Discussion Forum

I don't think you can do that, at least not directly. Comsol computes the solutions of partial differential equations subject to various constraints, boundary conditions, and initial conditions. Thus, you would need to express your breakdown physics in terms of those kinds of equations.

That said, you can certainly model the electric fields in and on your dielectric, and then, given a breakdown condition (e.g., such that if E > some user-entered threshold value, you would call it a breakdown condition), you could predict whether or not breakdown would be likely in any particular geometry. But that would not be modeling the actual physics of breakdown itself, which is a very complicated phenomenon. In fact, there is no shortage of literature, and ongoing research, in this subject.

I have been thinking about a shortcut by creating a non-linear material with conductivity depending on the magnitude of the E field. The simplest way to do it would be to have a type of step function that raises the conductivity once Enorm goes over a certain value.

To create a material I tried following the example of the "Soft Iron (with losses)" from the Materials Library (within the AC/DC materials. The soft iron with losses calculates the norm of the B and H, based on the three values for Bx, By, and Bz (called B1, B2, and B3 within the Model inputs).

In the material properties I went to the Quantities>Model Inputs, right clicked "Electric Field" and added it. This allows you access to Ex,Ey, and Ez (called E1,E2,and E3). You then calculate normE by sqrt(E1^2+E2^2+E3^2). Then, you can make the Electrical conductivity (sigma) dependent of this normE. However, when I try running a test, I get a buch of errors. Of course it must be some kind of chicken and egg problem. There must however be a way to run this as a parametric or time dependent model in which you start with normE=0 and allow it to increase as the potential is raised.

Any ideas on doing this?

I am also interested in this. Any ideas anyone?

Thanks,
John

Hi,

with a field dependent material as indicated by Jose you may be able to see where a breakdown starts if you already know the breakdown fieldstrength in the material of interest.
But as Robert already pointed out this is far from simulating the breakdown process itself.

Cheers
Edgar

--
Edgar J. Kaiser
emPhys Physical Technology

I would also add that in order to simulate the details of breakdown, this is getting into plasma type physics. This is quite involved as its own thing.

There are several books available for high voltage design work. You could seek those out. There are some books over at:

archive.org/details/texts

that may help. Search for Dielectric Phenomena by FW Peek. It is an older book and its free. It will help with some of the math behind the scenes with simpler structures. It is also a theory book.

I agree that the best you can do in comsol for something like this is to determine the E field strength. It would be up to you to determine if that strength is too much.

OK. This is what I was thinking too. I understand that electric breakdown is an extremely complicated process.

Thanks,
John