The Tap-scan Damage Detection Method For Bridge Structures

Damages are quite common in bridge structures, especially in the concrete ones. They will also develop as time pass by, and lead to terrifying disasters. With the increasing number in bridge collapse accidents, developing a damage identification method which is reliable and easy to use is essential to the safety of large scale bridges. In this thesis, a new damage identification method, called tap-scan method, is proposed, which is characterized with high sensitivity to small defects and strong anti noise capability, together with the capability to identify damage under the interface of external traffic flow.First of all, the theoretical dynamic response of a vehicle passing through a bridge with external exciting force is obtained. This result is verified by numerical simulations. It is confirmed that the damage information does exist in the response of the vehicle, hidden in the signal of acceleration signal, which provides theoretical basis for this tap-scan method. Following the sensitivity analysis, the rules for selecting the exciting force is also determined, i.e. the exciting frequencies should be near the bridge fundamental frequencies.Secondly, the whole identification process is simulated numerically. The signal of the vehicle passing a damaged bridge is obtained, and certain pattern recognition process is applied to identify the damage location. The result of sensitivity analysis is also verified by comparing the identification results from different system parameters, and it provides plans for the following experiments. In some cases, the angular acceleration signal is more sensitive to damages compared to the translational one, which has been demonstrated in the simulation of a 3D shell bridge model. To obtain better identification results, one can consider fusing the identification results from different signals.Finally, the tap-scan method is verified on small scale model experiments. On the wooden plate model, this method can locate the saw damage and the bridge pier at the same time. In this case, the pier is simply divided into the damage class, one can get the proper damage identification result after removing the effects of the piers. Using more complex pattern reorganization might leads to better results, telling out the differences between the damage and the pier directly. With another concrete box beam model, different cases are tested such as single damage, multiple damages on the bridge, together with extra traffic noise interference. The pavement overlay is also simulated by adding a rubber layer on the surface of the concrete beam. These experiments demonstrate the feasibility of the method on a real bridge, and will provide valuable experience for full scale tests.