I try to use Gambit now for 3 months and I'm not very happy with it. I'm calculating turbomachinery flow and have b-spline-surfaces and lines/curves from an external (unknown) modeller with iges. My typical problems are: 1.) Even simple elements like curves and radii, that are connected in the modeller, are not any more "real" connected in gambit and connected virtual geometry is bad for further operations. 2.) The "revolve" only works with planar lines or surfaces. Why ? In Turbomachinery you do not have much planar lines and surfaces. 3.) The "split surface" works sometimes and sometimes not (error) and has total different ways to work in real and virtual. Why can't a virtual face split a virtual face ? 4.) Booleans don't work with real geometry, so I have almost no advantages of the modern volume-modeler. 5.) The are not enough CAD-tools. Conclusion: If you have simple geometry (90-degree-angles, simple geometry, some points, geometry that can be built from predefined volumes ..) gambit is a good tool, but the other cases cause big problems. ... Now I have a new "state-of-the-art-modeller" and it takes much more time (if it works (!!)) to mesh a geometry than with p-cube. As I heard, p-cube is not supported next year. What are you doing ? Do you mesh turbomachinery succesfully with gambit ? What is your experience ?

A frustrated gambit user tom

Dear Tom,

I can undestand your pain and frustrations because you are working in a very complex area. Only the wearer knows where the shoe pinches. Only think we can do is try to solace your pain. At this moment, I would recommend you to go through some jounals in turbomachinery to improve your conceptions.

i'd probably recommend you contact FLUENT and let them help you with your problem. i have colleagues who have had problems with this software doing simpler things than what you're doing so you aren't alone. if you have customer support use it.

I have similar problems with GAMBIT as Tom. Even the new release 1.1 has a lot of bugs. I cannot mesh 3D surfaces, the mesh always collapses. Why cannot I revolve nonplanar surfaces? And many more wonders. Not very useful program for turbomachinery. I was trying to create a mesh of Kaplan turbine during vacation. The result was a surface mesh and volume mesh had to be created in TGrid. It seems we work with some beta version. Do you think, Fluent will improve Gambit until the end of millenium ? Recently they have announced a release of TurboGambit which should be a special modeling tool for turbomachinery. I have to say that I am very skeptic about it.

Hi,

I was reading your messages about mesh generation for Fluent and how ICEM P-Cube is not supported by Fluent in 2000. P-cube is written by ICEM CFD Engineering and will still be supported by us. Have a look at;

www.icemcfd.com under the GeoMesh link.

regards

David

Tom,

> 1.) Even simple elements like curves and radii,
> that are connected in the modeller, are not any
> more "real" connected in gambit and connected
> virtual geometry is bad for further operations.

Gambit's healer can connect "real" edges in many cases. It depends on the quality of the imported IGES model. I highly recommend that you contact your local Fluent office or distributor for help. (You did not include your email address, so I have no way of knowing your company/university and your location.) We have experts all over the world who can help with such difficulties.

> 2.) The "revolve" only works with planar lines or
> surfaces. Why ?

This is not generally true. Here are Gambit journal commands to create a nonplanar curve and revolve it.

vertex create coordinates 0 0 0 vertex create coordinates 0 1 1 vertex create coordinates -1 2 0 vertex create coordinates 0 3 0.5 edge create nurbs "vertex.1" "vertex.2" "vertex.3" "vertex.4" interpolate face create revolve "edge.1" dangle 45 vector 0 0 1 origin 0 0 0

There may be a problem with the specific curve or surface you are using. If you contact your local Fluent office/distributor, they will gladly investigate.

Also note that virtual and faceted edges can be swept and revolved in Gambit 1.1, i.e., you are not limited to "real" edges.

> 3.) The "split surface" works sometimes and
> sometimes not(error) and has total different
> ways to work in real and virtual. Why can't
> a virtual face split a virtual face ?

The "real" split operations are implemented with the Acis geometry kernel. We cannot rely on the Acis kernel to perform the booleans and splits on virtual and faceted entities because Acis does not understand them. Virtual geometry is a topological overlay, very much like the faces you create in P-Cube. Both are used to cover-up gaps and other problems in your underlying CAD geometry. Just like you cannot split a face with a face in P-Cube, you cannot do so with virtual faces in Gambit. That said, we will be writing our own code to implement booleans and splits on virtual and faceted geometry in a future release to eliminate the differences between "real" and the other geometry types.

> 4.) Booleans don't work with real geometry,

I assume you mean virtual, not real. Please see my comments above. Also note that the Spatial IGES translator (new in Gambit 1.1) always produces all real geometry from IGES import.

> 5.) The are not enough CAD-tools.

Gambit 1.2, due in early November, has more CAD tools. These include edge/edge intersections, projection of edges onto faces, additional arc creation methods, direct creation of rectangular/circular/elliptical/polygonal faces, and more.

> Do you mesh turbomachinery succesfully with gambit ?

We are working on specific enhancements to Gambit for turbomachinery applications. The processes of flow volume creation, decomposition, and meshing will be greatly automated. I encourage you to contact your local Fluent office or distributor for more information.

Regards, Scott

"pavel" writes:
> I cannot mesh 3D surfaces, the mesh always collapses.

You may need to adjust the faceting tolerance in Edit Defaults -> Geometry -> Tolerance. Please contact your local Fluent office or distributor for assistance.

> Why cannot I revolve nonplanar surfaces?

Please see my response to "tom".

> Recently they have announced a release of TurboGambit
> which should be a special modeling tool for
> turbomachinery. I have to say that I am very skeptic
> about it.

I am sorry you are skeptical about it. I can assure you that these capabilities will work well for turbomachinery applications. Our test matrix contains many difficult cases that will fully exercise the software, and we are interested in having more cases. If you are willing, please send one of your cases to your local Fluent office or distributor and we will incorporate it into our test matrix to ensure that this product will work successfully on your case when it is released.