Prediction of losses in electrical steels

The book chapter shows that a `good' model is one which only needs a small amount of reference experimental input magnetisation data and can predict accurately over a wide magnetisation range. It could be argued that, for many applications, it is not necessary for the model to be based on physical processes as long as it can be used effectively to predict the loss of a steel to a sufficient accuracy. The chapter summarises some approaches to loss modelling and shows that they each have their own virtues which can be assessed in terms of speed, complexity, wide applicability and accuracy. Loss separation approaches and hysteresis models are incorporated into some commercial computational electromagnetic software used for electrical machine performance predictions.