I am trying to solve some 2D electrostatic problem (Poisson Equation) using Gripap.jl and gmsh for mesh generation. I solve for the potential in space bounded by conductors, which have surface potentials prescribed as Dirichlet boundary conditions U=1 on the outer boundary and U=0 on the inner electrode.
When the inserted electrode is “small” (only one mesh cell per boundary segment), the solver seems to ignore the potential assigned to the small electrode (U=0) as shown in the figure below.
Apparently, the mesh cell number on each side is important as also illustrated by the figure below, where only the sides with two cells per side are correctly represented.
My source code is available in the following gist:
Is there some solver parameter to fix this issue?
I am using the gmsh boundary markers to assign the boundary conditions. Note that in Fenicsx ( fenicsproject.org ) I can solve on the same mesh without issues.
My software versions are
Thanks in advance for any help.