Quality and reliability of LES of convective scalar transfer at high Reynolds numbers

Numerical studies were performed to assess the quality and reliability of wall-modeled large eddy simulation (LES) for studying convective heat and mass transfer over bluff bodies at high Reynolds numbers (Re), with a focus on built structures in the atmospheric boundary layer. Detailed comparisons were made with both wind-tunnel experiments and field observations. The LES was shown to correctly capture the spatial patterns of the transfer coefficients around two-dimensional roughness ribs (with a discrepancy of about 20%) and the average Nusselt number (Nu) over a single wall mounted cube (with a discrepancy of about 25%) relative to wind tunnel measurements. However, the discrepancy in Re between the wind tunnel measurements and the real-world applications that the code aims to address influence the comparisons since Nu is a function of Re. Evaluations against field observations are therefore done to overcome this challenge; they reveal that, for applications in urban areas, the wind-tunnel studies result in a much lower range for the exponent m in the classic Nu ∼ Rem relations, compared to field measurements and LES (0.52-0.74 versus ≈ 0.9). The results underline the importance of conducting experimental or numerical studies for convective scalar transfer problems at a Re commensurate with the flow of interest, and support the use of wall-modeled LES as a technique for this problem that can already capture important aspects of the physics, although further development and testing are needed.