Reduced-scale numerical simulation method and its application to urban-scale buoyancy-driven flows

Urban-scale flows such as urban heat island circulation dominate the dissipation of heat and pollutants in urban areas under calm and stable weather conditions. The study of such buoyancy-driven flows at an urban scale is commonly conducted using water tank modeling and numerical simulations. This paper adopts an approach called "numerical water tank" which utilizes computational fluid dynamics to simulate the water tank model. By adopting this method, the research aims to investigate and analyze urban-scale buoyancy-driven flows. This paper proposes an evaluation criterion to assess eight turbulence models in various aspects. The results show that the SST k-omega model and LES perform well. LES is more accurate than RANS, and compared with LES, SST k-omega model can save 1/3-1/2 of the calculation time. For two-equation turbulence models (k-epsilon and k-omega), which are known for their easier convergence, it is recommended to maintain a Courant number equal to or below 1 (time step 0.5 s). In the case of LES, which is known for its challenging convergence, it is recommended to maintain a control Courant number below 0.2 (time step 0.1s). On this basis, the numerical water tank method based on large eddy simulation is applied to quantitatively study the effect of canopy heat flux attenuation on urban-scale wind field characteristics in valley cities. This paper combines the strengths of water tank modeling and numerical simulation, resulting in a more efficient approach for studying large-scale flows driven by natural thermal pressure.