CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] Meshing the Fluid Volume of a Fibrous Membrane/Media--Help with errors/settings? (http://www.cfd-online.com/Forums/ansys-meshing/120514-meshing-fluid-volume-fibrous-membrane-media-help-errors-settings.html)

ANT July 8, 2013 18:18

Meshing the Fluid Volume of a Fibrous Membrane/Media--Help with errors/settings?
 
Hello CFDOnline,

I've been struggling for what must've been weeks to script a loop of sweep commands in AutoCAD in order to generate my geometry without producing problematic geometry or errors, and that's finally done.

Now, what's left is to generate a mesh for use in Fluid-flow analysis in Fluent. With less dense fibers spaced more generously from each other, I could subtract these fibers (swept circles along paths) from a solid block in order to define my fluid volume, and then simply use a CutCell mesh from the ANSYS Meshing component.

However, when I'm working with less porous/more "dense" fibers, CutCell meshes, Tetra meshes, etc. all fail in the ANSYS Meshing component (errors about geometry tolerances and such--basically, it fails when meshing surfaces that are extremely close together, such as a fiber that's been subtracted just inside the fluid-zone boundary).

So, I've turned to ICEM, figuring that more control over the meshing would avoid these problems.

Now, I have two versions of the geometry with which I can work. The first "version" is just the fiber geometry--a whole bunch of swept bodies exported from AutoCAD in ACIS format. This is what it looks like:

http://i.imgur.com/gzb8Cho.png
http://i.imgur.com/gzb8Cho.png

Much easier, though, is to simply perform the Boolean Subtractions in AutoCAD, and then export the resulting fluid volume itself as an ACIS (.sat) file. This is what it looks like when I import this into DesignModeler in order to create the named selections I need:

http://i.imgur.com/NZ39BtB.png
http://i.imgur.com/NZ39BtB.png

This second method (doing the boolean subtraction beforehand) seems easier to me since I can add in the fiber-free inlet "buffer" zones ahead of and behind the fiber-filled area as well. The problem is, after creating even a basic OCTREE mesh in ICEM to test the result, Fluent is throwing a whole bunch of errors like:

5: WARNING: Cell with id 11487648 of thread 15 is missing face 3.
Primitive Error at Node 5: Build Grid: Aborted due to critical error.


Note: It is possible that this case or mesh file needs to be first
processed by the serial solver. To do this, please read
file into the serial solver and then save it. Next,
read saved file into the parallel solver.

4: WARNING: Cell with id 2344787 of thread 15 is missing face 3.
Primitive Error at Node 4: Build Grid: Aborted due to critical error.

Error: Build Grid: Aborted due to critical error.
Error Object: #f

Real overflow

Real underflow

Real overflow

Real underflow

Real overflow

Real underflow

No error handler available

Error: Cortex received a fatal signal (SEGMENTATION VIOLATION).
Error Object: ()


The meshed volume (well, only the surface elements visible), with the top and bottom hidden, looks as follows:

http://i.imgur.com/Aj07xqu.png
http://i.imgur.com/Aj07xqu.png

And here's a closeup, on which I don't see any glaring issues (holes? etc.) with which Fluent would have a problem:

http://i.imgur.com/QlxqNOT.png
http://i.imgur.com/QlxqNOT.png

The fluid volume is 100x100x500 in size, fibers are 4 units in thickness (here, everything is set in meters, I scale down to micrometers in Fluent). Some fibers slightly intersect each other where they meet, and others have a small gap (~0.5 units) between their surfaces at some points.

How can I produce a mesh that Fluent can use? Is there anything that I seem to be doing wrong? The options I have set are as follows:

http://i.imgur.com/ApsaMny.png
http://i.imgur.com/xdaapMy.png
http://i.imgur.com/c1PaBTd.png
http://i.imgur.com/ApsaMny.png
http://i.imgur.com/xdaapMy.png
http://i.imgur.com/c1PaBTd.png

Thanks very much in advance for your help!

kad July 9, 2013 10:03

Did you perform any checks in ICEM? Holes or missing faces should be detected by the the check mesh function. Also it shows you where the error occurs. Further ICEM manual states, that "Ignore wall thickness" option can lead to problems e.g. holes.

ANT July 9, 2013 16:58

Yes, I performed mesh checks/repairs; when errors came up, I chose the "Fix" option. Still, I got the same errors in Fluent about missing faces.

The CAD geometry shouldn't have any real "holes"; the geometry is simply the product of a boolean subtraction, after all. It should be airtight, since there's no apparent gaps in the geometry's surfaces. Any "holes" in the sides of the rectangular block would be formed by the cylindrical fibers passing through the outer walls, and thus these holes should align with the edges of the cylindrical surfaces passing through the block, and it looks like this is indeed the case when viewing the model surfaces.

Besides trying again with "Ignore Wall Thickness" disabled, is there anything else you would recommend? All I want is for the volume within the rectangular prism--which is bounded by the outer planar surfaces of the block as well as the surfaces of the tube-like bodies passing through--to be meshed without any odd gaps or errors being formed. I thought the Octree method would do a flood fill of this region and produce a valid mesh without any issues besides perhaps too few cells around boundary layers, which I could fix by refining the mesh.

kad July 9, 2013 19:30

What to do depends strongly on where the error occurs. So maybe you could give us a hint if it appears in critical gaps or something like that. And how many faces are missing?

For a low number of missing faces you can just create them manually. For a larger number I would try a little workaround.

First delete all shell elements. Then run check and it gives you missing faces error. Fix and put the new faces in a part like "missing". Associate mesh (only shell elements) with geometry (surfaces).

After both run checks again of cource to see if it worked.

stuart23 July 10, 2013 00:20

Hi Ant,

Try increasing your Edge Criteria for the Octree Algorithm (try 0.35 and 0.5). This has the chance of creating worse looking (serrated edges) mesh, but is usually more robust.

Good Luck,
Stu


All times are GMT -4. The time now is 12:24.