Seismic reliability of 3D base-isolated structures through FPS with elastic superstructure

The friction pendulum system (FPS) is becoming a widely used technique for seismic protection and retrofit of buildings, bridges and industrial structures. Experimental data also showed that the coefficient of friction depends on several effects (i.e., sliding velocity, apparent pressure, air temperature, cycling effect) so that it can be assumed as a random variable. The aim of the study consists in evaluating the seismic reliability of a 3D base-isolated structure with FP isolators considering both isolator properties (i.e., coefficient of friction) and earthquake main characteristics as random variables. Assuming appropriate density probability functions for each random variable and adopting the Latin Hypercube Sampling (LHS) method for random sampling, the input data set has been defined. Several 3D non-linear dynamic analyses have been performed in order to evaluate the system response. In particular, monovariate and multivariate (joint) probability density and cumulative distribution functions have been computed and, considering the limit state thresholds and domains (performance objectives) assumed as earthquake damage parameter (EDP) according to performance-based seismic design, the exceeding probabilities (structural performances) have been evaluated. Estimating the reliability of the superstructure, substructure and isolation level led to define and propose reliability-based abacus and equations useful to design the FP system.