The Design and Optimization of Additively Manufactured Windings Utilizing Data Driven Algorithms for Minimal Loss in Electric Machines

Permanent magnet (PM) electrical machines are an ever increasingly utilised motor topology for numerous industries such as automotive, aerospace, manufacturing, energy and premium consumer goods. PM electrical machines exhibit high power density, high operating efficiency and high torque to current ratio, whilst remaining robust and fault tolerant. However, the stator windings account for a significant proportion of the overall motor losses. There are many avenues for machine designers to potentially reduce winding losses whilst increasing overall efficiency and performance. Such avenues include, thermal management systems, novel winding materials and novel winding manufacturing methods. Additive manufacturing is generally recognised as a transformative manufacturing technology, especially in the design of electric machines. The ability to create a wide array of geometric shapes offers a level of design freedom that was previously unattainable. Additive manufacturing is therefore utilised in this paper to produce novel, optimised winding designs that have been configured to minimise total machine loss and maximise machine efficiency. This paper investigates the use of an algorithmic optimisation process within Ansys Optislang, and automated using python scripting. The optimisation process consists of sensitivity analysis utilising an efficient hybrid 2D FEA-Analytical model, meta-modelling and genetic algorithm to search the design space for optimal winding designs. The optimal designs are then validated against 2D and 3D FEA high precision motor models within Ansys maxwell and MotorCad and compared against a benchmark winding configuration. It was found that the most optimal winding design produced a motor efficiency of 97%.