Damage Localization in Inhomogeneous Plates Using a Sparse Array of Ultrasonic Transducers

A variety of Lamb wave methods have been shown to be effective in localizing damage in homogeneous plates. However, many structural components of interest are not simple plates of constant thickness, but are plate-like in geometry with inhomogeneities present in the form of thickness variations. Here we consider aluminum plates with thickness transitions typical of realistic structures. A sparse array of piezoelectric disc transducers is mounted on the plate, and defects are simulated via glued-on masses at various locations. Signals recorded from the sparse array after damage is introduced are compared to baseline signals recorded from the undamaged plate. Both beamforming and feature-based algorithms are implemented to localize damage. It was found that beamforming algorithms are ineffective in localizing damage when received signals are complex; however, images based upon an energy arrival feature were successful in localizing damage and contained virtually no artifacts. Results are shown for two plates, each with a sparse array of six transducers, and with simulated damage located both within and outside of the array. These results are significant because recorded signals are of the complexity expected for real structures.