von karman instability

[kevinarden]

I ran the von karman instability tutorial 2 years ago and it worked fine. I have re-inastalled elmer fem this month on both Linux and windows and have redone the tutorials as a refresher. The von karman is not the same, as provided it goes steady state even though it is a transient. If I try to increase the Reynolds number slightly it diverges.

Has anybody done this tutorial in the last month?

[raback]

Hi

Do you have same timestep size? The case is critical to the timestep. With longer steps the instability vanishes.

-Peter

[kevinarden]

I followed the tutorial verbatim.

[kevinarden]

I copied the tutorial files instead of re-creating them and it worked. I will see what I did wrong on the re-creation.

[kevinarden]

I typed in the wrong number on the re-mesh step at the beginning of the tutorial so I ended up with 4000 elements instead of 6000 elements. The instabilities were missed due to mesh density.

[Gary R]

Hi all;
I've been trying to get this tutorial to work for several days. I have found that the problem converges if you reduce the Nonlinear System relaxation to 0.1. Unfortunately, the output is 400 vtu files with every other one blank instead to 200 files. The first file seems to be correct i.e. initial condition. (Not setting initial conditions.) The second file is blank and the third file to about the 7th file shows animation but then reached a steady state condition. There is no vortex shedding. With the exception of the relaxation setting, I have exactly duplicated what is in the manual.

Debian Linux
Elmer built from the latest GitHub development release (8.4?)

Help will be sincerely appreciated

Gary R.

[kevinarden]

Would have to see the mesh files and sif. Also you could install the Elmer virtual machine or the CAElinux iso to see if you get the same results. Both come with Elmer installed.

[Gary R]

Hi,
I installed the CSC-Elmer VirtualBox Mint .ova system. This took minimal fiddling to get it started. The keyboard came out Finnish which took some work to correct but works fine now. Unfortunately, the same problem still exists with the exception that the blank frames are gone (only 200 frames now.) Attached is the sif file. I tried to increase the fineness of the mesh but that didn't help. The convergence monitor converges after about 20 iterations and then bottoms out to a straight line. This does not seem right. As I said before the first and second frames look like the result I would get with a steady state solution. The rest of the frames are exactly the same as the first frame.

All help will be appreciated.

Gary R

[kevinarden]

The 8/200 time step did not work for me, resulted in a time step of 8. I had to type in 0.04

[kevinarden]

I took your sif and typed in a time step of 0.01 and I change the linear convergence to 1.E-7 and the non-linear convergence to 1.E-5. The vortex shedding occurred

vonkarman.png

(103.26 KiB) Not downloaded yet

at about 2 seconds.