time waveform analysis
TRANSCRIPT
1
TIME AND FREQUENCY ANALYSIS� Machinery vibration analysis techniques� Time domain analysis� Frequency analysis� Demodulation
TIME DOMAIN ANALYSIS� Use of time domain analysis� Signal processing and presentation� Phase measurement� Instrument setup� Time waveform shape analysis� Synchronous vs nonsynchronous data� Random noise and vibration� Conclusions
USE OF TIME DOMAIN ANALYSIS� A graphic description of the overall physical behavior of a
vibration structure as a function of time� Clarification of FFT processed data� The position of the measurement point at each instant of
time relative to the position at rest� Overall peak amplitude� Phase and amplitude relationships of different
frequencies and different positions� The nature of amplitude modulation or frequency content
USE OF TIME DOMAIN ANALYSIS (cont.)� The symmetry of a signal; this relates to the
linearity of the vibrating system, the nature of the forcing function, and the severity of the vibration
� A measure of damping in the system� Direction of the initial exciting force
SIGNAL PROCESSING and PRESENTATION� Instrumentation� Presentation� Presentation setup� Differential time
Waveform Characteristics
� Objectives• Describe five waveform characteristics.• Identify waveform symmetry using APD.• Discuss waveform modulation and how it translates
to the FFT.
12-1
2
Waveform Characteristics
12-1
• A number of different displays "averaging modes" use the time domain. Displays such as synchronous time averaged data is averaged in the time domain.• APD (Amplitude Problability Distribution) this is a function ofWavepak, displays the symmetry and skewness of the waveform signal.• Each defect type has a characteristic waveform, which subsequently translates to the frequency domain.• There are characteristics and specific events that do nottranslate to the frequency domain as discrete peaks. In order to truly understand this limitation, the analyst must first understand how the time domain data is gathered and transformed into a spectrum through the Fast Fourier Transform (FFT) process.
Waveform Characteristics
12-2
• Time domain data, raw transducer output, signal voltage and many other terms refer to waveforms.• Waveform or time domain data is comprised of amplitude with respect to time. Signals with an amplitude, whether vibration, current, voltage changes, or other signal types, change with time.
Waveform Characteristics
12-2
There are certain things to look for when conducting waveform analysis, the waveform provides specific characteristics for defects of a single or multiple nature.
BAL - C-20 FLOAT ROLL FAN
C-20 FLOAT-FIH FAN BEARING INBOARD HORIZONTAL
Waveform Display
25-APR-96 09:36
RMS = 1.28
LOAD = 100.0
RPM = 3550.
RPS = 59.17
PK(+) = 6.94
PK(-) = 5.84 CRESTF= 5.40
0 20 40 60 80 100 120 140
-8
-6
-4
-2
0
2
4
6
8
Time in mSecs
Acc
eler
atio
n in
G-s
Time:
Ampl:
135.09
.00000
Waveform Characteristics
12-3
Note: The waveform is only as good as its definition. If the resolution of your waveform lacks definition, the data can be worthless, or poor at best.
• Once the characteristics have been properly identified, the analyst can rule out certain fault types.
• For example:• If a waveform is periodic (sinusoidal)
• looseness• cracks• resonance• antifriction bearings
Could probably be ruled out. You may not know what the problem is, but you know what it is not.
Waveform Characteristics
12-3
Listed below are waveform characteristics an analyst should lookfor when analyzing the waveform:
• Amplitude• Asymmetry• Electrical vs Mechanical• Distortions
• Periodic• Spikes/Impacts• Non-Periodic• Modulation
• Complexity• Discontinuities• Low Frequency Events• Truncation/Restrictions to Motion
Amplitude
12-4
• When diagnosing machinery faults using the time waveform, similar to spectral data, we are concerned with the amplitude of the waveform.• When we are discussing bearing and gear waveforms, we use the peak to peak amplitude of the waveform. This is often referred to as g swing. • The g swing is the sum of the absolute value of the maximum positive and negative amplitude in that period.• MasterTrend calculates this value and gives us the ability to trend and alarm based on this and other waveform values.
3
AmplitudeC-20 - C-20 FLOAT ROLL FAN
C-20 FLOAT-FIH FAN BEARING INBOARD HORIZONTAL
Waveform Display
25-APR-96 09:36
RMS = 1.28
LOAD = 100.0
RPM = 3550.
RPS = 59.17
PK(+) = 6.94
PK(-) = 5.84
CRESTF= 5.40
0 20 40 60 80 100 120 140
-8
-6
-4
-2
0
2
4
6
8
Time in mSecs
Acc
eler
atio
n in
G-s
ALERT
ALERT
FAULT
FAULT
Time:
Ampl:
76.72
-.109
12-4
Periodic
12-5
• Sometimes referred to as a deterministic simple signal, this is an ideal signal which repeats itself exactly after a fixed period.
• This is not possible in the real world. However, there are some machinery faults which have this characteristic.
• A single plane balance problem will have a very periodic waveform due to the mass rotational center and the rotor shaft of other component center line differences.
Periodic
12-5
BAL - ZONE 6 EXHAUST
C-30 Z6X -FOH FAN BEARING OUTBOARD HORIZONTAL
Waveform Display
13-JUN-95 14:52
RMS = .1390
LOAD = 100.0
RPM = 1000.
RPS = 16.67
PK(+) = .3672
PK(-) = .4322
CRESTF= 3.11
0 100 200 300 400 500
-0.6
-0.4
-0.2
-0.0
0.2
0.4
0.6
Time in mSecs
Acc
eler
atio
n in
G-s
Complexity
12-6
• To determine the complexity of the waveform, establish whether the signal is:
• periodic in nature• estimate the harmonic content• determine if the signal is synchronous• non-synchronous• identify whether the waveform correlates directly to the spectral data.
Complexity
12-6
C-20 - C-20 FLOAT ROLL FAN
C-20 FLOAT-FOH FAN BEARING OUTBOARD HORIZONTAL
Label: LOOSE, OUT OF BALANCE
Waveform Display
25-APR-96 09:37
RMS = 1.59
LOAD = 100.0
RPM = 3508.
RPS = 58.47
PK(+) = 4.88
PK(-) = 5.25
CRESTF= 3.30
0 20 40 60 80 100 120 140
-6
-4
-2
0
2
4
6
Time in mSecs
Acc
eler
atio
n in
G-s
Impacts/Spikes
12-7
• Impacts or Spikes may or may not be repetitive in nature. • The non repetitive spikes generate white noise. • Repetitive impacts or spikes, such as those produced by rolling element bearing defects or broken gear teeth, may excite discrete frequencies and therefore show up well in the spectrum. • This characteristic is best detected by defining a waveform amplitude type in acceleration. Acceleration data is proportional to force.• The crest factor, which is equal to the maximum peak (positiveor negative) divided by the RMS of the waveform, is a good indicator of the impacting. This value can be setup as an analysis parameter and trended in MasterTrend.
4
Impacts/Spikes
12-7
Repetitive Spikes
Discontinuities
12-8
• This characteristic is usually associated with faulty equipment due to the discontinuous nature of the data.• Data with this characteristic has breaks in the data where there appears to be a loss of input signal or a significant increase/decrease in amplitude.• This is not a uniform change such as resonance, load changes, or even sudden component failures. • Discontinuous data is typically unpredictable, and very distinct.• If you see this type of waveform pattern( YOU HAVE A PROBLEM )
Discontinuities
12-8
Asymmetry
12-9
• Asymmetry refers to the relationship between the positive and negative energy. • A waveform is asymmetric when there is more energy in the positive plane than the negative or vice versa.• Asymmetry refers to the direction of movement relative to the transducer mounting with a positive signal representing energy into ( towards ) the accelerometer and a negative signal representing away.• A tool which is designed to check this type of characteristic is the APD, Amplitude Probability Distribution.
Asymmetry
12-9
MISC - #1 H2O BOOSTER
4661 -MIV MOTOR INBOARD VERTICAL
Label: LOOSE BASE
Waveform Display
16-NOV-95 10:18
RMS = .5155
LOAD = 100.0
RPM = 1789.
RPS = 29.82
PK(+) = 2.24
PK(-) = 1.43
CRESTF= 4.35
0 60 120 180 240 300
-2.0
-1.5
-1.0
-0.5
0
0.5
1.0
1.5
2.0
2.5
Time in mSecs
Acc
eler
atio
n in
G-s
Asymmetry
12-10
Select the Analyze Data feature in Diagnostics Plotting when in Waveform Analysis.
5
APD
12-10
APD
12-11
Amplitude Probability Distribution• An APD or Amplitude Probability Distribution is similar to aHystorgram. • The signal is broken down into amplitude percentages, and thenthe amplitude is plotted. • The X-Axis is the amplitude and the Y-Axis is the percentage of the signal that falls into that amplitude range. • The APD is typically used for acoustical analysis.It can also be used for machine vibration analysis to find the balance of the signal (asymmetries), the direction, and possiblythe location of a specific defect especially those that may not stand out in the waveform or the spectrum.
Sinewaves
• Sinewaves are very symmetrical, which means there is a balance of energy in the positive and negative planes.
• If most of the vibration signal is evenly distributed and sinusoidal, there is a strong possibility it is due to a synchronous component such as imbalance, misalignment, gears, blades, etc.
• The waveform and APD show the shape of asinewave and the probability related to this type of signal.
12-11
Sinewaves
12-11
• Notice that the APD at the bottom of the above display shows aset of peaks at the maximum and minimum amplitude locations. • This could also be called a Hysteresis display. The probability of the signal being in the ± 10 volt location is much more probablethat the signal being at the zero location of the display.
Triangle Wave
12-13
• With a triangle wave, we see the relationship of the waveform and a different type of APD display. • Note that the data is skewed to the negative plane. Again, this provides the analyst with the direction of motion.
The following illustration displays the direct relationship between the waveform and the APD. Bear in mind that the APD provides another tool to determine location, direction, and asymmetry.
Triangle Wave
12-13
6
Squarewave
12-14
• The squarewave on the next slide provides some insight into the use of the APD for checking asymmetries. Remember that symmetry refers to the balance of energy. Therefore, with a slightly more complex signal, this becomes more important especially when performing Root Cause Failure Analysis(RCFA).• In the next illustration, the signal is asymmetric, and there is more energy in the positive plane than the negative. • The energy in the positive plane shows movement toward the transducer, and the negative plane is obviously the opposite.
Squarewave
12-14
Truncation/Restrictions to Motion
12-15
• Truncation means to abruptly shorten, or to appear to terminate. • In waveform analysis, this characteristic indicates restrictive motion.
Modulation
12-15
• All the waveform characteristics up to this point have dealt with signals of a constant amplitude.
• A varying signal will cause the waveform to become modulated. The type of modulation occurring determines its classification. Commonly referred to as Beat frequencies, these may be broken into three specific categories.
• Amplitude• Beating• Frequency
Amplitude
12-16
• The spectrum will have a peak at the signal's frequency with one peak on each side spaced at the frequency of the amplitude change. These peaks are referred to as sidebands.• Amplitude modulation is common when analyzing inner race bearing defects. This occurs when the defective bearing component passes in and out of the bearing load zone. The middle of the load zone is typically where the highest amplitudes in the waveform show up.
Amplitude
12-16
•The spectrum and waveform show slot pass frequency from an AC induction motor. The primary signal at 34xTS is marked with a vertical line. The sideband cursors mark the amplitude changeat 120 Hz.
7
Beating
12-17
• A beat is comprised of two unrelated single frequency signals, closely spaced in frequency.• Beating is often found in two pole induction AC motors. The close proximity of two times line frequency and the second harmonic of turning speed cause this beat.• An example of beating is shown next. The 2x RPM and 2x line frequency are separated by less than .5 Hz. The waveform shows the amplitude modulation associated with beating.
Beating
12-17
WAVEFORM DISPLAY
06-DEC-94 10:15
RMS = .0678 PK(+) = .1300
PK(-) = .1790
CRESTF= 2.64
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
-0.20-0.15-0.10-0.05-0.00
0.050.10
Time in Seconds
Acc
eler
atio
n in
G-s
AMGL - CENTAC 3 STAGE COMPRESSOR
#1 CENTAC -1BA MOTOR OUTBD AXIAL TO 200 Hz
REFERENCE SPECTRUM
06-DEC-94 10:15 OVRALL= .0581 V-DG
PK = .0380
LOAD = 100.0
RPM = 3575.
RPS = 59.58
60 80 100 120 140 160 180
0
0.01
0.02
0.03
0.04
Frequency in Hz
PK
Vel
in In
/Sec
Freq:
Ordr:
Spec:
Dfrq:
119.00
1.997
.02520
1.000
Frequency
12-18
• Rarely seen in a routine environment, this is a change in frequency without a change in the signal amplitude. Frequency modulation typically occurs in gearmeshingvibration, due to the small speed fluctuations caused by tooth spacing errors and faults as they develop. A very wide spread of sidebands in the spectrum is usually an indication that significant frequency modulation is present.• On our example shown next. The vertical line in the spectrum marks gearmesh frequency at 24xTS. The sideband cursors mark the output shaft speed with labels identifying sidebands spaced at input shaft speed. The waveform has been expanded to show the frequency modulation occurring. A good illustration is shown between 170 and 180 msecs.
Frequency
12-18
Frequency modulation
Low Frequency Events
12-19
When performing detailed analysis, you need to be able to collect and analyze data in excess of one minute for low frequency problems. This is extremely important when the machine in question has an operational speed below 200 RPM. The challenge in identifying low frequency defects is having sufficient time in the waveform. A low frequency event may only appear once in the collected time domain. As discussed earlier, this event will not be transformed into the spectrum.
Low Frequency Events
12-20
8
Electrical vs. Mechanical
12-20
• Determining if the source of energy is mechanical or electrical is sometimes difficult.• Appropriately set up waveforms can be a great help. Setting up for a long enough time to capture the operational conditions and the machine shutoff point can identify the source.• The advantage of using the time domain as opposed to the frequency domain is there is no need to worry about the screen update time or sampling rate.
Electrical vs. Mechanical
12-20
Waveform and Spectrum Relationships
12-21
• Each spectrum has an associated waveform. The spectrum is made of this waveform. As discussed earlier in this section, some of the characteristics in the waveform do not translate to the FFT due to the way the calculations are made. The assumption is that there is a repetitive cycle of events made up of sines and cosines. However, this is not actually the case.• If an event happens only once, then this event has no frequency; therefore, the spectral representation is a continuous spectrum.• In the waveform shown next, there is no repetition in the event; therefore, there is no frequency.
Waveform and Spectrum Relationships
12-21
Modulated Waveforms
12-22
• Finally, when modulation is involved, there is a direct relationship between the waveform and the spectrum depending on the differential time (∆t). • Knowledge of the modulation ∆t helps determine the resolution required for detailed spectral analysis. Also, from our previous discussion on modulation, we know there is a carrier frequency that the modulation must follow.Gears, bearings, and electrical defects each have carrier frequencies. For gears the carrier is the frequency where the gears mesh. However, a carrier frequency for an electrical defect could be the line frequency (FL ) or 2 * FL.
Modulated Waveforms
12-22
9
Waveform Analysis As Confirmation
12-23
• Every fault condition has a corresponding waveform characteristic. • Unbalance, for example, has a sinusoidal pattern with one major event per revolution.• Misalignment, which is primarily offset, typically has harmonic activity with the waveform having the same number of events per cycle as the spectral data has peaks. A misalignment condition generating a second and possibly a third order peak shows two or three sinewaves per revolution.• Looseness will have a complex waveform with many peaks within one revolution. This will confirm the spectral characteristics of multiple harmonics of turning speed.
Vertical Turbine PumpUnbalance Example
12-24
MOH
MIH
MOV
MIV
Vertical Turbine PumpUnbalance Example
• The multiple point spectrum plot below shows radial and axial measurements taken from the top of the vertical motor.
FWEL - FRESH WATER BOOSTER PUMP 1
131-546-03 - PTS=MOH MOV MOA
PK
Vel
ocity
in In
/Sec
Frequency in Order
0 3 6 9 12 15 18 21 24 27
Max Amp
.65
PlotScale
0
0.7
09-FEB-96 09:22
131-546-03-MOH
09-FEB-96 09:22
131-546-03-MOV
09-FEB-96 09:22
131-546-03-MOA
12-25
Mul
ti-sp
ectra
l -D
ata
Com
paris
on
Vertical Turbine PumpUnbalance Example
12-26
• The sharpness of the peak indicates that it has been created from a waveform dominated by a single frequency.
FWEL - FRESH WATER BOOSTER PUMP 1
131-546-03-MOV MOTOR OUTBOARD VERTICAL
Route Spectrum
09-FEB-96 09:22
OVRALL= .6466 V-DG
PK = .6464
LOAD = 100.0
RPM = 1776.
RPS = 29.60
0 5 10 15 20 25 30 35 40 45 50
0
0.2
0.4
0.6
0.8
1.0
Frequency in Order
Ordr:
Freq:
Spec:
1.000
1776.2
.646
Sin
gle
Spe
ctru
m -
Am
plitu
de R
elat
ions
Vertical Turbine PumpUnbalance Example
12-27
• Approximately 270 milliseconds of time (8 shaft revolutions) shows the clear one per revolution signal generated by the unbalance condition. FWEL - FRESH WATER BOOSTER PUMP 1
131-546-03-MOV MOTOR OUTBOARD VERTICAL
Waveform Display
09-FEB-96 09:22
RMS = .2679
LOAD = 100.0
RPM = 1776.
RPS = 29.60
PK(+) = .5932
PK(-) = .6215
CRESTF= 2.32
0 60 120 180 240 300
-0.8
-0.6
-0.4
-0.2
-0.0
0.2
0.4
0.6
0.8
Time in mSecs
Acc
eler
atio
n in
G-s
Tim
e W
avef
orm
-S
inus
oida
l
Fan BearingLooseness Example
12-28
The fan bearing looseness data provides the initial spectral data for diagnostics and the waveform data to confirm the looseness fault diagnosis.
The fan ran in an out of balance condition for two years. The bearings now have excess clearance, allowing the shaft to move around.
Direct Driven Overhung Fan
10
Fan BearingLooseness Example
12-29
Note the small amounts of harmonic activity and axial data amplitude. C-20 - C-20 FLOAT ROLL FAN
C-20 FLOAT - PTS=FIH FIV FIA FOH FOV FOA
PK
Vel
ocity
in In
/Sec
Frequency in Order
0 2 4 6 8 10 12 14 16
Max Amp
.46
PlotScale
0
1.0
14-JUN-95 08:08
C-20 FLOAT-FIH
14-JUN-95 08:09
C-20 FLOAT-FIV
14-JUN-95 08:09
C-20 FLOAT-FIA
14-JUN-95 08:09
C-20 FLOAT-FOH
14-JUN-95 08:10
C-20 FLOAT-FOV
14-JUN-95 08:10
C-20 FLOAT-FOA
Ordr:
Freq:
Sp 1:
1.000
3499.0
.395
Mul
ti-sp
ectra
l -D
ata
Com
paris
onFan BearingLooseness Example
12-30
The spectral plot below shows vibration in the horizontal direction on the fan outboard bearing.
C-20 - C-20 FLOAT ROLL FAN
C-20 FLOAT-FIH FAN BEARING INBOARD HORIZONTAL
Label: HARMONICS-BALANCE/LOOSENESS
Route Spectrum
14-JUN-95 08:08
OVRALL= .5095 V-DG
PK = .5065
LOAD = 100.0
RPM = 3498.
RPS = 58.30
0 2 4 6 8 10 12 14 16
0
0.1
0.2
0.3
0.4
0.5
0.6
Frequency in Order
PK
Vel
ocity
in In
/Sec
Ordr:
Freq:
Spec:
1.000
3499.0
.395
Sin
gle
Spe
ctru
m -
Am
plitu
de R
elat
ions
Fan BearingLooseness Example
12-30
• The cursor markers note the locations of harmonics of running speed. • Virtually all the vibration energy in this spectrum is caused by turning speed and harmonics. The sides, or skirts, of this peak are also very narrow. • The number of harmonics tells us that the spectrum is derived from a complex, repetitive time waveform.
Fan BearingLooseness Example
12-31
C-20 - C-20 FLOAT ROLL FAN
C-20 FLOAT-FIH FAN BEARING INBOARD HORIZONTAL
Label: HARMONICS-BALANCE/LOOSENESS
Waveform Display
14-JUN-95 08:08
RMS = 1.06
LOAD = 100.0
RPM = 3498.
RPS = 58.30
PK(+) = 3.08
PK(-) = 3.01
CRESTF= 2.88
0 30 60 90 120 150 180 210 240
-4
-3
-2
-1
0
1
2
3
4
Time in mSecs
Acc
eler
atio
n in
G-s
Tim
e W
avef
orm
-S
inus
oida
l Cha
ract
er
Fan BearingLooseness Example
12-31
• A clear and repeatable waveform occurs once per shaft revolution, 1 x RPM. • There is also multiple peaks within one revolution The waveform shows the acceleration created on the bearing housing by the looseness. • The repeatability of the waveform in time with respect to the shaft turning speed and amplitude means that the vibration force is tied to the shaft running speed.
Motor to PumpMisalignment Example
12-32
The pump has had high vibration since installation and numerous seal/packing and bearing failures. The maintenance personnel stated that the alignment was “difficult” because the base was drilled incorrectly at the manufacturers facility.
M1HM1VM1A
M2HM2V
P1HP1V
P2HP2V
P2A
P1A M2A
11
Motor to PumpMisalignment Example
12-33
At first glance, the problem might appear to be unbalance. If we take a closer look we see that 2X running speed peaks are present in all directions. #1 - TIMBERLINE BOOSTER (PROSPECT
TIMBSTRPRO - PTS=MOH MOV MIH MIV MIA
PK
Vel
ocity
in In
/Sec
Frequency in Order
0 5 10 15 20 25 30 35 40 45 50 55
Max Amp
.43
PlotScale
0
0.5
21-JUN-95 16:11
TIMBSTRPRO-MOH
21-JUN-95 16:11
TIMBSTRPRO-MOV
21-JUN-95 16:11
TIMBSTRPRO-MIH
21-JUN-95 16:11
TIMBSTRPRO-MIV
21-JUN-95 16:12
TIMBSTRPRO-MIA
Mul
ti-S
pect
ral -
Am
plitu
de C
ompa
rison
Motor to PumpMisalignment Example
12-34
• Harmonics of running speed are denoted by the fault frequency markers (dashed lines).• The first through sixth orders of running speed are visible with the 2X T.S. predominant. #1 - TIMBERLINE BOOSTER (PROSPECT
TIMBSTRPRO-MIV MOTOR INBOARD VERTICAL
Reference Spectrum
21-JUN-95 16:11
OVRALL= .1780 V-DG
PK = .1771
LOAD = 100.0
RPM = 1768. RPS = 29.47
0 3 6 9 12 15 18 21 24 27
0
0.03
0.06
0.09
0.12
0.15
0.18
0.21
0.24
Frequency in Order
PK
Vel
oci
ty in
In/S
ec
Ordr: Freq:
Spec:
1.004 1774.9
.01562
A=MOTOR HARMONIC
: 1.00
A A A A A
Sin
gle
Spe
ctru
m -
2xTS
Motor to PumpMisalignment Example
12-35
The waveform is repetitive for each revolution with two distinctpeaks for each period. #1 - TIMBERLINE BOOSTER (PROSPECT
TIMBSTRPRO-MIV MOTOR INBOARD VERTICAL
Waveform Display
21-JUN-95 16:11
RMS = .1784
LOAD = 100.0 RPM = 1768.
RPS = 29.47
PK(+) = .5682
PK(-) = .5457
CRESTF= 3.19
0 60 120 180 240 300
-0.8
-0.6
-0.4
-0.2
-0.0
0.2
0.4
0.6
0.8
Time in mSecs
Acc
eler
atio
n in
G-s
1 2
Tim
e W
avef
orm
-Tw
ice
per
Rev
olut
ion
Pump BearingLooseness Example
12-36
• The diagram above shows a centerhung pump with bearing housing dimensions worn oversize . • The worn housings makes the pump very loose .• Typical of many looseness problems, this has grown worse over time. A small dimension problem has gradually made itself worse.
Speed 1775 RPMH.p. 150
Pump BearingLooseness Example
12-37
Many harmonics of running speed are visible on all measurement positions. Baseline or floor energy is also very visible.
CWTR - COOLING WATER PUMP 1
341-545-01 - PTS=PIV PIH POV POH POA
PK
Vel
ocity
in In
/Sec
Frequency in Hz
0 400 800 1200 1600
Max Amp
.14
PlotScale
0
0.14
18-APR-96 08:46
341-545-01-PIV
18-APR-96 08:46
341-545-01-PIH
18-APR-96 08:46
341-545-01-POV
18-APR-96 08:47
341-545-01-POH
18-APR-96 08:47
341-545-01-POA
Mul
ti-sp
ectra
l -B
road
band
Pump BearingLooseness Example
12-38
CWTR - COOLING WATER PUMP 1
341-545-01-POA PUMP OUTBOARD AXIAL
Route Spectrum
18-APR-96 08:47
OVRALL= .3663 V-DG
PK = .3675
LOAD = 100.0
RPM = 1775.
RPS = 29.58
0 400 800 1200 1600
0
0.02
0.04
0.06
0.08
0.10
0.12
0.14
Frequency in Hz
PK
Vel
ocity
in In
/Sec
Freq:
Ordr:
Spec:
29.58
1.000
.03901
Sin
gle
Spe
ctra
l -9-
15xT
S a
nd B
road
band
12
Pump BearingLooseness Example
12-38
• A cursor is positioned at 1x running speed and on the harmonics of running speed. • The peaks are broad and have wide skirts.• Notice, no individual peak exceeds .1 in/sec, but the overall energy is .3663 in/sec.• This is common with looseness. Broad humps of energy show up in the 9X to 15x running speed range. • This indicates that the time waveform cannot be cleanly transformed into a spectrum. Therefore, the waveform must have random, non-periodic energy present.
Pump BearingLooseness Example
12-39
• There is no similarity in its pattern from revolution to revolution. Non-periodic, random patterns do not convert well in the FFT process. It is very difficult to assign specific frequencies and amplitudes to patterns in waveforms like the one on the next page.• This difficulty leads to the broadband energy humps in the spectrum. Broader humps indicate more random energy. Higher humps indicate more impacting in the waveform.
Pump BearingLooseness Example
12-39
CWTR - COOLING WATER PUMP 1
341-545-01-POA PUMP OUTBOARD AXIAL
Waveform Display
18-APR-96 08:47
RMS = 2.12
LOAD = 100.0
RPM = 1775.
RPS = 29.58
PK(+) = 7.68
PK(-) = 6.42
CRESTF= 3.63
0 60 120 180 240 300
-8
-6
-4
-2
0
2
4
6
8
10
Time in mSecs
Acc
eler
atio
n in
G-s
Tim
e W
avef
orm
-R
ando
m E
nerg
y
Rolling Element Bearing Example
12-40
• Maintenance personnel reported vibration from the back end of the motor after only 200 hours operating time on a newly installed drive.• The analyst investigated and found visible flakes of a bronze colored material near the back end of the motor. The motor manufacturer was contacted to determine if the 6330 bearings had a bronze retainer and the reply given was no.
Ski Lift Motor850 HP DC Motor Right Angle Gearbox
6330 Bearings
Rolling Element Bearing Example
12-41
• All the levels appear very low in amplitude, but notice the location of the dominant peaks.• There appears to be groups of many peaks closely spaced in the mid to higher frequency range. These “mounds of energy” can indicate bearing defects.
NSTR - BACKSIDE QUAD
BACKSIDEQD - PTS=MOV MOA MOH
PK
Vel
ocity
in In
/Sec
Frequency in Hz
0 400 800 1200 1600 2000
Max Amp
.19
PlotScale
0
0.20
05-JAN-96 08:46
BACKSIDEQD-MOV
05-JAN-96 08:57
BACKSIDEQD-MOA
05-JAN-96 08:56
BACKSIDEQD-MOH
Mul
ti-sp
ectra
l -N
on-S
ynch
rono
us E
nerg
y
Rolling Element Bearing Example
12-42
• The fault frequencies for the 6330 bearing ball pass frequency outer race are marked. Notice the number of peaks surrounding the higher frequency defect harmonics. • The large number of harmonics and sidebands will be created from a complex waveform. NSTR - BACKSIDE QUAD
BACKSIDEQD-MOH MOTOR OUTBOARD HORIZONTAL
Label: OUTER RACE FREQUENCIES W/CAGE SB
Analyze Spectrum
05-JAN-96 08:56
PK = .3611 LOAD = 100.0
RPM = 1298.
RPS = 21.64
0 400 800 1200 1600 2000
0
0.06
0.12
0.18
0.24
0.30
Frequency in Hz
PK
Vel
ocit
y in
In/S
ec
Freq:
Ordr:
Spec:
77.50
3.582
.03297
>SKF 6330
C=BPFO : 77.64
C C C C C C C C C C
Sin
gle
Spe
ctru
m -
Bad
Bea
ring
13
Rolling Element Bearing Example
12-43
NSTR - BACKSIDE QUAD
BACKSIDEQD-MOH MOTOR OUTBOARD HORIZONTAL
Label: OUTER RACE FREQUENCIES W/CAGE SB
Waveform Display
05-JAN-96 08:56
RMS = 1.75
LOAD = 100.0
RPM = 1298.
RPS = 21.64
PK(+) = 5.81
PK(-) = 5.13
CRESTF= 3.32
0 40 80 120 160 200
-6
-4
-2
0
2
4
6
8
Time in mSecs
Acc
eler
atio
n in
G-s
Tim
e W
avef
orm
-B
ad B
earin
g
Rolling Element Bearing Example
12-43
� The number and height of the spikes in the time waveform confirm the presence of severe impacting.
� The waveform shape is random and complex. This shape cannot be transformed into a clean spectrum, so the spectrum on the previous page with broad humps of energy is created.
� The bearing cage turned out to be bronze! It was deteriorating and did not have much life left. The outer race had major spalls from impacting balls. The bearing was replaced.
12-44
This is an example of Unbalance.
12-44
This is an example of Unbalance.� The cursor on the previous slide is marking 1xTS
(1 Order) at 59.34 Hz in the Spectrum. � How does that frequency relate in the Waveform.� The discussions on waveform analysis are not
intended for the analyst to discard the Spectral analysis.
� The Spectrum is Amplitude vs. Frequency. The Time Waveform is Amplitude vs. Time.
12-45
Now let us look at the Waveform in “Time”.
This is an example of Unbalance.
12-45
This is an example of Unbalance.� The cursors are marking the harmonics of
the turning speed frequency, harmonic cursors was selected. The frequency at 59.34 Hz 0r 59.35 Hz. = (1 Order).
� The �time is 16.85 msec. 16.85 msecdivided by 1000 = .01685 sec, this is the Time to complete 1 revolution.
� Frequency = 1 divided by the Time
� 1 divided by .01685 = 59.347 Hz = turning speed of the rotor.
14
This is an example of Unbalance.
12-46
• Change the display to “Revolutions” of the shaft. • Notice the �time is now 1.000 that is (1 Order). • Viewing the Waveform in “Revolutions” can often make analyzing a little simpler.
Misalignment Example
12-47
• Now look at a Misalignment example.• The cursors are marking harmonics of turning speed. The peak at 2x turning speed is the highest amplitude. We have 3 or 4 peaks per revolution of the shaft in the Time Waveform.
Misalignment Example
12-48
• Take a closer look at the misalignment waveform pattern.• Harmonic cursors are marking the harmonics of what frequency?• From this display you still do not really know! You only know that these marked peaks are harmonic.
Misalignment Example
12-49
• From this display the same frequency was marked and the“Set Mark” enabled, and the “Difference” cursors was selected.
Misalignment Example� Look at the �time, it is 13.05 msec. � 13.05 msec divided by 1000 = .01305 sec� 1 divided by .01305 sec = 76.63 Hz � 76.63 Hz x 60 = 4598 rpm� When we look at the freq: 76.65 in the display at
lower right hand corner we can see the frequency has already been calculated for us. The harmonics displayed are harmonics of 76.63 Hz. This the frequency of 1xTS.
12-49
Misalignment Example
12-50
• Change the display to “Revolutions” of the shaft. • Mark the same frequency, “Set Mark” select “Difference”cursors.
15
Misalignment Example
12-51
• We can control the cursor and look at the “����time”.• The “����time” is in Orders .999 orders.• We must remember it is very difficult marking exact frequencies in the Time Waveform.
Misalignment Example
12-52
• 1x turnining speed is at 76.63 Hz. 76.63 Hz x 60 = 4598 rpm.• We can see two events occurring in 1 revolution of the shaft. • How often is the second event occurring in the Time Waveform• We will mark the 1st event, select “Set Mark”, select “Difference” cursors
Misalignment Example
12-52
� Move cursor to the peak representing the 2nd event in one revolution.– Look at the “����time” between these two frequencies. It is
6.523 msec. 6.523 divided by 1000 = .006523 sec. – 1 divided by .006523 sec = 153.3 Hz– 153.3 Hz x 60 = 9,196 rpm – The 1x TS was 4,598 rpm, 4,598 x 2 = 9,196 rpm
� Now, it is easy now to see that this frequency is occurring at 2 x TS of the rotor. It is repeated every revolution of the shaft.
Bearing Problem
12-53
� On the following slide the cursor is marking 1xTS, we have peaks at the bearing defect frequencies.
� Also displayed on the following slide is the Spectrum with Fault Frequencies for the BPFI .
� The Primary calculated defect frequency for the BPFI is 5.91 orders. There are about 10 harmonics of 5.91 orders in the spectral data.
Bearing Problem
12-53
Bearing Problem
12-54
• How do the bearing frequencies relate in the Time Waveform? • This display shows the Fault Frequencies for the BPFI displayed.
• We must realize that the dotted lines do not automatically fall on the defect frequency we may want to mark. Just any frequency was selected. Notice where the fault lines are now.
16
Bearing Problem
12-55
• In the plot displayed below the cursor was placed on a different frequency before the fault lines where brought up. We can see that the fault lines will fall where we place the cursor.
Bearing Problem� Our main concern is knowing the spacing of the defect
frequencies. This is what is displayed when we bring up the fault frequencies in the Time Waveform, the “Spacing”.
� Let us examine the Waveform further:� The Primary calculated defect frequency for the
BPFI = 134.4 HZ, so the repetition rate of the impacts would calculate to 134.4 Hz.
� We still have to find the impacts that are occurring at that spacing. This will take some time for the analyst to develop this ability to spot the equal spacing.
12-55
Bearing Problem
12-56
• When initially viewing the Waveform we look for events that arerepeated, we also look for events that are equally spaced. In this plot there are several events that are repeated and equally spaced.
Bearing Problem
12-56
• We know from the Spectral display that we have an inner race defect. Let’s display the fault frequency for the BPFI, first without a cursor marking any event.
Bearing Problem
12-57
• All we are trying to do with this display at this point is to look for impacts that may represent the BPFI. There could be BPFO’s,BSF’s also. We will focus on the BPFI’s.
Bearing Problem
12-57
• After placing the cursor on a peak we suspect is an impact froma BPFI, then displaying the fault frequency for the BPFI, we can see we have several peaks that match up.
17
Bearing Problem
12-58
• We can view an expanded plot to see this a little clearer.
Bearing Problem
12-58
• Place the cursor on an impact that matches up with a fault line. Select the “Set Mark” option. Select “Difference”cursor. Move the cursor to the next fault line, now look at the Freq: at lower right hand corner. This should be very close to the Primary Calculated Freq. for the BPFI. In this example it is very close.
Bearing Problem
12-59
• Alarms can also be utilized in Waveform analysis.Select Set-Upfrom Tool Bar and you can set the Alarms and display them in theWaveform.
Bearing Problem
12-59
Bearing Problem
12-60
• The value for the Crest-Factor has been set to Peak 1.5 for the display seen below.
Bearing Problem
12-60
18
pumps, fans, steam flow, late life bearingsrandom vibration and noise
nonsynchronous frequencies cause moving, non stationary waveformsynchronous vs nonsynchronous
truncation of signal by bearings, supports, foundations or couplings – nonlinear behaviortruncated beats
bearings, gears, rolls – natural frequencies or forcing frequency modulated by low frequency that is generated by the fault
modulated pulses
motor faults, gears, bearings – a forcing frequency is modulated by a fault frequencymodulated frequencies
bearings, recips, flat spots, gear teeth (broken) – some functional; some fault basedpulses
grinders, motor driven fans, pumps where two forcing frequencies are closebeats
generators (slot passing), gears, vane pass, bearings, naturally generated harmonics superimposed on 1xmultiple harmonics
heavy 1x behavior can excite order located natural frequenciesorder excited natural frequencies
misalignment, looseness, generator faultsorders
rubs, oil whirl, resonance, trapped fluid hysteresis, loosenesssubharmonics
gear mesh, blade pass, natural frequencies, nonlinear behaviortruncated harmonics
excessive mass unbalance, thermal growth, bearing clearance problems, pedestal nonlinearity, rubstruncated 1x
mass unbalance, resonance, eccentricity, misalignment, bow, blade/diffuser interactionharmonic
MECHANISMSHAPE
TABLE 4.5. TIME WAVEFORM SHAPE ANALYSIS
Part 1 - Summary
12-61
• Waveform data may be used for much more than what is typically seen in industry. The ability to check for specific characteristics such as periodicity and modulation, helps the analysis process.
• Energy balance (asymmetry) may be checked for direction of signal and for the predominant traits of the signal.
• Overall waveform is much more understandable and useful than most would lead us to believe. However, this section enhances your analysis abilities using the time waveform.
DIGITIZED TIME DOMAIN— TRENDS
What is this spectrum Lines?
DIGITIZED TIME DOMAIN— DETAILS
What is this spectrum Lines?
PRESENTATION OF TIME WAVEFORM
TIME (sec.) DISPLAY PURPOSE
T/100 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - DETAILS OFT/80 HIGHER FREQUENCYT/20
T/10 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -- -TRENDS OFT/3 HIGHER FREQUENCYT/2
T - - - - - - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - BALANCING/PHASE2T3T
10 T- - - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - PHASE TRENDS20T80T
100T - - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - LONG TERM TIME TRENDS
Table 4.4. An Approach to the Presentation of a Standardized Time Waveform4.1
TIME DOMAIN WAVEFORM — TREND
4000 HP Induction Motor with 200T Display
19
TIME DOMAIN WAVEFORM — SHORT TERM
4000 HP Electric Motor with 20T Display
TIME DOMAIN WAVEFORM — BALANCING
4000 HP Electric Motor with T Display
TIME DISPLAY — SMALL MOTOR 400T
Small Motor with a 400T Display
SMALL MOTOR 100T
Small Motor with a 100T Display
GEARBOX 10T
Gearbox with 10T Display
GEARBOX — 1T ON MESH
Gearbox with T Display Shows Gearmeshing Effects
20
PRESENTATION SETUP� Visual process� Setup to accommodate visual analysis
– to evaluate• periodicity
– to evaluate amplitude changes
� Processing types– dual processing– expansion
STANDARD SETUP
Standard Time Waveform Display from an FFT Analyzer
DUAL PROCESSING
Dual Processing to Enhance the Time Waveform
� SPECTRUM:
– 10x operating speed
– fmax = 250 Hz
� TIME WAVEFORM:
– Period = = 0.0421
– Display = 8 cycles x 0.0421 0.336 sec
– Use 0.4 sec then
DUAL PROCESSING
�����
���
���
�
�����
���
� �����������
������������ ==
Dual Processing means:the capability to produce eachSpectrum and Time waveform data independently
(spectrum Fmax not equal to waveform Fmax)
PHASE MEASUREMENT
Use of Cursor to Measure Period and Amplitude
TRUNCATED 1X
Single Frequency Due to Mass Unbalance withClipping Due to Excessive Amplitude
21
TRUNCATED HARMONIC
Clipped Vane Pass Signal from Hull of a Ship
SUBHARMONIC
Loose Bearing Housing — ½ Orders and Multiples
ORDERS
Nonlinear Generator Pedestal Response to Differing Vertical Stiffness
FAR REMOVED ORDER
Slot Passing Frequency, 36X, Generated by Air Gap Variation (120Hz)
BEATS
Beats Caused by Two Adjacent Cavitated Vacuum Pumps
PULSES
Pulses from a Bearing Defect
22
AMPLITUDE MODULATION
Sidebands Caused by Amplitude Modulation — Broken Rotor Bar
PULSE INDUCED NATURAL FREQUENCIES
Pulse Induced Natural Frequencies in Printing Roll
TRUNCATED BEATS
Truncated Beat Waveform from a Motor Driven Fan
SYNCHRONOUS ORDERS
Exciter to Generator Misalignment Causing 1x and 2x
RANDOM NOISE and VIBRATION — RMS AVERAGING
Excessive Flow Noise Caused byPump Operating Off the Curve-Recirculation
CONCLUSIONS
� True physical behavior� Determine origin of frequencies� Determine severity
23
NONSYNCHRONOUS MULTIPLE FREQUENCIES
Boiler Feed Pump Drive — Nonsynchronous Second Order and Multiples
SINUSOIDAL AMPLITUDEMODULATION
Amplitude Modulation by a Single Frequency
NONSINUSOIDAL AMPLITUDE MODULATION
Amplitude Modulation in a Gearbox — Nonsinusoidal
MACHINE RESPONSE TO IMPACT EXCITATION
Response of a Machine to Impulse Excitation
IMPACT INDUCED NATURAL FREQUENCIES
Broken Gear Tooth Yielding Pulse Induced Natural Frequencies andAbrupt Pulse Loading in Time Domain
LIGHT IMPACTINDUCED ORDERS
Gradual Entering Pulse Caused by Misalignment, Eccentricity, or Tooth Profile Wear in a Gearbox
24
FREQUENCY MODULATION
Torsional Vibration a Form of Frequency Modulation
DIFFERENCE FREQUENCIES
Two Lobed Blower Generated Difference Frequencies — Pressure Pulsations Generated by Lobes Passing Discharge Port
MECHANISMS FORORDER GENERATION
� Natural excitation� Nonlinear parameters� Signal truncation
BEAT MECHANISM
Figure 4.50. Beat Mechanism
TRUNCATED BEATS
Hypothetical Vibration Response Exhibiting Beat Frequency
SUM and DIFFERENCEFREQUENCY TABLE
25
SUM and DIFFERENCEFREQUENCY MECHANISMS
Rotating Machinery Fault Diagnosis Using Sum and Difference Frequencies (Sidebands) (After Eshleman 4.2)
SUM and DIFFERENCE FREQUENCIES —PISTON PUMP/ENGINE
Pulsating Torque from an Engine-Pump Unit
SUM and DIFFERENCE FREQUENCIES —TURBINE PIPING
Frequency Domain Record of Line ShaftTurbine Piping Vibration
SPECTRUM SHAPE
Misalignment Induced Air Gap Vibration
DEMODULATION
� Mechanism� Rolling element bearings� Techniques
MODULATION MECHANISM� High frequency vibration amplitude is altered due
to mechanical defects� Low frequency modulator points to problem area� Used extensively for gears and bearings
26
GEARBOX MODULATION
Amplitude Modulation in a Gearbox — Nonsinusoidal
IMPACT INDUCED MODULATION
Pulse Induced Natural Frequencies in Printing Roll
DEMODULATION
Figure 4.58. Spectrum of a Demodulated Signal
BEARING DEFECT
Outer Race Defect on Rolling Element Bearing — Pulsation Effect
BEARING DEFECT FREQUENCY ZONES
Machine Vibration Response to Bearing Faults
DEMODULATIONOF A SIGNAL
Demodulation of a Signal ©© Copyright 2002 by Ronald L. Eshleman. All rights reserved. This figure may not be reproduced without permission of
Ronald L. Eshleman.