Pump Vibration Analysis
Pump Vibration Analysis
Pump Vibration Analysis
Related content
- Fault Diagnosis using Audio and Vibration
Diagnosis of Centrifugal Pump Faults Using Signals in a Circulating Pump
P Henrquez, J B Alonso, M A Ferrer et al.
Vibration Methods - Electrical Motor Current Signal Analysis
using a Modulation Signal Bispectrum for
the Fault Diagnosis of a Gearbox
To cite this article: A Albraik et al 2012 J. Phys.: Conf. Ser. 364 012139 Downstream
M Haram, T Wang, F Gu et al.
albraik@live.co.uk
Abstract. Pumps are the largest single consumer of power in industry. This means that faulty
pumps cause a high rate of energy loss with associated performance degradation, high
vibration levels and significant noise radiation. This paper investigates the correlations
between pump performance parameters including head, flow rate and energy consumption and
surface vibration for the purpose of both pump condition monitoring and performance
assessment. Using an in-house pump system, a number of experiments have been carried out
on a centrifugal pump system using five impellers: one in good condition and four others with
different defects, and at different flow rates for the comparison purposes. The results have
shown that each defective impeller performance curve (showing flow, head, efficiency and
NPSH (Net Positive Suction Head) is different from the benchmark curve showing the
performance of the impeller in good condition. The exterior vibration responses were
investigated to extract several key features to represent the healthy pump condition, pump
operating condition and pump energy consumption. In combination, these parameter allow an
optimal decision for pump overhaul to be made [1].
1. Introduction
The current methods for monitoring (simple vibration analysis) are not adequate enough to predict
incipient faults in a pump and avoid frequent breakdowns and outages that are causing the shutdown
of large pumps. The condition of components such as pump shafts and impellers, roller bearings and
drive parts is monitored by evaluation of specific machine vibrations, vibrations due to flow
excitation, and structure borne sound in roller bearings. The examination also involves operating
parameters such as the flow rate, suction pressure, output pressure, drive power, speed, bearing
temperatures and leakage monitoring. Condition information is automatically transmitted to the system
[2].To investigate the detection of incipient faults in the in-house horizontal centrifugal pumps, the
focus was on the use of state-of the art technology/system of vibration analysis and electrical motor
current signals for fault detection and diagnosis [3,4].
the working pump is generated by both mechanical and hydrodynamic sources. The mechanical
sources are invariably generated by rotation of unbalanced masses and friction in the bearings.
Hydrodynamic vibration is due to fluid flow perturbations and interaction of the rotor blades
particularly with the volute tongue and/or guide vanes. The generated vibration will cause the pump
surface to vibrate which will then act as a loudspeaker radiating airborne noise. Thus the basic
mechanisms generating both structure borne vibration and airborne noise are the same. Chudina et.al
[5] have shown that pump vibration contains both broadband noise and discrete frequency peaks.
The broadband content is ascribed to flow turbulence and vortex shedding particularly in the
narrow spaces between the pump rotor and adjacent stationary parts of the casing. There will also be a
contribution from such mechanical sources as the rotation of the pump shaft and bearings. Turbulent
noise will be at a minimum when the pump works with maximum efficiency, i.e. the pump operates at
its design point. Off-design flow rates generate additional hydraulic noise, particularly for very low
flow rates when internal recirculation occurs between the suction and discharge areas of the pump.
When the flow rate is greater than the design flow rate, flow turbulence and boundary layer vortex
shedding both increase [6,7]. The discrete components in the pump spectrum are primarily due to
interaction of the rotor blades with adjacent stationary objects (e.g. volute tongue) in the flow due to
the discrete nature of the pumps rotor blades. These mechanisms generate spectral peaks at the
rotational frequency (RF) and/or blade passage frequency (BPF) of the pump and their higher
harmonics. At or near the design point flow turbulence is at a minimum so the discrete components,
particularly the lower harmonics, tend to dominate the spectrum [8]. Away from the design point the
turbulent noise will increase and will, eventually, even exceed the tonal noise [5,7].
2
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
above NPSHr by making changes in the system design or operation to reduce or eliminate cavitation.
Cavitation does not always produce pump vibration, and the induced vibration often is random and
unmeasurable [15]. One of the other causes of flow-induced vibration at blade passing frequencies in
centrifugal diffuser pumps is the inappropriate radial gap between impeller and volute vanes. A small
gap may be preferable for pump performance, head and efficiency. However, it may initiate strong
impeller/volute interaction, resulting in high pressure pulsation inside the pump and consequent high
vibrations to the pump components [16].
3
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
4
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
For monitoring pump performances, a shaft encoder was used to measure pump speed. A flow
sensor was installed in the discharge line. In addition, two pressure sensors were installed in the
suction and discharge lines respectively for pump delivery head measurement.
The predicted characteristics between the NPSHa and NPSHr for this system were obtained by
throttling the valve in the discharge line progressively while the pump speed is at 2900rpm and the
valve in the suction line was fully open (100%) as suggested in ISO 3555. Figure 3.3 shows the
relation between NPSHa, NPSHr and flow rate. NPSHr (from pump head) increases with flow rate and
increases rapidly above 220 l/m. The measured NPSHa decreases with flow rate and the NPSHa-
NPSHr intersection occurs at a flow rate of approximately 320 l/m and head 5.75(m).
5
Head(m)
NPSHa-From Formula
2
NPSHr-From Pump
NPSHa-Measured
1
50 100 150 200 250 300 350
Flow rate(l/m)
5
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
Figure 4.1. Impeller 1 In good condition. Figure 4.2. Impeller 2 with gap between two plates.
40
35
Head(m)
30
25
20
15
10
50 100 150 200 250 300 350
Flow-rate(l/min)
6
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
50
398
396 45
394
50 100 150 200 250 300 350 40
Flow rate(l/m)
Relation between A & Q
Acceleration(m/s 2)
35
12
Current(A)
10 30
8
50 100 150 200 250 300 350 25
Flow rate(l/m)
Relation between Electric Power & Q
4 20
Power(KW)
3.5 15
3
50 100 150 200 250 300 350 10
Flow rate(l/m) 50 100 150 200 250 300 350
Flow rate(l/m)
(b) Voltage, current, power vs flow rate. (c) Vibration acceleration vs flow rate.
Figure 4.3. Collected data at different flow rates ( impeller 1 vs impeller 2).
Figure 4.3 (a) compares flow rate against head the impeller in good condition (blue line) with
defective impeller 2: a hammer blow on its edge reduced the gap between two plates at that point to
2.4mm (red line). The healthy impeller had a gap of 3.2mm. Impellers 1 and 2 are the same geometry.
The red line shows a higher head compared to the blue line and the deviation is greater at higher flow
rates. In Figure 4.3 (b) the red line shows that the faulty condition consumes more voltage than the
healthy condition. The current consumption is almost the same for both conditions. Power
consumption is higher for the faulty condition particularly for flow rates higher than 250 l/m.
Vibration acceleration is generally greater for the faulty condition for flow rates between 100 l/m and
320 l/m, as can be seen from the blue and red lines in Figure 4.3 (c). Above a flow rate of 320 l/m the
situation is reversed and the measured vibration acceleration is greater for the healthy pump. The
vibration acceleration for the faulty pump is almost linearly proportional to flow between 50 to 220
l/m and then increases rapidly up to 320 l/m flow rate after which it decreases rapidly. The vibration
acceleration for the healthy pump increases almost linearly between flow rates of 50 and 275 l/m then
increases rapidly.
Figure 4.4. Impeller 3 with gap between two plates reduced to 0.7mm at a single point by a hammer
blow.
7
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
40
Head(m) 35
30
25
20
15
10
50 100 150 200 250 300 350
Flow-rate(l/min)
60
500
55
50
0
50 100 150 200 250 300 350
Flow rate(l/m) 45
Relation between A & Q
Acceleration(m/s 2)
20 40
Current(A)
10 35
0 30
50 100 150 200 250 300 350
Flow rate(l/m)
25
Relation between Electric Power & Q
5
Power(KW)
20
15
0
50 100 150 200 250 300 350 10
Flow rate(l/m) 50 100 150 200 250 300 350
Flow rate(l/m)
(b) Voltage, current and power vs flow rate. (c) Vibration acceleration vs flow rate.
Figure 4.5. Collected data at different flow rates (impeller 1 vs impeller 3).
Figure 4.5 (a) compares flow rate against head the impeller in good condition (blue line) with
defective impeller 3: a hammer blow on its edge reduced the gap between two plates at that point to
0.7mm (red line). Impellers 1 and 3 are the same geometry. Figure 4.5 (a) compares the healthy
impeller (blue line) with the defective impeller (dented and with reduced gap between two plates (red
line). Once again the faulty impeller had a higher head than the healthy impeller at all flow rates, but
particularly above about 300 l/m. Figure 4.5 (b) shows the voltage remains almost the same across
flow rates from 50 to 325 l/m, but current and power increase slightly with flow rate. Figure 4.5 (c)
shows the vibration acceleration is almost the same in both healthy and faulty impeller for flow rates
up to about 220 l/m; for flow rates 220 to 275 l/m the vibration acceleration is greater for the healthy
condition, but above 275 l/m the defective impeller produced a greater level of vibration. The
vibration acceleration for the defective impeller is almost linearly proportional to flow rate between 50
to 220 l/m, then increases rapidly up to 320 l/m then decreases.
8
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
Figure 4.6. Impeller 4 with gap between two plates reduced to 0.3mm at a single point by a hammer
blow.
Head vs. Flow Rate
50
Impeller1
Impeller4
45
40
35
Head(m)
30
25
20
15
10
50 100 150 200 250 300 350
Flow-rate(l/min)
60
400
55
395
50
390
50 100 150 200 250 300 350
Flow rate(l/m) 45
Relation between A & Q
Acceleration(m/s 2)
10 40
Current(A)
9 35
8 30
50 100 150 200 250 300 350
Flow rate(l/m)
25
Relation between Electric Power & Q
4
Power(KW)
20
3.5
15
3
50 100 150 200 250 300 350 10
Flow rate(l/m) 50 100 150 200 250 300 350
Flow rate(l/m)
(b) Voltage, current and power vs flow rate. (c) Vibration acceleration vs flow rate.
Figure 4.7. Collected data at different flow rates (impeller 1 vs impeller 4).
9
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
Figure 4.7 (a) compares flow rate against head for the healthy impeller (blue line) with defective
impeller 4: a hammer blow on its edge reduced the gap between two plates at that point to 0.3mm (red
line). Impellers 1 and 4 are the same geometry. As previously, the faulty impeller provides a higher
head compare to the healthy impeller on flow rate above about 150 l/m, the difference is very marked
at flow rates above 300 l/m. Red consumes more AC volt compared to blue in Figure 4.7 (b), where
current consumption is slightly higher in Red compared to Blue. Power consumption is high in Red
compared to blue. In figure 4.7 (c), acceleration is found to be more in blue compared to red for the
flow rate 100-275 (m/l). Above 275 (m/l) flow rate, acceleration is more in red compared to Blue.
Acceleration in blue is almost linearly proportional between 50 to 220 (m/l) flow rate then, increases
rapidly up to 320 (m/l) flow rate and after that decreases rapidly. Acceleration in red is almost linearly
proportional between 50 to 275 (m/l) flow rate then, increases rapidly up to 315 (m/l) flow rate and
then increases rapidly.
Figure 4.8. Impeller 5 with gap between two plates reduced by a hammer blow at three equally spaced
points.
Head vs. Flow Rate
50
Impeller1
Impeller5
45
40
35
Head(m)
30
25
20
15
10
50 100 150 200 250 300 350
Flow-rate(l/min)
10
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
50
410
400 45
390
50 100 150 200 250 300 350 40
Flow rate(l/m)
Relation between A & Q
Acceleration(m/s 2)
35
12
Current(A)
10 30
8
50 100 150 200 250 300 350 25
Flow rate(l/m)
Relation between Electric Power & Q
5 20
Power(KW)
4 15
3
50 100 150 200 250 300 350 10
Flow rate(l/m) 50 100 150 200 250 300 350
Flow rate(l/m)
(b) Voltage, current and power vs flow rate. (c) Vibration acceleration vs flow rate.
Figure 4.9. Collected data at different flow rates (impeller 1 vs impeller 5).
Figure 4.9 (a) compares flow rate against head for the healthy impeller (blue line) with defective
impeller 5: three equally spaced hammer blows around the circumference reduced the spacing to
2.6mm, 2.9mm and 3.0mm, respectively, as shown. Impellers 1 and 5 are the same geometry. Unlike
the previous results, the healthy impeller produces a greater head up to about 300 l/m, for flows above
this the defective impeller produced a higher head. Figure 4.9 (b) shows that the pump containing the
faulty impeller consumes more power, requiring a higher AC voltage and current than the healthy
pump. Figure 4.9 (c) shows that vibration acceleration is much greater in the pump with defective
impeller for flow rates up to about 280 l/m between about 300 l/m and 325 l/m the vibration
acceleration is greater for the healthy pump, but above 325 l/m the acceleration is much the same for
both pumps.
5. Conclusions
Experiments were carried out on a high-pressure centrifugal pump (single stage) to study the effect on
pressure fluctuations due to different impellers faults at different flow rates. The effect on pump
performance is also examined. The impeller/volute interaction is an important design parameter in
developing high-energy pumps. The results for all tests show that the vibration level increases with
increased of flow rate and with different readings, this is due to a different type of defect on each
impeller. The experimental study has shown that data obtained from impellers with different gaps are
different, even though the impellers are geometrically similar and for the same pump. This is because
impellers 2, 3, 4 and 5 have different faults both the depth of the dents and the number of dents.
Further research work will be conducted to extract more data/features from the pump using different
techniques. The pump faults may then be identified using these data/features.
References
[1] Centrifugal Pumps, accessed, 29/09/2011, Impeller Reverse Design, webhome.idirect.com
[2] Production maintenance technology, PRUFTECHNIK, Condition Monitoring Gmbh,
www.pruftechnik.com.
[3] A brief Introduction to Centrifugal pumps, accessed, 27/09/2011, Joe Evans, Pacific Liquid and
Air System.www.Pacificliquid.com
[4] Relating Centrifugal Pump Efficiency, accessed, 30/09/2011,Variable Speed Drives and
Hydraulic Improvement to Energy Dollars. Steven Bolles, Process Energy Services, LCC
and Don Casada Diagnostic, LCC. processenergy.com
[5] Chudina M, and Prezelj J, (2009), Detection of cavitation in operation of kinetic pumps, Use of
discrete frequency tone in audible pump spectra using audible sound. Journal Applied
Acoustic, Vol. 70, Issue 4, pp 540-546, April
11
25th International Congress on Condition Monitoring and Diagnostic Engineering IOP Publishing
Journal of Physics: Conference Series 364 (2012) 012139 doi:10.1088/1742-6596/364/1/012139
[6] Chudina, M. , 2003, Noise as an indicator of cavitation in a centrifugal pump. Acoustical Physics,
Vol 49 Issue 4, pp. 1335-1347 [7] Cernetic, J, The use of noise and vibration signals for
detecting cavitation in kinetic pumps. Proc. IMechE Vol. 223 Part C: Jnl. Mech. Eng.
Science pp1645-1655, 2009
[8] Cudina, M. , 1998, Noise as an indicator of cavitation instability in .centrifugal pumps, Journal of
Mechanical Engineering, Vol. 45 [9] Lecture 12, Vibration Monitoring and Machine
Failure,www.scribd.com, accessed, 03/11/2011.
[10] Centrifugal Pumps. 1992, Design and Application, second Edition, by Gulg Publishing
Company, Houston, Texas
[11] Dennis H. Shreve, 1994, Introduction to vibration technology, IRD Mechanalysis, Inc.,
Columbus, Ohio 43229
[12] Jensen, J. and K. Dayton, , 2000, Detection cavitation in centrifugal pumps, in ORBIT
[13] koivula, T. 2000., On cavitation in fluid power. in 1st FPNI-PhD. Hamburg
[14] Escaler, X., et al., , 2006, Detection of cavitation in hydraulic turbines. Mechanical Systems and
Signal Processing
[15] Examining Causes of Pump Vibration, By STANKUTINITT Residential & Commercial Water
Morton Grove, Ill
[16] Attia E . Khalifa *, Amro M. Al-Qutub. The Effect Of Impeller-Volute Gap On Pressure
Fluctuations Inside A Double-Volute Centrifugal Pump Operating At Reduced Flow Rates, ,
King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabia
[17] Edited by Dale B. Andrews, Jan 2008, Newsletter on pumps and reliability, Diagnose Low
Discharge Pressure.
[18] L. Mongeau, D. E. Thompson, D. K. Mclaughlin , 1993, Sound Generation by Rotating Stall in
Centrifugal Turbo machines. Journal of Sound and Vibration.
[19] Variable Speed Pumping A Guide to Successful Applications, U.S. Department of Energy,
Energy Efficiency and Renewable Energy uropump Industrial Technologies Program.
12