mike cannon automated bearing dli engineering wear … auto bearing fault.pdf · 2016-08-04 · d l...
TRANSCRIPT
©D
LIE
ngr
Cor
p-
1
Automated BearingWear Detection
Mike Cannon
DLI Engineering
©D
LIE
ngr
Cor
p-
2
Vibration: an indicator of machine condition
©D
LIE
ngr
Cor
p-
3
Vibration Contents by Frequency
FFT Spectral Analysis
Identify Peaks & Patterns -
Relationship to
machinery components
Amplitude of peak – comparison tobaseline
Diagnosis and severity of machine problem
Repair recommendations –
Specific to problem
Priority and Remaining life
Narrow band Vibration Analysis
©D
LIE
ngr
Cor
p-
4
Time Domain & Frequency Domain
Vibration Wave Form
F r e q u e n c yA m p l i t u d e v s . t i m e
V i b r a t i o n S p e c t r u m
A complex signal is the sum of several sine waves
©D
LIE
ngr
Cor
p-
5
Vibration SignalVibration Signal
Vibration Data Acquisition & ProcessingAnalog SignalAnalog Signal
is Digitizedis Digitized
©D
LIE
ngr
Cor
p-
6
Forcing Frequencies Rotation Rate of Fan
Five-Blades on Fan
Fly swatter against the blades
©D
LIE
ngr
Cor
p-
7
18
Forcing Frequencies in Spectra
0 1 2 3 4 5 6 7 8 9 10
10 Orders
IPS
1X RPMFly Swatter
©D
LIE
ngr
Cor
p-
8
Results = Problems Identified
Imbalance Misalignment Bearings (BAD /COCKED)
Bent shaft Looseness
Out-of-round Journal Gear Problems Impeller Blade Problems Magnetic Noise Electrical Noise
Motor
GearCprsr
Detection
Diagnosis &
Prognosis
of specific problems
©D
LIE
ngr
Cor
p-
9
Accurate results from quality data
Triaxial Sensor
All 3 axes - more complete analysis
Improves accuracy of diagnosis
Permanently Mounted Stud
Excellent frequency response
Repeatability = accurate trending
Barcoding
Faster and more accurate
Prevents human error
100 100
WATCHMAN TEST POINT
DO NOT PAINT
100 100
WATCHMAN TEST POINT
DO NOT PAINT
©D
LIE
ngr
Cor
p-
10
Baseline Data to define the healthy machine
800 Alarm Levels
©D
LIE
ngr
Cor
p-
11
Order Normalization
Conventional Vibration Spectrum1800 rpm component
140
130
120
100
90
80
70
60
110-
0 3 6 9 12 15 18
Frequency 0-18,000 RPM
Am
plitu
deV
dB
©D
LIE
ngr
Cor
p-
12
Order Normalization
Order Normalized Vibration Spectrum
0 1 2 3 4 5 6 7 8 9 10
140
130
120
110-
100
90
80
70
60
Frequency in Orders (multiples of rotation rate)
Am
plitu
dein
VdB
©D
LIE
ngr
Cor
p-
13
Results from Vibration Analysis
Level IReports
Level IReports
Level IITrends
Level IITrends
Level IIISignatures
Level IIISignatures
©D
LIE
ngr
Cor
p-
14
1. MANDATORY Repair Recommendations
AIR CONDITIONING SALTWATER CIRCULATING PUMP #3Mandatory: Overhaul unit.
AUXILIARY MACHINERY COOLING WATER PUMP #2BMandatory: Align Motor and Pump.
2. IMPORTANT Repair Recommendations
MAIN CONDENSATE PUMP (MD) #1AImportant: Replace motor bearings.Desirable: Monitor pump bearings for increased vibration.
LUBE OIL SERVICE PUMP (ATTACHED) #1CImportant: Inspect bevel gears for proper meshing; ifsatisfactory, check intermediate shaft bearings for looseness.
TURBO-GENERATOR (60HZ) #8Important: Inspect accessory drive shaft for proper bearing fit and clearances.
MCA Results - Summary Report
©D
LIE
ngr
Cor
p-
15
Trend Reports
Greased Bearing
Replaced Bearing
Increased FrequencyIncreased Frequencyof Data Collectionof Data Collection
©D
LIE
ngr
Cor
p-
16
FFT spectrum, cepstrum &demod
Triaxial, order-normalized data
Machine specific, statisticalbaselines
Feature extraction &exceedance tables
4700 fault templates
Score & trend specific faultseverities
Input
Vibration sensors
Accelerometers & proxprobes
Process andperformance data
Portable or Onlinesystems
OSI Pi Database
Internal Functions Output
Specific faults & severity
Repair recommendations
Email & pager ALERTs
Network replication
OPC & XML connectivity
Web Connect browseraccess
Automated Diagnostic SystemsAutomated Diagnostic Systems
©D
LIE
ngr
Cor
p-
17
Incoming Spectra
Machinery Fault ReportMachinery Fault Report
What machine is this &what are its vibration sources?
How does this spectra compareto that of a healthy machine?
What, if anything, is wrong withthis machine and how bad is it?
Complete sets of high resolution vibe datafor FFT, cepstrum & demod
Machine configuration data,(VTAGs) coupled withcomponent specific fault templates
Detection diagnosis & prognosis
Automated Diagnostic System Flow Chart
Actionable Information
Feature extraction andexceedance table
©D
LIE
ngr
Cor
p-
18 15Couplings
Component Codes
1Motor Driven
Close CoupledPumps/Fans
2Turbines
4Gearboxes
5Linked Drives(Belt/chain)
6Pumps
Centrifugal
7Rotary Thread
Pumps
8Sliding Vane
Pumps
9Axial Piston
Pumps
10Fans
11Centrifugal
Compressors
12Piston
Compressors
13Generators
14Centrifugal
Purifiers
NOTE:This illustrated set
is not complete
3Motors
3Motors
15Couplings
6Pumps
Centrifugal
©D
LIE
ngr
Cor
p-
19
Features from comparison to baseline
AVERAGEAVERAGEAVERAGE AVERAGE + 1 SIGMAAVERAGE + 1 SIGMAAVERAGE + 1 SIGMAAVERAGE + 2 SIGMASAVERAGE + 2 SIGMASAVERAGE + 2 SIGMAS
©D
LIE
ngr
Cor
p-
20
Extracted Forcing Frequencies
©D
LIE
ngr
Cor
p-
21
Extracted Bearing Tones (3.1x)
©D
LIE
ngr
Cor
p-
22
Fault Template Diagnosis Rules
Misalignment (Parallel)
1. 2X Radial or Tangential > 95VdB&
> 2 VdB above Baseline on both sides of the coupling
2. 2X Radial > 1X Radialor
2x Tangential > 1X Tangential for one side of the coupling
3. The Maximum 2X > 101VdBor
The sum of 2X exceedances > 18 VdB for at least one side ofthe coupling.
©D
LIE
ngr
Cor
p-
23
CSDMCSDM
CSDMCSDM
CSDMCSDM
MOTORMOTOR
COUPLINGCOUPLING
CENTRIFUGAL PUMPCENTRIFUGAL PUMP
COMPONENTCOMPONENTCODESCODES
SCREENINGSCREENINGOUTPUT TABLEOUTPUT TABLE
(CSDM = Component Specific Data Matrix)(CSDM = Component Specific Data Matrix)
MACHINEDATA
MACHINEDATA
FAULTFAULTTEMPLATESTEMPLATES
Fault Models / Diagnostic Flow
©D
LIE
ngr
Cor
p-
24
Results from Automated Diagnostics
Provides the most efficient machinery fault detection,diagnosis & prognosis to insure: Immediate local fault detection & actionable information
Optimal asset reliability and machine life extension throughprognosis of how & when machine should be repaired
©D
LIE
ngr
Cor
p-
25
Rolling Element Bearing DetectionFault Indicators in Vibration Spectra
Bearing TonesNon-integer peaks and harmonicsrelative to component rotation rate
Sidebands
Noise floor elevation
High frequency indicatorsImpact and RingingModulation
©D
LIE
ngr
Cor
p-
26
Snap ring groove
Inner race
Inner ring
Cage
Outer race
Outer ring
Seal/shieldgroove & notch
Bore
Inner ringface
Outer ringface
Outsidediameter
Ball
Rolling Element Bearing Geometry
©D
LIE
ngr
Cor
p-
27
Bearing Tones
Pitc
hD
iam
eter
(PD
)
Ball Diameter (BD)
Ball Pass Frequency – Inner Race
BPFI
Ball Pass Frequency – Outer Race
BPFO
Ball Spin Frequency
BSF
Bearing Train Frequency
FTF
©D
LIE
ngr
Cor
p-
28
Rolling Element Bearings
©D
LIE
ngr
Cor
p-
29
Bearing Tones
Pitc
hD
iam
eter
(PD
)
Ball Diameter (BD)
BPFIn B
PRPM
BPFOn B
PRPM
BSFPB
BP
RPM
FTFBP
RPM
d
d
d
d
d
d
d
d
d
d
21
21
21
12
1
22
cos
cos
cos
cos
©D
LIE
ngr
Cor
p-
30
Simplified Bearing Tones
Non-Synchronous Components BPFO = Rollers x rpm x 0.4
BPFI = Rollers x rpm x 0.6
FTF = rpm x 0.4
Sidebands Usually 1X Around BPFI
May be FTF Around Bearing Tones
©D
LIE
ngr
Cor
p-
31
Rolling Element Bearings
For each shaft revolution:
4.6 balls will pass an outer race defect
The ball will make 2.02 revolutions
7.4 balls will pass an inner race defect
The cage will make 0.38 revolutions
©D
LIE
ngr
Cor
p-
32
Calculated Bearing Tones
Bearing tones do not always appear at the expected,calculated frequencies
User does not always have correct bearing information
Diagnostic System designed to work independently ofcalculated bearing frequencies
User may overlay calculated frequencies on graph ifdesired
©D
LIE
ngr
Cor
p-
33
Non-Synchronous Tones
Shaft Rotation
Shaft RateHarmonics
Bearing Tone 4.6 x
Sidebands 2.6x, 3.6x, 5.6x
0 1 2 3 4 5 660
70
80
90
100
110
120
130
140
Orders
VdBBearing Tone -- 4.6X
Synchronous Componentsare integer multiples ofshaft rotation rate
Bearing tones arenon-integer multiplesof shaft rate
©D
LIE
ngr
Cor
p-
34
Automated Spectral Peak Extraction
Identify Shaft Rate Peak(s)
Identify Harmonics of Shaft Rate(s)
Identify Forcing Frequencies
Identify External Vibration
Extract non-defined, non-synchronous peaks aspossible bearing tone candidates.
©D
LIE
ngr
Cor
p-
35
Summary of Analysis Techniques
Data Normalization
Automated Spectral Peak Extraction
Cepstrum Analysis
Noise Floor Calculation
Rule-Based Diagnostics
Demodulation
©D
LIE
ngr
Cor
p-
36
Cepstrum Analysis
The Spectrum of the Spectrum
0 10 20 30 40
CPM (x 1000)
60
70
80
90
100
110
120
130
140VdB
.006 .003 .002 .0015
Time, Seconds
60
70
80
90
100
110
120
130
140CdB
1X HarmonicsBearing Tones
©D
LIE
ngr
Cor
p-
37
The Cepstrum
Extracts Periodicities From the Spectrum
Separates Different Harmonic Series
Separates Sideband Families
Is Useful in Automating Bearing Wear Detection
Bearing Model Is Not Required!!!
Is this bearing tone candidate part of a family of peaks?
“#” = “Yes”
©D
LIE
ngr
Cor
p-
38
Spectral Data: Bearing Tone at 3.1x
©D
LIE
ngr
Cor
p-
39
Put it all together (Cont’d)
3.1X 6.2X 24.8X9.3X
52.7X
15.5X12.5X
©D
LIE
ngr
Cor
p-
40
Extracted Bearing Tones (3.1x)
©D
LIE
ngr
Cor
p-
41
Summary of Analysis Techniques
Data Normalization
Automated Spectral Peak Extraction
Cepstrum Analysis
Noise Floor Calculation
Rule-Based Diagnostics
Demodulation
©D
LIE
ngr
Cor
p-
42
High Noise Floor
0 2 4 6 8 10 12
CPM ( X 1000 )
60
70
80
90
100
110
120VdB
©D
LIE
ngr
Cor
p-
43
Summary of Analysis Techniques
Data Normalization
Automated Spectral Peak Extraction
Cepstrum Analysis
Noise Floor Calculation
Rule-Based Diagnostics
Demodulation
©D
LIE
ngr
Cor
p-
44
Amplitude Modulated Wave Form and itsSpectrum
©D
LIE
ngr
Cor
p-
45
Bearing Impact Faults
High-Frequency Damped Oscillation
Acceleration Wave Form
Spall
Demodulaterd Waveform
©D
LIE
ngr
Cor
p-
46
Demodulation Bearing Analysis
©D
LIE
ngr
Cor
p-
47
Bearing Faults
BPFO Frequency
©D
LIE
ngr
Cor
p-
48
Demodulation Data: Bearing Tone at 3.1x
©D
LIE
ngr
Cor
p-
49
Amplitude Demodulation
Early Warning of Bearing Wear
Confirmation of Bearing Wear
Better Information on Which Bearing is Defective
©D
LIE
ngr
Cor
p-
50
Bearing Faults…Typical Ranges
©D
LIE
ngr
Cor
p-
51
Diagnostic Report
FIRE PUMP #2Acquired: 8/19/03 10:35:38 AM 1xM = 1191 RPM Averages: 3RECOMMENDATIONS:MANDATORY: REPLACE MOTOR BEARINGSDIAGNOSES:EXTREME MOTOR BEARING WEAR
0.056 (+0.042) in/s at 3.11xM on 2A in low range0.028 (+0.025) in/s at 6.22xM on 2A in low range0.039 (+0.030) in/s at 9.33xM on 2R in low range0.079 (+0.077) in/s at 9.33xM on 2T in low range0.001 (+0.000) in/s at 24.8xM on 2A in high range0.088 (+0.084) in/s at 3.11xM on 2R in low range0.022 (+0.021) in/s at 15.5xM on 2R in high range0.022 (+0.011) in/s at 12.5xM on 2R in high range0.039 (+0.036) in/s at 3.11xM on 2T in low range0.001 (+0.001) in/s noise floor0.002 (+0.001) in/s noise floor0.001 (+0.001) in/s noise floor
MOTOR BALL BEARING DEMODBearing tone harmonics in demod spectrum with direct match between
regular and demod spectrum in 2R21 dB peak at 9.33x on 2R
©D
LIE
ngr
Cor
p-
52
Put it all together
©D
LIE
ngr
Cor
p-
53
Bearing – The Movie
The life and times of a troubled bearing
©D
LIE
ngr
Cor
p-
54
Progression of Bearing Faults
0 1 2 3 4 5 6
ORDERS
60
70
80
90
100
110
120
VdB Machine without Bearing Problems
11X
2X
3X4X
©D
LIE
ngr
Cor
p-
55
Progression of Bearing Faults
0 1 2 3 4 5 6
ORDERS
60
70
80
90
100
110
120
VdB Slight Bearing Tones at 3.1X
1X
2X
3X4X
©D
LIE
ngr
Cor
p-
56
Progression of Bearing Faults
0 1 2 3 4 5 6
ORDERS
60
70
80
90
100
110
120
VdB As defect worsens - 2nd harmonic rises
1X
2X
3X4X
©D
LIE
ngr
Cor
p-
57
Progression of Bearing Faults
0 1 2 3 4 10 12
ORDERS
60
70
80
90
100
110
120
VdB Increased machine looseness - 1X harmonics increase
1X
2X
3X
4X
©D
LIE
ngr
Cor
p-
58
Progression of Bearing Faults
0 2 4 6 8 10 12
ORDERS
60
70
80
90
100
110
120
VdB 1X Sidebands around the Bearing Tone
1X
2X
3X4X
©D
LIE
ngr
Cor
p-
59
Progression of Bearing Faults
0 2 4 6 8 10 12
ORDERS
60
70
80
90
100
110
120
VdB “Haystack” in Noise Floor appears as Wear Progresses
1X
2X
3X4X
©D
LIE
ngr
Cor
p-
60
Bearing Tone Sidebands
0 1 2 3 4 5 660
70
80
100
110
120
VdB
ORDERS
90
Bearing Tone
Sidebands spaced at FTF
140
130
©D
LIE
ngr
Cor
p-
61
Progression of Bearing Faults
0 2 4 6 8 10 12
ORDERS
60
70
80
90
100
110
120
VdB Entire Noise Floor rises in advanced stages
1X
2X
3X
4X
©D
LIE
ngr
Cor
p-
62
Progression of Bearing Faults
0 2 4 6 8 10 12
ORDERS
60
70
80
90
100
110
120
VdB Machine Failure
©D
LIE
ngr
Cor
p-
63
Summary of Analysis Techniques
Data Normalization
Automated Spectral Peak Extraction
Cepstrum Analysis
Noise Floor Calculation
Rule-Based Diagnostics
Demodulation
©D
LIE
ngr
Cor
p-
64
Key elements in a successful MCA
Information, not Data Measurement Methodology
Triaxial Data Collection with mounting pad (Complete data set &improved high frequency signal: Portable/Walk-around approach)
Superior Data Screening (Statistic baseline in hi/low freq range &automated, narrow band data comparison)
Analysis / Diagnostics / Prognosis
Automated Diagnostic Report with severity, fault and prioritizedrepair recommendation (information NOT just data)
Information Everywhere Distribution of Information
Simple, machine condition report/fault notification available via:
Internet or automated pager, email, text message
Delivering Machine Intelligence
©D
LIE
ngr
Cor
p-
65
Delivering Machine Intelligence
Visit our web site at:www.DLIengineering.com