Available online at www.sciencedirect.

com

Solar Energy 85 (2011) 22172227

www.elsevier.com/locate/solener

A comprehensive MATLAB Simulink PV system simulator

with partial shading capability based on two-diode model
Kashif Ishaque a, Zainal Salam a,, Syafaruddin b
a
Faculty of Electrical Engineering, Universiti Teknologi Malaysia, UTM 81310, Skudai, Johor Bahru, Malaysia
b
Kumamoto University, 2-39-1 Kurokami, Kumamoto 860-8555, Japan

Received 4 January 2011; received in revised form 24 May 2011; accepted 6 June 2011
Available online 5 July 2011

Communicated by: Associate Editor Igor Tyukhov

Abstract

This paper proposes a comprehensive MATLAB Simulink simulator for photovoltaic (PV) system. The simulator utilizes a new two-
diode model to represent the PV cell. This model is known to have better accuracy at low irradiance level that allows for a more accurate
prediction of PV system performance during partial shading condition. To reduce computational time, only four parameters are
extracted for the model. The values of Rp and Rs are computed by an ecient iteration method. Furthermore, all the inputs to the sim-
ulators are information available on standard PV module datasheet. The simulator supports a large array combination that can be inter-
faced to MPPT algorithms and power electronic converters. The accurateness of the simulator is veried by applying the model to ve PV
modules of dierent types (multi-crystalline, mono-crystalline, and thin-lm) from various manufacturers. It is envisaged that the pro-
posed work can be very useful for PV professionals who require simple, fast, and accurate PV simulator to design their systems. The
developed simulator is freely available for download.
2011 Elsevier Ltd. All rights reserved.

Keywords: PV module; Partial shading; Bypass diode; Blocking diode; Simulator; MATLAB Simulink

1. Introduction (Chaouachi et al., 2010; Enrique et al., 2010; Messai

et al., 2011). However, for a proper design of MPPT con-
Large and small scale PV power systems have been com- troller, an accurate simulation model of the module is
mercialized due to its potential long term benets. Their required. This is especially the case when the peak power
growth rates have been accelerated by the generous fed-in point changes continuously due to environmental varia-
tari schemes and other initiatives provided by various tions. In particular, one important situation to be consid-
governments to promote sustainable green energy. In large ered is the substantial drop of power yield during partial
PV power generation, systems are dominated by grid- shading condition.
connected; examples can be seen in (Albuquerque et al., The most important component that aects the accuracy
2010; Beser et al., 2010; Chen et al., 2010; Mellit and of simulation is the PV cell modeling. For improved
Pavan, 2010; Yoon et al., 2011). To ensure optimal use accuracy, the two-diode model (with Rp and Rs) has been
of the available solar energy, maximum power point track- proposed by several researchers (Hovinen, 1994; Gow
ing (MPPT) scheme is applied to the power converters and Manning, 1999; Hyvarinen and Karila, 2003; Nishioka
et al., 2003, 2007; Kurobe and Matsunami, 2005;
Corresponding author. Tel.: +60 7 5536187; fax: +60 7 5566272. Chowdhury et al., 2007). Although more to that of
E-mail addresses: [email protected] (K. Ishaque), single diode model, the inclusion of an additional diode
[email protected] (Z. Salam). increases the number of computed parameters. Several

0038-092X/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2011.06.008
2218 K. Ishaque et al. / Solar Energy 85 (2011) 22172227

computational methods are proposed (Hovinen, 1994; values of Rp and Rs are estimated by an ecient iteration
Gow and Manning, 1999; Chowdhury et al., 2007), but method. The accurateness of the model is compared with
in all of these techniques, new coecients are introduced two other modeling methods namely the Rs- and Rp-model.
into the equations, increasing the computing burdens. An In addition, the proposed work supports large array simu-
alternative approach to describe the two-diode model is lation that can be interfaced with MPPT algorithms and
by investigating its physical characteristics such as the elec- actual power electronic converters. This allows for the
tron diusion coecient, minority carriers lifetime, intrin- designer to evaluate the overall system performance when
sic carrier density, and other semiconductor parameters interacting with other components within the systems. It
(Hyvarinen and Karila, 2003; Nishioka et al., 2003, 2007; is envisaged that the proposed work can be very useful
Kurobe and Matsunami, 2005). While these models are for PV professionals who require fast and accurate PV sim-
useful to understand the behavior of the cell, information ulator to design their system.
about these parameters is not always available in commer-
cial PV datasheets. 2. PV module modeling
A comprehensive PV system simulator should fulll the
following criteria: (1) it should be fast, yet can accurately 2.1. Modeling using the two-diode model
predict the IV and PV characteristic curves, including
partial shading condition. (2) It should be a exible tool The two diode model is depicted in Fig. 1. The output
to develop and validate the PV system design, inclusive current of the module can be described as (Chih-Tang
of the power converter and MPPT control. Although exist- et al., 1957)
ing software packages like P-Spice, PV-DesignPro, Solar-    
Pro, PVcad, and PVsyst are available in the market, they V IRS
I I PV  I o1 exp 1
are relatively expensive, unnecessarily complex and rarely a1 V T 1
     
support the interfacing of the PV arrays with power con- V IRs V IRs
verters (Ishaque et al., 2011b).  I o2 exp 1  1
a2 V T 2 Rp
Over the years, several researchers have studied the
characteristics of PV modules under partial shading condi- where IPV is the current generated by the incidence of light.
tions. In (Alonso-Garca et al., 2006), an experimental Io1 and Io2 are the reverse saturation currents of diode 1
study was carried out but was limited to module-level shad- and diode 2, respectively. The Io2 term is introduced to
ing. The eect of shading on the output of the PV modules compensate for the recombination loss in the depletion re-
and the changes in their IV characteristics was investi- gion as described in (Chih-Tang et al., 1957). Other vari-
gated by (Kawamura et al., 2003). A numerical algorithm ables are dened as follows: VT 1,2 (= Ns kT/q) is the
was proposed by (Quaschning and Hanitsch, 1996) to sim- thermal voltage of the PV module having Ns cells con-
ulate the complex shading characteristics of the arrays nected in series, q is the electron charge
whereby every element (each cell of the module, bypass (1.60217646  1019 C), k is the Boltzmann constant
diode, blocking diode, etc.) was represented by a mathe- (1.3806503  1023 J/K), and T is the temperature of the
matical expression. The results were attractive but the sim- pn junction in K. Variables a1 and a2 represent the diode
ulation was considerably complex, thus limiting its ability ideality constants; a1 and a2 represent the diusion and
to predict the performance of larger systems. A MATLAB recombination current component, respectively. Typically,
based modeling to study the eects of partial shading based three points at Standard Test Conditions (STC) (Isc, 0),
on the single diode model was proposed in (Patel and (Vmp, Imp) and (Voc, 0) are provided by the manufacturers
Agarwal, 2008). Although the model is relatively simple, datasheet. An accurate estimation of these points for other
it exhibits serious deciencies when subjected to high tem- conditions is the main goal of every modeling technique.
perature variations. Moreover, it does not account for the Although greater accuracy can be achieved using this
open circuit voltage coecient, KV (Ishaque et al., 2011a). model (compared to the single diode model), it requires
Considering the importance of this issue, this paper pro- the computation of seven parameters, namely IPV, Io1,
poses a fast, accurate, and comprehensive PV system simu- Io2, Rp, Rs, a1, and a2. To simplify, several researchers
lator using the MATLAB Simulink environment. The assumed a1 = 1 and a2 = 2. The values are chosen based
availability of the simulator in the MATLAB platform is on the approximations of the SchokleyReadHall
seen as an advantage from the perspective of researchers
and practitioners alike as this software has almost become I
a de-facto standard in various engineering discipline. An +
ID1 ID2 Rs
important contribution of this work is the application of
the two-diode model that is known to have better accuracy, V
IPV Rp
especially at low irradiance level. It allows for a more accu-
rate prediction of PV system performance during partial
-
shading conditions. To reduce computational time, the
model parameters extraction is reduced to four while the Fig. 1. Two diode model of PV cell.
 K. Ishaque et al. / Solar Energy 85 (2011) 22172227 2219

recombination in the space charge layer in the photodiode 2.2.3. Determination of Rp and Rs values
(Chih-Tang et al., 1957). Although this assumption is From Eq. (5), the expression for Rp at maximum power
widely used but not always true (Keith et al., 2000). point (MPP) can be rearranged and rewritten as:
V mp I mp Rs
Rp n h     i o
2.2. Extraction of model parameters V mp I mp Rs V mp I mp Rs P max;E
I PV I o exp VT
exp p1V T
2  V mp

2.2.1. Determination of IPV current 6

The equation for PV current as a function of tempera-
ture and irradiance can be written as: It can be seen that Eq. (6) is an algebraic equation  for
any nite value of Rs, the value of Rp can be calculated eas-
G ily. Therefore, by taking the advantage of this algebraic
I PV I PV STC K i DT 2
GSTC expression, a simple and linear iterative method can be
where I PV STC (in Ampere) is the light generated current at adopted. The remaining two parameters of Eq. (5), i.e.,
STC, DT T  T STC (in Kelvin, TSTC = 25 C), G is the Rp and Rs, are obtained through matching technique
surface irradiance of the cell, and GSTC (1000 W/m2) is the (Ishaque et al., 2011a). Although this approach has been
irradiance at STC. The constant Ki is the short circuit cur- used in the single diode model, it has not been applied
rent coecient, normally provided by the manufacturer. for two-diode model. The idea behind this method is to
match the calculated peak power (Pmp,C) and the experi-
mental (from manufacturers datasheet) peak power
2.2.2. Determination of saturation current and diode ideality (Pmp,E) by iteratively increasing the value of Rs while simul-
factors taneously calculating the Rp value.
For a single diode model, the equation to describe the To start the iteration process, the suitable initial values
reverse saturation current that considers the temperature (typically a low value to ensure better results) for Rp
variation is given by (Villalva et al., 2009), i.e., and Rs are assigned. Hence, Rs is chosen to be zero and
I PV STC K i DT Rp is calculated by following equation:
Io 3
exp V oc;STC K v DT =aV T   1    
V mp V ocn  V mp
Rpo  7
The constant Kv is the open circuit voltage coecient; this I scn  I mp I mp
value is available from the datasheet. For the two diode
model, several researchers have calculated the values of where the rst term in bracket in Eq. (7) is the slope of the
Io1 and Io2 using iteration. The iteration approach greatly line segment between the short circuit and the MPP point,
increases the computation time, primarily due to the non- while the second is the slope of line segment between the
suitable values of the initial conditions (Jinhui et al., open circuit voltage and MPP point. For a typical PV mod-
2009). In (Ishaque et al., 2011a), a new analytical equation ule, the dierence between the nominal values of Vocn and
of both saturation currents was derived. Both saturation Vmp is not large; consequently, the second term in Eq. (7)
currents can be calculated as: will always be very low. Furthermore, the rst term will
also result in a very low value compared to the actual value
I o1 I o2 I o
of Rp. This ensures that Eq. (7) will always return a low va-
I PV STC K i DT lue of Rp, which is a good initial guess for the iteration.
4
expV oc;STC K v DT =fa1 a2 =pgV T   1 For each iteration, the value of Rp is calculated simulta-
neously using Eq. (6). With the availability of all the seven
The equalization simplies the computation because no parameters, the output current of the cell can now be deter-
iteration is required; the solution can be obtained analyti- mined using well known numerical techniques. In our
cally. In accordance to Shockleys diusion theory, the dif- work, we have utilized standard NewtonRaphson
fusion current, a1 must be unity (Chih-Tang et al., 1957). method. This algorithm has the advantage of a very quick,
The value of a2, however, is exible. Based on extensive quadratic convergence for initial values near the root, so
simulation carried out, it is found that if a2 P 1.2, the best that a good solution can be calculated within a few itera-
match between the proposed model and practical IV curve tion steps (Quaschning and Hanitsch, 1996). The owchart
is observed. Since (a1 + a2)/p = 1 and a1 = 1, it follows that that describes the Pmp matching algorithm is given in
variable p can be chosen to be P2.2. This generalization Fig. 2.
can eliminate the ambiguity in selecting the values of a1 Using Eqs. (2) and (4), all the four parameters of this
and a2. Eq. (1) can be simplied in terms of p as. model can be readily extracted, i.e., IPV, Io, Rp, and Rs.
      From these, only Rp and Rs need to be determined by iter-
V IRs V IRs
I I PV  I o exp exp 2 ation. IPV and Io are obtained analytically. Variable p can
VT p  1V T
  be chosen to be any number value greater than 2.2. These
V IRs simplications have greatly improved the computational
 5
Rp speed of the simulator.
2220 K. Ishaque et al. / Solar Energy 85 (2011) 22172227

3.1. Verication of the two-diode model for dierent

irradiation levels

Fig. 3 shows the IV curves for a single KC200GT mod-

ule for dierent levels of irradiation (per unit quantity:
Sun = 1 equivalent to 1000 W/m2). The calculated values
from the proposed two-diode and Rp-models are evaluated
against measured data from the manufacturers datasheet.
Comparison to the Rs-model is not included to avoid over-
crowding of plot. However, the results for the Rs-model
will be analyzed later in the performance evaluation
between the three models. The proposed two-diode model
and the Rp-model exhibit similar results at STC. This is
to be expected because both models use the similar maxi-
mum power matching algorithm at STC. However, as the
irradiance goes lower, more accurate results are obtained
from the two-diode model, especially in the vicinity of
the open circuit voltage. At Voc, the Rp-model shows depar-
ture from the experimental data, suggesting that Rp-model
is inadequate when dealing with low irradiance level. This
variation is envisaged to have signicant impact during
partial shading.
Fig. 4 analyzes the relative error of Voc and the maxi-
mum power point Pmp for KC200GT module at dierent
Fig. 2. Matching algorithm. irradiance levels. The temperature is set to STC. The rela-
tive error is dened as the dierence between simulated and
measured Voc and Pmp values. The dierence is then divided
3. Validation of the two-diode model
by the measured value. As can be seen at STC irradiance,
there is a very small dierence in the Voc values among
The two diode model and its parameters extractions
the three models. However, as the irradiance is reduced,
described in this paper are validated by measured parame-
a signicant deviation is observed for the Rs and Rp-mod-
ters of ve PV modules of dierent brands/models. These
els. Similar results can be seen for the Pmp. On the other
include the multi-, mono-crystalline, and thin-lm types.
hand, the proposed two-diode model accurately calculates
The specications of the modules are summarized in
Pmp at all irradiance values.
Table 1. The detailed modeling results for the aforemen-
tioned PV modules (except SM55) using this model can
be seen in (Ishaque et al., 2011a). The SM55 PV module 3.2. Verication of the two-diode model for temperature
will be used later in the validation of partial shading variations
results. The computational results are compared with the
Rs (Walker, 2001) and Rp (Villalva et al., 2009) models. The performance of the models when subjected to tem-
Table 2 shows the parameters for the proposed two-diode perature variation is considered. All measurements are con-
model. The actual number of parameters computed is only ducted at STC irradiance of 1000 W/m2. The proposed
four because Io1 = Io2 = Io while a1 = 1 and p can be cho- model is compared to the Rs-model. The comparison
sen arbitrarily, i.e., p P 2.2. specically is chosen to highlight the problem with the

Table 1
STC specications for the modules used in the experiments.
Parameter Multi-crystalline Mono-crystalline Thin-lm
Siemens Kyocera Shell Shell Shell
SM55 KG200GT S36 SP-70 ST40
Isc 3.45 A 8.21 A 2.3 A 4.7 A 2.68 A
Voc 21.7 V 32.9 V 21.4 V 21.4 V 23.3 V
Imp 3.15 A 7.61 A 2.18 A 4.25 A 2.41 A
Vmp 17.4 V 26.3 V 16.5 V 16.5 V 16.6 V
Kv 77 mV/C 123 mV/C 76 mV/C 76 mV/C 100 mV/C
Ki 1.2 mA/C 3.18 mA/C 1 mA/C 2 mA/C 0.35 mA/C
Ns 36 54 36 36 42
 K. Ishaque et al. / Solar Energy 85 (2011) 22172227 2221

Table 2
Parameters for the proposed two-diode model.
Parameter Multi-crystalline Mono-crystalline Thin-lm
Siemens Kyocera Shell Shell Shell
SM55 KG200GT S36 SP-70 ST40
Isc 3.45 A 8.21 A 2.3 A 4.7 A 2.68 A
Voc 21.7 V 32.9 V 21.4 V 21.4 V 23.3 V
Imp 3.15 A 7.61 A 2.16 A 4.24 A 2.41 A
Vmp 17.4 V 26.3 V 16.7 V 16.5 V 16.6 V
Io 2.232  1010 A 4.218  1010 A 2.059  1010 A 4.206  1010 A 1.13  109 A
IPV 3.45 A 8.21 A 2.3 A 4.7 A 2.68 A
Rp 144.3 X 160.5 X 806.4 X 91 X 263.3 X
Rs 0.47 X 0.32 X 0.89 X 0.51 1.6 X

Proposed Two-diode Model Rp-Model Experimental Data Rs-model when subjected to temperature changes. The
KC200GT PV module tested here for analysis. As can be
KC200GT Multi-Crystalline PV M odule seen in Fig. 5, the curves IV computed by the two-diode
8 model t accurately to the experimental data for all tem-
Sun=1 perature conditions. In contrast, at higher temperature,
results from the Rs-model deviate from the measured val-
6 ues quite signicantly.
Sun=0.8 Fig. 6 shows the performance of the three models for the
Voc and Pmp when subjected to variations in module tem-
I (A)

4
Sun=0.6 perature. There is no signicant dierence between the Rp
and the two-diode models. However, the Rs model exhibits
poor performance for both Voc and Pmp calculations.
Sun=0.4
2

3.3. Verication of model for various PV technologies

Sun=0.2
0
0 10 20 30 To further evaluate the eectiveness of the proposed
V (V)
model for dierent silicon technologies, comparisons
Fig. 3. IV curves of Rp-model and proposed two-diode model of the between S36 (mono-crystalline), SP70 (multi-crystalline),
KC200GT PV module for several irradiation levels. and ST40 (thin-lm) are carried out. All these modules
are manufactured by Shell. For this test, the irradiance is
maintained constant at STC. Fig. 7a shows the relative
Relative Error @ Voc (%)

3
Rs-Model
Rp-Model
2 KC200GT Multi-Crystallline PV Module
Proposed Two-Diode Model

1 8

0
1000 800 600 400 200
Solar Irradiance (W/m2) 6
(a) Proposed Two-diode Model
Rs-Model
I (A)

Experimental Data
Relative Error @ Pmp(%)

6 4
o o
4 75oC 50 C 25 C

2 2

0
1000 800 600 400 200
Solar Irradiance (W/m 2) 0
0 10 20 30
(b) V (V)

Fig. 4. Relative error for Voc and Pmp, for Rs, Rp and the proposed two- Fig. 5. IV curves of Rs and proposed two-diode model of the KC200GT
diode model for KC200GT PV module. PV module for several temperature levels.
2222 K. Ishaque et al. / Solar Energy 85 (2011) 22172227

4
Relative Error @ Voc (%)

Rs Model Rp Model Proposed 2-D Model

Rs-Model 10
3
Rp-Model
Proposed Two-diode Model 5
2

1 0
50 25 0 -25

Relative Error @ Pmp (%)

0 5
25 50 75
Temperature (C)
(a)
4 0
Relative Error @ Pmp(%)

50 25 0 -25
3 40

2
20
1
0
50 25 0 -25
0
25 50 75 Temperature (C)
Temperature (C)
(a)
(b)
Rs Model Rp Model Proposed 2-D Model
Fig. 6. Relative error for Voc and Pmp, for Rs, Rp and the proposed two- 40
diode model for KC200GT PV module.
20

error for Pmp for a wide variation of temperature (25 C 0

1000 800 600 400 200
Relative Error @ Pmp (%)

to +50 C). From the data, it can be concluded that more

accurate results are obtained from the two-diode model for 40
all silicon technologies. Furthermore, it can be noted that
20
using the Rs-model, considerably errors occur for Vmp
and Pmp. This is particularly severe for the thin lm tech-
0
nology. Fig. 7b depicts a comparison of all three modeling 1000 800 600 400 200
techniques when the irradiance is varied from 200 to 40
1000 W/m2. A temperature of 25 C is maintained in these
analyses. As can be seen, superior results are obtained with 20
the proposed two-diode model. As expected, the Rs-model
shows a signicant variation with the ST40 PV module. 0
1000 800 600 400 200
The extensive experimental verication above has pro- Solar Irradiance (W/m )
2

ven that the two diode model is superior to that of single (b)
diode model with Rs and Rp. This justies its usage in the
proposed simulator. Fig. 7. Relative error for Pmp of Rs, Rp and the proposed two-diode model
for S36 (multi-crystalline), SP70 (mono-crystalline) and ST40 (thin lm)
(a) for irradiance variation and (b) temperature variation.
3.4. Computational time

Table 3 shows the computation time of three modeling Table 3

approaches. Shortest computation time is observed for Computation time (ms).
the Rs-model; this is to be expected as the calculation does PV module Rs Rp Proposed two-diode
not involve any iteration process. For the proposed model, Type Model Model Model
due to the inclusion of an extra diode current in Eq. (5), the SM55 8.1 108.1 215.6
computational time, in average, is about 1.6 times longer KC200GT 21.1 105.3 221.7
that of the Rp-model. The increase is much less than one S36 8.71 256.8 424.6
order of magnitude, and hence computational-wise, can SP-70 7.97 183.8 276.2
be regarded as insignicant. The relatively short computa- ST40 1.1 430 646.5
tional time is due to the fewer number of parameters to be
calculated compared to the previous work on two-diode
model. Furthermore, the iteration method to compute the improved accuracy achieved using the proposed model,
values of Rs and Rp is very ecient. By considering the the slight increased in computational time is justied.
 K. Ishaque et al. / Solar Energy 85 (2011) 22172227 2223

8 2 0  1
4. PV simulator development < V IRs N ss
N pp
I N pp I PV  I o 4exp @ A
4.1. Large array conguration : V T N ss
0  1 39 0  1
In a PV power generation, a centralized inverter topol- V IRs NNppss = V IRs NNppss
exp @ A  25  @   A 8
ogy is utilized due to economic reasons. A single (central- P  1V T N ss ; Rp NNppss
ized) inverter is connected to a large number of PV
modules, typically, in a series parallel conguration. This
topology is known as the seriesparallel (SP) topology as where I PV ; I 0 ; Rp ; Rs ; p are the parameters of an individual
shown in Fig. 8. PV module. Note that Nss and Npp are series and parallel
Based on Fig. 8, a modied equation for the current for PV modules, respectively in a given SP array. Using the
the SP conguration can be written as: large array conguration described above, a generalized
simulation tool is developed in MATLAB Simulink. The
structure of the simulator is shown in Fig. 9. The simulator
is able to cater for any number Nss  Npp array, where Nss
and Npp represent the number of series and parallel mod-
ules for each group, respectively. The simulator has the
capability of showing the eects of shading, temperature,
and diodes (bypass and blocking).
For illustration, an example of large array simulation
with partial shading condition is shown here. Fig. 10 shows
a PV array composed of four groups namely, groups A, B,
C, and D. Each group can be congured for a dierent
shading and temperature conditions.
Fig. 11 depicts the mask implementation of group A. It
has been developed by based on Eq. (8). To simulate the
Fig. 8. Series parallel combination in PV array. shading eects of PV array, role of bypass diode is very

Group A BL_di 1
Current
Volt
Out1
s
Sun_A Irradiance + + v Switch
- - PV_Data
Shading
Pattern A Blocking To Workspace
Group B
Current Diode
Volt
Sun_B Irradiance

Shading
1 Pattern B Group C
s Current
In1 + + i
- - Volt
Sun_C Irradiance

Shading
Pattern C
Group D I-V Curve of PV Simulator
Current
Volt Add
Sun_D Irradiance

Shading Pattern D
P-V Curve of PV Simulator

Fig. 9. Simulink implementation of the PV simulator.

Shading Pattern A Shading Pattern B Shading Pattern C Shading Pattern D

Blocking
Diode

Group A Group B Group C Group D

Fig. 10. SP connection for 20  3 PV array.

2224 K. Ishaque et al. / Solar Energy 85 (2011) 22172227

Fig. 11. A mask implementation of PV simulator for group A.

vital. Its eect is added by the voltage equation of bypass

diode that is given as:
 
I PV
V BD AV T ln 1 9
Io

where A is the diode ideality factor (for group A) and IPV is

the diode current of the bypass diode. Once the shading
condition will occur, current will ow from the bypass
diode instead of PV module and hence multiple peaks will
be observed at the output of PV array.
Fig. 12 shows the input parameter template of the pro-
posed simulator. The module used in this experiment is
SM55. All parameters for the inputs are available from
manufactures datasheet. The other inputs are the shading
pattern, temperature, number of series modules (Nss), and
number of parallel modules (Npp). Bypass diode of any
group can be removed by selecting in the popup menu in
the parameter template of the PV simulator.

5. Results and discussion

5.1. Eects of bypass diode

Under shaded conditions, the shaded modules behave as

a load instead of generator, which produce the hot spot
problem. This problem can be avoided by driving the cur-
rent of non-shaded PV modules through the bypass diode. Fig. 12. PV simulator Block parameters window in simulink.
The proposed simulator must be capable to replicate the
eects of this diode. For the given parameters of the simu-
lator as shown in Fig. 12, the IV and PV curves for three diode, and (3) with the same shading conditions of (2) but
dierent cases are shown in Fig. 13a and b. (1) With no without bypass diode. It can be observed that number of
shading (G = 1 kW/m2), (2) with partially shaded peaks equal to the irradiance imposed for each shading pat-
(Group A = 1 kW/m2, Group B = 0.75 kW/m2, Group tern. However, more precisely, it depends on the tempera-
C = 0.5 kW/m2, Group D = 0.25 kW/m2) and with bypass ture of the modules, the irradiation level, the shading
 K. Ishaque et al. / Solar Energy 85 (2011) 22172227 2225

Fig. 13. (a) IV curves, (b) PV curves for SM55 from PV simulator for parameters in Fig. 11.

Fig. 14. (a) IV curves, (b) PV curves for SM55 of PV simulator with and without blocking diode.

pattern, and the array conguration (Patel and Agarwal, circuit current, it becomes reversed biased and a small leak-
2008). age current ows through the diode. Besides removing the
current imbalance condition, blocking diode prevents the
module from loading the battery at night by avoiding cur-
5.2. Eect of blocking diode rent ow from the battery through the PV array.

Bypass diodes, which limit the hot spot eect, oer no

protection to the current imbalance. In normal conditions 5.3. Validation of partial shading
(i.e., no shading), bypass diodes are reverse biased and
remain in this position. In order to protect against current In order to verify the accurateness of the proposed
imbalance, a blocking diode is used in series with the PV model for partial shading conditions, the result is com-
array such as shown in Fig. 8. Blocking diode modify the pared with the work carried out by (Syafaruddin et al.,
forward characteristics of a PV array. During normal oper- 2010), which is based on Articial Neural Network. In
ation, it is forward biased and conducts the PV array cur- (Syafaruddin et al., 2010), a real-time PV emulator was
rent hence produce a small voltage drop that is shown in designed to emulate PV characteristics curves with a
Fig. 14. When the PV array current goes beyond the short special focus on partial shading conditions. For the PV
2226 K. Ishaque et al. / Solar Energy 85 (2011) 22172227

modules model, the PV characteristic was generated based IPV L D

on Sandias PV module electrical performance model. The + +
main component of the developed emulator consists of two PV
personal computers (PCs) with AnalogDigital (AD) and Array VPV SW C Vo R
DigitalAnalog (DA) hardware under a real-time dSPACE
platform. Several inhomogeneous irradiance distributions
were used to investigate the behavior of the proposed
MPPT Controller
system.
Here, SM55 PV module is used for the validation. It is
expected that once the accuracy of PV module is established Fig. 16. PV system description utilizing designed simulator.
(with the two-diode model), the simulator can be extended
to any type of PV module. This is a common practice by results to the data provided by (Syafaruddin et al. (2010)). In
other researchers too, for example (Patel and Agarwal, particular, it shows close resemblance especially in the vicin-
2008; Syafaruddin et al., 2010). It can be observed in ity of the global peak. However, there appear some discrep-
Fig. 15 that, in general, proposed simulator gives the similar ancies in the PV curves at the local peaks. The following
reasons could be the sources of disagreement. First, a silicon
solar cell was used as an irradiance sensor, which is known
to be inferior in accuracy. Secondly, due to the constraint
Data (Syafaruddin, Karatepe et al. 2010) Proposed Model
12 of measurement sampling, the change in irradiation over
time during may not be recorded.
10

5.4. Simulation with converter and MPP controller

8
Array Current (A)

One of the most important requirements of a PV system

simulator is its ability to interface with the power electronic
6
systems with their associated control algorithms. Fig. 16
depicts a simulation example of a PV system together with
4 its boost dcdc converter and MPPT controller. The SM55
PV modules are used in the simulation for a 20  3 array
2 conguration. The MPPT controller utilizes the conven-
tional Perturbation and Observe (P&O) algorithm. The
performance of P&O to track the global MPP is evaluated
0
0 50 100 150 200 250 300 350 400 for the partial shading condition shown in Fig. 13b.
Array Voltage (V)
The simulation results are shown in Fig. 17. Initially, the
(a) array receives a uniform irradiation of (G = 1 kW/m2). It is
observed that until t = 0.4 s, at which shading occurs, the
Data (Syafaruddin, Karatepe et al. 2010) Proposed Model

1200 Shading occurs here

400
(a)
Voltage (V)
PV Array

1000
200
Array Power (W)

800 0
0 0.2 0.4 0.6 0.8 1
10
Current (A)

600
(b)
PV Array

5
400
0
0 0.2 0.4 0.6 0.8 1
200 4000
(c)
Power(W)

2000
0
0 50 100 150 200 250 300 350 400
Local Maxima
Array Voltage (V)
0
(b) 0 0.2 0.4
Time(s)
0.6 0.8 1

Fig. 15. Simulation and experimental results (a) IV characteristics and Fig. 17. (a and b) Output voltage and current from the PV array, (c) PV
(b) PV characteristics. array power (solid line) and dcdc converter power (dash line).
 K. Ishaque et al. / Solar Energy 85 (2011) 22172227 2227

PV arrays voltage and current are retained at 309 V and Chowdhury, S., Taylor, G.A., et al., 2007. Modelling, simulation and
9.44 A, respectively. This corresponds to the MPP of performance analysis of a PV array in an embedded environment. In:
Universities Power Engineering Conference, 2007. UPEC 2007. 42nd
2917 W, i.e., point A in Fig. 13b. Due to the shading con- International.
ditions (at t = 0.4 s), operating point is shifted to a new Enrique, J.M., Andujar, J.M., et al., 2010. A reliable, fast and low cost
MPP at 760 W (point B). This new operating point is maximum power point tracker for photovoltaic applications. Solar
clearly a local maxima. This example highlights the inabil- Energy 84 (1), 7989.
ity of a conventional MPPT scheme (i.e., P&O) to distin- Gow, J.A., Manning, C.D., 1999. Development of a photovoltaic array
model for use in power-electronics simulation studies. Electric Power
guish between the global and local maxima. Applications IEE Proceedings 146 (2), 193200.
Hovinen, A., 1994. Fitting of the solar cell IV-curve to the two diode
6. Conclusion model. Physica Scripta T54, 2.
Hyvarinen, J., Karila, J., 2003. New analysis method for crystalline silicon
In this paper, a MATLAB Simulink PV system simula- cells. Photovoltaic energy conversion, 2003. In: Proceedings of 3rd
World Conference on.
tor based on an improved two-diode model is proposed. To Ishaque, K., Salam, Z., et al., 2011a. Simple, fast and accurate two-diode
reduce computational time, the input parameters are model for photovoltaic modules. Solar Energy Materials and Solar
reduced to four, and the values of Rp and Rs are estimated Cells 95 (2), 586594.
by an ecient iteration method. Furthermore, the inputs to Ishaque, K., Salam, Z., et al., 2011b. Modeling and simulation of
the simulator are information available on standard PV photovoltaic (PV) system during partial shading based on a two-
diode model. Simulation Modelling Practice and Theory 19 (7), 1613
module datasheet. The simulator supports large array sim- 1626.
ulation that can be interfaced with MPPT algorithms and Jinhui, X., Zhongdong, Y., et al., 2009. Design of PV array model based
actual power electronic converters. The accurateness of on EMTDC/PSCAD. In: Power and Energy Engineering Conference,
the simulator is veried by ve PV modules of dierent 2009. APPEEC 2009. Asia-Pacic.
types (multi-crystalline, mono-crystalline, and thin-lm) Kawamura, H., Naka, K., et al., 2003. Simulation of IV characteristics of
a PV module with shaded PV cells. Solar Energy Materials and Solar
from various manufacturers. It is observed that the two- Cells 75 (34), 613621.
diode model is superior to the Rp and Rs models. Further- Keith, M., Peter, A., et al., 2000. Depletion-region recombination in
more, a PV system, together with the power converters and silicon solar cells: when does mDR=2? In: 16th European Photovoltaic
controllers, is simulated. The results are found to be to be Solar Energy Conference and Exhibition, pp. 251254.
in close agreement with theoretical prediction. The Kurobe, K.-i., Matsunami, H.AU, 2005. New two-diode model for
detailed analysis of multicrystalline silicon solar cells. Japanese Journal
designed simulator is available for free and can be down- of Applied Physics 44 (12), 8.
loaded from the following Web site: http://sites.google. Mellit, A., Pavan, A.M., 2010. A 24-h forecast of solar irradiance using
com/site/drkishaque/Downloads. articial neural network: application for performance prediction of a
grid-connected PV plant at Trieste, Italy. Solar Energy 84 (5), 807
Acknowledgment 821.
Messai, A., Mellit, A., et al., 2011. Maximum power point tracking using a
GA optimized fuzzy logic controller and its FPGA implementation.
The authors would like to thank Universiti Teknologi Solar Energy 85 (2), 265277.
Malaysia for providing the facilities and research grant to Nishioka, K., Sakitani, N., et al., 2003. Analysis of the temperature
conduct this research and the reviewrs for thier valueable characteristics in polycrystalline Si solar cells using modied equiva-
comments. lent circuit model. Japanese Journal of Applied Physics 42 (Part 1, No.
12), 5.
Nishioka, K., Sakitani, N., et al., 2007. Analysis of multicrystalline silicon
References solar cells by modied 3-diode equivalent circuit model taking leakage
current through periphery into consideration. Solar Energy Materials
Albuquerque, F.L., Moraes, A.J., et al., 2010. Photovoltaic solar system and Solar Cells 91 (13), 12221227.
connected to the electric power grid operating as active power Patel, H., Agarwal, V., 2008. MATLAB-based modeling to study the
generator and reactive power compensator. Solar Energy 84 (7), eects of partial shading on PV array characteristics. Energy Conver-
13101317. sion, IEEE Transactions on 23 (1), 302310.
Alonso-Garca, M.C., Ruiz, J.M., et al., 2006. Experimental study of Quaschning, V., Hanitsch, R., 1996. Numerical simulation of current
mismatch and shading eects in the IV characteristic of a photovol- voltage characteristics of photovoltaic systems with shaded solar cells.
taic module. Solar Energy Materials and Solar Cells 90 (3), 329340. Solar Energy 56 (6), 513520.
Beser, E., Arifoglu, B., et al., 2010. A grid-connected photovoltaic power Syafaruddin, Karatepe, E., et al., 2010. Development of real-time
conversion system with single-phase multilevel inverter. Solar Energy simulator based on intelligent techniques for maximum power point
84 (12), 20562067. controller of photovoltaic system. International Journal of Innovative
Chaouachi, A., Kamel, R.M., et al., 2010. A novel multi-model neuro- Computing, Information and Control (IJICIC) 6, 20.
fuzzy-based MPPT for three-phase grid-connected photovoltaic sys- Villalva, M.G., Gazoli, J.R., et al., 2009. Comprehensive approach to
tem. Solar Energy 84 (12), 22192229. modeling and simulation of photovoltaic arrays. Power Electronics,
Chen, Y., Athienitis, A.K., et al., 2010. Modeling, design and thermal IEEE Transactions on 24 (5), 11981208.
performance of a BIPV/T system thermally coupled with a ventilated Walker, G., 2001. Evaluating MPPT converter topologies using a
concrete slab in a low energy solar house: Part 1, BIPV/T system and MATLAB PV model. Journal of Electrical & Electronics Engineering,
house energy concept. Solar Energy 84 (11), 18921907. Australia 21 (1), 8.
Chih-Tang, S., Noyce, R.N., et al., 1957. Carrier generation and Yoon, J.-H., Song, J., et al., 2011. Practical application of building
recombination in PN junctions and PN junction characteristics. integrated photovoltaic (BIPV) system using transparent amorphous
Proceedings of the IRE 45 (9), 12281243. silicon thin-lm PV module. Solar Energy 85 (5), 723733.