[louisgag]

Hi all,

I'm working on a mesh generated with pointwise, but I am excluding the conversion process from my suspicion because I get the same nonOrthogonality values when I examine the mesh directly in pointwise.

My question is: can someone enlighten me as to why on the sharp corner of my 2D mesh shown in the figures is my nonOrtho so high? I read the code cellQuality.C and it seems clear that the intended "angle between cell centers vector and shared face normal" is respected by the projection done there. I'm using OpenFOAM-v1812, but I get the same values with other versions as well...


If I look at my first vertical cell on the right boundary layer cell, it has a nonOrtho of roughly 60°, but none of the "nonOrtho" angles with its 3 neighbors, from visual inspection, seem to go beyond 20°... What I am not understanding correctly?


I initially suspected numerical accuracy (the cell size on the wall is approx. 1e-6*1e-3), but my write format is double precision (16 digits) (and binary), and pointwise does report the same nonOrtho values before any exporting. I also modified cellQuality.C to make both concerned vectors unit-vectors before doing the dot product and taking the inverse sine, but the result stays the same.


The figures are 1. corner with cell centers highlighted; 2: zoom on the corner where nonOrtho goes to roughly 60°; 3: zoom on the face below the corner cell; 4: zoom on the center of the corner cell.



There must be a really simple explanation, can someone open my eyes?


Kind regards,

[louisgag]

Short update:


I did not find the explanation for the apparent discrepancy between displayed mesh and the nonOrthogonality calculation but I did fix it by using a different extrusion method in pointwise (which yields an extremely similar mesh in the first points).