[ssa_cfd]

Dear all,

I am simulating a flow inside a closed room. I have done the simulation using RANS k-epsilon model and the simulated results are well in agreement with the experiment results.

But in case of LES, the flow field obtained using field averaging is not accurate. I am using smagorinsky model, pimplefoam. (courant number 1, yPlus avg 0.23) Attached are images from the simulation.

Any help to rectify the error would be appreciated. I have tried different turbulence models, different schemes, but nothing worked out.

Thanks

[mAlletto]

How does the mesh look like? It is also important to resolve the shear layer.

[Santiago]

Comments:

1. Dont use pimpleFoam, the underrelaxation terms for LES plus the outer correctors will give you wtong results

2. The only valid scheme for div(phi,U) is Gauss linear, fourth, and any of the leastSquares family.

3. PISO implemented in FOAM is inconsistent in the tratment of the convective fluxes, thus your Co must be less than 0.5

4. The cell peclet number must be less than 2, due to the 2nd comment.

5. Is your Ksgs/K < 0.2?

Finally, have in mind that LES is a completely different animal, wrt RANS. The latter is more foolproof (no offense), as all scales of turbulence are modelled

4.

[ssa_cfd]

Hi Santiago,

I calculated the peclet number from the last time step and it is more than 2. How to reduce this.? Attached is the peclet number at the center plane.

Thanks.
ssa.

[ssa_cfd]

Originally Posted by mAlletto

How does the mesh look like? It is also important to resolve the shear layer.

I have used a simpleGrading (1 1 1) with 2 million cells. yPlus is less than 1.

[ssa_cfd]

Originally Posted by Santiago

5. Is your Ksgs/K < 0.2?

I am new to LES and I have mostly used RANS. Could you please explain how to calculate this value Ksgs/K from openfoam.

[Santiago]

Originally Posted by ssa_cfd

I am new to LES and I have mostly used RANS. Could you please explain how to calculate this value Ksgs/K from openfoam.

turbulence->k() gives you ksgs
turbulence->k0() gives you k

... in foam-ext 1.6...

[Santiago]

Originally Posted by ssa_cfd

I have used a simpleGrading (1 1 1) with 2 million cells. yPlus is less than 1.

For free sheared flows (ie jets), y+ at the walls is not important! Hows the kolmogorov number in the shear region?

[ssa_cfd]

Hi Santiago,

I have tried with piso with Courant number 0.1 and the results didn't change. If you have time, could you check the case file. ( Case download from Google drive )
and tell your suggestions to improve.


Thanks,
Senthil