search

OpenCMISS / iron

Source code repository for OpenCMISS-Iron
9 stars 62 forks source link

Issue with simplex elements with quadratic interpolation #107

Open PrasadBabarendaGamage opened 6 years ago

PrasadBabarendaGamage commented 6 years ago

There is an issue with the quadraticTet mechanics example in the old examples repo. The model converges but the undeformed and deformed configuration look distorted. Unfortunately, have not had the time to investigate further. quadratic_tet_mechanics.zip

Potential steps for debugging:

  1. Check if linear tet mesh created with OpenCMISS's inbuilt generatedMesh routines are created/visualised correctly (most likely as I think these are used for fluids)
  2. Check if quadratic tet mesh created with OpenCMISS's inbuilt generatedMesh routines are created/visualised correctly
  3. Solve mesh using zero load and visualise to confirm if deformed field matches undeformed field
  4. Repeat with 2. with the original loading condition in the example.
PrasadBabarendaGamage commented 6 years ago

screenshot-2017-12-15-10 09 38-selection_001

beauof commented 6 years ago

I think this goes back to what we have discussed during one of the conference calls.

It looked like the inconsistency in the iron and zinc/cmgui numbering schemes is causing this issue. This should probably be addressed by changing the export routine for quadratic tets.

rchristie commented 6 years ago

Zinc/EX cycle all local nodes for all bases in xi1 fastest then xi2 then xi3. Iron uses Zienkiewicz winding for simplex elements. The attached image shows the local node ordering for quadratic tetrahedra with Cmgui/EX (centre) and Iron/Zienkiewicz winding; VTK quadratic tetrahedra winding is different again. All are the same for linear tetrahedra. Be careful when you export in the Cmgui/EX order that you look up the equivalent node in the Iron sequence from the diagram. tetrahedron_winding