Evaluation of the CFD simulation method for wind prediction in complex urban spatial forms

Urban spatial form plays a crucial role in shaping ventilation conditions within cities. Wind environment predictions are commonly conducted during architectural and planning design to optimize design schemes based on urban spatial forms. Computational Fluid Dynamics (CFD) is widely used in both research and engineering for wind environment predictions. Although many studies have evaluated the reliability of CFD simulations, research examining this reliability in relation to urban spatial forms is lacking. This study classified urban spaces into main streets, inner streets, and non-street spaces. The reliability of CFD simulations was assessed by comparing results across different urban spaces with wind tunnel tests, focusing on morphological characteristics. The results show that the standard k-epsilon model captured overall wind patterns well, but the consistency of CFD predictions varied based on factors such as the angle between streets and wind direction, changes in street width, and height differences between buildings. The CFD showed higher prediction consistency in street spaces when the streets were aligned with the wind direction. When the street width varies from narrow to wide, underestimations in predictions may occur even the streets were aligned with the wind direction (for main streets, K = - 79% to - 88%; for inner streets, K = - 82% to - 93%). Conversely, the street width varies from wide to narrow tend to result in overestimations (for main streets, K = 44% to 211%). The prediction results can be reassessed when conducting similar research by taking into account the prediction deviation thresholds identified in this research.