- Email: [email protected]
Vibration and wear detection in rotating machinery by acoustic analysis
Applied Acoustics 28 (1989) 213-219
Technical Note Vibration and Wear Detection in Rotating Machinery by Acoustic Analysis
ABSTRACT This technical note illustrates the value of acoustic monitoring. Three investigations in differing areas of engineering are described. The first concerns excessive noise in the drive of a low speed wind tunnel During runs at differing fan speeds a specific bearing was identified as the noise source. On replacement the overall noise level fell several decibels. The second case concerns a universal milling machine which began to emit excessive noise. Eighteen different spindle speeds, involving nine possible gear pairs, can be selected. Knowledge of gear meshing frequencies enabled two specific gear pairs to be identifiedfrom analysis of the noise at various spindle speeds. The third study concerns an oilfired burner, intended for domestic heating, the noise levels of which were of concern. The sliding vane compressor of this unit was identified as the prime noise source. Frequency analysis indicated that vane vibration was the source of the problem. Redesign o f the vane and slot was recommended.
INTRODUCTION In recent years the value of condition or health monitoring of machinery has been increasingly recognised. 1'2 The quantity monitored may be contaminants (wear particles), vibration or the acoustic signature. Incipient failure may be predicted and, in the latter two methods, frequency analysis used to indicate the likely site of wear. 213 Applied Acoustics 0003-682X/89/$03"50 O 1989 Elsevier Science Publishers Ltd, England. Printed in Great Britain
C. W. Stammers
214
This note describes three undergraduate projects concerned with engineering problems, in which acoustic monitoring was a valuable diagnostic tool.
W I N D T U N N E L F A N DRIVE 3 The fan for the tunnel is driven by a constant-speed induction motor via an electromagnetic clutch (Fig. 1). Fan speed is controlled by means of a brake applied to the fan shaft. Some time after installation, drive noise began to increase, and a noise study was undertaken. The motor was disconnected from the drive line and its own noise peaks recorded. Peaks were found at 25, 200 and 625 Hz. Noise measurements were then made with the fan connected and running at a speed of 600 rev/min (Fig. 2). The noise peaks recorded were found to dominate those due to the motor. One peak was at 600 Hz. The motor speed 1'~'~;~,
L.H. BFAR I NG
@
I
,
f
I
I
I
I
[:E~ /
r---n ~
N~
~EARING
2-?
Fig. 1.
Bearing locations.
II
L
-ITi !
r--
I 1
~Od~
8hOHz 91rlBI ._
89dl
~, d II
,
" I
~r
rOSOl- z ~ 1 4 ~ ~.fB
//F'~,, '~i~ ~d~q~,~- •
di M~k
~F:'--
' i i
q In il~ X!
,j 1
"1
i i i
300
500
1000
Frequency
Fig. 2.
H
ILl I L92clB ' "~ I:!~OH~.
---
, r
R.H. BEARI N6
F-'~ ,.
f 96dB
CLUTCH R.H. BEARING
(Hz)
N o i s e spectrum.
ii
Z
Vibration and wear detection
215
was 1488rev/min while the clutch outer casing, which rotates with the motor, has 24 slots. Hence it appeared that the 600 Hz peak was a siren effect. However, when the slots were taped over, the noise peak fell by only 1 dB, and the peak had to be attributed to a bearing. Sound levels were recorded at each bearing at a number of fan speeds. The left-hand bearing (Fig. 1) turns at motor speed, the right-hand bearing at fan speed, and the clutch inner bearing at the difference of the two. In this way, the defective bearing was identified as the left-hand bearing of the drive. The bearing was dismantled and found to be worn. On replacement the overall sound pressure level was reduced by several decibels.
UNIVERSAL MILLING MACHINE 4 A milling machine of Russian manufacture began to emit noise of a sufficient level to prevent conversation. While there was a service panel on the miller, it was not possible to inspect the gears without dismantling the gearbox. The gearbox layout is shown in Fig. 3. Eighteen different spindle speeds, involving nine possible gear pairs, can be selected (Fig. 4). Gears 2 and 3 are
Fig. 3.
Main gearbox layout.
216
C. W. Stammers
--g
0
.=_ e,q
E
o
E
/ e~
217
Vibration and wear detection
always employed. There is a choice of three pairs of gears at the second and third stages, while the final drive is via gears 14 and 15 or gears 13 and 16. Final spindle speed varies from 40-5 rev/min to 1590rev/min. The meshing frequencies for each gear pair were calculated (Table 1). Frequency analysis of the noise at each spindle speed enabled gear pair 2 and TABLE 1 Gear and Shaft Speed Data
Shaft
Shaft speed (rev/min)
Meshing gears
(Hz)
Tooth meshing frequency
Previousshaft frequency (Hz)
(nz) A B
1 440 734
C
309 398 504
D
24.0 12.2
2/3 2/3
648 648
-24-0
5"15 6'64 8'41
6/17 4/7 5/18
196 232 269
12'2 12"2 12.2
114 225 451
1"9 3-76 7-52
11/9 10/8 12/17
88 139 196
5"15 5"15 5"15
D
147 291 582
2-45 4"84 9"70
11/9 10/8 12/17
113 179 252
6-64 6"64 6"64
D
186 386 737
3.11 6-13 12"3
11/9 10/8 12/17
143 227 319
8'41 8'41 8"41
E
31-5 40.5 51"3 62.0
0"53 0-67 0-86 1-03
15/14 15/14 15/14 15/14
36"1 46.7 59-0 71.4
1-90 2'45 3' 11 3'76
E
80.0 101 124 160
1-33 1-69 2"07 2"67
15/14 15/14 15/14 15/14
92'6 116 143 184
4'84 6"13 7"50 9'70
E
203 246 318 402
3-38 4-10 5-29 6-70
15/14 13/16 13/16 13/16
233 156 201 255
12.30 1'90 2'45 3"11
E
486 627 794 974
8"11 10-5 13-2 16-2
13/16 ! 3/16 13/16 13/16
308 397 503 617
3'76 4.84 6"13 7-50
E
1 256 1 590
20-9 26-5
13/16 13/16
795 1 007
9"70 12.30
218
C. W. Stammers
3 to be identified as the chief noise source. This was not surprising, as this pair is engaged at all spindle speeds. Gear pair 12 and 17 was also identified as a significant noise source.
OIL-FIRED BURNER 5 This particular type of burner has been successfully operated in an industrial role. The manufacturer wished to introduce a version for domestic heating, but noise levels were evidently too high for a home environment. An electric motor drives a fan and sliding vane compressor which supplies air to atomise the fuel. The motor speed is variable, and is controlled automatically to modulate the burner output to match demand. In the tests the controller was bypassed and the speed set manually. Noise measurements were made at the maximum motor speed. The burner was mounted on a turntable and the local noise level was recorded by a microphone sited 225 mm from the centre of the burner. This test identified the compressor as the chief noise source. Measurements were also made across the speed range of the motor using three microphones. The overall noise level increased with motor speed in a fairly regular manner. Frequency analysis indicated a dominant noise peak at 12 k H z and peaks at the 12th and 16th harmonics of the motor speed (Fig. 5). The latter two peaks could have been caused by wear in the compressor bearings. However, since the compressor has four sliding vanes which have to cross the inlet and outlet part on each revolution, vane vibration was suspected. This was confirmed by running with vanes removed (the noise level fell appreciably).
70-
111
~a
f10-
2ua
t.~
50-
~-
011
J
J
1 FR£OU£NEY / kHz
Fig. 5. Compressor noise spectrum.
Vibration and wear detection
219
A stroboscopic study of the compressor with the end plate removed indicated that vane lateral vibration was occurring, principally on the lowpressure side. The dominant peak at 12 kHz (which did not alter with motor speed) was also identified with vane vibration, and is possibly the natural frequency of vibration of the vane in its slot.
REFERENCES 1. Bruel & Kjaer, Machine-condition monitoring using vibration analysis, a case study from an iron ore mine. Application note BO 0178-11. 2. Astridge, D. G., Helicopter HUM cuts costs and down time. CME, May 1986, pp. 39-42. 3. Wilson, A. J. & Bramwell, A., Noise analysis of an electro-magnetic clutch. School of Engineering Report, University of Bath, June 1984. 4. Ashworth, R. S. & Bishop, N. N., An investigation into noise sources ina milling machine gearbox. School of Engineering Report, University of Bath, June 1985. 5. Painter, D. I. & Zakary, P. C., Noise reduction of a domestic oil burner. School of Engineering Report, University of Bath, June 1987. C. W. Stammers School of Mechanical Engineering, University of Bath, Bath BA2 7A Y, UK (Received 16 October 1988; revised version received 9 January 1989; accepted 3 May 1989)
Technical Note Vibration and Wear Detection in Rotating Machinery by Acoustic Analysis
ABSTRACT This technical note illustrates the value of acoustic monitoring. Three investigations in differing areas of engineering are described. The first concerns excessive noise in the drive of a low speed wind tunnel During runs at differing fan speeds a specific bearing was identified as the noise source. On replacement the overall noise level fell several decibels. The second case concerns a universal milling machine which began to emit excessive noise. Eighteen different spindle speeds, involving nine possible gear pairs, can be selected. Knowledge of gear meshing frequencies enabled two specific gear pairs to be identifiedfrom analysis of the noise at various spindle speeds. The third study concerns an oilfired burner, intended for domestic heating, the noise levels of which were of concern. The sliding vane compressor of this unit was identified as the prime noise source. Frequency analysis indicated that vane vibration was the source of the problem. Redesign o f the vane and slot was recommended.
INTRODUCTION In recent years the value of condition or health monitoring of machinery has been increasingly recognised. 1'2 The quantity monitored may be contaminants (wear particles), vibration or the acoustic signature. Incipient failure may be predicted and, in the latter two methods, frequency analysis used to indicate the likely site of wear. 213 Applied Acoustics 0003-682X/89/$03"50 O 1989 Elsevier Science Publishers Ltd, England. Printed in Great Britain
C. W. Stammers
214
This note describes three undergraduate projects concerned with engineering problems, in which acoustic monitoring was a valuable diagnostic tool.
W I N D T U N N E L F A N DRIVE 3 The fan for the tunnel is driven by a constant-speed induction motor via an electromagnetic clutch (Fig. 1). Fan speed is controlled by means of a brake applied to the fan shaft. Some time after installation, drive noise began to increase, and a noise study was undertaken. The motor was disconnected from the drive line and its own noise peaks recorded. Peaks were found at 25, 200 and 625 Hz. Noise measurements were then made with the fan connected and running at a speed of 600 rev/min (Fig. 2). The noise peaks recorded were found to dominate those due to the motor. One peak was at 600 Hz. The motor speed 1'~'~;~,
L.H. BFAR I NG
@
I
,
f
I
I
I
I
[:E~ /
r---n ~
N~
~EARING
2-?
Fig. 1.
Bearing locations.
II
L
-ITi !
r--
I 1
~Od~
8hOHz 91rlBI ._
89dl
~, d II
,
" I
~r
rOSOl- z ~ 1 4 ~ ~.fB
//F'~,, '~i~ ~d~q~,~- •
di M~k
~F:'--
' i i
q In il~ X!
,j 1
"1
i i i
300
500
1000
Frequency
Fig. 2.
H
ILl I L92clB ' "~ I:!~OH~.
---
, r
R.H. BEARI N6
F-'~ ,.
f 96dB
CLUTCH R.H. BEARING
(Hz)
N o i s e spectrum.
ii
Z
Vibration and wear detection
215
was 1488rev/min while the clutch outer casing, which rotates with the motor, has 24 slots. Hence it appeared that the 600 Hz peak was a siren effect. However, when the slots were taped over, the noise peak fell by only 1 dB, and the peak had to be attributed to a bearing. Sound levels were recorded at each bearing at a number of fan speeds. The left-hand bearing (Fig. 1) turns at motor speed, the right-hand bearing at fan speed, and the clutch inner bearing at the difference of the two. In this way, the defective bearing was identified as the left-hand bearing of the drive. The bearing was dismantled and found to be worn. On replacement the overall sound pressure level was reduced by several decibels.
UNIVERSAL MILLING MACHINE 4 A milling machine of Russian manufacture began to emit noise of a sufficient level to prevent conversation. While there was a service panel on the miller, it was not possible to inspect the gears without dismantling the gearbox. The gearbox layout is shown in Fig. 3. Eighteen different spindle speeds, involving nine possible gear pairs, can be selected (Fig. 4). Gears 2 and 3 are
Fig. 3.
Main gearbox layout.
216
C. W. Stammers
--g
0
.=_ e,q
E
o
E
/ e~
217
Vibration and wear detection
always employed. There is a choice of three pairs of gears at the second and third stages, while the final drive is via gears 14 and 15 or gears 13 and 16. Final spindle speed varies from 40-5 rev/min to 1590rev/min. The meshing frequencies for each gear pair were calculated (Table 1). Frequency analysis of the noise at each spindle speed enabled gear pair 2 and TABLE 1 Gear and Shaft Speed Data
Shaft
Shaft speed (rev/min)
Meshing gears
(Hz)
Tooth meshing frequency
Previousshaft frequency (Hz)
(nz) A B
1 440 734
C
309 398 504
D
24.0 12.2
2/3 2/3
648 648
-24-0
5"15 6'64 8'41
6/17 4/7 5/18
196 232 269
12'2 12"2 12.2
114 225 451
1"9 3-76 7-52
11/9 10/8 12/17
88 139 196
5"15 5"15 5"15
D
147 291 582
2-45 4"84 9"70
11/9 10/8 12/17
113 179 252
6-64 6"64 6"64
D
186 386 737
3.11 6-13 12"3
11/9 10/8 12/17
143 227 319
8'41 8'41 8"41
E
31-5 40.5 51"3 62.0
0"53 0-67 0-86 1-03
15/14 15/14 15/14 15/14
36"1 46.7 59-0 71.4
1-90 2'45 3' 11 3'76
E
80.0 101 124 160
1-33 1-69 2"07 2"67
15/14 15/14 15/14 15/14
92'6 116 143 184
4'84 6"13 7"50 9'70
E
203 246 318 402
3-38 4-10 5-29 6-70
15/14 13/16 13/16 13/16
233 156 201 255
12.30 1'90 2'45 3"11
E
486 627 794 974
8"11 10-5 13-2 16-2
13/16 ! 3/16 13/16 13/16
308 397 503 617
3'76 4.84 6"13 7-50
E
1 256 1 590
20-9 26-5
13/16 13/16
795 1 007
9"70 12.30
218
C. W. Stammers
3 to be identified as the chief noise source. This was not surprising, as this pair is engaged at all spindle speeds. Gear pair 12 and 17 was also identified as a significant noise source.
OIL-FIRED BURNER 5 This particular type of burner has been successfully operated in an industrial role. The manufacturer wished to introduce a version for domestic heating, but noise levels were evidently too high for a home environment. An electric motor drives a fan and sliding vane compressor which supplies air to atomise the fuel. The motor speed is variable, and is controlled automatically to modulate the burner output to match demand. In the tests the controller was bypassed and the speed set manually. Noise measurements were made at the maximum motor speed. The burner was mounted on a turntable and the local noise level was recorded by a microphone sited 225 mm from the centre of the burner. This test identified the compressor as the chief noise source. Measurements were also made across the speed range of the motor using three microphones. The overall noise level increased with motor speed in a fairly regular manner. Frequency analysis indicated a dominant noise peak at 12 k H z and peaks at the 12th and 16th harmonics of the motor speed (Fig. 5). The latter two peaks could have been caused by wear in the compressor bearings. However, since the compressor has four sliding vanes which have to cross the inlet and outlet part on each revolution, vane vibration was suspected. This was confirmed by running with vanes removed (the noise level fell appreciably).
70-
111
~a
f10-
2ua
t.~
50-
~-
011
J
J
1 FR£OU£NEY / kHz
Fig. 5. Compressor noise spectrum.
Vibration and wear detection
219
A stroboscopic study of the compressor with the end plate removed indicated that vane lateral vibration was occurring, principally on the lowpressure side. The dominant peak at 12 kHz (which did not alter with motor speed) was also identified with vane vibration, and is possibly the natural frequency of vibration of the vane in its slot.
REFERENCES 1. Bruel & Kjaer, Machine-condition monitoring using vibration analysis, a case study from an iron ore mine. Application note BO 0178-11. 2. Astridge, D. G., Helicopter HUM cuts costs and down time. CME, May 1986, pp. 39-42. 3. Wilson, A. J. & Bramwell, A., Noise analysis of an electro-magnetic clutch. School of Engineering Report, University of Bath, June 1984. 4. Ashworth, R. S. & Bishop, N. N., An investigation into noise sources ina milling machine gearbox. School of Engineering Report, University of Bath, June 1985. 5. Painter, D. I. & Zakary, P. C., Noise reduction of a domestic oil burner. School of Engineering Report, University of Bath, June 1987. C. W. Stammers School of Mechanical Engineering, University of Bath, Bath BA2 7A Y, UK (Received 16 October 1988; revised version received 9 January 1989; accepted 3 May 1989)