When finite element method of electromagnetic is used for some inhomogeneous medium, results of medium with different magnetic susceptibilty is very confused. However, when only conductivity of medium is different, results are good.
First tell me, where r u using FEM? because FEM is used for boundary value problems. and as you would know, at the interface of any surface where the dielectric constant has changed, your radiation, losses will increase and this is due to the skin effect. Better, you can use FDTD or MOM, if you haven't decided the boundary yet. it is very specific condition of material having fixed permeability and permittivity. So, kindly, give details of your problem.
Zhang, I believe you're not using standard FEM codes like FEMM. Material inhomogneity is addressed accurately in such codes.
I believe, if you're trying to code your own FEM program, you're probably considering the relative permeability of a material as a simple multiplier. In that case, the results would be erroneous and would not obey Ampere's law.
While solving for magnetic field in presence of non-linear materials the permeability has to be modeled as a function of field intensity H, be it in strong formulations or weak.
Thanks for your answers. My problem is about geophysics. While there is logging tool called induction logging tool to measure conductivities of formation. It is quasi-static magnetic problem. The sources are coil antenna with alternating current. The values in areas which is far away from sources are zeros. Formation is low magnetic susceptibilty. So usually out models do not consider about it and set it to zero , also the dielectric constant is fixed. But now, we find out there maybe some troubles from magnetic susceptility. So ,whether there are some problems with considering about magnetic susceptibity for quasi-static magnetic problem.