Performance-based Bi-objective optimization of structural systems subject to stochastic wind excitation

• Published in: Mechanical systems and signal processing Vol. 160; p. 107893
• Main Authors: Subgranon, Arthriya, Spence, Seymour M.J.
• Format: Journal Article
• Language: English
• Published: Berlin Elsevier Ltd 01.11.2021; Elsevier BV
• Subjects: Bi-objective optimization; Constraint modelling; Design optimization; Fragility; High-dimensional problems; Mathematical analysis; Performance-based design; Simulation; Statistical analysis; Statistical methods; Stochastic wind loads; System-level loss assessment; Wind effects; Wind engineering; Wind loads; Stochastic wind loads; Performance-based design; Bi-objective optimization; Wind engineering; High-dimensional problems; System-level loss assessment
• ISSN: 0888-3270; 1096-1216
• DOI: 10.1016/j.ymssp.2021.107893

•A bi-objective design optimization scheme for wind excited structures is outlined.•A system-level loss model is proposed for buildings subject to stochastic wind loads.•Closed form expressions are derived for treating inter-component correlations.•Cases studies are presented illustrating the potential of the proposed approach.•The importance of modeling inter-component correlations is highlighted. This paper outlines the development of a stochastic simulation-based design optimization approach for dynamic wind excited structures in which correlations between component damages and losses are explicitly treated. The proposed approach integrates a bi-objective design optimization scheme with a probabilistic performance-based wind engineering methodology which systematically accounts for the various sources of uncertainties involved in system loss estimation. Through the ∊-constraint technique, the bi-objective optimization problem is transformed into a series of single-objective stochastic optimization problems. To solve each ∊-constraint optimization problem, a pseudo-simulation scheme is proposed that allows for the formulation of an approximate sub-problem that can be solved sequentially to identify solutions that define a set of Pareto optimal designs. In the proposed scheme, samples of engineering demands are approximated in terms of auxiliary variable vectors, which are by-products of an augmented simulation carried out in a fixed design point. Analytical expressions are derived that relate the engineering demand samples to the second-order statistics of wind-induced losses based on the concept of fragility. Potential correlations between the component capacities and component losses are explicitly treated. The effectiveness of the proposed approach and its scalability to high-dimensional problems are illustrated through optimal designs of moment-resisting frames subject to stochastic wind loads.