Electromagnetic Modeling of Ferrites Using Shell Elements and Random Grain Structures

We present a novel approach for stochastic finite-element (FE) modeling of the electromagnetic field in ferrites, combining a thin shell model (TSM) for highly permittive grain boundaries with a Voronoi-tessellation-based geometry generation algorithm. The TSM is validated in the case of a periodic grain structure in a linear 2-D time-harmonic case over a frequency range of 10 kHz-1 GHz, and problems related to standard FE discretization are discussed. The TSM is then applied in a stochastic study for simulating the effect of varying grain structure on the effective resistivity, losses, and reactive power densities of a ferrite core.