Conference Agenda

Optimizing electrical systems represented by ODE and events, using their frequency spectrum is an important issue for designers. This paper presents a methodology to answer to this issue. Using gradient-based optimization algorithm, the paper proposes to simulate the electrical system according time, and then to compute its frequency spectrum. To optimize it by SQP, Automatic Differentiation is mainly used to compute the model gradients.

Magnetic gears can be considered as possible power transmission systems, in substitution of classical mechanical transmissions. They are able to transmit torque, between two mechanical axes, in a contactless way, through the interaction of two coaxial permanent magnets rotors with a set of ferromagnetic poles. The performance of magnetic gears depends on several geometric and material parameters and their sizing can be approached by optimization.
In this work a multi-objective optimization procedure is used to compute the minimum volume encumbrance to meet a value of the transmitted torque. The optimization is based on a constrained single objective approach and carried out by a deterministic optimization strategy. The analysis of the magnetic structure is performed by two dimensional magnetostatic nonlinear magnetic field analysis that is used to evaluate the maximum value of transmitted torque. Optimization loop is managed by a deterministic constrained technique working on the thickness values of the active and passive parts of the two rotor structures.
Results obtained allow to size the device, finding the minimal radial and axial encumbrance of the gear needed to obtain a given value of transmitted torque.

We present a design method for a 3D printed induction heating coil that is optimized by parameterizing the coil path and cross-section. The objective functions of the optimization are the temperature rises in the heated workpiece in the angular and longitudinal directions. The coil path optimization ensures homogeneous heating in the angular direction, while the cross-section optimization enables local heating in the longitudinal direction. We combine the optimized parameters and verify that the final shape can be printed with 3D printers.

We propose an innovative technique for dealing with optimal shape design problems characterised by curved boundaries between ferromagnetic and dielectric sub-regions. The proposed approach relies on the ability of the Virtual Element Method in handling meshes with polygonal elements having curved edges. The well-known TEAM 25 benchmark problem is considered as case study.