Viscous Layers problem
Viscous Layers problemPosted by Kristin Newlands at September 29. 2011
I'm having problems with applying the Viscous Layers hypothesis to my annular geometry and I would really appreciate some help.
Originally, I created the annular cylinder with R1=12.5, R2=25 and L=75, as I was under the (wrong) impression that Salome's units were mm. This caused me confusion when I set-up my case in Code_Saturne (which, as far as I understand, uses metres), so I re-created my geometry with R1=0.0125, R2=0.025 and L=0.075. However, when I proceed to mesh it, exactly as I had done for the previous model (see attached Screenshot), it succeeds in computing a mesh when I don't apply the Viscous Layers, but crashes when I enable it.
Is the problem due to the size of my model? I really need to practice a basic case before I increase the complexity, so I wanted to keep the domain small for faster simulations.
Maybe the reason of the crash is a too high Thickness of the Viscous Layers?
Thank you very much for your reply. I did question whether the thickness of the viscous layers was too high, but even though I reduce it, I am not successful in obtaining a mesh. I don't understand where the problem is, since all I changed are the dimensions of the geometry and I don't know what else to do to avoid the crash.
I desperately need this mesh to work and I attach a python dump of my geometry and mesh setup. Any advice that would help me make some progress would be greatly appreciated.
If I change
Viscous_Layers_1.SetTotalThickness( 4 ) -> Viscous_Layers_1.SetTotalThickness( 0.001 )
in your script, the mesh is computed OK after replaying your script.
Of course, that makes perfect sense. I was making a very silly mistake.
Thank you for taking the time to help me,
Dear Salome users,
I am actually new to the forum since I recently started using Open Foam, with Salome to make the meshing. A colleague of mine gave me a tutorial for using Salome in combination with Open Foam. This makes a Submarine, which should be calculated. Now at a certain point I have to give the mesh itś viscous layers, but then it tells there is a "segmentation violation in address 0". I have tried practically everything I could think of with my little knowledge of Salome. I am using the the latest version 6.3.1. Is there someone who knows how to solve this problem?
On the Internet I was reading about segmentation violation, and there they say that for some reason my CPU can access a certain memory, or something. I have no idea of what is going on with it.
Please help, I can continue my learning process because of this error.
The viscous layer is set up as thickness 10, number of layers 8, stretch factor of 1.2.
Further I am using Netgen 1D-2D for meshing in 2D, and afterwards I am meshing with Tetrahedron (Netgen) in 3D. on the fourth step I am adding the viscous layers and then it gives an error. With the 2D meshing I am giving Hypothesis in Netgen 2D simple parameters, with number of segments set to 8 for the farfield, and local length of 25 for the body of the submarine, and local length of 5 for the small rounded edges.
I am unable to upload the model because it is about 10 MB.
Thanks a lot,
Re: Viscous Layers problemPosted by Duncan Roy van der Heul at December 01. 2011
viscous layer algorithm
You have to download the patch and rebuild the SMESH module.
Re: Viscous Layers problemPosted by DAVID Gilles at December 01. 2011
Or you can wait a little for the next version (6.4.0) to come, in which the bug is fixed.
It should be a matter of days
Ohw thanks. In Salome 6.3.0 it isn't useable is it? Well I am not really sure how to rebuild the SMESH module. I actually don't have a lot of knowledge of programming, or Ubuntu. So I think I will have to wait a few days then.
Thanks for the quick replies.
Re: Viscous Layers problemPosted by Jeroen van der Vossen at December 14. 2011
Well after a little waiting, I am wondering if there is somebody who can tell me how to rebuild that SMESH module? Is there an easy way to do this, a sort of step by step explanation or something?
Activate by Kristin Newlands on Sep 29, 2011 06:23 PM