Surrogate-based design optimization of steel plate shear walls

Steel plate shear walls (SPSWs) are frequently utilized as a lateral load resisting system in high rise buildings for resisting wind and earthquake loading. The initial stiffness, lateral load-bearing capacity, and seismic energy dissipation capacity of SPSWs are primarily provided through the yielding of the infill plates attached to the boundary elements of the SPSW system. A detailed analysis of such systems requires the use of highly expensive computer simulations that lead to fundamental problems when trying to compare the responses of all combinations of design variables. This problem becomes very acute when using optimization schemes. This paper presents a surrogate-based optimization approach for achieving the best infill plate parameters in the design of SPSW under lateral loading conditions. The surrogate models have been developed using Gaussian and Moving Least Squares (MLS) methods. The efficacy of the developed models is checked through 150 full finite elements (FE) simulations of steel plate shear walls under different configurations. The optimization of SPSWs has been carried by the Mixed Integer Non-Linear Programming Approach (MINLP). It performs the discrete optimization of decisions simultaneously with the continuous optimization of parameters. Three independent design variables of SPSW, namely thickness, yield strength, and Young’s modulus, have been considered as input variables. The objective function of the optimization is the minimization of the structural mass of the steel infill plate (directly related to cost function) while satisfying the design constraints placed upon initial stiffness, yield point, ultimate strength, and ductility to obtain a global minimized solution. It is observed that the models with thin infill plate and nominal yield strength resulted in the minimization of the objective function. Finally, the optimal infill plate design variables for a target (design) performance is identified as the one with the minimum expected cost.