ROBUST OPTIMIZATION OF LINEAR DYNAMIC SYSTEM WITH RANDOM PARAMETERS UNDER STOCHASTIC EARTHQUAKE EXCITATION

Abstract

The structural optimization under random load is dealt in the form of non-linear optimization with stochastic performance measures. The analysis considers the effect of randomness in the earthquake load only, considering other parameters as deterministic. But uncertainty is inevitable in any structural system. The primary objective of present study is to propose a robust optimization strategy in the framework of adaptive response surface method for linear dynamic system characterized by parameter uncertainties subjected to stochastic earthquake load. Essentially, the proposed formulation becomes two-criterion equivalent deterministic optimization problem, where the weighted sum of the mean and variance of desired objective is optimized. The associated stochastic constraints are derived by imposing a limit on failure probability. To avoid repeated evaluations of complex dynamic responses and their sensitivities, the adaptive response surface based approximation method is used to obtain the stochastic constraint of the related optimization problem. The formulation is elucidated by optimizing a three-storied concrete frame. The numerical results are presented to study the differences in optimum solutions obtained by the conventional and adaptive response surface method. The robust optimization results are compared with the results of stochastic optimization with deterministic system parameters to study the effect of parameter uncertainty.