Hi there,

I want to simulate an airfoil in a 2D situation. When i build up my Mesh with Icem, i have to build a pseudo 3D Mesh with only 1 cells in the not considered direction.

Because of SST-Turbulence Model i need y+ of 1 in the boundary layer.

So i get very very small elements there, but in the far wake the are much bigger. Considering the third direction (with 1 Cells) elements with bad aspect ratios occur (about 2000).

How do i avoid this? How "high" should the 1 cell be (1mm, 10cm, 1m)? (the domain is about 50mx50m)

I have some convergence problems. Could this happen because of the aspect ratio, or is it unimportant because the third dimension in unconsidered.

Thanks in advance.

Try this!! Put the same length of smaller cell of your mesh in the direction that 1 cells in the not considered (i think that is z direction) and try not outdo the aspect ratios of 1.05!!

Forgive my bad english!

Do a search in google for this paper,

has a picture of a typical airfoil mesh, should help

Detached Eddy Simulation of Flow around A-Airfoil

STEFAN SCHMIDT and FRANK THIELE

Hi there,

thanks for your help first.

Costa when i do so, making the z-direction that small, like the smallest cells around the airfoil (boundary layer), the aspect ratio around the airfoil will be good. But in the far wake the aspect ratio will reach high levels, because there the cells are bigger (in the x and y direction). Do you think one has no problems with that?

1.05 aspect ratio?? Do you really mean the aspect ratio (for example X-length/z-length or y-length/z-length of one cell) or do you mean a scaling factor from one cell to another? I thought aspect ratio of 200 are just fine.

An other question, when i follow your hints the overall size of the elements will be very very small. Isn't that a problem when calculating forces for example, because you need a surface (which will be very very small then) to do so. So one gets really small values ca. 10e-06 or something like that, isnt that a problem of numerical accurancy.

Thanks in advance.

Kramer

don´t have a problem if in the far wake the aspect ratio will reach high levels, the important is that the residuals in condinate z be very very small!

forgive! 1.05 is the maximum scaling factor of you need use and not the aspect ratio!! The aspect ratio import when the bonduary conditions are importants near to close of bigs scales of aspects ratios. If not occur this, don´t have a problem.

I think that that numerical accurancy don´t be affected if your mesh will be very very small and the same coomb to overall size of the elements!

fogive bad english!!

You should probably aim for a "slice thickness" of about 2mm. It should be possible to get a reasonable y+ and aspect ratios of a few hundred. In my experience, CFX-10 will run on such a mesh quite happily and gives good agreement with experiment in the cases I have done. Your convergence difficulties could be coming from any number of other sources, so don't automatically assume it's due to aspect ratio (although 2000 is very high). Best of luck.

The aspect ratio is important when too the lenght scale of turbulence is small. This signify that, for example, if you have 1m of lenght scale, you need put 10 volumes of your mesh inside of 1m and your mesh need have 0,1 x 0,1 in maximum.