CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Virtual/Real geometry. (

Jack Keays June 13, 2000 11:27

Virtual/Real geometry.

I am wondering why it is often recomended to use real geometry in Gambit and try not to use virtual geometry. I ask, because I have an IGES file I imported into Gambit which will consist of mainly Virtual geometry due to the nature of the model. Also, what advantages does one type of geometry have over the other. Virtual and Real have confused me since I began using FLuent, and I would appreciate any replies. Jack.

John C. Chien June 13, 2000 12:22

Re: Virtual/Real geometry.
(1). I am not currently using the GAMBIT, but I have with me the training notes. I will just copy down the notes for you here. (2).Virtual geometry is typically not needed when geometry is created from scratch within GAMBIT. (so, the one you created from scratch is REAL geometry, using ACIS -a library of geometry creation modification routines. I would say, it is just geometry) (3). Virtual(FluentInc. propreitary) is another library of geometry creation modification routines providing additional functionality. (To do what?) merging edges together within a specified tolerance, for example, which CAN NOT be done under Real ACIS library routines. (4). Virtual geometry do not possess their own geometrical descriptions, they derive their geometrical description by "reference to one or more real entities(called the Host)".(5). When to use Virtual geometry? "Clean up Imported Geometry", "Simplify Geometry", "Decompose Geometry", " Modify the Mesh". (In other words, if it is not created in-house, it is not real, so use Virtual library to handle it).(6). When an IGES file is imported into GAMBIT , it is converted into real geometry except for trimmed surfaces which are usually converted to virtual geometry. The geometry usually does not get connected together using the ACIS tolerance of 1E-06(since most CAD packages use a tolerance of 1E-03). (7). Can not perform Boolean operations and other real geometry operations on Virtual geometry. (well, there is no need to re-invent the wheel. So, it is a good idea to build your own real geometry.)

Jurek June 13, 2000 13:39

Re: Virtual/Real geometry.
If you know exact what you want to do and what gambit is able to do with virtual geometry, then work with it. But even if you are a "virtual professional" sometimes there occure situations where you cannot even copy a vertex any more; then the whole model is corrupt. The best is if you construct your geometry in gambit with arcs, splines, surfaces which you built up on your imported geometry.

Aaron J. Bird June 13, 2000 15:55

Re: Virtual/Real geometry.
I agree. I have had success using the import as a template, then rebuilding in Gambit using real geometry.

jkc June 14, 2000 00:10

Re: Virtual/Real geometry.
The difficulty with Virtual Geometry is it often becomes a one way street. Once you introduce virtual geometry it becomes difficult to make modifications to the "real" geometry. Virtual Geomtry is required in certain cases but it is best used as the final step in your model creation and if at all possible should be avoided

Jurek June 14, 2000 00:18

Re: Virtual/Real geometry.
I also always make a complete real model and at the end (when I definitely do not have to make any construction anymore) I merge surfaces to avoit small angles and merge edges.

Jack Keays June 15, 2000 07:08

Re: Virtual/Real geometry.
I guess I should import my real model as a set of edges and vertices and leave out the surfaces. Would I then be able to use arcs, splines and surfaces to keep it real. My biggest problem is that my model is complex. It imvolves large numbers of curved surfaces. I am modelling a pump with an unusual single vane impeller. It is for pumpig wastewater/solids. ANyway, ye firmly believe that itr would be best to construct the geometry in Gambit using a basic imported geometry? Also, thanks for the help, I now see why virtual geometry should be avioded. I knew it caused problems with subsequent boolean operations, but I didn't think it would be too difficult to get around it! How wrong I was!

Aaron J. Bird June 15, 2000 08:21

Re: Virtual/Real geometry.
Jack, Though it may be complicated, what is the possibility of first simplifying your geometry -in order to keep it real- solving the transient flow to convergence (including grid convergence and validation if possible) then piece-by-piece adding the unusual geometric forms and subsequently solving those to convergence? Each of these solutions may not necessarily be the flow you desire, but you will be able to maintain good mesh and work toward the desired grid knowing that your solutions of the simpler geometries are mesh independent. At some point you may find you will need to use a virtual geometry for a particularly complicated or unusual part, but hopefully it will be limited and then you would have greater confidence that the geometry and its mesh are giving you what you desire. There may be faster ways to try, but I am not skilled enough to describe them in detail, so the above is what I would do. Good luck to you.


Daniel Malone June 15, 2000 15:46

Re: Virtual/Real geometry.

I operate a business, CAD Repair, that heals, repairs and simplifies CAD neutral files. We can do model defeaturing and we can increase the resolution (tolerance) of the CAD model to make your job easier. Our website is

For instance we can provide several versons of your original geometry with different entities defeatured so that you can import them into GAMBIT to do a trade study.

Scott Gilmore June 15, 2000 23:39

Re: Virtual/Real geometry.
Virtual geometry is a topological overlay, sitting "on top of" underlying CAD geometry. For those of you who had used GeoMesh, virtual in Gambit is very analogous to what you created in P-Cube. You could not do booleans in P-Cube, just as you cannot do with virtual in Gambit.

Virtual geometry's best use is defeaturing---removing insignificant details, sliver surfaces, etc., by merging surfaces together. This is a very automatic, convenient process. This is usually the last step before meshing, when booleans are no longer needed.

Virtual can also be used to close gaps by connecting edges within tolerance. This is sometimes the only choice for very-dirty CAD surface geometry. However, healing in Gambit 1.1 and later can close the gaps in reasonable cases.

Note that IGES import in Gambit no longer creates virtual geometry. This was the case in Gambit 1.0, but the new "Spatial" translator in Gambit 1.1 and later always creates "real" geometry, even for curved surfaces.

Beginning with Gambit 1.3, to be available for download very soon and on CD in July, you can import solid geometry in Parasolid and STEP formats. These choices are far superior to IGES.

Finally, virtual edges can be swept to form real faces, and virtual faces that are mappable can be converted to real faces in Gambit 1.2.4 and later.

All times are GMT -4. The time now is 18:54.