Active Magnetic Bearings technology is becoming more and more a standard in Oil and Gas turbo machinery. Touchdown bearings are an important element of all machines levitated by magnetic bearings. Frequently implemented as hybrid ball bearings, they constitute the only mechanical link between the rotor spinning at high peripheral speed and the stator. Since both sides carry fragile parts, any uncontrolled contact must be avoided.
Extensive testing has been done in industry. When unplanned down-time would have a high system impact or when repair options are limited (upstream equipment, subsea operation), elaborate testing campaigns are common. A complement, if not an alternative, are computer based transient rotor drop simulations. They allow reducing uncertainties and help qualifying designs. In most cases, simulations use rotor dynamics models combined with contact models allowing an estimation of shaft behavior during a drop. We present a new approach for transient rotor drop simulations, where detailed calculation of the interior ball bearing mechanics is at the center of the model. This is possible by using a multiple-bodies simulation code, which was initially developed for pure ball bearing calculations and implements contact models including lubrication, wear, and geometry imperfections. Here, the same code is used to model entire shaft systems, associated with preloaded touchdown bearings. This technique allows not only to understand and to analyze failure patterns that determine bearing lifetime (contact stress, heat, and deformation), but also to take into account interactions of the bearing itself with the rest of the system.
After a feasibility study for validating the capability of simulating rotor drops, three models were realized for three different types of machines of increasing complexity that were all validated by actual drop tests. The last model is a supercritical horizontal shaft system (200kg) representing an actual Oil and Gas application in reduced scale. Simulated shaft behavior was successfully compared to drop test recordings. The output was analyzed for internal bearing mechanics (contact pressures and contact angle, sliding or rolling friction) and could be correlated with evidence found on the bearings from the drop tests.