Monitoring and Control of False Brinelling

Résumé

False brinelling is a bearing defect type in railway systems that causes excessive noise and vibration. It occurs because of exposure of the bearing to vibration while it is stationary during the process of transportation of a new train by road or sea. These vibrations squeeze the lubricant from between the raceway and the rolling element, causing direct metal-to-metal contact. The resultant fretting causes wear damage on the surfaces of the raceways and rolling elements that, if left unchecked, will cause fatigue and bearing failure in service. Reliable prediction of false brinelling is yet to be achieved. The aim of this research is to develop a tool to investigate the conditions under which false brinelling occurs. This tool was developed first by establishing a mathematical model for predicting the occurrence and growth of the false brinelling phenomenon that was based on the underlying contact mechanics and wear . Then, validation and tuning experiments were conducted to validate the accuracy of the model. Finally, the tool for optimising the mitigation of false brinelling was developed. In the first objective of this project, an analytical model for predicting the false brinelling mechanism of the bearing was developed. Bearing contact mechanics and the wear on each rolling element were modelled based on energy dissipation theory, in particular Mindlin’s [1] and Fouvry’s [2] theories. In the model, the local volume and depth of the wear were calculated based on the occurrence conditions. The model also predicts wear under lateral and axial vibrations and in combination. The tangential input can be either displacement or force. In the second objective, the friction coefficient and wear coefficient were experimentally determined using a Tribolab and Profilometer (Talysurf) under different circumstances of load, contact pressure, sliding displacement and sliding frequencies. The experiments were designed using the Taguchi method. The experiments were conducted for 10,000, 20,000, 30,000, 40,000 and 50,000 cycles. The friction and wear coefficients were modelled and used to tune the analytical model for predicting false brinelling. The threshold energy for wear for high-speed steel was investigated during these experiments. Also, a test-rig was built to simulate the transportation conditions, in which parameters obtained from the field trial were used. This test-rig was used to validate the analytical model for both lateral and axial directions. The third objective introduced a tool to minimise the occurrence of false brinelling. In this tool, the radial load, friction coefficient, radial clearance and stiffness of the mounting plate of the train during transportation were investigated, which led to recommended changes to the system and layout of the train and shipping environments to appropriately constrain the bearings. Overall, the most important contribution of this thesis is the development of an analytical model for predicting false brinelling. This model predicts the wear depth and volume under the lateral and axial vibrations and their combination. Also, the model is around 10,000 times faster than the existing FE model. The second contribution of this thesis is experimental validation and tuning of the analytical model. Using a novel test-rig, experiments under lateral and axial oscillations were conducted. Also, sets of experiments investigated the behaviour of the friction, wear coefficient and threshold energy for wear activation under varying false brinelling occurrence conditions, which helped to tune the analytical model. The third contribution of this thesis is an optimisation tool that uses the analytical model to minimise false brinelling. The tool makes it possible to optimise the wear depth under the combination force (lateral and axial) and to mininise the transmission of the vibration during transportation of the train using the stiffness of the mounting plate of the train.