Vibration Analysis Level - 1 Final
Vibration Analysis Level - 1 Final
Vibration Analysis Level - 1 Final
INTRODUCTION OF CVCM
• Unbalance
• Misalignment
• Mechanical looseness
• Structural problems
• Bent shaft
• Bearing faults
Typical Machinery Problems that Can Be Found Using Vibration
Analysis
• Gear faults
• Belt problems
• Lubrication problems
• Electrical motor faults
• Cavitation and turbulence
• others
Overview
of
Condition Monitoring
(Maintenance Philosophies)
Definition of Maintenance
• Vibration measurement
• Electrical testing
• Oil Analysis
• Thickness testing
• Visual inspection
• Collect data
NO
YES
Create Regular Compare Fault Fault
Ref. Meas. limits Diagnostics correction
Input
m/c
specs
Create
New Ref. & Limits
Overview
of
Condition Monitoring
CBM Techniques
What is CBM & Why ?
• To try and maximise the plants production and increase the
mean time between outages many industries are moved
towards a ‘Condition Based Maintenance’ approach.
– Vibration Analysis
– Tribology
– Sonics
– Thermography
Diagnosing a machine is just like a person…
Total
Picture
Overview of Technologies
• Vibration Analysis
– Used to Detect, Analyse and Confirm plant machinery problems. This
can be done in three ways:
• On-line (CSI 2600 & CSI6500) for automated and continuous monitoring
and protection of critical plant items
• Portable (2130 /2140 Analyser) Route based data collection and analysis
• Wireless used for remote monitoring of moving or inaccessible equipment
Overview of Technologies
• Tribology
– Is the analysis of ‘interactive surfaces in relative motion’.
• Lubricants are analysed on-site using the 5300 mini-lab series.
• The results are plotted in a simple to understand tri-vector plot showing the
‘Chemistry’ ‘Contamination’ and ‘Wear’ of each lubricant, this allowing the lubricant
to be changed on condition rather than on a time-based interval.
Wear
Contamination Chemistry
Tribology and Oil Analysis
• Lubricant consolidation
• Extended oil change intervals
• Extended machine life
• Power consumption
Tribology Technology, a 3D Approach to
Industrial Fluid Analysis
Oil Chemistry
and
Viscosity Iron and Water
Tribology Parameters
• Particle counts
• Ferrous Index
• Large ferrous and non Ferrous indication
• Contaminant Index
• Estimated Water and free water droplet indication
• Chemical Index
• Viscosity at 40 degree C
• Percentage change in viscosity
Overview of Technologies
• Sonics
– Through a process known as ‘heterodyning’ Ultrasonic sounds that are
non-audible to human ears are converted back down to a frequency that
is audible to human ears, allowing the operator to hear and recognise
faults developing within plant operating systems, such as:
• Condition Monitoring
• Research and Development
• Quality control and process monitoring
• Non-destructive testing
• Security Systems
• And many others!
Overview of Technologies
• Wear Particle Analysis
• Wear particle analysis is a machine condition analysis technology that is
applied to lubricated equipment.
• It provides an accurate insight an accurate condition of a machine’s
lubricated components by examining particles suspended in the lubricant.
• By trending the size, concentration , shape and composition of particles
contained in a systematically collected oil samples, abnormal wear related
conditions can be identified at an early stage.
• Spectroscopy only recognizes particles up to 6 microns and ignores the
larger abnormal wear particles. For this reason it is a good practice to have
Wear Particle analysis performed on oil samples
Overview of Technologies
Motor Testing
• Motor Current Signature Analysis (MCSA) is a condition monitoring
technique used to diagnose problems in induction motors.
• Concept originates from early 1970s and was first proposed for use in
nuclear power plants for inaccessible motors and motors placed in
hazardous areas.
• It is rapidly gaining acceptance in industry today.
Accurate Results
What Is Vibration?
Upper Limit
Neutral Position
Lower Limit
Vibration in Every Day Life
• Concrete compactors
• Mobile phones
• Massagers
• Component feeders
• Phase. When
Vibration Amplitude
• Peak Amplitude
• Peak-Peak Amplitude
• RMS Amplitude
47
General Description – Measuring Response
• You can also look at vibration as the amount of ‘Time’ it takes to complete a
particular cycle
– If we examine the motion of a forcing function on a fan blade ‘Heavy Spot’ over
a period of time a distinct signature will occur.
• This is known as a
‘Time Waveform’
– Amplitude versus Time
Time Waveforms
• Unfortunately there are multiple sources of forcing functions that can emit
from a machine or component.
– Thus resulting in the time waveform becoming complex in nature
A8 - Example 15
Ex15 -F2V Fan Outboard Vertical
0.4
waveform. 0.2
1.0
EX 8
Los - Example 8
-P2V Pump Outboard Vertical
RPM = 832. (13.86 Hz)
PK(+) = .3263
Analyze
PK(-) =Spectrum
.3572
Acceleration in G-s
RMS = 1.27
0.0
0.8 LOAD = 100.0
0.6
vibration data. -0.2
PK(-) = .3572
0.1 CRESTF= 3.38
Acceleration in G-s
0.6
0.0
-0.1 0.4
-0.2
0.2
-0.3
-0.4
Time: 120.44 0
0 50 100 150 200 250 300 350 Ampl: -.07595 Freq: 736.86
Time in mSecs 0 6000 12000 18000 24000 30000 Ordr: 1.000
Frequency in CPM Spec: .245
Label: Looseness
Fast Fourier Transform – FFT Process
• When a problem starts to develop within a rotating component it will
generate a vibration signature. This signature should be captured in the time
waveform
– Distinguishing that signature can be very difficult when looking at a
time plot
• To understand the problem we need to understand the frequency
– ‘How often is it occurring?’
• The ‘FFT’ is a process that determines the frequency of a signal from a time
waveform.
• The FFT is named after an 18th century mathematician named ‘Jean
Baptise Joseph Fourier’. He established:
– ‘Any periodic signal can be represented as a series of sines and
cosines’.
– Meaning if you take a time waveform and mathematically calculate the
vibration frequency, it can be converted to a more familiar format
Vibration Spectrum
Tim
e
Amplitude
Amplitude
y
e nc
e qu
Tim Fr
e
Fast Fourier Transform (FFT) Spectrum
1x 2X 3X Frequency
54
Fast Fourier Transform (FFT) spectrum (Contd.)
40
-40
0
0 400 800 1.2k 1.6k 2k
[H z]
Ai
2
1 2 2 2
RSS i 1
A1 A2 A3 ...
N BF 1.5
56
Measuring Amplitude and Frequency
• You can measure amplitude from a time waveform as shown:
0 to Peak
+
RMS
Average
Amplitude
Time ‘t’
-
Peak to Peak
• The period ‘t’ is the time required for one revolution of the shaft in this
illustration, which equals one cycle of the waveform
– During this period, the amplitude of the waveform reaches a positive (+) peak,
returns to rest, and reaches a negative (-) peak before returning to rest
Measuring Amplitude and Frequency
• Peak (Pk) – Amplitude measured from the ‘at rest’ position (0) to the
highest value (0 to Peak)
• Peak to Peak (Pk-Pk) – Amplitude measured from the peak positive (+)
value to the peak negative (-) value
• RMS (Root Mean Square) – obtained by averaging the square of the signal
level over a period of time and then taking the square root result
• Average (Avg) – Amplitude value that averages the peak values of the
waveform
0 to Peak
• You can calculate the different RMS
amplitudes when one of the values are Average
known:
Amplitude
– RMS = 0.707 times the peak value Time ‘t’
+
– Avg = 0.637 times the peak value
– Pk-Pk = 2 times the peak value
-
Peak to
Peak
Measuring Amplitude and Frequency
• Severity of a vibration problem can be determined by the amplitude of
the vibration.
• We can measure amplitude in one of three ways
1. Displacement – measures the distance the shaft moves in relation to a
reference point.
2. Velocity – measures the displacement of the shaft in relation to time
3. Acceleration – measures the change in velocity in relation to time
0.25
100 ranges, velocity data
LOAD = 100.0
RPM = 831. (13.85 Hz)
Displacement measures
low frequency events
AccelerationininMicrons
0.20
3
80
Increasing
machine the frequency that
condition
ignoring high frequencies
Displacement
Upper Limit
Peak Peak to Peak
Mean Position
RMS
Lower Limit
Multiply
Number → PEAK –
PEAK RMS
BY PEAK
To Obtain ↓
PEAK-PEAK 1.000 2.000 2.828
PEAK 0.500 1.000 1.414
RMS 0.354 0.707 1.000
61
Displacement, Velocity & Acceleration Conversion
Formulas
METRIC UNITS
V = DF/19,100 D = 19,100V/F
V = 93,640 A/F A = VF/93,640
A = DF2/1,790,000,000 D = 1,790,000,000A/F2
Where,
D = Peak-To-Peak Displacement (Mils Pk-Pk)
V = Peak Velocity (In/Sec-Pk)
A = Peak Acceleration (G’s-Pk)
F = Frequency (CPM)
62
When To Use Displacement, Acceleration & Velocity
Dis
p l ac
em on
en t r at i
c ele
Ac
LOG AMPLITUDE
Velocity
63
Vibration Nomograph
64
Frequency Units
• Frequency refers to how often something occurs:
– How often a shaft rotates?
– How often a rolling element hits a defected race?
• There are three ways to express frequency:
1. CPM – Cycles Per Minute
– 1CPM = 1RPM
2. Hz – Cycles Per Second
– CPM / 60
3. Orders – Multiples of Turning Speed
– Frequency/Turning Speed
Los - Example 3
EX3 -P2V Pump Outboard Vertical
1.0
Analyze Spectrum
15-Nov-95 10:00:16
0.4
• Therefore: 0.2
– When the primary cursors is located on 1Order all the harmonics will be
synchronous 0
Freq: 736.86
– Harmonic cursors can be used to show non-synchronous and sub-
0 6000 12000 18000
Frequency in CPM
24000 30000 Ordr: 1.000
Spec: .245
0.3
0.2
0.1
0
Freq: 2.937
0 20 40 60 80 100 120 140 160 Ordr: 1.000
Frequency in kCPM Spec: .01038
Synchronous Energy
Los - Example 8
EX 8 -P2V Pump Outboard Vertical
1.0
Analyze Spectrum
15-Nov-95 10:00:16
0
Freq: 3888.9
0 6000 12000 18000 24000 30000 Ordr: 10.24
Frequency in CPM Spec: .748
Label: Outer Race Defect
Priority: 1
• Examples of sub-synchronous
energy are:
• Belt Frequencies
• Other Machine Speeds
• Cage Frequencies
Energy in a Spectrum
Synchronous
– N x RPM where N is an integer
Sub-synchronous
– <1 x RPM
Non-synchronous
– F x RPM where F is >1x RPM but not integer
Causes of Sub Synchronous Energy
• Frequencies that show
below the rotational
frequency (Less than 1
Order) are sub synchronous.
– Another component
– Cage frequencies
– Primary belt frequency
– Oil whirl (plain bearings)
Causes of Synchronous Energy
• Frequencies that are equal
too or a direct multiple of
running speed are
Synchronous
• Possible causes of
Synchronous energy are: Los - Example 8
–
EX 8 -P2V Pump Outboard Vertical
Imbalance
1.0
Analyze Spectrum
15-Nov-95 10:00:16
RMS = 1.27
–
0.8 LOAD = 100.0
Looseness
–
0.4
Gears etc
0
Freq: 736.86
0 6000 12000 18000 24000 30000 Ordr: 1.000
Frequency in CPM Spec: .245
Label: Looseness
Causes of Non Synchronous Energy
• Frequencies above (but not
integer multiples of) turning
speed are non synchronous.
• Possible causes of non
synchronous energy are:
– Another component
– Antifriction bearings BF - Example 5
E5 -R4A ROLL BRG. #4 - AXIAL
–
2.0
Electrical 1.8
Route Spectrum
12-Jul-96 17:16:42
–
1.6 RMS = 2.69
1.0
0.8
0.6
0.4
0.2
0
Freq: 3888.9
0 6000 12000 18000 24000 30000 Ordr: 10.24
Frequency in CPM Spec: .748
Label: Outer Race Defect
Priority: 1
Lines of Resolution
(LOR)
Lines of Resolution
• Lines of Resolution (LOR) determine how clear the peaks(data) are
defined within our spectrum.
• The more lines we have over the same F-max (Maximum frequency
scale). The more accurate our data will be
• Example.
– The diagram below shows data that has been collected using 400 LOR.
Notice how the top of the peaks are capped. When the LOR are
increased the data becomes more accurate.
Lines of Resolution
L2 - TA 16
TA16 -M1H Motor Outboard Horizontal
0.20
0.5
Analyze Spectrum
13-Mar-01 09:14:16
09:13:53
PK = .3852
.7078 • The spectrum shown displays
0.16
0.4 LOAD = 100.0
RPM = 1497.
1496.
RPS = 24.95
24.94
data at 800 L.O.R with an
Fmax of 1600 Hz
PK Acceleration in G-s
PK Acceleration in G-s
0.12
0.3
0.08
0.2 The second spectrum
displays the same data but
0.04
0.1 with 3200 L.O.R over the
same Fmax
0
0 400 800 1200 1600
Frequency in Hz
Lines of Resolution
• The range of LOR settings that we can choose from on the analyzer
starts at 100 Lines and go up to 12800 Lines.
Sub Synchronous
Amplitudes
1.2
Resolution
Velocity
1.0
Clarity of the spectral
0.8 data
RMS
Acceleration,
0.6
0.4
0.2
0
Freq: 3888.9
0 6000 12000 18000 24000 30000 Ordr: 10.24
FrequencyFrequency in CPM Spec: .748
Label: Outer Race Defect
Priority: 1
Hz (CPS), CPM, Orders
Phase
• The position of a vibrating part at a given instant with
reference to a fixed point or another vibrating part
83
Phase (Contd.)
1 2
86
Introduction to Vibration Analysis
(Data Collection)
Transducers and Mounting Techniques
• Although there are many different types of transducers
available, the most common type used for day to day data
collection are Accelerometers.
• These transducers provide an electrical charge proportional to
acceleration by stressing piezoelectric crystals typically
100mV/g sensors are used.
Transducers for Vibration Detection
• Accelerometers; Measures Acceleration
Advantages Disadvantages
• Measures displacement • Value of displacement is small at high
• Non contacting frequencies
• Variation in geometric irregularities &
• No moving parts
magnetic properties of shaft give
• No wear incorrect results
• Permanent installation
Velocity Pickup
Advantages Disadvantages
• Measure Velocity • Moving parts prone to wear
• Self Generating • Large size
• Low Impedance • Sensitive to orientation
• High lower limiting frequency (> app. 10
hz)
Piezoelectric Accelerometer
Advantages
• Rugged Disadvantages
• High impedance output
• Very large dynamic range
• No true DC response
• Wide frequency range
• Compact, often low weight
• High stability
• Can be mounted with any
orientation
Importance of Accelerometer Mounting
• Adhesive Mounting
• Magnetic Mounting
• Extension Probe
Four Different Methods of Accelerometer Mounting
Hand-held Mount
Magnet Mount
Stud Mount
Adhesive Mount
94
Handle the Accelerometer Carefully
95
Data Quality
• Whether it is your job to collect the data and/or analyse the
data it is important to understand that the technologies will
not give you the answer to a machines problem unless you
have collected meaningful, quality data
A1 - Recirculation Pump #5
RCP#5 -M2H MOTOR INBOARD BRG. - HORIZONTAL
0.025 Trend Display
FAULT 36-65xTS
PK In/Sec
0.020
0.015
0.010
ALERT
0.005
• 0
0 The
Eachmachines
technologydue
100 to beand
is stored
200 monitored are defined
analysed from
300 within the
a single software
400 platform,
500
software.
Days: 11-Aug-95 To 11-Dec-96
0.05
allowing the analyst to: Route Spectrum
11-Dec-96 17:33:57
PK In/Sec
0.04
OVERALL= .0604 V-DG
–
1.0 Route Waveform
Collaborate all data into one single report.
Acc in G-s
• These are older instruments used for vibration analysis not commonly
used in these days.
• In these analyzers the frequency is tunable for range of interest.
• Selector switches are provided for selecting the desired parameter of
vibration amplitude, full-scale amplitude range and filter "bandwidth"
characteristics.
• A hard copy print out of the spectrum can be obtained directly from the
analyzer
• Not able to store data and producing better resolved frequency spectra
• Not have the compatibility with predictive maintenance software.
• Outdated these days and replaced by FFT Analyzers
Swept Filter Analyzer
FFT Data Collectors
110
Vibration Analysis & Diagnostics
111
Vibration Analysis
112
Define The Problem
114
Determine Machine Details
• Types of bearing
• Type of coupling
• Leaking seals
• Obtain the data on both drive end and non drive end of each of the
machine component in all the three directions i.e.
– Horizontal
– Vertical
– Axial
• Define frequency and amplitude parameters
• Identify the problem component based on frequency and amplitude
• Reducing the list of possible problems based on frequency
• Consult time waveform to confirm the problem
117
Fault Diagnostics
IF - Example 2
Ex2 -F1H Fan Inboard Horizontal
6
Route Spectrum
16-Sep-99 08:36:29
5
• What should the OVRALL= 4.58 V-DG
RMS = 4.56
LOAD = 100.0
4
waveform show? RPM = 3000.
RPS = 50.00
RMS Velocity in mm/Sec
0
Freq: 3000.0
0 20000 40000 60000 80000 Ordr: 1.000
Frequency in CPM Spec: 4.539
Imbalance Waveform Data
• Despite the waveform being displayed in Acceleration
– Default unit for route based waveform data
• There is still a predominant sinusoidal waveform pattern
– 1 x Revolution sine wave
IF - Example 2
Ex2 -F1H Fan Inboard Horizontal
1.0
Waveform Display
02-Feb-00 15:13:51
0.8
PK = .5289
0.6 LOAD = 100.0
RPM = 2985.
RPS = 49.76
0.4
PK(+) = .8332
PK(-) = .8893
0.2
CRESTF= 2.38
Acceleration in G-s
-0.0
-0.2
-0.4
-0.6
-0.8
-- Baseline --
Value: 3.063
10
Date: 07-Apr-00
RMS Velocity in mm/Sec
8
FAULT
0
Date: 21-May-01
0 100 200 300 400 500 Time: 14:24:29
Days: 07-Apr-00 To 21-May-01 Ampl: 11.21
Imbalance Problem - Practical
• The following fan unit has an imbalance present on the rotor.
– 1xTs Peak in the Spectrum
– 1xTs Peak in the Waveform
Imbalance
IF - Example 2
Ex2 - F1H Fan Inboard Horizontal
6
Route Spectrum
16- Sep- 99 08: 36: 29
R M S V e lo c it y in m m /S e c
5 OVRALL= 4.58 V- DG
RM S = 4.56
LOAD = 100.0
RPM = 3000.
RPS = 50.00
4
0
Freq: 3000.0
0 20000 40000 60000 80000 Ordr: 1.000
Frequency in CPM Spec: 4.539
IF - Example 2
Ex2 -F 1H Fan Inboard Horizontal
1.0
Waveform Display
02-Feb-00 15: 13: 51
0.8
PK = .5289
0.6 LOAD = 100.0
A c c e le r a t io n in G - s
RPM = 2985.
RPS = 49.76
0.4
PK(+ ) = .8332
PK(-) = .8893
0.2
CR ESTF = 2.38
-0.0
-0.2
-0.4
-0.6
-0.8
-1.0
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Revolution Number
0
Freq: 1246.3
0 20000 40000 60000 80000 100000 Ordr: 1.000
Frequency in CPM Spec: 4.673
Imbalance Case Study 1
• The waveform from this data shown on the following page
represents a sinusoidal waveform clearly shown once per
revolution of the shaft – here the waveform is displayed in
velocity.
Bm/c - TOP RECIRC FAN
TRF B m/c -F2H Fan Outboard Horizontal
8
RMS = 4.66
problem. The 0
PK(+) = 7.03
PK(-) = 7.40
Velocity in mm/Sec
CRESTF= 1.59
amplitudes should be
checked in both radial -4
directions to confirm -8
this problem
-12
0 1 2 3 4 5
Revolution Number
Imbalance Case Study 1
• The plot shown indicates a multi spectral plot showing all the
radial directions.
Bm/c - TOP RECIRC FAN
TRF B m/c - Multiple Points (08-Nov-04)
Max Amp
4.27
Plot
Scale
TRF B m/c -F2V
5
RMS Velocity in mm/Sec
amplitudes are
common to both 0 TRF B m/c -F1V
bearing – Inboard
and Outboard in all 0 8000 16000
TRF B m/c -F1H
24000
recommended to be 2
FAULT
available opportunity
and for it to be re- 0
0 300 600 900
Days: 24-Nov-00 To 08-Nov-04
1200 1500
Date: 08-Nov-04
Time: 14:16:38
Ampl: 4.688
tested afterwards
Imbalance Summary
Offset – Where the shaft centrelines are parallel but they do not meet
producing a radial 2xTs peak
0
Freq: 2925.0
0 10000 20000 30000 40000 50000 Ordr: .983
Frequency in CPM Spec: 2.046
Angular Misalignment Spectral Data
• The spectral data below represents a simple misalignment
plot.
– The primary cursor denotes the 1xTs peak while the data was taken
in the axial direction. This type of data is common to that of Angular
Misalignment
B29 - PUMP NO 3
3601PUM003-M2A Motor Inboard Axial
8
Route Spectrum
04-Aug-04 08:49:05
7
OVERALL= 6.33 V-DG
RMS = 6.31
LOAD = 100.0
6
RPM = 1071. (17.84 Hz)
RMS Velocity in mm/Sec
0
Freq: 1.071
0 30 60 90 120 Ordr: 1.000
Frequency in kCPM Spec: 5.966
Offset Misalignment Waveform Data
• The waveform above is showing two clear peaks per
revolution of the shaft. This type of waveform resembling an
‘M’ or ‘W’ shape is common to offset misalignment.
– Data shown in velocity
PK(+) = 30.66
Velocity in mm/Sec
PK(-) = 26.81
10
CRESTF= 1.82
-10
-20
-30
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Revolution Number
Misalignment Waveform
• The waveform data shown above is predominantly showing
one sinusoidal waveform per revolution of the shaft.
– Here the data is shown Acceleration
B29 - PUMP NO 3
3601PUM003-M2A Motor Inboard Axial
0.8
Route Waveform
04-Aug-04 08:49:05
0.6
PK = .2596
LOAD = 100.0
RPM = 1071. (17.84 Hz)
0.4
PK(+) = .6277
PK(-) = .5683
0.2 CRESTF= 3.42
Acceleration in G-s
0.0
-0.2
-0.4
-0.6
-0.8
Rev : .680
0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 Ampl: -.306
Revolution Number
Case Study 3 – Kiln Main Motor Gearbox
• Introduction
• The Kiln drive gearbox motor had been replaced during a
planned plant shutdown.
• During the start up of the plant after the shutdown it was
noted that the motor and gearbox were excessively noisy.
Vibration data was taken during the run up of the plant to
determine the cause of the problem.
Main Motor
Kiln Gearbox
Case Study 3 – Kiln Main Motor Gearbox
• The spectral plot shown above is the data taken from the drive
end of the motor. Here there is a dominant 2xTs peak.
04 - Kiln Drive
0804 -M2H Motor Inboard Horizontal
2.4
Route Spectrum
In addition to the misalignment the excessive forces 2.1
29-Mar-01 11:33:43
0
29-Mar-01
0.9 09:40: 20
After 29-Mar-01
09:40: 09
0.3 23-Jan-01
15:02: 00
0 Before 25-Oct-00
09:04: 17
Freq: 1418.3
0 20000 Shutdown 40000 08-Aug-00
60000 Ordr: 1.000
Frequency in CPM 14:06: 56 Spec: 1.346
0 60 120 180 240 300
Frequency in kCPM
Case Study 3 – Kiln Main Motor Gearbox
• During data collection it was also observed that the grouting
around the front feet of the motor had begun to crack as a
result of the excessive force being applied to the motor base
and feet due to the misalignment.
• Conclusion
– It was confirmed the engineers that replaced the motor during the
shutdown and assumed as the motor was a like for like swap, as long as
they kept the shims in the correct place then alignment was not
necessary.
– Corrective action was required and production was stopped so the
motor could be re-aligned and the mountings re-secured.
Misalignment Summary
2.5
RMS Velocity in mm/Sec
2.0
1.5
1.0
0.5
0
Freq: 10.58
0 200 400 600 800 1000 Ordr: 1.000
Frequency in Hz Spec: 3.088
Looseness Spectral Data (Component)
• The spectral plot shown is demonstrating rotational
Looseness.
– The primary cursor is on 5xTs peak
• The 5 Order peak is vane pass frequency (5 vanes on the impeller)
– Multiple harmonics of 5xTs are shown indicating the impeller has
come loose. Ex 9
L1 - Example 9
-P2A Pump Outboard Axial
1.5
Route Spectrum*
17-Aug-01 08:52:02
1.2
The raised noise level around
OVERALL= 6.62 V-DG
RMS = 6.13
common to a pumping
RMS Velocity in mm/Sec
0.9
0
Freq: 14.88
0 40 80 120 160 200 240 Ordr: 5.002
Frequency in kCPM Spec: .742
Label: Centrifugal Pump - Medium
Looseness Waveform Data
• Here the waveform is demonstrating a lot of energy and
appears to be more random and non-periodic.
– Displaying the waveform in velocity may help to show the random
non-periodic pattern.
40 - Kiln Main Drive
M4441 -G2H Shaft 01 Outboard Horizontal
1.2
Route Waveform
06-Nov-02 11:02:11
PK(+) = .9797
0.4 PK(-) = .9874
CRESTF= 3.11
Acceleration in G-s
0.0
-0.4
-0.8
-1.2
0 50 100 150 200 250 300 350 400
Time in mSecs
Looseness Trend Data
• Here the trend plot is showing the parameter labelled as the 3-
15xTs. This is measuring the amount of energy from 3 orders to
15 orders, which is where the harmonics of looseness will
appear.
-- Baseline --
6
Value: .837
Date: 28-Feb-02
RMS Velocity in mm/Sec
FAULT
4
ALERT
3
0
0 10 20 30 40 50
Days: 28-Feb-02 To 16-Apr-02
Case Study 4 – Reciprocator Fan
• Introduction
• Data had been collected on the following fan for several
months as part of a routine periodic vibration routine. During a
routine visit to the machine it was observed that there was a
lot of low frequency activity showing around the bearing on
the inboard of the fan (F1H)
Case Study 4 – Reciprocator Fan
• The multiple plots shown above indicate the change over time
from the data taken on F1H.
– It is quite apparent that the data shown here is indicating multiple
harmonics of the 1xTs frequency (the rise energy as you move further
away from the 1xTs).
– This type of data is common to that of a looseness problem.
40 - Precip Fan
M2237 -F1H Fan Inboard Horizontal
0
29-Oct-02
11:00:02
18-Sep-02
09:13:26
29-Aug-02
15:30:18
29-Oct-02
11:00:02
RPM= 998.9
22-Aug-02
11:14:48
Freq: 16.65
0 300 600 900 1200 Ordr: 1.000
Frequency in Hz Sp 4: 2.811
Case Study 4 – Reciprocator Fan
• The waveform data taken for this particular point is not
showing a random type of waveform pattern which you
would expect from Structural looseness, but there is a more
a repeatable (timed interval) pattern.
40 - Precip Fan
M2237 -F1H Fan Inboard Horizontal
3
Analyze Waveform
18-Sep-02 09:24:16
2 RMS = .3747
LOAD = 100.0
RPM = 998. (16.63 Hz)
1
PK(+) = 2.36
PK(-) = 2.83
CRESTF= 7.55
Acceleration in G-s
-1
-2
-3
-4
0 100 200 300 400 500 600 700 800
Time in mSecs
Case Study 4 – Reciprocator Fan
• This type of waveform would more be indicating Component
looseness and may indicate a problem with a loose bearing.
• Conclusion
• It was recommended that the bearing should be inspected at
the next available opportunity.
– Upon inspection it was found that the bearing was a ‘Taper-Lock’
bearing and the taper lock was loose, thus resulting in excessive
clearance between the bearing and the rotor.
Looseness Summary
Fault Diagnostics
Gears, Bearings, Peakvue, Electrical, Belts,
Resonance
Gear Defects
• There are many different types of gears and gear combinations
available for various speed and power requirements.
• Regardless of gear type they all produce the same basic
vibration patterns and characteristics when a defect is present
0.3
0
Freq: 386.98
0 200 400 600 800 1000 Ordr: 15.00
Frequency in Hz Spec: .864
Calculating GMF – Single Reduction
• Single Reduction Gear Train
– The GMF is simply defined as the number of teeth on a gear multiplied
by its turning speed
GMF = (#teeth) x (Turning speed)
• Example:
– Consider the following gear train,
Gear 1= 44 Teeth
OUTPUT
Gear 2= 71 Teeth
Input = 1490RPM
INPUT
Gear 1 = 15 teeth
Gear 2 = 21 teeth
Gear 3 = 19 teeth
Gear 4 = 54 teeth
OUTPUT
Calculating GMF – Multi Reduction
Input = 1490RPM
INPUT
Gear 1 = 15 teeth
Gear 2 = 21 teeth
Gear 3 = 19 teeth
Gear 4 = 54 teeth
OUTPUT
Gear 3 = 10 teeth
Gear 4 = 20 teeth
OUTPUT
0.6
0.4
Sidebands
0.2
0
Freq: 18363.
0 8000 16000 24000 Ordr: 59.23
Frequencyin CPM Spec: .564
Dfrq: 310.82
Gears – Waveform Data
• Gears can produce different types of waveforms, the one
shown below is indicating gear wear.
– As the defective teeth come into mesh the noise generated increases
showing an increase in amplitude in the vibration data
FPP - SAND MILLS (OLD)A
X401A -G3A Shaft 02 Inboard Axial
1.5
Route Waveform
07-Nov-02 09:11:53
1.2
PK = .4580
0.9 LOAD = 100.0
RPM = 311. (5.19 Hz)
PK(-) = 1.13
CRESTF= 3.91
0.3
-0.3
-0.6
-0.9
-1.2
-1.5
0 1 2 3 4 5 6
Revolution Number
Case Study 5 –Gearbox
• The following case study is from a motor gearbox unit that
drives a roller.
– Product (Fibre) is fed along the top of the roll while being washed
through a series of baths.
– There are several of these Wash Nip Rollers in a continuous stream,
failure of any one of them results in lost production
• Data is collected on a
fortnightly basis as part of
a routine data collection
route
Case Study 5 –Gearbox
• The spectral data shown below is taken from the motor in the
axial direction
– (As the motor is mounted directly into the gearbox the first helical gear
is mounted on the end of the motor shaft).
L1NG - WASH LINE NIP UNIT 3
3-32J03 -MIA MOTOR INBOARD AXIAL
0.6
Route Spectrum
0
Ordr: 49.00
0 10 20 30 40 50 60 70 80 Freq: 57551.
Frequency in Orders Spec: .275
Dord: .00649
Case Study 5 –Gearbox
• The waveform data is showing a distinct pattern commonly
associated with gears.
• The amplitude increases In noise as the damaged teeth come
into mesh
– Producing over 2G-s of force in both the positive and negative direction
Case Study 5 –Gearbox
• The gears were inspected due to the critical nature of the
asset. It was found the gear to be severely damaged.
• A new gearbox was fitted and new data was taken showing the
difference between the good and bad gear
Bearing Defects
Rolling Element
Plain Bearings
Peakvue
Rolling Element Bearings
• Rolling element bearings have specific bearing failure modes
that can be observed in the spectral and waveform data.
Outer Race
Inner Race
How Bearing Faults Generate Vibration
How Bearing Faults Generate Vibration
Rolling Element Bearings
• Bearing defect frequencies are calculated based upon the geometry of the
bearing these calculations may include:
– Number of rolling elements
– Pitch Circle Diameter
– Rolling element diameter
– Contact angle
• Defined within Machinery Health Manager there are over 100000 predefined bearing
stored in the CSI bearing warehouse
• Bearing
FTF & BSF
BPFI & BPFO
Rolling Element Bearings - Waveform
• As a bearing becomes defected then the amount of
noise/force generated as the rolling elements impact the
defective area increases.
– This can show significant G-levels in the time waveform. This value is
trended in the software as the Peak-Peak value
• Conclusion
• The motor was reported as having a bearing
defect to the engineering group. As the
fundamental defect frequency was present
and the trend had shown sudden increases it
was recommended to change the bearing at
the next available opportunity.
Bearing Defects
Rolling Element
Plain Bearings
Peakvue
Plain Bearings
• Rotating elements are not used in sleeve (plain) bearings;
rather the shaft rides on a layer of lubricating oil inside the
bearing journal.
– Therefore the fundamental frequencies seen from antifriction bearings
do not apply to sleeve bearings.
will be.
– A good sleeve bearing will still show
8
a few harmonics as there is a small
clearance between the shaft and
bearing
4
0
Ordr: 1.000
0 3 6 9 12 Freq: 15.68
Frequency in Orders Spec: 7.494
Plain Bearings – Spectral Diagnostics
• Oil Whirl
– One of the major problems encountered with these types of bearings is
the possibility of hydraulic instability of the shaft within the bearing;
known as oil whirl or oil whip.
– Oil Whirl is a result of turbulent flow within the oil resulting in the oil
pushing the shaft around of centre.
Fu - Turbine Brg Thrust End • The dominant peak within the spectral
TBT -R1Y Radial 'Y' Direction
16
Route Spectrum* data will be typically at 0.4 orders.
27-Jul-04 14:08:21
(.40-.48)
Oil Whirl at 0.4 orders OVERALL= 2.93 V-DG
P-P = 22.71 – This defect is sub-synchronous data.
LOAD = 100.0
12
RPM= 941. (15.69 Hz) – When the amplitude of the oil whirl is
equal to or greater than the 1xTs peak a
P-P Displacement in Microns
problem exists
8
• In this instance oil whirl can be
corrected by:
– Properly loading the bearing
4
– Change the oil viscosity
– Change the oil pressure
0
Ordr: 1.000
0 3 6 9 12 Freq: 15.68
Frequency in Orders Spec: 7.494
Oil Whirl
Bearing Defects
Rolling Element
Plain Bearings
Peakvue
Peakvue Processing
• The detection of bearing and gear defects is one of the primary
expectations of a predictive maintenance program.
– As analysts we can spend a lot of time tying to determine these faults.
– Peakvue is a process that concentrates on these defects to help the analysts
determine potential faults developing
• Peakvue stands for the Peak Value and is a technique that detects high
frequency stress waves generated from metal to metal contact, such as:
– Bearing defects – Rotating elements striking a defect on the race
– Gear defects – Damaged teeth in mesh
– It is the detection of these high frequency stress waves that will aid with
analysis
Peakvue Processing - Filters
• In order to capture the stress wave signal the process requires
the use of a filter to remove all unwanted noise that can
dominate the data
Noise removed by
filter
Peakvue Processing – Waveform Data
• As stress waves are small in amplitude severity of the problem
can be judged using the time waveform
– Peak Value of force from the impact
• The waveform can resemble a spectrum as there is no
negative half to the data
B42 - ZONE 5 DF FAN 1
16/16EXT01-M2P Motor Inboard Horz Peakvue
0.8
0.7 Route Spectrum
N N N N N N N N N
RMS Acceleration in G-s
09-Jul-03 09:50:49
0.6 (PkVue-HP 1000 Hz)
OVERALL= 1.37 A-DG
For Peakvue analysis
0.5
0.4
RMS = 1.37
LOAD = 100.0
0.3 RPM = 1342. (22.37 Hz)
0.2
>NTN 6217
N=BPFO -OB
Use the Spectrum
0.1
0 – Diagnose the defect
0 200 400 600 800 1000
Frequency in Hz Use the Waveform
8 Route Waveform
7
09-Jul-03 09:50:49
(PkVue-HP 1000 Hz)
– Determine the severity
Acceleration in G-s
6 RMS = 2.97
5 PK(+) = 8.35
4 CRESTF= 2.81
3
2
1
0
Freq: 1.250
0 4 8 12 16 20 24 28 32 36 Ordr: .05587
Revolution Number Spec: .01367
Label: Bearing Fault - BPFO NTN6217
Case Study 7 – Peakvue on Fan Bearing
• The following machine is a
pre-heater fan designed to
heat the product prior to it
entering a Kiln
– There is no standby for this
machine
– Failure results in stopped
production
• The following data was taken from the above fan unit.
– The problem bearing resided on the fan inboard bearing.
– Data was collected on a monthly basis. Both conventional vibration
data and Peakvue data were taken during the route collection.
Case Study 7 – Peakvue on Fan Bearing
• The data shown below is taken using conventional vibration
methods on the inboard bearing of the fan
– 1x peak is highlighted showing amplitudes of 4mm/sec
– Waveform is showing less than 1G of force both +/-
40 - Preheater Fan
M4425 -F1H Fan Inboard Horizontal
5
Route Spectrum
4 29-Oct-02 11:19:26
OVERALL= 4.18 V-DG
RMSVelocity in mm/Sec
RMS = 4.18
3
LOAD = 100.0
RPM = 825. (13.75 Hz)
2
CRESTF= 3.11
0
-0.5
-1.0
-1.5
Ordr: 1.000
0 1 2 3 4 5 Freq: 13.75
Revolution Number Spec: 3.721
Case Study 7 – Peakvue on Fan Bearing
• The Peakvue data above is taken from the same point as the
previous data.
– This particular reading is using a 1000 Hz High Pass filter.
40 - Preheater Fan
M4425 -F1P Fan Inboard Horz Peakvue
0.7
Route Spectrum
0.6 F F F F
0.5
29-Oct-02 11:15:59
(PkVue-HP 1000 Hz) • Here the data is showing there
RMSAcceleration in G-s
5
4
PK(+) = 7.47 as oppose to the 1G from the
3
CRESTF= 2.25
DCoff = -3.08
previous data.
2
1
0
Ordr: 8.176
0 10 20 30 40 50 Freq: 113.14
Revolution Number Spec: .194
Case Study 7 – Peakvue on Fan Bearing
• Conclusion
• There is significant bearing damage relating the outer race of
the bearing.
• As the machine was critical to the process, the bearing was
changed on the next available opportunity that tied in with
process requirements.
Electrical Defects
Electrical Defects
• A motor can be simply broken down into two key components
– Rotor
– Stator • The stator is stationary
– Consists of wire wound in coils
and placed in slots of an iron
core.
– The stator produces a rotating
magnetic field.
0.4
RPM= 2967. (49.44 Hz)
showing a peak at 100Hz
(6000cpm)
RMS Velocity in mm/Sec
0.3
– 2xLf
0.2
– This can be mistaken for
misalignment
0.1
0
Freq: 100.00
0 500 1000 1500 2000 Ordr: 2.023
Frequency in Hz Spec: .386
Electrical Defects – Waveform Data
• The waveform data from a 100Hz peak will show a sinusoidal
pattern like the waveform shown below
Ex7 - Example 7
0
produce higher force (Higher
-0.5 waveform levels) than those
from electrical defects due to
-1.0
the stress being applied to
-1.5 the machine
-2.0
0 1 2 3 4 5 6
Revolution Number
Electrical Defects - Causes
• Common fault types that can produce the 2xLf peak are as
follows:
• Dynamic Eccentricity – Usually Rotor Related
• Static Eccentricity – Usually Stator Related
• Loose Iron or Slot Defect – Rotor or Stator
• Open or Shorted Windings
• Insulation Breakdown or Imbalanced Phase
• Loose Connectors
Electrical Defects - Peakvue
• Peakvue data also shows electrical defects at the 2xLf peak.
– This may be due to the rotor or stator bowing; due to heat build up.
• The spectral plot below is indicating a 100Hz peak using
Peakvue with a 1000Hz filter.
Case Study – Electrical Defect
• The following case study was taken from a glass manufacturer.
The data was from the ‘Electric Front Wall Cooling Fan’.
– This fan unit is a critical fan to the process and has no standby unit.
– In this particular instance the motor failed shortly after the data was
collected.
• Conclusion
– As the motor failed shortly after
data collection no action was
taken to prevent failure.
– The investigation in the motor
showed one of the connectors had
come loose causing the motor to
burn out.
Belt Defects
V-Belts
Timing Belts
Belt Defects
• Belts are the most common low cost way to transmit power
from one shaft to another.
– Belt drives rely on friction between the belt and pulley to transmit
power between drive and driven shafts
• The ability of belt to transmit power depends upon
1. Belt Tension (tension on the belt holds it tightly against the sheave)
2. Friction between the belt and sheave
3. The arc of contact between the belt and sheave (Wrap)
4. The speed of the belt
Offset Misalignment
Angular Misalignment
>Belt Freqs
J=Belt 1 Freq
0.3
• The primary cursor is
highlighting the 1xbelt
X - Motor speed
with several
x - Fan speed
0.2
harmonics.
• The 2xbelt is very
0.1 dominant suggesting
there is damage to the
belts.
0
Freq: 835.69
0 8000 4000 12000 16000 Ordr: .657
Frequency in CPM Spec: .04393
Label: Belt defect/worn belts & sheaves
Case Study 9 – Belt Defect
• As this is a critical machine it was recommended on the next
available opportunity that the belts needed to be checked for
damage and re-aligned.
• The machine was stopped and the belts were inspected based upon the
recommendation.
• Significant damage was found to several of the belts during this
inspection as well as worn pulleys. Both the belts and pulleys were
replaced and correctly aligned before re-starting the machine.
Resonance
Resonance
• Resonance is defined as:
An excitation of a natural frequency by a periodic forcing
function.
Before Excitation
Frequency
Resonance Curve
Frequency
Amplified Signal
Frequency
Resonance
• There are two factors that determine the natural frequency of
an asset these are;
1. Mass – The heavier an object the lower the natural frequency
2. Stiffness – The more rigid a structure the higher the natural frequency
21
RMS = 19.85
LOAD = 100.0
we need to check the amplitude
RPM = 1484. (24.73 Hz)
18 of the 1xTs 90 degrees to this
RMS Velocity in mm/Sec
0
Freq: 24.72
0 500 1000 1500 2000 Ordr: 1.000
Frequencyin Hz Spec: 19.50
Resonance – Multi Plot
• The multi point plot allows the analyst to display several measurement
points on the same plot. Here we are showing all the radial points from the
motor.
– It is very clear that the amplitudes of the 1xTs peak are excessive in the
horizontal direction when compared to the vertical. This is a characteristic of a
resonant condition. 40 - No 1 GCTCompressor
M4551 - Multiple Points (13-Feb-03)
24
Max Amp 20
22.0
16
12
0
RMSVelocity in mm/Sec
M2V 10:15
M2H 10:14
M1V 10:14
Point= M2H
13-Feb-03
10:14:46
RPM= 1484.
M1H 10:14
Freq: 25.00
0 500 1000 1500 2000 Ordr: 1.011
Frequency in Hz Sp 3: 19.35
Case Study 10 – Resonance
• The following case study is taken from a motor and a
reciprocating compressor. The unit is mounted on a steel
frame which, in turn sits on spring mounts designed for
dampening
• Recently the motor had been replaced due to bearing
defect; however the new motor was smaller and lighter but
delivered the same power as the previous motor.
• When the compressor was put back into service it was
noted there was excessive vibration coming from the unit.
The unit was left to run like this for several months until the
vibration became to excessive.
Case Study 10 – Resonance
• Data was taken across the unit using route based data
collection.
SL - Compressor
CP1 -M1H Motor Outboard Horizontal
60
Route Spectrum
02-Feb-04 15:09:54
30
20
10
0
Freq: 24.83
0 300 600 900 1200 1500 1800 Ordr: 1.000
Frequency in Hz Spec: 45.19
• The plot above is taken from the motor showing a 1xTs peak
in excess of 40mm/sec.
Case Study 10 – Resonance
• This data is very high in amplitude.
• The data was then displayed in a multi plot format to show
how the amplitude was across the radial plains.
• Due to the vastly different amplitudes at the 1xTs frequency
the defect on this motor was Resonance.
SL - Compressor
CP1 - Multiple Points (02-Feb-04)
50
Max Amp
44.1
40
20
10
0
M2H 15:26
M2V 15:26
0 4000 8000 12000 16000
Frequency in CPM
Case Study 10 – Resonance
• Recommendation
• It was determined that the change in motor size may be the
cause of the resonance as the mass had been altered. A visual
inspection of the frame work also revealed that one of the
support beams had cracked along the weld – this altering the
stiffness of the structure. The support was welded and
strengthened and more data was acquired to determine if any
effect on the resonance had occurred.
Case Study 10 – Resonance
• The spectra, shows the ‘Before’ and ‘After’ plot of the motor
inboard horizontal. It shows a significant drop in amplitude of
the 1xTs peak.
– By stiffening the structure the natural frequency had increased moving
it away from the 1xTs peak thus resulting in a significant drop in
amplitude. CP1
SL - Compressor
-M2H Motor Inboard Horizontal
50
Max Amp
44.1
40
30
20
PK Velocity in mm/Sec
10
0
07-May-04
10:08:05
02-Feb-04
15:26:38
0 1000 2000 3000 4000
Frequency in Hz
Summary of Faults
Resonance
Electrical
Imbalance
t
Misalignmen
Frequencies
Misalignment
Severe Looseness
c c c c c c c
Frequency
In Terms
Of RPM Most Likely Causes Other Possible Causes & Remarks
11 0.44
7.1 0.28
4,5 0.18
3,5 0.11
2,8 0.07
2,3 0.04
1.4 0.03
0,71 0.02
113 4.45
90 3.54
71 2.80
56 2.20
45 1.77
36 1.42
28 1.10
22 0.87
18 0.71
11 0.43
10 times = 20dB
Permissible 11.2
Just Tolerable 7.1
Just
Tolerable 4.5
Just Tolerable Allowable 2.8
Allowable 1.8
Allowable Good 1.12
Good Large Machines
15 kW<
Medium Machines
with rigid and heavy 1.71
foundations whose
Good <75kW
natural Frequency
0.45
Small <300 kW on special exceeds 0.28
Machines< 15 kW foundations machine speed
0.18
Group K Group M Group G
Setting Alarm Limits
Alarm criteria:
• The best alarm criteria will be based on prior data from the same machine,
tested under identical conditions.
• You can trend data
• You can compare to identical machines
• The goal here is to use a low cost, low tech method to see if overall vibration
levels are changing over time and then to take action based on an alarm.
• This type of monitoring may be done for a number of reasons, such as for
compliance with regulators.
Data collected:
• RMS overall levels
Alarm criteria:
• ISO or other standards
• Trending
ISO Alarm Limit
• The ISO 10816.1 General Machines severity chart has limits for four classes of
machines. The alarm limits are in Velocity and shows the values in both metric
and imperial units. Note that both are RMS values.
ISO 10816.1 Severity Chart: These are for overall values from 3 Hz to 1000 Hz.
Spectrum Alarm Limits
• The Overall value is a good number to trend. But it does not catch everything. It
does not let us know about small amplitude values that may indicate severe or
critical conditions.
• A better method is to have the system scan particular regions of the spectrum,
and compare against a level for that region.
• Which method should be used…Band, envelope, or statistical?
• Reference is still needed to compare against particularly when starting out.
• There are basically two ways to set a reference alarm level.
• The first is to utilize published alarm limits and set fixed alarm limits.
• The second is to start with existing vibration readings and perform a calculation
to derive the alarm limit.
• There are pros and cons to both approaches.
ISO 2372/10816.1 General Machinery Standard
• ISO limits are defined for overall limits. Because 1X vibration normally
dominates, it can be used for 1X limits.
• But limits are needed for the bearing frequencies, etc.
Band Alarms-1
• How fixed alarm limits and statistical calculations can be used to compute values
that can be applied to vibration limits.
• This is an area where the approach taken by the software package being used.
• In brief, there are "band alarms", "envelope (or mask) alarms", "expert systems"
and "artificial intelligence systems".
Six to twelve bands may be available for setting alarms. Bands are often defined by Analysis Parameter Sets.
Band Alarms-2
• Simply put, band alarms work on the principle that it is possible to consider
different portions of the spectrum, and different scalar measurements (overall
readings, bearing measurements, etc.) and apply different alarms to each band.
• For example, in a spectrum, one band could be created around the running speed
peak.
• The software will focus on the vibration level between 0.9X to 1.1X, for example.
• The software will look at the vibration in that frequency band and compute the
maximum level, an average level, the RMS level, or some other parameter and
see if it exceeds a limit.
• The same technique is also applied to other "scalar" data such as overall readings,
waveform pk-pk readings, crest factor calculations, bearings measurements and
other useful parameters.
Frequency Band Descriptions for Pumps
• One limitation of Band Alarms is that the bands are not necessarily sensitive enough.
• The RMS calculations are dominated by the highest peaks and harmonics and
sidebands may be ignored.
• Small peaks can grow next to big ones and the alarm will not be tripped.
• This is especially true of bearing frequencies.
Frequency Band Descriptions
• Envelope alarms (also known as mask alarms, and not to be confused with
envelope detection used in bearing analysis), take a different approach.
• Rather than breaking the spectrum into individual bands, an alarm limit is applied
to the entire spectrum.