Epistemic uncertainty in the seismic performance assessment of isolated buildings due to high damping rubber bearing model selection

Abstract

Seismic isolation has gained popularity for its proven effectiveness in protecting buildings and their contents during earthquakes. The mechanical filtering produced by the isolation system-which in turn depends on the force-deformation constitutive behavior of the seismic isolators-controls the seismic demand on the superstructure and its frequency content. Consequently, numerical models that accurately characterize this behavior are critical for predicting the seismic response of isolated buildings. High Damping Rubber Bearings (HDRBs) are widely used in practice, but their highly nonlinear behavior is challenging to model. As a result, several numerical models have been developed to simulate their behavior, each based on different assumptions. Currently, model selection is often guided by professional judgment and, more importantly, by the availability of models in commercial software. However, the impact of this choice on the performance assessments of buildings has not been systematically quantified. To address this gap, this study first evaluates the accuracy of eight HDRB models in reproducing experimental responses of a single device, and then propagates the resulting epistemic uncertainty into the seismic response of a benchmark isolated structure under ground motions from two seismic sources: subduction zones and crustal faults. The selected models represent formulations with varying levels of detailing and account for different phenomena, including stiffness degradation at large deformations, strain rate-dependency and axial-shear coupling. To minimize any potential bias, all model parameters were calibrated against experimental data using an optimization algorithm. Results show significant differences between models in predicting isolation system shear forces and superstructure demands, though most models estimate similar peak lateral displacements of the isolation system. More elaborate models predict significantly larger demands, an effect that is more significant for ground motions from crustal earthquakes. These increased demands potentially influence the fragility estimates of the structure and could change its failure mechanisms, highlighting the critical importance of model selection for reliable seismic performance and risk assessments of isolated buildings.