Solving city and building microclimates by fast fluid dynamics with large timesteps and coarse meshes

Abstract

Many urban and building microclimate problems are featured with a big computational domain and a large number of computing grids, so a fast and accurate solution by computation fluid dynamics (CFD) remains a challenge. The simulations are often only steady-state and, if there exists a transient model, the solution is often limited for a short period. For these computationally expensive simulations, large timesteps and coarse meshes (LTCM) seem preferable among users with limited computing resources, whereas many existing CFD models may not perform well for LTCM problems. This study developed a new fast fluid dynamics (FFD) model based on the semi-Lagrangian approach with high-order temporal and spatial schemes designed for LTCM problems. For large timesteps, a 2nd-order temporal scheme estimates the characteristic curve of each fluid's particle by accounting for the acceleration of the flow field. For coarse meshes, a 4th-order spatial scheme applies a backward-forward sweep technique for better accuracy in space. The performance of the proposed method is verified and validated by five benchmark cases, including a pure advection problem, forced convection in a single room, transient wind flow around a cylinder, airflow around a group of low-rise buildings, and airflow around a high-rise building. Then, we demonstrated the model for the study of the urban wind flows in a city downtown area with 2500 buildings and 35 million grids. As one of the first studies of applying FFD to urban aerodynamics, this study shows its strong potential for solving big transient problems with large timesteps and coarse meshes.