Stochastic fatigue crack growth analysis of metallic structures under multiple thermal–mechanical stress levels

Abstract This paper proposes an analytical–empirical framework for stochastic assessment of fatigue crack growth. The study offers solutions for drastic problems in studying fatigue field including dispersion and scarcity of fatigue data, and impact of several factors on fatigue crack growth. The research composes of two complementary phases. The first phase analyses the behavior of fatigue models under probabilistic conditions. The propagation of uncertainty is studied in fatigue models in this phase under cyclic loads. The research is based on two fundamental deterministic fatigue models: Walker and Forman correlations. The second phase employs probabilistic distributions for stochastic assessment of fatigue crack growth. A new approach is introduced to simulate fatigue crack growth stochastically under the effects of different variables through combining lognormal distribution and stress life models. Bayesian approach is applied for parameter estimation. The final presented models offer the ability of predicting fatigue crack length with desirable probability under multiple stress levels. Although the method is examined on an automotive application, the presented approach is universal and applicable for other industries dealing fatigue damage with multiple impacting factors.