I've been playing with Gambit for the last few weeks and I've quickly realised that there is a huge potential in the journal file concept. However, due to the many bugs still present in Gambit and also due to the very limited journal file language I've chosen to write perl scripts that generate journal files instead of just writing journal files directly. This makes it much easier to do parametric studies, alter modules to fit new CAD geometries etc (the journal files are dependent on the numbering of CAD elements - makes it very difficult to write general journal files).

If we could share these journal files or rather, the scripts that generate journal files, in a public database (perhaps here at CFD Online) this could be a potentially very useful resource. My question is really if there are any Gambit users out there who would be willing to share their journal-file modules with other users? I could set up this database here at CFD Online if there is an interestet and need for it.

Hi,

some questions about Gambit.

What is Gambit? What is it used for? How can I learn more about Gambit?

Ciao

Heinz

Gambit is Fluent's brand new preprocessor. It includes basic geometry creation tools based on the ACIS CAD kernel (same as Autocad uses) and mesh generation tools based on Fluent's own methods (T-grid etc.) ... all packaged with a nice "windows" like GUI. It supports most types of meshes and has the ability to work with "journal files" - a kind of programming language which is used to control gambit. You can record your work, save the generated journal file and then perhaps alter something in the journal file and re-run it in gambit. This makes it possible to easily to parametric studies, create general modules capable of automatically gridding certain type of geometries etc.

The best way to learn about Gambit is probably to attend one of Fluent's courses. They also have a bunch of nice tutorials. Be warned though, Gambit is still a bit too buggy to be really useful, although it seems very promising.

We have the journal file concept applied in MegaCads since 1994 and I made the experience that our scripts had to be documented very very well. Otherwise it is difficult to apply them to other configurations. May be that's easier with Gambit.

Probably true, I belive that you need a mix of interactivity and automation via scripts. Having script files that can interact with the user via a GUI would open up a whole lot of new possibilites. Gambit currently lacks this ability. Can MegaCads to that?

I'm afraid that I'm not very familiar with MegaCads. How many users does MegaCads have? Would you be interested in having a public scripts database here at CFD Online?

A database is a good idea. I have been using Gambit, since it was introduced - and it is a great program. Odds are, it will be more popular in the near future.

I am not quite sure what you are going to do with the GAMBIT journal files. But you did mention two things, one is "easier to do parametric studies" and the other is "alter modules to fit new CAD geometries etc". So, you are thinking that several different geometry models can be recreated by using the modified journal files ( which was originally created when you interactively put together the geometry model). For a simple geometry such as a cube or a sphere, you may be able to go back to the journal file ( ASCII script file), find the location where the dimension is defined and replace it by a variable name. In this way, you can change the value of the variable at the begining of the file in order to create a different geometry. I have not tried this approach yet. For complex geometry, since there are many different ways to put together the same final geometry model, it may not be practical to use someone else's journal file to create geometry models. For product design, the geometry data would normally be considered sensitive information. And a CAD designer's experience is normally measured by the total number of hours of hands-on experience. At least a couple of thousand hours are required for a 3-D CAD designer to do a good job. In short, even if there is such a data base for parametric geometry generation, You almost have to steal it in order to use it.

Jonas et al.,

I am sorry for jumping into this discussion so late. I'd like to address a few of the comments made previously by various individuals.

I think the idea of a repository for Gambit journals and related scripts is a great one. Certainly, as John Chien later pointed out, not all models can be easily parameterized and many are proprietary. Nevertheless, a reasonable number can probably be shared with others. Even if a journal (or related script) is not of use for your application, you may learn interesting and useful tricks from it. I would be very willing to explore the possibility of hosting the archive on fluent.com if Jonas prefers not to host it on cfd-online.

On the topic of journal interactivity (raised later by Jonas), we have received numerous requests from customers for such features. For example, allow a journal to popup a form in which the user would enter parameter values. Another example is a yes/no dialog allowing the user to interactively select a branch in the journal. We agree completely that there is much potential in adding such features, and hope to make some progress in this area as Gambit matures.

On the topic of stability, I am disappointed to learn that Jonas is still having problems. Surely, Gambit is immature and has instabilities in some areas, but we are making great strides with each new release. The latest release is 1.0.3. If you have a prior release, I encourage you to contact your local Fluent office for the latest. (They can provide download instructions if you prefer to get it from our ftp server. The instructions are too long to include here.) We are putting the finishing touches on 1.0.4, and expect it to become available in March.

I also strongly encourage you to report each and every bug and request to your local Fluent office. We take these reports very, very seriously. They are immediately seen by our development team as soon as they are entered into our database. The Gambit development manager immediately reviews the bug/request and assigns it to the appropriate developer. Once a bug is fixed, the person who logged it in our database (usually your technical support engineer in your local office) is immediately notified and can share the news with you.

Best regards, Scott Gilmore Fluent Inc.

I have used the original FLUENT/preBFC to create simple and very complex 3-D geometry models. So, when I went through a three-day training course, I realized that GAMBIT is not a simple program. It has (1). surface modeling ( traditional point, curve, and surface capability), (2). solid modeling (using geometry primitives to create more complex geometry through boolean operations), (3). CAD type editing capabilities, (4). controlled and automatic meshing ,(5). compatibility with FLUENT family of codes. (6). user-friendly GUI. I think, GAMBIT is a professional product. But, this also changed the environment in a different way. For example, a new user of FLUENT solvers normally would ask a designer (CAD) to create the geometry first. When it is done, he would ask the designer to export the IGES file to a mesh generation code so that he can easily distribute nodal points along a line segment and so forth. He can then create the mesh there or send it to the Tgrid to generate the mesh. And finaly, he can set up the boundary conditions, turbulence modelers, control parameters in the solver with the finished mesh. Therefore, his main job was to distribute nodal points on curves, which is not hard to do; and to set up the solver, which is not hard in FLUENTS solver at all. So the life was easy. With GAMBIT, the life is different. He must work as a designer and a CFD engineer. An average engineer can take just a few days to learn how to distribute points and set up solver parameters, BUT IT IS GOING TO TAKE YEARS TO BE A GOOD CAD DESIGNER. Very few CFD engineer can use PREbfc , keyboard and mouse to create very complex geometry models. So the question is: Is he going to learn 3-D CAD first, or Is he going to ask his old friend (CAD designer) to learn GAMBIT and build the model for him? ( maybe it is easier for a CAD designer to learn GAMBIT and become a CFD engineer. or is the CFD engineer going to take the design job away from the CAD designer?)

You can ask you CAD designer to export the model in the ACIS format. Then read into gambit and you still have a "real" geometry.

This format describes you model much better than IGES or STL because there are no tolerance problems.

A Question to the developers of Gambit: are ther plans to support parasolid on day or to switch the geometry kernel?

I've asked the same question to Fluent and the answer was no, no plans to switch from ACIS for the moment, but if parasolid continues to gain market share they might of course reconsider in the future.

Joern was correct to point out that you can still have the CAD designer build your geometries after switching from PreBFC to Gambit. However, not all CAD systems can directly export ACIS files. All can export IGES, and Gambit's IGES import works well for many models. We are working very hard to improve Gambit's geometry import. For example, we are testing a new "FTL" import for IDEAS models which has a lot of potential. We expect to deliver other major enhancements in the coming months. In addition, some fairly good direct translators are available for converting various CAD formats (e.g., Pro/E, Parasolid) directly to ACIS; your local Fluent office or distributor can provide more details.

If you prefer to mesh your surfaces in an outside CAD system, you can import them into Gambit just as you can into TGrid. Once inside Gambit, you can immediately generate the volume mesh, or replace the original surface meshes with new ones. (For example, import a triangular surface mesh and replace it with a quad mesh, then fill the volume with hexahedra. The original triangles define the geometry.)

Regarding Parasolid, we are actively and seriously considering direct support for it.

Scott Gilmore, Fluent Inc.

MegaCads has only a simple interactive editor to edit the script (MegaEdit). It's an additional program which will be started from MegaCads. But it is currently only in a very simple way coupled to the entities in the GUI. To my knowledge less than 25 people use MegaCads. But that may change because the executables are now freely available. So, currently there might be no interest in a public script library.

I should add how the script technique works in MegaCads. After/During the interactive grid generation construction, you can call the editor, which can show you the history of a selected entity. With a double click a parameter window opens and you can modify parameters. You can also delete or insert processes. The script handles variables for parameters (.e.g. file names, dimensions, etc.), has simple mathematical functions (e.g. arc length). Loops are possible to call several grid generations processes again, and 'sub-scripts' are also allowed, to use common grid construction procedures again. That is available since MegaCads 2.0 .