Radial Active Magnetic Bearing Design Optimization

This paper presents a design optimization approach to minimize the volume of a radial Active Magnetic Bearing (AMB) by comparing Genetic Algorithm (GA) and Pattern Search (PS) methods. The flexible rotor dynamic analysis is performed to determine AMBs dynamic load under different unbalance cases. Preliminary design parameters are generated and results are compared with optimization results, showing around 35% reduction in volume. The PS method resulted a bigger diameter but shorter bearing length compared with GA. Nevertheless, GA generated a thicker AMB with reduced external diameter. All designs (PD, PS and GA) satisfied design constraints as determined by rotor bearing dynamics while keeping the same bearing load capacity, also validating the PD methodology as a prototyping alternative to optimization strategies.