American Thermal Imaging

305 Plum Street
Red Wing, MN 55066

Phone:
651-385-0051

Fax: 651-385-9328

A Brief Summary of Vibration Analysis
Virtually every object we deal with vibrates. In some cases, the motion is minute, of little consequence while others are very perceptible, and affects our daily life. In most cases, our senses of touch and hearing are our detectors of vibration. These are keen senses but are limited in range and the ability to quantify the motions. An instrument called a spectrum analyzer does a much more detailed assessment of the motions. It shows both the frequency and the magnitude of even the most minute motions. The vibration analysis of equipment typically uses an accelerometer for sensing the motion. An extremely sensitive device, it translates motion into an electrical signal. The spectrum analyzer takes the signal from the accelerometer and translates it into a dynamic signature associated with a particular type of motion. As an example, when a hammer strikes a bell, the bell will resonate at its natural frequency and sends out sound waves that we can hear. If one were to place an accelerometer on the bell, it would measure the natural frequency of the bell which is the tone we hear. The signal would appear as a continuous waveform of constant frequency that decreases in amplitude as the ringing volume decreases. If one were to look at a spectrum (an FFT) of the ringing sound, it would show only one peak at the particular frequency that the bell was ringing. When viewing spectra, remember that one is seeing vibration caused by other elements in the machine and not just a single source such as a bearing etc. Some of this "other" information is useful in diagnosing equipment while some is nothing more than clutter to be ignored. For example, a fan mounted in the machine will show up as a vibration consisting of the speed of the fan multiplied by the number of fan blades (called the blade pass frequency). Its presence is usually of little consequence. Other background vibrations that are commonly displayed in a spectrograph include timing belts, SCR drive switching frequencies, gear trains, and structural vibration etc.

When viewing a spectrum, there are several important characteristics:

Fundamental frequency: For rotating machinery diagnostics it is assigned to the value of the shaft speed.
Harmonics: Integral multiples of the fundamental frequency. The second harmonic is 2X the shaft speed. The third harmonic is 3X the shaft speed etc.
Sidebands: For most rotating machinery applications, sidebands are spikes in the spectrum that are evident on either side of a frequency spike and whose value is often equal to the shaft speed.

When viewing a vibration analysis spectrograph, one must considers both the shape of the measured waveform and its magnitude. The magnitude by itself can be misleading simply because of the location of the accelerometer relative to the source of vibration. A long or interrupted path consisting of bolted joints etc will attenuate the signal and the magnitude will be diminished. It is similar in principal as when one moves away from a sound source, the volume decreases.

For condition monitoring, there are two general categories of faults revealed in a vibration analysis spectrum:

Rotating machine dynamics and bearing defect frequencies.

Rotating dynamics can be caused by imbalance, loose mounts, shaft misalignment, faulty coupling, and failing structural elements. These are easily seen as high-energy peaks at relatively low frequencies.

Because bearing analysis is one of the most important vibration measurements in predictive maintenance testing, it warrants more detailed discussion.

Bearing dynamics are characterized by very specific fault frequencies, which are based on the bearing geometry and speed . These fault frequencies are not present in normal operation and thus do not show up in a spectrum. If there is a defect in the bearing such as a spall or pit, the impact of the ball or roller going over this flaw will excite the frequency associated with that part. Therefore, spectrum analysis of a failing bearing is concerned with the presence and the relative amplitude of the defect frequencies including their harmonics.

The specific fault frequencies are defined as follows:

FTF: Fundamental Train Frequency relates to the rotation rate of the cage. It appears between 0.35 and 0.45 times the running speed. The FTF does not typically show significant energy in the spectrum but rather most often shows up as sidebands surrounding the other defect frequencies.
BSF: Ball spin frequency relates to the rotational rate of the balls or rollers. It is not to be confused with the ball pass frequency, which is the number of balls times the speed of rotation. The ball pass frequency does not typically show up in the spectrum.
BPFO: Ball Pass Frequency Outer Race is determined by the rate at which a ball passes over a fault in the outer race. The BPFO is approximately equal to 0.4 x RPM x number of balls or rollers.
BPFI: Ball Pass Frequency Inner Race is determined by the rate at which a ball passes over a defect in the inner race. The BPFI is approximately equal to 0.6 x RPM x number of balls or rollers.

It can readily be seen that the specific bearing geometry of all of the various bearing types becomes significant. For this reason, ATI subscribes to the worlds largest bearing database that contains all of the specific frequencies for any given bearing (over 265,000 part numbers).

Ultrasonic Testing: Ultrasonic testing takes place in the frequency range from two to five times higher than can be picked up by the human ear. It is of value in detecting bearing and gear faults far ahead of actual failure. The results of the testing are more qualitative than quantitative and as such, the technique is used primarily as a screening tool to help determine if further analysis is warranted. The readings that are recorded are relative measures and are meaningful only for future assessment comparison. The sound of the tone heard in the headset determines what action should be taken.