Performance-based design optimization of uncertain wind excited systems under system-level loss constraints

The aim of this paper is to define an efficient optimization methodology for the identification of minimum cost wind excited structural systems that satisfy system-level targets on the annual exceedance rate of expected losses. The main challenge to this vision is the need to integrate optimization models for solving design problems involving hundreds of free parameters with performance assessment models that require stochastic simulation over high-dimensional uncertain spaces. To overcome this challenge, a framework is proposed in this work that is based on decoupling the performance assessment model from the optimization problem. This is accomplished through the definition of a high-quality approximate optimization sub-problem that is scalable to high-dimensional design and uncertain spaces. By solving a limited series of these sub-problems, each formulated in the solution of the previous, a sequential optimization strategy is defined for identifying solutions to the original stochastic optimization problem. The efficiency of the method lies in the limited computational effort required to solve each sub-problem. To demonstrate the applicability and the scalability of the proposed framework, two examples are presented that involve the optimal design of the structural system of buildings subject to stochastic wind excitation and system-level constraints on the annual exceedance rate of expected losses.