[Swift]

Greetings to the Forum,

I have adapted the propeller example to suite my test case which is a mixer in a tank. I have a stator domain (region0) and a rotor domain (region1). I am able to run the model on a single core as well as in parallel processing. I notice however, that the parallel case is taking longer to solve than when it solves on a single core. I suspect it has to do with the way the solution is divided between the processors. I edited the decomposeParDict from the propeller example as follows to reflect the new orientation of my case:

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox           |
|  \\    /   O peration     | Version:  2.3.1                                 |
|   \\  /    A nd           | Web:      www.OpenFOAM.org                      |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    object      decomposeParDict;
}

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

numberOfSubdomains 3;



method          hierarchical;

hierarchicalCoeffs
{
    n               (1 1 3);
    delta           0.001;
    order           xyz;
}

/*numberOfSubdomains 3;

method          scotch;

distributed     no;

roots           ( );
*/
// ************************************************************************* //

You can see that this will divide the case in the z direction rather than the y direction like the propeller case.

I also tried the "scotch" method that is commented out in the code above.

I suspect the reason the parallel processing is taking longer is because of the way the solution is being divided up.

I would like to know if it is possible to to divide the solution such, that the stator region solves on one processor and the rotor region on the other. I realise that this would mean I only use two processors but I would like to see if this would improve the solution.

Thanks in advance for the help!

Thomas