[marico]

Hi all,

I'm new to OpenFOAM (just did tutorials) and want to ask you if it is possible to perform the following task (I did before in CFX) THEORETICALLY:

1. Solve a (geometrically complex) Problem with chtMultiRegionFoam (one Fluid, one Solid) with kOmegaSST model (as a ANSYS user it was me presented a the "ultimate" model for calculating heat fluxes in practical engineering problems) in STEADY STATE:

1a) Has the solver to be customized for that (as it is designed for transient...) or can it be modified in solver settings without recompiling?
1b) Has anyone experience with that?
1c) Whats the "Near wall treatment" in OpenFOAM generally, but especially for that turbulence model? In CFX it has an "automatic" treatment, which allows the use of fine AND rough (of course, not so nice) grids. Do I HAVE TO create yplus ~1 grids for this model?

2. Solve a STEADY STATE simulation after that with solidDisplacementFoam:

2a) solidDisplacementFoam is designed for transient but solidEquilibriumDisplacementFoam doesnt seem to include thermal stresses?! Which solver is appropriate if I want to calculate the displacements and stresses with the former calculated temperature field? Any experience?
2b) The procedure I estimate for that would be:
- Take the T field from first solution
- Set the field as constraint for displacements (How do I define that? For its not only boundaries but also the inner field that has to be set...)


Thanks a lot for your answers (and for that software)!!

Marco

[kati]

Hi Marco,

I can only reply following (note that I haven't tested the chtMultiRegionFoam):

1a) You might try ddtScheme CoEuler or SLTS. They locally limit the time step. See comments in their source code (you'll find ddtSchemes in OpenFOAM-1.5/src/finiteVolume/finiteVolume/). Or if your other schemes are diffusive enough (upwind etc) you will finally get a steady solution anyway.

1c) kOmegaSST in OpenFOAM is hardcoded with wall functions. So keep your y+ in logarithmic region. Most widely used models in OF are coded with wall functions, but there are some low-Re models, and SpalartAllmaras and LES models can be used as hybrid.

2a) If there is no temperature field in the solver, I think you'll have to design your own solver.

Regards,
Kati

[marico]

Hi Katy,

thanks a lot for your reply. In easy language, what does "keep y+ in logarithmic region" mean?
Value for y+ lower than 11?
What happens if it is too small or too big?

Marco

[marico]

Hi Katy,

thanks a lot for your reply. In easy language, what does "keep y+ in logarithmic region" mean?
Value for y+ lower than 11?
What happens if it is too small or too big?

Marco

[kati]

You should check a book on boundary layer theory to find out what is logarithmic region. Something like y+~30-400. If y+ is too small or too big, the assumptions of the model are not valid which then decreases the quality of the results. This is not OpenFOAM specific. Good luck.

Kati