Hard-linking faces in Gambit (periodics)
 

Hello.

I'm modeling a quarter section of a tank in Gambit and need to set up a periodic boundary on the inner faces (not the tank wall face, but what would be the two 2-D planes of fluid - at a 90 degree angle to each other). I have a question about hard-linking the face meshes:

- It has to be hard-linked BEFORE I mesh the face, right?

- Also, should I only mesh one of the faces?

*If you have experience with hard-linking, could you please shed some light on how to set it up and what possibly to LOOK OUT FOR. Thank you in advance!

-Eric R.

Reverse Orientation
 

Another thing....

What is the idea of "reverse orientation." For the quarter-tank model, is this needed?

Re: Hard-linking faces in Gambit (periodics)
 

Hi Eric,

It's probably best to talk to your Fluent/Gambit support person, but let me see if I can shed some light on your queries....

1) you need to hard-link the two periodic faces before you mesh.

2) you should mesh only one of the faces (the other face should be automatically meshed by Gambit because of the hard-link)

3) The "reverse orientation" refers to the mesh mapping between the two periodic faces. What you want is for the mesh on one periodic face to be "identical" to the mesh on the other periodic face such that if you rotate the mesh about the axis of rotation it will "exactly" (to some tolerance) match its counterpart periodic mesh.

When you hard-link you need to select 1) a periodic face followed by 2) a corner on that periodic face (make note of the orientation of the arrows when you select the corner); then select 3) the opposing periodic face followed by 4) the same "periodic" corner on that face (again note the orientation of the arrows when you select this corner). In order for your mesh to turn out periodic, the two selected face/corner pairs must have the same sense of orientation (e.g. arrows should be pointed in the same direction... both toward the axis of rotation, for example). This is where the "reverse orientation" toggle comes in. It permits you to reverse the orientation on a particular corner selection.

One suggestion...

Once you've "hard-linked" your two periodic faces, select one edge to mesh; pick a "large" value of the stretching factor (say 1.5) and mesh the edge. You should see Gambit automatically mesh the opposing periodic edge with the same stretching and orientation. If not, then you'll have to unlink the faces and try again.

Hope this helps!

Phil :)

Re: Hard-linking faces in Gambit (periodics)
 

Thank you Phil, I appreciate the response. I took a look at a few of the things you told me. It reaffirmed some of the things I was hesitant about.

It's weird... Although the hard links appear successful (from the mesh-edge checks), one of the meshed faces always comes out extremely skewed.

---- I apologize in advance if the following scenario confuses you. :)

I decomposed areas of the quarter-tank with the HEX/COOPER scheme, and I'm using the TET HYBRID around a volume for cooling coils (with a boundary layer on each coil face). Any faces of this volume that weren't meshed (by other volume's coopering) were meshed separately. The high skewness occurs on one periodic face of the volume containing the coils.

For this section, I selected each face to hard-link, selected one vertice of the 2-D periodic box surrounding the coils (from decomposing), and then had to select a vertice on the circular edge of each coil.

I'm not sure why I'm getting a very awkward and unrealistic looking skewness on one 2-D face mesh when the first one in the link is perfectly fine.

-Eric R.

Re: Hard-linking faces in Gambit (periodics)
 

Eric,

Apology accepted :)

First, a question...

Have you gone back to your model and unlinked abolutely every face, then started from the beginning and relinked each face; making sure that you pick corresponding periodic faces and one corresponding periodic vertex from each of the two periodic faces? Also, making sure that the orientation of vectors associated with each vertex are proper for periodicity? Sometimes I thought that I tripped the radio button for "reverse orientation", but actually didn't and that has messed up my periodic faces. Also, sometimes when this happens I have to go back and unlink absolutely every face (you can just drag a box around them all to select with one mouse-click) in order to get gambit to work properly.

Secondly, some comments...

1) You are aware that you need to perform all of your hard-linking before you do any meshing, otherwise you could end up with some peculiarly unwanted grid behavior.

2) Another thing to try is to split these troublesome faces and then hard-link and mesh to see if that helps. For example, you could split each corresponding periodic face using a surface aligned with a principal plane so that you are assured of periodicity with the split faces (I'm assuming here that your model axis is aligned with one of the principal coordinate axes).

3) Finally, gambit is a "work-in-progress", and I've found numerous bugs/problems with the software in my 2+ years of using it. It has improved over the years, but still surprises me sometimes. If you can't find a work-around on your own, to your current problem, I strongly recommend packaging your problem up and ftp-ing it to your Fluent support person to have them take a look at it.

Good luck!

Phil :)

Re: Hard-linking faces in Gambit (periodics)
 

Thanks Phil.

Yeah, I am currently speaking with my Fluent rep, and sending him some files. Thank you for your help, it is appreciated. :)

-Eric

One more question.
 

Phill,

I read your following comments and found it very helpful.

I had one more question for you regarding face mesh linking. What is the "periodic" option mean, which locate just below "Reverse orientation" option?

If I choose that, while linking the periodic faces together, then what does that mean? Does it interfere with the set periodic boundary condition in "Zone" panel?

Thanks for your help in advance.

Best,

Teymour

This option is per default enabled. Thus if you need to work with periodicity, you need to apply periodics BC by picking the periodic surfaces together (just one Boundary set).
But it will only succeed if the picked surfaces are hard-linked....

from help:
*******
The Link Face Meshes command includes a Periodic option that allows you to specify that the faces are periodically linked. Periodically linked faces are constrained such that they must behave identically to each other with respect to any virtual edge-split and vertex-move operations