CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Difference in results using different Mesh? (http://www.cfd-online.com/Forums/fluent/29950-difference-results-using-different-mesh.html)

Julie June 27, 2002 05:36

Difference in results using different Mesh?
 
Dear All,

I wonder if there will be a difference in results using hex-submap, hex-map, hex-cooper mesh?

The structure I created in gambit could not be meshed using hex-map or hex-cooper, errors like "there is no logical cube to be meshed" when using hex-map and "source faces are hard-linked" when using the hex-cooper. I do not understand what they mean, hence I did nothing to change my structure. I realise I could only mesh my structure using tri-pave or hex-submap.

Can anyone tell me the difference in results between the difference type of mesh method used, and what mesh is better prefered if I am doing a DPM in a rectangular room? Also, if anyone could tell me what the error message means. Thank you.

Mustafa Gelisli June 27, 2002 15:44

Re: Difference in results using different Mesh?
 
Hi Julie,

If your geometry is a rectangle you should be able to create a hex-map mesh. Hex-map mesher algorithm consider your geometry as a logical cube. it meshes the mutual faces using same interval size and than meshes the volume. When your geometry differs from a logical rectangle (for example, when you cut off o corner of the rectangle or you substract another volume from yor geometry etc...) Gambit gives "there is no logical cube to be meshed" error. That means you cannot mesh the geometry using hex-map scheme unless you divide your geometry into smaller- mappable volumes.

Secondly, cooper scheme consider your geometry as a logical barrel. When you first mesh one circular face of the barrel and then use cooper scheme, cooper imprint your existing mesh on the other circular face , meshes other faces(side faces) using map or sub map scheme and then meshes the volume. In your example you probably make hardlinks beetween your source faces(circular faces o logical barrel), so you have to unlink these face meshes before using cooper scheme. You can do that using mesh-face mesh-unlink face mesh command.

Finally, hexa map mesh is always gives more accurate results as long as you take care of mesh sizing. If you use adequate mesh in the high gradient regions, you will the most accurate results using hex-map scheme

Good Luck

Julie June 27, 2002 23:31

Re: Difference in results using different Mesh?
 
Hey Mustafa,

Thanks for responding, actually my model is not really just a rectangular block. It is a rectangular room with openings/vents at the ceiling. I created the vents using create volume->split volume->unite volume->delete volume but retain lower geometry->split volume with face to create openings at the ceiling of the rectangular volume

Also in this room, there is to be a partition wall.At first I create this wall by creating a face then split the volume of the room with this face. However when using cooper mesh I get error like source faces are hard-linked. The only mesh I could use was tri/pave

Then I modify the model by creating the partition wall using creating a thin volume and then subtract it from the room volume. And the only hex mesh I could use without getting any error message is submap

I have never use submap before, and I was skeptical about it. I was afraid that if I use this mesh I may run into problems later if I proceed on to DPM in fluent. Is submap reliable, as in if I ever run into any problems in solving in fluent, will it due to the fact that I have used a secondary mesh (submap) or as long as I could export the mesh successfully into fluent, no matter what kinda of mesh I should not have errors during solving ?

Mustafa Gelisli June 28, 2002 03:36

Re: Difference in results using different Mesh?
 
Hi Julie,

It is very difficult to say how to mesh your geometry unless I have it, but I can say that Submap is a very reliable meshing scheme. It is actually a different kind of hex-map scheme. You can obtain very accurate results using these scheme as long as you create adequately fine mesh in possible high gradient regions. Of course, I assume that your boundary conditions are ok.If you have a little bit more time to mesh your geometry, you can send me dbs file(only geometry please, not mesh) and I may be able to make another suggestions(But probably in 3 days maybe more- I am very busy nowadays).

Julie June 29, 2002 03:57

Re: Difference in results using different Mesh?
 
Hey thanks! I will send u my dbs file using the email next to your name.

Martin Shanel July 1, 2002 08:13

Re: Difference in results using different Mesh?
 
Hi Julie, if you want hex-map mesh so badly, why don't you split your volume into several smaller ones that can be meshed with hex-map? It's simple as that ... Good luck. Martin.

Mustafa Gelisli July 2, 2002 09:49

Re: Difference in results using different Mesh?
 
Hi Julie,

I have recieved your mails containing Gambit files but I could not open the files in Gambit. I don't know what the problem is (It may be the operating system. I use Win2000). Could you please send them to me again.(please first delete your mesh and then zip your dbs file to decrease the size of the files)

Regards, Mustafa

Julie July 2, 2002 10:28

Re: Difference in results using different Mesh?
 
yah, perhaps its the operating system. I am using Unix station in school.

Anyways I will send u them to u again tomorrow. But if you can't open them, then guess there is no other way then.Thanks.


All times are GMT -4. The time now is 11:55.