Abstract
The transport and dilution of pollutants at street level are dominated by the combined effects of the turbulent background flow and turbulence induced by the motion of vehicular traffic. A thorough understanding of these effects is crucial for the ability to predict the spatial distribution of pollutants correctly. In this study, a new, simple method for modeling vehicle-induced turbulence (VIT) in large-eddy simulations (LESs) of urban street canyon flows is described. The method is validated by comparing results to wind tunnel data of (1) flow around a resting car-shaped body and (2) a street-canyon setup with driving vehicles represented by moving obstacles. The LESs indicate that the main features of the wake flow, including velocity deficit and recirculation length, of a resting car-shaped vehicle is reproduced with good agreement to the experimental data, and shows convergence at a grid spacing of
= 0.05 m. In the street-canyon configuration, the simulations capture the primary vortex structure, roof-level separation and associated shear-layer structure, and the associated pollutant dispersion patterns. Moving traffic leads to enhanced vertical exchange and a reduction of in-canyon pollutant concentrations, which is consistently reproduced by the model. A systematic overprediction of streamwise velocity and pollutant removal is observed, suggesting sensitivity to the representation of roof-level momentum exchange. The additional computational cost introduced by the method remains limited (approximately 1%–8%). Overall, these results indicate that the new method provides a simple tool for investigating the influence of moving traffic on urban flow and pollutant dispersion in LES models.