Damage diagnosis in bridge structures using rotation influence line: Validation on a cable-stayed bridge

This study proposes a new damage identification technique for condition assessment of bridge structures. The method is based on the concept of rotation influence line (RIL) at the bridge bearing locations, and solely relies on measurements obtained from two points at either end of the bridge e.g., RILR and RILL. The sensitivity of the rotation measurement to damage is first investigated using a 1-D simply supported beam model and it is demonstrated that unlike conventional strain measurements, the rotation measurement is capable of providing useful information about damage even though it is far from the measurement point. Further, an existing cable stayed bridge is considered to validate the capability of RIL in damage identification. A comprehensive three-dimensional finite element model (FEM) of the bridge is established and calibrated using the measured static and dynamic responses. Numerous field tests are conducted on this large-scale structure to extract static and dynamic characteristics, including natural frequencies, mode shapes and influence lines (ILs). Sixteen hypothetical damage scenarios are induced in the numerical model, including symmetric and asymmetric cases of cable loss. A damage index based on the normalised discrepancy of RIL between the benchmark state and an unknown state is introduced and through extensive investigations it is demonstrated that regardless of damage location, either RILR or RILL (or both) can successfully identify the induced damage in all of the sixteen damage scenarios. In contrast, the success of strain-based measurement is highly dependent on the closeness of damage to the sensor location, thus a much higher number of strain gauge sensors is required not to misidentify damage. The contribution of this work is four-fold. First, a novel and robust damage identification technique based on RIL is proposed which has not been reported in the literature. Second, the method solely relies on two measurement points e.g., two tilt meters at either end of the bridge and it is capable of identifying damage, even far from the sensor location. Third, the validation of the technique is demonstrated through extensive numerical and field test investigations on a statically indeterminate cable-stayed bridge structure. Finally, it is demonstrated that the conventional strain-based measurement is very likely to misidentify cable damage even with extreme case of cable loss.