Load unstructured hex-mesh as blocking in icemcfd

korni · July 14, 2010, 07:23

Hi all,
does anybody know if it is possible to load an unstructured hex mesh as blocking in icemcfd? I found the option under the load blocking menue. But I get a segmentation fault...

Regards Michael

PSYMN · July 15, 2010, 21:45

Yup, it is possible. Not sure why you got a segmentation fault. Perhaps the resulting blocking was too complex or had unstructured regions or something and you ran out of memory while it was figuring everything out. It usually works for me and I have done some pretty interesting models.

Where did the model come from? Even though it was an unstructured mesh, did it come from a structured multi-block mesher (such as GridGen or TrueGrid)? Those usually work for me. I once had a spinal segment fail the first time (it was very complex), but there was a message (not a segmentation faliure) and it worked in the end. I don't recall what I had done wrong.

Purely Unstructured meshers may give you a mesh that can't be converted...

If you have access to TECH support, you could send it their way.

techsupp@ansys.com

Simon

korni · July 16, 2010, 05:28

Dear Simon,

thank you for your answer. It retried it on a 32 bit machine and it worked. The segfault appears only on 64 bit machines (2.6.26-2-amd64 Kernel with debian-Linux).
I want to use the pure hex mesher from Ansys12 to generate an unstructured hex mesh for a very complex case and use that mesh as blocking. Do you think there is a chance that this works?

Regards Michael

PSYMN · July 16, 2010, 17:15

I will submit a bug about this and see what testing says...

As for your particular mesh, if it can properly be broken up into mapped hexa blocks, you should be fine... If some blocks contain swept triangles or swept strange hexa patterns you may have difficulty.

Simon

PSYMN · July 21, 2010, 18:40

Hello Michael,

I checked with testing and they don't officially support debian-Linux (or test on it). Actually, this is the first I have even heard of it. However, they assure me that this functionality is well tested (manually and in our scripted regression test suite) and works on our supported platforms such as SuSE 10 64 bit and RedHat 4/5 Linux 64bit systems.

For a full list of supported platforms, please see the install guide. ICEM CFD may run on other configurations (I guess it *mostly* works on debian Linux

), but we have seen issues like this pop up.

Best regards,

Simon

July 14, 2010, 07:23	Load unstructured hex-mesh as blocking in icemcfd	#1
korni New Member Michael Kornhaas Join Date: Jun 2010 Posts: 2 Rep Power: 0	Hi all, does anybody know if it is possible to load an unstructured hex mesh as blocking in icemcfd? I found the option under the load blocking menue. But I get a segmentation fault... Regards Michael

July 15, 2010, 21:45	Should Work	#2
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	Yup, it is possible. Not sure why you got a segmentation fault. Perhaps the resulting blocking was too complex or had unstructured regions or something and you ran out of memory while it was figuring everything out. It usually works for me and I have done some pretty interesting models. Where did the model come from? Even though it was an unstructured mesh, did it come from a structured multi-block mesher (such as GridGen or TrueGrid)? Those usually work for me. I once had a spinal segment fail the first time (it was very complex), but there was a message (not a segmentation faliure) and it worked in the end. I don't recall what I had done wrong. Purely Unstructured meshers may give you a mesh that can't be converted... If you have access to TECH support, you could send it their way. techsupp@ansys.com Simon

July 16, 2010, 05:28	works on 32 bit machine	#3
korni New Member Michael Kornhaas Join Date: Jun 2010 Posts: 2 Rep Power: 0	Dear Simon, thank you for your answer. It retried it on a 32 bit machine and it worked. The segfault appears only on 64 bit machines (2.6.26-2-amd64 Kernel with debian-Linux). I want to use the pure hex mesher from Ansys12 to generate an unstructured hex mesh for a very complex case and use that mesh as blocking. Do you think there is a chance that this works? Regards Michael

July 16, 2010, 17:15	Will Submit a defect...	#4
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	I will submit a bug about this and see what testing says... As for your particular mesh, if it can properly be broken up into mapped hexa blocks, you should be fine... If some blocks contain swept triangles or swept strange hexa patterns you may have difficulty. Simon

July 21, 2010, 18:40	Unsupported platform	#5
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	Hello Michael, I checked with testing and they don't officially support debian-Linux (or test on it). Actually, this is the first I have even heard of it. However, they assure me that this functionality is well tested (manually and in our scripted regression test suite) and works on our supported platforms such as SuSE 10 64 bit and RedHat 4/5 Linux 64bit systems. For a full list of supported platforms, please see the install guide. ICEM CFD may run on other configurations (I guess it mostly works on debian Linux ), but we have seen issues like this pop up. Best regards, Simon

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Moving unstructured mesh with changing topology	meaton	OpenFOAM Running, Solving & CFD	6	April 27, 2008 08:56
Mesh motion Hex cells vs tets	kev4573	OpenFOAM Running, Solving & CFD	6	December 13, 2007 15:37
Hex Core Mesh	Bob	FLUENT	0	May 28, 2004 14:16
Mesh	Mignard	FLUENT	2	March 22, 2000 06:12
unstructured vs. structured grids	Frank Muldoon	Main CFD Forum	1	January 5, 1999 11:09