Learning Forum

[mskr]

I am doing Fluent Solution Data Export in ASCII and I can get quantities per node.

Now I want to do interpolation in my own code. Therefore I need the tetrahedra of my tet mesh.

How can I get the four nodes for each tetrahedron?

 

[Rob]

Cell ID might help you? Otherwise you'll have to figure out the maths to determine which nodes connect where. 

[mskr]

Thanks for the suggestions.

Cell ID takes floating point values when Location is set to node. I am not sure how to interpret that.

Could you elaborate on the second one? Is there even a unique solution to connecting unstructured points into tetrahedra?

 

This is how my exported file looks like. I have 4945 nodes in my mesh. By exporting cell centers, I can verify that I have 11643 cells.

[mskr]

I tried exporting from CFD Post, where there is the Line and Face Connectivity option, but my file looks exactly like when I don't tick this checkbox, i.e. it seems to have no function at all. Am I missing something?

[Rob]

Not sure about CFD Post as that uses an interpolated Fluent data set back onto the Fluent mesh due to it being node rather than cell based. 

Cell ID makes sense when you export cell centre data, I don't know what it represents for node data as it's not an integer. 

There's probably a unique solution, how you get it is an entirely different issue. You might be better off using the cell centre data and mapping that to your code?

[mskr]

Thanks for the thoughts. I try to export directly from Fluent instead of CFD Post then. But cell centre data is not sufficient as well, because I need all four corners of each tetrahedron to interpolate at arbitrary points. In my code, I first find the tet where the point falls into and then interpolate using barycentric coordinates.

It looks like I cannot do this with the Fluent ASCII format and I need to get ready to parse something more complex. If anyone has experience, I would appreciate the shared knowledge. I hope there were other people doing their own "post processing" before me.

EDIT: A friend told me I could do a delauney triangulation of the cell centres. Using cell centres is better because Fluent solves for those points and not for mesh nodes.