Systematic Methodology for Generating Natural Spall Faults on Rolling Element Bearings

As critical components in rotating machinery, the health condition of rolling element bearings needs to be estimated and monitored for maintenance decision-making such that machine downtimes due to catastrophic bearing failures can be minimised. Vibration-based condition monitoring is a well-established technique to monitor the bearing health condition. Due to practical challenges faced in different applications, a massive number of algorithms to process the vibration signals for bearing health monitoring purposes have been progressively developed and reported by numerous researchers in the last decades. Many of the fault diagnosis algorithms developed by the academic communities were tested on the vibration signals acquired from laboratory test rigs on which bearings with artificial (fabricated/manufactured) faults were used. As a consequence, the developed algorithms might not be effective to detect bearing faults in real-life situations. A standard accelerated life test (ALT) concept can be employed to generate realistic/natural faults on rolling-element bearings. However, the standard ALT is still very time-consuming (i.e. test duration is still relatively long) and thus expensive. In this paper, we examined the progression of surface rolling contact fatigue (RCF) based on recent works of researchers in the tribology research field. The gained physical insight into the RCF progression inspires us to propose a systematic methodology for generating realistic/natural spall faults on rolling element bearings with a reasonably short test duration and thus decreasing the test cost.