[Jernej Pirnar]

Hi all,

I am interested in incorporating the effect of a propeller using rotorDiskSource class. I am using OpenFOAM-5.x. I have found a work (report and input files) by Capitao Patrao [Capitao Patrao, A.: Description and validation of the rotorDiskSource class for propeller performance estimation. In Proceedings of CFD with OpenSource Software, 2017, Edited by Nilsson. H., http://dx.doi.org/10.17196/OS_CFD#YEAR_2017] from the Chalmers University of Technology. The work gives a fix to the error in how the local forces are calculated and implements more general blade tip correction. I have successfully modified the original OF code. The corrected rotorDiskSource works without any problems when the solver (simpleFoam) is run in serial. However, the difficulties begin in parallel where the output is strange. Below I paste several outputs of rotorDisk.

Serial case:
rotorDisk output:
min/max(AOA) = 1.6620832, 4.5369071
min/max(eps) = 25.235116, 59.281187
Rotor thrust = 671.50036
Rotor torque = 188.8365
Rotor power = 71041.172
Rotor propeller efficiency = 0.80412826
Effective drag = -576.07232
Effective lift = 671.50036

Parallel case (2 partitions):
rotorDisk output:
min/max(AOA) = 1.6619861, 4.5367581
min/max(eps) = 25.235298, 59.278707
Rotor thrust = 664.33664
Rotor torque = 186.83229
Rotor power = 70287.182
Rotor propeller efficiency = 0.80408372
Effective drag = -576.0604
Effective lift = 671.48486

Parallel case (3 partitions):
rotorDisk output:
min/max(AOA) = 1.6621062, 4.5363851
min/max(eps) = 5.729578e+16, -5.729578e+16
Rotor thrust = 0
Rotor torque = 0
Rotor power = 0
Rotor propeller efficiency = nan
Effective drag = -576.04504
Effective lift = 671.45967

Parallel case (6 partitions):
rotorDisk output:
min/max(AOA) = 1.6617572, 4.5364101
min/max(eps) = 25.235737, 45.521711
Rotor thrust = 186.92322
Rotor torque = 52.674087
Rotor power = 19816.238
Rotor propeller efficiency = 0.80247451
Effective drag = -576.03327
Effective lift = 671.44934


Does anybody have some experience in fixing the issue and can give a piece of advice?

Thank you.

Jernej

[jf_vt]

I had experiments with case running in serial and posing problem in parallel when using ggi coupling interface.

In my case it was traced back to a bug in openmpi implementation

I have no idea if your problem is of the same kind though

Can you post the case?

JF

[Jernej Pirnar]

Hi Jeff,

I am using the original case (prop_incomp_refCase) as described in Proceedings of CFD with OpenSource Software, 2017 (Edited by Nilsson H. http://dx.doi.org/10.17196/OS_CFD#YEAR_2017). Following the link the case files can be downloaded.

Angle of attack, effective lift and drag are invariant to the number of partitions (as expected). What bothers me is that eps, rotor thrust, torque, power, and propeller efficiency are not .

Regards,
Jernej

[lebc]

How does your flow field look like?

If you generate a slice in one of the rotor planes for all the cases you are running, do they look the same?

It may be just a matter of taking the information from all the processors for the calculation of the variables (usually the sumop() - or something like this - part of the code)

Try the slices, check if the physics make sense.

Best Regards,
Luis

[alexey_sw]

Hi!
I have a very similar experience using rotorDisk. Like you, I successfully modified the original OF code using Capitao Report. Anyway with corrections or without rotorDisk dont work adequately. Besides problems with parallel run, there are also some oscillation when run simpleFoam in serial. Oscillation in rotor forces generate oscillation in field U which leads to resonance (it`s my explanation ).

You can see how it look like in slice through the rotor (colored by force magnitude).
https://www.youtube.com/watch?v=D7cN2Dcwixw
https://www.youtube.com/watch?v=zQzQw7gU5XE
It was observed in case similar to tutorials simpleFoam/rotorDisk, was slightly changed mesh only.

I managed to get rid of oscillations by decreasing the relaxation factor U to 0.45. Maybe its not good method? Have anyone thoughts about the reason of this strange phenomena?

[vLhomme]

Hi !

I think I solved the issue !

With OF1812 and using the same files and following the same report.

I tried several decompositions of the case using SIMPLE method, in decomposeParDict I have for example :

numberOfSubdomains 3;

method          simple;

simpleCoeffs
{
    n               (1 3 1);
    delta           0.001;
}

I give below the output of my attempts with the 'n' parameter for the simple decomposition :

n (1 1 1) (undecomposed, reference)

rotorDisk output:
min/max(AOA)   = 1.6603615, 4.5343289
min/max(eps) = 25.238325, 59.282909
Rotor thrust = 671.04952
Rotor torque = 188.71667
Rotor power = 70986.411
Rotor propeller efficiency = 0.80420828
Effective drag = -575.71016
Effective lift = 671.04952

n (1 2 1)

rotorDisk output:
min/max(AOA)   = 1.660361, 4.5343286
 min/max(eps) = 25.238325, 59.282909
 Rotor thrust = 335.52474
 Rotor torque = 94.358332
 Rotor power = 35493.204
 Rotor propeller efficiency = 0.80420828
 Effective drag = -575.71012
 Effective lift = 671.0494

n (2 1 2)

rotorDisk output:
    min/max(AOA)   = 1.6603613, 4.5343287
 min/max(eps) = 25.238325, 59.282909
 Rotor thrust = 167.76238
 Rotor torque = 47.179168
 Rotor power = 17746.603
 Rotor propeller efficiency = 0.80420828
    Effective drag = -575.71014
    Effective lift = 671.0495

n(3 1 1)

rotorDisk output:
    min/max(AOA)   = 1.6603616, 4.534329
 min/max(eps) = 25.238325, 59.282909
 Rotor thrust = 274.3083
 Rotor torque = 77.168146
 Rotor power = 29027.057
 Rotor propeller efficiency = 0.80394276
    Effective drag = -575.71017
    Effective lift = 671.04953

From the cases above I would deduct that the forces are computed for only one of the processor domain (cause we obtain fractions of the values we get for the reference case) for example in the case (2 1 2), we divided the rotorDisk into quarter disk and we got the thrust, power and torque equal to the quarter of reference values.

n (1 3 1)

  min/max(AOA)   = 4.0015841, 7.9121525
 min/max(eps) = 5.729578e+16, -5.729578e+16
 Rotor thrust = 0
 Rotor torque = 0
 Rotor power = 0
 Rotor propeller efficiency =

and gave an error message

[v501:32819] *** Process received signal ***
[v501:32819] Signal: Floating point exception (8)
[v501:32819] Signal code:  (-6)
[v501:32819] Failing at address: 0x3ef00008033
[v501:32819] [ 0] /lib/x86_64-linux-gnu/libc.so.6(+0x354b0)[0x7f32e132d4b0]
[v501:32819] [ 1] /lib/x86_64-linux-gnu/libc.so.6(gsignal+0x38)[0x7f32e132d428]
[v501:32819] [ 2] /lib/x86_64-linux-gnu/libc.so.6(+0x354b0)[0x7f32e132d4b0]
[v501:32819] [ 3] /home/cfd/OpenFOAM/cfd-v1812/platforms/linux64GccDPInt32Opt/lib/libfvOptions.so(_ZNK4Foam2fv15rotorDiskSource9calculateINS_17geometricOneFieldEEEvRKT_RKNS_5FieldINS_6VectorIdEEEERKNS7_IdEERSA_bb+0xd5a)[0x7f32e3a3120a]
[v501:32819] [ 4] /home/cfd/OpenFOAM/cfd-v1812/platforms/linux64GccDPInt32Opt/lib/libfvOptions.so(_ZN4Foam2fv15rotorDiskSource6addSupERNS_8fvMatrixINS_6VectorIdEEEEi+0x396)[0x7f32e3a2d476]
[v501:32819] [ 5] simpleFoam[0x444659]
[v501:32819] [ 6] simpleFoam[0x420603]
[v501:32819] [ 7] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf0)[0x7f32e1318830]
[v501:32819] [ 8] simpleFoam[0x423f29]
[v501:32819] *** End of error message ***

From this case I think I can deduce that the values outputed here are taken from a node's domain that does not contain the rotating zone, values computed by the lines added in the report are not added from one processor to the other ...


Yes !! It's seems to be it ! I propose this modification in the file RotorDiskSourceTemplates.C (search for '//NEW MODIFICATION') :

//scalar etaProp = thrustEff*refVelEta_/powerEff; //NEW LINE //NEW MODIFICATION comment this as it can result in a division by 0
    if (output)
    {
        reduce(AOAmin, minOp<scalar>());
        reduce(AOAmax, maxOp<scalar>());
        reduce(dragEff, sumOp<scalar>());
        reduce(liftEff, sumOp<scalar>());
        reduce(epsMin, minOp<scalar>());//vlhomme 02/09/2019 MODIFICATION 
        reduce(epsMax, maxOp<scalar>());//vlhomme 02/09/2019 MODIFICATION 
        reduce(thrustEff, sumOp<scalar>());//vlhomme 29/08/2019 MODIFICATION 
        reduce(torqueEff, sumOp<scalar>());//vlhomme 29/08/2019 MODIFICATION 
        reduce(powerEff, sumOp<scalar>()); //vlhomme 29/08/2019 MODIFICATION 
        scalar etaProp = thrustEff*refVelEta_/powerEff; //vlhomme 29/08/2019 MODIFICATION  : I add this line cause here there is no risk of division by 0 
        Info<< type() << " output:" << nl
            << "    min/max(AOA)   = " << radToDeg(AOAmin) << ", "
            << radToDeg(AOAmax) << nl
            << " min/max(eps) = " << radToDeg(epsMin) << ", " //NEW LINE
            << radToDeg(epsMax) << nl //NEW LINE
            << " Rotor thrust = " << thrustEff << nl //NEW LINE
            << " Rotor torque = " << torqueEff << nl //NEW LINE
            << " Rotor power = " << powerEff << nl //NEW LINE
            << " Rotor propeller efficiency = " << etaProp << nl //NEW LINE
            << "    Effective drag = " << dragEff << nl
            << "    Effective lift = " << liftEff << endl;
    }
}

I now do have the same output with the decomposition I tried before, and even using scotch with 20 CPUs I still got reference results .

[farzadmech]

Dear Friends
I want to use rotorDisk example in my case which is a lagrangian solver(I have already added fvOptions and I can add external momentum sources). according to the code;

https://github.com/OpenFOAM/OpenFOAM...stem/fvOptions

which I copied here;

/*--------------------------------*- C++ -*----------------------------------*\
  =========                 |
  \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox
   \\    /   O peration     | Website:  https://openfoam.org
    \\  /    A nd           | Version:  6
     \\/     M anipulation  |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    object      fvOptions;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

disk
{
    type            rotorDisk;

    selectionMode   cellZone;
    cellZone        rotatingZone;

    fields          (U);    // Names of fields on which to apply source
    nBlades         3;      // Number of blades
    tipEffect       0.96;   // Normalised radius above which lift = 0

    inletFlowType   local;  // Inlet flow type specification
    inletVelocity   (0 1 0);

    geometryMode    specified;

    origin          (0 0 0);
    axis            (0 1 0);

    refDirection    (0 0 1);  // Reference direction
                              // - used as reference for psi angle
    rpm             1000;
    // pointAbove    (0 0 0.25);

    trimModel       fixedTrim;  // fixed || targetForce

    rhoRef          1000;
    rhoInf          1;

    fixedTrimCoeffs
    {
         theta0         0;
         theta1c        0;
         theta1s        0;
    }

    flapCoeffs
    {
        beta0           0;  // Coning angle [deg]
        beta1c          0;  // Lateral flapping coeff (cos coeff)
        beta2s          0;  // Longitudinal flapping coeff (sin coeff)
    }

    blade
    {
        data
        (
            (profile1 (0.1 -6 0.02))
            (profile1 (0.25 -6 0.02))
        );
    }

    profiles
    {
        profile1
        {
            type lookup;
            data
            (
                (-90 0.21 1.45)
                (-18 0.21 1.45)
                (-16 0.165 1.3)
                (-14 0.125 1.1)
                (-12 0.092 0.95)
                (-10 0.07 0.8)
                (-8 0.05 0.64)
                (-6 0.04 0.5)
                (-4 0.028 0.32)
                (-2 0.022 0.18)
                (0 0.02 0)
                (2 0.022 0.18)
                (4 0.028 0.32)
                (6 0.04 0.5)
                (8 0.05 0.64)
                (10 0.07 0.8)
                (12 0.092 0.95)
                (14 0.125 1.1)
                (16 0.165 1.3)
                (18 0.21 1.45)
                (90 0.21 1.45)
            );
        }
    }
}

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

I can add effect of blades in my simulation, but I can't and I can not see any effect. can anyone help me?


Thanks,
Farzad

Using rotorDisk(or any other blade simulator) for mixing purpose by applying fvOption

[AlxB]

Many thanks vLhomme for the fix you proposed.

Have you managed to make this working with OFv2112 as well ?

From my side, following the same files and report, it seems there is now something wrong when compiling.