CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Mesh generation -purpose of tetra transition (http://www.cfd-online.com/Forums/main/108472-mesh-generation-purpose-tetra-transition.html)

eishinsnsayshin October 24, 2012 11:27

Mesh generation -purpose of tetra transition
 
Hello everyone,

I don't have the greatest background in mesh generation algorithms and am relatively new with CFD in the automotive industry. I've just recently begun to mess with Hexa-interior mesh using ANSA for its benefits such as flow alignment and mesh size reduction.

Take the example of a rectangular box, surface meshed entirely with quads. Wall conditions on the sides, an inlet, and an outlet. Pretty simple model! Using this, I've noticed that if you have a quad surface element, and you do a "hexa interior" or "hexa poly" volume mesh, pyramids will be used on each of those quad surface elements, followed by a few tetras to transition to the actual hexa elements. These tetras make a nice "squared off" flat surface for each of the hexas to go on such that their edges are aligned with the flow. I've also noticed ANSA will use tetras/polys at times if you have specified a growth rate for your hexas to transition between the different hexa sizes (kind of obvious...it has to use something)

Now, lets say I wanted my rectangle domain volume meshed completely with hexas. Is there a way in ANSA to do this? I am specifying a growth rate of 1 (so...no growth). Using quad surfaces for the inlet, outlet and sides still will result in pyramids on the surface elements, a few tetras, and then perfect little cube hexas of the same size all the way to the center. My question is why the pyramid/tetra transition...the nodes end up in pretty much the same distance from the wall as if the whole thing was just hexas. Is there a CFD solving advantange for ANSA to do this? It's certainly not helping with resolving the boundary layer, since the first nodes off the wall of each pyramid is way too far away anyway. (I would never really expect good wall results doing this; I would in all practicality use prism layers...but still...the pyramid/tet transition would still exist).

Any clue? or is that just the way it is? Also, I tried to be clear with my question but I understand it's confusing, so let me know if you need any clarification!

Thanks in advance!

vangelis October 25, 2012 04:02

Hi there,

HexaInterior and HexaPoly algorithms are meant to generate hexas inside the main core of the volume for complex geometries where having pure hexa mesh is not possible. Therefore there is this transition zone with pyramids and tetras, so that the mesh can adjust to the complex surface of the model. Prior to using these algorithm one probably also generates boundary layers from the surfaces. You can refer to tutorial meshing for external aero.

If you have a simple geometry, like a square box, meshed with map quads then it is better to use VOLUMEs>MAP
You select one side and confirm, the opposite side and confirm and then the remaining round sides and confirm.
ANSA will generate a pure hexa mesh in this case.

Another option to generate pure hexa meshes is the HexaBlock menu of ANSA, please refer to respective tutorial of ANSA.

Hope this helps
Vangelis


All times are GMT -4. The time now is 18:16.