Identification of damage parameters during flood events applicable to multi-span bridges

Abstract

During flood events, the dynamic interaction between flowing water and bridges generates random loadings that force bridges to vibrate in all six degrees of freedom. It is difficult for a structural damage detection method to select a degree of freedom, or damage feature, to accurately describe and predict damage. The methodology presented here identifies damage-sensitive features and uses them to monitor bridge health. A small-scale physical model of a multi-span highway bridge was constructed to satisfy geometrical, Cauchy, and Froude similarities, and six-dimensional hydrodynamic forces induced by simulated flood events were investigated as an input excitation in a tilting flume. It was determined that pitch, roll, and surge motions can be used as damage features during the inundated stage, while pitch, roll, surge, and heave can be used before the inundated stage. In addition, angular velocity signals exhibited more consistent damage indices than acceleration. Using the damage features, the proposed algorithm could successfully detect damage and damage severity during simulated flood stages. Identifying damage features can reduce the size of the collected data and inform emergency responders’ decisions. This case study can be used to test methods at full scale on similar structures to develop automated health-monitoring systems.