Page 1 of 1
Maillage structuré ou non-structuré
Posted: Thu Mar 10, 2016 10:53 am
by gbenefice
Bonjour,
Le code saturne accepte les maillages structurés (plus difficile mais on peut le paramétrer) et non-structurés (utilisation de Salome).
Je voudrais savoir si cela est utile de réaliser un maillage structuré . Est ce que Saturne se sert du maillage structuré ou le transforme-t-il en un maillage non-structuré juste avant le calcul ?
Merci
bonne journée
Guillaume Benefice
Re: Maillage structuré ou non-structuré
Posted: Thu Mar 10, 2016 11:48 am
by Yvan Fournier
Hello,
Code_Saturne transforms unstructured meshes in structured meshes at computation start.
The only advantage of a structured mesh for the code might be a better convergence and more precise results for a given refinement. Even this is not always true, depending on how regular or "warped" the mesh may be.
Regards,
Yvan
Re: Maillage structuré ou non-structuré
Posted: Thu Mar 10, 2016 12:00 pm
by gbenefice
Thanks you for your answer
Another question: Salome can create a structured mesh and export this mesh ? The "hexahedron (i,j,k)" function can create a structured mesh but I don't export my mesh to visualize with paraview structured function (Extract subset function).
If the anwser is not, I will try OpenFOAM to mesh my structure.
Regards,
Guillaume
Re: Maillage structuré ou non-structuré
Posted: Thu Mar 10, 2016 3:31 pm
by Yvan Fournier
Hello,
Yes, Salome's "hexahedron (i,j,k)"" builds a mesh which is exported as unstructured, so you can use that for Code_Saturne (we use it very frequently).
Regards,
Yvan
Re: Maillage structuré ou non-structuré
Posted: Fri Mar 02, 2018 10:43 am
by valentin.derrien
Hello,
We are learning how to use Salome and Code_Saturne on simple geometries. We managed to run a convergent simulation using a unstructured mesh (Netgen 3D algorithm on Salome). Then we tried to run another simulation using a structured mesh (Body Fitting with constant spacing on Salome) but we face a problem during the checkmesh operation.
We are wondering if Code_Saturne can handle such meshes or if we did a mistake during the meshing process.
Please find attached the processor.log file of the structured simulation.
Thanks in advance for your answers.
Valentin
Re: Maillage structuré ou non-structuré
Posted: Fri Mar 02, 2018 1:59 pm
by Yvan Fournier
Hello,
The important message is at the end of the preprocessor log... CGNS NFACE_n elements not handled yet.
Those type of element representations were quite rare in CGNS files until a year ago (and that addition to the CGNS spec is not well though out). I already have a few examples of files using such elements, and hope to add support for those this year.
If you built the mesh with SALOME, you should export it to the MED format, so you will avoid this issue.
Best regards,
Yvan
Re: Maillage structuré ou non-structuré
Posted: Thu Mar 15, 2018 10:28 am
by valentin.derrien
Hello,
Thank you for your answer. Indeed we use SALOME to create our meshes, we tried to export them into med files but we still face problems.
By lurking through the Documentation part of Saturne website we have noticed that people often use .des meshes. Which software should we use to open and/or create those files ?
Regards,
Valentin
Re: Maillage structuré ou non-structuré
Posted: Thu Mar 15, 2018 6:56 pm
by Yvan Fournier
Hello,
.des files are generated by Simail, a very old tool. Which problems do you have with .med files ? Those should work best, so it is important to know what type of issues you have.
Regards,
Yvan