7/29/2019 Detecting Bearing Faults, Part 2

http://slidepdf.com/reader/full/detecting-bearing-faults-part-2 1/2

58  april/may11

How Much of a Risk AreYou Willing to Take?

What are your goals? Do youwant to know that a bearing mayail just days beore it is likely toail, with no prior warning? Orwould you like to know that abearing has been poorly lubri-cated, or has a minor deect thatwill develop into a major ault?With the techniques described inthis article you could learn thesethings months (certainly weeks)beore the bearing is likely to ail.With that extra time you couldchange the lubrication, orderparts, orga-

nize the labor, and look or the best op-portunity to perorm the bearing replace-ment. The result is a saer plant with lessdowntime, less stress, and higher prots.

Brief RecapIn the previous article, a ew important

points were made that are pertinent tothis article:

1. As bearings begin to ail, the vibra-tion is very low in amplitude, and therequency is very high (beyond your ability to hear, evenwith the best screw driver).

2. Simple spectrum analysis will not reveal the ault until ithas developed to stage three, unless you take special pre-cautions (listed later in the article).

3. To measure high requency vibration you must mount thesensor correctly.

Beyond Your Hearing: UltrasoundThe ultrasound technique is very easy to implement. The

measurement tool listens or very high requency vibration andprovides an indication o amplitude. It also amplies the vibra-

tion and shits (heterodynes) the requency so that you canhear it through headphones. Thereore, you can listen or thetelltale sounds o poor lubrication and bearing distress. Whenused correctly and appropriately, ultrasound instruments canbe very complimentary to the other techniques described inthis article.

What Are “Stress Waves,”and Why Should I Care?Beore describing the next two techniques it is important to

briey introduce the concept o the stress wave (also known asshock pulse). Metal-to-metal contact sets o a ripple eect: astress wave races through the metal components, causing thecomponents to vibrate due to resonance. The stress wave isa very short-duration, low-amplitude, high-requency wave.Every time the rolling elements roll over the damaged area onthe inner and/or outer race (or as the damaged areas on the

rolling ele-ments contactthe raceways),a stress wavewill be gener-ated. We canseek to detectthat wavewith tech-niques such asShock Pulse,PeakVue, and

SWAN (Stress Wave ANalysis, not discussed urther in this brie article). The vibration that results can also be detected via theenvelope method, and as the ault develops urther, via thetime waveorm and spectrum.

There is one very key point you must be aware o: we are talk-

ing about very high requencies, and as such the vibration sensormust be mounted correctly . Unless specically designed or thepurpose (e.g., Shock Pulse), a handheld probe is horribly inad-equate. Even a two-pole magnet mounted directly to the ma-chine surace is not adequate! All o the analyzer vendors will tellyou, you must properly prepare the surace and use an attach-ment pad (or stud mount) in order to achieve the best results.

It is also important to note that there are other deects that will gener-ate stress waves and high-requency vibration, including looseness, gearwear, and cavitation. That can help us to detect those conditions, but itcan conuse our attempts to detect bearing and lubrication aults.

Shock PulseThe vibration sensors provided by SPM and PRÜFTECHNIK are de-

signed to ampliy (through resonance) high-requency vibration (at ap-

 proximately 35 kHz). As noted earlier, lubrication and physical deects(including wear/spalls) will generate vibration around this requency.The vibration can be displayed as an amplitude to be trended, or a spec-trum can be displayed in order to better understand the specics o thedeect: inner race, outer race, etc.

Spike EnergyThe Spike Energy (units o gSE) technique aims to utilize the acceler-

ometer’s mounted resonance to ampliy the high requency vibration.However, in more recent years, the accelerometers provided have notbeen manuactured to have a repeatable resonance characteristic. Whatthat means is that when you change your accelerometer, the amplitudeswill change.

PeakVueThe PeakVue technique, developed by Emerson Process Management

(CSi Division), is also designed to detect the stress wave; however, it isperormed in a dierent way. The signal rom the accelerometer is digi-tally sampled (converted rom analog voltages to digital numbers) at avery high rate so that the very short duration stress waves can be de-tected and quantied. The PeakVue waveorm and spectrum providean indication o the bearing deect. As with all o the techniques, theaccelerometer must be mounted correctly, and the lter settings (usedto “tune in”to the bearing vibration) must be set correctly.

EnvelopingAlso known as “demodulation,”the enveloping technique, which is

used by a large number o vibration analyzer vendors, has been opti-mized to measure the low-amplitude, high-requency bearing vibration.

See Figure 1.The envelope spectrum is then checked or signs o the ault condi-

tion. Similar to the spectrum that results in the Shock Pulse, Spike En-ergy, and PeakVue systems, we are looking or peaks, sidebands, andharmonics that are related to the our characteristic bearing requencies:Ball Pass Frequency Outer race (BPFO), Ball Pass Frequency Inner race(BPFI), Ball (or roller) Spin Frequency (BSF), and Fundamental Train (orcage) Frequency (FTF). See Figure 2 or a summary o the progressionwe expect to see.

Spectrum AnalysisI we do not use one o these techniques and simply view a spectrum,

then we may have limited success unless we take precautions:

1. Acceleration is most sensitive to high-requency vibration, so i we view the spectrum in units o acceleration (Gs or mm/s2) andhave a high Fmax (70X or higher) and, better yet, we view thespectrum in logarithmic ormat, then we will achieve the best re-sults (with a spectrum alone).

2. I we view the spectrum in units o velocity (in/sec or mm/s), thenwe may need to wait until the bearing is at stage three until wesee positive signs o the ault. Increasing the Fmax and viewingthe spectrum in logarithmic ormat will help signicantly.

When viewing the velocity or acceleration spectrum (or any spectrumrom PeakVue, enveloping, etc.) there are a ew techniques that help toachieve the best results:

1. Look or peaks at requencies that are non-integer multiples o the shat speed (e.g., 3.09X, 4.65X, 7.89X, etc.).

Detecting 

BearingFaults

 vibration

 ViB

 c o n d i t i o n

 m o n i t o r i n g 

Fig

Descripton

In this example, we have a

smooth vibraon signal with

the weak bearing signal over-

laid. In reality, the vibraon

would not be as smooth, and

the bearing vibraon would

not be visible.

First, the relavely high-

amplitude, lower-frequency

vibraon must be removed

from the accelerometer signal.

This is typically achieved with

a high-pass or band-pass

lter. The me between

these “pulses” is equal (in this

example) to the me between

balls rolling past the damaged

area on the outer race, which

represents the “ball pass

frequency” for the outer race:

BPFO.

The signal is then reced (all

of the negave-going vibraon

is turned into posive-going

vibraon).

The signal is then “enveloped”

(thus the name) via a low-pass

lter. There is a lot that could

be said about this process, and

there are dierences from one

vendor to the next, but the

end result is a low-frequency

me waveform that, whentransformed to a spectrum

(FFT), will indicate that a fault

exists on the outer race (in this

example).

Metal-to-metal contact setso a ripple eect: a stress waveraces through the metal compo-nents, causing the componentsto vibrate due to resonance.

    P   a   r   t    2

Jason Tranter

This article is the second in a

 series o our. The rst article

 provided a summary o how 

the vibration patterns change

as the bearing ails. Ater very 

briefy recapping the basics,

this article will discuss the

detection and analysis tools

that can be used to determine

the nature and severity o the

bearing ault: ultrasound,

Shock Pulse, PeakVue, envel-

oping, and spectrum and time

waveorm analysis.

7/29/2019 Detecting Bearing Faults, Part 2

http://slidepdf.com/reader/full/detecting-bearing-faults-part-2 2/2

60  april/may11

2. There should be harmonics o those requencies (e.g., peaks at3.09X, 6.18X, 9.27X, etc.).

3. Check or sidebands o the turning speed o the shat. I they exist,then suspect a ault on the inner race. I there are no sidebands,suspect an outer race ault.

4. Check or sidebands o the undamental train requency (slightlyless than hal the turning speed o the shat). I t hey exist, thensuspect a ault on the rollers/balls.

Time waveform analysisIt is typically possible to view the time waveorm rom the Shock Pulse,

PeakVue, and envelope process, but I’ll ocus on t he raw waveorm romthe accelerometer. In the early stages o the ault condition it will be verydifcult to detect the ault with a time waveorm. However, as the aultdevelops, an acceleration waveorm can reveal the ault, especially whentaken rom low-speed machinery. As the ault develops, the waveormwill have characteristic “pulses”and patterns that indicate the conditiono the bearing ault. In the later stages o the ault, a waveorm in velocityunits can display the deect quite clearly.

Characteristic “modulated”pattern inthe accelerationwaveorm

(otencalledthe “angelfsh”pattern).

“Spikes”in the velocity waveormin

ConclusionI hope this article has helped to pro

techniques. They have all been used bearing aults at a very early stage. rectly, choose the correct settings, an

Figure 2

Descripton Vsal

If the lubricaon is

OK and the bearing

is faultless, then

the envelope spec-

trum will contain

“noise” – no peaks

to speak of.

As a fault develops,

peaks will rise up

out of the noise

oor at frequencies

described in the

spectrum analysis

secon.

When the peaks

are clearly above

the noise oor the

bearing should be

replaced.

As the fault

develops further,

the noise oor will

begin to li.

When the noise

oor swallows the

peaks, the bearing

is in late stage four

and is about to fail.

The surface of the

bearing is severelydamaged; there is

no longer period

vibraon.

One-day Training Sessions:

• Vibration Control

• Gearbox Design Application and Analysis

National Technical Training Register Today

June 21-24, 2011The Menger Hotel

San Antonio, TX

For more information or to register:

Phone: 630.654.2254  E-mail: vibinst@att.net

Website: www.vibinst.org

34TH ANNUAL MEETING OF THE VIBR

Keyn

 A Tri

by M

 Additional sessions include:

• Pump Design • Vibration-based Nuc

• Pump Vibration • Machine Isolation

• Pump Case Histories • Permanent Continuo

• Pump Rotor Dynamics • Electrical Signal Ana

• FFT & Signal Processing • Electric Motor Vibrat

 Jason Tran

and autho

ing materiinvolved in

his native A

Mobius Ins

consulting

monitoring

Practical Plant Failure Analysis is a three-day, reasonably-priced, practical seminar forenginee rs and skilled plant personnel. In it weuse hundreds of failed shafts, bearings, gears,belts, chains, and corrosion examples. As partof the class, small groups do hands-on analysisof a series of pieces, diagnosing how and whythey failed, and how to prevent another.

Te next public session will be on September28th –30th in Syracuse, NY. Private plantsessions range from two to five days and can beheld at your site.

Training from the ReliabilityProfessionals who “wrote the book on

practical failure analysis.”

For more details about f ailure analysisor training sessions, contact Dale Gamba

at 315-487-4390 or email us atreliable@twcny.rr.com

Sachs, Salvaterra & Associates, Inc6171 Airport Road

Syracuse, NY 13209

Lots of people talk about doing “Root CauseFailure Analysis”. With hundreds of hands-on

examples, we show you how to do it!

 

: :

Te next public session will be on September20th-22nd, 2011 in Syracuse, NY. Privateplant sessions range from two to five daysand can be held at your site.

Receive the“FREE”bi-weeklynewsletter of ultrasonic stories,applications, and case histories

as told by Jim Hall ofUltra-Sound Technologies

Ultra-Sound Technologies forUltrasound Level I & II Training

www.Ultra-soundtech.com