Identifying the stages of bearing damage

The early identification of troublesome conditions such as inadequate lubrication or misalignment enables an analyst to apply proactive corrective measures to extend bearing service life. The onset of bearing deterioration often occurs very early on, as raceways begin to wear, developing micropitting in the load zone.

Micropitting at this very early point does not necessarily reduce operational life, but is often a good indication that progression to stage one is imminent.

Stage One is still a "good" bearing. However, after a substantial portion of the bearing life has passed, micropitting results in the degeneration of the bearing to the point where very small craters develop in the raceways.

These small defects are not always impacted with enough force to generate measurable vibration signals in the velocity domain. As the bearing degrades and the rolling elements impact the fault, harmonics of the damage frequency will begin to show in the FFT spectrum.

Stage Two is a bearing with some wear as noted by the harmonics. There is no reason to change the bearing at this point. In fact, bearings have been pulled at this stage and the only damage apparent is pinpoint spalling in the raceways. As the harmonics increase in amplitude it may be prudent to increase the frequency of data collection. Bearing degradation is usually linear for a period of time and can be trended, but as service life is shortened, it becomes non-linear.

In Stage Three the bearing is becoming terminal. The FFT spectra show the fundamental defect frequency and the harmonics will often begin to indicate sidebands of bearing shaft rotating speed. This is particularly true of the BPFI (Ball Pass Frequency Inner Race) where the defect rotates through the bearing load zone. The vibration increases as the defect goes through the load zone and the signal is modulated producing the rotational sidebands. The BPFO (Ball Pass Frequency Outer Race) signal generally has constant bearing loading until bearing looseness, imbalance, misalignment or bent shaft amplitude modulates the defect signal often resulting in rotational sidebands.

For example using a BPFO of 107.6 Hz on a shaft turning 1800 RPM or 30 Hz, the sidebands will be at 77.6 Hz and 137.6 Hz (107.6 plus and minus 30) and the second harmonic will have sidebands at 185.2 Hz and 245.2 Hz (215.2 plus and minus 30). Further progression of the damage will generate additional sidebands at 2X running speed (47.6 and 167.6). When the sidebands overlap, the spectra become more difficult to analyse. But beware! The bearing is in the last days of its life and should be changed as soon as possible. During Stage 3, in addition to the spectrum information, the overall amplitudes provide a clue to the condition of the bearing.

Stage Four The service life at this point is extremely short and requires immediate corrective action. It is often characterised in the velocity or acceleration spectral domain as "haystack" amplitudes (broadband noise) in the bearing defect region. In acceleration enveloping spectra there will appear high amplitude defect frequency components as well as the 1X, 2X speed sidebands (indicating looseness) about the BPFO and in the extreme case, cage defect frequency components will appear.

Bearing Damage - Examples

The following spectra illustrate bearings in various stages as compared to standard velocity readings. Although we normally do not look in the enveloped time domain, the bearing defect signals will be evident if the defect is large enough. In Figure 1 the inner race defect shows up as each roller passes over the defect. The markers show the interval to be 0.00748 seconds which equals the frequency of 133.5 Hz. The bearing was damaged during installation..

Figure 1 - Acceleration Enveloping / Time

Figure 2 is a vibration signature of the same bearing, Pt# 6201. In velocity (IPS) we do see the BPFO since the damage was extensive, but there are no harmonics and the amplitude is low. The 60 Hz and harmonics are power line noise components.

Figure 2 - Velocity

Figure 3 is the same data process using enveloped acceleration.

Figure 3 - Acceleration Enveloping

In looking at these three plots one might wonder if all of these readings are necessary. However, when a decision must be made on whether or not to shutdown a machine, it is advisable to use all available diagnostic tools. The more qulaity information we have, the better our decisions.

Figure 4 is the spectrum of an MRC bearing that had been improperly stored. It remained on the shelf for seven years in an acidic environment which resulted in etching the balls as well as the side of the race. When the bearing was installed, the balls generated the Ball Spin Frequency (BSF) at 95.0 Hz. There was not a BSF component in the velocity spectrum as shown in Figure 4. It only appears in the enveloped spectrum of Figure 5

Figure 4 - Velocity

Figure 5 - Acceleration Enveloping

Although the spectrum amplitudes are very low, the defective rolling element produced clicking noises as the machine ran.

Figure 6 is the spectrum of a bearing with a damaged cage. The owner was unhappy with the high vibration amplitudes and the analysis confirmed the cage defect. It was discovered that the bearing was 20 years old! One additional problem here was that the eighth harmonic of Fundamental Train Frequency (FTF) overlaid the third harmonic of the rotating speed. The bearing was changed and the vibration was reduced to acceptable levels.

Figure 6 - Velocity

The next set of examples are from tests run with a compressor manufacturer who wanted proof that bearing problems could be detected.

Before we arrived he had purposely damaged one of the six bearings and asked us to find the modified bearing. The measurements confirmed the existence of the defect.

Figure 7 is the acceleration spectrum (G's). The fundamental BPFO does not show up but the harmonics do. This would alert the analyst that something is going on, but most people don't take routine bearing measurements in acceleration "G's".

Figure 7 - Acceleration

One additional note on bearing troubleshooting. Most bearings will have their BPFO and BPFI at non-integers of rotating speed. In other words, with the rotating speed at 1X, the BPFO will be at 3.56X or 4.73X. This is a valuable clue when looking at a machine when you have no idea of the type of bearings that are installed.

Figure 8 is the velocity measurement which is the method most people use. There is absolutely nothing here that would cause one to worry. Of the six bearings in the unit, the owner did not tell us which one was damaged, but because of the different sizes, the bearing defect frequencies are all different. It turned out the BPFO of the bad bearing was 400 Hz.

Figure 8 - Velocity

Figure 9 is the same reading processed using acceleration enveloping. This clearly shows the BPFO which was not visible in the velocity reading.

Figure 9 - Acceleration Enveloping

The next four spectra are from very slow speed bearings. this application is probably the most valuable for enveloping as opposed to velocity since there is no integration and therefore no amplification of the low frequency noise present in all data collectors. The low frequency cut-off can be set at 0 Hz and the spectrum will not have the usual velocity spectrum "ski slope".

These spectra are from a conveyor drive gear box. The bottom bearing had a history of failure and the customer was required to keep 3 days inventory on hand to maintain his just in time shipments if the bearing failed. Knowing the condition of these bearings enabled an assured repair schedule to include a reduction of just in time inventory of expensive machine components.

The shaft speed for this unit is 8.3 RPM. The BPFO varied between 1.4 to 1.8 because of speed changes while four different readings were being taken.

Figure 10 is the acceleration envelope spectrum with the BPFO at 1.8 Hz and two associated harmonics. Since there are no running speed sidebands, the bearings were diagnosed as having some light damage.

Figure 10 - Acceleration Enveloping

Figure 11 is the normal velocity spectrum. Notice there are no indications of bearing damage at the BPFO of 1.7 Hz. Figure 12 is an acceleration spectrum of the same bearing showing no damage indicators. On the other hand, Figure 10 clearly shows the BPFO with harmonics. Our advice was to keep running the machine but increase the observations to once per week rather than once a month. Considering the critical nature of this operation, this bearing should be changed at the next opportunity.

Figure 11 - Velocity

Figure 12 - Acceleration

The next example, Figure 13, is the other bearing in the same gearbox. It is heavily loaded and just staring to show some harmonics of the BPFO defect. It is difficult to obtain examples of stage three damage. If the observer is paying attention at all then he can hear that something is not right. the bearing example in Figure 14 is a bearing that was damaged enough to cause sidebands. The frequency gap between the sidebands and the fundamental equals a shaft speed of 40 Hz. These same sidebands also appear along with the harmonics components.

Figure 13 - Acceleration Enveloping

Figure 14 - Acceleration Enveloping

The final example, Figure 15, is a gear box that was in a stage three condition. The owner asked for a demonstration of the enveloping process to show how it could be used in a route setup. When the frequency range was set, we did not know the gear mesh frequency so we selected a range of 1000 Hz. This would be fine for "routine" data collection. Had we been looking for a problem in the gear box we would have selected a higher frequency range - probably about 2000 Hz. We barely caught the second harmonic at 990 Hz, but the sidebands are very clear at 25 Hz, the shaft rotating speed.

Figure 15 - Acceleration Enveloping

When we expressed our concern that it looked to be a major problem with the gear box, the owner just smiled and said "I know, I just wanted to see if your data collector was any good". It is.


Acceleration Enveloping is a relatively new measurement approach. Nonetheless, it has proven to be very valuable diagnostic indicator of a wide range of machine problems. In addition to velocity and acceleration as standard measurement parameters, these envelope methods are being universally adopted in predictive maintenance programs around the world.

Although care has been taken to ensure the accuracy of the data compiled on these pages, the apt Group does not assume any liability for error or omissions.