Numerical simulation of aerosol deposition from turbulent flows using three-dimensional RANS and LES turbulence models

In this study, three-dimensional computational fluid dynamics (CFD) simulations of particle deposition from turbulent flows in a vertical straight pipe were carried out, using the STAR-CCM + v5.04 CFD package. The highlight of the study is the development of a post-processing approach for quantitatively assessing the accuracy of the Lagrangian particle tracking scheme. Three Reynolds-averaged Navier-Stokes (RANS) models and a large eddy simulation (LES) model were employed in the simulations, in conjunction with two wall treatment schemes and several near-wall mesh conditions. The particle deposition velocity was obtained based on the CFD simulation results, and compared to the experimental results. In post-processing, a particle responsiveness factor, defined as the ratio of particle mean square velocity fluctuation to fluid mean square velocity fluctuation for the wall-adjacent cells, was quantified using an Eulerian particle transport formulation. The particle responsiveness factor of the CFD simulations was then compared with that obtained using an empirical equation. The most accurate aerosol deposition results were obtained with a fine near-wall mesh (y(+) approximate to 1) and resolved near-wall flow (no wall function). The LES model and the Reynolds stress model (RSM) produced the most accurate deposition velocity results, but the computation time of the former was up to ten times longer than that of the latter. The lower accuracy of the isotropic RANS models was attributed to their overprediction of the near-wall turbulence intensity gradient, and less accurate particle tracking as suggested by the particle responsiveness factor. The particle responsiveness factor, introduced for the first time in this study, was shown to be a useful index for evaluating accuracy of the Lagrangian particle tracking scheme due to its independence of specific knowledge of CFD algorithm and coding.