Structural health monitoring methods for bridges using ambient vibration and impedance measurements

Abstract

This thesis presents two gobal damage detection methods using modal flexibility obtained from ambient acceleration measurements, and a local damage detection method using impednace measurements under temperature variations. Firstly, a vibration-based damage detection method is proposed based on the modal flexibility matrix from ambient vibration measurements for reliable and practical structural health monitoring. The proposed method has three distinct advantages over the convensional methods: (1) the explicit inverse mapping from the damage-induced deflection to the damage location, (2) no requrement for the curvature calculation weak to noises, and (3) no requirement for the intact FEM model. The proposed method utilizes the modal flexibility matrices to estimate deflections under a positive bending inspection load (PBIL) which has been devised to simplify the damage detection procedure in this study. Then, damage can be localized by using the damage-induced deflection and the explicit inverse mapping from the damge-induced deflection to the damage location.Through a series of numerical and experimental studies on continuous beam systems and steel box-girder bridge models with a single and two spans, it has beenfound that the damage locations can be successfully localized by the proposed method. The damage detection performance was found to be superier than the modal strain energy-based damage index method under measurement noisy conditions. Secondly, a substructural damage detection method which requires acceleration measurements within a substructure, is proposed by extending the basic theory of the modal flexibility-based damage detection method to substructural damage detection problems. Novel concepts of the pseudo-modal flexibility and pseudo-modal deflection are introduced for damage detection on a substructure. It has been also found that the damage locations in the two-span steel box-girder bridge model were successfully localized by the pro...