| Electrical motors are ubiquitous in industries and are very energy intensive. US DoE (Department of Energy) reports that in 2000 electrical motors accounted about 64% of the electricity in the US, of which residential and commercial sector motor applications accounted for 27% and 26%, respectively, representing the biggest single load. In developing nations, such as South Africa, electrical motors consume about 47% of the total electricity. This high motor population represent huge potential for energy and dollar savings through improved motor designs and better operational behaviors. In electrical motors core loss account for a significant portion of the total losses, estimated at 25% in induction machines (with sinusoidal excitations) and even higher for newer designs such as switched reluctance motors (SMRs).; Engineers have had to design around core losses, while material scientists and physicists have studied the core loss phenomenon. The subject is largely left to material scientists, yet, engineers apply these magnetic materials to the design of devices. Consequently, there is a wide knowledge gap between lamination producers and electromagnetic engineers. When this happens, potential efficiency improvements in motors are foregone since motor designers do not fully utilize the material capabilities or there is a material-application mismatch, resulting in sub-optimal designs. In most cases, design engineers resort to using scalars to correct their models, and often these scalars are based on experience.; In this thesis, it is sought to understand the origin of the prevailing core loss models and to study the behavior of loss coefficients with frequency and flux density. This is done by adopting the loss separation principle and considering each component (hysteresis loss, eddy current loss and excess loss) separately. The knowledge of these coefficients leads to a proposal of new way to separate core loss. Hysteresis loss obtained by extrapolation shows conformity to the developed theory in this thesis. Eddy current loss coefficient must be modeled as frequency dependent to account for skin effect at higher frequencies. The excess loss coefficient has been found to be a function of both frequency and flux density. Following the separation is a development of core loss prediction models suitable for implementation in a commercial motor design software package. Physical prediction models presented here are based on measured data on a variety of common laminations used in industry. Attempts to predict core loss with nonsinusoidal excitations such PWM supplies are made. |
