Hi,

What is suitable mesh size in automotive aerodynamics analysis by a software such as star-cd and fluent (Finite Volume Method)?

What about CFX Software? (Finite Element Method)?

Thanks,

I don't have any idea about automotive applications but in general one could say that your flow should pass cell in a number of timesteps. So if you have a flow velocity of 1 m/s and your cells are about 1cm your timestep should be limited to say about 0.01 m /(1.0m/s*C) where C 2 or 3. C is often referred to as the Courant number. A second test should be the mesh independency of your solution. If you refine your mesh and do the same computation your solution should not change considarable.

If you have any other phenomena in your simulation like mass tranfer of diffusion you should also have a look at these timescales. Always use the smalest.

This does not depent on the software package although some might be smarter/faster in taking bigger timesteps.

hope this helps Michiel

Hi David,

Mesh size will depend on many parameters. The only way to be sure is to refine the grid until grid independance is acheived.

Please note, however, that CFX does not a Finite Element code. Rather, CFX-5 and CFX-TASCflow use a Finite Volume Method, but employ an element based discretization. This element based discretization provides significantly more integration points per control volume than cell based FVM and yields more accurate local gradients for diffusion and sub-grid source resolution. The net effect is that CFX-5 requires fewer nodes to acheive a grid independant solution.

Best regards, Robin