[Spiro]

by the discretisation of each problem?
I need to make some mathematics about refining/rescaling the mesh. Can anyone help me?
thanks in advance

[chegdan]

Originally Posted by Spiro

by the discretisation of each problem?
I need to make some mathematics about refining/rescaling the mesh. Can anyone help me?
thanks in advance

For refining the mesh, what about start looking at the code for the utility refineMesh? You might also look at the createMesh.H file or even the functionality of the interFoam solver. I know that there was (or I think there was) some functionality that refined the mesh at the interface. And , you could always look into the moving mesh source. I'm not an expert, but I would start there if I were trying to solve the problem. Good Luck.

Dan

[Spiro]

my problem (ok ONE of my problems) is that i can FORTRAN but I m the absolute dummy in C. I just want to create coef. matrices for one single problem (with different scalings) and then feed my own fortran program with these matrices for further investigation. Is this possible? Are these matrices not stored at all after an openfoam run?

[chegdan]

Originally Posted by Spiro

my problem (ok ONE of my problems) is that i can FORTRAN but I m the absolute dummy in C. I just want to create coef. matrices for one single problem (with different scalings) and then feed my own fortran program with these matrices for further investigation. Is this possible? Are these matrices not stored at all after an openfoam run?

Sure. I have successfully passed the ldu coefficient matrix (along with x and source b) that OpenFOAM defines to a CUDA linear system solver. One needs to define an extern function, have a container that both languages understand, and pass the info. Its easier for CUDA, since it has compatibility with C++. Im not sure about FORTRAN. There are numerous posts on the forum about connecting some fortran code to OF. There are also numerous posts on the structure of the ldu matrix.

Dan