See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/331855081

Condition Monitoring of Centrifugal Pumps Based on Pressure Measurements

Article  in  Periodica Polytechnica, Mechanical Engineering · March 2019

DOI: 10.3311/PPme.12140

CITATIONS READS

0 1,790

2 authors:

Csanád Kalmár Ferenc Hegedűs

Budapest University of Technology and Economics Budapest University of Technology and Economics
3 PUBLICATIONS   22 CITATIONS    35 PUBLICATIONS   319 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

GPU accelerated bubble dynamics and sonochemistry View project

Integration of independent ODE system on GPUs View project

All content following this page was uploaded by Ferenc Hegedűs on 08 June 2020.

The user has requested enhancement of the downloaded file.

 Periodica Polytechnica
Mechanical Engineering
Condition Monitoring of Centrifugal
Pumps Based on Pressure Measure-
x(y) pp. x-x, (year)
https://doi.org/10.3311/PPme.XXX
ments
Creative Commons Attribution
b
Csanád Kalmár1, Ferenc Hegedűs2

RESEARCH ARTICLE

Abstract 1 Introduction
The purpose of the present study is the investigation of condition
Centrifugal pumps are one of the mostly used turbomachines
of centrifugal pumps via pressure signals. Instead of vibration
in the industry and in households as well [1]. Because of the
measurement on the housings that is widely used in industry, our
massive application and the increasing aim for reducing energy
method is based on pressure signal measurement on the pressure
consumption, it is necessary to develop continuous condition
side of the pump. Fourier transforming such a signal can get us
monitoring techniques on pumps. By detecting several produc-
to make conclusions about the behavior of the pump. By chang-
tion or mounting failures, we have the chance to intervene in
ing the operating point along a characteristic curve, we can cre-
time; thus, granting good operation efficiency or preventing se-
ate waterfall diagrams that provides useful information about
rious accidents.
the pump at constant rotational speed. For example, it is possible
In the industry, the widely used method for vibration moni-
to differentiate the mechanical and the hydrodynamical effects
toring is the measurement of acceleration on the bearings [2, 3].
predicting the occurrence of many constructional failures (such
This technique is mostly applied because of its simple and
as unbalance, angular misalignment, bearing misalignment, mo-
straightforward utilization, but has a quite big disadvantage that
tor instability, etc.); thus, preventing heavy damage of the equip-
some surface treatment is necessary to mount the sensors. Such
ment.
treatment causes the loss of the guarantee, which is unaccepta-
Keywords ble in most cases. This problem can be avoided by applying a
condition monitoring, centrifugal pump, spectral analysis, diag- method based on pressure signal measurement on the pressure
nostics, waterfall diagram side of a pump, which is easy to implement in most cases.
Pressure signal measurements are mainly used in order to de-
tect cavitation in pipeline systems [4]. Ĉudina [5] made meas-
urements based on acoustic signals, and found correlation be-
tween acoustic emission and cavitation. However, according to
the best knowledge of the authors, condition monitoring tech-
niques based on pressure measurements has not yet been pub-
lished in the literature. Therefore, the main aim of the present
study is to develop and test such a method by comparing the
results obtained on two different centrifugal pumps having very
different health conditions.

2 The test rigs

1
Department of Hydrodynamic Systems,
Faculty of Mechanical Engineering,
For simplicity, the two pumps are indicated with signs ’A’
Budapest University of Technology and Economics and ’B’. Pump ’A’ (BMS 24/48) is a much used, quite old
H-1111 Budapest, Műegyetem rkp. 3., Hungary
(e­mail: [email protected]) machince. Its driving motor and the pump itself are assembled
2
Department of Hydrodynamic Systems, manually; consequently, parallel or angular misalignments and
Faculty of Mechanical Engineering, other mechanical failures can easily occur. In contrast, pump ’B’
Budapest University of Technology and Economics,
H-1111 Budapest, Műegyetem rkp. 3., Hungary (Grundfos TPE 65-340/2) is a much newer, more compact
(e­mail: [email protected]) equipment with much higher precision requirements. The

Condition Monitoring of Centrifugal Pumps year Vol No 1

 hand side is zero. 𝑝𝑝2 is calculated as the average of the measured
signal of the transducer, while 𝑝𝑝1 is measured by a U-tube mer-
cury-filled manometers for both pumps.
The efficiency of the pumps is calculated as
𝑃𝑃 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻
𝜂𝜂 = 𝑃𝑃 𝑢𝑢 = 𝑃𝑃𝑖𝑖𝑖𝑖
, (3)
𝑖𝑖𝑖𝑖

where 𝑃𝑃𝑖𝑖𝑖𝑖 is the input power and 𝑃𝑃𝑢𝑢 is the useful power. It is
Fig. 1 The sketch of the two measurement test rigs, with pump ‘A’ on the left important to note that the input powers are measured differently
and pump ‘B’ on the right. The pumps (P) circulate water from the tanks (R) on the two equipment. In the case of pump ‘A’, the mechanical
through a throttle valve (V). The pressure transducers (Tr) are placed at the input power can be measured directly by balancing the motor.
pressure side of the pumps. This means that the driving motor is fixed on its bearings, thus
the housing can rotate. The torque acting on the shaft is propor-
number of blades is 7 and 6 for pump ’A’ and ’B’, respectively. tional to the torque acting on the house, which can be measured
Figure 1 shows the sketch of the two test rigs. The pumps (P) by balancing it to a steady state. In contrast, in the case of pump
circulate water from the reservoirs (R). The rotational speed of ‘B’, only the measurement of the electric power of the motor
pump ‘B’ can be set with a frequency converter. However, in was possible due to its compactness. This means that in the case
the case of pump ‘A’, the rotational speed of the driving motor of pump ‘A’, efficiency refers to only the pump itself, but in the
is regulated via a thyristor. Consequently, the control system can case of pump ‘B’ it belongs to the motor-pump unit as a whole.
become unstable at certain regions of rotational speeds. This can This results in a slightly different definition for the efficiencies
be identified in the spectra of the measured pressure signal as a of the two pumps, but it has no effect to the drawn conclusions
frequency modulation discussed in more details later. The flow in the next sections.
rate can be adjusted with a throttle valve (V) and measured by a The uncertainties of the estimated quantities were calculated
standard orifice plate (O). The same pressure transducer (Tr) is based on the laws of error propagation. The highest absolute er-
mounted on the pressure side of each pump. The calibration di- rors of the flow rate, head and efficiency were ∆𝑄𝑄 = ±0.51 L/s,
agram of the pressure transducer (HBM P6A) is shown in the ∆𝐻𝐻 = ±0.9 m, ∆𝜂𝜂 = 0.04, respectively. During the evaluation
Appendix (see Figure A1), together with its confidence bands of the results, we will focus on the amplitude spectra of the
corresponding to 95% confidence level. The signal of the trans- measured pressure signals; therefore, detailed uncertainty anal-
ducer is sampled with a data acquisition system (HBM Spider ysis is omitted and error bars on the diagrams are not indicated.
8), the sampling rate of the transducer was 9600 Hz. The strategy to present the results is to calculate the spectrum
The quantities for the characteristic curves are calculated as of the pressure signal on the pressure side along the characteris-
follows. It is known that the flow rate of the system is propor- tic curves at several operation points applying different rota-
tional to the square root of the pressure drop on a standard meas- tional speeds. In this way, one series of spectra can be associated
uring orifice (MSZ EN ISO 5167-1:2003). This orifice is an to each characteristic curves. We made measurements on 17 dif-
equipment with standard diameter, and prescribed straight seg- ferent rotational speeds from 630 to 3000 1/min on pump ‘A’,
ments behind and ahead, too. The flow rate (Q) through the ori-
fice is calculated as
𝑑𝑑 2 𝜋𝜋 2∆𝑝𝑝𝑜𝑜
𝑄𝑄 = 𝛼𝛼 � , (1)
4 𝜌𝜌

where 𝛼𝛼 is the discharge coefficient calculated by an iterative

method according to the standard. 𝑑𝑑 = 0.03 m is the diameter
of the orifice for both pumps, 𝜌𝜌 = 998 kg/m3 is the density of
water at ambient temperature of 𝑇𝑇0 = 22 °C. The pressure drop
∆𝑝𝑝𝑜𝑜 on the orifice is measured by a single-tube mercury filled
manometer.
The head of the pumps (H) are calculated as
𝑝𝑝2 −𝑝𝑝1 𝑐𝑐2 2 −𝑐𝑐1 2
𝐻𝐻 = + + ℎ2 − ℎ1 , (2)
𝜌𝜌𝜌𝜌 2𝑔𝑔

where 𝑝𝑝 is the pressure, 𝑐𝑐 is the velocity, and ℎ is the geodetic

Fig. 2 Three characteristic curves of the pumps with pump ‘A’ on top, and
height. Here, indices 1 and 2 refer to the suction and the pressure
pump ‘B’ on bottom. The best efficiency points are indicated by the blue dots
sides, respectively. The diameters of the suction and pressure
on each rotational speed. These points determine the nominal flow rates. Pres-
sides are equal for both pumps, thus the second term of the right
sure signals were taken at every operation point marked by the crosses.

2 Period. Polytech. Mech. Eng. Csanád Kalmár, Ferenc Hegedűs

Fig. 3 Time signal and amplitude spectrum in an operating point of pump ‘A’ (𝑛𝑛 = 2100 1/min, 𝑄𝑄 = 2.49 L/s). The sharp peaks and wideband regions contain
useful information about the condition of the pump.

and 8 different rotational speeds from 1300 to 3000 1/min on Figure 4 shows a typical waterfall diagram of pump ‘A’ at
pump ‘B’. The flow rates were in the range of 0 to 𝑄𝑄max , where 𝑛𝑛 = 3000 1/min, in which the evolution of the peak amplitudes
𝑄𝑄max is the flow rate at a totally opened valve with value de- as a function of the flow rate can be clearly seen. The series of
pending on the current rotational speed. Three measured char- such diagrams with varying rotational speed gives a good over-
acteristic curves of the pumps can be seen in Figure 2. The best all insight into the behavior of the equipment in a wide range of
efficiency point at a certain rotational speed is indicated by the the operation parameters. The waterfall diagrams that are corre-
blue dot. These points define the nominal flow rate (𝑄𝑄N ) at each sponding to all the measured rotational speeds are shown in the
rotational speed. Appendix.
Before a detailed analysis, it is important to summarize that
3 Analysis process what kind of basic physical effects can be associated to certain
The basic elements of the analysis are the spectra obtained by features of the spectra and the waterfall diagrams. These are di-
Fourier transforming the measured pressure signals. An exam- vided mainly into two groups: mechanical and hydrodynamical
ple is shown in Figure 3 for pump ‘A’ at rotational speed 𝑛𝑛 = [7], see also Table 1. Mechanical effects can be productional or
2100 1/min and flow rate 𝑄𝑄 = 2.49 L/s. The sharp peaks and constructional failures like unbalance, bent shaft, pump and mo-
their harmonics can provide valuable information for condition tor misalignment or bearing failures. These types of effects have
monitoring of the machines discussed in more details during the a common nature that they cause sharp peaks in the spectra at 𝑓𝑓𝑟𝑟
next sections. and its integer multiples (up to 2x – 6x). If the presence of any
The measured spectra at the operation points along a charac-
teristic curve can be organized into a three dimensional waterfall
diagram, where the 2D spectra is plotted as function of the flow
rate. In the following, it is helpful to make use of dimensionless
forms of the frequency and the flow rate [6] indicated by an as-
terisk. Therefore, we let
𝑄𝑄
𝑄𝑄∗ = , (4)
𝑄𝑄𝑁𝑁

and
𝑓𝑓
𝑓𝑓 ∗ = , (5)
𝑓𝑓𝑟𝑟

where 𝑓𝑓𝑟𝑟 is the current rotational speed in Hz unit. With this

Fig. 4 Waterfall diagram of pump ‘A’ at 𝑛𝑛 = 3000 1/min. A waterfall dia-
nondimensionalization, 𝑄𝑄 ∗ = 1 belongs to the best efficiency
gram contains overall information about the behaviour of the pump on a spe-
point. Similarly, 𝑓𝑓 ∗ = 1 refers to the rotational speed.
cific rotational speed.

Condition Monitoring of Centrifugal Pumps year Vol No 3

mechanical effects is significant, these peaks dominate the spec- fluid as pressure waves, consequently the pressure values (and
tra. Hydrodynamical effects are mainly the impact of impeller- all other fluid quantities, too) fluctuate with the same frequen-
volute interaction, cavitation or non-uniform flow distribution cies as the mechanical vibration itself. This wave propagation
around the blades. They are presented even in a well-designed mechanism is mostly nonlinear and the waveform is deformed
pump. Mostly, these effects cause wideband regions in the spec- during the propagation, but this phenomenon can be considered
trum, or a sharp peak at the so-called blade passage fre- as a part of the aforementioned nonlinear effects. In this sense,
quency(BPF)[8], defined as it can be declared that the spectra of pressure signals are mostly
𝑓𝑓𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑧𝑧 ∙ 𝑓𝑓𝑟𝑟 , (6) resemble to those of vibration measurements.
Naturally, the final form of a spectrum is resulted by the sum
where 𝑧𝑧 is number of the blades. This frequency is generated of every operating effect. If a frequency component is generated
mainly due to the gap between the impeller and the tongue of by multiple different physical sources, their amplitudes are
the volute. The optimal size of the gap is around 4-6% of the added. In general, it is hard to separate these individual effects;
impeller diameter. Decreasing the gap results in higher effi- however, it can be stated that if a pump contains many mechan-
ciency, but the pressure fluctuations and the transmitted noise ical failures, the waterfall diagrams are dominated by sharp
also increases [9, 10]. peaks at 𝑓𝑓r and its integer multiples. Otherwise, mostly broad-
It has to be mentioned that due to instability reasons, the con- band regions and a peak at the blade passage frequency are can
stant rotational speed of the driving motor is not always guaran- be observed in the waterfall diagrams as a result of hydrody-
teed during a measurement causing a phenomenon called fre- namical effects.
quency modulation [11]. As a result, small amplitudes appear in
the spectrum slightly under and above of the frequency 𝑓𝑓𝑟𝑟 and 4 Results
its multiples. First of all, it is important to emphasize that most of the state-
Another important phenomenon have to be considered is the ments in this paper are often only assumptions that need more
nonlinearity of the system, which causes peaks also at integer detailed investigation, but they can be a good basis of further
multiples of 𝑓𝑓r called harmonics. It is quite difficult to separate research and applications.
whether a frequency component is generated directly by a phys-
ical effect or it is only a harmonic of some lower frequency. 4.1 Pump ‘A’
Fortunately, the harmonics of a physical effect usually have ex- As it is mentioned before, pump ‘A’ is a much more used,
ponentially decreasing amplitudes [12]. Therefore, the non- worn pump. Consequently, the possibility of many mechanical
monotonic decrease of the peaks at integer multiples of 𝑓𝑓𝑟𝑟 indi- failures and misalignments are very likely. Figure 5 shows the
cates the presence of mechanical vibrations in addition to unbal- waterfall diagram of pump ‘A’ at low rotational speed 𝑛𝑛 =
ance. 630 1/min. It can be observed that a sharp peak dominates the
It is important to mention that the values of Table 1 corre- spectra at 𝑓𝑓 ∗ = 1, most probably due to unbalance. The ampli-
spond to vibration measurements on the housings of the bear- tude of this peak increases monotonically with the flow rate. The
ings. However, every mechanical vibration propagates into the harmonics are presumably the effect of nonlinearity due to their
exponentially and monotonically decreasing amplitudes. The
Table 1 Summary of the physical effects acting on pumps (based on [7]) significant peak at 𝑓𝑓 ∗ = 7 is the blade passage frequency as
pump ‘A’ has seven blades. According to Figure 5, pump ‘A’
Physical effect Frequency component
unbalance 𝑓𝑓𝑟𝑟
bent shaft (1 − 2) ∙ 𝑓𝑓𝑟𝑟
Mechanical

pump and motor mis-

(1 − 4) ∙ 𝑓𝑓𝑟𝑟
alignment
bearing misalignment (2 − 3) ∙ 𝑓𝑓𝑟𝑟
internal assembly
(2 − 6) ∙ 𝑓𝑓𝑟𝑟
looseness
impeller-volute inter-
Hydrodynamical

𝑓𝑓𝐵𝐵𝐵𝐵𝐵𝐵
action
non-uniform flow
𝑓𝑓𝐵𝐵𝐵𝐵𝐵𝐵
distribution
cavitation high frequency wideband
region Fig. 5 Waterfall diagram of pump ‘A’ at 𝑛𝑛 = 630 1/min. A dominant excita-
motor instability frequency modulation tion (most probably unbalance) can be observed at f ∗ = 1. Upper harmonics
Other
non-linearity upper harmonics are decreasing exponentially. A slightly dominant peak can be seen at 𝑓𝑓BPF.

4 Period. Polytech. Mech. Eng. Csanád Kalmár, Ferenc Hegedűs

 Fig. 6 Waterfall diagram of pump ‘A’ at 𝑛𝑛 = 920 1/min. Heavy mechanical
and acoustic noise occurred during the measurements, which can be seen in the Fig. 8 Waterfall diagram of pump ‘A’ at 𝑛𝑛 = 2250 1/min. Effects of other
spectra. This phenomenon disappears as flow rate increases. mechanical excitations become stronger, as multiples of 𝑓𝑓 ∗ = 1 does not de-
crease exponentially.
might be heavily loaded with unbalance. Similar spectra can
also be observed also at 𝑛𝑛 = 760 1/min (see the Appendix for Figure 5); that is, the harmonics of 𝑓𝑓 ∗ = 1 are decreasing expo-
the waterfall diagram). nentially due to the nonlinearity. The flow rate dependence of
An additional interesting phenomenon appears on the water- such a heavy mechanical vibration can imply mounting prob-
fall diagram at 𝑛𝑛 = 920 1/min shown in Figure 6. High me- lems of the impeller. As the flow rate varies, the unbalanced ax-
chanical and acoustic noise were experienced during the meas- ial and radial forces acting on the impeller changes as well. In
urement. The mechanical vibrations were so heavy that it was case of bad fitting, these forces might “kick” the system to a
audible and caused tremors in the laboratory. This vibration can heavy vibrational state. The traces of this serious mechanical ef-
be easily noticed in the waterfall diagram, as a low-amplitude fect is presented in the waterfall diagrams approximately be-
wideband zone is suppressing the peak at the blade passage fre- tween rotational speeds 𝑛𝑛 = 720 1/min and 𝑛𝑛 = 1650 1/min.
quency (at 𝑓𝑓 ∗ = 7). This is probably the result of the experi- On some specific rotational speeds, due to control issues, the
enced heavy vibrations on the machine. Observe, however, that driving motor is unable to keep the rotational speed at a constant
the other peaks at 𝑓𝑓 ∗ = 4 or 5 can still be seen, perhaps as a value, but it fluctuates with low frequency resulting in the afore-
result of the increasing effect of some other mechanical effects. mentioned frequency modulation. This effect can be noticed for
It can be also noticed that the peaks at 𝑓𝑓 ∗ = 2 and 3 are much instance at 𝑛𝑛 = 1350 1/min shown in Figure 7. It can be clearly
weaker resulting in a nonmonotomic nature of the harmonics. seen that near the sharp peaks at multiples of 𝑓𝑓 ∗ = 1, smaller
That is, the harmonics are influenced by additional (external) “spikes” appear in the spectra that is a presumable indicator for
sources besides nonlinearity. Interestingly the heavy vibrations frequency modulation (see again [11]). This phenomenon ap-
disappear at high flow rates. Above 𝑄𝑄∗ = 1.5, the spectra is pears over the range of the rotational speed between 𝑛𝑛 =
quite similar to the one presented at 𝑛𝑛 = 630 1/min (see again 1350 1/min and 𝑛𝑛 = 1650 1/min. If stable operation is man-
datory, this range should be avoided.
At higher rotational speeds, from approximately 𝑛𝑛 =
1820 1/min up to 𝑛𝑛 = 3000 1/min, the integer multiples of
𝑓𝑓 ∗ = 1 in the spectra does not decrease exponentially as can be
seen in Figure 8 at 𝑛𝑛 = 2550 1/min. This plausibly means that
the effect of other mechanical failures (other than the unbalance)
that excite on multiples of 𝑓𝑓r tend to be more and more signifi-
cant on higher rotational speeds. A potential reason can be that
the mechanical forces are proportional to the square of the rota-
tional speed in general; thus, the effect of such mechanical is-
sues become notable only at higher rotational speeds.
Generally, it can be stated that all of the waterfall diagrams
of pump ‘A’ are mostly dominated by heavy mechanical effects
such as unbalance, bent shaft, assemble looseness, etc. As a re-
Fig. 7 Waterfall diagram of pump ‘A’ at 𝑛𝑛 = 1350 1/𝑚𝑚𝑚𝑚𝑚𝑚. Heavy frequency
sult, hydrodynamical noise is buried in the spectra by the sharp
modulation appears at certain rotational speeds due to control issues.

Condition Monitoring of Centrifugal Pumps year Vol No 5

 Fig. 11 Waterfall diagram of pump ‘B’ at 𝑛𝑛 = 2650 1/min. A wideband re-
Fig. 9 Waterfall diagram of pump ‘B’ at 𝑛𝑛 = 1300 1/min. There is no notice-
gion dominates the spectra up to 𝑓𝑓 ∗ ≈ 4. The only sharp peak is the one at the
able sharp peak at 𝑓𝑓 ∗ = 1 or its multiples, but a wideband region can be ob-
blade passage frequency.
served.

with Figure 7).

peaks caused by the heavy mechanical effects.
The only rotational speed where sharp peaks appear on pump
‘B’ is at 𝑛𝑛 = 1600 1/min, see in Figure 10. Here, a peak at
4.2 Pump ‘B’ 𝑓𝑓 ∗ = 1 can be noticed, similarly to the general behaviour of
In the case of pump ‘B’, since it is a less used and newer ma- pump ‘A’. It is very probable that in this case, the natural (or
chine, we expect less mechanical effects on the spectra. Figure critical) frequency of the system is excited, which is definitely
9 shows a waterfall diagram of pump ‘B’ at low 𝑛𝑛 = an undesired operation as it can cause unnecessary load on the
1300 1/min rotational speed. It can be clearly seen that the bearings as well as other mechanical parts of the pump. An in-
spectra is quite different from the ones of pump ‘A’. There is no teresting question is whether this situation could have been
noticeable sharp peak at 𝑓𝑓 ∗ = 1 or its integer multiples, mean- avoided by proper design of the pump. The amplitude of the
ing that mechanical effects are conceivably negligible in this harmonics are decreasing exponentially possibly indicating the
case. On the other hand, a wideband region can be observed at effect of the nonlinearity.
low frequencies up to 𝑓𝑓 ∗ ≈ 4. The only sharp peak appearing in At higher rotational speeds, the sharp peak at 𝑓𝑓 ∗ = 1 disap-
the waterfall diagram is at 𝑓𝑓 ∗ = 6, which is the blade passage pears. It can be observed at 𝑛𝑛 = 2650 1/min on Figure 11 that
frequency. This might also confirm that hydrodynamical effects the wideband region dominates the spectra again up to 𝑓𝑓 ∗ ≈ 4.
are more significant in the case of pump ‘B’. Observe also that In addition, a sharp peak is presented at 𝑓𝑓 ∗ = 6 (blade passage
the amplitudes in the waterfall diagram near the same rotational frequency). The amplitude of this frequency tends to decrease
speed for pump ‘A’ are an order of magnitude higher (compare towards 𝑄𝑄∗ = 1, which is in good agreement with results of
other researchers [6, 13].
In summary, it can be concluded that the spectra in pump ‘B’
is most likely dominated by hydrodynamical effects, since there
are no sharp peaks in the spectra at 𝑓𝑓 ∗ = 1 and its integer mul-
tiples in general (except at 𝑛𝑛 = 1600 1/min). In contrast, the
waterfall diagrams contain wideband regions, and only one peak
at the blade passage frequency, which may also confirm the
dominance of hydrodynamical effects. The detailed analysis of
this hydrodynamical wideband frequency region is beyond the
scope of the present study as it is focusing on the vibration mon-
itoring and health issues of pumps, which mainly have mechan-
ical origin.

5 Conclusions

Fig. 10 Waterfall diagram of pump ‘B’ at 𝑛𝑛 = 1600 1/min. This is the only
By measuring the pressure signals on two different centrifu-
rotational speed where sharp peaks appear at 𝑓𝑓 ∗ = 1 and its multiples, due to
gal pumps, conclusions were drawn about their health condition
one of the natural frequencies of the system.

6 Period. Polytech. Mech. Eng. Csanád Kalmár, Ferenc Hegedűs

through dimensionless waterfall diagrams along their character- of Centrifugal Pump Faults Using Vibration Methods".
istic curves. As a result, it has been found that pump ‘A’, which Journal of Physics: Conference Series, 364, pp. 012139.
2012. DOI:10.1088/1742-6596/364/1/012139
is a more used, older machine, is most probably loaded by sev-
[4] Jensen, J., Dayton, K. "Detecting Cavitation in
eral mechanical vibrations, since all of the spectra are dominated Centrifugal Pumps". Orbit, 2nd Qrt., pp. 26–30. 2000.
by sharp peaks at 𝑓𝑓r and its integer multiples. It can be clearly [5] Čudina, M. "Detection of Cavitation Phenomenon in a
seen that a quite significant load is acting on the pump corre- Centrifugal Pump Using Audible Sound". Mechanical
sponding to 𝑓𝑓 ∗ = 1, which is seemingly due to the unbalance of Systems and Signal Processing, 17(6), pp. 1335–1347.
the impeller. On lower rotational speeds, the harmonics of 𝑓𝑓 ∗ = 2003. DOI:10.1006/mssp.2002.1514
[6] Parrondo, J. L., González, J., Fernández, J. "The Effect
1 are decreasing exponentially supposedly because of the effect
of the Operating Point on the Pressure Fluctuations at
of non-linearity (see again the discussion of Section 3). How- the Blade Passage Frequency in the Volute of a
ever, at higher rotational speeds, these multiples gradually be- Centrifugal Pump". Journal of Fluids Engineering,
come more significant; that is, other physical effects (not only 124(3), pp. 784–790. 2002. DOI:10.1115/1.1493814
the unbalance) that generate peaks between 𝑓𝑓 ∗ = 2 − 4 may [7] Scheffer, C. "Session Five : Pump Condition
have more influence at higher rotational speeds. The instability Monitoring Through Vibration Analysis". In PUMPS:
Maintenance, Design, and Reliability Conference 2008
of the operation of the driving motor was also identified as a
– IDC Technologies (pp. 1–20). 2008.
frequency modulated spectra. [8] Zhang, N., Yang, M., Gao, B., Li, Z., Ni, D.
In the case of pump ‘B’, which is a much newer machine, it "Experimental and Numerical Analysis of Unsteady
has been found that with the exception of one rotational speed Pressure Pulsation in a Centrifugal Pump with Slope
(𝑛𝑛 = 1600 1/min), mostly hydrodynamical effects impact on Volute". Journal of Mechanical Science and
the spectra, due to the precise assembly and installation of the Technology, 29(10), pp. 4231–4238. 2015.
DOI:10.1007/s12206-015-0919-y
system. Most of the waterfall diagrams are dominated by a
[9] González, J., Parrondo, J., Santolaria, C., Blanco, E.
wideband region, and a sharp peak only at the blade passage "Steady and Unsteady Radial Forces for a Centrifugal
frequency. Pump With Impeller to Tongue Gap Variation". Journal
Comparing the waterfall diagrams of the two pumps, one can of Fluids Engineering, 128(3), pp. 454. 2006.
observe that pressure fluctuations on pump ‘A’ are significantly DOI:10.1115/1.2173294
higher. For example, comparing Figure 7 and 9, the maximum [10] Solis, M., Bakir, F., Khelladi, S. "Pressure Fluctuations
Reduction in Centrifugal Pumps: Influence of Impeller
value of the amplitudes are approximately 0.2 and 0.006 bar for
Geometry and Radial Gap". Proceedings of the ASME
pump ‘A’ and ‘B’, respectively, which is more than one order Fluids Engineering Division Summer Conference 2009,
of magnitude difference. Consequently, it might become possi- FEDSM2009, 1. 2009. DOI:10.1115/FEDSM2009-
ble to separate mechanical and hydrodynamical effects acting 78240
on a centrifugal pump. It is clear that mechanical effects cause [11] Guicking, D. Schwingungen (Vibration). [German].
sharp peaks in the spectra at 𝑓𝑓r and its integer multiples up to 2016.
[12] Horgratz, S. H. Nonlinear Dynamics and Chaos.
𝑓𝑓 ∗ = 2 − 4. Moreover, if the harmonics decrease exponentially
Perseus Books Group. 1994.
then those multiples are only the effect of non-linearity. [13] Hodkiewicz, M. R., Norton, M. P. "The Effect of
In summary, we have the opportunity to identify the appear- Change in Flow Rate on the Vibration of Double-
ance of a mechanical failure on a pump. By continuously moni- Suction Centrifugal Pumps". Proceedings of the
toring the amplitudes of the peaks at 𝑓𝑓r or its integer multiples, Institution of Mechanical Engineers, Part E: Journal of
and compare them to former ones, it is possible to act in time Process Mechanical Engineering, 216(1), pp. 47–58.
2002. DOI:10.1243/095440802760075030
and perform maintenance on the system to prevent heavy struc-
tural damage. The authors are aware of the fact that in order to
Appendix
apply this method confidently, it has to be improved further, for
Here is the calibration diagram of the pressure transducer
example comparing with vibration measurements, or generating
with its confidence bands corresponding to 95% significance
some of the mentioned physical effects intentionally.
level, and the equation of calibration:
References
[1] Gülich, J. F. Centrifugal Pumps (Third edit.). Berlin,
Heidelberg: Springer Berlin Heidelberg. 2014.
DOI:10.1007/978-3-642-40114-5
[2] Behzad, M., Bastami, A. R., Maassoumian, M. "Fault
Diagnosis of a Centrifugal Pump by Vibration
Analysis". In Volume 3 (pp. 221–226). ASME. 2004.
DOI:10.1115/ESDA2004-58534
[3] Albraik, A., Althobiani, F., Gu, F., Ball, A. "Diagnosis

Condition Monitoring of Centrifugal Pumps year Vol No 7

Fig. A1 Calibration diagram of the applied pressure transducer with its confi-
dence bounds corresponding to 95% confidence level.

Next, we present all measured waterfall diagrams for the two

pumps:
Pump ‘A’:

8 Period. Polytech. Mech. Eng. Csanád Kalmár, Ferenc Hegedűs

Condition Monitoring of Centrifugal Pumps year Vol No 9
Pump ‘B’:

10 Period. Polytech. Mech. Eng. Csanád Kalmár, Ferenc Hegedűs

View publication stats