CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   IcemCFD: cell = element? (

Melisa January 14, 2004 12:28

IcemCFD: cell = element?

I have some question about ICEMCFD again. I notice that when I run tet mesh, the message box will tell me the number of cells being generated (before slow transition). I initially thought it's the number of mesh elements for my geometry. However, when I write the output file, I just realise that the number of elements is much less than the no of cells shown in the message box.

Anyone know what the message means when it said 500K cells? :O

Glenn Horrocks January 14, 2004 17:16

Re: IcemCFD: cell = element?
Hi Melissa,

The initial step of the ICEM tet meshing procedure is to mesh the entire extent of the geometry, with no regard to the shape of the geometry. The only thing which is used initially is the minimum and maximum x,y,z extent of the domain. This initial mesh can have many times more elements in it than the final mesh.

After the initial mesh is done it then looks at each cell and determines whether they are in the geometry, or if they are outside the geometry (the "ORFN" block). Elements entirely in the ORFN block are deleted, elements entirely in the geometry are kept and elements split between the two are trimmed at the geometry boundary.

This is discussed in more detail on pages 15-21 of the user manual.


Melisa January 14, 2004 21:33

Re: IcemCFD: another question
Thanks again for your helpful comments, Glenn.. :)

I have another quick question:

I read the notes about scale factor and Global mesh size but I still don't quite understand why we need two parameters instead of one.

I know the max element size = scale factor*Global mesh size but why don't we just have max element size?

Any significance about having 2 parameters?

I don't see the difference when I change mesh size = 20, scale factor =1 to mesh size =1, scale factor =20.

The mesh looks exactly the same. :O

derrek January 14, 2004 22:44

Re: IcemCFD: another question
It is the same exact thing. Look at the formula, 20*1 = 1*20. The idea is that you set one number, mesh size. You then use the scale factor to refine or coarsen the mesh..



Melisa January 15, 2004 00:06

Re: IcemCFD: another question
Thanks, derrek. This clear the things out. :)

Glenn Horrocks January 15, 2004 00:44

Re: IcemCFD: another question
Hi Melisa and Derek,

As Derek said, they are the same. The scale factor is a factor which scales all mesh sizes across your entire mesh. If you increase the scale factor by two you can double every mesh element edge length in the entire model with just one parameter. Useful for doing mesh refinement studies.

A few things to remember: 1) There are some mesh dimensions which do not scale with scale factor. 2) The global edge length must be a power of two. If it is not a power of two it gets rounded to the nearest power of two. If you want to adjust the global edge length by a factor not equal to two you should use the scale factor.

As always, look in the documentation for more details.


All times are GMT -4. The time now is 21:02.