See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/262382589

A detailed modeling of photovoltaic module using MATLAB

Article in NRIAG Journal of Astronomy and Geophysics · June 2014

DOI: 10.1016/j.nrjag.2014.04.001

CITATIONS READS

555 5,882

3 authors, including:

Moulay Fatima
University of Sidi-Bel-Abbes
2 PUBLICATIONS 558 CITATIONS

SEE PROFILE

All content following this page was uploaded by Moulay Fatima on 12 December 2015.

The user has requested enhancement of the downloaded file.

NRIAG Journal of Astronomy and Geophysics (2014) xxx, xxx–xxx

National Research Institute of Astronomy and Geophysics

NRIAG Journal of Astronomy and Geophysics

www.elsevier.com/locate/nrjag

A detailed modeling of photovoltaic module

using MATLAB
a,*
Habbati Bellia , Ramdani Youcef b, Moulay Fatima b

a
Universite Bechar, Algeria
b
Universite Sidi-Bel-Abbes, Algeria

Received 13 June 2013; revised 14 February 2014; accepted 9 April 2014

KEYWORDS Abstract The PV module is the interface which converts light into electricity. Modeling this device,
Irradiance; necessarily requires taking weather data (irradiance and temperature) as input variables. The output
Temperature; can be current, voltage, power or other. However, trace the characteristics I(V) or P(V) needs of
I(V)/P(V) characteristic; these three variables. Any change in the entries immediately implies changes in outputs. That is
MATLAB/Simulink why, it is important to use an accurate model for the PV module. This paper presents a detailed
modeling of the effect of irradiance and temperature on the parameters of the PV module. The cho-
sen model is the single diode model with both series and parallel resistors for greater accuracy. The
detailed modeling is then simulated step by step using MATLAB/Simulink software due to its
frequent use and its effectiveness.
ª 2014 Production and hosting by Elsevier B.V. on behalf of National Research Institute of Astronomy
and Geophysics.

1. Introduction physical nature of the cell behavior are necessary to establish

a mathematical model of the PV cell and the PV module, in
The number of unknown parameters increases when the equiv- addition of course, to the use of that information given by
alent circuit of the chosen model becomes more convenient the constructors. The objective of this paper is to present use-
and far from being the ideal form. But most of the manufac- ful work to those who want to focus their attention on the PV
turers’ data sheets do not give enough information about the module or array as one device in a complex ‘‘electro-energetic
parameters which depend on weather conditions (irradiance system’’. So, the goal is to obtain at any time, the maximum
and temperature). So, some assumptions with respect to the power but also the more precise, therefore, the closest to the
experimental value.
* Corresponding author. Tel.: +213 049819193. The characteristic I(V) is a non-linear equation with multi-
E-mail addresses: [email protected] (H. Bellia), ple parameters classified as follows: those provided by con-
[email protected] (R. Youcef), [email protected] structors, those known as constants and the ones which must
(M. Fatima). be computed. Sometimes, searchers develop simplified meth-
Peer review under responsibility of National Research Institute of ods where, some unknown parameters cannot be calculated.
Astronomy and Geophysics.
They are thus assumed constant. For example, in Walker
and Geoff (2001) the series resistance RS was included, but
not the parallel resistance for a model of moderate complexity.
Production and hosting by Elsevier The same assumption is adopted in Benmessaoud et al. (2010),

2090-9977 ª 2014 Production and hosting by Elsevier B.V. on behalf of National Research Institute of Astronomy and Geophysics.
http://dx.doi.org/10.1016/j.nrjag.2014.04.001

Please cite this article in press as: Bellia, H. et al., A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal of Astronomy
and Geophysics (2014), http://dx.doi.org/10.1016/j.nrjag.2014.04.001
2 H. Bellia et al.

Atlas and Sharaf (1992), Beckman et al. (xxxx), Bryan (1999),

Table 1 Ideality factor (A) Huan-Liang et al., 2008, [16].
Bouzid et al. (2005), by considering the parallel resistance very
large. Other authors neglect both parallel and series resis- Technology Ideality factor
tances; the former due to being very large, the latter being very Si-mono 1.2
small. On the other hand, there are in the literature other Si-poly 1.3
papers, in which, these two internal characteristics of the PV a-Si-H 1.8
module are very important and have to be determined more a-Si-H tandem 3.3
accurately as in Townsend (1989), Alsayid and Jallad (2011), a-Si-H triple 5
cdTe 1.5
Kashif Ishaque and Syafaruddin (2011), Gazoli et al. (2009),
CTs 1.5
De Soto (2006) and Chouder et al. (2012). In addition to the
AsGa 1.3
series and parallel resistances and according to the authors,
two or three other parameters are to be determined; the photo
current (Iph), the saturation current (I0) and the ideality factor
(A). temperature and so, vary with varying conditions. In this
work, this term is designed by ‘a’ and called the thermal volt-
2. Presentation and modeling of PV module age (V), the ideality factor, is considered constant and is cho-
sen in Table 1 according to technology of the PV cell. The
2.1. In the Fig. 1a thermal voltage ‘‘a’’ is presented by equation (4)
NS  A  k  TC
a¼ ¼ N S  A  VT ð4Þ
The model does not take into account the internal losses of the q
current. A diode is connected in anti-parallel with the light In Chouder et al. (2012)), ‘a’ is called ‘‘the modified ideality
generated current source. The output current I is obtained factor’’ and is considered as a parameter to determine, while A
by Kirchhoff law: is the diode ideality (See Table 2).
I ¼ Iph  Id ð1Þ
2.2. In the (Fig. 1b)
Iph is the photocurrent, Id is the diode current which is pro-
portional to the saturation current and is given by the equation
(13): In reality, it is impossible to neglect the series resistance Rs and
    the parallel resistance RP because of their impact on the effi-
V ciency of the PV cell and the PV module. When RS is taken
Id ¼ I0 exp 1 ð2Þ
A:Ns :VT into consideration, equation (2) should take the next form:
   
V is the voltage imposed on the diode. V þ I  RS
Id ¼ I0 exp 1 ð5Þ
VT ¼ k  Tc =q ð3Þ a

I0 is the reverse saturation or leakage current of the diode Of course, Fig. 1b is a simplified form, easy to implement in
(A), VTc = 26 mV at 300 K for silisium cell, Tc is the actual simulators. But Fig. 1c is the most representative of the PV
cell temperature (K), k Boltzmann constant 1.381 · 1023 cell.
J/K, q is electron charge (1.602 · 1019 C).
VT is called the thermal voltage because of its exclusive 2.3. In the (Fig. 1c)
dependence of temperature (Anne and Michel, 2006; Sheik
Mohammed, 2011). By applying Kirchhoff law, current will be obtained by the
Ns: is the number of PV cells connected in series. A is the equation:
ideality factor. It depends on PV cell technology and can be
I ¼ Iph  Id  Ip ð6Þ
chosen in Table 1. It is necessary to underline that A is a con-
stant which depends on PV cell technology. IP, is the current leak in parallel resistor.
All the terms by which, V is divided in equation (2) According to the equation (7), the output current of a
under exponential function are inversely proportional to cell module containing Ns cells in series will be:

a) Ideal single diode model b) Practical model with Rs c) Practical model with Rs and Rp

Fig. 1 The equivalent circuit of a solar cell and PV device.

Please cite this article in press as: Bellia, H. et al., A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal of Astronomy
and Geophysics (2014), http://dx.doi.org/10.1016/j.nrjag.2014.04.001
A detailed modeling of photovoltaic module 3

3.2. Determination of I0
Table 2 PWX 500 PV module (49 W) characteristics.
Parameters Values
The shunt resistance RP is generally regarded as great, so the
Pmp (W) 49 last term of the relationship (8) should be eliminated for the
Imp (A) 2.88 next approximation. By applying equation (8) at the three
Vmp (V) 17
most remarkable points at standard test condition: the voltage
Isc (A) 3.11
Voc (V) 21.8
at open circuit (I = 0, V = Voc,ref), the current at short circuit
RS (O) 0.55 (V = 0, I = Isc,ref), and the voltage (Vmp,ref) and current
Noct C 45 (Imp,ref) at maximum power, the following equations can be
lsc (K) 1.3 Æ 103 written:
Kd (K) 72.5 Æ 103    
Isc;ref  Rs
NS 36 Isc;ref ¼ Iph;ref  I0;ref exp 1 ð12Þ
aref
   
Voc
   0 ¼ Iph;ref  I0;ref exp 1 ð13Þ
V þ I:RS V þ Rs I aref
I ¼ Iph  I0 exp 1  ð7Þ
a Rp    
Vpm;ref þ Ipm;ref Rs
Ipm;ref ¼ Iph;ref  I0;ref exp 1 : ð14Þ
It is not easy to determine the parameters of this transcen- aref
dental equation. But this model offers the best match with
experimental values. The (1) term has to be neglected because it is very smaller
than the exponential term. According to equation (11), and by
substituting (Iph,ref) in equation (14):
3. Determination of the parameters  
Voc;ref
0  Isc;ref  I0;ref exp ð15Þ
The number of parameters varies depending on the chosen aref
model and on the assumptions adopted by the searchers. For So:
example, in Beckman et al. (xxxx) and De Soto (2006), it is  
considered that Iph, I0, RS, RP and the factor ideality are five Voc;ref
I0;ref ¼ Isc;ref exp ð16Þ
parameters that depend on the incident solar radiation and a
the cell temperature. While in Bryan (1999), and Townsend The reverse saturation current is defined by:
(1989), the unknown parameters are Iph, I0, RS and c. Where qe 
G
c = A Æ NS I0 ¼ DT3C exp ð17Þ
In this work the four parameters that have to be evaluated Ak
are also Iph, I0, RS, RP.

3.1. Determination of Iph Imput T,G

Imput Imp, Vmp, Isc, Voc, μsc, Ns
Initialisation Rs=0, Rp by eq (21)
According to Fig. 1a, the output current at the standard test
conditions (STC) is:
   
V
I ¼ Iph;ref  I0;ref exp 1 ð8Þ Calculate Iph by eq (11)
aref
and I0 by eq (19)
This equation allows quantifying Iph,ref which cannot be
determined otherwise. When the PV cell is short-circuited:
   
0
Isc;ref ¼ Iph;ref  I0;ref exp  1 ¼ Iph;ref ð9Þ No
aref
Error > Tolerance
But this equation is valid only in ideal case. So, the equality
is not correct. And then, equation (10) has to be written as:
Yes
Iph;ref  Isc;ref ð10Þ
The photocurrent depends on both irradiance and
temperature: Newton Raphson solution of eq (21)
Err=|Pmax-Pmax,ex|
G
Iph ¼ ðIph;ref þ lsc  :DTÞ ð11Þ
Gref
G: Irradiance (W/m2), Gref : Irradiance at STC = 1000
W/m2, DT = Tc  Tc,ref (Kelvin), Tc,ref : Cell temperature at
STC = 25 + 273 = 298 K, lSC : Coefficient temperature of End
short circuit current (A/K), provided by the manufacturer,
Iph,ref : Photocurrent (A) at STC. Fig. 2 Iteration flow chart.

Please cite this article in press as: Bellia, H. et al., A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal of Astronomy
and Geophysics (2014), http://dx.doi.org/10.1016/j.nrjag.2014.04.001
4 H. Bellia et al.

1
G
Product 1
Iph
0.0013 Divide 2

Temp coef
2
T 1000
298
273
Tref
273
3.11

Iscref1

Fig. 3 Detailed Iph implementation.

36

Ns
1.3805 *10 ^(-23 )

k 1.3 3.11
1
A Iscref
Io ,ref
1
u
T e

Math
273
Function
21 .8
273
Voc 1
1.6021 *10 ^(-19 )

q Io ,ref 1

Fig. 4 Detailed I0,ref implementation.

1.6021 *10 ^(-19 )

q
1.12

Eg
exp (u(1))

1.3

1.3805 *10 ^(-23 )

1/298

Io 1
1

273

1
T
298 (u(1))^3 1
Io
2
Io ,ref

Fig. 5 Detailed I0 implementation.

Please cite this article in press as: Bellia, H. et al., A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal of Astronomy
and Geophysics (2014), http://dx.doi.org/10.1016/j.nrjag.2014.04.001
A detailed modeling of photovoltaic module 5

Rp

Ramp 1

Rs
1.6*10 ^(-19 )

q1
2
3 Iph
T

273 exp(u)
1
273 Fcn1 I

1
1.3805 *10 ^(-23 )
1
k Io

1.6
I1
A

36

Ns

Fig. 6 Detailed model with Rp term.

1.6021 *10 ^(-19 )

Ramp
q1 Rs

1
3
T I
2
Iph
273

273

1.3805 *10 ^(-23 )

I1
k
exp (u)
1.3
Fcn1
A
1
36
1
Ns
Io

Fig. 7 Detailed model without Rp term.

In order to eliminate the diode diffusion factor, equation

(18) is computed twice; at Tc and at Tc,ref. Then, the ratio of
the two equations is written in the next expression:
 3   
1000 G I Tc qeG  1 1
I0 ¼ I0;ref exp  ð18Þ
I
Tc;ref A  K Tc;ref Tc
Irradiance
  3
Voc;ref Tc
I0 ¼ ISC;ref exp
a Tc;ref
P   
qeG  1 1
 exp  ð19Þ
A  K Tc;ref Tc
25 T V
Equation (20) presents I0 with some parameters provided
temperature V
by the manufacturers as (Voc, ref, Tc,ref), others, related to the
technology of the PV cell, as (A, eG) and some constants.
PV Module
But ‘‘a’’ and Tc are dependents of actual temperature. That
is why; I0 has to be determined at real time.
Fig. 8 Presentation of the whole PV model.
3.3. Determination of RP and RS

eG: Material band gap energy (eV), (1.12 eV for Si) In order to make the proposed model more credible, RP and
D = diode diffusion factor RS are chosen so that the computed max power Pmp is equal

Please cite this article in press as: Bellia, H. et al., A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal of Astronomy
and Geophysics (2014), http://dx.doi.org/10.1016/j.nrjag.2014.04.001
6 H. Bellia et al.

Io Io

T Io ,ref Io,ref

1 G
Io
Io ,ref G
Iph Iph I 1
2 I
T 2 T
V
Ramp
Iph
T

Module output Current

Fig. 9 Grouped subsystems of the PV model.

Fig. 11 I(V) characteristic in Rp model.

Fig. 10 I(V) characteristic in Rs model.

to the experimental one Pmp,ex at STC conditions. So it is pos-

sible to write the next equation:
Imp;ref ¼ Pmp;ref =Vmp;ref ¼ Pmp;ex =Vmp;ref
   
Vmp;ref þ Imp;ref  RS
¼ Iph;ref  I0;ref exp 1
a
Vmp;ref þ Rs Imp;ref
 ð20Þ
Rp

V þI R
RP ¼ n mp;refh mp;ref S io ð21Þ
Vmp;ref þRS Imp;ref Voc;ref
Isc;ref  Isc;ref exp a
þ

Isc;ref exp ðVoc;ref =aÞ  Pmax;ex =Vmp;ref
The iteration starts at RS = 0 which must increase in order to
move the modeled Maximum Power Point until it matches
with the experimental Maximum Power Point. The corre-
sponding RP is then computed. There is only one pair (RP,
RS) that satisfies this condition.
The implementation presented in Fig. 6. was used to simu-
late the proposed model by incrementing RS until matching Fig. 12 I(V) characteristic in RS.

Please cite this article in press as: Bellia, H. et al., A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal of Astronomy
and Geophysics (2014), http://dx.doi.org/10.1016/j.nrjag.2014.04.001
A detailed modeling of photovoltaic module 7

Fig. 15 I(V) characteristic for different RS.

Fig. 13 I(V) characteristic in RP model.

Fig. 16 I(V) characteristics by varying irradiance.

Fig. 14 P(V) characteristic for different RS.

Pmp with Pmp,ex. Three curves for three different values of Rs 4. Simulation of the PV model
are presented in Fig. 14. The experimental value of maximum
power at STC, provided by the manufacturer of PWX 500 PV - First, equation (11) is substituted in equation (12) which
module (49 W) was used in equation (22). The iterative method gives the photocurrent and then, equation (12) is imple-
to compute the pair (RS, RP) gave RS = 0.45 O, mented in MATLAB/Simulink environment. The result is
RP = 310.0248 O. These two values make the proposed model represented in Fig. 3
the most representative of the chosen PV module. - The reverse saturation current at STC I0,ref is implemented
In order to simulate another PV module, its respective too, according to equation (17). It is represented in Fig. 4.
experimental maximum power is introduced in the equation - This allows the simulation of I0 which is represented in
(22) and then, the iterative method is used again to determine Fig. 5. It is a schematic form of equation (20).
the appropriate pair (RS, RP) which makes this same model the - Equation (8) is represented in two different forms; with and
most representative. Now, the RP model can be used to simu- without the third term containing the parallel resistance RP.
late the given module at different temperatures and irradiances The both forms of equation (8) are simulated and repre-
(See Fig. 2). sented respectively in Fig. 6 and 7.

Please cite this article in press as: Bellia, H. et al., A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal of Astronomy
and Geophysics (2014), http://dx.doi.org/10.1016/j.nrjag.2014.04.001
8 H. Bellia et al.

Fig. 17 I(V) characteristic by varying temperature. Fig. 19 P(V) characteristics by varying irradiance.

proportion to the series resistance. This is in accordance with

the fill factor relation (Anne and Michel, 2006; Bernard, 2004):
Pmp
FF ¼ ð22Þ
Voc  Isc
If Isc and Voc are constant, the fill factor FF changes only
according to Pmp. which is with respect of both Rs and Rp in
equation (21).
As shown in Fig. 12, it is confirmed that the manufacturer
did not take into consideration the parallel resistance, because
the peak power of the Rs model in this work matches with the
experimental peak power given in the data sheet for
Rs = 0.55 O. But for the Rp model, the peak power is logically
less than the experimental one (See Fig. 13).
The proposed Rp model is more accurate and the most
appropriate to simulate PWX 500 PV module (49 W) and
any other PV module. For PWX 500 PV module (49 W), all
the parameters are available to compute iteratively Rs and
Rp. The values were applied in the detailed Rp model presented
Fig. 18 P(V) characteristic by varying temperature. in Fig. 6. The results are presented in Fig. 14 and 15.
The proposed Rp model is now used to simulate the PV
module at different values of irradiance and temperature.
- The complete model is presented in Fig. 8. Irradiance and The I(V) characteristics are presented in Fig. 16 by varying
temperature are the inputs while the outputs are current irradiance from 200 W/m2 to 1000 W/m2 and taking The
and voltage. STC temperature. In Fig. 17, the temperature varies from
- To make simulation well organized, subsystems are 0 C to 60 C at STC irradiance.
grouped and presented in Fig. 9. P(V) characteristics are then presented in Fig. 18 by varying
temperature from 0 C to 60 C at STC under STC irradiance.
In Fig. 19 the irradiance varies from 200 W/m2 to 1000 W/m2
5. Curves and interpretation under STC temperature.

First, I(V) characteristic is plotted for Rs equal to zero and 6. Conclusions

Rs = 0.55 O. The last value of the series resistance is provided
by manufacturer. It is possible to point out some comments in The presented work is a detailed modeling and simulation of
Figs. 10 and 11; as the fact that neither Isc nor Voc are affected the PV cell and module. It is implemented under MATLAB/
by the change of the series resistance. In spite of this, the shape Simulink environment; the most used software by researchers
moves to the rectangular form when Rs decreases. The and engineers. This model is first drafted in accordance with
Maximum Power Point moves to the right, so, Pmp is in reverse the fundamentals of semiconductors and the PV cell

Please cite this article in press as: Bellia, H. et al., A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal of Astronomy
and Geophysics (2014), http://dx.doi.org/10.1016/j.nrjag.2014.04.001
A detailed modeling of photovoltaic module 9

technology. In other words, the PV module parameters have University Of Wisconsin Solar Energy Laboratory. 1500 Engineer-
been selected according to their variation with illumination ing Drive, Madison, U.S.A.
and temperature. It means that for any type of PV module, Bryan F., 1999, Simulation of grid-tied building integrated photovol-
one can use this model and determine all the necessary param- taic systems. MS thesis. Solar Energy Laboratory, University of
Wisconsin, Madison.
eters under any new conditions of irradiance and temperature
Bouzid, A., Chenni, R., Kerbache, T., Makhlouf, M., 2005. A Detailed
and then, obtain the I(V) and P(V) characteristics. This model Modeling Method for Photovoltaic Cells Energy. Elsevier.
can be considered as a tool which can be used to study all types Townsend, T.U., 1989, Method For Estimating The Long-Term
of PV modules available in markets, especially, their behavior Performance Of Direct-Coupled Photovoltaic Systems. M.S. The-
under different weather data of standard test conditions sis, Mechanical Engineering, U. Of Wisconsin-Madison.
(STC). It is important to compute RS, even if it is given by a Alsayid, B., Jallad, J., 2011. Modeling and simulation of photovoltaic
manufacturer because the experimental Maximum Power cells/modules/arrays. Int. J. Res. Rev. Comput. Sci. (IJRRCS) 2
Point does not match with the computed one. For each itera- (6).
tion, a pair of (Rs, RP) is obtained. But only one pair satisfies Kashif Ishaque, Syafaruddin, Zainal Salam, 2011. A comprehensive
the condition of matching the modeled and the experimental MATLAB Simulink PV system simulator with partial shading
capability based on two- diode model. Sol. Energy 85, 2217–2227.
peak power. So, Rs is iteratively increased until satisfying the
Gazoli, J.R., Ruppert, E., Villalva, M.G., 2009. Modeling and circuit –
condition. The proposed Rp model gave (Rs = 0.45 O, based simulation of photovoltaic arrays. Braz. J. Power Electron.
Rp = 310.0248 O) instead of (Rs = 0.55 O, and Rp not pro- 14 (1), 35–45.
vided). So it can be estimated to be more accurate in simulat- De Soto, W., 2006. Improvement And Validation Of A Model For
ing the PV module. Photovoltaic Array Performance By Solar Energy, 80, 78-88.
Chouder, A., Rahmani, L., Sadaoui, N., Silvestre, S., 2012. Modeling
References and simulation of a grid connected PV system based on the
evaluation of main PV module parameters. Simul. Model. Pract.
Theory 20, 46–58.
Walker, Geoff, 2001. Evaluating MPPT converter topologies using a
Anne, Labouret, Michel, Villoz., Energie photovoltaı̈que. (Dunod
matlab PV model. Aust. J. Electr. Electron. Eng. 21 (1).
3ème édition 2006).
Benmessaoud, M.T., Boudghene Stambouli, A., Midoun, A., Zegrar,
Bernard Equer, Energie solaire photovoltaı̈que (Ellipses 2004).
M., Zerhouni, F.Z., Zerhouni, M.H., 2010. Proposed methods to
Sheik Mohammed, S., 2011. Modeling and simulation of photovoltaic
increase the output efficiency of a photovoltaic (PV) system. Acta
module using MATLAB/Simulink. Int. J. Chem. Environ. Eng. 2
Polytech. Hung. 7 (2), 11.
(5).
Atlas, H., Sharaf, M., 1992. A fuzzy logic power tracking controller for
Huan-Liang Tsai., Ci-Siang Tu., Yi-Jie Su., 2008. Development of
a photovoltaic energy conversion scheme. Electr. Power Syst. Res.
generalized photovoltaic model using MATLAB/Simulink. Pro-
25, 227–238.
ceedings of the World Congress on Engineering and Computer
Beckman, W.A., Klein, S.A., Kou, Q., A method For Estimating The
Science October 22 - 24, San Francisco, USA.
Long-Term Performance Of Direct-Coupled Pv Pumping Systems,

Please cite this article in press as: Bellia, H. et al., A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal of Astronomy
and Geophysics (2014), http://dx.doi.org/10.1016/j.nrjag.2014.04.001
View publication stats