[ErikNij]

Hi all,


Long time lurker, but first time poster, but I hope that I do this right, and that someone can help me and that I can help someone in the future.

In any case, I am trying to develop my own solver in OpenFOAM. This is a solver that should calculate the magnitude of induction heating due to a coiled wire with AC current.


There are of course many equations to be solved, however the first one I need to solve is simply:

curl(H) = J_Coil. J Coil is defined by the user, so this is a simple 1 unknown, 1 equation deal.


I tried to solve it in OpenFOAM in the following way. First, define the needed varibles

volVectorField HCoil
        (
            IOobject
            (
           "HCoil",
            runTime.timeName(),
            mesh,
            IOobject::MUST_READ,
            IOobject::AUTO_WRITE
        ),
        mesh
    );


 volVectorField JCoil
    (
        IOobject
        (
           "JCoil",
            runTime.timeName(),
            mesh,
            IOobject::MUST_READ,
            IOobject::AUTO_WRITE
        ),
        mesh
    );

And then the solver itself

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

#   include "setRootCase.H"
#   include "createTime.H"
#   include "createMesh.H"
#   include "createFields.H"
#   include "initConvergenceCheck.H"

    Info<< "\nStarting time loop\n" << endl;
    while (runTime.loop())
    {
        Info<< "Time = " << runTime.timeName() << nl << endl;

        tmp<fvVectorMatrix> HCoilEqn
        (
            fvc::curl(HCoil) - JCoil
        );
        solve(HCoilEqn);
        #include "convergenceCheck.H"

        runTime.write();
    }

    Info<< "End\n" << endl;

    return 0;

Sadly, this does not compile, and gives the following compile time error

MyEmFoam.C: In function ‘int main(int, char**)’:
MyEmFoam.C:88:9: error: no matching function for call to ‘Foam::tmp<Foam::fvMatrix<Foam::Vector<double> > >::tmp(Foam::tmp<Foam::GeometricField<Foam::Vector<double>, Foam::fvPatchField, Foam::volMesh> >)’
   88 |         );
      |         ^
In file included from /home/erik/foam/foam-extend-5.0/src/foam/lnInclude/tmp.H:140,
                 from /home/erik/foam/foam-extend-5.0/src/foam/lnInclude/PtrListI.H:29,
                 from /home/erik/foam/foam-extend-5.0/src/foam/lnInclude/PtrList.H:310,
                 from /home/erik/foam/foam-extend-5.0/src/foam/lnInclude/List.C:30,
                 from /home/erik/foam/foam-extend-5.0/src/foam/lnInclude/List.H:291,
                 from /home/erik/foam/foam-extend-5.0/src/foam/lnInclude/labelList.H:36,
                 from /home/erik/foam/foam-extend-5.0/src/foam/lnInclude/Pstream.H:43,
                 from /home/erik/foam/foam-extend-5.0/src/foam/lnInclude/OPstream.H:35,
                 from /home/erik/foam/foam-extend-5.0/src/foam/lnInclude/parRun.H:35,
                 from /home/erik/foam/foam-extend-5.0/src/finiteVolume/lnInclude/fvCFD.H:4,
                 from MyEmFoam.C:61:
/home/erik/foam/foam-extend-5.0/src/foam/lnInclude/tmpI.H:74:8: note: candidate: ‘Foam::tmp<T>::tmp(const Foam::tmp<T>&, bool) [with T = Foam::fvMatrix<Foam::Vector<double> >]’
   74 | inline Foam::tmp<T>::tmp(const tmp<T>& t, bool allowTransfer)
      |        ^~~~
/home/erik/foam/foam-extend-5.0/src/foam/lnInclude/tmpI.H:74:8: note:   candidate expects 2 arguments, 1 provided
/home/erik/foam/foam-extend-5.0/src/foam/lnInclude/tmpI.H:50:8: note: candidate: ‘Foam::tmp<T>::tmp(const Foam::tmp<T>&) [with T = Foam::fvMatrix<Foam::Vector<double> >]’
   50 | inline Foam::tmp<T>::tmp(const tmp<T>& t)
      |        ^~~~
/home/erik/foam/foam-extend-5.0/src/foam/lnInclude/tmpI.H:50:40: note:   no known conversion for argument 1 from ‘Foam::tmp<Foam::GeometricField<Foam::Vector<double>, Foam::fvPatchField, Foam::volMesh> >’ to ‘const Foam::tmp<Foam::fvMatrix<Foam::Vector<double> > >&’
   50 | inline Foam::tmp<T>::tmp(const tmp<T>& t)
      |                          ~~~~~~~~~~~~~~^
/home/erik/foam/foam-extend-5.0/src/foam/lnInclude/tmpI.H:41:8: note: candidate: ‘Foam::tmp<T>::tmp(const T&) [with T = Foam::fvMatrix<Foam::Vector<double> >]’
   41 | inline Foam::tmp<T>::tmp(const T& tRef)
      |        ^~~~
/home/erik/foam/foam-extend-5.0/src/foam/lnInclude/tmpI.H:41:35: note:   no known conversion for argument 1 from ‘Foam::tmp<Foam::GeometricField<Foam::Vector<double>, Foam::fvPatchField, Foam::volMesh> >’ to ‘const Foam::fvMatrix<Foam::Vector<double> >&’
   41 | inline Foam::tmp<T>::tmp(const T& tRef)
      |                          ~~~~~~~~~^~~~
/home/erik/foam/foam-extend-5.0/src/foam/lnInclude/tmpI.H:32:8: note: candidate: ‘Foam::tmp<T>::tmp(T*) [with T = Foam::fvMatrix<Foam::Vector<double> >]’
   32 | inline Foam::tmp<T>::tmp(T* tPtr)
      |        ^~~~
/home/erik/foam/foam-extend-5.0/src/foam/lnInclude/tmpI.H:32:29: note:   no known conversion for argument 1 from ‘Foam::tmp<Foam::GeometricField<Foam::Vector<double>, Foam::fvPatchField, Foam::volMesh> >’ to ‘Foam::fvMatrix<Foam::Vector<double> >*’
   32 | inline Foam::tmp<T>::tmp(T* tPtr)
      |                          ~~~^~~~
make: *** [MyEmFoam.dep:736: Make/linux64GccDPInt32Opt/MyEmFoam.o] Error 1

From what I can understand, there are 6 versions of the fvVectorMatrix, 3 of which take 2 inputs, which I'm not doing, and 3 more that take 1 argument, which should be of the type Foam::fvMatrix<Foam::Vector> >, and I see that my one input is of the type Foam::tmp<Foam::GeometricField<Foam::Vector<double >, Foam::fvPatchField, Foam::volMesh> >


However, from other times that I have seen solve be used (in simpleFoam for example) it seems like they are doing something similar to me, so I am confused as to what is going wrong.



At the risk of sounding stupid, I thought that fvVectorMatrix would make the matrix, so to me, I dont understand why it needs a matrix as an input, as I thought that it makes the matrix.



Anyways, I hope that someone can help for. For the record, I am using foam-extend 5.0, as the old electromagnetic solver used fe. However, I can switch off of it if needed, but I think it should still work.

[NiklasW]

Hi
I think the problem isn't the fvVectorMatrix itself, but that you're trying to construct it as a temporary object with that tmp<> wrapper.
So if there is no specific reason for you to make that matrix temporary, I'd suggest to leave that tmp<> wrapper away. I hope that helps.


Edit: I rethought this and no longer think that is the problem. Sorry about that

[ErikNij]

Ah thanks anyways.

I did try something, I changed fvc::curl to fvm::lapcian, and then it did compile, however the mathematics are of course wrong. So perhaps this indicates that fvc::curl is not expecting as I expect it to. What do you think about this?


Edit:

I think I might have figured it out after looking at the fourm: problem with fvc:div

TL;DR fvc::curl (and other fvc:: operators) are explicit, that is to say they don't output a matrix, only fvm:: operators do that. So I want to solve my problem, I will need to find out how to express my problem only using fvm:: operators. Back to the books I guess