Search

The Partial Element Equivalent Circuit (PEEC) method is promising numerical technique for three-dimension electromagnetic modeling across various application fields. In the framework of the PEEC method, the partial elements modeling the magnetic and electric field coupling between elementary volumes and surfaces are computed by double-folded volume and surface integrals. Assuming the quasi-static hypothesis and an orthogonal mesh, the integrals have been computed by the analytical formulas derived in literature, which significantly reduces the computational time in comparison to the numerical integration. However, the existing analytical formulas are affected by significant numerical errors for certain PEEC structural mesh necessary to model the skin and proximity effects with a higher accuracy. To utilize the full potential of the PEEC method, the calculation of partial elements has to be carefully addressed, which has not been investigated in a comprehensive way so far. Accordingly, this paper presents a systematic accuracy analysis of the existing closed-form analytical formulas and methods for calculating the self and mutual inductances between two rectangular conductors. Additionally, a new strategy to select a proper analytical formula depending on the dimensions and positions of two conductors is proposed, which allows the mutual inductance extraction with a relative error of less than 0.1 % . The new method is systematically validated on examples of 3-D dense PEEC systems using the quadruple precision arithmetic as reference., IEEE Transactions on Electromagnetic Compatibility, 63 (1), ISSN:0018-9375, ISSN:1558-187X

This paper proposes the novel approach employing the kernel method as regression model to describe the dependency of magnet flux linkage on applied current, which is suitable for magnet temperature estimation. The model estimates flux linkage values with mean relative error of less than 2% compared to values calculated from finite element analysis (FEA). Magnet temperature is estimated by comparing a magnet flux linkage under loaded condition to the values from the regression models built under fixed temperatures. Results of the magnet temperature estimation method is about the same accuracy of the results using look-up table (LUT), hence it suggests the approach is suitable for non-linear motor property modeling.

A simulation methodology to reduce computational time of pre-stressed harmonic analysis of radio frequency (RF) microresonators is demonstrated. The methodology is based on the application of model order reduction to a system of ordinary differential equations obtained after spatial discretization by finite element software. Model order reduction produces a low dimensional approximation of the original system and hence enables a substantial reduction of simulation time while maintaining a very small approximation error. The approach allows to perform rapid device design and optimization. Once the device design and working conditions have been defined its reduced model can also be used to implement a behavioural model that can be employed in system level simulations.