Quantification of the accuracy of different approaches to modelling anisotropy of grain oriented electrical steels

Abstract
In order to accurately model the anisotropy of grain-oriented electrical steel it is necessary to consider the variation of magnetisation characteristics at various angles to the rolling direction. This affects the accuracy to which localised loss and flux distributions can be calculated in laminated cores. Many approaches have been attempted including modelling the spacial phase difference between B and H at different angles to the rolling direction. This is a close representation of the manner in which internal magnetisation occurs in practice and it has been shown to produce good agreement between experiment and calculation in relatively simple geometries. In this paper various approaches of the past 5 years are summarised before considering the particular geometry of a 3 phase transformer core in which the flux distribution has been measured experimentally and by the basic and the complex modelling approaches. The accuracy of the two approaches is compared and it is shown that large differences still occur between experimental and computed flux distributions even using the more accurate representation of the magnetisation process. This shows that further development is needed to predict losses sufficiently accurately for machine core designs.