[cbcoutinho]

I'm creating a new fvOptions source option based on the `meanVelocityForce` derived source in OpenFOAM 4.x. When I added a new scalar to the .H and .C files, I get a compilation error related to an incorrect conversion occurring in the constructor. More specifically, the error message I'm seeing is:

meanMassSource.C: In constructor ‘Foam::fv::meanMassSource::meanMassSource(const Foam::word&, const Foam::word&, const Foam::dictionary&, const Foam::fvMesh&)’:
meanMassSource.C:95:16: error: cannot convert ‘Foam::ITstream’ to ‘Foam::scalar {aka double}’ in initialization
     rAPtr_(NULL)
                ^
/home/redclient04/OpenFOAM/OpenFOAM-4.x/wmake/rules/General/transform:8: recipe for target 'Make/linux64GccDPInt32Opt/meanMassSource.o' failed
make: *** [Make/linux64GccDPInt32Opt/meanMassSource.o] Error 1

If I comment out the new scalar addition to the dictionary that is created in the constructor, the code compiles without an issue. Below are snippets of my fvOption files named meanMassSource:

meanMassSource.H:

#ifndef meanMassSource_H
#define meanMassSource_H

#include "autoPtr.H"
#include "topoSetSource.H"
#include "cellSet.H"
#include "fvMesh.H"
#include "volFields.H"
#include "cellSetOption.H"

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

namespace Foam
{
namespace fv
{

/*---------------------------------------------------------------------------*\
               Class meanMassSource Declaration
\*---------------------------------------------------------------------------*/

class meanMassSource
:
    public cellSetOption
{
protected:

    // Protected data
        vector Ubar_;                       //- Average velocity
        scalar gradC0_;                     //- Concentration gradient before correction
        scalar dGradC_;                     //- Change in concentration gradient
// vvv molarFlux_ is the new scalar I want to add vvv
        scalar molarFlux_;                  //- Molar flux at membrane walls in [mol/m2.s]
        vector flowDir_;                    //- Flow direction
        scalar relaxation_;                 //- Relaxation factor
        autoPtr<volScalarField> rAPtr_;     //- Matrix 1/A coefficients field pointer

.
.
.

And the associated meanMassSource.C file:

#include "meanMassSource.H"
#include "fvMatrices.H"
#include "DimensionedField.H"
#include "IFstream.H"
#include "addToRunTimeSelectionTable.H"

// * * * * * * * * * * * * * Static Member Functions * * * * * * * * * * * * //

namespace Foam
{
namespace fv
{
    defineTypeNameAndDebug(meanMassSource, 0);

    addToRunTimeSelectionTable
    (
        option,
        meanMassSource,
        dictionary
    );
}
}


// * * * * * * * * * * * * * Private Member Functions  * * * * * * * * * * * //

void Foam::fv::meanMassSource::writeProps
(
    const scalar gradC
) const
{
    // Only write on output time
    if (mesh_.time().writeTime())
    {
        IOdictionary propsDict
        (
            IOobject
            (
                name_ + "Properties",
                mesh_.time().timeName(),
                "uniform",
                mesh_,
                IOobject::NO_READ,
                IOobject::NO_WRITE
            )
        );
        propsDict.add("gradient", gradC);
        propsDict.regIOobject::write();
    }
}


// * * * * * * * * * * * * * * * * Constructors  * * * * * * * * * * * * * * //

Foam::fv::meanMassSource::meanMassSource
(
    const word& sourceName,
    const word& modelType,
    const dictionary& dict,
    const fvMesh& mesh
)
:
    cellSetOption(sourceName, modelType, dict, mesh),
    Ubar_(coeffs_.lookup("Ubar")),
    gradC0_(0.0),
    dGradC_(0.0),
    molarFlux_(coeffs_.lookup("molarFlux")),    // Again, here is where the new scalar is being looked up and put into the dictionary
    flowDir_(Ubar_/mag(Ubar_)),
    relaxation_(coeffs_.lookupOrDefault<scalar>("relaxation", 1.0)),
    rAPtr_(NULL)
.
.
.

I snipped the remaining source of both files because they are so far exactly the same as meanVelocityForce - the implementation hasn't been added yet. If I comment out the two offending lines where molarFlux_ is defined in the header file and is placed into the dictionary in the .C file, the fvOption compiles fine.

I've noticed that various derived sources have differing number of entries in their dictionaries, so I didn't think the number of entries is hard-coded somewhere. Since both the header and source files match up (at least to my untrained eyes), I can't tell where the problem is.

Can someone tell me where I'm missing something?

[cbcoutinho]

I spent a little more time digging around, and I found a fantastic answer to my question by Prof. Jasak - see his post here

Essentially the issue is that the lookup function involves a constructor of a class, but scalars are just doubles, which don't have constructors because doubles are primitive types. You need to wrap the lookup function with a readScalar function:

Change the offending line:

molarFlux_(coeffs_.lookup("molarFlux")),

to:

molarFlux_(readScalar(coeffs_.lookup("molarFlux"))),

[frantov]

Thanks Chris


for my case, this would be the solution assuming we are reading the scalar "height" from liquidDict



OF7
const scalar height(readScalar(liquidDict.lookup("height")));



OF8 (and onwards)

const scalar height(liquidDict.lookup<scalar>("height"));

[XieDaxia]

Originally Posted by cbcoutinho

I'm creating a new fvOptions source option based on the `meanVelocityForce` derived source in OpenFOAM 4.x. When I added a new scalar to the .H and .C files, I get a compilation error related to an incorrect conversion occurring in the constructor. More specifically, the error message I'm seeing is:

meanMassSource.C: In constructor ‘Foam::fv::meanMassSource::meanMassSource(const Foam::word&, const Foam::word&, const Foam::dictionary&, const Foam::fvMesh&)’:
meanMassSource.C:95:16: error: cannot convert ‘Foam::ITstream’ to ‘Foam::scalar {aka double}’ in initialization
     rAPtr_(NULL)
                ^
/home/redclient04/OpenFOAM/OpenFOAM-4.x/wmake/rules/General/transform:8: recipe for target 'Make/linux64GccDPInt32Opt/meanMassSource.o' failed
make: *** [Make/linux64GccDPInt32Opt/meanMassSource.o] Error 1

If I comment out the new scalar addition to the dictionary that is created in the constructor, the code compiles without an issue. Below are snippets of my fvOption files named meanMassSource:

meanMassSource.H:

#ifndef meanMassSource_H
#define meanMassSource_H

#include "autoPtr.H"
#include "topoSetSource.H"
#include "cellSet.H"
#include "fvMesh.H"
#include "volFields.H"
#include "cellSetOption.H"

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

namespace Foam
{
namespace fv
{

/*---------------------------------------------------------------------------*\
               Class meanMassSource Declaration
\*---------------------------------------------------------------------------*/

class meanMassSource
:
    public cellSetOption
{
protected:

    // Protected data
        vector Ubar_;                       //- Average velocity
        scalar gradC0_;                     //- Concentration gradient before correction
        scalar dGradC_;                     //- Change in concentration gradient
// vvv molarFlux_ is the new scalar I want to add vvv
        scalar molarFlux_;                  //- Molar flux at membrane walls in [mol/m2.s]
        vector flowDir_;                    //- Flow direction
        scalar relaxation_;                 //- Relaxation factor
        autoPtr<volScalarField> rAPtr_;     //- Matrix 1/A coefficients field pointer

.
.
.

And the associated meanMassSource.C file:

#include "meanMassSource.H"
#include "fvMatrices.H"
#include "DimensionedField.H"
#include "IFstream.H"
#include "addToRunTimeSelectionTable.H"

// * * * * * * * * * * * * * Static Member Functions * * * * * * * * * * * * //

namespace Foam
{
namespace fv
{
    defineTypeNameAndDebug(meanMassSource, 0);

    addToRunTimeSelectionTable
    (
        option,
        meanMassSource,
        dictionary
    );
}
}


// * * * * * * * * * * * * * Private Member Functions  * * * * * * * * * * * //

void Foam::fv::meanMassSource::writeProps
(
    const scalar gradC
) const
{
    // Only write on output time
    if (mesh_.time().writeTime())
    {
        IOdictionary propsDict
        (
            IOobject
            (
                name_ + "Properties",
                mesh_.time().timeName(),
                "uniform",
                mesh_,
                IOobject::NO_READ,
                IOobject::NO_WRITE
            )
        );
        propsDict.add("gradient", gradC);
        propsDict.regIOobject::write();
    }
}


// * * * * * * * * * * * * * * * * Constructors  * * * * * * * * * * * * * * //

Foam::fv::meanMassSource::meanMassSource
(
    const word& sourceName,
    const word& modelType,
    const dictionary& dict,
    const fvMesh& mesh
)
:
    cellSetOption(sourceName, modelType, dict, mesh),
    Ubar_(coeffs_.lookup("Ubar")),
    gradC0_(0.0),
    dGradC_(0.0),
    molarFlux_(coeffs_.lookup("molarFlux")),    // Again, here is where the new scalar is being looked up and put into the dictionary
    flowDir_(Ubar_/mag(Ubar_)),
    relaxation_(coeffs_.lookupOrDefault<scalar>("relaxation", 1.0)),
    rAPtr_(NULL)
.
.
.

I snipped the remaining source of both files because they are so far exactly the same as meanVelocityForce - the implementation hasn't been added yet. If I comment out the two offending lines where molarFlux_ is defined in the header file and is placed into the dictionary in the .C file, the fvOption compiles fine.

I've noticed that various derived sources have differing number of entries in their dictionaries, so I didn't think the number of entries is hard-coded somewhere. Since both the header and source files match up (at least to my untrained eyes), I can't tell where the problem is.

Can someone tell me where I'm missing something?

Hi~ I'm trying to simulating a periodic case with heat transfer. I have tried to use a constant heat sink for the Temperature(scalar) field, but the temperature keep increasing, so I also want to use a method similar to the meanvelocityforce for the scalar field.
Could you plase show me the complete code for study? Thanks a lot, whether yes or not.