About turbulent model

[chengan.wang]

Hello everyone

When I select the LES model, the momentum equation is automatically changes with turbulent viscosity. But for the scalar equation, like temperature, if I need add some turbulent model (like considering the turbulent thermal diffusivity) to the source term?

Best regards

Chengan

[Martin FERRAND]

Dear Chengan,

I'm not sure I understand well your question. If you select a LES modeling, then the continuity equation and the momentum equation and the transport equation for any scalar are those of the LES averaged variables. That is to say, you do not need to add any term to the transport equation of the temperature. If you want more details, I would advice you to have a look to Pope's book.

Best Regards,

Martin

[chengan.wang]

Hello Martin,
when I select a LES modeling, then the continuity equation and the momentum equation and the transport equation for any scalar are those of the LES averaged variables. For the momentum equation, we add a turbulent viscosity ut in source term of the equation, as described in page 13 of "Theory and programmer's Guide". For energy equation, after filterd, for example in the book "turbulence in Fluides" written by M. Lesieur, there is another term called subgrid eddy diffusivity kt, this term is not specified on page 15 of "Theory and programmer's Guide" in the equations for scalars. So I wonder that if I need add the term kt in enegry equation, or in the code_saturne, it builds automatically a relationship between turbulent viscosity ut and subgrid eddy diffusivity kt with turbulent Prandtl number Pr(t), so I don't need to change anything when I select LES modeling.
Besides, I'am working on the natural convection (Boussinesq approximation) with a high Ra number 1.5x10⁹ in a square geometry(1mx1m).The mesh is 100x100. left side is hot 300.5k, right side is 285.5k, the other two walls are adiabatic. When I finish the calculation, I find that the temperature on the left and right side are slightly changed, not exactly the temperature I fixed. all the three LES models (smagorinsky,wale,dynamic) have the same problem. When I seclect DNS( no model), this problem dosen't exist, the temperature on the boundary are exactly the value I set. If I forget some part when I use LES model?
Thank you very much
Chengan

[Martin FERRAND]

Dear Chengan,

you are right when writing that it is not well described in the theory guide. The transport equation (and you can look at the code of the subroutine covofi.f90) of a scalar take into account a total diffusivity (which is where ). The equation of Energy is then correct as well, you don't need to add yourself any term.

Concerning your natural convection test case, the temperature your see at the wall is indeed not exactly the one you imposed. It is in fact only a matter of post-processing: given you are using a LES model (it would be the same in RANS) you use a wall function at the wall. Consequently, the temperature you impose at the wall is used to compute the corresponding flux at the wall. In order to impose that computed flux, the code set where and are the boundary condition coefficient. That is that which is post-processed (and not the one you impose).
If you do a DNS, then no wall functions are used, and is exactly the one you impose.

What is the in your simulation?

Best Regards

[chengan.wang]

Hello Martin,
Thank you for your reply. I still have some questions:
1)I don't very clear what kind of wall function is used automatically by Saturne when I use LES. And how I could ouput the exact temperature on teh boundary?
2)based on the temperature obtained(filterd), if I still could use usproi.f90 to caculate heat flux correctly? For example to calculate Thermal balance.
3)For y+, when I use a mesh 100x100, in the file "listing",
!****************************************************************
** CONDITIONS AUX LIMITES EN PAROI LISSE
----------------------------------------
------------------------------------------------------------
Minimum Maximum
------------------------------------------------------------
Vitesse rel. en paroi uiptn : -0.00000E+00-0.00000E+00
Vitesse de frottement uet : 0.44190E-03 0.30878E-01
Vitesse de frottement uk : 0.00000E+00 0.00000E+00
Distance adimensionnelle yplus : 0.26168E-01 0.19741E+01
------------------------------------------------------
Nbre de retournements de la vitesse en paroi : 0
Nbre de faces en sous couche visqueuse : 400
Nbre de faces de paroi total : 400
------------------------------------------------------------
!*****************************************************************
For the model of dynamic,If the value is proper for calculation?

Best Regards,
Chengan