LES near wall mesh

tav98f · January 19, 2011, 13:38

I am running LES on an internal flow. I am consiering adding prism layers to my mesh to have good resolution near the wall. I know that tetrahedra are favorable over hexa because isotropy in the mesh is important for LES. Does this also mean that prisms are going to dampen or create turbulence? In other words, is a prism layer a bad idea for LES?

Thank you in advance for your help,
-Timothy

ghorrocks · January 19, 2011, 20:25

Where did you get the idea that tets are better than hexes from? It is wrong. If you can use hexes then do so as they are much better than tets. If isotropy is important then use 1:1:1 hexes.

The dissipation in tets, prisms and hexes of similar quality and size is similar. Hexes will be a bit less dissipation in some circumstances. But a big difference is hexes will use less memory.

tav98f · January 20, 2011, 14:33

I may have misrepresented what I was trying to say. What I meant is simply the anisotropy is not good. I was trying to mimic what the CFX documentaion says which is, "... consider the use of isotropic grids, perhaps using tetrahedral rather than hexahedral elements."

The reason that I am worried about using a prism layer is that the prism layer a set of layers that are thin extusions of the surface mesh. The size of the 2D projection of an element onto the wall does not change much from layer to layer. The length of the edges perpandicular to the wall, however, increases for each layer further from the wall. So the mesh that is closest to the wall is not isotropic.

ghorrocks · January 20, 2011, 22:04

The anisotropy is not due to tet vs hex. The anisotropy is due to poor mesh quality (ie squashed hexes). A 1:1:1 hex is better than a equilateral tet, but reasonable quality tets may be better than high aspect ratio hexes.

Prism meshes are a good idea next to the wall as anisotropy in hex/prism meshes is mainly a problem when the flow is in an artibrary direction. Next to the wall the flow is parallel to the wall, so high aspect ratio elements work OK as they are aligned with the flow.

January 19, 2011, 13:38	LES near wall mesh	#1
tav98f Member Timothy Join Date: Apr 2010 Posts: 33 Rep Power: 16	I am running LES on an internal flow. I am consiering adding prism layers to my mesh to have good resolution near the wall. I know that tetrahedra are favorable over hexa because isotropy in the mesh is important for LES. Does this also mean that prisms are going to dampen or create turbulence? In other words, is a prism layer a bad idea for LES? Thank you in advance for your help, -Timothy

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
unstructure mesh size control near to wall using gambit	lalit kumar	FLUENT	4	July 6, 2010 16:13
NEAR WALL MESH	kostas	FLUENT	8	December 8, 2008 07:11
Numerical Implementation of LES Wall models	dragon	Main CFD Forum	2	March 14, 2005 01:53
Multicomponent fluid	Andrea	CFX	2	October 11, 2004 06:12
turbulence wall law on unstructred mesh	Quain Tchew	Main CFD Forum	4	March 20, 2002 10:50

January 19, 2011, 20:25		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	Where did you get the idea that tets are better than hexes from? It is wrong. If you can use hexes then do so as they are much better than tets. If isotropy is important then use 1:1:1 hexes. The dissipation in tets, prisms and hexes of similar quality and size is similar. Hexes will be a bit less dissipation in some circumstances. But a big difference is hexes will use less memory.

January 20, 2011, 14:33		#3
tav98f Member Timothy Join Date: Apr 2010 Posts: 33 Rep Power: 16	I may have misrepresented what I was trying to say. What I meant is simply the anisotropy is not good. I was trying to mimic what the CFX documentaion says which is, "... consider the use of isotropic grids, perhaps using tetrahedral rather than hexahedral elements." The reason that I am worried about using a prism layer is that the prism layer a set of layers that are thin extusions of the surface mesh. The size of the 2D projection of an element onto the wall does not change much from layer to layer. The length of the edges perpandicular to the wall, however, increases for each layer further from the wall. So the mesh that is closest to the wall is not isotropic.

January 20, 2011, 22:04		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	The anisotropy is not due to tet vs hex. The anisotropy is due to poor mesh quality (ie squashed hexes). A 1:1:1 hex is better than a equilateral tet, but reasonable quality tets may be better than high aspect ratio hexes. Prism meshes are a good idea next to the wall as anisotropy in hex/prism meshes is mainly a problem when the flow is in an artibrary direction. Next to the wall the flow is parallel to the wall, so high aspect ratio elements work OK as they are aligned with the flow.