Model Validation of Flow and Dispersion Around a Cube

This paper compares results for flow over a cube between laboratory experiments and two numerical simulations. One of the simulations is a Reynolds-averaged Navier-Stokes (RANS) calculation, the other a large eddy simulation (LES). Both the structure of the flow and dispersion of a source behind the cube are compared. It was found that both simulations performed well when mean flows are compared. For dispersion, the LES performed better than the RANS simulation in that it was able to capture the effect of vortex shedding and produce a wider dispersion pattern. The plume in the RANS simulation is very similar to instantaneous realizations of the plume in the LES. Near the cube, the results were very similar. This model validation study suggests that the high cost of LES computations may be warranted when detailed time-varying solutions are of high interest, However the high fidelity RANS approach is a cost-effective alternative to LES in obtaining time-average mean field results