CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   tetra hexa mesh (http://www.cfd-online.com/Forums/fluent/95755-tetra-hexa-mesh.html)

 hamid1 January 1, 2012 08:40

tetra hexa mesh

Hi Guys,

i have to model a pipe in 3-d but with tetra mesh it seems that i have to dense it very very much to make the solution independent of grid,
do u think in hexa mesh it radically improves?

for grid independent study what is the role of thumb for the criteria to use, in terms of which parameter is better to be measured , how do u measure it (e.g. weighted average etc) and how much difference between denser and coarser mesh is acceptable?

thank you
hamid

 Far January 1, 2012 11:22

Quote:
 Originally Posted by hamid1 (Post 337490) Hi Guys, i have to model a pipe in 3-d but with tetra mesh it seems that i have to dense it very very much to make the solution independent of grid, do u think in hexa mesh it radically improves? for grid independent study what is the role of thumb for the criteria to use, in terms of which parameter is better to be measured , how do u measure it (e.g. weighted average etc) and how much difference between denser and coarser mesh is acceptable? thank you hamid
It is better to use hexa mesh. You can get the mesh independence at relatively low density in hexa.

 BigBen January 3, 2012 20:05

The idea when you are using hexa compare to tetra is to align your cells with the flow pattern. In this case, there is less information diffusion between the cells.

Now, in Fluid mechanics, everybody knows that the main pressure gradient is in the boundary layer, close to the wall.
Hexa mesh allows you to refine in the boundary layer thickness and to strech the mesh in the flow direction. Hexa mesh is anisotropic.

At the inverse, tetra mesh is an iotropic mesh. Means, if you want to refine in the boundary layer then you need to refine in the 3D dimensions (lot of cells).

2 "good" points: Hexa needs less cells and get less diffusion if flow aligned with the cells
1 bad point: it is not always easy to create an hexa mesh.

Tetra with never be as good as hexa

 Kamu April 11, 2012 11:49

Mesh Independent solution

I am using hexa mesh for pipe flow but for all my element numbers the solution is grid independent. Am modelling turbulent flow with standard wall functions. When I change the distance from the wall using inflation layers the solution remains the same as previous one! The results seem sensible.

 BigBen April 11, 2012 12:13

Fine,
If you obtain the same result whatever is the mesh, then it is perfect. Find the coaser mesh you can to reduce time consumption in the futur.
Which CFD code are you using?
Depending on the turbulence you are using, all the mesh are not recommanded.

The mesh size is based on the Y+.
You can decomposed the boundary layer in 3 layers:
Viscous sublayer
Outlayer and in the middle the log layer

Basically
Y+=1: sublayer linear layer u+=y+
30<Y+<300:log layer

You have to know where are valid your turbulence model and the wall function you are using.

 banty April 12, 2012 11:23

Hi,

For the same number of cells in hexahedral and tetrahedral mesh, average cell centroid distance in tetrahedral mesh is larger compared to the hexaherdal mesh. thats why tetrahedral mesh need more no of cells to be grid independence.

Anyway hexahedral mesh gives more accurate result compared to tetrahedral mesh it does not matter which solver u are using because pure orthogonal quality of tetrahedral mesh.

Quote:
 Originally Posted by hamid1 (Post 337490) Hi Guys, i have to model a pipe in 3-d but with tetra mesh it seems that i have to dense it very very much to make the solution independent of grid, do u think in hexa mesh it radically improves? for grid independent study what is the role of thumb for the criteria to use, in terms of which parameter is better to be measured , how do u measure it (e.g. weighted average etc) and how much difference between denser and coarser mesh is acceptable? thank you hamid

 All times are GMT -4. The time now is 11:37.