Refined seismic fragility curves of substation equipment considering ground motion classifications

Abstract

Seismic fragility of substation equipment is vital in risk analyses, as well as post-earthquake prediction of failure probabilities of multiple equipment for deciding the inspection order. However, the existing fragility curves of equipment are commonly generated without adequately considering intensity measures (IMs) of ground motions. During the post-earthquake use, they are deemed applicable to any earthquake occurring in the region, which is insufficiently precise. Thus, this paper employs a new method to derive more refined fragility curves for substation equipment based on ground motion classifications. It uses principal component analysis to convert IMs into a small number of principal components (PCs). Taking the PCs as indices, the selected ground motions are clearly categorized into distinct classes using the K-means clustering algorithm. Then, a simulation model of equipment is developed and the time-history dynamic analyses are performed to calculate seismic responses under different classes of ground motions. Finally, the refined fragility curves are derived via multiple stripe analysis for each ground motion class. A 500 kV power transformer is used as a case to implement the method. The results show that a small number of PCs can effectively reflect the multiple ground motion characteristics. Different ground motion classes have distinct distributions of IMs and PCs. There are notable discrepancies in the fragility curves of the power transformer when subjected to different ground motion classes. The significant disparities between the refined and original unclassified fragility curves highlight the importance of considering ground motion classifications.