Experiences and recommendations for numerical analyses of notch stress intensity factor and averaged strain energy density

This paper aims at providing recommendations for the practical application of relatively novel fatigue assessment approaches. Although the concept of notch stress intensity factor (N-SIF) itself was proposed decades ago to provide the analytical solution of the stress field in way of sharp notches, only recently it has been considered as the basis for fatigue strength assessment of welded joints. N-SIF itself and derived quantities such as the equivalent peak stress or the averaged strain energy density (SED) can be assumed as fatigue governing parameters and applied in everyday engineering practice being supported by powerful computation facilities. The N-SIF approach is still too cumbersome for practical applications; the SED approach seems promising to overcome shortcomings, to be applied flexibly to different joint geometries and possibly to automate the calculation procedure in finite element analysis. Software having different features and different refinement strategies are applied in order to benchmark advantages and disadvantages of possible practical solutions for the approaches based on N-SIF and averaged SED, thus offering various element types and modelling strategies. In general, high computational accuracy has been verified, despite relatively coarse meshes, if the element type, shape and mesh pattern is selected properly. However, the approach based on the averaged SED can be unsuitable for studying the notch effect of varying geometries if the averaging area increases more than the local stress concentration rises, as shown for the simplified geometry of a butt joint