Determination of Magnetic Bearing Power Losses via Parameter Estimation

The most common and accurate magnetic bearing power loss estimation technologies consist of laboratory test rigs which utilize spin-down data or dynamometer data to compute the losses. However, these technologies do not lend themselves to effective determination of core power losses arising in field machines. An on-line method that could be used for existing bearings in field machines would be beneficial to both end-users and equipment manufacturers. In this paper, a method of this kind is presented which requires only temperature data to compute the losses. This method uses a sequential parameter estimation technique to reduce the data. The new method is applied to a high speed experimental apparatus to estimate power losses at various combinations of rotational speed and coil currents. These power losses are due to windage, coil ohmic heating, eddy currents due to ripple from switching amplifiers, and journal rotation in the magnetic field. Results using the new method are presented for two different bearings: a silicon iron design and a cobalt iron design. Combined eddy current and hysteresis losses have been determined for speeds up to 3.1 million DN (mm-rpm) and flux densities just below saturation for each bearing. To the authors' knowledge, this is the first accounting of core loss in cobalt iron at high speeds.