Rolling Element Bearings
Rolling element bearings are ubiquitous in small to medium sized equipment. Detecting faults in them is a major justification for
monitoring. The major components of a rolling element bearing are the inner race, outer race, bearing cage, and the rolling elements
themselves. Faults may appear on any of these components.
Cause
Incorrect selection, poor lubrication, electrical discharge, misalignment, resonance, and supply harmonics can all increase the rate of
bearing deterioration. Root causes should be identified long before bearing faults start to present.
Effect
Bearing problems cause excitation forces that may affect other components, as well as local heating and contamination of any lubrication
in the bearing, which will accelerate the rate of bearing failure. Complete failure of the bearing will prevent the equipment from rotating
and will cause further secondary damage.
Diagnosis
Bearings have a number of characteristic frequencies, including inner/ outer race ball pass frequency (the rate at which the rolling
elements pass a fixed point on the race, abbreviated to BPIR and BPOR), and train frequency (the rotational rate of the bearing cage/ train).
These numbers are usually given as a fraction of rotational speed.
These characteristic frequencies can be obtained from datasheets or through bearing calculators. Approximations are typically valid (outer
race frequency is 0.4 * shaft speed * number of elements, inner race frequency 0.6 * shaft speed * number of elements, train frequency
0.4 * shaft speed). The exact coefficient used to calculate these frequencies precisely depends on the shape of the rolling elements, but for
the outer race this number is always below 0.5 and generally above 0.35; this number represents the relative rate of movement between
the outer race and the rolling element. In a perfect roller this would be 0.5 exactly, but this is never achievable in practice.
Where there is a fault, these frequencies appear on the spectrum as sidebands on a carrier frequency, e.g. at line frequency + BPIR *
rotational speed. The complete formula for this kind of fault is:
𝑛 𝑓𝑙𝑖𝑛𝑒 ± 𝑚 𝑓𝐵𝑒𝑎𝑟𝑖𝑛𝑔
Here n and m are integers, and most often n = 1, which means bearing faults appear either side of line frequency. F_bearing is the bearing
frequency, either BPIR, BPOR or train frequency, described above. Visually on the PSD this looks like two or more smaller peaks centred
around a line harmonic.
As long as the correct bearing information is entered into the P100, peaks at these frequencies are detected automatically.
Diagnostic parameter – Bearing
Bearing information can be entered into the equipment information table under driver, driven 1 or driven 2. If the bearing numbers are
known, this should be entered, and the P100 will use a bearing database to complete the bearing information .
Action
Actions should be proactive, dealing with root causes in time to prevent bearing damage. Once damage begins, offline oil filtering may
extend bearing life with careful monitoring so that replacement can be carried out at minimum cost. Thermal imaging may be able to
detect late stage bearing problems – but at this point the bearing may need replacement asap.
