Discussion Forum

Hello,
I have some convergence problem in my 3D study. I build a geophisycal magnetotelluric model, a stratified earth (layers at different electrical conductivity ). I will measure the impadance at the earth surface.
I use the Magnetic Field PhisycI impose a magnetic field for example along y direction on the top of a air layer up to the stratified earth. Then I impose Perfect Magnetic Conductor condition at the lateral boundaries parallel to the x direction, and impose Magnetic Insulation Condition at the bottom of the model and at the lateral boundaries parallel to the y direction.
I use a finer mesh near the evaluation points and a coarse far from the evaluation points. Near the evaluation point mesh edge are about 1/3 the skin depth of the higher frequencies used. For example for 100Hz like higher frequencies and 0.01 ohm*m (conductivity of the upper layer of earth), the mesh edges are about 200 m near the evaluation points.
The total lateral dimension of model are about 50 km in the three dimensions.

The problem is that I use a big interval of frequencies for the study, from 10^-2 Hz to 10^2 Hz, and I have problem with convergence. in some cases I couldn't reach 0.05 of relative tolerance , and in other cases i reach only 1 of relative tolerance for convergence!

It is possible that a fine mesh make the study difficult to converge?
I noticed that if I make a larger mesh this problem is not . But a need a very fine mesh for my study!
Thank you ,bye
Patrizia