RANS model calibration using stochastic optimization for accuracy improvement of urban airflow CFD modeling
Mohammadreza Shirzadi, Parham A. Mirzaei, Yoshihide Tominaga
Abstract
In this study, a systematic calibration methodology is proposed for enhancing the accuracy of urban airflow simulations using computational fluid dynamics (CFD) models based on the Reynolds-averaged Navier-Stokes (RANS) equations. In the calibration process, high-quality data from different sources are used to define the validation metrics, which are then utilized as the objective function in a stochastic optimization solver to find optimal values for closure coefficients of the RANS turbulence model. The proposed calibration method is applied to three different urban case studies, including an unstable atmospheric boundary layer (ABL) around a high-rise building, a sheltered cross-ventilated low-rise building, and a group of low-rise buildings located in a highly packed urban area. The significant advantage of using the obtained calibrated coefficients is observed over the existing coefficients embedded in CFD tools as well as the ones recommended by other calibration methods in literature. Thus, this study proves the necessity of finding a group of customized optimum closure coefficients for RANS turbulence models suitable for a wide range of urban flow problems.