Synthesis of vibration analysis and tribology in machine health monitoring

Oil analysis has long been considered to be a supplement to vibration analysis for machine condition monitoring, giving information about wear mechanisms and the type, shape, size, and shedding rate of wear particles. However, it has only recently been realised that tribological considerations can be used in symbiosis with vibration analysis to greatly improve prognostics. There are three major types of fault development in machines for which there exist prediction equations, abrasive wear, fatigue pitting and single crack development. The first two produce wear particles with distinctive properties allowing them to be distinguished by wear debris analysis, while the third, typified by tooth root cracks in gears, do not necessarily produce many wear particles, but their growth can be modelled by the methods of fatigue analysis and fracture mechanics. However, the prediction equations require a knowledge of parameters, such as surface profiles, loads and roughness, sliding velocities etc., which change with time and require updating. Vibration analysis can not only identify which individual components are failing, but also their current state of deterioration. By comparison of simulated and measured vibrations it is often possible to update not only the dynamic equations to estimate changing internal forces, but also the changing parameters of the tribological equations required for the prediction of wear and other fault development. This paper illustrates the advances made by our group in applying these principles to prognostics of IC engine bearings, gears, and rolling element bearings.