[peresah]

Hi there

Does anyone know if the Star-CCM+ solver accepts a structured (single or multiblock) volume mesh.

I've managed to export an unstructured volume mesh from ICEM CFD into Star-CCM+, but not a structured one. I can't find any clear info on this in the Star-CCM+ documentatation.

Thanks!

[JBeilke]

You can load any mesh into ccm+. You only have to convert it into one of the supported formats. CCM+ then treats all meshes as unstructured (face based).

[peresah]

You seem to agree that Star-CCM+ only supports unstructured grids, right?

[bramv101]

In a way you can say it supports them, however you need to convert...

[peresah]

Ok, thanks for clarifying :-)

[chrism]

You can import single block or multi-block plot3d files, those are structured grids.

[peresah]

Okay, although I have by now found papers stating that Star-CCM+ works on unstructured grids only and ICEM doesn't allow export of structured grids when Star-CCM+ is chosen as solver.

[Subhadeep]

Originally Posted by peresah

Okay, although I have by now found papers stating that Star-CCM+ works on unstructured grids only and ICEM doesn't allow export of structured grids when Star-CCM+ is chosen as solver.

The key is to identify the difference between grid and solver. StarCCM+ is an unstructured solver. So when you have a structured grid and either read in directly into StarCCM+ (Plot3D for example) or ask ICEM (which creates structured grids) to create the StarCCM+ compatible grid they get converted to unstructured format.

BUT physically your grid used is still structured with respect to location of the grid points and the 3D volumes. You are still getting nice cubes of volume from your structured grid. If this was your original goal to have the structured grid for computational domain then stick with it. All that is happening is the format change in data storage. Your structured grid is not being converted to a different unstructured grid.

This is true for all unstructured solvers. Most commercial codes are unstructured solvers. Only in house research codes are structured solvers. There only way to read in is structured grids like Plot3D format.

[peresah]

Hi Subhadeep

Thanks for the discussion. :-) I think we agree over-all on the content, although not on definitions :-).

As far as I know, the conventional definitions for structured/unstructured grids (or solvers) are dependent on the way the data is stored, not directly on spatial distribution:

• Structured grid/solver: Cells are implicitly defined by the node numbering. In one dimension, node numbers 1, 2, 3, 4, 5, 6 could define cells 1, 2, 3, 4, 5.
• Unstructured grid/solver: Cells are explicitly defined by a topology matrix. In the above example, the same cells could be numbered as 1, 5, 6, 2, 3 and a matrix would specify the cell/node connections. I guess, this storage format slows down the algorithm, but allows more complex domains.

Thus, a mesh consisting of pure cubes is not structured per definition, as you say, but could be "formatted" as either structured or unstructured.