Investigation of wide band Fiber Bragg
grating accelerometer use for rotating AC
machinery condition monitoring
Sinisa Djurovic a, Peter Kung b et al.
a
School of Electrical and Electronic Engineering, The University of Manchester, UK;
b QPS Photronics, Pointe Claire, Canada.
Outline
1. Introduction
2. Motivation---Wind generator failure statistics
3. FBG wide band vibration monitoring system
4. Generator test rig
5. Bearing fault specific vibration frequencies
6. Measured Vibration spectra under bearing fault conditions
7. Generator electrical fault specific vibration frequencies
8. Measured Vibration spectrum for stator short-circuit and open
circuit fault
9. Conclusions
Motivation
• Vibration analysis frequently used to detect
mechanical and electrical faults in rotating
machines
• Wind generators larger than 2MW show
combined bearing and stator winding failures of
more than 73%
• Cost effective and robust wide bandwidth
vibrations required to replace currently used
costly piezoelectric (PE) accelerometers
Fibre optics Vibration sensing
• Immune to EMI and high voltages
• Longer life
• Proven for monitoring low frequency end
winding vibration in large generators
• Direct electrical output
• Distributed vibration sensing using Long
Gauge technology
• Single sensor measures both temperature
and vibration
Large wind turbines failure statistics
UpWind: Design limits and solutions for very large wind turbines, EWEA 2011
Wind generator failure modes
1-2 MW
>2 MW
Wide Band Vibrofibre
Bandwidth of ≈1000 Hz to cover bearing,
stator/rotor electrical faults frequencies achieved
by changing the diving board material to
Polycarbonate
Fiber Cavity
Cross Section
Measured Vibration Amplitude (um)
Performance and Characteristics
Interrogator unit and software
interface
15
10
5
0
20
40
60
80
Applied Vibration Amplitude (um)
100
Response at 100 Hz vibration frequency
Measured Vibration Acceleration (m/s2)
Accelerometer Reflection Spectrum
20
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
0
200
400
600
800 1000 1200 1400
Frequency (Hz)
FBG accelerometer frequency response
Wind turbine generator test rig
Laboratory test bed
(viewed from above)
Stator terminal
box
Drive end
bearing
Sensor performance testing
• Benchmarked against Bruel&Kjaer Pulse vibration
platform utilising piezo-electric(PE) accelerometers
• Both sensor types mounted on generator frame
• Typical winding and bearing faults artificially
introduced
Sensor mounting
Rolling bearing race frequencies
Outer race

N  D
f o  f r b 1  b cos  
2  Dc

Inner race

Nb  Db
fi  f r
cos  
1 
2  Dc

fr=rotational frequency, Nb=number of rolling elements
Db=ball diameter, Dc=cage diameter, β=contact angle
Bearing Fault Emulation
• Various severity of bearing outer race fault
introduced in experiments
Drive-end
SKF 6313
Nb = 8
fo=3.07fr
fi=4.93fr
Machined bearing fault
(localised outer race fault)
Laboratory generator bearing
design data
Spectrum showing bearing fault effects
B&K, bearing fault vibration spectrum, 1590 rpm
0
10
3mm fault
healthy
-1
163.3
244.8
326.6
489.8
408.1
734.7
571.4
979.5
653.1
10
Acceleration [m/s 2]
898
816.4
-2
10
-3
10
-4
10
0
100
200
300
400
500
Frequency [Hz]
600
700
800
900
1000
QPS, bearing fault vibration spectrum, 1590 rpm
0
10
-1
3mm fault
healthy
326.3
163.3
Acceleration [m/s 2]
10
489.8
244.8
571.4
734.7
816.4
898
979.5
408.1
653.1
-2
10
-3
10
-4
10
0
100
200
300
400
500
Frequency [Hz]
600
700
800
900
Piezoelectric (top) and FBG (bottom) vibration frequency spectra
3mm outer race bearing fault at 1590 rpm
1000
Zoom-in View
B&K, bearing fault vibration spectrum, 1590 rpm
0.12
healthy
3mm fault
2
571.4
489.8
0.08
PE sensor
0.06 408.1
0.04
0.02
420
440
460
480
500
520
Frequency [Hz]
540
560
580
600
QPS, bearing fault vibration spectrum, 1590 rpm
0.08
489.8
2
0
400
Acceleration [m/s ]
Acceleration [m/s ]
0.1
FBG sensor
0.06
healthy
3mm fault
571.4
0.04
408.1
0.02
0
400
420
440
460
480
500
Frequency [Hz]
520
540
560
580
600
Typical Stator Winding Faults
U1U1 U1
U2U2 U2
W1
W1W1
V2V2 V2
W2W2
V1V1 V1 W2
Winding configurations:
Winding
a)a) a)
Healthy
Torque frequencies
6k 1  s  f s
Balanced
Unbalanced
2  6k 1  s  f s
k
1  s  f s
p
k
2  1  s  f s
p
U1U1 U1
U2U2 U2
V2V2 V2
W1
W1W1
V2V2 V2
W2W2
V1V1 V1 W2
b)b) b)fault
Open-circuit
U2U2 U2
W1
W1W1
W2W2
V1V1 V1 W2
c)c) c) fault
Short-circuit
• Achieved experimentally
using stator terminal box
k=0,1,2,3…
s=slip
p=pole pairs
fs=supply frequency
Spectrum showing short circuit fault effects
B&K, stator winding short-circuit fault, 1590 rpm
0
10
58.59
[(3),k=6]
158.8
[(2),k=6]
376.1
[(3),k=18]
258.8
[(3),k=6]
476.1
[(2),k=18]
576.3
[(3),k=18]
693.6
[(3),k=30]
-1
Acceleration [m/s 2]
10
893.8
[(3),k=30]
793.6
[(2),k=30]
-2
10
-3
10
short-circuit
healthy
-4
10
0
100
200
300
400
500
Frequency [Hz]
600
700
800
900
1000
QPS, stator winding stator short-circuit fault, 1590 rpm
0
10
258.8
[(3),k=6]
158.8
58.59
[(2),k=6]
[(3),k=6]
376.1
[(3),k=18]
476.1
[(2),k=18]
576.3
[(3),k=18]
693.6
[(3),k=30]
-1
Acceleration [m/s 2]
10
893.8
[(3),k=30]
793.6
[(2),k=30]
-2
10
-3
10
short-circuit
healthy
-4
10
0
100
200
300
400
500
Frequency [Hz]
600
700
800
900
Piezoelectric (top) and FBG (bottom) vibration frequency spectra
stator short-circuit fault at 1590 rpm
1000
Spectrum showing open circuit fault effects
B&K, stator winding open-circuit fault, 1590 rpm
1
10
0
58.59
[(3),k=6]
Acceleration [m/s 2]
10
258.8
[(3),k=6]
158.8
[(2),k=6]
376.1
[(3),k=18]
476.1
[(2),k=18]
576.3
[(3),k=18]
693.6
[(3),k=30]
793.6
[(2),k=30]
893.8
[(3),k=30]
-1
10
-2
10
-3
10
open-circuit
healthy
-4
10
0
100
200
300
400
500
Frequency [Hz]
600
700
800
900
1000
QPS, stator winding stator open-circuit fault, 1590 rpm
1
10
0
58.59
[(3),k=6]
Acceleration [m/s 2]
10
158.8
[(2),k=6]
376.1
[(3),k=18]
258.8
[(3),k=6]
476.1
[(2),k=18]
576.3
[(3),k=18]
693.6
[(3),k=30]
793.6
[(2),k=30]
893.8
[(3),k=30]
-1
10
-2
10
-3
10
open-circuit
healthy
-4
10
0
100
200
300
400
500
Frequency [Hz]
600
700
800
900
Piezoelectric (top) and FBG (bottom) vibration frequency spectra
stator open-circuit fault at 1590 rpm
1000
Variable Speed Operation
• Emulating realistic wind turbine variable speed
operating conditions
• Open-circuit winding fault introduced and PE
and FBG platforms compared
1500
Speed [RPM]
1480
1460
1440
1420
1400
1380
0
2
4
6
8
10
Time [s]
12
14
Typical measured generator speed profile
16
18
20
Transient vibration signal spectrum
1000
HEALTHY
900
FAULTY
800
Frequency [Hz]
700
F
A
U
L
T
600
500
400
300
200
100
0
0
2
4
6
8
10
Time [sec]
12
14
16
18
8
9
1000
FAULTY
HEALTHY
900
800
Frequency [Hz]
700
600
500
400
300
200
100
0
0
1
2
3
4
5
Time [sec]
6
7
Piezoelectric (top) and FBG (bottom) vibration short-time FFT spectra for
variable speed operation with and without open-circuit fault
F
R
E
Q
U
E
N
C
I
E
S
Single fault frequency zoom-in
FAULT FREQUENCY
3d spectrogram, PE sensor
M
A
G
N
I
T
U
D
E
FAULT FREQUENCY
FREQUENCY
TIME
3d spectrogram, FBG sensor
M
A
G
N
I
T
U
D
E
FREQUENCY
TIME
Summary
• Wideband VibroFibre was shown to provide
comparable performance to high cost PE sensor
under electrical and mechanical fault conditions
• Improvements ongoing in sensor characteristics
and signal processing to further enhance
performance and bring it closer to PE benchmark
• Future work will show distributed vibration sensing
and investigate sensor design with simultaneous
temperature and vibration sensing ability
• VibroFibre can become a competitive alternative to
current high cost sensing solutions
Thank You
Questions?