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/)
-   -   [ANSYS Meshing] Hex meshing producing poor quality elements on multibody part (http://www.cfd-online.com/Forums/ansys-meshing/84726-hex-meshing-producing-poor-quality-elements-multibody-part.html)

aarvay February 7, 2011 13:09

Hex meshing producing poor quality elements on multibody part
 
I'm attempting to do a fully hexahedral volume mesh on geometry for a fuel cell. Here are some reference images for the geometry.
http://lljk.org/cfd/reference.png
http://lljk.org/cfd/geo1.png
http://lljk.org/cfd/channel3.png
http://lljk.org/cfd/channel2.png

Ive been somewhat successful in producing the mesh. It generates a very nice looking surface mesh (http://lljk.org/cfd/surfacemesh.png) and I am trying to give the middle sections (sections 3-7 in the reference picture) thin layers. I was able to sweep the middle most 3 layers (4, 5, 6) and apply # of divisions control sizing. But the outer inner layers ( 7 and 3) cannot be swept. I've tried applying edge sizing but that doesn't work as shown here http://lljk.org/cfd/edgefail.png. the edge sizing works only in that local area and doesn't split the entire body. i want those layers (3 and 7) to be cut just like the middle layers as shown in this image http://lljk.org/cfd/sweep.png.

In addition to that, the volume elements inside look like hell. it throws a warning about exceeding the aspect ratio warning limit. I think its pretty clearly demonstrated in this cut plane. http://lljk.org/cfd/crap.png

So I have two important questions. First, how can I control the thickness of the elements in those inner layers that cannot be swept and edge sizing does not work.

Second, how can I get rid of those high aspect ratio elements and have a more uniform hex cells.

I'm pretty new to modeling and meshing so I may be taking the wrong approach. What I have done was apply sweeping method to the bodies that can be swept (2,4,5,6,8) and applied hex dominant method to the others. its also vital that the mesh be conformal between bodies.

Thanks for any info

PSYMN February 8, 2011 14:08

Have you tried multizone on this assembly...

It is the Hexadominant mesher that is giving you the crap, but the Multizone mesher may be able to imprint and hex mesh much more of this model.

aarvay February 8, 2011 17:45

Thanks for the tip.

Initially I tried to use multizone but it failed due to bad input parameters so I gave up and tried this method of hex dominant and sweep. Looking back through the instructions I think its a bit more clear what I need to do so I'll give the multizone method another try.

aarvay February 9, 2011 19:35

multizone doesn't work
 
Well I tried implementing a multizone strategy in a number of ways but i couldn't get it to work. I was able to get it to work on each individual body separately but then when i combined them all together the process failed. I think that the interface between the current collector and the GDL counts as parallel loops. Whatever the reason, multizone isn't happening. Workbench meshing worked all right for very simple fuel cell geometries but now that I'm trying to mesh more complicated ones I think I'm going to have to figure out how to use ICEM CFD, that seems like it would be more suited to meshing my highly square geometry.

carlp March 12, 2012 00:58

If you still working on it, I think pinch-control might solve a part of the problem...

// Meshing User's Guide // Global Mesh Controls // Defeaturing Group // Pinch

I also had problems with the multizone/hex-dominant meshing but with infaltion, but I solved it by do the multizone first, and then add one inflation at the time and just update the mesh.

Did ICEM CFD worked? I have similar problems...

aarvay March 12, 2012 01:25

Quote:

Originally Posted by carlp (Post 348852)
If you still working on it, I think pinch-control might solve a part of the problem...

// Meshing User's Guide // Global Mesh Controls // Defeaturing Group // Pinch

I also had problems with the multizone/hex-dominant meshing but with infaltion, but I solved it by do the multizone first, and then add one inflation at the time and just update the mesh.

Did ICEM CFD worked? I have similar problems...

Yeah this is kind of a thread necromancy. I don't remember what exactly was going wrong. I did end up switching to ICEMCFD and using the cartesian grid method and that fixed these problems but then added a bunch of different problems.

The problems I was talking about here were with version 12.1 of the workbench mesher, I don't recall seeing any of those options you mentioned although they may be there and I just missed them. I think in version 13 the workbench mesher got a significant upgrade and now i guess v14 came out recently. I haven't tried anything beyond 12.1 yet myself since my institution is kind of slow on the upgrades.

mrdelaunay March 16, 2012 13:09

you should go into dm, slice the model up and sweep everything. It should only take about 10 minutes to slice then reform a multibody part. The sweeper will work like a charm

aarvay March 16, 2012 15:56

you just reminded me of the problem I had with trying to mesh each part separately. I needed the mesh to be conformal at the boundaries between parts, otherwise the solver would shit itself.

PSYMN March 16, 2012 16:29

Sure, no problem...

After slicing and dicing, select all the parts (hold down CTRL for multi select) and right click to create a multi-body part in DM...

When the mesher sees the multi-body part, it will generate a conformal mesh... The slicing just helps it figure out the sweeps better.


All times are GMT -4. The time now is 14:35.