Optimal intensity measure selection in incremental dynamic analysis: methodology improvements and application to a high-speed railway bridge

This paper proposes a modified method to select the optimal intensity measure (IM) in the seismic fragility analysis by the incremental dynamic analysis (IDA). The performance of an IM is evaluated by its efficiency, the efforts on fragility estimation, and sufficiency, the ground motion information represented by the IM. The results by IDA in various IMs are generally obtained by the conversion from that with the scaling factor IM using a given selected ground motions suite. The present study shows that the inter-IM conversion leads to an incompatibility of IM value range at different IDA curves, hindering the normalization for eliminating the difference of unit and range in various IMs. To tackle this issue, a normalization procedure is developed specifically for IDA, employed in the definition of a normalized and relative efficiency metric. Regarding the sufficiency of IM, the information-theory-based metric, the approximate relative sufficiency measure (aRSM), is adopted. It is calculated by the EDP-based approach, consistent with its original definition, rather than the IM-based approach in the literature. By the Bayes' theorem, the prerequisite to neglecting this inconsistence of the IM-based aRSM calculation is derived. It is the approximated equality in the occurrence probabilities of the two compared IMs. A case study is performed in a typical simply-supported high-speed railway bridge and track system to explore the IM performance in different structural components by local EDPs. The feasibility and validity of the proposed method is demonstrated, particularly for the normalization procedure and the EDP-based aRSM calculation.