Hi there,

I have some questions regarding grid resolution near the column wall (assuming large columns) in dispersed bubbly flow. I assume in order to simulate boundary layer effects the meshsize here should be smaller then at the center (For simplicity the grid is not addaptively refined here). I made this assumption since in the flow simulation around (for example) a wing section the grid-density is higher near the surface of the wing (Boundary layer eq.) in order to reduce CPU time the far field is solved with different equations and the fields are then matched. In the bubble column the usage of different eq. is/can not be done, in order to simulate the turbulence inducing effect of bubbles interacting with the carier fluid.

question 1:

What should be the (functional) ratio of the meshsizes also related with LES-modelling?

question 2:

is this commonly done in process eng. (please put some references)? Since I still haven not found papers where researchers did this.

reactions (especially from related field) are encouraged!

Thank you.

Ridwan.

Hi Ridwan..

I'm answering the question 1 posed by you; how to get a filter size for LES model from non-uniform mesh ?(i'm changing the form of question ... but i hope this is what you meant) A. Scotti (in his dissertation at johns' hopkins) has shown that grid aspect ratio has a significant effect on the anisotropic grids and the coefficient for models calculated by dynamic procedure should take care of this factor.

Ghosal and Moin ( CTR research briefs) mentioned about the problem of commutative errors using a filter which is function of position ...(in your case it will be !!)

Vasiliyev and Lund (CTR research briefs) formulated a general class of commutative filters ...( the problem with this formulation is that filters thus constructed are not realizable ...)

I would suggest you to have a look at the above mentioned works, it will give a better insight about filtering as well as choosing the filter size from the anisotropic grid.

mayank