Abstract
This paper evaluates a strain measurement technique using raw signals for estimating the spall size in rolling bearings, enhancing condition-based predictive maintenance. Traditional methods, reliant on acceleration for defect severity assessment, struggle with unpredictable factors like transfer paths and varying operational conditions. Alternative approaches using shaft-housing displacement, housing strain, and load cells have shown potential in identifying spall length. However, these methods do not provide the rolling element contact forces during spall over-rolling, which potentially contain more detailed information about spall geometry, propagation rate, and potential quality loss and consequential damage.
The study introduces a thorough analysis of bearing strain signals extracted at the bearing outer ring surface in the presence of spalls. A dynamic model with experimental validation is used to analyze the impact of various parameters on strain signal features and rolling element contact forces. The results indicate that bearing strain signals offer distinct and consistent defect severity features, largely unaffected by speed and load, addressing the primary challenges of acceleration-based methods. A single strain raw measurement signal reveals the spall size and the reduction in contact forces.
This method can be an important enabler for physics-based prognostics, leveraging actual defect geometry and contact force data to predict spall progression. Consequently, bearing strain-based monitoring facilitates effective, true condition-based predictive maintenance.