[FarisAJ]

Dear All,

Since 2 months, I'm having an issue with running a URANS simulation whenever the mesh exceeds 100k. It is a 2D case with dynamicMesh (library: libfvMotionSolvers and solver: solidBody) and AMI. The case should run for 3 sec, but it always stops between 0.02 - 0.075 sec, with the following text:

mpirun noticed that process rank 12 with PID **** on node *** exited on signal 8 (Floating point exception).

Surprisingly, in my case the solution doesn’t blow up, i.e diverges, it just simply stops working.
After searching a lot about it, I found some saying that there might be a bug, but I can not be sure. Anyway it is really important for me to make it work. I tried the following solutions but non worked:

• Trying different meshing techniques. (By checking the mesh no issues come up)
• Solution schemes:

• Introduce various relaxation factors, e.g. 0.3 for P and 0.7 for U,k,omega.
• Trying different solvers: GAMG, smoothSolver, PBiCGStab.
• change convergence criteria (Tolerance).

• discretization schemes:

• Use 1st order schemes for divergence, i.e upwind.
• Use different time schemes: backward, Euler, and CrankNilson.

• Run the simulation with Mapped fields (initialized by RANS simulation) and
without.
• Use a different version of OpenFOAM both (.org and .com).
• Use large and small timeStep, fixed or variable.
• Add limitations on Courant Number.
• Run the solution in serial.

Note: The solution for finer meshes (about 400k) worked normally when I run the case without dynamicMesh,i.e non roatiting AMI, or when using the library (libsixDoFRigidBodyMotion.so) and (sixDoFRigidBodyMotion) solver.

A solution or suggestions are highly appreciated.


Another question: Is there a trick to use the libsixDoFRigidBodyMotion library, but ensure a constant rotational speed (Omega) ?

Thanks in advance