Buoyant flows in street canyons: Comparison of RANS and LES at reduced and full scales

Abstract

Reduced-scale experiments and full-scale field measurements show contradictory buoyancy effects from heated windward walls. Reduced-scale experiments exhibit significant thermal effects, but full-scale field measurements show a negligible thermal effect on the overall flow fields. This paper investigates this discrepancy by using at both scales Computational Fluid Dynamics simulations with Reynolds-Averaged Navier-Stokes (RANS) and Large Eddy Simulation (LES). Compared to experimental and field measurements, RANS models perform well at reduced scale but over-predict the thermal effects of heated windward walls at full scale. On the other hand, LES results agree well with measurements at both scales. Therefore, LES should be used for full-scale simulations of street canyon flows with heated windward walls. To date, there has been no explanation for the discrepancy of opposing thermal effects between reduced-scale experiments and full-scale field measurements although Richardson number similarity is satisfied. We provide an explanation by showing that in reduced-scale experiments with heated windward walls, the assumption of Reynolds number independence is invalid. In canyon flows with thermally induced buoyancy, unless the flow is proven independent of both Reynolds number and Grashof number, we should not generalize results from reduced-scale experiments to full-scale street canyons. Keywords: Urban street canyon; Large eddy simulation; Reynolds number independence; Dimensionless group similarity; Full-scale simulation; Buoyant flow