A CFD Investigation of wind- and buoyancy-driven flows in urban street canyons: Assessment of wall-modelled LES and Reynolds number independence

Source abstract

Accurately modelling street canyon airflow is vita for studying urban heat islands and ventilation. Such airflow dynamics have been widely modelled using large-eddy simulation (LES), ideally performed with a wall-resolved (WR) approach. Although highly accurate, this approach becomes computationally expensive for full-scale flows characterised by high Reynolds numbers ( R e ). There are two feasible approaches to mitigate this cost: (1) conducting wall-modelled large-eddy simulation (WMLES), or (2) conducting WRLES at reduced or intermediate scales and extrapolating the conclusions to full scale based on the assumption of R e independence. However, the applicability of these approaches to flows driven by both wind and buoyancy forces in street canyons has not been systematically assessed. Starting from a Computational Fluid Dynamics (CFD) validation against a reduced-scale wind tunnel experiment, this study compares the performance of WMLES with WRLES in predicting these flows at reduced (1:100) and intermediate (1:10) scales. The results indicate that WMLES provides reliable predictions of the flow and temperature fields. It is followed by assessing the validity of R e independence by comparing LES results at reduced, intermediate and full scales ( R e ≈ 10 4 , 10 5 , and 10 6 , respectively). Significant differences in flow structure and temperature distribution across scales indicate that the assumption of R e independence is invalid. Consequently, conclusions drawn from reduced-scale models using CFD or wind/water tunnels, or from intermediate-scale field tests for flows driven by such wind-buoyancy forces cannot be directly extrapolated to full-scale conditions. Full-scale WMLES simulations are recommended for these flows in realistic urban environments.