Currently, I have developed Large-eddy model in Lattice Boltzmann Method. When i use 1024 * 512 mesh, it works fine and reasonable match with literature is found. However, by increasing the mesh quality, drag and lift values got disturbed and values move far away from the literature. I do not know why fine mesh is causing problem. I think i need to add kind of filter in my coding to filter the distribution function or something. Can anybody help me how to solve this problem. (Using low quality mesh 512 * 256), result matches with literature.
I need fine mesh around the body for complex bodies, like, square with chamfer ratios.
Hi Muhammad, First of all you should know how to perform a mesh sensitivity analysis. One of the most recommended method is to decrease the cell size by increasing the number of cells, this would lead you to desirable results. Which means the cell size from which you cannot get any high difference in magnitude of various physical values. It should be noted that an over sensitive mesh could lead you to high time consumption and error.
Hi
if you get reasonable answer and the mesh indecency has been done, before you got error, it means after that by decreasing the cells you should change your turbulence model probably. Because the inflation is affected the number of cells and change Y+, so the results will go to gain errors.
The best method is to decrease the cell size (which do not change magnitude of various physical values) by increasing the number of cells .
Muhammad,
I guess you should not think of using any filter. Normally, a finer grid should lead to more accurate solutions, therefore your problem may come from somewhere else. To clarify where the problem might be, you need to perform a grid convergence test at first. Please be aware that LBM is very sensitive to the aspect ratio of the discretization! Then, you had better make sure that the boundary conditions are well formulated and imposed.
Good luck!
In LBM, one has to be very particular about the time step size too. With refinement did you adjust the time step size? I presume that you are familiar with the stability criteria in LBM.
- Badges
- Science topic
- Similar topics
- Control Systems Engineering
- Filtering