Performance-Based Design Optimization of Steel Braced Frame Using an Efficient Discrete Algorithm

Performance-based design optimization (PBDO) of steel braced frames (SBF) is a computationally intensive task, especially when nonlinear time history analysis is applied. In this paper, an efficient discrete optimization algorithm is proposed for PBDO of SBF utilizing a deformation-based method. Two difficulties exist in PBDO of SBF, and the first one is that multiple performance constraints are imposed on the structures. To avoid tackling all constraints simultaneously, a strategy is proposed in which the deformation constraints of beams and columns are strictly followed throughout the optimization process and the brace deformation constraints are checked again at the end of the optimization. The second difficulty is that the search for an optimum design is conducted in a discrete design space since the structural elements are usually taken from standard sections. A common practice is to use regression models for the sections, at the expense of removing sections that cannot fit into the regression models from the design space. In this paper all standard sections are preserved by using the cross-sectional area (Area) and moment of inertia (Ix) as the design variables, thus any standard section can be uniquely defined by its Area and Ix. To investigate the effectiveness of the proposed algorithm, three numerical examples are presented. Compared to the results achieved by genetic algorithm (GA) and differential evolution (DE), the proposed algorithm can achieve better or comparable structural designs. Furthermore, the convergence rate of the proposed algorithm is much higher than GA and DE, proving that the proposed algorithm is an efficient optimization method for PBDO of SBF.