CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Karman vortex street (http://www.cfd-online.com/Forums/fluent/29922-karman-vortex-street.html)

 Zhipeng June 18, 2002 21:18

Karman vortex street

Hi, everybody.

I have a trouble with the calculation of the Karman vortex street. It is a problem of flowing over a cylinder. The parameters that I used in Fluent(2D Segregated) are listed below: ++++++++++++++++++++++++++++++++++++++++++++++++++ +++++

The size of domain :length:40m width:15m

The diameter of cylinder :2m

The location of cylinder :(15,7.5)

Boundary conditions :

left boundary -- Velocity Inlet

right boundary -- Pressure Outlet

top and bottom boundary -- Wall

cylinder -- Wall

The properties of air :density:1.225kg/m3

viscosity:1.7894e-05kg/m-s

X-Velocity in Velocity Inlet :7.3037e-04m/s (constant),so Re=100.

Viscous Model :Laminar

Time step size :0.01s

Number of Time steps :20000 ++++++++++++++++++++++++++++++++++++++++++++++++++ +++++

But I did not see the Karman vortex street. What shall I do? Please give me some advice, thanks a lot.

 Laika June 19, 2002 04:17

Re: Karman vortex street

use realizable k-eps-turbulence-model instead of laminar flow.

good luck,

Laika, still orbiting

 Zhipeng June 19, 2002 20:46

Re: Karman vortex street

I have done the calculation with your method, but it doesn't work. Any suggestions?

 Steve June 19, 2002 21:52

Re: Karman vortex street

At that low reynolds number the flow may be steady. Either perturb the flow with asymmetric initial conditions. Or try a slightly higher Reynolds number. From memory I think it is more like Re=300 when the vortex street develops.

Don't use a turbulence model because the flow isn't turbulent and turbulence with just damp unsteadyness.

Steve

 Jin-Wook LEE June 20, 2002 00:04

Re: Karman vortex street

Anyway, at first, I think that you should try to give pertubation to the steady solution to see the unsteady flow. Otherwise, even though the flow is unsteady, you can see the unsteady flow after (maybe) couple of millions time step, at which time, numerical truncation error can be 'artificial pertubation.

Sincerely, Jinwook

 Jeff Moder June 24, 2002 11:00

Re: Karman vortex street

Besides the other comments on making sure you start your time-accurate simulation from some "perturbed" state, you may also want to increase your time step to something larger that 0.01 s, such as 16 sec.

I say increase your time step based on the following:

Let time setp dt = (period of lift)/1000

(actually even a few hundred time steps per period is probably enough, but 1000 should be plenty).

period of lift = 1/freq = D/(U * St)

D = cylinder diameter = 2 m U = uniform inlet speed = 7.3037e-04 m/s St = Strouhal number = 0.165 for Re=100 cylinder

period of lift = 1.66e+04 sec

dt = 1.55e+04/1000 = 16.6 sec

 All times are GMT -4. The time now is 17:47.