Hi allThe problem is that it's hard to obtain analytical solutions for more complicated geometries. For more complicated geometries, the only recourse could be to look at solutions from other codes for comparison. Some finite element vendors have published example problems on the web, e.g http://www.quickfield.com/advanced/big_mesh.htm or http://www.fieldp.com/library/estat50.pdf
I've been putting Bela through some extensive tests (FEMM 4.0 beta) to confirm some results I'm getting with my 3D BEM electrostatic solver and I'm pleased to say I get good agreement. What I'm looking for is example geometries where there are known analytic solutions for the capacitance, other than the standard concentric sphere, coaxial conductor etc. Can anyone suggest a good source of reference? I've done extensive searches with google & I'm just finding lots of the well known ones.
TIA Mark
Well, I haven't yet incorporated any notion of automatic mesh refinement (i.e. multiple solution steps in which the mesh is refined on the basis of a posteriori error estimates, say like Ansoft's Maxwell 2D), but I might at some point. Triangle already has the capability to refine meshes based on some sort of error criteria, so it might not be all that hard to implement. I view this as more important to a 3D code--although you can overwhelm just about any 2D problem with elements due to the current price of computer speed and memory, the same is not yet the case with 3D problems (and probably won't be for some time).p.s. David thanks for the acknowledgement in the BELA user's manual. Bela is equal to Quickfield IMHO.
p.p.s It would be nice to have a little more control (refinement) over the mesh. I think the speed at which the mesh size grows is too fast ? is this not controllable in triangle ?