Andrew Kasha, PhD Student, Purdue University Scott Sudhoff, Michael and Katherine Birck Professor of Electrical and Computer Engineering, Purdue University Multi-Objective Design Optimization of a Surface-Mounted Heterogeneous-Pole Permanent-Magnet (HPM) Machine SELECT Annual Meeting and Technology Showcase – Logan, Utah – September 27-28, 2016 FLUX DENSITY DISTRIBUTION SHAPING • Shaping air gap flux can potentially: • Reduce cogging torque • Improve torque density • Improve machine efficiency • Can be achieved using: • Halbach magnetization • Magnet segmentation • Mixed magnet grades • Physical pole shaping • The HPM uses mixed grade magnets of disparate shape atop a tiered rotor backiron to achieve FDD shaping THE S YMMETRIC HPM • The SHPM is a subvariant with: • Three magnet segments • Constant magnet depth • Identical outer PM materials • Identical outer magnet widths • Outer magnets of lower grade than inner magnet • No rotor backiron tiers • Constant air gap • Symmetric forward and reverse operation APPLICATION • The SHPM can be used to improve torque density of EV propulsion systems • The topology also has the potential to reduce torque ripple SHPM ANALYTICAL MODEL • Improved stator leakage model directly accounts for flux crossing the slot • Better predicts the leakage flux density at the top and bottom of stator tooth • Improves the predicted flux density waveforms in the stator, leading to better core loss prediction • Nonlinear model converges even under saturation and high current density conditions • Accounts for DC and AC conduction losses as well as steel core losses • Can be used to rapid evaluation of designs SHPM CROSS-SECTION MULTI OBJECTIVE OPT . B ASED DESIGN • Choose design variables which fully describe machine • Establish constraints to ensure performance and viability • Select metrics for comparing individual designs • electromagnetic mass • aggregate loss • Construct fitness function CASE S TUDY • Constant 2.5 HP • 5:1 speed range • Fixed rotor, stator, and conductor material • Aggregate losses evenly weighted at each operating point • Study repeated 5 times T OPOLOGY COMPARISON • Optimization studies for SHPM and SMPM (traditional single surface mounted magnet per pole) were run • SHPM outperforms SMPM in the low mass region • A 10.7% reduction in aggregate losses were observed for low mass designs FEA V ALIDATION • Analytical model validated using 2D FEA in Ansys Maxell • 72 magnetostatic evaluations of machine were performed across a full electrical cycle in 5 degree increments • Less than 4% error between average torque from FEA and analytical model • Flux density waveforms in stator show strong agreement between new analytical model and FEA, even at operating points exhibiting high slot leakage • Maximum average-to-peak torque ripple of 4% 17.86 8.07 3.57 17.29 7.96 3.6 0 2 4 6 8 10 12 14 16 18 20 OP1 OP2 OP3 Torque, Nm FEM 3.30% 0.63% -1.11% FUTURE WORK • Expansion of analytical model to include generalized HPM • Development of 3D thermal equivalent circuit (TEC) to predict temperature increase in stator • Construction of SHPM prototype to validate analytical and TEC models • Redesign SHPM using TEC