Identification of cracks in thin-walled structures by means of wavenumber filtering

Abstract

This research is related to a signal processing of full wavefield data as an effective tool for detection, localization and visualization of a crack growth in thin-walled structures. Full wavefield data of propagating Lamb waves in structures such as plates and shells made out of metallic alloys and composite laminates contain a wealth of information about wave pattern anomalies due to occurrence of a damage. The aim is to demonstrate a method for enhancing damage visualization in structures such that estimation of the length and orientation of the crack can be easily obtained. The proposed signal processing involves application of discrete fast Fourier transform, wavenumber domain filtering and inverse discrete Fourier transform. The method is further enhanced by a technique for compensation of the wave attenuation so that the effects of structural damage have the same influence regardless of the location. The concept is first illustrated on numerically simulated data, and then tested on experimental results. In the experiments, full wavefield measurements are obtained using a scanning laser Doppler vibrometer, which allows the measurement of displacements and/or velocities along three axes over a user-defined grid. In the proposed method only out-of-plane velocities are used. Tests performed on simple aluminum and composite plates with artificially introduced longitudinal cracks confirm the effectiveness of the method and its potential for application to the inspection of a variety of structural components.