CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Scale Problem and Moving Walls in PCube/TGrid/Fluent (http://www.cfd-online.com/Forums/main/1303-scale-problem-moving-walls-pcube-tgrid-fluent.html)

Christian Tollschein September 17, 1999 03:34

Scale Problem and Moving Walls in PCube/TGrid/Fluent
 
I tried to create an unstructured grid for a komplex geometry, which consisted of elements ranging from 0.1 mm to 20 meters. TGrid was not able to create the volume mesh in the area, where the small elements were located. Does anybody know how to solve this problem?

Secondly, this geometry was like a conveyer belt pulled through an acid bath and the simulation should show how the acid moves with the belt because of frictional forces. How can this moving wall problem be solved with Fluent?

Thanks a lot. Christian Tollschein

Joern Beilke September 17, 1999 04:49

Re: Scale Problem and Moving Walls in PCube/TGrid/Fluent
 
For the first part: If you want a tet mesh then switch to ICEM Tetra which works much better.


Jonas Larsson September 17, 1999 06:45

Re: Scale Problem and Moving Walls in PCube/TGrid/Fluent
 
If the moving wall is flat and moves at a constant velocity parallell to the wall you can just specify this when you set up your wall boundary condition in Fluent.

About the TGrid problem. TGrid can surely generate grids with varying cell size. Generating tets only is usually no problem. Growing prisms can be a bit tricky. Often it takes some trial-and-error and you have to adjust the surface meshes slightly a few times before it works.

Jonas Larsson September 17, 1999 06:50

Re: Scale Problem and Moving Walls in PCube/TGrid/Fluent
 
This is a very interesting statement. Could you perhaps elaborate a bit on when ICEM Tetra works better, which problems you had in TGrid that Tetra solved etc. I'm sure that many people would be interested in this.

I have only experience from TGrid and, sure there are problems, but generally I've found it to be a reasonably stable and robust volume mesh-generator.

Joern Beilke September 17, 1999 08:43

Re: Scale Problem and Moving Walls in PCube/TGrid/Fluent
 
There is no need to create a surface mesh in ICEM Tetra. This might save you a lot time. You just say how big the biggets cell should be and how small the smallest cell should be and also how fast the transistion between small and big is wanted.

These options can be specified either global or region wise. Thats nearly all you have to do. After this the mesh can be smoothed an the prism layers will be included.

Complex meshes can be created in less than 1 hour after cleaning the cad surface.

Just ask ICEM for a trial license.

Jonas Larsson September 17, 1999 10:11

Re: Scale Problem and Moving Walls in PCube/TGrid/Fluent
 
Sounds nice. How well does it handle IGES files with bad precision and no solids (only surfaces)? Is it stable? Is the prism-generation working better than it does in TGrid?

Jin Wook LEE September 17, 1999 10:12

Re: Scale Problem and Moving Walls in PCube/TGrid/Fluent
 
Fluent has the cability to model very thin wall, in the computational domain, which can be treated as zero thickness. You don't need to create volume for the very thin wall. Instead, treat it zero thickness wall. I have had several experiences which is very similar to your problem, but for stationary walls. I guess the concept of thin wall in Fluent might work for the moving wall. Ask support group of Fluent company whether 'the concept of thin wall in Fluent' can be applied for moving wall or not'. If they say 'YES', I think it is just for your case and it can save your effort and computational time very much.

Sincerely, Jinwook


John C. Chien September 17, 1999 12:18

Re: Scale Problem and Moving Walls in PCube/TGrid/Fluent
 
(1). I agree with you that it is not easy to create un-structured mesh automatically when the geometry is very complex. (2). Based on my experience, the volume mesh would fail when the mesh generating process was nearly complete, if the surface mesh is not good enough. But this was because my geometry was extremely complex. (3). The only way around this problem is to cut the domain into smaller pieces first. And then try to generate good surface mesh for each bolcks. Good surface meshes are essential for the successful generation of the volume mesh. I think, there are lot of rooms for good automatic mesh generation code.

Joern Beilke September 17, 1999 13:01

Re: Scale Problem and Moving Walls in PCube/TGrid/Fluent
 
I tried Tetra some time ago, so things may have changed a bit.

Surface meshes are just fine if they come from a solid modelling cad package. I always used stl-files, but IGES works (some people told me that they only use iges).

Bad iges files (from catia) will always cause some problems and additional work. But they should have a cad repair tool by now.

The prism mesher worked when we tried it but I have no comparison to Tgrid there.

It is very important to attend a course, because ICEM Tetra/Hexa/... are very powerfull tools.

They also have native interfaces to some cad packages (unigraphics, proe, ??). There are no more problems with a bad surface description any more because it works directly on the cad data.



All times are GMT -4. The time now is 14:26.