Dynamic sub-grid scale turbulent Schmidt number approach in large eddy simulation of dispersion around an isolated cubical building

In this paper, the effects of the sub-grid scale (SGS) turbulent Schmidt number, ScSGS, on the large eddy simulation of dispersion on and around an isolated cubical model building with a flush vent located on its roof are examined. Constant and dynamic ScSGS approaches for SGS turbulent mass flux modeling are employed. Simulation results are compared with the available wind tunnel measurements. Furthermore, the influence of the grid resolution on the accuracy of results predicted by the dynamic ScSGS approach is investigated. Detailed statistical analysis of ScSGS demonstrates that the dynamically computed ScSGS at different locations varies by a factor of almost 5 and a considerable deviation of ScSGS from its common values of 0.5 and 0.7 occurs. Particularly, in the vicinity of the building where the concentration gradients are noticeable, ScSGS has a larger variation. Also, the probability of occurrence of 0.2 < ScSGS <1.5 is more than 90 percent and the ScSGS mean values are nearly around 0.8 to 1 with a maximum variance of 0.2. In addition, by refining the grid, the differences between the predictions of constant and dynamic ScSGS approaches decrease. This is due to the reduction of sub-grid scales contribution to turbulent dispersion. It is confirmed that dynamic ScSGS approach is a practical alternative to the constant ScSGS approach, effectively eliminating a user-defined model coefficient.