Fast evaluation of kurtosis and skewness by modal decomposition for fatigue analysis in frequency domain under non-Gaussian random loads

In vibration fatigue, spectral methods are commonly used to assess fatigue damage of structures subjected to random vibrations. However, such methods often crash when non-Gaussian and non-stationary loads occurs, since all of them are based on the assumption of stationary and Gaussian loads. In literature several methods exist to overcome such issue but, among them, the most promising are those namely correction coefficients approaches. All the available correction coefficients are function of the kurtosis and skewness of the system response that can address only by time-consuming transient time domain analysis. This necessarily clash with the main advantages of spectral methods that is the computational efficiency. To this aim, this paper proposes a time-domain method that, exploiting the modal decomposition allows to efficiently calculate the central moments.