Issues with border elements when transitioning from structured to unstructured elements
I don't have an MWE for this issue, so I will do my best to describe it. I have a large 3D mesh, where a point of interest needs to consist of structured HEX8 elements. Elsewhere, the elements can be coarse. Ideally, these would be unstructured HEX8 elements. However, it seems there is no algorithm in place to recombine unstructured elements in 3D. So, I am using the Delaunay algorithm to have HEX8 structured elements bordered by PYR5 elements and then with coarse TET4 elements everywhere else.
Visually, the mesh looks ok. The issue is that there will always be a few pyramid elements with a negative Jacobian. Looking at the offending elements, they appear to be totally flat.
I cannot find a way to optimize these border elements or to avoid this issue. Is there a way to fix these elements? Or is there another way to approach this mesh that avoids using border elements altogether? I've attached a picture of the border region, just in case it is of any use.