Flows across 3D urban street canyons: Revisiting Reynolds number independence by large eddy simulation

• Flow structures in 3D urban street canyons evolve with the Reynolds number ( Re ), becoming more similar as Re increases. • The critical Re required to achieve Re -independent flow increases with H/W and L/W . • In-canyon vortex structures are influenced by H/W and L/W , while vortex core locations shift based on vortex interactions. • Spanwise flow remains significant in long urban street canyons even at L/W = 10. The Reynolds number ( Re ) independence assumption allows flows across urban street canyons to be studied at much lower Re in reduced scale experiments relative to Re at full scale. Re of 10,000 is often taken to be the critical Re ( Re c ) to achieve Re -independent flows, but recent studies suggest that this Re c is not universal. Moreover, most previous research has focused on Re -independence in 2D street canyons of infinite length, whereas real urban street canyons are inherently 3D, formed by buildings with varied lengths. This study investigates the influence of height-to-width ( H/W ) and length-to-width aspect ratio ( L/W ) of canyons on flow structures and Re c in urban street canyons, using validated large eddy simulations (LES). Our flow pattern analysis reveals that vortex structures in 3D urban street canyons differ significantly from those in 2D canyons. Unlike 2D canyons, where vortex cores are centralized, the vortex cores in 3D urban street canyons shift due to the interaction between spanwise flow and vertically aligned vortices. The results show that Re c increases with H/W and L/W . For example, a canyon with H/W of 1 and L/W of 1 has Re c between 15,000 and 30,000, while a canyon with H/W of 3 and L/W of 10 has Re c between 180,000 and 360,000. These findings indicate that Re c required to achieve Re -independent flows in 3D urban street canyons depends on the geometrical parameters such as H/W and L/W . Care should be taken when applying reduced-scale results at relatively low Re to full-scale scenarios.