Shape optimization of PM devices using constrained gradient based inverse problem methodology

Permanent magnets (PMs) are widely used in a variety of industrial equipment and devices. In magnet design, the shape of a PM plays an important role and minimization of the leakage flux improves the performance of the device. The shape optimization of PM devices using gradient based inverse problem methodology (GIPM) is presented. The paper describes for the first time the use of analytical sensitivities for shape optimization of PM devices. Furthermore, the adjoint method of the direct differentiation approach is used for the computation of the sensitivities of the object function. Two case studies are presented. The first involves a magnetic circuit with an air gap and PM excitation, the second is that of a PM pole face. In both cases, design optimization is employed to obtain a desired flux density profile in the air gap with a minimum leakage flux and minimum size of the PM material.