[Salome] IdeasUNVToFoam - big meshes

malcolm.akner@hotmail.com · February 26, 2021, 09:59

Hi everyone!

I have created a big mesh of about 70 million cells for a heat exchanger, model visible here:https://cad.onshape.com/documents/03...a1b4f5594c73d6

I created the model in Onshape, imported it and meshed it in Salome, downloaded it as a UNV file and am using IdeasUNVToFoam conversion to be able to simulate it in OpenFOAM. The ideasUNVToFoam script has been running for 3 days now and it's not advancing anymore. On smaller models (~1 Million cells) the script has worked perfectly, and the step that I'm currently stuck at is usually the longest step in those other, successful conversions.

I was wondering:
1) Is there any way to make ideasUNVToFoam run in parallel?
2) Are there any inherent limitations in ideasUNVToFoam that I should be aware of, size wise for the model?
3) Are there any flags I could set while running ideasUNVToFoam to make it only do the absolute necessary steps to convert my mesh to a .foam format?

Thank you and I wish you a great weekend!

/Malcolm Akner

olesen · March 2, 2021, 03:41

Even with 900 million cells there is no way that it should take anywhere near that long. There is either a programming error in the convertor or bad syntax in input file. Either way, it's getting stuck somewhere.

malcolm.akner@hotmail.com · March 2, 2021, 06:04

Thank you for your reply olesen.

So what can I do? I need this model in .foam format, what would be the best way of achieving that?

malcolm.akner@hotmail.com · March 3, 2021, 08:44

Thank you for your reply olesen.

So what can I do? I need this model in .foam format, what would be the best way of achieving that?

olesen · March 3, 2021, 15:00

The first thing I normally do is old-school debugging. In your case that would mean adding some progress statements to the convertor. Eg, printing out every 10.000 or even 100.000 points being read and converted. Same thing for the cells being read and converted. Sure, may produce lots of output but also will give you an idea how far it gets before stalling.
After that you can start seeing which part of the code let OKs suspicious or perhaps even check that your file content looks reasonable at that location.
Beyond that, you'll just have to see where it leads you. Much like programming, it is a bit of craftsmanship to debugging as well.

February 26, 2021, 09:59	IdeasUNVToFoam - big meshes	#1
malcolm.akner@hotmail.com New Member Malcolm Akner Join Date: Feb 2021 Posts: 4 Rep Power: 5	Hi everyone! I have created a big mesh of about 70 million cells for a heat exchanger, model visible here:https://cad.onshape.com/documents/03...a1b4f5594c73d6 I created the model in Onshape, imported it and meshed it in Salome, downloaded it as a UNV file and am using IdeasUNVToFoam conversion to be able to simulate it in OpenFOAM. The ideasUNVToFoam script has been running for 3 days now and it's not advancing anymore. On smaller models (~1 Million cells) the script has worked perfectly, and the step that I'm currently stuck at is usually the longest step in those other, successful conversions. I was wondering: 1) Is there any way to make ideasUNVToFoam run in parallel? 2) Are there any inherent limitations in ideasUNVToFoam that I should be aware of, size wise for the model? 3) Are there any flags I could set while running ideasUNVToFoam to make it only do the absolute necessary steps to convert my mesh to a .foam format? Thank you and I wish you a great weekend! /Malcolm Akner

March 2, 2021, 03:41		#2
olesen Senior Member Mark Olesen Join Date: Mar 2009 Location: https://olesenm.github.io/ Posts: 1,714 Rep Power: 40	Even with 900 million cells there is no way that it should take anywhere near that long. There is either a programming error in the convertor or bad syntax in input file. Either way, it's getting stuck somewhere. malcolm.akner@hotmail.com likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
bad momentum source, vortice direction with twoPhaseEulerFoam and non-ortho meshes	cutter	OpenFOAM Running, Solving & CFD	7	November 17, 2016 15:17
coefficient calculation (a_P, a_nb) in unstructured meshes vs. structured meshes(FVM)	t.teschner	Main CFD Forum	2	January 15, 2015 13:10
Read parallel meshes	dl6tud	OpenFOAM Programming & Development	2	March 7, 2014 04:45
[ICEM] How to separate two tri meshes with a line	Bollonga	ANSYS Meshing & Geometry	12	October 9, 2013 06:16
k and epsilon, how big is too big?	NJG	OpenFOAM Running, Solving & CFD	0	April 6, 2013 15:38

March 2, 2021, 06:04		#3
malcolm.akner@hotmail.com New Member Malcolm Akner Join Date: Feb 2021 Posts: 4 Rep Power: 5	Thank you for your reply olesen. So what can I do? I need this model in .foam format, what would be the best way of achieving that?

March 3, 2021, 08:44		#4
malcolm.akner@hotmail.com New Member Malcolm Akner Join Date: Feb 2021 Posts: 4 Rep Power: 5	Thank you for your reply olesen. So what can I do? I need this model in .foam format, what would be the best way of achieving that?

March 3, 2021, 15:00		#5
olesen Senior Member Mark Olesen Join Date: Mar 2009 Location: https://olesenm.github.io/ Posts: 1,714 Rep Power: 40	The first thing I normally do is old-school debugging. In your case that would mean adding some progress statements to the convertor. Eg, printing out every 10.000 or even 100.000 points being read and converted. Same thing for the cells being read and converted. Sure, may produce lots of output but also will give you an idea how far it gets before stalling. After that you can start seeing which part of the code let OKs suspicious or perhaps even check that your file content looks reasonable at that location. Beyond that, you'll just have to see where it leads you. Much like programming, it is a bit of craftsmanship to debugging as well.