ANN-based optimization framework for the design of wind load resisting system of tall buildings. (15th June 2023)

• Record Type: Journal Article
• Authors: Alanani, Magdy; Elshaer, Ahmed
• Is Part Of: Engineering structures. Volume 285(2023)
• Publication Date: 2023-06-15
• Subjects: Structural optimization -- Topology optimization -- Tall buildings -- Wind load -- Genetic algorithm -- Neural networks; Structural engineering -- Periodicals; Structural analysis (Engineering) -- Periodicals; Construction, Technique de la -- Périodiques; Génie parasismique -- Périodiques; Pression du vent -- Périodiques; Earthquake engineering; Structural engineering; Wind-pressure; Periodicals; 624.105
• DOI: 10.1016/j.engstruct.2023.116032
• Languages: English
• ISSNs: 0141-0296

Abstract

Highlights: An automated Structural Wind Optimization Framework (SWOF) was developed. SWOF considered different wind load cases during the optimization process. Lateral Load Resisting System (LLRS) layout was optimized for wind load. The use of Artificial Neural Network (ANN) reduced the SWOF computational cost. Abstract: Conventional design methodology for tall buildings is a time-consuming and repetitive trial-and-error procedure with a limited probability of yielding an optimal solution that satisfies architectural, structural and serviceability requirements. Tall buildings are typically slender structures and mainly depend on lateral load resisting systems (LLRS) (e.g., shear walls, cores, and bracing systems) to withstand the lateral load of earthquakes and wind events, where a minor change in their layout, size, or shape will affect the cost tremendously. Consequently, an automated layout optimization procedure will result in a more economic and sustainable design. This paper develops a novel structure-wind optimization framework (SWOF) to find the optimal shear wall layout of tall buildings subjected to wind loads. SWOF is considered a genetic algorithmbased framework that uses an Artificial Neural Network (ANN) surrogate model to evaluate its constraints and objective function. These surrogate models rely on a training dataset prepared using the Finite Element Method (FEM), which has been created using an open application program interface (OAPI) MATLAB code. An optimization problem; is presented to show SWOF's efficiency. SWOF showed significant capabilities in recognizing load direction, critical load cases and inertia concepts without explicitly defining them through the developed code.