Advanced Control of Active Bearings - Modelling, Design and Experiments

Abstract

In all rotating machines relative movements between the stationary parts and the rotating parts imply energy loss and, in many critical cases, vibration problems. This energy loss leads to higher overall energy consumption of the system. Research activities towards the reduction of friction, the enhancement of damping, the extension of operating range and the minimisation of critical vibrations in machine elements are of fundamental importance. The main component to tackle the energy-loss-related problems is the bearing. The area of design of active bearings, while very promising, is still in its early development mainly because of its high complexity and its multiphysics nature. The state-of-the-art models derived from first principles and axioms of mechanics are complex and often subject to significant parameter uncertainties. They are challenging to develop and not easily used for feedback control design. One example is the controllable radial gas bearing, where the lubricant air is injected through controllable injectors to levitate the rotor on an air film. Feedback control of the injection can improve upon the poor damping to reduce the disturbance sensitivity and vibrations near the critical speeds. The feedback control law is preferably designed from a simple model, which captures the dominant dynamics of the machine in the frequency range of interest. This thesis offers two main original contributions in the field of active bearings. First, an experimental technique is proposed for "in situ" identification of low complexity models of the entire rotor-bearing-actuator-sensor system. The approach employs grey-box identification techniques and is easily applied to industrial rotating machinery with controllable bearings. The approach is applied for identification of a linear parameter-varying model of a rotor supported by an active gas bearing. Second, is the application of model-based control techniques for controllable gas bearings. The parameter-varying model is shown to suit the design of classical and modern control including observer and state-feedback, H1, LPV and gain-scheduled H2 control designs to improve upon the dynamic properties of the gas bearing test rig. Experimental results using the control designs show that the controllers can increase the damping significantly. The damping enhancing controllers are shown to extend the range of safe operation by a 70% increase in shaft angular velocity, thereby allowing safe operation in and above the regions of the first and second critical speeds.