Development of a probabilistic health model representing variable amplitude fatigue loading damage in austenitic stainless steel nuclear components

Abstract

Fatigue scatter models allow the computation of uncertainties and confidence intervals linked to fatigue failure prediction. This phenomenon can be linked to a microscopic crack growth mechanism that is not modeled in S–N curves fatigue assessment approaches. The main fatigue scatter models found in the literature only allow the linear dependence of the cyclic load amplitudes to be modeled. The first contribution of this article is the development of a fatigue model that allows the prediction of nonlinear scatter dependence on load amplitude. More precisely, the proposed dependence structure is based on a partially affine function with a threshold effect, such that there is no dependence for load amplitudes sufficiently high. The model is successfully tested on a large fatigue database. A cumulative damage model is then obtained by adding two assumptions extracted from the literature. It is based on representing fatigue damage as the decrease in a structure’s fatigue health. The constructed model presents nonlinear cumulative damage properties and is successfully tested on two amplitude fatigue tests extracted from the literature. The whole fatigue failure prediction framework is finally applied to a real structure subjected to variable amplitude loadings.