Residual stress development in a repaired welded connection

Abstract

Research shows that the fatigue strength of a Very High Strength Steel (VHSS) can be significantly negatively effected by welds in a structure. The fatigue lifetime may be extended after cracking occurs by means of fatigue crack repair. A better insight by means of Finite ElementModelling (FEM) can help in understanding the effects of a repair procedure in awelded connectionwhich can results in the optimisation of the repairing process. The FEM model needs to be verified with experiment but measurement of residual stresses in a specimen requires either expensive or highly specialised equipment. Therefore, a more simple approach for model verification could aid in the accessibility of modelling of welding. The aim of this thesis is to establish a finite element model for a repair weld in a welded connection, and investigate to what extent the model can be verified by welding distortion. A weld repair experiment, consisting of an initial weld, segment removal and repair welding, is performed to retrieve experiment data to be used as verification and input for the FEM model. First the modelling of the initial weld is described, consisting of a thermal analysis and subsequent thermomechanical analysis. Regarding the thermal analysis, a mesh sensitivity analysis is performed as well as 2Dmodel results are compared with 3D results. The thermal analysis is performed by Goldak heat distribution and after calibration of the heat input, show a good agreement with temperature data retrieved from experiment. Regarding the thermo-mechanical analysis, an Abaqus subroutine was established to retrieve information of the weld pool during the thermal analysis. This information is used in a subroutine during the thermo-mechanical analysis to assign differentmaterial behaviour to different elements of the weld, based on thermal history. In the thermo-mechanical model distortion results show a good agreement with experiment distortion results. As a result of unexpected movement during clamping of the experiment, calibration was required of the mechanical boundary conditions in the model to find matching results. Modelling of the weld segment removal was performed, and similar trends of distortion between experiment and model results were found. However, further measurement of residual stresses is required to further evaluate the correlation between residual stress and distortion. Results of the thesis show that for a defect or faulty weld section, the measurement of the residual stresses is of great importance. Modelling of the full weld repair procedure proved to be beyond the scope of this thesis, since Finite Element work was more time consuming than initially anticipated.