Development of a computational design tool for the automatic routing of hairpin end-windings

Hairpin windings are growing in popularity because of their enhanced performance characteristics and the highly automatable manufacturing process. Due to their complexity, hairpin end-windings are challenging to geometrically model efficiently. Additionally, for high performance applications, additive manufacturing broadens the feasible design space by lifting traditional geometrical constraints, suggesting the potential for uniquely optimised solutions that were previously unattainable. The dramatic increase in valid candidate solutions requires automation to efficiently generate, determine, and assess the most applicable results, as well as optimise traditionally constrained designs. This paper documents the development of an algorithm that generates end-winding geometry from a winding pattern input considering the constraints of conventional wire bending machines. Additional functionality is implemented to utilise the geometric freedom additive manufacturing provides, such as varying conductor cross-sections. A case study is undertaken which compares the two outputs to a manually generated end-winding design, and initial geometry shows a 20% reduction in the total conductor length and a 6.6 mm reduction in height, implying benefits to both the compactness of the machine and the volumetric power density. Introducing non-uniform shapes facilitated by additive manufacturing is shown to reduce the height by a further 0.8 mm.