Validation study of cross-ventilation in a realistic building geometry: RANS, SAS and LES

The validation of computational fluid dynamics (CFD) simulations of natural cross-ventilation flow with wind tunnel (WT) measurements is important in view of accurate and reliable numerical simulations. A review of the literature indicates that the majority of previous CFD and WT measurement studies employed a simplified generic single-zone building with a prismatic shape. The objective of this study is the validation of isothermal CFD simulations of two different realistic building models resembling a pitched roof single-story house, both without (case 1) and with internal partition (case 2). The CFD simulations were conducted using the 3D steady Reynolds-averaged Navier-Stokes (RANS) approach with the SST k-ω, RLZ k-ε and RNG k-ε turbulence models, scale-adaptive simulations (SAS) with the SST k-ω model, and large eddy simulations (LES) with the Smagorinsky-Lilly subgrid-scale model. The evaluation was performed in two parts: impact of turbulence model and impact of internal partition. The results show that LES and SAS exhibit a good agreement with WT results, outperforming RANS for the two cases. When considering only indoor streamwise mean velocity, for case 1, 97% and 73% of the sampled LES and SAS velocities fall with the uncertainty band of the WT measurements. For case 2, these values are 92% and 75% for LES and SAS, respectively. Steady RANS provides an agreement of only 56% and 63% for case 1 and case 2, respectively.