In industry, the operation of turbomachinery supported by active magnetic bearings (AMBs) requires a robust and simple controller to meet increased performance requirements and stringent regulations. This study proposes an innovative low-order Multiple-Input Multiple-Output (MIMO) controller. Its structure is derived from the decentralized augmented Proportional-Integral-Derivative (PID) controller, enhanced with fixed terms in the skew-diagonals of the controller matrix. The resulting controller couples the information acquired by the AMB sensors on the same control axis to enhance the overall performance. The parameters of the new MIMO structure are tuned using a model-based procedure that exploits a non-smooth optimization algorithm, with the rotor model adjusted on experimental measurements to represent the real dynamics of the system. The novel controller performance is evaluated through two case studies: first, an expander-compressor system; second, a centrifugal turbo compressor designed for oil and gas applications, facing challenges associated with the observability and controllability of the second bending mode. The performance is then compared with that obtained using a decentralized augmented PID controller whose parameters have been tuned with the same non-smooth optimization algorithm. The new controller demonstrates a significant reduction in vibration caused by rotor bending modes. For example, in the analyzed case studies, unbalance responses were reduced by up to a factor of 10. Moreover, the proposed controller increases robustness compared to the decentralized case, as demonstrated by the second case study where controllability and observability issues are addressed.
Read full abstract