CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   tetrahedral or hexahedral? (

rayy April 6, 2007 01:36

tetrahedral or hexahedral?
Hi all,

wanna ask, why we need to choose which meshing type that we need to use. why need to choose tet than hex. or the other way round. what effect that they bring to simulation?

thanks rayy

Gernot April 6, 2007 03:01

Re: tetrahedral or hexahedral?
The biggest disadvantage of a tet mesh is that you need much more cells ---> that means longer calculation time. And in many cases you get a better skewnes with hex cells.

Phil April 9, 2007 05:29

Re: tetrahedral or hexahedral?
but in some cases you want more cells for better accuracy!

Hex cells generally have less errors than tet but can only be used for simple geometries. You should split the model into areas that can be meshed well using any method - circles, triangles and rectangles that are as uniform as possible then use any combination to get the lowest skewness. Just takes some practice. If 99% of cells are lower than say 0.5 skewness and NONE over 0.7 skewness then you have a good mesh and there shouldn't be many AVOIDABLE errors. CFD naturally isn't 100% accurate for many reasons.

Bearing in mind that hex uses less cells, results wont be as comprehensive and so avoid switching from tet to hex cells in square areas if they are important areas of interest. You want a dense mesh in these areas. If though for example you are using all hex cells you can lower the mesh spacing to increase the number of cells and achieve the same detailed analysis as tet cells.

3D - THE BEST TACTIC HERE IS TO SPLIT THE GEOMETRY AS SUGGESTED ABOVE IN 1 PLANE THEN EXTRUDE THE MESH THROUGH THE MODEL USING COOPER MESHING. BY SPENDING TIME LEARNING TO USE THE "LINK FACE MESHES" FUNCTION YOU CAN ACHIEVE A RIDICULOUSLY GOOD GRID. My current gas turbine combustor model has 179000 cells with 99% of cells below around 0.35 skewness and this is modeling fuel injection, a swirler and a combustor. Only a 45 degree segment with periodic boundaries.

All times are GMT -4. The time now is 20:13.