Optimization of Building Wind Environment Using CFD and Genetic Algorithm

The architectural and environmental characteristics of urban areas significantly impact human comfort, safety, and structural integrity. Hence, they constitute essential considerations during the initial planning and design phases of urban architecture. Although the Computational Fluid Dynamics (CFD) simulations offer several advantages over atmospheric boundary layer wind tunnel (ABLWT) experiments, such as controllable boundary conditions and the ability to conduct full-scale simulations, the extensive array of cases often renders the CFD approach cost-prohibitive, particularly in optimization contexts. To address this challenge and achieve fast optimization of building wind environment, this study integrates CFD simulations with artificial neural networks (ANN) and genetic algorithms (GA). It conducts a simulation-based investigation of architectural wind patterns of two adjacent buildings for varying building heights, spacing and the inflow wind direction. By evaluating two metrics: the area of uncomfortable wind speed regions for pedestrians and the area of unacceptable wind speed regions, the building wind environment under the building layout and height is assessed. The findings indicate that this approach effectively economizes the optimization process, thereby saving costs, and computational resources and the optimal building layout and height is determined.