[siw]

Hi,

I've some questions about conducting Large Eddy Simulations using OpenFOAM (current CFX/Fluent user) as so far I've only run the two RANS tutorials:

1. Is it possible to initialise a LES with the results from a steady-state RANS solution? This is something I do in CFX/Fluent to help the LES get started.

2. When OpenFOAM solves it only writes text to the Terminal, so that there aren't any live plots of the equation residuals and monitor points, then how do users conducting LES assess running averages of important solution parameters to determine when such parameters become statistically stable? Thereby knowing that the LES has run for long enough. Again in CFX/Fluent I ignore, say, the first 1000 iterations of a LES before starting the running averages over the next thousands of iterations.

3. Is it possible to specify an unsteady velocity inlet boundary condition for LES? This is to define a mean velocity gust.

4. From reading the OpenFOAM user guide I see that LES and DES/DDES/IDDES are possible. However, can embedded/zonal LES amd Wall-Modelled LES (WMLES) be conducted?

5. I want to simulate the flowfield around an object which is moving past other stationary objects. For this it may be more efficient to have a moving mesh rather than re-meshing the domain. I assume OpenFOAM can do that?

Thanks.

[Sylv]

Hello siw,

Here some hints from my experience of LES with OpenFOAM.

1. Yes.
a. Create a copy of your LES case (named RANSinit for example).
b. Do the necessary changes in the dictionnaries (you can even change the mesh to make it more RANS compatible)
c. Run the case
d. Once it's converged, use the tool "mapFields" to interpolate the RANSinit solution to the LES case. Check the result of the interpolation in paraview...

2. Equation residual should/must always be very low (10e-6) at the end of each timestep, so they are useless to check when your simulation is "statistical stationary". I use some probes, which write a big text file. You can use matlab or python to parse such file and plot the time series.

3. Yes. OpenFOAM includes a so-called turbulent inlet, but it is a big crap: it just adds a white noise to the bulk velocity... Use the LEMOS inflow generator from the University of Rostock in Germany instead.

4. no clue...

5. AFAIK OpenFOAM has such capabilities, but I never use them.

Cheers,
Marcel

[siw]

Thanks for your comments.

[RodriguezFatz]

Hi,
to 2) you may find this usefull:
http://www.cfd-online.com/Forums/ope...ng-script.html