Elmer diverges when changing the coordinate system!

Numerical methods and mathematical models of Elmer
Post Reply
foadsf
Posts: 81
Joined: 20 Sep 2016, 23:56
Antispam: Yes
Contact:

Elmer diverges when changing the coordinate system!

Post by foadsf »

Dear Elmerers,

I have used the Vortex shedding – von Karman instability example:

Directory: VonKarmanGUI
Solvers: FlowSolve
Tools: ElmerGUI
Dimensions: 2D, Transient
and have changed the geometry to have an step in the middle.
step.png
step.png (43.4 KiB) Viewed 4227 times
when I run this simulation it works just fine (case_h.sif, step_h.in2d)
sol.png
(59.31 KiB) Not downloaded yet
but when I change the geometry in a way to have x and y axises replaced (case_v.sif, step_v.in2d) it diverges! I'm confused!

I would would appreciate if you could help me know what am I doing wrong?

P.S. The reason I want to change the geometry to vertical is because later I want to change th experiment to axisymmetric. so if you can help me do that in one step that would be a great help.
Attachments
step.zip
(2.7 KiB) Downloaded 305 times
Last edited by foadsf on 06 Feb 2018, 16:19, edited 1 time in total.
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: simulation diverges when changing the coordinate system!

Post by mzenker »

Hi,

it may be that for 2D it makes a difference which plane (x-y, x-z) you use. There has bee a discussion on that quite some time ago: viewtopic.php?f=7&t=3711&p=12668#p12670. That problem should be fixed, but maybe not for every solver?
What happens if you just set to axisymmetric and rerun?

HTH,

Matthias
foadsf
Posts: 81
Joined: 20 Sep 2016, 23:56
Antispam: Yes
Contact:

Re: Elmer diverges when changing the coordinate system!

Post by foadsf »

Dear Matthias,

I tried to go straight to an Axisymmetric model where it diverged too. that's why I decided to go one step back and try the vertical model. I can try it again and put the .sif file here. In the meantime would you be so kind to check my .sif files and let me know if I'm making any obvious mistakes?
foadsf
Posts: 81
Joined: 20 Sep 2016, 23:56
Antispam: Yes
Contact:

Re: Elmer diverges when changing the coordinate system!

Post by foadsf »

Dear Matthias,

The axisymmetric model also diverges:
case.sif
(2.13 KiB) Downloaded 422 times
F.
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: Elmer diverges when changing the coordinate system!

Post by mzenker »

Hi,

I don't see any obvious error in your sif file, but I am no Navier-Stokes expert.
Have a look at the solver log and see where it diverges. Sometimes it helps just to increase the number of iterations (e.g. 1000 instead of 500 for the linear part). Also you can check if your length units are what you expect (Model->Summary). Elmer assumes m. If your geometry is in mm, then put the line

Coordinate Scaling = 0.001

in the free text field of the simulation section (Model->Setup).

HTH,

Matthias
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Elmer diverges when changing the coordinate system!

Post by raback »

Hi

Well, it seems that at least the step_v.in2d had negative x-coordinates. For axi symmetry (x,y) corresponds to (r,z) and there really cannot be negative radius.

Generally, it is easier to debug with the mesh.files attached. When the geometry it is difficult to know what steps were taken to turn it into mesh.

-Peter
Post Reply