Machinery Vibration Analysis

Category (Level 1)
 Course Contents
Chapter 1: Introduction
Chapter 2: Vibration Basics
2.1‐ What is vibration
2.2‐ Frequency and period
2.3‐ Time waveform (time domain)
2.4‐ Describing vibration (peak, peak to peak & RMS)
2.5‐ Vibration measurement units
2.6‐ Displacement‐velocity‐acceleration relation ship
2.7‐ Spectrum (frequency domain)
2.8‐ Introducing phase
2.9‐ Natural Frequency and resonance
Chapter 3: Understanding signals
3.1‐ Pure sine wave
3.2‐ Summation of waves & beating
3.3‐ Single Impulse
3.4‐ Pulse train
3.5‐ Square wave
3.6‐ Clipped or truncated wave
3.7‐ Synchronous impacting
3.8‐ Random impacting
3.9‐ Modulation
3.10‐ Beating versus modulation
Chapter 4: Data acquisition
4.1‐ Vibration monitoring system online/offline
4.2‐ vibration sensors
4.3‐ Selecting a transducer
4.4‐ Survey the machine
4.5‐ Accelerometer mounting
4.6‐ Measurement locations
4.7‐ Repeatability
4.8‐ Recognizing bad data

Chapter 5: Signal Processing

5.1‐ Understanding Analog & digital signals.
5.2‐ Sampling & processing signal
5.3‐ Filters
5.4‐ Analog digital converter
5.5‐ Sampling & Aliasing
5.6‐ Frequency domain
5.7‐ Windowing
5.8‐ Averaging
5.19‐ Spectral integration
5.10‐ Guide for data collection settings for routine monitoring
5.11‐ Overall readings
Chapter 6: Analysis Process
6.1‐ Introduction about vibration analysis
6.2‐ Vibration monitoring and analysis concepts
6.3‐ Vibration analysis techniques
6.4‐ Types of vibration data‐ based on machine state
6.5‐ Steady state vibration plot formats
6.6‐ Spectrum analysis for steady state data
6.7‐ Time waveform analysis for steady state data

Chapter 7: Fault Diagnosis

7.1‐ Diagnosing resonance
7.2‐ Diagnosing unbalance
7.3‐ Diagnosing eccentric pulleys
7.4‐ Diagnosing misalignment
7.5‐ Diagnosing cocked bearing
7.6‐ Diagnosing soft foot
7.7‐ Diagnosing bent shaft
7.8‐ Diagnosing looseness
7.9‐ Diagnosing rolling element bearing faults
7.10‐ Diagnosing gearbox faults
7.11‐ Diagnosing belt drive faults
7.12‐Diagnosing electric induction motor faults
7.13‐ Diagnosing Hydraulic & Aerodynamic faults in Pumps‐Fans‐Compressors
Chapter 8: Alarm Settings & Acceptance Criteria
8.1‐ Alarm Types
8.2‐ Acceptance Criteria‐Acceptance Limits

Chapter 9: Condition Monitoring Program

9.1‐ Equipment monitoring deployment process
9.2‐ Monitoring (Measurement ) frequency
9.3‐ Condition Monitoring reports
Chapter 1. Introduction
1.1 Condition monitoring:

Condition Monitoring is defined as the measurement and analysis of a

number of equipment parameters, which are used to determine the current
health of the equipment and thus the correct timing and extent of
equipment maintenance (Condition based Maintenance CBM-Predictive
Maintenance PDM), there are various parameters which are known to be
sensitive to the change in equipment condition like vibration, temperature,
wear particles in lubrication and hydraulic oil, performance & process
parameters, electric motor current, acoustic energy (ultrasound). The
change in the trended values of these monitored parameters means that a
primary failure starts to occur, if left unattended this failure will result in
costly secondary failures.
1.2 Condition monitoring Techniques:

Condition monitoring utilizes combination of techniques to detect

change in condition of the machine
The following are examples of condition monitoring techniques:
 Vibrating Monitoring
 Lube Oil Monitoring
 Thermography (thermal imaging)
 Performance Monitoring
 Electrical Signature Analysis (ESA) or Motor current
Signature Analysis MCSA
 Ultrasonic Monitoring (Acoustic Emission Monitoring)
1.3 Condition Monitoring and CBM Benefits:

 Detecting faults in an early stage and the recommended

maintenance action would then be scheduled and implemented
at convenient time according to the production demands
 Avoid sudden failures
 Avoid Infant Mortalities (Failures After overhauls)
 Avoid consequent secondary failures
 Increase machinery reliability
 Reduce Maintenance costs
 Reduce risks of catastrophic failure
1.4 Vibration Monitoring-Vibration Analysis:

This is the most commonly used and effective condition

monitoring technique to detect a wide range of machinery
problems. Vibration measurement is an effective method to
monitor machine condition during startups, shut downs and
normal operation
Chapter 2. Vibration Basics
2.1 What is vibration?
 It is the unwanted movement of a machine or a machine part
back and forth from its neutral position (position at rest)
 It is the unwanted machines motion resulting from the dynamic
forces that are created by machinery problems (Exciting Forces).
 These dynamic forces could be Imbalance, Misalignment,
bearing faults, gear faults, vane/blade passing rate ….
 Vibration pattern depends on the type of machinery fault
 The magnitude of vibration depend on the value of the exciting
force together with the machine characteristics like mass,
stiffness and damping.
 Stiffness: Is defined as the force per unit deflection or the force
required to make unit deflection, it is function of the elasticity of
the material used in manufacturing the machine components
Stiffness = force (lb)/deflection or displacement(inch)
= lb/inch or N/mm

 Damping: Is defined as the ability of the system to dissipate

vibration energy. It is proportional to velocity.
Damping = force(lb)/velocity (inch/sec)
= lb/inch/sec or N/mm/s
Free Undamped Vibration

Free Damped Vibration

 T

2.2 Frequency and Period

The frequency (F)of an event: is the number of occurrence of this event

in a defined period of time
Frequency units: cycles per second(Hz), cycles per minute (CPM),
orders of running speed EX: 1X or 2X
Ex: If the machine running speed was 1500 RPM and the vibration
frequency was 50 HZ (3000 CPM), then the frequency=2 order of
running speed=2X
The period (T): is the duration of one cycle
Period units: Seconds or milliseconds

1
T=
F
 2.3 Time Waveform (Time Domain)

Amplitude
+
Amplitude of vibration are
described in terms of
• Displacement
• Velocity
• Acceleration 0
 DISPLACEMENT - How far it moves
Time
(Microns or Mils)

 VELOCITY - How fast it moves -

(mm/sec or in/sec)

 ACCELERATION - How quickly velocity Period

changes 1 Cycle
(g’s or m/s2 or mm/sec2 or in/sec2)
2.4 Describing Vibration (Peak, Peak to Peak &RMS)

There Amplitude of vibration can be expressed in several ways:

• RMS (root mean Square)
• PK (peak)
• PK‐PK (peak to Peak)
• Average
For Sinusoidal (Pure Sine wave )only:
 RMS = PK/ √2 = PK/1.414 = Pk x 0.707
 PK‐PK= 2x PK
 Average value = 0.637 x PK
• Peak to Peak (PK to Pk) or (Pk‐PK): The peak to peak amplitude is the
amount measured from the bottom of the trough to the top of the peak
• Peak (PK): The peak is the amount from the zero line to the top of the peak
(or the bottom of the trough, which ever happens to be greater
• Root Mean Square (RMS): The RMS value represents the area under the
curve of the time wave form ( effective energy content in the signal), the
RMS is measured from TWF by Analogue method by using analogue
circuitry that solves the RMS integral equation (in previous slide), Another
measurement method from TWF is digital which is actually reverse of its
name i.e. the values of all samples in TWF are squared then averaged then
the square root is obtained. Also RMS overall value is some times
calculated digitally by root sum square of all spectrum lines (called FFT
overall or digital overall, will be expressed later in signal processing section)
• Average Value: it is the average value of the absolute value of the wave
form. This term is seldom used in vibration today although it is available in
some vibration meters
For Complex Wave (Non Sinusoidal):
Crest Factor: The crest factor of a wave form is the ratio of the
peak value to the RMS value, it is some times called the peak
to rms ratio

 For Sinusoidal Wave: Crest factor= Pk/RMS=1.4

 For Non Sinusoidal: Crest factor=True Pk/ RMS ; True Pk is

the Max PK in the time wave form

 Increase in crest factor means increase of the ratio between

the peak and rms value which indicates that there is spikes
or impacts in the wave form (Ex bearing fault)
2.5 Vibration Measurement Units:
Conventionally, vibration (Vibration Amplitude) measurement units are:

 Displacement: microns µm Pk‐Pk or Mils Pk‐Pk ; 1 Mil=25.4 µm

 Velocity: (mm/s) peak or (mm/s) rms or (in/s) pk or (in/s) rms

 Acceleration: (mm/s2) Pk or (mm/s2 )rms or (g’s)rms or (g’s )pk

 Velocity and acceleration are measured in RMS or PK values

 Displacement measured in PK‐PK values

2.6 Displacement, Velocity, Acceleration Relationship

Displacement X = D sin Ѡ t
.
Velocity X = D Ѡ cos Ѡ t
..
Acceleration X = ‐ D Ѡ2 sin Ѡ t

0 90 180 270 360

Acceleration Velocity Displacement
0 90 180 270 360
Acceleration Velocity Displacement

 Acceleration signal is leading velocity signal by 90 Deg

 Velocity signal is leading displacement signal by 90 Deg
 Phase shift between Acceleration and velocity is 90 Deg
 Phase shift between velocity and displacement is 90 Deg
 Phase shift between Acceleration and displacement is 180
Deg
 Data collectors can integrate acceleration signal to get velocity signal and
can integrate velocity signal to get displacement signal or double integrate
acceleration to get the displacement, but usually the reverse is not done by
the data collector. (may be in the analysis software for the spectrum units‐
Spectrum discussed later).

In Data collectors
 Acceleration is measured by Accelerometer

 Velocity can be measured by Accelerometer or velocity transducer

 Displacement can be measured by accelerometer, velocity transducer, or

displacement probe

 Accelerometer and velocity transducers are installed on the external casing

and can only measure absolute vibration.

 Displacement probes are inserted in the fluid film bearing and can only
measure the relative vibration between the shaft and the casing (bearing and
its journal)
 The frequency range of interest
Acceleration
(high freq.)

Velocity
(general purpose)

Displacement
(low freq)
 Displacement: Displacement is used as the measure for low
frequency vibration less than 20 HZ (120 CPM) on bearing caps,
structures. The exception is that for fluid film bearing machines the
displacement is recommended if displacement probe installed and in
this case displacement is used with higher frequencies, for Example:
4000 Hz‐240 000 CPM).
 Velocity: velocity is used as the measure for machine casing vibration
in the frequency range 10Hz (600 CPM)‐1000 Hz (60 000 CPM), In
general velocity is a good measure of vibration across most machine
speeds and frequencies of interest. Velocity measure can be used for
rolling element bearing machines or for fluid film bearing machines
that do not have displacement probes installed.
 Acceleration: Acceleration is used as the measure for machine casing
vibration at frequencies greater than 1000 Hz (60 000 CPM),
Acceleration used in high frequency analysis such as bearing and gear
box analysis.
 Low Speed Machines below 600 RPM: Displacement measurement is
commonly used

 Machine Speed 1000‐10 000 RPM: velocity measurement is

commonly used

 High Speed Machines greater than 10 000 RPM: Acceleration

measurement is commonly used.
Continue: Displacement, Velocity, Acceleration Relationship
For Pure Sine Wave
Displacement X = D sin Ѡ t
.
Velocity X = D Ѡ cos Ѡ t
..
Acceleration X = ‐ D Ѡ2 sin Ѡ t

Displacement Peak Amplitude = D

Velocity Peak Amplitude D 2

Acceleration Peak Amplitude = D 4 2 2

PK = √2 RMS =1.414 RMS & PK‐PK= 2PK

;Ѡ 2
; Ѡ in rad/s
; in Hz
Example1:
Convert Displacement of 2 mils PK‐PK @ 1775 CPM to i) velocity in mm/s rms and to
ii) velocity in IPS Pk
 Frequency 1775/60 = 29.58 Hz
 Displ=2 mils Pk‐Pk= 1 mil Pk= 25.4 µm Pk =25.4/1000 000 m Pk
 Vel=Displ x (2 )= (25.4/1000 000) x (2 x 3.14 x 29.58)=0.00472 m/s Pk
Vel= 0.00472 m/s Pk = 4.72 mm/s Pk = (4.72 /1.414) mm/s rms = 3.34 mm/s rms
Vel= 4.72 mm/s Pk = (4.72/25.4) in/s Pk = 0.186 in/s Pk

Example2:
Convert Acceleration of 1 g rms@ 1500 CPM to i) velocity in IPS Pk and to ii)
Displacement in mils Pk‐Pk
 Frequency 1500/60 = 25 Hz
 Accel=1 g rms = 1.414 g Pk= (1.414x 9.81) m/s2 Pk =13.87 m/s2 Pk
 Vel=Accel / (2 )= (13.87) /(2 x 3.14 x 25)=0.08830 m/s Pk = 88.3 mm/s Pk
Vel= (88.3/25.4) in/s Pk = 3.48 IPS Pk
 Displ= Vel/ (2 )= 0.08830/ (2 x 3.14 x 25) m Pk = 0.0005621 m pk = 562 µm Pk
Displ= 562 µm Pk = (562 /25.4) mils PK = 22.13 mils Pk = (22.13 x 2) mils Pk‐PK
Displ= 44.26 mils Pk‐Pk
2.7 Spectrum (Frequency
Domain) (FFT) Amplitude

Frequency
Time
Amplitude

Amplitude
Amplitude

Time
Spectrum (Frequency Domain) (FFT)

The FFT (Fast Fourier Transform) process separates the individual

sine waves from the composite (complex) wave form and displays
them according to their frequencies, the vibration spectrum is as if
we are looking at those separated waveforms from the end.
This mathematical function performs a transposition of the signal
from the time domain to the frequency domain. The signal
representation thus obtained is called a frequency spectrum.

• Amplitude
• Frequency
• Phase
 COMPARISON OF FREQUENCY AND TIME DOMAINS

Simple Wave forms

Complex Waveform

FMAX
Amplitude
9X
5X
3X Frequency
1X
Spectrum
Plot
Time Domain
(Sec or Min) Frequency Domain
TMAX (CPM or Hertz)
ALARM LEVEL = 0.1 IN/SEC

PEAK - RMS
OVERALL VALUE
Important Terms & Definitions used in vibration Spectrum Analysis
 Synchronous & non Synchronous Energy in the spectrum
 Harmonics
 Modulation in TWF and Side bands in the spectrum
 Maximum frequency (fmax) in the spectrum & Spectrum Resolution

All the energy in a spectrum can be grouped into one of three categories:
 Synchronous Energy: Energy that is an integer (Whole number) multiple of
running speed EX: 1X, 2X, 3X, 4X, 5X
Ex: imbalance, misalignment, bent shaft, Gear meshing faults
 Non Synchronous Energy: Energy that is a fractional number of running speed
EX: 1.1X, 2.5X, 3.9X
Ex: Roller Element Bearing defects like Inner race, outer race and ball defects
 Sub Synchronous Energy: Energy that is below running speed EX: 0.48X, 0.5X,
0.6X
Ex: Oil whirl, rubbing (in sleeve bearings), belt defects & roller brg Cage Freq.

Harmonics: harmonics are a series of evenly spaced peaks. Their amplitude

could vary, but the peaks will be spaced at integer multiples of the first in the
series (Fundamental Peak).
Synchronous, Non Synchronous & Sub‐synchronous

Non
Synchronous
Peaks
Synchronous
Peaks
 Sub‐
Synchronous
Peaks
Synchronous
Peaks
Harmonics In Spectrum

Fundamental Harmonics from

Peak (BPOR) BPOR
 Modulation in TWF & Side Bands In Spectrum

 Transform of specific signals: Amplitude-modulated signal.

T=1/Fm The number of sidebands

Modulating depends on the shape of the
signal modulation signal.
Freq. Fm F
Fm Fm
FFT

T=1/F
Carrier
Freq. F
Modulation in TWF & Side Bands In Spectrum ‐Data of Pump With
Bearing BPIR Defect (Pump Speed = 3000 RPM)
Modulation in TWF & Side Bands In Spectrum ‐Data of Pump With
Bearing BPIR Defect (Pump Speed = 3000 RPM)

Side Bands
Carrier
Freq.=BPIR
Maximum frequency (fmax) in the spectrum & Spectrum Resolution
Some forcing frequencies associated with machines:
Source Frequency (Multiple of RPM)

Fault Induced

Mass Imblalance 1X

Misalignment 1X, 2X

Bent Shaft 1X

Mechanical Looseness 1X, 2X, 3X,..

Structural Looseness (Flexibility) 1X

Rotor Rub (Fluid Film Bearing) 0.5X

Antifriction Bearings Bearing frequencies (not Integer ones)

Design Induced

Universal Joints 2X

Asymmetric Shaft 2X

Gear Mesh (n teeth) nX

Coupling (m Jaws) mX

Fluid Film Bearings (oil Whirl) 0.39X ‐ 0.49X

Blades and Vanes (m) mX

Reciprocating Machines Half and Full Multiples of speed, depending on design

Maximum frequency (fmax) in the spectrum & Spectrum Resolution

 The spectrum frequency Span fmax is selected based on machine RPM and other
machine frequencies (see table in the previous slide for forcing frequencies)

 The selected spectrum No of lines of resolution should allow sufficient resolution to

analyze the operating speed frequency, synchronous and non synchronous energy,
side bands and harmonics

 Resolution= fmax/No of Spectra lines or

Resolution= (fmax/No of Spectra lines) X window factor (will be expressed later)
Resolution & fmax, are related to sampling and sampling rate of TWF, (will be
expressed later)

 If adequate resolution is not available, multiple data measures for the same point
with different frequency spans must be acquired, it is some times desirable to split
the data point into two or three frequency spans
Maximum frequency (fmax) in the spectrum(Default Freq. Span)

Component Frequency Span (fmax)

Shaft Vibration 10 x RPM

Gear Box 3 x GM

Rolling Element Bearings 10 x BPI

Pumps 3 x Vane Pass Freq.

Motor / Generators 3 x 2Lf

Fans 3 x Blade Pass Freq.

Sleeve (journal) Bearings 10 x RPM

Remember
As mentioned before about the Root Mean Square, The RMS value represents
the area under the curve of the time wave form ( effective energy content in
the signal), the RMS is measured from TWF by Analogue method by using
analogue circuitry that solves the RMS integral equation. Another
measurement method from TWF is digital which is actually reverse of its name
i.e. the values of all samples in TWF are squared then averaged then the
square root is obtained. Also RMS value is some times calculated digitally by
root sum square of all spectrum lines (called FFT overall or digital overall) but
this is not True RMS and its value may differ from the true RMS obtained from
the time wave form because it depends on fmax selected for the spectrum.
Remember: Acceleration, Velocity & Displacement application based on Freq. range

F= 159 Hz
1 g
10 mm/s
10 µm Acc.

Acc. Vel.
Vel.
Displ.
Displ.

100 1000 10000 F (Hz)

Remember

Integration

Integration
reduces high
frequency
vibration and
amplifies low
frequency
vibration, and
introduces 90
Deg Phase
change
2.8 Introducing Phase

 Phase is an excellent vibration analysis tool, phase is used often in

diagnosing machinery problems, it is specially helpful in determining
imbalance, misalignment, structural looseness, resonance conditions and
others. Of course it is also used in balancing machinery
 Phase is the time relationship, measured in degrees, between vibrations of
the same frequency.
 Phase is a measure of the time difference between two sine waves.
 The time difference is calculated from the peak of one sine wave to the
peak of another
 Phase could be compared only for vibrations of the same frequency and
same units (accel to accel or vel to vel or displ to displ).
 Even though phase is truly a time difference, it is almost measured in terms
of angle (from 0 – 360 Deg)
 The phase difference between two waves forms is often called phase shift.
 Phase is a measure of the time difference between two sine
waves.
 The time difference is calculated from the peak of one sine
wave to the peak of another
Types of Phase:

Relative Phase: is the difference in time between the same event in two
signals. The two signals are the same frequency and the same events are
when they both reach their maximum.
Relative phase measurement are used to identify the phase shift between the
vibration signals of two locations on the machine (for example DE and NDE or
horizontal and vertical positions or across coupling of machine.
Relative Phase is measured by dual channel Analyzer or single channel and
tachometer to produce reference signal and measuring the absolute phase by
moving with the vibration sensor from one location to another and then
subtract the phase readings at different locations to get the relative phase
between these locations

Absolute Phase: it compares one signal to a fixed reference signal. The

reference signal is usually Tachometer pulse signal that fires one per
revolution. The phase reading is between the peak of the vibration signal and
the tach (tachometer) signal.
Relative (Cross Phase)
Relative (Cross Phase)
Amplitude - A Relative (Cross Phase) Amplitude - A
+ +

0 Time 0 Time

- -
Amplitude - B Amplitude - B
+ +

0 Time 0 Time

- -
In Phase: 180 Out of Phase:
o

A B A B
Absolute Phase
Absolute Phase

Reference (Trigger) Signal

 Strobe light ( stroboscope). ( Not recommended – Not accurate)
 Laser phase sensor (lasertach or laser tachometer).
 Optical phase sensor (photo-cell or phototach.,optical
tachometer).
 Key phasor on-line systems Non contact pickup NCPU.
Absolute Phase

Example: In the above wave form, the time between pulses is the period of
the reference signal (period of machine Speed), if the machine speed was
1500 RPM, which is 25 Hz, the time between pulses (period T) will be 1/25
sec = 0.04 sec, if there is 0.01 sec between the reference pulse and the next
maximum positive peak of the wave, then there is phase shift equivalent to
quarter the period or quarter cycle of the wave, then the phase shift will be:
Phase= (0.01/0.04) x 360 = 90◦
Absolute Phase

Phase is measured between the tachometer pulse and the next positive
peak in the vibration signal
Example: In the above wave form, the phase reading was 150◦
Phase= 150◦
Absolute Phase

Phase is measured between the tachometer pulse and the next positive peak in
the vibration signal
Example: In the above wave form, the phase reading was 340◦
Phase= 340◦
 Absolute Phase: True & Reversed Phase

True Phase= 150 Deg True Phase= 150

Reverse Phase=210 Reverse Phase=210

True Phase is measured from the tachometer pulse to the next positive peak in
the vibration signal
Reverse Phase is measured from the positive peak to the next tachometer pulse
 Absolute Phase: True & Reversed Phase

 True Phase is measured from the tachometer pulse to the next

positive peak in the vibration signal. (That’s why called Phase Lag)

 Reverse Phase is measured from the positive peak to the next

tachometer pulse.

 Most Vibration data collectors report true phase.

 If Strobe light is used in measuring phase , it provides reverse phase

(but as mentioned before the strobe is not accurate and not
recommended in measuring phase).
2.9 Natural Frequency & Resonance

 Natural Frequency:
The frequency of free vibration of a system.
the frequency at which a part likes to vibrate.
 Natural frequencies are frequencies at which a system
(machine or structure) will absorb energy easily leading to
build up of heat, motion, and stresses causing failure.

 For a simple mass and spring system (No

damping) the natural freq:
fn=(1/2 ) √(K/M) Hz
Ѡ n= √(K/M) rad/s
; K is the stiffness N/mm
 Natural frequency is a property of a system and is dependent on:
1.Mass
2.Siffness
3.Damping
 Increasing Mass moves the natural frequency to a lower frequency
 Increasing stiffness moves the natural frequency to a higher frequency
 Increasing damping does not move the natural frequency (does not
move it much-very slight) but acts as an absorber and lowers the
amplitude
 In reality, each machine, rotor or structure has several natural
frequencies

Bump
Test
Mode Shapes
 Mode Shapes: Mode shapes of a system are associated with its
natural frequency. The shape of the system as it vibrates at a natural
frequency is called its mode shape.
 Mode Shapes For Cantiliver Beam with one Fixed end

Mode Shapes For a rotor

Resonance
 Resonance: is the act of exciting the natural frequency.
 If a forcing frequency occurs at or close to a natural frequency
resonance exists.
 If there is impacting, the impacting inject energy at all frequencies or
at broad band of frequencies which might excite natural frequency.
 Resonances amplify vibration, the measured vibration level may be 3
to 50 times higher than they would be normally, so instead of
vibrating at 0.5 mm/s, for example, the machine could vibrate at up
to 25 mm/s.
Critical Speed
 Critical Speed: Critical speed is often defined simply as the speed
that excites a resonance. if 1X (running speed frequency) forcing
frequency excites a natural frequency, this speed is termed a critical
speed, and you should not operate at this speed, Otherwise the
machine will resonate.
 The term critical speed is typically used regarding large rotors such
as turbine rotors.
 Vibration increases dramatically near and at critical speed. As these
machines are run up to their operating speed they must run through
the critical speeds quickly to avoid catastrophic damage.
 The machine should not operate with in 20 % of a critical speed.
Chapter 3. Understanding Signals
3.1 Pure Sine Wave
Pure Sine wave is a single frequency producing one peak in the spectrum
3.2 Summation Of waves
Cont. Summation Of waves
Beating
Beats are seen when two equipment are installed beside each other or on the
same skid and operating at slight difference RPM, their signals combine
together to form beating.

Two Different Frequencies

Different Frequencies
combine to form beats

Beat frequency products

contain lower frequency
components than the
original frequencies
Cont. Summation Of waves
Beating
 Some times beating produce sum and difference frequencies in addition to
the two close frequencies
 Beat frequency = difference between the two close frequencies.
 If the spectrum resolution was not high the two close frequencies may
appear as one peak but this peak may be seen rising and falling while
viewing the spectrum life due to beating.
Cont. Summation Of waves
Beating
Cont. Summation Of waves
Summation 1X, 2X & 3X
Note the formation of an M W shape in the wave form due to summation of 1X
with 2X or 3X component, this appears in the time wave form of misaligned
equipment due to presence of 1, 2, 3 X vibration components components

1X and 2X
components
Summation

1X and 3X
components
Summation
 Cont. Summation Of waves
Summation
Misalignment Example

TWF showed M W
shape due to
presence of 1X, 2X, 3X
component

Spectrum
showed 1X, 2X,
3X component
presence of
3.3 Single Impulse
Single impulse will inject energy at all or broad band frequencies increasing
noise floor in all the spectrum with much more increase (humps) at regions
where the natural frequencies exist, an application of this is the Bump test to
identify natural frequencies.

Natural
Spectrum For Frequencies
Single Impulse
or Bump Test

Natural
Spectrum For Frequencies
Single Impulse
or Bump Test
 3.4 Pulse Train

TWF showed
Pulse train at
frequency =
200 Hz

Spectrum time
showed
Fundamental
Peak at
frequency =
200 HZ and
series of
harmonics from
the 200 Hz
 TWF showed
Pulse train at
frequency =
200 Hz=12000
CPM
Time Sec 0.05 0.1

Spectrum
showed
Fundamental
Peak at
frequency =
12000 CPM=
200 HZ and
series of
harmonics from
the 12000 CPM
 3.5 Square Wave
A square wave will produce in the spectrum its fundamental frequency in
addition to odd harmonics from the fundamental frequency

Amplitude

Time

Amplitude

Frequency‐Orders
3.6 Clipped or Truncated Wave
A truncated wave form will produce all integer harmonics in the spectrum, not
just the odd (As square wave),a wave will be clipped or truncated due to two
reasons:
 There is restriction in motion in one direction and motion free in the other
direction and this sometimes happens when there is looseness in the
machine, or rotor rubbing.
 The second reason if the dynamic range (amplitude range was not adjusted
correctly causing this distortion in the wave form, so its recommended to
adjust the dynamic range (Amplitude range) of data collector on the auto
ranging setting to avoid this distortion which may make cause confusion.
Cont. Clipped or Truncated Wave
 3.7 Synchronous‐Impacting Synchronous Impacting

Base Mounting
Looseness Example
Synchronous
Impacting in time
wave form due base
mounting looseness

Spectrum showed strong

Harmonics from 1X up to
30 order due base
mounting looseness, In
severe looseness cases
there will be sub order
harmonics 1/3 or 1 /2 X
sub harmonics
 3.8 Random‐Impacting
Pump Cavitation Example

Random Impacting
in time wave form
due to cavitation

Spectrum showed
increased noise floor
due to random
impacting from pump
cavitation
Cont. Random‐Impacting

Spectrum showed
increased noise floor
due to random
impacting from pump
cavitation
Cont. Random‐Impacting
3.9 Modulation

Two types of modulation sometimes happen in vibration:

 Amplitude Modulation: This happens when there is a fault with certain
forcing frequency called carrier frequency and the amplitude of vibration
due to this fault is increasing and decreasing periodically at lower frequency
called Modulated frequency as the component defect enters and exit from
loading zone.

Example: A bearing with an inner race defect (notch) the carrier frequency
will be BPIR (Ball pass inner race) frequency and the amplitude will increase
gradually as the inner race rotate and the notch approaches the loaded zone
and then will decrease gradually as the notch becomes far away from the
loaded zone, the following periodic amplitude increase and decrease will be
at modulated frequency which is equal to the rotational speed of the
bearing inner race which is actually the machine or shaft speed RPM. In this
case we will have two frequency components, Carrier frequency which is
the higher frequency at BPIR and Modulated frequency which is the lower
frequency at shaft speed.
Cont. Amplitude Modulation
 Transform of specific signals: Amplitude-modulated signal.

T=1/Fm The number of sidebands

Modulating depends on the shape of the
signal modulation signal.
Freq. Fm F
Fm Fm
FFT

T=1/F
Carrier
Freq. F
Cont. Amplitude Modulation

Carrier
Side Bands Freq.=BPIR
Cont. types of modulation

 Frequency Modulation This happens when the machine speed changes

periodically (due to for example cyclic process load increase) causing the
forcing frequency to change also periodically, in this case the carrier
frequency (higher frequency) will be the forcing frequency and the
modulated frequency (lower frequency) will be the frequency of periodic
change of speed or load.
Example: A machine with gear drive its speed is changing periodically
due to cyclic change in load causing the gear meshing frequency to
increase and decrease periodically, in this case we will have two
frequency component carrier frequency is the gear meshing frequency
(higher frequency) and modulated frequency which is the rate of change
of speed or cyclic load (lower frequency).
 Cont. Frequency Modulation
 Transform of specific signals: Frequency-modulated signal.
Frequency Modulation Period
Tm F
Fm Fm

FFT

Note: Spectrum of frequency modulation is quite similar to the spectrum of

amplitude modulation except that in frequency modulation there might be
much more side bands
3.10 Beating Versus Amplitude Modulation
Beating Beat Frequency = 2.1 Hz, period = 0.48 Sec

Amplitude Modulation Modulated Frequency= 49.2 Hz, Period = 0.02 Sec

 Cont. Beating Versus Amplitude Modulation
 Beating Frequency is very small because it is the difference between two
frequencies very close to each other (Beating often not more than 4 Hz),
modulated frequency is at higher frequencies.
 Beating has a long period in the wave form compared to the modulation
period.
 Beating in some cases may lead to amplitude near to zero at the end of
each beat cycle (if the magnitude of the two close frequencies are very
near, when they are 180 out of phase they cancel each other).
 Beating does not produce side bands in the spectrum.
Chapter 4. Data Acquisition
 4.1 Vibration Monitoring System Condition Monitoring Analysis
Software ‐ Data base Management
I ‐ Online Monitoring System
Software
II – Offline Monitoring System

Portable data collector(FFT Analyzer)

Online Protection & Monitoring Rack

Protection Alarm
Permanent
Sensor or Shut down
Installed
Signal sent to
Sensor
Machine control
system

General Equipment Critical Equipment

Cont. Vibration Monitoring System
I ‐ Online Monitoring System:
The online monitoring system is permanently connected to the measurement
sensors, Intervals between measurements (sampling rate) can be short and can be
considered as continuous thus a number of equipment parameters are automatically
measured, sampled, analyzed and stored in a database (Data base Management
program, which is known as condition monitoring software, is used). The sampling
can be conducted on a time basis, an alarm basis, when predetermined change in a
measured value is detected and during transient conditions (run up, coast down,
change in speed...).The measured parameters are analyzed and stored in the data
base so that on a comparison basis a trend line of the machine condition can be
established.

II – Offline Monitoring System:

Offline monitoring system differs from online monitoring in that the parameters
are measured on an intermittent time basis and not collected automatically.
Typically Offline system consists of portable data collectors and a central data
processing and storage system (Data base Management program which is known
as Condition monitoring software), this could be the same software as used for the
online condition monitoring system although this is not mandatory.
Cont. Vibration Monitoring System

Vibration Monitoring & Protection Racks

 Monitoring or Protection: Can be used for monitoring and provide analysis

data for the condition monitoring software or can be used for protection only
without connection to any condition monitoring software and generate alarm or
protection SD signals to be sent to the control system to shut down the machine
if there is dangerous vibration.

 BNC output: Most of the Online Monitoring racks have Buffer out put (BNC
terminals) to enable connection of portable data collectors(For offline data
collection). The BNC’s signals in most cases are the raw signals before any data
processing by the rack.
Online Monitoring System Layout
Online Protection System Layout
Offline Monitoring

Permanent Installed Sensors if no access for measure

Cont. Offline Monitoring

 Old vibration data collectors (analyzers) was

single channel
 Today analyzers are dual or multichannel
analyzers

Dual Channel
Analyzer Dual Channel
Analyzer
Cont. Offline Monitoring
 On Route Measurement
 Off Route Measurement

 On Route Measurement: The Route is a list of machines that you select from your
machine hierarchy (defined on your condition monitoring software ‐database),
this list is uploaded to your data collector, which will route you during vibration
measurement (data acquisition) from one location to next one for example :
Motor1 Non Drive End MNDE H, MNDE V, MNDE X, Motor1 Drive End MDEH,
MDEV, MDE X, Pump1 Drive End PDEH, PDEV, PDEX, Pump1 Non Drive End PNDE
H, PNDEV, PNDE X, and then Motor 2 Non Drive End,…….., and so on. After data
collection you connect the data collector to the condition monitoring software
(data base) and then download all vibration data measured on this route (machine
list) to the condition monitoring software.

 Off route Measurement: All data collectors support this option, that if you need
to take extra vibration measurement for extra locations (not on your route) or the
same machine locations which included in your route, but you need to add some
measurements with different measurement settings, in these cases you will use
the off route option and collect the required data and then will be downloaded to
your software or data base under the name of off route data.
Multi Channel Analyzers
4.2 Vibration Sensors

 Vibration sensor also called vibration probe or vibration transducer or vibration

pickup, they are all the same name

 There are three types of vibration sensors:

Accelerometer

Velocity Transducers (Same Name: Velometer)

Displacement transducers (Same Names: Proximity Probe, Non Contacting Probe,

Eddy Current Probe)

 The sensor is selected based on machine type and machine speed

Cont. Vibration Sensors:

Vibration Sensors

 Accelerometer:
Measures the acceleration vibration, but the acceleration signal can be
integrated by the data collector to obtain velocity signal or double integrated by
the data collector to obtain displacement signal

 Velocity Transducer:
Measures the velocity vibration, but the velocity signal can be integrated by the
data collector to obtain displacement signal.

 Displacement Transducer
Measures the displacement vibration only.

Note: Displacement transducers can not be used to measure velocity or acceleration,

also velocity transducers can not be used to measure acceleration, because the data
collectors can not differentiate the signals they can only integrate the signal.
(conversion for spectrum units may be done in the analysis software for the
spectrum units).
Cont. Vibration Sensors:

Absolute (Same name : Casing‐Or Seismic) Vibration:

Accelerometer and velocity transducers measure the casing (seismic)

vibration of the machine.

Relative Vibration:

Displacement probes (Non contacting probe) measure the vibration of

the shaft or the rotor relative to the fluid film bearing or the casing. The
non contacting probe are inserted in the bearing.
4.2.1 Displacement Probe (Proximity or Non Contacting or
Eddy current Probe)

 Displacement transducers measure the relative movement between the shaft and
the tip of the sensor, they are drilled into the journal or sleeve bearing.
Therefore, they are permanently mounted
 Proximity probes measure distance
 Between Probe and shaft
 Non contacting
 Magnetic energy absorbed proportional to distance
 The most commonly used proximity probes measure distances between the
probe tip and the shaft over an 80 mil range and change in distance cause
changes in output volts dc – Most commonly Sensitivity 200mv/mil
(7.87mv/µm).
 The most commonly used proximity probes measure distances between the
probe tip and the shaft over an 80 mil range and change in distance cause
changes in output volts dc – Most commonly Sensitivity 200mv/mil
(7.87mv/µm).
 When installed proximity probe must be gapped properly, In most
radial vibration application, adjusting the gap of the transducer to the
centre of the linear range is adequate. Proximity probe gap volts during
installation are set to about -8 to -10 volts, this voltage setting will place
the probe nearly in the middle of its linear range, and will allow the probe
to sense movement in the positive and in the negative directions, (i.e -9
volts creates a clearance between the probe and shaft about 45 mils or1.1
mm)
 There are three components a driver (sometimes called proximiter), a probe and a
cable between them, a voltage is supplied to the driver that produces an RF signal.
That signal is transmitted through the cable to the probe which acts as an antenna and
radiates the high frequency energy into the gap, it sets up a magnetic field. The shaft
of the machine (as a conductive material) absorbs the energy and eddy currents are
setup in the shaft. When the shaft is close to the probe, more eddy currents are
generated. When the shaft is farther away (within the effective range), less eddy
currents are generated. The loss of energy (the absorbed energy by the shaft) reduces
the RF signal. The absorption of the field (energy loss- Eddy current) cause the output
of the probe to decrease in proportion to the gap distance. As the distance to the shaft
changes dynamically , so does the output signal.
 Amplitude is at a minimum when distance (Gap) between probe and shaft is at a
minimum. Maximum eddy current flow occurs.
 Amplitude is at a maximum when distance (Gap) between probe and shaft is at a
maximum. Minimum eddy current flow occurs.
 Change in distance should be within the effective range of the transducer commonly
80 Mil range
 If the shaft is moving slowly within the RF field, the signal amplitude increases or
decreases slowly. If the shaft is moving rapidly within the RF field, the signal
amplitude increases or decreases rapidly. Oscillatory movement of the shaft causes
the RF signal to modulate.
 The demodulator circuit deals with slowly or rapidly changing signal amplitude in the
same way. If the target (shaft) not oscillating , as might be the case with a thrust
probe, the proximiter output is a constant (DC) voltage, called the gap. If the target
(shaft) is oscillating (Gap changing slowly or rapidly) the proximiter will provide
both a DC (Gap) and AC (vibration) components in the output signal. When the
shaft is vibrating the DC component represents the average position of the shaft and
the AC (dynamic) output represents the dynamic instantaneous position of the shaft.
 When we plot this signal against time, we get a time base wave form. The “ peak to
peak” amplitude for a proximity transducer system is simply how close, and then how
far away the shaft is from the probe in its vibration cycle.
 The AC or alternating current signal is the vibration signal generated by the
instantaneous change in distance from the tranducer to the shaft. The DC or direct
current component ( also known as the “ Gap” ) is the average distance from the
probe to the shaft ( in terms of probe voltage). If the shaft is not actually vibrating, the
DC component is the actual distance between the probe and the shaft, once again , in
terms of voltage.
 The most commonly used proximity probes measure distances between the probe
tip and the shaft over an 80 mil range and change in distance cause changes in
output volts dc – Most commonly Sensitivity 200mv/mil (7.87mv/µm).
 A typical system frequency response is from 0 Hz to 10 KHz, Newer proximity
system transducers have responses up to 12 KHz
 Advantages:
 It measures the actual relative displacement of the shaft within the
bearing.
 Low frequency response (to 0 Hz)
 They are reliable (when installed correctly)
 Disadvantages:
 They are expensive and difficult to install
 Their calibration (determining the ratio between output voltage and actual
displacement) is dependent on the shaft material (different materials
absorb the energy at different rates.
 Shaft run out and surface glitches produce false signal
 Applications:
 Generally used for fluid film bearing machines

 Notes:
 Proximity probes are installed in X‐Y
configuration.
 X, Y Probe Configuration (Convention):
Looking from driver to driven (Looking
downstream), Locate X and then 90◦
Counter clock wise to locate Y

 They are useful as a key phasor

Note: Signals from 2 Probes X, Y which are 90 Deg Apart in the presence of a
key phasor can be used to plot an orbit plot

 Units:
 The units are mils Pk‐Pk or microns Pk‐Pk
 4.2.2 Velocity Transducers

 The basic components are a magnet,

coil, and a suspension mechanism
(Spring and damper)
 The relative movement of the coil wire
with respect to the magnetic field
generates electrical signal proportional
to the vibration velocity.
 No external power required, it
generates electricity.
 Medium frequency range
 Sensitivity (mv/mm/s) or (mv/in/s)
Velocity pickup frequency response
 Advantages:
 No external power required, it generates electricity
 The signal output is powerful
 It is easy to use, not as sensitive to mounting problems
 Ability to operate at higher temperatures
 Disadvantages:
 Not suitable for low or high frequency measurements.
 They are sensitive to change in temperature
 The moving parts in the design (magnet or coil) may lead to internal wear
which can shorten the life.
 Size are quite large
 Applications
 Velocity transducers were very popular, but they are no longer the
transducers of choice now a days. Accelerometers that are internally
integrated to velocity (velometers) are more common.
 Units
 mm/s or in/s rms
 mm/s or in/s pk
4.2.3 Accelerometer

 The most common types of transducers used in measuring vibration.

 A piezoelectric crystal (quartz or ceramic crystal is frequently used) is
supported within the accelerometer case, mounted above the crystal is a
mass, as the case of the accelerometer vibrates, the mass mounted above
the crystal also vibrates applying a force on the crystal, the crystal is
compressed and produces an electric charge which is proportional to the
force applied by the mass on the crystal and therefore proportional to the
acceleration (based on Newton second law, force is proportional to the
acceleration of the mass). An amplifier is then required to convert charge
output to voltage output signal.
 Accelerometer sensitivity is in mv/g
Charge Mode and ICP types Accelerometers
 Several years ago, most accelerometers were charge mode piezoelectric
which required an external charge amplifier, these accelerometers are
called charge mode accelerometers and are rarely used now a days.
 Today accelerometers have the amplifier built in, and they are called
Integrated circuit piezoelectric transducers (ICP transducers).
 The ICP transducer type requires to be powered from the data collector
although the piezoelectric crystal is self generating for electric charge, this
is because the internal amplifier needs power to be operated, the amplifier
is powered by the data collector using around 18 to 30 Vdc (bias voltage)
 The data collector can check the sensor and cable to verify whether they are
ok via the bias voltage

Charge Mode Accelerometer ICP Accelerometer

Types of Accelerometers based on frequency range (frequency
response)

 General purpose (2Hz-2.6kHz, Sensitivity about 100 mv/g)

 Most of the accelerometers supplied with condition monitoring
systems and, used in permanent monitoring applications use 100
mv/g accelerometer.
 Low frequency (0.1Hz-1 kHz, High Sensitivity about 500-1000 mv/g)
 For very low speed machine or precision machine like machine tool
with vibration at low frequency and with low amplitude.
 Low frequency accelerometers employ a large seismic mass to
increase the output from the sensing element (Piezoelectric crystal)
and allows higher voltage output from the sensor (high sensitivity
1000 mv/g)
 High frequency (0.6-60khz, Low sensitivity about 10-20 mv/g)
 For very large noisy machines, the sensitivity need to be much lower
Low Frequency Cut off in Accelerometer
Two reasons for low frequency cut off, the first is the presence of amplifier which has
limited low frequency response, the low frequency roll off of the amplifier is typically 1 Hz
for most general purpose accelerometer, there are some that are specially designed to go to
0.1 Hz if very low frequency data is required. The second reason is the low frequency noise
generated due to environmental effects such as mechanical parts thermal expansion, to get
rid of these unwanted low frequency noise a high pass filter consisting of a resistance and
capacitance is incorporated in the electronics of piezoelectric accelerometers

Note: LF accelerometers has also built in Low pass LP filter to get rid of high
frequency vibration (High Frequency Cut off) which may saturate the accelerometer ,
note that the LF accelerometers has high sensitivity 500 or 1000 mv/g and can be
overloaded (saturated) with significant high frequency vibration.

Accelerometer Temperature Range

 Due to the presence of amplifier and electronics which is affected by high
temperature, there is usually a maximum operating temperature, For General purpose
accelerometers it is typically 120 or 150 Deg C, for applications with higher
temperature than this, special charge mode accelerometers with external amplifiers
are used and the amplifiers are placed at a place far from the hot equipment.
Frequency Response
Piezoelectric velocity transducer
 Advantages:
 Very wide frequency range
 Wide amplitude range
 Small size and less weight in comparison with velocity transducers
 Commercially attractive
 Velocity and displacement available as output (through built in
integrator in the sensor or by integration by data collector)
 Disadvantages:
 Not responsive down to 0 Hz
 Temperature limitation due to use of internal amplifier
 Possible radio interference and hence noise in signal
 Long Settling time with low frequency accelerometers.
 Applications
 Accelerometers are the most common types of transducers used in
measuring vibration. From portable data collectors to permanent
monitoring systems.
 As mentioned before most of the available velocity transducers now a days
are Accelerometers that are internally integrated to velocity (velometers)

 Units
 G’s rms
 G’s PK

 Calibration
 Excessive heat will damage the accelerometer, as will dropping them on a
hard surface. IF an accelerometer is dropped more than a few feet on to a
hard surface the piezo electric crystal inside the accelerometer can crack
 It is a good idea to have the accelerometers checked and calibrated once
each year.
4.3 Selecting a Transducer

Selecting transducer is based on :

 Application: Selecting of the suitable transducer type (Proximity, velocity,
accelerometer) will depend on the application (type of machine) as
mentioned in the previous slides.
 Displacement (Proximity) probes are typically used in permanent
monitoring of fluid film bearing machines.
 Accelerometers are the most common types of transducers used in
measuring vibration. From portable data collectors to permanent
monitoring systems.
 Frequency range: The selected transducer frequency range should be
suitable for the application
 Amplitude range: The selected transducer amplitude range (Dynamic
range) should be suitable for the application.
Frequency Range ‐ Transducer Frequency Response

Frequency Response Curve for 100 mv/g Accelerometer

 There is a low frequency limit, a flat usable (operating) region, and

a sensor resonance at high frequency end.
Typical Frequency Response Curves for various sensitivities
Amplitude Range – Dynamic Range

 Sensors have amplitude limits too

 The upper amplitude limit is the point where the output begins to
exceed the capabilities of the sensor . The output becomes saturated
if vibration levels (amplitude) exceed sensor output specifications
 The lower amplitude limit is level below which the noise of the
sensor electronics becomes too great and swamps the data.
 For Precision machines and low speed machines where vibration
level are very low, a very sensitive sensor is needed that can
amplify the signals sufficiently, and it must have very low
electronic noise.
4.4 Survey the machine

 Gather important technical data about the machine like speed, load,
bearing type and number, no of vanes, no of blades, Belt length, No
of gear teeth, No of rotor bars of motor, …., this will help to adjust
your frequency range for data collection , and will help to select the
measurement locations, and will help in analysis and fault
diagnosis.
4.5 Accelerometer Mounting
Screw (Stud) Mounting
This provides the most accurate, high frequency,
and repeatability measures

wrong wrong

Right
right

wrong wrong
Adhesive Cement Mounting

epoxy,
Epoxy or Cement
Adhesive
adhesive

epoxy,
 Adhesive cement,
adhesive
 Very good frequency response range if proper type is used with the right
thickness is applied .
 For Permanent Monitoring, or Permanent installed on non accessible
machine location with extension cables to switch box (BNC selector box)
at accessible area for offline monitoring.
 High repeatability.
Magnetic Mounting Base

 Convenient means of providing a quick or temporary mount

 Adequate frequency range
 Reduces frequency response rating by approximately 50 % versus
stud mount
Sensor Disks (Pads) for Magnetic Mounting

 Useful for machines without magnetically attractive surfaces.

Sample Disk
(overhead view)

Sample Disk
(side profile)

note adhesive fillet between

machine surface and sample disk
Hand Held Probe Mounting

 Rapid and convenient

 Subject to many sources of error
 Use only as a last resource
4.6 Measurement Locations

 Vibration is measured at the bearing housing as close as possible to the

bearing location
 Bearings are the machine parts where there exist the shortest path for rotor
vibration to the bearing housings.
 Measurement Planes

radial
vertical
horizontal
axial
4.7 Repeatability

 The sensor must be mounted on the machines in the same way and at the
same locations every time, so that the only reason for vibration change will
be a change in machine condition.
 Repeatability of data collection at the same load , same speed as possible.
 If the machine operates at different loads, create multiple machine in your
database (ex: Machine Train A full load and another machine for half
load,.., or ex: Machine train A 6000 RPM and another machine for 4500
RPM)
4.8 Recognizing Bad Data

 Bad data can be caused by:

 Sensor Settling time; Insufficient settling time
 Impacting Sensor; saturation or overloading of sensor due to impacting
sensor
 Thermal shock of sensor; sensor is moved from a hot surface to cold or vice
versa
 Bad cable
 Bad cable connection
 Cable movement
 Sensor mounting; Loose sensor mounting
 Sensor Fault
 Bad Set up of data collection
Sensor Settling Time

Modern vibration
sensors have built‐in
amplifiers and
integrated circuits that
require a short period of
time to “settle” after
powering them; if data
is collected immediately
after powering the
sensor, it will result in
the errors illustrated in
Figures 1 and 2.
Sensor Impacting and saturation
Another Example of Sensor Impacting and Saturation
A mechanical shock can also cause measurement errors; this usually
occurs when a sensor is bumped or knocked. It is always a good idea
to allow sensors to settle before taking measurements.
Sensor overloading can occur due to extremely high vibration
amplitudes, which saturate the sensor and lead to false results.
Saturation can be caused by excessive vibration from adjacent
machines or by extremely high frequency vibration sources such as
cavitation.
Another Example of Sensor Overload due to cavitation
Sensor Overload due to Thermal Shock

 Another source of bad data is when sensors receive a thermal shock;

this may occur when transferring sensors from hot to cold surfaces,
and vice versa. The sudden temperature change affects the circuitry
performance and it is important to let the sensor temperature
stabilize.
 Thermal transients cause similar ski slopes to sensor saturation due to
mechanical shocks or impacting sensor
Bad Cable Connection
Sensor Loose Mounting
Loose mounting can result in unexpected harmonics in the spectral data.
This typically occurs when a sensor is not attached firmly to a machine.
Poor mounting can also prevent the sensor from detecting certain
frequencies, although this may not be obvious from observing the data.
ICP Sensor Fault for ICP sensors

 The transducers are powered from the data collector by adding DC bias
voltage to the leads of the transducer, by checking the bias voltage (option
existing in most data collectors), it is possible to detect sensor or cabling
connection faults.
 Commonly the bias voltage if the is no faults is between 18 to 30 V DC
 For damaged sensors, the bias voltage will be from 2 to 5 volts.
 Bias voltage 0 volts, this indicates no power or cable or connection problem
Bad data collection setup
Good data for analysis
Chapter 5. Signal Processing
5.1 Understanding Analog & Digital Signals
Cont. Understanding Analog & Digital Signal
 The analyser is a digital instrument, so the data must be converted into
digital signal so the analyser will have numbers to deal with.

 Various vibration signals coming from the rotating equipment which are
“analog” in nature, need to be transformed into digital domain.

 The term “ analog” refers to the continuous nature of the signal

 A digital signal on the other hand, is not continuous, as it is formed by finite

number of amplitude levels, separated by finite time intervals.

 Sample rate (sampling Frequency): The rate at which the wave form is
sampled. It is the number of samples made in one second.
 Time Domain: To represent the time wave form, the data collector only has
the samples to work with. Any thing that happened between the samples
is lost. When the data collector or software draws the time waveform it
simply connects the dots. This is called the time domain.
Correct Signal & Aliased Signal
AAF : Anti aliasing Filter and Sampling rate Convention to
avoid Aliasing Note: will be discussed later

Two sample rates A. Low sample rate that distorts the original sound
wave. B. High sample rate that perfectly reproduces the original sound
wave. Image from Adobe Audition Help.
 5.2 Sampling & Processing Signal

 The process through which an analog signal is transferred into digital is

usually referred to as sampling, some times called digitizing signal.
 The flow diagram below shows the whole sampling process which starts
with an analog signal and ends up as a digital signal.
5.3 Filters

 Filtering: Filtering is a process that removes some frequencies from a signal

in order to suppress interfering frequencies and reduce noise.

 Noise: Sources of noise

 Noise can come from the process itself or from external machines
 Noise can come from electronics and transducers.
 Low Frequency noise will be very much amplified if signal is
integrated (converted from acceleration to velocity or displacement).
There are four types of commonly used filters:
1 - Low pass filters
Low pass filters allow low frequencies to pass through. Low pass filters are the
most common filter type because of the popularity in removing alias signals,
and for other aspects of data acquisition and signal conversion.
2 - High pass filters
High pass filters allow high frequencies to pass through. High pass filters are
normally used in early bearing wear detection. A high pass filter is useful to
block the high amplitude, lower frequencies to enable to “amplify” to the low
amplitude levels of early bearing wear in the higher frequencies.
3 - Band pass filters
Band pass filters allow frequencies within a band to pass through. Band pass
filters transmit only those signal components within around a center frequency.
Band pass filters are usually applied in situations that require extracting a
specific tone, such as a test tone, from adjacent tones or broadband noise.
4 - Band stop filters
Band stop filters block frequencies within a band from passing through. Band
stop filters transmit all signals except those between specified ranges
LP Filter

HP Filter
Band Pass Filter

Band Stop Filter

Tracking Filters

 A tracking filter is a low pass or band pass filter which automatically

tracks the input signal versus the rpm .

 They are used in order tracking to follow the machine speed during changes,
its application in run ups and coast downs to capture phase data along with
the peak vibration.

 Some tracking filters can track multiples of 1X

Filters Analog or Digital

 Two types of filters are common used digital and analog filters
 Digital filters are achieved with software.
 Analog Filters are created with electronic components such as capacitors
and resistors (Settling time is important for this reason)

Settling Time

 When data is applied to the filter circuit, it causes the output of the circuit
to ring, that’s why it is required to set the settling time to a duration long
enough to allow the output to settle to normal levels before capturing the
data.
 Filters roll off (cut off) and Transition Band: Ideally, filters would block
unwanted frequencies and provide a clean cutoff and keep out of unwanted
signals. However this is not the case. In reality, there is a transition region
where some frequencies will be attenuated, but not blocked.
5.4 Analog Digital Converter ADC

 ADC is device incorporated in all modern data acquisition instruments, and

is used to convert the input analog signal into a digital (discrete) signal so
that it can be further processed
 The voltage level of the signal is quantized through analog-to-digital
converter (ADC) where in the voltage is divided into discrete levels
(sampling process)
 The most important parameters for ADC are input range and dynamic range
 Input range : Example ± 5volts
 Dynamic range: is a measure of the ability to see small signals in the
presence of large ones, it is the ratio of the smallest signal to the biggest
signal and is typically represented in decibel (dB)
Dynamic range= 20 log (smallest signal / biggest signal)
 Related to the dynamic range is the bit number n, the higher the bit number
the higher the dynamic range , and the higher the amplitude resolution
(detection) as well.
 The bit shows how finely it slices the configured full measurement range,
the number of digital levels in which the range is divided to 2n divisions.
 Example: 16 bit ADC, divides the full input range into 216 (means 65536)
divisions, So if the input signal is 1volt (or 1000mv), we can still detect
changes as small as 0.015 mv in the presence of the 1volt (1000 mv), the
dynamic range of the analyser will be 20 log (1 division/65536 divisions) =
= 20 log ( 0.015mv/1000mv) = 96 dB
 Gain: Data collectors do not have only one input range, there is usually an
amplifier (a gain stage) with the ADC to increase the amplitude to best suit
the input range of the ADC converter, gain is simply amplification of a
signal.

 The measurement Quality of a data collector is actually a measure of :

1) resolution of the ADC and 2) its ability to amplify the signal so that the
collector utilizes the majority of its input range.

 Analysers with 16 bit ADC will certainly do the job correctly. Even though
bit number as high as 24 or 32 are available today.
 Input Range Manual or Auto: While most data collectors will allow you to
manually set the input range (manual set the gain), in most cases you will
allow the collector to auto range, it will determine the best range
automatically.

 Carefully-should you know your vibration level and ADC input range first
Example: if we have 100mv/g accelerometer, ± 5volts input range,
then if the actual vibration signal was ± 50 g’s this will clip the signal.
 Auto Range Versus Auto Scale:

Auto range is the process of selecting optimum gain setting, Auto scaling is
a simple graphical display which takes the spectrum and displays it with the
optimum display range
5.5 Sampling & Aliasing
 The process through which analog signal is transferred into digital signal.
 There are number of consideration
 How quickly should the signal be sampled
 how many samples are kept
Low Sample rate will cause Aliasing, Aliased signal will be generated

Two sample rates, Wave A with low sample rate that distorts the original
wave. Wave B with high sample rate that perfectly reproduces the
original wave.
 Aliasing & Nyquist Theorem

A 30 Hz Analog signal sampled with sample rate 60 Hz ( 60 sample per second),

resulted in a digital signal straight line
In case b) Aliased lower signal (green), resulting from a low sample rate. At this
point , it is important to state that once the aliasing of a signal has occurred, there
is no way to know whether the reconstruction signal is real or not.
 Aliasing: If we do not sample fast enough the reconstruction of the wave form
will not match the event (the actual signal), the reconstruction curve has the right
amplitude but the wrong period (or wrong frequency).this effect is called
aliasing.
 It is important to state that once the aliasing of a signal has occurred, there is no
way to know whether the reconstruction signal is real or not. The signal will
appear real in the spectrum but does not really exist.
 Nyquist Theorem: Nyquist theorem states that the wave form must be sampled
at a rate greater than twice the input signal. This frequency is often referred to as
the Nyquist Frequency
Fs > 2 f in ;Fs sampling frequency (sample rate) which is the number of
samples per second
; Fin the actual Analog signal.
 Note: Most digital analysers sample at 2.56 times the maximum frequency of
interest Fs = 2.56 fmax ; fmax is the max frequency of the spectrum.
Coping With Aliasing
• First Cause : a frequency is present in the signal that is not being
sampled fast enough ( more than 2 times the signal frequency

• Solution: To use Anti Aliasing filter AAF to remove any high

frequency component in the signal, this filter is low pass filter set at
the fmax , which will filter out any signal with a frequency higher
than our fmax. Most Digital analysers have built in AAF. the second
action is to sample the rest of the remaining signal at an adequate
sampling rate Fs > 2 fmax. Most digital analysers sample at 2.56
times the maximum frequency of interest Fs = 2.56 fmax ; fmax is
the max frequency of the spectrum.
• Second Cause:

(Just Hint- Need Further Discussion in advanced Course)

Two Signals are said to alias if the difference of their frequencies falls in
the frequency range of interest (Fmax in our spectrum). In other words if
the (Fs sampling rate or sampling frequency – Fin actual signal frequency)
falls below our selected f max setting, an aliased signal with frequency
equal to the difference frequency (Fs-Fin) is always generated in the
spectrum.

• Solution:
Actual number of lines in the spectrum are equal to half Number of
samples in wave form (Number of samples /2), There are FFT calculations
for half the number of samples, however to cope for the aliasing effect,
due to fold-over phenomena mentioned above, only number of lines equal
to Number of samples divided by 2.56 are kept (Number of lines=Number
of samples/2.56), and the rest of the lines are discarded to cope for
Alaising due to the cause above.
(Just Hint- Need Further Discussion in advanced Course)

Ideal and Real antialiasing filters

Aliasing Fold‐Over Phenomena

Coping With Aliasing Summary
AAF: Use anti Aliasing filter to get rid of any
signal with a frequency greater than our
frequency range of interest (f max), all data
collectors have some form of AAF

Adequate Sample Rate : Fs = 2.56 fmax

Solution for Sample with a sampling rate high enough, all
Eliminating data collectors sample at a rate equal to 2.56
Aliasing times the adjusted f max

Spectrum LOR= Number of samples/2.56,

Only number of line of resolution in the
spectrum which is equal to the number of
samples in the wave form divided by 2.56, only
these lines are kept and the rest of lines or FFT
calculations are discarded (actual FFT lines are
half No of samples in the wave form)
Two Rules are Applied by the Data Collector to Prevent
Aliasing

First Rule: Sample Rate Fs = 2.56 Fmax

Second Rule: No of Spectrum Lines LOR = No of samples/2.56

These two rules are convention applied by data collectors

manufactures, in the presence of built in AAF, will eliminate
the aliased signals.
 Time resolution is basically controlled by how fast the instrument collects
the data (sampling frequency). In contrast with the amplitude resolution
(depends on the no of bits of ADC), time resolution can be modified when
configuring the data acquisition device. The below figure shows that
sampling frequency (sampling rate) of 3 samples per second results in a
closer representation of the original analog signal, however when the
sampling frequency is three times slower, the reconstructed signal becomes
a poor representation of the analog signal.
Overall Amplitude
6 mils PP

Signal appears to
be quite smooth

Digital signal sampled at 640 samples per second or 640 HZ Same

Analog
Signal

Overall Amplitude
8.5 mils PP

Signal shows
some spikes

Digital signal sampled at 25600 samples per second or 25600 HZ

Time Span For The time block

Time Span T = No of samples/Sample rate

= LOR/f max

; Sample Rate Fs = 2.56 f max

; No of samples = 2.56 No of LOR

5.6 Frequency Domain
 The FFT process takes the time wave form and creates the spectrum which is
referred to as the frequency domain
Simple Wave forms
Complex Waveform

FMAX
Amplitude
9X
5X
3X Frequency
1X
Spectrum
Plot
Time Domain
(Sec or Min) Frequency Domain
TMAX (CPM or Hertz)
 The vibration spectrum is the result of a Fourier transform, named after the
mathematician who developed the equation which turns a complex wave into its
harmonic components
 Two other mathematicians, Cooley and Turkey developed an algorithm to
effectively speed up the processing of the Fourier transform and thus the name
Fast Fourier transform, to achieve the speed the number of samples in the time
series must be a power of 2: 28 = 256, 29= 512, 210 = 1024, 2211 = 2048, 212 =
4096,…………
 Number of Spectrum lines (LOR) = No of samples in TWF / 2.56
(as mentioned before to avoid Aliasing)

• Amplitude
• Frequency
• Phase
• To achieve the speed of FFT calculation, the number of samples in the time
series must be a power of 2: 28 = 256, 29= 512, 210 = 1024, 211 = 2048, 212 =
4096,…………
Important Note:
No Of Samples In TWF Number Of Lines in the Spectrum LOR=
Spectrum Number of
samples/2.56, Only
number of line of
resolution in the
N =29 = 512 Samples 256
spectrum which is
equal to the number of
samples in the wave
N =210 = 1024 Samples 512
form divided by 2.56,
only these lines are
kept and the rest of
N =211 = 2048 Samples 1024 lines or FFT
calculations are
discarded (actual FFT
lines are half No of
N =212 = 4096 Samples 2048
samples in the wave
form)
Important Note:
Spectrum LOR=
Number of
samples/2.56, Only
number of line of
resolution in the
spectrum which is
equal to the number of
samples in the wave
form divided by 2.56,
only these lines are
kept and the rest of
lines or FFT
calculations are
discarded (actual FFT
lines are half No of
samples in the wave
form)
• Number of line of resolution in Spectrum LOR is equal to Number of samples in
TWF divided by 2.56, LOR = No of Samples /2.56

No Of Samples In TWF Number Of Lines of

Resolution in the
Spectrum (LOR)

N =29 = 512 Samples 200

N =210 = 1024 Samples 400

N =211 = 2048 Samples 800

N =212 = 4096 Samples 1600

Time Domain and Frequency Domain Rules

• Sample Rate Fs = 2.56 f max

• No of samples = 2.56 No of LOR

• Time Span T = No of Samples/Sample Rate

= No of LOR / fmax
5.7 Windowing

 One assumption made in FFT calculation is that the time record is

continuous. That is the signal just before the captured time record, and the
block immediately after our time record are identical.

Time signal Time signal Time signal

Time Block
• Leakage Example:

Time signal
1 0.5
Amplitude [V]
0 -0.5
-1

0 100 200 300 400 500 600 700 800 900 1000
Time [ms]
• What is Leakage?
 Caused when the time waveform signal does NOT begin and end at
same point, introducing spurious frequencies
• The signal is discontinuous. It seems to have a step increase in level and
looks similar to an impact to the FFT calculation. It generates a peak like a
skirt, that is spread over a wide frequency band similar to an impact

Leakage
 The Window function attenuates the signal towards the edge of the
window minimizing leakage.

• Hanning Window:
Time signal
1 0.5
Amplitude [V]
0 -0.5
-1

0 100 200 300 400 500 600 700 800 900 1000
Time [ms]
Spectrum From Spectrum from
Windowed TWF Non windowed
Signal TWF Signal

Leakage
Window Types
 Hanning Window

 Hamming Window

 Flat Top Window

 Rectangular Window (Uniform or No window)

 Exponential Window
 Hanning Window:
 Best Frequency accuracy, window factor = 1.5
 Affects the amplitude, amplitude accuracy, error ≈ 18%
 Best Compromise between frequency resolution and amplitude
accuracy for steady state machinery analysis.
 Is the most commonly used in the routine vibration monitoring
 Hamming Window:
 Similar to Hanning window except the ends do not go to zero
amplitude.
 Flat Top Window:
 Window factor = 3.8
 Best Amplitude Accuracy – Error ≈ 1%
 Frequency accuracy is poor
 Flat top window primary use for calibration and accuracy
(Amplitude)
 Rectangular Window (Uniform or No window):

 Window Factor = 1
 Like No Window
 Amplitude Accuracy: error ≈ 56%
 Commonly used in Bump Testing and resonance checking
 Exponential Window:

 A Special windowing function for minimizing leakage in lightly

damped structures that is used in impact testing.

 In a lightly damped structure, during its impact testing some

times its oscillations may not die out within the sampled time
block, which results in a leakage error in the spectrum. An
exponential window adds damping to the time signal to force it
to die out within the time block, thus minimizing leakage. The
added damping is then removed mathematically after the signal
is processed.
Cont. Exponential Window:
Cont. Exponential Window:
Applications for Different Window Types

Application
(Type of Signal) Window Type

General Purpose Application (Steady State Hanning

Machinery analysis & Routine Monitoring)

Calibration and Amplitude Accuracy Fat Top

Transient (Bump Testing ‐ Impact Testing Rectangular

Resonance checking) Whose duration is shorter
than the length of the window (time block)

Transient Whose duration is Longer than the Exponential

length of the window (time block)
Effect of Windowing on Spectrum resolution

• The window function affects the frequency resolution of the

spectrum but we still need to use it to reduce leakage.

• The FFT always has a defined number of lines or bins (Lines of

resolution) 100, 200, 400, 800, 1600, 3200, …etc

• Resolution = fmax /No of lines

• Band width = Resolution X Window Factor

Window Factor = 1.5 for hanning window

= 1 for uniform (Rectangular) window
= 3.8 for flat top window

Window Factor is multiplier to apply when determining resolution

Bin or line
Effect of Hanning Window on Spectrum resolution
• Although hanning window has the best frequency accuracy, it still
affects it. There will also be leakage.

• Resolution = Fmax / No of lines of Resolution(bins)

• Band width = Window Factor X Resolution = 1.5 X fmax / LOR

• Separating Frequency: In order to completely resolve two discrete

frequencies
Separating freq ≥ 2 band width
≥ 2 window factor X Resolution
≥ 2 X 1.5 X fmax / LOR
Example: 0 - 400 Hz using 800 Lines

& Hanning Window

Band Width = ??

Answer:

Band Width = (400 / 800) 1.5 = 0.75 Hz / Line

Example: The student wishes to measure two frequency disturbances that are
very close together.

Frequency #1 = 29.5 Hz
.
Frequency #2 = 30 Hz
.
The instructor suggests a hanning window and 800 lines.

What frequency span is required to accurately measure

these two frequency disturbances ?
Answer
• Separating freq = 0.5 Hz
• Separating freq ≥ 2 band width
≥ 2 window factor X Resolution
≥ 2 X 1.5 X fmax / LOR

• Separating Freq ≥ 2 X 1.5 X fmax /800

0.5 Hz ≥ 3 fmax/ 800

0.5 Hz X 800 / 3 ≥ fmax

133.33 Hz ≥ fmax fmax should 133.33 Hz or less

Answer : fmax = 133 Hz
• Note: Therefore the frequency span must be 133 Hz or less to measure
the desired 0.5 Hz, if Fmax is greater than 133.33 Hz, the two frequencies
will not be completely resolved and their band width will combine.
5.8 Averaging
 Vibration is constantly changing slightly and there is noise in the signal

 To ensure more representative spectrum we need to make averaging

 Averaging helps in reducing noise from the signal and maintaining

repeatability, so that two measurements taken 5 minutes apart should be the
same.
Types of Averaging
 Linear Averaging

 Peak Hold Averaging

 Time Synchronous Averaging (TSA)

 Exponential Averaging
Linear Averaging
 Each spectrum is added to the next and the sum is divided by the total
number of spectra, this is done for each frequency line in the spectrum,
random back ground noise will be reduced.

 For Example if 4 Averages is required:

Averaged Spectrum = (S1 + S2 + S3 + S4) / 4

 The Linear Averaging is the most commonly used type in routine

monitoring.

 Number of averages: For routine monitoring usually 4 – 6 averages

Linear Averaging with overlapping
 In order to increase the speed of averaging, and to ensure that we do not lose
important data, we use overlap averaging.
 30 % to 50 % overlap is commonly used,
 Typically in routine monitoring, we use 4-6 averages and 50 % overlap.
.
Peak Hold Averaging

 peak hold is not a true averaging method, the peak value registered in each
analysis cell (Spectrum bin), from all the spectrum sets, is captured and
displayed

 Peak hold averaging is very useful in viewing transients ( Bump test, impact
test, run-up/coast down tests)
Time Synchronous Averaging

 Time Synchronous averaging (TSA) is performed by the TWF rather than

with the spectra, it uses a tach (trigger) signal to synchronize starting of data
collection of each TWF block with reference to tach pulse, TSA and
averages to zero all frequencies that are not integer multiples of the tach
signal. The remaining energy is turning speed and its harmonics.

 TSA averaging is very useful in gear box analysis, large number of averages
may be required to completely filter our non synchronous components.
 If we average TWF without synchronizing, they may cancel each other,
Tachometer pulse to synchronize wave forms is required
 TWF signals are synchronized by use of tachometer pulse, TSA average can
be done.
Linear Averaging with overlapping Data collection time
 As mentioned previously, the time for data collection (Time span in time block)
is equal to No of samples in TWF divided by the sampling rate, or equal to no of
lines (bins) in the spectrum divided by fmax, but this is the time for one average,
if we selected 4 averages, this time will be multiplied by 4.
• Sample Rate Fs = 2.56 f max

•No of samples = 2.56 No of LOR

•Time Span (for one average) = No of Samples/Sample Rate

“Time Window” = No of LOR / fmax

• Total Time of data collection T=(No of Samples/Sample Rate)*No of averages

•In case of overlapping this total time for data collection will be reduced based
on the % overlap.
Example:

How long will it take for 10 averages

at 75% overlap using a 800 line analyzer

and a 200 Hz frequency span?

75% Overlap ?
10 Averages
75% Overlap
800 Lines
200 Hz
Average #1 = 800 / 200

Average #1 = 4 seconds

Average #2 - #10 = (4 x 0.25)

Average #2 - #10 = 1 second each

Total time = 4 + (1 x 9)

Total time = 13 seconds

5.9 Spectral Integration
Spectral Integration
Displacement Spectrum

Integration
Integration
reduces
high
frequency
vibration Velocity Spectrum
and
amplifies
low
frequency
vibration,
and
introduces Acceleration Spectrum
90 Deg
Phase
change
Spectral Integration

• Why is the
“Ski‐Slope
produced from
Integration?

• How do we
solve this
problem?
Spectral Integration

• Integrating Acceleration to
get Velocity pops out a
constant value, which is
manifested as a “DC”
component because it has
no frequency dependence
Spectral Integration

• How can we
solve this
problem?

Truth is – we can’t!
It’s PHYSICS!

• What we can do
is…“Zero” the
first 5 or so
Spectral Bins!
5.10 Guide for Data Collection Settings For Routine Monitoring

Remember:
First Rule: Sample Rate Fs = 2.56 Fmax
Second Rule: No of samples = 2.56 No of Spectrum Lines LOR
Third Rule: Time Span T = No of samples/Sample rate
= LOR/f max
 Most of analyzer give you the option either to adjust
fmax and no of lines, or to adjust sample rate and no of
samples, you adjust two parameters and the analyzer
determine the other two parameters from the
mentioned 2.56 rule. Also the time span of the TWF
will depend on the selected two parameters as per
above rule.
Important Notes:
 The information in this section (Guide for Data Collection
Settings For Routine Monitoring) is just a guide, not rules,
There are no fixed rules for adjusting data collection
settings, it can differ from one application to another.

 In some applications, this guide may not be efficient.

 There are different ways and different approaches in

adjusting data collection settings, fmax, Spectrum and TWF
Resolution, time span.
Cont. Important Notes:
 A great many less-experienced vibration analysts just stick
with one frequency range for example 1000 HZ, regardless
of the machine being analyzed, this can lead to missing key
frequencies

 After Base Line Measurement and reviewing data for the

machine and the identical machine units, further tuning for
data collection settings may be required
Some forcing frequencies associated with machines:
Source Frequency (Multiple of RPM)

Fault Induced

Mass Imblalance 1X

Misalignment 1X, 2X

Bent Shaft 1X

Mechanical Looseness 1X, 2X, 3X,..

Structural Looseness (Flexibility) 1X

Rotor Rub (Fluid Film Bearing) 0.5X

Antifriction Bearings Bearing frequencies (not Integer ones)

Design Induced

Universal Joints 2X

Asymmetric Shaft 2X

Gear Mesh (n teeth) nX

Coupling (m Jaws) mX

Fluid Film Bearings (oil Whirl) 0.39X ‐ 0.49X

Blades and Vanes (m) mX

Reciprocating Machines Half and Full Multiples of speed, depending on design

Fmax Setting for TWF and Spectrum
 Machine Configuration: You must look at your machine configuration (Speeds,
bearings, gears, belts, vanes, blades, vanes, ….etc.)
 Machine Forcing Frequencies: The spectrum frequency Span fmax is selected based
on machine RPM and other machine forcing frequencies (see table in the previous
slide for forcing frequencies).
 Harmonics of Forcing Frequencies: you should also take in consideration that you
need to analyze harmonics (multiples) of these forcing frequencies.

Default Frequency Span fmax Guide

Component Span
Shaft Vibration 10 X RPM
Gear Box 3 X GM
Rolling Element Bearings 10 BPFI
Pumps 3 X VP
Motors / Generators 3 X 2Lf
fans 3 X BP
Sleeve bearings 10 X RPM
Fmax Setting for Spectrum
 General Recommendation: If adequate resolution is not available, multiple data
measures for the same point with different frequency spans must be acquired, it is
some times desirable to split the data point into two or three frequency spans

 All Points Spectrum‐2o orders Fmax: (All Measurement points ), a 20 order

frequency range will give you a number of harmonics blade/vane pass, In addition
that it will help you to analyze non synchronous bearing tones harmonics and side
bands.

 One or two points Spectrum‐10 Orders but with high Resolution: (in Load Direction
for example ),this will help you to distinguish between peaks which are very close to
each other , Ex 2X and 2 Lf , 4X and 2Lf, Synchronous and non synchronous energy
which could be confused together. Harmonics of belt frequencies and harmonics of
running speeds which also could be confused together.

 One or two points spectrum ‐High range (High fmax) 50 to 100 order : (In load
direction for example), this will help you in early bearing detection (bearing
harmonics at high frequency region), analyzing rotor bar passing frequencies, Gear
mesh harmonics and side bands.
Resolution Settings for Spectrum

 All Points Spectrum: (All Measurement Points) a Spectrum resolution of 0.5 HZ (30
CPM) seems suitable for common analysis and trending , this will have Band width of
1.5 times the resolution (with Hanning window), which will be equal to 0.75 Hz (45
CPM), this spectrum can be defined for all points.

 One or Two points Hi resolution Spectrum 0.1 Hz (6 CPM) Res, Bandwidth 0.15 HZ
(9 CPM): (in Load Direction for example ), a hi Res Spectrum is recommended to be
used, this will help you to distinguish between peaks which are very close to each
other , Ex 2X and 2 Lf , 4X and 2Lf, Synchronous and non synchronous energy,
Harmonics of belt frequencies and harmonics of running speeds, it will also help you
in analyzing side bands.

 One or two points spectrum ‐High range (High fmax) 50 to 100 order, Normal
Resolution, The resolution in this high Freq range spectrum can be 1 HZ: (In load
direction for example), this will help you in early bearing detection (bearing
harmonics at high frequency region), analyzing rotor bar passing frequencies, Gear
mesh harmonics and side bands.
TWF Setting

 TWF Resolution: TWF resolution is controlled by Sampling rate, Time span T (Time
window) for TWF is controlled by No of samples and sample rate (or No of line and
fmax) ; T = No of samples/Sample rate = No of lines/ fmax

 Sampling Rate (related to fmax): is selected based on the machine forcing

frequencies and their harmonics as mentioned previously.

 Typically a time span (Time Window) equivalent to 6 ‐10 Revolution of shaft is

required, don’t set the time window too long.

 2048 samples is usually considered enough sampled points.

Other Data collection Settings for Spectrum & TWF

 Units
• For Spectrum units up to 1000 HZ can be better seen in velocity units

• For Spectrums units above 1000 HZ can be displayed in both units velocity
and acceleration, acceleration in this region is more sensible.

• For Time wave form units, it is usually better to be seen in acceleration units,
as the integration to velocity will suppress the high frequency components
related to the bearings, and you would like to see the bearing defect impacts
in the TWF

Note: for fluid film bearings Rules are different where displacement proximity
probes are used, displacement is the unit of choice.
TWF Units: Effect of Integration on TWF Pattern (Shape)
Important
Note:
Acceleration
TWF shows
Data from clearly the
same high
measurement frequency
location but in impacting of
different units bearing
(Acceleration
& velocity) for
defected
bearing.
Velocity TWF
See how the is dominated
integration by Low freq.
from components
acceleration to 1X and 2X,
Velocity high freq.
affected the Bearing
shape of the impacts not
TWF. clear
Cont. Other Data collection Settings for Spectrum & TWF

 Windowing

• For Routine monitoring and steady state, use hanning window

 Averaging

• For Routine monitoring and steady state, linear (some times called Spectral)
averaging with No of averages 4‐6

 Overlapping Averages

• For Routine monitoring and steady state, 30 ‐ 50 % overlapping averaging is

recommended.

 Filtering

• For Routine Monitoring a HP filter 1 HZ seems suitable to remove low freq

noise, but you need to get sure that this region will not include any actual
forcing frequencies in your machine
5.11 Overall Readings
For Complex Wave (Non Sinusoidal):

X(t) could be velocity v(t) or Acceleration a(t) or displacement D(t)

Overall RMS Readings

 Overall RMS reading can be calculated by one of two ways :

Analogue Overall RMS
Digital Overall RMS

Overall RMS

Analog Overall RMS Digital Overall RMS

From Digitized TWF From FFT

Signal (Spectrum)
 Analogue Overall RMS : the RMS is measured from the raw analog signal by
using analogue circuitry that solves the RMS integral equation below
 Digital Overall RMS reading can be calculated by one of two ways :
 Overall RMS from Digitized wave form
 Overall RMS from FFT

 Digital Overall RMS is calculated digitally from the digitized TWF or

some portable analyzer manufacturers provide overall RMS reading
calculated from the spectrum

Digital Overall RMS

From Digitized TWF From FFT

Signal (Spectrum)
 Digital Overall RMS from Digitized TWF : Digital Overall RMS is calculated
digitally from the digitized TWF by a way which is actually reverse of its name (Root
Mean Square) i.e. the values of all samples in TWF are squared then averaged then
the square root is obtained .
 Digital Overall RMS Spectrum (FFT Overall) : FFT Overall is done by a
process known as the root sum square RSS, which is actually reverse of its name
(Root sum square), i.e. the values in each frequency box (spectrum line) is squared
,then summed and then the square root is obtained.

∑amplitude2
FFT Overall =
.

; 1.5 factor is used to correct for Hanning window

 Overall RMS

Analog Overall RMS Digital Overall RMS

From Digitized TWF From FFT

Signal (Spectrum)

• Which is better for overall RMS values

calculation??
Which is better for overall RMS values calculation??

 Analog Overall RMS:

 Analog overall calculated from the raw analog signal
 With Analog overall results are exact and not subject to data collection
setup settings.
 Digital Overall RMS from Digitized TWF
 is calculated digitally from the digitized TWF
 Readings will be quite near to that obtained from Analog overall
depending on manufacture convention while digitizing the analog signal,
whether the data collector will use its max frequency range ( and max
sampling rate ) or not.
Which is better for overall RMS values calculation??
• Digital Overall RMS from FFT ( FFT Overall )

 is calculated from the FFT line amplitudes

 There will be some error in the reading (deviation from that Analog
calculated overall) if the actual signal contains higher frequency
component higher than the fmax setting of the spectrum.

 There will be some error in the reading (deviation from that Analog
calculated overall) if the actual signal contains very low frequencies
which may be zeroed in the first three or four bins of the spectrum
(usually HP filter is set up to remove low frequency noise)
Which is better for overall RMS values calculation??

• Best Practice
 To trend both Analog overall RMS and FFT overall RMS values

 Analog overall is more accurate and will give you an indication if

there is something happening in the machine causing increase in the
overall while not affecting the FFT overall because of its fmax or HP
filter settings

 FFT overall will be more sensitive to the increase in vibration due to

a frequency in your frequency range of interest (fmax) in the FFT

 Excellent reason to calculate overalls from FFT is to use band

alarming where the energy in a frequency range is used as indicator
of trouble (Ex unbalance band, misalignment bands , bearings band,
……etc.
Overall PK and PK‐PK and True PK‐PK Readings

 Discrepancies between different data collectors depending on the

convention used by the manufacturers of the data collectors.
 Early portable data collectors often provide a peak reading that was simply
the RMS valueX1.414, and this is not true peak, it is correct only for pure
sine wave, but most of the real vibration waves, the positive peak ≠ negative
peak ≠ RMS X1.414
 Many analyst in the past had become accustomed to this PK=RMSX1.414 ,
and had a large data base and historical trends depending on this relation
PK=RMS X 1.414, so called derived peak (or sometimes called calculated
peak) and the option was retained under these names.
 Later, more options for wave form amplitude detection was added to
determine the true Pk-PK values from the wave form but also different
conventions for true PK . Discrepancies in True PK values still present.
 Some manufacturers convention for the true PK value is simply the true
PK-PK value divided by two. (True PK= True PK-PK/2), in our Figure
above the True PK will be 2.5
 Some manufacturers convention defines true Pk as the larger of either the
positive or negative pk. In our figure above the True PK will be 3
 Discrepancy between the two convention with about 20% in the values of
the true PK
 There is no right or wrong convention and when a symmetrical
wave form is used both convention will provide identical results
 Because true Peak and RMS are completely unrelated, instruments
capable of accurately providing both readings must employ one
type of circuit or algorithm for computing True peak and distinctly
different one for computing RMS
5.12 Revision on Signal Processing
Chapter 6. Analysis Process
6.1 Introduction About Vibration Analysis

 Objective of this chapter: To know important concepts in vibration

monitoring and analysis and to become familiar with the most
important vibration analysis techniques and tools that analyst can
use for effective vibration monitoring and analysis
6.2 Vibration Monitoring and Analysis Concepts

 First Concept Trending: Vibration Monitoring (or in general condition

monitoring) is all about trending (comparison of current data with historical
data).

 Second Concept Comparison: this is very important in vibration monitoring

and analysis, diagnosis and assessment as well.

• Comparison with reference data (base line after commissioning when

the machine was newly installed, or after an overhaul).
• Comparison with historical data (trending) as mentioned above
• Comparison between driver and driven equipment vibration (across
coupling for example).
• Comparison between vibration from the same equipment and same
location but different axes (Horizontal‐Vertical ‐Axial). (Whole
Machine Approach taking into account the machine orientation).
• Comparison between identical machines under the same operating
conditions
 Cont. Second Concept: Comparison:
 Important Notes:
• For Comparison purpose, whether comparison with historical data
(trending) or identical machines, there should be same operating
conditions i.e. same speed, same load, after thermal stability,…..

• If you were not able to achieve the same conditions, you can define
two data bases for the same machine on your condition monitoring
software for example one machine to be named Gas Compressor @
oper Speed 5000rpm, and another machine to be named Gas
compressor @ Oper Speed 6000 rpm, or one machine to be named
Elect Generator Half load, and another machine to be named Electr.
Generator full load, and so on.

• Create two different routes (lists), depending on the machine mode of

operation choose which route you will use in your data collection.
Cont. Vibration Monitoring and Analysis Concepts

 Third Concept Respond to maintenance activities: It is very important to

collect vibration data just before and just after major maintenance activities.

 Fourth Concept: One maintenance action at the time for investigation: In

some cases there are more than a probable cause of vibration, in such cases
start with the easier recommended maintenance action or check, and re‐
measure and make your analysis just after applying the maintenance
action. Try to agree with maintenance team about one action at the time.
This will help you to eliminate the probable causes and finally to put your
hand on the actual cause. This data will be recorded and notes will be
registered. This will also help you in the future if a similar case is repeated
for this machine or the sister machines.
6.3 Vibration Analysis Techniques
 Vibration Analysis Techniques: The most important vibration analysis
techniques are:

• Spectrum Analysis
• Time wave form TWF Analysis
• Phase analysis
Which is better????
In some cases one of the above technique is better in diagnosing certain fault
than the other, in other cases you might need further information from the
other techniques to eliminate the suspected faults and correctly diagnose the
existing fault. So depending on the situation, combining more than a
technique or altogether will provide you the best effective diagnosis of fault
conditions.

 Important Note: A great many less‐experienced vibration analysts just stick

with the spectrum, and neglect the time wave form and phase analysis, the
spectrum will tell you part of the story, but to know the whole story you have
to study the time wave form and phase as well.`
6.4 Types Of Vibration data based on machine state
 Four machine states that we need to collect data at, and make analysis for this
data:

• Steady state vibration data : Fixed speed (usually the operating speed)
• Transient vibration data: Run‐up/Coast down
• At rest data : Zero Speed
• Slow roll data: Very low Speed
 This Course we will focus on Spectrum ,Time wave form for steady state
machine operation. A hint will be given about orbit plot. Details about orbit
analysis are not covered in this course.

 Phase analysis is important in steady state data and transient data analysis as
well, phase analysis will be covered in the machinery fault diagnostics and
resonance sections

 Transient data analysis ,a hint will be given about transient analysis in the
resonance and testing for critical speed sections
 General about Vibration Analysis techniques and used plots in steady state
and transient vibration analysis:

• Under the three mentioned main techniques (spectrum, TWF, and

phase), there is a lot of plots with different data formats, which analyst
can use in his analysis, some of these plots are used in steady state
vibration analysis (Fixed speed), others are used in resonance testing
including transient data analysis (run‐up/coast down data). Other plots
are specialized for fluid film bearing machines where non contacting
displacement (proximity) probes and key phasor are installed.
6.5 Steady state vibration plot formats
The following are the common data and plots used in steady state vibration
analysis, All the below data and plots are generated from Spectrum, or
TWF, or phase, so interpreting or making analysis by these plots lies
under the umbrella of the main three techniques mentioned previously
(Spectrum , TWF and phase analysis).

• Trend (overall, Band Overall, 1X , 2X , 1X Phase, 2X Phase,…)

• Spectrum
• Water fall (depends on spectral data)
• Time Wave form
• Absolute phase data (for fluid film bearing machines which have key
phasor installed)
• Relative (cross) phase data
• Operating deflection shape (depends on relative phase data)
• Orbit Plot ( Generated from time wave forms, plot used for fluid film
bearing machines which have two orthogonal proximity probes
installed and key phasor)
6.5.1 Vibration Trend Plot

 Trend graph provide a quick visual view to the changes that are occurring, it shows
the change of a certain vibration parameter over time.

 Trend plot can be used to show the vibration level over a long time for monitoring
steady state conditions, or some times is used with transient run‐up/coast down
to compare run‐up vibration with a previous run up (compare with run up data
taken from 2 months ago for example)

 Trying to diagnose a problem from single data set with out having trends and
historical data showing how the condition has changed over time is very difficult.

 When the trend shows that there is increase in a certain vibration parameter, you
can not know the reason for the increase depending on the trend, trend is not the
answer, but it does indicate whether a problem is developing, So you will need to
make in depth analysis, as mentioned previously, with the three main analysis
techniques : Spectrum, TWF and phase data.
6.5.1 Cont. Vibration Trend Plot
 Trending overall values only is not enough monitoring for the vibration condition.
 Trend graphs can be used to trend the following:
• Trend for vibration overall values (Some times overall values are called Direct
values).

• Trend for certain spectral band overall value (for example band from 0.7 X to
1.3 X or band from 3.5X to 8.5 X)

• Trend for 1X, or 2X, or vane pass, (or……) vibration components

• Trend for phase data 1X phase or 2X phase or,………..
• Trend for gap voltage for fluid film bearings with displacement proximity
probes

• ……………………………………………………………………………………………………………….
• ………………………………………………………………………………………………………………
6.5.1 Cont. Examples for Vibration Trend Plots
6.5.1 Cont. Examples for Vibration Trend Plots
6.5.2 Spectrum Plot (Further Discussion on Spectrum Analysis
later in this chapter)
6.5.3 Spectra Water Fall Plot
 Water Fall plot: This plot helps to study a large samples of spectra at different dates
and shows how the spectral peaks and patterns changed over a period of time
6.5.3 Cont. Examples for Spectra Water Fall Plot
6.5.4 Time Wave Form TWF Plot (Further discussion for TWF
Analysis‐Later In this Chapter)
6.5.4 Cont. Time Wave Form TWF Plot (Further discussion for
TWF Analysis‐Later In this Chapter)
6.5.5 Operating Deflection Shape ODS Analysis (Further
discussion later in Resonance and Diagnosis Section)
 ODS Analysis : Need ODS Software for plotting animating the Operating Deflection
Shape
6.5.6 Orbit Plot (Further discussion for Orbit Analysis‐Later In
this Chapter)

Direct (Unfiltered Orbit) (Overall Orbit)

6.5.6 Cont. Examples for Orbit Plot (Further discussion for
Orbit Analysis‐Later In this Chapter)

Direct (Unfiltered Orbit) (Overall Orbit)

6.5.6 Cont. Orbit Plot (Further discussion for Orbit Analysis‐
Later In this Chapter)

1X Orbit (1X Filtered Orbit)

 This Chapter we will focus on Spectrum and Time wave form
analysis for steady state machine operation. Details about orbit
analysis are not covered in this course.

 Phase analysis is important in steady state data and transient data

analysis as well, phase analysis will be covered in the machinery
fault diagnostics and resonance sections

 Transient data analysis will be covered in the resonance and

testing for critical speed sections.
6.6 Spectrum Analysis for Steady State Data
 In this section will focus more on Spectrum analysis for steady state vibration data
(fixed running speed – Usually the same operating speed and under the same
operating load). (Refer to 6.2 Second Concept‐Comparison)

 Steps for Spectrum Analysis:

1. Validate the data
2. Identify the running speed
3. Quick Scan for three main regions (Synchronous, non Synchronous, and sub
synchronous)
4. Look for harmonics and subharmonics
5. Look for side bands
6. Check noise floor
7. Trending & comparison concept : review previous spectrums, use water fall
plot to see how the spectrum pattern has changed, and make comparison
with identical equipment.
8. Interpret spectrum pattern against known diagnostics rules (rule or pattern of
Imbalance, rule or pattern of misalignment, rule or pattern of brg faults,…….)

 Note: Try as much as you can to relate what you are seeing in the spectrum to the
TWF, and phase data, for more confidence in diagnosis.
6.6.1 Validate the Data
 Recognize bad data, bad data can be caused by: (refer to section 4.7)

 Sensor Settling time; Insufficient settling time

 Impacting Sensor; saturation or overloading of sensor due to impacting
sensor
 Thermal shock of sensor; sensor is moved from a hot surface to cold or vice
versa
 Bad cable
 Bad cable connection
 Cable movement
 Sensor mounting; Loose sensor mounting
 Sensor Fault
 Bad Set up of data collection
Example of Bad Data
Cont. Validate the Data
 Ensure that the spectrum has peaks and not just noise or electrically related
peaks (at line frequency and multiples), make sure that there is some
information about the mechanical state of the machine.
 Look for classic ski-slope
• Electronic Transient : Insufficient settling time
• Mechanical transient: Sensor shock
• Thermal transient
• Sensor Saturation (overload)
• ………. Refer to section 4.7
 6.6.2 Identify the Running Speed

 Locating the running speed is vital to the successful analysis of vibration spectrum,
Many peaks which can show up in the spectrum are directly related to the running
speed of the machine.

 In many cases the 1X peak will be obvious. When it is not , it may be necessary to
either look for a peak at a known multiple of the running speed and work
backwards (find vane bass or blade pass frequency peak at 6X for example and
then divide this frequency by 6 to locate the running speed frequency 1X peak), or
switch to other machine location where the 1X peak is more obvious.

 Some times you have more than one speed if there is belt or gear box, you will
have to locate the speed for driver (Ex motor) and speed for driven (Ex fan)
1X Peak (Running
Speed Peak)
 Data from vibration
measurement on electric motor
driving fan cooler (The fan
cooler is belt driven)

1X Fan @ 193
CPM

1X Motor @ 1490 CPM

6.6.3 Quick Scan for three main regions
 Quick Scan for three main regions (Synchronous, non Synchronous, and sub
synchronous)

 This is the first step to eliminate the probable faults and to differentiate between
them, by classifying the vibration components into the Synchronous, non
Synchronous, and sub synchronous.

 All the energy in a spectrum can be grouped into one of three categories:
• Synchronous Energy: Energy that is an integer (Whole number) multiple of
running speed EX: 1X, 2X, 3X, 4X, 5X Ex: imbalance, misalignment, bent shaft,
Gear meshing faults, vane (blade) passing faults, base mount looseness,
mechanical looseness, Structural looseness (flexibility).

• Non Synchronous Energy: Energy that is a fractional number of running speed

EX: 1.1X, 2.5X, 3.9X Ex: Roller Element Bearing defects like Inner race, outer race
and ball defects

• Sub Synchronous Energy: Energy that is below running speed EX: 0.48X, 0.5X,
0.6X Ex: Oil whirl, rubbing (in sleeve bearings), belt defects & roller brg Cage
Freq.
 Order normalization for the spectrum (frequency axis to be in orders of running
speed) can help to relate the peaks in the spectrum to the running speed and
identifying if it is Synchronous, non synchronous or sub synchronous

Sub‐
Synchronous
Peaks
Synchronous
Peaks

Frequency Axis is order

normalized
 Quick Scan for three main regions (Synchronous, non Synchronous, and sub synchronous)
 Use Harmonic cursor and/ or order normalized frequency axis

1X (running
Speed)Peak
Synchronous
Peaks Non
Synchronous
Peaks

Sub
Synchronous
Peaks
 Quick Scan for three main regions (Synchronous, non Synchronous, and sub synchronous)
 Logarithmic scale for amplitude can help in showing small peaks

Linear
Amplitude
Scaling

Logarithmic
Amplitude
Scaling
6.6.4 Look for Harmonics and Subharmonics
 Harmonics: harmonics are a series of evenly spaced peaks. Their amplitude
could vary, but the peaks will be spaced at integer multiples of the first in the
series (Fundamental Peak).

 The Fundamental frequency in the series is often referred to as the first

harmonic.

 Harmonics are produced in the spectrum for one of two reasons:

• Clipped (truncated or distorted ) wave form: if there is restriction in one
direction and freedom for structure or rotor to move in the other direction,
like in the case of mechanical looseness or base mounting looseness, or in
the case of pipe strain.

• Transient impacting in the wave form: Transients are a very steep increase
in amplitude with an equally steep decrease in level, possibly with some
ringing afterwards. As the impacting is stronger the harmonics are much
produced and are bigger in amplitudes. Example of Transient impacting
cases are looseness, gear defects , bearing defects. Looseness transient
impacting is synchronous impacting. In severe looseness cases sub
harmonics 1/3 X or 1/2 X may be produces. Bearing defect impacting is
non synchronous impacting.
Harmonics Clipped (Truncated) TWF (Base Mount
will be Looseness Case)
produced
in the
Spectrum
due to
clipping in
TWF

Transients in TWF (Synchronous Impacting) (Base Mount Looseness Case)

Harmonics
will be
produced
in the
Spectrum
due to
Transient
impacting
in TWF
 Transients in TWF (Non Synchronous Impacting‐BPIR impacts) (bearing defect Case)

Harmonics
will be
produced
in the
Spectrum
due to
transient
impacting
in the
TWF
 Conditions which produce harmonics are:
• Looseness (Harmonics from running speed and in severe cases suborder
harmonics 1/3 X or 1/2 X).

• Rubbing (Harmonics from running speed and some times sub order
harmonics depending on the type of rub 1/3 X or 1/2 X or ….)

• Misalignment (Harmonics of running speed, usually 1X, 2X , 3X only but

depending on alignment condition in severe cases may be up to 8X

• Bearing Wear (Harmonics from bearing defect frequency, harmonics from

BPIR, or BPOR, or BSF, or harmonics from 2BSF)

• Gear Faults (Harmonics from Gear meshing or some times called tooth
meshing frequency)

• Belt Wear (harmonics from belt rate or some times harmonics from 2
times the Belt rate)
 Harmonics: The most common harmonics are multiples of the shaft turning
speed. However harmonics can be multiples of any frequency. Harmonics can
be multiples of non synchronous frequency like bearing defect frequency
(BPIR or BPOR, or BSF), and can be multiple of sub synchronous frequency like
belt frequency. As the impacting is stronger more harmonics will be produced
and their amplitude level will increase.

 Sub Harmonics: In severe looseness cases sub order harmonics 1/3 X or 1/2 X
may be produced, the same in the case of rotor rub (in Fluid film bearing
machines) depending on the type of rub sub order harmonics 1/3 X or 1/2 X
may be produced.

 Harmonic Cursor: Use Harmonic cursor in the spectrum to identify harmonics

(Harmonics of motor speed, harmonics of fan speed, other harmonics
mentioned in the previous slide).
 Example of Faults indicated by Harmonics of running speed
 Shaft Misalignment Example

Spectrum is dominated by 1X, 2X, and 3 X component. 2X is the highest in

amplitude.
 Example of Faults indicated by Harmonics of running speed

MECHANICAL LOOSENESS
 MULTIPLE HARMONICS OF RUN SPEED

RUN SPEED HARMONICS

2X 1X, 2X, 3X, 4X, 5X, 6X, 7X, 8X
3X
1X 6X
5X 8X
7X
4X
 Example of Faults indicated by Harmonics of running speed
 Base Mounting Looseness Example
 Base TWF showed Synchronous impacting transients and clipping also, which are the
reasons for harmonics in the spectrum
Mounting
Looseness
Example

Important
Note:

Combine
Spectrum Spectrum showed strong harmonics from the running speed
data with (Fundamental frequency) due to transient impacting and clipping of TWF

TWF Data
 Looseness
Base Mounting mm/s

Looseness:
10
Loose Foundation
3.1
1X Harmonics and
1
sometimes
Sub-harmonics 0.31

½ X or 1/3 X or,…
.5X 1X 1.5X 2X 3X

mm/s
Mechanical Looseness:
Loose shaft 10

1X Harmonics and 3.1

sometimes 1
Sub-harmonics
0.31
½ X or 1/3 X or,…

.5X 1X 1.5X 2X 3X
 Rotor Rub
mm/

10

3.1

0.31

Truncated Wave form

.5X 1X 1.5X 2X 3X

• Symptoms same as Mechanical

Looseness
• Subharmonics ½ ,1/3 etc.
• Strong Harmonic pattern
Caused by truncation
 Example of Faults indicated by Harmonics of Non Synchronous Peak
(BPOR Bearing Defect)

Fundamental Harmonics from

Peak (BPOR) BPOR
6.6.5 Look for Side Bands
 Side Bands: Side bands are produced in the spectrum as a result of
amplitude modulation or frequency modulation

 Amplitude Modulation: This happens when there is a fault with certain

forcing frequency called carrier frequency and the amplitude of vibration
due to this fault is increasing and decreasing periodically at lower
frequency called Modulated frequency as the component defect enters
and exit from loading zone.

Example: A bearing with an inner race defect (notch) the carrier

frequency will be BPIR (Ball pass inner race) frequency and the
amplitude will increase gradually as the inner race rotate and the notch
approaches the loaded zone and then will decrease gradually as the
notch becomes far away from the loaded zone, the following periodic
amplitude increase and decrease will be at modulated frequency which
is equal to the rotational speed of the bearing inner race which is
actually the machine or shaft speed RPM. In this case we will have two
frequency components, Carrier frequency which is the higher
frequency at BPIR and Modulated frequency which is the lower
frequency at shaft speed.
 Example of Amplitude Modulation in TWF due to BPIR (Bearing Inner Race Defect)
Bearing
Inner Race
Defect
Example

Side bands
are spaced Modulation Rate or Modulation frequency is
from the equal to the running speed
carrier
frequency Example of Side bands produced in Spectrum due to amplitude modulation in
peak by a TWF(Bearing Inner Race Defect)
value equal
to the Side or
Carrier Carrier Freq.=
modulated Bands Freq.=BPIR Harmonic of
frequency ????????
BPIR Side
(running
speed in Bands
this
example)
 Side Bands: Side bands are produced in the spectrum in the form of 2 or
more (may be group or family of ) peaks which will be equally spaced around
the carrier frequency peak, and the spacing will be equal to the modulated
frequency.

 Carrier Frequency : Will be the Bearing Defect Frequency or may be

Harmonic from Bearing Defect Frequency (EX: BPIR or harmonic from BPIR in
case of inner race defect), or (Ex: BSF or harmonic of BSF in case of ball fault).

 Modulated Frequency : This will be the running speed in case of inner race
fault and will be the Cage (FTF) rate in case of ball fault.

 Harmonics of Bearing Defects Frequency: In some cases you will find that the
harmonics of the bearing defect frequency in the spectrum are greater in
amplitude than the fundamental peak, specially in early stages of bearing
fault, in these cases you might find the carrier frequency (harmonic of bearing
defect frequency for Example 4 BPIR or 4 BSF) is more clear in the TWF and its
amplitude will be modulated in wave form as mentioned above by running
speed or cage rate depending on type of fault.
Cont. Look for Side Bands
Side band Cursor: Use side band cursor to identify the side bands around
center frequency (carrier frequency)

Side Band
Cursor
Used Carrier
Freq.=BPIR
Gear Box Modulation in TWF & Side Bands In Spectrum ‐Data of Aux G.B of
Example Turbine (GG Speed = Gear Speed nearly 13000 RPM)

LS Gear Rate

HS Gear Rate

Modulation present , Modulated Freq.= Gear Speed

Modulation is present , Carrier Frequency = GM=4048 HZ (242 914 CPM),

Modulated Frequency= 13000 CPM ( GG speed=Gear Speed )
Gear Box Modulation in TWF & Side Bands In Spectrum ‐Data of Aux G.B of
Example Turbine (GG Speed = Gear Speed nearly 13000 RPM)

High
Speed
Modulated
Gear Side
Freq.= Gear
Bands
Speed Carrier
Carrier
Freq.=GM.
Freq.= GMF

Low Speed Gear Side Bands

 Carrier Frequency : Will be the Gear Meshing (tooth Meshing) Frequency or
may be Harmonic from GM Frequency (EX: 2GM or 3GM)

 Modulated Frequency : This will be the running speed of the offending gear.
 Harmonics of GM Frequency: In some cases you will find that the harmonics
of the GM frequency in the spectrum are greater in amplitude than the
fundamental peak, in these cases you might find the carrier frequency
(harmonic of GM frequency for Example 2 GM or 3GM) is more clear in the
TWF and its amplitude will be modulated in wave form as mentioned above
by running speed of offending gear.
 Frequency Modulation This happens when the machine speed changes
periodically (due to for example cyclic process load increase) causing the
forcing frequency to change also periodically, in this case the carrier
frequency (higher frequency) will be the forcing frequency and the
modulated frequency (lower frequency) will be the frequency of periodic
change of speed or load.

Example: A machine with gear drive its speed is changing periodically

due to cyclic change in load causing the gear meshing frequency to
increase and decrease periodically, in this case we will have two
frequency component carrier frequency is the gear meshing frequency
(higher frequency) and modulated frequency which is the rate of change
of speed or cyclic load (lower frequency).
 Cont. Frequency Modulation
 Transform of specific signals: Frequency-modulated signal.
Frequency Modulation Period
Tm F
Fm Fm

FFT

Note: Spectrum of frequency modulation is quite similar to the spectrum of

amplitude modulation except that in frequency modulation there might be
much more side bands
 Conditions which produce Side bands are:
• Bearing Inner Race and Ball faults: BPIR side banded by speed RPM or
BSF side banded by FTF (cage rate).
• Bearing outer race fault in the application where the bearing housing or
outer race rotating and shat or inner race stationary: BPOR side banded
by speed RPM.

• Gear Faults: GM frequency side banded by gear speed.

• Induction Motors: In some cases there is modulation by the pole passing
frequency (PPF side bands) and other cases there is modulation by twice
the line frequency
6.6.6 Look for Noise Floor in the Spectrum
 Noise Floor : The floor of the spectrum in some regions is slightly raised
up, and looks like a sea with small peaks out, like small island mountains
rising out of the sea.

 Noise Floor Lifts up Region: Noise Floor will raise up in the entire
spectrum or in some regions depending on the fault or the reason.

 Some causes of Raised Noise Floor

• Bearing Wear Later or final stages
• Rolling elements skidding in the bearing.
• Resonance
• Pumps Cavitation
• Severe synchronous impacting might cause raised noise floor
• Think about any cause for random impacting
• Think about any cause for friction.
• Other Sources of noise could be neighbor machine or process in the
plant .
 Some Causes of Raised Noise Floor
• Bearing Wear in the later or final stages of bearings just before failure
when the impacting is more stronger due to excessive clearance of the
inner and outer races and simultaneous multiple faults in inner, and outer
and balls all exist, in this cases there will be severe synchronous
impacting (looseness) and there will be random impacting as well, strong
synchronous impacting and random impacting both will add energy in the
entire spectrum or in several regions and might excite resonance in some
areas causing amplification of the noise, all this will cause lifting of the
noise floor up in the entire spectrum or several regions, you will find this
Lifted noise floor in several measurement directions horizontal, vertical ,
axial.

Raised Noise
Floor
 Cont. Some Causes of Raised Noise Floor
• Bearing Rollers Skidding: If a bearing is not correctly selected for its
application or the lubricant is not functioning correctly, or there is no
adequate load on the rolling elements (bearing lightly loaded), then the
rolling elements may slide or skid from time to time. This is more common on
non‐drive‐end bearings, especially on vertical machines, and far more
common with cylindrical roller bearings (as against deep groove ball
bearings). In this case, Look for the raised noise floor in the region of 100‐180
kCPM with BPFO or BPFI peaks protruding above the "hump”.
 Cont. Some Causes of Raised Noise Floor

• Resonance You will find the forcing frequency peak for example 1X or
blade pass or 3X very much amplified and sitting on a hump due to
noise floor amplification. If this hump exists and you find that the
amplification of the forcing frequency is directional (Ex in horizontal
only or in vertical direction only), it is suspected that there is resonance
in this specific area causing increase in the noise floor in this specific
area.

Peak sitting on hump indicates that Natural frequency

exists but not yet so much excited

Note the increase in noise floor

 Cont. Some Causes of Raised Noise Floor

• Pump Cavitation When the fluid cavitate reaching the vapor

pressure and formation of bubbles which burst, this causes random
impacting which appears in the TWF, the random impacting due to
cavitation will add energy to the spectrum floor specially in the
high frequency range from 10X to 30X of the pump speed, so if the
noise floor is based toward the high frequency range you may have
cavitation, process or flow noise.
 Pump
Cavitation
Example
Random Impacting
in time wave form
due to cavitation

Spectrum showed
increased noise floor
due to random Raised Noise Floor
impacting from pump
cavitation
Pump Cavitation Example
Review on Spectrum Analysis for Steady State Data

 Steps for Spectrum Analysis:

1. Validate the data
2. Identify the running speed
3. Quick Scan for three main regions (Synchronous, non Synchronous, and sub
synchronous)
4. Look for harmonics and subharmonics
5. Look for side bands
6. Check noise floor
7. Trending & comparison concept : review previous spectrums, use water fall
plot to see how the spectrum pattern has changed, and make comparison
with identical equipment.
8. Interpret spectrum pattern against known diagnostics rules (rule or pattern of
Imbalance, rule or pattern of misalignment, rule or pattern of brg faults,…….)

 Note: Try as much as you can to relate what you are seeing in the spectrum to the
TWF, and phase data, for more confidence in diagnosis.
Cont. Review on Spectrum Analysis for Steady State Data

Usually Keep in Mind

Related to Repeated
Harmonics (Periodic) Impacting or
Motion Truncation

Related to Modulation
(Bearing Faults, Gear
Side Bands
box Faults, Induction
Motor Faults

Random Impacting or
Friction or Severe
Raised Noise Floor
Synchronous
Impacting
6.7 Time Wave Form TWF Analysis for Steady State Data

 In this section will focus more on TWF analysis for steady state vibration data
(fixed running speed – Usually the same operating speed and under the same
operating load). (Refer to 6.2 Second Concept‐Comparison)

 As mentioned previously the three main techniques for vibration analysis are
TWF, Spectrum, and phase analysis, in some cases one of these techniques is more
efficient in revealing the required information, in some cases TWF is stronger in the
analysis than the spectrum. And as mentioned previously you get the best analysis
results when combining these techniques together
6.7.1 Time Wave Form Settings
 In order to have a good TWF that can be used for effective vibration analysis, the
TWF data collection settings should be well adjusted, three parameters that you
control : (Note: Refer to Signal processing chapter, Section 5.1, TWF settings for
further details about TWF data collection settings)

• First Parameter : Sampling rate (2.56 fmax)

• Second Parameter: No of samples (2.56 No of lines)
• T time span (time window) for the TWF = No of samples /Sample rate = No of
lines /Fmax , so the time span depends on the above two parameters

• Third Parameter whether to integrate the signal or not

 For most data collectors, adjusting TWF data collection settings depends on the
spectrum setting (2.56 rules mentioned above), they collect the wave form with
the spectrum measurement and they save the first or the last wave form during
the averaging process. Other collectors fix the TWF No of samples to be 2048
sample (for an 800 lines spectrum). In another data collectors the TWF is taken as
a separate measure and has separate data collection settings.
Cont. TWF Setting

 TWF Resolution: TWF resolution is controlled by Sampling rate, Time span T (Time
window) for TWF is controlled by No of samples and sample rate (or No of line and
fmax) ; T = No of samples/Sample rate = No of lines/ fmax

 Sampling Rate (related to fmax): is selected based on the machine forcing

frequencies and their harmonics as mentioned previously.

 Typically a time span (Time window) equivalent to 6 ‐10 Revolution of shaft is

required, don’t set the time window too long.

 2048 samples is usually considered enough sampled points.

 Units: For Time wave form, it is usually better to be seen in acceleration units, as
the integration to velocity will suppress the high frequency components related to
the bearings, and you would like to see the bearing defect impacts in the TWF

 Units Note: for fluid film bearings Rules are different where displacement
proximity probes are used, displacement is the unit of choice.
Cont. TWF Setting: Effect of Sampling Rate on TWF Resolution

Overall Amplitude
6 mils PP

Signal appears to
be quite smooth
Digital signal sampled at 640 samples per second or 640 HZ Same
Analog
Signal

Overall Amplitude
8.5 mils PP

Signal shows
some spikes

Digital signal sampled at 25600 samples per second or 25600 HZ

Cont. TWF Setting: Effect of Integration on TWF Pattern (Shape)
Important
Note:
Acceleration
TWF shows
Data from clearly the
same high
measurement frequency
location but in impacting of
different units bearing
(Acceleration
& velocity) for
defected
bearing.
Velocity TWF
See how the is dominated
integration by Low freq.
from components
acceleration to 1X and 2X,
Velocity high freq.
affected the Bearing
shape of the impacts not
TWF. clear
6.7.2 Time Wave Form Patterns
Pure Sin wave Single Frequency
 A single Frequency in the time wave form (1 pure sin wave) produces a single peak
in the spectrum
Imbalance Pattern‐Pure Sine Wave Pattern‐Pure 1X vibration

 Rotor Imbalance force produces 1X per revolution excitation (Pure 1X Vibration)

1X, 2X and 3X Signals in Time wave Form
Note the formation of an M W shape in the wave form due to summation of 1X
with 2X or 3X component, this appears in the time wave form of misaligned
equipment due to presence of 1, 2, 3 X vibration components

1X and 2X
components
Summation

1X and 3X
components
Summation
 Cont. 1X and 2X Signals in Time wave Form
Misalignment Pattern‐1X and 2X vibration
 Summation of the 1X and 2X signals will result in MW shape in the resultant Time
wave form, this will be more clear in the velocity TWF (Low frequency
components), however as mentioned previously it is recommended to see the TWF
in acceleration units if the measurement by accelerometer.

Non Symmetric TWF

data indicates the
motion is constrained
due to misalignment
Cont. Misalignment Pattern‐1X and 2X vibration
Cont. Misalignment Pattern‐1X and 2X vibration
Cont. Misalignment Pattern‐ M W Shapes
 The main component of misalignment are the 1X, 2X, 3X, but depend on the
misalignment condition , there might be 4X and 5X, and in severe misalignment up
to 10X harmonics might be present in the spectrum, these harmonics is produced
due to periodic impacts (constrain in movement and coupling binding).

TWF showed M W
shape due to
presence of 1X and 2X
components
Note the M W shapes

Spectrum
showed 1X and
2X components
Note:

Frequency (Hz) = 1/ T ; T Period (Seconds)

Use double (dual) cursor or Side band Cursor to

measure periods and corresponding frequencies
Beating Pattern in TWF
 Beats: When two waves of slightly different frequency overlap a phenomenon
known as beat results. The effect is a kind of throbbing sensation, which can
sometimes be heard.
 Beats are seen when two equipment are installed beside each other or on the
same skid and operating at slight different RPM, their signals combine together to
form beating
 The closer the two frequencies the lower the frequency of the beats.

 Consider two waves of slightly different frequencies f1 and f2 (f1 > f2) but of the
same amplitude. Figure 1 is a diagram of the two waves and their resultant.

 Beat Frequency ( f ) = f1 ‐ f2
Cont. Beating Pattern in TWF
 f1 = 222 Hz, f2 = 218 Hz, Beat Frequency ( f ) = f1 ‐ f2 = 4 Hz ,
Beat Period (T) = 1 /4 sec
(T)
Example of Beating Pattern‐Interplay between FBP & 1X Motor
Example of Beating Pattern‐Interplay between Motor 2X & 2lf
 The time period between the beats on the below waveform is 0.5 s. From this
information the frequency of the beat is calculated to be 2Hz (120 CPM). This
represents the frequency difference between the two source frequencies In this
case the beat was caused by interaction between a 2 X RPM vibration source and a
2 x fL vibration source on an induction motor.

T = 0.5 s
Beating Pattern in Spectrum
 Some times beating produce sum and difference frequencies in addition to
the two close frequencies
 Beat frequency = difference between the two close frequencies.
 If the spectrum resolution was not high the two close frequencies may
appear as one peak but this peak may be seen rising and falling while
viewing the spectrum life due to beating.
Amplitude Modulation Pattern in TWF
 Amplitude Modulation: This happens when there is a fault with certain
forcing frequency called carrier frequency and the amplitude of vibration
due to this fault is increasing and decreasing periodically at lower frequency
called Modulated frequency as the component defect enters and exit from
loading zone

Example Bearing Inner Race Defect Case : A bearing with an inner race
defect (notch) the carrier frequency will be BPIR (Ball pass inner race)
frequency and the amplitude will increase gradually as the inner race rotate
and the notch approaches the loaded zone and then will decrease gradually
as the notch becomes far away from the loaded zone, the following periodic
amplitude increase and decrease will be at modulated frequency which is
equal to the rotational speed of the bearing inner race which is actually the
machine or shaft speed RPM. In this case we will have two frequency
components, Carrier frequency which is the higher frequency at BPIR and
Modulated frequency which is the lower frequency at shaft speed.
 Example of Amplitude Modulation in TWF due to BPIR (Bearing Inner Race Defect)
Bearing
Inner Race
Defect
Example

Modulation Rate or Modulation frequency is

equal to the running speed

Example of Side bands produced in Spectrum due to amplitude modulation in

Side bands
TWF(Bearing Inner Race Defect)
are spaced
from the Side
carrier
Carrier or Carrier Freq.=
Bands Freq.=BPIR Harmonic of
frequency
peak by a ???????? BPIR Side
value equal Bands
to the
modulated
frequency
(running
speed in this
example)
Cont. Amplitude Modulation Pattern in TWF

Example Bearing Ball Defect Case : A ball with some damage, as it rolls
around, the damaged area impacts the inner and outer race, which
generates vibration (at carrier frequency – Ball spin frequency BSF or
2BSF) but the amplitude of vibration is not constant , because as the ball
rolls into the load zone of the bearing, the impact forces are greater and
as it rolls out of the load zone the force of the impacts are reduced. The
result is the amplitude of vibration rises and falls periodically at frequency
equal to the rate at which the ball moves around the bearing (Modulated
frequency – Cage or Fundamental Train Frequency FTF)
Bearing
Ball Defect
Example

Carrier
Freq.=
2BSF
Modulation Rate or Modulation frequency is equal
to FTF

Side bands
are spaced
from the
carrier Harmonics of
Carrier
frequency BSF
Freq.= 2BSF Side
peak by a
value equal Bands
to the
modulated
frequency
(FTF in this
example)
Cont. Amplitude Modulation Pattern in TWF
Example Misaligned Gears : Another classic case of amplitude modulation is
misaligned gears, the amplitude of tooth meshing (gear meshing) vibration will
be modulated by the rate of the gears turning speeds, the non uniform meshing
of gear teeth due to misalignment will cause amplitude to rise and fall at the
rate of gears turning speeds, the carrier frequency will be the tooth meshing
frequency (Gear mesh frequency GMF) and the modulated frequency will be the
rate of the gears speeds.

Gear Box Example

Modulation present , Modulated Freq.= Gear Speed

Gear Box Modulation in TWF & Side Bands In Spectrum ‐Data of Aux G.B of
Example Turbine (GG Speed = Gear Speed nearly 13000 RPM)
LS Gear Rate
HS Gear Rate

Modulation present , Modulated Freq.= Gear Speed

Modulation is present , Carrier Frequency = GM=4048 HZ (242 914 CPM),

Modulated Frequency= 13000 CPM ( GG speed= High Speed Gear Speed)

Modulated Carrier HS Gear Side

Freq.= HS Freq.= GMF bands
Gear Speed

LS Gear Side bands

Beating Versus Amplitude Modulation
Beating Beat Frequency = 2.1 Hz, period = 0.48 Sec

Amplitude Modulation Modulated Frequency= 49.2 Hz, Period = 0.02 Sec

Cont. Beating Versus Amplitude Modulation
 Beating Frequency is very small because it is the difference between two
frequencies very close to each other (Beating often not more than 4 Hz),
modulated frequency is at higher frequencies.
 Beating has a long period in the wave form compared to the modulation
period.
 Beating in some cases may lead to amplitude near to zero at the end of
each beat cycle (if the magnitude of the two close frequencies are very
near, when they are 180 out of phase they cancel each other).
 Beating does not produce side bands in the spectrum.
Mechanical OR Base Mounting Looseness Pattern in TWF
 Mechanical or base mounting looseness produce repetitive, periodic synchronous
impacting.
 Sometimes you will find the wave form non symmetric (clipped at one direction).

TWF showed Synchronous impacting transients and clipping also, which are the reasons for
harmonics in the spectrum
Cont. Mechanical OR Base Mounting Looseness Pattern in TWF

 Base TWF showed Synchronous impacting transients and clipping also, which are the reasons for
Mounting harmonics in the spectrum
Loosenes
s Example

Spectrum showed strong harmonics from the running speed (Fundamental frequency)
Important due to transient impacting and clipping of TWF
Note:

Combine
Spectrum
data with
TWF Data
 Non Synchronous Transient impacting Pattern in TWF
 Bearing
Non synchronous transient impacting in TWF due to
Outer bearing outer race defect
Race
Defect
TWF showed Non
Synchronous
impacting transients,
which are the reasons
for harmonics in the
spectrum

Impacts at rate of BPOR

Important Fundamental Peak Harmonics from BPOR

Note: (BPOR)

Combine
Spectrum
data with
TWF Data
Belt Defect Pattern in TWF
 When a belt defect happens the defective area makes an impact as it passes over
the pulleys.
 It is common for impacts to show in the wave form at the belt frequency or twice
the belt frequency because the defect passes over two pulleys in one rotation of he
belt.
 Belt Rate = Sheave Diam X Sheave RPM / Belt Length (CPM)

 In the below example ∆T=Period = 0.309 Sec, BF= 1/0.309= 3.236 HZ = 194 CPM
Cont .Belt Defect Pattern in TWF

Belt Defect Impacts at Belt Rate 194 CPM

Example

Important
Note:
Belt Defect is
more clear in
the TWF
(Impacts at
the belt rate)

In this
Example
Defect is not
clear in the
spectrum
 Cont .Belt Defect Pattern in TWF
Another Example : Belt Defect showed in TWF but at 2 times BF
BF = 304 CPM, 2 BF = 608 CPM

Impacts at 2 times the Belt Rate (608 CPM)

Important
Note:
Belt Defect is
more clear in
the TWF
(Impacts at
the 2belt
rate)

In this
Example
Defect is not
clear in the
spectrum
Non periodic (Random) Impacting Pattern in TWF
Pump
 Wave form data shows many pulses (impacts), but this time the pulses are not
periodic, they are random.
Cavitation
Example  It is common to find this random impacting pattern in the case of pump cavitation

Random
Impacting in
time wave
form due to
cavitation

Spectrum
showed
increased
noise floor
due to
random
impacting
from pump
cavitation
Gear Boxes: Broken Cracked tooth Gear Pattern in TWF
 The wave form is very powerful tool when attempting to diagnose gear faults

 time synchronous averaging are also very useful analysis tools.

 For Healthy Gear Box: The time waveform will be largely sinusoidal with limited
modulation and there should be no distinct transients

 Cracked or Broken gear tooth might not be clear in the spectrum, the best way to
detect such fault is from the TWF.

 Cracked or Broken Tooth: If you view a time waveform that contains approximately
6‐10 revolutions of the gear in question then the vibration change as the tooth
comes into mesh should be clear to see.
Gear Boxes: Broken Cracked tooth Gear Pattern in TWF
 The impact wave form in this pattern is occasionally referred to “angel fish”(Figure
below). This distinctive pattern shows an impact followed by smaller “ring down
"peaks giving the time data its pattern.

Synchronous
impacting at the rate
of Speed of the gear
with cracked or
broken teeth
 For Healthy Gear Box: The time waveform will be largely sinusoidal with limited
modulation and there should be no distinct transient
Review on TWF Analysis for Steady State Data

Usually Keep in Mind, Combine with Spectrum Data

Wave Form Patterns:
Probable Causes:
Pure Sinusoidal Imbalance or bent
Shaft if 1X

Probable Causes:
Presence of 2X and 3X
M, W Shapes
with the 1X, probable
misalignment

Mechanical or base
mounting Looseness
Periodic Synchronous
or rubbing, or Broken
Impacting
tooth gear in gear
application
Review on TWF Analysis for Steady State Data
Usually Keep in Mind
Periodic Non Bearings Defects, Belt
Synchronous Defects (at Belt Rate
Impacting Sub synchronous)

Non periodic Cavitation in Pumps

(Random) Impacting Application

Neighbor machine
interfering signal, 2X
and 2Lf in motor
Beating
application due to soft
foot or electric
problem
Review on TWF Analysis for Steady State Data

Usually Keep in Mind

Bearings or Gear
Boxes Faults, Rotor
Modulation
Bar problems in
Motors application
Chapter 7. Fault Diagnosis
7.1 Diagnosing Resonance
7.1.1 Introduction about Resonance in Machinery
 wide variety of fault conditions are either caused by or affected by
“natural frequency” may be up to 50%
 Resonance Definition: Resonance is the act of exciting the natural
frequency.
 If a forcing frequency occurs at or close to a natural frequency
resonance exists. Good design will take in consideration that the
excitation forces (Mach speed, 1X, 2X,…..any other forcing
frequency) will be 20 % away from the machine or structure natural
frequency
 If there is impacting, the impacting inject energy at all frequencies or
at broad band of frequencies which might excite natural frequency.
 Resonances amplify vibration, the measured vibration level may be 3
to 50 times higher than they would be normally, so instead of
vibrating at 0.5 mm/s, for example, the machine could vibrate at up to
25 mm/s.
Cont. 7.1.1 Introduction about Resonance in Machinery

 Natural Frequency:
The frequency of free vibration of a system.
the frequency at which a part likes to vibrate.
 Natural frequencies are frequencies at which a system
(machine or structure) will absorb energy easily leading to
build up of heat, motion, and stresses causing failure.

 For a simple mass and spring system (No

damping) the natural freq:
fn=(1/2 ) √(K/M) Hz
Ѡ n= √(K/M) rad/s
; K is the stiffness N/mm
Cont. 7.1.1 Introduction about Resonance in Machinery
 Natural frequency is a property of a system and is dependent on:
1.Mass
2.Stiffness
3.Damping
 Increasing Mass moves the natural frequency to a lower frequency
 Increasing stiffness moves the natural frequency to a higher frequency
 Increasing damping does not move the natural frequency (does not
move it much-very slight-damped natural frequency) but acts as an
absorber and lowers the amplitude
 In reality, each machine, rotor or structure has several natural
frequencies
Forced Harmonic Vibration Response of SDOF (single degree of
freedom System (Applied harmonic force F=Fo Sin (Ѡt)
Cont. 7.1.1 Introduction about Resonance in Machinery

Critical Speed
 Critical Speed: Critical speed is often defined simply as the speed
that excites a resonance. if 1X (running speed) forcing frequency
excites a natural frequency, this speed is termed a critical speed, and
you should not operate at this speed, Otherwise the machine will
resonate.
 The term critical speed is typically used regarding large rotors such
as turbine rotors.
 Vibration increases dramatically near and at critical speed. As
these machines are run up to their operating speed they must run
through the critical speeds quickly to avoid catastrophic damage.
 The machine should not operate with in 20 % of a critical speed.
Cont. 7.1.1 Introduction about Resonance in Machinery
Mode Shapes
 Mode Shapes: Mode shapes of a system are associated with its
natural frequency. The shape of the system as it vibrates at a natural
frequency is called its mode shape.
Cont. 7.1.1 Introduction about Resonance in Machinery

Mode Shapes For Cantiliver Beam with one Fixed end

Mode Shapes For a rotor

Cont. 7.1.1 Introduction about Resonance in Machinery
Mode Shapes Nodes & Antinodes

Node: Vibration measured at node

will be zero

Antinode: Vibration measured at

antinode will be max (maximum
deflection will be at the antinode )

Antinode node Antinode

node

node

Mode Shapes For a rotor

node
Antinode
7.1.2 Detecting Resonance Problems
Signs Of Resonances
 Unusual Failures: if you have machinery failures that seem to be as
a result of fatigue and you do not have any other explanations, then
you should consider resonance as a possible cause. Example of
failure modes include:
• Broken welds
• Cracked and leaking pipes
• Premature machine failures (bearings, seals,…)
• Broken or cracked shafts
• Foundation cracks.
 Signs in the spectrum: There are three signs in the spectrum
• See Next Slide
Cont. Detecting Resonance Problems
 Signs in the spectrum: There are three signs in the spectrum
• A peak that is so much amplified and sitting on high stack or
hump.
• Directional ( as the stiffness may be different from horizontal to
vertical to axial directions, and so on the natural frequencies)
• A peak that its amplitude changes very much with slight
change in the speed.
Cont. Detecting Resonance Problems
Examples of Spectrums indicating presence of resonance

Peak sitting on broad band hump indicates that

natural frequency exists but not yet so much excited
Cont. Detecting Resonance Problems
Examples of Spectrums indicating presence of resonance

Amplified Peak sitting on hump indicates Resonance

Cont. Detecting Resonance Problems
Resonance Case: Motor Driving Fan cooler (Belt Driven)

Motor speed 1470

Fan Speed 191
Coupling Type Belt
# of blades 7
Motor HP 37 Kw
Motor Bearings DE 6313 SKF
Motor Bearing NDE 6312 SKF
Cont. Detecting Resonance Problems
Cont. Resonance Case: Motor Driving Fan cooler (Belt Driven)
 Resonance at motor structure at 580 CPM, excited by impacting
due to fan excessive mechanical looseness (fan excessive bearing
clearance), 3X fan component vibration (580 CPM) transmitted
through the structure excited resonance (natural frequency exists at
approx. 580 CPM) at motor causing high directional vibration at
motor in direction perpendicular to belt direction at 580 CPM
 Vibration on motor NDE was much higher than the vibration on the
fan (source of excitation) as the resonance was at the motor structure
amplifying the vibration.
 After repair of fan and solving the problem of 3X fan (replacement
of fan shaft and bearing), resonance problem at motor cleared.
Cont. Resonance Case: Motor Driving Fan cooler (Belt Driven)

Motor Vibration Before Fan Problem

Data on motor Broad band Peak sitting on hump indicates that
(in belt Natural frequency exists but not yet so much excited

direction)before
fan problem
occurs

Motor Vibration After Fan Problem

Data on motor Resonance Occurs on Motor Structure when natural

frequency is excited by 3X fan impacting, 3X fan was
(in belt 3X Fan so much amplified (10 times from normal)
direction)after
fan problem
occurs 1X Fan
Cont. Resonance Case: Motor Driving Fan cooler (Belt Driven)

Data on motor (in Motor Vibration After Fan Problem

Direction
Resonance Occurs when natural frequency is excited by 3X
perpendicular to fan, vibration is more directional in the direction
the belt)After fan perpendicular to the belt

problem occurs
Hump under
peak
indicates
resonance

Data on motor (in Motor Vibration After Fan Problem

Direction of the
belt)After fan Resonance Occurs when natural frequency is excited
problem occurs by 3X fan

Hump under peak

indicates resonance
Cont. Resonance Case: Motor Driving Fan cooler (Belt Driven)

Bump Test on Motor

Natural Frequency exist at 540 CPM (value near

to 3X fan 580 CPM)
7.1.3 Special Tests To identify Natural Frequencies
Bump Test
 Striking a machine will inject energy at all frequencies.
 Usually a soft tip hammer is used. A piece of hard wood timber
works well but you should be careful not to damage the machine or the
structure. Soft tip inject more energy at lower frequencies. A metal bar
inject more energy at higher frequencies As the stiffness of the hammer
tip increases more energy is injected at higher frequencies and less
energy at lower frequencies and vice versa.
 Striking the machine will inject energy at all frequencies, this will
cause the floor across the entire frequency range to be raised, and so
natural frequencies will be excited amplifying that noise floor and
causing peaks and humps in the spectrum.
 Bump test can be done while machine is running, but best results from
bump test is normally obtained while the machine is off.
 The vibration resulting from the strike dies quickly, hence should be
captured quickly before it dies, data collection time (time span) should
be small, see next slide for recommended settings for data collection.
Examples of Spectrum and TWF of Bump Test
Cont. Bump Test
Examples of Spectrums of Bump Test

Natural Frequencies
Cont. Bump Test
Examples of Spectrums of Bump Test

Natural Frequencies
Cont. Bump Test‐ Recommended Settings for Bump Test.
 Use Rectangular (uniform-as no windowing) window
 Set the analyzer to peak hold averaging
 Low resolution data is ok (400 lines or less), to get fast data collection
t = No of lines/ fmax to capture vibration before it dies. Low resolution is ok as
you don’t need to know the exact value of natural frequency, you need to know
an approx. value, excitation force should be 20 % away form natural freq.
 Select a higher frequency range, also for fast data collection.
 No of averages can be set to 20 averages, Note you should not strike more than
once per time window or average. So try not to make double impacting in one
average, this will cause distortion and incorrect peaks may result in the
spectrum.
 If possible, set an appropriate gain setting and turn off the auto-ranging feature.
If auto-ranging is left on , additional strikes will have to me made to allow the to
analyser adjust the gain.
 Some analyser have a pre-trigger feature , setting this to 30 or 40% moves the
impulse of the Bump to the centre of the time window and therefor the measured
vibration is not affected by the windowing.
Cont. Bump Test‐ Recommended Settings for Bump Test.

Remember (Some points to be taken in

consideration during the test)
 Machine vibrate in three axes, unless it is already
suspected that the problem is dominant in one direction, then
the tests should be done for the three directions.
 Not to strike at a node and not to measure vibration at a
node as there will be no vibration to measure, you have to
make several trials to avoid nodes which may hide the
response.
 1 strike per each average and avoid double impact, there
will be no problem if one or two average missed the strikes
(No strikes), but the reverse if the striking rate is very quick
that it causes more than one strike per average, this may
cause unreal data or peaks in the spectrum
Cont. 7.1.3 Special Tests To identify Natural Frequencies
Impact Test and Frequency Response Function FRF
 For the impact test, same as bump test , the structure is impacted by a
special hammer for impact test (Impact hammer is equipped with
load cell to measure the input force).
 Two (Cross) channel measurement will be required, one channel will
be connected to the impact hammer to measure the input (impact) force,
and the other channel will be connected to the vibration accelerometer to
measure the vibration response due to the impact force.
 In the impact test, we will measure the FRF at different structure points
. The FRF is the ratio of the output to the input, i.e. is a measure of the
output vibration response due to the input excitation (impact) force. We
will get the spectrum of the FRF in units of vibration/force , EX g’s/N,
or Ex: ips/kgf. FRF is some times called TF (transfer function).
 Also phase will be measured, phase between the trigger which is the
impact force and all vibration components (at all frequencies of
interest), the result that you will have the FRF including amplitudes of
vibration per unit force and the phase readings at all frequencies of
interest.
Cont. Impact Test and Frequency Response Function FRF
 Natural frequency will be identified from amplification of the response,
there will be a considerable phase shift. If you are measuring at an
antinode point on the structure for certain mode ( certain natural
frequency), the amplification of the response will be clear, the phase
shift will be 180 from before natural frequency to after natural frequency
with 90 Deg phase shift at the natural frequency.
 Impact test is also used in the modal Tests (analysis) where the test is
done at large number of points of the structure and measurement are
taken for every point at three axes (H, V, A ), all the measured FRF
(magnitudes and phases) at all points are then imported to modal analysis
software to obtain the natural frequencies (modes), and mode shapes
(shape of structure deflection) at each mode, and animate the modal
response (mode shape vibration). and estimate also the amplification
factor or damping values at each mode as well.
 A difference between Impact test and bump test , is that the impact
test will confirm the natural frequency by seeing amplification and
180 Deg phase shift, but bump test will indicate natural frequency but
not 100 percent confirmation for that.
Cont. Impact Test and Frequency Response Function FRF
Cont. Impact Test and Frequency Response Function FRF
Cont. Impact Test and Frequency Response Function FRF
Cont. Impact Test and Frequency Response Function FRF

Bode Plot of Amplitude and Phase of a FRF function. Amplitude has peaks
corresponding to natural frequencies/resonances of test object. Phase has shift at
resonant frequency.
Remember (Some points to be taken in
consideration during the test)
 Machine vibrate in three axes, unless it is already
suspected that the problem is dominant in one direction, then
the tests should be done for the three directions.
 Not to strike at a node and not to measure vibration at a
node as there will be no vibration to measure, you have to
make several trials to avoid nodes which may hide the
response.
Cont. 7.1.3 Special Tests To identify Natural Frequencies
Run Up / Coast Down Tests with order tracking
 The purpose is to use the running speed (1X of residual imbalance)
during the run up or coast down to excite the natural frequencies, as
the 1X coincides with each natural frequency it will excite the natural
frequency and its amplitude will be very much amplified and the 1X
phase will be shifted 180 deg from well before to well after resonance
and the 90 deg phase shift will be at the natural frequency.
 The analyzer should be supported by order tracking, 1X tracking
filter will be used, to apply a filter around the 1X value and track its
frequency and amplitude as it changes with different speeds during run
up and coast down, the tracking filter will need that a tachometer or
key phasor to be present for speed measurement.
 Also analyzer should support transient data collection, for fast data
collection (fast sampling rate based on machine speed, sample rate will
be No of samples per revolution, for example sample rate 128
sample/rev or 64 sample /rev, sample rate will depend on machine
speed)
Cont. Run Up / Coast Down Tests with order tracking

 Several plot formats will support in detecting natural frequencies

from transient (run up /coast down data ), example of these plots are
bode plot, polar plots, cascade and water fall plots.
 Bode and polar plots will confirm the presence of natural frequency
(critical speed) since they will show the amplitude amplification and the
180 Deg phase shift.
 Cascade and water fall plots will just indicate natural frequencies but
will not confirm it hundred percent.
Cont. Run Up / Coast Down Tests with order tracking
Run Up / Coast Down Plot Formats‐Bode and polar Plots

Bode Plot Polar Plot

Bode Plot for 1X , direct amplitudes and 1X phase , and polar plot for 1X amplitude and
phase during coast down Coast down , data showing presence of natural frequency (or
critical speed –or mode) at 4500 CPM, note the amplification and 180 phase shift.
Run Up / Coast Down Plot Formats‐polar Plot

The same
example of
the previous
slide, but
focus on
polar plot.

Note the amplification in 1X amplitude and the 180 Deg phase shift with 90 Deg shift at
the critical speed.
Run Up / Coast Down Plot Formats‐Bode ‐ polar Plot

Bode Plot Polar Plot

 Bode Plot for 1X , direct amplitudes and 1X phase , and polar plot for 1X amplitude
and phase during coast down Coast down , data showing presence of natural
frequency (or critical speed –or mode) at 7900 CPM, note the amplification and 180
phase shift.
 See cascade plot for the same coast down data in the next slide.
 Run Up / Coast Down Plot Formats‐Cascade Plot‐Coast down
data for the same previous example in previous slide

The
same
example
of the
previous
slide,
but
cascade
plot data

 Amplification appears in the range of the 8000 CPM , quite near to what has been
seen in bode and polar plots in the previous slide.
Run Up / Coast Down Plot Formats‐Water Fall Plot‐Run up data

Waterfall Plot showing Critical Speeds During Startup Test

Cont. 7.1.3 Special Tests To identify Natural Frequencies
Changing machine speed – running up – or coasting down –
with out order tracking and without phase data‐by using No
of peak hold averaging
 You can try this method if your analyzer does not support transient data
collection and order tracking, specially if you can control the speed of
the machine during the run up or coasting down, to ramp up or coast
down slowly, and you set up the analyzer to collect a lot of number of
averages with peak hold averaging, fast collection setting (low
resolution), as the 1X running speed or any excitation force coincides
with the natural frequency they will be amplified, the resultant spectrum
will be like that of the bump test showing amplification of peaks around
each of the natural frequencies.
7.1.3 Special Tests To identify Natural Frequencies
Modal Testing
 This is out of the scope of the course
 But a quick hint.
 Modal testing is applied to identify the natural frequencies ( modes) and
the associated rotor or structure mode shape with each mode, and
animate the mode shape, in addition to estimation for the amplification
factor (or the damping ratio) at each mode.
 By viewing and animating the mode shape, and visualizing location of
nodes and antinodes, you can visualize how the resonance problem can
be solved and where is the points that need stiffening by braces or
supports.
 Modal testing depends on the impact test that will be done at huge
number of points in three axes (H-V-A) and FRF measured (Magnitude
and phase at all these points, all the measured data and the drawing (with
dimension), will be input to modal analysis software to identify the
modes and animate the mode shapes and calculate the other parameters
(amplification, damping).
Cont. Modal Testing‐Mode Shapes
Cont. Modal Testing‐Mode Shapes‐Rigid & Flexible
(Bending) Modes
Cont. Modal Testing‐Mode Shapes‐Rigid & Flexible Modes
7.1.4 Correcting Resonances
Solving Resonance Problems

fn=(1/2 ) √(K/M) Hz
Ѡ n= √(K/M) rad/s
; K is the stiffness N/mm
Cont. 7.1.4 Correcting Resonances
Cont. Solving Resonance Problems
 As mentioned in previous slide that natural frequency depends on
properties of the system (mass and stiffness), and the amplification of
vibration when resonance happens depends on the amount of damping
in the system. And resonance occurs when forcing frequency happens to
coincide with natural frequency.
 Resonance can be solved with one of ways of below ways depending on
the situation.
• Preventing the excitation, correcting the fault causing excitation,
this fault could be any mechanical problem, for example bearing
fault, belt wear, ….etc.
• Altering the excitation force, for example if the excitation force is
due to blade passing frequency and I can make slight change in the
speed to shift the blade passing frequency away from the natural
frequency, but this could be in applicable from the process point of
view. It can be applied as a temporarily solution.
• Cont. in next slide
Cont. Solving Resonance Problems
• Altering the natural frequency by changing mass or stiffness, so that
the excitation forces will no longer coincide with the natural frequency,
adding mass will shift the natural frequency down , adding stiffness
(bracing or supporting the structure) will shift the natural frequency up,
typically adding stiffness (supports or braces) is preferred than adding
mass, because lowering the natural frequency (by adding mass) may
make it coincide with other forcing frequencies, it is usually
recommended (if applicable) to move the natural frequency away from
the present forcing frequencies. Stiffening the structure is the most
used in solving resonance problems
• Isolating the Excitation Vibration from reaching resonant structure
by using of vibration isolators, Example 1X motor vibration causing
resonance of the mounting structure of the motor, and the solution of
stiffening the structure is not applicable, motor can be mounted on
isolators under the motor feet, which will isolate the motor vibration
(excitation) and reduce it very much that it does not reach the structure.
• Cont. next slide
Cont. Solving Resonance Problems
• Adding Damping Material to Reduce the amplification,
Adhesive damping material are attached or painted on the structure
to add damping to the structure.
• Altering the Natural frequency by use of dynamic absorbers or
tuned mass dampers mounted on the structure.

 Stiffening the structure is the most used solution in solving resonance

problems, when it is not applicable, we begin to think about other
solutions like isolators, damping material, and dynamic absorbers or
tuned mass dampers
Cont. Solving Resonance Problems
 Stiffening the structure is the most used solution in solving resonance
problems, when it is not applicable, we begin to think about other
solutions like isolators, damping material, and dynamic absorbers or
tuned mass dampers
 Selection of location of stiffening adding supports or braces either by
using modal Testing (analysis) and understanding the mode shape, or
Finite element analysis FEA model to propose design modification and
calculate the effect of bracing or supporting on shifting the natural
frequency, or operating deflection shape analysis ODS to visualize
how the structure is vibrating in order to estimate the required structural
modification. See Next Slide Hint about ODS and FEA.
 Selection of support location, If you don’t have the resources to go
through the above approaches, this will need experience and well
understanding of the resonance problem, in order to propose and try a
solution, Wrong estimation may make the problem worse.
 Typically the bracing or supports are added between a point that is
not moving (the foundation or the base) and the point of the structure
where the vibration is highest ( antinode).
Cont. Solving Resonance Problems
Examples of Structure Stiffening by supports
Cont. Solving Resonance Problems
Finite Element Analysis‐Just Hint
 FEA is an analytical technique that utilizes a mathematical model of
a structure to predict its natural frequencies, damping and mode shapes,
once the FEA model has been created, it is used to evaluate the
effectiveness of proposed structural modifications.
 A lot of information is required to build the model, these information
include dimensions, material properties, stiffness's, masses,…etc.
Cont. Solving Resonance Problems
Operating Defection Shape ODS analysis ‐Just Hint
 ODS is an application of cross channel (relative phase)
measurement, the ODS show how the vibration (or the structure) deflect
under operating conditions due to vibration at 1X running speed or
ODS can be created for the vibration at any offending frequency.
 One of the application of ODS when you believe that there is a
resonance problem and you want to understand the operating deflection
shape under normal operating conditions, in order to propose a certain
structural modification.
 ODS does not confirm resonance it is just measurement for relative
phase (at certain frequency, may be 1X or other) of many points on the
structure (at H-V-A directions) and a reference point on the structure, all
this vibration data together with the drawing and dimension of the
structure is imported to an ODS software for creating and animating the
operating deflection shape.
Cont. Operating Defection Shape ODS analysis ‐Just Hint
7.2 Diagnosing Unbalance
7.2.1 Introduction about Unbalance
Unbalance Definition:
 Unbalance is a condition where shaft’s geometric (rotational) centerline
and mass centerline do not coincide.
 The non uniform distribution of a rotor mass about its geometric centerline
 A condition in which a rotor imparts vibration force to its bearings as a
result of centrifugal forces.
• Fc = mrw2 = M e w2
• Fc = centrifugal force
• m = mass (imbalance mass) & M = rotor mass
• r = radius of imbalance mass from center of rotation
• e = eccentricity (distance) between shaft geometric
center and center of gravity (Mass)
• w=angular rate (angular speed) rad/s
; w=2 N/60 & N is the rotational speed (RPM)
Cont.7.2.1 Introduction about Unbalance

Unbalance Condition

Center of Shaft = Center of Mass

Heavy Spot

Center of Mass Center of Shaft

Rotation
Cont.7.2.1 Introduction about Unbalance
Causes Of Unbalance
 Damaged components (check condition before attempt to balance)
 Uneven dirt accumulation (on fans)
 Uneven corrosion or erosion of a rotor
 Missing balance weights
 Missing stud bolt, washer,… (check, parts Example coupling before
attempt to balance)
 Uneven mass distribution of electrical winding (wounded rotor motors)
 Manufacturing or design issues-(variation within material variable
density)-Non symmetry (build up rotor) -Key and keyways.
 Manufacturing defects (from machining or casting)
Cont.7.2.1 Introduction about Unbalance
Results Of Unbalance
 Put stresses on bearings and seals
 May cause other fault condition like rubbing.
 Fatigue in support structures
 Decreased product quality
 Power losses
 Disturbed adjacent machinery

Balancing Definition:
 The adjustment of the mass distribution in a rotating part so that the
axis of rotation and mass centerline of the rotating part are coincident.
7.2.2 Unbalance Vibration Characteristics
 High radial peaks (Horizontal and vertical directions) at 1X in the spectrum
(because imbalance cause high radial centrifugal force).
 Pure 1X vibration spectrum to a great extent.
(low amplitude values of harmonics of 1X could be present from other
residual faults, like misalignment, looseness, bearing or shaft asymmetry,
gravity effect in overhung disks or rotors,…etc.).
 Horizontal and vertical (radial) vibration amplitudes are quite similar,
since the imbalance centrifugal force is acting in the radial direction, the force
m r w2 will cause quite similar vibration in the horizontal and vertical
directions (There could be some difference in horizontal and vertical vibration
levels as the machine stiffness could differ from horizontal to vertical
direction).
 1X Amplitude is proportional to the square of the speed to some extent, so
if the speed is doubled you expect that the vibration will be increased by a
factor of four.
 Low axial vibration at 1X in the spectrum (except for overhung machines)
 Cont. next slide
Cont.7.2.2 Unbalance Vibration Characteristics
 Time wave form will be sinusoidal to a great extent (1X component)

 Relative (cross) 1X Phase Readings between horizontal and vertical

direction will be nearly 90 degrees (90 ±30) at the same bearing.

Or 270 Deg.
IMBALANCE
RADIAL VIBRATION
AT 1 X TURNING
SPEED IS MUCH
MORE SIGNIFICANT
THAN IN THE AXIAL
DIRECTION

Freq: 60.04
Order: 1.010
Spec: .390
Cont.7.2.2 Unbalance Vibration Characteristics
Example of Imbalance Case: Bleed (Seal)Fan Motor
MNDE H Spectrum
MNDE V Spectrum
Dominant (Pure) 1X Dominant (Pure) 1X

MNDE H TWF MNDE V TWF

Note that: Horizontal and vertical probes 1X vibration level is quite near
TWF is nearly Sinusoidal.
7.2.3 Unbalance Types
Unbalance Types:

 Static Unbalance

 Couple Unbalance

 Dynamic Unbalance
Cont. 7.2.3 Unbalance Types
Static Unbalance
 An imbalance condition where the mass centerline of a rotating part is
parallel to the axis of rotation but offset from it. This causes 'in-phase‘
repeating forces to act on the support bearings i.e. the force acting on one
bearing is always pointing in the same direction as that acting on the
other bearing. As a result, all points on the rotating part vibrate in a
synchronized manner (in phase) . Static imbalance can be corrected by
adding one correction weight to the appropriate location on the rotating
part.
 Static unbalance is equivalent to a heavy spot at a single point in the
rotor.
 Static unbalance, also called force unbalance
 The name static unbalance is because it will show up even if the rotor is
not running, if the rotor is placed on two frictionless knife edges, it would
rotate and come to rest when the mass or the heavy spot, is on the bottom.
Cont. 7.2.3 Unbalance Types
Cont. Static Unbalance

static imbalance -
vibration across machine
“in phase”

The relative phase between H-V

vibration at each brg location is
90 Degree (or 270 Deg.)

Note: For all phase rules, there

is tolerance ± 30 Deg.
Cont. 7.2.3 Unbalance Types
Couple Unbalance
 An imbalance condition where the mass centerline of a rotor is not parallel to
the axis of rotation but intersects it at center of gravity. This is caused by two
heavy spots one located at each end of the rotor and which are on opposite sides of
the rotor surface. When rotated, the centripetal forces associated with the
oppositely positioned heavy spots give rise to a couple that rotates at the rotational
speed of the rotor. The rotating couple in turn causes out-of-phase repeating
forces to act on the support bearings i.e. the force acting on one bearing is
always pointing in a direction opposite to that acting on the other bearing. As
a result, the rotor rocks from side to side (out of phase vibration from side to
side) . Couple imbalance can be corrected by adding two correction weights to
the appropriate locations on the rotor.
 Couple unbalance happens if there are two equal unbalance masses on the
opposite ends of the rotor, but 180 degrees opposite each other.
 A rotor with couple unbalance may be statically balanced (it may seem to be
perfectly balanced in a stationary condition when placed in frictionless bearings).
But when shaft operates (rotates), it produces centrifugal forces on bearings , and
they will be opposite phase.
Cont. 7.2.3 Unbalance Types
Cont. Couple Unbalance

couple imbalance -
vibration across machine
“180 degrees out of
phase”

The relative phase between H-V

vibration at each brg. location
is 90 Degree (or 270 Deg.)

Note: For all phase rules, there

is tolerance ± 30 Deg.
Cont. 7.2.3 Unbalance Types
Dynamic Unbalance
 An imbalance condition involving both static and couple imbalance.
The mass centerline is both offset from and not parallel to the axis of
rotation. Most cases of imbalance in machines are dynamic imbalance.
Dynamic imbalance can not be corrected in one plane
 Dynamic unbalance can not be corrected in one plane
Dynamic imbalance -
vibration across machine
(end to end)
“are neither in phase nor
180 out of phase”

The relative phase between H-V

vibration at each brg. location
is 90 Degree (or 270 Deg.)

The relative phase between H-H vibration (end to end) will be any angle,
relative phase V-V will be any angle, but these angles will be quite near
7.2.4 Unbalance of Vertical Machines
 Vertical machines, such as pumps, are usually cantilevered
from their foundation, and they usually show maximum
1X levels at the free end of the motor regardless of
which component is actually out of balance
 The spectrum again will show a strong 1X peak when
measured in the radial direction (horizontal or tangential),
the piping may cause some stiffness in their direction, so
you may find horizontal vibration much greater than the
tangential vibration.
 The pattern of imbalance in the case of vertical
machines will be quite near to the static imbalance in
the relative phase relation ship
 Rel. 1X phase H-H will be in phase (0 ± 30) Deg
 Rel 1X phase V-V will be in phase (0 ± 30) Deg
 Rel 1X phase H-V phase may be 90 (Deg if circular
motion i.e. H-V vibration levels are quite near, if stiffness
differ very much in the two directions, you might find the
vibration so much directional and causing relative phase H-
V to be near to 180 Deg. (In this case combined structural
looseness i.e. flexibility and imbalance problem together).
7.2.4 Unbalance of Overhung Machines
Cont. 7.2.4 Unbalance of Overhung Machines

 A high 1X in radial and axial directions

 The high 1X in axial is present because unbalance creates a bending moment on the
shaft causing bearing housing to move axially.
 Rel 1X Axial phase readings from brg to brg. tend to be in phase (be careful of
probe orientation-See note in Next slide).
 For Rel 1X Radial phase readings from brg to brg , No rule because often will be
dynamic imbalance - combination of static and couple, if one component whether
static or couple is dominant its relative phase rules will be present.
 Rel 1X phase at the same brg H-V will be 90 (± 30) Deg
 Overhung rotors often have both force (Static) and couple unbalance, each of
which may require correction, correction is best carried out by static couple
balancing method.
Important Note about Axial Relative Phase Measurement:

 Generally we should take into account the phase

difference that is attributed to the pick up direction. So
be careful of the probe orientation during axial phase
measurement, if you are not able to put the 2 probes in
the same axial direction, you should compensate the
measured rel. axial phase readings by adding or
subtracting 180 Deg.
7.2.5 Rotor Dynamic Balancing Hint.
Prior to Balancing
• Is Balance the Problem?
• Perform a complete analysis
prior to balancing
• Remember rotating Rotn
component did not go out of
balance by itself (check
cond.)
Beware of false indicators -
Misalignment & Resonance O
Resonance and Flexible 180 O
Structures Complicate the Deg. Deg.
Picture
Equipment Required
Vibration Transducer
Once-Per-Revolution Sensor
Filter capable of measuring
Phase
Speed, Amplitude, and Phase
Marker or Paint Stick
Polar Graph Paper and
Triangles
Balance Weights and Scale
 Cont. 7.2.5 Rotor Dynamic Balancing Hint.
Example Single Plane Balancing
O = 5 mil @ 190 Deg
TW= 75 gram @ 30 Deg
O+T = 3 mil @ 150 Deg Rotn
T = 3.4 mil @ 46 Deg
; T is the effect vector

O
180 Deg. O Deg.

CW = TW * O/T = 75 * 5/3.4 = 110.29

grams
CW = 110.29 grams @ 354 Deg.

Notes
• U (Actual Heavy spot before balancing) = Phase
110.29 grams @ 174 Deg.
• Influence vector I = T vector/ TW vector = 0.0453 mil/gr
Angle of Influence vector (phase lag) =16 Deg
• Actual Heavy spot is leading high spot O vector by 16 Deg.)
• TW is leading its effect T vector by 16 Deg
• CW is leading its effect (‐ O vector ) by 16 Deg
 7.3 Diagnosing Eccentric Pulleys (in Belt Driven Machines)
 Eccentricity occurs when the center of rotation is offset from the geometric
centerline of a sheave (pulley).
 Eccentric sheaves will generate strong 1X radial components, especially
in the direction parallel to the belts. This condition is common, and
mimics unbalance.
 There will be a high 1X vibration level on both components (motor and fan
for example), however due to the change in speed, these will be at two
different frequencies.
 Spectrum: It is best to collect spectra parallel and perpendicular to belts
instead of pure vertical and horizontal. Expect a high 1X peak. Amplitude
will be higher in the direction parallel with belts.
 Waveform: The waveform will be very sinusoidal when viewed in velocity.
 Phase: Phase readings taken parallel and perpendicular to belts will either
be in phase or 180° out of phase.
Cont. 7.3 Diagnosing Eccentric Pulleys (in Belt Driven Machines)
7.4 Diagnosing Misalignment
7.4.1 Introduction about Misalignment
Misalignment Definition:
 Shaft are misaligned when their rotational centerlines are not collinear
when the machines are operating under normal conditions

parallel misalignment angular misalignment

Cont.7.4.1 Introduction about Misalignment
Misalignment Effect:
 Misalignment is the root cause of majority of machine breakdowns:
bearing failures, damaged seals, shafts and couplings.
 Misalignment causes increase in power consumption
 It is important to check prior to alignment, that the machine has good
foundation, there are no soft foot problems, and there are no looseness,
run out, or other problems.
Misalignment Types
 Parallel (Offset) Misalignment: Misaligned shaft centerlines are parallel,
then the misalignment is called parallel (Offset) misalignment.
 Angular Misalignment: Misaligned shaft centerlines meet at a point, then
the misalignment is called angular (Gap) misalignment.
 Almost all misalignment seen in practice are a combination of these two
types
parallel misalignment angular misalignment
7.4.2 Misalignment Vibration Characteristics
 Resulting vibration depends on the radial stiffness of the components
(bearings, shafts, seals, couplings).
 Spectrum The main component of misalignment are the 1X , 2X or 1X,
2Xand 3X. Also depend on the misalignment condition , there might be 4X
and 5X, and in severe misalignment up to 10X harmonics might be present
in the spectrum, these harmonics is produced due to periodic impacts
(constrain in movement and coupling binding).
 2nd. Order may exceed 1st. Depending on misalignment severity and
misalignment type (offset or angular).
 M, W shapes in TWF, Specially velocity TWF
 High 1X axial vibration (out of phase) across coupling is a symptom for
both types of misalignment (parallel or angular) (See Note in Next Slide).
 Relative phase between vertical and horizontal will not be 90 Deg as the
case of imbalance but may be 0 or 180 in some cases. (not to be confused
with resonance cases).
Remember about Axial Relative Phase Measurement:

 Generally we should take into account the phase

difference that is attributed to the pick up direction. So
be careful of the probe orientation during axial phase
measurement, if you are not able to put the 2 probes in
the same axial direction, you should compensate the
measured rel. axial phase readings by adding or
subtracting 180 Deg.
Cont. 7.4.2 Misalignment Vibration Characteristics

Spectrum
showing 1X ,
2X and 3X
components
Cont. 7.4.2 Misalignment Vibration Characteristics
Cont. 7.4.2 Misalignment Vibration Characteristics

TWF showed M W
shape due to
presence of 1X and 2X
components
Note the M W shapes

Spectrum
showed 1X and
2X components
 MISALIGNMENT

2X IS DOMINANT
LOCATION OF
VIBRATIONS

VANE PASS IS DUE TO 11

BLADE PUMP IMPELLER
VANE PASS

Freq: 59.50
Order: 2.010
Spec: .640
7.4.3 Diagnosing Offset (Parallel) Misalignment
 It produces high radial vibration (vertical and horizontal).
 Radial vibration across coupling 1X rel. phase is close to 180 deg..
 Axial vibration across coupling 1X rel. phase is close to 180 deg.
 2x is often larger than 1x. Difference is subject to coupling design/type.
 Axial 1X and 2X amplitudes will be low for pure (dominant) parallel
misalignment
 Check Radially
Spectrum in
radial direction
showed high 1X
and 2X and 3X
components
2X is often the
larger
1X 180 out of
phase radially

Spectrum in
Axial direction
showed low 1X,
2X, 3X
1X 180 out of
phase axially
Radial (and axial 1X phase also) 1X rel. phase approx. 180 ° phase
shifted across the coupling
2X often highest peak
7.4.4 Diagnosing Angular (Gap) Misalignment
 High axial vibration
 Axial vibration across coupling 1X rel. phase is close to 180 deg.
 Radial vibration across coupling is nearly in phase 1X rel. phase is close to
0 deg.
 1X is often larger than 2X and 3X
 Radial vibration will also be fairly strong.
 Check axially.
Spectrum in
axial direction
showed high 1X
and 2X and 3X
components
1X is often the
larger
1X rel. phase
are 180 out of
phase axially

Spectrum in
radial direction
showed fairly
high 1X, 2X,
1X rel phase
approx. o deg.‐
in phase
radially
7.4.5 Diagnosing Severe Misalignment
 Severe misalignment will cause higher order components to show on the
spectrum (up to 8x).
 It could go even to a rich spectrum similar to that caused by mechanical
looseness.
 Coupling design would greatly influence the shape of the spectrum in case
of severe misalignment.
7.4.6 Distinguish Misalignment from Imbalance

Using Phase to distinguish Misalignment from

Imbalance
Cross Coupling Phase

180 deg out In Phase

GRRR!
Distinguishing Unbalance from
Misalignment
Unbalance Misalignment
High 1X response in frequency
High harmonics of 1X relative to 1X.
spectra.
Low axial vibration levels. High axial vibration levels.

Measurements at different Measurements at different locations

locations are in phase. are 180° out of phase.
Vibration levels are dependent on
Vibration levels are independent
temperature. (change during warm
of temperature.
up)
Vibration level at 1X increases Vibration level does not change with
with rotational speed. rotational speed.
Centrifugal force increases as the Forces due to misalignment remain
square of the shaft rotational relatively constant with changes in
speed. shaft rotational speed.
7.4.7 Distinguish Misalignment from Looseness

 Severe misalignment will cause higher order components to show on the

spectrum (up to 8x).
 It could go even to a rich spectrum similar to that caused by mechanical
looseness.
 Misalignment harmonics will not be as strong as those of looseness
 In Misalignment spectrum noise floor will not be increases, but looseness
with its strong impacts adds energy to the entire spectrum floor and
excites some resonances in the system causing humps.
 In looseness case , phase will be erratic.
7.5 Diagnosing Cocked Bearing
 Cocked Bearing is a form of misalignment but couldn’t be corrected by
shaft alignment, bearing must be removed and reinstalled correctly.
 Bearings on one side of coupling are misaligned
 Not mounted in the same plane
 Not normal to shaft
Cont. 7.5 Diagnosing Cocked Bearing
 Generate considerable axial vibration at both bearings.
 Axial vibration phase difference of same bearing top to bottom and/or
side to side of nearly 180 degrees.
 Bearings must be removed and installed correctly.

Axial Phase Side to side

or top to bottom

Or
7.6 Diagnosing Soft
 Soft foot is a condition that occurs when not all of the machine's feet
lay on the same plane. They may not be all the same height or it there
may be bent (angular soft foot).
 When the machine is bolted down a distortion takes place which causes
an increased vibration level.
 This condition will affect alignment and will appear on the spectrum as a
high peak at 1X RPM in the radial direction and often 2X and 3X
component as well.
Cont. 7.6 Diagnosing Soft
 For AC motors, the soft foot will cause non uniform air gap
between the rotor and the stator causing unbalanced magnetic
forces which will cause dominant vibration at twice the line
frequency (2 Lf ) in the radial direction. Hi resolution spectrum
will be required to discriminate between 2Lf and 2X for motors
(3000 rpm nominal speed) or between 2lf and 4X for motors
1500 rpm nominal speed
 For Pumps, the soft foot may cause uneven clearance between
the rotating vane and the diffuser (housing) which may cause
dominant vibration at Vane/Blade pass frequency.
 This condition will affect alignment and will appear on the
spectrum as a high peak at 1X RPM in the radial direction and
often 2X and 3X component as well.
Cont. 7.6 Diagnosing Soft

AC induction Motor
Dominant 2 Lf vibration
due to soft foot problem

 Hi resolution spectrum will be required to discriminate between 2Lf

and 2X for motors (3000 rpm nominal speed) or between 2lf and 4X
for motors 1500 rpm nominal speed
7.7 Diagnosing Bent Shaft
 Causes high axial vibration
 Across rotor axial phase difference (end to end) close to 180 degrees
 Across rotor , radial vibration (end to end) will be in phase
 Dominant component is at 1X when the bent is near to the centre of the
shaft
 Dominant component is 2X if bent is near to the coupling.
Cont. 7.7 Diagnosing Bent Shaft

axial measurements across machine

typically 180 degrees out of phase,

radial measurements across machine

are typically in phase
7.8 Diagnosing Looseness
 Looseness

• Mechanical Looseness (Rotating or Non Rotating)

 rotating looseness is the looseness caused by excessive clearance
between two parts normally one of them is rotating and one is stationary
like looseness between journal (shaft) and journal bearing, or looseness
due to excessive roller bearing clearance causing looseness of the shaft
within the bearing
 Non rotating looseness, is looseness between two parts normally the
both are stationary , like looseness due to excessive clearance between
bearing outer race and bearing housing.

• Base Mounting Looseness

• Structural Looseness (Structural Flexibility or Structure Weakness)

Cont.7.8 Diagnosing Looseness

Base Mounting Looseness

Looseness: mm/s

Loose Foundation
10
1X Harmonics and
3.1
sometimes
1
Sub-harmonics
½ X or 1/3 X or,… 0.31

.5X 1X 1.5X 2X 3X

mm/s
Mechanical Looseness:
Loose shaft 10

1X Harmonics and 3.1

sometimes 1
Sub-harmonics
0.31
½ X or 1/3 X or,…

.5X 1X 1.5X 2X 3X
Cont. 7.8 Diagnosing Looseness
Mechanical OR Base Mounting Looseness Pattern in TWF
 Mechanical or base mounting looseness produce repetitive, periodic synchronous
impacting.
 Sometimes you will find the wave form non symmetric (clipped at one direction).

TWF showed Synchronous impacting transients and clipping also, which are the reasons for
harmonics in the spectrum
Cont. Mechanical OR Base Mounting Looseness Pattern in TWF

 Base TWF showed Synchronous impacting transients and clipping also, which are the reasons for
Mounting harmonics in the spectrum
Loosenes
s Example

Spectrum showed strong harmonics from the running speed (Fundamental frequency)
Important due to transient impacting and clipping of TWF
Note:

Combine
Spectrum
data with
TWF Data
Cont. Mechanical OR Base Mounting Looseness Pattern in TWF

 Unlike Imbalance and Misalignment, the phase readings in case of

mechanical or base mounting looseness will be erratic , unstable.
Cont.7.8 Diagnosing Looseness
Structural Looseness (Structure Flexibility)
 Structure looseness (flexibility) will cause high 1X radial vibration at
the direction with lower stiffness, the vibration at this direction will
be greater than 4 times the vibration at the other perpendicular
direction
 For horizontal machines , usually the high 1X vibration will be in the
Horizontal direction.
 Structure Flexibility or weakness is due to: bad design or may be due
to deterioration, cracking or corrosion of the structure.
 If it is a design issue the structure will need stiffness by bracing or
supporting.
 This condition is different from imbalance as you will not find 90 Deg.
rel. phase between horizontal and vertical, but you might find 0 or
180 deg. ( not to be confused with resonance)
Cont.7.8 Diagnosing Looseness
Cont. Structural Looseness (Structure Flexibility)
Dominant 1X but
directional (high in
one radial
Horizontal Direction direction only) Vertical Direction
7.9 Diagnosing Rolling Element Bearing Faults
 When a rolling element passes over a defect on inner ring, outer ring, or one
or more of the rollers are defected, pulse like forces are generated that
result in one or a combination of bearing frequencies.
 Bearing frequencies:
 BPFO (ball passing frequency outer race) generated by rolling elements passing
over a defective outer race. This frequency is non synchronous and will typically
be in the range of 4.2X to 11.3X
 BPFI (ball passing frequency inner race) generated by balls or rollers passing over
a defect on the inner race. This frequency is non synchronous and will typically
be in the range of 6.3X to 13.8X
 BSF (ball spin frequency) generated by a ball or roller defect. This frequency is
non synchronous and will typically be in the range of 1.9X to 5.9 X
 FTF (fundamental train frequency) generated by cage defects. This frequency is
sub‐synchronous and will typically be in the range of 0.33X to 0.48X (inner race
rotating Application), or 0.52X to 0.67X (outer race rotating application).
 Φ
Φ

 If
 Φ=contact angle.
 N= number of rolling elements.
 P= pitch diameter.
 B= ball or roller diameter, average for tapered ones.
 RPS = speed in revolution per second.
• FTF=(RPS/2)[1- (B/P) cos Φ]
• BPFI=(N/2)(RPS) [1+ (B/P) cos Φ]
• BPFO=(N/2)(RPS) [1- (B/P) cos Φ]
• BSF=(P/2B) RPS [1- (B/P)² cos² Φ]
 Bearing frequencies:

 BPFO
 BPFI
 BSF
 FTF

 Bearing frequencies approximate estimation:

• BPFI = 0.6 No of Balls ( units in orders of machine speed)
• BPFO= 0.4 No of balls (units in orders of machine speed)

 Most Vibration Analysis software has the option to calculate bearing

forcing frequencies(from its data base), you input to the software the
bearing model (bearing number) and speed and the software will
calculate the forcing frequencies
Symptoms for Bearing Faults

 Non Synchronous peaks are symptoms for bearing problems

 High g’s level in acceleration spectrum and TWF is symptom for

bearing fault (other reasons could cause increase in g’s level as well).

 Modulation in TWF is a symptom for inner race or ball defect.

7.9.1 Bearing Inner Race , Outer Race and Ball Defects
Patterns in Spectrum & TWF
Inner Race Defect
Modulation in TWF & Side Bands In Spectrum ‐Data of Pump With
Bearing BPIR Defect (Pump Speed = 3000 RPM)
 Example of Amplitude Modulation in TWF due to BPIR (Bearing Inner Race Defect)
Bearing
Inner Race
Defect
Example

Modulation Rate or Modulation frequency is

equal to the running speed

Example of Side bands produced in Spectrum due to amplitude modulation in

Side bands
TWF(Bearing Inner Race Defect)
are spaced
from the Side
carrier
Carrier or Carrier Freq.=
Bands Freq.=BPIR Harmonic of
frequency
peak by a ???????? BPIR Side
value equal Bands
to the
modulated
frequency
(running
speed in this
example)
Outer Race Defect

Fundamental Harmonics from

Peak (BPOR) BPOR
Bearing
Ball Defect
Example

Carrier
Freq.=
2BSF
Modulation Rate or Modulation frequency is equal
to FTF

Side bands
are spaced
from the
carrier Harmonics of
Carrier
frequency BSF
Freq.= 2BSF FTF Side
peak by a
value equal Bands
to the
modulated
frequency
(FTF in this
example)
7.9.2 Bearing Failure stages
Cont. 7.9.2 Bearing Failure stages

Stage I
Stage I has some very high
frequency content in the Spike
Energy region. Indicates metal-to-
metal contact. May indicate a lack
of lubrication. This zone is in the
ultrasonic region(20,000 -
60,000Hz).Physical inspection
of the bearing at this stage may
not show any identifiable
defects. (Subsurface damage).

High Frequency Detection HFD

Techniques is required:
AE (Acoustic Emission)
SE (Spike Energy) Spike Energy or Stress
PeakVue Waves or Shock waves
Shock Pulse Method SPM (very High Freq. ,very
short duration pules
Cont. 7.9.2 Bearing Failure stages
Stage II
Stage II begins to generate signals
associated with natural resonance
frequencies (2000 - 8000 Hz) of the
bearing parts as bearing defects begin
to "ring" the bearing components. A
notable increase in Zones 3 and 4
region signals is associated with this
stage. this may be months ahead of
failure. Beginning signs of defects
will be found upon inspection.
Highly critical machines should be
replaced/repaired at the next
available opportunity, A reduced Resonance of one of bearing
monitoring interval should be components or mounting
established with the CM provider. magnet and sensor .Ringing
repeated by the rate of the
Detection Techniques required:
bearing defect‐ or in other
Enveloping (Demodulation) words natural frequency is
Acceleration Spectrum and g’s level in modulated by bearing defect
TWF frequency
Cont. 7.9.2 Bearing Failure stages
Stage III
Stage III condition has the fundamental
bearing defect frequencies present.
These frequencies are those discussed
previously in this paper. Harmonics of
these frequencies may be present
depending upon the quantity of defects
and their dispersal around the bearing
races. The harmonic frequencies will be
modulated, or side banded, by the shaft
speed or FTF. Zone 4 signals continue to
grow throughout this stage. At this stage,
if you remove the bearing, you can clearly
see the defects in the raceways and/or
rolling elements. Bearings in critical and
non-critical applications should be
replaced at this stage
Detection Techniques required: Fundamental bearing
Defect frequencies or
Velocity Spectrum
harmonics, side bands of
Acceleration Spectrum and g’s level in RPM or FTF.
TWF
Cont. 7.9.2 Bearing Failure stages
Stage IV
Stage IV is the last condition before
catastrophic failure of the bearing. Noisy
operation (rough).This stage is associated with
numerous modulated fundamental
frequencies and harmonics indicating that the
defects are distributed around the bearing
races. Due to the increased degradation of the
bearing the internal clearances are greater and
vibration pattern begins to look like
mechanical rotating looseness (Running
speed harmonics increase). During later
phases of stage IV, the bearing fundamental
frequencies will decline and be replaced with
random noise floor or "hay stack" at their
frequencies. Zone 4 signal levels will actually
Running Speed Harmonics increase
decrease. Bearings requiring routine vibration and increase in noise floor.
monitoring should not be allowed to reach Fundamental bearing Defect
this stage. frequencies or harmonics may be
Detection Techniques required: reduced at the end of stage 4 prior to
Velocity Spectrum catastrophic failure.
7.9.3 Envelope (Demodulation) Technique for Early
Detection of Bearing Faults

 Step1: HP (or
Bandpass) Filter
 Step2 : Rectify The
signal
 Step3 : Low Pass
Filter (Envelope or
Demodulation)
 Step 4 : Enveloped
FFT Spectrum
7.10 Diagnosing Gear Box Faults
 Some Gear box Fault Conditions:
• Tooth Wear
• Tooth Load
• Gear Eccentricity
• Backlash
• Gear Misalignment
• Broken or cracked teeth

 Forcing frequencies in gears:

• Input Gear Speed (1X Input Gear)
• Out put Gear Speed (1X Out put Gear)
; Speed In X No of Teeth In = Speed Out X No of Teeth Out
• Gear Meshing ( Tooth Meshing Frequency )
GM = No of Teeth of Input Gear X Speed of Input Gear
= No of teeth of Output Gear X Speed of Output Gear
Cont. 7.10 Diagnosing Gear Box Faults
 These forcing frequencies will be prominent in the radial direction for
spur gears and in the axial direction for helical gears

Helical Gear

Spur Gear

Helical Gear
Cont. 7.10 Diagnosing Gear Box Faults
7.10.1 Healthy gear Box Vibration Pattern (Spectrum & TWF)
 There will normally be peaks at shaft speeds (1X input and 1X output) and
gear mesh frequency GM and 2GM and there may be side bands of shaft
speed around the gear mesh frequency, however they will be low level.
 There frequency will be most prominent in the radial direction for spur
gears and in the axial direction for helical gears.
 Time wave form should be largely sinusoidal with limited modulation
(because of evenly tooth meshing), in healthy gear boxes there should be
no distinct transients.
Cont. Healthy gear Box Vibration Pattern (Spectrum & TWF)
7.10.2 Vibration Analysis for Gear Boxes
 Most faults are detected by studying GM, 2GM and 3GM with their side
bands
 Set f max high enough to see 3 times GM and side bands
 For Healthy Gear box Side bands may be present but low levels and limited
modulation in TWF
 Wave form is very powerful tool in monitoring and diagnostics of gear box
faults. Wave form will indicate whether the tooth meshing is happening
evenly or there is increase and fall (modulation), or whether one tooth is
different in its meshing from all teeth due to crack or broken tooth for
example.
 Time Synchronous averaging TSA is commonly used when attempting to
diagnose gear faults, TSA will average away all vibration that is not
synchronous with tach pulse, so other vibration from bearings are
removed.
7.10.3 Tooth Wear

 Wear is indicated by excitation of gear natural frequency (impacting

during meshing excite fn.) along with side bands of 1X RPM of bad
gear. (Multiple side bands will appear and increase in level).
 Side bands are a better wear indicator than GM
 GM may not change in amplitude when wear occurs
 Multiple side bands may appear at 2GM and 3GM frequency as well.
7.10.4 Gear Backlash

 Gear backlash also is indicated by excitation of gear natural frequency

(impacting during meshing excite fn.) along with side bands
 Multiple side bands may appear at 2GM and 3GM frequency as well.
 GM peak and gear natural freq. fn. Will often decrease with load
increase if this problem exist.
7.10.5 Tooth Load

 GMF frequency are often sensitive to load (amplitude dependent

on alignment and load).
 High GMF do not necessarily indicate problem but excessive load
may result in wear and damage prematurely.
7.10.6 Eccentric Gears

 if a gear is eccentric (or the shaft is bent), there will be an increase in the
amplitude of the sidebands (of offending gear) around the GM
frequency. The key is that rather than witnessing a family of sidebands,
there may only be one dominant sideband on either side of the gear
mesh peak.
 GM and 3GM will be dominant
 Eccentricity can result in backlash, therefore you may also witness a
peak at the natural frequency of the gear. This peak (whose frequency is
unknown) will also have 1X sidebands (of offending gear).
 Modulation in TWF at the rate of the offending gear.
7.10.7 Misaligned Gears

 Misalignment of gears will cause varying load and varying amplitude of

meshing (GM modulation by the RPM )
 2GM and 3GM will increase (more dominant), 2GM is the highest in
amplitude
 Side bands level will increase due to modulation caused by varying
load.
Gear Box Modulation in TWF & Side Bands In Spectrum ‐Data of Aux G.B of
Example Turbine (GG Speed = Gear Speed nearly 13000 RPM)
LS Gear Rate
HS Gear Rate

Modulation present , Modulated Freq.= Gear Speed

Modulation is present , Carrier Frequency = GM=4048 HZ (242 914 CPM),

Modulated Frequency= 13000 CPM ( GG speed= High Speed Gear Speed)

Modulated Carrier HS Gear Side

Freq.= HS Freq.= GMF bands
Gear Speed

LS Gear Side bands

7.10.8 Cracked or Broken Tooth

 Cracked or Broken gear tooth might not be clear in the spectrum, the
best way to detect such fault is from the TWF.
 Increase in 1X amplitude of gear with broken tooth
 Impact during the meshing of the broken or cracked tooth may excite
natural frequency fn
Cont.7.10.8 Cracked or Broken Tooth

Synchronous
impacting at the rate
of Speed of the gear
with cracked or
broken teeth
 Cracked or Broken gear tooth might not be clear in the spectrum, the best way
to detect such fault is from the TWF.
 Cracked or Broken Tooth: If you view a time waveform that contains
approximately 6‐10 revolutions of the gear in question then the vibration change
as the tooth comes into mesh should be clear to see.
 The impact wave form in this pattern is occasionally referred to “angel fish”(Figure
below). This distinctive pattern shows an impact followed by smaller “ring down
"peaks giving the time data its pattern.
7.11 Diagnosing Electric Induction Motors Faults
7.11 Diagnosing Electric Induction Motors Faults
 Forcing frequencies in Induction Motors:
• FL = Line Frequency 50 Hz or 60 Hz in Egypt (50 Hz), but usually in
vibration we don’t see Line Frequency but twice the line frequency
(2FL ) because the magnetic forces (causing vibration) rises and falls
at this rate during one cycle of AC current.
• Rotor Bar Passing Frequency RBP = No of rotor bars X RPM
• Ns= Synchronous Speed (nominal) = 120 FL / P
;Ns = CPM or RPM
;FL (Hz)
; P = No of poles
Ex: 2 pole , line freq.= 50 Hz, Ns =120 * 50/2 = 3000 RPM
Ex: 4 pole , line freq.= 50 Hz, Ns =120 * 50/4 = 1500 RPM
• Fslip = Slip Frequency = Actual Speed N – Synchronous speed Ns
• PPF=Pole Pass Frequency = Slip Freq. * No of poles
7.11.1 Vibration Characteristics for Induction Motor Faults
 Twice Line Frequency 2FL Most of the Electrical problems in induction
motors will appear at twice the line frequency (2FL ) because the magnetic
forces (causing vibration) rises and falls at this rate during one cycle of AC
current.
 Hi resolution Spectrum required
 Some faults causes PPF side bands around the 2Lf peak , which will
require high resolution spectrum (PPF low frequency) to be resolved
and seen beside the 2Lf peak.
 Most of the electrical faults cause increase in 2 FL , which could be
confused with 2X (2 RPM) of motor for 3000 RPM motors, or could be
confused with 4X (4RPM) of motor for 1500 RPM motors.
 Hi fmax required
 Some faults appear at rotor bar passing frequency which could be 50
rotor bar and might be more ( if unknown we can take a high freq.
range spectrum with 300 000 CPM (5000 Hz) and check the spectrum.
 Some Induction Motor Fault Conditions:
• Static ( Stator ) Eccentricity
• Dynamic Eccentricity (Eccentric rotor)
• Rotor Bow
• Cracked/ broken rotor bar/Loose rotor bars
• Shorted Lamination
• Loose connections
7.11.2 Stator Problems

 Stator Problems Produce high peak at 2FL (Twice the line frequency)
 Stator eccentricity produces uneven stationary air gap, vibration is very
directional. (at radial direction where there will be the biggest and the
smallest air gap).
 Beating sound may be heard (beating may happen, due to interplay
between the 2X and 2Lf
 Soft foot can produce an eccentric stator.
7.11.3 Eccentric Rotors (Dynamic Eccentricity)

 Eccentric rotors produce a rotating variable air gap, this induces

pulsating vibration ( increase and decrease in magnetic forces by the PPF
rate – modulation by the rate of the PPF
 Again you will see high 2LF but this time will be side banded by PPF
 7.11.4 Rotor Bar Problems

 For Cracked or broken rotor you will often see a busy spectrum, 1X
harmonics of running each with skirt of PPF side bands.
 For Loose Rotor Bar there will be increase in rotor bar passing frequency
peak and side banded by 2Lf
Note: Normally in many motors you will see the rotor bar pass frequency
exist and side bands by 2Lf appears in vibration, that does not mean a
problem, but should be low level for good motor condition (trend is
important).
7.11.5 Rotor Bow (Thermal Bow)

 Rotor bow on a motor will present like static unbalance. Broken rotor bars or
rotor shorted lamination or any other fault conditions that result in uneven
current flow in the rotor will result in localized heating, which will in turn
cause the rotor to bow. The bow creates an unbalance condition (uneven
weight distribution around the centerline of the shaft). Therefore we see a
high amplitude peak at 1X turning speed in the radial (vertical and horizontal)
directions.
 You may also see the effect of a rotating air gap – high peak at twice line
frequency (100 or 120 Hz) with pole‐pass sidebands.
 It can be detected by the fact that it goes away when the motor is cold.
7.11.6 Loose Connections (Phasing)

 Phasing problems due to loose connectors can cause excessive vibration at

2Lf with side bands of Lf /3
7.12 Diagnosing Belt Drives Problems
7.12.1 Diagnosing Defected Belt (Belt Wear)

 Spectrum will show peak at the belt frequency and harmonics (with twice the BF
may be the highest).
 Belt Rate = Sheave Diam X Sheave RPM / Belt Length (CPM)
 Belt Frequency is sub synchronous frequency (less than 1X of both pulleys).
Cont. 7.12.1 Diagnosing Defected Belt (Belt Wear)
 When a belt defect happens the defective area makes an impact as it passes over
the pulleys.
 It is common for impacts to show in the wave form at the belt frequency or twice
the belt frequency because the defect passes over two pulleys in one rotation of he
belt. Belt Frequency is sub synchronous frequency
 Belt Rate = Sheave Diam X Sheave RPM / Belt Length (CPM)
 In the below example ∆T=Period = 0.309 Sec, BF= 1/0.309= 3.236 HZ = 194 CPM
 Cont .Belt Defect Pattern in TWF
Another Example : Belt Defect showed in TWF but at 2 times BF
BF = 304 CPM, 2 BF = 608 CPM

Impacts at 2 times the Belt Rate (608 CPM)

Important
Note:
Belt Defect is
more clear in
the TWF
(Impacts at
the 2belt
rate)

In this
Example
Defect is not
clear in the
spectrum
7.12.2 Diagnosing Pulleys Misalignment

 Belt Misalignment produce 1X vibration in both radial and axial directions.

Predominantly in the axial direction.
 1X relative Phase shift in the axial direction from driver to driven pulleys
are 180 Deg
7.12.3 Diagnosing Eccentric Pulleys (in Belt Driven Machines)
7.13. Diagnosing Hydraulic and Aerodynamic problems in Pumps‐
Fans‐ Compressors

 Fluid generates forces causing vibration at Blade pass or vane pass frequency
 Blade pass or vane pass frequency = No of blades/vanes X RPM
 Blade pass or vane pass will normally be present in the spectrum but with
low amplitude, an increase in blade pass or vane pass will indicate a fault
condition
7.13.1 Blade/Vane Pass Fault Diagnosis

Reasons for Blade/Vane pass increase:

 Non uniform gap between the rotor and the stator (impeller and diffuser) or
fan blade and plenum (blade tip clearance),…., which means eccentric hosing
will cause high blade pass frequency, a protrusion in the housing may
increase the vibration at this frequency as well
 Cont. Next slide
Cont. 7.13.1 Blade/Vane Pass Fault Diagnosis
 Pipe strain, misalignment can cause increase in vane pass, as it might cause
non uniform gap
 Flexibility in the fan housing.
 High wind with some fan housing orientation may cause increase in
vibration at blade pass frequency
 Non uniform variable blade pitch angle
 Blade or vane wear.
 Impeller loose on shaft (Vane Pass freq. with side bands of RPM)
 Inlet or discharge line restriction (elbow, ….) blockage.
7.13.2 Pump Cavitation
• Pump Cavitation When the fluid cavitate reaching the vapor
pressure and formation of bubbles which burst, this causes random
impacting which appears in the TWF, the random impacting due to
cavitation will add energy to the spectrum floor specially in the
high frequency range from 10X to 30X of the pump speed, so if the
noise floor is based toward the high frequency range you may have
cavitation, process or flow noise.
 Pump
Cavitation
Example
Random Impacting
in time wave form
due to cavitation

Spectrum showed
increased noise floor
due to random Raised Noise Floor
impacting from pump
cavitation
7.13.3 Flow Turbulence

 Flow Turbulence in fans, pumps and compressors will show in the spectrum
as broad band sub synchronous peaks in the range from 0.4 to 0.8 X
 Causes of Turbulence:
• Obstruction in flow path, sharp turns (elbows), Abrupt diameter change
• Blockage in lines.
Chapter 8. Alarm Settings &
Acceptance Criteria
 8.1 Alarm Types

Analysis Parameter Sets

1xRPM - BALANCE
1.8
1.5 2xRPM - ALIGNMENT
1.2 3-6xRPM - LOOSENESS
0.9
GEARS & ANTI FRICTION BEARINGS
0.6
6-30xRPM 30-50xRPM
0.3

1000 2000 3000 4000

Frequency
Cont. 8.1 Alarm Types
Vibration Spectrum and Parameter Bands
(Analysis Parameter Sets)

Sub-
Harmonic 1X 2X Bearing Bearing Gears Bearing
Amplitude

1x 2x 50x
Trend of
Trend of Bearings
Balance
5
mm/sec Alarm
1
Time mm/sec Time
(Days) (Days)
Cont. 8.1 Alarm Types

SELECTED FREQUENCY BAND ALARMING

Sub-
Harmonic 1X 2X Bearing Bearing Gears Bearing
Amplitude

1x 2x 50x
Trend of
Trend of Bearings
Balance
5
mm/sec Alarm
1
Time mm/sec Time
(Days) (Days)
Cont. 8.1 Alarm Types

OVERALL ALARM
 AREA (ENERGY) UNDER THE SPECTRUM
– 'SHOTGUN' METHOD, NO DIAGNOSTIC INFORMATION

ALARM LEVEL = 0.1 IN/SEC

PEAK - RMS
OVERALL VALUE

 BEARING AND GEAR PROBLEMS WILL NOT GIVE EARLY

WARNING WITH OVERALL ALARMS
 Cont. 8.1 Alarm Types

ENVELOPE ALARMING

 CONSTRUCTED FROM BASELINE SPECTRA

 "VARIABLE SHAPE ENVELOPING"

 EFFECTIVE IN FINDING SMALL CHANGES

0.1 in/sec 0.3 in/sec

Increase Absolute
Value
200 % rise
over Reference .02 in/sec
Minimum
Value
Cont. 8.1 Alarm Types

Statistical Alarms
 Some of the modern soft wares has the power to
calculate an effective and practical alarm level for
different machines that are included in the data base
after collecting the first 4 sets of readings.
 However, re-adjusting the alarm in a PMP system
should be reviewed on a regular basis during the first
6 months of implementation.
8.2 Acceptance Criteria‐Acceptance Limits:
The initial alert/alarm criteria should be set to give the earliest possible indication of
the occurrence of a fault. Alert/alarm criteria may also result from the processing of
several measurements. Alert/alarm criteria should be optimized over time as an
iterative process.
The alert/alarm limits will depend on 4 main factors:
 OEM alarm limits, commonly machine in good condition shows measurement
values below or around half the OEM alarm value, CM alarm should be adjusted
less than OEM alarm level to facilitate diagnosis and taking the proper corrective
action as early as possible.
 Base line data; Trending or successive measurements and comparison to the
base line data (Condition monitoring software statistical alarms will be useful in
adjusting CM alarm values).
 Comparison between identical machine condition (identical machines‐trains
running with the same operating conditions).
 International standards (ISO, API, ASTM…) acceptance limits.
Cont. 8.2 Acceptance Criteria‐Acceptance Limits:

Information about Standards for acceptance limits

 Vibration: Information on alert/alarm criteria is contained in ISO 10816 and

ISO 7919. Also information about vibration acceptance limits is included in the
API standards of different machine categories like API 616 gas Turbines, API617
axial and centrifugal compressors, API 661 Air cooled heat exchangers, etc.

 ISO 10816 Mechanical vibration‐Evaluation of machine vibration by

measurements on non‐rotating parts
 ISO 7919 Mechanical vibration of non‐reciprocating machines‐Measurements
on rotating shafts and evaluation criteria
Cont. 8.2 Acceptance Criteria‐Acceptance Limits:

ISO 10816‐1 Evaluation of Machine Vibration‐

Part 1‐ General Guide Lines
Cont. 8.2 Acceptance Criteria‐Acceptance Limits:
Machine Classification

Class I: Individual parts of engines and machines integrally connected to the

complete machine in its normal operating condition. (Production electrical
motors of up to 15 KW are typical examples of machines in this category.)

Class II: Medium‐sized machines (typically electrical motors with 15KW to 75

KW output) without special foundations, rigidly mounted engines or machines
(up to 300 KW) on special foundations.

Class III: Large prime‐ movers and other large machines with rotating masses
mounted on rigid and heavy foundations which are relatively stiff in the
direction of vibration measurements.

Class IV: Large prime‐movers and other large machines with rotating masses
mounted on foundations which are relatively soft in the direction of vibration
measurements ( for example, turbo generator sets and gas turbines with
outputs greater than 10 MW).
Cont. 8.2 Acceptance Criteria‐Acceptance Limits:

Zone A: The vibration of newly commissioned machines would normally fall

within this zone.

Zone B: Machines with vibration within this zone are normally considered
acceptable for unrestricted long term operation.

Zone C: Machines with vibration within this zone are normally considered
unsatisfactory for long‐term continuous operation. Generally, the machine may
be operated for a limited period in this condition until a suitable opportunity
arises for remedial action.

Zone D: Vibration values within this zone are normally considered to be of

sufficient severity to cause damage to the machine.
Chapter 9. The Condition
Monitoring Program
9.1 Equipment Monitoring Deployment Process:

Equipment Monitoring Deployment Process:

The ISO 17359 Standard presents an overview of a generic procedure
recommended to be used when implementing a condition monitoring program
and a typical condition monitoring program flow chart is provided (See Fig 2
Below).
 Equipment Technical data gathering & machine Audit.
Gathering all the required technical data about selected equipment and its
components (Mechanical drawing, datasheets, Speed, power, gears, blades,
vanes, bearings, couplings, belts….) from the OEM manual and from onsite
equipment audit, This step is important for identifying the parameters which
should be monitored and the best condition monitoring techniques to be
applied on the equipment.

 Data base set up for the equipment on condition monitoring software.

Defining the gathered technical data, defining the monitored parameters,
and adjusting the measurement (Data Collection) setup (range,
resolution….).

 Measuring or collecting base line data for the selected parameters.

For machines with several operational states, it may be necessary to
establish baselines for each of these states.
 Verifying the data and apply setup enhancement or tuning if required.
 Trending successive set of data.
 Analysis, assessing condition & fault diagnosis.
 Issue analysis reports and work orders for recommended actions.
 Verify that the maintenance action done is correct by recollection of data and
analysis
after applying the recommended action. It is also useful to inspect the replaced
components to confirm that the initial diagnosis or prognosis was correct.
 Continue monitoring with fixed frequency defined in the CMMS.
 Review Equipment Monitoring Program
Condition monitoring process is an ongoing process, and techniques that may not
have been available, or considered to be too costly at the time or unfeasible in some
other way (lack of access, safety, etc.), may on review become feasible, similarly the
effectiveness of techniques being undertaken in the program should be assessed and
any technique considered no longer necessary removed. Alert/alarm criteria may
also need revision due to change or modification in machine.
9.2 Monitoring (Measurement) Frequency:

An effective CM program will identify developing issues that can affect

equipment performance and availability and will provide sufficient time for
intervention before the issue develops to a state where a shutdown is
required. Consideration should be given to whether continuous or periodic
sampling is required.
The measurement interval primarily depends on the type of fault, its rate of
progression (and thus the rate of change of the relevant parameters).
However, the measurement intervals also influenced by factors such as
historian data trending, how stable is the equipment condition, history of
failures, duty cycles, cost and failure criticality.
9.3 Condition Monitoring Reports:
The CM reports shall include but not limited to the following:
 Equipment Tag No and S.N
 Essential data describing the machine (Equip Type, Rated Speed, Rated power, Shaft
coupling, configuration such as direct, belt shaft driven, lubricant, component such as
bearing ,sealing and gear).
 Measurement position (vibration measurement location, Lube oil sampling point,
thermal image location).
 Instruments used in measurements (instrument type i.e. vibration proximity or vel
transducer or accelerometer, instrument sensitivity).
 Measurement units for vibration (g’s , mm/s, microns, Peak, RMS,….) for oil ( % by
weight, PPM, centistoke, … )for thermography (Temp. Deg K or Deg C, …).
 Date and time information of measurement or sampling.
 Operating conditions during measurement or sampling.
 Measurement (collected) data (vib spectrum , TWF, Phase , orbits, Polar Bode, …), for
oil (Viscosity, Color coding, TAN, wear paticle, ISO paticle count, Foaming…).
 Measurement settings (vib freq range, resolution…), (Thermography Emissivity,
Distance, Reflected Temp…).
 Background information about the history or problem.
 Diagnosis and data interpretation.
 Conclusion of findings.
 Recommended action.