[Jerryfan]

Hi,

I read the code of fvMesh recently. I got confused about one thing. I realized that its class data attributes like CPtr_ (cell centers), CfPtr_ (face centers) and SfPtr_ (face area vectors) are all defined as slicedGeometricField pointers, but magSfPtr_ (cell surface area magnitude) is defined as a GeometricField pointer.

//- Face area vectors
mutable slicedSurfaceVectorField* SfPtr_;

//- Mag face area vectors
mutable surfaceScalarField* magSfPtr_;

//- Cell centres
mutable slicedVolVectorField* CPtr_;

//- Face centres
mutable slicedSurfaceVectorField* CfPtr_;

My point is that since SfPtr_ includes the surface centers of all the cell surfaces (both the internal cell surfaces and boundary cell surfaces), why it is still defined as a slicedGeometricField variable? whereas, magSfPtr_ is defined as a GeometricField variable as expected as it includes all the cell surface areas. In my opinion, slicedGeometricField after all, it's a sliced part of a whole field. Since SfPtr_ already covers all the cell surfaces, what's the purpose of defining it as a slicedGeometricField variable? And it would be a sliced part of what field since it includes the content of the whole field? Then what's the essence of defining it as a sliced GeometricField? The same thing applies also to CPtr_ and CfPtr_. Can anyone explains why?

[Jerryfan]

It seems that I got the answer. It is all because of the way how it handles the slicing for the boundary fields. Below is what I got from the source code.

The complete field is the surface area vector field. The slicing for the boundary is achieved by using the slicedBoundaryField function in the slicedGeometricField constructor. As for why they do it this way, I think it might because the GeometricField class doesn't have a proper constructor so that we can just use a complete field as an argument to generator a GeoemtricField. Though GeometricField class does have a constructor that can take a Field type variable argument, the problem is the field variable used in this constructor must contain the internal field only, the boundary filed should not be included. I think this is the main reason, they have to use slicedGeometricField class here. While for magSfPtr_, it just simply use the magnitude of SfPtr_, which is already a kind of GeometricField. That's why magSfPtr_ is defined as a GeometricField pointer. Any comment is welcomed towards this discussion. Thanks.

SfPtr_ = new slicedSurfaceVectorField
(
IOobject
(
"S",
pointsInstance(),
meshSubDir,
*this,
IOobject::NO_READ,
IOobject::NO_WRITE,
false
),
*this,
dimArea,
faceAreas()
);

slicedGeometricField constructor:

Foam::SlicedGeometricField<Type, PatchField, SlicedPatchField, GeoMesh>::
SlicedGeometricField
(
const IOobject& io,
const Mesh& mesh,
const dimensionSet& ds,
const Field<Type>& completeField,
const bool preserveCouples
)
:
GeometricField<Type, PatchField, GeoMesh>
(
io,
mesh,
ds,
Field<Type>(),
slicedBoundaryField(mesh, completeField, preserveCouples)
)
{
// Set the internalField to the slice of the complete field
UList<Type>:perator=
(
typename Field<Type>::subField(completeField, GeoMesh::size(mesh))
);
correctBoundaryConditions();
}

In the makemagSf function of fvMesh class:

magSfPtr_ = new surfaceScalarField
(
IOobject
(
"magSf",
pointsInstance(),
meshSubDir,
*this,
IOobject::NO_READ,
IOobject::NO_WRITE,
false
),
mag(Sf()) + dimensionedScalar("vs", dimArea, VSMALL)
);