Most problems associated with creating pyramid layers manifest themselves in the subsequent process of generating the tetrahedral mesh.
Rapid changes in the sizes of cells have a negative influence on the convergence and accuracy of the numerical solution. The pyramid layer creation can produce rapid variations in cell volume in the following situations:
If the quadrilateral surface mesh has faces with rapid changes in size.
If there is great disparity between the sizes of the quadrilateral faces and the neighboring triangular faces used in the pyramid creation.
You can avoid the rapid variation in volume by creating quadrilateral and neighboring triangular grids with smooth variations in face size.
If the quadrilateral surface used to create pyramids has highly concave corners, the resulting pyramids may pierce each other and/or neighboring boundary faces:
In such cases, you can either increase the resolution to prevent
the intersections or alter the meshing strategy. An alternative is
to separate the quadrilaterals in the concave corner into another
zone (using techniques described in Separating Boundary Zones or Modifying the Boundary Mesh), create triangular faces from the quadrilateral faces using the /boundary/remesh/triangulate text command, and then
create pyramids.
The skewness-based pyramid creation will use the existing triangular faces and avoid the intersection problem (see Figure 14.13: Fixed Intersecting Pyramid Cells Using Triangular Faces).
Creating pyramids on quadrilateral faces with very high aspect ratios results in highly skewed triangular faces. Subsequent attempts to create a tetrahedral mesh from these elements will produce a poor-quality mesh.
Irrespective of the method used to generate the quadrilaterals, modify the meshing strategy to reduce the aspect ratio using an external grid generation package or the prism layer capability.