Generalized Conditional Mean Spectrum‐Based Ground Motion Records Selecting Considering Multiple Order Modes for High‐Rise Structure

ABSTRACT In the performance‐based earthquake engineering, the selection of ground motion (GM) records is reasonably of great importance for reliably estimating the response of the structure, and the conditional mean spectrum (CMS) provides an effective way for GM records selection because it can match the specific seismic hazard level of a given site. However, the spectral acceleration corresponding to the first mode of the structure is usually adopted by the traditional CMS‐based GM records selection method as the conditional intensity measure (IM), which may fail to consider the influence of multi‐order modes of the high‐rise structure and the period shift caused by structural nonlinearity. In addition, the severity of structural damage under earthquake excitation depends not only on the spectral characteristics but also on the amplitude and time domain of GM records, while the CMS is generally only used to characterize the GM records by matching the spectral acceleration. To solve the problems above, a new generalized conditional intensity measure (NGCIM) is proposed to select GM records by combining an average spectral acceleration ( AvgSa ) based‐CMS, duration, and amplitude parameters. Within the framework of the proposed method, the average of the spectral acceleration considering multi‐order modes and period shift of the structure is taken as the conditional IM. At the same time, the generalized GM models that can be used to predict the significant duration and amplitude at a specific seismic hazard level are incorporated so as to supply the construction of NGCIM. Finally, the Hyper‐Latin sampling is adopted in this paper to simulate the target IM distributions and thus to select GM records. To verify the effectiveness of the proposed NGCIM‐based GM records selection method, a super high‐rise structure is selected as an example to carry out the fragility analysis and risk assessment. The results show that compared with the traditional CMS‐based GM records selection method, the GM records selected by the proposed NGCIM have a small number and can produce the results of structural damage prediction with less uncertainty and higher reliability.