[Bibill]

Dear all,

I am implementing an own EpsilonWallFunction for a k-epsilon simulation. Inside the WF I would like to compute both the strain rate and vorticity invariants.

Here is the structure of the code :

void Foam::MyepsilonWallFunctionFvPatchScalarField::calculate
(
    const turbulenceModel& turbModel,
    const List<scalar>& cornerWeights,
    const fvPatch& patch,
    scalarField& G0,
    scalarField& epsilon0
)
{

    const volVectorField& Ut = turbModel.U();

    /* Strain rate invariant */
    volSymmTensorField S = symm(fvc::grad(Ut));

    /* Vorticity invariant */
    volTensorField W = skew(fvc::grad(Ut));

...

}

If we consider a rectangular wind tunnel, the WF works well for a single wall (let's say the ground), for two different and unconnected walls (e.g the ground and the top wall) but it crashes for two connected walls (e.g the ground and a side wall).

The errors are:

Starting time loop

Time = 1

smoothSolver:  Solving for Ux, Initial residual = 1, Final residual = 0.000410734772, No Iterations 5
smoothSolver:  Solving for Uy, Initial residual = 0.999853259, Final residual = 0.000909577149, No Iterations 4
smoothSolver:  Solving for Uz, Initial residual = 0.999796967, Final residual = 0.000793379745, No Iterations 4
GAMG:  Solving for p, Initial residual = 1, Final residual = 0.000988755628, No Iterations 23
time step continuity errors : sum local = 4.17199017e-05, global = 1.25994915e-05, cumulative = 1.25994915e-05
[c06c4f9bf5aa:13379] 1 more process has sent help message help-mpi-errors.txt / mpi_errors_are_fatal
[c06c4f9bf5aa:13379] Set MCA parameter "orte_base_help_aggregate" to 0 to see all help / error messages
[c06c4f9bf5aa:13385] *** An error occurred in MPI_Waitall
[c06c4f9bf5aa:13385] *** reported by process [1137901569,0]
[c06c4f9bf5aa:13385] *** on communicator MPI COMMUNICATOR 3 SPLIT FROM 0
[c06c4f9bf5aa:13385] *** MPI_ERR_TRUNCATE: message truncated
[c06c4f9bf5aa:13385] *** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,
[c06c4f9bf5aa:13385] ***    and potentially your MPI job)

NB:
• After some tests, the error occurs at the gradient computation line;
• It runs in serial;
• It runs for 5 processors but not for 15;


I would truly appreciate any help or comments!

thanks in advance!