[chen112p]

Hello all, I have been suffering from this issue for a while.

Currently I am working on external flow past bluff bodies. I admit that the geometries can be pretty complex: concave corners, convex corners, mixing of small feature ~0.5m and large scale geometry >100m and so, which gives a lot of stress on the mesher. Also, I would like to use prism layers on the bottom boundary, preferably on the geometry too.

RANS can converge pretty easily, most meshers can do a relatively good job. But when running LES/DES (PISO), simulations become much more unstable and unpredictable. Below I will list out the meshers (both commercial and open sourced) that I have tried with their issues/shortages. I hope someone can give me some advise on better using any of these meshers.

Snappy is a good tool but it's layer function is pretty funky, even if I only want to create prism layers on the ground boundary. I haven't found any success yet. Using graded background blockMesh very often messes up the geometry during castellation.

Star-CCM + ccm26ToFoam: the trimmer mesher is highly unstable when running LES. Most of the time sims diverged on corners and prism layers almost immediately. Meshes from the tet mesher is much more stable than the trimmer mesher, but 30% to 40% of the sim still failed pretty randomly (sometimes very long time after the sim started). Correct me if I am wrong, tet is not a preferred mesh type in FOAM. Also tet cells from Star-CCM are not created using delauney method which cant be converted to poly cells by polyDualMesh.

GMESH: Correct me if I am wrong, GMESH is not capable of generating 3D prism layers, and it is very hard to keep tracks of imported surfaces after the boolean subtraction

Thanks a lot

JT

[fumiya]

Hi,

You might want to try if velocityDampingConstraint helps:
https://www.openfoam.com/documentati...y-damping.html

Hope this helps,
Fumiya