I experienced some strange results when using a Hex mesh, created with the MakeStructured3DMesh command, for incompressible solid mechanics. Hence, I want to formulate a standard Taylor-Hood-mixed element for a displacement-pressure-formulation. Setting "hexes=False” in the attached MWE runs like a charm, but “hexes=True” leads to totally wrong results or divergence of the Newton solver. The deformation for converged states looks quite “interesting” (wrong), see screenshot. Is there some known issue with Hex elements or do I miss something?
Thanks for your comment. The code indeed runs with L2 pressure. However, to the best of my knowledge, it should run with a continuous H1 pressure. In fact, the quadratic H1 displacement and linear L2 pressure combination only runs for “hexes=True” and fails for “hexes=False”. The latter is my expected result, as this combination should be unstable. For “hexes=False”, quadratic H1 displacement and linear H1 pressure works as expected (as also explained in the NGSolve tutorials for Stokes flow).
