A Hybrid Strategy for the Design and Optimization of Coaxial Magnetic Gears

Magnetic gear transmissions are promising alternatives to mechanical ones due to their contactless power transfer, reduced acoustic noise and vibration, inherent overload protection, and improved reliability. However, their design requires fast but accurate tools. While three-dimensional finite-element models offer good accuracy, their complexity hinders their use for design purposes. Two-dimensional representations, on the other hand, tend to overestimate performance due to the lack of end effects in the axial direction. This paper proposes a hybrid design and optimization framework for coaxial magnetic gears that couples a two-dimensional optimizer based on a genetic algorithm with a three-dimensional parametric model. The former model helps identify promising combinations in the design variable space. Then, specific selections are refined through the three-dimensional model. Numerical results show that both approaches exhibit consistent parameter trends, with a resulting prototype yielding a torque density of 213 Nm/L in an envelope contained within 90 mm of diameter and 16.57 mm of active length.