Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
7157062 | Computers & Fluids | 2015 | 41 Pages |
Abstract
Two different DNS-approaches (direct numerical simulation of turbulent flow, i.e. without turbulence modeling) are used to determine the detailed structure of turbulent flow through channels with smooth and rough walls. They are a finite volume method (FVM) with grid refinement toward the walls and a Lattice Boltzmann method (LBM) with a uniform grid. With the LBM-approach high resolutions with up to 300Â million grid points are possible. Compared to the FVM-approach with less grid points and grid refinement toward the wall much higher resolutions away from the walls occur in the LBM solutions. Due to these high resolutions hairpin vortices can be clearly identified for these flows. They are known to be important building blocks in boundary layers. In channel flows, however, their dense existence has only been detected by experimental studies, whereas numerically it still has not been clearly visualized, especially in rough wall channel flows. Our LBM results show that hairpin vortices and their organizations in large scale and very large scale motions (LSM, VLSM) prevail in both smooth wall and rough wall channel flows. The similarity of the vortical structures in the outer layers over smooth walls and over the present k-type rough walls supports Townsend's wall similarity hypothesis.
Related Topics
Physical Sciences and Engineering
Engineering
Computational Mechanics
Authors
Y. Jin, M.F. Uth, H. Herwig,