Health Monitoring for Flywheel Rotors Supported By Active Magnetic Bearings

Abstract

Magnetic bearings are not a new technology in themselves, yet the control and implementation of such devices is still a budding science. Magnetic bearings are particularly attractive for some applications because of the low energy loss associated with their frictionless operation. By using electromagnetic forces, a rotor can be levitated without mechanical contact between the rotor and its supports. One major application of magnetic bearings is mechanical energy storage devices using flywheels, which potentially have a substantially higher energy storage density than more standard devices such as chemical batteries. Flywheels equipped with magnetic bearings can not only be used to store energy, but can also provide actuation for attitude control in satellite applications. The conceptual designs for such flywheel systems generally employ a disk consisting of a hub (either metal or composite) and a high strength composite rim spinning at high rotation speeds, 50kRPM or higher. This causes substantial stresses to be applied to the rim and hub. In addition, as energy is added or withdrawn from the system, the rotor speed changes over a wide range, resulting in cyclic stresses on the disk and possible fatigue induced cracks. The initiation and growth of such cracks has potentially disastrous implications, possibly causing the entire structure to be destroyed. Accordingly, health monitoring is of critical importance in maintaining the integrity of such devices. In this thesis, a health monitoring strategy based upon the acquisition and analysis of vibration measurements is described and evaluated. A common technique in this regard is to track changes in the synchronous vibration due to imbalance. However, such an approach must consider the controller strategy used with the magnetic bearings. Herein, a simulation model is developed that consists of a flywheel system supported by magnetic bearings, which are controlled using an adaptive strategy that suppresses synchronous vibration. The interaction between the rotor vibration and the controller responses are evaluated in order to provide insight into indicators of crack initiation and growth. The results and conclusions are also validated using an experimental test rig. Some insights and guidelines as to appropriate strategies for crack detection in rotor systems interacting with active bearing controllers are presented and discussed.