Discussion Forum

Dear community,
i want to calculate the current distribution and the magnetic field distribution of a moving conductor between two conducting disks. I'am interested in the outside magnetic field and the induced eddy currents in the conductive parts.

I'am using the Electric Current Interface(Ground BC and Normal Current Density BC) and the Magnetic Field interface (External Current Density Domain with ec.Jx, ec.Jy, ec.Jz).

I attached some pictures to clarify my needs:
-> contactBolts_ringVarCondu.png shows the contact disks and bolts. A Normal Current BC is applied to the top bolt and a Ground BC is applied to the lower bolt. In between is a segmented ring. The conductivity of the segments will vary over time in that way that the resulting current path will describe a circle between the two contact disks.
-> complete.png shows the whole model with a cylinder jacket around the contact geometry and an surrounding air domain where the outside magnetic field will be measured.
-> eddyCurrentsCylJacket.gif this top view shows the induced eddy currents in the conductive cylinder jacket which are one part i'am interested of.

By now i think this approach will work with real contact geometries since every topic is covered. And this simplified geometry is close to my real geometries. Thanks to Lingling Tang who pointed me to this solution in this thread www.comsol.com/community/forums/general/thread/42099/ .

Now i have two to three questions.

I modeled the 12 segments of the ring by hand and created 12 different materials and assigned them to the segments. The time varying conductivities look like sigma = lb_basisleitw+lb_leitw*rect1(t-3[ms]) [S/m] with different t-x[ms] for the different segments. The first term is to provide, that the conductivity never becomes zero. Since 12 segments for the ring is very coarse i would like to have like 100 of elements that the current moves finer.
Question: Is there a way to create different materials and assign them to the segments in are more programmable style? In that way that this process can be automated. I don't like the idea of create more that 12 materials by hand and assign them to the segments.

First i tried to build this with a moving mesh approach with a single cylindric conductor between the contact disks and let the conductor move over time. The model is also attached. With the Moving Mesh and the Electric Current Interface every thing worked well, i get a nice moving current path between the contact disks. When i add the Magnetic Field interface and introduce there an External Current Density and try to solve the problem with the Time Dependent Solver it never converges.
Question: Does someone have another idea how a moving current can be modeled instead of using variable conductivities? It would be nice if i could use moving domains somehow.

I appreciate any suggestion or push in the right direction.

Thank you very much
Florian Brabetz

Edit: The support pointet out that the Form Assembly functionality should not be used together with the Magnetic Field interface -> "The use of vector elements in 3D Magnetic Fields interface makes such formulation numerically unstable. ".