Numerical simulation of flow characteristics and pollutant distribution in 3D urban street canyons

Rapid urbanization has produced dense building clusters and narrow street canyons that inhibit pollutant dispersion. This study employs high-resolution three-dimensional computational fluid dynamics (CFD) simulations, using an unsteady hybrid Delayed Detached-Eddy Simulation (DDES) approach, of a representative street canyon on Huaihai Middle Road (Shanghai) to investigate flow structures and CO 2 dispersion for inflow angles and inlet wind speeds. The numerical model was validated against wind-tunnel measurements and reproduces primary vortex formation, bimodal vertical concentration profiles, and shear-layer separation. Key findings are that oblique inflow (notably 30° and 60°) induces asymmetric dual-vortex and multi-vortex states and multiple spatially distinct concentration maxima; low wind speeds produce enlarged vortices, longer residence times, and persistently high mean concentrations, whereas higher speeds improve mean dilution but increase instantaneous variability and the frequency of short-lived high-exposure events. Velocity and concentration fields are strongly coupled: coherent turbulent structures synchronize vortex shedding and concentration fluctuations. These results imply that monitoring and mitigation should address both mean and transient exposures. Dense or mobile sensor arrays with high temporal resolution are required to capture pulsed peaks, and urban-design measures (cross-canyon connectivity, targeted façade permeability, and strategically sited ventilation corridors) can substantially alter pedestrian-level exposure.