View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0004271 | OpenFOAM | Bug | public | 2025-09-02 12:20 | 2025-09-03 12:58 |
| Reporter | peth | Assigned To | will | ||
| Priority | normal | Severity | minor | Reproducibility | always |
| Status | closed | Resolution | suspended | ||
| Platform | GNU/Linux | OS | Ubuntu | OS Version | 22.04 |
| Product Version | dev | ||||
| Summary | 0004271: Non-moving NCC projected volume issue | ||||
| Description | This is a continuation of the previous bug report https://bugs.openfoam.org/view.php?id=4247 As suggested in the reply of the previous report, I swapped the owner-neighbour order of the air and solid domains in the createNonConformalCouples dictionary to lower the projected cell volume fraction in the air. Now the air is the neighbour, where a boundary layer mesh is defined. However, the max projected volume fraction using polyhedral mesh is still seems to be problematic (too high). I created new simplified sample simulations. I found the following: - If I use tetrahedral cells in the air and solid with prism cells on the air boundary layer, the max volume fraction in the air is close to zero. In this case the omega residual plot looks normal. - If I use polyhedral cells in the air with the corresponding boundary layer with a similar resolution as in the tetra case, the max volume fraction is usually orders of magnitude higher than in the previous case (I tested this for different geometries). This is strange, as the mesh resolution and boundary layer settings are kept similar. Besides, in the simulation results the omega residual looks flat. I share two simulation cases: - "test_bump": A low cell count case to highlight the problem. The projected max volume fractions in the tetra and polyhedral cases can be checked at log_mesh/createNonConformalCouples.log. No simulations were ran here, this was just created to highlight the problem. - "test_cube": This is a higher cell count case, where the simulations have been run too. The projected max volume fractions can be checked here at the log_mesh/createNonConformalCouples.log. The residual plot can be found at "graphs_residuals.png". In the residual plots, the flat curve for omega happens at the polyhedral case, but it looks normal in the tetra case. All meshes were generated in ANSYS Fluent Meshing. Cell sizings in the polyhedral and tetrahedral cases, as well as the mesh boundary layer settings are kept the same in the poly/tetra cases. | ||||
| Steps To Reproduce | - The described test cases are accessible here: https://www.eet.bme.hu/~palovics/simulations/ - The simulations can be run with the ./Allrun script, while the removing of the previous results can be done with the ./clean script. Residual plot can be generated with the ./postproc_residuals script (it needs awk and gnuplot). | ||||
| Tags | No tags attached. | ||||
|
|
The problem is that your wall faces are not flat. Take a look at the air cell at about (1.53309, -0.726246, 0.666235) in the test_bump_polyhedral case. It is substantially curved, relative to its thickness. If a face is not flat then its surface is not well defined and the surface used to construct the cell volume will not (in general) be the same as that which results from the NCC intersection process. So, the NCC process results in the adjacent cells experiencing a change in volume. This change registers as a non-zero projected volume fraction, even though there is no actual projection of the neighbour. Because the cells in your case are both curved and very thin, this change in the surface geometry registers as a significant error. It's not a problem with the triangles because triangles are flat by definition. You could fix it by triangulating the polygonal faces of the wall patch, and leaving the cells otherwise unchanged. This would remove the projected volume fraction error, but it would also increase the non-orthogonality of the boundary cells, so it's not without a trade-off. To have the code avoid this entirely would require the triangulation used to generate the cell volumes to be made exactly consistent with the triangulation used by NCC to perform the intersection. That's not out of the question, but it would be a significant development effort which would have to be paid for. Please contact CFD Direct if you wish to pursue this. |
