Improving Repeatability of Tests for Iron Loss Prediction and Electrical Machines Model Calibration

Prediction of of iron loss in electric machines is known to be a challenging task. The concept of a building factor to account for discrepancy between predicted and measured loss is widely used. This paper describes an extensive experimental campaign on permanent magnet synchronous machines at no-load, with magnetized and dummy rotors. Several identified challenges may be addressed by repeatabilty investigations combined with temperature monitoring of rotor core and bearing temperatures. For accurate prediction of iron loss impacted by manufacturing (cut edge or shrink-fit for example), current models require electrical steel properties (BH and iron loss density) as a function of tensile and compressive stress/strain. Such measurements have not yet been standardized. Repeatability issues with a bespoken single sheet tester were reduced by pre-cycling at three tensile stress levels. Analysis of measured impact of stress for different grades confirmed that a specifc set of data per grade and supplier may be required.