How can we export a complete mesh that can be imported in GAMBIT from a .cas and .dat file?

File/Import/Mesh --> Type Fluent

Do you mean import a .cas file? Yes I've already tried but it doesn't work. I get am error: Gambit is unable to open...

Don't we need to export the fluent grid in a .msh or dbs file??? so that it can be read in Gambit

Please help me.

no you can't export in *.msh or *.dbs how did you import the mesh file into fluent?

I suppose with a .dbs or .msh file from Gambit... The point is that I don't know because I took directly a .cas and .dat file from a colleague who doesn't work in my office anymore and nobody knows where the Gambit grid is... Now I'd like to know how to import this grid in Gambit again from a .cas file to modify it...

ok, I create a simple cube in gambit, and I meshed it and exported it in hexa, tetra, tetra-hexcore and I converted also the last one with "tpoly" utility. I could re-import the *.cas in Gambit for hexa, and tetra meshes. With the 2 others, it is not possible, because of the hanging-nodes , and/or polyedral cells. So I assume your mesh contains hanging nodes or polyedra. Check with your support, for fixing this problem. Sorry for not being able to help you furthermore

Check if you can export the mesh in CGNS format. If yes, try to import it, into gambit (this is in case of hanging nodes (no polyedra))

Yes it does contains polyedral cells... So in all the cases, thanks for the info. I'll look at the problem more deeply with the support

Polyhedral or hanging node (eg adapted) meshes cannot be read back into TGrid or Gambit. Fluent case files can be read into Gambit (assuming neither of the above) by writing out case as unzipped and in ASCII format. Then use File>Read>Mesh in Gambit to read in mesh into Gambit and virtual geometry will be created from mesh (creates faceted geometry). You may be able to convert to real but this depends on the case. You can unadapt adapted meshes in Fluent back to original density before transfer but polyhedral meshes cannot be converted back so need to be saved before conversion (even based on skewness only changes or a tploy'd hexcore mesh).