Fault-to-Failure Progression Modeling of Propulsion System and Drive Train Bearings for Prognostic and Useful Performance Life Remaining Predictions

Military and commercial aircraft operators are moving toward predictive maintenance to minimize costs while maximizing asset availability and safety. Key technological prerequisites for this new approach are the ability to monitor the condition of critical components, and the ability to accurately forecast their remaining useful life. Phase I of this project clearly demonstrated the feasibility of accurately predicting the remaining life of rolling element bearings. During Phase II, Sentient Corporation will complement in-house test capabilities by teaming with The Timken Company to assemble a comprehensive, high quality database of bearing spall initiation and propagation for tapered roller, angular contact, and radial bearings, while Purdue University will research material and component level phenomena. New understanding of bearing failure processes resulting from this data will be used to further develop the physics-based model created during Phase I. The goal is for the model to represent the physics of failure accurately enough that it will work in virtually any bearing application without modification. This will be demonstrated by evaluating model performance across the entire experimental database. In addition, the model will be enhanced to utilize diagnostic data for refining model predictions and estimating uncertainty.