Machinery Vibration Analysis

(Training Manual & Practical Case Studies)

2nd edition - 2025

Prepared by

Wael Salman
Condition Monitoring Consultant
Approved Vibration Analysis Instructor – Mobius Institute
PhD in Rotating Equipment Dynamics & Vibrations
Machinery Vibration Analysis
(Training Manual & Practical Case Studies)
2nd edition - 2025

Prepared by

Wael Salman
Condition Monitoring Consultant
Approved Vibration Analysis Instructor – Mobius Institute
PhD in Rotating Equipment Dynamics & Vibrations
 Acknowledgement

Thanks to Almighty ALLAH for giving me strength and ability to understand, learn, and complete
this book. I also admire the help and guidance of all my professors and teachers throughout my
educational life. I am very thankful to my wife and parents specially my mother who praying for me
to achieve any success in my life.
 About The Author

Wael Salman is a condition monitoring Consultant and Assistant Professor in the faculty of

mechanical engineering. His experience is divided into two main branches; 1) Academic experience

in teaching different academic, maintenance, and industrial courses; 2) Service and maintenance of

rotating equipment in different industries. He is specialized in predictive maintenance of rotating

equipment based on advanced vibration analysis and other technologies in addition to corrective

maintenance works such as shaft & pulley laser alignment and dynamic balancing in site. He has been

granted a PhD degree in dynamics and vibrations of rotating equipment in 2022. After that, he got

the accreditation of Mobius institute to be an approved vibration analysis Instructor in 2023. During

the academic and industrial life, he published scientific papers in local and international journals to

discuss the fault diagnosis of rotating equipment based on vibration analysis and Artificial

Intelligence (AI) techniques.

Author Contacts
Email: waelsalman@vibsolutions.net
Linkedin: https://www.linkedin.com/in/wael-salman-96428390/
 Table of Contents
1 Chapter (1) Introduction………………………………… ............................................................ 7
1.1 Maintenance……………….. ..................................................................................................... 7
1.2 Maintenance Techniques……………….. ................................................................................. 7
1.2.1 Run-to-failure (breakdown) maintenance technique………………… .............................. 8
1.2.2 Preventive maintenance technique………………… .......................................................... 8
1.2.3 Predictive maintenance technique…………………......................................................... 10
1.2.4 Proactive maintenance technique…………………. ......................................................... 13
2 Chapter (2) Fundamentals of Vibration……………. ................................................................ 15
2.1 Introduction……………….. .................................................................................................... 15
2.2 The Basic Vibration Signal……………….. ............................................................................ 15
2.3 Frequency……………….. ....................................................................................................... 16
2.4 Amplitude……………….. ...................................................................................................... 18
2.5 Displacement, Velocity, and Acceleration………………....................................................... 19
2.6 Time Domain……………….. ................................................................................................. 21
2.7 Frequency Domain……………….. ......................................................................................... 22
3 Chapter (3) Vibration Transducers…………… ........................................................................ 24
3.1 Introduction……………….. .................................................................................................... 24
3.2 Vibration Transducer……………….. ..................................................................................... 24
3.3 Types of Transducers……………….. ..................................................................................... 24
3.3.1 Proximity Probe………………… .................................................................................... 24
3.3.2 Velocity Transducers………………… ............................................................................ 26
3.3.3 Piezoelectric Accelerometers………………… ................................................................ 28
4 Chapter (4) Concept of Vibration Measurement…………… ................................................... 35
4.1 Introduction………………… .................................................................................................. 35
4.2 Types of Monitoring System……………….. ......................................................................... 35
4.2.1 Intermittent Monitoring System………………… ............................................................ 35
4.2.2 Permanent monitoring system…………………............................................................... 35
4.3 Time Interval between Measurements……………….. ........................................................... 35
4.4 Mounting the Accelerometer……………….. ......................................................................... 36
4.5 The Influence of Environments on Accelerometers……………….. ...................................... 38
4.5.1 Base Strain………………… ............................................................................................ 39
4.5.2 Nuclear Radiation………………… ................................................................................. 39
4.5.3 Magnetic Field………………… ...................................................................................... 40
4.5.4 Humidity………………… ............................................................................................... 40
4.5.5 Corrosive Substances………………… ............................................................................ 41
4.5.6 Acoustic Noise………………… ...................................................................................... 41
1
 4.5.7 Transverse Vibrations………………… ........................................................................... 41
4.5.8 Temperature Transients………………… ........................................................................ 42
4.5.9 Sensitivity Change due to Temperature………………… ................................................ 42
4.5.10 Earth Loops………………. ............................................................................................ 43
4.5.11 Triboelectric Noise……………….................................................................................. 43
4.6 Accelerometer Mounting Directions………………................................................................ 43
5 Chapter (5) General Fault Analysis……………. ....................................................................... 46
5.1 Type of Vibration Signals……………….. .............................................................................. 46
5.2 Interpretation of Vibration Data………………....................................................................... 46
5.3 Vibration monitoring……………….. ..................................................................................... 47
5.4 Monitoring Techniques……………….. .................................................................................. 47
5.4.1 Trending………………… ................................................................................................ 47
5.4.2 Comparative Analysis…………………. .......................................................................... 49
5.4.3 Baseline Data………………… ........................................................................................ 49
5.4.4 Signature Analysis………………… ................................................................................ 49
5.5 Quantifying the Vibration Level…………………. ................................................................. 49
6 Chapter (6) Understanding Data Collection……………........................................................... 51
6.1 Introduction……………….. .................................................................................................... 51
6.2 Methods of Vibration Data Collection……………….. ........................................................... 51
6.2.1 Automated Method………………… ............................................................................... 51
6.2.2 Manual Method………………… ..................................................................................... 51
6.3 Analog to Digital Converter……………….. ........................................................................... 52
6.4 Sampling rate……………….. ................................................................................................. 53
6.5 Lines of Resolution……………….. ........................................................................................ 54
7 Chapter (7) Signal Processing…………….. ................................................................................ 55
7.1 Introduction……………….. .................................................................................................... 55
7.2 Windowing……………….. ..................................................................................................... 55
7.2.1 Rectangular Window………………… ............................................................................ 55
7.2.2 Hanning and Hamming Window…………….. ................................................................ 56
7.2.3 Blackman-Harris Window………………… .................................................................... 57
7.2.4 Flat Top Window………………… .................................................................................. 58
7.3 Averaging……………….. ....................................................................................................... 58
7.3.1 Linear averaging………………… ................................................................................... 58
7.3.2 Peak Hold Averaging…………………. ........................................................................... 60
7.3.3 Time Synchronous Averaging………………… .............................................................. 61
7.4 Anti-aliasing Filter………………… ....................................................................................... 61

2
8 Chapter (8) Signal Analysis……………….................................................................................. 63
8.1 Introduction……………….. .................................................................................................... 63
8.2 Time Domain Analysis……………….. .................................................................................. 63
8.3 Frequency Domain Analysis……………….. .......................................................................... 66
8.3.1 Half Spectrum Plot………………… ................................................................................ 67
8.3.2 Full Spectrum………………… ........................................................................................ 69
8.4 Phase Analysis…………………. ............................................................................................ 72
8.4.1 Why Is Phase Important?........................... ....................................................................... 73
8.4.2 The Keyphasor Event…………………. ........................................................................... 74
8.4.3 Phase Measurement…………………. ............................................................................. 75
8.4.4 Absolute Phase………………….. .................................................................................... 76
8.4.5 Relative Phase………………… ....................................................................................... 77
8.5 Cepstrum Analysis……………….. ......................................................................................... 79
8.6 Envelope Analysis……………….. ......................................................................................... 80
8.6.1 Load Variation Modulation…………………................................................................... 81
8.6.2 Envelope Detection………………… ............................................................................... 81
8.7 Crest Factor………………… .................................................................................................. 82
8.8 Kurtosis Analysis……………….. ........................................................................................... 83
9 Chapter (9) Fault Diagnostics…………….. ................................................................................ 84
9.1 Introduction……………….. .................................................................................................... 84
9.2 Unbalance……………….. ...................................................................................................... 85
9.2.1 Static Unbalance………………… ................................................................................... 85
9.2.2 Couple Unbalance………………… ................................................................................. 86
9.2.3 Dynamic Unbalance………………… .............................................................................. 87
9.2.4 Unbalance – Overhung Rotors………………… .............................................................. 87
9.3 Bent Shaft………………......................................................................................................... 87
9.4 Eccentric rotor……………….. ................................................................................................ 88
9.5 Misalignment……………….. ................................................................................................. 89
9.5.1 Parallel Misalignment………………… ........................................................................... 89
9.5.2 Angular misalignment………………… ........................................................................... 89
9.6 Soft Foot………………........................................................................................................... 90
9.6.1 Parallel Soft Foot………………… .................................................................................. 90
9.6.2 Angular Soft Foot………………… ................................................................................. 91
9.6.3 Squishy Foot………………… ......................................................................................... 91
9.6.4 Stress-Induced Soft Foot………………… ....................................................................... 91
9.7 Looseness…………………. .................................................................................................... 91
9.7.1 Structure Looseness (Non-Rotating Looseness - Type A)………………… ................... 91
3
 9.7.2 Machine Components / Base Plate Looseness (Non-Rotating Looseness - Type B) ....... 92
9.7.3 Internal Assembly Looseness (Rotating Looseness - Type C)………………… ............. 93
9.8 Belt Drive Defects………………............................................................................................ 93
9.8.1 Pulley Misalignment………………… ............................................................................. 94
9.8.2 Belt Wear………………… .............................................................................................. 95
9.8.3 Eccentric Pulleys………………… ................................................................................... 95
9.8.4 Belt Resonance…………………...................................................................................... 95
9.9 Gear Defects………………..................................................................................................... 96
9.9.1 Normal Profile………………… ...................................................................................... 96
9.9.2 Gear Tooth Wear…………………................................................................................... 97
9.9.3 Gear Tooth Load………………… ................................................................................... 98
9.9.4 Eccentric Gears and Backlash……………………… ....................................................... 98
9.9.5 Misaligned Gears…………………… .............................................................................. 99
9.9.6 Cracked or Broken Teeth………………… ...................................................................... 99
9.9.7 Hunting Tooth Frequency………………… ................................................................... 100
9.9.8 Load Changes…………………...................................................................................... 101
9.10 Bearing Defects……………….. .......................................................................................... 102
9.10.1 Stage One……………….. ............................................................................................ 103
9.10.2 Stage Two……………….. ........................................................................................... 104
9.10.3 Stage Three………………. .......................................................................................... 104
9.10.4 Stage Four………………. ............................................................................................ 106
9.10.5 Cocked Bearing: on Shaft………………. .................................................................... 107
9.10.6 Cocked Bearing: in Housing………………. ................................................................ 107
9.10.7 Inner Race Sliding on Shaft………………. ................................................................. 108
9.10.8 Outer Race Loose in Housing………………. .............................................................. 108
9.11 Oil Whirl……………… ...................................................................................................... 109
9.12 Oil Whip………………....................................................................................................... 109
9.13 Electrical Problems……………… ...................................................................................... 110
9.13.1 Cracked/Broken Rotor Bars………………. ................................................................. 110
9.13.2 Eccentric Rotor………………. .................................................................................... 111
9.13.3 Eccentric Stator / Soft Foot………………. .................................................................. 111
9.14 Cavitation……………… ..................................................................................................... 112
10 Chapter (10) Practical Case Studies in Different Industries………… ................................ 113
10.1 Case Study (1) Hot Oil Centrifugal Pump (Oil and Gas) - Root Failure Analysis ............. 113
10.1.1 Equipment Historical Data………………. ................................................................... 113
10.1.2 Vibration Analysis and Diagnostics………………...................................................... 114
10.1.3 Piping Strain and Alignment Measurements………………. ....................................... 115
4
 10.1.4 Piping Offset Measurements………………. ................................................................ 116
10.1.5 Shaft Alignment Measurements………………. ........................................................... 123
10.1.6 Methods of Piping Alignment………………. .............................................................. 125
10.1.7 Visual Inspection for Interior Components of Pump……………….. .......................... 126
10.1.8 Maintenance and Correction Results………………. ................................................... 126
10.2 Case Study (2) Cooling Fan (Cement Industry) - Severe Fan Bearing Looseness ............. 129
10.2.1 Analysis and Diagnosis………………. ........................................................................ 129
10.2.2 Recommendations and Actions………………............................................................. 129
10.3 Case Study (3) Water Pump (Chemical Industry) – Pump Bearing Defect ........................ 132
10.3.1 Analysis and Diagnosis………………. ........................................................................ 132
10.3.2 Recommendations and Actions………………............................................................. 132
10.4 Case Study (4) Dynamic Separator (Cement Industry) - Severe Wear in Impeller ............ 134
10.4.1 Analysis and Diagnosis………………. ........................................................................ 134
10.5 Case Study (5) Water Pump (Water Station) – Soft Foot in Base Plate ............................. 137
10.5.1 Analysis and Diagnosis………………. ........................................................................ 137
10.5.2 Recommendations and Actions………………............................................................. 137
10.6 Case Study (6) Cooling Tower Fan (Chemical Industry) – Damaged Coupling Element . 140
10.6.1 Analysis and Diagnosis………………. ........................................................................ 140
10.7 Case Study (7) Hot Water Pump (Hospital Utilities) – New Pattern Due to Severe
Damage…………………………… ............................................................................................ 142
10.7.1 Analysis and Diagnosis………………. ........................................................................ 142
10.8 Case Study (8) Cement Mill (Cement Industry) – Damaged Tooth in Girth Gear ............. 144
10.8.1 Analysis and Diagnosis………………… ..................................................................... 144
10.9 Case Study (9) Agitator (Chemical Industry) – Resonance Phenomenon .......................... 146
10.9.1 Analysis and Diagnosis………………. ........................................................................ 146
10.9.2 Recommendations and Actions………………............................................................. 146
10.10 Case Study (10) Circulation Multi-Stage Pump (Oil and Gas) – Defected Bearing ........ 148
10.10.1 Analysis and Diagnosis………………. ...................................................................... 148
10.10.2 Recommendations and Actions………………........................................................... 148
10.11 Case Study (11) Vertical Pump (Chemical Industry) – Looks Like Cantilever Beam ..... 150
10.11.1 Analysis and Diagnosis……………… ....................................................................... 150
10.12 Case Study (12) Chiller (Chemical Industry) – Early Fault Detection based on
Envelope…………………………… .......................................................................................... 151
10.12.1 Analysis and Diagnosis……………… ....................................................................... 151
10.13 Case Study (13) Gas Reciprocating Compressor (Oil and Gas) – Normal Pattern .......... 153
10.13.1 Analysis and Diagnosis…………………. .................................................................. 153

5
 10.14 Case Study (14) Water Pump (Chemical Industry)–Motor High Load Due to
Misalignment……………………….. ......................................................................................... 155
10.14.1 Analysis and Diagnosis……………… ....................................................................... 155
10.15 Case Study (15) Blower (Chemical Industry) – Ruptured Belt ........................................ 157
10.15.1 Analysis and Diagnosis……………………. .............................................................. 157
11 References……………… .......................................................................................................... 159

6
1 Chapter (1) Introduction
1.1 Maintenance
Past and current maintenance practices in both the private and government sectors would imply
that maintenance is the actions associated with the equipment repair after it is broken. The dictionary
defines maintenance as follows: "the work of keeping something in proper condition; upkeep." This
would imply that maintenance should be actions taken to prevent a device or component from failure
or to repair normal equipment degradation experienced with the operation of the device to keep it in
a proper working order. Unfortunately, over the past decade indicates that most private and
government facilities do not expend the necessary resources to maintain equipment in a proper
working order. Rather, they wait for equipment failure to occur and then take the necessary actions
to repair or replace the equipment.
The main objective of maintenance is to
 Improve equipment availability/reliability.
 Increase equipment effective life time.
 Increase resource utilization.
 Increase productivity.
 Reduce operating cost
 Reduce total cost rate
 Increase profitability ratio
In another words, the maintenance targets are to:
o Avoid costly/unplanned downtime due to equipment failures.
o Achieve efficient machinery operation and eliminate the cost of unnecessary energy usage.
o Improve the accuracy of machines by keeping them clean and running in top form.
o Understand the equipment capabilities so that planned outputs can be optimized.

1.2 Maintenance Techniques

The following types of maintenance programs are performed on most types of equipment:
 Breakdown or run to failure maintenance
 Preventive or time-based maintenance
 Predictive or condition-based maintenance
 Proactive or design out maintenance.

7
 Figure 1.1 Maintenance technique.

1.2.1 Run-to-failure (breakdown) maintenance technique

 This method has no continuous activity associated with it. Essentially, no maintenance activity
is performed on machinery until it fails or produces unacceptable product.
 It is the most expensive because it requires massive amounts of overtime and spare parts
inventory.
 It generates the highest lost production rates.
 An unexpected failure can be dangerous to personnel and the facility.

Figure 1.2 Run to failure maintenance technique.

1.2.2 Preventive maintenance technique

 From run to failure maintenance technique, we progress to periodic preventive maintenance
which is sometimes called "historical" maintenance. This is where the histories of each
machine type are analyzed and periodic overhauls are scheduled to occur before the
statistically expected problems occur.
8
 Preventive maintenance can be defined as follow: Actions performed on a time- or machine-
run-based schedule that detect degradation of a component or system with the aim of
sustaining or extending its useful life through controlling degradation to an acceptable level.
 In other words, preventive maintenance is a maintenance activity repeated at a predetermined
frequency. The frequency may be based on calendar time or other occurrences such as service
hours or number of starts. For example, someone may change the oil in a car every 3000 miles
or every 3 months. Also it is a time-based system of parts replacement. That is, after a period
of time has elapsed, parts will be replaced.
 Preventive maintenance activities are set up based on experience from past failures, equipment
Characteristics, and vendor recommendations. Preventing past failures from recurring decides
many procedures. If a pump has failed a number of times from alignment problems due to
fasteners coming loose, a Preventive maintenance procedure might be set up to check fastener
tightness for the pump each month.

Figure 1.3 preventive maintenance technique.

1.2.2.1 Advantages of the preventive maintenance technique
 Cost effective in many capital intensive processes.
 Additional cost savings are realized because manpower and any heavy equipment are
scheduled.
 Increased component life cycle.
 Generates Energy savings.
 Reduced equipment or process failure.
 Flexibility allows for the adjustment of maintenance

1.2.2.2 Disadvantages of the preventive maintenance technique

 It results in unnecessary replacing components in satisfactory condition.

9
  Sometimes, a machine in good condition is disassembled for preventive maintenance, parts
replaced, and then returned to service in poorer condition because of errors during assembly.
 Catastrophic failures still likely to occur.
 Is more labor intensive
 Includes performing unneeded maintenance activities, which has the potential to result in
incidental damage to components

1.2.2.3 Preventive maintenance tasks

 Checking and cleaning

 Inspecting
 Adjustments
 Lubrication
 Parts replacement and servicing
 Calibration
 Repair of components and equipment

1.2.3 Predictive maintenance technique

 Predictive maintenance can be defined as follow: Measurements that detect the onset of a
degradation mechanism, thereby allowing casual stressors to be eliminated or controlled
prior to any significant deterioration in the component physical state.
 It is maintenance actions based on machine condition data.
 Condition-based maintenance is related to predictive maintenance. In condition-based
maintenance, the equipment is inspected and based on a condition, further work or
inspections are done.
 Scientific application of proven predictive techniques increases equipment reliability and
decreases the costs of unexpected failures.
 Predictive maintenance involves monitoring the machine's vibration characteristics or
symptoms to diagnose its condition.
 The machine's condition also determines the required replacement parts.

1.2.3.1 Advantages of the predictive maintenance technique

 Minimizes or eliminates costly downtime - increases profitable uptime.

10
  Minimizes or eliminates catastrophic machinery failures - damage from catastrophic failure
is usually much more extensive than otherwise would have been.
 Reduces maintenance costs.
 Reduces unscheduled maintenance - repairs can be made on time that least affect
production.
 Reduces spare parts inventories - many parts can be purchased just in time for repairs to be
made during scheduled machinery shutdowns.
 Optimizes machinery performance - machinery always operates within specifications.
 Reduces excessive electric power consumption caused by inefficient machinery
performance - saves money on energy requirements.
 Reduces need for standby equipment or additional floor space to cover excessive downtime
- less capital investment required for equipment or plant.
 Increases plant capacity.
 Reduces depreciation of capital investment caused by poor machinery maintenance - well
maintained machinery lasts longer and performs better.
 Reduces unnecessary machinery repairs - machines are repaired only when their
performance is less than optimal.
 Minimizes or eliminates the possibility that machinery repairs were the wrong repairs.
 Reduces repair time and unnecessary repairs
 Reduces penalties that result from late deliveries caused by broken down or poorly
performing machinery.
 Reduces warranty claims due to poor product quality caused by poorly performing
machinery.
 Reduces the possibility of accepting recently purchased new or used machinery with defects
 Increases machinery safety - injuries are often caused by poorly performing machinery.
 Reduces insurance rates because well maintained machinery increases safety.

1.2.3.2 Disadvantages of the predictive maintenance technique

 Increases investment in diagnostic equipment.

 Increases investment in staff training.

 Savings potential not readily seen by management.

11
 1.2.3.3 Steps for implementing predictive maintenance
 Detection
o Detection involves measuring and trending levels for each machine included in the
program on a regularly scheduled basis. The objective is to reveal significant
increases in the machine level to warn of developing problem
 Analysis or Fault Diagnosis
o Analysis helps to pinpoint specific machinery problems by revealing their unique
characteristics.
 Correction
o Corrective action is taken after specific problem has been detected and identified
by planning and scheduling all activities to ensure that machine downtime is kept
to the absolute minimum.
1.2.3.4 Predictive maintenance programs
A variety of technologies can be used as a part of comprehensive predictive maintenance
program. Since mechanical systems or machines account for the majority of plant equipment,
vibration monitoring is generally the key component of most predictive maintenance programs.
However, vibration monitoring cannot provide all the information required to perform a successful
predictive maintenance. Therefore, other predictive maintenance programs have to be included.

Vibration analysis
Vibration frequency measurement provides information about entire machine operation and
condition monitoring for machine fault vibration patterns created by:
o Wear on rotating elements - bearings, shafts, gears, linkages
o Broken/missing parts - gearboxes, electric motors, compressors, pumps, conveyor drives
o Lubrication deficiency
o Impeller/fan blade material build-up
o Improper parts assembly
We will discuss this technique in more details.

Thermography

Thermography is a predictive maintenance technique that can be used to monitor the condition
of plant machinery, structures and systems. It uses instrumentation designed to monitor the emission
of infrared energy, i.e. temperature, to determine their operating condition. By detecting thermal
anomalies, i.e. areas that are hotter or colder than they should be, an experienced surveyor can locate

12
and define incipient problems within the plant. Infrared thermography employs computerized thermal
measurement and imaging equipment to detect abnormal heating due to faulty conditions. An infrared
survey can detect a problem before it manifests itself into a costly failure. Readings are taken as a
part of the predictive maintenance routine and tracked over time. Failure shows up as a change in
temperature. Infrared thermography helps to identify the following problems:
 Unbalanced or overloaded circuits
 Faulty electrical contacts or connections
 Improper grounding
 Bearing and coupling wear
Examples of what can be scanned:
 Electrical switch gear, breakers, bus connections, and contacts.
 Transformer connections.
 Motor and generator connections, windings, feeders and exciters.
Ultrasonic monitoring
Ultrasonic monitoring is used to detect high frequency signatures since the normal monitoring
range for vibration analysis is (1Hz to 20,000Hz), but the ultrasonic technique monitors the frequency
range between 20,000Hz and 100kHz. Leak (turbulent flow of liquids and gases through a restricted
orifice) is mainly detected by the ultrasonic monitoring since it produces high frequency signatures
that can be easily identified. In addition, ultrasonic monitoring can be used to detect early faults in
elements of rotating bearings.
Oil analysis (tribology)
Oil analysis is used to determine foreign media in machinery lubricants where periodic check
of oil samples will detect wear before failure occurs. Oil is the lifeblood of machinery oil analysis
provides valuable information about the condition of the machinery. Degradation of hoses, filters,
seals, air intakes, engine blocks, and oil can assist in performing inexpensive tests. By monitoring
wear particles produced by everyday machinery operation, we can advise which components require
replacement.

1.2.4 Proactive maintenance technique

 The latest innovation in the field of predictive maintenance is so called proactive
maintenance.
 It uses a variety of technologies to extend the operating life of machines and to virtually
eliminate reactive maintenance.

13
  The major part of a proactive program is root cause failure analysis which is the
determination of the mechanisms and causes of machine faults.
 The goal of proactive maintenance is to apply advanced investigation and corrective
technology toward solving the root cause of any machinery problem.
 In its ideal form, its goal is to eliminate failures in plant components.

1.2.4.1 Advantages of proactive maintenance technique

 Can be the most efficient maintenance program.

 Lower costs by eliminating unnecessary maintenance or overhauls.
 Minimize frequency of overhauls.
 Reduced probability of sudden equipment failures.
 Able to focus maintenance activities on critical components.
 Increased component reliability.
 Incorporates root cause analysis.

1.2.4.2 Disadvantages of proactive maintenance technique

 Can have significant startup cost, training, equipment, etc.

 Savings potential not readily seen by management 1].
Table 1.1 Comparison of maintenance techniques

14
2 Chapter (2) Fundamentals of Vibration
2.1 Introduction
Vibration is the periodic, back and forth motion (oscillation) of an object. We encounter
vibration in many different ways in our daily lives. Nearly all musical instruments utilize the periodic
vibration of mechanical elements to make sound; for example, pianos and guitars use the vibration of
a string and connected soundboard, clarinets use the vibration of a small reed, and trumpets use the
vibration of the player’s lips. Vibration also exists in nature. The motion of the tides is an example of
a very low frequency vibration that is produced by the gravitational force of the moon and sun. This
motion is an example of forced vibration (and resonance, in the case of the Bay of Fundy). A sudden
gust of wind acting on a tall pine tree can also produce a periodic, low frequency vibration of the tree,
an example of free vibration resulting from an initial impulse. The wind blowing on aspen leaves
produces a continuous, periodic motion of the leaves, an example of self-excited vibration. Machines,
because of internal and external forces, also vibrate.
Machinery vibration involves the periodic motion of rotors, casing, piping, and foundation
systems, all at the same time. Usually this vibration is so small that sensitive equipment is needed to
detect it. To illustrate the small size of machinery vibration, we can compare it to the diameter of a
human hair. The average diameter is approximately 130μm (about 5mil). This is an unacceptable
vibration level on some steam turbine generator sets that are the length of a house. Vibration in
machines causes periodic stresses in machine parts, which can lead to fatigue failure. If the motion
due to vibration is severe enough, it can cause machine parts to come into unwanted contact, causing
wear or damage. Because of this, the control of vibration is an important part of machinery
management, and the concepts underlying vibration must be thoroughly understood by the machinery
professional.
This chapter will discuss the basic concepts of linear vibration, the vibration of machines,
rotation, and the most commonly used vibration measurement units. Then, we will move onto a
discussion of the concepts of free vibration at a natural frequency, forced vibration, and that most
interesting marriage of the two, resonance. Finally, we will discuss self-excited vibration, where a
system can internally transfer energy to produce vibration at a natural frequency.

2.2 The Basic Vibration Signal

A vibration transducer is a device that converts mechanical motion into an electronic signal. A
displacement transducer can be used to measure the displacement, or position, of an object relative to
the transducer. For most transducers this is, ideally at least, a one-dimensional measurement. If the

15
object is vibrating, the position of the object relative to the transducer will change in a repeating
pattern over time. Figure 2.1 shows an object that is vibrating toward and away from a transducer.
The different images show the evolution of the system over time. The transducer converts the position
of the box on the transducer’s sensitive axis to an output voltage, which is displayed at the top of the
figure. The transducer output voltage, or signal, is proportional to the distance from the transducer to
the object, the gap. This changing voltage signal represents the relative position of the vibrating object
versus time.

Figure 2.1. The relationship of a displacement vibration signal to the motion of an object.
There are two primary characteristics that we can measure on this signal, frequency and amplitude. This
signal has a simple, sinusoidal shape, and it contains only one frequency. More complex (and typical) signals
contain several frequencies of vibration with different amplitudes.

2.3 Frequency
Frequency is the repetition rate of vibration per unit of time. The vibration signal in Figure 2.1
has only one frequency. The frequency of this signal is found by measuring the amount of time it
takes to complete one cycle of vibration (Figure 2.2). This length of time is called the period (T) and
is shown in the figure. It has units of seconds per cycle of vibration. The frequency (f) has units of
cycles/second, or hertz (Hz) and is the reciprocal of the period in seconds:
1
𝑓(𝐻𝑧) = (2.1)
𝑇
16
Figure 2.2. Three methods of amplitude measurement of a single-frequency (sine wave) signal.
In rotating machinery applications, we are often interested in expressing the frequency in cycles
per minute, or cpm, so that the frequency can be directly compared to the rotating speed of the
machine, measured in revolutions per minute, or rpm. The frequency in cpm can be calculated from
the period using this expression:
𝑐𝑦𝑐𝑙𝑒𝑠 𝑠 60
𝑓(𝑐𝑝𝑚) = (𝑓 ) (60 )= (2.2)
𝑠 𝑚𝑖𝑛 𝑇
There are 2π radians or 360° in a circle; this concept can be extended to say that there are 2π radians
or 360° in one cycle of vibration. Thus, the frequency can also be expressed in radians/second (rad/s):
𝑟𝑎𝑑 𝑟𝑎𝑑 𝑐𝑦𝑐𝑙𝑒𝑠 2𝜋
𝜔( ) = (2𝜋 ) (𝑓 )= (2.3)
𝑠 𝑐𝑦𝑐𝑙𝑒 𝑠 𝑇
The frequency ω (Greek lower case omega) is sometimes called the circular frequency. Several terms
are commonly used to describe frequency ranges in machinery (Figure 2.3):
Synchronous
The same as rotor speed. The “X” is equivalent to a mathematical multiplication symbol. Thus, 1X
can be read as “1 times rotor speed.”
Nonsynchronous
Any frequency except 1X.
Sub-synchronous
Any frequency less than 1X. This can include simple integer ratios such as 2/3X, 3/4X, etc., decimal
ratios such as 0.48X, 0.37X, etc., or subharmonics (see below).
Super-synchronous
Any frequency greater than 1X. This can include simple integer ratios such as 3/2X, 5/2X, etc.,
decimal ratios such as 1.6X, 1.8X, etc., or super-harmonics (see below).
Super-harmonic
17
A frequency greater than 1X that is an integer multiple: for example, 2X, 3X, 4X, etc.
Subharmonic
A frequency less than 1X that is an integer ratio with one in the numerator: for example, 1/4X, 1/3X,
1/2X, etc.

Figure 2.3. Machinery vibration frequency definitions.

2.4 Amplitude
Amplitude is the magnitude of vibration expressed in terms of signal level (for example,
millivolts or milliamps) or in engineering units (for example, micrometers or mils, millimeters per
second or inches per second, etc.). The amplitude can be measured using several methods. One is to
measure the total voltage change from the minimum of the signal to the maximum of the signal. This
method is used for displacement signals and is referred to as double amplitude, or peak-to-peak,
abbreviated pp. In Figure 2.2, the peak-to-peak voltage change represents a total change in position
of 120μm (4.7mil). Often, the machinery diagnostician simply needs to know how much the machine
rotor is vibrating relative to the available diametric clearance in the machine. Peak-to-peak
measurement makes this comparison relatively easy.
The single amplitude, or peak method, abbreviated pk, measures the voltage change from the
middle of the signal to the maximum value of the signal. This method yields an amplitude that is one
half of the peak-to-peak value. This method of measurement is commonly used for velocity and
acceleration vibration signals, but is not well-suited to the measurement of displacement signals for
the reason given above.
The root-mean-square method, abbreviated RMS, describes the amplitude of a continuously
changing signal as a form of average. As the name suggests, it is calculated by taking the square root
of the mean, or average, of the squares of the signal values. If, and only if, the signal is a sine wave
(single frequency), the RMS amplitude will be equal to 0.707 times the peak amplitude, and it will
be equal to 0.354 times the peak-to-peak amplitude. If the signal is not a sine wave, then the RMS
value using this simple calculation will not be correct. Most machine vibration signals are not sine
waves. Instead, they contain a mixture of different frequencies.
18
2.5 Displacement, Velocity, and Acceleration
Displacement describes the position of an object. Velocity describes how rapidly the object is
changing position with time, and acceleration describes how fast the velocity changes with time.
Figure 2.4 shows an oscillating pendulum observed by a displacement transducer and plots of
displacement, velocity, and acceleration. The displacement of the pendulum is measured relative to
the vertical, rest position.
Note that the peak velocity occurs when the pendulum passes through the vertical position and
is moving toward the transducer. As the pendulum reaches the end of its motion and is closest to the
transducer (the maximum positive value in the displacement plot), the velocity is zero, momentarily.
The pendulum starts moving back in the opposite direction with negative velocity. When the
pendulum passes through the vertical position again, the velocity reaches its maximum negative
value. When the pendulum reaches the opposite displacement extreme, the displacement is at the
minimum on the displacement plot, and the velocity is again zero.

Figure 2.4. The motion of a pendulum. Curve of displacement (green), velocity (blue), and
acceleration (orange).
Because the pendulum is driven by gravity, the acceleration is zero when the pendulum is in the
vertical position. As the displacement approaches the positive peak near the transducer, the force due
to gravity is acting in such a way as to reduce the velocity. Thus, the acceleration is negative, and it
reaches its maximum negative value (the minimum on the plot) when the displacement is maximum.
Then, as the pendulum starts back away from the transducer, the negative acceleration becomes
smaller, heading up toward the zero crossing on the plot. When the pendulum reaches the opposite
extreme, gravity is trying to stop the pendulum and push it back toward the transducer. At this point,
the acceleration reaches a maximum positive value on the plot. For single-frequency signals (sine
19
waves) only, such as shown in the illustration, there is a simple mathematical relationship between
displacement, velocity, and acceleration:

𝑑 = 𝐴𝑠𝑖𝑛(𝜔𝑡)
𝑣 = 𝐴𝜔 sin(𝜔𝑡 + 90𝑜 ) = 𝑑̇ (2.4)

̇ 𝑑̈
𝑎 = 𝐴𝜔2 sin(𝜔𝑡 + 180𝑜 ) = −𝐴𝜔2 sin(𝜔𝑡) = 𝑣 =
where d is the displacement, v is the velocity, a is the acceleration, A is the amplitude of the
displacement (the maximum displacement possible, expressed as either PK, PP, or RMS), and ω is
the frequency of vibration in rad/s. The numbers 90° and 180° represent the relative phase, or timing
between the signals.
In Equations 2.4, velocity leads displacement by 90°; that is, it reaches its maximum one quarter
of a cycle, or 90°, before (the reason for the plus sign) the displacement maximum. Figure 2.4 shows
a set of plots of Equations 2.4; note that the velocity maximum occurs when the displacement is zero
and rising.
Acceleration leads displacement by 180°, and acceleration leads velocity by 180° − 90° = 90°.
In the plot, the acceleration maximum occurs when the velocity is zero and rising, and the
displacement and acceleration change in opposite directions. Important note: the phase angles here
are based on a mathematical definition of phase, not on phase measured by instrumentation. As shown
in Equations 2.4, the amplitude of the velocity is related to the amplitude of displacement by a factor
of ω. Similarly, the amplitude of acceleration is related to the amplitude of displacement by a factor
of ω2. This has important implications for transducer selection, because the amplitude of velocity and
acceleration signals can become very small at low frequencies [2]. The displacement, velocity, and
acceleration conversion are displayed in Table 2.1. Also, the relations between the form of amplitude
measurement are shown in Table 2.2.
Table 2.1. Unit Conversion (British unit)

20
 Table 2.2 The relation between amplitude measurement forms

2.6 Time Domain

Vibration data, plotted as amplitude versus time, are referred to a time domain data profile.
Examples of simple and complex time domain are shown in Figure 2.5 and Figure 2.6, respectively.
Time domain data are helpful in detecting some mechanical faults and sometimes it’s so difficult to
interpret the pattern since the time signal includes all kinematics and noise initiated in the machine
during the measurement.

Figure 2.5. Example of simple time domain.

21
 Figure 2.6. Example of complex time domain.

2.7 Frequency Domain

frequency domain displays how much of the signal exists within a given frequency band
concerning a range of frequencies. The signature of frequency domain signal is displayed in Figure
2.7. A frequency-domain representation can include information on the phase shift that must be
applied to each sinusoid to be able to recombine the frequency components to recover the original
time signal.

Figure 2.7. Typical frequency-domain vibration signature.

22
 Frequency-domain data are obtained by converting time domain data using a mathematical
algorithm called fast Fourier transform (FFT). FFT allows each vibration component of a complex
machine-train spectrum to be shown as a discrete frequency peak. The frequency-domain amplitude
can be the displacement per unit time related to a particular frequency which is plotted on the y-axis
against the frequency on the x-axis. The relation between the frequency and time domain is illustrated
in Figure 2.8.

Figure 2.8. Relationship between time and frequency domain.

23
3 Chapter (3) Vibration Transducers

3.1 Introduction
At present in the industry like research and development, the ability of monitoring, measuring
as well as analyzing the vibration is very important. Unfortunately, the suitable techniques for making
a measurement system for vibration with precise repeatable are not always clear. There are some
challenges related while measuring the vibration which includes a selection of suitable component,
the configuration of the system, signal conditioning, analysis of waveform, and setup. This chapter
discusses what is a vibration transducer, types, working principle, and applications.

3.2 Vibration Transducer

A transducer is a device that converts one type of energy, such as vibration, into a different type
of energy, usually an electric current or voltage. Commonly used transducers are velocity pickups,
accelerometers and Eddy current or proximity probes. Each type of transducer has distinct advantages
for certain applications, but they all have limitations as well. No single transducer satisfies all
measurement needs. One of the most important considerations for any application is to select the
transducer that is best suited for the job. The various vibration transducers are discussed below.

3.3 Types of Transducers

There are different types of transducers based on the parameters of vibration source

3.3.1 Proximity Probe

The Proximity Probe, also called an "Eddy Current Probe" or "Displacement Transducer", is a
permanently mounted unit. It includes all sensors that perform non-contact detection in comparison
to sensors, such as limit switches, that detect objects by physically contacting them. Proximity
Sensors convert information on the movement or presence of an object into an electrical signal. There
are three types of detection systems that do this conversion: systems that use the eddy currents that
are generated in metallic sensing objects by electromagnetic induction, systems that detect changes
in electrical capacity when approaching the sensing object, and systems that use magnets and reed
switches.

24
 Figure 3.1. Proximity probe.
3.3.1.1 Inductive Proximity Sensors
The operating principle of inductive proximity sensors is based on a coil and oscillator that
creates an electromagnetic field in the close surroundings of the sensing surface (see Figure 3.2). The
presence of a metallic object (actuator) in the operating area causes a dampening of the oscillation
amplitude. The rise or fall of such oscillation is identified by a threshold circuit that changes the
output of the sensor. The operating distance of the sensor depends on the actuator's shape and size
and is strictly linked to the nature of the material.

Figure 3.2. Schematic diagram of inductive proximity sensor.

3.3.1.2 Capacitive Proximity Sensors
Capacitive proximity sensors detect changes in the capacitance between the sensing object and
the sensor. The amount of capacitance varies depending on the size and distance of the sensing object.
An ordinary capacitive proximity sensor is similar to a capacitor with two parallel plates where the
capacity of the two plates is detected. One of the plates is the object being measured (with an
imaginary ground) and the other is the sensor's sensing surface. The changes in the capacity generated

25
between these two poles are detected. The objects that can be detected depend on their dielectric
constant, but they include resin and water in addition to metals.

Figure 3.3. Schematic diagram of capacitive proximity sensor.

3.3.1.3 Features
 Proximity sensors detect an object without touching it, and they therefore do not cause abrasion
or damage to the object
 No contacts are used for output, so the Sensor has a longer service life (excluding sensors that
use magnets).
Proximity sensors are suitable for use in locations where water or oil is used.
 Proximity Sensors can be used in a wide temperature range.
 Proximity Sensors are not affected by colors 3].
3.3.1.4 Limitations
o Proximity Probes are difficult to calibrate absolutely.
o A variation in magnetic properties of shaft and geometric irregularities of shaft gives erroneous
signal components.
o Have limited frequency range.

3.3.2 Velocity Transducers

Today, a vast majority of machinery vibration information is recorded and quantified in terms
of velocity. Most vibration measurements in the process industries are analyzed in terms of inches
per second (ips) in the United States, or mm/sec on the SI systems. Velocity readings are generally
recommended for measurements in the 100 to 30,000 CPM (1.7 to 500 Hz) frequency band. Accurate
and reliable vibration sensors are critical for successful machinery monitoring programs.

3.3.2.1 Piezo-Velocity Transducers

Piezo-Velocity Transducers (PVT) perform general purpose accelerometers and electrodynamic
velocity pickups on slow speed equipment. In the 90 to 3600 CPM range, PVT internal integration
provides greater signal fidelity than standard accelerometers. Solid state PVTs are more reliable and

26
measure broader frequencies than electrodynamic pickups. For many permanent sensor installations
in paper mills, steel, and power generation facilities, a PVT sensor is the best investment. The PVT
is essentially a piezoelectric accelerometer with an on-board velocity converter. The transducer
employs a piezoceramic sensing element and dense seismic mass to produce a charge output
proportional to acceleration. The high impedance charge signal is converted within the sensors to a
low impedance voltage output and integrated to velocity [4]. Section views of compression and shear
mode PVTs are shown in Figure 3.4

Figure 3.4. Two types of piezoelectric sensors.

3.3.2.2 Velocity Pickup Transducers
Velocity pickup transducer consists of coils, wire, and magnet and was arranged so that if the
housing is moved, the magnet tends to remain stationary due to its inertia (see Figure 3.5). The relative
motion between the magnetic field and the coil induces a current that is proportional to the velocity
of motion and the unit produces a signal directly proportional to vibration velocity.

Figure 3.5 Velocity sensor: a) Geometry of two pickup coils, b) Finite element modeling.

27
 3.3.2.3 Features
 Simple to install
 Good response in middle range frequencies
 Stand high temperature
 Do not require external power
 Lowest cost
3.3.2.4 Limitations
o Low resonant frequency and phase shift
o Cross noise
o Big and heavy
o Require electronic integration for displacement

3.3.3 Piezoelectric Accelerometers

Piezoelectric accelerometer is a self-generating device. The idea is that when a piezoelectric
material is subjected to a mechanical stress (shear, compression, or bending), an electric signal
generates across its faces and proportional to the stress (see Figure 3.6). Its frequency range is from
below 1 to 10000 Hz. Certain natural and manufactured materials like quartz, tourmaline, lithium
sulfate, and Rochelle salt generate electrical charge when subjected to a deformation or mechanical
stress. The electrical charge disappears when the mechanical loading is removed. Such materials are
called piezoelectric materials and the transducers, which take advantage of the piezoelectric effect,
are known as piezoelectric transducers. The charge generated in the crystal due to a force Fx is given
by:
𝑄 = 𝑘𝐹𝑥 = 𝑘𝐴𝑃𝑥 (3.1)
Where k is called the piezoelectric constant, A is the area on which the force Fx acts, and Px is the
pressure due to Fx. The output voltage of the crystal is given by:

𝐸 = 𝑣𝑡𝑃 (3.2)
Where v is called the voltage sensitivity, and t is the thickness of the crystal. The values of the
piezoelectric constant and voltage sensitivity for quartz are 2.25*10-12C/N and 0.055 volt-meter/n,
respectively. These values are valid only when the perpendicular to the largest face is along the x-
axis of the crystal.

28
 Figure 3.6. Piezoelectric accelerometer.
3.3.3.1 Piezoelectric Materials
The two basic piezoelectric materials used in vibration sensors today are synthetic piezoelectric
ceramics and quartz. While both are adequate for successful vibration sensor design, differences in
their properties allow for design flexibility. For example, natural piezoelectric quartz has lower charge
sensitivity and exhibits a higher noise floor compared to the modern "tailored" piezoceramic
materials. Now, most vibration sensor manufacturers use piezoceramic materials developed
specifically for sensor applications. Special formulations yield optimized characteristics to provide
accurate data in extreme operating environments. The exceptionally high output sensitivity of
piezoceramic material allows the design of sensors with increased frequency response compared to
quartz. Both quartz and piezoceramics exhibit an output during a temperature change (pyroelectric
effect) when the material is not mounted within a sensor housing. Although this effect is much lower
in quartz than in piezoceramics, the elements are isolated from fast thermal transients when they are
mounted within the sensor housing. The dominant thermal signals are caused by metal case expansion
strains reaching the base of the crystal. These erroneous signals are based on the mechanical design
29
rather than sensing material (quartz or piezoceramic). Some sensor designs isolate strains and
minimize the thermal induced signals.
A typical piezoelectric transducer (accelerometer) is shown in Figure 3.6. In this figure, a small
mass is spring loaded against a piezoelectric crystal. When the base vibrates, the load exerted by the
mass on the crystal changes with acceleration and hence the output voltage generated by the crystal
will be proportional to the acceleration. The piezoelectric accelerometer uses the characteristics of
crystal material which responds to an applied force in accordance with Newton's second law (force =
mass x acceleration). It produces a voltage and electric charge proportional to the applied force that
is proportional to acceleration with a constant mass.

The piezoelectric accelerometer is very stable over long periods of time and maintain its
calibration if it is not abused. The accelerometer can be damaged if it is subjected to excessive heat
and dropped onto a hard surface. If it is dropped more than a few feet onto a concrete floor or steel
deck, the accelerometer has to be re-calibrated to insure the integrity of crystal. The small crack
reduces the sensitivity and affects the resonance, and thus the frequency response. It is a good idea to
calibrate accelerometers about once a year if they are in service with portable data collectors.

The frequency range of the accelerometer is very wide, extending from very low frequencies
in some units to several tens of kilohertz. The high-frequency response is limited by the resonance of
the seismic mass coupled to the springiness of the piezo element. This resonance produces a very
high peak in the response at the natural frequency of the transducer (near 30kHz for commonly used
accelerometers). A rule of thumb is that an accelerometer is useable up to about 1/3 of its natural
frequency as shown in the Figure 3.7.

Figure 3.7. Accelerometer frequency response.

30
 The resonant frequency of an accelerometer is strongly dependent on its mounting. The best
type of mounting is always the stud mount -- anything else will reduce the effective frequency range
of the unit. Today, most accelerometers used in industry are of internal integrated circuit preamplifier
(ICP) as shown in Figure3.6. This preamplifier is powered by a DC polarization of the signal lead
itself, so no extra wiring is needed.

Figure 3.8. ICP accelerometer.

3.3.3.2 Accelerometer Mass
When the accelerometer is mounted on the test object, it will increase the mass of the vibrating
system and the influence of the mechanical properties of the test object. The accelerometer mass
should be less than the dynamic mass of the vibrating part onto which it is mounted. The effect of the
accelerometer mass on the measured acceleration and system resonance frequency is as follow:
𝑀+𝑀𝑎
𝑎= 𝑎𝑎 (3.3)
𝑀

𝑀+𝑀𝑎
𝐹 = (√ ) 𝐹𝑎 (3.4)
𝑀

𝑎−𝑎𝑎 𝑀𝑎
𝑒= × 100 = × 100 (%) (3.5)
𝑎 𝑀+𝑀𝑎
Where aa is the acceleration measured with accelerometer mounted; a is the acceleration without
accelerometer; Fa is the resonance frequency with accelerometer mounted; F is the resonance
frequency without accelerometer; Ma is the accelerometer mass (typical value: (0.7  Ma  500g); M
is the mass of the vibrating system.

31
 3.3.3.3 Sensitivity
Sensitivity is accelerometer output per unit input in mv, (m/s2) for voltage or in pc, (m/s2) for
charge. The sensitivity of condition monitoring accelerometers typically ranges between 10 and 500
mV/g; higher and lower sensitivities are also available. To choose the correct sensitivity for an
application, it is necessary to understand the vibration amplitude levels to which the sensor will be
exposed during measurements. As a rule of thumb, if the machine produces high amplitude vibrations
(greater than 10 g RMS) at the measurement point, a low sensitivity (10 mV/g) sensor is preferable.
The typical low sensitivity/high frequency accelerometers are used on equipment with geared drives.
However, even lower speed machines may require less sensitive accelerometers if gear mesh
frequencies correspond to structural resonance. If the maximum vibrations are less than 10 g RMS
and the shaft speed greater than 600CPM (10Hz), a 100 mV/g sensor is probably the best choice. Low
frequency accelerometers with sensitivities of 500 mV/g or higher should be used on precision
spindles taming less than 600CPM. The sensitivity of different accelerometers and their natural
frequencies are displayed in Figure 3.9. It was observed that as accelerometer size increases, its
natural frequency decreases, its frequency range decreases, and its sensitivity increases.

Figure 3.9. Sensitivity of different accelerometers and related natural frequency.

3.3.3.4 Dynamic range

The dynamic range determines the largest and smallest signals that can be simultaneously
measured. The accelerometer gives an electrical output which is directly proportional to acceleration
over a very wide dynamic range (approximately 1:108). The lower end of the dynamic range is only
limited by the noise in cables and preamplifiers. The lower dynamic limit is determined by electrical
32
noise from connecting cables and amplifier circuitry. The lower limit is about 0.001g. The upper
dynamic limit is determined by the accelerometer's structural strength. For general purpose
accelerometers, the upper limit is up to 500g.

3.3.3.5 Frequency range

Mechanical systems tend to have much of their vibration energy contained in the relatively
narrow frequency range between 10 Hz to 1000 Hz but measurements are often made up to say 10
kHz because there are often interesting vibration components at these higher frequencies. Therefore,
it is important to ensure that the accelerometer covers the range of interest. The frequency range over
which the accelerometer gives a true output is limited at the low frequency by two factors. The first
one is the low frequency cut-off of the amplifier which follows it. This is not normally a problem as
the limit is usually well below 1Hz. The second factor is the effect of ambient temperature fluctuations
with respect to sensitive accelerometer. This effect is minimal for modern shear type accelerometers.
The upper limit is determined by the resonant frequency of the mass-spring system of the
accelerometer itself. As a rule of thumb, if we set the upper frequency limit to one-third of the
accelerometer's resonance(natural)frequency, we know that vibration components measured at the
upper frequency limit will be in error by no more than + 12% (see Figure 3.10).

Figure 3.10. Accelerometer frequency range.

3.3.3.6 Accelerometer Types
A. Compression type accelerometers
This design is especially used for accelerometers which are intended for measurement of very
high shock levels and other special purpose (see Figure 3.11)

33
 Figure 3.11 .Compression type accelerometer.
B. Planar Shear accelerometers
The main feature of planar shear accelerometer is its small size (see Figure 3.12). Therefore, it
is used in places where the space is limited. Also, it is used to measure on very light structure.

Figure 3.12. Planar shear type accelerometer.

C. Delta shear accelerometers
The delta shear accelerometer can be used for virtually any application. The advantage of the
delta shear accelerometer is its excellent specifications and very low sensitivity to environmental
influences.

Figure 3.13. Delta shear type accelerometer.

34
4 Chapter (4) Concept of Vibration Measurement
4.1 Introduction
Excessive vibrations are often a product annoyance and a symptom of machinery in poor
working condition. Low vibration means longer operating life, fewer maintenance problems, more
comfort, and lower noise levels. Today, vibration measurements and analysis are the most widely
used test, modeling, modification and diagnostic technique. For the study, understanding, and
reduction of this phenomenon. Machinery condition measurement can tell us when the machinery
condition starts to degrade. This allows maintenance to only occur when it is needed, rather than the
regular schedule maintenance.

4.2 Types of Monitoring System

4.2.1 Intermittent Monitoring System
Intermittent monitoring system also called an off-line condition monitoring system is based on
measuring vibration at selected time intervals in the field. Then, the analysis will be performed either
in the field or later after data collection. Intermittent monitoring provides information at an early stage
about incipient failure and usually is used where very early warning of faults is required, advanced
diagnosis is required, measurement must be made at many locations on a machine, and machines are
complex.
4.2.2 Permanent monitoring system
Permanent monitoring system also called an on-line condition monitoring system is based on
measuring vibration continuously at selected points of the machine and comparing the measurements
with the acceptable levels of vibration. In the permanent monitoring system, transducers are mounted
permanently at each selected measurement point. Usually, permanent monitoring system used in
critical applications where:
o No personnel are available to perform measurements (off shore, remote pumping stations, etc.).
o It is necessary to stop the machine before a breakdown occurs in order to avoid a catastrophic
accident.
o An instantaneous fault may occur that requires machine shutdown.
o The environment (explosive, toxic, or high-temperature) does not permit the human involvement

4.3 Time Interval between Measurements

Selection of time interval between measurements requires knowledge of the specific machine.
Some machines develop faults quickly and others run trouble-free for years, a compromise must be
found between the safety of the system and the time taken for measurements and analysis. The
35
following rough rule of thumb is useful to select a time interval between measurements which is one-
sixth to one-tenth the expected period between overhaul. In any case, measurements should be made
frequently in the initial stages of condition monitoring program to ensure that the measured vibration
levels measured are stable and no developed faults. When a significant change is detected, the time
interval between measurements should be reduced sufficiently so as not to risk a breakdown before
the next measurement. The trend curve will help in determining when the next measurement should
be performed.

4.4 Mounting the Accelerometer

The method of mounting the accelerometer to the measuring point is one of the most critical
factors in obtaining accurate results from practical vibration measurements. Sloppy mounting results
in a reduction in the mounted resonant frequency which can severely limit the useful frequency range
of the accelerometer. The main goal is to achieve close mechanical contact between the accelerometer
base and the attached surface to which it is to be attached. The ideal mounting is by a threaded stud
onto a flat, smooth surface as shown in Figure 4.1. A thin layer of grease applied to the mounting
surface before tightening down the accelerometer will usually improve the mounting stiffness. The
tapped hole in the machine part should be sufficiently deep so that the stud is not forced into the base
of the accelerometer. The resonant frequency attained is almost as high as the 32kHz.

Figure 4.1. Threaded stud mounting.

A commonly used alternative mounting method is the use of a thin layer of bees-wax for sticking
the accelerometer into place as shown in Figure 4.2. The resonant frequency is only slightly reduced
(to 29kHz). Because bees-wax becomes soft at higher temperatures, the method is restricted to about
40°C. Bees-wax mounting method achieves acceleration levels up to 100 m/s2 for clean surfaces.

36
 Figure 4.2. Bees wax mounting.
Where permanent measuring points are to be established on a machine and it is not wished to
drill and tap fixing holes, pad can be used as shown in Figure 4.3. It is attached to the measuring point
by means of a hard glue. Epoxy and cyanoacrylate types are recommended as soft glues can
considerably reduce the usable frequency range of the accelerometer.

Figure 4.3. Adhesive pad mounting.

The permanent magnet mounting method is a simple attachment method where the measuring
point is a flat magnetic surface (see Figure 4.4). This method reduces the resonant frequency of the
accelerometer to about 7 kHz. The holding force of the magnet is sufficient for vibration levels up
to 1000 to 2000 m/s2 depending on the size of the accelerometer.

Figure 4.4. Flat magnet mounting.

A hand-held probe with the accelerometer mounted on top as shown in Figure 4.5 is very
convenient for quick-look survey work, but can give gross measuring errors because of the low
overall stiffness. A low-pass filter should be used to limit the measuring range at about 1000 Hz.
The typical frequency response curves for different accelerometer mounting are displayed in Figure
4.6.

37
 Figure 4.5. Hand held mounting.

Figure 4.6. Frequency response for different accelerometer mounting.

4.5 The Influence of Environments on Accelerometers

Modern accelerometers and accelerometer cables are designed to have the minimum possible
sensitivity to many external influences shown in Figure 4.7. Nevertheless, in more severe
environments, special accelerometers are sometimes necessary.

38
 Figure 4.7. External influences of environments on accelerometers.

4.5.1 Base Strain

When an accelerometer is mounted on a surface that is undergoing strain variations, an output
will be generated as a result of the strain being transmitted to the sensing element. Accelerometers
are designed with thick and stiff bases to minimize this effect: Delta Shear types have a particularly
low base strain sensitivity because the sensing element is mounted on a center post rather than directly
to the accelerometer base.

Figure 4.8. Influence of base strain.

4.5.2 Nuclear Radiation

Most accelerometers can be used under gamma radiation doses of 10kRad/h up to accumulated
doses of 2 M Rad without significant change in characteristics. Certain accelerometers can be used
in heavy radiation with accumulated doses in the excess of 100 M Rad.
39
 Figure 4.9. Influence of nuclear radiation.

4.5.3 Magnetic Field

The magnetic sensitivity of piezoelectric accelerometers is very low, normally less than 0.01 to
0.25 m/s2 per k Gauss in the least favorable orientation of the accelerometer in the magnetic field.

Figure 4.10. Influence of magnetic field.

4.5.4 Humidity
The accelerometer is sealed by epoxy bonding or welding to ensure reliable operation in humid
environments. Teflon sealed accelerometer cables are recommended in case the measurement is
performed in liquids for short duration. Also, the accelerometer connector should be sealed with an
acid free room temperature vulcanizing silicon rubber or mastic. Industrial accelerometers with
integral cables should be used for permanent use in humid or wet areas.

Figure 4.11.Influence of humidity.

40
4.5.5 Corrosive Substances
Special materials, which are resistant to most corrosive substances, are used in the construction
of the accelerometer.

Figure 4.12.Influence of corrosive substances.

4.5.6 Acoustic Noise

Noise levels presented in machinery are not sufficiently high to cause any significant error in vibration
measurements. Normally, the acoustically induced vibration in the structure, where the accelerometer
is mounted, is far greater than the airborne excitation.

Figure 4.13.Influence of acoustic noise.

4.5.7 Transverse Vibrations

Piezoelectric accelerometers are sensitive to vibrations acting in directions other than coinciding
with their main axis. In the transverse plane, perpendicular to the main axis, the sensitivity is less than
4% of the main axis sensitivity (see Figure 4.14). As the transverse resonant frequency normally lies
at about 1/3 of the main axis resonant frequency, it should be considered where high levels of
transverse vibration are present.

41
 Figure 4.14.Influence of transverse vibrations.

4.5.8 Temperature Transients

Temperature transients (rapid fluctuations) can cause an electrical output from the
accelerometer, but this effect has been considerably reduced in the delta shear accelerometer. The
charges developed on the piezoelectric material due to temperature transients are mainly developed
on surfaces normal to the polarization of the piezoelectric material which are not measured. This is
the reason why the delta shear accelerometer is a good choice.
4.5.9 Sensitivity Change due to Temperature
Typical general purpose accelerometers can tolerate temperatures up to 125°C. At higher
temperatures the piezoelectric ceramic will begin to depolarize so that the sensitivity will be
permanently altered. Such an accelerometer may still be used after recalibration if the depolarization
is not too severe. For temperatures more than 125°C, accelerometers with a special piezoelectric
ceramic are available. Alternatively, a heat sink and mica washer can be inserted between the base
and the measuring surface and the stream of cooling air can provide additional assistance.

Figure 4.15.Sensitivity change due to temperature.

42
4.5.10 Earth Loops
If the accelerometer is fixed to a test object which is connected to the ground, an earth loop will
exist causing noise to superimpose the vibration signal from the accelerometer. This situation can be
avoided by mounting the accelerometer by the aid of an isolating mounting method. Another way of
avoiding earth loops is to use an accelerometer which has its piezoelectric material isolated from the
housing.
4.5.11 Triboelectric Noise
Triboelectric noise is often induced into the accelerometer cable by mechanical motion of the
cable itself. It originates from local capacity and charge changes due to dynamic bending,
compression, and tension of cable layers. This problem can be avoided by using a proper graphited
accelerometer cable and taping or gluing it down as close to the accelerometer as possible.

Figure 4.16. Influence of accelerometer cable movement.

4.6 Accelerometer Mounting Directions

Vibration is a result of dynamic forces in machines which have moving parts connected to
machine structures. Different machine parts vibrate with various frequencies and amplitudes. These
vibration data are collected from points on bearings, preferably in two directions, radial measurements
on bearing housing and axial measurements on the ends of bearing housing where the bearings take
up axial forces (see Figure 4.17). The axial direction (A) is parallel to the rotational axis of the
machine. The radial directions (R) is perpendicular to the rotational axis of the machine. The same
concept of axial and radial directions is applied to the vertical mounting machines as shown in Figure
4.18. The accelerometer should be mounted so that the desired measuring direction coincides with its
main sensitivity axis. Accelerometers are also slightly sensitive to vibrations in the transverse
direction, but this can normally be ignored as the transverse sensitivity is typically less than 4% of
the main axis sensitivity.

43
 Figure 4.17.Vibration measurement direction.

Figure 4.18.Measurement directions of vertical machines.

44
 Vibration measurement locations are defined by the rigid member such as bearing housings,
bearing pedestals, and machine casings. Each component has at least two measurement points. For
example, motor-pump machine shown in Figure 4.19 has two measurement points for motor and
pump. The point nears coupling is called drive end and the other point is non drive end.

Figure 4.19 .Vibration measurement locations for a typical horizontal machine.

45
5 Chapter (5) General Fault Analysis
5.1 Type of Vibration Signals
There are three types of vibration signals as shown in Figure 5.1; periodic (i.e., machinery shaft
speed), random (i.e., varying), transient (i.e., pump cavitation due to improper system line-up).

Figure 5.1. Basic types of signals.

5.2 Interpretation of Vibration Data

The key of using vibration signature analysis for predictive maintenance, diagnostic, and other
applications is the ability to differentiate between normal and abnormal vibration profiles. Vibrations
are usually found in rotary machines due to shaft rotation, rotors contact with bearings, gear-mesh,
and else. However, specific problems with machinery generate abnormal vibrations such as rotor
unbalance, loose bolts, misaligned shafts, and worn bearings. Predictive maintenance utilizes
vibration signature analysis based on the following facts:
1. All common machinery problems and failure modes have distinct vibration frequency
components that can be isolated and identified.
2. A frequency-domain vibration signature is generally used for the analysis since it includes
discrete peaks representing the specific vibration source.
3. There is a reason for each frequency component in the vibration signature of machine train.
4. When the machine vibration signature is compared over the time, it will be similar until some
event changes the vibration pattern.

46
While the increase of amplitude may indicate degradation of machine train, there are other parameters
have significant effect on change the amplitudes of vibration signature such as the load and operating
condition. In addition, it is important to note that a lower amplitude does not necessarily indicate an
improvement in the mechanical condition of the machine train. Therefore, it is important to
understand clearly the source of all amplitude variations.

5.3 Vibration monitoring

Most vibration based predictive maintenance programs rely on one or more monitoring techniques.
These techniques include broadband trending, narrowband trending, or signature analysis. The
following events are essential in machine condition monitoring:
o Date of time based maintenance
o Initial fault indication and tentative diagnosis
o Diagnosis fault and predicting date of breakdown
o First alarm level (warning)
o Planned repair effected
o Second alarm level (shutdown)
o Machine breakdown

5.4 Monitoring Techniques

5.4.1 Trending
Trend monitoring of vibration level is a more useful maintenance tool than a one-time survey
of the absolute magnitudes of features only. A slight linear increase of feature values turns to
exponential when the point of failure is approaching. The time of failure depends strongly on
machines and the stress caused by operating conditions. The slope of the feature values or different
order derivatives can be used in control in order to reduce the stress imposed on process equipment.
Since the remaining useful life can be estimated, the effects of the operating point on overall
equipment effectiveness can be calculated. Examples of healthy and faulty machine trending are
displayed in Figure 5.2 and Figure 5.3, respectively.

47
 Figure 5.2.Example of healthy machine (constant trending).

Figure 5.3.Example of faulty machine (linear increase trending).

5.4.1.1 Broadband Trending
Broadband trending acquires overall or broadband vibration readings from selected points on
the machine train. This data is compared to either a baseline reading taken from a new machine or to
vibration severity charts to determine the relative condition of the machine. Broadband measurement
between 10 and 1000Hz is usually used in this type of analysis. Broadband or overall RMS data
present the total vibration level for the machine at a specified measurement point where the data was
acquired. It does not provide any information about the individual frequency components or machine
dynamics that created the measured value.

48
 5.4.1.2 Narrowband trending
Narrowband trending, like broadband, monitors the total energy for a specific bandwidth of
vibration frequencies. Unlike broadband, narrowband analysis utilizes vibration frequencies that
represent specific machine components or failure modes. This method provides the means to quickly
monitor the mechanical condition of critical machine components, not just the overall machine
condition. This technique provides the ability to monitor the condition of gear sets, bearings, and
other machine components without manual analysis of vibration signatures.

5.4.2 Comparative Analysis

Comparative analysis directly compares two or more data sets in order to detect changes in the
operating condition of mechanical or process systems. Vibration signatures such as time domain and
frequency domain signals are used in in this type of analysis. The comparison is performed with
baseline data, known machine condition, or industrial reference data. It is important to use the same
units of frequency and amplitude to make a correct comparison.

5.4.3 Baseline Data

Vibration measurements are collected for each machine train after new installation. These data
set can be used as a reference or comparison data set for all future measurements. However, such data
set must be representative of the normal operating condition of each machine-train. There are three
criteria use the baseline comparisons: reset after maintenance, proper identification, and process
envelope.

5.4.4 Signature Analysis

Signature analysis provides visual representation of each frequency component generated by
the machine train. Vibration signatures are helpful to determine the required maintenance. Most
predictive maintenance based vibration programs use some forms of signature analysis in their
programs. However, the majority of these programs rely on comparative analysis rather than full root
cause techniques.

5.5 Quantifying the Vibration Level

The overall vibration level, which describes the vibration severity, can be quantified using RMS for
velocity measurements. RMS is the most relevant measurement of overall vibration based on
international standards organization (ISO). After taking measurement values of vibration velocity
levels in mm/s (RMS), they are evaluated based on ISO standard. In general, ISO 10816-3 is used to
evaluate the majority rotary machines.
49
Figure 5.4.Vibration severity ISO 10816-3.

50
6 Chapter (6) Understanding Data Collection
6.1 Introduction
Vibration data collection is one of the most important steps in analyzing issues caused by
machinery vibration. If done correctly, your analysis, recommendations and reporting become
much more accurate. It involves the ability to safely collect vibration data from a given
machinery under normal operating condition. This is useful for the analysis of the health
condition of the equipment and to provide corresponding recommendations to correct any
noted anomalies. To correctly and safely collect the vibration data, we need to understand how
this process is implemented. In this chapter, you will find the methods of colle cting data and
the sequence of processing this data in the collector.

6.2 Methods of Vibration Data Collection

6.2.1 Automated Method
In this method, accelerometers are permanently mounted on machines and terminated into
data acquisition Units, which collect and process data at designated intervals (every minute,
hourly, daily, weekly, etc.), depend on the criticality of the machine. As shown in Figure 6.1,
this method has a simplified layout with less personnel involvement in the day-to-day
operations of the process. Data collected can be made available instantly online for all
stakeholders to view.

Figure 6.1 .Schematic diagram of automated collection sequence.

6.2.2 Manual Method

This entails the use of trained personnel to download routes manually from software on
a computer onto a data collector. The data collector is carried around the plant (from one
machine to the other) and used to collect data from each designated location on a
machine. Data is collected on the average interval of once per month, or quarter, while under
normal operating condition. After data has been collected, the device is connected to a
computer and the routes uploaded back into the software for analysis. As depicted in Figure
6.2, this method relies on a lot of manual involvement by trained personnel to ensure the
process is efficient. Data collected is commonly exported and sent by email to other
stakeholders who may not have access to the local computer where the software is installed.
51
 Figure 6.2 .Schematic diagram of manual collection sequence.

6.3 Analog to Digital Converter

The vibration waves collected by transducers are analog signals. Analog signals have to be converted
into digital values for further processing (see Figure 6.3). This conversion from an analog signal to a
digital signal is done by an Analog to Digital (A/D) converter. The A/D conversion is essentially done
by microprocessors based on binary numbers system (the power of 2). A 12-bit A/D converter
provides 4096 (212) discrete intervals whereas a 16-bit A/D converter provides 65536 (216) intervals.
The greater the number of intervals, the better the amplitude resolution of the signal. A 12-bit A/D
converter would result in a resolution of 0.025% of the full scale, whereas a 16-bit A/D converter
would yield a resolution of 0.0015%. It is thus possible to collect a signal with large and small
amplitudes accurately.

52
 Figure 6.3. Analog to digital converter.

6.4 Sampling rate

Sampling is the process of recording the amplitude of a wave at given instants, and then generating a
curve from the recorded points. Thus, the collected discrete sampled data points (digital) are used to
reconstruct the wave which was originally in an analog form. The reconstructed digital wave has to
look similar to the original wave or in other words, take samples so that the digitized wave is an exact
replica of the original analog wave. The answer lies in the Nyquist sampling theorem, which states
that:

‘If we are not to lose any information contained in a sampled signal, we must sample at a
frequency rate of at least twice the highest frequency component of interest.’

Figure 6.4 (left) displays examples for low and high sampling rate. It was observed that when the
sampling rate is too low (under sampling rate), another wave can be generated with a different
frequency rather than the original wave frequency. The sampling rate needs to increase to make a
correct representation for the original wave as shown in Figure 6.4 (right). The phenomenon of a
lower frequency wave due to under sampling is called aliasing. All data collectors/analyzers have
automatically selected built-in sampling rates and fitted with an antialiasing filter. The antialiasing
filter is a low pass electronic filter which allows low frequencies to pass and prevents higher ones.
The filter removes all vibrations in the analog signal that have frequencies greater than half the
sampling rate. It is automatically tuned to the proper values as the sampling frequency is changed
(this occurs when the frequency range of the analyzer is changed by the user). It is very important to
note that filtering has to occur before digitization of the analog commences.

53
 Figure 6.4. Example of low sampling rate (left), high sampling rate (right).

6.5 Lines of Resolution

Since the digitized time wave forms are sometimes difficult to interpret, another form of data is
presented based on Fast Fourier Transform (FFT) technique. The FFT is a spectrum of amplitude vs
frequency. The resolution is the number of lines (bins or LOR) that are used to display the frequency
spectrum. If we have N samples in the time record, FFT produces a spectrum with N/2 lines. So, a
2048 sample time record would produce a 1024-line spectrum. But due to the nature of anti-alias
filters, we actually only keep 800 lines. The accepted rule is that the number of lines is equal to the
number of samples in the time record divided by 2.56. The LOR could be 200, 400, 800, 1600, 3200,
6400 and 12800. It can now be deduced that when resolution is high, there is a better distinction
between frequency peaks. Another important parameter is the required maximum frequency (F-max)
which has a direct effect on the sampling rate since the sampling rate equals 2.56 of maximum
frequency. The time required for collecting the data varies inversely with F-max. The higher F-max,
the quicker the FFT can be displayed. As a general guideline, the following advises are provided to
select F-max value:
o For general rotating machinery such as pumps, fans, blowers, and motors, set the F-max to 20x
or 40x where x is the running speed.
o When measuring vibrations on gearboxes, the F-max setting should be at least three times higher
than the gear mesh frequency, where the gear mesh frequency is the number of teeth of pinion
and gear times their respective running speeds.
o If the analysis on the machine is conducted for the first time, it is advisable to take two spectra,
one by selecting F-max at 10x and the other at 100x. This is to ensure that no important
frequencies are lost in the low or high frequency zone. Once the range of the suspicious
frequencies is noted, F-max could be selected accordingly.

54
7 Chapter (7) Signal Processing
7.1 Introduction
Since the captured data from the transducer are continuous analog signals, some methods of
signal processing have to be performed for further analysis such as windowing, averaging, filtering,
and else. In this chapter, windowing and its types are presented. Then, the averaging and forms of
averaging are included. Finally, the concept of anti-aliasing filter is identified.

7.2 Windowing
In signal processing, a window function is a mathematical function that is zero-valued outside
of some chosen interval, normally symmetric around the middle of the interval, usually near a
maximum in the middle, and usually tapering away from the middle as shown in Figure 7.1. In actual
practice, the segment of data within the window is first isolated, and then only that data is multiplied
by the window function values. Thus, tapering, not segmentation, is the main purpose of window
functions. The reasons for examining segments of a longer function include detection of transient
events and time-averaging of frequency spectra. The duration of the segments is determined by
requirements like time and frequency resolution. But that method also changes the frequency content
of the signal by an effect called spectral leakage. There are different types of window functions which
will be discussed in the following sections.

Figure 7.1. A form of window function.

7.2.1 Rectangular Window

55
 Even if you use no window, the signal is convolved with a rectangular-shaped window of
uniform height, by the nature of taking a snapshot in time of the input signal and working with a
discrete signal. This convolution has a sine function characteristic spectrum. For this reason, no
window is often called the uniform or rectangular window because there is still a windowing effect.

Figure 7.2. Rectangular window function.

7.2.2 Hanning and Hamming Window

The Hanning and Hamming window functions both have a sinusoidal shape. Both windows
result in a wide peak but low side lobes. However, the Hanning window touches zero at both ends
eliminating all discontinuity. The Hamming window doesn’t quite reach zero and thus still has a slight
discontinuity in the signal. Because of this difference, the Hamming window does a better job of
cancelling the nearest side lobe but a poorer job of canceling any others. These window functions are
useful for noise measurements where better frequency resolution than some of the other windows is
wanted but moderate side lobes do not present a problem. The Hanning window is the most commonly
used window in machinery analysis.

56
 Figure 7.3. Hanning and hamming window functions.

7.2.3 Blackman-Harris Window

The Blackman-Harris window is similar to Hamming and Hann windows. The resulting spectrum has
a wide peak, but good side lobe compression. There are two main types of this window. The 4-term
Blackman-Harris is a good general-purpose window, having side lobe rejection in the high 90s dB
and a moderately wide main lobe. The 7-term Blackman-Harris window function has all the dynamic
range you should ever need, but it comes with a wide main lobe.

Figure 7.4. Blackman-Harris window function.

57
7.2.4 Flat Top Window
The flat top window is sinusoidal as well, but it actually crosses the zero line. This causes a
much broader peak in the frequency domain, which is closer to the true amplitude of the signal than
with other windows.

Figure 7.5. Flat top window function.

When application of the flat top window is compared to when no window (rectangular window
or uniform) is applied, a broader peak is observed in the FFT. The Hann window also results in
broadening of the peak, but to a lesser extent than the flat top. Discerning between two very close
frequencies becomes very difficult due to the broadness of the peaks. When the intent is to identify
the presence of a signal component (a peak) at a specific frequency, it is best to apply a rectangular
window to do the analysis. But, if the magnitude of the peak is important, the flat top window is
clearly the best.

7.3 Averaging
Signal averaging is a signal processing technique applied in the time domain, intended to
increase the strength of a signal relative to noise that is obscuring it. By averaging a set
of replicate measurements, the signal-to-noise ratio (SNR) will be increased, ideally in proportion to
the square root of the number of measurements. There are different types of signal averaging which
will be discussed in the following sections.

7.3.1 Linear averaging

58
 Each FFT spectrum collected during a measurement is added to one another and then divided
by the number of additions. This helps in obtaining repeatable data and tends to average out random
noise. This is the most commonly used averaging technique.

Figure 7.6. Linear averaging.

The start and end of the time record is windowed and pushed to zero. You could say that the
vibration information at the start and end of the record is being wasted. So what if we overlapped the
waveforms? Instead of using an entirely new time record for each average, we could use part of the
previous record. This illustration shows just how much time is saved. We have not needed to use the
data shaded gray. That means less time spent next to the machine.

59
 Figure 7.7. Overlap averaging.

7.3.2 Peak Hold Averaging

Peak hold develops an envelope of the highest spectral line amplitude measured for any average.
This technique is used for viewing transients such as impact and run up / coast down tests.

Figure 7.8. Peak hold averaging.

60
7.3.3 Time Synchronous Averaging
Time synchronous averaging (TSA) is a signal processing technique that extracts periodic
waveforms from noisy data. The synchronizing signal is usually a pulse generated by a photocell or
an electromagnetic pickup at a reference position on the shaft circumference. The TSA is well suited
for gearbox analysis, where it allows the vibration signature of the gear under analysis to be separated
from other gears and noise sources in the gearbox that are not synchronous with that gear.

Figure 7.9. Peak hold averaging.

7.4 Anti-aliasing Filter

According to the Nyquist sampling theorem, the sampling rate should be at least twice the
maximum frequency component of the signal of interest. In other words, the maximum frequency of
the input signal should be less than or equal to half of the sampling rate.

How do you ensure that this is definitely the case in practice? Even if you are sure that the signal
being measured has an upper limit on its frequency, pickup from stray signals (such as the power line
frequency or from local radio stations) could contain frequencies higher than the Nyquist frequency.
These frequencies may then alias into the appropriate frequency range and thus give you erroneous
results.

61
 To be sure that the frequency content of the input signal is limited, a low pass filter (a filter that
passes low frequencies but attenuates the high frequencies) is added before the sampler and the A/D
converter. This filter is an anti-alias filter because by attenuating the higher frequencies (greater than
the Nyquist frequency), it prevents the aliasing components from being sampled. Because at this stage
(before the sampler and the A/D converter) you are still in the analog world, the anti-aliasing filter is
an analog filter.

An ideal anti-alias filter passes all the appropriate input frequencies (below f1) and cuts off all
the undesired frequencies (above f1). However, such a filter is not physically realizable. In practice,
filters look as shown in illustration (b) below. They pass all frequencies < f1, and cut-off all
frequencies > f2. The region between f1 and f2 is known as the transition band, which contains a
gradual attenuation of the input frequencies. Although you want to pass only signals with
frequencies < f1, those signals in the transition band could still cause aliasing. Therefore in practice,
the sampling frequency should be greater than two times the highest frequency in the transition band.
This turns out to be more than two times the maximum input frequency (f1). That is one reason why
you may see that the sampling rate is more than twice the maximum input frequency.

Figure 7.10. Difference between a) ideal and b) practical anti-alias filter.

62
8 Chapter (8) Signal Analysis
8.1 Introduction
It is highly desirable to be able to detect all types of faults during the operation of rotating
machinery. Such faults range from vibrations at very low frequencies (sub synchronous components
indicating looseness, oil whirl, faulty belt drive, etc.), to vibrations at very high frequencies (tooth-
meshing frequencies, blade-passing frequencies, frequencies of structural resonance excited by faulty
rolling-element bearings, etc.). Such detection should be applicable to all of machines in the plant
which operate from very low to very high speed. This requires to select some kind of signals each
one has exclusive characteristics to identify and detect certain types of faults. In this chapter, different
vibration analysis techniques are presented such as time domain analysis, frequency domain analysis,
phase analysis, envelope (demodulation) analysis, and Cepstrum analysis. Summary of vibration
evaluation techniques is displayed in Table 8.1.

Table 8.1. Summary of vibration analysis techniques.

Technique Application Description Fault/machine
Time domain Waveform visualization for Amplitude vs. time Rubbing, impact,
analysis identification of distortion clipping, cracked
teeth
Frequency Direct frequencies, natural Amplitude vs. frequency General Fault
domain frequencies, sidebands, diagnosis
analysis beats, sub harmonics, sum
and difference frequencies
Phase analysis Detection of developing Phase vs. frequency at General Fault
cracks in shafts varied machine location diagnosis
Balancing
Cepstrum Identification and separation Inverse Fourier transform Rolling element
analysis of families of harmonics of the logarithmic power bearings, bladed
Identification and separation spectrum machines, gearboxes
of families of sidebands
Envelope Signal analysis Amplitude demodulation Rolling element
analysis observation of a low bearing, electrical
frequency amplitude machines, gear boxes
modulation happening at
high frequency

8.2 Time Domain Analysis

The time base plot is the most fundamental graphic presentation of machinery dynamic data. It
shows how a single parameter (most often displacement, velocity, or acceleration, but also any other
dynamic measurement) from a single transducer changes on a very short time scale, typically a
fraction of a second. This is in contrast to trend plots, which display the value of a slowly changing
parameter (for example, axial position) over a much longer time scale, typically hours to months.
63
 A time base plot represents a small slice of time in the vibration history of the machine. Usually,
the amount of time involves only a few revolutions of the rotor. During this short length of time
(about 17ms for one revolution of a 3600 rpm machine), the overall behavior of the machine is not
likely to change significantly. However, unfiltered time base plots can clearly show a change in
machine response if sudden events occur in the machine or if the machine is rapidly changing speed
(such as an electric motor startup).

Time base plots have several important uses. They have the advantage in being able to clearly
display the unprocessed output from a single transducer. This allows us to look for noise on the signal
or to detect the presence of multiple frequency components. An important use of a time base plot is
to identify the presence and timing of short term transient events.

Multiple time base plots can allow us to establish timing relationships at different axial locations
along the machine train. Or, the time base plots from a pair of XY transducers can be used to
determine the direction of precession of the rotor shaft.

The time base plot is a rectangular (Cartesian) plot of a parameter versus time (Figure 8.1).
Time is on the horizontal axis, and elapsed time increases from left to right; events occurring later in
time will be to the right of earlier events. Because of the time scales encountered in rotating
machinery, the elapsed time is typically displayed in milliseconds (ms). The measured parameter,
converted from voltage to engineering units, is on the vertical axis (in the figure, this is 1μm/div).

Figure 8.1. Unfiltered and filtered time base plot.

64
 In unfiltered time base plots, digitally sampled signal voltages are first divided by the transducer
scale factor to convert them to equivalent engineering units. Then, the converted values are plotted
on the time base plot. The resulting waveform describes the instantaneous behavior of the measured
parameter from one moment to the next.

Filtered time base plots are constructed from the amplitude and phase of vibration vectors. The
plot is synthesized by computing a sine wave with the correct frequency, amplitude, and phase. This
synthesis process assumes that conditions in the machine don’t change significantly over the period
of time represented by the synthesized waveform. This is usually, but not always, a correct
assumption.

Compensation of time base plot is a process to remove unwanted signal content (noise) that is
unrelated to the machine behavior that we want to observe. This noise, electrical and mechanical
runout (glitch), bow, etc., can partially or completely obscure the dynamic information. Shaft
scratches or other surface defects create a pattern of signal artifacts that repeats every revolution. It
can be very useful to remove this noise to better reveal the important dynamic information. One type
of compensation is slow roll compensation of vibration vectors. Most often, we wish to remove the
effects of any 1X slow roll response that may be present in the signal so that we can see the 1X
response due to unbalance.

Figure 8.2. Slow roll compensation of filtered time base plot.

65
 Slow roll compensation is primarily applied to eddy current displacement transducer data
because these transducers have a significant output at slow roll speeds. At these speeds, output from
velocity and acceleration (seismic) transducers is extremely low, and there is usually no measurable
slow roll signal. For this reason, slow roll compensation is rarely, if ever, performed on seismic
transducer data. Filtered time base plots can be slow roll compensated using a 1X, 2X, or nx slow roll
vector. The slow roll vector is subtracted from the original vibration vector, and the new, compensated
vibration vector is used to synthesize the filtered waveform. The end result is a filtered time base plot
that is slow roll compensated.

8.3 Frequency Domain Analysis

Vibration vectors are filtered to multiples of running speed, 1X, 2X, 3X, etc., and provide us
with information about the behavior of a machine at one of those particular frequencies. This
information is presented to us as the amplitude and phase of the vibration at that frequency. However,
machines can vibrate at many different frequencies simultaneously. These frequencies can be related
or unrelated to running speed and include both sub-synchronous and super-synchronous frequencies.
Since these frequencies are associated with the operating condition of the machine, the machinery
diagnostician must have some way to determine the frequency content of a vibration signal in order
to make an accurate diagnosis.

Vibration frequencies sometimes appear as a series of harmonics. The series consists of the
lowest frequency in the series, called the fundamental, and a number of frequencies at integer
multiples of the fundamental. In a typical series, the amplitude of higher order frequencies will decline
rapidly. To avoid confusion, we will define a harmonic as any frequency that is an integer multiple
of the fundamental. The first harmonic is the fundamental, the second harmonic has a frequency of
twice the fundamental, the third harmonic has a frequency of three times the fundamental, etc. Often,
the term harmonics will be used as a general term to indicate integer multiple frequencies that are
above the fundamental. The fundamental vibration frequency of a series can be any vibration
frequency. Often, the fundamental is 1X, but it can also be any sub-synchronous or super-synchronous
frequency. For example, a series could be based on a fundamental at ½X and include 1X, 3⁄₂X, 2X,
⁵⁄₂X, etc. In this series, the ½X is the fundamental or first harmonic, 1X (= 2⁄₂X) is the second
harmonic, 3⁄₂X is the third harmonic, etc. Such a series can be generated by a ½X rub.

In the past few years, an important new tool, the full spectrum plot, has been developed that
uses the signals from a pair of orthogonal, shaft relative, vibration transducers. The full spectrum plot
compares to a conventional spectrum plot in the same way that the orbit plot compares to a time base

66
plot. The full spectrum plot contains much more information than the spectrum plot, including
vibration precession direction and orbit ellipticity. It is so important for machinery diagnostics that
we now refer to the conventional spectrum plot as the half spectrum plot.

Spectrum plots are used to identify the frequency components that are present in complex
vibration signals and to trend changes in the amplitude of frequency components. These frequencies
include running speed, multiples of running speed, line frequency electrical noise, gear mesh
frequencies, gear defect frequencies, rolling element bearing frequencies, and vane and blade pass
frequencies. Rotor system natural frequencies that are excited will also show up on the spectrum plot.
Sub-synchronous frequencies that are often associated with fluid-induced instability, compressor
rotating stall, compressor surge, or rub, and super-synchronous frequencies that are often associated
with rubs and shaft cracks can also be identified.

8.3.1 Half Spectrum Plot

Let’s start with a machine operating at a constant speed. Because of a combination of unbalance
and other effects, the machine is vibrating in a complicated way. The unfiltered vibration signal from
a transducer on this machine will reveal all of the complexity of the machine vibration that falls within
the bandwidth of the data collection system. A filtered signal from this transducer will be a sine wave
at the filter frequency with some amplitude and phase. Figure 8.3 shows a complex, time base
vibration signal (red) and a series of sine waves (blue) that add up to produce the time base signal.
Using the Fourier transform, the frequency, amplitude, and phase of these sine waves (called
components) can be computed from a digital sample of the original time base signal. The phase for
each signal is measured with respect to the trigger signal that starts the sampling process at time t0.
The Fourier transform output is equivalent to the output of a series of band-pass filters that have been
set to integer multiples of the lowest frequency signal, f1. (In practice, the lowest displayed frequency
of a spectrum plot will usually be below the lowest measured vibration frequency.)
The amplitude scale can be either linear or logarithmic. Logarithmic scaling is useful when there
is a need to compare signals with both very large and very small amplitudes. This scaling will clearly
display all signal components and the noise floor. However, when applied to rotating machinery work,
logarithmic scaling makes it more difficult to quickly discriminate between significant and
insignificant vibration components. Linear scaling has the advantage of showing the most significant
components; weak, insignificant, and low-level noise components are greatly reduced in scale or
eliminated. Because of its advantage for machinery work, linear scaling will be used for the spectrum
plots in this book.

67
 Figure 8.3. Time and frequency domain for a complex signal.
The frequency scale can be displayed in several frequency units. Most spectrum analyzers
display the frequency in hertz (Hz). This is useful when comparing machine vibration frequencies to
line frequencies, such as in induction motor or steam turbine generator diagnostics. Some software
packages can display units of Hz, cpm, or orders of running speed. Cpm is a very convenient unit
when working with machinery, because it is easy to compare a frequency in cpm to the running speed
of the machine in rpm. Spectrum plots that are displayed in orders of running speed (1X, 2X, etc.) are
also easy to interpret.
A digitally calculated spectrum consists of discrete frequency bins, or lines, of finite width. The
width of these lines, the resolution of the spectrum, is an important consideration. The maximum
resolution of a spectrum is determined by the ratio of the spectrum span (the range of displayed
frequencies) to the number of spectrum lines that are displayed:
𝑆𝑝𝑎𝑛
𝑅𝑒𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 = (3.1)
𝑁𝑜 𝑜𝑓 𝐿𝑖𝑛𝑒𝑠

The spectrum plot is a collection of these lines, arranged side by side. The width of each line is equal
to the resolution of the spectrum. For example, a 400 line spectrum with a span of zero to 200 Hz will
have a resolution of

68
 200 𝐻𝑧
𝑅𝑒𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 = = 0.5 𝐻𝑧/𝐿𝑖𝑛𝑒
400 𝐿𝑖𝑛𝑒𝑠
Thus, each frequency line will, ideally, represent only the spectral energy in a 0.5 Hz (30 cpm) wide band from
0.25 Hz below to 0.25 Hz above the center frequency of the line. Accuracy in the displayed amplitude and
frequency of a spectrum line will depend on where the actual vibration frequency is with respect to the center
frequency and which window function is used.
The limited resolution of spectrum plots means that there is always an uncertainty associated
with any frequency we wish to measure. In the example above, a frequency actually located at, for
example, 99.75 Hz, is displayed at 100 Hz. A spectrum plot with poor resolution will have a
corresponding large uncertainty in the measured frequency. Even good resolution spectra may not be
able to discriminate between vibration frequencies of exactly ½X and 0.49X, an important distinction
for malfunction diagnosis. Higher resolution (zoomed) spectra can help, but orbits with Key phasor
dots can sometimes be superior to spectrum plots for making this kind of discrimination.
Noise can be a problem in spectrum plots. The Fourier transform of a spike is a series of
spectrum lines extending to very high frequency. Thus, anything that produces a sharp corner in the
signal will produce a series of spectrum lines. Sharp corners can result from shaft rebound at a rub
contact point or from an inadequate sampling frequency (causing a corner where a smooth transition
really exists), among other things. Spikes or steps in the signal can originate from electrical noise
problems or from scratches on the shaft. Spectrum plots are calculated from uncompensated
waveforms, which may contain significant slow roll or glitch content. In general, the appearance of
spectrum lines as a series of harmonics should be viewed with caution. Use time base, orbit, or
cascade plots (below) to validate the data.

8.3.2 Full Spectrum

The half spectrum is a spectrum of a single time base waveform. The full spectrum is the
spectrum of an orbit. It is derived from the waveforms from two, orthogonal, shaft relative
transducers, combined with knowledge of the direction of rotation. The information from the two
transducers provides timing (phase) information that allows the full spectrum algorithm to determine
the direction of precession at each frequency. Because the timing information is critical, the two
waveforms must be sampled at the same time.
The full spectrum is calculated by performing an FFT on each transducer waveform. The results
are then subjected to another transform that converts the data into two new spectra that represent
frequencies of precession, one spectrum for X to Y precession and one for Y to X precession. The
last step uses the direction of rotation information to determine which of the spectra represents

69
forward and which represents reverse precession frequencies. When this process is completed, the
two spectra are combined into a single plot, the full spectrum plot (Figure 8.4).
Figure 8.5 shows the relationships among time base waveforms, half spectra, the orbit, and the
full spectrum. The Y and X time base waveforms and their half spectra are at the top. The two
waveforms combine to produce the orbit at bottom left. The data used to generate the half spectra are
further processed to produce the full spectrum at bottom right. Note that you cannot generate the full
spectrum by combining the two half spectra. In the full spectrum plot, the spectrum of forward
precession frequencies is on the positive horizontal axis and the spectrum of reverse precession
frequencies is on the negative horizontal axis. Thus, for each frequency, there are two possible
spectrum lines, one forward, and one reverse. The relative length of the spectrum lines for each
frequency indicates the shape and direction of precession of the orbit filtered to that frequency.

Figure 8.4. Calculation of the full spectrum.

70
Figure 8.5. Time base waveforms and their half spectra, the orbit and its full spectrum.

71
8.4 Phase Analysis
We have seen that vibration can be measured with two parameters, frequency and amplitude.
These are good and useful measurements. However, vibration never occurs in isolation; there is
usually a root, or fundamental, cause of vibration in a machine. The machinery specialist needs to
identify the root cause of any vibration problem, and it is often difficult to do this on the basis of
frequency and amplitude alone. More information is needed. One piece of information that can be
very useful is the timing difference, or phase, between events. If we know the timing between a root
cause and its effect, we can use our knowledge of rotor behavior to deduce the possible root causes
of what we can measure. This gives us a powerful tool for the diagnosis of rotating machinery.
Phase is another name for the relative timing between two events in different signals. For
example, in Figure 8.6, two similar vibration signals are shown. These two signals reach the positive
peaks at different times. This timing difference is referred to as a phase difference. In machinery
applications, the phase difference of equivalent events on different vibration signals is called relative
phase. Absolute phase compares the timing of an event on the vibration waveform to a different type
of reference signal, produced by a once-per-turn marker on a shaft. Both methods are commonly used,
and both have their particular applications. Even though phase compares the timing of events, it is
expressed in units of degrees of the vibration cycle. In machinery instrumentation, the timing
difference is usually expressed as a fraction of the complete 360° vibration cycle. This is a positive
number with increasing time delay and is called positive phase lag. For example, in Figure 8.6, the
green signal lags the blue signal by 60°.

Figure 8.6. Relative phase difference between signals.

72
8.4.1 Why Is Phase Important?
In an automobile engine, there may be spark and fuel, but if the timing is off, the engine may
not run. In rotor behavior, the timing (phase) is just as important as vibration amplitude and frequency
for effective diagnosis of machine behavior. It is also necessary for efficient balancing of machinery.
Balancing requires us to know the angular location of the unbalance (the heavy spot). We deduce this
location by using phase measurement of the vibration response of the machine combined with our
knowledge of rotor behavior. Without phase information, we would have to perform many more runs
to calculate an initial balance solution. When rotors vibrate, they deflect away from the machine
centerline. When the vibration is 1X, the point on the shaft which is on the outside of the deflected
shaft is called the high spot (Figure 8.7). The timing of the rotor high spot passage under a transducer
(the positive peak of displacement) provides important information about rotor behavior. It can be
compared to the timing at different axial positions in the same machine. The amplitude and phase
information can be combined to produce a picture of the deflection shape, or mode shape, of the rotor
at running speed. Phase can also be used to document mode shape of the casing or structure in a
similar manner.

Figure 8.7. The high spot of vibration on a rotor shaft.

The vibration at the source of a machine problem always happens earliest in time. As vibration
propagates away from the source location, it experiences a time delay (phase lag). Typically, the
farther from the source, the longer the phase lag. Thus, by measuring the relative phase between
different axial positions in a machine and looking for the earliest signal, we can sometimes determine
the location closest to the source of the problem. A healthy machine should operate and vibrate with
a repeatable pattern day after day. Once the baseline vibration characteristics of a machine are known
(including changes with load or other measurable factors), changes in vibration that break the pattern

73
indicate that something may be wrong with the machine. Changes in phase are just as important as
changes in vibration amplitude or frequency, and one may change independently of the others.
Changes in phase, amplitude, or frequency can warn that something is happening to the machine. For
all of these reasons, phase is an important tool in machinery diagnostics.
8.4.2 The Keyphasor Event
The most common vibration in rotor systems is associated with rotor unbalance. The unbalance
acts as a one-cycle-per-revolution rotating force on the rotor. This 1X forcing produces a 1X, or
synchronous, vibration response in the machine. Because unbalance is so common, it is desirable to
have a fixed, timing reference signal so that we can make phase measurements.
An eddy current displacement transducer looking at a keyway or key serves this purpose
perfectly. Such a transducer is called a Keyphasor transducer. While a Keyphasor transducer is
usually an eddy current transducer, it can be any type of transducer, as long as it provides a repeatable,
once-per-turn reference signal. Figure 8.8 shows a Keyphasor transducer observing a keyway and the
resulting signal. As the leading edge of the notch passes by the transducer, the observed distance will
increase suddenly, and the transducer signal voltage will abruptly become more negative. When the
trailing edge of the notch passes by the probe, the transducer signal voltage will return to normal.
Other shaft configurations for generating the Keyphasor signal are possible. This pattern of voltage
changes occurs once every revolution of the rotor and is referred to as the Keyphasor event. The
Keyphasor event can be thought of as a timing signal. When the event occurs, a timing clock is set to
zero and started. This event is used to measure the elapsed time between the Keyphasor event and an
event on another signal. Each time the rotor completes a revolution, the Keyphasor event occurs
again, resetting the imaginary timer. This once-per-turn event is the timing reference used by
instrumentation to measure the absolute phase of vibration signals at 1X and integer multiples (2X,
3X, …). It is also used to measure rotor speed and other important characteristics of the dynamic
response of the rotor.

Figure 8.8. The Keyphasor event.

74
8.4.3 Phase Measurement
Phase is used to compare the timing of two events in different signals. If an event occurs later
in time than a reference event, then it lags the reference event. Similarly, if an event occurs before
the reference event, then it leads the reference event. In order to make meaningful phase
measurements, the signals being used must consist of a single primary frequency component or, in
the case of the Keyphasor signal, one clearly identifiable reference event. For this reason, signals are
usually filtered to the frequency of interest before making the measurement, although unfiltered
signals can be used if they are dominated by one frequency.
The convention used in most vibration measurement instrumentation is to measure phase lag
with a positive number, sometimes called positive phase lag. For example, take the vibration signal
shown in Figure 8.9. The spike in the Keyphasor signal provides the timing reference. The event of
interest is the first positive peak of the vibration signal that occurs after (to the right of) the Keyphasor
event. Because the first positive peak occurs after the Keyphasor event, the time delay between the
two events is referred to as the phase lag.

Figure 8.9. The Keyphasor event and absolute phase.

75
 The numeric value of the phase lag is found by taking the measured time delay and comparing
it to the time for one cycle of vibration:

𝛷 = [𝑡2−𝑡 1 𝑚𝑠
𝑇 𝑚𝑠/𝑐𝑦𝑐𝑙𝑒
] [360
𝑑𝑒𝑔
𝑐𝑦𝑐𝑙𝑒
] (3.2)

where Φ is the phase lag in degrees, t1 is the time in milliseconds at which the reference event takes
place, t2 is the time at which the event of interest takes place, and T is the time in milliseconds for
one cycle of vibration. Phase measurements are usually reduced to numbers between 0° and 360°.
This is automatic in many instrumentation systems. For example, a calculated phase measurement of
395° would be reduced to 395° − 360° = 35°. The measurement of phase using this instrumentation
convention (positive phase lag) differs from the measurement of phase using a standard mathematical
convention. In the mathematical system, phase lag would be a negative number, and phase lead would
be positive.

8.4.4 Absolute Phase

Absolute phase is the phase angle measured from the Keyphasor event to the first positive peak of
the waveform. For 1X vibration in a circular orbit, this peak occurs when the rotor high spot is nearest
the vibration transducer (the high spot and the orbit will be discussed in more detail in a later chapter).
Thus, for 1X vibration, the absolute phase is sometimes said to represent the phase of the rotor high
spot. The absolute phase is found by measuring the time between the Keyphasor event and the time
of the first positive peak (t2 − t1), measuring the time for one complete cycle of vibration, T, and using
Equation 3.2. In Figure 8.9, t1 is measured as about 2.25 ms, and t2 is about 5.0 ms. T is 6.25 ms −
2.25 ms = 4.00 ms. Applying Equation 3.2, the phase, Φ, is
5 − 2.25 𝑚𝑠 𝑑𝑒𝑔
𝛷=[ ] [360 ]
6.25 − 2.25 𝑚𝑠/𝑐𝑦𝑐𝑙𝑒 𝑐𝑦𝑐𝑙𝑒
𝛷 = 250𝑜 𝑝ℎ𝑎𝑠𝑒 𝑙𝑎𝑔
Absolute phase can only be measured on vibration signals with a frequency that is a harmonic
(integer) multiple of running speed, such as 1X, 2X, 3X, etc. Attempting to measure absolute phase
for frequencies that are not integer multiples, or that are sub-synchronous, can lead to ambiguity. For
example, in Figure 8.10, a vibration signal is shown that has a frequency slightly higher than 1X.
Measuring absolute phase from successive Keyphasor marks results in different phase values. For
signals with frequencies that are not integer multiples of running speed, absolute phase measurements
are meaningless. Note that, if the instrument cannot combine the Keyphasor signal with the vibration
signal, it is still possible to measure absolute phase by displaying both waveforms. The top two plots

76
in Figure 8.9 can be used to measure the absolute phase; the basic definition of absolute phase is
unchanged.

Figure 8.10. Absolute phase and nonharmonic signals.

8.4.5 Relative Phase

Relative phase is the time delay between equivalent events (peaks, zero crossings, etc.) on two
separate signals, and doesn’t use the Keyphasor event. Figure 8.11 shows an example of a relative
phase measurement. The two vibration signals have been filtered to the same frequency and represent
the displacement vibration at different axial positions on a machine (but the same transducer
orientation). Select one of the signals as a reference, and then select a convenient measurement point
on it. Here, the positive peak is used. The equivalent point is located on the second signal, and the
relative phase is calculated using Equation 3.2. The result is, “Signal B lags signal A by 130°.” It is
important to note which signal leads (or lags) the other and by how much, or the description is not
correct or complete. Normally, the reference signal is stated last. In this case, signal A was the
reference signal, and signal B was measured relative to the reference. For another example, suppose
a relative phase measurement is made on the signals in Figure 8.12. Initially the “A” signal is selected
as a reference. The relative phase is measured as “B lags A by 270°.” The result can be restated as “B
leads A by 90°.” Either is correct. In order to make a relative phase measurement, three conditions
should be met. First, the two signals must have the same frequency, but it can be any frequency, not
just integer multiples of running speed. If we attempt to measure the relative phase between signals
with different frequencies, we will obtain different results depending on where we make the
measurement, and the result will be meaningless. Second, relative phase measurements are most often
applied to vibration signals with the same units of measurement. If we want to compare the relative
phase of the rotor vibration at different axial positions, we would typically compare two displacement,

77
two velocity, or two acceleration signals. (Since the phase relationship of acceleration and velocity,
and velocity and displacement are always 90°, we could compare these signals by applying the right
correction.) Third, vibration transducers should have the same radial orientation if they are in different
axial planes. Relative phase measurements can be made between transducers with different
orientations, as long as they are in the same plane, to determine the direction of precession of a rotor.

Figure 8.11. Relative phase between equivalent points on signals.

Figure 8.12. Equivalent measurements of relative phase.

78
8.5 Cepstrum Analysis
In Fourier analysis, the cepstrum is the result of computing the inverse Fourier transform (IFT)
of the logarithm of the estimated signal spectrum (see ). The method is a tool for investigating
periodic structures in frequency spectra. The term cepstrum looks like a spelling error and is often
corrected by proof readers. It is not an error; the term has been chosen to give the function a scent of
“spectrum” because it is advantageous to consider the cepstrum as the spectrum of (the logarithm of)
the autospectrum. To support this perception, all terms related to the cepstrum are warped versions
of the corresponding spectrum terms:

• Spectrum: cepstrum

• Frequency: quefrency

• Harmonics: rahmonics

• Low-pass filter: short-pass lifter

• High-pass filter: long-pass lifter

The cepstrum is useful for all periodic manifests in a response spectrum from a rotating machine
like amplitude modulation of gear-tooth meshing frequency due to load variation, gear eccentricity
or general wear of gears and, of course, for identifying local faults.

Figure 8.13. Different between spectrum and cepstrum.

79
 Example: Local Fault in a 5:3 Gear: - The input gear spins with 3000 RPM = 50 Hz, which
means that a fault on a tooth will generate a repeated impact with the repetition time t = 1/50 Hz = 20
ms. The output gear spins with 3000 × 3/5 RPM = 1800 RPM = 30 Hz. A faulty tooth on this gear
will give a repeated impact every t = 1/30 Hz = 33.3 ms. The noise from the gear suggests a local
fault, and the vibration was measured on the housing of the gear. The cepstrum of the vibration
spectrum will reveal a possible periodicity in the spectrum. The green curve in Figure 8.14 is the
cepstrum and it reveals two rahmonic families indicating two local faults with a different fault rate: •
A rahmonic family with the fundamental rahmonic = 20 ms (50 Hz) which means a local fault on the
input gear (a faulty tooth) and a rahmonic family with the fundamental rahmonic = 30 ms (33.3 Hz),
indicating a faulty tooth on the output gear. The reason for the rahmonic component at quefrency 60
ms being higher than the other components is that both local faults contribute to this component: The
3rd rahmonic of the fault on the input gear and the 2nd rahmonic of the fault on the output gear.

Figure 8.14. Vibration Cepstrum.

8.6 Envelope Analysis

Envelope Detection or Amplitude Demodulation is the technique of extracting the modulating
signal from an amplitude-modulated signal. The result is the time history of the modulating signal
(see Figure 8.15). This signal may be studied/interpreted as it is in the time domain or it may be
subjected to a subsequent frequency analysis. Envelope Analysis is the FFT frequency spectrum of
the modulating signal.
Envelope Analysis can be used for diagnostics/investigation of machinery where faults have an
amplitude modulating effect on the characteristic frequencies of the machinery. Examples include
faults in gearboxes, turbines and induction motors. Envelope Analysis is also an excellent tool for
diagnostics of local faults like cracks in rolling element bearings.

80
 Figure 8.15. Envelope analysis concept.

8.6.1 Load Variation Modulation

The radial load in the bearing determines the strength of the impact from rolling over a fault. A
fault in a stationary bearing race will be subjected to the same force at each roll and consequently all
the pulses in the pulse train will be of equal strength/height. On the other hand, a fault in a rotating
race will be subjected to a varying force, the variation repeating itself with the RPM of the race. This
means that the pulse train will be amplitude-modulated with the RPM of the race, and in turn all the
harmonics in the line spectrum, ball passing frequency outer race (BPFO), or ball passing frequency
inner race (BPFI) will appear amplitude-modulated by the RPM of the race. Likewise, the ball fault
frequency (BFF) caused by a ball/roller fault, will be amplitude-modulated by the fundamental train
frequency (FTF). If there is more than one fault of a kind, the line spectrum will still contain the
harmonics of the bearing frequency. Only the shape of the spectrum will change depending on the
relative positions of the faults.

8.6.2 Envelope Detection

The bearing frequencies are present throughout the spectrum (the 1/T line spectrum), but
obscured at lower frequencies by other vibrations. However, there is a technique that makes it possible
to extract the bearing frequencies from the part of the vibration spectrum where the 1/T line spectrum
is dominant, that is, amplitude demodulation: A band-pass filter, with center frequency fc, filters out

81
the selected part of the spectrum, the output is shifted (heterodyned) to low frequency (fc → DC) and
subjected to envelope detection.
If the band-pass filter encompasses a range where the 1/T line spectrum is dominant, the
resulting time history will be dominated by the envelope of the original pulse train. This envelope
time history can now be subjected to FFT analysis for easy identification of Bearing Frequencies. The
figures below illustrate these properties. A synthesized time signal is composed as follows:
A pulse with a repetition time of 25 ms (~ 40 Hz) is subjected to a certain amplitude modulation with
a repetition time of 250 ms (~ 4 Hz = 240 RPM) plus random noise, 0 – 1 kHz, with substantially
higher power. This resembles a BPFI = 40 Hz fault, load variation modulated by the Shaft-speed =
240 RPM, the bearing situated in a ‘noisy’ machine. The modulation of the BPFI may be perceived
as contaminating the clean 40 Hz, BPFI, spectrum, but it is, on the other hand, very informative. In a
real situation where bearing frequencies would not be known in advance, this spectrum is the
signature of a rotating race fault, most often the inner race. In the example, the BPFI is the 10 th
harmonic of the shaft speed. This fact can be used to identify a load modulated race fault.

8.7 Crest Factor

The Crest Factor (CF) is used to characterize the quality of a waveform, and in other cases, CF
is used to show if a waveform contains a great deal of peaks. CF is defined as the ratio of the peak
level of the input signal to the RMS level.
𝑃𝑒𝑎𝑘 𝑙𝑒𝑣𝑒𝑙
𝐶𝑟𝑒𝑠𝑡 𝐹𝑎𝑐𝑡𝑜𝑟 = (3.3)
𝑅𝑀𝑆
Therefore, peaks in the time series signal will result in an increase in the CF value. This feature is
used to detect changes in the signal pattern due to impulsive vibration sources such as tooth breakage
on a gear or defect in a bearing.
When the discrete bearing frequencies cannot be identified because of high background noise
and/or widespread damage, the kurtosis and CF are the best techniques for the condition monitoring
of bearings. The curve in Figure 8.16 shows a typical trend for CF as bearing condition deteriorates.
Initially, there is a relatively constant ratio of peak to RMS value. As a localized fault develops, the
resulting short bursts increase the peak level substantially, but have little influence on the RMS level.
The peak level will typically grow to a certain limit. As the bearing deteriorates, more spikes will be
generated per ball-pass, finally influencing the RMS level, even though the individual peak levels are
not greater. Towards the end of bearing life, the CF may have fallen to its original value, even though
both peak and RMS levels have increased considerably. The best way to trend the data is as illustrated;
RMS and peak levels on the same graph, with CF inferred as the difference between the two curves.

82
 Figure 8.16. Crest Factor concept.
8.8 Kurtosis Analysis
Kurtosis is defined as the 4th moment of a distribution and measures the relative peakedness or
flatness of a distribution compared to a normal distribution. Kurtosis provides a measure of the size
of the tails of distribution and is used as an indicator of major peaks in a set of data. The equation for
kurtosis is given by:
∑𝑁
𝑛=1[𝑌(𝑛)−𝜇]
4
𝑘= (3.4)
𝑁×(𝜎2 )2

Where Y(n) is the raw time series at point n, is the mean of the data, σ2 is the variance of the data,
and N is the total number of data points.
Typical values of the signal kurtosis range from 3 to 45 depending upon the condition of the
bearing. As a general rule, variations in kurtosis closely follow variations in the CF and the values
higher than about 4 are indicative of damage. Kurtosis provides a wider dynamic range and is
unaffected by changes in speed and loading because the kurtosis is based upon detecting
impulsiveness.

83
9 Chapter (9) Fault Diagnostics
9.1 Introduction
Some of the machinery defects detected using vibration analysis are listed below:
Unbalance Oil whip/whirl
Bent shaft Electrical faults
Eccentricity Cavitation
Misalignment Shaft cracks
Looseness Rotor rubs
Belt drive problems Resonance
Gear defects Hydraulic and aerodynamic forces.
Bearing defects

We will now look at each one of the above cases in detail and see how they manifest in
vibration analysis.

Figure 9.1. Common faults in rotary machines.

84
9.2 Unbalance
Vibration due to unbalance of a rotor is probably the most common machinery defect. It is
luckily also very easy to detect and rectify. The International Standards Organization (ISO) define
unbalance as: That condition, which exists in a rotor when vibratory, force or motion is imparted to
its bearings as a result of centrifugal forces. It may also be defined as: The uneven distribution of
mass about a rotor’s rotating centerline. There are two new terminologies used: one is rotating
centerline and the other is geometric centerline. The rotating centerline is defined as the axis about
which the rotor would rotate if not constrained by its bearings (also called the principle inertia axis
or PIA). The geometric centerline (GCL) is the physical centerline of the rotor. When the two
centerlines are coincident, then the rotor will be in a state of balance. When they are apart, the rotor
will be unbalanced. There are three types of unbalance that can be encountered on machines, and
these are:
1. Static unbalance (PIA and GCL are parallel)
2. Couple unbalance (PIA and GCL intersect in the center)
3. Dynamic unbalance (PIA and GCL do not touch or coincide)
4. Unbalance – overhung rotor

Figure 9.2. Unbalance criteria.

9.2.1 Static Unbalance

For all types of unbalance, the FFT spectrum will show a predominant 1X frequency of
rotational speed in the radial direction. Vibration amplitude at 1X will vary proportional to the square
of the rotational speed. It is always present and normally dominates the vibration spectrum (Figure
9.3).

85
 Figure 9.3. Static unbalance.

9.2.2 Couple Unbalance

In a couple unbalance (Figure 9.4), the FFT spectrum displays a single 1X frequency of
rotational speed. The amplitude at 1X varies proportional to the square of speed. This defect may
cause high axial and radial vibrations. Couple unbalance tends to be 180° out of phase on the same
shaft. Note that almost a 180° phase difference exists between two bearings in the radial directions.

Figure 9.4. Couple unbalance.

86
9.2.3 Dynamic Unbalance
Dynamic unbalance is the most common type of unbalance and is a combination of static and
couple unbalances. It occurs when the principal axis of inertia and the axis of rotation are skew lines.
This unbalance can be understood as two different masses placed arbitrary on the rotor. FFT spectrum
displays a single 1X frequency of rotational speed. The amplitude at 1X varies proportional to the
square of speed. This defect may cause high axial and radial vibrations. Dynamic unbalance gives
different phase on the same shaft.

9.2.4 Unbalance – Overhung Rotors

The axial phase on the two bearings will seem to be in phase whereas the radial phase tends to
be unsteady. A special case of dynamic unbalance can be found in overhung rotors (see Figure 9.5).
Dynamic unbalance in overhung rotors causes high 1X levels in radial and axial direction due to shaft
bending which causes axial movement.

Figure 9.5. Unbalance-overhung rotor.

9.3 Bent Shaft

A bent shaft and bowed rotor are actually the same phenomena. The bent shaft is measurable
outside the machine housing while the bowed rotor is the same condition inside. These defects
sometimes develop on a motor that has been allowed to sit stationary for a long time. When a bent
shaft is encountered, the vibrations in the radial as well as in the axial direction will be high. Axial

87
vibrations may be higher than the radial vibrations. The FFT will normally have 1X and 2X
components. If the amplitude of 1X is dominant, then the bend is near the shaft center as shown in
Figure 9.6. If the amplitude of 2X is dominant, then the bend is near the shaft end.

Figure 9.6. Bent shaft near the shaft center.

9.4 Eccentric rotor

Eccentricity occurs when the center of rotation is at an offset from the geometric centerline of
a sheave, gear, bearing, motor armature or any other rotor. The maximum amplitude occurs at 1X of
the eccentric component in a direction through the centers of the two rotors (see Figure 9.7)

Figure 9.7. Eccentric rotor.

88
9.5 Misalignment
Misalignment, just like unbalance, is a major cause of machinery vibration. Some machines
have been incorporated with self-aligning bearings and flexible couplings that can take quite a bit of
misalignment. However, it is not uncommon to come across high vibrations due to misalignment.
There are basically two types of misalignment:
1. Parallel misalignment: the shaft centerline of the two machines is parallel to each other and
have an offset.
2.Angular misalignment: the shaft centerline of the two shafts meets at angle with each other.
3.Combination of parallel and angular misalignments

9.5.1 Parallel Misalignment

Parallel misalignment results in two hits per cycle and therefore, 2X vibration in the radial
direction is dominant. Parallel misalignment shows high radial vibration that approaches a 180° phase
difference across the coupling in the radial direction. Thus, we will see both the 1X and 2X peaks.
Coupling construction will often significantly influence the shape of the spectrum if misalignment is
severe (see Figure 9.8).

Figure 9.8. Parallel (offset) misalignment.

9.5.2 Angular misalignment

Angular misalignment primarily subjects the driver and driven machine shafts to axial
vibrations at the 1X, high 2X, and 3X in addition to 180o phase difference across the coupling in the
axial direction.

Figure 9.9. Angular misalignment.

89
 9.6 Soft Foot
Soft foot is one of the most prevalent conditions found in rotating machinery. This condition, if
not corrected, makes an alignment job much more difficult and sometimes impossible. Soft foot can
cause high levels of vibration. Soft foot can be seen as looseness and exhibit a high 1X vibration
signature and 180o phase difference between base plate and concrete base. Soft foot can be occurred
due to base plate problem or bad concrete base leveling as shown in Figure 9.10. There are four types
of soft foot as discussed in the following sections.

Figure 9.10. Causes of soft foot.

9.6.1 Parallel Soft Foot

Parallel soft foot or (short foot) exists when the foot simply does not reach the base and creates
a gap between the foot and base. The bottom of the foot is parallel to the base plate. Tightening the
hold-down bolts will result in a distortion of the machine’s frame as the soft foot is drawn down to
the base. This condition is easiest to detect using either a feeler gauge or dial indicator.

90
9.6.2 Angular Soft Foot
Angular soft foot can occur when the foot is touching the base on either the outside or inside
portion of the foot, but the other side of the foot is bent away creating an angle between the base and
the bottom of the foot. This type can be corrected by using plastic shims to fill the angular gap.

9.6.3 Squishy Foot

Squishy foot, sometimes called spring foot, exists when the gap between the foot and base has
already been filled with corroded / soft shims. The machine will appear to be fixed of soft foot
problems until the hold-down bolts are tightened. This type can be corrected by replacing the corroded
/ soft shims by integrated / hardened once.

9.6.4 Stress-Induced Soft Foot

Perhaps the most difficult soft foot condition to detect is caused by forces that are external to
the machine. This is referred to as stress-induced soft foot. It can be the result of pipe strain or stresses
induced by the electrical connections as well as drastic misalignment. Stress-induced forces can be
created during any stage of the alignment process therefore, eliminating this kind of soft foot may
require more than one check such as piping strain check.

9.7 Looseness
Mechanical looseness or the improper clearance between component parts is generally
characterized by a long string of harmonics of running frequency with abnormally high amplitudes.
In some machines, vibration levels may be excessive as a consequence of components being
assembled too loosely, for example in the case of a bearing, which is not properly secured. Mechanical
looseness can be classified to three categories:
1) Structure looseness.
2) Machine / base plate looseness.
3) Internal assembly looseness.
9.7.1 Structure Looseness (Non-Rotating Looseness - Type A)
This type of looseness is caused by structural looseness or weakness in the machine’s feet, base
plate, or foundation. It can also be caused by deteriorated grouting, loose hold down bolts at the base,
and distortion of the frame or base (soft foot). Phase analysis may reveal approximately 180° phase
shift between vertical measurements of machine’s foot / base plate and foundation as shown in Figure
9.11. When the soft foot is suspected, an easy test to confirm for it is to loosen each bolt, one at a
time, and see if this brings about significant changes in the vibration. In this case, it might be

91
necessary to re-machine the base or install shims to eliminate the distortion when the mounting bolts
are tightened again.

Figure 9.11. Structural looseness.

9.7.2 Machine Components / Base Plate Looseness (Non-Rotating Looseness -

Type B)
This problem is associated with loose pillow-block bolts, cracks in the frame structure, or the
bearing pedestal as displayed in Figure 9.12. This generates higher harmonics due to the rocking
motion of the pillow block with loose bolts. Since the movement is restricted in one direction, we can
see a truncated time wave form (This is the reason why harmonics appear in spectrum).

Figure 9.12. Non- rotating looseness – Type B.

92
9.7.3 Internal Assembly Looseness (Rotating Looseness - Type C)
Internal assembly looseness is normally caused by an improper fit between bearing component
parts which produce many harmonics in the FFT due to the nonlinear response of the loosed parts to
the exciting forces from the rotor. The phase is often unstable and can vary broadly from one
measurement to another. Looseness will often cause sub-harmonic multiples at 0.5X and 1.5X as
shown in Figure 9.13.

Figure 9.13. Internal (rotating) looseness.

9.8 Belt Drive Defects
Worn, loose and mismatched belts are considered the main problem in the rotating machines.
Belt defect frequencies are of the sub-harmonic type. Belt-drive problems, which include shaft
misalignment, pulley misalignment, belt wear, belt resonance, belts too tight, belts too loose, pulley
eccentricity and bent shafts, can be relatively straight forward to detect but can be far more difficult
to specifically diagnose and correct. That is mainly due to the wide variety of problems that can occur
in the installation and assembling of the belt drive, the difficulty of doing field testing on belts, and
the possibility of other influences (i.e. the base) having some effect. With timing belt drives, it is
useful to know that high amplitudes at the timing belt frequency indicate wear or pulley misalignment.
The belt frequency (Figure 9.14) is given by the formula:
𝑃𝑢𝑙𝑙𝑒𝑦 𝑟𝑝𝑚×𝑃𝑢𝑙𝑙𝑒𝑦 𝑃𝑖𝑡𝑐ℎ 𝐷𝑖𝑎𝑚𝑒𝑡𝑒𝑟
𝐵𝑒𝑙𝑡 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 = (4.1)
𝐵𝑒𝑙𝑡 𝐿𝑒𝑛𝑔𝑡ℎ

93
 Figure 9.14. Belt frequency with sub-harmonic belt frequencies.

9.8.1 Pulley Misalignment

There are two types of misalignment as shown in Figure 9.15. These conditions not only result
in destructive vibration but also cause accelerated wear of both the belt and the sheaves. Misalignment
of sheaves produces high vibration at 1X, predominantly in the axial direction. The ratio of amplitudes
of driver to driven rpm depends on the measurement position, relative mass, and the frame stiffness.
With sheave misalignment in fans, the highest axial vibration will be at the fan rpm. When the belt
drives an overhung rotor, which is in an unbalanced condition, it will have to be resolved with phase
analysis.

Figure 9.15. Pulley misalignment.

94
9.8.2 Belt Wear
Worn sheaves or belts may also increase vibration at certain rotational frequencies; however,
belt wear is more often indicated by increased slippage and drive wear. Belt defects appear in the
vibration signature as sub-synchronous peaks, often with harmonics. Figure 9.16 shows a typical
spectral plot (i.e., vibration profile) for a defective belt. If a belt is worn or lose, you will witness a
peak at the belt rate, and harmonics. The highest amplitude peak in the series will often be the twice
the belt rate frequency. The key forcing frequency is known as the "belt rate" or "fundamental belt
pass frequency". It is the rate at which a point on the belt passes a fixed reference point. It is always
less than the speed of either sheave.

Figure 9.16. Belt wear rate with harmonics.

9.8.3 Eccentric Pulleys

Eccentric or unbalanced sheaves cause maximum vibration at 1X of the sheave rotating speed
(see Figure 9.7). To resolve this condition, it may sometimes be possible to balance eccentric sheaves
by attaching washers to taper lock bolts. However, even if balanced, the eccentricity will still induce
vibration and cause fatigue stresses in the belt.

9.8.4 Belt Resonance

Resonance in the belt can cause high amplitudes if the belt's natural frequency should happen
to coincide with either the driving or driven sheave RPM. The natural frequency can be altered by
either changing the length or tension of the belt.
Spectrum: If the belt’s natural frequency coincides with the RPM of either driven component, then
you expect that the 1X peak to increase in amplitude as shown in Figure 9.17. The base of the peak
may be more broad than normal.
95
 Figure 9.17. Belt resonance.

9.9 Gear Defects

All gear sets create a frequency component referred to as gear mesh. The fundamental gear
mesh frequency is equal to the number of gear teeth times the running speed of the shaft. In addition,
all gear sets create a series of sidebands or modulations that are visible on both sides of the primary
gear mesh frequency.
9.9.1 Normal Profile
In a normal gear set, each of the sidebands is spaced by exactly the 1X running speed of the
input shaft, and the entire gear mesh is symmetrical as shown in Figure 9.18. In addition, the
sidebands always occur in pairs, one below and one above the gear mesh frequency and the amplitude
of each pair is identical. If we split the gear-mesh profile for a normal gear by drawing a vertical line
through the actual mesh (i.e., number of teeth times the input shaft speed), the two halves would be
identical. Therefore, any deviation from a symmetrical profile indicates a gear problem; however,
care must be exercised to ensure that the problem is internal to the gears and not induced by outside
influences. External misalignment, abnormal induced loads, and a variety of other outside influences
destroy the symmetry of a gear-mesh profile. For example, a single-reduction gearbox used to
transmit power to a mold-oscillator system on a continuous caster drives two eccentric cams. The
eccentric rotation of these two cams is transmitted directly into the gearbox, creating the appearance
of eccentric meshing of the gears; however, this abnormal induced load actually destroys the spacing
and amplitude of the gearmesh profile. A gearbox is a piece of rotating equipment that can cause the
normal low-frequency harmonics in the vibration spectrum, but also show a lot of activity in the high
frequency region due to gear teeth and bearing impacts. The spectrum of any gearbox shows 1X and

96
2X, along with the gear mesh frequency (GMF). The GMF will have running speed sidebands relative
to the shaft speed to which the gear is attached. Gearbox spectrums contain a range of frequencies
due to the different GMFs and their harmonics. All peaks have low amplitudes and no natural gear
frequencies are excited if the gearbox is still in a good condition. Sidebands around the GMF and its
harmonics are quite common.

Figure 9.18. Sidebands (paired and equal).

9.9.2 Gear Tooth Wear

Since the spacing between the sidebands is erratic and is no longer evenly spaced by the input
shaft speed frequency, the sidebands for a worn gear set tend to occur between the input and output
speeds and are not evenly spaced. An important characteristic of gear tooth wear is that gear natural
frequencies are excited with sidebands around them. These are spaced with the running speed of the
bad gear. The GMF may or may not change in amplitude, although high-amplitude sidebands
surrounding the GMF usually occur when wear is present. Sidebands are a better wear indicator than
the GMF itself (see Figure 9.19).

Figure 9.19. FFT spectrum analysis for gear tooth wear.

97
9.9.3 Gear Tooth Load
As the load on a gearbox increases, the GMF amplitude may also increase. High GMF
amplitudes do not necessarily indicate a problem, particularly if sideband frequencies remain low and
no gear natural frequencies are excited as shown in Figure 9.20. It is advised that vibration analysis
on a gearbox be conducted when the gearbox is transmitting maximum power.

Figure 9.20. FFT spectrum for gear tooth load.

9.9.4 Eccentric Gears and Backlash

If a gear is eccentric (or the shaft is bent), there will be an increase in the amplitude of the
sidebands around the gear mesh 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. In addition,
it is not uncommon to see an increase in the 1XGMF and 3XGMF peaks (see Figure 9.21).
Eccentricity can result in backlash therefore; you may also witness a peak at the natural frequency of
the gear. This peaks (whose frequency is unknown) will also have 1X sidebands.

Figure 9.21. FFT spectrum for eccentric gears and backlash.

98
9.9.5 Misaligned Gears

Gear misalignment always excites second order or higher GMF harmonics, which will have
sidebands spaced with the running speed. It will often show only small amplitudes at 1XGMF, but
much higher levels at 2X (see Figure 9.22). It is important to set the frequency range of the FFT
spectrum more than 3XGMF.

Figure 9.22. FFT spectrum for Gear misalignment

9.9.6 Cracked or Broken Teeth

When the gear has a cracked/broken tooth and when this tooth meshes the mating gear, the
clearance between two meshed teeth increases leading to low amplitude sidebands in the left side of
gear mesh frequency. When the next undamaged teeth are meshed, higher energy impact is generated
leading to higher amplitude sidebands to the right side of gear mesh frequency as shown in Figure
9.23. As a result, the paired sidebands have non symmetrical amplitude due to disproportional
clearance and impact energy.
The cracked/broken gear tooth generates high amplitude at 1X rpm of this gear and excites the
gear natural frequency with sidebands spaced with its running speed. The time wave form is the best
indicator for this fault since the impact shocks appear clearly and periodically repeated every one
revolution of defected gear as shown in Figure 9.24.
It is best
detected in the time domain, which will show a pronounced spike every time the problematic
tooth tries to mesh with teeth on the mating gear. The time between impacts will correspond to
1/speed of the gear with the broken tooth. The amplitude the impact spike in the time waveform
will often be much higher than that of the 1 gear rpm in the FFT spectrum (Figure 7.36).
99
 Figure 9.23. Non symmetrical sidebands profile due to broken tooth.

Figure 9.24. The pattern of cracked/broken tooth

9.9.7 Hunting Tooth Frequency

If each one of two mating gears has a defected tooth, the hunting tooth frequency (HTF) is
generated which is the rate of a tooth in one gear mates with a particular tooth in the other gear. HTF
is located in the low frequency as shown in Figure 9.25. A small defect in one tooth will repeatedly
contact the same teeth in the other gear causing localized wear on those teeth. In practice, the HTF is
used to detect faults on both the gear and pinion that may have occurred during manufacturing or as
a result of mishandling.

100
 Figure 9.25. Hunting tooth frequency

9.9.8 Load Changes

The energy and vibration profiles of gear set change with load. When the gear is fully loaded,
the profiles exhibit the amplitudes discussed previously. When the gear is unloaded, the same profiles
are present, but the amplitude increases dramatically (see Figure 9.26). The reason for this change is
the gear-tooth roughness. In normal practice, the backside of the gear tooth is not finished to the same
smoothness as the power or drive side. Therefore, more looseness is presented on the non-power or
back side of the gear.

Figure 9.26. Unloaded gear has much higher vibration levels

101
9.10 Bearing Defects
Bearings are an important machine element that supports the rotor system. They are designed
to provide less friction at the supports and carry the loads. In machineries, the two most common
types of bearings are used—the antifriction bearing or the rolling element bearing, and the journal
bearing working on the principle of hydrodynamic lubrication. We will focus on fault diagnosis in
rolling element bearing.
Bearing defects are one of the most common faults identified by vibration-monitoring programs
less than 10% run for design lifetime, ~40% fail due to improper lubrication, and 30% due to
misalignment. Although bearings wear out and fail, defects are normally symptoms of other problems
within the machine train or process system. The real problem of bearing elements can be identified
as follow:
Lubrication: Insufficient lubrication.
Wear/cracks/spalls: Damage on the inner or outer race and damage on the rolling
elements. Damaged cages are difficult to detect with vibration analysis.
Cocked bearing: Inner race not true on shaft or outer race not true in housing.
Excessive clearance: Due to excessive wear or poor bearing selection.
Skidding: Rolling elements skidding over inner race when out of the load zone.
Loose on shaft: Inner race sliding on shaft.
Loose in housing: Outer race turning in housing.

Four rotational frequencies are associated with rolling element bearings as shown in Figure
9.27. These components frequencies are fundamental train frequency (FTF), ball spin frequency
(BSF), ball pass frequency outer race (BPFO), and ball pass frequency inner race (BPFI). There is a
large number of fault conditions that can be associated with rolling element bearings such as bearing
wear, lubrication problems, cocked bearing, fluting/EDM, skidding, and looseness. These faults have
a significant effect and the effectiveness level was classified into four stages which will be discussed
in the following sections.

102
 Figure 9.27. Bearing components frequencies

9.10.1 Stage One

The earliest indications of bearing problems including poor lubrication occur in the ultrasonic
frequency (greater than 20 kHz). This is the stage of sub-surface damage which produces friction and
minor impacts It has a significant pattern in the frequency range 5 to 40 kHz as shown in Figure 9.28.
This requires to use ultrasound measurement devices. Traditional spectrum analysis and time
waveform techniques will not detect the fault in this stage. High frequency techniques such as
enveloping, demodulation, PeakVue™, Shock Pulse™, and Spike Energy™ can detect the fault in
Stage One.

Figure 9.28. Stage one: very low amplitude-very high frequency vibration.

103
9.10.2 Stage Two

As the bearing fault develops, techniques such as enveloping, demodulation, and acceleration
spectrum are more successful to use in this stage. The velocity spectrum still ineffective indicator to
detect the faults in this stage. Small defects in the bearing begin to "ring" – they excite the natural
frequencies of the bearing components, which are mainly in the range from 500 Hz to 2 kHz (see
Figure 9.29). These may also be resonances of supporting parts of the bearing. At the end of the
second stage, sidebands around the resonance peaks appear. This stage can be detected using
demodulated high frequency envelope spectra.

Figure 9.29. Stage two: Very low amplitude-high frequency vibration

9.10.3 Stage Three

As the wear increases, bearing defect frequencies and their harmonics occur in addition to
sidebands around these harmonics. This stage can be determined from the vibration velocity spectra.

9.10.3.1 Stage Three: Outer Race

When the bearing fault reaches stage three, the damage is more severe and will be visible if the
bearing is removed. The velocity spectrum can be used to detect the fault in addition to the time
waveform and high frequency techniques (demodulation, enveloping, and acceleration spectra). If the
damage is on the outer race, there will be an impact each time the ball or roller comes into contact
with the damaged area and 1XBPFO and its harmonics appear as shown in Figure 9.30. The amplitude
should be constant, therefore no modulation (or sidebands). If the outer race is rotating, the forces in
the impact will no longer be consistent. When the damaged area travels through the load zone, the
impacts will be stronger compared with the impacts when the damaged area is at the top of the bearing
where it may be lightly loaded. This rise and fall in the vibration amplitude generates sidebands in
the spectrum. Because the outer race would be rotating at the turning speed, the sidebands would be
spaced at 1X as shown in Figure 9.31.

104
Figure 9.30. The pattern of outer race defect in FFT spectrum (case of inner race is rotating).

Figure 9.31. The pattern of outer race defect in FFT spectrum (case of outer race is rotating).

9.10.3.2 Stage Three: Inner Race

If the damage is on the inner race, there will be an impact each time the ball or roller comes into
contact with the damaged area and 1XBPFI and its harmonics appear as shown in Figure 9.32. The
amplitude will be highest when the damaged area is in the load zone, therefore there will be 1X
sidebands.

105
 Figure 9.32. The pattern of inner race defect in FFT spectrum
9.10.3.3 Stage Three: Ball Damage

If the damage is on the balls or rollers, there will be an impact each time the damaged area
comes into contact with the inner or outer race, so the defected ball appears at 1XBSF and its
harmonics as shown in Figure 9.33. The amplitude will be highest when the damaged ball is in the
load zone, therefore there will be FTF (cage) sidebands.

Figure 9.33. The pattern of roller defect in FFT spectrum

9.10.4 Stage Four

When the bearing fault reaches stage four, the bearing has significant damage and should be
replaced. With time the tell-tale harmonics and sidebands may disappear. As the clearance in the
bearing increases due to wear, you will see signs of looseness (1X and its harmonics) as shown in
Figure 9.34. High frequency techniques become less effective as the condition worsens. Overall levels
will increase and the velocity spectrum will show the fault clearly. Spectrum will become very noisy
and the noise floor will lift up ("haystacks” will appear in certain areas).

106
 Figure 9.34. Stage four: outer race fault

9.10.5 Cocked Bearing: on Shaft

A cocked bearing, which is a form of misalignment, will generate considerable axial vibration.
Peaks will often be seen at 1X, 2X, as well as 3X as shown in Figure 9.35. The bearing can be cocked
on the shaft or in the housing. If it is cocked on the shaft, then it will "wobble” as it rotates generating
a rotating vibration. Given that there is such a strong axial vibration, it can be confused with
misalignment, and with unbalance in an overhung pump or fan. The presence of peaks at 2X and 3X
would indicate a cocked bearing condition over unbalance. Relative phase measurement is the key
indicator to differentiate between misalignment and cocked bearing.

Figure 9.35. Cocked bearing on shaft at inner race

9.10.6 Cocked Bearing: in Housing

If the bearing is cocked in the housing, the vibration will more closely mimic misalignment.
There will be two points on either side of the shaft on the face of the component (axial direction)
where the amplitude will be the highest and the vibration will be 180° out-of-phase as displayed in
Figure 9.36.

107
 Figure 9.36. Cocked bearing in the housing

9.10.7 Inner Race Sliding on Shaft

If the bearing is not installed correctly, the inner race may slide on the shaft – i.e. it does not
always turn at the same speed of the shaft. Depending upon the nature of the sliding, the vibration
spectrum may have an elevated 3X and its harmonics (6X, 9X, etc.) as displayed in Figure 9.37. If
the sliding motion is intermittent, i.e. it slides a little then stops, the change in the spectrum may not
be observed. If it is constantly sliding, the vibration pattern should change accordingly.

Figure 9.37. Bearing loose on shaft for inner race

9.10.8 Outer Race Loose in Housing

If the bearing is not installed correctly within the housing, the outer race may move relative to
the housing; it may begin to spin and even "rattle” within the housing. The peaks at 4X running speed
may be elevated as shown in Figure 9.38. If it is possible to view the bearing within the machine (by
removing the cover if allowable), the outer race can be seen with a relative movement to the housing.

108
 Figure 9.38. Bearing loose in housing
9.11 Oil Whirl
Oil whirl is a self-excited fluidic malfunction that typically occurs in plain sleeve bearings. It
may occur for a number of reasons including:
1) Misalignment – bearing unloaded.
2) Excessive clearance
3) Lightly loaded, low damping
Oil whirl generates a characteristic source of vibration in the range of approximately 0.38X to 0.48X
as shown in Figure 9.39.

Figure 9.39. Oil whirl

9.12 Oil Whip
Oil whip is a very destructive condition. This condition can occur when the speed of the machine
is over twice the critical speed (or the first balance rotor resonance); therefore, the critical speed is
close to the oil whirl frequency. The oil whirl excites the resonance and a violent condition arises as
shown in Figure 9.40. Vibration levels are very high and the machine must be stopped or else
catastrophic failure may occur.

109
 Figure 9.40. Oil whip
9.13 Electrical Problems
Vibrations of electrical machines such as motors, generators, and alternators can be either due to
mechanical or electrical issues. Almost of mechanical problems have been discussed in above
sections. Some of electrical defects can be investigated from the vibration spectrum such as:
Cracked/broken rotor bar
Eccentric rotor
Eccentric stator / soft foot
Phase loss
Unequal air gap between stator and rotor

9.13.1 Cracked/Broken Rotor Bars

Cracked/broken rotor bars, shorted end rings or rotor laminations, or loose rotor bar joints
develop a series of harmonics of turning speed with sidebands of the pole pass frequency (PPF) as
displayed in Figure 9.41. Since the PPF has a very low frequency, it is necessary to use a high
resolution spectrum. Also, audible and visible beating is appearing in the time waveform. There is
another pattern can be observed with cracked/broken/loosed rotor bars which is two. Line frequency
(2XLF) sidebands surrounding the rotor bar pass frequency (RBPF).
Hint:
PPF = No of poles × slip frequency
No of poles = 2 (3000 rpm) | 4 (1500 rpm) | 6 (1000 rpm)
Slip frequency = synchronized motor speed – actual speed
RBPF = Number of rotor bars × actual motor speed
110
 Figure 9.41. Broken/cracked/loosed rotor bar

9.13.2 Eccentric Rotor

Eccentric rotors produce a rotating variable air gap between the rotor and stator which induces
pulsating vibrations. The key indicator is the presence of the pole-pass sidebands around 1X and
2XLF as observed in Figure 9.42.

Figure 9.42. Eccentric rotor

9.13.3 Eccentric Stator / Soft Foot

Stator eccentricity produces an uneven stationary air gap between the rotor and stator and may
be occurred due soft foot problem. This produces a very directional source of vibration generating
from two-line frequency (100 or 120 Hz) as presented in Figure 9.43.

111
 Figure 9.43. Eccentric stator
9.14 Cavitation
Cavitation is a phenomenon in which the static pressure of a liquid reduces to below the
liquid's vapor pressure, leading to the formation of small vapor-filled cavities in the liquid. When
subjected to higher pressure, these cavities, called "bubbles" or "voids", collapse and can
generate shock waves that may damage machinery. These shock waves are strong when they are very
close to the imploded bubble, but rapidly weaken as they propagate away from the implosion.
Cavitation normally generates random and high-frequency broadband energy which is superimposed
with the vane pass (VP) frequency and sometimes with its harmonics as shown in Figure 9.44.

Figure 9.44. Pump cavitation

112
10 Chapter (10) Practical Case Studies in Different Industries
10.1 Case Study (1)
Hot Oil Centrifugal Pump (Oil and Gas) - Root Failure Analysis

10.1.1 Equipment Historical Data

In 2008 (WO. 426092) & 2013 (WO.1808304), the pump shaft was locked and all of bearings,
bushings, shaft sleeve, wear rings, and mechanical seals were damaged. In 2016, some vibration
measurements were collected at different flow rate to investigate the effect of process parameters on
the pump performance. It was found higher vibration levels beyond the critical level. The pump was
inspected in workshop (WO. 2452767) to find:
 Wear signs on impeller wear ring and casing wear ring - DE side
 some scratches on the casing wear ring - NDE side
 Impeller outer diameter is 10 mm smaller than its size mentioned in pump data sheet
 Wear signs on bearing house of DE and NDE sides

113
The pump was overhauled and some considerations were recommended during startup and operation
process to avoid liquid turbulence and high vibration levels. The vibration levels still vary between
high and medium levels with respect to mass flow rate variation.
Current and recent vibration measurements (December, 2024) display high vibration levels all the
time and no significant improvement when changing mass flow rate as occurred previously. The next
sections present the vibration measurements and analysis, causes of vibration patterns, special
measurements to identify the root causes, and correction methods.

10.1.2 Vibration Analysis and Diagnostics

The two identical pumps (A & B) were employed in this analysis to investigate the performance
criteria. To ensure all process parameters are similar for two pumps, all pressure gauges were
calibrated and mass flow rate is equalized. After that, the mass flow rate is changed to observe its
effect on vibration levels. Table (1) and Figure (1) display the vibration patterns at different mass
flow rate (1450, 1550, 1650 ton/hr). It was found that the dominant forcing frequency in vibration
spectra is Vane Passing Frequency (VPF). VPF is generated within the machine by pressure variations
as the rotating vane passes the cutwater, causing turbulence. This is inherent to the machine design
and generates a non-uniform disturbance in the liquid flow.
Table (1): Comparison of Vibration levels at different process parameters

Mass Suction Discharge Suction Discharge Overall Vibration levels ISO 10816-3
Flow Pressure Pressure Pressure Pressure Pump (B) VPF
Rate Pump (A) Pump (A) Pump (B) Pump (B) PNDE - Vertical Direction (mm/s)
(Ton/hr) (Bar) (Bar) (Bar) (Bar) (mm/s) Alarm Danger
(mm/s) (mm/s)

1460 1.22 12.1 1.22 10.5 6.62 5.13 4.5 7.1

1550 1.22 12.1 1.22 10.5 7.12 6.06 4.5 7.1

1600 1.22 12.1 1.22 10.5 6.61 5.75 4.5 7.1

114
 Figure (1) Vibration spectra at different process parameters - PNDE - Vertical direction

Although this phenomenon is normal, it is important to be in a limited level to avoid its impact
on pump itself and surrounding equipment. There are some causes leading to increase VPF:
1) Cavitation due to lack of process parameters or pump design
2) Piping forces on suction and/or discharge lines.
3) Impeller looseness and/or wear (to be performed by rotating department)
The first one was checked and adjusted as discussed above. The piping strain (forces) has to be
measured and estimate the effect of these forces on pump performance and alignment. The next
section presents some measurements to check this point.

10.1.3 Piping Strain and Alignment Measurements

These measurements can be performed based on the following sequenced procedures:
1. Measure the offset shift between two connected flanges for two lines. The measurements
will be taken from all radial directions and for complete assembled pump.
2. Take laser shaft alignment measurements for complete assembled pump and record the
readings of offset and angular misalignment.
3. Disassemble the bolts of two flanges between the pump and suction line.
4. Re-take laser shaft alignment measurements and record the readings of offset and angular
misalignment.
115
 5. Disassemble the bolts of two flanges between the pump and discharge line.
6. Re- Measure the offset shift between two connected flanges for two lines by the same
method of point (1).
7. Re-take laser shaft alignment measurements and record the readings of offset and angular
misalignment.
8. Compare and analyze the results

10.1.4 Piping Offset Measurements

The flanges are divided into four locations of measurements with 90o separation angle as shown in
Figure (2). The status of flanges offset was captured in assembled and dismantled cases for suction
and discharge lines. Figure (3) to Figure (6) present the shape of this shift.

Figure (2) A schematic diagram displaying locations of measurement points

116
Location (1) Location (2)

Location (3) Location (4)

Figure (3) Status of two assembled flanges - Suction Line

117
Location (1) Location (2)

Location (3) Location (4)

Figure (4) Status of two assembled flanges - Discharge Line

118
Location (1) Location (2)

Location (3) Location (4)

Figure (5) Status of two disassembled flanges - Suction Line

119
 Location (1) Location (2)

Location (3) Location (4)

Figure (6) Status of two disassembled flanges - Discharge Line

120
Deep and more accurate measurements were performed using feeler gauge as shown in Figure (7) to
identify the values of shift from four mentioned locations on each line. The main objective of these
measurements is to get bolt holes offset and comparing with the values mentioned in API 686. After
dismantling two lines, the gap between flanges was also measured to estimate the vertical forces from
pipes on the pump. All of these measurements are arranged in Table (2). In addition, the internal
distance between two pipes of suction and discharge lines was measured in assembled and
disassembled cases as shown in Figure (8) to investigate the parallesm.

Figure (7) Offset measurement between two flanges using feeler gauge

Table (2) Offset and gap measurements between pump and pipelines flanges
Suction Line Flanges Discharge Line Flanges
Bolt Gap Bolt Gap
Location Assembled Disassembled holes thickness Assembled Disassembled holes thickness
(mm) (mm) offset (mm) (mm) (mm) offset (mm)
(mm) (mm)
1 5.4 -6.75 12.15 10.12 8.1 -1.85 9.95 11.46
2 9 23.5 14.5 11.46 2.55 11.5 8.95 8.45
3 5.85 7.7 1.85 10.39 4.95 4.25 0.7 9.74
4 7.7 4.95 2.75 11.11 7.15 4.55 2.6 10.49

121
 Assembled Case Disassembled Case
Figure (8) Internal distance between two lines in mm

According to API 686: Pipe flange bolt holes shall be lined up with machinery nozzle bolt
holes within 1.5mm maximum offset from the center of the bolt hole to permit insertion of bolts
without applying any external force to the piping. The intent of this requirement is to ensure that
flange bolts can be easily installed without the application of external force (Chapter 6 - Section
4.6.2). Flange face separation shall be within the gasket spacing plus or minus 1.5mm. Only one
gasket per flanged connection shall be used (Chapter 6 - Section 4.6.4).
Based on these instructions, the values of bolt holes offset are out of tolerance for the majority
of measurements and this can be displayed clearly from the plan view as shown in Figure (9). The
gap thickness between flanges is too high which produces high vertical forces on pump.

122
 Figure (9) Amount of bolt holes offset

10.1.5 Shaft Alignment Measurements

Based on the measurements procedures mentioned above, the shaft alignment measurements
were taken three times; Full assembled alignment, after disassembling suction line, and after
disassembling discharge line to investigate the shaft movement due to piping forces. Figure (10) to
Figure (12) display the readings of three measurements, respectively. According to API 686: The
maximum shaft movement in either the vertical or horizontal directions after the flange is tightened
shall be 0.05mm or less. If the shaft movement is more than 0.05mm, the piping flange shall be
loosened from the machinery and corrections made to the piping or supports. Movement greater than
0.05mm is permissible during the tightening procedure (Chapter 6 - Section 4.8.5). It was found that
the pump shaft moved downward in the vertical direction by 0.34 mm when disassembling the suction
line and moved further by 0.51mm when disassembling the discharge line. This means that the values
of shaft movement are out of tolerance and there are high piping forces on the pump.

123
 -0.21 mm 0.02/100 mm -0.09 mm 0.0/100 mm

Figure (10) Cold alignment measurements for assembled flanges

0.13 mm 0.01/100 mm -0.01 mm 0.0/100 mm

Figure (11) Cold alignment measurements after dismantling suction line flanges

0.3 mm 0.04/100 mm -0.05 mm 0.0/100 mm

Figure (12) Cold alignment measurements after dismantling discharge line flanges

124
10.1.6 Methods of Piping Alignment
According to API 686: Moving the pump to achieve piping alignment is not acceptable and shall not
be permitted (Chapter 6 - Section 4.7.1). Bringing the flanges of the pipe into alignment may be
done by a number of means; however, all temporary supports for piping alignment (such as chain
falls and wedges) shall be removed during final alignment readings and piping bolt-up (Chapter 6 -
Section 4.7.2). The methods of piping alignment, which are suitable to the nature of work
environment, include the following:
1. Shim the supports to move the pipes in the direction of alignment (applicable to the equipment
and the correction was displayed in Figure (13)). If the shift is too high, piping support spacers
to be installed.
2. Adjust spring hanger tie-rod turnbuckles
3. After tuning bolt holes to be aligned as soon as possible, smaller bolt diameters are used with
bolt holes of slight offset.
4. Finally, release the base bolts of pump and assemble the suction and discharge lines with a
specified torque wrench. after that, soft foot check on pump foots is applied and corrected
before tightening the pump bolts.

Figure (13) Shifting and shimming of suction and discharge lines

125
10.1.7 Visual Inspection for Interior Components of Pump
The pump was completely dismantled in the workshop and visually inspected regarding any
sign of wears or damage. It was found that both impeller and housing wear rings have a sign of deep
scratches as shown in Figure (14) due to friction between rotating and stationary components. Overall
measurements for clearance and installed components were performed and complete overhaul was
carried out to include the replacement of impeller, impeller wear ring, housing wear ring, and
bearings. In addition, balancing and shaft run out was checked and corrected (for more details, see
the technical report of rotating department).

Figure (14) Signs of wear and scratches in impeller and housing wear rings

10.1.8 Maintenance and Correction Results

The piping alignment was performed for suction and discharge lines to avoid any pipe strains on the
pump and the pattern of correction appears in Figure (15). According to the instructions of pump
manufacturer mentioned in general arrangement drawing (see Figure (16)), the thermal expansion
leading to moving the pump shaft 0.18 mm upward and increasing its length by 0.3 mm. Therefore,
thermal expansion compensation has to be considered during cold alignment by putting the motor
shaft 0.18±0.05 mm upward. In addition, the adjustment of Distance Between Shaft End (DBSE)
includes the variation of shaft length. Figure (17) shows the alignment measurements after
overhauling the pump and before assembling the suction and discharge lines. It displays a readings
comparison for the case before and after pipe alignment. This measurement is used as a reference to
identify the effect of pipe strain after assembling the suction and discharge lines as shown in Figure
(18) and (19), respectively. It was found that the shift in alignment readings due to pipe strain is less
126
than or equal 0.05 mm and this is accepted according to API 686 rules mentioned above. The final
cold alignment shown in Figure (19) achieves the Manufacturer instructions regarding the thermal
expansion and the rules of API 686 (motor shaft offset by 0.18±0.05 mm).

Before correction After correction

Figure (15) Amount of bolt holes offset

Figure (16) Thermal expansion criteria according to Manufacturer instructions

0.03/100
0.3 mm 0.04/100 mm -0.05 mm 0.0/100 mm 0.11 mm -0.01/100 mm 0.03 mm
mm

Before correction After correction

Figure (17) Cold alignment measurements before assembling suction and discharge line flanges

127
 0.14 mm 0.00/100 mm -0.04 mm 0.03/100 mm

Figure (18) Cold alignment measurements after assembling suction line flange

0.15 mm 0.00/100 mm 0.00 mm 0.03/100 mm

Figure (19) Final cold alignment measurements after assembling suction and discharge
line flanges

128
10.2 Case Study (2)
Cooling Fan (Cement Industry) - Severe Fan Bearing Looseness

10.2.1 Analysis and Diagnosis

The overall vibration levels were collected from motor and fan bearings as arranged in Table (3).
There is high internal looseness at FNDE bearing resulting 1X and 0.5X harmonics in spectrum and
envelope as shown in Figure (20) and Figure (21), respectively. This looseness has a significant effect
on the vibration levels of motor bearings. Two fan bearings give the same pattern in spectrum and
envelope and FNDE bearing takes the largest vibration levels.

10.2.2 Recommendations and Actions

The FNDE has to be replaced to avoid the transferred vibration to other components. It was found
that the overall vibration of velocity levels is decreased and slight decrease in acceleration levels.
Peak to peak acceleration in time waveform is the best indicator to detect severe looseness in bearings.
The level decreased from 49g to 5g after replacing the loosed bearing as shown in Figure (22).

129
 Table (3) Overall vibration level according to ISO 10816-3 and ISO 14694
Cooling FAN Value Alarm threshold
Point Unit Current Previous Alarm Danger
11/2/2020 10/26/2020
Param RPM RPM
(19.5Hz) (22.5Hz)
1-MT-Ax
Overall Velocity mm/s 0.886 2.02 2.8 4.5
Overall Acceleration g 0.463 0.704 2 4
1-MT-RH
Overall Velocity mm/s 1.22 3.95 2.8 4.5
Overall Acceleration g 0.306 0.725 2 4
1-MT-RV
Overall Velocity mm/s 1.04 3.47 2.8 4.5
Overall Acceleration g 0.267 0.502 2 4
2-MT-Ax
Overall Velocity mm/s 0.522 2.07 2.8 4.5
Overall Acceleration g 0.351 0.354 2 4
2-MT-RH
Overall Velocity mm/s 1.59 3.28 2.8 4.5
Overall Acceleration g 0.272 0.313 2 4
2-MT-RV
Overall Velocity mm/s 2.26 3.22 2.8 4.5
Overall Acceleration g 0.559 0.458 2 4
3-FN-Ax
Overall Velocity mm/s 0.849 1.74 7.1 9
Overall Acceleration g 0.236 0.608 4 8
3-FN-RH
Overall Velocity mm/s 0.789 3.11 7.1 9
Overall Acceleration g 0.467 1.03 4 8
3-FN-RV
Overall Velocity mm/s 0.644 1.75 7.1 9
Overall Acceleration g 0.293 0.683 4 8
4-FN-Ax
Overall Velocity mm/s 0.725 3.43 7.1 9
Overall Acceleration g 0.367 1.12 4 8
4-FN-RH
Overall Velocity mm/s 1.23 8.25 7.1 9
Overall Acceleration g 0.805 1.86 4 8
4-FN-RV
Overall Velocity mm/s 1.04 5.16 7.1 9
Overall Acceleration g 0.319 0.942 4 8

130
 Figure (20) Frequency spectrum – FNDE – horizontal direction

Figure (21) Envelope spectrum – FNDE – horizontal direction

Figure (22) Time waveform – FNDE – horizontal direction – red (before) – blue (after)
131
10.3 Case Study (3)
Water Pump (Chemical Industry) – Pump Bearing Defect

10.3.1 Analysis and Diagnosis

The vibration measurements were collected from motor and pump bearings to investigate the
equipment integrity. It was found that the bearing defect frequencies appear in the envelope spectrum
of PNDE. These frequencies resulting from 2X ball spin frequency (BSF), fundamental train
frequency (FTF), and BSF sidebands with FTF as shown in Figure (23). In addition, impacts in time
wave resulting from 2X ball spin frequency and cage frequency as displayed in Figure (24).

10.3.2 Recommendations and Actions

The appearance of FTF is the most severity stage of bearing damage. Furthermore, it was
recommended to replace the two pump bearings. Another measurement was collected on 26 th
December, 2020 after pump bearings replacement and it was found a superior improvement since the
bearing defect frequencies and impacts disappeared as displayed in Figure (24).

132
 Figure (23) Envelope spectrum – PNDE – horizontal direction

Figure (24) Time waveform – PNDE – vertical direction – red (before) – blue (after)

133
10.4 Case Study (4)
Dynamic Separator (Cement Industry) - Severe Wear in Impeller

10.4.1 Analysis and Diagnosis

The main function of separator is to divide a given material path into two separated paths using air as
the carrying medium. One path contains only fine particles, and the other only coarse particles. A
distribution plate was used to guide the feed material into the separating space. The mechanical air
separator generates the circulating air inside the separator itself which employed as the classifying
force,
The overall vibration level is not accepted on motor and gearbox according to ISO 10816-3 as
displayed in Table (4). The main source of vibration on motor is 1X running speed (see Figure (25))
and blade pass frequency (BPF) with several harmonics on gearbox (see Figure (26)) in addition to
high impacting in time waveform (see Figure (27)). This observation is the pattern of flow disturbance
around impeller blades and this occurred due to wear or process issue. After visual inspection, it was
found that the blades of separator impeller have severe wear.

134
 Table (4) Overall vibration level according to ISO 10816-3
Dynamic Separator Value Alarm threshold
Point Current
Status Unit 4/4/2021 Alarm Danger
Parameter 2:05 PM
1-MT-Ax
Overall Vibration Velocity Dg mm/s 8.06 2.8 4.5
1-MT-RH
Overall Vibration Velocity pAl mm/s 2.07 2.8 4.5
1-MT-RV
Overall Vibration Velocity Dg mm/s 7.57 2.8 4.5
2-MT-Ax
Overall Vibration Velocity Dg mm/s 7.07 2.8 4.5
2-MT-RH
Overall Vibration Velocity pAl mm/s 1.71 2.8 4.5
2-MT-RV
Overall Vibration Velocity Al mm/s 2.91 2.8 4.5
3-ANG-Ax
Overall Vibration Velocity Al mm/s 3.95 2.8 4.5
3-ANG-RH
Overall Vibration Velocity Dg mm/s 8.17 2.8 4.5
3-ANG-RV
Overall Vibration Velocity Dg mm/s 8.31 2.8 4.5
4-ANG-Ax
Overall Vibration Velocity Al mm/s 3.98 2.8 4.5
4-ANG-R=
Overall Vibration Velocity Dg mm/s 5.02 2.8 4.5
4-ANG-RT
Overall Vibration Velocity Dg mm/s 5.4 2.8 4.5

Figure (25) Frequency spectrum – MNDE – vertical direction

135
Figure (26) Frequency spectrum – Gearbox DE – horizontal direction

Figure (27) Time waveform – Gearbox NDE – radial direction

136
10.5 Case Study (5)
Water Pump (Water Station) – Soft Foot in Base Plate

10.5.1 Analysis and Diagnosis

Firstly, this pump has recorded around 6 mm/s overall vibration level. After that, a base modification
was performed by adding C beam in the middle of base mount as shown in Figure (28). The vibration
reached 13.85 mm/sec after performing this modification (Negative Effect). It was found inclined
anchor bolt(s) as configured in Figure (29) leading to increase the vibration specifically in the axial
direction. Diagnosis may be confused between base plate issue and internal looseness in bearing due
to the pattern of 1X and its harmonics in the spectrum. However, the bearing looseness has severe
harmonics in addition to floor noise in the spectrum.

10.5.2 Recommendations and Actions

Machine base plate was reconstructed and leveled to prevent the soft foot problem. In addition, grout
was added to stiffen the base plate. Superior improvement was achieved as shown in Figure (30) and
Figure (31).

137
Figure (28) Base plate modification – negative effect

Figure (29) Configuration of inclined anchor bolt

138
Figure (30) Frequency spectrum – MNDE – axial direction

Figure (31) Vibration trending – MNDE – axial direction

139
10.6 Case Study (6)
Cooling Tower Fan (Chemical Industry) – Damaged Coupling Element

10.6.1 Analysis and Diagnosis

It was observed an increase in the vibration level on motor due to rotor bar frequency (RBF) in
addition to impacts in time waveform (see Figure (32)). The envelope spectrum includes 1X of
running speed and its harmonics side banded by pole pass frequency (PPF) as shown in Figure (33).
All of these patterns are related to electrical fault issue however, it was found a severe damage in
coupling rubber element (see Figure (34)). Confusion in diagnosis can be occurred in this case
between damaged coupling element and induction motor faults (cracked/broken/loosed rotor bar(s)).

140
 Figure (32) Time waveform – MNDE – axial direction

Figure (33) Envelop spectrum – MNDE – axial direction

Figure (34) Damaged and new coupling rubber element

141
10.7 Case Study (7)
Hot Water Pump (Hospital Utilities) – New Pattern Due to Severe Damage

10.7.1 Analysis and Diagnosis

The integral rubber element of coupling installed between motor and pump has 8 fingers as shown in
Figure (35). If the rubber element suffering a damage, it is expected to see 8X of motor running speed
in the frequency spectrum. Unfortunately, the rubber element has a severe damage leading to broking
4 fingers. The reminding 4 fingers contributes a new pattern of 4X of motor running speed and its
harmonics (see Figure (36)) in addition to periodic impacts in time waveform (see Figure (37)). This
case can be classified as an ambiguous diagnosis case.

142
 Figure (35) Integral and damaged rubber element

Figure (36) Frequency spectrum – PDE – radial direction

Figure (37) Time waveform – PDE – radial direction

143
10.8 Case Study (8)
Cement Mill (Cement Industry) – Damaged Tooth in Girth Gear

10.8.1 Analysis and Diagnosis

This type of rotating equipment has a large gear called girth gear to produce very low rotating speed
to match the process of cement mill. Abnormal sound was detected at certain time while the mill
rotating. This sound is related to damaged tooth appears when contacting the mated gear. The best
tool for detecting broken/damaged gear teeth is the time waveform (TWF). Unfortunately, the default
recorded time of TWF does not include sufficient number of revolutions to detect the impacting force
in the girth gear (see Figure (38)). To improve the ability of diagnosis, TWF was setup to include at
least 4 cycles of mill rotation as shown in Figure (39). It was observed that each time the damaged
tooth contacts the mated gear; it produces a pulse. Therefore, the good setup of vibration signals
facilitates the diagnosis process.

144
Figure (38) Time waveform (T=5.1 sec) – pinion bearing – vertical direction

Figure (39) Time waveform (T=16 sec) – pinion bearing – vertical direction

145
10.9 Case Study (9)
Agitator (Chemical Industry) – Resonance Phenomenon

10.9.1 Analysis and Diagnosis

The agitator works at different operating speeds based on Variable Frequency Drive (VFD) and has
a natural frequency close to 10.5Hz. There is a resonance when the blade passing frequency (BPF)
of 4-blade impeller shown in Figure (40) coincides the natural frequency at which agitator speed
reaches. In this case, vibration levels are very high as shown in Figure (41) due to resonance
phenomenon.
10.9.2 Recommendations and Actions
To avoid the resonance region, all agitator kinematics have to be located ± 20% far away from the
natural frequency. Next measurements were collected at running speed makes BPF far away from the
resonance region. It was found a superior decrease in vibration level as shown in Figure (42). The
main objective of this case study is to take into your account the resonance with variable speed
equipment.
146
 Figure (40) 4-blade Agitator impeller

Figure (41) Frequency spectrum when BPF coincides natural frequency

Figure (42) Frequency spectrum when BPF far away from natural frequency

147
10.10 Case Study (10)
Circulation Multi-Stage Pump (Oil and Gas) – Defected Bearing

10.10.1 Analysis and Diagnosis

Lean amine circulation pump is a multi-stage pump used to increase the pressure from 13.5 bar to 90
bar as a part of gas sweetening process. It suffered a sudden increase in vibration and noise. At the
first instant, the DE and NDE pump bearings indicating a high vibration level due to a pattern of
bearing(s) internal looseness (more harmonics in addition to floor noise as shown in Figure (43) and
Figure (44)). The quick recommendation is to change two pump bearings which leading to lost one
of the main objective of performing predictive maintenance (PdM) to save maintenance cost. One of
the main vibration concepts is the transferring from the defected component to other healthy
components. Furthermore, the DE bearing has a significant pattern for internal looseness and higher
vibration levels compared to NDE bearing.

10.10.2 Recommendations and Actions

The first decision in this case of in between pump is to replace just the DE bearing and keep
monitoring. It was found that the pattern of internal looseness is completely disappeared in NDE
bearing. By this method, we can achieve the features of PdM regarding saving spare parts, reducing
labors, reducing equipment downtime, and increasing productivity by reducing machine downtime.

148
Figure (43) Frequency spectrum – Pump DE – horizontal direction

Figure (44) Frequency spectrum – Pump NDE – horizontal direction

149
10.11 Case Study (11)
Vertical Pump (Chemical Industry) – Looks Like Cantilever Beam

10.11.1 Analysis and Diagnosis

If we investigate the configuration of vertical equipment, we can find that it looks like a cantilever
beam due to its clamped end (base plate side) and free end (Motor side). The maximum deflection of
cantilever beam occurs at the free end and zero defection at the clamped end. The maximum vibration
level was collected at motor NDE and the levels were decreasing gradually till reaching Pump DE
(Pump NDE was not measured due to high temperature issue and needs external amplifier
accelerometer). This means that the variation of vibration levels through the length of vertical
equipment centerline is considered a normal pattern due to the nature of machine construction.
However, we have to monitor from time to time the bearings and base tightening status.

150
10.12 Case Study (12)
Chiller (Chemical Industry) – Early Fault Detection based on Envelope

10.12.1 Analysis and Diagnosis

There is no question that the Envelope / Demodulation is one of the employed tool to diagnose the
faults in rotating equipment specifically in early fault detection. Unfortunately, the incorrect selection
for Filter Central Frequency (FCF) and band pass filter width to construct the Envelope Spectrum
losses important data in the frequency spectrum which are difficult to interpret. The shown case was
performed on compressor in York Chiller. It was found that there is some vibration energy
concentrated in the range of (1500 – 3500) Hz which is so difficult to interpret (see Figure (45)). Two
Envelope setup were implemented; the first one by selecting FCF and band pass filter in the range of
interest (around 2000 Hz) as shown in Figure (46) and the second one was away of any vibration
energy (around 7500 Hz) as shown in Figure (47). Good results were investigated from the former
one and displayed readable peaks. Furthermore, the competitive solution for this issue is to collect
firstly the vibration measurement in the range 0 – 20KHz without collecting Envelope spectrum. After
that, the selection of FCF and band pass filter is based on where the energy is concentrated in the
spectrum.

151
 Figure (45) Frequency spectrum of high frequency range

Figure (46) Envelope spectrum setup at range of interest

Figure (47) Envelope spectrum far away from range of interest

152
10.13 Case Study (13)
Gas Reciprocating Compressor (Oil and Gas) – Normal Pattern

10.13.1 Analysis and Diagnosis

Single acting gas compressor performs compression on the gas either by the head end or the crank
end of the piston to emit one pulse of gas per revolution of the crank shaft. Thus, the frequency of the
pulsations being generated is one times the fundamental frequency (i.e. compressor speed).
Double acting gas compressor uses both the head end and crank end of the piston in compression
process leading to emitting two pulses of gas per revolution (see the internal configuration in Figure
(48)). The frequency of the pulsations produced by the double-acting cylinder is two times the
fundamental frequency.
Based on this configuration, single acting compressors usually have pulsation components that will
be higher at the odd multiples (1X, 3X, 5X, etc.) as displayed in Figure (49). on the other hand, double
acting compressors have pulsation components higher at the even multiples (2X, 4X, 6X, etc.). These
multiples can be monitored by using band or mask alarm tools to detect the variation in their values.
In addition, there are periodic pulses in time waveform as presented in Figure (50). All of these
patterns are considered normal according to the nature of process in reciprocating compressors.
Frequencies up to 2 kHz are associated with mechanical looseness in running gear (connecting rod to
crosshead and crosshead to piston rod). Valve noise and process issues are included above 2 KHz.

153
Another advanced tool, called Kurtosis, can be applied to diagnose this type of rotating equipment.
Kurtosis is a parameter to measure the signal's peakedness since if Kurtosis is high, there are more
peaks in the signal and their amplitudes are greater. Kurtosis value is measured and trended at
different conditions of gas compressor and some statistics are used to identify the acceptable, alarm,
and danger Kurtosis level.

Figure (48) Internal configuration of reciprocating compressor

Figure (49) Frequency spectrum pattern collected from reciprocating compressor

Figure (50) Time waveform pattern collected from reciprocating compressor

154
10.14 Case Study (14)
Water Pump (Chemical Industry)–Motor High Load Due to Misalignment

10.14.1 Analysis and Diagnosis

It's a normal pattern to see rotor bar frequency (RBF) (the number of rotor bars times the RPM) in
the frequency spectrum since each rotor bar passes slight disruptions in the magnetic field due to the
current's path through the rotor and stator. If the number of rotor bars cannot be identified, RBF lies
between 36x and 96x since AC motors normally have 36 - 96 bars based on the size.
A high RBF indicates cracked or broken rotor bars or defective or loose rotor bar joints. This case
study represents a misalignment problem contributing high RBF as shown in Figure (51). The fact
that the misalignment between motor and pump shafts increasing the disruptions of magnetic field
leading to increase RBF. Misalignment fault can be investigated by
1) High 1x and 2x in vibration spectrum
2) High temperature on bearings around coupling
3) Stop the machine and take alignment measurement (See Figure (52))
In case of high RBF, we have to consider at the first instant the mechanical faults such as
misalignment, piping strain, and soft foot. After that, consideration of electrical faults due to electric
motor reliability and its low percentage of fault occurrence probability.

155
Figure (51) Rotor bar frequency before and after alignment correction

Figure (52) Offset and angular laser alignment measurements

156
10.15 Case Study (15)
Blower (Chemical Industry) – Ruptured Belt

10.15.1 Analysis and Diagnosis

As a first stage of spectrum analysis, we need to identify all forcing frequencies in the rotating
equipment to detect the nature of fault(s). One of these frequencies is called Belt Rate Frequency
(BRF) and can be observed in the spectrum due to fault or manufacturing defect in the belt as shown
in Figure (53). The unexpected pattern appears when 2XBRF is dominant as shown in Figure (54)
rather than 1XBRF leading to ambiguous diagnosis since the value of 2XBRF is not considered in
forcing frequencies list. The reason is due to the relation between the belt fault and the existence of
two pulleys since the fault strikes the two pulleys (see Figure (55)) and produces two events in one
belt revolution leading to increase 2XBRF. Based on this physical phenomenon, we have to consider
2XBRF and its harmonics in our analysis.

157
 Figure (53) Ruptured Belt

Figure (54) Frequency spectrum

Figure (55) The kinematics of defected belt

158
11References
1. Blanchard, B., & Fabrycky, W. (2008). Reliability-centered maintenance guide: for facilities and
collateral equipment.
2. Bently, D. E., & Hatch'Charles, T. (2003). Fundamentals of rotating machinery
diagnostics. Mechanical Engineering-CIME, 125(12), 53-54.
3. https://www.ia.omron.com/data_pdf/guide/41/proximity_tg_e_6_2.pdf
4. Industrial vibration sensor selection: Piezovelocity transducers, Wilcoxon Research
5. https://www.fluidlife.com/blog-vibration-data-collection-manual-automated-methods/
6. Anti-Aliasing Filters and Their Usage Explained, NI Engineer Ambitiously.
7. Envelope and Cepstrum Analyses for Machinery Fault Identification, Hans Konstantin-Hansen
and Henrik Herlufsen, Brüel and Kjær, Nærum, Denmark.
8. Hans Konstantin-Hansen, Envelope Analysis for Diagnostics of Local Faults in Rolling Element
Bearings, Brüel&Kjær, Denmark.
9. Detecting Faulty Rolling Element Bearings-Application Note, Brüel&Kjær Vibra
10. Essam Bahgat, MECHANICAL FAULT DIAGNOSIS PART 2, KAU.
11. Training modules - Mobius institute.
12. API 686, Recommended Practices for Machinery Installation and Installation Design.

159