Stay one step ahead of bearing failure

Barden UK's Trevor Morris gives the background to why noise and vibration measurement is so important in maintaining bearing performance.

Preventive maintenance is crucial with industrial bearings, if downtime is to be minimised, and there's no question that the best way of doing this is by analysing noise and vibration.

Abnormal operation, predicting the potential need for a replacement bearing, is signalled early on by this method.

But why is becoming increasingly important? The drive for continuous improvement in modern manufactured products means that the demands on rolling bearings for reduced noise and vibration become more acute, year on year.

This is particularly true with small and miniature bearings that are used in instrumentation, domestic electrical equipment, including white goods, and also vehicles where modern design methods mean that structures are lighter and thinner and therefore more likely to resonate with external vibration.

Here, by using the term "vibration", rather than "noise", the indication is that the amplitude of the occurrence is less than 1kHz, which is the generally accepted boundary between noise and vibration: anything above 1kHz is vibration, and below 1kHz is noise.

What is important to understand is that all bearings, even those in perfect condition, produce "noise" as their elements roll over the raceways and contact against the internal cage and flanges.

This noise is generated at high frequency and low amplitude.

The bearing housing amplifies the noise to a point where, using a sensitive accelerometer it can be "heard".

By careful detection and filtering, the noise signal can be amplified and represented as a frequency series in real time.

Most bearing faults occur with the rolling elements, cage or raceways.

The frequency of the fault has a direct relationship with the geometry of the bearing and the relative speed of each individual raceway.

By comparing the actual noise to the ideal bearing signature it is possible to highlight problems with the in-service bearing.

The method of presenting the data is either by visually representing it on an oscilloscope screen or as a number on a digital display.

When considering the effects of vibration on bearing operation two types must be considered: vibration arising from exposure to external sources and that from from self-generated frequencies.

In the first case, bearings either fail or their performance degrades in modes known as false brinelling, wear oxidation or corrosion fretting.

Such problems arise when loaded bearings operate without sufficient lubrication at very low speeds.

They are also the result of oscillation and can occur even when bearings are stationary.

When vibration is added, surface oxidation and selective wear result from minute vibratory movement and limited rolling action in the ball-to-raceway contact areas.

These conditions can be relieved by properly designed isolation supports and adequate lubrication.

Self-generated vibration is the result of nanometre variations of circular form in the bearing balls and raceways.

At operating speed, low-level cyclic displacement can occur as a function of these variations, in combination with the speed of rotation and the internal bearing design.

The magnitude of this cyclic displacement is usually less than the residual unbalance of the supported rotating member, and can be identified with vibration measuring equipment.

The presence of a pitched frequency in a bearing can also excite a resonance in its supporting structure.

However, whilst the principal frequencies of ball bearing vibration can be readily identified from the bearing design, frequency analysis of supporting structures is usually more difficult, but can be accomplished experimentally.

The difficulty in achieving resonant-free performance even in simple bearing support structures highlights the problems faced by engineers in far more complex mechanical systems that integrate shafts, bearings and housings, and are subjected to external loads.

The vibration performance of these systems is complex and often unpredictable.

For example: a lightly damped resonance can put performance outside acceptable criteria at specific speed ranges.

However, it must be said that the interaction of system resonances and bearing events in these systems is most pronounced in less-than-ideal designs (long, slender shafts, overhung rotor masses etc).

These designs are relatively uncommon, and require a lot of engineering effort to resolve.

They are usually solved through a series of iterations, by altering ball counts, adjustments to radial and axial stiffness, and sometimes by changing the natural frequencies of the bearings themselves.

(Updated by CR, May 2007).

Back to top