Evaluation of efficacy of intensity measures for fast and reliable seismic fragility analysis of bridge columns

Intensity measure (IM) is a crucial factor of probability seismic performance assessment, particularly in the development of fragility curves. This study establishes an evaluation framework to assess the efficacy of IMs, defined as the ability to enable fast and reliable fragility analysis. A k-means clustering algorithm, using various IMs as the clustering features, is employed to select representative ground motions for the fragility analysis of two bridge columns. Kullback-Leibler (KL) divergence is used to quantify the differences between the fragility curves obtained from different IMs or subsets of ground motions and a benchmark fragility curve generated using 1736 natural ground motions to assess the efficacy of each IM in fast and reliably compute fragility curves of bridge column. The superiority of the geometric mean of acceleration response spectrum ( Sagm ) as an IM from the perspective of least-squares sense is demonstrated, and it can represent the effects of different ground motions on structural seismic response. The results indicate that Sagm exhibits a high efficacy in fragility curve computation of bridge column. Further, the optimal period range for Sagm with best efficacy is 0.2T 1 -2.5T 1 (T 1 is the fundamental period of the structure). • From the perspective of least squares, Sagm may be a relatively superior IM. Combined with the elastic response spectrum, Sagm can then represent the impact of different ground motions on structural response. • An evaluation framework of IM was constructed based on fragility curves that focuses on the efficacy of IM, defined as the ability to conduct fragility curve fast and reliably. Sagm is an efficacy IM and can be used as the clustering feather of k-means clustering for ground motion selection to achieve fast and reliable fragility analysis of column. • The number of ground motions required for reliable fragility analysis may vary for different columns. In this study, the impact of the number of ground motions on the reliability of fragility curve with Sagm as IM was relatively small when the number of ground motions exceeds 120. • The acceleration response spectrum may not be suitable as a clustering feature in k-means clustering for ground motion selection to fragility analysis column. • The upper period limit has a greater impact on the efficacy of Sagm compared to the lower period limit, and the range of 0.2T 1 to 2.5T 1 may be a suitable period range.