Influence of uncertainties on the seismic performance assessment of chevron braced frames

Abstract

In the framework of the probabilistic seismic response assessment of buildings, a key aspect is the evaluation of the influence of uncertainties on the mean annual frequency of exceedance of assigned limit states. In this regard, this study is focused on the seismic performance of concentrically braced frames in the chevron configuration. Out of the possible sources of uncertainties related to the numerical model (e.g. modelling of infills, soil-structure interaction), the effects of modelling of connections at the ends of braces are investigated by considering three different models: a model with pinned braces, a model with braces connected to beams and columns by means of rigid links simulating the gusset plates and a model with braces connected to beams and columns by means of elements with axial and flexural stiffnesses based on the real size of the gusset plates. In the two latter models out of plane buckling of braces is allowed. All the considered models are first validated against an experimental test and then used to estimate the fragility curves and the mean annual frequency of exceedance of Damage Limitation, Significant Damage and Near Collapse limit states of four case-study buildings. In addition to the uncertainty in the main characteristics of the seismic input, the uncertainties in the yield strength of material, equivalent viscous damping ratio, permanent and variable loads are also considered. The effects of these additional sources of uncertainty on the structural response are determined by comparing results provided by numerical models with random values of the uncertain variables with results provided by a single model with median values of the same variables. Finally, a simplified approach is proposed to simulate the effects of the additional uncertain variables in analyses of models with only variability in the seismic input.