[memento]

Hi Foamers,

I implemented my own custom Temperature boundary condition in OpenFOAM. While the incompressible case is running fine and the solution is indeed plausible, my problem emerges when I port the bc to a compressible solver.


My BC is for cyclic cases, I'm using AMI.
I use the fixedJumpAMI condition and developed my BC based on that.

As I said above when runnning in simpleFOAM everything works fine and as intended.
But during the thermo.correct() in EEqn.H the solver crashes with SigFpe.


So as I debugged, multiple things seemed odd:

• In one case I decomposed the domain and usually the solver crashes as mentioned above during the thermo.correct().
  Here I started writing the gMin/gMax values of the used fields (T, p, psi, rho, mu, he and alpha) on all boundaries. Now the solvers crashes with SigFpe during evaluation of gMin/gMax on the processor patches.

• I let the solver write those fields before the thermo.correct() and the values of he seemed odd (the he transport eq. is solved at this point). All values of e are in the range of -1e8 but on the inlet patch there are 2 patch face values with +4e6 and +2e6. The list reads as follows:

inlet
 {
  type         energyJumpAMI;
  patchType cyclicAMI;
  jump         uniform -17957379;
  value        nonuniform List<scalar>
80
(
4771773.9
2337384.7
-1849468.6
-4666132.8
...
)
}

Does anybody already have experience implementing a custom BC for Temperature in compressible cases?
Is there something I haven't considered?
I'm really stuck and moving in circles.