CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Hexa meshing or Tetra meshing??? (http://www.cfd-online.com/Forums/cfx/26210-hexa-meshing-tetra-meshing.html)

NITIN DEWANGAN August 5, 2008 01:23

Hexa meshing or Tetra meshing???
 
what would be the change in the solution when we make geometry thru hexa meshing or thru tetra meshing?

which one is most accurate and easy to make ?

thanks for help...


Rogerio Fernandes Brito August 5, 2008 06:46

Re: Hexa meshing or Tetra meshing???
 
Both have the same accuracy solution. But tetra mesh wastes more time of cpu and the most number of elements. with hexahedrical mesh, you can use a coarse mesh and you will have a good result. I made this type of comparison between tetraedron and hexahedrical meshes in solid simulations (without fluid air or water...).

MAB August 5, 2008 07:11

Re: Hexa meshing or Tetra meshing???
 
also the number of tetra is about 60,00000

John August 5, 2008 09:25

Re: Hexa meshing or Tetra meshing???
 
I wouldn't say tets and hexes give the same solution. I'm troubleshooting a combustion problem right now where I have one mesh which is 100% hex and the other is tet with an inflated prism layer.

The hex mesh runs just the way you'd expect with a clear flamefront all the way to the wall, but the tet mesh has a lot of numerical diffusion with the turbulence equations at the tet/prism boundary which is giving some non-physical results (such as preventing the flame from spreading into the inflated boundary).

If you have a copy of the CFX manual I'd recommend taking a looking at the section on discretization errors in the Theory manual and you'll see that tets and prisms are fundamentally treated differently by the solver.

Rogerio Fernandes Brito August 5, 2008 09:54

Re: Hexa meshing or Tetra meshing???
 
It depends on your problem, of course!

John August 5, 2008 09:59

Re: Hexa meshing or Tetra meshing???
 
I agree. For the most part the amount of numerical diffusion at the tet/prism boundary is neglible and does not have too great an impact on the solution; however, it seems that in flows with large gradients in shear stress or large differences is the rate of shear strain that these diffusion effects are more readily apparent.


All times are GMT -4. The time now is 02:47.