G Model

AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS

Agricultural Water Management xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Agricultural Water Management

journal homepage: www.elsevier.com/locate/agwat

Energy and water management for drip-irrigation of tomatoes in a

semi- arid district
Imene Yahyaoui a,∗ , Fernando Tadeo b , Marcello Vieira Segatto a
a
Technological center, University of Espiritu Santo, Brazil
b
Industrial Engineering School, University of Valladolid, Spain

a r t i c l e i n f o a b s t r a c t

Article history: In this research, an autonomous off-grid system for irrigation in semi- arid areas is presented and dis-
Received 13 May 2016 cussed. In these areas precise irrigation is essential: as they are characterized by availability of solar
Received in revised form 12 July 2016 radiation, solar irrigation (supported by photovoltaic panels and batteries) is considered here. The cor-
Accepted 2 August 2016
rect operation of these installations is a necessity to ensure the correct crop irrigation and to extend
Available online xxx
the components lifetimes (batteries in particular). These objectives can be ensured by a management
system that correctly handles energy and water requirements. In this research, the energy and water
Keywords:
management for a photovoltaic water pumping installation used for irrigating tomatoes is developed
Water
Tomatoes irrigation
by integrating fuzzy logic inside the control system. This management system first evaluates the water
Photovoltaic energy volume needed by tomatoes during the vegetative cycle considering a detailed model for the tomatoes
Fuzzy logic evapotranspiration and irrigation frequency, following the site and crops characteristics. Based on this
Management and the energy availability a control algorithm decides the switching of the relays which connect the
Water losses main plant components (panels, batteries and water pumps). The control algorithm fulfills the objectives
by considering criteria related to the water volume needed to irrigate the crops, to the safe operation
of the batteries and the continuous operating of the pump. The algorithm is tested in two cases study:
during normal operation and during faults related with water losses. The obtained results confirm that
the irrigation demand is fulfilled, and autonomy is ensured during the vegetative season with a reduced
use of the batteries.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction
Tomato is a drought sensitive plant, since its yield decreases con-
siderably after short periods of water deficiency (Rodríguez et al.,
2014). The regularity in watering the plants is important, especially
during flowering and fruit formation (Reca et al., 2013; Rinaldi et al.,
2013). Indeed, the needed water amount depends essentially on the
type of the soil, the site and the weather characteristics, namely the
amount of rain, the humidity and the temperature (Hillel, 2012).
In semi-arid regions, generally, farmers use furrow or drip irriga-
tion, which is a common method for irrigating tomatoes, thanks to
its economic advantages in saving water and increasing the yield
production (Hillel, 2012; Reca et al., 2015). Farmers adopt this tech-
nique for both greenhouses and outdoor cultivation, for which the
frequency and the water volume for tomatoes irrigation depends
∗ Corresponding author.
E-mail addresses: imene.yahyaoui@ufes.br (I. Yahyaoui),
fernando@autom.uva.es (F. Tadeo), segatto@ele.ufes.br (M.V. Segatto).
on the growing stage of the plant, the rainfall and the irrigation
installation characteristics (Raes et al., 2000; Farneselli et al., 2015;
Liu et al., 2013).
In remote agriculture areas, stand-alone plants are frequently
used for electricity generation for systems which provide the
water volume needed for tomatoes irrigation. For agriculture appli-
cations, diesel engines are generally used for water pumping,
especially in isolated and remote areas, since they are reliable,
easily available and easy to use (Al-Smairan, 2012). However,
experience demonstrated that there are significant limitations
associated with using gensets for power generation. For instance,
the high operation and maintaining costs are its main disadvan-
tages (D’Ambrosio and Ferrari, 2015; Kumar et al., 2016) and the
environmental pollution (Chen and Hashim, 2016; Yahyaoui et al.,
2015a). Hence, taking advantage of the decrease in the renew-
able energy cost, consequently, water pumping installations based
on renewable energies are increasingly deployed in remote areas
(Campana et al., 2015).
Photovoltaic Powered Electric Water Pumping Systems
(PPEWPS) is the most common installation used for water pump-

http://dx.doi.org/10.1016/j.agwat.2016.08.003
0378-3774/© 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
2 I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx

applications, namely fuzzy logic (Sami et al., 2014; Paucar et al.,

Nomenclature1 2015).
In fact, this tool (fuzzy logic) showed its efficiency in con-
a temperature coefficient K −1 trol issues, namely deciding the irrigation schedule and nutrient
AC Alternating current (A) injection, depending on the climatic parameters (solar radiation,
CR Remaining battery capacity (A.h) humidity, etc.) (Reca et al., 2015; Chung et al., 2015), controlling
Cp Peukert capacity (A.h) the internal climatic variables in greenhouses (Márquez-Vera et al.,
DC Direct current (A) 2016) and for the energy management of autonomous water pump-
dod Depth of discharge ing installations, in which Fuzzy Management Algorithms (FMA)
FMA Fuzzy management algorithm have been used to maximize the pumped water, optimize the use
G solar radiation (W/m2 ) of renewable energy and ensure a safe operation of the battery bank
Gref Reference solar radiation (W/m2 ) (Ouada et al., 2013; Yahyaoui et al., 2014).
Ibat Battery current (A) The efficiency of this tool in various applications is given by its
Ii Instantaneous current supplied to the load (A) ease of use. For instance, in energy management works, fuzzy logic
Iph Generated photo-current at a given irradiance G (A). is a good decision tool, since it gives the possibility to describe sys-
Ir Reverse saturation current for a given ambient tem- tem behaviors, and decide control decisions using linguistic rules
perature (A) (Casillas et al., 2013; García et al., 2013). In addition, based on the
Ir Tref reverse saturation current for the temperature of expert knowledge, the fuzzy rules are written using a simple lin-
reference (A) guistic manner, which describes the adopted approach in taking
Isc short circuit current for a given temperature Ta (A) control decision (Yager and Zadeh, 2012; Bezdek et al., 2012).
Isc Tref Short circuit current for the temperature of refer- Hence, this paper presents a continuation of previous pub-
ence (A) lished works, in which an energy management algorithm for water
K Boltzmann constant pumping system destined to tomatoes irrigating has been stud-
MPPT Maximum power point tracking ied (Yahyaoui et al., 2015c,d). The present research work focuses
Rs Serial resistance of a photovoltaic module () on the energy and water management of an autonomous photo-
Rp the parallel resistance of the photovoltaic module voltaic irrigation plant in case of faults related with water losses
(), in the reservoir and discharged battery bank (Fig. 1). Hence, in
kp Peukert constant this research, the correction of the water losses that can occur
M Function modes 1, 2, 3, 4, 5, 6 is studied. In this research work, an autonomous water pumping
1, 2, 3, 4,
5, 6 pump plant composed of photovoltaic panels coupled to a lead-
n Coefficient of ideality acid battery bank is considered and used to ensure the energy
np Number of parallel photovoltaic modules availability between the system components, even while low or
Pbat Battery power (W) intermittent solar radiation, and to supply a centrifuge pump,
Pi Instantaneous power supplied to the load (W) which pump water into a reservoir. These components are linked
Ppump Power of the pump (W) via controllable relays, which are used to decide the energy flow
Ppv Photovoltaic power (W) between the energy sources. These objectives are performed using
PVP PhotoVoltaic panel an Energy Water Management Algorithm (EWMA) that ensures
q Electron energy (C) pumping the sufficient water volume needed for tomatoes irri-
R1 , R2 , gation. The EWMA is performed using fuzzy logic, which is used
Three switching relays to generate the relays control signals, depending on the measured
R3
Ta Ambient temperature at the panel surface (◦ C) Photovoltaic Panel Generation, the depth of discharge of the bat-
Tref Temperature of reference at the panel surface (◦ C) tery bank, the water level in the reservoir and the water flow. In
V Pumped water volume (m3 ) this research, Mamdani-type fuzzy logic is used within the manage-
Vc Open circuit voltage of a photovoltaic module (V) ment algorithm, since it is simple to use with little technical training
Vg Gap energy (e.V) and can be implemented using standard components, namely Pro-
Vt Ta Thermal potential at the ambient temperature ( C)
o grammable Industrial Controllers (Yager and Zadeh, 2012; Bezdek
t Pumping duration (h) et al., 2012).
Using meteorological measurements, namely the solar radia-
tion and the ambient temperature, and the water volume needed
for the crops irrigation, the EWMA decides the switching of the
ing (Benghanem et al., 2013; Olcan, 2015; Yahyaoui et al., 2015b). relays, which link the installation’s elements. Hence, the water vol-
PPEWPS are promising solutions, especially in small scale instal- ume needed for the crops irrigation can be pumped, the continuous
lations in regions characterized by good amounts of solar energy pump supply and the safe battery bank operation can be guaranteed
over the year (Masoudinejad et al., 2015): it is recommended that, (Fig. 1).
for installing Solar Photovoltaic (SPV) pumps, the average daily The paper is organized as follows: Section 2 details the toma-
solar radiation in the least sunny month should be greater than toes irrigation characteristics. The system components models are
3.5 kW/m2 on a horizontal surface (Henrik, 2007). Hence, this type described in Section 3. The Energy Water Management algorithm
of installations is used in isolated agriculture area, to provide the principle is explained in Section 4, in which the management strat-
water volume needed for irrigation, where the photovoltaic energy egy and the algorithm’s execution are detailed in depth. Obtained
generated should be optimally used. In this sense, several tools results of the EWMA are presented and discussed in Section 5.
have been used to optimize the use of the PV energy in agricultural Finally, Section 6 gives the conclusion.

1
x0i , d0k ,y0j ,f0s ,e0n andO0l are, respectively, the values of the variables x, d, y, f
,e and O in the membership intervals; and εx0i , εd0k ,εy0j , εf0s , εe0n and εO0l are the
range values of x0i , d0k ,y0j , f0s , e0n and O0l , respectively.

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx 3

Fig. 1. Scheme of the off-grid photovoltaic irrigation plant.

2. Tomatoes irrigation p and the mean monthly air temperature T for the corresponding
month, as it is described now (Pereira et al., 2015) Eq. (1):
Generally, drip and furrow irrigation are the most used methods
ET0 = K p(0.46 T + 8.13) (1)
for tomatoes irrigation (Biswas et al., 2015; Lamm, 2015). Although
mulching irrigation contributes to crop production by way of influ- where ET0 is the crops evapotranspiration (mm) and K is the cor-
encing soil productivity and weed control (Biswas et al., 2015), rection factor, expressed by (Pereira et al., 2015):
drip irrigation, which is characterized by its suitability for small
K = 0.03T + 0.24 (2)
and frequent irrigation applications, is selected here. Indeed, drip
irrigation allows the fruit production to be increased and the fruit To obtain the necessary gross water, it is essential to estimate
quality to be enhanced, since the exact water volume needed by the the irrigation losses. For this, an additional water quantity must be
crops is used for irrigation (Ding et al., 2015). Hence, small but fre- provided for the irrigation to compensate for those losses. Thus, the
quent water applications enable the plant to grow well, without any final water volume V needed to irrigate tomatoes is given by Eq. (3)
effect from water-stress, thanks to the frequent water applications (Pereira et al., 2015; Wichelns and Qadir, 2015):
between consecutive irrigation periods (Ding et al., 2015).
1 − lf (1 − LR )
Several researchers have focused on various crops yield V = (kc ETo − rm ) (1 + ) (3)
lf (1 − LR )
improvement using drip irrigation, especially tomato. Indeed, it has
been reported that drip irrigation allows 30- 50% higher tomato where:
yields (Biswas et al., 2015; Chukalla et al., 2015) and its use, either rm : the average monthly rain volume (mm),
alone or in combination with mulching methods, increases the lf : leaching efficiency coefficient as a function of the irrigation
tomato yield over the normal method of irrigation, which rep- water applied
resents 44% savings in irrigation water (Chukalla et al., 2015). LR : the leaching fraction given by the humidity that remains in
Therefore, the irrigation method mainly affects the crops yield pro- the soil, expressed by (Wichelns and Qadir, 2015):
duction.
ECw
Moreover, an efficient irrigation must fulfill the water volume LR = (4)
5 ECe − ECw
needed by the crops. Therefore, irrigation requires a good knowl-
edge of the meteorological parameters of the target region. Among where:
them, the reference crop evapotranspiration (ET0 ) and the rainfall ECw : the electrical conductivity of the irrigation water (dS. m−1 ).
rm , which can be planed for a given 10-days period (Olsen et al., ECe : the crop salt tolerance (dS. m−1 ).
2015; Linquist et al., 2015).
In fact, in the literature, several models have been used 3. System components modelling
to describe the crops evapotranspiration. For instance, some
researchers used the Penman-Monteith method, which depends As it has previously been mentioned, the studied plant is com-
essentially on the net radiation at the crop surface, the mean air posed of PV panels, a battery bank interconnected via controllable
temperature, the soil heat flux, the saturation and mean actual relays (Fig. 1). The power generated supplies a centrifuge pump,
vapor pressure, the water density and the wind speed (Fleischer which pump water to a reservoir. The system components models
et al., 2015). Other works presented the evapotranspiration as a are now explained:
function of the sunlight duration and the air temperature (Obid
et al., 2013). For instance, the Blaney-Criddle method includes the 3.1. PV panels model
seasonal crop coefficient kc , in addition to the sunlight duration and
the air temperature, which provides better patterns of the needed A one-diode based non-linear model is used for the management
water volume (Pereira et al., 2015). Hence, in this research study, algorithm, using an ideality factor to describe the diode’s perfor-
the Blaney-Criddle method is used to model the evapotranspira- mance (Adamo et al., 2011). The model uses the radiation G(t),
tion of tomatoes, since it is simple to evaluate (few parameters are the ambient temperature Ta (t) at the panel surface, and the panel
needed), its performance has been validated with experiments in parameters to evaluate the photovoltaic power Ppv . The model is
the literature (Pereira et al., 2015), and it takes into account of the described byEq. (1)–(5) (Adamo et al., 2011):
growing stage of the crops. Indeed, in this model, the reference crop Vc (t) + Rs Ic (t) Vc (t) + Rs Ic (t)
Ppv (t) = ns np Vc (t)(Iph (t) − Ir (t)(exp ( ) − 1) − ) (5)
evapotranspiration ET0 depends on the ratio of the mean daily day- Vt Ta Rp
time hours for a given month to the total daytime hours in the year G(t)
Iph (t) = Isc (t) (6)
Gref

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
4 I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx

Isc (t) = Isc Tref (1 + a(Ta (t) − Tref )) (7) components (Fig. 2). Hence, the fuzzy algorithm is to decide the
interconnection time of the system elements using only the expert
3
Ta (t) n −qVg 1 1 knowledge (Yager and Zadeh, 2012).
Ir (t) = Ir Tref ( ) exp( ( − )) (8)
Tref nKB Ta (t) Tref In fact, the EWMA is based on four steps: the knowledge base
of the expert, the fuzzification, the inference diagram and the
Isc Tref defuzziyfication (Yager and Zadeh, 2012; Yahyaoui et al., 2014). The
Ir Tref = qVc T
(9)
interconnection time decision of the system components is made
exp( nK T ref )−1
B ref by means of fuzzy rules that fulfill the following objectives:

3.2. Battery model
The photovoltaic panel produces electric energy only when the
solar radiation is available. Hence, the use of a battery bank is nec-
essary to complete the remaining power to the load supply on the
one hand, and to store the excess photovoltaic energy, on the other.
In this paper, a non-linear model for modeling the lead- acid bat-
tery is used (Chaabene, 2009; Yahyaoui et al., 2015c). In addition to
its simplicity, this model has the advantage of using both the bat-
tery current and voltage to describe precisely the battery behavior
when charging or discharging. Its performance is then evaluated
from its depth of discharge dod given by:
The stored charge in the battery CR is given by (Chaabene, 2009):
CR(k)
dod(k) = 1 − (10)
Cp
where:
∂k
kp
CR(k) = CR(k−1) + I (11)
3600 bat(k)
where ∂k is the time between instant k-1 and k and kp is the Peukert,
and Cp is the Peukert capacity, considered constant (A.h).
3.3. Pump
1) Provide the required irrigation when needed, by storing water
in the reservoir.
2) Ensure a continuous power supply, especially during weather
changes.
3) Minimize the use of the battery bank.
4) Protect the batteries against the excessive charge and discharge,
by disconnecting them, respectively, from PVs and the pump
when they are not used.
As photovoltaic power is used to supply the pump and the
batteries, the water pumping is normally performed during the
daylight, to minimize the battery use. This facilitates are to keep
the depth of discharge of the battery bank dod between two fixed
values dodmin and dodmax , for a continuous pump operation (that
stops when the tank is full or the battery discharged).
Hence, the management algorithm decides the switching times
of the three relays Rb , Rl and Rlb , which connect the photovoltaic
system elements (Fig. 3). Thus, it is necessary to establish some cri-
teria that define the algorithm efficiency. These criteria are related
to:
i. The water volume in the reservoir L.
ii. The photovoltaic energy produced by the panel Ppv .
iii. The battery depth of discharge dod.

Generally, water pumps supplied by induction machines (IM)

The management criteria are defined as follows:
are commonly used, thanks to the simplicity in control and its cheap
price. Hence, the mechanical power Ppump of the water pump is
a) When the reservoir contains enough water, store the excess of
given by (Zulkifli et al., 2015):
photovoltaic energy in the batteries.
VgH b) Maintain a high water level in the reservoir to guarantee the
Ppump = (12)
p t water volume needed for the crop irrigation.
c) Ensure a depth of discharge dod less than dodmax to protect the
where:
battery against deep discharge, and greater than dodmin to pro-
Ppump : the pump power (W),
tect it from excessive charge.
V: the pumped water volume (m3 ),
d) Ensure a margin of 10% of the photovoltaic power: the pump
g: the gravity acceleration (m/s2 ),
can be connected only to the panel if the measured photovoltaic
: the water density (Kg/m3 ),
power is 10% higher than the required power by the pump, to
Hh : the head height (m),
guarantee a continuous power supply for the pump.
p : the pump efficiency,
t: the water pumping duration (h).
During the day, the instantaneous power P pump verifies that:

4. Energy water management algorithm Ppump = Ppv + P̄Bat (13)

To ensure pumping the water volume needed to the crops
irrigation and a safe operation for the system, an Energy Water
Management Algorithm (EWMA) is proposed here. The EWMA aims
to fulfill the water volume needed to irrigate the crops and optimize
the use of the electrical energy produced from the photovoltaic
system (Fig. 1). Since the management is based on a case study,
therefore, it is obvious to choose fuzzy logic as a control tool. The
Fuzzy Management Algorithm (FMA) is explained now (Yager and
Zadeh, 2012; Yahyaoui et al., 2014).
4.1. Management strategy
According to the fourth criterion, the panel supplies the load
alone if it can provide at least 110% of the demand. This criterion is
to guarantee the stability of the supply. Thus:
Ppv ≥ 1.1Ppump (14)
When the water volume in the reservoir is not sufficient for the
tomatoes irrigation, water pumping is also performed during the
night using the battery bank. Hence, the pumping duration t is
evaluated based on the water pumping flow and the water volume
to be pumped. This is can be described by:
t = tf -ton = (L-Vi )/Q (15)

A management algorithm is established to meet the crops’ water where:

need through the control of the relays, which link the system’ tf : the time of finishing the water pumping (h),

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx 5

Fig. 2. The structure of the proposed energy management algorithm.

Fig. 3. Energy and water management strategy.

ton : the time of starting the water pumping (h),
L: the water need for the tomatoes irrigation (m3 ),
Vi : the initial water volume in the tank (m3 ),
Q: the water flow of the pump (m3 /h).
4.2. Switching mode
The proposed energy management algorithm is performed via
two steps: The first step consists in the acquisition of the climate-
related installation site parameters, which allows the photovoltaic
power Ppv to be estimated. The second step is to deduce the load
connection times and duration to the power sources (Fig. 3). Hence,
following the objectives listed above, six operating modes for the
three relays Rb , Rl and Rlb have been defined:
3) The third mode (mode 3) consists in pumping water and charg-
ing the battery with the energy in excess. In this case, the relays
Rl and Rb are on.
4) When the reservoir is full, the photovoltaic energy produced by
the panel is used in total to charge the battery. This is possible
when the battery is discharged and it corresponds to mode 4.
5) The relay Rl is switched on during the fifth mode (mode 5),
to allow the pump supplying. This is possible when the panel
produces the sufficient power to the pump with an excess of
10%.
6) During mode 6, only the relay Rlb is switched on. This mode is
possible during the night when the water volume in the reser-
voir is less than the volume needed to irrigate the crops for the
corresponding month.

1) At night, in normal conditions, the volume in the tank is full, so

all the switches are off (mode 1). This mode is maintained during
the irrigation period where the tank volume decreases.
2) In the early hours of the morning, mode 2 is possible since the
battery and the panels provide the pump with electric power to
ensure the water pumping. In this case, the relays Rl and Rlb are
on.
4.3. Fuzzy management algorithm
Fuzzy decisions are built upon four steps (Yahyaoui et al., 2014):
the creation of the knowledge base, the fuzzification, the infer-
ence diagram, and the defuzzification. These four steps are now
presented in detail.

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
6 I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx

4.3.1. Knowledge base To decide the switching of the relays Rl , Rb , Rlb , depending on
The knowledge base is generated on the basis of specifications the fuzzy variables x, d and v, two fuzzy sets are planned O =
analysis: (on, off ). They cover the domain O = [0, 1] and verify ∀ o ∈ O:

• Photovoltaic power Ppv  off rl

(o) +  on rl (o) = 1

{  off rb
(o) +  on rb (o) = 1 (22)
The photovoltaic generated power Ppv is periodically measured
and then partitioned in three fuzzy sets that cover the interval X =  off rlb
(o) +  on rlb (o) = 1
[0, Ppv max ] at low, medium and high generation levels, respectively:
where the switching controls given to relays are provided by
∀x ∈ X , L (x) + M (x) + H (x) = 1 (16) the membership functions corresponding to rl , rb , rlb respectively,
evaluated at o.
where L (x), M (x) and H (x) are, respectively, the low, medium
Based on this structure, the fuzzy rules for the relays’ switching
and high membership functions at the measured power level x.
time are classified according to three intervals of dod:
• Battery dod dod ∈ X = [0, ddLmax ]: the panels and/or the battery bank sup-
ply the pump,
dod ∈ Y = [ddLmin , ddMmax ]: supplying the pump is preferred
It is composed of three fuzzy sets that cover the interval D =
than charging the battery,
[0, dod max ] at low, medium and high production levels, respec-
dod ∈ Z = [ddMmin , ddLmax ]: charging the battery bank is pre-
tively, and verify:
ferred to supplying the pump, when the panel produces insufficient
∀d ∈ D , dL (d) + dM (d) + dH (d) = 1 (17) power to the pump.

where dL (d), dM (d) and dH (d) are, respectively, the low,
medium and high membership functions of dod d. 4.3.2. Fuzzification
• Photovoltaic power Ppv
• Stored water v

The membership functions of L (x0i ), M (x0i ) , H (x0i ) corre-

The third partition is composed of three fuzzy sets in the interval
sponding to Ppv are expressed as follows:
V = [0, Vmax ] which verify:

∀v ∈ V , vL (v) + vM (v) + vH (v) = 1 (18) 1 if 0 < x < xLmin
x −x
where vL (v), vM (v) and vH (v) are, respectively, the member- L (x0i ) = { 0i if xLmin < x < xLmax (23)
εx0i
ship functions of v.
As the definition of low, medium and high depends on the use 0 otherwise
of the auxiliary sets, the following fuzzy variables are defined:
x − x0i
if xMmin 1 < x < xMmin 2
εx0i
 Month M:
1 if xMmin 2 < x < xMmax 1
M (x0i ) = { (24)
This partition is composed of as many fuzzy sets as months, x0i − x
if xMmax 1 < x < xMmax 2
given by the interval M = (m1 , m2 , ..., mt ) and verify: εx0i
0 otherwise
∀m ∈ M, m1 (m) + m2 (m) + ... + mt (m) = 1 (19)

where mi (m) are the membership functions corresponding to the 1 if x > xHmax
month m. x − x0i
H (x0i ) = { if xHmin < x < xHmax (25)
εx0i
 Water level L
0 otherwise

This partition is composed of as many fuzzy sets as months,

• Battery depth of discharge dod
denoted by the interval L = (l1 , l2 , ..., lt ). The interval of the pos-
sible water L = [0, Lmax ] is covered by these sets and verify:
The membership functions of dL (d0k ), dM (d0k ), dH (d0k ) cor-
∀l ∈ L, l1 (l) + l2 (l) + ... + lt (l) = 1 (20) responding to dod are expressed as follows:
where li (l)is the membership function corresponding to li evalu-
1 if 0 < d < ddLmin
ated at l.
d0k − d
dL (d0k ) = { if ddLmin < d < ddLmax (26)
• Power difference P εd0k
0 otherwise
This partition is composed of two fuzzy sets F = (f1 , f2 ) and ver-
ify: d − d0k
if ddMmin 1 < d < ddMmin 2
εd0k
∀f ∈ F, f1 (f ) + f2 (f ) = 1 (21)
1 if ddMmin 2 < d < ddMmax 1
where fe (f ) is the membership function corresponding to fe eval- dM (d0k ) = { (27)
uated at f . d0k − d
if ddMmax 1 < d < ddMmax 2
εd0k
• Relays 0 otherwise

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx 7

1 if d > ddHmax explained previously. The control signals of the relays are given
following this equation:
d − d0k
dH (d0k ) = { if ddHmin < d < ddHmax (28)
εd0k 1
0 otherwise ron rl on dron
0
• Water volume v r0l,b,lb = (36)
1
rl on dron
The membership functions of vL (v0j ), vM (v0j ), vH (v0j ) corre-
sponding to the water volume v are expressed as follows: 0

1 if 0 < v < vvLmin 4.3.4. Defuzzification

v0j − v The control of the three relays is deduced by (Fig. 3):
vL (v0j ) = { if vvLmin < v < vvLmax (29)
εv0j If rl,b,lb < 0.5 then Rl,b,lb is off (37)
0 otherwise
If rl,b,lb > 0.5 then Rl,b,lb is on (38)
v − v0j
if vvMmin 1 < v < vvMmin 2 5. Results and discussions
εv0j
1 if vvMmin 2 < v < vvMmax 1
vM (v0j ) = { (30) To test the EWMA efficiency, the algorithm is validated using
v0j − v measured climatic data of an agricultural land planted with toma-
if vvMmax 1 < v < vvMmax 2
εv0j toes, situated in Northern Tunisia (latitude: 36.64◦ , longitude:
0 otherwise 9.60◦ ) and characterized by its semi-arid climate. This application
is prompted by the fact that tomatoes must be irrigated regularly,
1 if v > vvHmax especially during flowering and fruit formation. The irrigation is
v − v0j gravity-based: 200 m3 /h just before sunrise, to irrigate a 10 ha field
vH (v0j ) = { if vvHmin < v < vvHmax (31) by a low- pressure gravity-driven drip system.
εv0j
0 otherwise
5.1. Algorithm parameterization

• Power difference P
5.1.1. Photovoltaic Power Ppv
The photovoltaic power Ppv is classified as follows:
The membership functions of fL (f0e ), fH (f0e ) corresponding to If Ppv ∈ [0 10] then Ppv is considered low
P are expressed as follows: If Ppv ∈ [10 4500] then Ppv is considered medium
1 if 0 < f < ffLmin If Ppv ∈ [4500 10000] then Ppv is considered high

f0e − f
fL (f0e ) = { if ffLmin < f < ffLmax (32) 5.1.2. Battery depth of discharge dod
εf0e The battery’ non-linear model detailed in the subsection 3.2 is
0 otherwise used here to evaluate the dod, which is classified as follows:

1 if f > ffHmax If dod ∈ [0 0.02] then dod is considered low .

f − f0e
fH (f0e ) = { if ffHmin < f < ffHmax (33) If dod ∈ [0.02 0.9] then dod is considered medium .
εf0e
0 otherwise
If dod ∈ [0.9 1] then dod is considered high .
• Switching control of the relays Rb , Rl , Rlb
5.1.3. Water volume V needed for crops irrigation
The relay membership functions off rl ,rb ,rlb (o0z )and Using the water need model given in Section 2, the water volume
on rl ,rb ,rlb (o0z ) corresponding to the relays Rb , Rl , Rlb are V corresponding to each month of tomatoes’ vegetative cycle at the
expressed as follows: target location is described in Table 2. The values are measured and
provided by the agriculture administry of Medjez El Beb, Tunisia
1 if 0 < o < ooffmin (Table 1).
o0z − o The mean water volume of March (m1 ) is l1 = 60 m3 /day.
off rl ,rb ,rlb (o0z ) ={ if ooffmin < o < ooffmax (34)
εo0z The mean water volume of April (m2 ) is l2 = 100 m3 /day.
The mean water volume of May (m3 ) is l3 = 179 m3 /day.
0 otherwise
The mean water volume of June (m4 ) is l4 = 241 m3 /day.
1 if o > oonmax The mean water volume of July (m5 ) is l5 = 321 m3 /day.
The fuzzification of the water volume depends on the month
o − o0z
on rl ,rb ,rlb (o0z ) = { if oonmin < o < oonmax (35) and is described in Table 3.
εo0z
0 otherwise 5.2. Results and discussion

4.3.3. Inference diagram The management algorithm was implemented and tested by
Based on the fuzzified inputs, the rules set serve to decide simulations using measured data (solar irradiation, ambient tem-
the relays’ switching control, which are deduced using the modes perature, rainfall, etc.) from the target location (Medjez El Beb,

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
8 I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx

Fig. 4. Algorithm response in the case study for a day in March.

Fig. 5. Algorithm response in the case study for a day in April.

Northern Tunisia) for the irrigation season from March to July
(Fig. 2). Obtained results (Fig. 4-10) prove that the algorithm ful-
fills the objectives: relays switching ensures the system autonomy.
The water demand is fulfilled and the battery and load are cor-
rectly disconnected when not used. In fact, the case of unexpected
water extraction between the instants tfi and tfd from the reservoir
is studied here for March and April (Fig. 4 and Fig. 5). The miss-
ing water volume in the reservoir is less than the volume needed
to irrigate the crops. Hence, the starting water pumping time is
evaluated using equation (15). Then, the missed water volume is
pumped using a constant water flow Q = 200 m3 /h, and used to irri-
gate the crops. Thus, the developed algorithm allows the relay Rlb
to be switched on, which connects the battery bank to the pump, so
as to compensate the loss in water volume while the dod is less than
0.8. In this case, the pumping is performed to have the water needed
for the plants irrigation. Hence, to minimize using the battery bank
during the night and to keep it charged, the EWMA ensure pump-
ing only the missing water volume needed for the crops irrigation.
Hence, the EWMA fulfills the objectives O1 and O3.
The FMA is also tested in case of empty reservoir and charged
battery bank during May and June (Fig. 6 and Fig. 7). Indeed, the
battery bank is used to pump water until the water volume needed
to irrigate tomatoes is pumped. In this process, the irrigation start-
ing and finishing times are taken into consideration: the irrigation

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx 9

Fig. 6. Algorithm response in the case study for a day in May.

Fig. 7. Algorithm response in the case study for a day in June.

is finished one hour before sunrise to allow a better absorption of
the water by the crops. Thus, during irrigation, the water volume
decreases, following the constant irrigation flow rate (200 m3 /h). In
the sunrise, since the battery bank is discharged, the photovoltaic
energy generated is used to charge the batteries. Then, when it is
medium charged, the photovoltaic energy generated is used to sup-
ply the water pump and to charge the battery bank. Hence, the
available photovoltaic energy is used to charge the battery since the
battery bank is not full charged, so the relays Rl andRb are switched
on. During all of these modes, the dod is always maintained between
the prefixed values (0.02 and 0.8), which guarantees the battery
safety.
Using meteorological data of July, the case of full reservoir and
empty battery bank is tested (Fig. 8 and Fig. 9). In fact, the since
the battery bank is empty, the water pumping starts when there is
an excess in the photovoltaic power generated. In this case, both
the relays Rl and Rb are switched on, which corresponds to mode
3. Moreover, Fig. 9 shows that the control signals ensure relays
complementary switching (relays Rb and Rlb ) since each relay is
considered on when the membership degree for the relay control
signal is higher than 0.5 otherwise it is off, enabling then a con-
tinuous power supply for the pump and the system autonomy,
where the EWMA ensures pumping the water volume expected
for July and the dod is maintained less than 0.8. Moreover, the

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
10 I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx

Fig. 8. Algorithm response in the case study for a day in July.

Fig. 9. Algorithm response in the case study for three days in July.

relays switching shows that even in rapid changing in atmospheric 6. Conclusion

conditions, the panel is able to operate around the optimal value.
Using the global meteorological data of the solar radiation and
the ambient temperature, the proposed algorithm is evaluated from
March to July, as this is the growing season for tomatoes in the tar-
get location. The results show that the use average of batteries is
minimized, since the battery bank maximum contribution in sup-
plying the pump represents 26% of the panels contribution (Fig. 10).
Moreover, it is clear that the EWMA ensures pumping more water
volume than needed, especially during March and April. This proves
the algorithm efficiency in keeping the battery bank charged and
minimizing its use.
A fuzzy algorithm for the water and energy management
of an autonomous off-grid solar irrigation installation has been
presented and tested. The algorithm makes decisions on the inter-
connection time of the main components (photovoltaic panels,
batteries and water pumps) by controlling the switching of the
relays, and taking into account some constraints related with the
photovoltaic power generated, the battery depth of discharge, the
operating month and the amount of stored water.
This algorithm has been tested for a specific installation for
tomatoes’ irrigation during the vegetative cycle months of toma-

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx 11

Fig. 10. Energy generation evaluation during the tomato vegetative cycle.

Table 1 Table 2
Fuzzification of the knowledge base. Daily water volume needed for tomatoes irrigation.

dod is dL Month

Ppv March April May June July

v L M H ET0 (mm) 3.8 4.44 5.31 6.5 6.56

rm (mm) 0.68 0.88 0.65 0.38 0.14
rlb is on rlb is on rlb is off kc (%) 50 65 80 80 100
vL rl is on rl is on rl is on V (m3 /ha) 6 10 17.9 24.1 32.1
rb is off rb is off rb is on
rlb is on rlb is on rlb is off
vM rl is on rl is on rl is on Table 3
rb is off rb is off rb is on Water fuzzification corresponding to each month M.
rlb is off rlb is off rlb is off
vH rl is off rl is off rl is off M
rb is off rb is off rb is off
L m1 : March m2 : April m3 : May m4 : June m5 : July

dod is dM l1 low low low low low

l2 high medium low low low
Ppv
l3 high high medium low low
v L M H l4 high high high medium low
l5 high high high high medium
rlb is on rlb is on rlb is off
vL rl is on rl is on rl is on
rb is off rb is off rb is on
toes in Tunisia (from March to July), including extreme situations
rlb is on rlb is on rlb is off
vM rl is on rl is on rl is on that cause insufficient water volume in the reservoir or a depleted
rb is off rb is off rb is on battery bank. Using measured data from the target location, the
rlb is off rlb is off rlb is off results show that the algorithm ensures pumping the water vol-
vH rl is off rl is off rl is off ume needed by tomatoes, the system autonomy and the increases
rb is on rb is on rb is on
the batteries lifetime. Moreover, it is important to notice that the
dod is dH
proposed algorithm is general, in the sense that it can be used for
PV irrigation systems of different sizes, by providing the monthly
Ppv
water demand and the energetic requirements.
v L M H As a general conclusion, the efficiency of fuzzy logic has been
rlb is off rlb is off rlb is off demonstrated in combining energy and water management for
vL rl is off rl is off rl is on off-grid pumping installation, and that a simple management algo-
rb is on rb is on rb is on rithm system can improve the operation of off-grid PV systems.
rlb is off rlb is off rlb is off
vM rl is off rl is off rl is on
rb is on rb is on rb is on Acknowledgements
rlb is off rlb is off rlb is off
vH rl is off rl is off rl is off The authors thank Mr. Lamine Yahyaoui and the agriculture
rb is on rb is on rb is on administration of Medjez El Beb, Tunisian Ministry of Agriculture,
for providing us with data. They also acknowledge the support of
the Electric Engineering Department of the Federal University of
Espiritu Santo of Brazil. Dr Yahyaoui is funded by a grant and a
project (FAPES 0838/2015) from the Fundação de Amparo à Pesquisa

Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
G Model
AGWAT-4574; No. of Pages 12 ARTICLE IN PRESS
12 I. Yahyaoui et al. / Agricultural Water Management xxx (2016) xxx–xxx
e Inovação do Espirito Santo (FAPES), Brazil. Pr. Tadeo is supported by
a project from the Spanish ministry of economy and competitive-
ness (MiCInn DPI2014-54530-R, co- funded by FEDER). Pr. Segatto
is partially supported by CNPq/Brazil (CNPq 307470/2012- 1).
References
Adamo, F., Attivissimo, F., Di Nisio, A., Spadavecchia, M., 2011. Characterization and
testing of a tool for photovoltaic panel modeling. IEEE Trans. Instrum. Meas. 60
(5), 1613–1622.
Al-Smairan, M., 2012. Application of photovoltaic array for pumping water as an
alternative to diesel engines in Jordan Badia, Tall Hassan station: case study.
Renew. Sustain. Energy Rev. 16 (7), 4500–4507.
Benghanem, M., Daffallah, K.O., Joraid, A.A., Alamri, S.N., Jaber, A., 2013.
Performances of solar water pumping system using helical pump for a deep
well: a case study for Madinah, Saudi Arabia. Energy Convers. Manage. 65,
50–56.
Bezdek, J., Dubois, D., Prade, H., 2012. Fuzzy sets in approximate reasoning and
information systems, 5. Springer Science & Business Media.
Biswas, S.K., Akanda, A.R., Rahman, M.S., Hossain, M.A., 2015. Effect of drip
irrigation and mulching on yield, water-use efficiency and economics of
tomato Plant. Soil Environ. 61 (3), 97–102.
Campana, P.E., Li, H., Yan, J., 2015. Techno-economic feasibility of the irrigation
system for the grassland and farmland conservation in China: photovoltaic vs.
wind power water pumping. Energy Convers. Manage. 103, 311–320.
Casillas, J., Cordón, O., Triguero, F.H., Magdalena, L. (Eds.), 2013. Interpretability
issues in fuzzy modeling, 128. Springer.
Chaabene, M., 2009. Gestion énergétique des systèmes photovoltaïques. Master
course at the National School for Engineers of Sfax, Tunisia.
Chen, K.S., Hashim, J.H., 2016. A review on co-benefits of mass public
transportation in climate change mitigation. Sustain. Cities Soc., in press.
Chukalla, A.D., Krol, M.S., Hoekstra, A.Y., 2015. Green and blue water footprint
reduction in irrigated agriculture: effect of irrigation techniques, irrigation
strategies and mulching. Hydrol. Earth Syst. Sci. 19 (12), 4877–4891.
Chung, C.C., Chen, H.H., Ting, C.H., 2015. Fuzzy logic for accurate control of heating
temperature and duration in canned food sterilisation. Eng. Agric. Environ.
Food.
D’Ambrosio, S., Ferrari, A., 2015. Effects of exhaust gas recirculation in diesel
engines featuring late PCCI type combustion strategies. Energy Convers.
Manage. 105, 1269–1280.
Ding, R., Kang, S., Zhang, Y., Hao, X., Tong, L., Li, S., 2015. A dynamic surface
conductance to predict crop water use from partial to full canopy cover. Agric.
Water Manage. 150, 1–8.
Farneselli, M., Benincasa, P., Tosti, G., Simonne, E., Guiducci, M., Tei, F., 2015. High
fertigation frequency improves nitrogen uptake and crop performance in
processing tomato grown with high nitrogen and water supply. Agric. Water
Manage. 154, 52–58.
Fleischer, E., Bölter, J., Klemm, O., 2015. Summer evapotranspiration in western
Siberia: a comparison between eddy covariance and Penman method
formulations. Hydrol. Processes 29 (20), 4498–4513.
García, P., Torreglosa, J.P., Fernández, L.M., Jurado, F., 2013. Optimal energy
management system for stand-alone wind
turbine/photovoltaic/hydrogen/battery hybrid system with supervisory
control based on fuzzy logic. Int. J. Hydrogen Energy 38 (33), 14146–14158.
Henrik, Lund, 2007. Renewable energy strategies for sustainable development.
Energy 32, 912–919.
Hillel, D., 2012. Soil and Water: Physical Principles and Processes. Elsevier.
Kumar, B.R., Saravanan, S., Rana, D., Nagendran, A., 2016. A comparative analysis
on combustion and emissions of some next generation higher-alcohol/diesel
blends in a direct-injection diesel engine. Energy Convers. Manage. 119,
246–256.
Lamm, F.R., 2015. Cotton, tomato, corn, and onion production with subsurface drip
irrigation-A review. In: 2015 ASABE/IA Irrigation Symposium: Emerging
Technologies for Sustainable Irrigation-A Tribute to the Career of Terry Howell,
Sr. Conference Proceedings, American Society of Agricultural and Biological
Engineers, pp. 1–20.
Linquist, Bruce, Snyder, Richard, Anderson, Frank, Espino, Luis, Inglese, Guglielmo,
Marras, Serena, Moratiel, Ruben, Mutters, Randall, Nicolos, Placido, Rejmanek,
Please cite this article in press as: Yahyaoui, I., et al., Energy and water management for drip-irrigation of tomatoes in a semi- arid
district. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.08.003
Honza, Russo, Alfonso, Shapland, Tom, Song, Zhenwei, Swelam, Atef, Tindula,
Gwen, Hill, Jim, 2015. Water balances and evapotranspiration in water-and
dry-seeded rice systems. Irrig. Sci. 33 (5), 375–385.
Liu, X., Shao, L., Sun, H., Chen, S., Zhang, X., 2013. Responses of yield and water use
efficiency to irrigation amount decided by pan evaporation for winter wheat.
Agric. Water Manage. 129, 173–180.
Márquez-Vera, M.A., Ramos- Fernández, J.C., Cerecero-Natale, L.F., Lafont, F.,
Balmat, J.F., Esparza-Villanueva, J.I., 2016. Temperature control in a MISO
greenhouse by inverting its fuzzy model. Comput. Electron. Agric. 124,
168–174.
Masoudinejad, M., Emmerich, J., Kossmann, D., Riesner, A., Roidl, M., Hompel, M.,
2015. A measurement platform for photovoltaic performance analysis in
environments with ultra-low energy harvesting potential. Sustain. Cities Soc.
Obid, Kareem. R, Khaleel, Basim, 2013. Nife, Kadhim The Comparison between
different methods for estimating consumptive use of water in Iraq. J. Babylon
Univ./Eng. Sci. 21, 27–36.
Olcan, C., 2015. Multi- objective analytical model for optimal sizing of stand-alone
photovoltaic water pumping systems. Energy Convers. Manage. 100, 358–369.
Olsen, J.L., Stisen, S., Proud, S.R., Fensholt, R., 2015. Evaluating EO-based canopy
water stress from seasonally detrended NDVI and SIWSI with modeled
evapotranspiration in the Senegal River Basin. Remote Sens. Environ. 159,
57–69.
Ouada, M., Meridjet, M.S., Talbi, N., 2013. Optimization photovoltaic pumping
system based BLDC using fuzzy logic MPPT control. In: Renewable and
Sustainable Energy Conference (IRSEC), 2013 International, IEEE, pp. 27–31.
Paucar, L.G., Diaz, A.R., Viani, F., Robol, F., Polo, A., Massa, A., 2015. Decision support
for smart irrigation by means of wireless distributed sensors. In: Microwave
Symposium (MMS), 2015 IEEE 15th Mediterranean, IEEE.
Pereira, L.S., Allen, R.G., Smith, M., Raes, D., 2015. Crop evapotranspiration
estimation with FAO56: Past and future. Agric. Water Manage. 147, 4–20.
Raes, D., Sahli, A., Van Looij, J., Mechlia, N.B., Persoons, E., 2000. Charts for guiding
irrigation in real time. Irrig. Drain. Syst. 14 (4), 343–352.
Reca, J., García-Manzano, A., Martínez, J., 2013. Optimal pumping scheduling for
complex irrigation water distribution systems. J. Water Resour. Plann. Manage.
140 (5), 630–637.
Reca, J., García-Manzano, A., Martínez, J., 2015. Optimal pumping scheduling
model considering reservoir evaporation. Agric. Water Manage. 148, 250–257.
Rinaldi, M.M., Thebaldi, M.S., da Rocha, M.S., Sandri, D., Felisberto, A.B., 2013.
Postharvest quality of the tomato irrigated by different irrigation systems and
water qualities. IRRIGA 18 (1), 59–72.
Rodríguez, D., Reca, J., Martínez, J., Lao, M.T., Urrestarazu, M., 2014. Effect of
controlling the leaching fraction on the fertigation and production of a tomato
crop under soilless culture. Sci. Hort. 179, 153–157.
Sami, M., Shiekhdavoodi, M.J., Pazhohanniya, M., Pazhohanniya, F., 2014.
Environmental comprehensive assessment of agricultural systems at the farm
level using fuzzy logic: a case study in cane farms in Iran. Environ. Modell.
Softw. 58, 95–108.
Wichelns, D., Qadir, M., 2015. Achieving sustainable irrigation requires effective
management of salts soil salinity, and shallow groundwater. Agric. Water
Manage. 157, 31–38.
Yager, R.R., Zadeh, L.A., 2012. An introduction to fuzzy logic applications in
intelligent systems, 165. Springer Science & Business Media.
Yahyaoui, I., Sallem, S., Kamoun, M.B.A., Tadeo, F., 2014. A proposal for off-grid
photovoltaic systems with non-controllable loads using fuzzy logic. Energy
Convers. Manage. 78, 835–842.
Yahyaoui, I., Tina, G., Chaabene, M., Tadeo, F., 2015a. Design and evaluation of a
renewable water pumping system. IFAC-Papers OnLine 48 (30), 462–467.
Yahyaoui, I., Chaabene, M., Tadeo, F., 2015b. Evaluation of maximum power point
tracking algorithm for off-grid photovoltaic pumping. Sustainable Cities and
Society 25, 65–73.
Yahyaoui, I., Chaabene, M., Tadeo, F., 2015c. Energy management for photovoltaic
irrigation with a battery bank. Int. J. Energy Optim. Eng. (IJEOE) 4 (3), 18–32.
Yahyaoui, I., Chaabene, M., Tadeo, F., 2015d. Fuzzy energy management for
photovoltaic water pumping system. Int. J. Comput. Appl. 110 (9).
Zulkifli, A.S., Razali, M., Miza, N., Marsadek, M., Yahya, Z., Juhana Tengku Hashim,
T., 2015. Power plant performance based on power-efficiency curve of
hydroelectric power plant. Appl. Mech. Mater. 785, 516–520.