Magnetic circuit modelling of transformer core induction – resolution and accuracy

Abstract

As a tool for optimisation of transformer cores, numerical modelling is used to estimate local induction distributions. Compared to finite element method (FEM), the magnetic circuit modelling (MACC) uses much lower amounts of elements which raises questions of resolution and accuracy. This study compares MACC models of different resolutions, two flux paths (2FP) and four paths (4FP). Both are characterised by low computing times. The low resolution of 2FP offers compact information about global flux distributions, as a function of modified core geometry or material. The increased resolution of 4FP favours the assessment of regional concentration of rotational and circular magnetisation. A rough comparison with FEM illustrates that quantitative FEM studies of such mechanisms are impossible without specific post-processing. With respect to the overall accuracy of MACC, it is not determined by the number of elements but on the accuracy of geometrical assumptions and allocated permeability functions. For the transverse direction and overlap regions, the corresponding accuracy of related terms may be close to 10%, due to product tolerance or mechanical stress. Both high meshing densities and low approximation values are of academic nature. The attractiveness of modelling is not given by accuracy, but by revealing complex distributions with good approximation of instantaneous data.