A Comprehensive Approach for Predicting Remaining Useful Life in Electric External Gear Pumps

This paper presents a new methodology for predicting the Remaining Useful Life (RUL) of commercial electric pump units, aimed at overcoming limitations observed in existing approaches. In particular, it does not require a full-life test, its output is stochastic, it can be scaled to different operating conditions, and it can be used to guide design changes. In the context of Prognosis and Health Management, having a model that correctly forecasts RUL can enable predictive maintenance-based strategies that can reduce costs and improve operability. As known in the literature and observed by the experimental data collected in this study, although degradation is monotonic, its effect in terms of measured output is not. The proposed methodology leverages the Brownian model to predict the degradation path. The proposed framework incorporates Monte Carlo Simulations to provide a stochastic output with failure probabilities and survival plots. The model effectively captures the non-monotonic consequences of degradation, particularly influenced by particle interactions, by accounting for randomness. Furthermore, Computational Fluid Dynamics (CFD) simulations with a dispersed phase have been proven to identify erosion-prone areas and suggest design modifications to enhance pump robustness. Combining experimental insights, stochastic modeling, and predictive simulations, this comprehensive approach establishes a robust foundation for RUL predictions and informed design enhancements in hydraulic pumps.