On the Use of Sound Spectral Analysis for the In Situ Assessment of Structural Timber

This paper focuses on the possibility of using sound spectral analysis as a diagnostic technique for the assessment of existing timber structures. The basic idea is to perform a spectral analysis, through the Fast Fourier Transform (FFT) algorithm, of the sound (recorded by a smartphone microphone) that is produced by hitting the wood surface of the element with a hammer or an equivalent tool. The sound spectrum is compared to a reference spectrum obtained for sound wood in standard environmental conditions in order to detect the presence of damage or decay. Differently from "classic" stress wave propagation methods that are based on the determination of the time-of-flight of the stress wave, the proposed technique relies on the local acoustic properties of wood. A first insight on the method applicability was obtained by testing the technique on several timber specimens in different conditions, from sound to completely decayed. Registration of the signals was performed by using two different mid-market smartphone models. Moment analysis, by calculating the spectral and frequency energy centroid of the sound spectra, was used to evaluate and compare numerous experimental recordings. The parameters that have the most significant impact on the sound signal (e.g. impact tool, microphone distance, boundary conditions, etc.) and that can negatively influence the assessment were then identified and investigated. Finally, pros and cons of the presented method and future developments are examined and discussed.