[FSVH]

Hi,

My objective is to perform a convergence analysis of a numerical method in a series of progressively refined unstructured grids. In order to do this, consistent refinement is important, that is, all grids should maintain the same structure. I found that the ICEM option Edit Mesh => Adjust Mesh Density => Refine All Mesh was the right tool for doing that.

Things were fine with tetrahedral grids. I used a coarse base-grid and applied successively the Refine All Mesh tool for obtaining increasingly refined grids. I obtained excellent results in that way.

The problem arose when I tried to do the same with hexahedral grids. After refining once and exporting the refined grid, I realized that the parts representing boundaries weren’t saved by ICEM. I checked the grid with the diagnostic tool in ICEM and it said that there were a lot of uncovered faces. ICEM tried to fix the problem covering the exposed faces and putting the new faces into a new part. Apparently, all faces on all boundaries were put in the new part. The weird thing is that those faces weren’t uncovered at all, because there were boundary parts with quadrilateral elements coincident with the faces of the adjacent hexahedra.

I did several tests with different geometries and different base-grids and I observed that the problem appears whenever boundaries have quadrilateral elements, even when they are mixed with triangles. I don’t know if I’m doing something wrong or it is an implementation issue in ICEM, related only to the refinement of hexahedral or quadrilateral elements.

Any help would be greatly appreciated. By the way, I’m using ICEM 15.0.

[FSVH]

After close examination, I found the source of the problem. The Refine All Mesh tool creates more nodes than would be necessary when the elements are hexahedral. Some nodes on the boundary parts are actually repeated. When that happens, the connectivity of an adjacent hexahedron is defined with one node, whereas the connectivity of an adjacent boundary quadrilateral is defined with other node located at the same position. I have no idea why this happens, but after merging the coincident nodes, the diagnostic tool didn't find any further problem with the grid.