CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Complex Mesh In Gambit. (

Ranjeet March 24, 1999 18:54

Complex Mesh In Gambit.
Has anybody meshed a complex volume in gambit with structred mesh on few of the faces and unstructred mesh inside the volume.? need few tips?

Jonas Larsson March 25, 1999 03:43

Re: Complex Mesh In Gambit.
Yep, I've meshed a 1.5 stage turbine using a hybrid grid on the hub/shroud surfaces (quads around the blades and tris further out) and "cooper" (ie structured) in the radial direction. Works well, although bugs in the boundary layer handling makes it difficult to get all boundary layers correct. If you are having problems with boundary layers get the 1.0.4 release due out in a week or two, it is much better than 1.0.3 at boundary layers, although several problems still remain.

Ranjeet March 25, 1999 11:09

Re: Complex Mesh In Gambit.
It seems the mesh You are talking about is a 2-D mesh.What I am trying to do is say a cube is there and You mesh quads on few of the faces and tris inside the volume.Here Gambit gives me error that some x nodes not meshed and so it cant mesh the geometry.I have been playing around with this for a week now. need help?

John C. Chien March 25, 1999 14:07

Re: Complex Mesh In Gambit.
(1). there is no simple method to generate complex geometry and 3-D mesh automatically !!! (2). but if you plan ahead, and divide (cut) the computational domain into smaller blocks ( the surface patches will be smaller and simpler than one large complex surface), the nodal distribution, the surface mesh generation and the volume mesh generation will be easier to handle. (3). it is always a practical method to build the geometry and the mesh from ground up ( create points, curves, surfaces, volumes, and associated meshes).(4). I have used more than 500 surface patches to generate a complex 3-D geometry and mesh, using the very old FLUENT/preBFC module. ( the generation before Geomesh and Gambit) The mesh generator was tgrid module.( it can handle hybrid mesh) (5). The current generation like Gambit can handle the geometry and mesh generation in two ways: one is the traditional ground up method, and the other is CAD like top down solid modeling. There are many rules involved in the automatic mesh generation using the solid model top down approach. (6). My suggestion is: a). the first ground up method will always give you the answer with a lot of control over the geometry and the mesh, if you can spend more time to study the geometry of your problem and simplify it by subdivision of the computational domain. b). In order to use the second method, you really have to study every commands and the procedures of the user's guide in order to take advantages of the top down method. (7). So, in the ground up method, you control the mesh generation. On the other hand, you have to understand and follow the program's rules in order to use the top down method. (8). a few weeks or a few months is not uncommon to generate a complex 3-D mesh. (9). I have not started using the new Gambit program, but I think, if you follow the rule exactly, your chance of getting the final 3-D mesh should be relatively high. In this area, be patient.

ranjeet March 25, 1999 21:50

Re: Complex Mesh In Gambit.
Thanks for the reply,I hope now I will be able to tackle this.

Jonas Larsson March 26, 1999 03:29

Re: Complex Mesh In Gambit.
Nope the mesh I'm talking about is a full 3D mesh of a 1.5 stage axial turbine (two stators and one rotor). I meshed it using a mix or tris and quads on the hub surface and then extending them in the 3rd direction using the "cooper scheme". I have also started adding rim-seals and cavities using fully unstructured tets.

Ranjeet March 26, 1999 11:18

Re: Complex Mesh In Gambit.
In my case its the other way round, I have to make 2 sides as rot.periodic and two sides of the geometry as trans. periodic so I need quads on the outer surface and tris in the core.

John C. Chien March 26, 1999 13:25

Re: Complex Mesh In Gambit.
(1). If the outer boundary of the domain is covered by six sided bricks(hex.), then, in order to grow the four-sided cell(tet.) inside this closed boundary, you need a transition layer of five-sided cells(pyramid.) (2).you need to find out whether the program can do this automatically or not.

Ranjeet March 26, 1999 15:53

Re: Complex Mesh In Gambit.
That I have checked with the Gambit tech support and they say that Gambit automatically makes these pyramids. That is it supports quads on surface and tets inside the volume.

Kevin Cahill April 29, 1999 17:18

Re: Complex Mesh In Gambit.
The error message you are getting is typical of a mismatch in number of elements on your outter and inner surfaces. You will have to "play" with the number of tets on your inner surface and try remeshing the voulme until gambit can handle the transition. Remember that you need more tets on the tet-meshed surface than hexs on the hex-meshed surface. Try a ratio of about 2:1, but the spacing between the surfaces is also important. Keep trying. I have been throught this cycle, and it will work with a little luck.

All times are GMT -4. The time now is 20:36.