Dynamic Modeling and Wear-Based Remaining Useful Life Prediction of High Power Clutch Systems

Abstract

A model-based technique is presented for remaining useful life (RUL) prediction of highly dynamic, high-power dry clutch systems by combining physicsbased simulation and wear-prediction models. Primary load and engagement shear drivers (i.e., torque, speed, and clutch surface temperature) are modeled using a first principles approach. An extension of Archard’s law is presented in which life usage is predicted by using multiple stochastic models to determine a wear coefficient for each applicable wear mechanism. These models consider the physical wear process, including debris-particle and protective-layer formation, using parameters such as surface roughness, particle size, and surface temperature. These stochastic variables are evaluated in a probabilistic framework, using statistical methods such as Monte Carlo and importance sampling, which consider both measurement and modeling uncertainty. Confidence interval prognostic results are provided to predict the RUL of the clutch throughout its limited life in near-real time.