You are on page 1of 4

31st European Photovoltaic Solar Energy Conference and Exhibition


J. Muñoz(*), J.M. Carrillo, F. Martínez-Moreno, L.M. Carrasco, L. Narvarte.

Grupo de Sistemas Fotovoltaicos. Instituto de Energía Solar – Universidad Politécnica de Madrid (IES-UPM)
Address: ETS Ingeniería y Diseño Industrial (ETSIDI). Ronda de Valencia, 3. 28012 Madrid (Spain)
Orcid ID:

ABSTRACT: This paper describes the modelling of components for large PV pumping systems, which have been
implemented in an online and free-software simulator of PV systems called SISIFO, which is publicly available at Among other features, this software tool allows the prediction of water pumped as function of PV
size and type and the analysis of system performance.
Keywords: PV System, PV Pumping, Water-Pumping, Software, Modelling, Simulation.


The reduction of cost for PV modules caused by the
massive installation of grid-connected systems has The configuration of the simulated water pumping
increased the economic competitiveness of conventional PV system is displayed in figure 1, which is composed of
decentralised PV applications. In particular, there is a a PV generator, a variable-frequency inverter and an AC
renewed interest in water PV pumping systems, which centrifugal pump.
are the object of this paper, and PV-diesel hybrid
systems, which are discussed in a concurrent paper of this
conference [1].
Large PV water pumps, in the range of hundreds of
kW, are emerging as commercial products for water PV Frequency AC centrifugal
irrigation applications and, even in industrialised generator converter pump
countries, they may compete or complement the Figure 1: Configuration of a water pumping PV
connection to the electricity grid. system.
For example, in Spain, despite the hybridization of
grid and PV pumps is not allowed by the present The detailed modeling of the PV generator and the
regulatory framework, some regional agriculture inverter is beyond the scope of this paper, but it is
associations are studying the possibility of integrating described elsewhere by the authors [2][3]. It is just worth
large stand-alone PV pumps in their current irrigation mentioning that SISIFO allows the simulation of three
systems. static and six sun-tracking PV generators. For example,
These irrigation systems are usually composed by static ground-mounted or building-integrated PV
several centrifugal pumps of several hundreds of kW generators in roofs or façades, and conventional one-axis
with an aggregated capacity in the MW range, and horizontal or two-axes sun-trackers. Sun-trackers
associations are decided to supply one of more of these generate more energy at higher equipment cost and some
pumps with PV, which would reduce the contractual of them, e.g., one-axis horizontal solar trackers, may be
power required from the electricity grid at noon when cost-effective in particular energy scenarios at present PV
kWh prices, besides, are relatively high. modules prices.
To assess this integration and analysing technical as We focus here only on the modelling of the system
well economic aspects, it is necessary to predict the curve (pipeline characteristic) and the AC centrifugal
volume of water pumped by the PV system, typically, for pump, which is composed by an AC motor and a
a yearly period. mechanical centrifugal pump. Figure 2 shows the scheme
This paper describes the modelling of components for of the AC pump including some of the power
large PV pumping systems, which have been terminology defined below in this paper.
implemented in an online and free-software simulator of
PV systems called SISIFO, which is publicly available at PH
P1 P2
This simulation tool allows the prediction for water
pumping as function of PV size and type and the analysis AC motor
of system performance. Finally, a simulation example of Centrifugal
a real 20kW PV pumping demonstrator installed in the
Irrigator Community of Alto Vinalopó, Alicante (Spain), Figure 2: Scheme of an AC centrifugal pump.
is presented. This demonstrator includes a North-South
horizontal tracker which improves the efficiency of the 2.1 AC motor model
PV pumping system in terms of m3 of pumped water per The AC motor is characterized by its rated output
kWh of the incident irradiance. mechanical power (P2,NOM), or rated shaft power, and its
The presentation of this paper has been organised power conversion efficiency, M, which is calculated as:
with the following structure. Section 2 describes the
modeling of the system components and Section 3 P2 p2
presents the simulation example. M   (1)
P1 p 2  (k m 0  k m1 p2  k m 2 p22 )

31st European Photovoltaic Solar Energy Conference and Exhibition

Where p2=P2/P2,NOM being P2 the output mechanical The operating point on pump curve may be calculated
power of the motor and P1 the electric input power of the either graphically or numerically. Obviously, the second
motor. Parameters km0, km1 and km2 are fitted from the method is used in simulations, but let us first describe the
motor efficiency curve, which is normally provided by graphic method, which will be useful to illustrate the
the manufacturer. problem.
Graphically, the operating point is in the intersection
2.2 Centrifugal pump model of pump curve (Equation 2) and system curve (Equation
The characteristics of the pumps are normally 4). Figure 4 shows an example, where the operating point
provided by the manufacturers at a constant, nominal or at rated speed (1=NOM) would be the point A.
rated, speed. Figure 3 shows an example of the variation
of typical pump parameters as a function of the H
volumetric water flow, Q, which is usually expressed in Affinity parabola
[l/s] or [m3/h]. Pump curve
The parameters displayed in Figure 3 are: C
H Head. The curve H-Q of the pump is usually H1 A System curve
called the "pump curve".
P1 Electric input power.
B 1   NOM
P2 Shaft power. H2
P Pump efficiency. It is defined as the ratio
PH/P2, where PH is the hydraulic power,
which is equal to PH=k·H·Q. The constant k
 2  1
depends on the dimensions of H and Q.
Q2 Q1 Q
MP Motor-pump efficiency (MP=MP). It is
defined as the ratio PH/P1.
Figure 4: Graphical example for determining the
Head operation point of the pump with variable speed.

However, the operation point is not constant because

the speed of the pump varies according to the PV power,
which itself depends on solar radiation. The variation of
the pump speed modifies the pump curve and the
Efficiency P operating point. Figure 4 shows the new curve and new
operating point B at a lower speed (2<1). As the
MP system curve does not change, the operating point is
always in this curve.
Hence, the calculation of the operating point requires
Power P1 the determination of pump curves as a function of pump
speed. The solution is obtained using the well know
affinity laws for pumps, which, assuming that the
impeller diameter and water density remain constant,
state that:
Q (Flow)
Figure 3: Typical pump characteristics at constant
Q1  1
speed provided by manufacturers. 
Q2  2
For simulation purposes, some of these 2
H 1  1 
characteristics are fitted with second order polynomials  
H 2   2 
[4]. For example, H-Q and P2-Q curves may be written,
at rated speed, as: 3
P 21   1 
 
H (Q )  k B 0  k B1Q  k B 2Q 2 (2) P 2 2   2 

P 2(Q )  k P 0  k P1Q  k P 2Q 2 (3) In the previous equation, Q is the volumetric water

flow, H is the head and P2 is the shaft power. Subscripts
In order to determine the operating point of the pump 1 and 2 indicate two different operating points, which are
is necessary to know the relationship between H and Q in a parabola that passes through the origin. The equation
for the hydraulic components (pipeline, valves, etc.), of this “affinity” parabola is obtained by solving
which is usually so-called the "system curve". The simultaneously the first two affinity equations:
system curve has been modelled with the following
2 2
equation: H1  1   Q1 
    or
H 2  2   Q2 
H S (Q )  k S 0  k S 2Q 2
(4) (6)
Q 
Where HS is the system head, and kS0 and kS2 are H1  H 2  1 
constants that represent, respectively, static head and  Q2 
friction losses.

31st European Photovoltaic Solar Energy Conference and Exhibition

For example, in Figure 4, points C and B are in the

same affinity parabola. It can be observed that the point C
is not in the system curve (because the static head is
different from zero) and it can not be an operating point.
Nevertheless, it is required as intermediate calculation in
the numerical procedure described below. Finally, it is
worth pointing out that affinity laws assume that pump
efficiency remains constant for points 1 and 2, i.e., that
The final goal of the pump modelling is obtaining the
relationship between the output water flow, Q, and input
shaft power, P2. For this purpose the following numerical
procedure is applied:
1. Assume an initial flow equal to Q2.
2. Calculate the corresponding head on system Figure 6: North-South horizontal one-axis PV tracker
curve, H2=HS(Q2) (point B(Q2,H2)). installed in the Irrigator Community of Alto Vinalopó,
3. Determine the affinity parabola (Equation 6) that Alicante (Spain).
passes through point B.
4. Calculate the intersection of the affinity parabola
and pump curve (Equation 2), to obtain the point
5. Determine the hydraulic power at point C,
6. Determine the shaft power at point C, P2 (Q1).
7. Calculate pump efficiency at point C,
ηP1=PH1/P2(Q1), which is equal to the efficiency
at point B, ηP2= ηP1.
8. Calculate the hydraulic power at point B, PH2. (a)
9. Finally, determine the shaft power at point B as
P22=PH2/ ηP2.
Varying Q2, a set of discrete points for P22 are
obtained, which allow determining the relationship
between flow and shaft power. For simulation purposes,
these points are fitted with a third degree polynomial:

Q ( P 2)  kQ 0  kQ1 P 2  kQ 2 P 2 2  kQ 3 P 2 3 (7)

The previous equation has the typical shape displayed (b)

in Figure 5, where the parameter P2MIN is the minimum
required shaft power for water pumping.



Figure 5: Typical shape of the relationship between

flow, Q, and shaft power, P2. The parameter P2MIN is
the required minimum power for water pumping.
This section presents a simulation example of a real Figure 7: Characteristics of the AC centrifugal pump.
20kWp PV pumping demonstrator installed in the
Irrigator Community of Alto Vinalopó, Alicante (Spain), The characteristics of the centrifugal pump are
which pumps water from a borehole whose static head is displayed in Figure 7. Points marked with circles have
250m. been obtained from manufacturer information and solid
This demonstrator includes a North-South horizontal lines are the simulation models. Figure7-a shows the
one-axis tracker (see Figure 6) which improves the pump curve at the rated speed (2900rpm) together with
efficiency of the PV pumping system in terms of m3 of the system curve. Figure 7-b represents the motor
pumped water per kWh of the incident irradiance. efficiency as a function of the shaft power, which reaches

31st European Photovoltaic Solar Energy Conference and Exhibition

a maximum around 80%. Figure 7-c displays the Table I: Yearly grid-connection costs from a
mechanical shaft power as a function of the flow at rated representative borehole provided by the Irrigator
speed. And, finally, Figure 7-d shows the calculated Community of Alto Vinalopó.
curve using the above described numerical procedure, but Item Value
here as a function of the input electric power, P1. Operation time 2,613h per year
Figure 8 displays the yearly Sankey diagram for the Water pumping 738,679m3
system, where energy losses are indicated as a percentage Electricity consumption 958,939kWh
of the reference yield. In AC, the final yield reaches 2049 Total electricity cost 94,247€
kWh/kWp, of which 1158 kWh/kWp are converted into Per unit costs 9.8 c€/kWh
useful hydraulic energy. In terms of water pumping, they 12.8c€/m3
are equivalent to 1637m3/kWp or 32740m3 per year.
The performance ratio, PR, is defined here as the
ratio of useful hydraulic energy to the reference yield and 14

is equal to 46%. This efficiency may be substantially 12

improved in larger systems, where the pump and motor
efficiencies are higher. 10
The initial investment cost of the demonstrator is

Flow [m /h]

2,4€/Wp which, assuming a system lifetime of 20 years,
translates into a pumping cost equal to 7.3c€/m3. 6
Obviously, this cost should decrease for larger systems 4
and by economies of scale, and should be increased with
additional costs (capital, operation and maintenance, 2
etc.). Anyway, the final cost could be economically
attractive taking into account that, at present, the irrigator -10 -5 0 5 10
True solar time [hours]
community where this demonstrator has been installed
spends 12.8c€/m3 (see Table I).
Figure 5: Daily variation of water flow during the
characteristic days of the year.

This paper has presented the modelling of large PV
pumping systems, which has been implemented in
SISIFO, an online and free-software simulator of PV
systems that is publicly available at
Besides, a simulation example of real case study is
described, which illustrates some of the capacities of this
simulation tool.


This work has been possible thanks to the funding of

the FP7 European Programme (Energy) in the project
PhotoVoltaic Cost reduction, Reliability, Operational
performance, Prediction and Simulation (PVCROPS),
Project reference 308468. (


[1] J. Muñoz, JM Carrillo. Modeling and sizing of

large PV-diesel hybrid systems without energy
storage. EU PVSEC 2015, Hamburg.
[2] Muñoz, J., Marroyo, L., Collares-Pereira, M.,
Tyutyuyndzhiev, N., Conlon, M., Elmoussaoui, A.,
and Wilkin, B. An Open-Source Simulation Tool of
Grid-Connected PV Systems. 28th European
Photovoltaic Solar Energy Conference and
Exhibition, 2013, 3882–87.
Figure 8: Yearly Sankey diagram for the simulated PV doi:10.4229/28thEUPVSEC2013-5BV.4.18.
system. Energy losses are indicated as percentage of the
[3] J. Muñoz et al. SISIFO: An online simulator of PV
reference yield.
systems. Technical Reference Manual v1.0.
Available at:
Finally, it is worth mentioning that SISIFO also
allows a detailed analysis of system performance. For [4] Suehrcke, H., J. Appelbaum, and B. Breshef.
example, the user may access to the time series of any Modelling a Permanent Magnet DC
simulated variable (powers, losses, water flow, pump Motor/centrifugal Pump Assembly in a
speed, etc.). For example, Figure 9 shows the daily Photovoltaic Energy System. Solar Energy. Vol.
variation of water flow during the characteristic days of 59, no. 1–3, 1997: 37–42.
the year. doi:10.1016/S0038-092X(96)00117-X.