A Maximum-Entropy-based Multivariate Seismic Vulnerability Analysis Method for Power Facilities: A Case Study on a ±1100-Kv Dry Type Smoothing Reactor
Jiawei Cui,Ailan Che,Sheng Li,Yongfeng Cheng
DOI: https://doi.org/10.1016/j.engfailanal.2022.106740
IF: 4
2022-01-01
Engineering Failure Analysis
Abstract:Strong earthquakes can damage power facilities, resulting in power outages and major economic losses. Seismic vulnerability analysis is a common method to assess the seismic performance and risk of power facilities. However, this method uses only one seismic intensity measurement (IM), such as peak ground acceleration (PGA), and assumes that structural responses obey the log-normal distribution, which may ignore both the uncertainty of seismic excitation and the structural response. In this study, we developed a composite method of the maximum entropy method (MEM), multivariate, correlation analysis, and reliability theory for the seismic vulnerability analysis of power facilities, and evaluated a ±1100-kV dry type smoothing reactor (UHVDTSR), an important power facility in the ultrahigh-voltage (UHV) converter station. Using the proposed method, we determined the probability density function (PDF) of the seismic responses through statistical analysis rather than artificially assuming the distribution type. Furthermore, more IMs could be introduced to consider the uncertainty of seismic excitation. Shaking table tests were performed and Pearson and partial correlation analyses were used to optimize a reasonable vector-(IM1, IM2). The results indicate that the proposed method not only provides the probabilities of seismic damage states of the UHVDTSR based on a vector-(IM1, IM2), but also demonstrates greater efficiency and accuracy than classical methods.