A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No.

1 (2014) 107–125

MODELLING AND SIMULATION OF STANDALONE

SOLAR POWER SYSTEMS
A. M. DIZQAH, A. MAHERI, K. BUSAWON & A. KAMJOO
Faculty of Engineering and Environment, Northumbria University, Newcastle, UK.

ABSTRACT
In the design of the controllers of hybrid renewable energy system (HRES), the system dynamics and
constraints need to be modelled and simulated in conjunction with the controller itself. This paper
presents mathematical and equivalent electrical models taking into consideration all system dynamics
and constraints for the solar branch of HRES. This branch consists of photovoltaic (PV) array, load
and battery connected through a boost-type DC–DC converter. The probabilistic behaviour of the solar
irradiance, which intrinsically includes the effect of cloud shading, and the dynamics of the battery are
also modelled. The platform developed for dynamic simulation of the solar branch of HRES can be
employed for design of DC–DC converter controllers as well as design of energy management systems.
Keywords: Battery, boost-type DC–DC converter, electrical load, energy management system, HRES,
hybrid renewable energy system, photovoltaic, PV.

1 INTRODUCTION
A system including photovoltaic (PV), converter and battery cannot be modelled with a sys-
tem of ordinary differential equations (ODEs). Due to considerable algebraic constraints, the
system can only be modelled with a system of differential algebraic equations (DAEs).
Moreover, in spite of using some indirect techniques, the current versions of available
simulators are not able to solve systems with DAEs directly and this is one of the major
challenges in simulating hybrid renewable energy system (HRES).
In order to design controllers and converters for a HRES, it is necessary to model the entire
of the system. Altas and Sharaf [1] proposed a simplified model for PV array in Simulink and
employed it for simulation of the PV at a specific operating point powering constant
alternating current and direct current (DC) loads. They used the linearization technique intro-
duced by Buresch [2] to take into account the effect of solar irradiation on cell temperature.
Villalva et al. [3] proposed two simulation scenarios, both based on equivalent electrical
circuits, to simulate a PV module. They also proposed an algorithm for PV identification to
find the parameters of equivalent electrical circuit of the PV module. The introduced param-
eter identification algorithm by Villalva is based on minimization of the norm of the difference
between the calculated power at maximum power point (MPP) and the value provided by the
manufacturer. Ishaque et al. [4] introduced a two-diode model for a PV array aiming at
increasing the accuracy of the model at low solar radiation. However, their model does not
take into account the thermodynamics of the PV panel and its effect on the cell temperature.
Guasch and Silvestre [5] proposed a comprehensive model for lead-acid batteries and an
equivalent electrical circuit for simulations.
This paper proposes a comprehensive model of the solar branch of HRES considering the
dynamics and constraints that arise due to connecting the components together. A platform
for simulation of the solar branch of HRES for the entire of one day is also presented. The
developed simulator extracts the required parameters for solar irradiance and PV module
modelling from meteorological data and the module datasheet, respectively.

© 2014 WIT Press, www.witpress.com

ISSN: 2046-0546 (paper format), ISSN: 2046-0554 (online), http://journals.witpress.com
DOI: 10.2495/CMEM-V2-N1-107-125
108 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

2 COMPONENTS MODELLING
Figure 1 illustrates the standalone solar power system topology selected for this study. It is
based on a parallel topology that has superior performance comparing with other topolo-
gies [6]. The PV array is connected to the DC bus, where the wind turbine (WT) can also
be connected, via an MPP tracker (MPPT) equipped converter. Without losing general-
ity, the battery bank can be connected to DC bus directly; however, it can also be
connected through a controller. While the PV array and the WT always generate power,
the electrical load consumes energy and the battery bank exhibits both producing and
absorbing behaviours.

2.1 Solar irradiance

In order to simulate a PV array the solar irradiance must be modelled. Karaki et al. [7] intro-
duced a stochastic model for hourly averaged global insolation absorbing by a PV array. They
modelled the distribution of global solar irradiance as a beta probability distribution function
as follows:
a −1 b −1
Γ (a + b ) ⎛ S x ⎞ ⎛ Sx ⎞
f (S x ) = ⎜⎝ 1 − S ⎟⎠
(1)
Γ (a )Γ ( b ) ⎜⎝ Smax ⎟⎠ max

where:
Γ the gamma function
a and b the shape parameters of beta distribution
Sx the current solar irradiance W/m2
Smax the extraterrestrial solar irradiance W/m2
[ f(Sx)] the power distribution function of the solar irradiance Sx

The solar irradiance for a specific location and time can be modelled if there are enough
sample data to find the exponents a and b [7–9].

IPV IPV-DC
+ +
DC-DC
VDC ILoad
VPV converter
- +
-
VLoad Load

IBat -
+
VBat
-

DC Bus
Figure 1: Standalone HRES with parallel topology.
 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014) 109

2.2 PV module and PV array

A PV cell is a P–N junction that can be modelled as an electrical circuit [10]. Normally,
manufacturers produce PV modules consisting of several PV cells in series. In practice, a PV
array which is a combination of several photovoltaic modules in series and parallel
arrangement is used to provide appropriate power for a site. A PV panel is a mechanical
framework encompasses the PV array. The structure of PV panel and the construction
materials influence the performance of PV array due to the thermodynamics effects.
Figure 2 shows the equivalent electrical circuit of a PV module based on a single-diode
model. This model offers a good compromise between accuracy and simplicity [3]. Iph and Id
are, respectively, the photocurrent generated by the module and the diode current. Rs and Rp
are the equivalent series and parallel resistance of the module. In case of having several PV
modules connected in series and parallel arrangement, there will be a stray capacitance which
can be modelled as capacitor C in the model of Fig. 2 [11].
Equations (2)–(5) used by Villalva et al. [3] model the voltage-current characteristics of a
PV module. Equation (2) models the I – V curve of PV module (Fig. 3a).

(2)

S x (t )
I ph (t ) = ( I ph, stc + K i ΔTc (t )) (3)
Sstc

R p + Rs
I ph, stc = I sc, stc (4)
Rp

I sc , stc + K i ΔTc (t )
I 0 (t ) = (5)
⎛V + K v ΔTc (t ) q ⎞
exp ⎜ oc, stc −1
⎝ nd N s KTc (t ) ⎟⎠

Figure 2: The single-diode equivalent electrical circuit of PV modules or PV arrays.

110 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

where VPV and IPV are, respectively, the voltage and current of the PV module and all other
symbols are defined as follows:
q The electron charge (1.602 × 10−19C)
nd Diode ideally factor (−)
K The Boltzmann constant (1.38 × 10−23 J/K)
Io Reverse saturation current of diode at cell temperature Tx (A)
Id The diode current (A)
Iph,stc Photocurrent at STC (A)
Iph Photocurrent at cell temperature Tx and solar irradiance Sx (A)
Isc,stc Short circuit current of PV at STC (A)
Voc,stc Open circuit voltage of PV at STC (V)
Ns Number of PV cells in series to build PV module (−)
Rs Series resistor of array (Ω)
Rp Parallel resistor of array (Ω)
Ki The factor of the slope of the change in photocurrent due to a change in the cell
temperature (−)
Kv The slope of the change in PV voltage due to a change in the solar irradiance (−)
Sstc The solar irradiance at STC W/m2
Sx Current amount of solar irradiance W/m2
Tc Current amount of cell temperature (K)
ΔTc The difference between current cell temperature and cell temperature at STC (K)

The strictly controlled test conditions (STCs) are a series of metrics in which the performance
of PV modules are measured. These test conditions are as follows:

• Cell temperature of 25°C

• Global solar irradiance of 1000 W/m2
• Air Mass of 1.5

I-V charactristic of PV module

9
Isc MPP 8
Impp
7 [22.649,7.545], Tc = 328.15
PV Output Current
PV Current (A)

6 [23.873,7.566], Tc = 318.15
[25.094,7.586], Tc = 308.15
5
[26.341,7.598], Tc = 298.15
4
[27.615,7.601], Tc = 288.15
3
Isc : Short-Circuit current
2
Voc : Open-Circuit voltage
MPP : Maximum Power Point 1
0
Vmpp Voc 0 5 10 15 20 25 30 35
PV Voltage (V) PV Output Voltage
(a) (b)

Figure 3: The I – V curve of the PV module for (a) eqn (2) and the MPP [3], (b) constant solar
irradiance and cell temperature in the range from 15°C to 55°C [12].
 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014) 111

Applying the Thevenin theory, a PV array can be modelled with the same equivalent electrical
circuit with different circuit parameters as follows [12]:

⎧ I 0,array = I 0 × N pvp
⎪
⎪ I ph ,array = I ph × N pvp
⎪n = nd × N pvs
⎪⎪ d ,array
⎨ N pvs (6)
⎪R p ,array = Rp ×
⎪ N pvp
⎪ N pvs
⎪ R s ,array = Rs ×
⎪⎩ N pvp

where Npvp and Npvs, respectively, are number of PV modules in parallel and series.
In practice the cell temperature of the PV array, Tc, is changing due to variation of ambient
temperature, Ta, and the insolation. That means that the PV array does not necessarily
operate under the STC. Fig. 3b illustrates the variation of I – V characteristic of a PV array
for a constant 1000 W/m2 solar irradiance and the cell temperature in the range of 15°C–55°C.
It can be seen that for the sample PV module the open-circuit and MPP voltages reduce by
about 18% when the cell temperature increases in this range. Alam and Alounai [13]
introduced eqn (7) to model the variation of cell temperature while neglecting the effect of
wind speed.

dTc (t ) 1
= ⎡(ta − h ) G(t ) − Ul (Tc (t ) − Ta (t ))⎤⎦ (7)
dt Ct ⎣

In this equation, h stands for the electric conversion efficiency of the PV cell, typically
around 15% for currently available cells. While h is an electrical characteristic, other
parameters are thermal parameters. The transmittance absorptance product of PV panel, ta,
is the ratio of absorbed to incident energy. Ct and Ut are respectively, the overall thermal
capacitance and the overall heat loss coefficient per unit of the PV panel in W/°Cm2

2.3 Lead-acid battery

Battery is an essential element in HRES to overcome the output power fluctuation caused by
uncertainty in renewable resources or sudden changes in load demand. Moreover, the most of
the MPPT algorithms which are switching-based DC–DC converters depend on the constant
voltage providing by connected battery to DC bus (Fig. 1). There are different types of
batteries for backup/storage systems in HRES. Comprehensive reviews of different battery
technologies and their applications in HRES can be found in Beaudin et al. [14] and Divya
and Ostergaard [15]. Guasch and Silvestre [5] modelled lead-acid batteries for different
working zones (Fig. 4) as in eqns (8)–(16).
112 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

Over- Saturation Zone

charging
Zone

Battery voltage Transition zone

Charging Zone

Discharging Zone

Over-
discharging
Zone

Exhaustion
Zone
0
0 Time

Figure 4: Working zones of a lead-acid battery.

dSOC (t ) h (t )I bat (t )
= − s SOC (t ) + c (8)
dt C (t i )

C (t ) =
C nominal Ctcoef
Bcap (1 + a ΔT (t ) + b ΔT (t ) )
c c
2
(9)
⎛ | I (t ) | ⎞
1 + A cap ⎜ bat ⎟⎠
⎝ I nominal

⎛ ⎞
⎜ acmt ⎟ C
hc (t ) = 1 − exp ⎜ ⎟ (SOC (t ) − 1) , I nominal = nominal (10)
I
⎜ bat (t )
+ bcmt ⎟⎟
n
⎜⎝ I ⎠
nominal

• Discharging, Over discharging and exhaustion zones: The battery voltage in all these
zones can be modelled with the same equation. For the exhaustion zone, the SOC must be
non-zero to avoid the singularity.

(11)

• Charging zone
I bat (t ) ⎛ P1c P3c ⎞
Vbat (t ) = (Vboc + K boc SOC (t )) + ⎜ + + P5c ⎟ (12)
C10 ⎝ 1 + I bat (t ) P2 C
(1 − SOC (t ))P4 C
⎠
× (1 − arc ΔT (t ))

• Overcharging zone: The battery is still charging in this zone in spite of the fact that it has
been fully charged before. Overcharging starts the electrolyzing process inside the battery
which leads to produce hydrogen gas that may cause explosion. The battery enters to this
 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014) 113

zone when Vbat ≥ Vg, where Vg is the gassing voltage of the battery and the voltage of the
battery increases up to the saturation voltage (Vec).
⎡ ⎛ C ⎞⎤
⎢ SOC (t ) nominal C (t ) − SOCV g (t )C (t ) ⎥
⎜ C ⎟
Vbat (t ) = V g (t ) + (V ec (t ) − V g (t )) × ⎢1 − exp ⎜ n
⎟⎥ (13)
⎢ ⎜ I bat (t )t(t ) ⎟⎥
⎢ ⎜⎝ ⎟⎠ ⎥
⎣ ⎦

⎡ ⎛ I (t ) ⎞ ⎤
V g (t ) = ⎢ A gas + B gas ln ⎜1 + bat ⎟ ⎥ 1 − agas ΔT (t )
⎝ C nominal ⎠ ⎦⎥
( ) (14)
⎣⎢

⎡ ⎛ I (t ) ⎞ ⎤
V ec (t ) = ⎢ A fonsc + B fonsc ln ⎜1 + bat ⎟ ⎥ (1 − afc ΔT (t )) (15)
⎢⎣ ⎝ C nominal ⎠ ⎥⎦

A tSC
t(t ) = (16)
1 + B tsc ( )
I bat (t ) C tsc
C nominal

where:
Cnominal The rated battery capacity at nominal discharging current (h)
Inominal The nominal discharge current corresponding to Cnominal (A)
DT The temperature variation from the reference value of 25°C (K)
n The time (h)
ac The temperature coefficient (K−1)
bc The temperature coefficient (K−2)
Ctcoeff ,Acap,Bcap ,acmt ,bcmt The model constants (−)
hc The charging efficiency (−)
I10 The nominal discharge current for the period of 10 hours (A)
P2dc ,P2c ,P5dc ,P5c The model constants (Vh)
P1dc ,P1c The model constants (VAh)
P2dc ,P2c ,P4dc ,P4c ,Ctsc The model constants (−)
Pbodc ,Kboc ,Vbodc ,Vboc
Agas ,Bgas ,Afonc ,Bfonc The model constants (V)
ardc , arc , agas , afc The temperature coefficient (K−1)
t The time factor (−)
Atsc , Btsc The model constants (h)
Vg The nominal gassing voltage of the battery (V)
Vec The saturation voltage of the battery (V)
SOCVg The SOC of the battery at the gassing voltage (−)
SOC The battery state of charge (–)
Ibat The battery current (A)
Vbat The battery voltage (V)
s The self-discharge rate (0.2% per day)
C The battery instantaneous capacity (Ah)
C10 The battery 10-hour rated capacity (Ah)
114 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

2.4 Boost-type DC–DC converter

The boost-type DC–DC converter is implemented based on the switching-mode circuit

technology containing at least one energy storage and two semiconductor switches. However,
in the ideal case a single-pole double-throw switch can be used [16]. Figure 5 illustrates the
electrical circuit for a boost-type DC–DC converter with an ideal switch.
Middledbrook and Cuk [17] proposed a state-space averaging approach to model the boost
type DC–DC converter. They suggested two states, Il and Vc for the continuous conduction model
in which the instantaneous inductor current is always greater than zero. According to the proposed
approach there is a state-space system for the converter at each state of the switch and the overall
state-space model of the converter is the weighted average of these two models. The weighting
factor is the duration of time that the circuit remains in each state. Equations (17) and (18) provide
the average model of the boost-type DC–DC converter using ideal switch and diode.
⎡ Rl Rc (1 − D) 1− D ⎤
⎢− L − R
−
R ⎥
⎢ L (1 + c ) L (1 + c ) ⎥ I (t ) ⎡ 1 ⎤
⎡ I&l (t ) ⎤ ⎢ R R ⎥ ⎡ l ⎤ + ⎢ ⎥ V (t )
⎢& ⎥=⎢ ⎥ ⎢⎣Vc (t )⎥⎦ ⎢ ⎥ in
L (17)
⎣Vc (t )⎦ ⎢ 1− D 1
− ⎥ ⎣0⎦
⎢ C (1 + Rc ) RC (1 + )
Rc ⎥
⎣⎢ R R ⎦⎥

⎡ ⎤
⎢ R (1 − D) 1 ⎥ ⎡ I l (t ) ⎤
Vout (t ) = ⎢ c ⎥⎢ ⎥ (18)
⎢ (1 + c ) 1 + Rc
R ⎥ ⎣Vc (t )⎦
⎢⎣ R R ⎥⎦

where D is the duty of time that switch SW is in L position and the inductor is charged, R is
the external electrical load, and all other parameters are according to Fig. 5.
While state-space averaging approach is simple to analyse and implement, it does not model
the hybrid nature of the converter. A one-stage DC–DC converter (Fig. 5) works in two differ-
ent modes of the operation with respect to the value of the discrete state Sd [18]. For instance,
when this state is on, which means that its value is one, the switch SW is in the L position.
Defining the same state vector as the averaging approach, which is XT = [Il Vc]T , the boost-type
DC–DC converter can be modelled as the following hybrid system. It presents an affine state
space model coupled with a linear output equation for each mode of the operation [18]:

S
Figure 5: Electrical circuit of boost-type DC–DC converter with ideal switch.
 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014) 115

⎧ F X (t ) + f1Vin (t ); Sd = 1
X& (t ) = ⎨ 1 (19)
⎩ F2 X (t ) + f2Vin (t ); Sd = 0

⎧ gT X (t ); Sd = 1
Vout (t ) = ⎨ T1 (20)
⎩ g2 X (t ); Sd = 0

⎡ Rl Rc 1 ⎤
⎡ Rl ⎤ ⎢− L − R
−
Rc ⎥
⎢− L 0 ⎥ ⎢ 1+ c L (1 + ) ⎥ ⎡1⎤
F1 = ⎢ ⎥ , F2 = ⎢ R R ⎥, f = f = ⎢ ⎥
L (21)
⎢ 0 1 ⎥ ⎢ 1 1 ⎥ 1 2
⎢ ⎥
⎢ − ⎢ − ⎥ ⎣0⎦
⎣ C (Rc + R ) ⎥⎦ ⎢ C (1 + Rc ) Rc ⎥
RC (1 + )
⎢⎣ R R ⎥⎦

T T
⎡ ⎤ ⎡ ⎤
⎢ 1 ⎥ ⎢ Rc 1 ⎥
g1 = ⎢0 ⎥ , g2 = ⎢ ⎥ (22)
⎢ 1 + Rc ⎥ ⎢1 + Rc R
1+ c ⎥
⎣⎢ R ⎦⎥ ⎢⎣ R R ⎦⎥

2.5 Load profile

An electrical load profile indicates the hourly averaged load demand by a consumer over a
period of time. The hourly value and the shape of load demand depend on the consumer, hour
of the day and season. For domestic buildings separate graphs illustrate the load profile for
different seasons and also for weekdays as well as weekends. However, for other types of
consumers such as farms there may be just one load profile for all days of year. Figure 6
illustrates the load profile for a farm in the UK [19] used in this study with the assumption
that there is only a DC load.

Load profile of a farm

4000

3000
Load (W)

2000

1000

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour of the day

Figure 6: Load profile [19].

116 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

3 SIMULATION
Different components of the standalone solar site can be simulated with general blocks and
equivalent electrical circuits. The first level of the simulation model is given in Fig. 7. As it
can be seen there are five main blocks in this model. While SolarIrr simulates solar irradiance,
PVArray simulates an array of PV modules. In addition, the block with the name of Boost-Type
Converter provides the simulation of DC–DC converter. Battery and Load blocks, respectively,
simulate a stack of lead-acid battery cells and the load profile.

3.1 Solar irradiance simulation

The simulation proposed in this paper for solar irradiance is based on the hourly stochastic
model of eqn (1). Although, there are inaccuracies in fitting solar irradiance into beta
distribution still there are two advantages the simulation can benefit from stochastic modelling.
The main advantage of stochastic modelling of solar irradiance is its simplicity for
simulation. Once the shape parameters are learned for all hours of the year, the value of
insolation can be generated easily by using random generators. Another advantage
of generating random values for solar irradiance is that uncertainties due to cloud shading or
incident angle are intrinsically taken into account and the generated values of the solar
irradiance are directly fed into the PV cell.
In calculation of the shape parameters for hourly solar irradiance, the meteorological
sample data available from British Atmospheric Data Centre [20] between 1990 and 2010
have been used. The calculated shape parameters load into lookup tables and used to generate
random values for hourly solar irradiance (Fig. 8).
There are two challenges in simulating the insolation coupled with other components.
First, it causes the simulation to be multi-rate. While the time step for all components is small
enough to simulate the system dynamics, the simulation rate of the solar irradiance is one
hour, which is significantly higher. In this study, this challenge has been overcome by feeding
a separate clock to SolarIrr block (Fig. 7). Second, singularity can occur during simulation
by hourly stepwise solar irradiance. An interpolator has been employed to solve this problem
with linearizing the transition of solar irradiance between hours.

3.2 PV array simulation

A modified version of the equivalent electrical circuit proposed by Villalva et al. [3] is used
for simulating PV array. Figure 9 illustrates this model including a current source controlled
with eqn (23), where Im is the sum of the photocurrent, Iph, and the diode current, Id. The
proposed circuit is modified by adding a new block to simulate the PV cell temperature, Tc.
The PV cell temperature has been simulated using the state-space block and its output is then
fed into the PVArray block.

⎡ ⎛ Vpv (t ) + Rs I pv (t ) q ⎞ ⎤
I m (t ) = I ph (t ) − I 0 (t ) ⎢exp ⎜ − 1⎥ (23)
⎣⎢ ⎝ nd N s KTc (t ) ⎟⎠ ⎥⎦
 Vpv
Goto1
Icell
SolIrr
Goto4 C o n tin u o us Initialize M odel
From 2 Id e a l Switch
Iph
p o we rg u i Initializer Display
Goto5

Ki
Ta Goto6
From 1 Duty
Kv 0.0 Cycle
Goto7
Tc
Goto8 Goto11 Vout
Vd duty iL Ipv Vout v +-
[0] Goto9 Goto2
Ipv
Vin vC
From 3 Vpv Iout
IC
GN D GN D i IvpGoto10
Vout +
-

Boost-T ype converter Iout

PVArray
Iout1
i
–+

Current Filter1 Goto3

SolarIrr Iload
1 Location#(1..3)
Location Num ber
8
Bus
Load
GND

Clock

6 ExternalM onth Sx SolIrr

#(1..12) Goto
Month Variable
7 Battery 1
Load
-K- u+b Ta
Clock
Bias Goto12
Clock(sec) Gain
Bus
Load

GND

SolarIrr
A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

Clock(sec)1

Figure 7: First level of the simulation model.

117
118 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

Minute

MinuteTrigger

MinuteCounter

alpha alpha Minute

1 alpha disturbed_alpha
Location
Location# beta betta disturbed_betta beta
Sx
Sx Sx_k double 1
(1..3)
rmax disturbed_rmax
Sx
rmax rmax Sx_k-1
MonthNumber Data Type
Calcultion Disturbance Interpolator Conversion
muDist muDist
disturbance
HourNumber
sigmaDist sigmaDist

Scaling_Shaping SolarGenrator
month Factors Calculator

monthTrigger

2 MonthCounter Sx_kSx_k-1

Clock
Storage
hour

hourTrigger

HourCounter

Figure 8: Solar irradiance generator module.

8
Control
Logical
4 Operator
ScST C Divide1
3
Sc_STC
T aST C Iph 3
5 Sx Iph_signal
Sx Ta_STC Tc_Coef f _Iph 5
7 Tc TcCoeff_I_Signal
Tc Iph_STC
SolarIrr_Coef f _Iph 4
1 SolIrrCoeff_I_Signal
IphST C IphCalculator (eqn. 4)
g
Vpv_simport
2 1
1
Rs sw
+

iph Iph
iph Im
i0 I0 i0 Rp
s
-

NaNCheck ipv + v
ipv Ipv
NpvpEffect - 1
v pv
Ta_STC I0 Im Calculator Vpv
v pv Vpv Im VM
a
rs
Tc a
NaN Check
nser
Vpv_gnd
Rs npar
2
2 Ns Tc_Coef f _Voc Rs1
NS Im (eqn. 23)
I0Calculator (eqn.7)
10
Nser 6
NpvpEffect1 TcCoeff_V_Signal
6
Ipv uu
Saturation 2
Icell_signal

9
NP

Figure 9: The electrical equivalent circuit of PV array module.

The PV identification method proposed by Villalva et al. [3] with the following modifications
is adopted in this work.

• The norm of error has been changed to the difference of both the voltage and current at the
MPP with the values provided by the manufacturers.
• The step-size of RS declines dynamically due to decreasing the norm of error.
• A nonlinear transform has been applied to RS improving the speed of MPP search
algorithm.
 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014) 119

Although the PV array as a component can be simulated with its equivalent electrical
circuit, there is still a challenge in its connection to DC–DC converter. Since the PV array is
an instantaneous system [eqn (2)], its connection to the converter casts an algebraic constraint
on the converter which is not solvable with ordinary ODE integrators. It requires employing
a DAE integrator to simulate such a system directly. In the present study, however, a
combination of an ordinary integrator coupled with a Newton type method, which solves the
algebraic constraints, have been employed. It requires using a small step-size for the
simulation (i.e. 10−8 sec.) and setting manually a consistent initial value.

3.3 DC–DC converter simulation

Figure 10 shows the simulation circuit of the boost-type DC–DC converter including pulse
width modulation (PWM) block and ideal switches. Evidently, the simulation of the
converter has not been simplified with averaging approach [eqn (18)], instead it has been
modelled according to eqn (20) as a hybrid system. The transistors are replaced with two
ideal switches for simplicity’s sake. However, to prevent cases in which both switches are
opened causing the inductor to be disconnected, a snubber resistor has been added to Ideal
Switch1.

3.4 Battery simulation

The lead-acid battery is modelled with its equivalent electrical circuit as shown in Fig. 11.
The voltage source is controlled by the block called Dynamics according to eqns (9) and (12)
through (14). For simplicity, the self-discharge rate approximated as the constant loss equals
to 0.2% of its charge per day and the temperature effect is ignored.
As in the PV array, connecting the battery to the DC–DC converter introduces certain algebraic
constraints [eqns (11)–(13)] leading to a DAE-based system. In this case, the system can be
simulated by reducing the step-size of simulation and adding a slight R-L element (Fig. 7).

3.5 Load profile simulation

The DC load has been modelled as an electrical circuit. It consists of an optional constant
RLC load coupled with a variable load. In Fig. 12, it can be seen that the load profile has been

1 Sig na l(s ) P uls e s

duty
PWM Generator

1
iL

g
1 + -i 2 3
1
Vin Vout
iL_meas Ideal Switch Diode
L RL

Rc
1
g

+
- v
Ideal Switch1
vout_meas
2

+
- v
vin_meas

+ 2
C - v
vC
vC _meas

2
GND

Figure 10: Electrical circuit of boost-type DC–DC converter module.

120 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

V Current
current
Isd

Dynam i cs (Eqns. 9, 12, 13, 14)

i LoadBus
+ Current
- 2
Rg

-
s
Vbat Self_Disch

+
s

1
GND

Figure 11: Equivalent electrical circuit for lead-acid battery module.

2
LoadBus

RL RL
DC DC
Load2 Load3 RL RL
DC DC
LoadProfile Load Load1

1 Clock Load double

Clock
Data T ype Conversion
-
s

M Variable Load
+

1
GND

Figure 12: DC load module.

modelled using a current source controlled by the profile values, which are loaded into a
lookup table. A series inductor filter has been added to the model to suppress the sudden
changes of variable load.

4 SIMULATION RESULTS
MATLAB/SIMULINK environment was used to simulate the model given in Fig. 7 for a
24-hour period. The built-in algebraic loop solver and the SimPowerSystems toolbox in
SIMULINK can integrate the model of the solar branch of HRES. However, it also requires
applying small step-size of simulation and an initial condition block across the loop. Table 1
lists the simulation parameters including the day of the year, PV array configuration including
the PV module name and number of modules in parallel (Npvp), duty cycle of the converter,
battery parameters, and the type of the load profile. The model includes no DC–DC converter
controller as it is a generic simulation platform for any controller. The effect of this controller
in changing the duty cycle can be simulated manually.
Figure 13a and b shows the I – V and P – V curves of the sample PV module for different
values of solar irradiance and cell temperature. The MPP locations are indicated as a star on
both figures and vary due to changes in either the solar irradiance or the cell temperature.
Figure 14 shows the simulated insolation for the first of January and the first of October. It
also indicates variation in cell temperature of the PV array due to changes of solar irradiance.
In Fig. 14a the sudden changes in solar irradiance caused by natural obstacles such as clouds
have been indicated.
 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014) 121

Table 1: Simulation parameters.

Parameter Value
Date 1st June
PV model KC200GT
Npvp 10
D 50%
Battery Rated Capacity 550 Ah
Battery Nominal Voltage 48 VDC
Load profile Farm
Sampling time 1 × 10–8 sec
PWM Frequency 100 × 103 Hz

12
250 →[27.604,252.058]
W = 1200, Tc = 288.15 - 328.15

10 W = 1200, Tc = 288.15 - 328.15

→[27.615,209.901]
W = 1000, Tc = 288.15 - 328.15
[22.702,9.055]Æ ¨ [27.604,9.131] 200
[22.702,205.560]↑
W = 1000, Tc = 288.15 - 328.15
8
[22.649,7.545]Æ ¨ [27.615,7.601] →[27.535,167.256]
PV Output Current

PV Output Power

W = 800, Tc = 288.15 - 328.15

[22.649,170.897]↑
W = 800, Tc = 288.15 - 328.15 150

6 [22.527,6.030]Æ ¨ [27.535,6.074]
[22.527,135.831]↑ →[27.335,124.261]
W = 600, Tc = 288.15 - 328.15
W = 600, Tc = 288.15 - 328.15
[22.277,4.512]Æ ¨ [27.335,4.546] 100
4 [22.277,100.522]↑ →[26.955,81.160]
W = 400, Tc = 288.15 - 328.15
¨ [26.955,3.011] W = 400, Tc = 288.15 - 328.15
[21.783,2.995]Æ
50 [21.783,65.239]↑
2

0 0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
PV Output Voltage PV Output Voltage

Figure 13: (a) I – V and (b) P – V curves of Kyocera KC200GT PV module generated
by the simulator for solar irradiance and cell temperature in the ranges of
400–1200 W/m2 and 15°C to 55°C, respectively.

Figures 15 and 16 show the simulation results for the whole system. Figure 15a and b
shows the step-response of the output current and power of the DC–DC converter for differ-
ent values of switching duty-cycle. In Fig. 15a and b, it can be seen that the generated energy
of the PV array changes with respect to the DC–DC converter switching duty-cycle. The
small boxes inside the Fig. 15a and b illustrates the outputs fast variation as a result of the
DC–DC converter high frequency switching. The open-loop system, which is the plant with
no controller that manages the duty-cycle, can be simulated for the case that both the solar
irradiance and load demand are variable (Fig. 15c and d). It can be seen from Fig. 15c and d
that having a constant duty-cycle (i.e. D = 0.5 ), the generated power does not follow the trend
of the solar irradiation that has a peak in the middle of the day. The duty-cycle is fixed and
limits the output power. However, the impedance seen by the PV array changes due to
variation of the load and it may be nearer to the PV module internal impedance. Consequently,
the generated power follows the three peak demands of Fig. 6. In order to be able to
differentiate the effects of variable insolation and load demand, the simulation results are
122 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

Solar Irradiance (W/m2)

Cell Temperature (K)

315
250 298 310
200 296
294 305
150
100 0.5 11.5 22.5 300
50 -6 295
x 10
0 290
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour)
(a)
Solar Irradiance (W/m2)

Cell Temperature (K)

1000
800 298 340
296 330
600 294
320
400 0.5 11.5 22.5 310
200 -6 300
x 10
0 290
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour)
(b)

Figure 14: The simulated insolation and the cell temperature for (a) 1st of January, and (b) 1st
of October.

3.5 180

160

3 140

120
2.5
Power (W)
Current (A)

100
2.9
80 2.8
2 2.8
2.7
2.7 60
2.6
2.6 40
1.5 2.5
2.5
20
0.0862 0.0863 0.0863
0.08620.08630.0863
1 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Time (sec.) Time (sec.)
(a) (b)

12 350
Output Current Output power
PV current 300 PV power
10

250
8
Current (A)

Power (W)

200
6
150

4
100

2
50

0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour) Time (hour)
(c) (d)

Figure 15: (a) The step-response of output current and (b) power of the DC–DC converter for
constant load and insolation with D = 0.45, 0.5, 0.55, 0.6. The output (c) current
and (d) power of the PV and the DC–DC converter for simultaneous variable load
and insolation with D = 0.5
 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014) 123
55 12 350
50 Output Current Output power
10 PV current 300 PV power
45

40 250
8
35

C u rre n t (A )

P o w e r (W )
200
V oltage (V )

30
6
25 150
20 4
100
15

10 Output voltage 2 50
PV voltage
5
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour) Time (hour) Time (hour)
(a) (b) (c)

55 12 350
Output Current Output power
50
PV current 300 PV power
10
45

40 250
8
35
C urrent (A ) 200

P ow er (W )
V oltage (V )

30
6
25 150
20 4
100
15

10 2
Output voltage 50
5 PV voltage

0 0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour) Time (hour) Time (hour)
(d) (e) (f)

5 250 81

80.5
4 200
80

3 150 79.5

79
100
P ow er (W )

S O C (% )
Current (A )

2
78.5
1 50
78

0 77.5
0
77
-1 -50
76.5

-2 -100 76
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour) Time (hour) Time (hour)
(g) (h) (i)

30
800 80
Battery Current
Load Current
25 700
75
600
20
500 70
C urrent (A )

P ow er (W )

S O C (% )

15 400
65
300
10
200 60

5 100
55
0
0
-100 50
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour) Time (hour) Time (hour)
(j) (k) (l)

Figure 16: (a) Output voltage, (b) output current, and (c) output power of the PV and the
DC–DC converter for constant load with variable insolation. (d) Output voltage,
(e) output current, and (f) output power of the PV and the DC–DC converter for
variable load with constant insolation. (g) Current, (h) power, and (i) SOC of the
battery for constant load with variable insolation. (j) Current, (k) power, and (l)
SOC of the battery for variable load with variable insolation.

organized into two separate cases: (i) a constant load coupled with varying insolation, and (ii)
a constant insolation with variable load.
Figure 16a through c, respectively, illustrates the output voltage, current and power of the
PV array and the DC–DC converter for constant load and variable insolation. It can be seen
that although the solar irradiation changes, the generated power (and current) is almost con-
stant during the daylight as a result of constant duty-cycle. The duty-cycle is fixed to a value,
which is proper for the low insolation, and it limits the output power. While the PV voltage
124 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014)

changes during the day, the output voltage of the DC–DC converter, which is connected
directly to the battery terminal, is almost constant (Fig. 16a). Moreover, there is around 20%
difference between the PV and the DC–DC output powers which is a power loss due to the
internal resistor of the solenoid, Rl (Fig. 5). It is assumed that there is excess of energy dur-
ing the daylight that is stored into the connected battery. Figure 16g through l shows the
current, power, and the state of charge (SOC) of the battery. The direction of the battery
current in Fig. 16g shows that the battery is charged and the battery SOC increases during
the daylight.
On the other hand, Fig. 16d through f, respectively, depicts the system variables for the
constant insolation and variable load. Although the solar irradiation is constant, any load
demand fluctuation leads to the change in the input impedance seen by the PV module (Fig. 1)
when the duty-cycle is constant. Therefore, the generated power by the PV array slightly
changes. Figure 16g through l shows the current, power, and the SOC of the battery. Unlike
the previous case, this scenario introduces deficit of energy that is supplied by the battery.
The direction of the battery current and power in Fig. 16j and k, respectively, indicates that
battery is in discharging mode. Moreover, Fig. 16l shows that the battery SOC continuously
declines with different rates corresponding to the amount of the supplied energy.

5 CONCLUSION
To design different controllers of DC–DC converter as well as energy management system
(EMS) for HRES, modelling and simulating of HRES as the plant are essential. The model
needs to include not only the system dynamics of each component but also the algebraic con-
straints rising from connecting these components all together. The simulation must be stable
for at least 24 hours to be able to cover the variation of solar irradiance and load demands.
This paper demonstrates a model for the solar branch of HRES including different system
dynamics and algebraic constraints. It also proposes a flexible platform to simulate standalone
solar power systems which is capable to propose simultaneous accuracy in both electronic
components and coarse time-scale phenomena like cloud shading, load demand, and battery
behaviour. It simulates the complete system with constant and variable loads for 24 hours or
more which is used as the platform to study different controllers as well as EMS of HRES.
The results of the simulation with different scenarios including variable and constant solar
irradiance or electrical load besides operational characteristic curves have also been presented.

ACKNOWLEDGEMENT
This research was partially funded by Synchron Technology Ltd.

REFERENCES
[1] Altas, I.H. & Sharaf, A.M., A photovoltaic array simulation model for Matlab-Simulink
GUI environment. In Proc. IEEE Int. Conf. Clean Elect. Power (ICCEP) IEEE,
pp. 341–345, 2007.
[2] Buresch, M., Photovoltaic Energy Systems Design and Installation, McGraw-Hill: New
York, 1983.
[3] Villalva, M.G., Gazoli, J.R. & Filho, E.R., Comprehensive approach to modelling
and simulation of photovoltaic arrays. IEEE Transaction on Power Electronics, 24,
pp. 1198–1208, 2009. doi: http://dx.doi.org/10.1109/TPEL.2009.2013862
[4] Ishaque, K., Salam, Z. & Taheri, H., Accurate MATLAB simulink PV system simulator
based on a two-diode model. Journal Power Electronics, 11, pp. 179–187, 2011. doi:
http://dx.doi.org/10.6113/JPE.2011.11.2.179
 A. M. Dizqah, et al., Int. J. Comp. Meth. and Exp. Meas., Vol. 2, No. 1 (2014) 125

[5] Guasch, D. & Silvestre, S., Dynamic battery model for photovoltaic applications.
Journal Progress in Photovoltaics: Research and Applications, 11, pp. 193–206, 2003.
doi: http://dx.doi.org/10.1002/pip.480
[6] Ashari, M. & Nayar, C.V., An optimum dispatch strategy using set points for a
photovoltaic-diesel-battery hybrid power system. Journal Solar Energy, 66, pp. 1–9,
1999. doi: http://dx.doi.org/10.1016/S0038-092X(99)00016-X
[7] Karaki, S.H., Chedid, R.B. & Ramadan, R., Probabilistic performance assessment
of autonomous solar-wind energy conversion systems. IEEE Transaction on Energy
Conversion, 14(3), pp. 766–772, 1999. doi: http://dx.doi.org/10.1109/60.790949
[8] Sulaiman, M.Y., Oo, W.M.H., Wahab, M.A. & Zakaria, A., Application of beta
distribution model to Malaysian sunshine data. Journal Renewable Energy, 18,
pp. 573–579, 1999. doi: http://dx.doi.org/10.1016/S0960-1481(99)00002-6
[9] Dizqah, A.M., Maheri, A. & Busawon, K., An assessment of solar irradiance stochastic
model for the UK. In 2nd International Symposium on Environment Friendly Energies
and Applications (EFEA) Newcastle Upon Tyne, 2012.
[10] Deshmukh, M.K. & Deshmukh, S.S., Modelling of hybrid renewable energy systems.
Journal Renewable & Sustainable Energy Reviews, 12, pp. 235–249, 2008. doi: http://
dx.doi.org/10.1016/j.rser.2006.07.011
[11] Martinez, J., A state space model for the dynamic operation representation of small-scale
wind-photovoltaic hybrid systems. Journal Renewable Energy, 35, pp. 1159–1168,
2010. doi: http://dx.doi.org/10.1016/j.renene.2009.11.039
[12] Chatterjee, A., Keyhani, A. & Kapoor, D., Identification of photovoltaics source models.
IEEE Transactions on Energy Conversion, 26(3), pp. 883–889, 2011. doi: http://dx.doi.
org/10.1109/TEC.2011.2159268
[13] Alam, M.S. & Alounai, A.T., Dynamic modeling of photovoltaic module for real-time
maximum power tracking. Journal of Renewable and Sustainable Energy, 2, pp. 1–16,
2010. doi: http://dx.doi.org/10.1063/1.3435338
[14] Beaudin, M., Zareipour, H., Schellenberglabe, A. & Rosehart, W., Energy storage for
mitigating the variability of renewable electricity sources: an updated review. Jour-
nal Energy for Sustainable Development, 14, pp. 302–314, 2010. doi: http://dx.doi.
org/10.1016/j.esd.2010.09.007
[15] Divya, K.C. & Ostergaard, J., Battery energy storage technology for power systems –
an overview. Journal Electric Power Systems Research, 79, pp. 511–520, 2009. doi:
http://dx.doi.org/10.1016/j.epsr.2008.09.017
[16] Su, J.H., Chen, J.J. & Wu, D.S., Learning feedback controller design of switching
converters via MATLAB/SIMULINK. IEEE Trans on Education, 45, pp. 307–315,
2002. doi: http://dx.doi.org/10.1109/TE.2002.803403
[17] Middledbrook, R.D. & Cuk, S., A general unified approach to modelling switching-
converter power stages. In Proc. of IEEE Power Electronics Specialist Conference, 1976.
[18] Mariethoz, S., Almer, S., Baja, M., Beccuti, A.G., Patino, D., et al., Comparison of
hybrid control techniques for buck and boost DC–DC converters. IEEE Trans on
Control Systems Technology, 18, pp. 1126–1145, 2010. doi: http://dx.doi.org/10.1109/
TCST.2009.2035306
[19] Lopez, R.D., Agustin, J.L.B., Uson, L.C. & Aguarta, I.A., Demand side management
in hybrid systems with hydrogen storage in several demand. In 17th World Hydrogen
Energy Conference Brisbane, 2008.
[20] BADC, British Atmosphere Data Centre (BADC), 2011.