Skip to content

Wrong wall distance calculation for ACMI cases

Summary

When running cases with ACMI patches, the wall distance field is not correctly calculated by the methods meshWave and exactDistance, where the ACMI patch is interpreted as a solid wall and the wall distance is set accordingly (i.e. close to zero). In contrast, the wall distance method Poisson seems to work correctly for such cases. This can become a problem for routines which require the wall distance field, e.g. certain turbulence models such as kOmegaSST or SpalartAllmaras. An artificial damping of the turbulent eddy viscosity field nut can then be seen near the ACMI patch as a consequence (strength of damping is somewhat case-dependent).

Steps to reproduce

Take the standard OpenFOAM tutorial oscillatingInletACMI2D, switch to the kOmegaSST model and write out the wall distance field for selected time steps.

Example case

I prepared a modified version of the standard OpenFOAM tutorial oscillatingInletACMI2D to reproduce the problem: oscillatingInletACMI2D_SST_wallDist_OF-v2312.tar.gz

The setup is for OpenFOAM-v2312, but we have observed this behaviour also in OpenFOAM-v2212. The method to calculate the wall distance is selected in system/fvSchemes. A coded function object declared in system/controlDict is used to write out the wall distance field for each write time.

What is the current bug behaviour?

The wall distance field is incorrectly calculated for the method meshWave: fig_oscillatingInletACMI2D_wallDist_meshWave

The wall distance field is incorrectly calculated for the method exactDistance: fig_oscillatingInletACMI2D_wallDist_exactDistance

The method Poisson produces an accurate wall distance field (dependent on how well the Laplacian equation converges): fig_oscillatingInletACMI2D_wallDist_Poisson

I have not tested the advectionDiffusion method yet, but I would assume that it works as well.

What is the expected correct behavior?

The wall distance field should be calculated correctly for all wall distance methods.

Relevant logs and/or images

See section What is the current bug behaviour?

Environment information

  • OpenFOAM version : v2312
  • Operating system : openSUSE
  • Hardware info : not relevant
  • Compiler : gcc 7 & 11

Possible fixes

Unfortunately, I haven't had the time yet to dig deeper into the code to come up with a proposal for a fix.

bug

Edited by Marian Fuchs