I am doing a 3d model of a cylindrical plug flow reactor. There is a narrow annulus reactant inlet section followed by a large diameter preheating section, then a spool with small inlets for the other reactant, then a large diameter reaction zone.

The width of the annulus is narrow enough to require a high node density to mesh the volume (I am using tet/hybrid), While the preheat zone is large enough to make the same node density impractical. If I use two different node spacings, and set an 'outflow' boundary condition after the first mesh, I get an 'underflow error' when I iterate. If i try to merge or link the meshes, GAMBIT says it cannot perform this task. Does anyone know how to unite volume meshes with different node spacings? I would greatly appreciate any help as I am having this problem with several models.

I believe that the interface of two different mesh zones has to have same mesh in order to merge them. You may want to decompose the preheat zone (large volume), so you can have a transition domain from large volume to samll volume. I am not sure your "underflow error". Are you trying to define "outflow" BC in between the domains?

I think you do not need to define the boundary conditions between the domains. Could you explain your problem more detailed?

Did you try to use the non-conformal interface boundary?

Seems you have topologically disconnected volumes in the model with multiple faces at the interface. This is ok only if you want to use a nonconformal mesh. In that case, you should define an interface boundary as suggested by Volker.

For conformal mesh, you should do a "face connect" to connect coincident faces at the interface. Then remesh the volumes. You do not need to define the boundary condition at the interface, which should be of interior type.

Thanks, everybody, for your suggestions. I used a nonconformal grid and that seems to be the fix. I am currently working on the reactor I described and a couple of other aerosol reactor geometries for my M.S. project, and I am going to put them on my website if it would be interesting to anybody. I'll post the URL here when I am done.

Thanks,

Brant