An Efficient Methodology for Fatigue Reliability Analysis for Mechanical Components

This paper presents an efficient methodology to solve a fatigue reliability problem. The fatigue failure mechanism and its reliability assessment must be treated as a rate process since, in general, the capacity of the component and material itself changes irreversibly with time. However, when fatigue life is predicted using the $S$-$N$ curve and a damage summation scheme, the time dependent stress can be represented as several time-independent stress levels using the cycle counting approach. Since, in each counted stress cycle, the stress amplitude is constant, it becomes a random variable problem. The purpose of this study is to develop a methodology and algorithm to solve this converted random variable problem by combining the accumulated damage analysis with the first-order reliability analysis (FORM) to evaluate fatigue reliability. This task was tackled by determining a reliability factor using an inverse reliability analysis. The theoretical background and algorithm for the proposed approach to reliability analysis will be introduced based on fatigue failure modes of mechanical components. This paper will draw on an exploration of the ability to predict spectral fatigue life and to assess the corresponding reliability under a given dynamic environment. Next, the process for carrying out this integrated method of analysis will be explained. Use of the proposed methodology will allow for the prediction of mechanical component fatigue reliability according to different mission requirements.