CFD Online Discussion Forums - [mesh manipulation] Generating nonuniform meshes with sizefunction

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)

- OpenFOAM Meshing & Mesh Conversion (https://www.cfd-online.com/Forums/openfoam-meshing/)

- - [mesh manipulation] Generating nonuniform meshes with sizefunction (https://www.cfd-online.com/Forums/openfoam-meshing/61812-generating-nonuniform-meshes-sizefunction.html)

I modified the end of my code this way:

Code:

mesh.movePoints(points);



runTime.functionObjects().off();



runTime++;



mesh.write();

And I don't have errors related to fieldAverage anymore. A new time directory is created, but, only points are exported, not the full mesh with boundaries, owners, neighbours, etc... which doesn't make all this very useful.

I tried to understand a little how all this work.

In an fvMesh, the write() method is a reimplementation of the virtual write() inherited from polyMesh, itself inherited from objectRegistry (and itself from regIOobject).

But, when fvMesh::write() is called, the following code is executed

Code:

bool fvMesh::write() const

{

    return polyMesh::write();

}

and thus the regIOobject::write() is called, in the end.

The problem in my case is that the only written file is points.

When the fvMesh mesh is created, with

Code:

Foam::fvMesh mesh

(

    Foam::IOobject

    (

     Foam::fvMesh::defaultRegion,

     runTime.timeName(),

     runTime,

     Foam::IOobject::MUST_READ

    )

);

the IOobject provided is used by the inherited constructor of polyMesh, which creates the faces_, owner_, etc... each with their own IOobject, and registering all them on the fvMesh object. I don't see where is the difference of treatment, and why only the "points" file is exported.

Plus, if I try a

Code:

mesh.names()

I get:

Code:

11

(

cellZones

fvSolution

meshPhi

points

neighbour

faces

faceZones

fvSchemes

pointZones

boundary

owner

)

Which made me expect to find those files in the time directory.

John,

it's all about saving disk space.

When OF write a mesh, it only saves data that has changed after the last dump. On read OF goes backwards through the time dumps and collects the necessary data.

All what you have to do in order to obtain a "normal" mesh is to copy 1/polyMesh/points into constant/polyMesh and delete the "1" time directory.

Henrik

Thank you for the information !

Well rewriting the whole polyMesh forced me to dive a little into OF structure which wasn't a bad thing, after all :)

Something that might be attractive (my 2 cents...) would be to have the same kind of rule for field reading. If I apply a preprocess on the mesh, which outputs to other time directories, the process of field reading would detect and use the latest version so I don't have to manually copy everything on the last time directory.

Hi
I am also performing LES on channel flow and i would like to use ane hyperbolic mesh does someone has some know how to do that how to implement the routine in the code

Hello Alberto,

I am very new to OF and would like to know if it is possible for you to send me your code about "hyperbolic" (or any other analytical function" meshing? Because I would like to implement it on a variable cross-sectional cylinder, with the number of cells depending on the radius, for instance...

Thank you very much!

Best,

Bertrand

Hello guys,
could you please upload final working code...:)

Quote:

Originally Posted by alberto (Post 204647)

Thanks Hrv. It works! :-)

Hello Alberto
I know this is a very old thread. But could you possibly share your code of how to move the internal points of a uniform mesh?

Thank you,

Kaufman