A probabilistic methodology for fatigue life prediction

Abstract

This paper describes efforts in the probabilistic modelling of fatigue. The methodology proposed deals with crack nucleation from surface defects within a fracture mechanics framework. The methodology provides a quantitative understanding of the effects of bulk defects and their size distribution on the variation of fatigue life. It also considers the contributions of the variations of fatigue crack nucleation life and of crack growth resistance to the variation of fatigue life. A PM Ni-based superalloy Udimet 720 was used to evaluate the probabilistic methodology by an analytical solution and a Monte Carlo simulation. The model predictions were found to be in good agreement with the experimental results. However, it is found that Monte Carlo simulation provides a closer agreement with the experimental life distribution than the analytical solution. This is because the Monte Carlo simulation removes some untrue assumptions which have to be assumed to obtain the analytical solution.