Characterization of the fretting wear damage under random excitation

Abstract

Conventional sinusoidal excitation is commonly used for investigating oscillatory fretting wear. However, in real life applications, the excitation force is random. This work provides a methodology for analyzing random fretting wear damage accumulation.The proposed wear energy characterization is based on the identification of the root mean square (RMS) displacement response of the fretting tribo-system to a random Gaussian excitation. To identify each sine component of the random signal, the Fourier transform is applied on the displacement, showing the signal magnitude as a function of frequency components. Based on the Power Spectral Density (PSD) definition, each frequency component can be related to specific sine amplitudes composing the original random signal. From this decomposition into sine components, new simplified approximations of the work rate and accumulated dissipated energy are proposed to characterize the fretting wear damage under true random vibrations.Tests were performed using two crossed cylinders (proprietary Zr alloy/nickel-based alloy 718) in the fretting wear configuration at high temperature (315 °C). True random tests were then performed using a Gaussian acceleration-control strategy. Imposing the PSD defining the acceleration level, the ratio between partial and gross slip domains is controlled through the mean and variance of the Gaussian displacement response, which can be related to the RMS displacement excitation parameter. This set of tests consisted of 6 × 106 cycles in the gross slip regime. Additional random simulation cases were generated to validate the proposed energy-based characterization parameter.Exact and simplified approximations of both work rate and accumulated dissipated energy are compared for real test conditions and simulated cases. The proposed approximations are function of the predominant frequency within the random signal response, the gross slip regime signal portion, and the RMS of the Gaussian random displacement response. Very good correlations were obtained, validating both work rate and energy loss simplified approximations.