Study of the Centrifugal Pump Efficiency

at Throttling and Speed Control

Levon Gevorkov and Valery Vodovozov
Department of Electrical Engineering
Tallinn University of technology
Tallinn, Estonia
levon.gevorkov@ttu.ee

AbstractCentrifugal pumping systems are the common The VSD losses consist of the frequency converter losses
hydraulic devices mainly driven by electric drives. Their and motor losses. Since the useful power produced by the
efficiency varies significantly with the change of working point centrifugal pump is a part of the total power applied to its shaft,
location that depends on the positions of the performance and it is important to measure and maximize the pumping system
system curves. This paper is devoted to the study of the pumping efficiency. As it was shown in [8], [9], to achieve the minimal
system efficiency at throttling and speed control. A model-based pumping lifecycle cost, it is necessary to operate it as close to
calculation method for the single-drive centrifugal pump head the best efficiency region of a pump as possible.
and pressure control is described. The proposed Matlab model
was tested with the help of a real test bench. The results obtained An impact of the VSD on the system efficiency is in the
confirm the model validity and accuracy. focus of this paper. One study of such the impact was presented
in [10] where the losses of an induction motor of 2.2 kW,
Keywords energy efficiency, pumps, modelling, pressure 14 Nm fed from the frequency converter was explored. It was
control, simulation, variable speed drives. shown that its maximal efficiency depends on the motor and
converter size. Typically, it is of 92 98 % for 1 400 kW
I. INTRODUCTION VSDs. A decreasing effect on the VSD efficiency when the
Pumping systems represent complicated electromechanical motor is driven at a low rotational speed was demonstrated in
equipment including mainly the centrifugal pumps, variable [10] as well.
speed drives (VSD), pipelines, and various types of To analyse an influence of the VSD efficiency associated
sensors [1], [2]. They are among the most significant power with the changeable speed-torque conditions to the total
consumers and commonly work with variable loads those pumping system performance, in [11] the efficiency maps have
characteristics depend on many factors, such as pipe length and been adopted instead of using particular working points, such
material, a number of pumps, valves, drives, etc. [3]. These as the rated power at the rated torque and the rated speed. The
systems increase the kinetic energy of the liquid running required map was built there by processing the efficiency data
through the pump and thus create an appropriate pressure at the from [10] for the set of three-phase induction motors. Using
pump outlet. During the pump system study, the focus is this map, two situations were explored, both for the 30 % drop
mainly on the head, pressure, and flow control systems [4]. of torque. The former study was related to the stable (load-
Different methods and approaches are used to explore independent) speed levels of 250, 400, and 1350 rpm
pumping systems. Among the most efficient ways, the respectively. The latter study concerned the load-dependent
Simulink models are presented [5], [6]. These types of models situation where the speed dropped by 10 % due to the speed
are very useful tools for the pumping station testing and tuning. control inaccuracy. According to the analysis, in the first case
They allow replicating the processes that take place in real the efficiency decreased by 10 % at low speeds and 5 % at high
systems. Simulation cuts down financial costs and time during speeds, whereas in the second case the efficiency decreased by
the assessment process. The control systems developed with 20 % and 15 % respectively. This is the reason why the total
the help of the Simulink tools can be applied for the process efficiency of every pumping system has its individual non-
optimisation and for the real-time management without linear dependence on the flowrate and speed, which cannot be
expensive hardware prototyping. solved analytically.

According to recent European Union statistics [7], the total To assess the efficiency of the pumping system, the
share of energy consumed by the centrifugal pumps varies complex Simulink- and DriveSize-based model is proposed in
from 22 to 30 percent of the worldwide-developed volume. At this paper. It has some advantages useful for the pumping study
the same time, significant energy losses occur inside the pumps and process optimization. The paper is divided into the
mainly related to the volumetric losses inside the casing of a following parts. The next section explains the efficiency-
centrifugal pump and to the losses in the VSDs feeding the oriented head control of a pump. Here, an impact of the VSD
pumping units. efficiency is analyzed. Then, the topology of the model and its
subsystems are explained. The final part deals with
experimental verification of the model.

978-1-5090-20/16/$31.00 2016 IEEE 15th Biennial Baltic Electronics Conference (BEC2016)

Tallinn, Estonia, October 3-5, 2016
199
 II. EFFICIENCY-ORIENTED HEAD CONTROL OF A PUMP throttling using a regulation valve,
The efficiency of a separate centrifugal pump in an the speed control using the VSD.
arbitrary working point is as follows [12]: The first one affects the system curve. It moves the working
ghq pq , point to the left up along the performance curve at the rated
pump = = (1) speed while the valve closes and to the right down along the
P P same curve while the valve opens. The second approach affects
where the performance curve. It moves the working point to the left
h total head, m down along the system curve while the speed drops and to the
q flowrate, m3/h right up along the same curve while the speed grows.
liquid density, kg/m3 Let us Wr is a working point corresponding to the reference
g acceleration due to gravity, m/s2 head hr at the rated speed nr when the pump operates at its
p liquid pressure at the pump intake, Pa highest efficiency max. This point is usually presented in the
P brake power on the pump shaft, W pump manufacturers documentation.
A pump working point Wi is located at the intersection of Any time when the reference head h+ is higher than hr, the
the pump performance curve and the pipeline system curve only control approach is throttling because the speed cannot be
(Fig. 1). increased above the rated level. At this control, the working
point moves left up along the rated performance curve to W+
h, p,
whereas the flowrate q+ drops below the rated value qr.
Performance curves max When the reference head h- is lower than hr, both the
throttling and the speed control may be used. At the former
approach, the working point moves right down along the rated
h+ W+
performance curve to W1- whereas the flowrate q1- grows above
Wr the rated value qr. At the latter approach, the working point
hr
W2- W1- moves left down along the system curve to W2- whereas the
h- flowrate q2- drops below the rated value qr.
Compare these approaches to understand which one is
Efficiency curve better from the efficiency point of view. As follows from
Fig. 1, at throttling the efficiency decreases as the working
point moves outside Wr along the performance curve. And vice
System curves q
versa, as follows from (1) and (2) the pump efficiency does not
q+ q2- qr q1- change at the speed control as the working point moves outside
Wr along the system curve. It follows from the pump
characteristics that, to increase the head above hr, throttling is
to be used whereas to decrease the head below hr, the speed
control is preferable.
Next, involve in this study the VSD feeding the pump from
the supply grid [13]. Taking into account the VSD, the losses
of both system parts are to be taken into account to estimate the
efficiency. It means that the total efficiency of the pumping
Fig. 1. Characteristics of a centrifugal pump.
system is as follows:
The performance curve of a pump represents an hq
characteristic published by a manufacturer for the rated speed = VSD pump , (3)
nr of the pump shaft rotation [13]. Together, a manufacturer
In contrast to the separate pump, the efficiencies of the
usually publishes an efficiency curve of the pump. The similar
characteristics may be obtained for other speeds using the frequency converter and the motor of the VSD depend non-
affinity laws: linearly on both the load and the speed. By linking the load
torque to the speed and power [15],
2 3
qis ni , his ni , Pis ni , P
= = = (2) T= , (4)
qrs nr hrs nr Prs nr n

where an index s designates the coordinates of the points one more affinity law may be added to (2):
located on the s-th system curve, index r the points of the 2
Tis ni . (5)
rated performance curve, and i the points of the performance =
curve for the i-th speed. Trs nr

To adjust the head or pressure, two approaches may be Accordantly (5), the torque changes with a speed in a similar
used [14]: way as the head does. It is clear from this analysis, that to

200
select the proper control method throttling or speed control consumption in different pumping zones, and to emphasise an
every working point has to be assessed and compared from the importance of the VSD efficiency.
VSD efficiency point of view.
In order to implement the models of pumping units, two
separate modelling environments were used. The centrifugal
III. A MODEL OF THE CENTRIFUGAL PUMPING SYSTEM FOR pump and all hydraulic equipment including the pressure and
EFFICIENCY ESTIMATION flow sensors, pipeline, and tanks were modelled in
To verify operability of the method, computer simulation Matlab/Simulink. The model of the VSD consisting of an
was used [16]. The simulation was based on the pump and induction motor and a frequency converter was prepared in the
VSD models. The purposes of the simulation include the ABB DriveSize software tool [17]. Both models were brought
following: to demonstrate the distribution of losses between all together in the Simulink environment. Fig. 2 demonstrates the
devices in the pumping system, to compare the energy simulation architecture in brief.

Drive Size Matlab / Simulink

QH lookup table

Parameters of Calculating
application H
hydraulic
power
Q

Parameters of
converter S vs Pm lookup table
Motor power

Parameters of
motor
S
Calculating
overall
Lookup table of Calculating efficiency
output power S vs P lookup table input power and
optional
Power losses

Lookup table of efficiency

VSD losses
S

Fig. 2. The structure of the developed model for efficiency estimation

Here, the QH lookup table presents the tabulated pump

manufacturers performance curve, S vs Pm lookup table is IV. EXPERIMENTAL VALIDATION
the tabulated pump manufacturers power curve, and S vs P For the model validation, an experimental test bench was
lookup table are the tabulated VSD manufacturers loss used described in [18].
curves. To calculate the hydraulic power, the model (1) (3) is
used. To calculate the input consumed power, (4), (5) are To estimate the efficiency of the pumping system applying
applied. throttling, the rotational speed of the pump was maintained at
the constant level corresponding to its rated speed. On the
During the simulation, the pump was set to operate with a contrary, to estimate the efficiency of the pumping system
defined speed, depending on the test case. The current speed of applying the speed control, the frequency converter was
pumps was applied to the VSD model expressed in the form of adjusted.
the lookup tables derived from the characteristic graphs. These
lookup tables provide the current losses and input power
P, bar

depending on the pump speed and torque for each test case.
The special functionality was computed from the hydraulic
power at the output of the pumping system based on the
provided flow and pump head. The value of the current flow
was measured by the metering functionality of SimHydraulics. BEP
The achieved pump head was calculated by the pump model 1.55
implemented as the lookup table, which expresses the
performance curves of the pump. The input power was
estimated using the hydraulic power, pump input power, and 0.55
VSD losses obtained from the motor and frequency converter
model. To this aim, the model contains the modules for the
separate throttling and speed control. Q, m3/h
Fig. 3. Pressure control.

201
 To analyse data from the model and the experimental test ACKNOWLEDGMENT
bench, the working point of the pump was initially located This research work has been supported by Estonian
close to the BEP as Fig. 3 shows. Ministry of Education and Research (Project SF0140016s11).
After measuring the efficiency at the BEP, the pressure
setpoint was moved from 1.55 bar to a lower level, 0.55 bar. REFERENCES
Both methods were applied the mechanical and the speed [1] I. J. Karassik and T. McGuire, Centrifugal Pumps, Springer Science &
control to adjust the required pressure. Business Media, 2012, 780 p.
The comparative table (Table I) and diagram (Fig. 4) for [2] L. Nelik, Centrifugal & Rotary Pumps: Fundamentals With
Applications, CRC Press, 1999, 152 p.
the simulation and experimentation are shown below.
[3] F. J. T. E. Ferreira, J. A C. Fong and A. T. De Almeida, Ecoanalysis of
variablespeed drives for flow regulation in pumping systems, Industrial
TABLE I. COMPARISON OD EFFICIENCY ESTIMATES Electronics, IEEE Transactions vol. 58, no. 6, pp. 21172125, 2011
[4] V. S. Lobanoff and R. R. Ross, Centrifugal Pumps: Design and
Efficiency, % Application, Gulf Publishing Company, 1992, 577 p.
Pressure and control
Simulation Experimentation [5] M. D. Z. Izquierdo, J. J. S Jimenez and A. M. del Sol, Matlab software
1.55 bar (BEP) 37.5 34.0 to determine the saving in parallel pumps optimal operation systems, by
using variable speed drives, Energy 2030 Conference, Atlanta, USA,
0.55 bar (VSD) 36.2 32.6 2008, pp. 1 8.
[6] J. Ghafouri and H. Khayatzadeh, Dynamic modeling of variable speed
1.55 bar (Valve) 25.8 21.4 centrifugal pump utilizing MATLAB/SIMULINK, International
Journal of Science and Enineering. Investigation, vol. 1, no. 5, pp. 17,
2012.
[7] International Energy Agency. Tracking Industrial Energy Efficiency and
CO2 Emissions, Available at: www.iea.org/publications/freepublications
/publication/tracking_emissions.pdf.
[8] F. Attivissimo, M. Di Febo, P. Paganini, and G. di Mango, Model
based control and diagnostic system for centrifugal pumps, 2014 IEEE
Workshop on Environmental, Energy, and Structural Monitoring
Systems Proceedings, 2014, pp. 16.
[9] J. Viholainen, J. Tamminen, T. Ahonen, J. Ahola, E. Vakilainen, and
R. Soukka, Energy-efficient control strategy for variable speed-driven
paralllel pumping systems, Energy Efficiency, 2012, pp. 115.
[10] P. Angers, Variable frequency drive testing, Motor Efficiency
Performance Standards, Sydney, Australia, February 2009.
[11] V. Vodovozov and I. Bakman, Performance improvement of pumps fed
by the variable speed drives, Electrical, Control and Communication
Engineering, 2013, 4, pp. 45 50.
[12] D. Kaya, E. A. Yagmur, K. S. Yigit, F. C. Kilinc, A. S. Eren and
Fig. 4. Efficiency measured using the simulation and the experimental C. Celik, Energy Efficiency in pumps, Energy Conversion and
method near the BEP at the rated speed of 2760 rpm and for the pressure Management, vol. 49, no. 6, pp. 16621673, Jun 2008.
maintenance of 0.55 bar.
[13] E. Finnemore and J. B. Franzini, Fluid Mechanics with Engineering
Applications, NY: McGraw Hill, Inc., 2002, 500 p.
V. CONCLUSIONS [14] M. Al-Khalifah and G. McMillan, Control valve versus variable-speed
drive for flow control, International Society of Automation, 2013,
A new model for efficiency estimation in the centrifugal http://isa.org/
pump is proposed in the paper. It was tested with the help of [15] V. Vodovozov, Electrical Drive: Performance, Design and Control,
the real centrifugal pump test bench during the mixed pressure Lambert Academic Publishing, Saarbrcken, 2014, 320 p.
control. The model shows enough high accuracy at multiple [16] I. Bakman, High-Efficiency Predictive Control of Centrifugal Multi-
simulation modes of the processes that take place in a real Pump Stations with Variable-Speed Drives, PhD thesis, Tallinn, TTU,
centrifugal pump system. An advantage of the model is based 2016, 149 p. Available at: http://digi.lib.ttu.ee/i/?5352.
on the ability to predict energy efficiency of a centrifugal pump [17] DriveSize, [Online]. Available at: http://ds.nodus.info/starter.aspx
system in various working points during the pressure regulation [18] L. Gevorkov, I. Bakman, and V. Vodovozov, Hardware-in-the-loop
using either the throttling or the speed control dependently on simulation of motor drives for pumping applications, 2014 Electric
the required pressure level and pipeline resistance. The novelty Power Quality and Supply Reliability Conference (PQ), 2014, pp. 203
208.
of developed model is based on the possibility to combine data
from two simulation environments ABB DriveSize and
original Matlab model and to make calculations based on
them. The model can be also helpful at the design stage for the
efficiency calculation and drive sizing.

202