Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The tensile properties and toughness of microstructures relevant to the HAZ of X80 linepipe steel girth… Gaudet, Michael J. 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2015_september_gaudet_michael.pdf [ 70.4MB ]
JSON: 24-1.0166539.json
JSON-LD: 24-1.0166539-ld.json
RDF/XML (Pretty): 24-1.0166539-rdf.xml
RDF/JSON: 24-1.0166539-rdf.json
Turtle: 24-1.0166539-turtle.txt
N-Triples: 24-1.0166539-rdf-ntriples.txt
Original Record: 24-1.0166539-source.json
Full Text

Full Text

The Tensile Properties and Toughness of Microstructures Relevant tothe HAZ of X80 Linepipe Steel Girth WeldsbyMichael J. GaudetB. Applied Science in Materials Science and Engineering, University of Toronto, 2007A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDoctor of PhilosophyinTHE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES(Materials Engineering)The University of British Columbia(Vancouver)August 2015c©Michael J. Gaudet, 2015AbstractIn the transportation of oil and gas products from and over remote locations, such as Canada’s Arcticenvironment, pipelines are used. Girth welding to join sections of steel pipelines creates a substan-tial heat affected zone (HAZ) within the base pipeline steel. While there is significant concern asto the fracture and mechanical properties of the HAZ as whole, detailed knowledge about the me-chanical properties of the microstructural constituents is lacking. For this research, measurementsof the temperature time profile in the HAZ of single and dual torch welds were made. This was thenused to guide heat treatments of X80 steel in a Gleeble simulator to create samples of 8 differentbulk microstructures with differing amounts and morphologies of bainite, ferrite and martensite-retained austenite (MA). From the heat treated samples tensile and Kahn tear test specimens weremade for testing at ambient, -20◦C, and -60◦C. The highest strength microstructure proved to bethe finest, lower bainitic microstructure, while the lowest strength microstructure was the coarsest,upper bainitic sample containing a significant amount of MA. This was observed to be true at alltesting temperatures. As part of the tensile behaviour investigation, the Bouaziz dislocation basedmodel for work hardening was applied and shown to fit well across all temperatures and conditions.The Kahn tear test, a machine notched, thin-sheet, slow strain rate test, showed all tests failed in aductile manner. Relative toughness measurement from this test showed that the fine, lower bainiticmicrostructure was the toughest and the coarse, ferritic microstructure was the least tough. Thiswork presents a novel measurement of dual torch temperature time profiles in a real HAZ, an exten-sive mechanical testing program of isolated microstructures relevant to the X80 HAZ at potentialpipeline operating temperatures, and an applied a robust model to fit the work hardening behaviourfor all conditions. This work has the potential for future application in microstructure evolution-property models, and in a combined mechanical model of the different microstructures to furtherimprove understanding of HAZ mechanical responses.iiPrefacePortions of the text were taken and modified with permission of myself, as the author, from the publi-cations of M.J. Gaudet and W.J. Poole, ”Tensile Behaviour and Fracture Properties of X80 LinepipeSteel”, Materials Science & Technology 2011, pages 717-724, M.J. Gaudet and W.J. Poole, ”Ten-sile Behaviour and Fracture Properties of X80 Linepipe Steel”, Iron & Steel Technology, August2012, 99-104, and M.J. Gaudet and W.J. Poole, ”Tensile and fracture properties of X80 steel mi-crostructures relevant to the HAZ.” 2012 9th International Pipeline Conference, American Societyof Mechanical Engineers, 2012. Figures 4.6, 4.7, 4.12, 5.6, and 5.7 were modified from these pub-lications and used within. In these publications, I wrote the entirety of the manuscripts, created thefigures, and performed the experiments with my supervisor, W.J. Poole, providing discussion pointsand editing support.Dr. Warren Poole provided much input and support that guided this work, but specificallyassisted in the initial analysis using the Bouaziz model. Gary Lockhart provided help with thedesign, and technical support for the weld trials held at Evraz NA Inc in Regina. Chris Pennistonformerly of Evraz NA Inc was in charge of the selection of weld procedures to use, as well as thesetup and operation of the welding equipment during the weld trials. Etienne Caron developed themethods used to analyze the data from the weld trials that I modified and applied. Fateh Fazeliassisted in the modification of the cooling setup of the Gleeble 3500 for my experiments. JenniferReichert provided assistance in the methods used for the characterization of my microstructuresas well as data from her own characterization work for comparison. Quentin Puydt modeled thetriaxial stress state of flat strip tensile specimens that allowed for my correction to observed finalfracture stress. Thomas Garcin provided microstructure evolution model results using my Gleebleheat treatment data to compare with my observed microstructures from metallography. Design ofthe Kahn testing rig was made by myself and fabricated by the Materials Engineering machine shop.iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.1 Steel in Oil and Gas Pipelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.2 Pipeline Girth Welds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.3 Dual Torch Welding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.4 Welding Temperature Measurements . . . . . . . . . . . . . . . . . . . . . . . . . 112.5 Heat Affected Zone in Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.6 Fracture Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.7 Kahn Tear Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.8 Microstructure-Property Relationship . . . . . . . . . . . . . . . . . . . . . . . . 232.8.1 Microstructure-Strength Relationships . . . . . . . . . . . . . . . . . . . . 232.8.2 Microstructure-Fracture Behaviour Relationships . . . . . . . . . . . . . . 272.9 HAZ Mechanical Properties & Fracture Toughness . . . . . . . . . . . . . . . . . 293 Scope and Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354.1 Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354.2 Metallography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354.3 Weld Trials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394.4 Gleeble Heat Treatments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41iv4.5 Tensile Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424.6 Kahn Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495.1 Welding Trials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495.2 Gleeble Heat Treatments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535.3 Microstructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555.4 Tensile Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605.5 Kahn Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 705.6 Fractography Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 796 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 916.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 916.2 Weld Trials and Gleeble Heat Treatments . . . . . . . . . . . . . . . . . . . . . . 916.3 Kocks-Mecking Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 966.4 Microstructure-Tensile Properties Relationships . . . . . . . . . . . . . . . . . . . 1106.5 Kahn Tear Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1167 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1237.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1237.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126vList of TablesTable 2.1 Steel chemistries in literature . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Table 4.1 Steel chemistry for this research . . . . . . . . . . . . . . . . . . . . . . . . . . 35Table 4.2 Weld trial thermocouple placement . . . . . . . . . . . . . . . . . . . . . . . . 40Table 4.3 Welding trial procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40Table 5.1 Measured tensile properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Table 5.2 Kahn tear test properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78Table 5.3 Fracture surface observations . . . . . . . . . . . . . . . . . . . . . . . . . . . 90Table 6.1 Work hardening operational parameters . . . . . . . . . . . . . . . . . . . . . . 102Table 6.2 Work hardening model parameters . . . . . . . . . . . . . . . . . . . . . . . . 109viList of FiguresFigure 2.1 Fe-C phase diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Figure 2.2 Schematics of welding setups and resulting multipass cross-sections . . . . . . 10Figure 2.3 Time-temperature schematics of a point in the HAZ of single and dual torch welds 12Figure 2.4 Schematic showing the relation between HAZ position, peak temperature, HAZregion, and Fe-C phase diagram for a single torch weld . . . . . . . . . . . . . 13Figure 2.5 Schematic showing the interaction between initial and following torches on theHAZ regions in a multipass weld . . . . . . . . . . . . . . . . . . . . . . . . . 15Figure 2.6 Schematic of a continuous cooling transformation diagram showing how vari-ous factors influence the final microstructure . . . . . . . . . . . . . . . . . . 17Figure 2.7 Schematic of representative Kahn tear test data showing low and high tear re-sistance results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21Figure 4.1 Classification of microstructures . . . . . . . . . . . . . . . . . . . . . . . . . 37Figure 4.2 Le Pera thresholding method . . . . . . . . . . . . . . . . . . . . . . . . . . . 38Figure 4.3 Schematics of the embedded thermocouple and its placement . . . . . . . . . . 39Figure 4.4 Weld pass definition for single and dual torch welds used in this study. . . . . . 40Figure 4.5 Schematics of Gleeble sample designs and mechanical testing placements . . . 42Figure 4.6 Thermal cycles used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43Figure 4.7 Schematic of the subsized tensile specimen. . . . . . . . . . . . . . . . . . . . 43Figure 4.8 True stress-strain of model input based on the measured as-received conditioncompared to model output. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Figure 4.9 Ratio of model output to the input based on the measured as-received condition. 45Figure 4.10 Example of resulting true stress-strain data without and with correction to thefinal fracture point. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46Figure 4.11 Schematic of the Kahn tear test specimen. . . . . . . . . . . . . . . . . . . . . 46Figure 4.12 Schematic of data measured during the Kahn tear test. . . . . . . . . . . . . . 47Figure 5.1 Observed single and dual torch weld cross-sections. . . . . . . . . . . . . . . . 50Figure 5.2 Measured thermal cycles from single torch and 2.75” spaced dual torch welds . 51Figure 5.3 Measured peak temperature versus t8−5 from weld trials. . . . . . . . . . . . . 51Figure 5.4 Instantaneous cooling rates through austenite decomposition temperatures. . . 52Figure 5.5 Gleeble heat treatment thermal cycle results . . . . . . . . . . . . . . . . . . . 54Figure 5.6 Nital etchings of samples with 5 µm prior austenite grain size . . . . . . . . . 56Figure 5.7 Nital etchings of samples with 42 µm prior austenite grain size . . . . . . . . . 57Figure 5.8 Le Pera etchings of samples with 5 µm prior austenite grain size . . . . . . . . 58Figure 5.9 Le Pera etchings of samples with 42 µm prior austenite grain size . . . . . . . 59Figure 5.10 Measured microstructural constituents of the Gleeble heat treated samples. . . 60viiFigure 5.11 Engineering stress-strain of samples with 5 µm prior austenite grain size . . . . 61Figure 5.12 Engineering stress-strain of samples with 42 µm prior austenite grain size . . . 62Figure 5.13 True stress-strain and strain hardening rate of samples with 5 µm prior austenitegrain size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64Figure 5.14 True stress-strain and strain hardening rate of samples with 42 µm prior austen-ite grain size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Figure 5.15 Example images of failure from tensile testing . . . . . . . . . . . . . . . . . . 66Figure 5.16 True stress-strain to failure of samples with 5 µm prior austenite grain size . . 67Figure 5.17 True stress-strain to failure of samples with 42 µm prior austenite grain size . . 68Figure 5.18 Temperature dependence of tensile properties . . . . . . . . . . . . . . . . . . 69Figure 5.19 Kahn tear test results of samples with 5 µm prior austenite grain size . . . . . . 71Figure 5.20 Kahn tear test results of samples with 42 µm prior austenite grain size . . . . . 72Figure 5.21 Curve fitting of crack length versus adjusted displacement . . . . . . . . . . . 73Figure 5.22 Etear curves determined based on fitted fitted crack data . . . . . . . . . . . . . 74Figure 5.23 SEM images of Kahn fracture surfaces . . . . . . . . . . . . . . . . . . . . . . 75Figure 5.24 Etear and UPEd0−1 as a function of thickness for the as-received material . . . 76Figure 5.25 Fitting curves for Etear and UPEd0−1 as a function of thickness . . . . . . . . . 76Figure 5.26 Corrected Etear and corrected UPEd0−1 as a function of thickness . . . . . . . 77Figure 5.27 SEM tensile and Kahn fracture surfaces of the condition with a prior austenitegrain size of 5 µm with Nb in precipitates cooled at 10 ◦C/s tested at ambienttemperatures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79Figure 5.28 SEM tensile and Kahn fracture surfaces of the condition with a prior austenitegrain size of 42 µm with Nb in solution cooled at 50 ◦C/s tested at -20 ◦C. . . . 80Figure 5.29 As-received fracture surfaces at ambient, -20 ◦C, -60 ◦C. . . . . . . . . . . . . 80Figure 5.30 Condition with a prior austenite grain size of 5 µm, Nb in precipitates cooled at10 ◦C/s fracture surfaces at ambient, -20 ◦C, -60 ◦C. . . . . . . . . . . . . . . 81Figure 5.31 Condition with a prior austenite grain size of 5 µm, Nb in precipitates cooled at50 ◦C/s fracture surfaces at ambient, -20 ◦C, -60 ◦C. . . . . . . . . . . . . . . 82Figure 5.32 Condition with a prior austenite grain size of 5 µm, Nb in solution cooled at 10◦C/s fracture surfaces at ambient, -20 ◦C, -60 ◦C. . . . . . . . . . . . . . . . . 83Figure 5.33 Condition with a prior austenite grain size of 5 µm, Nb in solution cooled at 50◦C/s fracture surfaces at ambient, -20 ◦C, -60 ◦C. . . . . . . . . . . . . . . . . 84Figure 5.34 Condition with a prior austenite grain size of 42 µm, Nb in precipitates cooledat 10 ◦C/s fracture surfaces at ambient, -20 ◦C, -60 ◦C. . . . . . . . . . . . . . 86Figure 5.35 Condition with a prior austenite grain size of 42 µm, Nb in precipitates cooledat 50 ◦C/s fracture surfaces at ambient, -20 ◦C, -60 ◦C. . . . . . . . . . . . . . 87Figure 5.36 Condition with a prior austenite grain size of 42 µm, Nb in solution cooled at10 ◦C/s fracture surfaces at ambient, -20 ◦C, -60 ◦C. . . . . . . . . . . . . . . 88Figure 5.37 Condition with a prior austenite grain size of 42 µm, Nb in solution cooled at50 ◦C/s fracture surfaces at ambient, -20 ◦C, -60 ◦C. . . . . . . . . . . . . . . 89Figure 6.1 Weld trial HAZ microstructure of a 70 mm (2.75”) torch spacing compared toGleeble heat treated microstructures. . . . . . . . . . . . . . . . . . . . . . . . 92Figure 6.2 Weld trial HAZ microstructure of a single torch sample compared to Gleebleheat treated microstructures. . . . . . . . . . . . . . . . . . . . . . . . . . . . 93Figure 6.3 Weld trial HAZ microstructure of a different single torch sample compared toGleeble heat treated microstructures. . . . . . . . . . . . . . . . . . . . . . . . 94Figure 6.4 Example true stress-strain data with labelling of notable properties. . . . . . . 97viiiFigure 6.5 Schematic of the θ vs σtrue showing the definition of operational parameters. . 98Figure 6.6 Analysis of tensile tests data with respect to work hardening . . . . . . . . . . 100Figure 6.7 Plots of θ vs σ −σYS for each major primary microstructure . . . . . . . . . . 101Figure 6.8 Work hardening correlation plots . . . . . . . . . . . . . . . . . . . . . . . . . 103Figure 6.9 Initial dislocation density plotted against transformation start temperature . . . 105Figure 6.10 ξ versus testing temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . 106Figure 6.11 Comparison plots of work hardening model to real tests of 5 µm prior austenitegrain size tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107Figure 6.12 Comparison plots of work hardening model to real tests of 42 µm prior austenitegrain size tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108Figure 6.13 Yield strength, tensile strength and uniform elongation at ambient temperaturesplotted for each microstructure. . . . . . . . . . . . . . . . . . . . . . . . . . 110Figure 6.14 Corrected and normalized Etear plotted against corrected and normalizedUPEd0−1118Figure 6.15 Corrected Etear plotted against the testing condition. . . . . . . . . . . . . . . 119Figure 6.16 σ0.5%ys plotted against corrected Etear. . . . . . . . . . . . . . . . . . . . . . . 120Figure 6.17 Tensile properties compared to Kahn properties . . . . . . . . . . . . . . . . . 121ixList of Symbolst8−5 Time to cool from 800 ◦C to 500 ◦C.γ The austenite phase of steel.α The ferrite phase of steel.MA The mixed phase of martensite and austenite in steel.Tstart The transformation start temperature upon cooling from the austenite phase insteel.σ0 The intrinsic lattice strength of a microstructure.σss The solid solution strengthening component of a microstructure.σppt The precipitate strengthening component of a microstructure.σdis The dislocation strengthening component of a microstructure.σgb The grain boundary strengthening component of a microstructure.ρdis The dislocation density.σys 0.5% offset Yield strength as determined using the 0.5% offset strain method.σUTS The ultimate tensile strength.εUEL Engineering strain at the point of final uniform elongation (i.e. necking point).ε f rac True strain at the point of final fracture in a tensile test.σ f rac True stress at the point of final fracture in a tensile test.θ The work-hardening rate.σys 0 Work-hardening model parameter related to initial yield stress.ξ Characteristic length scale related to capture distance for dynamic recovery.σtear The tear strength from a Kahn tear test.UPEd0−1 The unit propagation energy in a Kahn tear test from max load to 1 mm of furtherpin-to-pin displacement.Etear The energy of tearing in a Kahn tear test, calculated in this study from 0 to 3 mmof crack length.xAcknowledgmentsThis research was made possible through the support of the project industry sponsors, Evraz Inc.NA and TransCanada PipeLines Ltd., and the Natural Sciences and Engineering Research Councilof Canada (NSERC).I am grateful to be surrounded by a great number of people that have been supportive andunderstanding during my research. I wish to first thank my supervisor Dr. Warren Poole for hiscontinued support and inspiring discussions throughout the entirety of my time here. I wish tothank all of the members that have been a part of the X80 project here at UBC, but in particular fortheir assistance of my own work by action or through discussion, Dr. Matthias Militzer, my friendJenny Reichert, Mehran Maalekian, Gary Lockhart, Fateh Fazeli, Thomas Garcin, Quentin Puydt,Morteza Toloui, and Etienne Caron. Jacob Kabel provided assistance to me while working with theSEM, and has grown to be a friend. The support of the machine shop, Ross Mcleod, Carl Ng, andDave Torok at UBC was crucial for my work, particularly for helping me with my Kahn tear testingrig. Wonsang Kim consistently provided quick assistance with minor issues with testing equipment.Michelle Tierney was always a friendly expert at helping me with handling the administrative sideof UBC from the start to the end.Outside of those that have been directly involved with my work I am thankful to have a numberof people that I have been able to rely on for support. I give a very deep thanks to my colleaguesand best of friends, Beth Sterling and Phil Tomlinson whose experience, intelligence, humour andinsight have provided much motivation and support at the most challenging times. I have beenfortunate to have had many friends outside of the program, of these I wish to thank Stephen Kim,Alison Schatz, Joanna Palermo, Elaine Fuertes and Patrick Bielstein, Seth and Claire Gilchrist,Sabrina Higgins, and Raymond Jones for their support, kindness and friendship. While I may notbe the best at staying in contact, I wish to thank those friends I made from back in my life back inxiToronto and Oakville who always are able to pick up our friendship where we left off; Lily Cheng,Geoff Young, Michael Favero, Clarence Chiu, and Thomas Arato.Finally, may these few words impart the depth of my gratitude; I wish to give the biggest thanksof all to my family, both those living and those passed, for their unwavering support that has alwaysprovided a comfort and for which I will remain eternally grateful. Except for my brother Andrew,who despite our friendship apparently never got over the fact that his bike broke because of a poorweld in 2007 [A. Gaudet, UBC Doctoral Thesis, 2010, pg. xv]. Considering there was that time heforgot to gas up the Jimny in the mountain range south of San Jose´, Costa Rica in 2012, forcing usto coast down a mountain in neutral for half an hour in the rain, I think we are even.xiiCHAPTER1Introduction1.1 IntroductionIn North America, significant oil and gas reserves can be found in remote locations that have yetto be fully developed. To access these reserves requires transportation of these resources from theextraction sites to refineries and distribution centres. Pipelines have proven to be the most efficient,safe, and economical way to transmit these resources over the long distances required. Howeverin more extreme environments, such as the Arctic, there are a number of technical and financialchallenges facing pipeline design, construction, and operation. In order to make future projectsviable, these challenges must be overcome through the research and development, in particular, ofmaterials.Steel is well established as the material of choice for long distance transmission of oil andgas resources. Higher strength steels, such as X80 or X100 grades have been shown to be morecost effective than conventional steels and are of importance for the so-called strain based designapproach [1–5] . Strain based design for pipelines puts an additional demand on linepipe steels bysetting a limiting strain value, typically 0.5%, to which the steel must remain structurally stable.Axial strains arise in all pipelines, but for pipelines traversing permafrost like the Alaskan Pipeline,Mackenzie Valley Pipeline, and the Northern Gateway Pipeline, these axial strains are of particularconcern due to issues such as frost heave and thaw settlement [6–9]. It is in important to gain a morecomplete understanding the mechanical behaviour, specifically the yield and initial work hardening1CHAPTER 1. INTRODUCTIONresponse, for these new strain-based design grades of steel.In construction of transmission pipelines, joining sections of pipe requires girth welding in thefield. With more remote locations, and in extreme environments, any improvements to weldingproductivity results in significant time and thus cost savings. Pipeline girth welding is typicallyperformed using a gas metal arc welding (GMAW) process in which a single weld torch depositsweld metal in the joint. To fully weld the thickness of a typical transmission pipeline requiresmultiple passes of the single weld torch. One proposed change to this process is to add a secondwelding torch which trails the first torch at a set distance (typically 50 to 200 mm) during certainpasses. Consequently fewer passes would be required in this so-call dual torch welding compared tothe typical welding procedure and this productivity gain resulting in a significant cost reduction forwelding in the more remote and extreme environments. For comparison, if a single pass of arounda 30” diameter pipe requires 2.5 minutes and requires 7 passes in total, replacing two passes withthe dual torch procedure could potentially reduce total welding time by 5 minutes i.e. greater then25% savings in welding time.Both single and dual torch welding cause changes to the microstructure, mechanical properties,and toughness in the region of the base linepipe steel adjacent to the weld. This region, the so-calledheat-affected zone (HAZ), exhibits variations dependent on the temperature-time profile, termed thethermal cycle, experienced at a given point in the HAZ. The HAZ in a dual torch weld may be morecomplex then a single torch HAZ due to the fact that heat flows from the two torches interact tocause complicated thermal cycles resulting in greater variation of microstructure and properties. Aswell, the influence of multiple passes of the welding torch also leads to variations of microstructureand properties. Regardless of which process is used, the HAZ from girth weld operations will resultin a highly complex situation with relations between the welding process, the resulting thermalcycles, and the final observed microstructures and properties.Considering the cost reduction opportunities of using higher strength steel in strain based de-signed pipelines and utilizing new, higher productivity welding practices it is of interest to gainfurther understanding into the most pertinent challenges. The HAZ, being a complex region thatmay show less desirable properties is of particular concern to applying these higher strength gradesand more productive welding processes. While there is a large amount of literature on testing ofwelds, there is very little research on the local response within the HAZ beyond the use of simple2CHAPTER 1. INTRODUCTIONhardness values. The research proposed here is to examine the microstructures observed in the HAZin multipass welds with both single and dual torch procedures, isolate and then generate pertinentmicrostructures in bulk samples that then can be tested for their mechanical properties. This re-search will use a commercially produced steel pipe of X80 chemistry that is appropriate for strainbased design.3CHAPTER2Literature Review2.1 Steel in Oil and Gas PipelinesHigh strength low alloy (HSLA) steels with microalloying additions have been used extensivelyto meet design criteria as pipelines moved to increasingly high strength grades. These alloys aretypified by low carbon contents (i.e. between 0.05-0.25 wt%), with an amount of manganese upto 2 wt%, and microalloying constituents of molybdenum, nickel, copper, niobium, titanium, vana-dium, and boron which total less then 1 wt% [10, 11]. The highest strength grades for pipelinestypically have lower carbon contents, in the range of 0.03-0.06 wt% . These steels are fabricatedinto plates for pipelines using thermomechanical controlled processing (TMCP) [12]. TMCP is aseries of processes starting with heating a steel slab into the austenitic phase then deforming thesteel at a temperature below the recrystallization temperature. This deformation refines the grainsize and shape as well as increases the dislocation density which impacts the transformation uponcooling. Immediately following this deformation, accelerated cooling is applied to the slab beforecoiling to control the phase transformation. With careful control over the alloying, deformation, andcooling the final microstructure is typically very fine, composed ideally of acicular ferrite or bainite,resulting in high strengths [10,13]. For X80 steels, the typical microstructure observed is a mixtureof mainly fine grain ferrite and bainite [12]. To form the plates into the pipelines, typically either aspiral mill process [3], or the double-bend then expansion process [14] is used. Spiral mill pipes aremade by applying a constant bending radius to the steel plate at a certain angle and then welding4CHAPTER 2. LITERATURE REVIEWthe seams along the formed tube. In the double bend then expansion process the plate is formedinto a U, then O, and then is expanded and welded along the longitudinal seem and is termed theUOE process. These processes both create a length of pipeline that is then shipped to the pipelineconstruction site, ready to be girth welded to form the transmission pipeline.A basic understanding of the metallurgy of these HSLA pipeline steels provides a context to dis-cuss the effects of welding and the HAZ, particularly with regard to how it relates to mechanical andfracture properties. This understanding begins with a discussion of typical phases and microstruc-tures observed in HSLA linepipe steels, and then the role steel chemistry with respect to specificalloying elements.Figure 2.1 shows the typical phase diagram of the iron-carbon system at low carbon con-tents. Over a range of elevated temperatures and low carbon content, only the face-centered cu-bic (FCC) austenite (γ) phase is seen. This ductile phase has a high solubility for carbon. Cool-ing from austenite results in solid phase transformation into various microstructures. At ambientFigure 2.1: Fe-C phase diagram with C content relevant for steels, based on data from [15].5CHAPTER 2. LITERATURE REVIEWtemperatures the primary equilibrium phase consists of α-ferrite, which has a body-centered cubic(BCC) structure with a low carbon content. Relative to other phases that can be produced, α-ferriteis a ductile, low strength phase [10]. The other equilibrium phase at ambient temperatures is iron-carbide (Fe3C), also called cementite. Fe3C is a strong but brittle phase that will be found as platesin a ferrite-Fe3C sandwich structure (i.e. pearlite) if the cooling rate is slow enough.As non-equilibrium cooling is typical in both the production of HSLA pipeline steels and intheir HAZ, the microstructures and phases that can be produced differ vastly from what is repre-sented in the the equilibrium phase diagram. At slow cooling rates, the austenite decomposes intoa primarily ferritic phase where the grains are fairly equiaxed and is often called polygonal ferrite.At slightly faster cooling rates, an orientation relationship between the growing ferrite and par-ent austenite phase develops resulting in an acicular ferrite microstructure. At faster cooling rates,bainitic microstructures are seen. Upper bainite structures are bainitic structures formed at highertemperatures and are typified by packets of lath-shaped sheaves of ferrite with carbides that havehad time to form between sheaves of the progressing ferrite-austenite interface [16]. Lower bainite,which forms at lower temperatures and faster cooling rates, is observed again to be packets of lath-shaped sheaves of ferrite but with the less time for diffusion, carbides form much finer and withinthe ferrite sheaves [16]. At even faster cooling rates, a diffusionless transformation takes place toform metastable martensite. In low carbon steels (<0.2% C), the form of martensite seen is a lath-like structure. This martensite phase, with its body-centered tetragonal (BCT) crystal structure andhigher carbon content is a considerably harder phase. The relation of these microstructures to themechanical behaviour is discussed in a later section.The role of alloying (or residual) elements in HSLA steels and their relation to the microstructure-property relationship is complex. They can directly effect strength through solid solution strengthen-ing, or they may form precipitates that can also add to the strengthening. Beyond direct interactionson physical properties like strength or toughness these alloying elements and/or their precipitatescan indirectly effect properties through their interaction with phase transformations, for instanceby promoting stability of certain phases or slowing the transformation kinetics. The following is ageneral summary of the individual role of each these alloying elements, while later sections con-tain separate and more specific discussion of chemistry effects on microstructure evolution, and ofchemistry effects on physical properties.6CHAPTER 2. LITERATURE REVIEWCarbonCarbon is found either dissolved into the solid solution at interstitial sites in the lattice or as aprecipitate (or carbide) with other elements, such as Fe3C, MoC, NbC, and TiC. As an interstitialalloying element it gives a very strong solid solution strengthening effect in ferritic microstructures.Carbon precipitates (carbides) are often brittle and with increasing amounts and sizes will lead todecreasing toughness of the steel [10]. The solubility of carbon differs with the structure of the ironlattice. Carbon is more soluble in the austenite phase and will tend to stabilize this phase.Carbon is a potent strengthener both directly and indirectly through its promotion of martensite,a microstructure often associated with low toughness. A carbon equivalent (CE) is often used tomake an empirical comparison of steel chemistries’ ability to form high hardness microstructures,and thus is a measure of hardenability. As high hardness microstructures are related to hydrogeninduced cold cracking in welds, the carbon equivalent is used also as a measure of the steel’s abilityto be welded without significant defects, simply referred to as a steel’s weldability. Yurioka [17]has provided a good review of weldability and carbon equivalent equations. One such common andsimple equation to determine weldability in low carbon steels is given by [17]:Pcm = %C+%Si30+%Mn20+%Cu20+%Ni60++%Cr20+%Mo15+%V10+5%B (2.1)where Pcm is the composition parameter, and the percentages of each element are in weightpercent. Higher Pcm values will be expected to have a higher potential for martensite which istypically an undesirable phase due to its low toughness. Pcm values in combination with empiricalevidence can be used then to classify a steel’s weldability.NitrogenNitrogen is an interstitial alloying element, and like carbon is very effective as a solid solutionstrengthener [10]. Similar to carbon, it has greater solubility in the austenite phase then the ferritephase causing it to precipitate out if the cooling rate is slow enough. Nitrogen, like carbon, willprecipitate with other elements. Significant amounts of nitrogen in solution is associated with lowertoughness, so it is often controlled by adding nitride forming alloy elements.7CHAPTER 2. LITERATURE REVIEWManganeseManganese is a common alloying element due to its affinity with sulphur (i.e. prevents harmfuliron-sulphide inclusions), and its low cost [10]. Furthermore, it provides a moderate solid solutionstrengthening effect while being a stabilizer to the austenite phase.SiliconSilicon is an inexpensive, strong deoxidizer and also acts as a moderate solid solution strength-ener [10]. The presence of silicon promotes ferrite formation. At high Si contents, a significantdecrease in toughness is observed.NickelNickel is a strong austenite stabilizer, and thus has a tendency to form martensitic structures [10].It also promotes acicular ferrite structures over equiaxed ferrite structures, which in turn increasestoughness. In a ferrite microstructure, it provides minor solid solution strengthening.MolybdenumMolybdenum is a ferrite former, but promotes acicular ferrite over equiaxed ferrite microstruc-tures. This is resulting from its strong retarding effect of the austenite-to-ferrite phase transforma-tion [18]. It has a strong solid solution strengthening effect, but also forms carbide precipitates [10].ChromiumChromium is an expensive alloying element that stabilizes ferrite phases and also will formcarbides [10]. As a solid solution strengthener it is not effective.Titanium, Niobium, and VanadiumTitanium, niobium and vanadium all form carbides and nitrides, often in complexes with eachother when all elements are present [10]. With the high temperature stability of the TiN precipitates,austenite grain growth at elevated temperature can be limited. Niobium carbides, nitrides, andcarbonitrides add significant precipitation strengthening [10]. These precipitates of Nb are stableto moderately high temperatures in the HAZ and can help keep a refined austenite grain size. Nbhas a moderate solid solution strengthening effect. Its presence has been noted to help promotebainitic transformations. The presence of V and its precipitates has been seen to improve strengthand toughness through promotion of finer structures such as acicular ferrite and bainite [19]8CHAPTER 2. LITERATURE REVIEWResidual ElementsCopper is typically a residual element often found in steels produced in electric arc furnacesdue to their use of recycled steel. Copper provides solid solution strengthening, while at higherconcentrations it forms precipitates but may also cause issues with embrittlement at high tempera-tures [10]. Sulphur is a residual element that is associated with low toughness and production issuesin steel due to the formation of iron-sulphides, a precipitate that causes a number of problems [10].Additions of manganese are typical, as discussed previously, so as to form manganese-sulphidespreferentially over the detrimental iron-sulphides. Phosphorus acts as a strong substitutional solidsolution strengthener; however, it segregates to grain boundaries, leading to an embrittlement effectand low toughness [10]. Thus, efforts are taken to limit the amount of phosphorus and sulphur.2.2 Pipeline Girth WeldsIn order to construct a pipeline on site, the individual lengths of pipeline are welded at their endsaround the girth typically using a mechanized multipass gas metal arc weld (GMAW) procedure.These girth welded joints, including the resulting heat affected zone (HAZ), need to meet the de-mands set by the pipeline design [20–24]. Since these joints represent a portion of the pipeline withsignificantly different properties compared to the base plate, it is an area that has been the focus ofmany investigations. Research on weld joints in pipeline steel can be broadly categorized into thosethat study welding process developments [23–25], weld metal properties [26–28], complete jointassessment [29–34], and HAZ properties and structure which will be discussed later. Of interest topipeline construction is how to reduce the costs of completing a girth weld. One such method is toreplace the commonly used single torch GMAW welding with higher productivity weld processes,such as dual torch welding. With a high productivity weld process, more welds can be completedin the same amount of time and field labour costs can be reduced.2.3 Dual Torch WeldingDual torch welding is a high productivity welding process which has a second welding torch thattrails the lead torch at a set distance. The result is that there is significantly more weld metal thatcan be deposited on a single pass of the welding equipment, which allows for increased weld-ing productivity. Figure 2.2 schematically shows the difference between single torch and dual torch9CHAPTER 2. LITERATURE REVIEWFigure 2.2: Schematics of (a) single torch welding, (b) dual torch welding, (c) the cross-section of a multipass single torch girth weld, (d) the cross-section of a multipass dualtorch girth weld.welds. For a dual torch weld, the parameters of each torch may vary in order to achieve desirablemechanical and fracture results of the final weld. One of these parameters that is being consideredis shown in Figure 2.2 (b), which is the inter-torch spacing in a dual torch weld. It is of interest toconsider how the heat from the lead torch influences the peak temperatures and cooling rates in theHAZ of the second torch. With decreasing inter-torch spacing, this influence would become greater,resulting in higher peak temperatures and slower cooling rates if measured from the same locationin the HAZ. Figures 2.2 (c) and (d) showing schematic diagrams of the cross-section of the depositsfrom a typical pass schedule for single and dual torch pipeline girth welds and the expected HAZthat would result from it. Specifically, note the wider HAZ expected in the dual torch weld.While there is significant research into development of dual torch procedures, there is relativelylittle on the effects of this procedure on the HAZ. The most substantial work can be found in the10CHAPTER 2. LITERATURE REVIEWdoctoral thesis by Hudson [35]. In this work dual torch and dual torch tandem weld proceduresare used to produce joints on X100 strength steel. Tandem welding is a high productivity weldprocedure where two wires feed a single welding torch. In Hudson’s work he showed that for thewelding speed and heat input applied that a 100 mm torch spacing meets the desired properties. Healso showed that a 50 mm torch spacing produces a lower toughness in the HAZ than is desired.Hamad et al. [36] investigated dual torch girth welds for X80 with a 50 mm torch spacing. Theirwork also showed a decrease in toughness in comparison to single torch weld that was related to theHAZ. Gianetto et al. [26] and Hamada et al. [28] investigated the properties of the weld metal fordual torch processes. These studies revealed some interesting aspects of dual torch welds, but lackcomprehensive link between dual torch procedures, microstructure development, and mechanicalproperties.2.4 Welding Temperature MeasurementsThe microstructure evolution in the HAZ is determined by the time-temperature profile, the so-called thermal cycle, that it is exposed to. The thermal cycle varies with weld procedure, weldparameters, and position. For a single torch pass the thermal cycle at a point will rapidly reach apeak temperature and cool from there. In the case of a single pass of a dual torch weld, two peaksare reached. Figure 2.3 is a schematic showing thermal cycles of a single torch and a dual torchpass. In industry practice the time to cool between 800 ◦C and 500 ◦C (t8−5) is used to characterizeif a thermal cycle will potentially result in detrimental microstructures in steel. In the dual torchpass thermal cycle, an individual torch should give a similar thermal cycle to that of a single torch.However, the thermal cycle of the second or trailing torch of a dual torch pass may be affected bythe initial or leading torch, potential slowing the cooling rate which would increase the t8−5. Whilethere is much research into predicting the thermal cycles in the HAZ of a single torch pass, such asthe most commonly used analytical solutions of Rosenthal [37], there has been little work to predictthermal cycles in a dual torch pass. The work by Chen et al. [38] investigated a dual torch thermalcycle using FEM models but provides no predictive tools to expand to other procedures.With no established predictive tool to estimate dual torch thermal cycles for a given proce-dure, direct temperature measurements provide a solution. For thick plate welds, like those in apipeline, embedded thermocouples are necessary for these direct measurements. Embedded ther-11CHAPTER 2. LITERATURE REVIEWFigure 2.3: Time-temperature schematics of a point in the HAZ of (a) a single torch weld, (b)a dual torch weld.mocouples for measuring weld thermal cycles have a number of potential issues. Ensuring goodcontact between the thermocouple junction and the end of the hole is difficult. Welding processesalso introduce a large amount of electrical noise. The presence of a hole can influence the heatflow and may introduce error into the measurements. In spite of the difficulties, direct temperaturemeasurements have been used to validate weld heat flow models in single torch thick plate welds.The work of Hudson [35] used direct temperature measurements to determine thermal cycles inX100 GMAW girth welds, including some dual torch procedures. His measurements involved bothplunging thermocouples into the molten weld pool and embedding the thermocouples under theweld via drilled holes. In the investigation of Nuruddin [39], embedded thermocouples were usedto investigate both tandem metal inert gas welds and submerged arc welds of X65, X70 and X100linepipe steels. Bilat et al. [40] measured thermal cycles of a dual torch GMAW girth weld usingembedded thermocouples. With this thermal cycle data they fit a model to the observed thermal cy-cles and used this to guide laboratory thermal simulations. Poorhaydari-Anaraki [41] drilled holesin X100 plate and embedded thermalcouples to measure thermal cycles from gas tungsten welds ina bead-on-plate procedure. Kou and Le [42] embedded thermocouples in the bottom of a 6061-T612CHAPTER 2. LITERATURE REVIEWaluminum plate to validate their heat flow mode. There are numerous other works that have alsoused embedded thermocouples in thick plates in order to determine HAZ thermal cycles [43–46].While each technique varies slightly in terms of hole design, how it makes a junction with the plate,the material used, and so forth, it is seen that direct temperature measurements through embeddedthermocouples can be used to determine thermal cycles in the HAZ.2.5 Heat Affected Zone in SteelThe heat affected zone (HAZ) is the region of material adjacent to a weld where metallurgicalchanges have occurred. The thermal cycle and specific steel chemistry will determine the extent ofthe HAZ and the microstructural evolution within the HAZ. Reference books by Grong [37] andEasterling [47] provide excellent reviews of microstructural development in the HAZ. For a singlepass of a welding torch with a steel base plate the HAZ is typically categorized into four regions.From closest to furthest from the fusion line, these regions are the coarse grain HAZ (CGHAZ), finegrain HAZ (FGHAZ), intercritical HAZ (ICHAZ), and subcritical HAZ (SCHAZ). Comparison ofpeak temperature reached to the iron-carbon phase diagram, as shown in Figure 2.4 can help ratio-Figure 2.4: Schematic showing the relation between HAZ position, peak temperature, HAZregion, and Fe-C phase diagram for a single torch weld, after [47]13CHAPTER 2. LITERATURE REVIEWnalize the HAZ categories. It should be noted that welding is a non-equilibrium practice. Thus,non-equilibrium values, such as austenite formation start and finish temperatures (Ac1 and AC3 re-spectively), should be used instead of the equilibrium phase diagram. In the diagram it can beseen that the final observation of the CGHAZ is related to high peak temperature of the weld passcausing significant grain growth in the austenite grain size prior to cooling and forming the finaltransformation products. The FGHAZ is seen to be related to when the steel is fully austenitized buthas not seen significant grain growth in the prior austenite grains before cooling and forming thefinal transformation products. The ICHAZ represents a portion of the microstructure where peaktemperatures have not allowed for the microstructure to become fully austenitized. The SCHAZis a region that has not be austenitized at all, but has perhaps had coarsening of carbides or othertempering effects on the microstructure at the elevated temperature.In a multipass or dual torch weld there are some additional complexities as HAZ regions fromeach pass or torch may interact in such a way so as to create potentially different microstructures.Figure 2.5 shows a schematic of a multipass weld microstructure related to the peak temperaturesreached. While there may be only slight differences in microstructure of most of these multipassregions compared to single torch, the intercritically reheated coarse grain HAZ (ICCGHAZ) haspotential to be a drastically different microstructure. As a result the ICCGHAZ as well as theCGHAZ has been the focus of a significant amount of research, particularly with regards to strengthand toughness, which will be discussed later.The classification of HAZ regions gives a broad understanding of what is occurring in the HAZ.A more detailed understanding of the HAZ comes with knowledge of how the thermal cycle andsteel chemistry lead to austenite decomposition products. In a given region of the HAZ for pipelinesteels, the final microstructure typically consists of a complex microstructure of at least two ofthe following decomposition products, which are listed in order of decreasing transformation tem-perature: polygonal ferrite, Widmansta¨tten ferrite, acicular ferrite, upper bainite, lower bainite,martensite.As martensitic structures typically have a portion of retained austenite which is extremely dif-ficult to distinguish from the martensite phase, and thus are often referred to as the martensite-austenite (MA) phase. In general, characterization of the amounts of each structure in the HAZis difficult due to the presence of multiple microstructures that are difficult to differentiate based14CHAPTER 2. LITERATURE REVIEWFigure 2.5: Schematic showing the interaction between initial and following torches on theHAZ regions in a multipass weld, after [48].on optical observations particularly in finer morphologies of the microstructures. Further difficultyin characterization of microstructures in the HAZ is that there is significant changes in the mi-crostructures over very short distances in the HAZ. In literature it is seen that there is often alsodifferent terminology used to describe microstructures [49–52] which emphasizes the uncertaintyin microstructure classification. Often, laboratory thermal simulations are used to isolate the ther-mal cycle at a specific point within a HAZ and use a combination of metallography and dilatometrytechniques to characterize the microstructure.The use of thermal simulations also allows for the construction of a continuous cooling transfor-mation (CCT) diagram for a given composition, which can be used to predict the final microstruc-tures along an HAZ cooling path [37, 47, 53]. However, caution must be used in the strict usage ofCCT diagrams. In the work by Spanos et. al. [54] and the study by Shome and Mohanty [55], CCTdiagrams for HSLA-80 and HSLA-100 grades of steel differ between the CGHAZ and FGHAZ.This difference is attributed to the differences in prior austenite grain size and the dissolution of15CHAPTER 2. LITERATURE REVIEWprecipitates resulting in alloying elements being put into solid solution.Austenite grain size is important to the transformation to final microstructures in the HAZ. Ithas been noted that the larger grain sizes typical of the CGHAZ promote bainitic and martensitictransformation products [56]. The prior austenite grain size in the HAZ is affected by a numberof factors. For HSLA steels there are typically pinning precipitates that are used to limit graingrowth during reheating and hot rolling during production, and in the HAZ during welding. Theseprecipitates can coarsen or dissolve at higher peak temperatures in the HAZ, lessening the Zenerpinning force and allowing the austenite grains to grow to sizes consistent with the CGHAZ. Onesuch set of precipitates in niobium-bearing steels are niobium-carbonitride (Nb(C,N)) complex pre-cipitates. These precipitates have been noted to dissolve at temperatures in the range of 1000 - 1300◦C [56–58]. Titanium-nitride (TiN) precipitates have shown to be much more stable than Nb(C,N),however these can coarsen at high temperatures (above 1300 ◦C) [59]. Moon et. al. [60] showedthat Nb and Ti can also form complex (Ti,Nb)(C,N) precipitates and that these can result in largeraustenite grain sizes than just TiN precipitates alone.The effect of the thermal cycle is extremely important in determining the final microstructure inthe HAZ. This is primarily through the effect of cooling rate during the austenite decomposition butas well through the temperature effect on austenite grain size and state of precipitates. Figure 2.6shows schematically how all of these effects interact on a CCT diagram to promote or slow certaintransformations in the HAZ.In multipass welds the final microstructure can prove to be even more complex. This is most trueof the ICCGHAZ as the formation of austenite upon reheating will contain both the microstructureof the first pass that produced CGHAZ and new decomposition products from the second pass. Ininvestigations by Lee et al. [61], Davis and King [62], Li et al. [63] and Bonnevie et al [64] it is seenthat the ICCGHAZ typically has higher amounts of MA and a higher number of large or massiveMA compared to CGHAZ. Later passes in a multipass weld were shown to reduce the amount ofMA in the ICCGHAZ by Bayraktar and Kaplan [65].There is an extensive number of HAZ studies across all types of steel chemistries and weldingconditions, most of which are not relevant (i.e. very different chemistry and welding conditions)to this study. The studies with relevant cooling rates in the HAZ and steel chemistries are thoseof Hamad et al. [36], Gianetto et al. [66], Fairchild et al. [67], and Moeinifar et al. [68]. Table 2.116CHAPTER 2. LITERATURE REVIEWFigure 2.6: Schematic of a continuous cooling transformation diagram showing how variousfactors influence the final microstructure, after [47].Table 2.1: Steel chemistries (wt %)C N Mn Si Ni Mo Cr Ti Nb V S P CuHamad et al. [36] 0.06 <60ppm 1.6 - N/A* 0.2 N/A* N/A* N/A* - - - N/A*Gianetto et al. [66] 0.1 <0.0025 1.5 0.18 0.68 <0.03 0.02 0.006 0.027 - 0.004 0.01 0.18Fairchild et al. - A [67] 0.11 0.0041 1.53 0.23 0.12 0.011 0.012 0.011 0.028 0.076 0.001 0.011 0.26Fairchild et al. - B [67] 0.094 0.0049 1.4 0.31 0.014 0.011 0.011 0.048 <0.001 0.005 <0.001 0.008 0.011Moeinifar et al. [68] 0.072 0.004 1.805 0.25 0.26 0.29 0.02 0.012 0.035 0.003 0.001 0.008 0.009reports the steel chemistries of the steels and a review of the cooling conditions and observed mi-crostructures follows.The work by Hamad et al. [36] is relevant to the current study as it was done at an industrysponsor using similar chemistry and weld procedures. In their work, the single torch and bothtorches of a dual torch weld on a X80 pipeline steel used a low heat input of approximately 0.5kJ/mm. While there is limited measured thermal data provided, it is seen that the second torch givesan average cooling rate of 30 ◦C/s (t8−5 of 10 seconds). In their observation it is seen that the dualtorch HAZ microstructures showed significantly larger prior austenite grain sizes compared to thesingle torch samples. The dual torch microstructure also shows to be largely upper bainitic while17CHAPTER 2. LITERATURE REVIEWthe single torch microstructure is lower bainite and lath martensite.Gianetto et al. [66] investigated microstructures that developed in the CGHAZ and ICCGHAZfrom submerged arc welds with two heat inputs. The two heat inputs investigated lead to estimatedaverage cooling rates of 41 ◦C/s (t8−5 of 7.3 seconds) and 21 ◦C/s (t8−5 of 14.5 seconds). CGHAZmicrostructures in both cases gave a bainitic microstructure with some isolated intragranular ferriteregions. With the higher heat inputs a larger prior austenite grain size was seen. In revealing theICCGHAZ it was seen that carbides were seen at prior austenite grain boundaries with some smalleramounts transformed in the grain interior. The carbides in the ICCGHAZ were observed to increasein size with temperature in the ICCGHAZ.In the study by Fairchild et al. [67] the CGHAZ was investigated for different alloys. At anaverage cooling rate of 48 ◦C/s (t8−5 of 6.2 seconds) the microstructure observed in both alloys wasseen to be primarily bainite with an amount of lath martensite. With a slower average cooling rateof 8.8 ◦C/s (t8−5 of 34 seconds) the microstructure was seen to still be primarily bainite but with nomartensite. At an average cooling rate of 5.6 ◦C/s (t8−5 of 54 seconds) both alloys showed to be amixture of different types of ferrites and upper bainite.The CGHAZ and reheated CGHAZ was investigated by Moeinifar et al. [68]. The cooling rateof 3.75 ◦C/s (t8−5 of 80 seconds) resulted in blocky MA as well as MA forming in stringer-likepatterns in the CGHAZ. In the ICCGHAZ it was seen that the blocky MA at prior austenite grainboundaries had coarsened. CGHAZ that was reheated just into the fully austenitic region showedMA content as well, but were less blocky and did not form the thick network at prior austenite grainboundaries as in the ICCGHAZ. For CGHAZ that was reheated below the austenite transformationstart temperature a similar observation is made.2.6 Fracture TestingThe emphasis on preventing failure in pipelines has resulted in the usage of a wide range of teststo asses the fracture properties of linepipe steels, their welds and their HAZs. The Charpy V notch(CVN) [69] test is the most common toughness assessment method used, but other common testsinclude the compact tension (C(T)) [70], single edge bend (SE(B)) [70], and drop-weight tear test(DWTT) [71]. A brief review of the toughness parameters obtained from the CVN, C(T) and SE(B)tests are presented below. A description of the DWTT will not be considered in detail, in spite18CHAPTER 2. LITERATURE REVIEWof the technique being continually used and improved for pipeline fracture testing as its sampledimensions are not favourable for HAZ testing or simulated HAZ testing [72–76].The CVN is an impact test where the energy absorption of breaking the sample, measured byenergy loss of the hammer, is used to quantify toughness. For ferritic steels, tests are done over arange of temperatures; as the temperature decreases, the fracture mechanism may change and theductile-to-brittle transition temperature (DBTT) can be observed with higher DBTT representinglower toughness qualities [77]. The difference of the ductile and brittle fracture mechanisms will bediscussed in a later section. The DBTT can occur over a range of temperatures and is often asso-ciated with significant scatter in absorbed energy values. Alternatively, this transition temperaturecan be determined by the observation of shear area on the fracture surface. In the ductile region,the energy to failure is called the upper shelf energy and is also used as a measure of toughness.Instrumented CVN tests provide load-time data, which is used to determine energy related to crackinitiation (EI) and crack propagation (EP) where the peak load reached separates the portion of theload-time data relating EI and EP [77]. Less commonly, precracked CVN tests can be used to deter-mine the dynamic critical stress intensity factor (KId) using a relation that require measurements ofthe critical stress intensity factor (KIC) from statically loaded tests [77].The C(T) test is the most common test to determine KIC values and involves static loading ofthe sample in tension. With the load being applied by two pins ahead of the notch and precrackthis creates a bending stress state on the remaining ligament with the root of the crack being placedunder tension. Similarly, SE(B) tests are three-point bending tests loaded statically with a notch andprecrack oriented on the side of the specimen that will be in tension. Using two of either the load,crack mouth opening displacement (CMOD) or crack length measurements the KIC value can becalculated [70]. The critical J-integral (JIC) can be determined using C(T) tests, but requires a largesample set or a more difficult procedure with careful measurement of crack extensions [77]. Cracktip opening displacement (CTOD) measurements can be made; these characterize toughness whenthey reach a critical value (δc) [78]. The critical crack tip opening angle (CTOAcr) has been appliedin the ductile fracture of pipelines and it has recently been shown that this value can be obtainedusing load-displacement data from thin-sheet C(T) samples [79]. The review by Zhu and Joyce [80]provides a review of these typical testing methods to obtain these fracture parameters, and on therelationship between these values.19CHAPTER 2. LITERATURE REVIEWThe advantages of the CVN tests are that they are easy to perform, requires a relatively smallamount of material, and minimal preparation. The drawback is that they provide only a relativemeasure of toughness from a dynamic loading situation. Conversely the advantage of C(T) andSE(B) testing is that it provides toughness values that are more robust and transferable to the finalcomponent. Unfortunately, regular C(T) and SE(B) tests require significantly more amounts of ma-terial, and are more difficult to prepare specimens due to having to create a fatigue precrack. Thereis opportunity for a fracture toughness measurement technique that can provide more informationthan a CVN test, but does not require extensive preparation as C(T) or SE(B) tests.2.7 Kahn Tear TestsThe Kahn tear test utilizes notched sheet samples which are then pulled in tension along the sameaxis as the initial root of the notch to initiate then propagate a crack through the sample. Thespecimen is similar to that of a C(T) test in that a tensile load is applied off of the center of thesample. However the differences are that for the Kahn test the initial notch length to uncrackedligament is lower, load is applied in alignment with the root of the machined notch to start and thatthe notch is made with a constant angle from the edge to the root of the notch. The original sampledesign by Kahn and Imbembo was used for investigation of a variety of steel chemistries with platesof thicknesses from 12.5 to 25 mm [81–83]. The test provides load-deflection data which givesthe tear strength (σtear), unit propagation energy (UPE) and unit initiation energy (UIE) of a crack,given by:σtear =4 Pmaxb t(2.2)UIE =initiation energyb t(2.3)UPE =propagation energyb t(2.4)where Pmax is the maximum load, b is the width from notch root to back edge of the specimen,20CHAPTER 2. LITERATURE REVIEWFigure 2.7: Schematic of representative Kahn tear test data showing low and high tear resis-tance results, after [84].initiation and propagation energies are taken as the area under the load-deflection curve prior to andafter maximum load respectively. Figure 2.7 shows how these values are defined.The study by Kaufman and Knoll [84] reduced the dimensions of Kahn’s original specimen inorder to evaluate its potential to provide crack toughness data. Their study on aluminum alloys withthicknesses between 1.52 and 1.65 mm demonstrated a linear proportionality between UPE, and thesum of UPE and UIE from Kahn tests with the critical strain-energy release rate (Gc) determinedfrom middle cracked toughness tests. The ratio of tear strength of the material to the yield stresswas also shown to have a potential relation to Gc. The work by Kaufmann and Knoll constitutes thebasis of the current ASTM standard [85] for the Kahn tear testing of aluminum products.While studies such as those by Kirman [86], and Senz and Spuhler [87] utilize the same analysisas Kaufmann to characterize fracture behaviour between various aluminum alloys, the importantworks by Dumont, Deschamps et al. [88–90] on 7000 series aluminum alloys expands the analysis.Their first work [88] revealed the proportionality between the Gc from regular compact tension testand the UIE of the Kahn tear test. In their major subsequent work [89] they went on to use this datato formulate a model based on energy dissipation per unit area. The model accurately predicted UIEevolution with respect to yield stress of samples quenched slowly after different amounts of aging.However, the model had limited success in predicting the UIE evolution for the tougher samplesobserved when using fast quench rates after aging.21CHAPTER 2. LITERATURE REVIEWThe work by Bron, Besson and Pineau [91] analyzed the Kahn test for aluminum alloys in adifferent way. They considered the crack energy dissipation rate (R) by correlating crack extensionmeasurements to the load deflection data. They observed that the R value increases disproportion-ately more with yield strength for the middle cracked tension (M(T)) samples in comparison to theKahn test. Their plausible explanation is that this is due to the plastic zone size increasing morewith yield for the small scale yielding condition in the M(T) in comparison to the Kahn tests wherethe large scale yielding limits the plastic zone sizes. Fractography demonstrated that the failuremechanisms were identical for both the M(T) and Kahn samples which supports the notion of arelation between the two R values. Bron and Besson [92] subsequently modeled the M(T) and Kahntests by applying continuum damage mechanics based on parameters determined from their earlierwork. The model has some drawbacks, such as the large number of fitting parameters and shows adelay in crack initiation, but does provide fairly good accuracy in the prediction of load and crackgrowth rate. Asserin-Lebert, Besson and Gourgues [93] improved upon the measurement of R fromKahn test by limiting the crack extension measurements to the region of stable crack propagationfor their investigation of specimen thickness on toughness in an aluminum 6056 alloy.The study by Pirondi and Fersini [94] investigated the possibility of using crack tip openingangle (CTOA) measurements as a way to transfer between Kahn and M(T) test for aluminum alloys.They proved successful in applying a finite element model to achieve this relation when steady-stateCTOA values from Kahn testing are compared to those of smaller sized M(T) tests. This is due totheir similar evolution of crack propagation with respect to the ratio of applied force to plastic limitload. For regular sized M(T) tests, such as those used in the Bron et al. study [91], the relation isseen to simply work only after the fully plastic load is reached.The modern Kahn tear test has been shown to be capable for determining fracture propertiesof aluminum alloys but has seen very limited application for testing steels. The work by Devgun,Carrillo and Roosen [95] has applied the modern Kahn tear test for their limited investigation oflow carbon mild steels. Their analysis shows promising relationships between crack initiation andpropagation energies and the yield strength, ultimate tensile strength and Charpy impact values.More recently, the work by Ying et al. [96] used Kahn tear tests in the investigation of temperingtemperature effects on toughness in a high strength boron steel. Their analysis made use of the UPEand σtear, both of which showed a dependence on thickness in the range tested between 1 and 2 mm.22CHAPTER 2. LITERATURE REVIEWTheir work did not investigate the Kahn tear testing technique with any significant detail. Lack ofextensive investigations of the Kahn tear test on steels gives the prospect for novel studies.2.8 Microstructure-Property RelationshipThis microstructure-property relationship defines the variation in properties seen in the HAZ. Theuse of phenomenological models affirm our knowledge of the crucial factors in the microstructure-property relationship, while empirical models based on microstructural measurements also providesinsight into these relationships in absence of a valid physical based model. Gladman [10] hasdiscussed microstructure-property relationships and their modelling for HSLA steels. Of particularconcern to the HAZ is toughness, which is related to the mechanical behaviour (i.e. strength andstrain-hardening). Microstructure relations to the mechanical behaviour are discussed first, followedby their relations to ductile fracture, brittle fracture and the transition between ductile and brittlefracture. As linepipe steels may operate in sub-zero temperatures, considerations of a decreasingtemperature on the microstructure-property relation will be emphasized.2.8.1 Microstructure-Strength RelationshipsDeformation in steel is defined by the resistance to plastic flow (i.e. the yield stress) and the evolu-tion of this resistance with increasing strain, known as the strain hardening behaviour. The relationbetween microstructure and mechanical behaviour can be understood using an isolating mechanismsof strengthening, modeling these mechanisms, then combining the separate models to provide anoverall strength of the microstructure. For a single phase material (e.g. ferrite), most of these modelsapply superposition of separate strengthening mechanisms taking the form of:σtot = σ0 +σss+σppt +σdis+σgb (2.5)where σtot is the strength of the phase, σ0 is the inherent strength of the lattice to dislocation move-ment, σss is solid solution strengthening, σppt is precipitation strengthening, σdis is strengtheningfrom dislocations (both those generated from phase transformation and from work hardening), andσgb is the grain boundary strengthening. Such an approach is used by numerous authors [10,97–102]23CHAPTER 2. LITERATURE REVIEWfor steels.The inherent strength of the lattice to dislocation movement, σ0, can be qualitatively understoodthrough the Peierls-Nabarro stress which is calculated according to [103]:τp =2G1−ν · exp(−2piwb)(2.6)where τp is the Peierls stress, G is the shear modulus of the material, ν Poisson’s ratio, b is themagnitude of dislocation’s Burger’s vector (i.e. the unit slip distance), and w is the width of a dislo-cation. As the Peierls stress is a function of the width of the dislocation it is seen that FCC structureswith large dislocation widths will typically have a smaller Peierls stress while BCC structures withsmall dislocation widths will result in high Peierls stresses. In the case of a high Peierls stress, asin steels, a strong temperature dependence is observed [77]. For BCC structures, the movement ofdislocations occurs through the nucleation of a kink-pair, and then the motion of these kinks [103].The formation of the double-kink pair has an activation energy and it is assumed that this gives riseto the temperature dependence of the intrinsic stress to move the dislocation. Unlike BCC metals,FCC metals have a weakly temperature-dependent Peierls stress.Solid solution strengthening is broadly a complex interaction of chemical, electrical and geo-metrical effects of the solute atom, the parent lattice, and dislocation movement. However it canbe generally stated that the solid solution strengthening effect is a consequence of the lattice di-lation that the solute elements incur [10] and the elastic interaction of the solute atoms with thedislocations. It is typical to simplify the problem by assuming the strengthening due to solutescan be represented as the solute atoms acting as a set of weak obstacles on the glide plane. Thestrengthening contribution for solutes then can take the form of [10, 104]:σss = ∑ki · cni (2.7)where ci is the solute concentration of element i, ki is its strengthening contribution, and n is aconstant between 1/2 and 1. For a low concentration, n is approximately unity [10], otherwise it24CHAPTER 2. LITERATURE REVIEWcan be taken to be 1/2 [105, 106] or 2/3 [107]. Since HSLA steels, such as X80, have low soluteconcentrations this empirical approximation is appropriate at room temperatures, though the valuesof the coefficient ki will vary with the parent phase (e.g. ferrite vs. austenite). In BCC iron,interstitial elements (i.e. C and N) impose an asymmetric strain (i.e. both hydrostatic and shearcomponents) which allows them to interact with both edge and screw dislocations making them themost potent solid strengtheners by weight. Substitutional solute atoms give rise to only hydrostaticdistortions and will interact primarily with edge dislocations. By weight, it is seen that P is a strongstrengthener (though the mechanism appears to be unrelated to lattice dilation for this element),while Si, Cu, and Mn are seen to be moderate solid solution strengtheners. Mo is seen to be tobe a weak strengthener, while Ni is seen to have little or no solid solution strengthening effect. Cractually shows a negative contribution to solid solution strengthening, though it is unclear if this isa result of it removing solute N from solid solution, or due to electron bond weakening of the latticeby its presence [10]. This highlights the need to differentiate between the alloy content levels andthe portion of that content that is in solid solution or in precipitates.Precipitates are present in HSLA steel and can cause strengthening through a number of possiblemechanisms. In HSLA steels the primary particles of interests are carbides (e.g. Fe3C, MoC,NbC) or nitrides (e.g. NbN, TiN), or complex carbonitrides (e.g. Nb(C,N), (Nb,Ti)(C,N)). Thestrengthening of these precipitates relate to how they interact with dislocations, which categorizesprecipitates into those shearable and non-shearable by a dislocation. Also important is the size,spacing, and distribution of precipitates. The Orowan-Ashby equation considers only non-shearableprecipitates, which is appropriate for hard carbides, nitrides and carbonitrides, and takes the formof [108]:σppt =0.538Gb f 1/2X· ln(X6.125×10−4)(2.8)where G is the shear modulus, b is magnitude of the Burgers vector, f is volume fraction of pre-cipitates, and X is the diameter of precipitates. As the activation energy for a dislocation energy tobow or bypass is quite large, there is effectively little effect of temperature on precipitate strength-ening [109].25CHAPTER 2. LITERATURE REVIEWThe strength to move a dislocation increases with dislocation density according to the wellknown Taylor relation [110–113]:σdis = αMGbρ1/2dis (2.9)where α is a constant dependent on temperature, M is the Taylor factor and ρdis is the dislocationdensity. Due to volumetric changes during austenite decomposition, certain phases have been ob-served to have high dislocation densities (e.g. martensite, acicular ferrite) prior to any extra plasticdeformation of a sample. The generation, storage and annihilation of dislocations during plastic de-formation leads to a change in the dislocation density giving rise to the strain-hardening behaviourobserved in stress-strain curves. The evolution of dislocation density can be described by the modelof Kocks and Mecking [114]:dρdisdεp= k1ρ1/2dis − k2ρdis (2.10)where εp is plastic strain, k1 is a constant related to dislocation storage, and k2 is a constant relatedto the dynamic recovery of dislocations. While dislocation storage is not temperature dependent,dynamic recovery is via thermal activation aiding in the cross-slip of screw dislocations allowing fordislocations to meet and annihilate. As a result, with decreasing temperatures less dynamic recoveryis expected which should increase the strain-hardening response (i.e. an increase in strength for acertain amount of plastic deformation for lower temperatures).The well-known Hall-Petch relation is used extensively for determining strengthening due tograin size, given by [10, 113]:σgb = kH−P d−1/2 (2.11)where kH−P is the Hall-Petch constant and d is the grain size. There have been a number of proposedtheories to explain the Hall-Petch equation, however the physical basis for the relation remains un-26CHAPTER 2. LITERATURE REVIEWcertain [113]. Nonetheless, the Hall-Petch equation adequately describes the grain size strengthen-ing effect for more practical application. The Hall-Petch constant kH−P has been observed to varywith alloy contents and with test temperature. Note that in the case of a complex microstructure dmay be related to other specific features of the microstructure, such as the lath size in bainite.As it is common to have more then one phase present in steel (e.g. MA is always present inat least some amount), there must be a consideration for determining the composite strength of themicrostructure. One example comes from Bouquerel et al. [100] who determined the overall stress-strain behaviour of a transformation-induced plasticity steel by using stress mixture laws. Theirapproach was more detailed than most as they computed the strengthening for each of the polygo-nal ferrite, bainitic ferrite, retained austenite and strain-induced martensite phases using a methodsimilar to equation 2.5 prior to invoking rule of mixtures. As the rule of mixtures is typically usedto determine the composite strength of a steel, understanding the strengthening mechanisms withineach phase provides insight into the overall microstructure-mechanical behaviour relationship.2.8.2 Microstructure-Fracture Behaviour RelationshipsFracture occurs either through a process with extensive plastic deformation, referred to as ductilefracture, or with minimal plastic deformation, referred to as brittle fracture. In ductile fracture, alarge amount of plasticity occurs prior to failure and requires relatively high amounts of energy. Inbrittle fracture, the amount of plasticity is limited and will require relatively low energy to frac-ture. While brittle fracture is absolutely undesirable, ductile failure in pipelines can lead to runningcracks under certain conditions, and thus ductile fracture properties are also of concern [115]. It isimportant to understand how the microstructure affects toughness with respect to ductile fracture,brittle fracture and the transition from the ductile to brittle fracture.Ductile fracture occurs via the nucleation, growth and coalescence of voids. The nucleation ofthese voids are associated with inclusions or second phase particles (e.g. precipitates, martensiteislands) through particle cracking or interface decohesion [116]. As these voids grow, the volumefraction of voids is found to progress with a dependency on the initial volume fraction and shapeof voids, which in turn are related to the fraction and shape of particles or hard phases actingas void nucleation sites [117]. At some point in the void growth process, mechanical stabilityin the remaining ligaments is lost resulting in voids coalescing and a ductile crack forming then27CHAPTER 2. LITERATURE REVIEWpropagating from void to void [116]. The strain to final fracture has been observed to decreasewith increasing volume fraction of particles or hard phases, and as the shape of these change fromelongated in the tensile loaded direction to elongated in a direction normal to the tensile direction[116].Brittle fracture in steels typically occurs through a cleavage mechanism. Cleavage, as describedoriginally by Cottrell, occurs in a crystal through a separation along specific crystallographic planeswhen stress reaches a critical level [10]. The nucleation of a crack may be associated with stressintensification processes such as particle (e.g. carbonitrides) cracking, debonding between phases,or at dislocation pile-ups. At a critical stress, this initial crack will propagate across the grain.Depending on the temperature and the grain size, this cleavage crack across a grain may be arrestedby a grain boundary until the applied stress along with the stress concentration at this crack tip israised enough to cause cleavage in the next grain [118]. Consequently it is seen that the stress tocause cleavage fracture is inversely proportional to the grain size [77]. This is important as whileother strengthening mechanisms do not also improve cleavage fracture resistance, microstructurerefinement provides an improvement both to strength and cleavage fracture resistance.Brittle fracture is of particular concern to the HAZ of pipeline steels. Measured DBTT fromCVN tests are typically used to determine susceptibility to brittle fracture. This transition tempera-ture has typically been characterized by empirical models, for example one used by Koo et al. [99]for high strength pipeline steels is given by:DBTT = F(σ0 +σss+σppt +σdis)− kCVN d−1/2e f f (2.12)where F and kCVN are constants and de f f is the facet size in cleavage, which is the grain size in asimple ferritic structure, or related to the size of low-angle boundary domains in complex structurescontaining bainite or martensite.As the stress concentration of a crack causes the development of a plastic zone which influ-ences the stress field, the stress-strain behaviour including strain-hardening is important to consider.Strain-hardening considerations have been applied previously, and are well reviewed in the book byThomason [116]. It is seen that the strain-hardening behaviour is included through Hollomon-based28CHAPTER 2. LITERATURE REVIEWequations. To this author’s knowledge, only the work of Spa¨tig et al. [119] applied the Kocks-Mecking view of strain-hardening to an assessment of fracture. Their study on a ferritic-martensiticsteel used Kocks-Mecking analysis in part to determine constitutive temperature behaviour of theirsteel. Subsequently, they used a 2D FEM model of a C(T) test in the plane strain condition in com-bination with the determined constitutive behaviour to estimate the stress-strain field ahead of thecrack tip. They considered cleavage to occur when a critical tensile stress is reached over a criticalarea ahead of the crack tip. Comparison of the resulting effective critical stress intensity (Ke f f )from the model with experimental data showed good predictions at temperatures below the DBTT.This highlights the practical use of investigating the relations between microstructure, mechanicalbehaviour, and fracture.2.9 HAZ Mechanical Properties & Fracture ToughnessAs mechanical and fracture properties are directly related to microstructure, and the HAZ is noted tobe a gradient of microstructures, it can be expected that a gradient in properties will exist in the HAZ.Testing to determine properties in a HAZ subregion from an actual weld is not straightforward, asthe microstructural gradient changes over a very small distance. Mechanical properties are mostcommonly investigated using hardness measurements, which provide only limited information [47].While there are studies that have investigated other methods of testing, such as subsize or microtensile testing [120–122], microshear testing [123], and shear punch testing [124], they are subjectto potential errors by testing a gradient in microstructure across the HAZ and have not seen wideacceptance. Typically CVN and SE(B) tests are used to investigate real HAZ toughness properties.It has been discussed by many authors [36,47,125–127] that placing a fatigue crack or even a notchin the correct subregion of a real weld HAZ is extremely difficult. Measurements based on thesetechniques are clearly sensitive to sample preparation and may not properly isolate an individualsubregion of the real HAZ. As such it is common to make use of laboratory thermal simulations toinvestigate the microstructures and mechanical properties of linepipe steels.In the investigations of steel HAZ’s, it has been noted that certain subregions are associated withsignificantly lower toughness. These areas of low toughness are most often associated with eitherthe CGHAZ or ICCGHAZ [62–65, 68, 128–131], and it is seen in the majority of HAZ studies thatthe presence of MA is associated with low toughness [62–65, 131–137]. Some studies have found29CHAPTER 2. LITERATURE REVIEWother sources of low toughness, such as that by Fairchild et al. [67] which showed TiN particlesinitiated brittle fracture in the CGHAZ, or that by Ohya et al. [128] which found cracks in theCGHAZ and ICCGHAZ to initiate at the intersections of bainite packets (though often near largeparticles).The work by Lee et al. [61] showed a decrease in the dcr and CVN values for the actual CGHAZcompared to the base plate. Their study also showed decreased CVN energy values in the laboratorysimulated CGHAZ and ICCGHAZ in comparison to all regions of the HAZ and they correlated thedecrease in toughness values to the increase in volume fraction of MA constituent.Shi and Han [132] investigated a 800 MPa HSLA steel using laboratory thermal simulationswith a single peak temperature and cooling rates from 6 seconds to 240 seconds to generate CVNspecimens. They found a decrease in toughness with increasing austenite grain size and MA content.While mentioned only briefly, they noted that as MA content increases the morphology changesfrom elongated to blocky in shape. They also found that peak temperatures in the ICHAZ produceda network of MA at grain boundaries and deteriorated the toughness severely.The work by Davis and King [62] used CVN tests on laboratory simulated ICCGHAZ andshowed that it is not simply the amount of MA, but also its distribution, morphology and hardnessdifference in comparison to the matrix phase which are important to brittle fracture initiation. Themost detrimental situation is reported to be a near-connected grain boundary network of largerblocky shaped MA which is substantially harder than the matrix.Li and Baker [136] investigated the effects of MA morphology on fracture toughness of ICCG-HAZ using laboratory thermally simulated CVN and CTOD tests. They found that both crackingand debonding of the MA phase occurred. Importantly, they found that in regions of high triaxialstress states that cracking of blocky MA phase is likely to occur whereas low triaxial stress statesresult in decohesion of the elongated stringer MA which leads to crack formation. They also foundthat carbides and inclusions did not significantly influence the toughness properties.The thesis by Mohseni [138] investigated the DBTT in the CGHAZ, and ICCGHAZ of a X80steel using SE(B) for CTOD measurements and CVN on laboratory simulated specimens. In hiswork, the ICCGHAZ showed lower fracture toughness values than the CGHAZ at each temperaturetested. Close examination of the fracture surface revealed that MA was observed at the initiationpoints of cleavage cracks, and that these initiation sites were either due to MA debonding or from30CHAPTER 2. LITERATURE REVIEWa region between two closely-spaced blocky MA constituents with the ICCGHAZ showing moresevere structures.In the work by Goodall [139] laboratory thermal simulations were used extensively in the ex-amination of HAZ transformation of X80 and X100 steels. In one part of his work, a X80 steel wassubjected to two thermal cycles that simulate a dual torch weld where the torch spacing is so closethat the second thermal cycle starts partway through the austenite decomposition of the first thermalcycle. The peak of the second thermal cycle for these tests did not reach full austenitization temper-atures. That is, the heating of the second thermal cycle occurs between Ar1 and Ar3 of the first cycleand reaches a peak between Ac1 and Ac3 in the second thermal cycle. Thermal cycles were based ona Rykalin 3D models reaching an initial peak of 1350 ◦C giving an average cooling rate of 50 ◦C/sfor the first peak and average cooling rate of 25 ◦C/s for the second peak. This was compared withsamples simulating CGHAZ and ICCGHAZ with the same cooling profiles. From CVN testing interms of the DBTT the CGHAZ gave the best toughness, ICCGHAZ gave lower toughness, and theinterrupted ICCGHAZ gave the lowest toughness. The ICCGHAZ did show slightly greater uppershelf energy values compared to the interrupted ICCGHAZ samples. This difference was explainedin terms of the lack of a hard MA phase along with acicular ferrite becoming a constituent in theinterrupted ICCGHAZ giving rise to a slightly higher upper shelf energy for the ICCGHAZ, whilethe coarseness of the interrupted ICCGHAZ led to the higher DBTT.The study by Hamad, Collins and Volkers [36] used relevant steel and welding conditions tothis study. They tested the real CGHAZ of single, tandem and dual torch girth joints using CVNand CTOD from C(T) tests and showed δc and CVN values for upper shelf energy and DBTT werepoorest for the dual torch process. The δc values were subject to scatter which they attributed todifficulties in crack placement or testing near the DBTT in the case of dual torch samples. Theyreported no significant variation from sample to sample in amounts of MA, segregation and largeparticles such as TiN which otherwise could explain the scatter. The study suggests that the largerprior austenite grain sizes helps shift the DBTT to higher temperatures, but does not provide amechanism for why this occurs.So far discussion has centered on toughness measurements of the HAZ. Mechanical propertyinvestigations of the various subregions of the HAZ are typically limited to hardness measurements.Recently only works by Zhao et al. [140] and Qiao et al. [141] used laboratory thermal simulations31CHAPTER 2. LITERATURE REVIEWto generate tensile specimens for investigation. The work of Zhao et al. [140] on a X90 steel pro-vides engineering stress-strain curves for their specimens. Typical properties of yield stress, tensilestrength and total elongation are plotted against the heat input used to calculate the thermal cycle,and a decrease in tensile strength with heat input is observed. Despite having full tensile test data onX100 thermally simulated samples, Qiao et al. [141] simply report tensile strength values comparedto peak temperature for the different simulated process parameters. They show that a minimumof strength is observed at a peak temperature of 950 ◦C for each heat input case which is relatedto the FGHAZ. The lack of detailed investigation of the mechanical properties in the HAZ givesopportunity for novel work to be introduced.32CHAPTER3Scope and Objectives3.1 ScopeThis study investigates the mechanical behaviour of a X80 linepipe steel with microstructures pro-duced that are relevant to a typical girth weld HAZ. The study uses a commercially produced pipe,specifically a Mackenzie Valley prototype strain based design X80 (650 MPa yield stress) steel.The welds of interest are GMAW produced girth welds with industrially relevant welding condi-tions (e.g. similar torch speed, heat input, number of passes). Furthermore, there is interest incomparing single torch welds to high productivity dual torch welds with different torch spacings.The mechanical properties of the HAZ microstructures that are of interest include the yield strength,work hardening behaviour, and tear resistance. These properties are to be investigated within therange of potential operating temperatures (20◦C to -60◦C).3.2 ObjectivesThe objective of this work is to investigate the mechanical behaviour of microstructures that may befound in the HAZ adjacent to industrially produced GMAW girth welds. To achieve this objective,a number of related subobjectives were defined, summarized below.• Characterize the time-temperature profile in the HAZ of typical girth welds for both singletorch and for dual torch with different spacing.33CHAPTER 3. SCOPE AND OBJECTIVES• Develop the thermal cycle profiles for Gleeble heat treatments to reproduce relevant HAZmicrostructures.• Produce these microstructures in bulk specimens suitable for mechanical testing.• Characterize the microstructure of the Gleeble heat treated specimens.• Measure the tensile properties and tear resistance (using the Kahn tear test) of the thermallytreated specimens at ambient temperature, -20 ◦C, and -60 ◦C.• Characterize the fracture surfaces of both tensile and Kahn specimens• Develop a model for the work hardening behaviour that relates to the characteristics of themicrostructures observed and is applicable to all temperatures testedIn the achievement of these objectives, there is an impact to the current state of scientific andtechnical knowledge. Measurements of the HAZ thermal cycle profile, particularly with respect todual torch welds represent a novel data set. The tensile and tear property measurements of specificmicrostructures found to be relevant to the HAZ furthers understanding. The development of awork hardening model represents an extension to the current state of work hardening models. Onthe technical side, HAZ temperature measurements as done in this study are a technical achieve-ment in terms of their accuracy. The creation of bulk specimens suitable for testing required therefinement of Gleeble heat treatment techniques. This also produced a new set of data regardingtensile mechanical properties relevant to HAZ microstructures. The use of the Kahn tear test forHAZ steel specimens is a novel approach.34CHAPTER4Methodology4.1 MaterialThe material used in this work is an X80 grade linepipe steel supplied by Evraz Inc NA. The chemi-cal composition of this steel is given in Table 4.1. Using Nital and Le Pera etchings on this material,Tafteh [142] revealed the as-received microstructure of this steel, shown to be 97.5 % ferrite and2.5 % martensite and/or retained austenite. In the work by Banerjee et al. [143] on this steel it wasseen that there are three main types of precipitates in the as-received condition. There are largecuboidal TiN particles (with radii of approximately 61 nm) which remain stable through the hightemperatures [143]. Nb(C,N) particles are seen in both large (approximately 69 nm) and small(approximately 2 nm) sizes [143]. Mo2C particles are also observed with an average diameter ofapproximately 53 nm [143].Table 4.1: Steel chemistry (wt %)C Mn Nb Mo Ti N0.06 1.65 0.034 0.24 0.012 0.0054.2 MetallographyTo investigate the microstructures of this steel, samples were ground, polished, then etched. Wetgrinding was performed progressively on 60, 180, 300, 420, 600, 800, and 1200 grit paper. Polishing35CHAPTER 4. METHODOLOGYwas then carried out using 5 µm then 1 µm diamond polish and then rinsed with distilled water anddenatured ethyl alcohol. Two types of etching were carried out after the grinding and polishing,Nital and Le Pera. Nital etching uses a solution made of 2% by volume of concentrated nitric acid inabsolute ethyl alcohol, with concentrated nitric acid being comprised of approximately 70% HNO3by weight in water. The sample was then immersed for approximately 30 seconds and then rinsedwith distilled water and denatured ethyl alcohol. Le Pera etching involves two solutions that aremixed immediately before etching the sample. The first solution was a 4% weight/volume of Picricacid in absolute ethyl alcohol, and was left to stir slowly while polishing the sample. The secondsolution was a 1% weight/volume metabisulphite in distilled water, also made prior to polishing thesample. After the polishing and rinsing was complete, the sample was heated using a hair dryerwhile mixing and gently swirling 5mL of each of the two solutions together. The sample was thenimmersed for approximately 45 seconds. If the sample’s microstructure was not fully revealed forboth Nital or Le Pera etches, immediate reimmersion for short intervals was sometimes performed.If that proved unsuccessful the sample was ground, polished and re-etched starting from 800 gritpaper.Nital etching reveals the complex mixture of ferrite, upper bainite and lower bainite found inthe microstructures of these steels. For this study, Figure 4.1 exemplifies how these constituentsare classified. Irregular ferrite appears to have grains with irregular shape and no internal structurevisible. Upper bainite shows packets of grains with an internal structure of aligned but disconnectedparticles or carbides. Lower bainite shows packets of grains with an aligned and very fine structure.In order to quantize the amounts of these constituents, a point counting technique was used. From asample, multiple images (five at the minimum) taken at 500X magnification have a grid overlayedof them, and then the type of microstructures at the intersections are counted.Le Pera etching reveals the martensite and retained austenite (MA) as bright islands against adarker matrix. The difference between martensite and retained austenite is not revealed using thistechnique. To measure the amount of MA, multiple images (again, five at the minimum) takenat 500X magnification are image-processed then measured. A brightness threshold was appliedusing the GNU Image Manipulation Program (GIMP) using the linear scale, which gives a blackbackground with the MA phase that passes the threshold in white. The image was cropped toremove portions of the image that are darker in the outer regions due to vignetting. In the case36CHAPTER 4. METHODOLOGYFigure 4.1: Classification of observed microstructures:(a) Irregular ferrite microstructure (b)Upper bainitic microstructure (c) Lower bainitic microstructure [144]where it was difficult to identify a singular best threshold, two thresholding images are made toset the upper and lower bounds and analysis on both proceeds as follows and was then averaged atthe end to determine a singular value. The thresholded image was loaded into ImageJ, set to 8-bit,then a threshold setting of (129,255) was applied, and then was converted to a mask. After thispreprocessing, the ”Analyze Particles” command was used for particles with sizes from 0 to infinityand circularity from 0 to 1. This gives an area fraction of MA. The MA fraction was then subtractedfrom the fraction of the most appropriate phase as measured from the Nital quantization. Figure 4.2shows the thresholding process and match up for a typical Le Pera image.37CHAPTER 4. METHODOLOGYFigure 4.2: Le Pera thresholding method: (a) Initially cropped Le Pera image (b) Thresholdedimage display MA as black (c) Lower threshold overlayed as red on initial image givesMA content of 16.7% (d) Higher threshold overlayed as red on initial image gives MAcontent of 11.2%.38CHAPTER 4. METHODOLOGY4.3 Weld TrialsWeld trials to measure the thermal cycles in the HAZ of a typical girth weld were performed withthe assistance of Evraz Inc NA. Initially, Evraz supplied a cross-section of the a typical weld whichwas then Nital etched to reveal the HAZ and weld microstructure. This guided experimental designfor the weld trial. The welds were done to join 16 mm thick flat plates of X80 steel with a lengthof approximately 610 mm along the welding line, and a width of approximately 305 mm. Sixholes were drilled into the plates, spaced 50 mm from the end of the plate, 25 mm away from eachother along the weld line, and at an angle close to the bevel so that the end of the hole was placedinto the expected HAZ at different distances to the fusion line and at different depths. During thewelding trial, type S for their high temperature measurement range, or N thermocouples for theirhigh temperature stability but lower cost, were spot welded to the end of the drilled holes and wereprotected by a ceramic sheath. A schematic of the weld trial setup is shown in Figure 4.3, withdetails of the setup given in Table 4.2.Figure 4.3: Schematics of (a) an embedded thermocouple in a plate ready for welding and (b)approximate placement and labelling of thermocouples.Welding trials were performed at Evraz’s Regina research and development facility using theirrobotic welding setup. Single torch welds as well as dual torch welds with inter-torch spacings of2.75”, 4”, and 7” were performed to simulate the girth welding procedures for pipe as specifiedby TransCanada Pipelines Ltd. During the welding passes thermocouple data was recorded at asampling rate of 244 Hz. As placement of the thermocouples were done to fit the wide range of HAZ39CHAPTER 4. METHODOLOGYTable 4.2: Weld trial thermocouple placementWeld Type Thermocouple # Relative depth from bottom Distance from bevel (mm) Thermocouple typeSingle Torch1 3/8 1.75 S2 3/8 2.25 N3 1/2 1.75 S4 1/2 2.25 N5 11/16 1.40 S6 11/16 1.90 NDual Torch1 3/8 1.25 S2 3/8 1.75 N3 1/2 1.15 S4 1/2 1.65 N5 11/16 1.50 S6 11/16 2.00 Nfor different torch spacings, and considering there is variation from test to test, multiple tests for eachtorch spacing was performed to improve chances of obtaining useful thermal cycle data. Weldingprocedures used in the trial, guided by those used in actual procedures by TransCanada PipelinesLtd., follow the pass schedule schematically shown in Figure 4.4 and the details are reported inTable 4.3. Note the different heat inputs for the single and dual torch procedures.Figure 4.4: Weld pass definition for (a) single torch and (b) dual torch welds used in this study.Table 4.3: Welding trial proceduresSingle torchPass Root Hot Fill 1 Fill 2 Fill 3 Fill 4 CapHeat Input (kJ/mm) 0.23 0.15 0.26 0.26 0.26 0.26 0.26Travel Speed (mm/s) 12.5 23 17 17 17 17 17Preheat/Interpass (◦C) Room - 100 100 100 100 100Dual torchPass Root Hot Fill 1 lead Fill 1 trail Fill 3 Cap lead Cap trailHeat Input (kJ/mm) 0.2 0.17 0.57 0.57 0.56 0.36 0.36Travel Speed (mm/s) 12.5 23 8 8 7 8.5 8.5Preheat/Interpass (◦C) Room - 100 - 100 100 -Signal processing was conducted on the temperature-time data from the weld trials. First, a40CHAPTER 4. METHODOLOGYParks-McClellan filter was applied to remove noise from the raw signal, passing signals with fre-quencies under 12.2 Hz and completely attenuating noise with a frequency of 24.4 Hz and above(which represent 5% and 10% of the sampling frequency respectively). It was seen that there wasa voltage shift of the measured data due to the activation of the welding equipment which wascorrected for simply by addition of the shift to data for the duration of the welding equipment acti-vation.4.4 Gleeble Heat TreatmentsHeat treatments were carried out using the Gleeble 3500 thermomechanical simulator at UBC. Thisallowed for the design of complex thermal cycles to create microstructures relevant to the HAZ.The design of the large strip testing sample used is shown in Figure 4.5. Samples were machinedfrom the supplied X80 plate to a size of 200 mm X 50 mm X 1.20 mm with the length of theGleeble specimen aligned in the rolling direction of the plate. Thermocouples were spot welded tothe top surface to measure and control the temperature during the heat treatments. The center S-typethermocouple were to control the Gleeble’s system to meet the programmed thermal cycles with theother, K-type thermocouples providing data on temperature gradients in the sample. The thermalcycles used are shown schematically in Figure 4.6. The experimental design was chosen to producemicrostructures consisting of primarily one constituent relevant to the microstructures observed inthe HAZ. This was accomplished by varying between two grain sizes of austenite prior to cooling,between the precipitate-solid solution conditions of Nb, and between two cooling rates.The austenite grain sizes produced in the final thermal cycle were 5 µm and 42 µm as determinedin the work of Tafteh [145] for the same thermal cycles. The conditions of Nb were to be primarilyin solid solution or primarily in precipitates. In the case of Nb in solution, this required two thermalcycles, the first being to dissolve the Nb(C,N) particles and put Nb into the solution. In the caseof a prior austenite grain size of 42 µm with Nb in precipitates, solutionizing was required in orderto achieve the prior austenite grain size. For this condition, the final thermal cycle includes a 20minute hold at an elevated temperature (900 ◦C) in the fully austenitic regime but below the Nb(C,N)formation temperature in order to precipitate an amount of Nb out of solution. While initially it wasthought that this time would be enough to nearly fully precipitate, recent work by Reichert [144]has shown that this precipitation cycle results in approximately 50% of the Nb being precipitated.41CHAPTER 4. METHODOLOGYFigure 4.5: (a) Schematic of a Gleeble large strip design and (b) schematic of where mechan-ical testing specimens are removed from the strips.While this condition will be referred to as Nb being in precipitates, it is important to keep in mindthat the Nb will only partially be in precipitate form. The cooling rates selected were 10 ◦C/s and 50◦C/s. In order to cool the strip at these rates, the sample is quenched along the bottom of the sampleusing compressed helium set at a valve pressure of 12 psi and applied using a three-nozzle system.In order to prevent the strip from buckling during the testing, the Gleeble’s air ram was set at 1.0kNin tension.After Gleeble heat treatments, electrical discharge machining (EDM) was used to produce atensile and Kahn specimen from the original sample as in Figure 4.5 (b). Samples for metallography(Nital and Le Pera etching) were taken from the material between the root of the Kahn notch andthe middle of the tensile specimen.4.5 Tensile TestingTensile tests were carried out on subsized specimens as shown in Figure 4.7. Ambient temperaturetests were carried out using an Instron 8872 testing machine with an applied crosshead speed of1.5 mm/min which gives an approximate strain rate of 2 x 10−3 s−1. Testing at -20 ◦C and -60 ◦Cwere carried by immersing the samples in an ethanol-glycol mixture cooled to the point of freezingby dry ice. The exact mixture of ethanol-glycol for a given testing temperature was determined by42CHAPTER 4. METHODOLOGYFigure 4.6: Thermal cycles for (a) Nb in precipitates with 5 µm austenite grain size, (b) Nb inprecipitates with 42 µm austenite grain size, (c) Nb in solution with 5 µm austenite grainsize, and (d) Nb in solution with 42 µm austenite grain size.trial and error. The solution temperature was monitored throughout the testing process and foundto be very stable, keeping within 2 ◦C of the aim temperature. Cold temperature tests were carriedout on a Instron TM-L load frame with screw driven crosshead speed of 0.51 mm/min resulting inan approximate strain rate of 1 x 10−3 s−1. An extensometer was used in all cases to measure thedisplacement in the gage of the specimen, including being immersed when needed.Figure 4.7: Schematic of the subsized tensile specimen.43CHAPTER 4. METHODOLOGYIn this study, the yield strength that will be used differs from the regular definition. Whereas thenormally used definition of yield strength is based on the 0.2% offset method, this study will use thesame offset method but with a 0.5% offset instead. As it will be shown, the initial portions of thestress-strain curves are very rounded. During the course of this investigation it was found that the0.2% offset yield strength did not provide consistent measurements for these rounded stress-straincurves while the 0.5% offset yield strength did. The use of 0.5% yield strength results in a higherreported yield strength compared to the normal 0.2% and the reader should be aware of this whenmaking comparisons of the yield strength in this study to those from other studies.After testing, the fracture surface was examined using secondary electron imaging on a HitachiS-2300 scanning electron microscope. From low magnifications images, an area measurement ofthe final fracture surface is made. This is used to determine the final fracture stress and the finalfracture strain, computed from:σ f rac =Ff racA f rac(4.1)ε f rac = lnA0A f rac(4.2)where σ f rac is the stress at fracture, Ff rac is the load at fracture, A f rac is the fracture surface area,and A0 is the initial cross-sectional area.In the determination of fracture stress using this equation, the triaxial stress state in the neck isnot taken into consideration. To do so requires a correction. To correct for the effect of triaxiality inthe neck of these specimens at fracture, a finite element analysis of the tensile specimen was madeby Puydt [146]. This model used the true stress-strain curve from the as-received condition. Usingthis data, he used the point of necking to the point of observed fracture to linearly extrapolate the truestress-strain data from the point of necking. The original stress-strain data with this extrapolationwas used as an input for the model to be fit. The output of the model is a true stress-strain curvecorrected for triaxiality. Note that this curve is the same up until the point of necking, then diverges44CHAPTER 4. METHODOLOGYfrom this point. Figure 4.8 shows the true stress-strain curve from the input and the modeled truestress-strain curve in the neck. It is seen that the model true stress in the neck is higher than theinput true stress, as expected. The difference between the model corrected true stress and the inputtrue stress can be represented as a ratio. Figure 4.9 is the ratio of the computed true stress to themodel true stress versus true strain. It can be seen in this figure that after a strain of about 0.9 aplateau of this ratio is reached at a value of 1.14. This factor, 1.14, then is used in the correctionfrom the measured values of true fracture stress to account for the effect of triaxiality in the neck.An example of the effect of this correction of the fracture stress on the true stress-strain curve isshown in Figure 4.10.Figure 4.8: True stress-strain of model input based on the measured as-received conditioncompared to model output.Figure 4.9: Ratio of model output to the input based on the measured as-received condition.45CHAPTER 4. METHODOLOGYFigure 4.10: Example of resulting true stress-strain data without and with correction to thefinal fracture point.4.6 Kahn TestingTo assess fracture properties, Kahn tests were used. The Kahn specimen, shown in Figure 4.11, waspulled in tension at a continuous rate of 0.4 mm/min which results in the initiation and propagationof a crack. Kahn tests were performed on a MTS 312.21 servohydraulic testing machine. Load linedisplacement was measured by attaching an extensometer to the pins passing through the specimen.Figure 4.11: Schematic of the Kahn tear test specimen.46CHAPTER 4. METHODOLOGYIn order to investigate the effect of thickness on the properties measured, tests were carried outon the as-received material with additional thicknesses of 1.5 mm, 2.0 mm, and 2.5 mm. Lowtemperature testing were carried out using the same immersion solution of ethanol-glycol cooledby dry ice. As with the tensile tests, the fracture surface was examined using secondary electronimaging on a Hitachi S-2300 scanning electron microscope.Analysis of the Kahn specimens uses the load-displacement data. A schematic of typical Kahnmeasurement data is shown in Figure 4.12. Tear strength (σtear) for this loading geometry is calcu-lated according to the equation:σtear =4 PmaxB0 t(4.3)where Pmax is the maximum load, B0 is the uncracked ligament length, and t is the specimen thick-ness.Figure 4.12: Schematic of data measured during the Kahn tear test.Also using the load-displacement, the unit propagation energy from the load-displacement curvefor a displacement of 1 mm from the maximum load (UPEd0−1) is calculated according to:UPEd0−1 =∆UpB0 t(4.4)47CHAPTER 4. METHODOLOGYwhere the ∆Up is the propagation energy as determined by the load-displacement energy calculatedby numerical integration using the trapezoidal rule from the displacement at maximum load to 1 mmfrom that displacement. The use of 1 mm for this measurement comes from the fact that bucklingof the specimen was observed to occur in specimens sometime after this displacement.While UPEd0−1 values represent a quick method to investigate crack propagation energies, amore rigorous method was determined to be desirable in order to validate the usage ofUPEd0−1. Tothat end, crack length measurements were used in the calculation of the tearing energy, Etear whichis calculated by:Etear =1t∆Up∆a(4.5)where t is the specimen thickness, ∆Up is the load-displacement energy calculated over the rangeof the crack length measurement, and ∆a is the change of the crack length at two measured lengths.∆Up is the propagation energy calculated as discussed previously. The crack length measurementswere done for samples at ambient temperatures. To measure the crack length for these tests, testswere interrupted intermittently past peak load by holding crosshead displacement constant. Withthe crosshead displacement held constant, the crack length is measured by observing the distancebetween position notch root and tip as observed by a travelling optical microscope. For each setof thicknesses, a polynomial fit to the crack length-displacement data from the point of maximumload and 1 mm of displacement is made. Using this polynomial fit, Etear is calculated from a= 0 toa= 3 mm. For cold temperature tests, crack measurements were not possible. It is assumed that thecrack growth-displacement relations measured at ambient temperatures will be valid throughout allthe temperatures tested in this work so that the Etear values can be calculated for all tests.48CHAPTER5Results5.1 Welding TrialsIn the welding trial, a total of nine welds were performed with successful temperature measure-ments. These consisted of at least two repeats of single torch, 2.75”, 4”, and 7” dual torch spacings.Cross-sections at the thermocouples were made, polished and then etched in 2% Nital to reveal themicrostructures and to determine the position of the thermocouple relative to the fusion line. Fig-ure 5.1 shows example macrographs and micrographs of a single torch and a 2.75” dual torch weldsection. These two conditions show the smallest and largest HAZs with widths from the fusionline of approximately 1.2 mm and 2.6 mm respectively. The micrograph also shows that in thesecases the thermocouples were placed within the HAZ, close to the fusion line. This is not true of allthermocouples, however, as in a number of cases where thermocouple measurements were lost dueto molten weld metal penetrating the thermocouple hole. For surviving thermocouples it is seen thatnot every welding pass results in a HAZ measurement, which is expected as the initial placementwas planned to measure the weld passes in the middle of the plate. For single torch welds, thermalcycle measurements in the HAZ are associated with the second and third fill pass. For the dualtorch welds, it is primarily the first fill pass, which is a dual torch pass, that results in thermal cyclemeasurements within the HAZ.Examples of measured temperature-time data and its associated filtered and corrected thermalcycle data from the weld trial for a single torch and a 2.75” dual torch case are shown in Figure 5.2.49CHAPTER 5. RESULTSFigure 5.1: (a) Single torch weld macro picture and micrograph, (b) 2.75” dual torch weldmacro picture and micrograph.The correction applied to the raw data is two fold, as discussed previously in Section 5.1. First,filtering for noise from the welding process is applied, where Figure 5.2 clearly shows the weld on-off times by the appearance of this noise in the raw signal. Once the noise is removed a clear drop inthe thermocouple signal is observed while the welding is occurring, which is then corrected for byadding the difference between the initial average signal temperature before welding begins and theinitial average signal temperature after welding begins. The filtered and corrected peak temperatureprovides a more accurate value since it removes the noise that would at certain data points givehigher peak temperatures than what is actually existing at the measurement location. The resultingfiltered and corrected thermal cycle data is then used in further analysis.To investigate the differences between single torch and dual torch, as well as the effect of torchspacing, Figure 5.3 summarizes the most pertinent thermal cycle data measured during the tests.This figure displays peak temperatures observed in the fully austenitic region and the associated50CHAPTER 5. RESULTSFigure 5.2: (a) Single torch and (b) 2.75” spaced dual torch weld thermal cycle measurements.Figure 5.3: Measured peak temperature versus t8−5 from weld trials.time to cool between 800 ◦C and 500 ◦C (t8−5). Note that t8−5 is effectively an inverse measure ofaverage cooling rate over this temperature range. It is seen that the t8−5 decreases slightly with peaktemperature, which is to say average cooling rates increase with peak temperatures. It also showsthat t8−5 times can be placed into three categories. The shortest t8−5 times are seen in single torchcases, which have an average of 3 seconds, giving an average cooling rate of 100 ◦C/sec. The leadtorch of dual torch passes also have low t8−5 times, showing an average time of 5.6 second, whichis equivalent to an average cooling rate of 54 ◦C/sec. The longest t8−5 times, with an average of 1751CHAPTER 5. RESULTSseconds, were seen in the trailing torch of dual torch passes, which gives an average cooling rate of18 ◦C/sec. It is seen that the highest t8−5 time for trail torches is from a 2.75” torch spacing, whilethe lowest is seen to be from a 7” torch spacing. While the weld parameters are the same for eachof these cases, the difference in t8−5 times makes sense as the lower torch spacings mean less timefor heat from the lead torch to dissipate away from the weld line and thus act as an effective preheatfor the the trailing torch.For a more detailed look at cooling data, numerical differentiation was done on the temperaturetime data. This gives an instantaneous cooling rate with respect to temperature. Figure 5.4 shows anexample of this data for each condition. In addition this figure also displays the region of transfor-mation start temperatures as determined from the data of Reichert [147]. This shows that the coolingrate can significantly change with temperature in the temperature region that transformation is occur-ring. For instance, for the single torch case cooling rates over the transformation start temperaturerange begin at a high cooling rate of 135 ◦C/s and drops to a low value of 65 ◦C/s. Interestingly, it isseen in the trailing torch cases that a minimum value around 10 ◦C/s is reached and then maintainedat temperatures above the end of the transformation start temperature range. Similar to the discus-sionFigure 5.4: Instantaneous cooling rates through austenite decomposition temperatures.52CHAPTER 5. RESULTSof Figure 5.3 it is seen that the trailing torches give the lowest cooling rates, with a difference ofabout 60 ◦C/s at the start of the range of transformation start temperatures. This change in coolingrate with temperature over the range of transformation start temperatures is important as it willinfluence the final transformation products.5.2 Gleeble Heat TreatmentsIn order to investigate the mechanical properties of microstructures in the HAZ, bulk specimens arerequired. Using the large testing strips with the Gleeble 3500 thermal-mechanical physical simu-lator, bulk microstructures that can be used for mechanical and fracture testing may be produced.Early in this work, a number of issues were addressed in using the large strips such as heating ratecontrol, overshooting or undershooting peak temperatures, control over dwell times at peak tem-peratures, cooling rate consistency, temperature gradients along the length of the specimen, stripflatness, strip buckling and strip thinning. The 100 ◦C heating rate was selected because it wasseen to be reproducible for the large strip specimen used. While this rate is significantly lowerthan measured heating rates of 1000 to 3000 ◦C/s during the weld trials in the actual HAZ, thework of Banerjee et al. [143] showed that heating rates of 100 ◦C/s and 1000 ◦C/s produced simi-lar austenitic grain sizes. This data suggests that the use of 100 ◦C/s heating rate for Gleeble heattreatments would have little impact on the results of this study, which makes the use of large stripsin this study feasible. Improvements were made to the temperature control programming in order toensure peak temperatures even at the highest temperatures were consistent and dwell times kept to aminimum. A modification was made to the helium quenching system which provided improvementto gradient cooling effects as well as helping to maintain a consistent cooling rate. The thermalgradient across the width of the specimen was investigated but was found to be negligible. Thechange in the cooling system also helped with strip flatness issues by distributing the quenchinggas over a wider area and with less pressure. Strip buckling or thinning at high temperatures wasminimized with proper selection of piston air ram tension pressure to allow the thermal expansionwithout causing excessive strain.Thermal cycle data measured during a typical heat treatment is shown in Figure 5.5 (a) and(b). These graphs show the first and second thermal cycles used in the creation of one of the mostextreme conditions, the condition with Nb in solution with a prior austenite grain size of 42 µm and53CHAPTER 5. RESULTSFigure 5.5: Thermal cycle data from a Gleeble heat treatment showing (a) stage one (b) stagetwo. Peak temperature recorded from all Gleeble heat treatments for centerline and 5mm on either side off centerline for (c) 5 µm and (d) 42 µm thermal cycles.a 50 ◦C/s cooling rate. This was also the most difficult condition to produce, due to the high tem-peratures for two thermal cycles in combination with the high heating and cooling rates observed.In both thermal cycles it is seen that the thermal gradient along the specimen is relatively minor,varying by about ±10 ◦C in the peak temperature region, while the time at maximum temperatureis consistent at measured locations along the length of the specimen. The data also shows a kink inthe cooling curve in both graphs at approximately 500 ◦C. This delay in the cooling is due to the re-54CHAPTER 5. RESULTScalescence due to the heat of the phase transformation occurring as the austenite phase decomposes.Earlier in the project, an effort was made to maintain a constant measured cooling rate in spite of therecalescence by increasing the gas quenching rate. While it proved possible to do so if prior infor-mation on the transformation temperature was known, it was determined that maintaining a constantquenching rate is more reliable and consistent. Figure 5.5 (c) and (d) shows the consistency in peaktemperatures and thermal gradients for all tests. It is seen in Figure 5.5 (c) that for tests with a goalof 5 µm austenite grain size the peak temperature of 950 ◦C is consistently reached with ±10 ◦Cfor the middle thermocouple with the gradient also typically being less then 10 ◦C along the length.For tests with the goal of 42 µm austenite grain size, shown in Figure 5.5 (d), less consistency inobtaining the desired peak temperature of 1350 ◦C is seen. The deviation reaches as much as 34◦C for the center thermal measurement from the aimed peak temperature. The gradient is also seento be larger for these heat treatments, typically showing a gradient within ± 15 ◦C of the centrethermocouple measurement. The issues as discussed previously, such as peak temperature control,dwell time control, strip flatness, thinning or buckling caused these tests with the aim of 42 µm tobe extremely difficult to produce consistently. Furthermore, the use of K-type thermocouples forthe outer measurements through both thermal cycles needed to produce 42 µm austenite grain sizescould result in a greater error in temperature measurement at these positions due to the high tem-peratures leading to diffusion at the thermocouple tips leading. The S-type thermocouples proveddifficult to ensure a good connection suitable to control the Gleeble 3500 throughout the entire ther-mal cycle, particularly once the quenching system is turned on and the chamber fills suddenly withhelium. Considering that the temperature is very high, (1350 ◦C), and the difficulties in producingthese samples as discussed, and the fact that material was limited, the deviation from aimed peaktemperature and the thermal are deemed acceptable for the purposes of this study.5.3 MicrostructuresThe Gleeble heat treatments produced samples in eight different conditions in order to create mi-crostructures for mechanical and fracture testing. To investigate what microstructures were pro-duced, the main microstructural constituents were observed with samples etched with a 2% Nital so-lution in an optical microscope. Figures 5.6 and 5.7 show the Nital etched microstructure of the con-55CHAPTER 5. RESULTSFigure 5.6: 2% Nital etchings of 5 µm prior austenite grain size with Nb in precipitates withcooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with cooling rates of (c)10 ◦C/s and (d) 50 ◦C/s.ditions with austenite grain sizes of 5 µm, and conditions 42 µm, respectively. The conditions witha prior austenite grain size of 5 µm and with Nb in precipitates show a mainly irregular ferritic mi-crostructure. The effect of increasing cooling rate from 10 ◦C/s to 50 ◦C/s results in a refinement inthe size of the microstructure. The refinement of microstructure with increased cooling rate is seenthroughout all comparable conditions. The conditions with a prior austenite grain size of 5 µm andNb solution present a very difficult microstructure to characterize. While there is a mix of ferriteand of upper bainite in these specimens, the refined microstructure in these samples in combinationwith similarities between the two phases made differentiation using the method used here not pos-sible. The upper bainite phase, noted by irregularly shaped grains with internal structures is clearlyseen throughout Figures 5.7 (a) and (b), which are specimens with Nb in precipitates with a prior56CHAPTER 5. RESULTSFigure 5.7: 2% Nital etchings of 42 µm prior austenite grain size with Nb in precipitates withcooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with cooling rates of (c)10 ◦C/s and (d) 50 ◦C/s.austenite grain size of 42 µm. In these micrographs lower bainite is also seen, which is observed bygrain packets with lath-like structures and has a more continuous substructure compared to upperbainite. Lower bainite is seen more clearly in Figures 5.7 (c) and (d) as it makes up the main con-stituent of the microstructures with prior austenite grain size of 42 µm and Nb in solution. It is alsoof note that in the cases that bainite forms the primary microstructure, as in all images in Figure 5.7,remnants of the prior austenite grain boundary can be seen as the bainite packets do not completelyremove the prior structure. Prior structures are not seen in the ferritic specimens.Le Pera etching reveals the MA constituent of the microstructure. Le Pera etched microstruc-tures are shown in Figures 5.8, and 5.9 for conditions with an austenite grain size of 5 µm, and57CHAPTER 5. RESULTSFigure 5.8: Le Pera etchings of 5 µm prior austenite grain size with Nb in precipitates withcooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with cooling rates of (c)10 ◦C/s and (d) 50 ◦C/s.conditions with an austenite grains of 42 µm respectively. MA is observed in these images by beingbrighter or white compared to the darker matrix. Little MA is seen in the conditions with Nb in pre-cipitates and 5 µm prior austenite grain sizes. The MA in these ferritic specimens is disconnected,and dispersed fairly evenly throughout the microstructures. The amount of MA is noticeably greaterfor the mixed ferrite/upper bainite cases with Nb in solution and prior austenite grain size of 5 µmas seen in Figures 5.8 (c) and (d). The MA in these conditions is also more connected and lessequiaxed. Some similarity in the MA appearance is seen between these conditions with austenitegrain size of 5 µm and Nb in solution and the conditions with a prior austenite grain size of 42 µmand Nb in precipitates shown in Figure 5.9 (a) and (b). The MA phase in these specimens, as well asthe conditions with prior austenite grain size of 42 µm and Nb in solution are at times difficult to de-58CHAPTER 5. RESULTSFigure 5.9: Le Pera etchings of 42 µm prior austenite grain size with Nb in precipitates withcooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with cooling rates of (c)10 ◦C/s and (d) 50 ◦C/s.termine due to a lack of contrast between the MA phase and the matrix. Notable though, conditionswith prior austenite grain sizes of 42 µm and Nb in solution show less MA that is more elongatedand appears to follow prior austenite grain boundaries.Quantification of the microstructures according to the method outlined in Section 4.2 summa-rizes the previous discussion of observed microstructures. Figure 5.10 shows that the Gleeble heattreatments have produced microstructures with one of four primary constituents. The primary con-stituents are found to be either ferrite, upper bainite, lower bainite, or a complex mix of ferriteand bainite. Interestingly, it is seen that the highest MA fractions correspond with the primaryconstituent being of mixed ferrite-bainite or of upper bainite.59CHAPTER 5. RESULTSFigure 5.10: Measured microstructural constituents of the Gleeble heat treated samples.5.4 Tensile ResultsIn the previous section it was seen that the Gleeble heat treatments produced a diverse range ofmicrostructures. Tensile testing was carried out on these at three temperatures (ambient, -20 ◦C,-60 ◦C) on subsized tensile specimens for each condition. The engineering stress-strain curves foreach condition at each testing temperature conditions with an austenite grain size of 5 µm and con-ditions with an austenite grains of 42 µm are shown respectively in Figures 5.11 and 5.12. Table 5.1summarizes the important tensile testing data. All conditions show a ductile stress-strain responsefor each testing temperature. In some tests at the lower two temperatures, the immersible exten-someter used reached its operating limit prior to failure. In these tests, the extensometer arms weremoved during the test as this limit was being approached, and the strain past this point is not us-able. In all tests where this movement was required, it occurred after necking has occured. Thismeans that the true stress-strain data using the extensometer is valid since it can only be calculatedup to the point of necking where the localization of strain causes thinning which cannot be measured60CHAPTER 5. RESULTSFigure 5.11: Engineering stress-strain of 5 µm prior austenite grain size with Nb in precipi-tates with cooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with coolingrates of (c) 10 ◦C/s and (d) 50 ◦C/s. Note that the ’X’ marks the last usable extensometerdata during the test (load measurements continued).by the extensometer. It is also noted that there is a difference in the extensometer gauge lengthused between tests at ambient temperatures and the tests at -20 ◦C and -60 ◦C tests. The ambienttemperature tests used a gauge length of 12.5 mm while the colder temperatures used a gauge lengthof 9 mm. The difference in gauge length does not influence measured strains during the uniformdeformation prior to necking, but after the localization of strain in the neck the shorter gauge lengthswill result in greater engineering strains compared to using larger gauge length.From the engineering stress-strain plots (Figures 5.11 and 5.12) it can be seen that the ultimatetensile strength for every condition increases with decreasing testing temperature. Samples from all61CHAPTER 5. RESULTSFigure 5.12: Engineering stress-strain of 42 µm prior austenite grain size with Nb in precipi-tates with cooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with coolingrates of (c) 10 ◦C/s and (d) 50 ◦C/s. Note that the ’X’ marks the last usable extensometerdata during the test (load measurements continued).the heat treatments show a rounded curve throughout. In determining the yield stress with the typical0.2% offset method, significant variation between samples were observed due to the sensitivity ofthis method to these rounded stress strain curves. This was resolved by using a 0.5% strain offset,which provided consistent results. Table 5.1 summarizes all measured tensile properties.To further investigate the mechanical behaviour up until necking, the strain hardening rate wasdetermined for each test and plotted with the true stress-strain data. This is shown in Figures 5.13and 5.14 for conditions with an austenite grain size of 5 µm and conditions with an austenite grainsof 42 µm respectively. All conditions show that necking occurs, which is defined by the point ofplastic instability where the increase in strength due to strain hardening cannot keep in balance with62CHAPTER5.RESULTSTable 5.1: Measured tensile propertiesType Temperature σys 0.5% offset (MPa) σUTS (MPa) εUEL εfrac σfrac (MPa) Number of testsNb in precipitates, 5 µm @ 10 ◦C/s Ambient 497 660 0.108 1.00 1443 5Ferrite -20 ◦C 495 693 0.142 1.36 2157 1-60 ◦C 522 732 0.122 0.87 1245 2Nb in precipitates, 5 µm @ 50 ◦C/s Ambient 547 707 0.092 1.20 1846 6Ferrite -20 ◦C-60 ◦CNb in solution, 5 µm @ 10 ◦C/s Ambient 511 669 0.113 1.14 1604 3Mixed Ferrite/Bainite -20 ◦C 518 705 0.160 1.11 1685 1-60 ◦CNb in solution, 5 µm @ 50 ◦C/s Ambient 563 715 0.097 1.10 1582 2Mixed Ferrite/Bainite -20 ◦C 590 756 0.092 1.14 1853 1-60 ◦C 628 805 0.102 0.97 1697 1Nb in precipitates, 42 µm @ 10 ◦C/s Ambient 510 630 0.091 1.42 1853 2Upper Bainite -20 ◦C 522 662 0.105 1.61 2397 1-60 ◦C 523 709 0.116 1.08 1466 1Nb in precipitates, 42 µm @ 50 ◦C/s Ambient 589 697 0.075 1.39 2110 3Upper Bainite -20 ◦C 602 722 0.084 1.56 2729 1-60 ◦C 633 770 0.082 1.02 1596 2Nb in solution, 42 µm @ 10 ◦C/s Ambient 602 692 0.061 1.30 1835 3Lower Bainite -20 ◦C 625 746 0.078 1.46 2814 1-60 ◦C 640 776 0.090 1.16 1796 2Nb in solution, 42 µm @ 50 ◦C/s Ambient 716 806 0.046 1.50 2633 4Lower Bainite -20 ◦C-60 ◦C63CHAPTER 5. RESULTSFigure 5.13: True stress-strain and strain hardening rate of 5 µm prior austenite grain size withNb in precipitates with cooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solutionwith cooling rates of (c) 10 ◦C/s and (d) 50 ◦C/s.the increased stress from a thinning section, referred to as the Conside`re condition. As there is nomeasurement of the cross-sectional area in the necked region, no true stress data is usable after thispoint with exception of the final point at fracture which will be discussed later. From these figuresit is seen that at a given amount of strain the strain hardening rate is higher as testing temperaturedecreases. Exceptionally, for the two mixed ferrite/bainite microstructure conditions with Nb insolution and an austenite grain size of 5 µm there is not much strain hardening rate differencewith decreasing temperature (i.e. Figure 5.13 (c) and (d)), whereas all other conditions did showsignificant differences.The true stress-strain data from tests has so far provided information on the tensile behaviour upuntil the point of necking. Post-test examinations were used to investigate post necking behaviour64CHAPTER 5. RESULTSFigure 5.14: True stress-strain and strain hardening rate of 42 µm prior austenite grain sizewith Nb in precipitates with cooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb insolution with cooling rates of (c) 10 ◦C/s and (d) 50 ◦C/ a single point, the point of failure. This was accomplished by measuring the final fracture areausing SEM images. Figure 5.15 (a) and (b) show examples of typical SEM images of a tensilefracture surface used in the measurement of the final fracture measurement. As part of the post-testinvestigation, observations of the failure surface were also made and tests that showed significantdeviation from a flat fracture surface at failure were not used to calculate true fracture strain andstress. Macrographs in Figure 5.15 (c) and (d) show typical flat and invalid slant failure respectively.With the final fracture surface area measured for the cases of flat failure, calculation of the true stressand true strain is determined, and corrected according to the method discussed in Section 4.5.With the true fracture stress corrected, Figures 5.16 and 5.17 plot true stress-strain until fractureat each testing temperature for conditions with an austenite grain size of 5 µm and conditions with65CHAPTER 5. RESULTSFigure 5.15: SEM images of tensile fracture surface of (a) Nb in solution with cooling rate of50 ◦C/s and (b) Nb in precipitates with cooling rate of 50 ◦C/s. (c) Typical flat fracturesurface and (d) unacceptable slant austenite grains of 42 µm respectively. These graphs show that microstructures formed with aprior austenite grain size of 42 µm generally have higher values of true stress and true strain atfracture than microstructures formed with a prior austenite grain size of 5 µm. In Figure 5.16 (c)and (d) it is seen that the conditions with a prior austenite grain size of 5 µm with Nb in solution,microstructures that are composed of a mix of ferrite/bainite, have similar values for true stress andtrue strain at fracture across all testing temperatures.The effect of temperature on tensile properties is summarized in Figure 5.18. It is seen that thetensile strength and yield strength both increase with decreasing temperature, with an average of 76MPa and 37 MPa respectively across all conditions when going from ambient to -60 ◦C. However,there are some interesting deviations from this trend. The case with Nb in solution with prioraustenite of grain size of 42 µm, which is a lower bainite microstructure, shows a large increasein tensile strength (122 MPa) compared to the average. With respect to engineering strain at thepoint of necking (i.e. maximum engineering strain under uniform elongation), which is shownin Figure 5.18 (c), it is seen that this value generally increases with decreasing temperature. Thelargest exception to this is the case with Nb in precipitates with prior austenite of grain size of 5µm which is a ferritic microstructure where uniform elongation engineering strain increases then66CHAPTER 5. RESULTSFigure 5.16: True stress-strain to failure of 5 µm prior austenite grain size with Nb in precipi-tates with cooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with coolingrates of (c) 10 ◦C/s and (d) 50 ◦C/s.67CHAPTER 5. RESULTSFigure 5.17: True stress-strain to failure of 42 µm prior austenite grain size with Nb in precip-itates with cooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with coolingrates of (c) 10 ◦C/s and (d) 50 ◦C/s.68CHAPTER 5. RESULTSFigure 5.18: Temperature dependence of (a) yield strength, (b) ultimate tensile strength, (c)engineering uniform elongation, and (d) true strain at fracture.decreases as temperature decreases. The two cases with upper bainite as the main constituent,that is the case with Nb in solution with a prior austenite grain size of 42 µm show a behaviourwhere the engineering strain at the point of necking only increases a small amount with decreasingtemperature. The observation of final fracture strains in Figure 5.18 (d) show a different trend withtemperature. From ambient temperature to -20 ◦C there is an increase in true strain at fracture forall samples except the condition with a prior austenite grain size of 5 µm and Nb in solution cooledat 10 ◦C/sec where the value decreases. From -20 ◦C to -60 ◦C it is seen that the true fracture strain69CHAPTER 5. RESULTSdrops for every condition. The drop is most severe for the condition with a prior austenite grain sizeof 5 µm with Nb in precipitates and a cooling rate of 10 ◦C/sec which is a ferritic microstructure.Larger drops over this temperature range are also noted in the conditions with a prior austenite grainsize of 42 µm and Nb in precipitates which are primarily upper bainitic microstructures.5.5 Kahn ResultsKahn tests for each microstructural condition were carried out at different temperatures. For com-parison, load divided by thickness is plotted against the displacement adjusted so that the peak loadis reached at zero displacement. These plots are shown for the 5 µm prior austenite grain size and42 µm austenite grain size conditions in Figures 5.19 and 5.20 respectively. Note that a numberof the ambient temperature tests in these figures show small load drops; these are associated withholding displacement constant during optical crack length measurements. Results of the Kahn testsshow similarities in trends to the tensile tests. Decreasing temperatures result in increased maxi-mum normalized loads reached and thus tear strengths observed, trending similar to the effect seenwith ultimate tensile strength in the tensile testing. The condition of 42 µm prior austenite grainsize and 50 ◦C/s cooling rate, a microstructure with primarily lower bainite, shows the highest tearstrength, similar to the tensile tests where it showed the highest ultimate tensile strength. The con-dition of a 5 µm prior austenite grain size with Nb in and 10 ◦C/s cooling rate shows the lowesttear strength, similarly in the tensile tests this condition showed the second lowest ultimate tensilestrength. Crack length data during testing was limited by the travel length of the microscope. Withinthe crack measurements observed it is seen that crack lengths increase at approximately the samerate with displacement for each condition.From the Kahn test’s load and displacement data, UPEd0−1 and Etearvalues were calculatedaccording to the method discussed in Section 4.6. While UPEd0−1 values represent a quick methodto investigate crack propagation energies, Etear provides a more rigorous method which was usedto validate the usage of UPEd0−1 and investigate thickness effects. In order to determine the valuesof a used in the calculation of Etear, crack measurements of all specimens were categorized andcombined by sample thickness into groupings close to 1.1 mm, 1.25 mm, 1.5 mm, 2.0 mm, and 2.5mm in thickness. Then a polynomial fit to the crack displacement data was made between the pointof maximum load and 1 mm of pin-to-pin displacement from that point. These fits to crack length70CHAPTER 5. RESULTSFigure 5.19: Kahn tear test results of 5 µm prior austenite grain size with Nb in precipitateswith cooling rates of(a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with cooling ratesof (c) 10 ◦C/s and (d) 50 ◦C/s71CHAPTER 5. RESULTSFigure 5.20: Kahn tear test results of 42 µm prior austenite grain size with Nb in precipitateswith cooling rates of(a) 10 ◦C/s and (b) 50 ◦C/s, and Nb in solution with cooling ratesof (c) 10 ◦C/s and (d) 50 ◦C/sdisplacement data are shown in Figure 5.21. These figures show that with thicker specimens, cracklength increases slightly less with pin displacement compared to thinner specimens. The thinnerspecimens also show a greater scatter in crack length measurements overall. It is assumed that theroom temperature crack length measurements and fits to crack length will remain applicable to testsat colder temperatures. This assumption is made because crack measurements were not possibledue to the nature of immersion testing. It is is believed that this assumption is reasonable since alltests remain ductile and that the crack growth should be geometrically driven.The fit to crack length data was then used to determine the Etear values. This was done over eachdata point for the first millimeter of adjusted displacement, which can be used to construct Etearcurves such as the two shown in Figure 5.22. Note in Figure 5.22 (a) that the appearance of load72CHAPTER 5. RESULTSFigure 5.21: Curve fitting of crack length versus adjusted displacement for tests with thicknessclose to (a) 1.15 mm (b) 1.25 mm (c) 1.5 mm (d) 2.0 mm (e) 2.5 mm73CHAPTER 5. RESULTSdrops in this figure clearly show that they represent a very small deviation from the otherwise regularcurve and thus it is reasonable to assume that negligible crack growth occurred during the cracklength measurements. The decrease in Etear value with increasing crack length may be attributedto the fact that thinning of the specimen is not accounted for in the calculation. Figure 5.23 showstwo examples of the Kahn surface after fracture. With the notch of the specimen on the right side ofthese images it is clear that the sample has thinned after the initial crack growth toward the left fromthe notch as part of the fracture process. If Etear used the current thickness instead of the initiallymeasured thickness, the Etear may not show the decrease as seen in Figure 5.22. For comparison toUPEd0−1 values, the Etear value was taken at 1 mm of adjusted displacement without any correctionto thinning of the specimen.Figure 5.22: Etear curves determined based on fitted fitted crack data for (a) Nb in precipitates,42 µm prior austenite grain size with cooling rate of 10 ◦C/s and (b) Nb in solution, 42µm prior austenite grain size with cooling rate of 50 ◦C/s.The role of thickness on Etear and UPEd0−1 values was investigated using the as-received con-dition. Figure 5.24 shows that Etear and UPEd0−1 values for the as-received conditions increasewith thickness. Thus, to use Etear and UPEd0−1to compare results, a correction for the thicknesseffect was needed. To do so required normalizing each test by a test of the same condition that wasclosest to a set thickness. The set thickness that the correction would be made to was decided tobe 1.2 mm as this was close to the average thickness for most thermally treated specimens. Witheach test normalized, tests that were close in thickness (within 0.05 mm) had their values averaged.To this normalized Etear and UPEd0−1 values we reduced to an average point for specimens close74CHAPTER 5. RESULTSFigure 5.23: SEM images of Kahn fracture surface of (a) Nb in precipitates, 42 µm prioraustenite grain size with cooling rate of 10 ◦C/s and (b) Nb in solution, 42 µm prioraustenite grain size with cooling rate of 50 ◦C/ thickness and a curve was empirically fitted to the data. Fitting of the Etear and UPEd0−1 withrespect to thickness is shown in Figure 5.25, and the fitting equation takes the form of:Normalized Energy Fit = c1(1− exp(−c2(t− c3))) (5.1)where t is thickness, c1, c2, and c3 are constants, and the normalized energy fit value is calculatedseparately for Etear and UPEd0−1 with differing values of the constants. For Etear fitting, c1 is 1.11,c2 is 2.6, and c3 is 0.4. For UPEd0−1 fitting, c1 is 1.135, c2 is 2.5, and c3 is 0.4.With a curve fitting for Etear and UPEd0−1 with respect to thickness, a correction for thicknessto these values was applied. The effect of this correction is shown in Figure 5.26. In these graphsit can be seen, particularly when considering the as-received material which covers a wide rangeof thicknesses, that the trend of Etear and UPEd0−1 to increase with thickness has been reduced.From these figures it is seen that there are differences in using Etear andUPEd0−1. For instance, themaximum Etear values are seen to be from the condition with with a prior austenite grain size of 42µm and Nb in solution with a cooling rate of 50 ◦C/s which is primarily lower bainite structure with75CHAPTER 5. RESULTSFigure 5.24: (a) Etear and (b) UPEd0−1 as a function of thickness for the as-received materialFigure 5.25: Fitting curve for (a) Etear and (b) UPEd0−1 as a function of other conditions reaching similar values of Etear. In the case of maximum UPEd0−1 values, it isseen this condition has still some of the highest values, but are now comparable to the conditionswith prior austenite grain size of 5 µm cooled at 50 ◦C/s with Nb in precipitates and with Nb insolution which are ferritic and mixed ferrite/bainite microstructures respectively. The lowest valueof Etear are seen to be the condition with a prior austenite grain size of 5 µm with Nb in precipitatescooled at a rate of 10 ◦C/s which is a ferritic microstructure. The lowest value of UPEd0−1 thecondition with a prior austenite grain size of 42 µm with Nb in precipitates cooled at 10 ◦C/s which76CHAPTER 5. RESULTSis a microstructure that is primarily upper bainite. Table 5.2 summarizes measured Kahn tear testingproperties including the uncorrected and corrected Etear and UPEd0−1 values.Figure 5.26: (a) Corrected Etear and (b) corrected UPEd0−1 as a function of thickness, ex-panded to view the thermally treated conditions of both respectively in (c) and (d).77CHAPTER5.RESULTSTable 5.2: Kahn tear test propertiesType Temperature σTear UPEd0−1 CorrectedUPEd0−1 Etear Corrected Etear Number of tests(MPa) (J/mm2) (J/mm2) (J/mm2) (J/mm2)As-Received Ambient 1203 0.256 0.256 1.80 1.82 9-20 ◦C 1251 0.271 0.274 1.81 1.74 2-60 ◦CNb in precipitates, 5 µm @ 10 ◦C/s Ambient 1080 0.241 0.259 1.49 1.62 3Ferrite -20 ◦C 1156 0.267 0.267 1.84 1.86 2-60 ◦C 1203 0.264 0.261 1.84 1.84 1Nb in precipitates, 5 µm @ 50 ◦C/s Ambient 1219 0.264 0.275 1.77 1.86 5Ferrite -20 ◦C 1261 0.286 0.284 1.98 1.99 3-60 ◦C 1314 0.297 0.292 2.05 2.05 1Nb in solution, 5 µm @ 10 ◦C/s Ambient 1171 0.250 0.256 1.72 1.78 3Mixed Ferrite/Bainite -20 ◦C 1135 0.245 0.278 1.55 1.78 2-60 ◦C 1197 0.247 0.288 1.57 1.84 1Nb in solution, 5 µm @ 50 ◦C/s Ambient 1245 0.269 0.275 1.81 1.87 3Mixed Ferrite/Bainite -20 ◦C 1275 0.264 0.304 1.67 1.93 2-60 ◦C 1309 0.271 0.300 1.77 1.97 1Nb in precipitates, 42 µm @ 10 ◦C/s Ambient 1115 0.245 0.248 1.70 1.74 3Upper Bainite -20 ◦C 1151 0.256 0.251 1.77 1.76 2-60 ◦C 1181 0.262 0.256 1.81 1.78 1Nb in precipitates, 42 µm @ 50 ◦C/s Ambient 1244 0.254 0.268 1.64 1.74 2Upper Bainite -20 ◦C 1278 0.282 0.276 1.96 1.95 2-60 ◦C 1313 0.284 0.281 2.00 2.00 1Nb in solution, 42 µm @ 10 ◦C/s Ambient 1189 0.219 0.259 1.50 1.78 4Lower Bainite -20 ◦C 1311 0.288 0.278 1.99 1.97 2-60 ◦C 1354 0.290 0.282 2.02 2.00 1Nb in solution, 42 µm @ 50 ◦C/s Ambient 1386 0.274 0.281 1.94 2.00 3Lower Bainite -20 ◦C 1442 0.283 0.294 1.93 2.03 2-60 ◦C 1511 0.308 0.308 2.23 2.25 178CHAPTER 5. RESULTS5.6 Fractography ResultsFracture surfaces of samples after tensile and Kahn testing were investigated using SEM. The frac-ture surface as examined at low magnifications for tensile and Kahn specimens were discussed inprevious sections (i.e. see Figures 5.15 and 5.23). These figures were examples of the primarilyductile fracture surface seen at low magnification in all cases for both tensile and Kahn tests. Fig-ure 5.27 and 5.28 show high magnification micrographs for several exemplary testing conditionsand testing temperatures for the comparison of tensile to Kahn fracture surfaces. From these figuresit can be seen that for a given microstructure and testing temperature the fracture surface betweentensile tests and Kahn tests showed very similar fracture surfaces. For simplicity then, tensile frac-ture surfaces will be used here to compare fracture across conditions except where there is no validtensile test for that condition wherein the Kahn fracture surface image is used instead.At higher magnifications the void structure of the fracture surface is revealed. Figure 5.29shows high magnification of the as-received material. What is seen in this figure is that the voidsize remains comparable at all temperatures tested. Figures 5.30 and 5.31 show the high magni-fication micrographs of the conditions with a prior austenite grain size of 5 µm with Nb precip-itated and cooling rates of 10 ◦C/s and 50 ◦C/s respectively. These micrographs show that theseconditions which are ferritic microstructure have comparable void sizes with only the occasionalappearance moderately sized voids when compared to the as-received material which is also a fer-ritic microstructure. Fracture surfaces with the conditions with a prior austenite grain size of 42Figure 5.27: (a) Tensile and (b) Kahn fracture surfaces as examined by SEM for the conditionwith a prior austenite grain size of 5 µm with Nb in precipitates cooled at 10 ◦C/s testedat ambient temperatures.79CHAPTER 5. RESULTSFigure 5.28: (a) Tensile and (b) Kahn fracture surfaces as examined by SEM for the conditionwith a prior austenite grain size of 42 µm with Nb in solution cooled at 50 ◦C/s testedat -20 ◦C.µm and Nb in solution with cooling rates of 10 ◦C/s and 50 ◦C/s are shown in Figures 5.32 and5.33 respectfully. What is seen from these images is that the void size remains fairly similar to theconditions with a prior austenite grain size of 5 µm with Nb in precipitates with mainly smallervoids and occasional moderately sized voids.Figure 5.29: As-received fracture surfaces for (a) tensile test at ambient temperature (b) Kahntest at -20 ◦C (c) tensile test at -60 ◦C80CHAPTER 5. RESULTSFigure 5.30: Condition with a prior austenite grain size of 5 µm, Nb in precipitates cooled at10 ◦C/s fracture surfaces for (a) tensile test at ambient temperature (b) tensile test at -20◦C (c) tensile test at -60 ◦C81CHAPTER 5. RESULTSFigure 5.31: Condition with a prior austenite grain size of 5 µm, Nb in precipitates cooled at50 ◦C/s fracture surfaces for (a) tensile test at ambient temperature (b) Kahn test at -20◦C (c) Kahn test at -60 ◦C82CHAPTER 5. RESULTSFigure 5.32: Condition with a prior austenite grain size of 5 µm, Nb in solution cooled at 10◦C/s fracture surfaces for (a) tensile test at ambient temperature (b) tensile test at -20◦C (c) Kahn test at -60 ◦C83CHAPTER 5. RESULTSFigure 5.33: Condition with a prior austenite grain size of 5 µm, Nb in solution cooled at 50◦C/s fracture surfaces for (a) tensile test at ambient temperature (b) tensile test at -20◦C (c) tensile test at -60 ◦C84CHAPTER 5. RESULTSFigures 5.34 and 5.35 show high magnification micrographs of conditions with a prior grainsize of 42 µm and Nb in precipitates with cooling rates of 10 ◦C/s and 50 ◦C/s respectively. Theseconditions, which are primarily upper bainitic microstructures, have both smaller voids like in theprevious conditions, but now large void structures are seen often across the fracture surface. Thehigh magnification examination of conditions with a prior austenite grain size of 42 µm and Nb insolution with cooling rates of 10 ◦C/s and 50 ◦C/s are shown in Figures 5.36 and 5.37 respectively.These figures show that these conditions which are primarily lower bainite microstructure give afracture surface with some smaller voids and a number of large voids. These lower bainitic fracturesurfaces show similarities to the fracture surfaces seen in the upper bainitic cases. The large voidsin the condition with a prior austenite grain size of 42 µm with Nb in solution cooled at 50 ◦C/s areseen to be a little larger than those of the other conditions with a prior austenite grain size of 42 µm.All samples showed an amount of small voids on their fracture surfaces. For the ferritic andferrite/bainite mixture microstructures that were formed with a prior austenite grain size of 5 µmthe small voids were seen to dominate the fracture surface. In the upper and lower bainite specimens,which are specimens formed with with a prior austenite grain size of 42 µm, large voids are seenoften across the fracture surface. In all conditions, the testing temperature was seen to have littleeffect on the size of the voids. Table 5.3 summarizes all observations made of the fracture surfaces.85CHAPTER 5. RESULTSFigure 5.34: Condition with a prior austenite grain size of 42 µm, Nb in precipitates cooled at10 ◦C/s fracture surfaces for (a) tensile test at ambient temperature (b) tensile test at -20◦C (c) tensile test at -60 ◦C86CHAPTER 5. RESULTSFigure 5.35: Condition with a prior austenite grain size of 42 µm, Nb in precipitates cooled at50 ◦C/s fracture surfaces for (a) tensile test at ambient temperature (b) Kahn test at -20◦C (c) Kahn test at -60 ◦C87CHAPTER 5. RESULTSFigure 5.36: Condition with a prior austenite grain size of 42 µm, Nb in solution cooled at 10◦C/s fracture surfaces for (a) tensile test at ambient temperature (b) tensile test at -20◦C (c) Kahn test at -60 ◦C88CHAPTER 5. RESULTSFigure 5.37: Condition with a prior austenite grain size of 42 µm, Nb in solution cooled at 50◦C/s fracture surfaces for (a) tensile test at ambient temperature (b) tensile test at -20◦C (c) tensile test at -60 ◦C89CHAPTER5.RESULTSTable 5.3: Fracture surface observationsType Temperature Tensile Fracture Surface Kahn Fracture SurfaceNb in precipitates, 5 µm @ 10 ◦C/s Ambient Moderate sized voids (5-10 µm) Small to moderate sized voids throughout(5-10 µm)Ferrite -20 ◦C Smaller and moderate sized voids Mainly small voids, some moderate voids-60 ◦C Smaller and moderate sized voids Small and moderate voids throughoutNb in precipitates, 5 µm @ 50 ◦C/s Ambient Mainly small voids Large voids near notch, small voids else-whereFerrite -20 ◦C Mainly small voids, some moderate voids-60 ◦C Small and moderate voids throughoutNb in solution, 5 µm @ 10 ◦C/s AmbientMainly small voids with some moderateand large voids Small and moderate sized voidsMixed Ferrite/Bainite -20 ◦C Mainly small voids and some moderatevoidsMainly small voids, some moderate voids-60 ◦C Moderate size voids throughoutNb in solution, 5 µm @ 50 ◦C/s Ambient Mainly small voids Small and moderate sized voidsMixed Ferrite/Bainite -20 ◦C Mainly small voids with some moderatevoidsMainly small voids, some moderate voids-60 ◦C Smaller sized voids and deep cracksSmall voids and moderate voids, somesmall cracks or ridges parallel to crackpropagation directionNb, in precipitates, 42 µm @ 10 ◦C/S Ambient Mainly smaller voidsModerate sized voids and some large voidsthroughoutUpper Bainite -20 ◦C Mainly small voids with a few larger voidsSmall and moderate to large voidsthroughout-60 ◦CSmaller and moderate voids, and somedeep cracksLarge voids throughoutNb, in precipitates, 42 µm @ 50 ◦C/S Ambient Small voids with plenty of larger voids Large voids near notch, moderate voidselsewhereUpper Bainite -20 ◦C Small voids and some larger voids Very large voids near notch, large voidsfurther in-60 ◦C Smaller voids and some cracks Large voids throughoutNb in solution, 42 µm @ 10 ◦C/s Ambient Mainly small voids with some large voids Large voids throughoutLower Bainite -20 ◦C Small voids with some large voids Large voids and a few very large voids-60 ◦C Small and large voids with numerouscracksLarge voids near notch, some small cracksfurther inNb in solution, 42 µm @ 50 ◦C/s Ambient Some small voids and numerous largevoidsVery large voids throughoutLower Bainite -20 ◦C Very large voids near notch, large voidsfurther in-60 ◦CLarge and very large voids near notch,ridges and cracks parallel to crack prop-agation direction further inVoid diameter sizing is approximated as follows: small (1-5 µm), moderate (5-10 µm), large (10-15 µm),and very large (15-20 µm).90CHAPTER6Discussion6.1 IntroductionFrom the test results, there were many interesting findings which will be discussed in this chap-ter with particular attention given to the relation between microstructure and observed properties.Where appropriate, analysis is extended further to provide insights. Also of importance to the dis-cussion is examining the applicability of test and analysis methods used here so as to be clear aboutthe potential and limitations. As such, the most crucial aspect of testing is first addressed, i.e. theability to consistently create isolated bulk microstructures for testing that are of relevance to realobserved HAZ microstructures.6.2 Weld Trials and Gleeble Heat TreatmentsWeld trials provided the heating and cooling rate data that guided Gleeble heat treatments. Consid-ering this, the HAZ microstructures developed in weld trials can be compared to the microstructuresgenerated through Gleeble heat treatments to determine if these heat treated microstructures are ofrelevance. Figures 6.1, 6.2, and 6.3 shows typical microstructures seen in the HAZ from the trialsalong with microstructures from Gleeble heat treatment that appear similar from metallographiccomparison. It is seen that the HAZ microstructures in close proximity to the fusion line (i.e. mi-crostructures in the CGHAZ and FGHAZ) largely show features (i.e. constituents, coarseness, andmorphology) comparable to those produced by the Gleeble heat treatments. While the purpose of91CHAPTER 6. DISCUSSIONFigure 6.1: (a) Weld trial HAZ microstructure of a 70 mm (2.75”) torch spacing compared to(b) Gleeble heat treated microstructures.92CHAPTER 6. DISCUSSIONFigure 6.2: (a) Weld trial HAZ microstructure of a single torch sample compared to (b) Glee-ble heat treated microstructures.93CHAPTER 6. DISCUSSIONFigure 6.3: (a) Weld trial HAZ microstructure of a different single torch sample compared to(b) Gleeble heat treated microstructures.94CHAPTER 6. DISCUSSIONthis study is not to reproduce exactly the individual microstructures in the HAZ using Gleeble heattreatments, it is seen that the microstructures that have been produced are of relevance to the HAZ.Furthermore, it is seen in these figures that Gleeble heat treatments that would be expected to onlyhave relevancy further from the fusion line in some cases produced microstructures that are in factshowing similarities to the microstructure observed close to the fusion line. While it is not clearwhy this is the case, it is probable that subsequent weld passes have affected the final microstructurein these positions, making a straightforward relation between distance from the fusion line and theGleeble heat treatment not possible. Plausible scenarios for the observed thermal history in eachcase will follow. In Figure 6.1, the heating cycle in the weld may not have been enough to dissolveall of the Nb containing particles in the CGHAZ prior to cooling, resulting in a microstructure thatis similar to the shown thermal cycles with 42 µm prior austenite grain size and Nb in precipitates.In Figure 6.2, the single torch weld observed close to the fusion zone is similar to the Gleeble heattreatment with Nb in solution with a large prior austenite grain size. The weld microstructure inFigure 6.3 shows fine features in the area close to the fusion zone. It is likely that in this case,the microstructure was affected by a subsequent pass of the weld torch placing the region into theFGHAZ which is well represented by the observed Gleeble heat treatments that use a small prioraustenite grain size. By design, the Gleeble heat treatments did not attempt to encompass all thermalcycles and their resulting microstructures. For example, microstructures in the ICCGHAZ, the zonenoted in literature [61–64,68] to be a potential site of low toughness in the HAZ, are not investigatedin this study due to this microstructure being only a small portion of the HAZ, and due to constraintsof time and material.It was seen in Figure 5.5 (d) that there was some spread in peak temperatures for Gleeble heattreated samples with a prior austenite grain size of 42 µm grain size (peak temperature of 1350◦C). To assess the significance of this variation on prior austenite grain size, the microstructureprediction model of Maalekian et al. [148] was used on tests giving the lowest peak temperature(1320 ◦C), close to the aim peak temperature (1350 ◦C), and the highest peak temperature (1380◦C). For the thermal cycle giving the lowest peak temperature, the model predicts that the austenitegrain size would be 39 µm which is close to the goal of reaching a 42 µm austenite grain size. Forthe thermal cycle resulting in the highest peak temperature, the model predicts that the austenitegrain size should be 50 µm. This is noticeably higher than the aim of 42 µm. Austenite grain size95CHAPTER 6. DISCUSSIONinfluences the transformation start temperatures; an increase in the austenite grain size correlateswith a decrease of the transformation temperature, which is to say that it promotes lower bainiticand even martensitic transformation products. From the model, the predicted microstructure for thesample with the significantly higher grain size remains fully lower bainitic with a small amount ofmartensite, which is the same microstructure constituents observed from specimens closer to theaim peak temperature of 1350 ◦C. The austenite grain size also influences the packet size of thelower bainite laths. While this packet size could affect the cleavage fracture toughness, it is notexpected to affect the tensile properties or ductile fracture properties. Thus the difference betweensamples with 39 µm and 50 µm austenite grain sizes are not significant in the context of this work.6.3 Kocks-Mecking AnalysisHaving shown that there is confidence in the Gleeble heat treatments to generate microstructuresof relevance to the HAZ, the rest of the discussion focuses on the results of mechanical testing.This discussion will first examine the results from tensile testing, and then move on to the relationbetween microstructure and tensile properties in the following section.The tensile testing results in Section 5.4 showed some interesting behaviour. Figure 6.4 showsthe schematic of one test condition at two temperatures that typifies the stress-strain curves as ob-served from the thermally treated specimens. It is seen that the elastic-plastic transition for thesestress-strain curves is very rounded, which is to say that there is no distinct yield point observed andthat yielding occurs over a range of strain. This makes determination of the yield strength difficult,which lead to the selection of using the 0.5% offset yield strength over the typical 0.2% as this gavemore consistent values across tests of the same test conditions. The strict use of an offset yieldstrength to characterize the yield point however ignores the details related to micro-yielding.After yielding begins, work-hardening occurs and is measured up to the Conside`re condition(i.e. start of necking), which was reached in all test conditions as typified by the results shown inFigure 6.4. Work hardening is represented by the work hardening rate given by:θ =dσdε(6.1)96CHAPTER 6. DISCUSSIONFigure 6.4: Example true stress-strain data from Nb in precipitates, 5 µm prior austenite grainsize cooled at 10 ◦C/s at two different temperatures with labelling of notable properties.with the Conside`re condition being defined as:dσdε= σ (6.2)After the Conside`re criteria is reached, the cross-sectional area in the neck changes from theinitial value and to observe the work-hardening would require measurement of the changing cross-sectional area in the necked region. In this study, no such measurements were made and thus work-hardening can only be measured up to the point of necking, with the exception of the measurementsmade after the test and thus relating to the point of final fracture. The work-hardening behaviour inthis study, typified in Figure 6.4, shows that as testing temperature decreases from ambient temper-atures, greater work-hardening rates are observed, greater elongation at the Conside`re condition isobserved, and greater stress at the Conside`re condition is observed. There are some exceptions tothis work hardening behaviour which will be discussed shortly.97CHAPTER 6. DISCUSSIONA more complete analysis of the work hardening behaviour follows the framework of Kocks-Mecking [112,114,149], plotting the work-hardening rate against stress and is shown schematicallyin Figure 6.5. From such a plot, various stages of a steel’s deformation can be observed. As the stressfirst increases, the steel is within the linearly elastic region and θ should be equal to the Young’smodulus of the material. With stress increasing further, the steel experiences the elastic-plastictransition and θ rapidly decreases. Following the elastic-plastic transition, stage III work hardeningis seen with a constant (i.e. linear) decrease of θ with increasing stress, which is the stage of workhardening seen in the typical tensile test and is the most important for this investigation of workhardening behaviour. The stage IV low work hardening rate is important to high strains, which areobserved in the plots of true stress-strain including the true fracture stress.Figure 6.5: Schematic of the θ vs σtrue showing the definition of operational parameters.The stage III deformation regime, with some reasonable assumptions, can give parameters witha physical basis. The strengthening contribution of work hardening was previously defined by:σdis = αMGbρ1/2 (6.3)98CHAPTER 6. DISCUSSIONwhere the dislocation density, ρ , evolves according to:dρdεp= k1ρ1/2− k2ρ (6.4)The theoretical dislocation hardening saturation stress, θdis s, at which no further dislocation strength-ening occurs via stage III mechanisms, can be seen in Figure 6.5, and calculated by:σdis s =αMGbk1k2(6.5)The initial work hardening rate for dislocation contribution to flow stress is given by:θdis 0 =αMGbk12(6.6)As the steel in this investigation does not have a clear yield point, it is useful to follow the workof Cheng et. al. [150]. Similar to Cheng’s work, θmax is defined by the intersection point betweenthe linear extrapolation of the elastic-plastic transition region and of the stage III region on the θversus σtrue graph as shown in Figure 6.5.From the relations above, the physical basis of these parameters can be seen. θmax is propor-tional to k1 (the dislocation storage rate) according to equation 6.6, and thus can be used by directcomparison to examine this term. The dynamic recovery term of the work hardening equation, k2,can be determined from the slope in stage III according to:dθdσ=−k22(6.7)Importantly, k1 represents an athermal term related to θmax and k2 represents a thermally dependentterm related to dθ/dσ . Consequently the value of ∆σθmax (as shown in Figure 6.5) represents acombination of these terms. With these terms the effect of test temperature on work hardening can99CHAPTER 6. DISCUSSIONbe examined critically.To give an example of the analysis from the current study, Figure 6.6 shows an example of thedata from a single condition tested at the three testing temperatures. As it is not typical to see valuesplotted against stress it is helpful to recognize the trends between the more common stress-strainand θ -strain plots and θ -stress plots. At the start of the tests the hardening rate is equal to the elasticmodulus, and drops from there as the material progresses into the plastic region. This effect is seenin both Figure 6.6 (b) and (c) which show hardening rate plotted against true strain and true stressrespectively. The value in continuing the analysis becomes evident in Figure 6.6 (d). This plot of θversus σ −σys reveals the θmax and dθ/dσ at a glance which are in turn directly related to the k1and k2 respectively.Figure 6.6: Analysis of tensile tests data from the condition with Nb in solution with a prioraustenite grain size of 42 µm grain size and 10 ◦C/s cooling rate (lower bainite primarymicrostructure). Plots of (a) True stress vs true strain (b) θ vs true strain (c) θ vs truestress and (d) θ vs σ −σys.100CHAPTER 6. DISCUSSIONFigure 6.7 shows the comparison of the θ versus σ −σys plots at all testing temperatures foran example condition of each primary microstructure. From this figure, an important observation isthat width of the curves, which is related to the extent of work hardening, decreases from the ferriteto the mixed ferrite/bainite, to the upper bainite and finally to the lowest in the lower bainite.Figure 6.7: θ vs σ −σYS of conditions with (a) ferrite primary microstructure (Nb in precip-itates, 5 µm prior austenite grain size, 10 ◦C/s) (b) mixed ferrite/bainite primary mi-crostructure(Nb in solution, 5 µm prior austenite grain size, 50 ◦C/s), (c) upper bainiteprimary microstructure (Nb in precipitates, 42 µm prior austenite grain size, 10 ◦C/s),and (d) lower bainite primary microstructure (Nb in solution, 42 µm prior austenite grainsize, 10 ◦C/s).For more detailed analysis, Table 6.1 presents the work hardening parameters for all tests andconditions while Figure 6.8 displays this data with respect to material properties (e.g. yield strength)or testing conditions (i.e. temperature).Figure 6.8 (a) and (b) shows that the extent of work hardening (∆σθmax) has correlation to both101CHAPTER 6. DISCUSSIONTable 6.1: Work hardening operational parametersType Temperature θmax (MPa) −dθ/dσ ∆σθmax (MPa)Nb in precipitates, 5 µm @ 10 ◦C/s Ambient 6730 27 254Ferrite -20 ◦C 6860 20 341-60 ◦C 7660 21 373Nb in precipitates, 5 µm @ 50 ◦C/s Ambient 5730 24 236FerriteNb in solution, 5 µm @ 10 ◦C/s Ambient 6710 24 277Mixed Ferrite/Bainite -20 ◦C 6610 22 302Nb in solution, 5 µm @ 50 ◦C/s Ambient 7280 28 257Mixed Ferrite/Bainite -20 ◦C 7360 26 285-60 ◦C 7280 25 296Nb, in precipitates, 42 µm @ 10 ◦C/S Ambient 5430 26 206Upper Bainite -20 ◦C 6360 24 263-60 ◦C 7100 23 311Nb, in precipitates, 42 µm @ 50 ◦C/S Ambient 5350 32 166Upper Bainite -20 ◦C 6600 27 240-60 ◦C 5260 22 240Nb in solution, 42 µm @ 10 ◦C/s Ambient 5020 33 150Lower Bainite -20 ◦C 4320 25 171-60 ◦C 4650 22 213Nb in solution, 42 µm @ 50 ◦C/s Ambient 5400 47 114Lower Bainiteyield strength and testing temperature. ∆σθmax is seen to decrease with increasing yield strength inFigure 6.8 (a) and increasing with decreasing temperatures in Figure 6.8 (b). The increase of ∆σθmaxwith decreasing temperatures can be explained in part by considering the effect of temperature ondynamic recovery as represented by −dθ/dσ . The −dθ/dσ value is seen to to decrease withtemperature as in Figure 6.8 (c). If θmax is fixed, the influence of a lower −dθ/dσ would cause anincrease in ∆σθmax with decreasing temperature.In Figure 6.8 (d) it can be seen that there is dependence of θmax with microstructure. Themicrostructures with notable amounts of MA (i.e. the mixed ferrite/bainite, and the upper bainitemicrostructures) show a lower value of θmax compared to the other microstructures. All θmax valuesfall within the range of 5000 to 7500 MPa which compares well with literature on BCC metalswhich estimates θmax to be equal to the shear modulus divided by 10 to 20, which works out toapproximately 4000 to 8000 MPa [151] for steels. With regard to temperature, θmax appears to102CHAPTER 6. DISCUSSIONFigure 6.8: Plots of (a) ∆σθmax vs σys 0.5% offset at ambient testing temperature, (b) ∆σθmax vstesting temperature,(c) −dθ/dσ vs temperature, and (d) θmax vs temperature independent (i.e. within ±10%) with two exceptions. The exceptions are thoseconditions of Nb in precipitates with a prior austenite grain size of 42 µm with cooling rates of10 and 50 ◦C/s which are microstructures of mixed ferrite/bainite with significant MA which maysignificantly contribute to the work hardening. As θmax is related to the k1 term a temperaturedependency is not expected for single phase materials.A close examination of −dθ/dσ in Table 6.1 shows that the dynamic recovery term decreasesmost significantly in the lower bainite specimen, mainly upper bainite condition with the finer struc-103CHAPTER 6. DISCUSSIONture, and the ferrite specimen. The microstructures of primarily mixed ferrite/bainite, and upperbainite with a coarser structure show very little decrease of −dθ/dσ with decreasing testing tem-perature. This effect is clearly seen in Figure 6.8 (c), however it is not immediately clear whyphysically there would be such a significant difference in these specimens with respect to dynamicrecovery processes between the microstructure types.A recent approach by Bouaziz [152] to the Kocks-Mecking analysis proposes a modified dislo-cation density evolution law which can be applied to the current data. Bouaziz’s proposed disloca-tion density evolution law is given as:dρdεp= M(kb√ρ exp(−ξ√ρ))(6.8)where M is the Taylor factor, k is a constant describing dislocation accumulation similar to k1 fromKocks-Mecking analysis, and ξ is a characteristic length scale which can physically be interpretedas the capture distance for dynamic recovery. Note that this equation reduces to the Kocks-Meckingdislocation evolution equation if ξ√ρ  1.Using this evolution law, along with the dislocation density strengthening contribution a simplemodel of the stress strain curve can be made. In this model, the initial yield stress is set and thedislocation strengthening contribution is added to it, changing with incremental steps of strain. Thismodel has four input parameters which are, k, the initial yield strength σys 0, initial dislocation den-sity ρ0, and ξ value. Fitting these to the experimental curves is then done with three reasonable re-strictions. Firstly, the k value is taken as a constant of 0.2 throughout all conditions and temperaturesas it is assumed that the storage of dislocations in these microstructures are the same and is temper-ature and strain rate independent. Secondly, initial dislocation densities for a given microstructureremain the same at all test temperatures. That is to say the initial dislocation density of a microstruc-ture is related to the transformation start temperature for given microstructure. Initial dislocationdensities are selected to be within an reasonable range for a given microstructure, increasing withdecreasing transformation start temperature (Tstart) [16]. The relation between the selected ρ for themodel and approximate Tstart of the microstructures is shown in Figure 6.9. The third restriction isthat for a given test temperature all test conditions should have the same ξ value. The ξ value is104CHAPTER 6. DISCUSSIONFigure 6.9: The relation between selected initial dislocation density and the approximate trans-formation start temperature based on the experiments of Reichert [147].related to the rate of dynamic recovery for a given overall dislocation density. It is assumed thatdynamic recovery has a temperature dependency as shown in Figure 6.10. The remaining fittingparameter then is the initial strength which is temperature dependent which is selected to give thebest fit of the model to the current condition at the current temperature. By numerically integrat-ing equation 6.8 the dislocation density evolution is determined with incremental steps of plasticstrain. This contribution to the flow stress is then calculated using the Taylor equation as discussedpreviously and repeated here:σdis = αMGbρ1/2 (6.9)To compare this model with the true stress-strain curves from experiments, the plastic strain forthe experiments is determined according to:εplastic = εtotal− σtrueE (6.10)where εplastic is the plastic strain, εtotal is the measured total strain, σtrue is the measured true strain,and E is the Young’s modulus. Note the second term on the right hand side of the above equationrepresents the elastic strain. With εplastic now defined, a comparison of the model to the experiment105CHAPTER 6. DISCUSSIONFigure 6.10: The selected value of ξ (i.e. related to dislocation capture distance for dynamicrecovery) versus testing temperaturecan be meed. Figures 6.11 and 6.12 shows this at all testing temperatures for tests from microstruc-tures with a prior austenite grain size of 5 µm and 42 µm, respectively. Table 6.2 summarizes themodel parameters used in fitting the model.From the figures showing the model and experimental stress-strain curves it is seen that themodel using the Bouaziz evolution dislocation density evolution law provides a good fit to all workhardening data. As this works for such a wide number of conditions with very few changes to pa-rameters, this model is quite robust in describing work hardening behaviour. This model can beseen as a simple constitutive relation that accounts for test temperature via the initial yield strengthand ξ values. However, the model does not well describe the observed elastic-plastic transition(e.g. figure 6.12 (d)), and this presents a limitations of the model. Further discussion of mi-crostructure relations to work hardening behaviour will be revisited in the following discussionof the microstructure-property relationships.106CHAPTER 6. DISCUSSIONFigure 6.11: Model compared to true stress-plastic strain of 5 µm prior austenite grain sizewith Nb in precipitates with cooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, and Nb insolution with cooling rates of (c) 10 ◦C/s and (d) 50 ◦C/s.107CHAPTER 6. DISCUSSIONFigure 6.12: Model compared to true stress-plastic strain to failure of 42 µm prior austenitegrain size with Nb in precipitates with cooling rates of (a) 10 ◦C/s and (b) 50 ◦C/s, andNb in solution with cooling rates of (c) 10 ◦C/s and (d) 50 ◦C/s.108CHAPTER 6. DISCUSSIONTable 6.2: Work hardening model parametersType Temperature ρ0 (1/m2) σys 0 (MPa) ξ (m)Nb in precipitates, 5 µm @ 10 ◦C/s Ambient 1.0x1010 425 2.15x10−7Ferrite -20 ◦C 1.0x1010 430 1.95x10−7-60 ◦C 1.0x1010 440 1.65x10−7Nb in precipitates, 5 µm @ 50 ◦C/s Ambient 5.0x1010 485 2.15x10−7FerriteNb in solution, 5 µm @ 10 ◦C/s Ambient 4.0x1011 435 2.15x10−7Mixed Ferrite/Bainite -20 ◦C 4.0x1011 440 1.95x10−7Nb in solution, 5 µm @ 50 ◦C/s Ambient 8.0x1011 500 2.15x10−7Mixed Ferrite/Bainite -20 ◦C 8.0x1011 515 1.95x10−7-60 ◦C 8.0x1011 535 1.65x10−7Nb, in precipitates, 42 µm @ 10 ◦C/S Ambient 5.0x1012 405 2.15x10−7Upper Bainite -20 ◦C 5.0x1012 420 1.95x10−7-60 ◦C 5.0x1012 425 1.65x10−7Nb, in precipitates, 42 µm @ 50 ◦C/S Ambient 2.0x1013 485 2.15x10−7Upper Bainite -20 ◦C 2.0x1013 490 1.95x10−7-60 ◦C 2.0x1013 500 1.65x10−7Nb in solution, 42 µm @ 10 ◦C/s Ambient 3.5x1013 470 2.15x10−7Lower Bainite -20 ◦C 3.5x1013 505 1.95x10−7-60 ◦C 3.5x1013 505 1.65x10−7Nb in solution, 42 µm @ 50 ◦C/s Ambient 5.0x1013 605 2.15x10−7Lower Bainite109CHAPTER 6. DISCUSSION6.4 Microstructure-Tensile Properties RelationshipsWith results of the microstructures and the mechanical properties from the Gleeble heat treatedsamples having been presented, the relation of how the former leads to the latter can now be dis-cussed. This discussion will focus on how microstructure relates to the properties determined fromthe tensile tests at ambient temperatures. Where prominent, the discussion is expanded to examinetemperature effects on the microstructure-tensile property relationship.The tensile properties up until necking can be seen to be primarily affected by two microstruc-ture features. These two features are the constituents and the characteristic length scale of the mi-crostructure. This can be seen for example in Figure 6.13 which shows the different yield strength,tensile strength and uniform elongation against the primary microstructural constituent at ambienttesting temperatures. From this graph it is seen that the lowest strengths but highest uniform elon-gations are given by the samples with primarily ferrite as the major constituent, with a coarse mor-Figure 6.13: Yield strength, tensile strength and uniform elongation at ambient temperaturesplotted for each microstructure.110CHAPTER 6. DISCUSSIONphology. At the other extreme, the highest strengths, which correlates with lowest uniform elon-gation, are found in the microstructure with lower bainite as the main constituent with the fineststructure. Of samples with similar microstructures, it is seen that as the microstructure is refinedboth yield and tensile strength increases, while uniform elongation decreases.To discuss the relation of microstructure and tensile properties, it was seen in Section 2.8 thatequation 2.5 well represented the individual contribution mechanisms to the strength of a specimen.This equation is repeated here:σtot = σ0 +σss+σppt +σdis+σgb (6.11)To examine further the differences in strength resulting from microstructural changes, it is usefulto discuss the influence of the various microstructures on the strengthening terms of equation 6.11.To quickly review, the strengthening contributions can be separated into the inherent strength of thelattice to dislocation movement (σ0), solid solution strengthening (σss, precipitation strengthening(σppt), dislocation interaction strengthening (σdis), and grain boundary strengthening σgb). Eachof these contributions will now be discussed individually with respect to the microstructural con-stituents observed. As the primary microstructures (ferrite, upper bainite, lower bainite) are BCCmatrices, it would be expected that σ0 would be similar for each of these phases. On the otherhand, the MA phase will have a different σ0, as both of the structures that make up this phasehave a different lattice structure (body centered tetragonal (BCT) for martensite, FCC for austen-ite). As the fraction of MA is relatively low in the ferrite and lower bainite specimens (i.e. lessthan 5%) and to a first approximation can be ignored. The more significant MA fractions (i.e.≈10%) in the mixed ferrite/bainite as well as upper bainite specimens mean that the σ0 may needto be accounted for. The relative amount of martensite and retained austenite remains difficult tocharacterize. It is not clear how significant the influence of these phases will be on the compositestrength of the microstructures, although recent preliminary work by Reichert [153] using EBSDsuggests that martensite primarily makes up the MA in these cases. The BCC structures of ferrite,upper bainite, and lower bainite along with the BCT structure of martensite would be expected tohave a relatively high Peierls-Nabarro strength and strong temperature dependence. On the other111CHAPTER 6. DISCUSSIONhand the FCC austenitic structure would not be expected to have a strong temperature dependence.Specimens with a high amount of MA (i.e. mixed ferrite/bainite, and upper bainite) may show lesstemperature dependence in the σ0 term if the austenite is the predominate constituent of the MAphase.Solid solution strengthening changes in two ways for this study, firstly by design wherein thepresence of Nb(C,N) is controlled, and secondly how the solid solution strengthening elementsare distributed in the microstructure with respect to its constituents. While Nb is noted to have amoderate solid solution strengthening effect, it is relatively unimportant compared to the interstitialsolutes of C and N. Therefore it is most important to consider the amount present and the distributionof these elements, and in particular the carbon. The primary phases of ferrite, upper bainite, andlower bainite are known to have low solubilities of C. It is expected, regardless of the presence ofNb(C,N), that C which is in solution becomes distributed into the MA phase or precipitated out inanother form (such as interlath carbides in upper bainite, or intralath carbides in lower bainite), orsome combination thereof. In the case of C enriching the MA phase, a significant increase of theσss component for each of these phases would be expected. In samples where Nb(C,N) has beendissolved there is potential for the MA phase to reach higher carbon contents, and consequentlyhigher solid solution strengthening contributions. However, as there is only a limited amount ofNb in this chemistry, it can be expected that the difference is not significant. Consequently, thecontribution of solid solution strengthening is likely to be similar and limited in significance for theprimary microstructure constituents of ferrite, upper bainite and lower bainite.Consideration of changes to the strengthening contribution of precipitates follows a similar ar-gument to that of solid solution. That is, changes to σppt are determined on one hand by the amountof Nb(C,N) precipitates present by design, and secondly by which precipitates form and how theyare distributed in each constituent. Regarding the Nb(C,N) precipitates, these are found in the as-received material to be well distributed throughout the microstructure and consisting of two setsof sizes (i.e. 2 nm and 69 nm respectively) [143]. It is assumed that the majority of Nb in the as-received material is contained within the Nb(C,N) precipitates. In the case of Nb in precipitates witha 5 µm prior austenite grain size, no solutionizing treatment was applied and the final microstruc-tures (which are the samples with the main constituent of ferrite) are expected to have the samedistribution Nb(C,N) precipitates. In the sample with Nb in precipitates and a 42 µm prior austenite112CHAPTER 6. DISCUSSIONgrain size, which are the specimens with upper bainite as the primary constituent, the solutionizingtreatment combined with the 20 minute reprecipitation only resulted in about 50% of the Nb beingin precipitates [147]. As these precipitates are formed in austenite, they are expected to be relativelycoarse and thus their strengthening contribution would be low.Besides Nb(C,N) precipitates, there are many others that may contribute to the strengthening ef-fect such as TiN, or MoC. Importantly, TiN is not significantly influenced by the thermal treatmentsapplied here and will remain consistently distributed in all microstructures. MoC on the other handwill dissolve in solutionizing treatments and will not precipitate under the cooling conditions ap-plied, which means this distribution should remain consistent across the microstructures as well. Animportant difference is recognized in the distribution of carbides between upper bainite and lowerbainite constituents exists. In upper bainite, larger sized carbides are formed between bainite laths.As these are not in the matrix of the upper bainite, they will not interfere directly with the dislocationmotion and their contribution to σppt will be limited. In lower bainite, carbides also form betweenthe bainite laths, although these precipitates are smaller. Furthermore, in lower bainite carbides alsoform within the laths, but as these are relatively coarse they will not likely add significant strengthto σppt .As previously stated, there is a significant increase in strength observed for samples with thesame primary microstructure but different characteristic length scales. For cases where only thelength scale is changing with the same primary microstructure then the most significant change onthe tensile strength would be through the grain size effect on strengthening. This contribution typi-cally defined by the Hall-Petch equation, as given in equation 2.11, shows that grain size strength-ening is inversely proportional to the square of the effective grain size. In ferrite specimens theeffective grain size can be taken to be equal to the actual grain size. For microstructures with upperand lower bainite, the effective grain size can be associated with the lath size. In every case, theeffect of a higher cooling rate but with a same final primary microstructure constituent results in asignificant refinement of microstructure (i.e. effective grain size) and consequently a higher strengthfor these structures. Grain size thus has a critical influence on the strengthening contributions.The microstructure relation to the dislocation strengthening contribution (σdis) will be discussedfirst in terms of the initial dislocation density which influences the yield strength, and then in termsof the evolution of the dislocation density (and thus strain hardening behaviour). The initial dis-113CHAPTER 6. DISCUSSIONlocation density differs between the microstructural constituents. The initial dislocation densityincreases progressively with lower temperature transformation products, which is to say ferrite hasa lower initial dislocation density than upper bainite, which is less than lower bainite, which is lessthan martensite. As a result, the initial dislocation strengthening contribution will follow the sametrend. This effect is likely to be a major factor in this study across the various microstructures, andindeed the observed yield strength follows this trend.The strengthening contribution and resulting work hardening behaviour is complexly related tothe microstructure. The dislocation strengthening contribution, previously presented in equation 2.9,is repeated here:σdis = αMGbρ1/2 (6.12)The dislocation evolution equation from Bouaziz, previously introduced in equation 6.8, and re-peated here:dρdεp= M(kb√ρ exp(−ξ√ρ))(6.13)It is of interest to discern how the initial dislocation density, dislocation storage (k), and dynamicrecovery processes (ξ ) vary with microstructure. Utilizing the discussion presented on the Kocks-Mecking analysis of the results, some speculation can be made as to how microstructure affects thevarious parameters. The initial dislocation density varies with microstructure according to the strainof transformation. It is expected that the initial dislocation density increases with decreasing trans-formation start temperature of the microstructural constituent as discussed previously, and indeedcorrelates with the yield strengths observed from the thermally treated specimens. Alongside thegrain size strengthening contribution, the dislocation strengthening contribution also represents aprimary influence on the strength differences between microstructures.So far the discussion of the dislocation strengthening contribution has focused on the initialyield strength, and thus the initial dislocation density, of the material. The evolution of strength114CHAPTER 6. DISCUSSIONwith strain, and thus the evolution of dislocation density is important and is seen to vary withmicrostructure constituent. In the Bouaziz model it was seen that the dislocation storage term k isrelated to θmax while dynamic recovery processes represented by the ξ term are related to−dθ/dσ .First considering the dislocation storage, it is seen that there is a weak correlation at least at ambienttemperatures of θmax on yield strength, with θmax decreasing slightly with increased yield strength.This trend is at best weakly correlated with the initial dislocation density relation to transformationstart temperatures (i.e. ferrite specimens with low initial dislocation densities give higher θmaxvalues and lower bainite specimens with higher initial dislocation densities give lower θmax values).The dynamic recovery, or the ξ term, is related to −dθ/dσ value from the Kocks-Meckinganalysis. This value does not show any consistent trend with microstructure constituent. The ma-jority of the−dθ/dσ values at ambient temperatures fall within the range of 24 to 33, but the lowerbainite microstructure with the fine length scale show a significantly higher−dθ/dσ of 47. It is notimmediately clear why −dθ/dσ for this exceptional condition is so high. The fact that this excep-tional condition shows a very small amount of work hardening means that there is a small range toevaluate, as well the non-distinct elastic-to- plastic transition may explain some uncertainty in thisvalue. Interestingly, it was shown that when the Bouaziz dislocation model is used, the fit even tothis extreme condition appears to be fairly good with one value of ξ (i.e. capture distance for dy-namic recovery) at ambient temperature. This supports the assertion that this point of −dθ/dσ forlower bainite is an outlier due to analysis sensitivities. Furthermore, the use of the model highlightsthe fact that dynamic recovery is insensitive to the differences between these microstructures.The dynamic recovery term is also expected to have a temperature dependency. This depen-dence is seen with −dθ/dσ decreasing with decreasing test temperature. This in turn gives thework hardening behaviour a temperature dependence. This effect is significant as work hardeningwas seen to differ with temeprature clearly from the experimental stress-strain curves. Using theBouaziz model, a clear temperature dependency of the ξ value was seen to fit all curves well. Fromobservations, the decrease in −dθ/dσ with decreasing temperature is noted to be greatest in sam-ples with primary microstructures of lower bainite, and of upper bainite with a finer microstructure.It is uncertain why these cases in particular show a stronger decrease of −dθ/dσ by observations,but as the ξ fit to the stress strain curves is robust it is likely that the stronger decrease may again,be due to an issue with sensitivity in the analysis.115CHAPTER 6. DISCUSSIONThe grain boundary strengthening effect was previously discussed on the basis of comparingbetween microstructures of the same primary constituent. It was discussed that this effect was quitesignificant within the basis of consideration. Considering now the difference between microstruc-tures on the basis of their effective grain sizes (i.e. related to the typical grain size definition inferrites, or related to lath spacings in bainites), it is reasonable to correlate the increase of yieldstrength microstructures with the decrease in effective grain sizes when comparing microstructures.As the effective grain sizes vary significantly with microstructure (ferrite given the largest, andlower bainite the lowest), the strength can be expected as well to vary significantly. This is indeedobserved in the results.6.5 Kahn Tear TestingIn Kahn tear testing it was seen that all conditions failed in a ductile manner. This is in line withthe ductile failure seen in all conditions of the tensile tests. As such, Kahn results detail differencesin ductile failure between the different microstructures. In distinction from the tensile tests, thepresence of a notch then growing crack allows for the examination of this ductile failure in a stateof higher stress triaxiality. Before such comparisons are made, a critical look at the testing methodis presented.The Kahn tear test was selected for this study for numerous reasons, but particularly that ofits compatibility with Gleeble specimen dimensions and simplicity in setup and measurement. Inturn, there were a number of issues related to using this test as a fracture test. Foremost is thefact that the plane-strain condition is not reached. A quick calculation estimates that the planestrain condition would require a thickness over 17 mm. This thickness cannot be met due to bothlimitations of the Gleeble testing, and the simple fact that this is actually thicker than the suppliedplate. Consequently, there will be an influence of thickness and plasticity on the ability to measurea true fracture toughness in a simple manner. It was seen in testing (see Figure 5.24) that even therelative measures of toughness of Etear and UPEd0−1 showed a dependence on thickness. Whilethese values were corrected for thickness previously, and henceforth these corrected values arereferred to as Etear and UPEd0−1, they still do not represent true fracture toughness. As both ofthese values are determined using an energy taken directly from the load displacement curve theydo not only represent of the energy associated with the fracture process. Consequently, there is no116CHAPTER 6. DISCUSSIONsimple analytical relation between these values and typical true toughness measures.The inability of Etear and UPEd0−1 to measure a true fracture toughness does not preclude theuse of these values to potentially represent the toughness in a relative but quantitative way of thesamples tested here. As the tested conditions have somewhat similar tensile properties (i.e. yieldingand strain hardening is comparable for all specimens), it is reasonable to expect that any observeddifferences in these values between conditions will be the result of microstructural differences onductile fracture. That is to say the effects from different tensile properties on plastic zone size devel-opment with respect to both in-plane plastic zone size and the extent of thinning at the crack tip isrelatively insignificant over the range of yield strengths, tensile strengths, and strain hardening ratesas observed in tensile testing. In addition, it is important to consider the error in these measure-ments. Specifically, the use of a crack measurement in Etear calculations introduces a possibility forgreater error associated with accuracy in the crack length measurement. In developing the mastercurve for crack elongation with displacement at a given thickness it was seen that there was moder-ate but acceptable scatter in the crack length measurements. Additionally, there was an assumptionof applicability of these crack curves from ambient temperatures to the lower testing temperatures.As the material largely shows comparable stress-strain responses at these temperatures it seems rea-sonable that crack growth will occur in a similar manner. Considering then that Etear and UPEd0−1can be taken as relative measures of ductile fracture toughness, the discussion can move on to ex-amine the effectiveness of using these values in practice to compare ductile fracture resistance ofdifferent microstructures.Between the values of Etear and UPEd0−1 there are some differences in ranking the microstruc-tures. Etear represents a more rigorous method of measuring relative toughness to which UPEd0−1value can be compared. However, given the practical difficulties with determining the Etear valuethe question remains if UPEd0−1 can adequately be used. This comparison can easily be observedby plotting Etear against UPEd0−1 normalized by their maximum values, as shown in Figure 6.14.Note for this figure, and future figures referring to Kahn properties, error bars are defined basedon the standard deviation of nine tests of the as-received material. In this figure, if these two val-ues were perfectly proportional the data would collapse to a direct relation. In doing so, somedifferences are noted. Notably, for the condition with Nb in precipitates, prior austenite grainsize of 5 µm and 10 ◦C/s cooling show quite low values of Etear, indeed the lowest values, but117CHAPTER 6. DISCUSSIONfrom the UPEd0−1 this condition is not so low compared to the other conditions. The condi-tion with Nb in solution, prior austenite grain size of 5 µm and 50 ◦C/s cooling shows a signif-icant increase in ranking when comparing Etear and UPEd0−1 values. In fact, this condition re-sults in one highest corrected UPEd0−1. Consequently, the use of UPEd0−1 values is not recom-Figure 6.14: Corrected and normalized Etear plotted against corrected and normalizedUPEd0−1. The symbol in the upper left represents the uncertainty for each data point inthe figure.mended over Etear for comparing relative toughness values as it will lead to a potentially differentranking. Only when the difficulty of obtaining crack length measurements prohibits determinationof Etear values should the UPEd0−1 value be used.Using the Etear value then, it is possible to look into microstructure property relationship. Fig-ure 6.15 plots the Etear value at various temperatures against the testing condition. There are threenotable trends. First, decreasing testing temperatures tends to be associated with an increase in theEtear value, with the most notable exception of this trend being the condition of Nb in precipitateswith a cooling rate of 10 ◦C/s. It is possible considering the error within the experimental data thatthis is not actually an exception, and other may be exceptions to this trend, however it is expectedthat for every condition there would be no exception to the stated trend. Secondly, there is an in-crease of Etear with a higher cooling rate which creates a finer final microstructure. Finally, as theEtear appears to have a modest increase with increasing amounts of bainite, with the highest valuesbeing observed in lower bainite.In ductile failure, voids nucleate, then grow and finally coalesce causing fracture. In stable crack118CHAPTER 6. DISCUSSIONFigure 6.15: Corrected Etear plotted against the testing condition.growth, this process occurs continuously. The mechanism by which ductile voids grow involve theplastic motion of dislocations. As such, influences on yield strength will also effect the ductile frac-ture energy. It is unsurprising to find that a decrease in temperature, known and shown previouslyhere to cause an increase in yield stress, has resulted in the observed positive correlation betweentest temperature and Etear. Similar to this, the decrease in coarseness of a microstructure correlateswith an increase to yield strength, and as such can explain the increase of Etear with increasingcooling rate for a given primary microstructure. In general, this trend of yield strength correlatingto an increase of Etear seems to hold fairly well and may also in part explain the differences betweenmicrostructures. This is exemplified in Figure 6.16 which plots the tensile yield strength againstthe Etear value. Note there are two exceptional cases, that of the fine upper bainite microstructureand that of the coarse lower bainite microstructure, showing lower Etear values then expected fortheir σ0.5%ys values. Unexpectedly, this relatively low Etear value is seen only for these conditionsat ambient temperatures.119CHAPTER 6. DISCUSSIONFigure 6.16: σ0.5%ys plotted against corrected Etear.Figure 6.16 showed that there may be relations that can be made between Kahn and tensile tests,specifically between σ0.5%ys and Etear. It may be insightful to examine data between these tests toreveal any other potential relations. Figure 6.17 shows a number of plots comparing tensile to Kahntest properties. First, Figure 6.17 (a) and (b) show the expected relation of strength values (yield andtensile strengths respectively) to the tear strength. The fact that these both relate to the tear strengthis unsurprising but lends confidence to the consistency of the testing method. Figure 6.17 (c) showsthe tensile fracture strain plotted against Etear while Figure 6.17 (d) shows the corrected tensilefracture stress plotted against the tear strength. What is seen is a lack of any correlation betweenmeasured tensile properties at fracture and properties from the Kahn tear test. This shows thepractical limitation of tensile testing to characterize the ductile tearing behaviour. It also validatesthe use of Kahn tear tests in this work as it provides insight into the ductile fracture properties thatotherwise would not have been observed.To emphasize again, no brittle behaviour is observed. As such, the relative comparison oftoughness is based on the ductile fracture of these specimens. While it was not possible to expandthe testing matrix due to limited materials and time, it would have been informative to continueKahn testing to colder temperatures or increase the crosshead speed so as to observe and comparethe conditions under which these microstructures undergo the ductile to brittle transition. Whilethicker specimens would also have been desirable (e.g. would have provided more constraint atthe notch root), producing these microstructures in a controlled and consistent way as was done in120CHAPTER 6. DISCUSSIONFigure 6.17: Tensile properties compared to Kahn properties. (a) σ0.5%ys plotted against σtear,(b) σUTS plotted against σtear, (c) ε f rac plotted against corrected Etear, and (d) correctedσ f rac plotted against σtear. The symbol in the bottom right of (c) and (d) represents theuncertainty for each data point in the figure.121CHAPTER 6. DISCUSSIONthis study would not have been possible. Ultimately, this leaves a significant gap in evaluating thetoughness of these microstructures relevant to the HAZ. On the other hand, the testing conditionsreached a fairly extreme temperature (-60 ◦C) with all microstructures showing a ductile failure.This in and of itself suggests that these microstructures are inherently tough, and consequently thesteel chemistry used here is robust.122CHAPTER7Concluding Remarks7.1 SummaryThe HAZ of single and dual torch welds experience a very complicated thermal history. This his-tory varies spatially around the weld and produces a gradient of microstructures. The mechanicalproperties are directly related to the microstructure and as a result any tests on actual welds willsample a range of microstructures. The objective in this work was to create bulk specimens of indi-vidual microstructures so that detailed characterization of the structure property relationship couldbe carried out. This was accomplished by first making temperature measurements in the real HAZof a full thickness pipeline GMAW weld performed with parameters expected to be in use in fieldconditions, and with single torch, as well with dual torch with spacings of 2.75”, 4”, and 7”. Basedon microstructure work from the real HAZ, and from project work from colleagues, specific mi-crostructures were selected then reproduced in bulk specimens for mechanical testing. Tensile andKahn tear testing was performed at ambient, -20 ◦C, and -60◦C in order to investigate how thesemicrostructures perform under the potentially extreme service conditions that they may be exposedto.The following summarizes the primary results of this work:• Cooling rates in the HAZ for the main fill passes were observed to be on average 100 ◦C/sec insingle torch welds, on average 54 ◦C/sec in leading torch of dual torch welds, and on average123CHAPTER 7. CONCLUDING REMARKS18 ◦C/sec in the trailing torch of a dual torch welds. Microstructures observed in the HAZshowed varying amounts of ferrite, upper bainite, lower bainite, and MA.• Thermal simulations for creating bulk specimens suitable for mechanical testing were refinedin order to produce microstructures relevant to the observed HAZ. Microstructures consistingprimarily of ferrite, primarily of upper bainite, and primarily of lower bainite were producedwith at least two different refinements. All microstructures were characterized with respect tothe amount of primary microstructural constituents as well as for their amount of MA phase.• Tensile testing showed that the highest strengths and lowest ductility was observed in mi-crostructures with the primary constituent of lower bainite. Ferrite, mixed ferrite/bainite andupper bainite showed comparable strength levels, however it is seen that ductility was de-creased for samples with a primarily upper bainitic microstructure.• The Bouzaiz dislocation based model for work hardening was shown to provide a good fitacross all temperatures and conditions with minimal fitting parameters. Interestingly, it wasseen that the ξ value, which is related to the dislocation capture distance for dynamic recovery,was observed to linearly decrease with temperature.• The Kahn tear testing was investigated in terms of its ability to study the fracture of steel. Itwas found that the method cannot easily provide fracture toughness parameters such as KIC,but can provide a relative measure of toughness. Kahn tear testing showed that even in thepresence of a stress concentration at cold temperatures (-60 ◦C) that all microstructures failedin a ductile manner at the strain rates used.There are a number of important outcomes from this work. The temperature measurements ofthe dual torch welds in the HAZ represents a completely novel data set. This data can be used, as wasthe case in this work, to guide microstructure evolution work through the use of thermal simulations.It was a technical achievement to produce thermally treated specimens of a bulk microstructurebased on this data that was suitable in size for both tensile and Kahn tear testing. Most importantly,by examining conditions and microstructures that bound those observed in the real HAZ and thenmeasuring tensile and Kahn tear properties there is potential to utilize this work to form the basisof a full microstructure-property relationship to determine the local properties of microstructures124CHAPTER 7. CONCLUDING REMARKSin the HAZ at a wide range of potential operating temperatures. This is particularly useful if amicrostructural evolution model, as has been developed in the group at UBC, is created and used asan input for such a model. Furthermore, there is the potential to use this data to model the overallHAZ mechanical behaviour by combining each of the local microstructural responses.7.2 Future WorkThis work has left a number of potential opportunities for future work to follow:• Conduct additional testing of existing conditions where tests either failed or were limited innumber in order to improve confidence of results observed in this work.• Investigate opportunities for improvement to Kahn tear testing, both in general and with re-spect to properties relevant to this study. Specific to this study, it would be of interest toextend testing to colder temperatures, or at higher strain rates in order to investigate the tran-sition from ductile to brittle fracture. Using digital image correlation to examine crack growthand plastic zone development could be considered for a greater understanding of ductile frac-ture at room temperatures. The use of anti-buckling guides to prevent or delay buckling at thefar side of the ligament could be a way to extend the range over which useful information canbe observed. Additional analysis, such as utilizing two extensometers and accounting for thegeometry in rotation, could potentially be of interest in trying to obtain a more suitable directmeasure of fracture toughness properties.• Perform additional fracture testing with other methods, such as Charpy V-notch. This wouldbe useful in providing toughness properties that could be compared to the relative toughnessrankings observed from Kahn tear testing of the microstructures in this study.• Integrate the observations from this study into a combined model to determine overall me-chanical properties in the HAZ of a weld. In one such model, the welding parameters wouldbe taken as inputs to determine the thermal cycles at different positions in the HAZ, whichthen would be used to determine the microstructure evolution. The microstructures modeledat different positions are then used as an input into a mechanical model that utilizes observa-tions from this study to determine the mechanical properties in the HAZ.125Bibliography[1] Bing Liu, X.J. Liu, and Hong Zhang. Strain-based design criteria of pipelines. Journal ofLoss Prevention in the Process Industries, 22(6):884–888, November 2009. (Cited onpage 1.)[2] V. N. Marchenko and B. F. Zinko. Current trends in the development and production ofsteels and pipes for gas and oil pipelines: An international scientific-technical conference.Metallurgist, 52(3-4):167–174, 2008. (Cited on page 1.)[3] Laurie E. Collins, Dengqi Bai, Fathi Hamad, and Xiande Chen. High strength spirallinepipe for strain-based pipeline designs. The Seventeenth International Offshore and PolarEngineering Conference, January 2007. (Cited on pages 1 and 4.)[4] M. K. Gra¨f, H. Hillenbrand, C. J. Heckmann, and K. A. Niederhoff. High-strengthlarge-diameter pipe for long-distance high-pressure gas pipelines. International Journal ofOffshore and Polar Engineering, 14(1):69–74, 2004. (Cited on page 1.)[5] A. C. Walker and K. A. J. Williams. Strain based design of pipelines. The FourteenthInternational Conference on Offshore Mechanics and Arctic Engineering, 5:345–350, 1995.(Cited on page 1.)[6] Koui Kim, Wei Zhou, and Scott L. Huang. Frost heave predictions of buried chilled gaspipelines with the effect of permafrost. Cold Regions Science and Technology,53(3):382–396, August 2008. (Cited on page 1.)[7] Jean-Marie Konrad. Estimation of the segregation potential of fine-grained soils using thefrost heave response of two reference soils. Canadian Geotechnical Journal, 42(1):38–50,February 2005. (Cited on page 1.)[8] Scott L. Huang, Matthew T. Bray, Satoshi Akagawa, and Masami Fukuda. Fieldinvestigation of soil heave by a large diameter chilled gas pipeline experiment, fairbanks,alaska. Journal of cold regions engineering, 18(1):2–34, 2004. (Cited on page 1.)[9] A. G. Razaqpur and Daiyu Wang. Frost-induced deformations and stresses in pipelines.International journal of pressure vessels and piping, 69(2):105–118, 1996. (Cited onpage 1.)[10] T. Gladman. The Physical Metallurgy of Microalloyed Steels. Maney for the Institute ofMaterials, 1997. (Cited on pages 4, 6, 7, 8, 9, 23, 24, 25, 26, and 28.)[11] ASM International Handbook Committee. ASM Handbook, Volume 01 - Properties andSelection: Irons, Steels, and High-Performance Alloys. ASM International, 1990. (Cited onpage 4.)126BIBLIOGRAPHY[12] Daniel B. Lillig, Douglas S. Hoyt, Martin W. Hukle, John P. Dwyer, Agnes M. Horn, andKeith Manton. Materials and welding engineering for ExxonMobil high strain pipelines. InThe proceedings of the Sixteenth (2006) International Offshore and Polar EngineeringConference ISOPE, volume 4. International Society of Offshore and Polar Engineers, 2006.(Cited on page 4.)[13] A. J. DeArdo. Metallurgical basis for thermomechanical processing of microalloyed steels.Ironmaking & Steelmaking, 28(2):138–144, 2001. (Cited on page 4.)[14] J. P. Ormandy, M. Strangwood, and C. L. Davis. Effect of microalloying additions on steelplate to pipe property variations during UOE linepipe processing. Materials Science andTechnology, 19(5):595–601, May 2003. (Cited on page 4.)[15] H. Okamoto. C-Fe (Carbon - Iron), C (Carbon) Binary Alloy Phase Diagrams. In AlloyPhase Diagrams, ASM Handbook, volume 3. ASM Handbooks Online, ASM International, 1992. (Cited on page 5.)[16] Harry Bhadeshia and Robert Honeycombe. Steels: Microstructure and Properties:Microstructure and Properties. Butterworth-Heinemann, April 2011. (Cited on pages 6and 104.)[17] Nobutaka Yurioka. Physical metallurgy of steel weldability. ISIJ international,41(6):566–570, 2001. (Cited on page 7.)[18] W.T. Reynolds, F.Z. Li, C.K. Shui, and H.I. Aaronson. The Incomplete transformationphenomenon in Fe-C-Mo alloys. Metallurgical Transactions A, 21(6):1433–1463, 1990.(Cited on page 8.)[19] Seung Youb Han, Sang Yong Shin, Chang-Hyo Seo, Hakcheol Lee, Jin-Ho Bae, Kisoo Kim,Sunghak Lee, and Nack J. Kim. Effects of Mo, Cr, and V Additions on Tensile and CharpyImpact Properties of API X80 Pipeline Steels. Metallurgical and Materials Transactions A,40(8):1851–1862, August 2009. (Cited on page 8.)[20] Yong-Yi Wang, David Rudland, Rudi Denys, and David Horsley. A preliminary strain-baseddesign criterion for pipeline girth welds. In 2002 4th International Pipeline Conference,pages 415–427. American Society of Mechanical Engineers, January 2002. (Cited onpage 9.)[21] William Mohr, Robin Gordon, and Robert Smith. Strain-based design guidelines forpipeline girth welds. The Fourteenth International Offshore and Polar EngineeringConference, 1:10–17, January 2004. (Cited on page 9.)[22] Martin W. Hukle, Agnes M. Horn, Douglas S. Hoyt, and James B. LeBleu. Girth weldqualification for high strain pipeline applications. In ASME 2005 24th InternationalConference on Offshore Mechanics and Arctic Engineering, pages 369–374. AmericanSociety of Mechanical Engineers, 2005. (Cited on page 9.)[23] D. P. Fairchild, M. D. Crawford, W. Cheng, M. L. Macia, N. E. Nissley, S. J. Ford, D. B.Lillig, and J. Sleigh. Girth welds for strain-based design pipelines. In Proceedings of the18th International Offshore and Polar Engineering Conference, Vancouver, BC, Canada,pages 48–56. International Society of Offshore and Polar Engineers, 2008. (Cited onpage 9.)127BIBLIOGRAPHY[24] Fathi Hamad, Xiande Chen, and Laurie Collins. High toughness, submerged arc girth weldfor northern pipeline applications. In 2006 International Pipeline Conference, pages917–925. American Society of Mechanical Engineers, January 2006. (Cited on page 9.)[25] R. A. Teale and Milton Randall. Automatic Welding Procedures For X80 Pipeline Systems.The Third International Offshore and Polar Engineering Conference, pages 441–444, 1993.(Cited on page 9.)[26] J. A. Gianetto, J. T. Bowker, D. V. Dorling, and D. Horsley. Structure and Properties of X80and X100 Pipeline Girth Welds. In 2004 International Pipeline Conference, pages1485–1497. American Society of Mechanical Engineers, January 2004. (Cited on pages 9and 11.)[27] Susan R. Fiore, James A. Gianetto, Mark G. Hudson, Suhas Vaze, Shuchi Khurana, John T.Bowker, and David V. Dorling. Development of optimized weld metal chemistries forpipeline girth welding of high strength line pipe. In 7th International Pipeline Conference,pages 417–428. American Society of Mechanical Engineers, 2008. (Cited on page 9.)[28] Masahiko Hamada, Hidenori Shitamoto, and Hiroyuki Hirata. Tensile properties andmicrostructure of girth welds for high strength linepipe. The Nineteenth InternationalOffshore and Polar Engineering Conference, January 2009. (Cited on pages 9 and 11.)[29] J. T. Bowker, J. A. Gianetto, G. Shen, W. Tyson, and D. Horsley. Cross-weld tensileproperties of girth welds for strain-based designed pipelines. In 2006 International PipelineConference, pages 485–496. American Society of Mechanical Engineers, 2006. (Cited onpage 9.)[30] Hiroyuki Motohashi and Naoto Hagiwara. Effect of Strength Matching and StrainHardening Capacity on Fracture Performance of X80 Line Pipe Girth Welded JointSubjected to Uniaxial Tensile Loading. Journal of Offshore Mechanics and ArcticEngineering, 129(4):318–326, 2007. (Cited on page 9.)[31] T. A. Netto, A. Botto, and M. I. Loureno. Fatigue performance of pre-strained pipes withgirth weld defects: Local deformation mechanisms under bending. International Journal ofFatigue, 30(6):1080–1091, June 2008. (Cited on page 9.)[32] T. A. Netto, M. I. Loureno, and A. Botto. Fatigue performance of pre-strained pipes withgirth weld defects: Full-scale experiments and analyses. International Journal of Fatigue,30(5):767–778, May 2008. (Cited on page 9.)[33] Satoshi Igi, Teruki Sadasue, Takahiro Kubo, Nobuyuki Ishikawa, and Nobuhisa Suzuki.Effect of Crack Geometry And Tensile Properties On Tensile Strain Limit of X80 Linepipe.The Eighteenth International Offshore and Polar Engineering Conference, January 2008.(Cited on page 9.)[34] James A. Gianetto, John T. Bowker, David V. Dorling, David Taylor, David Horsley, andSusan R. Fiore. Overview of Tensile and Toughness Testing Protocols for Assessment ofX100 Pipeline Girth Welds. In 2008 7th International Pipeline Conference, pages 463–472.American Society of Mechanical Engineers, 2008. (Cited on page 9.)[35] Mark G. Hudson. Welding of X100 linepipe. PhD thesis, Cranfield University, 2004. (Citedon pages 11 and 12.)128BIBLIOGRAPHY[36] Fathi Hamad, Laurie Collins, and Riny Volkers. Effect of GMAW procedure on theheat-affected zone (HAZ) toughness of X80 (grade 550) linepipe. In 7th InternationalPipeline Conference, pages 553–569. American Society of Mechanical Engineers, 2008.(Cited on pages 11, 16, 17, 29, and 31.)[37] Øystein Grong. Metallurgical Modelling of Welding. Maney Publishing, 2nd edition, 1997.(Cited on pages 11, 13, and 15.)[38] Y. Chen, Y. Wang, and J. Gianetto. Thermal and microstructure simulation of high strengthpipeline girth welds. In Proceedings of the 18th International Conference on Offshore andPolar Engineering (ISOPE 2008), Vancouver, Canada, pages 33–39, 2008. (Cited onpage 11.)[39] Ibrahim K. Nuruddin. Effect of welding thermal cycles on the heat affected zonemicrostructure and toughness of multi-pass welded pipeline steels. PhD thesis, CranfieldUniversity, 2012. (Cited on page 12.)[40] A. S. Bilat, A. F. Gourgues-Lorenzon, J. Besson, and A. Pineau. Brittle Fracture inHeat-Affected Zones of Girth Welds of Modern Line Pipe Steel (X100). In Fracture ofNano and Engineering Materials and Structures, pages 109–110. Springer, 2006. (Cited onpage 12.)[41] Kioumars Poorhaydari-Anaraki. Microstructure and Property Examination of the Weld HAZin Grade 100 Microalloyed Steel. PhD thesis, University of Alberta, 2005. (Cited onpage 12.)[42] S. Kou and Y. Le. Three-dimensional heat flow and solidification during the autogenousGTA welding of aluminum plates. Metallurgical Transactions A, 14(11):2245–2253, 1983.(Cited on page 12.)[43] S Murugan, P. V Kumar, B Raj, and M. S. C Bose. Temperature distribution duringmultipass welding of plates. International Journal of Pressure Vessels and Piping,75(12):891–905, October 1998. (Cited on page 13.)[44] V.K. Goyal, P.K. Ghosh, and J.S. Saini. Process-controlled microstructure and castmorphology of dendrite in pulsed-current gas-metal arc weld deposits of aluminum andal-mg alloy. Metallurgical and Materials Transactions A, 38(8):1794–1805, July 2007.(Cited on page 13.)[45] V.K. Goyal, P.K. Ghosh, and J.S. Saini. Analytical studies on thermal behaviour andgeometry of weld pool in pulsed current gas metal arc welding. Journal of MaterialsProcessing Technology, 209(3):1318–1336, February 2009. (Cited on page 13.)[46] Abhay Sharma, Ajay Kumar Chaudhary, Navneet Arora, and Bhanu K. Mishra. Estimationof heat source model parameters for twin-wire submerged arc welding. The InternationalJournal of Advanced Manufacturing Technology, 45(11-12):1096–1103, December 2009.(Cited on page 13.)[47] Kenneth Easterling, editor. Introduction to the Physical Metallurgy of Welding.Butterworth-Heinemann, 2nd edition, 1992. (Cited on pages 13, 15, 17, and 29.)[48] B. de Meester. The Weldability of Modern Structural TMCP Steels. ISIJ International,37(6):537–551, 1997. (Cited on page 15.)129BIBLIOGRAPHY[49] Shi Zhong Liu. Metallography of HSLA steel weldments. Key Engineering Materials,69-70:1–20, 1992. (Cited on page 15.)[50] E. A. Wilson. The gamma-alpha transformation in low carbon irons. ISIJ International,34(8):615–630, 1994. (Cited on page 15.)[51] George Krauss and Steven W. Thompson. Ferritic microstructures in continuously cooledlow- and ultralow-carbon steels. ISIJ International, 35(8):937–945, 1995. (Cited onpage 15.)[52] G. Thewlis. Classification and quantification of microstructures in steels. Materials Scienceand Technology, 20(2):143–160, February 2004. (Cited on page 15.)[53] P. L. Harrison and R. A. Farrar. Application of continuous cooling transformation diagramsfor welding of steels. International Materials Reviews, 34(1):35–51, January 1989. (Citedon page 15.)[54] G. Spanos, R. W. Fonda, R. A. Vandermeer, and A. Matuszeski. Microstructural changes inHSLA-100 steel thermally cycled to simulate the heat-affected zone during welding.Metallurgical and Materials Transactions A, 26(12):3277–3293, 1995. (Cited on page 15.)[55] M. Shome and O. N. Mohanty. Continuous cooling transformation diagrams applicable tothe heat-affected zone of HSLA-80 and HSLA-100 steels. Metallurgical and MaterialsTransactions A, 37(7):2159–2169, 2006. (Cited on page 15.)[56] M. Shome, O. P. Gupta, and O. N. Mohanty. Effect of simulated thermal cycles on themicrostructure of the heat-affected zone in HSLA-80 and HSLA-100 steel plates.Metallurgical and Materials Transactions A, 35(3):985–996, 2004. (Cited on page 16.)[57] E. J. Palmiere, C. I. Garcia, and A. J. De Ardo. Compositional and microstructural changeswhich attend reheating and grain coarsening in steels containing niobium. Metallurgicaland Materials Transactions A, 25(2):277–286, 1994. (Cited on page 16.)[58] L. J. Cuddy and J. C. Raley. Austenite grain coarsening in microalloyed steels.Metallurgical Transactions A, 14(10):1989–1995, 1983. (Cited on page 16.)[59] Kook-soo Bang, Chan Park, and Stephen Liu. Effects of nitrogen content and weld coolingtime on the simulated heat-affected zone toughness in a Ti-containing steel. Journal ofMaterials Science, 41(18):5994–6000, September 2006. (Cited on page 16.)[60] Joonoh Moon, Sanghoon Kim, Hongchul Jeong, Jongbong Lee, and Changhee Lee.Influence of Nb addition on the particle coarsening and microstructure evolution in aTi-containing steel weld HAZ. Materials Science and Engineering: A, 454-455:648–653,April 2007. (Cited on page 16.)[61] Sunghak Lee, Byung Chun Kim, and Dongil Kwon. Correlation of microstructure andfracture properties in weld heat-affected zones of thermomechanically controlled processsteels. Metallurgical transactions. A, Physical metallurgy and materials science, 23A(10):2803–2816, 1992. (Cited on pages 16, 30, and 95.)[62] C. L. Davis and J. E. King. Cleavage initiation in the intercritically reheated coarse-grainedheat-affected zone: Part I. Fractographic evidence. Metallurgical and materials transactionsA, 25(3):563–573, 1994. (Cited on pages 16, 29, 30, and 95.)130BIBLIOGRAPHY[63] Y. Li, D. N. Crowther, M. J. W. Green, P. S. Mitchell, and T. N. Baker. The effect ofvanadium and niobium on the properties and microstructure of the intercritically reheatedcoarse grained heat affected zone in low carbon microalloyed steels. ISIJ international,41(1):46–55, 2001. (Cited on pages 16, 29, and 95.)[64] E. Bonnevie, G. Ferrire, A. Ikhlef, D. Kaplan, and J.M. Orain. Morphological aspects ofmartensiteaustenite constituents in intercritical and coarse grain heat affected zones ofstructural steels. Materials Science and Engineering: A, 385(1-2):352–358, November2004. (Cited on pages 16, 29, and 95.)[65] Emin Bayraktar and Dominique Kaplan. Mechanical and metallurgical investigation ofmartensiteaustenite constituents in simulated welding conditions. Journal of MaterialsProcessing Technology, 153-154:87–92, November 2004. (Cited on pages 16 and 29.)[66] J. A. Gianetto, J. E. M. Braid, J. T. Bowker, and W. R. Tyson. Heat-affected zone toughnessof a TMCP steel designed for low-temperature applications. Journal of Offshore Mechanicsand Arctic Engineering, 119(2):134–144, May 1997. (Cited on pages 16, 17, and 18.)[67] D. P. Fairchild, D. G. Howden, and W. A. T. Clark. The mechanism of brittle fracture in amicroalloyed steel: Part i. inclusion-induced cleavage. Metallurgical and MaterialsTransactions A, 31(3):641–652, 2000. (Cited on pages 16, 17, 18, and 30.)[68] S. Moeinifar, A.H. Kokabi, and H.R. Madaah Hosseini. Influence of peak temperatureduring simulation and real thermal cycles on microstructure and fracture properties of thereheated zones. Materials & Design, 31(6):2948–2955, June 2010. (Cited on pages 16, 17,18, 29, and 95.)[69] E28 Committee. Test methods for notched bar impact testing of metallic materials.Technical report, ASTM International, 2012. (Cited on page 18.)[70] E08 Committee. Test method for linear-elastic plane-strain fracture toughness KIC ofmetallic materials. Technical report, ASTM International, 2012. (Cited on pages 18 and 19.)[71] E08 Committee. Test method for drop-weight tear tests of ferritic steels. Technical report,ASTM International, 2008. (Cited on page 18.)[72] S. Xu, R. Bouchard, and W.R. Tyson. Simplified single-specimen method for evaluatingCTOA. Engineering Fracture Mechanics, 74(15):2459–2464, October 2007. (Cited onpage 19.)[73] Ph.P. Darcis, C.N. McCowan, H. Windhoff, J.D. McColskey, and T.A. Siewert. Crack tipopening angle optical measurement methods in five pipeline steels. Engineering FractureMechanics, 75(8):2453–2468, May 2008. (Cited on page 19.)[74] Hyo Kyung Sung, Seok Su Sohn, Sang Yong Shin, Sunghak Lee, Nack J. Kim, Seung HwanChon, and Jang Yong Yoo. Effects of finish rolling temperature on inverse fractureoccurring during drop weight tear test of API X80 pipeline steels. Materials Science andEngineering: A, 541:181–189, April 2012. (Cited on page 19.)[75] Robert Eiber. Drop-weight tear test application to natural gas pipeline fracture control.Journal of Pipeline Engineering, 12(3), 2013. (Cited on page 19.)131BIBLIOGRAPHY[76] C. Hari Manoj Simha, Su Xu, and W.R. Tyson. Non-local phenomenologicaldamage-mechanics-based modeling of the drop-weight tear test. Engineering FractureMechanics, 118:66–82, March 2014. (Cited on page 19.)[77] Richard W. Hertzberg. Deformation and fracture mechanics of engineering materials. JohnWiley & Sons, Inc, Hoboken, NJ, 5th edition, 2013. (Cited on pages 19, 24, and 28.)[78] E08 Committee. Test method for crack-tip opening displacement (CTOD) fracturetoughness measurement. Technical report, ASTM International, 2008. (Cited on page 19.)[79] S. Xu, N. Petri, and W.R. Tyson. Evaluation of CTOA from load vs. load-line displacementfor c(t) specimen. Engineering Fracture Mechanics, 76(13):2126–2134, September 2009.(Cited on page 19.)[80] Xian-Kui Zhu and James A. Joyce. Review of fracture toughness (G, K, J, CTOD, CTOA)testing and standardization. Engineering Fracture Mechanics, 85:1–46, May 2012. (Citedon page 19.)[81] N.A. Kahn and E.A. Imbembo. Method of evaluating transition from shear to cleavagefailure in ship plate and its correlation with large-scale plate tests. Welding Journal,27(4):169–182, 1948. (Cited on page 20.)[82] N.A. Kahn and E.A. Imbembo. Notch sensitivity of steel evaluated by tear test. WeldingJournal, 28(4):153–165, 1949. (Cited on page 20.)[83] N.A. Kahn and E.A. Imbembo. Further study of navy tear test. Welding Journal,29(2):84–96, 1950. (Cited on page 20.)[84] J.G. Kaufman and A.H. Knoll. Kahn-type tear tests and crack toughness of aluminum alloysheet. Materials Research and Standards, 4(4):151–156, 1964. (Cited on page 21.)[85] B07 Committee. Test method for tear testing of aluminum alloy products. Technical report,ASTM International, 2013. (Cited on page 21.)[86] I Kirman. The relation between microstructure and toughness in 7075 aluminum alloy.Metallurgical Transactions, 2(7):1761–1770, 1971. (Cited on page 21.)[87] Richard R. Senz and Edwin H. Spuhler. Fracture mechanics’ impact on specifications andsupply. Metal Progress, 107(3):64–66, 1975. (Cited on page 21.)[88] D. Dumont, A. Deschamps, and Y. Bre´chet. On the relationship between microstructure,strength and toughness in AA7050 aluminum alloy. Materials Science and Engineering: A,356(1-2):326–336, September 2003. (Cited on page 21.)[89] D. Dumont, A. Deschamps, and Y. Bre´chet. A model for predicting fracture mode andtoughness in 7000 series aluminium alloys. Acta Materialia, 52(9):2529–2540, May 2004.(Cited on page 21.)[90] D. Dumont, A. Deschamps, Y. Bre´chet, C. Sigli, and J. C. Ehrstrm. Characterisation ofprecipitation microstructures in aluminium alloys 7040 and 7050 and their relationship tomechanical behaviour. Materials Science and Technology, 20(5):567–576, May 2004.(Cited on page 21.)132BIBLIOGRAPHY[91] F Bron, J Besson, and A Pineau. Ductile rupture in thin sheets of two grades of 2024aluminum alloy. Materials Science and Engineering: A, 380(1-2):356–364, August 2004.(Cited on page 22.)[92] F. Bron and J. Besson. Simulation of the ductile tearing for two grades of 2024 aluminumalloy thin sheets. Engineering Fracture Mechanics, 73(11):1531–1552, July 2006. (Cited onpage 22.)[93] A. AsserinLebert, J. Besson, and A.F. Gourgues. Fracture of 6056 aluminum sheetmaterials: effect of specimen thickness and hardening behavior on strain localization andtoughness. Materials Science and Engineering: A, 395(1-2):186–194, March 2005. (Citedon page 22.)[94] A. Pirondi and D. Fersini. Simulation of ductile crack growth in thin panels using the cracktip opening angle. Engineering Fracture Mechanics, 76(1):88–100, January 2009. (Cited onpage 22.)[95] Mohan S. Devgun, Antonio J. Carrillo, and Tomas A. Roosen. Mechanical properties andtensile fracture characteristics of partially annealed low carbon mild steel sheet. Materialsscience and engineering, 59(1):49–57, 1983. (Cited on page 22.)[96] Liang Ying, Ying Chang, Ping Hu, Guo Zhe Shen, Li Zhong Liu, and Xiao Dong Li.Influence of low tempering temperature on fracture toughness of ultra high strength boronsteel for hot forming. Advanced Materials Research, 146-147:160–165, October 2010.(Cited on page 22.)[97] D. V. Edmonds and R. C. Cochrane. Structure-property relationships in bainitic steels.Metallurgical transactions a, 21(6):1527–1540, 1990. (Cited on page 23.)[98] C.H. Young and H.K.D.H. Bhadeshaa. Strength of mixtures of bainite and martensite.Materials Science and Technology, 10(3):209–214, 1994. (Cited on page 23.)[99] J. Y. Koo, M. J. Luton, N. V. Bangaru, R. Petkovic, D. P. Fairchild, C. W. Petersen,H. Asahi, T. Hara, Y. Terada, and M. Sugiyama. Metallurgical design of ultra high-strengthsteels for gas pipelines. International Journal of Offshore and Polar Engineering,14(1):2–10, 2004. (Cited on pages 23 and 28.)[100] J Bouquerel, K Verbeken, and B De Cooman. Microstructure-based model for the staticmechanical behaviour of multiphase steels. Acta Materialia, 54(6):1443–1456, April 2006.(Cited on pages 23 and 27.)[101] M. Delince´, Y. Bre´chet, J.D. Embury, M.G.D. Geers, P.J. Jacques, and T. Pardoen.Structure-property optimization of ultrafine-grained dual-phase steels using amicrostructure-based strain hardening model. Acta Materialia, 55(7):2337–2350, April2007. (Cited on page 23.)[102] I.A. Yakubtsov, P. Poruks, and J.D. Boyd. Microstructure and mechanical properties ofbainitic low carbon high strength plate steels. Materials Science and Engineering: A,480(1-2):109–116, May 2008. (Cited on page 23.)[103] Derek Hull and D. J. Bacon. Introduction to Dislocations. Elsevier, February 2011. (Citedon page 24.)133BIBLIOGRAPHY[104] F.B. Pickering and T. Gladman. Investigation into some factors which control strength ofcarbon steels. Technical report, Iron and Steel Institute, 1963. (Cited on page 24.)[105] R. L Fleisgher. Solution hardening. Acta Metallurgica, 9(11):996–1000, November 1961.(Cited on page 25.)[106] R. L Fleischer. Substitutional solution hardening. Acta Metallurgica, 11(3):203–209, March1963. (Cited on page 25.)[107] R. Labusch. A statistical theory of solid solution hardening. physica status solidi (b),41(2):659–669, January 1970. (Cited on page 25.)[108] T. Gladman. Precipitation hardening in metals. Materials Science and Technology,15(1):30–36, 1999. (Cited on page 25.)[109] Harold J. Frost and M. F. Ashby. Deformation-Mechanism Maps: The Plasticity and Creepof Metals and Ceramics. Pergamon Press, Oxford Oxfordshire ; New York, 1st edition,October 1982. (Cited on page 25.)[110] G. I. Taylor. The mechanism of plastic deformation of crystals. part I. theoretical.Proceedings of the Royal Society of London. Series A, 145(855):362–387, July 1934. (Citedon page 26.)[111] G. I. Taylor. The mechanism of plastic deformation of crystals. part II. comparison withobservations. Proceedings of the Royal Society of London. Series A, 145(855):388–404,July 1934. (Cited on page 26.)[112] H. Mecking and U. F. Kocks. Kinetics of flow and strain-hardening. Acta Metallurgica,29(11):1865–1875, November 1981. (Cited on pages 26 and 98.)[113] Marc A Meyers and Krishan K Chawla. Mechanical Behavior of Materials. CambridgeUniversity Press, 2009. (Cited on pages 26 and 27.)[114] U. F. Kocks and H. Mecking. Physics and phenomenology of strain hardening: the FCCcase. Progress in Materials Science, 48(3):171–273, 2003. (Cited on pages 26 and 98.)[115] David J Horsley. Background to the use of CTOA for prediction of dynamic ductile fracturearrest in pipelines. Engineering Fracture Mechanics, 70(34):547–552, February 2003.(Cited on page 27.)[116] P. F. Thomason. Ductile Fracture of Metals. Pergamon Press, 1990. (Cited on pages 27and 28.)[117] A. Pineau. Modeling ductile to brittle fracture transition in steelsmicromechanical andphysical challenges. International Journal of Fracture, 150(1-2):129–156, June 2008.(Cited on page 27.)[118] A. Pineau. Development of the local approach to fracture over the past 25 years: Theory andapplications. International Journal of Fracture, 138(1-4):139–166, March 2006. (Cited onpage 28.)[119] P. Spa¨tig, G. R. Odette, E. Donahue, and G. E. Lucas. Constitutive behavior and fracturetoughness properties of the F82H ferritic/martensitic steel. Journal of nuclear materials,283:721–726, 2000. (Cited on page 29.)134BIBLIOGRAPHY[120] Jung-Suk Lee, Jang-Bog Ju, Jae-il Jang, Woo-Sik Kim, and Dongil Kwon. Weld crackassessments in API X65 pipeline: Failure assessment diagrams with variations inrepresentative mechanical properties. Materials Science and Engineering A,373(1-2):122–130, 2004. (Cited on page 29.)[121] K. L. Murty, P. Q. Miraglia, M. D. Mathew, V. N. Shah, and F. M. Haggag. Characterizationof gradients in mechanical properties of SA-533B steel welds using ball indentation.International journal of pressure vessels and piping, 76(6):361–369, 1999. (Cited onpage 29.)[122] Hakan Aydin and Tracy W. Nelson. Microstructure and mechanical properties of hard zonein friction stir welded X80 pipeline steel relative to different heat input. Materials Scienceand Engineering: A, 586:313–322, December 2013. (Cited on page 29.)[123] X.P. Zhang and L. Dorn. Estimation of the local mechanical properties of pipeline steels andwelded joints by use of the microshear test method. International Journal of PressureVessels and Piping, 75(1):37–42, 1998. (Cited on page 29.)[124] G.R. Stewart, A.M. Elwazri, R. Varano, N. Pokutylowicz, S. Yue, and J.J. Jonas. Shearpunch testing of welded pipeline steel. Materials Science and Engineering: A,420(1-2):115–121, March 2006. (Cited on page 29.)[125] M. Kocak, L. Chen, G. Terlinde, G. Gnirss, and K.-H. Schwalbe. CTOD testing of HAZ andanalysis of pop-in behavior. Journal of Offshore Mechanics and Arctic Engineering,112(3):214–222, 1990. (Cited on page 29.)[126] J.H. Kim and E.P. Yoon. Notch position in the HAZ specimen of reactor pressure vesselsteel. Journal of Nuclear Materials, 257(3):303–308, 1998. (Cited on page 29.)[127] Y.C. Jang, J.K. Hong, J.H. Park, D.W. Kim, and Y. Lee. Effects of notch position of thecharpy impact specimen on the failure behavior in heat affected zone. Journal of MaterialsProcessing Technology, 201(1-3):419–424, 2008. (Cited on page 29.)[128] Kenji Ohya, Jongseop Kim, Ken’ichi Yokoyama, and Michihiko Nagumo. Microstructuresrelevant to brittle fracture initiation at the heat-affected zone of weldment of a low carbonsteel. Metallurgical and Materials Transactions A: Physical Metallurgy and MaterialsScience, 27 A(9):2574–2582, 1996. (Cited on pages 29 and 30.)[129] Koji Homma, Chitoshi Miki, and Hang Yang. Fracture toughness of cold worked andsimulated heat affected structural steel. Engineering Fracture Mechanics, 59(1):17–28,1998. (Cited on page 29.)[130] H. Qiu, H. Mori, M. Enoki, and T. Kishi. Fracture mechanism and toughness of the weldingheat-affected zone in structural steel under static and dynamic loading. Metallurgical andMaterials Transactions A, 31(11):2785–2791, 2000. (Cited on page 29.)[131] Sangho Kim, Young-Roc Im, Sunghak Lee, Hu-Chul Lee, Sung-Joon Kim, and Jun HwaHong. Effects of alloying elements on fracture toughness in the transition temperatureregion of base metals and simulated heat-affected zones of mn-mo-ni low-alloy steels.Metallurgical and Materials Transactions A, 35(7):2027–2037, 2004. (Cited on page 29.)135BIBLIOGRAPHY[132] Yaowu Shi and Zhunxiang Han. Effect of weld thermal cycle on microstructure and fracturetoughness of simulated heat-affected zone for a 800MPa grade high strength low alloy steel.Journal of Materials Processing Technology, 207(1-3):30–39, October 2008. (Cited onpages 29 and 30.)[133] Fukuhisa Matsuda, Kenji Ikeuchi, Hitoshi Okada, Ivan Hrivnak, and Hwa-Soon Park. Effectof m-a constituent on fracture behavior of 780 and 980MPa class HSLA steels subjected toweld HAZ thermal cycles. Transactions of JWRI, 23(2):231–238, December 1994. (Citedon page 29.)[134] Astrid Lambert-Perlade, Anne-Franoise Gourgues, Jacques Besson, T. Sturel, and AndrPineau. Mechanisms and modeling of cleavage fracture in simulated heat-affected zonemicrostructures of a high-strength low alloy steel. Metallurgical and Materials TransactionsA, 35(13):1039–1053, 2004. (Cited on page 29.)[135] Amer E. Amer, Min Y. Koo, Kyong H. Lee, Sang H. Kim, and Soon H. Hong. Effect ofwelding heat input on microstructure and mechanical properties of simulated HAZ in aCucontaining microalloyed steel. Journal of Materials Science, 45(5):1248–1254, March2010. (Cited on page 29.)[136] Y. Li and T. N. Baker. Effect of morphology of martensiteaustenite phase on fracture ofweld heat affected zone in vanadium and niobium microalloyed steels. Materials Scienceand Technology, 26(9):1029–1040, September 2010. (Cited on pages 29 and 30.)[137] Yang You, Chengjia Shang, Nie Wenjin, and Sundaresa Subramanian. Investigation on themicrostructure and toughness of coarse grained heat affected zone in X-100 multi-phasepipeline steel with high Nb content. Materials Science and Engineering: A, 558:692–701,December 2012. (Cited on page 29.)[138] Peyman Mohseni. Brittle and Ductile Fracture of X80 Arctic Steel. PhD thesis, NorwegianUniversity of Science and Technology, 2012. (Cited on page 30.)[139] Graeme Robertson Goodall. Welding High Strength Modern Line Pipe Steel. PhD thesis,McGill University, 2011. (Cited on page 31.)[140] Wengui Zhao, Wei Wang, Shaohui Chen, and Jinbo Qu. Effect of simulated welding thermalcycle on microstructure and mechanical properties of X90 pipeline steel. Materials Scienceand Engineering: A, 528(24):7417–7422, September 2011. (Cited on pages 31 and 32.)[141] Gui Ying Qiao, Ming Zheng, Yu Gu, Xiu Lin Han, Xu Wang, Fu Ren Xiao, and Bo Liao.Mechanical properties of the heat affected zone of an x100 grade pipeline steel. MaterialsScience Forum, 773-774:795–802, November 2013. (Cited on pages 31 and 32.)[142] Reza Tafteh. Personal communication. (Cited on page 35.)[143] Kumkum Banerjee, Matthias Militzer, Michel Perez, and Xiang Wang. Nonisothermalaustenite grain growth kinetics in a microalloyed X80 linepipe steel. Metallurgical andMaterials Transactions A, 41(12):3161–3172, December 2010. (Cited on pages 35, 53,and 112.)[144] Jenny Reichert. Personal communication, 2014. (Cited on pages 37 and 41.)136BIBLIOGRAPHY[145] Reza Tafteh. Austenite decomposition in an X80 linepipe steel. PhD thesis, University ofBritish Columbia, 2011. (Cited on page 41.)[146] Quentin Puydt. Personal communication, 2013. (Cited on page 44.)[147] Jennifer M. Reichert, Thomas Garcin, Matthias Militzer, and Warren J. Poole. Formation ofMartensite/Austenite (M/A) in X80 Linepipe Steel. In 2012 9th International PipelineConference, pages 483–489. American Society of Mechanical Engineers, September 2012.(Cited on pages 52, 105, and 113.)[148] M. Maalekian, R. Radis, M. Militzer, A. Moreau, and W. J. Poole. In situ measurement andmodelling of austenite grain growth in a Ti/Nb microalloyed steel. Acta Materialia,60(3):1015–1026, February 2012. (Cited on page 95.)[149] H Mecking, UF Kocks, and H Fischer. Hardening, recovery, and creep in FCC mono-andpolycrystals. In Presented at the 4th Intern. Conf. on Strength of Metals and Alloys, Nancy,30 Aug.-3 Sep. 1976, volume 1, pages 334–339, 1976. (Cited on page 98.)[150] L. M. Cheng, W. J. Poole, J. D. Embury, and D. J. Lloyd. The influence of precipitation onthe work-hardening behavior of the aluminum alloys AA6111 and AA7030. Metallurgicaland Materials Transactions A, 34(11):2473–2481, November 2003. (Cited on page 99.)[151] Javier Gil Sevillano. Flow stress and work hardening. In RW Cahn, P Hassen, andEJ Kramer, editors, Materials Science and Technology. A Comprehensive Treatment, vol. 6,VCH, pages 19–88. Weinheim, 1993. (Cited on page 102.)[152] O. Bouaziz. Revisited storage and dynamic recovery of dislocation density evolution law:Toward a generalized KocksMecking model of strain-hardening. Advanced EngineeringMaterials, 14(9):759–761, 2012. (Cited on page 104.)[153] Jennifer M. Reichert, Matthias Militzer, Warren J. Poole, and Laurie Collins. A newapproach using EBSD to quantitatively distinguish complex transformation products alongthe HAZ in X80 linepipe steel. In 2014 10th International Pipeline Conference. AmericanSociety of Mechanical Engineers, 2014. (Cited on page 111.)137


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items