Vortex Shedding at Re~30000 in 2D?
 

Hello, I try to simulate the vortex shedding frequency at a bluff body in a rectangular duct.I use the RSM Model with Enhanced Wall Treatment, PISO, Second order upwind and defined some vertixes around the bluff body, where I let me show the dynamic pressure (vertex average). In the insteady simulation i used about 20 Time Steps/expected period. With these settings I cant observe a periodic pressure characteristic. Does anyone know how to make the simulation show me the frequency? If anyone knows an Article, where the procedure for a similar problem is described it would help me a lot. It would be very helpful to know, where I can see the periodic pressure course and how high I can expect the amplitude to be. Best Regards Gerd Sattler

Re: Vortex Shedding at Re~30000 in 2D?
 

Hi Gerd!

I am very interested in your work, because I'm also studying a flow past a bluff body (my case is a 2D car shape). I started steady simulations, but when I swich to 2nd order schemes, solution starts to oscillate, so I think I have an unsteady case. Can I make you some questions? (sorry for my off-topic)

1. When you reach convergence in your unsteady simulations, how do you know it? I mean, if other parameters like CD, Cl, etc. are oscillating because of the unsteadiness of the simulation, what's your convergence criteria? Residuals? In my unsteady simulations they oscillate too, so I don't know if my problem has converged

2. Do you use 1st or 2nd order time discretization?

3. How do you compute your time step in advance in order to know which time step is the best to your unsteady simulations?

Thanks a lot Gerd!

Re: Vortex Shedding at Re~30000 in 2D?
 

Hello, I will try to answer ur questions as good as I can, even though I am no expert. I was learning by doing or asking more experienced users (who hopefully know more than me).

1. In my case the flow into the inlet doesent change in time. I have the same problem as u and I didnt find a person who could answer this question. As the residual doesnt seem to give any information about convergence, I defined a few vertexes and observed the pressure and velocity values at these vertexes. When they dont change any more, the Time Step solution is assumed to be converged. So the simulation of the next time step can start.

2. I use 2nd order upwind. 3. I used as one time step the 1/20 of the expected time period of the vortex shedding. But this is just because my main interest is in simulating the frequency. I think the fluent manual propses to choose the time step size, so that after 10 - 20 interations the solution of one time step is converged. They also give some other suggestions which could help u.

In fact I dont know what u mean by writing that u think u have an unsteady case. I used the unsteady simulation because it is the only way to find out the frequency and not because it is a "unsteady case". I usually first let the steady simulation (it computes a time average) converge and then I switch to unsteady, so that the steady simulation is the starting point for the unsteady simulation.

Hope I could help u a bit. Kind Regards Gerd

Re: Vortex Shedding at Re~30000 in 2D?
 

Thanks a lot Gerd! You've grown my interesting in your case, because of the last paragraph you wrote in your reply.

In your case, are you telling me that you've done steady state simulation and in 2nd order scheme (previous to your unsteady one) and it has converged?

The fact is that in my case, when I switch to 2nd order schemes solutions start to oscillate periodically (residuals, CD, Cl...) and then when I run an unsteady sim. to see what was happening I made some plots and I realized that my case is not steady because plots shows that vertex shedding is present (I think the oscillations in my steady state warned me about this.)

Now, I'm confused, because you got the convergence in 2nd order schemes and your real case is unsteady. Do you think my steady state may converge as yours?

And the last thing, in some regions I get a flow velocity of 40m/s. Which is the limit to consider air as compressible? Mach 0.1, o.2...? I think is 0.2, but I has many convergence problems during unsteady simulations (they diverge when inlet flow is 20m/s and some regions have 40m/s). Could it be due to I'm considering air incompressible and it is not?

Many thanks!