The purpose of this paper is to understand the capability and consistency of large eddy simulation (LES) in Eulerian–Lagrangian studies aimed at predicting inertial particle dispersion in turbulent wall-bounded flows, in the absence of ad hoc closure models in the Lagrangian equations of particle motion. The degree of improvement granted by LES models is object of debate, in terms of both accurate prediction of particle accumulation and local particle segregation; therefore, we assessed the accuracy in the prediction of the particle velocity statistics by comparison against direct numerical simulation (DNS) of a finer computational mesh, under both one-way and two-way coupling regimes. We performed DNS and LES at friction Reynolds number Re τ= 180 in smooth and rough channels, tracking particles with different inertia, with the aim to conduct a parametric study that examines the accuracy of particle statistics obtained from LES computations. The issue has been widely analysed in turbulent flow bounded by smooth walls, whereas the effect of rough boundaries on momentum coupled two-phase flows has been much less investigated until now. The action of the roughness of the wall is studied in terms of both turbulence modification and particle interaction with the wall surface due to particle rebounding off the solid boundary, without the introduction of a virtual rebound model. Results show that resolved LES adequately predicts particle-induced changes in both fluid and particle statistics in rough channels, at least for the range of parameters considered here. © 2020, Springer-Verlag GmbH Austria, part of Springer Nature.
|Number of pages||24|
|Publication status||Published - 2020|
All Science Journal Classification (ASJC) codes
- Computational Mechanics
- Mechanical Engineering