Theory and experimental verification of a resultant response-based method for assessing the critical seismic excitation direction of curved bridges

Previous studies have shown that the seismic incidence angle imposes a non-negligible impact on the seismic performance of curved bridges. The computational efficiency of some current methods for determining the critical angle needs to be improved and their applicability in practical engineering projects remains to be examined. For this reason, a resultant response-based (RRB) method is developed herein for assessing the critical excitation direction of curved bridges. To validate the feasibility of this method in an actual seismic design context, a 1/62.5-scale model of a three-span curved bridge is designed and a multi-angle shaking table test is implemented. Meanwhile, the finite-element model of the test specimen is set up, and the RRB method as well as the linear response-history analysis (LRHA) are comparatively assessed. The results indicate that the RRB method can capture the critical excitation direction of curved bridges with sufficient precision (error does not exceed 10% compared to LRHA). The associated computational effort is also substantially reduced given that RRB requires analysis solely along two orthogonal directions as the incidence angles, compared to standard response history analyses where ground motion excitation is applied at multiple ground motion orientations. The above observation is further verified by a well-designed experimental campaign, which demonstrates the accuracy and practicability of the RRB method for the case of realistic bridge configurations.