CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] Improve prism mesh at trailing edge (

siw August 24, 2009 02:46

[ICEM] Improve prism mesh at trailing edge
1 Attachment(s)

How can I improve the structured mesh at the trailing edge of a wing? The image is what I have. The prism mesh height is based on a y+ calculation, but I have not yet split the mesh in the required number of elements.


PSYMN August 27, 2009 13:35

3 options.
The Y+ calc should give you the first layer height, not the total height, so I guess you calculated using the ratio and number of layers to find the total height?

This method of generating one or a few layers and then subdividing is good for speed, but you give up some quality.

Try doing it with at least three or 5 layers (and then split). You will see that the software can then curve the prism layers to better handle these situations…

Another fun trick is to create a trailing edge surface that would go back half a chord or so… This surface will be a zero thickness baffle surface, so don’t forget to make it an “int wall” under the Mesh tab => Params by parts. Include this for growing prisms and your prism layer will nicely walk straight back from the trailing edge. There will be pyramids at the end of the trailing edge surface, but that should be fine with most solvers. Then go back and delete the shells for the baffle surface.

Another option would be to reduce the Max Prism angle. Currently it must be set to 180 degrees because it is wrapping around this trailing edge. If you reduce it sufficiently, it won’t wrap and will just open up (with a pyramid cap) instead…

Personally, I like the baffle idea best.

siw August 28, 2009 02:32

Thanks PSYMN, I appreicate your help once again.

I think my explanation was unclear but I did use the Y+ calc for the first layer height - the CFX Mesh guide gives a handy equation. I just then worked out a total layer height with an expansion factor of 1.2 and 15 individual layers.

I don't think I fully understand your baffle surface. Are you saying that you extend a surface rearwards from the trailing edge that has zero thickness and put prism layers on both sides of this surface? And that the prism are unable to curve around from the upper and lower sides of this surface?


PSYMN August 28, 2009 09:20

Right, so I take the trailing edge curve and extrude it backwards into a new surface and grow prisms on it. We are better off with the prisms on the baffle surface stair stepping down (via pyramids) along this baffle rather than trying to wrap them around the ends…

I don’t have a handy non-proprietary example to show you and I don’t have time to make one up right now, but try it an you will see what happens.

Prakash.Paudel August 18, 2011 15:17

just a quick question on PSYMN's suggestion on the baffle surface. Before tossing the mesh out to the solver, do i need to delete the baffle surface as well along with the shells on that surface ?

PSYMN August 20, 2011 23:56

Up to you...
You never need to delete the "surface"... Only the mesh goes out to the solver and it may be handy to keep that surface in case you want to remesh again later.

But you may want to delete the shells. It is up to you. You could also just set those shells as an internal wall in the solver and then you could use it for post processing...


All times are GMT -4. The time now is 11:48.