Evaluation of fema p-58 overturning fragility curves for freestanding building components

Next generation performance based seismic design procedures have been recently presented in FEMA P-58, which allow to estimate the seismic risk of buildings by explicitly accounting for the contribution of both structural and nonstructural damage. Within this proposed framework for risk analysis, the vulnerability of nonstructural components is expressed in probabilistic terms by means of fragility curves. In the case of unanchored components, sliding and overturning fragilities, adapted from ASCE/SEI 43-05, are provided. Fragility parameters include the median peak total floor velocity at which the limit state initiates, and a dispersion reflecting the uncertainty of the prediction. In this study, the fragility function for overturning proposed by FEMA P-58 is evaluated through comparisons with curves obtained from numerical simulations on a set of rigid blocks characterized by different geometries. For each block the fragility is derived from an incremental dynamic analysis that uses a bin of records to represent the ground motion variability. In order to determine the rocking response of the block, a nonlinear model that concentrates energy dissipation at the instant of impact is adopted. The assumed restraint condition allows for two-sided rocking with no sliding. The filtering effect of a structure that supports the block is investigated by analyzing a case study building consisting of a six-story reinforced concrete frame. Dynamic interaction between the structure and the block is neglected, and fragility is calculated from the response of the latter to the floor acceleration histories derived by exciting the frame with the assumed bin of records. The obtained results show that FEMA P-58 largely overestimates the dispersion of the overturning capacity, that the floor acceleration spectral intensity is equally efficient but easier to estimate, and that the building filtering effect cannot be ignored because significantly reduces the scatter of the block response.