyster Crk. Vt. Yankee BeaverValley

NYPA

Transaction

Pri

ce $

/kW

Hydroelectric

Geothermal

Oil

Gas

Coal

Electricity Industry Restructuring (2005)

lPower Marketers

l“National” Grid

lVarious Generators lAll Customers

lLDC’s

1-11

Current 2000 Cal PX Revenues

$0

$25,000,000

$50,000,000

$75,000,000

$100,000,000

$125,000,000

$150,000,000

$175,000,000

$200,000,000

$225,000,000

$250,000,0001 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57

Ranked Revenues Days

Dai

ly R

even

ues

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

Rev

enu

es

1-12

EPRI Nuclear Market Share(Equivalent Technology)

• U.S. market--97% membership• Latin America--1/3 membership• Asia--some project participation• Europe--51% membership/40% participation

Europe (134,244 MWe)

51%

40%

9%

Full Members

Funders

Non-participants

Asia (51,358Asia (51,358 MWe MWe))

75%

25%

Funders

Non-participants95%

5%

Full Members

Funders

EPRI ParticipationNorth America, MWe = 119,199

1-13

Why is EPRI Nuclear Successful in aDeregulated Business Environment?

• The global nuclear “fleet” sinks or sails together– Safety must be maintained

– Good public relations is still mandatory

– Need to use international benchmarks, concepts andsolutions to optimize the utilization of resources

• The real competition is not other nuclear plants,but natural gas-fired combined-cycle plants

• No one country (or company) has sufficientresources to go it alone

• It seems to work, but EPRI has to work harderand smarter each year

1-14

Management of NuclearManagement of NuclearRisk and Resources inRisk and Resources inCommercial NuclearCommercial Nuclear

Power PlantsPower Plants

Cost ManagementCost Management

SafetySafety

RiskRiskCommunicationCommunication

FuelFuel OperationsOperations MaintenanceMaintenance

Environmental BenefitsEnvironmental Benefits

Radiation &Radiation &Waste ManagementWaste Management

Low Level WasteLow Level Waste

Radiation FieldRadiation FieldReductionReduction

Spent FuelSpent FuelManagementManagement

AgingAgingManagementManagement

BWRBWRMaterialsMaterials

BOPBOPMaterialsMaterials

PWRPWRMaterialsMaterials

AssetAssetManagementManagement

Life CycleLife CycleManagementManagement

DecommissioningDecommissioning

Capital ProjectsCapital Projects

Profitable electricity andProfitable electricity andancillary services sales,ancillary services sales,

bibi-lateral, short-term,-lateral, short-term,spot, etcspot, etc

EPRI Nuclear Power Business OrganizationReflects Our Member’s Business

30%

30%

15%

10%

10%

1-15

Opportunities for New NuclearVarious Scenarios in E-EPIC

0.00

500.00

1000.00

1500.00

2000.00

2500.00

1990

1993

1996

1999

2002

2005

2008

2011

2014

2017

2020

2023

2026

2029

2032

2035

2038

2041

2044

2047

2050

Year

Cap

acit

y

Nuclear

Advanced Nuclear: Pres Admin

Advanced Nuclear: 2XC Seq+CapEx

Advanced Nuclear: Cap Ex

1-16

EPRI’s Future Direction in Nuclear

• To focus on high value-added work for the existing plantsand long term (>40 years) operation

– Shared risks and rewards

– Leveraged resources

• To expand the membership base internationally to bringeven greater value to all members

• To conduct cooperative R&D involving all stakeholders

• To reduce our cost of doing business, similar to all of ourmembers

• To expand the strategic part of our Nuclear R&D programto support the EPRI Energy Technology Roadmap andembrace the future

1-17

Conference Issues and Opportunities

• Fundamental understanding of fatigue is essential for thesuccess of current and future nuclear plants from safetyand cost perspectives

• Collaboration and cooperation on a global scale is criticalto our success, including sharing of data, test programsand theories

• Universal pressure to reduce R&D in general when wemay be on the verge of a nuclear renaissance

• Nuclear programs are in different states in differentregions of the world

• Electricity demand will grow, deregulation andprivatization will be the norm leading to new generationrequirements and real business opportunities

2-1

2 US UTILITY PROGRAM FOR FATIGUE MANAGEMENT

Michael R. RobinsonDuke Energy Corporation

Stan T. RosinskiJohn J. Carey

EPRI

Arthur F. DeardorffStructural Integrity Associates

presented atInternational Conference on Fatigue of Reactor Components

July 31 – August 2, 2000Silverado Country Club

Napa, California

2-3

US UTILITY PROGRAMS FORFATIGUE MANAGEMENT

by


Stan T. RosinskiJohn J. Carey

EPRI

Arthur F. DeardorffStructural Integrity Associates

presented at

International Conference on Fatigue of Reactor ComponentsJuly 31 – August 2, 2000Silverado Country Club

Napa, California

Technical PaperSIT-00-007

2-4

US UTILITY PROGRAMS FOR FATIGUE MANAGEMENT

Michael R. Robinson John J. CareyDuke Energy Corp. EPRI526 So. Church St. MC EC090 3412 Hillview Ave.Charlotte, NC 28292 Palo Alto, CA 94304

Stan T. Rosinski Arthur F. DeardorffEPRI Structural Integrity Associates1300 Harris Boulevard 3315 Almaden Expressway, Suite 24Charlotte, NC 28262 San Jose, CA 95118-1557

ABSTRACTThermal fatigue is one of the main causes of leakage from reactor coolant system piping. Although

nuclear plants were designed to accommodate the effects of cyclic operation, experience has shown thatsome loading conditions were not known at the time of initial plant design. As a result, fatigue relatedcracking and leakage is periodically encountered. In US plants, the response to cracking issues that aregeneric in nature are generally handled by vendor owner groups or through other industry-directedactivities. Thermal fatigue in non-isolable branch piping in pressurized water reactors is currently beingaddressed by the Materials Reliability Project Thermal Fatigue Issue Task Group (MRP TF-ITG). TheEPRI-managed project is targeted to release a Thermal Fatigue Management Guideline in late 2001. Thisproject, with related background information, is described.

BACKGROUNDMetal fatigue is one of the degradation mechanisms that was explicitly addressed in the design of US

nuclear power plants. In the formative years of nuclear plant design, it was recognized that there wouldbe significant pressure/temperature cycling in nuclear power plants. As a result, Section III of the ASMEBoiler and Pressure Vessel Code [1] was developing, requiring that the effects of cyclic operation onreactor vessels be considered. Requirements for nuclear piping system design were concurrentlydeveloped in USAS B31.7 [2], incorporated into ASME Section III in the early 1970’s. Because of theserequirements, modern reactor coolant pressure boundary components of nuclear plants were designed tomeet Code requirements for a specified set of cycles, expected to be bounding at the time of the initialplant design. At most plants, the number of design transients was listed in plant Technical Specifications,with the number of design cycles being a limitation on plant operation.

It was also recognized that inspection of nuclear systems for evidence of cracking or leakage wouldprovide additional assurance of pressure boundary integrity. Thus, Section XI of the Boiler and PressureVessel Code [3] was prepared, mandating a program of visual, surface and volumetric inspection at keylocations. In the original Section XI Code, high stress/high usage factor locations were given the highestpriority for inspection.

The plant design, operation and inspection requirements have generally been successful, in thatleakage or cracking due to fatigue has not resulted from conditions considered in the original design.However, there have been a number of occurrences of fatigue due to conditions not known or expected tooccur at the time of the original plant design. In each case, the nuclear industry has responded withresearch, inspections, analysis, and other actions to understand the reasons for the fatigue cracking and toput in programs place to manage the potential degradation due to fatigue.

2-5

OVERVIEW OF KEY THERMAL FATIGUE ISSUESThe following provides an overview of some of the significant thermal fatigue issues that have

affected more than a single plant. These show examples of how utilities have performed evaluations tounderstand and to resolve an issue and to develop an effective approach for managing fatigue.

PWR Steam Generator Feedwater Nozzle Cracking:In many pressurized water reactor (PWR) plants, the feedwater nozzles on the steam generators are

used for both the main feedwater and for auxiliary feedwater. During hot standby operating conditions,flow variations of the auxiliary feedwater resulted in cycling stresses due to feedwater nozzlestratification, leading to cracking in the nozzle-to-safe-end weld region [4, 5]. Utilities have respondedwith improved operational techniques, modified feedwater sparger inlets and increased inspections tomanage this issue.

BWR Reactor Vessel Feedwater Nozzle Cracking:In early boiling water reactor (BWR) plants, the inlet to the vessel internal feedwater sparger was a

slip-fit configuration. Due to leakage past the sparger inlet, high cycle fatigue initiated cracks on thenozzle bore and blend radius regions [6]. A special program was initiated by the utilities to understandthe cause of the cracking and to develop a modified design and loading specification that could be usedfor plant modifications. Many utilities installed a monitoring program to detect for leakage. With theimproved designs and with more operating experience, the issue is currently managed by nozzleinspection, supported by plant-specific flaw tolerance analysis to justify the inspection interval.

BWR Control Rod Drive Return Nozzle Cracking:In early BWR plants, a reactor vessel nozzle (approx. 3-inch diameter) was provided to return control

rod drive water to the reactor vessel. Cracking in these nozzles was discovered, due to low flow cyclingand turbulent mixing with the reactor downcomer water [6]. Utilities responded by showing that the linewas not needed and capping the nozzle to avoid the thermal mixing that lead to the fatigue cracking.

B&W Plant Safety Injection/Makeup Nozzle Cracking:Through-wall leakage has occurred in the nozzle/safe-end region at two plants [7, 8]. This cracking

occurs as a result of loosening of the nozzle thermal sleeve, probably due to 1) low normal flowfluctuations allowing hot/cold temperature cycling in the nozzle region and 2) flow induced vibration.The issue is currently managed by plant operations that assure adequate flow in the makeup line and aninspection program that assures that there is no loosening of the thermal sleeves in the nozzle.

Farley Safety Injection Piping :NRC Bulletin 88-08 [9] notified PWR operators that certain lines might be susceptible to cracking as

a result of valve leakage. Inleakage of cold water from the charging system caused a through-wall leak inSafety Injection piping at the Farley plant. Related leakage had also occurred in foreign plants (includinga failure due to out-leakage from an RHR suction line). As a result of this leakage, plant-specificassessments were conducted at all plants. EPRI initiated the TASCS (Thermal Stratification, Cycling andStriping) program [10], performing testing and evaluations to better understand the sources of stresscycling that can occur in normally stagnant branch lines attached to reactor coolant piping. Many utilitiesimplemented monitoring programs, some of which are still in place to assure that significant leakage isnot occurring.

PWR Surge Line Stratification:At several plants, excess movements were observed in pressurizer surge lines, indicating that the lines

were significantly stratified. Bulletin 88-11 [11] was issued, prompting utility Owner Groups to performtesting at a number of plants to quantify the loadings and number of stratification cycles occurring. The

2-6

issue was resolved based on Owners Group analyses for each of the specific vendor types of reactors. Forthis issue, no cracking or leakage was ever observed.

Drain Line Leaks:At TMI-1 and Oconee-1, through-wall leakage has been observed in small stagnant drain lines

attached to reactor coolant piping. Testing at TMI showed that cyclic turbulent penetration into the smalluninsulated line was producing stratification locally at the elbow, although there was evidence of a pre-existing defect at the weld between the elbow and horizontal piping where the leakage occurred. AtOconee, the cracking occurred in the elbow and no pre-existing defects were identified. A similar crackoccurred in a foreign plant.

RECENT INDUSTRY PROGAMSIn addition to Owners Group activities that addressed some of the specific issues noted in the

previous section, EPRI has been involved in a number of industry- supported programs to address fatigue.In addition, ASME Section XI has been active in developing Appendix L [12] that specifically providesguidelines for evaluating potential fatigue issues.

EPRI TASCS ProgramThe Thermal Stratification, Cycling and Striping (TASCS) program was initiated by EPRI soon after

the issuance of NRC Bulletin 88-08 [9]. The objective of the program was to develop models,correlations and guidelines to evaluate TASCS phenomena in reactor coolant piping systems. Based on asurvey of existing literature, testing, and plant monitoring results that existed at that time, the TASCSprogram was developed to conduct additional testing and to develop correlations and models to predictthe thermal-hydraulic conditions in nominally stagnant lines attached to reactor coolant systems.

The testing program produced a database for correlating turbulence penetration into branch piping,which was identified as a dominant mechanism contributing to thermal cycling. Analytical efforts led tomodels that would predict stratification height and heat transfer effects in the presence of stratified flow.A handbook was developed that provides screening criteria and evaluation methodology for pipingsystems potentially affected by TASCS [10]. Key subjects covered in the included:

- Height of Stratified Flows- Heat Transfer in the Presence of Leakage Flow- Turbulence Penetration Length and Thermal Cycling- Thermal Striping- Free Convection and Stratification Heat Transfer- Test Results, Correlations and Data Analysis- Example Problems.

The final report was issued in early 1994. This handbook has been used by many utilities as a basisfor evaluating branch piping systems to determine the susceptibility to TASCS. The screening criteriahave been incorporated into risk-informed inservice inspection programs [13].

EPRI Fatigue Management ProgramAs an expansion of the TASCS program to other areas of operating nuclear plants, and to cover all

types of fatigue mechanisms that might be encountered in plants, EPRI initiated a project to produce aFatigue Management Handbook in 1992. The major developments of this program were:

- A comprehensive handbook provided to help utilities implement fatigue management programs- Screening criteria for thermal and vibrational effects- Methods for managing fatigue degradation and its impact on plant availability- Cost-effective methods to ensure and extend component fatigue life

2-7

The resulting handbook was provided in four volumes, including:- Volume 1: Fatigue concepts, regulatory issues and experience summary- Volume 2: Thermal/vibrational fatigue screening criteria with operating plant fatigue database- Volume 3: Corrective Measures for fatigue problems- Volume 4: Bibliography and copies of key fatigue references

As part of the project, the operating plant fatigue database was developed, showing that a largenumber of fatigue failures were mechanical, primarily resulting from vibration at small socket weldconnections. This finding lead to further research to investigate the fatigue aspects of socket-weldedfittings[14].

ASME Section XI ActivitiesIn the early 90’s, the ASME Section XI Working Group on Operating Plant Criteria began efforts to

develop criteria that could be used in operating plants to justify continued operation when a fatigueconcern was identified. At the time, utilities were struggling with how to handle issues such as 1)predicted fatigue usage factor greater than 1.0, 2) number of cycles greater than considered in design, or3) transients encountered with severity greater than considered in design. The outcome of this effort wastwo reports published by EPRI [15, 16] and the recently added Appendix L to ASME Section XI.

MRP THERMAL FATIGUE PROGRAMThe Materials Reliability Project (MRP) Thermal Fatigue Industry Issue Group (TF-ITG), is currently

developing Thermal Fatigue Management Guidelines (TFMG) that can be used by utilities to addresspotential thermal fatigue in piping systems that are non-isolable from the reactor coolant system piping inPWR’s. The MRP thermal fatigue project is composed of 13 separate tasks. Figure shows the inter-relationship between the tasks and how the outcome of each tasks relates to assessing the potential forthermal fatigue or performing additional inspections, monitoring, maintenance, or plant modifications.The overall schedule for the project is shown in Figure 2.

Tasks 1 and 2 – Project ManagementThis is the overall effort by EPRI to manage the project. Under EPRI leadership, the project is to be

completed on time and within the budget established by the member utilities. This task also includesmeetings with industry and Nuclear Regulatory Commission to assure that the program outputs will beacceptable for use in performing plant evaluations at project completion.

Task 3 – Industry Operating ExperienceExisting industry failure and leakage databases are assembled primarily based on leaks or identified

flaws. Collection of industry experience and sharing of monitoring data and information of events that donot meet some higher reporting threshold will provide added value and new insights into where thermalfatigue may be a problem and where it is definitely not a problem. A database is being developed that willcollect utility observations regarding plant monitoring or events that do not reach a level requiringreporting. This database will be available on the EPRI web site. It will contain historical experience andwill include provisions to capture and make available similar information in the future.

Task 4 – Thermal Fatigue ScreeningThis task is expanding on the EPRI TASCS models and results from other industry programs to

develop improved fatigue screening methodology. The screening approach will identify the thermalfatigue phenomena that might be possible, factors necessary for thermal fatigue to occur, and the logicprocess for making this determination. When screening identifies a location as susceptible to thermalfatigue, a conservative prediction of the thermal loading will be made, determining an allowable operatingtime prior to the cumulative usage factor exceeding 1.0. The methodology will have a technicallydefensible basis and will be validated against known instances of thermal fatigue failures.

2-8

Task 5 – Thermal Fatigue Monitoring GuidelinesThis task will provide guidance to utilities in implementing effective monitoring when analytical

means and screening show that thermal fatigue may be an issue. All types of montoring will bediscussed. State-of-the-art monitoring technology will be identified. Criteria for effective placement ofmonitoring sensors will be developed. Methods for interpretation of monitoring data will be provided.Most importantly, criteira will be provided for discontinuance of monitoring programs. The outcome willbe practical guidance on the use of monitoring to detect potential thermal fatigue phenomena.

Task 6 – NDE Inspection GuidelinesThis task is assembling previous guidance on nondestructive examination (NDE) methodologies such

as (RT) or ultra sonics (UT) and evaluating potential new technologies. Recommendations will bedeveloped for specific NDE technology and variables, especially for small diameter piping.Recommendations will be made on the appropriate qualification of NDE examiners and procedures. Therecommended means of evaluating NDE data and reporting levels are also being provided. Contacts weremade internationally to determine any difficulties in applying NDE for thermal fatigue and to identifylaboratory investigations to verify NDE performance. Research has been completed on determiningcapabilities for producing thermal fatigue cracks. Based on this research, mockups were designed andfabricated. A set of mockups with individual cracks due to thermal fatigue, along with mockupscontaining thermal craze cracking, were fabricated and used in qualification of techniques for detectingthermal fatigue cracking. The outcome of this effort will be guidance on NDE methodologies andrecommendations of specific NDE technology and variables to use when inspecting for suspected thermalfatigue damage. As an adjunct to this task, an interactive training module is being developed to traininspectors on the characteristics of thermal fatigue and how to detect it using the recommendedmethodologies.

Task 7 – Plant Operations/Maintenance GuidelinesThis task focuses on how operations and maintenance (O&M) practices can lead to thermal fatigue

damage and how potential changes can eliminate or minimize thermal fatigue damage potential. PWRoperating experience will be used to identify plant practices and corrective actions implemented ataffected plants. Guidance will be developed for use by utility engineers to aid in identifying O&Mpractices which may lead to fatigue damage and what actions may be taken to mitigate the consequences.A key area for the research will be valve maintenance. This will result in guidance for plant personnel toidentify how O&M practices can create or minimize the potential for thermal fatigue cracking.

Task 8 – Thermal Fatigue EvaluationThis task will develop analytical approaches that may be used to evaluate components where

screening can not be conservatively used to demonstrate an acceptable fatigue life. The evaluationmethodology will include both a simplified disposition evaluation and a more rigorous analysis guide toaid engineers in evaluating thermal fatigue situations. The analysis guide will document the techniquesand describe comprehensive methodologies for analytical reconciliation of thermal fatigue phenomenausing or based on ASME Section III methodology.

Task 9 – Plant Modification GuidelinesThis task will identify and describe plant modifications that might be implemented for avoiding

thermal fatigue. Implementation of cost effective plant changes could potentially eliminate the need forfuture monitoring and augmented piping inspections where there is significant probability of futurecracking.

2-9

Task 10 – International Technical ExchangeThis task will identify and pursue participation in important foreign R&D activities that could

contribute to resolution of thermal fatigue issues. This activity is intended to provide awareness of andaccess to foreign information of value to US utilities for detection, assessment, mitigation, and preventionof thermal fatigue damage. Similarly, this activity can provide information to non-domestic utilities toresolve fatigue issues. The international workshop on thermal fatigue at which this paper is presented isone of the outcomes of this task.

Task 11 – Thermal Fatigue Management GuidelinesThis task represents the principal product of the thermal thermal fatigue project. It will assemble the

results of the other tasks, document conclusions drawn from that work, and provide recommendations formanaging thermal fatigue. An interim version of the guideline will be provided prior to development ofthe final products from Tasks 4 and 8. The Thermal Fatigue Management Guideline will be a compilationof methods for assessment, screening, monitoring, analysis, and remediation and/or management ofthermal fatigue.

Task 12: TrainingThis task will develop training applying the Thermal Fatigue Management Guideline. This training

will be aimed at utility personnel (operations, maintenance, systems and design engineering) to increasetheir overall knowledge and awareness of cyclic thermal fatigue and how plant operations andmaintenance may be contributors to the phenomena. This will result in more knowledgeable andexperienced personnel and more effective management of thermal fatigue issues.

Task 13: ASME Activities MonitoringThis task provides for monitoring ASME Section III and XI activities related to thermal fatigue and

inspection. Inspection guidance developed by the project will be reviewed with appropriate Code groupsto determine if such guidance should form the basis for future Code revisions.

Other ActivitiesThe MRP TF-ITG has just undertaken the responsibility to provide further review of EPRI efforts to

conduct fatigue related research in other areas. The specific scope of these activities is underdevelopment.

LICENSE RENEWAL ACTIVITIESEPRI has conducted several projects in the past to support plants preparing for license renewal. In

earlier projects, studies were performed to evaluate the effects of light water reactor environments onreactor coolant components in PWR’s and BWR’s based on design and plant monitoring data [17-22]. Aproject has just been initiated to develop a consensus criteria for preparation of license renewalapplications. This will develop a common approach for addressing fatigue in license renewal submittals,eliminating some of the uncertainties and providing costs savings for utilities.

CONCLUSIONSFatigue is not a new issue for US utilities with operating nuclear plants. Even though nuclear plants

were designed with consideration of the potential effects of fatigue, operating experience periodicallyidentifies an issue that was not considered in the design phase. As a result, various vendor and industryresearch programs and ASME Code activities have been undertaken in recent years to understand andresolve fatigue issues.

With the advent of deregulation of the utility industry, utility personnel must assure that fatigue-related leakage does not occur, both from the standpoint of plant safety and economics. The current MPRthermal fatigue project is one aimed at assuring that leakage due to thermal fatigue in non-isolable pipingsystems does not occur. Planning is underway for additional programs to support goals of minimizing

2-10

fatigue failures in the future, and showing that fatigue can be adequately managed during an extendedoperating period.

REFERENCES1. ASME Boiler and Pressure Vessel Code Section III, “Nuclear Vessels,” 1963.2. USA Standard Code for Pressure Piping, “Nuclear Power Piping,” USAS B31.7-1969.3. “Draft ASME Code for Inservice Inspection of Nuclear Reactor Coolant Systems,” ASME, October

1969 (Predesessor of ASME Section XI).4. U.S. Nuclear Regulatory Commission, “Cracking in Feedwater System Piping,” IE Bulletin 79-13,

Revision 2, October 16, 1979.5. U.S. Nuclear Regulatory Commission, “Investigation and Evaluation of Cracking Incidents in

Pressurized Water Reactors,” NUREG-0691, September, 1980.6. U.S. Nuclear Regulatory Commission, “BWR Feedwater Nozzle and Control Rod Drive Return Line

Nozzle Cracking,” NUREG-0619, April, 1980.7. U.S. Nuclear Regulatory Commission, “Cracking in Piping of Makeup Coolant Lines at B&W

Plants,” IE Information Notice No. 82-09, March 31, 1982.8. U.S. Nuclear Regulatory Commission, “Unisolable Crack in High Pressure Injection Piping,”

Information Notice 97-46, July 9, 1997.9. U.S. Nuclear Regulatory Commission, “Thermal Stresses in Piping Connected to Reactor Coolant

Systems,” NRC Bulletin No. 88-08, June 22, 1988, Supplement 1, June 24, 1988, and Supplement 2,August 4, 1988.

10. TR-103581, “Thermal Stratification, Cycling and Shaping (TASCS); EPRI, March 1994.11. U.S. Nuclear Regulatory Commission, “Pressurizer Surge Line Thermal Stratification,” NRC Bulletin

No. 88-11, December 20, 1988.12. Appendix L to ASME Boiler and Pressure Vessel Code Section XI, “Operating Plant Faituge

Assessment,” 1995 Edition, 1996 Addenda.13. TR-112657, “Risk-Informed Inservice Inspection Evaluation Procedure,” Rev. B-A, EPRI, December

1999.14. TR-107455, “Vibration Fatigue of Small Bore Socket-Welded Pipe Joints,” EPRI, June 1997.15. TR-100252, “Metal Fatigue in Operating Nuclear Power Plants,” EPRI, April 1992.16. TR-104691, “Operating Nuclear Power Plant Fatigue Assessments,” EPRI, April 1995.17. TR-107515, “Evaluation of Thermal Fatigue Effects on Systems on Systems Requiring Aging

Management Review for License Renewal for the Calvert Cliffs Nuclear Power Plant,”, January1998.

18. TR-105759, “An Environmental Factor Approach to Account for Reactor Water Effects in LightWater Reactor Pressure Vessel and Piping Fatigue Evaluations,” December 1995.

19. TR-110043, “Evaluation of Environmental Fatigue Effects for a Westinghouse Nuclear Power Plant,”April 1998.

20. TR-110356, “Evaluation of Environmental Thermal Fatigue Effects on Selected Components in aBoiling Water Reactor Plant,” April 1998.

21. TR-110356, “Environmental Fatigue Evaluations of Representative BWR Components,” May 1998.22. TR-110356, “Effect of Environment on Fatigue Usage for Piping and Nozzles at Oconee Units 1,2,

and 3,” December 1999.

2-11

ScreeningTool

4

Screencomponents

Knowledge &Experience

12

Maintenance/Ops

Guide

7

Tryto

EliminateTF Initiator

MonitoringGuidelines

5

MonitoringPredict

TF Loading(temp, rates, flows, cycles)Predictive

Tool

4

DispositionTool

8

AnalysisMethods

Guide

8

TrySimplified Tool

Vulnerable

Success

Success

Full Plant Life

Not Successful

Only Partial Plant Life Predicted

Permanent or Recurring or Repair/Replace Monitoring Inspections Mod (Flaw Tolerance)

Not Successful

Not

Vulnerable

OKContinue

Operations

Confirmed (1)

(CUF<1.0)

Events DetectedFailures / Flaws

OccurResearch

Valve Testing3, 10

InspectionGuidelines

6

Repair/Replace

Guidelines9

RigorousAnalysis

License ActionD

eliv

ered

ITG

Tas

ks

(1) Permanent/semi-permanent monitoringmay be recommended to confirm predicted

significance

Figure 1. Project Flowchart

2-12

Thermal Fatigue ITGLevel 1 Schedule

1999 2000 2001 ID Task Name Start Finish A S O N D J F M A M J J A S O N D J F M A M J J A S O N

1 & 2 Project Management 7/28/99 11/1/01 3 Industry Operating Experience 1/3/00 11/30/00 4 Thermal Fatigue Screening 10/1/99 8/31/01 5 TF Monitoring Guidelines 1/3/00 10/31/01 6 TF Inspection Guidelines 9/1/99 9/29/00 7 Plant O&M Guidelines 1/3/00 10/31/00 8 Thermal Fatigue Evaluation TBD TBD 9 Plant Modification Guidelines 1/3/00 10/31/00

10 International Technology Exchange 10/1/99 10/1/01

11a Interim Fatigue Management Guidelines 6/1/00 10/31/01

11b Thermal Fatigue Management Guidelines 6/1/00 8/31/00

12 Develop & Delivery of Training 9/1/01 11/1/01

13 ASME Section XI WG Activities 8/2/99 8/31/01

Figure 2. Project Schedule

US UTILITY PROGRAMS FORFATIGUE MANAGEMENT

Authors:Arthur F. Deardorff*

Structural Integrity Associates

Stan T. Rosinski and John J. Carey EPRI


International Conference on Fatigue of Reactor ComponentsNapa, California USA

July 31 – August 2, 2000

*Presenting Author [email protected]

2-13

OUTLINE

• Welcome

• Background and Historical Perspective

• Recent Industry Activities

• Materials Reliability Project (MRP) FatigueActivity

2-14

BACKGROUND

• Nuclear Plant Reactor Coolant Systems Designedfor Cyclic Operation

♦ Vessels – ASME Section III♦ Piping – USAS B31.7/ASME Section III

• Key Locations Inspected Per ASME Section XI

• However, Fatigue-related Failures Occur♦ Loadings not known at time of initial design♦ Thermal and vibration

• Industry Has Responded to Manage the Issues

2-15

REVIEW OF KEY THERMALFATIGUE ISSUES

PWR Steam Generator Feedwater Nozzle Cracking

• ≈≈≈≈ 1980• Cracking at Nozzle/Safe-end Region• Caused by Flow Cycling/Stratification

♦ Flow cycling of auxiliary feedwater♦ Counterbore geometry caused high local stresses

• Managed by:♦ Improved operations♦ More frequent inspection♦ Modified designs

2-16


BWR Feedwater Nozzle Cracking

• ≈≈≈≈ 1980• Cracking In Nozzle Bore/Blend Radius Area• Caused by High Cycle Fatigue

♦ Bypass of cold leakage flow past sparger inlet

• Managed by:♦ Modified designs♦ Leakage monitoring♦ Inspection interval based on flaw tolerance evaluation

2-17


B&W Safety Injection/Makeup Lines

• 1982 / 1997• Cracking at Safe-End / Piping Weld• Caused by Loosening of Thermal Sleeve

♦ Flow induced vibration♦ Flow cycling allowed hot RCS water into region

• Managed by:♦ Inspection for thermal sleeve tightness♦ Maintaining adequate flow in nozzle

2-18


Safety Injection Piping

• 1988 + Others in Non-US Plants• NRC Bulletin 88-08 / Cracking in Elbow/Horizontal

Pipe Region• Caused by Cold Water Inleakage/Turbulence

Penetration• Managed by:

♦ Monitoring♦ Valve maintenance♦ EPRI TASCS Program

2-19

OTHERS

• BWR Control Rod Drive Nozzle Cracking♦ Flow to nozzle eliminated

• PWR Surge Lines♦ Re-evaluated per NRC Bulletin 88-11

• Drain Lines♦ Current MRP issue

2-20

EPRI PROGRAMS

• Thermal Stratification, Cycling and Striping(TASCS) Program

♦ Initiated to address Bulletin 88-08 issues

♦ Performed testing/analysis to address stratification andturbulence penetration

♦ Final report in 1994

• EPRI Fatigue Management Handbook♦ Followed TASCS Program Looking at Remainder of

Plant

♦ Also addressed vibrational fatigue

♦ Led to further programs° socket weld fitting fatigue testing

° improved stress indices/stress intensification factors

2-21

ASME SECTION XI ACTIVITIES

• Started Efforts about 1990 to Develop CodeApproaches for Addressing Fatigue Concerns

♦ Recognized that stress reports produced to showCUF<1.0

♦ Stressed component re-analysis° reduce conservatisms° use actual transients

♦ Allowed fatigue monitoring♦ Included flaw tolerance assessment as an alternate to

usage factor analysis

• Appendix L Approved in 1996• Efforts Continuing to Gain Regulatory

Acceptance

2-22

MATERIALS RELIABILITY PROJECTTHERMAL FATIGUE PROGRAM

• MRP TF-ITG: Thermal Fatigue Issue Task Group

• Approved by Utilities in Mid-1999

• EPRI Managed Program♦ Providing guidelines for managing thermal fatigue

♦ Addressing non-isolable PWR piping

• Mike Robinson is Task Group Chairman

2-23

MRP-TF PROJECT GOALAND SCOPE

• Provide EPRI member utilities with a consistent set ofguidelines and methodology for addressing piping thermalfatigue issues in 2001

• Scope is thermal fatigue issues for those portions of ASMEclass 1 piping systems that are:

♦ Connected to the reactor coolant pressure boundary

♦ Greater than 1” diameter

♦ Not isolable from the reactor coolant pressure boundary

• Similar to NRC Bulletin 88-08 scope

2-24

STRATEGIC APPROACH

• Respond to NRC request for industry leadership and action

• Provide the industry ability to manage thermal fatigueconcerns through current license life and any renewalperiods

♦ validate methodology against known failures

• Take advantage of earlier thermal fatigue work by EPRI andsupplement it with new ideas and enhanced thermal fatiguemanagement capabilities

2-25

MRP-TF TASK 3:INDUSTRY OPERATING EXPERIENCE

• Description: Existing industry databases are currentlybased on leaks or identified flaws

♦ Data from plant monitoring or other experience also beingcollected

• Outcome: A database that collects utility experience andobservations which do not reach the level of recordingthrough LERs or other means

♦ Data to be available on EPRI web site

♦ Process to be established for continued update

2-26

MRP-TF TASK 4:THERMAL FATIGUE SCREENING

• Description: This task will review existing programresults for needed enhancements for improved screeningand evaluation capability

♦ Thermal fatigue phenomena

♦ Factors for thermal fatigue to occur

♦ Logic process for screening

♦ Load magnitudes/cycles to be conservatively estimated

• Outcome: Technically defensible methods fordetermining when and where significant thermal fatiguedamage may occur in PWR piping systems

2-27

MRP-TF TASK 5:THERMAL FATIGUE MONITORING

GUIDELINES

• Description: This task will provide guidance formonitoring

♦ Basis for implementing a monitoring program

♦ Identification of state-of-the-art monitoring technology

♦ Effective placement of monitoring sensors

♦ Interpretation of monitoring data

♦ Basis for discontinuing monitoring

• Outcome: Practical guidance on the use of monitoring todetect potential thermal fatigue phenomena

2-28

MRP-TF TASK 6:NDE INSPECTION GUIDELINES

• Description: This task will assemble previous guidanceon NDE methodologies (such as RT or UT) and makerecommendations for specific NDE technology andvariables

♦ Mockups constructed representing thermal fatigue in smallpiping

♦ Various advanced NDE approaches evaluated

♦ Special considerations for detecting thermal fatigue crackingand crazing investigated

• Outcome: Guidelines on NDE methodologies andrecommendation of specific NDE technology and variablesto use when inspecting locations for suspected thermalfatigue damage

2-29

MRP-TF TASK 7:PLANT O&M GUIDELINES

• Description: This task focuses on how O&M practicescan lead to thermal fatigue damage and identifiesnecessary changes to eliminate the damage potential

♦ Valve maintenance

♦ Operations generally effects systems normally used, such ascharging and auxiliary spray

• Outcome: Guidance for use by plant personnel to identifyhow O&M practices can create or minimize the potential forcausing thermal fatigue events

2-30

MRP-TF TASK 8:THERMAL FATIGUE EVALUATION

• Description: This task will develop both a simplifiedevaluation methodology and a more rigorous analysisguide to aid engineers in evaluating situations wherethermal fatigue may potentially be occurring

• Outcome: An analysis guide, documenting thetechniques and describing comprehensive methodologiesfor analytical reconciliation of thermal fatigue phenomenausing or based on ASME Section III methodology, will bedeveloped for evaluating location subject to cyclic thermalloadings

2-31

MRP-TF TASK 9:PLANT MODIFICATION GUIDELINES

• Description: This task will identify and describe plantmodifications that may be used for avoiding the potentialfor thermal fatigue

• Outcome: Guidelines for identification of cost effectiveplant changes that would eliminate need for futuremonitoring and piping augmented inspections

2-32

MRP-TF TASK 10:INTERNATIONAL TECHNICAL

EXCHANGE

• Description: This task will support identification of andpossible participation in important foreign R&D activitieswhich could contribute to resolution of the thermal fatigueissue

• Outcome: Awareness of and access to foreigninformation that could be of value for detection,assessment, mitigation, and prevention of thermal fatiguedamage (This workshop is an outcome of this task)

2-33

MRP-TF TASK 11:THERMAL FATIGUE MANAGEMENT

GUIDELINES

• Description: This task represents the principal product ofthis project, assembling the results of the other tasks,documenting conclusions drawn from that work, andproviding recommendations for managing thermal fatigue

• Outcome: The “TFMG” will be a compilation of methodsfor assessment, screening, monitoring, analysis, andremediation for and management of thermal fatigue

♦ The ITG will seek NRC staff acceptance of guideline.

♦ An interim TFMG will be issued later this year.

♦ Final TFMG available late 2001.

2-34

MRP-TF TASK 12:DEVELOP & DELIVER TRAINING

• Description: This task develops and delivers the trainingfor utility engineers and others in applying the “TFMG”andwill be provided so that utility personnel (operations,maintenance, engineering) can increase their overallknowledge and awareness of cyclic thermal fatigue andhow plant operations and maintenance may be contributorsto the phenomena

• Outcome: More knowledgeable and experiencedpersonnel and more effective management of potentialthermal fatigue issues

2-35

PROJECT FLOWCHART

ScreeningTool

4

Screencomponents

Knowledge &Experience

12

Maintenance/Ops

Guide

7

Tryto

EliminateTF Initiator

MonitoringGuidelines

5

MonitoringPredict

TF Loading(temp, rates, flows, cycles)Predictive

Tool

4

DispositionTool

8

AnalysisMethods

Guide

8

TrySimplified Tool

Vulnerable

Success

Success

Full Plant Life

Not Successful

Only Partial Plant Life Predicted

Permanent or Recurring or Repair/Replace Monitoring Inspections Mod (Flaw Tolerance)

Not Successful

Not

Vulnerable

OKContinue

Operations

Confirmed (1)

(CUF<1.0)

Events DetectedFailures / Flaws

OccurResearch

Valve Testing3, 10

InspectionGuidelines

6

Repair/Replace

Guidelines9

RigorousAnalysis

License Action

Del

iver

ed IT

G T

asks

(1) Permanent/semi-permanent monitoringmay be recommended to confirm predicted

significance

2-36

PROJECT SCHEDULE

ID Task Name Start Finish A S O N D J F M A M J J A S O N D J F M A M J J A S O N

1 & 2 Project Management 7/28/99 11/1/01

3 Industry Operating Experience 1/3/00 11/30/00

4 Thermal Fatigue Screening 10/1/99 8/31/01

5 TF Monitoring Guidelines 1/3/00 10/31/01

6 TF Inspection Guidelines 9/1/99 9/29/00

7 Plant O&M Guidelines 1/3/00 10/31/00

8 Thermal Fatigue Evaluation TBD TBD

9 Plant Modif ication Guidelines 1/3/00 10/31/00

10 International Technology Exchange 10/1/99 10/1/01

11a Interim Fatigue Management Guidelines 6/1/00 10/31/01

11b Thermal Fatigue Management Guidelines 6/1/00 8/31/00

12 Develop & Delivery of Training 9/1/01 11/1/01

13 ASME Section XI WG Activities 8/2/99 8/31/01

2000 20011999

2-37

MRP LICENSE RENEWAL SUPPORT

• Recent Addition to Scope

• Established to Prepare Criteria for AddressingEnvironmental Effects of Fatigue Beyond 40Years

• Scope and Approach Under Development

2-38

SUMMARY

• Fatigue in Operating Plants is Not a New Issue

♦ Utilities have generally been effectively dealing withfatigue as issues surface

♦ Industry groups respond to perform research/evaluationas required

• Even though fatigue not identified as safetyissue, utilities need to take appropriate actions toavoid forced outages due to fatigue failure

2-39

3-1

3 ACTIVITIES OF THE OECD-NEA IN THE AREA OFTHERMAL FATIGUE IN LWR PIPING

E. MathetOECD Nuclear Energy Agency

Le Seine St Germain12 boulevard des Iles

F-92130 Issy-les-Moulineaux

3-3

Activities of the OECD-NEA in the areaof Thermal Fatigue in LWR Piping

E. MathetOECD Nuclear Energy Agency

Le Seine St Germain12 boulevard des Iles

F-92130 Issy-les-Moulineaux

Abstract

The OECD Nuclear Energy Agency has 27 Member countries. Under the auspices of theNuclear Safety Division are two senior committees dealing with regulatory aspects(CNRA) and technological aspects (CSNI). Under these two committees activitiesrelevant to thermal fatigue in LWR piping are carried out under the two WorkingGroups (WG's): WG on Operating Experience, WG on Integrity of Components andStructures. There is also co-operation with the former task group on thermalhydraulicsapplication. These groups make recommendations to the senior committees.

WG on Operating Experience is mainly concerned with the analysis of safety significantincidents, but it undertakes special studies as well. WG on Integrity of Components andStructures (Integrity and Ageing, IAGE) has activities in fracture mechanics, Non-Destructive Examination and material degradation, as being the three aspects ofstructural integrity for metal reactor components. It also has sub-groups dealing withthe ageing of concrete structures and the seismic behaviour of structures. In the area ofthermal fatigue, it has already organised a Specialist Meeting in Paris in 1998, and itcurrently has a programme of work approved by CSNI.

This paper provides background on the Agency itself, while focusing on the thermalfatigue activities within NEA carried out mainly through the Committee on the Safetyof Nuclear Installations and the groups of technical experts on operating experience,structural integrity, and thermal hydraulic.

3-4

INTRODUCTION

The Nuclear Energy Agency (NEA) is one of the 15 bodies that make up theOrganisation for Economic Co-operation and Development (OECD), located in theParis area in France. The Members of the OECD/NEA are a group of 27 like -minded,developed countries which at the end of 1999 operated 348 reactor units. Today, nuclearenergy accounts for 25 per cent of all electricity produced in OECD countries (17 per centworld-wide).

The mission of the OECD/NEA (Ref 1) is to assist its Member countries in maintainingand further developing, through international co-operation, the scientific, technologicaland legal bases required for a safe, environmentally friendly and economical use ofnuclear energy for peaceful purposes, as well as to provide authoritative assessmentsand to forge common understandings on key issues, as input to government decisionson nuclear energy policy and to broader OECD policy analyses in areas such as energyand sustainable development.

In carrying out this mission, the NEA facilitates exchange 'and comparison' of nationalregulatory practices, reviews selected technical and economic aspects of nuclear powerdevelopment and sponsors joint international research projects. In addition, the NEAthrough its studies and the development of technical standards and codes of practiceprovides a source of objective information and advice to its Members.

The broad direction of the programme of work of the OECD/NEA is set by the SteeringCommittee for Nuclear Energy. This Committee is assisted by a number of specialisedstanding committees such as the Committee on the Safety of Nuclear Installations(CSNI) and the Committee on Nuclear Regulatory Activities (CNRA).

In a typical year, these two committees with the assistance of the NEA secretariatorganise 12 to 15 workshops focused on well defined topics of current high interest tonuclear safety; produce about 35 to 40 reports - several of which are state-of-the artreports - and organise and co-ordinate a number of International Standard Problem(ISP) exercises with the aim of increasing confidence in the prediction of the computercodes used to assess the safety of nuclear power plants.

In general, the emphasis of the OECD/NEA safety programme is related to safetyresearch. The CSNI, (ref 2), is made up of senior scientists and engineers, with broadresponsibilities for safety technology and research programmes. The technical fields ofnuclear reactor safety interest into which the CSNI has designated specific WorkingGroups (items 1 to 4) and Special Expert Groups (items 5 and 6) are:

1. Risk Assessment,

2. Analysis and Management of Accidents,

3-5

3. Integrity of Components and Structures,

4. Operating Experiences,

5. Human and Organisational Factors

6. Fuel Safety Margins

along with expert groups formed from time to time.

This paper will focus specifically on those activities of Working Group on Integrity andAgeing (IAGE) relevant to thermal fatigue in LWR piping. IAGE WG deals with theintegrity of structures and components, and has three sub-groups, dealing with theintegrity of metal structures and components, ageing of concrete structures, and theseismic behaviour of structures. It should be noted by most activities are done in co-operation with other Working Groups, in particular the CSNI WG on Operatingexperiences which guides the operation of the Incident Reporting System (IRS) and alsoevaluates the events and identifies significant safety issues to be brought to theattention of CSNI. There is also co-operation with the former task group onthermalhydraulic applications.

The IAGE working group on metal components has dealt with issues related more orless closely to thermal fatigue in piping by sponsoring: round robins or benchmarks onfatigue crack growth in plates and pipes under mechanical loading, on crack openingbehaviour and leak rates in pipes, on pipe leak and break probabilities, and NDE in arange of components (the former PISC project, jointly with the CEC); Specialistsmeetings or workshops on Leak Before Break, reactor coolant system leakage andfailure probabilities and experience with thermal fatigue in LWR piping caused bymixing and stratification; and a report on monitoring.

NEA works in co-ordination with other international exercises such as the CEC, IAEAand WANO to avoid duplication and to ensure the appropriate participation. This co-ordination takes the form of liaison between the secretariats, joint sponsorship ofmeetings or projects, and exchange of future programmes of work.

THERMAL FATIGUE IN LWR PIPING

Thermal fatigue due to stratification and mixing of hot and cold water is a recurringphenomenon, but with a relatively low frequency. It can affect safety related piping ofmany different systems (ECCS, RHR, AFW, pressuriser surge line and safety / reliefvalve discharge lines) at various locations. It may happen that these phenomena arenot included in the design conditions. Recently, similar events have occurred insections of piping in the primary circuit and therefore the plant owners and the safety

3-6

authorities are now alerted over some conditions not being consistent with licensingbasis and inspection commitments. Of particular concern are those problems arisingfrom unanticipated thermal fatigue in unisolable piping connected to the reactorcoolant system. None of the mentioned incidents led to radioactivity propagation to theenvironment, but the safety significance of these events results from the leakage ofprimary coolant through the second barrier but inside the reactor containment.

Thermal fatigue problems can be seen as challenging issues for plant owners. Becauseof the high potential impact on safety, cost and radiation exposure, these issues have tobe addressed more effectively. This is possible only with a very close co-operationbetween researcher, designer and plant owner and, among plant personnel, betweenmaintenance and operation staff. Having all these people working together is the keypoint to keep the risk of thermal fatigue under control and, more generally, to ensuresafe and effective plant operation. There is a need to encourage co-operation betweenthe different disciplines to solve the problem.

PAST ACTIVITIES OF THE GROUP

The IAGE working group on the integrity of metal structures and components has anon-going activities in the area of thermal fatigue in LWR piping.

In 1997 this IAGE WG decided a Specialist Meeting, (ref 3), on experience with thermalfatigue caused by mixing and stratification of hot and cold water would be timely.There have been many incidents (cracks) caused by stratification. The IRS had about 25reports up to 1996, and during the year before there were several events with the samecause. Thermal fatigue cracks in BWR internals have been of concern for some time.Other events such as Dampierre-1 had raised questions about earlier assumptions.

The Specialist Meeting was hosted by IPSN, France under the sponsorship of WANOand CSNI WG on Operating experiences and CSNI WG on Integrity and Ageing jointly.As the topic arises out of operating experience, it was essential to involve the utilitiesfully through the participation of WANO.

There had been sessions on operating experience, thermal hydraulic phenomena,material and structural response, monitoring, inspection, mitigation and preventionand the safety implication. There had been about 70 participants.

Outputs of the Specialist Meeting .

Outputs of the Specialist Meeting were written down into conclusions andrecommendations that were agreed by participants and approved by CSNI. Althoughsome progress in the knowledge have been made since this meeting they are still validand served as a basis to draw up the IAGE WG action plan. They are reproducedbelow.

3-7

"- There is a need to develop further accepted methods to identify locations withpotential risk of thermal fatigue. There are proposals for simplified screening criteriabased on semi-empirical models to determine the areas where there is risk of thermalfatigue, but these are not generally accepted. As there are many uncertainties, it ispossible to consider a probabilistic treatment, although the data for this are alsolimited. As first step in this direction, the use of best estimate analysis methods shouldbe considered, so that attention is focused on the areas most at risk for thermal fatigue,although these may be difficult to define. Then, at those locations, monitoring oftemperature and of pressure if necessary should be implemented. In addition, periodicverification of the leaktightness of the nearest valves could further reduce the risk ofthermal fatigue.

- Monitoring of temperature fluctuations can be seen as an important part of thedefence and, at present, it remains the most reliable method to avoid unanticipatedincidents. At present, there are different strategies in use and no single methodprovides defence. There is a need for combining redesign and revised operatingpractices. A small internal cold water leakage into a hot section of pipe can lead to aquick propagation of cracks by thermal fatigue in some sensitive zones. Manufacturingprocess probably has a very important impact on the rate of crack initiation. It is notpossible to draw up simple criteria for such parameters as allowable valve leak rates orlimiting pipe diameters, as there is a great variety in the systems and operatingprocedures.

- Concerning the experience from non-destructive examinations, ultrasonic testing gavenumerous false calls, and for certain geometries and material conditions, performanceof present NDE methods to detect fatigue cracks is limited. If a greater reliance is to beplaced on NDE, the development of qualified methods is needed."

Follow up actions approved by CSNI.

Discussions at the Specialist Meeting held in Paris in June 1998, and within the IAGEworking group have identified a need to develop screening criteria to guide monitoringand inspection. That is, there is a need to develop reliable methods to identifypositions in the reactor coolant and associated systems, which are sensitive to thermalstratification and thermal mixing.

It was proposed to work towards this objective, by preparing a report or technicalposition paper on present practices. Topics to be discussed are:

� Design for conservative estimates of thermal loads

� Design and operation practices that reduce thermal stratification/mixing problems

3-8

� Actions taken to mitigate thermal stratification/mixing problems

� Thermal stratification/mixing in conventional power plants

� Thermal stratification/mixing problems with Russian designed reactors

Other topics that could be investigated subsequently include:

� Basic causes of fatigue cracking by thermal stratification and thermal mixing

� Effective inspection methods

� Development of improved and reliable methods for the detection of fatigue cracksespecially in critical components like elbows

� Advanced methods for the detection of the early phases of fatigue cracking

� Consideration of benchmarks for comparing present design methods

Other action undertook by the CSNI was the co-sponsorship along with the US NuclearRegulatory Commission of the "International Conference on Fatigue" host by EPRI inNapa, California on July 31 - August 2, 2000.

CURRENT AND FUTURE ACTIONS.

At its last meeting in June 2000 in Paris, the IAGE Working Group heard presentationsfrom France, Germany and Japan showing this topic kept on being of high interest forboth the industry and regulatory bodies. The group discussed thoroughly thesignificance of (1) screening criteria to determine critical locations, (2) management ofinterfaces between system engineers, thermal hydraulic and mechanical engineers, (3)code changes and (4) experimental information available after all those years of work.

He then draw up a programme of work that would address the subject with aninternational perspective as it is the purpose of the OECD-CSNI.

One aspect of the program is to get a broad picture of technical issues. Outputs of the"International Conference on Fatigue" and of the Specialist Meeting in Paris in 1998would serve to update understanding of the issues by sharing views and problems.Some topics listed above in the follow-up actions paragraph would be then addressed.

As to the other aspect, it will focus on actions that would complement and foster co-operation between countries. To start with, it was decided to conduct an internationalsurvey on:

� screening criteria,

� experiments available that address high cycle thermal fatigue,

3-9

� the tendency for code changes, if any.

Future activities considered by the group after the above mentioned actions beingcompleted would be to conduct international benchmarks on technical issues (e.g.thermal hydraulic, crack initiation and propagation, NDE techniques and capabilities).

CONCLUSIONS

The contents of this paper provide an overview of both the framework and structure ofthe OECD Nuclear Energy Agency and in particular the activities being performed inthe area of thermal fatigue in LWR piping.

On issues of thermal fatigue, the main NEA safety technology committee, theCommittee on the Safety Nuclear Installations, organises relevant activities based onthe aspects of operating experience, structural integrity and thermalhydraulics. In allthese areas, groups of experts review at regular intervals the state of the art andorganise studies, workshops, preparation of reports, etc., aimed at discussing issues ofmutual concern and at arriving at common technical positions. The use of specialistgroups and specialist meetings to examine in depth specific questions of nuclear safetyand the use of international standard problems to provide a vehicle for discussion andcomparison are the important characteristics of the OECD/NEA programme of work.

Although the OECD/NEA obligations are primarily towards its Member countries,most of the information generated by its activities is made available to all countries.Furthermore, in line with the general policy of the OECD, there is a trend towardsincreasing participation by countries which are not currently Members of the OECD.

In the area covered by this paper, the OECD-CSNI decided to promote a program ofwork by organising or sponsoring conferences and workshops and also by promotingexchange of information on critical topics by conducting a survey in 2000. Follow-upactions would be scheduled after completion of those tasks.

References

1 The Strategic Plan of the Nuclear Energy Agency , OECD/NEA/NE(99)1, 1999

2 The Strategic Plan for the Committee on the Safety of Nuclear Installations . OECD/NEA/CSNI/R(2000)3,January 2000.

3 Proceedings of the Specialists Meeting on Experience with thermal fatigue in LWR piping caused by mixingand stratification, Paris 1998. OECD/NEA/CSNI/R(98)8, December 1998.

International Conference onFatigue, NAPA July31 - Aug 2,2000 July 31, 2000Eric Mathet, IAGE Secretary

International Conference on FatigueInternational Conference on FatigueNAPA, CA, USANAPA, CA, USA

July 31 - Aug 2, 2000July 31 - Aug 2, 2000

Organized by EPRICo-sponsored by USNRC

andOECD Nuclear Energy Agency/Committee on

the Safety of Nuclear Installations (CSNI)

3-10


OECDOECD■ Organization for Economic Co-operation and

Development■ W. Europe, N. America, Japan■ 29 members - in 1996 the Czech Republic,

Hungary, Poland and S. Korea joined◆ OECD is

✦ a tool for intergovernmental co-operationmainly in the economic field

✦ a place for policy makers to compare point ofviews and experiences

3-11


NEA (Nuclear Energy Agency)NEA (Nuclear Energy Agency)■ Division of OECD■ 27 Member Countries■ 80 staff■ Mission is to assist its Member countries

◆ in maintaining and further developing, throughinternational co-operation, a peaceful andeconomical use of nuclear energy

◆ to provide authoritative assessments and to forgecommon understandings on key issues

3-12


NEA (Nuclear Energy Agency)/CSNINEA (Nuclear Energy Agency)/CSNI

■ Nuclear Safety Division provides secretariat

for CSNI

■ CSNI works through four WGs and two

Special Expert Groups (SEGs)

■ WG on Structural Integrity of Components

and Structures

3-13


NEA NEA StrenghtsStrenghts

■ Homogeneity in membership◆ small club◆ like-minded approach◆ climate of mutual trust◆ relatively non political

■ Provides added value and is cost effective■ Strong scientific/technical/legal work

◆ narrow focus◆ does not deal with proliferation, safeguards

■ Work methods flexible and responsive to memberneeds

3-14


Four Working Groups andFour Working Groups andtwo Special Expert Groupstwo Special Expert Groups

■ Risk assessment■ Analysis and management of Accidents■ Integrity of Components and Structures■ Operating experiences■ Human and organisational factors■ Fuel Safety Margins■ Co-ordination with CEC, IAEA, WANO etc

3-15

Fuel Cycle Safety

COMMITTEE on the SAFETY OF NUCLEAR INSTALLATIONS (CSNI)Chairman: M. Livolant - Secretary: G.M. Frescura CSNI Bureau: A. Thadani, A. Alonso

OperatingExperiences

Integrity ofComponents

and Structures

Chairman:Secretary:

Chairman:Secretary:

Chairman:Secretary:

Chairman:Secretary:

Analysis andManagementof Accidents

Chairman:Secretary:

Fission ProductBehaviour Seismic Behaviour

of Structures

Concrete StructuresAgeing

Integrity of MetalComponents &Structures

Design Basis

DATA BasesIncident ReportingSystem (IRS)ICDE Project -CommonCause Failure DataComputer-Based SafetySystemsFuel Incident & ReportingSystem (FINAS)

Fire Risk Assessment

Human Reliability -Errors of Commission

RiskAssessment

CSNI Programme Group

Severe Accidents

Chairman:Secretary:

Working Groups Special Expert Groups

Human andOrganisational

FactorsFuel Behaviour

NEA Co-operationwith CEEC and NIS

Fuel Cycle Safety

COMMITTEE on the SAFETY OF NUCLEAR INSTALLATIONS (CSNI)Chairman: M. Livolant - Secretary: G.M. Frescura CSNI Bureau: A. Thadani, A. Alonso

OperatingExperiences

Integrity ofComponents

and Structures

Chairman:Secretary:

Chairman:Secretary:

Chairman:Secretary:

Chairman:Secretary:

Analysis andManagementof Accidents

Chairman:Secretary:

Fission ProductBehaviour Seismic Behaviour

of Structures

Concrete StructuresAgeing

Integrity of MetalComponents &Structures

Design Basis

DATA BasesIncident ReportingSystem (IRS)ICDE Project -CommonCause Failure DataComputer-Based SafetySystemsFuel Incident & ReportingSystem (FINAS)

Software Reliability

Human Reliability -Errors of Commission

Risk Monitor

RiskAssessment

CSNI Programme Group

Severe Accidents

Chairman:Secretary:

Working Groups Special Expert Groups

Human andOrganisational

Factors

Fuel SafetyMargins

SANDIA LOWERHEAD FAILURE

PROJECT

CABRI WATERLOOP PROJECT

HALDEN REACTORPROJECT

RASPLAV

Passive Systems PSA

New CSNI structure3-16


WG on Integrity of Components andWG on Integrity of Components andStructuresStructures (IAGE) (IAGE)

Since 1996 has 3 sub-groups

◆ Integrity of metal components and structures

◆ Aging of concrete structures

◆ Seismic behavior of structures

3-17


Typical IAGE WG activitiesTypical IAGE WG activities

■ Workshops, Specialists Meetings■ Benchmarks/round robins/ISPs■ State of the Art Reports, Topical reports

3-18


■ Conference on fatigue, IPSN,Paris,1998

■ CSNI Approved proposal (Dec. 99)■ Co-sponsorship of the

International Conference onFatigue, EPRI, July-Aug 2000

■ IAGE WG program of work (June 2000)

Activities on Thermal FatigueActivities on Thermal Fatigue

3-19


◆ need to develop further accepted methods toidentify locations with potential risk of thermalfatigue;

◆ Monitoring of temperature fluctuations can beseen as an important part of the defence and, atpresent, it remains the most reliable method toavoid unanticipated incidents;

◆ If a greater reliance is to be placed on NDE, thedevelopment of qualified methods is needed.

OUTPUTS of the Conference on fatigue, Paris, 1998

3-20


a need to develop screening criteria toguide monitoring and inspection.

I.e.

a need to develop reliable methods toidentify locations in the reactor coolant andassociated systems, which are sensitive tothermal stratification and thermal mixing.

CSNI Approved Proposal, Dec 1999CSNI Approved Proposal, Dec 1999

Based on:

3-21


■ Topics to be discussed are

! Design for conservative estimates of thermal loads;

! Design and operation practices that reduce thermalstratification/mixing problems;

! Actions taken to mitigate thermal stratification/mixingproblems;

! Thermal stratification/mixing in conventional power plants;

! Thermal stratification/mixing problems with Russiandesigned reactors.


3-22


■ Other topics that could be investigated subsequently include:

" Basic causes of fatigue cracking by thermal stratification andthermal mixing;

" Effective inspection methods;

" Development of improved and reliable methods for the detection offatigue cracks especially in critical components like elbows;

" Advanced methods for the detection of the early phases of fatiguecracking;

" Consideration of benchmarks for comparing present designmethods.


3-23


At its last meeting in June 2000, the Metal sub-groupdiscussed thoroughly the significance of

◆ screening criteria to determine critical locations;

◆ management of interfaces between system

engineers, thermal hydraulic and mechanical

engineers;

◆ tendency for design code changes;

◆ experimental information available after all those

years of work.

IAGE WG discussionsIAGE WG discussions

3-24


➀ Conduct an international survey on:

" screening criteria;

" experiments available that address high cycle thermalfatigue;

" the tendency for code changes, if any.

➁ Conduct international benchmarks on technicalissues (e.g. thermal hydraulic, crack initiation andpropagation, NDE techniques and capabilities).

Follow-up actionsFollow-up actions

3-25

4-1

4 FATIGUE OF REACTOR COMPONENTS: NRCACTIVITIES

D. KalinouskyJ. Muscara

U.S. Nuclear Regulatory CommissionWashington, D.C., 20555

4-3

Fatigue of Reactor Components: NRC Activities



4-4

Fatigue of Reactor Components: NRC Activities



Abstract

The NRC has been dealing with the fatigue of reactor components since the late 1970’s.The NRC has initiated a Fatigue Action Plan and has identified three Generic SafetyIssues dealing with fatigue; GSI-78, “Monitoring of Design Basis Transient FatigueLimits for Reactor Coolant System,” GSI-166, “Adequacy of Fatigue Life of MetalComponents,” and GSI-190, “Fatigue Evaluation of Metal Components for 60-YearLife.” It has also supported research on fatigue at the Argonne National Laboratory.

It has been found in the above studies that the risk of core damage from fatigue failuresof primary system components is small and does not justify any additional actions bythe licensees. Although, for a twenty-year license renewal period, an increase in thefrequency of fatigue failures has been identified. This increase in failure frequencyneeds to be addressed during license renewal through an aging management program.

The ASME design curves have also been found to be non-conservative under certainreactor coolant environments and loading conditions. This is an issue that the NRC staffhas discussed with ASME Code committees over the last 12 years. The NRC has askedthe ASME Code to take actions to address this issue and has recommended twoacceptable methods for incorporating the effects of the environment in code fatigueanalyses. The first method is to develop the design curve in the conventional mannerusing the data from specimens tested in reactor environments, and the second methodis to use a fatigue life correction factor.

4-5

Introduction

Concerns about fatigue in reactor components arose in the late 1970’s with reports offatigue crack initiation and growth occurring in operating reactor components. Theseoccurrences were discussed in the NRC’s IE Bulletin 79-13 (1). A further concern wasthe possibility of thermal or mechanical transients that were not included in theoriginal fatigue design basis. NRC Bulletin 88-08 (2) and NRC Bulletin 88-11 (3) wereissued to initially address this concern.

To further assess these concerns, the NRC initiated several actions. It identified severalGeneric Safety Issues dealing with fatigue and developed a Fatigue Action Plan toresolve these issues; in addition, research was initiated on the various fatigue topics.This research has been used to resolve these generic issues and to identify issues forfurther investigation. Currently, there is ongoing fatigue research sponsored by theNRC under its Environmentally Assisted Cracking program.

Fatigue Action Plan

In 1993, the Commission instructed the NRC staff to treat fatigue as a potential safetyissue for operating reactors. The staff’s response was the development of the FatigueAction Plan (FAP). The FAP was developed to resolve three principal issues:

• Many older vintage nuclear power plants have components of the reactorcoolant pressure boundary (RCPB) that were designed to industry codes that didnot require the explicit fatigue analysis presently required by the AmericanSociety of Mechanical Engineers (ASME) Boiler and Pressure Vessel Code. Aconcern was raised regarding the fatigue resistance of these components for theplant design life.

• Test data showed that the design fatigue curves of the ASME Code were notconservative for nuclear power plant primary system environments (4). Aconcern was also raised regarding the fatigue resistance of components designedusing these ASME Code curves.

• The appropriate corrective action to be taken when the fatigue allowable limit isexceeded (calculated Cumulative Usage Factor (CUF) > 1) has been a subject ofcontroversy. The staff identified a need to develop a staff position on this subject.

One of the tasks under the FAP was to conduct a survey to determine the number ofoperating plants that have a fatigue analysis of the vessel, primary system components,and piping. This was done by a review of the available NRC licensing documentation.The survey found that approximately 40 percent of the operating plants have primarysystem components designed to a piping code that did not require a formal fatigueanalysis.

4-6

To address the technical issues, the FAP called for the fatigue evaluation of a sample ofRCPB components. The Idaho National Engineering Laboratory (INEL), under contractto the NRC, conducted assessments of sample components selected from seven powerplants. The components were selected from an EPRI report (5) listing typical areas inPWRs and BWRs for which high fatigue usage factors have been calculated. In thisanalysis, interim fatigue design curves developed at Argonne National Laboratory(ANL) were used (4). The NRC published the results of the INEL componentevaluations in NUREG/CR-6260 (6).

The evaluation of the sample of components with high design usage factors, using theinterim design fatigue curves, indicated that the ASME fatigue criteria (CUF<1) couldbe met for the majority of components when conservatisms in the original componentdesign were taken into account. A detailed finite element analysis on two of thecomponents indicated that additional conservatisms could be removed. In addition,data from transient monitoring instrumentation indicated that design transients forsome components were conservative. Even considering detailed finite element analysisand plant transient monitoring, there are still a limited number of piping locationswhere the ASME Code fatigue criteria may be exceeded at the end of design life.

The staff took the position that when the CUF for a component exceeded 1, GenericLetter (GL) 91-18, “Information to Licensees Regarding Two NRC Inspection ManualSections on Resolution of Degraded and Nonconforming Conditions and onOperability,” (7) would apply. The GL stipulates that the failure to meet a codecriterion specified in the FSAR is a nonconforming condition, which requires acorrective action. Since the license basis code is specified in the FSAR, the guidance inGL 91-18 is applicable.

On the basis of the accomplished FAP tasks, the staff found that no immediate staff orlicensee action was necessary to deal with the fatigue issues addressed. The FAP wascompleted in September of 1995 (8).

Generic Safety Issue - 78

Generic Safety Issue (GSI) – 78, “Monitoring of Design Basis Transient Fatigue Limitsfor Reactor Coolant System,” was developed in 1992 to determine whether transientmonitoring was necessary at operating plants. The goal of the transient monitoringaddressed by GSI-78 was to provide assurance that components do not exceed theirlicensing basis during the lifetime of the plant. The GSI-78 transient monitoring concernwas addressed as part of the FAP.

The resolution of GSI-78 was accomplished by focusing on the evaluation of risk fromfatigue failure of selected reactor coolant system components over a 40-year period. AGSI-78 resolution package was presented to the Advisory Committee on Reactor

4-7

Safeguards with a discussion of the results of a study using the PRAISE code. Thisstudy demonstrated that the risk from fatigue failure of the reactor coolant systemcomponents was very low. Thus, GSI-78 was closed in February of 1997.

Generic Safety Issue – 166

GSI-166, “Adequacy of Fatigue Life of Metal Components,” was established in responseto questions raised by the staff in SECY-93-049, “Implementation of 10 CFR Part 54,‘Requirements for Renewal of Operating Licenses for Nuclear Power Plants,’” (9) aboutreassessing the fatigue design of metal components. The output from the FAP and GSI-78 provided essential technical information for the resolution of GSI-166.

In resolving GSI-166, the staff obtained the records of transient monitoring from theseven plants selected in the FAP assessments. Based on these records and conservatismsidentified in the component analyses, the staff did not believe there existed a significantsafety concern that current licensing basis fatigue criteria had been exceeded atoperating plants over a 40-year life. Thus, no action was required for the licensees as aresult of the evaluations performed under GSI-166 and the GSI was closed in Februaryof 1997.

Generic Safety Issue – 190

GSI-166 and GSI-78 only applied to the original 40-year plant life. To address thepossibility of a 20-year license renewal period, the staff opened GSI-190, “FatigueEvaluation of Metal Components for 60-Year Life,” analyzing the same componentsfrom the reactor coolant system as for GSI-166, but for a 60 year life in the LWRenvironment. GSI-190 addressed the environmental effects on design basis fatiguetransients, studying the probability of fatigue failure, and the associated core damagefrequency (CDF) for a 60-year plant life. It did not address all aspects of fatigue relateddegradation, including those outside the design basis. Events involving thermal ormechanical stresses that result in unanticipated cyclic loads are outside the scope of thedesign basis transients and were not studied under GSI-190.

The approach to resolving GSI-190 built upon the approaches used in the FAP and theprevious GSIs. The Pacific Northwest National Laboratory (PNNL), under contract tothe NRC, made use of the most recently developed fatigue curves in simulated LWRenvironmental conditions from Argonne National Laboratory (ANL) (10, 11, 12).PNNL performed calculations of the probability of component failure and the CDFassociated with these failures. These results are reported in NUREG/CR-6674, “FatigueAnalysis of Components for 60-Year Plant Life” (13).

4-8

Probabilistic fatigue calculations using the pc-PRAISE code were performed on 47sample components from 6 locations in seven example plants (five PWR and two BWRplants). During this work the staff identified that the pc-PRAISE code could not modellarge aspect ratio cracks (ratio of crack length to crack depth), nor could it model thejoining of several small cracks to make a large crack which could subsequentlypropagate through the wall thickness of the component. The staff concluded that theeffect of large aspect ratios was an important factor in fatigue analyses and there was aneed to modify the pc-PRAISE code to model this. The modified pc-PRAISE program(version 4.2) is available via the NRC’s ADAMS system, in a package titled pcPRAISE.

Using the modified pc-PRAISE program, PNNL performed a probabilistic analysis forcrack initiation and through wall crack growth in the components mentioned above for40 and 60-year plant lives, considering both air and LWR (water) environments.Calculations for the air environment and the 40-year life were done to provide abaseline from which to compare the results of a 60-year life, incorporating the effect ofthe revised fatigue curves coupled with the modified pc-PRAISE program. Anevaluation was performed to estimate the conditional CDF from the fatigue failure ofthese components. Given a through wall crack, the objective was to estimate theconditional probabilities of a small leak, of a large leak, and of a pipe break. Data onpipe failure events indicate that only a small fraction of through-wall flaws result inlarge leaks or breaks. Probabilistic fracture mechanics models, like the one contained inthe pc-PRAISE program, predict that fatigue failures will usually be in the mode ofsmall leaks rather than large leaks or breaks. From the conditional probabilities ofsmall leak, large leak, and pipe break, the conditional CDF was estimated for the sevenexample plants based on results extracted from probabilistic risk assessments (PRA).The major findings from the calculations include the following points:

• Many of the components have cumulative probabilities of crack initiation andcumulative probabilities of through-wall cracks that approach unity within the40 to 60-year time period. However, some components, often with similarvalues of fatigue usage factors, show much lower failure probabilities. This isrelated to the fact that the usage factors address only crack initiation and do notaddress the specific factors for each component that determine how likely it isthat an initiated crack will grow through wall.

• The maximum failure rate is in the range of 10-2 through-wall cracks per reactoryear, and those failures were associated with high CUF locations. The 10 highestcomponent failure rates for 40-years in air, 40-years in water, and 60-years inwater are shown in Figure 1.

• Failure rates for other components having much lower failure probabilities arechanged by as much as an order of magnitude from 40 to 60-years, but thesecomponents make relatively small overall contributions to the cumulative CDFestimates.

4-9

• The maximum CDF based on these calculated failure rates is about 10 -6 per year.These maximum values correspond to components with very high cumulativefailure probabilities, and the failure rates do not change much from 40 to 60years. The range of CDF was between 10-6 to 10-15. The 10 highest CDF rates for40-years in air, 40-years in water, and 60-years in water are shown in Figure 2.

Based upon these low CDFs, the staff concluded that they could not be used as a basisfor a cost/benefit backfit analysis to justify imposition of a new regulatory requirementon operating reactors. However, the calculations supporting resolution of this issue,which included consideration of environmental effects, indicates the potential for anincrease in the frequency of pipe leaks as plants continue to operate. Thus, the staffconcluded that, consistent with existing requirements in 10 CFR 54.21, licensees shouldaddress the effects of the coolant environment on component fatigue life as agingmanagement programs are formulated in support of license renewal.

NRC Research

ANL has been conducting a research program for the NRC on environmentally assistedcracking since 1978. As part of this program, the environmental effects on fatigue crackinitiation in pressure vessel and piping steels has been studied since 1986. Several keyvariables have been identified that influence the steel’s fatigue life; such as: steel type,sulfur content, strain amplitude, strain rate, temperature, and dissolved oxygencontent. Test data from ANL and Japan have shown potentially significant effects ofLWR environments on the fatigue resistance of pressure vessel and piping steels.Based on the existing fatigue strain vs. life (S-N) data, interim fatigue design curveshave been developed that account for environmental effects and the results werereported in NUREG/CR-5999. A more rigorous statistical analysis of the available datawas performed and statistical models were developed in NUREG/CR-6335 forestimating the fatigue lives of carbon and low-alloy ferritic steels, austenitic stainlesssteels, and Alloy 600 in LWR environments. The results have been used to estimate theprobability of fatigue cracking. As more experimental data became available, thestatistical models have been modified and/or optimized with a larger fatigue database.The ANL work performed more recently on fatigue in LWR environments has beensummarized in topical reports NUREG/CR-6583 (14) for carbon and low-alloy steels,and NUREG/CR-5704 (15) for wrought and cast austenitic stainless steels. The reportspresent an overview of the existing fatigue S-N data on these steels and describe theinfluence of key service and material parameters on fatigue life, the mechanism of crackinitiation in LWR environments, and procedures for incorporating environmentaleffects in ASME Section III fatigue evaluations.

Two possible procedures have been identified for incorporating the effects of LWRenvironments into fatigue evaluations. The first procedure calls for environmentallyadjusting the ASME design curve, such that it takes into account the shorter fatigue life

4-10

of steels in the LWR environment. The design curve is developed in the conventionalmanner by applying the factor of 2 on stress or 20 on cycles to the S-N test datadeveloped from specimens tested in LWR environments. The second procedure,originally developed in Japan by Higuchi and Iida (16), and later proposed in the U. S.by GE and EPRI (17) uses a fatigue life correction factor, Fen, which is the ratio of thefatigue life in air at room temperature to that in water at service temperature. Eitherprocedure provides acceptable results and the NRC staff can support their use in theASME Code.

Further research to be sponsored by the NRC at ANL over the next few years willprovide information on such topics as the synergistic effects of surface finish andenvironment on fatigue life; the combined effects of loading sequence andenvironment; and fatigue crack initiation in sensitized stainless steel. A comprehensiveevaluation of stainless steel fatigue test specimens will be performed to explain whyenvironmental effects are more pronounced in low dissolved oxygen than highdissolved oxygen water. Several of these planned research tasks are:

• Fatigue tests on low-alloy steel specimens with differing surface finishes andtreatments will be carried out in air and water environments. This will provideinformation on the effects of surface roughness on the fatigue life.

• Metallographic evaluations of stainless steel specimens will be conducted to helpexplain why environmental effects are more pronounced in low dissolvedoxygen water than in high dissolved oxygen water. The metallographicevaluations will also help to understand the mechanism of fatigue crackinitiation in a LWR environment.

• Fatigue tests will be carried out on sensitized Type 304 stainless steel.

• The effect of intermediate temperatures on the fatigue life of austenitic stainlesssteels will be investigated.

• Existing fatigue data does not include the combined effects of environment andsequential loading. Therefore, fatigue tests on low-alloy steels in high dissolvedoxygen water will be conducted to investigate these effects on fatigue life.

Need to Address Environmental Effects and NRC Interaction with ASME

Results from studies in Japan and those at ANL illustrate potentially significant effectsof LWR coolant environments on the fatigue life of carbon and low–alloy steels and ofaustenitic stainless steels (Figures 3 and 4). Under certain loading and environmentalconditions the reactor water environmental effects alone substantially exceed thereductions in the current design curve to account for the differences between specimen

4-11

tests and component behavior. In addition, the current Code design curve for stainlesssteel in air does not accurately represent the available experimental data and thusincludes a reduction of only 1.5 and ~10 from the mean air curve, not the 2 and 20originally intended (Figure 5). Based on these studies and findings, NRC and ANL staffhave been recommending to ASME Code committees and to the Board on NuclearCodes and Standards (BNCS) since the late 1980’s that the ASME Code fatigue designcurves need to be updated to incorporate the effects of reactor coolant environments.The need to address this issue continues to become more important in light of the agingof plants, changes in other parts of the fatigue analyses addressed by the Code, and therenewal of plant licenses for an additional 20 years of operation. Ample data now existto support the required changes.

Although the existing fatigue design curves are non-conservative for certain LWRconditions and environments, the current design procedures, e.g., stress analysis rules,cycle counting, etc., are conservative enough that the overall assessment of fatigue lifehas been conservative. However, the Code permits new improved approaches forfatigue evaluations, e.g., finite element analyses, fatigue monitoring, improved Kefactors, etc., which can significantly reduce the conservatism in the other elements ofthe present design methods. To ensure that the overall assessment maintains a degreeof conservatism consistent with that chosen for the fatigue limit in air, the currentunderstanding regarding environmental effects should be incorporated into the ASMECode for fatigue evaluations.

The ASME Code fatigue design curves, given in Appendix I of Section III, are based onstrain–controlled tests of small polished specimens at room temperature in air. Thefatigue design curves were developed from the mean curves for the experimental databy reducing the fatigue life at each point on the mean curve by a factor of 2 on strain or20 on cycles, whichever was more conservative. This reduction was intended toaccount for data scatter (heat–to–heat variability), effects of mean stress or loadinghistory, and the differences in surface condition and size between the test specimensand actual components. As explicitly noted in Subsection NB–3121 of Section III of theCode, the effects of the LWR coolant environment are not addressed in the currentCode fatigue design curves. The Subsection states that the owner's designspecifications should provide information on any reduction to fatigue design curvesnecessitated by environmental conditions.

In 1991, the ASME BNCS requested the PVRC to examine the existing worldwidefatigue strain vs. life (S–N) data and develop recommendations for ASME. The PVRCcommittee on cyclic life and environmental effects has evaluated the issue and, at itsJune 15, 1999 meeting in Columbus, Ohio (18), the committee endorsed a method forincorporating the effects of LWR coolant environments into the ASME Code fatigueevaluations. The Executive Director of PVRC has transmitted their recommendationsand approach to implement environmental fatigue procedures to the ASME BNCS by

4-12

letter from Hollinger to Ferguson dated October 31, 1999. The methodology, proposedby EPRI/GE (17), is based on the use of fatigue life correction factors (Fen) to account forenvironmental effects on fatigue life.

The PVRC recommendation also defines an “effective” F en, Fen/Z, where Z is a factorthat constitutes the perceived conservatism in the Code design curves, i.e., the portionof that factor of 20 that is judged not actually necessary to account for data scatter,surface finish, etc. PVRC proposed the factor Z as 3.0 for carbon and low–alloy steelsand 1.5 for stainless steels. Recent reports from Japan (19) and also findings from theANL work (20) indicate that the entire margin of 20 is expended by factors other thanthe environmental effects. Also, the effective F en approach presumes that all otheruncertainties have been anticipated and accounted for. Therefore we do not believe it isprudent to apply the factor Z to Fen.

The NRC sponsored work at ANL has also studied methods for incorporatingenvironmental effects into ASME Code Section III fatigue evaluations (NUREG/CR-6583, NUREG/CR-5704). Two procedures have been proposed; (a) use ofenvironmentally adjusted fatigue design curves or (b) use Fen to adjust the currentASME Code fatigue usage values for environmental effects. Although estimates offatigue lives based on the two methods may differ somewhat because of differencesbetween the ASME mean curves used to develop the current design curves and thebest–fit curves to the existing data used to develop the environmentally adjustedcurves, the NRC can support either of these methods as they provide an acceptableapproach to account for environmental effects.

In December 1999, the NRC again requested the ASME Code to address the effects ofthe LWR environment on fatigue life, and to update the current design curve forstainless steel in air. NRC staff and contractors are participating in Section III and XIactivities to address and incorporate the effects of environment into ASME Code fatigueevaluations. The Section III Subgroup on Fatigue Strength has begun to revise thefatigue design curves. ASME Section XI Task Group on Operating Plant FatigueAssessments is considering the possible inclusion of a methodology for addressing theeffects of the environment on fatigue life in its revision of Non-Mandatory Appendix L.

Thermal Fatigue

In 1997, an event occurred at Oconee 2, which involved the unit being shut down dueto cracking and leakage from a weld location in the 2½-inch diameter ( NPS 2½ ), Class1 portion of a combination makeup and high-pressure injection line (equivalent to aportion of the High Pressure Safety Injection (HPSI) system as designated in the ASMECode).

4-13

After the failed nozzle assembly was removed from service and a metallurgicalexamination performed, it was determined that the crack consisted of a 360 degreeinside surface flaw with minimum depth of 30 percent through-wall. The crack hadalso penetrated completely through-wall over an arc length of 77 degrees. The licenseefor Oconee attributed the cracking to thermal cycling and flow-induced vibration.

As a result of this event, the inservice inspection (ISI) requirements given in Section XIof the ASME Code for butt-welded connections in HPSI piping were reexamined.Subsection IWC of Section XI of the ASME Code requires that Class 2 HPSI pipingdown to NPS 2 receive both a volumetric and a surface examination as part of afacility’s ISI program. Vessels, piping, pumps, valves and other components NPS 1½and smaller in HPSI systems of pressurized water reactor plants are, however,specifically exempted from this requirement. Subsection IWB of Section XI of theASME Code requires only surface examination for Class 1 butt-welded pipingconnections less than NPS 4, with the one exemption provision excluding piping ofNPS 1 and smaller from examination. Therefore, for the HPSI system, the inspectioncriteria for Class 2 piping between NPS 4 and NPS 2 (or NPS 1 ½), inclusive, are morecomprehensive than those for Class 1 piping of the same size range. Moreover, theexamination requirements specified for Class 1 HPSI piping may not be adequate todetect thermal fatigue cracking initiating on the inside diameter in a timely mannersince volumetric examinations are not required.

In response to issues raised by the NRC on the inspection criteria and the thermalfatigue event at Oconee 2, the industry proposed an initiative to address the issue ofthermal fatigue in nuclear power plant piping. The program was initiated under theElectric Power Research Institute (EPRI) Pressurized Water Reactor (PWR) MaterialsReliability Project (MRP). The goal of the Thermal Fatigue Issue Task Group(established/approved in March 1999 by the MRP) is to provide a consistent set ofguidelines and a methodology for addressing thermal fatigue issues. The NRC and theEPRI MRP periodically meet to discuss progress and issues related to thermal fatigue.It is expected that the task group will develop screening, monitoring, inspection, andmanagement guidelines addressing thermal fatigue in the 2000-2001 timeframe.

Because thermal fatigue events that are not part of the design basis were not evaluatedunder the generic safety issues on fatigue, and because thermal fatigue cracking eventshave recently been reported in other plants, including France and Japan, the NRC willfollow closely the work, guidelines and implementations being developed under thisindustry initiative.

Conclusion

Analysis of the fatigue issues through the resolved GSIs and the completed FAP haveresulted in no new requirements being imposed on the licensees for their current

4-14

licensing period. For license renewal, it has been found that the increase in throughwall crack frequency with an additional 20-year period of operation makes it necessaryfor the licensees to consider fatigue in their aging management programs.

Research sponsored by the NRC has provided results that have helped resolve genericissues related to fatigue. It is providing data, results and methodologies that could beused to update the ASME Code for incorporating the effects of the LWR environment infatigue analyses or for the NRC to develop its own guidance.

References

1. U. S. Nuclear Regulatory Commission, IE Bulletin 79-13, Cracking in FeedwaterSystem Piping, October 16, 1979.

2. U. S. Nuclear Regulatory Commission, NRC Bulletin 88-08, Thermal Stress inPiping Connected to Reactor Coolant Systems, August 4, 1988.

3. U. S. Nuclear Regulatory Commission, NRC Bulletin 88-11, Pressurizer Surge LineStratification, December 20, 1988.

4. S. Majumdar, O.K. Chopra, and W.J. Shack, Interim Fatigue Design Curves forCarbon, Low-Alloy, and Austenitic Stainless Steels in LWR Environments, NUREG/CR-5999,1993.

5. Electric Power Research Institute, Report TR-100252, Metal Fatigue in OperatingNuclear Power Plants, April 1992.

6. A.G. Ware, D.K. Morton, and M.E. Nitzel, Application of NUREG/CR-5999 InterimFatigue Curves to Selected Nuclear Power Plant Components, NUREG/CR-6260, 1995.

7. U. S. Nuclear Regulatory Commission, GL 91-18, Information to LicenseesRegarding Two NRC Inspection Manual Sections on Resolution of Degraded andNonconforming Conditions and on Operability, November 7, 1991.

8. SECY-95-245, Completion of the Fatigue Action Plan, September 25, 1995.

9. SECY-93-049, Implementation of 10 CFR Part 54,’‘Requirements for Renewal ofOperating Licenses for Nuclear Power Plants ,” March 1, 1993.

10. J. Keisler, O.K. Chopra, and W.J. Shack, Statistical Analysis of Fatigue Strain-LifeData for Carbon and Low-Alloy Steels, NUREG/CR-6237, 1994.

4-15

11. J. Keisler, O.K. Chopra, and W.J. Shack, Fatigue Strain-Life Behavior of Carbon andLow-Alloy Steels, Austenitic Stainless Steels, and Alloy 600 in LWR Environments,NUREG/CR-6335, 1995.

12. O.K. Chopra and F.A. Simonen, Private Communication, “Updated FatigueDesign Curves for Austenitic Stainless Steel in LWR Environments,” 1998.

13. M.A. Khaleel, F.A. Simonen, et al., Fatigue Analysis of Components for 60-Year PlantLife, NUREG/CR-6674, 2000.

14. O.K. Chopra and W.J. Shack, Effects of LWR Coolant Environments on FatigueDesign Curves of Carbon and Low-Alloy Steels, NUREG/CR-6583, 1998.

15. O.K. Chopra, Effects of LWR Coolant Environments on Fatigue Design Curves ofAustenitic Stainless Steels, NUREG/CR-5704, 1999.

16. M. Higuchi and K. Iida, Fatigue Strength Correction Factors for Carbon and Low-Alloy Steels in Oxygen-Containing High-Temperature Water, Nucl. Eng. Des., 129, 1991.

17. Electric Power Research Institute, Report TR-105759, An Environmental FactorApproach to Account for Reactor Water Effects in Light Water Reactor Pressure Vessel andPiping Fatigue Evaluations, December 1995.

18. Welding Research Council Program Report, Vol. LIX No. 5/6, May/June 1999.

19. M. Higuchi, Fatigue Curves and Fatigue Design Criteria for Carbon and Low-AlloySteels in High-Temperature Water, Probabilistic and Environmental Aspects of FractureMechanics, PVP Vol. 386, 1999.

20. O.K. Chopra and J. Muscara, Effects of Light Water Reactor Environments on FatigueCrack Initiation in Piping and Pressure Vessel Steels, ICONE-8, Baltimore, MD, April 2000,Paper No. 8300.

4-16

1. GE Old – Recirc RHR return line T2. Westinghouse New - RHR line inlet transition3. CE New - Surge line elbow4. Westinghouse New - Charging nozzle nozzle5. CE Old - Surge line elbow6. Westinghouse New - RPV outlet nozzle7. B & W - RPV outlet nozzle8. CE Old - RPV outlet nozzle9. GE New – RHR line straight pipe10. GE Old – Core spray system safe end

Figure 1: Through Wall Crack Frequencies

1E-07

1E-06

1E-05

1E-04

1E-03

1E-02

1E-01

1E+00

1 2 3 4 5 6 7 8 9 10

Component

Air: 40 Yr Water: 40 Yr Water: 60 Yr

Through Wall Crack Frequency

4-17

1. Westinghouse New - RHR line inlet transition2. CE New - Surge line elbow3. CE Old - Surge line elbow4. Westinghouse New - Charging nozzle nozzle5. Westinghouse New - RPV outlet nozzle6. B & W - Makeup/HPI nozzle safe end7. B & W - RPV outlet nozzle8. CE Old - RPV outlet nozzle9. B & W - Decay heat removal line reducing T10. Westinghouse Old - RPV outlet nozzle inside surface

1E-12

1E-11

1E-10

1E-09

1E-08

1E-07

1E-06

1E-05

1 2 3 4 5 6 7 8 9 10

Component

Air: 40 Yr Water: 40 Yr Water: 60 Yr

Core Damage Frequency

Figure 2: Core Damage Frequencies

4-18

Figure 3: S-N Data for Carbon Steel in Water

Figure 4: S-N Data for Stainless Steel in Water

4-19

Figure 5: Current ASME Mean Curve

Fatigue of Reactor Components:NRC Activities

Joe MuscaraDoug Kalinousky

U. S. Nuclear Regulatory CommissionOffice of Nuclear Regulatory ResearchWashington, DC 20555

INTERNATIONAL CONFERENCE ON FATIGUEOF REACTOR COMPONENTS31 July – 2 August, 2000Napa, California

4-20

Introduction

• GSI-78, GSI-166, & Fatigue Action Plan

• GSI-190

• Research program accomplishments

• Need to address environmental effects

• NRC interaction with ASME

• Future activities4-21

GSI-78

• “Monitoring of Design Basis Transient FatigueLimits for Reactor Coolant System”– Originated in May 1983– Determine if transient monitoring was necessary at

operating plants– Resolved as part of FAP by analyzing risk– Risk of fatigue failure over 40 years is low

• No regulatory actions necessary

4-22

GSI-166

• “Adequacy of Fatigue Life of MetalComponents”– June 1993

– Resolved by identifying conservatisms incomponent analysis

– Based on a risk analysis, no significant safetyconcern that licensing criteria had been exceededfor 40 years

• Resolved with no licensee actions required4-23

Fatigue Action Plan

• Originated: July 1994; Completed: Sept 1995

• Used to resolve GSI-78 & GSI-166

• Three principal issues– Some components did not require a fatigue analysis

– ASME fatigue design curves not conservative in aLWR environment

– What to do when CUF>1

4-24

GSI-190

• “Fatigue Evaluation of Metal Components for60-Year Life”

• Carried on GSI-78 & GSI-166 for a 20-yearlicense renewal period

• Considered environmental effects

• Estimated CDF due to fatigue failures fromdesign basis transients

4-25

GSI-190 (cont)

• Modified pcPRAISE (available in ADAMS)– Initiation: multiple sites, varying times

– Growth: from initiation until failure– Linking: can form large aspect ratio cracks

• Maximum failure rate, 10-2 per year• Maximum CDF, 10-6 per year• Address environment on component fatigue

life in aging management programs4-26

Research Program Accomplishments

• NRC initiated studies at ANL in 1986 on the effects of environment andloading on the fatigue life of LWR materials

• Obtained fatigue life data for different materials and test conditions– material: carbon steels; A106B, A333-B

low-alloy steels; A533-B, A302-BSSs; types 304 & 316NGcast SS; CF8M (with & w/o thermal aging)

– environment: air & water

– temperature: 288 C and room temperature

– dissolved oxygen: low DO; <10 ppbhigh DO; 400-800 ppb

– strain range: 1.0 - 0.32%

– strain rate: 0.4 - 0.00004%/s

• Characterized growth of small cracks in smooth fatigue test specimens todevelop a mechanism of crack initiation in CSs and LASs 4-27

Research Program Accomplishments (cont)

• Developed interim design curves for different operating parameters;such as, strain rate, DO, S, & temperature (NUREG/CR-5999, 93)

– SS curve based only on high DO data, interim curve considered

bounding for low DO environment

– Results used for fatigue evaluations of selected reactor components(NUREG/CR-6260, 95)

• Developed statistical models from all existing data for estimatingfatigue life (& probability of fatigue cracking) in carbon & low-alloy steels, austenitic SSs, and Alloy 600 in LWR environments(NUREG/CR-6335, 95)– Results used to assess fatigue life and risk for 40-year plant operation

(GSI-78 & GSI-166)

4-28

Research Program Accomplishments (cont)

• Statistical models were updated with evolving research results– NUREG/CR-6583, 98: CSs & LASs with additional data on DO

– NUREG/CR-5704, 99: additional data in low DO for wrought SSs &new data on cast SSs

• Results from above reports were used in resolution of GSI-190– Evaluation of fatigue life and risk for 60-year plant operation

– Leak frequency for 40 - 60 year operation

• From the above studies and data two procedures have evolved forincorporating effects of LWR environments into fatigue evaluations– Environmentally adjusted design fatigue curves

– Fatigue life correction factor, Fen (Japan/EPRI)

4-29

Need to Address Environmental Effects andNRC Interaction with ASME

Historical Perspective• Results from studies in Japan (Higuchi & Iida, 91) and ANL

(NUREG/CR-4667, 90; NUREG/CR-6583, 98; NUREG/CR-5704,99) illustrate potentially significant effects of LWR coolantenvironment on fatigue life of steels

• The current code design curve for stainless steel in air does notaccurately represent the available experimental data (Jaske &O’Donnell, 78) and thus includes a reduction of only 1.5 and ~10from the mean air curve, not the 2 and 20 originally intended

4-30


S-N Data for Carbon Steel in Water

4-31


S-N Data for Stainless Steel in Water

4-32


Current ASME Mean Curve

4-33


Historical Perspective (cont)• Although existing code fatigue design curves are non-conservative

for certain LWR conditions, design practices for fatigue life havebeen conservative (e.g., higher design loads, larger number ofcycles)

• With improved approaches for fatigue evaluations (e.g., finiteelement analyses, fatigue monitoring, etc.) environmental effectshould be incorporated into code fatigue evaluations to ensure adegree of conservatism consistent with that originally intended

4-34


Historical Perspective (cont)• Since late 80s, NRC staff have been involved in discussions

with ASME Code committees, PVRC, and technicalcommunity to address issues related to environmental effectson fatigue

• In 1991, ASME BNCS requested the PVRC to examineexisting worldwide fatigue S-N data and developrecommendations

• In 1995, resolution of GSI-78 and GSI –166 established that– Risk to core damage from fatigue failure of RCS is very small;

no action required for current plant design life of 40 years– The NRC staff believe that the fatigue issues should be evaluated

for extended period of operation for license renewal (under GSI-190) 4-35


Historical Perspective (cont)• In 1999, resolution of GSI-190 established that:

– Based on probabilistic analyses, sensitivity studies, interactions withindustry, and different approaches available to manage the effects ofaging, no generic regulatory action is required

– However, calculations supporting resolution of this issue, serviceexperience (NUREG/CR-6582), and nature of age related degradationindicate potential for increase in frequency of pipe leaks as plantscontinue to operate

– Consistent with requirements of 10 CFR 54.2, aging managementprograms for license renewal should address component fatigueincluding the effects of coolant environment

• In December 1999, the NRC requested the ASME Code toaddress the effects of the LWR environment on fatigue life,and to update the current design curve for Stainless Steel in air

4-36


Guidance for Managing Environmental Effects on Fatigue• The NRC will review aging management programs for fatigue in

support of license renewal on a case-by-case basis untilacceptable industry guidance is available

• We believe it would be useful for the ASME Code to developmethods for incorporating the environmental effects in fatiguelife analyses– NRC staff and contractors are working with ASME code committees in

Section III and XI for incorporation of such methods

• In the absence of ASME action in this area, the NRC may haveto develop the needed regulatory guidance

4-37


Methodology Available for IncorporatingEnvironmental Effects into Fatigue Evaluations

• Two methods are available– Using environmentally adjusted fatigue design curves– By applying environmental correction factor, Fen, to existing ASME

Code fatigue analyses

• Environmentally adjusted fatigue design curves andcorrelations for calculating Fen are contained in NUREG/CR-6583 and NUREG/CR-5704

• PVRC steering committee on Cyclic Life & EnvironmentalEffects endorsed Fen approach (May/June 1999); andtransmitted recommendations to ASME BNCS (Oct 1999)

• NRC can accept either method4-38

Future Activities

Complete NRC Research• Characterize growth of small cracks in SS specimens

– Better understanding of different behavior with respect to DO

• Evaluate the combined effects of loading sequence and environment onfatigue

• Effect of surface roughness / finish• Effect of intermediate temperature (EAC data)

Other Activities• Follow the industry research on thermal fatigue being conducted by

EPRI under the PWR Materials Reliability Project• Follow the new industry research to be conducted by EPRI to address

environmental fatigue issues and component testing• Continue to work with ASME Code committees to incorporate effects of

the environment into code fatigue analyses 4-39

5-1

5 A REGULATOR'S VIEW ON THE FATIGUE ISSUE

Louis Van der WielMinistry of EnvironmentThe Hague, Netherlands

Oral Presentation only. No presentation slides or technical paper available.

5-3

A REGULATOR’S VIEW ON THE FATIGUE ISSUE

Louis Van der WielMinistry of EnvironmentThe Hague, Netherlands

Oral Presentation only. No presentation slides or technical paper available.

THERMAL FATIGUE I

6-1

6 THERMAL FATIGUE IN FRENCH RHR SYSTEM

C. Faidy, T. Le CourtoisElectricité de France

EDF-SEPTEN

E. de FraguierElectricité de France

EDF-CNEN

J-A Leduff, A. LefrançoisFRAMATOME

J. DechelotteElectricité de France

EDF-DPN

6-8

The conclusions of these fracture analyses are :

- damage mechanism confirmed : high cycle thermal fatigue

- no fabrication defect

- deep cracks (one through wall crack and some others up to 80% of the wall thickness)along longitudinal and circumferential welds, starting at the root of the weld on the innersurface

- lot of small cracks in the weld area, mainly in the ground counterbore areas and highcompressive residual stresses regions (less than 3 mm depth)

- just few cracks in the base metal, generally in ground area, with very limited depth

Fatigue evaluation

In order to evaluate the lead factors of the different degradations, we reviewed the design andconstruction data and the operating experience of the RHRS.

The leak occurred on the first industrial cycle, at intermediate shutdown, with RHRS connectedand a RCS temperature of 180°C. CIVAUX 1 had more than 1500 hours of operation with alarge ∆T(over 150°C) in this mixing tee.

If we compare the different similar plants and the 2 trains of each of them, we have informationfor duration between 310 and 2800 hours (table 1).

N4 Plants Hours of operation

Train A / B

Maximum crack depth in the mixing area

Nominal thickness : 9.3 mm

1 2700 / 2800 between 3.2 and 5.8 mm

2 1600 / 1700 between 0.9 and 4.2 mm

3 1500 / 1500 between 0.6 and 7.3 mm + 1 through wall

4 450 / 310 between 0. and 1.8 mm

Table 1 : Maximum crack depth in different location of the mixing areafor different similar plants

The maximum crack depth is in a reasonnable agreement with the number of hours ofoperation at large ∆T (over 150°C) with one critical location on plant 3 in the elbow (throughwall crack after 1500 hours) and more limited degradations on plant 1 with more than 2800hours of operation.

The welding process is in general TIG weld, except for some longitudinal welds that usedplasma root pass (figure 6) without weld metal. No fabrication defect has been found in anylocation.

- 8 -

6-12

Conclusions

A reasonable understanding of the CIVAUX 1 event has been obtained through a large anddetailed analysis program. A new design has been proposed for N4 plants. All repair piecesare designed and fabricated with specific requirements (compressive residual stresses on theinner surface and polished surface finished), with that improvement the expected life of thenew design will be over 4000 hours at large ∆T. Nevertheless, in-service inspection andmonitoring are required with shorter periods by French Safety Authority to include safetyfactors.

A specific attention has been done to all mixing tees with large ∆T in all safety class pipingsystems. A step by step procedure is yet under application, a screening criteria has beenproposed based on simple model and code fatigue curves.

A large complementary 5-years Research and Development program has been launched in1999, the first results are expected end of this year. It will support all our past analysis and hasto proposed some reliable procedure for flaw evaluation under high cycle thermal fatigue.

All the French Codes (RCC-M and RSE-M) will be periodically updated to take into account thelast developments on this subject : high cycle thermal fatigue.

_________________________________

- 12 -

7-1

7 RESULTS OF THERMAL STRATIFICATIONMEASUREMENTS AT NUCLEAR POWER PLANTBEZNAU

Suresh SahgalNOK, Nuclear Power Plant Beznau

Switzerland

8-1

8 LEAKAGE FROM CVCS PIPE OF REGENERATIVEHEAT EXCHANGER INDUCED BY HIGH CYCLETHERMAL FATIGUE AT TSURUGA NUCLEAR POWERSTATION UNIT 2

Takeshi SakaiThe Japan Atomic Power Company

Tsuruga Power StationMyoujin-cho 1-Banchi

Tsuruga-shi, Fukei-ken, Japan

8-3

LEAKAGE FROM CVCS PIPE OF REGENERATIVE HEAT EXCHANGER

INDUCED BY HIGH-CYCLE THERMAL FATIGUE

AT TSURUGA NUCLEAR POWER STATION UNIT 2

Takeshi SakaiThe Japan Atomic Power Company

Tsuruga Power StationMyoujin-cho 1-Banchi

Tsuruga-shi, Fukei-ken, Japan

ABSTRACTOn July 12, 1999 while Tsuruga-2, PWR 4-loop plant, was operating at full power (1,160 MWe),

unidentified leakage inside the primary containment vessel was detected. As the leakage was identified, theplant promptly started to proceed to cold shutdown. Visual inspection after an isolation of the CVCS(Chemical & Volume Control System) revealed that the leakage was from a connecting pipe between themiddle and lower stage in the CVCS regenerative heat exchanger.

The CVCS regenerative heat exchanger has three shells, i.e. the upper, the middle and the lower shell.Each heat exchanger shell has an inner cylinder containing a heat exchanger tube bundle. Reactor coolant iscooled while streaming inside the inner cylinder, however, the temperature of the coolant which flows outsidethe inner cylinder as a bypass flow keeps high. These two coolant flows are mixed around the outlet of theinner cylinder.

Thermal-hydraulic mock-up tests simulating internal flows in the heat exchanger were conducted alongwith thermal and structural analyses. As a result, the mechanism induced cracks was determined to be high-cycle thermal fatigue.

1. IntroductionOn July 12, 1999 while Tsuruga-2, PWR 4-loop plant, was operating at full power (1,160 MWe), the

containment sump alarm was also initiated, indicating an increasing leakage rate. As a result of a visual surveyinside the PCV, it was recognized that the source of leakage was the CVCS (Fig. 1).

The leakage was from a connecting pipe between the middle and lower stages of the CVCS regenerativeheat exchanger, as shown in Fig.2.

The detailed investigation of the cracked pipe revealed that 12 circumferential and longitudinal cracksincluding a through-wall crack were identified, as shown in Fig.3. A lot of cracks were found on the innersurface of the middle stage heat exchanger shell. The fracture surface of crack a in the connecting pipe isshown in Fig.4.

The fracture surfaces of the circumferential and longitudinal cracks had metallic crystal structure marksand beach marks, which are characteristics of high-cycle, low alternating stress level fatigue.

The cracks on the inner surface of the middle stage heat exchanger shell were found to be many, small andmulti-directional characteristics of thermal fatigue. As the result of crack investigation, it was believed that allcracks in the connecting pipe and the heat exchanger shell were due to high-cycle thermal fatigue.

8-4

180° 270° 90°0° 180°Unit (mm)

n

e

j

a

i

k

g

b

f

d

c

h

a

Flow Flow

Weldment

Fig.3 Cracks in the Connecting Pipe

a ~ d

j , k , n

e ~ i

ΛLongitudinal Crack

ΛCircumferential Crack

Fig.2 CVCS Regenerative Heat Exchanger

Letdown Inlet

Letdown Outlet

Charging Outlet

Charging Inlet

Cra

ck

Elbow

Flow

ConnectingPipe

Refueling WaterStorage Tank

Containment SpraySystem

High Pressure SafetyInjection System

Residual HeatRemoval System

Hx

Hx

PressurizerRelief Tank

Containment Spray

AccumulatorTank

Pressurizer

Steam Generator

To Turbine

From Feed WaterPump

Reactor VesselRecirculationSump

Containment Sump

Containment Vessel

CVCSRegenerative Hx

RCP

CH/SIP

DemineralizerVolume ControlTank

Boric Acid Tank

PrimaryDemineralizedWater Tank

CHP

Chemical and Volume Control System

Fig.1 Outline of Tsuruga Power Station Unit-2 Reactor System

8-5

Fig.4 Fracture Surface of Longitudinal Crack a in the Connecting Pipe

Fig.5 The Results of the Liquid Penetrant Examination inthe Middle Stage Shell

180°

90°

0°27

0°18

0°

154.5574.5

#2 Support Ring

Outlet Nozzle

Welding Line

Inner Surface of Middle Stage Shell

Area of UT indicationsArea of PT indicationsPT indicationArea of UT Impossible

0 100mm

#1 Support Ring

(Bottom)

(Top)

(Top)

⋅ Striation like mark in thefracture surface.

Fracture surface− Fracture surface was flat.− Beach mark on the fracture surface.− Crack was started from multi-crack initiation sites.

Initiation BInitiation A Micro simple

Crack opening length 47mm

Upstream side

Inner surface

Fracture surface

⋅ Grow up to big crack from small crack.⋅ Crack without branch.⋅ Austenitic material without defect.⋅ Cold worked layer on the surface by grinding.

Initiation C

Downstream side

⋅ The initiation site had no defect.⋅ Metallic crystal structure marks

on the fracture surface.

8-6

2. Structure of Regenerative Heat ExchangerThe regenerative heat exchanger consists of 3 horizontal shells and connecting pipes. Fig.6 shows the

structure of the damaged middle stage shell. Each shell has an inner cylinder containing the tube bundle toguide the main flow effectively.

The inner cylinder has six support rings to support the internal structure and to limit the bypass flow in theannulus between the inner cylinder and shell to as small as possible. The inner cylinder containing a heatexchanger tube bundle is inserted into the shell in the fabrication process. It results in a small gap betweeneach support ring and the shell.

The measured gap between a support ring and the shell was approximately 1.5mm (approximately 3mm indiameter), which allowed a large bypass flow in the annulus between the inner cylinder and the shell, i.e. about40 percent of the total flow.

The bypass flow rate was estimated by the heat-balance calculation based on the actual operatingtemperature of inlet and outlet of the regenerative heat exchanger. The temperature of the bypass flow is veryhigh (about 250ºC), but the velocity in the annulus is only 7cm/s. On the other hand, the mean temperature ofthe main flow passing inside the inner cylinder is approximately 170ºC as a result of heat-exchange by the tubebundle.

As the result of reviewing the structure, it was identified that the two coolant flows (i.e. the lowertemperature main flow and the higher temperature bypass flow) could be mixed around the outlet of the innercylinder, and that there could be a potential for relatively high thermal stress due to the temperature difference.Therefore, thermal-hydraulic mock-up tests simulating internal flows in the heat exchanger were conducted.

Fig.6 Structure of the Regenerative Heat Exchanger (the middle stage)

Bypass Flow Bypass Flow Bypass Flow(Narrow Part)

Let DownLine Inlet

InnerCylinder Inlet Main Flow

InnerCylinderOutlet

Let DownLine Outlet

Calculated Value Based onDesign Pressure Drop(Approximately 23%)

0.04m/s 1.3m/s 2.6m/s 0.8m/s 0.4m/s 0.7m/s 2.4m/s

Calculated Value Based onOperating Data(Approximately 40%)

0.07m/s 1.6m/s 2.6m/s 0.6m/s 0.3m/s 0.5m/s 2.4m/s

PassPartition

Charging Line Outlet Let Down Line Inlet

Inner Cylinder

#6 Support Ring Bypass Flow#1 Support Ring

Inner CylinderOutlet

Charging Line Inlet

Inner Cylinder Inlet

Bypass Flow(Narrow Part)

Fixed Support Leg

Main Flow

Sliding Support Leg

Let Down Line Outlet

Thermal Shield Plate#2 Support Ring

8-7

3. Investigation of the Causes3.1 Thermal-Hydraulic Mock-up Tests

Thermal-hydraulic mock-up tests were performed in order to understand the internal flow behavior in theheat exchanger (hereinafter “Hx”). The mock-up test facility, as shown in Fig.7, was fabricated, to representhalf of the middle stage Hx and the connecting pipe between the middle and lower stages. All the dimensionswere same as the actual Hx in order to simulate the internal flow. Table 1 shows the test conditions comparedwith the actual flow conditions. Therefore, it was considered that the Richardson number, which is the ratio ofbuoyancy and inertial force, was a key parameter to simulate the internal flow. The velocity in the mock-uptest facility was determined so that the Richardson number could be the identical between mock-up test andactual Hx.

The shell and the connecting pipe were made of acrylic resin to observe the flow pattern inside the Hx.And the inner cylinder and the support rings were made of stainless steel in consideration of the actual thermalconduction. The acrylic resin for the shell and the connecting pipe was selected, because it was estimated thatthe thermal conduction of acrylic resin is almost the same as the actual condition by the insulation outside ofthe shell and the connecting pipe.

Two types of flow patterns, shown in Fig.8, were observed as a result of the test. Flow patterns wereidentified to depend on the gap at the bottom of the #2-support ring (hereinafter “#2-gap”). The gap isdefined as the distance between the bottom of the #2-support ring and inner surface of the shell. Flow pattern1 was observed in the case when the #2-gap was relatively small. In this flow pattern, a cold water regionappeared in the lower part of the shell between the #2 and #3 support rings. This cold region was producedby a cold main flow, i.e., water in this region was stagnant and cooled by thermal conduction to main cold flowinside the inner cylinder via the inner cylinder wall.

In the region between the #1 and #2 support rings, hot bypass water flowed downward from the top of the#2 support ring to the connecting pipe. And cold water flowed downward from the outlet of the inner cylinderto the connecting pipe symmetrically toward the 90º and the 270º sides of the shell. The temperaturedistribution of the connecting pipe is shown in Fig.8. The 270º side was hot and the 0º side was cold.

On the other hand, flow pattern 2 was observed in the case the #2-gap was relatively large. In this flowpattern, the cold water region disappeared in the lower side of the shell between the #2 and #3 support rings,because the hot bypass flow blew out the cold water produced by the cold main flow through the innercylinder. In the region between the #1 and #2 support rings, hot bypass water flowed asymmetrically. Hotbypass water flowed on the 90º side and all cold water from the outlet of the inner cylinder flowed on the 270ºside of the shell. Therefore, hot water flowed from the 0º side of the connecting pipe so that hot water wasrotated towards downstream in the connecting pipe, as shown in Fig.8. As figures viewing some cross-sections of the connecting pipe shown in fig.8, hot water occupied in the 0º side at the inlet, in the 90º side atthe outlet of the elbow. It was also obtained from the mock-up tests showing a transition region between thetwo flow patterns. Flow pattern 1 occurred if the #2-gap was less than 1.5mm and flow pattern 2 occurred ifthe #2-gap was larger than 1.7mm, so that the transition region was between 1.5mm and 1.7mm of the #2-gap.

Temperature fluctuations were recorded. The largest fluctuation at the shell and the connecting pipe was±23ºC and ±14ºC respectively, as shown in Fig.9. The period of the fluctuations was almost several seconds.This data was used in a stress analysis.

8-8

2iu

D/gR

ρρ∆=

Table 1 Comparison of Flow Conditions Between theMock-up and the Actual Hx

Flow Rate 6.8m3/hMain Flow Temperature 30ºCFlow Rate 4.6m3/hBypass Flow Temperature 70ºC

NondimensionalDensityDifference(∆ρ/ρ)

0.018

RepresentativeSize*(D) 0.08m

RepresentativeVelocity*(u) 0.028m/s

Ri 18

Flow Rate 17.1m3/hMain Flow Temperature 170ºCFlow Rate 11.4m3/hBypass Flow Temperature 250ºC

NondimensionalDensityDifference(∆ρ/ρ)

0.112

RepresentativeSize*(D) 0.08m

RepresentativeVelocity*(u) 0.070m/s

Ri 18

Actual Hx Mock-up

RichardsonNumber

* Representative size and velocity correspond to the annulus gap and the velocity of thebypass flow in the annulus between the inner cylinder and the shell.

Location ofThermocouples

#2 Support Ring

#3 Support Ring

#1 Support Ring

Location ofThermocouples

Middle Stage HeatExchanger

Fig.7 Mock-up Test Facility

Lower Stage HeatExchanger

Fig.8 Flow Patterns in the Regenerative Heat Exchanger

#3 #2 #1 180º

270º

90º

0º180º

90º

90º

90º90º

180º

180º180º

0º

0º

0º0º

270º

270º270º

Head

Hot Water

Cold Water

Offsetcondition

2.5mm(180º)

1.5mm(90º)

0.5mm(0º)

1.5mm(270º)

Cold Water Region

LowerOffset

Flow Pattern 2

#3 #2 #1 180º

270º

90º

0º180º

90º

90º

90º90º

180º

180º180º

0º

0º

0º0º

270º

270º270º

Head

Hot Water

Cold Water

OffsetCondition

0.5mm(180º)

1.5mm(90º)

2.5mm(0º)

1.5mm(270º)

Cold Water RegionVanishes

UpperOffset

270º270º

Flow Pattern 1

0º90º

180º270º

0º

90º180º

270º

0º90º

180º270º

0º

90º180º

270º

Fig.9 Temperature Fluctuation Data

(a) Shell

∆T±23ºC

150

200

250

0:00 0:01 0:02 0:03 0:04 0:05Time ( hour : min. )

Tem

pera

ture

(ºC

)

(b) Elbow

150

200

250

0:00 0:01 0:02 0:03 0:04 0:05

Time ( hour : min.)

Tem

pera

ture

(ºC

)

∆T±14ºC

8-9

3.2 Simulation of Flow Pattern Change MechanismThe flow pattern change mechanism was studied based on the results of the mock-up tests. Fig.10 shows

the flow pattern change mechanism. The gaps at the bottom of each support ring are relatively small becauseof gravity. From the investigation of the actual Hx, the #2-gap was 0.6mm and the #4 support ring touchedthe inner surface of the shell. Therefore, in an initial condition, the bypass flow could be stagnant in the lowerportion between each support ring without a region between the #1 and #2 support rings. This conditionproduces cold water in the lower portion of the shell as shown in Fig.10 (STEP 1). Deformation of the shelloccurs due to the temperature difference between the top and the bottom of the shell (STEP 2). Each bottomgap increases by this deformation so that cold water produced in the lower portion of the shell starts to beswept by bypass flow through the bottom gap (STEP 3). In the next step (STEP 4), as the top and the bottomtemperature of the shell become uniform, each gap starts to decrease. Then, deformation of the shell starts tovanish because of this temperature difference decreasing (STEP 5). As the bottom gap decreases, cold waterregion starts to be produced again, and then condition inside of the Hx returns to STEP 2.

A simulation was conducted in order to evaluate this flow pattern change mechanism and to obtain itsfrequency and the amount of gap change. Fig.11 shows the finite element model used for the simulation. Ineach section between each support ring, the thermal-hydraulic conditions were input.

Some thermal-hydraulic conditions were obtained by the mock-up tests. There were two importantparameters. One was the change rate of temperature differences between the top and the bottom of the shell.The experimental data is shown in Fig.12.

As the #2 gap decreases (i.e. the #2 support ring moves downwards), the flow of hot water through thelower gap decreases and the cold water region is produced. It results in increasing the temperature differencesbetween the top and the bottom of the shell. The change rate of temperature differences becomes positive.

On the other hand, the rate becomes negative when the #2 gap increases. The important point was thatone #2 gap had two values of the rate.

Another important parameter was the “lag time” of the changing temperature of the bottom of the shellafter finishing the support ring position change. Fig.13 shows experimental data on the lag time τ. Afterlifting the support ring, the temperature increased in lag time τ after the end of lifting support ring (Fig.13 (a)).And also, the temperature decreased in lag time τ after the end of lowering support ring (Fig.13 (b)).

Fig.14 shows the results of the computer simulation. According to the result, the #2-gap changed betweenabout 1.4mm and 1.8mm and this range was wider than the transition region which says two flow patternscould occur repeatedly. The period was about 500 seconds.

8-10

Hot

Cold Cold Cold Cold

Hot Hot Hot

Cold Cold Cold Cold

Cold

Cold Cold Cold

HotHot Hot Hot

Hot Hot Hot Hot

Cold

Cold

Cold

Cold Cold Cold Cold

Hot Hot Hot

Hot Hot Hot Hot

#1Support

Ring #2 #3 #4 #5 #6

Gap Increasing

UniformTemperature

Gap DecreasingDeformation of theshell almost vanishes

Sliding SupportLeg

Gap Decreasing Fixed Support Leg

STEP1 (Initial state)⋅ Cold water region grow up in the lower

position between each support ring

STEP2

STEP3

STEP4

STEP5

Repeating STEP2 to STEP5

ThermalShield Plate

Fig.10 Flow Pattern Change Mechanism

#2 SupportRing

Sliding Support Leg(Rotation Free)

Center of theModel

Fixed Support Leg(Rotation Free)

Fig.11 Finite Element Model (Half Model)

#1 SupportRing

#3 SupportRing

#4 SupportRing

#5 SupportRing

#6 SupportRing

ThermalShield Plate

Inner CylinderShell

Fig.12 Experimental Data on the Change Rate of Temperature Difference between Topand Bottom of Hx. Shell

68

67

66

65

64

63

62

61

60

59

58

68

67

66

65

64

63

62

61

60

59

58

0 100 200 300 400 0 100 200 300 400

Tem

pera

ture

(º C

)

Tem

pera

ture

(ºC

)

Time(sec) Time(sec)

180º

0º

180º

0º

61.6ºC

59.5ºC

0.15ºC/s(Rate convertedto the actual condition)Start of lifting

support ring

End of liftingτ

180º

0º

60.4ºC

65.5ºC0.24ºC/s(Rate convertedto the actual condition)τ

Start of loweringsupport

End of lowering

(a) Gap Increasing from 0.5mm to 1.5mm (b) Gap Decreasing from 2.5mm to 0.5mm

Fig.13 Experimental Data on Lag Time τ

0.4

0.3

0.2

0.1

0

-0.1

-0.2

-0.3

-0.40 0.5 1 1.5 2 2.5 3

0.5 1 1.5 2 2.5 3

+σσσσ

-σσσσ

+σσσσ

-σσσσ

: Decreasing Gap : Increasing Gap

d(∆T)dt

(ºC/sec)

Lower Gap of the #2 Support Ring(mm)

8-11

2.0

1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

#2

#3

#4

0 500 1000 1500 2000 2500 3000

Low

er G

ap (

mm

)

Time (sec)

Flow Pattern 2

Transition RegionFlow Pattern 1

Bypass Flow:40%

3.3 Evaluation of Crack Initiation and PropagationCrack initiation and propagation analyses were conducted to evaluate the fatigue life due to thermal stress

produced by the flow pattern change. This was based on the results of the thermal-hydraulic mock-up tests,the simulation of flow pattern change mechanism and the metallurgical investigation.

The thermal stress for the crack propagation analysis was calculated by superposition of the lower frequenttemperature change due to flow pattern change and the higher frequency temperature fluctuation due to themixture of the bypass flow and the main flow.

The high average stresses due to operating pressure, thermal transient and residue stress by welding and soon were applied to the cracked pipe and shell.

The stress analysis and the experimental method estimated the average stresses.The fatigue life decreased in consideration of the effect of average stress by the hardness and the surface

finish investigation at the cracked pipe and shell. The average stresses were compared with the thermal stressdue to the flow pattern change. It was evaluated that stresses due to the above-mentioned flow pattern changemechanism were high enough to cause fatigue initiation.

Results of the evaluation of crack initiation and propagation, as shown in Table 2, revealed that both cyclesof thermal stress were in the order of 105.

It was determined from the flow pattern change mechanism simulation (approximately 500 seconds) thatthe cycles calculated by using the plant operating time of Tsuruga-2 (approximately 95,000 hours) were on theorder of 105.

The calculation results corresponded to the cycles obtained by the crack initiation and propagationanalyses.

It was assumed that the cause of the cracked pipe and shell was the high-cycle thermal fatigue, because theresults of thermal-hydraulic mock-up tests and the stress analysis corresponded to the fracture surface ofcracks and the plant operating history.

Fig.14 Results of the Simulation of Flow Pattern Change Mechanism

8-12

4. ConclusionAs a result of the investigation, the cause of the leakage from the connecting pipe was considered to be as

follows;(1) Flow out of the lower temperature bypass flow region occurred repeatedly at the lower part of the shell,

which yielded a cyclic deformation of the shell due to thermal expansion and shrinkage.(2) This cyclic deformation caused a cyclic change of the gap between the inner cylinder support ring and shell,

and consequently the cyclic change of the flow pattern at the region where the bypass flow and main flowmixed.

(3) Superposition of lower frequent temperature change due to the change of flow pattern and higher frequenttemperature fluctuation due to the mixture of the bypass flow and main flow caused high-cycle thermalfatigue cracking.

5. Reference(1) T.Hoshino, T,Aoki, T.Ueno and Y.Kutomi, “Leakage from CVCS Pipe of Regenerative Heat Exchanger Induced by High-Cycle Thermal Fatigue at Tsuruga Nuclear Power Station unit-2”, ICON-8615,presented at 8 th International Conference on Nuclear Engineering, Baltimore, MD USA (April 2000)

Table 2 Evaluation Results of Crack Initiation andPropagation

Investigation Results ofFracture Surface

Evaluation Results (×105 cycles)

Cracks Depth ofCracks(mm)(1)

Number ofBeachMarks

Fatigue Lifefor Initiation

Fatigue Lifefor

Propagation

TotalFatigue Life

a 12.4 8 2.0~ 5.0~6.8 7.0~

d 11.6 6 0.3~ 3.1~5.0 3.4~

e 9 - 1.9~ 1.6~2.0 3.5~

Shell 12.0 8 2.4~ 1.7~2.4 4.1~

Note(1) Depth values except for crack a were based on the cracksinvestigated.

9-1

9 OPERATING EXPERIENCE REGARDING THERMALFATIGUE OF UNISOLABLE PIPING CONNECTED TOPWR REACTOR COOLANT SYSTEMS

Paul HirschbergArthur F. Deardorff

Structural Integrity Associates3315 Almaden Expressway, Suite 24

San Jose, CA 95118-1557

John CareyEPRI Project Manager1300 Harris BoulevardCharlotte, NC 28262

9-3

OPERATING EXPERIENCE REGARDING THERMAL FATIGUE OF UNISOLABLE PIPINGCONNECTED TO PWR REACTOR COOLANT SYSTEMS

Paul Hirschberg John CareyArthur F. Deardorff EPRI Project ManagerStructural Integrity Associates 1300 Harris Boulevard3315 Almaden Expressway, Suite 24 Charlotte, NC 28262San Jose, CA 95118-1557

ABSTRACT

The U.S. Nuclear Regulatory Commission issued Bulletin 88-08 in response to thermal fatiguefailures that occurred at three plants in unisolable portions of piping systems attached to the reactorcoolant loop piping. The failure mechanism was thermal stratification caused by valve leakage, andthermal cycling, caused by turbulence penetration. The EPRI Materials Reliability Project (MRP) iscurrently undertaking a project to develop improved guidelines for screening, evaluation, inspection, andmonitoring of thermal fatigue. One of the first steps in developing such guidelines has been to collectoperating experience data from domestic PWR plants, to gain an understanding of the quantity andseverity of thermal fatigue damage that has actually occurred. By reviewing the failure locations,mechanisms, monitoring results, and corrective actions, more effective tools that address the root causesof actual failures can be developed.

The operating experience data collected consists of two groups: major leak events in domestic andforeign plants, and precursors to leakage, such as thermal monitoring results and observed anomalies indomestic plants. Both sets of data will be available to utilities for review in a database format. Theemphasis in this paper will be on the latter group, reviewing the experiences of domestic utilities inimplementing monitoring and inspection activities in response to the Bulletin. Each PWR plant wascontacted to determine what thermal and other monitoring was implemented, the results, whether leaks orcracks were found, modifications that were made, and unusual occurrences observed that were attributedto thermal fatigue. The scope was limited to the Bulletin 88-08 applicable piping systems. This papersummarizes the results of this survey.

INTRODUCTIONNRC Bulletin 88-08 [1] cited two types of thermal fatigue failures:

1) Inleakage events, in which a leaking isolation valve in the high pressure injection systemallowed colder fluid to leak into and stratify in the unisolable portion of the reactor coolantsystem. The upstream pressure of the injection system, which is driven by the charging pumps,is higher than RCS pressure, thus allowing inleakage. Although not fully understood at thetime, cracking was caused by turbulence penetration from the RCS flow into the branch pipe,resulting in thermal cycling between the hot RCS flow and cold leakage flow. The events thatprecipitated the Bulletin occurred at the Farley and Tihange plants.

2) Outleakage events, in which intermittent leakage out of the stem leakoff line of the isolationvalve in the Residual Heat Removal system allowed hot RCS fluid to leak out of the unisolableportion of the RCS. The intermittent leakage caused stratification and thermal cycling in theunisolable section, resulting in a crack. A leak event at the Genkai plant was the source of theconcern.

9-4

The Bulletin required all PWR plants to:

1) Review all systems connected to the reactor coolant loop that are normally stagnant to identifyunisolable sections that are potentially susceptible to cracking from the thermal fatiguemechanisms described in the Bulletin

2) In locations that may have been subjected to high thermal stresses, perform nondestructiveexaminations of the welds, heat affected zones, high stress points, and geometric discontinuitiesto assure that there are no existing flaws

3) Implement a program to assure that thermal fatigue cracking will not occur in these lines byeither: a) monitoring the lines to measure thermal stratification and evaluating the results againstacceptance criteria, b) preventing pressure upstream of isolation valves from exceeding RCSpressure, or c) installing permanent modifications to enable the system to withstand the stresses.

Since the issuance of the Bulletin, other leakage events have occurred, mostly either in foreign plants,or due to scenarios that were different than those anticipated by the Bulletin. Domestic plants respondedto 88-08 in a variety of ways, either instrumenting some or all of the lines with thermocouples, installingpressure monitoring systems, measuring valve leakage, or installing various piping system modificationsto preclude thermal fatigue. Insight has been gained from the results of the temperature monitoringperformed at these plants. This paper summarizes the results of a comprehensive review of the operatingexperience related to the Bulletin 88-08 thermal fatigue issue, both from the leak events worldwide, andthe domestic monitoring experience and anomalies observed in the course of operation.

DATA COLLECTION METHODOLOGY AND SCOPEOver 30 references were reviewed to obtain the details of the leak events that have occurred

worldwide. This included NRC documents, licensee event reports, conference proceedings, NUREGreports, previously published EPRI reports, operating experience reports, and individual plant failureanalyses. Three databases previously compiled by EPRI [2, 3, 4] were searched for relevant items. TheNRC’s Bibliographic Retrieval System was another source of event information.

The primary source of information for individual plant thermal monitoring experience was aquestionnaire that was sent to all domestic PWR plants. The questionnaire solicited information onsystems monitored, degree of stratification observed, NDE examination results, leaks and cracks found,other monitoring, and modifications made to reduce thermal fatigue. Responses to the questionnaire werefollowed up with telephone conversations to obtain additional details. Other information was obtainedfrom ASME technical papers, stress analyses performed by Structural Integrity Associates, operatingexperience reports, plant discrepancy reports, and conversations with cognizant plant engineers.

Other thermal fatigue and thermal stratification events have occurred which are beyond the scope ofthis review. Stratification in the Pressurizer Surge line and Steam Generator Feedwater nozzles aremanaged by other means and were not included in the scope of the MRP program. Some foreign plantleak events occurred in systems connected to the RCS but were in isolable locations, and were also notincluded.

RESULTS OF REVIEW - GENERALWorldwide, there have been fourteen leak events recorded in unisolable portions of stagnant systems

connected to the Reactor Coolant system [5, 6] in PWR plants. Only seven of these were caused by themechanisms described in Bulletin 88-08. Of these, after the original Farley crack that precipitated theBulletin, all occurred in foreign plants. From this standpoint, the actions taken by domestic plants inresponse to the Bulletin were effective in preventing additional failures.

The remaining seven leaks were due to other scenarios not anticipated in the Bulletin but arenevertheless related to the original issue. Four of these pertained to design features that are somewhatunique, and the remaining three had essentially a common root cause. These will be discussed belowunder the relevant systems.

9-5

Domestic plants’ actions in response to Bulletin 88-08 varied. Two-thirds of the plants instrumentedone or more lines to measure thermal stratification. In most cases, one or two operating cycles weremonitored, and if minimal stratification was observed, monitoring was discontinued. One-fourth of theplants performed valve leakage monitoring, either by direct leakage measurement at a test connection, orby applying pressure in a closed section and measuring pressure reduction, or by fluid inventorybalancing. In most cases, such monitoring is continuing. Only one plant monitors high pressure injectionsystem pressure upstream of the isolation valve to assure it does not exceed RCS pressure. However,most CE plants, which do not have safety injection systems that operate above RCS pressure, monitorpressure to prevent RCS pressure from entering lower pressure systems due to backleakage through thecheck valves. About 40% of the plants took some action to mitigate thermal fatigue, such as reroutingpiping, adding isolation valves, relocating valves, or improving valve maintenance to prevent leaks.

One of the requirements of the Bulletin was for the plants to review their systems and determinewhich were susceptible to the thermal fatigue issue. The following were the systems considered by theplants to be the most susceptible and /or required temperature monitoring:

1. Auxiliary Spray2. High Pressure Safety Injection3. Alternate Charging4. Charging / Makeup5. Intermediate Head (Hot Leg) Injection6. RHR / Shutdown Cooling / Decay Heat Suction7. Main Pressurizer Spray8. Low Pressure Injection9. Drain Lines / Excess Letdown (after 1995)

RESULTS OF REVIEW BY SYSTEM

High Pressure Safety InjectionThis is the system in which the original 88-08 failures occurred, at the Farley and Tihange plants.

The system is driven by the charging (or HPI) pumps at pressures higher than RCS pressure. Theinjection nozzles are smaller diameter than the lower pressure injection nozzles, as low as 1½” on theWestinghouse four loop plants. The lines enter the RCS either from above or from the side.

At Farley, the crack occurred in the heat affected zone of the weld between the first elbow and thehorizontal pipe, about 3 feet from the RCS cold leg nozzle, on a 6 inch NPS line. The crack was causedby hot RCS fluid from turbulence penetration interacting at the bottom of the pipe with cold valve leakagefluid that had stratified. The failure was caused by high cycle fatigue, however the current understandingof turbulence penetration is not sufficient to be able to predict the cycling frequency or the depth ofpenetration with any certainty. The general rule of thumb is that turbulence penetration effects extendbetween 5 and 25 pipe diameters into the branch pipe.

Four other leak events occurred that were of a similar nature. At Tihange, the crack was locatedabout 2 feet from the RCS hot leg nozzle, in the elbow base metal. At Obrigheim, the crack was locatedeven closer to the RCS, in the nozzle to elbow weld. It had as a contributing cause a deep notch in thecircumferential weld that originated in the fabrication process. At Dampierre 2, the crack was in the weldbetween the check valve and straight pipe just upstream of the hot leg nozzle. At Dampierre 1, the leakoccurred in the base metal of the horizontal run between the hot leg and the check valve, about 2 feetfrom the nozzle. After the failed pipe was replaced, another crack occurred in the same location only ninemonths later. The replaced section was found to have high residual stresses, and the isolation valve hadnot stopped leaking.

9-6

A somewhat different failure was found at Biblis, where the crack occurred at a tee that connects ahot and a cold injection line. There were contributing causes of a bound up snubber placing high tensilestress on the pipe, and the presence of pump mechanical vibrations. This system design is not found indomestic plants.

It should be noted that none of the safety injection cracks occurred in 1½” nozzles as used in theWestinghouse four loop design. Also, Combustion Engineering plants can be considered as notsusceptible to inleakage in the safety injection system because the system is not driven by the higherpressure charging pumps; the highest pressure in the system is about 1500 psi, from the SI pumps.Babcock and Wilcox plants have thermal sleeves in the high pressure injection nozzles, protecting most ofthe region subject to thermal fatigue.

The results of temperature monitoring in U.S. plants for this system indicated very little thermalstratification during normal operation unless the isolation valve was leaking. The entire pipe from theloop nozzle to the isolation valve generally remained hot. There was only one case where significantstratification was reported: this was due to valve leakage and resulted in a top to bottom temperaturedifference of 215°F, with cycling having a 2-20 minute period. The valve leakage was corrected and thestratification disappeared.

Although there was little thermal stratification during normal operation, several plants reportedstratification during plant heatup for a different reason. During periods when not all reactor coolantpumps are running, a loop with the pump turned off experiences a rise in static pressure. These plantsreported that the associated check valve did not seat tightly and RCS flow leaked backwards past thecheck valve and into the other loops. This resulted in stratification gradients of up to 170°F. Thisstratification does not pose a concern because it only occurs for a short time and has only a limitednumber of thermal cycles.

Only one plant installed some form of pressure monitoring on this system. Pressure is checked by aweekly surveillance and is relieved when system pressure exceeds RCS pressure. Two plants installed asecond isolation valve to prevent inleakage.

Intermediate Pressure Safety Injection (From SI Pumps)This system is fed by the safety injection pumps, which operate at about 1500 psi, and usually injects

through six inch nozzles in the hot legs. Babcock and Wilcox plants do not have this system, andWestinghouse three and two loop plants use the high pressure injection nozzles for both functions.Combustion Engineering plants use the same nozzle for intermediate head injection, accumulatorinjection, and low head (shutdown cooling) injection.

A number of plants measured a moderate amount of stratification on these lines but not due to valveleakage. Most of the lines connect to the reactor coolant loop from above, followed by a horizontal piperun. Top to bottom temperature gradients of up to 100°F were reported in this horizontal section, due tonatural convection from the RCS fluid rising to the top of the pipe, and heat loss to the ambient coolingthe fluid at the bottom. The stratification appeared to be constant with no cycling, and therefore is not aconcern for a thermal fatigue failure.

There were incidents of check valve backleakage, which caused the upstream piping to pressurize toRCS pressure. Most of the CE plants have installed pressure control systems that relieve to the reactorcoolant drain tank if backleakage causes the pressure to exceed SI pump pressure. No failures haveoccurred in these lines.

Low Pressure Safety Injection (RHR and Accumulators)The low pressure injection nozzles usually perform the combined function of accumulator injection

(at about 600 psi) and RHR / Shutdown Cooling injection (at about 350 psi). On B&W plants, DecayHeat Removal / Core Flood injection is done directly into the reactor pressure vessel. The low pressurenozzles are the largest injection nozzles, between 10 and 14 inches in diameter.

9-7

No leaks have occurred in these lines. Of the plants that monitored these lines, two reportedobserving thermal stratification due to natural convection, with no valve leakage and no cycling. Theselines tend to have longer horizontal runs with the isolation valve located a longer distance from the loopnozzle. One plant had a check valve leak backwards during heatup, causing stratification in the otherlines, when reactor coolant pumps were being cycled. In the B&W plants, only a small amount ofstratification occurred in the lines connected to the reactor vessel, with no cycling, as flow in the vesselpast the nozzle is insufficient to cause turbulence penetration. None of the above are concerns for thermalfatigue, as there is insufficient cycling to cause a failure.

RHR Suction / Shutdown Cooling / Decay Heat DropThe RHR Suction line typically drops vertically from the RCS Hot Leg, then travels horizontally for

some distance to the isolation valve. Bulletin 88-08 Supplement 3 was issued because of a leak at theGenkai plant. The internals of the isolation valve alternately shrunk and expanded causing intermittentoutleakage of RCS fluid through the stem packing and leakoff line. The crack, which occurred at theweld between the first elbow downstream of the hot leg nozzle and the horizontal run, was concluded tohave been caused by the intermittent leakage periodically introducing hot fluid at the elbow. Thehorizontal run loses heat to the ambient, thus causing the outleakage to stratify. Had the leakage not beenintermittent, there would not have been enough thermal cycles to cause a failure according to the rootcause described in the bulletin.

The elbow was 9 pipe diameters from the loop nozzle, well within the range of turbulence penetrationfrom the hot leg flow. In hindsight, it is possible that the cracking was assisted by turbulence penetrationcausing periodic incursions of hot leg fluid into the horizontal portion of the system, which resulted in thehorizontal run periodically stratifying, and the elbow being submitted to cyclic thermal shocks. Most ofthe responses to the 88-08 Bulletin focused on the valve leakage issue, particularly on stem packingleakage. A number of plants implemented improved valve maintenance programs to reduce packingleaks. Although these programs were effective in stopping packing leaks, they may not have addressedthe real root cause of thermal stratification and thermal cycling in the RHR system.

The results of thermal monitoring indicated that numerous plants measured thermal stratification inthe RHR suction line without any valve leakage occurring. Top to bottom gradients of up to 350°F weremeasured at some plants. A more common temperature gradient was 120°F, with the upper temperaturecycling +/- 40°, 5-15 cycles per day. The cause of this stratification is turbulence penetration of the hotleg fluid extending into the horizontal pipe run, which stratifies due to natural convection. In order tostratify, the length of the vertical run has to be short enough for the hot fluid to reach the horizontal run,but not so short that the horizontal run is always hot, if the isolation valve is located close to the elbow. Anumber of plants have long horizontal runs, which increases the propensity for stratification due to heatlosses. Some plants noted stratification only during heatup or cooldown, probably because changes inRCS flow during those periods changed the depth of turbulence penetration. One plant noted that duringpower reductions, the greater the reduction in power, the longer was the turbulence penetration distance.Another plant noted that the turbulence penetration was not cyclic but manifested itself in a randomhelical pattern.

Charging / Alternate Charging / MakeupTwo leak events occurred in plants having the Babcock and Wilcox Makeup system nozzle design.

At Crystal River and Oconee 2, the thermal sleeve became loose and allowed turbulent mixing of hotRCS fluid with cold makeup fluid behind the sleeve. In the B&W design, makeup is not heated by aregenerative heat exchanger and thus the temperature difference between makeup and RCS flow is large.The thermal sleeve was installed with a press fit, and plastic deformation due to hot and cold transientscaused it to loosen. The B&W plants have installed an improved thermal sleeve design and implementeda periodic inspection program, such that the problem is considered resolved.

9-8

Westinghouse plants typically run either charging or alternate charging, but not both. The concern isthat the line that is not in operation could experience inleakage from the charging pump discharge andhave the same mechanism for thermal cycling at the high pressure safety injection lines. Of the B&Wplants, one uses two nozzles for normal makeup while the others use one. CE plants use both chargingpaths, so this issue is not a concern for those plants.

The results of thermal monitoring on these lines indicated very little thermal stratification. Eitherthere was no valve leakage, or the distance from the isolation valve to the loop nozzle was short enoughthat the stagnant fluid did not cool significantly between the regenerative heat exchanger outlet and theloop nozzle.

The B&W plant that uses two makeup nozzles reported stratification of up to 325°F during heatup,due to backflow caused by pressure differences in the two lines when reactor coolant pumps are notrunning.

Pressurizer Spray (Main and Auxiliary)The auxiliary spray system in Westinghouse and CE plants draws from the charging system and is

stagnant during normal operation. If the isolation valve leaks, charging system pressure can produce flowtoward the tee with the main spray system. The fluid in the auxiliary spray system is typically coldbecause the piping length is long enough to cool to ambient. The main spray comes from the reactorcoolant loop cold legs, thus there is the potential for a significant temperature difference between themain and auxiliary spray near the tee, and thermal stratification cycling could result if there is unsteadymixing of the flows from the main spray and the leakage.

The results of temperature monitoring indicated that thermal stratification was minimal during normaloperation, unless the isolation valve was leaking. Two plants reported top to bottom temperaturegradients of 200°F caused by valve leakage. When the leakage was corrected, the stratificationdisappeared. Testing and analysis by one plant indicated that the flow velocity of the main spray line wasinsufficient to cause turbulence penetration to enter the auxiliary spray line. Therefore, although it ispossible to have some stratification in the auxiliary spray line from natural convection during normaloperation, it is not a concern due to the lack of a cycling mechanism.

During heatup and cooldown, several plants noted significant stratification occurring in these lines.When none or an insufficient number of reactor coolant pumps are running to generate sufficient mainspray flow, hot steam from the pressurizer enters the lines. This occurs despite the presence of the “gooseneck” loop seal in some of the pressurizer designs. When auxiliary spray was used, a top to bottomtemperature gradient of up to 260°F was measured. In addition, auxiliary spray tended to be cycled onand off repeatedly especially during heatups, which results not only in thermal stratification cycles butalso thermal shocks at the pressurizer nozzle and auxiliary to main spray tee.

One plant noted significant stratification during letdown isolations. Apparently there was someisolation valve leakage of charging flow, which was significantly cooler due to the lack of letdown flowin the regenerative heat exchanger.

One leak event occurred in Finland at the Loviisa plant. The crack was in a Z-type isolation valve inthe auxiliary spray line, vertically above the main spray tee. During heatup and cooldown, pressurizersteam entered the valve outlet, and auxiliary spray entered the inlet; the hot outlet was at a lower elevationthan the cold inlet, and stratification cycling occurred internally in the valve. There had also been apreexisting material inclusion defect in the valve that facilitated cracking.

Reactor Coolant Loop Drains / Excess LetdownThe reactor coolant loop drain lines and the excess letdown line were not recognized as being

susceptible to developing unisolable leaks due to thermal stratification cycling in any of the plant reviewsin response to Bulletin 88-08. However, four leak events have occurred in these lines. At Three MileIsland, a leak occurred in a cold leg drain line, in the weld between the first elbow downstream of the loopnozzle and the horizontal pipe run. At Oconee, a leak happened in almost the exact same location, thistime near the center of the elbow extrados. The TMI drain line was 1½” NPS, with an increase to 2” in

9-9

the horizontal run; the distance from the inside surface of the reactor coolant loop to the crack was 14”.Similarly, at Oconee the drain line was 1½” and the distance to the crack was 13”. The crack was causedby turbulence penetration from the RCS intermittently extending into the horizontal run. In both cases thepipe was not insulated, which made it easier for the horizontal portion to stratify. The vertical pipe lengthwas just right for the hot fluid to periodically extend into the horizontal run, but not so short as to keep thehorizontal pipe warm all the time.

At Mihama, the same mechanism caused a leak in the excess letdown line. The line was 2” NPS andthe crack was located 15” from the reactor coolant loop inside surface. One difference was that theMihama line was insulated; in this case, however, the length of the horizontal run to the isolation valvewas very long, resulting in an equivalent amount of heat loss to ambient as a shorter, uninsulated line.There is also evidence that there was an unsteady stratification layer in the elbow. The Mihama elbowwas also found to have had high tensile residual stress induced in the fabrication process. The correctiveaction for this event was to shorten the vertical run such that the turbulence penetration boundary waswell into the horizontal run, away from the elbow, which is a point of stress concentration.

The other leak event occurred at Loviisa, in a cross tie line between a hot leg drain and a cold legdrain. The crack was in a weld between a reducer and the tee that joins the two drains. Leakage past thecross tie valve allowed hot leg fluid to flow into the cold leg drain. Intermittent thermal expansion of thevalve internals caused thermal cycling at the tee. This system design is not found in domestic plants.

Since none of the domestic plants initially identified the RCS drains as a potentially susceptiblesystem, none of them performed any temperature monitoring. However, one plant had a leak occur in theexcess letdown line in a similar location as in Mihama. In this case the crack was in the heat affectedzone of the weld between the first off elbow and the horizontal run, 13” from the reactor coolant loop.The utility concluded that the cause was a thermal interference between a flange on the line and a floorsupport plate, which generated a stress cycle each time the reactor coolant loop heated up or cooled down.At another plant, a crack was found in a hot leg nuclear sampling system isolation valve. The crack wascaused by admitting 600°F samples on a daily basis into a normally stagnant line.

CONCLUSIONSThermal stratification and cycling is still a concern for fatigue cracking in lines connected to the

reactor coolant loop. However, the susceptible systems can probably be limited to high pressure safetyinjection, RHR suction, and reactor coolant loop drain lines, or similar lines. More work needs to be doneto improve the understanding of turbulence penetration effects, particularly how to determine thepenetration distance and the rate of cycling.

The details of the worldwide leak events and the domestic plant monitoring experience are beingmade available in database format on EPRI WEB. The database will be kept up to date by adding newoperating experience as submitted by users.

REFERENCES

1. “Thermal Stresses in Piping Connected to Reactor Coolant Systems”, Bulletin 88-08 andSupplements 1, 2, and 3, U.S. NRC, 1988 – 1989.

2. EPRI Fatigue Management Handbook, TR-104534 Vol. 2, EPRI Palo Alto, December 1994.3. “Nuclear Reactor Piping Failures at U.S. Commercial LWRs: 1961 – 1997”, Report TR-110102,

EPRI Palo Alto, December 1998.4. “Piping System Failure Rates and Rupture Frequencies for Use in Risk Informed Inservice Inspection

Applications”, Report TR-111880, EPRI Palo Alto, March 1999.5. “Experience with Thermal Fatigue in LWR Piping Caused by Mixing and Stratification”, Specialists

Meeting Proceedings, OECD Nuclear Energy Agency, Paris, June 1998.6. “Assessment of Pressurized Water Reactor Primary System Leaks”, NUREG/CR-6582, INEEL,

December 1998.

OPERATING EXPERIENCE:THERMAL FATIGUE OF UNISOLABLE

PIPING CONNECTED TO PWRREACTOR COOLANT SYSTEMS

Paul Hirschberg, Art Deardorff


John Carey

EPRI

July 2000

9-10

NRC Bulletin 88-08

• Inleakage past a closed isolation valve betweenthe charging system and the safety injectionnozzle caused an unisolable leak at Farley andTihange

• Outleakage through the RHR suction isolationvalve leakoff line caused cracking in the firstelbow off the reactor coolant loop (Genkai)

• Scope of Bulletin is the ASME Class 1 piping thatis attached to the Reactor Coolant Loop and isnot isolable from the loop, and is usuallystagnant and subject to cyclic thermalstratification

9-11

NRC Bulletin 88-08 - Actions

• PWR plants identify systems susceptible tothermal stratification / cycling mechanisms

• Perform ISI examinations of welds in susceptiblelocations

• Assure cracking will not occur by either:♦ Instrumenting lines to monitor for thermal stratification

♦ Monitor pressure upstream of SI isolation valve

♦ Valve leakage monitoring

♦ Hardware modifications

9-12

Other Thermal Stratification IssuesNot in Scope

• Pressurizer Surge Line stratification

• Steam Generator Feedwater nozzle stratification

9-13

EPRI MRP Industry OperatingExperience Review

• Collect details of Bulletin 88-08 related leakevents that occurred worldwide

• Survey domestic utilities to obtain results ofmonitoring programs, identify conditions thatcould be precursors to thermal fatigue cracking,and capture non-reportable events andobservations

• Present information in a database format, to bemaintained on EPRI WEB

9-14

Sources of Operating ExperienceData

• Questionnaire sent to all domestic PWR plants,responses followed up by telephone

• Three databases previously compiled by EPRI(Risk informed ISI, SKI, Fatigue ManagementHandbook)

• NRC Bibliographic Retrieval System (LERs)

• Operating experience reports

• Published technical papers

• Evaluations by utility and consultants

• NUREGs, conference proceedings

9-15

Results of Review - General

• 14 leak events have occurred worldwide:♦ Only 7 were due to scenarios cited in Bulletin 88-08

♦ After Farley, none of these occurred in domestic plants

♦ Of the remaining 7 events:° 4 were in domestic plants

° 2 were specific to B&W plants

° 2 in designs unique to European plants

° 3 had a common root cause

9-16

General Results - Plant Responsesto Bulletin 88-08

• 67% instrumented one or more systems tomeasure thermal stratification♦ Most monitored for only one or two cycles

• 27% perform some form of valve leakagemonitoring

• 12% have means to monitor and control systempressures

• 40% took some action to prevent thermal fatiguefailures: modified piping geometry, improvedvalve maintenance, added or changed valves

9-17

General Results - Systems Considered byPlants to be Susceptible to Thermal Fatigue

1. Auxiliary Spray

2. Safety Injection from charging pumps

3. Alternate Charging

4. Charging / Makeup

5. Safety Injection from SI pumps

6. RHR Suction / Shutdown cooling / DHR drop

7. Main Pressurizer Spray

8. Low Pressure Injection

9. RCS Drain lines / Excess letdown (after 1995)

9-18

Safety Injection from Charging Pumps Leak Events

• Farley (1987) - HAZ of elbow to pipe weld

• Obrigheim (1986) - nozzle to elbow weld

• Tihange (1988) - elbow base metal

• Dampierre 2 (1992) - valve to pipe weld

• Dampierre 1 (1996) - straight pipe base metal(cracked again in 9 months)

• Biblis (1995) - tee connecting hot and coldinjection lines

• None in Westinghouse 1 1/2" nozzles

9-19

Farley and Tihange

♦ In-leakage from higher pressures in charging system♦ Interactions between cold in-leakage and hot Reactor Coolant System fluid caused

thermal cycling

9-20

Safety Injection from Charging PumpsMonitoring Experience

• Monitoring indicated that stratification wasinsignificant under normal conditions unless avalve was leaking

• Stratification occurred during heatup; severalplants reported backflow through the check valvewhen the associated RCP was off and others wererunning. Cross flow to other loops causedstratification, max. 170°°°° F

• CE plants are not subject to inleakage

• Only one plant installed a pressure control systemto prevent upstream pressure from exceeding RCSpressure

9-21

Safety Injection from SI Pumps andRHR / Accumulator Injection

• No leak events

• Several plants reported a moderate amount ofstratification♦ Most cases caused by natural convection; no cycling

♦ Maximum ∆∆∆∆T = 200°°°° F in long straight run, no leakage

♦ One case was due to a leaking check valve whichallowed outleakage during heatups

♦ CE plants have pressure relief systems to preventoverpressurization from check valve backleakage

♦ B&W plant lines injecting into vessel have negligiblestratification with no turbulence penetration

9-22

RHR Suction / Shutdown Cooling /Decay Heat Removal

• Only leak event was at Genkai♦ Attributed to intermittent outleakage of RCS hot leg fluid

through isolation valve stem packing

♦ May have also been due to turbulence penetration of hotleg fluid into horizontal run

9-23

Genkai

♦ Out-leakage occurred at isolation valve leak-off line♦ Cyclic stratification occurred in horizontal piping

9-24

Hypothetical Turbulence PenetrationCycling

9-25

RHR Suction / Shutdown Cooling /Decay Heat Removal

• Many plants reported significant thermalstratification, up to 350°°°° F♦ Most cases were due to turbulence penetration and not

the Genkai leak-off line mechanism

♦ One case due to a packing leak in the isolation valve

♦ Some reported cycling but the period was long (0.2 - 8days)

♦ Some cases were significant only during powerreduction; the greater the power reduction, the longerthe turbulence penetration distance

9-26

Charging / Makeup / Alt. Charging

• Leak Events:♦ Crystal River (1983) and Oconee 2 (1997)

♦ In Makeup / HPI system, B&W plants only

♦ Caused by leakage past thermal sleeve allowingturbulent mixing of cold makeup flow with RCS flow

♦ Managed by a B&W Owners Group program

9-27

Charging / Makeup / Alt. Charging

• Monitoring Experience:♦ No stratification in charging / makeup except when both

makeup trains in use, and crossflow occurs duringheatups due to pressure differences from pump cycling

♦ Alternate charging had insignificant stratification

♦ Some plants use both charging and alternate charging,or equalize their use

9-28

Pressurizer Spray / Auxiliary Spray

• Leak event at Loviisa♦ Crack in auxiliary spray isolation valve due to internal

thermal cycling of pressurizer steam and cold auxiliaryspray

♦ Pre-existing material defect

♦ Valve design not used in U.S.

9-29

Pressurizer Spray / Auxiliary SprayMonitoring Experience

• Stratification is not significant during normaloperation unless isolation valve leaks

• No turbulence penetration from main spray intoaux. spray

• During heatups and cooldowns, significantstratification (260°°°° F) occurs when RCPs are offand main spray flow is insufficient to fill the pipe.Pressurizer steam refloods the line, and cyclingoccurs with each auxiliary spray injection.

• One plant reported higher auxiliary spraystratification (236°°°° F) during periods whenletdown is isolated; isolation valve leaked by andcharging flow is colder

9-30

Reactor Coolant Loop Drains

• Leak Events:♦ TMI 1 - 1995

♦ Oconee 1 - 2000

• Caused by high cycle fatigue due to turbulencepenetration

• Vertical length was about 8 - 9 diameters

• Horizontal portion of drain line was uninsulated

• Horizontal portion stratifies due to loss of heat toambient and intermittent incursion of reactorcoolant

9-31

TMI Cold Leg Drain

Isometric of drain line showing details of leak area and location of damaged support

9-32

Excess Letdown

• Excess Letdown Leak Event:♦ Mihama - 1997

• Caused by high cycle fatigue due to turbulencepenetration, similar to drain line events

• Vertical length about 8 - 9 diameters

• Horizontal portion of drain line was insulated, butisolation valve was far away (18 feet)

• Modification was to shorten the distance from theRCL to the first elbow, to avoid the turbulencepenetration boundary

9-33

Modification to Avoid TurbulencePenetration

Before Modification After Modification

High TemperatureRegion

Interface at elbow piperesults in high thermal stress

Interface at horizontal piperesults in low thermal stress

Low TemperatureRegion

Low TemperatureRegion

High TemperatureRegion

15.5

in.

15.7

5 in

.

10.2

4 in

.

9-34

RCL Drains / Excess LetdownMonitoring Experience

• One plant had a leak in 1995 on the ExcessLetdown line at the first elbow. It was in a similarlocation to the Mihama event. It was attributed tothermal interference during heatup / cooldown

• On an RCL drain / sampling line, one plantreported finding a crack in 1994 due to thermalfatigue, caused by admitting 600°°°° F samples intoa stagnant line on a daily basis

9-35

Conclusions

• Mechanisms for thermal stratification and cyclinginduced fatigue failures are still present

• Concern primarily in high pressure safetyinjection, RHR Suction, and RCL drains

• Need ongoing monitoring to prevent valveinleakage

• RCL drains and excess letdown lines should beinsulated

9-36

10-1

10 AUXILIARY FEEDWATER LINE STRATIFICATION ANDCOUFAST SIMULATION

Authors :STÉPHAN J.-M., MASSON J.C.

Électricité de France - Industry Branch - Research and Development DivisionMechanics and Components Technology77818 Moret-sur-Loing CEDEX - France

10- 3

AUXILIARY FEEDWATER LINE STRATIFICATION AND COUFAST SIMULATION

Authors :

STÉPHAN J.-M., MASSON J.C.

Électricité de France - Industry Branch - Research and Development Division

Mechanics and Components Technology

77818 Moret-sur-Loing CEDEX - France

ABSTRACT :

EDF conducted in the last years different studies on the stratification problems in pipes. One of them consisted ofsubjecting a metallic full-scale mockup of the steam generator feedwater system to different regimes ofstratification. Very useful data were obtained on thermal and mechanical effects of stratification. Then 4000cycles of fatigue were applied between two stable states of stratification. The 2 levels of the stratificationinterface were chosen to obtain the maximum possible variation of stresses during the cycle. At the end of thetests, destructive examinations revealed small cracks due to fatigue. The usage factors were calculated usingelastic and cyclic elastic-plastic computations. Considering the small depths of the cracks (1.4 to 2 mm) theusage factors (1.3 to 1.9) can be accepted, even if margins are weaker than those obtained in other fatiguestudies of mockups subjected to thermal shocks.

1 - INTRODUCTION

A thermal stratified flow is characterized by 3 superposed horizontal layers : a cold layer at the bottom of thepipe, a hot layer at its top and an interface between these two layers, which shows a vertical temperaturegradient. The conditions required for the creation of a stratified flow are mainly the low velocities of fluids. Theyexist in certain pipings of nuclear power plants especially in the steam generator (SG) feedwater lines, in somedead ends (safety injection, residual heat removal,…), and in some pressurizer surge lines [1, 2, 3].

In the case of a SG feedwater line, the cold flow is generated by water supplied to the steam generator throughthe auxiliary feedwater system during hot shutdowns. The hot flow results from the heating of water in theinternal feedwater ring. A counter-current recirculation can be established in the upper section of the feedwaterline. The incoming cold flow does not always suffice to prevent the hot water from flowing backward [1].

The main disturbances due to thermal stratification on nuclear units have been piping displacements and fatiguecracks on pipings. If large lengths of pipe are stratified, the displacements or the forces applied on supports maybe considerable (TROJAN power plant [2]).

Cracks result from the fatigue damage caused by the variations of the stratified state. In the case of the SG feedline, the variations are due to modifications of the feedwater flow rate. The main cracks mentioned have beennoticed in COOK2 (1978) [1], SEQUOYAH (1992), and DIABLO CANYON (1992).

Studies of thermal stratification have formed the subject of articles for congresses or reviews [4, 5, 6]. The studypresented in this paper concerns the stratification in the SG feedwater system. It is original in that, at the end of aknowledge acquisition phase, a cycling of the mock-up under fatigue conditions has been carried out.

2 - DESCRIPTION OF THE COUFAST MOCK-UP

The COUFAST mock-up is made up of a 90° elbow connected to 2 straight pipes (figure 1). Its dimensions andmaterial are similar to those used in the SG feedwater system of the French PWR 900. It differs principally in thelength of the horizontal section, and in the downward section which is vertical instead of being tilted at 45°.

It has been built according to the rules applicable to the French nuclear constructions [13]. The only exceptionsare the 3 welds (among 5) in the horizontal section. Their designs differ deliberately from the standard rules,

10- 4

either in the taper angle or the radius, or in the shot-peening finish (figure 2). The aim was to study the effects ofgeometry or shot-peening on the fatigue resistance.

The cold water is injected in the lower part of the mock-up through a diffuser to minimize the disturbance in theelbow. The hot water inlet nozzle is set four meters away from the elbow. The outlet nozzles are set at the end ofthe horizontal section at a distance of 6 meters from the elbow. When the stratification takes place in the mock-up, a part of the hot water flows toward the elbow in a counter-current direction. It forms the upper hot layer ofthe stratified flow. This flow reproduces the flow back from the SG ring toward the elbow. As this flowrate is notknown, it was necessary to investigate various hot water inflows, combined to various cold water inflows, tocheck the situations existing in the PWR plants.

Mechanically speaking, the model is supported in 2 points. The mock-up is embedded on a concrete bed-plate onthe horizontal section about 5 meters away from the elbow. A weight-support in the middle of the vertical sectiontakes up partially the weight of the mock-up (figure 1). It was adjusted to give no momentum at the fixed pointwhen the mock-up is filled with cold water. The connections with the water feed loop leave the model free tobend under effect of thermal stratification.

3 - THE INSTRUMENTATION OF THE MOCK-UP

A great number of thermocouples (about 60) are set on the outer wall, particularly on the elbow and its weld, andin front of the 3 thermocouple rods (figure 3). The inner wall is fitted with 3 sets of 6 thermocouples in front of thethermocouple rods n°3 and 4 (fig. 4). One of them emerge in the water but the others are embedded at a depthof 0.5 mm in the inner wall. They are very close to each other, spaced by only 0.5 mm. These 3 sets are used tomeasure the heat exchanged between the fluid and the inner wall in the different layers of the stratified flow.

The vertical profiles of fluid temperature are measured along three vertical diameters (figure 3). Two oppositevertical rods, each bearing 18 thermocouples (1 per cm), are set in each diameter (figure 5).

Fifty strain gauges are set on the outer wall as shown on figure 4. The follow-up of cold thermal shock tests hasmade it necessary to set 25 additional strain gauges and 10 associated thermocouples on the inner wall.

4 - THE PERFORMING OF THE TESTS

The nominal thermal-hydraulic conditions for testing were as follows:

- pressure of 80 bars,

- cold water at 60°C with flow rates ranging from 0.1 to 9 m3/h,

- hot water at 280°C with flow rates ranging from 0.1 to 5 m3/h.

The thermal-hydraulic capacities of the water feed loop have limited the sum of the cold and the hot inflows to 10m3/h. Figure 6 shows the flow diagram of the SUPERBABETTE loop which feeds the model.

Four tests have been carried out at hot water temperatures other than 280°C so as to get closer to stratifiedstates measured on the plants. Complementary tests of fast thermal shocks were made at the end-of-life of themodel. Some modifications of the loop were made to obtain an instantaneous increasing of cold water inflow.

5 - TESTS AT STABLE STRATIFIED STATES

Some of the results presented in this paragraph have already been published [7].

5.1 - Thermal-hydraulic aspects

The tests have been carried out by keeping constant the cold and the hot incoming flow rates until thetemperatures in the model had stabilized. Half an hour at least is necessary to let the temperatures stabilize. Thestratification level stabilizes in a short period of time, but the vertical thermal gradient in the stratificationinterface increases slowly. Because thermal stresses depend on the thermal gradient, it is necessary to wait for astable state to obtain the maximum stress values.

Different combinations of incoming flow rates have been investigated as shown in figure 7. Below a cold flow rateof 5 m3/h, the level of stratification is independent of the hot inflow (if this one is over 0.5 m3/h). Above a coldflow rate of 5 m3/h, results are more and more dependent of the hot inflow, and the comparison is not satisfactory

10- 5

with in situ measurements. This discordance is assumed to result from the diameters of the outlet nozzles whichare too small.

The position of the stratification interface and the maximum vertical thermal gradient (see the profile in figure 8)both increase as the cold flow rises (see also § 5.3). When the cold water inflows is below 0.5 m3/h the cold waterstream is heated by the wall. For example, in the case of a cold flow rate of 0.1 m 3/h at 60°C, the temperature ofthe bottom increases to 100°C just after the elbow and then rises as a function of the distance from the elbow.This makes the thermal stresses smaller for very low cold inflows.

No significant variations in the temperature of the stratification interface have been noticed during the tests. Themaximum peak-to-peak variation was 30°C at a frequency ranging from 0.1 to 1 Hz.

Thermal-hydraulic calculations have been made successfully on the model configuration [8].

5.2 - Thermomechanical aspects

A heat exchange coefficient cannot be easily defined in the zone of the stratification interface because the fluidtemperature varies abruptly with the altitude. However the measurements on the mock-up enable adetermination of a global « transfer coefficient ». Its value was about 3500 W/m²/°C, that is much higher thanthose determined in the other layers of the flow. No quantitative explanation has yet been found.

At the interface level, the circumferential thermal gradients on the wall are softened. Their ratio relative to thethermal gradient in the fluid, is about 1/2 on the inner wall and 1/4 on the outer wall (less for low stratifications).

The global displacements of the model were measured at the point where the support takes up the weight. Theyare considerable and can reach 100 mm in the vertical direction and 50 mm in the horizontal direction. Themeasurements of stresses is reported in the chapter 8.

5.3 - Comparison with in situ measurements

The figure 9 summarizes the comparison with measurements made on the SG feedwater system in actual PWRplants. The positions of the stratification interface are quite comparable. The thermal gradients exceed thosemeasured in the plants, because of care taken to obtain stabilization. As indicated in § 5.1 the thermal-hydraulicbehavior of the COUFAST mock-up is representative of the behavior of SG feedwater systems, only if the coldwater flowrate is less than 5 to 6 m3/h (and the hot water flowrate more than 0.5 m3/h). Then, with unchangingtemperatures, the height H of the interface level correlates remarkably with the cold water flowrate.

Based on the COUFAST results and on the measurements in 3 EDF plants, it has been possible to establish asimple mathematical model which gives the relative height H/D depending on the Froude number (figure 9).

6 - TESTS OF QUICK ESTABLISHING OF STRATIFICATION

In the SG feedwater systems the flow rate may change frequently and quickly. So it was advisable to evaluatethe value of stresses on the inner wall in these conditions. Modifications of the facilities were achieved to applycold thermal shocks to the mock-up. Four tests have been carried out, starting from two different initial statesand for two cold water flow rates. In a first phase a cold front spreads out horizontally over a height about 8 cm(figure 10). Then, in a second phase, the cold front rises up, turning into a horizontal stratification, while thevertical thermal gradient gradually increases in the fluid, reaching temporarily 180°C/cm.

Thermal variations measured during the thermal shocks (maximum peak-to-peak ∆θ = 80°C) are small. Theadditional stresses measured on the inner wall during thermal shocks are weak (about 10 MPa). From the fatigueviewpoint, they do not cause any other significant damage because they set up in zones which differ from thoseaffected during stable stratified states.

7 - CYCLING TESTS UNDER THERMAL FATIGUE CONDITIONS

7.1 - Choice of the cycling conditions

The most damaging cycle (figures 7 and 11) has been defined on the basis of measurements made during stablestratified tests, and was confirmed by numerical studies. It was found to be a cycle between 2 stratified statescorresponding to positions of the stratification level of about 50° and 70° (azimuthal angle from the bottom). The

10- 6

greatest amplitude of stresses in the straight section (about 320 MPa) is reached on the inner wall at anazimuthal angle of 60°.

Water chemical conditions during these cycling tests were imposed to comply with the maximum valuesauthorized for SG feedwater systems in the French nuclear power plants. Stratification can occur only when thesteam generator is supplied through the Auxiliary Feedwater System as during hot shutdowns.

In this case the ultimate values are :

- pH (at 25°C) : 8.8 to 9.3,

- oxygen content lower than 100 µg/l,

- cationic conductivity lower than 2 µS/cm.

7.2 - Realisation of cycling

This cycle has been applied about 4000 times on the model in four series of about 1000 cycles. Non destructiveexaminations (gamma radiographic on welds and ultrasonic on the elbow) have been made prior to the cyclesand after each series. No evolution has been noticed and no indication was found.

The respect of the oxygen content is of a prime importance for fatigue life of ferritic steels. The readings haveproven that the maximum value reached was about 30 ppb (or µg/l) during only a very short period of time. Theaverage value was about 5 ppb, which corresponds to the maximum admissible value under normal operatingconditions (power > 25% nominal power). The condition imposed in French PWR plants has been respected andso it is not useful to apply the correction proposed for example by EPRI (environment factor) when, among otherthings, the dissolved oxygen in water is higher than 100 ppb [9, 10, 11].

7.3 - Examination after cycling

At the end of its life, the destructive examinations of the model revealed small cracks in the welds tapers. Thecracks were situated on each weld in the area affected by the variations of the stratification interface position*.

reference of weld S1 S2 S3 S4 S5

height / diameter (H /D) 0.27 0.2 0.19 0.17 0.13

crack lengths (mm) 57 30 33 42 (39)

depths (mm) {left / right) 1.35 / 1.25 1.4 / 2.0 1.95 / 1.8 2.8 / 0.75 (1.0)

* Note : Taking into account the curvature of the mock-up induced by the stratification, the heights of the cracks have to be corrected. Reportingthe relative heights to the location of the rods 3 & 4, H/D* = 0.19 which is exactly the average of the 2 stratification levels.

The biggest crack is 2.8 mm deep and is located on the most severe taper. The weld S5 which is shot-blasteddoesn't show any crack in the tapers, but only one (1 mm) at the junction with the weld fillet.

8 - NUMERICAL SIMULATIONS OF THE TESTS

8.1 - Numerical simulation of static tests

All the calculations described later were made with the Code_Aster developed by EDF R&D Division [12].

The work first consisted of validating the optimum meshing in the straight section of the piping before starting acomplete 3D calculation. We used a 2D finite element model with generalized plane deformations. Themeasurements in the section of the 3 & 4 rods (figure 3) were used for the validation because this sectionincluded many thermocouples. It was possible to compare the fluid temperature profiles, the temperatures on theinner and the outer wall, and the stresses measured by strain gauges at 6 points on the outer wall.

The validation is made for seven stratified states giving different positions of the stratification level. Using thespecific transfer coefficient (see § 5.2) the temperature profiles calculated on the inner and the outer wall areclose to those measured (fig. 12). Calculated and measured stresses are also in good conformity (fig. 13). On theinner wall, calculated stresses are the highest on the flank just below (120 MPa in traction) and just above (200

10- 7

MPa in compression) the stratification level. They depend on the level of the stratification in amplitude but moresignificant is the variation of the position of the extremum of these stresses which follows the level of thestratification. Consequently, the maximum stress amplitude is reached when stratification is cycled between twostable stratified states. The optimal states were determined to be the S_43 and the S_26 (see figure 7).

A complete calculation using a 3D finite element model has then been made for the same seven stratified states.The comparison between stress measurements (located in the current zones) and calculations was satisfactory,too. The highest stresses calculated were in the weld tapers, so the following computations concentrated on thesimulation of these tapers (3 geometric shapes) by 3D elastic finite elements. It should be noted that the stressintensification factor Kt depends on the level of stratification (1.5 to 2). For the conditions of the tests, Kt ≈ 1.8close to the factor K3 = 1.7 tabulated in ASME III & RCC-M [13] for as-welded joints.

8.2 - Numerical elastic calculations of the fatigue generated by thermal cycles

For fatigue calculation it is essential to know the stress amplitude range at the location where it is the highest.Elastic computations performed for the 3 different shapes of the tapers allowed to select these locations for thechosen cycle of stratification (between the states S_43 - S_26). The variation of stresses during a cycle isobtained by the difference between stresses calculated for the 2 stabilized stratified states.

The vertical profiles of temperature for these 2 states were imposed as loading conditions (figure 11). The meshused for the computation of the mock-up (1/4 of the model with regard to symmetry conditions) is shown onfigures 14 and 15. The boundary conditions were "no longitudinal displacements" in the center of the weld (planeof symmetry), and "free end" displacements at the other end ( lengthened to simulate an continuous pipe).

The vertical profile of the stress intensity (Tresca) variation during a cycle shows (figure 16) a steep peak whichis situated just between the minimal and maximal heights of the stratification interface for the 2 endings of thecycle. The interpretation of the results was made according to the chapter B 3200 of the RCC-M (or ASME III)because in the chapter B3600 (Piping analysis) it is not convenient to introduce non-axisymmetric loading (asstratification). It gives the following results for the weld S2 (standard weld on the straight section) :

- range of total stress Sp : 540 MPa- range of primary plus secondary stress Sn : 420 MPa

The stresses mentioned above were calculated taking into account the dependence of thermo-elasticcharacteristics (Young's modulus and coefficient of expansion) on the temperature. They lead to a usage factorof about 1.3 for the 4000 cycles carried out on the mock-up. We used also the new rule introduced in the RCC-Mcode for austenitic stainless steels (June 1994) to evaluate its impact (The specific rule for ferritic steels is underdevelopment). The plastic concentration factor Ke increases then from 1.3 to 1.5. Thus the usage factorbecomes 1.9 which is in accordance with the small depth of the cracks. Without dependence of characteristics ontemperature, the usage factor are respectively 0.7 and 1.4 (see the recap in the table § 8.3).

Note : The main advantage of the new rule is to be more realistic and less sensitive to methods of calculation. Using thenew rule, the usage factor of the weld with an excessive tapering (S3 with an angle of 20°) decreases from 8.7 to 4,while it increases from 1 to 1.5 for the normal tapers S1 and S2 (these calculations were made with the samehypotheses and methods which were differing from those of the recent calculations reported herein).

8.3 - Numerical elastic-plastic calculations of the fatigue generated by thermal cycles

Considering the important role of the factor Ke (for simplified elastic-plastic analysis), elastic-plasticcomputations have been made according to RCC-M method (calculation of the range of the deformation).

The elastic-plastic cyclic behavior of the A42 ferritic steel (~ SA 106 carbon steel) has been estimated at roomtemperature by fitting different curves of ferritic steels given in literature. It has been precisely measured at themaximum temperature of the cycle (280°C) on specimens of the same material used for the mock-up making.The figures 17 and 18 present the cyclic curves (used just as they are in the non linear isotropic modeling) andthe linearization used in the linear kinematic hardening modeling.

The incremental computations take into account the internal pressure (first stage). The second stage consists ingoing from the isothermal cold state to the first stratification state (S_26) with an intermediate state (see figure11) to avoid an abnormal plastification. Then the fatigue cycles (S_26 - S_43 - S-26) are applied.

Four different cases are calculated. The first uses an isotropic modeling and the material behavior (stress-straincurve) is defined by averaging for each value of strain the stresses at 20°C and 280°C. The cyclic stress-strain

10- 8

curve at the most loaded point is shown on figure 19. The behavior becomes completely elastic during thesecond part of the first cycle (plastic adaptation) so the computations were not carry on beyond. The second casetakes into account the dependence of the behavior curves on the temperature by interpolating (linearly) betweenthe 2 basic curves. The figure 20 shows a plastic adaptation from the very first cycle.

The third case is a calculation using a linear kinematic model with a cyclic curve taken as the average of the 2basic curves. The computation needed four cycles to obtain a stabilization (this was probably due to numericalproblems). The figure 21 shows the plastic accommodation (non linear open cycle) and the figure 22 shows theevolution of the equivalent strain (based on Tresca and assigned with an algebraic sign). The fourth case with thelinear kinematic model used the dependence of curves on temperature (linear interpolation of the parameters ofthe 2 models at 20°C and 280°C). There is again a plastic accommodation but after optimization of the numericalparameters, the stability is obtain from the first cycle (S_26 - S_43 - S_26).

The following table gives the results in term of equivalent strain range and the usage factors obtained by fullelastic-plastic analysis and by simplified elastic-plastic analysis (called "elastic" in the last column ; see § 8.2) :

reference of calculationequivalent strain

range (%)usagefactor

Ke* = εpl / ε el

(full elastic-plastic)RCC-M rulefor Ke (Sn)

Ke factor"elastic"

usage factor"elastic" analysis

isotropic at average T 0.29 0.60 1.13 before 1994 1.19 0.69

kinematic at average T 0.29 0.62 1.13 new (austenitic) 1.48 1.37

isotropic depending on T 0.32 0.78 1.15 before 1994 1.32 1.28

kinematic depending on T 0.41 1.75 1.46 new (austenitic) 1.5 1.9

It is well known that the isotropic modeling of the cyclic behavior gives an underestimation of the actual behaviorof the structure, and the kinematic modeling gives an overestimation.

The Ke factor used following an elastic calculation is determined as being the upper boundary of the rate (Ke*)between strain range obtained by elastic-plastic calculation and that obtained by elastic calculation. This isalways verified for the new rule introduced in the RCC-M. For the previous rule (the same as in ASME Code) theKe factor is under the elastic-plastic Ke* obtained with the kinematic model depending on temperature.

Only the usage factor obtained by this last modeling is upper 1 and could be considered as conservative.

9 - CONCLUSION

The study of the COUFAST mock-up was conducted in full scale (dimensions, temperatures, pressure) of thesteam generator feedwater systems. Much important knowledge was acquired during the tests at stable stratifiedstates (temperature profiles in the fluid, thermal transfer coefficients on stratification interface…). The elasticcalculations were validated by comparisons with measurements. After that, fatigue tests were conducted bycycling between the 2 stable stratified states giving the most severe alternating stress range (very severe incomparison with actual situations in plants). The destructive examinations after 4000 cycles revealed smallcracks in the tapers of welds. The usage factors were calculated with different methods. They are acceptable inrespect with the depth of cracks if they are calculated with the new rule introduced in the RCC-M. Elastic-plasticcomputations are conservative only by using a kinematic hardening model for material cyclic behavior. It isstrongly recommended to take into account the dependence of behavior on temperature.

REFERENCES

[1] US Nuclear Regulatory Commission : Cracking in feedwater system piping, Bulletin 79-13, June , 1979.

[2] USNRC : Thermal stresses in piping connected to Reactor Coolant Systems, Bulletin 88-08, June 1988.

[3] USNRC : Pressurizer Surge Line Thermal Stratification Bulletin 88-11, December 20, 1988.

[4] Deardorff A.F., Kim J.H., Roidt R.M. : Thermal stratification in Nuclear reactor piping system, ICONE, Japan,November 1991

[5] Wolf L., Schygula U., Geiss M., Handjosten E. : Thermal stratification tests in horizontal feedwater pipelines,NUREG/CP-0091, Volume 5, 1987.

[6] Wolf L., Häfner W., Schygula U., Geiss M., Handjosten E. : Results of HDR-experiments for pipe loads underthermally stratified flow conditions, Nuclear Engineering and Design 137 (1992)

10- 9

[7] Stéphan J.M., Caron J. : COUFAST : Experimental study of mechanical consequences of thermalstratification on a piping elbow, ASME PVP , New Orleans, June 1992

[8] Péniguel C., Stéphan J.M. : Thermal hydraulic and mechanical behaviour of an elbow in presence of astratified flow, NURETH-5, Salt Lake City, September 1992

[9] Chopra O. K. : Status of Fatigue Issues at Argonne National Laboratory, EPRI Conference on OperatingPower Plant Fatigue Issues & Resolutions, Snowbird (Utah), August 1996

[10] Gosselin S. R. : An environmental Factor Approach to Account for Reactor Water Effects in LWR PressureVessel and Piping Fatigue Evaluations, EPRI Conference, Snowbird, August 1996

[11] Mehta H.S., Gosselin S.R. : An environmental Factor Approach to Account for Reactor Water Effects inLWR Pressure Vessel and Piping Fatigue Evaluations, PVP-Vol. 323, Volume 1, ASME 1996

[12] Lévesque J.R. : Code_ASTER®, © EDF DER 1996, http://www.edf.fr/der/html/produits/production

[13] AFCEN : Règles de Conception des Chaudières nucléaires-Mécanique (RCC-M), 1993, AFNOR, Paris

[14] Masson J.C., Stephan J.M. : Fatigue induced by thermal stratification - Results of tests and calculations ofthe COUFAST model, OCDE workshop on thermal stratification, Paris, june 1998

10-10

AUXILIARY FEEDWATER LINESTRATIFICATION

AND COUFAST SIMULATION

J.M. STEPHAN , J.C. MASSONEDF/R&D Division

Mechanics and Technology of Components

International Conference on Fatigue of Reactors Components July 2000

10-11

General overview of COUFAST model and of testingconditions

Stable states of stratification

Fatigue cycling

Numerical interpretation

Conclusion

Plan

10-12

1.1 - Different types

Thermal Mixing TurbulentStratification Zones Penetration

COOK 2 CIVAUX FARLEYDIABLO CANYON TIHANGESEQUOYAH DAMPIERRE

Thermal Hydraulics phenomena on PWR pipings

10-13

1.2 - Main difficulties :

- Local phenomena

- Inside piping phenomena

- Difficulties to obtain thermal characteristics withCFD calculations

Needs for experimental evaluation in complement of onsite measurements

Thermal Hydraulics phenomena in PWR pipings

10-14

2.1 - General overview of COUFAST model and of testingconditions

2.2 - Stable states of stratification

2.3 - Fatigue cycling

2.4 - Numerical interpretation

2.4 - Discussion on methods for prediction of fatigue

2.5 - Conclusion

Exemple : Thermal Stratification on COUFAST model

10-15

General overview of the COUFAST mock-up

hot water

! outlet

"

cold wateroutlet

" inlet (280°C)

! cold water inlet (60°C)

embedded

stratificationinterface

cross-section

280°C

cold water(60°C)

Diameter 406.4 mmThickness 21.4 mm

Ferritic A42 steel

weightsupported

WELDS

S1 S4S2 S5S3

10-16

pH (at 20°C) : 9,0 +/- 0,3

oxygen content : < 31 µg/l

Cationic conductivity : < 2 µµµµS/cm

Water Chemistry

10-17

reference

of the weld

S1

(elbow)

S2

(straight)

S3

(20 °)

S4

(r = 0)

S5

(shot)

crack length (mm) 57 30 33 42 (39)

crack depth (mm)

(elbow side)1.35 1.4 1.95 2.8 (1.0)

crack depth (mm)

(oulet side)1.25 2.0 1.8 0.75 0

results of destructive examinations

10-18

General overview of the COUFAST mock-up

hot water

! outlet

"

cold wateroutlet

" inlet (280°C)

! cold water inlet (60°C)

embedded

stratificationinterface

cross-section

280°C

cold water(60°C)

Diameter 406.4 mmThickness 21.4 mm

Ferritic A42 steel

weightsupported

WELDS

S1 S4S2 S5S3

10-19

1 - to adjust the values for heat exchange coefficients

cold water (forced circulation) : 750 W / m2K-1

hot water (natural circulation) : 2000 W / m2K-1

interface (special exchange) : 3500 W / m2K-1

2 - to optimize the mesh for 3D calculations

32 elements on the lower quarter of the cicumference

2 elements through the thickness

2D calculations of stable states of stratifications

10-20

- 7 states of stratification to analyse the influence of

the height of the stratification level

the sharpness of the gradient

- 4 different geometries to compare

the elbow and the weld S1 connecting it to the pipe

the tapers in welds S2, S3 and S4

3D calculations of stable states of stratifications

10-21

- good adjustment of the thermal model

- validation of mechanical calculations

- parametric studies (4 geometries - 7 states)

guided us to chose the most appropriate cycle for fatigue

gave us indications on stress intensification in the tapers

main results for stable states of stratifications

10-22

geometry

stratif. heightS1 S2 S3 S4

55 mm 1.98 1.78 2.63 1.88

70 mm 1.78 - 2.60 -

95 mm 1.78 - - -

115 mm 1.77 1.77 2.56 1.88

125 mm 1.75 - - -

150 mm 1.50 - - -

190 mm 1.72 1.73 2.66 1.92

stress intensification factors

10-23

↔↔↔↔

reference ofcalculation

Ke* usagefactor

RCC-M ruleKe (Sn)

Kefactor

usagefactor

isotropicaverage T

1.13 0.60 before 1994 1.19 0.69

kinematicaverage T

1.13 0.62 new rule(austenitic)

1.48 1.37

Ke* = epl / e el

plastic calculations elastic calculations

comparison of Ke and usage factors for average T

10-24

↔↔↔↔

reference ofcalculation

Ke* usagefactor

RCC-M ruleKe (Sn)

Kefactor

usagefactor

isotropicdepending on T 1.15 0.78 before 1994 1.32 1.28

kinematicdepending on T 1.46 1.75 new rule

(austenitic)1.5 1.9

Ke* = epl / e el

plastic calculations elastic calculations

comparison when depending on temperature

10-25

behavior's modelfor calculation

equivalentstrain range (%)

usagefactor

isotropic

average temperature0.29 0.60

linear kinematic

average temperature0.29 0.62

isotropic

depending on temperature0.32 0.78

linear kinematic

depending on temperature0.41 1.75

results of plastic calculations

10-26

- much knowledge on stratification

- validation of thermal and elastic calculations

- severe fatigue cycling ➞➞➞➞ small cracks

- usage factors acceptable if using the new RCC-M rule

- elastic-plastic calculations conservative if kinematic modeling

CONCLUSION

11-1

11 FATIGUE EVALUATION IN PIPING CAUSED BYTHERMAL STRATIFICATION

Eduardo Maneschy and Rodolfo SuannoEletronuclear S.A.

Universidade do Estado do Rio de JaneiroRio de Janeiro, RJ

Brazil

11-3

FATIGUE EVALUATION IN PIPING CAUSED BY THERMAL STRATIFICATION

Eduardo Maneschy 1 and Rodolfo Suanno 1,2

Eletronuclear S.A. 1

Universidade do Estado do Rio de Janeiro 2

Rio de Janeiro, RJBrazil

ABSTRACT This paper presents the fatigue evaluation of the thermal cycling stratification in the residual heatremoval (RHR) auxiliary line of Angra 1 nuclear power station. Since the phenomenon is plant specific,the investigation is conducted through an experimental study (temperature measurements) and numericalanalysis (heat transfer and stress calculations). The temperature profile is obtained using thermocouplesinstalled on the external surface of the pipe, at five circumferential locations. These data are used as aninput to a transient heat transfer model in order to obtain the temperature distribution through pipesection. The temperature difference in the section produces local stresses (due to the non-linear profileacross the pipe section) that are superposed to the global ones (moments due to restrained thermalexpansion) to obtain total response. A three-dimensional finite elements model is used for the thermaland structural analyses, being the numerical solution carried out by ANSYS program. The part of theRHR considered is class 1, and the fatigue calculation to obtain the residual life of the piping isperformed according to the ASME Code Section III Subsection NB-3200 requirements.

INTRODUCTION Since the late 80’s thermal stratification in piping has been one of the most important concerns in thenuclear industry. At least two plants, Farley 2 (USA) in 87 and Tihange 1 (Belgium) in 88, have reportedprimary coolant leakage due to cracks in elbow piping of an emergency core cooling system. Thedamage is attributed to high cycle thermally induced fatigue mechanisms, which produces through-wallcracks in welds (Farley) and base metal (Tihange). At Genkai 1 (Japan) in 88 a similar defect wasreported in the weld of an elbow caused by thermal cycling in the unisolable part of the RHR system,[1]. Additional information of these events is available in the USNRC Bulletin 88-08, [2]. Dampierre 2in 92 and Dampierre 1 in 96, both in France, [3] and [4], and Oconee 2 (USA) in 97, [5], alsodeveloped through-wall crack in unisolable parts of piping system due to thermal stratification.

The current industry data indicate that the thermal cycling is critical for fatigue and appears to dependon piping configuration and on the power level of the plant. This phenomenon eventually is responsiblefor cracks in horizontal section of the piping caused by intermittent valve leakage and/or turbulentpenetration, [6]. As a result of these problems, and to avoid future leakage, corrective and preventiveactions were considered in many plants.

To take into account the thermal stratification, a transient not included in the original designspecification, a temperature monitoring program was implemented in Angra 1. The pressurizer surge

11-4

line was the first system selected to be investigated. The analysis was conducted by Westinghouse in1992, and concluded that stratification load would not impact the original design, [7].

Recently, in response to a question raised by the Brazilian Licensing Authority, a screening wasperformed in Angra 1 to find out which systems could be susceptible to stratification. The evaluationindicated the RHR suction line between the reactor coolant system and isolation valve as a potentiallocation. In order to quantify the severity of the phenomenon, a temperature monitoring system wasinstalled on the piping outside surface. Since the result of the plant measurements has shown a high levelof thermal cycling, a structural analysis was conducted to evaluate the stratification effect on the lifetimeof the line.

The present paper focuses on experimental and numerical approaches adopted to perform the fatigueassessment of the RHR piping. The results of the monitoring program are used as an input to heattransfer and stress evaluation. Since the length of horizontal run upstream of the isolation valve is short,stratification induces local thermal stress through the pipe wall. The resulting stress should be combinedwith global stresses caused by bending deflection (restrained thermal expansion). Both thermal andstructural analysis are developed employing three-dimensional finite elements models. Numericalsolution is performed by ANSYS program, [8], and the stress analysis is conducted according to ASMESection III Subsection NB-3200 Code, [9].

MONITORING TEMPERATURE IN ANGRA 1 Angra 1 is a 657 MW Westinghouse two loops plant. The RHR system comprises a 8 inches sch 160pipe (outside diameter is 219 mm and thickness is 23 mm) made of stainless steel (ASME SA 376 TP304). The isolation valve is located in the horizontal section, approximately ten times the outsidediameter away from the connection with main reactor coolant piping (hot-leg). Part of the layout isshown in Fig.1.

The temperature monitoring system consists of thermocouples mounted on the outside surface of thepiping, in a section located 0,20 m upstream of the valve. The sensors are installed at five differentangular points around the pipe, according to the schematic representation indicated in the Fig.1. The dataare obtained continuously at every two minutes since the end of 1998. Figure 2a is a typical resultshowing the temperatures measured at 100% power operation as a function of time. The result during aday is nearly the same since the beginning of the measurements program. For loop # 2 there are twodistinct responses between the different fluid layers. The thermocouple at the top of the pipe shows thatthe response is cyclic, and the temperature changes from 225oC to 300oC in approximately 40 minutes.The remaining thermocouples indicate constant temperatures, with values at the four circumferentiallocations equal to 198oC, 167oC, 157oC and 144oC. The cyclic response is strongly dependent on thepower level of the plant, as can be observed comparing Figure 2a (100% operation) and Figure 2b (60%operation). A smaller level of stratification, without oscillation, is observed in loop # 1.

The temperature distribution at 100% power is plotted as a function of relative height (sensor locationnormalized to the outside pipe diameter) and depicted in the Fig.3. Configurations that correspond to thefluctuations between maximum (300oC) and minimum (225oC) temperature at the top are shown. Notice

11-5

Fig. 1 – Piping layout and thermocouples locations

Pipe cross section AA (• Thermocouples positions)

0,2 m

M

Section AA

RCS(Hot Leg)

60°

60°

11-6

2a – 100% power operation

0

50

100

150

200

250

300

350

00:00 04:00 08:00 12:00 16:00 20:00 00:00

Time (hr)

Te

mp

era

ture

(oC

)Top

Bottom

2b – 60% power operation

0

50

100

150

200

250

300

4:00 8:00 12:00 16:00 20:00 0:00

Time (hr)

Tem

pera

ture

(o C

)

Top

Bottom

Fig. 2 – Measured temperatures at pipe outside diameter as a function of hours of day

11-7

-1

-0,5

0

0,5

1

0 50 100 150 200 250 300 350

Temperature (oC)

Rel

ativ

e he

ight

(s

enso

r lo

catio

n/o

uts

ide d

iam

ete

r)

T maxT min

Fig. 3 – Maximum and minimum temperature stratification through pipe section (relative height = 1 top ; relative height = −1 bottom)

Fig.4 − Piping layout for three-dimensional model

0°

45°90°

Section AA

0°

Pipe cross section(viewed from valve side)

11-8

that top-to-bottom maximum temperature difference on the outside wall reaches approximately 156oC,being the profile through the section continuous. By contrast, a discontinuous one is observed in thepressurizer surge line.

It is important to mention that at least two walkdowns inside containment building during operationwere conducted, and a leakage through the isolation valve was difficult to be identified. Besides that, thelevel of the reactor coolant drain tank was also checked, and no reportable variation was noticed. Itappears that the Angra 1 thermal cycling might not be produced by the active mechanism that occurredin Genkai 1, [1], but can be similar to those developed due to turbulent penetration induced by the hot-leg flow, [6]. Accordingly, temperature fluctuations at the top are caused when a hot turbulent flow fromthe hot-leg penetrates into the RHR branch line and interacts with the cooler fluid at the horizontal partof the piping.

THERMAL ANALYSIS Because the instrumentation is mounted on the external surface of the pipe, the fluid condition insideis not known and the temperature distribution through the pipe section can not be determined. In order toovercome this difficult, a practical alternative should be adopted. In this paper the basic idea is toassume an initial condition for the fluid and verify, using transient heat transfer calculation, if thetemperature response matches the values measured outside. To achieve this goal, an iterative procedureis necessary.

Numerical method to solve the problem is based on the three-dimensional finite element method. Thethermal model is constructed using the eight-node element (ANSYS SOLID70), with the pipe thicknessmeshed using four elements. The system is simulated in the stratified region and includes part of thevertical section of the line, elbow, horizontal portion and pipe-valve transition zone as schematicallyrepresented in Fig.4.

The flow is restricted to a small area located in the upper part of the horizontal pipe, and the boundarycondition is the temperature profile presented in Fig.3. At this area the water is supposed to heatup infive seconds from 225oC to 315oC (hot-leg temperature). The values in other points through the sectionare unchanged. The temperature history at the pipe is obtained adopting a heat transfer coefficient equalto 1600 W/m2 oC, a number selected in a conservative basis, and considered to cover a wide range offlow rates. These values are consistent with the measured temperatures and represent the final picks afterseveral trials, which include variation in the fluid temperature and its time to reach maximum. Thethermal model for the minimum stratification (initial stage, beginning of the transient) is represented inFig.5. It can be noticed that at the external surface the temperature in the upper part is 225oC, and decaysto the quantities readings in the other thermocouples (locations and results as shown in Fig.1 and Fig.3).The temperature distribution at final stage (end of the transient), when the value at pipe outside reaches300oC, is shown in Fig.6. This is the maximum stratification and occurs 300 seconds after starting thecalculation.

The corresponding temperatures at these instants are reproduced in Fig. 7 and 8 as a function of pipeinternal circumference. The numerical solution is obtained for three sections at the elbow and the sectionwhere the thermocouples were installed (see Fig.4). It can be observed that such as the measured data,the change in calculated temperature is limited to a small area located at the top portion of the pipe.

11-9

Fig. 5 – Temperature distribution at the minimum stratification level (beginning of the transient, 225oC at top pipe)

Fig. 6 – Temperature distribution at the maximum stratification level (end of the transient, 300oC at top pipe)

11-10

Fig. 7 – Temperature distribution around pipe internal circumference (beginning of the transient, 225oC at top pipe)

Fig. 8 – Temperature distribution around pipe internal circumference (end of the transient, 300oC at top pipe)

0

50

100

150

200

250

300

350

0 100 200 300 400 500 600

Pipe circumference (mm)

Tem

pera

ture

(o C

)

Section AA

Elbow 90 Degrees

Elbow 45 Degrees

Elbow 0 Degrees

0

50

100

150

200

250

300

350

0 100 200 300 400 500 600

Pipe circumference (mm)

Tem

pera

ture

(oC

)

Section AA

Elbow 90 Degrees

Elbow 45 Degrees

Elbow 0 Degrees

11-11

A close examination of the data confirms that the numerical simulation approximately reproduces themeasured temperatures (see Fig.9), giving an indication that the through-wall solution is adequate forstructural analysis. Notice that the highest stresses are induced during steady-state response, when thegreatest top-to-bottom temperature difference is achieved. Therefore, the stratification analysis differsfrom thermal shock problems, when the stresses associated with through-wall gradients occurs duringthe transient portion of the loading, and tends to disappear when the steady-state is reached.

STRUCTURAL ANALYSIS The structural analysis is conducted with a three-dimensional finite element model, constructedemploying a eight-node element (ANSYS SOLID45). The mesh is represented in Fig.10, and is builtwith 21120 solid elements. The thermal stresses are evaluated using the most unfavorable temperaturegradient through the pipe section, derived from the heat transfer results, which occurs at the end of thetransient.

In order to save computer running time only a part of the piping layout is modeled, with boundaryconditions imposed using substructuring technique. Therefore, the effect of the removed parts arecalculated previously and used as stiffness matrices at the ends. This implies that the restraints are notrigid and the analysis considers the flexibility of the removed pipe. The applied load is represented bythe temperature distribution along the piping, imposed in three different regions: from the hot-legconnection down to the vertical pipe (0,80 m before the elbow, see Fig.1) it is assumed as 315oC.Between this point and the pipe-valve interface the stratification profile as obtined by the thermalanalysis is imposed. After the isolation valve a 40oC (ambient temperature) is adopted.

Global stress resulting from the moments produced by the ends restraints is combined with local stresscaused by the thermal gradient through the pipe section. The total stress is obtained summing up thesetwo components. In the present work it is assumed that the mechanical loads (dead weight and seismic)are negligible when compared to the thermal stresses. Therefore, they are not included in thecalculations.

The input for stress analysis considers the values generated by the finite element solution and isperformed using the rules of the ASME III NB-3200 Code [9]. Accordingly, primary plus secondarystress intensity range is evaluated first (shakedown criterion), followed by the peak stress intensity rangecalculation. Finally, the ASME fatigue curves are employed to determine the cumulative usage factor,which shall be less than the allowable value of 1,0.

RESULTS The study is performed considering the minimum and maximum stratification illustrated in Fig.3.Because the stress is proportional to the temperature difference, the values corresponding to theintermediates levels of stratification are obtained by linear interpolation. On the other hand, the originalstress analysis performed by Westinghouse, [10], calculated the usage factor very small, which meansthat the design transients are negligible compared with thermal cycling loads.

The analysis considering the upper part of the pipe at 225oC is performed first. The results of thecalculation are presented in Fig.11. As next, the top of the pipe is assumed to be at 300oC. The analysis

11-12

-1

-0,5

0

0,5

1

0 50 100 150 200 250 300 350

Temperature (oC)

Rel

ativ

e he

ight

(sen

sor

loca

tion/

outs

ide

diam

eter

)

Fig. 9 – Measured (◊) and numerical (•) data comparisom at the maximum stratification (relative height = 1 top ; relative height = −1 bottom)

Fig. 10 – Solid model for structural analysis

11-13

is conducted and the resulting stresses are shown in Fig.12. The detailed observation related to the stressplots defines the horizontal pipe and isolation valve interface as the most critical location. Anothercomponent which is highly stressed is the elbow.

The fatigue analysis is carried out using the stress range when the internal pressure and thetemperature goes from one condition to another in time. Since the shakedown criterion is met, theconcentration factor Ke is 1,0, and the alternating component is half of the peak stress range. Thealternating stress intensity is amplified by 2,0, a number supposed to cover the effect of the structuraldiscontinuity at the pipe-valve interface, and by the ratio of the modulus of elasticity given on theASME fatigue curve to the value used in this analysis (Ec/Eh=1,11).

Two cases have to be considered. The first one is when the operation changes from zero loadcondition (cooldown) to 100% power, an event assumed to occur 200 times. The pressure plus thermalfluctuation stress range is 329 MPa, plotted in Fig.13, and the usage factor is 0,008.

The second case to analyze is when the temperature fluctuation at the top of the pipe takes place.During the transient the pressure is essentially constant and does not contribute to the total stress. At100% power the stresses due to thermal cycling are determined considering the differences at two loadconditions: transient down (225oC) and up (300oC) at the top part of the pipe, being the remaining valuesat other regions held constant (see Fig.3). This is supposed to occur nearly 36 times a day or at a roughguess, 420000 for plant life (40 years and availability factor of 0,8). As shown in Fig.14, the stress rangeat the pipe-valve location is 182 MPa. The calculated cumulative usage factor is 0,84.

As it can be noticed the stress range is controlled by the temperature oscillation at the upper part ofthe pipe. However, the temperature amplitude at this portion is not constant (see Fig.2a) and,consequently, the above usage factor is overly conservative. A better life estimation is obtained if anappropriated cycle counting, such as rainflow method, is considered. If this technique is adopted, andneglecting the small fluctuations (less than 30oC), only 20 meaningful cycles per day have to be counted(235000 cycles during plant lifetime). As a result, the cumulative usage factor is reduced to 0,47.

Despite of the above number being less than 1,0, a nondestructive examination was conductedduring the last Angra 1 outage (May 2000). To investigate if a crack developed due to thermal cycling,the ultrasonic method was adopted, being the locations selected the elbow and the horizontal pipingsections upstream the isolation valve. The examination was concentrated on the top of the pipe, theregion where the highest stresses take place (see Figs.11-14). All welds and base metal were inspected,and no indications were found.

Anyhow, the mitigation of the problem shall continue to be achieved. To reduce the stress fluctuations,operational and maintenance procedures, structural modifications or a combination among them, arehighly desired. Even though this does not eliminate thermal cycling, the damage would be less severe.

CONCLUSION The thermal cycling stratification phenomenon at Angra 1 RHR piping was analyzed by anexperimental and numerical approach. The data used in the study were obtained at 100% poweroperation, measured for a period from the end of 1998 to mid of 1999.

11-14

Fig. 11 – Stress distribution due to minimum stratification level (beginning of the transient, 225oC at the top pipe)

Fig. 12 – Stress distribution due to maximum stratification level (end of the transient, 300oC at the top)

11-15

Fig. 13 – Stress distribution due to pressure and maximum stratification when the operational condition changes from zero load to 100% power

Fig. 14 – Stress distribution at 100% when the stratification level changes from minimum (225oC) to maximum (300oC) at the top pipe

11-16

Since no leakage indication was found, it appears that the thermal fluctuation occurs due to turbulentpenetration. Two other reasons support this idea: 1) the isolation valve location related to hot-leg iswithin the limit to turbulent penetration, which is assumed to be between 10 and 25 branch line internaldiameters ; 2) the characteristic of the temperature distribution through the cross section shows a mixturebetween hot and cold fluid (by contrast, during leakage hot fluid is concentrated at the top of the pipe).However, additional investigation has to prove this statement.

The temperature profile on the pipe surface was used as input to thermal and structural three-dimensional finite element models. For the fatigue evaluation it was assumed 420000 cycles throughoutthe plant lifetime. This is a conservative estimation, based on stratification between maximum andminimum levels equal to 36 times a day. If the rainflow technique is adopted this value is reduced to 20cycles per day.

From the stress analysis results, the critical location is the interface between the pipe and isolationvalve. At this point the calculated usage factor is equal to 0,84 (conservative) or 0,47 (realistic), bothless than the allowable value. This indicates that the thermal cycling stratification observed in Angra 1does not have impact on the integrity of the RHR piping.

REFERENCES1. Shirahama, S., “Failure to the Residual Heat Removal System Suction Line Pipe in Genkai Unit 1Caused by Thermal Stratification Cycling”, Experience with Thermal Fatigue in LWR Piping Caused byMixing and Stratification, June 8-10, 1998, France, p.73-81.2. USNRC, “Thermal Stresses in Piping Connected to Reactor Coolant System”, Bulletim 88-08 andSupplements 1, 2, and 3, U.S. Nuclear Regulatory Commission, June 22, 1988.3. Navarro, G., “Determination of the Thermal Loading Affecting the Auxiliary Lines of the ReactorCoolant System in French PWR Plants”, Experience with Thermal Fatigue in LWR Piping Caused byMixing and Stratification, June 8-10, 1998, France, p.358-362.4. Gauthier, V., “Thermal Fatigue Cracking of Safety Injection System Pipes Non Destructive TestingInspections Feedback”, Experience with Thermal Fatigue in LWR Piping Caused by Mixing andStratification, June 8-10, 1998, France, p.436-453.5. Lund, A.L. and M. Hartzman, “USNRC Regulatory Perspective on Unanticipated Thermal Fatiguein LWR Piping”, Experience with Thermal Fatigue in LWR Piping Caused by Mixing and Stratification,June 8-10, 1998, France, p.41-48.6. EPRI, “Thermal Stratification, Cycling, and Striping (TASCS), TR-103581, March 1994.7. Westinghouse, “Structural Evaluation of the Angra Unit 1 Pressurizer Surge Line, Considering theEffects of Thermal Stratification”, WCAP-13685, Westinghouse, 1994.8. ANSYS, “ANSYS Finite Element Code Users' Manual-Revision 5.4”, Swanson Analysis SystemsInc., Houston PA, 1997.9. ASME, “ASME Boiler and Pressure Vessel Code, Section III Subsection NB and Appendices”,American Society of Mechanical Engineers, 1989.10. Westinghouse, “ASME Section III Piping Stress Analysis for the Angra Nuclear Generating StationUnit 1”, WCAP-9630, Westinghouse, 1994.

THERMAL FATIGUE II

12-1

12 CONSIDERATION OF THERMAL FATIGUE AND CYCLEMONITORING OF A B31.1 PLANT

Kenneth ChangAmerican Electric Power

Oral presentation only. No presentation slides or technical paper available.

12-3

Consideration of Thermal Fatigue and Cycle Monitoring of a B31.1 Plant

Kenneth ChangAmerican Electric Power

Oral presentation only. No presentation slides or technical paper available.

13-1

13 MAIN RESULTS OF EDF’S EXPERIENCE ON IN SITUMEASUREMENTS RELATED TO THERMAL FATIGUEON PWR REACTOR COOLANT PIPING

Frédéric DULCERE, Gilles NAVARROElectricité de France, Research and Development Division

6, quai Watier - 78401 CHATOU FRANCE- 33 .(0)[email protected], [email protected]

13-3

INTERNATIONAL CONFERENCE ON FATIGUE OF REACTOR COMPONENTSJuly 31 - August 2, 2000 Napa, California

Frédéric DULCERE, Gilles NAVARROElectricité de France, Research and Development Division6, quai Watier - 78401 CHATOU FRANCE- 33.(0)1.30.87.78.06

[email protected], [email protected]

MAIN RESULTS OF EDF’S EXPERIENCE ON IN SITUMEASUREMENTS RELATED TO THERMAL FATIGUE ON

PWR REACTOR COOLANT PIPING

1. INTRODUCTION

The various incidents imputed to thermal fatigue, which occurred throughout the world on theauxiliary lines of the Reactor Coolant System (SIS, RHR, CVC), led EDF to urge a research program inorder to determine the origins and the consequences of these problems. The stakes related to theseproblems concern, at the same time, the safety and the availability of the units.

This program was based on instrumentations of the nuclear power plants in operation whoseprincipal results are presented hereafter.

2. DEFINITION OF THE INSTRUMENTATIONS

Specific instrumentations were installed on the auxiliary lines of several power plantsrepresentative of the French nuclear capacity. The strategy of standardized plants enabled us to equip alimited number of installations while keeping a global sight of the phenomena affecting the whole of ournuclear plants. Thus, two 900 MW units (Blayais 1 and Cruas 4), one 1300 MW unit (Golfech 2), andmore recently one 1400 MW unit (Chooz B1) were instrumented.

The international experience feedback showed that the defects had occurred near or on the welds.Therefore, it is close to these that the majority of the sensors was installed. The choice of the positions ofthe transducers was defined according to four principal objectives:

• characterization of the thermal loadings supported by the auxiliary lines underoperation,

• recording of the transients for comparison with the conception transients,• highlighting of thermal cyclings,• discovery of unexpected phenomena.

The instrumentations mainly consisted of thermocouples welded on the external skin of the pipes(30 to 140 according to the site) and pressure pick-up measurements (approximately ten). For each

13-4

instrumentation, the principal signals of exploitation were also recorded in order to be correlated with thelocal data.

For each instrumented unit, uninterrupted acquisitions lasted several cycles of operation in order tobe ensured of the representativeness of the observations carried out. This monitoring was made withacquisition periods ranging between 1 and 20 seconds which made it possible to accurately record theconsequences of the main operating conditions on the instrumented lines.

3. UNISOLABLE PIPING LOADINGS

A good knowledge of the unisolable piping damage constitutes a very significant stake from asafety point of view. The availability is the second objective as incidents, on such part of the lines, inducelong stops of the units (More than 50 days for Dampierre 1 SIS crack).

The incident that occured at Dampierre 2 in 1992 was in accordance with discovered crackings atFarley and Tihange. Then, the instrumentation program aimed at determining these pipes damage risks byimproving our knowledge of the thermohydraulic phenomena which sit there.

3. 1. TURBULENT PENETRATION

The principal phenomenon affecting the unisolable piping of the non-discharging lines connected tothe main primary system is related to the shearing of the principal flow at the nozzle. It results in thepenetration of primary fluid in the auxiliary line and structure in a gimlet shape (vortex). The penetrationlength varies between 15 and 20 times the hydraulic diameter of the pipes. Even if this distance variesaccording to the thermohydraulic conditions of the flow and the lines geometry, the phenomenon wasobserved on the whole of the instrumented lines.

3. 2. CONSEQUENCES UNDER NORMAL OPERATION

3. 2. 1. HEAT PROPAGATION

According to the length of the unisolable piping and the vortex penetration depth, the head of thevortex can reach the first isolation valve. In this case, the temperature of the latter is close to the primarytemperature.

When the turbulent penetration does not reach the first isolation valve, two cases arise:• either the head of turbulence is situated in a vertical pipe and the propagation of heat is

stopped (the vertical pipe is hot at the top and cold at the bottom),• or the head of turbulence is in a horizontal pipe and a free convection current forms and

propagates heat to the valve or to the first vertical part going down.

Figure 1 shows an example of conditioning of a horizontal auxiliary line where the vortex isrelayed by a free convection current which leads to a very high temperature at the level of the valve.

13-5

LD =2 19

LD =1 14

302 °C

302 °C

272 °C

275 °C 265 °C

LD =3 24

301 °C

Hot Leg

Figure 1 : Heat propagation in the unisolable piping

The turbulent penetration does not lead to any fluctuation of temperature in so far as it only mixeshot fluid. If it meets a lower temperature fluid, coming for example from a leaking valve, slight cyclingscan be observed.

It is what the instrumentation highlighted, in particular on SIS lines and on draining of crossoverleg lines. These cyclings are regular and do not vary with operating conditions. They do not presentparticular risks.

3. 2. 2. VORTEX INSTABILITY

An unexpected phenomenon particularly caught our attention. It relates to an accumulator injectionline in Cold Leg (figure 2) where a complex thermohydraulic phenomenon creates repeated temperatureshocks upstream of a vertical part (figure 3). Its appearance frequency is random and it even completelydisappeared during several months. No correlation with operation of exploitation was found.

13-6

RCP 321 VP

LD =

3 21

LD =

1 5

LD =

2 18

RPE

BF

3

Figure 2 : Instrumentation of the accumulator injection line

These thermal shocks are attributed to length penetration variations of the primary vortex whosehead is near the elbow. Turbulence head movements occur, which, from time to time, go beyond theelbow and generate hot shocks in the higher part of the horizontal pipe, then go back causing slowcoolings of this part of the line.

Figure 3 : Example of thermal shocks

For this line, and some others, the designer (FRAMATOME) carried out fatigue calculations. All inall, these calculations showed, that under normal operation, the integrity of the auxiliary lines was notcalled into question in spite of the difference between the loadings expected and those measured.Nevertheless, some lines (of which the line seeing the temperature random peaks) are today the object ofa particular control program, because the fatigue resistance margins are from now on reduced.

13-7

3. 3. DAMAGED VALVES CONSEQUENCES : FARLEY-TIHANGE PHENOMENON

3. 3. 1. SIMULATION OF A SIS/CVC LEAK AT BLAYAIS 1

With an aim at better understanding the risks related to a leak of the isolation valves between themain primary system (155 bars) and the CVC (180 bars), in particular at the origin of the incidents ofFarley and Tihange, a specific installation was installed at Blayais 1 (figure 4). It consisted in the creationof a bypass of one of the isolation valves, equipped with two manual valves and a flowmeter, in order tosimulate a leakage with variable flow from CVC towards the primary system.

MD

MHHSI

1234567

1234567

12345

1

HL 1

HL 2

HL 3

ManualRegulating

ValveFlowmeter

Figure 4 : Functional diagram of the injection test

Injections staged between 0 and 300 l/h were carried out. They led to significant modifications ofthe lowest line thermal loadings, in this case the SIS line in Hot Leg 1. For the other lines, no sign of coldfluid was detected in the range of studied flow rate.

13-8

Position of the manual valve Section 1 Section 2 Section 3

Cycling measured at the invert of the pipe

Closed

1/2 tr

2 tr

Opened

Hot Leg 1

Figure 5 : Cyclings evolution during the injection test

More precisely, the observation of the unisolable part of this line teaches us about the interaction ofthe primary vortex and the cold water leakage coming from the high pressure system. As the leak-flowincreases, the area of interaction moves away from the valves. The more important the cold water leak is,the more the zone of interaction moves towards the RCS. In the range of studied flow rate, the amplitudeof cyclings also increases with the leak-flow. Upstream of the valve, the cold fluid injection results in animportant cooling of the pipe.

The thermomechanical calculations based on these results enabled us to determine anapproximation of the flow rate limit leading to welds damage.

3. 3. 2. FEEDBACK FROM THE DAMPIERRE 1 INCIDENT

Double cracking at Dampierre 1 in 1997 was especially surprising by its localization in the straightpart of the pipe. It mainly showed that our knowledge of the phenomena was still imperfect. Until then,the welds had always been regarded as the most sensitive points of the auxiliary lines. In fact, theinteraction between the primary vortex and the cold water leakage is very local and the instrumentation ofBlayais 1 did not make it possible to extrapolate the results obtained at the two ends of the horizontal pipeto the straight part of the line.

13-9

Among the investigations carried out following the discovery of the cracks, temperaturemeasurements during hot shutdown on the damaged line, allowed, by comparison with Blayais 1 injectiontests data, to identify the origin of the problem as being the SIS/CVC isolation valve leakage.

While waiting for the generalization of a modification equivalent to that of Tihange (pressure well),these temperature measurements are now carried out with each starting in complement of the tests aimingat highlighting a possible cold water leakage.

A more precise instrumentation was installed on the SIS line in Hot Leg 1 of Dampierre 1 in orderto improve our knowledge in the accused zone and to compare, in a wider way, the line loadings withthose of Blayais 1. The valve separating the SIS and CVC circuits having been repaired, thisinstrumentation showed that without leak, thermal loadings are equivalent between the two units.

4. DEAD-END BRANCH LINE LOADS

The damages noted in the dead-end branch lines (in between the two isolating valves) are generallylow, and few through-wall cracks were highlighted in these zones. The safety aspect is less critical therethan on the unisolable part, and the principal stake concerns maintenance and plant availability.

4. 1. HEAT PROPAGATION PHENOMENON

The instrumentations showed that, by conduction through the first isolation valve when it is hot,then by free convection, the temperature of most of auxiliary lines is high in the dead-end branch line.This propagation can extend over significant lengths on horizontal piping. Temperatures higher than100°C were recorded at more than 150 D from primary nozzle (figure 6).

Only a vertical pipe going down can stop heat conduction. The insulation removal associated with agood ventilation of the building can also contribute to the reduction of the temperatures, but the resultremains random.

RCP 422 VP

187°C

RCP 418 VP

107°C

BU2

RPE

L

DLD

L = 156D

2 L = 140D

1

Figure 6 : Heat propagation on a draining of crossover leg line

13-10

4. 2. CONSEQUENCES

4. 2. 1. WATER/STEAM OR WATER/AIR DIPHASIC ENVIRONMENT

The main risk related to a high temperature in the dead-end branch line is a chemical risk becausethe temperature favors the process of corrosion. The probability that this phenomenon leads to defectswould be extremely limited if piping contained only water because the chemical elements likely tocorrode metal remain in weak concentration.

The risk worsens when a diphasic environment is established in the dead leg. This diphasicenvironment can have two origins:

• A bad venting involving a water/air environment. This probability is related to thealignment and the vent valves positioning which do not always allow a perfect venting.

• A temperature in the dead leg higher than the saturation temperature, which generates awater/steam environment. Indeed, the pressure variations in some lines can be quite important(between 1 and 155 bars) when it is not imposed by the auxiliary line operation (example:pressure forced at 42 bars on the accumulator injection lines).

In both cases, the presence of a water/steam or water/air interface causes the concentration ofcorrosive elements at the interface (boric acid for example). This phenomenon results in the appearanceof level lines which can evolve in generalized corrosion according to the concentration of the corrosiveelements and the environment oxygen content. These damages are slow and represent a stake mostly froma point of view of availability and maintenance costs.

Besides, oxygen presence associated with a high level of stress (average stress or residual stress)favors the appearance of stress corrosion cracking. Unlike generalized corrosion, this degradation modecan be fast and involves several through-wall cracks (Bugey 3 SIS, draining of crossover leg lines ofCattenom 1 and Fessenheim 1).

4. 2. 2. THERMAL STRATIFICATION

The presence of a free convection current in the dead-end branch line causes the appearance ofthermal stratification. Maximum amplitudes of 80°C were observed. They are at the origin of strain andstress, the effects of which combine with the stress normally supported by the piping.

13-11

LD =7 75

LD =5 31

LD =6 72

LD =4 29

79 °C

110 °C

168 °C

109 °C

119 °C

130 °C

195 °C

Figure 7 : Stratification in the dead-end branch line

5. CONCLUSIONS

The whole of the instrumentations results helped us improve our knowledge of thermohydraulicphenomena affecting RCS auxiliary lines of units in operation.

The unisolable pipes are subjected to a hot vortex influence which, combined with a cold waterweak source, can generate thermal fatigue damage. The difficulty in having a perfect control of the coldwater arrivals led EDF to urge a modification on all 900 MW units in order to remove this source ofdegradation.

Concerning the dead-end branch line, the presence of a diphasic environment mainly related to hightemperatures can lead to the appearance of degradations of chemical origin from level lines till stresscorrosion cracking.

From a lawful point of view the gathered data are used to check the fatigue behavior of sensitiveauxiliary lines in spite of significant differences between the real loadings and those used for the design.

These results enabled us to improve our inspection strategy and preventive maintenance of theauxiliary lines. They are also used to determine in a realistic way the stresses supported by a given line, orto justify the behavior of some defects discovered during inspections.

The results of these instrumentations are finally used for future units design by taking into accountseveral cycles of acquisition on the 900 MW units, but also on the 1300 MW and 1450 MW units, wheresome innovations resulted from the experience feedback of first measurement series.

14-1

14 EVALUATION OF OCONEE-2 HIGH PRESSUREINJECTION/NORMAL MAKEUP (HPI/NMU) LINE WELDFAILURE

Bret L. BomanFramatome Technologies

Lynchburg, VA

James R. SmotrelFramatome Technologies

Lynchburg, VA

J. Michael DavisDuke EnergyCharlotte, NC

Timothy D. BrownDuke EnergySeneca, SC

14-3

Evaluation of Oconee-2 High Pressure Injection/Normal Makeup (HPI/NMU) Line Weld Failure

Bret L. BomanFramatome Technologies

Lynchburg, VA

James R. SmotrelFramatome Technologies

Lynchburg, VA

J. Michael DavisDuke EnergyCharlotte, NC

Timothy D. BrownDuke EnergySeneca, SC

ABSTRACTAn evaluation of the 1997 Oconee Nuclear Station -

Unit 2 High Pressure Injection/Normal Makeup (HPI/NMU)line weld failure is described herein. Computational FluidDynamics (CFD) analyses formed an integral part of the rootcause investigation. CFD results were used to “test” each ofthe postulated failure causes. Ultimately, the failure wasdetermined to be the result of a loose thermal sleeve thatallowed cyclic thermal mixing between cold makeup flow andhot reactor coolant system (RCS) water to occur. This cyclicmixing resulted in thermal fatigue that caused cracks to initiateand propagate.

INTRODUCTIONOn April 21, 1997, the Oconee Nuclear Station - Unit

2 experienced a through-wall crack in the weld between theHigh Pressure Injection/Normal Makeup (HPI/NMU) pipingand the nozzle safe-end, Figure 1. The crack caused a leak inthe 2A1 line. Plant operators correctly diagnosed the leak andbrought the plant to safe shutdown. Evaluations, repairs, andcorrective actions were conducted to understand the failure andpreclude further occurrences.

SYSTEM/COMPONENT DESCRIPTIONEach of the four RCS cold legs is equipped with high-

pressure injection piping for use during an emergency conditionsuch as a loss-of-coolant accident for the reactor. Two of these2-½ inch schedule 160 (66.6 mm ID) stainless steel piping linesprovide the normal makeup flow to the RCS to replace thewater being removed and purified via the letdown piping. Thenozzle is comprised of two parts, the base nozzle and a safe-end, Figure 1.

To prevent thermal cycling of the base metal, eachnozzle is equipped with a 1.5-inch (38.1 mm) ID thermalsleeve. Whereas the thermal sleeve experiences thermalcycling as the makeup flow varies to meet the demands of theRCS pressurizer level control, it does not experience pressureloading and had been perceived to be immune from significantthermal fatigue.

Two thermal sleeve designs have been employed inmost B&W-designed plants, Figure 2. The old design wasinstalled without prescriptive guidelines for maintaining a tightfit between the sleeve and nozzle. The upstream end of thethermal sleeve (away from the RCS) was contact-expanded intothe nozzle safe-end. As a result of flow-induced vibration,thermal expansion/contraction, or other phenomena many ofthe old design thermal sleeves loosened. The new designthermal sleeves have been hard-rolled into the nozzle with aspecified wall thinning to provide good contact force. To date,the new design thermal sleeves have not experienced anyloosening.

Unique features of the Oconee units’ system designs(relative to other B&W units) are that: (1) normal makeup isfed through two cross-tied makeup lines, (rather than one linein other B&W units) and (2) each makeup line has a “warming”or minimum flow line (WL) to ensure continuous, positivemakeup flow. (Other B&W units employ a bypass valvearound the makeup control valve to ensure a continuousminimum flow.)

The weld failure, safe-end cracking, and thermalsleeve cracking are not unique to the Oconee units. HPI/NMUnozzles have experienced similar failures at other B&W units.

14-4

SYSTEM OPERATIONUnder full power conditions, the cold leg water

temperature is approximately 557°F (292ºC). Normal makeupflow is typically 17 gpm (1.1 l/s) but can vary between 3 and 35gpm (0.2 and 2.2 l/s). Makeup flow is typically 90°F (33ºC).Under emergency conditions, high pressure injection flow isusually in excess of 100 gpm (6.3 l/s) per nozzle andtemperatures can be as low as 40°F (5ºC).

INSPECTION RESULTSAfter safe shutdown of the plant, the nozzle safe-end,

thermal sleeve, and portions of the makeup piping werereplaced. The removed components were examined, visuallyand via metallurgical evaluation.

The weld crack was observed to have initiated at theinside surface. Its circumferential extent was 360 degrees onthe inside surface and 77 degrees on the outside surface, Figure3. It is believed that the cracking had propagated over a longtime period.

Other components in the system also experienceddamage. The region of contact expansion between the sleeveand the safe-end was found to have loosened to the point ofproducing an annular gap of approximately 0.03 inches (1 mm.)The thermal sleeve was also severely damaged. Numerousdeep cracks were observed and a “window” of missing thermalsleeve was identified. There was significant wear of the thermalsleeve on the downstream (RCS) side. Within the safe-end,both outboard of and beneath the thermal sleeve contact-expanded area , severe cracking was observed, with somecracks extending nearly 30 percent through wall. In addition,the makeup piping in the vicinity of the warming line wasfound to have shallow surface cracks or crazing.

TEMPERATURE MEASUREMENTSBecause of concerns relative to thermal fatigue due to

unanticipated thermal stratification in RCS attached pipingsystems (USNRC Bulletin 88-08,Reference 2) temperaturemeasurements were made on the Oconee Unit 1 HPI/NMUlines in 1989-1990. The measured data showed no evidence ofsignificant thermal stratification except during plant heatupsand cooldowns. During some combinations of reactor coolantpump (RCP) operation, hot flow from one cold leg to anothercold leg would occur via the cross-tied makeup piping. Amaximum top-to-bottom pipe temperature difference of 327°F(181.7°C) was recorded in the makeup line adjacent to the idleRCP. This cross-flow occurred due to: (1) the pressuregradient between cold legs during certain partial RCPoperation, and (2) reverse flow through leaking, oversizedcheck valves. Since the number of thermal cycles occurringwith these thermal stratified flow conditions was relativelysmall, this was not perceived, at that time, to be a significantthermal fatigue damage mechanism.

Post-failure (1997 to present) temperaturemeasurements (at the weld and on the piping on both sides ofthe replaced, properly-sized check valves) have not shown thiscross-flow induced thermal stratification since the replacementof the check valves. In fact, current measurements do not showany significant thermal activity. Maximum line temperaturesand top-to-bottom temperature differences have been less than190°F (87.8°C) and 20°F (11.1°C), respectively. However, dueto physical constraints, temperature monitoring cannot revealthermal stratification and cycling within the thermal sleeveitself.

POSTULATED CAUSES OF FAILUREAfter review of the inspection results, temperature

measurements, and other B&W plant operating experiences, thefollowing potential causes of the weld crack and other observedthermal fatigue damage were postulated.

1. Partial RCP operation during plant heat-up/cool-down withresulting back-flow through the HPI/NMU boundary checkvalve of the non-operating RCP cold leg and into theHPI/NMU nozzle of the operating RCP cold leg in thesame loop: It was thought that this “cross-flow” couldhave caused hot RCS water to flow backward from the idlecold leg, into that makeup nozzle, stratify and mix withthat makeup line’s cold warming line flow, and exit theother makeup nozzle, after also stratifying and mixing withthe second makeup line’s warming line flow.

2. Turbulent penetration during full power operation: It wasthought that this mechanism could have caused thermalstratification and cycling in the thermal sleeve and at theweld.

3. In-leakage of hot RCS water through a loose thermalsleeve gap between the thermal sleeve and nozzle ID: Itwas thought that this “annular back-flow” could havecaused hot RCS water to cyclically mix with the coldmakeup flow in the vicinity of the weld.

Cycling mechanisms for the latter two failure causesincluded makeup flow variations, system pressure and flowvariations, and the basic unsteady nature of the turbulent flowregime.

CFD ANALYSISBecause CFD could potentially provide greater insight into thefailure than could be achieved strictly through post-failuretemperature monitoring, three series of CFD analyses,corresponding to each of the postulated failure causes, wereconducted. CFD would provide a continuous temperaturedistribution (especially in areas that could not be measured

14-5

(e.g., within the thermal sleeve)) and would allow parametricevaluations to be performed.

A three-dimensional finite element model of thepiping and nozzle was constructed and the CFD code ANSYSFLOTRAN5.3, Reference 3, was used to calculate the flowpatterns and temperature distributions in the nozzle and thermalsleeve region. The model included of 120,000 finite elementnodes and used a thermal, turbulent, incompressible solution ofthe Navier-Stokes and energy equations. The model consistedof a portion of the cold leg, thermal sleeve, warming line, andmakeup line. The standard K-ε turbulence equations were usedwith the Viollet “buoyancy terms” activated, Reference 4. The steady state results (typically 500 iterations) continued toshow some cyclic variation with additional iterations,suggesting unsteady flow.

CFD RESULTSThe CFD analysis results consisted mainly of water

temperature distributions within the thermal sleeve, nozzle, andmakeup piping.

Series 1 Results - The cross-flow cases modeled the jetpump suction that acts on the end of the thermal sleeve in thecold leg with the operating RCP. This resulted in aredistribution of the flow rate in the warming lines and thepotential for back-flow through a leaking HPI/NMU checkvalve from the loop with the non-operating RCP to theHPI/NMU line of the operating RCP. In the HPI/NMU linewith the idle RCP, the leaking check valve will result in RCSwater being drawn into the thermal sleeve and HPI/NMU linedue to a venturi like effect1. At, and upstream of the warmingline interface, significant thermal gradients can exist as the hotRCS water flows over the top of the colder warming line flow,which enters the bottom of the HPI/NMU line. This can lead tosubstantial thermal stratification in the HPI/NMU line and highthermal gradients in the vicinity of the warming lineconnection, Figure 4. For the HPI/NMU piping with theoperating RCP, the hot flow combines with the colder warmingflow resulting in substantial thermal gradients also in thevicinity of the opposite warming line connection, Figure 5.Thermocouple data recorded in 1989 and 1990 support thisconclusion.

1 The pressure difference between the operating pump and idlepump loop is comprised of a static pressure difference and avelocity pressure difference. The static pressure in theoperating pump loop is greater than the static pressure in theidle pump loop. However, due to the large differences in coldleg velocity between the operating pump loop and the idlepump loop, the aspirating flow effect due to the velocitypressure difference overwhelms the static pressure difference.This causes the makeup flow to be drawn from the idle pumploop to the operating pump loop.

Series 2 Results - The turbulent penetration cases investigatedthe thermal hydraulic conditions for a range of warming lineand normal makeup line flow rates. This series of casesevaluated whether turbulent penetration during full poweroperation alone could result in thermal conditions at the safeend HPI/NMU piping weld that could lead to the Unit 2 failure.The results of those analyses showed that turbulent penetrationeffects produce high thermal gradients within the cantileverportion of the thermal sleeves at normal HPI/NMU flow rates.Normal HPI/NMU flow per nozzle is approximately 17 gpm(1.1 l/s), but can vary between 3 and 35 gpm (0.2 to 2.2 l/s).Only at very low flow rates (< 1 gpm) were significant thermalgradients calculated in the safe end-HPI/NMU line weld regionupstream of the thermal sleeve, Figures 6-8. This indicates thatduring full power operations (with a tight thermal sleeve), thesafe end piping weld is unaffected by turbulent penetration,since it is very unlikely that HPI/NMU flow would be less than1 gpm.

An additional turbulent penetration case was developed todetermine the effect of RCS geometry on the length andorientation of the turbulent penetration. At the attachmentpoint of the HPI nozzle, the RCS cold leg is inclined downwardat 45 degrees to the horizontal. It was thought that thisgeometry might explain the temperature distribution in thesleeve. Turbulent penetration cases conducted within thisinvestigation showed the tendency of the hot water to enter atthe bottom of the thermal sleeve (downstream side), and realignto the top portion of the sleeve as the penetration moved furtherinward. To observe the effect of gravity, an additional case wasmodeled in which gravity was realigned to be exactly oppositethe direction of cold leg flow. Results of this case, Figure 9,demonstrated that hot RCS water again entered the downstreamside, which was now at the top of the thermal sleeve but did nothave to reorient itself as with the previous cases. Thedistribution of the flow and the length of the penetration, in thiscase, compared favorably with previous work by EPRI in theThermal Stratification, Cycling, and Striping program,Reference 5.

Series 3 Results - The annular back-flow cases wereanalyzed similarly to the turbulent penetration cases, exceptthat an additional source of upstream RCS flow was modeledthrough the gap between the thermal sleeve and safe end. Theresults showed that if the annular back-flow is of the samemagnitude as the HPI/NMU flow rate, some hot water couldpropagate upstream (a separate hydraulics analysis calculatedleakage rates of 1 to 12 gpm (0.06 to 0.8 l/s) depending on thethermal sleeve gap height and eccentricity). This regionbetween the warming line and the thermal sleeve is alreadyvery turbulent due to the merging flows from the HPI/NMUline and the warming line. Annular back-flow would increasethe turbulence and further disrupt the flow field. Significanttime-dependent fluctuations of the thermal gradients would be

14-6

expected in this region, which contains the safe end toHPI/NMU pipe weld, Figure 10. These results are especiallyrelevant since the pipe upstream of the safe end exhibited abroad region of multi-directional cracks in the piping basemetal.

CORROBORATION OF RESULTSFLOTRAN’s ability to predict thermally stratified

flows was demonstrated by benchmarking to test data,Reference 6. In addition, following a similar weld failure at theCrystal River Unit 3 plant in 1982, weld temperatures weremeasured as a function of makeup flow. Results showed thateven at the lowest makeup flow rate of 1.5 gpm, no stratifiedflow was observed. This is consistent with the CFD results thatshowed turbulent penetration does not reach the weld until themakeup flow decreases below 1 gpm.

Also, in all cases where weld and safe-end crackinghas occurred, a loose thermal sleeve has been present. Thus,the conclusion that the loose thermal sleeve created conditionsfor high cycle thermal fatigue is supported by plant experiencesand the CFD results.

CONCLUSIONS AND RECOMMENDATIONSThe CFD analyses provided insight into the failures

that would have been difficult, if not impossible, to achievewith post-failure measurements. The CFD analyses confirmedthat no single mechanism appears to be clearly responsible forall of the conditions observed in the Unit 2 nozzle, thermalsleeve, and piping. Since turbulent penetration during normalpower operations caused the largest thermal gradients in thethermal sleeve, this is likely the mechanism responsible for thedamage noted to the thermal sleeve. However, the CFDanalysis demonstrates that turbulent penetration does not leadto significant thermal gradients at the safe end to HPI/NMUline weld except at uncharacteristically low HPI/NMU flows.

Annular back-flow through a loose thermal sleeve canproduce significant thermal gradients in the highly turbulentflow region at the safe end to HPI/NMU line weld and is likelythe primary mechanism leading to the weld crack. However,annular back flow would not typically produce large thermalgradients in the thermal sleeve or HPI/NMU piping upstream ofthe warming line.

During unbalanced partial RCP operation, cross flowcan result in substantial thermal gradients in the vicinity of bothwarming line connections, and appears to be the most likelycandidate responsible for the noted degradation at thoselocations.

To preclude subsequent failures, it is recommendedthat the thermal sleeves undergo regular inspection to ensurethe integrity of the thermal sleeve/nozzle joint (i.e., no

continuous gap that would allow hot RCS water to mix with thecold makeup flow). It is noteworthy that none of the re-designed thermal sleeves have exhibited the looseningexperienced by the old thermal sleeve design. In addition, thethermal sleeve should either be treated as a consumable, havean inspection program implemented to insure its integrity, or beprovided with makeup flow rates sufficient to preclude thermalcycling within the sleeve.

REFERENCES

(1) Shah, V.N., et al, “Assessment of Pressurized WaterReactor Primary System Leaks,” NUREG/CR-6582,December 1998.

(2) USNRC Bulletin 88-08, “Thermal Stresses in PipingConnected to Reactor Coolant Systems.”

(3) FLOTRAN Users Guide, Revision 5.1, DN:S261:511st Revision, Sept. 30, 1994, ANSYS, Inc., Houston,PA. (This Users Guide is also applicable to version5.3)

(4) Viollet, P.L., "Modelling of Turbulent RecirculationFlows," Nuclear Engineering and Design (1987), pp.365-377.

(5) Electric Power Research Institute (EPRI) Report TR-103581, “Thermal Stratification, Cycling, and Striping(TASCS),” March 1994.

(6) Smotrel J.R., "Turbulent Thermal Stratification In ALong Horizontal Pipe," Proceedings of SeventhInternational ANSYS Conference, May, 1996.

15-1

15 CURRENT ACTIVITIES ON GUIDELINES OF HIGH-CYCLE THERMAL FATIGUE IN JAPAN

Takao NakamuraThe Kansai Electric Power Co.,Inc.

Current Activities on Guidelines ofHigh-Cycle Thermal Fatigue in Japan

Takao Nakamura TheKansai Electric Power Co.,Inc.

TEL : +81-3-3591-9261 (ex.256)FAX : +81-3-3593-0586E-mail : [email protected]

International Conference on Fatigue of Reactor Components

31 July-2 August,2000

Silverado Country Club & Conference Center

Napa,California

15-3

Background (1)

• Voluntary countermeasures have been made based ontroubles in domestic and foreign countries– Thermal stratification in Pressurizer Surge line and SG

Feedwater line– Thermal stratification due to Valve leakage– Thermal stratification due to turbulence penetration to the

stagnant line, etc

• Leakage due to thermal fatigue occurred in spite of theefforts

- Mihama-2 excess letdown line in 4/1999 - Tsuruga-2 connecting pipe of regenerative heat exchanger in 7/1999

• MITI required utilities to reinforce countermeasures

15-4

• Additional requirement in MITI ordinance No.62– New requirement ; “Structures and components should be designed

to prevent the failure due to high cycle thermal fatigue.”

• Detailed actions to satisfy the new requirementdescribed in the MITI guideline– Guideline established on the basis of current utilities’ practice

– Voluntary countermeasures turned to regulatory guideline

• More investigation is necessary for phenomena such asCivau-1 case and turbulence penetration

• New group organized in JSME to establish voluntarydesign standards to be quoted to MITI ordinance

Background (2)15-5

Phenomena Example of Experienced Plant

Mihama-2 in 1999

Farley-2 in 1987

Genkai-1 in 1988

Surge Trojan in 1988

Feedwater flow D.C.Cook-2 in 1979

Thermalstriping

Civaux-2 in 1998

Operation

Mixing of hot and cold water

Thermalstratification

Cause

Turbulence penetration

Leakage from valve seat

Leakage from valve grand

High-cycle Thermal Fatigue Phenomena

15-6

JSME Activities• Utilities - vendors joint research for PWR and BWR

– 2 year program starting from April 2000

– Includes hydraulic tests and analytical investigation

– Detailed program being developed

• Scope (Phenomena to be considered)– Thermal fatigue due to temperature oscillation in a

mixing zone of high and low temperature water

– Thermal fatigue due to oscillation of thermalstratification caused by turbulence penetration flow tothe stagnant pipe

• Establishment of design standards by utilizing datafrom the joint research and actual plants

15-7

Schedule for Establishment of High-cycle Thermal Fatigue Evaluation Standards2000 2001 2002

1. Thermal Striping(1) Investigation of past studies

(2) Investigation of types of confluence in actual plants(Investigation of points to be evaluation)

(3) Planning Test , Fabrication of test loop and device

(4) Thermal striping test-Measurement of temperature fluctuation

(5) -Thermal hydraulic analysis

(6) Development of evaluation method

2. Thermal Stratification in stagnant pipe(1) Investigation of past studies

(2)Investigation of configulations of stagnant part and operatingconditions in actual plants (Investigation of points to be evaluation)

(3) Planning Test , Fabrication of test loop and device

(4)Thermal stratification test in stagnant pipe-Observation of themal stratification by the test-Study correlation between major flow parameters and the penetration depth

(5) Development of evaluation method

3.Development of high-cycle thermal fatigueevaluation standard

15-8

Investigation of past studiesInvestigation of points to be evaluation(configurations, operating conditions)

Clarification of factors to cause temperature fluctuationStudy of the scope and structure for standard

Thermal striping test Thermal stratification testin stagnant pipe

Develop evaluation method Develop evaluation method

Preliminary evaluationof actual components

Preliminary evaluationof actual components

Results of actual plantcomponent inspectionfor validation

Development of the draft high-cycle thermal fatigue evaluation guideline

Guideline for Evaluation of High-cycle Thermal Fatigue R&D Efforts

Test plan

Measurement oftemperature fluctuation

Thermal hydraulicanalysis

• Develop evaluation method of the distribution of temperature and stress• Develop evaluation charts

Study correlation between major flow parametersand characteristics of temperature fluctuation

Test plan

Measurement of thermal stratificationby the test

Study correlation between major flowparameters and the penetration depth

15-9

BWR

Thermal Striping Test (1)Investigation of Types of Confluence in Actual Plants

-Confluence type

-Pipe Diameter (D1,D2)

-Flow Rate (U1,U2)

-Temperature (T1,T2)

-Upstream Turbulence

(valve, elbow, diffuser)

Basic 3 types

PWRSame Diameter,

Collision

Same Diameter,

Confluence

Different Diameter,

Confluence

(D1,U1,T1) (D1,U2,T2)

(D1,U1,T1)

(D1,U1,T1)

(D1,U2,T2)

(D2,U2,T2)

15-10

Visual Test Fluid Temperature Fluctuation Test

Pipe mat. Acrylic resin Metal (SUS)

model

(D1,U1,T1) (D1,U2,T2)

(D1,U2,T2)

(D1,U1,T1)

(D2,U2,T2)

(D1,U1,T1)

Type A Type B Type C

TestParameter

Flow Rate Ratio: U2/U1 , Main Stream Flow Rate : U1Diameter Ratio: D2/D1(Type-C) , Upstream Turbulence (Valve,Elbow,Diffuser) , etc.

(Constant) Temperature Difference T and Main stream pipe Diameter D1

Measurement Fluid Temperature , Metal Temperature

EvaluationConfirmation of Flow Condition,Fluid Temperature FluctuationProfile

Amplitude and Period of Fluid TemperatureFluctuationAmplitude and Period of Pipe WallTemperature Fluctuation

Thermal Striping Test (2)Test Plan

15-11

Draft Guideline for Evaluation of Thermal Stratification in Stagnant Pipe

Length to Horizontal Part : L/d, Branch Pipe Diameter : d, Main Pipe Diameter : D Main Flow rate : V, Main Flow Temperature : T, Horizontal Part Length : l, etc.

T Tcr ?

L/d L1/d ?

L/d L2/d ?

Individual Detail EvaluationOK

Determination of L1

Determination of L2

(Parameter) V, d/D, T, (density),

(kinematic viscosity),

etc.

(Parameter) V, d/D, T, l , etc.

L1:

d [ L/d L2/d ]

[ L/d L2/d ]

L/d l

Major pipe

L/d

Design Condition

L2 is maximum length when penetrationflow enough reach into horizontal part .

Yes

Yes

Yes

No

No

No

Penetrationdepth ofcavity-flow

Tcr [by calculating] is maximum temperaturethat stress amplitude by thermal stratificationis less than fatigue limit in whole pipe region.

T Tcr a lim

15-12

Thermal Stratification Test in Stagnant Pipe

1. Investigation of Branch types and its Flow Conditions in Actual Plants

2. Planning Test Matrix Parameter: V, T, D, d, L, l (horizontal part length) , Branch type, Upstream elbow, , ,etc.

3. Thermal Stratification Test (1) Visual Test (acrylic pipe) Case1 : main pipe temperature T1=branch pipe temperature T2 =RT Case2 : T2-T1= about 40

(2)Actual Temperature and Pressure Test (SUS pipe)

Evaluation of L1 and L2

15-13

Summary

- More investigation is necessary to evaluate high-cycle

thermal fatigue quantitatively

- Data (from researches including tests and calculations )

will be collected by the Japanese joint research in 2 years

- JSME design standard will be established using collected Data

- International information exchanging is highly expected

15-14

Design condition : Diameter ratio D2/D1 Temperature difference ∆tin

Velocity ratio U2/U1 Upstream elbow/valve

∆Tin < ∆Tcr ?

Consideration to attenuation by mixing Chart I : AF (attenuation factor) ∆Tf = ∆Tin AF

∆Tf mod = ∆Tf Mod. Factor for upstream elbow/valve

∆Tf mod < ∆Tcr ?

Yes

No

Yes

No

∆Tcr : Critical temperature difference

Heat transfer coefficient h in steady flow

Consideration to heat transfer enhancement in unsteady condition Chart II : EF (enhancement factor)

Structural calculation assuming most effective frequency Chart III : ∆σ*

σalt = 1/2 Kt ∆σ* Eα∆Tf mod / (1-ν)

σalt < σlim ?Yes

No

(σlim : Fatigue limit)

Consideration to spectrum of ∆Tf using Chart IV : Power spectrum of ∆Tf and Chart III

Structural calculation considering power spectrum of ∆Tf i : i-th frequency component exceeding fatigue limit∑ ∆

×=i i

ifUFσfor repetition allowable

timeoperation

UF < 1 ?

Acceptable design Yes

RedesignNo

Draft Guideline forEvaluation of Thermalstriping

15-15

Nondimensional axial distance

AF

(at

tenu

atio

n fa

ctor

)1.0

0.5

0.00 5 10

U2/U1 = 10

1

0.1

Chart I : Attenuation of fluid temperaturefluctuation by mixing

Chart II : Heat transfer enhancement in unsteady condition

Reynolds number

Enh

ance

men

t fac

tor

1

2

3

U2/U1 = 10

U2/U1 = 1

U2/U1 = 0.1

0.1 1.0 10.0 100.0f *

1.0

0.5

0.0

∆σ *

Bi = 10

5

1

Chart III : Thermal stress assumingmost effective frequency

Strouhal number.001 .1 10.

Pow

er s

pect

rum

den

sity

of

tem

pera

ture

flu

ctua

tion

Chart IV : Power spectrum densityof temperature fluctuation

Draft Guideline for Evaluation of Thermal striping15-16

FATIGUE MONITORING/EVALUATION

16-1

16 STATUS AND UPDATE OF THE EPRI FATIGUEPROFATIGUE MONITORING PROGRAM

Stan T. RosinskiEPRI

1300 Harris BoulevardCharlotte, North Carolina 28262

Gary L. StevensArthur F. Deardorff


San Jose, CA 95118-1557

16-3

STATUS AND UPDATE OF THE EPRI FATIGUEPROFATIGUE MONITORING PROGRAM


1300 Harris BoulevardCharlotte, North Carolina 28262

Gary L. StevensArthur F. Deardorff


San Jose, CA 95118-1557

ABSTRACT

FatiguePro has been developed as an advanced tool for tracking cycles and fatigue usage atkey reactor coolant pressure boundary components. Using data from the plant process computer,FatiguePro can logically determine the types of cycles occurring and can compute stress- timehistories for fatigue-sensitive locations. Initially developed in 1986, many new features haverecently been incorporated. Continued user input is collected through the ongoing FatigueProUser’s Group to enhance the software and to ensure that it maintains its usability with changingcomputer technology. This paper summarizes the capabilities of FatiguePro, how it is beingused at operating nuclear plants and the continued utility support to keep the software up to date.

INTRODUCTION

Fatigue usage accumulation resulting from plant transient operation contributes to aging ofcritical equipment in light water reactor (LWR) nuclear power plants. During the design ofmodern nuclear plants, the effects of stress and fatigue were evaluated and bounded using designrules contained in Section III of the ASME Boiler and Pressure Vessel Code [1] and/orUSAS B31.7 [2]. End-of-life cumulative usage factors were determined, in accordance withthese rules, showing that fatigue usage factors were less than unity. This fatigue assessment wasbased on an assumed conservative set of design transients to assure that components would notexceed the allowable cumulative usage factor of one throughout their lifetime (usually fortyyears).

To assure that design safety margins remain adequate throughout the operating life of the plant,each plant's license generally included requirements to count cycles or otherwise demonstratethat actual operating experience remains bounded by that assumed in the original plant design.Typically, significant design transient operating cycles are logged and counted to assure that thedesign fatigue limits are not exceeded. In practice, however, many of the actual plant operatingcycles are not well characterized by the design transients and, in fact, are often much less severe.Classification of individual plant events into one of the design transient categories is also a

16-4

difficult task, for which plant operators are given relatively little guidance. Further, many of theevents included in the design analysis had only a minor contribution to fatigue usage. As a result,some operating plants have approached the limit for the number of analyzed design transientsearly in plant life. In other cases, plant operating cycles may have been classified incorrectly orinconsistently, resulting in a poor estimate of cumulative usage accumulation. Finally, therehave been occurrences of loadings not considered in the original design basis (e.g., thermalstratification), that have a significant contribution to usage factors when considered in a designanalysis.

In most cases, the design transients very conservatively bound plant operation. In other cases,transients have occurred that exceeded the design basis. In the early years of nuclear plantoperation, there was no practical means by which plant operators could easily evaluate theeffects of the difference between the transients actually occurring and those considered in theplant design. In 1985, EPRI initiated a project to develop a prototype system for monitoringcumulative usage factors in nuclear power plant components. In this project, a methodology wasdeveloped for accessing plant instrument data and converting this directly into peak stress versustime at plant locations of interest. The key to this technique was a transfer function approach,using Green’s Functions and transfer matrices to convert plant data to peak stress versus timebased on the actual plant data [3]. Another innovation was implementation of the ordered overallrange (OOR) cycle counting methodology to account for actual cycle sequence, overcoming thelimitation in the ASME Code which considers the most conservative possible cycle sequence.The methodology was developed into a specialized software system called FatiguePro , withdemonstrations at two plants [4,5].

Through continued plant applications and improvements to the system, many additional featureshave been included to more fully address plant cycle tracking requirements. To date, thefollowing features have been incorporated into the FatiguePro software:

• Acquisition of plant operating data (from existing plant instrumentation)• Calculation of peak stress versus time and cumulative usage factor for fatigue-critical plant

locations• Automated cycle counting• Computation of cumulative usage factors at fatigue-critical components using a cycle-based

fatigue approach• Prediction of growth of actual or hypothetical cracks• Real time analysis and detailed review capabilities of all input plant data and results• Summary reporting capability for permanent documentation purposes

This paper provides an update on the FatiguePro software and describes how it can be used tofulfill cyclic tracking and/or other margin assessment requirements to demonstrate that fatigue isbeing properly managed.

FATIGUEPRO LIFE ASSESSMENT APPROACH

Typically, nuclear plants are required to track plant transients against cycle limits. Theserequirements are specified in the plant licensing bases and/or Technical Specifications. The

16-5

intent of these requirements is to ensure that actual plant operation remains within the envelopeassumed in the design basis. When properly implemented, these requirements are consistentwith the recommended compliance of Generic Safety Issue 78 [6] and other regulatorydocuments.

In many cases, plants have implemented manual or computer-based transient monitoring systemsto fulfill plant-specific requirements for tracking cycle accumulation. The intent of theFatiguePro software development was to provide an industry-approved tool that could be usedby plant engineers to fulfill plant fatigue life tracking requirements. In addition, the softwareprovides the capability to perform a life assessment based on actual plant data. This lifeassessment may be based on any combination of the three following approaches:

• Counting, categorizing and tracking plant transient events, and comparing the result to thenumber of cycles assumed in the design basis.

• Computing cumulative usage factors, based on either real-time stress time history predictionor on a cycle-based approach, and demonstrating that usage factors less than unity aremaintained for all monitored locations.

• Using a Section XI, Appendix L flaw tolerance approach [7] based on real-time plant data todemonstrate that actual or postulated flaws remain within acceptable values.

All of these approaches are intended to demonstrate that Code margins for fatigue-sensitivecomponents are maintained during actual plant operation. This approach generally demonstratessignificantly more margin than is shown in component stress reports, since actual transients aretypically less severe than those assumed in the original design.

The applicability of EPRI’s fatigue monitoring methodology was demonstrated throughimplementation and field testing of a prototype system at San Onofre Unit 2, a pressurized waterreactor (PWR) [4]. A similar installation and field test was also performed at Quad Cities Unit 2,a boiling water reactor (BWR) [5].

DEVELOPMENTS

FatiguePro was developed and first applied at nuclear plants in the late 1980’s. At that time,personal computers were much less powerful than today, and the Windows operating system wasin its early stages of development. Even with this lack of computing power, the methodologybuilt into FatiguePro was quite efficient at utilizing and logically evaluating plant instrumentdata to predict local temperatures and loading conditions at fatigue-sensitive locations. Ascomputer power has increased, newer capabilities have been added to the FatiguePro softwareshell and into plant unique applications. Some of the significant advances include the following:

• Modern Windows Interface. The software interface now allows the user to perform anevaluation of plant data to determine if there are data errors or missing data that might affectthe results. The plant data is archived in special compressed data files that may be re-evaluated at a later time. The current fatigue status is available in a set of standard reports.

• Automated Cycle Counting. Whereas the original FatiguePro software predicted fatigueusage only based on predicted stress time history, the current version also has an automated

16-6

cycle counting module. The software logically evaluates actual plant data and counts design-basis cycles, such as heatup, cooldown, plant trips, etc. The key characteristics of each eventare saved in the records, and/or are used to predict a fatigue usage contribution. Thus, plantdata can be evaluated to show that actual transients are less severe and contribute to a lowerfatigue accumulation than predicted in the original stress analysis.

• Piping Transient Temperature Response Modeling. In many cases, the determination oftemperatures in a piping system must be based on remote instrumentation. In addition, whenthere is no flow, the piping system will slowly cool to ambient conditions or will equalizewith the temperature of a connected system. A standard model has been implemented inFatiguePro that will properly account for these situations. This more-exact modeling offluid systems allows for better prediction of local thermal stresses that are a major contributorto fatigue usage accumulation [8].

• Environmental Effects. Although not yet implemented into the standard FatigueProsoftware, a special version of the software was developed to assess the effects of reactorwater on fatigue usage of selected components. This required that local temperature effectsand strain rate be evaluated in the fatigue analysis [9,10,11].

FATIGUEPRO USER'S GROUP (FPUG)

Following the initial FatiguePro development in 1986 and subsequent evolution to FatigueProversion 2.0 in 1997, there was a desire on the part of many utility users to form a group thatwould direct future enhancements of the software. This group could also provide a forum forsharing, solving and responding to industry fatigue-related issues. In particular, a way todisseminate all of the lessons learned by the large population of users was desired. To respond tothese desires, EPRI formed the FatiguePro User's Group (FPUG) in 1998. The FPUG has thefollowing objectives:

• Provide methods for management of fatigue margins at member utility nuclear plants,including the use of EPRI's FatiguePro software.

• Develop updated FatiguePro software that is compatible with evolving computerinfrastructure and operating environments.

• Enhancing member utility fatigue management capabilities through networking, exchangingof ideas and experiences, and presentation of topics in the area of fatigue.

Membership is composed of one utility representative for each subscribing plant site. The FPUGis initially chartered for four years and includes member-directed FatiguePro softwaredevelopment and enhancement. Software development and enhancement will include theimplementation of three major architectural changes to FatiguePro, thereby “modernizing” thesoftware, and allowing for more efficient code maintenance over subsequent years. Theseinclude:

• Implementation of a 32-bit design which will operate under current computer operatingsystems and use the improved interface features standard to these operating systems.

• Implementation of a full-featured database engine for storing configuration data and programresults using standard SQL format.

16-7

• Implementation of integrated data review and graphics modules such that reliance on otherexternal software will no longer be needed to perform these functions.

The culmination of this effort will be the next generation of the FatiguePro software (Version3.0). Semi-annual meetings linked with annual fatigue seminars are planned for the duration ofthe FPUG term to enhance member participation.

The intent of the FatiguePro software development is to provide an industry-approved tool thatcan be used by plant engineers to fulfill plant cyclic duty tracking requirements by using anycombination of approaches. All of the approaches used by FatiguePro are intended todemonstrate, in an accurate, reliable, and retrievable fashion, that structural design margins forall critical components are maintained during actual plant operation. Therefore, FatigueProprovides the technical tools that may be used to assure that structural design margins aremaintained.

TYPICAL USE OF SOFTWARE

Reactor coolant system components in nuclear plants are designed and analyzed in accordancewith ASME Section III Class 1 code design rules or similar rules for older plants. The designprocess involves analyzing each applicable component for a set of design transients that boundthe expected plant operation in terms of temperature and pressure profiles for the life of theplant. Each component is analyzed to meet all of the applicable design requirements, includinglimits on primary and secondary stresses and cyclic duty limits. During the license period of theplant, the plant operator is responsible to ensure that plant components remain within thelicensing basis. The plant Safety Analysis Report (SAR) usually specifies the set of designtransients that initially define the plant licensing basis.

Requirements for tracking plant operations stem from the necessity for the plant owner todemonstrate that the plant operates within the licensing basis. Excursions outside the boundariesof the licensing basis require special analysis to demonstrate that structural design margins aremaintained over the operating life of the plant, thus ensuring continued safe and reliableoperation.

Early in the life of most present-day operating plants, the most direct and easiest form ofensuring that operation remained within the licensing basis was considered to be counting andcategorization of plant transient events. Such an approach was considered straightforward, andeliminated the need to re-perform costly, labor-intensive fatigue evaluations based on plant-unique operating history. Later in plant life, however, many plants may experience certain plantevents in quantities that either exceed the number assumed in the licensing basis, or accumulateat a rate that is projected to exceed the number assumed in the licensing basis prior to the end ofthe desired operating period. In addition, events or loads may be experienced that were notconsidered in the original design. In these cases, simple cycle counting is not sufficient todemonstrate acceptable design margins. In order to demonstrate acceptable design margins inthese instances, the plant owner has several options. First, each component can be re-analyzed toa revised set of design transients that represent the observed operation of the plant in terms ofnumber and severity of events. Alternatively, the plant owner can initiate a condition

16-8

assessment program that accounts for the actual operation of the plant, and addresses the effectof actual operation on the structural margin of the affected components. These approaches areconsistent with those recommended in ASME Code, Section XI, Nonmandatory Appendix L [7],as depicted in Figure 1.

A few selected fatigue-critical components can be used to bound the fatigue behavior of entirecomponents or systems by virtue of the most severe loadings and/or highest stressconcentrations. Thus, a detailed fatigue analysis of these bounding components using actualplant transient temperatures and pressures substitutes for a complete design basis analysis of anentire component or system.

The FatiguePro software is intended to provide an industry-approved tool that can be usedwithin an integrated management program to show that design safety margins are maintained.FatiguePro can be used to fulfill plant cyclic duty tracking requirements, perform componentstructural margin evaluations, and accommodate flaw tolerance evaluations of components withcumulative usage factor values that exceed allowable limits. FatiguePro can be implemented(i.e., installed) at any time during the operating lifetime of a power plant. Through a"baselining" procedure that can be performed as a part of installation, monitored componentfatigue duty can be determined from the start of plant operation up to the point of installation.This duty can be reflected in FatiguePro, and "on-line" monitoring can be performed from thatpoint forward. Trending can be used to estimate the remaining life expectancy of the monitoredcomponents.

To accomplish condition assessment, FatiguePro incorporates technical capabilities that addressthe following three requirements:

• Plant Licensing Basis Cycle Counting. This requirement is accomplished by consistently andaccurately counting, categorizing, and tracking plant transient events for comparison to theevents assumed in the licensing basis. This activity provides a direct measure that plantcycles remain within cyclic limit requirements, and an indirect measure of structural designmargin. Automatically recording the occurrence of plant cycles may eliminate the need forplant operating personnel to do so manually, and reduces the inaccuracies and unnecessaryconservatism inherent in manual cycle counting.

• Determine Actual Fatigue Margins. This determination is accomplished by computingcumulative usage factors based on actual plant operation, and demonstrating that these valuesremain less than the design allowable for all monitored components. This method goes onestep further than cycle counting alone, in that it provides a direct measure of fatigue designmargin. In addition, assessing the fatigue status of the reactor coolant pressure boundary canbe readily accomplished subsequent to an event that exceeds the number of allowable cyclesspecified in the plant design basis.

• Demonstrate Actual or Postulated Flaws Remain within Allowable Limits. Components inthe plant that are expected to accumulate abnormally high cumulative usage factors may beclosely monitored utilizing a fatigue crack growth fracture mechanics methodology. Plantowners can adjust plant operational procedures and inservice inspection programsaccordingly to ensure that design structural margins are maintained.

16-9

All of these approaches are intended to demonstrate, in an accurate, reliable, and retrievablefashion, that structural design margins for all fatigue-critical components are maintained duringactual plant operation. FatiguePro provides the technical tools that may be used to assure thatstructural design margins are maintained in accordance with the approach appropriately chosenby the plant owners.

CONCLUSIONS

FatiguePro addresses the need for assuring plant design safety margins by providing thefollowing capabilities:

• Automatically and reliably records the occurrence of plant thermal cycles.• Determines actual fatigue margins.• Assesses the structural integrity of the reactor coolant pressure boundary after an event that

exceeds the operating pressure and temperature limits or number of allowable cycles, asidentified in the plant Technical Specifications.

• Closely monitors areas of the plant that are expected to accumulate abnormal cumulativeusage factors, enabling reactor operators to adjust plant operational procedures and inserviceinspection programs accordingly.

Pilot plant studies have been performed using FatiguePro for lead BWR and PWR plants [4, 5,12]. Each of these pilot projects included plant-specific application of the FatiguePro software,as well as detailed evaluation of several years’ worth of plant data. The objective of theseevaluations was to rigorously test the FatiguePro methodology and provide further confidencein the plant-specific application of the software and its ability to fulfill plant cyclic duty trackingrequirements. In addition, fatigue duty extrapolation schemes were developed for generic use.These methodologies estimate cumulative usage factors in instances where plant data are notavailable (i.e., as in the case of establishing the cumulative usage factor at a time when data arenot retrievable). These studies concluded that fatigue monitoring for a few selected fatigue-critical components is a technically acceptable alternative to the cycle counting requirementscontained in most plant technical specifications, and can be used to fulfill the relatedrequirements associated with the plant licensing basis.

Since the original development of FatiguePro, there have been continuing projects at individualplants to upgrade fatigue monitoring programs. Table 1 reflects the plants with FatigueProinstalled or under development.

All of the approaches used by FatiguePro are intended to demonstrate, in an accurate, reliable,and retrievable fashion, that structural design margins for all critical components are maintainedduring actual plant operation. Therefore, FatiguePro provides the technical tools that may beused to assure that structural design margins are maintained.

16-10

REFERENCES

1. ASME Boiler & Pressure Vessel Code, Section III, “Rules for Construction of NuclearPower Plant Components,” Division I, Subsection NB, “Class 1 Components,” 1989 Edition(or earlier editions).

2. USAS B31.7, “Nuclear Power Piping”, 1969.3. A. Y. Kuo, S. S. Tang, and P. C. Riccardella, "An On-line Monitoring System for Power

Plants: Part 1 – Direct Calculation of Transient Peak Stress Through Transfer Matrices andGreen's Functions," Proceedings, 1986 Pressure Vessels and Piping Conference andExhibition, PVP-Vol. 112, pp. 25-32, ASME, Chicago, IL, July 1986.

4. EPRI Report NP-5835, “FatiguePro: On-Line Fatigue Usage Transient MonitoringSystem,” May 1988.

5. EPRI Report NP-6170-M, “FatiguePro On-Line Fatigue Monitoring System:Demonstration at the Quad Cities BWR,” January 1989.

6. NUREG-0933, “A Prioritization of Generic Safety Issues,” U. S. Nuclear RegulatoryCommission, April 1999.

7. ASME Boiler & Pressure Vessel Code, Section XI, Nonmandatory Appendix L, “OperatingPlant Fatigue Assessment,” 1995 Edition.

8. Stevens, G., Gerber, D. and Rosinski, S., “Latest Advances in Fatigue MonitoringTechnology Using EPRI's FatiguePro Software,” in Proceedings, 15th InternationalConference on Structural Mechanics in Reactor Technology, August 15-20, 1999,Seoul, Korea..

9. EPRI Report TR-110356, “Evaluation of Environmental Thermal Fatigue on SelectedComponents in a BWR Plant,” EPRI, Palo Alto, California, April 1998.

10. EPRI Report TR-110043, "Evaluation of Environmental Fatigue Effects for a WestinghouseNuclear Power Plant, EPRI, Palo Alto, California, April 1998.

11. EPRI Report TR-107515, “Evaluation of Thermal Fatigue Effects on Systems on SystemsRequiring Aging Management Review for License Renewal for the Calvert Cliffs NuclearPower Plant,” EPRI, Palo Alto, California, December 1997.

12. EPRI Report TR-107448, “FatiguePro, Version 2: Fatigue Monitoring Software,”December 1997.

16-11

Table 1. FatiguePro Installations

PWR FATIGUEPRO PROJECTS

Plant NSSSVendor

FatigueAnalysis

CycleCounting

Fatigue CrackGrowth

ANO-2 C-E X XCalvert Cliffs 1/2 C-E X X XSan Onofre 2/3 C-E X

Waterford 3 C-E XFt. Calhoun C-E XMillstone 3 W X

Maanshan 1/2 (Taiwan) W X XDiablo Canyon1/2 W X X X

Sequoyah 1/2 W X X XPrairie Island 1/2 W XPoint Beach 1/2 W X X

Vogtle 1/2 W X XCallaway W X X

Wolf Creek W X XMcGuire 1/2 W XCatawba 1/2 W X

Vandellos (Spain) W X XSurry 1/2 * W

North Anna 1/2 * WKewaunee * W

ANO-1 B&W XOconee 1/2/3 B&W X

Crystal River 3 B&W X*recommendation report completed or in progress

BWR FATIGUEPRO PROJECTS

Plant FatigueAnalysis

CycleCounting

Fatigue CrackGrowth

Quad Cities 2 (Demonstration) XOyster Creek X X

Susquehanna 1/2 X XKuosheng 1/2 (Taiwan) X X

WNP-2 X XBrunswick 1/2 X

Chin-Shan 1/2 (Taiwan) X XRiver Bend X XGrand Gulf X X

Toshiba Plants (Japan) XSanta Maria de Garoña (Spain) X

Cofrentes (Spain) XFitzPatrick *

Clinton X X*recommendation report completed or in progress

16-12

Typical Remaining Life Assessment

Model Component

Define Loading / Collect Operating Data

Stress AnalysisFE or Closed Form

Damage Model

Crack Growth Model

Σ Damagevs. Time

Σ Crack Growth vs. Time

Remaining Life Prediction

Material Properties

Inspection Data

Repeatfor New

Operating Data

95093r0

Figure 1. Typical Remaining Life Assessment

Status and Update of the EPRI FatigueProFatigue Monitoring Program


Gary L. Stevens Arthur F. DeardorffStructural Integrity Associates


31 July - 2 August 2000

Napa, California

16-13

Introduction

● Fatigue accumulation is an aging concern foroperating nuclear plants

● Set of design cyclic conditions specified for plant life

● Plant duty tracking requirements typically specifiedin Technical Specifications or FSAR

● Tracking is difficult■ Actual transients are often different from (and less severe

than) assumed transients

■ Classification performed inconsistently by differentengineers

● Tracking does not account for implicit margin

● In response to these issues, EPRI initiatedFatiguePro project in 1985

16-14

Introduction

● FatiguePro development initiated in 1985■ Methodology developed for accessing plant instrument data

and converting directly into peak stress versus time

◆ Transfer function approach

■ Implementation of ordered overall range (OOR) cyclecounting methodology to account for actual cycle sequences

■ Specialized software system developed and demonstratedat San Onofre (1988) and Quad Cities (1989)

16-15

FatiguePro Features

● Features incorporated into FatiguePro■ Acquisition of plant data

■ Stress peaks/valleys computed for stress-based usagefactors

■ Automated cycle counting for cycle-based usage factors

■ Prediction of growth of actual or hypothetical cracks

■ Real time analysis and detailed review

■ Comprehensive documentation

16-16

Fatigue Monitoring Using FatiguePro

● When properly implemented a fatigue monitoringprogram complies with requirements specified invarious regulatory documents

■ Provides knowledge of fatigue duty

■ Demonstrates margins for continued operation

■ Provides valuable insight to guide plant operation

■ Identifies previously unknown loadings

■ Addresses license renewal concerns

● FatiguePro developed to provide integratedassessment as part of overall fatigue management

■ Assessment based on actual plant data

16-17

FatiguePro Life Assessment Approach

● FatiguePro can be integratedinto component lifeassessment strategies

■ Consistent with ASME SectionXI, Appendix L

■ Assure design margins aremaintained

● Extensive industry utilizationof FatiguePro as part ofoverall fatigue managementprogram

Typical Remaining Life Assessment

Model Component

Define Loading / Collect Operating Data

Stress AnalysisFE or Closed Form

Damage Model

Crack Growth Model

Σ Damagevs. Time

Σ Crack Growth vs. Time

Remaining Life Prediction

Material Properties

Inspection Data

Repeatfor New

Operating Data

95093r0

16-18

PWR FatiguePro Installations

PWR FATIGUEPRO PROJECTS

PlantNSSS

VendorFatigueAnalysis

CycleCounting

Fatigue CrackGrowth

ANO-2 C-E X XCalvert Cliffs 1/2 C-E X X XSan Onofre 2/3 C-E X

Waterford 3 C-E XFt. Calhoun C-E XMillstone 3 W X

Maanshan 1/2 (Taiwan) W X XDiablo Canyon1/2 W X X X

Sequoyah 1/2 W X X XPrairie Island 1/2 W XPoint Beach 1/2 W X X

Vogtle 1/2 W X XCallaway W X X

Wolf Creek W X XMcGuire 1/2 W XCatawba 1/2 W X

Vandellos (Spain) W X XSurry 1/2 * W

North Anna 1/2 * WKewaunee * W

ANO-1 B&W XOconee 1/2/3 B&W X

Crystal River 3 B&W X*recommendation report completed or in progress

16-19

BWR FatiguePro Installations

BWR FATIGUEPRO PROJECTS

PlantFatigueAnalysis

CycleCounting

Fatigue CrackGrowth

Quad Cities 2 (Demonstration) XOyster Creek X X

Susquehanna 1/2 X XKuosheng 1/2 (Taiwan) X X

WNP-2 X XBrunswick 1/2 X

Chin-Shan 1/2 (Taiwan) X XRiver Bend X XGrand Gulf X X

Toshiba Plants (Japan) XSanta Maria de Garoña (Spain) X

Cofrentes (Spain) XFitzPatrick *

Clinton X X*recommendation report completed or in progress

16-20

FatiguePro Update

● Significant recent advances to FatiguePro■ Incorporation of sophisticated thermal-hydraulic models

◆ Eliminate unrealistic “steps” associated with designtransients

◆ Better prediction of local thermal stresses

■ Incorporation of fracture mechanics solutions for fatiguecrack growth assessments

■ Development and implementation of environmentalfatigue algorithms

◆ Not yet included into standard FatiguePro

◆ Prototype developed for scoping studies

■ Formation of a FatiguePro User’s Group

16-21

FatiguePro User’s Group

● FPUG formed in 1998 following development ofFatiguePro version 2.0 (1997)

● FPUG Objectives■ Provide methods for management of fatigue margins at

member utility nuclear plants, including the use of EPRI'sFatiguePro software.

■ Develop updated FatiguePro software that is compatiblewith evolving computer infrastructure and operatingenvironments.

■ Enhancing member utility fatigue managementcapabilities through networking, exchanging of ideas andexperiences, and presentation of topics in the area offatigue

16-22

FatiguePro User’s Group

● FPUG membership consists of 23 domestic/3international plant sites

● FPUG directing enhancement of FatiguePro■ Implementation of a 32-bit design

■ Implementation of a full-featured database engine forstoring configuration data and program results usingstandard SQL format

■ Implementation of integrated data review and graphicsmodules such that reliance on other external software willno longer be needed to perform these functions

■ Extensive user-interface enhancements

● Next generation FatiguePro (version 3.0) to bereleased December 200116-23

Conclusions

● FatiguePro addresses the need for assuring plantdesign safety margins by providing the followingcapabilities

■ Automatically and reliably records the occurrence of plantthermal cycles

■ Determines actual fatigue margins

■ Assesses structural integrity of the reactor coolant pressureboundary after an event that exceeds the operatingpressure and temperature limits or number of allowablecycles, as identified in the plant Technical Specifications.

■ Closely monitors areas of the plant that are expected toaccumulate abnormal cumulative usage factors, enablingreactor operators to adjust plant operational proceduresand inservice inspection programs accordingly

16-24

Conclusions

● Approaches in FatiguePro are intended todemonstrate, in an accurate, reliable, andretrievable fashion, that structural design marginsfor all critical components are maintained

● FatiguePro is an effective tool that may be used toassure that structural design margins aremaintained

16-25

17-1

17 REMARKS TO THE DIFFERENT FACTORSINFLUENCING FATIGUE ANALYSIS AND FATIGUEDESIGN CURVES

E. RoosK.-H. Herter

S. IsslerStaatliche Materialpruefungsanstalt (MPA)

University of StuttgartPfaffenwaldring 32

70569 Stuttgart, Germany

17-3

REMARKS TO THE DIFFERENT FACTORS INFLUENCING FATIGUEANALYSIS AND FATIGUE DESIGN CURVES

E. RoosK.-H. Herter

S. IsslerStaatliche Materialpruefungsanstalt (MPA)

University of StuttgartPfaffenwaldring 32

70569 Stuttgart, Germany

Abstract

Technical codes used for construction, design and operation of nuclear components and systemsprovide detailed stress analysis procedures, materials data and a design philosophy which guaran-tees a reliable behaviour of the structural components throughout the lifetime. Especially for c y-clic stress evaluation the different codes provides fatigue analyses to be performed consideringvarious loading histories and geometric complexities of the components. In order to fully unde r-stand the background of fatigue analysis included in the codes as well as the fatigue design curvesused as a limiting criteria (fatigue life usage factor), it is important to understand the history andthe methodologies which are available for the design engineers.

Using design by analysis in the codes a simplified elastic plastic fatigue analysis is recommendedwhen the range of primary plus secondary stress intensity exceeds the mS3 ⋅ limit. For that casethe fictitious alternating stress amplitude aS is calculated by multiplying the range of primary plussecondary plus peak stress intensity nS with the stress dependent plastification factor eK . In nu-clear and non nuclear codes different methods calculating eK values are available. The safetymargins of this simplified elastic plastic fatigue analysis have been studied by using experimentaland numerical results.

The design fatigue curves in the nuclear codes are based on uniaxial cycling failure data. A factorof 2 on stress and a factor of 20 on cycles, whichever is more conservative at each point, was a p-plied to the best fit curve. Nowadays numerous experimental data is available to check the con-servatism of the fatigue design curves concerning the influence of welds, environment, surfacefinish, loading, temperature, mean stress and size. The effects of the different factors influencingthe fatigue analysis and the fatigue design curves will be discussed.

17-4

1 Introduction

Technical codes and standards like ASME-Code Section III [1], French RCC-M Code [2], BritishStandard BS 5500 [3] or German KTA Safety Standards [4] are the basis for construction, designand operation of nuclear components and systems. The general philosophy in the design of com-ponents and structures is to demonstrate that the function and the is guaranteed throughout thelifetime. It is important that the design concept accounts for most possible failure modes and pro-vides rational margins of safety against each type of failure. Some of the potential failure modeswhich component and structure designers should take into account are for example:

Excessive elastic deformation including elastic instability,Excessive plastic deformation,Brittle fracture,Fatigue,Corrosion.

During design stage a complete picture of the state of stress within the component and structureobtained by calculation or measurement of both mechanical and thermal stresses during transientand steady state operation has to be created. It has to be demonstrated that all stresses (primary,secondary) as well as environmental loading are within the allowable stress limits, and the usagefactor developed by a fatigue analysis (peak stresses) is well below the limiting value.

It is possible to prevent failure modes caused by fatigue by imposing distinct limits on the peakstresses at the highest loaded regions of the component and structure since fatigue failure is r e-lated to and initiated by high local stresses or by reducing the load cycles. The design rules ac-cording to the technical codes and standards [1,2,3,4] provides for explicit consideration of cyclicoperation, using design fatigue curves of allowable alternating loads (allowable stress or strainamplitudes) vs. number of loading cycles (S/N-curves), specific rules for assessing the cumulativefatigue damage (cumulative fatigue life usage factor) caused by different specified or monitoredload cycles. The influence of different factors like welds, environment, surface finish, temperature,mean stress and size must be taken into consideration.

2 Use of Design Fatigue Curves

Reviewing fatigue rules and codes for nuclear pressure vessels and piping it becomes apparentthat the majority are similar to or identical with those in the ASME-Code Section III [1], like theGerman KTA Standards [4]. The ASME design fatigue curves for carbon and low alloy steels aswell as austenitic stainless steels are based on stress amplitude and cycles to failure data whichwere obtained from small smooth-machined specimens tested under strain control loading, mainlyin bending in room temperature and air environment [5,6,7]. The design curves were derived byintroducing factors of 2 on stress and 20 on cycles, whichever gave the lowest curve and is meantto account for real effects (size, environment, surface finish, scatter of data) occuring during plantoperation. The fatigue design curve in the British Standard BS 5500 [3] was derived from fatigue

17-5

test data obtained under axial load from welded specimens. The reason therefore was that thepresence of a weld could reduce fatigue strength because of the inevitable presence of weld d e-fects. But all of the pressure vessel and piping fatigue design rules are based essentially on thesame approach based on data from primarily low-cycle fatigue (LCF) tests carried out on ma-chined specimens, mainly with plain unwelded specimens tested under strain control. ConservativeS/N-curves are developed and used for the fatigue analysis in conjunction with stress concentra-tion factors tK or fatigue strength reduction factors fK to take into account the structural di s-continuities in the components and structures including welds [8].

Different procedures exist in the German technical rules for pressure vessels AD-Merkblatt [9]and the European Standard EN 13445 [10] for unfired pressure vessels. The approach uses alsoS/N-curves with stress concentration factors like the ASME Code but much more advice is givenabout the use of the stress concentration factors to be adopted for weld details. Additional explicitfactors in form of an equation or a curve are given to account for the influence of temperature,surface finish and weldment, size and mean stresses. Further German codes and rules used in m e-chanical engineering and machinery are the FKM-Guidelines [11] and the RKF-Guidelines [12]with detailed requirements for the determination of alternating stress amplitudes.

Fatigue data are generally obtained from unwelded specimens at room temperature and are plot-ted in the form of nominal stress amplitude aS vs. number of cycles to failure. The total strainrange tε∆ obtained from the tests is converted to nominal stress range by multiplying the strainrange by the room temperature modulus of elasticity E

2ES t

aε∆⋅= (1)

Most of the S/N-curves given in the codes and standards are to be applied for specific steels (e.g.distinguish between steels of different ultimate tensile strength mR ).

Influence of Temperature

The use of fatigue design curves is restricted in the nuclear codes to a specific maximum te m-perature below the creep range. Using design fatigue curves it is necessary to adjust the allowablestresses if the modulus of elasticity E at operating temperature is different from the one of the de-sign curves. The stress amplitude aS must be multiplied by the ratio of the modulus of elasticitygiven by the design fatigue curve to the value of the modulus of elasticity used in the analysis.

Another approach is given in the German AD S2 rules [9], where the influence of temperaturemust be adjusted by a cycle depending factor Tf .

17-6

Influence of Surface Finish and Welds

Design curves in the nuclear codes include a factor of 2 on stress or 20 on cycles relative to themean of the test data to account for differences between specimen test conditions and real vesselsand piping. This includes effects of surface finish and welds. Furthermore there are in the nuclearcodes specific requirements concerning the surface finish of components especially for welded r e-gions and for different vessel and piping products and different joints. Stress indices are availablefor use of the code equations determining the stress amplitudes.

A special regard to the influence of surface finish depending upon peak-to-valley height zR andnumber of cycles is given in German AD S2 rule. The influence of the surface finishing is d e-scribed by the surface factor which is defined by

)m 6R(

)R(f

zf,a

zf,ao µ<

=σ

σ , (2)

where f,aσ is the sustainable stress amplitude for different zR values. The requirements for de-termining of according to AD S2 for a material with ultimate tensile strength of 500 MPa areshown in Fig. 1.

0,6

0,7

0,8

0,9

1

1000 10000 100000 1000000 10000000Cycles N

Sur

face

fact

or f o

AD S2

R m=500 MPa 200

100

50

10

6

Rz<6 µm

Figure 1: Influence of surface finish according to AD S2 [9]

Experimental values for surface factors of derived by specimens with different zR values areshown in Fig. 2. The experimental data has been evaluated for the endurance limit (N=2·106) ofdifferent materials.

17-7

0,5

0,6

0,7

0,8

0,9

1

0,5 0,6 0,7 0,8 0,9 1

S urface factor fo

Sur

face

fact

or f

o

expe riments byM PA S tuttgart

safe

stee ls w ith UTS

570 M Pa<Rm <1300 M P a

Figure 2: Surface finish factor of for the endurance limit of stress (fatigue strength)

Influence of Size

Most of the material and failure behaviour has been determined using small laboratory specimens.However failure stress amplitudes are lower for components because of size effects caused bydifferent stress gradients or statistical effects of material characteristics. Size effects are covered inthe nuclear codes [1,4] by the factors 2 on stress or 20 on cycles. A different approach is includedin German AD S2 rule, Fig. 3

0 , 6

0,7

0,8

0,9

1

1000 10000 100000 1000000 10000000Cycles N

S

i

z

e

f

a

c

t

o

r

f

d

A D S2non we lded com ponents

< 2 5

100

50

30

wall th ickness s=150 m m

Figure 3: Influence of size according to AD S2 [9]

17-8

Experimental data concerning the influence of size are available from [13], Fig. 4. It is evidentthat comparatively large scatter emerge, especially for the tests with larger specimens.

0,6

0 ,7

0 ,8

0 ,9

1

0 20 40 60 80 100Equivalen t d iam eter / m m

Siz

e fa

ctor

f d

Exp. C rN iMo steel (K t=1)Exp. C rN iMo steel (K t=2)Exp. C rN iMo steel (K t=5)AD S2

Figure 4: Size factor df for the fatigue strength of round solid specimens under repeated (r e-

versed) bending stresses [13] depending upon the equivalent diameter (for pipes: equivalent d i-ameter = s/2)

Influence of Mean Stress

If a component is stressed by an alternating stress greater than the yield strength 2,0pR of thematerial, it makes no difference whether there is a present nominal mean stress mσ or not. In thisstress state the true mean stress always will be zero. Therefore the fatigue design curves are ad-justed to include the maximum effect of the mean stress only in the part of the fatigue curve lyingbelow an alternating stress amplitude 2,0pa R=σ .

In all the ASME-based codes the fatigue design curves are plotted in terms of stress amplitudeindependent of mean stress, the curves are showing already the full effect of maximum meanstress. The evaluation of the effect of mean stresses is accomplished by use of the modified Lan-ger-Goodman Diagram, where mean stress is plotted as the abscissa and the amplitude of the a l-ternating stresses is plotted as the ordinate. Thus, for the adjusted fatigue curve there should notbe any mean stress present which will cause fatigue failure in less than the given cycles.

In non-nuclear codes the influence of mean stresses is taken into account individually. A simpleequation was proposed by Wellinger and Dietmann [14]

m

mf,af,a

R1 )1R()R(

σσσ −⋅−== (3)

with R as the stress ratio. The influence of mean stresses in the FKM-Guidelines [11] is describedby the mean stress sensitivity σM which was introduced by Schütz [15] as

17-9

)0R(

)0R()1R(M

f,a

f,af,a

==−−=

=σ

σσσ . (4)

If there is no experimental data available, the mean stress sensitivity for steels can approximatelybe determined by the equation

1.0R00035.0M m −⋅=σ , (5)

with mR in dimension MPa. Equation (5) is illustrated in Fig. 5.

AlM

gSi1 3.

1354

.53.

1254

.73.

4354

.73.

4364

.7

AlM

g5

St 3

7 St 5

2 Ck

4 5S

AE

413

0 GS

25

CrM

o4

NiC

oMo

gegl

üht

41 C

r4

1.66

04.5

1.77

04.6

PH

15-

7 M

o AM

355

GS

NiC

oMo

NiC

oMo

00 500 1000 1500 2000

U pper tensile stress R / M P am

0 ,2

0 ,4

0 ,6

0 ,8

1 ,0

1 ,2

Mea

n st

ress

sen

sitiv

ity M

σ

scatte ra lum inium alloys

steelstitan ium a lloys

casted stee ls

Figure 5: Mean stress sensitivity σM versus ultimate tensile strength [15]

For the endurance limit of stress (fatigue strength) the mean stress effect on the alternating stressamplitude f,aσ can be adjusted by the Haigh's diagram. Fig. 6 shows the proposal by [11] and[14] compared with experimental data for a high strength low alloy rotor CrMoV-steel.

R =0

R =0,5

R =

σa,f / MPa

Mean stress / MPaσm

100-500 -100 500

500

R = 1

arctan Mσ

yield lim it

arctan Mσ3

R mR p0,2

FKM-Guidelines [11]W ellinger/D ietm ann [14]Exp. M PA Stuttgart

Figure 6: Influence of mean stress according to the fatigue strength diagram (Haigh's diagram)

17-10

Influence of Environment

Despite of the factors 2 and 20 there have been relatively few corrosion fatigue failures in carbonor low-alloy steel components in LWR's and quite a lot of discussions are under way concerningthe influence of environment to the fatigue design curves (crack initiation and crack growth underenvironmental conditions). Data from specimens testing indicated that fatigue life shorter than thefatigue design curve values are possible, if the tests are carried out under low frequency loadingconditions in oxygenated water environment at elevated temperatures [16,17]. The investigationsperformed to determine corrosion-assisted crack growth rates for pressure boundary materialsexhibit a big scatter [18].

3 Calculation of Stress Intensity Range

The four equations used in Class 1 piping to calculate the stress intensity are the code equation 9addressing primary stress margins

mio

2o

1 S5,1I2

MDB

t2

pDB ⋅≤+ (primary stress limit for design conditions), (6)

the code equation 10 addressing the shake down stress limit

mmio

2oo

1n S3)T(RangeStressThermalI2

MDC

t2

DpCS ⋅≤∆++= (7)

(limit for the primary + secondary stress intensity range) and the code equation 11 defining thepeak stress range for fatigue analysis

)T,T,T(RangeStressThermalI2

MDCK

t2

DpCKS 21m

o22

oo11p ∆∆∆++= (8)

with the stress amplitude

2

SS

pa = (9)

Research is still under way concerning the categorization of the stresses directly influencing theresult of a fatigue analysis.

Thermal Stresses

Most of the fatigue relevant stresses in piping systems are caused by thermal loading. The differ-ence between the density of the fluid caused by the temperature gradient from bottom to top ofthe pipe cross section (eg. pressurizer coolant and that of the somewhat cooler hot leg coolant)combined with low flow rates can result in thermal stratification in the horizontal portions of a

17-11

piping system. The hot and cold fluid levels of the stratified flow conditions are separated by ainterface or mixing layer. On the other hand high flow rates can cause a temperature gradient inpipe longitudinal direction (jump of temperature) and result in a thermal shock loading on the in-side pipe surface constant throughout the pipe cross section. To calculate thermal stresses in pipesthe code equations 10 and 11 are available.

Thermal stratification

Thermal stratification in piping system causes an cirumferentially varying temperature distributionin the pipe wall resulting in local through wall axial stresses (through wall radial temperature gra-dient) and global bending stresses in the piping system (axial expansion forces and thermal mo-ments). Maximum local thermal stress is found when a thin interface layer occurs in the upper orlower parts of the pipe cross section. Maximum global thermal bending stress is found when a thininterface layer occurs in the middle of the pipe cross section.

The ASME-Code Section III [1] or KTA Standards [4] rules calculating thermal stresses may notbe completely applicable for the thermal stratification loading.

Thermal shock

The ASME-Code Section III [1] or KTA Standards [4] rules calculating thermal stresses are a p-plicable for the thermal shock loading.

Plastification Factor Ke

For nuclear power plant components which are a subjected to cyclic loading a fatigue analysis inaccordance to different safety rules [1-3] has to be performed. If elastic-plastic deformation is tobe expected then generally costly non-linear FE-calculations have to be carried out. However un-der specific conditions a simplified elastic-plastic fatigue analysis may also be performed usingplastification factors eK . This is considerably simpler since it is based on linear elastic materialbehaviour. The determination of eK values according to the German KTA safety rules, whichhave been largely adopted from the ASME code is shown in the following.

Characteristic stresses and strains are represented by a hysteresis loop during a cycle, Fig. 7.

The fictive elastic strains elε∆ are brought into line with the actual, elastic-plastic strain tε∆ bymultiplying them by the plastification factor eK which is defined by

el

teK

εε

∆∆= . (10)

According to KTA 3201.2, Section 7.8.4 the plastification factor has to be calculated as follows:

17-12

≥

<<

−⋅

−−+

≤≤

=

m3S

S for

n

1

m3S

S3 for 1

S 3

S

)1m( n

n11

3S

S0 for 1

K

m

n

m

n

m

n

m

n

e (11)

with mS as design stress intensity value for the material used.

∆εpl

εm ax

εaεa

εm in

∆εt

∆εel

ε

σσ m

ax

σ mεm

σ minσ a

σ a

Figure 7: Hysteresis loop

For example, for ferritic respectively martensitic steels the characteristic material value mS is cal-culated as

= 0,3

R ;

7,2

R ;

5,1

R MinS mRTmTT2,0p

m . (12)

nS is the fictive elastic equivalent stress range of primary and secondary stresses. The value ofthe material parameters m and n are to be taken from Tab. 1.

Table 1: Material parameters

Material Group m n C/Tmax °low alloy carbon steel, martensitic stainless steel 1 2,0 0,2 370carbon steel, austenitic unalloyed stainless steel 2 3,0 0,2 370nickel-based alloy 3 1,7 0,3 425

17-13

In Fig. 8 the dependence of eK on the load proportional reference magnitude mn S/S is shownfor various material groups.

0

1

2

3

4

5

6

0 2 4 6 8 10

R a tio S n / S m

Pla

stifi

catio

n fa

ctor

Ke

g ro up 1

gro up 2

gro up 3

K TA 320 1.2

Figure 8: eK factors according to ASME [1] and KTA [4]

An evaluation of experimental results from LCF tests with smooth specimens )1K( t = for group1 materials (ferritic and martensitic steels) is shown in Fig. 9. It’s obvious that the scatter is ac-cording to the available wide scale of different heat treatments and strengths relatively small. Thecalculation of eK values according to ASME/KTA is evidently very conservative.

0

2

4

6

8

0 5 10 15 20 25

R a tio Sn / S m

Pla

stifi

catio

n fa

ctor

Ke

K TA 3201.2 (g roup 1 ) E xp . 20 M nM oN i 5 5E xp . X 20 C rM oV 12 1 E xp . 15 N iC uM oN b 5E xp . 15 M nN i 6 3 E xp . 10 C rM o 9 10E xp . 13 C rM o 4 4 E xp . 14 M oV 6 3E xp . 15 M o 3

LC F tes ts a tro om tem pe rature

(n =1 41)

Figure 9: eK factors from LCF tests compared with the values of KTA [4]

17-14

4 Conclusions

Fatigue is a potential failure mode for components and structures of nuclear power plants. For theexplicite consideration of cyclic operational loads (mechanical, thermal) in different technicalcodes and standards specific fatigue analysis methods are available. As for the nuclear codes andstandards the following conclusions can be drawn:

- The influence of temperature is adequate addressed by considering the ratio of the modulus of ela sticity using the S/N-curves.- The design curves were derived by introducing factors of 2 on stress and 20 on cycles to account for real effects (size, environment, surface finish, scatter of data) occuring during plant operation. This implies that during manufacturing and design the specific requriements are met and during plant operation the conditions for environmental effects are monitored and controlled.- To account for thermal stresse within a fatigue analysis the code equations may not be completely applicable for the thermal stratification loading- Compared to experimental data the calculation of the plastification factor eK according to ASME/KTA is evidently very conservative.

References

1. "Rules for Construction of Nuclear Power Plant Components". ASME Boiler and PressureVessel Code, Section III, The American Society of Mechanical Engineers, 1998 Edition

2. "French Design and Construction Rules for Mechanical Components of PWR Nuclear Islands(RCC-M)". AFCEN - Association Française pour la Construction des Ensembles Nuclé-aires, Paris

3. "Unfired Fusion Weld Pressure Vessels - BS 5500". British Standard Institution

4. "Safety Standards of the Nuclear Safety Standards Commission (KTA)". KTA Rules 3201and 3211, Carl Heymanns Verlag KG, Cologne, latest edition

5. B. F. Langer, "Design of Pressure Vessels for Low-Cycle Fatigue", Journal of Basic Engi-neering, Vol. 84, No. 3, September 1962, pp. 389-402

6. C. E. Jaske, W. J. O'Donnell: "Fatigue Design Criteria for Pressure Vessel Alloys", Journalof Pressure Vessel Technology, November 1977, pp. 584-592

7. D. R. Diercks: "Development of Fatigue Design Curves for Pressure Vessel Alloys using aModified Langer Equation", Journal of Pressure Vessel Technology, Vol. 101, November1979, pp. 292-298

8. "Fatigue strength reduction and stress concentration factors for welds in pressure vessels andpiping", Welding Research Council Bulletin, WRC 432, June 1998

17-15

9. German Technical Rules for Pressure Vessels (AD-Merkblätter), AD-S1 and AD-S2, CarlHeymanns Verlag KG, Cologne, latest edition

10. European Standard for "Unfired Pressure Vessels" EN 13445-3 (Part 3 - Design), CEN, 1999

11. "Rechnerischer Festigkeitsnachweis für Maschinenbauteile (FKM)", Teil III - Ermüdungs-festigkeitsnachweis, Heft 183-2, Forschungskuratorium Maschinenbau e.V., Frankfurt 1994

12. "Richtlinienkatalog Festigkeitsberechnung Behälter und Apparate (RKF)", Teil 5 und 6 - Er-müdungsfestigkeit, Linde GmbH, Dresden, 1986

13. K.-H. Kloos, B. Fuchsbauer, W. Magin, D. Zankov: "Übertragbarkeit von Probestab-Schwingfestigkeitseigenschaften auf Bauteile", VDI-Berichte, Nr. 354, 1979, pp. 59-72

14. K. Wellinger, H. Dietmann: "Festigkeitsberechnung - Grundlagen und technische An-wendung", Alfred Kroener Verlag, 1976

15. W. Schütz: " Über eine Beziehung zwischen der Lebensdauer bei konstanter und veränderli-chen Beanspruchungsamplituden und ihre Anwendbarkeit auf die Bemessung von Flugzeug-bauteilen", Zeitschrift für Flugwissenschaften, Band 15, 1967

16. J. M. Keisler, O. K. Chopra, W. J. Shack: "Statistical models for estimating strain-life be-haviour of pressure boundary materials in light water reactor environment", Nuclear Engi-neering and Design 167 (1996), pp. 129-154

17. O. K. Chopra, W. J. Shack: "Low-cycle fatigue of piping and pressure vessel steels in LWRenvironments", Nuclear Engineering and Design 184 (1998), pp. 49-76

18. K. Kussmaul, D. Blind, V. Läpple: "New observations on the crack growth rate of low alloynuclear grade ferritic steels under constant active load in oxygenated high-temperature wa-ter", Nuclear Engineering and Design 168 (1997), pp. 53-75

18-1

18 THE RUSSIAN REGULATORY APPROACHES IN THEFATIGUE EVALUATION OF NPP CONSTRUCTIONCOMPONENTS

I. KaliberdaN. Karpunin

Gosatomnadzor, Moscow, Russia

No presentation slides or technical paper available.

18-3

The Russian Regulatory Approaches in the Fatigue Evaluation ofNPP Construction Components

I. KaliberdaN. Karpunin

Gosatomnadzor, Moscow, Russia

No presentation slides or technical paper available.

19-1

19 THE EVALUATION SYSTEM OF THERMALSTRATIFICATION STRESS USING OUTER SURFACETEMPERATURE

Itaru MuroyaKiminobu Hojo

Takasago Research & Development Center Mitsubishi Heavy Industries, Ltd.

Sigeki SuzukiToshimitsu Umakoshi

Kobe Shipyard and Machinery Works Mitsubishi Heavy Industries, Ltd.

NDE/NDT

20-1

20 MICROSTRUCTURAL CHANGES OF PRESSUREVESSEL STEEL DURING FATIGUE IN HIGHTEMPERATURE WATER ENVIRONMENT

Chie Fukuoka, Yukiya G. Nakagawa, Makoto HiguchiIshikawajima Harima Heavy Industries, Co. Ltd.

Tokyo, Japan


Charlotte, North Carolina, USA

20-3

MICROSTRUCTURAL CHANGES OF PRESSURE VESSEL STEEL DURINGFATIGUE IN HIGH TEMPERATURE WATER ENVIRONMENT

Chie Fukuoka, Yukiya G. Nakagawa, Makoto HiguchiIshikawajima Harima Heavy Industries, Co. Ltd.

Tokyo, Japan


Charlotte, North Carolina, USA

Abstract

An effective method for measuring the fatigue damage accumulation state in a structural materialwas applied to samples removed from an A533 Grade B, Class 1 plate material fatigue cycled ina reactor water environment. The method, Selected Area Diffraction (SAD), is a microstructuralexamination technique that is used for identifying small cell-to-cell angular misorientations inthe crystal lattice. Samples were fatigue cycled as a function of strain amplitude and strain ratein a reactor water environment and angular misorientation measurements taken utilizing the SADtechnique. SAD measurements were then correlated with total fatigue damage accumulation.The angular misorientation was found to increase as the fatigue deformation increased, rapidlyincreasing in the early stages of cyclic loading followed by a more gradual increase duringsubsequent cycling. This profile is consistent with previous SAD measurements taken onsamples fatigue tested in air. For samples fatigued to failure in a reactor water environment asimilar value of angular misorientation was measured, independent of the testing strain rate.This suggests that measurement of a threshold angular misorientation value is feasible forcomponent life assessment of materials exposed to reactor water environments.

Introduction

Reduction in forced outages and avoidance of expensive repairs would be facilitated by aninspection technique that could predict when fatigue macro crack initiation will occur in themetallic structures, systems, and components of power plants. As nuclear power plants age,evaluating pre-fatigue damage becomes increasingly important.

A microstructural technique for measuring the early stages of fatigue in reactor vessel materialshas been developed1. The method, Selected Area Diffraction (SAD) is based on microstructuralexamination by electron diffraction for identifying dislocation cell to cell angular misorientationin the crystal lattice of quenched and tempered high strength low alloy steels. It was observedthat this misorientation increases as the fatigue deformation accumulates, which is a prerequisitefor fatigue crack initiation2.

1 Y.G. Nakagawa, H. Yoshizawa, M.E. Lapides, Metall. Trans. A, 1990, vol.21A, pp.1769-732 C. Fukuoka, H. Yoshizawa, Y.G. Nakagawa, M.E. Lapides, Metall. Trans. A, 1993, vol.24A,pp.2209-16

20-4

It has been demonstrated that the fatigue lifetime of steels in a pressurized high temperaturewater environment is shorter than in air3,4,5. These studies also showed that the fatigue life wasstrongly influenced by strain rate and dissolved oxygen (DO) content. A formula for therelationship between fatigue lifetime and strain rate has been suggested4:

N25w = A(dεT / dt)P (1)

where N25w is defined as the fatigue life, measured as a 25% reduction in applied peak load, in

high temperature water and dεT / dt is the tensile strain rate.

The objective of this work was to investigate the effect of reactor water environment on themicrostructural changes during fatigue. The SAD was used to correlate the impact of theenvironment on the component fatigue life by measuring microstructural changes in the materialsexposed to reactor water environment. Additional details of this work have already beenpublished elsewhere6

Experimental Procedure

Fatigue Testing in High Temperature Water

Fatigue test specimens were machined from an A533 Grade B Class 1 (A533B-1) reactorpressure vessel plate material. This material had been previously prepared to specificallysimulate Japanese early vintage reactor material and contains a higher sulfur content andgenerally exhibits a lower toughness. The chemical composition and mechanical properties ofthis material are summarized in Table 1. Figure 1 (a) and (b) provide dimensions of the fatiguetest specimens that were utilized during testing in air and in LWR water environments,respectively.

Fatigue damage was induced in the test specimens by axial strain controlled cycling at twodifferent strain rates, 0.004%/s and 0.4%/s, and at different strain amplitudes. For fatigue testingat the slow strain rate (0.004%/s) a slow-fast saw-tooth shaped cycling pattern was utilized inorder to reduce the overall fatigue testing time. The fast portion of the saw-tooth shape wasutilized during the compressive portion of each fatigue cycle. It has been previously suggestedthat only the tensile strain portion of the loading had significant influence on the fatigue

3 M. Higuchi and H. Sakamoto, Low Cycle Fatigue Of Carbon Steel In High Temperature PureWater Environment, Tetsu to Hagane (Journal of Iron and Steel Institute of Japan), vol.71, no.8,1985, pp.101-1074 M. Higuchi and K. Iida, Fatigue Strength Correction Factors For Carbon And Low-AlloySteels In Oxygen-Containing High-Temperature Water, Nuclear Engineering and Design,vol.129, 1991, pp.293-3065 M. Higuchi and K. Iida, Reduction In Low-Cycle Fatigue Life Of Austenitic Stainless Steels InHigh-Temperature Water, ASME PVP-vol.353, 1997, pp.79-856 Measuring Fatigue Damage in Materials – Phase 2 , EPRI, Palo Alto, CA: 1999, TR-110251.

20-5

lifetime7. Fatigue testing in air was conducted either at room temperature (RT) or at a nominalLWR operating temperature, 289°C.

Fatigue tests in simulated LWR water were performed in an autoclave with a closed watercirculating loop. The DO level in the water was established by controlling the rate of bubblingargon gas and oxygen into deionized water. Table 2 summarizes the environmental conditionsapplied during testing.

During fatigue testing both in air and simulated LWR water, sample failure (represented by thenumber of cycles to failure, N25) was taken to be the point beyond which the applied peak stressfell 25% under the same strain conditions.

SAD Measurements

The SAD method used in this study is schematically illustrated in Figure 2 and has beendescribed elsewhere1. This method uses transmission electron microscopy to measure theaverage cell to cell misorientation in grains oriented around the <111> zone axis. An example ofthe statistical analysis performed on data obtained from the SAD technique is shown in Figure 3.The normal distribution of the maximum angular deviation, θ, is illustrated for an as-receivedSA508 sample and for samples previously fatigued at a total strain range of 0.78 % to N/N f = 50,100 % (Nf =2,100). The fraction of life, N/Nf, expresses the state of fatigue damage, where N isthe number of the cycles applied to the sample and Nf is the number of cycles to failure. Themean value of θ, equivalent to 50% probability in each plot, is considered to represent anaverage angular deviation of the cells from the reference direction, <111>.

Figure 4 illustrates typical SAD measurements performed on SA508 samples of different heatorigins and fatigued at different strain ranges. As shown in Figure 4, the θ values at N/Nf =100% are similar, 4 to 5 degrees, for all tests, suggesting that a critical orientation value mayexist for this material.

Those samples whose fatigue tests were interrupted for microstructural examination were firstvisually inspected for cracking. The gage section was examined with an optical microscope (x10magnification).

All test bars were cut perpendicular to the stress axis within the gage section. Small disks (3 mmdiameter x 0.1 mm thickness) were fabricated and electropolished to prepare TEM and SADsamples.

Microstructural damage was evaluated by the SAD method and correlated to the fatigue testingconditions. Fatigue lifetime changes due to the different types of testing conditions werecorrelated to the microstructural observations by the SAD method and surface observations.

7 Higuchi, M., Iida, K., and Asada, Y., Effect Of Strain Rate Change On Fatigue Life Of CarbonSteel In High-Temperature Water, ASTM STP 1298, American Society of Testing and Materials,1997

20-6

Results

The results of fatigue testing performed in air on the A533B material, including applicableangular misorientation values measured via the SAD technique, are provided in Table 3. Resultsof fatigue testing performed in a simulated LWR environment, including applicable angularmisorientation values measured via the SAD technique, are provided in Table 4.

Fatigue Test Results

Strain amplitude versus fatigue lifetime curves of SA508-2 in air and A533B-1 in air and inwater are plotted in Figure 5. Open and solid symbols represent fatigue test results in water andin air, respectively. A fatigue curve and its formula are also shown in this figure, which wasobtained as the mean data curve for Japanese low alloy steels in room temperature air 8. Closedcircles represent fatigue test results for SA508-2 in air, triangle markers represent testing resultsat normal strain rate (0.4%/s) for A533B-1, and square markers represent testing results at slowstrain rate (0.004%/s) for A533B-1. It is evident that fatigue lifetime is significantly reduced inthe LWR environment, especially at slow strain rate (0.004%/s).

Figure 6 shows the relationship between strain rate and fatigue life. The slope of the curvesshown in the figure is parameter P in equation (1). The P value (0.476 for strain amplitude of0.6%, 0.747 for that of 0.31%, and 0.746 for that of 0.24%) measured by this study was muchlarger than previous observations for low alloy steels (reported P values for low alloy steelstypically ranged from 0.31 to 0.44)9. This may be due to the relatively high sulfur content of theplate used in this experiment. It has been demonstrated that sulfur content of steels has asignificant effect on fatigue crack growth rate10. Higuchi and Iida also proposed the P value ofcarbon and low alloy steels can be roughly estimated by a following empirical formula8:

PC = 0.864 + 14.6 * S (%) - 0.00092 * UTS (MPa) (2)

where PC is the P value at 289°C and 1-8 ppm oxygen content. This formula implies that thestrain rate more significantly impacts fatigue life with an increase in the sulfur content. The PC

value of the A533B used for this experiment is 0.588.

SAD and Microstructural Examination Results

Figure 7 shows representative fracture surfaces of fatigue specimens from this study. Fracture

8 Higuchi, M., Fatigue Curves And Fatigue Design Criteria For Carbon And Low Alloy Steels InHigh-Temperature Water, ASME PVP Vol. 386, 1999, pp.161-169.9 Higuchi, M. and Iida, K., Effects of Strength and Sulfur Content on Fatigue Strength of CarbonSteel Weldments in Oxygenated High Temperature Water, International Conference on PressureVessel Technology Vol.1, ASME, 199610 Kitagawa H., Nakajima H., Nagata N., Sakaguchi Y., and Iwadate T., The Japanese DomesticRound Robin Tests On Cyclic Crack Growth Of A533B-1 Steels In Simulated LWR PrimaryCoolants, Proceedings of the second IAEA specialists’ meeting on subcritical crack growth, heldin Sendai, May 15-17, 1985, NUREG/CP-0067, MEA-2090, Vol.1, 1986

20-7

surfaces are relatively flat and smooth for the specimens fatigued in air (a) or at normal strainrate (0.4%/s) in LWR environment (b). Those specimens fatigued at a lower strain rate in LWRenvironment always exhibit a rough fracture surface (c). Multiple cracks initiate for thespecimen tested in the LWR water environment when strain rate is slow, many of whichsubsequently propagate simultaneously and ultimately link.

Multiple hair cracks and corrosion pits were seen on the surface of the specimens fatigued in theLWR water environment at a strain rate of 0.004%/s as seen in Figure 8 (a). This condition wasobserved for most of the specimens fatigued at the slow strain rate (0.004%/s) in water.However, only a small number of corrosion pits were observed on the surface of specimensfatigued at the higher strain rate (0.4%/s) even in LWR water environment, as shown in Figure8(b). Figure 8 (c) shows the surface of the specimens fatigued in air, where no corrosion pits orhair cracks were observed. It has been reported that crack initiation occurs in the early stages offatigue life when tested in air, but the number of crack initiation sites are few, and usually verydifficult to observe at the early stage of the fatigue life 11.

Mean misorientation change of the cells during fatigue life is shown in Figure 9. Closed circlemarkers represent the SA508 specimens fatigued in air at normal strain rate (0.4%/s), and closedsquare markers represent the A533B-1 specimen fatigued in air at slow strain rate (0.004%/s).The mean misorientation is slightly lower for A533B-1 for both as-received and fatiguedsamples, and this is in good agreement with the fact that the cell structure is less developed forA533B-1 as compared to SA508. Open markers represent the specimens fatigued in the LWRenvironment. All of the misorientation (SAD) values of the failed specimens are smaller than thevalue of the specimens fatigued in air, but the misorientation (SAD value) also increases duringthe cycling in LWR environment. The observation that the SAD increases quickly in the earlystage of the fatigue life followed by slower increase later in the life, is quite similar to the SADchanges of the specimens fatigued in air.

Discussion

The results show that the SAD method can effectively be utilized to monitor in-servicecomponents in contact with the reactor coolant and to determine if and when the components arein a fatigue damaged state even without detection of cracks. Life assessment is also possible forthe components provided that the change of the SAD as a function of N/N f and the threshold ofthe SAD value in a water environment are determined.

It is important to note that the microstructural changes during fatigue tests in water, even at thevery slow strain rate, occur at the same order of magnitude as the changes observed duringfatigue tests in air. This suggests that the mechanical damage is essential in determiningcomponent fatigue life, regardless of environment. The chemical (corrosion) factor in the waterenvironment is secondary but considered to increase crack velocity during cycling. Thus, we canapply the microstructural conditioning concept for fatigue crack initiation proposed in the

11 Miller, K.J., Initiation And Growth Rates Of Short Fatigue Cracks, Fundamentals ofDeformation and Fracture, Eshelby Memorial Symposium, Sheffield, 2-5 April 1984, pp.477-499

20-8

analyses of the relationship between the SAD and N/Nf in air to testing in water. Since the SADat N/Nf=1 in water (3-3.5 degree) is slightly smaller than the SAD values observed in air (4-4.5degrees), the chemical factor contributes to reduce the critical or threshold value for materialconditioning.

The mechanism of the physical damage process during fatigue is yet to be fully understood. Ithas been commonly accepted and has been confirmed in the past studies of this project thatmicro-cracks exist from the first several cycles of fatigue tests, initiating at small inclusions onthe surface of the samples fatigued in air12. The crack sizes found on the specimen surface weremostly the same (about 0.05 mm), i.e. crack initiation occurs in the very early stage, but thecracks did not grow until near the end of the lifetime, Nf. At this point small cracks started togrow abruptly leading to sample failure.

One of the characteristic differences observed for the sample surfaces tested in water at the slowstrain rate (0.004%/s) was that the density of the small cracks was much higher than those offatigue samples tested in air. This is likely due to the high density of specimen surface corrosionpits as seen in Figure 8 (a) that provided additional crack initiation sites. It was also observed forthe samples tested in water that the small initiated cracks started to grow from the beginning tothe end of the fatigue life, through the continuous process of the growth and combination of themicro-cracks to develop one or two major cracks in the order of 5-10 mm length, resulting in anorder of magnitude reduction in the fatigue life. The strain rate difference of the fatigue tests inwater seems to influence the pitting density, and thus the micro crack growth behavior. For thenormal strain rate fatigue tests, 0.4%/s, the number of surface pits and micro cracks wereobserved to be significantly less than fatigue tests at the low strain rate, 0.004%/s. The crackgrowth behavior at the normal strain rate was closer to that for fatigue tests in air. This mayexplain the difference in the crack surface morphology of the samples after tests. Specificallythe samples tested in water at the slow strain rate exhibited a rough fracture surface while thefracture surfaces of the samples tested at the normal strain rate and in air were relatively flat asshown in Figure 7.

Figure 10 schematically summarizes the results of the present study, where the crack growthbehavior and the mean angular misorientation in the microstructure are plotted as a function offatigue cycles for both tests in air and in water (at slow strain rate). Fatigue crack initiation andthe micro crack growth during fatigue is surface sensitive and strongly depends on theenvironment and strain rate. The microstructural change measured by the SAD techniquerepresents change in the bulk property. Thus, the SAD value is independent of the testingenvironment and only depends on the mechanical strains induced in the sample. The criticalSAD value is given by the value at N=Nf in any testing environment.

Conclusions

1. The SAD value of the specimens fatigue tested in the water environment (DO = 1ppm)increases from about 1.5 (as received) to above 3.0 (N/Nf = 1) degrees, irrespective of thestrain rate. The change is less than the SAD changes observed for specimens tested in air but

12 Measuring Fatigue Damage in Materials-Phase I; EPRI, Palo Alto, CA: 1998. TR-110250

20-9

well above the level of the distinction between the virgin and fatigue-failed state.2. The observation in the specimens fatigue tested to N/N f < 1 that the SAD values increase

quickly in the early stage of the fatigue life followed by a slower increase in later life, is quitesimilar to the SAD changes of the specimens fatigued in air .

3. At the very slow strain rate (0.004%/s) marked reductions in fatigue life are observed in thesamples fatigue tested in the LWR water environment, namely the number of cycles tofailure are an order of magnitude smaller than those tested in air or tested at the strain rate of0.4%/s.

20-10

Table 1Chemical Composition and Major Mechanical Properties of A533B-1

(a) Chemical Composition (weight percent)

C Si Mn P S Ni Cr Mo Cu

0.2 0.3 1.5 0.02 0.020 0.5 0.1 0.5 0.1

(b) Mechanical PropertiesYield Strength

(MPa)Tensile Strength

(MPa)Elongation

(%)Reduction of Area

(%)

481 617 26 61

Table 2Environmental Fatigue Test Conditions

Environmental Conditions Fatigue Test Conditions

Medium

Temperature

Pressure

Electrical Conductivity

pH

Dissolved Oxygen Content

Water Flow Rate

Deionized Water

289

8.0 MPa

<0.2 µS / cm

5-7

1 ppm

60 liter/h

Control Mode

Wave Shape

Strain Rate

Rising Phase

Falling Phase

Strain Ratio

Axial Strain

Triangle/Saw Tooth

0.4-0.004 %/s

0.4 %/s

-1

Table 3Fatigue Test and SAD Results for A533 in Air

Specimen # Temperature(°C)

εεεεa

(%)dεεεεT/dt(%/s)

N25 cycles N/Nf SAD(deg)

as-received - - - - 0 1.42

SQV2A-A1 RT 0.6 0.4 1680 1

SQV2A-A2 RT 0.31 0.4 6740 1

SQV2A-A3 289 0.31 0.4 7380 1

SQV2A-A4 289 0.31 0.004 5026 1 3.86

SQV2A-A5 289 0.31 0.004 N=2500 0.5

R=-1, wave shape:triangle

20-11

Table 4Fatigue Test and SAD Results for A533 in LWR Water Environment


εεεεa



as-received - - - - 0 1.42

SQV2A-W3 289 0.6 0.4 845 1

SQV2A-W4 289 0.6 0.004 94 1 2.8

SQV2A-W5 289 0.31 0.4 4800 1 3.05

SQV2A-W6 289 0.31 0.004 154 1 3.51

SQV2A-W7 289 0.31 0.004 1 0.01 2.26

SQV2A-W8 289 0.31 0.004 15 0.1 2.14

SQV2A-W9 289 0.31 0.004 38 0.25 2.33

SQV2A-W10 289 0.31 0.004 77 0.5

SQV2A-W11 289 0.24 0.4 10895 1

SQV2A-W12 289 0.24 0.004 350 1

R=-1, wave shape: triangle & saw tooth, DO=1 ppm

20-12

Figure 1Fatigue Test Specimens Dimensions (unit: mm)

Figure 2Schematic Illustration of the SAD Procedure

20-13

Figure 3Normal distribution of the cell orientation for 0, 50, and 100% of fatigue life for samplescycled at total strain range of 0.78%

Figure 4Mean Misorientation Difference Between Cells Measured by SAD Versus N/Nf

20-14

Figure 5Strain-life Curve for A533

Figure 6Relationship Between Strain Rate and Fatigue Life

20-15

Figure 7Fractography of Fatigue Samples, Tested (a) in Air, (b) in Water at Normal Strain Rate,and (c) in Water at Slow Strain Rate

Figure 8Surface Observation of Fatigued Samples by Optical Microscopy, Tested (a) in Water atSlow Strain Rate, (b) in Water at Normal Strain Rate, and (c) in Air

Figure 9Mean Misorientation Change During Fatigue Life

20-16

Figure 10Schematic Illustration of Relationship Between Fatigue Crack Growth and MisorientationChange During Fatigue Life

Microstructural Changes Of Pressure VesselSteel During Fatigue In High Temperature

Water Environment

C. Fukuoka, Y. Nakagawa, M. Higuchi

Ishikawajima Harima Heavy Industries, Inc.- Japan

S. Rosinski

EPRI - USA


31 July - 2 August 2000

Napa, California

20-17

Objectives

● Develop a technique to measure microstructuralchanges due to accumulated fatigue damage

● Correlate microstructural changes with measuredchanges in fatigue life

■ Various loading/cycling histories - above and belowfatigue limit

■ Consider reactor water environment

● Assess potential for application to improved lifeassessment capabilities

20-18

Introduction

● Joint program between EPRI and IHI

● Previous activities under joint program■ Fatigue testing performed on SA508 reactor pressure

vessel forging material

■ Influence of loading history on fatigue life investigated -combination of cycling above and below fatigue limit

■ Microstructural changes correlated with measuredchanges in fatigue life

■ Results published in EPRI TR-110250, MeasuringFatigue Damage in Materials - Phase 1, March 1998

20-19

Program Scope

● Scope of Phase 2 program■ Investigate the effect of reactor water environment on

microstructural changes that occur during fatigue

■ Correlate microstructural changes with fatigue damageaccumulation for A533B material exposed to reactorwater environment

■ Samples were fatigue cycled as a function of strainamplitude and strain rate

■ Microstructural changes measured using Selected AreaDiffraction (SAD)

20-20

Selected AreaDiffraction (SAD)

● Microstructuralexamination technique

■ Identify small angularmisorientations incrystal lattice

● Angular misorientationsobserved through TEM

■ Misorientationmeasured betweengrains oriented in the<111> zone axis

20-21

Experimental Setup

Environmental Conditions Fatigue Test Conditions

Medium

Temperature

Pressure

Electrical Conductivity

pH

Dissolved Oxygen Content

Water Flow Rate

Deionized Water

289

8.0 MPa

<0.2 µS / cm

5-7

1 ppm

60 liter/h

Control Mode

Wave Shape

Strain Rate

Rising Phase

Falling Phase

Strain Ratio

Axial Strain

Triangle/Saw Tooth

0.4-0.004 %/s

0.4 %/s

-1

All dimensions in mm

20-22

Results


εεεεa



as-received - - - - 0 1.42

SQV2A-A1 RT 0.6 0.4 1680 1

SQV2A-A2 RT 0.31 0.4 6740 1

SQV2A-A3 289 0.31 0.4 7380 1

SQV2A-A4 289 0.31 0.004 5026 1 3.86

SQV2A-A5 289 0.31 0.004 N=2500 0.5

R=-1, wave shape:triangle


εεεεa



as-received - - - - 0 1.42

SQV2A-W3 289 0.6 0.4 845 1

SQV2A-W4 289 0.6 0.004 94 1 2.8

SQV2A-W5 289 0.31 0.4 4800 1 3.05

SQV2A-W6 289 0.31 0.004 154 1 3.51

SQV2A-W7 289 0.31 0.004 1 0.01 2.26

SQV2A-W8 289 0.31 0.004 15 0.1 2.14

SQV2A-W9 289 0.31 0.004 38 0.25 2.33

SQV2A-W10 289 0.31 0.004 77 0.5

SQV2A-W11 289 0.24 0.4 10895 1

SQV2A-W12 289 0.24 0.004 350 1

R=-1, wave shape: triangle & saw tooth, DO=1 ppm

Fatigue test and SADresults for A533B testedin air

Fatigue test and SADresults for A533B testedin reactor waterenvironment

20-23

Results

Strain amplitude versusfatigue lifetime of SA508-2 inair and A533B-1 in air and inwater

20-24

Results

Relationship between strainrate and fatigue life

20-25

Selected Area Diffraction Results

● SAD values increase quicklyearly in life followed by moregradual increase

● SAD values of A508/A533samples are lower when testedin reactor water environment

20-26

Discussion

● Previous studies suggested threshold SAD value(N/Nf=1) for testing in air

■ θ = 1.5 degrees as-received; 4-5 agrees at N/Nf=1

■ Independent of strain amplitude, strain rate

θθθθ vs N/Nf for A508-2 as a functionof strain amplitude

20-27

Discussion

● Similar threshold SAD value suggested for A533Bsamples tested in air, ~4 degrees

● For testing in reactor water environment, lowerthreshold SAD values (N/Nf=1) measured, ~ 3degrees

● Lower SAD values for both materials when tested inreactor water environment

■ Independent of strain rate

■ Suggests that chemical process contributes to reducecritical (threshold) SAD value

20-28

Conclusions

● Selected Area Diffraction successfully used to measuremicrostructural changes in A533B material exposed toreactor water environment and correlate with fatiguedamage accumulation

● SAD values increase with fatigue accumulation forsamples tested in air and reactor water environment

■ Independent of applied strain rate

■ Increase in SAD value is smaller for specimens tested inreactor water environment than in air

■ Change is above level of distinction between as-received andfatigue-failed (N/Nf=1) state; threshold value suggested

● SAD technique appears to be feasible to determinecomponent fatigue status, independent of environment

20-29

Recommended Activities

● For development and application as an engineeringtool, following activities recommended:

■ Determine SAD threshold values for wider spectrum offatigue test experimental variables

◆ Dissolved oxygen

◆ Strain rate

◆ Strain amplitude

■ Apply SAD to materials extracted from operating plants

20-30

21-1

21 MICROSTRUCTURAL INVESTIGATIONS ANDMONITORING OF DEGRADATION OF LCF DAMAGE INAUSTENITIC STEEL X6CRNITI 18-10

D. Kalkhof, M. Grosse, M. NiffeneggerPaul Scherrer Institute

Nuclear Energy and SafetyVilligen, Switzerland CH- 5232

D. Stegemann, W. WeberUniversity of Hannover

Institute of Nuclear Engineering and NDTElbestrasse 38A

Hannover, Germany D- 30419

21-3

EPRI / USNRC / OECD NEA CSNIINTERNATIONAL CONFERENCE ON FATIGUE OF REACTOR COMPONENTS

JULY 31 - AUGUST 2, 2000NAPA, CALIFORNIA, USA

MICROSTRUCTURAL INVESTIGATIONS AND MONITORING OFDEGRADATION OF LCF DAMAGE IN AUSTENITIC STEEL X6CRNITI 18-10

D. Kalkhof, M. Grosse, M. NiffeneggerPaul Scherrer Institute

Nuclear Energy and SafetyVilligen, Switzerland CH- 5232

D. Stegemann, W. WeberUniversity of Hannover

Institute of Nuclear Engineering and NDTElbestrasse 38A

Hannover, Germany D- 30419

Abstract

The microstructural changes in the pre-crack stage of low-cycle fatigue damage (LCF) in austenitic pipingsteels are characterised by neutron and X-ray diffraction. The LCF damage evolution in the metastableaustenitic steel causes a deformation-induced phase transformation from austenite to martensite. Thresholdsexist for the formation of martensite as a function of both, the load amplitude and the number of LCF cycles.Magnetic stray field and eddy current measurements were chosen to transfer the results of materialcharacterisation to NDT methods. The density and distribution of martensite obtained from neutrondiffraction experiments were used to adjust the NDT signals. Both NDT techniques were able to detect thevery low amount of martensite in the different aged specimens (0.5 - 3.1 vol.% martensite, usage factors from0 up to 1.0).

21-4

1 INTRODUCTION

The development of advanced diagnostic systems that are able to identify and detect material degradation inthe microstructures of steels is a new challenge for non-destructive testing (NDT). Especially in nuclearpower plants, early detection of material degradation can contribute to improve safety and reliability of theprimary circuit boundary as well as plant life time management of certain components. Lifetime extension ofnuclear power plants (NPP) has become an important topic and resulted in the requirement for advancedsafety management tools. Therefore systems for early detection of material degradation of pressurisedprimary loop components are demanded.

With regard to leaks in pressurised water pipes of nuclear power plants [e.g., 1-3], which were caused bythermal fatigue damage, deformation-induced changes in the microstructure of metastable austenitic steelswere investigated. It is well known that varying temperatures and thermal stratification in piping of powerplants might cause fatigue damage by thermal strains. Such strains promote the phase transition from a cubicface centred (fcc) austenite to cubic body centred (bcc) martensite [4].

Our structure investigations using metallography as well as neutron and X-ray diffraction are primarilyintended to identify and quantify martensite formation in the pre-crack stage of low-cycle fatigue damage(LCF) in austenitic piping steels and its influencing parameters. Martensite is a ferromagnetic phasedistributed in the austenitic matrix which itself is paramagnetic. This opens the possibility to detect themartensitic content by means of magnetometers. To monitor the pre-crack damage state, the application ofmagnetic NDT methods based on magnetic stray field and eddy current measurements is investigated.

2 MATERIALS AND TEST CONDITIONS

The investigations were performed on hour-glass specimens (according to ASTM E606, Fig.1). The materialunder investigation was the titanium stabilised austenitic metastable steel X6CrNiTi18-10, which is widelyused for vessels and piping. The chemical composition (in wt.-%) is given in Tab. 1. In order to homogenisethe microstructure, the specimens were annealed at 1040 °C (1 h) and afterwards quenched in oil. The fatiguetests were performed with alternating loading with a test frequency of 2 s-1 at room temperature and werecontrolled by the total strain amplitude. A first test with a strain amplitude of ±2.55 mm/m did not result incrack initiation. Therefore a strain amplitude of ±2.95 mm/m was applied referring to the position of thesmallest cross section. Different usage factors were realised by well defined cycles of strain loads in series ofhour-glass specimens. Tab.2 gives the cycle number for some investigated specimens and the correspondingusage factor D. The usage factor indicates how many of the life time is passed. D is defined to be equal to 1for specimens, where crack initiation has already occurred. The LCF tests were stopped when a 5% forcedrop was reached. Therefore the corresponding cycle numbers vary for specimens with D=1 due todifferences in the damage progression. For specimens without a crack, D is defined as the ratio of the appliedcycle number to an averaged cycle number representing crack initiation. This averaged value for crackinitiation is obtained in pre-testing on 5 specimens.

For the metallographic investigations, the specimens were cut lengthwise. After mechanical grinding andpolishing, the specimens were electrochemically etched using Beraha-II solution [5]. Fig.2 shows themicrostructure of specimen 1.6 at three axial positions. At middle positions dark streaks were found. Withincreasing distances from the middle of the specimen the width of the streaks decreases down to zero. Atpositions far from the middle, the microstructure is homogeneous.

The microstructure of the dark streak regions consists of martensite needles in an austenitic matrix (Fig.3a).Outside these streaks only austenite grains can be found (Fig.3b).

21-5

Table 1. Chemical composition in wt.-% of the X6CrNiTi18-10 specimens

C Mn P S Cr Ni Ti

0.05 1.08 0.039 0.019 17.55 9.86 0.39

75mm

d=18mm

220mm

32mm

Figure 1. Hour-glass specimens (ASTM E606)

-30mm from the middle of the sample

middle of the sample +30mm from the middle of the sample

Figure 2. Microstructure of specimen 1.6 at different axial positions

21-6

a) inside the streaks b) outside the streaks

Figure 3. Microstructure of specimen 1.6 in- and outside the dark streaks

3 STRUCTURE INVESTIGATIONS BY MEANS OF DIFFRACTION EXPERIMENTS

The martensite content formed by phase changes was measured by means of neutron and X-ray diffractionexperiments. Austenite and martensite differ in their microstructure. Austenite has a face centered cubiclattice cell whereas the lattice cell of martensite is body centered cubic. This makes it possible to determinethe phase composition in the steel by neutron and X-ray diffraction experiments.

Neglecting textures, the volume fraction of a phase v1 in a two phase system (austenite and martensite) isgiven by [6]:

v1=[(I{h,k,l}2 / I{h,k,l}1) K1,2 +1]-1 (1)

I{h,k,l} is the intensity scattered at the lattice plane {h,k,l}. K1,2 is a factor which considers the differentstructure type of martensite and austenite and the different angular positions of the reflexions. K1,2 factors aretabulated for X-rays in [6] and can be calculated for neutrons [7].

3.1 Neutron diffraction (ND)

The neutron diffraction experiments were performed at the powder diffractometer DMC at SINQ (SwissSpallation Neutron Source, PSI Villigen) [8]. The instrument is optimised for high intensity. It allows thedetermination of the phase content down to values below 1 %. With a neutron beam wave length ofλ = 0.38 nm, a range of the scattering angle 2Θ of 68° ≤ 2Θ ≤ 147° was analysed. The measurements on theLCF specimens were performed with a beam cross section of 40 mm (width) x 10 mm (height). Themeasuring position was in the middle of the LCF specimens (±5 mm in height). Results of martensite contentobtained by these measurements are summarised in Tab.2. At four specimens also the axial distribution ofthe martensite was measured. For this purpose a reduced beam height of 2 mm was used and the specimenswere scanned in axial direction.

21-7

Fig.4 shows the specimen cross section variation, the corresponding strain amplitude and the measuredmartensite content in dependence on the axial coordinate of the specimen. It can be concluded that themartensite formation is limited to the area of the smallest cross section. In Fig.5, the axial distribution of themartensite content measured with the reduced beam height of 2 mm is given for four specimens. Three ofthem, 1.6, 2.3 and 2.5 have D = 1. The maxima of the martensite content are detected at the crack position. Itseems that the crack formation results in an additional martensitic transformation. In the specimen 1.7(without crack) the maximum of the martensite content was found at the smallest cross section. Themartensite distribution in axial direction is unsymmetric for this specimen. This is also proved by the eddycurrent measurements (see 4.2).

In Fig.6, the measured martensite content is plotted as a function of the strain amplitude. Below a thresholdstrain amplitude (about 2.2 mm/m for specimens 1.7, 2.3 and 2.5 and 2.5 mm/m for specimen 1.6), nomartensite was detectable by neutron diffraction (detection limit of these experiments is about 0.5 vol.-%).For strains higher than the threshold strain amplitude, a nearly linear dependence of the martensite content onthe strain amplitude was found if the data measured at the crack locations were not included. The slope of thecurves depends on the cycle number. The dependence of the martensite content on the cycle number N isgiven in Fig.7. Also for the cycle number a threshold value (Nc = 13'000) exists, below which no martensitictransformation can be detected. At higher cycle numbers the martensite content depends nearly linearly on thecycle number.

21-8

Table 2. Martensite content obtained by neutron diffraction

Sample No. Cycle number Usage factor D Crack position Martensite (vol.-%)

1.2 0 0 - < 0.50

2.4 0 0 - < 0.50

1.8 11200 0.4 - < 0.50

1.4 15850 0.6 - 0.61

2.2 15850 0.6 - 0.54

1.3 22400 0.8 - 0.95

1.7 22400 0.8 - 0.93

2.5 26000 1.0 5 mm away from the middle 1.49

2.3 32000 1.0 in the middle 1.49

1.6 63200 1.0 in the middle 3.13

0

2

4

6

-40 -20 0 20 40axial sample position in mm

mar

ten

site

co

nte

nt

in v

ol.-

%,

cro

ss s

ecti

on

, str

ain

am

plit

ud

e

sample cross section

strain amplitude

content of martensite

0.0

1.0

2.0

3.0

4.0

5.0

-30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30

distance from the middle of the sample in mm

ma

rte

ns

ite

co

nte

nt

in %

sample 1.6: D=1.0,N=63000, crack at 0 mm


sample 1.7: D=0.8,N=22400, no crack

sample 2.5: D=1.0,N=26000, crack at + 5 mm

Figure 4.Axial distribution of the martensite contentfor specimen 1.6 together with the axialdependence of the specimen cross sectionand the local strain amplitude

Figure 5.Axial distribution of the martensite contentfor specimens 1.6, 1.7, 2.3 and 2.5

21-9

0

1

2

3

4

5

2.0 2.2 2.4 2.6 2.8 3.0 strain amplitude in mm/m

mar

ten

site

co

nte

nt

in %



sample 2.5: D=1.0, N=26000, crack at +5 mm

0

1

2

3

4

0 20000 40000 60000 80000

cycle number N

mar

ten

site

co

nte

nt

in %

Figure 6.Martensite content in dependence on the strainamplitude for the specimens 1.6, 2.3, 2.5

Figure 7.Martensite content in dependence on thecycle number

3.2 X-ray diffraction (XRD)

X-ray diffraction experiments using synchrotron light sources enable to analyse martensite distribution in twodimensions. The method guarantees a higher local resolution than neutron diffraction in shorter measurementperiods, but is limited to the surface area. X-ray diffraction results give additional information on the radialdistribution of martensite.

The X-ray diffraction experiments were performed at the ROBL beamline at ESRF Grenoble (France) [9]. Inorder to use the K1,2 values from [6], the wavelength of the Mo-Kα radiation (λ = 0.07107 nm) was applied.The investigated specimen 1.6 was cut into two halves parallel to the sample axis. A mapping of the axial andradial martensite distribution was gained by scanning the specimen through the fixed beam.

With the applied wave length and the beam cross section of 0.2 mm x 0.5 mm, the analysed volume wasabout 1 mm (axial) x 0.5 mm (radial) x 0.005 mm (in depth). For the determination of the martensite content,the intensity relations between the {111}-austenite and the {110}-martensite peak measured in XRDexperiments were used.

Due to the small beam size and the very small beam divergence in the XRD experiments and the relativelarge grain size of the austenite (diameter about 20 - 30 µm), the statistic condition for a polycrystal (infinitenumber of grains NG seen by the beam, in practice NG > 1000) is not completely fulfilled. A wide scatter ofthe intensity values of the {111}-austenite reflexion between the several specimen locations is theconsequence. The grain size of the martensite is small enough to meet this condition. As the content ofmartensite is very small compared with the content of austenite, the mean value of the austenite {111}-intensity can be used for the phase analysis. Using this mean value, the relative error of the martensite content(xM) is smaller than 6% (∆xM = 0.06 x xM).

21-10

Fig.8 shows the axial and radial mapping of the martensite content of specimen 1.6 determined from the X-ray diffraction data. Some typical scattering patterns are included. In agreement with the ND measurements,in the axial direction, the martensite is found at the smallest cross section of the specimen. In the radialdirection, the martensite content is concentrated near the surface (y= -6 and 7mm) and, as a broad area, in thecenter of the specimen (x,y = 0mm).

- 1 8 - 1 6 - 1 4 - 1 2 - 1 0 - 8 - 6 - 4 - 2 0 2

- 8

- 6

- 4

- 2

0

2

4

6

8

X A x i s

Y A

xis

3 . 5 0 0 - - 4 . 0 0 0 3 . 0 0 0 - - 3 . 5 0 0 2 . 5 0 0 - - 3 . 0 0 0 2 . 0 0 0 - - 2 . 5 0 0 1 . 5 0 0 - - 2 . 0 0 0 1 . 0 0 0 - - 1 . 5 0 0 0 . 5 0 0 0 - - 1 . 0 0 0

1 E + 2

1 E + 3

1 E + 4

9 . 6 9 . 7 9 . 8 9 . 9 1 0 1 0 . 1 1 0 . 2 1 0 . 3

1 E + 2

1 E + 3

1 E + 4

9 . 6 9 . 8 1 0 1 0 . 2 1 0 . 4

1 E + 2

1 E + 3

1 E + 4

9 . 6 9 . 8 1 0 1 0 . 2 1 0 . 4

1 E + 2

1 E + 3

1 E + 4

9 . 6 9 . 7 9 . 8 9 . 9 1 0 1 0 . 1 1 0 . 2 1 0 . 3

1 E + 2

1 E + 3

1 E + 4

9 . 6 9 . 8 1 0 1 0 . 2 1 0 . 4

crack tip position

Figure 8.Mapping of the radial and axial distribution of the martensite content and typical XRD patterns

21-11

4 TESTING FOR LCF DEGRADATION BY MEANS OF MAGNETIC NDT METHODS

By means of neutron diffraction experiments it was demonstrated that the martensite content depends on theusage factor for the realised loading conditions at room temperature. Based on these results, it appearschallenging to develop methods for the use in screening tests to identify areas with an increased amount ofmartensite. In this way, an early detection of LCF damaged zones could be managed.

In case of LCF damage in metastable austenitic steels, the application of magnetic methods is possible as thedeformation-induced martensite is of ferromagnetic and the austenitic matrix of paramagnetic nature. Theproblem is to measure very low amounts of martensite between 0.5 and 3 vol.-%. Therefore sensitivemagnetic sensors must be used to identify local differences in magnetic properties. For an advanced NDTtechnique it is also expected that the local distribution of the martensite can be visualised. In principle it ispossible to apply active and passive magnetic techniques.

Concerning the martensite content, results of the neutron diffraction experiments are used to adjust the NDTsignals. Note, that for the usage factor D = 1, the LCF damage is different with regard to the cycle numbers.

4.1 Measurements of magnetic stray field strength

Due to the ferromagnetism of martensite it is possible to apply a passive magnetic technique. Aftermagnetising of the specimens, the remanent field strength can be measured. The magnetisation of thespecimens was carried out in a field with a strength of about 11 kA/m. Afterwards all specimens wereexamined using high sensitive SQUID and fluxgate sensors. Measurements of the axial and normalcomponents of the magnetic stray field were performed in a distance of 20 mm from the bar axis. Thefluxgate sensor was sensitive enough to detect martensitic areas and the results did not show any differencescompared with the SQUID measurements.

The fluxgate output (axial component) is plotted as a function of the axial scanning position in Fig.9. Thisoutput shows a pronounced positive peak when the sensor is directly over the position of the smallest crosssection of the LCF samples. In the case of measurements on samples with cracks (D = 1), the amplitude offield strength depends on the cycle number. For the pre-crack stages the amplitude of the fluxgate outputincreased with the usage factor or with the corresponding cycle numbers. The LCF damage in the specimenswas detected down to the usage factor of D=0.6. The fluxgate output for the usage factors 0.6, 0.4 and 0.0 donot differ significantly.

21-12

-200 -150 -100 -50 0 50 100 150-1,5

-1,0

-0,5

0,0

0,5

1,0

1,5

2,0

2,5

3,0

4

6

5

7

3

2

11 : D = 1 .0 ( N = 6 3 0 0 0 )2 : D = 1 .0 ( N = 3 2 0 0 0 ) 3 : D = 1 .0 ( N = 2 6 0 0 0 ) 4 : D = 0 .8 ( N = 2 2 4 0 0 ) 5 : D = 0 .8 ( N = 2 2 4 0 0 )6 : D = 0 .6 ( N = 1 5 8 5 0 )7 : D = 0

Stre

ngth

(axia

l) of

mag

netic

stra

y fie

ld,

10-6 T

A x i a l p o s i t i o n , m m

Figure 9.Stray field measurements (fluxgate output) at LCF specimens of different usage factors,cycle numbers and martensite content

21-13

4.2 Measurements by means of eddy current techniques

Eddy current techniques seem to be very promising for detecting LCF degradation. If the eddy current field islocally disturbed by changes of permeability, the eddy current signal is changed. The induced voltageamplitude referred to a defined phase angle of the receiving coil was recorded. It is favourable to applyminiaturised sensors. The distance between the measuring sensor and the surface should be small. Theapplied manipulator system was able to follow the hour-glass shape of the specimens, thus the distancebetween sensor and surface was always 1 mm. An essential aspect for eddy current measurements is thechoice of the exciting frequency. For martensite visualisation a frequency of 50 kHz was used which providesa reasonable balance between high local resolution and depth of the eddy current field in the steel. The lowerand upper bound of 0.5 and 3.15 vol.-% martensite content define the adjustment of the eddy currentsensitivity.

For eddy current measurements, the end pieces of the LCF specimens were cut off and only the middlesections 75 mm in length were used. Additionally, the specimens were cut lengthwise into two halves. Thus itis possible to determine the martensite distribution also in the radial specimen direction and obtaininformation about the progress of the martensite formation. Some examples of eddy current output obtainedon the circumferencial surface and on the axial cross section are plotted in the Figs.10 and 11. The plotsvisualise the martensite distribution on the circumferential surface (Fig.10) and on the axial cross section(Fig.11). The axial specimen direction corresponds to the horizontal coordinate axis, the circumferential andradial specimen directions correspond to the vertical coordinate axis. Plots are shown for different usagefactors (specimen numbers 1.4, 1.7 and 2.3 with the corresponding usage factors 0.6, 0.8 and 1.0).

In Fig.10 the eddy current signal amplitudes at the circumferential surface are plotted for an angle range of180°. The martensite content of 0.61 vol.-% for specimen 1.4 was clearly detected. The martensite ishomogeneously distributed in the circumferential direction. The plot shows the beginning of the martensiteformation during LCF loading. The changes of the martensite density in the axial direction show that theformation of martensite was really deformation-induced. In the following damage stages D = 0.8 and D = 1.0,the martensitic pattern is the darker as the usage factor increases. For specimen 1.7, the martensite is not ashomogeneously distributed as for specimen 1.4. It is possible that the axial loading was accompained by asmall amount of bending. For specimen 2.3, loaded up to crack initiation, a single large crack was detected atthe position of the smallest cross section. Furthermore, some local areas with a very high martensiteconcentration were found. These are probably new sources for macroscopic cracks. The martensite ishomogeneously distributed. At the crack position an additional formation of martensite was observed.

In Fig.11, the eddy current signal amplitudes at the axial cross section surface are plotted. The sensor outputfor specimen 1.4 shows, that the formation of martensite is limited at the surface in the position of thesmallest cross section. No martensite was detected in depth.

For further damage progression (specimen 1.7), the martensite tends to spread in depth. The distribution ofmartensite in the axial direction is wider at the surface than in the bulk. A macroscopic crack initiation willbe effected by stress relaxation around this location. This results in a decrease of the martensite spreading inthis area which is shown for specimen 2.3. The shape of the martensitic pattern is broader at the opposite sideof the crack.

21-14

Figure 10.Eddy current measurements on thecircumferencial surface of LCF specimens

1.4 (D=0.6, xM=0.61 vol.-%)1.7 (D=0.8, xM=0.93 vol.-%)2.3 (D=1.0, xM=1.49 vol.-%)

Figure 11.Eddy current measurements on the axialcross section of LCF specimens

1.4 (D=0.6, xM=0.61 vol.-%)1.7 (D=0.8, xM=0.93 vol.-%)2.3 (D=1.0, xM=1.49 vol.-%)

21-15

5 CONCLUSION

All applied methods (neutron and X-ray diffraction, magnetic stray field measurements with SQUID andfluxgate sensors, eddy current measurement) allowed the detection of martensite. While the amount ofmartensite could be quantified with the diffraction methods, the magnetic techniques yield only relativeresults; that means, they have to be calibrated somehow, i.e. with results from neutron diffractionexperiments.

Neutron and X-ray diffraction are powerful tools to determine the deformation-induced martensite contentand its distribution in hour-glass specimens. Neutron diffraction is the only method to obtain the absolutevalue of martensite content in a specimen section covering the whole thickness.

It was shown that the content of martensite depends on the usage factor. The existence of thresholds for thecycle number and the total strain amplitude to form deformation-induced martensite could be concluded(total strain amplitude: 2.2 .. 2.5 mm/m, cycle number: 13'000). These thresholds are useful when examiningindustrial components under service conditions. It should be clarified whether thresholds also exist for otherconditions of loading and temperature.

Both applied magnetic techniques, stray field and eddy current measurements, were able to detect materialdegradation due to low-cycle fatigue. For our testing conditions the eddy current technique was able to detectLCF damage down to D=0.6, the stray field measurements down to D=0.8. No significant differences ofmeasuring signals for both techniques were found for D < 0.6, which corresponds to the results obtained byneutron diffraction experiments. For practical application, eddy current measurements in combination withneutron diffraction investigations are recommended.

The correlation between the usage factor and the NDT signal is clearly demonstrated which is ratherpromising for further investigations. However, further investigations namely for higher temperature rangesare needed before the application of the method for real components can be tested.

6 ACKNOWLEDGMENTS

This work was supported by the Swiss Federal Nuclear Safety Inspectorate, HSK.

REFERENCES

1. Experience with thermal fatigue in LWR piping caused by thermal mixing and stratification.Paris: OECD Nuclear Energy Agency, 1998. OECD NEA CSNI R(98)8

2. Leak on the Residual Heat Removal System at Civaux 1. Vienna: International Atomic Energy Agency,1999. IAEA TC 485.27

3. R. Rantala, Thermal Fatigue Experiences and Countermeasures in Finland. Paris: Meeting of the OECDNuclear Energy Agency CSNI Principal Working Group 3, 1999

4. R.W.K. Honeycombe, Steels, Microstructure and Properties. Edward Arnold Publishers Ltd., 1981

5. Metallographische Anleitung zum Farbätzen nach dem Tauchverfahren, Band2: Farbätzung nachBeraha und ihre Anwendungen. Düsseldorf: Deutscher Verlag für Schweisstechnik, Herausgeber,E.Weck, E.Leistner, 1983

21-16

6. G.Fanninger, U.Hartmann, Physikalische Grundlagen der Quantitativen RöntgenographischenPhasenanalyse (RPA). Härterei Technische Mitteilungen HTM 27, 1972, p. 233

7. G.E.Bacon, Neutron Diffraction, Oxford: Clarendon Press, 1975.

8. http://www1.psi.ch/www_sinq_hn/SINQ/instr/ DMC.html

9. W.Matz et al., ROBL- a CRG Beamline for Radiochemistry and Materials Research at the ESRF. JournalSynchrotron Radiation 6, 1999, p. 1076

22-1

22 NDE TECHNOLOGY FOR DETECTION OF THERMALFATIGUE DAMAGE IN PIPING

Stan WalkerPedro Lara

EPRI NDE Center1300 WT Harris Blvd.Charlotte, NC 28262

22-3

NDE Technology for Detection of Thermal Fatigue Damage in Piping

Stan WalkerPedro Lara

EPRI NDE Center1300 WT Harris Blvd.Charlotte, NC 28262

704/547-6081 [email protected]

ABSTRACT

Recent piping failures related to thermal fatigue have prompted a review of examinationpractices currently used for detection of thermal fatigue damage in piping, particularly insmall-diameter piping (less than 4-inches in diameter). Because volumetric examinationtechniques are difficult and most examinations of small-diameter piping are currentlylimited to surface examinations, the industry has expressed a desire to develop a plan forboth near- and long-term solutions. A program is being developed that includes criteriaand methods to improve the reliability of small-diameter piping examinations. Inresponse to these concerns, an Issues Task Group (ITG) was put together under EPRI’sPressurized Water Reactor Materials Reliability Project. The inspection-relateddeliverables of the ITG, Thermal Fatigue Inspection Guidelines, are the subject of thispaper. Specifically, this paper describes the results of the evaluation of candidatenondestructive evaluation (NDE) technologies for thermal fatigue crack detection andrecommends guidance for NDE examiner qualifications.

INTRODUCTION

Leaks of thermal fatigue origin in piping connected to the reactor coolant system haveoccurred at several power plants in the US, Europe, and Japan, and the NuclearRegulatory Commission (NRC) has expressed concern with the occurrences. Becausethese failures have included small diameter piping where volumetric inspectiontechniques were considered to be unreliable, the NRC has expressed their desire for theelectric industry to develop a plan for both near- and long-term solutions.

This activity was aimed at helping the industry reach consensus on the optimal NDEmethodology and its capability to detect thermal fatigue cracking in small diameterpiping. To accomplish this objective, the relevant experiences and currently-acceptedpractices were reviewed. Evaluations were performed on mockups containing thermalfatigue cracks to document the capabilities of various NDE techniques. From theseevaluations, guidance about the qualification of NDE examiners was established.

22-4

THERMAL FATIGUE CRACKING

Thermal fatigue damage typically occurs as a result of a component malfunction, whichcauses an unplanned mixing of hot and cold fluids. In addition to a faulty component,conditions for thermal fatigue require the formation of a temperature stratified fluidregion, with the stratification varying periodically. The temperature difference betweenthe mixing fluids must be sufficiently large so that the associated principal stress exceedsthe fatigue endurance limit value.

Although the crack distribution is somewhat random, some common patterns can beidentified from the reported events.

• The thermal fatigue cracks originate at the inner surface and grow to theoutside surface.

• The cracks initiate as a shallow network called craze. These craze cracks aretransgranular in nature and exhibit extensive branching. Although thedirection of these cracks is somewhat random, a “checker-board” appearancehas been observed in many situations.

• These craze cracks are reported as shallow, however, actual quantification oftheir depth is scarcely documented.

• The deeper cracks (responsible for the component failure) generally occur inthe midst of the craze cracking, although some exceptions to this pattern havebeen reported. The resulting crack system typically includes one or more deepcracks surrounded by an extensive network of shallower penetrating defects.

• Cracks that originate away from welds tend to have an axial orientation. Theseaxially-oriented cracks are skewed, with the skewness angle varyingsomewhat randomly.

• In the vicinity of welds, the crack orientation is likely to be circumferential.1

MOCKUPS

Six specimens with implanted thermal fatigue cracks were manufactured, with thepurpose of providing targets against which the effectiveness of the candidate NDEtechniques could be gauged, and to be used as part of the examiner qualificationprocedure.

The mockups were made to test the ability of the NDE technique to

• Perform a volumetric examination on high curvature piping and elbowsurfaces;

• Size the length of axial, circumferential, and skewed cracks located in themidst of a craze network;

• Size the extent of the craze crack network.

All specimens are made of austenitic stainless steel type 304L.

22-5

The crazing and crack penetration depths were selected to be at the low end of themeasurements reported at power plants which experienced failures: from 0.025 to 0.147inch. Accordingly, the specimens were made with craze depths of 0.02 or 0.04 inch, andcracks were produced with depths twice the depth of the surrounding craze.

RESULTS

From the evaluations performed to-date, the following trends and observations have beenderived

• Manual ultrasonic techniques were capable of detecting and measuring thelength of craze and thermal fatigue cracks in each mockup.

• The time-of-flight diffraction technique detected, length sized and depth sizedthe craze and thermal fatigue cracks in 5 of the 6 mockup samples tested. Thecraze was not detected in one sample.

• Conventional radiography detected the thermal fatigue cracks and some of thecraze cracks in the mockup specimens. Craze detection was sufficient tooutline the craze affected area.

• Ultrasonic spectroscopy detected the craze cracks in 4 of the 6 mockupsevaluated. The areal extension of the craze was accurately sized. Thethermal fatigue cracks were not detected.

• Pulsed Eddy Current detected the craze cracks with depths greater than 10percent. The technique did not detect the thermal fatigue cracks.

• Vibro-Modulation detected the presence of craze cracks in 2 of the 3 mockupspecimens tested. Areal extension of the craze could not be derived from thesignals.

• Conventional eddy current was judged not to be suitable for this applicationbecause of the limited penetration of the current field.

From these results the following conclusions were obtained:

• Manual ultrasonic examination techniques were found to perform best amongthe technologies considered in this evaluation. This technique is viable fordetecting and length-sizing cracks of thermal fatigue origin in small borepiping when applied in accordance with the EPRI procedure.

• Time-of-flight diffraction is a viable detection technique for scanning largeareas semi-automatically to be supplemented with manual ultrasonictechniques when more precise length-sizing evaluations can lead to a betterrepair schedule.

• Conventional radiography, with fine-grain film, is a viable technique fordetecting thermal fatigue cracks deeper than 10 percent of the wall thickness.This technique can be supplemented with manual ultrasonic techniques whenthe examination objective includes detecting craze damage in its early stages.

22-6

• Ultrasonic spectroscopy, pulsed eddy current, and vibro-modulation requirefurther development before they can be implemented in a power plantenvironment for thermal fatigue crack inspection.

Accordingly, a generic procedure for inspection using manual ultrasonic techniques wasdeveloped. The generic procedure was further tested on a field-extracted, 1½-inchdiameter elbow that exhibited thermal fatigue damage.

Finally, it is recommended that examiners receive approximately 4 hours ofindoctrination prior to performing examination in power plants. This indoctrinationshould be administered to examiners who have previously demonstrated proficiency inultrasonic examination of piping welds through some industry standard.1

QUALIFICATION OF ULTRASONIC EXAMINERS

Examiner qualification is an important aspect of NDE. Beginning in 1983, it has beenmandatory that all persons performing ultrasonic examination at any domestic boilingwater reactor (BWR) facility for detection of intergranular stress corrosion cracking(IGSCC) satisfactorily complete training and performance demonstration. This wasmandated because of the history of poor performance, and/or the occurrence of leakagedue to IGSCC within a short time after examination.

Currently, the majority of NDE personnel performing examinations at nuclear powerstations has received training for ultrasonic detection of IGSCC and has demonstratedsatisfactory performance on field-removed specimens. Because thermal fatigue crackinghas characteristics entirely different from IGSCC, and occurs from different causalfactors, it is recommended that all personnel performing ultrasonic examination forthermal fatigue cracking receive additional indoctrination (or training).

The indoctrination recommendations are based on knowledge of crack morphology,laboratory experimentation, and field experience. It is recommended that thisindoctrination be administered to examiners who have previously demonstratedproficiency in ultrasonic examination of piping welds through some industry standard.The industry standard could be a formal qualification such as the American Society ofMechanical Engineers (ASME) Section XI Appendix VIII qualifications as administeredby the Performance Demonstration Initiative (PDI) or some other industry-recognizedstandard. In addition to the completion of some existing piping qualification program(which may have been conducted on samples with IGSCC or other cracking), it isrecommended that examiners for thermal fatigue damage receive indoctrination ofapproximately 4 hours prior to performing examinations at power plants. Thisindoctrination should include information on formation of thermal fatigue cracking, theappearance of craze cracking, thermal fatigue crack morphology as contrasted to IGSCC,ultrasonic interaction with thermal fatigue cracking, and the use of specialized ultrasonicprobes for small diameter pipes. This indoctrination should also include a visualevaluation of thermal fatigue cracks and a hands-on ultrasonic examination of knownthermal fatigue cracking. A course is currently being developed to provide this

22-7

recommended indoctrination. It is not necessary for examiners to complete a practicaldemonstration of proficiency in addition to the one completed for the aforementionedindustry standard.1

REFERENCE

1. Draft 100015, NDE Technology for Detection of Thermal Fatigue Damage in Piping(MRP-TF-06), EPRI, Palo Alto, CA and MRP TF-ITG, May 2000

23-1

23 NDE METHOD FOR DETECTION OF FATIGUE DAMAGE

G. L. KesselDominion Generation

Innsbrook Technical Center500 Dominion Boulevard

Glen Allen, Virginia 23060

J. K. NaEnergy Service Group International, Inc.

8979 Pocahontas TrailWilliamsburg, Virginia 23185-6243

W. T. Yostand

J. H. CantrellNASA Langley Research CenterHampton, Virginia 23681-0001

Y. L. HintonU. S. Army Research LaboratoryVehicle Technology Directorate

MS 231Hampton, Virginia 23681-2219

23-3

NDE METHOD FOR DETECTION OF FATIGUE DAMAGE

G. L. KesselDominion Generation

Innsbrook Technical Center500 Dominion Boulevard

Glen Allen, Virginia 23060

J. K. NaEnergy Service Group International, Inc.

8979 Pocahontas TrailWilliamsburg, Virginia 23185-6243

W. T. Yostand

J. H. CantrellNASA Langley Research CenterHampton, Virginia 23681-0001

Y. L. HintonU. S. Army Research LaboratoryVehicle Technology Directorate

MS 231Hampton, Virginia 23681-2219

23-4

Abstract

An NDE tool measuring fatigue levels of steam turbine blades has been developedby NASA and Virginia Power. The background technology is based on the nonlinearacoustics that has been used to study fatigue damage in aluminum alloys at NASA. Asan initial effort for the joint development, a base line study was performed with 410Cbstainless steel by using the state of the art facility in NASA. In-situ data collection onsteam turbine blades at the Virginia Power’s power stations became possible when aportable fatigue measurement system was designed and built. The results show that thenonlinearity parameter, β, is useful to predict the fatigue life.

Introduction

Microstructural damages caused by fatigue in metallic materials are usuallyrelated to the accumulation of various types of dislocation structures. Depending on thematerial’s crystal structures, somewhat different dislocation substructures are formedduring the fatigue process. However, in general, dislocation dipole densities increase asthe fatigue process continues until the initiation of a crack. A pictorial diagram in Figure1 shows the relation between the acoustic nonlinearity parameter β and the fatigue cyclesas the fatigue process continues until the formation of a crack in wavy-slip materials. Ithas been found theoretically and experimentally that the values of β increase as thefatigue damage accumulates through a dynamical process involving dislocation dipoles[1, 2].

Figure 1. Relation between β and number of fatigue cycles for typical wavy-slip alloys.

Dislocation Dipoles

Dislocations

Pre PSB (vein)like structures

VacancyGenerationRegion

PSB

Frozen PSBβ

Log Cycles

Intersection ofPSB with MaterialDiscontinuity

StressSingularity

CrackFormatio

23-5

NASA Efforts in Nonlinear Acoustics and Early Work in Fatigue

For the last decade or so, NASA has been developing the nonlinear acoustics touse as a nondestructive evaluation technique for assessment of fatigue damages inaluminum alloys used in aircraft. The initial work in fatigued aluminum was performedin aluminum 2024-T4 and the results have been published [3, 4, 5].

The nonlinear acoustics technique referred in this paper relies on the detection ofthe second harmonic signals which generated from wave distortion caused in part by themicrostructure of materials. In order to generate harmonic signals, a transmittingultrasonic transducer is mounted on one side of the test specimen as shown in Figure 2(a).Initially, it launches a pure sinusoidal ultrasonic tone burst signal with a fixed frequencyinto the material. The medium distorts the wave during propagation and the distortedwaveform reaches the receiving transducer on the other side of the specimen. The signalis frequency analysed during which the amplitude of fundamental and second harmoniccomponents are determined. See Figure 2(b). The acoustic nonlinearity parameter, β, isthen determined by measuring the amplitudes of both fundamental and second harmonicsignals and using the following expression:

β = 8/(k2a) (A2/A12) (1)

where k is the wave number (2π/λ), a is the thickness of the test specimen, A2 and A1 arethe amplitudes of second harmonic and fundamental signals, respectively.

For calibration of reference samples, we use a capacitive detector (microphone) asthe receiving transducer placed on a carefully prepared sample. The use of the capacitivedetector permits absolute determination of the acoustic wave amplitudes.

The capacitive detection system has been used to study the fatigue damage inaluminum 2024-T4 induced by loading a cyclic tensile stress at 85% of the yield strengthon an MTS machine. The results of the test show that the value of β increases as thefatigue cycles increases. More details can be found in the references [6, 7].

Current Work between NASA and Virginia Power

NASA and Virginia Power have developed an NDE tool which can help todetermine the fatigue damage in steam turbine blades. This joint effort has beenexercised since 1993. As the first step of the investigation we performed a basic studywith 410Cb stainless steel as we did with the aluminum 2024-T4 alloy. The dog boneshaped fatigue specimens were examined by sectioning out the gauge area after they werefatigued on an MTS frame. The results showed the same trend in β as the aluminumalloys as shown in Figure 3. It is clear that the value if β increases as the fatigue damageincreases. After 10 million cycles of fatigue, the increase in β value was found to be

23-6

(a) Acoustic signal distorts as it propagates through the materials under test.

(b) Frequency analyzed of the received signal after transmission through the material under test.

Figure 2. Generation of harmonic signals due to wave distortion due to micro structures.

about 90%. However, the β value measured on a section cut out from the lower vanearea of an actual turbine blade, which was retired for cause, was approximately 500%higher than the β value of virgin material. This result indicates that the fatigue damageon the turbine blades are much higher than the simple tensile stress fatigue mechanism.In turbine blades, the fatigue mechanisms may be influenced by environmental factorsnot present during the fatigue of the 410 stainless steel used in the fatigue specimens.

The measurement system used for the basic studies uses generic apparatus foreach function and consequently takes almost two full racks of equipment. Since a bulkysystem is impractical for use in power plant environment, we identified a commercialmanufacturer (Ritec, Inc) of a system adoptable for field use. A block diagram for thismeasurement system is shown in Figure 4. A 5 MHz pure sinusoidal tone burst signal isgenerated by a gated amplifier and filtered by a high power low pass filter before it issent to the transmitting transducer. The signal received by the receiving transducer issplit into two band pass filters. One is for the fundamental component (5 MHz) and the

Specimen

TransmittingTransducer

ReceivingTransducer

Frequency(MHz)

Fundamental

SecondHarmonic

ThirdHarmonicA

mpl

itud

e (d

B)

23-7

other is for the second harmonic component (10 MHz). Once the amplitudes of these twocomponents are measured by a digital oscilloscope, the β value can be determined by areference technique developed for these measurements.

Figure 3. Nonlinearity parameter as a function of fatigue cycles for 410Cb stainless steel.

Figure 4. Block diagram for the fatigue damage measurement system.

0

10

20

30

40

50

60

70

80

10 1

y = 1.2 Log(x) + 13.0

nonlinearity parameter of an actual turbine blade

10 2 10 3 10 4 10 5 10 6 10 7

50 �Terminator

5 MHzLow Pass

Filter

10 MHzBand Pass

Filter

DigitalOscilloscope

(LeCroy)

Tone Burst Genrator/Receiver

1.27 cmTransmittingTransducer

(5 MHz)

0.48 cmReceivingTransducer

Sample

Second Harmonic

Fundamental

SignalMonitor

ReceiverNo.1 No.2

GatedAmplifier

Output

.

Splitter

5 MHzBand Pass

Filter

23-8

The reference sample measurement technique [8] uses a standard calibration sample todetermine β. Field measurements on the turbine blade are directly compared with equivalentmeasurements on the reference sample. Under these conditions, the field measurements areconverted directly to the nonlinearity parameter values, β.

The advantage of this reference sample method is that the voltage levels measuredfor the fundamental and second harmonic signals are directly used without making acomplicated calibration procedure normally required with the capacitive detectionmethod. This is a significant improvement in reducing the measurement time in the fieldbecause measurements on more than one hundred blades need to be tested in a limitedaccess time for non-destructive inspection during the outage.

Other factors which affect these field measurements include non-parallel surfacesand surface roughness. Blades are normally sand blasted with 220 grit aluminum oxidebeads for an inspection. In addition to this, the cross section of the blades is airfoilshaped with cant angle variable across the surface from 5 to 15 degrees. We havedeveloped correction formulas to compensate for theses effects. The experimental resultsare shown in Table 1 [9]. The general trends can be seen from this table are as follows:the rougher the surface, the lower the β value while the higher the cant angle, the higherthe β value.

Results of Turbine Blade Study

The nonlinearity parameter was measured on 403 stainless steel turbine bladesused for 17 years on Unit 5 at the Virginia Power’s Chesterfield Power Station during amaintenance outage. Fourteen blades were selected randomly from the L-0 stage blades.The measurements were taken on the boss areas around the two tie wire holes and aregion midway between the tie wire holes. The test results are shown in Figure 5. Fiveor more measurements at each location on each blade were taken. The standarddeviations are shown by the error bars. In most cases, the standard deviations are smallwhich demonstrates the reliability of the measurement. At a glance, we notice that the βvalues of the outer tie wire regions are slightly higher than the other two regions.Although each blade may have experienced different fatigue loading over the entireservice period, the overall trend seems to be that the outer tie wire area is subject to morefatigue damage.

For comparison of the β values of the three test regions, average β values for eachregion of all tested blades are calculated and plotted in Figure 6(a). The average β valuefor the outer tie wire is 41.3, an increase of nearly 500% over β of the virgin materialwhich ranges between 7.2 and 8. Near the inner tie wire, the average β is 27.3, whilemidway between the two tie wires the value is 32.2 (increases of about 300% and 400%over the virgin β value). This may indicate that the region around the outer tie wireexperiences more fatigue induced damage than the other two regions. The tie wires

23-9

Table 1. β values of canted specimens at various surface finishes

reduce the amount of flexural bending in the turbine blades during normal operation. Theregion around the inner tie wire would therefore experience less bending due toconstraints by the rest of the blade resulting in a lower β value compared to the outer wireregion. In addition, the boss area around the tie wire is of greater thickness, whichincreases the flexural stiffness and further reduces bending. The area between the two tiewires is slightly less restrained and able to bend more, and hence is subject to morefatigue damage accumulation than the inner tire wire area.

More nonlinearity parameter measurements were made on the lower vane sectionof randomly selected turbine blades of L-1 stage on Unit 3 at the Mount Storm PowerStation of Virginia Power and the results are shown in Figure 7. The β values of these10 year old blades are higher than the virgin material, with one blade showing a β valueof over 60. This is relatively high for a 10 year old blade. However, considering the factthat some blades have failed within 20 years of service life in the past, this result mayindicate that the particular blade may be experiencing a higher fatigue loading comparedto the other blades on the same rotor.

Recently, Virginia Power provided a 26 inch long L-1 stage steam turbine blade(410Cb stainless steel) retired from service from the Chesterfield Power Station. Theblade was designed and manufactured by General Electric and has one tie wire hole and asingle boss on the pressure side. The blade served for 18 years and was removed fromthe rotor because of a crack that was initiated from the tip of the tie wire hole on thesuction side and has grown toward the trailing edge approximately a half inch long. Thecrack is visible with naked eyes from both sides of the blade. An X-ray image showingthe crack is shown in Figure 8 with a picture of the corresponding region.

RMS 330

RMS 280

RMS 200

RMS 125

RMS 63

RMS 2

Optical

4.2 7.7 12.4 26.3

Roughness

4.8 7.2 12.2 23.6

5.3 7.5 11.6 20.4

5.2 3.4 11.5 18.7

5.8 4.0 10.7 17.2

6.5 4.5 9.9 15.4

7.3 5.1 9.3 13.6

0o 5o 10o 15o

23-10

The nonlinearity parameter β was measured on the vane section of the crackedblade as shown in the Figure 9. The rectangular image was constructed with the β valuesmeasured at the step of one eighth of an inch over the entire area. An enlarged image isshown in Figure 10. Higher β values were measured on the boss area compared to other

0

10

20

30

40

50

60

70Beta at Outer Tie Wire

ββββ

0

10

20

30

40

50

60

70Beta Between Tie Wires

ββββ

0

10

20

30

40

50

60

70Beta at Inner Tie Wire

ββββ

Individual Blades

Figure 5. β measured at three different locations on each of fourteen blades. Error barsindicate standard deviations.

23-11

Figure 6. Average β values at the three different locations on 17-year old Chesterfieldblades.

Figure 7. β values of 10-year old Mount Storm blades.

β

Material: 403 Stainless Steel Age: 17 Years

Outer Tie Wire Midspan Inner Tie Wire 0

10

20

30

40

50

60

Individual Blades

Material: 403 Stainless Steel

Age: 10 years

0

10

20

30

40

50

60

70

β

23-12

regions on the vane. From this result, it is clear that the boss area is subject to a higherlevels of fatigue damage. Some spots on the boss are even higher than 150 for the βvalue. The direction of crack propagation appears to intersect regions of maximum β inthe boss area. If one compares the X-ray image in Figure 8 and image in Figure 10, thecrack had propagated toward the two high β value spots indicated by the two hollowarrows. The β values on these spots are over 150.

Figure 8. An X-ray image and its corresponding photographic image of the crack at thetie wire hole of a retired blade. On the photographic image, the invisiblesuction side tie wire hole and the crack is drawn in for a better comparison.

Anticipated Benefits

The fatigue sensor results will be used by turbine owners to determine theoptimum time to replace components. The present method for blade replacement is torepair cracked components and then replace these blades as materials become available.The procurement time for some steam turbine blades can be over one year. The fatiguesensor results will allow blades to be replaced prior to crack initiation, and will allow theequipment to operate for longer periods between maintenance inspection intervals.

(a) X-ray image (b) Photographic image

Tie Wire Hole (suction side)

Tie Wire Hole(pressure side)

FatigueCrack

RootTip

23-13

5 0 . 0

1 0 0 . 0

1 5 0 . 0

9"

1.25"

Tie Wire Hole

Boss Area

Pressure Side

23-14

.

Bos

s A

rea

23-15

The down time due to a steam turbine blade failure can be weeks or even monthsin length, depending on the availability of parts. By being able to measure the fatigue ofa component, the life of a component can be predicted, and the part can be replaced priorto failure. The ability to replace blades prior to crack initiation will increase thereliability of the turbine, and will reduce the unexpected down time caused by bladefailures. This ability to optimise blade replacements will reduce the turbine operating andmaintenance costs

Acknowledgment

This work is supported by Virginia Power.

References

1. Hikata, A., Chick, B. B., “Dislocation contribution to the second harmonic generationof ultrasonic waves”, Journal of Applied Physics, Vol 36, 1965, p. 229.

2. Cantrell, J. H., Yost, W. T., “Acoustic harmonic generation from fatigue-induceddislocation dipoles”, Philosophical magazine A, Vol. 69, No 2, 1994, pp. 315-326.

3. William T. Yost and John H. Cantrell “Materials characterisation using acousticnonlinearity parameter and harmonic generation: engineering materials”, Review ofProgress in Quantitative Nondestructive Evaluation, Vol. 9, 1990, p. 1669.

4. William T. Yost and John H. Cantrell “Nonlinear acoustical assessment of precipitatenucleation and growth in aluminum alloy 2024”, Review of Progress in QuantitativeNondestructive Evaluation, Vol. 19, 2000, p. 1375.

5. William T. Yost and John H. Cantrell “Fatigue cycle induced variation of the acousticnonlinearity parameter in aluminum alloy 2024”, Review of Progress in QuantitativeNondestructive Evaluation, Vol. 18, 1999, p. 2345.

6. Jacob Philips and M. A. Breazeale, Physical Acoustics, Vol. XVII (Academic Press,New York, 1984), Ch. 1.

7. Jeong K. Na, John H. Cantrell and William T. Yost “Linear and nonlinear ultrasonicproperties of fatigued 410Cb stainless steel”, Review of Progress in QuantitativeNondestructive Evaluation, Vol. 15, 1996, p. 1347.

8. William T. Yost and John H. Cantrell “Calibration techniques for electronic-basedsystems used in measurement of nonlinearity parameters”, Review of Progress inQuantitative Nondestructive Evaluation, Vol. 18, 1999, p. 2345.

23-16

9. Jeong K. Na, John H. Cantrell and William T. Yost “Effects of surface roughness andnonparallelism on the measurement of the acoustic nonlinearity parameter in steamturbine blades”, Review of Progress in Quantitative Nondestructive Evaluation, Vol.19, 2000, p. 1417.

ENVIRONMENTAL FATIGUE I

24-1

24 ENVIRONMENTAL EFFECTS ON FATIGUE CRACKINITIATION IN PIPING AND PRESSURE VESSELSTEELS

Omesh K. ChopraArgonne National Laboratory, Argonne,

Illinois 60439

24-3

ENVIRONMENTAL EFFECTS ON FATIGUE CRACK INITIATION INPIPING AND PRESSURE VESSEL STEELS

Omesh K. Chopra Argonne National Laboratory, Argonne, Illinois 60439

Phone 630-252-5117

KEYWORDSFatigue Crack Initiation, Strain vs. Life (S–N) Curve, LWR Environment, Carbon Steel, Low–Alloy Steel, Austenitic Stainless Steel

ABSTRACTThe ASME Boiler and Pressure Vessel Code provides

rules for the construction of nuclear power plant components.Appendix I to Section III of the Code specifies fatigue designcurves for structural materials. However, the effects of lightwater reactor (LWR) coolant environments are not explicitlyaddressed by the Code design curves. Test data illustratepotentially significant effects of LWR environments on thefatigue resistance of carbon and low–alloy steels and austeniticstainless steels. This paper summarizes the work performed atArgonne National Laboratory on the fatigue of piping andpressure vessel steels in LWR coolant environments. Theexisting fatigue S–N data have been evaluated to establish theeffects of various material and loading variables, such as steeltype, strain range, strain rate, temperature, and dissolved–oxygen level in water, on the fatigue lives of these steels.Statistical models are presented for estimating the fatigue S–Ncurves for carbon and low–alloy steels and austenitic stainlesssteels as a function of material, loading, and environmentalvariables. Methods for incorporating environmental effectsinto the ASME Code fatigue evaluations are discussed.Differences between the methods and their impact on thedesign fatigue curves are also discussed.

INTRODUCTIONCyclic loadings on a structural component occur because

of changes in mechanical and thermal loadings as the systemgoes from one load set (e.g., pressure, temperature, moment,and force loading) to any other load set. For each load set, anindividual fatigue usage factor is determined by the ratio of thenumber of cycles anticipated during the lifetime of thecomponent to the allowable cycles. Figures I–9.1 through I–9.6 of Appendix I to Section III of the ASME Boiler andPressure Vessel Code specify design fatigue curves that definethe allowable number of cycles as a function of applied stress

amplitude. The cumulative usage factor (CUF) is the sum ofthe individual usage factors, and the ASME Code Section IIIrequires that the CUF at each location must not exceed 1.

The ASME Code fatigue design curves, given inAppendix I of Section III, are based on strain–controlled testsof small polished specimens at room temperature in air. Thefatigue design curves were developed from the best–fit curvesof the experimental data by first adjusting for the effects ofmean stress on fatigue life and then reducing the fatigue life ateach point on the adjusted curve by a factor of 2 on strain or 20on cycles, whichever was more conservative. As described inthe Section III criteria document, these factors were intendedto account for data scatter (heat–to–heat variability), effects ofmean stress or loading history, and differences in surfacecondition and size between the test specimens and actualcomponents. The factors of 2 and 20 are not safety marginsbut rather conversion factors that must be applied to theexperimental data to obtain reasonable estimates of the lives ofactual reactor components. However, because the meanfatigue curve used to develop the current Code design curve foraustenitic stainless steels (SSs) does not accurately representthe available experimental data (Jaske and O’Donnell, 1977;Chopra, 1999), the current Code design curve for SSs includesa reduction of only §�� DQG �� IURP WKH PHDQ FXUYH IRU WKH

SS data, not the 2 and 20 originally intended.As explicitly noted in Subsection NB–3121 of Section III

of the Code, the data on which the design fatigue curves(Figs. I–9.1 through I–9.6) are based did not include tests inthe presence of corrosive environments that might acceleratefatigue failure. Article B–2131 in Appendix B to Section IIIstates that the owner's design specifications should provideinformation about any reduction to design fatigue curves thathas been necessitated by environmental conditions. Existingfatigue strain–vs.–life (S–N) data illustrate potentiallysignificant effects of light water reactor (LWR) coolantenvironments on the fatigue resistance of carbon steels (CSs)

International Conference on Fatigue of Reactor ComponentsJuly 31 - August 2, 2000

Napa California, USA

24-4

and low–alloy steels (LASs) (Ranganath et al., 1982; Higuchiand Iida, 1991; Nagata et al., 1991; Van Der Sluys, 1993;Kanasaki et al., 1995; Nakao et al., 1995; Higuchi et al., 1997;Chopra and Shack, 1997, 1998a, b, c, 1999) and of austeniticSSs (Fujiwara et al., 1986; Mimaki et al., 1996; Higuchi andIida, 1997; Kanasaki et al., 1997a, b; Hayashi, 1998; Hayashiet al., 1998; Chopra and Gavenda, 1997, 1998; Chopra andSmith, 1998; Chopra, 1999) (Fig. 1). Under certainenvironmental and loading conditions, fatigue lives of CSs canbe a factor of 70 lower in the environment than in air (Higuchiand Iida, 1991; Chopra and Shack, 1998b). Therefore, themargins in the ASME Code may be less conservative thanoriginally intended.

Two approaches have been proposed for incorporating theeffects of LWR environments into ASME Section III fatigueevaluations: develop new design fatigue curves for LWRapplications, and use a fatigue life correction factor to accountfor environmental effects. Both approaches are based on theexisting fatigue S–N data for LWR environments, i.e., thebest–fit curves to the experimental fatigue S–N data in LWRenvironments are used to obtain the design curves or fatiguelife correction factor. As and when more data becameavailable, the best–fit curves have been modified and updatedto include the effects of various material, loading, andenvironmental parameters on fatigue life. Interim designfatigue curves that address environmental effects on fatiguelife of carbon and low–alloy steels and austenitic SSs were firstproposed by Majumdar et al. (1993). Design fatigue curvesbased on a rigorous statistical analysis of the fatigue S–N datain LWR environments were developed by Keisler et al. (1995,1996). Results of the statistical analysis have also been used toestimate the probability of fatigue cracking in reactorcomponents. The Idaho National Engineering Laboratoryassessed the significance of the interim fatigue design curvesby evaluating samples of components in the reactor coolantpressure boundary (Ware et al., 1995). Six locations wereevaluated from facilities designed by each of the four U.S.nuclear steam supply system (NSSS) vendors. Selectedcomponents from older vintage plants, designed according tothe B31.1 Code, were also included in the evaluation. Thedesign curves and statistical models for estimating fatigue livesin LWR environments have recently been updated for carbonand low–alloy steels (Chopra and Shack, 1998b, c, 1999) andaustenitic SSs (Chopra and Smith, 1998; Chopra, 1999).

The alternative approach, proposed initially by Higuchiand Iida (1991), considers the effects of reactor coolantenvironments on fatigue life in terms of a fatigue lifecorrection factor Fen, which is the ratio of the life in air to thatin water. To incorporate environmental effects into the ASMECode fatigue evaluations, a fatigue usage for a specific loadset, based on the current Code design curves, is multiplied bythe correction factor. Specific expressions for Fen, based onthe statistical models (Chopra and Shack, 1998b, c, 1999;Chopra, 1999; Mehta and Gosselin, 1996, 1998) and on thecorrelations developed by the Environmental Fatigue Data

Committee of the Thermal and Nuclear Power EngineeringSociety of Japan (Higuchi, 1996), have been proposed.

This paper summarizes the data that are available on theeffects of various material, loading, and environmentalparameters on the fatigue lives of carbon and low–alloy steelsand austenitic SSs. The two methods for incorporating theeffects of LWR coolant environments into the ASME Codefatigue evaluations are presented. Although estimates offatigue lives based on the two methods may differ because ofdifferences between the ASME mean curves that were used todevelop the current design curves and the best–fit curves to theexisting data that were used to develop the environmentallyadjusted curves, either method provides an acceptableapproach to account for environmental effects.

0.1

1.0

10.0

10 1 10 2 10 3 10 4 10 5 10 6

Str

ain

Am

plitu

de,

Ha

(%)

Carbon Steel

Fatigue Life (Cycles)

Mean CurveRT Air

ASME Design Curve

Temp. (°C)DO (ppm)Rate (%/s)S (wt.%)

: <150: Š0.05: •0.4: •0.006

150–2500.05–0.20.01–0.4•0.006

>250>0.2<0.01•0.006

10 1 10 2 10 3 10 4 10 5 10 6

0.1

1.0

10.0

Austenitic Stainless Steels

Fatigue Life (Cycles)

Mean CurveRT Air

ASME Design Curve

Temp. (°C)DO (ppm)Rate (%/s)

250–325-0.005Š0.01

: 100–200: -0.005: -0.01

260–325•0.2•0.4

Ha

(%)

Fig. 1. Fatigue S–N data for carbon steels and austeniticstainless steels in water; RT = room temperature

FATIGUE S–N DATA IN LWR ENVIRONMENTSCarbon and Low–Alloy Steels

The fatigue lives of both CSs and LASs are decreased inLWR environments; the reduction depends on temperature,strain rate, dissolved oxygen (DO) level in water, and Scontent of the steel. Fatigue life is decreased significantlywhen four conditions are satisfied simultaneously, viz., strainamplitude, temperature, and DO in water are above aminimum level, and strain rate is below a threshold value.The S content in the steel is also important; its effect on lifedepends on the DO level in water. Although themicrostructures and cyclic–hardening behavior of CSs andLASs differ significantly, environmental degradation of fatigue

24-5

life of these steels is very similar. For both steels, onlymoderate decrease in life (by a factor of <2) is observed whenany one of the threshold conditions is not satisfied. The effectsof the critical parameters on fatigue life and their thresholdvalues are summarized below.(a) Strain: A minimum threshold strain is required for

environmentally assisted decrease in fatigue lives of CSsand LASs (Chopra and Shack, 1998b, c, 1999). Thethreshold value most likely corresponds to the rupturestrain of the surface oxide film. Limited data suggestthat the threshold value is §�� KLJKHU WKDQ WKH IDWLJXH

limit for the steel.(b) Strain Rate: Environmental effects on fatigue life occur

primarily during the tensile–loading cycle, and at strainlevels greater than the threshold value required torupture the surface oxide film. When any one of thethreshold conditions is not satisfied, e.g., DO<0.05 ppm or temperature <150°C, the effects of strainrate are consistent with those in air, i.e., only the heatsthat are sensitive to strain rate in air show a decrease inlife in water. When all other threshold conditions aresatisfied, fatigue life decreases logarithmically withdecreasing strain rate below 1%/s (Higuchi and Iida,1991; Katada et al., 1993; Nakao et al., 1995); the effectof environment on life saturates at §��V �&KRSUD

and Shack, 1998b, c, 1999). The dependence of fatiguelife on strain rate for A106–Gr B CS and A533–Gr BLAS is shown in Fig. 2. For A533–Gr B steel, thefatigue life at a strain rate of 0.0004%/s in high–DOwater (§�� ppm DO) is more than a factor of 40 lowerthan that in air.

(c) Temperature: When other threshold conditions aresatisfied, fatigue life decreases linearly with temperatureabove 150°C and up to 320°C (Higuchi and Iida, 1991;Nagata et al., 1991; Nakao et al., 1995). Fatigue life isinsensitive to temperatures below 150°C or when anyother threshold condition is not satisfied.

(d) Dissolved Oxygen in Water: When other thresholdconditions are satisfied, fatigue life decreaseslogarithmically with DO above 0.05 ppm; the effect

saturates at §�� ppm DO (Nagata et al., 1991; Nakao etal., 1995). Fatigue life is insensitive to DO level below0.05 ppm or when any other threshold condition is notsatisfied.

10 2

10 3

10 4

10 -5 10 -4 10 -3 10 -2 10 -1 10 0

AirSimulated PWR-0.7 ppm DO

Fat

igue

Life

(C

ycle

s)

Strain Rate (%/s)

A106–Gr B Carbon Steel

288°C, Ha -0.75%

10 2

10 3

10 4

10 -5 10 -4 10 -3 10 -2 10 -1 10 0

AirSimulated PWR-0.7 ppm DO

Fat

igue

Life

(C

ycle

s)

Strain Rate (%/s)

A533–Gr B Low–Alloy Steel

288°C, Ha -0.75%

Fig. 2. Dependence of fatigue life of carbon and low–alloysteels on strain rate

(e) S Content of Steel: The effect of S content on fatiguelife depends on the DO content of the water. When thethreshold conditions are satisfied and for DO contentd1.0 ppm, the fatigue life decreases with increasing Scontent. Limited data suggest that environmental effectson life saturate at a S content of §�� wt.% (Chopra

Table 1. Fatigue test results for Type 304 austenitic SS at 288°C

TestNo.

Dis.xygena

(ppb)

Dis.Hydrogen(cc/kg)

Li(ppm)

Boron(ppm)

Pre–soak(days)

pHat RT

Conduc-tivityb

(PS/cm)

ECPa

teel mV(SHE)

Ten.Rate(%/s)

StressRange(MPa)

StrainRange

(%)

LifeN25

(Cycles)1805 – – – – – 4.0E-3 467.9 0.76 14,4101808 4 23 2 1000 1 6.4 18.87 –686 4.0E-3 468.3 0.77 2,8501821 2 23 2 1000 1 6.5 22.22 –693 4.0E-3 474.3 0.76 2,4201859 2 23 2 1000 1 6.5 18.69 –692 4.0E-3 471.7 0.77 2,4201861 1 23 – – 1 6.2 0.06 –610 4.0E-3 463.0 0.79 2,6201862 2 23 – – 5 6.2 0.06 –603 4.0E-3 466.1 0.78 2,4501863 1 – – – 5 6.3 0.06 –520 4.0E-3 476.5 0.77 2,250

aDO and ECPs measured in effluent.bConductivity of water measured in feedwater supply tank.

24-6

and Shack, 1998b). At high DO levels, e.g., >1.0 ppm,fatigue life seems to be insensitive to S content in therange of 0.002–0.015 wt.% (Higuchi, 1995). When anyone of the threshold conditions is not satisfied,environmental effects on life are minimal and relativelyinsensitive to changes in S content.

Austenitic Stainless SteelsThe fatigue lives of austenitic SSs are decreased in LWR

environments; the reduction depends on strain rate, level ofDO in water, and temperature (Chopra and Gavenda, 1997,1998; Chopra and Smith, 1998; Kanasaki et al., 1997a). Theeffects of LWR environments on fatigue life of wroughtmaterials are comparable for Types 304, 316, and 316NG SS.Although the fatigue lives of cast SSs are relatively insensitiveto changes in ferrite content in the range of 12 to 28%(Kanasaki et al., 1997a), the effects of loading andenvironmental parameters on the fatigue life of cast SSs differsomewhat. The significant results and threshold values ofcritical parameters are summarized below.(a) Strain: A minimum threshold strain is required for

environmentally assisted decrease in fatigue life ofaustenitic SSs. Limited data suggest that the thresholdstrain range is 0.32 to 0.36% (Chopra and Smith, 1998;Kanasaki et al., 1997b).

(b) Dissolved Oxygen in Water: For wrought austeniticSSs, environmental effects on fatigue life are morepronounced in low–DO, i.e., <0.01 ppm DO, than inhigh–DO, i.e., t0.1 ppm DO, water (Chopra and Smith,1998; Kanasaki et al., 1997a). In high–DO water,environmental effects are moderate (less than a factor of2 decrease in life) when conductivity is maintained at<0.1 PS/cm and electrochemical potential (ECP) of thesteel has reached a stable value (Fig. 3). For fatiguetests in high–DO water, the SS specimens must besoaked for 5 to 6 days for the ECP of the steel to reach astable value. Figure 3 shows that although fatigue life isdecreased by a factor of §� ZKHQ FRQGXFWLYLW\ RI ZDWHU

is increased from §�� WR �� PS/cm, the period forpresoaking appears to have a larger effect on life thanthe conductivity of water. In low–DO water, theadditions of Li and B, or low conductivity, orpreexposing the specimen for §� GD\V EHIRUH WKH WHVW� RU

dissolved H, have no effect on fatigue life of Type 304SS (Table 1). Also, for cast austenitic SSs, the effect ofDO content is somewhat different; the fatigue lives areapproximately the same in both high– or low–DO waterand are comparable to those observed for wrought SSsin low–DO water (Chopra and Smith, 1998).

(c) Strain Rate: In high–DO water (conductivity<0.1 PS/cm and stable ECP of the steel), fatigue life isinsensitive to changes in strain rate. In low–DO water,fatigue life decreases logarithmically with decreasingstrain rate below §��V; the effect of environment on

life saturates at §��V IRU ZURXJKW SSs (Chopraand Smith, 1998; Kanasaki et al., 1997b). Existing dataare too sparse to define the saturation strain rate for castaustenitic SSs.

10 3

10 4

10 -2 10 -1 10 0

Fat

igue

Life

(C

ycle

s)

Conductivity of Water (PS/cm)

Type 304 SS288°C, 'H -0.75%Strain Rate 0.004/0.4 %/sDO -0.84 ppm

ECP Steel Electrode mV(SHE)Open Symbols: 145–165 (-120 h soak)Closed Symbols: 30–145 (-20 h soak)

Fig. 3. Effects of conductivity of water and soak period onfatigue life of Type 304 SS in high–DO water

(d) Temperature: Existing data are too sparse to establishthe effects of temperature on fatigue life over the entirerange from room temperature to reactor operatingtemperatures. Limited data indicate that environmentaleffects on fatigue life are minimal below 200°C andsignificant at temperatures above 250°C (Kanasaki etal., 1997b); life appears to be relatively insensitive tochanges in temperature in the range of 250–330°C. Thepressure vessel research council (PVRC) steeringcommittee for cyclic life and environmental effects(CLEE) has proposed a ramp function to describetemperature effects on the fatigue lives of austenitic SSs;environmental effects are moderate at temperaturesbelow 180°C, significant above 220°C, and increaselinearly from 180 to 220°C (Yukawa, 1999).

OPERATING EXPERIENCE IN NUCLEAR POWERINDUSTRY

Experience with operating nuclear power plantsworldwide reveals that many failures may be attributed tofatigue; examples include piping components, nozzles, valves,and pumps (Kussmaul et al., 1983; Iida, 1992). In most cases,these failures have been associated with thermal loading due tothermal stratification and striping, or mechanical loading dueto vibratory loading. Significant thermal loadings due to flowstratification were not included in the original design basisanalysis. The effect of these loadings may also have beenaggravated by corrosion effects due to a high–temperatureaqueous environment. A review of significant occurrences ofcorrosion fatigue damage and failures in various nuclear powerplant systems has been presented in an EPRI report (Dooleyand Pathania, 1997); the results are summarized below.Cracking in Feedwater Nozzle and Piping

Fatigue cracks have been observed in feedwater pipingand nozzles of the pressure vessel in boiling water reactors(BWRs) and steam generators in pressurized water reactors

24-7

(PWRs) (Kussmaul et al., 1984; NRC 1979, 1993). Themechanism of cracking has been attributed to corrosion fatigue(Watanabe, 1980; Gordon et al., 1987) or strain–inducedcorrosion cracking (SICC) (Lenz et al., 1983). Case historiesand identification of conditions that lead to SICC of LASs inLWR systems have been summarized by Hickling and Blind(1986).

In BWR nozzle cracking, initiation has been attributed tohigh–cycle fatigue caused by the leakage of cold water aroundthe thermal sleeve junction area, and crack propagation hasbeen attributed to low–cycle fatigue due to plant transientssuch as startup/shutdowns and any feedwater on/off transients.The frequency of the high–cycle fatigue phenomenon due toleakage around the sleeve is §��±� Hz and therefore is notexpected to be influenced by the reactor coolant environment.Estimates of strain range and strain rates for typical transientsassociated with low–cycle fatigue are given in Table 2 (Ford etal., 1993). Under these loading and environmental conditions,significant reduction in fatigue life has been observed forcarbon and low–alloy steels (Chopra and Shack, 1998b, 1999).

Table 2. Typical chemical and cyclic strain transients

Component OperationDO

(ppb)Temp.(°C)

StrainRange (%)

StrainRate (%/s)

FW Nozzle Startup 20/200 216/38 0.2-0.4 10–2

FW Piping Startup 20/200 216/38 0.2-0.5 10–3–10–2

FW Piping Startup 20/200 288/38 0.07-0.1 4–8x10–6

FW PipingTurbineRoll <200 288/80 0.4 3–6x10–3

FW PipingHotStandby <200 288/90 0.26 4x10–4

FW Piping Cool Down <20 288/RT 0.2 6x10–4

FW PipingStratifica-tion 200 250/50 0.2-0.7 10–4–10–3

In PWR feedwater pipe cracking, cracking has beenattributed to a combination of thermal stratification andthermal striping (Dooley and Pathania, 1997). Environmentalfactors, such as high DO in the feedwater, are believed to alsohave played a significant role in crack initiation. The thermalstratification is caused by the injection of low–flow, relativelycold feedwater during plant startup, hot standby, andvariations below 20% of full power, whereas thermal stripingis caused by rapid, localized fluctuations of the interfacebetween hot and cold feedwater.

Lenz et al. (1983) showed that in feedwater lines, thestrain rates are 10–3–10–5%/s due to thermal stratification and10–1%/s due to thermal shock and that thermal stratification isthe primary cause of crack initiation due to SICC. Also, theresults from small–size specimens, medium–size components(model vessels), and full–size thermal–shock experimentssuggest an influence of oxygen content in pressurized water oncrack initiation behavior (Kussmaul et al., 1984).

A detailed examination of cracking in a CS elbowadjacent to the steam generator nozzle weld (Enrietto et al.,1981) indicates crack morphologies that are identical to thoseobserved in smooth specimens tested in high–DO water. For

example, the deepest crack was straight, nonbranching,transgranular through both the ferrite and pearlite regionswithout any preference, and showed considerable oxidationand some pitting at the crack origin. In fatigue test specimens,near–surface cracks grow entirely as tensile cracks normal tothe stress and across both the soft ferrite and hard pearliteregions, whereas in air, cracks grow at an angle of 45° to thestress axis and only along the ferrite regions (see Fig. 4 of acompanion paper at this conference by Chopra and Park). Theidentical crack morphologies indicate that environment playeda dominant role in crack initiation. Similar characteristics oftransgranular crack propagation through both weld and basemetal, without regard to microstructural features, have alsobeen identified in German reactors (Hickling and Blind, 1986).

Components tests have also been conducted to validate thecalculation procedures and the applicability of the test resultsfrom specimen to actual reactor component. Tests on pipes,plates, and nozzles, under cyclic thermal loading in aqueousenvironment (Kussmaul et al., 1983) indicate that crackinitiation in simulated LWR environments may occur earlierthan the values of the ASME Section III fatigue design curve;environmental effects are more pronounced in the ferritic steelthan in the austenitic cladding. Tests at the HDR–facility(Katzenmeier et al., 1990) have also shown good agreementbetween the fatigue lives applicable to specimens andcomponents, e.g., first incipient crack on pipes appeared in1200 cycles, compared with 1400 cycles for a test specimenmade of the same material and tested under comparableconditions (8 ppm DO).

Safety Injection System and Pressurizer Surge LineSignificant cracking has also occurred in unisolable pipe

sections in the safety injection system piping connected to thePWR coolant system (NRC 1988a, b). This phenomenon,which is similar to the nozzle cracking discussed above, iscaused by thermal stratification. Also, regulatory evaluationhas indicated that thermal stratification can occur in all PWRsurge lines (NRC 1988c). In PWRs, the pressurizer water isheated to §��& ��)�� 7KH KRW ZDWHU� IORZLQJ DW D YHU\

slow rate from the pressurizer through the surge line to thehot–leg piping, rides on a cooler water layer. The thermalgradients between the upper and lower parts of the pipe can beas high as 149°C (300°F).

Full–scale mock-up tests to generate thermal stratificationin a pipe in a laboratory have confirmed the applicability oflaboratory data to component behavior (Lenz et al., 1990). Thematerial, loading, and environmental conditions were simulatedon a 1:1 scale, using only thermohydraulic effects. Under theloading conditions, i.e., strain rate and strain range typical ofthermal stratification in these piping systems, the coolantenvironment is known to have a significant effect on fatiguecrack initiation (Chopra, 1999; Kanasaki et al., 1997a, b).

24-8

Steam Generator Girth Weld crackingAnother instance of thermal–fatigue–induced cracking

where environmental effects are believed to have played a rolein crack initiation has been observed at the weld joint betweenthe two shells of a steam generator (Foley et al., 1991). Thefeedwater temperature in this region is nominally 204–227°C(440–440°F), compared with the steam generator temperatureof 288°C (550°C). The primary mechanism of cracking hasbeen considered corrosion fatigue with possible slow crackgrowth due to stress corrosion cracking. A detailed analysis ofgirth–weld cracking indicates that crack initiation wasdominated by environmental influences, particularly underrelatively high–DO content and/or oxidizing potential(Bamford et al., 1991).

INCORPORATING ENVIRONMENTAL EFFECTSINTO ASME FATIGUE EVALUATIONS

Two procedures are currently being proposed forincorporating effects of LWR coolant environments into theASME Section III fatigue evaluations; develop a new set ofenvironmentally adjusted design fatigue curves (Chopra andShack, 1998b, 1999; Chopra, 1999; Chopra and Smith 1998)or use fatigue life correction factors Fen to adjust the currentASME Code fatigue usage values for environmental effects(Chopra and Shack, 1999; Chopra, 1999; Mehta and Gosselin,1996, 1998). For both approaches, the range and boundingvalues must be defined for key service parameters thatinfluence fatigue life. It has been demonstrated that estimatesof fatigue lives based on the two methods may differ because ofdifferences between the ASME mean curves used to developthe current design curves and the best–fit curves to the existingdata used to develop the environmentally adjusted curves.However, either of these methods provides an acceptableapproach to account for environmental effects.

Design Fatigue CurvesA set of environmentally adjusted design fatigue curves

can be developed from the best–fit stress–vs.–life curves to theexperimental data in LWR environments by using the sameprocedure that has been used to develop the current ASMECode design fatigue curves. The stress–vs.–life curves areobtained from the strain–vs.–life curves, e.g., stress amplitudeis the product of strain amplitude and elastic modulus. Thebest–fit experimental curves are first adjusted for the effect ofmean stress by using the modified Goodman relationships

cS a Sa

V u �V y

Vu � Sa

§

©¨ ·

¹¸ for Sa <Vy , (1a)

and cS a = Sa for Sa >Vy , (1b)

where cS a is the adjusted value of stress amplitude Sa, and Vyand Vu are yield and ultimate strengths of the material,respectively. Equations 1a and 1b assume the maximumpossible mean stress and typically yield a conservativeadjustment for mean stress, at least when environmental

effects are not significant. The design fatigue curves are thenobtained by lowering the adjusted best–fit curve by a factor of2 on stress or 20 on cycles, whichever is more conservative, toaccount for differences and uncertainties in fatigue life that areassociated with material and loading conditions.

Statistical models based on the existing fatigue S–N datahave been developed for estimating the fatigue lives ofpressure vessel and piping steels in air and LWR environments(Chopra and Shack, 1998b, 1999; Chopra, 1999; Chopra andSmith, 1998). In air at room temperature, the fatigue life N ofCSs is represented by

ln(N) = 6.564 – 1.975 ln(Ha – 0.113), (2a)

and of LASs by

ln(N) = 6.627 – 1.808 ln(Ha – 0.151), (2b)

where Ha is applied strain amplitude (%). In LWRenvironments, the fatigue life of CSs is represented by

ln(N) = 6.010 – 1.975 ln(Ha – 0.113) + 0.101 S* T* O* ÝH *,(3a)

and of LASs by

ln(N) = 5.729 – 1.808 ln(Ha – 0.151) + 0.101 S* T* O* ÝH *,(3b)

where S*, T*, O*, and ÝH * are transformed S, temperature, DO,and strain rate, respectively, defined as follows:

S* = 0.015 (DO > 1.0 ppm)S* = S (DO d 1.0 ppm & 0 < S d 0.015 wt.%)S* = 0.015 (DO d 1.0 ppm & S > 0.015 wt.%) (4a)

T* = 0 (T < 150°C)T* = T – 150 (T = 150–350°C) (4b)

O* = 0 (DO < 0.05 ppm)O* = ln(DO/0.04) (0.05 ppm d DO d 0.5 ppm)O* = ln(12.5) (DO > 0.5 ppm) (4c)ÝH * = 0 ( ÝH > 1%/s)ÝH * = ln( ÝH ) (0.001 d ÝH d 1%/s)ÝH * = ln(0.001) ( ÝH < 0.001%/s). (4d)

The discontinuity in the value of O* at 0.05 ppm DO is due toan approximation and does not represent a physicalphenomenon. In air at room temperature, the fatigue data forTypes 304 and 316 SS are best represented by Eq. 5a

ln(N) = 6.703 – 2.030 ln(Ha – 0.126), (5a)

and for Type 316NG, by Eq. 5b

24-9

ln(N) = 7.422 – 1.671 ln(Ha – 0.126). (5b)

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6

Design Curve Basedon Statistical ModelASME Code Curve

Str

ess

Am

plitu

de, S

a (M

Pa)

Number of Cycles, N

Carbon SteelRoom–Temp. AirVu = 551.6 MPa

Vy = 275.8 MPa

E = 206.84 GPa

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6

Design Curve Based on Statistical ModelASME Code CurveS

tres

s A

mpl

itude

, S

a (M

Pa)

Number of Cycles, N

Low–Alloy SteelRoom–Temp. Air

Vu = 689.5 MPa

Vy = 482.6 MPa

E = 206.84 GPa

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6 10 7

Statistical ModelASME Code Curve

Str

ess

Am

plitu

de, S

a (M

Pa)

Number of Cycles, N

Austenitic Stainless SteelRoom Temp. AirVu = 648.1 MPa

Vy = 303.4 MPa

E = 195.1 GPa

Fig. 4. Design fatigue curve developed from statistical modelin air at room temperature

In LWR environments, the fatigue data for Types 304 and 316SS are best represented by

ln(N) = 5.768 – 2.030 ln(Ha – 0.126) + T' ÝH ' O', (6a)

and for Type 316NG, by

ln(N) = 6.913 – 1.671 ln(Ha – 0.126) + T' ÝH ' O', (6b)

where T', ÝH ', and O' are transformed temperature, strain rate,and DO, respectively, defined as follows:

T' = 0 (T < 180°C)T' = (T – 180)/40 (180 d T < 220°C)T' = 1 (T t 220°C) (7a)

ÝH ' = 0 ( ÝH > 0.4%/s)ÝH ' = ln( ÝH /0.4) (0.0004 d ÝH d 0.4%/s)ÝH ' = ln(0.0004/0.4) ( ÝH < 0.0004%/s) (7b)

O' = 0.260 (DO < 0.05 ppm)O' = 0 (DO t 0.05 ppm). (7c)

The models are recommended for predicted fatigue lives ofd106 cycles.

The design fatigue curves were obtained from the best–fitcurves, represented by Eqs. 2a–3b for CSs and LASs, and byEqs. 5a and 6a for austenitic SSs. To be consistent with thecurrent Code design curves, the mean–stress–adjusted best–fitcurves were decreased by the same margins on stress andcycles that are imposed in the current Code curves, e.g., theadjusted best–fit curves were decreased by a factor of 2 onstress for CSs and LASs and by a factor of 1.5 for austeniticSSs. A factor of 20 on life was used for all of the curves,although the actual margin on life is 10–16 for SSs because ofthe differences between the ASME mean curve and the best–fitcurve to existing fatigue data.

The new design fatigue curves for CSs and LASs andaustenitic SS in air are shown in Fig. 4, whereas those in LWRcoolant environments are shown in Figs. 5–7; only the portionsof the environmentally adjusted curves that fall below thecurrent ASME Code curve are shown in Figs. 5–7. Becausethe fatigue life of Type 316NG is superior to that of Types 304or

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6

Statistical Model

ASME Code CurveStr

ess

Am

plitu

de, S

a (M

Pa)

Number of Cycles, N

Carbon SteelWaterWhen any one of the following conditions is true:Temp. <150°CDO <0.05 ppmStrain Rate •1%/s

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6

Statistical ModelASME Code Curve

Str

ess

Am

plitu

de, S

a (M

Pa)

Number of Cycles, N

Low–Alloy SteelWaterWhen any one of the following conditions is true:Temp. <150°CDO <0.05 ppmStrain Rate •1%/s

Fig. 5. Design fatigue curves developed from statistical modelunder service conditions where one or more criticalthreshold values are not satisfied

24-10

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6

0.10.010.001ASME Code Curve

Str

ess

Am

plitu

de, S

a (M

Pa)

Number of Cycles, N

Carbon SteelWaterTemp. 200°CDO 0.2 ppmSulfur •0.015 wt.%

Strain Rate (%/s)

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6


Str

ess

Am

plitu

de, S

a (M

Pa)

Number of Cycles, N

Low–Alloy SteelWaterTemp. 200°CDO 0.2 ppmSulfur •0.015 wt.%

Strain Rate (%/s)

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6


Str

ess

Am

plitu

de, S

a (M

Pa)

Number of Cycles, N

Carbon SteelWaterTemp. 250°CDO 0.2 ppmSulfur •0.015 wt.%

Strain Rate (%/s)

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6


Str

ess

Am

plitu

de, S

a (M

Pa)

Number of Cycles, N


Strain Rate (%/s)

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6

0.10.010.001ASME Code CurveS

tres

s A

mpl

itude

, S

a (M

Pa)

Number of Cycles, N

Carbon SteelWater

Temp. 288°CDO 0.2 ppmSulfur •0.015 wt.%

Strain Rate (%/s)

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6


Str

ess

Am

plitu

de, S

a (M

Pa)

Number of Cycles, N


Strain Rate (%/s)

Fig. 6. Design fatigue curves developed from statistical model for carbon and low–alloy steels under service conditions whereall critical threshold values are satisfied

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6 10 7

0.040.004Š0.0004

Str

ess

Am

plitu

de S

a (M

Pa)

Number of Cycles N

DO <0.05 ppm

•200°CStrain Rate (%/s)

ASME Code Design Curve

<180°C, All Strain Ratesor •220°C, 0.4%/s

10 2

10 3

10 1 10 2 10 3 10 4 10 5 10 6 10 7

Str

ess

Am

plitu

de S

a (M

Pa)

Number of Cycles N

DO •0.05 ppm

ASME Code Design Curve

All Temperatures & Strain Rates

Fig. 7. Design fatigue curves developed from statistical models for Types 304 and 316 SS in water with <0.05 and t 0.05 ppmDO

24-11

316 SS, the design curves in Figs. 4 and 7 will be somewhatconservative for Type 316NG SS. For carbon and low–alloysteels, a set of design curves similar to those shown in Fig. 6can be developed for low–S steels, i.e., steels withd 0.007 wt.% S. The results indicate that in room–temperature air, the current ASME Code design curves for CSsand LASs are conservative with respect to the curves based onthe statistical models, and those for austenitic SSs arenonconservative at stress levels above 300 MPa.

For environmentally adjusted design fatigue curves(Figs. 5–7), we define a minimum threshold strain, belowwhich environmental effects are modest. The threshold strainfor CSs and LASs appears to be §�� KLJKHU WKDQ WKH IDWLJXH

limit of the steel. This translates into strain amplitudes of0.140 and 0.185%, respectively, for CSs and LASs. Thesevalues must be adjusted for mean stress effects and variabilitydue to material and experimental scatter. The threshold strainamplitudes are decreased by §�� IRU CSs and by §�� IRU

LASs to account for the effects of mean stress, and by a factorof 1.7 on strain to provide 90% confidence for the variations infatigue life that are associated with material variability andexperimental scatter (Keisler et al., 1995). Thus, a thresholdstrain amplitude of 0.07% (or a stress amplitude of 145 MPa)is obtained for both CSs and LASs. The existing fatigue dataindicate a threshold strain range of §�� IRU DXVWHQLWLF SSs.This value is decreased by §�� WR DFFRXQW IRU PHDQ VWUHVV

effects and by a factor of 1.5 to account for uncertainties infatigue life that are associated with material and loadingvariability. Thus, a threshold strain amplitude of 0.097%(stress amplitude of 189 MPa) is obtained for austenitic SSs.The PVRC steering committee for CLEE (Yukawa, 1999) hasproposed a ramp for the threshold strain; a lower strainamplitude below which environmental effects are insignificant,a slightly higher strain amplitude above which environmentaleffects decrease fatigue life, and a ramp between the twovalues. The two strain amplitudes are 0.07 and 0.08% forcarbon and low–alloy steels, and 0.10 and 0.11% for austeniticSSs (both wrought and cast SS). These threshold values havebeen used to develop Figs. 6 and 7.

Fatigue Life Correction FactorThe effects of reactor coolant environments on fatigue life

have also been expressed in terms of a fatigue life correctionfactor Fen, which is the ratio of life in air at room temperatureto that in water at the service temperature (Higuchi and Iida,1991). A fatigue life correction factor Fen can be obtainedfrom the statistical model (Eqs. 2–7), where

ln(Fen) = ln(NRTair) – ln(Nwater). (8)

The fatigue life correction factor for CSs is given by

Fen = exp(0.554 – 0.101 S* T* O* ÝH *), (9a)

for LASs, by

Fen = exp(0.898 – 0.101 S* T* O* ÝH *), (9b)

and for austenitic SSs, by

Fen = exp(0.935 – T' ÝH ' O'), (9c)

where the constants S*, T*, ÝH * and O* are defined in Eqs. 4a–4d, and T', ÝH ' and O' are defined in Eqs. 7a–7c. A strainthreshold is also defined, below which environmental effectsare modest. The strain threshold is represented by a ramp, i.e.,a lower strain amplitude below which environmental effectsare insignificant, a slightly higher strain amplitude abovewhich environmental effects are significant, and a rampbetween the two values. Thus, the negative terms in Eqs. 9a–9c are scaled from zero to their actual value between the twostrain thresholds. The two strain amplitudes are 0.07 and0.08% for CSs and LASs, respectively, and 0.10 and 0.11% foraustenitic SSs (both wrought and cast SS). To incorporateenvironmental effects into the Section III fatigue evaluation, afatigue usage for a specific stress cycle, based on the currentCode design fatigue curve, is multiplied by the correctionfactor. The experimental data adjusted for environmentaleffects, i.e., the product of experimentally observed fatigue lifein LWR environments and Fen, are presented with the best–fitS–N curve in room–temperature air in Fig. 8.

A similar approach has been proposed by Mehta andGosselin (1996, 1998); however, they defined Fen as the ratioof the life in air to that in water, both at service temperature.The Fen approach, also known as the EPRI/GE approach, hasrecently been updated to include the revised statistical modelsand the PVRC discussions on environmental fatigueevaluations (Mehta, 1999). An “effective” fatigue lifecorrection factor, expressed as Fen,eff = Fen/Z, is defined whereZ is a factor that constitutes the perceived conservatism in theASME Code design curves. The Fen,eff approach presumesthat all uncertainties have been anticipated and accounted for.

CONCLUSIONSThe work performed at Argonne National Laboratory on

fatigue of carbon and low–alloy steels in LWR environments issummarized. The existing fatigue S–N data have beenevaluated to establish the effects of various material andloading variables such as steel type, strain range, strain rate,temperature, sulfur content in steel, orientation, and DO levelin water on the fatigue life of these steels. Statistical modelsare presented for estimating the fatigue S–N curves as afunction of material, loading, and environmental variables.Case studies of fatigue failures in nuclear power plants arepresented and the contribution of environmental effects oncrack initiation is discussed.

24-12

0.1

1.0

102 103 104 105 106 107

Str

ain

Am

plitu

de, H

a (

%)

Adjusted Fatigue Life, Fen x N25 (Cycles)

Carbon Steels

Statistical ModelRoom Temp. Air

0.1

1.0

102 103 104 105 106 107

Str

ain

Am

plitu

de,

Ha (

%)


Low–Alloy Steels


0.1

1.0

102 103 104 105 106 107

Str

ain

Am

plitu

de, H

a (

%)


Austenitic Stainless Steels


Fig. 8. Comparison of experimental data adjusted forenvironmental effects with best–fit fatigue S–N curvein room–temperature air

The current two methods for incorporating the effects ofLWR coolant environments into the ASME Code fatigueevaluations, i.e., the design fatigue curve method and thefatigue life correction factor method, are presented. Bothmethods are based on the statistical models for estimatingfatigue lives of carbon and low–alloy steels and austenitic SSsin LWR environments. Although estimates of fatigue livesbased on the two methods may differ because of differencesbetween the ASME mean curves used to develop the currentdesign curves and the best–fit curves to the existing data usedto develop the environmentally adjusted curves, either of thesemethods provides an acceptable approach to account forenvironmental effects.

The environmentally adjusted design fatigue curvesprovide allowable cycles for fatigue crack initiation in LWRcoolant environments. The new design curves maintain themargin of 20 on life. However, to be consistent with thecurrent ASME Code curves, the margin on stress is 2 forcarbon and low–alloy steels and 1.5 for austenitic SSs.

In the Fen method, environmental effects on life areestimated from the statistical models but the correction isapplied to fatigue lives estimated from the current Code designcurves. Therefore, estimates of fatigue lives that are based onthe two methods may differ because of differences in theASME mean curve and the best–fit curve to existing fatiguedata. The current Code design curve for CSs is comparable tothe statistical–model curve for LASs, whereas it is somewhatconservative at stress levels of <500 MPa when compared withthe statistical–model curve for CSs. Consequently, usagefactors based on the Fen method would be comparable to thosebased on the environmentally adjusted design fatigue curvesfor LASs and would be somewhat higher for CSs.

Figure 4 indicates that for austenitic SSs, the current Codedesign fatigue curve is nonconservative when compared withthe statistical–model curve, i.e., it predicts longer fatigue livesthan the best–fit curve to the existing S–N data. Therefore,usage factors that are based on the Fen method would be lowerthan those determined from the environmentally correcteddesign fatigue curves.

ACKNOWLEDGMENTSThis work was sponsored by the Office of Nuclear

Regulatory Research, U.S. Nuclear Regulatory Commission,Job Code W6610.

REFERENCESBamford, W. H., Rao, G. V., and Houtman, J. L., 1992,

“Investigation of Service Induced Degradation of SteamGenerator Shell,” Proc. 5th Intl. Symp. on EnvironmentalDegradation of Materials in Nuclear Power Systems –Water Reactors, American Nuclear Society, La GrangePark, IL.

Chopra, O. K., 1999, “Effects of LWR Coolant Environmentson Fatigue Design Curves of Austenitic Stainless Steels,”NUREG/CR–5704, ANL–98/31.

Chopra, O. K., and Gavenda, D. J., 1997, “Effects of LWRCoolant Environments on Fatigue Lives of AusteniticStainless Steels,” Pressure Vessel and Piping Codes andStandards, PVP Vol. 353, D. P. Jones, B. R. Newton, W.J. O'Donnell, R. Vecchio, G. A. Antaki, D. Bhavani, N. G.Cofie, and G. L. Hollinger, eds., American Society ofMechanical Engineers, New York, pp. 87–97.

Chopra, O. K., and Gavenda, D., J., 1998, “Effects of LWRCoolant Environments on Fatigue Lives of AusteniticStainless Steels,” J. Pressure Vessel Technol. 120, pp.116–121.

24-13

Chopra, O. K., and Shack, W. J., 1997, “Evaluation of Effectsof LWR Coolant Environments on Fatigue Life of Carbonand Low–Alloy Steels,” Effects of the Environment on theInitiation of Crack Growth, ASTM STP 1298, W. A. VanDer Sluys, R. S. Piascik, and R. Zawierucha, eds.,American Society for Testing and Materials, Philadelphia,pp. 247–266.

Chopra, O. K., and Shack, W. J., 1998a, “Low–Cycle Fatigueof Piping and Pressure Vessel Steels in LWREnvironments,” Nucl. Eng. Des. 184, pp. 49–76.

Chopra, O. K., and Shack, W. J., 1998b, “Effects of LWRCoolant Environments on Fatigue Design Curves ofCarbon and Low–Alloy Steels,” NUREG/CR–6583, ANL–97/18.

Chopra, O. K., and Shack, W. J., 1998c, “Fatigue CrackInitiation in Carbon and Low–Alloy Steels in Light WaterReactor Environments – Mechanism and Prediction,”Fatigue, Environmental Factors, and New Materials, PVPVol. 374, H. S. Mehta, R. W. Swindeman, J. A. Todd, S.Yukawa, M. Zako, W. H. Bamford, M. Higuchi, E. Jones,H. Nickel, and S. Rahman, eds., American Society ofMechanical Engineers, New York, pp. 155–168.

Chopra, O. K., and Shack, W. J., 1999, “Overview of FatigueCrack Initiation in Carbon and Low–Alloy Steels in LightWater Reactor Environments,” J. Pressure VesselTechnol., in press.

Chopra, O. K., and Smith, J., L., 1998, “Estimation of FatigueStrain–Life Curves for Austenitic Stainless Steels in LightWater Reactor Environments,” Fatigue, EnvironmentalFactors, and New Materials, PVP Vol. 374, H. S. Mehta,R. W. Swindeman, J. A. Todd, S. Yukawa, M. Zako, W.H. Bamford, M. Higuchi, E. Jones, H. Nickel, and S.Rahman, eds., American Society of MechanicalEngineers, New York, pp. 249–259.

Dooley, R. B., and Pathania, R. S., 1997, “Corrosion Fatigueof Water Touched Pressure Retaining Components inPower Plants,” EPRI TR–106696, Electric PowerResearch Institute, Palo Alto, CA.

Enrietto, J. F., Bamford, W. H., and White, D. F., 1981,“Preliminary Investigation of PWR Feedwater NozzleCracking,” Intl. J. Pressure Vessels and Piping, 9, pp.421–443.

Foley, W. J., Dean, R. S., and Hennick, A., 1991, “Closeout ofIE Bulletin 79–13: Cracking in Feedwater SystemPiping,” NUREG/CR–5258, US Nuclear RegulatoryCommission, Washington, DC.

Ford, F. P., 1986, “Overview of Collaborative Research intothe Mechanisms of Environmentally Controlled Crackingin the Low Alloy Pressure Vessel Steel/Water System,”Proc. 2nd Intl. Atomic Energy Agency Specialists'Meeting on Subcritical Crack Growth, NUREG/CP–0067,MEA–2090, Vol. 2, pp. 3–71.

Ford, F. P., Ranganath, S., and Weinstein, D., 1993,Environmentally Assisted Fatigue Crack Initiation inLow–Alloy Steels – A Review of the Literature and the

ASME Code Requirements, EPRI TR–102765,. ElectricPower Research Institute, Palo Alto, CA.

Fujiwara, M., Endo, T., and Kanasaki, H., 1986, “Strain RateEffects on the Low Cycle Fatigue Strength of 304Stainless Steel in High Temperature Water Environment,”Fatigue Life: Analysis and Prediction, Proc. of the Intl.Conf. and Exposition on Fatigue, Corrosion Cracking,Fracture Mechanics, and Failure Analysis, ASM, MetalsPark, OH, pp. 309–313.

Gavenda, D. J., Luebbers, P. R., and Chopra, O. K., 1997,“Crack Initiation and Crack Growth Behavior of Carbonand Low–Alloy Steels,” Fatigue and Fracture 1, Vol. 350,S. Rahman, K. K. Yoon, S. Bhandari, R. Warke, and J. M.Bloom, eds., American Society of Mechanical Engineers,New York, pp. 243–255.

Gordon, B. M., Delwiche, D. E., and Gordon, G. M., 1987,“Service Experience of BWR Pressure Vessels,”Performance and Evaluation of Light Water ReactorPressure Vessels, PVP Vol.–119, American Society ofMechanical Engineers, New York, pp. 9–17.

Hayashi, M., 1998, “Thermal Fatigue Strength of Type 304Stainless Steel in Simulated BWR Environment,” Nucl.Eng. Des. 184, pp. 135–144.

Hayashi, M., Enomoto, K., Saito, T., and Miyagawa, T., 1998,“Development of Thermal Fatigue Testing with BWRWater Environment and Thermal Fatigue Strength ofAustenitic Stainless Steels,” Nucl. Eng. Des. 184, pp.113–122.

Hickling, J. and Blind, D., 1986, “Strain–Induced CorrosionCracking of Low–Alloy Steels in LWR Systems – CaseHistories and Identification of Conditions Leading toSusceptibility,” Nucl. Eng. Des. 91, pp. 305–330.

Higuchi, M., 1995, presented at Working Group Meeting onS–N Data Analysis, the Pressure Vessel Research Council,June, Milwaukee.

Higuchi, M., 1996, presented at Working Group Meeting onS–N Data Analysis, the Pressure Vessel Research Council,April, Orlando, FL.

Higuchi, M., and Iida, K., 1991, “Fatigue Strength CorrectionFactors for Carbon and Low–Alloy Steels in Oxygen–Containing High–Temperature Water,” Nucl. Eng. Des.129, pp. 293–306.

Higuchi, M., and Iida, K., 1997, “Reduction in Low–CycleFatigue Life of Austenitic Stainless Steels in High–Temperature Water,” Pressure Vessel and Piping Codesand Standards, PVP Vol. 353, D. P. Jones, B. R. Newton,W. J. O'Donnell, R. Vecchio, G. A. Antaki, D. Bhavani,N. G. Cofie, and G. L. Hollinger, eds., American Societyof Mechanical Engineers, New York, pp. 79–85.

Higuchi, M., Iida, K., and Asada, Y., 1997, “Effects of StrainRate Change on Fatigue Life of Carbon Steel in High–Temperature Water,” Effects of the Environment on theInitiation of Crack Growth, ASTM STP 1298, W. A. VanDer Sluys, R. S. Piascik, and R. Zawierucha, eds.,

24-14

American Society for Testing and Materials, Philadelphia,pp. 216–231.

Iida, K., 1992, “A Review of Fatigue Failures in LWR Plantsin Japan,” Nucl. Eng. Des. 138, pp. 297–312.

Jaske, C. E., and O’Donnell, W., J., 1977, “Fatigue DesignCriteria for Pressure Vessel Alloys,” Trans. ASME J.Pressure Vessel Technol. 99, pp. 584–592.

Kanasaki, H., Hayashi, M., Iida, K., and Asada, Y., 1995,“Effects of Temperature Change on Fatigue Life ofCarbon Steel in High Temperature Water,” Fatigue andCrack Growth: Environmental Effects, Modeling Studies,and Design Considerations, PVP Vol. 306, S. Yukawa,ed., American Society of Mechanical Engineers, NewYork, pp. 117–122.

Kanasaki, H., Umehara, R., Mizuta, H., and Suyama, T.,1997a, “Fatigue Lives of Stainless Steels in PWR PrimaryWater,” Trans. 14th Intl. Conf. on Structural Mechanicsin Reactor Technology (SMiRT 14), Lyon, France, pp.473–483.

Kanasaki, H., Umehara, R., Mizuta, H., and Suyama, T.,1997b, “Effects of Strain Rate and Temperature Changeon the Fatigue Life of Stainless Steel in PWR PrimaryWater,” Trans. 14th Intl. Conf. on Structural Mechanicsin Reactor Technology (SMiRT 14), Lyon, France, pp.485–493.

Katada, Y., Nagata, N., and Sato, S., 1993, “Effect ofDissolved Oxygen Concentration on Fatigue CrackGrowth Behavior of A533 B Steel in High–TemperatureWater,” ISIJ Intl. 33 (8), pp. 877–883.

Katzenmeier, G., Kussmaul, K., Roos, E., and Diem, H., 1990,“Component Testing at the HDR-Facility for Validatingthe Calculation Procedures and the Transferability of theTest Results from Specimen to Component,” Nucl. Eng.Des. 119, pp. 317-327.

Keisler, J., Chopra, O. K., and Shack, W. J., 1995, “FatigueStrain–Life Behavior of Carbon and Low–Alloy Steels,Austenitic Stainless Steels, and Alloy 600 in LWREnvironments,” NUREG/CR–6335, ANL–95/15.

Keisler, J., Chopra, O. K., and Shack, W. J., 1996, “FatigueStrain–Life Behavior of Carbon and Low–Alloy Steels,Austenitic Stainless Steels, and Alloy 600 in LWREnvironments,” Nucl. Eng. Des. 167, pp. 129–154.

Kussmaul, K., Blind, D., and Jansky, J., 1984, “Formation andGrowth of Cracking in Feed Water Pipes and RPVNozzles,” Nucl. Eng. Des. 81, pp. 105–119.

Kussmaul, K., Rintamaa, R., Jansky, J., Kemppainen, M., andTörrönen, K., 1983, “On the Mechanism ofEnvironmental Cracking Introduced by Cyclic ThermalLoading,” IAEA Specialists Meeting Corrosion and StressCorrosion of Steel Pressure Boundary Components andSteam Turbines, VTT Symp. 43, Espoo, Finland, pp. 195–243.

Lenz, E., Liebert, A., and Wieling, N., 1990, “ThermalStratification Tests to Confirm the Applicability ofLaboratory Data on Strain Induced Corrosion Cracking to

Component Behavior,” 3rd IAEA Specialists Meeting onSub–Critical Crack Growth, Moscow, pp. 67–91.

Lenz, E., Stellwag, B., and Wieling, N., 1983, “The Influenceof Strain–Induced Corrosion Cracking on the CrackInitiation in Low–Alloy Steels in HT–Water – A RelationBetween Monotonic and Cyclic Crack InitiationBehavior,” IAEA Specialists Meeting Corrosion andStress Corrosion of Steel Pressure Boundary Componentsand Steam Turbines, VTT Symp. 43, Espoo, Finland, pp.243–267.

Majumdar, S., Chopra, O. K., and Shack, W. J., 1993,“Interim Fatigue Design Curves for Carbon, Low–Alloy,and Austenitic Stainless Steels in LWR Environments,”NUREG/CR–5999, ANL–93/3.

Mehta, H. S., 1999, “An Update on the EPRI/GEEnvironmental Fatigue Evaluation Methodology and itsApplications,” Probabilistic and Environmental Aspectsof Fracture and Fatigues, PVP Vol. 386, S. Rahman, ed.,American Society of Mechanical Engineers, New York,pp. 183–193.

Mehta, H. S., and Gosselin, S. R., 1996, “An EnvironmentalFactor Approach to Account for Reactor Water Effects inLight Water Reactor Pressure Vessel and Piping FatigueEvaluations,” Fatigue and Fracture Volume 1, PVP Vol.323, H. S. Mehta, ed., American Society of MechanicalEngineers, New York, pp. 171–185.

Mehta, H. S., and Gosselin, S. R., 1998, “EnvironmentalFactor Approach to Account for Water Effects in PressureVessel and Piping Fatigue Evaluations,” Nucl. Eng. Des.181, pp. 175–197.

Miller, K. J., 1985, “Initiation and Growth Rates of ShortFatigue Cracks,” Fundamentals of Deformation andFracture, Eshelby Memorial Symposium, CambridgeUniversity Press, Cambridge, U.K., pp. 477–500.

Miller, K. J., 1995, “Damage in Fatigue: A New Outlook,”International Pressure Vessels and Piping Codes andStandards: Volume 1 – Current Applications, PVP Vol.313–1, K. R. Rao and Y. Asada, eds., American Society ofMechanical Engineers, New York, pp. 191–192.

Mimaki, H., Kanasaki, H., Suzuki, I., Koyama, M., Akiyama,M., Okubo, T., and Mishima, Y., 1996, “Material AgingResearch Program for PWR Plants,” Aging ManagementThrough Maintenance Management, PVP Vol. 332, I. T.Kisisel, ed., American Society of Mechanical Engineers,New York, pp. 97–105.

NRC, 1979, “Cracking in Feedwater System Piping,” IEBulletin No. 79–13, U.S. Nuclear Regulatory Commission,Washington, DC.

NRC, 1988a, “Safety Injection Pipe Failure,” NRCInformation Notice 88–01, U.S. Nuclear RegulatoryCommission, Washington, DC.

NRC, 1988b, “Thermal Stresses in Piping Connected toReactor Coolant Systems,” NRC Bulletin No. 88–08, U.S.Nuclear Regulatory Commission, Washington, DC.

24-15

NRC, 1988c, “Pressurizer Surge Line Thermal Stratification,”NRC Bulletin No. 88–11, U.S. Nuclear RegulatoryCommission, Washington, DC.

NRC, 1993, “Thermal Fatigue Cracking of Feedwater Pipingto Steam Generators,” NRC Information Notice 93–20,U.S. Nuclear Regulatory Commission, Washington, DC.

Nagata, N., Sato, S., and Katada, Y., 1991, “Low–CycleFatigue Behavior of Pressure Vessel Steels in High–Temperature Pressurized Water,” ISIJ Intl. 31 (1), pp.106–114.

Nakao, G., Kanasaki, H., Higuchi, M., Iida, K., and Asada, Y.,1995, “Effects of Temperature and Dissolved OxygenContent on Fatigue Life of Carbon and Low–Alloy Steelsin LWR Water Environment,” Fatigue and CrackGrowth: Environmental Effects, Modeling Studies, andDesign Considerations, PVP Vol. 306, S. Yukawa, ed.,American Society of Mechanical Engineers, New York,pp. 123–128.

Ranganath, S., Kass, J. N., and Heald, J. D., 1982, “FatigueBehavior of Carbon Steel Components in High–Temperature Water Environments,” BWR EnvironmentalCracking Margins for Carbon Steel Piping, EPRI NP–2406, Electric Power Research Institute, Palo Alto, CA,Appendix 3.

Van Der Sluys, W. A., 1993, “Evaluation of the AvailableData on the Effect of the Environment on the Low CycleFatigue Properties in Light Water Reactor Environments,”Proc. 6th Intl. Symp. on Environmental Degradation ofMaterials in Nuclear Power Systems – Water Reactors, R.E. Gold and E. P. Simonen, eds., The MetallurgicalSociety, Warrendale, PA, pp. 1–4.

Ware, A. G., Morton, D. K., and Nitzel, M. E., 1995,“Application of NUREG/CR–5999 Interim Design Curvesto Selected Nuclear Power Plant Components,”NUREG/CR–6260, INEL–95/0045.

Watanabe, H., 1980, “Boiling Water Reactor FeedwaterNozzle/Sparger, Final Report,” NEDO–21821–A, GeneralElectric Co., San Jose, CA.

Yukawa, S., 1999, Meeting of the Steering Committee forCyclic Life and Environmental Effects (CLEE), thePressure Vessel Research Council, June, Columbus, OH.

25-1

25 MECHANISM OF FATIGUE CRACK INITIATION INLIGHT WATER REACTOR COOLANT ENVIRONMENTS

Omesh K. ChopraArgonne National Laboratory

Argonne, IL 60439 USA

Heung–Bae Park Korea Power Engineering Company, Inc.,

360–9 Mabukri, Kusongmyon, Yonginshi, Kyunggido, Korea

25-3

MECHANISM OF FATIGUE CRACK INITIATION INLIGHT WATER REACTOR COOLANT ENVIRONMENTS

Omesh K. Chopra1 and Heung–Bae Park2

1Argonne National Laboratory, Argonne, IL 60439 USA2Korea Power Engineering Company, Inc.,

360–9 Mabukri, Kusongmyon, Yonginshi, Kyunggido, Korea

KEYWORDSFatigue Crack Initiation, Strain–vs.–Life (S–N) Curve, LWR Environment, Crack Growth Rate, Slip Oxidation/Dissolution

ABSTRACTThe effects of a light water reactor (LWR) coolant

environment on the fatigue resistance of materials is notexplicitly addressed in the ASME Code design fatigue curves.Existing fatigue strain–vs.–life (S–N) data illustrate potentiallysignificant effects of LWR coolant environments on the fatiguelife of piping and pressure vessel steels. In this paper, wediscuss the influence of reactor environments on themechanism of fatigue crack initiation. Decreased fatigue livesof carbon and low–alloy steels in water with high dissolvedoxygen are caused primarily by the effects of environment onthe growth of short cracks that are <100 Pm deep. In LWRenvironments, the growth of these small cracks occurs by a slipoxidation/dissolution process. A fracture mechanics approachhas been used to evaluate the effects of environment on fatiguecrack initiation. The fatigue life, defined as the number ofcycles required to form an engineering–size crack, i.e., a 3–mm–deep crack, is considered to be composed of the growth ofmicrostructurally and mechanically small cracks. The growthof the latter is characterized in terms of 'J and crack–growth–rate (da/dN) data in air and LWR environments. The growthof microstructurally small cracks is expressed by a modifiedHobson relationship in air and by the slip oxidation/dissolutionmodel in water. The estimated fatigue S-N curves agree wellwith the experimental data for carbon and low–alloy steels inair and water environments.

INTRODUCTIONThe formation of surface cracks and their growth as shear

and tensile cracks to an engineering size (i.e., 3 mm deep)constitute the fatigue life of a material, which is represented bystress– or strain amplitude–vs.–fatigue life (S–N) curves.These curves define, for a given stress or strain amplitude, thenumber of cycles needed to form an engineering–size crack.

Cyclic loadings on a structural component occur becauseof changes in mechanical and thermal loadings as the systempasses from one load set (e.g., pressure, temperature, moment,and force loading) to any other load set. For each load set, anindividual fatigue usage factor is determined by the ratio of thenumber of cycles anticipated during the lifetime of thecomponent to the allowable cycles. Figures I–9.1 through I–9.6 of Appendix I to Section III of the ASME Boiler andPressure Vessel Code specify fatigue design curves that definethe allowable number of cycles as a function of applied–stressamplitude. The cumulative usage factor (CUF) is the sum ofthe individual usage factors, and the ASME Code Section IIIrequires that the CUF at each location must not exceed 1.

The current ASME Code design fatigue curves are basedon strain–controlled fatigue tests of small polished specimensin air at room temperature. The design fatigue curves havebeen obtained by first adjusting the best-fit curves to theexperimental data for mean stress effects and then decreasingthe adjusted curves by a factor of 2 on stress or 20 on cycles,whichever was more conservative, at each point on the curve.These factors were intended to account for the differences anduncertainties in relating fatigue lives of laboratory testspecimens to those of actual reactor components. The factorsof 2 and 20 are not safety margins but rather conversionfactors that must be applied to the experimental data to obtainreasonable estimates of the lives of actual reactor components.

The effects of light water reactor (LWR) coolantenvironments on fatigue resistance of a material are notexplicitly addressed in the Code design fatigue curves.Existing fatigue S–N data illustrate potentially significanteffects of LWR coolant environments on the fatigue resistanceof carbon steels (CSs), low–alloy steels (LASs) (Chopra andShack, 1997, 1998a, 1998b, 1999; Chopra and Muscara,2000), and austenitic stainless steels (Chopra and Gavenda,1998; Chopra and Smith, 1998; Chopra, 1999, Chopra and

International Conference on Fatigue of Reactor ComponentsJuly 31 - August 2, 2000

Napa California, USA

25-4

Muscara, 2000). The key parameters that influence fatigue lifein LWR environments are temperature, dissolved oxygen (DO)level in the water, loading or strain rate, and strain (or stress)amplitude; for carbon and low–alloy steels, the sulfur contentof the steel is also important. Under certain environmentaland loading conditions, the environmental effects alonesubstantially exceed the factor of 20 on life that is used toaccount for the differences between specimen tests andcomponent behavior.

The objective of this paper is to use crack growth data andfracture mechanics analysis to examine the fatigue S-Nbehavior of carbon and low-alloy steels in air and LWRenvironments. The influence of reactor environments on themechanism of fatigue crack initiation is discussed. Fatigue lifeis considered to be composed of the growth ofmicrostructurally small cracks (MSCs) and mechanically smallcracks. The growth of the latter has been characterized interms of the J–integral range 'J and crack–growth–rate (CGR)data in air and LWR environments. The growth ofmicrostructurally small cracks in air is expressed by a modifiedversion of the relationship presented by Hobson (1982) and bythe slip dissolution/oxidation process (Ford et al., 1993) inwater.

MECHANISM OF FATIGUE CRACK INITIATIONThe formation of surface cracks and their growth as shear

(Stage I) and tensile (Stage II) cracks to an engineering size (3mm deep) constitute the fatigue life of a material, which isrepresented by the fatigue S–N curves. The curves specify, fora given stress or strain amplitude, the number of cycles neededto form an engineering crack. During fatigue loading ofsmooth test specimens, surface cracks 10 Pm or longer formquite early in life (i.e., <10% of life) at surface irregularities ordiscontinuities either already in existence or produced by slipbands, grain boundaries, second–phase particles, etc. (Miller,1985; Tokaji et al., 1988; Gavenda et al., 1997; Obrtlik et al.,1997; Chopra and Shack, 1998a). Consequently, fatigue lifemay be considered to be composed entirely of crackpropagation (Miller, 1995).

Growth of these surface cracks may be divided into tworegimes; an initial period, which involves growth of MSCs,that is very sensitive to microstructure and is characterized bydecelerating crack growth (Region AB in Fig. 1), and apropagation period, that involves growth of mechanicallysmall cracks that can be predicted by fracture mechanicsmethodology and is characterized by accelerating crack growth(Region BC in Fig. 1). Mechanically small cracks, whichcorrespond to Stage II, or tensile, cracks are characterized bystriated crack growth and a fracture surface normal to themaximum principal stress. Conventionally, the former hasbeen defined as the initiation stage and is considered sensitiveto stress or strain amplitude, and the latter has been defined asthe propagation stage and is less sensitive to strain amplitude.

The characterization and understanding of both the crackinitiation and crack propagation stage are important foraccurate estimates of the fatigue lives of structural materials.

0 0.2 0.4 0.6 0.8 1

Cra

ck L

engt

h

Life Fraction

Microstructurally Small Crack (MSC)(Stage–I Shear Crack)

Mechanically Small Crack(Stage II Tensile Crack)

A

B

C

'V2

'V1

'V2 > 'V1

Crack Length

Microstructurally Small Crack

LEFM or EPFM

'V 1

Non–PropagatingCracks

'V3

'V2

'V3 > 'V 2 > 'V1

'V 1

Mechanically Small Crack

Fig. 1. Schematic illustrations of (a) growth of short cracks insmooth specimens as a function of fatigue life fractionand (b) crack velocity as a function of crack length.LEFM = linear elastic fracture mechanics; EPFM =elastic plastic fracture mechanics.

Reduction of fatigue life in high–temperature water hasoften been attributed to easier crack initiation, because surfacemicropits that are present in high–temperature water act asstress raisers and provide preferred sites for the formation offatigue cracks (Nagata et al., 1991). However, experimentaldata do not support this argument; the fatigue lives of carbonand low–alloy steel specimens that have been preoxidized at288°C in high–DO water and then tested in air are identical tothose of unoxidized specimens (Fig. 2) (Chopra and Shack,1998a). If the presence of micropits was responsible for thereduction in life, specimens preexposed to high–DO water andtested in air should show a decrease in life. Also, the fatiguelimit of these steels should be lower in water than in air. Dataobtained from specimens in high–DO water indicate that thefatigue limit is either the same as, or §�� KLJKHU� LQ ZDWHU

than in air (Chopra and Shack, 1998a, b).

25-5

0.1

1.0

10 1 10 2 10 3 10 4 10 5 10 6 10 7

F/FS/F

F/F in airF/F in <10 ppb DOS/F in <10 ppb DO

Tot

al S

trai

n R

ange

, 'H

t (%

)

Fatigue Life, N 25

Air

Strain Rates (%/s)S: 0.004 & F: 0.4

Preoxidized

A106–Gr B Steel288°C Water0.5–0.8 ppm DO

0.1

1.0

10 1 10 2 10 3 10 4 10 5 10 6 10 7

F/FS/F

F/F in airF/F in <10 ppb DO

Tot

al S

trai

n R

ange

, 'H

t (%

)

Fatigue Life, N 25

Air

Strain Rates (%/s)S: 0.004 & F: 0.4

Preoxidized

A533–Gr B Steel288°C Water0.5–0.8 ppm DO

Fig. 2. Effects of environment on formation of fatigue cracks in carbon and low–alloy steels. Preoxidized specimens wereexposed at 288°C for 30–100 h in water with 06–0.8 ppm dissolved oxygen.

-20

0

20

40

60

80

100

120

AirPWR>0.6 ppm DO

0.2 0.4 0.6 0.8 1.0

Num

ber

of C

rack

s

Strain Range (%)

A106 Gr–B Carbon Steel

Open Symbols: fast/fast testClosed Symbols: slow/fast or

fast/slow test

2.0-20

0

20

40

60

80

100

120

AirPWR>0.6 ppm DO

0.2 0.4 0.6 0.8 1.0

Num

ber

of C

rack

s

Strain Range (%)

A533 Gr–B Low–Alloy Steel

Open Symbols: fast/fast testClosed Symbols: slow/fast or

fast/slow test

2.0

Fig. 3. Number of cracks >10 Pm long along longitudinal section of fatigue specimens of (a) A106 Gr B carbon steel and(b) A533 Gr B low–alloy steel tested in various environments. Number of cracks represents the average value along a 7–mm gauge length.

Furthermore, if reduction in life is caused by easierformation of cracks, the specimens tested in high–DO watershould show more cracks. Figure 3 shows plots of the numberof cracks >10 Pm long, along longitudinal sections of thegauge length of A106–Gr B and A533–Gr B specimens as afunction of strain range in air, simulated PWR environment,and high–DO water at two strain rates. The results show thatwith the exception of the LAS tested in simulated PWR water,environment has no effect on the frequency (number per unitgauge length) of cracks. For similar loading conditions, thenumber of cracks in the specimens tested in air and high–DOwater is identical, although fatigue life is lower by a factor of§� LQ ZDWHU� 'HWDLOHG metallographic evaluation of the fatiguetest specimens indicates that the water environment has littleor no effect on the formation of surface microcracks.Irrespective of environment, cracks in carbon and low–alloysteels initiate along slip bands, carbide particles, or at theferrite/pearlite phase boundaries.

The enhanced growth rates of long cracks in pressurevessel and piping steels in LWR environments have been

attributed to either slip oxidation/dissolution (Ford, 1986) orhydrogen–induced cracking (Hänninen et al., 1986). Bothmechanisms depend on the rates of oxide rupture, passivation,and liquid diffusion. Therefore, it is often difficult todifferentiate between the two processes or to establish theirrelative contributions to crack growth in LWR environments.

Studies on crack initiation in smooth fatigue specimens(Gavenda et al., 1997) indicate that the decrease in fatigue lifeof CSs and LASs in LWR environments is caused primarily bythe effects of environment on the growth of cracks <100 Pmdeep. When compared with CGRs in air, growth rates inhigh–DO water are nearly two orders of magnitude higher forcracks that are <100 Pm deep and one order of magnitudehigher for cracks that are >100 Pm deep. Metallographicexamination of test specimens indicates that in high–DOwater, surface cracks <100 Pm–deep grow entirely as tensilecracks normal to the stress, whereas in air or simulated PWRenvironments, they are at an angle of 45° to the stress axis(Fig. 4) (Gavenda et al., 1997). Also, for CSs, cracks <100 Pmdeep propagate across both the soft ferrite and hard pearlite

25-6

(a) (b)Fig. 4. Photomicrographs of fatigue cracks along gauge sections of A106–Gr. B carbon steel in (a) air and

(b) high–DO water at 288°C

(a) (b)Fig. 5. Photomicrographs of fatigue cracks on gauge surfaces of A106–Gr. B low–alloy steel in (a) air and

(b) high–DO water at 288°C

regions, whereas in air, they propagate only along soft ferriteregions. The crack morphology on the specimen surface alsodiffers in air and water environments (Fig. 5); surface cracksin high–DO water are always straight and normal to the stressaxis, whereas in air or simulated PWR environments, they aremostly at 45° to the stress axis. The different crackmorphology, absence of Stage I crack growth, and propagationof near–surface cracks across pearlite regions indicate that inhigh-DO water, growth of MSCs occurs predominantly by theslip oxidation/dissolution process.

In high–DO water, crack initiation in CSs and LASs maybe explained as follows: surface microcracks form quite earlyin fatigue life. During cyclic loading, the protective oxide filmis ruptured at strains greater than the fracture strain of surfaceoxides, and the microcracks grow by anodic dissolution of thefreshly exposed surface to crack lengths greater than thecritical length of MSCs. These mechanically small cracksgrow to engineering size, and their growth, which ischaracterized by accelerating rates, can be predicted byfracture mechanics methodology.

FATIGUE S–N DATA IN LWR ENVIRONMENTSThe fatigue lives of both carbon and low–alloy steels are

decreased in LWR environments; the reduction in life dependson temperature, strain rate, DO level in the water, and sulfurcontent of the steel (Chopra and Shack, 1998a, 1998b, 1999;Chopra and Muscara, 2000; Higuchi and Iida, 1991; Higuchiet al., 1997; Higuchi, 1999). Statistical models based onexisting fatigue S–N data have been developed for estimatingfatigue lives of these steels in air and LWR environments(Chopra and Shack, 1998a, 1999; Chopra and Muscara, 2000).The effects of the critical parameters on fatigue life and theirthreshold values, as well as the statistical models forestimating fatigue life in air and LWR environments, havebeen presented in a companion paper at this conference.

INITIAL CRACK SIZEStudies on crack initiation in smooth fatigue specimens

indicate that surface cracks form quite early in life. Smith etal. (1996) detected 10–Pm deep surface cracks at temperaturesup to 700oC in Waspalloy. Hussain et al. (1994) examined the

25-7

growth of §��±Pm–deep surface cracks through four or moregrains. Tokaji et al. (1986a, b, 1988, 1992) defined crackinitiation as the formation of a 10–Pm–deep crack. Gavendaet al. (1997) reported that in room–temperature air, 10–Pm–deep cracks form early during fatigue life, i.e., <10% of fatiguelife. Suh et al. (1985, 1990) reported that a crack is said tohave initiated when any crack-like mark grows across a grainboundary, or when the separation of grain boundaries becomesclear. Based on these results, it is reasonable to assume theinitial depth of MSCs to be §�� Pm.

TRANSITION FROM MICROSTRUCTURALLY SMALLTO MECHANICALLY SMALL CRACK

Various criteria may be used to define the crack length fortransition from microstructurally small to mechanically smallcrack. They may be related to the plastic zone size, cracklength–vs.–fatigue life (a–N) curve, Weibull distribution of thecumulative probability of fracture, stress range–vs.–cracklength curve, or grain size. The results indicate that the cracklength for transition from MSC to mechanically small crackdepends on applied stress and microstructure of the material.Plastic Zone

de los Rios et al., (1992, 1996) and Lankford (1977, 1982,1985) defined the transition from small to large cracks as thecrack length at which the size of the linear elastic fracturemechanics (LEFM) plastic zone exceeds a grain diameter.Crack Length–vs.–Fatigue Life Curve

Obrtlik et al. (1997) divided the fatigue crack length a–vs.–fatigue life N curves into two regimes: MSCs, in which thedependence of crack length on fatigue life can be representedby a straight line; and mechanically small cracks, in whichfatigue crack growth is represented by an exponential functionfit of the experimental data.Weibull Distribution of the Cumulative Probability of Fracture

Suh et al. (1985, 1990) used the knee in the Weibulldistribution of cumulative probability of fracture to define thetransition from shear crack growth to tensile crack growth.The knee occurred in the range of 3–5 grain diameters.Stress Range–vs.–Crack Length Curve

Kitagawa and Takahashi (1976) and Taylor and Knott(1981) used the stress range–vs.–crack length curve todiscriminate a MSC from a mechanically small crack. Forcrack lengths >500 Pm, plots of the threshold stress range forfatigue crack growth ('Vth) vs. crack length yield a straightline, i.e., the threshold stress intensity factor ('Kth) isconstant. For crack lengths < 500 Pm, 'Vth deviates from thelinear relationship and approaches a constant value as thecrack length becomes smaller. The constant value of 'Vth isapproximately equal to the fatigue limit of a smooth specimenof the material. The crack length at which the 'Vth vs. cracklength curve changes from a linear relationship to a constant

value is used to define the transition from microstructurallysmall to mechanically small cracks.Grain Size

Tokaji et al. (1986a, 1988, 1992) estimated the transitioncrack length to be about 8 times the microstructural unit size.Ravichandran (1997) reported that large fluctuations in crackshape or aspect ratio occur at crack lengths of approximately afew grain diameters (typically five or fewer grain diameters).Hussain et al. (1994) observed that fatigue CGRs decreasedsystematically at microstructural heterogeneities up to a lengthof three or four grain diameters. Dowling (1977) reported thatthe J-integral correlation is not valid for surface crack lengths<10 crystallographic grain diameters.

The above studies indicate that the crack length fortransition from MSC to mechanically small crack is a functionof applied stress and microstucture of the material; actualvalue may range from 150 to 250 Pm. A constant value of§�� Pm was assumed for convenience, for both carbon andlow–alloy steels; it is the initial size for mechanically smallcracks.

FATIGUE CRACK GROWTH RATESAir Environment

The growth rates da/dN (mm/cycle) of MSCs, i.e., from 10to 200 Pm, in air can be represented by the Hobsonrelationship (Hobson, 1982; Miller, 1985; Brown, 1986;Carbonell and Brown, 1986)

da/dN = A1 'V� �n1 (d – a), (1)

where a is the length (mm) of the predominant crack, 'V is thestress range (MPa), constant A1 and exponent n1 aredetermined from existing fatigue S–N data, and d is thematerial constant related to grain size. The values of A1 andn1 for carbon and low–alloy steels at room temperature andreactor operating temperatures are given in Table 1. A valueof 0.3 mm was used for the material constant d. Also, becausegrowth rates increase significantly with decreasing cracklength, a constant growth rate was assumed for crack lengthssmaller than 0.075 mm. The applied stress range 'V isdetermined from Ramberg–Osgood relations given byEqs. A1–A5 of the Appendix, it represents the value at fatiguehalf–life.

Table 1. Values of the constants A1 and n1 in Equation 1Steel Type Temperature A1 n1

Carbon Room 3.33 x 10–41 13.13

Operating 9.54 x 10–34 10.03

Low–Alloy Room 1.45 x 10–36 11.10

Operating 1.07 x 10–43 13.43

25-8

The growth rates of mechanically small cracks in air areestimated from Eq. A8 of the Appendix. A factor of 1.22enhancement in growth rates was used at reactor operatingtemperatures.

LWR EnvironmentA model based on oxide film rupture and anodic

dissolution (or slip dissolution/oxidation model) was proposedby Ford et al., (1993) to incorporate the effects of LWRenvironments on fatigue lives of CSs and LASs. The modelconsiders that a thermodynamically stable protective oxidefilm forms on the surface to ensure that the crack willpropagate with a high aspect ratio without degrading into ablunt pit, and that a strain increment is required to rupture theoxide film, thereby exposing the underlying matrix to theenvironment. Once the passive oxide film is ruptured, crackextension is controlled by dissolution of freshly exposedsurfaces and by oxidation characteristics. Ford and Andresen(1995) proposed that the average crack growth rate da/dt(cm/s) is related to the crack tip strain rate Ý Hct (s–1) by therelationship

da/dt = A2 Ý Hct� �n2 , (2)

where the constant A2 and exponent n2 depend on the materialand environmental conditions at the crack tip. A lower limitof crack propagation rate is associated with blunting when thecrack tip cannot keep up with the general corrosion rate of thecrack sides or when a critical level of sulfide ions cannot bemaintained at the crack tip. The crack propagation rate atwhich this transition occurs may depend on the DO level, flowrate, etc. Based on these factors, the maximum and minimumenvironmentally assisted crack propagation rates have beendefined by Ford et al. (1993), Ford and Andresen (1995), andFord (1996). For crack–tip sulfide ion concentrations abovethe critical level, CGR is expressed as

da/dt = 2.25 x 10–4( Ý H ct )0.35; (3a)

for crack–tip sulfide ion concentrations below the critical level,it is expressed as

da/dt = 10–2( Ý H ct )1.0. (3b)

However, the growth rates predicted by Eqs. 3a and 3b aresomewhat higher than those observed experimentally(Gavenda et al., 1997). To be consistent with the experimentaldata, the constants in Eqs. 3a and 3b were decreased by afactor of 3.2 and 2.5, respectively. Assuming that Ý H ct isapproximately the same as the applied strain rate, Ý Happ , andcrack advance due to mechanical fatigue is insignificantduring the initial stages of fatigue damage, crack advance percycle from Eq. 3a for significant environmental effects is givenby

da/dN = 7.03 x 10–5('H – Hf)( Ý Happ )–0.65, (4a)

and from Eq. 3b for moderate environmental effects is given by

da/dN = 4.00 x 10–3('H – Hf), (4b)

where Ý Happ is the applied strain rate (s–1) and Hf is the thresholdstrain range needed to rupture the oxide film; Hf was assumedto be 0.0023 and 0.0029, respectively, for CSs and LASs. Forstrain rates >§��V� da/dN is lower from Eq. 4a than fromEq. 4b. Also, existing fatigue S–N data indicate that strainrate effects on life saturate at §��V �&KRSUD DQG 6KDFN�

1998a). Therefore, Eq. 4a can be applied at rates between0.003 and 0.00001 s–1, Ý

Happ is assumed to be 0.003 s–1 forhigher values, and 0.00001 s–1 for lower values. Equations 4aand 4b assume that the stress–free state for the surface oxidefilm is at peak compressive stress.

Studies on crack initiation and crack growth in smoothfatigue specimens indicate that the reference fatigue CGRcurves (Fig. A1 in the Appendix) for carbon and low–alloysteels in LWR environments are somewhat higher than thosedetermined experimentally from the growth of mechanicallysmall cracks in LWR environments (Gavenda et al., 1997).Furthermore, using the reference CGR curves and fracturemechanics analyses to examine the fatigue S-N behavior ofthese steels in LWR environments yields conservative results.Therefore, the reference fatigue CGR curves were modified toestimate the growth rates of mechanically small cracks; themodified curves are shown in Fig. 6. The threshold values of'K (MPa·m1/2) are given by

'Kb = 10.11 T��, (5a)

and

'Kc = 32.03 T0.326, (5b)

where rise time T is in seconds.

10 -6

10 -5

10 -4

10 -3

10 -2

10 -1

10 0

10 1 10 2

Cra

ck G

row

th R

ates

(m

m/c

ycle

)

'K (MPa·m 1/2)

Carbon & Low–Alloy Steels

'Kb 'Kc

Susceptible to EAC

da/dN = 5.67 x 10–8 'K3.07

Air Environment

da/dN = 3.78 x 10–9 'K3.07

Not susceptible to EAC

da/dN = 4.91 x 10–9 'K3.07

da/dN = 7.67 x 10–6 T 0.691 'K0.949

R = 0

T = 100 s

Fig. 6. Modified reference fatigue crack growth rate curvesfor carbon and low–alloy steels for LWR applications

25-9

10-2

10-1

100

101

10 100 1000

Cra

ck G

row

th R

ate

da/d

N (Pm

/cyc

le)

Crack Length (Pm)

Low–Alloy Steel288°C0.70% Strain Range0.01%/s Strain Rate

Air

Simulated PWR

Simulated BWR

Fig. 7. Crack growth rates during fatigue crack initiation inlow–alloy steels in air and simulated PWR and BWRenvironments

101

102

103

0 1000 2000 3000 4000 5000 6000 7000

AirBWR WaterPWR Water

Cra

ck L

engt

h (P

m)

Number of Cycles, N

Reactor Operating Temps.Strain Range 0.75%Strain Rate 0.1%/sOpen Symbols: Low–Alloy SteelClosed Symbols: Carbon Steel

101

102

103

0 1000 2000 3000 4000 5000 6000 7000

AirBWR WaterPWR Water

Cra

ck L

engt

h (P

m)

Number of Cycles, N

Reactor Operating Temps.Strain Range 0.75%Strain Rate 0.01%/sOpen Symbols: Low–Alloy SteelClosed Symbols: Carbon Steel

Fig. 8. Crack growth in carbon and low–alloy steels as afunction of fatigue cycles at two strain rates

Environmental effects on fatigue life are moderate whenany one of the threshold environmental conditions is notsatisfied, e.g., temperature <150°C, DO <0.05 ppm, strain rate>1%/s, or strain range is below the critical value. Formoderate environmental effects, the growth rates of

mechanically small cracks are represented by the curves formaterials not susceptible to environmentally assisted cracking(EAC), and those of MSCs, by either Eq. 4b or Eq. 1,whichever yields the higher value. For example, at high strainranges, growth rates determined from Eq. 1 can be higher thanthose determined from Eq. 4b, i.e., mechanical factors controlcrack growth and environmental effects are insignificant.

Environmental effects on fatigue life are significant whenall of the threshold conditions are satisfied, e.g., temperaturet 150°C, DO t 0.05 ppm, strain rate <1%/s, and strain rangeis above the critical value. A minimum threshold sulfurcontent of 0.005 wt.% was also considered, i.e., sulfur contentmust also be >0.005 wt.% for significant environmental effectson fatigue life. When all five threshold conditions aresatisfied, the growth rates of mechanically small cracks arerepresented by the curve for materials susceptible to EAC for'K values below 'Kb, by the curve for materials notsusceptible to EAC at 'K values above 'Kc, and by thetransition curve for in–between values of 'K. The growthrates of MSCs are represented by either Eq. 4a or Eq. 1,whichever yields the higher value.

ESTIMATES OF FATIGUE LIFEThe existing fatigue S–N data for carbon and low–alloy

steels in air and LWR environments were examined with thepresent model, in which fatigue life consists of the growth ofMSCs and mechanically small cracks. The former may bedefined as the initiation stage and represents the growth ofMSCs from 10 to 200 Pm. The growth of mechanically smallcracks may be defined as the propagation stage and representsthe growth of fatigue cracks from 200 to 3000 Pm. During theinitiation stage, the growth of MSCs is expressed by amodified Hobson relationship in air (Eq. 1) and by the slipdissolution/oxidation process in water (Eqs. 4a or 4b). Duringthe propagation stage, the growth of mechanically small cracksis characterized in terms of the J–integral range 'J and CGRdata in air and LWR environments (Fig. 3). The correlationsfor calculating the stress range, stress intensity range 'K, J–integral range 'J, and the CGRs for long cracks in air aregiven in the appendix. For the cylindrical fatigue specimens,the stress intensity ranges 'K were determined from the valuesof the J–integral range 'J. Although 'J is often computedonly for that portion of the loading cycle during which thecrack is open, in the present study, the entire hysteresis loopwas used when we estimated 'J (Dowling, 1977). The stressintensities associated with conventional cylindrical fatiguespecimens were modified according to the correlationsdeveloped by O'Donnell and O'Donnell (1995). Typical CGRsand crack growth behavior during fatigue crack initiation inair and water environments are shown in Figs. 7 and 8,respectively.

25-10

Experimental values of fatigue life and those predictedfrom the present model in air and low– and high–DO water areplotted in Fig. 9. The predicted fatigue lives in air showexcellent agreement with the experimental data; the predictedvalues in LWR environments, particularly in high–DO water,are slightly lower than the experimental values. Thedifferences in predicted and experimental fatigue lives in LWRenvironments are most likely due to crack closure effects thatare expected to be significant at low strain amplitudes. Thefatigue S–N curves developed from the present model andthose obtained from the statistical models in air and in PWRand BWR environments are shown in Figs. 10 and 11,respectively.

SUMMARYThe influence of reactor environments on the mechanism

of fatigue crack initiation is discussed. Decreased fatigue livesof carbon and low–alloy steels in high–DO water are caused

primarily by the effects of environment on the growth of smallcracks <100 Pm deep. In LWR environments, the growth ofthese small fatigue cracks in carbon and low–alloy steelsoccurs by a slip oxidation/dissolution process.

To predict the fatigue lives of carbon and low-alloy steelsin air and LWR environments, we used fracture mechanicsapproach in which we consider fatigue life to be composed ofthe growth of microstructurally and mechanically small cracks.The growth of the former cracks is very sensitive tomicrostructure and is characterized by decelerating crackgrowth, that of the latter, which can be predicted by fracturemechanics methodology, is characterized by accelerating crackgrowth, and has been characterized in terms of the J–integralrange 'J and CGR data in air and LWR environments. Thegrowth of MSCs is expressed by a modified Hobsonrelationship in air and by the slip dissolution/oxidation processin water. The crack length for transition frommicrostructurally small crack to mechanically small crackwas

10 1

10 2

10 3

10 4

10 5

10 6

10 7

Carbon SteelLow-Alloy Steel

10 1 10 2 10 3 10 4 10 5 10 6 10 7

Pre

dict

ed L

ife (

Cyc

les)

AirRoom Temperature

Observed Life (Cycles)

10 1

10 2

10 3

10 4

10 5

10 6

10 7

Carbon SteelLow-Alloy Steel

10 1 10 2 10 3 10 4 10 5 10 6 10 7

Pre

dict

ed L

ife (

Cyc

les)

AirOperating Temperature


10 1

10 2

10 3

10 4

10 5

10 6

10 7

<0.05 ppm DO•0.05 ppm DO

10 1 10 2 10 3 10 4 10 5 10 6 10 7

Pre

dict

ed L

ife (

Cyc

les)

Carbon SteelWaterOperating Temperature


10 1

10 2

10 3

10 4

10 5

10 6

10 7

<0.05 ppm DO•0.05 ppm DO

10 1 10 2 10 3 10 4 10 5 10 6 10 7

Pre

dict

ed L

ife (

Cyc

les)

Low–Alloy SteelWaterOperating Temperature


Fig. 9. Experimentally observed values of fatigue life of carbon and low–alloy steels vs. those predicted by the present modelin air and water environments

25-11

10-1

100

101

101 102 103 104 105 106

Statistical ModelPresent Model

Str

ain

Am

plitu

de, H

a (%

)

Number of Cycles, N

Air Carbon Steel Operating Temperture

10-1

100

101

101 102 103 104 105 106


Str

ain

Am

plitu

de, H

a (%

)

Number of Cycles, N

Air Low–Alloy Steel Operating Temperture

Fig. 10. Fatigue strain–vs.–life curves developed from the present model for carbon and low–alloy steels in air

10-1

100

101

101 102 103 104 105 106


Str

ain

Am

plitu

de, H

a (%

)

Number of Cycles, N

PWR EnvironmentCarbon Steel Operating Temperture

10-1

100

101

101 102 103 104 105 106


Str

ain

Am

plitu

de, H

a (

%)

Number of Cycles, N

PWR EnvironmentLow–Alloy Steel Operating Temperture

10-1

100

101

101 102 103 104 105 106


Str

ain

Am

plitu

de, H

a (

%)

Number of Cycles, N

BWR EnvironmentCarbon SteelOperating Temperture

Strain Rate 0.1%/s

Strain Rate 0.001%/s

10-1

100

101

101 102 103 104 105 106


Str

ain

Am

plitu

de, H

a (

%)

Number of Cycles, N

BWR EnvironmentLow–Alloy SteelOperating Temperture

Strain Rate 0.1%/s

Strain Rate 0.001%/s

Fig. 11. Fatigue strain–vs.–life curves developed from the present model for carbon and low–alloy steels in LWRenvironments

based on studies of small crack growth.. Fatigue livesestimated from the present model show good agreement withthe experimental data for carbon and low–alloy steels in airand LWR environments. At low strain amplitudes, i.e., fatiguelives of >104 cycles, the predicted lives in water are slightlylower than those observed experimentally, most likely becauseof the effects of crack closure.

ACKNOWLEDGMENTSThis work was sponsored by the Office of Nuclear

Regulatory Research, U.S. Nuclear Regulatory Commission,Job Code W6610.

APPENDIXCyclic Stress Range

The cyclic stress–strain response of carbon and low–alloysteels varies with steel type, temperature, and strain rate. Ingeneral, these steels exhibit initial cyclic hardening, followedby cyclic softening or a saturation stage. The CSs, with apearlite and ferrite structure and low yield stress, showsignificant initial hardening. The LASs, which consist oftempered ferrite and a bainite structure, exhibit a relativelyhigh yield stress, and show little or no initial hardening, mayexhibit cyclic softening at high strain ranges. At 200–370°C,these steels exhibit dynamic strain aging, which leads toenhanced cyclic hardening, a secondary hardening stage, and

25-12

negative strain rate sensitivity. Under the conditions ofdynamic strain aging, cyclic stress increases with decreases instrain rate.

The cyclic stress range vs. strain range relationship isexpressed by the modified Ramberg–Osgood relationship givenby

'H = 'V E� � + 'V A3� �n3 , (A1)

where E is Young’s modulus, constant A3 and exponent n3 aredetermined from the experimental data, and cyclic stress rangecorresponds to the value at half–life. At room temperature, therelationship of cyclic stress range 'V (MPa) to strain range'H �� for CSs may be represented by

'H = ('V/2010) + ('V/766.1)(1/0.207), (A2)

and for LASs, by

'H = ('V/2010) + ('V/847.4)(1/0.173). (A3)

The effect of strain rate on the cyclic stress–strain curve is notconsidered at room temperature. At 288°C, the cyclic stress–strain curves may be represented by the correlations developedby Chopra and Shack (1998a). For CSs, the curve is given bythe relationship

'H = ('V/1965) + ('V/Asig)(1/0.129), (A4a)

where Asig varies with the strain rate ÝH (%/s) expressed as

Asig = 1079.7 – 50.9 log( ÝH ). (A4b)

For LASs, the curve is given by the relationship

'H = ('V/1965) + ('V/Bsig)(1/0.110), (A5a)

where Bsig is expressed as

Bsig = 961.8 – 30.3 log( ÝH ). (A5b)

Stress Intensity Factor RangeFor cylindrical fatigue specimens, the range of stress

intensity factor 'K was determined from the value of the J–integral range 'J, which, for a small semicircular surfacecrack, is given by Dowling (1977) as

'J = 3.2 ('V2/2E) a + 5 ['V 'Hp/(S + 1)] a, (A6a)

where 'Hp is plastic strain range (%) (second term in theRamberg Osgood relationship) and S is the reciprocal of thestrain hardening exponent n in Eq. A1. The stress intensityfactor range 'K is obtained from

'K = (E 'J)1/2, (A6b)

where E is the elastic modulus. Equation A6a incorporates acombined surface and flaw shape correction factor Fs of 0.714,which is derived from equivalent linear elastic solutions;Eq. 3a is valid as long as the crack size is very small whencompared with the specimen diameter. For conventionalfatigue tests, life is defined as the number of cycles for thetensile stress to decrease 25% from the peak or steady–state

value, i.e., the crack–depth–to–specimen–diameter ratio can beas high as 0.4. Therefore, the geometrical correction factor Fsfor a small semicircular surface crack was modified accordingto the correlation developed by O'Donnell and O'Donnell(1995):

Fs = 0.6911 + 1.2685 (a/D) – 5.6638 (a/D)2 + 21.511 (a/D)3, (A7)

where D is specimen diameter. For conventional fatigue testson cylindrical specimens, Fs may increase up to 1.7.

The J–integral range 'J is calculated from the ranges ofcyclic stress and plastic strain, determined from stablehysteresis loops, i.e., at fatigue half–life. In general, 'J iscomputed only for that portion of the loading cycle duringwhich the crack is open. For fully reversed cyclic loading, thecrack opening point can be identified as the point where thecurvature of the load–vs.–displacement line changed before thepeak compressive load. In the present study, evidence of acrack opening point was observed for cracks that had grownrelatively large, i.e., near the end of fatigue life. Therefore, asrecommended by Dowling (1977), the entire hysteresis loopwas used in estimating 'J.

Crack Growth RateThe fatigue CGRs da/dN of structural materials are

characterized in terms of the range of applied stress intensityfactor 'K and are given in Article A–4300 of Section XI of theASME Boiler and Pressure Vessel Code. For a stress ratio Rin the range of –2 <R <0, the reference fatigue CGRs da/dN(mm/cycle) of carbon and low–alloys steels exposed to airenvironments are given by

da/dN = 3.78 x 10–9 ('K)3.07, (A8)

where 'K = Kmax, the maximum stress intensity factor(MPa·m1/2). However, the effect of temperature is notconsidered in Eq. A8; CGRs are generally higher at 288°Cthan at room temperature (Logsdon and Liaw, 1985). Theresults of Logsdon and Liaw indicate that for both CSs andLASs, CGRs are §�� KLJKHU DW ��& WKDQ DW URRP

temperature.Section XI of the ASME Code also includes CGR curves

for these steels exposed to LWR environments. The growthrates are represented by two curves for low and high values of'K. However, the curves do not consider the effects of loadingrate. Recent experimental results have shown the importanceof key variables of material, environment, and loading rate onCGRs in LWR environments. Fatigue CGR correlations havebeen developed to explicitly consider the effects of loadingrate, stress ratio R, 'K, and sulfur content in the steel (Easonet al., 1994). The new correlations, shown in Fig. A1, aredivided into two categories: (a) for materials not susceptible toenvironmental effects, e.g., when sulfur content in the steel islow, CGRs are a factor of 2.8 higher than those in air; and

25-13

(b) for materials susceptible to environmental effects, e.g.,when sulfur content in the steel is high, CGRs are defined interms of rise time T, stress ratio R, and 'K.

10 -6

10 -5

10 -4

10 -3

10 -2

10 -1

10 0

10 1 10 2

Cra

ck G

row

th R

ates

(m

m/c

ycle

)

'K (MPa·m 1/2)

Carbon & Low–Alloy Steels

'Ka 'Kb 'Kc 'Kd

Air Environment

da/dN = 3.78 x 10–9 'K3.07

Not susceptible to EAC

da/dN = 1.07 x 10–8 'K3.07

Susceptible to EAC

da/dN = 1.56 x 10–7 'K3.07

da/dN = 1.18 x 10–5 T 0.691 DK0.949

R = 0T = 100 s

Fig. A1. Proposed reference fatigue crack growth rate curvesfor carbon and low–alloy steels in LWR environmentsfor a rise time of 100 s and R = –1

The correlations in Fig. A1 correspond to a rise time of100 s and Kmin <0, e.g., fully reversed cyclic loading; R is setto zero. The various threshold values of 'K (MPa·m1/2) aregiven by

'Ka = 14.156 T0.125, (A9a)

'Kb = 7.691 T0.326, (A9b)

'Kc = 27.186 T0.326, (A9c)

'Kd = 44.308 T0.326, (A9d)where rise time T is in seconds.

REFERENCESBrown, M. W., 1986, “Interface Between Short, Long and

Non-Propagating Cracks,” Mechanical Engineering Pub.,pp. 423-439.

Carbonell, E. P., and Brown, M. W., 1986, “A Study of ShortCrack Growth in Torsional Low–Cycle Fatigue for aMedium Carbon Steel,” Fatigue Fract. Engng. Mater.Struct. 9, pp. 15-33.

Chopra, O. K., 1999, “Effects of LWR Coolant Environmentson Fatigue Design Curves of Austenitic Stainless Steels,”NUREG/CR–5704, ANL–98/31.

Chopra, O. K., and Gavenda, D., J., 1998, “Effects of LWRCoolant Environments on Fatigue Lives of AusteniticStainless Steels,” J. Pressure Vessel Technol. 120, pp.116–121.

Chopra, O. K., and Muscara, J., 2000, “Effects of Light WaterReactor Coolant Environments on Fatigue Crack Initiationin Piping and Pressure Vessel Steels,” Proc. 8th Intl.Conference on Nuclear Engineering, 2.08 LWR MaterialsIssue, Paper 8300, American Society of MechanicalEngineers, New York.

Chopra, O. K., and Shack, W. J., 1997, “Evaluation of Effectsof LWR Coolant Environments on Fatigue Life of Carbonand Low–Alloy Steels,” Effects of the Environment on theInitiation of Crack Growth, ASTM STP 1298, W. A. VanDer Sluys, R. S. Piascik, and R. Zawierucha, eds.,American Society for Testing and Materials, Philadelphia,pp. 247–266.

Chopra, O. K., and Shack, W. J., 1998a, Effects of LWRCoolant Environments on Fatigue Design Curves ofCarbon and Low-Alloy Steels, NUREC/CR-6583, ANL-97/18.

Chopra, O. K., and Shack, W. J., 1998b, “Low-Cycle Fatigueof Piping and Pressure Vessel Steels in LWREnvironments,” Nucl. Eng. Des. 184, pp. 49-76.

Chopra, O. K., and Shack, W. J., 1999, “Overview of FatigueCrack Initiation in Carbon and Low-Alloy Steels in LWREnvironments,” J. Pressure Vessel Technol. 121, pp. 49-60.

Chopra, O. K., and Smith, J., L., 1998, “Estimation of FatigueStrain–Life Curves for Austenitic Stainless Steels in LightWater Reactor Environments,” Fatigue, EnvironmentalFactors, and New Materials, PVP Vol. 374, H. S. Mehtaet al., eds., American Society of Mechanical Engineers,New York, pp. 249–259.

de los Rios, E. R., Navarro, A., and Hussain, K., 1992,“Microstructural Variations in Short Fatigue CrackPropagation of a C-Mn Steel,” Short Fatigue Cracks,ESIS 13, Mechanical Engineering Pub., pp. 115-132.

de los Rios., Wu, X. D., and Miller, K. J., 1996, “A Micro-Mechanics Model of Corrosion-Fatigue Crack Growth inSteels,” Fatigue Fract. Engng. Mater. Struct. 19, pp.1383-1400.

Dowling, N. E, 1977, “Crack Growth During Low-CycleFatigue of Smooth Axial Specimens,” ASTM STP 637,pp. 97-121.

Eason, E. D., Nelson E. E., and Gilman, J. D., 1994,“Modeling of Fatigue Crack Growth Rate for FerriticSteels in Light Water Reactor Environments,” ChangingPriorities of Code and Standards, PVP 286, ASME, NewYork, pp. 131–142.

Ford, F. P., 1996, “Quantitative Prediction of EnvironmentallyAssisted Cracking,” Corrosion 52, pp. 375-395.

Ford, F. P., and Andresen, P. L, 1995, Corrosion in NuclearSystems: Environmentally Assisted Cracking in LightWater Reactors, Marcel Dekker Inc., pp. 501-546.

Ford, F. P., Ranganath, S., and Weinstein, D., 1993,Environmentally Assisted Fatigue Crack Initiation in

25-14

Low–Alloy Steels – A Review of the Literature and theASME Code Requirements, EPRI Report TR–102765.

Gavenda, D. J., Luebbers, P. R., and Chopra, O. K., 1997,“Crack Initiation and Growth Behavior of Carbon andLow-Alloy Steels,” Fatigue and Fracture 1, PVP 350,ASME, New York, pp. 243-255.

Hänninen, H., Törrönen, K., and Cullen, W. H., 1986,“Comparison of Proposed Cyclic Crack GrowthMechanisms of Low Alloy Steels in LWR Environments,”Proc. 2nd Int. Atomic Energy Agency Specialists' Meetingon Subcritical Crack Growth, NUREG/CP–0067, MEA–2090, Vol. 2, pp. 73–97.

Higuchi, M., 1999, “Fatigue Curves and Fatigue DesignCriteria for Carbon and Low–Alloy Steels in High–Temperature Water,” Probabilistic and EnvironmentalAspects of Fracture Mechanics, PVP Vol. 386, S.Rahman, ed., American Society of Mechanical Engineers,New York, pp. 161–169.

Higuchi, M., and Iida, K., 1991, “Fatigue Strength CorrectionFactors for Carbon and Low–Alloy Steels in Oxygen–Containing High–Temperature Water,” Nucl. Eng. Des.129, pp. 293–306.

Higuchi, M., Iida, K., and Asada, Y., 1997, “Effects of StrainRate Change on Fatigue Life of Carbon Steel in High–Temperature Water,” Effects of the Environment on theInitiation of Crack Growth, ASTM STP 1298, W. A. VanDer Sluys, R. S. Piascik, and R. Zawierucha, eds.,American Society for Testing and Materials, Philadelphia,pp. 216–231.

Hobson, P. D., 1982, “The Formulation of a Crack GrowthEquation for Short Cracks,” Fatigue Fract. Engng. Mater.Struct. 5, pp. 323-327.

Hussain, K., Tauqir, A., Hashmi, S. M., and Khan, A. Q.,1994, “Short Fatigue Crack Growth Behavior in Ferrite-Bainitic Steel,” Metall. Trans. A 25A, pp. 2421-2425.

Kitagawa, H., and Takahashi, S., 1976, “Applicability ofFracture Mechanics to Very Small Cracks of the Cracks inthe Early Stage,” Proc. 2nd Int. Conf. on MechanicalBehavior of Materials, ASM, pp. 627-631.

Lankford, J., 1977, “Initiation and Early Growth of FatigueCracks in High Strength Steels,” Eng. Fract. Mech. 9, pp.617-624.

Lankford, J., 1982, “The Growth of Small Fatigue Cracks in7075-T6 Aluminum,” Fatigue Fract. Engng. Mater.Struct. 5, pp. 233-248.

Lankford, J., 1985, “The Influence of Microstructure on theGrowth of Small Fatigue Cracks,” Fatigue Fract. Engng.Mater. Struct. 8, pp. 161-175.

Logsdon, W. A., and Liaw, P. K., 1985, “Fatigue CrackGrowth Rate Properties of SA508 and SA533 PressureVessel Steels and Submerged Arc Weldments in Roomand Elevated Temperature Air Environments,” Eng. Frac.Mech. 22, pp. 509-526.

Miller, K. J., 1985, “Initiation and Growth Rates of ShortFatigue Cracks,” Fundamentals of Deformation andFracture, Cambridge Press, pp. 477-500.

Nagata, N., Sato, S., and Katada, Y., 1991, “Low–CycleFatigue Behavior of Pressure Vessel Steels in High–Temperature Pressurized Water,” ISIJ Intl. 31 (1), pp.106–114.

Obrtlik, K., Polak, J., Hajek, M., and Vasek, A., 1997, “ShortFatigue Crack Behavior in 316L Stainless Steel,” Int. J.Fatigue 19, pp. 471-475.

O'Donnell, T. P., and O'Donnell, W. J., 1995, “Stress IntensityValues in Conventional S-N Fatigue Specimens,”International Pressure Vessels and Piping Codes andStandards, PVP 313, pp. 195-197.

Ravichandran, K. S., 1997, “Effects of Crack Aspect Ratio onthe Behavior of Small Surface Cracks in Fatigue; Part II.Experiments on a Titanium (Ti-8Al) Alloy,” Metall.Trans. A 28A, pp. 157-167.

Smith, R. A., Liu, Y., and Garbowski, L., 1996, “Short FatigueCrack Growth Behavior in Waspaloy at Room andElevated Temperature,” Fatigue Fract. Engng. Mater.Struct. 19, pp. 1505-1514.

Suh, C. M., Lee, J. J., and Kang, Y. G., 1990, “FatigueMicrocracks in Type 304 Stainless Steel at ElevatedTemperature,” Fatigue Fract. Engng. Mater. Struct. 13,pp. 487-496.

Suh, C. M., Yuuki, R., and Kitagawa, H., 1985, “FatigueMicrocracks in a Low Carbon Steel,” Fatigue Fract.Engng. Mater. Struct. 8, pp. 193-203.

Taylor, D., and Knott, J. F., 1981, “Fatigue Crack PropagationBehavior of Short Cracks; The Effect of Microstructure,”Fatigue Fract. Engng. Mater. Struct. 4, pp. 147-155.

Tokaji, K., and Ogawa, T., 1992, “The Growth Behavior ofMicrostructurally Small Fatigue Cracks in Metals,” ShortFatigue Cracks, ESIS 13, Mechanical Engineering Pub.,pp. 85-99.

Tokaji, K., Ogawa, T., and Harada, Y., 1986a, “The Growth ofSmall Fatigue Cracks in a Low–Carbon Steel; The Effectof Microstructure and Limitation of Linear ElasticFracture Mechanics,” Fatigue Fract. Engng. Mater.Struct. 9, pp. 205-217.

Tokaji, K., Ogawa, T., Harada, Y., and Ando, Z., 1986b,“Limitation of Linear Elastic Fracture Mechanics inRespect of Small Fatigue Cracks and Microstructure,”Fatigue Fract. Engng. Mater. Struct. 9, pp. 1-14.

Tokaji, K., Ogawa, T., and Osako, S., 1988, “The Growth ofMicrostructurally Small Fatigue Cracks in a Ferrite-Bainitic Steel,” Fatigue Fract. Engng. Mater. Struct. 11,pp. 331-342.

26-1

26 ENVIRONMENTAL FATIGUE EVALUATION ONJAPANESE NUCLEAR POWER PLANTS

Hitoshi OhataPlant Operation Department,

Plant Management HeadquartersThe Japan Atomic Power Company

Tokyo, Japan

26-3

ENVIRONMENTAL FATIGUE EVALUATION ON JAPANESE NUCLEAR POWER PLANTS

ABSTRACT Fatigue evaluations considering LWR environmental effects on

key components for the oldest LWR in Japan, i.e., Tsuruga Unit 1,were carried out based on its operating experiences. Tsuruga Unit 1 isa BWR/2, non-jet pump type plant (see Table-1). A couple ofevaluation methods, i.e., Higuchi-Iida formula and the design fatiguecurves provided in NUREG/CR-6260, were adopted to calculateenvironmental fatigue usage factors. The evaluation results showedthat all calculated fatigue usage factors were well below the allowablelimit of 1.0.

From the evaluation results, it was concluded that the results aremainly due to the low frequency of forced shutdowns, and thatenvironmental fatigue will not be a significant safety issue even takinginto account 60 years of plant operation.

In this paper, the results of low-cycle thermal fatigue evaluationfor key components considering environmental effects are discussed,since it has been suggested that the fatigue life of components in theLWR water environment could be significantly shorter than that in anair environment.

Table-1 Outline of Tsuruga Unit 1Plant Tsuruga Unit 1Owner The Japan Atomic Power CompanyReactor Type BWR/2Main Contractor GEThermal Output (MWh) 1070Electric Output (Mwe) 357Construction permit Apr. 22, 1966First Criticality Oct. 03, 1969Commercial Operation Mar. 14, 1970

METHODOLOGY FOR ENVIRONMENTAL FATIGUEEVALUATION

LOCATIONS TO BE EVALUATEDWith respect to environmental fatigue evaluation, locations to be

evaluated for Tsuruga Unit 1 were selected in accordance with thefollowing steps.

In the first step, referring to the evaluation results in the designstage of the components of recent BWR plants, locations that aresubjected to low-cycle thermal fatigue among safety-related

components and plant availability-related components were selected.In accordance with MITI Code Notification No.501, which is similarto ASME Section III, low-cycle fatigue at those locations wasevaluated without considering environmental effects.

In the second step, the locations that could be subjected toenvironmental effects, or in contact with primary coolant, wereselected from the locations selected in the first step. Then, they wereevaluated considering environmental effects. In this evaluation,assuming that critical points in the air condition are also severe in thewater environment, only the severest point of each selected locationwas evaluated.

In accordance with the selection steps described above, the mainselected locations for environmental fatigue evaluation were asfollows.

• Reactor feed water nozzles (low alloy steel)• Reactor feed water piping and main steam piping (carbon

steel)• Recirculation piping (austenitic stainless steel)• Reactor core shroud and support (austenitic stainless steel and

Alloy 600)• PLR pump casing (austenitic stainless cast steel)

As an example of the evaluation locations, the reactor feed waternozzle is shown in Fig.1 and Fig.2.

Hitoshi OhataPlant Operation Department,

Plant Management HeadquartersThe Japan Atomic Power Company

Tokyo, Japan

26-4

Figure-1 Selection of the Evaluation Location on RPV in Tsuruga Unit 1 (BWR)

Figure-2 Evaluation Point on RPV Feedwater Nozzle inTsuruga Unit 1 (BWR)

THERMAL TRANSIENT CYCLESFor Tsuruga Unit 1 case, thermal transient cycles for

environmental fatigue evaluation were determined as following.First, various thermal transient events which had occurred in

Tsuruga Unit 1 since the pre-operational test period, were classifiedinto the design basis thermal transient events which were usuallyutilized in the thermal fatigue evaluation in the design stage. Then, theaverage frequency or event cycles per year for each of the design basisthermal transient events was calculated. Finally, the predicted numberof event cycles after 60 years of operation on each design basis eventwas obtained by linear extrapolation.

Both the design basis thermal transient cycles and the predictedthermal transient cycles in Tsuruga Unit 1 are shown in Table-2. Thistable shows that all of the predicted thermal transient cycles after 60years of operation are less than the design basis cycles.

Table-2 Thermal transient cycles for Tsuruga Unit 1 (BWR)Design basis thermal

transient event

Design basis

thermal

transient

cycles

actual thermal

transient

cycles in the

past (before

the end of

FY1993)

predicted

thermal

transient

cycles after 60

years of

operation

RPV Stud Bolt tensioning 123 28 69

Hydraulic pressure test 400 162 (49) 375(262)

Start-up (Heat-up) 240 92 (17) 208 (133)

Low power operation at

night (no smaller than 75%)

10,000 45 (8) 93 (56)

Low power operation at

weekend (less than 75%)

2,000 71 (26) 167 (122)

Rod pattern change 400 24 (15) 60 (51)

Loss of feed water heating

(by turbine trip)

10 0 (0) 0 (0)

Loss of feed water heating

(by feed water heater

bypass)

70 0 (0) 0 (0)

Reactor scram (by turbine

trip etc.)

187 58 (9) 90 (41)

Reactor scram (by feed

water pump trip)

10 4 (0) 9 (5)

Reactor scram (by improper

SRV actuation)

2 0 (0) 0 (0)

Shutdown (Cooling down) 238 92 (17) 208 (133)

RPV Stud Bolt

detensioning

123 28 69

Note: Figures in parentheses are applicable to the evaluation for the RPV feed water

nozzles. Since the inner surface metal of RPV feed water nozzles where thermal

fatigue had been accumulated was removed in 1985, predicted thermal transient

cycles after 60 years of operation were determined by using the actual thermal

transient cycles after metal removal..

EVALUATION BASED ON MITI CODE NOTIFICATION NO.501Fatigue evaluation without considering environmental effects for

Tsuruga Unit 1 was conducted in accordance with MITI CodeNotification No.501, as in the design stage.

ENVIRONMENTAL FATIGUE EVALUATION METHODOLOGYThe Higuchi-Iida formula was primarily selected as an

environmental fatigue evaluation method [1]. In case that the Higuchi-Iida formula is not applicable because material is carbon steel or lowalloy steel, or there is no data on the strain rate, the method inNUREG/CR-6260 was applied [2]. In the next article, details of theirapplicability of each method are described.

EVALUATION BASED ON HIGUCHI-IIDA FORMULAThe Higuchi-Iida formula is applicable to carbon steel and low

alloy steel. In addition, it is necessary to know the correspondingstrain rate at the location to be evaluated in order to calculate Fen

No. Locations in RPV Material Severity of Thermal SelectedType Transient Location

1 Vessel Head Spray Nozzle SUS Medium

2 Core Spray Nozzle SUS Small

3 Instrumantation Nozzle SUS Small

4 Recirculation Water Inlet SUS SmallNozzle

5 CRD Housing Stub Alloy Small600

6 ICM Housing Penatration SUS Small

7 Reactor Core Differntial SUS SmallPressure mesuring Nozzle

8 Recirculation Water Outlet SUS SmallNozzle

9 CRD Water Return LAS SmallNozzle

10 Feed Water Nozzle LAS Large �

11 Isolation Condenser SUS SmallNozzle

12 Main Steam Nozzle SUS Small

1

2

3

4

56

7

8

9

10

11

12

Nozzle (LAS)

Severest Point atFeedwater Nozzle

Safe End (CS)

Thermal Sleeve (Stainless Steel)Thermal Sleeve (Alloy 600)

C/ L

•

26-5

(fatigue life reduction factor for environmental effects) in the Higuchi-Iida formula by using the results of stress analysis.

As a result, in the environmental fatigue evaluation for TsurugaUnit 1, Higuchi-Iida formula was only applied to reactor feed waternozzle (low alloy steel).

EVALUATION BASED ON NUREG/CR-6260NUREG/CR-6260 is applicable to all kinds of material, therefore,

it is applicable to all evaluation locations, including the locationswhere the Higuchi-Iida formula was not applied. As a result,NUREG/CR-6260 was applied to the components which were made ofaustenitic stainless steel and Alloy 600, e.g., reactor core shroud andits support. In addition, NUREG/CR-6260 was also applied to thelocations where data on strain rates were not available, e.g.,recirculation system piping, feed water system piping and main steampiping.

EVALUATION RESULTSAfter the selection of the locations to be evaluated, determination

of the predicted number of thermal transient cycles after 60 years ofoperation, and the selection of environmental fatigue evaluationmethod, the environmental fatigue calculation was carried out (seeTable-3).

The results showed that all of the calculated CUFs, consideringenvironmental effects for Tsuruga Unit 1, were well below theallowable limit of 1.0 (see Table-4).

Table-3 Evaluation locations and applied evaluationmethods for Tsuruga Unit 1

Component/

Evaluation location

Selected

reason

Applied evaluation method and

evaluating condition

Method Higuchi-Iida formula

Temperature

(°C)

100 T 200

DO (ppm) 0.1 DO 0.2

Reactor Pressure

Vessel (RPV)/

Feed water nozzles

(low alloy steel)

Severest point

in RPV from

the viewpoint

of fatigue

Strain rate

(%/sec.)

Mean value for a pair

of strain peaks

Method NUREG/CR-6260

Temperature

(°C)

Applicable to all

temperatures

DO (ppm) Applicable to all DO

levels

Primary Loop

Recirculation (PLR)

system piping/

T joint, downstream

of PLR pump outlet

valve (austenitic

stainless steel)

Highest point

as to peak

stress intensity

Strain rate

(%/sec.)

0.001%/s curve in Fig.

3-18

Method NUREG/CR-6260

Temperature

(°C)

288°C

DO (ppm) 0.1

Main steam system

piping/

Main steam header

tee inside

Containment Vessel

(carbon steel)

Highest point

as to peak

stress intensity

Strain rate

(%/sec.)

Saturated curve in

Fig. 3-5

Table-4 CUFs after 60 years of operation consideringenvironmental effects

Component/ Evaluation location

CUFs after 60 years of

operation considering

environmental effects

RPV/Feed water nozzles 0.057

PLR piping/ T joint, downstream of PLR pump

outlet valve

0.146

Main steam piping/ Main steam header tee inside

Containment Vessel

0.449

DISCUSSIONAll of the calculated CUFs in the environmental fatigue

evaluation conducted for Tsuruga Unit 1 were well below theallowable limit of 1.0.

The reason for such favorable results is considered to come fromthe fact that Tsuruga Unit 1 has been operated steadily, for instance,the number of forced shutdowns were considerably less than designbasis thermal transient cycles as shown in Table-2.

Among the calculated CUFs considering environmental effectsafter 60 years of operation, the value of 0.449 for the main steampiping (carbon steel) in Table-4 seems to be relatively high. This isbecause, in this evaluation, the design fatigue curve in NUREG/CR-6260 for single-phase water was utilized instead of that in a water-steam two phase condition. This produced a significant fatigue lifereduction. Therefore, the actual CUF is believed to be less than thecalculated CUF.

Finally, as thermal transients utilized for stress analyses in thedesign stage are conservatively determined [3], it is believed that thereis further margin to the limit from the viewpoint of fatigue.

CONCLUSIONFrom the results of environmental fatigue evaluation for the

oldest LWR in Japan, i.e., Tsuruga Unit 1 (BWR), it is concluded thatthe low-cycle fatigue considering environmental effects will not be asignificant safety issue even taking into account 60 years of plantoperation.

REFERENCES[1] M. Higuchi and K. Iida, “Fatigue strength correction factors for

carbon and low-alloy steels in oxygen-containing high-temperaturewater” Nuclear Engineering and Design 129 (1991) 293-306, North-Holland

[2] A. G. Ware, D. K. Morton, M. E. Nitzel, “Application ofNUREG/CR-5999 Interim Fatigue Curves to Selected NuclearPower Plant Components”, NUREG/CR-6260, INEL-95/0045

[3] T. Sakai, K. Tokunaga, G. L. Stevens and S. Ranganath,“Implementation of Automated, On-Line Fatigue Monitoring in aBoiling Water Reactor”, PVP Volume 252, page 67-74, The 1993ASME Pressure Vessel and Piping Conference, Denver, Colorado,1993, ASME, New York

27-1

27 EVALUATION OF ENVIRONMENTAL EFFECTS ONFATIGUE LIFE OF PIPING

F.A. Simonen and M.A. KhaleelPacific Northwest National Laboratory

Richland, Washington 99352

D.O. Harris and D. DedhiaEngineering Mechanics Technology, Inc.

San Jose, California 95129

D.N. Kalinousky and S.K. ShaukatU.S. Nuclear Regulatory Commission

Washington, D.C. 20555

27-3

EVALUATION OF ENVIRONMENTAL EFFECTSON FATIGUE LIFE OF PIPING







27-4

EVALUATION OF ENVIRONMENTAL EFFECTSON FATIGUE LIFE OF PIPING







Abstract

Recent data indicate that light water reactor environments can significantly reduce thefatigue resistance of materials, and thereby show that design fatigue curves may beunconservative for reactor coolant environments. This paper describes calculations ofprobabilities of fatigue failures for a sample of components from five PWR and two BWRplants. The calculations used an enhanced version of the pc-PRAISE probabilistic fracturemechanics code. A wide range of failure probabilities were predicted, with somecomponents having end-of-life probabilities of through-wall crack of nearly 100 percentand others with probabilities less than 10-6. Inclusion of environmental effects (waterversus air) can increase calculated through-wall crack probabilities by as much as twoorders of magnitude. Sources of the uncertainties in the calculated probabilities wereidentified as coming from assumptions made in the fracture mechanics model itself andalso from the inputs to the model. Uncertain inputs include the design-basis data for cyclicstresses which could differ from the actual stresses occurring during plant operation, andassumptions regarding strain rates and environmental variables used for predicting offatigue crack initiation and growth. This paper describes sensitivity calculations thataddress the various sources of uncertainty.

Introduction

The American Society of Mechanical Engineers (ASME) Code Section III requires afatigue evaluation of the components of the reactor-coolant pressure boundary. Aspectsof the code fatigue methodology have come under review because recent test data indicatethat the effects of light-water reactor (LWR) environments could significantly reduce the

27-5

fatigue resistance of materials and show that the ASME design fatigue curves may notalways be conservative for nuclear power plant primary system environments. As a resultArgonne National Laboratory (ANL) has developed revised fatigue curves based onlaboratory test data from small, polished specimens cycled to failure in water withtemperatures, pressures, and chemistries that simulate LWR conditions (NUREG/CR-6335) (Keisler et al. 1995).

Using curves developed by ANL, the Idaho National Engineering Laboratory (INEL)investigated the significance of the interim fatigue curves as published in NUREG/CR-5999 (Majumdar et al. 1993) by performing deterministic fatigue evaluations for a sampleof components in the reactor coolant pressure boundary of LWRs. Cumulative usagefactors (CUFs) for each component were reported NUREG/CR-6260.

ANL has also developed statistical models for estimating the effects of various material,loading, and environmental conditions on the fatigue life of these materials. Probabilisticfatigue strain versus life (S-N) data for carbon steel (CS), low-alloy steel (LAS), andaustenitic stainless steels (SS) were published in NUREG/CR-6237. In the present paperthe statistical models from the ANL work were used by Pacific Northwest NationalLaboratory to estimate the probability of fatigue-crack initiation. The objective was tocalculate component-failure probabilities rather than deterministic usage factors.

The methodology and results of the PNNL study are documented in NUREG/CR-6674.This NUREG report includes a description of the plants and components that areaddressed by the fatigue analyses. This is followed by a discussion the fracture-mechanicsmethodology along with documentation of the probabilistic equations from ANL that wereused to predict the number of cycles to crack initiation. Another section of the reportfocuses on the consequences of small and large leaks and describes how calculations wereperformed to estimate core damage frequencies (CDF). These evaluations of CDFs areoutside the scope of the present paper. The final section of the NUREG reportsummarizes the results in terms of absolute and relative failure probabilities, givingparticular attention to how these calculated probabilities differ for a 40-year versus 60-year plant life. Failure probabilities for water versus air environments are then compared.

Appendices to NRUGEG/CR-6674 describe actual inputs and results of the calculations.Details of the modified pc-PRAISE code and a review the accuracy of crack-tip stress-intensity factors calculated by pc-PRAISE are documented in other appendices.Discussions of the fracture-mechanics model include assumptions made to account for theeffects of through-wall stress gradients on crack propagation and describe methods usedto estimate the fractions of through-wall cracks that become small leaks and large leaks.A final appendix of NURER/CR-6674 describes sensitivity calculations that evaluateeffects of uncertainties in the fracture-mechanics calculations, and calculations thatexercise the cracking linking model to show how calculated crack lengths change whenthe inputs to the linking model are changed.

27-6

The present paper begins with a summary of the probabilistic fracture mechanicsmethodology and the enhanced version of the pc-PRAISE computer code. Results andconclusions from the component specific calculations are then presented. The remainderof the paper is devoted to sensitivity calculations that are helpful in reconciling therelatively high values of calculated failure probabilities with more favorable serviceexperience that has not shown evidence of cracking or leakage.

Methodology for Through-Wall Crack Calculations

The calculations estimated the probability that fatigue cycles will result in through-wallcracks and leaks of various sizes in pressure boundary components of reactor coolantsystems (RCSs) of PWR and BWR plants. Only the contribution of initiated fatiguecracks was addressed, which excluded the contributions of pre-existing cracks. Themethodology consisted of two parts. The first part calculated the probability that a fatiguecrack (3 mm deep) will initiate as a function of time over the life of the plant. The secondpart evaluated the probability that these initiated cracks will grow to become through-wallcracks. Additional calculations for probabilities of large versus small leaks and for coredamage frequencies are described in NUREG/CR-6674. These calculations are beyondthe scope of the present paper.

Stress amplitudes and the numbers of stress cycles for the selected components during a40-year plant life were taken directly from NUREG/CR-6260 (Ware et al. 1995). Thetypes of transients and cyclic stresses for the 60-year plant life were assumed to be thesame as those during the 40-year plant life. The 60-year number of accumulated cycleswas calculated by multiplying the 40-year number of cycles by a factor of 1.5. Thenumber of cycles to crack initiation was a function of the material type, water/airenvironment, temperature, dissolved oxygen content, sulfur content and strain rate. Thematerial types were carbon steel, low-alloy steel, 304/316 austenitic stainless steel and316NG stainless steel. The statistical models of NUREG/CR-6335 (Keisler et al. 1995)were used to calculate the number of cycles to crack initiation corresponding to givenprobabilities (or percentiles) of the material S-N curves.

The ANL statistical distributions predicted the number of cycles to initiate a 3-mm crackfor a given cyclic stress amplitude. The parameters of the lognormal probabilistic fatigueinitiation curves were based on the ANL revised fatigue curves published in NUREG/CR-6335 (Keisler et al. 1995).

The crack propagation was assumed to start from a 3-mm deep initiated flaw, which canwas then grown to the critical size (through-wall) for component failure. This 3-mm sizewas the estimated crack size that can give a measurable 25 percent load drop in the testingof standard fatigue specimens. Sensitivity calculations were performed to evaluate theeffect of changing this crack depth from 3 mm to 2 mm or 4 mm. The resulting changes inthe calculated probabilities of through-wall cracks were at most a factor of two. It wasdecided not to include the initial crack depth as a simulated variable, in part because the

27-7

uncertainty in the initial crack depth is indirectly captured by the statistical scatter in thedata based on the load drop approach used to define crack initiation.

The cyclic stress levels from the INEL report were used to calculate both crack initiationand crack growth. These stresses included effects of stress concentrations in a mannerprescribed by the ASME Code approach of stress indices. In many cases the stress indicesaddressed very high local stresses (e.g. weld root stress concentrations) with values up to2.0. Because such surface stresses are not indicative of internal stress levels remote fromthe stress concentration, adjustments were made for deeper cracks to account for theeffects of the through-wall stress. Cyclic stresses were broken into a component ofuniform stress and a component of thermal gradient stress in accordance with a set ofrules.

Fatigue Crack Initiation

The present work used a crack-initiation model (NUREG/CR-6335) (Keisler et al. 1995).This model estimates the probability of initiating a 3-mm deep fatigue crack based onexisting fatigue (S-N) data, foreign and domestic, for carbon, low-alloy and stainless steelsused in the construction of nuclear power plant components.

ANL Crack-Initiation Correlation

Only data obtained on smooth specimens tested under fully reversed loading conditionswere considered. A statistical distribution was fitted by ANL to S-N data to describe thescatter in the fatigue data. The ANL statistical distributions of cycles to initiate a 3-mmcrack for a given cyclic stress were lognormal. The parameters of the probabilistic fatigueinitiation curves were based on the ANL revised fatigue curves published in NUREG/CR-6335 (Keisler et al. 1995). The equations for stainless steels included recent updates forfatigue life correlations provided by ANL.

The equations for cycles to failure were coded into a Fortran subroutine forimplementation into probabilistic fracture mechanic codes such as pc-PRAISE. Thecalling program needs to provide values for the stress amplitude, the material type, thesulfur content (for ferritic steels), temperature, whether the environment is water or air,oxygen content of the water, and the strain rate for the stress cycle. Figure 1 wasgenerated from data obtained from a series of calls to the subroutine. Each of the curvescorresponds to the indicated percentile of data having cycles to failure less than or equal tothe indicated percentile. The solid curve of Figure 1 is the median or 50th percentile curvefor cycles to crack initiation. Inputs for strain rates, oxygen, and sulfur were all assignedas bounding values that are unlikely to be present simultaneously at these maximum valuesfor any given component.

The enhanced version of pc-PRAISE addresses crack initiation at multiple sites bysubdividing the pipe circumference into a set of 5.08-cm (2-in.)-long zones. Theamplitude of cyclic stresses at each site can vary in a manner specified by user input such

27-8

that the fatigue cracks may initiate at some sites much sooner or later than at other sites.The model also assumes no correlation between the random scatter in crack initiationtimes from one site to the next. Thus, a different selection from the family of S-N curves(as shown by the example of Figure 1) is sampled at random for each of the various sitesaround the pipe circumference.

Treatment of Size Effects

The equations developed by ANL to predict probabilities of fatigue-crack initiation arebased on a statistical treatment of data from small specimen tests. An additional term ofln(4) is included to bring the equation into empirical agreement with some test data on22.86-cm (9-in.)-diameter vessels. This term is intended to account for size, geometry,and surface-finish differences between small fatigue test specimens and actual components.

The present calculations made use of the ANL equation, including the ln(4) term, for thosecases in which the model assumed only one initiation site. However, the revised pc-PRAISE model accounts for multiple initiation sites with each site covering some 5.08 cm(2 in.) of the pipe circumference. The probability of crack initiation therefore increases asthe number of specified initiation sites is increased. This means that the fracture-mechanicsmodel itself indirectly accounts for size effects, and inclusion of the ln(4) term in the ANLequation can result in a double counting of size effects.

The ln(4) term of the ANL equation was modified when used to address crack initiation atmultiple sites. The pc-PRAISE multiple-site model was calibrated to achieve agreement ofcalculated cycles to crack initiation with experimental data from the tests of the 22.86-cm(9-in.)-diameter vessels described in the ANL reports. The conclusion from thiscalibration effort was that the cycles to failure from the ANL equation needed to beincreased by a factor of about 3.0. The net result was a factor of 3/4 applied to thenumber of cycles to failure from the small specimen data. In contrast, the ANL equationuses a factor of 1/4, but bases the fatigue-life prediction on consideration of a singleinitiation site.

Fatigue Crack Growth

As for the calculations for crack initiation, the calculations for fatigue crack-growth werebased on data that accounted for the effects of environment on the growth rates. Theequations for fatigue crack growth rates were unchanged from the prior version of the pc-PRAISE code (Harris and Dedhia 1992). These equations did not address the specificfactors that enhance the crack-growth rates in the same level of detail and rigor as in theArgonne correlations for crack initiation.

27-9

Treatment of Correlations

It was assumed that the random variations in fatigue crack-growth rates were notcorrelated with the corresponding random variations in the cycles to crack initiation. Ifsuch correlations were to exist, the predictions for probabilities of through-wall crackscould be unconservative as indicated by sensitivity calculations described below. Thesimplifying assumption greatly facilitated the calculations, and has a good technical basisbecause the technical literature (Wire and Li 1996) provides evidence to support theassumption of independence. In general, crack initiation and crack growth involvedifferent material damage mechanisms, such that environment and loading rates affect themechanisms for crack initiation and growth differently.

It was assumed that crack growth occurred under conditions of zero R-ratio. While thisassumption will be conservative for transients with very high stress amplitudes, crack-growth rates for cases of high cycle/low stress fatigue could be underestimated. The pc-PRAISE model for predicting multiple crack initiations from site-to-site around thecircumference of a given weld also assumed that random variations in the number ofcycles to crack initiation were uncorrelated from site-to-site.

Treatment of Through-Wall Stress Gradients

The inputs for cyclic stress were the same stresses that were used in the NRC-fundedresearch project at INEL as described in NUREG/CR-6260 (Ware et al. 1995). The datagave only peak stresses for the surface locations at which the initiation of fatigue crackswas to be evaluated and did not describe the corresponding variations of the stressesthrough the section thickness. It was appropriate to use these peak stresses for theinitiation aspect of fatigue cracking. However, it was unrealistic to always assume thatthese peak stresses were uniformly distributed through the component thickness.

The pc-PRAISE calculations used a number of conservative assumptions and inputs,which were in part balanced by unconservative assumptions and inputs. The inputs forstress cycles were taken from the INEL report (NUREG/CR-6260 [Ware et al. 1995]) andare believed in most cases to conservatively bound the stresses experienced during actualplant operation. These stresses assumed bounding conditions for the severity of thermaltransients and other loads. The method used by INEL to derive load pairs from thetransients assumed a worst-case sequencing of loads. The method used to estimatethrough-wall stress distributions (uniform tension versus through-wall gradient) wasintended to overestimate the stress assigned to the uniform tension category.

The approach used for the present calculations was to decompose the peak stress into acomponent of uniform stress and a component of through-wall gradient stress. Details ofthe approach are described in NUREG/CR-6674. A standardized (quadratic) stressgradient was developed on the basis of stress solutions for heating and cooling ramps andstep changes in surface temperatures.

27-10

Since results of detailed stress calculations were not available from the work ofNUREG/CR-6260 (Ware et al. 1995), rules were developed to assign a fraction of thepeak stress to the uniform stress category. The remaining fraction was assigned to thethrough-wall gradient category. In many cases, the values of peak stresses were greaterthan 100 ksi, which implied that most of the stress was due to heating and coolingtransients or were due to geometric stress concentrations. In other cases, the number ofstress cycles was very large, which also suggested thermal transients. Anotherconsideration was that the ASME code limits do not permit membrane stresses (includingsecondary stresses) to be more than three times the code design stress (i.e. < 3Sm). Fortypical piping materials, the 3Sm limit implied that all stress ranges (or 2Sa) greater than 45ksi should be treated as gradient stresses.

The rules used to assign stresses to the uniform and gradient categories were:

• seismic stresses were treated as 100 percent uniform stress.

• stresses greater than 45 ksi were treated as having a uniform component of 45 ksi withthe remainder being assigned to the gradient category.

• For transients with more than 1000 cycles over a 40-year life, it was assumed that 50%of the stress was uniform stress and 50% through-wall gradient stress. In addition theuniform stress component was not permitted to exceed 10 ksi.

These rules were less conservative than assuming uniform stresses through the fullcomponent sections. The approach ensured that shallow initiated cracks would at first besubjected to the peak surface stresses, but allowed for a reduction in crack-growth rates asthe cracks grow into regions of lower stress levels.

Crack and Linking of Cracks at Multiple Sites

The enhanced version of pc-PRAISE simulated the initiation of fatigue cracks at multiplesites around the circumference of a weld, a phenomenon often seen in service inducedcracking of pipe welds. Details of the aspect of the fracture mechanics model aredescribed in NUREG/CR-6674. The crack growth calculations simulate the growth(depth and length) of the individual cracks and combine adjacent cracks into a single largercrack in accordance with proximity rules. One of the sensitivity calculations describedbelow provides an example of the crack linking.

27-11

Treatment of Inservice Inspections

The calculations of through-wall crack frequencies did not account for potential benefitsof inservice inspections or maintenance programs, even though the predicted cumulativeprobabilities of failure for many of the components attained levels late in plant life thatexceeded 50 percent. On the other hand, the pc-PRAISE model does take credit for l eakdetection, but this only decreased the probability that short through-wall cracks withinconsequential leaks will grow over time to cause much larger leaks. Leak detection hadno effect on the probabilities of through-wall cracks.

Calculations for Selected Components

PNNL performed calculations for the five pressurized-water reactor (PWR) plants andtwo boiling-water reactor (BWR) plants listed in Table 1 and for the components listed inTable 2.

Inputs to Calculations

The fracture-mechanics calculations were based on data compiled by INEL from stressreports for actual plants. However, NUREG/CR-6260 (Ware et al. 1995) did not revealthe identities of these plants. For each plant, four to nine locations were investigated,including locations within the reactor pressure vessel. Calculations with the pc-PRAISEpredicted probabilities of crack initiation and probabilities of through-wall cracks as afunction of time for plant operating periods up to 60 years.

For the PWR plants, the curves for high-sulfur steel (0.015 weight percent) and a low-oxygen environment (0.01-ppm) were used. For the BWR plants, the curves for highsulfur steel and a high-oxygen environment (0.10 ppm) were used. The strain rates forboth PWR and BWR components (low alloy and carbon steel) were assumed to be0.001% (see NUREG/CR-6260) (Ware et al. 1995). For 316 stainless steel, the strain ratewas 0.004%. For all components, the temperature was assumed to be 290 C.

Summary of Results

Table 3 provides the final results for all the components. Results are given for both a 40-year and a 60-year operating period. Many of the components have cumulativeprobabilities of both crack initiation and through-wall cracks that approach unity. Othercomponents, often with similar values of fatigue usage factors, show much lower failureprobabilities. As discussed in NUREG/CR-6674 the maximum failure rate (through-wallcracks per year) is about 5×10-2, and the maximum core-damage frequency based on thesecalculated failure rates is about 1.0×10-6 per year. These maximum values do not changesignificantly from 40 years to 60 years. In contrast, failure rates for other componentswith much lower baseline failure rates are seen to increase by as much as an order ofmagnitude from 40 years to 60 years.

27-12

Correlation of Failure Probabilities with Usage Factors

Figure 2 shows the correlation between calculated probabilities of through-wall cracks foreach of the 47 components with the fatigue usage factors reported in NUREG/CR-6260(Ware et al. 1995). The correlation is only approximate because the usage factors addressonly crack initiation. As such the usage factor calculations do not address factors thatdetermine how likely each initiated crack will grow to become a through-wall flaw.

The CUFs as reported in this paper were taken directly from the INEL work ofNUREG/CR-6260 (Ware et al. 1995), which made use of the fatigue (S-N) curves ofNUREG/CR-5999 (Majumdar et al. 1993). These curves accounted for environmentaleffects and predicted significant reductions in life compared to the fatigue curves of theASME code. As such, the INEL fatigue usage factors for the sample components aregenerally greater than the usage factors calculated when the components were originallydesigned.

The plots of Figure 2 indicate a potential for through-wall cracks (probability of failuregreater than say >10-1) even for usage factors less than one. Usage factors greater thanone can sometimes correspond to essentially 100 percent failure probability. On the otherhand, for usage factors of 0.1 or less Figure 2 indicates that the probabilities of failurebecome relatively low (10-3 or less). These overall trends are consistent the viewpoint thatcode usage factors should not be treated as precise predictors of cycles to fatigue failure,but rather as a method to establish acceptable designs. It should furthermore be noted thatplant operating experience has shown few if any fatigue failures for the types of loadingconditions addressed by the design calculations. Instead, fatigue failures have generallybeen due to stresses (vibration, thermal fatigue, etc.) that were not anticipated during plantdesign.

Probabilities at 60 Years Versus 40 Years

Figure 3 shows trends of the calculated results of Table 3. The data points above thedashed diagonal line indicate the wide range of the calculated failure probabilities andcompare the failure probabilities at the end of a 60-year plant life with the correspondingprobabilities for a 40-year plant life. The range of the through-wall crack probabilities isabout seven orders of magnitude.

The failure probabilities corresponding to a 60-year plant life can be a factor of 10 orgreater than those for a 40-year plant life. It should be noted that the increases aregreatest for components that have relatively small values at 40 years. In contrast, thereare only small increases when the 40-year probabilities are already quite large. In thesecases, the fatigue cracks initiate relatively early in life and there is a high potential ofleaking before the end of a 40-year operating period. However, it is unlikely in practicethat such high levels of fatigue damage would go unnoticed and unmitigated at asignificant number of fatigue sites. It is more likely that corrective action programsconsisting of augmented inspections, repairs and replacements, along with changes to

27-13

plant operating practice would be implemented before the end of a 40-year operatingperiod. Such programs would significantly decrease the failure frequencies from thosecalculated here.

Water Versus Air Environment

The main objective of the PNNL study was to compare predicted probabilities of fatiguefailures for a 60-year life versus a 40-year life with the effect of reactor coolantenvironments included in both evaluations. However, additional calculations wereperformed to establish the extent of environment effects independent of the issue of 40years versus 60 years pant life.

The data points plotted below the dashed diagonal line of Figure 3 compare the calculatedfatigue lives for water versus air environments. The 40-year life for the water environmentwas used as the baseline case. The data show that changing to an air environment giveslower probabilities of through-wall cracks (by about a factor of 100). In contrast,changing from a 40-year life to a 60-year life increased the probabilities, but the relativeincrease (a factor of about 10) is not nearly as large as that associated with theenvironmental effects.

Sensitivity Calculations for Surge Line Elbow

A detailed study was performed for the surge-line location in the newer-vintageCombustion Engineering plant. This was the location with the highest calculated failureprobability of any of the 47 locations addressed by PNNL.

Figure 4 presents the failure probabilities predicted by pc-PRAISE for the surge-lineelbow in terms of probabilities of crack initiation and of through-wall cracks, both as afunction of time. It is seen that cracks can initiate rather early in the plant life, with abouta 50-percent probability of initiating a fatigue crack after only 10 years of operation. Overthis same 10 year period, about 50 percent of the initiated cracks are predicted to becomethrough-wall cracks. The frequency of through-wall cracks (lower curve) increasessignificantly over this 10 year period and then remains relatively constant over theremainder of the 60-year plant life.

Figure 5 evaluates effects of the critical inputs that specify through-wall stress gradients.It was recognized that peak surface stresses, apply only to the initiation of fatigue cracks,and are not representative of the stresses that grow these initiated cracks through thethickness of the pipe wall. A range of assumptions on stress gradients was made for thecalculations of Figure 5. It should be emphasized that the probabilities of crack initiationremained the same for all the calculations because the cyclic surface stresses that governcrack initiation were the same for all cases.

The solid curve of Figure 5 shows the baseline through-wall crack probabilities asreported in Table 3. This calculation assumed that peak stresses greater than 45 ksi

27-14

should be treated as thermal gradient stresses. The most conservative assumption wouldtreat the peak stress as entirely uniform tension stress. As indicated by Figure 5, the moreconservative assumption increased the calculated failure probabilities by a factor of about2.0. The other bounding assumption was to treat the peak stress as 100 percent thermalgradient. This reduced the failure probabilities by a factor between five and ten. Perhapsthe most realistic assumption addressed by Figure 5 considered all stresses greater than 15ksi as thermal gradient stresses. This assumption decreased the through-wall crackprobabilities by a factor of about 2.0 relative to the baseline case.

A Study of Crack Lengths and Linking as Predicted by pc-PRAISE

The pc-PRAISE model for fatigue crack initiation was applied to simulate the initiation,growth and linking of thermal fatigue cracks for a small diameter pipe. The objective wasto demonstrate the ability of pc-PRAISE to predict realistic lengths for circumferentialcracks. A second objective was to perform sensitivity calculations to evaluate the effectsof modeling assumptions and alternative inputs. The calculated crack lengths were thencompared to the size and shape of the cracking reported for a small diameter pipe of thehigh-pressure injection system at the Oconee 2 plant (USNRC 1997).

The calculations addressed a stainless steel pipe with an inner diameter of 2.9 inch and awall thickness of 0.3 inch. The baseline case was intended to correspond to inputs andassumptions used for the 47 PWR and BWR components of the PNNL study. Thefollowing describes the baseline case:

• 100% of the cyclic stress assigned to the thermal gradient category

• no circumferential variation of cyclic stress

• cycles to crack initiation were sampled independently at each circumferential site

• 5 sites for crack initiation around the circumference of pipe

• 1 percent probability that the length of the initiated fatigue crack will exceed thelength of the initiation site

Table 4 lists the calculations for five variations of the input parameters from those of thebaseline case.

The magnitudes and numbers of the cyclic stresses were assigned to give calculatedprobabilities of through-wall cracks that approached 100 percent. The objective was notto predict failure probabilities per se, but to predict the circumferential extent of crackingthat develops late in the life of a highly stressed component.

It was not possible to make a direct comparison of the probabilistic calculations with thesingle observed case of Oconee-2 pipe failure (see Figure 6). The comparisons were

27-15

therefore based on calculated distributions of cracks corresponding to the time at whichthe probability of through-wall cracking attained a value of 50 percent.

A further complication was that the ANL fatigue correlations assume that initiated fatiguecracks start immediately with a depth of 3 mm. For the small 3-inch diameter pipe, this 3-mm crack is about 40 percent of the pipe wall. Much of the cracking in the Oconee-2event had depths less than 30 percent of the pipe wall. The output of the computer codefails to define the depths of these small cracks, prior to such time that they attain thethreshold depth of 3-mm. Therefore, that portion of the Oconee cracking with depths lessthan 3 mm was treated as uncracked in the context of the approach taken by the pc-PRAISE fracture mechanics model.

The crack length at the outer surface of the Oconee-2 pipe had become 21 percent of thepipe circumference (Figure 6) when the leak rate caused the plant operators to bring theplant into a shutdown mode. The pipe also had part-through cracking around theremaining circumference. About 47 percent of the pipe circumference had cracking thatexceeded the 3-mm threshold of the ANL correlations.

The columns of Table 4 further summarize trends from the detailed computer output.Several global measures of the extent of circumferential cracking are used to describe thecircumferential cracking as indicated by the column headings. The baseline case of Table 4predicts that 40 to 60 percent of the pipe circumference will be cracked (at a probability ofabout 50 percent), with the percentage depending on the measure selected to described thecircumferential cracking. The calculated percentages are generally consistent with theobserved cracking of the Oconee-2 pipe. The higher numbers (60%) of the final twocolumns of Table 4 are based on a weld-by-weld summary of the simulated data. Thesetabulations characterize only the total amount of circumferential cracking and do notconsider whether this cracking is in the form of one large crack or from the sum of severalsmaller unconnected cracks.

The second calculation assumed that part of the cyclic stress is uniform tension (i.e. 20%)rather than a pure through-wall thermal gradient stress. The predicted probability for longcracks decreases markedly, suggesting that the cracking pattern of the Oconee failure ischaracteristic of a pure thermal gradient stress.

The third calculation of Table 4 increased the number of potential crack initiation sitesfrom 5 to10. The length of the sites decreased from about 2 inch to about 1 inch. Thereis little predicted change in the overall amount of circumferential cracking, but the lengthsof the individual cracks tended to decrease. This means more cracks with the averagelength of each crack becoming shorter.

The fourth calculation of Table 4 kept the number of initiation sites at five, but increasedthe assigned probability that an initiated crack will span the entire 2-inch length of theinitiation site. The median length of the initiated crack was not changed from the baselinevalue of 0.6 inch. Table 4 shows little change in the calculated probabilities for largerlengths of circumferential cracking.

27-16

The fifth calculation of Table 4 assumed a perfect correlation between the number ofcycles to crack initiation for all sites in a given weld. That is , if one of the five sitesbecomes cracked, all of the other sites will crack at the same time. This assumptionignores the characteristic scatter in fatigue data. The predictions of Table 4 show veryhigh probabilities for cracking around a large fraction of the pipe circumference.

The final calculation of Table 4 expands on the previous calculation. The site-to-siterandomness of fatigue lives is again taken to be zero, but there is a 20% circumferentialvariation in the cyclic stress level. The circumferential variation gives a modest reductionin the probability for cracking around large fractions of the pipe circumference comparedto the previous case.

In summary the pc-PRAISE predictions have been compared to the cracking observed atOconee-2. The service failure had deep cracking over 20 to 50 percent of the pipecircumference, whereas the pc-PRAISE baseline calculation predicted deep cracking oversome 40 to 60 percent of the pipe circumference. It is concluded that the predictedcracking is generally consist and somewhat conservative relative to `the observedcracking.

Sensitivity Calculations Using Latin Hypercube Method

A comprehensive set of calculations for all locations of the seven plants addressed theeffects of changing various inputs and modeling assumptions. These calculations appliedthe Latin hypercube method described in an ASME paper (Khaleel and Simonen 1995),which permitted rapid calculations even for components with very small failureprobabilities. The streamlined Latin hypercube method was successfully benchmarkedagainst the pc-PRAISE code.

Baseline ParametersInputs for the baseline cases of the sensitivity calculations were essentially the same as theinputs later used for the final pc-PRAISE runs of the NUREG/CR-6674 report.

Initial Flaw Depth – The baseline depth of the initiated crack was assigned a deterministicvalue of 3 mm. Variations from this depth were part of the sensitivity calculations.

Correlations Between Crack Initiation and Crack Growth – The baseline case assumedthat the random variations in crack growth rates (da/dN) were independent of randomvariations in the number of cycles to crack initiation. This assumption was used for thecalculations with the pc-PRAISE code. Sensitivity calculations addressed the effect of aperfect correlation between crack initiation and crack growth. These calculations used thesame random number to sample from the distributions for crack initiation and crackgrowth.

Oxygen Content of Reactor Water – The baseline cases used a reactor water oxygencontent of 0.010 ppm (10 ppb) for PWR plants and 0.100 ppm (100 ppb) for BWR plants.

27-17

These values were considered to be realistic levels but somewhat higher than expected fortypical plant operating conditions. Sensitivity calculations considered somewhat lowerand more typical values for water chemistries.

Sulfur Content – The baseline value of sulfur content was 0.015 weight percent for lowalloy steels. For stainless steels the sulfur content does not appear in the equations forfatigue crack initiation. Sensitivity calculations considered the effects of somewhat lowerand more typical values of sulfur than the bounding value of 0.015 percent.

Strain Rate – Lower strain rates result in fewer predicted cycles to crack initiation. Thebaseline cases assumed a common strain rate of 0.001 percent per second for allcomponents and all transients. Somewhat lower values were used for sensitivitycalculations.

Sensitivity to Initial Flaw Depth

The ANL fracture mechanics model defines crack initiation as a surface flaw with a depthof about 3 mm. This depth was related to the 25 percent load drop method used to detectthe presence of a crack in the fatigue tests. The actual depth of the crack could be lessthan or greater than 3 mm. To address uncertainties in the initial crack depth, calculationswere performed for crack depths of 2 mm and 4 mm.

Each point on Figure 7 corresponds to one of the components from the sevenrepresentative plants. If initial flaw depth had no effect on the probability of through-wallcracks, then all points would fall on the diagonal line. However Figure 7 shows that thecalculated probabilities of through-wall crack increases somewhat for the 4-mm crack anddecreases somewhat for the 2-mm crack. The changes in through-wall crack probabilitiesare at most by a factor of 2, and are insignificant for components having relatively highfailure probabilities.

Simulation of the initial flaw depth as an additional stochastic variable was not consideredappropriate, because 1) the calculated failure probabilities were relatively insensitive to theassumed value of flaw depth, and 2) uncertainties in the flaw depths corresponding toexperimental data on cycles to crack initiation were considered to be adequatelyaccounted for in the variability in cycles to crack initiation.

Sensitivity to Initial Flaw Length or Aspect Ratio

The baseline Latin hypercube model assumed that all initiated cracks had aspect ratios of10:1 (a length of 30 mm for the initial flaw depth of 3 mm). It was also assumed that theaspect ratio of growing fatigue crack remained at 10:1 as the crack increased in depth.

Figure 8 shows the effect of replacing the 10:1 aspect ratio with a ratio of 3:1. The smallervalue was selected as bounding for growing fatigue cracks. The calculated probabilities ofthrough-wall cracks decreased by a factor as great as 10. The differences were greatest

27-18

for low failure probabilities and were relatively small for components with very high failureprobabilities. The results indicate that precise inputs for modeling of flaw lengths are notcritical to the calculations of through-wall crack probabilities (provided that assumedvalues of aspect ratio are not taken to be unrealistically small) but are critical toprobabilities of large leaks and breaks.

Effect of Wall Thickness

The results of Figure 9 were generated by arbitrarily changing the wall thickness of eachcomponent to 1.0 inch and then to 8.0-inch. Whereas the calculated probabilities ofinitiating a fatigue crack do not change, the thicker component was expected to have agreater fatigue life because more stress cycles are needed to growth the crack through thethicker metal path. The calculated results are consistent with this expectation. Thedifferences in failure probabilities are about a factor of 10 for relatively low failureprobabilities but become insignificant when the failure probabilities are relatively large.The results were based on a uniform through-thickness distribution of cyclic stress. Thepresence of large stress gradients will tend to offset the wall thickness effect seen in Figure9.

Effect of Through-Wall Stress Gradients

The baseline case conservatively assumed a uniform distribution of stress through the wallthickness, which meant that the peak surface stress that governs crack initiation also isavailable to grow the small initiated crack to become a through-wall crack. However, formost stress transients, the peak surface stress is associated with stress gradients. Figures10-13 show the sensitivity of calculated probabilities of through-wall cracks to themagnitude of the stress gradient.

Figure 10 assumed a relatively modest gradient consisting of a linear distribution of stresssuch that the stress at the outer surface is 50 percent of the peak stress at the innersurface. This modest gradient has only a small effect (factor of 2 or less) in terms ofdecreasing the calculated probabilities of through-wall cracks. Figure 11 increased thelinear stress gradient such that the outer surface stress became zero. In this case thecalculated probabilities of through-wall cracks decreased by as much as a factor of 10relative to the baseline case.

Figure 12 was based on a rather extreme assumption whereby the stress used forcalculating crack growth rates was assumed to be uniformly distributed through the wallthickness, but with this stress reduced to 50% of the peak surface stress used for theinitiation of the crack. Such an assumption could approximate the situation where crackinitiation is from very localized stress concentration. The predicted effect on through-wallcrack probabilities is substantial and amounts to 3-4 orders for magnitude of cases withvery low failure probabilities. The effect is much smaller for components with the higherfailure probabilities, but gives a factor of about 10 reduction in failure probability.

27-19

Figure 13 was based on an assumed stress gradient that was a more realistic estimate of astress gradient produced by a transient thermal stress. The predicted effect of this stressgradient that is about two orders of magnitude for lower failure probability componentsand about one order of magnitude for components with higher failure probabilities. It wasconcluded that realistic predictions of through-wall crack probabilities require realisticmodeling of through-wall stress gradients.

Correlation of Crack Initiation and Crack Growth

The source code was modified to assume a perfect correlation between the randomvariations in cycles to crack initiation with the corresponding variations in the crackgrowth rates. Such correlations were expected to increase the probability that a crackwhich initiates at a relatively small number of stress cycles will subsequently grow at afaster than average crack growth rate. The data points of Figure 14 confirm thisexpectation. The correlation increases the calculated failure probabilities by up to an orderof magnitude, with the effect being largest when the failure probabilities are small. Atfailure probabilities greater than 1.0E-01 the effect becomes negligible.

Environment and Material Characteristics

A number of inputs for environmental parameters must be assigned for application of theArgonne National Laboratory equations for fatigue crack initiation. The calculations ofFigures 15-20 address the effects of uncertainties in these inputs on calculatedprobabilities of through-wall cracks. The inputs addressed in sensitivity calculations werethe oxygen content of the reactor water, the sulfur content of the steel, and the strain rateassociated with the cyclic stresses. The calculations did not address the effects oftemperature.

Figure 15 addresses the effect of oxygen content with the baseline being an oxygen levelof 0.01 ppm that is typical of PWR conditions. The sensitivity calculations increased thislevel to 0.10 ppm (BWR conditions). It is should be noted that these calculationsarbitrarily assigned the same oxygen level to all of the 47 components without regardwhether they corresponded to a PWR or BWR plant. It is seen in Figure 15 thatincreasing the oxygen level over the selected range of uncertainty has at most an order ofmagnitude effect on calculated probabilities of through-wall cracks. In some cases theprobabilities increased (ferritic steel components) and other cases the calculatedprobabilities decreased (stainless steel components).

Figures 16 and 17 show the sensitivity of calculated failure probabilities to strain rates, andindicate that low strain rates can result in higher values of calculated failure probabilities.The default strain rate used in the baseline calculations was 0.001. Figure 16 shows theeffect of using a strain rate of 0.01, which would correspond to an actual transient with arelatively rapid loading rate. Figure 15 shows that some of the calculated failureprobabilities decreased by as much as a factor of 10. Failure probabilities for anothergrouping of components (ferritic steel) show little if any change. Figure 16 increased the

27-20

strain rate by a factor of 1000, and these results show more substantial decreases incalculated failure probabilities for the higher strain rate.

Figure 18 shows the effect of reducing the sulfur content of the steel to 0.0 percent from0.015 percent. Some failure probabilities show no increase (stainless steel) whereas othercomponents show only a modest increase (less that a factor of 2.0).

Figure 19 (low alloy steel) and Figure 20 (stainless steel) show the effect of changing theinputs from bounding values that govern environmentally assisted fatigue to moremoderate values that may be more typical of actual plant operation. The calculations usedbaseline inputs of strain rates of 0.001 %/sec, oxygen of 100 ppb and sulfur of 0.015percent. In constructing these figures the material type for all components was arbitrarilyset to either low alloy steel (Figure 19) or stainless steel (Figure 20). Within the relativelysmall range of uncertainty (as considered here) the changes in calculated failureprobabilities were relatively small (factor of 2 or less). The assumption in thesecalculations was that the total exclusion of environmental effects could not be justified forany of the components. Other sensitivity calculations of Figure 3 compared calculatedthrough-wall crack probabilities for air environment versus probabilities for waterenvironment, which shows differences by a factor of 10 or more.

Summary and Conclusions

Probabilities of fatigue failures have been estimated for RPV and piping components offive PWR and two BWR plants. These calculations were made possible by thedevelopment of a new version of the pc-PRAISE probabilistic fracture mechanics codethat has the capability to simulate the initiation of fatigue cracks in combination with asimulation of the subsequent growth of these fatigue cracks. The calculations gave a widerange of failure probabilities for the selected components, with some components havingend-of-life probabilities for through-wall cracks of nearly 100 percent and others withprobabilities of less than 10-6.

It is recognized that there are uncertainties in the calculated failure probabilities.Uncertainties come from assumptions made in the fracture mechanics models and from theinputs to the models. In addressing these uncertainties, the intent of the baselinecalculations was to perform best-estimate calculations. When best-estimate inputs werenot available, the approach was to select conservative values. In particular, the inputs forcyclic stresses were based on design-basis data, which could differ from the stressesoccurring during actual plant operation. Other areas of uncertainty included inputs forstrain rates and environmental variables used to predict fatigue-crack initiation.

Calculations were performed to address the effects of reactor water environments (versusair) and to compare these effects to the effects of extended plant operation from 40 yearsto 60 years. The environmental effects were predicted to increase through-wall crackprobabilities by as much as two orders of magnitude.

27-21

The calculated through-wall crack probabilities for the components with the very highestprobabilities of failure show essentially no increase in failure probability from 40 to 60years. On the other hand, other components with lower fatigue uasage showed significantincreases in failure probabilities. Here again, the increases associated with the water-environment are a factor of 10 or more greater than the corresponding increasesassociated with extended plant operation from 40 to 60 years.

Further work is needed to reconcile the relatively high calculated failure probabilities forsome components with the relatively good service experience for these components. It isexpected that inputs based on better data on the frequency and severity of cyclic stressesin combination with a better characterization of environmental factors will remove most ofthe differences between predicted cycles to failure versus service experience.

The higher values of though-wall crack probabilities were also based on conservativeassumptions relative to the scenario that governs the inspection and maintenance of high-fatigue locations. The model assumed that components with high-failure frequencies willremain in operation as components fail one-by-one or are retired at the end of plant life.In practice, the first failure of a group of similar components will cause other members ofthe group to be subject to an aggressive program of corrective actions such that theprobability of repeat failures is greatly reduced. Such corrective actions would includefrequent inservice inspections and changes to plant operational practices to reduce stresslevels. The present fracture-mechanics model does not address the benefits of suchcorrective actions in reducing failure frequencies.

Acknowledgments

Work Supported by the U.S. Nuclear Regulatory commission under Contract DE-AC06-76RLO1830; NRC JCN W6671.

References

Harris, D. O. and D. Dedhia. 1992 . A Probabilistic Fracture Mechanics Code forPiping Reliability Analysis (pc-PRAISE code), NUREG/CR-5864, U.S. NuclearRegulatory Commission, Washington D.C.

Keisler, J., O.K. Chopra and W. J. Shack. 1994. Statistical Analysis of Fatigue Strain-Life Data for Carbon and Low-Alloy Steels, NUREG/CR-6237, U.S. Nuclear RegulatoryCommission, Washington D.C.

Keisler, J., O.K. Chopra, and W. J. Shack. 1995. Fatigue Strain-Life Behavior ofCarbon and Low-Alloy Steels, Austenitic Stainless Steels, and Alloy 600 in LWREnvironments, NUREG/CR-6335, U.S. Nuclear Regulatory Commission, WashingtonD.C.

27-22

Khaleel, M. A. and F. A. Simonen. 1995. “A Model for Predicting Vessel FailureProbabilities Due to Fatigue Crack Growth,” ASME PVP Vol. 304, pp. 401-416, Fatigueand Fracture Mechanics in Pressure Vessels and Piping.

Khaleel, M.A., F.A. Simonen, H.K. Phan, D.O. Harris and D. Dedhia 2000. FatigueAnalysis of Components for 60-Year Plant Life, NUREG/CR-6674, U.S. NuclearRegulatory Commission, Washington D.C.

Majumdar, S., O. K. Chopra, and W. J. Shack. 1993. Interim Fatigue Curves to SelectedNuclear Power Plant Components, NUREG/CR-5999, U.S. Nuclear RegulatoryCommission, Washington D.C.

USNRC 1997. Unisolable Reactor Coolant System Leak at Oconee-2 April 21, 1997 ,Licensee Event Report LER No. 270/97-001, U.S. Nuclear Regulatory Commission.

Ware, A. G., D. K. Morton, and M. E. Nitzel. 1995. Application of NUREG/CR-5999Interim Fatigue Curves to Selected Nuclear Power Plant Components, NUREG/CR-6260, U.S. Nuclear Regulatory Commission, Washington D.C.

Wire, G. L., and Y. Y. Li. 1996. “Initiation of Environmentally-Assisted Cracking inLow Alloy Steels,” Fatigue and Fracture – Volume 1, PVP-Vol.323, pp. 269-289,American Society of Mechanical Engineers, New York.

27-23

Table 1. Plants Evaluated in the Fatigue Study

PWRs BWRs

Babcock and Willcox (B&W) General Electric (GE) – Newer Vintage

Combustion Engineering (CE) – NewerVintage

GE – Older Vintage

CE – Older Vintage

Westinghouse (W) – Newer Vintage

W – Older Vintage

Table 2. Components Selected for Fatigue Analysis

PWRs BWRs

1. Reactor pressure vessel shell andlower head.

1. Reactor pressure vessel shell andlower head.

2. Reactor vessel inlet and outletnozzles.

2. Reactor vessel feedwater nozzle.

3. Pressurize surge line (including hotleg and pressurizer nozzles).

3. Reactor recirculation piping(including inlet and outlet nozzles).

4. Reactor coolant piping chargingsystem nozzle.

4. Core spray line reactor vessel nozzleand associated class 1 piping.

5. Reactor coolant piping safetyinjection nozzle.

5. Residual heat removal class 1 piping.

6. Residual heat removal (RHR) systemclass 1 piping.

6. Feedwater line class 1 piping.

27-24

Table 3 Summary of Results for All Seven Plants – Water Environment

DET DET CUMULATIVE CUMULATIVE CUMULATIVE CUMULATIVE TWC/YEAR TWC/YEAR CDF CDFUSEAGE@40 USEAGE@60 PI@40 YR PI@60 YR PTWC@40 YR PTWC@60 YR @40 YR @60 YR @40 YR @60 YR

CE-NEW RPV LOWER HEAD/SHELL LAS 1.40E-02 2.10E-02 7.89E-06 4.82E-05 6.71E-15 1.44E-12 1.13E-15 1.90E-13 1.13E-16 1.91E-14CE-NEW RPV INLET NOZZLE LAS 4.75E-01 7.12E-01 1.40E-02 4.44E-02 5.90E-05 9.01E-04 7.50E-06 7.59E-05 2.03E-11 2.05E-10CE-NEW RPV OUTLET NOZZLE LAS 4.72E-01 7.08E-01 4.22E-01 6.89E-01 1.74E-03 2.90E-02 3.58E-04 2.57E-03 9.65E-10 6.93E-09CE-NEW SURGE LINE ELBOW 304/316 2.60E+00 3.90E+00 9.95E-01 9.99E-01 9.81E-01 9.98E-01 7.60E-02 9.38E-02 2.17E-06 2.67E-06CE-NEW CHARGING NOZZLE NOZZLE LAS 1.04E-01 1.56E-01 9.56E-04 3.84E-03 2.61E-06 5.50E-05 3.46E-07 5.06E-06 2.77E-12 4.05E-11CE-NEW CHARGING NOZZLE SAFE END 304/316 5.02E-01 7.53E-01 1.06E-02 6.75E-02 9.00E-05 1.03E-03 1.75E-05 1.15E-04 1.40E-09 9.21E-09CE-NEW SAFETY INJECTION NOZZLE NOZZLE LAS 4.57E-01 6.85E-01 1.01E-03 4.81E-03 1.00E-06 1.90E-05 3.75E-07 1.50E-06 1.88E-12 7.50E-12CE-NEW SAFETY INJECTION NOZZLE SAFE E 304/316 2.86E-01 4.29E-01 8.68E-03 3.16E-02 2.61E-06 5.50E-05 3.46E-07 5.06E-06 1.73E-11 2.53E-10CE-NEW SHUTDOWN COOLING LINE ELBOW 304/316 4.87E-01 7.30E-01 1.13E-02 5.75E-02 2.00E-05 4.53E-04 7.00E-06 4.40E-05 1.89E-10 1.19E-09CE-OLD RPV LOWER HEAD/SHELL LAS 1.30E-02 1.95E-02 2.68E-06 1.93E-05 6.36E-16 1.85E-13 1.07E-16 1.85E-13 1.08E-17 1.86E-14CE-OLD RPV INLET NOZZLE LAS 1.72E-01 2.58E-01 1.88E-03 7.89E-03 4.11E-07 1.33E-05 5.87E-08 1.33E-05 1.58E-13 3.59E-11CE-OLD RPV OUTLET NOZZLE LAS 5.53E-01 8.29E-01 5.91E-01 8.46E-01 7.05E-02 3.53E-01 8.98E-03 2.27E-02 2.42E-08 6.13E-08CE-OLD SURGE LINE ELBOW 304/316 6.61E-01 9.92E-01 9.39E-01 9.87E-01 6.27E-01 8.85E-01 4.36E-02 5.48E-02 1.24E-06 1.56E-06CE-OLD CHARGING NOZZLE SAFE END 304/316 5.62E-01 8.43E-01 1.18E-02 5.31E-02 4.10E-05 5.98E-04 8.75E-06 5.05E-05 7.00E-10 4.04E-09CE-OLD SAFETY INJECTION NOZZLE SAFE E 304/316 3.17E-01 4.75E-01 7.56E-03 3.59E-02 1.40E-05 2.00E-04 2.25E-06 1.85E-05 1.13E-10 9.25E-10CE-OLD SHUTDOWN COOLING LINE ELBOW 304/316 8.40E-02 1.26E-01 3.94E-02 1.19E-01 2.10E-04 2.36E-03 4.38E-05 1.98E-04 1.18E-09 5.35E-09B&W RPV NEAR SUPPORT SKIRT LAS 2.23E-01 3.35E-01 8.25E-03 2.50E-02 7.85E-06 1.52E-04 1.04E-06 1.36E-05 1.04E-07 1.36E-06B&W RPV OUTLET NOZZLE LAS 4.69E-01 7.04E-01 7.74E-01 8.99E-01 1.83E-01 5.44E-01 1.94E-02 3.35E-02 5.25E-08 9.03E-08B&W MAKEUP/HPI NOZZLE SAFE END 304/316 1.05E+00 1.58E+00 1.30E-01 4.79E-01 2.10E-03 3.09E-02 5.88E-04 2.22E-03 2.94E-08 1.11E-07B&W DECAY HEAT REMOVAL/REDUCING T 304/316 5.30E-01 7.95E-01 5.72E-02 2.08E-01 3.00E-03 2.54E-02 4.26E-04 1.79E-03 1.15E-08 4.82E-08W-NEW RPV LOWER HEAD/SHELL LAS 1.80E-02 2.70E-02 3.21E-05 1.71E-04 7.52E-13 9.64E-11 1.23E-13 1.21E-11 1.24E-14 1.50E-12W-NEW RPV INLET NOZZLE LAS 2.90E-01 4.35E-01 2.49E-03 1.05E-02 9.17E-07 2.84E-05 1.30E-07 2.83E-06 3.51E-13 7.64E-12W-NEW RPV OUTLET NOZZLE LAS 6.58E-01 9.87E-01 8.62E-01 9.49E-01 3.65E-01 7.42E-01 3.17E-02 4.50E-02 8.57E-08 1.22E-07W-NEW CHARGING NOZZLE NOZZLE 316NG 3.37E+00 5.06E+00 9.51E-01 9.83E-01 8.72E-01 9.63E-01 5.38E-02 5.66E-02 4.31E-07 4.53E-07W-NEW SAFETY INJEC NOZZLE NOZZLE 316NG 1.46E+00 2.19E+00 4.34E-03 3.69E-02 5.00E-04 1.09E-02 5.33E-05 1.30E-03 2.67E-10 6.50E-09W-NEW RESIDUAL HEAT INLET TRAN 304/316 2.73E+00 4.10E+00 9.58E-01 9.99E-01 7.80E-01 9.80E-01 6.25E-02 1.18E-01 1.69E-06 3.17E-06

W-OLD RPV LOWER HEAD SHELL LAS 8.91E-01 1.34E+00 1.11E-01 1.28E-01 7.20E-07 1.11E-05 8.38E-08 9.08E-07 8.44E-09 9.15E-08W-OLD RPV INLET NOZZLE INNER SURFACE LAS 3.02E-01 4.53E-01 3.91E-01 6.44E-01 4.38E-03 5.04E-02 7.53E-04 3.96E-03 2.03E-09 1.07E-08W-OLD RPV INLET NOZZLE OUTER SURFACE LAS 4.96E-01 7.44E-01 6.81E-02 1.11E-01 4.48E-04 3.32E-03 4.75E-05 2.18E-04 1.28E-10 5.89E-10W-OLD RPV OUTLET NOZZLE INNER SURF LAS 4.99E-01 7.48E-01 4.90E-01 7.53E-01 9.33E-03 9.60E-02 1.56E-03 7.54E-03 4.21E-09 2.04E-08W-OLD RPV OUTLET NOZZLE OUTER SURF LAS 3.47E-01 5.20E-01 1.63E-01 2.38E-01 7.77E-03 3.60E-02 6.99E-04 1.83E-03 1.89E-09 4.94E-09W-OLD CHARGING NOZZLE NOZZLE 304/316 3.19E-01 4.79E-01 4.67E-04 3.75E-03 3.00E-07 5.20E-06 7.50E-08 6.00E-07 6.00E-13 4.80E-12W-OLD SAFETY INJECTION NOZZLE NOZZLE 304/316 3.27E-01 4.90E-01 1.88E-03 1.31E-02 4.00E-06 8.80E-05 8.75E-07 1.05E-05 4.38E-12 5.25E-11W-OLD RESIDUAL HEAT REMOVAL TEE 304/316 2.05E-01 3.08E-01 1.34E-02 5.16E-02 1.15E-04 1.14E-03 1.63E-05 9.26E-05 8.13E-10 4.63E-09GE-NEW RPV NEAR CRDM PENETRATION LAS 6.28E-01 9.42E-01 7.89E-05 3.49E-04 7.88E-12 6.82E-10 1.25E-12 8.26E-11 1.25E-13 8.26E-12GE-NEW FEEDWATER NOZZLE SAFE END LAS 1.88E+00 2.82E+00 1.04E-01 2.53E-01 1.31E-03 1.47E-02 2.38E-04 1.23E-03 3.57E-11 1.84E-10GE-NEW RECIRC SYS - TEE SUCTION PIPE 304/316 8.30E-01 1.25E+00 4.23E-02 1.39E-01 4.80E-04 4.67E-03 7.13E-05 3.66E-04 1.07E-11 5.49E-11GE-NEW CORE SPRAY LINE SAFE END EXT LAS 4.36E-01 6.54E-01 3.83E-04 1.27E-03 1.45E-07 3.25E-06 1.97E-08 3.04E-07 7.09E-15 1.09E-13GE-NEW RHR LINE STRAIGHT PIPE LAS 1.13E+01 1.69E+01 4.73E-01 6.71E-01 4.10E-01 6.21E-01 1.35E-02 2.25E-02 2.54E-11 2.03E-10GE-NEW FEEDWATER LINE ELBOW LAS 3.69E+00 5.53E+00 1.59E-01 3.65E-01 1.01E-03 1.46E-02 1.69E-04 1.35E-03 3.04E-09 5.06E-09

GE-OLD RPV LOWER HEAD TO SHELL LAS 7.90E-02 1.19E-01 2.71E-10 2.76E-08 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00GE-OLD RPV FEEDWATER NOZZLE BORE LAS 3.17E+00 4.75E+00 7.27E-02 2.42E-01 1.00E-05 8.80E-04 2.50E-06 9.76E-05 3.75E-14 1.46E-12GE-OLD RECIRC SYSTEM RHR RETURN LINE 304/316 3.90E+00 5.85E+00 9.43E-01 9.99E-01 7.12E-01 9.85E-01 7.20E-02 1.23E-01 1.08E-08 1.84E-08GE-OLD CORE SPRAY SYSTEM NOZZLE LAS 5.20E-01 7.80E-01 1.41E-04 7.89E-04 1.91E-08 8.84E-07 2.85E-09 9.51E-08 6.41E-17 2.14E-15GE-OLD CORE SPRAY SYSTEM SAFE END 304/316 1.77E+00 2.65E+00 3.33E-01 7.64E-01 1.46E-02 1.10E-01 2.08E-03 8.04E-03 4.68E-10 1.81E-09GE-OLD RESIDUAL HEAT TAPERED 304/316 4.78E-01 7.17E-01 1.47E-03 7.89E-03 9.21E-05 1.02E-03 1.07E-05 7.82E-05 1.61E-12 1.17E-11GE-OLD FEEDWATER LINE - RCIC TEE LAS 6.98E+00 1.05E+01 3.86E-01 7.82E-01 2.99E-03 5.92E-02 6.96E-04 5.54E-03 1.04E-10 8.30E-10

PLANT COMPONENT MAT

27-25

Table 4 Comparison of Alternative pc-PRAISE Calculations with Oconee-2 Event

Measure of the Circumferential Extent of Cracking

Percent of deep cracks (a/t>80%) that are the result of 2or more linking of cracks from

adjacent sites(Q1)

Percent of deep cracks (a/t >80%) that are longer than 40%

of the circumference(Q4)

Percent of cracked welds thathave cracking over more than

60% of the innercircumference

(Q2)

Percent of the welds that havedeep cracking (a/t > 80%) thatextends the deep cracking over

more than 60% of thecircumference

(Q3)

Oconee-2 - 100%

0% if cracking <3-mmis neglected

100% if crack < 3-mm isincluded

0%

Baseline Calculation 78.0% 73.4% 51.8% 52.6%

Calculation with 80% thermal gradient stress and 20% uniformtension stress

31.1% 3.4% 0.2% 0.0%

Calculation with 5 circumferential sites for crack initiationincreased to 10 sites 19.8%

24.4% 64.0%59.3%

Calculation with length of initiated crack increased from 1%probability for 2 inch crack to 10% probability 78.9% 81.5%

59.0% 58.1%

Cycles for crack initiation 100% correlated from site-to-site (butno circumferential variation in cyclic stress) 100.0% 98.8%

84.2% 92.8%

Cycles for crack initiation 100% correlated from site-to-site (buta 20% circumferential variation in cyclic stress)

89.4% 92.8% 71.1% 77.4%

27-26

Figure 1 Example of Probabilistic S-N Curves for Low-Alloy Steel

1.E-12

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

0 1 2 3 4 5 6 7 8 9 10

Fatigue Usage Factor at 40 Years

Cu

mu

lati

ve P

rob

abili

ty o

f T

hro

ug

h-W

all C

rack

at

40 Y

rs

Figure 2 Comparison of Calculated Usage Factors with CalculatedThrough-Wall Crack Probabilities

1

10

100

1000

1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06

Cycles for Crack Initiation

Str

ess

Am

plit

ud

e, k

si

N0.0001%

N0.1%

N1%

N2%

N5%

N10%

N25%

N50%

N75%

N90%

N9999%Fatigue of Carbon Steel in Water at 550F;

Low Strain Rate; High Sulfur; High Oxygen

27-27

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Through-Wall Crack at 40-Years(Water Enviroment)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

- A

lter

nat

ive

Cas

e Crack Initiation and Crack Growth Statistically Independentpc-PRAISE

Cumulative Probability of TWC at 40-Years(Air Environment)

Cumulative Probability of TWC at 60-Years(Water Environment)

No Failures in 107 Trials

Figure 3 Comparison of Probabilities of Through-Wall Cracks for Air Versus WaterEnvironment and for 40-Year Life and 60-Year Life

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 10 20 30 40 50 60 70 80

Time, Years

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

an

d T

WC

per

Yea

r Cumulative Probability of Crack Initiation

Through-Wall Cracks per Year

New Vintage Cumbustion Engineering PlantSurge Line Elbow

Cumulative Probabilityof Through-Wall Crack

Initiated Cracks per Year

Figure 4 Calculated Probabilities of Crack Initiation and Through-Wall Crack for theSurge-Line Elbow of the Newer Vintage CE Plant

27-28

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 10 20 30 40 50 60 70 80

Time, Years

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

an

d T

WC

per

Yea

r Cumulative Probability of Crack Initiation

TWC All Uniform Tension

Newer VintageCumbustion Engineering PlantSurge Line Elbow

Cumulative Probabilityof Through-Wall CrackMembrane Stress < 45 ksi (Baseline Case)

TWC Membrane Stress < 15 ksi

TWC All Gradient Stress

Figure 5 Calculated Probabilities of Through-Wall Crack for the Surge-Line Elbow of the Newer Vintage CE Plant for Alternative Through-Wall Stress Distributions

27-29

Figure 6 Small Diameter Pipe with Cracking Caused Thermal Fatigue Stresses (LER No. 270/97-001)

27-30

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Through-Wall Crack at 40-Years(Depth of Initiated Crack = 3-mm)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears

(Dep

th o

f In

itia

ted

Cra

ck =

2-m

m a

nd

4-m

m)

Crack Initiation Probabilities by Method 2Crack Initiation and Crack Growth Statistically Independent

Latin Hypercube Method

a0 = 2-mm

a0 = 4-mm

Figure 7 Effects of Initial Crack Depth

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Through-Wall Crack at 40-Years(Crack Growth With Constant 10:1 Aspect Ratio)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears

(Cra

ck G

row

th W

ith

Co

nst

ant

3:1

Asp

ect

Rat

io)



Figure 8 Effect of Flaw Aspect Ratio

27-31

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Through-Wall Crack at 40-Years(All Components Assumed to be 1.0-Inch Thick)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears

(All

Co

mp

on

ents

Ass

um

ed t

o b

e 8.

0-In

ch T

hic

k)



Figure 9 Effect of Wall Thickness

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Through-Wall Crack at 40-Years(Stress for Crack Growth = Stress of Crack Initiation)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears



Linear Stress Gradient 1.0 to 0.5

Figure 10 Effect of Small Linear Stress Gradient

27-32

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00


Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears



Linear Stress Gradient 1.0 to 0.0

Figure 11 Effect of Larger Linear Stress Gradient

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00


Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears



Uniform Stress 50%

Figure 12 Effect of Reduced Stress for Growth of Fatigue Crack

27-33

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00


Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears



Nonlinear Stress Gradient

Figure 13 Effect of Nonlinear Stress Gradient

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of TWC at 40-Years(Initiation Time and Crack Growth Rates Statistically Independent)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears

(In

itia

tio

n T

ime

and

Cra

ck G

row

th R

ates

Co

rrel

ated

)

Crack Initiation Probabilities by Method 2(Monte Carlo Simulation)

Figure 14 Effect of Correlation Between Crack Initiation and Crack Growth

27-34

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Through-Wall Crack at 40-Years(All Components With Low Oxygen = 0.01 PPM)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears

(All

Co

mp

on

ents

Wit

h H

igh

Oxy

gen

= 0

.10

PP

M)



Figure 15 Effect of Increased Oxygen Content

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Through-Wall Crack at 40-Years(All Components With Low Strain Rate = 0.001 %/sec)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears

(All

Co

mp

on

ents

Wit

h M

ediu

m S

trai

n R

ate

= 0.

01 %

/sec

)



Figure 16 Effect of Small Increase in Strain Rate

27-35

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Through-Wall Crack at 40-Years(All Components With Low Strain Rate = 0.001 %/sec)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears

(All

Co

mp

on

ents

Wit

h H

igh

Str

ain

Rat

e =

1.00

%/s

ec) Crack Initiation Probabilities by Method 2

Crack Initiation and Crack Growth Statistically Independent


Figure 17 Effect of Large Increase in Strain Rate

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Through-Wall Crack at 40-Years(All Components With Low Sulfur = 0.0)

Cu

mu

lati

ve P

rob

abili

ty o

f T

WC

at

40-Y

ears

(All

Co

mp

on

ents

Wit

h H

igh

Su

lfu

r =

0.0

15)



Figure 18 Effect of Reduced Suffer Content

27-36

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Crack Initiation at 40-Years(All Components Assumed to be of Low Alloy Steel)

Cu

mu

lati

ve P

rob

abili

ty o

f C

rack

Inia

tio

n 4

0-Y

ears

(EA

C E

ffec

ts M

od

erat

ed o

r R

emo

ved

)


Crack Initiation Probabilities by Method 2Effects of EAC Moderated

Sulfur = 0.007; EDOT = 0.01 %/sec; O2 = 10 PPB

Figure 19 Effect of Reduced EAC for Low Alloy Steel

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E-11 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00

Cumulative Probability of Crack Initiation at 40-Years(All Components Assumed to be Stainless Steel)

Cu

mu

lati

ve P

rob

abili

ty o

f C

rack

Inia

tio

n 4

0-Y

ears

(EA

C E

ffec

ts M

od

erat

ed o

r R

emo

ved

)


Crack Initiation Probabilities by Method 2Effects of EAC Moderated

Sulfur = 0.007; EDOT = 0.01 %/sec; O2 = 10 PPB

Figure 20 Effect of Reduced EAC for Stainless Steel

28-1

28 FATIGUE LIFE REDUCTION IN PWR WATERENVIRONMENT FOR STAINLESS STEELS

K. TsutsumiH. KanasakiT. Umakoshi

Mitsubishi Heavy Industries, Ltd., Japan

T. NakamuraS. Urata

H. MizutaS. Nomoto

The Kansai Electric Power Co., Inc., Japan

ENVIRONMENTAL FATIGUE II

29-1

29 AN APPROACH FOR EVALUATING THE EFFECTS OFREACTOR WATER ENVIRONMENTS ON FATIGUE LIFE

Robert E. NickellApplied Science & Technology

Poway, CA

David A. Gerber Gary L. StevensStructural Integrity Associates

San Jose, CA

29-3

An Approach for Evaluating the Effects of ReactorWater Environments on Fatigue Life

By

Robert E. NickellApplied Science & Technology

Poway, CA

And

David A. Gerber Gary L. StevensStructural Integrity Associates

San Jose, CA

Presented At The

INTERNATIONAL CONFERENCE ON FATIGUE OF REACTOR COMPONENTS

July 31-August 2, 2000Silverado Country Club & Conference Center

Napa, California

I. Background.

Generic Safety Issue No. 166 (GSI-166), later renumbered as Generic Safety Issue No.190 (GSI-190), was identified by U. S. Nuclear Regulatory Commission (NRC) staff [1]because of concerns about the effects of reactor water environments on fatigue life duringthe period of extended operation. The GSI was closed in December 1999, based on amemorandum from NRC-RES to NRC-NRR [2]. During the time when the issue wasactively being evaluated, both the NRC and the industry carried out numerous studies todetermine the scope and magnitude of the effects. Some of these studies were generic,such as those conducted by EPRI and its contractors. Others, by necessity, were plantspecific. For example, the first two license renewal applicants – Baltimore Gas & ElectricCompany for the Calvert Cliffs Nuclear Power Plant (CCNPP) and Duke Energy for theOconee Nuclear Station (ONS) – were required to address the GSI in their applications ona plant-specific basis. Each of the applicants developed responses to the NRC staffquestions on environmental fatigue without the benefit of information from the GSIclosure.

With the closure of the issue, industry efforts have now shifted to the management ofpotentially increased through-wall cracking frequency and associated potential increases inleakage from fatigue-sensitive component locations. Management of these potentialenvironmental fatigue effects is a license renewal issue, based upon communications with

29-4

the NRC staff.

For example, an informal handout by the NRC license renewal staff at an industry meetingheld on November 17, 1999 [4] identified an additional requirement for license renewalapplicants “. . . to incorporate reactor water environmental effects into fatigue agingmanagement program decision criteria.” In other words, the determination of whether aparticular component location is to be included in a license renewal program for managingthe effects of fatigue, and the characteristics of that program (e.g., inservice inspectionmethod, frequency, and acceptance criteria) should incorporate reactor waterenvironmental effects for the 60-year period.

Reference 2 noted that the 60-year fatigue design basis for advanced light-water-cooledreactors (ALWRs) was considered by the NRC staff to be adequate without considerationof reactor water environmental effects, because of the conservatism in ASME Boiler andPressure Vessel Code, Section III, Subsection NB-3000 fatigue design procedures.However, Reference 2 also found that the potential for increased frequency of through-wall cracking and leakage from environmentally-assisted fatigue was such that agingeffects management programs for license renewal would need to consider reactor waterenvironmental acceleration of fatigue cracking.

This paper addresses the remaining regulatory concerns on reactor water environmentaleffects for fatigue of metal components during the license renewal period based on threeconsiderations. First, the nuclear industry has an incentive to minimize through-wallcracking and leakage for both economic and safety reasons during the license renewalperiod. Second, most nuclear plants have fatigue monitoring programs that can managethrough-wall cracking and leakage, provided that appropriate interpretation of reactorwater environmental effects is factored into those programs. Third, an industry consensuson that interpretation is useful from both a cost and a regulatory stability perspective. It isthe opinion of the authors that the industry program described in this paper is responsiveto all three of these concerns.

II. Fatigue Monitoring Program Screening Criteria.

Time Limited Aging Analysis. Fatigue design, when evaluated in terms of eitherexplicitly defined or implied numbers and severity of design-basis transients or operationalcycles, has been determined to be a Time Limited Aging Analysis (TLAA), even thoughno explicit time-limited assumptions are involved. The rules of the ASME Boiler andPressure Vessel Code, Section III, Subsection NB, for Class 1 nuclear power plantcomponents, represent a typical explicit fatigue design basis. The rules of ANSI B31.1,with fatigue strength reduction factors applied to piping stress allowables that aremeasured in terms of design-basis numbers of equivalent full-range thermal cycles,represent a typical implicit fatigue design basis.

Part 54.21(c)(1) of the License Renewal Rule identifies the following three acceptablemethods for resolving TLAAs:

i. The fatigue design analyses can be shown to remain valid for the period of 60years (e.g., the number and severity of explicitly defined design-basistransients for 60 years of operation are bounded by the number and severity of

29-5

explicitly defined design-basis transients for 40 years);

ii. The fatigue design analyses have been projected to the end of the period ofextended operation (e.g., the cumulative fatigue usage factor (CUF) can beshown to remain less than 1.0 for 60 years); or

iii. The effects of aging on the intended function(s) will be adequately managedfor the period of extended operation.

In the latter case, the license renewal applicant is required to demonstrate that theelements of a program to manage the effects of fatigue are in place or will be in placeduring the period of extended operation. The fatigue TLAA evaluation is shown on theright side of Figure 1 as a licensing basis assessment.

Plants with fatigue monitoring programs are able to show, in almost all cases, that thenumber of design-basis transients assumed in the fatigue design basis envelopes the actualnumber of transient events experienced for either 40 or 60 years of operation. In such acase, a formal finding that 10 CFR 54.21(c)(1)(i) is satisfied can be made. However,license renewal applicants then must introduce reactor water environmental effects into asubsequent fatigue evaluation of potentially fatigue-sensitive component locations. Thisevaluation is shown on the left side of Figure 1, and has been labeled "TechnicalEvaluation" since it is outside the bounds of the plant licensing basis. This brings in one orboth of the other two options for addressing fatigue – either an analytical re-evaluation ofthe cumulative usage factor (CUF) for 60 years (10 CFR 54.21(c)(1)(ii)) or ademonstration that fatigue effects are adequately managed (10 CFR 54.21(c)(1)(iii)).

Potential Approaches. Five possible approaches for incorporating reactor waterenvironmental effects into the subsequent fatigue evaluations have been used to-date byutilities that have submitted license renewal applications to the NRC. Two of theseapproaches are defined as explicit, and three are defined as implicit.

The two implicit approaches are exemplified by the Oconee and Hatch License RenewalApplications (LRAs). One of these implicit approaches proposes to track the number ofdesign-basis transients during the period of extended operation, and to reduce the numberof those transients or cycles that would trigger aging management program action by afactor. This factor would conservatively incorporate known reactor water environmentaleffects. The other implicit approach proposes to include component locations in the plantfatigue monitoring program based on a CUF that is reduced by a factor. Again, the factorwould conservatively incorporate known reactor water environmental effects. A commonfeature of these two implicit approaches is that design-basis transient severity is used, asopposed to actual operating transient profiles. Design-basis transient severity has beenshown to provide conservatism on the order of 10 to 20, and often more, in explicitfatigue design calculations [5].

The three explicit approaches are described in the following. The first of these explicitapproaches deterministically recalculates CUFs for a number of fatigue-sensitivecomponent locations, and incorporates reactor water environmental effects in either thecalculated increment of fatigue usage or in the fatigue design allowable. This explicitmethod, referred to as the Fen multiplier approach, was developed by EPRI and the

29-6

General Electric Company (GE) [6], and has been endorsed by the Pressure VesselResearch Committee’s (PVRC’s) Cyclic Life Environmental Effects (CLEE) SteeringCommittee. The method has been used by EPRI and its contractors to recalculate CUFsfor fatigue-sensitive component locations in an early-vintage Combustion Engineering(CE) PWR [5], an early-vintage Westinghouse PWR [7], a late-vintage GE BWR [8], andan early-vintage GE BWR [9]. This approach allows for the selective application ofreactor water environmental reductions to incremental CUF calculations, depending ontransient values of dissolved oxygen, temperature, strain range, and strain rate.References 5, 7, and 8 used actual plant transient profiles (as opposed to design-basistransient severity), as measured by plant instrumentation, in order to take advantage of thetransient definition conservatism. All of the EPRI studies took advantage of actualnumbers of plant transients.

The second explicit method involves a reduced set of fatigue design curves, such as thoseproposed by Argonne National Laboratory (ANL) [11, 12]. This method has been usedby Idaho National Engineering Laboratory (INEL, now INEEL) to recalculate the CUFsfor fatigue-sensitive component locations in early and late vintage CE PWRs, early andlate vintage Westinghouse PWRs, early and late vintage GE BWRs, and B&W PWRs.The results of the INEEL calculations were published in NUREG/CR-6260 [13]. TheINEEL calculations took advantage of the conservatism available in the numbers of actualplant transients, relative to the numbers of design-basis transients, but did not recalculatestress ranges based on actual plant transient profiles.

Another explicit approach can be used. This approach probabilistically determines thefatigue-sensitive component locations to be included in the fatigue aging managementprogram. For a plant that has incorporated risk-informed methodology to redefine itsplant inservice inspection (ISI) program, this involves assessing the impact of reactorwater environmental effects on the number of locations to be inspected, the type ofexamination method to be used, and the frequency of examination. This explicit approachis exemplified by the probabilistic calculations carried out by Pacific Northwest NationalLaboratory (PNNL) [3] in support of the NRC staff evaluation of GSI-190. All of thefatigue-sensitive component locations evaluated in NUREG/CR-6260 were re-evaluated,in probabilistic terms, in Reference 3. It should be pointed out that a variant on theprobabilistic approach is possible; that is, the deterministic results from NUREG/CR-6260and the EPRI generic studies can be reviewed and incorporated as “expert opinion” in therisk-informed inservice inspection (RI-ISI) process.

The justifications for the two implicit approaches depend upon negotiating the degree ofconservatism in the fatigue design cycle or fatigue CUF reduction factor with the NRCstaff. Industry calculations using the Fen (intermittently applied reactor waterenvironmental effects) approach would argue for a knockdown factor of 2 to 2.5, whichwould set a CUF limit of 0.4 to 0.5 as the decision criterion limit for fatigue monitoring orinservice examination. This limit is approximately the same as the Examination CategoryB-J threshold for Class 1 piping weld examination in Subsection IWB of the ASME CodeSection XI. If NRC contractor calculations (reduced S-N curves) are used, knockdownfactors from 6 to 10 are possible, which would argue for a CUF decision threshold of 0.1to 0.15.

29-7

In the following sections, an approach is outlined that uses all of the evaluationsperformed to-date, coupled with plant-specific fatigue evaluation and ISI program results,to systematically address the fatigue TLAA for license renewal and the incorporation ofreactor water environmental effects in a manner consistent with that shown in Figure 1.

III. NUREG/CR-6260 Results.

As a part of the effort to close GSI-166 (later GSI-190) for operating nuclear powerplants during the current 40-year licensing term, INEEL evaluated fatigue-sensitivecomponent locations at plants designed by all four U. S. nuclear steam supply system(NSSS) vendors – Westinghouse (PWR), CE (PWR), GE (BWR), and B&W (PWR). Forthe first three vendor designs, component locations from both early and late vintagedesigns were evaluated. Reference 13 provides the results of those evaluations. Theearly-vintage Westinghouse PWR calculations are chosen here for illustrating the availableinformation.

The fatigue-sensitive component locations chosen for evaluation in Reference 13 for theearly-vintage Westinghouse PWR calculations were: (1) the reactor vessel shell and lowerhead; (2) the reactor vessel inlet and outlet nozzles; (3) the pressurizer surge line,including the pressurizer and hot leg nozzles; (4) the reactor coolant system pipingcharging system nozzle; (5) the reactor coolant system piping safety injection nozzle; and(6) the residual heat removal system Class 1 piping. For the latter three componentlocations, INEEL performed representative design basis fatigue calculations, since early-vintage Westinghouse PWRs typically utilized an ANSI B31.1 design basis for most of theClass 1 piping.

Reference 13 found that four of the six fatigue-sensitive component locations hadcumulative usage factors (CUFs) less than 1.0 for both 40 and 60 years of operation,including the effects of the reactor water environment through a reduced set of fatiguedesign curves [11]. However, in two of these four cases (the charging nozzle and thesafety injection nozzle), the evaluators needed to use NB-3200 (design by finite elementanalysis) methods, instead of NB-3600 piping analysis methods, in order to obtain morerealistic stress distributions. Reference 13 states, in part: “Whereas the NB-3200 and NB-3600 results were comparable for the Salt computed for the nozzle-to-charging systemjunction region, the Salt was reduced by more than a factor of four in the nozzle body(considered to be a branch connection in the NB-3600 analysis) region using the NB-3200finite element analysis.” Similarly, the report states that the corresponding reduction inSalt for the safety injection nozzle body was more than a factor of 10.

These findings were corroborated in Reference 9, where a recirculation system teeconnection in a BWR plant was evaluated by both NB-3600 and NB-3200 analysismethods. The reduction in the CUF at the most fatigue-sensitive location (the insidecorner of the tee) was approximately a factor of 10. Such reductions in CUF are typical ofNB-3200 calculations, providing a clear demonstration of a portion of the conservatisminherent in standard ASME Code NB-3600 piping fatigue analyses.

The remaining two fatigue-sensitive component locations from the Reference 13evaluation of early-vintage Westinghouse PWRs had the following results. The design-

29-8

basis CUF for the inside surface of the reactor vessel lower head near the shell-to-headtransition, where core support guides are welded to the interior of the shell, wasdetermined to be 0.290. When the NUREG/CR-5999 interim fatigue curves were applied,the 40-year CUF increased to 0.891, with the 60-year CUF at 1.337. Since the majorcontributors to these CUFs were 200,000 hypothetical alternating cycles of frictional andvibration loads, and since these design-basis cycles are conservatively defined, thiscomponent location is not considered to represent an issue. The conservatism in thedefinition of the design-basis cycles more than compensates for the reactor waterenvironmental effects.

This is not the case for the other most fatigue-sensitive component location – the insidesurface of the hot leg to surge line nozzle safe end. The design-basis CUF, includingthermal stratification loads, was determined to be 0.900. When the revised interim fatiguecurves of NUREG/CR-5999 were used, with actual numbers of cycles instead of design-basis numbers of cycles, the 40-year CUF increased to 4.248. The 60-year CUF increasedto 6.372. The INEEL calculations were also based on 30-second stresses, implying thatthermal stratification loads approximate a thermal shock. Some reduction of the CUFvalues might be possible with more realistic transient definitions.

These results from Reference 13 for early vintage Westinghouse plants are typical of thefindings for other plant types. In general, the majority of fatigue-sensitive componentlocations were found to have 60-year CUFs less than 1.0. A relatively few componentlocations were found to have environmentally-adjusted CUFs for 60 years that exceeded1.0, using the reduced S-N curve approach. In this sense, the Reference 13 results matchup very well with the probabilistic results of Reference 3.

IV. EPRI Generic Studies.

The calculations reported in Reference 13 were based on the interim reduced fatiguedesign curves given in Reference 11. Such an approach penalizes the component locationfatigue analysis unnecessarily, since research has shown that a combination ofenvironmental conditions is required before reactor water environmental effects becomepronounced. The strain rate must be sufficiently low and the strain range must besufficiently high to cause continuing rupture of the passivation layer that protects theexposed surface area. Temperature, dissolved oxygen content, metal sulfur content, andwater flow rate are additional variables to be considered. In order to take theseparameters into consideration, EPRI and the General Electric Company jointly developeda method, called the Fen approach [6], that permits reactor water environmental effects tobe applied intermittently, as justified by parameter combinations.

The Fen approach was then applied by EPRI and its contractors to fatigue –sensitivecomponent locations in four types of nuclear power plants – an early-vintage CE PWR[5], an early-vintage Westinghouse PWR [7], and both early-vintage [9] and late-vintage[8] GE BWRs. Component locations similar to those evaluated in Reference 13 wereexamined in these generic studies. A later study on reactor water environmental effects onClass 1 branch piping also used the Fen approach [10].

The early-vintage Westinghouse PWR results from Reference 7 provide an excellent

29-9

example of the benefits of the Fen approach. Actual transient information from plantinstrumentation (e.g., hot leg temperature, pressurizer water temperature) over threecycles of operation (1994, 1995, and 1996) was extrapolated both backward and forwardin time, in order to calculate the environmental factor to be applied to the design-basisCUF. The maximum value for any of the surge line locations – pressurizer shell,pressurizer surge nozzle, pressurizer spray nozzle, pressurizer water temperatureinstrument nozzle, RCS hot leg surge nozzle, and charging nozzle – was 1.91. Thiscompares to an environmental multiplier of over 7 from Reference 13. The intermittentapplication of the environmental effects provides a reduction in CUF of almost 4. Theeffect of using actual transient information instead of design-basis transient definitions iseven greater, approaching a factor of 10.

These findings were confirmed in the other PWR study – of an early-CE PWR [5]. Again,the pressurizer surge line was studied in detail. This calculation provided a directcomparison with the same component location evaluated in Reference 13. Section 5.2.3of Reference 13 describes the results of calculating the fatigue CUF for the pressurizersurge elbow in older vintage CE plants. The component location is fabricated fromaustenitic stainless steel, with a design-basis CUF of 0.705 for 40 years of operation.Reference 13 cites a value for the 40-year CUF of 8.07 when the ANL interimenvironmental fatigue design curves [11] were used. This CUF value is more than tentimes the design-basis CUF. Based on the Fen approach, Reference 5 showed that theenvironmental multiplier was actually only about two. The difference between the twoenvironmental multipliers is again about a factor of 4.

Reference 13 also provided the results of CUF recalculation based on the removal of twodifferent elements of conservatism: (1) the use of actual numbers of operating transients(as opposed to prescribed numbers of design-basis transients), and (2) the use of revisedinterim environmental fatigue design curves recommended by ANL (as opposed to theNUREG/CR-5999 [11] interim environmental fatigue design curves). Examining only thesecond of these elements of conservatism, the difference between the interim and therevised interim fatigue curves is shown in Figures 3-18 and 3-19 of Reference 13 and canbe on the order of a factor of 2. In particular, the difference is especially significant forstress ranges up to 80 ksi. The alternating stress ranges for the two dominant load pairsare 44.89 and 36.99 ksi, respectively, so that the allowable number of design cycles are4,878 cycles and 15,639 cycles, respectively, when using the NUREG/CR-5999 [10]curve. These number of allowable cycle values increased to 7,095 cycles and 35,238cycles, respectively, when using the revised (0.001 %/sec strain rate) curve. Because thenumber of design-basis transients was 15,000 cycles and 72,025 cycles, respectively, forthe two load pairs, the contributions to the CUF decrease to 2.114 and 1.987,respectively, with a total CUF of approximately 4.5. This recalculation shows that theexcess conservatism in the NUREG/CR-5999 [11] stainless steel interim fatigue designcurve is of the order of 1.5 to 2.2, and could have been higher. The ratio between theReference 5 environmental multiplier and the revised Reference 13 environmentalmultiplier is now about 2, instead of 4.

Collectively, the EPRI generic studies show that the difference between intermittent andcontinuous application of reactor water environmental effects is approximately a factor of

29-10

four. This ratio, which represents the difference between the industry Fen method and thereduced S-N curve method, is one of the factors that can be used to implicitly or explicitlyincorporate reactor water environmental effects into fatigue evaluations and agingmanagement program decision criteria. The ratio of about 10 that represents the effect ofactual thermal transients versus severe design-basis thermal transients is another factor. Afactor representing the ratio of the number of actual thermal cycles experienced versus thenumber of design-basis thermal cycles is also available.

V. EPRI Branch Piping Studies.

Reference 10 evaluated six Class 1 small-bore branch piping systems at Oconee Units 1, 2,and 3, including reactor water environmental effects. The piping systems evaluated were:

• Core flood piping and the associated reactor pressure vessel nozzle

• Pressurizer spray piping and the spray nozzle in the pressurizer

• High pressure emergency injection (HPI) and normal makeup piping

• Decay heat removal piping

• Letdown piping, and

• Loop drain piping.

Because of the low amplitudes of cyclic stress, many of these piping systems were foundto be exempt from Class 1 fatigue analysis. These exemptions included:

• Portions of the core flood piping outboard of the isolation valve

• Decay heat line piping, including the hot leg nozzle

• Letdown piping, including the cold leg nozzle

• Loop drain piping, including cold leg nozzles, and

• Cold leg pressurizer spray nozzles.

The three systems requiring fatigue evaluation were the pressurizer spray piping, theHPI/emergency piping, and the HPI/makeup piping. For the HPI/emergency andHPI/makeup piping locations evaluated, reactor water environmental effects led to CUFson the order of 1.5 to 2.2. If a moderate environmental effects factor of either 1.5 or 2.0is available for these locations, fatigue enhanced by reactor water environmental effects isnot a concern.

The pressurizer spray system for PWRs represents a potential thermal fatigue concernbecause of severe thermal transient conditions, including potential stratification.Component locations for this system were not analyzed for the older vintage W plant inNUREG/CR-6260 [13]. However, the pressurizer spray nozzle and two other pressurizerspray line locations were evaluated for thermal fatigue at the Oconee plant [10], includingreactor water environmental effects and stratification transients. The procedure used forthese environmental fatigue calculations in the Oconee study was similar to that used forthe Calvert Cliffs pilot study [5], in that on-line monitoring of critical thermocouplelocations was performed to determine actual thermal transient behavior. The evaluation

29-11

also included a virtual dissolved oxygen instrument based on Oconee plant experience,calculation of Fen multipliers for the thermal transients during the baseline measurementperiod, and extrapolation to 40 and 60 years of operation.

All locations, except for the spray nozzle, were determined to have CUF values less than1.0 for 60 years, including significant Fen multipliers that ranged from 3.9 to 7.4.Therefore, with the exception of the spray nozzle, branch piping locations do not present amajor environmental fatigue concern.

For the Oconee pressurizer spray nozzle, the original design-basis CUF of 0.60 boundedthe CUF calculated for the inside and outside surfaces of the base metal (not exposed tothe reactor water environment). For the stainless steel cladding, which is exposed to thereactor water environment, the design-basis CUF was 0.15. Environmental and thermalstratification effects were such that, for auxiliary spray cooldown, the design basis CUF of0.15 increased to 0.973 (an Fen multiplier of 6.488) when the worst case thermal transientswere used. Extrapolating the environmentally-enhanced CUF for the stainless steelcladding to 60 years would cause the CUF to exceed 1.0 (CUF = 1.46).

If a moderate environmental effects factor of either 1.5 or 2.0 is available forconsideration, even the spray nozzle would be of no concern for reactor waterenvironmental effects.

VI. NRC Staff Issues. The NRC staff raised a number of issues relative to the EPRIgeneric studies that have been the subject of a number of meetings over the past twoyears. Those issues can be separated into four topics:

(1) The NRC staff has asked for revised Fen calculations for carbon and/or low-alloy steel components analyzed in the EPRI generic studies, using analternating strain range threshold of 0.07 % (as opposed to 0.10 % argued bythe industry), above which the more recent ANL data would apply.

(2) The NRC staff has asked that the moderate environmenta l effects factor bereduced to 3.0, rather than the factor of 4.0 argued by the industry.

(3) The NRC staff has asked for revised Fen calculations for austenitic stainlesssteel components analyzed in the EPRI generic studies, using an alternatingstrain range threshold of 0.097 % (as opposed to 0.10 % argued by theindustry), above which the more recent ANL data would apply.

(4) The NRC staff has asked that the moderate environmental effects factor bereduced to 1.5, rather than the factor of 2.0 argued by the industry.

In addition, the NRC staff has asked that the industry explicitly address differencesbetween the mean ASME air curve used as the basis for the ASME Code fatigue designcurve and the mean air curve established by ANL. In essence, this implies a “double”environmental correction – one correction to account for differences in fatigue lifebetween ANL air data and ANL simulated reactor water data, and the other correction toaccount for differences between the ASME Code design basis mean air curve and theANL mean air curve.

VII (a). EPRI Generic Study Recalculations -- Carbon and Low-Alloy Steels. More

29-12

recent laboratory fatigue data in simulated LWR reactor water environments have beengenerated by ANL for carbon and low-alloy steels, and published in NUREG/CR-6583[14]. These data do not differ substantially from the data used in the EPRI genericstudies. However, the change in strain threshold may have a significant effect, and thateffect has been evaluated.

The recalculation is based on one of the examples contained in EPRI TR-105759 [6], aBWR carbon steel feedwater piping location with a design-basis fatigue usage factor of0.1409 for 40 years. An alternating stress threshold of 30 ksi (approximating thealternating strain threshold of 0.10 %) was used initially to adjust the incremental fatigueusage for eight (8) out of thirty-one load pairs, giving an additional (environmental)fatigue usage of 0.0477, for a 40-year adjusted total of 0.1886. The overall Fen multiplierin this case was 1.38 (1.68 for the eight affected load pairs).

Reducing the alternating stress threshold to 21 ksi (approximating the alternating strainthreshold of 0.07 %) would require an environmental adjustment for at least six additionalload pairs. Assuming that the Fen multiplier of 1.68 would continue to apply for the 14load affected load pairs, the estimate for the adjusted fatigue usage factor would be0.1409 – 0.0803 + 1.68 (0.0803) = 0.1955. The overall Fen multiplier increases only to1.39.

Because the additional load pairs that would have to be included contribute relativelysmall increments to the total CUF, it is unlikely that changing the strain range thresholdwould present a significant issue. It should be pointed out also that the intermittent F enmultiplier is less than a moderate environmental effects factor of either 3.0 or 4.0.

VII (b). EPRI Generic Study Recalculations -- Austenitic Stainless Steels. More recentlaboratory fatigue data in simulated LWR reactor water environments have also beengenerated by ANL for austenitic stainless steels, and published in NUREG/CR-5704 [15].These data are substantially more penalizing than the data used in the EPRI genericstudies. The updated analytical expressions for the mean fatigue initiation life of austeniticstainless steel in both air and the laboratory-simulated light water reactor (LWR)environment are given below.

For the laboratory air environment, the number of cycle for fatigue crack initiation, N, is:

ln (N) = 6.703 - 2.030 ln (εa - 0.126) + T1*η*, (1)

where εa is the strain amplitude (%) and T1* and η* are transformed temperature andstrain rate, respectively, defined as follows:

T1* = 0 (T < 250°C)T1* = [(T-250)/525]0.84 (250 < T < 400°C)T = temperature, °C

η* = 0 (η > 0.4%/sec)η* = ln(η1/0.4) (0.0004 < η < 0.4%/sec)

29-13

η* = ln(0.0004/0.4) (η < 0.0004%/sec)

For the laboratory simulation of the reactor water environment,

ln (N) = 5.768 - 2.030 ln (εa - 0.126) + T2*η*O*, (2)

where T2* and O* are transformed temperature and dissolved oxygen (DO), respectively,defined as follows:

T2* = 0 (T < 200°C)T2* = 1.0 (T > 200°C)

O* = 0.260 (DO < 0.05 ppm)O* = 0.172 (DO > 0.05 ppm)

The analytical expression for Fen is then obtained as the following:

Fen = exp [0.935 + η*(T1* - T2*O*)] (3)

For the case of relatively low temperature (< 200oC), a low (bounding) strain rate, andeither high or low dissolved oxygen, the environmental shift is 2.55. For relatively hightemperature (> 200oC), low dissolved oxygen, and a low (bounding) strain rate, theenvironmental shift may be as high as 15.35, although there is a reduction above 250 oCwhere the environmental factor decreases to about 3.20 at 340 oC.

For most of the component locations evaluated in the EPRI generic studies, these mostrecent data do not pose a problem for the demonstration that the 60-year CUF is less than1.0, including reactor water environmental effects. Again, a significant benefit accrues tothe Fen approach in this regard, since most of the penalizing thermal transients lie belowthe threshold temperature of 200oC. Therefore, the environmental shift is relatively low,provided that a different multiplier is used for the portions of the transient that are aboveand below 200oC. However, for the most fatigue sensitive locations, (e.g., surge lineelbow in PWRs), the environmentally-adjusted CUF increases over that calculated in theEPRI generic studies by a factor of about 2. If a moderate environmental effects factor of2.0 is available, there are no component locations for which the CUF cannot be shown tobe less than 1.0 for 60 years. If a moderate environmental effects factor of 1.5 is available,the environmentally-adjusted CUF for one or two locations exceeds 1.0 for 60 years. Ifno moderate environmental effects factor is available, the most fatigue-sensitivecomponent locations will have 60-year CUFs that approach 2.0.

A CUF greater than 1.0 does not guarantee fatigue cracking, due to other conservatismsin ASME Code Class 1 fatigue design analyses. However, there is some potential forfatigue cracking to occur when the most recent data on austenitic stainless steels are usedin the environmental fatigue evaluations for one or two of the most fatigue sensitivecomponent locations.

29-14

VIII. NUREG/CR-6674 Results.

Reference 3 based the probabilistic risk assessments on the component location fatiguecalculations of NURGE/CR-6260 [13]. Forty-seven (47) component locations in sevendifferent types of light-water-cooled power reactors (e.g., new vintage CombustionEngineering PWRs, older vintage Combustion Engineering PWRs, Babcock & WilcoxPWRs, new vintage Westinghouse PWRs, older vintage Westinghouse PWRs, newvintage General Electric BWRs, and older vintage General Electric BWRs) wereevaluated. The probability of through-wall cracking after either 40 years or 60 years ofoperation, using a very conservative approach, is fairly high for nine of those componentlocations:

• Surge line elbows in new vintage and older vintage CE plants,

• RPV outlet nozzles in older vintage CE plants, B&W plants, and new vintage Wplants,

• Charging system nozzles in new vintage W plants,

• RHR inlet transitions in new vintage W plants,

• RHR line pipe straight pipe in new vintage GE plants, and

• Recirculation system RHR return lines in older vintage GE plants

However, for many of the other component locations, typical cumulative probabilities ofthrough-wall cracking for either 40 years or 60 years of operation are on the order of 10 -5

to 10-8. Such probabilities do not warrant any extraordinary inservice inspectionrequirements. Even for a component location such as a charging nozzle safe end in a newvintage CE plant, with a cumulative 60-year probability of through-wall cracking of about10-3, the efficacy of extraordinary inservice inspection may not be warranted. The 40-yearprobability is about 10-4. Therefore, based upon an interpretation of the very conservativethrough-wall cracking probabilities calculated in Reference 3, only a few componentlocations warrant a review and possible revision of the plant inservice inspection programto accelerate the inspection frequency.

The reason for the relatively large sample of component locations with a cumulativeprobability of through-wall cracking can be traced to the very conservative assumptionsused in the probabilistic calculations. While these assumptions are reasonable as the basisfor regulatory bounding calculations, the assumptions are too conservative to serve as thebasis for risk-informed inservice inspection (RI-ISI). One example should suffice.Detailed stress calculations – in particular, the stress distribution across the componentthickness were not available from NUREG/CR-6260. Only the peak cyclic stresses forsurface locations, many of which represent discontinuities and stress concentrations, wereavailable. Therefore, Reference 3 assumed that, for peak cyclic stresses with surfaceamplitude greater than 310 Mpa (45 ksi), the uniform (e.g., membrane) component was310 Mpa (45 ksi), with the remainder of the stress amplitude assigned to a “gradient”category. The gradient category was defined as a nonlinear distribution of stress acrossthe component thickness typical of step change thermal loading. This assumption is much

29-15

less conservative than assuming a uniform membrane stress across the componentthickness equal to the peak cyclic stress. However, uniform thermal stress amplitudes aretypically very small, in comparison to the linear component of stress and, in particular, thenonlinear component of stress.

By reducing some of these conservatisms to more accurate “best estimate” levels, it wouldbe possible to estimate the cumulative probability of through-wall cracking on a morerealistic basis, thereby providing improved information to RI-ISI programs at the plants.Until such modified probabilistic calculations are completed, Reference 3 has identified aconservative set of component locations that can be used as input for RI-ISIconsideration, based on a cumulative through-wall crack probability for 40 years of 10-3.

IX. Section XI Program.

The results from NUREG/CR-6260, NUREG/CR-6674, and from the EPRI genericstudies can be used to demonstrate that, for most component locations, the TLAAresolution requirements of either 10CFR54.21(c)(1)(i) or 10CFR54.21(c)(1)(ii) are met.Only a few exceptions can be identified, depending upon the conservatism of the analyticalmethods used. Two general classes of component locations have been identified for whichaging management program demonstrations would be required:

• Fatigue-Sensitive PWR Pressurizer Surge Line Locations

• Fatigue-Sensitive PWR Class 1 Branch Piping Locations

In addition, for some PWRs, the pressurizer spray line could be an exception because ofthermal stratification loads. For these three sets of fatigue-sensitive component locations,the requirements of 10CFR54.21(c)(1)(iii) should be used. In other words, it is necessaryto demonstrate that the effects of fatigue damage on the intended function(s) will beadequately managed for the period of extended operation.

A sample of the surge line welds is examined as a part of the plant ISI program every tenyears, in accordance with the requirements of the ASME Boiler and Pressure Code,Section XI, Subsection IWB, Examination Categories B-J and B-P. Surge line weldsselected for the ISI program, by nature of their size, require a volumetric examination. Anumber of the welds should have been examined ultrasonically during the first twoexamination periods at PWR plants, and a somewhat larger sample of welds, and perhapsbase metal, may need to be examined ultrasonically during subsequent ten-year periods.The justification for the enlarged sample could be based on RI-ISI methods.

Most RI-ISI programs implemented to-date do not explicitly include reactor waterenvironmental effects. Therefore, for the most fatigue-sensitive locations, the ten-year ISIinterval may be called into question. This issue can be addressed by using the results offlaw tolerance evaluations, coupled with previous ultrasonic (UT) examinations, assumingthey have not revealed any indications.

In addition to the fatigue-sensitive pressurizer surge line locations, the other set ofcomponent locations that are deemed to be fatigue-sensitive, especially when reactorwater environmental effects are considered, is fatigue-sensitive Class 1 small-bore piping.This set is defined as Class 1 piping and fittings with nominal pipe size (NPS) less than

29-16

four inches in diameter. In particular, the concern is for those small-bore piping locationswhere either (1) a geometric discontinuity is present and the thermal loading is representedby thermal shock, or (2) stratified flow conditions are known to exist.

In order to manage fatigue crack initiation and growth for this set of fatigue-sensitivesmall-bore piping locations, the plant may choose to develop a One-Time Inspection forASME Section III Class 1 Piping Less Than 4 Inch License Renewal NPS Program.

X. Summary.

The objective of this paper is to examine existing fatigue evaluations and establish aprocess for evaluating the fatigue TLAA for both license renewal and reactor waterenvironmental effects. Results from NUREG/CR-6260 have been coupled with resultsfrom industry generic and plant-specific studies.

First, utilizing the results of fatigue monitoring programs, it can usually be shown that thenumber of design-basis transients assumed in the fatigue design basis envelopes the actualnumber of transient events experienced for either 40 or 60 years of operation. Thus, theTLAA on fatigue for license renewal can be shown to be acceptable based on 10 CFR54.21(c)(1)(i) of the License Renewal Rule.

Second, a technical assessment that introduces reactor water environmental effects intothe fatigue evaluation can be performed based on the evaluations contained in plant-specific evaluations, NUREG/CR-6260, and the EPRI generic studies. The results of thistechnical assessment show that relatively few component locations are left that do notmeet allowable acceptance criteria. The major exceptions are the PWR surge line elbows,welds, and nozzles. Another possible exception is Class 1 small-bore piping undersignificant thermal stratification loading. Thus, many components are consideredacceptable through an analytical re-evaluation of the cumulative usage factor (CUF) for 60years (10 CFR 54.21(c)(1)(ii) of the Rule).

Finally, for those locations remaining that do not satisfy acceptance criteria when reactorwater environmental effects are included, an approach that demonstrates that fatigueeffects are adequately managed for the intended operating period is used (10 CFR54.21(c)(1)(iii) of the Rule). This is accomplished by incorporating all appropriatefatigue-sensitive locations into relevant ISI programs. RI-ISI programs can considerreactor water environmental effects during the expert panel decision criteria step to furthersupport this approach. Collectively, this process, which is outlined in Figure 1, can beused to address the TLAA on fatigue and also reactor water environmental effects andjustify operation for an extended period.

In order to address the remaining fatigue issues associated with license renewal, fouradditional generic tasks could be undertaken. The first task would be to complete theassembly of information and interpretation of data to support current license renewalapplications. The second task would be to review the technical basis for the Reference 3risk study, removing excess conservatism so that the results can be used more directly insupport of ISI programs for detection and sizing of environmentally-assisted fatigue crackinitiation and growth at fatigue-sensitive component locations. Third, for those fatigue-sensitive component locations for which the probabilities of environmentally-assisted

29-17

fatigue crack initiation, through-wall crack growth, and subsequent leakage are sufficientlyhigh – such as surge line elbows in PWR plants, a cooperative lead-plant inspectionprogram could be initiated. Since surge line elbows are fabricated from either wrought orcast austenitic stainless steel, the technical difficulties of ultrasonic examination to detectand size such cracks needs to be evaluated. Finally, the feasibility and cost of testinglaboratory-scale and perhaps full-scale fatigue test articles under realistic reactor waterenvironmental conditions should be assessed. This last task could be considered forpotential interaction with a much larger effort called the Japanese Environmental FatigueTesting Program.

XI. References.

1. SECY-95-245, “Completion of the Fatigue Action Plan ,” James M. Taylor,Executive Director for Operations, U. S. Nuclear Regulatory Commission,Washington, DC, September 25, 1995.

2. Memorandum, Ashok C. Thadani, Director, Office of Nuclear Regulatory Research,to William D. Travers, Executive Director for Operations, Closeout of GenericSafety Issue 190, “Fatigue Evaluation of Metal Components for 60 Year PlantLife,” U. S. Nuclear Regulatory Commission, Washington, DC, December 26, 1999.

3. NUREG/CR-6674 (PNNL-13227), “Fatigue Analysis of Components for 60-YearPlant Life,” Pacific Northwest National Laboratory for the U. S. Nuclear RegulatoryCommission, Washington, DC, June 2000.

4. Informal Handout at a Meeting Between U. S. Nuclear Regulatory CommissionStaff and Industry, Christopher I. Grimes, Director, License Renewal ProjectDirectorate, U. S. Nuclear Regulatory Commission, Washington, DC, November 17,1999.

5. “Evaluation of Thermal Fatigue Effects on Systems on Systems Requiring AgingManagement Review for License Renewal for the Calvert Cliffs Nuclear PowerPlant,” Report No. EPRI TR-107515, Structural Integrity Associates for EPRI,Palo Alto, CA, January 1998.

6. “An Environmental Factor Approach to Account for Reactor Water Effects in LightWater Reactor Pressure Vessel and Piping Fatigue Evaluations ,” EPRI TR-105759, General Electric Company for EPRI, Palo Alto, CA, December 1995.

7. “Evaluation of Environmental Fatigue Effects for a Westinghouse Nuclear PowerPlant,” Report No. EPRI TR-110043, Structural Integrity Associates for EPRI,Palo Alto, CA, April 1998.

8. “Evaluation of Environmental Thermal Fatigue Effects on Selected Components ina Boiling Water Reactor Plant,” Report No. EPRI TR-110356, Structural IntegrityAssociates for EPRI, Palo Alto, CA, April 1998.

9. “Environmental Fatigue Evaluations of Representative BWR Components,” ReportNo. EPRI TR-107943, General Electric Company for EPRI, Palo Alto, CA, May1998.

29-18

10. “Effect of Environment on Fatigue Usage for Piping and Nozzles at Oconee Units1,2, and 3,” Report No. EPRI TR-110120, Structural Integrity Associates forEPRI, Palo Alto, CA, December 1999.

11. “Interim Fatigue Design Curves for Carbon, Low-Alloy, and Austenitic StainlessSteels in LWR Environments,” NUREG/CR-5999 (ANL-93/3), Argonne NationalLaboratory for the U. S. Nuclear Regulatory Commission, Washington, DC, April1993.

12. “Statistical Analysis of Fatigue Strain-Life Data for Carbon and Low-Alloy Steels ,”NUREG/CR-6237 (ANL-94/21), Argonne National Laboratory for the U. S.Nuclear Regulatory Commission, Washington, DC, August 1994.

13. “Application of NUREG/CR-5999 Interim Fatigue Curves to Selected NuclearPower Plant Components,” NUREG/CR-6260 (INEL-95/0045), Idaho NationalEngineering Laboratory for the U. S. Nuclear Regulatory Commission, Washington,DC, March 1995.

14. “Effects of LWR Coolant Environments on Fatigue Design Curves of Carbon andLow-Alloy Steels,” NUREG/CR-6583 (ANL-97/18), Argonne National Laboratoryfor the U. S. Nuclear Regulatory Commission, Washington, DC, March 1998.

15. “Effects of LWR Coolant Environments on Fatigue Design Curves of AusteniticStainless Steels,” NUREG/CR-5704 (ANL-98/31), Argonne National Laboratoryfor the U. S. Nuclear Regulatory Commission, Washington, DC, April 1999.

29-19

Figure 1. Approach for License Renewal Fatigue Assessment

Robert E. NickellDavid A. GerberGary L. Stevens

Reactor Water Environmental Effects on Fatigue Life

AN APPROACH TO EVALUATING THE EFFECTS OF REACTORWATER ENVIRONMENTS ON FATIGUE LIFE

International Conference on Fatigue of Reactor ComponentsSilverado Country Club & Conference Center

Napa, California

July 31- August 2, 2000Robert E. Nickell, Applied Science & Technology

David A. Gerber and Gary L. Stevens, Structural Integrity Associates

29-20



Implicit Methods for Incorporating Reactor Water Environmental Effects

◆ Reduce the number of design-basis transients or cycles that wouldinitiate aging management program actions by a factor thatconservatively incorporates known and applicable reactor waterenvironmental effects.

◆ Use the design-basis CUF, reduced by a factor that conservativelyaccounts for known and applicable reactor water environmental effects,as a decision criterion for initiating aging management program actions.

◆ A common feature of both implicit approaches is that design-basistransient severity is used, as opposed to actual operating transientprofiles. Design-basis transient severity has been shown to provideconservatism on the order of a 10 to 20, or even more, relative to actualoperating transients.

◆ Industry calculations support a factor of 2.0 to 2.5 (intermittentapplication of environmental effects); regulatory calculations support afactor of 6 to 10.

An Approach for Evaluating the Effects ofReactor Water Environments on Fatigue Life

29-21



Explicit Methods for Incorporating Reactor Water Environmental Effects

◆ Deterministically recalculate the CUF for a number of fatigue-sensitivecomponent locations, incorporating reactor water environmental effectsin the recalculation through an environmental multiplier, Fen.

◆ Deterministically recalculate the CUF for a number of fatigue-sensitivecomponent locations, incorporating reactor water environmental effectsin the recalculation through a reduced set of fatigue design curves.

◆ Probabilistically evaluate the fatigue-sensitive component locations to beincluded in the aging management program, with reactor waterenvironmental effects incorporated into the probabilistic calculations.

◆ In all three cases, generic calculations are available to assist in theevaluation (NUREG/CR-6260, NUREG/CR-6674, EPRI generic studies).

◆ Calculations feed into risk-informed ISI programs as “expert”information.


29-22



NUREG/CR-6260 Generic Results

◆ Six or More Fatigue-Critical Component Locations Evaluated forSeven Reactor Types (Early and Late Vintage W PWRs; Early and LateVintage GE BWRs; Early and Late Vintage CE PWRs; B&W PWRs)

◆ For many component locations, direct application of the ANL fatiguecurves produced CUFs greater than 1.0 for both 40 and 60 years

◆ By reducing the number of cycles from the design basis to(extrapolated) actual cycles, or by making strain rate adjustments tothe ANL fatigue curves, the CUF was reduced below 1.0 for both 40and 60 years for most of the affected component locations

◆ NB-3200 finite element calculations for NB-3600 piping locations wereable to demonstrate substantial reductions in the design-basis CUF,even with reactor water environmental effects

◆ For a few component locations (e.g., surge lines), further calculations(e.g., elastic-plastic analysis) were thought to be necessary to reducethe CUF below 1.0


29-23



NUREG/CR-6674 Generic Results

◆ The PNNL risk study used appropriate bounding assumptions to estimatecontributions to core damage frequency (CDF) for 47 of the most fatigue-sensitive Class 1 component locations in seven types of LWRs

◆ The extension of these bounding calculations showed only sixcomponent locations (e.g., PWR surge line elbows, PWR RPV outletnozzles, BWR recirculation system RHR return lines) with sufficientlyhigh through-wall cracking frequency to serve as the basis for risk-informed inspection

◆ The extension of the CDF calculations to through-wall cracking frequencyand leakage rates, using the bounding assumptions, may be tooconservative to be used as the basis for risk-based management ofpotential reactor water environmental effects

◆ Reduction of some of the conservatism will limit the number of locationsto be inspected even further


29-24



NUREG/CR-6674 Generic Results (Continued)

◆ 60-year cumulative through-wall cracking probabilities > 10-4

❖ Surge-line elbows in newer vintage and older vintage CE plants

❖ RPV outlet nozzles in older vintage CE plants, B&W plants, and newer vintageW plants

❖ Charging system nozzles in newer vintage W plants

❖ RHR inlet transitions in newer vintage W plants

❖ RHR line piping in newer vintage GE plants

❖ Recirculation system RHR return lines in older vintage GE plants

◆ Most significant conservatism is probably the stress assumptions acrossthe component wall thickness for crack growth calculations


29-25



EPRI Initial Generic Studies: Reactor Water Environmental Effects

◆ On-Line Transient and Fatigue Usage Monitoring

“FatiguePro : On-Line Fatigue Usage Transient MonitoringSystem,” Report No. EPRI NP-5835M, Electric Power ResearchInstitute (1988)

❖ Environmental module (e.g., virtual dissolved oxygeninstrumentation) added to on-line monitoring system in 1997

◆ Reactor Water Environmental Effects Evaluation (Fen)Methodology

“An Environmental Factor Approach to Account for ReactorWater Effects in Light Water Reactor Pressure Vessel and PipingFatigue Evaluations,” Report No. EPRI TR-105759, Electric PowerResearch Institute (December 1995)


29-26



EPRI Generic Applications Studies: Reactor Water Environmental Effects

◆ “Evaluation of Thermal Fatigue Effects on Systems on SystemsRequiring Aging Management Review for License Renewal for theCalvert Cliffs Nuclear Power Plant,” Report No. EPRI TR-107515, ElectricPower Research Institute (January 1998)

◆ “Evaluation of Environmental Thermal Fatigue Effects on SelectedComponents in a Boiling Water Reactor Plant,” Report No. EPRI TR-110356, Electric Power Research Institute (April 1998)

◆ “Evaluation of Environmental Fatigue Effects for a WestinghouseNuclear Power Plant,” Report No. EPRI TR-110043, Electric PowerResearch Institute (April 1998)

◆ “Environmental Fatigue Evaluations of Representative BWRComponents,” Report No. EPRI TR-107943, Electric Power ResearchInstitute (May 1998)


29-27



Generic Findings: Reactor Water Environmental Effects

◆ Increase in CUF from Fen is between 1.0 to 1.6, and typically less than3.0, for older vintage CE plants, older vintage Westinghouse plants,and both older and newer vintage BWR plants

◆ Almost all modified CUFs fall within the moderate environmentaleffects margin in the ASME Code Section III fatigue design curves

◆ The conservatism associated with design-basis thermal transientdefinitions versus actual thermal transient profiles is worth a factor ofthe order of 2 to 20

◆ The actual number of cycles versus numbers of design-basis cyclesrepresents an added conservatism


29-28



Major Generic Findings: Reactor Water Environmental Effects

◆ The combination of actual thermal transients and actual numbers ofthermal transients, plus the application of the Fen methodology anddetailed finite element analysis, where necessary, provided a genericdemonstration that CUFs are less than 1 for both 40 and 60 years foralmost all fatigue-sensitive locations

◆ Concern about austenitic stainless steel component locations duringperiods of operation at low temperature and low dissolved oxygenlevels remains, and has a significant effect on EPRI generic findings

◆ Field experience does not confirm the generic austenitic stainlesssteel findings, probably because of applicability of some of thelaboratory simulations of reactor conditions, including flow rates andsurface strain distributions


29-29



EPRI Generic Studies: NRC Staff Issues

◆ Revised Fen calculations for carbon/low-alloy steels using an alternatingstrain threshold for reactor water environmental effects of 0.07 %, ratherthan 0.10 %, and NUREG/CR-6583 data (A recalculation from EPRI TR-105759 shows that these changes would have little effect on the EPRIgeneric results).

◆ Reduced moderate environmental effects factor of 3.0 (instead of 4.0)for carbon/low-alloy steels and 1.5 (instead of 2.0) for austeniticstainless steels (Recalculations show that these changes would havelittle effect on the EPRI generic results).

◆ Revised Fen calculations for austenitic stainless steels using analternating strain threshold for reactor water environmental effects of0.097 %, rather than 0.10 %, and NUREG/CR-5704 data (Recalculationsshow that the NUREG/CR-5704 data have an effect on the order of afactor of 2, thus significantly affecting the EPRI generic results).


29-30



EPRI Generic Applications Studies: Reactor Water Environmental Effects

◆ “Effect of Environment on Fatigue Usage for Piping and Nozzles atOconee Units 1, 2, and 3,” Report No. EPRI TR-110120, Electric PowerResearch Institute (December 1999)

❖ Exemption from Class 1 fatigue analysis for portions of the core flood pipingoutboard of the isolation valve; decay heat line piping, including the hot legnozzle; letdown piping, including the cold leg nozzle; loop drain piping,including cold leg nozzles; and cold leg pressurizer spray nozzles.

❖ Three systems required fatigue evaluation: (1) pressurizer spray piping; (2)HPI/emergency piping; and (3) HPI/makeup piping.

❖ All locations, except for the pressurizer spray nozzle, were found to haveCUFs (including reactor water environmental effects) < 1.0 for 60 years, evenwith environmental multipliers of 3.9 to 7.4.

❖ Pressurizer spray nozzle had 60- year CUF, including reactor waterenvironmental effects, of 1.46 (Fen = 6.49).


29-31



Application to a Westinghouse PWR

◆ NUREG/CR-6260 evaluated (1) the reactor vessel and lower head; (2) thereactor vessel inlet and outlet nozzles; (3) the pressurizer surge line,including the pressurizer and hot leg nozzles; (4) the reactor coolantsystem piping charging system nozzle; (5) the reactor coolant systempiping safety injection nozzle; and (6) the residual heat removal systemClass 1 piping.

◆ Four of the six locations had CUFs < 1.0 for both 40 and 60 years,including reduced fatigue design curves. Two of these four locations thatwere found to be < 1.0 needed NB-3200 calculations to reduce the CUF.

◆ The remaining locations were: (1) the inside surface of the hot leg tosurge line nozzle safe end (40-year CUF = 4.25; 60-year CUF = 6.37), and(2) inside surface of the vessel lower head near the shell-to-headtransition, where core support guides are welded to the shell interior (40-year CUF = 0.89; 60-year CUF = 1.34).


29-32



Application to a Westinghouse PWR (Continued)

◆ Application of the EPRI/GE Fen approach (intermittent reactor waterenvironmental effects, as opposed to reduced fatigue design curves)reduces the CUF by a factor of between 2 and 4.

◆ Application of laboratory data from NUREG/CR-5704 increases the CUFby a factor of about 2.

◆ The number of mechanical loading transients for the core support guidesis very conservative

◆ Therefore, the inside surface of the lower head of the vessel is not amajor fatigue issue.

◆ The inside surface of the hot leg to surge line nozzle safe end must beincluded in the plant inservice inspection program, with an inserviceinspection interval justified by flaw tolerance calculations.


29-33



SUMMARY

◆ At least five (two implicit and three explicit) approaches are available forimplementation by license renewal applicants as the basis for decisioncriteria to determine the fatigue-sensitive component locations for whichfatigue crack initiation and growth, including the effects of reactor waterenvironments, must be managed.

◆ Generic information available from both NRC and EPRI contractor reportsis able to show that almost all component locations have CUFs for 60years, including reactor water environmental effects, that are less than1.0.

◆ The remaining component locations (e.g., the inside surface of the hot legto surge line nozzle safe end in PWRs) must be included in the plantinservice inspection program, with an inservice inspection intervaljustified by flaw tolerance calculations, including appropriate reactorwater environmental effects.


29-34



SPARES


29-35



Presentation Outline

◆ Recent Developments - NRC Staff

-- 11/17/99: Risk Study Results, Grimes Statement

-- 12/1/99: Craig Letter to ASME BNCS

-- 12/3/99: NRC Staff Presentation to ACRS, Issue Closure

-- 12/26/99: Memorandum From Thadani to Travers

◆ Recent Developments - Industry

-- 1/10/00: Industry Action Meeting at NEI

-- 1/27/00: Industry Presentation to ASME BNCS

-- 2/25/00: EPRI Letter to ASME BNCS

-- 2/28-29/00: ASME Code Meetings, Daytona Beach


29-36



Recent Developments

◆ GSI-190 has been determined not to be a safety issue and is closed

◆ Licensees continue to be asked by NRC staff to address the effects ofreactor water environments on fatigue life as aging managementprograms are formulated in support of license renewal

◆ The PNNL risk study shows no significant difference in calculated coredamage frequencies (CDFs) between 40 and 60 years of life

◆ The NRC staff has determined that ALWRs certified to 10 CFR 52 that aredesigned to ASME Code fatigue requirements for 60 years of life havesufficient design conservatism to compensate for reactor waterenvironments

◆ The industry argues that license renewal regulatory oversight is notrequired for environmental fatigue other than a TLAA demonstration thatthe fatigue CLB is applicable for 60 years of service


29-37



Recent NRC Fatigue Activities

◆ Extension of probabilistic risk study to 60 years of life has now beencompleted

◆ “Fatigue Analysis of Components for 60-Year Plant Life,” F. A.Simonen, et al., has been published as NUREG/CR-6674, June 2000.

◆ Results were discussed at the NRC Fatigue Workshop on November17, 1999, which also included a distributed statement by Chris Grimeson the requirements for license renewal applicants.

◆ Results were used as the basis for closure of GSI - 190 and discussedwith the Advisory Committee on Reactor Safeguards (ACRS) onDecember 3, 1999

◆ ACRS issued favorable letter on the NRC staff findings


29-38



Chris Grimes’ Statement, Fatigue Meeting, 11/17/99

◆ “In particular, we anticipate that fatigue damage beyond the existing40-year license term may not have sufficient impact on the coredamage frequency to justify imposing a requirement to revise thecurrent licensing basis to include an environmental factor in all of thefatigue design locations.”

◆ “At least for the critical Class 1 locations, the staff would expect thatan environmental factor would be accounted for in the process bywhich licensees decide when to take corrective action in their fatiguemonitoring programs.”


29-39



Chris Grimes’ Statement, Fatigue Meeting, 11/17/99 (Continued)

◆ We encourage the industry ….. to establish an industry consensuson decision criteria to ensure that corrective actions will be takenbefore fatigue damage jeopardizes the ability of plant systems andcomponents to perform their intended functions.

◆ We would expect to continue to review the fatigue monitoringprograms on a plant-specific basis, relative to some accounting forenvironmental effects in the decision criteria, until an industryconsensus is established and endorsed by the NRC.


29-40



Memorandum, Ashok C. Thadani, Director, Office of Nuclear RegulatoryResearch, to William D. Travers, Executive Director for Operations,dated December 26, 1999

◆ The advanced light water reactors (ALWRs) that have been certifiedunder 10 CFR Part 52 were designed for a 60-year life expectancy.The associated fatigue analyses accounted for the design cyclesbased on a 60-year plant life but did not account for theenvironmental effects as addressed in GSI - 190. However, the staffhas concluded that there is sufficient conservatism in the fatigueanalyses performed for the generic 60-year ALWR life to accountfor environmental effects.


29-41



Initial Industry Studies: Analytical Conservatism - Transient Definitions

◆ Fatigue design analysis conservatism study

“Evaluation of Conservatisms and Environmental Effects in ASMECode, Section III, Class 1 Fatigue Analysis,” Report No. SAND94-0187, Sandia National Laboratories (August 1994)

◆ Findings

Conservatisms in the transient definitions, analytical procedures,fatigue design curves, and other elements of the ASME Code explicitfatigue design process appear to more than compensate for reactorwater environmental effects


29-42



Recommendations, January 10, 2000, Industry Meeting (Continued)

◆ The PNNL reactor water environmental effects risk study should beevaluated further by the industry

-- Reduce conservative assumptions used in bounding analyses

-- Perform sensitivity studies

-- Single component Monte Carlo calculations for benchmarking

◆ EPRI should continue its planning exercise to prepare for testinglaboratory-scale fatigue test articles under realistic reactor waterenvironmental conditions, while examining the potential forcoordination with the Japanese EFT program


29-43



Recommendations, January 10, 2000, Industry Meeting (Continued)

◆ As an alternative to a very expensive test program, EPRI shouldcontinue its evaluation of a cooperative industry program forinservice examination of the most fatigue-sensitive componentlocations (e.g., surge line elbows), considering base metalinspection capability, wrought versus cast stainless steel issues,and the potential for a lead plant program.

◆ EPRI should prepare a short white paper in support of the SNOCHatch LR proposal to use reactor water environmental fatigueeffects management decision criteria -- i.e., fatigue design basisCUF threshold less than 1.0 (e.g., 0.1) for determining locations tobe managed, with design-basis transient cycle monitoring andincluding non-CUF approaches as needed.


29-44



ASME Code Meetings, February 28-29, 2000, Daytona Beach, FL

◆ ASME Section XI Task Group on Operating Plant Fatigue Assessment,Monday, February 28, 2000

-- Presentation, Stan Rosinski, EPRI, on January 10th Recommendations

-- Presentation, Mike Davis, Duke Energy, on EPRI MRP Thermal Fatigue

◆ ASME BNCS has directed Subcommittee III to take responsibility for theissue, with Dick Barnes (SG Design) in charge of oversight group;meeting of oversight group on Monday night to gather information; NRCstaff (Joe Muscara) claim industry is “stonewalling” the issue, while NRChas worked on the issue for over ten years

◆ Subgroup on Fatigue Strength (SG-FS) of the service Subcommittee onDesign (SC-D) took several actions to implement reduced S-N curves


29-45



SUMMARY

◆ GSI - 190 is closed; no actions relative to the ASME Section III Class 1fatigue design basis could be justified by the NRC staff, for either 40 or 60years of service, because of the small conditional contributions to coredamage frequency

◆ The bounding risk calculations show a measurable increase in through-wall cracking and potential leakage for a few of the most fatigue-sensitivecomponent locations; the NRC LR staff feel justified in requesting thatreactor water environmental effects be incorporated into agingmanagement program decision criteria

◆ This increase in through-wall cracking and leakage has not been observedin actual service, probably because the risk study was extremelyconservative and because the laboratory data on reactor waterenvironmental effects are not directly applicable


29-46



NRC Fatigue Activities◆ Generic Safety Issue (GSI) 78, “Monitoring of Design Basis Transient

Fatigue Limits for Reactor Coolant System,” was identified in June1983 -- determine whether transient monitoring (cycle counting) isnecessary at operating plants

◆ GSI-166, “Adequacy of Fatigue Life of Metal Components,” wasidentified in April 1993, as a consequence of license renewal fatigueissues raised for current operating plants

◆ -- CUF > 1, reactor water environmental effects, Class 1 B31.1piping, adequacy of inservice inspection

◆ NRC Fatigue Action Plan, SECY-95-245, “Completion of the FatigueAction Plan,” September 25, 1995


29-47



NRC Fatigue Activities

◆ Memorandum, Eric S. Beckjord, Director, Office of Nuclear RegulatoryResearch, to Ashok C. Thadani, Associate Director for Inspection &technical Enforcement, Office of Nuclear Reactor Regulation,September 23, 1994

◆ Addressed Generic Issue No. 78, “Monitoring of Fatigue TransientLimits for the Reactor Coolant System”

◆ The impact on core-damage frequency of fatigue failure in piping,using the environmentally-adjusted fatigue curves of NUREG/CR-5999,was found to be negligible in comparison to that from RPV failure.

◆ Crack initiation does not insure through-wall cracking; the conditionalprobabilities of through-wall cracking are small (flaw tolerant piping).

◆ Contribution to core-damage frequency is insensitive to CUF, even upto CUF = 10.


29-48



NRC Fatigue Activities

◆ The probabilistic risk study results from Generic Issue 78 were laterincorporated into the NRC Fatigue Action Plan, which was completedand documented in SECY-95-245. The staff concluded that the riskfrom fatigue failure of the primary coolant pressure boundarycomponents was very small, based on a plant life of 40 years. Theimpact of a license renewal period of 20 years on fatigue of metalcomponents was to be considered in the resolution of Generic SafetyIssue 166. Later, Generic Safety Issue 190 was established to addressthis subject.


29-49



NRC SECY-95-245 Findings

◆ “the [NRC] staff believe that no immediate staff or licensee action isnecessary to deal with the fatigue issues addressed by the [FatigueAction Plan]”;

◆ “fatigue failure of piping is not a significant contributor to core-meltfrequency”;

◆ “the [NRC] staff does not believe it can justify requiring a backfit ofthe environmental fatigue data to operating plants”; and

◆ “the [NRC] staff believe that the [Fatigue Action Plan] issues shouldbe evaluated for any proposed extended period of operation forlicense renewal.”


29-50

30-1

30 DESIGN BASIS ENVIRONMENTAL FATIGUEEVALUATION AT OCONEE


1300 Harris BoulevardCharlotte, NC 28262

Arthur F. DeardorffWilliam F. Weitze


San Jose, CA 95118-1557

Tim BrownDuke Energy Corporation7800 Rochester HighwaySeneca, SC 29672-0752

30-3

DESIGN BASIS ENVIRONMENTAL FATIGUE EVALUATION AT OCONEE

Stan T. Rosinski Arthur F. Deardorff Tim BrownEPRI William F. Weitze Duke Energy Corporation1300 Harris Boulevard Structural Integrity Associates 7800 Rochester HighwayCharlotte, NC 28262 3315 Almaden Expressway, Suite 24 Seneca, SC 29672-0752

San Jose, CA 95118-1557

ABSTRACT

A Class 1 fatigue evaluation of piping systems was conducted at the Oconee nuclear plant. Thisanalysis included an assessment of the effects of proposed environmental fatigue rules. The developmentof environmental correction factors (Fen) for application to individual load set pairs considered strain ratesbased on quasi-steady, thermal transient and dynamic loading effects. The approach was applied toseveral components and showed that environmental effects would be quite significant when using theproposed environmental correction to the fatigue curves. A critical evaluation of the applicability of theenvironmental correction factors is presented.

INTRODUCTION

Recent studies have determined that reactor water environment can affect component fatigue life. Anevaluation methodology has been developed by EPRI where environmental correction factors (Fen) can beapplied to the partial usage factors determined in a conventional ASME Code fatigue analysis [1]. Thisproposed methodology has been used as part of an industry position regarding the management ofenvironmental fatigue. This methodology was applied in several system-specific analyses to demonstrateuse of the methodology and the general impact of reactor water effects on component fatigue [2-5].During these studies, modified procedures were developed as new research results became available [6],including revised equations for carbon/low-alloy steel [7] and for stainless steel [8].

To assess the effects of environment on fatigue in representative pressurized reactor components, aClass 1 fatigue analysis was performed on the nonisolable stainless steel piping attached to the reactorcoolant system at the Oconee Nuclear Station. The analysis included the effects of thermal stratificationin some portions of the piping that had been identified during in-plant testing. The proposedenvironmental rules were then used to develop environmental factors that could be applied to the ASMEClass 1 usage factor analysis.

COMPONENTS EVALUATED

ASME Code fatigue analysis was performed on Class 1 piping systems attached to the reactor coolantsystem out to and including the first isolation valve. Fatigue usage factors at a number of locations in theCore Flood, Pressurizer Spray and High Pressure Injection/Makeup piping systems were shown to requiredetailed Class 1 fatigue evaluation. The remaining piping systems were shown to be exempt from therequirements for a cyclic fatigue analysis and were not evaluated further since fatigue effects would beexpected to be minimal.

ENVIRONMENTAL CORRECTION FACTOR, Fen

The environmental correction factor (Fen) is defined as the ratio between the fatigue life in air to thatin water, both at the same temperature. It may be a function of several variables, including material,

30-4

water oxygen content, strain rate, and temperature. (For ferritic materials, the sulfur content is also afactor.) In performing an environmental fatigue evaluation, a separate Fen factor may be applied to eachof the partial usage factors (associated with each load set pair) determined in a conventional ASME Codefatigue analysis. The equations for describing Fen for stainless steels [11] used in this evaluation are asfollows:

Fen = exp [0.935 + η* (T1* - T2* O*)]

where:η* = 0 (for η > 0.4%/sec)η* = ln (η/0.4) (for 0.0004 ≤ η ≤ 0.4%/sec)η* = ln (0.0004/0.4) (for η < 0.0004 %/sec)η = strain rate, %/sec

T1* = 0 (for T < 250°C)T1* = [(T – 250)/525]0.84 (for 250 ≤ T < 400°C)T2* = 0 (for T < 200°C)T2* = 1.0 (for T ≥ 200°C)T = material temperature, °C

O* = 0.260 (for DO < 0.05 ppm)O* = 0.172 (for DO ≥ 0.05 ppm)DO = dissolved oxygen, ppm

A load set pair strain amplitude threshold was assumed (0.097 percent) below which there would beno effect of environment on fatigue usage [6].

ENVIRONMENTAL EVALUATION OF LOAD SET PAIRS

As shown in the above equations, the environmental correction factor is a function of temperature,strain rate, and dissolved oxygen content. These factors are easily definable in laboratory testing.However, in real component fatigue analysis, it is not so apparent as to how to choose the appropriatevariables to use in the analysis. The dissolved oxygen level may be different for the two events of a loadset pair. Temperature and strain rate are highly variable during a thermal transient. The stress peaks andvalleys of the transient pair may be far removed from one another (in time) and there may be other lesssignificant stress change transients between the load set pair peaks and valleys. Thus, some assumptionsmust be made in determining how to apply laboratory data to actual component evaluation.

Effective Environmental Factor (F′′′′en)

In actual reactor operating conditions, the conditions that exist for the peak and valley of a load setpair may be different and the strain rate and fluid and local surface temperature will vary during each ofthe contributing transients. Reference 7 presented an improved approach that determines an effectiveenvironmental factor as follows:

εε−ε

−+=′ ∫

ε

εd

1F1F

max

th thmax

enen or ε

ε−ε=′ ∫

ε

εd

FF

max

th thmax

enen

30-5

where:Fen = instantaneous environmental factor based on current conditions during transientε = strain, relative to that at the most compressive stress stateεmax = algebraic maximum strain for load set range pairεth = strain threshold value

For a load set pair, the strain range, equivalent to εmax - εmin, may be determined as:

E

S2 altminmax =ε−ε

where:εmin = algebraic minimum value of strain for load set pair, arbitrarily taken as zero for the

integration processSalt = alternating stress from fatigue analysis, ksiE = modulus of elasticity from fatigue curve, ksi

This equation offers the opportunity to evaluate the time history of conditions that occur during atransient pair, rather than just assuming the worst possible value for the conditions that might exist, whendetermining environmental effects.

Piping Analysis Equations

The peak stress range and alternating stress amplitude for a load set pair for piping analysis is givenby Equations 11 and 14 of ASME Section III, NB-3650 [9]:

213

bbaaab33i0

2200

11p

TE1

1TEK

)1(2

1

TTECKMI2

DCK

t2

PDCKS

∆αν−

+∆αν−

+

α−α++=

and

Salt = 2

SK pe

The factor Ke is determined based on Equation 10, that describes the range of primary plus secondarystresses for the load set pair. The nomenclature is given in the ASME Code and will not be repeated here.

As shown in the piping equations, the peak stress range may be a function of pressure changes,change in moments, and changes in thermal conditions that occur during the two conditions of a load setpair. When evaluating strain rate effects, strain rates associated with pressure and thermal expansionmoment changes are expected to be relatively low and may or may not occur simultaneously with thethermal effects. The strain rates due to thermal effects may be high or low, depending on the transient;high strain rates are expected to occur simultaneously with high stress range rapid thermal transients.Dynamic loading (e.g., due to an earthquake, if present) would be expected to have a very high strain rate.

30-6

In a piping analysis, the stress range is generally calculated based on the extreme range of pressureand moment for the two events for the load set pair. For the local thermal effects, the maximumcontribution of each of the terms may be added directly, or time phasing may be considered. The stressindices are defined such that the maximum multiplier is used for each individual term in the expressionfor stress intensity. Thus, the reported total stress range is a conservative assessment of the potentialstress intensity excursion between two load states. This approach has been shown to be conservative foruse in design of piping components, allowing the many locations in piping systems to be economicallyevaluated using modern computer techniques.

Transient Components for Strain Rate Analysis

Figure 1 depicts the stress time history between two states of a load set pair. State 1 is chosen as thatevent that causes relatively compressive thermal stresses when the thermal terms are combined. State 2creates a state of stress that is tensile relative to State 1. Dynamic (seismic) stresses can increase theoverall strain range and can be assumed in analysis to occur with one or both of the thermal stressextremes, usually concurrently with the most severe load set pair.

For evaluating an effective environmental factor (F′en), only the rising strain portion of the transientthat exceeds the threshold strain (εth) has been evaluated since the integration is from the positive strainthreshold to a larger value of strain. This is reasonable since it is the tensile straining that tends to rupturethe material surface layer that protects it from the water environment.

Evaluation of F′′′′en

The thermal stress time history in piping analysis may be determined based on a combination of thethree thermal terms from Equation 11 of ASME Section III, NB-3650. Thus, for the inside surface ofpiping, where the most significant thermal effects are experienced,

( )

α−α±

ν−∆α−

ν−∆α= bbaaab33

213eth TTECK

1

TE

)1)(2(

TEKKS

In this equation, each of the thermal terms is a function of time. For simplicity, it is assumed that theeffect of the elastic-plastic multiplier Ke is uniformly distributed over the total strain range. In reality, theexcess plasticity would occur only after exceeding 3Sm, but the analysis would be quite complicated toassume other than the uniform distribution. Since the piping analysis equations combine stress rangesusing absolute values, the sign of the third term is chosen as that which produces the maximum stressrange between two event pairs. This thermal stress time history is determined for each of the two load setstates.

The strain rate time history is determined from the derivative of the thermal stress time history:

dt

Sd

E

1

dt

d th=ε

The inside surface temperature, which can potentially be used to assess environmental factors, can beestimated from the time history of temperatures:

30-7

21ba

surf T2

T

2

TTT ∆+

∆−

+=

When evaluating the complete stress range between two thermal transients, the maximum thermalstress near the end of State 1 may not be equal to the minimum thermal stress near the beginning of State2. This difference of thermal stress must be considered, and can be considered to have a very slow strainrate since States 1 and 2 may not necessarily be closely related, or even occurring at the same time. Forexample, State 1 could represent heatup and State 2 could represent cooldown, with other less severetransients occurring between them.

The stress range due to other load changes (e.g., pressure and thermal expansion moments) can bedetermined from:

dynthalto SSS2S δ−δ−=δ

where:δSdyn = stress range due to dynamic loads, including effects of Ke

δSth = stress range due to thermal effects, including effects of Ke

The stress intensity range due to dynamic loads may be determined separately (e.g., by comparing thestress range for the load set pair with and without the seismic event).

In evaluating each load set and load set pair, State A is defined as that with the most compressivestress, or with an upward thermal transient. State B is defined as the other event. In addition, each of thestates may or may not include a significant thermal transient. Also, the load set pair may or may notinclude an earthquake or other dynamic event.

In Reference 1, any load set pair with seismic content was considered to exclude environmentaleffects. This assumes that the seismic strain range would be relatively large. If the seismic contributionis small, there may still be environmental effects for the load set pair. Thus, for any pair where thedynamic loading is a contributor, the F′en will be evaluated using a high strain rate for the dynamic strainrange contribution that will minimize environmental effects for the dynamic contribution. The seismicstress range will be considered to be with State A and/or State B per the original non-environmentalfatigue analysis. Thus, for determining environmental factors there are five strain contributions that areevaluated.

ε1 - Associated with a dynamic event at the relatively compressive state, existing only if a dynamicload is defined for State A.

ε2 - Associated with a thermal transient with State A, existing only if State A contains a thermaltransient.

ε3 - Associated with "slow" events and equal to the total strain range, minus the dynamic and thermalranges associated with both states A and B.

ε4 - Similar to ε2 except that it occurs with State B transient conditions.

ε5 - Similar to ε1, except that it occurs with State B.

30-8

In determining F′en, only the portion of the strains above the strain amplitude threshold areconsidered, consistent with the approach proposed in Reference 7. No evaluations were conducted thatevaluated the complete strain range in the integration to determine F′en although this might be an alternateapproach that could be considered.

Evaluation of Oxygen and Temperature Effects

During normal power operation in pressurized water reactors, coolant oxygen concentration istypically below the threshold of detection of about 0.002 ppm. Since the components analyzed are allstainless steel, low oxygen is controlling in that it yields a higher environmental factor. Since at least oneof the load states was always at normal operating conditions, environmental factors were based on lowoxygen (< 0.05 ppm).

Assessment of temperature affects on environmental factors is more complex and not so well defined.The question is: What temperature should one use in the fatigue equations to assess the values of Fen?Several choices existed:

1) Use instantaneous surface temperature in the integration of Fen,

2) Use maximum Fen over the fluid temperature range, or3) Use an average Fen over the fluid temperature range.

It is not clear if it is the fluid or metal temperature that produces the environmental effect, or if it isthe instantaneous metal temperature or the temperature that existed for a long time before the transientoccurred. This is one of the difficulties in translating the laboratory test data to a realistic fatigueevaluation. Four approaches were evaluated (Table 1) to show the sensitivity of the temperatureassumption on the analysis outcome. Different approaches were applied to the dynamic strain ranges ascompared to the quasi-steady portion of the ranges associated with each load set pair.

In determining the environmental factor for the quasi steady non-transient portions of the range, theapproaches to determine Fen were:

Cases 1 and 3: Fen was evaluated over the complete range of temperature associated with the twoassociated load states. The maximum Fen factor over the range was conservatively chosen.

Cases 2 and 4: An alternate means of determining the temperature effects was evaluated as suggestedin Reference 7:

1) If the maximum temperature was below the temperature threshold, then an average Fen wasdetermined over the actual temperature range, by choosing a number of temperature steps over theinterval and determining an integrated average. For the cases examined in this evaluation, Fen inthis case was constant and equal to exp(0.935) = 2.547.

2) If the maximum temperature was above the temperature threshold and the minimum temperaturewas below the threshold, then an average Fen was determined over the range between the maximumand the threshold, using an integrated average as above.

3) If the minimum temperature was above the temperature threshold, then an integrated average Fen

was determined over the actual temperature range.

For determination of the environmental factors for thermal transients, the actual surface temperatureduring the transient was used for Cases 1 and 2. As a more conservative alternative, the fluid temperaturein the range leading to the highest value of Fen considering both load sets were used for Cases 3 and 4.

30-9

For any dynamic portions of the strain range, the temperature for the associated quasi-steady orthermal transient condition was used.

Adjustment of F′′′′en

The current ASME Code fatigue curves contain nominal factors of 2 on stress and 20 on cycles belowthe developed mean curve. A portion of the factor of 20 on cycles was to account for moderateenvironmental effects. Thus, after computing F′en, for any load set pair evaluated using the ASME Codefatigue design curves, the value was adjusted to account for this factor. Thus, the final adjusted effectiveenvironmental factor is determined by:

F*en =

φ′enF

(but ≥ 1)

For carbon and low alloy steels, φ = 4[2]. For stainless steels, φ = 2.0 [2]. The NRC, in recentinteraction, indicated that the factor for carbon/low-alloy steels should be 3.0 and for stainless steels thefactor should be 1.5 [6]. In the analyses described in this report, φ = 1.5 was used for stainless steel.

For Alloy 600 materials the F′en was previously defined as a constant value of 1.49. It is reasonableto assume that the same environmental factor inherent in the ASME Code fatigue design curves forcarbon/low-alloy steels and stainless steels is also inherent in the fatigue design curve for Alloy 600.Assuming that the factor of 1.5 also applies to Alloy 600, there would be no net effect of environment forAlloy 600 since F′en is a constant (1.49). Thus, environmental effects of any Inconel 600 welds orcomponents were not evaluated.

RESULTS OF ENVIRONMENTAL EVALUATION

The reactor coolant system attached piping considered in this study was analyzed to the Class 1requirements of the 1983 ASME Code [9]. For performing the environmental evaluation, the pipinganalysis was not “fine tuned” to reduce any conservatisms prior to addressing environmental effects. Theanalysis was conducted using standard techniques, making reasonable assumptions to assure that theusage factors and other Code limits could be met without consideration of environmental effects. Somerepresentative locations with relative high computed cumulative fatigue usage were chosen to assess theeffects of environment as shown in Table 2.

Table 3 lists the transients analyzed, along with the transient numbers used in the fatigue tables. Insome cases, transients were grouped to reduce analytical effort where the transients were not especiallysevere. For spray transients, S1, S2, etc. in a transient number refer to a pressurizer spray event duringthat transient. For the transients, “up” indicates increasing temperature transients (causing compressivethermal stress) and “down” indicates decreasing temperature transients (causing tensile thermal stress).Some transients had both “up” and “down” portions. In addition, there were specific stratificationtransients identified for each system, identified with only a letter designator.

The environmental assessment was based on the fatigue tables taken from the Code fatigue analysisoutput. Load sets with no transient, or with a slow transient, or with stress always decreasing during thetransient were treated as quasi-steady state. For load sets with thermal transients, beginning and endtimes were chosen to capture only increasing thermal stresses, thus defining times of εmin and εmax.

30-10

Table 4 summarizes the results, showing all four cases. The difference in the approaches forchoosing Fen as a function of temperature were minor for the various cases. Tables 5 through 9 givedetailed fatigue usage calculations, both with and without environmental effects, for the mostconservative evaluation approach (Case 3). Details of determining Fen is provided in Reference 10.

COMMENTARY ON PIPING ANALYSIS AND ENVIRONMENTAL EFFECTS

ASME Class 1 piping analysis is conducted to the rules in ASME Section III, NB-3650. A set ofequations is provided that conservatively shows that the rules of NB-3200 are met. There are manyconservatisms in the “design by rule” approach using the piping equations. There have been no identifiedcases where the process is not conservative, except for cases where the loading conditions were notknown at the time of the initial plant design.

The proposed environmental rules have been developed using laboratory tests simulating a light waterreactor environment. In order to complete testing in a reasonable amount of time, the testing must beconducted continuously. The resulting stress(or strain) time history bears little resemblance to that whichoccurs in an operating nuclear plant, or the stress range pairing required by the ASME Code. Theproposed environmental rules provide another level of conservatism into the piping design equations, andconsiderably add to the complexity and labor needed to design and analyze piping systems.

Consider the fatigue analysis in Table 6. The earthquake loadings have been conservativelyconsidered to occur simultaneously with the most severe thermal transients. The first transient pairing isthat of the termination of an HPI injection with the initiation of an HPI injection. Certainly, these areclosely related transients, but for the environmental evaluation, the combination includes first thetermination and then the injection, which is backwards relative to reality. The next load set pair is end ofheatup paired with termination of HPI injection following a rapid depressurization. In this case, the rangeof pressures is certainly opposite of the thermal stresses, and the two transients are not mechanisticallyconnected. Further down the table, there is a combination of stratification and no stratification. For thestratification case, the events could be closely connected such that the compressive stress followed bytensile stress sequencing could occur.

For the HPI Makeup piping, the transient pairing could be mechanistic in that one of the controllingcombinations is a loss of makeup followed by reinitiation of makeup. In this case, the tensile transientwould closely follow the compressive transient. This type of transient is generally not the general rule inreactor systems, however. What is found in most fatigue analyses is that there is no relationship betweenthe transients such that there would be long periods of time between the two transients of a load set pairfor a material passivating layer to re-establish. This long period of re-passivating between the extreme ofthe strain cycles did not exist in the fundamental test data from which the environmental factors weredetermined.

The conservatism in the design analysis environmental approach was also demonstrated in industryevaluations of components at another PWR where actual plant data were evaluated using fatiguemonitoring [2]. For the fatigue monitoring environmental evaluation, actual plant transient sequence andmagnitudes were considered. The resulting multipliers on fatigue usage to account for environmentaleffects (without consideration of the factor φ) were 1.4 to 1.6 [2]. Thus, using environmental factors in adesign basis analysis is extremely conservative.

CONCLUSIONS

Increases in cumulative usage factor for piping fatigue analyses when environmental effects areconsidered in a design analysis were shown to be very significant in all cases, and ranged from a factor of

30-11

2.8 to 8.8. Cumulative usage factors using an environmental correction exceeded the ASME Code limitsfor some locations.

It is noted that the methodology for analyzing piping systems in the ASME Section III, NB-3650,contains a number of individual conservative assumptions which contribute to additional conservatism inan environmental effects analysis. For example, there is no consideration of cycle sequence wherecompressive transients might mechanistically follow tensile transients in event pairing. The analysis andthe proposed environmental fatigue curves also do not account for long periods of steady state stressconditions between transients that might mitigate some environmental effects.

The extent of the increases in cumulative usage factor leads to a prediction of piping failures due tofatigue. The fact that design basis transient fatigue failures are not common indicates that there is a largeamount of conservatism in the combination of the original fatigue calculations and in the formulas used tocalculate environmental effects.

REFERENCES

1. An Environmental Factor Approach to Account for Reactor Water Effects in Light Water ReactorPressure Vessel and Piping Fatigue Evaluations, EPRI, Palo Alto, CA: 1995, TR105759.

2. Evaluation of Thermal Fatigue Effects on Systems Requiring Aging Management Review for LicenseRenewal for the Calvert Cliffs Nuclear Power Plant, EPRI, Palo Alto, CA: 1997, TR-107515.

3. Environmental Fatigue Evaluations of Representative BWR Components, EPRI, Palo Alto, CA:1998, TR-1079434.

4. “Evaluation of Environmental Fatigue Effects for a Westinghouse Nuclear Power Plant, EPRI, PaloAlto, CA: 1998.

5. Evaluation of Environmental Thermal Fatigue Effects on Selected Components in a BWR Plant,EPRI, Palo Alto, CA: 1998, TR-110356.

6. Letter, Douglas J. Walters (NEI) to Christopher Grimes (NRC), “Request for Additional Informationon the Industry's Evaluation of Fatigue Effects for License Renewal,” April 8, 1999.

7. Chopra, O. K., and Shack, W. J., “Overview of Fatigue Crack Initiation in Carbon and Low AlloySteels in Light Water Reactor Environments,” Journal of Pressure Vessel Technology, Volume 121,February 1999.

8. NUREG/CR-5704, “Effects of LWR Coolant Environments on Fatigue Design Curves of AusteniticStainless Steels,” US NRC, April 1999.

9. ASME Boiler and Pressure Vessel Code, Section III, 1983 Edition with no Addenda.10. Effect of Environment on Fatigue Usage for Piping and Nozzles at Oconee Units 1,2, and 3, EPRI,

Palo Alto, CA: 1999. TR-110120.11. Mehta, H.S., "An Update on the EPRI/GE Environmental Fatigue Evaluation Methodology and its

Applications", PVP Vol. 356, Probabilistic and Environmental Aspects of Fracture and Fatigue,ASME 1999, pp. 183-193.

30-12

Table 1. Cases for Evaluating Fen Variation with Temperature

Case Non-Transient Loads Thermal Transients

1 maximum Fen over temperaturerange

surface temperature at eachtime point

2 average Fen over temperaturerange

surface temperature at eachtime point

3 maximum Fen over temperaturerange

maximum Fen over inputtemperature range

4 average Fen over temperaturerange


Table 2. Locations for Environmental Assessment

System Unit Location Description CUF

Core Flood 1-3 5 Nozzle safe-end to pipeweld

0.033

HPI/Emergency 1 175A Nozzle safe-end to pipeweld

0.193

HPI/Makeup 1 116 Pipe to valve weld 0.177HPI/Emergency 2&3 160 Pipe to valve weld 0.640

HPI/Makeup 2&3 305 Pipe to valve weld 0.291Pressurizer Spray 1 135C

130Pipe to aux. spray tee

weldPipe to valve weld

0.0520.042

Pressurizer Spray 2&3 210A 4" pipe to reducer weld 0.081

30-13

Table 3. List of Transients and Transient Groups

Number Transient Description1A Heatup, may include hydrotest (at pressure indicated)1ACF Core Flood Check Valve Test During Heatup1BCF Cooldown with Core Flood Decay Heat Removal Start1B Cooldown (zero load if so indicated)2A/2B Power Increase/Decrease3 Power Loading4 Power Unloading5 Step Load Increase6 Step Load Decrease7 Step Load Reduction8A Reactor Trip, Loss of Flow8B Reactor Trip, High Temp.8C Reactor Trip, High Press.8HPI Manual HPI Activation after Trip9 Rapid Depressurization10 Change of RCS Flow11 Control Rod Withdrawal12 Hydrotest14 Control Rod Drop16 Steam Line Failure15 Loss of Station Power17A Loss of Feedwater17B Turbine Bypass19A Feed and Bleed Operations20D Loss of Makeup22A HPI System Test22C Pressurizer Heat Loss EvaluationGroups M downand M up

Transients 2A/2B, 3, 4, 5, 6, and 10 for HPI and CoreFlood analyses

Group S down Transients 7, 8A, 8B, 8C, 14, 15, and 17A for HPI andCore Flood analyses

Group S up Transients 7, 8A, 8B, 14, and 15 for HPI and CoreFlood analyses

Group T down Transients 11 and 17B for HPI and Core Floodanalyses

Group T up Transients 8C, 11, 17A, and 17B for HPI and CoreFlood analyses

Group S1 down Transients 3, 5, and 10 for Pressurizer Spray analyses

Group S2 up Transients 2, 3, 4, and 14 for Pressurizer Sprayanalyses

Group S3 up Transients 7 and 8B for Pressurizer Spray analyses

Group S4 up Transients 11, 16, 17A, and 17B for Pressurizer Sprayanalyses

30-14

Table 4. Cumulative Fatigue Usage Factors with and without Environmental Effects

System Location Original Case 1 Case 2 Case 3 Case 4

Core Flood 5 0.033 0.092 (2.75) 0.092 (2.75) 0.092 (2.75) 0.092 (2.74)

HPI/Emergency

175A 0.193 0.844 (4.36) 0.756 (3.92) 0.904 (4.69) 0.812 (4.21)

HPI/Makeup 116 0.177 1.562 (8.83) 1.560 (8.82) 1.563 (8.84) 1.528 (8.64)

HPI/Emergency

160 0.640 2.131 (3.33) 1.963 (3.07) 2.443 (3.82) 2.275 (3.56)

HPI/Makeup 305 0.291 2.313 (7.95) 2.311 (7.94) 2.316 (7.96) 2.215 (7.61)

PressurizerSpray

130 0.042 0.165 (3.94) 0.148 (3.55) 0.208 (4.97) 0.191 (4.56)

PressurizerSpray

210A 0.081 0.508 (6.27) 0.425 (5.25) 0.601 (7.43) 0.519 (6.41)

Note: Numbers in parentheses are overall increase factors relative to no environmental effects

Table 5. Fatigue Results - Core Flood Location 5 (Case 3)

State A State B KeSalt

(ksi) Cycles Nallowed CUFair F*en CUFen

Trans. 1ACF up+OBE Trans. 1BCF+OBE 1 67.05 2 8412 0.00024 3.533 0.00085Trans. 1ACF up+OBE Trans. 1BCF 1 63.58 1 10330 0.00010 3.708 0.00037

Trans. 1ACF up Trans. 1BCF 1 60.12 357 13647 0.02616 3.017 0.07892Group T up Trans. 1ACF down 1 46.86 211 47092 0.00448 1.698 0.00761Trans. 1A P=3192 Trans. 1ACF down 1 45.61 1 54293 0.00002 1.698 0.00003Group S up Trans. 1ACF down 1 44.21 148 64392 0.00230 1.698 0.00391

Total = 0.033 Total = 0.092

Table 6. Fatigue Results - HPI/Emergency Location 175A (Case 3)

State A State B Ke Salt (ksi) Cycles Nallowed CUFair F*en CUFen

8HPI up+OBE 8HPI dn + OBE 1 129.25 2 785 0.00255 3.021 0.007708HPI up+OBE Trans. 8HPI dn. 1 126.66 1 833 0.00120 3.048 0.00366

Trans. 8HPI up Trans. 8HPI dn. 1 124.28 67 881 0.07605 2.909 0.22122Trans. 1A, P=3192 Trans. 9 dn. 1 118.73 1 1008 0.00099 3.267 0.00324Trans. 1A, P=2567 Trans. 9 dn. 1 117.30 14 1055 0.01327 3.156 0.04188Trans. 1A, P=2252 Trans. 9 dn. 1 116.57 25 1080 0.02315 3.098 0.07172Trans. 1A, P=2252 Trans. 16 dn. 1 102.04 1 1769 0.00057 3.367 0.00192Trans. 16 up Trans. 22A dn. 1 100.81 1 1851 0.00054 3.150 0.00170Trans. 1A, P=2252 Trans. 22A dn. 1 97.26 39 2115 0.01844 3.199 0.05900Trans. 9 up Trans. 12 1 60.81 5 12894 0.00039 10.232 0.00399

Trans. 9 up Trans. 1B 1 58.41 35 15747 0.00222 10.232 0.02272Trans. 22A up Trans. 1B 1 48.04 40 41638 0.00096 10.358 0.00994Trans. 1A, P=2252 Trans. 1B 1 46.88 280 46994 0.00596 10.206 0.06083No Stratification Stratification A 1 27.5 22302 - 0.03769 10.232 0.38565No Stratification Stratification B 1 24.01 14868 - 0.00887 1 0.00887

Total = 0.193 Total = 0.904

30-15

Table 7. Fatigue Results - HPI/Makeup Location 116 (Case 3)


1A, P=3192+OBE 20D up+OBE 2.11 229.11 1 147 0.00680 9.722 0.06611

1A, P=2567+OBE 20D up+OBE 1.96 209.91 1 189 0.00529 9.708 0.05136

1A, P=2567+OBE Trans. 20D up 1.73 179.54 1 299 0.00334 10.022 0.03347

Trans. 1A Trans. 20D up 1.51 150.99 7 498 0.01406 10.006 0.14069

Trans. 1A Trans. 20D dn. 1.27 133.47 5 714 0.00700 5.461 0.03823

Trans. 1A Trans. 20D dn. 1.19 124.67 5 872 0.00573 5.361 0.03072

Trans. 1A Trans. 16 down 1 62.01 1 11750 0.00009 1.698 0.00015

Trans. 1A Trans. 9 dn. 1 61.37 40 12391 0.00323 1.698 0.00549

Trans. 1A Trans. 8HPI dn. 1 59.08 70 15061 0.00465 1.698 0.00790

Trans. 1A Trans. 22A dn. 1 56.74 40 18518 0.00216 1.698 0.00367

Trans. 1A Trans. 9 up 1 56.01 40 19782 0.00202 1.698 0.00343

Trans. 16 up Trans. 1A 1 55.15 1 21417 0.00005 1.698 0.00009

Trans. 1A Trans. 19A dn. 1 54.45 148 22862 0.00647 1.698 0.01099

Trans. 8HPI up Trans. 1B dn., Zero 1 34.89 70 265321 0.00026 1.698 0.00044

Group T up Trans. 1B dn., Zero 1 29.49 170 756245 0.00022 1.698 0.00037

Trans. 19Bup Trans. 1B dn., Zero 1 28.62 120 911810 0.00013 1.698 0.00022

Trans. 1B up Trans. 19A (dn.) 1 26.59 630 1284179 0.00049 1 0.00049

No Stratification Stratification A 1 52.37 3186 27895 0.11421 10.232 1.16863

No Stratification Stratification B 1 21.37 2124 3248937 0.00065 1 0.00065

Total = 0.177 Total = 1.563

Table 8. Fatigue Results - HPI/Emergency Location 160 (Case 3)


8HPI dn. + OBE 8HPI up+OBE 2.428 241.1 2 123.42 0.016204 3.123 0.05061

Trans. 8HPI up 8HPI dn. + OBE 2.301 224.4 1 149.32 0.006697 3.062 0.02050

Trans. 8HPI up Trans. 9 down 2.238 212 40 173.63 0.230374 3.236 0.74543

Trans. 8HPI dn. Trans. 8HPI up 2.181 209 27 180.32 0.149731 3.088 0.46236

Trans. 8HPI dn. Trans. 16 up 2.151 202 1 197.38 0.005066 3.095 0.01568

Trans. 8HPI dn. Trans. 9 up 2.046 182.9 39 265.28 0.147012 3.188 0.46874

Trans. 9 up Trans. 16 down 1.762 141.8 1 572.85 0.001746 3.580 0.00625

Trans. 1A, P=3192 Trans. 22A down 1 94.2 1 2232.57 0.000448 2.544 0.00114



Trans. 1B, P = 0 Trans. 22A up 1 52.1 40 27532 0.001453 10.232 0.01487

Trans. 1A, P = 2252 Trans. 1B, P=0 1 42.8 320 76928 0.00416 10.194 0.04241

No Stratification Stratification A 1 >27.5 28674 - 0.05606 10.232 0.57368

No Stratification Stratification B 1 <27.5 8496 - 0.00444 1 0.00444

No Stratification Stratification C 1 23.42 1062 - 0.00048 1 0.00048

No Stratification Stratification D 1 24.15 6372 - 0.00330 1 0.00330

Total = 0.640 Total = 2.443

30-16

Table 9. Fatigue Results - Pressurizer Spray Location 210A (Case 3)


Trans. 1BS1 up+OBE Trans. 8AS1+OBE 3.16 186.472 2 268 0.00746 6.553 0.04889

Trans. 1BS4 up Trans. 8AS1+OBE 2.46 126.223 1 841 0.00119 7.392 0.00880

Trans. 1BS2 up Trans. 8AS1 1.94 88.1531 77 3046 0.02528 7.778 0.19664

Stratification A Trans. 1BS2 up 1.6 70.4849 13 6989 0.00186 5.918 0.01101

Stratification A Trans. 1BS4 up 1.6 70.4849 267 6989 0.03820 7.244 0.27674

Stratification A Trans. 1AS1 1.43 60.7184 80 12991 0.00616 9.474 0.05836

Trans. 1B Trans. 1BS4 down 1 26.7159 270 1432151 0.00019 1 0.00019

Trans. 1A Trans. 1AS2 1 26.5965 199 1475441 0.00013 1 0.00013

Stratification A Group S1 down 1 22.6826 1492 4248321 0.00035 1 0.00035

Stratification B Group S1 down 1 17.8494 1800 2.1E+07 0.00009 1 0.00009

Total = 0.081 Total = 0.601

30-17

Figure 1. Transient Pair Strain Range

DESIGN BASIS ENVIRONMENTALFATIGUE EVALUATION AT OCONEE

Authors:Arthur F. Deardorff* and William F. Weitze


Stan T. Rosinski EPRI

Tim BrownDuke Energy Corporation

International Conference on Fatigue of Reactor ComponentsNapa, California USA

July 31 - August 2, 2000

*Presenting Author [email protected]

30-18

PROJECT OBJECTIVES

• Perform a Modern Piping Stress Analysis

• Save All Transient Time Histories

• Develop Approach for Application of Environmental

Fatigue Rules

• Assess Environmental Effects Using Design Transients

30-19

SCOPE OF ANALYSIS

• Performed ASME Section III (NB-3600) Class 1 Analysis ofAttached Piping♦ core flood

♦ pressurizer spray

♦ high pressure injection/makeup

♦ decay heat removal

♦ letdown

♦ loop drains

• Last 3 Exempt From Cyclic Evaluation (NB-3630(d)(2)

• Remaining Affected by Significant Thermal Transients

30-20

EQUATIONS FOR ENVIRONMENTAL EFFECTS

• Developed Environmental Factor (Fen) for Each Load SetPair From Current Fatigue Analysis

• Based On NUREG/CR-5704 Laboratory Tests for StainlessSteel

where ηηηη* = strain rate function

T1* = temperature function 1

T2* = temperature function 2

O* = dissolved oxygen function

(((( ))))(((( ))))*O*T*T*935.0expF 21en −−−−ηηηη++++====

30-21

Fen (for DO < 0.05 ppm)

Dissolved Oxygen > 0.05 ppm

0

1

2

3

4

5

6

7

8

9

0 100 200 300 400 500 600 700

Temperature, F

Fen

Dissolved Oxygen < 0.05 ppm

0

2

4

6

8

10

12

14

16

18

0 100 200 300 400 500 600 700

Temperature, F

Fen

30-22

ASSESSMENT OF REAL TRANSIENTS

• Used Effective Environmental Factors

where: εεεε = strain, relative to most compressive

state of load set pair

εεεεmax = maximum relative strain

εεεεth = threshold relative strain

εεεεmax - εεεεmin = 2Salt/E

• Adjusted to Account for Code Factor

εεεε∫∫∫∫ εεεε−−−−εεεε====′′′′ εεεε

εεεε dF

F max

th

thmax

enen

enF′′′′

SSfor5.1/F*F enen ′′′′====

30-23

ASSESSMENT OF REAL TRANSIENTS

30-24

INFORMATION AVAILABLE FROMPIPING ANALYSIS

• Equation 11 of NB-3650

• Equation 14 of NB-3650

• Ke > 1.0 if Equation 10 of NB-3650 > 3Sm

213

bbaaab33i0

2200

11p

TE1

1TEK

)1(2

1

TTECKMI2

DCK

t2

PDCKS

∆∆∆∆αααανννν−−−−

++++∆∆∆∆αααανννν−−−−

++++

αααα−−−−αααα++++++++====

2

SKS pe

alt ====

30-25

DETERMINATION OF TRANSIENTPARAMETER TIME HISTORIES

• Thermal Stress Approximation

• Strain Rate

• Estimate of Material Inside Surface Temperature

(((( ))))

αααα−−−−αααα±±±±

νννν−−−−∆∆∆∆αααα−−−−

νννν−−−−∆∆∆∆αααα==== bbaaab33

213eth TTECK

1

TE

)1)(2(

TEKKS

dt

Sd

E

1

dt

d th====εεεε

21ba

surf T2T

2TT

T ∆∆∆∆++++∆∆∆∆−−−−++++====

30-26

DETERMINATION OF STRAIN RATES

• ASME Code Analysis Requires that Usage Factors beCalculated Based on Extreme Stress States

♦ tensile state may/may not follow compressive state

♦ may be years between two events

• So, Must Synthesize Load Set Pair Transient for Strain RateAnalysis

30-27

DETERMINATION OF STRAIN RATES

• Thus, can perform Fen integration with respect to strain εεεε (ifwe know temperature and oxygen content)

30-28

EVALUATION OF DO ANDTEMPERATURE

• Laboratory Testing Based on "Uniform Conditions"

• Conditions Different Between Real Load Set Pairs

• What Do We Put in Our Equation for Fen for T and DO?

30-29

CHOICE OF DISSOLVEDOXYGEN FOR Fen

• For PWR's it is Generally Easy

♦ most severe transients have DO << .05 ppm

• But, What if Transient Pair Includes One

DO > 0.05 ppm and Other with DO < 0.05 ??

♦ may be important for PWR surge line evaluations sincereactors may be just coming out of outages

30-30

CHOICE OF TEMPERATUREFOR Fen

• Many Choices Exist for Choosing the T to Use in Fen

Equation Integration♦ use temperature before each transient?

♦ use temperature instantaneously?

♦ use same or different for each state?

♦ use most conservative temperature?

♦ use material or surface or fluid temperature?

• Fen Varies Considerably With Temperature

30-31

EVALUATION OF TEMPERATUREASSUMPTIONS

Case Non-Transient Loads Thermal Transients

1maximum Fen overtemperature range

surface temperature ateach time point

2average Fen over

temperature rangesurface temperature at

each time point

3maximum Fen overtemperature range


4average Fen over

temperature rangemaximum Fen over input

temperature range

• Several Cases Evaluated

• Differences as High of a Factor of 2

30-32

SUMMARY OF RESULTS

• Core Flood: Effective Fen* ≈≈≈≈ 2.75♦ most of transients < 200°C

• HPI Emergency Inj.: Effective Fen* ≈≈≈≈ 3.0 to 4.7♦ controlling transients had high strain rates

• HPI Makeup: Effective Fen* ≈≈≈≈ 7.6 to 8.8♦ major contribution due to stratification transients

(from plant data)

• PZR Spray Piping: Effective Fen* ≈≈≈≈ 3.5 - 7.5♦ included both rapid and stratification transients

30-33

COMMENTARY ON PIPINGENVIRONMENTAL EFFECTS

• Piping Analysis is by Itself a Conservative "Design-by-Rule" Approach, Overstating Actual Fatigue Results -

♦ P + M + ∆∆∆∆T1 + ∆∆∆∆T2 + Ta + Tb♦ Ke

♦ cycle sequence not considered

• Laboratory Data is not Prototypical of Long-Term PlantOperation

• Current Environmental "Equations" Add Complexity♦ very difficult to deal with without adding considerable

conservatism

♦ may not be realistic

30-34

31-1

31 AN UPDATED METHOD TO EVALUATE REACTORWATER EFFECTS ON FATIGUE LIFE FOR CARBONAND LOW ALLOY STEELS

Makoto HIGUCHIIshikawajima-Harima Heavy Industries Co., Ltd.

Isogo-ku, Yokohama, 2358501 JAPAN

32-1

32 ENVIRONMENTAL FATIGUE, CRACK GROWTH RATESIN TITANIUM STABILIZED STAINLESS STEEL

Jussi Solin, Päivi Karjalainen-RoikonenVTT Manufacturing Technology

CODES AND STANDARDS

33-1

33 STRESS INTENSIFICATION FACTORS

E. A. WaisWais and Associates, Inc.

R. CarterEPRI

34-1

34 FATIGUE EVALUATIONS USING ASME SECTION XINON-MANDATORY APPENDIX L

S. R. Gosselin, P.E.Pacific Northwest National Laboratory

902 Battelle Blvd.Richland, WA 99352

34-3

FATIGUE EVALUATIONS USING ASME SECTION XINON-MANDATORY APPENDIX L



34-4

FAITIGUE EVALUSTAIONS USING ASME SECTION XINON-MANDATORY APPENDIX L



ABSTRACT

This paper describes an EPRI and USNRC collaborative project to evaluate the current ASMESection XI Appendix L fatigue evaluation procedures and propose changes to address severaldeficiencies. As components continue to operate beyond fatigue usage factors that exceeding 1.0,their failure frequencies are expected to increase unless augmented inspection strategies areimplemented. A probabilistic approach to address NDE performance and inspection frequencyissues related to damage tolerance assessments and component fatigue life extension will bedescribed. Probabilistic fracture mechanics calculations are presented to demonstrate that anaugmented level of inservice inspection can ensure that the failure probability of fatigue criticalcomponents will not increase if operation is continued beyond usage factors permitted by theASME Section III design code.

ASME SECTION III FATIGUE DESIGN

The discussion in this section was drawn from a presentation by Dr. William E. Cooper,Teledyne Engineering Services to the PVRC Workshop on Environmental Effects on FatiguePerformance in January 1992 (1). The essential points of his paper are summarized below:

The basic intent of the ASME Section III Code has not changed since before its inception - toestablish rules that address design requirements for new construction while providing reasonableassurance of reliable operation. These requirements were directed mainly at the Manufacturer,while the Owner was assigned the important responsibility of defining the operational conditions(i.e. anticipated service duty), which the Manufacturer must consider. These service conditionsare defined for the Manufacturer in the Design Specification.

Prior to World War II, the design of pressure retaining components was primarily based onselecting a thickness that satisfied pressure stress criteria of one-fifth ultimate strength. Duringthe war, this nominal factor of safety was reduced from five to four as an emergency measure. Asa result of the apparent success of this change, the Codes were revised to adopt the lower factor

34-5

of safety and efforts were initiated to determine if further reductions were practical. The SpecialCommittee to Review Code Stress Basis (SCRCSB) was tasked to determine what would berequired to reduce the nominal factor of safety from four to three. The results of the SCRCSBwork, Design by Analysis, were published in Section III in 1963 and Division 2 of Section VIII in1968.

The Design by Analysis procedure included a number of related considerations; however, thepurpose for adding fatigue as one of the failure modes was to assure that the reduction of thenominal factor of safety from four to three would not result in a decrease in reliability if the vesselwere subjected to cyclic operating conditions. The fatigue design procedures were intended toprovide confidence that the component could be placed in service safely, not necessarily toprovide a valid measure of actual component service life.

The cyclic loading conditions defined in the Owner's Design Specification were not intended torepresent a commitment on how the vessel was to be operated. Neither did it mean that theOwner was intended to be completely oblivious to the manner in which the vessel was operated,only that the design transient definitions should provide useful information. For example, if anOwner were able to show that the Design Specification included a cyclic event more severe thanactual operating loads, then it could be assumed that the vessel would not be subjected to anunevaluated condition.

ASME SECTION XI PHILOSOPHY

This section presents a philosophy, originally endorsed by the ASME Section XI ExecutiveCommittee, intended to guide future Code activities regarding fatigue and its impact oncomponent serviceability (2).

Once a nuclear plant begins commercial operation, subsequent components integrity assessmentsare performed to determine the fitness for service. The results of the fitness-for-service evaluationare then be used to assess component operability and establish a "fatigue operating basis".

ASME Section XI has made significant strides in developing acceptability criteria for majorcomponents based on improved understanding of material performance in actual operatingenvironments; for example, irradiation damage in vessels and stress-corrosion cracking in BWRpiping. The Section XI approach for protecting against fatigue damage will utilize a similarphilosophy. The key elements include:

1. Evaluation procedures and criteria that ensure adequate protection from pressure boundaryfailures at the end of a prescribed operating period.

2. Evaluation procedures that combine the results of analyses, inspections, and serviceexperience. A cumulative fatigue usage factor calculated in accordance with the rules inASME Section III may not, in itself, represent a limit on component serviceability.

34-6

3. Alternative procedures designed to complement, not replace, the cumulative fatigue usagefactor approach contained in ASME Section III. This would be accomplished by providingguidelines to address service conditions and loading mechanisms not anticipated in theOwner's Design Specification or accounted for in the original Design Report.

4. Procedures that include a fracture mechanics based flaw tolerance approach and consider bothcrack initiation and propagation.

In November 1991, the ASME Section XI Task Group on Operating Plant Fatigue Assessmentwas formed to study fatigue evaluation methods, develop procedural guidelines, and establishacceptance criteria that operating plants could use to assess component serviceability. The resultsof this work were included in ASME Section XI Non-Mandatory Appendix L (3). Of significancewas the introduction of a damage tolerance procedure to assess the serviceability of fatiguesensitive components. In this procedure the component is first inspected to verify the absence ofany relevant indication or flaw. Subsequent inspections are then based on deterministic fracturemechanics calculations and the time required for an assumed flaw to grow to unacceptable size.

ASME SECTION XI APPENDIX L

The fatigue assessment approach in Appendix L is based on satisfying one of two principles: 1)The component is designed and operated such that fatigue damage will be limited to an amountless than that required for crack initiation, or 2) the component is designed and operated in amanner that is “damage tolerant”. In the latter principle, “damage tolerant” means that it must beable to tolerate fatigue accumulation and crack growth without reducing the structural integritybelow acceptable limits.

Figure 1 (4) shows how these principles can interact when evaluating plant components suspectedto have a fatigue concern. In the majority of cases, operating plant fatigue concerns areassociated with the discovery of loading conditions not previously accounted for in the DesignReport (5). Faced with this problem, utilities will usually attempt to establish componentoperability by "requalifying" the component design. For those plants designed to ASME SectionIII, the analyst simply adds the new loading condition(s) to the other design transients assumed inthe Design Report and demonstrates that the revised design basis cumulative usage factor (CUF)remains below 1.0.1

Depending on the significance of the loading condition in question, requalifying the as-builtcomponent to the original design and licensing requirements may be extremely difficult and insome cases may not be possible. Examples of this are can be seen in the industry’s efforts to

1 For the many operating plants designed to earlier piping design codes (i.e., ANSI B31.7, ANSI B31.1, etc.)

without original Class 1 fatigue analyses, this approach to component fatigue qualification becomes significantly

more difficult and expensive.

34-7

address operating plant fatigue concerns in BWR and PWR feedwater nozzles, PWR surge lines,PWR spray lines, etc. Consequently, it may be necessary to deviate from the traditional “designqualification” analysis to confirm component serviceability. In these cases, a flaw toleranceapproach is used. This analysis examines the tendency for flaw growth of a postulate fatiguecrack assumed to be present in the component. The subsequent inspection schedule would thenbe based on the time required for this hypothetical flaw to grow to an unacceptable size.

Appendix L Flaw Tolerance Procedures

The flaw tolerance approach is well suited for operating plants because fatigue concerns aretypically limited to a few specific plant component locations. Also, with this approach futureperformance capabilities can be established independent of past loading histories. The procedurefor performing fatigue flaw tolerance evaluations is contained in Article L-3000 of Appendix L(3). The procedural steps include:

1. Verify the absence of any relevant indication exceeding the applicable flaw acceptancestandards in Table IWB-3410.

2. Postulate the existence of a hypothetical flaw according to the flaw model in L-3200.

3. Determine the stresses at the location of the postulated flaw.

4. Determine the postulated end of evaluation period flaw size and critical flaw sizes by usingexisting ASME Section XI the analytical procedures contained in Appendix A, Appendix C,or Appendix H as applicable.

5. Apply the appropriate flaw acceptance criteria contained in IWB-3600.

6. Continue operation and inspect the component at a frequency specified in L-3420.

Appendix L Flaw Model

The basis for the Appendix L flaw model is discussed in EPRI TR-104691, Operating NuclearPower Plant Fatigue Assessments (6). Since the assumed absence of any actual flaws is based oninspection results, the flaw model (size and shape) must be consistent with the inspectioncapabilities. In other words, the hypothetical flaw assumed in the flaw tolerance assessmentshould represent the largest flaw expected to be missed during the initial inspection. The flawmodel used in Appendix L is described in Table 1 (ASME Section XI 1995).

The postulated flaw sizes were based primarily on judgment and limited quantitative informationof ultrasonic inspection (UT) capabilities. At that time, limited performance demonstration datawere available for thick walled reactor pressure vessels and were considered most applicable whenthe component section thickness was greater than 4 inches. Since the industry’s Appendix VIIIperformance demonstration initiative (PDI) for piping was just starting, similar data was notavailable for thinner components. This was especially true for components whose section

34-8

thickness was below 1 inch. Consequently, conservative estimates were made regarding flawdetection capabilities for these components.

The Appendix L flaw model was subsequently reviewed by the ASME Section XI WorkingGroup on Procedure Qualification and Volumetric Inspection. In their opinion, the inner surfacefatigue cracks described in Table 1 could be detected in service for most materials that comprisethe reactor coolant system. However, the detection capability may not be appropriate forcomponents containing dissimilar (bimetallic or trimetallic welds) and statically or centrifugallycast austenitic piping components, due to their challenging metallurgical characteristics (6).

Table 1 – Appendix L Flaw Model

Appendix L Augmented Inspections

In Appendix L, flaw tolerance is measured in terms of the allowable operating period, P. Asdefined in the Appendix, P represents the time required for the hypothetical flaw to grow to adepth equal to the maximum allowable flaw size specified in ASME Section XI, IWB-3600.Successive inspections are then implemented at the location of concern according to the scheduleshown in Table 2 (3). The time intervals between inspections are designed to ensure that thecomponent will be re-examined before a growing hypothetical flaw that exceeds half the sizeallowed by ASME Section XI.

Table 2 – Appendix L Successive Inspection Schedule

Section Thickness, t (in.) Reference Flaw Depth, a/t (%)

0.5 30.01 20.02 15.03 11.7

4-12 10.0Notes:1. 6:1 minimum aspect ratio2. For t<0.5 inches, the postulated flaw depth, a = 0.15 inches3. For t>12 inches, the postulated flaw depth, a = 1.2 inches

Allowable OperatingPeriod, P

Successive InspectionFrequency

> 20 yearsEnd of each 10 year ISIInspection Interval

< 20 yearsEnd of P/2 operating years oreach operating cycle,whichever is greater

34-9

Appendix L Applications

Most recent applications of Appendix L have been associated with license renewal applications.The license renewal rule requires that the licensee perpetuate the current design basis to the end ofthe renewal period (i.e., 60 years). With regard to fatigue, this means that the calculated CUFusing ASME methods must be maintained less than 1.0. Emphasis is placed on the more fatiguesensitive areas in the plant that are selected based on anticipated loading conditions as well asindustry service experience. These fatigue sensitive locations are not necessarily limited to Class1 components and may include Class 2 or ANSI B31.1 piping components whose original designqualification was not based on a CUF analysis.

These industry applications have raised concerns regarding the effectiveness of the current flawtolerance procedures. In a report to the ASME Section XI Task Group on Operating PlantFatigue Assessment (7), BG&E indicated that, in license renewal space, the potential for large-scale replacements is real. Lines such as the surge line, main spray, auxiliary spray, and CVCShave the potential to exceed a CUF of 1.0 in a multitude of locations prior to the end of theextended design life. In many cases the Appendix L flaw tolerance procedures do not provide asufficiently long inspection interval to allow long-term plant operation.2 As such, the only viableoptions to deal with this would be to recalculate the CUF using more refined inputs, or at worst,repair/replace the affected items.

The general consensus within ASME Section XI Working Group Operating Plant Criteria andTask Group Operating plant Fatigue Assessments is that a stronger technical basis is needed forboth the flaw growth models and the rules used to establish inspection frequencies. Mostnoteworthy are the following issues:

1. Limited Effectiveness of Appendix L - Insights into flaw detection capabilities gained fromindustry performance demonstration initiatives (PDI) suggest that, in some cases, the NDEcapabilities of field teams performing piping inspections are better than assumed in AppendixL. In these cases, the flaw size assumptions in Appendix L may be overly conservative whichrender the flaw tolerance alternative ineffectual. This is especially true for components whosesection thickness is below 1 inch.

2. Flaw Model Applicability - The Appendix L flaw tolerance procedure flaw size assumptionsare common to both carbon steel and wrought austenitic stainless steel materials. Theassumed flaw depths are presented as a function of section thickness only. No distinction ismade with regard to differences in flaw detection and sizing capability for carbon and stainless

2 At BG&E, augmented inspection frequencies were found to be unacceptably short, less than one operating cycle,

when environmentally assisted fatigue crack growth curves were applied indiscriminately throughout the

evaluation period for stainless steel locations in the CVCS and pressurizer surge line (8).

34-10

steel materials or other NDE performance factors such as inspection access, inspectionprocedure, or inspector qualification.

3. Technical Basis – An improved technical basis is needed for both the flaw growth modelsand the rules used to establish inspection frequencies. The flaw model (i.e., size and shape)used in Appendix L should be closely tied to performance demonstration results and theinspection frequencies need to ensure that the likelihood of failure will not increase beyondthat provided in the original design code of record.

4. Probabilistic Fracture Mechanics Approach - There is a need for probabilistic fracturemechanics evaluation methods and acceptance criteria that can be used to evaluate componentserviceability and define augmented inspection frequencies.

PROBABILISTIC ASSESSMENT OF APPENDIX L INSPECTION STRATIGIES

In this section we show how a probabilistic approach can be applied to determine inspectionfrequencies that account for demonstrated NDE performance and to ensure that reliable pipingperformance is maintained throughout the component’s original or extended operating life (9). Inthe example described below, we assume that the weld location is subject to thermal fatigue. Theexample considers a stainless steel pipe (29 inch outside diameter by 2.5 inch wall) which isloaded at 5000 cycles per year such to give a CUF=1.0 after 20 years of operation given a weldroot stress concentration factor of 3.0. This corresponds to a nominal alternating stress of 27.3ksi and a peak alternating stress, at the weld root, of 81.9 ksi. The inspection frequency necessaryto maintain the component’s failure probability at or below that associated with the fatigue limitspecified in the ASME Section III design code (e.g., cumulative usage factor (CUF) must be lessthan 1.0) is then determined. The inspection frequencies are compared to those obtained from thedeterministic procedures in Appendix L.

Probabilistic Calculations

Probabilistic fracture mechanics calculations are presented to demonstrate that an augmented levelof inservice inspection can ensure that failure rates of fatigue critical components should notsignificantly increase as operation is continued beyond usage factors permitted by the design code.Uncertainties in flaw growth rates and in flaw detection were addressed by application of theprobabilistic fracture mechanics code pc-PRAISE (10). Suitable inspection frequencies wereestablished for a given flaw detection capability (probability of detection or POD curve) byadopting a goal for an acceptable piping failure probability (i.e. probability of through-wall crackper weld per year). Continued operation for calculated CUFs exceeding unity was considered tobe acceptable only if additional inspections are performed. The additional inspections are requiredto maintain calculated failure rates at levels less than or equal to calculated failure rates before theusage factors became 1.0.

34-11

The pc-PRAISE model assumed semi-elliptical surface flaws with aspect ratios of 12 and 20, anda Paris law for fatigue crack growth having a mean rate corresponding to constants of C = 9.14E-12 and m = 4. A simplified treatment of flaw initiation was assumed. At time = 0.0, very smallinner surface cracks were assumed to be present, with depths uniformly distributed between 0.005to 0.010 inch.

The alternative inspection frequencies were limited to the case of no inspections and inspectionsevery 2 or 4 years, with the inspection program being introduced after 20 years of operation. Thereliability for the ultrasonic NDE was described by the error function type curves used by the pc-PRAISE code to describe flaw detection. Two bounding curves were assumed for purposes ofthe demonstration calculations. The less effective NDE assumed a threshold detection capability(50% POD) for a 0.10t flaw (a* = 0.25 inch) whereas the more effective NDE had a 50% POD fora 0.05t flaw (0.125 inch). In each case the POD curve provided significantly better detectioncapabilities for flaws of greater depths, such that flaw depths 0.25 and 0.50 inch respectively, orabout twice the threshold size, could be detected with a probability of better than 90%.

Figure 2 shows the predicted cumulative probability of leak (through-wall crack) as a function ofthe operating time (0 to 40 years). At 20 years (when the calculated CUF becomes 1.0) thecumulative leak probability is about 1.0E-02, or one chance in one hundred that the weld wouldfail. If no inspections are performed, the cumulative failure probability curve continues to rise andwith an increasing failure rate. All of the alternative inspection scenarios (combinations of PODand inspection frequency) reduce the calculated failure probabilities, but some scenarios reducethe failure probability much more than others.

Figure 3 shows the calculated failure rates in terms of failures per weld per year of operation. Themost effective inspection (a* = 0.125-inch) reduces the failure rate by about an order of magnitudecompared to the alternative of no inspection. In this case the failure rates during the second 20-years of operation are actually substantially lower than the corresponding rates during the first 20-years of operation. Some of the other less rigorous inspections of Figure 3 are also sufficientlyeffective to maintain the calculate failure rates at or below the rate that exists at the time (20-years) when the CUF attains the limiting value of unity. For example, Figure 3 indicates that anAppendix L inspection with a 4-year frequency and a* = 0.125-inch would meet the probabilisticcriteria. The alternative of a 2-year frequency with a* = 0.25-inch would also meet the criteria.The most effective inspection (a* = 0.125 inch) reduces the failure rate by about an order ofmagnitude compared to the alternative of no inspection. In this case the failure rates during thesecond 20 years of operation are actually substantially lower than the corresponding rates duringthe first 20 years of operation. Therefore, in this extreme case where thermal fatigue loading issignificantly high, a 2 to 4 year inspection frequency will maintain the component’s reliability atdesign basis levels.

34-12

Deterministic Appendix L Calculations

Having shown with the probabilistic method that acceptable inspection frequencies range from 2to 4 years, the deterministic criteria of Appendix L were then applied. The flaw model was againa semi-elliptical surface flaw with aspect ratios of 6, 12 and 20. The postulated flaws had initialdepths of 0.326 inches and 0.250 inches. Fatigue crack growth was calculated using the ASMESection XI Appendix C curve for stainless steel at 550oF with no environmental correction. TheSection XI end of inspection interval allowable flaw was 0.75t (1.875 inch).

The results of fatigue crack growth calculations for the four cases considered are shown in Table3. In the first case, the existing Appendix L flaw model (aspect ratio of 6.0 and initial flaw depthof 0.326 inches) was applied. The Appendix L flaw depth was also assumed in cases 2 and 3 and

the aspect ratios were increased to 12 and 20 respectively. In the final case, the Appendix Lassumed flaw depth was reduced from 0.326 inches to 0.250 inches in order to be consistent withobserved detection capabilities (11) and those anticipated in pc-PRAISE (10 and 12). Theseresults were the basis for the Appendix L inspection frequencies of Table 3, which range from 2 to4 years and are in excellent agreement with the conclusions of the probabilistic methodology.

EPRI-NRC REASEARCH ACTIVITES

This section describes ongoing research activities sponsored by EPRI and the NRC to address theneeds and concerns expressed above (13). The objectives of this project are:

1. Improve the understanding of demonstrated NDE performance capabilities.

2. Upgrade the POD modeling capabilities in the pc-PRAISE code.

3. Understand how NDE performance factors impact piping reliability and associated augmentedinspection strategies.

Table 3 – Appendix L Inspection Requirements

Assumed Reference Initial Flaw Size Allowable Operating Period Inspection

Case a/t a AspectRatio

Cycles P P/2

1 13% 0.326” 6 33,207 6.65 yrs 3.32 yrs2 13% 0.326” 12 23,060 4.61 yrs 2.30 yrs3 13% 0.326” 20 17,572 3.51 yrs 1.75 yrs4 10% 0.250” 6 40,500 8.10 yrs 4.05 yrs

34-13

4. Evaluate the current Appendix L flaw tolerance procedure and its ability to define augmentedinspection strategies that can offset the increases in the failure frequencies that are expected tooccur as components continue to operate after calculated usage factors exceed 1.0.

5. Develop procedures and guidelines for a probabilistic fracture mechanics approach tocomponent fatigue life extension.

PDI Performance Data Evaluation

In this task experts from the EPRI NDE Center and Pacific Northwest National Laboratory(PNNL) will be consulted to estimate probability of detection (POD) curves for fatigue cracksbased on industry performance demonstration data. A matrix of POD curves will be developedfor various NDE performance categories. Each performance category will represent acombination of performance factors such as: material type (stainless steel or carbon steel), access(double-sided or single-sided), section thickness, procedure (automatic or manual), and inspectorqualification.

pc-PRAISE POD Model Evaluation and Upgrade

This task will evaluate the current pc-PRAISE POD model against industry performance data. Amenu of performance category POD curves and/or closed form functions will be prepared andincorporated into the pc-PRAISE probabilistic fracture mechanics code.

Augmented Inspection Strategy Sensitivity Study

In this task a matrix of probabilistic fracture mechanics (PFM) calculations will be performed toidentify augmented inspection strategies that would result in a failure frequency (failures per year)at or below the failure frequency associated with the design basis fatigue limit (e.g., cumulativeusage factor of 1.0). The calculations will consider high cycle and low cycle fatigue loading, airand reactor water environments, and high stress through-wall gradients versus uniform through-wall stress distributions. PFM models recently developed at PNNL will be adapted to address theissues associated with fatigue crack initiation and crack growth in combination with the benefitsof inservice inspections (Simonen et al. 1999).

Appendix L Flaw Tolerance Flaw Procedure Assessment

This task will assess whether the reductions in failure frequencies due to the Appendix L programof inspections (for locations with usage factors greater than 1.0) can offset the increases in thefailure frequencies that are expected to occur as components continue to operate after calculatedusage factors exceed 1.0. Recommended changes to Appendix L will be identified which willensure the following:

1. Flaw size assumptions consider various NDE performance factors and are consistent withperformance demonstration results.

34-14

2. Augmented inspection frequencies will maintain component reliability at a level consistentwith the component’s original design basis.

Probabilistic Alternative Approach

This task will develop a probabilistic alternative approach that may be used to address NDEperformance and inspection frequency issues related to damage tolerance assessments andcomponent fatigue life extension.

Industry Benefits

This program will establish inspection frequencies that are consistent with the service conditionsand the demonstrated performance levels of the NDE methods being applied. Case studies haveshown that when inspections are performed at appropriate frequencies with reliable NDEmethods, justification for continued operation of components with calculated fatigue usage factorsthat exceed original design limits can be made. This can provide a direct benefit to those plantsseeking license renewal.

In many cases it is expected that using performance demonstration data to optimize postulatedAppendix L flaw size assumptions will eliminate much of the unnecessary conservatism in theseprocedures and reduce the resulting augmented inspection frequencies. Also, by expandingAppendix L to consider a probabilistic fitness-for-service approach, special issues or concernsrelated to life extension beyond the original 40 years can also be addressed.

The PDI program has generated a very large set of data on the reliability of NDE as performed onoperating U.S. nuclear vessels and piping. This data is particularly valuable both because of thenumber of data points and because it is based on trials performed by the same inspectionpersonnel, procedures, and UT equipment employed in field inspections and because thecomponents, welds and flaws were intended to be represent realistic field conditions.

There would be great benefits to industry by documenting the collective quantitative trends of thePDI work, namely:

Industry could potentially take credit for the improved ISI to gain relief from present code andregulatory practices (i.e. the Appendix L improvements as described above, reduced inspectionintervals for Section XI and augmented inservice inspection programs, plant life extension, etc.)At present the fracture mechanics analyses, including probabilistic methods, may be based onpessimistic estimates for NDE performance, or no credit can be taken for the impact ofinspections on the structural integrity of components such as the RPV in the context of PTS risk

Industry could justify reductions in testing requirements for performance demonstration efforts bydemonstrating a high statistical confidence in collective NDE detection and sizing capability forthe cohort of inspections teams that are performing field inspections.

34-15

Nevertheless within these restrictions the collective trends of the data could be characterized inconsiderable detail and at a level that could be used to the benefit of the industry that wouldenable structural integrity evaluations to be performed in a much more realistic manner than isnow possible.

Fracture mechanics computer codes can now include the benefits of inservice inspections at givenintervals with specific techniques. Inputs are needed for probability of detection curves and insome cases the expected sizing errors. Such calculations have been performed as part of researchefforts to support applications of risk informed inservice inspections. However, the uncertaintiesin estimated POD curves have been large, such that it has been difficult to take credit for ISI interms of regulatory requirements. Any improvements in ISI reliability have essentially goneunrecognized with respect to use in flaw evaluations, establishing inspection intervals, plantspecific PTS evaluations, etc. The PDI data would permit POD curves and sizing errors to beestablished with a sound technical basis that reflects the current demonstrated NDE capabilities ofthe nuclear industry.

It is further noted that the current Section XI – Appendix VIII tables for the minimum number ofsamples and passing scores for detection, false calls and sizing errors were based on statisticalevaluations of round robin studies to estimate the performance capabilities of the inspectionprocess and did not assume a qualification effort of the type encompassed by the PDI activities.Rather the current bases have assumed that each organization would have an independentperformance demonstration activity and the statistical calculations did not take credit for theadded knowledge and confidence gained from subjecting a large group of organizations to acommon effort that would address the collective reliability of a population of inspections teams.Such an approach could reduce the need for each passing team to inspect as many samples or todemonstrate as high level of successful detection and sizing performance. In other words, havingthe inspection process quantified (personnel, equipment, and procedures) provides data necessaryto check the Appendix VIII assumptions for performance demonstration and to make it possibleto optimize the process to achieve the desired screening (pass/fail threshold) most economically.

CONCLUSIONS

The results for the above example show that inspection frequencies based on Appendix Levaluations are consistent with inspection frequencies derived from a probabilistic assessment. Itis seen that inspections at appropriate frequencies with reliable NDE methods can justifycontinued operation for components with calculated fatigue usage factors that exceed originaldesign limits. It is even possible with an aggressive inspection program to decrease failurefrequencies during the later periods of plant life to the same levels that existed relatively early inlife. It should be noted that the Appendix L approach was primarily developed for location withhigh fatigue for which there is only a perceived potential for cracking. The approach would notbe applied in other cases for which the operating plant has actually experienced degradation (i.e.

34-16

detection of small cracks or leakage). In such cases, the preferred strategy would be one ofrepair/replacement, augmented inspections, and other mitigative actions.

Future work will to expand the scope of the calculations to address other piping sizes, materials,and cyclic stress conditions. The goal will be to establish inspection frequencies that areconsistent with the service conditions and the demonstrated performance levels of the NDEmethods being applied. It is expected that, in many cases, the inspection frequencies required bythe simplified rules of Appendix L are overly conservative. As shown in this paper, usingperformance demonstration data to optimize the Appendix L flaw size could eliminate much ofthis conservatism. In other cases, the calculations may indicate a need for improved NDEmethods and/or for more frequent inspections.

By expanding Appendix L to consider a probabilistic fitness-for-service approach, special issuesor concerns related to life extension beyond the original 40 years can also be addressed. Forexample, the need to address potential fatigue cracking at outside surface locations, for which theNDE procedure will be based on dye penetrant and magnetic particle methods rather than UTmethods. In the end, the analyst could have a choice of 3 alternative fatigue qualificationmethods, each having a sound technical basis that will ensure piping reliability is maintainedthroughout its desired service life.

The flaw tolerance procedure contained in Appendix L is ineffectual, primarily due to veryconservative flaw size assumptions. An improved technical basis is needed for the flaw modelsand rules use to establish inspection frequencies in Appendix L. It is expected that NDEperformance data based on PDI will provide substantial information to develop appropriate flawsize assumptions and POD curves for use in structural integrity evaluations. The resultantproducts from this work will be especially useful for utilities that may have fatigue concerns or areplanning to seek license renewal.

ACKNOWLEDGMENTS

The author wishes to acknowledge Dr. Steven Doctor and Dr. Fred Simonen of Pacific NorthwestNational Laboratory. The insights they provided with regard to the industry performancedemonstration initiative and the development of Section XI Appendix VIII were incorporateddirectly into this paper.

REFERENCES

1. Cooper, W. E., “The Initial Scope and Intent of the Section III Fatigue DesignProcedures”, PVRC Workshop on Environmental Effects on Fatigue Performance,Clearwater Beach, FL, January 20, 1992, p. 1-6.

34-17

2. Gosselin, S. R., “ASME Section XI Philosophy Related to Operating Nuclear Power PlantFatigue Damage Protection”, PVP Volume 313-2, International Pressure Vessel PipingCodes and Standards: Volume 2- Current Perspectives ASME 1995.

3. American Society of Mechanical Engineers, ASME Boiler and Pressure Vessel Code,Section XI, Non-mandatory Appendix L, “Operating Plant Fatigue Assessments”, 1995Edition, New York, July 1995.

4. Gosselin, S. R., A. F. Deardorf, and D. W. Peltola, “Fatigue Assessments in OperatingNuclear Power Plants,” PVP-Vol. 286, pp. 3-18, Changing Priorities of Codes andStandards, American Society of Mechanical Engineers, New York, 1994.

5. EPRI TR-100252, ASME Section XI Task Group on Fatigue in Operating Plants, MetalFatigue in Operating Nuclear Power Plants, April 1992.

6. EPRI TR-104691 Final Report “Operating Nuclear Power Plant Fatigue Assessments”,Palo Alto, CA, April 1995.

7. Connor, J. T., BG&E Report to ASME Section XI Task Group Meeting on OperatingPlant Fatigue Assessments, August 1999.

8. EPRI TR-107515 Final Report, “Evaluation of Thermal Fatigue Effects on SystemsRequiring Aging Management Review for License Renewal for Calvert Cliffs NuclearPower Plant”, Palo Alto, CA, December 1997.

9. Gosselin, S. R. and Simonen, F. A., “A Probabilistic Basis for Damage ToleranceAssessments and Component Fatigue Life Extension”, PVP-Vol. 383, ASME, 1999.

10. Harris, D. O., and Dedhia, D. D., “Theoretical and User’s Manual for pc-PRAISE, AProbabilistic Fracture Mechanics Computer Code for Piping Reliability Analysis,”NUREG/CR-5864, U. S. Nuclear Regulatory Commission, Washington, D.C.

11. Heasler, P.G., and Doctor, S. R., “A comparative Analysis of Round Robin Studies”,Presentation at 1st International Conference on NDE in Relation to Structural Integrity forNuclear Pressurized Components, October 1998, Amsterdam, The Netherlands.

12. Simonen, F. A., and Khaleel, M. A., “A Model for Predicting Vessel Failure ProbabilitiesDue to Fatigue Crack Growth,” PVP-Vol. 304, Fatigue and Fracture Mechanics inPressure Vessels and Piping, pp. 401-416, American Society of Mechanical Engineers,New York, 1995.

13. Carter, R. G., Mayfield, M., Gosselin, S. R., and Simonen, F. A., “An Evaluation of FlawTolerance Procedures Used for Component Fatigue Life Qualification”, 8th InternationalConference on Nuclear Engineering, April 2000.

34-18

14. Simonen, F. A., Phan, H. K., Harris, D. O., Dedhia, D., Kalinousky D. N., and Shaukat, S.K., “Evaluation of Environmental Effects on Fatigue Life of Piping”, Presented at 27th

Water Reactor Safety Information Meeting, Bethesda, MD, October 25-27, 1999.

34-19

Operating PlantFatigue Concern

AreActual Load Cycles

< Design?

Section III FatigueCUF Evaluat ion

CUF<1.0?

Appendix L FlawTolerance Evaluation

Fitness forContinued Service

?

Continue OperationInspect at Normal

Section XI Frequency

Repair or Replace

No

Yes

No

Yes

Yes

No

Continue OperationInspect at Increased

Frequency

Figure 1: Operating Plant Fatigue Assessments

34-20

0.E+00

1.E-02

2.E-02

3.E-02

4.E-02

5.E-02

0 5 10 15 20 25 30 35 40 45

Time, Years

Cu

mu

lativ

e P

rob

abili

ty o

f Lea

k

C:\PRAISE96\APPENL\APL1.XLSDecember 1998

Baseline CaseNo Additional ISI

ISI @ 4 Yra* = 0.25 Inch



Fatigue Usage Factor > 1.0Inspections as per Appendix L



Figure 2: Cumulative Leak Probability

0.0E+00

5.0E-04

1.0E-03

1.5E-03

0 5 10 15 20 25 30 35 40 45 50

Time, Years

Lea

k F

req

uen

cy, L

eaks

per

Wel

d p

er Y

ear

C:\PRAISE96\APPENL\APL1.XLS

Fatigue Usage Factor > 1.0Inspections as per Appendix L

Baseline CaseNo Additional ISI


ISI @ 4 Yra* = 0.125


"Acceptable" Leak Frequency


Figure 3: Leak Frequency

35-1

35 AN UPDATE ON THE CONSIDERATION OF REACTORWATER EFFECTS IN CODE FATIGUE INITIATIONEVALUATIONS FOR PRESSURE VESSELS AND PIPING

Hardayal S. MehtaGE Nuclear EnergySan Jose, California

35-3

AN UPDATE ON THE CONSIDERATION OF REACTOR WATEREFFECTS IN CODE FATIGUE INITIATION EVALUATIONS

FOR PRESSURE VESSELS AND PIPING

Hardayal S. MehtaGE Nuclear EnergySan Jose, California

ABSTRACTReactor water environmental fatigue effects are currently a subject

of active discussion among the industry, the standards making bodiesand the regulatory bodies in U.S. and Japan. The PVRC has recentlyrecommended to the BNCS a procedure that can potentially beimplemented into the ASME Code. The objective of this paper is toprovide details of the procedure and discuss the rationale for some ofthe modifications incorporated into the procedure prior to therecommendation to BNCS. Also, a brief discussion of the currentRegulatory and ASME Code activities in progress on this subject isprovided.

INTRODUCTIONMehta and Gosselin (1995, 1998) provide the details of the

proposed EPRI/GE methodology to account for the reactor waterenvironmental fatigue effects in the ASME Code fatigue evaluations.The methodology essentially consists of applying an environmentalcorrection factor, Fen, that is computed using equations involvingenvironmental parameters, to each load set pair that does not meet thethreshold criteria. Argonne National Laboratory (ANL) proposedstatistical relationships were used to develop the mathematicalexpressions for Fen. The materials covered are carbon, low alloy andstainless steels. Mehta and Gosselin (1995) also included the formatfor a proposed non-mandatory appendix for potential implementationinto the ASME Code. Mehta (1998a, 1998b) provided an example ofthe application of the methodology to representative BWR locations.An update of the methodology including the impact of the recent ANLstatistical relationships, is described by Mehta (1999). Based on theirextensive research work, the Japanese researchers also haveindependently proposed a somewhat different approach for thecalculation of a Fen [e.g., Higuchi, 1999] for carbon and low alloysteels.

The Steering Committee on Cyclic Life and EnvironmentalEffects (CLEE) of the Pressure Vessel Research Council (PVRC) hasreviewed the EPRI/GE environmental fatigue evaluation methodologyand has endorsed it as a preferred approach to evaluate theenvironmental fatigue effects in any ASME Code fatigue evaluations[Yukawa, 1998]. Discussions in the various Task Groups of the CLEElast year resulted in several modifications to the methodology and aredescribed next.

RECENT PVRC ACTIONSLast year, the CLEE had a detailed discussion on the guidelines to

be used in developing the fatigue life environmental correction factorfor potential ASME Code implementation [WRC, 1999]. Items agreedon in the discussion include the following:

(i) The basis for the calculation of the fatigue life correction factor(Fen) will be the room temperature air S-N curve which means that Fen= ratio of fatigue life in room temperature air to life in the servicetemperature water.

(ii) Include the concept of an effective “Fen” for the purpose ofmaintaining the existing Code margins with respect to the mean watercurves. This can be done by defining an effective Fen = Fen/Z whereZ = 3 for carbon and low alloy steels and Z = 1.5 for austeniticstainless steels. At present, Z=1.5 will be applied to both cast andwrought stainless steels. Also, the value of effective Fen must be ≥ 1.0.

(iii) To minimize confusion in Code applications, the basis andrationale for effective Fen calculations would be separated from theprocedures for the calculation of basis Fen factors.

(iv) The following values of threshold strain amplitudes werediscussed and accepted: For carbon and low alloy steels 0.07% to0.08% with a ramp function between the two values; for wrought andcast austenitic steels 0.1% to 0.11% with a ramp between the twovalues.

35-4

(v) The temperature threshold for wrought and cast austeniticstainless steels was discussed and it was agreed to replace the presentthreshold = 200°C (derived from ANL correlations) by a ramp functionfrom 180°C to 220°C.

(vi) The criterion and threshold values for austenitic stainlesssteels were agreed on with a recognition of the scarcity of data. Afterthe meeting, Mr. Higuchi communicated that the Japanese project onenvironmental fatigue would review the available data in Japan andsend the results as soon as possible.

The proposed methodology of Mehta and Gosselin (1995) in theform of a non-mandatory appendix was modified to reflect thepreceding discussion and is included here as Appendix A. The PVRCforwarded to the Board on Nuclear Codes and Standards (BNCS) theproposed non-mandatory appendix text along with the suggestions forconsideration and potential Code implementation [WRC, 1999].

POTENTIAL CODE IMPLEMENTATION APPROACHESThis section presents the suggested changes in the appropriate

articles of ASME Section III and XI to provide stress analysts withenabling words to include environmental effects in the Code fatigueevaluations. The evaluation procedure itself could be first developedas a Code Case and in the future can potentially be incorporated in theform of a non-mandatory appendix to the Code.

Potential Section III ImplementationFor the evaluation of new designs, the environmental fatigue

procedures would be part of Section III. The enabling words could beadded in Paragraphs NB-3200 and NB-3600.

NB-3200. The procedure for fatigue analysis in NB-3200 iscontained in NB-3224.4(e), “Procedure for Analysis for CyclicLoading.” Paragraph (5) of NB-3224.4 (e) stipulates six steps forcalculating the cumulative fatigue damage when there are two or moretypes of stress cycles which produce significant stresses. Add thefollowing step to NB-3224.4(e):

Step 7: When the environmental effects on the fatigue life areconsidered significant, such effects may be accounted for by using themethods described in Appendix XX.

NB-3600. Paragraph NB-3610 specifies the general requirementsof piping design. A sub-paragraph NB-3614 worded as follows may beadded to provide enabling words:

NB-3614 Environmental Effects. When the environmental effectson the fatigue analysis required by NB-3650 are consideredsignificant, such effects may be accounted for by using the methodsdescribed in Appendix XX.

The other location that needs to be modified is Subparagraph3653.8 which might be added as follows:

NB-3653.8 Consideration of Environmental Effects. When theenvironmental effects are considered significant, the cumulative fatiguedamage shall be that calculated using the procedures of Appendix XX.

Potential Section XI ImplementationThe most logical place for enabling words in Section XI is

Paragraph IWB-3740 which pertains to operating plant fatigueassessment. For example, the IWB-3740(a) could be modified asfollows:

(a) Appendix L provides procedures that may be used to assessthe effects of thermal and mechanical fatigue concerns on componentacceptability for continued service. When the environmental effects onthe fatigue analysis are considered significant, such effects may beaccounted for by using the methods described in Appendix XX.

The reference to Appendix XX in the above sentence could be anon-mandatory appendix in Section III.

Potential Implementation as Section XI Code CaseThis could be in the form of an inquiry worded as follows:Inquiry: Section XI, Division 1, IWB-3740 has provision for

assessment of the effects of thermal and mechanical fatigue concernson component acceptability for continued service. What proceduresmay be used to account for the reactor water environment effects insuch assessments?

Reply: It is the opinion of the Committee that the thermal andmechanical fatigue concerns from the reactor water environmenteffects may be evaluated in accordance with the followingrequirements (Appendix A).

RECENT REGULATORY ACTIONSThe following excerpts from a recent letter written by the

Advisory Committee on Reactor Safeguards (ACRS) to the US NRC[Powers, 1999] provides a good summary of the current regulatorythinking in relation to environmental fatigue effects:

Recommendations• We agree with the staff’s proposal that GSI-190 be resolved

without any additional regulatory requirements.• The staff should ensure that utilities requesting license renewal

consider the management of environmentally assisted fatigue in theiraging management programs.

BackgroundThe effects of fatigue for the 40-year initial reactor license period

were studied and resolved under GSI-78, “Monitoring of FatigueTransient Limits for Reactor Coolant System,” and GSI-166,“Adequacy of Fatigue Life of Metal Components.”

The staff concluded that risk from fatigue failure of componentsin the reactor coolant pressure boundary was very small for 40-yearplant life. In our March 14, 1996 letter, we agreed with the staff’sconclusion.

GSI-190 was established to address the residual concerns of GSI-78 and GSI-166 regarding the environmental effects of fatigue onpressure boundary components for 60-years of plant operation. Thescope of GSI-190 included design-basis fatigue transients, studying the

35-5

probability of fatigue failure and its effects on core damage frequency(CDF) of selected metal components for 60-year plant life.

DiscussionResolution of GSI-190 was based on the results of an NRC-

sponsored study performed by the Pacific Northwest NationalLaboratory (PNNL). In that study, PNNL examined design-basisfatigue transients and the probability of fatigue failure of selected metalcomponents for 60-year plant life and the resulting effects on CDF.

The PNNL study showed that some components have cumulativeprobabilities of crack initiation and through-wall growth that approachunity within the 40- to 60-year period. The maximum failure rate(through-wall cracks per year) was in the range of 10-2 per year, andthose failures were associated with high cumulative usage factorlocations and components with thinner walls, i.e., pipes morevulnerable to through-wall cracks. There was only a modest increase inthe frequency of through-wall cracks in major reactor coolant systemcomponents having thicker walls. In most cases, the leakage from thesethrough-wall cracks is small and not likely to lead to core damage.Therefore, the projected increased frequency in through-wall cracksbetween 40- and 60-years of plant life does not significantly increaseCDF. Based on the low contributions to CDF, we agree with theproposed resolution of GSI-190.

Environmentally assisted fatigue degradation should be addressedin aging management programs developed for license renewal.Minimization of leakage is important for operational safety,occupational doses, and for continued economic viability of theplants.”

The PNNL probabilistic study the ACRS referred to wasconducted by Simonen, et al. (1999). Recently, the NRC wrote a letterto the BNCS [Craig, 1999] requesting ASME action to address issuesrelated to the effects of the reactor water environment on the reductionof fatigue life of light water reactor components.

ASME CODE ACTIVITYIn the Section XI Code space, the environmental fatigue issue is

currently being addressed by the Task Group on Operating PlantFatigue Assessments reporting to the Working Group on OperatingPlant Criteria. The PVRC-endorsed approach is under discussion inthe Task Group. The responsibility for the Code fatigue curves in theSection III space is with the Subgroup on Design and Subgroup onMaterials, Fabrication and Examination. However, the mainresponsibility for the Code S-N curves and the fatigue design methodsis with the Subgroup on Fatigue Strength reporting to theSubcommittee on Design. The Operating Plant Fatigue AssessmentsTask Group plans to hold joint meetings with the Subgroup on FatigueStrength to coordinate the activities related to the environmentalfatigue issue.

INDUSTRY ACTIVITIESThe industry concern with the environmental fatigue effects has

mostly been in the context of the plant license renewal. Baltimore Gas& Electric (Calvert Cliffs Nuclear Power Plant) and Duke Energy

(Oconee Nuclear Station) have essentially proposed to use the PVRC-preferred approach to evaluate the environmental fatigue effects for thelicense renewal terms [Tuckman, 1999; Cruse, 1999]. Given that asignificant number of plant owners are expected to apply for thelicense renewal of their units, the industry is currently working with thelicense renewal committee of the EPRI and the Nuclear EnergyInstitute (NEI) to develop an approach to this issue that does notimpose a significant burden while assuring safe plant operation.

DISCUSSIONA key difference between the PVRC-preferred approach outlined

in the Appendix and the Japanese approach [e.g., Higuchi, 1999] is thefact that the former uses a working definition of ‘moderate’environmental effects that are assumed to have already been accountedfor in the factor of 20 on cycles used in developing the original CodeS-N curves. Therefore, the calculated values of Fen are divided by afactor (‘Z’ factor in the Appendix) to obtain an effective value of Fen.Some of the supporting arguments in the favor of an effective Fenapproach are the following:

• A recent paper by Chopra and Shack (1999) provides a goodrationale for taking some credit out of the factor of 4 assigned tosurface finish, atmosphere, etc. They state, “Because carbon and low-alloy steels and austenitic SSs develop a corrosion scale in LWRenvironments, the effect of surface finish may not be significant, i.e.,the effects of surface roughness are included in environmentallyassisted decrease in fatigue life in LWR coolant environments. Inwater, the sub-factor on life to account for surface finish effects may beas low as 1.5 or may be eliminated completely; a factor of 1.5 on strainand 7 on cycles is adequate to account for the uncertainties that arisefrom material and loading variability. Therefore, the factor of 20 onlife that is used in developing the design fatigue curves includes, as asafety margin, a factor of 3 or 4 on life that may be used to account forthe effects of environment on the fatigue lives of these steels.”

• Cooper (1992) estimated an average factor of 3 on fatigueinitiation with respect to the design curve in the PVRC tests on lowalloy steel vessels tested with water at room temperature.

• The analytical expressions for Fen give values greater than 1even when the parameters are below the threshold. For example, in thecase of stainless steel, the Fen equation (Equation 3 in Appendix)predicts a value of 2.55 even at room temperature. Similarly, the valueis 2.47 for low alloy steels. Ideally, the Fen values should approach 1as the threshold values are reached.

Development of a sound technical basis for the effective Fenapproach is currently under consideration by the CLEE.

SUMMARYA PVRC-preferred procedure to address reactor water

environmental effects in pressure vessel and piping fatigue evaluations,was described in this paper. This subject is currently under activediscussion with the regulatory agencies, the ASME Code and theindustry. The paper briefly described the activities of these groups inrelation to this subject.

35-6

ACKNOWLEDGEMENTThe author would like to acknowledge many helpful discussions

with and encouragement received from Drs. Sumio Yukawa and W.A.Van Der Sluys.

REFERENCESChopra, O.K. and Shack, W.J., 1999, “Methods for Incorporating

Effects of LWR Coolant Environment into ASME Code FatigueEvaluations,” ASME PVP-Volume 386, pp. 171-181.

Cooper, W.E., 1992, “The Initial Scope and Intent of the SectionIII Fatigue Design Procedures,” PVRC Workshop on EnvironmentalEffects on Fatigue Performance, Clearwater Beach, FL, January 20.

Craig, J.W., Director, Division of Engineering Technology, officeof Nuclear Regulatory Research, NRC, Letter to J.H. Ferguson,Chairman, BNCS, December 1, 1999.

Cruse, C.S., Baltimore Gas and Electric Company, Letter toUSNRC Document Control Desk, Confirmatory Item 3.2.3.3-1, July 2,1999.

Higuchi, M., 1999, ” Fatigue Curves and Fatigue Design Criteriafor Carbon and Low Alloy Steels in High-Temperature Water,” ASMEPVP-Volume 386, pp. 161-169.

Hollinger, G.L., Executive Director, PVRC, Letter to J.H.Ferguson, Chairman, BNCS, October 31, 1999.

Mehta, H.S. and Gosselin, S.R., 1995, “An Environmental FactorApproach to Account for Reactor Water Effects in Light Water ReactorPressure vessel and Piping Fatigue Evaluations,” EPRI Report No. TR-105759. Also, ASME PVP-Volume 323 (1996).

Mehta, H.S. and Gosselin, S.R., 1998, “Environmental FactorApproach to Account for Water Effects in Pressure Vessel and PipingFatigue Evaluations,” Nuclear Engineering and Design Journal,Volume 181, pp. 175-197.

Mehta, H.S., 1998a, “Environmental Fatigue Evaluations ofRepresentative BWR Components,” EPRI Report No. TR-107943.

Mehta, H.S., 1998b “Application of EPRI/GE EnvironmentalFactor Approach to Representative BWR Pressure Vessel and PipingFatigue Evaluations,” ASME PVP-Vol. 360, pp. 413-425.

Mehta, H.S., 1999, ” An Update on the EPRI/GE EnvironmentalFatigue Evaluation Methodology and Its Applications,” ASME PVP-Volume 386, pp. 183-193.

Powers, D.A., Chairman ACRS, Letter to William D. Travers,Executive Director for Operations, US NRC, Dated December 10,1999,Subject: Proposed Resolution of GSI-190.

Simonen, F.A., et al., 1999, “Evaluation of Environmental FatigueEffects on Fatigue Life of Piping,” Presented at the 27th Water ReactorSafety Information Meeting, October 25-27.

Tuckman, M.S., Duke Energy Corporation, Letter to USNRCDocument Control Desk, response to SER Open Item 4.2.3-2, October15, 1999.

Welding Research Council Progress Reports, 1999, Volume LIX,No. 5/6, May/June, pp. 33-36.

Yukawa, S., 1998, ” PVRC Progress Report to BNCS, October 9,1998, Atlanta, Georgia,” Welding Research Council Progress Reports,Volume LIII, No. 9/10, Sept./Oct.

NONMANDATORY APPENDIX XX

FATIGUE EVALUATIONS INCLUDINGENVIRONMENTAL EFFECTS

ARTICLE X-1000

SCOPEThis Appendix provides methods for performing fatigue usage

factor evaluations of reactor coolant system and primary pressureboundary components when the effects of reactor water on fatigueinitiation life are judged to be significant.

X-1100 ENVIRONMENTAL FATIGUE CORRECTIONThe evaluation method uses as its input the partial fatigue usage

factors U1, U2, U3, .....Un, determined in Class I fatigue evaluations. InClass I design by analysis procedure, the partial fatigue usage factorsare calculated for each type of stress cycle in paragraph NB-3222.4(e)(5). For Class I piping products designed using NB-3600procedure, Paragraph NB-3653 provides the procedure for thecalculation of partial fatigue usage factors for each of the load set pairs.

The cumulative fatigue usage factor, Uen, considering theenvironmental effects is calculated as the following:

Uen = U1•Fen, 1 + U2•Fen, 2 + U3•Fen, 3 ... Ui•Fen, i ....+ Un•Fen, n

where, Fen,i is the effective environmental fatigue correction factorfor the ith stress cycle (NB-3200) or load set pair (NB-3600).

X-1200 ENVIRONMENTAL FACTOR DEFINITIONX-1210 The nominal values of environmental fatigue

correction factors are to be calculated using the expressions below.

Carbon Steel

Fen,nom = [exp (0.559 - 0.101S*T*O*ε’*)] (1)

Low Alloy Steel

Fen,nom = [exp (0.903 - 0.101S*T*O*ε’*)] (2)

Stainless Steels (wrought and cast)

Fen,nom = exp [0.935 - T*O*ε’*)] (3)

X-1260 The effective environmental fatigue correctionfactor, Fen, is obtained by dividing the nominal value calculated inX-1210 with a material-specific factor which accounts for moderate

35-7

environmental fatigue effects already included in the S-N curves ofFigures I-9.1 and I-9.2.

Fen = Fen,nom/Z, but no less than 1.0

Where, Z = 3.0 for carbon and low alloy steels and 1.5 forwrought and cast stainless steels.

X-1300 EVALUATION PROCEDURESFor some types of stress cycles or load set pairs any one or more

than one environmental parameters are below the threshold value forsignificant environmental fatigue effects. The value of theenvironmental fatigue correction factor, Fen for such types of stresscycles or load set pairs shall be equal to 1.0. Article X-2000 providesprocedure for threshold criteria evaluation.

The procedures for the evaluation of Fen factors for design byanalysis and for Class I piping products fatigue evaluations areprovided in X-3000.

X-1400 NOMENCLATUREThe symbols adopted in this Appendix are defined as follows:E = Young’s Modulus, psiFen = Effective environmental correction factor applied to

fatigue usage calculated using Code fatigue curvesFen(τ) = Environmental correction factor calculated at a specific

instant in time, τ.Fen,int = Environmental correction factor based on integrated

approach.DO = Dissolved oxygen content of water (ppm)O* = Transformed oxygen contentS = Sulfur content of carbon and low-alloy steels, weight %S* = Transformed sulfur contentSalt = Alternating stress amplitude, psiSrange = Range of stress intensity associated with a transient

cycle, psiT = Temperature (°C)T* = Transformed temperature.Ta = Average temperature on side ‘a’ during a temperature

transientTb = Average temperature on side ‘b’ during a temperature

transientTc = Sum of |Ta - Tb|, |∆T1| and |∆T2| for temperature transient

producing compressive stresses at the component surfacein contact with fluid

Tm = Metal temperature during a temperature transient atsurface in contact with fluid

Tt = Sum of |Ta - Tb|, |∆T1| and |∆T2| for temperature transientproducing tensile stresses at the component surface incontact with water

∆T1 = Linear temperature gradient through a component wallduring a temperature transient

∆T2 = Nonlinear temperature gradient through a componentwall during a temperature transient

tt = Elapsed time between the start of temperature transientand the time when Tt is reached, seconds

tT,th = Elapsed time between the start of decreasing temperaturetransient and the time when metal surface in contact withfluid reaches threshold temperature, seconds

Uen = Cumulative fatigue usage factor including theenvironmental effects

Ui = Cumulative fatigue usage factor for load set pair ‘i’obtained by using Code fatigue curves

εi = Strain range for load set pair i, %ε’ = Strain rate, %/secondε’* = Transformed strain rateε’*(τ) = Transformed strain rate at elapsed time equal to τ

ARTICLE X-2000

ENVIRONMENTAL FATIGUE THRESHOLDCONSIDERATIONS

X-2000 SCOPEThis Article provides procedure for screening out types of stress

cycles or load set pairs for which any one or more than oneenvironmental parameters are below the threshold value for significantenvironmental fatigue effects. The value of the environmental fatiguecorrection factor, Fen for such types of stress cycles or load set pairsshall be equal to 1.0.

X-2100 STRAIN AMPLITUDE THRESHOLDSX-2110 The strain amplitude threshold for carbon and low

alloy steels is 0.07%. Fen values shall be used at strain amplitudesequal to or exceeding 0.08%. A linear interpolation may be used tocalculate Fen values for strain amplitudes between 0.07% and 0.08%.

X-2120 The strain amplitude threshold for wrought andcast stainless steels is 0.10%. Fen values shall be used at strainamplitudes equal to or exceeding 0.11%. A linear interpolation may beused to calculate Fen values for strain amplitudes between 0.10% and0.11%.

X-2130 Calculate the strain amplitude, εi associated with atype of stress cycle or load set pair ‘i’ by multiplying the alternatingstress intensity Salt i by 100 and dividing by the modulus of elasticity E.The value of E shall be obtained from the applicable design fatiguecurves of Figs. I-9.0.

X-2140 If the value of εi calculated in X-2130 for a load setpair is less than or equal to appropriate value from X-2110 or X-2120,that load set pair satisfies the threshold criterion and the value of Fen i is1.0. No further evaluation with respect to other threshold values needbe made for this load set pair.X-2200 STRAIN RATE THRESHOLD

The strain rate threshold is 1.0%/second for carbon and low alloysteels, and 0.4%/second for wrought and cast stainless steels. A loadset pair involving only the seismic loading satisfies the strain ratethreshold criterion for strain rate and the value of Fen i is 1.0. No

35-8

further evaluation with respect to other threshold values need be madefor this type of stress cycle or load set pair.

If the strain rate associated with the tensile stress load set for anyother load set pair exceeds the threshold value, Fen is 1.0 for that loadset pair.

X-2300 TEMPERATURE THRESHOLDX-2310 The temperature threshold for carbon and low alloy

steels is 150°C.X-2320 The temperature threshold for wrought and cast

stainless steels is 180°C.X-2330 Define the effective temperature, T associated with

a type of stress cycle or load set pair ‘i’ as equal to the higher of thehighest temperatures in the two transients or load sets constituting thetype of stress cycle or load set pair.

X-2340 If the temperature calculated in step (b) is less thanor equal to the threshold value, the stress cycle or load set pair satisfiesthe threshold criterion for temperature and the value of Fen i is 1.0.

X-2400 DISSOLVED OXYGEN THRESHOLDThis is applicable only to carbon and low alloy steels.(a) Define the effective dissolved oxygen content, DO ass ociated

with a type of stress cycle or load set pair ‘i’ as equal to the higher ofthe highest oxygen content in the two transients or load setsconstituting the type of stress cycle or load set pair.

(b) If the value of DO determined in step (a) for a type of stresscycle or load set pair is less than or equal to 0.05 ppm, that type ofstress cycle or load set pair satisfies the threshold criterion and thevalue of Fen i is 1.0.

ARTICLE X-3000

ENVIRONMENTAL FACTOR EVALUATION

X-3100 SCOPEThis Article provides procedure for calculating the Fen factors for

types of stress cycles (NB-3200) or load set pairs (NB-3600). Only thetypes of stress cycles or load set pairs that do not meet the thresholdcriteria of X-2000 need to be considered for Fen calculation.

X-3200 EVALUATION PROCEDURE FOR DESIGN BYANALYSIS

X-3210 Determination of Transformed Strain Rate

X-3211 The strain rate (%/sec) for a stress cycle isdetermined as the following:

ε’ = Srange i •100/E• tmax

where, Srange i is the stress difference range for cycle‘i’ asdetermined in NB-3224.4(e)(5) and the tmax is the time in seconds whenthe stress difference reaches a maximum from the start of thetemperature transient. This calculation is performed only for the stepdown temperature transient or other tensile stress producing cycle inthe stress cycles constituting a pair.

X-3212 The transformed strain rate ε’* for carbon and lowalloy steels is obtained as the following:

ε’* = 0 (ε’ > 1%/sec)ε’* = ln(ε’) (0.001 < ε’ < 1%/sec)ε’* = ln(0.001) (ε’ < 0.001%/sec)

X-3213 The transformed strain rate ε’* for stainless steelsis obtained as the following:

ε’* = 0 (ε’ > 0.4%/sec)ε’* = ln(ε’/0.4) (0.0004 < ε’ < 0.4%/sec)ε’* = ln(0.0004/0.4) (ε’ < 0.0004%/sec)

X-3220 Determination of Transformed TemperatureX-3221 The temperature, T associated with a stress cycle

‘i’ is equal to the higher of the highest metal temperatures in the twotransients constituting the stress cycle or load set pair.

X-3222 The transformed temperature T* for carbon andlow alloy steels is obtained as the following:

T* = 0.0 (T< 150°C)T* = T-150 (T> 150°C)

X-3223 The transformed temperatures T* for stainlesssteels are obtained as the following:

T* = 0.0 (T<180°C)T* = (T-180)/40 (180°C<T<220°C)T* = 1.0 (T>220°C)

X-3230 Determination of Transformed DOX-3231 For carbon and low alloy steels, the effective

dissolved oxygen content, DO associated with a load set pair ‘i’ isequal to the higher of the highest oxygen level in the two transientsconstituting the load set. The transformed DO, O* is obtained asfollows:

O* = 0 (DO<0.05 ppm)O* = ln(DO/0.04) (0.05 ppm < DO < 0.5 ppm)O* = ln(12.5) (DO > 0.5 ppm)

X-3232 For wrought stainless steels, the effective dissolvedoxygen content, DO associated with a load set pair ‘i’ is equal to thelower of the oxygen level in the two transients constituting the load set.The transformed DO, O* is obtained as follows:

35-9

O* = 0.260 (DO < 0.05 ppm)O* = 0.172 (DO > 0.05 ppm)

X-3233 For cast stainless steels, O* = 0.260

X-3240 Determination of Transformed Sulfur for Carbon &Low Alloy Steels

The sulfur content S in terms of weight percent might be obtainedfrom the certified material test report or an equivalent source. If thesulfur content is unknown, then its value shall be assumed as 0.015%.The transformed sulfur, S* is obtained as the following:

S* = S (0<S<0.015 wt%)S* = 0.015 (S>0.015 wt%)

X-3250 Determination of Fen

The environmental correction factor F en i for a type of stress cycleand the cumulative fatigue usage factor shall be calculated usingequations given in X-1200.

X-3260 Determination of Fen Based on Damage ApproachProcedure similar to that described in X-3660 may be used to

remove some of the conservatism built into the Fen i determined inX-3250.

X-3600 EVALUATION PROCEDURE FOR PIPINGThe procedures in this Article use the input information and the

partial fatigue usage results from the NB-3650 fatigue evaluation. Theexample of specific load set information needed is: internal pressure,the three moment components, |Ta-Tb|, ∆T1 and ∆T2. When thedetailed results of one-dimensional transient heat transfer analyses areavailable in the form of time history of |Ta-Tb|, ∆T1 and ∆T2, suchresults may be used to reduce conservatisms in the calculated values ofenvironmental correction factor.

X-3610 Determination of Strain RateThe strain rate (%/sec) for a load set pair ‘i’ is determined as the

following:

εi’ = 200•Salt i • [Tt/(Tt + Tc )]/(E• tt)

where, Salt i is the alternating stress intensity for load set pair ‘i’calculated in NB-3653.3. This calculation is performed only for thestep down temperature transient in a load set pair.

The transformed strain rate εi’* shall be obtained as described inX-3210.

X-3620 Determination of Transformed TemperaturesThe transformed temperatures shall be obtained as described in

X-3220.

X-3630 Determination of Transformed DOThe transformed DO shall be obtained as described in X-3230.

X-3640 Determination of Transformed Sulfur for Carbonand Low Alloy Steels

The transformed sulfur shall be obtained as described in X-3240.

X-3650 Determination of Fen

The environmental correction factor F en i shall be calculated usingequations given in X-1200.

X-3660 Determination of Fen Based on Integrated ApproachWhen the results of detailed transient analyses are available to

predict strain rate, such results may be used to reduce conservatisms inthe calculated values of Fen. The following expression or equivalentshall be used:

tT,th

Fen,int = (1/tT,th) ∫0 [Fen(τ)]dτ

The preceding value of Fen may be used in lieu of the Fen valuecalculated in X-3650. Fen(τ) is the appropriate environmental factorderived from X-1200, with time dependent properties/factors for thetime in the transient where the temperature exceeds the thresholdvalue.

36-1

36 CODES AND THERMAL FATIGUE: STATUS ANDON-GOING DEVELOPMENT

C. FaidyEDF-SEPTEN

VIBRATION/HIGH CYCLE FATIGUE

37-1

37 VIBRATION FATIGUE TESTING OF SOCKET WELDS,PHASE II

Paul Hirschberg, Peter C. RiccardellaStructural Integrity Associates

Michael Sullivan, Jon SchletzPacific Gas & Electric Co.

Robert CarterEPRI

37-3

VIBRATION FATIGUE TESTING OF SOCKET WELDS, PHASE II

Paul Hirschberg Michael Sullivan Robert CarterPeter C. Riccardella Jon Schletz EPRI Project ManagerStructural Integrity Associates Pacific Gas & Electric Co. 1300 Harris Boulevard3315 Almaden Expressway, Suite 24 3400 Crow Canyon Rd. Charlotte, NC 28262San Jose, CA 95118-1557 San Ramon, CA 94583

ABSTRACTThis paper describes the results of the second phase of an EPRI sponsored program to perform high

cycle fatigue testing of socket welds in order to quantify the effects of various factors upon the fatiguestrength. Analytical results had demonstrated that the socket weld leg size configuration can have animportant effect on its high cycle fatigue resistance, with longer legs along the pipe side of the weldgreatly increasing its predicted fatigue resistance. Other potentially important factors influencing fatiguelife include residual stress, weld root and toe condition, pipe size, axial and radial gaps, and materials ofconstruction. The second phase of the program tested 27 additional socket weld specimens of variousdesigns by bolting them to a vibration shaker table and shaking them near their resonant frequencies toproduce the desired stress amplitudes and cycles. Another objective of the second phase of testing was toevaluate various methods of in-situ modification or repair of socket welds, which could be used asalternatives to replacement with butt welds. The results of the program are presented which includecomparisons of the various socket weld designs with standard Code socket welds, butt welds, ASMEmean failure data, and recent test data published by Higuchi et al. Fatigue Strength Reduction Factors arecalculated based on the testing results.

INTRODUCTIONFailures of small bore piping connections continue to occur frequently in U.S. nuclear power plants,

resulting in degraded plant systems and unscheduled plant downtime. Prior research [1] has indicated thatthe majority of such failures are caused by vibration fatigue of socket welds. In order to better understandand characterize this phenomenon, investigations have been performed in the U.S. [1,2,3] and overseas[4,5,6]. Analytical results reported in Reference [3] have demonstrated that the socket weld leg sizeconfiguration can have an important effect on its high cycle fatigue resistance, with longer legs along thepipe side of the weld greatly increasing its predicted fatigue resistance. Other factors which potentiallyinfluence socket weld fatigue life are residual stress, weld root and toe condition, loading mode, pipe size,axial and radial gaps, and materials of construction.

The test program described in this paper was initiated in 1997 under EPRI sponsorship to study theimportance of these factors. A large number of socket weld samples were vibration-fatigue tested tofailure on a high frequency shaker table. The objectives of the testing were to improve the industry’sunderstanding and characterization of the high cycle fatigue resistance of socket welds and to developappropriate fatigue strength reduction factors for such welds reflecting the effects of those factors listedabove which prove to be significant. The ultimate goal of this research is to develop recommendeddesign and fabrication practices that can be used to enhance socket weld fatigue resistance in vibration-sensitive locations, as well as to provide guidelines for screening out and preventing vibration-fatiguefailures in existing welds.

The first phase [7], completed in 1998, investigated the variation in high cycle fatigue resistance as afunction of weld leg length, pipe diameter, and piping material. The effect of an additional weld pass,post-weld heat treatment, and eliminating the ASME Code-required axial gap were also studied. The testsetup and results have been described in a previous paper [8]. The second phase investigated remedialactions for existing socket welds, in order to determine their effectiveness in extending the useful life ofsocket welded joints, in lieu of making expensive modifications. The effect of varying toe conditions was

37-4

also studied in Phase II, and additional data at higher loads was collected to further quantify the benefit ofincreased weld leg length. This paper describes the second phase of the test program, and provides thecumulative results for both phases.

BACKGROUND AND APPROACHThe probability of fatigue failure at a weld is a function of many considerations: type of weld, weld

profile, material, weld size, wall thickness, heat treatment, grinding after welding, etc. Socket welds arecommonly used in small bore piping (2” and under) joints due to their ease of assembly, but their fatiguestrength is generally considered to be less than that of a butt weld. The fatigue strength reduction factor(FSRF) is a means of quantifying the effect of a local structural discontinuity on the fatigue strength ofthe joint. It is defined as the fatigue strength of the component without the discontinuity divided by thefatigue strength with the discontinuity. The FSRF is usually determined by test, and is generallycalculated by dividing the endurance limit of the ASME mean failure curve for polished bar specimens,by the endurance limit of the joint in question. The endurance limit is used because the FSRF varies withfatigue life; the value at the endurance limit is bounding. For practical reasons, the endurance limit isoften defined as the fatigue strength at some large, predefined number of cycles, such as 2x107. ArticleNB-3600 of Section III, which provides rules for the design of Class 1 piping, accounts for the variationin fatigue strength between different welds and fittings by multiplying the calculated alternating stress bystress indices, before entering the fatigue curves. For primary plus secondary bending moment loading,the applicable stress index is the product of C2 and K2. C2K2 is approximately equivalent to the FSRF,although there are some differences in derivation relating to testing requirements and statistical treatmentof the data. For socket welds (1998 Edition of the Code), C2 = 1.93 for a Code minimum weld accordingto the rules of the fabrication section (NB-4427), and the product C2K2 is 3.86, or about an FSRF of 4 fora socket weld.

PARAMETERS TESTEDWhile the ASME piping design Code provides stress indices for socket welds, these represent the

predicted fatigue strength of standard weld designs. This test program included welds of the standardCode design as well as several variations in weld design. The following is an overview of the parametersthat were tested.

Weld Leg SizeAnalysis described in Reference 3, and testing reported in Reference 10, have identified variations in

socket weld design that can significantly affect the fatigue life of the joint under high cycle loading. Oneof the most important factors is the weld leg size. Extending the weld leg length along the pipe side of theweld by a factor of two over the Code minimum of 1.09 tn (where tn is the nominal pipe wall thickness)has resulted in a significant improvement in fatigue resistance. In this testing program, the effect of weldsize on fatigue resistance was studied by testing samples fabricated with oversized legs on the pipe side,and comparing them to control samples at nominal Code dimensions. Testing was also done to study howconventional small bore butt welded fittings compare to new 2 x 1 welds, and to 1 x 1 welds modified to2 x 1 weld dimensions, under comparable loading conditions. The butt welded specimens consisted ofpipes welded to standard weld neck flanges, which is similar to a typical method used in power plants forupgrading socket welded fittings when vibration problems have occurred.

Weld Modification and RepairAn important purpose of the testing program was to study repair or modification actions that could be

taken to improve the fatigue resistance of an existing socket welded joint. Repair concepts for leakingsocket welds that would allow plants to continue operating until the next outage before implementing apermanent repair were also tested. Weld overlay repairs were applied to four of the welds that developedleaks in the first phase of testing. The leaks were peened with water still in the pipe, and a seal weld wasapplied when the leak stopped. The overlay design applied sufficient additional reinforcement to cover

37-5

the possibility of either a toe or a root failure having occurred. The design is shown in Figure 3. Theweld was applied using a shielded metal arc welding process to simulate the most likely welding processthat would be used for a temporary repair in a power plant. The pipe side toe of the weld did not have ablending radius, producing a sharp transition, as would likely be found in a field repair. The repairedwelds were tested at their original load levels to determine if they could survive until the next outage, orconvenient time for a permanent repair. Specimens chosen for these tests included carbon and stainlesssteel welds that had experienced root and toe failures.

Another remedial approach tested was to modify existing standard Code welds to a 2 x 1 leg lengthconfiguration after having been fatigue cycled in their original configuration but before any obviouscracking or fatigue damage was observed. This is illustrated in Figure 2. Testing compared the degree ofimprovement for the modified welds relative to new 1 x 1 and 2 x 1 welds.

Weld Profile and Toe ConditionThe profile of the weld and the condition of the weld toe can have a significant effect on socket weld

fatigue life. The ideal toe condition is for the weld to blend smoothly with the pipe with no discontinuityor undercut. Premature toe failures are generally the result of a discontinuity or small flaw in the toe,which propagates through the base metal. Grinding the weld can help by removing small surfaceimperfections, which could be sites for crack initiation. Conversely, heavy “abusive” grinding canworsen the situation by leaving a cold worked condition and tensile residual stresses. In the first phase oftesting, several specimens exhibited toe failures prior to the expected number of cycles. When toefailures occurred, they were generally associated with minor welding discontinuities at the toe, whichwould normally be acceptable by Code workmanship standards. In the second phase, samples withpolished toes, as-welded toes, and intentionally poor toes were tested to quantify the effect of the toecondition. One of the butt welds was also fabricated with a toe defect to see if this is a concern for buttwelds.

Residual StressThe residual stress in the weld can have an important effect on fatigue life in the high cycle regime.

Residual stress acts as a mean stress, which reduces the allowable number of cycles at a particularalternating stress. This is more important in the high cycle region than in low cycle fatigue, where theeffects of plasticity act to relieve the mean stress. The residual stress in a socket weld can be altered byvarying the welding technique, such as by adding a final cover weld pass. “Last Pass Improvement”refers to a technique in which a normal Code socket weld is improved by adding a last pass on the pipeside of the weld, which changes the residual stress in the weld root to compressive and extends the leglength along the pipe somewhat. Tests were conducted to determine the amount of benefit gained by thisprocess.

Post-Weld Heat TreatmentPost-weld heat treatment (PWHT) is another method of reducing the residual stress in a weld. The

ASME Code requires PWHT for most ferritic steel welds, but exempts socket welds and austenitic steels.One of the reasons is that heating austenitic steels can sensitize them to intergranular stress corrosioncracking (IGSCC). However, in view of the importance of residual stresses in high cycle fatigue life,applying PWHT to socket welds in vibration sensitive locations that are not exposed to an IGSCCconducive environment can be a viable means of reducing residual stress.

Axial GapThe ASME Code requires that an axial gap of 1/16” be provided between the pipe end and the socket.

The reason is that if the pipe carries hot fluid, differential thermal expansion between the pipe and thefitting may add significant stress to the weld if there is no gap. Also, without the gap, shrinkage of thefillet weld could produce residual stresses in the weld, pipe, and fitting wall. However, some recenttesting [4] has indicated that absence of the gap has a negligible effect on fatigue life.

37-6

Pipe SizeThe ASME Code design fatigue curves do not reflect any differences in fatigue strength as a function

of pipe diameter. However, the Phase 1 testing showed that ¾” pipe had superior fatigue strength to the2” pipe.

Piping MaterialAlthough the endurance limit of carbon steel is lower than that of stainless, previous test data has

indicated that some of the fatigue strength reduction factors for stainless steel are higher. While themajority of the samples tested in this program were stainless steel, one-third of the 2” specimens werecarbon steel. The purpose was to note whether any of the modifications to socket weld design are more orless effective in carbon steel than in stainless.

TESTING ARRANGEMENTThe test setup was previously described in detail in [8]. The testing method used specimens vertically

cantilevered on a shaker table. All specimens were fabricated from Schedule 80 piping and compatiblecomponents. Loading amplitudes were selected based on fatigue data available in the literature, with thetarget of generating failures in approximately 106 to 107 cycles. As a large number of runouts (tests endedwithout failure) were obtained in the first phase of testing, the stress levels in the second phase of testingwere increased.

Sets of nine socket weld specimens were bolted to the shaker table and shaken simultaneously neartheir resonant frequencies to produce the desired stress amplitudes and cycles. A cantilevered specimenof the type illustrated in Figure 1 was used, with the test weld being the weld at the lower end of thespecimen between the pipe and the flange used to bolt the specimen to the table. Different loadamplitudes were applied to different samples in the same test by fine tuning the specimen naturalfrequencies relative to the shaker table excitation frequency. The shaker table typically ran at 100-110 Hzand the test specimens had natural frequencies that were nominally 4-8 Hz lower than the excitationfrequency. By adding or subtracting small masses, such as nuts and washers, the frequencies of the testspecimens were moved enough off resonance to adjust each individual response acceleration. The flangeconfigurations were modified to produce socket weld details typical of the socket welded fittings used onsmall bore piping in nuclear plants (tees, elbows, weldolets, couplings, etc.). The specimens werepressurized with air to a moderate pressure of approximately 50 psig. When depressurization occurred,indicating a failure, the specimen was removed from the table at the next convenient test stoppage, andthe testing was then resumed with only the remaining specimens. This test method is directly comparablewith the plant loading mechanism of concern (vibration fatigue), whereas conventional fatigue testingtechniques (rotating beam or four point bending) can have considerable variability with respect to eachother [5,6], and possibly with respect to in-plant vibration.

The stress in the test sample was determined from the measured acceleration response and testfrequency, using beam formulas superimposing the concentrated weight and distributed weight effects.Nominal pipe wall but actual lengths and weights were used. It was observed that when the crack formed,the natural frequency of the specimen reduced, causing the acceleration response and the resulting stressto decline (this usually affected only the last 3-5% of the cycles). The stress reported in the test results isan average over all of the cycles.

TEST RESULTSTable 1 is a summary of the results of the Phase II tests. The Phase I results have been reported

previously in [8]. However, the results of both phases of testing are plotted in Figures 4-8. Trend curvesfrom socket weld fatigue testing reported in [4,5,6] are also shown on the plots, labeled “HiguchiCurves”. The ASME mean failure curves for polished bar specimens are also shown for comparisonpurposes. When specimens exhibited toe failures (solid points in the figures) they tended to fail somewhatprematurely, relative to the more common root failures.

37-7

Table 1 - Phase II Test Results

Test Series 4 – Repairs and Mods. - 2” SS and CSSpecimen Sa(ksi) Nf Comments1 – Mod. to 2x1 SS 17.4 2.57E+07 Runout2 – Mod. to 2x1 SS 17.6 5.05E+06 Toe Failure3 – Prev. Runout SS 16.3 1.33E+07 Root Failure4 – Mod. to 2x1 CS 14.7 2.57E+07 Runout5 – Mod. to 2x1 CS 11.1 2.57E+07 Runout6 – Weld Repair CS 8.9 2.57E+07 Runout7 – Weld Repair CS 8.0 2.57E+07 Runout8 – Weld Repair SS 11.6 1.10E+07 Failure at Orig. Toe9 – Weld Repair SS 11.0 2.57E+07 RunoutTest Series 5 – Toe Conditions - 2” SSSpecimen Sa(ksi) Nf Comments1 – Smooth Toe 24.2 1.26E+06 Toe Failure2 – Smooth Toe 23.6 1.04E+06 Toe Failure3 – Poor Toe 18.2 1.62E+06 Toe Failure4 – Poor Toe 20.5 3.80E+05 Toe Failure5 – Polished Toe 21.4 3.24E+05 Toe Failure6 – Polished Toe 22.5 9.10E+05 Root Failure7 – Smooth Last Pass 22.8 3.56E+05 Toe Failure8 – Smooth Last Pass 23.3 1.62E+06 Root Failure9 – Smooth Last Pass 22.0 1.96E+06 Root FailureTest Series 6 – 2x1 vs. Butt Weld - 2”Specimen Sa(ksi) Nf Comments1 – Butt Weld SS 23.0 1.22E+06 I. D. Root Failure2 – Butt Weld SS 22.3 6.61E+05 I. D. Root Failure3 – Butt Weld CS 15.3 2.19E+06 Toe Failure4 – Butt Weld CS 16.8 1.97E+06 Toe Failure5 – Butt SS Undercut 22.6 1.00E+06 Toe Failure6 – 2x1 SS 24.0 1.04E+06 Toe Failure7 - 2x1 SS 23.5 2.11E+07 Runout8 – 2x1 CS 16.5 2.38E+06 Root Failure9 – 2x1 CS 16.9 3.75E+06 Root Failure

In general, failures at a lower number of cycles and/or higher stress tended to originate at the toewhile the higher cycle / lower stress failures tended to occur at the root.

The following sections describe in detail the test results for each of the parameters studied.

Effect of Weld SizeFigures 4 and 5 show the results of testing of standard Code size welds (1 x 1) and enhanced, 2 x 1 welds,which have a leg dimension along the pipe of twice that of the Code minimum leg dimension. Alsoshown for comparison are the test results for the butt welds. Figure 4 shows the 2” stainless steel data;Figure 5 shows the 2” carbon steel data. In general, the nominal Code dimension (1 x 1) specimensyielded data somewhat above the corresponding “Higuchi Curve”. The failures at the lower numbers ofcycles were toe failures and those at higher cycles were root failures (open points in the figures). The twobutt weld failures at low cycles in this figure originated in the inside diameter of the pipe, at the edge ofthe weld root.

37-8

The 2 x 1 specimens were significantly stronger than the standard welds. All exhibited runouts at thelower stress levels, even though tested at stress amplitudes 30% higher than those applied to the standardCode specimens. At the higher stress levels, in the stainless steel tests, one was a runout at 23.5 ksi, whilethe other failed at only 1 x 106 cycles, and 24 ksi. The latter specimen was a toe failure, however, andperformed about the same as the standard Code specimens tested at this stress level, which also exhibitedtoe failures. In the carbon steel tests at higher stress levels, both of the 2 x 1 welds performedsignificantly better than the standard welds.

The butt welds were tested for the purpose of comparison against the 2 x 1 welds. However, the buttwelds did not perform as well as expected. In the stainless steel tests, they were about as strong as thestandard Code socket welds, while in the carbon steel tests, they were somewhat better than the standardwelds, but not quite as good as the 2 x 1 welds. The butt weld results can be explained by the fact that thetests were comparing a socket welded flange to a butt welded, weld neck flange, and not to a pipe-to-pipebutt weld. The weld neck flange acts as a tapered transition, and the ASME Code stress indices reflect areduced fatigue strength for such a joint. For the as-welded standard socket weld (1998 Code), C2K2 =3.86; for the as-welded butt welded transition, C2K2 = 3.78. For an as-welded pipe-to-pipe butt weld,C2K2 = 2.57. Thus the stress indices for the butt welded transition are only slightly better than that of asocket weld. Socket welds are rarely used in pipe-to-pipe connections but rather at elbows, branchconnections, or other fittings which would result in thickness transitions even if replaced with butt welds.Therefore the practice of replacing socket welds with butt welds to “fix” vibration fatigue problemsappears to be of little value, if the butt weld is to a fitting with a geometric transition.

Two of the stainless steel butt welds failed by a crack that originated at the inside diameter of thepipe, at the edge of the weld root. The crack grew perpendicular to the pipe axis, through the weld metal.Further examination of the weld root indicated the presence of a number of discontinuities, which mayhave served as crack initiation sites. The other stainless steel butt weld had an undercut intentionallyplaced at the toe; although the specimen failed at the toe, the fatigue strength was equivalent to the otherbutt welds.

It was concluded from these tests that the 2 x 1 welds offer a significant improvement in fatiguestrength over standard Code welds. They also provide a greater strength improvement than replacementwith butt welded fittings, in addition to easier fitup and construction.

Toe ConditionThe results of testing the three different toe conditions are shown in Figure 6. The poor toe

conditions produced failures at or below the Higuchi data curve. As expected, they were toe failures. It isinteresting to note that the two polished toe specimens did not show any improvement over the smooth,as-welded pieces. One of the polished toes failed at the toe right on the Higuchi trend line, which wasbelow all of the standard, as-welded samples. Examination of the fracture surface under magnificationshowed multiple crack initiation sites along the scratch marks generated by the polishing. It is concludedfrom these tests that while a poor toe condition definitely reduces the fatigue life, polishing the toe doesnot seem to improve it. Possibly using a finer grit sandpaper and polishing in the axial direction wouldhave produced better results. Four of the five specimens with a final weld pass at the toe performedsomewhat better than the standard welds, with one failing early near the Higuchi curve.

Weld Process EnhancementsFigure 7 compares the results of the weld process enhancements for 2” stainless socket welds. Figure

8 shows the same for the carbon steel tests. The enhancements were: adding a “last pass” at the weld toeon the pipe side to improve the residual stress; post-weld heat treatment of the weld; and eliminating theCode required axial gap. The testing results for these specimens were previously reported in [8]. The lastpass improved specimens yielded somewhat mixed results. In general, where premature failures occurredin last pass improved specimens, they were due to toe failures, indicating that the last pass welding mayhave left a discontinuity or stress raiser at the toe. Three of the five last pass samples performed betterthan the standard socket welds. Post weld heat treatment appears to increase the fatigue life of the

37-9

standard Code specimens. One of the stainless steel runouts from Phase I was retested at a higher load intest series 4. Despite having withstood 23 million cycles at 12 ksi, it lasted another 13 million cycles at16 ksi. The ASME Code required gap appears to have no effect on high cycle fatigue resistance.

Modified and Repaired WeldsTwo stainless steel and two carbon steel 1x1 socket welds that had been tested to runout in the second

or third test series had additional weld metal applied to them to make them 2x1 welds. The modifiedwelds were then retested at loads that were approximately 50% higher than in the previous tests. Thepurpose was to determine whether an existing Code standard weld can be increased to a 2x1 weld in situto improve its fatigue strength, and how it would compare against a new 2x1 weld. The results are shownon Figures 4 (stainless) and 5 (carbon steel). One of the stainless welds was a runout at 17.4 ksi. Theother failed at the toe after 5 million cycles at 17.6 ksi. The first specimen was a former Code standardweld, and the second was a post-weld heat treated specimen. It should be noted that the latter weld wassectioned, and examinations indicated that the original weld did not have any indications of cracking priorto the buildup. Both of the carbon steel welds were runouts, the first being a former post-weld heat treatedspecimen tested at 14.7 ksi, and the second, a no-gap specimen tested at 11 ksi. The results clearly showan improvement in fatigue strength over standard Code welds. As for comparing the built up 2x1 weldswith new 2x1 welds, most of the welds of both categories were runouts, so a definitive comparison wasnot possible. However, the built up welds appear to be at least as good as the new 2x1 welds.

The results of the tests of the weld overlay repaired welds are shown on Figures 7 and 8. One of thestainless steel repaired welds lasted about 11 million cycles at 11.6 ksi. The original weld had been a last-pass specimen that had failed at the toe at 10 million cycles under the same load. This time, the weldfailed not at the new toe, but at the continuation of the original weld crack. Although the overlay did notarrest the original toe crack, it was still successful in restoring all of the fatigue life of the original weld.The second stainless specimen had originally been a Code standard weld that had failed at the root at 10million cycles at 10.5 ksi. In this test, it was a runout at a stress of 11 ksi. The weld overlay wassuccessful in arresting the original crack. The repaired weld was thus better than the original. In thecarbon steel tests, both repaired welds were runouts. They had both originally been standard Code weldsthat had failed at the root after 7 million and 10 million cycles at 8 ksi. The repaired welds withstood 26million cycles at 8 and 9 ksi respectively and did not show any evidence of crack initiation. It can beconcluded from these tests that the weld overlay repair process was not only successful in restoring theoriginal fatigue strength of the specimen, but it actually improved the weld’s fatigue resistance. The weldoverlay design used was more effective in arresting root cracks than toe cracks.

Fatigue Strength Reduction FactorsFatigue strength reduction factors (FSRFs) were calculated as a means for quantifying the test results

of the various socket weld designs. The FSRFs are approximate, as they are based on estimates of theendurance limit, a number of tests that were runouts, and a limited number of test points. Table 2 is asummary of the FSRFs for each of the categories of tests, based on the ratio of the endurance limit fromthe ASME Code mean failure curve to the apparent endurance limit from the testing results. Since thetesting was terminated at approximately 2 x 107 cycles, the alternating stress corresponding to this numberof cycles was taken as the endurance limit for the purposes of this calculation. For the 2” stainless steelspecimens, the standard Code socket welds had an FSRF of approximately 3.4. The 2 x 1 leg sizespecimens improved upon this to a value of 2.3, versus 2.9 for the butt welded specimens. Similarly, forthe carbon steel specimens, the standard weld FSRF was 3.5. The 2 x 1 welds reduced this value to 2.0,versus 2.4 for the butt welds. An apparent break in this trend is that, as observed previously in Reference5, ¾” standard Code specimens showed less of an improvement in going to the 2 x 1 weld geometry.

This is due to the fact that the standard welds in this pipe size have a higher ratio of weld sectionmodulus to pipe section modulus than in 2 “ pipe and are consequently stronger. It appears that it isreasonable to use an FSRF of 3.9 for standard Code socket welds, and an FSRF of 2.6 for 2 x 1 leg sizesocket welds in vibration fatigue applications, independent of pipe size. This is consistent with the

37-10

current ASME Section III Code requirement of C2K2 = 3.9 for standard socket welds and the lower boundof 2.6 for larger leg length welds. This applies only to welds with no root defects or lack of fusion.Testing of the samples with toe defects in this program resulted in an FSRF of about 5. The EPRIHandbook recommends a FSRF of 8.0 for “Poor” welds, which is intended to encompass welds with rootor toe defects.

A comparison of the various weld treatments indicates that post-weld heat treatment had the mostconsistent benefit, with an FSRF of 2.5 for 2” stainless, 2.5 for carbon steel, and 2.6 for ¾” stainless. The“last pass” welds had FSRFs of about 3, and the no-gap welds about 3.5.

The modification of the previously tested 1 x 1 welds, by building up the weld on the pipe side to a 2x 1 profile, produced welds that were as good as or better than new 2 x 1 welds. For stainless steel, theFSRF for the built-up 2 x 1 welds was 2.2, versus 2.3 for new 2 x 1 welds and 3.4 for new 1 x 1 welds.For carbon steel, the FSRFs were 1.8 for the built-up 2 x 1, 2.0 for the new 2 x 1, and 3.5 for the Codestandard welds. The weld overlay repairs were also successful in improving the fatigue strength ofleaking, standard Code welds. The stainless steel welds that were repaired had an FSRF of 3.2, which isbetter than new standard welds, at 3.4. The results were even better in the carbon steel specimens, withan FSRF of 2.8 for the repaired welds versus 3.5 for new Code welds.

Table 2 - Fatigue Strength Reduction Factors

Category No. Tested Avg. FSRF2” Stainless Steel TestsCode Standard (1 x 1) 2 3.42 x 1 5 2.31 x 1 Built-up to 2 x 1 2 2.2Butt Weld 3 2.9Weld Overlay Repair 2 3.2PWHT 2 2.5Last Pass 2 3.1No Gap 2 3.7Polished Toe 2 3.6Smooth As-Welded Toe 2 2.8Smooth Last Pass Toe 3 2.9Poor Toe 2 4.82” Carbon Steel TestsCode Standard (1 x 1) 2 3.52 x 1 5 2.01 x 1 Built-up to 2 x 1 2 1.8Butt Weld 2 2.4Weld Overlay Repair 2 2.8PWHT 1 2.5No Gap 1 3.1¾” Stainless Steel TestsCode Standard (1 x 1) 2 2.42 x 1 3 2.2PWHT 1 2.6Last Pass 2 2.3No Gap 1 2.4

37-11

CONCLUSIONS AND RECOMMENDATIONSOn the basis of the testing, it is concluded that socket welds with a 2 to 1 weld leg configuration

(weld leg along the pipe side of the weld equal to twice the Code required weld leg dimension) offer asignificant high cycle fatigue improvement over standard ASME Code socket welds. This weld designoffers a superior improvement in fatigue resistance than does replacement of socket welded fittings withbutt-welded fittings. Since vibration fatigue of socket welds has been a significant industry problem, it isrecommended that this improved configuration be used for socket welds in vibration critical locations.

The majority of the test failures occurred due to cracks that initiated at weld roots. However, toeinitiated failures occurred in tests at higher stress levels that were premature in comparison with identicaltests in which root failures prevailed. Therefore, care must be taken with socket welds of any design toavoid metallurgical or geometric discontinuities at the toes of the welds (such as undercut or non-smoothtransitions). Such discontinuities promote a tendency for toe failures which greatly reduces fatigue life.Tests of welds with intentionally poor toes clearly demonstrated the early failures that such discontinuitiescan produce. Polishing the toes did not provide any benefit, causing cracking originating at scratchmarks. Because of the importance of the toe condition, the last pass improvement process (in which afinal pass is added to the pipe side toe of a standard Code weld) cannot be given an unqualifiedrecommendation at this time. The last pass improved specimens had a tendency to develop toe failures.

Other conclusions drawn from this program are that the code required axial gap in socket welds(1/16”) appears to have little or no effect on high cycle vibration fatigue resistance (thermal expansioneffects were not part of the test), and that post weld heat treatment appears to increase the fatigueresistance of standard Code specimens. Although post-weld heat treatment consistently showed improvedresults, it has the downside of potentially sensitizing austenitic welds for IGSCC in certain environments.Therefore, PWHT is not recommended for situations where IGSCC may be a damage mechanism.

The test data supports the use of the ASME Section III stress indices for standard Code welds(currently C2K2 = 3.9). It also indicates that this factor may be reduced to two-thirds that value (2.6) for 2to 1 leg size welds. (The current Code lower bound of C2K2 = 2.6 is based on weld leg length, but it is afunction of the shortest leg length.) Both of these values are appropriate only if the weld roots are free ofdefects such as lack of fusion or lack of penetration.

Testing of modification and repair concepts indicated that these approaches were successful inimproving the strength of already installed socket welds without having to replace them with butt welds.Code standard 1x1 welds that were built up to a 2x1 profile performed as well as new 2x1 welds. Weldoverlay repairs of leaking standard welds not only provided enough fatigue resistance for the welds to lastto the next outage, but actually improved their fatigue strength to better than new standard Code welds.The weld overlay process was somewhat more successful repairing root failures than toe failures.

REFERENCES1. EPRI 1994, ”EPRI Fatigue Management Handbook,” EPRI TR-104534-V1, -V2, -V3.2. Smith, J. K., 1996, “Vibrational Fatigue Failures in Short Cantilevered Piping with Socket-Welding

Fittings,” ASME PVP 338-1.3. EPRI 1997, “Vibration Fatigue of Small Bore Socket-Welded Pipe Joints,” EPRI TR-107455.4. Higuchi, M. et al, 1995, “Fatigue Strength of Socket Welded Pipe Joints,” ASME PVP 313.5. Higuchi, M. et al, 1996, “A study on Fatigue Strength Reduction Factor for Small Diameter Socket

Welded Pipe Joints,” ASME PVP 338-1.6. Higuchi, M. et al, 1996, “Effects of Weld Defects at Root on Rotating Bending Fatigue Strength of

Small Diameter Socket Welded Pipe Joints,” ASME PVP 338-1.7. EPRI 1998, “Vibration Fatigue Testing of Socket Welds”, EPRI TR-111188.8. Riccardella, P. C. et al, 1998, “Vibration Fatigue Testing of Socket Welds”, ASME PVP 360.9. ASME Boiler and Pressure Vessel Code, Section III, Subsection NB, 1998 Edition.10. WRC Bulletin 432, 1998, “Fatigue Strength Reduction Factors and Stress Concentration Factors for

Welds in Pressure Vessels and Piping”, Welding Research Council.

37-12

Figure 1. Test Setup

Figure 2. 1 x 1 Weld Built Up to 2 x 1

37-13

Figure 3. Weld Overlay Design

Figure 4. Weld Size, Stainless Steel

EPRI SOCKET WELD VIBRATION TESTS2" Stainless Steel Welds

2x1 vs. Code Standard Welds

1

10

100

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N(Cycles)

Sa(

ksi)

Higuchi Curve

ASME Mean Failure

Code 1 x1

2x1 Welds

1x1 Built up to 2x1

Butt Welds

Runout

Solid Symbols are Toe Failures

37-14

EPRI SOCKET WELD VIBRATION TESTS2" Carbon Steel Welds


1.00

10.00

100.00

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N (Cycles)

Sa

(ksi

)

Higuchi Curve

ASME Mean Failure

Code 1x1

2x1 Welds

1x1 Built up to 2x1

Butt Welds

Runouts


Figure 5. Weld Size, Carbon Steel


Toe Conditions

1

10

100

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N(Cycles)

Sa(

ksi)

Higuchi Curve

ASME Mean Failure

Code Smooth As Welded

Polished Toe

Poor Toe

Last Pass

Runout


Figure 6. Toe Conditions

37-15


Weld Process Enhancements

1

10

100

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N(Cycles)

Sa

(ks

i)

Higuchi Curve

ASME Mean Failure

Code Standard

PWHT

Retested PWHT Runout

No Gap

Last Pass

Weld Overlay Repaired

Runout


Figure 7. Weld Enhancements, Stainless Steel



1.00

10.00

100.00

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N (Cycles)

Sa

(ksi

)

Higuchi Curve

ASME Mean Failure

Code Standard

PWHT

No Gap


Runouts


Figure 8. Weld Enhancements, Carbon Steel

VIBRATION FATIGUE TESTING OFSOCKET WELDS, Phase II

Paul Hirschberg, Pete RiccardellaStructural Integrity AssociatesMike Sullivan, Jon SchletzPacific Gas & Electric Co.

Robert CarterEPRI

July 2000

37-16

Background

• Leaks of small bore socket welds due to vibrationare a common problem in nuclear plants

• Can cause unscheduled outages, resulting insignificant cost impact

• EPRI sponsored a research program to testimprovements in socket weld design

• Phase I completed 1998, Phase II in 1999 (thispaper)

37-17

Purpose

• Test different socket weld designs on a highfrequency shaker table

• Determine how variations in socket weld designaffect fatigue life

• Develop fatigue strength reduction factors

• Develop guidelines for avoiding fatigue failures invibration sensitive locations

• Test in-situ repair options

37-18

Scope of Testing - Phase I

• Increased leg length (2 x 1)

• 2" and 3/4" NPS

• Carbon steel and stainless steel

• Additional weld pass at toe

• Post weld heat treatment

• Eliminating axial gap

37-19

Scope of Testing - Phase II

• Increased leg length (2 x 1) at higher loads♦ Compared against standard Code welds

♦ Compared against butt welds

♦ Carbon steel and stainless steel

• Toe condition♦ Smooth / as-welded toe

♦ Polished toe

♦ Poor toe condition

♦ Smooth additional weld pass

37-20

Scope of Testing - Phase II, cont’d

• Modified welds♦ Increase previously cycled standard welds to 2 x 1


• Repaired welds♦ Repaired leaking welds with weld overlay


• All phase II tests were 2" NPS Sch 80

37-21

Test Setup

• Specimens vertically cantilevered on shake table♦ Similar to vents and drains

• Amplitude of vibration set to cause failure in 10E6-10E7 cycles

• Nine specimens tested at a time

• Individual load amplitude varied by tuningspecimen natural frequency vs. table frequency

• Specimens pressurized with air; failure identifiedas loss of air pressure (through-wall leak)

• Applied stress determined from beam formulasusing measured accelerations and frequencies,specimen length and weight

37-22

Custom Drilled Plate on Shaker Table

L

Test Weld

Socket WeldingFlange Modified fromStandard Flange

Note:M & L will be adjusted to yielddesired natural frequency (~ 100 Hz)

SCH 80 Pipe

Mass (M)

PressureGauge

Signal fromAccelerometer

Direction OfOscillations

98039r0

Test Specimen Detail

37-23

Test Apparatus(Six of Nine Specimens Shown - Test Control Computer on Right)

37-24

1 x 1 Weld Built up to 2 x 1

37-25

Weld Overlay Design

37-26

Weld Overlay Repair (B2-2SS-2)

37-27

Results - Weld Size

• 2 x 1 welds much stronger than standard Codewelds

• 2 x 1 welds stronger than butt welds

• Root failures at high cycles, lower amplitude; toefailures at lower cycles, higher amplitude

• Two butt welds failed at low cycles due to IDcrack along side weld root

37-28



1

10

100

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N(Cycles)

Sa

(ksi

)

Higuchi Curve

ASME Mean Failure

Code 1 x1

2x1 Welds

1x1 Built up to 2x1

Butt Welds

Runout


2" (5.08 cm) Stainless Steel Weld Sizes

37-29



1.00

10.00

100.00

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N (Cycles)

Sa

(ks

i)

Higuchi Curve

ASME Mean Failure

Code 1x1

2x1 Welds

1x1 Built up to 2x1

Butt Welds

Runouts


2" (5.08 cm) Carbon Steel Weld Sizes

37-30

Butt Weld Failure

37-31

Results - Toe Condition

• Poor toe condition resulted in premature toefailures

• Three additional weld pass specimens lastedlonger than standard welds, two failed earlier atthe toe

• Polished toe specimens failed earlier thanstandard welds

37-32


Toe Conditions

1

10

100

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N(Cycles)

Sa

(ksi

)

Higuchi Curve

ASME Mean Failure

Code Smooth As Welded

Polished Toe

Poor Toe

Last Pass

Runout


Toe Conditions

37-33

Polished Toe Failure

37-34


• Additional weld pass improves fatigue life as longas toe blends smoothly

• Post weld heat treatment improves fatigue lifesignificantly; may cause sensitization or heattreatment may be difficult to achieve

• Eliminating axial gap has no consistent effect -temperature effects not tested

37-35

2" (5.08 cm) Stainless Steel Weld Enhancements



1

10

100

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N(Cycles)

Sa

(ksi

)

Higuchi Curve

ASME Mean Failure

Code Standard

PWHT

Retested PWHT Runout

No Gap

Last Pass


Runout


37-36

2" (5.08 cm) Carbon Steel Weld Enhancements



1.00

10.00

100.00

1.00E+05 1.00E+06 1.00E+07 1.00E+08

N (Cycles)

Sa

(ks

i)

Higuchi Curve

ASME Mean Failure

Code Standard

PWHT

No Gap


Runouts


37-37

Size Comparison

• Code standard 3/4" socket welds have longerfatigue life than standard 2" welds

• Improved weld configurations show less of animprovement in the 3/4" size

37-38

Modified and Repaired Welds

• Three of four previously cycled socket welds thatwere built up to 2 x 1 would not fail

• Built up welds clearly superior to new standardwelds, as good as new 2 x 1 welds

• Weld overlay repair at a minimum restored full lifeof weld; three of four tests did not producefailure

• Weld overlay repair more successful arrestingroot cracks; toe crack continued to propagate

37-39

Weld Overlay of Root Crack

37-40

Weld Overlay of Toe Crack

37-41

Fatigue Strength Reduction Factors

• Method to estimate relative benefit of variousweld design improvements

• Calculated by dividing endurance limit stress ofpolished bar specimen by endurance limit stressof weld

• Roughly similar to ASME Code stress indicesC2K2

37-42

Category No. Tested Avg. FSRF

2” Stainless Steel Tests

Code Standard (1 x 1) 2 3.4

2 x 1 5 2.3

1 x 1 Built-up to 2 x 1 2 2.2

Butt Weld 3 2.9

Weld Overlay Repair 2 3.2

PWHT 2 2.5

Last Pass 2 3.1

No Gap 2 3.7

Polished Toe 2 3.6

Smooth As-Welded Toe 2 2.8

Smooth Last Pass Toe 3 2.9

Poor Toe 2 4.8

2” Carbon Steel Tests


2 x 1 5 2.0

1 x 1 Built-up to 2 x 1 2 1.8

Butt Weld 2 2.4

Weld Overlay Repair 2 2.8

PWHT 1 2.5

No Gap 1 3.1

¾” Stainless Steel Tests


2 x 1 3 2.2

PWHT 1 2.6

Last Pass 2 2.3

No Gap 1 2.4

Fatigue Strength Reduction Factors

37-43

Conclusions

• 2 x 1 welds offer significant improvement infatigue strength over Code standard welds

• Replacement of socket welds with butt weldsprovides little improvement

• Important to avoid discontinuities in weld toe toprolong fatigue life

• Polishing the weld toe did not provide any benefit

• Additional pass provided benefit but was subjectto toe initiated cracks

• Post weld heat treatment helps, but may sensitizeaustenitic material

37-44

Conclusions, cont’d

• Eliminating axial gap had no effect

• Previously cycled standard welds built up to 2 x 1were as good as new 2 x 1 welds

• Weld overlay repaired welds were better than newstandard welds

• Weld overlay repairs were more successful withroot cracks than toe cracks

37-45

Fatigue Reactor Components

Documents

Transcript of Fatigue Reactor Components