[evgenii]

Hi, all.
Does anyone work with curl in OpenFoam?
I have some problems with it.
Hope to get some help or advice.

I wrote:

solve(
mu0*fvm::ddt(E)
+ fvc::curl(H)
);

solve(
epsilon0*fvm::ddt(H)
+ fvc::curl(E)
);


E,H - volVectorField, mu0, epsilon0 - dimensionedScalar

When I try to compile following error appears:
no matching function to call for solve(...)

Where am I mistaken?
Thank you.
Eugene

Hello Eugene,

Did you try fvm instead of fvc:

solve(
mu0*fvm::ddt(E)
+ fvm::curl(H)
);

I think it's not a problem with curl but a syntax problem. Had you also define the equation to solve... for example:

tmp<fvvectormatrix> EEqn
(
mu0*fvm::ddt(E)
+ fvm::curl(H)
);
EEqn().relax();
EEqn().solve();

Try this and tell me :-).

Fred

Or this...

tmp<fvvectormatrix> EEqn
(
mu0*fvm::ddt(E)
==
- fvc::curl(H)
);
EEqn().relax();
EEqn().solve();

The variable to solve in the left part of the equation and the source term on the right one.

Fred

[evgenii]

Thank you so much! It's now compiling succesfully (How it works I'll know later )
There is only fvc::curl, there's no fvm::curl.
But you are right, I needed to define an equation in manner you gave.

[hjasak]

Curl only comes in the explicit version (fvc) because the implicit one couples the components of the vector. The current matrix support cannot handle that but the new version (block matrix) will come in due course). So, there's no fvm::curl at the moment.

As for the calculation of the curl, have a look at fvcCurl.C. There are two implementations

- the one based on the Gauss Theorem is commented out.

- the one with the Hodge dual is active and allows you to use your favourite gradient calculation algorithm

Enjoy,

Hrv

Thanks Hrv for your comment and your advices .

[evgenii]

Thank for comments Hrv, it is very useful as usual

[gort]

Hi all,
i am trying to implement the following equation in the solver
curl(B)=mu0*j
B and j are volVectorField and mu0 is dimensionedScalar
mu0 and j are known.
Corresponding entry in the solver is
solve (fvc::curl(B)==mu0*j)
The "response" of compiler is
error: no match for 'operator==' in 'Foam::fvc::curl(const Foam::GeometricField<type,>&) [with Type = Foam::Vector<double>]() == 0'
/shared/home/gort//OpenFOAM/OpenFOAM-1.3/src/OpenFOAM/lnInclude/cellModelI.H:124 : note: candidates are: bool Foam::operator==(const Foam::cellModel&, const Foam::cellModel&)
/shared/home/gort//OpenFOAM/OpenFOAM-1.3/src/OpenFOAM/lnInclude/edgeI.H:147: note: bool Foam::operator==(const Foam::edge&, const Foam::edge&)

What is the right FOAM syntax for this equation? Will be happy for any advises.

Best regards,
Sergej

[mvoss]

hi,

since this is looong time ago... did you managed to solve your curl(B)=mu0*j with OF ??

neewbie

[makaveli_lcf]

Dear Hrv,
is there a point why Hodge dual is taken in favour of the Gauss reconstruction of curl?
Thank you in advance!
Best regards,
Sasha

Originally Posted by hjasak

Curl only comes in the explicit version (fvc) because the implicit one couples the components of the vector. The current matrix support cannot handle that but the new version (block matrix) will come in due course). So, there's no fvm::curl at the moment.

As for the calculation of the curl, have a look at fvcCurl.C. There are two implementations

- the one based on the Gauss Theorem is commented out.

- the one with the Hodge dual is active and allows you to use your favourite gradient calculation algorithm

Enjoy,

Hrv

[Kummi]

Hello Foamers,
It's been a long while - however I have some query related to this thread !!

I am working with curl operator. My equation looks like:

And for this equation, I written the matrix structure as,

{
fvVectorMatrix UEqn
(
dummy*(fvm::ddt(U)) // (1) fvm::ddt(U) // ADDITIONAL implicit term - helps in compilation
+ (fvc::curl(U)) // (2)
- ((U^gradU)/(0.4343*(God))) // (3) volScalarField God = f1+(f2*rho*mag(U)); volVectorField gradU(fvc::grad(God));
);
UEqn.solve();
}

I have compiled successfully, but I couldn't able to predict the velocity - shows constant magnitude during post-processing.


At the top, the similar matrix was created and compiled. In this (Below) vector matrix, after compilation - is there any change in vector variables (E and H) ? not sure. I feel that, its because the curl is defined explicitly.

tmp<fvvectormatrix> EEqn
(
mu0*fvm::ddt(E)
==
- fvc::curl(H)
);
EEqn().relax();
EEqn().solve();

As Hrv stated, for implicit curl operators, block matrix implementation seems necessary. Its in Foam-Extend versions based on fully implicit method (FIM).
I haven't worked with it before. Looking ahead !!

If someone have exposed with block matrix method for simple linear equation, as like stated above., kindly share some thoughts and ideas.

Thank you !!