I was wandering if anybody had any luck using Modulus to solve simple electro(magneto) dynamics problems. I have tried applying Modulus to a simple set-up of an electro-magnet (c-shaped iron core with copper windings on one side through which current j is applied all in 2D). In this setting Maxwell equations can be reduced to a Poisson equation with the current j being a “force” term. The trick is that the difference in magnetic permeability material properties between iron and air have to be encoded as a Neumann boundary condition to the border between air and iron. Of course we can add those to our loss, however I’m having problems getting the correct solution out. I think it is due to the fact that the magnetic permeability of iron is about 7000 more than that of air, so the boundary condition wants to enforce a normal gradient of the function to be about 7000 more steep on the iron side than on the air side and it fails.
Anyway I was just curious if anybody had any luck using Modulus for electromagnetics (special domain, not frequency domain)?
Hello @jaroslaw.rz, The only thing we have tried related to this is solving some Maxwell-Vlasov equations. We were able to get pretty good results with this but have not released them. We have solved some coupled fluid solid thermal problems though. In these cases we are similarly solving the Poisson equation. For the boundary conditions between the two we have a Neumann for the flux and Dirichlet for temp. The ratios of permeability are going to be tough to deal with though. In our thermal problems the ratios of diffusivity are on the order of 3-6 magnitudes. The only wave we have managed to handle these is to use a complex iterative method call hFTB ((PDF) A NOVEL METHOD FOR THE COMPUTATION OF CONJUGATE HEAT TRANSFER WITH COUPLED SOLVERS). We were using this for the limerock example where the heat sink material is copper and the fluid is air.