CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Mixed models in LES (http://www.cfd-online.com/Forums/main/5894-mixed-models-les.html)

Mixed models in LES

Hi guys,

I have a question regarding mixedmodels in LES.

I understood that in LES we only model the deviator part of the subgrid tensor in the case of eddy viscosity SGS models (Smagorinsky type). we don't model the isetropic part and it is added to the pressure, i.e.we will have a modiefied pressure.

but in the case of scale similarity models like Baldina we consider entire subgrid tensor.i.e,

tauij=scale similarity term from bardina (=Lij).

but In the case of mixedmodels we have the formula like

deviator part of tauij=smagorinsky model+(Lij-1/3Lkk*deltaij)

where deltaij=1 when i=j otherwise it is 0.

my question is Are we going treat subgrid tensor similar way like Smagorinsky model,i.e, we will have a modified pressure or we required some other approach.

bye

Rajani

 Mayank Tyagi March 26, 2003 16:10

Re: Mixed models in LES

Hi, You are right in treating SGS tensor by using the deviatoric part only (yes, you'll get a modified pressure). In evaluating the model coefficient, we need the deviatoric part...ref: Vreman et al (1996), On the formulation of dynamic mixed models, physics of fluids. However, you could gain more insight from Lkk (trace of scale similar term) about the subgrid turbulent kinetic energy...ref: Knaepen and Carati (2001), physics of fluids (also in proceedings of 3rd AFOSR conf. on DNS/LES 2001) hope this helps Mayank

 euler March 27, 2003 04:15

Re: Mixed models in LES

Hello

In principle you can model also the full tensor, not only the anisotropic part of the tensor. With Smagorinsky-like models this is not possible because S_ij is traceless. And then you model only the anisotropic part of the tensor.

In mixed models you can model the full tensor or only the anisotropic part. It is up to you. For example in Sarghini et al 99 Physics of Fluids the model is:

tau_ij = L_ij + eddy viscosity term

Of course the Pressure will not be the real Pressure but some mixed function of P and the subgrid turbulent kinetic energy. (Because the eddy viscosity term remains traceless)

Regards

 S. Crofts. March 27, 2003 06:56

Re: Mixed models in LES

Hi there,

Where can I find a text (preferably a text book) that describes how the trace part of the tensor is absorbed into the pressure term? Also,

Thanking you all.