Abstract

Optimal design of electric machines, having complex 3-D magnetic flux path, such as transverse flux or axial flux motors, is computationally demanding owing to the unavailability of a fast 2-D finite-element method (FEM), which leads to the indispensable use of costly 3-D FEM models. With multiobjectives and large design parameter vector, this task is further exacerbated. To accelerate the design procedure, unlike conventional methods, this article proposes the use of two information sources, an expensive but accurate high-fidelity 3-D-FEM model and a fast but inaccurate low-fidelity 2-D-equivalent FEM model to exploit corresponding opposite merits of each and construct a multifidelity surrogate model (response surface) with minimum number of expensive 3-D-FEM simulations. The low-fidelity 2-D model, which intrinsically does not exist for such machines, is generated by ensuring equal magnetic reluctance along main magnetic path despite deformation of model's geometry. The proposed methodology is applied to optimize a novel magnetically levitated transverse flux permanent magnet linear synchronous motor, considering suspension characteristics (lateral and normal stiffnesses) along with thrust as objectives and a seven parameter design vector. Results show significant design time-saving without compromising surrogate accuracy. The optimized design is fabricated and relevant experimental measurements are taken.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call