Hi,

I used my own RK4 coeff., but these are not fixed. Have a look at the book from Blazek for example, there the combination of RK4 coeff. and CFL-numbers is described in detail (Table 6.1 and 6.2).

CgnsToFoam will fail for sure for the SKV. Because the SKV has just a 2D mesh, but you'll need to extend this in circumferential direction. There is a little bit of hacking required in order to do this. Maybe I can get my old python scripts next week and maybe I find the time to write some kind of HowTo. So it is basically working, but not out of the box.

Oliver

Hi Oliver,

Yes, for now I am using Blazek coeffs and CFL.

As for the SKV, I started to dig into pyCGNS and cgnsTools. It will take me a couple of weeks so if you have some python scripts it would be very helpfull.

Antonio

Hi,

you asked some times ago about the SRFVelocity patch. As this one should live in libfiniteVolume.so, I think you can just use it in the solvers. But please test it and report it here.

Converting the Throughflow result into OpenFOAM®:

1.) Convert cgns to vtmb with the script in the src directory of the throughflow code: scm2vtmb.py
2.) Expand the 2D vtmb mesh in circumferential direction with the expandScmMesh class from the TurboVTK project sourceforge.net/p/turbopost/turbovtk, there is also an example at the tutorials section: expandScmMesh_test.py
3.) Convert the 3D VTK mesh with CandyStore (sourceforge.net/projects/candystore/) in OpenFOAM® format. After importing the vtmb or vtu file, I think you have to type somthing like "send all foam" in order to do the export. But I can't remember the correct command at the moment.
4.) Resample the raw output files to an OpenFOAM® project. Maybe you have to rename some of the files.
5.) optional: You want to use transformPoints in order to have your point coordinates in Meter.
6.) The RotatingVelocity is the RelativeVelocity in cylindrical coordinates. The Velocity field is the AbsoluteVelocity in cylindrical coordinates (x=r,y=u,z=z). You want to calculate the cartesian components of at least of one these velocity fields with funkySetFields and name it U or URel.
7.) Use mapFields to map the "2D" result to your mesh.

Please test it,
Oliver

Hi everyone,

concerning the initial solutions there is a bit of python code I would like to share.
Short version: Basically it uses a numpy array of data from the S2m plane and puts it into a funkySetFieldsDict.
Longer version: Using field data in a structured S2m grid, the code uses linear interpolation to create a "center node" for each cell of the structured grid. This way each cell can be divided into 4 triangles, which enables easy barycentric interpolation for any point within that triangle. All the interpolation formulas and "point inside triangle checks" are jammed into a funkySetFieldsDict.
This approach worked for my purposes on a 2x2 cells S2m numpy grid.
I've attached the code with a working example.