Simplified method for seismic risk assessment of buildings with consideration of aleatory and epistemic uncertainty

Abstract

A simplified method for seismic risk assessment with consideration of aleatory and epistemic uncertainties is proposed based on the widely used closed-form solution for estimating the mean annual frequency of exceeding a limit state (LS). The method for the determination of fragility parameters involves a non-linear static analysis of a set of structural models, which is defined by utilising Latin hypercube sampling, and non-linear dynamic analyses of equivalent single degree-of-freedom models. The set of structural models captures the epistemic uncertainties, whereas the aleatory uncertainty due to the random nature of the ground motion is, as usual, simulated by a set of ground motion records. Although the method is very simple to implement, it goes beyond the widely used assumption of independent effects due to aleatory and epistemic uncertainty. Thus, epistemic uncertainty has a potential influence on both fragility parameters, and not only on dispersion, as has been assumed in some other approximate methods. The proposed method is applied to an example of a four-storey reinforced concrete building, where it is shown that the effects of epistemic uncertainties, in addition to aleatory uncertainty, increase with the severity of the LS, so that, for the near collapse LS, the risk with consideration of both sources of uncertainty is more than double if compared to the risk, which was determined solely by the consideration of aleatory uncertainty.