2nd Croatian Conference on Earthquake Engineering 2CroCEE

Since the prediction of the seismic response of structures is subject to large uncertainties, the need for a probabilistic approach is obvious, especially for estimating the critical parameters of the seismic response. Considering the uncertainties that exist in the material and geometric form of reinforced concrete structures (RC), reliability analyses using the finite element method (FEM) have been carried out in the context of Performance-Based Earthquake Engineering (PBEE).
In this study, the possibilities of non-linear modelling of the plane reinforced concrete frame (RC) and its reliability analysis using different numerical methods were presented and compared: Mean-Value First-Order Second-Moment (MVFOSM), First-Order Reliability Method (FORM), Second-Order Reliability Method (SORM) and Monte Carlo Simulation (MCS). The calibrated numerical models used were based on a previous experimental test of a planar RC frame subjected to cyclic horizontal loading. For the reliability analysis, parameters with random variables (RV) were added to the numerical models. Pushover analyses were performed using the implicit limit state function (LSF) by controlling the horizontal interstory drift ratio (IDR).
It was found that the reliability analysis results are sensitive to the reliability analysis method. The reliability analysis results show that in a nonlinear region after exceeding the yield strength of the longitudinal reinforcement, the parameters of the section geometry were more important compared to the parameters of the material properties. The results also show that epistemic (knowledge-based) uncertainties significantly affect the dispersion and response of the median of the estimated parameters. The MCS sampling method is recommended, but the first-order reliability method (FORM) applied to a response model can also be used with good accuracy. It has been shown that reliability analysis using FEM can directly implement geometric and material nonlinearities to cover epistemic (knowledge-based) uncertainties.