Motor-pump system modelization

A. Hadj Arab
a,
*
, M. Benghanem
b
, F. Chenlo
c
a
Centre of Development of Renewable Energies, BP 62 Bouzareah, Algeria
b
Institute of Electronics, University of Sciences and Technology Houari Boumediene,
BP 32, El-Alia, Bab-Ezzouar 16111, Algiers, Algeria
c
Departamento de Energ as Renovables-CIEMAT. Avda. Complutense 22, Madrid 28040, Spain
Received 9 April 2005; accepted 2 June 2005
Available online 20 July 2005
Abstract
This paper presents two mathematical motor-pump models for photovoltaic (PV) applications.
These models will allow us to obtain the operating point with the PV array and the ow rate of the
pumped water. The modelled motor-pump characteristics are currentvoltage and owvoltage. The
models are established for centrifugal and positive displacement pumps coupled to DC motors.
The experimental data is obtained in our pumping test facility. A simplied method is proposed to
obtain the parameters of these models in order to reduce the number of experimental measurements
to the minimum.
q 2005 Elsevier Ltd. All rights reserved.
Keywords: Motor-pump; PV pumping; Modelization
1. Introduction
Photovoltaic (PV) water pumping is one of the most typical PV application in
developing countries and has the potential to become a major force for social and
economic development. In Algeria many remote villages are not yet connected to the
electric grid and face severe problems of water for drinking and irrigation purposes [1].
There are several theoretical models of pumps and motors [2,3]. Nevertheless, to obtain
the parameters of the models applied to concrete motor-pumps, we need to separate
Renewable Energy 31 (2006) 905913
www.elsevier.com/locate/renene
0960-1481/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2005.06.003
* Corresponding author. Tel.: C213 21 901503; fax: C213 21 901654.
E-mail address: hadjarab@hotmail.com (A. Hadj Arab).
physically the motor from the pump and to measure their currentvoltage (IV) and ow
voltage (QV) characteristics in relation to the torque and the velocity.
For these reasons, we have decided to use an empirical model for the complete motor-
pump system in which parameters are obtained in a pumping test facility, built for this
purpose. It allows us to characterize any pumping system in the ranges from 0 to 100 m for
the pumping head and from 0 to 30 m
3
/h for the ow rate and from 0 to 50 A and from 0 to
250 V for the PV array [4].
In this pumping test facility we obtain, for each xed pumping head, the pumped ow
versus input voltage, simultaneously we measure the consumed current and therefore, we
get the relation between the curves IV and QV.
2. Motor-pump modelization
2.1. Modelization for each pumping head
The experimental part is based on the measurements of two centrifugal pumps and two
positive displacement pumps CP1, CP2, DP1 and DP2. The characteristics of these pumps
are summarized in Table 1. These pumps have been completely characterized in the
pumping test facility in order to proceed to their modelization. To obtain these curves, we
have xed the pumping head, varying the input voltage, and we measured the following
parameters:
The input DC voltage and current
The output ow rate and pumping head
Then we xed another head and we repeated the same process of measurement. In this
way, we obtained two measured curves IV and QV of the motor-pump assembly of each
head.
Figs. 1 and 2 show an example of the experimental values obtained with DP1, which
represents the dependence of the current and the pumped ow versus the voltage for each
head. From these results, we have tried to nd simple mathematical expressions, which
allow us to adjust these experimental values. In the case of the current versus the voltage,
we found that it is linear. In the case of the ow, the experimental points present a certain
curvature, therefore it seems more suitable to use a second-degree equation. For this
Table 1
Technical characteristics of the modelled pumps according to the manufacturer data
Type Floating centrifu-
gal and single stage
Floating centrifu-
gal and multistage
Positive displacement
and submersible
CP1 CP2 DP1 DP2
Motor Brushless DC DC DC DC
Nominal power (W) 450 400 120 300
Range voltage (V) 0100 048 24 1245
Maximum current (A) 8.4 13 4 10
A. Hadj Arab et al. / Renewable Energy 31 (2006) 905913 906
reason, we suggested the following equations [4,5]:
IV ZaV Cb (1)
QV ZcV
2
CdV Ce (2)
0
1
2
3
4
5
6
7
0 5 10 15 20 25 30
Voltage (V)
F
l
o
w

(
l
/
m
i
n
)
DP1
h = 65m
h = 51m
h = 42m
h = 22m
h = 17m
h = 12m
h = 7m
Fig. 2. Example of characteristics owvoltage for DP1 for different heads.
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 5 10 15 20 25 30
Voltage (V)
C
u
r
r
e
n
t

(
A
)
DP1
h = 7m
h = 12m
h = 17m
h = 22m
h = 42m
h = 51m
h = 65m
Fig. 1. Example of characteristics currentvoltage for DP1 for different heads.
A. Hadj Arab et al. / Renewable Energy 31 (2006) 905913 907
with I and Q being positive. Where a, b, c, d, and e are parameters obtained by means of an
iterative Newton technique.
The comparison between the N values of the measured data and calculated is done by
using the value R-square, R
2
.
In the Table 2, we have the example of the parameters a, b, c, d and e of IV and QV
characteristics, for each pumping head for the case of DP1 pump. The Eqs. (1) and (2)
agree very well with the experimental values, and this according to the obtained values of
the regression coefcient R
2
.
2.2. General model for any pumping head
The results obtained with the simple model described have been found to agree very
well with the experimental values. However, the whole values obtained are useful only for
the given experimental heads and we do not obtain information for other heads found by
installer in practice, intermediate values of these are given. For the use of the model in
simulation programs, it is useful to account for a unique whole of parameters,
characteristic of each pump, that allow us to reproduce its behavior in relation to the
head with sufcient reliability. The generalization of the described model for the case of
variable heads for any value, will allow also to reduce greatly the number of experimental
measurements necessary to characterize a real pump model.
To generalize the described model by the Eqs. (1) and (2), it is necessary to introduce
the dependence of the parameters a, b, c, d and e with the head. We have studied other
mathematical ways to analyze the dependence of I(V) and Q(V) to the head, like the
addition of terms dependent on h, for example, this has proved to be less reliable and more
complex.
In the general model that we propose here, the ve parameters a, b, c, d, and e of the
Eqs. (1) and (2) are related to the pumping head h by second-degree equations of the
following form:
ah Za
0
Ca
1
h Ca
2
h
2
(3)
bh Zb
0
Cb
1
h Cb
2
h
2
(4)
ch Zc
0
Cc
1
h Cc
2
h
2
(5)
Table 2
Example of characteristics currentvoltage I(A)V(V) and owvoltage Q(l/min)V(V) of DP1 motor-pump unit
for different heads
Head (m) a b R
2
c d e R
2
7 0.028 0.863 0.98 K0.043 1.758 K12.142 0.99
12 0.029 1.101 0.98 K0.033 1.453 K10.216 O0.99
17 0.029 1.342 0.99 K0.035 1.521 K10.883 0.99
22 0.029 1.570 0.99 K0.029 1.361 K10.193 0.99
42 0.030 2.351 0.98 K0.027 1.404 K12.357 0.99
51 0.028 2.768 0.98 K0.030 1.526 K13.962 O0.99
65 0.021 3.379 0.92 K0.033 1.674 K16.295 O0.99
A. Hadj Arab et al. / Renewable Energy 31 (2006) 905913 908
dh Zd
0
Cd
1
h Cd
2
h
2
(6)
eh Ze
0
Ce
1
h Ce
2
h
2
(7)
The relations for a(h), b(h), c(h), d(h), e(h) are adjusted by the least squares method.
The constants a
0
, a
1
, a
2
, b
0
, b
1
, b
2
, c
0
, c
1
, c
2
, d
0
, d
1
, d
2
, e
0
, e
1
, e
2
of IV and QV
characteristics for the case of CP1, CP2, DP1 and DP2 pumps are given in the Tables 3
and 4.
3. Simplied method for the parameter measurements
It is obvious that for each pump we have measured between 6 and 9 times different
pumping heads and for each measured curve in order to obtain between 15 and 90 values.
These parameters are conditioned by the capabilities of the pumping test facility. The
question is if it is necessary to use a sophisticated pumping test facility which allow us to
do all the measurements, to obtain all the parameters valid for each pump?. To resolve
Table 3
Example of characteristic currentvoltage I(A)V(V) for different pumps
CP1 CP2 DP1 DP2
A
2
K0.0002 0.0006 K7!10
K6
K4!10
K6
A
1
0.01 K0.002 0.0004 0.0004
A
0
0.06 0.016 0.03 0.04
R
2
0.97 O0.99 0.93 0.95
B
2
K0.03 K0.04 K7!10
K6
6!10
K5
b
1
K0.56 0.36 0.04 0.03
b
0
0.13 K2.52 0.59 0.75
R
2
0.95 O0.99 O0.99 O0.99
Table 4
Example of characteristic owvoltage Q(l/min)V(V) for different pumps
CP1 CP2 DP1 DP2
c
2
K0.03 K0.0002 K1!10
K5
3!10
K6
c
1
0.23 K0.004 0.001 K0.0001
c
0
K0.51 0.005 K0.05 K0.01
R
2
0.95 0.96 0.81 0.82
d
2
4.13 0.04 0.0004 K0.001
d
1
K31.24 0.33 K0.03 0.005
d
0
71.24 2.02 1.83 0.10
R
2
0.96 0.97 0.73 0.79
e
2
K145.89 K1.57 K0.003 0.0008
e
1
1082.60 K3.65 0.13 K0.07
e
0
K2342.40 K27.60 K12.07 K9.63
R
2
0.97 0.98 0.94 0.41
A. Hadj Arab et al. / Renewable Energy 31 (2006) 905913 909
this question, we have proposed to use the same model but with more reduced data. If the
general model is valid with less values, it can constitute a useful tool for laboratories or
installations with less technological means. The idea is to reduce the number of
experimental measurements to the minimum and apply the general model to obtain the
characteristic parameters. Then, we compare the obtained results and those obtained with
the complete curves for all the heads.
This simple methodology has been called simplied method because, though the
applied equations are the same as in the general case, the number of the experimental
values are reduced to only nine points of measurement (dened by V, I, Q, h): We have
chosen, in a random manner, three points for different voltages. The only imposed
condition has been that the three values of each curve are sufciently representative, that
is, for values of low ow, high and average, so that the form of each form may be
maintained. In the same way, the chosen heads correspond to low values, high and average
so that the rank application of the simplied model is as extensive that in the general case.
Figs. 3 and 4 show an example of the experimental values obtained for DP1 for three
different heads
Thus to dene the IV and QV characteristics, we need to solve a system of nine linear
equations. For this, at three different heads, we have three points of different
measurements of I, V, and Q with Q being positive.
The aim of this example is to compare whether the results obtained using the general
model or the simplied method instead of the models for each head are valid. The PV
water pumping system consists of a PV array, a motor-pump, a water well, a tank and a
load. The chosen site is Algiers (Algeria). We have developed a computer program with an
0
0.5
1
1.5
2
2.5
3
3.5
0 5 10 15 20 25 30
Voltage (V)
C
u
r
r
e
n
t

(
A
)
DP1
H = 7m
H = 17m
H = 42m
Fig. 3. Example of characteristics currentvoltage for DP1 for three different heads (simplied method).
A. Hadj Arab et al. / Renewable Energy 31 (2006) 905913 910
hourly time step, which predicts the performances of different systems [6]. The Typical
Meteorological Year (TMY) was used for the solar radiation and temperature. The solar
radiation in a tilted surface is calculated using both Frutos [7] and Hay/Mckay [8]
correlations. An analytical model has been used to perform the PV array at different
irradiance and temperature conditions [911]. The tank capacity used was two autonomy
days. The operating point of the system is the intersection point of the currentvoltage
characteristic of both the motor-pump assembly and the PV array. The intersection is
calculated by means of an iterative Newton technique.
Table 5
RECM and EM using the general model and the simplied method for a daily volume of the monthly averages
produced with CP1 for different heads with a PV generator whose characteristics are I
SC
Z12.04 A; V
OC
Z
71.78 V; I
m
Z11.28 A; V
m
Z58.30 V
Head (m) RECM (%) EM (%)
General model Simplied method General model Simplied method
3 3.11 2.69 K3.10 2.72
4 8.69 13.42 8.66 13.39
5 1.65 2.00 K1.66 1.98
6 8.68 5.29 K8.33 K4.84
7 13.99 19.12 13.61 18.96
8 7.57 2.01 K8.28 K0.33
0
1
2
3
4
5
6
7
0 5 10 15 20 25 30
Voltage (V)
DP1
H = 42m
H = 17m
H = 7m
F
l
o
w

(
l
/
m
i
n
)
Fig. 4. Example of characteristics owvoltage for DP1 for three different heads (simplied method).
A. Hadj Arab et al. / Renewable Energy 31 (2006) 905913 911
Table 6
RECM and EM using the general model and the simplied method for a daily volume of the monthly averages
produced with CP2 for different heads with a PVgenerator whose characteristics are I
SC
Z9.03 A; V
OC
Z86.14 V;
I
m
Z8.46 A; V
m
Z69.96 V
Head (m) RECM (%) EM (%)
General model Simplied method General model Simplied method
5 0.48 1.19 K0.47 K1.19
6 1.83 0.79 K1.87 K0.75
7 3.03 1.54 K3.12 K1.54
8 1.08 2.37 1.07 2.44
9 5.71 5.89 5.90 6.11
14 11.65 19.20 K11.89 K19.55
15 4.09 6.10 4.47 K6.15
18 6.67 6.16 6.78 K6.08
23 2.26 12.14 K2.30 K12.23
Table 7
RECM and EM using the general model and the simplied method for a daily volume of the monthly averages
produced with DP1 for different heads with a PV generator whose characteristics are I
SC
Z4.52 A; V
OC
Z21.
53 V; I
m
Z4.23 A; V
m
Z17.49 V
Head (m) RECM (%) EM (%)
General model Simplied method General model Simplied method
7 1.75 2.68 K1.76 2.68
12 1.65 5.50 1.62 5.51
17 0.30 3.07 K0.08 3.12
22 1.78 4.55 1.79 4.59
42 2.59 0.72 K2.65 K0.52
51 0.29 4.70 0.21 4.69
65 1.68 13.35 1.62 14.30
Table 8
RECM and EM using the general model and the simplied method for a daily volume of the monthly averages
produced with DP2 for different heads with a PV generator whose characteristics are I
SC
Z9.03 A; V
OC
Z43.
06 V; I
m
Z8.46 A; V
m
Z34.98 V
Head (m) RECM (%) EM (%)
General model Simplied method General model Simplied method
12 1.19 0.23 K1.20 K0.06
17 2.62 3.43 2.65 3.46
27 4.03 3.60 K4.09 K3.65
32 5.14 5.55 5.20 5.62
37 2.26 1.92 K2.25 K1.90
47 0.27 0.49 0.26 0.48
A. Hadj Arab et al. / Renewable Energy 31 (2006) 905913 912
The comparison between the calculated values of the models for each head and those
calculated by the general model or the simplied method is given by using the root of the
average square error, RECM, and the average error, EM.
EM and RECM for each pump, are shown in Tables 58. As we can see, in general, the
comparison is very favorable for all the pumps and on a large scale of heads.
4. Conclusions
The development of the two motor-pump unit models allows us to obtain the operating
point of the system and the pumped ow rate. The two models are based on experimental
results of several photovoltaic pumps, which have been characterized completely in the
pumping test facility in order to obtain the parameters of the models. The models are
established for centrifugal and positive displacement pumps with DC motors.
It has been demonstrated that measuring nine points (corresponding to three voltages
and three pumping heads) is sufcient to obtain all the parameters; this is very useful if we
do not have a sophisticated pumping test facility.
Acknowledgements
The authors would like to thank very much Professor Lorenzo E. and for valuable
discussions in this work.
References
[1] Posorski R. Photovoltaic water pumps, an attractive tool for rural drinking water supply. Solar Energy 1996;
58(46):15563.
[2] Kou Q, Klein SA, Beckam WA. A method for estimating the long-term performance of direct-coupled PV
pumping systems. Solar Energy 1998;64(13).
[3] Hsiao YR, Blevins BA. Direct coupling of Photovoltaic power source to water pumping system. Solar
Energy 1984;32(4):48998.
[4] Hadj Arab A, Chenlo F, Benghanem M. Loss-of-load probability of photovoltaic water pumping systems.
Solar Energy 2004;76(6):71323.
[5] Hadj Arab A. Optimisation of low power photovoltaic pumping systems Doctorate in telecommunication.
Spain: Polytechnic University of Madrid (ETSIT-UPM); 1999.
[6] Hadj Arab A, Chenlo F, Lorenzo E. Comparison of different PV water pumping systems 2nd world
conference and exhibition on photovoltaic solar energy conversion. Vienna 1998.
[7] Frutos F, Ruiz V, Gutirrez J. Correlation between the solar diffuse and global radiation for Madrid. Revue
Internationale dHeliotechnique. Premier Semestre 1985;712.
[8] Hay JE, McKay DC. Estimating solar irradiance on inclined surfaces: a review and assessment of
methodologies. Int Solar Energy 1985;3:20340.
[9] Chan Jr DSH, Philips JR, Phang JCH. A comparative study of extraction methods for solar cell model
parameters. Solid State Electron 1986;29(3):32937.
[10] Lorenzo E. Solar electricity Engineering of photovoltaic systems. Progensa, Spain 1994.
[11] Hadj Arab A, Chenlo F, Mukadam K, Balenzategui JL. Performance of PV water pumping systems. Renew
Energy 1999;18(2):191204.
A. Hadj Arab et al. / Renewable Energy 31 (2006) 905913 913