CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CD-adapco (http://www.cfd-online.com/Forums/cd-adapco/)
-   -   karman vortex street (http://www.cfd-online.com/Forums/cd-adapco/53300-karman-vortex-street.html)

michael April 28, 2003 05:00

karman vortex street
 
hi i've some problems with the simulation of a karman vortex street behind a cylinder. i want to do the 2-d case. the results for y+ as well as the results for the courant number are ok.but the frequency, i get through a fft, is not in the range it should be. i tried it also with more cells but the results aren't better. can anybody give me a hint what i'm doing wrong or what i can change for better results?

Anton Lyaskin April 29, 2003 05:42

Re: karman vortex street
 
Hi!

I have some experince with modelling vortex streets behind cylinders and other bluff bodies with Star-CD, so some questions first:

What is your Reynolds number?

Flows with vortex streets have a complex physics, it concerns mainly turbulence. You can have laminar flow in the vicinity of the cylinder and in the near wake and in the fare wake there is a transition to turbulence. So fully turbulent simulation is valid only for supercritical Re, i.e. higher than 1.0e7. And "classic" Karman vortex street usually means much lower Re.

How do you determine you frequency?

Do you take fft from some velocity component or velocity magnitude? Usually the frequency for vortex street means vortex shedding frequency and oscillations of some velocity component can have a frequency twice higher. Besides, there are secondary vortices wich also affects the frequency of velocity oscillations in the near wake.

Where do you measure the oscillations?

The frequency varies with the distance from the cylinder and you should check the paper describing the experiments with wich you want to compare your results for the location of the probe.

Regards, Anton

michael April 30, 2003 03:44

Re: karman vortex street
 
hi anton thanks for your answer we did the the simulation with reynolds numbers of 200 and 140000. so we wanted to have a laminar and turbulent vortex street. we messured our oscillations on top of the cylinder surface. and we messured the oscillation of pressure! your right that there is a big influence of the distance ! we also found out that you can't do the 2-d simulations for high reynolds numbers. you have to use 3-d models. what you think? regards, michael

Anton Lyaskin April 30, 2003 04:20

Re: karman vortex street
 
The highest Re I've tried for 2D a circular cylinder was 50000. I was doing full turbulent simulation though it's a subcritical Re. The averaged darg was ok, but I didn't check the frequency (I think, frequency is more sensible to inacuracies than drag). Your Reynolds number is very close to critical and it's difficult to get even experimental results for it because many additional factors like roughness make a large influence. In some reference books I have there are no solid curves at Drag vs Re and Frequency vs Re plots - just some gray spots. I guess transition phenomena can be critical for such Re. So, probably the answer that it is impossible to correctly model such flow with existing capabilities.

As I wrote earlier, there can be secondary vortices which can impact on pressure oscillations on top of the cylinder. Later (far downstream) they can dissolve or merge with primary vortices. Did you try measuring oscillation in the wake?

Regards, Anton

michael April 30, 2003 04:44

Re: karman vortex street
 
not yet. but i will try it today. and for what reynolds numbers did you get acceptabel results? did you ever try to do this simulations with a 3-d modell? what kind of turbulence modell did you use? we did our simulation with k-e-high reynolds number and k-e-square...with an implicit time discretisation regards. michael

Anton Lyaskin April 30, 2003 05:04

Re: karman vortex street
 
As I said the highest Re was 50000. I was doing 3D simulation for higher Re but not for the circular cylinder - there were diffent prisms with corner separation which in terms of turbulence modelling is "simpler" than separation from the cylinder surface. The agreement for frequency was quiet good (I was measuring crossflow oscillations at 0.5D behind the body). I used k-epsilon RNG, for circular cylinder - with non-equilibrium wall function. I was advised to try also low-Re model, but didn't have time for it yet.


All times are GMT -4. The time now is 13:45.