The CFX11 Solver Guide lists the orthogonality angle as one of the measures of mesh quality. It's defined as the "area weighted average of angles between n and s vectors for all integration point surfaces of a control volume". The accompanying diagram then shows an example of orthogonality where n and s align, and non-orthogonality where n and s do not align. If non-orthogonality is "bad" then why is the acceptable range for the orthogonality angle >20 degrees? Don't you want n and s to align thereby producing a small orthogonality angle?

Tristan

90 degrees = orthogonal

This is the angle between the line connecting two nodes and the face (as opposed to the face normal). This definition was used because it was more consistent with the more common measure of face angle.

-CycLone

CycLone,

Face? Which face? Are you talking about an integration point surface. I apologize for the the confusion but it seems that the CFX-solver checks 3 measures of mesh quality (orthogonality angle, aspect ratio, mesh expansion factor) that you can't look at directly in CFX-post because it only provides "related" measures (min/max face angle, edge length ratio, element volume ratio). Of course, the smoothing options in ICEM are for different "related" measures. If I ever make a grid for a "real" geometry that actually passes the CFX-solver checks, it will be a miracle!

Thanks,

Tristan