View Issue Details

Jump to NotesJump to History
IDProjectCategoryView StatusDate SubmittedLast Update
0004247OpenFOAMBugpublic2025-06-18 14:552025-06-18 14:55
Reporterpeth Assigned To 
PrioritynormalSeverityminorReproducibilityalways
Status newResolutionopen 
PlatformGNU/LinuxOSubuntuOS Version22.04.5 LTS
Product Version12 
Summary0004247: OmegaWallfunction issues on non-moving NCC boundary in a steady-state CHT simulation
DescriptionOmega field blows up in case of parallel run of turbulent k-omega SST simulation using omegaWallfunction on NCC static fluid-solid boundary in a steady-state CHT simulation, while it runs with single core on the same mesh. In other cases, although the simulation is calculated, the omega residual plot looks strange.
    
I share some simulations. All of them are "dummy" simulations, created only to investigate the issue.
Simulation details:
        - Meshes were generated with ANSYS Fluent Meshing and ANSYS Meshing (the latter was used only for the solid part of simA).
        - steady-state simulation
        - k-omega SST RANS model
        - solver mode: SIMPLE
        - atmospheric boundary layer settings at the inlet and at the bottom
        - resolved boundary layer around the target object
        - target object is heated with a given volumetric power source
       
simA:
        - The omega field blows up close to the solid-fluid NCC interface in case of parallel run after a couple of iterations. In this case omega at certain locations has very large values in the cells close to the air-solid boundary, while it has large negative values on the corresponding fluid-solid boundary face, which is nonphysical (see the attached simulation).
        - In case of single core there is no blowup, the simulation seems to be iterating, but the omega residual plot is flat (see the attached simulation).
        - In previous simulations I also experienced that a too rough solid mesh resolution can also cause problems in the air domain too, even if the air mesh resolution and quality is good (I guess through the non-conformal coupled faces (?)). However, I think this issue is not directly related to the described omega blowup, as in the attached simulations the omega field starts to have bad values on the smooth part of the solid surface, where both the air and solid mesh surfaces are planar (see the uploaded pictures at the link in the next section).

simB:
        - Two simulations: conformal and nonconformal
        - This is a simplified case, where the omega blowup does not happen in the parallel case.
        - However the residual plot of the omega field in the nonconformal case looks again strange: it falls to a small value in the first couple of iterations, then it remains constant, similarly as in the other simulation in the single core case. This issue does not happen if the mesh is conformal.
        - The calculated U field look roughly similar in the nonconformal and conformal cases. On the air-solid surface the omega values are different, although this can be caused by the different meshes.
Steps To Reproduce1. Simulations with results are accessible here: https://www.eet.bme.hu/~palovics/simulations/
2. 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).
3. I uploaded the parallel simA simulation without results too (parallel_mesh_and_settings.tar.gz), if smaller download size is needed.
4. I also uploaded two pictures created in ParaView of the omega field blowup in the parallel case to highlight where the problem happens.
TagsNo tags attached.

Activities

There are no notes attached to this issue.

Issue History

Date Modified Username Field Change
2025-06-18 14:55 peth New Issue