Effective Use of Magnetization Data in the Design of Electric Machines With Overfluxed Regions

Electric machines often have overfluxed regions that carry large magnetic fields that are beyond the maximum value for which magnetization data are available. These fields are close to magnetic saturation, so extrapolation to saturation is required to accurately map the overfluxed regions. However, the extrapolation methods used in most magnetic field software do not consider the physics of saturation, so they are known to cause significant errors. The simultaneous exponential extrapolation procedure presented herein incorporates the physics of saturation. It essentially converts the extrapolation problem into an interpolation problem by forcing it to pass through a point close to saturation, thereby minimizing errors. This paper also discusses potential hidden noise in the measured data that can cause numerical instability. It outlines a graphical procedure to remove such hidden noise to facilitate faster convergence. These extrapolation to saturation and the hidden-noise elimination procedures facilitate more accurate mapping of the overfluxed regions and allow a better assessment of their impact of such regions on the performance of electric machines.