A Practical Study of Fourier Analysis for Monitoring Fatigue Progression in Elements Subjected to Random Loads

It is sometimes necessary to determine the manner in which materials deteriorate with respect to time; for instance when quantifying a material's ability to withstand sustained dynamic loads. In such cases, it is well established that loss of structural integrity is reflected by variations in modal characteristics such as natural frequency. The research presented herein uses the system's 'instantaneous' frequency response function (FRF), captured using a modified short-time Fourier transform (STFT), to continuously monitor shifts in the system's natural frequency. This approach requires a compromise to be made between the accuracy of spectral estimates and how often they can be made. The limitations brought forth by this compromise can be significantly increased if unsuitable inputs for parameter extraction are applied. This paper uses the results from numerous physical and numerical experiments to address some practical limitations of Fourier analysis with respect to temporal resolution and the uncertainties associated with extracting variations in modal parameters. Results which aid the analyst in selecting suitable inputs for parameter extraction, including those of spectral enhancement techniques, namely data overlapping and zero-padding, are also included.