LAUSR

Abstract The large-scale practical application of Large-Eddy Simulation (LES) for predicting long-term wind flow and pollutant dispersion in urban areas is inhibited mainly by the associated very large computational costs. To overcome this difficulty, the present study, for the first time, applies transport-based recurrence Computational Fluid Dynamics (rCFD) to simulate atmospheric pollutant dispersion around a building. A novel diffusion model is proposed to accurately predict pollutant transport with rCFD. To illustrate the feasibility and advantages of rCFD, pollutant dispersion around an isolated cubical building with a rooftop vent, immersed in neutral atmospheric boundary layer flow is used as a case study and both LES and rCFD simulations are conducted. It is shown that rCFD simulation results agree well with those from LES both in terms of mean and fluctuating concentrations while the simulation wall-clock time drops from 222 h to 16 min. The application of four evaluation metrics (FAC2, FB, NMSE and R) indicates very good agreement between LES and rCFD results. Another major advantage of rCFD is that different pollutant events can be simulated promptly once the database has been stored for a given flow configuration, as shown by the comparison of LES and rCFD results for two other cases with different release locations. This study extends the application of transport-based rCFD to pollutant dispersion in the built environment and indicates that rCFD is a promising approach to facilitate the large-scale practical application of LES for this type of applications.