Bob Jones
Principal Applications Engineer
SKF Condition Monitoring
San Diego, Calif.

When bearings show signs of fatigue, such as spalls or surface cracks, many engineers think they should simply replace the bearing. Failed bearings are often unfortunate signs of problems elsewhere in a machine. For example, repeated bearing failure can result from unbalanced shafts or loose components.

Vibration analysis helps spot these problems early, preventing unexpected shutdowns and often allowing bearing replacements during scheduled machine outages. Monitoring wear also maximizes bearing life because experienced technicians can estimate how much wear a bearing will sustain and can then run the bearing as long as possible.

Detection tools
Vibration-monitoring devices are either portable units or permanent on-line setups. Portable devices are often the starting point for vibration monitoring because the equipment is less expensive and does not require installation. On-line systems, however, overcome several shortcomings of portable devices. Permanently mounted sensors can be installed in hard-to-reach machine areas or in harsh environments. Permanent connections also improve measurement accuracy. And on-line systems don’t require trained personnel to take measurements. This makes it easy to increase the reading frequency, which is recommended after detecting defects.

To interpret bearing vibration data, engineers must determine each bearing’s fault or defect frequencies. Published software programs output this data for visual inspection. Each bearing component vibrates at a unique frequency. A defect on any component emits vibration peaks at the component’s defect frequency. A peak at a defect frequency or its harmonics indicates a problem with that component of the bearing. All rolling-element bearings have four defect frequencies corresponding to each bearing component. These are called ball-pass frequency outer race (BPFO), ball-pass frequency inner race (BPFI), cage or fundamental-train frequency (FTF), and ball-spin frequency (BSF).

Although vibration-monitoring software contains defect frequencies for bearings from major manufacturers, equations are also available. BPFO, for example, is calculated with:

BPFO = n/12 X (rpm)/60
X (1-Bd/Pd X cos(A))

where n = number of rolling elements; Bd = ball diameter, in.; Pd = pitch diameter, in.; and A = contact angle.

After determining the bearing’s defect frequencies look for peaks in its vibration output. A peak at a defect frequency or at harmonics indicates a possible bearing problem. When rolling elements repeatedly pass over a defect such as a spall or crack, they generate an impulsive vibration. This impulse repeats anywhere from five to l0 times shaft speed and is nonsynchronous with shaft speed. Nonsynchronous repetition results because of the mathematics involved in the calculations. The cage that separates the rolling elements rotates at about 0.35 times shaft speed.

Vibration signals generate harmonics at multiples of the defect’s fundamental frequency. For example, if a BPFO is 6.5 times shaft speed, repeated vibration signals should occur at 13, 19.5, 26 times shaft speed and so on. Initial defects can produce vibration signals at harmonics up to 50 times the fundamental defect frequency.

Engineers can set alarms in on-line monitoring equipment to warn of changes in bearing conditions. When an alarm goes off experts recommend additional checks using portable test devices. This helps pinpoint the source of the problem. Monitoring equipment can output data as overall time waveforms, displacement, velocity, acceleration, and enveloped acceleration spectrums.

Overall vibration readouts plot a combination of all machine vibration signals in a time waveform plot. These are fairly impractical for finding bearing defects because they do not isolate specific vibration signals. A more useful method is a Fast-Fourier Transformation. FFTs display overall vibrations in component signals and plot them on a frequency scale. This isolates each vibration signal at its frequency and lets analysts check for specific machine faults. The frequency indicates the vibration source and amplitude indicates severity. When peak amplitudes arise at specific frequencies, suspect the component that generates its vibration signal at that frequency.

Most analysts use a combination of velocity and enveloped-acceleration measurements. Velocity measurements are general-purpose indicators that highlight low-frequency vibrations such as imbalance and misalignment. Enveloped acceleration, on the other hand, captures low amplitude, high-frequency signals like bearing defects. These are typically lost among higher-energy machine noise. The technique of enveloping filters out most nonrepetitive machine vibrations. Enveloping then enhances the repetitive vibrations, letting analysts detect faults early. Several methods determine a fault’s severity. For example, compare current vibration readings with past ones and with similar machines running under the same conditions. Look for sideband peaks of shaft speeds near defect harmonic and fundamental frequencies. Sidebands typically indicate severe bearing damage.

Causes of problems
After noticing questionable frequencies technicians must look for machine problems. Some primary sources of excessive vibration are imbalance, misalignment, and mechanical looseness. Experienced personnel may discover faulty equipment by looking, feeling, or even smelling. Finding less obvious problems, however, requires understanding how various components vibrate. Imbalance, for example, can produce vibrations with high amplitude at frequency equal to shaft speed. Initial bearing defects, on the other hand, generate low amplitudes and high frequencies. Gear-mesh vibrations also have low amplitudes with frequencies depending on shaft speed and the number of gear teeth.

Vibration from pure imbalance is a once-per-revolution sinusoidal waveform. On an FFT spectrum, this appears as a high 1X amplitude without harmonics. Although other faults can produce high 1X amplitudes, they usually produce harmonics as well. A variety of factors can lead to imbalance, including improper manufacture, uneven debris build up on rotors, vanes, or blades, and installing shaft fittings without counterbalancing.

Misalignment is either angular or parallel and often a combination of the two. Angular misalignment causes axial vibration at running-speed frequency. Parallel misalignment produces radial vibration at twice or three times running-speed frequency. 1X and 2X readings often appear simultaneously because of combined angular and parallel misalignment. Few faults other than misalignment produce excessive 3X vibration. Possible misalignment causes include thermal expansion after aligning cold machines, coupling properly aligned machines to misaligned ones, and uneven, settling foundations. Imbalance and misalignment can cause bearings to carry higher dynamic loads than their design specifications.

Mechanical looseness also leads to bearing failure. It generally appears on an FFT spectrum as a long string of rotating-frequency harmonics or 1⁄2 rotating-frequency harmonics. Machines with loose mountings or loose internal components are two types of mechanical looseness. If looseness results from a component, such as a fan blade, the part could detach and cause secondary damage. Indirect sources, such as looseness and imbalance, are not the only causes of bearing failure. Problems come from excessive loads, improper lubrication, handling, and installation.

Initial bearing fatigue produces shear stresses that appear immediately below the load-carrying surface. These stresses eventually cause cracks that gradually propagate to the surface. Fragments break away as rolling elements pass over cracks. This is either spalling or flaking. Although bearings with light spalling or flaking may not need replacing, the conditions will likely worsen until the bearing is unusable. Excessive loads or improper lubrication may also produce surface cracks that grow into the material. Fatigue and surface distress impact bearing behavior, producing noise and vibration detectable by monitoring equipment.

© 2010 Penton Media, Inc.