Instability with Riemann BC, possibly coupled with periodics?

[rpuenter]

I am trying to simulate a 2.5D airfoil set-up (though the example I show has a cylinder currently). I am trying to reduce the cell count by having the inlet and outlet relatively close to the airfoil (3 and 5 chords respectively). To prevent reflections I'm trying the Riemann BC. The rest of the patches of the box are periodic in such a way that the upper and lower, and front and back patches are communicated.

I get what seems to be a strange even-odd decoupling at the inlet (see screen grab attached). I thought it would be related to #1090, but changing the front and back periodics to symmetry does not fix it, so I assume it must be a problem with the Riemann.

I attach 2 cfg files (one with a steady state with symmetry planes, used to get the initial solution for the unsteadyc, but it also develops the same instability if I set them periodic ) and the gmsh geo file that generates the mesh.

su2bug.zip

System info:

• OS: Windows 10, Ubuntu wsl2
• C++ compiler and version: g++ (GCC) 9.4.0
• MPI implementation and version: OpenMPI 4.0.3, but I'm running serial
• SU2 Version: 8.0.0

[rpuenter]

I seem to have solved it by reducing the aspect ratio of the inlet cells. It was not very large compared to the maximum in the mesh, but I guess it was enough to cause an instability. So I'd say it not a bug but a robustness issue. Anyway meshes should be good quality, so I guess it's my bad.