Towards Realistic and Efficient Computational Fluid Dynamics Simulations for Urban Wind Applications and Pedestrian Wind Comfort Assessments

Abstract

Urban wind comfort and safety have become critical considerations in modern urban planning, particularly in response to increased urbanization and highrise buildings, and the potential impact of climate change. Urban development can significantly alter pedestrian-level wind conditions, creating safety and comfort issues imposed by architectural features that channel high-velocity winds. Computational fluid dynamics has emerged as a vital tool for predicting and analyzing complex wind flows and providing insight into the planning stage of building projects to improve wind comfort in the urban environment. However, challenges persist in accurately andefficientlymodeling wind in urban environments. This thesis seeks to address challenges related to pedestrian wind comfort. Specifically, it explores the influence of building geometric detail, urban morphology, and the number of simulated wind directions on modeled pedestrian wind comfort. The objective is to develop practical guidelines that can support architects and urban planners in designing safer, more comfortable public spaces. This thesis utilizes advanced computational fluid dynamics methodologies to enhance the accuracy and efficiency of urban wind simulations. High-fidelity geometric models were generated to represent the terrain, buildings, and vegetation layers within the simulated environments. The computational setup followed established best practices in urban wind simulation, including carefully defined domain sizing, grid generation, boundary conditions, and flow physics approximations to enhance accuracy and minimize errors. The incompressible, three-dimensional, steady Reynolds-averaged Navier-Stokes equations were solved using the finite volume method within the OpenFOAM toolbox, an open-source CFD software package. Both rectangular and cylindrical computational domains were utilized, with meshing strategies designed to accurately capture complex flow features around buildings and terrain. The findings from this research provide practical advancements in simulating pedestrian wind comfort, demonstrating methods that improve both predictive accuracy and computational efficiency in urban wind analysis. Findings show that simplified geometric building models, such as those derived from digital building data, are sufficient for accurately predicting pedestrian-level wind speeds and general flow patterns. Furthermore, highly detailed models, which are time-consuming to produce, may not be necessary for many practical applications, thereby streamlining the simulation process. Coarse, unstructured meshes can effectively capture overall flow features, allowing flexibility in simulation strategies while maintaining a high degree of accuracy. The research also assesses how the number of simulated wind directions affects pedestrian wind comfort maps. Using the 12 wind directions standard leads to 5 % – 10 % misclassification in wind comfort maps, and urbanization does not appear to dictate wind direction count for reliable assessments. To minimize misclassification risks, the study recommends using at least 8 wind directions for basic assessments, 24 for high-accuracy evaluations, and 36 for safety-focused studies. The thesis also validates its simulation methodologies against wind tunnel and field experiments, proving their reliability for various urban scenarios. Comparing real wind data ensures that the approaches developed can be widely applied in urban wind simulations. Future research could explore the development of machine learning techniques for more rapid prediction of wind conditions, studying the impact of geometric detail on other urban challenges, such as heat islands and pollution dispersion, and refining wind direction studies by incorporating site-specific atmospheric stability conditions. These directions can further enhance the accuracy, efficiency, and applicability of urban wind simulations.