[wstapel]

Dear Foamers,


I've got a problem with coding a parallel calculation of the flow-rate which is to be evaluated on the run (not by integratePatchField).

My overall purpose is the following: I want the flow-rate to influence the inlet boundary conditions of an internal flow.

So far I've actually managed to evaluate the flow-rate, but it only works if I'm performing parallel calculations on my local machine or on one single node of a cluster. However if I try to run the very same calculation on a cluster distributed over several nodes, it always yields a zero flow-rate.
So obviously the problem must have something to do with parallelisation or with mpi ...

Finally here is the code, I added to the original icoFoam solver:

// Declaration and Initialisation of Variables

word outletPatchName = "OUTLET";

// search for label of the outlet patch by the name within the boundary dictionary

label inletPatchi = mesh.boundaryMesh().findPatchID(outletPatchName);

// Calculate volumetric flow-rate

scalar volFluxOutlet = sum(phi.boundaryField()[outletPatchi]);

// Write flow-rate

Info<< "Volumetric Flowrate: " << volFluxOutlet << " m^3/s" << endl;

/**********************************************/

I'd be glad for any helpful comments.

Waldemar

[kmooney]

Hi Waldemar,

Try using

gSum(...)

instead of

sum(...)

I believe that gSum has built-in mpi reduction operators while 'sum' does not.

I hope that helps!
Cheers,
Kyle

[wstapel]

Hi Kyle,

it works fine. Thank you very much!

[kmooney]

Excellent! Good to hear.