Another Questions about BC
 

hey guys,

i'm very new in CFD and even newer in OpenFoam and I have a question about the BC in turbulence modelling. My task is the comparison of different turbulence models in different geometries.

It would be very good to know how the parameters of epsilon and k in the k/epsilon-model have to be set. Is there any formula for calculation these parameters or are there any standard parameters for walls and in-/outlets?

It's the same problem for other turbulence models like k/omega etc.

thank you

Hi,

turbulent fields are generally set based on some pre-defined input, say turbulent intensity. If you are simulating pipe flow, you know roughly how much fluctuations you can expect and you calculate the fields based on that.

To know how to do it, you can follow : http://www.cfd-online.com/Wiki/Turbu...ary_conditions

achyutan

Make sure you pay attention to the turbulence intensity and turbulent length scale and how to estimate them, e.g. like advised here:

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

and here:

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

those pages are great, thanks for that.

But how bad can the results become, if I put wrong parameters for k, epsilon and omega in?

With other words: Is it true, that those parameters are only the start parameters for time step 0 and they will be improved during the simulation? Am I right, if I think that the solution converged in less time with "good" start parameters?

thank you
Hendrik

I am not an experienced user so I may be wrong but I think you are right in saying that these are just the starting values. Question is, will you run your simulation long enough for them to reach the correct values? Hence the importance of setting them right in the first place (also this speeds up the convergence).

At the moment I'm running my simulation as long as it stops because the solution is converged and OpenFoam ends automatically the simulation.

perhaps this is too long? I don't know!

Ah, fair enough. I'm used to running transient solvers. If you are running a steady one and it converges then you should be OK I think.