Vibration Level IIfinal
Click here to load reader
description
Transcript of Vibration Level IIfinal
Training Course on Vibration Analysis Level-II
Centre for Vibration Analysis & Machine Condition Monitoring (CVCM)[email protected]
Ph:051-9246021Fax:051-9218114
Cell:0300-8561060
by
Course Facilitator
S. Zafar Hussain Kazmi, Director (CVCM) Ghulam Mustafa, Manager (CVCM) Muhammad Haroon, Manager(CVCM) Hamid Raza , Senior Engineer Abid Khan, Senior Engineer
Hand’s on Training Aftab Ahmed, Principal Tech Altaf Hussain, Principal Tech Zaheer Taj, Senior Tech
Review of Maintenance Practices Review of Condition Monitoring Technologies Principles of Vibration Data Acquisition Signal Processing Vibration Analysis Fault Analysis Equipment Testing and Diagnostics Corrective Action Running a Successful Condition Monitoring Program Acceptance Testing Review of ISO Standards
Course Contents
Review of Maintenance Practices
Maintenance-1
Reactive Maintenance
― Often called ‘Breakdown Maintenance’ and has the concept ‘fix it when it breaks’.
― This is probably the most common type of maintenance in industry today but can be the most costly, especially on critical machines.
― Maintenance costs are usually higher due to the catastrophic failure that occurs.
Maintenance-2
Planned Maintenance
― Also known as ‘Shutdown Maintenance’.
― This is based upon ‘Timed Intervals’ between maintenance.
― Can be very effective if maintenance and resources are aimed at the machines that need it the most.
― However it can be very difficult to distinguish which machines actually need maintenance.
Maintenance-3
Predictive Maintenance
― Also known as ‘Condition Based Maintenance’.
― This approach uses non-intrusive technologies to determine the actual condition of a machine and its rate of failure.
― This can be very effective in extending machine life with big financial savings if implemented properly.
Maintenance-4
Proactive Maintenance
― Often referred to as ‘Root Cause Analysis’.
― This philosophy works hand in hand with Predictive Maintenance, eliminating the source of the fault to try to prevent it from re-occurring.
Today’s Industrial Demand
― It should be unacceptable to deliver
• less performance for more money
• same performance for more money
― It could be acceptable to deliver
• same performance for less money
• more performance for the same money
• more performance for more money
The desire is More Performance for Less Money!!!!
Predictive Maintenance Objectives
― To confirm good-condition machines
― To detect developing problems
― To determine the nature and severity of the problem
― To schedule repairs that can best fit with production and maintenance needs
Predictive Maintenance Techniques
― Vibration measurement
― Electrical testing
― Motor current analysis
― Reciprocating machine testing
― Thickness testing
― Visual inspection
― And many more…
Predictive Maintenance Basic Facts
― Every mechanical or electrical faults on a machine has a distinct vibration behavior.
― Any change in the vibration signature indicates changes in the dynamic operating condition of the machine.
Predictive Maintenance Mechanism (VA)
― Establish a database of all the machines that need to be monitored
― Establish a data collection route that best optimize the data collection time
― Download route into the data collector
― Collect data
― Upload collected data into the database
Predictive Maintenance Mechanism(1)
― Run exception reports to detect the problematic machines
― Analyze only the machines in the exception reports
― Generate repair work to be performed
― Again collect data on the machine on which work is being done.
Start
Create Ref.
RegularMeas.
Inputm/c
specs
FaultDiagnostics
Fault correction
Compare limits
YES
NO
Rules +
Experience
Create New Ref. & Limits
Predictive Maintenance Mechanism(2)
‘What is Vibration?’
Basics of Vibration
Introduction:
What is Vibration? (1)
What is Vibration ?― Vibration is the motion of a body
about a reference position caused by a force
In simple terms vibration is :-
― ‘A response to some form of excitation’.
― The free movement of shaft in a journal bearing will cause it to vibrate when a ‘forcing function’ is applied
What is Vibration?
― Vibration is a pulsating motion of a machine or a machine
part from its original position of rest and can be
represented by the formula:
Vibration Amplitude Response =
Dynamic Force ______________
Dynamic Resistance
Principle of Vibration Analysis
Mechanical faults generate unique vibration ― Geometry of the machine
― diameter of the shaft, number of bearing elements, etc.
― Turning speed (e.g. RPM)
Mechanical Defects detected with vibration analysis
Imbalance Belt drive faults
Misalignment Machine resonance
Bent shaft Cavitation
Looseness Shaft Rub
Bearing Defects including:
- cage defect
- outer race defect
- inner race defect
- rolling element defect
Gear defects
Electrical faults
Vibration from Mechanical Faults
Vibration from Mechanical Faults
Vibration from Mechanical Faults
Vibration from Mechanical Faults
Vibration from Mechanical Faults
Vibration from Mechanical Faults
Vibration Fundamentals
How Much Vibration is Too Much ?
1 ― Use Absolute Vibration Levels
• Given by machine makers• Published Vibration Severity Standards
e.g. ISO 2372, VDI 2056, BS 4675
2 ― Use Relative Vibration Levels
ISO 10816-3
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
mm/s rms inch/s rms
rigid flexible rigid flexible rigid flexible rigid flexible Foundation
pumps > 15 kW medium sized machines large machines
radial, axial, mixed flow 15 kW < P 300 kW 300 kW < P < 50 MW Machine Type
integrated driver external driver motors motors 160 mm H < 315 mm 315 mm H
Group 4 Group 3 Group 2 Group 1 Group
A newly commissionedB unrestricted long-term operation
C restricted long-term operation
D vibration causes damage
ISO 10816-3
140 5.51
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
µm rms mil rms
rigid flexible rigid flexible rigid flexible rigid flexible Foundation
pumps > 15 kW medium sized machines large machines
radial, axial, mixed flow 15 kW < P 300 kW 300 kW < P < 50 MW Machine Type
integrated driver external driver motors motors160 mm H < 315 mm 315 mm H
Group 4 Group 3 Group 2 Group 1 Group
A newly commissioned
B unrestricted long-term operation
C restricted long-term operation
ISO 10816-3
Vibration standards are guidelines
NotPermissible
NotPermissible
NotPermissible
GoodLarge Machines with rigid and heavyfoundations whose natural Frequency exceedsmachine speed
Just Tolerable
Allowable
GoodSmall Machines< 15 kW
Just Tolerable
Allowable
Just Tolerable
Allowable
Good15 kW< Medium Machines <75kW
<300 kW on specialfoundations
2.5
tim
es =
8d
B
10 t
imes
= 2
0dB
Group K Group M Group G
45281811.27.14.52.81.81.121.710.450.280.18
Vel
ocit
y m
m/s
RM
S
ISO2372 ( BS 4675 , VDI 2056 )
Displacement
― Displacement is a measure of the actual distance an object is moving from a reference point.
― Displacement is expressed in “mils” 1 mil = .001 inch
― Displacement is also frequency related, in that 10 mils @ 1000 rpm is not the same as 10 mils @ 10000 rpm.
Velocity
― Velocity is the rate of change in position.
― Typical velocity units are: IPS (Inches Per Second), mm/sec (millimeters per second).
― Velocity is the most accurate measure of vibration because it is not frequency related. 0.5 IPS @ 1000 rpm is the same as 0.5 IPS @ 10000 rpm.
Acceleration
― Acceleration is the rate of change of velocity and is the measurement of the force being produced.
― Acceleration is expressed in gravitational forces or “G’s”, (1G = 32.17 ft/sec/sec)
― Acceleration is frequency related, in that 1 g @ 1000 rpm is not the same as 1 g @ 10000 rpm.
Orbit-1
― Orbits used to measure relative shaft movement within a journal-type
bearing.
― The shape of the orbit told the analyst how the shaft is behaving within
the bearing as well as the probable cause of the movement.
― This accomplished using proximity probes usually mounted through the
bearings with a 90-degree separation
― With modern analyzers, it is possible to also collect an orbit using case-
mounted velocity probes or accelerometers to see how the machine
housing is moving.
Orbit-2
― An orbit is usually collected while the machine is at its normal operating
state or speed, but it can also be collected while the machine is increasing
or decreasing in speed, such as during a coast-down or startup.
― The data can be collected in a steady state, in what is known as an
unfiltered orbit, requiring no tachometer or at multiples of running
speeds such as first, second or third order to look for issues relating to
that or another specific frequency.
Unfiltered displacement orbit Unfiltered velocity orbit Filtered orbit
Orbit-3
― Orbits are Lissajous patterns of time domain signals that are
simultaneously plotted in the X–Y coordinate plane of an oscilloscope
or vibration analyzer.
― In this form of display, it is very difficult to trace the start of the orbit
as it appears to be an endless loop.
― In order for us to determine the direction of rotation, a phase trigger is
employed.
― The trigger will show the direction of rotation by looking at the dot on
the orbit as the starting point of 1× RPM and the blank space as the
end point.
Orbit-3
Figure: Vibration Pickups in Orbit Analysis Application
Orbit-3― Orbit analysis is the vibration measure of any rotor system in an X–Y plot
(Figure ).
― In most applications, the unit of measurement is displacement which is
measured directly using proximity probes.
― These types of measurements are relative vibration readings.
― Relative readings are considered vibration measurements of the shaft with
respect to the bearing housing.
― As the probes are clamped firmly to the housing, there is no relative motion
between the probe and the housing. Thus, the orbit is achieved.
Orbit-3― With that in mind, orbit plots give a visual graph of the actual shaft centerline
movement inside the bearing housing.
― Accelerometers and velocity pickups can also be used to create orbits. These are
external transducers, which require mounting on the outside of the bearing
housing.
― These types of measurements are called case orbits. Case orbits are useful to
separate shaft and case vibrations. This can provide absolute shaft motion
(relative to space).
― Orbit may be done for the overall signals as measured or it can be done for
filtered signals where it is required to show orbit for specific frequency such as
the frequency of rotation or its multiples.
Orbit-1 (Understanding Orbit Plot)
― To understand orbits, waveforms and their relationship to orbits is
necessary to understand.
― Let us begin with waveforms.
― The waveform plot shown in Figure has two sine waves, Y and X.
― The Y plot is on the top and the X plot is at the bottom.
Orbit-2 (Understanding Orbit Plot)
― The waveform signature runs left to right and the amplitudes change from negative to positive, whatever the case may be.
― The changes in the waveform cause the orbit to form.
― An orbit is made up of an X- and Y-axis with zero in the center.
― Starting from the center, up is positive and down is negative.
― Right is positive and left is negative.
― Now that we know waveform and orbit conventions, let us trace the waveforms and create an orbit.
― The Y plot is on the top and the X plot is at the bottom.
Orbit-1 (Effect of Probes Direction and Keyphasor )
― In many cases, it is not possible to mount probes easily in the desired 90
degrees out of phase horizontal and vertical orientation.
― Probes often have a 45-degree deviation instead.
― The following illustration shows the common mounting positions for X
and Y probes.
Orbit-2 (Effect of Probes Direction and Keyphasor )
― When using keyphaser signal to specify the starting point of orbit plot,
the resulting plot will be referenced to the keyphaser location.
― Therefore, if the keyphaser is positioned in the same angle as the X-
sensor, the result is the actual orbit of shaft movement, otherwise, the
plot will be shifted by angle between keyphaser and X-sensor.
Orbit-1 (Applications of Orbit Analysis)
― Orbit plots can efficiently be used in vibration diagnosis where other techniques, such as FFT and time waveform, may not provide sufficient information.
― In the following, some vibration problems will be discussed.
Unbalance:-
Unbalance will generally produce 1xRPM vibration with 90˚ phase shift
between the horizontal and vertical directions. This will result is ellipse-shaped
orbit as that shown in the Figures.
Orbit-2 (Applications of Orbit Analysis)
Misalignment:-― When radial preloads due to misalignment, gravity, fluid forces and
other causes increase in magnitude, the orbit will become acutely
ellipsoid.
― A bearing preload due to a cocked assembly can also cause the orbit
to have lower amplitude in one axis that makes the ellipse look thinner.
― The average shaft centerline will move from the normal position to ― the upper left quadrant, for example, all points on the orbit are moving
clockwise (which is the same as the direction of rotation) and therefore
the orbit is still in forward precession.
Orbit-3 (Applications of Orbit Analysis)
― If the preloading increases further, it will result in the orbit’s shape to
resemble a number 8 character as shown in Figure .
― In this case, it is also interesting to follow the average shaft centerline
position, which has now moved further upwards into the left quadrant.
― If this orbit is carefully studied, it will be noticed that if a point on the
orbit begins its journey from the dot, it is moving counter-clockwise
initially, whereas the shaft is rotating in the clockwise direction.
― Thus, heavy preloading due to misalignment can cause the shaft to go
into reverse precession.
Orbit-4 (Applications of Orbit Analysis)
― Forward precession is normal, reverse is not.
― If the trajectory of our imaginary point on the trace of the orbit is
continued, one can visualize that precessions keep changing
continuously.
Orbit-5 (Applications of Orbit Analysis)
Rotor Rub:-― Orbit analysis is a good tool to identify rubs.
― As mentioned earlier, partial or complete rubs can occur when a rotating shaft comes in
contact with stationary parts like seals or in abnormal cases of bearing (and/or
instrumentation) failures.
― The rub causes the orbit to take on different shapes.
― From a number 8 to a full circle to something like the orbit shown in Figure
Orbit-6 (Applications of Orbit Analysis)
Oil Whirl :-― Oil whirl is basically a sub-synchronous fluid instability.
― When viewed in the orbit domain, it is shown with the characteristic two dots.
― When viewed with an oscilloscope, the two dots do not appear stationary, but seem to be rotating instead.
― This is because the frequency is marginally less than 0.5X.
― An oil whirl phenomenon generates a vibration precession, which is always forward as shown in Figure.
Orbit-7 (Applications of Orbit Analysis)
Oil Whip:-― The oil whip phenomenon occurs when the rotor is passing through its critical speed.
― Oil whip is a destructive bearing defect.
― The precession of vibration is in the forward direction in this case, but some reverse 1X and sub-synchronous components are present due to anisotropy (changes in response when operating conditions change) of the bearing pedestal stiffness.
― The period of this self-excited defect may, or might not, be harmonically related to the rotating speed of the shaft.
― When it is not harmonically related, the dots appear to be moving randomly as shown in Figure. When it is harmonically related they appear stationary.
Beating (1)
― If two vibration components are quite close together in frequency and if
they are present at the same time at the same place, they will combine in
such a way that their sum will vary in level up and down at a rate equal to
the difference in frequency between the two components.
― This phenomenon is known as beating, and its frequency is the beat
frequency.
― There is confusion in some areas between beating and amplitude
modulation, which also can produce an undulating vibration level.
Beating (2)
― Amplitude modulation is different from beating, and is caused by a high-frequency component being multiplied by a lower-frequency component and is thus a nonlinear effect, whereas beating is simply a linear addition of two components whose frequencies are close to one another.
Modulation(1)
― Amplitude modulation is defined as the multiplication of one time-
domain signal by another time-domain signal.
― The signals may or may not be complex in nature, i.e., either or both
signals may contain harmonics components.
― It is impossible to have amplitude modulation unless at least two
different signals are involved.
― Modulation is inherently a non-linear process, and always gives rise to
frequency components that did not exist in either of the two original
signals.
Modulation(2)
― Examples of machines that produce amplitude modulation are gearboxes, where
the tooth mesh frequency is modulated by the turn speed of each gear, and
rolling element bearings, where bearing tones can be modulated by turning
speed or the fundamental train frequency of the bearing.
VIBRATION ANALYSIS TECHNIQUES
Vibration Characteristics-1
Amplitude How Much
Frequency How Often
Phase When
Vibration Characteristics-2Amplitude:-― Amplitude is the magnitude of vibration expressed in terms of signal level (millivolts or
milliamps) or in engineering units ( Micron, mils, milli meter per second or inch per second).
― There are many ways of measuring vibration amplitude levels, the most common are:
• peak to peak, zero to peak,
• root mean square (RMS),
• average and crest factor.
― Zero to peak or peak is the measurement from the zero line to the top of the positive peak or
the bottom of the negative peak.
• Peak = 1.414 x RMS
• RMS = 0.707 x PEAK VALUE
• Peak to peak is the distance from the top of the positive peak to the bottom of the negative peak. This measurement is used most often when referring to displacement amplitude .
• Pk - Pk = 2 x PEAK VALUE
Vibration Characteristics-3
― The average value is 0,637 times the peak of a sin wave; average values are measured
by most analog meters.
• Avg = 0.637 x PEAK VALUE
― The crest factor is determined by dividing the peak value by the RMS value. For a true
sine wave.
• Crest Factor = 1/.707 = 1.414
Am
plit
ude
+
Time ‘t’
Peak to Peak
0 to Peak
RMSAverage
-
Vibration Characteristics-4
― There are three types of measurements used to display amplitude.
― These are:
Displacement Velocity Acceleration
Vibration Characteristics-5
Displacement:- is the distance that shaft moves in relation to
reference point. The total movement of the shaft is measured in Peak to
Peak.
Velocity:- is the displacement of the shaft in relation to time? It is
measured in RMS (Root Mean Square) or Peak.
Acceleration:-is defined as the change in velocity over time. With this
value we want the maximum impact (Force) generated, so we use the
Peak or RMS measurement.
Vibration Characteristics-6
Amplitude Units ‘ Metric’
Displacement µm Pk-Pk
Velocity mm/sec RMS
Acceleration g’s Pk
Vibration Characteristics-7
Amplitude Units ‘ English’
Displacement mils Pk-Pk
Velocity inch/sec Pk
Acceleration g’s RMS
Conversion of Units-1
Metric Units
where:
D = Peak-Peak Displacement (μm Pk-Pk)
V = Peak Velocity (mm/sec Pk)
A = Peak Acceleration (g’s Pk)
F = Frequency (CPM)
V = DF / 19,100 V = 3690A / F A = DF2 / 70,470,910
D = 19,100V / F A = VF / 3690 D = 70,470,910 / F2
Conversion of Units-2
English Units
where:
D = Peak-Peak Displacement (mils Pk-Pk)
V = Peak Velocity (inch/sec Pk)
A = Peak Acceleration (g’s Pk)
F = Frequency (CPM)
V = DF / 19,100 V = 93,640A / F A = DF2 / 1,790,000,000
D = 19,100V / F A = VF / 93,640 D = 1,790,000,000 / F2
Vibration Characteristics - Amplitude Relationships-1
― The three types of amplitude measurements used to display data are directly related
to each other.
― Changing from one amplitude unit to the next alters the way in which the data is
displayed. • Velocity is the default unit for standard data collection techniques
– High and low frequency events can be seen
A8 - Example 15Ex15 -F1H Fan Inboard Horizontal
Label: Large Fan Unit - Easy
Route Spectrum 22-Aug-02 11:30:50
OVERALL= 3.45 V-DG RMS = 3.44 LOAD = 100.0 RPM = 831. (13.85 Hz)
0 20000 40000 60000
0
1
2
3
4
5
Frequency in CPM
RM
S V
elo
city
in m
m/S
ec
For normal operating speed ranges, velocity data provides the best indication of machine condition
A8 - Example 15Ex15 -F1H Fan Inboard Horizontal
Label: Large Fan Unit - Easy
Route Spectrum 22-Aug-02 11:30:50
OVERALL= 3.45 V-DG P-P = 104.98 LOAD = 100.0 RPM = 831. (13.85 Hz)
0 20000 40000 60000
0
20
40
60
80
100
120
140
Frequency in CPM
P-P
Dis
pla
cem
ent
in M
icro
ns
Low frequencies require very little force to move an object
A8 - Example 15Ex15 -F1H Fan Inboard Horizontal
Label: Large Fan Unit - Easy
Route Spectrum 22-Aug-02 11:30:50
OVERALL= 3.45 V-DG PK = .3909 LOAD = 100.0 RPM = 831. (13.85 Hz)
0 20000 40000 60000
0
0.05
0.10
0.15
0.20
0.25
0.30
0.35
Frequency in CPM
PK
Acc
eler
atio
n in
G-s
Increasing the frequency that the objects move with the same velocity, the force needed to move it increases, thereby reducing the distance it can travel
Displacement measures low frequency events ignoring high frequencies– Relative shaft motion
Acceleration accentuates the high frequencies ignoring the low frequencies– Good for early bearing
detection (Whenever there is Metal to Metal Impacting involve)
Vibration Amplitude Measuring Units Acceleration
G’s or in/s2
(180 deg phase lead)
Velocity
mm/sec or inches/sec(90 deg phase lead )
Displacement
Acceleration, a
Velocity, v = a /2 f
Displacement = a /4 2 f 2
VelocityAcceleration
90 o
Time
90 o
Displacement
μm or inch or Mils
Vibration Characteristics - Amplitude Relationships-2
Vibration Characteristics – Frequency Relationships-1
Frequency― Measure of the number of cycles of vibration that occur in a specific
period of time :-
• Tells us at what rate the vibration is occurring
• Reciprocal of the Period (T)
• Measured in Hz /CPM
• Converted by a factor of 60
• CPM relates directly to machine RPM
― The time required to complete one full cycle of vibration
Frequency =1
T=
Cycles
Second
1
Period=
Vibration Characteristics – Frequency Relationships-2― 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
― Consider a motor has a rotational speed of 1485RPM, in terms of frequency this
equates to:• 1485 CPM (1rpm = 1cpm)• 24.75 Hz (1485/60) (minutes to seconds)• 1 Orders (1 x revolution of the shaft)
Vibration Characteristics – Frequency Relationships-3Frequency
― The table below demonstrates the relationship between the different
frequency units over a range of frequencies.
CPM 1500 2250 3000 6000 12000
Hz 25 37.5 50 100 200
Orders 1 1.5 2 4 8
― Stress = Displacement 0-600 CPM
― Fatigue = Velocity 600-120,000 CPM
― Force = Acceleration Above 120,000 CPM
Vibration Institute recommendation
Vibration Characteristics – what to use ?
Vibration Characteristics
Significance of Frequency:-
― Essential to pinpoint the cause of a machine problem.
― The forces that causes vibration are usually generated through the
rotating motion of the machine parts.
― These changes in direction and amplitude according to rotational
speed of the machine components, most vibration problem will have
frequencies that directly related to the rotational speed.
― Vibration frequency is an analysis or diagnostics tool
Vibration Characteristics - What is Phase-1?
― Phase is the measure of time
difference expressed in degrees
between two events occurring at the
same frequency.
― Phase is the relationship of vibration
motion with respect to an other
vibration part or fixed reference point
1 Cycle = 360
180°
A
B
90°
A
B
A & B are180 degreesout of phase
A
B
A
B
A & B are90 degreesout of phase
Vibration Characteristics - What is Phase-2?
Two Types of Phase
• Absolute phase Absolute phase is the relationship of the Peak of vibration and
a fixed reference Signal (once per revolution)• Relative Phase Relative phase is the relationship between two Peaks of
vibration signals
Absolute Phase
Phase lag angle between once per turn marker andfirst positive peak in a vibration waveform
• Express in degrees phase lag• Must be filtered to multiple of turning speed
a
b
Absolute Phase = a/b X 360o
Relative Phase
Phase lag angle between positive peaks of twoseparate vibration signals (equivalent events)
• The two signal must be same vibration unit (eg. vel & vel or displ & displ)
Signal B lags signal A by 110o
Velocity signal A
Velocity signal B
110o
Significance of Phase
• Phase measurements are not taken during routine data collection of
predictive maintenance
• However, when developing problems are found comparative phase
readings can provide valuable information pinpointing the specific
problem
Vibration Phase Analysis-1 (Bubble Diagram)
• Rather than recording the phase readings numerically, record them visually.
• It can be difficult to look at a series of numbers and interpret the movement of the machine.
• However using graphical symbols makes this task easier.
• By drawing a circle and a tail at the desired angle, it is easy to quickly determine the angle with
a quick glance, as shown in figure.
• Do not need to write down the phase angle.
• Just draw the tails; either inside or outside the circle, as shown in Figure.
• The two readings are 180° out of phase.
• Often the angle is written above the horizontal line and the amplitude is written below the
line.
Vibration Phase Analysis-2 (Bubble Diagram)
• This data can be used in a number of ways, but one common method is called
the bubble diagram(developed by Ralph T. Buscarello), as illustrated in Figure.
• Take readings around the machine and enter them into the diagram, adding the tails
according to the angle.
Vibration Phase Analysis-3 (Bubble Diagram)
• You must be careful when comparing phase readings taken at opposite ends of a
machine, or when comparing phase readings taken across a coupling.
• Phase readings are sensitive to direction. Therefore you have to add 180° to your
readings if the accelerometer is turned 180°.
• You must also be familiar with the phase convention used by your data collector.
Figure illustrates one such convention.
Vibration Phase Analysis-4 (Bubble Diagram)
• Also note that when talk about the phase relationships between certain points machine.
• That the phase readings should be in-phase, 90° or 180° out of phase.
• These are only approximate values.
• The actual readings may be up to 30° higher or lower and the rule still holds.
• For example, if the difference between two readings was between 150° and 210°, then you can consider the readings to be 180° out-of-phase.
• Also, if the difference between two readings is approximately 270°, then that is equivalent to a 90° phase difference.
• Likewise the phase difference of -180° is equivalent to a 180° phase difference.
• It all depends upon the direction of rotation, the setup of the data collector, and the convention used by the data collector.
Vibration Phase Analysis-4 (Bubble Diagram)
Vibration Phase Analysis-5 (Bubble Diagram)
Operating Deflection Shape-1 (ODS)
• Operating Deflection Shapes (ODS) are used for visualization of the vibration pattern of a structure under real life operating conditions.
• Vibration measurements are performed at different points and directions on the structure known as degrees of freedom (DOFs) and the vibration pattern can be shown in a number of formats including an animated geometry model of the structure.
• Following Figure shows an example of a geometry model before animation.
• Unlike modal analysis techniques which only help visualize the inherent resonant characteristics of a product.
• ODS is a very powerful tool that can solve problems related to forced vibrations .
Operating Deflection Shape-2 (ODS)
• The method of investigation employs a software package that allows the user to observe, analyze and document the dynamic behaviour of machines and mechanical structures.
• The software displays spatially acquired vibration on a 3D model of the test structure.
• Measurements on the structure are carried out using a real time multi-channel frequency analyzer and exported into the display software where the ODS are viewed.
• All of the geometrical drawing and the measurement process are carried out in a pre-built real time multi-channel frequency analysis project.
• The software enables the user to investigate the operating deflection shapes of the machine under test.
• By establishing the way in which the machines are vibrating we will be able to identify the most appropriate mounting locations for transducers to evaluate the greatest risk of vibration exposure.
• Locating the transducers on a node where there is little or no vibration will result in an underestimate of the vibration magnitude associated with the use of the tool.
Operating Deflection Shape-3 (ODS)
• Traditionally, ODS have been defined as the deflection of a structure at a particular frequency.
• However, ODS can be defined more generally as any forced motion of two or more DOFs on a structure.
• All vibration is a combination of both forced and resonant vibration. • Forced vibration can be due to:
• 1. Internally generated forces • 2. Unbalances • 3. External loads
• An operating deflection shape contains the overall vibration for two or more DOFs on a structure.
• An ODS therefore contains both forced and resonant vibration components whereas a mode shape characterizes only the resonant vibration at two or more DOFs .
Operating Deflection Shape-4 (ODS)
• Real continuous structures have an infinite number of DOFs and an infinite number of modes.
• From a testing point of view, a real structure can be sampled spatially at as many DOFs as desired but there is a limitation when looking at small structures such as sanders.
• The more we spatially sample the surface of the structure by taking more measurements within the limitations, the more definition we will give to its ODS as less interpolation will be required between measured points.
• ODS analysis is a method to model the motion of a structure as influenced by its own operating forces and/or those from external sources.
• They can be viewed for a specific moment in time or a specific frequency. • For our investigations the most relevant frequency is the operating frequency
under real conditions. Unlike modal testing, measurements are obtained during normal operation.
Operating Deflection Shape-5 (Example ODS)
• In order to show exactly how the software works and to show that it outputs true ODS, a simple example of a steel bar is used with the following procedure followed: • Calculate the expected natural frequencies of the bar via a known
equation. • Fasten the bar to a shaker and attach 6 transducers. • The transducer closest to the pivot will be the reference and so all
vibration will be relative to this transducer. • Tap the steel bar with a hammer close to the free end and observe
the FFT using a real time multi-channel frequency analyzer to show that the calculated frequencies are present in the bar.
• In turn drive the bar at the calculated frequencies via the shaker, observe the mode visually by viewing the bar, and then observe the ODS produced by the software and see if they appear the same.
Natural frequency of a beam:-
Operating Deflection Shape-6 (Example ODS)
Operating Deflection Shape-7 (Example ODS)
• Using the values of A in 1), 2) and 3) gives the frequencies of the first three modes of the bar to be 54Hz, 341Hz and 938Hz respectively.
• Natural frequency calculations can be made with
Operating Deflection Shape-8 (Example ODS)
• The geometry model is drawn in the software and involves drawing a long, thin box with the correct dimensions and has the right number of points to assign the transducers.
• Following Figure shows a real picture of the bar and a geometry model of the bar.
Picture of steel bar and the geometry model
Resonance Transducer
Resonance Transducer
Operating Deflection Shape-9 (Example ODS)
The Fast Fourier Transform (FFT)
The FFT showing the frequencies of the first three modes
Operating Deflection Shape-10 (Example ODS)
• The FFT is acquired via the real time multi-channel frequency analyzer.
• The steel bar is tapped repeatedly close to the free end and the FFT from one of the closest transducers is recorded.
• The resulting FFT is displayed in the Figure and shows three dominant peaks corresponding to the first three modes of the bar.
• The frequencies of the modes are shown to be 52Hz, 355Hz and 916Hz, which are very close to the calculated values.
• The calculations do not take into account the mass of the six transducers and so this could explain the difference.
Operating Deflection Shape-11 (Example ODS)
Obtaining the ODS• The steel bar is driven in turn at each of the three frequencies
and a single measurement set is recorded. • The data is then exported into the display software where the
ODS are viewed. • The geometry model is interpolated and viewed at the modal
frequency. • Interpolation involves the computation of points or values
between the ones that have been measured, using the data from the surrounding points or values.
Operating Deflection Shape-12 (Example ODS)
From Left to Right, Top to Bottom: The ODS of the first mode of the bar at 52Hz
Figure shows the ODS of the steel bar at 52Hz. This is the first mode of the bar with the shape being directly compared to picture-1
in mode shape Figure. It should be noted that the interpolation of the bar is not quite correct at the free end of the bar in given below Figure.
Reason: Transducer is not located right at the edge of the bar. No data were collected at this point.
Operating Deflection Shape-13 (Example ODS)
From Left to Right, Top to Bottom: The ODS of the second mode of the bar at 355Hz
Figure shows the ODS at 355Hz, the second mode of the bar. This should be directly compared with picture-1 in mode shape Figure.
Operating Deflection Shape-14 (Example ODS)
From Left to Right, Top to Bottom: The ODS of the third mode of the bar at 916Hz
Figure shows the ODS at 916Hz, the third mode of the bar. This should be compared with picture-1 in mode shape Figure. This is not as clear as the other two modes as more measurement points
would be needed in order to gain more definition at this mode. Less interpolation would be needed between measured points if more
measurements were taken.
Enveloping-1 (Demodulation)
• Also known as “demodulation,” the enveloping technique, which is used by a large number of vibration analyzer vendors, has been optimized to measure the low-amplitude, high-frequency bearing vibration.
• The envelope spectrum is then checked for signs of the fault condition. • Similar to the spectrum that results in the Shock Pulse, Spike Energy, and
PeakVue systems, we are looking for peaks, sidebands, and harmonics that are related to the four characteristic bearing frequencies:
• Ball Pass Frequency Outer race (BPFO), • Ball Pass Frequency Inner race (BPFI), • Ball (or roller) Spin Frequency (BSF), • and Fundamental Train (or cage) Frequency (FTF). • The envelope spectrum is then checked for signs of the fault condition.
Enveloping-2 (Demodulation)
There are two types of bearings :
• Rolling Element
• Anti-Friction” Bearings
• Fluid Film Bearings
• Rolling Element: Low cost, simple to apply. • But are capable of only moderate speeds and relatively light
loads. • Rotor dynamics aren’t bad but diagnostics can be complex
due to all those spinning balls. • Fluid Film: Capable of supporting very high loads, high
temperatures, high speed. • Expensive and associated rotor dynamics are very complex.
bearing
bearing housing
Accelerometer
bearing
Oil Wedge (load zone)
Soft Metal (Babbitt)
Eddy Current Probe
Enveloping-3 (Demodulation)
• What happens when there is a fault or defect on the inner or outer race of the bearing?
bearing
bearing housing
Accelerometer
Fault or Defect on Outer Race
We feel an impact anytime a ball or roller passes over the defect
This impact energy is typically very low amplitude
Enveloping-4 (Demodulation)
• In fact, the vibration energy from a bearing fault is so small sometimes that it gets hidden by all the other machine vibration going on:– Unbalance, Looseness, Misalignment, etc
Enveloping-5 (Demodulation)
• If we pass the signal through the right Envelope or High Pass Filter.
• We could theoretically leave only the vibration generated by our bearing fault.
Band Pass Filter
Am
pli
tud
e
Frequency
Enveloping-6 (Demodulation)
• Our Enveloped or High Pass Filtered Signal would look like this:
Am
pli
tud
e
Enveloping-7 (Demodulation)
• We need to know what frequency or frequencies we are trying to isolate
• The frequencies are generated by the impact of the ball or roller as it passes over the defect on the race
• So what frequency is this?
Enveloping-8 (Demodulation)
• If we envelope properly, we should be able to eliminate all the higher amplitude, low frequencies that are present in the signal:– Unbalance Frequency (1X)– Misalignment Frequencies (1X and 2X)– Looseness Frequencies (1X and 2X and possibly more running
speed harmonics)– Fundamental Bearing Defect Frequencies (Non-
harmonics from around 3X to around 40X) So what frequency is this?
Am
pli
tud
e
Enveloping-9 (Demodulation)
• If we envelope properly, all we should have left is the bearing natural frequency response to the impacts that are occurring as the balls or rollers pass over the defect(s)
• Why?– Because this gives us a measure of the energy
generated by any impacts or impulses on the system– Since we are measuring the amount of resonance
occurring in the system, it will be very sensitive to the severity of the impacts and hence, the severity of the fault
– If measured properly, we should see almost all bearing related energy
Enveloping-10 (Demodulation)
• Rockwell Automation (Entek and IRD Brands)– Use Analog High Pass Filters– 100 Hz, 200 Hz, 500 Hz, 1 KHz, 2 KHz and 5 KHz
• SKF– Use Analog Envelope Filters– 5 to 100 Hz, 50 to 1000 Hz, 0.5 to 10 kHz, 5 to 40 kHz
and 250 to 350 kHz• CSI
– Use Analog High Pass Filters– 500 Hz, 1 KHz, 2 KHz and 5 KHz
• Both CSI and Rockwell Automation then apply a digital low pass filter to the signal to create the Envelope
Enveloping-11 (Demodulation)
What are These?
Am
pli
tud
e
Enveloping12 (Demodulation)
• Where Do the Sidebands come from?
Sideband or Modulating Frequencies
Am
plitu
de
Carrier Frequency
Frequency
Enveloping-13 (Demodulation)
• If we apply a Digital Low Pass Filter as the upper end of the Envelope, we can then extract the modulating frequencies from the impact waveformA
mp
litu
de
Frequency
Digital Low Pass Filter (FMAX)
The Key Question: What frequencies are left?The Key Question: What frequencies are left?
Shock Pulse-1
• Of the methods used to assess the operating condition of rolling element bearings, one of the most successful and popular techniques is that of Shock Pulse evaluation.
• Shock Pulses are a special type of vibration that can be clearly distinguished from ordinary machine vibrations:
• The actual Shock Pulse is the pressure wave generated at the moment when one metallic object strikes another (Figure 1).
• The bulk of the impact momentum, however, acts to deform the target object, which then oscillates at its natural frequency.
• This vibration ultimately dissipates primarily as heat due to internal friction material damping (Figure 2).
Fig-1: Shock pulse Lubricated Fig-2: Deformation Wave
Shock Pulse-2
Shock Pulses in Bearings• Shock Pulses occur during bearing operation when a rolling element
passes over an irregularity in the surface of the bearing race. • Of course, there is no such thing as a perfectly smooth surface in
real life• Therefore, even new bearings emit a signal of weak Shock Pulses in
rapid succession.• This Carpet Value rises when the lubrication film between rolling
elements and their races becomes depleted.• A defect on the surface of a rolling element or bearing race
produces a strong Shock Pulse reaction with up to 1,000 times the intensity of the Carpet Value.
• These clusters of high amplitude peaks or Maximum Value stand out clearly from the background noise and are ideal indicators of bearing damage.
Shock Pulse-3
Measurement• Shock Pulses propagate within a much higher frequency range
than that of ordinary machine vibration, and their energy content is relatively low.
• Therefore, the accelerometer used for Shock Pulse measurement is tuned with a 36 kHz resonance frequency that lies precisely within the Shock Pulse frequency range.
• In addition, a 36 kHz band pass filter is applied to the accelerometer signal to help filter out lower frequency mechanical vibration.
• When Shock Pulse is present the tuned accelerometer resonance is excited and amplifies the Shock Pulse signal resulting in an excellent indication of bearing lubrication and damage.
• Shock Pulse is responsive even when far more energetic machine vibration is present.
• Therefore, lower frequency mechanical conditions such as unbalance, shaft misalignment or vibration from adjacent machines have little effect on Shock Pulse.
Shock Pulse-4
Measurement• In addition, high-frequency signals tend to dissipate rapidly so very
little interference is encountered from adjacent bearings.
Carpet Level = Bearing Noise
Signal Peaks above Carpet Level
Shock Pulse-5
Evaluate Bearing Condition• Just as with other condition evaluation methods, the Shock Pulse
technique reaches its conclusions via certain defined parameters. • Shock Pulse is influenced by factors such as bearing size, rpm,
signal damping and lubrication.• Shock Pulse readings generally should be compared with
‘signature’ readings taken when the condition is known to be good.• However, through the years, reliable normalization methods have
been developed which correct the effect of bearing size and rpm on Shock Pulse.
• Shock Pulse measurements will track up and down with speed and down for smaller bearing diameters and up for larger bearing diameters.
• By entering the bearing diameter and shaft speed, a normalization factor or dBn is automatically calculated and applied.
Shock Pulse-6
• This corrects for speed and bearing differences and allows one to properly evaluate a bearing on the first Shock Pulse measurement.
• The application of normalization also allows for the use of standardized alarm levels on a machines running at different speeds and with varying bearing diameters.
Shock Pulse-7
Quantitative Analysis• Two normalized parameters are used to determine bearing condition:• • Carpet Value—Indicates deterioration or a poor lubrication
condition. This can be caused by a number of conditions such as insufficient lubrication, lubrication contamination, shaft misalignment, improper bearing load or improper bearing installation.
• • Maximum Value—Indicates damaged bearing elements. Maximum Value is generated by clusters of higher amplitude Shock Pulses. This is caused when the bearing elements hit defects within the bearing such as a spall on a bearing raceway.
• Maximum Value will always be greater than Carpet Value. • Carpet and Maximum Values will increase in amplitude by roughly the
same amount if lubrication breaks down. • If bearing defects are present then Maximum Value will increase or trend
upward faster than Carpet Value, and the separation between the two will increase.
• To keep things simple, Carpet Value normally indicates lubrication condition, and Maximum Value most often relates to bearing damage.
Spike Energy-1
• As bearings, gear teeth, and other machine components wear, they develop microscopic cracks and spalls, which in turn cause bumpier operation.
• The mechanical knocking produces short pulses, or spikes, of vibratory energy that excite component natural frequencies.
• The impacts of microscopic cracks and spalls also excite the natural frequencies of spike energy accelerometers gathering vibration signatures from around the system; – acting as carrier frequencies, they lead the machine defect
frequencies that flutter with them. Impact energies (labeled in acceleration units gSE of spike energy) are registered by the accelerometers as functions of spike amplitude and repetition rate, and are sent on for further analysis.
Spike Energy-2
• Spike energy is a tool which is very sensitive/effective in identifying the unique types of high frequency energy present in repetitive impacting type events.
• An increase in spike energy will usually be one of the earliest indicators of a rolling-element bearing defect.
• Furthermore the distinctive fault frequency associated with the defect (BPFO, BPFI, 2xBSF, FTF) will show up in a clearer and easier-to identify pattern in the spike energy spectrum.
• Usually at it's fundamental frequency as compared to the regular spectrum (usually a haystack of harmonics of fault frequencies in a wide range around 2000Hz)
Spike Energy-3
• Conventional vibration parameters (displacement, velocity, acceleration) typically fall within the linear frequency response range of most transducers, and are therefore fairly easy to measure.
• But spike energy detects frequencies beyond the linear range of most industrial transducers.
• Because mounting methods affect higher frequencies, spike energy results vary with different setups.
• Impact-induced resonant frequencies of industrial accelerometers typically range from 10 to 50 kHz, varying greatly with construction and mounting.
• If two accelerometers had the same frequency response characteristics, it would be a coincidence; thus, spike energy readings made with different accelerometers shouldn’t be compared.
• Because of spike energy’s great sensitivity to setup, the most meaningful way to use spike energy for machinery condition monitoring applications is to observe trends in the returned signal.
• For consistency, the same accelerometer, mounting method and measurement location should be used throughout any data collection.
Peakvue-1
What is Peakvue? Peakvue is a technology unique to Emerson and means ‘Peak
Value’ Such as the Peak Value of an impact generated by a bearing
defect in a time waveform - (True Peak Value) If you have a 21centuray analyzer you have the capability to
acquire ‘Peakvue Data’ The ‘True Peak Value’ is obtained by concentrating on ‘Stress
Wave Analysis’ rather than conventional vibration data. These stress waves travel further than conventional vibration
signals so a truer indication of fault severity is obtained.
Peakvue-2
What is a Stress Wave? Stress waves accompany metal-metal impacting. These stress waves are short-term (fractional to a few milliseconds)
transient events, which introduce a ripple effect on the surface machinery as they propagate away from the initial event.
If you think of a stone being dropped into a pool of water. The stone is the initial impact generated by the fault. The effect of the stone being dropped into the water cause a ripple on the
surface of the water which, spreads over a wide area.
Initial Impact
Peakvue-3
If a bearing has a sub-surface defect (early bearing wear), when a rolling element passes over the defect it bends the race slightly and then as the rolling element passes it restores back to it’s natural state.
This event causes a high frequency (1-50KHz) short duration stress wave.
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-4
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-5
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
1. Conventional Vibration Signals that are filtered from the Peakvue Signal
· Imbalance· Misalignment· Gears· Bearings· Resonance
2. Peakvue filter removing low frequency noise from the stress wave data
· This is to prevent low frequency noise consuming the stress wave activity
3. High frequency stress wave activity occurring in the 1000Hz - 20000Hz frequency range at a rate governed by a low frequency event
· Bearings· Gears
Peakvue-6
There are two types of filters available
Band Pass FiltersThe band pass filter removes all the data above and below the filter corner values
High Pass FilterThe high pass filter removes all data lower in frequency to that of the filter selection allowing only the high frequency stress waves to pass through
After the filtering process what should remain is the high frequency stress wave activity that is occurring at the rate of the excitation – such as from a bearing.
f
f
Peakvue-7
A comparison can be made of the sampling to
show how data is collected through both
methods of data acquisition, normal and
Peakvue.
FFT
High
Pass
Filter
Full
Wave
Rectify
Digital
Peak
Impact
Detection
Vibration Signal
Peakvue-8
The diagram below shows sampling of data using normal data collection.
Instantaneous Samples
Stress wave- this is missed under normal conditions
Peakvue-9
• The diagram below shows sampling of data using Peakvue™ data collection.
Peakvue Samples
Stress wave- this is missed under normal conditions
Peakvue-10
• Peakvue measures the highest amplitude found in a stress waves (Pk Value) and holds that data
• The waveform data is then passed through a high pass filter to remove the unwanted, low frequencies– Imbalance, Misalignment, Looseness, resonance etc.
• This just leaves us with the high frequency impacting data (Peak) above the machine noise level
• The data is then brought back to fundamental frequency. (this allows analysis of the data to be done quicker and easier)
• The waveform should contain enough time to include at least 15 shaft revolutions to resolve cage frequency in the spectrum for rolling element bearings.
(The waveform time length is determined by the lines of resolution divided by the f-max)
Peakvue-11
• The F-max should be set at least 3 or 4 times the highest expected defect frequency (usually inner race defect for rolling element bearings)
• One average should be used when taking PeakVue data
• Transducer mounting should be consistent for trend able data.
• At minimum the surface should be clean (free of paint, dirt, etc.), stress waves are easily attenuated.
Peakvue – Spectrum-1
• Here is a typical Peakvue spectra plot.
• This is typically a GOOD spectrum
1. Broad band energy - Filtered Noise
2. Units should be ‘acceleration’ (Very high frequency analysis)
3. Amplitude values are low. Severity of fault is not determined in the spectra
Peakvue – Spectrum-2
• This is a Peakvue spectrum where high frequency stress waves are being detected
1. Broad band energy - Filtered Noise
2. Units still in ‘acceleration’ (Very high frequency analysis)
3. Amplitude values are low. Remember severity of fault is not determined in the spectra
Notice the Impacts passing through the filtered noise
• This is indication of a fault developing
Peakvue (Spectrums and Waveforms Diagnostics Techniques-1)
• Shown below is a typical Peakvue spectrum with a defect present
Good Spectrum will show only a noise level
Noise removed by filter
Stress waves are showing clearly in the data at 4.6 Orders
The filter used is shown in the top right hand corner
Peakvue (Spectrums and Waveforms Diagnostics Techniques-2)
• 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
Route Waveform 09-Jul-03 09:50:49 (PkVue-HP 1000 Hz) RMS = 2.97 PK(+) = 8.35 CRESTF= 2.81
0 4 8 12 16 20 24 28 32 36
0
1
2
3
4
5
6
7
8
Revolution Number
Acc
eler
atio
n in
G-s
B42 - ZONE 5 DF FAN 116/16EXT01-M2P Motor Inboard Horz Peakvue
Label: Bearing Fault - BPFO NTN6217
Route Spectrum 09-Jul-03 09:50:49 (PkVue-HP 1000 Hz) OVERALL= 1.37 A-DG RMS = 1.37 LOAD = 100.0 RPM = 1342. (22.37 Hz)
0 200 400 600 800 1000
0
0.1
0.2
0.3
0.4
0.5
0.6
0.70.8
Frequency in Hz
RM
S A
ccel
erat
ion
in G
-s
Freq: Ordr: Spec:
1.250 .05587 .01367
>NTN 6217 N=BPFO -OB
N N N N N N N N N
For Peakvue analysis
Use the Spectrum
• Diagnose the defect
Use the Waveform
• Determine the severity
Peakvue (Spectrums and Waveforms Diagnostics Techniques-3)
• Waveforms can be confused with spectrums, as the waveform is only plotting the peak value and does not show a full wave.
A1 - Example 1EX 1 -D3P Tail Roll Non D/S Peakvue
Label: Easy
Analyze Waveform 16-Mar-01 12:03:14 (PkVue- HP 500 Hz)
PK = .0556 LOAD = 100.0 RPM = 80. RPS = 1.33
PK(+) = .5599 PK(-) = .0397 CRESTF= 14.25
0 3 6 9 12
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
Revolution Number
Acc
eler
atio
n in
G-s
2. Peak Value Impacts
1. Filtered Noise Level
4. Acceleration as default units
3. No Peak Negative Value
Peakvue (Spectrums and Waveforms Diagnostics Techniques-3)
• Diagnosing a Peakvue spectrum and waveform is not to dissimilar to that of conventional data.
• However there are a few differences which can be a bit confusing at first, these are:
1. Do not try to locate 1xTurning Speed, as this is low frequency data and will be filtered out.
• Turning speed should be entered using the conventional spectral data.
2. Multiple harmonics are often present within a spectrum due to the way peakvue samples the data.
• These do not indicate ‘Looseness’
Peakvue (Spectrums and Waveforms Diagnostics Techniques-4)
3. Spectral amplitudes are always low in amplitude but should not be used to judge severity. Use the spectrum to diagnose the fault.
4. Waveforms indicate the severity of the problem.
5. Ensure the same filter setting is used in both the spectrum and waveform.
Potential faults can be missed or overlooked if different filters are used.
6. Cage Defects show up well in peakvue data and is normally an indication the bearing is under stress.
7. All low frequency faults are removed from the data and will not be seen in a Peakvue spectrum and waveform
Imbalance, Misalignment, Looseness, Resonance - All Gone.
Peakvue (Spectrums and Waveforms Diagnostics Techniques-5)
ANALYZE WAVEFORM 16-Mar-01 12:03:14 (PkVue- HP 500 Hz) PK = .0556 PK(+) = .5599 PK(-) = .0397 CRESTF= 14.25
0 3 6 9 12
-0.1
0
0.1
0.2
0.3
0.4
0.50.6
Revolution Number
Acc
eler
atio
n in
G-s
A1 - Example 1EX 1 -D3P Tail Roll Non D/S Peakvue
Label: Easy
ANALYZE SPECTRUM 16-Mar-01 12:03:14 (PkVue- HP 500 Hz) PK = .0484 LOAD = 100.0 RPM = 80. RPS = 1.33
0 20 40 60 80 100
0
0.004
0.008
0.012
0.016
Frequency in Hz
PK
Acc
eler
atio
n in
G-s
Freq: Ordr: Spec:
7.284 5.463 .01018
1.Spectral data indicating a defect at 5.463 Orders
2. Impacting also being detected at 0.6G-s
3. Very Slow RPM
Peakvue (Spectrums and Waveforms Diagnostics Techniques-6)
ANALYZE WAVEFORM 16-Mar-01 12:03:14 (PkVue- HP 500 Hz) PK = .0556 PK(+) = .5599 PK(-) = .0397 CRESTF= 14.25
0 3 6 9 12
-0.1
0
0.1
0.2
0.3
0.4
0.50.6
Revolution Number
Acc
eler
atio
n in
G-s
A1 - Example 1EX 1 -D3P Tail Roll Non D/S Peakvue
Label: Easy
ANALYZE SPECTRUM 16-Mar-01 12:03:14 (PkVue- HP 500 Hz) PK = .0484 LOAD = 100.0 RPM = 80. RPS = 1.33
0 20 40 60 80 100
0
0.004
0.008
0.012
0.016
Frequency in Hz
PK
Acc
eler
atio
n in
G-s
Freq: Ordr: Spec:
7.284 5.463 .01018
>NSK 6207 F=BPFI -IB
F F F F F F F F F F F F F
4.Fault Frequencies Indicate a BPFI Defect
Peakvue (Spectrums and Waveforms Diagnostics Techniques-7)
• For machines running between speeds of 900 - 3600RPM recommended guidelines for setting initial warning levels in the Peakvue™ time - waveform are as follows:
Alert Value Fault Value
Inner Race 3.0g's 6.0g's
Outer Race 6.0g's 12.0g's
Rolling elements fault 4.5g's 9.0g's
Cage frequencies If evident then the bearing is usually under stress.
PeakVue-1
Figure 3: Photograph of defective bearing from the inlet of pinion stand gearbox
• The impacting levels trended from 18 g's in July to a high of 37 g's in September.
PeakVue-2
• In this case, normal vibration data did identify the fault; however, the low levels observed did not place the fault at a level of significant concern. The impacting levels identified in PeakVue, excess of 30 g's, raised the concern level and initiated planning for replacement.
Figure 6: Photograph of defective service water pump bearing, showing inner and outer race spalling.
Time Waveforms
You can also look at vibration as the amount of ‘Time’ it takes to complete a particular cycleIf 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 motion is called a sine wave. – The horizontal axis is
measuring Time– The vertical axis is
measuring Amplitude 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
• The plot shown on the right is a complex time waveform.
– Amplitude versus Time• This is just one format (domain)
for analysing vibration data. • Data can also be analysed in a
‘Spectrum’ – (Amplitude Vs Frequency) through a process known as the FFT
A8 - Example 15Ex15 -F2V Fan Outboard Vertical
Route Waveform 22-Aug-02 11:33:16
PK = .1495 LOAD = 100.0 RPM = 832. (13.86 Hz)
PK(+) = .3263 PK(-) = .3572 CRESTF= 3.38
0 50 100 150 200 250 300 350
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
Time in mSecs
Acc
eler
atio
n in
G-s
Time: Ampl:
120.44 -.07595
Los - Example 8EX 8 -P2V Pump Outboard Vertical
Label: Looseness
Analyze Spectrum 15-Nov-95 10:00:16
RMS = 1.27 LOAD = 100.0 RPM = 737. RPS = 12.28
0 6000 12000 18000 24000 30000
0
0.2
0.4
0.6
0.8
1.0
Frequency in CPM
RM
S V
elo
city
in m
m/S
ec
Freq: Ordr: Spec:
736.86 1.000 .245
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 sine's and cosines’.
– Meaning if you take a time waveform and mathematically calculate the vibration frequency,
it can be converted to a more familiar format
Frequency
Am
plit
ude
How the Vibration Spectrum is Created
Time
Amplitude
Time
Am
plit
ude
Frequency Domain
• The frequency domain (Spectrum) plots the data as ‘Amplitude’ in the (Y) axis and ‘Frequency’ in the (X) axis. This data is derived from the time domain – mathematical manipulation of the time waveform.
• Recall the waveform and spectrum from the previous slide. If you tried to determine all the frequencies from the waveform plot, you would need all day just to analyse one point of data.
• As the FFT plots the frequencies from the waveform for you the analysis of this data becomes easier and reduces the amount of time needed for analysis of each point.
A8 - Example 15Ex15 -F2V Fan Outboard Vertical
Route Waveform 22-Aug-02 11:33:16
PK = .1495 LOAD = 100.0 RPM = 832. (13.86 Hz)
PK(+) = .3263 PK(-) = .3572 CRESTF= 3.38
0 50 100 150 200 250 300 350
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
Time in mSecs
Acc
eler
atio
n in
G-s
Time: Ampl:
120.44 -.07595
Los - Example 8EX 8 -P2V Pump Outboard Vertical
Label: Looseness
Analyze Spectrum 15-Nov-95 10:00:16
RMS = 1.27 LOAD = 100.0 RPM = 737. RPS = 12.28
0 6000 12000 18000 24000 30000
0
0.2
0.4
0.6
0.8
1.0
Frequency in CPM
RM
S V
elo
city
in m
m/S
ec
Freq: Ordr: Spec:
736.86 1.000 .245
Los - Example 3EX3 -P2V Pump Outboard Vertical
Analyze Spectrum 15-Nov-95 10:00:16
RMS = 1.27 LOAD = 100.0 RPM = 737. RPS = 12.28
0 6000 12000 18000 24000 30000
0
0.2
0.4
0.6
0.8
1.0
Frequency in CPM
RM
S V
elo
city
in
mm
/Sec
Freq: Ordr: Spec:
736.86 1.000 .245
Here the primary cursor is at 1 Order (1xTs). All the other cursors are harmonics (exact multiples of the primary cursor)
Harmonic - Orders
• Harmonics are cursors that are exact multiples of the primary frequency– They are used to locate other frequencies related to the primary cursor
• Therefore:– When the primary cursors is located on 1Order all the harmonics will be
synchronous– Harmonic cursors can be used to show non-synchronous and sub-
synchronous harmonics depending upon the energy of the primary frequency
Energy in the Spectrum
C1 - Example 4E4 -MOH MOTOR OUTBOARD HORIZONTAL
Route Spectrum 09-Feb-00 12:41:33
OVRALL= .5785 V-DG RMS = .5716 LOAD = 100.0 RPM = 2937. RPS = 48.95
0 20 40 60 80 100 120 140 160
0
0.1
0.2
0.3
0.4
0.5
Frequency in kCPM
RM
S Ve
loci
ty in
mm
/Sec
Freq: Ordr: Spec:
2.937 1.000 .01038
Synchronous Energy
• Synchronous energy - related to turning speed.
• All the other peaks are harmonics off, which means they are related to the first peak
• We can see from the spectrum that the first peak is at 1 Orders (which means it is 1 x turning speed)
Examples of synchronous energy:1) Imbalance 2) Misalignment 3) Gearmesh
Los - Example 8EX 8 -P2V Pump Outboard Vertical
Label: Looseness
Analyze Spectrum 15-Nov-95 10:00:16
RMS = 1.27 LOAD = 100.0 RPM = 737. RPS = 12.28
0 6000 12000 18000 24000 30000
0
0.2
0.4
0.6
0.8
1.0
Frequency in CPM
RM
S Ve
loci
ty in
mm
/Sec
Freq: Ordr: Spec:
736.86 1.000 .245
Non-Synchronous Energy
• Non-synchronous energy - not related to turning speed
• We can see from the spectrum that the first peak is at 10.24 Orders. This is not related to turning speed.
• Examples of non-synchronous energy:• Bearings Multiples of belt frequency Other Machine Speeds
BF - Example 5E5 -R4A ROLL BRG. #4 - AXIAL
Label: Outer Race DefectPriority: 1
Route Spectrum 12-Jul-96 17:16:42
OVRALL= 2.63 V-DG RMS = 2.69 LOAD = 100.0 MPM = 3225. RPM = 380.
0 6000 12000 18000 24000 30000
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Frequency in CPM
RM
S V
elo
city
in m
m/S
ec
Freq: Ordr: Spec:
3888.9 10.24 .748
Sub-Synchronous Energy
• Sub-synchronous energy - Less than turning speed
• The spectrum shows the first impacting peak below 1 Order. This is sub-synchronous energy
• Examples of sub-synchronous energy are:• Belt Frequencies• Other Machine Speeds• Cage Frequencies
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
Energy in a Spectrum
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:
– Imbalance
– Misalignment
– Looseness
– Vane pass frequency
– Gears etc
Los - Example 8EX 8 -P2V Pump Outboard Vertical
Label: Looseness
Analyze Spectrum 15-Nov-95 10:00:16
RMS = 1.27 LOAD = 100.0 RPM = 737. RPS = 12.28
0 6000 12000 18000 24000 30000
0
0.2
0.4
0.6
0.8
1.0
Frequency in CPM
RM
S V
elo
city
in m
m/S
ec
Freq: Ordr: Spec:
736.86 1.000 .245
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
– Electrical
– System resonances
– Multiples of belt frequency
BF - Example 5E5 -R4A ROLL BRG. #4 - AXIAL
Label: Outer Race DefectPriority: 1
Route Spectrum 12-Jul-96 17:16:42
OVRALL= 2.63 V-DG RMS = 2.69 LOAD = 100.0 MPM = 3225. RPM = 380.
0 6000 12000 18000 24000 30000
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Frequency in CPM
RM
S V
elo
city
in m
m/S
ec
Freq: Ordr: Spec:
3888.9 10.24 .748
Data Acquisition Principles
What is Data Acquisition
The sampling of the real world to generate data that can be manipulated by a computer
Data Acquisition System
Data Acquisition System(1)
• Sensor/transducer for measurement of physical variables.
• Signal-Conditioner/transmission circuitry, that enables conversion of signal outputs from transducers to a readable form for Data Acquisition/interface modules.
• The Data Acquisition Hardware comprising Multiplexer, Amplifier, A/D Converter, Buffer Memory, etc. to digitize the analog signals for CPU.
• Computer/CPU to process the digital data for data processing, display, outputs (control), storage, transmission, etc.
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.
Transducer Types
Seismic:- Bearing relative to space.• Velocity Pickups • Accelerometers• Piezoelectric velocity pickups
Relative:- Shaft relative to bearing.• Non-contact Eddy Current Displacement Probes
Absolute:- Shaft relative to space.• Shaft Contact Displacement Probes (including Shaft Sticks and Shaft Riders)
Vibration Transducers
Vibration Transducers
Sensors…Transducers…Probes…What is it?
….It basically converts mechanical vibration to an electrical signal
AccelerometerCharge Type &Line DriveConstant Voltage &Constant Current
Velocity Probe DisplacementShaft Riders
Proximity Probes(Eddy Current Probes)
Velocity Sensors
Accelerometers
Displacement Probes (non-contact, eddy current probes)
Seismic Transducer
VELOCITY PICKUP
Velocity Pickups
Note :- There are two types of velocity pickups the above advantages do not apply to piezoelectric velocity transducers.
ADVANTAGES
Self Generating – no power supply required Magnet inside coil generates velocity proportional to vibration Spring mass system 10 Hz. to 1000 Hz. Phase change 900 Directional mounting Large & Heavy Output = mV/inch/sec Wide range of available outputs
Piezoelectric Velocity Pickup
ADVANTAGES
Remember everything that you just learned about an accelerometer
The output of the accelerometer has been integrated to velocity and has a 900 phase change
100 mV/inch/sec (4 mV/mm/sec)
500 mV/inch/sec (20 mV/mm/sec)
SESMIC TRANSDUCERS
ACCELEROMETERS
AMPAMP
IEPE– Internal Amplifier– Industrial
Charge Mode– External Amplifier– High Temperature
Accelerometers - advantages
• No moving parts, no wear.
• Rugged.
• Very large dynamic range.
• Wide frequency range.
• Compact, often low weight.
• High stability.
• Can be mounted in any orientation.
• Measures casing vibration
• Measures absolute vibration
• Integrate to Velocity
• Easy to mount
• Large range of frequency response
• Available in many configurations
Accelerometer Types
The three most common are :-• Compression Type • Inverted Compression Type• Shear Type
Compression type accelerometer
Electric connector
Seismic Mass
Preload Stud
Acoustic Shield
Piezoelectric Material
ICP AmplifierMounting StudReceptacle
Base
Compression Type Accelerometers
Advantages• Relatively low cost
Disadvantages• Sensitive to base strain• Sensitive to Thermal transients• Can cause over-saturation and transducer settling problems
Widely used
Inverted compression type
Piezoelectric MaterialICP Circuit
Mounting stud receptacle
Seismic Mass
Preload Sleeve
Shear Type Accelerometer
Electric connector
Seismic Mass
Post
Acoustic ShieldPiezoelectric Material
ICP CircuitMounting StudReceptacle
Base
Advantages - Shear Type
• Lower sensitivity to base strain• Large dynamic range • Much less sensitive to temperature transients• Stabilizes quickly when taking measurements at low frequencies.
Disadvantage: -• Generally higher cost due to added components
Typical Accelerometer Parameters/Specifications
Typical Accelerometer Frequency Response
Frequency Response & Mounting Technique
Sen
siti
vity
Freq
StudMount
HandProbe
Dual RailMagnet
FlatMagnet
MountingPad
1.5KHz 10KHz 32KHz
Frequency Response & Mounting Technique (1)
Frequency Response & Mounting Technique (2)
What is the frequency range of your…Instrument…Cables …Sensor …Sensor Coupling
What is the fault frequency you are looking for ?
Sensor freq = 12 KHz
Instrument freq= 80 KHz
Cable length ?
Sensor coupling ?
Frequency Range, Sensitivity & Application
Frequency Range, Sensitivity & Application (1)
Non-contact Eddy Current Probe (Relative)
PICKUPCOIL
MAGNETIC FIELD
SHAFT
GAPMETER
DC SIGNAL SENSOR
DISPLACEMENTSIGNAL - TO ANALYSER ORMONITOR
NON-CONTACT PICKUP
DETECTOROSCILLATOR AMPLIFIER
Non-contact Pickups
Non-contact Eddy Current Displacement Probes
USED FOR:-• Relative Shaft Vibration.• Radial & axial shaft position.• Differential expansion between case and rotor.
Especially effective on machinery with high mass rigid casings and relatively low mass rotors supported in journal type bearings.
N.C.P Problems & precautions
• Only Measures Displacement - Sensitive Only to low frequency defects.
• Subject to Mechanical and Electrical Run-out .
• Units must be pre calibrated for specific shaft materials.
Shaft Contact Displacement Probes (Absolute)
• Shaft Sticks• Hardwood, fish-tail, fixed to accelerometer or velocity pickup• Measures vibration amplitude & phase• Shaft Riders
SHAFT SURFACE
NON-METALLIC TIP
MACHINE HOUSING
SHAFT RIDER ASSEMBLY
PICKUP MOUNTING STUD
Shaft Rider
Direct Contact : Absolute Measurements
• Shaft Riders (permanently installed)• Shaft Sticks or Fishtails
– safety issue– very useful below coupling of vertical pumps
Radial CasingVibration
Radial ShaftVibration & Position
Axial ShaftVibration & Position
Typical Uses of Vibration Transducers
Measurement Parameter
• Find the “flattest” spectrum
• Normally velocity is used
• For very slow running machine (<
600 RPM) displacement is
preferred.
• For High frequency diagnostics use
acceleration
• Always use acceleration for
Envelope analysis.
Acceleration
Velocity
Displacement
Monitoring Techniques
Acceleration
Velocity
Displacement
Displacementaccentuates LOW frequencies,and attenuates HIGH frequencies.
Velocity“flat” treats all frequencies equally.
Accelerationaccentuates HIGH frequencies,and attenuates LOW frequencies.
Freq
Vib
Easy to install Good for detecting high
frequency faults No moving parts Good dynamic/frequency range Small/light weight Withstands high temperatures
Needs double integration to displacement
Needs external power source Provides limited information on shaft
dynamic motion Not good for slow speed machines
Measures directly shaft motion Same transducers for axial thrust,
speed and radial vibration Measures directly in
displacements units Measures DC (shaft position) No moving parts
Pro
xim
ity
Pro
bes
Acc
eler
omet
ers
Advantages Disadvantages
Runout problems Sensitive to shaft materials Installation Limited freq. range. No detection of
rolling element bearing faults Temperature restrictions External proximitor needed
Comparison of Transducers
Gear
Blades
Rolling ElementBearings
ShaftRotatingSpeed
2x
3x
JournalBearingsinstability
1 KHz 3KHz 25KHzNon Contact Displacement
Velocity Probe
Accelerometer
Vibration Pickups
Analog to Digital Convertors
ADC Analog-to-Digital Converters(1)
ADC (Analog to Digital Converter) is an electronic device that converts a continuous analog input signal to discrete digital numbers (binary)
Analog– Real world signals that contain noise– Continuous in time
Digital– Discrete in time and value– Binary digits that contain values 0 or 1
ADC Analog-to-Digital Converters(2)
All microcontrollers store information using digital logic
Compress information to digital form for efficient storage
Digital data transfer is more efficient
Provides a link between real-world signals and data storage
ADC Analog-to-Digital Converters(2)
How ADC Works
Sampling – Sample-Hold Circuit– Aliasing
Quantizing and Encoding– Resolution
Binary output
ADC Analog-to-Digital Converters(2)
Sampling
Reduction of a continuous signal to a discrete signal
Achieved through sampling and holding circuit
Switch ON – sampling of signal (time to charge capacitor)
Switch OFF - voltage stored in capacitor (hold operation)
Must hold sampled value constant for digital conversion
ADC Analog-to-Digital Converters(2)
Sampling (cont’d)
Sampling rate depends on clock frequency
Use Nyquist Criterion
Increasing sampling rate increases accuracy of conversion
Possibility of aliasing
Sampling Signal:
Sampling Period:
Nyquist Criterion:
ADC Analog-to-Digital Converters(2)
Aliasing
High and low frequency samples are indistinguishable
Results in improper conversion of the input signal
Usually exists when Nyquist Criterion is violated
Can exist even when:
Prevented through the use of Low-Pass (Anti-aliasing) Filters
max2 ffs
ADC Analog-to-Digital Converters(2)
Quantizing and Encoding Approximates a continuous range of values and replaces it with a
binary number Error is introduced between input voltage and output binary
representation Error depends on the resolution of the ADC
ADC Analog-to-Digital Converters(2)
V
resolutionQerror
5.
2/
Resolution Maximum value of quantization error Error is reduced with more available memory
Vrange=Input Voltage Range
n= # bits of ADC
)12/( nrangeVresolution
Example:
)12/(71
3
0.7
3
VV
n
VVrange
Resolution
ADC Analog-to-Digital Converters(2)
Resolution (cont’d) Increase in resolution improves the accuracy of the conversion
Minimum voltage step recognized by ADC
Analog Signal Digitized Signal- High Resolution
Digitized Signal- Low Resolution
ADC Analog-to-Digital Converters(2)
• Dynamic Range– Usually defined in dB, depends on the number of
bits used by the ADC• For example, a 12 bit ADC has 212 possible data
values, or 4,096 “steps” between the lowest and highest values the ADC can see (0 to 5 Volts, typ.)
• 8-bit is 256 steps• 16-bit is 65,536 steps, so more is better, right?
ADC Analog-to-Digital Converters(4)
• Dynamic Range– For a 12 bit ADC…20 log (4095/1) = 72 db
• Theoretical only, electronic noise reduces to 65 db
– For a 16 bit ADC…20 log (65536/1) = 96 db• Electronic noise may make this only 80 db
• Massively more data to manipulate w/o much practical gain in Dynamic Range.
Selection of ADC (Analog-to-Digital Converters)
Important Considerations– Input Type – Differential or Single Ended– Resolution - Most Important– Scaling - Allows the user to divide or multiply the input voltage to
more closely match the full scale range of the ADC– Sample Rate - The sample rate must be at least twice the
frequency the you are measuring, but 5 times is much better– Channel Scan Rate - The channel scan rate is the maximum rate
that the ADC can select a new channel and make a measurement. many ADCs have a relatively slow scan rate (when compared to the sample rate.)• To achieve a sample rate of 600Hz on three channels, you will
need a channel scan rate of at least 1.8kHz
Signal Processing
Raw Signal
Amp
ACOutput
Integrator1x, 2x
High PassFilter
Low PassFilter
DCOutput
Amp DetectorP-P or RMS
DisplayReading
Accelerometer
Signal Processing-Flow Chart
Signals & Signal Processing
A signal is a function of independent variables such as time, acceleration, position, velocity and displacement etc.
A signal carries information, and the objective of signal processing is to extract useful information carried by the signal.
Signal processing is concerned with the mathematical representation of the signal and the algorithmic operation carried out on it to extract the information present.
Stationary Signals
Non - Stationary Signals
- Vibration from rotating machines
- Vibration from reciprocating machines (short term)
- Vibration from run-ups and coast-down
Machine Signal Types
Time Signal
Absolute Vibration with Free-Space
Machine Vibration Signal
AC Signal
DC Signal
Relative Vibration with mounting position of Prox. Probe
Machine Vibration Signal
Frequency Filtering
Filters
A filter is a function that in the frequency domain has a value close to 1 in the range of frequencies that the analyst wishes to retain and close to zero in the range of frequencies that the analyst wishes to eliminate.
The filter can be applied in the time domain or in the frequency domain but their function is best understood in the frequency domain
If the filter is a mechanical or electrical device which operates on the continuous time physical signal it is called an analogue filter.
If the filter is a numerical algorithm or mechanical device which operates on sampled data it is called a digital filter.
High-Pass filters - As the name imply, a high pass filter allows high frequencies to pass. (lower frequency limit)
Low-Pass filters - Allow low frequencies to pass through (upper limit)
Bandpass filters - Allows only frequencies within the band
Anti-aliasing filters - Low pass filter at half the sampling frequencies
Types of filters:
f
f
Frequency Filtering
Fast Fourier Transform (FFT) • Fast Fourier Transform (FFT) is a method of taking a real world, time-varying
signal and splitting it into components, each with an amplitude, a phase, and a frequency.
• In modern instruments today, the FFT is more commonly used to provide frequency domain.
• All waveforms, no matter how complex, can be expressed as the sum of sine waves of varying amplitudes, phase, and frequencies information.
FFT - Fast Fourier Transform is an efficient means of transforming a time signal into a frequency spectrum.
1. Aliasing - high frequencies appearing as low frequencies
2. Leakage - Memory contents forced to be periodic. Can give discontinuities when ends joined
3. Picket fence effect – Actual spectrum sampled at discrete frequencies, peaks may be missing
1. Aliasing - high frequencies appearing as low frequencies
2. Leakage - Memory contents forced to be periodic. Can give discontinuities when ends joined
3. Picket fence effect – Actual spectrum sampled at discrete frequencies, peaks may be missing
FFT - Pitfalls
Sampling rate too slow
High frequency analysis results in false low frequency signal
Solution: Use Anti-aliasing filterTypically a 1K (1024 point) transform, 512 frequency components are calculatedand 400 lines displayed. Similarly a 2K transform 800 lines are displayed.
FFT pitfalls - Aliasing Effect
1st Sample
2nd Sample
-ve
-ve
+ve
+ve
…..give discontinuities when ends joined
Use Hanning Window
FFT pitfalls - Leakage
ActualSpectrum
MeasuredSpectrum
FFT pitfalls - Picket Fence Effect
Fast Fourier Transform
• To under stand the FFT digital sampling process ,we must have the under standing of:
– Fmax– Number of Averages– Number of Lines– Average Type– Percent Overlap– Low Frequency Corner– Window Type
Sampling Rate
Sampling Rate: the sampling rate (SR) is the rate at which amplitude values are digitized from the original waveform.
– High-quality sampling rate :
SR = 44,100 samples/second– Medium-quality sampling rate:
SR = 22,050 samples/second– Low-quality sampling rate :
SR = 8,192 samples/second
Higher sampling rates allow the waveform to be more accurately represented
Nyquist Theorem and Aliasing
Nyquist Theorem:
We can digitally represent only frequencies up to half the sampling rate.
– Example:
SR=44,100 Hz
Nyquist Frequency = SR/2 = 22,050 Hz
– Example:
SR=22,050 Hz
Nyquist Frequency = SR/2 = 11,025 Hz
Nyquist Theorem and Aliasing
• Frequencies above Nyquist frequency "fold over" to sound like lower
frequencies.
This fold-over is called aliasing.
• Aliased frequency f in range [SR/2, SR] becomes f ': f ' = |f - SR|
f ' = |f - SR|– Example:
SR = 20,000 Hz
Nyquist Frequency = 10,000 Hz
f = 12,000 Hz --> f ' = 8,000 Hz
f = 18,000 Hz --> f ' = 2,000 Hz
f = 20,000 Hz --> f ' = 0 Hz
Hertz
Nyquist Theorem and Aliasing
Graphical Example 1a:
– SR = 20,000 Hz– Nyquist Frequency = 10,000 Hz– f = 2,500 Hz (no aliasing)
Graphical Example 1b:
– SR = 20,000 Hz– Nyquist Frequency = 10,000 Hz– f = 5,000 Hz (no aliasing)
(left and right figures have same frequency, but have different sampling points)
Nyquist Theorem and Aliasing
Graphical Example 2:
– SR = 20,000 Hz
– Nyquist Frequency = 10,000 Hz
– f = 10,000 Hz (no aliasing)
Graphical Example 2:
– BUT, if sample points fall on zero-crossings the sound is completely cancelled out
Nyquist Theorem and Aliasing
Graphical Example 3:
– SR = 20,000 Hz
– Nyquist Frequency = 10,000 Hz
– f = 12,500 Hz, f ' = 7,500
Graphical Example 3:
Fitting the simplest sine wave to
the sampled points gives an aliased
waveform (dotted line below):
Fmax
• Highest frequency captured and displayed by the instrument.
• In choosing the Fmax, we also set other parameters. One of these is called the anti-aliasing filter.
Lines of Resolution• ‘Lines of resolution’ determine the clarity of the spectral data • Typical values are 100, 200, 400, 800, 1600, 3200, 6400, and 12,800.• Each line will cover a range of frequencies (bin), and the resolution of each
line can be calculated simply by dividing the overall frequency (Fmax) by the number of lines.
• For example, an Fmax of 120,000 CPM and 400 lines gives a resolution of 300 CPM per line.
L2 - TA 16TA16 -M1H Motor Outboard Horizontal
Analyze Spectrum 13-Mar-01 09:13:53
PK = .7078 LOAD = 100.0 RPM = 1496. RPS = 24.94
0 400 800 1200 1600
0
0.1
0.2
0.3
0.4
0.5
Frequency in Hz
PK A
ccel
erat
ion
in G
-s
L2 - TA 16TA16 -M1H Motor Outboard Horizontal
Analyze Spectrum 13-Mar-01 09:14:16
PK = .3852 LOAD = 100.0 RPM = 1497. RPS = 24.95
0 400 800 1200 1600
0
0.04
0.08
0.12
0.16
0.20
Frequency in Hz
PK A
ccel
erat
ion
in G
-s
Lines of Resolution
• The spectrum shown displays data at 800 L.O.R with an Fmax of 1600 Hz
The second spectrum displays the same data but with 3200 L.O.R over the same Fmax
Bandwidth
BW = Fmax / LOR
Energy is summed up within a Bin and plotted at the centre frequency
Cen
tre Freq
uen
cy
Bandwidth
The Bandwidth can be defined by:
(Frequency Span / Analyzer Lines) Window Function
Uniform Window Function = 1.0
Hanning Window Function = 1.5
Flat Top Window Function = 3.8
Example: 0 to 400 Hz using 800 Lines & Hanning Window
Answer = (400 / 800) 1.5 = 0.75 Hz / Line
DATA CAPTURE TIME
– As the parameters Fmax and lines of resolution are selected, the total sample time for
capturing valid FFT data is determined.
– For a 400-line FFT, due to the calculations involved, we need to take 1024 points on
the waveform. This number (N = 2.56*(#lines)) is
derived from the following calculations:
– Bandwidth (BW) = Fmax/(#lines) ; T(obs) = 1/BW = (#lines)/Fmax
– T(obs) = N*T(sample) = N*(1/(2.56* Fmax)) ; N = 2.56*(#lines)
Where:
(#lines) = total number of lines of FFT resolution
– Fmax = highest analyzed frequency (Hz.)
– N = number of samples collected
– T(sample) = sample period (sec.)
– T(obs) = observation time (sec.).
DATA CAPTURE TIME
If we assume we want an Fmax of 120,000 CPM and 400 lines of resolution,
we can now determine how long our sampled time waveform must be.
• To avoid aliasing, a low pass filter of 120,000 CPM is selected
• To avoid aliasing, we sample at 307,200 CPM (=2.56 x 120,000).
• There are 1024 samples to yield 400 lines of resolution
The section of time waveform observed will be 1024 samples at a sample time of 2
msec., for a total of 0.2 sec. Thus, we need an instrument with
at least 5 KHz sampling rate (1024 samples in 0.2 secs = 5120 samples/sec).
As another example, a 400 line FFT with an Fmax of 6000 CPM would
require an observed time waveform calculated as follows:
– T(obs) = NxT ( sample ) = N/2.56xFmax = 1024/2.56x100 Hz.
– = 1024x(1/256) = 4 seconds.
DATA CAPTURE TIME
• To illustrate the relationship between the length of the time waveform we
need to observe and the resolution achieved, consider how you would need
to examine a signal made up of two waveforms with very close frequencies.
• If the waveforms started off in phase, it would take a long time before they
separated enough to show their different frequencies.
• For example, this can be heard as "beats" when two machines run at nearly
the same speed.
• The bottom line is: In order to achieve high resolution in the frequency
domain, long sample times are required.
Averaging
• Instrument uses a digitized time waveform and performs the mathematical
operation to produce FFT.
• Observing only one section of time waveform may exclude some peak caused
by a random vibration influence.
• To minimize this, it is common to look at several sections of the time
waveform, calculate several FFTs, and display an average result.
• Four averages are commonly taken.
• Averaging provides more repeatable results in data collection for early
warnings of machine deterioration.
• Types of averaging include: linear, exponential, peak hold averaging and etc.
Averaging
• Linear Averaging: In linear averaging, each instantaneous spectrum is added to the next and the sum
is divided by the total number of spectra. This method is useful in obtaining repeatable data for fault
trending, as used in most predictive maintenance programs.
• Peak-hold Averaging: Peak hold is not a true averaging method. During sampling time, the peak value
registered in each analysis cell is captured and displayed. This method is very useful in viewing
transients or for stress analysis calculations
• Exponential Averaging: This technique takes the most recent spectrum collected and weighs it more
heavily than the past data. It is useful in observing conditions that are changing slowly with respect to
sampling time i.e., a steady-state process.
• Synchronous Averaging: This method utilizes a synchronizing signal from the machine being
analyzed. The synchronizing signal is usually derived from a photocell, electromagnetic pickup, or
some other form of tachometer input The vibration input is sampled at precisely the same moment
with respect to shaft rotation during the averaging time period. This method can prove to be a useful
tool for filtering out random background vibrations.
Overlap Averaging
• When more than 1 average is used to calculate the FFT, it is possible to use overlapping
samples, as shown in Figure below:
• This works well since the first part and last part of the sample have their amplitudes reduced in
normal averaging, while the overlapping sample takes full readings at these positions.
• The reduction in accuracy is very small, and for FFTs with a low Fmax and a lot of averages,
collection times can be reduced considerably.
• For example, an FFT with 400 lines, an Fmax of 6000 CPM, and 8 averages without
overlapping takes 32 seconds to gather the samples. With 50% overlap averaging, sampling
requires only 18 seconds.
Windowing and Leakage
• FFT based measurements are subject to errors from an effect known as
leakage.
• This effect occurs when the FFT is computed from of a block of data
which is not periodic.
•To correct this problem appropriate windowing functions must be
applied.
• The user must choose the appropriate window function for the specific
application.
• When windowing is not applied correctly, then errors may be introduced
in the FFT amplitude, frequency or overall shape of the spectrum.
Windowing and Leakage
What is leakage?
Leakage is caused when the time waveform signal does NOT begin and end at the
same point, introducing spurious frequencies.
The Window or weighting function attenuates the signal towards the edge of the
window – minimizing leakage.
Figure : Comparison of non periodic sine wave and FFT with leakage (left) to windowed sine wave and FFT showing no leakage (right)
Windowing and Leakage
Required to solve “Leakage”
Several Types• Uniform• Hanning – Most Commonly used• Hamming• Blackman-Harris
Why do we use the Hanning Window?
Best compromise between frequency resolution and amplitude accuracy for
steady-state machinery analysis
Uniform or Flat-Top is the best choice for transient machinery analysis.
Windowing and Leakage
• The most common windows and their features are given below. This table
can be used to choose the best windowing function for each application.
Peak
PeaktoPeak
RMS
Avg
Always ask.... Are you measuring RMS or Peak , etc ?? What is the frequency range ?? How much averaging?
Freq. = 1/Time
Freq. = Hz= rev. per second
Machine Freq are function of RPMie. rev. per minute
Bandpass Measurement
RMS
True peak - peak
ATa t dtRMS
T
1 2
0
( )
A Apeak RMS 2 *
For Sine waves only:
a
T = averaging period
RMS
ApeakApeak peak
Apeak
Apeak peak
Detector
Vib
rati
on A
mp
litu
de
?
Freq / Orders?
Lower Freq.limit?
UpperFreq.limit?
No. of lines? Avgs?
Frequency Range Selection?
Time Waveforms
Time Waveforms
• The time waveform is the electrical signal from the sensor.
• It is a trace of the voltage changes as the instantaneous vibration changes from
moment to moment.
• This voltage is graphed with time. Thus the name Time waveform. The waveform
provides a view into exactly how that point is moving or vibrating over time.
Time Waveforms Analysis
• Just like the spectral there are certain patterns and characteristics to look for when conducting waveform analysis.
• Once the characteristics have been identified, the analyst can rule out certain faults
e.g: if the waveform is periodic faults like Looseness, Bearing defects, Cracks could be ruled out.
• Data from the time plot will indicate what type of vibration is present. The five types of vibration are harmonic, periodic, beating, impulsive, or random
Time Waveforms Analysis
Harmonic Vibration Periodic Vibration
Time Waveforms Analysis
Pulsating or Beating Vibration Impulsive Vibration
Time Waveforms Analysis
Random Vibration Asymmetric Vibration
Waveform Analysis
Distortion
Waveform Analysis
Electrical vs Mechanical
Waveform Analysis
Noise
Waveform Analysis
Extended time
Waveform Analysis
Extended time
Waveform Analysis
Low frequency
Waveform - Beats
• A beat is the result of two closely spaced frequencies going into and out of phase• The wideband spectrum will show one peak pulsating up and down• The difference between the peaks is the beat frequency which itself will be present
in the wideband spectrum
• A beat is the result of two closely spaced frequencies going into and out of phase• The wideband spectrum will show one peak pulsating up and down• The difference between the peaks is the beat frequency which itself will be present
in the wideband spectrum
WIDEBAND SPECTRUM
ZOOMSPECTRUM
F1 F2
Crest Factor
• The Crest Factor is equal to the peak amplitude of a waveform divided by the
RMS value. The purpose of the crest factor calculation is to give an analyst a
quick idea of how much impacting is occurring in a waveform. Impacting is
often associated with roller bearing wear, cavitation and gear tooth wear.
• In a perfect sine wave, with an amplitude of “1”, the RMS value is equal
to .707, and the crest factor is then equal to 1.41. A perfect sine wave contains
no impacting and therefore crest factors with a value higher than 1.41 imply
that there is some degree of impacting
Crest Factor
• The Problem with the Fast Fourier Transform (FFT) The definition of the Fast Fourier Transform implies that any signal can be approximated by
the sum of a set of sine waves. Unfortunately, this doesn’t work so well when one has a signal that consists of non-periodic events, impacts or random noise . Both impacts and random noise appear the same in the spectrum although they mean different things in the context of machinery vibration analysis. The crest factor is therefore useful in giving the analyst a quick idea of what is occurring in the time waveform.
Crest Factor
• Comparison of 2 Waveforms In below figures we can see an example of the use of the Crest Factor. The waveform in
figure on left has a crest factor of 3.01. The waveform in figure on right has a crest factor of 1.61. The data in figure on left represents a machine with serious rolling element bearing wear, and the crest factor is relatively high due to the amount of impacting occurring within the bearing. The data in figure on right represents a machine with an unbalance, but no impacting related to bearing wear.
Crest Factor
• Conclusion The Crest Factor is a quick and useful calculation that gives the analyst an idea of how much
impacting is occurring in a time waveform. This is useful information that is lost if one is only viewing a spectrum as the FFT cannot differentiate between impacting and random noise. Impacting in a time waveform may indicate rolling element bearing wear, gear tooth wear or Cavitation. Quite often, the Crest Factor is trended over time in order to see if the amount of impacting is increasing or not.
• Equipment Testing and Diagnostics
Impact Testing (bump tests)
Vibration Analysis
Signal Analysis (diagnostics)Vibration responses of the machine or the structure under investigation aremeasured during operation conditions
System Analysis (bump test)Structure or a machine part is put into vibration by means of known excitation forces, often out of its working environment
What is a bump test
• A bump test is an impact test carried out on a machine or product to excite the structure
• Bump test is the measured response of an impact to an object.• The force of the impact is not controlled or measured.• The response of the object is not controlled, BUT IS MEASURED.
Why do a bump test
Excessive levels of noise or vibration
If problems appear under certain conditions, i.e. different machine
speeds or load conditions.
To excite and measure the natural frequency(s) of an object.
To identify a resonance
To understand a change in mass
To understand a change in stiffness
To understand a change in damping
Identification of modal parameters to compare the experimentally
obtained data with corresponding data obtained by FEM or other
theoretical methods
Bump Test Equipment
An impact hammer with a load cell attached to its head to measure the
input force.
An accelerometer to measure the response acceleration at a fixed point
& direction.
A 2 or 4 channel FFT analyzer to compute FRFs.
Post-processing modal software for identifying modal parameters and
displaying the mode shapes in animation.
Measurement Setup
Measurement Setup(1)
• UNIFORM WINDOW• Take your time – Bump around• Do not over range or clip the input signal• 800 – 1600 lines of resolution• Try some different frequency spans• Only 1 bump for each time record• About 4 averages (depends on noise)
How to do a Bump Test ?
How to do a Bump Test ?
The test normally carried out with the machine switched off. An accelerometer is placed on the part of the structure that is
suspected of causing significant resonant frequencies. The structure is repeatedly hit (gently); and during these impacts, a
measurement is taken from the accelerometer which is recording the responding ring from the structure.
An analysis is made of the responding vibration to measure its frequency content.
If a natural frequency coincides with a running speed of the machine shaft or any of its multiples, then there is a chance that high noise & vibration will occur.
A harmonic cursor will assist in displaying where these frequencies occur in the spectrum.
Avoiding Bad Data
• The magnet should be firmly screwed onto the accelerometer. Any looseness between the magnet and accelerometer will corrupt the reading.
• Gently slide the accelerometer onto the measurement position.
• The magnet should be in firm contact with machine’s surface. Any movement of the magnet will be falsely recorded as vibration data. Try sliding/rotating the magnet until a firm seating is achieved.
• Avoid disturbing the accelerometer while taking the measurement.
What do you Impact with?
What do you impact with ?
Energy Value vs. Frequency
• The item used to deliver the impact to the object under test will determine the energy that is delivered to the object.
– Large objects with considerable mass should be impacted with rubber or wood. This will generate high energy low frequency responses. (cow plops)
– Small objects with considerable stiffness should be impacted with metal or hard plastics. This will generate low energy high frequency responses. (pin drops)
How does it work ?
• Bump testing or impact testing works because the bump or impact contains all of the individual frequencies or sign waves.
• When you bump or impact the object under test, you will excite all of the natural frequencies of that object.
How does it work ?
Sine Waves
Bumps from Sine Waves ?
Fundamental
2nd Harmonic
3rd Harmonic
4th Harmonic
5th Harmonic
10th Harmonic
20th Harmonic
50th Harmonic
100th Harmonic
Why the Uniform Window ?
What to Bump (example)
• 1” diameter steel round stock • 36” length• Clamped in “V” blocks at each end• Accelerometer stud mounted on center (100
mV/g)
Bump It ! Two Responses
Bump Test-Application
Stator Generator End Winding Testing
150 100 50 0 50 100 150
30°
60°120°
150°
210°
240° 300°
330°
Turbine Blade Testing
Autospectrum(Signal 2) - InputWorking : Input : Input : FFT Analyzer
0 200 400 600 800
0
2
4
6
[Hz]
[m/s²] Autospectrum(Signal 2) - InputWorking : Input : Input : FFT Analyzer
0 200 400 600 800
0
2
4
6
[Hz]
[m/s²]
Autospectrum(Signal 2) - InputWorking : Input : Input : FFT Analyzer
0 200 400 600 800
0
2
4
6
8
[Hz]
[m/s²] Autospectrum(Signal 2) - InputWorking : Input : Input : FFT Analyzer
0 200 400 600 800
0
2
4
6
8
[Hz]
[m/s²] Autospectrum(Signal 2) - InputWorking : Input : Input : FFT Analyzer
120 160 200 240 280 320 360 400
0
400m
800m
1.2
1.6
[Hz]
[m/s²] Autospectrum(Signal 2) - InputWorking : Input : Input : FFT Analyzer
120 160 200 240 280 320 360 400
0
400m
800m
1.2
1.6
[Hz]
[m/s²]
Comparison of Modal ParametersExperimental & Theoretical
Modal Test of Pipe Assembly
Autospectrum(Signal 4) - InputWorking : Input : Input : FFT Analyzer
0 20 40 60 80
0
10m
20m
30m
40m
[Hz]
[m/s²] Autospectrum(Signal 4) - InputWorking : Input : Input : FFT Analyzer
0 20 40 60 80
0
10m
20m
30m
40m
[Hz]
[m/s²] Time(Signal 3) - InputWorking : Input : Input : FFT Analyzer
0 200m 400m 600m 800m
-200
-100
0
100
200
[s]
[N] Time(Signal 3) - InputWorking : Input : Input : FFT Analyzer
0 200m 400m 600m 800m
-200
-100
0
100
200
[s]
[N]
Autospectrum(Signal 2) - InputWorking : Input : Input : FFT Analyzer
0 20 40 60 80
0
40m
80m
120m
160m
200m
[Hz]
[m/s²] Autospectrum(Signal 2) - InputWorking : Input : Input : FFT Analyzer
0 20 40 60 80
0
40m
80m
120m
160m
200m
[Hz]
[m/s²] Autospectrum(Signal 3) - InputWorking : Input : Input : FFT Analyzer
0 20 40 60 80
0
20m
40m
60m
80m
100m
[Hz]
[m/s²] Autospectrum(Signal 3) - InputWorking : Input : Input : FFT Analyzer
0 20 40 60 80
0
20m
40m
60m
80m
100m
[Hz]
[m/s²]
Pump Base Plate Resonance Responsible for Piping Vibration & Component Failures
Problem Statement
Two overhang Centrifugal Pumps working in parallel
– Pumps driven by 75 hp motors, 3600 rpm
– Over 15 years of operation, experienced repeat failures on piping system, valves & couplings.
– Routine vibration measurement for trending do not show any pattern
– Maintenance Personnel observed high vibration on the piping system where many of the failures occurred.
– Initial thought was that piping resonance was excited.
– It had been a practice to tighten the pipe hanging supports to reduce vibration temporarily.
Problem Statement (cont’d)
Vibration Measurement on South PumpBearing Housing, Coupling End
Vibration Measurement on South PumpBearing Housing, Coupling End
Vibration Measurement on South PumpBearing Housing, Coupling End
Vibration Measurement on South PumpBearing Housing, Coupling End
Vibration Measurement on South PumpBearing Housing, Coupling End
Vibration Analyses
All Vibration at running speed (1X)
Previous attempts at balancing did not yield any significant
improvements
Maintenance Personnel thought the problem was a piping
resonance
Impact Testing
Impact testing on the Inlet and discharge piping did not
show any resonance in the vicinity of running speed
The baseplates of both pumps were impacted in several
directions.
The natural frequency on the baseplate of the south pump
was found to be at the running speed of that pump.
Repairs
Repaired the pump foundation that had some cracks. Changed the baseplate Corrected the local practice of tightening the pipe hangers
to control vibration The pumps have been in operation for 5 years without
major issues
Summary
• Take your time• Choose your weapon• Bump around• Uniform Window• Look at the time waveform• Look at the frequency
spectrum
• Calculate the amplification factor
• Change the mass• Change the stiffness• Add damping• Bump around
• Phase Analysis
Phase Analysis
• A phase study is a collection of phase measurements made on a machine or structure and evaluated to reveal information about relative motion between components.
• In vibration analysis, phase is measured using absolute or relative techniques.
Absolute phase
• Absolute phase is measured with one sensor and one tachometer referencing a mark on the rotating shaft .
• At each measurement point, the analyzer calculates the time between the tachometer trigger and the next positive waveform peak vibration.
• This time interval is converted to degrees and displayed as the absolute phase.
• Phase can be measured at shaft rotational frequency or any whole number multiple of shaft speed (synchronous frequencies).
• Absolute phase is required for rotor balancing.
Absolute phase
Absolute Phase Measurement
Absolute phase
Absolute phase is calculated between the tach signal and vibration waveform
Relative Phase
• Relative phase is measured on a multi-channel vibration analyzer using two or more (similar type) vibration sensors.
• The analyzer must be able to measure cross-channel phase.
• One single-axis sensor serves as the fixed reference and is placed somewhere on the machine (typically on a bearing housing).
• Another single-axis or triaxial sensor is moved sequentially to all of the other test points.
• At each test point, the analyzer compares waveforms between the fixed and roving sensors.
• Relative phase is the time difference between the waveforms at a specific frequency converted to degrees.
• Relative phase does not require a tachometer so phase can be measured at any frequency.
Relative Phase
Relative Phase Measurement
Relative Phase
Relative Phase Calculated Between Two Vibration Waveforms
When to use Phase Analysis
• Everyone needs phase analysis.
• A phase study should be made on problem machines when
the source of the vibration is not clear or when it is necessary
to confirm suspected sources of vibration.
• A phase study might include points measured only on the
machine bearings or it can include points over the entire
machine from the foundation up to the bearings.
• How phase can help analyze vibration is given In the
• subsequent slides.
Soft Foot
• The term soft foot is used to describe machine frame distortion.
• It can be caused by a condition where the foot of a motor, pump or other component is not flat, square and tight to its mounting, or many other things, such as machining errors, bent or twisted feet and non-flat mounting surfaces.
• Soft foot increases vibration and puts undue stress on bearings, seals and couplings.
• Soft foot on a motor distorts the stator housing creating a non-uniform rotor to stator air gap resulting in vibration at two times line frequency.
Soft Foot
• A good laser shaft alignment system should be used to verify soft foot by loosening the machine feet one at a time.
• Phase can be used to identify soft foot while the machine is in operation.
• Measure vertical phase between the foot and its mounting surface.
• If the joint is tight, the phase angle is the same between surfaces.
• If the phase angle is different by more than 20 degrees, the foot is loose or the machine frame is cracked or flimsy.
Phase identifies in-plane or twisting bearing motion.
Phase shift across a soft foot.
Cocked Bearings and Bent Shafts
• Phase is used to detect cocked bearings and bent shafts.
• Measure phase at four axial locations around the bearing housing.
• If the bearing is cocked or the shaft is bent through the bearing, the phase will be different at each location.
• If the shaft is straight and the bearing is not twisting, the phase will be the same at each location.
Cocked Bearings and Bent Shafts
Confirm Imbalance
• A once-per-revolution radial vibration usually means rotor unbalance.
• Use phase to prove imbalance is the problem.
• To confirm imbalance, measure the horizontal and vertical phase on a shaft or bearing housing.
• If the difference between the phase values is approximately 90 degrees, the problem is rotor unbalance.
• If the phase difference is closer to zero or 180 degrees, the vibration is caused by a reaction force.
• An eccentric pulley and shaft misalignment are examples of reaction forces.
Confirm Imbalance
Horizontal to Vertical Phase Shift of about 90 Degrees Confirms Unbalance
Looseness, Bending or Twisting
• Phase is used to detect loose joints on structures and bending or twisting due to weakness or resonance.
• To check for looseness, measure the vertical phase at each mechanical joint as indicated by the arrows in Figure.
• When joints are loose, there will be a phase shift of approximately 180 degrees. The phase angle will not change across a tight joint.
Looseness, Bending or Twisting
A phase shift between bolted joints indicates looseness.
Shaft Misalignment
• Shaft misalignment is easily verified with phase.
• Measure each bearing in the horizontal, vertical and axial directions.
• Record the values in a table or bubble diagram as shown in Figure.
• Compare the horizontal phase from bearing to bearing on each component and across the coupling.
• Repeat the comparison using vertical then axial data.
• Good alignment will show no substantial phase shift between bearings or across the coupling.
• The machine in Figure has a 180-degree phase shift across the coupling in the radial directions.
• The axial directions are in-phase across the machine. The data indicates parallel (offset) shaft misalignment.
Shaft Misalignment
Phase Data Indicates Parallel Shaft Misalignment
Fault Analysis
Resonance
• Resonance is defined as:
An excitation of a natural frequency by a periodic forcing function.
• All assets contain natural frequencies that vary depending upon the
stiffness and mass. --- Resonance can be considered to be a vibration amplifier, that takes the
force level of the periodic forcing function and amplifies it; which significantly increases the movement of the asset.
If Vibration is a Fire, The Resonance is a Fuel
Example of Resonance• The example shown represents the effect on amplitude of the forcing
function when in resonance.– In plot 1 the 1xts is running below the natural frequency (Fn).– Fn can be seen in plot 2. – Plot 3 shows the increase in amplitude of the forcing function when
run at the natural frequency – this is resonance
Frequency
Frequency
Frequency
Before Excitation
Resonance Curve
Amplified Signal
1
2
3
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 frequency2. Stiffness – The more rigid a structure the higher the natural frequency
• Resonance is becoming more of a problem in industry in recent years due to:– Older equipment having to run faster to meet current production
demands (often above what it was designed for)– Equipment is being built cheaper and lighter
• This is resulting in amplification of the forcing function creating excessive machine movement resulting premature machine failure.
Effects of Resonance
• The ODS data is showing a steel frame structure deflecting at one corner in the vertical direction due to a resonant condition.
Characteristics of Resonance• Characteristics of Resonance
– Resonance is very directional in nature (Movement may be greater in one plain than the other)
– Vastly different amplitudes of the forcing function from one direction to the other (between Horizontal and Vertical – Rule of thumb ratio is 3:1 difference)
– Resonance is very speed sensitive (small changes in speed can show large differences in amplitude of the forcing function)
– Resonance can occur at any frequency but most commonly associated with the 1xTs
– 180 phase change occurs when shaft speed passes through resonance
Resolving a Resonance• There are a number of alterations to the system that can be made to
resolve a resonance condition. – However if structural changes are to be made you need to be careful
you don’t excite another natural frequency once the change has been made?
• Once you are sure you have a resonant condition it can be corrected by one of the following methods:– Change the Mass– Change the Stiffness– Remove the forcing function– Dampen the structure
Dampening is a method used to convert mechanical energy into thermal energy. It does not remove the resonant condition only controls the amount of movement.
Resonance – Spectral Data• The spectrum is showing the 1xTs peak of the motor with amplitudes reaching
19mm/sec. – This is high for the 1xTs.
• Very often this type of data can be mistaken for Imbalance as this defect can also produce a high 1xTs peak. – However Imbalance is a centrifugal force and should show similar amplitudes in
both radial plains where as resonance is very directional.
• In order to help resolve this issue we need to check the amplitude of the 1xTs 90 degrees to this point (horizontal to vertical) – This can easily be done by
using the ‘multi point plot’ in the software
40 - No 1 GCT CompressorM4551 -M2H Motor Inboard Horizontal
Route Spectrum 13-Feb-03 10:14:46
OVERALL= 19.95 V-DG RMS = 19.85 LOAD = 100.0 RPM = 1484. (24.73 Hz)
0 500 1000 1500 2000
0
3
6
9
12
15
18
21
24
27
Frequency in Hz
RM
S V
elo
cit
y in
mm
/Se
c
Freq: Ordr: Spec:
24.72 1.000 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.
RM
S V
elo
cit
y in
mm
/Se
c
Frequency in Hz
40 - No 1 GCT CompressorM4551 - Multiple Points (13-Feb-03)
0 500 1000 1500 2000
0
4
8
12
16
2024
Max Amp 22.0
M1H 10:14
M1V 10:14
M2H 10:14
M2V 10:15
RPM= 1484. 10:14:46 13-Feb-03 Point= M2H
Freq: Ordr: Sp 3:
25.00 1.011 19.35
Imbalance• Imbalance (Unbalance) occurs when the centre of mass differs from the
centre of rotation.• If the centre of mass changes on the rotor due to a heavy spot or some
other influence then a centrifugal force is produced. This results in the centre of rotation being offset from the centre of mass causing the vibration to increase at the rotational frequency.
Unbalance
• Primary Types
• Static or Forced• Coupled• Dynamic
Imbalance (Types)Static Imbalance
Dynamic Imbalance
Couple Imbalance
Static Imbalance
• The radial vibration readings are the highest amplitudes with the axial vibration generally much lower in amplitudes
• Static Imbalance will show a 0º phase shift across the rotor (vertical to vertical or horizontal to horizontal) and 90º phase shift from vertical to horizontal at the same bearing location
• The phase angle will change the same amount the heavy spot changes if the system is linear
Dynamic Imbalance
• Any thing other than static• Requires more than one correction planes• Rule of thumb• If D/W < 4 two plan is required• D = Diameter of rotor, w = width of rotor• Two unequal/equal heavy spots 180º apart in separate planes
on the same rotor or located at some spacing other than 180º. Amplitudes will differ or will be related to the location and amount of heavy spot
Unbalance
• Causes of Imbalance Improper Assembly Material build up / dirt Wear to components Broken or missing partsAll of the above conditions will result in an unbalanced state
• Diagnostic Rules for unbalance Periodic non-impacting sinusoidal waveform Spectral peak at 1xTs (1 Order) Very little axial vibration in case of static imbalance but high in case of overhung rotor Similar amplitudes between horizontal and vertical plains for static imbalance and differ in case of
dynamic imbalance 90º phase shift from vertical to horizontal Synchronous fault type Amplitudes will increase with speed Very low harmonics of 1xTs
Static unbalance
• Force unbalance will be in-phase and steady• Amplitude will increase with the square of speed• 1X RPM always present and normally dominates• Can be corrected by the placement of one weight in one plane
Dynamic/coupled unbalance
• 0-180 out of phase on the same shaft for dynamic & 180 out for coupled
• 1X RPM always present and normally dominates• Amplitude varies with square of increasing speed• Can cause high axial as well as radial amplitudes• Balancing requires Correction in two planes
• 1X RPM present in radial and axial directions• Axial readings tend to be in-phase but radial readings might
be unsteady• Overhung rotors often have both force and couple unbalance
each of which may require correction
Overhung Rotor unbalance
IF - Example 2Ex2 -F1H Fan Inboard Horizontal
Route Spectrum 16-Sep-99 08:36:29
OVRALL= 4.58 V-DG RMS = 4.56 LOAD = 100.0 RPM = 3000. RPS = 50.00
0 20000 40000 60000 80000
0
1
2
3
4
5
6
Frequency in CPM
RM
S V
elo
city
in m
m/S
ec
Freq: Ordr: Spec:
3000.0 1.000 4.539
Unbalance Spectral Data The spectrum shown represents a simple unbalance state Single peak at 1xTs (1 Order) Little indication of harmonics
What should the waveform show?
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
Changing the units to velocity would reduce the amount of high frequency noise residing on the waveform
IF - Example 2Ex2 -F1H Fan Inboard Horizontal
Waveform Display 02-Feb-00 15:13:51
PK = .5289 LOAD = 100.0 RPM = 2985. RPS = 49.76
PK(+) = .8332 PK(-) = .8893 CRESTF= 2.38
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
-1.0
-0.8
-0.6
-0.4
-0.2
-0.0
0.2
0.4
0.6
0.8
1.0
Revolution Number
Acc
eler
atio
n in
G-s
E02N - JB1420C CONDY RECOVERY PUMPJB1420C -M1H Motor Outboard Horizontal
Trend Display of 1xTS
-- Baseline -- Value: 3.063 Date: 07-Apr-00
0 100 200 300 400 500
0
2
4
6
8
10
12
14
Days: 07-Apr-00 To 21-May-01
RM
S V
elo
cit
y in
mm
/Sec
FAULT
Date: Time: Ampl:
21-May-01 14:24:29 11.21
Imbalance Trend Data The trend data is a good way of determining if there has been a change in
condition, as this plots amplitude against time (where time is in days) Here the 1xTs parameter is being trended Vibration has been steady at 3mm/sec for a period of time A sudden change instate should alert the analyst to a fault developing
Imbalance Problem - Practical The following fan unit has an imbalance present on the rotor. 1xTs Peak in the Spectrum 1xTs Peak in the Waveform
What would happen to the data if the following occurred to the fan?
ImbalanceIF - Example 2
Ex2 -F1H Fan Inboard HorizontalRoute Spectrum 16-Sep-99 08:36:29
OVRALL= 4.58 V-DG RMS = 4.56 LOAD = 100.0 RPM = 3000. RPS = 50.00
0 20000 40000 60000 80000
0
1
2
3
4
5
6
Frequency in CPM
RM
S V
elo
cit
y in
mm
/Sec
Freq: Ordr: Spec:
3000.0 1.000 4.539IF - Example 2
Ex2 -F1H Fan Inboard HorizontalWaveform Display 02-Feb-00 15:13:51
PK = .5289 LOAD = 100.0 RPM = 2985. RPS = 49.76
PK(+) = .8332 PK(-) = .8893 CRESTF= 2.38
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
-1.0
-0.8
-0.6
-0.4
-0.2
-0.0
0.2
0.4
0.6
0.8
1.0
Revolution Number
Accele
ratio
n in
G-s
Bent Shaft
• Bent shaft problems cause high axial vibration• 1X RPM dominant if bend is near shaft center• 2X RPM dominant if bend is near shaft ends• Phase difference in the axial direction will tend towards 1800 difference
Bend near the Bearing
1
High axial vibrations at 1 x rpm Axial phase readings are different on the same
bearing housing.
14 2
3
Phase
1 30
2 50
3 120
4 140
Axial Phase showing twisting axial motion
Bend due to Shaft Bow
• High vibration at 1x rpm in axial direction• Phase is steady in axial direction on the same
bearing housing• But the two bearing housings supporting the
rotor are out of phase by nearly 180o in axial90o
270o
Misalignment• When two mating shafts do not share the same collinear axis then
misalignment is induced.
• Misalignment is one of the primary reasons for premature machine failure. The forces that are exerted on the machine and its components when in a misaligned state are greatly increased from normal operating conditions
Misalignment
• Operational Deflection Shape (ODS) is a technique that machine
movement based upon the phase and magnitude of data collected from
the analyser. Shown below is an image from the ODS illustrating the forces
that are exerted onto the machine and components when running in a
misaligned condition
Misalignment• Misalignment can be broken into three basic categories, these are:
• Angular – Where the shaft centrelines cross producing a 1xTs peak axially
Offset – Where the shaft centrelines are parallel but they do not meet producing a radial 2xTs peak
More commonly seen – A combination of the above
Misalignment
Misalignment• Another common problem associated with
alignment is ‘bearing misalignment’.• Bearing misalignment occurs when the bearings
are not mounted in the same plain possibly due to: one or more of the bearings being cocked in the
housing The machine itself distorts due to thermal
growth or soft foot conditions Misalignment at the drive causes shaft bending.
Misalignment• Diagnostic Rules for Misalignment
– High axial levels of vibration at 1xTs(often .5-2 times the radial readings)– High radial levels of vibration at1xTs and/or 2xTs, 3x & 4x may also be present– Repeatable period sine waveform showing 1, 2,3,4 clear peaks per revolution (Most likely
“M” or “W” shape)– Data can usually be seen across the coupling– Phase reading will differ by 180º in axial or radial directions– Other visual observations may also be present like:• Excessive bearing temperature• Oil Leakage around the seals• Coupling worn /wear
• Diagnostic Rules for Bearing Misalignment– High levels of vibration at 1xTs and 2xTs– Repeatable periodic sine waveform showing 1 or 2 clear peaks per revolution– Data usually shown either the driver or driven component
Offset Misalignment Spectral Data
• The spectral data shown represents a simple misalignment plot. – The primary cursor denotes the 1xTs peak while the harmonic cursors
indicate a larger 2xTs peak. This type of data is common to that of Offset Misalignment
ST.1 - Raw Water PumpP029 -M2H
Route Spectrum 15-FEB-93 11:04:18
OVRALL= 6.50 V-DG RMS = 6.47 LOAD = 100.0 RPM = 2976. RPS = 49.61
0 10000 20000 30000 40000 50000
0
1
2
3
4
5
6
7
Frequency in CPM
RM
S Ve
loci
ty in
mm
/Sec
Freq: Ordr: Spec:
2925.0 .983 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 33601PUM003-M2A Motor Inboard Axial
Route Spectrum 04-Aug-04 08:49:05
OVERALL= 6.33 V-DG RMS = 6.31 LOAD = 100.0 RPM = 1071. (17.84 Hz)
0 30 60 90 120
0
1
2
3
4
5
6
7
8
Frequency in kCPM
RM
S Ve
loci
ty in
mm
/Sec
Freq: Ordr: Spec:
1.071 1.000 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
ST.1 - Raw Water PumpP029 -M2H
Waveform Display 26-MAR-93 13:32:52
RMS = 17.00 LOAD = 100.0 RPM = 2996. RPS = 49.93
PK(+) = 30.66 PK(-) = 26.81 CRESTF= 1.82
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
-30
-20
-10
0
10
20
30
40
Revolution Number
Velo
city
in m
m/S
ec
Misalignment• 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 33601PUM003-M2A Motor Inboard Axial
Route Waveform 04-Aug-04 08:49:05
PK = .2596 LOAD = 100.0 RPM = 1071. (17.84 Hz)
PK(+) = .6277 PK(-) = .5683 CRESTF= 3.42
0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
Revolution Number
Acc
eler
atio
n in
G-s
Rev : Ampl:
.680 -.306
Angular Misalignment
• Characterized by high axial vibration• 180 phase change across the coupling• Typically high 1 and 2 times axial vibration• Not unusual for 1, 2 or 3X RPM to dominate• Symptoms could indicate coupling problems
Parallel Misalignment
• High radial vibration 1800 out of phase• Severe conditions give higher harmonics• 2X RPM often larger than 1X RPM• Similar symptoms to angular misalignment• Coupling design can influence spectrum shape and
amplitude
Radial
1x 2x4x
Bearing Misalignment
• Vibration symptoms similar to angular misalignment• Attempts to realign coupling or balance the rotor will not alleviate
the problem.• Will cause a twisting motion with approximately 180 phase shift
side to side or top to bottom
Vibration due to Eccentricity
• The distance of the geometric center of a rotating body from the axis of rotation
• Some eccentricity or out of roundness is present in all type of rotors• Is the major cause of unbalance and can be corrected by balancing • But balancing more eccentric rotors result in reducing vibration in one
direction and increasing in the other.• Common type of eccentric rotors are eccentric pulleys, gears, pump
impellers, motor rotors.
Vibration due to Eccentricity• High vibration at 1xrpm of eccentric component in radial direction• Eccentricity is high directional in nature, highest vibration is in the direction
of belt tension in case of eccentric pulleys• Comparative phase readings between horizontal to vertical usually differs
by 0 or by 180 degrees.• Eccentric motor rotors show vibration at 2FL with pole pass frequency side
bands.• Eccentric pump impellers show high vibration at 1 rpm and also at Vane
pass frequencies and their harmonics• Eccentric gears results high vibration at 1 rpm of eccentric gears. Phase may
be used to differentiate between unbalance and eccentricity. Eccentric gears also exhibit high levels of gmf with 1 rpm side bands.
• Largest vibration at 1X RPM of eccentric rotor • Horizontal and vertical phase readings differ by 0 or 180• Attempts to balance will cause a decrease in amplitude in one
direction but an increase may occur in the other direction
Eccentric Rotor
How would looseness ?
Looseness• Looseness can be broken down into two main categories, Structural and
Component
Structural looseness occurs when there is free movement within the machines support structure causing excessive vibration. This can be a result of:– Loose support bolts to the components feet and supports– Cracked welds– Deterioration of the base itself.
Component looseness generally occurs when there is excessive clearance to the components within the machine, such as:– Excessive clearance between the shaft and bearings– Excessive clearance between the shaft and an impeller etc.
Looseness• Diagnostic Rules for Looseness
– Multiple harmonics of the 0.5xTs and/or 1xTs peak - Structural– Multiple Harmonics of the component that is loose - Component– Number of harmonics will increase as the looseness progresses– Random, non-periodic waveform - Structural– Waveform shows predominant impacts – Component– May also truncation in the waveform– Phase varies and unsteady– Raised noise level around the 1xTs + harmonics– Half harmonics may also be present– Can be present in all Directions but often high in vertical direction
Looseness Spectral Data (Structural)
– The spectral plot shown is demonstrating Looseness. – The 1xTs peak has been highlighted by the primary cursor and the
relevant harmonics have been displayed.– Multiple harmonics of 1xTs are shown up to around 10 orders of 1xTs.
40 - Kiln Main DriveM4441 -G2H Shaft 01 Outboard Horizontal
Route Spectrum 06-Nov-02 11:02:11
OVERALL= 5.22 V-DG RMS = 5.22 LOAD = 100.0 RPM = 635. (10.58 Hz)
0 200 400 600 800 1000
0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Frequency in Hz
RM
S V
elo
cit
y in
mm
/Sec
Freq: Ordr: Spec:
10.58 1.000 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.
L1 - Example 9Ex 9 -P2A Pump Outboard Axial
Label: Centrifugal Pump - Medium
Route Spectrum* 17-Aug-01 08:52:02
OVERALL= 6.62 V-DG RMS = 6.13 LOAD = 100.0 RPM = 2974. (49.57 Hz)
0 40 80 120 160 200 240
0
0.3
0.6
0.9
1.2
1.5
Frequency in kCPM
RM
S V
elo
city
in m
m/S
ec
Freq: Ordr: Spec:
14.88 5.002 .742
The raised noise level around the vane pass frequency is common to a pumping problem known as Cavitation– This would be the likely
cause of the impeller problem
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 DriveM4441 -G2H Shaft 01 Outboard Horizontal
Route Waveform 06-Nov-02 11:02:11
RMS = .3174 LOAD = 100.0 RPM = 635. (10.58 Hz)
PK(+) = .9797 PK(-) = .9874 CRESTF= 3.11
0 50 100 150 200 250 300 350 400
-1.2
-0.8
-0.4
0.0
0.4
0.8
1.2
Time in mSecs
Acc
eler
atio
n in
G-s
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.
40 - Kiln Main DriveM4441 -G2H Shaft 01 Outboard Horizontal
Trend Display of 3-15xTS
-- Baseline -- Value: .837 Date: 28-Feb-02
0 10 20 30 40 50
0
1
2
3
4
5
6
7
8
Days: 28-Feb-02 To 16-Apr-02
RM
S V
elo
city
in m
m/S
ec
ALERT
FAULT
Mechanical Looseness
• Caused by structural looseness of machine feet• Distortion of the base will cause “soft foot” problems• Phase analysis will reveal aprox 180 phase shift in the vertical
direction between the base plate components of the machine
Mechanical Looseness
• Caused by loose pillow block bolts• Can cause 0.5, 1, 2 and 3X RPM• Sometimes caused by cracked frame structure or bearing block
Mechanical Looseness
• Phase is often unstable• Will have many harmonics• Can be caused by a loose bearing liner, excessive bearing
clearance or a loose impeller on a shaft
Rotor Rub
• Similar spectrum to mechanical looseness• Usually generates a series of frequencies which may excite
natural frequencies• Sub harmonic frequencies may be present• Rub may be partial or through a complete revolution.
Truncated waveform
Rolling Element and Plan Bearing Defects
Rolling Element Bearings
– Rolling element bearings have specific bearing failure modes that can
be observed in the spectral and waveform data.
– Bearing frequencies differ from most other frequencies present
within the spectral data because unless the bearing has a defect
there will be no frequency peaks in the data relating to the bearing.
Only if the bearing has a defect will frequencies show in the spectral
data.
There are four main fundamental bearing defect frequencies these are:
Rolling Element Bearings
Inner Race
Outer Race
How Bearing Faults Generate Vibration
• BALL SPIN (BSF)
• CAGE (FTF)
• INNER RACE (BPFI)
• OUTER RACE (BPFO)
FTF & BSF
BPFI & BPFO
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
BEARINGS in CSI Warehouse:
c:\RBMsuite\SysData\CSI_CMP.WH ****************************************************
BRG ID Bearing Type #B/R FTF BSF BPFO BPFI 12143 RHP 6218 11 0.418 2.967 4.598 6.402 24421 SKF 6313E 8 0.376 1.894 3.009 4.991 25372 SKF I-26313 19 0.433 3.568 8.219 10.781
Rolling Element Bearings• Characteristics of Bearing Defects
– High frequency raised noise level (Hump of energy)– Non-Synchronous harmonic peaks (Both low and high frequency)– Time waveform will show a lot of noise/impacting – Early stages of bearing wear may show better if viewed in acceleration
in the frequency domain– Fundamental bearing defect frequency (First calculable frequency)
may not be present in the spectral data– Sidebands surrounding BPFO are much more serious than sidebands
surrounding BPFI (for fixed outer race)
Rolling Element Bearings
• The appearance of defect frequencies USUALLY starts with…
• BPFO & BPFI
• Followed by BSF
• Followed by FTF• These scenarios presume the absence of manufacturing errors on rolling
element bearing components
• When a roller or ball defect is present from the start, BSF may well appear in the spectrum WITHOUT any progression similar to these scenarios
Frequencies Generated By REBs
• Random HF to ultrasonic
–5KHz to 60 KHz
• Component Fn
–30KCPM to 120KCPM• 54K to 96K for most
• Defect Frequencies
• Sum & Difference / Sidebands
Failure Mode 1• The early stages of bearing defects
produce low amplitudes of vibration at higher frequencies – (Appears on the right hand side of
the spectrum). • These are normally humps of energy or
peaks that are harmonics to the fundamental frequency. – (The fundamental frequency should
not be visible at this stage).
Stage 1: Fault Onset
ZONEA
ZONE BBEARINGDEFECTFREQ. REGION
ZONE CBEARINGCOMPON.NATURAL FREQ. REGION
30K
120K
1X2X
3X
STAGE 1
Vibration Analysis (typical):1) Standard FFT: no visible indication in spectrum2) Spike Energy: slight increase in value (e.g. 0.25 gSE)3) PeakVue: bearing frequency peak(s) corresponding to fault type
amplitude at 2-7 g’s depending on type and location
ZONE BBEARINGDEFECTFREQ. REGION
ZONE CBEARINGCOMPON.NATURAL FREQ. REGION
ZONEA
30K
120K
BPF
O/B
PFI
3x2x
STAGE 1
Oil Analysis (typical):1) Readings: Slight increase in elemental Fe, particle count, and WPC2) Visual Ferrography:
• Small platelet shaped particles (<30 μ) from contact fatigue• Small spherical shaped particles (<5 μ) from surface fatigue
Failure Mode 2
• Distinct harmonics of Non-Synchronous peaks appear. – (These should appear lower down the scale
of the spectrum – towards the left / middle of the plot)
• Sidebands may appear around these frequencies usually equating to turning speed. – (The fault frequencies may not match exactly
with the peaks in the spectrum due to the fact that the bearing geometry will have changed).
Stage 2: Intermediate Wear
ZONE BBEARINGDEFECTFREQ. REGION
ZONE CBEARINGCOMPON.NATURAL FREQ. REGION
1X2X
3X
STAGE 2
Bea
ring
f
ZONEA
30K
120K
Vibration Analysis (typical):1) Standard FFT: bearing defect ‘rings’ in Zone C (natural freq.)2) Spike Energy: increase in value (e.g. 0.50 gSE)3) PeakVue: bearing frequency peak(s) with increasing harmonics
amplitude at 3-10 g’s depending on type and location
Oil Analysis (typical):4) Readings: elemental Fe stable but increase in particle count, WPC, and PLP5) Visual Ferrography:
• Platelet shaped particles (30-50 μ) from contact fatigue• Possible spherical shaped particles (<5 μ) from surface fatigue
ZONE BBEARINGDEFECTFREQ. REGION
ZONE CBEARINGCOMPON.NATURAL FREQ. REGION
ZONEA
ZONE BBEARINGDEFECTFREQ. REGION
ZONE CBEARINGCOMPON.NATURAL FREQ. REGION
ZONEA
30K
120K
BPF
O/B
PFI
3x2x 4x 6x5x
STAGE 2
Failure Mode 3
• The fundamental frequency normally appears at this stage – (First calculable frequency of the bearing – towards
the left-hand side of the spectral plot). This is classed as advanced stages of bearing wear.
• Sidebands may be visible that equate to other bearing frequencies – BSF, FTF etc).
Stage 3: Severe Wear
ZONE BBEARINGDEFECTFREQ. REGION
ZONE CBEARINGCOMPON.NATURAL FREQ. REGION
1X2X
3X
STAGE 3
Bea
ring
f
BPF
I
BPF
I
BPF
O
ZONEA
30K
120K
Vibration Analysis (typical):1) Standard FFT: bearing frequencies with harmonics and sidebands in Zone B2) Spike Energy: increase in value (e.g. 1.0 gSE)3) PeakVue: bearing frequency peak(s) with increasing harmonics and sidebands
amplitude climbs to 5-10 g’s or higher (depending on type and location)
Oil Analysis (typical):4) Readings: small change in elemental Fe, substantial increase in WPC and PLP 5) Visual Ferrography:
• Sharp increase in large particles (>30μ), both platelets and cutting wear• Increased three-dimensional appearance to wear particles
ZONE BBEARINGDEFECTFREQ. REGION
ZONE CBEARINGCOMPON.NATURAL FREQ. REGION
ZONEA
30K
120K
BPF
O/B
PFI
2x 4x
6x5x3x
7x 8x
STAGE 3
Failure Mode 4
• The bearing degrades so much that the
spectrum becomes a mass of noise. At this
point the bearing will fail at any point (If it
last this long – most fail around Mode 3).
Stage 4: Imminent Failure
ZONE BBEARINGDEFECTFREQ. REGION
ZONE CBEARINGCOMPON.NATURAL FREQ. REGION
RANDOM HIGHFREQ. VIBRATION
STAGE 4
1X2X
3X
ZONEA
30K
120K
Vibration Analysis (typical):1) Standard FFT: discrete bearing frequencies replaced by broadband noise2) Spike Energy: falling levels until just before failure, then levels rise sharply3) PeakVue: bearing frequency peak(s) with increasing harmonics and sidebands
amplitude climbs to 10 g’s or higher (depending on type and location)
Oil Analysis (typical):4) Readings: small change in elemental Fe, substantial increase in WPC and PLP 5) Visual Ferrography:
• Broad range of huge particles (75μ+) from fatigue and adhesion• Particle counts/ferrous density are excessive
?
ZONE BBEARINGDEFECTFREQ. REGION
ZONE CBEARINGCOMPON.NATURAL FREQ. REGION
ZONEA
30K
120K
BPF
O/B
PFI
2x 4x 6x5x3x
7x 8x
STAGE 4
Rolling Element Bearings - BPFI• Typical data showing a defected inner race
– Fundamental frequency showing– Harmonics low and high frequency + sidebands
Rolling Element Bearings - BPFO• Data showing a defect related to the BPFO
– The fundamental frequency is showing– Harmonics from low to high frequency
Rolling Element Bearings - BSF
• Bearing defect showing the BSF – Rolling elements– Sidebands around the BSF = FTF
Rolling Element Bearings - FTF• The FTF is the only bearing frequency that is sub-synchronous
– May not detect then with conventional vibration data– FTF defect at 0.4 orders shown in Peakvue
• Bearing
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
• This data is taken from a pump with a damaged
• bearing– The force levels are
reaching 40G-s
Rolling ElementPlain Bearings
Peakvue
Bearing Defects
Plain Bearings
............................................................................................................
• Since there is no contact between the bearing
and the shaft monitoring of sleeve bearings for
vibration analysis usually requires the use of
displacement probes mounted 45 degrees
either side of top dead centre.
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.
Plain Bearings
• As there are no rotating components in the bearing that produce high
frequency noise (force) there is no need to monitor a high frequency range.
Usually 10 to 15 orders of turning speed will be sufficient.
• Sleeve bearings have specific defects that contribute towards bearing
failure, these are:
Excessive clearance
Hydraulic instability (oil whirl)
Plain Bearings – Spectral Diagnostics Excessive Clearance When there is excessive clearance between the rotor and the bearing
then this will have an effect on the system vibration. When the bearings have excessive clearance then a ‘looseness’ occurs.
The spectral data shown below is indicating a sleeve bearing with excessive clearance.
As the clearance increases then the harmonics of 1xTs will increase and can go up to 10–15xTs. – Like looseness the more
harmonics there are the more severe the problem will be.
– A good sleeve bearing will still show a few harmonics as there is a small clearance between the shaft and bearing
Fu - Turbine Brg Thrust EndTBT -R1Y Radial 'Y' Direction
Route Spectrum* 27-Jul-04 14:08:21
OVERALL= 2.93 V-DG P-P = 22.71 LOAD = 100.0 RPM = 941. (15.69 Hz)
0 3 6 9 12
0
4
8
12
16
Frequency in Orders
P-P
Dis
pla
ce
me
nt
in M
icro
ns
Ordr: Freq: Spec:
1.000 15.68 7.494
Fu - Turbine Brg Thrust EndTBT -R1Y Radial 'Y' Direction
Route Spectrum* 27-Jul-04 14:08:21
OVERALL= 2.93 V-DG P-P = 22.71 LOAD = 100.0 RPM = 941. (15.69 Hz)
0 3 6 9 12
0
4
8
12
16
Frequency in Orders
P-P
Dis
pla
cem
ent
in M
icro
ns
Ordr: Freq: Spec:
1.000 15.68 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.
• The dominant peak within the spectral data will be typically at 0.4 orders. (.40-.48)
– This defect is sub-synchronous data. – When the amplitude of the oil whirl is
equal to or greater than the 1xTs peak a problem exists
• In this instance oil whirl can be corrected by:– Properly loading the bearing– Change the oil viscosity– Change the oil pressure
Oil Whirl at 0.4 orders
Plain Bearings – Spectral Diagnostics
Oil Whirl
• Usually occurs at 42 - 48 % of running speed• Vibration amplitudes are sometimes severe• Whirl is inherently unstable, since it increases
centrifugal forces therefore increasing whirl forces
• Oil whip may occur if a machine is operated at 2X the rotor critical frequency.
• When the rotor drives up to 2X critical, whirl is close to critical and excessive vibration will stop the oil film from supporting the shaft.
• Whirl speed will lock onto rotor critical. If the speed is increased the whip frequency will not increase.
oil whirl
oil whip
Oil Whip Instability
Rolling ElementPlain Bearings
Peakvue
Bearing Defects
Peakvue™
What is Peakvue™
• What is Peakvue? Peakvue is a technology unique to CSI and means ‘Peak Value’ Such as the Peak Value of an impact generated by a bearing defect in a time
waveform - (True Peak Value) If you have a 21XX analyzer you have the capability to acquire ‘Peakvue
Data’
These stress waves travel further than conventional vibration signals so a truer indication of fault severity is obtained.
The ‘True Peak Value’ is obtained by concentrating on ‘Stress Wave Analysis’ rather than conventional vibration data.
What is Peakvue™
What is a Stress Wave?
Stress waves accompany metal-metal impacting. These stress waves are short-term (fractional to a few milliseconds) transient events, which introduce a ripple effect on the surface machinery as they propagate away from the initial event. – If you think of a stone being dropped into a pool of water. The stone is
the initial impact generated by the fault. The effect of the stone being dropped into the water cause a ripple on the surface of the water which, spreads over a wide area.
Initial Impact
What is Peakvue
• If a bearing has a sub-surface defect (early bearing wear), when a rolling element passes over the defect it bends the race slightly and then as the rolling element passes it restores back to it’s natural state.
• This event causes a high frequency (1-50KHz) short duration stress wave.
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 theuse of a filter to remove all unwanted noise that can dominate the data
1. Conventional Vibration Signals that are filtered from the Peakvue Signal
· Imbalance· Misalignment· Gears· Bearings· Resonance
2. Peakvue filter removing low frequency noise from the stress wave data
· This is to prevent low frequency noise consuming the stress wave activity
3. High frequency stress wave activity occurring in the 1000Hz - 20000Hz frequency range at a rate governed by a low frequency event
· Bearings· Gears
Peakvue Processing - Filters• There are two types of filters available• Band Pass Filters
The band pass filter removes all the data above and below the filter corner values
• High Pass FilterThe high pass filter removes all data lower in frequency to that of the filter selection allowing only the high frequency stress waves to pass through
After the filtering process what should remain is the high frequency stress wave activity that is occurring at the rate of the excitation – such as from a bearing.
f
f
How Does It Work?
A comparison can be made of the
sampling to show how data is collected
through both methods of data
acquisition, normal and Peakvue™.
FFT
High
Pass
Filter
Full
Wave
Rectify
Digital
Peak
Impact
Detection
Vibration Signal
How Does It Work?
• The diagram below shows sampling of data using normal data • collection.
Stress wave- this is missed under normal conditions
Instantaneous Samples
Peakvue Samples
How Does It Work?• The diagram below shows sampling of data using Peakvue™ data collection.
Stress wave- this is missed under normal conditions
How Does It Work?
• Peakvue measures the highest amplitude found in a stress waves (Pk Value) and holds that data
• The waveform data is then passed through a high pass filter to remove the unwanted, low frequencies– Imbalance, Misalignment, Looseness, resonance etc.
• This just leaves us with the high frequency impacting data (Peak) above the machine noise level
The data is then brought back to fundamental frequency. (this allows analysis of the data to be done quicker and easier)
PeakVue® How does it work?
• The waveform should contain enough time to include at least 15 shaft
revolutions to resolve cage frequency in the spectrum for rolling element
bearings. (The waveform time length is determined by the lines of
resolution divided by the f-max)
PeakVue® How does it work?
• The f-max should be set at least 3 or 4 times the highest expected
defect frequency (usually inner race defect for rolling element bearings)
One average should be used when taking PeakVue® data
PeakVue® How does it work?
• Transducer mounting should be consistent for trend able
data. At minimum the surface should be clean (free of
paint, dirt, etc.), stress waves are easily attenuated.
Types of filter availableFilter CalculationsFilter Guidelines
Filters
Selecting the wrong type of filter will result in poor quality data– To much noise filtered through (the spectrum becomes very noisy)– To much is filtered out (The stress wave is not allowed to pass
through)
Filters Options
• There are two types of filter options in Peakvue, these are:– 1. Band Pass Filter– 2. High Pass Filter
• Each of the filters are designed to remove unwanted data out of the signal at the appropriate levels
One of the key elements in acquiring meaningful peakvue data is the selection of ‘filters’
Filter Options - ‘High Pass Filter’
• High Pass Filters remove all frequencies from the data below the filter
setting but allow the high frequency stress wave to pass through.
All low frequencies are removed from the input signal
Stress Wave data is allowed to pass through the filter
High Pass Filter
Filter Options - ‘Band Pass Filters’
• Looks for stress waves within a parameter defined by the filter setting.
Frequencies above and below this setting are removed from the data
Data is filtered out of the signal
Data is filtered out of the signal
Data passes through filter
Filter Selection
• To select the correct filter we need to consider the highest operational defect
frequency that we want to measure/detect. Then select the next available
filter above that frequency
• E.g.
• Consider a typical motor / pump arrangement. We have:
•1 - 4 Pole A.C. Induction Motor
•2 - 3 Jaw Coupling
•3 - Centrifugal Pump
--- Typically the highest defect frequency
to emit from this machine would be?1 - BPFI - Bearing Defect
Filter Selection
• 4 Pole Motor A.C Induction fitted with bearings SKF 6313
• Defect Frequencies (Orders)
•FTF - 0.384
•BSF - 2.037
•BPFO - 3.071
•BPFI - 4.929 Typically we would want to see the 10th Harmonic of the BPFI
– Highest defect frequency:• (BPFI x 10) x Turning Speed (Hz)• (4.929 x 10) x 25• 1232.3 Hz
– We would then select the next available filter setting above the frequency
Available filters
• High Pass Filters
• 500hz
• 1000hz
• 2000hz
• 5000hz
• 10000hz
• 20000hz
• Band Pass Filters
• 20hz – 150hz
• 50hz – 300hz
• 100hz – 600hz
• 500hz – 1khz
Note: the meter will only allow you to select the next filter above the specified Fmax.
From our previous calculation of 1232Hz, What filter setting would we select?
Filter uses (Band Pass) - Guidelines
• Band Pass Filters
• 20hz – 150hz Felt problems on paper machines
• 50hz – 300hz Certain structural resonance excitation,
modulation of gearmesh in low speed machinery
• 100hz – 600hz Gearmesh modulation in intermediate speed
machinery.
• 500hz – 1khz Gearmesh modulation
Tip: use bandpass filters when the event of interest is the excitation of a structural resonance, or the modulation of known frequencies – such as gearmesh.
Filter uses (Highpass) - guidelines
• High Pass filters
• 500hz Low speed machinery having <125hz. Bearing & gearing problems
• 1000hz Intermediate speed machinery (<2000 rpm) with gear mesh <300hz
• 2000hz Medium speed machinery (<4000rpm) with gear mesh ,600hz
• 5000hz Machinery up to 9000rpm and gear mesh to 1500hz, Requires attention be
paid to how the sensor is mounted as well as the sensors frequency response.
• 10000hz High speed machinery with gear mesh up to 3000hz sensor must be
permanently mounted with a frequency response of 3db in the 30kHz or higher
range.
• 20000hz High speed machinery with gearmesh up to 6000hz. Sensor must be high
frequency and permanently mounted.
Tip: Use highpass filters when the objective is to detect stress waves which are emitted by metal on metal impacting.
Filter Selection - Question
• Consider:– Motor running at a speed of 1000RPM– Driving a fan unit via pulley belts– Fan Speed is 1350RPM
• Motor Bearings = SKF 3095 - BPFI 4.855• Fan Bearings = SKF 6210 - BPFI 5.907
• Calculate what Filter setting would be required for both the motor and the fan bearings? – Filters Available:
• 500 Hz, 1000Hz, 2000Hz, 5000Hz, 10000Hz, 20000Hz. (High Pass)• 20-150Hz, 50-300Hz, 100-600Hz, 500-1KHz. (Band Pass)
Filter Selection - Answers
• Motor Speed = 1000CPM / 60 = 16.667Hz• Fan Speed = 1350CPM / 60 = 22.5Hz
• Filters Available:• 500 Hz, 1000Hz, 2000Hz, 5000Hz, 10000Hz, 20000Hz. (High Pass)• 20-150Hz, 50-300Hz, 100-600Hz, 500-1KHz. (Band Pass)
Motor.
– BPFI = 4.855– Defect Frequency = (BPFI x 10) x Turning Speed (Hz)– Defect Frequency = (4.855 x 10) x 16.667– Defect Frequency = 809.18 Hz
Filter Selection - Answers
• Motor Speed = 1000CPM / 60 = 16.667Hz• Fan Speed = 1350CPM / 60 = 22.5Hz
• Filters Available:• 500 Hz, 1000Hz, 2000Hz, 5000Hz, 10000Hz, 20000Hz. (High Pass)• 20-150Hz, 50-300Hz, 100-600Hz, 500-1KHz. (Band Pass)
Fan
– BPFI = 5.907– Defect Frequency = (BPFI x 10) x Turning Speed (Hz)– Defect Frequency = (5.907 x 10) x 22.5– Defect Frequency = 1329.07Hz
Spectrums and WaveformsDiagnostics Techniques
Peak-vue Data
Peakvue - Spectrum
• Here is a typical Peakvue spectra plot.
• This is typically a GOOD spectrum
1. Broad band energy - Filtered Noise
2. Units should be ‘acceleration’ (Very high frequency analysis)
3. Amplitude values are low. Severity of fault is not determined in the spectra
Peakvue - Spectrum
• This is a Peakvue spectrum where high frequency stress waves are being detected
• This is indication of a fault developing
1. Broad band energy - Filtered Noise
2. Units still in ‘acceleration’ (Very high frequency analysis)
3. Amplitude values are low. Remember severity of fault is not determined in the spectra
Notice the Impacts passing through the filtered noise
Peakvue Processing – Spectral Data
• Shown below is a typical Peakvue spectrum with a defect present• The filter used is shown in the top right
hand corner
Stress waves are showing clearly in the data at 4.6 Orders
Noise removed by filter
Good Spectrum will show only a noise level
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
Route Waveform 09-Jul-03 09:50:49 (PkVue-HP 1000 Hz) RMS = 2.97 PK(+) = 8.35 CRESTF= 2.81
0 4 8 12 16 20 24 28 32 36
0
1
2
3
4
5
6
7
8
Revolution Number
Acc
eler
atio
n in
G-s
B42 - ZONE 5 DF FAN 116/16EXT01-M2P Motor Inboard Horz Peakvue
Label: Bearing Fault - BPFO NTN6217
Route Spectrum 09-Jul-03 09:50:49 (PkVue-HP 1000 Hz) OVERALL= 1.37 A-DG RMS = 1.37 LOAD = 100.0 RPM = 1342. (22.37 Hz)
0 200 400 600 800 1000
0
0.1
0.2
0.3
0.4
0.5
0.6
0.70.8
Frequency in Hz
RM
S A
ccel
erat
ion
in G
-s
Freq: Ordr: Spec:
1.250 .05587 .01367
>NTN 6217 N=BPFO -OB
N N N N N N N N N
For Peakvue analysis Use the Spectrum
– Diagnose the defect Use the Waveform
– Determine the severity
Peakvue - Waveforms
• Waveforms can be confused with spectrums, as the waveform is only plotting the peak value and
does not show a full wave.A1 - Example 1
EX 1 -D3P Tail Roll Non D/S Peakvue
Label: Easy
Analyze Waveform 16-Mar-01 12:03:14 (PkVue- HP 500 Hz)
PK = .0556 LOAD = 100.0 RPM = 80. RPS = 1.33
PK(+) = .5599 PK(-) = .0397 CRESTF= 14.25
0 3 6 9 12
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
Revolution Number
Acc
eler
atio
n in
G-s
1. Filtered Noise Level
2. Peak Value Impacts
3. No Peak Negative Value
4. Acceleration as default units
Peakvue - Diagnostics
• Diagnosing a Peakvue spectrum and waveform is not to dissimilar to that of
conventional data.
• However there are a few differences which can be a bit confusing at first, these are:
1. Do not try to locate 1xTurning Speed, as this is low frequency data and will be
filtered out.
Turning speed should be entered using the conventional spectral data.
2. Multiple harmonics are often present within a spectrum due to the way peakvue
samples the data.
These do not indicate ‘Looseness’
3. Spectral amplitudes are always low in amplitude but should not be used to judge
severity. Use the spectrum to diagnose the fault.
4. Waveforms indicate the severity of the problem.
Peakvue - Diagnostics
• Continued…..
5. Ensure the same filter setting is used in both the spectrum and waveform.
Potential faults can be missed or overlooked if different filters are
used. 6. Cage Defects show up well in peakvue data and is normally an indication the
bearing is under stress.
7. All low frequency faults are removed from the data and will not be seen in a
Peakvue spectrum and waveform
Imbalance, Misalignment, Looseness, Resonance - All Gone.
Peakvue - Diagnostics
ANALYZE WAVEFORM 16-Mar-01 12:03:14 (PkVue- HP 500 Hz) PK = .0556 PK(+) = .5599 PK(-) = .0397 CRESTF= 14.25
0 3 6 9 12
-0.1
0
0.1
0.2
0.3
0.4
0.50.6
Revolution Number
Acc
eler
atio
n in
G-s
A1 - Example 1EX 1 -D3P Tail Roll Non D/S Peakvue
Label: Easy
ANALYZE SPECTRUM 16-Mar-01 12:03:14 (PkVue- HP 500 Hz) PK = .0484 LOAD = 100.0 RPM = 80. RPS = 1.33
0 20 40 60 80 100
0
0.004
0.008
0.012
0.016
Frequency in Hz
PK
Acc
eler
atio
n in
G-s
Freq: Ordr: Spec:
7.284 5.463 .01018
1.Spectral data indicating a defect at 5.463 Orders
2. Impacting also being detected at 0.6G-s
3. Very Slow RPM
Peakvue - Diagnostics
ANALYZE WAVEFORM 16-Mar-01 12:03:14 (PkVue- HP 500 Hz) PK = .0556 PK(+) = .5599 PK(-) = .0397 CRESTF= 14.25
0 3 6 9 12
-0.1
0
0.1
0.2
0.3
0.4
0.50.6
Revolution Number
Acc
eler
atio
n in
G-s
A1 - Example 1EX 1 -D3P Tail Roll Non D/S Peakvue
Label: Easy
ANALYZE SPECTRUM 16-Mar-01 12:03:14 (PkVue- HP 500 Hz) PK = .0484 LOAD = 100.0 RPM = 80. RPS = 1.33
0 20 40 60 80 100
0
0.004
0.008
0.012
0.016
Frequency in Hz
PK
Acc
eler
atio
n in
G-s
Freq: Ordr: Spec:
7.284 5.463 .01018
>NSK 6207 F=BPFI -IB
F F F F F F F F F F F F F
4.Fault Frequencies Indicate a BPFI Defect
Peakvue™ Amplitudes - Rolling Element Bearings
• For machines running between speeds of 900 - 3600RPM recommended
guidelines for setting initial warning levels in the Peakvue™ time -
waveform are as follows:
Alert Value Fault Value
Inner Race 3.0g's 6.0g's
Outer Race 6.0g's 12.0g's
Rolling elements fault 4.5g's 9.0g's
Cage frequencies If evident then the bearing is usually under stress.
Peakvue™ Amplitudes - Rolling Element Bearings
• For machines running at speeds <900RPM recommended guidelines for setting
initial warning levels in the Peakvue™ time- waveform are as follows:
32.7
2.52.2
1.91.6
1.3
1.0
0.60.50.3
0.1
6
5.5
5.0
4.4
3.9
3.3
2.6
1.9
1.20.9
0.50.2
4.5
4.1
3.7
3.3
2.9
2.4
2.0
1.5
0.90.7
0.40.2
0
1
2
3
4
5
6
7
900800700600500400300200100753510
Inner race -Amplitude (g's)
Outer race -Amplitude (g's)
Rolling elements -Amplitude (g's)
Levels for concern for machines running below 900 RPM
RPM
Acce
lera
tion
g's
• Cavitation will generate random, high frequency broadband energy superimposed with VPF harmonics
• Normally indicates inadequate suction pressure• Erosion of impeller vanes and pump casings may occur if left unchecked• Sounds like gravel passing through pump
• Cavitation will generate random, high frequency broadband energy superimposed with VPF harmonics
• Normally indicates inadequate suction pressure• Erosion of impeller vanes and pump casings may occur if left unchecked• Sounds like gravel passing through pump
CAVITATION
Hydraulic and Aerodynamic Forces
• If gap between vanes and casing is not equal, Blade Pass Frequency may have high amplitude
• High VPF may be present if impeller wear ring seizes on shaft• Eccentric rotor can cause amplitude at VPF to be excessive
• If gap between vanes and casing is not equal, Blade Pass Frequency may have high amplitude
• High VPF may be present if impeller wear ring seizes on shaft• Eccentric rotor can cause amplitude at VPF to be excessive
VPF = VANE PASS FREQUENCY
Hydraulic and Aerodynamic Forces
• Flow turbulence often occurs in blowers due to variations in pressure or velocity of air in ducts
• Random low frequency vibration will be generated, possibly in the 50 - 2000 CPM range
• Flow turbulence often occurs in blowers due to variations in pressure or velocity of air in ducts
• Random low frequency vibration will be generated, possibly in the 50 - 2000 CPM range
FLOW TURBULENCE
Hydraulic and Aerodynamic Forces
• Surge can occur if the pressure developed by the compressor is not equal to or greater than the downstream pressure
• Random low frequency vibration will be generated, possibly at 30-45 % of compressor of compressor speed
• Surge can occur if the pressure developed by the compressor is not equal to or greater than the downstream pressure
• Random low frequency vibration will be generated, possibly at 30-45 % of compressor of compressor speed
SURGE
Hydraulic and Aerodynamic Forces
Vibration due to Induction Motor Problems
Vibration due to Induction Motor Problems
Stator Eccentricity
Shorted Laminations
Loose Iron
Eccentric Rotor
Broken or Cracked Rotor Bars or Shorting Rings
Shorted Rotor Laminations
Loose Rotor Bars
A Typical Electrical Motor Section
Some Commonly Used Terms
• FL= Line Frequency = 50 Hz or 3000cpm
• Ns= Synchronous Speed = 120 FL/P
• P = Number of Poles 2,4or 6 etc.
• Fs = Slip Frequency = Ns-rpm
• FP = Pole Pass Frequency= #of Poles x Slip Freq.
• RBF= Rotor Bar Pass Frequency = No. of Rotor Bars x Rpm
Stator Problems Diagnosis
• Key Frequency is twice line frequency 2FL
• High amplitude at 2FL regardless of no. of poles of motor
• Needs a zoom spectrum or very high resolution spectrum to differentiate between 2xrpm and 2FL in case of 2Pole motor.
• Stator problems will often show significant vibration at rotating magnetic field frequency (rpm of motor)
• Vibration is highly directional
Motor with unequal air gap Spectrum
Rotor Problems Diagnosis
• Eccentric Rotor exhibits high vibration at 2FL
with Pole Pass Frequency side bands• Requires adjustment of bearing housing or
machining or rotor journals• Cracked or Broken rotor bars exhibits high
vibration at 1xrpm with pole pass frequency side bands.
• Loose rotor bars show vibration at RBF.
Eccentric Motor Rotor Spectrum
Vibration due to Gear Problems
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
• The following topic will discuss the basic characteristics for the following types of gears:
Spur Gears Helical Gears Bevel Gears
Spur Gears
• Spur Gears are most commonly thought of when diagnosing gears. The teeth are cut parallel to the shaft. These gears are good at power transmission and speed changes but are noisier than other gear types.
• Spur Gear Advantages– High efficiency– Low heat generation
• Spur Gear Disadvantages– Can be very noisy
Helical Gears
• Helical Gears have teeth cut at an angle to the shaft. These gears are much quieter than spur gears but due to the angular nature of the gear meshing, axial thrust and therefore axial vibration is higher than those of spur gears
– Sometimes to counter act the axial thrust these gears can be double up and are known as ‘Double Helical’ or ‘Wishbone Gears’
• Helical Gear Advantages– Quiet Operation
• Helical Gear Disadvantages– Less power transmission efficiency and
greater heat generation than spur gears– Axial loading on bearings
Bevel Gears
• Bevel Gears are used to transmit power and speed to an output shaft perpendicular to the
drive shaft. These gears use a bevel design to transmit the power better.
– These gears are most commonly seen on right angle gearboxes (where the input shaft is
at 90 degrees to the output shaft)
• Bevel Gear Advantages– Converts the direction of power transmission
• Bevel Gear Disadvantages– Less efficient
– Higher heat generation
Gear Analysis
– Vibration analysis of gears can provide a wealth of information about the
mechanical health of the gears. This section discusses the basic
frequencies that may be present within a gearbox.
• Gear Mesh Frequency Spectral Data
– The gear mesh frequency (GMF) refers to the frequency at which to
mating gears interact with each other and is the most commonly
discussed gear frequency.
– However, GMF by itself is not a defect frequency. The GMF should
always be present in the spectral data regardless of gear condition.
What is important is the amplitude as this may vary depending upon
gear condition or loading of the gear.
Gears
– Two mating gears will generate a frequency known as the GMF and
will show in the spectral data regardless of gear condition.
40 - Kiln Main DriveM4441 -G1V Shaft 01 Inboard Vertical
Route Spectrum* 08-Jun-02 23:11:51
OVERALL= 2.22 V-DG RMS = 2.14 LOAD = 100.0 RPM = 1548. (25.80 Hz)
0 200 400 600 800 1000
0
0.3
0.6
0.9
1.2
Frequency in Hz
RM
S Ve
loci
ty in
mm
/Sec
Freq: Ordr: Spec:
386.98 15.00 .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,
GMF = #teeth x turning speed
GMF = 44teeth x 1490 RPM
GMF = 65560 CPM or 65560/60 = 1092.6 Hz
INPUT
OUTPUT
Input = 1490RPM
Gear 1 = 44 Teeth
Gear 2 = 71 Teeth
Calculating GMF – Multi Reduction
– Calculating the GMF for gearboxes that have multiple trains use the following.
GMF = (#teeth) x (Turning speed)Gear Ratio = (#teeth in) / (#teeth out)Speed out = (Speed in) x (Gear Ratio)
• Example:– Consider the following gear train:
OUTPUT
Input = 1490RPM
Gear 1 = 15 teethGear 2 = 21 teeth
Gear 3 = 19 teethGear 4 = 54 teeth
INPUT
Calculating GMF – Multi Reduction
OUTPUT
Input = 1490RPM
Gear 1 = 15 teethGear 2 = 21 teeth
Gear 3 = 19 teethGear 4 = 54 teeth
INPUT
Gear Ratio 1 = 15 teeth / 21 teeth = 0.714Speed Out = 1490 RPM x 0.714 = 1064.28 RPM
Gear Ratio 2 = 19 teeth / 54 teeth = 0.351Speed Out = 1064.28 RPM x 0.351 = 374.47 RPM
GMF 1 = 1490 RPM x 15 teeth = 22350 CPMGMF 2 = 1064.28 RPM x 19 teeth = 20221.32 CPM
GMF Calculation Exercise
Calculate – Speeds of all shafts– All GMF from the following gearbox arrangement
– Gear Ratio 1 = 10/40 = 0.25– Shaft 2 speed = 1000 x 0.25 = 250 RPM– Gear Ratio 2 = 10/20 = 0.5– Shaft 3 Speed = 250 x 0.5 = 125 RPM– GMF 1 = 1000 x 10 = 10000 CPM– GMF 2 = 250 x 10 = 2500 CPM
OUTPUT
Input = 1000 RPM
Gear 1 = 10 teethGear 2 = 40 teeth
Gear 3 = 10 teethGear 4 = 20 teeth
INPUT
Gears – Sideband Frequencies
– Sidebands are the most common indication that a gear is defected.– Sidebands are equally spaced frequencies in the spectral data that materialise either side
of the main GMF peak.– The sideband frequency spacing is equal to either the turning speed of the input gear or
the turning speed of the output gear.
– Sidebands show in the data when either the gear is worn, loose or eccentric.
– The speed of the shaft with the bad gear on it will produce the most dominant sidebands in the spectral data.
FPP - SAND MILLS (OLD)AX401A -G3A Shaft 02 Inboard Axial
Route Spectrum 07-Nov-02 09:11:53 (SST-Corrected)
OVERALL= 2.18 V-DG RMS = 2.17 LOAD = 100.0 RPM = 310. (5.17 Hz)
0 8000 16000 24000
0
0.2
0.4
0.6
0.8
1.0
Frequency in CPM
RM
S V
elo
cit
y in
mm
/Se
c
Freq: Ordr: Spec: Dfrq:
18363. 59.23 .564 310.82
Gears
– The spectral data shows GMF with sideband data. – The sidebands are equally spaced at intervals of 310 CPM. This is indicating the gear that
rotates at 310 RPM is the one that is worn or damaged.
GMF
Sidebands
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)AX401A -G3A Shaft 02 Inboard Axial
Route Waveform 07-Nov-02 09:11:53
PK = .4580 LOAD = 100.0 RPM = 311. (5.19 Hz)
PK(+) = 1.27 PK(-) = 1.13 CRESTF= 3.91
0 1 2 3 4 5 6
-1.5
-1.2
-0.9
-0.6
-0.3
0
0.3
0.6
0.9
1.2
1.5
Revolution Number
Ac
ce
lera
tio
n in
G-s
Shaft Misalignment
Phase Data Indicates Parallel Shaft Misalignment
Corrective Action
General Maintenance Activities
• No predictive maintenance program is complete until it has the three basic components:
• Detection• Analysis• Correction
General Maintenance Activities
• Already discussed vibration detection and analysis of machinery faults in detail.
• Statistics indicates that a very large percentage of machinery vibrations are due to unbalance and misalignment alone.
• The balancing of rotors is possible in the field, but can also be done with dedicated machines.
• Similarly, misalignment is also a major cause of unwanted vibration. Alignment correction also requires special techniques.
• When excessive vibrations due to resonance are encountered, it is often difficult to find an easy solution to the problem.
• The use of dynamic absorbers as a possible tool for controlling resonance-induced vibration (already discussed).
• Later slides will show you the correction process of balancing and alignment faults.
Review of Balancing Process
Review of Balancing Process for
Single Plane
Sensors & Cable connecting
Confirmation of Unbalance
• Before starting balancing job, first make sure that high vibration is due to unbalance– High horizontal vibration– Dominant vibration is 1x RPM– Horizontal to Vertical ratio is less than 3:1– Horizontal Vertical Phase Difference is 90 deg– High Axial vibration (in case of overhung rotors)
• When Unbalance is confirmed, Start balancing procedure by defining Balancing job
Play
Define Job (1)
• Machine & Area information• Weight Planes: 1• Measurement Planes: 1• Measurement Points: 1• RPM of Machine: 1790• Tach. Location: 0 deg• Sensors Location: IBH• Weight Plane Setup
– Discrete Locations i.e. No of Blades: 12
Define Job (2)
Define Job (3)
Make Measurement
• Reference Run (or As Is Run)
MPT CH SPEED MAG PHASE
IBH 1 1792 RPM 147.3 µm 107 degree
Trial Run
• Calculate Trial Weight– Weight of Rotor– Radius of Trial Weight
• Attach trial weights to the rotor• Take measurement of trial run• Analyzer will give correction weights at the
end of trial run
Tolerance Check
• Enter applied weights– Applied weights (obtained from previous step) will
be entered here• Check Results
MPT CH SPEED MAG PHASE
IBH 1 1791 RPM 6.05 µm 32 degree
Trim Correction
• Further refinement can be performed by Trim Correction
• Tolerance Check will provide trim weights which can be applied to the rotor for refinement
• Effects of Trim Weights can be checked by repeating tolerance check again
Review of Shaft Alignment Procedure
Confirmation of Misalignment
• Before starting alignment job, first make sure that high vibration is due to misalignment– Dominant vibration is 2x RPM– High Axial Vibration– Across the coupling Phase Difference is 180 deg
• When Misalignment is confirmed, Start alignment procedure by defining alignment job
Alignment Procedure
Analyzer Screen
Physical Setup
• Fasten the Laser head A and B on the shaft one on each side of coupling
• Center both Lasers on by adjusting thumb wheels on the front of each sensor head
• Measure dimensions between Foot Bolts, Laser heads and coupling
Job Definition
• Enter the RPM and dimension of laser heads,
foot bolts and coupling as shown in following
figure:
Check Soft Foot
• Loose and then tight the highlighted bolt one
by one, analyzer will give severity of soft foot
by letters X, XX, XXX where number of X shows
Severity.
Acquire Data and Machine Moves
• Now take the reading by analyzer and analyzer will provide the moves for the machine
• After providing the required moves, acquire data again to verify the alignment and repeat the steps for refinement