Jiacai Zhu, Jiafen Cao, Wanlin Guo

International Journal of Fatigue  ,  173, 107697， 2023.

Abstract: Fatigue is one of the greatest threats to structural integrity. It is sensitive to the uncertainties of material quality, manufacturing quality and service load, making fatigue reliability analysis important and challenging. Here, a reliability analysis method considering these uncertainties is proposed. Firstly, the distribution of equivalent initial crack sizes in the material was obtained from the S-N data by back-extrapolation based on three-dimensional (3D) fatigue crack growth prediction. Then, the sample set for reliability analysis was obtained by randomly combining the applied stresses and crack sizes under the assumption that the crack sizes and applied stresses are independent. Further, the fatigue life of every sample is predicted with the 3D crack growth method, and the cumulative distribution of fatigue life was obtained to analyze the reliability. Based on the obtained initial crack distribution with stress ratio R ≥ 0, the predicted P-S-N diagrams of different stress ratios and stress concentrations were in accordance with the experimental data, indicating that the obtained initial crack size distribution has a certain universality. Finally, the influence of sampling numbers on the reliability analysis is analyzed with five types of commonly used metallic alloys. It is shown that the cumulative distribution of fatigue lives could be obtained with a less sampling number. Compared with Monte Carlo and importance sampling methods, this method can reduce the number of samples effectively and achieve the reliable probability of different fatigue life intervals.

 Highlights: •The equivalent initial cracks were calculated from S-N data with 3D crack growth method.

   •An efficient sampling method was proposed to obtain the sample set.

   •The rationality of initial crack distribution was verified by P-S-N diagrams.

   •Obtain the fatigue life distribution with a small sample size.

   •Calculate the reliable probability of different fatigue life intervals.

Link: https://www.sciencedirect.com/science/article/pii/S0142112323001986