The strain-life approach is a powerful toot for the assessment of steel constructions experiencing low cycle as well as high cycle fatigue. However, among the various modifications and variations of the original concept, the decision to choose a proper one is challenging. Especially since common guidelines do not provide clear recommendations or do not cover the strain-life approach at all. In this contribution, we compared widely accepted versions as well as somewhat newer ideas of the approach regarding the determination of local strains, mean stress correction or strength hypothesis in an extensive numerical study including three geometries with different sharp notches. We essentially emphasized the incorporation of plastic material behavior directly in the finite element (FE) calculation using a plastic material law. There are partly similar investigations, but they focus exclusively on elastic calculations with subsequent plasticity corrections. Based on the results, recommendations for the application of the approach are given and aspects worth focusing on in further research are suggested. Additionally, a local and global sensitivity analysis of the material parameters was performed for both, that is elastic calculations combined with a plasticity correction as well as plastic FE-computations. The performed analysis identifies the fatigue strength exponent and the fatigue ductility exponent as the parameters most influencing the estimated fatigue life, given their maximum correlation coefficients of 0.56 and 0.91 respectively. The observed results seem to critically assess current approximation methods of strain-life material properties, since these two parameters are often assumed to be constant values for a range of steel grades. (C) 2018 Elsevier Ltd. All rights reserved.
