Probabilistic floor spectra for fully self-centering structures with flag-shaped hysteretic behavior

Self-centering structures are emerging structural systems developed for enhancing the seismic resilience of building structures by minimizing residual displacements. Various investigations were conducted to study the design methods and seismic performance of self-centering structural systems, but the research on nonstructural performance in self-centering structures is limited. This paper aims to develop the probabilistic floor spectra that can be used for designing the acceleration-sensitive nonstructural components in self-centering building structures under nearfault ground motions. To this end, comprehensive dynamic analyses were conducted to obtain the floor spectra. 320 near-fault ground motions were selected for considering the seismic excitation's uncertainty. The Anderson-Darling test was adopted to estimate the probabilistic distribution of the floor spectra. The test results indicate that the floor spectra follow the logarithmic normal distribution under earthquakes. The effects of the nonlinear properties of primary self-centering structures (including the initial period, hysteretic features, and structural damping) and the damping of the nonstructural components on the median and dispersion of floor spectra were discussed based on the parametric dynamic analysis results. The prediction model of the probability density function of the probabilistic floor spectra for self-centering structures was developed based on the artificial neural network algorithm. The software FloorSpectraNet was developed to facilitate practical applications.