corrosion issues in light water reactors: stress corrosion cracking (efc51)

Corrosion issues in light water reactors

i

European Federation of Corrosion PublicationsNUMBER 51

Corrosion issues inlight water reactors

Stress corrosion cracking

Edited byD. Féron and J.-M. Olive

Published for the European Federation of Corrosionby Woodhead Publishing and Maney Publishing

on behalf ofThe Institute of Materials, Minerals & Mining

CRC PressBoca Raton Boston New York Washington, DC

W O O D H E A D P U B L I S H I N G L I M I T E DCambridge England

iii

Woodhead Publishing Limited and Maney Publishing Limited on behalf ofThe Institute of Materials, Minerals & Mining

Woodhead Publishing Limited, Abington Hall, AbingtonCambridge CB21 6AH, Englandwww.woodheadpublishing.com

Published in North America by CRC Press LLC, 6000 Broken Sound Parkway, NW,Suite 300, Boca Raton, FL 33487, USA

First published 2007 by Woodhead Publishing Limited and CRC Press LLC© 2007, Institute of Materials, Minerals & MiningThe authors have asserted their moral rights.

This book contains information obtained from authentic and highly regarded sources.Reprinted material is quoted with permission, and sources are indicated. Reasonableefforts have been made to publish reliable data and information, but the authors andthe publishers cannot assume responsibility for the validity of all materials. Neitherthe authors nor the publishers, nor anyone else associated with this publication, shallbe liable for any loss, damage or liability directly or indirectly caused or alleged to becaused by this book.

Neither this book nor any part may be reproduced or transmitted in any form or byany means, electronic or mechanical, including photocopying, microfilming andrecording, or by any information storage or retrieval system, without permission inwriting from Woodhead Publishing Limited.

The consent of Woodhead Publishing Limited does not extend to copying for generaldistribution, for promotion, for creating new works, or for resale. Specific permissionmust be obtained in writing from Woodhead Publishing Limited for such copying.

Trademark notice: Product or corporate names may be trademarks or registered trademarks,and are used only for identification and explanation, without intent to infringe.

British Library Cataloguing in Publication DataA catalogue record for this book is available from the British Library.

Library of Congress Cataloging in Publication DataA catalog record for this book is available from the Library of Congress.

Woodhead Publishing ISBN-13: 978-1-84569-242-1 (book)Woodhead Publishing ISBN-13: 978-1-84569-346-6 (e-book)CRC Press ISBN-1: 978-1-4200-6001-0CRC Press order number: WP6001ISSN 1354-5116

The publishers’ policy is to use permanent paper from mills that operate asustainable forestry policy, and which has been manufactured from pulpwhich is processed using acid-free and elementary chlorine-free practices.Furthermore, the publishers ensure that the text paper and cover board usedhave met acceptable environmental accreditation standards.

Typeset by Replika Press Pvt Ltd, IndiaPrinted by TJ International Limited, Padstow, Cornwall, England

iv

Contents

Contributor contact details xiii

Series introduction xix

Volumes in the EFC series xxi

Preface xxvii

PART I Overviews

1 An overview of materials degradation by stresscorrosion in PWRs 3P. M. SCOTT, Framatome ANP, France

1.1 Introduction 31.2 Nickel base alloys in PWR primary water 51.3 Nickel base alloys on the secondary side of PWR steam

generators 111.4 Stainless steels in PWR primary circuits 151.5 Low alloy steels 191.6 Concluding remarks 211.7 References 22

2 Corrosion potential monitoring in nuclear powerenvironments 25A. MOLANDER, Studsvik Nuclear AB, Sweden

2.1 Introduction 252.2 Measurements in BWRs 262.3 PWR primary system 342.4 PWR secondary systems 362.5 Summary and conclusions 412.6 Acknowledgements 422.7 References 43

v

3 Kinetics of passivation of a nickel-base alloy in hightemperature water 44A. MACHET, A. GALTAYRIES and P. MARCUS, Laboratoire dePhysico-Chimie des Surfaces, France and P. JOLIVET, M. FOUCAULT,P. COMBRADE and P. SCOTT, Framatome ANP, France

3.1 Introduction 443.2 Experimental procedure 443.3 Results 463.4 Discussion 533.5 Conclusion 543.6 References 55

Part II Stress corrosion cracking: susceptibility and initiation

4 IASCC susceptibility under BWR conditions ofwelded 304 and 347 stainless steels 59M.L. CASTAÑO, CIEMAT, Spain, B. VAN DER SCHAAF, NRG, Holland,A. ROTH, Framatome ANP, Germany, C. OHMS, JRC-IE, Holland,D. GAVILLET, PSI, Switzerland and S. VAN DYCK, SCK·CEN, Belgium

4.1 Introduction 594.2 Experimental procedure 604.3 Results and discussion 634.4 Conclusions 684.5 References 69

5 The effect of lead on resistance of low alloy steel toSCC in high temperature water environments 70K. MATOCHA and G. ROžNOVSKÁ, VÍTKOVICE, Czech Republic andV. HANUS, NPP Czech Republic

5.1 Introduction 705.2 Testing material 705.3 Experimental procedure 715.4 Results and discussion 715.5 Conclusions 745.6 Acknowledgement 755.7 References 75

6 Effect of cold work hardening on stress corrosioncracking of stainless steels in primary water ofpressurized water reactors 76O. RAQUET and E. HERMS, CEA/Saclay, France and F. VAILLANT,T. COUVANT and J. M. BOURSIER, EDF/Les Renardières, France

6.1 Introduction 76

Contentsvi

6.2 Experimental procedure 776.3 Results and discussion 816.4 Conclusions 856.5 Acknowledgement 856.6 References 85

7 Effect of strain-path on stress corrosion cracking ofAISI 304L stainless steel in PWR primaryenvironment at 360 ∞C 87T. COUVANT, F. VAILLANT and J.M. BOURSIER, EDF R&D - MMC,France and D. DELAFOSSE, Ecole des Mines de St-Etienne, France

7.1 Introduction 877.2 Experimental procedure 887.3 Results 917.4 Discussion 987.5 Conclusions 1017.6 References 101

8 Dynamic strain ageing of deformed nitrogen-alloyedAISI 316 stainless steels 103U. EHRNSTÉN and A. TOIVONEN, VTT Technical Research Centre ofFinland, Finland and M. IVANCHENKO, V. NEVDACHA, Y. YAGOZINSKYY

and H. HÄNNINEN, Helsinki University of Technology, Finland

8.1 Introduction 1038.2 Experimental procedure 1048.3 Results 1068.4 Discussion of results 1148.5 Conclusions 1178.6 Acknowledgements 1178.7 References 117

9 Laboratory results of stress corrosion cracking ofsteam generator tubes in a ‘complex’ environment –an update 119O. HORNER, E.-M. PAVAGEAU and F. VAILLANT, EDF R&D, Franceand O. DE BOUVIER, EDF Nuclear Engineering Division, France

9.1 Introduction 1199.2 Experimental procedure 1209.3 Results 1209.4 Discussion 1279.5 Conclusions 1289.6 References 129

Contents vii

10 The effect of sulphate and chloride transients on theenvironmentally-assisted cracking behaviour oflow-alloy RPV steels under simulated BWRconditions 130S. RITTER and H.P. SEIFERT, Paul Scherrer Institute (PSI), Switzerland

10.1 Introduction 13010.2 Experimental procedure 13210.3 Results and discussion 13610.4 Summary and conclusions 14610.5 Acknowledgements 14710.6 References 147

11 Transgranular stress-corrosion cracking in austeniticstainless steels at high temperatures 149A. BROZOVA, Nuclear Research Institute, Czech Republic andS LYNCH, Monash University, Australia

11.1 Introduction 14911.2 Experimental procedure 15111.3 Results 15211.4 Discussion 15311.5 References 160

Part III Stress corrosion cracking: propagation

12 Crack growth behaviour of low-alloy steels forpressure boundary components under transientlight water reactor operating conditions – CASTOC,Part 1: BWR/NWC conditions 165S. RITTER and H.P. SEIFERT, Paul Scherrer Institute (PSI),Switzerland, B. DEVRIENT and A. ROTH, Framatome ANP GmbH,Germany, U. EHRNSTÉN, VTT Industrial Systems, Finland,M. ERNESTOVÁ and M. ŽAMBOCH, Nuclear Research Institute (NRI),Czech Republic, J. FÖHL and T. WEISSENBERG, StaatlicheMaterialprüfungsanstalt (MPA), Germany and D. GOMÉZ-BRICEÑO

and J. LAPEÑA, Centro de Investigaciones EnergéticasMedioambientales y Tecnológicas (CIEMAT), Spain

12.1 Introduction 16512.2 Experimental procedures 16612.3 Results and discussion 16912.4 Summary and conclusions 18312.5 Acknowledgements 18412.6 References 184

Contentsviii

13 Crack growth behaviour of low-alloy steels forpressure boundary components under transientlight water reactor operating conditions – CASTOC,Part 2: VVER conditions 186M. ERNESTOVÁ and M. ŽAMBOCH, Nuclear Research Institute (NRI),Czech Republic, B. DEVRIENT and A. ROTH, Framatome ANPGmbH, Germany, U. EHRNSTÉN, VTT Industrial Systems,Finland, J. FÖHL and T. WEISSENBERG, StaatlicheMaterialprüfungsanstalt (MPA), Germany, D. GOMÉZ-BRICEÑO

and J. LAPEÑA, Centro de Investigaciones EnergéticasMedioambientales y Tecnológicas (CIEMAT), Spain and S. RITTER

and H.P. SEIFERT, Paul Scherrer Institute (PSI), Switzerland


14 Effect of yield strength on stress corrosion crackpropagation under PWR and BWR environments ofhardened stainless steels 200M.L. CASTAÑO, M.S. GARCÍA, G. DE DIEGO and D. GOMÉZ-BRICEÑO,CIEMAT, Spain


15 Corrosion fatigue crack growth behaviour of low-alloyRPV steels at different temperatures and loadingfrequencies under BWR/NWC environment 211S. RITTER and H.P. SEIFERT, Paul Scherrer Institute (PSI),Switzerland


Contents ix

16 Effect of cyclic loadings on the stress corrosioncrack growth rate in alloy 600 in PWR primarywater 231C. GUERRE, O. RAQUET and L. DUISABEAU, CEA, France andG. TURLUER, IRSN, France

16.1 Introduction 23116.2 Materials and specimen 23116.3 Experimental procedure 23416.4 Results 23616.5 Discussion 24116.6 Conclusions 24416.7 Acknowledgments 24416.8 References 244

17 Pattern recognition model to estimate intergranularstress corrosion cracking (IGSCC) at crevices andpit sites of 304 SS in BWRs environments 245M. URQUIDI-MACDONALD, Penn State University, USA

17.1 Introduction 24517.2 Objective and procedure 24617.3 Effect of pH 24617.4 Effect of fluid velocity 24717.5 Effect of electrochemical corrosion potential (ECP) 24717.6 Effect of conductivity 24817.7 Effect of sensitization (EPR) 24817.8 Effect of stress intensity 24917.9 Data collection 25017.10 Non-deterministic approach: ANN 25017.11 Results 25217.12 Conclusions 25817.13 References 258

18 Fatigue crack growth in austenitic steel AISI 304L inPWR primary water at room temperature andelevated temperature 260I. NEDBAL, J. KUNZ and J. SIEGL Czech Technical University,Czech Republic

18.1 Introduction 26018.2 Fatigue experiments 26018.3 Macroscopic crack growth rate 26118.4 Fractographic analysis 26318.5 Conclusions 267

Contentsx

18.6 Acknowledgements 26818.7 References 268

Part IV Practical experience

19 Corrosion damage to 18Cr-9Ni-Ti steel after 25 yearsof operation in steam-water environments of theVK-50 reactor 273G.V. FILYAKIN, V.K. SHAMARDIN, YU.D. GONCHARENKO andV.A. KAZAKOV, FSUE ‘SSC RIAR’, Russia

19.1 Introduction 27319.2 Material – operation conditions 27419.3 Experimental results 27519.4 Discussion 28119.5 Conclusions 28719.6 References 288

20 Comprehensive investigation of the corrosion stateof the heat exchanger tubes of steam generators 289K. VARGA, Z. NÉMETH, A. SZABÓ, K. RADÓ, D. ORAVETZ andK. É. MAKÓ, University of Veszprém, Hungary, Z. HOMONNAY,E. KUZMANN and S. STICHLEUTNER, Eötvös Loránd University,Hungary and P. TILKY, J. SCHUNK and G. PATEK, Paks NuclearPower Plant Ltd., Hungary

20.1 Introduction 28920.2 Experimental procedure 29020.3 Results and discussion 29220.4 Conclusions 30020.5 Acknowledgements 30420.6 References 304

21 Stress corrosion cracking of a Kori 1 retired steamgenerator tube 306H. P. KIM, S. S. HWANG, D. J. KIM, J. S. KIM, Y. S. LIM andM. K. JOUNG, Korea Atomic Energy Research Institute, Korea

21.1 Introduction 30621.2 Experimental method 30621.3 Results and discussion 30721.4 Summary 31421.5 Acknowledgement 31421.6 References 314

Contents xi

22 A systematic study of the corrosion effects of theFRAMATOME CORD-UV technology 316K. RADÓ, K. VARGA, Z. NÉMETH, I. VARGA, J. SOMLAI, D. ORAVETZ

and K. É. MAKó, University of Veszprém, Hungary, Z. HOMONNAY

and E. KUZMANN, Eötvös Loránd University, Hungary, J. BORSZÉKI

and P. HALMOS University of Veszprém, Hungary and P. TILKY andJ. SCHUNK, Paks Nuclear Power Plant Ltd., Hungary


Index 328

Contentsxii

Contributor contact details

(* = main contact)

Editors

D. FéronCEA-SaclayDPC/SCCME, bât. 458, P.C. 5091191 Gif-sur-Yvette CedexFrance

E-mail: [email protected]

J.-M. OliveHYDROGENIUS-AIST-KyushuUniversity744 Moto-oka, Nishi-ku819-0395 FukuokaJapan


Chapter 1

P. M. ScottFramatome ANPTour AREVA92084 Paris La Défense CedexFrance


Chapter 2

A. MolanderStudsvik Nuclear ABSE-611 82 NyköpingSweden


Chapter 3

A. Machet, P. Jolivet and P. ScottFramatome ANPTour AREVAF-92084 Paris-la-DéfenseFrance

A. Galtayries* and P. MarcusLaboratoire de Physico-Chimie desSurfacesEcole Nationale Supérieure deChimie de Paris11 rue P. et M. CurieF-75005 ParisFrance


M. Foucault and P. CombradeFramatome ANPCentre TechniqueF-71205 Le CreusotFrance

xiii

Chapter 4

M.L. Castaño*, B. van der Schaaf,A. Roth, C. Ohms, D. Gavillet andS. van DyckCIEMATComplutense 2228040, MadridSpain


Chapter 5

Karel Matocha*, Václav Hanus andGabriela RožnovskáVÍTKOVICE – Research &Development, LtdV. HanusNPP TemelinCzech Republic


Chapter 6

O. Raquet and E. HermsCEA/SaclayDEN/DPC – 91191 Gif sur YvetteCedexFrance

E-mail: [email protected]@cea.fr

T. Couvant*, F. Vaillant andJ. M. BoursierEDF R&D - MMCAvenue des Renardières - Ecuelles77818 Moret-sur-Loing CedexFrance

E-mail: [email protected]@[email protected]

Chapter 7

T. Couvant*, F. Vaillant andJ. M. BoursierEDF R&D - MMCAvenue des Renardières - Ecuelles77818 Moret-sur-Loing CedexFrance


D. DelafosseEcole des Mines de St-Etienne157 Cours Fauriel42023 St-Etienne cedex 2France

Chapter 8

U. Ehrnstén* and A. ToivonenVTT Technical Research Centre ofFinlandIndustrial SystemsKemistintie 3P.O. Box 1704FIN-02044 VTTFinland


M. Ivanchenko, V. Nevdacha,Y. Yagozinskyy and H. HänninenHelsinki University of TechnologyDepartment of MechanicalEngineeringPuumiehenkuja 3P.O. Box 4200FIN-02015 HUTFinland

Contributor contact detailsxiv

Chapter 9

O. Horner*, E-M. Pavageau,F. Vaillant, EDF R&D - MMCAvenue des Renardières - Ecuelles77818 Moret-sur-Loing CedexFrance


O. de BouvierEDF Nuclear Engineering DivisionCentre d’Expertise et d’Inspectiondans les Domaines de la Réalisationet de l’Exploitation93206 Saint DenisFrance


Chapter 10

S. Ritter* and H.P. SeifertPaul Scherrer Institute (PSI)Nuclear Energy and Safety ResearchDepartmentLaboratory for Materials BehaviourCH-5232 Villigen PSISwitzerland


Chapter 11

A. Brozova*Nuclear Research Institute Rez, plc.25068 RezCzech Republic

E-mail: [email protected]@nri.cz

Contributor contact details xv

S. LynchSchool of Physics and MaterialsEngineeringMonash UniversityVictoria 3800Australia

Chapter 12



B. Devrient and A. RothFramatome ANP GmbHErlangenGermany

U. EhrnsténVTT Technical Research Centre ofFinlandIndustrial SystemsKemistintie 3P.O. Box 1704FIN-02044 VTTFinland


M. Ernestová and M. ŽambochNuclear Research Institute Rez, plc.25068 RezCzech Republic


Contributor contact detailsxvi

J. Föhl and T. WeissenbergStaatliche Materialprüfungsanstalt(MPA)StuttgartGermany

D. Goméz-Briceño and J. LapeñaCentro de InvestigacionesEnergéticas Medioambientalesy Tecnológicas(CIEMAT)MadridSpain

Chapter 13

M. Ernestová* and M. ŽambochNuclear Research Institute Rez, plc.25068 RezCzech Republic


B. Devrient and A. RothFramatome ANP GmbHErlangenGermany

U. EhrnsténVTT Technical Research Centre ofFinlandIndustrial SystemsKemistintie 3P.O. Box 1704FIN-02044 VTTFinland


J. Föhl and T. WeissenbergStaatliche Materialprüfungsanstalt(MPA)StuttgartGermany



D. Goméz-Briceño and J. LapeñaCIEMATNuclear Fission DepartmentStructural Materials ProjectAvda. Complutense 2228040 MadridSpain

Chapter 14

M.L. Castaño, M. S. García*,G. de Diego, D. Goméz-BriceñoCIEMATNuclear Fission DepartmentStructural Materials ProjectAvda. Complutense 2228040 MadridSpain


Chapter 15



Chapter 16

C. Guerre* and O. RaquetCEADEN/DPC/SCCME/LECAbât.45891191 Gif-sur-Yvette CedexFrance

E-mail: [email protected]@cea.fr

L. DuisabeauCEADEN/DMN/SEMI/LCMIbât.62591191 Gif-sur-Yvette CedexFrance


G. TurluerIRSNDSR/SAMSBP1792262 Fontenay-aux-roses CedexFrance


Chapter 17

M. Urquidi-MacdonaldPenn State University203 Earth-Engineering ScienceBuildingUniversity ParkPA 16801USA


Chapter 18

I. Nedbal*, J. Kunz and J. SieglCVUT - FJFI - KMATTrojanova 13PRAHA 2CZ 120 00Czech Republic

E-mail: [email protected]@[email protected]@fjfi.cvut.cz

Chapter 19

G. V. Filyakin, V. K. Shamardin*,Y. D. Goncharenko and V. A.Kazakov FSUE ‘SSC RIAR’Dimitrovgrad -10Ulyanovsk region433510Russia.


Chapter 20

K. Varga*, Z. Németh, A. Szabó andK. RadóUniversity of VeszprémDepartment of RadiochemistryH-8201 VeszprémP.O. Box 158Hungary

E-mail: [email protected]@almos.vein.hu

D. Oravetz and K. É. MakóUniversity of VeszprémDepartment of Silicate Chemistryand Materials EngineeringH-8201 VeszprémP.O. Box 158Hungary

Contributor contact details xvii

Z. Homonnay, E. Kuzmann andS. StichleutnerEötvös Loránd UniversityDepartment of Nuclear ChemistryH-1518 BudapestP.O. Box 32Hungary

P. Tilky, J. Schunk and G. PatekPaks Nuclear Power Plant LtdH-7031 PaksP.O.Box 71Hungary

Chapter 21

H. P. Kim, S. S. Hwang, D. J. Kim,J. S. Kim, Y. S. Lim*, M. K. JoungKorea Atomic Energy ResearchInstituteP.O. Box 105YusongTaejon305-600Korea


Chapter 22

Krisztián Radó, K. Varga,Z. Németh, I. Varga* and J. SomlaiUniversity of VeszprémDepartment of RadiochemistryH-8201 VeszprémP.O. Box 158Hungary

E-mail: [email protected]@almos.vein.hu

D. Oravetz and K. É. MakóUniversity of VeszprémDepartment of Silicate Chemistryand Materials EngineeringH-8201 VeszprémP.O. Box 158Hungary

Z. Homonnay and E. KuzmannEötvös Loránd UniversityDepartment of Nuclear ChemistryH-1518 BudapestP.O. Box 32Hungary

J. Borszéki and P. HalmosUniversity of VeszprémDepartment of Analytical ChemistryH-8201 VeszprémP.O. Box 158Hungary

P. Tilky and J. SchunkPaks Nuclear Power Plant LtdH-7031 PaksP.O. Box 71Hungary

Contributor contact detailsxviii

The EFC, incorporated in Belgium, was founded in 1955 with the purpose ofpromoting European co-operation in the fields of research into corrosion andcorrosion prevention.

Membership of the EFC is based upon participation by corrosion societiesand committees in technical Working Parties. Member societies appointdelegates to Working Parties, whose membership is expanded by personalcorresponding membership.

The activities of the Working Parties cover corrosion topics associatedwith inhibition, education, reinforcement in concrete, microbial effects, hotgases and combustion products, environment sensitive fracture, marineenvironments, refineries, surface science, physico-chemical methods ofmeasurement, the nuclear industry, the automotive industry, computer basedinformation systems, coatings, tribo-corrosion and the oil and gas industry.Working Parties and Task Forces on other topics are established as required.

The Working Parties function in various ways, e.g. by preparing reports,organising symposia, conducting intensive courses and producing instructionalmaterial, including films. The activities of Working Parties are co-ordinated,through a Science and Technology Advisory Committee, by the ScientificSecretary. The administration of the EFC is handled by three Secretariats:DECHEMA e.V. in Germany, the Société de Chimie Industrielle in France,and The Institute of Materials, Minerals and Mining in the United Kingdom.These three Secretariats meet at the Board of Administrators of the EFC.There is an annual General Assembly at which delegates from all membersocieties meet to determine and approve EFC policy. News of EFC activities,forthcoming conferences, courses, etc. is published in a range of accreditedcorrosion and certain journals throughout Europe. More detailed descriptionsof activities are given in a Newsletter prepared by the Scientific Secretary.

The output of the EFC takes various forms. Papers on particular topics,for example reviews or results of experimental work, may be published inscientific and technical journals in one or more countries in Europe. Conferenceproceedings are often published by the organisation responsible for theconference.

European Federation of Corrosion (EFC)publications: Series introduction

xix

In 1987 the, then, Institute of Metals was appointed as the official EFCpublisher. Although the arrangement is non-exclusive and other routes forpublication are still available, it is expected that the Working Parties of theEFC will use The Institute of Materials, Minerals and Mining for publicationof reports, proceedings, etc. wherever possible.

The name of The Institute of Metals was changed to The Institute ofMaterials on 1 January 1992 and to The Institute of Materials, Minerals andMining with effect from 26 June 2002. The series is now published byWoodhead Publishing and Maney Publishing on behalf of The Institute ofMaterials, Minerals and Mining.

P. McIntyreEFC Series EditorThe Institute of Materials, Minerals and Mining, London, SW1Y 5DB UK

EFC Secretariats are located at:

Dr B A RickinsonEuropean Federation of Corrosion, The Institute of Materials, Minerals andMining, 1 Carlton House Terrace, London, SW1Y 5DB, UK

Dr J P BergeFédération Européenne de la Corrosion, Société de Chimie Industrielle,28 rue Saint-Dominique, F-75007 Paris, FRANCE

Professor Dr G KreysaEuropäische Föderation Korrosion, DECHEMA e.V., Theodor-Heuss-Allee25, D-60486 Frankfurt, GERMANY

Series introductionxx

1 Corrosion in the nuclear industryPrepared by the Working Party on Nuclear Corrosion

2 Practical corrosion principlesPrepared by the Working Party on Corrosion Education (out of print)

3 General guidelines for corrosion testing of materials for marineapplicationsPrepared by the Working Party on Marine Corrosion

4 Guidelines on electrochemical corrosion measurementsPrepared by the Working Party on Physico-Chemical Methods ofCorrosion Testing

5 Illustrated case histories of marine corrosionPrepared by the Working Party on Marine Corrosion

6 Corrosion education manualPrepared by the Working Party on Corrosion Education

7 Corrosion problems related to nuclear waste disposalPrepared by the Working Party on Nuclear Corrosion

8 Microbial corrosionPrepared by the Working Party on Microbial Corrosion

9 Microbiological degradation of materials – and methods ofprotectionPrepared by the Working Party on Microbial Corrosion

10 Marine corrosion of stainless steels: chlorination and microbialeffectsPrepared by the Working Party on Marine Corrosion

11 Corrosion inhibitorsPrepared by the Working Party on Inhibitors (out of print)

Volumes in the EFC series

xxi

Volumes in the EFC seriesxxii

12 Modifications of passive filmsPrepared by the Working Party on Surface Science and Mechanisms ofCorrosion and Protection

13 Predicting CO2 corrosion in the oil and gas industryPrepared by the Working Party on Corrosion in Oil and Gas Production(out of Print)

14 Guidelines for methods of testing and research in high temperaturecorrosionPrepared by the Working Party on Corrosion by Hot Gases andCombustion Products

15 Microbial corrosion (Proc. 3rd Int. EFC Workshop)Prepared by the Working Party on Microbial Corrosion

16 Guidelines on materials requirements for carbon and low alloysteels for H2S-containing environments in oil and gas productionPrepared by the Working Party on Corrosion in Oil and Gas Production

17 Corrosion resistant alloys for oil and gas production: guidance ongeneral requirements and test methods for H2S ServicePrepared by the Working Party on Corrosion in Oil and Gas Production

18 Stainless steel in concrete: state of the art reportPrepared by the Working Party on Corrosion of Reinforcement inConcrete

19 Sea water corrosion of stainless steels – mechanisms andexperiencesPrepared by the Working Parties on Marine Corrosion and MicrobialCorrosion

20 Organic and inorganic coatings for corrosion prevention – researchand experiencesPapers from EUROCORR ’96

21 Corrosion – deformation interactionsCDI ’96 in conjunction with EUROCORR ’96

22 Aspects on microbially induced corrosionPapers from EUROCORR ’96 and the EFC Working Party on MicrobialCorrosion

23 CO2 corrosion control in oil and gas production – designconsiderationsPrepared by the Working Party on Corrosion in Oil and Gas Production

Volumes in the EFC series xxiii

24 Electrochemical rehabilitation methods for reinforced concretestructures – a state of the art reportPrepared by the Working Party on Corrosion of Reinforcement in Concrete

25 Corrosion of reinforcement in concrete – monitoring, preventionand rehabilitationPapers from EUROCORR ’97

26 Advances in corrosion control and materials in oil and gasproductionPapers from EUROCORR ’97 and EUROCORR ’98

27 Cyclic oxidation of high temperature materialsProceedings of an EFC Workshop, Frankfurt/Main, 1999

28 Electrochemical approach to selected corrosion and corrosioncontrolPapers from 50th ISE Meeting, Pavia, 1999

29 Microbial corrosion (Proc. 4th Int. EFC workshop)Prepared by the Working Party on Microbial Corrosion

30 Survey of literature on crevice corrosion (1979–1998): mechanisms,test methods and results, practical experience, protective measuresand monitoringPrepared by F. P. Ijsseling and the Working Party on Marine Corrosion

31 Corrosion of reinforcement in concrete: corrosion mechanisms andcorrosion protectionPapers from EUROCORR ’99 and the Working Party on Corrosion ofReinforcement in Concrete

32 Guidelines for the compilation of corrosion cost data and for thecalculation of the life cycle cost of corrosion – a working partyreportPrepared by the Working Party on Corrosion in Oil and Gas Production

33 Marine corrosion of stainless steels: testing, selection, experience,protection and monitoringEdited by D. Féron on behalf of Working Party 9 on Marine Corrosion

34 Lifetime modelling of high temperature corrosion processesProceedings of an EFC Workshop 2001. Edited by M. Schütze,W. J. Quadakkers and J. R. Nicholls

35 Corrosion inhibitors for steel in concretePrepared by B. Elsener with support from a Task Group of WorkingParty 11 on Corrosion of Reinforcement in Concrete

Volumes in the EFC seriesxxiv

36 Prediction of long term corrosion behaviour in nuclear wastesystemsEdited by D. Féron and Digby D. Macdonald on behalf of WorkingParty 4 on Nuclear Corrosion

37 Test methods for assessing the susceptibility of prestressing steels tohydrogen induced stress corrosion crackingPrepared by B. Isecke on behalf of Working Party 11 on Corrosion ofSteel in Concrete

38 Corrosion of reinforcement in concrete: mechanisms, monitoring,inhibitors and rehabilitation techniquesEdited by M. Raupach, B. Elsener, R. Polder and J. Mietz on behalf ofWorking Party 11 on Corrosion of Steel in Concrete

39 The use of corrosion inhibitors in oil and gas productionEdited by J. W. Palmer, W. Hedges and J. L. Dawson

40 Control of corrosion in cooling watersEdited by J. D. Harston and F. Ropital

41 Metal dusting, carburisation and nitridationEdited by H. Grabke and M. Schütze

42 Corrosion in refineriesEdited by J. Harston

43 The electrochemistry and characteristics of embeddable referenceelectrodes for concretePrepared by R. Myrdal on behalf of Working Party 11 on Corrosion ofSteel in Concrete

44 The use of electrochemical scanning tunnelling microscopy (EC-STM) in corrosion analysis: reference material and proceduralguidelinesPrepared by R. Lindström, V. Maurice, L. Klein and P. Marcus on behalfof Working Party 6 on Surface Science

45 Local probe techniques for corrosion researchEdited by R. Oltra on behalf of Working Party 8 on Physico-ChemicalMethods of Corrosion Testing

46 Amine unit corrosion in refineriesPrepared by J. D. Harston and F. Ropital on behalf of Working Party 15on Corrosion in the Refinery Industry

47 Novel approaches to the improvement of high temperaturecorrosion resistanceEdited by M. Schütze and W. Quadakkers on behalf of Working Party 3on Corrosion in Hot Gases and Combustion Products

Volumes in the EFC series xxv

48 Corrosion of metallic heritage artefacts: investigation, conservationand prediction of long term behaviourEdited by P. Dillmann, G. Béranger, P. Piccardo and H. Matthiessen onbehalf of Working Party 4 on Nuclear Corrosion

49 Electrochemistry in light water reactors: reference electrodes,measurements, corrosion and tribocorrosion issuesEdited by R.-W. Bosch, D. Féron and J.-P. Celis on behalf of WorkingParty 4 on Nuclear Corrosion

50 Corrosion behaviour and protection of copper and aluminium alloysin seawaterEdited by D. Féron on behalf of Working Party 9 on Marine Corrosion

51 Corrosion issues in light water reactors: stress corrosion crackingEdited by D. Féron and J.-M. Olive on behalf of Working Party 4 onNuclear Corrosion

52 (to come)

53 Standardisation of thermal cycling exposure testingEdited by M. Schütze and M. Malessa

Stress corrosion cracking is one of the major localised corrosion issues inlight water nuclear reactors. The various structural materials used in nuclearpower plants including low alloy steels, stainless steels, nickel base alloys,Zirconium base alloys, had shown cracks developed under the combinedaction of a mechanical stress and more or less polluted water. Both pressurisedwater reactors (PWRs) and boiling water reactors (BWRs) had to face stresscorrosion cracking phenomena even if the water chemistry and the materialsare different. In both cases, stress corrosion cracking is under control eitherby changing the material (use of Alloy 690 in PWRs conditions) or thechemistry (hydrogenated/noble water chemistry in BWRs). These are directfeedback of the research and development programmes where Europeanteams played a significant role. Moreover, nuclear power plants were designedto operate for 30–40 years, and the extension of their life time to 60 years isnow being envisaged which is a longer life time than in many other industries:the materials used for components and circuit pipes which typically rely ontheir passivity in the aqueous environment for corrosion protection, are beingor are intended to be used for significantly longer periods than initiallyplanned. Further research and developments are then needed to predict theirbehaviour, to prepare remedial and repair actions. The obtained data will behelpful for the industry to define a fitness-for-service strategy.

The objective of this EFC book No. 51 is to give an overview of recentdevelopments on stress corrosion cracking performed mainly by Europeanteams, from laboratory investigations to field applications. The book hasbeen divided in four main parts: (i) overviews, (ii) stress corrosion cracking:susceptibility and initiation, (iii) stress corrosion cracking: propagation and(iv) practical experience. The current state-of-the-art is described not onlyfor stress corrosion cracking, but also for two main related subjects: corrosionpotential monitoring and passivation. The book also covers topics rangingfrom initiation and susceptibility to propagation. It includes low alloy steels,stainless steels and nickel base alloys, boiling water and pressurised waterreactor conditions.

The editors would like to thank the authors who presented and wrote

Preface

xxvii

Prefacexxviii

chapters of outstanding scientific and technical content and who respondedenthusiastically to the questions and comments raised by the reviewers.They would also like to thank the members of the Working Party ‘NuclearCorrosion’ (EFC WP4) and of the Working Party 5 ‘Environment SensitiveFracture’ (EFC WP5) of the European Federation of Corrosion who reviewedthese chapters. They would like also to thank Mylene Belgome, Secretary ofthe EFC WP4, who helped the editors in reviewing, correction and secretarialprocedures.

The editors hope that this book will be useful to scientists and engineersin the development of understanding of and resolution of stress corrosioncracking phenomena that they have to face in light water reactors.

Damien FéronChairman of the EFC WP4

andJean-Marc Olive

Chairman of the EFC WP5

Part I

Overviews

1

Corrosion issues in light water reactors2

3

1.1 Introduction

Most of the world’s nuclear steam supply systems for generating electricityare based on water cooled and moderated systems of which the most widespreaddesigns are the Pressurized Water Reactor (PWR) and the Boiling WaterReactor (BWR). Such power production systems are initially designed tooperate for up to 40 years and extension to 60 years is now being envisagedin many cases. It is perhaps often overlooked that the materials of constructionthat typically rely on their passivity in the aqueous environment for corrosionprotection are being, or are intended to be used for significantly longerperiods than in most other industries. Thus, long-term operating experienceis really only now being gained as many such nuclear power plants havereached 20 to 30 years old. It is not surprising, therefore, that as plants haveaged, some serious corrosion problems have been encountered and remediedor repaired, of which one of the most serious is stress corrosion cracking.

The main difference between a PWR and a BWR is that in the former,sub-cooled primary water cools the nuclear fuel and exchanges its heat viasteam generators to create steam to drive a turbine and alternator in a secondarycircuit. In the latter, water is boiled directly by the nuclear fuel and the steamis then separated and dried before passing directly to the turbine. Operatingtemperatures range between about 280 and 320 ∞C except for the PWRprimary circuit pressurizer which operates at 343 ∞C. The fundamentals ofwater reactor chemistry treatment and control are described in reference [1]and a recent overview of PWR water chemistry operating experience inreference [2].

From a corrosion perspective, the operating environments in PWRs andBWRs are radically different as illustrated in Fig. 1.1 on a Pourbaix diagramfor nickel and iron at 300 ∞C. (This figure also indicates the corrosion conditionsfor some common stress corrosion phenomena in both PWRs and BWRs thatwill be described later.) Thus, in PWRs, the water of the primary and secondarycircuits are alkali treated and essentially oxygen-free to ensure minimum

1An overview of materials degradation by

stress corrosion in PWRs

P. M. S C O T T, Framatome ANP, France


general corrosion and corrosion product release rates of the structural materials.PWR primary water also contains about 3 ppm of dissolved hydrogen tosuppress water radiolysis and, as a consequence, primary circuit corrosionpotentials are about 200 mV lower compared to the secondary side, in bothcases being close to the H2/H+ redox potential for virtually all structuralmaterials. In direct cycle BWRs by contrast, extremely pure water is used toensure the lowest possible general corrosion rates. For those BWR plants onNormal Water Chemistry (NWC), radiolytic decomposition of water incombination with removal of non-condensable gases at the turbine condenser

–2.00

–1.50

–1.00

–0.50

0.00

0.50

1.00

1.50V(SHE)

pH14121086420

NiO2

BWRNWC

Ni3O4

O2

H2O

H+

1 ppbH21 atm8.2 ppm

Ni++

Ni

Fe++

Fe

Ni(OH)3

Caustic IGA

PWSCC

FeFe3O4

BWRHWC

NiOPWRSecondary sidePrimary side

Causticcracking

AcidSO4cracking

1.1 Simplified Pourbaix diagram for nickel and iron at 300 ∞Cshowing the principal pH-potential combinations for PWR primaryand secondary water, BWR Normal Water Chemistry (NWC) and BWRHydrogen Water Chemistry (HWC) and the modes of stress corrosioncracking of Alloy 600.

An overview of materials degradation 5

establishes electrochemically significant concentrations of dissolved oxygenand hydrogen peroxide in the recirculating water and consequently corrosionpotentials are around 500 mV more positive than in PWR primary coolantcircuits. In the case of the Hydrogen Water Chemistry (HWC) variant forBWRs, hydrogen at about 10% of the concentration typical of PWR primarycircuits is used to depress corrosion potentials to values intermediate betweenthose of BWR NWC and PWR primary circuits, specifically with the intentionof protecting sensitized and cold worked stainless steels from IntergranularStress Corrosion Cracking (IGSCC), as described briefly later.

The main emphasis of the examples of corrosion related material failuresdescribed hereafter come mainly from the author’s experience of interpretingand modelling stress corrosion of structural materials in PWR systems.However, some examples of BWR experience are also provided for comparisonas well as to illustrate the significantly different experience in many casesbetween the two water cooled nuclear reactor systems.

1.2 Nickel base alloys in PWR primary water

The most severe stress corrosion problem to affect PWRs is IGSCC of Alloy600 in PWR primary water (sometimes called PWSCC for Pressurized WaterStress Corrosion Cracking, as in Fig. 1.1). It has become a generic issuerivalling that of IGSCC of sensitized and/or cold worked stainless steels inBWRs in terms of unanticipated outages and cost of repairs. In addition,high strength nickel base alloy fasteners and springs fabricated fromprecipitation hardened Alloys X750 or 718 are used extensively in PWRprimary circuits and some service failures of these items have also occurred.

Alloy 600, a nickel base alloy containing 14–17% Cr and 6–10% Fe plusvarious minor elements was initially adopted for use in PWRs for steamgenerator tubes because of its excellent resistance to chloride cracking (fromthe secondary side) compared to stainless steel. It was also attractive forprimary circuit components because of the close similarity of its coefficientof thermal expansion to that of the low alloy steel used to fabricate thereactor pressure vessel, pressurizer and steam generator shells.

The susceptibility of Alloy 600 to IGSCC in operational service in PWRprimary water was first revealed in steam generator tubing in the early 1970sin tight U-bends and in rolled, cold-worked transitions in diameter within orjust above the tube sheet [3]. This then became a major cause of steamgenerator tube cracking in the 1980s, and later, premature steam generatorretirement and replacement. IGSCC of pressurizer nozzles and Control RodDrive Mechanism (CRDM) penetrations in the upper heads of PWR reactorpressure vessels followed in the late 1980s and has continued for over adecade [4, 5]. Apparently interdendritic, but in fact intergranular, stresscorrosion cracking of the compatible weld metals Alloys 182 and 82, the


former having a composition similar to that of Alloy 600, has also beenobserved more recently in major primary circuit welds of several PWRplants, often after very long periods in service ranging between 17 and 27years [5]. To these can be added the experience of extensive IGSCC in theg ¢ strengthened analogue of Alloy 600, Alloy X750, which is used for splitpins attaching the CRDM guide tubes to the upper core plate. Even Alloy718, a high strength nickel base alloy containing 17–21% Cr, which is normallyconsidered a very reliable high strength material in PWR primary water use,has occasionally exhibited IGSCC [4].

A common feature of service failures of Alloy 600 and its compatibleweld metals is the presence of very high residual stresses exceeding thenominal yield strength, usually coupled with a roughly machined or heavilyground surface finish. High residual stresses may be induced by rollingoperations as with steam generator tube expansion into the steam generatortube sheet mentioned above or by nearby welding operations as in the caseof CRDM nozzles. If thermal or mechanical plastic straining results in aplastic compression/tension hysteresis cycle, then very high tensile stresseseasily up to 1000 MPa can be generated. By contrast, stress relief (in practiceof attached low alloy steel components) has a very favourable effect onIGSCC resistance and no failures of Alloy 600 components so stress relievedhave occurred in service. The other major factors influencing IGSCCsusceptibility are the material microstructure and the temperature, an activationenergy of 44 kcal/mole being generally admitted for crack initiation. Muchresearch into the metallurgical parameters affecting IGSCC of Alloy 600 andsimilar materials in PWR primary water has shown that chromium carbidesprecipitated on the grain boundaries improve resistance while intragranularcarbides have the opposite effect. Thus material procurement specificationswere developed to ensure that products were delivered with the carbonprecipitated as far as possible as carbides on grain boundaries. Even ‘sensitized’materials, that is those with grain boundary carbides but an adjacent narrowzone of chromium depletion have improved IGSCC resistance in PWR primarywater, in sharp contrast to their very poor resistance in oxygenated BWR NWC.

The generic mechanism IGSCC of the nickel base Alloy 600 and its highstrength analogue, Alloy X750, in PWR primary water has been extensivelystudied. Despite considerable experimental efforts, no consensus exists as tothe nature of the cracking mechanism [1] and both life modelling and remedialmeasures have relied on empirical, phenomenological correlations. In additionto the major influencing parameters of stress, cold work, temperature andcarbide morphology mentioned above, a profound influence of hydrogenpartial pressure (or corrosion potential) has been identified with a worst casecentred on corrosion potentials near the Ni/NiO equilibrium (Fig. 1.1). Themechanism of cracking also does not apparently change between 300 ∞Csub-cooled water and 400 ∞C superheated steam.


It is interesting to note that despite the intense debate concerning themechanism of IGSCC of Alloy 600 in PWR primary water, the most recentmodels incorporate the idea that solid state grain boundary diffusion is ratecontrolling [6]. This is independent of whether the mechanistic model considersthat cracks advance by an oxidation process at the crack tip or due toembrittlement caused by hydrogen discharged by the matching cathodicreaction. Such models provide physically based support for the high value ofthe apparent activation energy, which is typical of solid state grain boundarydiffusion in nickel. Physical support for a fourth power dependency of IGSCCon applied stress comes mainly from studies of grain boundary sliding (itselfdependent on grain boundary diffusion) observed during primary creep inAlloy 600 at temperatures between 325 and 360 ∞C. Grain boundary slidingrates are also observed to depend on grain boundary carbide coverage, greatercoverage being associated with slower grain boundary sliding rates and higherresistance to IGSCC.

Various empirical models have been developed to predict IGSCC of Alloy600 and similar materials in PWR primary circuits until, as sometimes is thecase, replacement becomes unavoidable. The only presently perceived sureremedy for susceptible Alloy 600 components is replacement, usually byAlloy 690 (28–31% Cr and 7–11% Fe) and its compatible weld metals,Alloys 152 and 52, which have proved to be resistant to IGSCC in PWRprimary water both in severe laboratory tests and, to date, after up to 15years in service.

Predictive equations for IGSCC in Alloy 600 were first developed forsteam generator tubes and later extended to pressurizer nozzles and upperhead CRDM penetrations [7, 8]. Both deterministic and probabilistic methodshave been developed. Modelling of Alloy 600 component life is often basedon the following empirical equation:

t CI

ERTf

m = exp

–4s ( ) 1.1

where:tf is the failure time (hours),C is a constant,s is the applied stress (MPa),Im is a material susceptibility index (e.g. Table 1.1),E is the apparent activation energy (44 kcal/mole),R is the universal gas constant (1.987 cal/mole/∞K),T is the absolute temperature (∞K).

Establishing the stress including residual fabrication stress on a givencomponent is not trivial, but well tried and proven approaches based onfinite element stress analysis or experimental techniques applied to mock-


ups are available. Dealing with material variability in susceptibility to IGSCCis not so straightforward, however, and in the case of classification of thesusceptibility of CRDM nozzle cracking in US PWRs, has been ignored.

One method to account for variability in material resistance to IGSCC hasbeen based on a system of material indices, Im, in equation (1.1) [9]. At itssimplest, with no direct information about IGSCC susceptibility of individualheats, the guidelines given in Table 1.1 were adopted. They were based onobservations of minimum times to failure of plant components or, in caseswhere no service failures have been observed, of laboratory specimens inaccelerated tests of representative plant materials. The constant C in equation(1.1) was adjusted so that an index of unity corresponds to a minimumfailure time of 10,000 hours at a temperature of 325 ∞C and an applied stressof 450 MPa, as observed in practice in plant and in laboratory tests. Inaddition, temperature and stress indices were defined relative to the referenceconditions of 325 ∞C and 450 MPa consistent with equation (1.1) as follows:

IE

R Tq = exp– 1 – 1

598ÊË

ˆ¯ ( )È

ÎÍ˘˚ Is

s = 450

4( ) 1.2

Thus: tI I If

m = 10000

◊ ◊q s1.3

In this way, the minimum time to cracking of each generic Alloy 600primary circuit component was assessed after determining its operatingtemperature and stress. The results for different generic components of PWRprimary circuits are shown in Table 1.1. Appropriate surveillance strategieswere then established.

The quantification of variability of Alloy 600 heat susceptibility to IGSCChas been developed further to assess cracking encountered in the upper headCRDM nozzles of French PWRs and extended to other large Alloy 600primary circuit components [8]. Three main types of microstructure wererecognized and related to the carbon content, thermal treatment, especiallythe temperature at the end of forging or rolling operations, and yield strengthafter hot-working:

∑ class A with mainly intergranular carbide precipitates;∑ class B re-crystallized with carbides mainly on a prior grain boundary

network;∑ class C re-crystallized with randomized intragranular carbides as well as

carbides on prior grain boundaries.

These classes were then linked to their IGSCC resistance (i.e. materialsusceptibility index) as determined from operating experience or in acceleratedlaboratory tests of archive materials mainly at 360 ∞C.

Inevitably, such an approach to assessing IGSCC susceptibility reveals

An overview

of materials degradation

9

Table 1.1 Minimum failure times for IGSCC of Alloy 600 components in PWR primary circuits [9]

Ref. Alloy 600 parts Material Stress Temperature Overall Time Observationindex index index index (hours) *

1 Hydraulic expansion 0.2 0.4 1 0.08 80000 NC2 Divider plate 0.5 0.3 0.9 0.14 80000 NC3 Hard rolling on cold leg (Ringhals 2) 2 2.2 0.1 0.44 48000 C4 Pressurizer nozzle (San Onofre 3) 0.5 0.1 3.3 0.17 56000 C5 Nozzle (San Onofre) 0.5 0.9 3.3 1.49 8000 C6 Pressurizer nozzle (ANO1) 0.5 0.3 3.3 0.5 84336 C7 Pressurizer nozzle (Palo Verde 1) 0.5 0.4 3.3 0.66 33320 C8 Nozzle (Palo Verde 2) 0.5 1.5 1.1 0.83 25000 C9 Explosive expansion Fessenheim 1 1 0.4 1 0.4 75000 C

10 Hard rolling on SG hot leg (Gravelines 6) 0.5 2.2 1 1.1 30000 C11 Hydraulic expansion (Doel 2) 2 0.4 1 0.8 30000 C12 Small U-bends Vallourec 2 2.2 0.3 1.32 30000 C13 Small U-bends Westinghouse 2 10 0.3 6 6000 C14 Sensitive hard rolling on SG hot leg 1 2.2 1 2.2 20000 C15 Very sensitive hard rolling on SG hot leg 2 2.2 1 4.4 8000 C16 1300 MW Pressurizer Nozzle 0.5 3.2 3.3 5.28 8000 C17 Mechanical pluggs 0.5 1 1 0.5 40000 C18 French CRDM Nozzles 0.5 1.5 0.5 0.4 80000 C

0.5 1.5 0.5 0.4 26800 C1.1 2.8 0.08 0.24 72909 C1.1 2.5 0.08 0.22 48427 C1.1 2.5 0.08 0.22 58868 C1.1 2.5 0.08 0.2 90777 C

*: NC: non cracked; C: cracked.


significant scatter in the susceptibility indices for different heats about themean associated with each class. This dispersion in material propertiescombined with the dispersion of stress values for any particular componentgives rise to a distribution of failures with time that can be fitted to anappropriate function such as the Weibull distribution. The main advantage ofthe Weibull distribution is that it has a linear transform that can be fitted tothe early failures in order to give a reliable prediction of the increase in stresscorrosion failures with time [7, 8].

Further improvements in estimating the progression of IGSCC failures inAlloy 600 with time as well as the uncertainty in those predictions havecome about by applying the Monte Carlo simulation technique of randomlysampling distributions of the input parameters in equation (1.1) [8]. Anexample of the results using the Monte Carlo approach is shown in Fig. 1.2in the form of a Weibull distribution comparing the results of these simulationswith the inspection results for upper head penetrations in each susceptibilityclass of Alloy 600. When the Monte Carlo simulations are repeated manytimes, the dispersion in the resulting Weibull distribution of failure times isrelatively small because the number of penetrations considered for eachPWR plant series is quite large (over 1000). It can be shown that the progressionof the problem for each design series of PWRs has relatively little inherentuncertainty. On the other hand, if the problem is considered on a plant by

SimulationInspection results

1300 MW Units: Circles 12 to 14

Susceptibility C

Susceptibility B

Susceptibility A

Influence of time and susceptibility99.99

99

90

70

50

30

10

1

0.01

Cu

mu

late

d p

erce

nta

ge

of

crac

ked

pen

etra

tio

ns

Time (hours)20000 50000 80000 200000

1.2 Results of Monte Carlo simulations of IGSCC in upper headCRDM penetrations of 1300 MWe French PWRs and comparison withinspection results for each class of alloy 600 [8].


plant basis, then the statistical uncertainty in predictions of the proportionthat will crack in a given operating time is much greater because there areless than a hundred CRDM penetrations per upper head. For a given upperhead, this statistical uncertainty can be of the order of ±1 to ±5 on the meanprediction, which is easily demonstrated and quantified in a probabilisticsense using the Monte Carlo simulation technique.

Once a stress corrosion crack has been detected by non-destructiveexamination in a PWR primary circuit component, an essential step in thejustification of structural integrity and further operation without repair orreplacement of the affected component is an assessment of crack growthduring the next few operating cycles. Practical approaches to assessing crackgrowth by IGSCC in Alloy 600 components have relied on empiricalmeasurements of crack growth rates as a function of crack tip stress intensity,KI, of the form [10, 11]:

dadt

C KIn = ( – 9)◊ ( in MPa )K mI 1.4

The values of the coefficients C and n for given practical circumstances varybetween different publications but there is a reasonable consensus that theapparent or effective activation energy to be used for adjusting the coefficientC for temperature is ~31 kcal/mole.

1.3 Nickel base alloys on the secondary side of

PWR steam generators

The main type of PWR steam generator in general use is the verticalRecirculating Steam Generator (RSG) with tube bundles, depending on age,made from either mill annealed or thermally treated Alloy 600, thermallytreated Alloy 690, or Alloy 800. Thermal treatment is carried out at ~700 ∞Cwith the objective of precipitating dissolved carbon as chromium carbideson the grain boundaries. Sub-cooled primary water flows through inside ofthe tubes and boils secondary water on the shell side of the tubes. The steamquality of the water-steam mixture entering the steam driers of RSGs istypically 10% and the superheat across the tubes may vary from 10 to 40 ∞C.

Vertical PWR steam generators have experienced a variety of corrosion-induced problems and many have been replaced, usually because of corrosioninduced cracking of mill annealed Alloy 600 steam generator tubes. Only avery few thermally treated tubes have experienced such problems and theyappear to be due to isolated failures of the thermal treatment to ensure anadequate grain boundary carbide microstructure. Some steam generators withmill annealed Alloy 600 tubes have been replaced after only 8 to 12 years ofoperation, which is well short of the usual initially licensed plant operatingperiod of 40 years. New or replacement RSGs are supplied with thermally


treated Alloy 690 or Alloy 800 tubing, which to date have resisted bothprimary and secondary corrosion problems.

Secondary side steam generator tube corrosion problems involving millannealed Alloy 600 include denting, wastage, intergranular attack, IGSCC,and pitting on the outside surfaces of the steam generator tubes [3, 12]. Theevolution of steam generator tube corrosion with time in terms of relativeimportance of each damage mechanism is shown in Fig. 1.3.

Many secondary side corrosion problems with mill annealed Alloy 600tubes have been associated with the interstices between the tubes and thetube supports. The tube support structures for most of the early units weremade of carbon steel, while later units switched to Types 405, 409, and 410ferritic stainless steels for greater corrosion resistance. Tube support structuresof early units used plates with drilled holes, then plates with trefoil or quatrefoilbroached holes, initially with concave lands and then flat lands, or latticebars (egg crates). The objective of the more open tube support designs is toreduce the accumulation of impurities in the interstices by the phenomenonof hideout (see later).

Another corrosion sensitive zone for steam generator tubes has been inand just above the tubesheet. In some of the very early RSG designs, thetubes were only partly expanded just above the seal weld with the lowertubesheet face, thus leaving a crevice between the outside diameter of thetube and the inside diameter of the hole in the tubesheet. Later, the tubeswere expanded into the tubesheet along nearly their full length in order toclose all but the last ~4 mm of the tube to tubesheet crevice. Tube expansionhas been achieved by various methods, mechanical rolling, hydraulic, andexplosive. Each expansion method generated its own characteristic residualstress fields in the tubes that have influenced subsequent stress corrosionbehaviour if, or when, impurities concentrate by hideout either in the tubesheet crevice or under sludge that accumulates on the upper face of thetubesheet.

The underlying cause of all forms of localized corrosion observed on thesecondary side of steam generators is the phenomenon of hideout of lowvolatility solutes in superheated crevices with restricted water circulation.Most impurities entering recirculating steam generators in the feed water arerelatively insoluble in the steam phase and can concentrate by potentiallymany orders of magnitude in occluded superheated crevices by a wick boilingmechanism. Due to the potential variety of impurities entering the steamgenerators, many complex mixtures of concentrated chemicals can beenvisaged. This severely complicates the task of understanding the mechanismsof tube attack and defining adequate remedies. Tube damage such as wastage,pitting and denting has been attributed to the local formation of strong acidsand, evidently, has been largely eliminated by appropriate management ofsecondary water chemistry (Fig. 1.3). By contrast, the steadily rising trend in


IGA/IGSCC (Fig. 1.3) suggests that counter measures have not been completelyeffective, probably because the mechanism has not always been correctlyidentified.

The morphology of IGSCC in mill annealed Alloy 600 steam generatortubing consists of single or multiple major cracks with minor-to-moderateamounts of branching that are essentially 100% intergranular. Experiencesuggests that secondary side IGSCC requires stresses greater than 0.5 yieldin order to propagate rapidly. At lower levels, propagation rates may approachzero, or the corrosion may take the form of intergranular attack (IGA). IGAis the second generally recognized form of secondary side corrosion attackof mill annealed Alloy 600 where there is substantial volumetric attack ofevery grain boundary. Stress is not strictly necessary for IGA to occur, whichdistinguishes it from IGSCC. Nevertheless, the two are clearly closely related.

IGA/IGSCC varies greatly with height of the tube support plate inrecirculating steam generators, being much more prevalent at the lower levelswhere the temperature difference between the primary and secondary fluidsis greatest. This is clearly strong evidence for the importance of impurityhideout, which increases as a function of the available superheat on thesecondary side. Broached tube support plates minimize the extent of thenarrow gap between the tube and its support plate and hence substantiallyreduce the tendency for impurity hideout in such locations.

When the fraction of tubes affected by IGA/IGSCC at tube support plateintersections is plotted as a function of time on Weibull distribution coordinates,it is observed that the slopes of the Weibull plots are rather high, typically

1.3 Worldwide causes of steam generator tube plugging [12].

73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96Year

100

90

80

70

60

50

40

30

20

10

0

Per

cen

tag

e

SCC/IGA (OD)

Fretting

WastageDenting

Other

SCC (ID)

Pitting


between 4 and 9 [7, 13]. This indicates that once IGA/IGSCC starts, itsprogression to other tubes is rather rapid and relatively consistent betweendifferent plants. On the other hand, incubation periods before cracking startsvary considerably. In some cases, IGA/IGSCC has not been observed at all,even on mill annealed Alloy 600 tube bundles after very long periods ofoperation. There is a tendency to attribute this variability between plantsmainly to differences in secondary water chemistry and impurities. However,heat to heat variability in sensitivity of mill annealed Alloy 600 to IGA/IGSCC is very important in this respect and the proportion of very sensitiveheats varies markedly between different plants [4, 13].

Following the retirement of some steam generators with degraded tubing,it has been possible to extract and observe metallographically complete tube/tube support plate intersections [14]. These studies have revealed that thecrevice between the tube and tube support plate is typically plugged at itsentrance and exit with a very low porosity (<10%) solid mixture of magnetiteand silica. In the centre of the crevice, the deposit is mainly magnetite andits porosity is much higher at around 50%. This fouling and plugging of thecrevices between tubes and carbon steel tube support plates with cylindricaltube holes is generally acknowledged to be widespread. Very high forceshave systematically become necessary to extract tubes for destructiveexamination, indicating that the tubes no longer slide easily in the tubesupport plates as intended by the design. Extensive detailed examinationshave also been made of the deposits found on extracted steam generatortubes [15]. On the tube free spans, magnetite deposits are observed overlyinga protective nickel/chromium spinel oxide. Within the tube support platecrevices, thin layers rich in alumino-silicates have been observed on the heattransfer surfaces associated with poorly protective oxide films and the presenceof IGA/IGSCC.

At least seven classes of environmental contaminants have been postulatedat various times to explain the occurrence of IGA/IGSCC of mill annealedAlloy 600 [16, 17]:

∑ high concentrations of sodium hydroxide (NaOH) and/or potassiumhydroxide (KOH);

∑ the products from the reaction of sulphate ions with hydrazine or hydrogen(reactive sulphur-bearing species are postulated);

∑ the products of thermal decomposition of ion exchange resins (sulphatesand organic residuals);

∑ highly concentrated salt solutions at neutral or nearly neutral pH (thesesalt solutions are the natural consequences of condenser leakageconcentrated to high levels by the boiling process in the steam generator);

∑ alkaline carbonates and/or their reaction or hydrolysis products andalumino-silicate deposits (believed to affect the nature of the passivefilm on the alloy surface);


∑ lead contamination;∑ polluted steam.

All these different modes of secondary cracking of Alloy 600 have recentlybeen extensively reviewed [18]. Evaluation and modelling of mill annealedAlloy 600 tube damage by IGA/IGSCC has, nevertheless, traditionally beenbased on the assumed formation of solutions in occluded superheated creviceswith extreme values of pH less than 5 or greater than 10 at temperature. Inpractice, most cases have been attributed to caustic cracking, and extremecare is now taken to restrict as much as possible sodium impurities enteringsteam generators. A few cases of stress corrosion cracking in operating steamgenerators have been clearly caused by lead, sometimes, but not necessarily,with a marked transgranular component to the cracking. Lead induced crackingoccurs across the whole feasible range of pH; it is one of the few types oftube degradation for which there is unequivocal evidence that it occurs in themid-range, moderately alkaline pH targeted by the secondary water chemistrytreatment to minimize general corrosion. Whether the minor amounts of leadfound in practically every steam generator have a critical influence on IGA/IGSCC behaviour of Alloy 600 as distinct from aggravating another underlyingdegradation mechanism remains unresolved [19]. The latter option seemslikely in the view of the widely varying and erratic distributions of leadtraces found on steam generator tubes.

Modification of the crevice environment appears at first sight to be themost straightforward method of preventing or arresting secondary side corrosionalthough implementation can be complicated due to existing deposits impedingaccess of secondary water to the occluded zone. Attempts to modify thecrevice environment have included several factors, such as lowering thetemperature, adding a pH neutralizer or buffering agent such as boric acid,removing the aggressive species by flushing or soaking, and changing theconcentration and/or anion to cation ratio of bulk water contaminants [3].Laboratory studies with model boilers have shown the benefit of several ofthese corrective measures and some have been applied to operating steamgenerators. Minimizing sludge entry and fouling of steam generators alsocontributes to reducing the hideout and concentration of impurities.

1.4 Stainless steels in PWR primary circuits

1.4.1 Primary pressure boundary

Type 304 and 316 austenitic stainless steels are the main materials used forthe pressure boundary piping of PWR primary circuits. The internal surfacesof low alloy steel components are also clad with Type 308/309 stainless steelweld overlays. Operating experience with these stainless steels over manytens of years has generally been excellent. Those stress corrosion failures


that have occurred have in most cases, if not all, been due to internal orexternal surface contamination by chlorides or to out-of-specification chemistryin dead-legs or other occluded volumes where primary water chemistry controlcan be difficult (such as the transient presence of oxygen for significantperiods) [4]. Excessive cold work with the attendant risk of martensite formationin Type 304 stainless steel has also been a contributing factor in some cases.

CRDM housings above the main reactor vessel and associated canopyseals that ensure the leak tightness of threaded joints in the housings are anexample of dead-leg locations that have experienced some stress corrosion,mainly Transgranular Stress Corrosion Cracking (TGSCC) attributed primarilyto chloride contamination. However, sulphate either as a surface impurity onthreaded surfaces or from thermal decomposition of any resin fines that findtheir way accidentally into the hot parts of the primary circuit may alsocontribute since sulphate in combination with oxygen is well known to causestress corrosion in BWRs, albeit usually intergranular.

Although low carbon grades of Types 304 and 316 stainless steels haveoften been used to minimize the risk of sensitization (by grain boundarychromium depletion) of weld heat affected zones, there is no doubt that suchsensitized materials exist in many older PWRs. Nevertheless, practicalexperience shows that de-oxygenated, hydrogenated PWR primary waterdoes not cause IGSCC in such sensitized materials, in contrast to BWRexperience with oxygenated NWC water.

1.4.2 Core internals

Another major use of Type 304 and 316 austenitic stainless steels is for thestructures supporting the nuclear core in the reactor pressure vessel. This isgenerally a bolted structure of horizontal formers and vertical baffle platesthat, because of its proximity to the nuclear fuel, is very heavily neutronirradiated. Unlike the stainless steel components of fuel elements that aredischarged and replaced after a few reactor cycles, the core support structureis intended to remain for the whole reactor life.

Irradiation-Assisted Stress Corrosion Cracking (IASCC) is a term thatdefines cracking phenomena in core structural materials of water cooledand/or moderated nuclear power reactors in which neutron and/or g irradiationcontributes directly to the initiation and propagation of stress corrosion cracking.By implication, in the absence of material damage by fast neutrons and/ormodification of the environmental chemistry by ionizing radiations, crackingeither does not occur or is significantly less severe. Laboratory and field datashow that intergranular stress corrosion cracking of austenitic steels canresult from long-term exposure to high-energy neutron radiation in bothPWR and BWR systems [20, 21].

Neutron irradiation causes atom displacements from their equilibrium


crystallographic locations thereby creating point defects (vacancies andinterstitials) that may either recombine or diffuse to traps such as grainboundaries, dislocations and second phase interfaces. The diffusion andagglomeration of point defects leads to significant changes in microstructureand mechanical properties that alter resistance to stress corrosion cracking.One consequence is a significant hardening of materials due to the formationof many interstitial (Frank) dislocation loops of nanometre dimensions.Hardening saturates after fast neutron doses of about 5 ¥ 1021 n/cm2

(E > 1 MeV) with yield stresses typically in the range 800 to 1100 MPa.Point defect trapping at grain boundaries leads to changes of local elementalcomposition in a zone about ±5 nm wide due to atoms of different elementsexchanging at different rates with the diffusing point defects. Typicallychromium, iron and molybdenum depletion and nickel and silicon enrichmentare observed. More generalized changes in elemental composition may alsobe caused by nuclear transmutation reactions

In the case of the oxygenated coolants of BWRs, the modification of grainboundary composition due to neutron irradiation, particularly chromiumdepletion, has been shown to be an important precursor of IASCC. Neutrondoses exceeding 5 ¥ 1020 n/cm2 (E > 1 MeV) are associated with the occurrenceof IASCC in BWRs, this being the dose required to develop sufficientirradiation-induced chromium depletion at grain boundaries. (Note that themaximum end-of-life dose to the core internals of BWRs is about 8 ¥ 1021

n/cm2 (E > 1 MeV), which is about an order of magnitude less than thatanticipated for PWRs due to the wider water gaps between the fuel andinternals in the former case.) In addition, the formation of oxidizing species,oxygen and hydrogen peroxide, by radiolysis plays an important role in thismanifestation of IASCC in BWRs, which is absent in PWRs due to thehydrogen added to PWR primary water.

Nevertheless, PWR field experience has also shown that intergranularcracking of highly irradiated core components can occur. Type 304 claddingof control rods and cold worked Type 316 core baffle-former bolts of somefirst generation (CP0 series) 900 MWe French PWRs have crackedintergranularly in service [21]. Fast neutron doses of >2 ¥ 1021 n/cm2 (E > 1MeV), strains >0.1 %, and absence of water circulation around the boltshanks in the affected plants have been implicated in the cracking. Clearly,the absence of oxidizing species, oxygen and hydrogen peroxide, is an obviousenvironmental difference compared to BWRs that renders grain boundarychromium depletion of no particular consequence in PWR primary water.However, the considerable hardening that occurs very probably plays animportant role (as indeed it also does in BWRs) [22].

In addition to the phenomena of radiation induced hardening and changesto grain boundary composition, other radiation damage processes could havean important influence on the development of IASCC. Helium bubble


formation, particularly if it occurs on grain boundaries, may have an adverseeffect on intergranular stress corrosion resistance. Irradiation creep can relaxresidual and applied stresses and is independent of temperature in the rangeof interest to light water reactors. Swelling, hitherto only considered ofimportance to fast reactors, could in principle also appear at the high neutrondoses associated with the second half of life of PWRs and affect the loadsapplied to components such as baffle bolts due to differential swelling ratesbetween Type 304 and 316 stainless steels. Thus, although significant advanceshave been made in the understanding of IASCC, much remains to be learned,and it is today a very active field of research in the context of both BWR andPWR plant aging.

1.4.3 High strength fasteners

Precipitation hardened high strength nickel base alloys have already beenmentioned earlier but high strength stainless steels are also widely used inPWRs for components such as bolts, springs and valve stems. The main onesare A286 precipitation hardened austenitic stainless steel, A410 and similarmartensitic stainless steels, and 17-4 PH precipitation hardened martensiticstainless steel. Small numbers of such components have cracked over theyears due to stress corrosion or hydrogen embrittlement and on occasions,loose parts have been generated in the primary circuit.

A286, an austenitic, precipitation hardened, stainless steel is strengthenedby g ¢, Ni3(Ti, Al), formed during aging at 720 ∞C. Its use is favoured wherethe expansion coefficient relative to other austenitic stainless steels is animportant design factor. Unfortunately, it is susceptible to IGSCC in PWRprimary water when loaded at or above the room temperature yield stress,typically 700 MPa. [23, 24]. Cold work prior to aging in combination withthe lower of two commonly used solution annealing temperatures of 900 and980 ∞C has a particularly adverse effect on resistance to IGSCC. Hot headingof bolts, which can create a heat-affected zone between the head and shank,is another known adverse factor. Nevertheless, even if these metallurgicalfactors are optimized, immunity from cracking cannot be assured unless thestresses are maintained below the room temperature yield stress, whichnecessitates strictly controlled bolt loading procedures. There is also strongcircumstantial evidence that superimposed fatigue stresses can lower themean threshold stress for IGSCC even further. Finally, impurities, includingoxygen introduced during plant shut down and possibly consumed onlyslowly in confined crevices, may help crack initiation. Once initiated, cracksgrow relatively easily even in well-controlled PWR primary water.

Components such as valve stems, bolts and tie rods requiring rather highstrength combined with good corrosion resistance in PWR primary circuitwater have been typically fabricated from martensitic stainless steels such as


Type 410 and 17-4 PH. Significant numbers of failures of Type 410 andsimilar martensitic stainless steels have occurred [25]. In most cases, theaffected components have usually entered service too hard due to temperingat too low a temperature. No in-service aging seems to have been involved,however, in the case of Type 410 and similar martensitic stainless steels. Anadditional problem has been caused by galvanic corrosion with graphitecontaining materials in the packing glands of valves, sometimes leading tovalve stem seizure. The preferred replacement material has often been 17-4PH with its higher chromium and molybdenum content no doubt conferringbetter resistance to crevice corrosion.

Service failures of 17-4 PH precipitation hardening stainless steel havealso occurred in PWR primary water [26, 27]. Initially, intergranular crackingby stress corrosion/hydrogen embrittlement was associated with the lowesttemperature aging heat treatment at 480 ∞C (900 ∞F) designated H900. Thisgives a minimum Vickers hardness value of 435HV clearly in excess of thelimit of 350HV commonly observed to limit the risk of hydrogen embrittlement.The 593 ∞C (H1100) aging heat treatment was subsequently widely adoptedand normally yields a hardness value below 350HV. Nevertheless, a smallnumber of failures have continued to occur. The origin of these failuresappears to be thermal aging in service rather than ‘reversible temperembrittlement’ that is related to the diffusion of phosphorus to grain boundariesat aging temperatures generally above 400 ∞C. Thermal aging of precipitationhardened stainless steels such as of 17-4 PH arises from an intra-granulardecomposition of the martensitic matrix into two phases, a which is rich iniron, and a¢ which is chromium rich. A progressive generalized increase inhardness is observed with corresponding increases in strength and ductile/brittle transition temperature and loss of fracture toughness. The hardeningcannot be reversed without re-solution annealing. French studies have shownthat this aging mechanism can occur in 17-4 PH steels on time scales relevantto the design lives of PWRs at temperatures exceeding 250 ∞C and quantitativemodels for component assessment have been developed [26]. Intergranularfailures have been associated with hardness values following in-service agingthat have significantly exceeded 350HV and have also been apparentlyaggravated by impurities coming from valve packing gland materials.

1.5 Low alloy steels

1.5.1 Secondary circuit components

A small number of potentially serious failures caused by transgranular stresscorrosion/corrosion fatigue have occurred in low alloy steel steam generatorshells and carbon steel feedwater piping that are directly exposed to secondarywater. The combination of fabrication and operational factors necessary for


such cracking to occur in carbon and low alloy steels in steam-raising planthas ensured that it has in reality been highly plant specific. Extensivecircumferential cracking of the upper shell to cone girth welds of all theIndian Point 3 steam generators was found in 1982 following a steam leakthrough one of more than a hundred circumferential cracks [28]. Subsequently,the steam generator shells of six other plants located in the United States andEurope were also observed to be cracked in the same location. In somecases, cracking recurred after local repairs had been made by contour grinding.The steam generator shell cracking was caused by an environmentally assistedcracking mechanism and has been variously called ‘corrosion fatigue’, ‘stresscorrosion cracking’, or ‘strain-induced corrosion cracking’. In fact, the lastterm seems most appropriate since it recognizes that although the crackingis environmentally controlled, a dynamic strain is necessary to maintaincrack propagation [29]. Consequently, crack extension tends to occurintermittently alternating with pitting at the crack tip during quiescent periods.

This environmentally assisted cracking (EAC) mechanism observed forsteam generator shell materials is well known and characterized both forbainitic low alloy steels as well as for ferritic-pearlitic carbon manganesesteels used extensively in both conventional and nuclear steam-raising plant[29–31]. In addition to the dynamic loading requirement usually caused bylarge thermal transients, cracking has been associated in practice with highresidual welding stresses due to poor or non-existent stress relief. The worstaffected plants had been weld repaired during fabrication of the final closureweld. In one case, the girth weld had to be completely remade and stressrelieved at a higher temperature of 607 ∞C compared to 538 ∞C originally.Water chemistry transients, particularly oxygen ingress, occurring at thesame time as dynamic loading have also been strongly linked to the observedcracking. The effect of oxygen was observed to be greatly exaggerated ifcopper corrosion products (e.g. from brass condenser tubes) were also present[28]. The only metallurgical factors that appeared to play a role were thesulphur impurity content of the steel in the form of manganese sulphide,where the risk of cracking was greater the higher the sulphur content, andpossibly also the free nitrogen content via the phenomenon of strain aging[31]. Practical resolution of steam generator shell cracking has been mainlyachieved by contour grinding of existing cracks and by ensuring that auxiliaryfeed water is properly de-oxygenated prior to use, particularly during plantstart-up.

In addition to these reported incidents of steam generator shell cracking,a very large technical literature exists concerning EAC of carbon and lowalloy steels in both nuclear and conventional steam-raising plant [29–31].The observed cracking is usually transgranular cleavage-like in appearancealthough can occasionally be intergranular without any obvious involvementof other chemical pollutants.


1.5.2 High strength fasteners

High strength martensitic and maraging steels are used in many externalfastener applications in nuclear power plants as well as for some internalfasteners in PWR secondary circuits. A significant number of corrosion relatedfailures of external fasteners used for support bolting and pressure boundaryflanges have occurred [32]. Failures of low alloy (AISI 4340 and 4140) andmaraging steel support bolting have been attributed mainly to hydrogenembrittlement. Steels with ultra high yield strengths greater than 1000 MPahave failed due to a combination of too high applied stresses and humid orwet environments collecting around the bases of components. Pitting oftenprecedes cracking in such cases. Steels with lower yield strengths have alsofailed due to poor heat treatment or material variability. Hydrogen crackingis usually avoided by specifying an upper bound strength limit (normallydefined by a hardness level acceptance criterion of <350HV).

A second category of bolt failures has concerned the integrity of theprimary pressure boundary at locations such as flanges of manway coversand valves. Most of these incidents have been caused by erosion-corrosionin PWR primary water leaks. A small number of failures among this categoryof bolt have been associated, however, with environment assisted crackingrather than wastage [32]. The ferritic bolting steels involved were not out ofspecification but had been in contact with molybdenum disulfide lubricants.It has been postulated that the lubricant dissociated on contact with hot waterto yield hydrogen sulphide, which is a severe hydrogen embrittling agent forferritic steels. Prevention of this type of failure therefore includes avoidingleaks at flanges by improved gasket design and eliminating the use of sulphidecontaining lubricants.

1.6 Concluding remarks

The aging of light water cooled and moderated nuclear power plants such asBWRs and PWRs has been accompanied by many cases of corrosion-relatedmaterial failures, particularly stress corrosion cracking. A small proportionof these material failures have arisen because existing knowledge was notapplied and have then been remedied by tightening quality assuranceprocedures. Others were not predictable in advance because of the very longoperating times involved and have sometimes proved to be widespread andgeneric. These have been carefully studied and effective predictive modelshave been developed in parallel with practical and economic repair strategies.This is an essential continuing process that will increase in importance asthese power plants enter the second half of their original design lives. Inmany cases, plant life extension beyond the original design life is a practicaland economic option but continued vigilance for unexpected long-term agingand corrosion processes will be essential.


1.7 References

1. P. Cohen, Water Coolant Technology of Power Reactors, 2nd printing, AmericanNuclear Society (1980).

2. F. Nordmann, A. Stutzmann, J.-L. Bretelle, ‘Overview of PWR chemistry options’,Proceedings of Chimie 2002, SFEN (2002) Paper 94.

3. S. J. Green, ‘Steam generator failure or degradation’, Corrosion in the NuclearPower Industry, ASM Handbook, Volume 13, ‘Corrosion’, (1987) 937–945.

4. P. M. Scott, ‘Stress corrosion cracking in Pressurized Water Reactors – Interpretation,modeling and remedies’, Corrosion, 56 (2000) 771–782.

5. W. Bamford, J. Hall, ‘A review of Alloy 600 cracking in operating nuclear plants:historical experience and future trends’, Proceedings of 11th Int. Conference onEnvironmental Degradation of Materials in Nuclear Power Systems – Water Reactors,Stevenson, American Nuclear Society (2003) 1071–1079.

6. P. M. Scott, P. Combrade, ‘On the mechanism of stress corrosion crack initiation andgrowth in Alloy 600 exposed to PWR primary water’, Proceedings of 11th InternationalConference on Environmental Degradation of Materials in Nuclear Power Systems– Water Reactors, Stevenson, Washington, American Nuclear Society (2003) 29–35.

7. R. W. Staehle, J. A. Gorman, K. D. Stavropoulos, C. S. Welty, ‘Application ofstatistical distributions to characterizing and predicting corrosion of tubing in steamgenerators of Pressurized Water Reactors’, Proceedings of Life Prediction of CorrodibleStructures, ed. R. N. Parkins, NACE International (1994) 1374–1439.

8. P. Scott, C. Benhamou, ‘An overview of recent observations and interpretation ofIGSCC in nickel base alloys in PWR primary water’, Proceedings of 10th Int.Symposium on Environmental Degradation of Materials in Nuclear Power Systems– Water Reactors, Lake Tahoe, NACE International (2001).

9. C. Amzallag, S. Le Hong, C. Pagès, A. Gelpi, ‘Stress corrosion life assessment ofAlloy 600 components’, Proceedings of 9th International Symposium on EnvironmentalDegradation of Materials in Nuclear Power Systems – Water Reactors, NewportBeach, CA, The Metallurgical Society (1999) 243–250.

10. C. Amzallag, F. Vaillant, ‘Stress corrosion crack propagation rates in reactor vesselhead penetrations in Alloy 600’, Proceedings of 9th International Symposium onEnvironmental Degradation of Materials in Nuclear Power Systems – Water Reactors,Newport Beach, CA, The Metallurgical Society (1999) 235–241.

11. G. A. White, J. Hickling, L. K. Mathews, ‘Crack growth rates for evaluating PWSCCof thick-walled Alloy 600 material’, Proceedings of 11th International Conferenceon Environmental Degradation of Materials in Nuclear Power Systems – WaterReactors, Stevenson, Washington, American Nuclear Society (2003) 166–179.

12. D. R. Diercks, W. J. Shack, J. Muscara, ‘Overview of steam generator tube degradationand integrity issues’, Nuclear Engineering and Design, 194 (1999) 19–30.

13. P.M. Scott, ‘A discussion of mechanisms and modeling of secondary side corrosioncracking in PWR steam generators’, Proceedings of Chemistry and Electrochemistryof Corrosion and Stress Corrosion Cracking: A Symposium honoring the Contributionsof R. W. Staehle, Ed R. H. Jones, The Metallurgical Society (2001) 107–122.

14. L. Albertin, F. Cattant, A. Baum, P. Kuchirka, ‘Characterization of deposits inDampierre-1 steam generator support plate crevices’, Proceedings of 7th InternationalSymposium on Environmental Degradation in Nuclear Power Systems – WaterReactors, Breckenridge, Colorado, NACE International (1995) 399–408.

15. B. Sala, P. Combrade, A. Gelpi, M. Dupin, ‘The use of tube examinations and


laboratory simulations to improve the knowledge of local environments and surfacereactions in TSPs’, Control of Corrosion on the Secondary Side of Steam Generators,Eds. R. W. Staehle, J. A. Gorman and A. R. McIlree, NACE International (1996)483–497.

16. S.J. Green, ‘Thermal, hydraulic and corrosion aspects of PWR steam generatorproblems’, Heat Transfer Engineering, 9 (1988) 19–68.

17. Q. T. Tran, P. M. Scott, F. Vaillant, ‘IGA/IGSCC of Alloy 600 in complex mixturesof impurities’, Proceedings of 10th Int. Symposium on Environmental Degradationof Materials in Nuclear Power Systems – Water Reactors, Lake Tahoe, NACEInternational (2001).

18. R. Staehle, J. A. Gorman, ‘Quantitative assessment of submodes of stress corrosioncracking on the secondary side of steam generator tubing in Pressurized WaterReactors: Parts 1, 2 and 3’, Corrosion, 59 (2003) 931–994, 60 (2004) 5–63, and 60(2004) 115–180.

19. L. E. Thomas, V. Y. Gertzman, S. M. Bruemmer, ‘Crack-tip microstructures andimpurities in stress-corrosion-cracked Alloy 600 from recirculating and once-throughsteam generators’ Proceedings of 10th Int. Symposium on Environmental Degradationof Materials in Nuclear Power Systems – Water Reactors, Lake Tahoe, NACEInternational (2001).

20. R. M. Horn, G. M. Gordon, F. P. Ford, R. L. Cowan, ‘Experience and assessment ofstress corrosion cracking in L-grade stainless steel in BWR internals’, NuclearEngineering and Design, 174 (1997) 313–325.

21. P. M. Scott, M-C. Meunier, D. Deydier, S. Silvestre, A. Trenty, ‘An analysis ofbaffle/former bolt cracking in French PWRs’, ASTM STP 1401 EnvironmentallyAssisted Cracking: Predictive methods for Risk Assessment and Evaluation of Materials,Equipment and Structures, Ed. R. D. Kane, ASTM (2000) 210–223.

22. S. M. Bruemmer, E. P. Simonen, P. M. Scott, P. L. Andresen, G. S. Was, J. L. Nelson,‘Radiation-induced material changes and susceptibility to Intergranular failure oflight-water-reactor core internals’, Journal of Nuclear Materials, 274 (1999) 299–314.

23. US NRC Information Notice, ‘Stress corrosion cracking of reactor coolant pumpbolts’ 90 – 68 (1990) and supplement 1 (1994).

24. J. B. Hall, S. Fyfitch, K. E. Moore, ‘Laboratory and operating experience with AlloyA286 and Alloy X750 RV internals bolting stress corrosion cracking’, Proceedingsof 11th International Conference on Environmental Degradation of Materials inNuclear Power Systems – Water Reactors, Stevenson, Washington, American NuclearSociety (2003) 208–215.

25. US NRC Information Notice, ‘Valve stem corrosion failures’, 85–59, (1985).26. B. Yrieix, M. Guttmann, ‘Aging between 300 and 450 ∞C of wrought martensitic 13–

17 wt% Cr stainless steels’, Materials Science and Technology, 9 (1993) 125–113.27. H. Xu, S. Fyfitch, ‘Aging embrittlement modeling of Type 17-4 PH at LWR

temperature’, Proceedings of the 10th International Conference on EnvironmentalDegradation of Materials in Nuclear Power Systems – Water Reactors, Lake Tahoe,CA, NACE International (2001).

28. W. J. Bamford, G. V. Rao, J. L. Houtman, ‘Investigation of service-induced degradationof steam generator shell materials’, Proceedings of 5th International Symposium onEnvironmental Degradation of Materials in Nuclear Power Systems – Water Reactors,Monterey, American Nuclear Society (1992) 588–595.

29. J. Hickling, D. Blind, ‘Strain-induced corrosion cracking of low-alloy steels in LWR


systems – Case histories and identification of conditions leading to susceptibility’,Nuclear Engineering and Design, 91 (1986) 305–330.

30. P. M. Scott, D. R. Tice, ‘Stress corrosion in low steels’, Nuclear Engineering andDesign, 119 (1990) 399–413.

31. H. P. Seifert, S. Ritter, J. Hickling, ‘Environmentally-assisted cracking of low-alloyRPV and piping steels under LWR conditions’, Proceedings of 11th InternationalConference on Environmental Degradation of Materials in Nuclear Power Systems– Water Reactors, Stevenson, Washington, American Nuclear Society (2003) 73–88.

32. C. J. Czajkowski, ‘Corrosion and stress corrosion cracking of bolting materials inlight water reactors’, Proceedings of 1st Int. Symposium on Environmental Degradationof Materials in Nuclear Power Systems – Water Reactors, Myrtle Beach, NACE(1984) 192–208.

25

2.1 Introduction

Corrosion potential measurements were introduced in nuclear power plantswhen IGSCC in sensitized Type 304 stainless steel was the dominating typeof failure in BWRs. The corrosion potential (often called ECP) was found tobe the prime parameter controlling IGSCC. Hydrogen water chemistry (HWC)was developed and introduced as a remedy. Electrochemical measurementswere introduced during so-called HWC mini-tests, which were performed todetermine the necessary injection rate of hydrogen to reduce the corrosionpotential below the critical potential for IGSCC. ECP monitoring equipmentwas first installed in remote sampling systems and the desired response tohydrogen was easily obtained. With time, the effect of hydrogen peroxidewas understood and monitoring locations with short transport time or in-situmeasurements were implemented. Corrosion potential measurementsdeveloped to a routine technology necessary for safe reactor operation and atool, not only for the chemist, but also for the control room personnel. Inspite of the cost of the measuring equipment, the measurements initiallyrequested by the authorities were also shown to be economically attractivesince such measurements permitted the hydrogen injection rate to beoptimized.

Electrochemical measurements have also been performed in PWR systemsand mainly the feedwater system on the secondary side of PWRs. Themeasurements performed so far have shown that electrochemical measurementsare very sensitive tools to detect and follow oxygen transients in the feedwatersystem. Also, determinations of the minimum hydrazine dosage to the feedwaterhave been performed. However, PWR secondary side monitoring has not yetbeen utilized to the same level as BWR hydrogen water chemistry surveillance.The future use of corrosion potential monitoring in PWR secondary systemis considered to be dependent on the direction of secondary side chemistrydevelopments.

Both in a BWR on hydrogen water chemistry and in the PWR secondary

2Corrosion potential monitoring in nuclear

power environments

A. M O L A N D E R, Studsvik Nuclear AB, Sweden


system the corrosion potentials show a large variation between differentsystem parts. To postulate the material behavior at different locations thelocal chemical and electrochemical conditions must be known. Thus, modelingof chemical and electrochemical conditions along reactor systems is important,as actual measurements cannot be performed at every point of interest. However,different relations between the corrosion potential of stainless steel and theoxidant concentration have been published and only recently an improvedunderstanding of the electrochemical reactions and other conditions thatdetermine the corrosion potential in BWR systems has been reached. Also,improved modeling for PWR secondary systems has recently been performedand published.

2.2 Measurements in BWRs

2.2.1 HWC monitoring: From autoclaves to in-pipemeasurements

When HWC was introduced and ECP measurements were performed withthe prime goal to follow the corrosion potential due to the hydrogen addition,monitoring was initially performed in autoclaves often installed remote fromthe reactor system. With time, it was realized that such measurements did notreflect the system conditions due to low flow rate in the autoclave anddecomposition of hydrogen peroxide and consumption of oxygen in thesampling line [1]. Consequently measured corrosion potentials were low.With time this was understood and better techniques were developed asdescribed below, but first an example of very early data is given.

Figure 2.1 shows an experimental set-up used during one of the very firsthydrogen water chemistry experiments. It was performed in the Oskarshamn2 BWR. A remote low flow autoclave was used connected with small diameterpiping, Decomposition of hydrogen, oxygen consumption and consequentlylower corrosion potentials will be measured in the autoclave compared to thereactor system itself. The results of the mini-test are given in Fig. 2.2. Whenhydrogen addition starts, oxygen contents drops quickly and a correspondingdecrease of the corrosion potential is obtained. The hydrogen addition wasno higher than 60 ppb in the reactor water but a total suppression of thecorrosion potential was obtained, that is down to the hydrogen line.

With time the sampling effects were understood and there was a developmentaway from autoclaves to in-pipe measurements or measurements inside theactual reactor component. In Sweden the first in-pipe electrode was installedin 1989 in the Barsebäck 1 BWR [2]. An example illustrating the effect oftransport time on ECP results was obtained during an experiment in theRinghals 1 BWR [3] performed to validate the in-pipe measurements and tocompare different measuring locations. Several monitoring points were used.

Corrosion potential m

onitoring in nuclear power environm

ents27

2333 600

Ref 1 Ref 2

System 331

50 g/s

250∞C

autoklav

Elektrokemi-Kylare

X-750 2343

Ref 3Pt

352

4 l/h

11/h

H2O2

H2

O2

N2SR

Torkrör

Flodes-matare

GasFlodes-matare

GasFlodes-matare

Flodes-matareKajon-

bytare

Kond mitning

TalliumMnO2Jontytare

AB

HerschMark 2Syre-

analysator

Gaskromato-

graf

Kond mat-brygga

Temp

1 l/hFlodes-mature

Torkror

2.1 Schematic experimental setup from an early HWC experiment in the Oskarshamn 2 BWR.


ECP measurements were performed at the five different locations shown inFig. 2.3.

The results, shown in Fig. 2.4, demonstrated the gradual decrease of thecorrosion potential along the flow direction. During conditions with a lowhydrogen addition the in-pipe potential is about 200 mV that is the same asfor NWC. In a high flow autoclave close to the reactor piping a potential ofabout 0 mV was measured and for more remote systems even lower potentialswere obtained. After such and similar efforts it was understood that HWCcorrosion potentials must be measured in-pipe. In Sweden all plants onHWC rely on such measurements performed in the PLR-system for theirHWC surveillance.

For the measurements in-pipe, platinum electrodes are installed in theprimary loop recirculation piping less than 1 s transport time from thedowncomer in the reactor pressure vessel, see Figs 2.5 and 2.6 (location B).As a back-up, electrodes in the water clean-up system are also used (locationC in Fig. 2.6). This position is 10–20 s from the RPV but the pipe dimensionis still 200 mm or more. Side stream autoclaves are no longer used for HWCmonitoring.

By the in-pipe technique the errors introduced by sampling effects wereavoided. Today these effects can reasonably be modeled, but at the time of

Hydrogen

Oxygen

Ecor Type 304

50

10

200

100

–100

–500

mV

SH

Ep

pb

O2

pp

b H

2

1200 2400 1200 2400 1200 2400 1200 2400 120001-07-06 01-07-07 01-07-08 01-07-09 01-07-10

2.2 Examples of results from an experiment in the Oskarshamn 2 BWR.

Corrosion potential m

onitoring in nuclear power environm

ents29

Flow direction when pump 2is in service

Flow direction when pump 1is in service

Pump 2

Pump 1

313 P4

Location 5

Transport time: 3–5 minFlow rate: 10–40 g/sVolume: 5 l

Location 4

Transport time: 50 s (45 s)Flow rate: 2 kg/sVolume: 30 l

Location 3

Transport time: 45 s (50 s)Flow rate: 2 kg/sVolume: 30 l

Transport times within brackets refer to conditions when pump 2 is in service

Location 2System 335

Transport time: 30 s (95 s)Flow rate: 2 kg/sø24 mm

Location 1System 321

Transport time: 22 sFlow rate: 105 kg/sø300 mm

2.3 Illustration of the locations of consecutive ECP measuring points in Ringhals 1 BWR, see [3].


the measurements it was an important demonstration of the need to improvein-plant monitoring and evidence for the need for in-pipe monitoring techniques.

When the variations of the corrosion potential at different monitoringlocations were demonstrated [4], certain modeling efforts were performed totry to model and predict these variations. The models were based on hydrogenperoxide decomposition rates and usually the mixed potential model. Withtime, new data were also collected and new phenomena were detected. Fromcompilations of HWC data from several reactor cycles it was noted that largedeviations between different years, with respect to the hydrogen demand,were sometimes obtained.

Location 1Location 2Location 3Location 4

90-7-2190-7-2090-7-1990-7-18Date (YY-M-DD)

Eco

rr (

mV,

SH

E)

–800

–600

–400

–200

0

200

400

2.4 Comparison of the results from four consecutive ECP measuringpoints in the Ringhals 1 BWR, see [3].

2.5 Schematic illustration of an in-pipe reference electrode.

Corrosion potential monitoring in nuclear power environments 31

For example Fig. 2.7 shows the hydrogen injection rate needed to obtaina potential of –300 mV SHE in the PLR system of the Oskarshamn 2 BWR.There is a large variation with time. Values vary generally between 0.5 and1.5 ppm in the feedwater, that is a factor of three. The wide variation of thenecessary hydrogen injection rate was not predicted by the existing models.

A BD

Bottom drain

E

In-core electrode

C

Feedwater system

To waterclean-upsystem

Residualheat removalsystem

Mainrecirculationsystem

2.6 Schematic illustration of monitoring points in a BWR.

H2

312 R901312 R902

2001-07-011999-07-011997-07-011995-07-011993-07-011991-07-01

2.0

1.5

1.0

0.5

0.0

H2-

do

s (p

pm

)

2.7 Hydrogen injection rate required for constant ECP (–300 mV SHE)in the Swedish BWR Oskarshamn 2.


However, it was found that a significant improvement was achieved if theactual power level along the core periphery was used for the radiolysiscalculations instead of a constant power level. As most of the recombinationis occurring in the downcomer, the power level at the core periphery is veryimportant. To reasonably predict the variation during a cycle, and betweencycles, the ECP/HWC model must be regularly and accurately updated fromfuel management codes.

Not only the radiolysis model but also the ECP model have been improvedover the years. An ECP model should work for both lab and in-plant conditions.In Fig. 2.8 the upper curve shows ECP vs. hydrogen peroxide concentration.This curve is a summary of results obtained both in-plant and in our lab [5].In Fig. 2.9 the results from Fig. 2.8 indicated by the squares are compared totwo other models [6], the Swedish BwrChem model and the GE semi-empiricalmodel. Neither of the models reproduces the experimental results [7].

In the Studsvik ECP model the database from in-plant monitoring as wellas lab monitoring is used. The model provides a mechanistic background tothe measured ECPs. The results of the model calculations come out as Evansdiagrams as shown in Fig. 2.10 for a fairly high H2O2 concentration. Line 1shows the hydrogen peroxide oxidation to oxygen and line 2 shows thehydrogen peroxide to reduction to water. Line 3 is the passive current ofstainless steel. The calculated ECP is shown as the dashed line at the intersectionof the two curves. A concentration change will only cause a vertical shift ofthe two curves leaving the ECP unaffected. Thus this Evans diagrams explains

Laboratory data (oxygen)In-plant dataThis work, B1This work, lab (oxygen)This work, lab (hydrogen peroxide)

100001000100101Dissolved oxidant concentration (ppb)

–700

–600

–500

–400

–300

–200

–100

0

100

200

300

Co

rro

sio

n p

ote

nti

al (

mV

SH

E)

2.8 Compilation of experimental ECP results from laboratory andplants [5]. The upper curve shows environments with hydrogenperoxide.


BC_REFGE-semiemp.SNAB-decr.SNAB-incr.

NWC Lab test – H2 = 0 ppb, 0.08 m/s, Peroxide injection

1000010001001010.1Conc of H2O2 (ppb)

–600

–500

–400

–300

–200

–100

0

100

200

300

EC

P (

mV

)

2.9 Comparison of Studsvik experimental results (squares) with twoECP-models (lines).

800 10006004002000–200–400–600–800–1000ECP (mV SHE)

3

2

1

1.E-02

1.E-03

1.E-04

1.E-05

1.E-06

1.E-07

1.E-08

1.E-09

1.E-10

1.E-11

1.E-12

i (A

/cm

2 )

2.10 Schematic Evans diagram for a BWR environment with 100 ppbof hydrogen peroxide. The unstable H2O2 disproportions into oxygenand water with electrode as catalyst. Both oxidation and reductionare first order reactions fi ECP independent of H2O2 concentration.


the plateau regime of the ECP in water with hydrogen peroxide. The abruptdrop to a low potential is also well reproduced.

The BWR experience can be summarized as follows:

∑ We now have a good understanding the basic mechanisms of ECP, HWCand radiolysis models.

∑ In spite of the improvements correct HWC modeling is very difficult.The modeling results are very sensitive to the input data.

∑ ECP measurements will still be needed for HWC supervision. However,ECP modeling provides a reliable estimation of the ECP upstream anddownstream of the measuring position which sometimes is very important.

2.3 PWR primary system

ECP measurements in a PWR primary system have been performed in theRinghals 4 reactor which is a Westinghouse PWR. The equipment was installedin a sampling line from the hot leg, see Fig. 2.11. During steady operation novariations of the potentials were noticed. Working electrodes of platinum,stainless steel, nickel-base alloys and carbon steel showed the same potential,that is the hydrogen equilibrium potential, see Fig. 2.12.

During start-up and shut down the potentials showed large changesdepending on the chemistry, see Fig. 2.13. For start-up the chemical degassingis easily followed and during shut down the shift to oxidizing conditions anda hydrogen peroxide dosage performed for crud removal is identified, seereferences [8] and [9].

The conclusions of the PWR primary side measurements were that:

∑ Corrosion potentials were found to be stable and all materials fell on thehydrogen line. pH could be monitored.

Autoclave

Water chemistry analysisTo volume control tank and

charging/local sampling tank

From reactor coolantsystem hot leg

Decay coil

Containment wall

2.11 The monitoring point in the Ringhals 4 primary system.


∑ During start-up and shut-down large variations with chemistry occurred.∑ The benefits of electrochemical monitoring were limited during present

water chemistry (high hydrogen). Equipment for ‘PWR mini test’ wasschematically designed and suggested for low hydrogen water chemistry.

∑ External silver chloride electrode developed for the work was demonstratedto be reliable over one reactor year. Comparisons with electrodes ofdifferent types were performed.

Date

700

300

250

200

150

100

50

0

Tem

per

atu

re ∞

C

TemperaturepH electrode

Type 304Alloy 600R 533-BPlatinum

500

300

100

–100

–300

–500

–700

–900

Co

rro

sio

n p

ote

nti

al (

mV

SH

E)

87-07-28 87-08-0287-07-2387-07-1887-07-13

2.12 Measured ECP in the primary sytem of the Ringhals 4 PWRunder steady operation.

300

250

200

150

100

50

0

Tem

per

atu

re ∞

C

87-08-1687-08-1587-08-1487-08-1385-08-1885-08-1685-08-14Dates

–900

–700

–500

–300

–100

100

300

500

700

Co

rro

sio

n p

ote

nti

al (

mV

SH

E) Temperature

Type 304Type 316Alloy 600Alloy X-750A 533-BPlatinum

Temperature

pH electrode

Type 304Alloy 600R 533-BPlatinum

2.13 Examples of ECPs measured during start-up and shut-down in aPWR primary system [8, 9].


2.4 PWR secondary systems

Measurements in the PWR secondary side were initiated in the beginning ofthe 1990s, bearing the experience from measurements in BWRs in mind.Autoclave equipment was used for the first measurements and three autoclaveswere installed at different points in the secondary system, see Fig. 2.14. Thesampling lines were kept as short as possible and were no longer than 1 m.

Condenser

Condensatepumps

Hydrazineaddition

Lowpressureheaters

Highpressureheaters

Feedwaterpumps

Steam generators

Analysis

Analysis

Autoclave 1T = 30 ∞C



R808 V31

R807 V43

R805 V41

Autoclave 1 was installed after the condensate pumps.At this location the temperature is close to ambient.

Autoclave 2 was installed after the feedwater pumps andthe low pressure feedwater heaters. The temperature atthis location is 120–130 ∞C.

Autoclave 3 was installed after the high pressurefeedwater heaters. The temperature at this location is210–230 ∞C.

2.14 Monitoring locations in a secondary system.


The results of the secondary side measurements at Ringhals can besummarized as follows:

∑ Oxidizing conditions at location 1 and 2 independent of hydrazine content.∑ Normally reducing conditions at location 3.∑ Fast response to transients. Transients not detected by other measurements

were identified.

Also an influence of the sampling lines was noted in spite of the shortlines used. To avoid the sampling problem it was decided to install measuringequipment directly into the feedwater piping. A suitable measuring location(a pressure monitoring point) was located close to the autoclave samplingline, see Fig. 2.15. The seals of the electrodes, see Fig. 2.16, were of a typepreviously verified in BWR measurements.

Examples of results are given in Figs 2.17 and 2.18. During normal steadyoperation there is only a very small difference between the two potentials butwhen an oxidant transient occurs the in-pipe electrode reacts much morestrongly than the electrode exposed in the autoclave. For example, in Fig.2.17 a small oxygen transient is detected by the in-pipe measurement only(to the left in the diagram). To the right in the diagram a larger transient

Steam generators

Local sampling point

In situmonitoringpoint

R808 V31

R807 V43

Feedwaterpumps

Highpressureheaters

hydrazine analysis

Oxygen and



2.15 The installation point for the in-pipe electrode close to thesampling line for the third autoclave. The installation was made in apressure monitoring gauge directly into the feedwater piping.


occurs and the in-pipe electrode goes up 150 mV but the autoclave electrodeincreases only about 25 mV. In Fig. 2.18 a similar transient occurs but hereit is also clearly seen that the transient is not only smaller, but is also detectedlater in the autoclave compared to the in-pipe measurement.

The results of the secondary side measurements are summarized as follows[10, 11]:

∑ Electrochemical measurements are more sensitive to redox variationsthan other methods used in-plant.

∑ Fast response to transients. Transients identified which were not detectedby other measurements.

2.16 The in-pipe electrode for PWR secondary system. The electrodecomprises a Pt electrode and a silver chloride electrode.

–600

–500

–400

–300

–200

–100

Pt autoclave 3

Pt in pipe

94-12-29 12.0094-12-28 12.0094-12-27 12.0094-12-26 12.00

–100

–200

–300

–400

–500

–600

Po

ten

tial

(m

V S

HE

)

2.17 Example of transient response in the autoclave and at the in-pipe monitoring point.

To pressuremonitoring gauge

Silver/silver chlorideelectrode

Platinumelectrode

Feedwaterpipe


∑ In-pipe monitoring is more sensitive than measurements in side-streamautoclaves.

∑ Reliable service has been demonstrated.∑ Electrochemical monitoring offers an improved water chemistry

surveillance.∑ Interesting method for routine use.

However, ECP measurements have not developed to such a routine methodfor PWR secondary system in the same way as ECP monitoring in BWRs onHWC. One reason for that is a very stable reducing chemistry in certainPWRs without or with only very few transients. During such conditionsECPs are low and stable and thus not so interesting to monitor. However, inother PWRs oxygen transients have been detected and successfully mitigated,for example by revision of plant operation guidelines. It should also beemphasized that in some in-plant installations, the oxygen consumption inthe sampling lines probably has been large and even so large that any transientswere leveled out. New types of equipment offers improved measurements.

2.4.1 Oxygenated water chemistry in PWRsecondary system

There are now suggestions for a new type of water chemistry, OxygenatedWater Chemistry (OWC) [12]. It is well-known that iron transport to thesteam generators (SG) can affect the integrity and the performance of theSG. The dominant source of iron is carbon steel used as a structural materialin the secondary system. The carbon steel corrodes and releases corrosion

–100

–200

–300

–400

–500

–600

–100

–200

–300

–400

–500

–600

Po

ten

tial

(m

V S

HE

)

94-10-18 2.00 94-10-18 3.0094-10-18 1.0094-10-18 0.00

Pt Autoclave 3

Pt in pipe

2.18 Example of transient response in the autoclave and at the in-pipe monitoring point.


products to the coolant due to flow assisted corrosion under AVT (all volatiletreatment) chemistry. OWC has successfully been adopted in BWRs, somefossil units and a limited number of other types of power plants to suppressflow assisted corrosion of carbon steel. It has also been shown that an increaseof the redox potential limits flow assisted corrosion in a secondary sideenvironment and that the redox potential is very sensitive to small amountsof oxygen.

Figure 2.19 illustrates the OWC concept for a PWR secondary system.Oxygen is added to the feedwater line at a given point to the left in thediagram. Oxygen will be gradually consumed during transport towards thesteam generators. Curve C shows a too low addition. Curve A shows a toohigh addition where oxygen will go into the SG. Curve B would be theoptimum according to this concept.

To obtain calibration data for an ECP model for PWR secondary system,measurements were performed on carbon-steel in a laboratory loop. Theresults have been presented elsewhere [12]. Figure 2.20 shows some of theresults obtained at 180 ∞C for different hydrazine contents. The agreementbetween measured and modeled ECPs are good for the given conditions.

Another way to illustrate the modeling accuracy is shown in Fig. 2.21. Itshows the calculated loop outlet oxygen concentration versus the measuredoutlet oxygen content for various hydrazine contents. The good results indicatethat the hydrazine oxygen system has been modeled in a good way.

For ECP-modeling work a so-called OWC simulator has been developedwhich in principle is very similar to the BWR ECP-model. Figure 2.22shows an example of calculations using the OWC simulator. It shows thedecrease of the oxygen content and the corrosion potential in a samplingline. The reactions are very fast in small diameter sampling lines as used in

Suitable oxygen rangefor FAC suppression

O2 injection

Distance from O2 injection point

O2

con

cen

trat

ion

Feedwater line (Carbon steel)

Area with FAC

SG

Curve A

Curve B

Curve C

2.19 Illustration of the OWC concept [12].


the modeling according to the figure, but in large diameter pipes the chemistryconditions can be stable in analogy with the situation in BWRs. Due to thatECP monitoring is needed to follow the effects of OWC.

2.5 Summary and conclusions

The current status of in-plant ECP monitoring is summarized as follows:

∑ In BWRs electrochemical monitoring is an extremely valuable tool forHWC surveillance.

5000 N2H41000 N2H4300 N2H4100 N2H4500 NH3 onlyPure water

1001010.10.01O2 conc. (ppb)

–600

–500

–400

–300

–200

–100

0

100

EC

P (

mV

SC

E)

2.20 Modeled ECP as a function of O2 concentration at 180 ∞C anddifferent levels of hydrazine (pH 9.2 except for pure water) [12].

10030010005000

1001010.10.01Measured loop outlet O2 level

Mo

del

ed lo

op

ou

tlet

O2

leve

l

100

10

1

0.1

0.01

2.21 Calculated outlet oxygen concentration vs. measured outletoxygen at 180 ∞C [12].


∑ In PWR primary system redox variations are small outside the core. Forlow hydrogen chemistry, electrochemical monitoring is of considerableinterest.

∑ In PWR secondary systems, electrochemical monitoring is a versatiletool for redox monitoring and redox mapping. Electrochemicalmeasurements are strongly recommended for any OWC application inPWR secondary system.

∑ We have a good understanding of the basic mechanisms behind ECP andsemi-quantitative modeling is possible.

∑ Accurate, quantitative modeling is, however, in principle impossible.ECP of SS and Ni-base alloys in reactors is a kinetically determinedquantity that depends on the surface properties of the electrode. Thesurface properties, in turn, may depend on the water chemistry and thehistory of the electrode.

∑ ECP modeling may provide reliable estimates of ECPs upstream anddownstream of an ECP measuring point.

2.6 Acknowledgements

Many colleagues at Studsvik and at Swedish power plants contributed to thiswork and their contributions are gratefully acknowledged. Also financialsupport from Swedish utilities, Swedish Nuclear Power Inspectorate andThe Japan Atomic Power Company is gratefully acknowledged.

2.22 Simulation of a 25 m long 10 mm sampling line at 180 ∞C with aflow rate of 20 g/s [12].

O2(

pp

p)


2.7 References

1. Ullberg, M. On corrosion potential measurement in BWRs, Proc. Fourth Int. Symp.on Environmental Degradation of Materials in Nuclear Power Systems – WaterReactors, NACE, 1990.

2. Molander, A. and Karlberg, G. Hydrogen water chemistry surveillance in a boilingwater reactor, Proc. Fourth Int. Symp. on Environmental Degradation of Materials inNuclear Power Systems – Water Reactors, NACE, 1990.

3. Molander, A. and Jansson, C. Conventional and in situ corrosion potential monitoringin a BWR, Proc 1991 Int. Conf. on Water Chemistry in Nuclear Power Plants, JAIF,(1991).

4. Molander, A. et al. Corrosion potential monitoring in Swedish BWRs on hydrogenwater chemistry, Proc. Ninth Int. Symp. on Environmental Degradation of Materialsin Nuclear Power Systems – Water Reactors, TMS, 1999.

5. Molander, A. et al., Comparison of the corrosion potential for stainless steel measuredin-plant and in laboratory during BWR normal water chemistry conditions, Proc1998 Int. Conf. on Water Chemistry in Nuclear Power Plants, JAIF, (1998).

6. Wikmark, G., Lundgren, K., Wijkström, H., Pein, K. and Ullberg, M. SKI Report2004: 27, 2004. (in Swedish).

7. Molander, A. and Ullberg, M. The Corrosion Potential of Stainless Steel in BWREnvironment – Comparison of Data and Modeling Results. Proc. Symp. on WaterChemistry and Corrosion in Nuclear Power Plants in Asia 2003, Atomic EnergySociety of Japan (2003).

8. Molander, A. et al., Corrosion potential measurements in reactor water of a PWR,Proc. Fourth Int. Conf. on Water Chemistry of Nuclear Reactor Systems, Bournemouth,BNES, (1986).

9. Molander, A. et al., Significance of corrosion potential monitoring in a PWR primarysystem, Proc. Fifth Int. Conf. on Water Chemistry of Nuclear Reactor Systems,Bournemouth, BNES, (1990).

10. Molander, A. et al., Electrochemical measurements in secondary system of Ringhals3 PWR, Proc. Sixth Int. Conf. on Water Chemistry of Nuclear Reactor Systems,Bournemouth, BNES, (1992).

11. Molander, A. et al., Studies of redox conditions in feedwater of PWR secondarysystems, EUROCORR’96.

12. Takiguchi, H., Kadoi, E. and Ullberg, M. Study on Application of Oxygenated WaterChemistry for Suppression of Flow Assisted Corrosion in Secondary Systems ofPWRs, Presented at the 14th Int. Conf. on the Properties of Water and Steam, Kyoto,Japan, 2004.

44

3.1 Introduction

The oxidation behaviour of nickel base alloys in high temperature and highpressure water, simulating the primary circuit of steam generators (SG) ofpressurised water reactors (PWR), has been studied by several authors [1–12]. It is generally recognised that the passivity of this alloy in such conditionsis due to the formation of a chromium-rich oxide layer which provides adiffusion barrier and reduces the corrosion rate. The passive layer reducesthe release of corrosion products, such as nickel cations, in primary sidewater. It is of crucial importance because of the activation of 58Ni into 58Coin the primary circuit, which increases the global radioactivity of the primarycircuit of PWRs. For safety reasons for the maintenance staff, it is veryimportant to control and limit the release of nickel species in the primarycircuit. Another important application of a better knowledge of the formationof the passive film is the understanding of stress corrosion cracking mechanisms[13]. Very few papers deal with the initial stages of passivation in hightemperature and high pressure water. The aim of this work was to identifythe nature of the oxide layer on Alloy 600 by X-ray Photoelectron Spectroscopy(XPS), from very short to longer times of passivation and to determine thekinetic law for the growth of the barrier oxide layer for both short and longertimes.

3.2 Experimental procedure

3.2.1 Materials

Polycrystalline samples of commercial Alloy 600 (Ni-16Cr-9Fe (weight %)or Ni-18Cr-9Fe (atomic %)) were cut either into rectangular coupons of15 ¥ 10 mm2 (1 mm thick) for shorter oxidation times or disks of 15 mmdiameter (1 mm thick) for longer oxidation times. The polycrystalline couponswere mechanically polished to a 1 mm diamond finish. Prior to the short

3Kinetics of passivation of a nickel-base

alloy in high temperature water

A. M A C H E T, A. G A LT AY R I E S and P. M A R C U S,Laboratoire de Physico-Chimie des Surfaces, France and

P. J O L I V E T, M. F O U C A U LT, P. C O M B R A D Eand P. S C O T T, Framatome ANP, France

Kinetics of passivation of a nickel-base alloy 45

passivation time periods, further cleaning was performed, in UHV conditions,in the XPS spectrometer, with argon ion sputtering (4 kV, 0.9 mA, 1 ¥ 10–4

Pa (1 ¥ 10–6 mbar), 60 minutes).

3.2.2 Passivation

The short passivation times in high temperature, high pressure water(325 ∞C, ~155 ¥ 105 Pa (155 bar)) were performed in a titanium microautoclave,dedicated to short times of treatment (a few tens of seconds and up), allowingtransfers of the samples to and from the XPS spectrometer without air exposure.The cooling in the microautoclave is rapid (~2 minutes) and is performedunder argon, as well as the transfer to the XPS, so that the possible changesof the surface composition are minimised. The longer passivation times (upto 400 hours) were performed in a static autoclave, and the samples wererinsed, dried and transferred in air before XPS analysis. The data for t = 0(blank tests) correspond to the surface analysis of a sample either after thedifferent transfers from the spectrometer to the microautoclave and back tothe spectrometer, without immersion in primary water (shorter oxidationtimes) or after heating up to 325 ∞C in PWR conditions and cooling (longeroxidation times). In both systems, the aqueous solution simulating unsaturatedPWR primary water conditions contained 2 mg.l–1 Li and 1200 mg.l–1 B. Ahydrogen overpressure of 0.3 ¥ 105 Pa (0.3 bar) was maintained to ensure adissolved H2 concentration of 35 cm3.kg–1 and a low oxygen content of<30 mg.kg–1. In our experimental conditions, the pH of the solution was 7.1and the potential was –0.808 V/SHE. This potential is calculated from theH2/H

+ equilibrium potential, close to the corrosion potential of Alloy 600[14].

3.2.3 Surface analysis

For the surface characterisation by XPS, the Ni 2p, Cr 2p, Fe 2p, O 1s, C 1s,B 1s and Li 1s core level spectra were recorded with a VG ESCALAB MkII X-ray photoelectron spectrometer, with an AlKa or MgKa radiation (hn =1486.6 and 1253.6 eV, respectively), at a pass energy of 20 eV. The spectrometerwas calibrated against the reference binding energies (BEs) of clean Ni andAu samples (Ni 2p3/2 and Au 4f7/2 lines set at 852.8 eV and 84.0 eV,respectively). The take-off angles of the photoelectrons were 90∞ and 45∞,with respect to the sample surface. In the XPS spectrometer, the base pressureof the analysis chamber was 3 ¥ 10–8 Pa (3 ¥ 10–10 mbar). The argon ionssputtering used to clean the surface prior to passivation was carried out in theanalysis chamber with an ion energy of 3 keV. To analyse the individualcontributions of the Ni 2p3/2, Cr 2p3/2, Fe 2p3/2, O 1s, C 1s, and B 1s corelevels, peak decomposition was carried out with a commercial computer


program (Eclipse provided by VG) using Gaussian/Lorentzian peak shapes,and a Shirley background. For longer passivation times, the depth profilemode was used (in the analysis chamber), with 3 keV argon ions, a targetcurrent of 2.5 mA.cm–2, and an Ar pressure of 1 ¥ 10–4 Pa (1 ¥ 10–6 mbar).

3.3 Results

3.3.1 Short oxidation times

Figure 3.1(a) shows the Cr 2p core level spectra recorded after passivationof the polycrystalline alloy for different time periods: from the blank test (0)to 8.2 min. With increasing exposure times, the intensity of metallic chromiumdecreases (low binding energy part), but the signal is still present after 8minutes of passivation. To get more detailed information about the changesin chromium surface species, the Cr 2p3/2 peaks have been systematically

Ni 2p3/2X

PS

inte

nsi

ty (

arb

. un

it)

Binding energy (eV)(b)

885 875 865 855 845

8.2 min

6.5 min

5.6 min

4.4 min

1.8 min

1.2 min

0.4 min

0 min

Cr 2p3/2

Binding energy (eV)(a)

570574578582586590594

8.2 min

6.5 min

5.6 min

4.4 min

1.8 min

1.2 min

0.4 min

0 min

XP

S in

ten

sity

(ar

b. u

nit

)

3.1 Evolution of the Cr 2p (a) and Ni 2p (b) core level peaks of a Ni-16Cr-9Fe (wt.%) alloy for different passivation times in themicroautoclave, in high temperature (325 ∞C) water (MgKa X-raysource, take-off angle of 90∞).


decomposed into up to three components: one located at a binding energy(BE) of 574.3 ± 0.1 eV, and two other ones located at BEs of 577.2 ± 0.5 eVand 577.9 ± 0.1 eV (see for example Fig. 3.2(a) for a passivation time of 8.2min). By comparison with published data [15–17], the signal at 574.3 eV isassigned to metallic chromium in the nickel-base alloy, the signal at 577.2eV to Cr3+ in Cr2O3 [1, 15, 18] and the signal at 577.9 to Cr3+ in Cr(OH)3

[1, 15]. From this systematic peak fitting, it comes that the signal of chromium(III) hydroxide disappears, around 4 min, in favour of chromium (III) oxideonly (Fig. 3.2(a)).

Figure 3.1(b) shows the Ni 2p core level spectra recorded after passivationof the polycrystalline alloy for different time periods in a similar comparison:from the blank test (0) to 8.2 min. The signal from metallic nickel is dominantfor 0 to 4.4 min of passivation. After 8.2 min., the metallic signal is stillpresent but the signal from oxidised nickel has become significantly moreintense. To get more detail about the nature and quantity of oxidised nickelspecies, two main components (and the associated satellites) are consideredfor the Ni 2p3/2 peak decomposition: a signal from metallic nickel in the

Binding energy (eV)(c)

Fe2O3

Fe 2p3/215

14

13

XP

S in

ten

sity

(ar

b. u

nit

.)

720 715 710 705Binding energy (eV)

(d)

XP

S in

ten

sity

(ar

b. u

nit

.)

14

10

542 538 534 530 526

O 1s

B2O3

Fe2O3

Cr2O3

Ni(OH)2


865 855 845

Ni∞

Ni 2p3/2

XP

S in

ten

sity

(ar

b. u

nit

.)

18

16

Ni(OH)2


Cr 2p3/2

Cr∞

Cr2O3

XP

S in

ten

sity

(ar

b. u

nit

.) 13

11

572576580584

3.2 Cr 2p3/2 (a), Ni 2p3/2 (b), Fe 2p3/2 (c) and O 1s (d) core level spectra(and the peak fitting) of a Ni-16Cr-9Fe (wt.%) alloy after 8.2 minutesof passivation in the microautoclave, in high temperature (325 ∞C)water (MgKa X-ray source, take-off angle of 90∞).


alloy at a BE of 853.1 ± 0.1 eV [8, 15], and another feature located at a BEof 856.8 ± 0.4 eV (see for example Fig. 3.2(b) after 8.2 min. of passivation).

The latter signal is attributed to Ni(OH)2 on the surface [19]. The absenceof Ni2+ signal corresponding to NiO in the Ni 2p core level spectra, whateverthe passivation time, is in agreement with the Pourbaix diagram of Ni at300 ∞C [20, 21], showing that NiO is not stable in the experimental E-pHconditions of this work. Core level spectra of both Cr and Ni indicate thegrowth of a surface oxide layer (composed of oxide and hydroxides).

The Fe 2p3/2 spectra (see for example Fig. 3.2(c) after 8.2 min. of passivation)show low intensity signals, due to the low concentration of iron in the alloyas well as on the surface after passivation (the surface concentration in theoxide layer is <5 at. %). The Fe 2p3/2 spectra are fitted by only one peak,located at 710.6 ± 0.2 eV, corresponding to oxidised iron in Fe2O3 [15].

The peak decompositions of the alloying elements are well correlatedwith the O 1s peak decomposition (see for example Fig. 3.2(d) for 8.2 min.of passivation), including the chromium oxide feature at a BE of 531.1 ± 0.6eV, the chromium hydroxide feature at a BE of 531.8 ± 0.2 eV (for passivationtimes less than 4 min.) and a feature at 532.2 ± 0.5 eV attributed to thehydroxide ions in nickel hydroxide. As a minor feature, one can systematicallysee a small oxygen signal of O2– in Fe2O3 (530.6 ± 0.3 eV).

The oxygen signal coming from a contamination by boron oxide is locatedat 533.1 ± 0.5 eV, with also a small contribution at very high binding energy(around 538 eV). The latter peak, strongly shifted, is due to differentialcharging effects on this compound [19].

The angle-dependent XPS data (not shown here) for the O 1s core levelindicate that the nickel hydroxide is located on the chromium oxide surface[19]. A simple layer model, based on the stratification of the passive layer,is shown in Fig. 3.3. This model has been used for the calculation of thethickness of the different layers present in the passive film.

It consists of an outermost layer of Ni(OH)2, an intermediate layer ofCr(OH)3 and an inner Cr2O3 oxide layer, in contact with the alloy. Due to itslow concentration (<5 at.%), and to the difficulty to locate the small amountof detected Fe2O3, it was not included in this model.

The equivalent thicknesses of the external layer (Cr(OH)3 and Ni(OH)2)and the Cr2O3 internal layer have been systematically determined from theXPS intensities of the metallic and oxide features. Figure 3.4 displays theresults corresponding to the Cr2O3 inner layer, plotted as a function ofpassivation time. From the examination of Fig. 3.4, three domains are observed:

(i) after the first exposure (0.4 min), the formation of chromium oxide,(ii) from 0.4 to ~4 min, there is a plateau corresponding to an ultra-thin

Cr2O3 oxide layer (~ 1 nm).(iii) beyond ~4 min, a re-oxidation is observed, with the growth of the

Cr2O3 layer.


The three domains are indicated on Fig. 3.4.After 8 minutes of passivation, the oxide layer is still growing, exhibiting

a duplex layer with islands of Ni(OH)2 on top of the continuous inner layerof Cr2O3 (2 nm thick). Complementary information on the structural aspectsof the inner Cr2O3 layer have been obtained by Scanning Tunneling Microscopy

Ni-Cr-Fe alloy

Cr2O3

Cr(OH)3

Ni(OH)2

3.3 Layer model used for the calculation of the thickness of thedifferent layers present in the passive film formed on the Ni-16Cr-9Fe(wt.%) (Ni-18Cr-9Fe at.%) alloy surface after a short oxidation in hightemperature (325 ∞C) water.

Experimental data for alloy 600Inverse logarithmic curve fittingParabolic curve fittingLogarithmic curve fitting

Plateau

21 33

2

1

0

Eq

uiv

alen

t th

ickn

ess

of

the

chro

miu

m o

xid

e la

yer

(nm

)

109876543210Passivation time (min)

3.4 Equivalent thickness of the Cr2O3 internal layer formed on a Ni-16Cr-9Fe (wt.%) alloy (Ni-18Cr-9Fe at.%) as a function of passivationtime in the microautoclave, and the parabolic, logarithmic andinverse logarithmic fitting of the growth of the Cr2O3 layer in Step 3.


(STM), indicating that this layer is crystalline and that its structure is consistentwith the hexagonal structure of the oxygen sub-lattice hc in the (0001)orientation of a-Cr2O3 [19].

Figure 3.4 displays the curve fitting of the kinetics of the Cr2O3 growth bythree classical kinetic models: parabolic, logarithmic and inverse logarithmic,the equations of which are the following:

∑ parabolic fitting: ( – 0.9) = 0.35 ( – 4.3)Cr O2

2 3d t¥∑ logarithmic fitting: dCr O2 3 = 0.9 + 0.39 ¥ ln(4.06 ¥ (t – 4.3) + 1)

∑ inverse logarithmic fitting: 0.6( – 0.9)

= 1.62 – ln – 4.3( – 0.9)Cr O Cr O

22 3 2 3

dt

d

At this point, it is not possible to discriminate between the three growthlaws, but it will be shown below that the discrimination becomes possible byextrapolating to longer oxidation times and comparing with the experimentaldata.

3.3.2 Longer oxidation times

The samples have been exposed from 0 (blank test) to 400 hours in hightemperature (325 ∞C) and high pressure water in a static autoclave. The XPSspectra of the oxidised samples (before sputtering) show that the signalscorresponding to the metallic alloy (Ni, Cr, Fe) are not detected (see forexample Fig. 3.5(a), (c) after 100 hours of passivation), which indicates thatthe oxide layer is thicker than the one measured for short oxidation times.XPS depth profiles were performed in order to get in-depth information onthe composition of the oxide layer.

To be consistent with the results obtained for short times of oxidation,before and during sputtering, for each passivation time, the Cr 2p3/2 corelevel was systematically decomposed into metallic chromium from the alloy, ata BE of 574.4 ± 0.1 eV, and Cr2O3 at a BE of 577.4 ± 0.1 eV, the Ni 2p3/2 corelevel into metallic nickel from the alloy at a BE of 853.1 ± 0.1 eV, andNi(OH)2 at a BE of 857.3 ± 0.1 eV. An example of peak decomposition isgiven in Fig. 3.5, after a passivation time of 100 hours in a static autoclave.The intensity of the Fe 2p core level was very weak so the peak decompositionwas difficult to perform, and only the intensity of the Fe 2p1/2 core level peakis reported in Fig. 3.6 (to avoid the overlapping with the Ni Auger lines inthe Fe 2p3/2 region, with the AlKa X-ray source). As regards the O 1s corelevel peak, it was decomposed into two constituents: the hydroxyl OH– at aBE of 532.8 ± 0.2 eV and the oxygen in oxide O2– at a BE of 531.1 ± 0.2 eV(Fig. 3.5(e) and (f)).

Figure 3.6 displays one typical example of the XPS profiles obtained onthe Ni-16Cr-9Fe (wt.%) alloy (Ni-18Cr-9Fe at. %) oxidised 100 hours inhigh temperature water, as a function of sputtering time. From the examination


of Fig. 3.6, one can observe that both the intensities corresponding to thehydroxyl contributions in O 1s and Ni 2p core levels decrease rapidly as afunction of sputtering time. At the same time, an increase of the oxidecontributions in the O 1s and Cr 2p3/2 core levels is observed. It is concludedthat the external layer is mainly composed of Ni(OH)2 while the inner layeris composed of Cr2O3. The weak intensity of the signal of the Fe 2p3/2 corelevel does not present any significant variation during sputtering.


Cr 2p3/2577.4 eV

Cr2O330

25

20XP

S in

ten

sity

(K

CP

S)

590 585 580 575 570

Ni 2p3/2

Binding energy (eV)(c)

857.3 eV

Ni(OH)2

875 865 855 845

XP

S in

ten

sity

(K

CP

S) 90

80

70

60

50

O 1s

Binding energy (eV)(e)

XP

S in

ten

sity

(K

CP

S)

80

60

40

20

540 536 532 528

Oxide

Hydroxide531.1 eV

532.8 eV

Cr 2p3/2


Cr∞

577.4 eV

Cr2O3

XP

S in

ten

sity

(K

CP

S) 80

70

60

50

40

30

20

574.6eV

590 585 580 575 570

Ni 2p3/2

Binding energy (eV)(d)

Ni∞853.1 eV

857.3 eVNi(OH)2XP

S in

ten

sity

(K

CP

S) 110

100

90

80

70

60875 865 855 845

Binding energy (eV)(f)

O 1s531.1 eV

Oxide

Hydroxide532.8 eV

540 536 532 528

XP

S in

ten

sity

(K

CP

S)

80

60

40

20

3.5 Cr 2p3/2, Ni 2p3/2, and O 1s core level spectra (and the peakfitting) of a Ni-16Cr-9Fe (wt.%) alloy after 100 hours in hightemperature (325 ∞C) water, in a static autoclave: before ionsputtering ((a), (c) and (e), respectively) and after 70 minutes of ionsputtering ((b), (d) and (f), respectively). AlKa X-ray source, sputteringconditions: Ar ions, 3 keV, 2.5 mA · cm–2.


In order to determine the thickness of the chromium oxide inner layerfrom the XPS profiles, we have used the intensity of oxygen in the oxide(O2–). The sputtering time between the maximum of the intensity (when thechromium oxide is reached), and the intensity at half maximum (assigned tothe oxide/alloy interface) is measured. The sputtering time is then convertedinto thickness, using the calibration obtained from Nuclear Reaction Analysis(NRA): 0.25 nm.min–1 for a target current of 2.5 mA.cm–2 [22].

After oxidation of Alloy 600, for 100 hours in high temperature water (Fig.3.6), the equivalent thickness of Cr2O3 is 10 nm ±1 nm. The same systematictreatment of the XPS depth profiles, after oxidation times from 0 (blank test)to 400 hours, has allowed us to obtain the thickness of Cr2O3 for eachpassivation time. Figure 3.7 shows the kinetics of chromium oxide growthfor passivation times up to 400 hours. The thickness of the inner Cr2O3 layerincreases significantly up to 20 hours, then it becomes almost constant.

3.3.3 Extrapolation of the growth laws calculated forshort oxidation times and comparison with theexperimental data

In order to relate the data of the kinetics of passivation of the alloy for shortand long oxidation times (up to 400 hours), the three kinetic laws fitting the

Ni (metallic)Cr (oxide)

Ni (hydroxide)Fe (total)

Cr (metallic)O (hydroxide)

O (oxide)

6050403020100Sputtering time (min)

70

60

50

40

30

20

10

0

XP

S In

ten

sity

(kC

PS

/s)

Internal layer: 10 nm

O (hydroxide)

O (oxide)

Ni (metallic)

Fe (total)Cr (metallic)

Ni (hydroxide)

Cr (oxide)

3.6 XPS depth profile of the oxide layer formed on a Ni-16Cr-9Fe(wt.%) alloy (Ni-18Cr-9Fe at.%) after 100 hours in high temperature(325 ∞C) water, in a static autoclave (AlKa X-ray source, sputteringconditions: Ar ions, 3 keV, 2.5 mA · cm–2).


data for the short oxidation times have been extrapolated, using the initialvalue of the Cr2O3 oxide layer (measured after the blank test). The threeequations are now:

∑ parabolic fitting: ( – 6) = 0.35 Cr O2

2 3d t¥∑ logarithmic fitting: ( = 6 + 0.39 ln (4.06 + 1)Cr O2 3d t¥ ◊

∑ inverse logarithmic fitting: 0.6( – 6)

= 1.62 – ln ( – 6)Cr O Cr O

22 3 2 3

dt

d

The three calculated curves are plotted on Fig. 3.7. It comes from thecomparison of the fitted and experimental data that the parabolic and inverselogarithmic laws are out of range, while a satisfactory agreement is obtainedwith the logarithmic law.

3.4 Discussion

The results of this work and of the STM data [19] show that, in the initialstage of oxidation, the mechanisms of passivation of the Ni-16Cr-9Fe (wt.%)alloy (Ni-18Cr-9Fe at. %) involve three steps:

∑ Step 1: selective dissolution of nickel from the metallic alloy, and nucleationand growth of Cr2O3 islands, covered by Cr(OH)3, in contact with theprimary water solution.

3.7 Kinetics of passivation of a Ni-16Cr-9Fe (wt.%) alloy (Ni-18Cr-9Feat.%) in high temperature (325 ∞C) water: thickness of the inner Cr2O3layer vs. passivation time. The three extrapolations from the dataobtained for short oxidation times are shown: parabolic, logarithmicand inverse logarithmic models.

Experimental data for Alloy 600Inverse logarithmic curve fittingParabolic curve fittingLogarithmic curve fitting

5004003002001000Passivation time (h)

160

140

120

100

80

60

40

20

0

Eq

uiv

alen

t th

ickn

ess

of

the

chro

miu

m o

xid

e la

yer

(nm

)


∑ Step 2: coalescence of the Cr2O3 islands and formation of a continuouslayer of Cr2O3 (about 1 nm thick), with an outer layer of Cr(OH)3. Atthis point, there is a temporary blocking of the growth of Cr2O3, evidencedby the plateau in the kinetics of Fig. 3.3.

∑ Step 3: the oxide growth starts again with the conversion of Cr(OH)3

into Cr2O3 according to: Cr(OH)3 + Cr Æ Cr2O3 + 3H+ + 3e– and thenthe further growth of Cr2O3.

The data for short oxidation times can be well fitted by classical growthmodels. To check the consistency of the models, the growth of Cr2O3 hasbeen followed from the depth profiles obtained for longer oxidation times.The comparison of the extrapolated growth law to longer times shows that inthe range of passivation times up to 400 hours, it is possible to rule outunambiguously the parabolic and inverse logarithmic laws. Only the logarithmiclaw fits well the experimental data. From a mechanistic point of view, thefact that the parabolic law does not fit the data shows that the hypothesis ofthermally activated solid state diffusion is not valid here. The mechanismassociated to a logarithmic law involves the mobility of ions in a high field(with the tunneling of electrons through the oxide). However, the thicknessof 10 nm, obtained here for the Cr2O3 oxide layer, seems too large for theelectron tunneling effect. However, one has to remember that the chromiumoxide is crystalline and grain boundaries are present in the oxide layer [19].Such defects, and possibly other defects, can allow the transfer of electrons.This approach has revealed that it is possible to follow the kinetics of thegrowth of the inner Cr2O3 layer in the initial stages of passivation. It is thefirst direct evidence of the key role of short oxidation times in the kinetics ofthe Cr2O3 oxide growth on nickel-base alloys in high temperature water.

In the context of the release of nickel species, it has been possible toestimate the amount of nickel released in the aqueous solution, during step1. The resulting value (0.2 mg.dm–2) is in satisfactory agreement with somerecent data [23], obtained with electropolished coupons of Alloy 690, after24 hours (the shortest time investigated in this work). In the context of stresscorrosion cracking [13], the three-step mechanism that is proposed here canbe used as a reasonable hypothesis to describe the repassivation process afterlocal breakdown of the passive layer.

3.5 Conclusion

The kinetics of passivation in high temperature and high pressure water of aNi-16Cr-9Fe (wt.%) alloy (Ni-18Cr-9Fe at. %) has been investigated. Thecomposition and the thickness of the surface oxide layer have been measuredfor very short oxidation times (minutes) and longer oxidation times (up to400 hours). The thickness of the inner chromium oxide barrier layer has been


measured as a function of oxidation times. From the three kinetic laws thatfit well the experimental data for the short oxidation times, only the logarithmiclaw can be retained from the comparison of the extrapolated growth lawswith the experimental data for long oxidation times. The logarithmic law ischaracteristic of an apparent mechanism of high charge field effect for thegrowth of the internal Cr2O3 layer. This approach has revealed that it ispossible to follow the kinetics of the growth of the inner Cr2O3 layer in theinitial stages of passivation.

3.6 References

1. J. E. Castle, C. R. Clayton, Passivity of Metals, R. P. Frankenthal and J. Kruger Eds.,The Electrochemical Society, Princeton, N. J., USA, 1978, 714–729.

2. N. S. McIntyre, D. G. Zetaruk, D. Owen, J. Electrochem. Soc., 1979, vol. 126, 750–760.

3. C. Y. Chao, L. F. Lin, D. D. Macdonald, J. Electrochem. Soc., 1981, vol. 128, 1187–1194.

4. L. F. Lin, C. Y. Chao, D. D. Macdonald, J. Electrochem. Soc., 1981, vol. 128, 1194–1198.

5. R. L. Tapping, D. Davidson, E. McAlpine, D. H. Lister, Corros. Sci., 1987, vol. 26,563–576.

6. P. Combrade, M. Foucault, D. Vançon, P. Marcus, J. M. Grimal, A. Gelpi, Proceedingsof the 4th International Symposium on Environmental Degradation of Materials inNuclear Power Systems – Water Reactors, D. Cubicciotti Ed., NACE, 1989, 79–95.

7. J. Robertson, Corros. Sci., 1991, vol. 32, 443–457.8. T. M. Angeliu, G. S. Was, J. Electrochem. Soc., 1993, vol. 140, 1877–1883.9. N. Hakiki, D. Colin, O. De Bouvier, E. Picquenard, G. Sagon, J. Corset, M. Da

Cunha Belo, Proceedings of the International Symp. Fontevraud III, Contribution ofMaterials Investigation to the Resolution of Problems Encountered in PressurizedWater Reactors, 1994, vol. 1, 327–336.

10. B. Stellwag, Corros. Sci., 1998, vol. 40, 337–370.11. F. Carette, M. C. Lafont, G. Chataignier, L. Guinard, B. Pieraggi, Surf. Interf. Anal.,

2002, vol. 34, 135–138.12. F. Carette, L. Guinard, B. Pieraggi, Proceedings of the International Conference on

Water Chemistry of Nuclear Reactor Systems, Operation Optimisation and NewDevelopments (2002).

13. F. P. Ford and P. L. Andersen, in Corrosion Mechanisms in Theory and Practice –Second Edition, Revised and Expanded, Ed. P. Marcus, Marcel Dekker, Inc., NewYork (USA), 2002, and reference therein.

14. N. Totsuka, Z. Szklarska-Smialowska, Corrosion, 1987, vol. 43, 734–738.15. P. Marcus, J. M. Grimal, Corros. Sci., 1992, vol. 33, 805–814.16. V. Maurice, W. P. Yang, P. Marcus, J. Electrochem. Soc., 1994, vol. 141, 3016–3027.17. A. M. Salvi, J. E. Castle, J. F. Watts, E. Desimoni, Appl. Surf. Sci., 1995, vol. 90,

333–341.18. N. S. McIntyre, D. G. Zetaruk, D. Owen, Appl. Surf. Sci., 1978, vol. 2, 55–73.19. A. Machet, A. Galtayries, S. Zanna, L. Klein, V. Maurice, P. Jolivet, M. Foucault, P.

Combrade, P. Scott, P. Marcus, Electrochimica Acta, 2004, vol. 49, 3957–3964.


20. B. Beverskog, I. Puigdomenech, Corros. Sci., 1997, vol. 39, 969–980.21. B. Beverskog, I. Puigdomenech, Corros. Sci., 1997, vol. 39, 43–57.22. A. Machet, A. Galtayries, P. Marcus, A. Gelpi, C. Brun, P. Combrade, J. Phys. IV,

2001, vol. 11, 79–88.23. F. Carette, PhD Thesis, Institut National Polytechnique de Toulouse (2002).

Part II

Stress corrosion cracking: susceptibilityand initiation

57

59

4.1 Introduction

Core components of light water reactors (LWR), made of austenitic stainlesssteels (SS) and nickel alloys, subjected to stress and exposed to relativelyhigh fast neutron flux may suffer a cracking process termed as IrradiationAssisted Stress Corrosion Cracking (IASCC). This degradation phenomenonis a time dependent process in which neutron and gamma radiation aredirectly implicated in the initiation and propagation of cracking [1]. Althoughthis type of cracking was first recognized in boiling water reactors (BWR),later service failures attributed to IASCC were also observed in pressurizedwater reactors (PWR) components [2].

Cracking of welded reactor pressure vessel (RPV) internal components,such as core shrouds, has increased in BWR during recent years. Most ofthese cracking incidents were associated with the heat-affected zone (HAZ)of the welded components. As cracking was located in the HAZ, some coreshroud failures have been attributed to classical Intergranular Stress CorrosionCracking (IGSCC) of thermally sensitized stainless steels, due to the significantgrain boundary carbide precipitation occurring in the HAZ during the weldingprocess. However, the intergranular cracking of stabilized and L-gradematerials, where carbide precipitation is minimized, cannot be sufficientlyexplained by the thermal chromium depletion mechanism [3].

Although the maximum end-of-life dose for BWR core shrouds is about3 ¥ 1020 n/cm2 [4], below the threshold fluence (5 ¥ 1020 n/cm2) for IASCCin BWR of annealed materials, the influence of neutron irradiation in theHAZ of welds is still an open question. As a consequence of the weldingprocess, residual stresses, microstructural and mechanical changes are inducedin the welded stainless steels. In addition, neutron radiation can lead tocritical modifications in material characteristics and in the surrounding waterenvironments, which can modify the stress corrosion cracking resistance ofthe components. While the IASCC of base materials is being widely studied,the specific conditions of weldments are rarely addressed.

4IASCC susceptibility under BWR conditions

of welded 304 and 347 stainless steels

M. L. C A S T A Ñ O, CIEMAT, Spain, B. VA N D E RS C H A A F, NRG, Holland, A. R O T H, Framatome ANP,

Germany, C. O H M S, JRC-IE, Holland, D. G AV I L L E T,PSI, Switzerland and S. VA N D Y C K, SCK·CEN, Belgium


An engineering and scientific approach to understand the behaviour ofwelds under irradiation was the base for the definition of the INTERWELDProject. The objective of the project is to explore the neutron radiationinduced changes in the welding heat-affected zones of components, such ascore shrouds of the BWR, that promote intergranular cracking. For thispurpose the relation between the development of weld residual stresses,microstructure, microchemical and mechanical properties during irradiationand the stress corrosion cracking behaviour of the materials is investigated.This work was performed with the financial support from the EuropeanCommission, EURATOM FP5 (contract number FIKS-CT-2000-00103). Amore detailed description of the project has been published previously [5].


4.2.1 Materials

Two austenitic stainless steels, type 304 and 347, were used for this experimentalwork. The chemical compositions of these materials are listed in Table 4.1.To produce the weldments, two plates of 2500 mm ¥ 200 mm ¥ 12 mm ofeach material, 304 and 347 SS, were joined by welding along the long sideof the rectangular strips. Figure 4.1 is a scheme of the welding profile. GasTungsten Arc Welding (GTAW) was used for the root pass (first pass) andShielded Metal Arc Welding (SMAW) for all filler passes (passes 2–5). A

Table 4.1 Chemical composition of the materials (wt.%)

Materials C Si Mn P S Cr Ni Nb Nb/C

304 0.042 0.310 1.63 0.03 0.010 18.36 9.50 – –347 0.03 0.46 1.22 0.034 0.005 17.69 10.34 0.49 16.3

60∞

12

2–4

5

2

3

4

1

Base metals:304 or 347

Base metals:304 or 347 SS

4.1 Scheme of the welding (mm). Source: Framatome ANP.

IASCC susceptibility under BWR conditions 61

double-V shape was selected to minimize the shrinkage and deformation ofthe weld plates.

4.2.2 Irradiation process

After the welding process, the two plates were divided and weld strips/slices, stress corrosion, tensile, microstructural and microchemical testspecimens were fabricated. All these samples (specimens and strips) wereinserted in the irradiation capsules and irradiated at 300 ∞C by NRG in theHigh flux Reactor (HFR) in Petten, NL, at two neutron fluences: 0.3 and 1.0dpa.

4.2.3 Materials characterization

Before and after irradiation the materials were fully characterized. The generalmicrostructure was determined by optical and Transmission ElectronMicroscopy (TEM), for both base metal and HAZ. The grain boundarycharacterization at the HAZ was performed by Auger Electron spectroscopy.Vickers micro-hardness and tensile properties were determined in the basemetal and HAZ. For mechanical characterization, flat micro-tensile samplesof 0.3 mm thickness and 5.5 mm gauge length were used. The mechanicalproperties at the HAZ were determined by fabrication of micro-tensile samplesevery 0.4 mm from the fusion line.

The sensitization degree of unirradiated stainless steel 304 and 347weldments was determined using the Double Loop-EPR (ElectrochemicalPotentiodynamic Reactivation) method. In order to get information aboutthe HAZ, examinations were performed by successively cut tangential crosssections along the weld edge (weld direction) starting nearest the fusion lineand proceeding in defined distance steps (0.2 mm) towards the unaffectedbase metals.

4.2.4 Residual stresses measurements

Weld residual stresses are determined by destructive and non-destructivemethods. Before irradiation, weld residual stresses were measured by neutrondiffraction and by the ring-core technique while after irradiation only neutrondiffraction was used.

The ring-core technique was used by Framatome ANP to determine thelocal weld residual stresses (WRS) in depth as a reference for the neutrondiffraction technique. For this purpose, surface strains are measured usingstrain gauges rosettes. A ring is machined around the strain-gauge-rosette byelectric discharge machining (EDM). The removal of materials around therosette leads to an elastic relaxation of the remaining core, on top of which


the strain-gauge is applied. With this procedure a depth profile of residualstresses can be determined. Measurements were performed locating the centreof the strain gauge at 1 mm from the fusion line. Three locations along theplate were measured: the centre and the two ends of the plate. More detailsabout the technique were published previously [6].

Neutron diffraction was used to determine the weld residual stresses beforeand after the irradiation. This technique is a powerful tool for non-destructivemeasurement of residual stresses deep within crystalline materials at reasonablespatial resolution. Bragg’s law is applied, which relates the average latticespacing at a point in a component, from which neutrons with a certainwavelength are diffracted. By measuring changes in the lattice spacing, onecan determine the residual strains, and then derive stresses. Measurements ina sufficient number of directions – normally at least three – have to be made.Facilities at the HFR of the JRC in Petten, NL, and at PSI in Villigen, CH,were used for these measurements.

4.2.5 Stress corrosion tests

To evaluate the susceptibility of the materials to stress corrosion cracking,SSRT tests were performed at a strain rate of 3.5 ¥ 10–7 s–1. Flat tensilespecimens with a gauge length of 16 mm and 1 mm thickness were fabricated.Samples were machined to contain part of the weld metal, the HAZ and thebase metal in the gauge length, Fig. 4.2.

The unirradiated materials and materials having two levels of irradiation(0.3 and 1.0 dpa), respectively, were tested or will be tested. By the time ofthis paper only results of the unirradiated material and the irradiated materialwith 0.3 dpa were available. SSRT are performed using ‘as welded’ materials.However, in the unirradiated condition, some samples of 304 and 347 SSwere tested with a post weld heat treatment (PWHT): 580 ∞C, 24 hours for304 SS and 450 ∞C, 24 hours for 347 SS. Tests have been carried out at 290∞C and 90 MPa, in pure water with inlet conductivity less than 0.1 mS/cm

16

47

15.5 15.5

13

18

∆4

WeldR3.5

4.2 SSRT sample geometry (mm). Source: CIEMAT.


and 200 ppb of dissolved oxygen (DO). In addition, slow strain rate tests atthe same temperature were performed in argon gas. Table 4.2 presents thetest matrix.

SSRT are carried out in a refreshed autoclave incorporated to a hightemperature/high pressure loop suitable to perform SCC with irradiatedmaterials. During the tests, stress and strain of every sample are recorded, aswell as water chemistry parameters and environmental conditions. All thetests have been carried out to the sample fracture at high temperature. Aftertesting, the fracture surface and the gauge length of the tested samples areexamined by Scanning Electron Microscopy (SEM), to identify the presenceof secondary cracking and the cracking morphology and to quantify thepercentage of Intergranular (IG), Transgranular (TG) and Ductile (D) fracturemode.

4.3 Results and discussion

4.3.1 Material characterization

The microstructure of the base metals for both 304 and 347 SS contains afully austenitic matrix with annealing twins. The number of twins decreasesin the heat affected zone. No grain boundary precipitation was detected inboth materials at any location. The grain size was 40 mm in the base metalof 347 SS and 50 mm in the base metal of 304 SS. In the heat affected zonean increase of grain size was observed: 60 mm for 347 and 65 mm for 304.The width of HAZ in terms of grain size variation can be estimated toapproximately 600 mm.

TEM observations of both unirradiated materials indicate a low dislocationdensity, typical for austenitic stainless steels (1013 m–2 for base metal 347 SSand 1014 m–2 for HAZ 347 SS). Some stacking faults and small precipitateswere found. Close to the fusion line recrystallized regions are observed,surrounded by a matrix with a high dislocation density. In general, in bothmaterials the dislocation density increases with decreasing distance from thefusion line. Welded 304 SS presents a slightly higher dislocation densitythan welded 347 SS. Some deformed samples were examined and the resultsindicate that the deformation takes place predominantly by twinning at roomtemperature and by dislocation motion at 300 ∞C, both in base metal and

Table 4.2 Slow strain rate test matrix for welded 304 and 347 materials

Unirradiated 0.3 dpaEnvironment As welded

PWHT As welded

Pure water + 200 ppb O2 x x xInert gas x x


HAZ of 304 and 347 weldments. The characterization of irradiated materialsis still in progress.

Vickers micro-hardnesses of unirradiated welded 304 and 347 SS, from 0to 3000 mm of the fusion line were determined. The hardness values rangefrom 200 to 250 Hv, but these variations does not allow the determination ofthe HAZ, in terms of hardness modifications.

Mechanical properties of unirradiated 304 and 347 SS were determined atroom temperature and at 300 ∞C, both in the base metal and HAZ. Significantmechanical properties variations are observed from base metal to the HAZ inboth materials. In general, an increase of yield strength (YS) is observed inthe HAZ, Fig 4.3, while ultimate tensile strength (UTS) and uniform elongation(UE) present no significant differences along the fusion line. In the basemetal, the YS of 347 SS, at both room temperature and 300 ∞C, is higher thanthe YS of 304 SS. However, the YS of 304 and 347 SS in the HAZ arecomparable.

Results of sensitization measurements show that all EPR values rangefrom 0.05 to 0.45%. Metallographic observations of the sample surfacesafter the EPR tests do not indicate specific grain boundary attack. Accordingly,no sensitization of the materials due to the welding process was produced in304 and 347 SS. Similar results were obtained by Auger Electron Spectroscopyused for grain boundary characterization of the unirradiated 304 and 347HAZ. No significant differences in the respective average of alloying elementsconcentrations, iron, nickel, chromium, in the intergranular area and theductile areas were observed in both materials. A slight segregation of

300025002000150010005000–500–1000Distance from the fusion line (mm)

600

500

400

300

200

100

0

Yie

ld s

tren

gth

(M

Pa)

Yield strength dependence of AISI 304 as afunction of the distance from the fusion line.

Ttest = 573 Ke = 7.46 ¥ 10–4 s–1

4.3 Yield strength of 304 SS as a function of distance from fusionline, at 300 ∞C. Source: PSI.


phosphorus was observed in a low percentage of the intergranular areasof 347 and 304. Examination of irradiated samples will indicate thepossible modification of the grain boundary microchemistry induced byirradiation.

4.3.2 Residual stresses measurements

Residual stresses measurements by ring-core technique indicate no significantdifferences in the residual stress state of the two materials, 304 and 347 SS.Figure 4.4 shows an example of the stresses measured by this technique. Ingeneral, the profiles show prevailing tensile stresses although at some locationscompressive stresses were also observed. The stresses parallel to the fusionline are significantly higher than stresses perpendicular to the fusion line.The stresses on the face containing the last welding pass are also significantlyhigher than those on the opposite face. During the sectioning of the platesinto strip samples for irradiation and corrosion testing, a significant amountof residual stresses were also released.

Residual stress measurements by neutron diffraction were performed inwelded 347 and 304 plates. Results on 347 SS indicate that the largestresidual stresses are in the welding longitudinal direction. Stresses as high as400 MPa were found in the long 347 SS specimens, while the shorter specimens,corresponding in size to the specimens irradiated at the HFR, exhibitedsignificantly lower longitudinal stresses – up to 200 MPa, Fig. 4.5. Welding

304.B.1 (centre/face 2 = welding pass #4)304.E (Y) (edge/face 2 = welding pass #4)304.F (X) (edge/face 2 = welding pass #4)304.B.1 (centre/face 1 = welding pass #5)304.E (Y) (edge/face 1 = welding pass #5)304.F (X) (edge/face 1 = welding pass #5)

0.0 1.0 2.0 3.0 4.0 5.0 6.0Depth (mm

400

300

200

100

0

–100

–200

Res

idu

al s

tres

sp

aral

lel t

o w

eld

(M

Pa)

4.4 Depth profiles of the residual stress parallel to the weld directionin the HAZ in ‘as welded’ 304 SS. Source: Framatome ANP.


transverse and plate normal stresses were found to be far less significant andhardly influenced by shortening of plates.

Measurements on welded 304 SS show a large scatter, attributed to thegrain size of the material. Nevertheless, it will be necessary to test, whetherthis situation changes during irradiation.

4.3.3 Stress corrosion cracking

Typical stress/strain curves of unirradiated and irradiated (0.3 dpa) 304 and347 SS, obtained at 290 ∞C in oxidizing water and inert gas, can be observedin Figs 4.6 and 4.7. In the case of unirradiated 347 SS, no differences werefound in the curves obtained in water and in inert gas. However, in the caseof unirradiated 304 SS lower strain to failure and maximum stress werefound in the samples tested in water than in the samples tested in inert gas.

Some samples of unirradiated 304 and 347 SS were tested with a PWHT.The results indicate no significant differences in the behaviour of both materialsin ‘as welded’ and PWHT conditions [6].

Concerning the irradiated material (0.3 dpa), Figs 4.6 and 4.7 show thatneutron irradiation produces an increase of YS and maximum stress and areduction of elongation in both materials. This hardening induced by neutronradiation is a well known effect on austenitic stainless steels. Accompanyingan increase in hardness is a decrease in the ductility and fracture toughness.The YS of 300 series stainless steels at 300 ∞C can reach 5 times the unirradiatedvalues for 7–10 dpa [7]. In fact, the increase in yield strength follows asquare root dependence on dose. For a neutron fluence of 0.3 dpa, bothmaterials, welded 304 and 347 SS, show a YS around 400–420 MPa. Thesevalues agree with the general tendency for 300 series stainless steels.

Longitudinal 270 mm

Longitudinal 30 mm

Longitudinal 4 mm

0 15 30 45–15–30–45Position (mm)

–200

–100

0

100

200

300

400

500

Res

idu

al s

tres

s (M

Pa)

4.5 As welded 347 SS: Mid-thickness longitudinal residual stressesfor varying lengths of plate. Source: JRC.


Unirradiated samples of 304 and 347 SS tested in inert gas show 100% ofductile fracture. However, samples of unirradiated 304 SS tested in purewater + 200 ppb DO, show some percentage of transgranular fracture, Fig.4.8, while samples of unirradiated 347 SS tested in the same water environmentalways present ductile fracture.

In general, the transgranular fracture observed in non-sensitized materials

AISI 304 Unirradiated200 ppb DO Inert gas

200 ppb DO Inert gasIrradiated 102

87

73

58

44

29

15

0

Str

ess

(ksi

)

50454035302520151050% Strain

700

600

500

400

300

200

100

0

Str

ess

(MP

a)

4.6 Stress/strain curve of 304 SS tested at 290 ∞C. Source: CIEMAT.

AISI 347 Unirradiated200 ppb DO Inert gas

200 ppb DO Inert gasIrradiated

50454035302520151050% strain

102

87

73

58

44

29

15

0

Str

ess

(ksi

)

700

600

500

400

300

200

100

0

Str

ess

(MP

a)

4.7 Stress/strain curve of 347 SS tested at 290 ∞C. Source: CIEMAT.


tested in high temperature water may be considered as an artifact of theextreme severe mechanical conditions during SSRT [8]. However, similarsevere mechanical conditions are achieved during testing of 347 SS and notransgranular cracking was observed. Probably, elongated inclusions in 304SS, observed in the fracture surface of tested samples, could have someinfluence on the initiation and propagation of the transgranular fracture.

Fracture surfaces of irradiated (0.3 dpa) 347 SS samples, tested at 290 ∞Cin inert gas and in pure water show 100% of ductile fracture. However, thefracture surface of irradiated (0.3 dpa) 304 SS tested in pure water + 200 ppbDO, shows some percentage of transgranular fracture, Fig. 4.9, while samplestested in inert gas show ductile fracture. No intergranular fracture was detectedin any of the tested samples.

According to the literature [9], a threshold neutron fluence of around 5 ¥1020 n/cm2 (equivalent to appr. 1 dpa) is necessary to induce any susceptibilityto intergranular cracking in annealed material. In spite of the hardeningobserved in the material, irradiated 304 and 347 weldments with 0.3 dpa donot show any intergranular cracking in the tested environment. Future SSRTtests with welded 304 and 347 SS irradiated to 1 dpa will show if thesematerials are susceptible or not to intergranular cracking.

4.4 Conclusions

The evolution of the weld residual stresses, microstructure mechanicalproperties and stress corrosion behaviour, induced by neutron irradiation, isbeing evaluated in the framework of the INTERWELD project. Material

4.9 Fracture surface of irradiated (0.3 dpa) 304 SS tested at 290 ∞C inpure water + 200 ppb DO. Source: CIEMAT.

25% Transgranular75% Ductile

14% Transgranular86% Ductile

4.8 Fracture surface of unirradiated 304 materials tested at 290 ∞C inpure water + 200 ppb DO. Source: CIEMAT.


characterization, residual stress measurements and stress corrosion crackingtests were performed and are partially still in progress. Preliminary resultsindicate that in spite of the hardening observed in the material, the irradiated304 and 347 weldments with 0.3 dpa do not show any susceptibility tointergranular cracking at 290 ∞C in pure water with 200 ppb of dissolvedoxygen, based on the results of SSRT.

4.5 References

1. S. Bruemmer. ‘New issues concerning radiation induced materials changes and irradiationassisted stress corrosion cracking in LWR’. 10th Int. Symp. on Environ. Degradation.of Materials in NPS-Water Reactors. Lake Tahoe (2001).

2. P. Scott, M. Meurier, D. Deydier, S. Silvestre, A. Trency. ‘An analysis of BaffleFormer Bolt cracking in French PWRs’. Environmental Assisted Cracking: PredictiveMethods for Risk Assessment and Evaluation of Materials Equipment and Structures.ASTM 1410. West Conshohocken, (2000).

3. T.M. Angeliu, P. Andresen, E. Hall, J.A. Sutliff, S. Sitzman, R.M. Horn. ‘IGSCC ofunsensitized stainless steels in BWR environments’. 9th Int. Symp. on EnvironmentalDegradation of Materials in NPS-Water Reactors. TSM. 1999, p. 311.

4. H.M. Chung, J.H. Park, W. Ruther, R. Strain, J. Sanecki, N, Zaluzec. ‘Crackingmechanism of type 304L stainless steels core shroud welds’. 9th Int. Symp. onEnvironmental Degradation. of Materials in NPS-Water Reactor. TSM. 1999, p. 973.

5. B. Van der Schaaf, A. Roth, C. Ohms, D. Gavillet, S. Van Dyck, M.L. Castaño.‘Irradiation effect on the evolution of the microstructure, properties and residual stressin the heat affected zone of stainless steels (INTERWELD)’. FISA-2003. EU Researchin Reactor Safety, Luxembourg, 10–13 November 2003.

6. A. Roth, B. Van der Schaaf, M.L. Castaño, C. Ohms, D. Gavillet, S. Van Dyck.‘INTERWELD: European Project to determine Irradiation Induced Materials Changesin the HAZ of Austenitic Stainless Steel Welds that influence the Stress CorrosionBehaviour in High Temperature Water’, MPA-Seminar, 9–10, October 2003.

7. G. Was. ‘Recent developments in understanding Irradiation Assisted Stress corrosionCracking’. 11th Intern. Symp. on Environ. Degradation of Materials in NPS-WaterReactors. TSM. Stevenson, WA. August 2003, p. 965.

8. A. Jenssen, L. Ljungberg. ‘IASCC of stainless steels alloys in BWR normal waterchemistry and hydrogen water chemistry’. 6th Int. Symp. on Environ. Degradation ofMaterials in NPS-Water Reactors. TSM. S. Diego, CA, August 1993, p. 547.

9. M. Kodama et al, ‘IASCC susceptibility of austenitic stainless steels irradiated to highneutron fluence’. 6th Intern. Symp. On Environ. Degradation of Material in NPS-Water Reactor. TSM. S. Diego , CA, August 1993, p. 583.

70

5.1 Introduction

Metallic lead or lead compounds have been identified in many cases asstress corrosion cracking initiators in tubing materials of operating PWRsteam generators [1–8]. However almost no information is available in literatureabout the susceptibility of low alloy steels to PbSCC in deaerated hightemperature water environments.

As the primary collector bodies of WWER 1000 steam generator, made ofa low alloy steel of type 10GN2MFA, rank among the most stress and corrosionexposed component of the horizontal steam generator, a research programmeinvolving the study of lead (in concentrations 1 ppm, 10 ppm and 100 ppm)on resistance of the steel to SCC was realized. Slow strain rate tests offatigue precracked C(T) specimens oriented in L-R direction were used fordetermination of the beginning of the stable crack growth and the averageenvironmentally assisted crack growth rate. Static autoclave tests at 278 ∞Cwere performed in water solutions simulating crevice environments havingpH278 5.5 and pH278 7.02.

5.2 Testing material

All test specimens were extracted from the bottom part of WWER 1000primary collector body made of 10GN2MFA low alloy steel with bainiticmicrostructure. Its chemical composition is shown in Table 5.1.

Tensile properties (in longitudinal direction) of the studied steel at laboratorytemperature and at 278 ∞C are summarized in Table 5.2.

5The effect of lead on resistance of lowalloy steel to SCC in high temperature

water environments

K. M AT O C H A and G. R O Ž N O V S K Á, VÍTKOVICE,Czech Republic and V. H A N U S, NPP Czech Republic

Table 5.1 Chemical composition of the studied steel (wt%)

C Mn Si P S Cu Ni Cr Mo V

0.09 0.89 0.24 0.011 0.008 0.07 2.23 0.22 0.48 0.04

The effect of lead on resistance of low alloy steel to SCC 71


SCC tests in deaerated high temperature water solutions were performed instatic autoclave 11 l in volume fitted with INOVA servohydraulic testingmachine. Chemical composition of the modeled crevice solutions investigatedis shown in Table 5.3. Pb2+ cations in concentrations of 1 ppm, 10 ppm and100 ppm were introduced as PbO. High temperature pH of water solutionswas calculated using MULTEQ programme in UJV Rez near Prague.

Slow strain rate tests of C(T) specimens were carried out under strokecontrol at a stroke rate of 9.2 ¥ 10–7 mm/s. The initiation of stable crackgrowth was monitored by the AC Potential Drop method. The averageenvironmentally assisted crack growth rate was calculated as

Va a

tCORair =

– D D

where t is the time from the beginning of the stable crack growth to the endof the test.

Fracture surfaces created by stable crack growth during autoclave testswere examined by scanning electron microscope.


The fracture behaviour of the studied steel in air at temperatures rangingfrom 20 ∞C to 300 ∞C is characterized by a ductile stable crack growth. Thevariation in d (crack tip opening displacement) with crack advance Da wasinvestigated in air at 290 ∞C using a multiple specimen method [9]. Theinitiation of stable crack growth expressed like stress intensity factor Kd forDa = 0.05 mm was found to be (Kd)in = 134 MPam1/2.

Table 5.2 Tensile properties of the studied steel (test specimens 3 mm indiameter, cross head speed 0.2 mm/min.)

Test temperature YS [MPa] UTS [MPa] Z [%]

+20 ∞C 518 620 75278 ∞C 439 580 72

Table 5.3 Chemical composition of modeled crevice solutions investigated

pH278 NaCl Na2SO4 H2SO4

7.0 329.9 ppm 739.6 ppm –5.5 329.9 ppm 739.6 ppm 102.1 ppm


Results of slow strain rate tests on fatigue precracked C(T) specimens insolutions with varying Pb2+ are shown in Table 5.4.

The results presented in this table demonstrate that lead cations (Pb2+)enhance the SCC susceptibility of 10GN2MFA low alloy steel in both hightemperature modeled crevice solutions. The rise of Pb2+ concentration from10 to 100 ppm in the water solution with pH278 7.02 caused the decrease ofthe of stable crack growth initiation and the increase of VCOR. However thesteel under investigation was not found susceptible to SCC in this solution inthe case where Pb2+ £ 1 ppm.

The susceptibility of the steel to SCC was also found to be affected by thehigh temperature pH of the solution. The alteration in high temperature pHfrom 7.02 to 5.5 entailed the decrease of the stable crack growth initiationfrom (Kd)in = 123 MPam1/2 to (Kd)in = 66 MPam1/2 and the subcritical crackgrowth due to SCC.

The increased concentrations of Pb2+ ions had almost no effect on thebeginning of the stable crack growth but affected significantly the averagecrack growth rate. Fractographic analysis of the fracture surfaces created bySCC in dearated solutions (see Fig. 5.1) revealed the same fractographicfeatures formerly observed on fracture surfaces of test specimens tested inaerated distilled water [9].

The higher crack growth rates were accompanied by the significantoccurrence of transverse microcracks on fracture surface (see Fig. 5.2, Table5.5).

Metallographic evaluation of test specimen cross sections showed that thetransverse microcracks were initiated in the process zone ahead of the growingcrack and probably contributed to the increase of crack growth rate (seeFig. 5.3).

Table 5.4 Results of slow strain rate tests of fatigue precracked C(T) spec.,u = 9.2 ¥ 10–7mm/s

Modeled Pb2+ a0 Vpl Da d (Kd)in VCOR

solution [mm] [mm] [mm] [mm] [MPam] [mm/s]

Secondary 0 24.17 0.21 0.07 0.12 138 0waterpH278 7.02 0 25.24 0.10 0.05 0.08 123 0pH278 7.02 1 ppm 25.24 0.25 0.09 0.13 141 0pH278 7.02 10 ppm 23.83 0.23 0.23 0.13 90 2.9 ¥ 10–7

pH278 7.02 100 ppm 24.10 0.24 2.60 0.12 75 6.3 ¥ 10–6

pH278 5.5 0 23.39 0.18 0.32 0.11 66 5.1 ¥ 10–7

pH278 5.5 1 ppm 25.00 0.18 0.67 0.10 62 1.1 ¥ 10–6

pH278 5.5 10 ppm 24.10 0.25 0.33 0.13 64 4.1 ¥ 10–7

pH278 5.5 100 ppm 24.00 0.34 6.46 0.13 58 1.3 ¥ 10–5


5.1 Transgranular brittle fracture with river patterns. Modeled crevicesolution, pH278 5.5, u = 9.2 ¥ 10–7 mm/s, t = 278 ∞C, 0 ppm Pb2+

5.2 Modeled crevice solution, pH278 7.02, u = 9.2 ¥ 10–7 mm/s,t = 278 ∞C, 100 ppm Pb2+.


5.5 Conclusions

From the results obtained in this study it follows that:

∑ The susceptibility of 10GN2MFA low alloy steel to SCC in deaeratedhigh temperature water environments with increased concentrations ofCl–, Na+ and SO4

2– is affected by high temperature pH.∑ Lead cations (Pb2+) enhance the SCC susceptibility of the steel in both

solutions investigated (pH278 7.02 and pH278 5.5).∑ The presence of Pb2+ affected both the beginning of stable crack growth

and VCOR.∑ The higher crack growth rates were accompanied by the significant

occurrence of transverse microcracks probably initiated in the processzone ahead of the growing crack.

Table 5.5 The relation between environmentally assisted crack growthrate and the frequency of transverse microcrack occurrence

Modeled solution VCOR [mm/s] Frequency of microcrackoccurrence

pH278 7.02, 100 ppm Pb2+ 6.3 ¥ 10–6 significantpH278 5.5, 0 ppm Pb2+ 5.1 ¥ 10–7 isolatedpH278 5.5, 1 ppm Pb2+ 1.1 ¥ 10–6 significantpH278 5.5, 10 ppm Pb2+ 4.1 ¥ 10–7 isolatedpH278 5.5, 100 ppm Pb2+ 1.3 ¥ 10–5 significant

5.3 The occurrence of transverse microcracks on fracture surface andin process zone of the growing crack.


5.6 Acknowledgement

This work was supported by the Ministry of Education of the Czech Republicthrough project LN00B029.

5.7 References

1. Hélie, M. Lead Assisted Stress Corrosion Cracking of Alloys 600, 690, and 800. Proc.of Sixth International Symposium on Environmental Degradation of Materials inNuclear Power Systems – Water Reactors. Edited by R.E. Gold and E.P. Simonen. Apublication of the Minerals, Metals and Materials Society. ISBN Number0-87339-258-2, 1993, p. 179.

2. Castaňo-Marín, L., Gomez-Briceňo, D., Hernández-Arroyo, F. Influence of LeadContamination on the Stress Corrosion Resistance of Nickel Alloys. Proc. of SixthInternational Symposium on Environmental Degradation of Materials in Nuclear PowerSystems – Water Reactors. Edited by R.E. Gold and E.P. Simonen. A publication ofthe Minerals, Metals and Materials Society. ISBN Number 0-87339-258-2, 1993,p. 189.

3. Costa, D. and co-workers. Interaction of Lead with Nickel-Base Alloys 600 and 690.Proc. of Seventh International Symposium on Environmental Degradation of Materialsin Nuclear Power Systems – Water Reactors. NACE International. ISBN:1-877914-95-9, 1995, Vol. 1, p. 199.

4. Wright, M.D and co-workers. Embrittlement of Alloy 400 by Lead in Secondary SideSteam Generator Environments. Proc. of Seventh International Symposium onEnvironmental Degradation of Materials in Nuclear Power Systems – Water Reactors.NACE International, ISBN: 1-877914-95-9, 1995, Vol. 1, p. 209.

5. Chung, K.K. and co-workers. Lead Induced Stress Corrosion Cracking of Alloy 690in High Temperature Water. Proc. of Seventh International Symposium on EnvironmentalDegradation of Materials in Nuclear Power Systems-Water Reactors. NACE International.ISBN: 1-877914-95-9, 1995, Vol. 1, p. 233.

6. Hélie, M., Lambert, I., Santarini, G. Some Considerations about the Possible Mechanismsof Lead Assisted Stress Corrosion Cracking of Steam Generator Tubing. Proc. ofSeventh International Symposium on Environmental Degradation of Materials in NuclearPower Systems – Water Reactors. NACE International, ISBN: 1-877914-95-9, 1995,Vol. 1, p. 247.

7. Hwang, S.S., Kim, K.M., Kim, U. Ch. Stress Corrosion Cracking Aspects of NuclearSteam Generator Tubing Materials in the Water Containing Lead at High Temperature.Proc. of the Eighth International Symposium on Environmental Degradation of Materialsin Nuclear Power Systems – Water Reactors. American Nuclear Society, Inc. ISBN:0-89448-626-8, Vol.1, p. 200.

8. Sarver, J.M., Miglin, B.P. A Parametric Study of the Lead-Induced SCC of Alloy 690.Proc. of the Eighth International Symposium on Environmental Degradation ofMaterials in Nuclear Power Systems – Water Reactors. American Nuclear Society,Inc. ISBN: 0-89448-626-8, Vol. 1, p. 208.

9. Matocha, K., Wozniak, J. Analysis of WWER 1000 Collector Cracking Mechanisms.Topical Meeting on WWER 1000 Steam Generator Integrity, Tokyo, Japan, 25–29September 1995.

76

6.1 Introduction

The factors which have been identified to possibly contribute to IrradiationAssisted Stress Corrosion Cracking (IASCC) of austenitic stainless steels(ASSs) in Pressurized Water Reactors (PWRs), via a synergic effect, may becategorized as either radiation water chemistry (water radiolysis and thesubsequent increase in the electrochemical potential – g heating and localconcentration of ions), direct microstructural effects (radiation inducedhardening – hydrogen and helium embrittlement) irradiation creep or micro-compositional effects (radiation-induced segregation of impurities andredistribution of major alloying elements).

To complement IASCC tests performed in hot cells, a R&D programme iscarried out in CEA and EDF laboratories to investigate separately the effectsof factors which could contribute to IASCC mechanisms. In the frameworkof this study, the influence of cold work on Stress Corrosion Cracking (SCC)of ASSs in PWR primary water is studied to supply additional knowledgeconcerning the contribution of radiation hardening on IASCC of ASSs. Testingof unirradiated materials in primary water conditions is intended to play asignificant role in the attempt to discriminate the specific effect of materialhardening on the susceptibility of ASSs to SCC. Susceptibility of sensitizedASSs to SCC is a well known phenomenon identified in BWR oxidizingenvironments. Long term research carried out in the case of BWR-typeconditions has shown that respective minimum levels of chloride and oxygenare required for SCC susceptibility [1, 2]. Electrochemical potential identifiedin the case of ASSs exposed to hydrogenated PWR are usually considered astoo low to promote SCC susceptibility. Solution annealed ASSs, essentiallyof type AISI 304(L) and AISI 316(L), are generally considered as immune toSCC in hydrogenated primary water and these materials are thus widely usedin PWRs. The specific role of hardening could be thus envisaged as especiallyimportant in the case of IASCC of ASSs in deoxygenated PWR environmentswhere irradiation induced grain boundary depletion of chromium appears

6Effect of cold work hardening on stresscorrosion cracking of stainless steels in

primary water of pressurized water reactors

O. R A Q U E T and E. H E R M S, CEA/Saclay, France andF. V A I L L A N T, T. C O U VA N T and

J. M. B O U R S I E R, EDF/Les Renardières, France

Effect of cold work hardening on stress corrosion cracking 77

not directly able to explain the susceptibility to SCC, the corrosion potentialin these conditions remaining well below the critical cracking potentialrecognized for sensitized ASSs [3]. The detrimental role of cold-work wasclearly recognized in pure water BWR conditions in presence of oxygen [4–9]. Some rather recent data [10, 11] have then demonstrated that ASS specimensincluding a cold deformed hump were susceptible to SCC in hydrogenatedboric acid environment under CERTs conditions. The objective of this chapteris to detail the results obtained in the framework of the CEA-EDF researchprogramme dealing with SCC of cold-worked ASSs. The main focus of thisprogramme was to determine conditions in terms of types of cold-workleading to the susceptibility of ASSs to SCC in nominal hydrogenated PWRconditions, to define dedicated primary criteria of SCC susceptibility and toexplore the range of susceptibility of these materials.


6.2.1 Materials

Commercial purity AISI 304L plate material was the main material used forthe study. Average grain size is 50 mm and level of ferrite 5%. The chemicalcomposition of the plate materials was measured as 0.026% C, 19.23% Cr,9.45% Ni, 0.17% Cu, 0.24% Mo, 1.49% Mn, 0.52% Si, 0.027% P, 0.002%S, 0.064% N (in wt %). Some tests were performed on commercial AISI316L stainless steels of the following composition: 0.027% C, 17.2% Cr,12.15% Ni, 0.12% Cu, 2.34% Mo, 1.76% Mn, 0.48% Si, 0.23% P, 0.001%S, 0.064% N (in wt %). The samples were fully annealed 1050 ∞C during 30minutes then quenched by argon flux under vacuum. Solution treatment wascarried out before application of any cold-working procedure.

6.2.2 PWR test procedure

All of the tests performed in the framework of this study were carried outinside dedicated static 316L autoclaves including direct measurement ofhydrogen partial pressure via the use of in-situ Ag-Pd probes. Hydrogenconcentrations tested were located inside the range of 25 to 35 cc/kgH2O.STP.The level of pollutants (chloride, fluoride, sulfate) is controlled after eachtest in autoclave via ionic chromatography in order to verify that concentrationof each pollutant remain below the specified value of 50–100 ppb.

Analysis of the fracture surface was performed using Scanning ElectronMicroscopy (SEM) in order to determine the morphology of fracture surfaceand the average depth of SCC. When mentioned, the average SCC growthrate was defined and calculated as follows: average SCC growth rate (mm/hour) = (maximum crack depth on fracture surface (mm)) / (time to failure(hours)).


6.2.3 CERTs on specimens not cold-worked

As a basis regarding susceptibility of the material to SCC in PWR conditionsfor the whole study, CERT test was performed on a smooth specimen of AISI304L ASS not cold-worked before tensile testing. The cold-worked layerassociated with previous manufacturing was suppressed by full annealing.The specimen was additionally electropolished before tensile testing. Theinitial surface hardness of the material in these conditions is lower than150HV (0.98N). CERT test was performed in nominal primary water conditionswith a deformation rate of 1.1 ¥ 10–7s–1 at 360 ∞C. A reference test on a samespecimen was additionally performed in inert gas at 360 ∞C.

In order to analyze the incidence of stress triaxiality on the susceptibilityof material to SCC, some tests were performed on smooth specimenscomprising a circumferential V-notch perpendicular to the tensile axis. Thesespecimens were fully annealed after the machining of the V-notch and thusno residual cold-work is intended.

6.2.4 CERTs on cold pressed V-humped specimens

The specimens cold-worked by a V-hump were manufactured from AISI304L and AISI 316L of the composition mentioned above. These sampleswere prepared from initial flat tensile specimens fully annealed. The colddeformed hump was performed in the centre of the gauge length by the useof a dedicated die. The velocity of the crosshead of the tensile machine wasfixed to 2 ¥ 10–4 mm.min–1 during these experiments which corresponds toa local deformation rate of 10–6 s–1 inside the hump. The maximum measuredinitial surface hardness is located at the inlet of the V-hump and correspondsto a value of 340HV (1.96N) before tensile tests for the 304L material.

In order to complement this study, V-humped specimens were preparedfrom 304L plates initially cold-worked by cross-rolling (reduction of thickness89%, initial surface hardness 380HV). These specimens were then tested byCERT in primary water at 360 ∞C.

6.2.5 CERTs on specimens cold-worked by fatigue

These specimens were extracted from 8 mm diameter, 16 mm long samplespreviously cold-worked by fatigue at ambient temperature. The characteristicsof the fatigue cold-working applied in the first stage are the following:tensile-compressive loading cycle, total deformation ± 2.4%, extension rate4 ¥ 10–3 s–1, 50 cycles. This treatment lead to a uniform cold-work inside the4 mm diameter specimens extracted from the above samples. The resultingsurface hardness is 320HV (1.96N) and Ferriscope® measurements showedformation of 6% of martensite phase from austenite in the cold-worked


material. Then specimens were tested in CERTs at 360 ∞C and for a constantextension rate of 1 ¥ 10–7 s–1. In order to obtain references on the role ofmartensitic phases formed during cold-working regarding the susceptibilityof the material to SCC, some specimens were extracted from samples previouslycold-worked by fatigue at 200 ∞C. At this temperature formation of martensiteby cold-working is indeed not intended. In this latter case, the initial surfacehardness of the material is lower compared to specimens cold-worked atlaboratory temperature 255HV (1.96N) due to the absence of martensitictransformation. Due to the reduction of yield strength with temperature, theincrease of the dislocation density generated by fatigue loading is higherinside the specimen cold-worked at 200 ∞C.

The specimens cold-worked by fatigue were tested by CERT at 360 ∞Cand 1 ¥ 10–7 s–1. All specimens were manufactured from AISI 304L of thecomposition mentioned above.

6.2.6 CERTs on specimens cold-worked by countersinking and cold rolling

For comparison purposes, some specimens were cold-worked by conventionalmachining and rolling. The procedure in the latter case is a cross-rolling witha final reduction of thickness to 89%. The resulting hardness after cold-working is 380HV (1.96N) in the whole specimen thickness. The specimenwas then tested in CERT at 360 ∞C and 2.5 ¥ 10–8 s–1.

The surface hardness resulting from cold-working by counter sinking is400HV (0.98N), the depth of the cold-worked layer being roughly 250 mm.The specimen was then tested in CERT at 360 ∞C and 1 ¥ 10–7 s–1. Allspecimens were manufactured from AISI 304L of the composition mentionedabove.

6.2.7 CERTs on specimens cold-worked by shotpeening

Shot-peening treatment was selected in order to reproduce practical cases ofsuperficial cold-working. The procedure of shot-peening provided a highinitial surface hardness of 474HV (0.49N). The material hardness decreaseswith depth as shown in Fig. 6.1 (profile of micro-hardness measurements).The cold-worked layer has a total depth of 250 mm. The maximum residualstresses in surface of the material were in the range of 850 MPa (compressive).In the aim to study the specific effect of the initial surface hardness onmaterial susceptibility to SCC, dedicated specimens were prepared fromshot-peened samples in order to obtain progressive lower surface hardness.Electropolishing was used to eliminate a part of the outer hard layer obtainedby shot-peening and obtain test specimens with a respective initial surface


hardness of 300HV (0.49N) and 270HV (0.49N). All specimens were madefrom AISI 304L of the composition above and were tested by CERT at360 ∞C and 1 ¥ 10–7 s–1.

6.2.8 Constant load tests

These tests were performed on AISI 304L samples of the compositionmentioned above.

The procedure for specimen cold-working is based on theses used forCERTs. The objective of the long-term constant load tests described here isto assess the susceptibility of ASS material under static conditions. Oneconstant load test was performed on a specimen cold-worked by shot-peening(initial surface hardness 474HV (0.49N)). The load applied is 550 MPa (coreof sample) which corresponds to a total deformation of 18%.

The second test was performed on a cylindrical smooth specimen cold-worked by fatigue at ambient temperature (see procedure above). An additionalcircumferential V-notch was machined around the specimen to increase theseverity of the test. The initial surface hardness before testing was 340HV(1.96N). The load applied corresponds to 80% of the maximum value reachedduring previous CERT tests on the circumferential V-notched specimensmentioned above. The constant load tests were carried out at 360 ∞C innominal primary water conditions.

6.2.9 Constant deformation tests

Systematic constant deformation tests were performed on AISI 304L and316L U-bends and 4-points bending specimens (respective deformation at

6.1 Evolution of material hardness in the case of specimens cold-worked by shot-peening, note the initial high surface hardness.

100

304L CW by shot-peening

500 150 200 250 300Depth (mm)

Har

dn

ess

(HV

0.4

8N)

500

400

300

200

100

0


the apex 10% and 4%). In the case of each material the following previouscold-working modes were applied: shot-peening (same procedure thanmentioned above), rolling (50% reduction of thickness), bending and tensiledeformation (20%). Additionally, U-bends of each material were preparedfrom plates not previously cold-worked.


6.3.1 Results obtained on not cold-worked specimens

The measured deformation to failure for the annealed specimen tested (CERT,1 ¥ 10–7 s–1) in primary water is 39.5%. The fracture surface was ductile.Observation of lateral surfaces however, showed the presence of smallintergranular defects located in the striction area of the sample. Preparationand observation of a specimen cross section showed that the average depthof these defects is always below 30 mm. Observation of a similar specimentested in inert gas did not allow detection of a similar type of defect. A slightintergranular initiation appears then to be possible on annealed ASS materialsin hydrogenated primary water under severe testing conditions (CERTconditions, striction area) but no propagation occurred. Existence of strain/stress localization and triaxiality were, on the contrary, identified as havinga decisive effect on material susceptibility to cracking. The smooth specimencircumferentially notched and annealed revealed a great susceptibility tocracking by SCC when tested in CERT conditions. The initiation of cracksis located inside the notch and the mode of fracture obtained is intergranularwith a slight transgranular initiation area. Conditions of stress/strain triaxialityon annealed specimen appear thus sufficient to lead to a noticeable susceptibilityto cracking by SCC in PWR primary water conditions.

6.3.2 Influence of the cold-work process

Tests on V-humped specimens

Tests carried out on V-humped specimens confirmed the strong susceptibilityof ASSs cold deformed in these conditions. The deepest crack reached morethan 1 mm for a test duration of 660 hours which corresponds to an averagecrack growth rate of 1.5 mm/h. Similarly to what was observed in the case ofcircumferentially notched specimens, the mode of fracture is fully intergranularfor AISI 304L with noticeable traces of plasticity observed on the face ofgrain boundaries (traces of slip bands). In the case of AISI 316L, the modeof fracture obtained is a fully river pattern transgranular SCC mode.

As mentioned above, conditions of triaxiality of stress/strain under dynamicdeformation conditions (CERT) strongly favours the susceptibility of ASSsto SCC. A cold working process including an initial compressive stress state


(V-hump) appears to promote susceptibility of materials to cracking underCERT conditions.

On the contrary, V-humped specimens prepared from 304L plate cold-worked by cross-rolling and then tested by CERT at 360 ∞C in primary waterdid not show any susceptibility to SCC. The fracture surface obtained aftertests showed only dimple ductile fracture. Excessive cold-work (reductionof thickness 89% was obtained by cross-rolling) seems to prevent any SCCsusceptibility of the material in some specific conditions. The cold-workitself is then not intrinsically a decisive factor of sensitization of ASS materialto cracking by SCC in PWRs conditions. The susceptibility to SCC of V-humped stainless steels samples to SCC was originally observed by Smialowskaet al. [10, 11] in borated, hydrogenated water at 350 ∞C and more recently byArioka [12, 13] and Kaneshima [14] after CERTs on V-humped specimensconstructed from 316L stainless steel. The fracture mode observed byKaneshima on 316L after CERTs at 360 ∞C also consists of transgranularcracking but with some local intergranular area on samples previously annealed.Some specimens were also initially cold-worked by rolling before themanufacturing of the hump. The authors observed that the amount of SCCbrittle mode decreased when the initial deformation ratio increased, the fracturesurfaces observed being completely ductile when the initial deformationratio was beyond 50%. These observations are fully in agreement with thisstudy where excessively initially cold-worked V-humped specimens (crossrolling) did not show any susceptibility to SCC during CERTs.

The localization of deformation supplemented by the initial presence ofcompressive residual stresses appears as decisive factors promoting SCC ofASSs under dynamic deformation conditions (CERT).

CERTs on specimens cold-worked by fatigue, cold rolling andcounter sinking

Despite the particular severity of the cold-working conditions mentionedabove in the case of specimens cold-worked by counter sinking and coldrolling, CERTs in hydrogenated primary water conditions at 360 ∞C leadonly to ductile fracture. These two types of cold-work do not appear to besusceptible to sensitize the material to SCC. The nature of the cold-workingprocedure is then especially important regarding susceptibility of ASSs toSCC in primary water conditions. On the contrary, AISI 304L previouslycold-worked by fatigue is strongly susceptible to cracking during CERTs.Fracture surfaces of specimens cold-worked by fatigue at ambient temperaturebefore CERT at 360 ∞C showed large SCC propagation zones. The morphologyof cracking was mixed (intergranular and transgranular) with a dominanttransgranular mode of fracture. The measured average crack growth rate wasover 1.6 mm/h. As mentioned above in the case of V-humped samples, a


cold-working procedure including compressive deformation appears as amajor contributor to the SCC susceptibility of ASSs under dynamic deformationconditions (CERTs). Specimens previously cold-worked at 200 ∞C and testedby CERT in primary water at 360 ∞C showed that no major effect of the pre-existing martensite phase on SCC susceptibility exist in the framework ofthese experiments. The extension of cracking is indeed similar on specimenscontaining no martensite (cold-worked by fatigue at 200 ∞C before CERTs)with respect to specimens containing martensite (previously cold-worked atambient temperature). A very limited role of the martensite was similarlyunderlined by Andresen [8] after testing carried out on ASSs in pure waterunder oxygenated and hydrogenated conditions.

CERTs on specimens cold-worked by shot-peening

Specimens cold-worked by shot-peening and tested by CERT in primarywater at 360∞ C showed strong susceptibility to cracking. As mentionedabove, existence of residual compressive stress state before tests promotesusceptibility to SCC. Large SCC fracture mode is identifiable on specimenfracture surfaces. SCC propagation mode is transgranular and the measuredaverage crack growth rate on fracture surfaces was beyond 1 mm/h. A stronginfluence of the initial surface hardness on SCC susceptibility is observed.Figure 6.2 shows the evolution of the average crack growth rate dependingon initial surface hardness measured before CERTs. The average crack growthrate strongly decreases with surface hardness. A criterion of 300 ± 10HV(0.49N) can be proposed as a minimum required regarding SCC crackingsusceptibility of ASSs. This criterion is quite similar to those that were

6.2 Evolution of the average crack growth rate on specimens testedby CERT versus the initial surface hardness, CW by shot-peening,PWR primary water, 360 ∞C.

150

E250

E301312

100500 200 250 300 350 400 450 500Initial surface hardness (HV5)

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

Ave

rag

e cr

ack

gro

wth

rat

e (m

m/h

)

E294


proposed by Tsubota et al. [5] for SCC of ASSs in pure water containingoxygen at 288 ∞C. These authors observed that SCC of 304L under crevicedbent beam conditions (CBB tests) in pure water saturated with oxygen occurredbeyond a critical surface hardness of 270HV.

6.3.3 Influence of the mechanical solicitation

Constant load tests

The specimens tested under constant load in primary water at 360 ∞C did notshow any susceptibility to SCC. Slight initiation was, however, observed inthe case of the specimens cold-worked by shot-peening. The depth of theSCC cracks was observed by SEM inside the notch of the cold-workedspecimen tested under constant load in primary water for 17 000 hours.These latter observations showed that the maximum crack depth reachedwas less than 20 mm. The morphology of cracking is transgranular. Similarbehaviour was observed on shot-peened specimen tested under constant load:initiation of SCC cracks was observed after 4,000 hours of test but no additionalpropagation of these initiated cracks was reported after 7,000 hours of totaltest duration. Observation of cross sections of the tested sample showed thatthe maximum depth of initiated crack was lower than 20 mm.

Constant deformation tests

Similarly to what was observed during constant load tests, susceptibility ofASSs to SCC under constant deformation is particularly weak. No crackingwas observed in the case of all of the previously cold-worked AISI 304Lor 316L samples after 9,000 hours of total test duration in primary water at360 ∞C. Slight, dispersed initiation of SCC cracks was only observed in thecase of AISI 304L and 316L shot-peened specimens after 5,600 hours of testbut no further propagation was reported. It is important to note that thisslight initiation of SCC was then identified only in the case of heavily cold-worked specimens showing the strongest initial surface hardness among allof the cold-working procedures evaluated. No initiation was also reported onnotched specimens.

It is important to note that noticeable SCC of ASSs in primary water ofPWRs was thus only reported in the framework of this study under dynamicdeformation conditions (CERTs). Under static conditions or for an excessivelylow deformation rate corresponding to the natural creep rate at 360 ∞C, nopropagation of SCC cracks was reported but crack initiation identified. Mostof the results available in the open literature dealing with SCC of ASSs inPWRs conditions and where SCC propagation was identified, were obtainedunder dynamic deformation conditions. Crack propagation on pre-cracked


fracture mechanics CT-type specimens was, for instance, obtained undercyclic loading (regular periodic unloading stages) [15] or trapezoidal waveloading [16]. It can be envisaged that a minimum deformation rate should beeventually required to allow crack propagation in the case of cold-workedASSs exposed to PWRs conditions.

6.4 Conclusions

1. Strong SCC susceptibility of ASSs can be observed under dynamicdeformation conditions (CERTs) in hydrogenated primary water of PWRs.This susceptibility to cracking is promoted by cold-work and/or localizationof deformation.

2. Cold-work procedure including compressive stage (fatigue, shot-peening)strongly favours SCC susceptibility in PWRs conditions and under dynamicdeformation conditions.

3. For a given cold-working procedure, SCC susceptibility of ASSs materialsincreases with cold-work. A threshold of susceptibility can be identifiedin the case of the shot-peening procedure and for AISI 304L stainlesssteels in terms of initial surface hardness before CERT. SCC crackpropagation is observed beyond 300HV for shot-peened specimens. Forexcessively severe levels of cold-work by cold-rolling, material does notappear to be susceptible to cracking.

4. SCC initiation but no propagation is identified under static conditions(constant load, constant deformation). Dynamic deformation conditions(CERTs, cyclic loading) appear as a prerequisite for SCC susceptibilityof ASSs in PWRs.

6.5 Acknowledgement

The authors sincerely thank the ‘Conseil Général d’Île de France’ for hiscontribution to the financing of the SEM.

6.6 References

1. B.M. Gordon, Materials Performance, 19, 4, 1980.2. P. Combrade, in Corrosion sous Contrainte: Phénoménologie et mécanismes –

Bombannes, eds D. Desjardins and R. Oltra, 1990.3. P. Scott, Journal of Nuclear Materials, 211, 1994.4. J. Kuniya, I. Masaoka, R. Sasaki, Corrosion, 44, 1, 1988.5. M. Tsubota, Y. Kanazawa, I. Hitoshi, 7th International Symposium on Environmental

Degradation of Materials in Nuclear Power Systems – Water Reactors, 1995.6. H. Hanninen, Effect of sensitization and cold work on stress corrosion susceptibility

of austenitic stainless steels in BWR and PWR conditions, VTT Metals LaboratoryReport 88, May 1981.


7. P. Andresen, T.M. Angeliu, W.R. Catlin, L.M. Young, R.M. Horn, Corrosion 2000conference, paper 203.

8. P. Andresen, T.M. Angeliu, L.M. Young, Corrosion 2001 conference, paper 228.9. M.O. Speidel, R. Magdowski, 9th International Symposium on Environmental

Degradation of Materials in Nuclear Power Systems – Water Reactors, 1999.10. Z. Szkalarska-Smialowska, Z. Xia, Corrosion, 48, 1992.11. S. Sharkawy, Z. Xia, Z. Szkalarska-Smialowska, Journal of Nuclear Materials, 195,

1992.12. K. Arioka, Y. Kaneshima, T. Yamada, T. Terachi, 11th International Symposium on

Environmental Degradation of Materials in Nuclear Power Systems – Water Reactors,2003.

13. K. Arioka, Colloque International Fontevraud 5, September 2002.14. Y. Kaneshima, N. Totsuka, N. Nakajima, 10th International Symposium on

Environmental Degradation of Materials in Nuclear Power Systems – Water Reactors,2002.

15. M.L. Castaño, M.S. Garcia, G. de Diego, D. Gomez-Briceño, L. Francia, ColloqueInternational Fontevraud 5, September 2002.

16. T. Shoji, G. Li, J. Kwon, S. Matshushima, Z. Lu, 11th International Symposium onEnvironmental Degradation of Materials in Nuclear Power Systems – Water Reactors,2003.

87

7.1 Introduction

Austenitic stainless steels (ASS) are characterized by a good resistance togeneral corrosion at elevated temperature, permitting their widespread use inprimary and auxiliary circuits of Pressurised Water Reactors (PWRs). However,some components suffer stress corrosion cracking (SCC) under neutronirradiation. This degradation could be the result of the increase of hardnessor the modification of chemical composition at the grain boundary byirradiation. In order to avoid complex and costly corrosion facilities, theeffects of irradiation on the material are commonly simulated by applying astrain hardening on non-irradiated material prior to stress corrosion crackingtests. Numerous studies have demonstrated the susceptibility to stress corrosioncracking (SCC) of ASS in boiling MgCl2 solution [1] and in Boiling WaterReactor (BWR) environments [2]. Most of SCC tests have been made inorder to find the threshold values for SCC occurrence and to clarify theeffect of plastic pre-deformation. Particularly, transgranular SCC (TGSCC)in boiling MgCl2 solution initiation appears when the amount of elasticenergy reaches a threshold value, making the strain hardening essential forthe initiation and the propagation of SCC [3, 4]. Furthermore, it seems thatsusceptibility to SCC is not a monotonic function of pre-straining in thatenvironment [5]. In BWR environment, intergranular SCC (IGSCC) is alsodeeply correlated to strain hardening: crack growth tests have shown that thecrack growth rate (CGR) increases with the yield strength in several ASS[6]. On the other hand, SCC of ASS in a PWR environment is relativelypoorly known, mainly because of the restricted conditions for the occurrenceof the phenomena in that environment. Currently, the main cause of initiationof SCC in a PWR environment is related to materials sufficiently pre-strained[7–12]. Authors have tried to understand the effect of a pre-straining on SCCin a PWR environment using cold pressed humped specimens [13–15]. Butthe interpretation of slow strain rate tests (SSRTs) with this kind of specimenis not obvious, because of the combined influences of the mechanical parameters

7Effect of strain-path on stress corrosioncracking of AISI 304L stainless steel in

PWR primary environment at 360 ∞C

T. C O U VA N T, F. V A I L L A N T and J. M. B O U R S I E R,EDF R&D - MMC, France and D. D E L A F O S S E,

Ecole des Mines de St-Etienne, France


(pre-strain, stress, strain rate). In this context, this study has been conductedto highlight the effect of pre-shear hardening and strain path on initiation andpropagation of both IG and TGSCC of AISI 304L in PWR environment at360 ∞C. Pre-shearing is not the most common and straightforward methodused to pre-harden materials prior to SCC tests. But this method has allowedreproduction on flat specimens of the sequential deformation at the intradosof cold pressed humped specimens where initiation of SCC has been widelyobserved.


7.2.1 Material

The material is a sheet 30 mm thick. The chemical composition and mechanicalproperties of the austenitic alloy 304L tested in this work are given in Tables7.1 and 7.2 respectively. Mechanical properties are isotropic in the plane ofthe sheet as shown by tensile tests in roll and transverse directions. Nonsensitized 304L is solution annealed at 1150 ∞C and water quenched. Theresultant microstructure is characterized by a grain size of about 60 mm withno evidence of any carbide precipitate in the matrix and along the grainboundaries. The austenitic grain size was measured through metallographicetching (standard ASTM E 112 [16]). It contains less than 5% of residuald-ferrite and is subject to strain-induced martensite transformation (Ms =–133 ∞C, Md30 = –3 ∞C) as predicted by Angel [15]. The quantity of d-ferritemeasured by X-ray diffraction is higher in the middle of the sheet than nearthe skins.

7.2.2 Pre-shear hardening of the material

Pre-shearing tests at 25 ∞C were used to raise the yield strength of thematerial prior to SCC tests. Similarly to tensile test, we can considered thatshear tests lead to an homogeneous strain-hardening with a fair approximation,

Table 7.1 Chemical composition (wt%) of studied 304L

C Si Mn S P Cr Ni Co Ti Cu Al Mo N Fe

0.026 0.52 1.49 0.002 0.027 19.23 9.45 0.07 <0.005 0.17 0.033 0.24 0.064 Bal

Table 7.2 Mechanical properties at 360 ∞C of studied 304L

YS (MPa) UTS (MPa) El. (%) HV0.1

160 450 40 160

Effect of strain-path on stress corrosion cracking of AISI 304L 89

without any localization of the strain. Two samples (1 ¥ 200 ¥ 200 mm) havebeen cut in the middle of the sheet of 304L SS. Samples were silk-screenprinted with a 2 mm step grid, in order to control the homogeneity of theshear at the end of the pre-strain hardening test. Grids were eliminated priorto SCC tests. Pre-shearing were conducted with a frame composed of a fixedbody where a mobile tie was sliding at its centre. The shear rate at theambient air was 2.5 ¥ 10–3 s–1 and the shear amplitudes were respectively 0.2and 0.4 for the first and the second samples. Pre-shearing tests led to theformation of a ¢-martensite (5–7% measured by X-ray diffraction for g = 0.4)and to an increase of hardness (respectively 320 and 340 HV0.1 for g = 0.2and 0.4). Each sample had two gauge lengths (1 ¥ 40 ¥ 200 mm) in whichsecondary specimens have been cut for SCC tests after pre-shearing, forthree directions defined by f = 45∞, 90∞ and 135∞ (Fig. 7.1). Consequently,three strain paths were followed during subsequent SCC tests consisting oftensile tests.

Extensive studies concerning sequential deformations have already beenpublished [16, 17]. Schmitt et al. [18] have proposed a scalar parameter b tocharacterize a two-stage strain path. In our study b was defined as the doublecontracted tensor product between the plastic shear mode e1 during the pre-strain (in air, 25 ∞C) and the subsequent plastic tensile mode e 2 during theSSRT (PWR environment, 360 ∞C):

b e ee e

= :

| | | |1 2

1 2

˜ ˜˜ ˜◊

200 mm

40 mm

Rolling direction

f = 90∞

f = 45∞

f = 135∞

7.1 Machining of secondary specimens in pre-sheared materials inair at 25 ∞C.


For a b value close to 1 (f = 45∞) the strain tensors were almost identicalleading to a pseudo-monotonic test (no important changes were seen in thestress-strain curves). For a value close to –1 (f = 135∞), a reverse test, orpseudo-Baushinger test, was obtained. For a value close to 0 (f = 90∞), thestrain tensors of the two sequential deformation paths were perpendicular,i.e. the double-dot product of these tensors is zero. Then, the sequentialdeformation path led to a cross effect. The micro-mechanical aspects ofstrain path changes are not dealt with in this chapter.

7.2.3 Specimen preparation

Two types of specimens were used in this study. Specimens with 79 mmgauge length (type A) were used for SSRTs on non-pre-strained material.These 2 mm thick specimens were cut in the middle of the sheet, in itstransverse plane. The shape and dimensions of type B specimens (1 mmthick) employed for SSRTs are shown in Fig. 7.2. These secondary specimenswere cut in pre-sheared samples (see previous paragraph). In the middle ofthe gauge length of type B specimens, a double notch was machined by electrondischarge machining. The radius and depth of the notches were 150 mm. 3Dcalculations by finite elements have allowed determination of the strain,stress and triaxiality in the notches vs. the elongation of the specimen.

7.2.4 SSRT procedure

Specimens were ultrasonically rinsed in ethanol and then in distilled water.Tests were carried out in Hastelloy (C-276) autoclaves. Specimens wereisolated from the autoclave by oxidized zircalloy to avoid galvanic coupling.Experiments were conducted under open circuit conditions. The environmentwas primary water (1000 ppm B as boric acid, 2 ppm Li as lithium hydroxide)at 360 ∞C. Solution was previously de-aerated by evaporating 20% of the

∆6.1 mm

25 mm

75 mm

20 mm

12 mm

13.5 mm

3.5 mmr = 5 mm

7.2 Type B specimen for SSRTs on pre-sheared and non-pre-strained304L SS.


initial volume at 125 ∞C, then a hydrogen overpressure was introduced(30 cc/kg) and controlled using a Pd-Ag thimble. SSRTs, were conductedwith an apparent applied strain rate ( )ape of 5 ¥ 10–8 s–1. For notched specimens,e ap was defined as [Ln(1 + dl/l0)]/t, where dl was the elongation of thespecimen, l0 the initial length and t the duration of the test. During SSRTsthe load was measured vs. elongation. At the end of the tests, specimenswere rinsed in distilled water and then microscopically examined in order tofind any SCC. The depth of the main crack was measured on the fracturesurface of failed specimens, using a scanning electron microscope (SEM), oron cross-sections in the case of interrupted tests. The average crack growthrate was estimated by dividing the maximum crack depth by the duration ofthe test.

7.3 Results

7.3.1 Initiation and propagation of TGSCC in nonpre-strained specimens

Five tests have been carried out, four tests were interrupted before the ruptureof the specimen, for various elongations, allowing establishment of arelationship between the depth of the main stress corrosion cracks and thestrain hardening of the material resulting from the SSRT in PWR environment(Fig. 7.3). The observation of the fracture surface of test #657 by SEMpermitted to identify a purely transgranular stress corrosion cracking. Crackswere initiated and propagated on every face of the specimen. Furthermore,

Initiation of TGSCCPropagation of TGSCC

y = 3736.3x – 940.09

y = 311.19x – 32.746

Strain0.400.350.300.250.200.150.100.050.00

450

400

350

300

250

200

150

100

50

0

Dep

th o

f th

e d

eep

est

crac

k (m

m)

7.3 Initiation and propagation stages during SSRTs with non-pre-strained specimens (type A) in PWR environment (360 ∞C). Depth ofthe main crack vs. strain at the end of the test.


a large amount of cracks was equally observed on the gauge surface (ª130cracks.mm–2). More generally, cracks were uniformly distributed forelongations higher than 17%. For short elongations (< 17%), the identificationof the crack path may have been difficult, because depths were systematicallylower than the grain size. Very short cracks (2 mm) were detected for 0.10strain, while the transition from initiation to propagation was observed for acrack depth of 50 mm after 0.27 strain. At the rupture of the specimen (test#657), the elongation (0.34 strain) was rather close to the elongation observedin inert gas. The aTGSCC – e curve allowed to propose, for an apparent strainrate of 5 ¥ 10–8 s–1, a slow crack growth rate in the initiation stage, for astrain in the range 0.10–0.27, and a ‘rapid’ CGR in the propagation stage fora strain above 0.27. The following CGRs were calculated from the slope ofthe two corresponding segments represented on Fig.7.3.

˙

˙

˙

a

aStage1

Stage2

–8 –1

= 0.05 m/h

= 0.52 m/h

= 5 10 s

mm

¥

ÏÌÔ

ÓÔe

The accuracy of the CGR proposed for the initiation stage was clearlyhigher than for the propagation stage. Further tests will be necessary toprecisely ascertain the depth of the transition and the CGR in the rapidpropagation stage (with CT specimens). Crack depth versus the strain hardeningresulting from SSRTs is shown on Fig. 7.4. Strain hardening was quantifiedby Vickers micro-hardness measures near the initiation areas. Thus, a micro-hardness threshold for initiation was found close to 250 HV0,1 and close to310 HV0,1 for propagation. However, it was assumed that micro-hardnessdid not evolve in the vicinity of the edges of the cracks during the propagationstage, except at the crack-tip. So, this hypothesis led us to consider that themicro-hardness measured at the end of the test was representative of the

7.4 Micro-hardness threshold for initiation and propagation of SCCduring SSRTs in PWR environment (360 ∞C, e = 5 ¥ 10–8 s–1).

340300260220180Vickers microhardness at the crack-tip

350

300

250

200

150

100

50

0

Cra

ck d

epth

(mm

)


strain hardening for initiation. Similarly, the evolution of crack depth versusthe equivalent Von Mises stress have been represented on Fig. 7.5, on whichinitiation corresponded to a stress of 430 MPa (0.10 strain) and the transitionto propagation to a stress of 700 MPa (0.34 strain). Additional tests associatedto finite element calculations should precisely ascertain the value of thetransition stress during this kind of test. This stress threshold was above thestress measured by RX diffraction at room temperature at the apex of RUBswhere no initiation has been observed after 15 000 h. Consequently, theovershoot of these thresholds (stress and strain hardening) is not a sufficientcondition for initiation and propagation of SCC of SS, at least for shortduration tests in the laboratory.

7.3.2 Initiation and propagation of SCC during SSRTswith pre-sheared specimens

Type B specimens were used in this part of the study. Except one specimen(test #549), all specimens were double-notched. Above all, pre-shear hardeningassociated with complex strain paths allowed reproduction with flat specimensand a perfect homogeneous pre-strain, the mechanical state followed at theintrados of cold pressed humped specimens. The use of flat specimens alsoallowed a quantitative approach of the effect of strain hardening on SCC, bycontrast to the use of cold pressed humped specimens implying contactproblems, friction, high heterogeneity and weak reproducibility of the strainhardening. Nevertheless, the section of the specimen was limited to 1 mmthick because of technological considerations. Two pre-shearing levels wereemployed (g = 0.2 and g = 0.4) and three strain paths were followed: pseudo-monotonic (b = +1), reverse or pseudo-Baushinger (b = –1) and cross test(b = 0).

160 260 360 460 560 660 760Equivalent V.M. Stress (MPa)

350

300

250

200

150

100

50

0

Cra

ck d

epth

(mm

)

7.5 Stress threshold for initiation and propagation of SCC duringSSRTs in PWR environment (360 ∞C, e = 5 ¥ 10–8 s–1).


A cross test with a pre-deformation of 0.20 in simple shear and a subsequenttensile test led to a necking similar to that resulting from a pseudo-monotonictest subsequent to a 0.4 pre-shearing. After a 0.4 pre-shear, the necking ofspecimens were restricted during the subsequent tensile tests whatever thestrain path. Consequently, specimens had a limited capability of deformation.In Fig. 7.6 elongations to rupture of SCC specimens are shown as a functionof pre-shear and strain path. The equivalent cumulated strain (x-axis) wasthe sum of the equivalent pre-shear (25 ∞C) and the equivalent strain at theend of the SSRT (360 ∞C). Comparing the monotonic test with a test in inertgas at 360 ∞C (not represented on Figure) it could be assumed that elongationsto rupture were fairly reduced because of SCC. Thus, 30% of reduction wasobserved for {g = 0.4; b = –1} with SCC < 140 mm while an elongation of22% was noted for the monotonic test due to highest SCC (300 mm). Finally,the effects of the strain paths on yield stress were minor for g = 0.4, butsignificant for g = 0.2. More precisely, stress-strain curves of reverse testshave confirmed that the material presented a Baushinger behaviour.

Figure 7.7 describes the main stress corrosion crack depth initiated at thenotch of specimens as a function of their elongation and strain path. On thewhole, a strong effect of pre-hardening on SCC was observed and describedin the following paragraphs. Dots corresponding to {g = 0.2; b = 0} were inagreement with those corresponding to {g = 0.2; b = –1}: the shortest elongationto rupture was observed for the cross test (b = 0), which was the most severemechanically. CGRs were calculated in the initiation and propagation stageswhen possible. Thus, the trend of the curve in the case {g = 0.2; b = –1}indicated a CGR transition. Therefore, it was possible to evaluate the CGR,considering tests relating to reverse and cross test with 0.2 pre-shear and

g = 0, b = (+1)g = 0.2, b = –1g = 0.2, b = 0g = 0.4, b = –1g = 0.4, b = 0g = 0.4, b = +1

Cumulated strain0.320.280.240.200.160.120.080.040

600

500

400

300

200

100

0

Eq

uiv

alen

t V

M s

tres

s (M

Pa)

7.6 Stress-strain curves of SSRTs in PWR environment for severalstrain paths (360 ∞C, e = 5 ¥ 10–8 s–1). The equivalent stress and theequivalent cumulated strain are considered. g is the pre-shearing andb is the strain path.


considering an elongation rate of 4.7 mm.h–1. The following CGRs for theinitiation and propagation stages were obtained:

˙

˙

˙

a

aStage1

Stage2

–8 –1

= 0.22 m/h

= 6.10 m/h

= 5 10 s

mm

¥

ÏÌÔ

ÓÔe

These CGRs were clearly higher than those measured in non-pre-strainedspecimens, with a factor 4 for the initiation stage and a factor 12 for thepropagation stage. Therefore, strain hardening clearly led to an increasing ofCGR for any stage.

Effect of strain localization on SCC

In SSRTs, the specimen often fails soon after the initiation of the crackingand little information on crack propagation is obtained, especially in pre-strained materials. That’s why SCC was focused at particular locations bythe use of double notches in the gauge section of the specimen. The effect ofstrain-localization during SSRT appeared when comparing two specimens,one with notches and the other without (tests #549 and #583). Pre-strainhardening and subsequent strain path, characterized by {g = 0.4; b = –1},were identical. Results showed that notches favoured initiation of SCC andespecially TGSCC. Indeed, observations of fracture surfaces revealed amultitude of intergranular short cracks (< 40 mm) initiating at the surface ofthe smooth specimen and a main transgranular stress corrosion crack (80 mm)for the notched specimen. In the notched specimen, a lot of intergranularshort cracks were observed outside the notches. In brief, 0.4 pre-shear hardening

g = 0.2/b = –1g = 0.2/b = 0g = 0/b = (+1)g = 0.4/b = –1g = 0.4/b = 0g = 0.4/b = +1

Elongation (mm)76543210

1000

900

800

700

600

500

400

300

200

100

0

Max

cra

ck d

epth

(mm

)

7.7 Crack depth versus elongation for several strain paths. SSRTs onnotched type B specimens (360 ∞C, e = 5 ¥ 10–8 s–1).


associated with a subsequent reverse SCC test was favourable to intergranularinitiation, but the presence of the notch mostly enhanced TGSCC (Fig. 7.7).

Effect of the strain path on the CGR

Pre-shearing had different implications for trans- and intergranular crackdepths. In fact, the intergranular crack depth was an increasing function ofpre-shear hardening for complex strain paths (see Fig. 7.8). Moreover, forreverse and cross SCC tests, the IG crack depth was independent on thestrain path for 0.2 pre-shear. In return, for 0.4 pre-shear, IG crack was clearlydeeper for the most mechanically severe strain path (cross test or b = 0).Consequently, it could be concluded that a pseudo-monotonic strain pathdoes not significantly favour initiation of IGSCC whatever the pre-shearhardening level. Secondly, it could be assumed that IGSCC was enhanced bysevere strain paths.

Figure 7.9 shows the effect of strain path on transgranular stress corrosioncracks. First, it could be noticed that TGSCC initiated whatever the strainpath. For a pseudo-monotonic strain path (b = +1) the depth of transgranularcracking was independent of the level of the pre-shear hardening (ª 400mm). It means that CGR in a pre-strained material (g = 0.4) was clearly morerapid than in the non-pre-strained one (g = 0), since the durations of the testswere radically different (respectively 537 h and 1368 h for g = 0.4 and 0).Then the case of complex strain paths was observed. A significant reductionof the crack depth for (g = 0.4 ; b π 1) was noticed, compared to the pseudo-monotonic strain path (g = 0.4; b = +1). Thus, the depth of transgranularcracks was five times less important for a complex strain path than for apseudo-monotonic strain path. It was assumed that a complex strain path

b = –1b = 0b = +1

g0.40.30.20.10

160

140

120

100

80

60

40

20

0

Max

. IG

SC

C d

epth

(mm

)

7.8 Intergranular crack depth versus pre-shearing for several strainpaths. SSRTs on notched type B specimens. (360 ∞C, e = 5 ¥ 10–8 s–1).


associated to a significant pre-strain hardening reduced the propagation oftransgranular cracks. For g = 0.2, the two complex strain paths have beendistinguished. TGSCC was favoured by the pre-strain for b = –1, the crackingreached 1 mm depth. The moderate 0.2 pre-shear significantly increased theCGR. For the most severe strain path (b = 0), the depth of crack did notexceeded 400 mm: it could be concluded that a mechanically severe strainpath (b = 0) limited the propagation of a transgranular crack.

Stress and strain hardening thresholds for SCC

The whole results conducted with type B specimens were plotted on Fig.7.10. The first conclusion was that for a complex strain, the micro-hardnessthreshold for propagation of TGSCC was above 315 HV0,1, which was closeto the value estimated with SSRTs on non-pre-strained 304L. As for thepropagation threshold of IGSCC, it was less than 345 HV0,1. Considering thestrain hardening characterized by (g = 0.2; b = –1), we could assume that thestress threshold for apparent initiation was close to 470 MPa and the stressthreshold for propagation was below 680 MPa. These values referred toTGSCC, which was the deepest, and was in good agreement with thosefound on the non-pre-strain hardened material. On the whole, representativepoints related to complex strain paths seemed to be fitted by a curve increasingrapidly with the true equivalent stress, in the range 600–800 MPa.

Effect of triaxiality

The deepest transgranular cracks were observed for the highest values of thestress triaxiality. In particular, the reverse strain path {g = 0,4, b = –1} was

b = –1b = 0b = +1

g0.40.30.20.10

1200

1000

800

600

400

200

0

Max

. TG

SC

C d

epth

(mm

)

7.9 Transgranular crack depth versus pre-shearing for several strainpaths. SSRTs on notched type B specimens. (360 ∞C, e = 5 ¥ 10–8 s–1).


considered, it was noticed that both TGSCC and IGSCC increased significantlywith triaxiality. Nevertheless, data were still missing to conclude on theeffect of triaxiality and to propose any possible threshold for initiation orpropagation of SCC.

7.4 Discussion

First, SSRTs have supported the primary idea that strain hardening was aprerequisite condition for SCC initiation and propagation. Several curveshave led to proposed thresholds for both initiation and propagation. No SCChas been observed for micro-hardness below 240 HV0.1, and no propagationunder 310 HV0.1. Likewise, an equivalent stress close to 700 MPa seemednecessary for propagation. Nevertheless, the overstepping of these thresholdswas not a guarantee for SCC initiation or propagation as demonstrated bysequential tests.

Second, SSRTs have demonstrated the important effect of the strain pathon SCC mechanisms and more precisely on the crack growth path. Briefly,the monotonic strain paths led to pure TGSCC (Fig. 7.11) while complexstrain paths (reverse and cross SCC tests) favoured IGSCC (Fig. 7.12).Furthermore, IGSCC was an increasing function of strain hardening whileTGSCC was first favoured by strain hardening, then decreased when thestrain hardening became too important.

Therefore, some mechanical considerations about the strain path changescould reveal some SCC aspects. During plastic strain, the most highly stressedslip systems were activated, leading to the dislocation motion in these planes.After a sufficient amount of monotonic deformation (b = +1) the dislocation

g = 0/b = (+1)g = 0.2/b = –1g = 0.2/b = 0g = 0.4/b = –1g = 0.4/b = 0g = 0.4/b = +1

400350300250200Vickers microhardness in initiation areas

1200

1000

800

600

400

200

0

Max

. SC

C d

epth

(mm

)

7.10 Vickers micro-hardness threshold for SCC initiation andpropagation for several strain paths. SSRTs on notched type Bspecimens. (360 ∞C, e = 5 ¥ 10–8 s–1).


structures evolved toward steady-state configurations as cell block boundaries(CBBs), where dislocations were stored. Generally CBBs were formed, infcc structure, along the most active {111}-cristallographic slip planes. In areverse test (b = –1), most of the slip systems that were active during the pre-strain were also active during the second loading, but were operating in theopposite sense. According to Hu [17], the beginning of the reverse loadinglead to the rapid disappearance of unstable dislocation pile-ups, giving riseto an asymmetry of slip resistance. In a cross test (b = 0), the active slipsystems from the first deformation path remained latent while new slip systemswere activated. A high resistance to dislocation motion was obtained, becausethe CBBs formed during the first stage operated as obstacles for the new slipsystems. Consequently, changes of strain paths could have two major effectson SCC: first, CBBs formed during the pre-deformation could lead to strongobstacles to dislocation motion and increase SCC in agreement with corrosionenhanced plasticity models [1]. Second, short transient behaviour such asBaushinger effect (decreasing of yield strength on reloading in a reversesense) or cross effect, resulting from micro-plasticity, could have majorimplications on the enhancement of SCC mechanisms in 304L. One of thenoticeable features was that the effects resulting from reverse and cross testsappeared to vanish after an equivalent tensile strain of 0.15–0.20. Afterwards,

7.11 Transgranular crack propagation. SSRTs on notched type Bspecimens. (360 ∞C, e = 5 ¥ 10–8 s–1).


the initial plastic anisotropy was totally replaced by the anisotropy inducedby the new deformation mode. In accordance to SCC observations, it couldbe assumed that TGSCC was dramatically reduced in the cross SCC testbecause the motion of dislocations into the grains, during the second strain,was slowed down by the dislocation forest induced by the pre-strain hardening.Furthermore, TGSCC could be favoured by planar glide at the crack-tip,while IGSCC would rather be enhanced by the strain incompatibilities.Additional tests and observations should be carried out to strengthen thishypothesis.

Comments could finally be made about the transition from initiation topropagation of SCC, observed during SSRTs. SCC of alloy 600 in primaryenvironment had shown that transition from initiation to propagation stagedepended on a critical default, related to a critical stress intensity factorKISCC. This notion originally defined in the LMF was adapted to the case ofenvironmentally assisted cracking. KISCC has been interpreted as the KI thresholdabove which SCC mechanism produced a local plasticity at the crack-tip, toauto-supply the SCC mechanism with necessary dislocations for depassivationof the material [19]. KISCC value currently admitted for alloy 600 in PWRenvironment was close to 9 MPa m [19]. According to the LMF theory, KI

values were valid only under very strict conditions (plane deformations,

7.12 Intergranular crack propagation. SSRTs on notched type Bspecimens. (360 ∞C, e = 5 ¥ 10–8 s–1).


plane default with a critical size, localised strain). Finally, KI could not becalculated from short cracks propagating in SSRT specimens where the plasticstrain was not localised and where the stress increased during the test. However,this transition could have another origin, as a morphological barrier (grainsize). Indeed, the depth of transition corresponded to the grain size of thematerial. Therefore it could be assumed imagine that cracks initiate andslowly propagate until a strong obstacle (grain boundary) was reached, thatcould be over passed with a collective effect only (coalescence of shortcracks to cross the barrier). Further tests and observations should be conductedto understand the significance of this transition (different loadings, grainsizes, etc.).

7.5 Conclusions

A series of tests was conducted with two types of specimens to make clearthe initiation and propagation stages in a non-pre-strain hardened and pre-strain hardened 304L SS. Pre-shearing tests were used to clarify the effect ofthe strain path on SCC and more precisely on CGP as suggested by previousstudies conducted with humped specimens. The main conclusions were asfollows:

∑ pre-straining was necessary for SCC∑ there was a strain path effect on the crack morphology∑ a monotonic strain path led to TGSCC∑ a complex strain path led to IGSCC∑ TGSCC depth was not a monotonic function of the pre-straining∑ IGSCC depth was an increasing function of the pre-straining∑ no stress effect was observed on crack morphology.

The benefit of sequential testing was to reduce incubation time to theonset of cracking in susceptible materials through the application of thedynamic plastic straining during the first sequence. Additional sequentialtests could be carried out with different constant stresses, pre-straining orstrain paths. Besides, efforts could be made to clarify the effect of the strainpath and the cold plastic deformation on the environment-sensitive crackingof austenitic stainless steels. In particular, the role played by austenitetransformation, the internal stresses and the substructure of the dislocationsshould be investigated.

7.6 References

1. T. Magnin, Chierragatti, Oltra, Acta Metallurgica et Materialia, 1990, vol. 38 no. 7,1313.

2. P. Ford, ‘Slip dissolution model’, Corrosion sous contrainte phénoménologie etmécanismes, Ed. de Physique 1990.


3. M. Puiggali, ‘Etude du mécanisme de fissuration par CSC des aciers inoxydablesausténitiques: influence de l’écrouissage, de la température et du potentielélectrochimique’, Thesis Bordeaux I University, 1981.

4. Y. De Curière, ‘Recherche d’une amélioration du comportement en CSC d’alliagesde structure CFC: influence d’une pré-déformation en fatigue oligocyclique sur lecomportement en CSC de l’acier inoxydable austénitique 316L dans une solutionbouillante de MgCl2 à 117 ∞C’, Thesis of ENSMSE and INPG, 2000.

5. T. Magnin, A. Chambreuil, Bayle, Acta Materialia, 1996, vol. 44 no. 4, 1457.6. R. Pathania, ‘Quantification of yield strength effects on IGSCC in austenitic stainless

steels and its implication to IASCC’, EPRI report 1007380, 2002.7. T. Couvant, ‘Corrosion sous contrainte en milieu primaire REP de l’acier inoxydable

austénitique écroui 304L’, Thesis of Ecole des Mines de Saint-Etienne, 2003.8. F. Vaillant, T. Couvant, J.M. Boursier, ‘Stress corrosion cracking of cold worked

austenitic stainless steels in laboratory primary water environment’, Pressure Vesselsand Piping Conference, San Diego, 2004.

9. S.W. Sharkawy, Z. Xia, Z. Szklarska-Smialowska, ‘Stress corrosion cracking ofAISI 304 and 316 stainless steels in lithiated water at 350 ∞C’, Journal of NuclearMaterials, 1992, 195, 184–190.

10. N. Totsuka, Z. Szklarska-Smialowska, ‘Hydrogen induced IGSCC of Ni-containingfcc alloys in high temperature water’, 3rd Environmental Degradation of Materialsin Nuclear Power Systems – Water Reactor, 1988, 691–696.

11. Z. Szklarska-Smialowska, Z. Xia, S.W. Sharkawy, ‘Comparative studies of SCC intwo austenitic stainless steels and alloy 600 on exposure to lithiated water at350 ∞C’, Corrosion, 1992, vol. 48, no. 6, 455–462.

12. K. Arioka, ‘Effect of temperature, hydrogen and boric acid concentration on IGSCCsusceptibility of annealed 316 stainless steel’, Fontevraud V, France, 2002,149–158.

13. T. Couvant, J.M. Boursier, F. Vaillant, D. Delafosse, O. Raquet, C. Amzallag, ‘Effectof prestraining on SCC resistance of austenitic alloys in PWR primary water’,Environmental Degradation of Engineering Materials, Bordeaux, 2003.

14. Annual Book of ASTM Standards, Part 11, ASTM, Philadelphia 1978, 205.15. Angel, ‘Formation of martensite in austenitic stainless steels’, Journal of the Iron

and Steel Institute, 1954, vol. 177, no. 1, 165–174.16. Peeters, Kalidindi, Van Houtte, Aernoudt, ‘A cristal plasticity based worked-hardening/

softening model for B.C.C. metals under changing strain paths’, Acta Materialia,2000, vol. 48, 2123–2133.

17. Hu, Rauch, Teodosiu, International Journal of Plasticity, 1992, vol. 8, 839.18. Schmitt, Aernoudt, Baudelet, ‘Yield loci for polycristalline metals without texture’,

Material Science and Engineering, 1985, vol. 75, 13–20.19. F. Foct, ‘Mécanismes de corrosion sous contrainte de l’alliage 600 polycristallin et

monocristallin en milieu primaire: rôle de l’hydrogène’, Thesis of Ecole des Minesde Saint-Etienne, 1999.

103

8.1 Introduction

In the early 1970s, numerous cases of intergranular stress corrosion crackingoccurred in boiling water reactors (BWR) in AISI 304 type austenitic stainlesssteels. The root cause for this cracking is a combination of tensile stresses, anoxidising environment and a sensitised material. The remedial actions takenhave involved all three major parameters, e.g. application of narrow-gapwelding technique to reduce residual stresses, increase of the overall purityof the primary water, application of hydrogen or noble metal water chemistry,as well as reducing the amount of carbon in the stainless steels to avoidsensitisation. Nitrogen is added to maintain the strength level of austeniticstainless steels with reduced carbon levels. In the early 1990s cases ofintergranular cracking in non-sensitised, low carbon stainless steel materialsof types AISI 316NG and AISI 304L were observed. Several cases have so farbeen reported and cracking has been observed both in the HAZ of the weldsas well as in the base metals far away from any weld. Although all affectingparameters are so far not known, deformation seems to be a common parameter.Several open questions are still connected to this type of cracking, such as apossible difference in the behaviour between different types of austeniticstainless steels, the effect of chemical composition, the effect of cold work(amount and temperature), the influence of constraint during welding, etc.

The affecting mechanisms may include dynamic strain ageing (DSA) andenvironmentally enhanced creep. Dynamic strain ageing occurs in alloyscontaining solute atoms, which can rapidly and strongly segregate todislocations and lock them during straining. The maximum effect of DSAcorresponds to such conditions, where the solute atoms can follow by diffusionthe changes of the dislocation structure. DSA phenomenon leads to aninhomogeneous plastic flow and serrated yielding during straining at elevatedtemperatures and often results in a remarkable degradation of mechanicalproperties for a number of engineering alloys.

Austenitic stainless steels show DSA behaviour in a wide range of

8Dynamic strain ageing of deformed

nitrogen-alloyed AISI 316 stainless steels

U. E H R N S T É N and A. T O I V O N E N, VTT TechnicalResearch Centre of Finland, Finland and M. I VA N C H E N K O,

V. N E V D A C H A, Y. Y A G O Z I N S K Y Y andH. H Ä N N I N E N, Helsinki University of Technology, Finland


temperatures (~200–800 ∞C), which depends on the actual strain rate. Interstitialcarbon and nitrogen atoms dissolved in the crystal lattice play a determiningrole in DSA of austenitic stainless steels in the temperature range between200 ∞C and about 600 ∞C [1–3]. Literature results have, however, also shownthat nitrogen alloying shifts the onset temperature of DSA to higher values[4].

The aim of the present investigation is to study the effects of nitrogenalloying and deformation on DSA phenomenon in austenitic AISI 316L stainlesssteel at ~ 300 ∞C. The effect of deformation of AISI 316NG steel on thecracking behaviour in BWR water was additionally investigated using risingand constant displacement loading.


Three model austenitic AISI 316L type stainless steel materials with differentnitrogen contents and a commercial nuclear grade AISI 316NG stainlesssteel were used in the study. A sensitised AISI 304 steel was additionallyused in the crack growth rate tests. The chemical compositions of the materialsare shown in Table 8.1. Details concerning the manufacturing of the modelmaterials can be found in [5]. The effect of deformation was investigated byprestraining the materials at room temperature in tension before preparationof test specimens.

All blanks for the tensile test specimens were cut from the plates, transverseto their rolling direction, and in the longitudinal direction from the AISI316NG stainless steel pipe. The microstructure and hardness (HV 10) weredetermined. Tensile tests for observing DSA were carried out using a 25 kNMTS 858 test machine equipped with a MTS High-Temperature Furnace653.02, at strain rates of 10–4, 10–5, 5 ¥ 10–6 and 10–6 s–1, and temperaturesof 200, 288 and 400 ∞C. All tensile test specimens were prepared accordingto ASTM standard E 8M (sheet-type sub-size specimens). Tensile tests wereperformed according to the standards SFS-EN 1002-1 and ASTM E21 (StandardTest Method for Elevated Temperature Tension Tests of Metallic Materials).Internal friction method was used in the study for evaluation of the freenitrogen content and its diffusion redistribution in the crystalline lattice ofthe studied stainless steels. Details of the test parameters are given in [5].

Crack growth rate tests in simulated BWR NWC environment (DOout 500ppb, kin < 0.1 mS/cm, T 290 ∞C, p 92 bar) were performed using rising andconstant displacement tests and 10 ¥ 10 ¥ 55 mm3 SEN(B) specimens. Sixspecimens, five made of AISI 316NG and one of sensitised AISI 304 (1050∞C/20 min + 680 ∞C/1 h + 500 ∞C/24 h) material were tested in the sameautoclave equipped with bellow loading devices. The AISI 316NG steel wastested in non-deformed (one specimen), and deformed conditions (twospecimens with 5% and two with 20% deformation). The sensitised AISI 304

Dynam

ic strain ageing of deformed nitrogen-alloyed A

ISI 316

105

Table 8.1 Chemical compositions of the studied stainless steels in weight %

Type Code C Si Mn P S Cr Ni Mo Cu Al O2 N2

AISI 316L 1042 0.022 0.51 1.47 0.026 0.002 16.8 11.0 2.1 0.20 0.02 0.004 0.028

AISI 316L 1043 0.022 0.52 1.50 0.027 0.002 16.8 11.1 2.0 0.19 0.02 0.004 0.085

AISI 316L 1045 0.022 0.53 1.53 0.027 0.002 17.0 11.2 2.1 0.18 0.02 0.005 0.176

AISI 316NG BB44 0.022 0.38 1.66 0.027 0.002 17.0 12.5 2.28 0.11 0.01 0.007 0.093

AISI 304 165 0.042 0.47 0.88 0.026 0.018 18.2 10.2


steel, used as a reference specimen to enable comparison of the test resultswith literature data, was tested in non-deformed condition. The tests werestarted with a displacement rate of 5.5 ¥ 10–8 mm/s, which was reduced to5.5 ¥ 10–9 mm/s when stable crack growth was detected and further withconstant displacement tests after about 600 h testing time. The total testingtime was 1198 h. The crack growth was continuously monitored using theDC-PD technique. After the tests, the cracks were opened by fatigue, thefinal crack lengths were measured and the cracking morphology was determinedusing SEM.

8.3 Results

The microstructure of all materials was austenitic. The grain size of themodel alloys was smaller than that of the commercial AISI 316NG steel,Table 8.2. The slope of the increase in hardness was similar for all alloys,Fig. 8.1.

Nitrogen alloying increases the strength properties of AISI 316L stainlesssteels in the testing temperature range and the elongation to fracture decreaseswith increasing nitrogen content except in the case of the commercial AISI316NG stainless steel, which demonstrates highest elongation to fracture in

Table 8.2 Grain sizes and hardness of the investigated materials

Material and code Grain size Hardness (HV 10)ASTM No/mm ——————————————————

0% def. 5% def. 20% def.

AISI 316L, 1042 7/36 136 172 229AISI 316L, 1043 6.5/43 159 191 255AISI 316L, 1045 8/25 179 215 284AISI 316NG, BB44 5/71.8 147 174 227AISI 304, 165 4/101 nd nd nd

1042, 0.028% N1043, 0.085% N1045, 0.176% NBB 44, 0.093% N

Degree of deformation (%)2520151050

300

250

200

150

100

50

0

Har

dn

ess

(HV

10)

8.1 Hardness versus degree of deformation for AISI 316L materials.

Dynamic strain ageing of deformed nitrogen-alloyed AISI 316 107

the whole range of testing temperatures, Fig. 8.2. Yield stress decreases withtesting temperature, while ultimate tensile stress is almost constant in thestudied temperature range. As a function of temperature, the elongation tofracture varies with the nitrogen content only slightly. The strain hardeningcoefficient increases with increasing testing temperature and decreasingnitrogen content, Fig. 8.2d.

Serrated yielding was observed in all AISI 316L stainless steels at testingtemperatures above 200 ∞C and strain rates slower than 10–4 s–1, Fig. 8.3.DSA serrations on the stress-strain curves are well-defined at testingtemperatures of 288 and 400 ∞C, while at 200 ∞C they appear only for thematerial with the lowest nitrogen content of 0.028 wt.%. The obtained stress-strain curves indicate that nitrogen alloying suppresses the DSA developmentin AISI 316L type stainless steels. Further, the amplitude of the stress pulsesdecreases markedly with the increase of nitrogen content and only a few

AISI 316 LN with 0.176 wt.% nitrogenAISI 316 LN with 0.085 wt.% nitrogenAISI 316 LN with 0.028 wt.% nitrogenAISI 316 NG with 0.093 wt.% nitrogen

Temperature (∞C)(b)

200 250 300 350 400

540

520

500

480

460Ult

imat

e te

nsi

le s

tres

s (M

Pa)

Temperature (∞C)(a)

200 250 300 350 400

200

225

175

150

125

Yie

ld s

tres

s (M

Pa)

0.48

Temperature (∞C)(d)

200 250 300 350 400

0.50

0.52

0.54

0.56

0.58

0.60

0.62

0.64

0.66

Str

ain

har

den

ing

co

effi

cien

t

Temperature (∞C)(c)

200350300250200

52

48

44

40

Elo

ng

atio

n (

%)

8.2 Temperature dependencies of yield stress (a), ultimate tensilestress (b), elongation to fracture (c) and strain hardening coefficientobtained for 13–28% strain (d) for the studied stainless steels.


0.1 mm/mm

Engineering strain(b)

600

500

400

300

200

100

0

En

gin

eeri

ng

str

ess

(MP

a)

200 ∞C 288 ∞C 400 ∞C

0.1 mm/mm

Engineering strain(c)

600

500

400

300

200

100

0

En

gin

eeri

ng

str

ess

(MP

a) 200 ∞C 288 ∞C 400 ∞C

0.1 mm/mm

Engineering strain(a)

600

500

400

300

200

100

0

En

gin

eeri

ng

str

ess

(MP

a) 200 ∞C 288 ∞C 400 ∞C

8.3 Engineering stress-strain curves obtained at the strain rate of10–5 s–1 for AISI 316L stainless steels with 0.028 (a) and 0.176 (b)wt.% of nitrogen and for AISI 316NG (c) stainless steel with 0.093wt.% of nitrogen.


pulses are present on the stress-strain curves of the stainless steels with0.093 and 0.176 wt.% of nitrogen at testing temperature of 288 ∞C. A similareffect of nitrogen on DSA in AISI 316LN stainless steels was obtained in [4]for higher strain rates of testing. Prestraining at room temperature leads notonly to an increase of yield and ultimate tensile stresses, but it reduces alsothe onset deformation of DSA, Fig. 8.4a. It seems that cold working facilitatesthe DSA development in nitrogen-alloyed stainless steels. DSA serrationsbecome visible on the stress-strain curve for AISI 316NG steel obtained attesting temperature of 200 ∞C after 5% prestraining, Fig. 8.4b, while muchless serrations appear in the as-supplied material, Fig. 8.3c.

The A-type serrations [1] observed correspond to quasi-regular separatepulses of flow stress. For evaluation of the average time between the pulses

0.1 mm/mm

Engineering strain(a)

0.1 mm/mm

Engineering strain(b)

500

600

400

300

200

100

0

En

gin

eeri

ng

str

ess

(MP

a)E

ng

inee

rin

g s

tres

s (M

Pa)

0

100

200

300

400

500

600200 ∞C 288 ∞C 400 ∞C

20%5%

0%

8.4 Engineering stress-strain curves obtained at 288 ∞C and strainrate of 10–5 s–1 for prestrained AISI 316NG stainless steel (a) and forAISI 316NG stainless steel with 5% prestraining at differenttemperatures (b).


the obtained stress-strain curves were transformed to the frequency dependencyusing Fourier analysis. Fourier spectra are shown in Fig. 8.5. In the presenceof quasi-regular serrations on the stress-strain curve, as it can be seen fortesting temperatures of 288 and 400 ∞C in Fig. 8.5a, some maxima arise inthe Fourier spectra, while no distinct maxima are present in the spectrum at200 ∞C, when serrations are missing. The maxima shown by arrows in Fig.8.5, correspond to the flow stress pulses, which reflect the repeated advancementof the Lüders band throughout the specimen. The average time betweenpulses, which is reciprocal to the frequency of maximum in the Fourierspectrum, can be estimated to be 2.7 ks for the testing temperature of 400 ∞C.

Nitrogen alloying suppresses the amplitude of DSA serrations as it is seenin Fig. 8.5b. High amplitude quasi-regular pulses of the flow stress observedin the stainless steel with 0.028 wt.% of nitrogen become smaller whennitrogen content increases to 0.176 wt.%, and the average time betweenpulses is then about 2.3 ks.

Internal friction (IF) in the studied stainless steels was mainly measuredto check the presence of interstitial nitrogen atoms in the crystalline latticeof the studied austenitic stainless steels. Two IF peaks were observed, Fig.8.6, situated at about –50 ∞C and 100 ∞C, which increase with the amount ofcold deformation. They presumably represent an anelastic response ofdislocations interacting with point defects produced in the austenite crystallinelattice by cold deformation [6]. It is well established [7] that IF peak in thevicinity of 350 ∞C is caused by a Snoek-like relaxation process due to elementaldiffusion jumps of interstitial nitrogen atoms in FCC crystalline lattice ofaustenite.

The amplitude of the Snoek peak is proportional to the free nitrogenconcentration. Thus, Fig. 8.6 reveals that free nitrogen atoms are present inthe AISI 316NG steel at 288 ∞C. The concentration of the free nitrogenatoms in the lattice increases with the amount of prestraining, in line with thetensile test results showing an earlier onset of DSA in deformed materials.The observed increase of the nitrogen Snoek-like peak amplitude in theprestrained stainless steel is reduced with ageing time at elevated temperatures,Fig. 8.7, due to escape of free nitrogen from the solid solution. The peakreduction process can be described as a sum of three exponential decayfunctions (shown by dotted lines in Fig. 8.7) with characteristic decay timesof 0.6, 2.6 and 14.2 ks. The origin of the fastest component of the process isstill unclear, while the second and third ones can be related to long-rangediffusion escape of nitrogen from solid solution to dislocations and, probably,to grain boundaries. Ageing also results in an increase of the normalisedshear modulus, indicating pinning of dislocations by nitrogen atoms due toageing [5, 7, 8].

The characteristic decay time of 2.6 ks, which represents the long-rangediffusion of nitrogen to dislocations, is close to the value of the average time


8.5 Fourier transformation spectra of the flow stress signal in thetensile tests of AISI 316NG stainless steel at a strain rate of 10–5 s–1

and different temperatures (a), and of AISI 316L alloys at a strain rateof 10–5 s–1 and 288 ∞C (b). Arrows in (a) correspond to quasi-regularseparate pulses.

2520151050Frequency ¥ 104 (s–1)

(b)

1.0

0.5

0.0

1.0

0.5

0.0

1.0

0.5

0.0

Am

plit

ud

e (a

rb. u

nit

s)

0.028 wt.% N

0.085 wt.% N

0.174 wt.% N

Am

plit

ud

e (a

rb. u

nit

s)

Frequency ¥ 104 (s–1)(a)

2520151050

400 ∞C

288 ∞C

200 ∞C

0.0

0.4

0.8

0.0

0.4

0.8

0.0

0.4

0.8


between serration pulses obtained above from stress-strain curve by Fourieranalysis (2.7 ks). It seems that the repeated pinning of dislocations by diffusionof mobile nitrogen atoms, which is related to the advancement of Lüdersbands, is a key element of DSA in AISI 316L steels at testing temperaturesused in this study.

0Temperature (∞C)100–200 –100 200 300 400 500

9

8

7

6

5

4

3

2

Inte

rnal

fri

ctio

n (

Q–1

¥ 1

04 )

as-supplied

2% prestraining

5%

8.6 Temperature dependencies of internal friction for AISI 316NGstainless steel in as-supplied state and after 5% and 20%prestraining.

1.004

1.002

1.000

No

rmal

ised

mo

du

lus

(G(T

)/G

(370

∞C)

100001000Time (s)

7

6

5

4

Inte

rnal

fri

ctio

n (

Q–1

¥ 1

04 )

8.7 Amplitude of the Snoek-like peak of nitrogen and normalisedshear modulus of AISI 316NG stainless steel as a function of ageingtime at 340 ∞C. Dotted lines represent the three components of thepeak amplitude decay.


In the crack growth rate tests fully intergranular cracking was obtained inthe sensitised AISI 304 stainless steel specimen and in one of the two 20%prestrained AISI 316NG steel specimens. Some IG fracture was also observedin one of the two 5% prestrained AISI 316NG steel specimens, Fig. 8.8. All

(a)

(b)

8.8 Fractographs showing fully intergranular cracking in the 20%prestrained AISI 316NG specimen (a) and mixed trans- andintergranular cracking in the 5% prestrained AISI 316NG specimen(b) after constant displacement testing in BWR NWC environment.


other specimens revealed transgranular cracking. The crack growth ratesobtained for the sensitised AISI 304 material are in the order of 10–7 mm/sand are similar to those obtained using 10 ¥ 10 ¥ 55 mm3 SEN(B) and25 mm C(T) specimens in reference [9]. The crack growth rates of sensitisedAISI 304 steel specimens depend on the loading mode: the crack growthrates are lower by a factor of 2 to 10 under constant displacement than underrising displacement conditions. All of the crack growth rates are also plottedas a function of loading rate, in terms of J-integral increase rate dJ/dt, in Fig.8.9b. dJ/dt is a measure of loading rate independent of the specimen size andloading geometry. The interconnections between crack growth rate, fracturemorphology and loading rate/type are discussed in more detail in reference[9].

The results revealed a higher tendency for the 20% deformed AISI 316NGsteel to intergranular environmentally assisted cracking (EAC) in BWR NWCenvironment compared to non-deformed, non-sensitised material. However,the susceptibility to EAC is much lower than that in sensitised stainlesssteels, in accordance with expectations. The crack growth rate at a similarloading rate (i.e., dJ/dt) is one order of magnitude higher in the sensitisedAISI 304 steel compared to that in 20% deformed AISI 316NG. The crackgrowth rate in 5% deformed AISI 316NG steel showing mixed transgranularand intergranular cracking was in the same order as in the sensitised AISI304 steel. However, there was a ripple loading fatigue component of R ~0.9and f ~1 Hz present during that test, which can be expected to result inpartially transgranular fracture morphology and also in enhanced crack growthrate. More tests are, however, needed in order to determine the EAC crackgrowth rates of non-sensitised stainless steels as a function of degree ofdeformation and chemical composition.

8.4 Discussion of results

The results obtained in the present investigation are in good accordance withliterature data on nitrogen effects on DSA in AISI 316L stainless steels. Amap of DSA, shown in Fig. 8.10, summarises the serrated flow appearancein AISI 316NG stainless steel at different strain rates and testing temperatures.The dashed line in Fig. 8.10 forming a boundary for testing parameters,where DSA occurs, extends to lower strain rates applied in the presentinvestigation as compared to those in [4].

The enthalpy calculated using the dashed line in Fig. 8.10 is about 1.24eV and its value approaches the enthalpy of nitrogen diffusion in the austenitelattice. This value is very close to the enthalpy of nitrogen diffusion calculatedfrom the Snoek-like IF peak, 1.45 eV at 350 ∞C [5].

The DSA-results showing that nitrogen alloying suppresses the onset strainand temperature range of DSA indicate that nitrogen alloying may also lower


This study, sensitised AISI 304This study, 5% prestrained AISI 316NGThis study, 20% prestrained AISI 316NGToivonen (2004), sensitised AISI 304

This study, sensitised AISI 304This study, 5% prestrained AISI 316NGThis study, 20% prestrained AISI 316NGToivonen (2004), sensitised AISI 304

Circled data points: constant displacementAll others: rising displacement

100806040200KJ (MPam1/2)

(a)

1e-5

1e-6

1e-7

1e-8

da/d

t (m

m/s

)da

/dt (

mm

/s)

1e-5

1e-6

1e-7

1e-8

dJ/dt (kJ/s2m)(b)

1e-58e-66e-64e-62e-60–2e-6

8.9 Crack growth rate as a function of KJ for the crack growth ratetests with observed intergranular cracking (a) and as a function ofthe loading rate in terms of dJ/dt (b). A fatigue component of R ~0.9and f ~1 Hz was present during the constant displacement phase ofthe 5% prestrained AISI 316NG specimen.


8.10 DSA-map of AISI 316NG stainless steel. Filled symbolscorrespond to strain rate and temperature values at which DSA(serrated yielding) was observed on stress-strain curves. Data pointsshown by triangles above the dotted line were obtained in [4]. Thedashed lines are the boundaries for the DSA appearance in thisstudy and in [4].

Temperature (∞C)200300400500600

1/T (K–1)0.0012 0.0014 0.0016 0.0018 0.0020 0.0022 0.0024

10–2

10–3

10–4

10–5

10–6

d(e

)/d

t (s

–1)

log

(d

(e)/

dt)

–2.0

–2.5

–3.0

–3.5

–4.0

–4.5

–5.0

–5.5

–6.0

the EAC susceptibility, if DSA is considered to be a part of the decisivemechanism. The crack growth rate test results obtained in this study are inline with literature and field experience showing increased susceptibility andincreased crack growth rates in non-sensitised stainless steels due to deformation[10]. However, the crack growth rates in deformed, non-sensitised stainlesssteel are lower than those in sensitised stainless steel in BWR NWCenvironment. Crack growth rate tests on stainless steels have revealed acorrelation between susceptibility to intergranular cracking, CGR and yieldstrength [10]. The yield strength increases as a function of deformation, butalso as a function of nitrogen content. As these materials are non-sensitised,corrosion must be less decisive and localisation of deformation to the grainboundaries more important than in the case of sensitised stainless steels.

A suppression of the DSA development in the studied stainless steelscaused by nitrogen alloying looks contradictory as DSA, e.g. in low alloysteels, is enhanced by free interstitials. The suppressive effect of nitrogen onDSA may be caused by the increase of the flow stress with nitrogen alloyingof the steel causing an increase of the actual stress and consequent possiblechanges in the deformation response. DSA is expected to result in localisationof plastic deformation to grain boundary regions. This is also the case indeformed materials, where DSA was observed at all studied nitrogen levels.A detailed mechanism of the role of DSA in EAC and the role of deformationas well as stainless steel composition needs further investigations to reveal


the main parameters affecting EAC in deformed, non-sensitised stainlesssteels in high temperature water such as BWR NWC.

8.5 Conclusions

∑ DSA in nitrogen-alloyed AISI 316L type stainless steels can occur in theinvestigated temperatures range of 200–400 ∞C, when strain rates areslower than 10–4 s–1.

∑ Nitrogen suppresses the DSA development in AISI 316L type stainlesssteels. The onset deformation of DSA serrations shifts to higher valuesof strain and the amplitude of the flow stress pulses decreases withincrease of nitrogen content.

∑ Prestraining at room temperature reduces the onset deformation of DSAin AISI 316NG stainless steel.

∑ An apparent activation enthalpy of DSA in AISI 316NG stainless steelis about 1.24 eV at temperatures around 300 ∞C. The value of enthalpyof DSA corresponds well to the enthalpy of nitrogen diffusion in AISI316NG steel obtained by the internal friction method being about1.45 eV.

∑ Prestraining increases the susceptibility of non-sensitised AISI 316NGstainless steel to intergranular stress corrosion cracking in BWR NWCenvironment. The crack growth rate is, however, lower than that forsensitised stainless steel.


This presentation is prepared within the project Structural operability andplant life management (RKK and XVO), which is coordinated by TeollisuudenVoima Oy. The work has been funded by the National Technology Agency(Tekes), Teollisuuden Voima Oy (TVO), Fortum Power and Heat Oy, FortumNuclear Services Ltd, FEMdata Oy, Neste Engineering Oy, Fortum Oil andGas Ltd and VTT. The Swedish Nuclear Power Inspectorate, SKI, alsoparticipated in this work. Their funding is gratefully acknowledged.

8.7 References

1. L. H. de Almeida, I. LeMay, P. R. O. Emygdio: Mater. Characterization, 40 (1998),pp. 137–150.

2. L.H. de Almeida, P.R.O. Emygdio: Scr. Met. et Mater., 31 (1994), pp. 505–510.3. R. Ilola, M. Kemppainen, H. Hänninen: Dynamic Strain Ageing of Austenitic High

Nitrogen Cr-Ni and Cr-Mn Steels, Proc. of 5th Int. Conf. ‘High Nitrogen Steels’98’,Mat. Sci. Forum, 318–320 (1999), pp. 407–412.

4. D. W. Kim, W. Ryu, J. Hwa Hong, S. Choi: Journal of Nuclear Materials, 254(1998), pp. 226–233.


5. M. Ivanchenko, U. Ehrnstén, V. Nevadacha, Y. Yagodzinskyy, H. Hänninen. ‘DynamicStrain Ageing of Nitrogen-alloyed AISI 316L Stainless Steel. Proceedings of the 7thInternational Conference on High Nitrogen Steels 2004, Ostend, Belgium, September19–22, 2004. GRIPS media GmbH. p. 641–649.

6. C.F. Burdett, I.J. Queen: Met. Rev., 43 (1970), pp. 47–65.7. Yu. Jagodzinski, S. Smouk, A. Tarasenko, H. Hänninen: Distribution of Interstitial

Impurities and their Diffusion Parameters in High-Nitrogen Steels Studied by Meansof Internal Friction, Proc. of 5th Int. Conf. ‘High Nitrogen Steels ’98’, Mat. Sci.Forum, 318–320 (1999), pp. 47–52.

8. A.S. Nowick, B.S. Berry: Anelastic Relaxation in Crystalline Solids, AcademicPress, N.Y., London, 1972, 677 p.

9. A. Toivonen: Stress Corrosion Crack Growth Rate Measurement in High TemperatureWater using Small Precracked Bend Specimens. VTT Publications 531, 2004, 206 p.+ App.

10. P. Andresen, P. Emigh, M. Morra, R. Horn: Effects of Yield Strength, CorrosionPotential, Stress Intensity Factor, Silicon and Grain Boundary Character on the SCCof Stainless Steels. 11th Symposium on Environmental Degradation of Materials inNuclear Power Systems – Water Reactors, Stenvenson, WA, Aug. 10–14, 2003, pp.816–833.

119

9.1 Introduction

Steam generator tubes of Pressured Water Reactors (PWR) suffer fromIntergranular Generalized Attack (IGA) and Intergranular Stress CorrosionCracking (IGSCC) in flow-restricted areas at the top of tubesheet or betweentubes and support plates, where water pollutants are likely to concentrateunder heat flux. During the first decade of operation, the resulting creviceenvironments were supposed to be alkaline. However, chemical specificationsand operating conditions used in PWRs have been improved (i.e., low sodiumcontent, introduction of polishing mix-beds for make-up water and boricacid injection) during the last decade, which has led to a less causticenvironment, according to hide-out return analyses of the secondary waterafter shutdown [1, 2] and from examinations of pulled tube [3].

Many investigations have been performed on pulled tubes from severalFrench plant units selected for different kinds of cooling water and chemicalconditioning of the secondary side in PWR. These investigations revealedIGA and IGSCC to occur under alumino-silicate deposits, with an underlyinghydroxide gel rich in chromium that is brittle and non-protective [4, 5].Laboratory experiments carried out independently by EDF or in collaborationwith CEA and Framatome ANP have succeeded in reproducing the maincharacteristics of such deposits and non-protective films on alloy 600 [6, 7].These experiments were carried out in an All Volatile Treatment (AVT)environment containing silica in combination with other pollutants like alumina,phosphate and acetic acid, called the ‘complex’ environment [6]. IGA andIGSCC of 600 mill annealed (MA) alloy also occur under alumino-silicatedeposits, which strongly suggests the detrimental effect of silica associatedwith the other pollutants. Nevertheless, doubt still remains on the role ofalumino-silicates in the corrosion process since cracking also occurs in AVTenvironment with phosphate and acetic acid. The relevance and relativeimportance of each pollutant has been recently discussed by EDF, especiallywith respect to plant experience [6].

9Laboratory results of stress corrosion

cracking of steam generator tubes in a‘complex’ environment – an update

O. H O R N E R, E.-M. P AVA G E A U and F. V A I L L A N T,EDF R&D, France and O. D E B O U V I E R,

EDF Nuclear Engineering Division, France


A review of the laboratory results obtained by considering the ‘complex’environment is presented here, together with an update (concerning the recentdata obtained at EDF). Indeed, several tests have been carried out in order tostudy the initiation and the slow and fast propagation processes for 600 MAand thermally treated (TT) alloy. The influence of some related factors liketemperature and environmental factors is also discussed. Finally, two hypotheses– i.e. the effect of alumino-phosphate and alumino-silicate compounds orcyano ligands – are proposed in order to explain the dissolution of nickel inalloy 600 during the corrosion process. Each hypothesis is discussed in thelight of the previous results obtained in the laboratory by EDF.


9.2.1 Materials

Specimens (C-ring, Wedge Open Loaded (WOL) and tube specimens) wereprepared from MA or 600 TT alloy steam generator tubes. The chemicalcomposition and the mechanical properties of specimens were checked atEDF laboratories. The chemical composition and mechanical properties ofthe main materials used in this study are respectively detailed in Tables 9.1and 9.2:

9.2.2 Environment

The reference ‘complex’ environment was prepared with de-ionized waterand is described in Table 9.3:

9.2.3 Analysis

After each test, specimens were examined and IGSCC depths were measuredby optical microscopy on a section of the considered specimen.

9.3 Results

9.3.1 Introduction

The risk of IGSSC occurrence is larger than the risk of IGA occurrence.Therefore, this study is focused on IGSSC in order to model the corrosionrate. According to some previous results obtained in a sodium hydroxideenvironment, IGSCC of alloy 600 involves three successive steps, namelyincubation and, after initiation of cracks, slow propagation and rapidpropagation, as shown is Fig. 9.1:

A clear assessment of incubation would require statistical analysis onnumerous specimens, which could not be afforded here. The empirical model

Stress corrosion cracking of steam

generator tubes121

Table 9.1 Chemical composition of the tubes (weight %)

C Si Mn S P Cr Ni Fe Co Ti Cu Al N

Alloy RCC-M 0.010 < < < < 14.00 > 6.00 < < < < –600 4101 0.050 0.50 1.00 0.015 0.025 17.00 72.00 10.00 0.10 0.50 0.50 0.50

U581 0.021 0.18 0.21 0.003 0.007 15.20 bal. 8.95 0.020 0.30 0.015 0.26 0.0046U573 0.022 0.14 0.23 0.0002 0.005 15.20 bal. 6.90 0.03 0.32 0.03 0.27 0.0046

Table 9.2 Mechanical properties of the tube in MA and TT-conditions

20 ∞C 350 ∞C

YS (MPa) UTS (MPa) Elong (%) YS (MPa) UTS (MPa) Elong (%)

Alloy 600 RCC-M4101 275<< 450 > 550 >30 > 215 – –U581 MA 232–239 638–644 46 198–201 600–583 43–38U581 TT 227–231 653–657 39 178–178 588–579 –U573 MA 307 728–735 38–40 252 664 25.7U573 TT 274–287 > 716 38–40 239–240 648–652 –

Table 9.3 Description of the reference ‘complex’ environment defined by EDF

Species NH3 N2H4 Al2O3 SiO2 Ca3(PO4)2 CH3COOH pH320∞C

ppb mol/L ppm mol/L g/L mol/L g/L mol/L g/L mol/L ppm mol/L

Concen- 500 2.6 ¥ 10–5 2 6.3 ¥ 10–5 1.3 0.013 6.15 0.103 2.6 0.008 10 1.7 ¥ 10–4 5.2tration


established for IGSCC in the reference ‘complex’ environment was supposedto involve the incubation, slow propagation and rapid propagation stages.Moreover, the transition between slow and rapid propagation was supposedto occur when the stress intensity factor at the tip of the crack reached thelevel of the threshold stress intensity factor KISCC (the depth of the crackcorresponding to this transition is called the critical depth dc, as shown inFig. 9.1).

9.3.2 Incubation stage

In order to determine the incubation time of C-ring specimens in alloy 600in the reference ‘complex’ environment, some tests were interrupted at selectedtimes, as shown in Fig. 9.2:

Figure 9.2 suggests that, in the reference ‘complex’ environment (T =320 ∞C), the incubation time for C-ring specimens in the 600 MA alloy islocated in the 0–2000 hours range and in the 2000–3000 hours range for C-ring specimens in the 600 TT alloy. This strongly suggests that the initiationtime for 600 TT alloy is longer than the one for 600 MA alloy in thisenvironment.

The initiation time for alloy 600 in the ‘complex’ environment (of theorder of a few thousand hours) is very low compared to the running time ofPWR.

9.3.3 Initiation

IGSCC of alloy 600 in sodium hydroxide (MA and TT conditions) andsulfate (MA and TT conditions) environments appears above a stress threshold(sth) which was expressed as an increasing function of the yield stress of the

t : time

dc

d : IGSCC depth

Initiation

Rapidpropagation

Slow propagation

Incubation

9.1 Scheme of the different stages of IGSCC of alloy 600.

Stress corrosion cracking of steam generator tubes 123

material [8, 9]. Therefore, constant load tests were carried out on alloy 600tubes in the MA and TT conditions, in the ‘complex’ environment at 320 ∞C,to determine a possible relation between stress threshold and yield stress. Anapproximate value of sth has been determined for two heats of 600 MA and600 TT alloy respectively, as shown Fig. 9.3:

Firstly, these results show that the stress threshold values are smaller inthe reference ‘complex’ environment than in the sodium hydroxide or thesulfate environment [9]. In particular, a very low value of sth (160 MPa) hasbeen determined in the case of one tube made of 600 MA alloy. The IGSCCinitiation process is more likely to occur in the case of a ‘complex’ environmentthan in the two other environments (sodium hydroxide and sulfate). Secondly,

U581 MAU581 MA no interruptionU573 MAU573 MA no interruptionU573 TTU573 TT no interruption

150

100

50

0

Cra

ck d

epth

(mm

/h)

Time (h)500040003000200010000

9.2 Results of stress corrosion cracking on C-rings of alloy 600 in thereference ‘complex’ environment at 320 ∞C.

600 MA600 TT

260240220200180160Yield stress at 350 ∞C (MPa)

300

250

200

150

100Str

ess

thre

sho

ld a

t 32

0 ∞C

(M

Pa)

9.3 Stress threshold (sth) versus yield stress at 350 ∞C (YS350 ∞C) for600 alloy tubes in the reference ‘complex’ environment at 320 ∞C.


the thermal treatment of alloy 600 does not seem to have any effect on thestress threshold value (the values of sth in the case of 600 MA and 600 TTalloy are very similar).

9.3.4 Rapid propagation

CERT results were exploited according to the procedure developed by Santarini[10]. This procedure consists of representing the number N(1 > L) of cracktraces of depth greater than L versus the length L. In the case of 600 MAalloy, it was not possible to discriminate between a slow and a rapid propagationstep, as shown in Fig. 9.4:

In addition, the same result was obtained in the case of 600 TT alloy. Inconclusion, a rapid propagation step is not determined in the case of alloy600 in the reference ‘complex’ environment. This result is further confirmedin the reference ‘complex’ environment by an investigation of propagationon WOL specimens. Indeed, no crack deeper than 40 mm was locally obtained,leading to a local crack growth rate of ca 0.02 mm/h, fully consistent with apropagation in the slow regime.

9.3.5 Slow propagation

The slow propagation step was studied in a ‘complex’ environment by usingC-ring specimens in 600 MA and TT alloy at 320 ∞C. The effects of severalparameters on slow propagation, including chemical conditions, potentialand temperature have been studied in detail.

L (mm)200150100500

1000

100

10

1

N (

DL)

9.4 Number N (N > l) of cracks traces of depth greater than L versusthe length L on CERT specimen in 600 MA alloy in the reference‘complex’ environment at 320 ∞C (strain rate of 5 ¥ 10–8 s–1).


Organic compounds effect

Firstly, the results obtained in a ‘complex’ environment at 320 ∞C (with orwithout acetate) suggest that the choice of the amine (ammonia or morpholine)is not decisive on IGSCC. Moreover, cracking of alloy 600 is reduced if alarge concentration of ammonia is used; acetate alone does not play a significantrole in the cracking.

Secondly, the results obtained in an AVT environment with or withoutphosphate confirm their inhibitor effect on cracking at a high concentration(7.8 g/L), whereas they are harmful at an intermediate concentration (2.6 g/L) [7]. Moreover, phosphates are not necessary in the ‘complex’ environmentfor the cracking to occur.

Thirdly, the effect of carbonate has also been studied. Indeed, carbonateincreases IGSCC at high pH (9.2). However, no effect is observed at neutralpH. It has been shown that no cracking is observed when carbonate is replacedby sodium hydroxide (pH = 9.2). Therefore, this effect is likely to be due tocarbonate and not to the pH value. Finally, several tests have also beencarried in the ‘complex’ environment at 320 ∞C by varying the Al/Si ratio(see Section 9.4), as shown in Fig. 9.5:

These tests, conducted in EDF laboratories, show that the Al/Si ratio haslittle effect on cracking in a ‘complex’ environment. However, some significantdetrimental effect was observed in CEA laboratories [6]. This different behaviorcould be due to different hydrogen concentrations (and potential) used duringthese tests.

9.5 Al/Si ratio effect on the crack length on C-ring specimens(600 MA alloy) in a ‘complex’ environment (T = 320 ∞C).

U581 MAU573 MAU581 MA

0.9 10 0.80.70.60.50.40.30.20.1Al/Si ratio

150

100

50

0

Cra

ck le

ng

th (mm

/h)

3191 h test, no break


Redox potential effect

Four tests carried out for C-ring specimens (600 MA and TT alloy) at differentpotential values between 0 mV and 360 mV/Ecorr. These tests strongly suggestthat the crack velocity decreases rapidly as the potential value increasesfrom 0 to 400 mV/Ecorr, as shown in Fig. 9.6:

Temperature effect

Four tests carried out for C-ring specimens (600 MA and TT alloy) at differenttemperature values between 305 ∞C and 335 ∞C show that the crack velocityvaries with temperature, as shown in Fig. 9.7:

In particular, the maximum of IGSCC for alloy 600 in the ‘complex’environment occurs between 312.5 ∞C and 320 ∞C. This effect, which hasalready been observed in the sulfate environment [9], could be due to achange in the corrosion mechanism (i.e., from IGSCC to general corrosion).In most tests, 600 TT alloy is a bit less sensitive than 600 MA alloy towardsIGSCC, but in some conditions (e.g. at 312.5 ∞C, see Fig. 9.7), 600 TT alloycan be more sensitive that 600 MA alloy towards IGSCC.

Material effect

The IGSCC results obtained for 600 MA alloy do not depend on the heatconsidered during the tests. Moreover, 600 TT alloy was revealed to begenerally a bit more resistant towards IGSCC than 600 MA alloy. In particular,

600 MA600 TT

E (mV/Ecorr)

Vm

ax (mm

/h)

4003002001000

0.04

0.03

0.02

0.01

0

9.6 Potential effect on the crack velocity of 600 alloy in the reference‘complex’ environment at 320 ∞C.


the ratio in the same experimental conditions between the crack velocitiesfor 600 MA and TT C-ring specimens in a ‘complex’ environment is0.45 ± 0.35.

9.4 Discussion

IGA and IGSCC of mill annealed alloy 600 occur under alumino-silicatedeposits. Such alumino-silicate deposits are expected to have similar structuresto zeolites which are widely used as catalysts or ion exchange products sincethey correspond to potential Lewis acids (compounds with an electron doubletvacancy). On a structural point of view, zeolites correspond to crystallineinorganic polymers based on a repeating framework of AlO4

– and SiO4–

tetrahedra linked by some common oxygen atoms. These compounds can bechemically described by the formula M (AlO , SiO ), H O1/

+2 2 2n

n y z , where Mn+

corresponds to an alkali, an ammonium or a transition metal cation.As a consequence, these compounds may accept cations from the oxidizing

surface of alloy 600, according to the following mechanism [6]:

∑ electron transfer reaction:

M (AlO , SiO ), H O1/+

2 2 2nn y z Æ zeolite-Al[]Lewis acid center + H2O

zeolite-Al[]Lewis acid center + Ni Æ Ni2+ + zeolite-Al[:]

∑ ion exchange reaction:

Ni + M (AlO , SiO ), H O2+1/

+2 2 2n

n y z Æ

Ni (AlO , SiO ), H O + 1/ M1/22+

2 2 2+y z n n

9.7 Temperature effect on the crack velocity of 600 alloy in the‘complex’ environment at 320 ∞C.

600 MA600 TT

T (∞C)340330 335325320315310305300

v max

(mm

/h)

0.04

0.03

0.02

0.01

0


According to this mechanism, the Al/Si ratio should have a strong influenceon IGSCC. An increase in the Al/Si ratio in the zeolite framework wouldoccur with an increase in the Al/Si ratio in the alumina and silica sources.However, some of the previous tests (see Section 9.3) show that the Al/Siratio has little effect on cracking in a ‘complex’ environment. One has toconsider that the experimental conditions used here are not optimal for thezeolite compound synthesis.

IGA and IGSCC also involve an underlying hydroxide gel rich in chromium,which is brittle and non-protective. Another mechanistic hypothesis considersthat the dissolution of Ni could be increased by the cyano coordination ofNi2+, according to the following mechanism:

∑ amid formation:

CH3COOH + NH3 Æ CH3COO– NH 4+

∑ amid deshydratation (probably via the formation of P2O5 from phosphates):

CH3COO– NH 4+ Æ CH3CN + 2H2O

∑ Ni dissolution by formation of the complex: [NiII(CN)4]2–

Indeed, IGSCC was detected in an AVT, phosphate and acetic acidenvironment, i.e. without any alumino-silicate deposits. However, some ofthe previous tests (see Section 9.3) show that phosphate is not necessary toobtain cracks.

9.5 Conclusions

The effects of several parameters such as chemical conditions, potentialvalue and temperature, on the slow propagation step for alloy 600 in a‘complex’ environment (T = 320 ∞C) have been investigated in detail. The‘complex’ environment succeeds in obtaining a cracking at the approximaterate of 0.02 mm/h, similar to field experience.

Based on the mechanistic hypotheses detailed in Section 9.3, furtherinvestigations are necessary to verify the detrimental effect of alumino-silicate zeolite compounds and cyano ligands on IGSCC of alloy 600. Indeed,the results obtained in this study do not allow discrimination with confidencebetween the two hypothesis

Finally, it seems to be impossible, with the results obtained in this study,to model the slow propagation step of alloy 600 in the ‘complex’ environment.Indeed, the effects of the previous parameters (in particular the chemicalconditions) are not definite enough to allow the modeling of IGSCC velocityfor alloy 600 in the reference ‘complex’ environment.


9.6 References

1. Vaillant F., Prieux B., Cattant F., Stutzmann A., Lemaire P., ‘Saint-Laurent B1:expertises, chimie en fonctionnement et essais de corrosion’, Contribution of MaterialsInvestigation to the Resolution of Problems Encountered in Pressurized Water Reactors,International Symposium, Fontevraud III, 12–16 September 1994, SFEN, pp. 383–393.

2. Ollar P., Viricel-Honorez L., ‘Better understanding flow-restricted environmentsfrom hideout return analyses’, Contribution of Materials Investigation to the Resolutionof Problems Encountered in Pressurized Water Reactors, International Symposium,Fontevraud IV, 14–18 September 1998, SFEN, pp. 465–476.

3. Cattant F., Dupin M., Sala B., Gelpi A., ‘Analysis of deposits and underlying surfaceson the secondary side of pulled tubes from a French plant’, Contribution of MaterialsInvestigation to the Resolution of Problems Encountered in Pressurized Water Reactors,International Symposium, Fontevraud III, 12–16 September 1994, SFEN, pp. 469–480.

4. Sala B. Henry K., Lancha A.M., Dupin M., Combrade P., Erre R., Gelpi A., ‘Analysisof the deposits and surface layers on tubes pulled from PWR French steam generators’,Proc. Eurocorr’96, Nice (France), September 1996, paper IX OR 14.

5. Sala B., Gelpi A., Chevallier S., Dupin M., ‘Complementary investigations concerningthe analysis of the deposits and underlying surfaces observed on French PWR steamgenerator pulled tubes’, Proc. Intern. Symp. Fontevraud IV, September 1998, SFEN,p. 553.

6. de Bouvier O., Vaillant F., Millet L., Scott P. M., Tran Q. T. ‘Duplication in laboratoryof deposits, films and IGA/SCC damage observed on pulled steam generator tubes’,Contribution of Materials Investigation to the Resolution of Problems Encounteredin Pressurized Water Reactors, International Symposium Fontevraud V, 23–27September 2002, SFEN, p. 1049.

7. Tran T., Scott P., Vaillant F., ‘IGA/SCC of Alloy 600 in complex mixtures of impurities’,Proceedings of the tenth International conference on environmental degradation ofmaterials in nuclear power systems – Water reactors, South Lake Tahoe (NV), USA,August 5–9, 2001.

8. Vaillant F., Pavageau E. M., Bouchacourt M., Boursier J M., Lemaire P, ‘Modelingthe secondary side corrosion of tubings: a help to the maintenance policy of PWRsteam generators’, Proceedings of the ninth International conference on environmentaldegradation of materials in nuclear power systems – Water reactors, Newport Beach(CA), USA, August 1–5, 1999, p. 673.

9. Pavageau E. M., Vaillant F., de Bouvier O., Bouchacourt M., Caire J. P., Dalard F.,‘Secondary side corrosion modeling of alloy 600 for steam generator tubes based onlaboratory tests in sulfate environments‘, Proceedings of the tenth Internationalconference on environmental degradation of materials in nuclear power systems –Water reactors, South Lake Tahoe (NV), USA, August 5–9, 2001.

10. Santarini G., ‘Comprehensive interpretation of CERTs: a method for the characterizationand the prediction of IGSCC’, Corrosion, vol. 45, no. 5, p. 369.

130

10.1 Introduction

The ageing of light water reactor structural materials has been one of themajor factors affecting the availability, cost and safety of nuclear power.Along with fatigue and irradiation embrittlement, environmentally-assistedcracking (EAC) is a potential RPV ageing mechanism. The EAC behaviourof low-alloy steel (LAS) pressure boundary components under transient-free, steady-state boiling water reactor (BWR) power operation conditions,which covers the largest part of the lifetime, is well established [1–5]. Theseinvestigations have shown very low susceptibility to SCC crack growthunder static loading conditions in oxygenated, high-purity, high-temperaturewater/BWR environments at temperatures around 288 ∞C and also formedthe basis for the definition of the BWRVIP-60 SCC disposition lines forSCC crack growth in LAS during BWR power operation (Fig. 10.1, [3]).

On the other hand, the EAC behaviour during and after water chemistrytransients has hardly been investigated so far, but is also of great practicalrelevance because BWR operation inevitably involves periodic short-termvariations in water chemistry and oxygen/corrosion potential (ECP).Conductivity and oxygen/ECP transients occur during start-up/shut-downand occasionally during steady-state power operation (ion exchanger resinintrusions, condenser leakages, etc.). Nowadays, the extent (magnitude, period)and frequency of such transients is strongly minimised by following thecurrent EPRI BWR water chemistry guidelines (Table 10.1, [6]).

The possible effect of relatively short-term water chemistry transients onthe transition EAC crack growth behaviour during and, in particular, after atransient under subsequent steady-state power operation is of great interestfor safety assessments. Effects of oxygen/ECP transients on EAC crackgrowth in LAS have already been widely investigated in the context ofhydrogen water chemistry/noble metal chemical addition (NMCA) by Andresenet al. in tests with periodical partial unloading (PPU) [7]. There is an obviouslack of qualified and well documented testing with sulphate and chloride

10The effect of sulphate and chloride

transients on the environmentally-assistedcracking behaviour of low-alloy RPV steels

under simulated BWR conditions

S. R I T T E R and H. P. S E I F E R T, Paul Scherrer Institute(PSI), Switzerland

The effect of sulphate and chloride transients 131

transients under highly oxidising BWR/normal water chemistry (NWC)conditions at high ECP, where the most distinct and severe long-term/hysteresiseffects might be expected.

Therefore, the EAC crack growth behaviour of three different low-alloyRPV steels during and after sulphate and chloride transients was investigatedunder simulated BWR/NWC power operation conditions by tests with PPUand experiments under constant load. These tests should indicate, if hysteresisor long-term effects might occur under these highly oxidising conditions andreveal information on the transition behaviour during and after such a transient(response times, incubation periods, delay times, acceleration of crack growth,etc.). Furthermore, the adequacy and conservative character of the BWRVIP-60 SCC disposition lines (Fig. 10.1, [3]) for SCC crack growth in LASduring and after water chemistry transients was evaluated and assessed inthe context of the current EPRI BWR water chemistry guidelines (Table10.1).

Stress intensity factor Kl (MPa·m1/2)100806040200

10–8

10–9

10–10

10–11

da/d

t scc

(m

/s)

BWRVIP-60 SCC DL 1:Stationary power operation

BWRVIP-60 SCC DL 2:During and 100 h after transientsof water chemistry and load

Table 10.1 EPRI water chemistry guidelines for reactor water during BWR/NWCpower operation [6]

Control Parameter Action level 1 Action level 2 Action level 3

Conductivity [mS/cm] > 0.3 > 1.0 > 5.0Sulphate [ppb] > 5 > 20 > 100Chloride [ppb] > 5 > 20 > 100

10.1 BWRVIP-60 SCC disposition lines [3] for SCC crack growth inLAS. Line 1: Stationary, transient-free BWR power operation (< EPRIaction level 1), line 2: During and 100 h after water chemistrytransients (> EPRI action level 1).



10.2.1 Materials

Three different types of low-alloy, nuclear grade RPV steels with either alow or high sulphur content were investigated (Tables 10.2 and 10.3). TheRPV steels had a granular, bainitic microstructure with an average formeraustenitic grain size of 10 to 20 mm. The spatial distribution and morphologyof the MnS-inclusions was fairly homogenous and similar in alloys B and Ccovering the range from small, spherical to large (up to a few 100 mm),elongated inclusions. In alloy A distinct sulphur segregation zones with largeclusters of MnS inclusions were observed. The local sulphur content thereforesignificantly deviated from the average bulk sulphur content of 0.015 wt.%and varied between 0.003 and 0.053 wt.%.

10.2.2 Specimens

25 mm thick compact tension specimens (1T-C(T)) were used for allexperiments. They were manufactured in the T-L or L-T orientation.The specimens were pre-cracked by fatigue in air at room temperature,using a load ratio R of 0.1. The maximal KI at the final load step was£ 15 MPa · m1/2.

10.2.3 Environmental parameters

The tests were conducted in modern high-temperature water loops [8]. Waterchemistry (oxygen content and conductivity) and flow rate were measured atthe autoclave inlet and outlet. Inside the autoclave pressure and temperaturewere measured. The stainless steel autoclave volume of 10 litres was exchangedthree to four times per hour. In the vicinity of the specimens a flow velocityin the range of mm/s was obtained. The concentration of dissolved oxygen(DO) was adjusted by adding an argon-oxygen mixture to the storage tank.After the demineralised water in the storage tank was purified by ionexchangers, active coal and microfilters, the conductivity was controlled bydosing 0.02 M Na2SO4 or NaCl to the high-purity (£ 0.06 mS/cm) water.Conductivity and concentration of DO were controlled at the inlet water.Ionic impurities of the water (grab samples at inlet and outlet) were analysedby Inductive Coupled Plasma – Atomic Emission Spectroscopy (ICP – AES)and Ion Chromatography (IC) several times during each test.

The ECP of the specimens and the redox potential (platinum probe) werecontinuously monitored by use of an external Ag/AgCl/0.01 M KCl referenceelectrode. The specimens were electrically insulated from the autoclave andfrom each other by ZrO2 spacers.

The effect of sulphate and chloride transients133

Table 10.2 Chemical composition (in wt.%) and heat treatment of the investigated low-alloy RPV steels (WQ = water quenched, FC = furnacecooled, AC = air cooled, PWHT = post weld heat treatment

Alloy C Si Mn P S Ni Cr V Mo Al N Heat treatment

20 MnMoNi 5 5 A 0.25 0.33 1.54 0.014 0.015* 0.62 0.18 0.024 0.68 0.021 0.004 900 ∞C/9h/WQ,(∫ SA 508 Cl.3) 650 ∞C/34h/AC,

660 ∞C/14h/AC+ PWHT

22 NiMoCr 3 7 B 0.22 0.20 0.91 0.008 0.007 0.88 0.42 0.010 0.53 0.018 0.008 895 ∞C/7h/WC,(∫ SA 508 Cl.2) 645 ∞C/17h/AC

+ PWHT

SA 533 B Cl.1 C 0.25 0.24 1.42 0.006 0.018 0.62 0.12 0.007 0.54 0.030 0.006 915 ∞C/12h/AC,(∫ 20 MnMoNi 5 5) 860 ∞C/12 h/WQ,

660 ∞C/12 h/FC+ PWHT

*local sulphur content: 0.003–0.053 wt.%)

Table 10.3 Mechanical properties (tensile tests in air, DIN 50125, B5 ¥ 50-specimens, Rp0.2 = yield stress, Rm = tensile strength, A5 = elongationat fracture, Z = reduction of area)

Room temperature 288 ∞C

Alloy Rp0.2 [MPa] Rm [MPa] A5 [%] Z [%] Rp0.2 [MPa] Rm [MPa] A5 [%] Z [%]

20 MnMoNi 5 5 A 512 663 19 56 462 618 17 51

22 NiMoCr 3 7 B 467 605 17 72 400 578 16 70

SA 533 B Cl.1 C 456 618 23 60 412 588 21 55


10.2.4 Crack growth monitoring and fractographical post-test evaluation

Crack advance was continuously monitored using the reversed direct currentpotential drop (DCPD) method with a resolution limit corresponding to roughly5 mm [8]. The crack growth increment was calculated by the Johnson formula[9]. The calculated crack length at the end of the experiment was then verifiedand, if necessary, corrected with regard to the mean final crack length <a0 +DaEAC> as revealed by post-test fractography [8]. The crack growth rates(CGR) were determined by linear fit of the crack increment versus timecurve. After the test, the specimens were broken apart at liquid nitrogentemperature for post-test evaluation. For fractographical analysis in the scanningelectron microscope, the oxide film on the fracture surface of one specimenhalf was removed by galvanostatic reduction in an ENDOX-bath [10].

10.2.5 Mechanical loading

Two pre-cracked specimens were investigated simultaneously under the testconditions in oxygenated high-temperature water in a daisy chain. The loadwas actuated with a screw-driven, electro-mechanical tensile machine withcomputer control. The KI values were calculated according to ASTM E 399by the measured load and by the actual mean crack length <a0 + DaEAC>,derived by the DCPD method and by post-test fractographical evaluation[8].

10.2.6 Test procedure of the sulphate transientexperiment (Test 1)

In the first test, the effect of a sulphate transient on an actively growing EACcrack in two low-alloy RPV steels with different sulphur contents (alloy Aand B) under low-flow and highly oxidising BWR conditions was investigatedunder PPU (constant load amplitude loading with trapezoid waveform)conditions. The four major experimental phases of this transient test areshown in Fig. 10.2. After achieving the desired environmental conditions,the specimens were pre-oxidised in the test environment (8 ppm DO, kinlet =0.06 mS/cm) under a small mechanical pre-load. Before applying the sulphatetransient, an EAC CGR in the range of the ‘low-sulphur SCC line’ of the GEmodel [11] was generated in the initially high-purity water by PPU(asymmetrical trapezoid loading at high load ratio R of 0.8 with a rise timeDtR of 1000 s and long hold time at maximum load DtH of 5 h). The testingphase in oxygenated, high-temperature water at 288 ∞C consisted on a sequenceof three different water chemistry conditions WC 1 to 3 with different sulphateconcentrations (Fig. 10.2 and Table 10.4). After 170 h in high-purity water,


the sulphate level was increased within 1 h from < 0.6 to 368 ppb by dosing0.02 M Na2SO4. The sulphate level was then kept constant for further 310 h(WC 2). Afterwards, the sulphate level was decreased within 2.6 h to a valueof < 0.6 ppb and the high-purity water chemistry conditions (WC 3) weremaintained for further 240 h before the specimens were unloaded.

10.2.7 Test procedure of the chloride transientexperiments (Tests 2 and 3)

The effect of chloride transients on the EAC behaviour of three low-alloyRPV steels with different sulphur contents under low-flow and highly oxidisingBWR conditions was investigated by two different tests under PPU andconstant load conditions.

Test 2 (50 ppb chloride, PPU)

In this test the effect of a chloride transient on an actively growing EACcrack in two low-alloy RPV steels with different sulphur contents (alloy Aand B) under low-flow and highly oxidising BWR conditions was investigatedunder PPU conditions. The four major experimental phases of this transienttest were similar to the sulphate transient test (see above) and are shown inFig. 10.2. Instead of sulphate, chloride was added to the high-purity waterfor 40 h (WC 2). The chloride level was increased within 1 h from < 0.4 to

10.2 Simplified schematic of the test procedures of tests 1 to 3 withthe major experimental phases.

Test 1 (368 ppb SO4 )Test 2 (50 ppb Cl–)

2–

Test 3 (20 ppb Cl–)

Time

Coolingphase

Testingphase

Conditioningphase

Heatingphase

Pre-test

phase

DtR

DtR

DtR

(CO2)

Load

Load

k(Na2SO4or NaCl)

O2(ECP)

T


50 ppb by dosing 0.02 M NaCl. Afterwards, the chloride level was decreasedwithin 1.6 h to a value of < 0.4 ppb and the high-purity water chemistryconditions (WC 3) were maintained for further 170 h before the specimenswere unloaded.

Test 3 (20 ppb chloride, constant load)

The EAC behaviour of alloy C and B during and after a chloride transientwas investigated in a third test under constant load. In this test 20 ppb chloridewas added for 290 h under pure constant load under otherwise similarenvironmental conditions as in test 1 and 2. After 240 h of NaCl dosing asingle partial unloading with a load ratio R of 0.7 was performed (DtFall =200 s, DtRise = 1000 s). Then the load was kept constant for further 1130 h.


10.3.1 Effects of a sulphate transient on the EACbehaviour

The average EAC CGR of the two specimens (alloy A and B) during thewater chemistry phases WC 1 to 3 are summarised in Table 10.4. In materialA no effect of the sulphate transient on the EAC CGR was observed. TheEAC CGR before, during and after the sulphate transient were almost identical(Table 10.4). The very small increase of the CGR from WC 1 to 3 wasprimarily the result of the increasing KI,max value. The PPU (R = 0.8, KI,max

= 69–85 MPa·m1/2, DtR = 1000 s, DtH = 5 h) under these highly oxidisingconditions resulted in stable EAC crack growth in the range of the BWRVIP-60 SCC disposition line 2. During the PPU at high KI,max values it waspossible to resolve the crack growth in each trapezoid cycle and to qualitativelydifferentiate between the EAC crack growth during the slow rising load andthe constant load part of the cycle (Fig. 10.3). The crack mainly grew duringthe rising load phase with some very minor crack advance (and very lowCGR) during the subsequent constant load phase. The crack arrested eitherduring the constant load or during the unloading part of the cycle and re-initiated again during the rising load part of the next cycle. In accordance tothe GE model [11], the absence of an acceleration of EAC crack growth maybe attributed to ‘high-sulphur’ crack chemistry conditions, which alreadyexisted during the high-purity water chemistry phase WC 1 because of thehigh ECP and the dissolution of MnS-inclusions in the enclave of the incipientcrack. Therefore, an increased bulk sulphate concentration did not result inan acceleration of the crack growth.

In material B continuous cessation of EAC crack growth was observed inthe high-purity water chemistry phase WC 1. The crack was growing with a

The effect of sulphate and chloride transients137

Table 10.4 Summary of the most important environmental parameters and the mean CGR during the three water chemistry phases oftest 1 (WC 1–3). WC 2 = sulphate transient

Alloy A Alloy B

Test phase WC 1 WC 2 WC 3 WC 1 WC 2 WC 3

<ECP> [mVSHE] +110 +70 +115 +130 +100 +120

<kinlet> [mS/cm] 0.07 1.00 0.06 0.07 1.00 0.06

SO 42– [ppb] < 0.6 368 < 0.6 < 0.6 368 < 0.6

KI,max [MPa·m1/2] 68.8–70.9 70.9–78.0 78.0–84.7 61.9 62.0 62.0–65.9

<da/dt> [m/s] 8.1 ¥ 10–10 1.4 ¥ 10–9 1.6 ¥ 10–9 1.4 ¥ 10–11 1.4 ¥ 10–11 1.6 ¥ 10–9


A: 20 MnMoNi 5 5, 0.015 wt.% ST = 288 ∞C, DO = 8 ppm, ECP = 71 mVSHEWC 2: k = 1.00 mS/cm, 368 ppb SO4

2–

Load42

40

38

36

34

32

Load

(kN

)

Time (h)725720715710

25.06

25.04

25.02

25.00

24.98

Cra

ck le

ng

th (

mm

)

10.3 Crack length of alloy A specimen during the sulphate transient.Test 1.

CGR just slightly above the detection limit of the DCPD (<da/dt> ª 1.4 ¥10–11 m/s) during the water chemistry phases WC 1 and 2. The sulphatetransient did not result in an acceleration of the very slow EAC crack growth.65 h after returning to high-purity water, EAC re-initiated. After re-initiationfollowed by a short transition period, stable and stationary EAC crack growthwith a CGR of 1.6 ¥ 10–9 m/s (51 mm/year) in the range of the ‘low-sulphurSCC crack growth curve’ of the GE model [11] was observed in alloy B. TheCGR was almost identical to that in alloy A. It is believed, that the re-initiation of fast EAC in alloy B was not the direct result of a delayed effectof the sulphate transient, but rather a result of the probabilistic nature of theEAC cessation/pinning/arrest phenomena and of the re-initiation process[12].

The same crack growth behaviour has also been observed by MPA Stuttgartin the framework of the CASTOC programme [13] and in many otherinvestigations of PSI [1, 4]. In these experiments under comparableenvironmental conditions (T = 288 ∞C, DO = 0.4 or 8 ppm, ECP = 0 to +150mVSHE), all investigated RPV steels with different sulphur contents revealedthe same crack growth behaviour with comparable CGR under identicalloading conditions in high-purity water and in water with very high levels ofsulphate (added as H2SO4 or Na2SO4). Even very high sulphate contents ofup to 1400 ppb (ca. 10 mS/cm) did not result in an acceleration of EAC crackgrowth. Fast growing cracks, triggered by cyclic or slow rising loadingarrested immediately after switching to constant load at stress intensity factorsof up to 53 MPa·m1/2. In spite of the absence of any accelerating effect underhighly oxidising BWR conditions, sulphate still remains a harmful species


for EAC in LAS, since it has been observed to affect EAC initiation fromsmooth surfaces in LCF and SSRT tests and to accelerate EAC crack growthunder reducing PWR conditions [5]. Furthermore it may help to overcomecrack pinning/arrest phenomena and re-initiation problems.

10.3.2 Effect of chloride on the EAC behaviour

Test 2 (50 ppb chloride, PPU, [14])

The mean EAC CGR of the two alloys during the individual water chemistryphases WC 1 to 3 are summarised in Table 10.5. In Fig. 10.4 the crackgrowth behaviour of material A during the chloride transient is shown. Anaccelerating effect of the chloride transient on the EAC CGR was observed.The PPU (R = 0.73, KI,max = 51–67 MPa·m1/2, DtR = 1000 s, DtH = 12 h)during the high-purity water chemistry phase WC 1 resulted in stable EACcrack growth slightly above the BWRVIP-60 disposition line 2.

In the high-purity water chemistry phases WC 1 and 3 the crack growthbehaviour was similar to that during the sulphate transient experiment, asdescribed before (Fig. 10.3). 3 h after adding NaCl to the high-purity water,onset of fast EAC occurred during the constant load phase of a PPU cycle(Fig. 10.5). 17 h later, the EAC crack growth reached a stationary stateduring the rising (da/dtRL = 1.4 ¥ 10–7 m/s) and constant load (da/dtCL = 1.8¥ 10–8 m/s) part of the next PPU cycle with a mean CGR of 2.2 ¥ 10–8 m/s(694 mm/year) at stress intensity factor values KI of 55 to 62 MPa·m1/2. Afterstopping the NaCl dosage, the crack was further growing with the same highCGR under constant load until the next partial unloading (Fig. 10.4). Thenthe crack growth started to slow down to a mean CGR in the same range (andwith the same behaviour: see Fig. 10.3) as observed before the chloridetransient. The slightly higher CGR was assigned to the higher KI,max values.The decay of the crack growth after the chloride transient might have beencaused by crack closure effects during the partial unloading. Therefore, apossible long-term effect of a chloride transient on the EAC crack growthbehaviour under constant load could not be fully excluded based on thisexperiment.

According to the EPRI BWR/NWC water chemistry guidelines (Table 1),the chloride concentration has to be reduced below the action level 2 within24 h from the time of occurrence by adequate correction actions. Otherwisean orderly shut-down shall be initiated. If it is foreseeable that the parameterwill be below the action level 2 value within the time period required toachieve an orderly shut-down, power operation can be maintained. Theexperimentally observed short incubation period of 3 h for acceleration ofEAC is therefore significantly shorter than the maximum allowable timeinterval for returning to normal operating conditions. The short incubationperiod for acceleration of EAC in combination with the very high SCC CGR

Corrosion issues in light w

ater reactors140

Table 10.5 Summary of the most important environmental parameters and the mean CGR during the three water chemistry phases oftest 2 (WC 1–3). WC 2 = chloride transient

Alloy A Alloy B


<ECP> [mVSHE] +105 +110 +125 +130 +125 +150

<kinlet> [mS/cm] 0.06 0.23 0.06 0.06 0.23 0.06

Cl– [ppb] < 0.4 50 < 0.4 < 0.4 50 < 0.4

KI, max [MPa·m1/2] 51.5–52.9 54.2–62.1 64.8–67.2 49.5–49.6 49.6–49.8 49.8–50.4

<da/dt> [m/s] 1.2 ¥ 10–9 2.2 ¥ 10–8 1.8 ¥ 10–9 2.7 ¥ 10–11 2.5 ¥ 10–9 4.6 ¥ 10–10*


under constant load of almost 1 mm/day arise some safety concern for severechloride transients under BWR/NWC conditions.

Test 3 (20 ppb chloride, constant load)

The mean EAC CGR of the investigated alloys C and B during the individualwater chemistry phases WC 1 to 3 are summarised in Table 10.6. In both

A: 20 MnMoNi 5.5, 0.015 wt.% ST = 288 ∞C, DO = 8 ppm,ECP = 110 mVSHE

(< 0.4 ppb Cl–)WC 3WC 2WC 1

(50 ppb Cl–)(< 0.4 ppb Cl–)

Inlet

1.8 ¥ 10–9 m/s2.2 ¥ 10–8 m/s

1.2 ¥ 10–9 m/s

Kl = 52 – 67 MPa.m1/2

0.25

0.20

0.15

0.10

0.05

0.00

Co

nd

uct

ivit

y (m

S/c

m)

Time (h)400 450 500 550 600

28.0

27.0

26.0

25.0

24.0

Cra

ck le

ng

th (

mm

)

10.4 Crack length of alloy A specimen with mean EAC CGR (linear fit)before during and after the chloride transient. Test 2.

WC 2WC 1(50 ppb Cl–)(< 0.4 ppb Cl–)

Load

Onset of fast EACunder constant load

32

30

28

26

24

22

Load

(kN

)

475470 480 485 490 495Time (h)

24.4

24.3

24.2

24.1

24.0

23.9

Cra

ck le

ng

th (

mm

)

A: 20 MnMoNi 5.5, 0.015 wt.% ST = 288 ∞C, DO = 8 ppm, ECP = 110 mVSHEk = 0.06/0.23 mS/cm, < 0.4/50 ppb Cl–

10.5 Crack length of alloy A specimen at the beginning of thechloride transient. Test 2.


ater reactors142

Table 10.6 Summary of the most important environmental parameters and the mean CGR during the three water chemistry phases oftest 3 (WC 1–3). WC 2 = chloride transient.

Alloy C Alloy B


<ECP> [mVSHE] +140 +140 +140 +145 +145 +145

<kinlet> [mS/cm] 0.06 0.13 0.06 0.06 0.13 0.06

Cl– [ppb] < 2 20 < 2 < 2 20 < 2

KI, max [MPa·m1/2] 38.0–38.3 38.3–44.2 44.2–52.4 32.1 32.1–32.3 32.3

<da/dt> [m/s] 1.2 ¥ 10–10 5.8 ¥ 10–9 7.0 ¥ 10–10 < 4.0 ¥ 10–12* 1.1 ¥ 10–9 < 1.4 ¥ 10–12*

*Close to detection limit of the DCPD measurement.


investigated materials no accelerating effect of the chloride transient couldbe observed for the first 240 h of chloride addition. Therefore a single partialunloading and reloading (R = 0.7) of the specimens was performed. Duringthe reloading onset of fast EAC was observed in both specimens.

The EAC crack growth behaviour of alloy C during and after the chloridetransient is presented in Fig. 10.6. In alloy C onset of fast EAC crack growthoccurred during the reloading due to dynamic straining of the crack-tip.After 4 h under pure constant load sustained, stable SCC with a CGR valueof 5.8 ¥ 10–9 m/s (183 mm/year) was observed (Fig. 10.7). The crack wasgrowing with same rate for further 40 h (delay time) after returning to high-purity water. Then the crack growth slowed down to a mean CGR of 7.0 ¥10–10 m/s (22 mm/year), which was still significantly higher than before thetransient. The crack growth after the chloride transient consisted of periodicsequences of longer phases with slightly lower CGR (3 to 5 ¥ 10–10 m/s) andshorter phases with significantly faster transient CGR (5 ¥ 10–9 m/s) than thelong-term mean value of 7.0 ¥ 10–10 m/s (Fig. 10.6). This special shape ofcrack growth was caused by local crack pinning (periods with slightly slowerCGR) and subsequent retarded and stepwise failure of these uncrackedligaments (periods with higher transient CGR) by EAC or ductile tearing,which was further confirmed by post-test fractography. Similar shapes ofcrack growth versus time curves were observed in many other PSI tests withfast SCC under constant load [4].

Stable, fast SCC with CGR significantly above the BWRVIP-60 SCCdisposition line 1 and 2 could be sustained for more than 1100 h after returning

C: SA 533 B Cl.1, 0.018 wt.% ST = 288 ∞C, DO = 8 ppm,ECP = 140 mVSHE

WC 2(20 ppb Cl–)

WC 1 WC 3(< 2 ppb Cl–)

Inlet

5.8 ¥ 10–9 m/s1.2 ¥ 10–10 m/s

7.0 ¥ 10–10 m/s

(4 ¥ 10–10 m/s)

(5 ¥ 10–9 m/s)

0.12

0.09

0.00

0.03

0.06

Co

nd

uct

ivit

y (m

S/c

m)

Time (h)2100180015001200900

27.0

26.5

26.0

25.5

25.0

24.5

24.0

Cra

ck le

ng

th (

mm

)

1 partialunloading(R = 0.7)

KI = 38–52 MPa.m1/2

10.6 Crack length of alloy C specimen with mean EAC CGR (linear fit)during and after the chloride transient. Test 3.


to high-purity water in alloy C. This showed that there might be a distinctivelong-term effect after severe (≥ EPRI action level 2) and prolonged chloridetransients.

Alloy B also showed accelerated EAC crack growth during the chloridetransient, although the CGR slowed down to values below the detection limitof the DCPD measurement 35 h after returning to high-purity water.

10.3.3 Comparison with the BWRVIP-60 SCC dispositionlines

The EAC CGR results of the different water chemistry transient tests underPPU and constant loading conditions with alloy A, B and C were comparedto the BWRVIP-60 SCC disposition lines [3]. In Fig. 10.8, the SCC CGRfrom constant load tests [1, 4, 12, 13] in oxygenated high-temperature water(0.4 or 8 ppm DO, k = 0.25 to 10 mS/cm, T = 288 ∞C) with alloy A, B andC and different amounts of sulphate (65 to 1400 ppb) added either as Na2SO4

or H2SO4 were compared to the BWRVIP-60 SCC disposition lines. Evenfor extremely high sulphate levels significantly above the EPRI action level3, the SCC CGR were well below the BWRVIP-60 SCC disposition line 2for water chemistry transients. Furthermore, fast growing EAC cracks, triggeredby cyclic or slow rising loading arrested immediately after switching toconstant load at stress intensity factors of up to 60 MPa·m1/2 [1–4]. Theseresults clearly confirmed the very conservative character of the BWRVIP-60SCC disposition line 2 for sulphate transients even exceeding the EPRIaction level 3.

Load

5.8 ¥ 10–9 m/s

Onset offast EAC

22

20

18

16

14

12

Load

(kN

)

Time (h)1065 1070106010551050

24.4

24.3

24.2

24.1

24.0

23.9

Cra

ck le

ng

th (

mm

)

C: SA 533 B Cl.1, 0.018 wt.% ST = 288 ∞C, DO = 8 ppm, ECP = 140 mVSHEWC 2: k = 0.13 mS/cm, 20 ppb Cl–

10.7 Crack length of alloy C specimen at the beginning of thechloride transient (partial unloading). Test 3.


In contrast to the sulphate transient experiment (368 ppb SO )42– , a relatively

low chloride concentration (50/20 ppb Cl–, ≥ EPRI action level 2) had adistinct accelerating effect on the EAC crack growth and resulted in sustained,stationary SCC crack growth under constant load under simulated highlyoxidising BWR conditions. As shown in Fig. 10.9, the SCC CGR of alloy A,B and C during chloride transients clearly exceeded the BWRVIP-60 SCC

BWRVIP-60 SCC DL 2BWRVIP-60 SCC DL 2

T = 288 ∞C, DO = 8/0.4 ppm,ECP = 70 – 140 mVSHE

: 0.25 mS/cmAlloy B (Na2SO4):

Alloy A (H2SO4, [14]):: 1 mS/cm: 5 mS/cm: 10 mS/cm

Alloy C (Na2SO4):

: 0.25 mS/cm: 1 mS/cm

102

101

100

10–1

da/d

t (m

m/y

ear)

da/d

t scc

(m/s

)

10–12

10–11

10–10

10–9

10–8

20 30 40 50 60 70 80 90 100Stress intensity factor Kl (MPa·m1/2)

10.8 Effect of sulphate on SCC CGR under constant load. Comparisonwith the BWRVIP-60 SCC DL [3].

10.9 Effect of chloride on SCC CGR under constant load. Comparisonwith the BWRVIP-60 SCC disposition lines [3].

High-sulphur line, GE modelBWRVIP-60 SCC DL 2BWRVIP-60 SCC DL 1

T = 288 ∞C, DO = 8 ppm,ECP = 70 – 140 mVSHE

Alloy A, 50 ppb Cl–

Alloy B, 50 ppb Cl–

Alloy B, 50 ppb Cl–

Alloy C, 20 ppb Cl–

Alloy C, <2 ppb Cl–, after Trans. of 20 ppb Cl–

104

103

102

101

100

10–1

da/d

t scc

(mm

/yea

r)

Stress intensity factor Kl (MPa·m1/2)70 806050403020

10–7

10–8

10–9

10–10

10–11

10–12

da/d

t ssc

(m/s

)


disposition line 2 for water chemistry transients. In several cases stable,stationary SCC with CGR above the BWRVIP-60 SCC disposition line 2could be sustained after severe (≥ EPRI action level 2) and prolonged chloridetransients for much longer periods (> 1000 h) than the 100 h interval suggestedby BWRVIP-60 and the CGR remained well above the BWRVIP-60 dispositionline 1 for stationary, transient-free BWR power operation. The short incubationperiod (few hours) under PPU for acceleration of SCC in combination withthe very high SCC CGR under constant load of up to 1 mm/day and possiblelong-term effects arise some concern for severe chloride transients underBWR/NWC conditions. These important results and the non-conservatismof the BWRVIP-60 SCC disposition lines for severe chloride transients shouldtherefore be further investigated at lower chloride concentrations adjacent tothe EPRI action level 1 and shorter transient periods, both for lower DO of0.4 ppm and lower KI values.


The adequacy and conservative character of the BWRVIP-60 stress corrosioncracking (SCC) disposition lines during and after water chemistry transientswas evaluated and assessed in the context of the current EPRI boiling waterreactor (BWR) water chemistry guidelines. For that purpose, the SCC behaviourof three nuclear grade low-alloy RPV steels during and after sulphate andchloride transients was investigated under simulated BWR power operationconditions by tests with periodical partial unloading (PPU) and experimentsunder constant load. Modern high-temperature water loops, on-line crackgrowth monitoring (DCPD) and fractographical analysis by scanning electronmicroscope were used to quantify the cracking response.

In oxygenated, high-temperature water (T = 288 ∞C, 8 ppm dissolvedoxygen (DO)), the addition of 368 ppb sulphate (≥ EPRI action level 3) didnot result in acceleration of crack growth under PPU and constant load in allmaterials and the SCC crack growth rates (CGR) under constant load duringsulphate transients were conservatively covered by the BWRVIP-60 SCCdisposition line 2. Both, in high-purity water or water with 368 ppb sulphate,no sustained SCC was observed. Under PPU conditions, the cracks wereonly growing during the rising load phase of the PPU cycles. The absence ofan acceleration of EAC crack growth was attributed to ‘high-sulphur’ crackchemistry conditions, which already existed before the sulphate transientbecause of the high corrosion potential and the dissolution of MnS-inclusionsin the pre-crack enclave.

The addition of 20 to 50 ppb chloride (≥ EPRI action level 2) resulted inacceleration of the SCC crack growth in all materials by at least one order ofmagnitude and in fast, stationary SCC under constant load in the investigatedstress intensity factor range KI from 32 to 62 MPa·m1/2 with CGR significantly


above the BWRVIP-60 SCC disposition line 2 and close to the ‘high-sulphurSCC crack growth curve’ of the GE model. In several cases stable, stationarySCC with CGR above the BWRVIP-60 SCC disposition lines 1 and 2 couldbe sustained after severe (≥ EPRI action level 2) and prolonged chloridetransients for much longer periods (> 1000 h) than the 100 h interval suggestedby BWRVIP-60. The short incubation period (few hours) under PPU foracceleration of SCC in combination with the very high SCC CGR underconstant load of up to 1 mm/day and possible long-term effects, at least aftersevere and prolonged chloride transients, arise some concern for severechloride transients under BWR/NWC conditions. These important resultsand the non-conservatism of the BWRVIP-60 SCC disposition lines for severechloride transients should therefore be further investigated at lower chlorideconcentrations adjacent to the EPRI action level 1 and shorter transientperiods, both for lower DO of 0.4 ppm and lower KI values.


This work has been performed within the CASTOC (5th EC FW programme,participants: MPA Stuttgart – D, CIEMAT Madrid – ES, NRI Řež – CZ, PSIVilligen – CH, Framatome ANP GmbH Erlangen – D, VTT Espoo – FIN,[13]) and RIKORR-I project. The financial support for this work by theSwiss Federal Office for Education and Science (BBW), the Swiss FederalNuclear Safety Inspectorate (HSK) and the Swiss Federal Office of Energy(BFE) is gratefully acknowledged. Thanks are also expressed to U. Ineichen,U. Tschanz, B. Gerodetti, and E. Groth (all PSI) for their experimentalcontribution to the project.

10.6 References

1. J. Heldt, H.P. Seifert, Nuclear Eng. & Design, Vol. 206, 2001, pp. 57–89.2. D. Blind, F. Hüttner, A. Wünsche (MPA Stuttgart), K. Küster (HEW), H.P. Seifert,

J. Heldt (PSI), A. Roth (Siemens KWU), P. Karjalainen-Roikonen, U. Ehrnstén(VTT), ‘European Round Robin Test on Constant Load EAC Tests of Low AlloySteels under BWR Conditions’, 9th Int. Conf. on Environmental Degradation ofMaterials in Nuclear Power Systems – Water Reactors, ANS/NACE/TMS, Aug. 1–5, 1999, Newport Beach, CA, USA, pp. 911–919.

3. F.P. Ford, R.M. Horn, J. Hickling, R. Pathania, G. Brümmer, ‘Stress CorrosionCracking of Low Alloy Steels under BWR Conditions; Assessments of Crack GrowthRate Algorithms’, 9th Int. Conf. on Environmental Degradation of Materials inNuclear Power Systems–Water Reactors, ANS/NACE/TMS, Aug. 1–5, 1999, NewportBeach, CA, USA, pp. 855–863.

4. H.P. Seifert, S. Ritter, ‘Environmentally-Assisted Cracking of Low-Alloy ReactorPressure Vessel Steels under Boiling Water Reactor Conditions’, PSI-Report 02-05,ISSN 1019-0643, Feb. 2002.


5. H.P. Seifert, ‘Literature Survey on the SCC of Low-Alloy Steels in High-TemperatureWater’, PSI-Report 02-06, ISSN 1019-0643, Feb. 2002.

6. M. Lasch, U. Staudt, ‘Die VGB-Richtlinie fuer SWR-Anlagen im internationalenVergleich’, VGB-Kraftwerkstechnik, Vol. 75, 1995, pp. 745–750.

7. P.L. Andresen, L.M. Young, ‘Crack-Tip Microsampling and Growth Rate Measurementsin Low-Alloy Steel in High-Temperature Water’, Paper No. 156, NACE Corrosion,1995.

8. H.P. Seifert, S. Ritter, ‘PSI Contribution to the CASTOC Round Robin on EACof Low-Alloy RPV Steels under BWR Conditions, PSI-Report 01-08, ISSN1019-0643, Aug. 2001.

9. H.H. Johnson, Materials Res. Stand., Vol. 15, 1978, pp. 89–111.10. P.M. Yukawich, C.W. Hughes, Practical Metallography, Vol. 20, 1997, pp. 1–12.11. F.P. Ford, ‘Environmentally Assisted Cracking of Low-Alloy Steels’, EPRI NP-

7473-L. Electric Power Research Institute, Jan. 1992.12. S. Ritter, H.P. Seifert, ‘Effect of a Sulphate Transient on the EAC Crack Growth

Behaviour of Low-Alloy RPV Steels under Simulated BWR Operating Conditions’,PSI Report No. 02-09, ISSN 1019-0643, Mar. 2002.

13. J. Föhl, U. Ehrnstén, M. Ernestová, D. Gómez-Briceño, J. Lapeña, S. Ritter, A. Roth,B. Devrient, H.P. Seifert, T. Weissenberg, M. Žamboch, ‘Crack Growth Behaviourof Low-Alloy Steels for Pressure Boundary Components under Transient Light WaterReactor Operating Conditions–CASTOC’, FISA-Conference on EU Research inReactor Safety, Luxembourg, Nov. 10–12, 2003.

14. S. Ritter, H.P. Seifert, ‘Effect of a Chloride Transient on the EAC Crack GrowthBehaviour of Low-Alloy RPV Steels under Simulated BWR Operating Conditions’,PSI Report No. 02-23, ISSN 1019-0643, Nov. 2002.

149

11.1 Introduction

Some pressurised-water nuclear-power plants are equipped with horizontalsteam-generators in which heat-exchange tubes, exposed to high-temperature(H-T) water, are made from titanium-stabilised stainless steel. There havebeen few instances of cracking during over 20 years of service provided thatappropriate water chemistries are maintained, but there are concerns thatstress-corrosion cracking (SCC) could become a significant problem beforedesired lifetimes are achieved. SCC is most likely to initiate and grow fromthe outside surfaces of tubes at crevices between the tubes and the tube-support plates where ionic impurities in the water can become concentrated.

Previous studies have shown that the susceptibility of stainless steels toSCC in H-T water is sensitive to impurities such as chloride ions,electrochemical potential, and degree of sensitisation of the steel [1]. One ofthe present authors (and co-workers) [2, 3] have determined threshold stress-intensity-factors (KIEAC) and crack-growth rates for a titanium-stabilisedstainless steel in high-temperature (270 ∞C) alkaline, neutral, and acidenvironments that could possibly occur in crevices (Fig. 11.1). Crack-growthrates were generally in the range 10–9 to 5 ¥ 10–8 m/s for a range of crack-opening displacement rates (1–100 mm/h) in all the environments (Fig. 11.2).Comparison of KIEAC values from these tests with the stress-intensity factorsfor small cracks that might develop under service loading suggested thatsmall (0.1 mm) cracks should not grow under normal service-loading conditionswithout taking into account residual stresses and the effects of stress bi-axiality on K IEAC

2 (Fig. 11.3).The most significant differences observed between the different

environments were (i) much shorter crack-initiation times for the acidenvironment compared with the neutral and alkaline environments, (ii)somewhat larger crack-growth rates for the alkaline environment comparedwith the neutral environment at low crack-opening displacement (COD)rates, and (iii) somewhat shorter crack-initiation time (and lower KIEAC) for

11Transgranular stress-corrosion cracking in

austenitic stainless steels at hightemperatures

A. B R O Z O VA, Nuclear Research Institute, Czech Republicand S. L Y N C H, Monash University, Australia


NeutralAlkaline

1000.0100.010.01.00.1Crack opening displacement rate (mm/hour)

08Kh18N10T in neutral and alkaline crevice solution,Na, SO4, Cl, H3SiO4, 270∞C

70

60

50

40

30

20

0

10

Kin

i (M

Pam

1/2 )

11.1 Results of rising displacement tests on SENT specimens [3] inneutral and alkaline solutions specified in Table 11.1: Plot of stress-intensity factor, Kini, at crack-growth initiation versus test rate. Thethreshold values, KIEAC, can be easily determined as minimum limitvalues of the data sets. Note the different rate dependencies inneutral and alkaline solutions.

NeutralAlkaline

Crack opening displacement rate (mm/hour)

08Kh18N10T in neutral and alkaline crevice solution,Na, SO4, Cl, H3SiO4, 270∞C

1000.0100.010.01.00.1

Cra

ck g

row

th r

ate

(m/s

)

1.E-06

1.E-07

1.E-08

1.E-09

1.E-10

1.E-11

11.2 Crack-growth rate versus test-rate data from RDT tests [3] inneutral and alkaline solutions specified in Table 11.1 showing asignificant rate-dependency in the alkaline solution but not in theneutral solution.

Transgranular stress-corrosion cracking 151

the alkaline environment compared with the neutral one [2, 3]. SCC wastransgranular (T) and fracture surfaces appeared to be similar in all theenvironments. Fracture surfaces were cleavage-like and exhibited crack-arrest markings (CAMs), as has been observed previously for some stainlesssteels tested in high-temperature aqueous environments [1, 4].

In this chapter, more detailed observations of fracture surfaces than havebeen made previously are reported. The aim of the observations was to geta better understanding of the mechanisms and kinetics of SCC. Proposedmechanisms for cleavage-like cracking are therefore reviewed in the light ofthe present observations and those made by others. A better fundamentalunderstanding of SCC can be valuable in terms of making engineeringjudgements regarding component lifetimes estimated on the basis of‘accelerated’ testing.


Details of experimental procedures and materials can be found elsewhere [2,3]. In summary, tests were carried out on C-ring and single-edge-notch tensilespecimens under rising-load displacements at 270 ∞C. An austenitic stainless

b

SGT with crack depth 0.1 mm

SENT, a/W = 0.2

SENT, a/W = 0.5

CT, a/W = 0.5

0.60.40.20–0.2–0.4–0.6

70

60

50

40

30

20

10

0

Th

resh

old

str

ess

inte

nsi

ty K

IEA

C (

MP

am1/

2 )

11.3 Effect of biaxiality b, calculated from T stress, crack length and K

value as: b p =

I

T aK

, on the threshold KIEAC value, previous

experimental results [2] on CT and SENT specimens with two crackdepths. Extrapolation is made for steam generator tube (SGT), ∆16 ¥1.4 mm, with outside surface crack depth 0.1 mm.


steel, equivalent to AISI 321, with a grain size ~13 mm and yield strength ofabout 280 MPa (at 300 ∞C) was used. The environment compositions, pHvalues, and potentials are listed in Table 11.1. Fracture surfaces were coveredwith thick oxide after testing and were cleaned either by immersion in boilingdistilled water or ammonium nitrate at 75–90 ∞C (standard ASTM G1-90procedure C3.4).

11.3 Results

Scanning-electron microscopy (SEM) at high magnifications (up to 60,000¥)showed that some of the fracture surfaces after cleaning were remarkablyuncorroded, and features as small as 20 nm, produced by the fracture process,were resolved. For example, numerous steps, some of which were serratedon a fine scale, and CAMs with spacings as small as 50 nm were observedin some areas (Fig. 11.4). Coarser steps, with undercutting, and numeroussecondary cracks were also evident. Occasional large dimples were observedin cleavage-like areas at high K values, suggesting that large, localised strainswere associated with cracking. (Other studies have also shown evidence ofsubstantial localised plasticity.) The fracture-surface appearance was generallysimilar for the different environments, but it was difficult to determine if

Table 11.1 Composition (mole/litre), pH, and potential (mV versus SHE) of testsolutions

Na Cl H3SiO4 SO4 pHT Potential

Acid 6.0 2.00 0.003 3.00 2.7 –310Neutral 2.7 1.91 0.0019 0.39 6.4 –540Alkaline 2.7 0.50 0.13 0.81 9.8 –990

11.4 SEMs of fracture surfaces produced in (a) alkaline environment,and (b) neutral environment, showing cleavage-like appearance withcrack-arrest markings.

(a) (b)1 mm 200 nm


there were subtle differences due to differing degrees of corrosion afterfracture and variations in appearance from one area to another on eachspecimen.

11.4 Discussion

The fractographic observations at high magnifications show smaller stepsand more closely spaced CAMs than have been reported previously forT-SCC in stainless steels. However, no strikingly new observations weremade, and any extremely fine detail that may have been produced by thefracture processes was either difficult to distinguish from superficial corrosionor obscured by thin films. The crystallography of cracking could not beestablished as there were no well defined etch pits on fracture surfaces toprovide a guide, and the facets were too small for other techniques to beapplied.

Previous observations of T-SCC in various stainless steels (usually testedin boiling MgCl2 at ~150 ∞C) suggest that cracking most often occurs on{100} planes in <110> directions, but also quite commonly occurs on {110}planes in both <100> and <110> directions [5, 6]. Facets parallel to {111}planes are sometimes observed, especially near crack origins [5, 6]. Otherfracture planes have occasionally been reported [5, 6]. On a fine scale, {100}and {110} facets sometimes exhibit a corrugated appearance due to crackingon alternate {111} planes [5]. However, such micro-facets were not apparentusing high-resolution SEM in the present work.

Cleavage-like fracture surfaces with CAMs have been observed afterenvironmentally assisted cracking in many ductile materials in a variety ofenvironments [5], and numerous mechanisms have been proposed to explainthis fracture morphology. These mechanisms have been based on adsorption,film-formation, localised dissolution, and hydrogen (or vacancy) generationand diffusion ahead of cracks. For stainless steels in H-T water, all thesematerial-environment interactions probably occur and, hence, establishingthe mechanisms of SCC is difficult. Some of the proposed mechanisms of T-SCC are outlined in the following section, with further details available inthe references cited. Whether or not these mechanisms are applicable to H-T SCC of stainless steels, or may be applicable with some modification, arethen discussed.

11.4.1 Outline of proposed mechanisms of T-SCC

Adsorption-induced dislocation-emission (AIDE) [7, 8]

The AIDE mechanism, illustrated in Fig. 11.5, involves weakening ofinteratomic bonds at crack tips due to adsorption of environmentalspecies such as hydrogen so that dislocation emission is facilitated.


Hydrogen atoms within several atomic layers of the crack-tip surface, aswell as hydrogen at the surface, probably facilitate dislocation emission. Theterm ‘adsorbed hydrogen’ used in the rest of the paper includes all thesesites. Crack growth occurs by alternate slip from crack tips, in conjunctionwith void formation around particles or at other sites in the plastic zoneahead of cracks. Diffusion or dislocation transport of hydrogen ahead ofcracks is not essential but may occur – resulting in ‘hydrogen adsorption’and dislocation emission from tips of voids. Macroscopic fracture planesbisect the angle between the active slip planes, and crack fronts lie along theline of intersection of crack planes and slip planes. For example, {100}fracture planes and <110> directions of cracking can be produced whenalternate slip occurs on {111} planes. On a very fine scale, fracture surfaceswould exhibit small dimples.

Hydrogen enhanced decohesion (HEDE) [9]

This mechanism is also based on weakening of metal-metal bonds at or nearcrack tips by high, localised concentrations of hydrogen, so that tensileseparation of atoms (‘decohesion’) occurs in preference to slip. Decohesioncould occur at several locations, viz (i) within a few atomic distances ofcrack tips due to ‘adsorbed hydrogen’, (ii) some distance ahead of cracks atpositions of maximum tensile or hydrostatic stress, and (iii) at particle-matrix interfaces. Some dislocation activity may accompany decohesion,and may increase stresses at decohesion sites, but local strains should be low.HEDE could occur along low-index planes in order to minimise surface andplastic-energy contributions to the fracture energy, as for cleavage in inertenvironments.

Hydrogen-enhanced localised plasticity (HELP) [9]

The HELP mechanism is based on the re-configuration of hydrogenatmospheres around dislocations, which reduces elastic stress fields so that

11.5 Diagram illustrating adsorption-induced dislocation-emission (AIDE)mechanism.

H adsorption

Anodic + Cathodicreactions

Ionicdiffusion

{110} <100>

{100} <110>

Void nuclei

Slipbands Vacancy

cluster

H and vacancy diffusion

Dislocation emission(alternate slip)


repulsive interactions are minimised when dislocations approach obstacles.Since hydrogen concentrations are localised near crack tips due to entry ofhydrogen at crack tips and hydrostatic stresses, deformation should be morelocalised during crack growth than in the absence of hydrogen. However, theprecise mechanisms of crack growth, and how cleavage-like fractures mightbe produced, have not been specified.

Corrosion-enhanced localised plasticity [10]

The following sequence of events to account for cleavage-like fracturesexhibiting fine-scale {111} facets is envisaged: (i) rupture of any oxide filmsat crack tips, (ii) localised dissolution along a {111} plane intersecting thecrack tip, (iii) enhanced plasticity along the {111} plane due to increasedstress, adsorption, or HELP, (iv) piling-up of dislocations along the slipplane against an obstacle such as a Lomer-Cottrell (L-C) lock, (v) cracknucleation and cleavage (perhaps involving HEDE) along the {111} planeback towards the main crack tip, and (vi) repetition of the above sequence ona differently inclined {111} slip plane.

Corrosion-assisted cleavage [11]

This proposed mechanism has similarities to the previous mechanism in thatdissolution is envisaged as occurring along a {111} slip plane to produce a‘dog-leg’ which, along with a dislocation pile-up, increases the stress on a L-C lock at the head of the pile-up so that a cleavage crack nucleates at thelock. However, unlike the previous mechanism, the cleavage crack is thoughtto propagate along a {110} plane, with further dissolution along theaforementioned {111} plane linking up the cleavage cracks.

Film-induced cleavage [1]

As the name implies, the mechanism involves (i) the formation of a brittle,epitaxial film (usually a nanoporous de-alloyed film ≥ 50 nm thick) at cracktips, (ii) rapid fracture of the film, (iii) continuation of cracking into theunderlying substrate by cleavage for up to 10 mm, with dislocation activitysuppressed by extremely high local strain rates, and (iv) crack-arrest due toincreasing extents of dislocation activity.

Selective-corrosion vacancy-creep [12]

High concentrations of vacancies (and di-vacancies) are produced by selectivecorrosion (de-alloying), and it has been proposed that these vacancies diffuseahead of cracks and promote dislocation climb and cross-slip thereby localising


deformation near crack tips. Vacancies could also cluster to form nano-voids, possibly along low-index crystallographic planes [13]. A combinationof vacancy-enhanced creep and vacancy clustering is thought to result incleavage-like cracking, but details of exactly how this might occur have notbeen clarified.

Dissolution mechanisms

Crack growth is considered to occur primarily by dissolution in these models– as opposed to the corrosion-enhanced plasticity and corrosion-assistedcleavage models where dissolution triggers mechanical fracture but does notmake a major contribution to crack advance. ‘Slip-dissolution’ models wereamongst the earliest ones proposed for SCC, and involve repetitive cycles ofdissolution, repassivation, and then rupture of oxide films at crack tips byslip [1]. Slip-dissolution mechanisms are generally considered to be moreapplicable to intergranular SCC rather than T-SCC.

A more recent dissolution model for T-SCC (stress-assisted directed-dissolution) [14] involves high elastic stresses that stretch interatomic bondsso that rates of dissolution are increased by about an order of magnitude –with dissolution directed along low-index crystallographic planes to minimisethe energy of bond breaking as atoms are detached from the lattice. Somedislocation emission from crack tips is considered necessary to open cracksand allow solution access and diffusion of ions away from crack tips. Formaterial-environment systems where hydrogen is generated, it has beensuggested that dissolution rates could also be increased due to bond weakeningby hydrogen [15].

11.4.2 Possible explanations for crack-arrest markings

CAMs are commonly observed on cleavage-like fractures produced byenvironmentally assisted cracking, but it must be emphasised that they arenot always observed, especially at low stress-intensity factors and for certaincrystallographic planes and directions [5]. CAMs can also be explainedregardless of which of the preceeding SCC mechanisms (or combination ofmechanisms) is operative [8]. One general mechanism for the formation ofCAMs involves the formation of ligaments of uncracked material behind themain crack tip. These ligaments act as a restraining force opposing crack-tipopening so that cracks may slow down or stop, and crack-tips probablyrepassivate, at least along part of the crack front. Ligaments could then failby SCC or tearing (producing the observed serrated steps or tear ridges onfracture surfaces), thereby increasing stresses at crack tips so that blunting,film rupture, and continued SCC can occur. The presence of intermittentmicrostructural barriers that inhibit crack growth could be another general


basis for the formation of CAMs. For example, dislocation cell walls orparticular arrangements of shielding and anti-shielding dislocations couldact as barriers to crack growth or, alternatively, produce a sudden avalancheof dislocation egress at crack tips resulting in blunting.

CAMs in some systems could be associated with ‘embrittlement’ of materialahead of cracks, with cracks suddenly advancing when a critical degree ofembrittlement is achieved, and then stopping when they encounter‘unembrittled’ material. ‘Embrittlement’ could be caused by a criticalconcentration of solute hydrogen, or a critical volume fraction of vacancy-induced or plasticity-induced voids ahead of cracks.

Discontinuous cracking could also occur due to varying environmentalconditions at crack tips. For example, localised dissolution could result in avery high concentration of complex ions near sharp crack tips, which mightstifle further dissolution (and other reactions) so that crack-arrest occurred.Diffusion of ions near crack tips into the bulk solution, perhaps aided byplastic blunting at crack tips due to creep, could then re-establish environmentalconditions conducive to further reactions, so that crack growth re-initiated.Alternatively, highly concentrated solutions produced by localised dissolutionmight become embrittling, perhaps with complex ions adsorbing at cracktips resulting in an increment of cracking by AIDE. Cracks could arrestwhen the supply of embrittling species became exhausted.

11.4.3 Applicability of proposed mechanisms to T-SCC instainless steel

The likelihood that hydrogen is generated at crack tips during T-SCCof stainless steels, along with observations that ‘stable’ stainless steelsexhibit cleavage-like cracking when tested in air at 20 ∞C after hydrogencharging [16], lend credibility to an H-based mechanism. Any lack ofcorrelation between the susceptibility of different stainless steels toembrittlement after hydrogen charging and susceptibility to SCC, and minimalembrittlement due to gaseous hydrogen in stable austenitic stainless steels[1], does not preclude an H-based SCC mechanism. These effects couldreflect differences in rate-controlling steps rather than differences inmechanisms.

For {100}<110> fractures, which seem to be observed most commonlyfor SCC of stainless steels, the AIDE mechanism (involving ‘adsorbedhydrogen’) is a strong contender as the mechanism is based, in part, onobservations of cleavage-like fractures with this crystallography incircumstances where only adsorption can occur, e.g. liquid-metal embrittlement(LME) of Al and Ni, and SCC at very high velocities (~10 mm/s) for Alalloys in aqueous environments [7, 8]. Cleavage-like cracking due to adsorption-induced LME has also occasionally been observed along {110} planes in


<100> and <110> directions [17]. Crack growth on {110} planes in the{100} direction could involve alternate-slip since slip planes intersect crackfronts, as occurs for {100} <110> fractures.

For crack growth on {110} planes in <110> directions, slip planes do notintersect crack fronts. In this case, ‘adsorbed’ species could result in decohesion(HEDE), or a variant of AIDE involving alternate-shear at crack tips couldoccur. Alternate-shear has been proposed to account for planar {110}<110>fracture surfaces produced by fatigue of stainless-steel single crystals in airat low growth rates [18]. Crack planes bisect two shear bands in whichmultiple slip occurs on all four {111} slip planes, and AIDE of small segmentsof dislocations intersecting crack tips (that are not atomically sharp) couldperhaps facilitate alternate shear.

For cracks growing by an alternate-slip process, nucleation and growth ofsmall voids ahead of cracks enable a small crack-tip-opening angle to bemaintained, and the presence of small, shallow dimples on cleavage-likefracture surfaces in some systems shows that such a process can occur [7, 8]For cleavage-like fractures in stainless steels, however, there is no evidenceof dimples on fracture surfaces, although their presence cannot be completelyruled out. Dimples can be extremely small and shallow, and could be obscuredby oxide films or obliterated by corrosion after fracture.

Besides void formation, other crack-tip re-sharpening processes, such asintermittent decohesion, fracture of brittle films, or dissolution at crack tips,could occur in conjunction with alternate-slip (AIDE) to produce cleavage-like fractures. For the AIDE mechanism (involving ‘adsorbed hydrogen’),some anodic dissolution must, of course, occur to balance the cathodicproduction of hydrogen. The possibility that dissolution or de-alloying couldfacilitate simultaneous dislocation emission also merits some consideration,especially as hydrogen effects can be discounted for cleavage-like SCCfractures in some materials [1]. Also, if crack tips become passivated, ruptureof oxide films must occur to allow AIDE. Thus, there are a number ofvariants to the originally proposed AIDE mechanism, as discussed in moredetail elsewhere [8], and one or more of them could be applicable to T-SCCin stainless steel.

Other proposed mechanisms are more problematical than the AIDEmechanism and its variants in being able to account for cleavage-like SCC,not only for stainless steels but also more generally. The HELP mechanism,for example, can be discounted on a number of grounds. Firstly, hydrogenatmospheres around dislocations become more dispersed at higher temperatures[9] and HELP would not be expected to occur at 270 ∞C, and would probablynot be significant at ~150 ∞C. Secondly, even in the temperature and strain-rate regimes where HELP could occur, it is difficult to envisage how justlocalising dislocation activity ahead of cracks could produce a change froma non-crystallographic ductile fracture in inert environments to a fracture


along low-index crystallographic planes in specific directions, e.g. {100}<110>[7, 8].

Localised deformation due to increased vacancy concentrations and thepresence of vacancy clusters could well be important, but seems unlikely, byitself, to be able to explain the crystallographic aspects of cracking, as discussedfor HELP. Also, heavily irradiated stainless steel, which should contain ahigh density of vacancy clusters, exhibits ductile fracture in air [19], suggestingthat vacancy clusters per se are not embrittling. However, vacancy clusters,perhaps stabilised by hydrogen, could be one mechanism for nucleating verysmall voids ahead of cracks which, in conjunction with AIDE, could producecleavage-like fractures.

Mechanisms based on the presence of L-C locks that block slip and initiatecleavage ahead of crack tips are questionable for several reasons. Firstly, thepattern of river lines on fracture surfaces suggests that cleavage cracksre-initiate from the existing crack-tip, rather than ahead of the crack [5].Secondly, recent 3-D modelling of dislocation interactions suggest that L-Clocks are not strong obstacles and can be easily ‘unzipped’ [20]. Mechanismsprimarily involving dissolution at crack tips are generally discounted on thebasis that they are not consistent with the detailed fractography, e.g. the veryfine-scale steps that have been shown to be interlocking on opposite fracturesurfaces.

Film-induced cleavage is questionable for SCC in stainless steels becauseany films formed at crack tips may be too thin, or may not have the requisiteproperties, to induce cleavage [1]. It is also unclear, in general, why cleavagein films should continue into an underlying ductile substrate, rather thanblunting immediately as has been observed in some model systems [8].However, observations in other systems do support a film-induced SCCmechanism, and further studies to clarify some of the issues concerning thismechanism are warranted.

11.4.4 Rate-controlling steps for SCC

Possible rate-controlling steps for H-T SCC in stainless steel include(i) transport processes in the environment, (ii) electrochemical-reactionand adsorption kinetics, (iii) kinetics of film-rupture and re-passivation,(iv) diffusion of hydrogen or vacancies ahead of cracks, and (v) crack-tipstrain rates, as is the case for other SCC systems. Previous work for H-TSCC in stainless steel [2, 3] has shown that there is no effect of COD rateson crack-growth rates in neutral environments whereas crack-growth ratesincreased with increasing COD rates in alkaline environments. Thus,crack-tip strain rates do not appear to be rate-controlling in neutralenvironments, but could well be in alkaline environments. In neutralenvironments, the lack of an effect of COD rate is possibly associated


with rate-control by hydrogen adsorption and diffusion kinetics. Thedifferent behaviour in different environments is probably associated withdifferences in hydrogen-ion concentration, and rates of dissolution and re-passivation at crack tips. Possible differences in the extent of crack-branchingcould also affect crack-growth rates. Further work is clearly requiredto understand these effects, and to understand the kinetics of SCC moregenerally.

11.5 References

1. R.C. Newman and A. Mehta, ‘Stress Corrosion Cracking of Austenitic Steels’,Environment Induced Cracking of Metals, R.P. Gangloff and M.B. Ives, eds, NACE-10, 1990, 489–509, and references therein.

2. A. Brozova, J. Burda, L. Papp and K. Pouchman, ‘Nature of Tube Degradation inHorizontal Steam Generators’, Proc. Int. Symp. on Materials Investigation to theResolution of Problems in Pressurised Water Reactors, Fontevraud 5, 2002, 23–27.

3. A. Brozova, ‘Role of Anodic Dissolution and Hydrogen Effects during EnvironmentallyAssisted Cracking of A321 Steel in Concentrated Solutions at 270 ∞C’, HydrogenEffects on Materials Behavior and Corrosion Deformation Interactions, Proc. Of theInt. Conf., ed. N.R. Moody et al., TMS publication, ISBN 0-87339-501-8, 2003,723–731.

4. H.D. Solomon, ‘Transgranular, Granulated and Intergranular Stress Corrosion Crackingin AISI 304 SS’, Corrosion, 1984, 40, 493–506.

5. J.I. Dickson, Li Shiqiong, J.P. Bailon and D. Tromans, ‘The Fractography ofTransgranular SCC in F.C.C Metals: Mechanistic Implications’, Parkins Symposiumon Fundamental Aspects of Stress Corrosion Cracking, S.M. Bruemmer et al., eds,TMS, 1992, 303–322, and references therein.

6. E.I. Meletis and R.F. Hochmann, ‘A Review of the Crystallography of Stress CorrosionCracking’, Corros. Sci., 1986, 26, 63–90, and references therein.

7. S.P. Lynch, ‘Environmentally Assisted Cracking: Overview of Evidence for anAdsorption-Induced Localised-Slip Process’, Acta Metall., 1988, 36, 2639–2661.

8. S.P. Lynch, ‘A Commentary on Mechanisms of Environmentally Assisted Cracking’,Corrosion-Deformation Interactions (CDI ’96), T. Magnin, ed., Inst. of Metals,1997, 206–219, and references therein.

9. H.K. Birnbaum, ‘Mechanisms of Hydrogen Related Fracture of Metals’ HydrogenEffects on Material Behavior, N.R. Moody and A.W. Thompson, eds, TMS, 1990,639–658, and references therein.

10. T. Magnin, A. Chambreuil and B. Bayle, ‘The Corrosion Enhanced Plasticity Modelfor Stress Corrosion Cracking in Ductile fcc Metals’, Acta Mater., 1996, 44, 1457–1470.

11. W.F. Flanagen and B.D. Lichter, ‘A Mechanism for Transgranular Stress CorrosionCracking’, Int. J. Fracture, 1996, 79, 121–135.

12. H. Leinonen and H. Hanninen, ‘Prediction of Stress Corrosion Cracking Susceptibilityof Austenitic Stainless Steels in 50% CaCl2 Solution’, Corrosion-DeformationInteractions (CDI ’96), T. Magnin, ed., Inst. of Metals, 1997, 131–139.

13. D.A. Jones, ‘A Unified Mechanism of Stress Corrosion and Corrosion FatigueCracking’, Metall. Trans. A, 1985, 16A, 1133–1141.


14. B.D Lichter, H. Lu and W.F. Flanagan, ‘Strain-Enhanced Dissolution: A Model forTransgranular Stress Corrosion Cracking’, 2nd Int. Conf. on Environment SensitiveCracking and Corrosion Damage, M. Matsumura et al. eds, Nishiki Printing Ltd.,Hiroshima, Japan, 2001, 271–278.

15. L. Qiao and X. Mao, ‘Thermodynamic Analysis on the Role of Hydrogen in AnodicStress Corrosion Cracking’, Acta metall. mater. 1995, 43, 4001–4006.

16. H. Hanninen and T. Hakkarainen, ‘Fractographic Characteristics of a Hydrogen-Charged AISI 316 Type Austenitic Stainless Steel’, Metall. Trans. A, 1979, 10A,1196–1199.

17. S.P. Lynch, Unpublished work, 2002.18. R. Rieux, J. Driver, and J. Rieu, ‘Fatigue Crack Propagation in Austenitic and

Ferritic Stainless Steel Single Crystals’, Acta Metall., 1979, 27, 145–153.19. M.R. Louthan, Jr., Savannah River Company, Aiken, SC, USA, private communication,

2002.20. V. Bulatov, F. Abraham, L. Kubin, B. Devincre and S. Yip, ‘Dislocation Junction and

Crystal Plasticity: Linking Atomistic and Mesoscale Simulations’, Nature, 1998,391, 669–672.

Part III

Stress corrosion cracking: propagation

163

165

12.1 Introduction

Ageing of pressure boundary components is one of the main factors controllingthe lifetime of nuclear power plants. EAC under certain circumstances canbe one of the major ageing mechanisms of LAS in high-temperature water.The project ‘Crack Growth Behaviour of Low-Alloy Steels for PressureBoundary Components under Transient Light Water Reactor OperatingConditions’ (CASTOC), was performed within the 5th EC frameworkprogramme and addressed the problem of EAC of Western and Eastern typesteels used for pressure boundary components [1, 2].

The objective of the CASTOC project was to screen the EAC behaviourof low-alloy reactor pressure vessel (RPV) steels in high-temperature waterduring transients of load and water chemistry that may occur during start-upand shut-down, steady-state operation and load following mode of commerciallyoperating LWRs. This is in contrast to the worldwide activities in the past,which focused mainly on either cyclic loading or static loading and steady-state operating conditions. The main focus of the project was directed to theinteraction between static and cyclic loading which was realised, e.g., bylow frequency corrosion fatigue (LFCF) phases followed by static load or byperiodical partial unloading (PPU) with different rise and hold times. Inconjunction with the different load spectra, the effect of transients in water

12Crack growth behaviour of low-alloy steelsfor pressure boundary components under

transient light water reactor operatingconditions – CASTOC, Part 1:

BWR/NWC Conditions

S. R I T T E R and H. P. S E I F E R T, Paul Scherrer Institute(PSI), Switzerland, B. D E V R I E N T and A. R O T H,Framatome ANP GmbH, Germany, U. E H R N S T É N,VTT Industrial Systems, Finland, M. E R N E S T O V Á

and M. Ž A M B O C H, Nuclear Research Institute (NRI),Czech Republic, J. F Ö H L and T. WEISSENBERG,

Staatliche Materialprüfungsanstalt (MPA), Germany andD. G O M É Z - B R I C E Ñ O and J. L A P E Ñ A,

Centro de Investigaciones Energéticas Medioambientales yTecnológicas (CIEMAT), Spain


chemistry was investigated. A more detailed description of the project isavailable in [1] and [2].

The results of the project were assessed with respect to quality, reliabilityand their application to plant concerns and possible code implementation.This comprised a comparison of the data from the CASTOC project withdata from literature and codes. With regard to the specific situation in Europe,the materials and water environment conditions were chosen to address boththe concerns of BWRs and VVERs.

In the current paper some important crack growth results obtained insimulated BWR/NWC environment are presented. The results of the testsperformed under simulated VVER conditions are summarised in [3]. For adetailed description of all results see [4–7].

12.2 Experimental procedures

The tests performed within the CASTOC project comply with the currentstate-of-the-art knowledge of science and technology in laboratory testing ofEAC processes. The investigated materials represent nuclear grade and lowerbound materials. With regard to the selected environmental conditions,enveloping parameters were applied.

12.2.1 Materials

The following LAS, used for the pressure retaining components of the primarycoolant of LWRs, were investigated under BWR/NWC conditions (Table12.1).

∑ Material A, base material: seamless forged ring of German type20MnMoNi5-5 RPV steel, equivalent to ASME SA 508 Grade 3 Cl. 1.

∑ Material A, weld metal: S 3 NiMo 1/OP 41 TT (wire/flux), fabricated asweld using narrow gap welding according to nuclear grade quality withoptimised wire and flux relevant to welding techniques, which wereused for RPVs of Western plants.

∑ Material B, base material: seamless forged vessel shell of German type22NiMoCr3-7 RPV steel, equivalent to ASME SA 508 Grade 2 Cl. 1,taken from a RPV fabricated for a nuclear power plant in Germany(‘Biblis C’), which has not been built, however.

All base materials have a fine-grained bainitic microstructure. In case ofmaterial A, a high variation in the local sulphur content was observed rangingfrom 0.003 to 0.053 wt.%. Furthermore, portions of intergranular fracturewere found on fracture surfaces produced under cyclic load in air givingindications for grain boundary segregations (P, Mo and Mn), which could beidentified by Auger Electron Spectroscopy.

Crack grow

th behaviour of low-alloy steels

167

Table 12.1 Chemical composition in wt.% and mechanical properties of the investigated materials

Material Chemical composition [wt.%] sYS (RT) sUTS (RT) sYS (288 ∞C)

C Mn Cr Ni P S V [MPa] [MPa] [MPa]

A (base) 0.25 1.54 0.18 0.62 0.014 0.015 0.024 512 663 462

A (weld) 0.07 1.15 0.10 1.04 0.014 0.005 0.010 496 572 440

B (base) 0.22 0.91 0.42 0.88 0.008 0.007 0.007 467 605 400

base = base material RT = room temperature sUTS = ultimate tensile strengthweld = weld metal sYS = yield strength


The specimens were predominantly taken from the 1/4 T to the 3/4 Tposition (T = thickness of the plate/forging) which corresponds to requirementsaccording to the acceptance testing procedure. One fracture mechanics specimenof material A was investigated with the crack located in the HAZ of the jointweld. The hardness determined on a metallographically prepared sectionnear the location where the specimen was taken reached values of about 340HV 0.1.

12.2.2 Experimental facilities

The laboratory tests of the six CASTOC partners were performed in autoclaveswith integrated loading systems attached to sophisticated refreshing or once-through high-temperature water loops, which enable precise adjustment andcontrol of the water chemistry and loading conditions. During the experimentsall important data were recorded continuously. 25 mm thick compact tensionspecimens (C(T)25) were used for most of the tests. By means of the reverseddirect current potential drop (DCPD) technique, on-line crack lengthmeasurements were performed with high resolution in crack length. Thedetection limit of the DCPD technique was of the order of 2 to 10 mm. Thiscorresponded to a detection limit of crack growth rates during a 300 h testperiod of about 10–12 to 10–11 m/s (60 to 300 mm/a).

12.2.3 Loading conditions

Tests with cyclic loading (LFCF), PPU cycles and under pure static loadwere conducted, whereas a cyclic load phase always preceded the static loadphases to generate an actively growing crack. In this way long incubationtimes for the development of stable crack growth could be avoided. The testsunder cyclic load were usually performed with a high number of cycles toachieve sufficient stable crack growth conditions and sufficient crack advanceto consider this data as reliable. Under static load, the duration of the testswas extended up to 500 h.

12.2.4 Environment

BWR conditions were mostly simulated with high purity water at a temperatureof 288 ∞C with an oxygen content of 400 mg/kg, representing steady-statepower operation in a conservative way. An oxygen content of 8000 mg/kgwas applied additionally to investigate the effect of oxygen during start-upphases in the plant and at the same time to simulate a realistic electrochemicalcorrosion potential (ECP). If not stated otherwise, the term ‘BWR water’ isused for water of high purity with an electrical conductivity of k < 0.2 mS/cm in the outlet water. With very few exceptions the CASTOC tests were

Crack growth behaviour of low-alloy steels 169

performed under quasi-stagnant flow conditions with a low refreshing rate togenerate conservative data with respect to plant conditions.

To simulate transient BWR power operating conditions, tests at temperaturesof 180 or 240 ∞C were performed as well as sulphate or chloride in terms ofNa2SO4/H2SO4 or NaCl, respectively, being added to the high purity waterin some experiments. For BWR/NWC power operation the main parametersof the cooling water are specified in the EPRI Water Chemistry Guidelines[8] or in the German ‘VGB-Richtlinie’ [8]. Accordingly, environmentalparameters were chosen which clearly exceeded the normal operation valuesfor stationary BWR power operation. In some experiments single valueseven exceeded the EPRI Action Level 3 limits, taking into account that thosewater chemistry conditions would not allow to continue operation of theplant but are necessary to evaluate the effect of worst case conditions. Therefore,the crack growth results of these experimental investigations are highlyconservative with respect to the water chemistry parameters of normal plantoperation.


12.3.1 Inter-laboratory comparison test

The project was designed in a way that different partners contributed resultspartly to the same objective but with different parameter sets. To assure thatthe individual results can be comprised to describe the complex interactionsof the different parameters, an inter-laboratory comparison test was performedto demonstrate the reproducibility and, at the same time, to generate datawhich describe the crack growth behaviour of material A in high purityBWR water [4, 9].

The tests were carried out under nominally identical conditions understatic loading after initial cyclic loading. It was confirmed that all tests wereperformed under proper control of all important ‘external’ parameters, e.g.load and environment, and that sufficient crack advance mostly was achievedin the cyclic phase to obtain reliable data [9].

After changing from cyclic to static load, immediate cessation of crackgrowth was observed in all labs at the applied stress intensity factors of upto almost 60 MPa÷m. An example is presented in Fig. 12.1. This confirmsthat material A, tested under these conditions, is not susceptible to EACunder static load (stress corrosion cracking, SCC).

The cycle-based crack growth rate data da/dN measured during cyclicloading revealed a range of up to two orders of magnitude in the simulatedBWR water environment under nominally identical conditions (Figs 12.2and 12.3). This range is higher than is generally expected for similar materialsand loading parameters in this test environment. However, a range of up to


Time (h)50040030020010020100

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

Cra

ck a

dva

nce

(m

m)

BWR water, 288 ∞C400 mg/kg O2Material A

Fatigue phase Constant load phase

Specimen A11

Specimen A10

Note the different scale before and after thex-axis break!

12.1 Crack advance according to the DCPD signal versus testing time,showing cessation of crack growth at static load immediately aftercyclic loading in the autoclave (oxygenated high purity BWR water at288 ∞C).

Da > 0.2 mmDa > 0.2 mm

ASME XI ‘Air’ASME XI ‘Wet’

( (

DKl (MPa ÷m)100101

R = 0.1

0.65 £ R < 1

BWR water, 288 ∞C400 mg/kg O2

Material Af = 8.3 ¥ 10–3 s–1

R = 0.7

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)

102

101

100

10–1

10–2

10–3

12.2 Range of crack growth rates of material A at a frequency of8.3 ¥ 10–3 s–1 in oxygenated high purity BWR water at 288 ∞C.


one order of magnitude is typically observed in this loading frequency rangefor RPV steels based on well-behaved crack growth of homogeneous materialin a similar environment [9, 12] and all data obtained in the present study forlarger crack extensions (Da ≥ 0.2 mm) also falls within this with the exceptionof one single data point. This ‘outlier’ was related to a highly uneven crackfront and a repetition of the test in the corresponding lab revealed a cycle-based crack growth rate da/dN in the expected range. Most of the dataobtained in the CASTOC project with low crack growth rates were relatedeither to load drops because of electric power interruptions, highly unevencrack front because of crack pinning, or short crack advances (Da < 0.2 mm)and were believed to be non-representative for this material.

EAC growth in simulated BWR environment is affected by the stressintensity, type of loading, the ECP related to the dissolved oxygen concentration,the concentration of specific anionic impurities ( SO4

2– , Cl–, H2S, S2–, HS–,etc.) reflected by the conductivity of the water, the flow rate passing thecrack mouth, the steel sulphur content, and the morphology/spatial distributionof the MnS-inclusions [12]. Variations of these parameters can result indifferentcrack growth rates. Apart from material properties, all of these factors werefairly similar in all labs. A relevant part of the large data range may thereforebe related to the inhomogeneous microstructure in this material: e.g., the

Da > 0.2 mmDa > 0.2 mm


0.65 £ R < 1 R = 0.1


Material Af = 8.3 ¥ 10–4 s–1

R = 0.7

DKl (MPa÷m)100101

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)

102

101

100

10–1

10–2

10–3

12.3 Range of crack growth rates of material A at a frequency of8.3 ¥ 10–4 s–1 in oxygenated high purity BWR water at 288 ∞C.


investigated specimens showed a strong variation in the amount of MnSinclusions and local islands of intergranular fracture, indicating localsegregation. Additionally the nature of the EAC cracking process itself canlead to local crack pinning, crack cessation/arrest and re-nucleation problems,and thus to lower crack growth rates than expected. EAC is a deterministicprocess, but singular results are influenced by the previously mentionedparameters, thus adding a probabilistic factor. Furthermore, crack growthdata, which were derived from test periods with not long enough crackadvance (Da < 0.2 mm) with regard to the microstrucutre (e.g., grain size)and to the resolution of fractographic crack length determination and DCPDtechnique for crack length measurement, in particular in the lower range ofda/dN, are associated with a higher uncertainty and thus are less reliable.This has to be considered with regard to practical applications. Several datapoints, from several specimens representing several locations in a LAS materialare needed before representative and conservative crack growth rate valuescan be derived with satisfactory confidence. From the results of the presentwork it was concluded, that a large enough (Da ≥ 0.2 mm) crack advance isneeded following any type of anomalous testing event before appropriatecyclic crack growth rate data are measured. If not further specified, thefollowing cyclic crack growth rate diagrams therefore only contain datapoints with crack advances ≥ 0.2 mm.

The data were compared with the prediction line as presented in theASME Boiler and Pressure Vessel Code, Section XI, Appendix A [10]. Thiscomparison clearly reveals that a significant number of data, in particularthose which were derived from test periods with sufficient crack advance,exceed the ASME prediction line. It can further be concluded that the lowercycling frequency (8.3 ¥ 10–4 s–1, Fig. 12.3) tends to cause higher crackgrowth rates, than the higher frequency (8.3 ¥ 10–3 s–1, Fig. 12.2). This isconsistent with the general experience that EAC is a strong time dependentprocess.

12.3.2 Results from cyclic loading and load transients

Effect of material

Material A was investigated extensively and showed consistently, that mostof the da/dN data exceed the ASME XI prediction curve in high purity water(Figs 12.2 and 12.3). Material B, representing optimised nuclear grade, showedsignificantly lower susceptibility to EAC and in most cases it was difficult toinitiate fast EAC crack growth under cyclic load. Therefore, the range of da/dN data were generally almost one order of magnitude lower than that ofmaterial A (Fig. 12.4), but if once fast crack growth was initiated, the crackgrowth rates of material B, e.g. during slow load transients, were in the sameranges as those of material A [6]. The lower susceptibility to EAC appeared


also in tests with PPU and during the applied chloride transient (comparepages 178–9) and was attributed to the more homogeneous microstructure,lower sulphur content and lower susceptibility to dynamic strain ageing(DSA) in this material.

The weld metal and the HAZ of the joint weld of material A showedsimilar behaviour as the base material A (Fig. 12.5), although due to thelimited number of tests and too short crack advances (Da < 0.2 mm) in somecases no clear conclusion for the cyclic crack growth behaviour could beestablished in the framework of the CASTOC project.

Effect of oxygen content

The effect of dissolved oxygen content was investigated on material A inhigh purity BWR water. No significant effect of oxygen on EAC was observedat concentrations between 400 and 8000 mg/kg (Fig. 12.6), which is consistentwith the only small differences in the measured ECP values.

Effect of temperature

In BWR water environment experiments were carried out with material Aand B at temperatures of 180 and 240 ∞C in addition to those at 288 ∞C. An

Material AMaterial BMaterial B withDa < 0.2 mm

f = 2.2¥10–4s–1

R = 0.2

f = 2.5¥10–3s–1

R = 0.8

ASME XI ‘Wet’, R = 0.2ASME XI ‘Wet’, R = 0.8


DKl (MPa÷m)100101

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)

102

101

100

10–1

10–2

10–3

12.4 Comparison of crack growth rates of material A and B under lowfrequency fatigue in oxygenated high purity BWR water at 240 ∞C.


Base material AHAZWeld metalWeld metal withDa < 0.2 mm

DKl (MPa÷m)100101

R = 0.1

0.65 £ R < 1

10–3

10–2

10–1

100

101

102

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)



f = 8.3 ¥ 10–3 s–1

R = 0.7

12.5 Crack growth behaviour of weld metal and HAZ compared todata for base material A under low frequency fatigue in oxygenatedhigh purity BWR water at 288 ∞C.

400 mg/kg O2

8000 mg/kg O2

DKl (MPa÷m)100101

R = 0.1

0.65 £ R < 1


BWR water, 288 ∞C

Material Af = 8.3 ¥ 10–4 s–1

R = 0.7 (0.8)

10–3

10–2

10–1

100

101

102

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)

12.6 Effect of oxygen on crack growth rate under low frequencyfatigue in BWR water at 288 ∞C.


example is presented in Fig. 12.7 for material A, whereas material B showeda similar behaviour. In this temperature range, no clear temperature trend onthe crack growth behaviour was observed due to the wide range of data, thelimited number of tests, and too short crack advances (Da < 0.2 mm) in somecases. However, literature data in BWR environment reveal an increase incrack growth rates with increasing temperature, sometimes with a maximumat intermediate temperatures (200 to 250 ∞C) in case of materials whichshow distinct DSA effects [11, 12].

Effect of load transients and frequency

By means of the DCPD measurement technique detailed insight into thecrack growth behaviour as a function of time was obtained. Tests performedwith PPU revealed very clearly that crack advance in high purity BWR wateroccurs only in the phases of rising load (e.g., Fig. 12.8).

Figure 12.9 shows an example of PPU tests in which the hold time wasvaried from 0 to 105 s (0 to 28 h). As long as EAC under static load duringthe hold time period at maximum load does not occur, the low frequencycorrosion fatigue (LFCF) crack growth rate da/dN depends only on the numberof loading events and thus the curve in Fig. 12.9 must result in a horizontalline. Differences appear in the crack growth rate due to different applied loadratios (R = 0.2 and 0.8) and different rise times to maximum load (100, 1000,and 4000 s). With the longer rise times (1000/4000 s), the load increase isslower, causing a lower strain rate in the crack tip region and hence thecontribution of EAC is more pronounced and results in a higher crack growthrate da/dN. This behaviour is consistent with the basic mechanism of strain-induced corrosion cracking (SICC).

In Fig. 12.10 cycle-based crack growth rates of material A are plottedversus the loading frequency and compared to the corresponding ASME XIprediction curves. The crack advance per fatigue cycle da/dN increases withdecreasing frequency, whereas sustained, stationary LFCF crack growth wasobserved down to very low frequencies of 10–5 s–1. The ASME XI predictioncurves are significantly exceeded for all investigated loading frequenciesunder low-flow and oxidising conditions at 240 ∞C.

Effect of sulphate

The effect of sulphate in BWR water, added as Na2SO4 and H2SO4 respectively,was predominantly investigated on material A. Although the sulphate contentwas realised up to values far beyond the onset of Action Level 3 of the EPRIWater Chemistry Guidelines according to the sulphate content of up to 1400mg/kg, no enhanced crack growth rates were observed (Fig. 12.11). This is ingood agreement with model predictions and most investigations under BWR/


288 ∞C240 ∞C180 ∞C180 ∞C withDa < 0.2 mmR = 0.1

0.65 £ R < 1

DKl (MPa÷m)100101


BWR water400 mg/kg O2

Material Af = 8.3 ¥ 10–4 s–1

R = 0.7

10–3

10–2

10–1

100

101

102

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)

12.7 Effect of temperature on crack growth rate under low frequencyfatigue in oxygenated high-purity BWR water.

32

30

28

26

24

22

Load

(kN

)

Load

da/dt =1.22¥10–9m/s

Material ABWR water, 288 ∞C, 8000 mg/kg O2

Time (h)390 400 410 420 430 440

24.0

23.9

23.8

23.7

23.6

23.5

Cra

ck le

ng

th (

mm

)

DCPD

12.8 Effect of load transients (PPU) on the crack advance during theload transient and during the hold time period; material A, highpurity BWR water at 288 ∞C.


R = 0.8, DtR = 1000 sR = 0.2, DtR = 4000 sR = 0.7, DtR = 100 s

DtH = 0 h

DtH DtR

Hold time at maximum constant load DtH (h)1010.10.01

t

P

PPU

Material A, BWR water, 240 ∞C, 400 mg/kg O2

DtR = 100/1000/4000s, variationof DtR

61.560.158.0

60.458.8

57.5

59.859.558.255.0

57.0

56.0

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)

10–2

10–1

100

101

102

103

12.9 Effect of hold time at maximum load during PPU on crackadvance per reloading event; numbers in the diagram indicate themaximum KI during the hold time.

R = 0.2R = 0.8

Material A, BWR water, 240 ∞C, 400 mg/kg O2

ASME XI ‘Wet’R = 0.2, DK = 43.4 MPa÷m

ASME XI ‘Wet’R = 0.8, DK = 12.5 MPa÷m

100

10

1

0.1

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)

Frequency (s–1)10–6 10–5 10–4 10–3 10–2 10–1 100 101

12.10 Effect of loading frequency on crack growth rate in oxygenatedhigh purity BWR water at 240 ∞C.


NWC conditions known from literature [12]. In some few other experimentsunder BWR/NWC conditions, an accelerating effect of sulphate was observed,in particular in LFCF tests with low-sulphur steels at lower corrosion potentials,where crack growth rates in high purity water were close to those measuredin air [12]. In the CASTOC experiments, however, the effect of sulphateaddition was shaded because the investigated material has relatively highsulphur content, so that the dissolved manganese sulphides intersected bythe plane of the growing crack strongly control the local water chemistry.The ranges of crack growth rate data without and with additional sulphateoverlap completely and there is no tendency that crack growth datacorresponding to EPRI Action Level 3 are higher than those correspondingto EPRI Action Levels 1 or 2.

Effect of chloride

To screen the effect of chloride on the crack growth behaviour of LAS a testwith PPU with a hold time of 12 h at maximum load was performed introducing

High purity waterSulphateSulphate withDa < 0.2 mm

R = 0.1

0.65 £ R £ 1

DKl (MPa÷m)100101


Material Af = 8.3 ¥ 10–4 s–1

R = 0.7


10–3

10–2

10–1

100

101

102

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)

12.11 Effect of sulphate content in BWR water at 288 ∞C on the crackgrowth rate of material A under low-frequency fatigue, sulphatecontent up to 150 mg/kg.


a chloride transient of about 50 mg/kg for a time period of 40 h. Accordingto the chloride content, this water condition corresponds to Action Level 2 ofthe EPRI Water Chemistry Guidelines, whereas the electrical conductivitywhich is monitored continuously in the plant, resulted in a value of about0.27 mS/cm, which is still below the EPRI Action Level 1 limit.

Starting in BWR water of high purity but with enhanced oxygen contentof about 8000 mg/kg, crack advance was observed typical for PPU tests, i.e.,crack growth occurred only in phases of rising load (compare page 175and Fig. 12.8). After an incubation time of about 3 h, onset of fast crackgrowth under static load (SCC) occurred in the specimen of material A (Fig.12.12). The specimen of material B, which was in daisy chain with thespecimen of material A in the same experiment, exhibited retarded crackinitiation only after 32 h, i.e. after two reloading events in chloride containingwater.

The detailed analysis of the DCPD signal showed that the chloride transientaffected both, the crack growth behaviour under static load da/dt and thecrack growth rate under cyclic load da/dN based on the number of reloadingevents (Fig. 12.13). The crack advance Da during reloading, however, wasonly slightly increased. This indicates that the mechanism of SICC which isacting during the reloading phase is not much affected by the presence ofchlorides. The high value for the cycle-based crack growth rate da/dN hasmainly to be attributed to the mechanism of SCC which occurs under staticload at the level of the maximum stress intensity factor KImax.

After returning to high purity water, crack growth in the phase of constantstatic load was still observed, however, with strongly decreasing tendencyover two more load cycles. This observation indicates that a long-term effect(‘memory effect’) of a chloride transient cannot be fully excluded. Similarbehaviour was observed for material B but as already mentioned with retardedcrack initiation, lower crack growth rate under static load and less crackadvances during reloading.

12.3.3 Results from static loading

Based on well-qualified data for simulated BWR environment [13, 14],disposition lines for SCC crack growth in LAS during BWR power operationwere proposed by an international group of experts, working within theframework of the EPRI BWRVIP Project, and accepted by the US NuclearRegulatory Commission as an interim position [13]. The BWRVIP-60 SCCDisposition Line (DL) 1 applies to crack growth in LAS under static loadingand transient-free, stationary BWR/NWC or hydrogen water chemistry poweroperation conditions, whereas the BWRVIP-60 SCC DL 2 may be used forestimating SCC crack growth during and 100 h after transients in waterchemistry (> EPRI Action Level 1 limit) or load transients not covered by


fatigue evaluation procedures. For comparison purposes crack growth ratedata of material A obtained from tests in high purity BWR water underconstant load are displayed together with these DLs in Fig. 12.14. For stressintensity factors in the range of up to about 60 MPa÷m crack growth was not

50 mg/kg Cl–< 0.4 mg/kg Cl–

Onset of fast EACunder static load

DCPD

Material ABWR water, 288 ∞C, 8000 mg/kg O2

32

30

28

26

24

22

Load

(kN

)

490 495485480475470Time (h)

24.4

24.3

24.2

24.1

24.0

23.9

Cra

ck le

ng

th (

mm

)Load

12.12 Crack initiation in a C(T)25 specimen of material A during thephase of static load due to a chloride transient (50 mg/kg) inoxygenated (8000 mg/kg O2) BWR water at 288 ∞C.

High purityphase 3

50 mg/kg chloridephase 2

High purityphase 1

da/d

N

da/d

t

da/d

N

da/d

N

da/d

t = 0

da/dtDecreasing tendency

with time

Material ABWR water, 288 ∞C8000 mg/kg O2Kl = 55 to 62 MPa÷m

3¥10–8

2¥10–8

1¥10–8

0

Cra

ck g

row

th r

ate

da/d

t (m

/s)

Cra

ck g

row

th r

ate

da/d

N (mm

/cyc

le)

2000

1500

1000

500

0

Contributionof SCC

12.13 Crack growth behaviour (da/dN and da/dt) of material A before,during and after a chloride transient in oxygenated BWR water at288 ∞C.


observed in any of the C(T)25 specimens, which were tested in high purityBWR water. These results in which the crack growth rate was below thedetection limit of the DCPD method of 10–12 to 10–11 m/s (60 to 300 mm/a)are displayed schematically in Fig. 12.14. When the stress intensity factorwas increased far beyond the validity limits for linear elastic fracture mode(LEFM) continuous crack growth was observed even in high purity BWRwater obviously due to yielding processes in the crack tip region. Althoughthis plastic deformation occurs under constant static load, the EAC processmay mechanistically be attributed to SICC. However, most of the data fallbelow the DLs. These data in general are not relevant for application tothick-walled pressure boundary components, which are loaded in the LEFMregime during operation, as e.g. the RPV, because the LEFM validity criteriaare violated. For thinwalled components, e.g. pipes, a careful assessment ofLEFM applicability has to be performed depending on the load and flawsize.

The most important result from the investigations under static load wasthe reproducible observation of immediate crack cessation of growing cracksafter changing from cyclic loading to static load for all investigated basematerials in oxidising high purity BWR water at stress intensity factorsbelow 60 MPa÷m (e.g., Fig. 12.1).

There were only few exceptions in which continuous crack advance wasdetected under static load. These are described in the following sections.

BWRMP-60 SCC DL 1 (stationary power operation)BWRMP-60 SCC DL 2 (during and 100 h after transients)

‘High-Sulphur line’

‘Low-Sulphur line’

Stress intensity factor Kl (MPa÷m)90 10080706050403020

LEFM range

all other dataunder constant loadbelow detection limit< 10–12 to 10–11 m/s

BWR water, 288 ∞CMaterial A

10–11

10–10

10–9

10–8

10–7

10–6

Cra

ck g

row

th r

ate

da/d

t (m

/s)

12.14 Crack growth behaviour of material A under constant staticload in high purity BWR water at 288 ∞C, range of plane-strain LEFMvalidity indicated.


Effect of material

As stated above, none of the investigated base materials showed anysusceptibility to EAC under constant static load in the relevant stress intensityrange and normal BWR water chemistry. This behaviour was also confirmedfor the weld metal of the joint weld in material A.

A different behaviour was observed in a test with a specimen from theHAZ of the joint weld in material A. Even at a stress intensity factor of KI

= 47 MPa÷m, using a C(T)25 specimen, sustained crack growth occurredwith a crack growth rate beyond the DL 2 (Fig. 12.15). Based on hardnessmeasurements at room temperature in the base material (235 HV) and theHAZ (340 HV) and also, based on general experience, it may be concludedthat the limit for plane-strain LEFM even at elevated temperature is significantlyhigher for the HAZ as compared to the base material. Therefore, the appliedstress intensity factor in this experiment is still supposed to fulfil the LEFMcriteria.

Effect of environment

As already stated (page 175), additional high sulphate content in BWR waterdid not enhance the crack growth rate under cyclic load. The same is true forstatic load. In the relevant range of stress intensity, crack cessation occurredin all cases even at a sulphate concentration corresponding to Action Level 3of the EPRI Water Chemistry Guidelines.

12.15 Summary of conditions under which continuous crack growthwas observed under static load in oxygenated high purity BWR andBWR water containing 50 mg/kg chloride.

HAZ, high purity waterMaterial A, chloride transientMaterial B, chloride transient

‘High-Sulphur line’

‘Low-Sulphur line’

S = start of testE = end of test

ES

ES

3020 40 50 60 70 80 90 100Stress intensity factor Kl (MPa÷m)

BWRVIP-60 SCC DL 1 (stationary power operation)

BWRVIP-60 SCC DL 2 (during and 100 h after transients)

BWR water, 288 ∞C

10–11

10–10

10–9

10–8

10–7

10–6

Cra

ck g

row

th r

ate

da/d

t (m

/s)


Clear evidence of crack growth under static load was obtained from testswith materials A and B when introducing a chloride transient. According tothe chloride content (50 mg/kg), the conditions corresponded to Action Level 2of the EPRI Water Chemistry Guidelines, whereas the electrical conductivitywas still in the range below Action Level 1. Crack initiation occurred duringthe phase of static load after a relatively short incubation time of 3 h and32 h respectively (compare pages 178–9). The crack growth rate da/dtsignificantly exceeds the DL 2 (Fig. 12.15). For material A a crack growthrate of 1.8 ¥ 10–8 m/s (570 mm/a) was determined at a stress intensity factorof about 55 MPa÷m. Material B showed a longer incubation time and lowercrack growth rate under static load of 2.5 ¥ 10–9 m/s (80 mm/a). Details arealready shown in pages 178–9 and Figs 12.12 and 12.13. After returning tohigh purity water, a decreasing tendency of the crack growth rate with somedelay was observed indicating a kind of ‘memory effect’ due to the residencetime of the chloride ions in the crevice.


The CASTOC project has given more insight into both the phenomenologyand the acting mechanisms on corrosion cracking in BWR water environmentand the effect of transients. With regard to the application of the results fromthe CASTOC project for the assessment of components in LWRs, the followingaspects should be considered:

∑ Low-alloy steel base materials for RPV application revealed resistanceto SCC crack growth under constant static load up to stress intensityfactors of about 60 MPa÷m in BWR/NWC environment.

∑ Under certain environmental and material conditions, however,experimental results from tests performed under static load give reasonfor a more careful consideration of the assessment of components. Thescreening experiments of this project revealed crack growth under constantload for the following conditions: HAZ of the joint weld of material A,and materials A and B during a water chemistry transient with 50 mg/kgchloride.

∑ The proposed BWRVIP-60 SCC Disposition Line 1 [13] for crack growthunder steady-state conditions was essentially confirmed for base materials.With respect to the Disposition Line 2 for transients in load and waterchemistry, however, further consideration is recommended based on theresults of this project.

∑ The project has revealed the general trend that the existing predictioncurve presented in the ASME Boiler and Pressure Vessel Code, SectionXI, Appendix A [10] for da/dN assessment of existing flaws in low-alloysteels may not be conservative under some specific conditions, e.g. low


loading frequencies, since the observed crack growth rates at lowfrequency cyclic loading significantly exceeded the maximum rates ofthe Code.


The financial support by the 5th Framework Programme of the EuropeanCommission under Contract No. FIKS-CT-2000-00048 and the Swiss FederalOffice for Education and Science (BBW) is gratefully acknowledged. Thevaluable contributions of all partners to data generation and discussion of theresults were the basis for a successful performance of the project.

12.6 References

1. J. Föhl, U. Ehrnstén, M. Ernestová, D. Gómez-Briceño, J. Lapeña, S. Ritter, A. Roth,B. Devrient, H.P. Seifert, T. Weissenberg, M. Žamboch, ‘Crack Growth Behaviourof Low Alloy Steel for Pressure Boundary Components under Transient Light WaterReactor Operating Conditions – CASTOC’, FISA Conf. on EU Research in ReactorSafety, Luxembourg, Nov. 12–14, 2001.

2. J. Föhl, U. Ehrnstén, M. Ernestová, D. Gómez-Briceño, J. Lapeña, S. Ritter, A. Roth,B. Devrient, H.P. Seifert, T. Weissenberg, M. Žamboch, ‘Crack Growth Behaviourof Low-Alloy Steels for Pressure Boundary Components under Transient Light WaterReactor Operating Conditions – CASTOC’, FISA Conf. on EU Research in ReactorSafety, Luxembourg, Nov. 10–12, 2003.

3. M. Ernestová, M. Žamboch, J. Föhl, U. Ehrnstén, D. Gómez-Briceño, J. Lapeña, S.Ritter, A. Roth, B. Devrient, H.P. Seifert, T. Weissenberg, ‘Crack Growth Behaviourof Low-Alloy Steels for Pressure Boundary Components under Transient Light WaterReactor Operating Conditions – CASTOC, Part II: VVER Conditions’, EUROCORR2004, Paper No. 241, Nice, France, Sep. 12–16, 2004.

4. U. Ehrnstén, ‘Crack Growth Behaviour of Low Alloy Steels for Pressure BoundaryComponents Under Transient Light Water Reactor Operating Conditions’, EU ProjectCASTOC, Technical Report WP1, 2002.

5. J. Lapeña, D. Gómez-Briceño, ‘Crack Growth Behaviour of Low Alloy Steels forPressure Boundary Components Under Transient Light Water Reactor OperatingConditions’, EU Project CASTOC, Technical Report WP2, 2003.

6. M. Žamboch, J. Föhl, ‘Crack Growth Behaviour of Low Alloy Steels for PressureBoundary Components Under Transient Light Water Reactor Operating Conditions’,EU Project CASTOC, Technical Report WP3, 2003.

7. A. Roth, B. Devrient, J. Föhl, ‘Crack Growth Behaviour of Low Alloy Steels forPressure Boundary Components Under Transient Light Water Reactor OperatingConditions’, EU Project CASTOC, Technical Report WP4, 2003.

8. U. Staud, M. Lasch, ‘Die VGB-Richtlinie für Chemie in SWR-Anlagen – aktuellerStand’; VGB-Konferenz ‘Chemie im Kraftwerk 1995’, VGB-Speisewassertagung 1995,Vortrag KKW3.

9. U. Ehrnstén, D. Gómez-Briceño, J. Lapeña, M. Ernestová, M. Žamboch, S. Ritter,H.P. Seifert, A. Roth, J. Föhl, F. Hüttner, T. Weissenberg, ‘Inter-Laboratory CrackGrowth Test on Pressure Vessel Steel 20MnMoNi5-5 in Simulated BWR Environment’,


11th Int. Conf. on Env. Deg. of Mat. in Nucl. Power Systems – Water Reactors,Stevenson, WA, USA, Aug. 10–14, 2003.

10. ASME Boiler and Pressure Vessel Code, Section XI, ‘Rules for In-service Inspectionof Nuclear Power Plant Components’, Appendix A, Article A-4000, ‘MaterialProperties’.

11. A. Roth, et al., ‘Experimental Investigations Concerning the Possible Effect ofDynamic Strain Ageing in the Environmentally Assisted Cracking of Low AlloySteels in Oxygenated High Temperature Water’, 29th MPA Seminar, Stuttgart, Germany,Oct. 09–10, 2003.

12. H.P. Seifert, S. Ritter, J. Hickling, Power Plant Chemistry, 6, pp. 111–123, 2004.13. F.P. Ford, R.M. Horn, J. Hickling, R. Pathania, G. Brümmer, ‘Stress Corrosion

Cracking of Low Alloy Steels under BWR Conditions; Assessments of Crack GrowthRate Algorithms’, 9th Int. Conf. on Env. Deg. of Mat. in Nucl. Power Systems –Water Reactors, pp. 855–863, Newport Beach, CA, USA, Aug. 1–5, 1999.

14. J. Heldt, H.P. Seifert, Nuclear Engineering and Design, 206, pp. 57–89, 2001.

186

13.1 Introduction

The ageing of pressure boundary components is one of the main factorscontrolling the lifetime of nuclear power plants. Under certain circumstancesEAC can be one of the major ageing mechanisms of LAS in high-temperaturewater. The project ‘Crack Growth Behaviour of Low-Alloy Steels for PressureBoundary Components under Transient Light Water Reactor OperatingConditions’ (CASTOC), was performed within the 5th EC frameworkprogramme and addressed the problem of EAC of Western and Eastern typesteels used for pressure boundary components [1, 2].

The objective of the CASTOC project was to screen the EAC behaviourof low-alloy reactor pressure vessel (RPV) steels in high-temperature waterduring load transients and water chemistry transients such as may occurduring start-up and shut-down, steady-state operation and the following modeof commercially operating LWRs. This is in contrast to the worldwide activitiesin the past, which focused mainly on either cyclic loading or static loadingand steady-state operating conditions. The main focus of the project wasdirected at the interactions between static and cyclic loading, which wasrealised, for example, by low frequency corrosion fatigue (LFCF) tests followedby static load or by periodical partial unloading (PPU) with different rise andhold times. In conjunction with the different load spectra, the effect of water

13Crack growth behaviour of low-alloy steelsfor pressure boundary components under

transient light water reactor operatingconditions – CASTOC, Part 2: VVER

conditions

M. E R N E S T O V Á and M. Ž A M B O C H, Nuclear ResearchInstitute (NRI), Czech Republic, B. D E V R I E N T and

A. R O T H, Framatome ANP GmbH, Germany,U. E H R N S T É N, VTT Industrial Systems, Finland,

J. F Ö H L and T. W E I S S E N B E R G, StaatlicheMaterialprüfungsanstalt (MPA), Germany, D. G O M É Z -

B R I C E Ñ O and J. L A P E Ñ A, Centro de InvestigacionesEnergéticas Medioambientales y Tecnológicas (CIEMAT),

Spain and S. R I T T E R and H. P. S E I F E R T, Paul ScherrerInstitute (PSI), Switzerland


chemistry transients was investigated. A more detailed description of theproject is available in references [1] and [2].

The results of the project should in particular be assessed with respect toquality, reliability and their application to plant concerns and possible codeimplementation. This comprises a comparison of the data from the CASTOCproject with data from literature and codes and to give indications where theresults may be considered in plant life management strategies. The testedmaterials and water environment conditions were chosen to address the concernsof both BWRs and VVERs.

In this chapter the crack growth results obtained in a simulated VVERenvironment on materials representing VVER RPVs are presented. The mainfocus of the tests performed in a simulated VVER environment was directedto the crack growth rate (CGR) data during cyclic and static loadings and toinvestigate the effect of materials, oxygen content and the effect of differentconstraint situations realised by different specimen size. The tests wereperformed at NRI within the CASTOC project subsequently after the inter-laboratory comparison test. The results of the tests performed under simulatedBWR/NWC conditions are summarised in [3]. For a detailed description ofall results see [4–8].


The tests performed within the CASTOC project comply with the currentstate-of-the-art knowledge of science and technology in laboratory testing ofEAC processes. The investigated materials represent nuclear grade materials.With regard to the selected environmental conditions, enveloping parameterswere applied.

13.2.1 Materials

The ferritic reactors LAS 15Ch2MFA (material C) and 15Ch2NMFA (materialD) as RPV steels used at VVER 440 and VVER 1000, respectively, wereinvestigated under simulated VVER conditions. The chemical compositionof the materials are summarised in Table 13.1. The steels were quenched andtempered followed by air cooling [6]. Both base materials had a fine-grainedbainitic microstructure. Mechanical properties are given in Table 13.2.

∑ Material C, base material: forged plate 15Ch2MFA (thickness of theplate 140 mm), fabricated according to nuclear grade quality with enhancedsulphur content.

∑ Material D, base material: forged plate 15Ch2NMFA (thickness of theplate 320 mm), fabricated according to nuclear grade quality.


ater reactors188

Table 13.1 Chemical composition (in wt.%) of the investigated materials

Base material Chemical composition [wt.%]

C Mn Si P S Ni Cr Mo Cu V Co As

Material C 0.15 0.40 0.24 0.013 0.015 0.30 2.78 0.64 0.08 0.29 0.009 0.011

Material D 0.14 0.45 0.25 0.009 0.007 1.23 2.15 0.57 0.05 0.08 * *

* not determined


13.2.2 Specimen preparation

The C(T)25 and C(T)50 samples were manufactured from forged plates,material C specimens in L-S direction and material D specimens in S-Ldirection. The specimen were pre-cracked in air using parameters whichfulfill the demand of ASTM E399, 1990 to the final a0/W ~ 0.5.

13.2.3 Experimental facilities

The test equipment at NRI consisted of a heated autoclave vessel with anintegrated bellows system to apply the mechanical load, a water refreshingsystem to adjust desired water conditions, a high pressure pump, and measuringequipment to control the water chemistry [6]. The autoclave had the capacityto install either two C(T)25 or C(T)50 specimens in a daisy chain, whereeach specimen was electrically isolated using ceramic and mica spacers.

The laboratory test unit enables control of the water and loading conditions.The reversed direct current potential drop (DCPD) system for on-line cracklength monitoring was used. The detection limit of the DCPD technique is ofthe order of 10 mm. On-line monitoring was performed on load, pressure andtemperature. The outlet conductivity and defining the outlet oxygenconcentration as the target value of oxygen were measured continuously.The external Ag/AgCl/deionate reference electrode and a platinum probewere used for continuous measurement of the corrosion and redox potentials.

13.2.4 Testing procedure and environment

Each autoclave test was divided into a stabilisation phase, covering the timeneeded for heating and pressurising the autoclave (about 100 h) and aconditioning phase taking at least 100 h at stable conditions before the testphase was started by fatigue loading followed by constant loading. The mainobjective of the fatigue loading part of the test was to create an activelygrowing crack before switching to constant load, and to investigate whethersustained crack growth occurs at desired K value in simulated VVER water.

Normal operation of VVER reactors is characterised by very low oxygen

Table 13.2 Mechanical properties of the investigated materials

Temperature Rp0.2 Rm [MPa] A5 [%] RA [%][∞C] [MPa] Yield Tensile Elongation Reduction

strength strength of area

Material C 20 545 651 22.6 75.0350 469 544 17.8 75.0

Material D 22 570 674 19.7 75.3


concentration (< 10 mg/kg). In order to investigate the effect of higher oxygenlevels – e.g., as residual oxygen after reactor start-up or as a result of oxygeningress during power operation – the tests were performed in oxygenated(~ 200 mg/kg) or oxygen-free (< 20 mg/kg) water with boric acid at 288 ∞Cto generate conservative data with respect to plant conditions. The VVERwater chemistry parameters are listed in Table 13.3.

The aim of the project was to obtain the data at stress intensity factorsoutside the range of linear elastic fracture mode (LEFM). The testing procedureincluded loading the specimens to a stress intensity ranging from 56 to88 MPa÷m starting of fatigue loading using a positive saw tooth waveformwith a rise time of 1000 s and a decline time of 200 s (f = 8.3·10–4 s–1) anda load ratio of R = 0.1, 0.2 or 0.8. After the crack activation, constant loadwas applied for at least 300 h.

After termination of the autoclave testing the cracks were opened at liquidnitrogen temperature. The pre-crack length a0 and EAC advance were measuredat 25/50 (C(T)25/C(T)50) equidistant locations on the fracture surface alongthe notch. Fracture surface investigations using a scanning electron microscopewere performed before and after electrochemical cleaning.


13.3.1 Results from cyclic loading

The CGRs obtained from the experiments were compared with the predictionline of the ASME Boiler and Pressure Vessel Code Case N 643, Section XI,Div. 1 [9]. In this Code Case the prediction line is determined by a thresholdvalue in DKI, the load ratio R and the rise time Dtr of the cycle. Despite thefact that the Code Case was established from Western type PWR waterenvironments, the data obtained from tests in simulated VVER environmentswithin this project are compared with the Code Case prediction line. Thecrack growth rates in mm/cycle have been calculated using the average crackincrement for the cyclic test period, the amount of cycles and the rise time.

Effect of material

The tested materials differed in sulphur, chromium, phosphorus and vanadiumcontent. Although material C had a higher sulphur content (0.015 wt.%)

Table 13.3 Water chemistry parameters in simulated VVER water

Boric acid Potassium Ammonia Conductivity O2 concentrationhydroxide (in the outlet) (in the outlet)

6.8 g/kg 23.5 mg/kg 20 mg/kg ~130 mS/cm < 20 mg/kg or~ 200 mg/kg


compared to material D (0.007 wt.%) the CGRs for material D tended to behigher than those of material C under comparable conditions (Figs 13.1 and13.2). This observation indicates that the sulphur content of the steel was notthe sole material parameter controlling EAC growth rates.

Effect of oxygen content

The effect of dissolved oxygen in VVER water environment on the crackgrowth behaviour can be established only from the tests performed withC(T)50 specimens. At low loading ratio (R = 0.2 and R = 0.1) the CGRstended to be slightly higher in oxygenated water than in oxygen-free water.At the higher loading ratio (R = 0.8) the CGRs in water with enhanceddissolved oxygen content were more than one order of magnitude higherthan in an oxygen-free environment (Figs 13.3 and 13.4).

DKl,max (MPa.m1/2)100101

ASME XI air lineR = 0.2

R = 0.8

R = 0.2/01

CT25 CT50

VVER water, 288 ∞C< 20 mg/kg or 200 mg/kg O2, f = 8.3 ¥ 10–4 s–1

Material C

ASME XI Code case N 643PWR water lineR = 0.2Dtrise = 1000 s

1.E+03

1.E+02

1.E+01

1.E+00

1.E+01

1.E–02

1.E–03

da/d

N (mm

/cyc

le)

13.1 CGRs of material C at a frequency of 8.3 ¥ 10–4 s–1 in oxygenatedand oxygen-free VVER water at 288 ∞C.


Effect of specimen size

The effect of specimen size can be evaluated from the tests performed inoxygenated water with a loading ratio of 0.2 and 0.1 (Fig. 13.5). It becomesobvious that the crack growth per cycle in small C(T)25 specimens is aboutone order of magnitude higher than that of the large C(T)50 specimens forboth materials. This might be caused by higher plastic deformation in thecrack tip area of the smaller specimens, in particular at the low load ratio Rsince the prevailing mechanism is strain induced corrosion cracking. Withregard to the transferability of laboratory results to large components thesmaller specimens obviously provide more conservative data.

13.3.2 Results from constant loading

Although the BWRVIP-60 SCC Disposition Lines (DL) [10] apply to BWR/NWC conditions the da/dt data obtained from the experiments were comparedto these DLs. In most of the tests the crack activated during cyclic loading



R = 0.8

R = 0.2/0.1

VVER water, 288 ∞C< 20 mg/kg or 200 mg/kg O2, f = 8.3 ¥ 10–4 s–1

CT25 CT50

Material D


1.E+03

1.E+02

1.E+01

1.E+00

1.E–01

1.E–02

1.E–03

da/d

N (mm

/cyc

le)

13.2 CGRs of material D at a frequency of 8.3 ¥ 10–4 s–1 inoxygenated and oxygen-free VVER water at 288 ∞C.


arrested after switching to static load. There were only a few exceptions inwhich continuous crack advance was detected under static load. The crackgrowth rates, da/dt, were calculated using the average crack increment forthe static test period and the whole static period test time.

Effect of material

The corrosion fatigue crack activated by cyclic loading arrested after thechange to constant static load during all tests in oxygenated (~ 200 mg/kg)and oxygen-free (< 20 mg/kg) simulated VVER water performed with C(T)25and C(T)50 specimens from material C. During the test phases at static loadof at least 300 h duration no crack growth could be detected at applied stressintensity factors ranging from 56 to 88 MPa÷m.

Concerning the material D, the corrosion fatigue crack activated by cyclicloading, arrested after switching to constant static load during all tests performedon C(T)50 specimens in oxygenated and oxygen-free simulated VVER water.Similar to the behaviour of material C no crack growth was observed during

CT25 CT50

Closed symbols: material Copen symbols: material D

VVER water, 288∞C, f = 8.3 ¥ 10–4 s–1

oxygen-free (< 20 mg/kg O2)


R = 0.8

R = 0.2/0.1



1.E+03

1.E+02

1.E+01

1.E+00

1.E–01

1.E–02

1.E–03

da/d

N (

mm

/cyc

le)

13.3 CGRs of material C and D at a frequency of 8.3 ¥ 10–4 s–1 inoxygen-free VVER water at 288 ∞C.


the phases of static load. However, crack growth under static load was observedin C(T)25 specimens tested in oxygenated VVER water at stress intensityfactors in the range from 58 to 70 MPa÷m.

For comparison the results are displayed together with the BWRVIP-60SCC DLs in Fig. 13.6. For stress intensity factors in the range of up to about56 MPa÷m no crack growth was observed in any of the specimens of materialC or in any of the C(T)50 specimens of material D, which were tested inoxygenated and oxygen-free VVER water. When during the constant loadperiod the stress intensity factor was increased beyond the validity limits forlinear elastic fracture mode (LEFM) continuous crack growth was observedon C(T)25 specimens of material D in oxygenated VVER water and the da/dt data (~10–8 m/s) are higher than the ones expected from DL 2.

The investigation of the fracture surfaces on C(T)25 specimens of materialD which showed crack growth under static load revealed a small portion ofintergranular (IG) cracking during cyclic phase and high portion of IG crackingduring the constant load phase (Fig. 13.7).

CT50

VVER water, 288 ∞C, f = 8.3 ¥ 10–4 s–1

oxygenated (200 mg/kg O2)



R = 0.8

R = 0.2



1.E+03

1.E+02

1.E+01

1.E+00

1.E–01

1.E–02

1.E–03

da/d

N (mm

/cyc

le)

13.4 CGRs of material C and D at a frequency of 8.3 ¥ 10–4 s–1 inoxygenated VVER water at 288 ∞C.


Effect of oxygen

The effect of oxygen could not be demonstrated on materials C and D inVVER water because of crack arrest on most of the specimens. The onlypossibility of examining the effect of oxygen is to use the da/dt data obtainedon C(T)25 specimens of material D in oxygenated (~ 200 mg/kg) VVERwater which could be compared with data in oxygen-free VVER water. Theadditional test with C(T)25 specimens of C and D materials in oxygen-freeVVER water is being carried out separately to the CASTOC project and willdocument the possible effect of oxygen.

Effect of specimen size

The effect of specimen size on the crack growth rate under static load couldnot be clearly demonstrated. In tests with material C no crack growth wasobserved for both specimen sizes (i.e. 25 mm or 50 mm thick) and in materialD crack growth was only detected in C(T)25 specimens, this was associated



C(T) 50

C(T) 25



CT 50CT 25

VVER water, 288 ∞C,200 mg/kg O2, f = 8.3 ¥ 10–4 s–1, R = 0.2/0.1

1.E+03

1.E+02

1.E+01

1.E+00

1.E–01

1.E–02

1.E–03

da/d

N (mm

/cyc

le)

13.5 CGRs of material C and D at a frequency of 8.3 ¥ 10–4 s–1 inoxygenated VVER water at 288 ∞C; C(T)25 and C(T)50 specimens.


with IG cracking. It can be concluded that for the smaller specimen sizeC(T)25 equal nominal stress intensity factors provided more conservativedata than the larger one, which suggests that the mechanism was straininduced corrosion cracking because more extended yielding is anticipated inthe smaller specimen at high KI values.


The PWR part of the CASTOC project addresses environmentally assistedcrack growth phenomena in the low-alloy steels used for pressure boundarycomponents in Russian-type pressurised water reactors (VVER). The numberof tests using sophisticated test facility and measurement technique for theon-line detection of crack advance have provided a more detailed understandingof the mechanism of environmentally assisted cracking and providedquantitative data for CGRs as a function of loading events and time. Thework was focussed on the evaluation of crack growth under cyclic load,crack growth and crack cessation under static load, and on determining theeffect of oxygen content, constraint and stress state outside the range ofLEFM. Nevertheless, regarding the application of the results from the CASTOCproject for the assessment of components in LWRs, the following aspectsshould be considered:

∑ Low-alloy steel base materials for RPV application revealed resistanceto crack growth under constant static load up to stress intensity factors

VIP DL 1 VIP DL 2

Kl (MPa.m1/2)9080706050403020

Closed symbols: material COpen symbols: material D

CT25 CT50

1.E–07

1.E–08

1.E–09

1.E–10

1.E–11

1.E–12

1.E–13

da/d

t (m

/s)

13.6 Crack growth behaviour of material C and D under constantstatic load in oxygenated and oxygen-free VVER water at 288 ∞C; thedata below detection limit ~ 10–13 m/s.


13.7 Typical intergranular attack in C(T)25 specimen of material Dtested in oxygenated VVER water environment at 288 ∞C (staticloading phase).

(a)

(b)

(c)

50 mm

50 mm

50 mm


of about 60 MPa÷m in VVER normal water chemistry. The observedcrack growth behaviour is consistent with plant experience in general,and especially in those cases, where austenitic stainless steel cladding isnot applied or was removed deliberately or by chance.

∑ Under certain environmental and material conditions, however,experimental results from tests performed under static load suggest thata more careful assessment of components should be made. The screeningexperiments carried out in this project revealed crack growth underconstant load for the following condition: material D in oxygenated(~ 200 mg/kg) VVER water and stress intensity factor beyond the validitylimit for linear elastic fracture mode.

∑ The proposed BWRVIP-60 SCC Disposition Line 1 [10] for crack growthunder steady-state BWR/NWC conditions was essentially confirmed aswell as for steady-state VVER conditions. However, with respect to theDisposition Line 2 for transients in load and water chemistry furtherconsideration is recommended based on the results of this project.

∑ The curves provided in the ASME Code Case N 643 for a PWR waterenvironment cover fairly well the data obtained in a VVER environmentfor Russian type RPV steels, even at increased oxygen content.

The CASTOC results provide an important contribution to the understandingof crack growth behaviour as a function of time and on as a consequence ofthe number and height of loading events. This is important in evaluatingtransient events, that may occur in a power plant.


The financial support by the 5th Framework Programme of the EuropeanCommision under contract No. FIKS-CT-2000-00048 is gratefullyacknowledged. The valuable contributions of all partners to data generationand discussion of the results formed the basis for successful performance ofthe project.

13.6 References

1. J. Föhl, U. Ehrnstén, M. Ernestová, D. Gómez-Briceño, J. Lapeña, S. Ritter, A. Roth,B. Devrient, H.P. Seifert, T. Weissenberg, M. Žamboch, ‘Crack Growth Behaviourof Low Alloy Steel for Pressure Boundary Components under Transient Light WaterReactor Operating Conditions – CASTOC’, FISA Conference on EU Research inReactor Safety, Luxembourg, November 12–14, 2001.

2. J. Föhl, U. Ehrnstén, M. Ernestová, D. Gómez-Briceño, J. Lapeña, S. Ritter, A. Roth,B. Devrient, H.P. Seifert, T. Weissenberg, M. Žamboch, ‘Crack Growth Behaviourof Low-Alloy Steels for Pressure Boundary Components under Transient Light WaterReactor Operating Conditions – CASTOC’, FISA Conference on EU Research inReactor Safety, Luxembourg, November 10–12, 2003.


3. S. Ritter, H.P. Seifert, B. Devrient, U. Ehrnstén, M. Ernestová, J. Föhl, D. Gómez-Briceño, J. Lapeña, A. Roth, T. Weissenberg, M. Žamboch, ‘Crack Growth Behaviourof Low-Alloy Steels for Pressure Boundary Components under Transient Light WaterReactor Operating Conditions – CASTOC, Part I: BWR/NWC Conditions’, PaperNo. 281, EUROCORR 2004, Nice, France, September 12–16, 2004.

4. U. Ehrnstén, ‘Crack Growth Behaviour of Low Alloy Steels for Pressure BoundaryComponents Under Transient Light Water Reactor Operating Conditions’, EU projectCASTOC, Technical Report WP1, 2002.

5. U. Ehrnstén, D. Gómez-Briceño, J. Lapeña, M. Ernestová, M. Žamboch, S. Ritter,H.P. Seifert, A. Roth, J. Föhl, F. Hüttner, T. Weissenberg, ‘Inter-Laboratory CrackGrowth Test on Pressure Vessel Steel 20MnMoNi5-5 in Simulated BWR Environment’,11th Int. Conf. on Env. Deg. of Materials in Nuclear Power Systems – Water Reactors,Stevenson, WA, USA, August 10–14, 2003.

6. M. Žamboch, J. Föhl, ‘Crack Growth Behaviour of Low Alloy Steels for PressureBoundary Components Under Transient Light Water Reactor Operating Conditions’,EU project CASTOC, Technical Report WP3, 2003.

7. J. Lapeña, D. Gómez-Briceño, ‘Crack Growth Behaviour of Low Alloy Steels forPressure Boundary Components Under Transient Light Water Reactor OperatingConditions’, EU project CASTOC, Technical Report WP2, 2003.

8. A. Roth, B. Devrient, J. Föhl, ‘Crack Growth Behaviour of Low Alloy Steels forPressure Boundary Components Under Transient Light Water Reactor OperatingConditions’, EU project CASTOC, Technical Report WP4, 2003.

9. Cases of ASME Boiler & Pressure Vessel Code, Case N-643; ‘Fatigue Crack GrowthRate Curves for Ferritic Steels in PWR Water Environment’, Section XI, Division 1,2000.

10. F.P. Ford, R.M. Horn, J. Hickling, R. Pathania, G. Brümmer, ‘Stress CorrosionCracking of Low Alloy Steels under BWR Conditions; Assessments of Crack GrowthRate Algorithms’, 9th Int. Conf. on Env. Deg. of Materials in Nuclear Power Systems– Water Reactors, Newport Beach, CA, USA, pp. 855–863, August 1–5, 1999.

200

14.1 Introduction

Core components of light water reactors (LWR), made of austenitic stainlesssteels (SS) and nickel alloys, subjected to stress and exposed to relativelyhigh fast neutron flux may suffer a cracking process termed Irradiation AssistedStress Corrosion Cracking (IASCC). This degradation phenomenon is a time-dependent process in which neutron and gamma radiation are directly implicatedin the initiation and propagation of cracking [1]. Although this type of crackingwas first recognized in Boiling Water Reactor (BWR), later service failuresattributed to IASCC were observed in Pressurized Water Reactor (PWR)components [2].

Among the material modifications induced by neutron irradiation, RadiationInduced Segregation (RIS) and Radiation Hardening have been identified asthe main contributors to the susceptibility to IASCC of irradiated stainlesssteels. Chromium depletion at grain boundaries produce by Radiation InducedSegregation can justify the IASCC response in oxidizing environments, suchas BWR normal water chemistry [3]. However, in non-oxidizing environments,such as PWR primary water or BWR hydrogen water chemistry, the roleplayed by chromium depletion at grain boundaries on IASCC behaviour ofhighly irradiated material seems to be irrelevant [4], and the influence ofmaterial hardening is becoming more strongly considered.

Radiation hardening can be simulated by mechanical deformation, in spiteof the significant difference observed in the microstructure of both types ofmaterials. Furthermore, it is accepted that the study of the SCC behaviour ofunirradiated austenitic steels with different hardening levels could contributeto the understanding of IASCC mechanism.

In this chapter, crack growth rate data of sensitized 304, 316L and 347with different hardening levels obtained by cold work will be presented anddiscussed. Sensitized 304 SS has been tested in BWR conditions to gainsome insight into the cracking behaviour of core shroud, in which the cracksappear associated to HAZ. 316L SS will be tested in typical PWR primary

14Effect of yield strength on stress corrosion

crack propagation under PWR and BWRenvironments of hardened stainless steels

M. L. C A S TA Ñ O, M. S. G A R C Í A, G. D E D I E G O andD. G O M É Z - B R I C E Ñ O, CIEMAT, Spain

Effect of yield strength on stress corrosion crack propagation 201

water and in high lithium primary water to study the behaviour of thismaterial in its use as baffle former bolts (BFB). Tests have been carried outin high lithium primary water to simulate the postulated environment around theBFB due to gamma heating. Niobium-stabilized type 347 SS has been testedin BWR and PWR to study the behaviour of this material in its use ascore shroud in the Siemens-KWU plants and in its use as BFB in Westinghouseplants.


14.2.1 Materials

Materials tested were three plates of commercial austenitic stainless steels,type 316L, 347 and 304. The chemical compositions and the mechanicalproperties, at room temperature, in ‘as-received’ condition, are shown inTables 14.1 and 14.2, respectively.

Stainless steel type 304 SS was subjected to a sensitization treatment of650 ∞C, 1 hour and air cooling, while no additional heat treatment wasapplied to 316L and 347 SS.

14.2.2 Hardening process

In order to produce materials with different yield strength, several degrees ofcold or warm work have been applied to the material by tensile deformation.Large tensile samples were strained in a 100 Tm tensile machine to several

Table 14.2 Mechanical properties (yield strength, ultimate tensile strength,percentage of elongation and Vickers hardness), at room temperature, in as-received condition

Material YS MPa (Ksi) UTS MPa (Ksi) Elongation (%) Hv (30 Kg)

316L SS 232 (33.9) 566 (82.6) 84 172347 SS 238 (34.8) 568 (82.9) 70 171Sen. 304 SS* 240 (35.1) 707 (103.2) 86 194

*Mechanical properties after the sensitization treatment: 650 ∞C, 1 hour, air cooling

Table 14.1 Chemical composition (%wt) of tested materials

Material C Cr Mn Mo N Ni P S Si Co Nb

316L SS 0.020 17.39 1.28 2.20 0.020 11.49 0.032 0.001 0.45 0.14 –

347 SS 0.028 18.10 1.79 – – 10.35 0.022 0.005 0.17 – 0.46

304 SS 0.059 18.02 1.63 0.20 0.084 8.10 0.022 0.022 0.44 0.15 –


% of deformation, before necking. Finally the gauge length of the strainedsamples was used for Compact-Tension (C-T), tensile and metallographicsamples fabrication. Stainless steels 316L and 347 were hardened by coldwork at room temperature whereas 304 was hardened by warm work at atemperature higher than 200 ∞C.

14.2.3 Test procedure

Crack growth test of hardened stainless steels, 316L, 347 and sensitized 304,with different levels of yield strength have been performed in BWR andPWR conditions. Table 14.3 shows the test matrix carried out. Two samplesper conditions were tested simultaneously.

Crack growth rate tests were performed using 12 mm CT specimens,fabricated according to the ASTM E-399. CT specimens were pre-cracked inair under a triangular wave (22 Hz and R = 0.1). Then, the samples were pre-cracked in high temperature water under a triangular wave (2 ¥ 10–2 Hz andR = 0.6) and, then, constant load with periodic unloading under trapezoidalwave was applied with a holding time of 9000 seconds and 5 and 45 secondfor the unloading and reloading process. In some of the tests in BWR conditions(347 and sensitized 304 SS with highest YS) the time loading and reloadingwas 50 and 450 seconds. Nominal stress intensity factor, Ki, was in the rangeof 25–35 MPam1/2. The apparent crack advance was on-line monitored byDirect Current Potential Drop (DCPD) technique.

After each crack propagation test in high temperature water, the CTspecimens were opened by fatigue in air, at room temperature, and the fracturesurface examined by Scanning Electron Microscopy (SEM). This observa-tion was used to identify the areas of crack propagation, to determine themorphology of the cracking and to verify the accuracy of the DCPD monitoringtechnique.

Table 14.3 Test matrix

PWR 340 ∞C BWR, 290 ∞C

1200 ppm B, 2 ppm Li 108 ppm B, NWC 200 ppb O2

3.2 ppm H2 7 ppm Li3.2 ppm H2

Material YS (MPa) YS (MPa) YS (MPa)316L CW 542 772 819 542 819 – – – –347 CW 518 639 705 – – 238 518 639 705Sen. 304 WW – – – – – 240 503 643 680

CW: Cold worked, WW: Warm worked,



14.3.1 Effect of yield strength in PWR conditions

Sixteen CT samples were tested at 340 ∞C in primary PWR. After testing, of230 to 580 hours duration, some samples exhibit a band of intergranularcracking with a regular initiation from the pre-cracking end and almost evencrack propagation. However, other specimens show localized initiations alongthe pre-cracking end, rapid propagation along elongated grains with thepresence of ligaments and growing finger-like shapes. Some samples onlyshow localized crack initiation along the pre-cracking crack front, identifiedas intergranular morphology. In all the cases the fracture morphology wasalways intergranular.

In addition to the crack propagation obtained by fractographic measurements,apparent crack growth rate was available using the DCPD. Agreement betweencrack growth rates obtained by both techniques has been discussed previously[5]. To establish comparisons among different conditions, maximum crackgrowth rates (CGR) were obtained by dividing the deepest fractographicintergranular propagation by total testing time. Figure 14.1 plots the crackgrowth rate of 316L and 347 SS at 340 ∞C obtained under standard and highlithium primary water conditions. All crack growth rate data are normalizedto a stress intensity of 30 MPam1/2 [6]. In both materials, as yield strengthincreases the crack growth rate increases. In the case of 347 SS the crackgrowth rates obtained are lower than in 316L SS. However, the effect ofyield strength seems to be more evident in 347 SS than in 316L SS. Resultsobtained are consistent with the published data for cold worked 316L SSobtained by Andresen [7], although in this reference tests were performed inhydrogenated pure water instead of primary PWR.

316L Primary water316L High Li347 Primary water

* Corrected CGR to K30

YS (MPa)900850800750700650600550500450

CG

R*

(mm

/s)

1E-6

1E-7

1E-8

14.1 CGR of CW stainless steels in PWR at 340 ∞C.


Crack growth rate was ª 3 times higher in high Li primary water than inPWR standard primary water. The SCC susceptibility of cold worked austeniticstainless steels in lithiated solutions at high temperature has been shownpreviously by Smialowska and co-workers [8]. The possibility of highlyalkaline environments by concentration of LiOH in the liquid phase has beenconsidered for some closed crevices, in particular for PWR core baffle bolts [9].

14.3.2 Effect of yield strength in BWR conditions

Fourteen CT samples were tested at 290 ∞C in BWR conditions Testtemperature, pressure, conductivity, dissolved gases and corrosion potential,using a reference electrode of Cu/Cu2O, were continuously monitored andrecorded. After the pre-cracking at high temperature, samples were exposedfor around 500 hours to oxidizing environment (200 ppb oxygen). Specimensof sensitized 304 with 240 MPa of YS, used as reference, presented a bandof intergranular cracking with regular initiation from the pre-cracking endand almost even crack propagation. However, sensitized and warm-worked304 SS presents partial crack initiation from the pre-cracking end, rapidpropagation along elongated grains, presence of ligaments betweenintergranular cracking and growing finger-like shapes. In the case of 347 SS,no intergranular cracking was observed in material with yield strength of238 MPa, but a clear intergranular propagation was observed for higheryield strength. The crack propagation in 347 SS shows similar characteristicsto those observed in 304 SS.

Following the same criteria as in PWR, maximum crack growth rate hasbeen used for comparisons. Figure 14.2 plots the crack growth rate of sensitized

Sen. AISI-304AISI-347


YS (MPa)800700600500400300200

1E-6

1E-7

1E-8

CG

R*

(mm

/s)

14.2 Crack growth rates of sensitized 304 SS and 347 SS tested at290 ∞C in BWR conditions.


304 and 347 SS as a function of yield strength in BWR conditions, at 290 ∞C.According to the fractographic values, the crack growth rate of sensitized304 SS and 347 SS increases as yield strength increases. However, thecapability of increasing the crack propagation is less pronounced in sensitized304 SS. This material shows significant crack growth rate in the non-warm-work conditions, due to its sensitization treatment. Apparently, the detrimentaleffect of yield strength could be overwhelmed by the marked and well-known high crack growth rate of sensitized material in oxidizing environments.A more significant effect of sensitization is reported by M. Spiedel [10], whoindicates that crack growth rate of heavily sensitized 304 SS is independentof the stress intensity and yield strength. In fact, in the present study, verysimilar CGR have been obtained for specimens with yield strength of 503and 680 MPa. However, it is important to indicate that yield strength of 680MPa in sensitized 304 was obtained by a two-step deformation and that theresultant increase of hardness was lower than expected. As a consequence,the crack growth rate could be affected by the straining procedure followedto produce the target yield strength.

Crack growth rate of 347 SS shows a significant dependence on the yieldstrength, and crack growth rate as high as 3.1 ¥ 10–7 mm/s was measured for347 SS with a YS of 705 MPa. These values are consistent with the publisheddata [11].

This material, 347 SS has been tested in PWR and BWR conditions. Inspite of the higher test temperature in PWR water, crack propagation washigher in BWR (NWC) water. The influence of yield strength in bothenvironments seems to be quite similar, Fig. 14.3. This similar dependencewith the yield strength at both low and high corrosion potential has also beenreported by Andresen [11].

347 BWR 290 ∞C347 PWR 340 ∞C


YS (MPa)900800700600500

CG

R*

(mm

/s)

1E-6

1E-7

1E-8

14.3 Crack growth rate of cold worked 347 SS, tested in PWR andBWR conditions.


14.3.3 Implications for the IASCC process

Hardening induced by neutron radiation is one of the effects observed onstructural materials exposed to radiation fields. The potential contribution ofradiation induced hardening to the initiation and propagation of crackingprocess is becoming more strongly considered, especially for environmentswhere other factors, such as microchemistry, have no significant influence.In order to compare the crack growth rate data obtained with hardened andirradiated materials, the correlation between yield strength and dpa shown inFig. 14.4 has been used [12]. The range of yield strength tested, from 500 to800 Mpa, could correspond to a radiation damage from 1 to 10 dpa.

Figure 14.5 shows crack growth rate data for hardened materials obtainedin this work and in other labs and available data for irradiated materials, inPWR conditions. Crack growth rate for irradiated 304 up to 12 and 35 dpawere obtained under constant load with one per day unloading (holding time>80.000 s.) in some steps of the test, at 335 ∞C [13]. Only data from valid Kvalues have been included. Crack growth rate data for irradiated 304 up to6 dpa under several program conditions were obtained at 340 ∞C [14]. Alldata were corrected to K30.

Irradiated stainless steels at higher yield strength presented a significant

304 SS Jenssen304 SS Jenssen304 SS Jenssen304 SS Jenssen316 SS Jenssen316 SS Jenssen316 SS Jenssen316 SS Jenssen316 SS Bergenlid310 SS Kodama316 SS Kodama347 SS Kodama304/316 SS Odette/LucasNeutron-irradiated300-Series Stainless Steel

76543210Square root (dpa)

1000

800

600

400

200

0

Mea

sure

d t

ensi

le y

ield

str

eng

th (

MP

a)

14.4 Increase of yield strength with dpa [12].


CW 316CW 347WW 316 L. Shoji [13]WW 304 L. Shoji [13]WW 304 L. Andresen [14, 15]WW 316 L. Andresen [14, 15]Irrad-304 Halden [16]Irrad-304 CIR-Content Load [17]Irrad-304 CIR-trapez-1000s [17]Irrad-304 CIR-trapez-10000s [17]

PWR

1E-6

1E-7

1E-8

CG

R (

mm

/s)

YS (MPa)11001000900800700600500

14.5 Comparison of CGR of irradiated material and hardened materialat 340–335 ∞C in PWR conditions.

dispersion, SCC behaviour is probably a multi-parameter phenomenon andthe influence of testing conditions could modify final results. However, thecrack growth rate for irradiated stainless steels and for hardened materialshow a similar trend when the yield strength increases. Therefore, hardenedmaterial seems to reproduce reasonably well the behaviour of irradiatedmaterial under stress corrosion cracking conditions, in PWR conditions.

Similar comparison has been performed for BWR conditions, Fig. 14.6.Crack growth rate data for irradiated 304 SS (13 dpa), 316NG (1.4 dpa) and347 SS were obtained in pure water with 6–7 ppm O2 at 288 ∞C [13, 15]Crack growth rate for irradiated 304 SS (1.4 and 3 dpa) and irradiated 316SS (2 dpa) were obtained in pure water with 300 ppb O2 at 289 ∞C, underseveral load programs [16]. Only data obtained under constant load or undercycling load with holding time of 7200 were included in the Fig. 14.6.Finally, irradiated 304 SS (12 dpa) was tested in pure water with 900 ppb O2.at 288 ∞C, under constant load [17].

Comparison of crack growth rates of irradiated material and hardenedmaterial shows that, in general, crack propagation rates of irradiated materialare above crack growth rates of hardened material. In addition, crack growthrates of irradiated material are also above crack growth rates of hardened andsensitized material. These results seem to suggest that hardened materialsand even hardened and sensitized materials are not appropriated to evaluatethe behaviour of irradiated materials in oxidizing conditions, representativeof BWR environments.


14.4 Conclusions

In PWR at 340 ∞C, crack growth rates increase as yield strength increases,both in 316L SS and 347 SS. The influence of yield strength is more pronouncedin 347 SS that in 316L SS. However the crack propagation obtained is lowerin 347 SS than in 316L SS. Crack growth rates are almost ª 3 times higherin high Li and high pH primary water than in PWR standard primary water.

The detrimental effect of yield strength on crack propagation, in BWRconditions, was less pronounced in sensitized 304 than in 347 SS. For thismaterial, crack propagation rates are higher in BWR (NWC) than in primaryPWR, in spite of the higher temperature of the latter conditions.

Radiation hardening and cold work produces similar effects on crackgrowth in PWR conditions. Hardened material seems to be appropriate toassess the stress corrosion cracking behaviour of irradiated material underthis condition. However, crack growth rate data obtained with hardened andsensitized material poorly reproduce the data of irradiated material and thereforeseems not to be appropriated to evaluate the behaviour of irradiated materialsin oxidizing conditions representative of BWR environments.

14.6 Comparison of CGR of irradiated material and hardened materialat 288 ∞C in BWR conditions.

YS (MPa)900 1000800700600500400300200

BWR1E-5

1E-6

1E-7

CG

R (

mm

/s)

Se. 304347316L Shoji [13]304 Shoji [13]347 Shoji [13]304L Andresen [14, 15]316L Andresen [14, 15]Irrad. 304, 316, 1.4 and 3dpa NRC [19]Irrad. 304, 12 dpa, Studsvik [20]Irrad. 304, 13 dpa Halden [16, 18]Irrad. 347, 2 dpa Halden [16, 18]Irrad. 316NG, 1.4 dpa Halden [16, 18]


14.5 References

1. P. L. Andresen. ‘Irradiation Assisted Stress Corrosion Cracking’. Stress CorrosionCracking Material Performance and Evaluation. R.H. Jones. ASM-210, p 182. (1992).

2. P. Scott, M. Meurier, D. Deydier, S. Silvestre, A. Trency. ‘An analysis of BaffleFormer Bolt cracking in French PWRs’. Environmental Assisted Cracking: PredictiveMethods for Risk Assessment and Evaluation of Materials Equipment andStructures.ASTM 1410.West Conshohocken, (2000).

3. M. Kodama, R. Katsura, J. Morisawa, S. Nishima, S. Suzuki, K. Takamori. ‘IASCCsusceptibility of Irradiated Austenitic Steels under very Low Dissolved Oxygen’.7th Int. Conf. on Environ. Degradation of Materials in NPS-Water Reactor, NACE,p 1121 (1995).

4. K. Fukuya et al. ‘Stress Corrosion Cracking on cold worked 316 stainless steelirradiated to high fluence’. 10th International Conference on Environmental Degradationof Materials in NPS – Water Reactors. Lake Tahoe (2001).

5. M. L. Castaño, M. S: Garcia Redondo, G. De Diego, D. Gómez-Briceño, ‘CrackGrowth Rate in BWR and PWR of Hardened Austenitic Stainless Steels’, 11thInternational Conference on Environmental Degradation of Materials in NuclearPower Systems – Water Reactors August 10–14, 2003.

6. P. Andresen, K. Gott, L. Nelson. ‘Stress Corrosion Cracking of Sensitized type 304Stainless Steels in 288 ∞C Water: A five Laboratory Round Robin’. 9th InternationalSymposium on Environmental Degradation of Material in NPS – Water Reactors.TSM, p 423 (1999).

7. P. Andresen, T. Angeliu, R. Catlin, L. Young, R. Horn. ‘Effect of Deformation onSCC of Unsensitized Stainless Steels’. NACE Corrosion 2000. Paper 203.

8. S. W. Sharkawy, Z. Xia, Z. Szklarska-Smialowska. ‘Stress Corrosion Cracking of AISI-304 and 316 Stainless Steels in Lithiated Water at 350 ∞C’. JNM 195, p 184 (1992).

9. P. M. Scott. 200 F. N. Speller Award Lecture: ‘Stress Corrosion Cracking in PressurizedWater Reactors – Interpretation, Modeling and Remedies’. Corrosion, Vol. 56, No.8, 771–782 (2000).

10. M. O. Spiedel, R. Magdowski. ‘Environmental Degradation Assessment and LifePrediction of Nuclear Piping Made of Stabilized austenitic stainless steels’. Proceedingsof the Inter. Symp. Plant Aging and Life Prediction of Corrodible Structures. Sapporo,Japan, p 951 (1995).

11. P. Andresen, T. Angeliu, L. Young, W. Catlin, R. Horn. ‘Mechanisms and Kinetics ofSCC in Stainless Steels’. 10th International Conference on Environmental Degradationof Materials in NPS – Water Reactors, Lake Tahoe (2001).

12. S. M. Brummer, E. P. Simonen, P. M. Scott, P. L. Andresen, G. S. Was, J. L. Nelson.‘Radiation Induced Material Changes and Susceptibility to Intergranular Failure ofLight Water Reactor Core Internals’. JNM 274, p 299 (1999).

13. T. Shoji, et al. ‘Quantification of Yield Strength Effects on IGSCC of AusteniticStainless Steels in High Temperature Waters’. 11th International Conference onEnvironmental Degradation of Materials in Nuclear Power Systems – Water ReactorsAugust 10–14, 2003.

14. P. L. Andresen. ‘Similarity of Cold Work and Radiation Hardening in Enhancing YieldStrength and SCC Growth of Stainless Steel in Hot Water’. NACE 2002 Paper No. 0509.

15. P. L. Andersen et al. ‘Stress Corrosion Crack Growth Rate Behavior of VariousGrades of Cold Worked Stainless Steel in High Temperature Water’. NACE 2002Paper No. 2511.


16. OECD Halden Reactor Project. Proprietary Information.17. CIR Proprietary Information.18. OECD Halden Reactor Project. Proprietary Information.19. O. K. Chopra, E. E. Gruber and W. J. Shack. ‘Crack Growth behavior of irradiated

austenitic stainless steels in high purity water at 289 ∞C’. 11th Int. Con. onEnvironmental Degradation on Material in NPS – Water Reactors. Skamania Lodge,August 2003.

20. A. Jenssen, P. Efsing, K. Gott, P. O. Anderson. ‘Crack growth behavior of irradiated304L stainless steel in simulated BWR environment’. 11th Int. Con. on EnvironmentalDegradation on Material in NPS – Water Reactors. Skamania Lodge, August 2003.

211

15.1 Introduction

Low-alloy steels (LAS) are widely used for the reactor pressure vessel (RPV)of light water reactors (LWR), which is the most critical pressure-boundarycomponent as far as safety and plant life are concerned. The possible effectof environmentally-assisted cracking (EAC) on RPV structural integritytherefore continues to be a key concern within the context of both reactorsafety and evaluation/extension of plant service life. The accumulated operatingexperience and performance of low-alloy primary pressure-boundarycomponents is very good world-wide [1–5]. The current fatigue design andevaluation codes (ASME III and XI) have been quite successful in preventingfatigue cracks and failures in LAS components and would therefore seem tobe adequate or conservative under most operating circumstances. Instancesof EAC have occurred particularly in boiling water reactor (BWR) service,most often in LAS piping and, very rarely, in the RPV itself [1–5]. Oxidisingagents, usually dissolved oxygen (DO), and relevant dynamic straining (e.g.,arising from thermal stratification, thermal and pressurisation cycles duringstart-up/shut-down, etc.) were always involved [1–6]. These cases wereattributed either to strain-induced corrosion cracking (SICC) or low-frequencycorrosion fatigue (LFCF) (Table 15.1) [2, 4].

Operational experience [1–5] and laboratory background knowledge [7]both indicate, that the current fatigue design curves in Appendix A of theASME Boiler and Pressure Vessel Code, Section XI, might be non-conservativefor certain critical, short-lived, BWR plant transients (start-up/shut-down,hot stand-by, thermal stratification, etc.) and that SICC or very low-cyclecorrosion fatigue covers the most important gap in the field of EAC of LAS.There is a relevant lack of quantitative SICC/LFCF crack growth data underthese critical conditions (i.e., at slow strain rates or very low cyclic frequencies(< 10–2 Hz), intermediate temperatures (150 to 270 ∞C) and high corrosionpotentials (ECP)) [1–9]. Under these combinations of temperature and strainrate, weld and weld HAZ materials could eventually reveal higher CF CGR

15Corrosion fatigue crack growth behaviour of

low-alloy RPV steels at differenttemperatures and loading frequencies

under BWR/NWC environment

S. R I T T E R and H. P. S E I F E R T, Paul ScherrerInstitute (PSI), Switzerland


than the base metal, because of dynamic strain ageing (DSA) [9, 10] orhydrogen-induced EAC (if hardness of HAZ > 350 HV) [11, 12]. An EACproject [9] was therefore started at Paul Scherrer Institute (PSI) to evaluateand assess the adequacy and conservatism of the current reference fatiguecrack growth curves in LWR coolant environment in Appendix A of theASME Boiler and Pressure Vessel Code, Section XI (‘ASME XI wet fatigueCGR curves’) [13] and of existing crack growth models (GE model [3])under these critical conditions. In this project different RPV steels and weldfiller/weld HAZ materials were investigated. This chapter presents someimportant results and conclusions of this experimental parameter study.Special emphasis is placed on loading frequency, temperature and materialeffects.


15.2.1 Materials

Five different types of low-alloy, nuclear grade RPV steels (base metal andHAZ) with either a low, medium, or high sulphur (and aluminium) contentand a RPV weld filler material were investigated (Tables 15.2 and 15.3) [8,9, 14]. The weld filler and weld HAZ materials were taken from thecircumferential core girth weld of a German pressurised water RPV (Biblis C,1976), which has not been commissioned.

All base materials were quenched and tempered. The weld filler, weld

Table 15.1 Basic types of environmentally-assisted cracking

Environmentally-assisted cracking (EAC)

Mechanism SCC SICC CFStress corrosion Strain-induced Corrosion fatiguecracking corrosion cracking

Type of loading Static Slow monotonically Cyclic:rising or very low- low-cycle, high-cyclecycle

LWR operation Transient-free,condition steady-state power Start-up/shut-down, Thermal fatigue,

operation thermal stratification thermalstratification, …

Characterisation BWRVIP-60 ? ASME XI,of crack growth Disposition Lines Code Case N-643

(PWR)

Characterisation ? Susceptibility ASME III,of crack initiation conditions: ECPcrit, Fenv-approach

de/dtcrit, ecrit

Corrosion fatigue crack grow

th behaviour of low-alloy R

PV

steels213

Table 15.2 Chemical compositions of investigated steels in wt.%

Material C Si Mn P S Cr Mo Ni V Al Cu

20 MnMoNi 5 5 A 0.210 0.25 1.26 0.004 0.004 0.15 0.50 0.77 0.008 0.0130 0.06

SA 508 Cl. 2 B 0.210 0.27 0.69 0.005 0.004 0.38 0.63 0.78 0.006 0.0150 0.16

SA 533 B Cl. 1 C 0.250 0.24 1.42 0.006 0.018 0.12 0.54 0.62 0.007 0.0300 0.15

22 NiMoCr 3 7 D 0.215 0.20 0.91 0.008 0.007 0.42 0.53 0.88 0.007 0.0180 0.04

Weld Filler E 0.054 0.17 1.19 0.013 0.007 0.04 0.55 0.94 0.006 0.0053 0.06

20 MnMoNi 5 5 F 0.260 0.32 1.44 0.016 0.015 0.15 0.61 0.63 0.020 0.0290 0.17

HAZ of D G 0.215 0.20 0.91 0.008 0.007 0.42 0.53 0.88 0.007 0.0180 0.04


ater reactors214

Table 15.3 Important properties of investigated LAS (WQ = water quenched; FC = furnace cooled; AC = air cooled; SR = stress relief heattreatment; DSA-index (T = 250 ∞C) = (Z1E-3 %/s – Z1E-1 %/s)/Z1E-1 %/s = ductility loss, +++: high, ++: medium, +: low DSA susceptibility)

Material S Al Nfree Heat treatment Micro- RP288 C∞ DSA-

[wt.%] [wt.%] [ppm] structure [MPa] index

20 MnMoNi 5 5 A 0.004 0.013 30 910–920∞C/6h/WQ, Bainitic 418 –12.3% +++(∫ SA 508 Cl.3) 640–650∞C/9.5h/FC

SA 508 Cl.2 B 0.004 0.015 2 900∞C/ 8h/WQ Bainitic/ 396 –16.4% +++(∫ 22 NiMoCr 3 7) 600∞C/9h/AC ferritic-

pearlitic

SA 533 B Cl.1 C 0.018 0.030 < 1 915∞C/12h/AC/860∞C/12h/WQ Bainitic 412 –8.9% ++(∫ 20 MnMoNi 5 5) 660∞C/12h/FQ/610∞C/40h/FQ

550∞C/12h/FQ/550∞C/12h/FQ

22 NiMoCr 3 7 D 0.007 0.018 3 890–900∞C/7h/WQ Bainitic 400 –0.58% +(∫ SA 508 Cl.2) 640–650∞C/17h/AC+SR*

S3 NiMo 1 E 0.007 0.005 16 *540–555∞C/59h/465∞C/ Ferritic 430 –2.6% +RPV weld filler 590–610∞C/21h/465∞C/

590–605∞C/11.25h/AC

20 MnMoNi 5 5 F 0.015 0.029 ? 900∞C/9h/WQ/650∞C/34h/AC/ Bainitic 439 –9.1% ++(∫ SA 508 Cl. 3) (0.003–0.053) 660∞C/14h/AC/550∞C/47h/

600∞C/8h/AC

Weld HAZ of D G 0.007 0.018 ? 540–555∞C/59h/465∞C/ – 640 ?590–610∞C/21h/465∞C/590–605∞C/11.25h/AC

Corrosion fatigue crack growth behaviour of low-alloy RPV steels 215

HAZ, and some base materials were post-weld heat-treated or stress relieved.The RPV steels had a granular, bainitic (alloy A, C, D, F) or a mixed bainitic/ferritic-pearlitic structure (alloy B) with an average former austenitic grainsize of 10 to 20 mm. The spatial distribution and morphology of the MnSinclusions was fairly homogeneous and similar in alloys A to D covering therange from small, spherical to large (up to a few 100 mm), elongated inclusions.Alloy F revealed distinct banded sulphur segregation zones with large clustersof MnS inclusions. The weld filler material E had a very fine-grained, ferriticmicrostructure with a mean grain size of £ 6 mm. This material revealed avery fine-dispersed distribution of extremely small (£ 1 mm), spherical MnSinclusions. The maximum hardness/microhardness and tensile residual stressin the region of the fusion line/HAZ [14] was limited to 320 HV1/350 HV0.5and to 30 to 40 MPa. Concerning the EAC behaviour, the steels mainlydiffered in their DSA susceptibility and sulphur content/MnS morphology(Table 15.3).

15.2.2 Specimens

25 mm thick compact tension specimens (1T-C(T)) according to ASTM E399were used for all experiments. The base metal specimens were manufacturedfrom forged ingots or hot-rolled steel plates mainly in T-L or L-T orientation.The weld and weld HAZ specimens were manufactured in the T-L or L-Tand T-S or T-L orientation. The specimens were pre-cracked by fatigue in airat room temperature, using a load ratio R of 0.1. The maximal KI at the finalload step was £ 15 MPa·m1/2. The fatigue pre-crack of the HAZ specimenswas positioned in the middle of the HAZ close to the peak hardness region.Because of the wavy form of the fusion line, small parts of the pre-crackplane were in some cases in the region of the fusion line or in the sub-criticalpart of the HAZ.

15.2.3 Environmental parameters

The tests were performed in modern high-temperature water re-circulatingloops under simulated BWR/NWC conditions, i.e., in oxygenated high-temperature water at temperatures of either 288, 250, 200, or 150 ∞C. Waterchemistry (oxygen content and conductivity) and flow rate were measured atthe autoclave inlet and outlet. Inside the autoclave pressure and temperaturewere measured. The stainless steel autoclave volume of 10 litres was exchangedthree to four times per hour. In the vicinity of the specimens a flow velocityin the range of mm/s was obtained. The concentration of DO was adjusted byadding an argon-oxygen mixture to the storage tank. After the demineralisedwater in the storage tank was purified by ion exchangers, the conductivitywas controlled by dosing 0.02 M Na2SO4 to the high-purity (£ 0.06 mS/cm)


water. Concentration of DO and conductivity were controlled at the inletwater and were varied from 0.4 to 8 ppm and 0.06 to 1.0 mS/cm (<1 to370 ppb SO4

2– ). Ionic impurities of the water (inlet and outlet) were analysedby Inductive Coupled Plasma – Atomic Emission Spectroscopy and IonChromatography about four times each test [8, 9].

The ECP of the specimens and the redox potential (platinum probe) werecontinuously monitored by use of an external Ag/AgCl/0.01 M KCl-referenceelectrode. The specimens were electrically insulated from the autoclave,from each other, and from the clip gauges by ZrO2 spacers. The ECP reacheda quasi steady state during the conditioning phase and only increased ca.30 mV during 1000 h [8, 9].

15.2.4 Crack growth monitoring and fractographical post-test evaluation

Crack advance was continuously monitored using the reversed direct currentpotential drop (DCPD) method with a resolution limit of about 5 mm. Thecrack growth increment was calculated by the Johnson formula. The meanpre-fatigue crack length was assigned to the potential drop at the point ofcrack growth initiation during initial loading in the test, as determined accordingto ASTM E1737. The calculated crack length at the end of the experimentwas then verified and, if necessary, corrected with regard to the mean finalcrack length as revealed by post-test fractography. In the case of fairly uniformcrack advance, the difference between calculated and fractographicallydetermined increment of crack advance was <1 to 5%. The specimens werebroken open at liquid nitrogen temperature for post-test evaluation. Forfractographical analysis in the SEM, the oxide film on the fracture surfacewas removed by galvanostatic reduction in an ENDOX-bath [8, 9].

15.2.5 Mechanical loading

Two pre-cracked specimens were investigated simultaneously under the testconditions in oxygenated high-temperature water in a daisy chain. The loadwas actuated with a screw-driven, electromechanical tensile machine withcomputer control. The KI values were calculated according to ASTM E399by the measured load and by the actual mean crack length, derived by post-test fractographical evaluation and by DCPD method [8, 9].

The different phases of the experiments are shown in Fig. 15.1. Thespecimens were loaded with a small mechanical pre-load of approximately9 kN, corresponding to a KI between 12 and 18 MPa·m1/2, and the autoclavewas heated in deoxygenated high-purity water (1). Thereupon the conditioningphase followed, where the environmental parameters were adjusted and thespecimens were pre-oxidised for ≥ 168 hours (2). The subsequent cyclic


loading in LFCF tests was performed under load control. Constant loadamplitude loading with a positive saw tooth waveform (slow loading, fastunloading) was applied (3). In most cases the K I

max values were below theASTM E647 limit. Finally the specimens were unloaded and the autoclavecooled down (4). [8, 9]


15.3.1 Effect of temperature and loading frequency

To investigate the effect of temperature and loading frequency on corrosionfatigue (CF), several LFCF tests with the RPV steels A to C and the weldfiller/weld HAZ material E/G at four different temperatures (150, 200, 250,and 288 ∞C) were performed in oxygenated high-temperature water (DO =8 ppm) with a conductivity of 0.25 mS/cm (65 ppb SO4

2– ). The ECP decreasedfrom +250 mVSHE at 150 ∞C to +150 mVSHE at 288 ∞C. A positive saw toothloading with a high R value of 0.8, a DK of 11.7 to 13.7 MPa·m1/2 andloading frequencies n of 10–5, 10–4, 8.3 ¥ 10–4, and 2.5 ¥ 10–3 Hz wereapplied.

For all frequencies and materials, both cycle-based CGR Da/DNEAC andtime-based CGR da/dtEAC increased with increasing temperature from 150to 250 ∞C. In alloy A, B and the HAZ material G no noticeable change inCF CGR was observed by further increasing the temperature from 250 to

4321

Asymmetrical sawtooth loading

DtR168 h

Pre-loadKl << Klscc

Load, Kl

Time t

Time t

Time t

Dtt = 0

Temperature

Environmental parametersO2, k, …

1: Heating phase 2: Conditioning phase3: Low-frequency fatigue phase (DtR) 4: Cooling phase

15.1 Schematic of the LFCF tests.


288 ∞C. In alloy C and the weld material E a maximum in CGR was observedat 250 ∞C and CF CGR decreased again by further increasing the temperaturefrom 250 to 288 ∞C. This is exemplarily shown in Figs 15.2 and 15.3 for thetime-based CGR in alloy A and the weld material E. In the temperature rangefrom 150 to 250 ∞C, an Arrhenius activation energy EA between 40 and50 KJ/mol was calculated for the different frequencies and materials [8].

For all materials and temperatures, the CF crack advance per cycleDa/DNEAC increased with decreasing frequency, whereas the time-based

n = 8.3 ¥ 10–4 Hz, DtR = 1000 sn = 1 ¥ 10–4 Hz, DtR = 10000 sn = 1 ¥ 10–5 Hz, DtR = 100000 s

Temperature (∞C)300250200150

20 MnMoNi 5 5, 0.004 wt.% S, AR = 0.8, DK = 11.7–13.7 MPa·m1/2

DO = 8 ppm, 65 ppb SO42–

da/d

t EA

C (

m/s

) 10–9

10–10

10–11

10–12

10–8

n = 8.3 ¥ 10–4 Hz, DtR = 1000 sn = 1 ¥ 10–4 Hz, DtR = 10000 sn = 1 ¥ 10–5 Hz, DtR = 100000 s

RPV weld, 0.007 wt. % S, ER = 0.8, DK = 12–13.4 MPa·m1/2

DO = 8 ppm, 65 ppb SO42–

Temperature (∞C)300250200150

10–12

10–8

10–9

10–11

10–10

da/d

t EA

C (

m/s

)

15.2 Effect of temperature on da/dtEAC from LFCF tests at differentfrequencies. Alloy A.

15.3 Effect of temperature on da/dtEAC at different frequencies. Weldfiller material E.


da/dtEAC decreased with decreasing frequency. In alloy B, C, and the weldmaterial E at 288 ∞C, in alloy B at 250 ∞C, and in the HAZ material G at 200/250 ∞C no noticeable change of the CF crack advance per cycle Da/DNEAC

was observed by a reduction of the loading frequency from 10–4 to 10–5 Hz.Based on the results from other temperatures and other materials, it is concludedthat this might be rather an experimental artefact than ‘critical frequencybehaviour’, since there are many reasons for cessation, crack arrest and localcrack pinning phenomena, which could feign such a behaviour. Dependingon temperature and material a power law relationship (Da/DNEAC = A·n–n)between crack advance per cycle Da/DNEAC and loading frequency n wasobserved in the loading frequency range form 10–5 to 10–2 Hz with an exponentof 0.4 to 0.65 (typically 0.5 to 0.6). In most cases, stable and stationary CFcrack growth was observed down to very low frequencies of 10–5 Hz. Thisbehaviour is exemplarily shown in Figs 15.4 and 15.5 for alloy A and theHAZ G.

The same frequency trends and very similar LFCF CGR have also beenobserved in high-purity (k £ 0.06 mS/cm, < 1 ppb SO /Cl )4

2– – , high-temperature water with a DO (ECP) of 8 (+150 to +200 mVSHE) and 0.4 ppm(0 to +60 mVSHE) at temperatures of 288 and 250/240 ∞C for low and highload ratios R of 0.2 and 0.8. Sustained CF crack growth has also beenobserved down to extremely low loading frequencies of 3 ¥ 10–6 Hz underthese conditions. The effect of loading frequency on the cycle-based CGRDa/DNEAC at a realistic DO concentration of 0.4 ppm is exemplarily shownin Fig. 15.6 for alloy F for a load ratio of 0.2 and 0.8. Additionally, the

15.4 Effect of loading frequency and temperature on Da/DNEAC inalloy A.

Frequency n (Hz)

250 ∞C288 ∞C200 ∞C150 ∞C

ASME XI ‘Wet’

20 MnMoNi 5 5 0.004 wt. % S, ADO = 8 ppm, 65 ppb SO4

R = 0.8, DK = 12.0–13.7 MPa·m1/2

2–

10–210–310–410–5

Da/D

NE

AC (mm

/cyc

le)

100

10

1

0.1


corresponding ‘ASME XI wet fatigue CGR’ and results of the RPV steel Aat a loading frequency of 8.3 ¥ 10–4 Hz are also shown. Under low-flowconditions, the ‘ASME XI wet fatigue CGR’ could be significantly exceededin high-purity, high-temperature water with a DO content of 0.4 and 8 ppmat loading frequencies £ 10–2 Hz and temperatures ≥ 150 ∞C (Figs 15.4 to15.6).

250 ∞C288 ∞C200 ∞C150 ∞C

ASME XI ‘Wet’

HAZ, 0.007 wt. % S, G, TSDO = 8 ppm, 65 ppb SO4

R = 0.8, DK = 12.5–14.6 MPa·m1/2

2–

Frequency n (Hz)10–210–310–410–5

100

10

1

0.1

Da/D

NE

AC (mm

/cyc

le)

15.5 Effect of loading frequency and temperature on Da/DNEAC inHAZ G.

400 ppb O2, k = 0.06 mS/cmECP = +50 mVSHE

ASME XI ‘Wet’

ASME XI ‘Wet’

R = 0.2

R = 0.2

20 MnMoNi 5 5, 0.015 wt.% S, F, 240 ∞C20 MnMoNi 5 5, 0.004 wt. % S, A, 250 ∞C

R = 0.2, DK = 42.2–47.7 MPa·m1/2

R = 0.8, DK = 10.9–13.0 MPa·m1/2

R = 0.8, DK = 12 MPa·m1/2

Frequency n (Hz)10–6 10210110010–5 10–4 10–3 10–2 10–1

Da/D

NE

AC (mm

/cyc

le)

100

10

1

0.1

15.6 Effect of loading frequency on Da/DNEAC in alloy F and A andcomparison to ‘ASME XI wet fatigue CGR’.


15.3.2 Effect of loading conditions

LFCF tests at different frequencies ranging from 2.9 ¥ 10–6 to 1.4 ¥ 10–2 Hzand at three different load ratio R (and DK) levels of 0.2 to 0.34 (22.9 to64.4 MPa·m1/2), 0.7 to 0.88 (7.8 to 18.4 MPa·m1/2), and 0.95 to 0.98 (3.2 to1.5 MPa·m1/2) were conducted at 150 to 288 ∞C in water with 8 or 0.4 ppmDO and 65 or < 1 ppb SO4

2– .In Fig. 15.7, CF crack growth increments per fatigue cycle for all materials,

temperatures, and all load ratios are plotted versus the applied stress intensityfactor amplitude DK and are compared to the corresponding ‘ASME XI wetfatigue CGR curves’. The CF CGR Da/DNEAC increased with increasing DKand load ratio and decreasing loading frequency [9]. For loading frequencies< 10–3 Hz, the cycle-based CGR Da/DNEAC in LFCF tests significantly exceededthe ‘ASME XI wet fatigue CGR curves’ by a factor of 2 to 100 for allmaterials as well as for low and high load ratios. Furthermore, the ripple loadtests at very high load ratios R of > 0.95 indicated an EAC-threshold DKEAC

of £ 2 MPa·m1/2 for highly oxidising conditions [9], which is significantlysmaller than the apparent thresholds of the ‘ASME XI wet fatigue CGRcurves’. Values below the ‘ASME XI wet fatigue CGR curves’, a noticeablemechanical fatigue crack growth contribution to the total CF crack growth,and associated fatigue striations on the fracture surface were only observedat loading frequencies ≥ 10–3 Hz. Figure 15.8 shows such an example offatigue striations on the fracture surface of an alloy B specimen. The excess

0.19 < R < 0.34

0.96 < R < 0.980.68 < R < 0.88

DO = 0.4 – 8 ppm< 1 or 65 ppb SO4T = 150–288 ∞C

2–

Heat-affected zone GWeld filler material E,

Solid symbols: 0.004–0.007 wt. % S, open symbols: 0.015–0.018 wt. % S

100101DK (MPa·m1/2)

ASME X1 ‘Wet’R ≥ 0.65

ASME XI ‘Wet’R £ 0.25

10–3

103

102

101

100

10–2

10–1

Da/D

NE

AC (mm

/cyc

le)

15.7 Effect of loading conditions on Da/DNEAC and comparison with the‘ASME XI wet fatigue CGR curves’. Alloy A–C, E, G.


difference from the ‘ASME XI wet fatigue CGR curves’ increased withdecreasing frequency and DK, increasing load ratio R and increasingtemperature with a maximum at 250 ∞C.

15.3.3 Effect of material parameters

At a DO ≥ 0.4 ppm, neither the sulphate nor the DO content had an effect onthe LFCF CGR [9]. Under these highly oxidising conditions (ECP > 0 mVSHE),the low- and high-sulphur RPV steels and the weld filler/weld HAZ materialsE/G showed a comparable CF crack growth behaviour over a wide range ofenvironmental (<1 to 370 ppb SO4

2– , 0.4 to 8 ppm DO) and loading conditions(DK, R, n). This is shown in Fig. 15.9 for a temperature of 288 ∞C anddifferent loading conditions. In Fig. 15.9 the measured CF CGR da/dtEAC areplotted versus the corresponding fatigue CGR da/dtAir in air under otherwiseidentical loading conditions. Air fatigue CGR have been calculated accordingto Eason [15]. All the CGR data of the different materials were within asmall scatter band of one half (n > 10–3 Hz) to one order (n £ 10–3 Hz) ofmagnitude over a wide range of loading conditions with different load ratiosR, stress intensity factor amplitudes DK, and loading frequencies n. Theobserved range of CGR data for the different materials/microstructures for agiven da/dtAir was in the same order of magnitude as the scatter of CF CGRin RPV steels with homogeneous sulphur distribution at a loading frequencyof 10–4 Hz. Therefore, neither the sulphate nor the sulphur content ormicrostructure had a significant effect on LFCF CGR under these highlyoxidising conditions. The same behaviour has also been observed in slowrising load (SRL) tests under identical system conditions [8].

15.8 SEM micrograph of fatigue striations on the fracture surface ofalloy B specimen (150 ∞C, 8 ppm DO, R = 0.8, v = 2.5 ¥ 10–3 Hz).

6 mm


A different trend in the LFCF crack growth behaviour under highly oxidisingconditions was only observed at intermediate temperatures (200 to 250 ∞C)and/or very low loading frequencies £ 3·10–5 Hz (Fig. 15.10). At 250 ∞C andat a loading frequency of 10–5 Hz the cycle-based CGR Da/DNEAC increasedwith increasing DSA susceptibility, which even seemed to dominate the

20 MnMoNi 5 5, 0.004 % S, ASA 508 Cl.2, 0.004 % S, BSA 533 B Cl. 1, 0.018 wt.% S, CWeld filler, 0.007 % S, EWeld HAZ, 0.007 % S, G

n = 3¥10–6–3¥10–3 HzR = 0.2–0.8,

DK = 11–62 MPa.m1/2

T = 288 ∞C65/<1 ppb SO4

DO = 8 ppm

2–

10–13 10–710–810–910–12 10–11 10–10

da/dtinert (m/s)

10–13

10–7

10–12

10–11

10–10

10–8

10–9

da/d

t EA

C (

m/s

)

288 ∞C, 1E-5 Hz250 ∞C, 8.3E-4 Hz250 ∞C,1E-5 Hz

8 ppm O2, 65 ppb SO42–

100

10

1

0.1SA 508 Cl. 2 20 MnMoNi 5 5 SA 533 B Cl. 1

0.018 % SDSA-Ind. = –8.9 %high S/med. DSA

C

0.004 % SDSA-Ind. = –12.3 %

low S/high DSAA

0.004 % SDSA-Ind. = –16.4 %

low S/high DSAB

Da/D

NE

AC (mm

/cyc

le)

Material

15.9 Comparison of da/dtEAC from LFCF tests with different alloys/microstructures.

15.10 Comparison of Da/DNEAC from LFCF tests with alloys A, B and C.


effect of steel sulphur content. The RPV steels A and B with a low sulphurcontent of 0.004 wt.% S and high DSA susceptibility revealed a higher cycle-based CGR Da/DNEAC under these conditions than the high sulphur steel C(0.018 wt.% S) with a moderate DSA susceptibility. With increasing loadingfrequency, the difference between the materials disappeared. At a loadingfrequency of 8.3 ¥ 10–4 Hz all materials revealed very similar CF CGR. At288 ∞C and a loading frequency of 10–5 Hz on the other hand, the cycle-based CGR Da/DNEAC seemed to better correlate to the steel sulphur contentthan to the DSA susceptibility and increased with increasing sulphurconcentration [9].

The CF crack growth behaviour at 288 ∞C may be explained by the GEmodel [3, 16] (film rupture/anodic dissolution mechanism) and a criticalsulphur-anion concentration in the crack-tip electrolyte for fast EAC and itsdependence on the steel sulphur content, bulk sulphur anion concentration,ECP and CGR/loading frequency. The maximum/plateau of CF CGR at/above 250 ∞C and the higher cycle-based CGR Da/DNEAC for the low-sulphursteels with a high DSA susceptibility at intermediate temperatures and verylow loading frequencies n £ 3 ¥ 10–5 Hz clearly indicated that DSA mightaffect the CF crack growth behaviour and eventually even dominate steelsulphur effects under certain temperature/loading frequency combinations.Similarly, a good correlation between DSA and SICC susceptibility in SRLtests [8] and between SCC CGR at intermediate temperatures in constantload tests [9] was observed in other PSI investigations. This further confirmedthe possible effect of DSA on EAC. DSA may result in a higher crack-tipstrain and strain rate than outside the DSA range or than in a material, whichis not susceptible to DSA [9, 10]. The inhomogeneous localisation ofdeformation, the increase in dislocation density and increase in planardeformation by DSA can result in a reduction of the local fracture toughnessand favour brittle crack extension, but also in the mechanical rupture of theprotective oxide film and therefore crack advance by anodic dissolution/hydrogen embrittlement mechanism [9, 10]. Therefore, DSA maysynergistically interact with both mechanisms to increase EAC susceptibility.The concentration of ‘free’, interstitial nitrogen and carbon, which mainlygovern the DSA susceptibility in LAS, might therefore be just as relevant forEAC susceptibility as the steel sulphur content.

15.3.4 Comparison to the GE model

As shown in previous papers [8, 9], the LFCF CGR data of all materials werelying between the ‘high- and low-sulphur line’ of the GE model [3, 16] andwere conservatively covered by the ‘high-sulphur line’ for all frequenciesand temperatures. The model correctly predicts most experimentally observeddata trends. There is now some increasing experimental evidence that the


transition curves between the ‘low- and high-sulphur line’ of the model donot conservatively cover the results under highly oxidising conditions. Thisis exemplarily shown in Figs 15.11 and 15.12.

15.12 Time-domain analysis of LFCF test data with superpositionmodel and comparison to GE model.

GE-Model, ‘High-Sulphur line’Superposition-Model

10–13 10–510–610–710–810–910–12 10–11 10–10

da/dtAir (m/s)

288 ∞C 250 ∞C240 ∞C

10–5

10–13

10–12

10–11

10–10

10–6

10–7

10–8

10–9

da/d

t EA

C (

m/s

)

SA 533 B Cl. 1, 0.018 % S, C20 MnMoNi 5 5, 0.004 % S, A20 MnMoNi 5 5, 0.015 % S, F

T = 240–288 ∞C0.4–8 ppm O2

<1 or 65 ppb SO4n = 3E-6-8E-3 Hz

R = 0.2–0.8DK = 11–62 MPa·m1/2

2–

ECP = + 200 mV

Frequency n (Hz)10–6 10210110010–110–5 10–4 10–3 10–2

ASME XI ‘Air’

ASME XI ‘Wet’

Da/D

NE

AC (mm

/cyc

le)

1000

100

10

1

0.1

0.01

Superposition-model with

GE-Model: High-Sulphur Line, TransitionCurve for ECP = +100 mVSHE, 0.020 %

20 MnMoNi 5.5, F, 0.015 % S, O2 = 400 ppbk = 0.06 mS/cm, ECP = +60 mVSHE

R = 0.8, DK = 10.9–13.0 MPa·m1/2

15.11 Comparison of Da/DNEAC from LFCF tests with GE model andASME XI.


In Fig. 15.11 the results of a LFCF test (R = 0.8, DK = 10.9 to 13 MPa·m1/2,10–5 to 3 ¥ 10–3 Hz) with the RPV steel F (0.015 wt.% S) in high-puritywater (£ 0.06 mS/cm) with a DO of 0.4 ppm (+60 mVSHE) are compared tothe predictions of the GE model [3, 16]. Additionally, the corresponding‘ASME XI wet fatigue CGR’ for these loading conditions in an inert (‘air’)and in high-temperature water environment (‘wet’) are plotted. The exacttransition lines for the given test conditions were not available. The plottedtransition line for an ECP of +100 mVSHE is based on a steel sulphur contentof 0.02 wt.%, high-purity water and quasi-stagnant flow conditions and shouldtherefore conservatively cover the test conditions. Under these conditions,the GE model would predict a critical frequency of ca. 10–3 Hz, which wasnot confirmed by test results, where an increase of the LFCF CGR Da/DNEAC

with decreasing loading frequency down to very low values of 10–5 Hz wasobserved. This clearly indicated that the transition curves of the model mightbe not conservative under highly oxidising conditions (ECP > 0 mVSHE).DSA, which is not considered in the GE model, might be one possible reasonfor this discrepancy. In susceptible materials, DSA may affect the EACcracking behaviour at temperatures from 150 to 300 ∞C, in particular at slowstrain rates/low loading frequencies < 10–4 s–1/ < 10–4 Hz (see pages 222–4).

15.3.5 Assessment of the current ‘ASME XI wet fatigueCGR curves’

The current ‘ASME XI wet fatigue CGR curves’ [13] are based on dataobtained prior to 1980. They depend explicitly on DK and R, but not on othervariables that are known to be important, such as loading frequency or ECP.As already shown in Figs 15.4 to 15.7, the CF CGR Da/DNEAC in LFCF testsunder highly oxidising (ECP = 0 to +250 mVSHE) and low-flow conditionssignificantly exceeded the current ‘ASME XI wet fatigue CGR curves’ forloading frequencies < 10–2 Hz and temperatures > 150 ∞C. Within theinvestigated parameter range, the excess difference to the ‘ASME XI wetfatigue CGR curve’ increased with decreasing frequency, increasing loadratio and temperature with a maximum around 250 ∞C.

The current ‘ASME XI wet fatigue CGR curves’ do not adequately describethe experimentally observed CF crack growth behaviour of LAS in oxygenatedhigh-temperature water. The curves either predict too low (e.g., n £ 10–2 Hzand ECP > 0 mVSHE or 10–2 Hz < n <10 Hz and high R/small DK) or too highCGR (e.g., n £10–2 Hz and ECP < –200 mVSHE or 10–1 Hz < n < 10 Hzand high DK) [9, 16]. System conditions (e.g., n > 10 to 100 Hz), wheresignificant environmental effects on fatigue crack growth can be neglectedor excluded, were not defined in ASME XI. For these reasons, a modificationof the ‘ASME XI fatigue CGR curves’ or the development of a new codecase for BWR/NWC should be taken into consideration. Based on the presented


results, the development of more realistic reference curves should considerboth the strong effect of loading frequency/strain rate and ECP. A differentiationby material parameters (e.g., steel sulphur content) does not seem to benecessary from an engineering point of view. Different curve sets could bedeveloped for several ECP regimes (e.g., BWR/NWC and BWR/HWC). Anyof such procedures would result in more complicated flaw tolerance evaluationsthan so far, since the loading frequency/strain rate of different transients hadto be considered in an adequate way, but it would have the potential toreduce both uncertainty and undue conservatism.

15.3.6 Superposition model and time-domain evaluationof LFCF CGR results

A simple linear superposition model [9], which considers both frequencyand ECP effects, is briefly outlined in the following paragraphs as one possibleway for the development of new reference curves. In this model, the cycle-based CGR in high-temperature water Da/DNEAC is just a simple linearsuperposition of the cycle-based CGR in air Da/DNAir by pure mechanicalfatigue and of the corrosion-assisted CGR Da/DNENV. The first contributionis a purely cyclic-controlled process and independent of loading frequency.The second contribution only occurs during the rising load part of the fatiguecycle and is strongly dependent on crack-tip strain rate de/dtCT (da/dtEAC =A · (de/dtCT)n) and loading frequency. Under cyclic loading conditions it isassumed, that the crack-tip strain rate is proportional to the experimentally-derived and known fatigue CGR in air (de/dtCT a da/dtAir) under otherwiseidentical loading conditions. For the onset of fast EAC, the fatigue CGR inair da/dtAir has to exceed a critical CGR da dt/ Air

crit , which is dependent on theECP. Based on these assumptions, the CF crack growth in high-temperaturewater can be described by the following equations, which are the basis of theso-called ‘time-domain analysis method’:

1. da dt da dt f/ < / = (ECP)Air Aircrit : da/dtEAC = da/dtAir = f (DtR, DK, R)

2. da dt da dt f/ / = (ECP)Air Aircrit≥ : da dt C da dt da dt/ ( / ) + /EAC Air

mAir◊

C, m = f (ECP)

These equations can be easily transformed in a cycle-based form by dividingthem by the loading frequency. The parameters C, m and da dt/ Air

crit have to beconservatively determined by experiments for different ECP regimes, (e.g.,BWR/NWC-, BWR/HWC- and PWR-conditions). In contrast to da dt/ Air

crit , Cand m are expected to be only slightly dependent on ECP. Based on da dt/ Air

crit ,thresholds DKEAC = f (n, R) and critical frequencies ncrit = f (DK, R) for theonset of EAC can be derived.

Such a time-domain analysis for a large data base of LFCF tests [9] in


oxygenated high-temperature water (simulated BWR/NWC operatingconditions) is shown in Fig. 15.12. The air fatigue CGR da/dtAir have beencalculated according to Eason [15]. The test conditions covered a wide rangeof environmental (240 to 288 ∞C, 0.4 to 8 ppm DO, k = 0.06 to 0.25 mS/cm,<1 to 65 ppb SO4

2– ), material (0.004 to 0.018 wt.% S) and loading parameters(DK = 11 to 62 MPa·m1/2, R = 0.2 to 0.8, n = 3·10–6 to 8 ¥ 10–3 Hz). Additionally,the ‘high-sulphur line’ of the GE model [3, 16] and the transition curve foran ECP of +200 mVSHE and a steel sulphur content of 0.02 wt.% are shown.Despite the wide range of parameters, all CF CGR data were lying in arelatively small scatter band of one half to one order of magnitude andwithin a factor of 5 of the calculated regression curve. The data indicated acritical CGR da dt/ Air

crit < 10–13 m/s and further confirmed the non-conservatismof the transition curve of the GE model under highly oxidising conditions.The corrosion fatigue crack growth behaviour of LAS in oxygenated, high-temperature water can therefore be reasonably described by the proposedmodel and by one single equation in the time-based form. Furthermore, itdirectly considers frequency effects and has the potential to define ‘immunityconditions’, where environmental effects on fatigue crack growth can beexcluded or neglected.


The SICC and LFCF behaviour of five different RPV steels and of a weldfiller and weld HAZ material were characterised under simulated transientBWR/NWC conditions by cyclic fatigue tests with pre-cracked fracturemechanics specimens. The experiments were performed in oxygenated high-temperature water at temperatures of either 288, 250, 200, or 150 ∞C. Thesetests revealed the following results for low-flow and highly oxidising conditions(ECP > 0 mVSHE, ≥ 0.4 ppm DO):

15.4.1 Temperature/loading frequency

For all frequencies and materials, both cycle-based CGR Da/DNEAC andtime-based CGR da/dtEAC increased with increasing temperature from 150to 250 ∞C, where a maximum/plateau could be observed with furtherincreasing the temperature to 288 ∞C. With decreasing frequency the CFcrack advance per cycle Da/DNEAC generally increased for all temperaturesand materials. Sustained EAC growth could be maintained down to lowfrequencies of 10–5 Hz. The time-based LFCF CGR da/dtEAC were in therange of 5 ¥ 10–11 m/s (1.6 mm/a) to 5 ¥ 10–8 m/s (160 mm/a) and decreasedwith decreasing loading frequency.


15.4.2 Material aspects

The RPV steels with low and high sulphur content and the weld filler/weldHAZ material showed very similar LFCF CGR over a wide range of loadingconditions. A possible effect of steel sulphur content was only observed at288 ∞C at very low loading frequencies n £ 3 ¥ 10–5 Hz, where Da/DNEAC

seemed to increase with increasing sulphur content. The maximum/plateauof CF CGR at/above 250 ∞C and the higher cycle-based CGR Da/DNEAC forthe low-sulphur steels with a high DSA susceptibility at intermediatetemperatures and very low loading frequencies n £ 3 ¥ 10–5 Hz clearlyindicated that DSA might affect the LFCF behaviour. The concentration of‘free’, interstitial nitrogen and carbon might therefore be just as relevant forEAC susceptibility as the steel sulphur content, at least under conditionswhere DSA is observed.

15.4.3 Comparison to the GE model and to ASME XI

The LFCF CGR of all materials were conservatively covered by the ‘high-sulphur line’ of the GE model for all temperatures and frequencies. Thetransition curves between the ‘low- and high-sulphur line’ seem to be non-conservative under these highly oxidising conditions and the model thereforepredicts too high critical frequencies ncrit and CGR da dt/ Air

crit . The current‘ASME XI wet fatigue CGR curves’ could be significantly exceeded bycyclic fatigue loading at low frequencies (< 10–2 Hz) for low- and high-sulphur steels as well as the RPV weld filler/HAZ materials and low andhigh load ratios in the temperature range between 150 and 288 ∞C. They donot adequately describe and conservatively cover the experimentally observedCF crack growth behaviour of LAS under BWR/NWC conditions. Thedevelopment of more realistic reference fatigue crack growth curves for anew BWR/NWC code case should therefore be taken into consideration. Asimple superposition model/time-domain evaluation method, which includesboth frequency and ECP effects, could be used for that purpose.


The financial support for this work by the Swiss Federal Nuclear SafetyInspectorate (HSK), the Swiss Federal Office of Energy (BFE) and the SwissFederal Office for Education and Science (BBW) is gratefully acknowledged.Thanks are also expressed to U. Ineichen, U. Tschanz, B. Gerodetti, andE. Groth (all PSI) for their experimental contribution to this work.

15.6 References

1. P. Scott, D. Tice, Nuclear Engineering and Design, 119, pp. 399–413, 1990.


2. J. Hickling, D. Blind, Nuclear Engineering and Design, 91, pp. 305–330, 1986.3. F.P. Ford, ‘Environmentally-Assisted Cracking of Low-Alloy Steels’, EPRI NP-

7473-L, Electric Power Research Institute, Jan. 1992.4. Y.S. Garud, S.R. Paterson, R.B. Dooley, R.S. Pathania, J. Hickling, A. Bursik,

‘Corrosion Fatigue of Water-Touched Pressure Retaining Components in PowerPlants’, EPRI TR-106696, Final Report, Nov. 1997.

5. H.P. Seifert, S. Ritter, J. Hickling, Power Plant Chemistry, 6, pp. 111–123, 2004.6. E. Lenz, N. Wieling, Nuclear Engineering and Design, 9, pp. 331–344, 1986.7. O.K. Chopra, W.J. Shack, Nuclear Engineering and Design, 184, pp. 49–76, 1998.8. S. Ritter, H.P. Seifert, Power Plant Chemistry, 5, pp. 17–29, 2003.9. H.P. Seifert, S. Ritter, U. Ineichen, U. Tschanz, B. Gerodetti, ‘Risskorrosion in

druckführenden Komponenten des Primärkreislaufes von SWR’, BFE-Final Report,PSI, Switzerland, Feb. 2003.

10. H. Hänninen et al., ‘Effects of Dynamic Strain Aging on Environment-AssistedCracking of Low-Alloy Pressure Vessel and Piping Steels’, 10th Int. Conf. on Env.Degr. of Mat. in Nucl. Power Systems – Water Reactors, CD-ROM, Paper No. 47,Lake Tahoe, NV, USA, Aug. 6–10, 2001.

11. M. Tsubota et al., ‘Intergranular Stress Corrosion Cracking of Low-Alloy and CarbonSteels in High-Temperature Water’, Proc. 6th Int. Conf. on Env. Degr. of Mat. inNucl. Power Systems – Water Reactors, pp. 53–58, San Diego, CA, USA, Aug. 1–5, 1993.

12. J. Hickling, ‘Wasserstoffinduzierte Spannungsrisskorrosion in niedriglegierten Stählen’,4th MPA-Seminar, Paper No. 7, Stuttgart, Germany, Oct. 4–5, 1978.

13. ASME XI Appendix A-4300, ASME Boiler & Pressure Vessel Code, Section XI,Rules for In-Service Inspection of Nuclear Power Plant Components, Appendix A,Article A-4000, Subsection A-4300: Fatigue Crack Growth Rate, ASME, New York,1998.

14. S. Ritter, H.P. Seifert, ‘Characterisation of the Lower Shell and Weld Material of theBiblis C Reactor Pressure Vessel’, PSI-Report 02-01, ISSN 1019-0643, Jan. 2002.

15. E.D. Eason et al., Nuclear Engineering and Design, 184, pp. 89–111, 1998.16. F.P. Ford, P.L. Andresen, ‘Corrosion Fatigue of A533B/A508 Pressure Vessel Steels

in 288 ∞C Water’, 3rd Int. IAEA Specialist’s Meeting on Subcritical Crack Growth,1, pp. 105–124, W. Cullen, Ed., NUREG/CP-0112, Moscow, USSR, May 14–17,1990.

231

16.1 Introduction

Alloy 600 is widespread in Pressurized Water Reactors (PWR). Results oftests, generally performed with monotonic loadings (constant load, constantdeformation or constant extension rate), show that this alloy is sensitive toStress Corrosion Cracking (SCC) in primary water of PWR. For a few years,there has been a growing interest in assessing the role and the effects ofcyclic loadings or stress transients on the Primary Water Stress CorrosionCracking (PWSCC) behavior of metallic components like Alloy 600. Forinstance, as French nuclear power plants operate at variable power that maygenerate cyclic loadings on the components, the determination of the effectsof cyclic loadings is of great importance.

Congleton et al. [1] concluded there was an accelerating effect of cyclingloading on Crack Growth Rate (CGR). Bosch and Vaillant [2–4] agreed withan environmentally assisted cracking for frequencies below 0.01 Hz.Concerning the influence of the wave form, Lidar [5] concluded that thesaw-tooth form was the most damaging one.

This study takes place in a program with the support of the French Institutefor Radiological Protection and Nuclear Safety (IRSN) that focuses on theeffect of low frequency and high R ratio cyclic loading conditions on Alloy600 in order to improve the knowledge of the conditions leading to enhancedCGRs and to assess some testing procedures allowing ‘gentle cycling’ orperiodic load discharge to generate steady CGRs. Two heats are tested, thefirst one (heat 3110439) was air melted by Allegheny Ludlum and the specimenswere machined in a hot rolled plate. The second one (heat WL344) wasproduced by Techphy.

16.2 Materials and specimen

The composition of the heat 3110439 is given in Table 16.1. The plate wascut in a hot rolled larger plate (30% cold-work). The microstructure (Fig.

16Effect of cyclic loadings on the stress

corrosion crack growth rate in alloy 600in PWR primary water

C. G U E R R E, O. R A Q U E T and L. D U I S A B E A U,CEA, France and G. T U R L U E R, IRSN, France


ater reactors232

Table 16.1 Chemical composition of the two heats in Alloy 600

C Mn Si S Ni Cr Cu Co Fe P Ti Mo N Al CB

RCC-M4101 <0.1 <1 <0.5 <0.015 >72 1417 <0.5 <0.1 610 <0.025 <0.5 – – <0.5 –

3110439 0.045 0.25 0.18 0.0001 75.9 15.6 0.01 0.03 7.4 0.008 0.25 0.02 0.005 0.12 0.01

WL344 0.06 0.82 0.31 <0.001 72.8 15.8 0.01 0.01 9.6 0.008 0.196 – – 0.164 –

Effect of cyclic loadings on stress corrosion crack growth 233

(a)

(b)

16.1 SEM images of the microstructure of the heat 3110439, (a) viewof the lateral side (b) view of the core.

16.1) was characterized by SEM. The grain size is between 4.3 and 4.5(ASTM). All the grain boundaries are covered with carbides. The distributionof intragranular carbides is very irregular. In the core of the plate, there areno intragranular precipitates as well as very near the surface, but on theedges of the plates (top and above), there are two areas of about one or two


millimeters thick where more than fifty percent of the grains contain highdensities of intragranular precipitates. In the core, even if most of the grainsdo not contain any intragranular precipitate, several bands that are dozens ofmicrons thick contain intragranular precipitates. The Vickers hardness isbetween 312 Hv and 320 Hv. The mechanical properties shown in Table 16.2have been measured at 325 ∞C.

The chemical composition of the heat WL344 manufactured by Techphyis given in Table 16.1 and its mechanical properties in Table 16.2. Thismaterial is tested ‘as forged’ with a mean grain size of 6 (ASTM). Themicrostructure of the material shows intergranular carbides and intragranularcarbides (Fig. 16.2). The Vickers hardness on the surface of the bar is around245 Hv.

Tests were conducted on CT specimens; the thickness was 25.4 mmaccording to the 1TCT standard. All the specimens were fatigue precrackedin air according to the ASTM E399 and ISO/DIS 7539 standard. After thetest, the specimens were broken to failure by fatigue in air. Then, the fracturesurface morphology was characterized with a SEM. The initial crack lengthbefore the SCC test and the final crack length were measured on the fracturesurface with a SEM.


The tests are performed in the Venus loop that is a high-temperature, high-pressure re-circulating loop. Venus is equipped with four independentautoclaves. Inside each autoclave up to three specimens in daisy chain canbe tested. For one specimen per autoclave, the crack growth is monitored bya Reverse Direct Current Potential Drop system (RDCPD).

The test was performed in primary water (1000 ppm boron as boric acid,2 ppm lithium as lithium hydroxide) at 325 ∞C. The water chemistry wascontrolled by ion exchange resins and several analyses by ionic chromatographyand by plasma impedance spectrometry were achieved before, during andafter the test. The chlorides, fluorides and sulfates levels were lower than150 ppb for all the analyses.

On each autoclave a hydrogen probe (palladium/silver membrane) isfixed for measuring the hydrogen partial pressure in the water. The dissolved

Table 16.2 Mechanical properties at 325 ∞C of the twoheats in Alloy 600

Rp 0,2% (MPa) UTS (MPa)

3110439 837 850WL344 395 650


hydrogen concentration is calculated on the basis of the hydrogen partialpressure and the hydrogen solubility at this temperature. The hydrogen contentwas kept within the range 25–50 cc.kg–1 by several injections in a by-pass inthe loop.

Four different mechanical loads were applied after either a sequence loadingor a direct loading in the corrosion loop (Table 16.3). The sequence loadingconsists of several sequences of mechanical cycling with an increasing Kand a decreasing frequency as described in Table 16.3. The objective of thisprocedure is to promote the propagation of a regular crack front and tofacilitate the transition between the transgranular fracture mode of theprecracking in air and the intergranular fracture mode of the SCC crack.Following either a direct loading or a sequence loading, three types of waveformwere investigated: constant load, triangular or saw-tooth wave form (Table16.3). The frequency of the triangular wave form was around 6 ¥ 10–4 Hzcorresponding to similar rise and fall time around 14 mn. The frequency ofthe saw tooth wave form was around 6 ¥ 10–4 Hz with a rise time around28 mn and a very fast fall time (10 s).

16.2 SEM image of the microstructure of the core of the WL344 heat.

20 mm


16.4 Results

16.4.1 Fracture morphology

Four 1TCT specimen in 3110439 heat Alloy 600 were tested with four differentmechanical loadings. The duration of the test was more than 1200 h forevery specimen.

Macroscopic observations are shown on Fig. 16.3. The specimens exhibitan irregular crack front for the SCC crack but also for the air fatigue precrack.A lateral view is shown on Fig. 16.4. This irregular crack front could beexplained by the heterogeneous microstructure of the material and, moreparticularly, by the various distributions of the intragranular carbides asshown before. On the contrary, the three WL344 specimens exhibit a regularcrack front for the air fatigue pre crack as well as for the SCC test (Fig.16.5). The distribution of carbides is homogenous for this heat.

The fracture mode is intergranular at constant load and with low frequencycycling (6 ¥ 10–4 Hz) for both the 3110439 heat and the WL344 heat (Fig.16.6). The fracture mode is mainly transgranular for the first sequences ofthe loading sequence (Fig. 16.7) tested only on the 3110439 heat.

The images on Figs 16.8 and 16.9 show the transition between the air

Table 16.3 Experimental procedure

(a) Experimental procedure

Autoclave 1 Autoclave 2 Autoclave 3 Autoclave 4

Loading Direct Sequence Direct DirectMechanical Constant F Constant F Triangular Saw toothload wave form wave form

f = 6 ¥ 10–4 Hz f = 6 ¥ 10–4 Hzand R = 0.8 and R = 0.8

Specimen 1 A600 A600 A600 A600 31104393110439 3110439 3110439

Specimen 2 A600 WL344 A600 WL344 A600 WL344

(b) Details of the sequence loading

Sequence Load ratio R Frequency Wave form

1 0.3 0.26 Hz Triangle2 0.5 0.34 Hz Triangle3 0.6 0.24 Hz Triangle4 0.7 0.27 Hz Triangle5 0.7 0.001 Hz Triangle6 Constant load7 0.7 Partial unloading Triangle

(f = 0.001 Hz)8 Constant load


16.3 Macroscopic observations of the fracture surface, Alloy 600(3110439 heat) DL: direct loading; C: constant load, SL: sequenceloading, T: triangular wave form, S: saw tooth wave form.

(a) DL+C

(b) SL+C(c) DL+T

(d) DL+S

Air fatigueprecrack

Crack growth atconstant load inPWR environment

5 mm

10 mm


16.4 Macroscopic observations of the lateral view of a CT specimen(3110439 heat) DL tested at constant load after a sequence loading.

16.5 Macroscopic observations of the fracture surface of the WL344heat tested at constant load (CL), triangular wave form (T) and saw-tooth wave form (S) after direct loading.

(a) CL (b) T (c) S

10 mm

10 mm


fatigue precrack and the constant load for the tests with direct loading. Noevidence of arrest or delay was found on the locations.

Concerning the specimen tested with low frequency mechanical cycling,an intermediate high frequency cycling (0.4 Hz, R = 0.8) was applied during100 h in the middle of the test. Figure 16.10 shows that this step leads to theductile fracture of remnant ligaments.

(a)

(b)

16.6 SEM image of the intergranular fracture mode observed on3110439 heat at constant load (a) and with low frequency cycling (b).


16.4.2 Crack growth rates

Concerning the 3110439 heat, due to the irregular crack front, the crackgrowth rates were measured in the center of the specimen where the crackfront is regular and the crack propagation occurred in mode I. For the WL344

16.7 SEM image of the transition between the air-fatigue precrackand the sequence loading characterized by transgranular fracture(3110439 heat).

16.8 SEM image of the transition between the air fatigue precrackand the stress corrosion cracking at constant load (3110439 heat).


heat, as the crack front is regular, the crack growth rate is the average valuebetween the maximum and the minimum crack length measured on the fracturesurface of the specimen.

The crack growth rates are presented on Fig. 16.11. The results of thispresent study are compared to those measured by Le Hong et al. [6] at320 ∞C on two heats (WF 675 and HB 400).

The crack growth rate defined by the DCPD method on the 3110439 heattested at constant load was 2 ¥ 10–7 mm.s–1. This value is three times largerthan those measured on the surface but due to the irregular crack front, theDCPD method was representative of an average value between the center ofthe specimen and the edges.

16.5 Discussion

Seven 1TCT specimens of different heats of Alloy 600 were tested with fourdifferent mechanical loadings. According to the crack growth measured onthe samples, the two alloys tested appear to be similarly sensitive to SCC atthis temperature.

Concerning the effect of the mechanical cycling, for the conditions tested,the crack growth is the highest for mechanical cycling for the 3110439 heat(about three times). This is not the case for the WL344 heat. The comparisonbetween the triangular and the saw-tooth loading shape lead to the conclusion

16.9 SEM image of the transition between the air fatigue precrackand the intergranular fracture at constant load for the WL344 heat.


(a)

(b)

16.10 SEM image of the transgranular fracture due to the highfrequency step during the low frequency test with triangular cyclingfor the 3110439 heat (a) and with saw-tooth cycling for the WL 344heat (b).


that the saw-tooth loading is more damaging than the triangular one inagreement with Lidar conclusion [5].

The high susceptibility for cracking of the tested materials is certainly thereason why the accelerating effect of the cyclic loading is not very important.Tests in progress are performed at a lower temperature and for a lowermechanical solicitation. Under these conditions, the accelerating effect ofthe cyclic loading could be more significant.

In this study, the DCPD method was applied to the 3110439 heat specimen.Due to the irregular crack front, this method was not efficient to follow thecrack growth during the test. The observations of the transition between theair fatigue precrack and the test at constant load in the primary water exhibit nosign indicating a crack arrest or a delay after a direct loading for the 3110439heat as well as for the WL344 heat. For these conditions (PWR at 325 ∞C),the loading sequence was not necessary to facilitate the transition betweenthe air fatigue precrack and the SCC and to promote a regular crack front.

The high frequency sequence used during the low frequency mechanicalcycling lead to the fracture of the unbroken ligaments and promote a regularcrack front for both the 3110439 heat and the WL 344 heat. As the DCPDmeasurements were applied to the 3110439 heat, it was influenced by theirregular crack front and then, the increase of the crack growth rate during thelow frequency cycling induced by the high frequency step could not be assessed.

16.11 Crack growth rates measurements for the two heats of Alloy600 at constant load (CSTT), with triangular cycling (TR) and withsaw-tooth cycling (DS) after direct loading.

K max (MPam1/2)4035302520

da/dt (mm.s–1)

1.E-08

1.E-09

1.E-07

1.E-06

WF 675 (Le Hong)

WL344-TR

3110439 - DS

3110439 - CSTT

HB 400 (Le Hong)

WL344-DS

WL344-CSTT

3110439 - TR


Concerning the specimen tested at constant load conditions, no unbrokenligaments were found.

16.6 Conclusions

An important IRSN/CEA program is performed in order to characterize theeffect of a cyclic loading on CGR in primary water of PWRs. The firstresults concern two heats of Alloy 600. Other materials will be tested suchas, for instance, Alloy 182, Alloy 82 and cold-worked stainless steels. TheVENUS loop used for the program makes it possible to test up to 12 CTspecimen during the same test.

For the two heats of Alloys 600, the accelerating effect of the cyclicloading was estimated at 325 ∞C and complementary tests are in progress fora lower temperature and lower mechanical solicitations. During these tests,the influence of the waveform will also be studied.

16.7 Acknowledgments

Financial support by the Directorate for Reactor Safety of the French Institutefor Radiological Protection (IRSN) and by CEA/DSNI is gratefullyacknowledged. The authors would also thank General Electric for the supplyof the 3110439 heat and the ‘Region Ile-de-France’ for the funding of thescanning electron microscope used in this study.

16.8 References

1. J. Congleton, E.A. Charles, Sui G., Review on effect of cyclic loading on environmentalassisted cracking of alloy 600 in typical nuclear coolant water, Corrosion science,volume 43, (2001).

2. C. Bosch, Etude de la relation entre la CSC et la FC basse fréquence de L’Alliage 600en milieu primaire REP, Ph.D. Thesis, University of Bordeaux, France (1998).

3. F. Vaillant, S. Le Hong, C. Amzallag, C. Bosch, Crack growth rate on vessel headpenetrations in alloy 600 in primary water, Colloque Fontevraud IV, 14–18 September1998.

4. F. Vaillant, J.M. Boursier, C. Amzallag, J. Champredonde, J. Daret, C. Bosch, Influenceof a cyclic loading on crack growth rates of alloy 600 in primary environment: anoverview, 11th International Conference on Environmental Degradation of Materialsin Nuclear Systems, Stevenson, WA, Aug. 10–14, 2003.

5. P. Lidar, Aspects of crack growth in structural materials in light water reactors, Ph.D.Thesis, Departement of Material Science and Engineering, Royal Institute of Technology,Stockholm, Sweden (1997).

6. S. Le Hong, F. Vaillant, C. Amzallag, Synthesis and comparison of crack growth ratemeasurements on tubes and plates in Alloy 600 in high temperature hydrogenatedprimary water, in: Advances in Mechanical Behavior, Plasticity and Damage, volume2, Euromat 2000, eds D. Miannay, P. Costa, D. François, A. Pineau, (2000).

245

17.1 Introduction

There are two approaches for making predictions. One is based on the theoreticalmodels (deterministic model); the other is based on experimental data (non-deterministic model). Data was collected from the open literature on boilingwater reactors (BWRs) or BWR environmental condition and was used tolearn the trends of crack growth rates with different parameters.

In order to have a full matrix of data, we had to convert parameters suchas electrolyte concentration into conductivity and pH, while H2, O2, andH2O2 concentrations, temperature, and flow velocity were translated intoelectrochemical potential (ECP). We homogenized all the units of the differentparameters, and de-convoluted the final crack length and integrated time ofthe cracking experiment to estimate crack growth rate (CGR) using theCouple Environmental Fracture Model (CEFM). Once the data washomogenized into a complete table, we had six main parameters to describethe experiments (pH, temperature, electrolyte conductivity, Stress IntensityFactor, Electrochemical Polarization Reverse (EPR) – EPR is a synonym foralloy sensitization, and metal sensitization), and one dependant variable (crackgrowth rate).

The non-deterministic approach used an Artificial Neural Network (ANN),which does not pre-suppose any model or assumption, but learns from thedata. The ANN’s job was to map the six parameters described above into thedependant variable and explore how the CGR was affected by each of thoseparameters.

The main goals of this chapter are to explore ANNs as a predictive toolfor IGSCC crack growth rates and analyze the predictions. This chapter is asummary of the work developed by my graduate student, C.P. Lu, in his1996 thesis [3, 4].

17Pattern recognition model to estimateintergranular stress corrosion cracking

(IGSCC) at crevices and pit sites of304 SS in BWRs environments

M. U R Q U I D I - M A C D O N A L D, Penn State University,USA


17.2 Objective and procedure

Intergranular stress corrosion cracking (IGSCC) of recirculation piping in aboiling water reactor (BWR) has been a major operating problem in theworld [1]. Three conditions have to exist simultaneously to cause IGSCC:The steel has to be in a sensitized condition, subjected to tensile stress, andthe environment must have the impurity for this type of corrosion [2].

The stainless steel has to be in a sensitized condition before IGSCCoccurs. Sensitization usually occurs after the recalculating pipes are welded.Upon welding, the temperature in the matrix adjacent to the weld exists forsome time in the range of 425 ∞C to 815 ∞C [5], resulting in the precipitationof chromium carbides (Cr23C7) at the grain boundary. The chromium carbideprecipitates are high in chromium, while the adjacent alloy is depleted of itschromium. This chromium-depleted alloy adjacent to the grain boundaries ismuch less corrosion-resistant than the interior of the grains, and intergranularattack (including IGSCC) can result. Stainless steel matrices having chromium-depleted grain boundaries are said to be ‘sensitized’ and therefore less resistantto IGSCC.

The necessity for applied mechanical stress is indicated by the fact that nointergranular attack occurs on unstressed samples. Tensile stress will build-up local stresses at the crack tip which causes slip along bands emanatingfrom the tip and therefore results in a sudden rupture of the passive film. Theexposed metal passivates as a passive film reforms, but repassivation is notan instantaneous process. Accordingly, a period exists over which a decreasingfraction of the surface is bare and able to support high dissolution currents.Metal dissolution results in crack advance, but eventually the surface becomespassive again and the crack stops growing. At this point, local stresses beginto build ahead of the crack tip once again, and at some critical level, slipoccurs and the passive film ruptures. Therefore, crack advance is cyclicaldue to the applied external stress.

17.3 Effect of pH

Several investigators have studied the effect of pH on IGSCC in sensitizedstainless steel. However, many of the studies cannot be accepted as ademonstration of pH effects alone since other parameters (e.g., conductivity)have varied simultaneously [6]. A few tests, however, have been run to showthat the effect is certainly attributable to pH. They show that crackingsusceptibility increases with decreasing pH, and that the major pH effect ison crack initiation [7]. In summary, low pH values enhance IGSCC of sensitizedstainless steel.

Andresen [8] ran tests in a simulated BWR environment using sensitizedstainless steel; Na2SO4, NaHSO4, and H2SO4 were added to the aqueous

Pattern recognition model to estimate intergranular stress 247

solutions. The solution conductivity in these tests was kept constant. Theexperiment showed that the susceptibility of sensitized 304 stainless steel toIGSCC increases with decreasing pH values, and that the major pH effect ison crack initiation.

Ohnaka et al. [6] ran constant load tests in Na2SO4 (pH = 6) and in H2SO4

(pH = 3) at 150 ∞C for sensitized 304 stainless steel. The time to fail at pH= 6 is larger than ten times the failure time at pH = 3.

Other tests run in weak alkaline water indicated that the crack propagationrate increases with decreasing pH beyond neutral pH [9].

17.4 Effect of fluid velocity

Fluid flow influences the rate of mass transport of oxygen to the steel surface,and could be expected to affect the rate of IGSCC. By de-convoluting theinitiation and propagation times for Constant Extension Rate Test (CERT)specimens, Choi et al. [10] showed that the initiation time of IGSCC forsensitized stainless steel in 250 ∞C water at the open circuit potential and ata flow velocity of 8 m/s is almost double that under static conditions. On theother hand, the crack growth rate was found to increase with increasing flowrate, particularly at low flow velocities.

A later study by Shim et al. [11] explored the effects of flow velocity andpH on IGSCC of sensitized 304 SS. They found that, in pure water and inacidic solutions, the crack growth rate was increased with increasing flowvelocity. However, in alkaline solutions, the crack growth rate was not affectedby flow velocity. All of these studies involved flow past the external surfaceof CERT specimens and did not attempt to direct flow to the crack tip, asmight occur under fatigue conditions due to the pumping action of the crackflanks. Accordingly, these findings may not hold for fatigue loading. It canbe argued that at higher frequencies the pumping action of the crack flanksis sufficiently severe so as to inhibit intergranular fracture by washing outthe crack enclave and, hence, by preventing the build-up of acidic conditionsat the crack tip.

Finally, Macdonald and Fuller [12] found that flow velocity has no detectableeffect on the critical potential for IGSCC in sensitized 304 SS in 0.01 MNa2SO4 (slightly alkaline media) at 280 ∞C.

17.5 Effect of electrochemical corrosion potential

(ECP)

The most important environmental parameter in IGSCC is the electrochemicalcorrosion potential (ECP) [13]. Sensitized type 304 SS will suffer IGSCConly at potentials (ECPs) above some critical value EIGSCC [14]. Additionally,the crack growth rate in sensitized type 304 SS at ECP>EIGSCC increases


strongly with increasing corrosion potential, probably due to an enhancedability of the external surfaces to consume charge emanating from the crackthrough the reduction of oxygen and hydrogen peroxide.

Macdonald [13] developed a mixed potential model (MPM) for non-equilibrium, high temperature aqueous solutions that exist in BWRs andsuccessfully calculates ECP data for BWR in-vessel components andrecirculation pipes. The ECP is a strong function of the temperature, flowvelocity, metal composition, and radiolytic species of higher O2, H2O2, andH2 concentration.

17.6 Effect of conductivity

The conductivity of the aqueous solutions plays a key role in determining therate of IGSCC in sensitized stainless steel in high temperature aqueous solutions[15, 16]. Previous work by Macdonald and Urquidi-Macdonald [1, 2] on theCoupled Environmental Fracture Model (CEFM) explained the observedincrease in the CGR with increasing conductivity in terms of an enhancementin the throwing power of the external environment. Thus, an increase in theconductivity of the external solution will increase the crack growth ratebecause the higher conductivity allows the positive current created by themetal dissolution reaction inside the crack to be consumed by oxygen reductionover a greater distance from the crack mouth. Accordingly, the externalsurface is capable of accommodating a larger dissolution current, therebyresulting in a larger crack growth rate. However, those same calculationspredict that this effect will saturate as the conductivity increases further andcontrol of the CGR will revert to the crack internal environment at sufficientlyhigh conductivities.

17.7 Effect of sensitization (EPR)

Sensitization of stainless steels towards IGSCC is a common problem duringwelding, particularly in those systems that cannot subsequently be solutionannealed to redissolve the precipitated chromium carbides. The cause ofcracking is the depletion of chromium at the grain boundaries on precipitationof the chromium carbides (Cr23C7) leaving a matrix of high chromium grains‘glued’ together with a lower chromium phase. As a result, the grain boundaryis less corrosion-resistant than the grains, and preferential stress-assistedcorrosion at the grain boundary may be sufficiently severe to cause rapidcrack propagation.

The width of the chromium-depleted zone increases and the chromiumcontent decreases with increasing sensitization time [22]. Furthermore, asthe sensitization time is increased, the critical potential for IGSCC becomesmore negative, i.e. the steel becomes increasingly susceptible to


IGSCC. However, for prolonged sensitization times, the back diffusion ofchromium may actually decrease the susceptibility of stainless steel to IGSCC.It has been measured on sensitized 304 SS in a autoclave attached to arecirculation piping of a BWR the critical potential for IGSCC is that of–230 mVSHE,T.

Vermilyea et al. [17–20] have observed that thicker oxide films are formedon the emergent boundary grains when sensitized type 304 stainless steel isexposed to dilute sulfuric acid and neutral solutions at 290 ∞C. As such, inthe absence of straining, the susceptible emergent boundaries are preferentiallyprotected. However, upon straining, these thicker films rupture to mark anucleation event. After rupture, the film on the matrix rapidly reforms as aresult of high re-passivation rate, while the higher dissolution rate and lowerre-passivation rate at the grain boundaries tend to confine the attack to theseareas.

The studies by Macdonald et al. [21] pointed out that at moderate strainrate, measured at constant extension rate tensile test (CERT), cracks propagatedtrans-granularly until they reached a sensitized grain boundary at whichpoint crack growth takes place inter-granularly. However, this is not alwaysthe case for tests which were performed at much lower strain rates.

There is consensus among the scientific community that a metal or metalalloy with an EPR of the order of 15 C/cm2 will present a sensitizedmicrostructure; however, there is not a clear consensus in the scientificcommunity of the effect of sensitization on IGSCC over a wide range ofEPRs. It is established in the corrosion community that sensitization enhancescrack initiation and accelerates crack growth in BWR conditions.

17.8 Effect of stress intensity

The effect of stress intensity on the susceptibility to IGSCC for sensitized304 stainless steel was studied extensively some years ago. These studiesshowed that a stress intensity of mode I (KI) has a strong influence on thecrack propagation rate [23]. The crack growth rate is strongly dependent onKI, in the range of 20–25 MPa m , but a plateau is observed for higher KI

values. This same data suggests that the threshold stress intensity forenvironment-assisted crack growth (KISCC) and, possibly, the crack velocityin the plateau region depends on the applied potential, at least in sulfateenvironments. However, the data is highly scattered, particularly with regardto the plateau of crack velocity.

Significant micro-branching occurs during the propagation of IGSCC forKI values greater than 35 MPa m . Accordingly, the calculated stress intensityvalues are not reliable in the plateau region, but the trends are significantbecause in this region the crack growth rate is least dependent on KI.


17.9 Data collection

Data, including experiments done in the laboratory under simulated BWRsenvironments and experiments performed on nuclear reactors themselves,were collected from the open literature. This represented a mammoth effort,since there is not a consensus on which parameters impact crack growthrates. After reviewing the literature and the models we developed based infirst principles, we concluded that the variables that affect the crack growthrate are: temperature, pH, electrolyte conductivity, metal alloy, environmentand environmental conductivity, and stress intensity applied to the pipe sections.

Kassner observes [24] that the influence of solution conductivity is not assevere as Ford described. However, in Kassner’s work, more emphasis isplaced on the phenomena of corrosion fatigue and little data is available onconstant load tests. During recent years reliable data have been generated forsensitized solution annealed and cold worked and sensitized SS; data that wehad not analyzed in this chapter.

The first algorithm used in this chapter was dedicated to calculate theroom temperature pH, since it was used instead of electrolyte composition inour analysis. The second challenge was determining the electrochemicalpotential of the metal/electrolyte as a function of the following variables:temperature, flow velocity, hydraulic diameter, and concentrations of hydrogen,oxygen, and hydrogen peroxide on the electrolyte from injection, make up water,and water radiolysis. The data available sometimes reported ElectrochemicalPolarization Reverse (EPR) or metal sensitization. If the metal sensitizationwas not reported and the experiment was performed in an autoclave, weassumed the EPR was 15 MPa*m1/2 corresponding to an un-sensitized material.When the stress intensity was not reported, it was set at 27.5 MPa*m1/2.Since, the crack growth rate was always reported for 304 SS, all crack growthrates were changed to the same units.

17.10 Non-deterministic approach: ANN

To date, several models have been developed using a mechanical orelectrochemical point of view to predict the crack growth rates of sensitizedtype 304 stainless steel. However, all these approaches do not have thecapability to predict every aspect of IGSCC because of the complexity of theproblem. For systems that are too complicated to be modeled mathematically,Artificial Neural Networks (ANN) offer an alternative way of making accuratepredictions through learning from quality data. One particular benefit is thatit can reasonably predict results that are beyond the range of the experimentaldata, provided a good understanding of problems and the limitation of theANN is possessed.

A system with multiple inputs and outputs can be modeled using ANNsby applying the system inputs to the network and using the system outputs


as the desired outputs. After an appropriate number of iterative learningcycles, the ANN then constitutes a non-algorithmic model of the processinvolved.

A neural network is a data processing system which consists of a numberof simple, highly interconnected processing elements in an architecture inspiredby the structure of the cerebral cortex portion of the brain. Hence neuralnetworks are often capable of doing things which humans or animals do wellbut conventional computers do poorly. Therefore, it can model complexsystems that cannot be modeled mathematically.

The configuration emulates cognitive processes of the brain in responseto external stimuli. The fundamental cellular unit of the nervous system and,in particular, the brain is the neuron [26]. A neuron is a processing elementwhich receives signals through structures known as dendrites. Thesedendrites receive information from other neurons through regulatingunits called synapses. After processing, the outputs are transmitted throughstructures known as axons to other neurons. The human brain consists of acomplex network of neurons processing and distributing information in thismanner.

The regulating activity of the synapses in this process is particularlyinteresting, since it represents an important part of the learning-recognitionprocess. The transmission of chemicals, or neurotransmitters, through synapsesis the primary means by which neurons communicate [26]. Synapses areknown to be excitory or inhibitory in nature. If a synapse is excitory, stimulationof the pre-synaptic neuron causes an increase in the probability of firing inthe post-synaptic neuron. If a synapse is inhibitory, it causes a decrease inthe probability of firing in the post-synaptic neuron. The level of informationcommunicated to other neurons is then controlled by the synapse. Synapticefficiency is modified through the actions of certain enzymes and is believedto be the basis for learning. In other words, as information is processed orknowledge is increased, the regulating activity of the synapses is progressivelyrefined.

To understand the cognitive process, it is convenient to compare memoryto a data base; i.e. information is stored at particular locations in the brain.When stimuli are received through the senses, the neural network is triggeredinto action. Through the excitory and inhibitory actions of synapses, signalsare transmitted toward the particular locations that stored the same or similarinformation. The strength of the signal at its destination represents howstrongly the external stimuli resemble the stored data.

The unit analogues to the neuron are the processing elements (PE). Typically,a PE has a number of input paths which represent dendrites. Input informationis weighted using random or specified weights. The weighted inputs are thencombined to form a single input which is then modified by a transfer functionand output to other PEs [27].


Two distinct phases of operation exist: learning and recall. Learning is theprocess of adapting or modifying the connection weights in response tostimuli being presented to the input layer. The activity of successively presentingpairs of input and output data to the network is known as training. This mayinvolve showing a network many examples hundreds of thousands of times.Recall refers to the network activity in processing a given input and creatingan output response. This typically follows learning activity and is based onthe effectiveness of the training-learning process.

Successive pairs of input-output data being presented to the network resultsin continuous refinement of connection weights in the PEs involved. This isespecially true for feedback networks, where information feeds back andforth through the network until some convergence criterion concerning inputand output is reached.

Different forms of summation and transfer functions may be adopted asjudged appropriate for the particular form of activity studied. At last, thenumber of PEs and layers in the network may be varied iteratively or in linewith some explanatory theory.

Type of learning: The network learns by means of a training rule whichmodifies the connection weights in response to inputs and the desired outputs.This method of learning is known as supervised learning. Several types oflearning rules exist.

Back-propagation is a technique that distributes the error in the outputlayer to other layers. A back-propagation network has at least three layers: aninput layer, an output layer, and at least one hidden layer. During the learningphase, both inputs and desired outputs are presented to the network. Usingthe input values, information is propagated forward through the network allthe way to the output layer. The actual output is then compared with thedesired output and the error calculated. Blame for this error is next distributedto all connecting processing elements. In this manner, errors are propagatedback to the input layer. Once the errors are known, the weights are modifiedto minimize the global error. [28].

17.11 Results

The data set collected contained 1322 datum point entries. Each datum pointcontained 6 inputs or parameters (pH, temperature, conductivity, stress intensity,EPR, ECP), and one dependant variable which was expressed as crack growthrate with a logarithm base of 10.

A back propagation containing one input layer of size 6, three hiddenlayers with 50 neurons each, and an output layer of size 2 was designed.There is not a methodology that dictates how to design a net; therefore, theprocess is trial and error.


The net was trained with the full available data for over 24 hours in a highspeed 32-bit, 1.53 GHz Windows XP laptop. Once the net was trained, theperformances of the learning were compared by plotting the CGR measuredversus the CGR predicted by the NN, as shown in Fig. 17.1.

Figure 17.1 shows that the learning is accurate for the data quality used. Anerror of ±5% was assumed to be inherent to the data. Small deviations to a 45degree straight line are observed at the lower and higher ends of the CGRs.Because the data came from several authors and there was no discussion of theerror that the data may contain, we were satisfied with the ANN performance.

Next, we artificially created data for which five out of the 6 parametersthat control or impact the CGR remained constant. We then varied one of theparameters between a minimum and maximum range of a typical operationalnuclear reactor.

EPR influence on CGR

Since the EPR test attacks chromium-depleted regions within the stainlesssteel microstructure, a correlation between the chromium content at the grainboundary and EPR value should exist. Therefore, it would be useful to finda way to relate the direct measurements of chromium depletion by ScanningTransmission Microscopy and Electron Diffraction Spectroscopy (STEM-EDS) to indirect measurements using EPR techniques.

300

AN

N p

red

icte

d c

rack

gro

wth

rat

e (A

ng

stro

m/s

)

250

200

150

100

50

0

‘Measured’ crack growth rate (Angstrom/s)300250200150100500

17.1 Measured CGR versus ANN predicted CGR. A/s indicatesAngstrom/s.


Bruemmer et al. [29] has done systematic studies of chromium depletiondevelopment and made comparisons between grain boundary depletionmeasurements and intergranular corrosion tests by STEM-EDS measurementand EPR techniques. Their studies showed that significant grain boundarydepleted regions, particularly those below 13 weight%, produced large EPRvalues. On the other hand, very low EPR values were measured when grainboundary Cr concentrations rose above 12.5 to 13.5 weight%. Thus, it isapparent that there is a critical Cr content above which sensitization does notoccur. Critical Cr concentrations for intergranular corrosion and IGSCCresistance will depend on both environmental and material variables. Thedata appears to indicate that for 304 SS the maximum sensitization is reachedabove an EPR of 30 C/cm2.

Figure 17.2 shows a very strong dependence of the CGR with the degreeof senzitation in the range of EPR 25 to 30 C/cm2. This results appears to bein agreement with the experiential observation that the maximum impact ofEPR on CGR is obtained about 30 C/cm2.

Stress intensity influence on CGR

The CEFM’s predictions (Congleton’s approach applied [30]) on the crackgrowth rate versus stress intensity are found to increase by a factor of about1.34 by increasing KI from 1 MPa m to 40 MPa m with the greatest rateof change occurring at low stress intensities. However, the correlation ofCongleton’s and Gerberich’s approaches cannot recognize a critical KISCC,below which environmentally-assisted stress corrosion cracking does notoccur, agreeing with experimental evidence. If we extrapolate the ANNprediction shown in Fig. 17.3 at 1 MPa m , we estimate a CGR of about 2E-7 cm/s, while 40 MPa m has a value of about 2.7E-7 cm/s. This is a rate ofabout 1.33, which is surprisingly close to the increase ratio predicted by theCEFM and Congleton’s approach.

pH influence on CGR

The variation of the CGR with pHT is shown in Fig. 17.4. The curve passeda maximum corresponding to a pH of about 6. At high pH, it is expected thatthe crack tip will not become acidified as it does at lower pH, and that themagnetite film is more compact than at lower pH. These two facts mayexplain the results predicted by the effect of pH on CGR. The prediction ofthe ANN is also in agreement with the observations by Andresen on BWRenvironments using sensitized stainless steel [8] in that IGSCC increaseswith decreasing pH values and the major pH effect is on crack initiation. Italso agrees with the observation made by Ohnaka et al. [6] cited earlier inthis chapter.


Temperature influence on CGR

Macdonald et al. [32] calculated the effect of temperature by using theCEFM and compared his results to experimental data measured by Andresen

Degree of sensitization (EPR)35302520151050

2.90E-07

2.70E-07

2.50E-07

2.30E-07

2.10E-07

1.90E-07

1.70E-07

1.50E-07

Log

(C

GR

, cm

/s)

17.2 Impact of sensitization (in C/cm2) on the CGR for a 304 SS pipeat T = 288 ∞C; KI = 27.5 MPavm; Electrolyte conductivity = 0.5 mS/cm;pHT = 5.67.

Stress intensity, MPa m^(1/2)605040302010

2.90E-07

2.70E-07

2.50E-07

2.30E-07

2.10E-07

1.90E-07

1.70E-07

1.50E-07

Log

(C

GR

, cm

/s)

17.3 Impact of stress intensity on the CGR for a 304 SS pipe atT = 288 ∞C; Sensitization = 30 C/cm2; Electrolyte conductivity = 0.5mS/cm; pHT = 5.67.


[31]. Macdonald observed that the CGR passes through a maximum withincreasing temperature at a temperature of about 180 ∞C, as shown in Fig.17.5. Macdonald et al. explain that ‘temperature dependence of the CGR isattributed to the competing effects of temperature on the thermally activatedprocesses that occur at the crack tip and the properties (including ECP andconductivity) of the external environment.’ The measurements andcalculation do not correspond to those of a nuclear reactor, but rather theconditions are those of a room temperature, under oxygenated, low conductivityelectrolyte.

The ANN prediction also shows a similar type of curve which passes bya maximum of about 160 ∞C (Fig. 17.6) for Type 304 SS in a nuclear reactorenvironment. At high temperatures, the oxide formed on steel is more compactand protective while at low temperatures (such as during cold shut down),the concentration of radiolytic species is highly suppressed, making theenvironment less favorable for IGSCC. These two combined phenomenamay explain the reason why IGSCC goes through a maximum of about160 ∞C. Figure 17.6 shows the variation of the CGR as a function of temperature.A maximum it is observed about 160 ∞C. The maximum corresponds to themaximum found by several authors during experimental and theoreticalapproaches [5, 31, 32] (see Fig. 17.5).

pH99887766554

2.90E-07

2.70E-07

2.50E-07

2.30E-07

2.10E-07

1.90E-07

1.70E-07

1.50E-07

Log

(C

GR

, cm

/s)

17.4 Impact of pHT on the CGR for a 304 SS pipe at T = 288 ∞C;Sensitization = 30 C/cm2; Electrolyte conductivity = 0.5 mS/cm; Stressintensity = 27.5 MPa m .


Experimental curve [5]Experimental data [5]‘CEFM - Congleton strain rate optionCEFM - Ford strain rate option

Temperature (∞C)350300250200150100500

1.00E-06

1.00E-07

1.00E-08

Cra

ck g

row

th r

ate

(cm

/s)

17.5 The effect of temperature on CGR in Type 304 SS pipe in dilutesulphuric acid solution having an ambient temperature conductivityof 0.27 mS/cm and a dissolved oxygen concentration of 200 ppb. Theexperimental data (curves) are taken from [31, 32].

17.6 Impact of temperature on the CGR for a 304 SS pipe at pH =5.75; Sensitization = 30 C/cm2; Electrolyte conductivity = 0.5 mS/cm;Stress intensity = 27.5 MPa m .

Temperature (∞C)

350300250200150100500

2.70E-07

2.90E-07

2.50E-07

2.30E-07

2.10E-07

1.90E-07

1.70E-07

1.50E-07

Log

(C

GR

, cm

/s)


17.12 Conclusions

By mining data from different authors and mapping that information intocrack growth rates, we can explore each of the variables or parametersmeasured by the different authors and their impact on the crack growth rates.This is in contrast to the different models currently available that do notaccount for variables that may be of interest for the sake of simplicity, suchas the effect of temperature sensitization on metals or metal alloys.

We demonstrated that ANNs can be a useful predicting tool for researchingthe impact of the parameters on IGSCC and exploring the relative importanceof each variable and the overall effect of water chemistry on IGSCC.

17.13 References

1. D.D. Macdonald and M. Urquidi-Macdonald, Corrosion Science, 32 (1991) 51.2. D.D. Macdonald and M. Urquidi-Macdonald, SRI Project PYC-4032, Final Report,

(1988).3. C.P. Lu, Penn State University, Engineering Science and Mecahnics, Ph. D. degree

thesis, (1996).4. D.D. Macdonald, P.C. Lu and M. Urquidi-Macdonald, ‘Stress Corrosion Cracking in

Type 304 SS in High Temperature Aqueous Systems: I. Artificial Neural NetworksAnalysis’, Corrosion, 95 (1995) 1.

5. A.D. Jones, Principles and Prevention of Corrosion, (1991) 290.6. N. Ohnaka et al., Corrosion, 39 (1983) 214.7. L.G. Ljungber and D. Cubicciotti, Corrosion, 41 (1985) 290.8. P.L. Andresen, EPRI NP-3384, Final Report (1983).9. P. Fejes, R. Ivars and J. Svensson, International Conference on Water Chemistry of

Nuclear Reactor System, (1983) 231.10. H.J. Choi, Y-H. Hu and D.D. Macdonald, Proceeding of The First International

Symposium Environmental Degradation of Materials in Nuclear Power Systems –Water Reactors, (1984) 532.

11. S.H. Shim and Z. Szklarska-Smialowska, Corrosion, 43 (1987) 280.12. D.D. Macdonald and G.A. Fuller, Corrosion, 40 (9184) 474.13. D.D. Macdonald, Corrosion (March 1992) 194.14. D.D. Macdonald and G. Cragnolino, Proceeding 2nd International Symposium

Environment Degradation of Materials in Nuclear Power Systems – Water Reactors,NACE, (1986).

15. F.P. Ford and P.L Andresen, Proceeding of The Third International Symposium onEnvironmental Degradation of Materials in Nuclear Power Systems – Water reactors,(1987) 789.

16. P.L. Andresen, Proceeding of The Third International Symposium on EnvironmentalDegradation of Materials in Nuclear Power Systems – Water reactors, (1987) 301.

17. P. Chung and Z. Szklarska-Smialowska, Corrosion, 37 (1981) 39.18. D.A. Vermilyea, Corrosion, 31 (1975) 421.19. M.E. Indig and D.A. Vermilyea, Corrosion, 31 (1975) 51.20. D.A. Vermilyea and M.E. Indig, Journal of Electrochemistry Society, 119 (1972) 39.21. L.F. Lin, G. Cragnolino, Z. Szklarska-Smialowska, and D.D. Macdonald, Corrosion,

37 (1981) 616.


22. P. Chung, A. Yoshitaake, G. Cragnolino and D.D. Macdonald, Corrosion, 41 (1985)159.

23. T. Shoji, ASME International Conference on Advances in Life Prediction Methods,Albany, N. Y. (1983) 127.

24. H.M. Chung, T.F. Kassner and S. Majumdar, NUREG/CR-4667, 14 (1992) ANL-92/30.

25. Private Communication with D.D. Macdonald.26. J.E. Dayhoff, Neural Network Architecture, (1990) 5.27. P.K. Simpson, Artificial Neural Systems, (1991) 8.28. J.M. Zurada, Introduction to Artificial Neural System, p.173.29. S.M. Bruemmer, L.A. Charlot and B.W. Arey, Corrosion Science, (1987) 328.30. J. Congleton, T. Shoji and R.N. Parkins, Corrosion Science, 25 (1985) 633.31. P.L. Andresen, Corrosion, 49 (1993) 714.32. M. Vankeerberghen* and D.D. Macdonald, ‘Predicting crack growth rate vs.

temperature behavior of Type 304 stainless steel in dilute sulphuric acid solutions’,Corrosion Science, 44 (2002) 1425.

260

18.1 Introduction

The material under study is an extra low carbon grade of an austeniticchromium-nickel steel with a good corrosion resistance and low susceptibilityto intergranular corrosion in as-welded condition. These properties warrantan application of this steel in power plant engineering, chemical or processindustry, etc. During the severe service conditions, components are oftenexposed to a combination of corrosion and fatigue.

An extensive high cycle fatigue experimental program was carried outwith the general aim to study the fatigue crack propagation under varioustesting conditions [1–4]. A part of the program, the results of which aresummarised in the paper, was focused on the influence of water environmentsimulating exploitation conditions of nuclear power plants both at room andelevated temperature. To explain the obtained macroscopic characteristics offatigue crack growth under the given conditions, results of microfractographicanalysis were taken into account.

18.2 Fatigue experiments

Servohydraulic testing system INOVA 100 kN was used for high-cycle fatiguetests. All fatigue tests of CT-specimens (of thickness B = 5 mm and twodifferent widths, W = 50 or 38 mm) were carried out at stress ratio R = 0.3and load frequency f = 1 Hz. Two sets of specimens were tested at room andelevated temperature (T = 300 ∞C) in the aqueous solution simulatingexploitation environment in PWR (B-water). For comparison, one set ofspecimens was tested in air at room temperature.

For the crack length monitoring, the compliance method or drop potentialmethod was used. By means of the secant method, the macroscopic fatiguecrack growth rate v = da/dN was determined from the measured crack lengtha versus applied number of cycles N.

Corresponding DK values were computed by the equation

18Fatigue crack growth in austenitic steel AISI

304L in PWR primary water at room andelevated temperature

I. N E D B A L, J. K U N Z and J. S I E G L, Czech TechnicalUniversity, Czech Republic

Fatigue crack growth in austenitic steel AISI 304L 261

D DK a PBW

aW

aW

( ) = 2 +

1 – 1/2 3/2◊

( )( )

¥ ( ) ( ) ( )ÈÎÍ

˘˚

0.886 + 4.64 – 13.32 + 14.72 – 5.62 3 4a

WaW

aW

aW

18.1

valid for 0 / 1£ £a W [5, 6], where DP = load range [N], B = specimenthickness [m], and W = specimen width [m].

18.3 Macroscopic crack growth rate

Macroscopic fatigue crack growth rates are presented as a function of stressintensity factor range DK (Figs 18.1 and 18.2). In Fig. 18.1, the resultsobtained in PWR primary water are compared with the ones in air, both atthe same temperature T = 20 ∞C. Fatigue crack growth rate in the aqueousenvironment is rather higher than in air, but the influence of aggressiveenvironment is gradually decreasing with increasing DK – there is no substantialdifference caused by the influence of environment in the range of higher DK.The influence of elevated temperature on the fatigue crack growth in PWRprimary water is presented in Fig. 18.2 – in whole DK range, the crackgrowth rate at T = 300 ∞C is substantially (approx. 3¥) lower than at T =20 ∞C, i.e. increasing temperature decelerates the fatigue degradation process

AirB-water

504030201010–4

10–3

10–2

10–1

100

101

DK [MPa.m1/2]

R = 0.3T = 20 ∞C

v [mm/cycle]

18.1 Fatigue crack growth rate v vs. DK in air and B-water at roomtemperature.


in the aqueous environment. The dispersion of CGR values at T = 300 ∞C isnotably larger.

The experimental data v vs. DK were fitted by regression function in theform of general Forman law [7]

v CK K K

R K K

mth

m

cm =

( ) ( – )[(1 – ) – ]

0 1

2◊ D D D

D, DKth £ DK £ (1 – R)Kc 18.2

where DKth and Kc are the regression parameters with formal physical meaningof threshold stress intensity factor range, and fracture toughness respectively.

The two mentioned parameters along with the three exponents mj (j =0, 1, 2) and constant C were estimated by means of the least squares method[8]. The final regression functions v = v(DK) at the same stress ratio R = 0.3are following:(a) in air at room temperature

vK K

K = 3.62

( ) ( – 15.34)(117.89 – )

2.12 0.17

2.34◊ D DD

(15.34 £ DK £ 37.61) MPa·m1/2 18.3.1

(b) in B-water at room temperature

vK K

K = 3.85 10

( ) ( – 14.96)(59.73 – )

3–1.41 0.58

1.90◊ ◊ D DD

(15.04 £ DK £ 44.52) MPa.m1/2 18.3.2

20 ∞C300 ∞C

R = 0.3B-water

v [mm/cycle]

DK [MPa.m1/2]

504030201010–4

10–3

10–2

10–1

100

101

18.2 Fatigue crack growth rate v vs. DK in B-water at T = 20 ∞C and300 ∞C.


(c) in B-water at elevated temperature T = 300 ∞C

vK K

K = 1.38 10

( ) ( – 15.06)(254.22 – )

87–3.75 0.58

35.46◊ ◊ D DD

(15.24 £ DK £ 54.72) MPa.m1/2 18.3.3

for v = da/dN in mm/cycle and DK in MPa.m1/2. The corresponding regressioncurves are also presented in Figs 18.1 and 18.2.

18.4 Fractographic analysis

Fracture surfaces of fatigued specimens were studied by means of lightoptical stereomicroscope (in magnification range 4 to 70 times) and scanningelectron microscope (10 to 30 000 times). Fractographic analysis was focusedon description of fractographic features characterising the crack growth withaim to explain the influence of aggressive environment and/or elevatedtemperature on FCG micromechanisms of the fatigue process in AISI 304Lsteel.

Aside from environment and temperature, multiple initiations along thetip of spark-machined notch in CT-specimens have always been observed.Also the first period of fatigue crack propagation (i.e., in the range of low DKvalues: DK < 17 MPa.m1/2) is independent of testing conditions – in closeneighbourhood of the notch, fracture micromorphology is predominantlyinfluenced by the steel microstructure (Fig. 18.3).

The next periods of fatigue crack propagation will be described for differentconditions separately.

18.3 Micromorphology of fatigue fracture close to the notch, at lowCGR.

5 mm


18.4.1 Air, T = 20 ∞CStriation patches prevail in the range of DK > 17 MPa.m1/2. Fracture surfaceis relatively smooth, boundaries between individual striation patches areformed by low steps (Figs 18.4, 18.5). The higher DK results in the highernumber of secondary microcracks generally created by the striationmicromechanism.

18.4.2 PWR primary water, T = 20 ∞CRoughness of fracture surface increases with increasing DK. Similarly as inair, striation patches dominate for DK > 17 MPa.m1/2. The higher DK results

18.4 Striation patches, a typical feature in all testing conditions(example: air/20 ∞C).

18.5 Detail from Fig. 18.4: striations with germs of secondarymicrocracks.

10 mm

1 mm


in the higher steps between adjacent striation patches and the higher frequencyof transversal secondary microcracks. The second main fractographic featureis intergranular facets (Figs 18.6 and 18.7). The microrelief of these facets ischanging with increasing DK: relative proportion of facets with striationsincreases instead of the smooth facets of intergranular separation. Theoccurrence of intergranular facets can explain the higher crack rate in aqueousenvironment (see Fig. 18.1). With increasing DK, the total area percentage ofintergranular fracture decreases (see Fig. 18.9) and this decline correspondsto the fall of ratio of two compared CGR (see Fig. 18.10 based on the resultsin Fig. 18.1). The third fractographic feature is a transgranular fracture(geometrically badly defined in conventional fractographic nomenclature –see Fig. 18.8). The occurrence of this indecipherable type of fracture is

18.6 Occurrence of intergranular facets in aqueous environment atT = 20 ∞C.

18.7 Detail from Fig. 18.6: smooth facets of intergranular separation.

50 mm

5 mm


unambiguously tied up to testing conditions (b) – it was never observed onfractures of specimens tested under conditions (a) or (c).

18.4.3 PWR primary water, T = 300 ∞CIn PWR primary water at elevated temperature, transgranular striation patchesare the main fractographic feature of the fatigue fracture in the area

18.8 Indecipherable transgranular fracture (only in B-water atT = 20 ∞C).

AISI 304 L steelT = 20 ∞C, R = 0.3

0

1

2

3

4

5

6

7

Inte

rgra

nu

lar

frac

ture

(%

)

DK (MPa.m1/2)403530252015

18.9 Area percentage of intergranular fracture in B-water versus DK.

5 mm


corresponding to DK > 17 MPa.m1/2. The abundance of secondary transversalmicrocracks is higher than in the two previous cases.

Contrary to the fracture of specimens fatigued at room temperature, bothin air and B-water, neither environmentally influenced transgranular fracture(as in Fig. 18.8) nor intergranular separation were observed, and intergranularfacets with striations occurred only rarely. In a final part of fatigue fracturecorresponding to DK > 35 MPa.m1/2, large ductile dimples with diameter upto 100 mm and serpentine glide on their walls were found.

18.5 Conclusions

Macroscopic and microfractographic studies of fatigue crack growth in stainlesssteel AISI 304L at the same stress ratio (R = 0.3) but under variousenvironmental conditions results in the following conclusions:

∑ In comparison with the corresponding data in air, the macroscopic fatiguecrack growth rates in the B-water at room temperature are higher. Thisfact corresponds to concurrence of other fracture micromechanisms onthe degradation process: in air, only a striation formation is the maincontrolling mechanism, whereas in B-water, furthermore intergranularseparation and transgranular fracture associated with the effect of

18.10 Ratio of fatigue crack growth rates in B-water and air versusDK.

AISI 304 L steelT = 20 ∞C, R = 0.3

DK (MPa.m1/2)403530252015

3

2

1

0

v B-w

ater

/vai

r (1)


aggressive environment simultaneously participate in fatigue crack growth.The influence of aggressive environment on fatigue crack growth rate ismost pronounced in low DK range and it is gradually disappearing withincreasing DK. This trend corresponds to fractographically assesseddecreasing proportion of intergranular fracture mechanism on the fatigueprocess in B-water (see graphs in Fig. 18.9 and Fig. 18.10).

∑ In B-water, elevated temperature decelerates the fatigue crack propagationin the whole DK range under study – the crack growth rate at T = 300 ∞Cis approximately three times lower than at T = 20 ∞C. This resultcorresponds to fractographic findings – in contrast to fatigue fracture atroom temperature, there is neither intergranular separation nor theindecipherable environmentally transgranular fracture at T = 300 ∞C.

∑ The presented relations v = v(DK), based on the extensive experimentaldata set, represent credible basic characteristics of fatigue properties ofsteel AISI 304 L. Among others, this information (e.g., in the form ofequations 18.3.1 to 18.3.3) can be used for an extended study in thedomain of quantitative fractography oriented to the fatigue crack growthin the material important for nuclear engineering applications.


This work is a part of the activity within the research project MSM6840770021.The authors gratefully acknowledge Electricité de France for the substantialmaterial and financial support.

18.7 References

1. I. Nedbal, P. Kopriva, J. Kunz, J. Siegl and M. Karlik, ‘Verification of Fatigue Propertiesof Stainless Steel 304L on CT Specimens’. [Report V-KMAT-452/98.] Czech TechnicalUniversity in Prague – Faculty of Nuclear Sciences and Physical Engineering –Department of Materials, 1998.

2. I. Nedbal, J. Siegl, J. Kunz and P. Kopriva, ‘Fatigue Properties of Austenitic Steel304L in Corrosive Environment’. [Report V-KMAT-471/99.] Czech Technical Universityin Prague – Faculty of Nuclear Sciences and Physical Engineering – Department ofMaterials, 1999.

3. I. Nedbal, J. Siegl, J. Kunz and P. Kopriva, ‘Fatigue Behaviour of Austenitic Steel304L in Air and Corrosive Environment’. [Report V-KMAT-484/00.] Czech TechnicalUniversity in Prague – Faculty of Nuclear Sciences and Physical Engineering –Department of Materials, 2000.

4. I. Nedbal, J. Siegl, J. Kunz and P. Kopriva, ‘Fatigue Behaviour of Austenitic Steel304L in B-water at Elevated Temperature’. [Report V-KMAT-509/01.] Prague, CzechTechnical University in Prague – Faculty of Nuclear Sciences and Physical Engineering– Department of Materials, 2001.

5. ASTM E 647-91. ‘Standard Test Method for Measurement of Fatigue Crack GrowthRates’, ASTM, 1991.


6. Y. Murakami et al., Stress Intensity Factors Handbook, Pergamon Press, 1987.7. R.G. Forman and T. Hu, ‘Application of Fracture Mechanics on the Space Shuttle’,

Damage Tolerance Metallic Structures: Analysis Methods and Applications, ASTMSTP 842, J. B. Chang and J. L. Rudd eds., ASTM, 1984, 108–133.

8. P. Kopriva and J. Kunz, ‘Statistical Processing of Experimental Data on Fatigue CrackGrowth’, Proc. CTU Seminar 94, Part C, CTU Prague, 1994, 129–130.

Part IV

Practical experience

271

273

19.1 Introduction

Taking into consideration that the life-time of the VK-50 boiling water reactoris assumed to be prolonged up to 50 years (to 2015), the problem of servicelife justification and safety operation of internal components acquires aparticular significance.

Sensitivity to stress corrosion cracking in corrosive environments andunder internal or external stresses [1–3] is one of the principal problemswhile using austenitic stainless steels as an internal structure material ofcores and steam generators of PWR and BWR [1–3]. This problem relates tochrome depletion of grain boundaries and adjacent zones on the sensitisedregions in the vicinity of welds due to the precipitation of Me23C6 and/orMe6C type chrome carbides [1, 2].

Neutron irradiation favors initiation of the stress corrosion cracking dueto the radiation-induced segregation and depletion at grain boundaries.Moreover, specific mechanisms, appropriate to irradiation conditions, aresuggested for sulfur releasing from MnS sulfides as a result of nucleartransformation of 54Mn in 56Fe, cascading failures on the sulfides-matrixinterface and additional penetration of manganese deep into the metal due tothe reverse effect of Kirkendall [4]. This is followed by the grain boundarypoisoning with such detrimental corrosive elements as sulfur, phosphorusand chlorine. In that case, the investigations of metal state and ensuring itssafe service become of great concern. The real elements of long-term operatedinternal components provide the most significant information on corrosionpropagation in a water coolant under irradiation. A measurement channel ofthe VK-50 reactor after 25 years of operation has been selected as an objectof the investigation. This chapter summarises and analyses the findings andcompares them to previous data obtained from testing a wrapper material ofan emergency assembly (08Cr-18Ni-10Ti) after 30 years of operation [5].The indicated element had failures of corrosion origin in the weld region.

19Corrosion damage to 18Cr-9Ni-Ti steel after

25 years of operation in steam-waterenvironments of the VK-50 reactor

G. V. F I LYA K I N, V. K. S H A M A R D I N,YU. D. G O N C H A R E N K O and V. A. K A Z A K O V,

FSUE ‘SSC RIAR’, Russia


19.2 Material – operation conditions

The measurement channel like most elements of internal components isfabricated of 18Cr-9Ni-Ti steel (Fig. 19.1) and represents a tube of ∆ 76 ¥5 mm, freely extending through all the core and above the reactor coverwhere it is fixed. The lower part of the tube is plugged by a welded plug. Thechannel was operated from the end 1971 to 1996 and was removed withoutany signs of its integrity loss. Operating temperatures were 250–280 ∞C.Operational environment: water under pressure, steam-water mixture andsteam. The chemical composition was as follows: water hardness – 1.2 mg-eq/l; pH – 6.0–6.2; Fe – 0.012 mg/l; chlorine-£0.05 mg/l; copper – 0.009–0.03 mg/l; NO – 0.011–0.03 mg/l; zinc – 0.01–0.04 mg/l; salt content –0.18–0.2 mC/cm.

Axial distribution of the neutron fluence (E > 0.1 MeV and E ≥ 0.5 MeV)in the measurement channel is given in Fig. 19.2.

Five rings of 76 ¥ 10 ¥ 5 mm were cut out to perform the material studyand estimate the material corrosion sensitivity as a function of irradiationand coolant:

Ring No. 1 – at a distance of 130 cm below the reactor core center; itincludes the weld, heat affected zone and base metal;Ring No. 2 – at a distance of 90 cm below the reactor core center;Ring No. 3 – 2 cm below the reactor core center;Ring No. 4 – 90 cm above the reactor core center;

Basket

Well

Shell

Reactor vessel

X18H9TPartition

MC

Upper lattice

Core

Lower lattice

200

180

160

140

120

100

80

60

Corebottom

Core Top

Hei

gh

t o

f ac

tive

zo

ne

(cm

)

15 ¥ 2M FA

19.1 Longitudinal section of the VK-50 reactor core with themeasurement channel.

Corrosion damage to 18Cr-9Ni-Ti steel 275

Ring No. 5 – 550 cm above the reactor core center, i.e. directly under thereactor cover. The maximum neutron fluence for ring 5 at E > 0.5 MeV was2.2 ¥ 109 cm–2 and at E > 0.1 MeV – 2.9 ¥ 109 cm–2.

While considering and analysing the investigation results, the previouspublished data [5], obtained from testing the 08Cr-18Ni-10Ti steel wrapperof the emergency assembly removed after 30 years of operation in the VK-50 reactor, have been used. The indicated element had failures of corrosionorigin in the weld region.

19.3 Experimental results

19.3.1 Metallography

Cracks are observed on ring 1 in the weld and heat affected zone (Fig.19.3a). The outer surface breaking cracks are not revealed. On the outersurface of rings 1 and 2 only the intergranular corrosion is observed.

The intergranular corrosion, corrosion cracking (Fig. 19.4a, b) and under-surface corrosion exfoliation are detected on rings 3–5 (Fig. 19.5a, b). Themaximum corrosion depth is at the core top level and achieves 3–3.5 mm.

19.3.2 Electron microscopy

Electron microscopy of samples cut out from ring 5, having a minimumneutron fluence, showed that the steel structure is characterised by the presenceof austenite equiaxial grains where one may find primary precipitates of

E > 0.1 MeV

E > 0.5 MeV

Core axis (cm)200180160140120100806040200

Flu

ence

(cm

–2)

1 ¥ 1021

0

2 ¥ 1021

3 ¥ 1021

4 ¥ 1021

5 ¥ 1021

6 ¥ 1021

19.2 Axial neutron fluence distribution in the measurement channelextending through the core.


19.3 Cracking and intergranular corrosion at the lower part of themeasurement channel (¥ 200): weld and heat affected zone of ringNo. 1(a); base metal of ring No. 2(b).

19.4 Corrosion cracking (¥ 200): ring No. 3(a), ring No. 4(b).

(a) (b)

(a) (b)

19.5 Under-surface corrosion (exfoliation) of the base metal (ring 5)¥ 200 (a); ¥ 5 (b).

(a) (b)

Magnification ¥ 200


excess phases up to 1 mm and clusters of linear dislocations. The radiationdefects are not revealed.

As shown in Fig. 19.6, the precipitation of the secondary phases from fewtens of micrometers to some hundreds of nanometers occurs in the 08Cr-18Ni-10Ti steel under irradiation. They are mainly located along the grainboundaries and in the adjacent areas (Fig. 9.6 a, b). Their chemical compositionis as follows: 37–46 Fe, 11–15 Cr, 32–45 Ti, 5–6 Ni at %. These precipitatesin terms of their structure are evidently Me23C6 or Me6C type carbides. Thelarge precipitate, shown in Fig. 19.6a, represents TiC. The fine-dispersedprecipitates of the rounded shape, presumably the G-phase, were discoveredin the grain body. Their average size is in the order of 8 nm, the density 5 ¥1015 cm–3. The average diameter of dislocation loops was 9 nm and theirdensity 1.4 ¥ 1016 cm–3. The clusters of the dislocation loops and lineardislocations are revealed in the samples from rings 2–4. The average size ofthe dislocation loops is ~10 nm, the maximum one is 20 nm. The loopconcentration is estimated as 4 ¥ 1016 cm–3.

The crystallographic and elemental analysis of the secondary phaseprecipitates showed that the precipitates, both large and fine dispersed, arefor the most part titanium carbides (TiC) with a FCC lattice.

19.3.3 Fractography

Fractographic observations were performed on the samples of 4 ¥ 4 ¥ 1 mm,cut out from rings 2–4. Before the examinations, the samples were mechanicallyfractured in order to obtain the fresh fracture surface.

A region, remarkable for the fracture nature and having all features ofbrittle fracture, was revealed on the fracture surface at the outer face of themeasurement channel. Its depth is ~3–5 mm for ring 2, 30 mm for ring 3(Fig. 19.7a) and 100 mm for ring 4 (Fig. 19.7b) that is in a good correlationwith the steam content in the coolant (Table 19.1).

19.6 Microstructure of the tube material after irradiation –Transmission electron microscopy.

(a) (b) (c)

500 nm

500 nm 200 nm


19.3.4 Auger spectroscopy

Auger spectroscopy was carried out on the samples cut out from ring 4 inorder to evaluate possible modification of the chemical composition of thesuperficial brittle layer. The choice was governed by the fact that the upperpart of the measurement channel is mostly affected by the corrosion attackand has the maximum thickness of the brittle superficial layer. The examinationswere carried out on the fresh fragment of the fracture surface of 2 ¥ 2 ¥ 0.3mm. To obtain the averaged concentration values of various elements on thesurface, the primary electron beam of about 3 mm diameter was scanned ina raster to light up a surface point of 200 ¥ 200 mm on the sample. Afterregistering Auger spectra (10–12 atomic layers) from the fracture surface,multiple ion etching of the examinated surface was performed for analysingthe elemental composition in depth. The thickness of the remote layer wascalculated from the etching rate, equal approximately to 20 Angstroms perminute. Before measuring Auger peak intensities of the alloying elements

(a) (b)

19.7 Fracture surface of the samples cut out from the ring 3 (a) and4 (b).

Table 19.1 Depth of the corrosion failure of the measurement channel at differentfluences and coolant densities

Ring Point of cutting out Neutron Coolant CorrosionNo. fluence density failure

E > 0.1 MeV (g/cm3) depth (mm)

1 130 cm below the core center 3.0¢1010 1.000 0.12 90 cm below the core center 2.7¢1021 0.665 0.13 3 cm below the core center 5.1¢1021 0.375 1.04 90 cm above the core center 1.7¢1021 0.305 3.55 550 cm above the core center 2.1¢109 £ 0.305 1.8


(Fe, Cr, Ni and Ti) and carbon, ion purification of the surface was carriedout. During the operation the concentration of the main alloying elements inthe material matrix was not modified, but very high carbon value was observedat the outer face washed by coolant (Fig. 19.8).

As shown in Fig. 19.9a, no chrome and nickel and a very high content oftitanium is observed on the fracture surface. Titanium value is about 60times as high as that one in the matrix. The depth of the modified layer variesfrom 0.2 to 1.2 mm. The carbon content decreases up to the initial state levelat the distance of ~1.2 mm, whereas the high level of oxygen is not modifiedand even at the depth of 1 mm is in the order of 40 at% (Fig. 19.9b). Thedepth of the sulfur and chlorine layer is approximately 0.2 and 0.4 mmrespectively and the copper layer is in the order of 1 mm.

The calcium content is strongly decreases in the layer of 0.2 mm thickness,but it is entirely disappeared at a depth of 1 mm (Fig. 19.9c).

To provide information on the uniformity or irregularity of various elementdistribution on the fracture surface, the registration of elemental charts serieswas performed at the accelerating voltages of 3 and 9 kV. Figure 19.10presents the elemental charts of the fracture surface of one of the grainslocated in the center of each photo. As seen in Fig. 19.10a, chlorine isdistributed uniformly. In contrast sulfur is concentrated mainly in the clustersof 5–6 mm (Fig. 19.10b). It is especially good seen with lower magnificationsat the accelerating voltage of 3 kV (Fig. 19.10c). As a rule, an elevatedconcentration of iron (Fig. 19.10e), chrome (Fig. 19.10h), as well as nickelis noted in the sulfur clusters. These sulfur clusters appear to representcomplex sulfide compounds containing iron, chrome and nickel.

Copper is distributed in the form of separate clusters along the edges ofthe grain selected for examination (Fig. 19.10d). An elevated copperconcentration is observed in several points, located along the fracture patternperimeter, and it is absent at the surface of the fracture itself. The carbon

Distance from the outer edge of the tube (mm)1000 200 300 400 500 600 700 800

5

0

10

15

20

25

Ato

mic

par

t o

f ca

rbo

n (

%)

19.8 Carbon distribution along the thickness of the measurementchannel wall (ring No. 4) in its upper part.


Cu

CaClS

(c)

1.210.80.60.40.20Distance from the failure surface (mm)

0

1

2

3

4

Co

nce

ntr

atio

n (

%)

(b)


Co

nce

ntr

atio

n (

%)

60

C

O

40

20

0

(a)


Co

nce

ntr

atio

n (

%)

Fe

Cr

NiTi

80

60

40

20

0

19.9 Concentration of the main added elements (a), minus the lightones and impurities), oxygen and carbon (b), as well as sulfur,chlorine, copper and calcium (c) in the irradiated steel as a functionof the distance from the fracture surface.


distribution chart (Fig. 19.10g), the Auger peak of which has the maximumintensity, in terms of topographic nuances, bears a strong resemblance to afracture surface image in absorbed or secondary electrons (Fig. 19.10i).There are no particular carbon clusters, unless the areas are available withsome lower carbon concentration in the points where an elevated oxygencontent is observed (Figs 19.10f, g). The elevated oxygen content in the rightlower part of the fracture patter (Fig. 19.10f) is accompanied by an elevatednickel content and a slightly heightened content of chrome (Fig. 19.10h).

19.4 Discussion

During the entire operation period the measurement channel has been subjectedto small longitudinal tensile stresses in the range of 0.5–0.7 Mpa, not able toprovoke stress corrosion cracking, even at the critical concentration of oxygenand chlorine in the coolant. Accumulation of chlorine and carbon at themetal/coolant interface and their diffusion in the metal thickness is a moreserious factor [2]. So, when examinating the emergency assembly wrapperoperated during 30 years in the same reactor, a higher chlorine content wasdetected at a depth up to 0.4 mm from the fracture surface (Fig. 19.9c). The

19.10 Charts of the elemental distribution on the fracture surface ofthe irradiated steel (a, b, d-i ¥ 1400; c ¥ 400; A, B, C, D, – pointswhere the local elemental composition is determined).

(a) Chlorine (b) Sulfur (c) Sulfur

(d) Copper (e) Iron (f) Oxygen

(g) Carbon (h) Chromium (i) Absorbed electrons

AB

C

D


results of this work show that the saturation of the measurement channelsurface with carbon up to 20 at.% is observed in the layer of 30–100 mmdeep, resulting in the brittle fracture. The carbonisation effect leds to generationof the internal stresses which may be sufficient to initiate and propagatestress corrosion cracking.

The properties of one or other section of the measurement channel dependon its location: the farther off the core bottom, the greater is the carbonsaturation and the thickness of the brittle layer (p. 2.3–2.4). It evidentlyrelates to the considerably variable coolant density along the axis. The loweris the density (at a high fluence), the greater is the depth of the corrosionfailure (Table 19.1).

The coolant density in the lower core part of the reactor under operationis 0.735 g/cm3. In this region, only intergranular corrosion is observed on themeasurement channel. The steam content increases approaching the corecenter, the coolant density decreases up to 0.375 g/cm3, and intergranularcorrosion and corrosion cracking achieves 1 mm depth. Futher decreasing ofthe coolant density to 0.304 g/cm3 at the level of the core top is followed bya sharp increase in the corrosion depth. It appears to be related to the elevatedoxygen concentration in steam, carbonisation of the tube surface, possibilityto concentrate the chloride impurities, when evaporating multiple waterportions, and to facilitate oxygen access.

The neutron fluence varies along the axis of the measurement channelfrom 2.1 ¥ 109 cm–2 to 5.1 ¥ 1021 cm–2, however all these areas are revealedto have corrosion failures. As seen in Table 19.1, the corrosion depth of themeasurement channel material is not a strong function of the fluence value.

The results lead to the following conclusions:

∑ Failure occurs exceptionally along the grain boundaries.∑ Ti segregates and the Cr and Ni content decreases on the grain boundaries.∑ There are harmful corrosive elements such as chlorine, sulfur and copper

on the failure surface.∑ Segregation thickness on the failure surface is about 1–1.5 mm; in any

case the segregation has practically disappeared after ion etching at adepth ~3 mm (Fig. 19.11).

The elemental charts give the qualitative representation of the elementaldistribution. The quantitative information can be provided from analysingthe local elemental composition in the separate points with an electron probediameter of 0.1–0.3 mm (in this study in the points ‘ACE’ in Fig. 19.10,Table 19.2).

The same table presents for comparison the integral measurements of theelemental composition in the raster 200 ¥ 200 mm, the surface of whichcomprises the examined fracture pattern. The results of the local measurementsare in a good agreement with the qualitative information provided by the


elemental charts. As shown in the table, only chlorine is rather uniformlydistributed on the failure surface. The concentration of the other elementsvaries in a wide range. The references [1–7] and the findings permit us toreveal three groups of factors: (1) chief factors, (2) governing factors and (3)related factors.

19.4.1 Chief factors

Tensile stresses

The cracks have been observed only in the upper part of the tube in thevicinity of the weld (Fig. 19.3a), where the tensile stresses, according to thework results, can reach 300 MPa. As the cracks have not been observedbelow, in the base metal, where the neutron fluence was increased, one mayconsider that the tensile stresses are of importance in the radiation-inducedintergranular corrosion cracking and prevail over the radiation-inducedprocesses of segregation/depletion.

19.11 Charts of sulfur and iron distribution on the failure surface ofthe irradiated steel after the ion etching at a depth in the order of3 mcm ¥ 1500.

(a) Sulfur (b) Iron (c) Absorbed electrons

Table 19.2 Local elemental composition of the failure surface(‘A’, ‘B’, ‘C’ and ‘D’ refer to Fig. 19.10)

Element Content at %

‘A’ ‘B’ ‘C’ ‘D’ Raster

Fe 4.0 12.2 1.7 7.4 2.4Cr 0.5 2.8 0.8 1.6 0.6Ni 3.8 5.1 1.4 6.5 0.3Ni 1.0 0.6 0.6 1.2 0C 75.9 61.7 82.5 47.0 91.5O 13.0 13.1 6.3 30.8 3.9Cl 0.5 0.4 0.6 0.3 0.3S 0.4 2.7 0.1 0 0.4Cu 0.8 1.4 6.2 3.0 0.5Ca 0 0 0 2.3 0


Sensitisation

The weld adjacent zone is heated to 550–800 ∞C under welding, whichresults in the precipitation of the large Me23C and/or Me6C type carbides ofsome hundreds nanometers size (Fig. 19.6), containing about 13% chromium,42% iron, 39% titanium and 6% nickel, on the grain boundaries and sub-boundary areas. As a result, the grain boundaries are depleted in chrome.Secondary source of depletion is the purely thermal low-temperaturesensitisation in the temperature range of 250–350 ∞C. As estimated, the timeneeded for its proceeding at the temperature of 285 ∞C is 12 years [6] that istwo and a half times lower than the service life of this tube. And the third,probably the most powerful source of the grain boundary depletion inchromium, is caused by the radiation-induced processes: (a) rising of theavailable carbides; (b) radiation-induced depletion of the grain boundary inchromium; (c) precipitation of the fine-dispersed phases containing chromium.In the work [7] the segregation processes are analysed, which occur on thegrain boundaries of steel 304 in the initial state and after irradiation at thetemperature of 300 ∞C (Table 19.3).

As indicated in Table 19.3, the results of the energo-dispersion analysis(EDA) show that at the initial state, a notable segregation of chromium andsilicon and a strong segregation of molybdenum and phosphorus are observedon the grain boundaries. According to the results of the Auger electronic

Table 19.3 Irradiation effect on the chemical composition modification of the grainbody and sub-boundary zones of steel 304 at the temperature of 300 ∞C

Element EDA AugerTEM electron(mass analysis%) (at.%)Initial 5 ¥ 1021 Initial 5 ¥ 1021

state n/cm2, state n/cm2

grainboundary

Grain Grain Grain Grain Grain Grainbody boundary body boundary body boundary

Fe Base 62–64 61 69.39 74.9 68.4 60.6Cr 18.54 24–25 16 19.58 16.4 18.7 14.2Ni 8.28 9 15–16 7.75 7.8 8.43 14.1Mo 0.32 1.7–1.9 1 0.18 n/d 1.53 0.2Mn 1.52 n/d n/d n/d n/d n/d n/dSi 0.55 1 4.2–4.5 1.08 n/d 1.25 8.3C 0.069 n/d n/d n/d n/d n/d n/dP 0.023 0.8 1.1 0.04 0.7 0.06 1.9S 0.021 n/d n/d n/d n/d n/d n/d

EDA – energo-dispersion analysisn/d – not determined


analysis, the segregation of chrome is not confirmed. However, the data ofthe energo-dispersion analysis show that after irradiation, the grain boundariesare depleted in chromium, molybdenum and enriched in nickel, phosphorusand especially silicon. The results of the energo-dispersion analysis andAuger spectroscopy are in qualitative agreement. The results from bothinvestigation methods are in a good agreement for chrome. But oppositeresults are obtained for nickel: in this study the grain boundary is observedto be depleted in nickel up to zero, where as, in work [7] the grain boundaryis observed to be enriched in nickel.

Thus, one may conclude that the radiation-induced segregation on thegrain boundaries is a second significant factor affecting radiation-inducedintergranular corrosion cracking.

19.4.2 Governing factors

Concentration of oxygen

A steam-water mixture as high as 0.2–0.5 mg/l in water and up to 20–50 mg/l in steam due to radiolysis [1, 2] favors increasing corrosion potential withcorrosion rates in intergranular channel peaks.

Chlorine precipitation on the boundaries

The coolant is probably the principal source of chlorine. In this case thepossibility to accumulate chlorine at the steam-water interface favors theinitiation of chloride cracking even when the ion-chlorine composition is<0,1 mg/l. The transmutation of sulfur in chlorine and the possible chlorinerelease from the disintegrated manganese sulfides is an additional althoughprobably not as efficient source of chlorine [5].

Grain boundary poisoning with sulfur

Sulfur, acting as a poison during electro-chemical processes and decreasingthe repassivation rate after the oxide film ruptures, favors the initiation of thelocal anodic processes. The coolant and sulfur, contained in the steel andunlinked in the sulfides, are probably the principal source of the sulfur. Theradiation-induced disintegration of MnS can be an additional source of sulfuras indicated in reference [5]. In several points of the degradation surface thesulfur content achieves 2.7%.

Copper effect

The two valent copper ions are known [1, 2] to favor chloride corrosioncracking. Besides ~0.3% copper content in the steel, copper is present in the


coolant (up to 0.03 mg/l). This quantity is sufficient to achieve the copperconcentration of 6.2% on several points of the degradation surface and in thevicinity of the boundary zones.

19.4.3 Related factors

Saturation

Saturation of the grain boundaries with carbon and oxygen when theirconcentration achieves 50 at %.

Effect of the steam-water interface

Under operation the investigated part of the tube was alternatively cooledwith water at a temperature of 5–7 ∞C below the boiling point and with asteam-water mixture at saturation temperature. Thus, this tube section wassituated at the steam-water interface in the area most sensitive for corrosioncracking, due to chloride accumulation as the result of multiple drying andeasy acces of oxygen.

The analysis of the findings and reference data, not claiming to be original,allows us to represent the mechanism of the radiation-induced intergranularcorrosion cracking of the austenitic stainless steel as follows:

∑ Initially, when the oxide film of metal has cracks, voids and other defects,general corrosion is going on at a relatively high rate. The low solublecorrosion products remedy gradually the oxide film defects on the grains,decreasing the area and number of the anodic regions. The anode locationmoves progressively to the sensitised grain boundaries. This period maybe considered as incubation one [1, 2].

∑ Furthermore the process is going at low rates corresponding to anodedissolution in the passive state. The maximum rate is observed on areaswith a minimum chrome concentration resulting in pitting. As theintergranular cracks become deeper, the corrosion products inhibit accessof new steam-water portions to the anodic areas and metal ion removalin the opposite direction. At the same time the access of cathodic depolariser(oxygen) is inhibited, which results in displacement of the cathodicprocess on the walls of the intergranular channels in the vicinity of thesurface [1, 2].

∑ Gradually the poisoning of the grain boundary with sulfur and chlorine,arising from the coolant and MnS radiation disintegration, gathers strength.The copper starts to precipitate on the surfaces of the formed intergranularcracks. The complex clusters are formed on the basis of sulfur andcopper also including iron, chrome, nickel, carbon and oxygen.

The radiation-induced processes of grain boundary depletion in chromium


and nickel and simultaneous enrichment in titanium, phosphorus and siliconare also important. The chemical composition of the sub-boundary regionsdiffers more and more from the chemical composition of the matrix. Thenucleus of the intergranular cracks grows in length and width, affected byresidual tensile weld stresses, achieving at the final stage the lentgh of sometens of millimeters and fusing in the one main transversal crack. Theintergranular corrosion cracking is completed by the transversal rupture ofthe tube.

As the superficial layer is saturated with carbon, the stresses arise as aresult of local lattice distortion, change of coefficient of temperature expansionand so on. In this case the longitudinal cracks appear on the outer tubesurface and the below-surface corrosion is initiated in the superficial layer.The intergranular corrosion moves gradually in the corrosion cracking. Acidityof the crack peaks environment increases due to the radiolitic formation andaccumulation of nitric acid. The cleavage effect, arising from the complexclusters (or phases) on the basis of sulfur and copper, introduces an additionalcontribution in the further opening and growth of the intergranular cracks[2].

19.5 Conclusions

1. After 25 years of operation an important corrosion of the outer surfaceof the measurement channel tube has occurred.

2. The corrosion failure of the tube is evidently a function of coolant densityand entailed carbonisation depth of the superficial layer.

3. The neutron irradiation effect is not a main factor in increasing the basemetal sensitivity of the measurement channel to the intergranular corrosionand corrosion cracking.

4. The main causes of the intergranular corrosion cracking are: (1) residualtensile weld stresses; (2) austenite sensitisation in the vicinity of weldduring the welding; (3) radiation-induced low-temperature sensitisationunder irradiation; (4) radiation-induced segregation and depletion on thegrain boundaries.

5. The chemical composition of the grain boundaries and ajacent areas of1 mm wide after irradiation has no common concern with the initialcomposition of the matrix. Strong depletion in chromium and nickel andenrichment in titanium occur. The width of these zones is differentfollowing the chemical elements and, as a rule, varies from 0.2 to 1 mm.

6. The wide application of the Auger spectrometry with constructing theelemental charts of the failure surface, raster and punctual elementalspectra permitted us to pool all the data using the quantitative analysisof the element-structure features directly on the grain boundaries.


19.6 References

1. I.I. Vasilenko, R.K. Melekhov, ‘Steel Corrosion Cracking’ Academie of Sciences ofUkrain, Physico-mechanical Institute, ‘Naukova Dumka’, 1977.

2. V.P. Pogodin, V.L. Bogoyavlensky, V.P. Sentyurev. ‘Intergranular Corrosion and CorrosionCracking of Stainless Steels in Water Environments’ Atomizdat, Moscow, 1970.

3. G.G. Ulig, R.Y. Revi, ‘Corrosion and corrosion control. Introduction to corrosionscience and engineering’. Chemistry, 1989.

4. F.A. Garner, L.R. Greenwood, H.M. Chung. ‘Irradiation-induced instability of MnSprecipitates and its possible contribution to IASCC in light water reactors’. Proceedingsof the of 8th International Symposium on Environmental Degradation of Materials inNuclear Power Systems – Water Reactors, August 10–14, 1997, Florida, USA, pp.857–860.

5. Yu. D. Goncharenko, V.A. Kazakov, G.V. Filyakin, V.K. Shamardin et al. ‘Study ofmaterial properties of an assembly wrapper (Cr-18Ni-10Ti steel) of the VK-50 reactorafter 30 years of operation. Proceedings of the 6th Inter-branch Conference on theReactor Material Study, September 11–15, 2000, Dimitrovgrad, SSC RF RIAR, Russia,pp. 49–68.

6. A.A. Nazarov. ‘Steel sensibility to intergranular cracking and modern method of itsestimation’, Review, ZNIIKM Prometey, 1991.

7. J.F. Williams, P. Spellward, J. Walmsley, T.R. Mager, M. Koyama, H. Mimaki, I.Suzuki. ‘Microstructural effects in austenitic stainless steel materials irradiated in apressurized water reactor’. Proceedings of the 8th International Symposium onEnvironmental Degradation of Materials in Nuclear Power Systems – Water Reactors,August 10–14, 1997, Florida, USA, pp. 812–822.

289

20.1 Introduction

In accordance with international trends, the life-cycle prolongation of thenuclear reactors type VVER-440/213 at Paks NPP is fundamental to Hungarianenergy policy. Recent investigations of this issue [1] have shown an additional20–25 years of operation is possible over the 30 years predicted earlier. Thispossibility may be essential in the development of the national economy.

In the enhancement of the power capacity and/or a possible extension ofthe life-cycle, the contamination and corrosion state of the steam generatorsof the VVER 440/213 type pressurized water reactors are considered to beone of the decisive factors [1, 2]. During the construction of the abovereactor blocks replacement of the steam generators was not taken intoconsideration; therefore, the replacement of even one steam generator couldresult in a considerable production loss and extreme investment cost. Someyears ago, evaluating the primary and secondary side water chemistry dataand the corrosion effects of the chemical decontamination procedures performedat NPP Paks, an intense demand emerged to perform overall estimation onthe corrosion state of the steam generators, i.e. to prepare a so-called ‘corrosionmap’ [3]. This ‘corrosion map’ takes a survey of the corrosion features ofthe heat exchanger tubes made of austenitic stainless steel in the steamgenerators.

Owing to the fact that there are no investigation methods available for thein-situ monitoring of the inner and outer surfaces of heat exchanger tubes, aresearch project based on sampling as well as on ex-situ electrochemical andsurface analytical measurements has been elaborated. The preliminary corrosionstudies were started in 2000, and so far a systematic investigation of thecorrosion state of 16 steel samples originating from different steam generatorsof Paks NPP has been performed [4–5].

20Comprehensive investigation of the

corrosion state of the heat exchangertubes of steam generators

K. V A R G A, Z . N É M E T H, A . S Z A B Ó, K . R A D Ó,D. O R AV E T Z and K. É. M A K Ó, University of Veszprém,

Hungary, Z. H O M O N N AY, E. K U Z M A N N andS. S T I C H L E U T N E R, Eötvös Loránd University, Hungary

and P. T I L K Y, J. S C H U N K and G. P AT E K,Paks Nuclear Power Plant Ltd., Hungary


The aim of this chapter is to summarize the experimental findings on thesurface characteristics (passivity, morphology, chemical composition andstructure, phase composition) of the steel specimens obtained byelectrochemical (voltammetry) and surface analytical (SEM-EDX, CEMS,XRD) methods.


20.2.1 Preparation of the samples

The experiments have been performed on 16 austenitic stainless steel specimens(type: 08X18H10T (GOST 5632-61), outer diameter: 16 mm, average wallthickness: 1.6 mm) originating from different steam generators of Paks NPP.The main characteristics of the samples are given in Table 20.1. The surfacedecontamination procedure (if any) of the above tubes was carried out atPaks NPP according to the AP-CITROX technology [6–7].

From the tube samples having a length of 41–548 mm, specimens of 20mm length were cut for the voltammetric, SEM-EDX, CEMS and XRDstudies. The tube pieces were cut into two halves with the help of a sawalong their diameter then – only for voltammetric and CEMS studies –planed gently. In order to keep the original oxide layer on the specimens nodegreasing procedure was applied on the surfaces. The corrosion propertiesof the stainless steel samples prepared via the above technique were studiedmainly on the inner surfaces, which were formerly in contact with the primarycoolant, by electrochemical, surface spectroscopic and microscopic methods.

Table 20.1 Main characteristics of the specimens

Number of Year of Year of Year of Length of thesample decontamination cutting investigation tube (cm)

1 2001 2002 2002 14.82 1996, 1997 2001 2001 34.73 1996 2000 2000 12.64 2001 2002 2002 24.15 2001 2002 2002 27.06 2001 2002 2002 14.87 – 1999 2001 4.38 2001 2002 2002 14.79 2001 2001 2001 30.0

10 1993 2000 2000 12.011 – 2000 2000 13.712 – 2001 2001 54.813 – 1998 2001 4.114 – 2000 2000 12.615 – 2001 2001 40.516 Inactive reference 1982 2001 4.65

sample

Corrosion state of heat exchanger tubes 291

20.2.2 Investigation of the corrosion state of tubespecimens by voltammetry

The passivity of the tube samples was studied by the potentiostatic polarizationmethod. The experiments were carried out by the means of a VoltaLab 40(RADIOMETER) type electrochemical measuring system controlled by aPC. To perform these investigations a special electrochemical cell wasdeveloped. In the course of potentiostatic polarization experiments the potential(E) of the specimen (working electrode) was continuously shifted towardsanodic direction at a constant rate of 10 mV·min–1 and the current density (i)related to the inner surface area of the specimen was recorded. Themeasurements were carried out in boric acid solution (c = 12 g·dm–3) inargon gas atmosphere (99.999 v/v % Ar). The schematic of the measuringsystem, the detailed experimental procedure and the determination of thecorrosion parameters (such as corrosion potential (Ec), corrosion currentdensity (ic), and corrosion rate (vc)) derived by the so-called Stern methodare described in our earlier work [7]. In addition, the average value of thepassivity current density (ipass) was determined in the potential range of0.60–0.80 V. The electrode potential values quoted in this paper are given onthe saturated calomel electrode (SCE) scale.

20.2.3 Study of the surfaces by SEM-EDX method

The morphology and chemical composition of the oxide layer developed onthe inner surfaces of the 16 stainless steel specimens were studied by scanningelectron microscopy (SEM), equipped with an energy dispersive X-raymicroanalyzer (EDX) (Type: JEOL JSM-50A, controlled with Röntec EDR288 software). In case of each specimen having a length of 20 mm, twodifferent surface areas were studied by making use of the combined SEM-EDX equipment. The comparative evaluation of surface morphology wasperformed by analyzing the SEM micrographs obtained at two differentmagnifications, M = 3000 and M = 1000 respectively. The chemical compositionof the sample surfaces was determined at least on two different areas of1 mm2 by EDX method.

20.2.4 CEMS analysis of the surface oxide layers

The 16 samples cut out from the heat exchanger tubes of the steam generatorswere measured by Conversion Electron Mössbauer Spectroscopy (CEMS) inorder to determine the phase composition of the surface oxide layers. TheCEMS method is based on the recoilless nuclear absorption of the 14.4 keVg-rays of 57Fe in the surface oxide layer, followed by the emission of conversionelectrons upon de-excitation of the 57Fe nucleus. The surface selective


measurement is guaranteed by the detection of low energy conversion electrons(<7 keV).

The CEMS spectra were recorded at room temperature with a conventionalMössbauer spectrometer (Wissel) in constant acceleration mode. Theconversion electrons were detected with a constant-flow type proportionalcounter specially designed for CEMS technique (Ranger). The counter gaswas a mixture of 96% He and 4% methane. A 57Co(Rh) source provided theg-rays. Calibration was done by measuring an a-Fe foil in transmissionmode, which is the reference of the isomer shifts given in this chapter.

Due to the absorption of the conversion electrons, information can beobtained from the ~300 nm thick surface layer of the samples. The error ofthe determination of the phase composition is ±5%; however, this does notinclude the depth dependent sensitivity of the detection process. Thereforethe phase composition (i.e., iron content of the different phases) obtained isalways an integral composition of the 300 nm thick surface layer, whichcontains an exponential weighing according to the depth of the particularphase.

20.2.5 XRD phase analysis

XRD as a standard technique in metallurgy was applied to show the formationof the a-phase in the austenitic steel. The XRD measurements were carriedout with a PHILIPS PW3710 type diffractometer (CuKa X-rays, voltage: 40kV, current: 40mA, goniometer speed: 0.02∞/s). Both the inner and outersurfaces were analyzed for almost all samples.

Taking into account the absorption characteristics of the CuKa X-rays(~8 keV), the phase analysis refers to a ~30 mm thick surface layer, which is100 times larger than in the case of the CEMS method. The exact quantitativephase analysis is hindered by depth dependent sensitivity and, in addition,by texture of the samples, possible existence of amorphous components,uneven surface, etc. The relative error of the determination of crystallinephases may be estimated as ±5%.

Non-crystalline phases, like amorphous Fe-oxides, -oxihidroxides cannotbe analyzed. Their presence is indicated mostly by line broadening and anincrease in the baseline intensity.



Some illustrative potentiostatic polarization curves of the stainless steelspecimens originating from different steam generators of the Paks NPP


measured in boric acid solution can be seen in Fig. 20.1. All the I vs. Ecurves were evaluated and the Ec, ic, vc and ipass values for all specimenswere calculated. The main corrosion parameters are summarized in Table20.2.

As clearly seen from the corrosion data summarized in Table 20.2, theinner surfaces of the samples have a passive character in a wide potentialinterval next to the corrosion potential. The calculated corrosion rate is verylow, and the corrosion current density does not exceed the value of ic = 4 ¥10–7 A·cm–2.

A careful inspection of the above data, however, reveals that the potentiostaticpolarization behavior of samples 2, 3 and 10 differs significantly from theothers. The average corrosion rate of their inner surface is beyond the valuesdetermined for the other samples (see Table 20.2). At this point it should beemphasized that the tube samples exhibiting unfavorable corrosion statewere decontaminated earlier – in some cases more than once – by the AP-CITROX technology at Paks NPP [6–7].

The tube sample 9 was also decontaminated, but immediately before thecutting procedure, so the parameters characterizing its passivity are stillfavorable. Similarly, the data derived from the potentiostatic polarizationcurves in Fig. 20.1 confirm that the corrosion parameters of the samplesdecontaminated in 2001 – about one year before cutting – are as good as theones not decontaminated earlier.

The average corrosion rates of the inner surfaces of samples 1, 4, 5, 6 and8 can be qualified similarly to the samples showing excellent corrosionfeatures (e.g., the inactive reference sample (16) and samples 7, 11, 12, 13,

0.80.60.40.20–0.2–0.4E/V

58

4

6

1

–5

–6

–7

–8

–9

Igi/A

(cm

–2)

20.1 Potentiostatic polarization curves measured at the inner surfaceof samples 1, 4, 5, 6 and 8 in boric acid solution (c = 12 g · dm–3).Scan rate: 10 mV · min–1.


ater reactors294

Table 20.2 Corrosion parameters determined from voltammetric curves

Determined Numbercorrosion ofparameters specimens

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Ec(mV) –82.8 –55.7 60.5 0.7 371.3 –64.8 143.7 307.6 –30.9 200.8 312.2 –115.7 119.4 –15.9 –68.4 –126.5

Ic(nA·cm–2) 35.0 300.4 335.0 37.7 37.4 37.5 40.7 60.2 49.3 160.0 25.0 70.4 33.4 – 63.3 35.1

ip 0.50 106.3 >1 0.83 0.38 <1 0.81 0.92 1.18 <1 <1 1.81 0.55 <1 1.86 0.65(mA · cm–2)

Vc 0.4 3.5 3.9 0.4 0.4 0.4 0.5 0.7 0.6 1.8 0.3 0.8 0.4 <1.8 0.7 0.4(mm · year–1)


14 and 15). The latter tube samples have never been decontaminated and sotheir average corrosion rates are extremely low (vc < 0,8 mm/year); evenbetter than the literature data which were measured for the stainless steeltype 08X18H10T (GOST5632-61) in water solutions at temperatures of 280–350 ∞C [8].

20.3.2 Study of the surfaces by SEM-EDX method

The morphology and chemical composition of the inner surfaces of thespecimens were studied by SEM-EDX method. The data measured haveproved that the samples 2, 3 and 10 which were decontaminated a few yearsago, have very similar surface characteristics. Some typical SEM-EDX resultsregarding this group of samples are illustrated by the findings gained in thecase of sample 3 (see Figs 20.2–20.3). The protective oxide layers formed onthe surfaces of the above samples are compact, thick (thickness > 1 mm),nevertheless contain many cracks and scattered deep failures. No presenceof any deposited crystals can be identified on the oxide films. The SEM-EDX data show that these oxide layers exhibit amorphous character and arerich in chromium and nickel. (The above statement is confirmed by theCEMS results, which provide evidence that amorphous iron-hydroxide(Fe(OH)3) is the dominant component in the passive layer.)

The results for the inner surface of sample 9 reveal that the surface layerfound on the freshly decontaminated sample is compact, covered mainly

20.2 SEM micrograph (3000¥) of the surface morphology ofsample 3.


with amorphous protective oxide layer, on which large crystals can be sparselyobserved. The size of some crystals exceeds the 10 mm value. The thicknessof the base oxide layer is below 0.5 mm, and the oxide film is contaminatedby a significant amount of manganese. As this sample was decontaminatedimmediately before cutting, it is possible that the manganese – probably inthe form of MnO2 – remained on the surface after the AP-CITROXdecontamination procedure.

On surfaces of the samples 7, 11, 12, 13, 14, 15, as well as of the inactivereference sample (16) there is a thin (thickness less than 0.5 mm) passivelayer with excellent protective characteristics. In the majority of the samplescracks cannot be identified and the surface consists of decisively crystallinephases (probably magnetite and hematite). Some illustrative results for theabove group are shown in Figs 20.4–20.5, which summarize the data measuredon sample 12.

In case of the samples (1, 4, 5, 6 and 8) which were decontaminated in2001 – about one year before sampling – a medium thick or thick (≥ 1 mm)compact layer showing basically amorphous structure can be observed. Thereare no crystalline deposits on the surfaces, and the chromium enrichment inthe various oxide layers is varied but significant. A little nickel enrichmentcan also be observed in the case of the most samples. It is to be noted that adeposit (probably Fe-oxide) with spongy structure covers the main part ofthe surface of sample 6.

20.3 EDX spectrum measured on the surface of sample 3.


20.4 SEM micrograph (3000¥) of the surface morphology of sample12.

20.5 EDX spectrum measured on the surface of sample 12.


20.3.3 Phase analysis of the surface oxide layers byCEMS

Figures 20.6–20.7 show 57Fe-CEMS spectra of some selected samples. Theresults of the phase analysis are summarized in Table 20.3 for all the 16samples.

It was found that on the inner surface of samples 7, 11, 13, 14, and 15 thedominant magnetic phases are magnetite and/or hematite (see Fig. 20.6 forsample 7). Magnetite is represented by two sextets in the Mössbauer spectradue to two different cationic sites for iron in the spinel structure, whilehematite can be described with one sextet only. The Mössbauer parametersobtained for these phases (especially for magnetite) were slightly differentfrom the literature data on the pure compounds, which was attributed to theeffect of Cr- and Ni-substitution.

Velocity (mm/s)–8.0 –4.0 0 4.0 8.0

13050000

13000000

12950000

12900000

12850000

12800000

Co

un

ts (

arb

itra

ry u

nit

s)

20.7 CEMS spectrum measured on the inner surface of sample 2.

20.6 CEMS spectrum measured on the inner surface of sample 7.

8.04.00–4.0–8.0Velocity (mm/s)

10580000

10560000

10540000

10520000

10500000

10480000

Co

un

ts (

arb

itra

ry u

nit

s)


Magnetite was missing on the inner surface of samples 9 and 12, while onsamples 2, 3, 10, and 16 magnetic phases could not be found at all withinexperimental uncertainty (Fig. 20.7).

The phase analysis of the samples, which were subjected to adecontamination process a year before this CEMS study (1, 4, 5, 6, and 8),reveals various amounts of magnetite besides the austenitic bulk steel in theupper 300 nm layer. In these samples the dominant magnetic phase isundoubtedly the magnetite, however, at the statistics of the spectra, the presenceof 5–10% hematite cannot be excluded.

In addition to the magnetic phases and/or the bulk steel, the presence ofanother paramagnetic phase was made obvious by the spectrum evaluationsin all samples. This phase could be evaluated by a doublet with parametersisomer shift, d = 0.35–0.45 mm/s and quadrupole splitting, D = 0.6–0.8 mm/s. This may allow an assignment to lepidocrocite (g-FeO(OH)) or amorphousiron(III)-hydroxide (Fe(OH)3). Since XRD did not show anything whichmight correspond to this phase (i.e. the crystalline lepidocrocite), we assignedit to amorphous Fe(OH)3.

The presence of the a-phase (ferrite, martensite) could not be confirmedin the upper 300 nm layer besides the oxide phases by CEMS. Interestingly,however, when the surface was cleaned from the oxide layer (inner and outersurface of sample 3, outer surface of sample 14), the ferrite phase showedup. The relative amount of this phase (i.e., that of iron contained in thephase) was very significant, namely 60% and 50% on the outer and inner

Table 20.3 Percentage phase distribution on the specimens’ surfaces determinedfrom the CEMS spectra

Number of Phases (m/m%)sample

Steel (austenite) g-FeOOH/Fe(OH)3 Fe3O4 a-Fe2O3

1 50 32 18 –2 29 71 – –3 38 62 – –4 20 36 44 –5 14 40 46 –6 42 42 16 –7 25 36 24 158 13 63 24 –9 57 23 – 20

10 54 46 – –11 53 13 13 2112 50 27 – 2313 20 18 15 4714 21 17 27 3515 22 19 22 3716 93 7 – –


surface of sample 3, respectively, and 22% on the outer surface of sample 14.The appearance of the a-phase can be attributed to various mechanical effects(cold rolling, shaping, polishing), high temperature effects such as welding,and possibly irradiation in the reactor, etc.

It can be assumed that the phase transformation observed at the surfacemight have spread over into deeper regions of the bulk steel. This is supportedby the XRD results in the next chapter. Since the penetration depth of theapplied X-rays is about two orders of magnitude higher than that of theconversion electrons, one can conclude that the g to g phase transformationis a bulk phenomenon in the samples studied.

20.3.4 XRD analysis

The inner and outer surfaces of all the samples (except 3, 10, 11, 14) wereinvestigated by XRD. The primary goal of the XRD studies was to detect thea-phase, if there is any, at the inner and outer surfaces of the samples. In thegroup of the investigated radioactive samples, the ferrite phase couldundoubtedly be shown on the inner surface of samples 1, 4, 5, and 12, as wellas on the outer surface of sample 13. It is remarkable that the presence of thea-phase was convincingly shown on both the inner and outer surface of theinactive reference steel sample (16), too. Because of the partial overlappingof the main XRD reflections of the austenite and ferrite phases, the presenceof ferrite in the other samples cannot be excluded up to ~7%.

20.4 Conclusions

In the framework of the comprehensive investigation of the general corrosionstate of heat exchanger tubes, 16 samples originating from different steamgenerators of Paks NPP have been studied. The main characteristics of thesamples are summarized in Table 20.4. From the electrochemical (voltammetric)and surface analytical (SEM-EDX, CEMS, XRD) experiments the followinggeneral statements can be made:

∑ The samples studied – based on their general corrosion state (corrosionrate, thickness as well as chemical and phase compositions of the protectiveoxide layer) – may be classified into three groups.

∑ Comparing the phase compositions of the inner (primary circuit side)and outer (secondary circuit side) surfaces, one can conclude that despitethe targeted reductive heat carrier medium, magnetite is very rarelyfound in the protective oxide layer on the inner surfaces. In contrast,magnetite is a dominant phase at the secondary circuit side. (The lattercan be explained if one takes into account that the steel used in thesecondary circuit is basically carbon steel.) A common phase component

Corrosion state of heat exchanger tubes

301

Table 20.4 Main corrosion and metallographic characteristics of the steel specimens studied

Number Average Thickness (mm) CEMS phase Presence of the Remark on Year ofof sample corrosion of the oxide distribution (%) a-phase (ferrite, decontamination investigation

rate of the layer formed on the inner martensite)* oninner surface on the inner surface the outer surface(mm/year) surface

1 ª 0.5 £ 0.5 Fe(OH)3 – 32 + (XRD, 19%) – (XRD) Decontaminated 2002Fe3O4 – 18 (2001)austenite – 50

2 ª 3.5 >1 Fe(OH3)3 – 71 – (XRD) –(XRD) Decontaminated 2001austenite – 29 (1996, 1997)

3 ª 3.9 >1 Fe(OH)3 – 62 + (CEMS, 50%) +(CEMS, 61%) Decontaminated 2000austenite – 38 (XRD) (1996)

4 ª 0.4 >1 Fe(OH)3 – 36 +(XRD, 11%) –(XRD) Decontaminated 2002Fe3O4 – 44 (2001)austenite – 20

5 ª 0.4 >1 Fe(OH)3 – 40 +(XRD, 6%) –(XRD) Decontaminated 2002Fe3O4 – 47 (2001)austenite – 13

6 ª 0.4 >1 (Feoxide Fe(OH)3 – 42 –(XRD) –(XRD) Decontaminated 2002deposits) Fe3O4 – 16 (2001)

austenite – 42

7 ª 0.5 <0.5 Fe(OH)3 – 36 –(XRD) –(XRD) – 2001Fe3O4 – 24Fe2O3 – 15austenite – 25


ater reactors302

8 ª 0.6 >1 Fe(OH)3 – 63 –(XRD) –(XRD) Decontaminated 2002Fe3O4 – 24 (2001)austenite – 13

9 ª 0.6 <0.5 Fe(OH)3 – 23 –(XRD) –(XRD) Decontaminated 2001Fe2O3 – 20 (2001)austenite – 57

10 ª 1.8 >1 Fe(OH)3 – 46 NA NA Decontaminated 2000austenite – 54 (1993)

11 ª 0.3 <0.5 Fe(OH)3 – 13 NA NA – 2000Fe3O4 – 13Fe2O3 – 21austenite – 53

12 ª 0.8 <0.5 Fe(OH)3 – 27 +(XRD, 7%) –(XRD) – 2001Fe2O3 – 23austenite – 50

13 ª0.4 £0.5 Fe(OH)3 – 18 –(XRD) +(XRD, 23%) – 2001Fe3O4 – 5Fe2O3 – 47austenite – 20

Table 20.4 Continued



Corrosion state of heat exchanger tubes

303

Table 20.4 Continued



14 < 1.8 <0.5 Fe(OH)3 – 18 NA +(CEMS, 22%) – 2000Fe3O4 – 26Fe2O3 – 35austenite – 21

15 ª 0.7 <0.5 Fe(OH)3 – 19 –(XRD) –(XRD) – 2001Fe3O4 – 22Fe2O3 – 37austenite – 22

16 ª 0.4 <0.5 Fe(OH)3 – 7 +(XRD, 13%) +(XRD, 7%) Inactive 2000austenite – 93 reference sample

NA: not available; – : not detectable; + : detectable


in all investigated samples is the amorphous Fe(OH)3. This is mostcharacteristic on the inner surface of the samples (2, 3, and 10) whichwere decontaminated earlier by the AP-CITROX procedure resulting ina weaker corrosion resistance of the steel.

∑ The effect of chemical decontamination on the corrosion state of theinner surface of the samples can be summarized as follows: All thesamples classified to the group having less favorable corrosion statewere decontaminated earlier by the AP-CITROX technology. During theapplication of the AP-CITROX method in the plant environment thechemical treatment of the steel surface is inhomogeneous, the quasiequilibration dissolution of the iron content from the surface oxide layerduring the oxalic-acid – citric-acid treatment cannot assure. In theknowledge of the technological parameters (0.4 cm3/cm2 decontaminationsolution to treated steel surface area ratio, temperature not more than95 ∞C) and considering the solubility of the Fe(II)-oxalate in hot water(0.026g/100 cm3 water) [9], it is probable that during the chemical treatmenta considerable amount of Fe(II)-oxalate precipitates on the inner surfaceof the heat exchanger tubes. Moreover, a significant part of the Fe(II)-oxalate deposits cannot be eliminated even during the cleaning andpassivation steps of the AP-CITROX technology. After the decontaminationthe heat exchanger tubes are open to the air for a longer period of time,consequently, the inner surfaces are covered with an aqueous solution(in some cases with a mixture of steam and water) saturated in dissolvedO2. During this time period, a considerable amount of amorphous Fe(III)-hydroxides is formed from the Fe(II)-species in the surface region of theprotective oxide-layer. Therefore, a part of the iron originally bound inthe form of stable oxides (magnetite, spinel, hematite) has been transformedinto amorphous Fe(III)-hydroxides and remained on the surface oxidelayer as an undesired result of the decontamination technology. Theamorphous Fe(III)-hydroxide layer is however loosely bound (mobile)on the surface; consequently, this ‘hybrid’ structure of the amorphousand crystalline phases in the oxide layer may meaningfully influence theextent of the radioactive contamination and the amount of the corrosionproducts in the coolant of primary circuit.


This work was supported by the Paks NPP Co. Ltd. (Paks, Hungary), and theHungarian Science Foundation (OTKA Grant No. T 031971/2000).

20.6 References

1. T. Katona, S. Rátkai, Á. Bíró Jánosiné and Cs. Gorondi, Fizikai Szemle 11, 341(2001).


2. D. Bodansky, Nuclear energy, AIP Press Woodbury, New York, 1996.3. Gőzfejlesztő dekontaminálások kritikai értékelése és hőátadó csőminták korróziós

vizsgálata (Report in Hungarian), University of Veszprém, Reg. No: 01850-140,Veszprém, Hungary, 2000.

4. K. Varga, Z. Németh, A. Szabó, D. Oravetz, P. Tilky, J. Schunk, Magy Kém. Folyóirat108, 444, 2002.

5. Z. Homonnay, E. Kuzmann, S. K. Sticleutner, É. Makó, K. Varga, Z. Németh, A.Szabó, P. Tilky, J. Schunk and G. Patek, Magy Kém. Folyóirat 108, 449, 2002.

6. K. Varga, P. Baradlai, G. Hirschberg, Z. Németh, D. Oravetz, J. Schunk and P. Tilky,Electrochim. Acta 46, 3783, 2001.

7. K. Varga, Z. Németh, J. Somlai, I. Varga, R. Szánthó, J. Borszéki, P. Halmos, J.Schunk and P. Tilky, J. Radioanal. Nucl. Chem. 254 (3), 589, 2002.

8. V.V. Geraszimov and A. Sz. Monahov, A nukleáris technika anyagai, (in Hungarian)Műszaki Könyvkiadó, Budapest, 1981.

9. CRC Handbook of Chemistry and Physics (Editor: D. R. Lide), CRC Press, London,1994.

306

21.1 Introduction

Kori 1 is a pressurized water reactor (PWR) with two recirculating steamgenerators and an electrical capacity of 600Mwe. Each steam generator has3388 steam generator tubes which were expanded to a full depth by mechanicalroll expansion. It has been operated since 1978 and its steam generators werereplaced in 1998 because of an extensive degradation of the steam generatortube made of a low temperature mill annealed (LTMA) Alloy 600. The mainforms of the degradation of the tubings were pitting, primary water stresscorrosion cracking (PWSCC), outer diameter stress corrosion cracking(ODSCC) and intergranular attack (IGA).

To mitigate the degradation of the steam generator tubing in Kori 1, thepreheater made of copper alloy and the condenser were replaced in 1988 andchemical cleaning was applied in 1990 and secondary water chemistry wasimproved. Furthermore, to assure the integrity of the steam generator tubes,plugging, sleeving and finally a steam generator replacement were performed.

This chapter addresses the evolution trends of the retired Kori 1 steamgenerator tube degradation such as pitting, PWSCC, and ODSCC based onthe repair of the tubings and presents a failure analysis of the tubes extractedfrom Kori 1.

21.2 Experimental method

The numbers of steam generator tubes repaired via either plugging or sleevingin Kori 1 during each overhaul period were classified according to thedegradation mechanism. Statistical distributions of each failure mechanismwere plotted using the Weibull distribution, which was found to be the optimumfor characterizing the corrosion of steam generator tubes [1].

Failure analysis of the tubes extracted from Kori 1 was performed toclarify the degradation mechanism and to establish remedial actions. Theextracted tubes that were selected based on the eddy current test (ECT)

21Stress corrosion cracking of a Kori 1 retired

steam generator tube

H. P. K I M, S. S. H WA N G, D. J. K I M, J. S. K I M,Y. S. L I M and M. K. J O U N G, Korea Atomic Energy

Research Institute, Korea

Stress corrosion cracking 307

signal during in-service inspection (ISI) were transferred to a hot laboratoryat the Korea Atomic Energy Research Institute.

Non-destructive examinations covering a visual inspection and a laboratoryECT were carried out and then destructive examinations followed. Chemicalcompositions of the sludge on the tube sheet were analyzed by inductivelycoupled plasma atomic emission spectroscopy (ICP-AES) and ionchromatography (IC) and structure of the sludge on tube sheet was analyzedby a X-ray diffractometry (XRD). Chemical compositions of the corrosionproduct on the tube and fracture surface were analyzed by SEM-EDS andAuger electron spectroscopy (AES).


21.3.1 Pitting

The number of tubes repaired due to pitting as a function of an effective fullpower year (EFPY) is shown in Fig. 21.1: 419 tubes were plugged due topitting in 1985 for the first time since it began operation in 1978. Thenumber of tubes plugged due to pitting decreased for the next two fuel cyclesuntil 1988 and then increased until 1990 and then finally decreased. Tomitigate the pitting in the SG tubes, both the condenser and the preheaterwith copper as the major alloying element in the secondary side were replacedin 1988. Analysis of the sludge on the secondary side in 1985 showed thatthe sludge contained a large amount of copper. Ingress of sea water into the

EFPY140 12108642

Nu

mb

er o

f tu

bes

rep

aire

d d

ue

to p

itti

ng

0

100

200

300

400

500

Sea water ingress

Chemical cleaning in 1990

Material and condenserreplacement in 1988

21.1 Number of tubes repaired with EFPY for pitting.


steam generator through the condenser tube made of copper alloy occurreduntil 1988 and then dramatically reduced after the replacement of the condensermade of copper alloy with that of Ti alloy in 1988. Ingress of sea water ledto an accumulation of the chloride ion in the sludge on the tube sheet. Acombination of the chloride ion and oxidizing agents such as copper seemsto enhance pitting. Even though ingress of sea water into the steam generatorvirtually stopped and the material with a copper source in the secondary sidewas replaced in 1988, the pitting rate increased during the period from 1988to 1990, so a chemical cleaning of the secondary side steam generator wasperformed in 1990. After the chemical cleaning, the pitting rate decreaseduntil the steam generators were replaced in 1998.

A cumulative Weibull distribution showing the fraction failed versus timefor the pitting on the secondary side is shown in Fig. 21.2. The cumulativeWeibull distribution clearly shows that the pitting trend remained unchangedduring the period from 1985 to 1990 even though the condenser and preheateron the secondary side were replaced in 1988 and ingress of sea water hadalmost stopped after the condenser replacement. However, the pitting trendin the cumulative Weibull distribution changed after the chemical cleaningin 1990, suggesting that the chemical cleaning had removed the chloride ionand/or copper in the sludge.

Three tubes (R26C38, R28C53 and R36C45) were extracted in 1988 andone tube (R25C29) in 1992 because of the pitting in the secondary side. Across-sectional area showing a pit under a sludge pile is shown in Fig. 21.3.In Fig. 21.3, some pitting indicated by the solid arrow, has penetrated throughthe wall, leading to a leak of the primary water, while other pitting, indicated

EFPY252015105

0.1

Frac

tio

n o

f tu

be

30

Chemical cleaning

21.2 Fraction of tubes failed vs. time in Weibull plot for pitting.


by dotted arrow, has partially penetrated the tube. Schematic profile of sludgepile on top of tubesheet is shown on left hand side of the steam generatortube and pit position represented by closed circle is plotted with angle fromarbitrary reference point and height from top of tubesheet on right hand sideof the steam generator tube in Fig. 21.4.

All the pits were located within the sludge pile on the tube sheet (Fig.21.4), suggesting that the pitting had occurred due to the concentrated impuritiessuch as the copper compound and the chloride ion at the interface betweenthe tube and the sludge pile.

Corrosion products within a pit in the pulled out tubes can be classifiedinto two types based on their appearance. One is generally a dark corrosionproduct and the other is a layered corrosion product composed of metalliccopper bands within the oxide. The corrosion product is shown by a brightband. The bright image matches well with the copper mapping. The numbersof copper bands in a corrosion product in a pit were generally fewer than thenumber of outages, therefore it might be related to the number of outages.However, a copper band in a corrosion product in a pit could be formed ina static autoclave test in a laboratory. So, the formation of a copper band ina pit for an extracted tube seems to be independent of the number of outages.

A mechanism for the copper band formation is proposed. An anodic reactionproceeds at the inner surface of the pit. A cathodic reaction proceeds at thecorrosion product in the pit and on the free surface around the pit. Based onthe fact that metallic copper band in the corrosion product is found in the pit,it can be presumed that copper ion should migrate into the pit. However,migration of the positively charged copper ion into the pit might not befeasible because the inside of the pit has excess positive charge that is

21.3 Pits in extracted tube, through wall pit is indicated by solidarrow and partially through walled pits are indicated by dottedarrows on cross section of tube.

Primary side

Secondaryside


produced by dissolution of the inner surface of the pit. So, negatively chargedcopper complex, which might be copper chloride complex, is introduced toexplain the presence of metallic copper band. Formation of copper chloridecomplex is thermodynamically possible [2]. The negatively charged coppercomplex and chloride ion move into the pit from the free surface to satisfythe charge neutrality. The dominant cathodic reaction would be an oxygenreduction if the concentration of the negatively charged copper complex isless than the critical concentration, and the negatively charged copper complexwould hardly be reduced. So, the corrosion product without a copper band isformed in the pit. A cathodic reaction of the negatively charged coppercomplex in addition to the oxygen reduction would start at the surface of thecorrosion product surface in the pit and form the copper band, if theconcentration of the negatively charged copper complex reaches a criticalconcentration. The layered copper band in the corrosion product seems to beformed by a repetitive action of the above two processes.

An electron microprobe was used to analyze the polished cross section ofthe pit. The analysis shows small but widely distributed amounts of bothsulfur and lead in the corrosion product. Significantly there was no sodiumand essentially no chloride in the corrosion product. The major elementswere chromium and nickel. However, the nickel content is low throughout

191

158

315∞3.2

3.2

1154.5

182∞

2.5

115

178∞313

15110

168

111

1

41∞235∞

42∞5.43

TubesheetTubesheet

Profile ofsludge pile

S/G B Hot leg R 28C 53

Top of thetubesheet

Steam generator tube

360∞ 270∞ 180∞ 90∞ 0∞

21.4 Schematic profile of sludge pile on top of tubesheet is shownon left hand side of the steam generator tube and pit positionrepresented by closed circle is plotted with angle from arbitraryreference point and height from top of tubesheet on right hand sideof the steam generator tube.


the deposit. Average ratio of the chromium over the nickel in the corrosionproduct is 23.9, while that in Alloy 600 is 0.22. This fact indicates that thecorrosion product in the pit was enriched in chromium as compared to nickelrelative to the nominal composition of the Alloy 600. Many previous workshave suggested that chromium is enriched on the outer layer of the corrosionproduct for Alloy 600 in an acidic environment while nickel is enriched onthe outer layer in an alkaline environment relative to the chemical compositionof the Alloy 600 [3–5]. Therefore, the pitting in the extracted tube occurredin an acidic environment.

Analysis of the extracted tube (R25C29) in 1992 shows that the chemicalcleaning performed in 1990 removed the copper deposit on the free surfacebut could not remove the corrosion product within a pit, probably due to thelow accessibility of the chemical cleaning solution into the pit.

21.3.2 PWSCC

A cumulative Weibull distribution showing the fraction failed versus timefor a PWSCC near the tubesheet on the primary side is shown in Fig. 21.5.SG tubes were repaired due to PWSCC for the first time in 1990. The hot legtemperature of Kori 1 is 319 ∞C. The Weibull characteristic time and slopeis 25 years and 4.5, respectively. While the mean value of the Weibullcharacteristic time and slope of the other plants with a hot leg temperature(Thot) of 324–326 ∞C are 11.3 years and 4.3, respectively. The Weibullcharacteristic time for a PWSCC of Kori 1 with Thot = 319 ∞C is 25 years andconsiderably higher than that of the other plants with Thot = 324–326 ∞C.

EFPY101

1E-3

0.01

0.1

Slope = 4.5theta = 25

Frac

tio

n o

f tu

bes

21.5 Weibull plot for PWSCC of Kori 1 (retired).


This is consistent with the suggestion that a decrease in the temperatureincreases the value of the Weibull characteristic time. However, the Weibullslope for PWSCC of Kori 1 is almost the same as that of the other plants witha higher temperature, even though there was a suggestion that the Weibullslope increases with temperature. At present, it is unclear what caused sucha situation.

Two tubes (R11C45 and R16C35) were extracted because of PWSCCfrom Kori 1 in 1992. All the PWSCC in the extracted tubes was located atthe roll transition as indicated by the ECT during an ISI. Three axial crackswere found on the R11C45 tube where one crack had penetrated throughwall and the other two cracks had partially penetrated. Several circumferentialPWSCC was found within a band at the roll transition on the R16C35 tube.Maximum depth of the circumferential PWSCC is about 53%. Lead, sulfurand chloride detected by a wave dispersive spectroscopy (WDS) in the corrosionproduct in the primary side near the through wall cracked PWSCC seems tohave migrated from the secondary side through the through wall PWSCC.Primary water in Kori 1 has been controlled according to the EPRI guidelines.The low temperature mill annealed (LTMA) Alloy 600 is less resistant toPWSCC when compared to the high temperature mill annealed Alloy 600 orthermally treated Alloy 600 [6]. Residual stress at the roll transition is higherthan that at the explosive or hydraulic expansion transition [7]. So, the lowerPWSCC resistance of the LTMA Alloy 600 in Kori 1 may be attributed to themicrostructure with an intragranular carbide and the roll expansion whencompared to the other plants.

21.3.3 ODSCC

A tube (R27C34) was pulled out because of ODSCC in Kori 1 in 1994. Eventhough ODSCC or PWSCC at the expansion transition were not indicated bythe ECT during the ISI, it was presumed that ODSCC or PWSCC at theexpansion transition would be found upon a destructive analysis because ofthe high residual stress at the roll transition. However, the destructive analysisshows that all the ODSCC proceeded above the expansion transition withinthe sludge pile and no ODSCC was found at the roll expansion transition.Destructive examination of the tubes shows that the intergranular corrosionattack (IGA) is confined to near the top of the tubesheet (TTS) (TTS ±2 mm), IGA and ODSCC are observed in a region from TTS to TTS + 50mm, and ODSCC is found in a region from TTS + 50 mm to TTS + 72 mmas shown in Fig. 21.6. Maximum depth of the ODSCC was about 60% of thetube thickness at a location from TTS to TTS + 50 mm and 100% in a regionfrom TTS + 50 mm to TTS + 72 mm. ODSCC rate is fast just below the topof the sludge, suggesting that the impurities concentrate preferentially in thatregion where wetting and boiling occurs but that the secondary water may


not penetrate below that region because the deposit might adhere very tightlyto the tube for the secondary water to penetrate or because a steam phasealways exists once the height of the sludge pile reaches 70mm above theTTS. Scratch, acting as a stress riser was not observed in the tube.

The secondary water chemistry was subject to volatile treatment withammonia and hydrazine. The environment near an ODSCC was estimated tobe caustic based on a chemical analysis of the leachates from the tube surfaceas well as AES analysis and the hideout and return data. Litmus paper waspressed on to the deposit surface after the deposit surface was wetted withdistilled water. pH of the deposit surface was about 8.5. pH of the distilledwater was about 6 probably because the carbon dioxide was dissolved in thedistilled water, which clearly indicates that the leachate is caustic. Chromiumfraction over the major alloying elements, Cr/(Ni + Cr + Fe) is less than 0.05for the tube surface deposit while the fraction is about 0.15 for Alloy 600. pHcalculated based on the hideout and return data was about 9.5 at 300 ∞Cunder the assumption that the concentration factor is 107 and silica is presentin the deposit. So, the ODSCC proceeded in a caustic environment.

From the failure analysis of the extracted tubes, it can be assumed that thecrevice chemistry was an acidic and oxidizing environment because of ingress

21.6 ODSCC in R27C34 tube.

Top

Bottom

TTS

–10

+20

+50

+70

+76

+91


of sea water and copper compound accumulation in the steam generatorfrom 1978 to early 1990, and then it changed to a caustic and slightly oxidizingenvironment after early 1990. In 1988, ingress of sea water was stoppedbecause the condenser was replaced with one made of Ti alloy and in 1990chemical cleaning was performed to remove the sludge on the tubesheet.

21.4 Summary

A material and condenser replacement in the secondary side and a chemicalcleaning of the steam generator changed the Weibull distribution for thepitting. Ingress of sea water via the condenser into the steam generator andan accumualtion of chloride in the steam generator induced the pitting. Amechanism for copper band formation within the corrosion product in a pitwas proposed. Pitting seems to have occurred in an acidic and oxidizingenvironment from 1978 to early 1990. The Weibull characteristic time andslope for the PWSCC is 25 years and 4.5, respectively. Crack shape andlocation strongly depends on the tube location. Axial PWSCC was onlyobserved in the R16C35 tube and circumferential PWSCC was only observedin the R11C45 tube at the roll expansion transition. Some tubes that experiencedan extensive ODSCC rather than a PWSCC in the roll transition seemed tobe due to the impurities concentrated in the crevice which induces ODSCC,even though the stress in the roll transition of the primary side is higher thanthat in the secondary side. ODSCC seems to have occurred in a caustic andslightly oxidizing environment from early 1990 to 1998.

21.5 Acknowledgement

This chapter is based, in part, on the Steam Generator Project of the Mid andLong-Term Program financially supported by the MOST in Korea.

21.6 References

1. R.W. Staehle, J.A. Gorman, K.D. Strvropoulos and C.S. Welty, Jr, ‘Application ofStatistical Distributions to Characterizing and Predicting Corrosion of Tubing in Steamgenerator of Pressurized Water Reactors’ Life Prediction of Corrodible Structures,R.N. Parkins. ed, NACE, Houston, Texas (1994) p. 411.

2. W. Liu and D.C. McPhail, ‘The thermodynamic properties of copper chloride complexesand copper transport in magmatic hydrothermal solutions’, Chemical Geology, 221(2005) p. 21.

3. J.B. Lumsden, S.L. Jeanjaquet, J.P.N. Paine and A. Mcliree, ‘Mechanism andEffectiveness of Inhibitors for SCC in a Caustic Environment’, 7th InternationalSymposium on Environmental Degradation of Materials in Nuclear Power Systems –Water reactors, Breckenridge, Colorado (1995) p. 317.

4. C. Laire, G. Platbrood and J. Stubbe, ‘Characterization of the Secondary Side deposits


of Pulled Steam generator tubes’, 7th International Symposium on EnvironmentalDegradation of Materials in Nuclear Power Systems – Water reactors, Breckenridge,Colorado (1995) p. 387.

5. J.M. Boursier, M. Dupin, P. Gosset and Y. Rouillon, ‘Secondary Side Corrosion ofFrench PWR Steam Generator Tubing: Contribution of Surfaces Analyses to theUnderstanding of the Degradation Process’, 9th International Symposium onEnvironmental Degradation of Materials in Nuclear Power Systems – Water reactors,Newport Beach, California (1999) p. 555.

6. H.P. Kim, S.S. Hwang, Y.S. Lim and J.S. Kim, Metals and Materials, 7 (2001) p. 55.7. U.C. Kim, et al., ‘Failure Analysis of Pulled out Tube from Kori 1’, KAERI, 1995.

316

22.1 Introduction

The surfaces of heat exchanger tubes (outer diameter: 16 mm, average wallthickness: 1.6 mm) provided by Paks NPP Ltd, were treated by a version ofthe FRAMATOME CORD-UV technology in order to ascertain the efficiencyof oxide layer removal. (It should be emphasized at this point that thisversion of the CORD-UV procedure was developed for the in-core removalof various non-spinel type oxides such as hematite and amorphous Fe-oxides(-hydroxides), as well as low substituted magnetite.) In addition, modellingof the reactor restart-up period (during follow-up treatment with boric acidsolution) was also performed.

The CORD-UV technology for oxide layer removal was carried out in apilot-plant model system (Fig. 22.1) [1] under the following conditions: 13g·dm–3 boric acid solution containing 2–2.2 g·dm–3 oxalic acid was circulatedinside the heat exchanger tubes for a period of 30 hours. The flow rate was3.0 m/s and the solution temperature ranged between 85 and 95 ∞C. Every20 minutes 33% of the solution was changed with the same volume of13 g·dm–3 boric acid solution containing 2–2.2 g·dm–3 oxalic acid. Thus, thedissolution of the main elements (Fe, Cr, Ni) of the oxide layer proceededunder quasi-equilibrium conditions.

Following the dissolution experiments of the oxide layer, time dependenceof the chemical concentration of the corrosion products removed into the13 g·dm–3 boric acid solution during the reactor restart-up period (time period:30 hours, flow rate: 3.0 m/s, solution temperature: 70–95 ∞C) was also studiedby modelling.

In the course of the above chemical treatments the radioactivity, morphology,chemical and phase compositions of the inner surface of the sample tubeswere studied. The investigation methods and time scale are shown in Table22.1.

22A systematic study of the corrosion effectsof the FRAMATOME CORD-UV technology

K. R A D Ó, K. V A R G A, Z. N É M E T H, I. V A R G A,J. S O M L A I, D. O R AV E T Z and K. É. M A K Ó,

University of Veszprém, Hungary, Z. H O M O N N AY andE. K U Z M A N N, Eötvös Loránd University, Hungary,

J. B O R S Z É K I and P. H A L M O S, University of Veszprém,Hungary and P. T I L K Y and

J. S C H U N K, Paks Nuclear Power Plant Ltd., Hungary

The FRAMATOME CORD-UV technology 317

300

Ø10

Ø10

Ø13

Ø10

1

2

300

3

4

5

3

2

6

78

22.1 Pilot-plant model system.

Table 22.1 Time scale of the methods

æÆ æÆ Magnetite Follow up æÆæÆdissolution æÆæÆæÆ treatment

æÆ with oxalic with boricacid acid solution æÆ

1 2 3 4g-spectrometry ICP-OES g-spectrometry ICP-OES g-spectrometrySEM-EDX (Samples SEM-EDX (Samples SEM-EDXXRD from XRD from XRD

solution) solution)

Voltammetry Voltammetry VoltammetryCEMS CEMS

1 Purging valve2 Ball-end (Ø 12,7)3 Silicone tube (Ø 14/20)4 Flow-meter (Type of

water-glass: Kent KSSW)5 Investigated heat

exchanger tube (Ø 13/16)6 Priming and outlet pipe7 Centrifugal pump (Cole-

Parmer Type U-75225-15)8 Speed control (pump

drive)(The diameter values aregiven in millimetres)



22.2.1 Examinations by g-spectrometry

The intensity of the g-radiation emitted by radionuclides incorporated in thesurface oxide of the tube samples was determined by g-spectrometry beforeand after applications of the oxide layer removal process. For detection aHPGe semi-conductor detector (Camberra, type: 7100, area: 50 mm2, thickness:5 mm) attached to multi-channel analyzer (Oxford Instruments Inc., type:PCA-Multiport (8k)) was used.

22.2.2 Investigation of corrosion state of tube specimensby voltammetry

The passivity of the tube samples was studied by a potentiostatic polarizationmethod by using a VoltaLab 40 (RADIOMETER) type electrochemicalmeasuring system controlled by a PC. The measurements were carried out inboric acid solution (c = 12 g·dm–3) under argon gas atmosphere (99.999v/v% Ar). The measured parameters were corrosion potential (Ec), corrosioncurrent density (ic), and corrosion rate (vc). The electrode potential valuesquoted in this paper are given versus the saturated calomel electrode (SCE)scale.

22.2.3 Examinations by SEM-EDX methods

The morphology and chemical composition of the oxide layer that haddeveloped on the inner surfaces of the stainless steel specimens were studiedby scanning electron microscopy (SEM), equipped with an energy dispersiveX-ray microanalyzer (EDX) (Type: JEOL JSM-50A, controlled with Röntecsoftware). For each specimen, two different surface areas were studied bycombined SEM-EDX experiments. The comparative evaluation of themorphology of the surfaces was performed using SEM micrographs obtainedat two different magnifications, M = 3000 and M = 1000. The chemicalcomposition of each sample surface was analyzed by EDX in at least at twodifferent surface areas of 1 mm2.


Samples cut out from the heat exchanger tubes of the steam generators weremeasured by Conversion Electron Mössbauer Spectroscopy (CEMS) in orderto determine the phase composition of the surface oxide layers. To preparesamples for the Mössbauer measurements, 2 cm long pieces were cut outfrom the tubes and then halved in the axial direction. The halves were then


carefully pressed mechanically to make them flat while keeping the innersurface (primary circuit side) free from any damage. The CEMS spectrawere recorded at room temperature with a conventional Mössbauer spectrometer(Wissel) in constant acceleration voltage mode. The conversion electronswere detected with a constant-flow type proportional counter specially designedfor the CEMS technique (Ranger). The counter gas was a mixture of 96% Heand 4% methane. A 57Co(Rh) source provided the g-rays.


XRD was applied to show the formation of the a-phase in the austeniticsteel. The 2 cm long pieces of the tubes were cut into halves but were notshaped to become flat as for the CEMS measurements. The XRD measurementswere carried out with a PHILIPS PW3710 type diffractometer (CuKa X-rays, voltage: 40 kV, current: 40 mA, goniometer speed: 0.02∞/s). The exactquantitative phase analysis was hindered by depth dependent sensitivity and,in addition, by the texture of the samples, possible existence of amorphouscomponents, uneven surfaces, etc. Non-crystalline phases, as well as amorphousFe-oxides, -oxihidroxides could not be analyzed. Their presence is indicatedmostly by line broadening and as increase in the baseline intensity.

22.2.6 Investigation of the composition of solutions withICP-OES and gravimetrical methods

Samples were taken from the solutions used in the course of the decontaminationtreatment of the heat exchanger tubes by the CORD-UV technology as wellas in the course of the follow-up treatment with boric acid solution. UsingICP-OES, the concentrations of alloying components dissolved into thesolutions as a function of the time were determined. Additionally, the solutionsremoved from the circulation system were filtered, and the amount of dispersedcorrosion product was weighed after one week drying.


22.3.1 Examinations by g-spectrometry

The decontamination efficiency of the CORD-UV technology was determinedby g-spectrometry. The decontamination factors (DF) taken from the g-spectrometric measurements of the surface of the heat exchanger tubes areshown in Table 22.2. The decontamination efficiency of this oxide dissolvingversion of the CORD-UV procedure is extremely low (DFCORD-UV = 1.10∏1.34).It is clear that the procedure applied – as a consequence of the low kinetic ofthe chemical processes (mainly dissolution of spinel-type oxide and complex


formation) – has low efficiency for the dissolution of the inner oxide layercontaminated predominantly by radionuclides.

The average decontamination factors were determined to a great extent bythe decontamination factors of Co-radionuclides (58Co, 60Co) and 54Mn. Theresults shows that DF values of above-mentioned nuclides were extremelylow; it is known that the Co-nuclides (mainly 60Co) are incorporated in deepregions of the oxide layer; consequently, it is probable that the present versionof the CORD-UV technology is not an efficient procedure to remove themain part of the spinel-type oxide layer from the surface.


The passive state of the surface of tube sample was examined by a potentiostaticpolarization method. The samples were measured before and after the CORD-UV procedure and the follow-up treatment with boric acid solution. Anillustrative potentiostatic polarization curve of the inner surface of the samplemeasured in boric acid solution (c = 12 g·dm–3) can be seen in Fig. 22.2. Thecorrosion parameters such as corrosion potential (Ec), corrosion current density(ic), and corrosion rate (vc) were evaluated from the polarization curves byusing the Stern method [1]. The main corrosion parameters are summarizedin Table 22.3.

The corrosion parameters summarized in Table 22.3 reveal that the inner

Table 22.2 Intensity and DF values measured by g-spectrometry

Nuclides E I1 I2 DF (I1/I2)

60Co 1173.2 59528 56950 1.051332.5 52581 51167 1.03

58Co 810.8 385873 228514 1.68110mAg 657.7 1581* 301* 5.25

884.2 973* – –54Mn 834.8 513627 432022 1.1959Fe 1099.2 4654 1848 2.52

1291.6 2998 1520 1.9795Nb 765.8 21259 8296 2.56124Sb 602.7 3629* 1327* 2.73

1691 736 472 1.5695Zr 756.9 1776* 684* 2.26

724.2 1929* 556* 3.47

Sum total 1051144 783356 1.34

Where:E = Emitted g-energy (keV)I1 = Intensity before the oxide dissolution (imp/2000s)I2 = Intensity after the oxide dissolution (imp/2000s)


surfaces of the samples had a passive character in a wide potential intervalnear to the corrosion potential. The calculated corrosion rate is very low, andthe corrosion current density does not exceed the value of ic = 65 nA·cm–2.

It should be highlighted at this point that no unfavourable tendencies inthe corrosion state of the sample were detected in the course of the chemicaltreatments. The passive character of the surface is essentially constant followingthe oxide dissolution by the CORD-UV method and the follow up treatmentwith boric acid solution. Therefore, it can be concluded that the removedparts (fraction) of the protective oxide layer do not affect the average corrosionrate of the steel surface.

Table 22.3 Corrosion parameters evaluated from the polarization curves

Corrosion data Sample X

X/1 X/1 X/3 X/4

Corrosion potential (Ec) (mV) –64.8 –53.2 –54.8 –66.5

Corrosion current density 37.5 47.7 43.6 38.8(ic) (nA · cm–2)

Corrosion rate (vc), (mm · y–1) 0.43 0.55 0.50 0.45

22.2 Potentiostatic polarization curves of the steel sample measuredin boric acid solution (c = 12 g · dm–3). Sweep rate: 10 mV min–1.

Legend: (X) number of the steel specimen,(1) original surface (two parallel samples),(3) surface obtained after the CORD-UV procedure,(4) surface obtained after the follow-up treatment with boric acid solution.

E/V–0.2–0.4 0.0 0.2 0.4 0.6 0.8

SampleX/4

SampleX/3

SampleX/1

lgi /

A (

cm–2

)

–5

–6

–7

–8

–9


22.3.3 Examinations by SEM-EDX methods

SEM micrographs clearly demonstrate the changes in surface morphology ofthe sample after the CORD-UV treatment.

Surface condition before treatment: the protective surface layer formedon the inner side of the heat exchanger tube had an amorphous structure witha considerable roughness factor. The surface was covered with spongy Fe-oxide deposits. The protective inner layer enriched in Cr was shielded by anouter Fe-oxide deposit. This means that the amorphous Fe-oxide was mainlyexcited by the EDX method. This fact is confirmed by the low intensity ofX-ray peaks in the EDX spectra. It is to be noted that thickness of the oxidelayer on the surface exceeds 1 mm.

As a consequence of the oxide layer treatment by the CORD-UV process,the surface roughness of the steel tube sample decreased, and the thicknessof the Cr rich inner oxide layer remained unchanged. Moreover, the amountof the spongy structured deposits on the surface decreased, too. The favourablechanges in surface morphology may be explained by the adequate applicationof the CORD-UV procedure used for oxide layer removal (i.e. Fe(II) dissolutionshould be performed under quasi-equilibrium conditions). On the other hand,it is probable that the multilayer oxide structure (so-called ‘duplex’ structure)formed from the bulk alloy on the inner surface of untreated stainless steelspecimens, which is also advantageous for the oxide dissolution procedure.


Due to the range of the conversion electrons in oxides, information may beobtained from a layer thickness of about 300 nm. The error in the determinationof the phase composition is ca. ± 5%; however, this does not include thedepth dependent sensitivity of the detection process. Therefore the phasecomposition (i.e., iron content of the different phases) obtained is always anintegral composition of a 300 nm thick surface layer, which contains anexponential weighting according to the depth of the particular phase. Figure22.3 shows illustrative 57Fe-CEMS spectra of the sample. Besides aconsiderable amount of amorphous – non-stoichiometric – Fe(OH)3 andaustenitic steel phase, ca. 16 mass percent magnetite (spinel) was found inthe upper 300 nm thick layer of the inner surface of the untreated samples.The surface of the sample obtained after the oxide layer dissolution containedsignificantly less magnetite and amorphous Fe(OH)3 phases.

It may be concluded that an adequate application of the CORD-UVmethod provides favourable alterations of the composition of the surfaceoxide layer; i.e. the amount of magnetite and amorphous Fe(III)- oxides(-hydroxides) decreases substantially. However, it is of special importanceto emphasize that even in case of the appropriate application of the


CORD-UV technology, substantial amounts of amorphous Fe(III)-hydroxidesremain on the surface.


The XRD diffractograms were taken of steel specimens after the three mainsteps of the experiment. Namely, the original surface condition, the surfacecondition after the CORD-UV and the surface condition after the boric acidtreatment were investigated. The characteristic XRD diffractograms are seenat Fig. 22.4. The distribution of the crystalline phases evaluated from theXRD diffractograms is shown in Table 22.4.

y (mm/s)–10 –5 0 5 10

SampleX/3

SampleX/1

Co

un

ts

8700000

8750000

9800000

9850000

1.34 ¥ 106

1.36 ¥ 106

1.37 ¥ 106

1.38 ¥ 106

22.3 CEMS spectra measured on the inner surface of the stainlesssteel specimens.

Legend: X/1: Before the CORD-UV technology;X/3: After the CORD-UV technology


On the inner side surface region of the untreated sample the dominantpresence of austenitic steel and slight amount of magnetite phase was observed.After the CORD-UV procedure a significant increase in the amount of thecrystalline phases can be detected. Similar but less intense changes may benoticed after the treatment with boric acid solution. Only crystalline austeniteand magnetite phases could be identified on the inner surface of every sample.As a consequence of the above treatments, the ratio of magnetite to austeniticsteel decreased continuously. These facts give further confirmation of thedeductions made previously.

X/3 inner surface

X/1 outer surface

S

MSM

MM

M

MM

70605040302010

1400

0

200

600

800

1000

1200

400

% Theta

Co

un

ts

X/3 inner surface

MM

SS

SS

M MM

M

706050403020100

200

1200

1000

400

600

800

% Theta

Co

un

ts

22.4 Diffractograms of the specimen’s surface before and after theCORD-UV procedure.


22.3.6 Investigation of solution compositions with ICP-OES and gravimetrical methods

The amount of dispersed (colloid) and/or dissolved corrosion products (Fe,Cr, Ni, Mn and Co) removed from the oxide layer into the solutions duringthe CORD-UV procedure and the boric acid treatment was studied by ICP-OES and gravimetric methods. In addition, the average thickness of theoxide layer dissolved into the boric acid solution was estimated from thetotal amount of corrosion products measured in the model solution of primarycoolant.

22.5 Time dependence of the total concentration of main alloyingcomponents dissolved into the boric acid solution (13 g · dm–3)containing oxalic acid (2.2 g · dm–3) in the course of the CORD-UVmethod.

Ni FeCr

302520151050Elapsed time (h)

0

100

200

300

400

500C

on

cen

trat

ion

(m

g d

m–3

)

Table 22.4 The percentage of the crystalline phases

Number of sample Crystalline phase (mass percent)

SS – 304 steel M – magnetite

X/1 outer side original surface 64 36

X/1 inner side original surface 94 6

X/3 inner side surface after the 96 4

CORD-UV technology

X/4 inner side surface after boric 97 3acid treatment


Figure 22.5 shows that dissolution of the main alloying components (Fe,Cr, Ni) into the boric acid solution containing oxalic acid proceededcontinuously without saturation. The total concentration of Cr and Ni dissolvedfrom the protective oxide layer was significant, but not more than 12 masspercent of Fe concentration. The average values of Fe concentration measuredin the solution – by taking into consideration the solubility of Fe(II)-oxalatein hot water [2] – give a strong indication that the dissolution of the surfaceoxide in the course of the CORD-UV procedure was performed under quasi-equilibrium conditions.

The average thickness of the removable surface layer (d ) is a characteristicof the mobility of the remnants of oxide layer following the chemical treatmentof the inner tube surfaces by the CORD-UV procedure. It was found thatonly a thin layer (d = 0.077 mm) could be removed from the surfaces of steelspecimens. This finding may be explained by (i) an appropriate applicationof the CORD-UV technology, (ii) the stable ‘duplex’ structure formed on theinner surface of untreated stainless steel specimens.

22.4 Conclusions

The aims of the present studies of the heat exchanger tubes were to obtaininformation on the efficiency of oxide layer removal, as well as on thecorrosion and solution chemical effects of the CORD-UV technology. (Thisversion of the CORD-UV procedure was developed for the in-core removalof various non-spinel type oxides such as hematite and amorphous Fe-oxides(-hydroxides), as well as low substituted magnetite.) From the results of thecomprehensive studies detailed earlier it can be concluded that:

∑ The decontamination efficiency of the oxide dissolving version of theCORD-UV procedure is extremely low (DFCORD-UV = 1.10∏1.34).Therefore, it is probable that the present version of the CORD-UVtechnology is not an efficient method to remove the main part of thespinel-type oxide layer from such surfaces.

∑ The voltammetric results reveal that no unfavourable tendencies in thecorrosion state of the sample were detected in the course of the chemicaltreatments. The passive character of the surface was essentially constantfollowing oxide dissolution by the CORD-UV method and the follow uptreatment with boric acid solution. Therefore, it can be concluded thatthe removed parts (fractions) of the oxide layer do not affect the averagecorrosion rate of the steel surface.

∑ The SEM-EDX and ICP-OES results indicate that the dissolution featuresof the surface oxide layer are basically dependent upon the chemicalstability (‘duplex’ structure) of the protective oxide layer as well as thecorrect application of the CORD-UV method.


∑ The CEMS and XRD studies show that proper application of the CORD-UV method may cause favourable changes in the composition of thesurface oxide layer; i.e., the amount of magnetite and amorphous Fe(III)-oxides (-hydroxides) decreases substantially. However, it is of specialimportance to emphasize that even in the case of the correct applicationof the CORD-UV technology, substantial amounts of amorphous Fe(III)-hydroxides remains on the surface.

22.5 References

1. K. Varga, Z. Németh, J. Somlai, I. Varga, R. Szánthó, J. Borszéki, P. Halmos, J.Schunk and P. Tilky, J. Radioanal. Nucl. Chem. 254, 589 (2002).

2. CRC Handbook of Chemistry and Physics (Ed. D. R. Lide), CRC Press, London,(1994).

328

10GN2MFA steel 70–517-4 PH precipitation hardened stainless

steel 19A286 austenitic stainless steel 18A410 martensitic stainless steel 18–19acetic acid 119–29adsorption-induced dislocation emission

(AIDE) 153–4, 157–8AISI 304 stainless steel 15–18, 104, 105,

113, 114, 115IASCC susceptibility under BWR

conditions 59–69pattern recognition model to estimate

IGSCC 245–59yield strength and crack propagation

200–10AISI 304L stainless steel

effect of cold working 76–86effect of strain path on SCC 87–102fatigue crack growth in primary water

260–9AISI 316L stainless steel 15–18

dynamic strain ageing 103–18effect of cold work hardening 76–86yield strength and crack propagation

200–10AISI 316NG stainless steel 103–18AISI 347 stainless steel

IASCC susceptibility under BWRconditions 59–69

yield strength and crack propagation200–10

Alloy 600

effect of cyclic loadings on crackgrowth rate in primary water231–44

crack growth rate measurements240–1, 243

experimental procedure 234–6fracture morphology 236–40, 241,

242materials and specimen 231–4

IGSCC in a ‘complex’ environment119–29

steps 120–7IGSCC in primary water 5–11kinetics of passivation 44–56low temperature mill annealed

(LTMA) and Kori 1 retiredsteam generator tube 306–15

mill annealed and secondary side11–15

Alloy 690 7, 12Alloy 800 12Alloy X750 6a-phase 292, 299–300, 301–3, 319,

323–4, 325alternate-shear 158alternate-slip 158alumino-silicates 14, 119, 127–8

aluminium/silicon ratio 125, 128ammonia 125amorphous iron hydroxides 304, 322–3amorphous oxide layers 295–7, 300–4, 322anodic dissolution 224, 286AP-CITROX technology 290, 304

Index

Index 329

artificial neural networks (ANN) 245,250–2, 253–7, 258

ASME Boiler and Pressure Vessel Code184, 198

ASME XI wet fatigue CGR curves211–12

and corrosion fatigue behaviour oflow alloy steels 220, 221–2,225, 226–7, 229

CASTOC project comparison withBWR/NWC conditions 172–9VVER conditions 190–2, 193, 194,

195Auger spectroscopy 278–81austenite 324austenitic stainless steels 87

dynamic strain ageing 103–18effect of cold work hardening

76–86fatigue crack growth in primary water

260–9heat exchanger tubes of steam

generators 289–305IASCC susceptibility under BWR

conditions of welded stainlesssteels 59–69

primary circuits 15–19TGSCC in at high temperatures

149–61VK-50 reactor measurement channel

273–88yield strength and crack propagation

200–10see also under individual types

autoclavesHWC monitoring 26, 27, 28PWR secondary systems 36–9

back-propagation 252, 253baffle former bolts (BFB) 200–10Baushinger strain path 90, 93–100biaxiality 149, 151boiling water reactors (BWRs) 3–5, 17,

103CASTOC project for BWR/NWC

conditions 165–85corrosion fatigue crack growth

behaviour of low-alloy steels211–30

corrosion potential monitoring 25,26–34, 41

IASCC susceptibility of weldedstainless steels 59–69

pattern recognition to estimate IGSCC245–59

simulated BWR conditions and effectof water chemistry transients onlow-alloy steels 130–48

yield strength and crack propagationin BWR conditions 204–5,207–8

Bragg’s law 62brain, human 251BWRVIP-60 SCC disposition lines 130,

131CASTOC project

BWR/NWC conditions 179–81,182, 183–4

VVER conditions 192–4, 196,198

comparison of water chemistrytransient tests with 144–6

calcium 279, 280carbides

Alloy 600 6, 8intergranular 8, 234intragranular 8, 234VK-50 reactor measurement channel

277, 284carbon 224, 229

VK-50 reactor measurement channel279–82

carbonate 125CASTOC project 165–99

BWR/NWC conditions 165–85environment 168–9experimental facilities 168inter-laboratory comparison test

169–72loading conditions 168materials 166–8results from cyclic loading and

load transients 172–9, 180

Index330

results from static loading 179–83VVER conditions 186–99

experimental facilities 189materials 187–9results from constant loading

192–6, 197results from cyclic loading 190–2,

193, 194, 195specimen preparation 189testing procedure and environment

189–90caustic cracking 14, 15cell block boundaries (CBBs) 99characteristic decay time 110–12chemical cleaning 306, 308, 311chemical decontamination 290, 301–3,

304chloride transients 130–48

CASTOC project 176–9, 180, 183comparison with BWRVIP-60 SCC

disposition lines 145–6constant load tests 136, 141–4effect on EAC behaviour of low-alloy

steels 139–44, 145–7periodical partial unloading 135–6,

139–41chlorine 279, 280, 281, 283

precipitation on boundaries 285chromium 295, 296

critical chromium content 254dissolved into boric acid solution

325–6Kori 1 retired steam generator 310–11passivation of nickel base alloys 46–7,

50–2VK-50 reactor measurement channel

279, 280, 281chromium depletion 6, 17, 200, 246, 248,

254, 273, 284chromium hydroxide 47, 48, 49, 53–4chromium oxide 322

passivation kinetics of nickel basealloys 47, 48–50, 51–2, 53–4

classical growth models 49, 50, 52–3, 54cleavage-like cracking 151, 152, 153

proposed mechanisms 153–6cobalt radionuclides 320

cold rolling 79, 82–3cold work hardening 76–86

CERTs 78–80, 81–4constant deformation tests 80–1, 84–5constant load tests 80, 84effect of yield strength on crack

propagation 200–10influence of cold work process 81–4materials 77non-cold-worked specimens 78, 81PWR test procedure 77

‘complex’ environment 119–29incubation stage 122, 123initiation of IGSCC 122–4rapid propagation 124reference ‘complex’ environment 120,

121slow propagation 124–7

conductivity 248constant deformation tests 80–1, 84–5constant extension rate tests (CERTs) 247

effect of cold work hardening 78–80,81–4

constant load testsAlloy 600 in primary water 235–9,

240, 241CASTOC project

BWR/NWC conditions 168, 169,170, 179–83

VVER conditions 192–6, 197effect of cold work hardening 80, 84water chemistry transients 136, 141–4

conversion electron Mössbauerspectroscopy (CEMS)

comprehensive investigation of heatexchanger tubes 291–2,298–300, 301–3

FRAMATOME CORD–UVtechnology 317, 318–19, 322–3,327

coolant density 278, 282copper

Kori 1 retired steam generator 307–8band formation 309–10

VK-50 reactor measurement channel279, 280, 281, 285–6

copper chloride complex 310

Index 331

CORD-UV technology seeFRAMATOME CORD-UVtechnology

core internals 16–18core shroud failures 59corrosion-assisted cleavage 155corrosion current density 291, 293, 294,

320–1corrosion-enhanced localised plasticity

155, 159corrosion failure depth 278, 282corrosion fatigue 211–30

assessment of current ‘ASME XI wetfatigue CGR curves’ 226–7,229

comparison to GE model 224–6, 229crack growth monitoring and

fractographical evaluation 216effect of loading conditions 221–2effect of material parameters 222–4,

229effect of temperature and loading

frequency 217–20, 228environmental parameters 215–16materials 212–15mechanical loading 216–17specimens 215superposition model and time-domain

analysis 227–8‘corrosion map’ 289corrosion potential 291, 293, 294, 320–1corrosion potential monitoring 25–43

BWRs 25, 26–34, 41PWRs 25, 34–41, 42

primary systems 25, 34–5, 42secondary systems 25, 36–41, 42

corrosion rates 291, 293–5, 301–3,320–1

counter sinking, cold working by 79,82–3

coupled environmental fracture model(CEFM) 248

crack-arrest markings (CAMs) 151, 152,153

possible explanations 156–7crack cessation 169, 170, 181–2, 183,

192–4

crack growth monitoring 134, 216crack growth rates (CGRs)

AISI 304L steel fatigue crack growthin primary water 261–3,267–8

AISI 316 stainless steels 113–14, 115,116

CASTOC projectBWR/NWC conditions 172–9,

180, 181, 182, 183inter-laboratory comparison test

169–72VVER environments 190–6

corrosion fatigue behaviour of low-alloy steels 217–29

effect of cold work hardening 83–4effect of cyclic loadings on for Alloy

600 in primary water 231–44effect of lead on low-alloy steels in

high-temperature waterenvironment 71, 72–4

effect of yield strength 203–8measurement for Alloy 600 11pattern recognition model for IGSCC

253–7, 258stainless steels in high temperatures

149, 150rate-controlling steps for TGSCC

159–60strain path and SCC of stainless steels

92, 94–5, 96–7water chemistry transients 134,

146–7chloride transients 139–44, 145–7comparison with BWRVIP–60

SCC disposition lines 144–6sulphate transients 136–9, 144–5,

146crack opening displacement (COD) rate

149, 150, 159–60critical chromium content 254critical depth 122critical stress intensity factor 100–1, 122,

149, 150, 151, 249cross test strain path 90, 93–100crystalline phases 296, 297, 304, 323–4,

325

Index332

cyano ligands 128cyclic loading

CASTOC projectBWR/NWC conditions 168,

169–71, 172–9, 180VVER conditions 190–2, 193, 194,

195effect on crack growth rate in Alloy

600 in primary water 231–44see also low frequency corrosion

fatigue (LFCF) tests

decay time, characteristic 110–12decohesion 154, 158decontamination, chemical 290, 301–3, 304decontamination factors 319–20, 326Delta rule 252dendrites 251denting 12–13deterministic approach 245, 246–9,

253–7, 258dimples 152, 158, 267direct current potential drop (DCPD)

method 134, 168, 189, 216discontinuous cracking 157dislocation density 63–4dislocation emission 153–4, 157–8dislocation loops 17, 277dislocation motion 98–100dissolution mechanisms 156dissolved/dispersed corrosion products

319, 325–6ductile fracture 67–8, 82duplex structure of oxide layer 49–50,

53–4, 322, 326dynamic strain ageing (DSA) 103–18,

223–4, 229experimental procedure 104–6results 106–14, 115

ECP modelling 32–4, 42electrochemical corrosion potential (ECP)

25, 247–8see also corrosion potential

monitoringelectrochemical polarisation reverse

(EPR) 246, 248–9, 254, 255

electron microscopy 275–7elongation to fracture 107embrittlement 157

hydrogen embrittlement 19, 21, 224liquid-metal embrittlement 157–8

energo-dispersion analysis (EDA) 284–5enthalpy 114environmental modification 15environmentally-assisted cracking (EAC)

20, 21, 211effects of water chemistry transients

130–48forms of 212

EPRI water chemistry guidelines 130,131

Evans diagrams 32–4

fasteners see high strength fastenersfatigue

cold working by 78–9, 82–3, 84corrosion fatigue see corrosion fatiguecrack growth in AISI 304L stainless

steel in primary water 260–9air at 20ºC 264fatigue experiments 260–1fractographic analysis 263–7macroscopic crack growth rate

261–3primary water at 20ºC 261, 262,

264–6, 267–8primary water at 300ºC 262, 263,

266–7, 268fatigue striations 221–2, 264–5, 266–7ferrite 299–300, 301–3film-induced cleavage 155, 159film rupture 224fluid velocity 247Forman law 262Fourier transformation spectra 110, 111fracture surfaces

cyclic loadings and Alloy 600 inprimary water 236–40, 241,242

effect of lead on resistance of low-alloy steel 72, 73, 74

fatigue crack growth in AISI 304Lsteel in primary water 263–7

Index 333

fatigue striations 221–2TGSCC in austenitic stainless steels

151, 152–3VK-50 reactor measurement channel

277, 278welded stainless steels 68

FRAMATOME CORD-UV technology316–27

CEMS analysis of surface oxidelayers 317, 318–19, 322–3,327

g-spectrometry 317, 318, 319–20ICP-OES and gravimetrical methods

317, 319, 325–6SEM-EDX methods 317, 318, 322,

326voltammetry 317, 318, 320–1, 326XRD phase analysis 317, 319, 323–4,

325frequency, loading 175, 178, 217–20,

228

g-spectrometry 317, 318, 319–20GE model 224

comparison of LFCF test data forlow-alloy steels to 224–6, 229

grain boundaries 284–5carbides on 6, 8

grain boundary diffusion 7grain boundary sliding 7grain boundary poisoning 273, 285, 286grain size 63, 106gravimetrical methods 319, 325–6

haematite 296–7hardening

cold work hardening see cold workhardening

yield strength and crack propagationof hardened stainless steels200–10

hardnessAISI 316 stainless steels 106welded stainless steels 64

heat 3110439 Alloy 600 231–44heat WL344 Alloy 600 231–44heat-affected zone (HAZ)

CASTOC project 168, 173, 174, 182corrosion fatigue crack growth

behaviour 212–29IASCC susceptibility of welded

stainless steels under BWRconditions 59–69

heat exchanger tubescomprehensive investigation of

corrosion state 289–305CEMS analysis of surface oxide

layers 291–2, 298–300, 301–3sample preparation 290SEM-EDX method 291, 295–7voltammetry 291, 292–5XRD phase analysis 292, 300

FRAMATOME CORD-UVtechnology see FRAMATOMECORD-UV technology

helium bubble formation 17–18high strength fasteners

baffle former bolts 200–10low-alloy steels 21stainless steels 18–19

hold time 174–5, 177human brain 251hydrogen

injection rate 31–2TGSCC in austenitic stainless steels

at high temperatures 153–5,157–8

hydrogen embrittlement 19, 21, 224hydrogen-enhanced decohesion (HEDE)

154, 158hydrogen-enhanced localised plasticity

(HELP) 154–5, 158–9hydrogen sulphide 21hydrogen water chemistry (HWC) 4–5,

25monitoring 26–34

ICP-OES 317, 319, 325–6impurity hideout 12–13in-pipe monitoring

HWC 26–34PWR secondary systems 37–9

incubation time 122, 123intergranular attack (IGA) 275, 276

Index334

Alloy 600 in secondary side of PWRs12–15

Kori 1 retired steam generator tube312

steam generator tubes in a ‘complex’environment 119–29

intergranular carbides 8, 234intergranular facets 265, 267intergranular fracture 68, 81–2, 249

Alloy 600 and cyclic loadings inprimary water 236, 239

CASTOC project 194, 195, 196fatigue crack growth in AISI 304L

steel 265, 266intergranular stress corrosion cracking

(IGSCC) 5, 25, 59, 103, 275,276, 286–7

A286 18Alloy 600

in primary water 5–11in secondary side 12–15

pattern recognition model 245–59stages for Alloy 600 120–7

incubation 122, 123initiation 122–4rapid propagation 124slow propagation 124–7

steam generator tubes in a ‘complex’environment 119–29

strain path and in stainless steels87–102

inter-laboratory comparison test 169–72intermittent microstructural barriers

156–7internal friction (IF) 110, 112INTERWELD project 59–69intragranular carbides 8, 234intragranular precipitates 233–4inverse logarithmic curve fitting 49, 50,

52–3, 54iron

dissolved into boric acid solution325–6

nickel and iron Pourbaix diagram3–5

passivation of nickel base alloys 47,48, 50, 52

VK-50 reactor measurement channel279, 280, 281, 283

iron hydroxides, amorphous 304, 322–3iron oxalate 304irradiation-assisted stress corrosion

cracking (IASCC) 16–18, 200effect of cold work hardening on in

stainless steels 76–86susceptibility of welded stainless

steels under BWR conditions59–69

yield strength and in hardenedstainless steels 206–8

irradiation creep 18

kinetic growth models 49, 50, 52–3,54

Kori 1 retired steam generator tube306–15

experimental method 306–7outer diameter SCC 312–14pitting 305–11primary water SCC 311–12

lead 15effect on resistance of low-alloy steel

in high-temperature waterenvironment 70–5

learning 252lepidocrocite 299ligaments 156linear elastic fracture mode (LEFM) 181,

182liquid-metal embrittlement (LME)

157–8lithium 204loading conditions 221–2loading frequency 175, 178, 217–20,

228localised dissolution 157localised plasticity 154–5, 158–9logarithmic curve fitting 49, 50, 52–3,

54Lomer-Cottrell (L-C) locks 155, 159low-alloy steels (LAS) 19–21

CASTOC project see CASTOCproject

Index 335

corrosion fatigue crack growthbehaviour 211–30

effect of lead on resistance to SCC70–5

effect of water chemistry transients130–48

high strength fasteners 21secondary circuit components 19–20

low frequency corrosion fatigue (LFCF)tests

CASTOC projectBWR/NWC conditions 168,

172–9, 180VVER conditions 190–2, 193, 194,

195low-alloy steels and corrosion fatigue

crack growth behaviour211–30

comparison to GE model 224–6,229

effect of loading conditions 221–2effect of material parameters

222–4, 229effect of temperature and loading

frequency 217–20, 228schematic 216–17time-domain analysis 227–8

low temperature mill annealed Alloy 600306–15

magnetic phases 296–7magnetite 14, 296–7, 300, 322, 324, 325manganese 273, 296martensite 89, 299–300, 301–3martensitic stainless steels 18–19material susceptibility indices 7–11mechanical properties 64, 133, 234

see also yield strengthmemory effect 179, 183mill annealed Alloy 600 11–15

‘complex’ environment 119–29mixed potential model (MPM) 248monitoring points

BWR 28, 31PWR

primary system 34secondary system 36

Monte Carlo simulation 10–11

neural networks see artificial neuralnetworks (ANN)

neurons 251neutron diffraction 62neutron fluence 278, 282

distribution 274, 275nickel 295, 296

dissolved into boric acid solution325–6

Kori 1 retired steam generator310–11

nickel and iron Pourbaix diagram3–5

passivation of nickel base alloys 46,47–8, 50–2

nickel base alloyskinetics of passivation 44–56in primary water 5–11in secondary side of PWR steam

generators 11–15nickel hydroxide 48, 49, 51nitrogen

alloying and DSA of AISI 316Lstainless steel 103–18

‘free’ interstitial 110, 111, 224,229

non-deterministic approach 245, 250–2,253–7, 258

normal water chemistry (NWC) 4–5CASTOC project under BWR/NWC

conditions 165–85corrosion fatigue under BWR/NWC

environment 211–30

organic compounds 125Oskarshamn 2 BWR 26, 27, 28, 31–2outer diameter stress corrosion cracking

(ODSCC) 312–14OWC simulator 40–1, 42oxidation times

longer 50–2short 46–50, 54

extrapolation of growth lawscalculated for 52–3

oxide layers

Index336

amorphous 295–7, 300–4, 322CEMS analysis 291–2, 298–300,

301–3, 317, 318–19, 322–3,327

comprehensive investigation of heatexchanger tubes 291–2,295–300, 301–3, 304

thickness 301–3passivation kinetics of nickel base

alloys 47, 48–50, 51–2, 53–4removal see FRAMATOME CORD-

UV technologySEM-EDX 291, 295–7, 317, 318, 322,

326oxygen

dissolved oxygen contentCASTOC project 173, 175, 191,

193, 194, 195and corrosion fatigue of low-alloy

steels 215–16, 222–4passivation of nickel base alloys 47,

48, 50–2VK–50 reactor measurement channel

279, 280, 281oxygen concentration 285

oxygen/ECP transients 87–9, 130oxygenated water chemistry (OWC)

39–41, 42

Paks NPP steam generator tubes 289–305parabolic curve fitting 49, 50, 52–3, 54passivation kinetics 44–56

extrapolation of growth lawscalculated for short oxidationtimes 52–3

longer oxidation times 50–2materials 44–5short oxidation times 46–50, 54surface analysis 45–6three-step mechanism 53–4

passivity current density 291, 293,294

pattern recognition model 245–59data collection 250deterministic approach 245, 246–9,

253–7, 258effect of conductivity 248

effect of ECP 247–8effect of fluid velocity 247effect of pH 246–7, 254–5, 256effect of sensitisation (EPR) 246,

248–9, 254, 255effect of stress intensity 249, 254,

255non-deterministic approach (ANN)

245, 250–2, 253–7, 258results 253–7

periodical partial unloading (PPU)CASTOC project

BWR/NWC conditions 168,172–9, 180

VVER conditions 190–2, 193, 194,195

water chemistry transientschloride transient 135–6, 139–41sulphate transient 134–5, 136–9

pH 246–7, 254–5, 256phase analysis

CEMS 291–2, 298–300, 301–3, 317,318–19, 322–3

XRD 292, 300, 301–3, 317, 319,323–4, 325, 327

phosphate 119–29pitting 12–13, 21, 286


point defects 17potentiostatic polarisation 291, 292–3,

317, 318, 320–1Pourbaix diagram 3–5precipitates 275–7

intragranular 233–4precipitation hardened stainless steels

18–19pre-shear hardening 88–90

initiation and propagation of SCC93–5

pressurised water reactors (PWRs) 3–5corrosion potential monitoring 25,

34–41, 42primary systems 25, 34–5, 42secondary systems 25, 36–41, 42

Kori 1 see Kori 1 retired steamgenerator tube

Index 337

primary water see primary waterSCC of steam generator tubes in a

‘complex’ environment119–29

secondary systems see secondarysystems

yield strength and crack propagationin PWR conditions 203–4,206–7, 208

primary pressure boundary 15–16, 21primary water 3–4

AISI 304L steeleffect of strain path on SCC

87–102fatigue crack growth 260–9

cold work hardening and stainlesssteels in 76–86

corrosion potential monitoring ofprimary systems 25, 34–5, 42

effect of cold work hardening on SCCof stainless steels in 76–86

effect of cyclic loadings on crackgrowth rate in Alloy 600231–44

nickel base alloys in 5–11SCC in Kori 1 retired steam generator

tube 311–12stainless steels in primary circuits

15–19processing elements (PEs) 251, 252pseudo-monotonic strain path 90,

93–100

radiation hardening 17, 66, 200, 206–8radiation induced segregation 200,

284–5rapid propagation 124rate-controlling steps 159–60reactor pressure vessel (RPV) 211

corrosion fatigue crack growthbehaviour of low-alloy steels211–30

effect of water chemistry transients onlow-alloy steels 130–48

recall 252recirculating steam generators (RSGs)

11–15

redox potential 126residual stresses 6, 20

measurements in welded stainlesssteels 61–2, 65–6

reverse strain path 90, 93–100ring-core technique 61–2Ringhals 1 BWR 26–8, 29, 30Ringhals 4 PWR 34–5, 36–9roll transition 312

saturation 286saw tooth wave form 235–6, 237, 238,

241–3scanning electron microscopy-energy

dispersive X-ray microanalysis(SEM-EDX)

comprehensive investigation of heatexchanger tubes 291, 295–7

FRAMATOME CORD–UVtechnology 317, 318, 322, 326

secondary microcracks 264, 265, 267secondary systems 3–4

corrosion potential monitoring 25,36–41, 42

low-alloy steels 19–20nickel base alloys 11–15water chemistry and Kori 1 retired

steam generator tube 306,313–14

selective-corrosion vacancy-creep 155–6,159

sensitisationpattern recognition model 246, 248–9,

254, 255VK-50 reactor measurement channel

284–5serrated yielding 107–10short oxidation times 46–50, 54

extrapolation of growth lawscalculated for 52–3

shot-peening 79–80, 83–4, 85silicon see alumino-silicatesslip-dissolution mechanisms 156slow propagation 124–7slow strain rate tests (SSRT) 62–3

low-alloy steel and effect of lead onresistance 70, 72

Index338

strain path and SCC of stainless steels90–1, 98

non-pre-strained specimens 91–3pre-sheared specimens 93–8

sludge pile 308–9, 310Snoek-like peak 110, 111solid state grain boundary diffusion 7solution composition, investigating 319,

325–6specimen size 192, 195–6stainless steels

austenitic see austenitic stainlesssteels

in PWR primary circuits 15–19strain path and SCC 87–102

static loading see constant loadingsteam generator shell cracking 19–20steam generator tubes 5

heat exchanger tubes see heatexchanger tubes


SCC in a ‘complex’ environment119–29

steam-water interface 286–7strain hardening coefficient 107strain hardening threshold 92–3, 97, 98strain-induced corrosion cracking (SICC)

20, 211, 212low-alloy steels under BWR/NWC

conditions 211–30strain localisation 95–6strain path 87–102

initiation and propagation of SCCwith pre-sheared specimens93–8

initiation and propagation of TGSCCin non-pre-strained specimens91–3

material 88pre-shear hardening 88–90specimen preparation 90SSRT procedure 90–1two-stage strain path and b parameter

89–90stress-assisted directed-dissolution 156stress intensity

factor range and AISI 304L steel261–3

pattern recognition model 249, 254,255

threshold stress intensity factor 100–1,122, 149, 150, 151, 249

stress relief 6stress-strain curves

AISI 316 stainless steels 107–10strain path and 94welded stainless steels 66–7

stress thresholdeffect of strain path 92–3, 97, 98IGSCC in ‘complex’ environments

122–4striations 221–2, 264–5, 266–7Studsvik ECP model 32–4sulphate transients 130–48

CASTOC project for BWR/NWCconditions 175–6, 179, 183

compared with BWRVIP–60 SCCdisposition lines 144–5

effects on EAC behaviour of low-alloy steels 136–9, 144–5, 146

test procedure 134–5sulphide-containing lubricants 21sulphur 273

corrosion fatigue crack growthbehaviour of low–alloy steels212–14, 222–4, 229

VK-50 reactor measurement channel279, 280, 281, 282, 283

grain boundary poisoning 285,286

superposition model 227–8supervised learning 252surface oxide layers see oxide layersswelling 18synapses 251

temperatureCASTOC project for BWR/NWC

conditions 173–4, 176corrosion fatigue crack growth

behaviour of low-alloy steels217–20, 228

effect on crack velocity of Alloy 600

Index 339

in a ‘complex’ environment126, 127

fatigue crack growth in AISI 304Lsteel in primary water 260–9

20ºC 261, 262, 264–6, 267–8300ºC 262, 263, 266–7, 268

lead and resistance of low-alloy steelin high-temperature water 70–5

pattern recognition model to estimateIGSCC 256–7, 258

TGSCC in austenitic stainless steels athigh temperatures 149–61

tensile stresses 6, 283thermal ageing 19thermally treated (TT) Alloy 600

119–29threshold stress intensity factor 100–1,

122, 149, 150, 151, 249time-domain analysis 227–8titanium 279, 280titanium carbides 277transgranular fracture 67–8, 113–14, 249

Alloy 600 and cyclic loading inprimary water 236, 240

cold work hardening of stainlesssteels 82, 83, 84

fatigue crack growth in AISI 304Lsteel 265–6

transgranular stress corrosion cracking(TGSCC) 16

in austenitic stainless steels at hightemperatures 149–61

applicability of proposedmechanisms 157–9

experimental procedure 151–2possible explanations for crack-

arrest markings 156–7proposed mechanisms 153–6rate-controlling steps 159–60results 152–3

strain path and in stainless steels87–102

initiation and propagation in non-pre-strained specimens 91–3

initiation and propagation withpre-sheared specimens 93–8

transients 20

CASTOC project see CASTOCproject

effect of sulphate and chloridetransients on low-alloy steels130–48

oxygen/ECP transients 37–9, 130transition from initiation to propagation

100–1transverse microcracks 72, 73, 74triangular wave form 235–6, 237, 238,

241–3triaxiality 97–8tube expansion 12tube support structures 12, 14

ultimate tensile stress 107under-surface corrosion 275, 276

V-humped specimens, cold pressed 78,81–2

vacancy clusters 155–6, 159vacancy-enhanced creep 155–6, 159vertical recirculating steam generators

(RSGs) 11–13VK-50 reactor

emergency assembly wrapper 273, 281measurement channel 273–88

Auger spectroscopy 278–81chief factors in corrosion 283–5electron microscopy 275–7fractography 277–8governing factors in corrosion

285–6material-operation conditions

274–5metallography 275, 276related factors 286–7

voltammetrycomprehensive investigation of heat

exchanger tubes 291, 292–5FRAMATOME CORD-UV technology

317, 318, 320–1, 326VVER environment 186–99

warm work hardening 201–2wastage 12–13water chemistry transients 20, 130–48

Index340

chloride transientseffect on EAC behaviour 139–44,

145–7test procedure 135–6

comparison with BWRVIP-60 SCCdisposition lines 144–6

crack growth monitoring 134environmental parameters 132fractographical analysis 134materials 132, 133mechanical loading 134specimens 132sulphate transients

effects on EAC behaviour 136–9,144–5, 146

test procedure 134–5Weibull distributions 10, 13–14

Kori 1 retired steam generator failedtubes 308, 311–12

weld filler 212–29weld metal 166, 167, 168, 173, 174welded stainless steels 20, 59–69

irradiation process 61materials characterisation 61, 63–5materials 60–1residual stresses measurements 61–2,

65–6

stress corrosion cracking 66–8stress corrosion tests 62–3

WWER 1000 steam generator 70–5

X-ray diffraction (XRD) phase analysiscomprehensive investigation of heat

exchanger tubes 292, 300,301–3

FRAMATOME CORD-UVtechnology 317, 319, 323–4,325, 327

X-ray photoelectron spectrometry (XPS)45–8, 50–2

yield strength 64, 66, 107crack propagation in hardened

stainless steels 200–10BWR conditions 204–5hardening process 201–2implications for IASCC process

206–8materials 201PWR conditions 203–4test procedure 202

zeolites 127–8

corrosion issues in light water reactors: stress corrosion cracking (efc51)

Documents