Probabilistic high cycle fatigue behaviour prediction based on global approach criteria

This paper presents an approach to predict the reliability of high cycle fatigue (HCF) behaviour of metallic parts using the multi-axial HCF criterion of Crossland, for the case of normally distributed in-phase fully reversed torsion and bending loading and HCF material characteristic parameters. The dispersions of: (i) the HCF criterion material characteristic parameters and (ii) the applied loading have been taken into account. The reliability of the HCF resistance was determined by using the “Strength Load” with first order reliability method (FORM). This approach gives iso-probabilistic Crossland diagrams (PCD) corresponding to different coefficient of variation (COV) of loading and material HCF characteristic parameters. An application has been carried out on a hard steel metal submitted to a fully reversed torsion and bending loading. Two types of various dispersed loadings, having different COV, are studied: (i) only random torsion amplitude loading and (ii) both random torsion and bending amplitude loading. The proposed method allows evaluation of the influences of different dispersions on the reliability of the HCF behaviour. It has been observed that, the proposed method is qualitatively consistent with the physical observations and leads to a more reliable HCF prediction compared to the deterministic approach, which takes into account separately two fatigue limits corresponding to a given reliability value, in the HCF criterion of Crossland.