CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [GAMBIT] Meshing in GAMBIT (non symmetrical geometry) (

sbrCFD November 2, 2011 12:16

Meshing in GAMBIT (non symmetrical geometry)
I am new to GAMBIT and I am working on a project that involves a lake. As a first attempt I have considered it as a big pool with some inlets and outlets (inflows/outflows). I have generated the geometry considering some points and then, I build some edges to form the pool. Moreover I have also considered some “holes” from which inlets and outlets are going to go in and out. The final geometry is not symmetrical. I have meshed the edges and finally, I intended to mesh the volume (I have considered one volume only). However, the volume mesh could not be done because une of the faces could not be meshed succesfully (according GAMBIT has told me).
Could anyone tell me the best way to deal with this kind of problem? Would it be better to consider several volumes (i.e., to divide the “pool” in diferent sections)? Should I have the same number of nodes in the parallel faces of the “pool”?
I will appreciate very much your help. Thanks!

-mAx- November 3, 2011 01:46

Gambit is giving to you an info that it could not mesh one surface.
But it also gives you the surface id.
So you can try to check this surface and you will see where is the problem.
Then if you can either try to mesh your volume on the fly, or split your domain into several subdomains as you mentionned.
If you need more help, don't hesitate to post pictures

sbrCFD November 3, 2011 11:42

Meshing in GAMBIT (2)
1 Attachment(s)
I have decided to split the domain in several subdomains (volumes). Then, I have tried to mesh one of the volumes (specifically volume 2), and an error has appeared with “face.16”. The error message was “Bad interval assignment for submapping face.16”. I do not know how to manage this error. Would it be necessary to have the same geometrical forms on the opposite faces of the volume (face.16 and face.9 in this case)?
I am also posting a file (please let me know if you are unable to open it).
Many thanks for you help.

-mAx- November 4, 2011 02:20

*delete all your volume meshes
*mesh surface 16
*check that edges 41 & 38 (also 46 & 43) have same nodes-counter
*mesh surface 8 and 17 (will fail if edges 41 & 38 - 46 & 43 don't have same nodes-counter)
*mesh surface 15
*mesh volume 2

sbrCFD November 4, 2011 15:10

Many thanks mAx. I was able to mesh the faces and the complete volume. I will try to improve the mesh now.

sbrCFD November 7, 2011 10:28

More about meshing a non-symmetrical geometry (strategy)
I could mesh the geometry with the strategy previously proposed by mAx. However, the mesh it too fine in some subdomains (as in volume 2), or more than I would like. I would like to have a “uniform” mesh pattern. I think that a uniform mesh pattern will allow me to solve the problem properly, obviously, with a finer mesh on the vicinity of the inflows/outflows of the geometry. I could mesh the complete geometry with the same pattern as volume 2 but I will obtain a geometry with too much nodes.
With that purpose in mind, I have created several volumes, then I meshed all the edges (with a finer pattern on the vicinity of the inflows/outflows of the geometry), and finally I tried to mesh each volume. As you can see from my previous posts, this strategy did not allow me to mesh one of the volumes (volume 2).
Could anyone tell me a “smarter or proper” strategy to achive my objective? I am not sure about how fine/sparse the mesh should be , but then I would like to refine the mesh. Do you think that having a geometry with such different dimensions between the “pool” and the inflows/outflows dimensions wouls oblige me to refine the full mesh to have a proper solution of the problem?
Many thanks for your help.

-mAx- November 8, 2011 02:46

You have a thin structure.
At least you need 8-10 cells in the depth (more if you treat it with a Boundary Layer).
This is your "pre-dominant" factor, and you cannot avoid it.
So you can "play" with the other directions to stretch your cells, but not to much, else you will have aspect-ratio issues.
Do you have any symmetry in your model? If yes use it (-them) to reduce your domain

sbrCFD November 11, 2011 14:01

Many thanks for your answer mAx!

All times are GMT -4. The time now is 00:58.