CFD simulations on urban microclimates: Validated by scaled outdoor experiments

Computational Fluid Dynamics (CFD) validations for urban microclimates often rely on full-scale experiments. However, spatial heterogeneities and anthropogenic influences in actual urban environments introduce significant data uncertainties, making it challenging to evaluate the impact of different turbulence, radiation, and heat conduction models on simulation results. Therefore, this study validated CFD simulations using scaled outdoor experiments with more controlled and reliable data. The unsteady Reynolds-Averaged Navier-Stokes (URANS) simulations were conducted by coupling radiative transfer, turbulent convection, and heat conduction. The simulations were evaluated against measured urban surface and air temperatures, as well as wind speeds. Differences among various radiation, turbulence and conduction models were compared. Results demonstrated that the simulations accurately captured diurnal variations in urban microclimates. Among the radiation models, the Discrete Ordinates (DO) and Surface-to-Surface (S2S) models provided better accuracy in predicting surface temperatures compared with the P-1 model. In turbulence modeling, the Standard (STD) and Realizable (RLZ) k- models performed similarly, while the Renormalization Group (RNG) k- and Shear-Stress Transport (SST) k- showed larger discrepancies, particularly in wind speed predictions below the urban canopy height. For heat conduction, the shell conduction model, which accounts for heat transfer in both normal and planar directions, yielded more accurate temperature predictions than the thin wall model with one-dimensional heat transfer. This research advances understanding of how numerical schemes affect CFD simulations of urban microclimates. The findings provide valuable guidance for model selection and improve simulation accuracy for urban planning.