CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   Surface normals in ICEM (

Chander November 9, 2010 11:19

Surface normals in ICEM
I have recently started using ICEM-CFD for meshing my domain. In fact this is the first time I am using a commercial code for CFD. Please advise what is the role of surface normals in ICEM. Is there a 'consistent' direction of surface normals which one should ensure while meshing the domain? I see a tool 'Modify surface normals'under the Repair geometry tab. Why do we need this?

PSYMN November 11, 2010 17:14

Thats normal.
ICEM CFD is a tool with lots of options... They are there for the experts who may want them. If you are new and haven't said "man, I wish I could modify surface normals", you probably shouldn't worry about it. It is rarely needed for CFD (it is taken care of automatically).

Just focus on the few simple steps you actually need to get the mesh you want. This keeps it from getting overwhelming...

But since you asked... Some solvers, particularly FEA solvers, are very concerned with surface normals because their properties have a direction... They also come in handy for certain operations, like extrude, where you can extrude by normal instead of giving a direction.

Chander November 17, 2010 13:33

Thanks Simon for your reply!
I was worried about surface normals as someone suggested that a possible reason for non-convergence of my simulation in CFX could be wrong direction of surface normals while creating geometry in ICEM.
Thanks again

PSYMN November 17, 2010 19:16

Oh yes, fixed.
Yes, I have seen that in CFX... Especially if You grow prisms thru a baffle against the wall (quads form on the baffle).

In older versions, the normal of the prisms was not consistent and each column of quad shells would orient based on the line element at the wall (which were not required to be uniform).

As a result, some columns had normals facing different than the rest and this caused the problem. But once we heard that this was important, we immediately fixed it so that the entire baffle, including all the prism quad columns, have the same orientation.

I have not heard any reported issues since.

All times are GMT -4. The time now is 17:25.