DNV Platform of Computational Welding Mechanics

Transcription

1 DNV REPORT No.: IIW Document Number: X Delegation of Norway and Sweden Commission X DNV Platform of Computational Welding Mechanics ABSTRACT Per R. M. Lindström DNV Materials Laboratory, Det Norske Veritas AS, Høvik, Norway / Department of Engineering Science, University West, Trollhättan, Sweden This document presents the DNV Platform of Computational Welding Mechanics, CWM, with its associated CWM-methodology. That has been developed, validated and implemented as a part of DNV s Technology Leadership program in the field of Structural Integrity and Materials Technology. A successful CWM implementation requires that the actual organisation has gained the knowledge and understanding of the following related topics: - Welding Engineering with an emphasis on the welding process and its thermodynamics - Weld process quality control such as calibration, validation as well as DAQ, (Data Acquisition) - Transient thermo-mechanical coupled FE-analyses and constitutive modelling - Computational platforms comprising the selection of hardware, operative system and FEM-code as well as suitable pre- and post-processing tools From that perspective there is a lack of reliable and/or hands-on oriented CWM Engineering Handbooks and best recommended practices available on the market. For that sake is the DNV CWMmethodology and its hands on solutions presented. The CWM-methodology described can not only be used for residual stress assessments, as presented in this report. It can also be used for various applications such as assessment of used and/or proposed WPS, Welding Procedure Specifications as well as optimisation of the manufacturing and production process of integrated metallic structures. From the results of a parametric CWM-study have three (3) factors been identified to drive and/or contribute to the magnitude of the weld residual stresses in ship steel plate materials. The contributing and/or driving factors identified are the: - Thermal- and Mechanical Boundary Conditions during the production welding - Yield stress difference between the base- and the weld filler material - Weld heat input, Q, which affects the weld cooling time IIW Keywords: Weld Residual Stresses, CWM, Computational Welding Mechanics, Weld Simulations, Kinematic Hardening, WPS Assessment, Distortion IIW 66 th Annual Assembly 2013

2 Table of Contents Page 1 INTRODUCTION DNV CWM PLATFORM Introduction HPC Linux Workstation FEM-Solver at CWM-Simulations Pre- and Post-Processing CWM Analysts DNV CWM MATERIAL MODEL Use of the CWM-material models Thermal CWM-material model Mechanical CWM-material model Residual Stress Releasing Function Residual Stress Release Temperatures Hardening Formulation Weld filler material activation DNV CWM METHODOLOGY Introduction Scrutinizing WPQR and WPS Test Welding and DAQ Weld Heat Calculations D Transient FE-Weld Simulations D Transient FE-Weld Simulations D Transient Thermo-Mechanical FEA Elastic Shakedown Analyses CWM-STUDY OF RESIDUAL STRESS CONTRIBUTING FACTORS Introduction Assessment of the WPS Weld Joint Geometry and CWM-Models Weld Heat Input Thermal and Mechanical Boundary Conditions Elastic Shakedown Analyses Description of the CWM Analyses Carried Out Material properties and modelling Weld Residual Stress Results - Elastic Shakedown Weld Residual Stress Results Alternative 1 and Q min Weld Residual Stress Results Alternative 1 and Q nom Weld Residual Stress Results Alternative 1 and Q max Weld Residual Stress Results Alternative 2 and Q min Weld Residual Stress Results Alternative 2 and Q nom Weld Residual Stress Results Alternative 2 and Q max Weld Residual Stress Results Alternative 3 and Q min Weld Residual Stress Results Alternative 3 and Q nom Weld Residual Stress Results Alternative 3 and Q max DISCUSSION Thermo- and Mechanical Boundary Conditions Weld Heat Input Yield Stress Difference IIW 66 th Annual Assembly 2013 i

3 7 CONLUSIONS REFERENCES APPENDIX A Material Certificate NV Grade EH32 APPENDIX B Material & Physical Data IIW 66 th Annual Assembly 2013 ii

4 IIW Document Number: X INTRODUCTION This document presents the DNV Platform of Computational Welding Mechanics, CWM, with its associated CWM-methodology. That has been developed, validated and implemented as a part of DNV s Technology Leadership program in the field of Structural Integrity and Materials Technology. In a scientific context FE-based, thermo-mechanical simulations to determine the temperature and stress fields present during welding have been performed for about forty years. The FEtechnology and the computer capacity (computations and storage) have developed to a state that Computational Welding Mechanics, CWM, is an established and mature process. /1/ /2/ It implies that CWM can be used as a fairly reliable tool at the assessment of Welding Procedure Specifications, WPS, and proposed weld process parameters. /3/ /4/ As well as parametric studies for the sake of optimised weld residual stress fields at the structural integrity design phase. /5/ /6/ /7/ A number of FEM-weld simulation reports with experimental verification /8/ /9/ /10/ /11/ and one (1) residual stress weld experiment /12/ has recently been presented. Areas where improvement may still be needed are: - Data Acquisitioning, DAQ, of weld process parameters used and Quality Control, QC, of the thermal- and mechanical boundary conditions affecting the actual weld test coupons to be used for the purpose of CWM-calibration, -verification and/or -validation. - Accurate modelling of the actual weld process parameters used or to be used - Accurate modelling of the actual thermal- and mechanical boundary conditions used or to be used - Material modelling in particular at high temperatures and at solid state phase transformations - Geometric modelling and mesh density optimisation of complex weld joint configurations - Modelling and/or definition of the transient thermal- and dynamic mechanical contact problems associated with welding processes A successful industrial CWM implementation requires that the actual organisation has gained the knowledge and understanding of the following related topics: - Computational platforms that comprises the selection of hardware, operative system and FEM-code as well as suitable pre- and post-processing tools /13/ - Welding Engineering with an emphasis on the weld process parameters and its thermodynamics /14/ - Quality Control, QC, of weld process related activities /15/ - Transient thermo-mechanical coupled FE-analyses and constitutive modelling /16/ Anyhow there is a lack of reliable and/or hands-on oriented CWM Engineering Handbooks and best recommended practices available on the market creating a barrier that is preventing engineers to gain understanding and knowledge in the topic. This has resulted in an extensive and expensive learning curve for any individual that wants to gain the knowledge and skill of CWM-analyses. The methodology here presented has been validated against scientifically published benchmark examples presented by IIW, International Institute of Welding /17/ /18/ as well as two (2) commercial available CWM-codes. /19/ /20/ IIW 66 th Annual Assembly

5 2 DNV CWM PLATFORM IIW Document Number: X Introduction This chapter describes various practical and theoretical aspects to be considered in conjunction with CWM, Computational Welding Mechanics. 2.2 HPC Linux Workstation Before an engineering organisation starts with CWM-activities one has to select and configure a computational platform that will works in the industrial reality. DNV Materials Laboratory has selected to use a 12 Core Fujitsu Celsius R670 HPC Work Station with dual boot of the operative systems Microsoft Windows 7 and Linux CentOS-6.4 with a ThinLinc Desktop terminal server as its default CWM computational platform. /21/ /22/ /23/ 2.3 FEM-Solver at CWM-Simulations DNV Materials Laboratory has selected to use the commercial FEM-software LS-Dyna as its default non-linear FEA-code for CWM-simulations. /24/ /25/ A disadvantage with the solver selected compared to the FEA-solver ABAQUS is the lack of a comprehensive user manual. ABAQUS is for the time being the DNV in-house code for non-linear FE-simulations that had to be abounded at CWM-simulation due to lack of numerical performances. The computational time for a 3D transient thermal TIG-bead on plate simulation was found to be about 6 8 times longer by the use ABAQUS compared to LS-Dyna. /26/ Anyhow in-depth descriptions of the nonlinear finite element analysis methodologies used by the code, such as Lagrangian, Eularian and arbitrary Lagrangian Eulerian are well and extensively described in the book Nonlinear Finite Elements for Continua and Structures. /27/ As well as the theoretical foundations of the inelastic material modelling with its numerical formulation and implementation are described in the book Computational Inelasticity. /28/ Whilst the mathematically modelling of macroscopic volume elements behaviour and the physics underlying the phenomena is presented in the book Mechanics of Solid Materials. /29/ The general LS-Dyna keyword users manuals, 2 volumes, can be downloaded free of charge from the website of LSTC. /30/ LS-Dyna is by default incorporating a double ellipsoidal weld heat source, commonly denoted the Goldak heat source after its formulator. /31/ The LS-Dyna version of the Goldak double ellipsoidal weld heat source can be used for thermo-mechanical staggered coupled simulations as well as thermal analyses only. The thermo-mechanical staggered coupled approach has been found beneficial as the control of the 3D-movements is handled as a mechanical problem and subsequently solved by the mechanical solver free from influences of the thermal solver and vice versa. The general principles of heat conduction in solids and boundary layer theory as well as its applications are described in /32/ /33/ /34/. A novel feature of the LS-Dyna s Goldak heat source implementation, from a welding engineering perspective, is that it is possible to orient the weld torch or electrode in any direction and model the arc pressure weld pool surface depression. This makes it possible to include the torch angle value stated in the WPS for the production of a specific weld joint as well as the arc pressures of various weld processes, Fig. 1. The first use of LS-Dyna s Goldak weld heat source adjustable parameters was presented 2012 /16/ /3/ and has been adopted here as well. IIW 66 th Annual Assembly

6 IIW Document Number: X Fig. 1. Illustration of how the heat flux emission can be controlled by the use of LS-Dyna s Goldak weld heat source implementation, the green colour half spheres are symbolising the weld heat flux field volume. 2.4 Pre- and Post-Processing CWM-analyses should be done by the use of a mapped Quad-/Hex mesh with transition elements in order to facilitate robustness toward to excessive element deformations, Fig. 2. Fig. 2. Illustration of a mapped Hexagonal CWM-mesh with transition elements General pre- and post-processing is done by the use of LS-PrePost from LSTC in combination with a suitable open source text editor such as Notepad++, NEdit or Vim. /35/ /36/ /37/ /38/ The book LS-DYNA for Beginners is a useful step by step self-study book for the learning of LS- PrePost and LS-Dyna. /39/ The pre-processor TrueGrid is recommended to be used at parametric studies of weld joint geometry influences etc. /40/ At demanding modelling activities is the reading of the extensive amount of user manual pdfdocuments facilitated by the use of an E-reader such as Kindle or Letto. /41/ /42/ IIW 66 th Annual Assembly

7 2.5 CWM Analysts IIW Document Number: X Even if the FE-technology and the computer capacity (computations and storage) have developed to a state that Computational Welding Mechanics, CWM, is an established and mature process in a scientific context. It is far from that case in an industrial context that not is defence and/or nuclear related. Resulting in a great need of CWM Handbooks targeting practicing welding-, material-, and mechanical engineers as well as metallurgists with a M.Sc. degree or similar. As existing CWM-books more or less are focusing on the scientific- and/or advanced readers. /1/ /2/ One other obstacle that has to be overcome is the recruitment and training of CWM-analysts. During the implementation of DNV s CWM-Methodology has two (2) fairly new examined engineers with a M.Sc. mechanics degree been given basic CWM-training. Based on the experiences gained it has been concluded that the On the job training time required should be about 6 weeks (enhanced with literature home studies when necessary). Preferable combined with some sort of formal welding technology education of about 6 weeks. Such as IWS- Diploma, International Welding Specialist and/or IWSD-Diploma, International Welded Structures Designer. /43/ Anyhow the most fundamental knowledge a CWM-analyst must gain before he or she can start to do FE-weld simulations is the reading, interpretation and understanding of the most basic welding engineering documents such as WPQR and WPS. /2/ /4/ /44/ /45/ /46/ /47/ For that sake it is highly recommended that a CWM-analyst has the access to some basic welding engineering related handbooks. /48/ /49/ /50/ /51/ 3 DNV CWM MATERIAL MODEL In order to utilise the full potential of the LS-Dyna solver for the sake of CWM has DNV Materials Laboratory developed a CWM-material model package that now is implemented into the LS-Dyna solver package. /52/ /53/ The CWM-material model package is intentionally compiled for the Implicit Double Precision MPP Solver of LS-Dyna and consists of two (2) distinctively separated material models specifically: - A thermal CWM-material model /52/ - A mechanical CWM-material model /53/ 3.1 Use of the CWM-material models At simulation of a multi pass weld joint is the CWM-material model used for: - Base material - Existing and solidified weld filler material - Weld passes to be produced The material activation temperatures, liquidus- and solidus temperature, are set to very low values for the base material and existing weld passes, typical values used are: - Liquidus temperature = 10-5 C - Solidus temperature = 10-6 C These setting results in that accumulated weld residual stresses in the base- and existing weld filler material will be released if the temperature pass through the residual stress release temperature interval (lower and upper residual stress release threshold). IIW 66 th Annual Assembly

8 3.2 Thermal CWM-material model IIW Document Number: X The user-defined features of the thermal CWM-material model include: - Weld filler material activation at a temperature interval defined by the analyst - A thermal conductivity of 150 W/(m C) between the solidus and liquidus threshold temperatures of the alloy in question - A thermal conductivity of 300 W/(m C) above the liquidus threshold temperature of the alloy in question As with the Mechanical CWM-material model, the activation occurs gradually during the time takes the thermal energy of the heat source to heat up the quiet weld filler material between its solidus and liquidus temperatures. 3.3 Mechanical CWM-material model The mechanical CWM-material model includes three (3) user-defined features: - A residual stress releasing function - Isotropic and/or linear kinematic hardening - Weld filler material activation The CWM-material model is a thermo-elastic-plastic model with kinematic hardening that allows for material creation as well as residual stress release triggered by temperature and the model is, for the time being, limited to solid elements only. The material is initially in an inactivated state that is called the ghost state, in the literature sometimes referred to as a quietand/or chewing-gum material. In this state the material has the thermo-elastic properties defined by the: - Ghost Young s modulus - Ghost Poisson s ratio - Ghost thermal expansion coefficient The name Ghost material has its origin from the fact that the CWM-analyst from time to time must reminds himself and the stakeholders that numerical artefacts may heritage from the inactivated weld filler material s properties. These values represent metal empty volumes i.e. vacuum or fluids (air and liquid). In theory, shall the Young s modulus value be small enough to not influence the surroundings but large enough to avoid numerical problems. A quiet material stress should never reach the yield point. When the temperature reaches the activation (birth) temperature, a history variable representing the indicator of the welding material is incremented, this variable follows Equation. 1. Equation. 1. ( ) ( ( ( ) ) ) This parameter is available as history variable 9 in the output database. The effective thermoelastic material properties are interpolated in accordance with Equation. 2 - Equation. 4. Equation. 2. ( ) ( ) Equation. 3. ( ) ( ) IIW 66 th Annual Assembly

9 IIW Document Number: X Equation. 4. ( ) ( ) The stress update then follows a classical isotropic associative thermo-elastic-plastic approach with kinematic hardening that is summarized in the following. The explicit temperature dependence is sometimes dropped for the sake of clarity. The stress evolution is given as described in Equation. 5. Equation. 5. ( ) Where C is the effective elastic constitutive tensor and ( Equation. 6) and ( Equation. 7) are the thermal and plastic strain rates. Equation. 6. Equation. 7. Equation. 7 include the deviatoric stress, the back stress κ and the effective stress that are involved in the plastic equations, Equation. 8. Equation. 8. ( ) ( ) The effective yield stress (σ y ) is given as expressed in Equation. 9 and plastic strains evolve when the effective stress exceeds this value. Equation. 9. ( ) ( ) The back stress evolves as described in Equation. 10, where p is the rate of effective plastic strain that follows from consistency equations. Equation. 10. ( ) ( ) When the temperature reaches the start annealing temperature, the material starts assuming its virgin properties. Beyond the start annealing temperature it behaves as an ideal elastic-plastic material but with no evolution of plastic strains. The resetting of effective plastic properties in the annealing temperature interval is done by modifying the effective plastic strain and back stress before the stress update as described by Equation. 11 and Equation. 12. Equation. 11. ( ( )) Equation. 12. ( ( )) IIW 66 th Annual Assembly

10 IIW Document Number: X Residual Stress Releasing Function The residual stress releasing function, also known as annealing- and/or recrystallization function, gradually releases the back stress tensor and the effective plastic strain when the material temperature pass through the residual stress release temperature interval (lower- and upper residual stress release threshold). The residual stress release function eliminates the prior accumulated hardening history, for example when the material goes from a solid state to a liquidus state or vice versa (or if there is a specific solid state phase transformation or recrystallization interval). Here, the user-defined values at the upper residual stress release threshold temperature are: - Back stress tensor set to the value zero [ ] - Effective plastic strain set to the value zero [ ] - Stresses d plastic strains remain unchanged [ ] Residual Stress Release Temperatures The residual stress release temperature interval, lower- and upper residual stress release threshold, are defined in the following welding engineering manner by the use of an Iron Carbon Phase Diagram, Fig. 3: - Residual Stress Release Start Temperature: (Lower residual stress release threshold) TMCP Ferritic steels: Use A1 value from Iron Carbon Phase Diagram Normalized Ferritic steels: Use A3 value from Iron Carbon Phase Diagram Austenitic steels: Use 1030 C (exact temperature can be discussed/argued) - Residual Stress Release Stop Temperature: (Upper residual stress release threshold) TMCP Ferritic steels: Use A3 value from Iron Carbon Phase Diagram Normalized Ferritic steels: Use A3 value + 25 C Austenitic steels: Use 1100 C (exact temperature can be discussed/argued) IIW 66 th Annual Assembly

11 IIW Document Number: X Fig. 3. Iron Carbon Phase Diagram indicating the A1-, A3-, Solidus- and Liquidus Lines Hardening Formulation The material hardening model can employ a mixed hardening with a range from 100% isotropic hardening to 100% linear kinematic hardening Weld filler material activation The weld filler materials of all weld passes are modelled and participate in the FE-simulation from the beginning. The weld filler material has two different mechanical properties: - Ghost, also known as quiet and chewing gum, i.e. inactive, - Active Initially, all weld filler material is in the ghost state. It becomes activate as a function of temperature /54/ in particular, the activation occurs gradually during the time it takes for the thermal energy of the heat source to heat the ghost weld filler material between its solidus and liquidus temperatures. The solidus and liquidus temperatures of the actual weld filler material should be identified by means of its chemical composition in combination with the use of an appropriate phase diagram. At the moment of that the weld filler material has become fully activated will it obtain liquid weld filler material properties and gradually transform to solid weld material properties when its temperature falls below its liquidus and solidus temperature. This implies that the ghost material model contains three (3) different material characteristics for one and single material: - Ghost - Liquid - Solid It also implies that the weld filler material only can build up weld residual stresses after that it has been activated and/or the temperature has fallen down below its solidus- and residual stress releasing temperatures. IIW 66 th Annual Assembly

12 IIW Document Number: X DNV CWM METHODOLOGY 4.1 Introduction The DNV CWM-methodology is a substantially improvement of the work carried out 1998 by L. Josefson and B. Brickstad. /54/ That, for the time being, is a dominating CWM-method adopted by the Swedish Radiation Safety Authority (2009) as well as TWI (2010). /55/ /56/ /57/ /58/ A number of practical-, commercial-, coding-, academically- and technically related reasons has contributed to DNV s industrial-cwm evolution ending up to use the commercial FEM-code LS-Dyna. Nevertheless, a major reason that clearly is differentiating LS-Dyna from other commercial general non-linear implicit solvers is that LSTC, Livermore Software Technology Corporation, from the very beginning has concentrated its effort to optimise its code for rapid solving times of transient dynamic material problems incorporating thermo-dynamic effects. This has been found very beneficial at FE-weld simulations as the transient thermal load of the weld heat flux results in an intricate/complex dynamic interaction of contracting weld filler material at the same time as the base material first expands and finally contract as a function of that the weld heat flux pass by. The DNV CWM-methodology package consists of the following approaches that can be combined in various constellations to meet the specific needs and demands of the actual project: - Scrutinizing of the actual weld joint configuration including, WPQR, WPS, base- and weld filler material certificates as well as the material producers recommendation - Experimental test welding and DAQ, Data Acquisitioning, based on the data of the actual weld joint configuration s documentation - 3D thermal FE-simulation of the weld heat flux and temperature fields with activation of weld filler material - 3D thermo-mechanical staggered coupled FE-simulation of the weld heat flux, temperature fields, weld residual stresses and deformations, including activation of weld filler material - Axisymmetric thermo-mechanical coupled FE-simulation of the weld heat flux, temperature fields, weld residual stresses and deformations, including activation of weld filler material. By the use of the weld heat flux, obtained from a 3D thermal FE-simulation, as the weld heat load - 2D thermo-mechanical coupled FE-simulation of the weld heat flux, temperature fields, weld residual stresses and deformations, including activation of weld filler material. By the use of the weld heat flux, obtained from a 3D thermal FE-simulation, as the weld heat load IIW 66 th Annual Assembly

13 IIW Document Number: X Scrutinizing WPQR and WPS Before the modelling and CWM-simulations begins shall the actual weld joint configuration including, WPQR, WPS, base- and weld filler material certificates as well as the material producers recommendations be reviewed and scrutinized. As the current conceptual idea of CWM is that the FE-weld simulations shall be based on weld test experimental data such those stated in a WPQR or WPS. /44/ /45/ /46/ /47/ 4.3 Test Welding and DAQ If required, experimental test welding and DAQ, Data Acquisitioning, shall be performed by weld process oriented Welding Engineers in accordance with a quality level that at least are in accordance with the most stringent requirements of the applicable WPS- and WPQR standard NORSOK M Weld Heat Calculations A fundamental part of CWM-analyses is the calculation and scrutinizing of the weld heat input, Q w : - Used at the WPT, Welding Procedure Test - Stated in the WPQR, Welding Procedure Qualification Record - Stated on the WPS, Welding Procedure Specification - Used or intended to be used in the reality From a welding engineering perspective is the weld heat input, Q w, the amount of energy used to produce a specific weld pass. The unit is kj/mm and Q W is calculated by Equation. 13. /60/ The back ground to the equation and the weld process thermal efficiency values has been described by L-E Svensson. /61/ Equation. 13. Nomenclature Equation 13: Q w = Weld Heat Input [kj/mm] η = Weld process thermal efficiency U = Arc Voltage [V] I = Current [A] v = Welding speed [mm/s] IIW 66 th Annual Assembly

14 IIW Document Number: X In the case of CWM-analysis should the weld heat input, Q w, of each weld pass be calculated as weld heat power, P w, with the unit W, calculated by Equation. 14. Equation. 14. Nomenclature Equation 14: P w = Weld Heat Power [W] η = Weld process thermal efficiency U = Arc Voltage [V] I = Current [A] D Transient FE-Weld Simulations A Goldak double ellipsoidal weld heat source model is used at 3D transient, thermal and thermomechanical staggered coupled, FE-weld simulations. The weld heat power, P w, is applied in the form of a 3D weld heat flux density, Q Φ3D, with the unit W/m 3, Fig. 4. /44/ Fig. 4. Illustration of LS-Dyna s Goldak double ellipsoidal weld heat source implementation The LS-Dyna implementation of the Goldak double ellipsoidal weld heat source model is described by Equation. 15 Equation. 18. /62/ /62/ Equation. 15. Equation. 16. ( ) Equation. 17. ( ) Equation. 18. Nomenclature Equation P w = Weld Heat Power [W] Q Φ3D = 3D Weld Heat Flux Density [W/m 3 ] = 3D Forward Weld Heat Flux Density [W/m 3 ] = 3D Aftward Weld Heat Flux Density [W/m 3 ] f f = Forward end weld heat power deposition factor, ( 0 < F f < 2) IIW 66 th Annual Assembly

15 IIW Document Number: X f a = Aftward end weld heat power deposition factor, ( 0 < F a < 2) a, b, c 1, c 2 = Radiuses describing the double ellipsoidal shape [m] x, y, z = Distances (Cartesian coordinates) with the arc emitting node as Origo [m] v = Weld head travel speed [m/s] t = Time [s] D Transient FE-Weld Simulations At 2D/Axisymmetric FE-weld simulations the weld heat power, P w, is applied in the form of an Equivalent 2D weld heat flux density, Q Φ2D, with the unit W/m 3. In the 2D/Axisymmetric CWM-methodology formulated by B. Brickstad and L. Josefson (1998) is P w converted to Q Φ2D by Equation Equation. 20. /54/ Equation. 19. Equation. 20. ( ) Nomenclature Equation P w = Weld Heat Power [W] Q Φ2D = Equivalent 2D Weld Heat Flux Density [W/m 3 ] V 2D = 2D Weld Pass Volume [m 3 ] A w = Weld Pass Crosse Section Area [m 2 ] v = Welding speed [m/s] t 1 = Time of the weld melt pool s forward edge to pass by the weld joint s cross section plane [s] t 2 = Time of the weld melt pool s aftward edge pass by the weld joint s cross section plane [s] The application of Equation Equation. 20 implies the simplified assumption of that the 2D weld heat flux density, Q Φ2D, has a uniform amplitude during the time it takes for the weld head and its associated weld melt pool to pass by the weld joint s cross section plane, Fig. 5. It should also be mentioned that it is fairly tricky to calculate correct time values of t 0 and t 1. Fig. 5. Equivalent 2D weld heat flux density load curve (left); weld joint s cross section plane (right) IIW 66 th Annual Assembly

16 IIW Document Number: X In the reality is this not the case as the weld heat flux, Q Φ, in the way of the cross section plane is a function of numerous and various weld melt pool related parameters. By the use of an improved 2D/Axisymmetric weld heat flux density modelling method, formulated by P. Lindström and L. Josefson (2012), better (more accurate) weld residual stress results are obtained. /2/ The improved 2D/Axisymmetric weld heat flux density modelling methodology is as follow: Step 1 By the use of 3D transient thermal FE-weld simulation is the sum of the resultant heat flux, Q Φ, of all nodes in the cross section plane of the actual weld pass analysed, Fig. 6. The unit of the Q Φ value is W/m 3. Fig. 6. Illustration of Q Φ -curve in the plane of a root pass cross section plane Step 2 The 3D transient thermal FE-weld simulation gives fairly accurate Q Φ -values for the time period when the weld melt pool pass over the cross section plane, Fig. 7. Fig. 7. Q Φ -values over time in the plane of a root pass cross section Step 3 The Q Φ -curve is used to calculate the Equivalent 2D weld heat flux density, Q Φ2D,. This is done IIW 66 th Annual Assembly

17 IIW Document Number: X by collecting and compiling the values from Fig. 7 into an Equivalent 2D Weld Heat Flux Density Load Curve Table, Table 1. Table 1 Equivalent 2D Weld Heat Flux Density Load Curve Time [s] Q Φ [W/m 3 ] 0 2,68E+07 0,7 5,65E+08 2,2 1,33E+10 3,9 2,75E+09 4,4 2,77E+08 5,1 2,28E+08 5,8 5,49E+07 f 2D [ - ] Q Φ2D [W/m 3 ] IIW 66 th Annual Assembly

18 Step 4 A 2D weld heat flux density scale factor, f 2D, is calculated by Equation. 21 IIW Document Number: X Equation. 21. Nomenclature Equation 21 f 2D = 2D weld heat flux density scale factor Q = Total weld heat energy used to produce the entire weld pass [J] = Integer of the resultant heat flux sum [J] t 0 = Time when total resultant heat flux start to affect the cross section plane [s] t 1 = Time when total resultant heat flux stop to affect the cross section plane [s] The sum of the weld heat energy for the entire weld pass, Q, is calculated by either Equation. 22 or Equation. 23. Equation. 22. Equation. 23. ( ) Nomenclature Equation Q = Total weld heat energy used to produce the entire weld pass [J] Q w = Weld heat input [kj/mm] P w = Weld heat power [W] lw = Weld pass length [m] t 0 = Start time of welding the pass [s] t 1 = Stop time of welding the pass [s] IIW 66 th Annual Assembly

19 IIW Document Number: X Whilst the integer of the resultant heat flux sum,, conveniently can be calculated by the use of LS-PrePost, Fig. 8. /27/ Fig. 8. Integer of the resultant heat flux sum,, in the plane of a root pass s cross section Step 5 The equivalent 2D weld heat flux density load curve table is updated with the f 2D -value, calculated by the use of Equation. 24, Table 2. Equation. 24. ( ) Table 2 Equivalent 2D Weld Heat Flux Density Load Curve Time [s] Q Φ [W/m 3 ] f 2D [ - ] Q Φ2D [W/m 3 ] 0 2,68E+07 2,17E-04 0,7 5,65E+08 2,17E-04 2,2 1,33E+10 2,17E-04 3,9 2,75E+09 2,17E-04 4,4 2,77E+08 2,17E-04 5,1 2,28E+08 2,17E-04 5,8 5,49E+07 2,17E-04 IIW 66 th Annual Assembly

20 IIW Document Number: X Step 5 The equivalent 2D weld heat flux density load curve table can now completed. This is done by calculation of Q Φ2D, Table 3. Table 3 Equivalent 2D Weld Heat Flux Density Load Curve Time [s] Q Φ [W/m 3 ] f 2D [ - ] Q Φ2D [W/m 3 ] 0 2,68E+07 2,17E-04 7,28E+03 0,7 5,65E+08 2,17E-04 1,53E+05 2,2 1,33E+10 2,17E-04 3,61E+06 3,9 2,75E+09 2,17E-04 7,47E+05 4,4 2,77E+08 2,17E-04 7,51E+04 5,1 2,28E+08 2,17E-04 6,19E+04 5,8 5,49E+07 2,17E-04 1,49E+04 Step 6 Finally is the equivalent 2D weld heat flux density load curve updated and adjusted with start and stop values in such way that it will fit its intended purposes. An example of that is given in Table 4. Table 4 Equivalent 2D Weld Heat Flux Density Load Curve Time [s] Q Φ [W/m 3 ] f QΦ2D [ - ] Q Φ2D [W/m 3 ] , , ,68E+07 2,17E-04 7,28E+03 1,7 5,65E+08 2,17E-04 1,53E+05 3,2 1,33E+10 2,17E-04 3,61E+06 4,9 2,75E+09 2,17E-04 7,47E+05 5,4 2,77E+08 2,17E-04 7,51E+04 6,1 2,28E+08 2,17E-04 6,19E+04 6,8 5,49E+07 2,17E-04 1,49E+04 6, IIW 66 th Annual Assembly

21 IIW Document Number: X A comparison of the two (2) Equivalent 2D weld heat flux density calculation methods, the original- versus the method adopted by DNV, demonstrates that the new Q Φ2D -method more accurately describes the rapid thermal transient associated with arc-welding processes. Thereby facilitating a better simulation of the weld joint s dynamic expansion and contraction as a function of that the weld melt pool passes by. Fig. 9. Fig. 9. Comparison of the Original Q Φ2D -method (left) versus the DNV Q Φ2D -method (right) IIW 66 th Annual Assembly

22 IIW Document Number: X D Transient Thermo-Mechanical FEA At 2D transient thermo-mechanical simulation of arc welding processes is DNV utilising a 2D/3D Hybrid element formulation, see Fig. 10. This formulation results in a 2D Generalized Plain Strain performance that is in accordance with the generalized plane strain theory used by the commercial FEA-code Abaqus. /63/ The theory assumes that the model lies between two bounding planes, which may move as rigid bodies with respect to each other, thus causing strain of the thickness direction fibres of the model. It is assumed that the deformation of the model is independent of position with respect to this thickness direction, so the relative motion of the two planes causes a direct strain of the thickness direction fibres only. Fig. 10. Illustration of a simply supported 2D/3D hybrid element CWM model resulting in 2D Generalised Plain Strain performance as defined by the commercial FEA-code ABAQUS. /63/ IIW 66 th Annual Assembly

23 4.6 Elastic Shakedown Analyses IIW Document Number: X Cyclic plastic stain behaviour is generally decomposed into three (3) regimes, see Fig. 11: - Elastic shakedown - Plastic shakedown - Plastic ratcheting Fig. 11. Illustration of ductile metallic material responses at cyclic loading Elastic shakedown (b) is defined as the stress or strain level below which there is a no cyclic plasticity during each cycle. In other words, the condition of the elastic shakedown is obtained when the plastic deformation occurs during the early cycles but the final steady stat behaviour is fully elastic due to the build-up of residual stresses. Plastic shakedown (c) is the condition in which the material experiences reversed plastic straining during cycling with no further accumulation of plastic deformation. Plastic ratcheting (d) describes the condition in which the material accumulates some plastic strain during each cycle. IIW 66 th Annual Assembly

24 IIW Document Number: X CWM-STUDY OF RESIDUAL STRESS CONTRIBUTING FACTORS 5.1 Introduction In order to identify the main contributing and/or driving factors to weld residual stress magnitudes in ship steel plate materials, the weld residual stresses in the way of an existing ship s bilge strake transversal butt weld have been approximated by CWM-analyses, Fig. 12. Fig. 12. The bilge strake location in a ship structure /64/ Due to uncertainties regarding the actual weld manufacturing process used it was found necessary to carry out a parametric CWM-study covering a broad range of base material properties as well as possible thermal- and mechanical boundary conditions at the time of the production welding. The weld joint analysed is a double sided butt weld located in the bilge plate (material quality NV A32), Fig. 13. Fig. 13. The keel strake transversal butt-weld joint analysed IIW 66 th Annual Assembly

25 5.2 Assessment of the WPS IIW Document Number: X Before the modelling and simulation work commenced should the actual WPS and its associated WPQR be scrutinized. Anyhow, a copy of the WPS used at the fabrication of the weld joint could not be identified and for that sake was the review limited to an assessment of the WPQR Weld Joint Geometry and CWM-Models Based on the combined information from the Ship s shell expansion plan and the weld joint geometry detail information in WPQR, see Fig. 14. It was decided to use a 3D transient thermal FE-weld simulation model as illustrated in Fig. 15 and Fig. 16. Fig. 14. The weld joint geometry details of the WPQR IIW 66 th Annual Assembly

26 IIW Document Number: X Fig. 15. Weld joint geometry used at the 3D transient thermal FE-weld simulation Fig D transient thermal FEA model used Dimensions: 406 x 306 mm and t1 = 30 mm ;t2 = 20 mm The 3D transient thermal model constituted of a structured mesh utilising LS-Dyna s 8-node hexahedral fully integrated selective reduced solid element /65/ with a total number of solid elements and nodes, Fig. 17. The mechanical element has a linear displacement approximation and the corresponding thermal element is a fully integrated 8-node element with linear temperature approximation. At the 3D transient thermal simulation only the thermal element was activated. Fig D transient thermal FEA model s cross section mesh density IIW 66 th Annual Assembly

27 IIW Document Number: X The 2D transient thermo-mechanical model constituted of a 2D/3D hybrid element formulation with a 2D Generalized Plain Strain performance in accordance with the generalized plane strain theory used by the commercial FEA-code Abaqus. /63/ The element type used was LS-Dyna s 8- node hexahedral fully integrated selective reduced solid element. /65/ For the full size plate model was a total number of 8476 solid elements and nodes used. The mechanical element has a linear displacement approximation and the corresponding thermal element is a fully integrated 8-node element with linear temperature approximation. Fig D transient thermo-mechanical FEA model used, the blue section is the part of the plate that is used at the final elastic plastic shakedown analyses. Dimensions: 4000 x 3000 mm and t1 = 30 mm ;t2 = 20 mm 2D/3D hybrid element formulation with a 2D Generalized Plain Strain performance /63/ It shall be noted that the cross section mesh density used for the 2D transient thermo-mechanical model ( Fig. 19) is about 4 times denser compared to the 3D transient thermal model, Fig. 17. For the sake of capturing local weld residual stresses and deformations as well as avoiding numerical stress and strain artefacts in the way of the HAZ. /66/ Fig D transient thermo-mechanical FEA model s cross section mesh density IIW 66 th Annual Assembly

28 IIW Document Number: X Weld Heat Input An initial 3D transient thermal simulation of the weld joint was carried out with the nominal weld heat input values stated in the WPQR, Fig. 20 and Table 5. Fig. 20. Weld arc energy parameters stated in the WPQR Table 5 Weld Heat Input Values Covered by the WPQR Pass [No.] I [A] U [V] η V [cm/min] Q min [kj/mm] Q nom [kj/mm] Q max [kj/mm] Root ,69 2,25 2,81 Run ,46 4,61 5,76 Run ,68 4,90 6,12 Run ,57 3,43 4,29 Run ,68 4,90 6,12 Based on the results of the initial 3D transient thermal simulation activity it is understood, beyond any doubt, that the actual weld joint not could have been produced with weld heat input values covered by the presented WPQR. As its nominal weld heat input values, Q nom, resulted in an extremely large weld melt pool, Fig. 21. IIW 66 th Annual Assembly

29 IIW Document Number: X Fig. 21. Weld melt pool temperature and distribution (through the plate penetration) of Run-5 For that sake it was decided to use three (3) different weld heat input magnitudes in order to illustrate the weld heat input s influences on the weld residual stress values in the way of the HAZ, Heat Affected Zone. The values used are presented in Table 6 here below and it shall be noted that Q-1 and Q-2 heritage from a DNV approved WPQR, Fig. 22 and Table 6. /67/ Table 6 Weld Heat Input Values Used at the CWM-analysis Q-1 Q-2 Q-3 Pass [No.] [kj/mm] v [cm/min] [kj/mm] v [cm/min] [kj/mm] v [cm/min] Root 1,260 50,0 1,680 50,0 1,20 32 Run-2 1,470 52,8 1,960 52,8 2,00 25 Run-3 1,470 52,8 1,960 52,8 2,25 25 Run-4 1,798 48,0 2,397 48,0 2,00 28 Run-5 1,362 54,5 1,817 54,5 2,25 25 IIW 66 th Annual Assembly

30 IIW Document Number: X Fig. 22. Weld arc energy parameters stated in WPQR No: 188:00:00 Table 7 Weld Heat Input Values Covered by WPQR No: 188:00:00 Pass [No.] I [A] U [V] η V [cm/min] [kj/mm] [kj/mm] [kj/mm] Root ,0 1 50,0 1,260 1,680 2,100 Run ,0 1 52,8 1,470 1,960 2,450 Run ,0 1 52,8 1,470 1,960 2,450 Run ,5 1 48,0 1,798 2,397 2,996 Run ,0 1 54,5 1,362 1,817 2,271 IIW 66 th Annual Assembly

31 IIW Document Number: X Thermal and Mechanical Boundary Conditions It is not known, in detail, how the butt weld joint in question has been manufactured. For that reason was the joint welding simulated with three (3) possible types of welding tables. Resulting in three (3) different thermo-mechanical boundary conditions at the time of production welding, the alternatives are: - The plates are placed on a cast steel welding table, tightly/rigidly fixated by wedges, Alternative 1 in Fig The plate are placed on a concrete floor reinforced with heavy steel bars to which the plates are tightly/rigidly fixated by wedges and/or tack weld joints, Alternative 2 in Fig The plates are welded in a flake line with mechanical compression locking Alternative 3 in Fig. 23 Fig. 23. Illustration of the three (3) alternative thermo-mechanical boundary conditions at manufacturing of the butt weld joint The welding simulation was done with an anticipated takt time of 600 seconds (10 min) in the following welding production sequence: Takt 1 Fixation of the plates to the welding table at 20 C room temp., Fig. 24 Takt 2 Takt 3 Takt 4 Welding of Root Pass and cooling at 20 C room temp., t = 600 s Welding of Pass-2 and cooling at 20 C room temp., t = 600 s Welding of Pass-3 and cooling at 20 C room temp., t = 600 s Takt 5 Flipping the plate as well as cooling at 20 C room temp., t = 600 s, Fig. 25 Takt 6 Fixation of the plates to the welding table at 20 C room temp., Fig. 26 Takt 7 Takt 8 Welding of Pass-4 and cooling at 20 C room temp., t = 600 s Welding of Pass-5 and cooling at 20 C room temp., t = 600 s Takt 9 Release of plate and cooling at 20 C room temp., t = 600 s, Fig. 27 IIW 66 th Annual Assembly

32 IIW Document Number: X Fig. 24. Takt 1: Fixation of plates to the three (3) alternative welding tables before welding the Root pass Fig. 25. Takt 5: Flipping the plates at completion of Pass-3 Fig. 26. Takt 5: Fixation of plates to the three (3) alternative welding tables before welding Pass-4 Fig. 27. Takt 9: Releasing the plates at completion of Pass-5 IIW 66 th Annual Assembly

33 IIW Document Number: X During all welding sequences is a 100% thermal- and mechanical bonded contact anticipated between the plates and the actual welding table. At the flip- and release sequences are the plates simulated as simply supported, Fig. 28, as described in Chapter 4.5, 2D Transient Thermo- Mechanical FEA of this report. Fig. 28. Illustration of a simply supported 2D/3D hybrid element CWM model At the welding of the passes Root, Pass-2 and Pass-3 it is anticipated that the weld joint preparation surfaces of the top side will be covered by weld flux. As well as the void space of Pass-4 and Pass-5 is filled up and packed with a backing flux or some sort of a ceramic backing bar, Fig. 29. It implies that there will be almost zero (0) energy transport in the form of conduction, radiation and/or convection from all weld joint preparation surfaces. Fig. 29. Illustration of a ceramic backing bar IIW 66 th Annual Assembly

34 IIW Document Number: X During all the weld simulation sequences is all plate and weld metal surfaces simulated to be exposed to the surrounding work shop atmosphere by the use of an apparent thermal convection boundary condition. Describing the total amount of heat transfer from the surfaces due to convection and radiation, Equation. 25 and Equation. 26. /54/ /56/ Equation. 25. Equation Elastic Shakedown Analyses At completion of the weld production related simulations was a 0,795 m long section of the plate in the way of the butt weld joint (see Fig. 30) subjected to an elastic shakedown analysis with boundary conditions and load spectra presented in Fig. 31 and Fig. 32. Fig. 30. The 0,795 m long plate section subjected to elastic shakedown Fig. 31. The boundary conditions used at elastic shakedown analyses Fig. 32. The load spectra used at the elastic shakedown analyses IIW 66 th Annual Assembly

35 5.3 Description of the CWM Analyses Carried Out IIW Document Number: X The CWM-analyses has been done with the weld filler material combination Y-D/NF310, that has a typical yield stress, σ y = 499 MPa at 20 C. /68/ The FE-weld simulations has been done with respect to the as elastic shakedown condition of weld residual stresses (σ xx -, σ yy - and σ zz - stresses) along Line A, -B, -C and D in the way of a butt weld joint, as illustrated in Fig. 33. Fig. 33. Illustration of the residual stress (σ xx ; σ yy ; σ zz ) sampling lines weld In order to study and explain the base materials influences on the resulting weld residual stress magnitude have a total number of 18 simulations been carried out for two (2) different types of base materials. Simulated to be welded with the weld filler material combination Y-D/NF310, that has a typical yield stress, σ y = 499 MPa at 20 C. /68/ The base materials are: - Grade NV A, with a minimum yield stress, σ y = 235 MPa at 20 C - Grade NV AH32, with a typical yield stress, σ y = 417 MPa at 20 C An overview of the 18 different simulations carried out is presented in Table 8 and Fig. 34 here below. IIW 66 th Annual Assembly

36 Alternative IIW Document Number: X Table 8 Weld Simulations Carried Out Weld filler material combination Y-D/NF310 (σ y = C) Grade NV A (σ y = 235 MPa at 20 C) Grade NV AH32 (σ y = 417 MPa at 20 C) Q-1 (Q min ) Q-2 (Q nom ) Q-3 (Q max ) Q-1 (Q min ) Q-2 (Q nom ) Q-3 (Q max ) 1 NVA Alt1-Q1 NVA Alt1-Q2 NVA Alt1-Q3 NVAH32 Alt1-Q1 NVAH32 Alt1-Q2 NVAH32 Alt1-Q2 2 NVA Alt2-Q1 NVA Alt2-Q2 NVA Alt2-Q3 NVAH32 Alt2-Q1 NVAH32 Alt1-Q2 NVAH32 Alt2-Q3 3 NVA Alt3-Q1 NVA Alt3-Q2 NVA Alt3-Q3 NVAH32 Alt3-Q1 NVAH32 Alt3-Q2 NVAH32 Alt3-Q3 Fig. 34. Illustration of the three (3) alternative thermo-mechanical boundary conditions at manufacturing of the butt weld joint Material properties and modelling The minimum yield stress (σ y ) of the base material NV A is 235 MPa at 20 C and it will experience kinematic hardening during the thermo-mechanical deformations of the weld process cycles. /69/ /70/ The residual stress release temperature interval used for NV A is C. A typical yield stress (σ y ) of the base material NV AH32 produced in Japan around year 1996/1997 has been found to be about σ y = 417 MPa at 20 C, Appendix A. The steel plate will experience kinematic hardening during the thermo-mechanical deformations of the weld process cycles. /69/ /70/ The residual stress release temperature interval used for NV AH32 is C. The yield stress (σ y ) of the weld filler material combination Y-D/NF310 is 499 MPa at 20 C and it will experience kinematic hardening during the thermo-mechanical deformations of the weld process cycles. /68/ The weld filler material activation temperature interval used for Y- D/NF310 is C and the residual stress release temperature interval is C. A 100% Linear Kinematic hardening formulation was used for the base- and the weld filler materials in order to simulate the hardening during the thermo-mechanical deformations of the weld process cycles. /2/ /54/ /55/ /56/ IIW 66 th Annual Assembly

37 Alternativ e 5.4 Weld Residual Stress Results - Elastic Shakedown IIW Document Number: X The sampling lines of As Shakedown residual stress (σ xx ; σ yy ; σ zz ) results in the way of the weld joint at 20 C are presented in Fig. 35. Fig. 35. Illustration of the weld residual stress (σ xx ; σ yy ; σ zz ) sampling lines Table 9 Weld Simulations As Shakedown Residual Stress (σ xx ; σ yy ; σ zz ) Results Weld filler material combination Y-D/NF310 (σ y = C) Q-1 (Q min ) Grade NV A (σ y = 235 MPa at 20 C) Q-2 (Q nom ) Q-3 (Q max ) Q-1 (Q min ) Grade NV AH32 (σ y = 417 MPa at 20 C) Q-2 (Q nom ) Q-3 (Q max ) 1 Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Table 10 Weld Heat Input Values Used at the CWM-analysis Q-1 Q-2 Q-3 Pass [No.] [kj/mm] [kj/mm] [kj/mm] Root 1,260 1,680 1,20 Run-2 1,470 1,960 2,00 Run-3 1,470 1,960 2,25 Run-4 1,798 2,397 2,00 Run-5 1,362 1,817 2,25 IIW 66 th Annual Assembly

38 5.4.1 Weld Residual Stress Results Alternative 1 and Q min IIW Document Number: X Fig. 36. Weld residual stresses (σ xx ) As Shake Down : NV A Left and NV AH32 Right Fig. 37. Weld residual stresses (σyy) As Shake Down : NV A Left and NV AH32 Right Fig. 38. Weld residual stresses (σ zz ) As Shake Down : NV A Left and NV AH32 Right IIW 66 th Annual Assembly

39 5.4.2 Weld Residual Stress Results Alternative 1 and Q nom IIW Document Number: X Fig. 39. Weld residual stresses (σxx) As Shake Down : NV A Left and NV AH32 Right Fig. 40. Weld residual stresses (σyy) As Shake Down : NV A Left and NV AH32 Right Fig. 41. Weld residual stresses (σ zz ) As Shake Down : NV A Left and NV AH32 Right IIW 66 th Annual Assembly

40 5.4.3 Weld Residual Stress Results Alternative 1 and Q max IIW Document Number: X Fig. 42. Weld residual stresses (σxx) As Shake Down : NV A Left and NV AH32 Right Fig. 43. Weld residual stresses (σyy) As Shake Down : NV A Left and NV AH32 Right Fig. 44. Weld residual stresses (σ zz ) As Shake Down : NV A Left and NV AH32 Right IIW 66 th Annual Assembly

41 5.4.4 Weld Residual Stress Results Alternative 2 and Q min IIW Document Number: X Fig. 45. Weld residual stresses (σxx) As Shake Down : NV A Left and NV AH32 Right Fig. 46. Weld residual stresses (σyy) As Shake Down : NV A Left and NV AH32 Right Fig. 47. Weld residual stresses (σ zz ) As Shake Down : NV A Left and NV AH32 Right IIW 66 th Annual Assembly

42 5.4.5 Weld Residual Stress Results Alternative 2 and Q nom IIW Document Number: X Fig. 48. Weld residual stresses (σxx) As Shake Down : NV A Left and NV AH32 Right Fig. 49. Weld residual stresses (σyy) As Shake Down : NV A Left and NV AH32 Right Fig. 50. Weld residual stresses (σ zz ) As Shake Down : NV A Left and NV AH32 Right IIW 66 th Annual Assembly

43 5.4.6 Weld Residual Stress Results Alternative 2 and Q max IIW Document Number: X Fig. 51. Weld residual stresses (σxx) As Shake Down : NV A Left and NV AH32 Right Fig. 52. Weld residual stresses (σyy) As Shake Down : NV A Left and NV AH32 Right Fig. 53. Weld residual stresses (σ zz ) As Shake Down : NV A Left and NV AH32 Right IIW 66 th Annual Assembly

44 5.4.7 Weld Residual Stress Results Alternative 3 and Q min IIW Document Number: X Fig. 54. Weld residual stresses (σxx) As Shake Down : NV A Left and NV AH32 Right Fig. 55. Weld residual stresses (σyy) As Shake Down : NV A Left and NV AH32 Right Fig. 56. Weld residual stresses (σ zz ) As Shake Down : NV A Left and NV AH32 Right IIW 66 th Annual Assembly

45 5.4.8 Weld Residual Stress Results Alternative 3 and Q nom IIW Document Number: X Fig. 57. Weld residual stresses (σxx) As Shake Down : NV A Left and NV AH32 Right Fig. 58. Weld residual stresses (σyy) As Shake Down : NV A Left and NV AH32 Right Fig. 59. Weld residual stresses (σ zz ) As Shake Down : NV A Left and NV AH32 Right IIW 66 th Annual Assembly

46 5.4.9 Weld Residual Stress Results Alternative 3 and Q max IIW Document Number: X Fig. 60. Weld residual stresses (σxx) As Shake Down : NV A Left and NV AH32 Right Fig. 61. Weld residual stresses (σyy) As Shake Down : NV A Left and NV AH32 Right Fig. 62. Weld residual stresses (σ zz ) As Shake Down : NV A Left and NV AH32 Right IIW 66 th Annual Assembly

47 6 DISCUSSION IIW Document Number: X Thermo- and Mechanical Boundary Conditions From the results of the parametric butt weld joint calculations it can be concluded that when one is using exactly the same weld process parameters and production sequence. For exactly the same base- and weld filler material will the magnitudes of the weld residual stresses depend on the thermo- and mechanical boundary conditions utilised at the manufacturing process, Fig. 63, Fig. 64 and Table 11. Fig. 63. Weld residual stresses (σ xx ) As Shake Down produced on welding table Alternative 1 Base material Grade NV AH32 (σ y = 417 MPa at 20 C) Weld filler material combination Y-D/NF310 (σ y = C) Fig. 64. Thermo-mechanical boundary condition Alternative 1 (left) and Alternative 3 (right) Table 11 Weld Heat Input Q-2 Pass [No.] [kj/mm] Root 1,680 Run-2 1,960 Run-3 1,960 Run-4 2,397 Run-5 1,817 IIW 66 th Annual Assembly

48 6.2 Weld Heat Input IIW Document Number: X It has also been noticed that the magnitudes of the weld residual stresses are dependent on the weld heat input, Q, which in combination with the base material s thermal properties and the surrounding thermal boundary conditions controls the weld cooling rate. /14/ A low weld heat input, Q min, results in higher weld residual stress magnitudes compared to a high weld heat input, Q max, Fig. 65, Fig. 66 and Table 12 Fig. 65. Weld residual stresses (σ xx ) As Shake Down produced on welding table Alternative 3 Q min left and Q max - right Base material Grade NV AH32 (σy = 417 MPa at 20 C) Weld filler material combination Y-D/NF310 (σy = C) Fig. 66. Thermo-mechanical boundary condition Alternative 3 Table 12 Weld Heat Input Q-1 Q-3 Pass [No.] [kj/mm] [kj/mm] Root 1,260 1,20 Run-2 1,470 2,00 Run-3 1,470 2,25 Run-4 1,798 2,00 Run-5 1,362 2,25 IIW 66 th Annual Assembly

49 6.3 Yield Stress Difference IIW Document Number: X The magnitudes of the weld residual stresses appear, to a minor extent, to depend on the yield stress difference between the base material and the weld filler material. A low yield stress difference value is indicated to result in a low residual stress value whilst a high yield stress difference value is indicated to result in a high residual stress value, Fig. 67, Fig. 68 and Table 13. Fig. 67. Weld residual stresses (σ xx ) As Shake Down produced on welding table Alternative 1 Weld filler material combination Y-D/NF310 (σ y = C) Base material Grade NV A (σ y = 235 MPa at 20 C - left Base material Grade NV AH32 (σ y = 417 MPa at 20 C) right Fig. 68. Thermo-mechanical boundary condition Alternative 1 Table 13 Weld Heat Input Q-1 Pass [No.] [kj/mm] Root 1,260 Run-2 1,470 Run-3 1,470 Run-4 1,798 Run-5 1,362 IIW 66 th Annual Assembly

50 7 CONLUSIONS IIW Document Number: X Based on the results of the parametric CWM analyses carried out it can be concluded that the following factors are contributing and/or driving the magnitude of the weld residual stress values: Thermal- and Mechanical Boundary Conditions during the production welding Yield stress difference between the base- and the weld filler material Weld heat input, Q, which affects the weld cooling time IIW 66 th Annual Assembly

51 8 REFERENCES IIW Document Number: X /1/. Goldak, J., Akhlaghi, M., 2005, Computational Welding Mechanics, ISBN: ,Springer-Verlag New York Inc., USA /2/. Lindgren, L.-E., 2007, Computational Welding Mechanics, ISBN: , Woodhead Publishing, UK /3/. Lindström, P. R. M., Josefson, B. L., 2012, 2D, Axisymmetric and 3D Finite Element Analysis Assessment of the IIW RSDP Round Robin Initiative, Phase 1 and Phase 2, IIW Document No. IIW-X , IIW-XIII and IIW-XV , IIW General Assembly 2010, Denver, USA /4/. Josefson, B. L., Lindström, P., and Molin, M., 2010, 2D and 3D Simulation of the IIW Round Robin Benchmark, IIW Document No. IIW-XIII , IIW General Assembly 2010, Istanbul, Turkey /5/. Cox, D. S., 2011, Repair of the NRU Reactor Vessel: Technical Challenges and Lesson Learned, International Conference on Research Reactors: Safe Management and Effective Utilization, November 2011, Rabat, Marocco /6/. Goldak, J., Yetisir, M., and Pistor, R., 2012, The Role of Computational Weld Mechanics in the Weld Repair of Canada s NRU Nuclear Reactor, Welding and Repair Technology for Power Plants, Tenth International EPRI Conference, EPRI, June 26-29, 2012, Marco Island, Florida, USA /7/. Pakhamaa, A., Wärmefjord, K., Karlsson, L., Soderberg, R., and Goldak, J., 2012, Combining variation with welding simulation for prediction of deformation and variation of a final assembly, International Journal of Computing and Information Science in Engineering, Vol. 12, to appear. /8/. Wohlfart, H., Nitscke-Pagel, Th., Diger, K., Siegel, D., and Brand, M., 2010, IIW Round Robin Residual Stress Calculations and Measurements FINAL REPORT, IIW Document No. IIW-X , IIW-XIII and IIW-XV , IIW General Assembly 2010, Istanbul, Turkey /9/. Smith, M.C., Bouchard, P.J., Turski, M., Edwards, L. and Dennis, R.J., 2012, Accurate prediction of residual stress in stainless steel welds, Computational Materials Science, Vol. 54, pp /10/. Muransky, O., Smith, M.C., Bendeich, P.J., Holden, T.M., Luzin, V., Martins, R.V., and Edwards, L., 2012, Comprehensive numerical analysis of a three-pass bead-in-slot weld and its critical validation using neutron and synchrotron diffraction residual stress measurements, International Journal of Solids and Structures, Vol. 49, pp /11/. Wohlfart, H., Nitscke-Pagel, Th., Dilger, K., Siegel, D., Brand, M., Sakkiettibutra, J. and Loose, T., 2012, Residual Stress Calculations and Measurements Review and Assessment of The IIW Round Robin Results, Welding in the World, 09/ , Vol. 56, pp /12/. Farajin, M., Nitscke-Pagel, Th., and Siegel, D., 2013, Welding Residual Stresses in Tubular Steel Joints and Their Behavior Under Multiaxial Loading, IIW Document No. IIW-XIII , IIW General Assembly 2013, /13/. Lindström, P. R. M., Faraij, M., Review and selection of a Finite Element Simulation Platform For Academic and Industrial Analyses of In-Service Welding operations (2004) ASME 23rd International Conference on Offshore Mechanics and Arctic Engineering, Vancouver, Canada /14/. Lindström, P. R. M., 2009, Heat Transfer Prediction of In-Service Welding in a Forced Flow of Fluid, ASME Journal of Offshore Mechanics and Arctic Engineering, Vol. 131, pp /15/. ISO , 2005, Quality requirements for fusion welding of metallic materials - Part 2: Comprehensive quality requirements /16/. Lindström, P. R. M., Josefson, B.L., Schill, M., and Borrvall, T., 2012, Constitutive Modeling and Finite Element Simulation of Multi Pass Girth Welds, NAFEMS NORDIC Conference: Engineering Simulation, Gothenburg, Sweden /17/. Janosch, J.-J., 2008, International Institute of Welding work on residual stress and its application to industry, International Journal of Pressure Vessels and Piping, Vol. 85, pp IIW 66 th Annual Assembly

52 IIW Document Number: X /18/. Wohlfahrt, H., 1997/2004 IIW Round Robin Update-Results for Residual Stress and Distortion Prediction, IIW-X/XII/XV-RSDP /19/. /20/. /21/. /22/. /23/. /24/. LSTC, Livermore Software Technology Corporation, USA, /25/. LS Dyna 971, Release 7.0.0, Double Precision MPP Solver, Livermore Software Technology Corporation, USA /26/. Jurinic, A., 2012, Arc welding simulation with Abaqus/Std, Simulia EuroNordics, Private communication /27/. Belytschko, T., Liu, W. K., and Moran, B., 2000, Nonlinear Finite Elements for Continua and Structures, ISBN: , John Wiley & Sons Ltd, UK /28/. Simo, J. C., Hughes, T. J. R., 2000, Computational Inelasticity, ISBN: , Springer- Verlag New York Inc., USA /29/. Lemaitre, J., Chaboche, J.-L., 1994, Mechanics of Solid Materials, ISBN: , Cambridge University Press, UK /30/. /31/. Goldak, J., Chakravarti, A., and Bibby, M., 1985, A Double Ellipsoid Finite Element Model for Welding Heat Sources, IIW Document No , International Institute of Welding /32/. Carslaw, H. S., Jaeger, J. C., 1986, Heat Conduction in Solids, ISBN: , Oxford University Press, UK, /33/. Shlichting, H., Gersten, K., Boundary Layer Theory, , Springer-Verlag Berlin and Heidelberg GmbH Co. K, Germany /34/. Eckert, E. R. G., Drake, R. M. Jr., 1987, Analysis of heat and mass transfer, ISBN: , Springer Verlag, Berlin, Germany /35/. LS-PrePost 3.2, 2012, Livermore Software Technology Corporation, USA /36/. /37/. /38/. /39/. Shah, Q., Abid, H., 2012, LS-Dyna for Beginners, ISBN: , LAP Lambert Academic Publishing AGCo KG, Germany /40/. /41/. /42/. /43/. /44/. Goldak, J., Paramjeet, G. K., and Bibby, M., 1990, Computer simulation of welding processes, ASME Winter Annual Meeting Symposium on Computer Modeling and Simulation of Manufacturing Processes, Producxtion of Engineering Division, pp. 193 /45/. Lindgren, L.-E., 2001, Finite element modeling and simulation of welding. Part 1 : Increased complexity, Journal of Thermal Stresses, Journal of Thermal Stresses, Volume 24, No. 2, pp /46/. Lindgren, L.-E., 2001, Finite element modeling and simulation of welding. Part 2 : Improved material modeling, Journal of Thermal Stresses, Journal of Thermal Stresses, Volume 24, No. 3, pp /47/. Lindgren, L.-E., 2001, Finite element modeling and simulation of welding. Part 3 : Efficiency and integration, Journal of Thermal Stresses, Journal of Thermal Stresses, Volume 24, No. 4, pp IIW 66 th Annual Assembly

53 IIW Document Number: X /48/. The Procedure Handbook of Arc Welding, 2012, 14th Edition, The James F. Lincoln Arc Welding Foundation, Cleveland, Ohio, USA /49/. SSAB Welding Handbook, 2012, 1st Edition, ISBN: , SSAB Oxelösund AB, Oxelösund, Sweden /50/. SSAB Sheet Steel Formin Handbook, 1998, SSAB Tunnplåt AB, Borlänge, Sweden /51/. SSAB Sheet Steel Joining Handbook, 2004, SSAB Tunnplåt AB, Borlänge, Sweden /52/. Material model: MAT_THERMAL_CWM (MAT_T07), LS Dyna 971, Release 7.0.0, Livermore Software Technology Corporation, USA /53/. Material model: MAT_CWM (MAT_270), LS Dyna 971, Release 7.0.0, Livermore Software Technology Corporation, USA /54/. B. Brickstad, B.L. Josefson, 1998, A parametric study of residual stresses in multi-pass buttwelded stainless steel pipes, International Journal of Pressure Vessels and Piping, Vol. 75, pp /55/. Zang, W., Gunnars, J., Mullins, J., Dong, P., and Hong, J. K. 2009, Improvement and Validation of Weld Residual Stress Modelling Procedure, Report 2009:15, ISSN: , Swedish Radiation Safety Authority, Stockholm, Sweden /56/. Mullins, J., Gunnars, J., 2009, Influence of Hardening Model on Weld Residual Stress Distribution, Report 2009:16, ISSN: , Swedish Radiation Safety Authority, Stockholm, Sweden /57/. Zang, W., Gunnars, J., Mullins, J., Dong, P., and Hong, J. K. 2009, Effect of Welding Residual Stresses on Crack Opening Displacement and Crack-Tip Parameters, Report 2009:17, ISSN: , Swedish Radiation Safety Authority, Stockholm, Sweden /58/. Wei, L., He, W., 2010, Comparison of Measured and Calculated Residual Stresses in Steel Girth and Butt Welds, Report no. 924/201, TWI Ltd, Cambridge, UK /59/. Wohlfahrt, H., 2009, Report on the Round Robin Tests on Residual Stresses 2009 Joint Working Group of Commission X/XIII/XV, IIW Document Nos. IIW- X , IIW-XIII , IIW-XV /60/. NS-EN :2009,Page 10 11, Ch.8 Sec.7, Heat input /61/. Svensson, L.-E., 1993,Control of Microstructure and Properties in Steel Arc Welds, ISBN: , CRC Press Inc, USA /62/. Shapiro, A. B., Heat Transfer in LS-Dyna, 2003, Proceedings of the 4th European LS-DYNA Conference. 22nd - 23rd May 2003, Ulm, Germany /63/. Abaqus Theory Manual 6.11, Chapter Generalized Plain Strain, Simulia /64/. IACS Recommendation 82, Surveyors Glossary Hull Terms & Hull Survey Terms, 2003, IACS, London, UK /65/. Page , Ch.3 Sec.4, Fully Integrated Brick Elements and Mid-Step Strain Evolution, LS Dyna Theory Manual, March 2009, Livermore Software Technology Corporation, USA /66/. Oddy, A. S., McDill, J. M. J., Goldak, J. A., 1990, Consistent Strain Field in 3D Finite Element Analysis of Welds, Journal of Pressure Vessel Technology, Vol. 112, pp /67/. WPQR nr: 188:00:00, 2008, Junoverken AB, Uddevalla, Sweden /68/. Report on the renewal approval test of automatic and semi-automatic welding materials, 2011, Report No.: 11124, 06 October 2011, Nippon Steel & Sumikin Weldding Co., Ltd., Narashino- City, Chiba-Pref., Japan /69/. DNV Rules for Classification of Ships - January 2013, Part 2 Materials and Welding, Ch. 2 Sec. 1 - Page 16, DNV, Høvik, Norway /70/. DNV-OS-B101 Metallic Materials, - October 2012, Ch.2 Sec.1 - Page 19, DNV, Høvik, Norway - o0o - IIW 66 th Annual Assembly

54 APPENDIX A NV GRADE EH32

55

56 APPENDIX B MATERIAL & PHYSICAL DATA

57 Density Reference: EN :2001 Thermal Conductivity Reference: X XV X-XV-RSDP IIW Round Robin on residual stress and Distortion prediction Phase 1 Results Specific Heat Capacity Reference: X XV X-XV-RSDP IIW Round Robin on residual stress and Distortion prediction Phase 1 Results

58 Young s Modulus Reference: X XV X-XV-RSDP IIW Round Robin on residual stress and Distortion prediction Phase 1 Results Poisson s Ratio Reference: X XV X-XV-RSDP IIW Round Robin on residual stress and Distortion prediction Phase 1 Results Yield Stress References: X XV X-XV-RSDP IIW Round Robin on residual stress and Distortion prediction Phase 1 Results DNV Rules for Ships, Part 2, Chapter 2 and 3 Report on renewal approval test of welding consumable flux combination Y-D / NF310, No , DNV

59 Plastic Modulus Reference: X XV X-XV-RSDP IIW Round Robin on residual stress and Distortion prediction Phase 1 Results Thermal Expansion Reference: EN :2001

60 Apparent Thermal Boundary Layer References: SSM Rapport Mullins, J., Gunnars, J., 2009, Influence of Hardening Model on Weld Residual Stress Distribution, Report 2009:16, ISSN: , Swedish Radiation Safety Authority, Stockholm, Sweden Apparent Boundary Layer Apparent Boundary 20 C T 500 C T h h T 500 C T h h 2 W m C W m C