Design Evaluation Method for Random Fatigue Based on Spectrum Characteristics

Abstract

This paper shows the new design approach for random fatigue evaluation based on spectral characteristics. Fatigue damage under random loading is usually evaluated according to the following steps: 1. decomposition of random wave to stress amplitude using the rainflow cycle counting method (RFC) 2. evaluation of fatigue damage using Palmgren-Miner’s linear summation rule. In design process, the fluctuation of load is usually characterized through power spectral density (PSD). Therefore, the design process is expected to be generalized if the fatigue damage is directly evaluated from the PSD together with S-N diagram of the material. In fact many properties related to fatigue damage, such as distribution of extreme values, can be derived theoretically from geometrical properties of PSD. However, concerning the distribution of stress amplitude counted by RFC, it is rather difficult to derive it theoretically due to its complicated procedure. In this paper, upper bound of stress amplitude distribution is confirmed for many random waves generated by numerical simulation for many types of PSD. Expressing the upper bound distribution by closed form function using PSD characteristics leads us to the direct evaluation of fatigue damage with safety margin if the fatigue damage by particular stress amplitude is approximated using some series expansion form. Simple approximation procedure of high cycle fatigue damage for austenitic stainless steel and ferritic steel will be proposed in the paper. Finally, design evaluation procedure for the fatigue damage evaluation from PSD together with S-N diagram is summarized.