[sharonyue]

Hi guys,

There is someth wrong upon this constructor:

volScalarField K
    (
        IOobject
        (
            "K",
            runTime.timeName(),
            mesh,
            IOobject::NO_READ,
            IOobject::NO_WRITE
        ),
        0.75*Cds*phase2->rho()*magUr*bp/d1
    );

The solver can be compiled with no error. but when it runs:

NO_READ specified for read-constructor of object n of class IOobject

More speciafically, Cds and magUr is:

volScalarField Cds
    (
        neg(Re - 1000)*(24.0*(1.0 + 0.15*pow(Re, 0.687))/Re)
      + pos(Re - 1000)*0.44
    );

dimensionedScalar rho_;

volVectorField Ur(U1 - U2);
volScalarField magUr(mag(Ur));

volScalarField bp(pow(Alpha2, -2.65));

volScalarField d1
    (
        IOobject
        (
            "d1",
            mesh.time().timeName(),
            mesh,
            IOobject::MUST_READ,
            IOobject::AUTO_WRITE
        ),
        mesh
     );

I dont know what happened, Thanks.

PS.
Even I change k to be:

volScalarField K
    (
        IOobject
        (
            "K",
            runTime.timeName(),
            mesh,
            IOobject::NO_READ,
            IOobject::NO_WRITE
        ),
        d1
    );

The same error

[Bernhard]

How do you know the problem is in the definition of "K"? I do not see that from your error message. Are you sure it compiles without warnings? (run wclean first)

[tomislav_maric]

You are trying to call a non-existing constructor for the GeometricField, by forgetting to pass the 'mesh' as a constant reference. I made the small example app and compiled it in 2.2.x (nothing has changed in GeometricField, for a looong time).

Bernhard was right, your code cannot compile.

Here is the fix:

#include "fvCFD.H"

// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
// Main program:

int main(int argc, char *argv[])
{
    #include "setRootCase.H"
    #include "createTime.H"
    #include "createMesh.H"

    // This code does't compile.
    //volScalarField K
    //(
        //IOobject
        //(
            //"K",
            //runTime.timeName(),
            //mesh,
            //IOobject::NO_READ,
            //IOobject::NO_WRITE
        //),
        //0.75 //  The rest doesn't matter *Cds*phase2->rho()*magUr*bp/d1
    //);
    
    volScalarField K
    (
        IOobject
        (
            "K",
            runTime.timeName(),
            mesh,
            IOobject::NO_READ,
            IOobject::NO_WRITE
        ),
        mesh, // You forgot to pass the const ref to the mesh.
        0.75 //  The rest doesn't matter *Cds*phase2->rho()*magUr*bp/d1
    );

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

    Info<< "\nEnd\n" << endl;
    return 0;
}

The latter definition of K compiles and runs.

Among the GeometricField constructors there isn't one with the signature you have tried to use : GeometricField(const IOobject& , T).

What happens in the working code above: the GeometricField(const IOobject&, const dimensioned<T>&, ...) get's called and the scalar '0.75' gets implicitly converted to the 'dimensioned<scalar>' by 'dimensioned<T>::dimensioned<T>(const T&)' constructor. Don't worry about '...', those arguments have predefined default values. If you want, you can redefine them - they are about defining types of BCs for the field K.

[triple_r]

Are your sure the problem is in K? The error message says it is in object "n". I think you should be fine defining "K" as you have if it compiles. Post your definition of "n" and I might be able to help you more.

[sharonyue]

Thanks you guys all, especially @tomislav_maric. Your validation is feasible.

1. This is my fault. It is caused by volScalarField n.

2. I tried @tomislav_maric's method, it worked.

[tomislav_maric]

Originally Posted by triple_r

Are your sure the problem is in K? The error message says it is in object "n". I think you should be fine defining "K" as you have if it compiles. Post your definition of "n" and I might be able to help you more.

Maybe there is another error message somewhere else, but the signature of the constructor used for "K" was wrong, try and uncomment that constructor in the code I posted, you'll se what I mean.

[triple_r]

You are absolutely right, in your example the mesh has to be specified. But the owner's original K initialization was using another scalar field, so it didn't need passing the mesh. And because he mentioned that there was no compile-time errors, a constructor was matching that pattern.

[sharonyue]

I think both you two are right.~

[tomislav_maric]

we are all right, the main thing is the problem got solved

@triple_r: of course, the other ctor call is valid

[triple_r]

Yup! As long as the post owner is happy and his problem is solved, we are all happy