Dynamics-based condition indicators for defective rolling elements in ball bearings: a multi-sensorial endurance test and simulation analysis

• First endurance test of rolling element defects under combined loading. • New CIs based on FTF rise and double-BSF decay identify failure onset. • Physics-based model links signal trends to orientation-dependent load capacity. • Continuous ODM serves as physical ground truth for the critical failure transition. • Multi-sensorial strategy provides robust precursors for RUL prediction. Due to their orientation-dependent dynamics and the scarcity of long-term experimental data, rolling element (RE) defects pose significant challenges in prognostics. To investigate RE defect degradation, this study combines extended endurance testing of an angular contact bearing under combined radial and axial loading with high-fidelity multibody dynamic simulation. Multi-modal measurements using acceleration, strain, and oil debris monitoring revealed the evolution of a step-like mass loss and a critical knee point marking the onset of severe degradation. A previously unreported relationship between the fundamental train frequency (FTF) and an alternating double ball spin frequency (2xBSF) component was identified. This relationship was physically explained using a validated ADAMS model, which attributed the behavior to orientation-dependent load-carrying variations of the defective RE. Based on these findings, we propose new physics-informed condition indicators that combine FTF growth and 2xBSF decay as precursors to the knee point, thereby advancing the understanding of RE defect dynamics and supporting future remaining useful life prediction. The main contributions of this work are threefold: (i) it provides rare long-term experimental evidence of RE fatigue under combined radial and axial loading, a configuration that is highly relevant to industrial applications, yet has seldom been studied; (ii) it introduces multi-indicator condition metrics that act as precursors to the knee point and support future remaining useful life prediction; and (iii) it establishes the physical mechanisms underlying these indicators through validation utilizing a high-fidelity multibody dynamic model. Collectively, these results advance the fundamental understanding of RE defect dynamics and contribute to a robust, physics-informed strategy for predictive maintenance of RE bearings.