CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   stoopid discontinuity (https://www.cfd-online.com/Forums/fluent/42518-stoopid-discontinuity.html)

Matthew R October 10, 2006 07:04

stoopid discontinuity
 
Hi all, i'm quite the novice at CFD. I've used Gambit to generate my mesh but i have a really frustrating problem. When i import the mesh to Fluent and run a few iterations i see that flow does not occur across the boundarys of different volumes. ie flow goes into one section and stops as if it thinks it is a solid wall. For the life of me i can't determine how to fix this! I don't think i need to Fuse in Fluent. Strangely, if i do a little test in a new file with 2 simple volumes with different meshing and run it in Fluent this type of discontinuity doesn't exist, so why in my big mesh? It is very very frustrating...

Regards, Matt

manoj October 10, 2006 07:34

Re: stoopid discontinuity
 
it is possible bcz of wall between two volume u can ensure same by checking grid in fluent. how to slove it? 1) u can unite all geometry mesh it get the result 2)if u have used split command then check whether connected option is ON or not. hope this may be useful.

Jason October 10, 2006 07:54

Re: stoopid discontinuity
 
The faces between the geometry aren't connected... you have two options...

1) connect all the overlapping faces (i.e. the faces where the two volumes should share)... if you have all real geometry, you can highlight everything and hit connect... then manually check each face to make sure it's connected (i.e., there is only one face where the two volumes come together). You'll have to remesh, because currently the faces that you're going to connect have different meshes, so when you connect the faces, one of the face meshes will be destroyed

2) use non-conformal interfaces. That's the simplest. But non-conformal interfaces pose their own problems (i.e. numerical problems).

Hope this helps, and good luck, Jason

Matthew R October 10, 2006 22:42

Re: stoopid discontinuity
 
Hi guys thanks for the replies. Still battling with it however. I have tried the connect faces thing, but it hasn't worked. I think it's due to the nature of my geometry. Basically the problem occurs at the boundary of two cubes with the same dimensions, except the bottom cube has beveled corners. This mean that there are two faces at the interface: one square face and one square with the corners cut off. Can't Gambit connect these (connect the parts that do overlap) someway?

Also, why can't we attach a picture!

Regards, Matt

sach October 11, 2006 01:42

Re: stoopid discontinuity
 
I think your problem is; you have virtual volumes so it may not split in gambit the best way is non confirmal interfaces. do one check I think there may be omore than two volumes or dummy faces one should do the following once complete the volume mesh its best practice to confirm the 1) no of volumes present 2) delete the faces from entire geometry this will delete the hanging faces

Matthew R October 11, 2006 04:31

Re: stoopid discontinuity
 
Hmmm, i'm learning a lot..

What i was doing was meshing a volume, and then meshing the volume on top of it and thinking they would be connected...but i was using two different faces at the interface! I needed to construct the top volume using the face from the bottom for them to connect.

Of course once i figured this out, i found that the nice quad mesh that i could previously use would not work, since the mesh now had to work with the meshed face on the other volume.

What i have done is used a small amount of tet mesh as an interface between the two volumes, this enables me to use quad mesh for the rest of the shape...is this valid? Is this a better way than using something like Tgrid or the non-conformal meshes i've read about to make the interface?

Thanks again for the replies...i'm such a novice!

regards, -matt

Jason October 11, 2006 07:56

Re: stoopid discontinuity
 
I guess I should've asked for more details when telling you to connect faces, but I assumed that you had identical faces where the volumes come together.

Gambit can only connect faces if they are coincident, have coincident edges, and coincident vertices (i.e. are copies of one another within the tolerance).

If the faces are different, then use the non-conformal interface. If your mesh size is refined enough and consistent across the boundary (you can even use tet on one side and hex on the other... as long as they're of similar size).

The only other option is to split the larger face with the smaller one. Now the large face has a chunk cut from it that looks exactly like the small face. You can connect these, and then mesh.

There's a lot of other information on the forum about connected geometry, non-conformal meshes, etc. Do a search and see what you find.

Good luck, Jason

Matthew R October 11, 2006 09:37

Re: stoopid discontinuity
 
thanks for the help..looks like its finally working, -matt

sach October 11, 2006 22:49

Re: stoopid discontinuity
 
one more suggestion; its not so that you have to extrude the same face please see the details of volume split operation (its a binary operation )


All times are GMT -4. The time now is 08:05.