Discussion Forum

Hello everyboy,

I’m a newbie in the COMSOL world.
I’m having some issue in the simulation of a slight variation of the library model “Eddy current”. I’ve changed the operative frequency from 50Hz to 10GHz and changed the dimension scale from millimeters to nanometers, no problems while computing. The issue occurs when I try to simulate in the same conditions but with a dielectric layer (made of SiO2) between the source and the piece of metal and in contact with the latter. Since I haven’t changed any study parameter, it takes the whole 10000 iterations to compute and show an error message which says:

“
Failed to find a solution.
Divergence of the linear iterations.
There was an error message from the linear solver.
Very ill-conditioned preconditioner.
The relative residual is more than 1000 times larger than the relative tolerance.
Returned solution is not converged.
- Feature: Stationary Solver 1 (sol2/s1)
“

Solution 1 is disabled.

I’ve applied no BC to the dielectric layer, mostly because in the Magnetic Fields “package” there are ones for metallic layer and some others are “not applicable”. I tried to apply some of them, the computation converged, but "Magnetic Insulation" or "PMC" BCs don't seem to be physically suitable for the solution of my issue.. By trying these two I figured out that the problem may be a missing BC on the Dielectric boundaries.
Can you tell me why is this happening? Should I try to find suitable Boundary Conditions in other “packages”? Can the problem be related to an enormous Skin Depth in SiO2? I encounter the same issue even if I change the medium of the cylinder from air to insulator.. Can you please suggest me something, some solution?
I really thank you for your time,

Pietro