CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Max Ratio in ICEM CFD (

Nicola July 13, 2006 07:32

Max Ratio in ICEM CFD
Hi, I'd want understand why max ratio has to be less than 100 with hexa mesh! There is a way in ICEM CFD to refine mesh near walls without refine in all domain to limite max ratio under 100!!! Thank you!!!

myron July 14, 2006 07:36

Re: Max Ratio in ICEM CFD
What do you mean by max ratio - the growth ratio in edge parameters or aspect ratio of the mesh?

If you mean the growth ratio specified in edge parameters: Why would you want a growth ratio more than 1.4 to 1.5? This is the rate of cell growth from one cell to the next. If you say 100, that means you would have the next layer by 100 X the previous cell. I know of no CFD code that can handle such a volume jump from cell to cell.

If you mean aspect ratio - I don't think there is a limitation on this. The aspect ratio limit would be specific to the code and type of problem. More than 100 doesn't seem wildly out of the ballpark for many codes.

Nicola July 14, 2006 08:07

Re: Max Ratio in ICEM CFD
Hi Myron,

max ratio is the maximum ratio of the lengths of any two edges that are adjacent to a vertex in an element. During a course about ICEM CFD and CFX someone told me max ratio has to be less 100. Sometimes you can accept max ratio values up to 1000 but in this case you have to use double precision numbers. So I think it's only a numerical problem! Near wall I need to refine mesh up to 0.3 mm and so if I want to have max ratio less than 100 I have to refine a lot on the other directions. This increases a lot mesh cells number and I can't launch my simulation because of limited memory! So do you know a way to refine mesh near wall without increasing a lot cells number in other directions? Can I accept max ratio higher than 100? Up to...???

Thank you

Glenn Horrocks July 16, 2006 17:35

Re: Max Ratio in ICEM CFD
Hi Nicola,

Whether your simulation will accept cells greater than 100 is problem dependant. Simulations which converge easily (eg low Reynolds number flows) would probably be OK but the same mesh on a high Reynolds number flow might not be.

The only way of being sure it to try it and find out. You will learn more about your simulation by exploring these questions yourself.

Glenn Horrocks

All times are GMT -4. The time now is 00:28.