Identification of damage-induced frequency decay on a large-scale model bridge

Planning the maintenance activity of a civil structure with a life cycle cost-effectiveness requires the evaluation of its functionality in service to identify possible damage states which need a retrofit, avoiding thus danger for the users. In this paper, this topic is explored by using an experimental 1:4 scale model bridge, in which some damage states on the main steel beams are introduced arbitrarily with the aim of detecting the effect of local cracks on the performance of the structure. By performing dynamic tests on the model bridge, it was possible to link the variability of the natural frequencies with the gradual stiffness reduction caused by the cracks. By processing the acceleration signals due to a random excitation of the 6.0 × 3.0 m2 deck, the main natural frequencies of the bridge have been extracted in several progressive damage states. The observed behaviour was explained by making use of a detailed Finite Elements numerical model. The frequency decay predicted by these methods is in close agreement with the experimental observations and permits to build a useful link between frequency data and damage detection.