Analysis and Prediction of Defect Size and Remaining Useful Life of Thrust Ball Bearings: Modelling and Experiment Procedures

Abstract

A new methodology based on full factorial design (FFD) for the prediction of the size of the spalls on the rolling contact surfaces of thrust ball bearings and its remaining useful life is presented in this research work. The FFD is used to analyse the effect of the following factors: rotational speed, load, spalling size, position of the sensors and frequency range on the vibrations level and the interaction effects between these factors. Data collected from vibratory signals generated by thrust ball bearings operating in the fatigue test rig are used. Statistical analyses show that the full factorial model with sixteen coefficients leads to more precise spalling size estimates. Analysis of variance showed that these results were statistically significant [P value is less than 0.0001 where significance level $$\alpha $$ ( $$\alpha = 0.05$$ )]. A modified exponential model used in the prediction of the remaining useful life shows its efficiency and applicability.