Optimal design of dynamically constrained structures

Abstract

A modified iterated simulated annealing (MISA) method is proposed for optimal design of structural systems with dynamic constraints. The MISA method uses a random sequence of designs to find the optimal design. In performing the optimization, the MISA method employs two new features: the first is automatic reduction of the search region; and the second is sensitivity analysis of the design variables. The MISA method deviates from traditional annealing algorithms because of these two features and is considered to be a new method. A two-story frame structure with dynamic constraints was used in evaluating the performance of the MISA method. Classical optimal design methods were used to solve the same problem and the results were compared to MISA. The comparisons show that MISA is able to provide the global minimum even when infeasible initial designs are attempted. By contrast, some classical optimization methods either fail to converge or converge to local minima.