code_saturne User's Forum

Hi everyone !

I am currently working on fluid-structure interaction for my master's thesis and I am quite new to code_saturne and all the SALOME environment. I understood that it was potentially feasible to couple code_saturne with code_aster in order to do numerical FSI simulations but I am struggling to find what method/interface is the most adequate to do that (and also on which version of both codes is it possible).

Is anyone familiar with this process or could help me to see clearer on that topic ? Maybe a tutorial case would be nice to look at.

Thank you in advance for any suggestion or advice.

Regards,

Léo

Hello Leo,

The easiest way to do a numerical FSI simulation is to chain code_saturne and code_aster.
Both codes can read the same format .med. So here's how you can do this:
1. Generate a mesh as a .med file, using Salome(if possible, hexahedral or with a good quality, as a finite volume solver code_saturne is very sensitive to your mesh).
2. Parametrize your CFD study and compute the results you are interested in. After it converged, code_saturne can also store the physical fields in a .med file. If you need tutorials for the CFD part, you can check the youtube channel of Simvia or of code_saturne. Here is a CFD tutorial I made a few months ago for code_saturne.
3. Now, you want to provide this code_saturne .med output as an input of code_aster. However, these two codes are using different discretisation scheme: code_saturne uses finite volume so the physical fields are stored on the mesh faces. code_aster uses finite element method, so the field data must be on the nodes of the mesh. You must therefore perform a projection face -> Node. Code_aster can do this in many way, either by directly working on the med (not recommanded), or using some code_aster commands. There are many discussion on the code_aster forum talking about this, here's one I found.
4. Then, run the code_aster simulation, and see the med result file code_aster generated. Your fields are now stored at the point but the study and physical interpretation are the same. Note you can also find code_aster tutorial on the youtube channel of Simvia.

It is possible to do more advance types of coupling, but let me know if this first answer helps.
Best,
Louis

Hello,

I have checked with the colleague working on code_saturne/code_aster coupling, and have a few more details.

- Regarding "chained" computations, there is an exemple/test case here: https://codeaster.gitlab.io/doc/docaste ... index.html

- Regarding the actual 2-way coupling, rewritten 2 years ago to now use MEDCoupling instead of YACS (and which should thus be usable on large cases in HPC environments), we still have a few bugs and missing features in MEDCoupling, and the code is partly working but still buggy. A final, validated version is only expected around late 2026 (as this involves multiple people from different departments who are already quite busy with other projects, progress is slow).

Best regards,

Yvan