CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] Pyramid or very low prism elemet (

juliom February 21, 2013 16:34

Pyramid or very low prism elemet
3 Attachment(s)
Dear all,
I have been working with ICEM CFD for allmost one year, so I consider myself a "fresher" in ICEM CFD or any other mesher.
I am working with a engine intake, but my mesh is a mess.
First I did a surface mesh using the pach independent method, after I smooth it, I got a volume mesh using Delaunay ( least quality about 0.4, so ok :) )
At the end of my checked and smoth volume mesh, I tried to get good prism elements but without luck.

I tried using surface mesh and curve mesh setup parameters in order to get a smaller element where I had these low elements (0 - 0.4 :mad:) nothing worked.

Then I tryed to change the ortho weight and so forth without luck, could you please give a hint??


Dear Friends something I missed out was to comment about a type of merge of different parts of the domain, I upload a file to show you.

BrolY February 22, 2013 05:22

PSYMN February 25, 2013 19:30

Based on what I can see in those pics, it looks like your issues are "curve" related.

ICEM CFD patch independent assumes you want nodes along your curves and it will force your mesh to align. It also assumes that you don't care about surface intersections if you don't include a curve...

It looks like you may have a curve along your symmetry plane that is forcing your prisms there to align and resulting in poor quality. If you don't need a curve, delete it, then mesh again.

It looks like you do not have a curve where the small cylinder meets the elbow. This means that you don't have a nice sharp feature line where the two sides meet and your prism will have a hard time negotiating that curve.

One good option would be to use "Geometry => Repair => Build Diagnostic Topology". Set a filter angle to something like 30 degrees and turn on the option to filter curves and points. This will remove curves and points where surfaces meet at less than 30 degrees (such as on your symmetry plane) and it will include curves where they meet at greater than 30 degrees, such as between the pipe and the elbow.

juliom February 26, 2013 08:59

Mr. Simon,
I really appreciate your time answering my question.
Indeed I used Geometry => Repair => Build Diagnostic Topology.
Since I am using a patch independent procedure I thought I wouldn´t hav eproblem ith this issue. I haven`t ever used a patach dependent process due to that I am not clomptley clear about this proceess of clean up CAD.
I alwways perfomr the topology check, deleting all "non red curves"
I need the curve along the symmetry plane, where both tubes are intersected, because is part of the topology of the domain.
Other thing I was looking at is that teh quality of the prism near by this intersection is about (0.1-0.4), I have read that due to the fact that the problem is located in a small part of the entire domain the numerical problem will not spread to the entire domain and it will not affect the global numerical performance.
BUT at the moment I see the aspect ratio of elements, I realized that the aspect ratio is very bad below to 0.2, I have read that AR must be always above of 0.2. I have tryed to use a curve parameter in the entire domain in order to make the prism length small, but without luck, I guess i need a super computer to get a very small tetra element in order to keep the AR balanced.
Any other suggestion ???

All times are GMT -4. The time now is 20:34.