Improving impact-echo method by using cross-spectral density

When the impact-echo method is used to detect internal flaws in materials and delaminations between layers, the signal which is collected by the transducer after rebounding in any interface is not always amenable to easy interpretation. The longitudinal waveform produced by multiple longitudinal wave reflections between two surfaces is affected by the interferences with the other waves generated by reflections (PS, SP and S waves) or with the Rayleigh wave and its rebounds from the borders of the material. This study applies signal processing techniques on the signals obtained, with finite elements simulation, in different points on the surface of a material. First, windowing is used to remove the Rayleigh wave, and after that, a new function called multicross-spectral density is defined applying cross-spectral density over the signals collected from points placed at several distances from the impact point. Two cases have been studied: calculation of the thickness of a concrete plate, and detection of a shallow crack (delamination) in the interface of a concrete slab with an asphalt overlay. The noise and the non-desired peaks in the frequency domain are reduced after the technique is applied facilitating the interpretation of results.