Approach to consider welding residual stresses in fatigue design using direct numerical simulations
The overall objective of this project is to develop an approach to include the influence of welding residual stresses in fatigue assessments through direct simulations. First, welding residual stresses in small-scale specimens and in large structures will be determined and compared. Residual stresses in small specimens used for experimental investigations are often relatively low due to the lack of structural constraints. The investigations will show in what range residual stresses are present in small-scale specimens used for fatigue tests and how they will differ in larger components. Using numerical welding simulations, a wide range of plate thicknesses and specimen dimensions will be assessed in order to identify the influence of geometrical parameters on residual stresses. The simulation results for the models of different dimensions, will allow a comparison between residual stresses in small-scale specimens and large structures. Since residual stresses may be redistributed or relaxed by superimposed external loads, for fatigue assessments the stabilized residual stresses after loading are relevant. Using numerical simulations, this redistribution is calculated for different load levels between compression and tension, i.e. different load ratios. As a result, the effective residual stresses affecting fatigue will be determined for different combinations of residual stress levels and load cases and compared for small and large-scale components. In fatigue tests it will then be investigated how residual stresses affect fatigue behaviour. Since residual stresses are generally not uniform over the plate thickness, the influence on crack initiation and propagation will be distinguished. Furthermore, two different weld geometries with different residual stress distributions and stress concentration factors will be compared: cruciform joints, transversal to the loading direction, and stiffeners longitudinal to the loading direction. Subsequently the residual stress influence will be compared to that of mean stresses. Overall, the test results will give information on the influence residual stresses have on different weld geometries compared to mean stresses. Finally, the results of the simulations and experiments will be evaluated to develop an approach to consider residual stresses in fatigue assessments using direct numerical simulations. For this purpose, it has to be determined which level of accuracy is needed in the simulations and how to read out residual stresses in order to obtain conservative outcomes. The effective stress ratio is determined by simulating the residual stress redistribution under loading. It is then used to calculate a stress ratio correction. Thus, it will be verified if such procedure based on a residual stress value on the plate surface is applicable or if further corrections are needed. The results of this project will further improve efficiency and safety in future design procedures.