CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] Aspect Ratio is too high. ICEM (http://www.cfd-online.com/Forums/ansys-meshing/116901-aspect-ratio-too-high-icem.html)

Dorasemon April 27, 2013 23:06

Aspect Ratio is too high. ICEM
 
Hi, everyone.
I want to know how to control the aspect ratio if the height of boundary layers is less than 1% of the global sizes.
I am simulating ventilation in office room and there are some energy sources in my case. I should ensure the y+<5 so the aspect ratio is very easy to be higher than 100. Especially in the domain far away from the boundary, I don't know whether this problem will cause wrong results in fluent.
By the way, I use blocking method.
Thank you in advanced for any information.

Far April 28, 2013 02:19

is it a 2d problem? AR of 100 is not too high. However to enhance accuracy use double precision solver with high aspect ratios.

Dorasemon April 28, 2013 07:56

Quote:

Originally Posted by Far (Post 423568)
is it a 2d problem? AR of 100 is not too high. However to enhance accuracy use double precision solver with high aspect ratios.

Thank you very much, I will try double precision.
It is a 3d problem. In my case, the temperature is about 1℃ lower than experiment . I can't find the reason so I thought the problem may be aspect ratio.

Far April 28, 2013 08:17

For 3d problem you can have AR upto 1000 or even higher.

Mesh is the problem but not always ;)

Dorasemon April 28, 2013 08:40

Quote:

Originally Posted by Far (Post 423620)
For 3d problem you can have AR upto 1000 or even higher.

Mesh is the problem but not always ;)

Thanks:).
I will try use other turbulence models or refine the mesh.

Ananthakrishnan April 28, 2013 11:15

Hi,
Just in case if you have not discovered this idea. Keep a uniform node distribution in the far field boundary. This will help reducing the aspect ratio as well. If not you might end up having reversed flow at the outlet in fluent.
hope this idea is relevant to you!!

Far April 28, 2013 11:31

Ananthakrishnan : Very relevant hint. I agree too

Dorasemon April 29, 2013 09:39

Quote:

Originally Posted by Ananthakrishnan (Post 423652)
Hi,
Just in case if you have not discovered this idea. Keep a uniform node distribution in the far field boundary. This will help reducing the aspect ratio as well. If not you might end up having reversed flow at the outlet in fluent.
hope this idea is relevant to you!!

Ananthakrishnan:
Thanks. In my case, their are some computers(rectangular box, use temperature as boundary condition in fluent) on a table(thickness=0) in the office.
I split the block and set the edge params to control the spacing and ratio, this cause the aspect ratio very high. I don't know how to keep uniform node distribution in the far field boundary and control the spacing and ratio at the same time.
Could you tell me how to mesh the boundary layers near the computer surfaces and table?
Thank you very much.

Far April 29, 2013 09:57

Instead of using copy to all parallel edges, set spacing parameters individually.

Give fine spacing/ratio on the wall surfaces (computers etc) and define uniform spacing in the farfield.

Dorasemon April 29, 2013 10:03

Quote:

Originally Posted by Far (Post 423888)
Instead of using copy to all parallel edges, set spacing parameters individually.

Give fine spacing/ratio on the wall surfaces (computers etc) and define uniform spacing in the farfield.

Far:

If set spacing parameters individually, would the grid orthogonality be another problem?

Ananthakrishnan April 29, 2013 11:29

Hi,
Theoritically it can cause some problems. Just check the pre mesh orthogonal quality. If its too bad then you might have to change your blocking strategy a little bit at those places..If not you can go ahead.

Generally it does not create such big problems. Go ahead and try it out!! because without that there are very high probabilities of reverse flow

Dorasemon April 29, 2013 11:49

Quote:

Originally Posted by Ananthakrishnan (Post 423931)
Hi,
Theoritically it can cause some problems. Just check the pre mesh orthogonal quality. If its too bad then you might have to change your blocking strategy a little bit at those places..If not you can go ahead.

Generally it does not create such big problems. Go ahead and try it out!! because without that there are very high probabilities of reverse flow

After trail and error, I get better result using hex-core method to generate mesh and Launder-Sharma model instead of k-e RNG turbulence model, even though mesh quality is very low(I don't know how to refine it, so I let it be).

Next step, I will try what you and Far suggested.

Dorasemon April 29, 2013 11:59

Quote:

Originally Posted by Far (Post 423620)
For 3d problem you can have AR upto 1000 or even higher.

Mesh is the problem but not always ;)

Far:
What you said is absolutely right!
When I change RNG k-e to Launder-Sharma model, I get better result. And I will try refine the mesh to find whether bether mesh can get good result or not.
I really appreciate you helping me.


All times are GMT -4. The time now is 18:43.