b'Industrial Machinery & Heavy Equipment | Engineer Innovationfollowing optimization objectives were selected:O : Maximize gravimetric power 1density [kW/kg] O : Maximize efficiency2 O : Minimize torque ripple3We developed scripts that link MATLAB optimization toolbox to Simcenter MAGNET via scripting interface. The optimization algorithm selects a set of geometric dimensions that completely define a machine design. Our scripts then use Simcenter MAGNET to evaluate the machines performance. The multi-objective genetic algorithm requires evaluation of a very large number of motor design candidates, making it essential to quickly evaluate FEA models. AFPM feature 3D flux paths, meaning that their complete physics can only be captured with 3D FEA models. Unfortunately, compared to 2D models, 3D models require significant computation time, which is prohibitive for optimization. So, we adapted a 2D FEA technique for model evaluation. A detailed description of this technique is presented in . As the number of computational slices used in the 2D model increases, the performance obtained using the 2D models tend towards that obtained from a 3D model as shown in figure 4. The number of computational slices used is a compromise between accuracy and evaluation time. A 2D model for a 16 pole, 24 slot single rotor machine is shown in figure 5. Optimization Results:The optimization was run for 150 generations with a population of 75 per generation. The 2D projections of pareto fronts from the final generation (shown in figure 6) indicate a clear trade-off between the gravimetric power density and efficiency. Typical PM motor designs reported in literature for passenger electric vehicles have maximum power density of approximately 3 kW/kg while high performance motors used in aircraft electrification reported up to 10 kW/kg . The optimization results in figure 6 clearly show that the single rotor AFPM machine that we are developing can potentially achieve power density well 45'