Parameter Sensitivity Study of Vibration Induced Fatigue Analysis in Time Domain and Frequency Domain Approaches

In the automotive industry many components face fatigue failure due to prolonged vibrations. This is commonly known as Vibration Induced Fatigue (VIF). There are two approaches to evaluate this; time & frequency domain. A straight forward and widely used method is the rainflow counting technique in the time domain. This counting algorithm is readily available and, apart from the time history, it needs only one variable input (the number of stress ranges). In case of high cycle fatigue, longer time histories are required for a statistically representative fatigue estimate, which makes the time domain approach consume large amounts of time and resources. This shifts our interest towards frequency domain methods. In the frequency domain, Dirlik's method is proven to be robust and gives closer results to the time domain. Since Dirlik's approach is probabilistic, there are different variables (such as stress ranges and scaling factor for RMS value to predict maximum stress) to be considered for the evaluation of cumulative fatigue damage. The degree of closeness of Dirlik's method to the rainflow counting method mainly depends on the above mentioned variables. Apart from the literature available on these methods, damage dependency on these parameters is still unclear. This paper attempts to demonstrate the variation of the cumulative damage with respect to these parameters and tries to find the optimum range of operation. An exhaust mounting bracket, with random loading, is the subject of this analysis to compare these methods.