Random Vibration Testing with Specified Fatigue Damage Spectrum and Preserved Power Spectral Density

Abstract

The fatigue damage spectrum (FDS) model characterizes how the damage potential is distributed over the excitation frequency range, similarly to how the power spectral density (PSD) characterizes the distribution of the excitation level. However, reproducing the in-service PSD during vibration testing on shakers does not necessarily mean that the FDS would be also reproduced. This is because some peaks, higher than those in a signal generated from the PSD, occur in vibrations of automobiles, trucks, and railway vehicles. Presence of these peaks in real in-service vibrations and their absence in the PSD-based random testing is the reason why the FDS obtained by the ordinary PSD control of the shaker is different in shape and usually lies below the in-service FDS. It is shown in the paper that the FDS shape as a function of frequency can be controlled by manipulating some of the IFFT phases instead of prescribing all of them randomly. Since the phase manipulation does not affect the excitation PSD, the FDS can be controlled with the PSD preserved. It was demonstrated in the paper for an example of automobile vibration that, when the ordinary PSD imitation failed in the numerical simulations, the developed PSD + FDS imitation was able to match not only the FDS of in-service vibration with non-Gaussian high peaks but also the accelerated FDS profile, which was 3 times higher than the in-service FDS. Numerical results have also shown that the FDS precision is maintained if the testing continues longer than the duration of the measured data record.