Analytical Evaluation of Spectral Moments and Dirlik’s Damage Model to Allow Comparison of Life Testing With Dissimilar PSD Vibration Curves

Abstract

In the application of random vibration testing and the inverse power law to test products and extrapolate equivalent field life, the shape of the Power Spectral Density (PSD) vibration curve frequently introduces prediction errors and increased testing. A field PSD profile found to be highly dissimilar in shape to that used in the original accelerated testing raises questions of how to extrapolate equivalent life and often results in the need to repeat long duration vibration testing. Past work by the author has proposed utilizing the evaluation of zero crossings in two dissimilar PSD curves as a way to equate their testing impact and life predictions and therefore not require re-testing. This paper expands on this theory and explores utilization of spectral moments and Dirlik’s damage model to compare dissimilar PSD curves. Finite Element Analysis (FEA) and other analytical methods are utilized to expose a mechanical system to vibration inputs of different shapes, compute their fatigue life and then evaluate spectral moments and Dirlik’s damage model’s ability to account for dissimilar PSD curves.