A comparative study on the performance of the damage detection methods in the frequency domain

During last two decades, a vast number of damage detection methods have been proposed either in frequency or time domain. These methods normally have their own advantages and limitations or suitable applications; the purpose of this study is to examine the performance of the some popular methods on localisation a possible damage on a sample structure. All of the chosen methods are based on the frequency domain data and work based on proposing a damage sensitive indicator which contains spatial information. Mode shape curvature, frequency response functions' curvature, modal strain energy, flexibility matrix and spatial wavelet transform were amongst those damage detection methods were chosen for this study. The case study considers a clamped-clamped beam which was modelled by solid elements in order to define several damage stages based on different crack depth. Damage was simulated by reduction in elastic modulus of the elements in damage zone. The transient response of the structure due to an external impact excitation was obtained by ANSYS and then polluted by different percentages of white noise. The time-domain responses at selected evenly-spaced locations was then processed by MATLAB to achieve the FRFs and mode shapes respectively by applying Fourier transform and eigenvalue realization algorithm (ERA). Based on the obtained results, it was found that despite some of these methods were suggested by so many researchers, they completely fail in localising damage in the structure even at high level of damage severity.