Improving with probabilistic and scale features the Basquin linear and bi-linear fatigue models

Abstract

Although mechanical fatigue is considered stochastic in nature and there is a trend in industry towards reliability-based design, most popular high cycle stress-based fatigue models applied by industry still relies on deterministic methods. This is evidenced by the absence of integral probabilistic approach for predicting local and global probabilities of failure of structures in fatigue analysis commercial codes. In this paper, a probabilistic extension of the classical Basquin linear and bi-linear fatigue models, which are the most commonly promoted by standards (ASTM, UNE…) and Guidelines (FKM, DNV GL, VDI…), is proposed. The proposed models include the scale effect and the probabilistic character of the S-N field. This seems to be a judicious and recommendable option for the practicing engineer to face the real fatigue design of components under varying loading while ensuring safety requirements. The proposed enhanced model is implemented into the NCode2020 software to illustrate the possible implementation in general commercial codes focused on fatigue design. Finally, the applicability of the procedure proposed is illustrated by means of a practical example that includes the evaluation of experimental results and the prediction of failure for an Open-Hole-Plate.