In this work, the reduced scalar potential is utilized to derive a finite element formulation capable of handling hysteretic nonlinear material laws. The Fréchet derivative is employed to deduce a quasi-Newton scheme in the weak form for solving the nonlinear partial differential equation that describes the magnetostatic field. Methods for evaluating the Jacobian are presented, and their performance is compared on a 3-D domain under uniaxial and rotational excitation. The numerical results demonstrate the necessity of a flexible approximation to overcome the non-uniqueness of the Jacobian at reversal points, which naturally occurs in hysteresis loops. Consequently, an exact or excessively localized evaluation would give rise to difficulties in states of material magnetization characterized by these reversal points.
30 January 2025 in Computers & Mathematics with Applications
Incremental models for magnetic vector hysteresis have been developed in previous works in accordance with basic principles of thermodynamics. In this article, we derive an equivalent representation of the associated hysteresis operator in terms of a co-energy function which is useful for magnetic field computations based on a scalar potential. Using the convex duality, we further define the corresponding energy functional and the associated inverse hysteresis operator which is required for computations based on the vector potential. The equivalence of the two representations with the energy-based hysteresis models proposed in earlier works is demonstrated and numerical results for some typical test problems are presented obtained by finite element simulation of corresponding scalar and vector potential formulations.
In this work, the Simplified Multiscale Model (SMSM) is incorporated into the Energy-Based (EB) quasistatic vector hysteresis model to represent the anhysteretic part of the material behavior. This approach enables the inclusion of effects such as mechanical stress, magnetostriction, material anisotropy, and crystallographic texture. By integrating the anhysteretic model into the EB framework, it becomes possible to account for dissipative effects (in our case domain wall pinning) while utilizing detailed material information. To solve the EB model in conjunction with the SMSM, two approaches are pursued: a numerical optimization of a free energy functional and an explicit approximate variant, known as the Vector Play Model (VPM). Both methods are compared in terms of computational performance, and the differences in results are demonstrated through the simulation of the cross-section of an electric machine. Furthermore, the local as well as global behavior is investigated. It is shown that in an electrical machine configuration, the VPM provides a very satisfactory approximation to both local and global responses, together with a reduced computation time compared to the EB model. In the provided application case, it is shown that a shrink-fitting operation can lead to a 30%-increase of the overall hysteresis losses.
The parameter identification of energy-based hysteresis models is extremely important regarding the accuracy of simulation results of, e.g., electrical machines and transformers. This paper aims to propose an extended version to existing parameter identification techniques in order to correctly depict rotational losses up to high saturation levels. On the basis of measurements obtained by a rotational single sheet tester (RSST), the measured magnetic field strength is split into its reversible and irreversible part. Using this information the parameters of an adapted vector hysteresis model are identified by means of a least squares minimization. The model results are compared to the measurements and show a good agreement concerning the vanishing rotational losses at saturation level and the behaviour of the magnetic flux density.
2 December 2024 in e+i Elektrotechnik und Informationstechnik