Indian

Indian Institute
December of
of Technology
9-11, 2015.
Institute Delhi, India
Technology
9th IFAC symposium
December on Control of Power and Energy Systems
December 9-11,
9-11, 2015.
2015. Delhi,
Delhi, India
India Available online at www.sciencedirect.com
Indian Institute of Technology
December 9-11, 2015. Delhi, India

ScienceDirect
IFAC-PapersOnLine 48-30 (2015) 462–467

Design and Evaluation of a Renewable Water Pumping System

Design and Evaluation of a Renewable Water Pumping System
Design and Evaluation of a Renewable Water Pumping System
Imene Yahyaoui Giuseppe Tina Maher Chaabene Fernando Tadeo
Imene
Imene
Dept. SystemsYahyaoui
YahyaouiEng and Giuseppe
Giuseppe
Dept. Electric, Tina
Tina Maher Chaabene
MaherElectric
Dept. Chaabene Eng Fernando
SystemsTadeo
Fernando
Dept. Tadeo
Eng and
Dept.
Dept. Systems
Systems
Automatic Eng
Eng and
and ElectronicDept.
Dept. Electric,
Electric,
and Computer Dept.
Dept.
National Electric
Electric
Eng School Eng
Eng Dept.
Dept. Systems
Systems
Automatic Eng
Eng and and
Imene Yahyaoui
Automatic ElectronicGiuseppe and Tina
Computer Maher
National Chaabene
Eng School Fernando
Automatic Tadeo
Dept. Automatic
Industrial
Systems EngEng School and Electronic
Univ.Dept. and
Catania, Computer
Electric, Italy National
Univ.
Dept. Sfax, Eng
Electric School
Tunisia
Eng Automatic
Industrial
Dept. Systems EngEng Schooland
Industrial
Industrial Eng
Eng School
School Univ.
Univ. Catania,
Catania, Italy
Italy Univ.
Univ. Sfax,
Sfax, Tunisia Industrial Eng
Eng School
Univ. Valladolid,
Automatic Spain Giuseppe.tina@dieei.uni
Electronic and Computer National EngTunisia
maherchaabane@yahoo. School Industrial
Univ. Valladolid,
Automatic School
Spain
Univ.
Univ. Valladolid,
Valladolid,
imene@autom.uva.es Spain
Spain Giuseppe.tina@dieei.uni
Giuseppe.tina@dieei.uni
ct.it Italy maherchaabane@yahoo.
maherchaabane@yahoo.
comTunisia Univ.
Univ. Valladolid,
Valladolid,
fernando@autom.uva.es Spain
Spain
Industrial Eng School Univ. Catania, Univ. Sfax, Industrial Eng School
imene@autom.uva.es
imene@autom.uva.es ct.it
ct.it com
com fernando@autom.uva.es
fernando@autom.uva.es
Univ. Valladolid, Spain Giuseppe.tina@dieei.uni maherchaabane@yahoo. Univ. Valladolid, Spain
imene@autom.uva.es ct.it com fernando@autom.uva.es
Abstract: The design and evaluation of a stand-alone photovoltaic system for well pumping in agricultural
applicationThe
Abstract:
Abstract: The design
design and
is presented. and evaluation
evaluation
Given of
of aa stand-alone
environmental stand-alone
(irradiance photovoltaic
photovoltaic
and ambient system
system for
for well
temperature), well pumping
pumping
system (PV in
in agricultural
agricultural
and battery
application
application
technology, is
is presented.
presented.
array geometry) Given
Given and environmental
environmental
load (monthly (irradiance
(irradiance
daily and
and
demand) ambient
ambient
data, the temperature),
temperature),
optimal size system
system
of main (PV
(PV and
and battery
components batteryof
Abstract: The design and evaluation of a stand-alone photovoltaic system for well pumping in agricultural
technology,
technology,
PV systems array
array
are geometry)
geometry)
obtained by and
and
a load
load
sizing (monthly
(monthly
algorithm; daily
daily demand)
demand)
specifically the data,
data, the
the
output optimal
optimal
are: the size
size
surface of
of main
main
of PV components
components
array and of
of
the
application is presented. Given environmental (irradiance and ambient temperature), system (PV and battery
PV
PV systems
systems are
are obtained
obtained by aa sizing
bydecisionsizing algorithm;
algorithm; specifically
specifically ofthe
the output
output are: the surface of
of PV array and
and the
battery
technology, packarray capacity.
geometry)The and loadis made ondaily
(monthly the basis
demand) the
data, the are:
estimatedoptimalthe
power surface
sizegeneration,
of main PVcomponents
array
the required the
of
battery
battery
power pack
pack
for the capacity.
capacity.
load, the The
The
water decision
decision
needed is
is
by made
made
the on
on
crops the
the
andbasis
basis
the of
of
batterythe
the estimated
estimated
requirements. power
power The generation,
generation,
proposed the
the
design required
required
is then
PV systems are obtained by a sizing algorithm; specifically the output are: the surface of PV array and the
power
power
evaluated for
for the
using load,
thecapacity.
load, the water
the simulations,
yearly needed
waterdecision
neededonis by the
the crops
byhourly crops
base, and
and the
the battery
battery
performed by requirements.
requirements.
aestimated
specialized The proposed
Thegeneration,
commercialproposed design
design
software, is
is then
then
named
battery pack The made on the basis of the power the required
evaluated
evaluated
PVSyst,fortothe using
using
show yearly
yearly simulations,
simulations, on
on hourly
hourly base,
base, performed
performed by
by a
a specialized
specialized commercial
commercial software,
software, named
named
power load,thatthethewaterproposed
neededoptimal
by the crops size ensures
and the also
battery an requirements.
high reliabilityThe evaluated
proposedbydesign two indices:
is then
PVSyst,
PVSyst,
number of to show
tousing
showyearly
autonomous that
that thethe
daysproposed
proposed
(NAD)on optimal
optimal Losssize
andhourly size
of ensures
ensures
Load also
also by
Probability an high
an a(LOLP).
high reliability
reliability evaluated by
by two
evaluated software, two indices:
indices:
evaluated simulations, base, performed specialized commercial named
number
number of of autonomous days (NAD) and
and Loss of
of Load Probability (LOLP).
PVSyst, toautonomous
show that the days (NAD) optimal
proposed Losssize Load Probability
ensures also an (LOLP).
high reliability evaluated by two indices:
Keywords: photovoltaic panels; algorithm; sizing; pumping; PVSyst.
© 2015, of
number
Keywords:
Keywords: IFAC (International
autonomous
photovoltaic
photovoltaic days Federation
panels;
panels; (NAD) and
algorithm;
algorithm; of Loss
Automaticof Load
sizing;
sizing; Control) Hosting
Probability
pumping;
pumping; PVSyst.
PVSyst. by Elsevier Ltd. All rights reserved.
(LOLP).
has also been combined with artificial neuronal networks and
Keywords: photovoltaic panels; algorithm; sizing; pumping; hasPVSyst.
has also been
been combined
INTRODUCTION
IINTRODUCTION
alsoalgorithms
genetic [9, 10].with
combined with artificial
artificial neuronal
neuronal networks
networks and and
NTRODUCTION genetic
genetic algorithms
algorithms [9,
[9, 10].
10].
The main components of a stand-alone photovoltaic (PV) has However,
also been these combined methods withmay artificial
result neuronal networks
in an oversized and
system
The
The main
system, main components
components
that supplies a given of
of aaload
INTRODUCTIONstand-alone
stand-alone
are: a PV photovoltaic
photovoltaic
array, a battery (PV)
(PV) However, these methods may result in an oversized system
genetic
for one algorithms
However,locationthese [9,
and 10].undersized
methods
an may result one in an
for oversized
another system
location
system,
system, that
that supplies
supplies aa regulator.
given
given loadThe are: aa PV array, aa battery
pack,
pack,
anmain
The MPPT/
an used
MPPT/
charge
components
charge of aload
regulator.
are:MPPT/charge
stand-alone
The
PV
MPPT/charge
array, regulator
photovoltaic battery
(PV)
regulator
for
[15].one
for However,
one The location
location
oversized
these
and
and an
an undersized
case
methods undersized
results
may resultinone
one forinstallation
in for
high another
another location
an oversized location
costs.
system
pack,
must
system,an
be MPPT/
that to charge
operate
supplies a regulator.
correctly
given load The
with
are: MPPT/charge
both
a PV PV modules
array, regulator
a and
battery [15]. The oversized case results in high installation costs.
must be used to operate correctly with both PV modules and [15].
Whereas, The oversized
an undersized case results
case, the in
for one location and an undersized one for another location high
installation installation
is unable costs.to
must
pack, be
batteries.
an used
Thistoappliance
MPPT/ operateregulator.
charge correctly
performsThe with both
onlyPV
notMPPT/charge themodules
charge
regulatorand Whereas, an undersized case, the installation is unable to
batteries.
batteries. This
This appliance
appliance performs
performs not
notavoidonly
onlyPV the charge
themodules
charge and
and Whereas,
supply
[15]. Thethe an
load undersized
oversizedwith the
case case,
energy
resultsthe in installation
needed high [16, 17].is
installation unable
Moreover,
costs.to
discharge
must be of the
used to batteriescorrectly
operate in orderwith to both damage or poor
and supply the load with the energy needed [16, 17]. Moreover,
discharge
discharge of
of the
the batteries
batteries in
in order
order to
to avoid
avoid damage
damage or
or poor
poor supply
the
Whereas, the
installationload
an with
lifetime
undersized theis energy
shorter,
case, needed
the due to [16,
the
installation 17].
excessive
is Moreover,
unable use of
to
energy performance
batteries. This appliance but performs
also the not maximumonly thepower chargepoint and the installation
installation lifetime isreasons,
shorter,the duesizesto the
themust
excessive use ofof
energy
energy
tracking performance
performance
(MPPT) in but
but
such also
also
a way the
the tomaximum
maximum
have an power
power
efficient point
point
PV the
supplybatteries.
the load lifetime
For these
with theis shorter,
energy due
needed to [16, excessive
17]. beMoreover,use
carefully
discharge
tracking of the batteries
(MPPT) ina order to avoid damage or poor the batteries. For these
these isreasons, theand sizes must [15].be
be carefully
tracking
conversion.
energy (MPPT) in
performance in such
such
but a way
also way the
to have
have an
tomaximum efficient
an power
efficientpoint PV
PV the batteries.
selected for each
the installation For specific
lifetime reasons,
shorter,the
application duesizes themust
to location
excessive carefully
use of
conversion. selected
selected
In for
for
[18], each
each
the specific
specific
deterministic application
applicationmethod and
and location
location
and the [15].
[15].
probabilistic
conversion.
tracking the batteries. For these reasons, the sizes must be carefully
Sizing(MPPT)
the components in suchofa these way PV to installations
have an efficient affect theirPV In
In [18],
[18], the
the deterministic
deterministic method
method andonthe probabilistic
Sizing
Sizing the
approach
selected areeach
for used to analyze
specific applicationthe andand
impact thedesign
location probabilistic
[15]. the PV
conversion.
autonomy andcomponents
the cost [1, 2]. of
components of these
theseitPV
Hence, PV is installations
necessary toaffect
installations affect their
their
fix during approach
approach
system andare
are theused
used to
to
battery analyze
analyze
storage the
the
for impact
impact
three on
on design
design
geographic the
the PV
sites PV
in
autonomy
autonomy
the design and
and cost
cost
adequate [1,
[1, 2].
2].
values Hence,
Hence,
for the it
it is
is necessary
necessary
components to
to
sizes, fix
fix during
during
such as In [18], the deterministic method and the probabilistic
Sizing the components of these PV installations affect their system
system
Italy, and
and the
the
characterized battery
battery
by storage
storage
different for
for
valuesthree
three ofgeographic
geographic
sun radiation sites
sites in
in
and
the
the design
design and adequate
adequate
photovoltaic panel values
values
surface for the
for and components
theitthe
components sizes,
sizes,
battery capacity such
such as
as
[3, 4]. approach are used to analyze the impact on design the PV
autonomy cost [1, 2]. Hence, is necessary to fix during Italy,
Italy,
ambient characterized
characterized
temperature. by different
byTodifferent
find the values
values
best of sun
ofgeographicradiation
sun radiation
comprise between and
and
two
the
the photovoltaic panel surface and the battery capacity [3, 4]. system and the battery storage for three sites in
In
In the photovoltaic
fact, for
design
fact, adequate
for
panel
agricultural surface
values
agricultural for and
the the
applications, battery
components
applications,
during
during
capacity
sizes,thesuch
the
[3, 4].
crops
cropsas ambient
ambient
conflicting temperature.
temperature.
obiecjetives To (reliability
To find the
find the best
best and comprise
comprise
costs) abetween
between
fuzzy two
two
logic
In fact, for
vegetative cycle,agricultural
the photovoltaic applications,installationduring sizetheselected
crops Italy, characterized by different values of sun radiation and
the photovoltaic cycle,panel surface and the battery capacity [3, 4]. conflicting
conflicting obiecjetives
obiecjetives (reliabilityapproachand costs)
costs)
wasaaused fuzzy logic
vegetative
vegetative
must guaranteecycle, the water
the
the photovoltaic
photovoltaic
volume installation
installation
needed sizetheselected
size
for selected
crops based multi-objective
ambient temperature. To (reliability
optimization
find the best and comprise fuzzy
between logic
in [19].
two
In
must fact, for
guarantee agricultural
the water applications,
volume needed during for the
the crops
crops based
basedThis multi-objective
multi-objective
paper presents optimization
optimization
a continuation approach
approach of was
was
previous used
used in
in [19].
[19].
published
must
irrigation, guarantee
the the
system water
autonomy volume and needed
the for
battery the
bank crops
safe conflicting
This obiecjetives (reliability and costs) a fuzzy logic
vegetative the
irrigation,
irrigation,
operating
cycle,
[3]. system
the Indeed,
system
the photovoltaic
autonomy
autonomy
knowing theand
and
installation
water the battery
the volume
battery
sizebank
bank
needed
selected
safe
safe
for works
basedThis bypaper
paper
some presents
presents
multi-objective
aa continuation
continuation
the optimization
authors 17],of
[16,approach previous
of where
previous
was used an published
published
ain sizing
[19].
must
operating guarantee
[3]. the
Indeed, water
knowing volume
the needed
water volume for the
needed crops
for works
works
algorithm by
by some
some
has been the
the authors
authors
presented. [16,
[16,
Here, 17],
17],
the where
where
algorithm an
an is aadetailed
sizing
sizing
operating
irrigating
irrigation, [3].
the Indeed,
crops,
thecrops,
system the knowing
site
autonomy the water
characteristics,
and the battery volume
the solar needed
radiation
bank for
safe This
algorithm paper presents
has been
been a
presented. continuation of
Here, probabilisticprevious
the algorithm published
algorithmsimulations,
is detailed
detailed
irrigating
irrigating the
the the site
site characteristics, the solar radiation algorithm
and by has
validated by the presented.
means of hourlyHere, the is
and
operating
and the [3].crops,
the photovoltaicIndeed,
photovoltaic
the panel
knowing
panel
characteristics,
type,thesizing
type, wateraims
sizing
the to
volume
aims
solar
to
radiation
provide
needed
provide
the
for
the
works
and
and
that validated
validated
spans
some
one by
by
year,means
means
authors
using of
ofa
[16, probabilistic
hourly
hourly
widely
17], where an
probabilistic
used general
a sizing
simulations,
simulations,
software for
and
adequate
irrigatingthe photovoltaic
values
the crops, of the the panel
panel type, sizing
surface,
site characteristics, batteryaims to
thecapacity
solarprovide and the
radiation(in algorithm
that spans has
one been
year, presented.
using Here,used the algorithm
aa (Grid-connected,
widely general is detailed
software for
adequate
adequate
some values of
values of thepanel
instances) the panel
panel surface,
surface,
reservoir battery
battery
volume. capacity
capacity
Into this and
and (in
(in
sense, that
the
and spans
study
validatedone
of year,
PV
by meansusing
systems widely used
of (Grid-connected,
hourly probabilistic general software
stand-alone
simulations, for
or
and
some the photovoltaic
instances) the type,
reservoir sizing
volume. aims In provide
this the
sense, the
the study
study
hybrid), of
of
named PV
PV
PVSyst systems
systems
[20] (Grid-connected,
(Figure 1). stand-alone
stand-alone or
or
some
adequate instances)
researchers values the
haveofestablished
the reservoir
panel various
surface, volume. methods
battery In to
capacitythisoptimize
sense,
and (in that spans
hybrid), one year,
named PVSyst using
[20] a (Figure
widely 1). used general software for
researchers
researchers
photovoltaic have
have established
established
installations various
various
components methods
methods
[5]. ForIninstance,to
to optimize
optimize
some hybrid),
the study named
of PVSyst
PV [20]
systems (Figure 1).
(Grid-connected, stand-alone or
some instances)
photovoltaic the reservoir
installations components volume.
[5]. For this sense,
instance, some
photovoltaic
works have
researchers installations
focused on
have established components
developing [5].
analytic
various methods For instance,
methods based
to based some
optimize on hybrid), named PVSyst
I.S IZING [20]
A (Figure
LGORITHM 1).
P RINCIPLE
works
aworks simplehave
have focused
focused
calculation on
onofdeveloping panels analytic
developing
the analytic
surface methods
methods
andinstance,
battery based bankon
on I.S IZING A
I.SIZING LGORITHM P
ALGORITHM PRINCIPLE
RINCIPLE
photovoltaic
acapacity
simple installations
calculation of components
the panels [5].
surface For
and battery some
bank A good sizing must fulfill the electrical demand of the load
aworks
simplehave calculation
using the
focused of
energetic the
on developing panels
balance surface
[6-8].
analytic and
Other
methods battery
works based bank
haveon A
A good
good sizing
sizing I.S must
must fulfill
fulfill
A
[15]. Hence, the main objective is to RINCIPLE
IZING the
the
LGORITHM electrical
electrical
P ensure demand
demand
the load of the
of the load
load
supply
capacity
capacity using
using the
the theenergetic
energetic balance
balancesurface[6-8].
[6-8]. Other
Other works
works bank have
have [15]. Hence, the main objective is to ensure the load supply
aconcentrated
simple
concentrated
on
calculation
on the of cost
the
cost
versus
panels reliability and question
battery [9]. [15]. Hence,
throughout
A good the the the
sizingday,main
mustwhileobjective
charging
fulfill is to
the
the electrical ensure
battery the
demandwith load
the
of supply
excees
the load
concentrated
Moreover,
capacity using some on the
theresearchers cost versus
energetic versus
have proposed
balance
reliability
reliability
[6-8]. Othersizingquestion
question
algorithms
works
[9].
[9].
have throughout
throughout
of theHence,
energythetheday,
day,
and while
while
guaranteeing charging
charging the the
the
waterbattery
battery
volume with
withloadthe
the excees
needed exceesfor
Moreover,
Moreover,
based on the some
some researchers
researchers
minimization have
have
ofversus proposed
proposed
cost functions, sizing
sizing
using algorithms
algorithms
the Loss[9].of [15].
of the energy and main objective
guaranteeing is
the to ensure
water volumethe needed supplyfor
concentrated
based on the on
the minimizationthe
minimization cost of cost
cost[10-14].reliability
functions, using question
theapproach
Loss of of of
the the
throughout energy
irrigation. and
the day,The guaranteeing
scheme
while of
charging thethe water volume
proposed
the proposed
battery with needed
approach
the excees for
is
based
Load
Moreover, on
Probability (LLP)
some researchers of
concept functions,
have[10-14].
proposed using
This LLP
sizing the Loss
algorithms the
the irrigation.
irrigation.
presented in Figure The
The scheme
scheme
1 [16, 17]. The of
of the
the proposed
algorithm depends approach
approach
on: for is
is
Load
Load Probability (LLP) concept This LLP approach of the energy and guaranteeing the water volume needed
based Probability
on the minimization(LLP) concept of cost[10-14].
functions, This
usingLLPtheapproach
Loss of presented
presented in
in Figure
Figure 1
1 [16,
[16, 17].
17]. The
The
the irrigation. The scheme of the proposed approach is
algorithm
algorithm depends
depends on:
on:
Load Probability (LLP) concept [10-14]. This LLP approach 462 presented in Figure 1 [16, 17]. The algorithm depends on:
Copyright © 2015 IFAC
Copyright
Copyright ©© 2015
2015 IFAC
IFAC 462
462
2405-8963 © 2015, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.
Peer review©under
Copyright 2015 responsibility
IFAC of International Federation of Automatic
462Control.
10.1016/j.ifacol.2015.12.422
 IFAC CPES 2015
December 9-11, 2015. Delhi, India Imene Yahyaoui et al. / IFAC-PapersOnLine 48-30 (2015) 462–467 463

 the water volume needed, Step 5 Calculation of the water needed V [16, 17]:
 the site characteristics,  1  f i 1  L R  
 the panel characteristics. V   k c E T o  rm   1   (2)
 f i 1  L R  

The algorithm aims to find the panels surface S opt and where:
k c : crop growth coefficient for month M,
the battery capacity C bat opt that guarantee the installation
ET o : reference evapotranspiration average for month M,
autonomy when supplying the pump. Hence, the idea consists
rm : average rain volume for month M,
in searching the optimal components sizes that ensure the
balance between the charged and the extracted energies E c f i : leaching efficiency,

and E e , respectively. In fact, the battery bank supply the load L R : leaching fraction given in the soil.
by E bat when the panel does not generate the sufficient
Step 6 Calculation of the pumping duration [17]:
energy E load , and is charged with a part of the PV energy
V
produced E pv (Figure 2). The balance between the t  (3)
Q
accumulated and the extracted energies does not guarantee the where Q is the water flow (m²/h).
autonomy, due to the fluctuation in the solar radiation and the
energy losses. Thus, to ensure the autonomy and provide the Step 7 Calculation of the minimum panel surface S i and
energy demanded by the load, the algorithm is performed by the initial battery capacity C bat i using equations (4) and
adopting an efficiency coefficient  (slightly greater than 1). (5) respectively:
Hence, the energy balance can be expressed as follows: P pum p  t  d 
Ec   Ee (1) Si   1  aut 
2
W pv  bat  l  pv  reg  inv  opt ther  m atching  d rech 
The sizing algorithm is performed using two sub
algorithms during the vegetative cycle: the Algorithm 1 (4)
determines the sizes of the panel surface S M and the battery E d d aut
C bat  (5)
capacity C bat M for each month M. Then, Algorithm 2 is
i
V bat  dod max

performed to deduce the final sizes based on the sizes with:

determined for each month and the available components, Ppum p : pump power (W),
providing the numbers of panels and batteries needed. d aut : requested days of autonomy,
Algorithm 1 is detailed now following Figure 4. d rech : days needed to recharge the battery,
a) Algorithm 1: Determination at each month of the
minimum panel surface and battery bank capacity W pv : average daily radiation (Wh/ m 2 / day),
Step 1 Estimation of the diffused and direct radiation.  bat : electrical efficiency of the battery bank,
Step 2 Deduction of the hourly daily solar radiation
l : electrical efficiency of the rest of the installation
distribution H t  h , d  in a tilted panel [17].
(includes ohmic wiring and mismatching wiring losses),
Step 3 Estimation of the hourly cell temperature T c h 
 pv : efficiency of each photovoltaic panel,
[16].
 reg : regulator performance,
Step 4 Deduction of the hourly panel efficiency  pv  h 
[17].  inv : inverter performance,
Ensure the Charge
Protect thethe energy against
batteries Guarantee the
demanded by the load
energy balance deep discharge water needed  optther : panel performance - optical and thermal effects (%),
 matching : panel matching performance (%),
Objective
Water needed Ed : daily energy consumption (Wh),
Outputs

Panels surface S opt

Inputs

SIZING
Site characteristics
ALGORITHM
Panel characteristics Battery bank capacity nbat V bat : battery voltage (V),
opt

C bat
Constraints
 dod : maximum permitted variation of the depth of
max

Ec   E AM  EPM  dod  0.78 Vreservoir  V

discharge dod .
Ec   Ee max

Step 8 Calculation of P p vi corresponding to the

Fig. 1 Planning of the proposed sizing algorithm minimum panel surface S i , using (6) [17]:
P pv i   pv S i H t (6)

463
IFAC CPES 2015
464
December 9-11, 2015. Delhi, India Imene Yahyaoui et al. / IFAC-PapersOnLine 48-30 (2015) 462–467
E pv Eload
Load
E bat
E load
Photovoltaic
panels Battery
Fig. 2 Energy balance principle
Step 9 Calculation of the energies expected to be daily
stored and extracted from the battery E c and E e .
Step 10 If the extracted energy is higher than the stored
energy, the algorithm increases the panel surface by the
minimum increment of the PVP size commercially
available: the algorithm looks for the best configuration to
guarantee the balance between the demanded and the
produced energies, by equalizing the energies stored E c
and extracted energies E e in the battery bank (1).
Step 11 Battery capacity C bat M deduction for month M
[17]:
Ec
C bat  (7)
M
V bat
b) Algorithm 2: Calculation of the minimum panel surface and
battery bank capacity for the whole vegetative cycle
Using Algorithm 2 (Figure 3), the final values of the panel
surface S o p t and the battery bank capacity C bat opt , are
deduced. S o p t corresponds to the maximum value of the panel
surface obtained during the months. The final battery capacity
is the corresponding value for S o p t , since it is the most critical.
II.APPLICATION TO A CASE STUDY
The proposed algorithm is applied now to evaluate the
components sizes of a case study: the proposed algorithm is
tested during the months that correspond to the vegetative
cycle of tomatoes (March to July), using data of the target
area (Northern of Tunisia: latitude: 36.39°, longitude: 9.6°).
Following Algorithm 2, the Algorithm 1 was first
evaluated for all the months in the vegetative cycle: the solar
radiation accumulated on a tilted panel is evaluated; then, the
panel yield is calculated for each month; in parallel, the water
needed V is evaluated depending on the vegetative cycle and
the site [17]. Then, if the stored energy is higher than the
extracted energy, the surface is increased by the minimum
2
surface in the market (in our case, the increment is 0.5 m ).
Algorithm 1 results are summarized in Table 1, which
shows that the proposed strategy always ensures the water
needs, respects the limits on the battery-bank’ depth of
discharge and the energy balance (1). This has been tested
during the months of tomatoes vegetative cycle: the
efficiency coefficient  is around the fixed values throughout
all the considered months. For this value,  dod m ax is
guaranteed to be equal to 0.78.
464
For instance, in July, the minimum  is 1.46, and the
value obtained with Algorithm 1  1 is equal to 1.47. On the
other hand, in March, the generated photovoltaic power
during the morning supplies the pump together with the
battery bank during the pumping duration. After that, the
photovoltaic power generated charges the battery bank. The
quotient between the cumulated and extracted energies is
1.66, which is near to target value 1.7. We must point out that
for the energy balance, an error coefficient is used to consider
the clouds. Hence, in our study, we take into account the
possibility of having cloudy days. For example, in April the
loss of energy each day is 23.23 %. The obtained results
(Table 1) prove that the panels surface and battery bank
capacity obtained using the proposed Algorithm 1 satisfy the
energy balance. This is possible thanks to the calculation of
the battery capacity, which is done by considering the same
 dod m ax value that can be reached. Since July is the most
critical month for irrigation, the system components sizing of
July is selected. The obtained size allows the load to be
supplied during the requested pumping duration  t , and also
provides the energy E c needed to charge the battery bank.
III.VALIDATION USING PVSYST
The installation size has been also tested using PVSyst,
since the solar radiation, the ambient temperature and the load
requirements of the target cite can be manually choosed. This
tool allows determine and validate installations components
sizes. In addition, it takes into account varios losses related
with components or climatic parameters. Hence, PVSyst
evaluate the size efficicency using the solar fraction (SF),
which determines whether the panel surface is able to supply
the load with the needed energy. Moreover, it performs a
more detailed evaluation of the installation size: the system
losses (Ls), the unused energy (Lu) and the energy supplied
to the user (Yf). The PVSyst simulation shows that the
adopted size (S= 101.5 m 2 and C bat =1680 A.h) gives good
results. In fact, Figures 5 and 6 show that during the crops
vegetative cycle, the solar fraction (SF), which determine
whether the solar installation provides the load with the
sufficient energy, is pratically equal to one, except in June
and July, in which it is equal to 0.962 and 0.934, respectively
(Table 2). This leak of energy can be covered by considering
an additional water volume in the reservoir.
For M=1: M max , evaluate S M and C bat M
using Algorithm 1.
S M , C bat
M
S opt  max S 1 , S 2 ,.., S max 
C bat  C bat when S opt  S M
opt M
Fig. 3 Sizing Algorithm 2
 IFAC CPES 2015
December 9-11, 2015. Delhi, India Imene Yahyaoui et al. / IFAC-PapersOnLine 48-30 (2015) 462–467 465

Moreover, the obtained size is tested during the year The results show that the choosen size allows supplying
(Figure 7), by taking into account all the possible losses the load and having no more than 3.4 % of load losses, which
related to the components or climatic parameters. represents a good result.
Site characteristics

Step 1: Estimation of the diffused and the Step 5: Calculation of the needed water volume V
direct radiations H d and H b [17]. [16, 17].
Step 2: Deduction of the solar radiation in a Step 6: Calculation of the pumping duration  t
tilted panel H t [16, 17]. [16, 17].
Step 3: Estimation of the temperature Tc
[17].
Step 4: Deduction of the panel’ efficieny  p v
[16].
Ht  pv V t

Step 7: Calculation of the initial panel’ surface

Si and the initial battery bank capacity C bat i
S  Si
and initialize .
Step 8: Calculation of the photovoltaic power P pv i corresponding to the initial surface [17].

P pv i

Step 9: Calculation of the energies charged and extracted from the batteryE c bank and E e .

E c and E e

SM  S Yes No
Step 10: Ec  Ee

Yes No
Ec  Ee
Step 11 : Deduction of C bat M
(7) Increment S
Decrement S

Fig. 4 Sizing Algorithm 1 for each month M

10 1.2
Lu: Unused energy (full battery) 1.91 kWh/kWp/day PR: Performance Ratio (Yf/Yr):0.211
Normalized energy [kWh/kWp/day]

Lc: Collection loss (PV-array losses) 1.07 kWh/kWp/day SF: Solar Fraction (ESol/ELoad):0.966
Ls: system losses and battery charging 1.34 kWh/kWp/day
Yf: Energy supplied to the user 1.15 kWh/kWp/day 1.0
8
Performance Ratio PR

0.8
6
0.6

4 0.4

2 0.2

0
0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Fig. 6 Performance Ration and solar fraction using PVSyst
Fig. 5 Normalized production using PVSyst

465
IFAC CPES 2015
466
December 9-11, 2015. Delhi, India Imene Yahyaoui et al. / IFAC-PapersOnLine 48-30 (2015) 462–467

Loss diagram over the whole year

1808 kWh/m² Horizontal global irradiation

+10.5 % Global incident in coll. plane

-2.9 % IAM factor on global

1939 kWh/m² * 81 m² Cells Effective irradiancy on cells
Efficiency at STC= 15.80 % PV conversion
24.68 MWh Array nominal energy (at STC effic)
-3.5 % PV loss due to irradiance level

-10.9 % PV loss due to temperature

-1.2 % Module quality loss

-1.1 % Module array mismatch loss
-0.6 % Ohmic wiring loss
0.0 % Loss by respect to the MPP
running

-42.7 % Unused energy at the output

11.52 MWh
of the array
Effective energy at the output of the array
-5.5 % Converter loss during operation (efficiency)
0.0 % Converter loss due to power threshold
(efficiency)
0.0 % Converter loss over nominal conv. voltage
Missing energy 10.89 MWh
Stored
0.0 % Converter loss due to voltage threshold
3.4 % 70.6% +0.2 % Battery stored energy balance
0.2 MWh

Battery efficiency loss

Direct use 29.4%

-35.7 %

-23.8 % Battery efficiency loss

0.0 % Battery self-discharge current

5.33 MWh
5.51 MWh
Fig. 7 Loss diagram over the whole year
IV.CONCLUSION
A sizing algorithm to decide on the sizing of the
installation elements was presented and validated using
PVSyst tool. The algorithm is tested for a 10 ha land surface
in the northern of Tunisia. The sizing results ensures
supplying the pump during the pumping period, the energy
needed by the load and the needed water volume for crops
irrigation.
ACKNOWLEDGEMENTS
Miss Yahyaoui is funded by a grant from the Spanish
ministry of economy MICInn BES-2011-047807.
V.REFRENCES
[1] Kaldellis, J. K; Zafirakis, D; & Kondili, E. (2010). “Optimum sizing
of photovoltaic-energy storage systems for autonomous small islands”.
International Journal of Electrical Power & Energy Systems, 32(1),
24-36.
[2] Sidrach-de-Cardona, M; & Mora López, Llanos. (1998). “A simple
model for sizing stand-alone photovoltaic systems”.Solar Energy
Materials and Solar Cells,55(3), 199-214.
[3] Khatib, Tamer; Mohamed, Zah.A; & Sopian, K. (2013). “A review of
photovoltaic systems size optimization techniques”. Renewable and
Sustainable Energy Reviews, 22, 454-465.
[4] Jakhrani, Abdul Qayoom; Othman, Al-Khalid; Rigit, Andrew; Ragai.
Henry; Samo, Saleem. Raza; & Kamboh, Shakeel. Ahmed. (2012). “A
Energy supplied to the user
Energy need of the user
novel analytical model for optimal sizing of standalone photovoltaic
systems”. Energy, 46(1), 675-682.
[5] Acakpovi, Amevi; Xavier, Fifatin. Francois; & Awuah-Baffour,
Robert. (2012). “Analytical method of sizing photovoltaic water
pumping system”. In the proceedings of the 4th IEEE International
Conference on Adaptive Science & Technology, 65-69.
[6] Shrestha, G. B; & Goel, L. (1998). “A study on optimal sizing of
stand-alone photovoltaic stations”. IEEE Transactions on Energy
Conversion, 13(4), 373-378.
[7] Barra, Luciano; Catalanotti, Sergio; Fontana, F; & Lavorante, F.
(1984). “An analytical method to determine the optimal size of a
photovoltaic plant”. Solar Energy, 33(6), 509-514.
[8] Groumpos, P. P; & Papageorgiou, G. (1987). “An optimal sizing
method for stand-alone photovoltaic power systems”. Solar Energy,
38(5), 341-351.
[9] Mellit, A; Benghanem, M; Hadj Arab, A; & Guessoum, A. (2003).
“Modelling of sizing the photovoltaic system parameters using
artificial neural network”. In the proceedings of the IEEE Conference
on Control Applications, 353-357.
[10] Yang, Hongxing; Zhou, Wei; Lu, Lin; & Fang, Zhaohong. (2008).
“Optimal sizing method for stand-alone hybrid solar–wind system
with LPSP technology by using genetic algorithm”. Solar Energy,
82(4), 354-367.
[11] Khatib, Tamer; Mohamed, Azah; Sopian, K; & Mahmoud, M. (2012).
“A new approach for optimal sizing of standalone photovoltaic
systems”. International Journal of Photo Energy.
[12] Klein, S. A; & Beckman, W. A. (1987). “Loss-of-load probabilities for
stand-alone photovoltaic systems”. Solar Energy, 39(6), 499-512.

466
 IFAC CPES 2015
December 9-11, 2015. Delhi, India Imene Yahyaoui et al. / IFAC-PapersOnLine 48-30 (2015) 462–467 467

[13] Abouzahr, I; & Ramakumar, R. (1991). “Loss of power supply
probability of stand-alone photovoltaic systems: a closed form solution
approach”. IEEE Transactions on Energy Conversion, 6(1), 1-11.
[14] Maghraby, H. A. M; Shwehdi, M. H; & Al-Bassam, G. K. (2002).
“Probabilistic assessment of photovoltaic (PV) generation systems”.
IEEE Transactions on Power Systems, 17(1), 205-208.
[15] Mellit, A; Benghanem, M; & Kalogirou, S. A. (2007). “Modeling and
simulation of a stand-alone photovoltaic system using an adaptive
artificial neural network: Proposition for a new sizing procedure”.
Renewable Energy, 32(2), 285-313.
[16] Yahyaoui, Imene; Chaabene, Maher;& Tadeo, Fernando.(2013). “An
algorithm for sizing photovoltaic pumping systems for tomatoes
irrigation”. In the proceedings of the IEEE International Conference
on Renewable Energy Research and Applications (ICRERA), 1089-
1095.
[17] Yahyaoui, Imene; Ammous, Mahmoud; & Tadeo, Fernando. (2015).
“Algorithm for optimum sizing of a photovoltaic water pumping
system”.Int. Journal of Computer Applications (IJCA), 11(6), 21-28.
[18] Capizzi, Giacomo.; Bonanno, Francesco; & Tina, Giuseppe. Marco.
“Experiences on the Design of Stand-Alone Photovoltaic System by
Deterministic and Probabilistic Methods”. Proceedings of the
International Conference on Clean Electrical Power (ICCEP), 328–
335.
[19] Giuseppe. Tina; Conti, Stefania; & Ragusa, Carlo (2002). “Optimal
sizing procedure for stand-alone photovoltaic systems by Fuzzy
Logic” Journal of Solar Energy Engineering, 124, 77-82.
[20] PVsyst software. CUEPE, University of Geneva. www.pvsyst.com.

Table 1 Panel surface and battery capacity for each month M

Months March April May June July

Results
Cloud coverage Ac (%) 30.15 23.23 28.38 13.03 14.11

 erro r 1.30 1.23 1.28 1.13 1.14

SM m 
2
37.5 41.5 54.5 61.5 101.5

C bat (Ah) 840 1050 840 1050 1680

M

Ec
1  1.66 1.57 1.64 1.44 1.46
E e AM  E e PM

Table 2 Energy balance and main PVSyst results

Results GlobHor GlobEff E Avail E Unused E Miss E User E Load
2 2 Sol Frac
(kWh/ m ) (kWh/ m ) (MWh) (MWh) (MWh) (MWh) (MWh)
Months

January 78.0 117.8 1.221 0.630 0.000 0.000 0.000 1.000

February 89.1 116.0 1.185 0.608 0.000 0.000 0.000 1.000

March 140.0 161.3 1.587 0.959 0.000 0.419 0.419 1.000

April 164.1 161.8 1.708 1.089 0.000 0.540 0.540 1.000

May 208.1 183.6 2.055 0.855 0.000 1.116 1.116 1.000

June 225.0 187.0 1.966 0.466 0.056 1.429 1.485 0.962

July 237.0 201.8 1.973 0.099 0.129 1.824 1.953 0.934

August 208.0 196.7 1.891 0.968 0.000 0.000 0.000 1.000

September 166.0 184.2 1.767 0.905 0.000 0.000 0.000 1.000

October 128.0 166.7 1.661 0.851 0.000 0.000 0.000 1.000

November 89.9 138.6 1.412 0.724 0.000 0.000 0.000 1.000

December 75.1 123.8 1.266 0.649 0.000 0.000 0.000 1.000

Year 1808.3 1939.5 19.692 8.801 0.185 5.328 5.513 0.966

467