EXPERIMENTAL ASSESSMENT OF VIBRATION-BASED METHODS FOR DAMAGE LOCALIZATION

This paper reports results relevant to the experimental comparison of a number of vibration-based methods for damage localization applied using responses recorded during forced vibration tests on a composite beam. Damage is intended herein as a stiffness change. Four methods from the literature were compared: two model-based methods, namely the direct stiffness determination method and the model updating method and two response-driven methods, namely the modal interpolation error method and the modal strain energy method. The tested specimen is a composite beam made of timber and plaster boards, with a total span of 4.0 m and simply supported at the ends. During the forced vibration tests responses in terms of acceleration were measured at 24 points along the beam axis. Responses were recorded in both the reference states and in several modified states in which local variations of bending and shear stiffness were simulated adding a stiffener or unscrewing a number of bolts. Using the accelerometric responses the comparison of the four considered methods was performed in terms of the accuracy of damage localization.