Induction machine performance improvements: Design-oriented approaches

Special-purpose induction machines have not enjoyed the same attention in inverter-fed applications as permanent-magnet and switched-reluctance machines, due to of their low power density and low efficiency. Their performance capabilities are limited because they need to satisfy standards for general-purpose machines. Fully integrated computer-aided design (CAD) tools for special-purpose machines that overcome these limitations are required, featuring machine models with reasonable accuracy and small computational effort.; Three different modeling approaches for CAD have been used in the past. Analytical machine models, most common, have low computational effort and complexity but are inaccurate outside certain design restrictions. FEA has, in principle, respectable accuracy, but it requires a massive computational effort. Magnetic equivalent circuits (MEC) provide a compromise between analytical models and FEA. They have reasonable accuracy combined with moderate computational effort, are flexible in size, and are easily parameterized and extended into 3-D models. This dissertation provides a feasibility study for a general 3-D MEC modeling approach intended for design. Theory, permeance network generation, modeling of motion, and force calculation based on the Maxwell Stress Tensor method are discussed.; To verify the modeling approach, inductor current waveforms and power loss are estimated with simple 2-D MEC models and compared to measurements and FEA. Then, the force on an electromagnet is calculated with 2-D and 3-D MEC models and compared to measurements, 2-D and 3-D FEA, and analytical models. All of the models provide adequate accuracy in nonsaturated conditions, but are inaccurate when saturation occurs (except the 3-D FEA model).; Finally, an induction machine model featuring a variable permeance element mesh is introduced. Results for a 500 W motor are compared to measurements, analytical models, and FEA. This machine has limited nominal torque capabilities due to saturated stator and rotor teeth, discovered neither by the design software (analytically based) nor FEA. The 3-D MEC model, about a magnitude smaller in complexity and size than the FEA model, more accurately estimates the limited torque. All the results indicate that MEC, corrected for local saturation, is a promising option for a design tool.