I've got a ProE model of a turbine blade that I'm trying to mesh in ICEM. The Model is translationally periodic in the "y" direction. Originally the periodic boundaries were made up of three surfaces on the top of the domain and three on the bottom.

After loading the mesh from the blocking and running the "check mesh" under the "edit mesh" tab I get errors saying that:

"maximum distance from periodic 7.62939e-006 between vertices 309611 and 374922; Fixed"

When I try to run the elliptic hex smoother it completely corrupts mesh.

I've tried a simplified example to see if my procedure is correct and I've managed to successfully mesh a rectangular domain with an O-grid around a blob that is defined by a closed spline. Periodicity is fine and the elliptic smoother output looks great. If I try to angle the top and bottom of the domain (making it skewed when viewing the x-y plane), or try to define the top and bottom of the domain by a curved spline I get the same errors as above (different distances and node numbers).

I've played around with the tolerances in both ProE and ICEM and it doesn't seem to make a difference...

I've searched the forums and can't find anyone that's encountered a similar problem with periodicity. How are people generating their domains? Importing them to ICEM?

Any input is greatly appreciated.

-Danny

Are you actually specifically applying a periodicity condition in ICEM?

The smoothing in ICEM can be very poor - use at risk is what I always say. First make sure you have all your hard corner vertices tied to geometry points, edges to curves, etc. Lock all hard features to be preserved via associations/projections.

You may also want to try the multi-block smoother, which can be even more painful - but often works.

Good luck.

Joe,

Yes, I'm applying the periodicity condition in ICEM under the mesh parameters, then assigning the periodicity to the appropriate nodes of the blocking. (All the vertices on the periodic surfaces have been assigned as periodic). The vertices that lie on the inlet and outlet of the domain are associated to the points that define the corners while the 8 vertices that mark the leading/trailing edges of the blade (or the blob) are snapped to the curves which define the periodic surface. I've essentially followed the turbine blade demo in the ICEM Demo room...

In terms of importing the ProE model I've tried using the workbench reader for the actual ProE file, or the IGES and parasolid file formats. (The iges file format was used for the blob configuration successfully).

The ProE models I've used in my tests are all periodic as far as I can tell. The dimensions defining the periodic surfaces are all constrained and identical tangent constrains are used on both the top and bottom of the domains that use splines.

Is there any limitation on what type of surface ICEM can treat as periodic? For example in that periodic turbine blade demo, the rotationally periodic surfaces have some sort of curvature associated with them, any idea how those surfaces were created?