Influence of modelling methods of a fluctuating weld geometry on the fatigue assessment using the effective notch stress approach

The notch geometry is influencing the fatigue of a welded joint, due to its local stress concentrating effect. This local stress can be considered in detail by local fatigue approaches. A widely used approach for the fatigue assessment of weld joints is the effective notch stress approach. The measurement of the notch geometry and the implementation of the same into a finite element analysis are crucial for the accuracy of the predicted fatigue life or number of load cycles to failure. During the elaborate process of calculating the notch stresses, time intensive tasks are capturing of the real weld geometry, building the finite element model, and positioning of a reference radius in the simulation model. This process is simplified by a method, as it deploys a statistical evaluated probabilistic weld geometry. These probabilistic geometries are believed to be representative for welds produced under equal conditions. This paper aims to validate the use probabilistic weld geometries by investigating the notch stresses and the fatigue assessment results for a group-based geometry in contrast to ones based on single specimens. The theoretical predicted fatigue lives are compared to experimental data to evaluate the prediction quality. The calculation of notch stresses for specimen-based geometries shows the best approximation to experimental data. However, the group-based models are able to provide comparable results.