Optimal control of reduced-order finite element models of rotor-bearing-support systems

The need of rotating machines to operate at higher speeds requires better techniques of vibration control. This paper presents how optimal control is applied to rotor-bearing-support systems in which the rotor finite element models have relatively large degrees of freedom (DOF). To conduct the control design for such rotor finite element models with large DOF is challenging and expensive. At this point, the order reduction of the model has its role. In this work, two types of rotor-bearing-support system were used: one with tilting pad journal bearings and the other one with plain full journal bearings which is typically less stable. For cost consideration, the large DOF rotor was reduced into smaller DOF. The quality of the model reduction was evaluated by comparing some first natural frequencies, modal damping ratios, critical speeds, and response of both the full system and the reduced system, along the frequency range of interest. To control vibration, linear quadratic regulator (LQR) control technique was used. From the closed-loop responses, it is shown that the LQR controller suppresses the resonance quite well. In this case, two different pairs of weighting matrices were used. It is shown that the first pair is better at lower speeds, whereas the second pair is better at higher speeds.