Conference Agenda

The modern high-frequency drive of induction motors made eddy-current field analyses time-consuming. The computation time can be drastically reduced using the model order reduction (MOR) methods, which need to incorporate nonlinear characteristics of magnetic saturation for an application to induction motors. We established the nonlinear MOR of induction motors by parameterizing the multi-port Cauer Ladder Network (CLN). The parameterized multi-port CLN was applied to the transient analysis of a rotating induction motor.

From solutions of Finite Element (FE) simulations depending on parameters, a surrogate odel of the FE solution can be build based on the Proper Orthogonal Decomposition (POD) combined with the Radial Basis Functions (RBF) interpolation. Then, a fast approximation of a FE solution can be computed for any parameter values. In order to optimize the number of FE solutions used to define the surrogate model, a hierarchical multilevel approach based on an iterative algorithm is developed. In this communication, a hierarchical multilevel (H-POD-RBF) approach is developed for a nonlinear magnetostatic problem and is applied in the case of a 3D three-phase inductance.

We present a numerical method to solve generic problems in electromagnetic modelling (from statics to high frequency problems) based on approximating fields with spline functions, in the framework of Isogeometric Analysis (IGA). The recurrence relations involved in derivatives of basis spline functions allow the discretisation of differential operators as very sparse incidence matrices. Contrary to other methods with the same property, this holds for spaces with arbitrarily high approximation properties. Additionally the so-called dual differential operators need not be built on a different mesh if the appropriate spline spaces are chosen. The main challenge lies in to the construction of appropriate discrete constitutive equations (discrete Hodge–Star operators).

In this article, three different 3-D eddy current formulations are derived based on the magnetoquasistatic model: a b-conforming, an h-conforming and an eh-conforming formulation. These three formulations are compared, in high-temperature superconductor simulations, based on their effectiveness and robustness with nonlinear conducting and nonlinear magnetic materials.

The striking growth of powerful computing resources allows time-efficient solution of computationally demanding problems. In particular, advances in high-performance computing have made stochastic approaches to real-world applications more practical. The belief propagation algorithm is a probabilistic method typically used in information theory and artificial intelligence. This paper exploits the probabilistic message passing attribute of belief propagation for solving stochastic linear systems that naturally arise from finite element formulation of stochastic partial differential equations, establishing

an explicit connection between the two fields for the first time. The accuracy of the algorithm is validated by comparison to the well-known Mont Carlo method.