Weight optimisation of a surface mount permanent magnet synchronous motor using genetic algorithms and a combined electromagnetic-thermal co-simulation environment

This paper proposes a new approach for the design optimisation of a naturally ventilated, vertically mounted surface mount permanent magnet synchronous motor (PMSM). This approach is based on a parallel computing (electromagnetic and thermal) environment and non-linear constrained optimisation problem solving using genetic algorithms (GA). The optimisation is simultaneously applied to both the electromagnetic and thermal design parameters rather than optimising them sequentially. It will be shown that the new design approach allows a weight reduction of more that 4% compared to the approach where the electromagnetic and thermal designs are done separately based on state-of-the art typical parameters values.