Open schwarz-e opened 2 years ago
@schwarz-e While the spatial variation is not surprising when using FEM over unstructured meshes, the range is intriguing. Have you tested refining the mesh, particularly in the interior? Can you plot element WSS that is computed over the face (#62)?
I used Martin's method to calculate the WSS at the faces on a more refined mesh- the mid-slice range is a lot better (only about 5%) but there is a large axial variation when it should be constant.
Axial variation is indeed strange. This didn't happen in your previous test case. Is there a chance for a bug in your element-based calculation around the inlet plane?
It seems like this stems from the velocity solution at the outlet which has non-zero tangential components. This doesn't seem to be a bug, its a consistent steady-state solution when Direchlet at the inlet and Neumman at the outlet is perscribed. However, it's still undesirable.
I am running a simple rigid pipe simulation with a mesh generated with a prismatic boundary layer/outer wall.
However, the WSS field generated is not uniform. Ignoring effects at the end caps, within one axial slice the WSS can vary by 10-20%.
Slice taken at mid-vessel:
Here are the input file and mesh for reference:
setup.zip