Seismic Fragility Evaluation Of P.C. And Steel-Concrete Bridges In Permafrost Regions

This paper focused on the seismic fragility comparison between Prestressed Concrete (P.C.) bridges and steel-concrete bridges in permafrost regions. Incremental Dynamic Analysis (IDA) and Latin Hypercube Sampling (LHS) were used to simulate the uncertainty. The seismic fragility curves were obtained, in which machine learning approaches and different seasons were also involved. The results show that the vulnerable component rank in the frozen ground in descending order are: the bottom of pier, the top of pier, the exterior shear key, and finally the rubber bearing, and it applies to both P.C. and steel-concrete bridges. The seismic fragility of bridges in permafrost regions is significantly different in winter and summer. The damage probability of bridge piers in winter is significantly higher than in summer. However, the damage probability of bearings and exterior shear keys are lower. The improvement of the fragility of the pier increases the demand for a higher performance bearing and exterior shear key of the steel-concrete composite bridge. The seismic vulnerability predictions by the machine learning approach are stable. The BPANN method is more reliable than LSSVM in terms of errors and performances.