Dynamic reliability analysis for nonlinear structures using stochastic finite element method

Abstract

An efficient and accurate algorithm is developed to evaluate reliability in the time domain for nonlinear structures subjected to short duration dynamic loadings, including earthquake loading. The algorithm is based on the nonlinear stochastic finite element method (SFEM). Uncertainties in the dynamic and seismic excitation, and resistance-related parameters are incorporated by modeling them as realistically as possible. The uncertainty in them is explicitly addressed. The proposed algorithm intelligently integrates the concepts of response surface method (RSM), finite element method (FEM), first-order reliability method (FORM), and an iterative linear interpolation scheme. This leads to the stochastic finite element concept. It has the potential to estimate the risk associated with any linear or nonlinear structure that can be represented by a finite element algorithm subjected to seismic loading or any short duration dynamic loadings. In the context of the finite element method, the assumed stress-based finite element algorithm is used to increase its efficiency. Two iterative response surface schemes consisting of second order polynomials (with and without cross terms) are proposed. A mixture of saturated and central composite designs is used to assure both efficiency and accuracy of the algorithm. Sensitivity analysis is used to improve the efficiency further. The unique feature of the algorithm is that it is capable of calculating risk using both serviceability and strength limit states and actual earthquake loading time histories can be used to excite structures, enabling a realistic representation of the loading condition. The uncertainty in the amplitude of the earthquake is successfully considered in the context of RSM. Uncertainty in the frequency content of an earthquake is considered indirectly by conducting a parametric study to quantify the effect of uncertainty in the frequency content of earthquakes on the overall reliability of structures. The algorithm has been extensively verified using the Monte Carlo simulation technique. The verified algorithm is used to study the reliability of structures excited by actual earthquake time histories. The results of the numerical examples show that the proposed algorithm can be used accurately and efficiently to estimate the risk for nonlinear structures subjected to short duration time-variant loadings including seismic loading.