CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS (http://www.cfd-online.com/Forums/ansys/)
-   -   Help! Emergency!! Karman Vortex using ANSYS (http://www.cfd-online.com/Forums/ansys/81374-help-emergency-karman-vortex-using-ansys.html)

altomos October 25, 2010 13:43

Help! Emergency!! Karman Vortex using ANSYS
 
Hello everyone.

To begin with, I must say that I am such an idiot ANSYS learner :(. I've tried hard to simulate Karman Vortex Street with CFX. But now, I get no result at all.

Please, anyone that have experienced this before, guide me step by step from the modelling, meshing... But, I found biggest difficulty in meshing. I use ANSYS 12.1

I appreciate your advices much, many thanks to you!!!

Josh October 25, 2010 19:59

Hi -

Nobody is going to give you a step-by-step guide. Capturing the von Karman vortex street is pretty challenging for a new user. You'll need to post specific questions and give details of particular problems.

http://www.cfd-online.com/Wiki/Ansys_FAQ

altomos October 26, 2010 00:09

Oh, thank you very much, Mr. Josh.
My biggest problem is I cannot generate a fine mesh (or suitable mesh) for the simulation.
Actually, I learn everything by myself but this Karman Vortex simulation is really challenging as you said. That's why I need help.

Do you have any advice for me in meshing for Karman Vortex simulation?

Josh October 26, 2010 01:51

I understand your struggle. I had to learn everything myself, as well.

What meshing program are you using? I'd recommend avoiding CFX's built-in meshing program. Use ICEM (or Gambit), something that can create a nice structured mesh and allows for optimal control.

Get a copy of Spalart's "Young-Person's Guide to Detached-Eddy Simulation Grids" (2001). Although it focuses on DES, the grid for the cylinder would be excellent for URANS, as well. Make sure the grid refinement is sufficient in the boundary layer and wake regions. Do this by creating at least three meshes, each doubling in resolution (e.g., mesh 1: 50000 nodes; mesh 2: 100000 nodes; mesh 3: 200000 nodes). Once the desired results (e.g., skin friction) have stopped changing with further refinement, you have obtained a grid-independent solution. Also, ensure the grid quality is sufficient.

Cylinder simulations are very common. Search this website. Here's one page about cylinders: http://www.cfd-online.com/Wiki/Flow_...cular_cylinder

Make sure the simulation is transient (unsteady).

Good luck.

altomos October 26, 2010 02:28

Many thanks to you. I'll try harder. Anyway, this problem is sure uneasy. Thanks for your advice, I do appreciate that much!

Josh October 26, 2010 03:33

You're welcome! Let us know what you figure out.

akhokhay October 28, 2010 08:37

the most important part is (as josh mentioned): Make sure the simulation is transient (unsteady).!!!
this, what I waste lots of time to figure that out!
also I believe if u do it for urself the mesh would not be much difficult maybe even a 2d can do the trick!

altomos November 10, 2010 12:50

Thank you all. I finally solved the problem. I made a mesh of ~420.000 nodes, after 8 hours running the simulation on a Core i3-330 laptop, I got the Karman Vortex Street. SO HAPPY!!!

Josh November 10, 2010 15:31

Congrats. A video would be great. If you're feeling generous, you can describe your setup in greater detail (e.g., unstructured/structured mesh, boundary layer resolution, Reynolds number, timestep size, discretization scheme...)

galap November 11, 2010 16:01

I did a transient simulation in CFX. I used an unstructured "2D" Mesh generated with ICEM. In fact it was no big deal to get the desired result. The key point was a sufficient small time step. I estimate this by the seperation frequency which can be calculated by the strouhal correlation for several bodies.

example: http://www.youtube.com/watch?v=TGNQL0WCVDo

akhokhay November 11, 2010 17:39

I cr8 group on fb CFD(wonder y no one else build it before me) and 1 of the members post 2 interesting video
http://www.facebook.com/pages/CFD/173392319338461
I am lazy to grab links from youtube and post it separately!

altomos November 12, 2010 03:34

Quote:

Originally Posted by galap (Post 283131)
I did a transient simulation in CFX. I used an unstructured "2D" Mesh generated with ICEM. In fact it was no big deal to get the desired result. The key point was a sufficient small time step. I estimate this by the seperation frequency which can be calculated by the strouhal correlation for several bodies.

example: http://www.youtube.com/watch?v=TGNQL0WCVDo

Yeah, thanks for your video. But it seems your Karman Vortex Street is not yet developed clearly. The swirling strength seems not very strong. It's just my opinion.

altomos November 28, 2010 03:21

Everyone, thank you a lot for helping me out. After many runs, I finally relatively satisfied with this video of my simulation.

Please take a look and give me advice for better development. I appreciate much.

http://www.youtube.com/watch?v=bfF2Ktn5oQA

akhokhay November 28, 2010 06:41

congrats.
I think it is just ok and fine!

Josh November 29, 2010 03:51

Looks good to me. Good job.

altomos December 6, 2010 03:02

This is my 2nd simulation and it's a square cylinder:
http://www.youtube.com/watch?v=BFHzHGwL3I8


All times are GMT -4. The time now is 02:09.