Roller bearing prognosis requires the detection of a bearing defect signature in the earliest possible stage in order to avoid a minor or catastrophic mechanical failure. Defects can occur in any of the bearing parts, inner and outer race, cage and rolling elements. It is possible to identify the defective component of the bearing based on the specific vibration frequencies that are excited. However, the pattern of vibration spectrum changes as the bearing deteriorates through different stages. Depending on which failure stage the bearing is in, different techniques are required to find fault signatures in different frequency ranges. Techniques such as enveloping analysis that works in the high frequency region require higher data sampling rates and therefore more expensive data acquisition hardware than techniques conducted in low frequency region. This paper compares two popular rolling element bearing diagnostics techniques — spectrum analysis in the bearing characteristic frequency range and enveloping analysis in the high frequency range — using aircraft engine test rig data. The techniques are compared both in terms of the time of detection and data sampling requirement; this analysis provides guidance for technology adoption in future field deployment. Results demonstrate that enveloping analysis is able to detect bearing defects much earlier than the spectrum analysis, but it requires a higher data sampling rate. The bearing defect characteristic frequency is detectable in low frequency spectrum only in the late stage of the failure and it is contaminated by other harmonics such as shaft unbalance. From a practical perspective, the final choice of the technology adopted for deployment should be based on an analysis of hardware requirements and tolerance of detection latency.

This content is only available via PDF.
You do not currently have access to this content.