- What is your Reynolds number?

- Do you work on Sunday? Or do you work in Japan where it is already Monday. Where is .se?

Astrid

*LOL* No....I'm doing my master thesis and my supervisor is like a slave master.....he he he....I have to work day and night... .se is Sweden....I'm in Stockholm at the Royal institute of technology...and you? My Re is 2.0417E+06...

- I'm in The Netherlands. I am just doing this for the lol, on a Sunday after-afternoon

- Quite high Reynolds number. Don't you expect any 3D-effects? Unsteady flow behavior or flow separation? Just curious.

Astrid

You seem to know a great deal about CFD though....I'm just an novice....but I'm trying... Well...I guess I have a some 3D effects but considering what I'm solving from my point of view those are negligble... Do you work with CFD?

Dimitris,

I think you need to spend a bit of time diagnosing why k and epsilon residuals are stalling. If that is the case then I would be suprised if fluids residuals are not stalling at some point. Since you are simulating flow around a cylinder at fairly high Reynolds number you might expect this flow to be transient, i.e. vortex shedding.

If you want to "smooth" this out so that you can average out the vortex shedding I suspect that you might have to play with the inlet turbulence levels. Do you have data for this from your experiments? Two things to check are whether or not the turbulence is decaying within a few cells downstream of your inlet, and also what does the turbulent viscosity field look like? Or more importantly, the turbulent viscosity ratio, mu_T/mu. If mu/mu_T is mostly 1 throughout your domain (or in the area immediately surrounding the cylinder) then the turbulence model is basically doing nothing for you in these flow regions. Hence, you may need to play with the inlet turbulence levels to get things "diffusive" enough to stablise the solution.

One other issue with these flows around cylinders is domain confinement. If it is over confined then you could also run into problems. i.e. you will get the wrong shedding frequency because the shedding wavelength can adopt to the channel width. Does your geometry exactly match your experimental setup, or is it a small cut out? This might not be important for you if the shedding frequency does not matter.

Neale.

Dimitris,

please read the latest posting of Neale very carefully. As vortex shedding might be important, you should consider switching to a full 3D description. Then use the SST model which is capable of calculating a decent Cd coefficient and shedding frequency. Then you have to fully integrate everything to the wall. However......, for the flow around a cilinder I have only tested SST up to Re = 1e5. So, this differs from your case.

Astrid