MODELING AND CIRCUIT-BASED

SIMULATION OF PHOTOVOLTAIC ARRAYS

Marcelo Gradella Villalva, Jonas Rafael Gazoli, Ernesto Ruppert Filho

University of Campinas (UNICAMP), Brazil
[email protected], [email protected], [email protected]

Abstract - This paper presents an easy and accurate This text presents in details the equations that form the the
method of modeling photovoltaic arrays. The method is I-V model and the method used to obtain the parameters of
used to obtain the parameters of the array model using the equation. The aim of this paper is to provide the reader
information from the datasheet. The photovoltaic array with all necessary information to develop photovoltaic array
model can be simulated with any circuit simulator. The models and circuits that can be used in the simulation of power
equations of the model are presented in details and the converters for photovoltaic applications.
model is validated with experimental data. Finally, sim-
ulation examples are presented. This paper is useful for II. MODELING OF PHOTOVOLTAIC ARRAYS
power electronics designers and researchers who need an
effective and straightforward way to model and simulate A. Ideal photovoltaic cell
Fig. 1 shows the equivalent circuit of the ideal photovoltaic
photovoltaic arrays.
cell. The basic equation from the theory of semiconductors
[1] that mathematically describes the I-V characteristic of the
ideal photovoltaic cell is:
Keywords – PV array, modeling, simulation.
   
I. INTRODUCTION qV
I=Ipv,cell − I0,cell exp −1 (1)
akT
A photovoltaic system converts sunlight into electricity. | {z }
Id
The basic device of a photovoltaic system is the photovoltaic
cell. Cells may be grouped to form panels or modules. Panels where Ipv,cell is the current generated by the incident light (it is
can be grouped to form large photovoltaic arrays. The term ar- directly proportional to the Sun irradiation), Id is the Shockley
ray is usually employed to describe a photovoltaic panel (with diode equation, I0,cell [A] is the reverse saturation or leakage
several cells connected in series and/or parallel) or a group current of the diode [A], q is the electron charge [1.60217646 ·
of panels. Most of time one are interested in modeling pho- 10−19 C], k is the Boltzmann constant [1.3806503·10−23J/K],
tovoltaic panels, which are the commercial photovoltaic de- T [K] is the temperature of the p-n junction, and a is the diode
vices. This paper focuses on modeling photovoltaic modules ideality constant. Fig. 2 shows the I-V curve originated from
or panels composed of several basic cells. The term array used (1).
henceforth means any photovoltaic device composed of sev-
eral basic cells. In the Appendix at the end of this paper there B. Modeling the photovoltaic array
are some explanations about how to model and simulate large The basic equation (1) of the elementary photovoltaic cell
photovoltaic arrays composed of several panels connected in does not represent the I-V characteristic of a practical pho-
series or in parallel. tovoltaic array. Practical arrays are composed of several con-
The electricity available at the terminals of a photovoltaic nected photovoltaic cells and the observation of the charac-
array may directly feed small loads such as lighting systems teristics at the terminals of the photovoltaic array requires the
and DC motors. Some applications require electronic con- inclusion of additional parameters to the basic equation [1]:
verters to process the electricity from the photovoltaic device.
These converters may be used to regulate the voltage and cur- practical PV device
rent at the load, to control the power flow in grid-connected
ideal PV cell I
systems and mainly to track the maximum power point (MPP)
of the device.
Ipv Id Rs
Photovoltaic arrays present a nonlinear I-V characteristic Rp V
with several parameters that need to be adjusted from experi-
mental data of practical devices. The mathematical model of
the photovoltaic array may be useful in the study of the dy-
namic analysis of converters, in the study of maximum power Fig. 1. Single-diode model of the theoretical photovoltaic cell and
point tracking (MPPT) algorithms and mainly to simulate the equivalent circuit of a practical photovoltaic device including the
photovoltaic system and its components using simulators. series and parallel resistances.

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1244

 Ipv Id I
equation, provide only a few experimental data about electri-
- = cal and thermal characteristics. Unfortunately some of the pa-
rameters required for adjusting photovoltaic array models can-
V V V
not be found in the manufacturers’s data sheets, such as the
light-generated or photovoltaic current, the series and shunt
resistances, the diode ideality constant, the diode reverse sat-
Fig. 2. Characteristic I-V curve of the photovoltaic cell. The net
cell current I is composed of the light-generated current Ipv and the uration current, and the bandgap energy of the semiconductor.
diode current Id . All photovoltaic array datasheets bring basically the following
information: the nominal open-circuit voltage Voc,n , the nom-
inal short-circuit current Isc,n , the voltage at the maximum
power point Vmp , the current at the maximum power point
voltage
(0, Isc ) source Imp , the open-circuit voltage/temperature coefficient KV , the
MPP
(Vmp , Imp )
short-circuit current/temperature coefficient KI , and the max-
imum experimental peak output power Pmax,e . This informa-
I

tion is always provided with reference to the nominal or stan-

current dard test conditions (STC) of temperature and solar irradia-
source (Voc , 0) tion. Some manufacturers provide I-V curves for several irra-
V diation and temperature conditions. These curves make easier
the adjustment and the validation of the desired mathematical
Fig. 3. Characteristic I-V curve of a practical photovoltaic device I-V equation. Basically this is all the information one can get
and the three remarkable points: short circuit (0, Isc ), maximum from datasheets of photovoltaic arrays.
power point (Vmp , Imp ) and open-circuit (Voc , 0).
Electric generators are generally classified as current or
voltage sources. The practical photovoltaic device presents an
hybrid behavior, which may be of current or voltage source de-
    pending on the operating point, as shown in Fig. 3. The practi-
V + Rs I V + Rs I cal photovoltaic device has a series resistance Rs whose influ-
I = Ipv − I0 exp −1 − (2)
Vt a Rp ence is stronger when the device operates in the voltage source
region, and a parallel resistance Rp with stronger influence in
where Ipv and I0 are the photovoltaic and saturation currents the current source region of operation. The Rs resistance is
of the array and Vt = Ns kT /q is the thermal voltage of the the sum of several structural resistances of the device [24].
array with Ns cells connected in series. Cells connected in The Rp resistance exists mainly due to the leakage current of
parallel increase the current and cells connected in series pro- the p-n junction and depends on the fabrication method of the
vide greater output voltages. If the array is composed of Np photovoltaic cell. The value of Rp is generally high and some
parallel connections of cells the photovoltaic and saturation authors [11–14, 17, 18, 25–28] neglect this resistance to sim-
currents may be expressed as: Ipv =Ipv,cellNp , I0 =I0,cell Np . plify the model. The value of Rs is very low and sometimes
In (2) Rs is the equivalent series resistance of the array and Rp this parameter is neglected too [26, 29–31].
is the equivalent parallel resistance. This equation originates The I-V characteristic of the photovoltaic device shown in
the I-V curve seen in Fig. 3, where three remarkable points Fig. 3 depends on the internal characteristics of the device
are highlighted: short circuit (0, Isc ), maximum power point (Rs , Rp ) and on external influences such as irradiation level
(Vmp , Imp ) and open-circuit (Voc , 0). and temperature. The amount of incident light directly affects
Eq. (2) describes the single-diode model presented in Fig. the generation of charge carriers and consequently the current
1. Some authors have proposed more sophisticated models generated by the device. The light-generated current (Ipv ) of
that present better accuracy and serve for different purposes. the elementary cells, without the influence of the series and
For example, in [2–6] an extra diode is used to represent the parallel resistances, is difficult to determine. Datasheets only
effect of the recombination of carriers. In [7] a three-diode inform the nominal short-circuit current (Isc,n ), which is the
model is proposed to include the influence of effects which maximum current available at the terminals of the practical
are not considered by the previous models. For simplicity the device. The assumption Isc ≈ Ipv is generally used in photo-
single-diode model of Fig. 1 is studied in this paper. This voltaic models because in practical devices the series resis-
model offers a good compromise between simplicity and ac- tance is low and the parallel resistance is high. The light-
curacy [8] and has been used by several authors in previous generated current of the photovoltaic cell depends linearly on
works, sometimes with simplifications but always with the the solar irradiation and is also influenced by the temperature
basic structure composed of a current source and a parallel according to the following equation [19, 32–34]:
diode [9–23]. The simplicity of the single-diode model with
the method for adjusting the parameters and the improvements G
proposed in this paper make this model perfect for power elec- Ipv = (Ipv,n + KI ∆T ) (3)
Gn
tronics designers who are looking for an easy and effective
model for the simulation of photovoltaic devices with power where Ipv,n [A] is the light-generated current at the nominal
converters. condition (usually 25 ◦ C and 1000W/m2), ∆T = T − Tn
Manufacturers of photovoltaic arrays, instead of the I-V (being T and Tn the actual and nominal temperatures [K]), G

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1245

[W/m2 ] is the irradiation on the device surface, and Gn is the voltage according the the practical voltage/temperature coef-
nominal irradiation. ficient. This equation simplifies the model and cancels the
The diode saturation current I0 and its dependence on the model error at the vicinities of the open-circuit voltages and
temperature may be expressed by (4) [32, 33, 35–38]: consequently at other regions of the I-V curve.
The validity of the model with this new equation has been
 3    tested through computer simulation and through comparison
Tn qEg 1 1
I0 = I0,n exp − (4) with experimental data. One interesting fact about the correc-
T ak Tn T
tion introduced with (6) is that the coefficient KV from the
where Eg is the bandgap energy of the semiconductor (Eg ≈ manufacturer’s datasheet appears in the equation. The volt-
1.12 eV for the polycrystalline Si at 25 ◦ C [11, 32]), and I0,n age/temperature coefficient KV brings important information
is the nominal saturation current: necessary to achieve the best possible I-V curve fitting for
temperatures different of the nominal value.
Isc,n
I0,n =   (5) If one wish to keep the traditional equation (4) [32, 33, 35–
Voc,n 38], instead of using (6), it is possible to obtain the best value
exp −1
aVt,n of Eg for the model so that the open-circuit voltages of the
model are matched with the open-circuit voltages of the real
with Vt,n being the thermal voltage of Ns series-connected
array in the range Tn < T < Tmax . By equaling (4) and (6)
cells at the nominal temperature Tn .
and solving for Eg at T = Tmax one gets:
The saturation current I0 of the photovoltaic cells that com-
pose the device depend on the saturation current density of the
   
semiconductor (J0 , generally given in [A/cm2 ]) and on the ef- Isc,Tmax Tn 3
fective area of the cells. The current density J0 depends on the  I0,n Tmax 
intrinsic characteristics of the photovoltaic cell, which depend Eg = − ln 
   ·

qVoc,Tmax
on several physical parameters such as the coefficient of diffu- exp −1 (7)
aNs kTmax
sion of electrons in the semiconductor, the lifetime of minority akT n Tmax
·
carriers, the intrinsic carrier density, and others [7]. This kind q (Tn − Tmax )
of information is not usually available for commercial photo-
voltaic arrays. In this paper the nominal saturation current I0,n
is indirectly obtained from the experimental data through (5), where Isc,Tmax = Isc,n + KI ∆T and Voc,Tmax = Voc,n +
which is obtained by evaluating (2) at the nominal open-circuit KV ∆T , with ∆T = Tmax − Tn .
condition, with V = Voc,n , I = 0, and Ipv ≈ Isc,n .
The value of the diode constant a may be arbitrarily chosen. D. Adjusting the model
Many authors discuss ways to estimate the correct value of this Two parameters remain unknown in (2), which are Rs and
constant [8, 11]. Usually 1 ≤ a ≤ 1.5 and the choice depends Rp . A few authors have proposed ways to mathematically de-
on other parameters of the I-V model. Some values for a termine these resistances. Although it may be useful to have
are found in [32] based on empirical analysis. As [8] says, a mathematical formula to determine these unknown parame-
there are different opinions about the best way to choose a. ters, any expression for Rs and Rp will always rely on exper-
Because a expresses the degree of ideality of the diode and it is imental data. Some authors propose varying Rs in an iterative
totally empirical, any initial value of a can be chosen in order process, incrementing Rs until the I-V curve visually fits the
to adjust the model. The value of a can be later modified in experimental data and then vary Rp in the same fashion. This
order to improve the model fitting if necessary. This constant is a quite poor and inaccurate fitting method, mainly because
affects the curvature of the I-V characteristic and varying a Rs and Rp may not be adjusted separately if a good I-V model
can slightly improve the model accuracy. is desired.
This paper proposes a method for adjusting Rs and Rp
C. Improving the model
The photovoltaic model described in the previous section based on the fact that there is an only pair {Rs ,Rp } that war-
can be improved if equation (4) is replaced by: ranties that Pmax,m = Pmax,e = Vmp Imp at the (Vmp , Imp )
point of the I-V curve, i.e. the maximum power calculated
Isc,n + KI ∆T by the I-V model of (2), Pmax,m , is equal to the maximum ex-
I0 =   (6) perimental power from the datasheet, Pmax,e , at the maximum
Voc,n + KV ∆T
exp −1 power point (MPP). Conventional modeling methods found in
aVt
the literature take care of the I-V curve but forget that the
This modification aims to match the open-circuit voltages P -V (power vs. voltage) curve must match the experimental
of the model with the experimental data for a very large range data too. Works like [26, 39] gave attention to the necessity
of temperatures. Eq. (6) is obtained from (5) by including of matching the power curve but with different or simplified
in the equation the current and voltage coefficients KV and models. In [26], for example, the series resistance of the array
KI . The saturation current I0 is strongly dependent on the model is neglected.
temperature and (6) proposes a different approach to express The relation between Rs and Rp , the only unknowns of (2),
the dependence of I0 on the temperature so that the net effect may be found by making Pmax,m = Pmax,e and solving the
of the temperature is the linear variation of the open-circuit resulting equation for Rs , as (8) and (9) show.

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1246

 250

200
V: 26.3
P: 200.1
180
200
160 MPP V: 26.3
P: 200.1

140

Pmax [W]
P [W]

150
120

100
Rs increasing 100
80

60

40 50

20

0 0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35

V [V] V [V]

Fig. 4. P -V curves plotted for different values of Rs and Rp . Fig. 5. Pmax,m vs. V for several values of Rs > 0.

V: 26.3
8 I: 7.61

 Pmax,m
7
  =   MPP
q Vmp + Rs Imp 6
= Vmp Ipv − I0 exp −1
kT aNs − (8) 5
Vmp + Rs Imp I [A]
− = Pmax,e 4
Rp
3

Rp = Vmp (Vmp  + Imp Rs )/ 

2

(Vmp + Imp Rs ) q 1
{ Vmp Ipv − Vmp I0 exp + (9)
Ns a kT 0
0 5 10 15 20 25 30 35
+Vmp I0 − Pmax,e }
V [V]
Eq. (9) means that for any value of Rs there will be a value
of Rp that makes the mathematical I-V curve cross the exper- Fig. 6. I-V curves plotted for different values of Rs and Rp .
imental (Vmp , Imp ) point.
E. Iterative solution of Rs and Rp The iterative method gives the solution Rs = 0.221 Ω for
The goal is to find the value of Rs (and hence Rp ) that the KC200GT array. Fig. 5 shows a plot of Pmax,m as a func-
makes the peak of the mathematical P -V curve coincide with tion of V for several values of Rs . There is an only point, cor-
the experimental peak power at the (Vmp , Imp ) point. This responding to a single value of Rs , that satisfies the imposed
requires several iterations until Pmax,m = Pmax,e . condition Pmax,m = Vmp Imp at the (Vmp , Imp ) point. Fig. 7
In the iterative process Rs must be slowly incremented start- shows a plot of Pmax,m as a function of Rs for I = Imp and
ing from Rs = 0. Adjusting the P -V curve to match the ex- V = Vmp . This plot shows that Rs = 0.221 Ω is the desired
perimental data requires finding the curve for several values solution, in accordance with the result of the iterative method.
of Rs and Rp . Actually plotting the curve is not necessary, as This plot may be an alternative way for graphically finding the
only the peak power value is required. Figs. 4 and 6 illustrate solution for Rs .
how this iterative process works. In Fig. 4 as Rs increases
the P -V curve moves to the left and the peak power (Pmax,m )
220
goes towards the experimental MPP. Fig. 5 shows the con-
tour drawn by the peaks of the power curves for several val- 218

ues of Rs (this example uses the parameters of the Kyocera 216

KC200GT solar array [40]). For every P -V curve of Fig. 4 214

there is a corresponding I-V curve in Fig. 6. As expected 212

Pmax [W]

from (9), all I-V curves cross the desired experimental MPP 210

point at (Vmp , Imp ). 208

Plotting the P -V and I-V curves requires solving (2) for 206
I ∈ [0, Isc,n ] and V ∈ [0, Voc,n ]. Eq. (2) does not have a 204
direct solution because I = f (V, I) and V = f (I, V ). This 202 R: 0.221
transcendental equation must be solved by a numerical method P: 200.1
200
and this imposes no difficulty. The I-V points are easily ob- 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

tained by numerically solving g(V, I) = I − f (V, I) = 0 for a Rs [Ω]

set of V values and obtaining the corresponding set of I points.
Obtaining the P -V points is straightforward. Fig. 7. Pmax = f (Rs ) with I = Imp and V = Vmp .

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1247

 8 V: 0
I: 8.21
TABLE II
7
V: 26.3
I: 7.61 Parameters of the adjusted model of the KC200GT
6
solar array at nominal operating conditions.

5 Imp 7.61 A
I [A]

4
Vmp 26.3 V
Pmax,m 200.143 W
3
Isc 8.21 A
2
Voc 32.9 V
1 I0,n 9.825 · 10−8 A
V: 32.9

0
I: 0
Ipv 8.214 A
0 5 10 15 20 25 30 35
a 1.3
V [V]
Rp 415.405 Ω
Fig. 8. I-V curve adjusted to three remarkable points. Rs 0.221 Ω

200

180
V: 26.3
P: 200.1 best model solution, so equation (10) may be introduced in the
160
model.
140
Rp + Rs
120 Ipv,n = Isc,n (10)
P [W]

Rp
100

80 Eq. (10) uses the resistances Rs and Rp to determine Ipv 6=

60
Isc . The values of Rs and Rp are initially unknown but as the
40
solution of the algorithm is refined along successive iterations
20
the values of Rs and Rp tend to the best solution and (10)
V: 0 V: 32.9

0
P: 0 P: 0
becomes valid and effectively determines the light-generated
0 5 10 15 20 25 30 35
current Ipv taking in account the influence of the series and
V [V] parallel resistances of the array. Initial guesses for Rs and Rp
are necessary before the iterative process starts. The initial
Fig. 9. P -V curve adjusted to three remarkable points.
value of Rs may be zero. The initial value of Rp may be given
by:
Figs. 8 and 9 show the I-V and P -V curves of the
Vmp Voc,n − Vmp
KC200GT photovoltaic array adjusted with the proposed Rp,min = − (11)
Isc,n − Imp Imp
method. The model curves exactly match with the experi-
mental data at the three remarkable points provided by the Eq. (11) determines the minimum value of Rp , which is
datasheet: short circuit, maximum power, and open circuit. the slope of the line segment between the short-circuit and the
Table I shows the experimental parameters of the array ob- maximum-power remarkable points. Although Rp is still un-
tained from the datasheeet and Table II shows the adjusted pa- known, it surely is greater than Rp,min and this is a good initial
rameters and model constants. guess.

F. Further improving the model G. Modeling algorithm

The model developed in the preceding sections may be fur- The simplified flowchart of the iterative modeling algorithm
ther improved by taking advantage of the iterative solution of is illustrated in Fig. 10.
Rs and Rp . Each iteration updates Rs and Rp towards the
III. VALIDATING THE MODEL
As Tables I and II and Figs. 8 and 9 have shown, the de-
TABLE I veloped model and the experimental data are exactly matched
Parameters of the KC200GT PV array at at the nominal remarkable points of the I-V curve and the ex-
25 ◦ C, AM1.5, 1000 W/m2 . perimental and mathematical maximum peak powers coincide.
Imp 7.61 A The objective of adjusting the mathematical I-V curve at the
Vmp 26.3 V three remarkable points was successfully achieved.
Pmax,e 200.143 W In order to test the validity of the model a comparison with
Isc 8.21 A other experimental data (different of the nominal remarkable
Voc 32.9 V points) is very useful. Fig. 11 shows the mathematical I-V
curves of the KC200GT solar panel plotted with the exper-
KV −0.1230 V/K
imental data at three different temperature conditions. Fig.
KI 0.0032 A/K
12 shows the I-V curves at different irradiations. The circu-
Ns 54
lar markers in the graphs represent experimental (V, I) points

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1248

 Inputs: T , G
KYOCERA KC200GT - 25 °C
I0 , eq. (4) or (6)
9
Rs = 0
1000 W/m2
Rp = Rp,min , eq. (11) 8

7 2
800 W/m

END εP max > tol

600 W/m2

I [A]
5
no
yes 4
400 W/m2
3
Ipv,n , eq. (10)
Ipv and Isc , eq. (3) 2

Rp , eq. (9) 1
Solve eq. (2) for 0 ≤ V ≤ Voc,n
Calculate P for 0 ≤ V ≤ Voc,n 0
0 5 10 15 20 25 30
Find Pmax
εP max = kPmax − Pmax,e k V [V]
Increment Rs
Fig. 12. I-V model curves and experimental data of the KC200GT
solar array at different irradiations, 25 ◦ C.
Fig. 10. Algorithm of the method used to adjust the I-V model.

KYOCERA KC200GT - 1000 W/m2

9
SOLAREX MSX60 - 1000 W/m2
8
4.5
7
75 °C 50 °C 25 °C 4
6 25 °C
3.5
I [A]

5
3 75 °C
I [A]

4
2.5

3 2

2 1.5

1 1

0 0.5
0 5 10 15 20 25 30

0
V [V] 0 5 10 15 20

V [V]
Fig. 11. I-V model curves and experimental data of the KC200GT
2
solar array at different temperatures, 1000 W/m . Fig. 13. I-V model curves and experimental data of the MSX60
solar array at different temperatures, 1000 W/m2 .

extracted from the datasheet. Some points are not exactly

matched because the model is not perfect, although it is ex-
act at the remarkable points and sufficiently accurate for other SOLAREX MSX60 - 1000 W/m2
60
points. The model accuracy may be slightly improved by run-
25 °C
ning more iterations with other values of the constant a, with-
50
out modifications in the algorithm.
Fig. 13 shows the mathematical I-V curves of the Solarex 40 75 °C
MSX60 solar panel [41] plotted with the experimental data
P [W]

at two different temperature conditions. Fig. 14 shows the 30

P -V curves obtained at the two temperatures. The circular
markers in the graphs represent experimental (V, I) and (V, P ) 20

points extracted from the datasheet. Fig. 14 proves that the

model accurately matches with the experimental data both in 10

the current and power curves, as expected.

0
0 5 10 15 20
IV. SIMULATION OF THE PHOTOVOLTAIC ARRAY
V [V]
The photovoltaic array can be simulated with an equivalent
circuit model based on the photovoltaic model of Fig. 1. Two Fig. 14. P -V model curves and experimental data of the MSX60
simulation strategies are possible. solar array at different temperatures, 1000 W/m2 .
Fig. 15 shows a circuit model using one current source

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1249

 I I
Rs
+ -

I + V
Im Rp + -
V -

V
V I
    Ipv Numerical solution of eq. (2)
V + Rs I
Ipv − I0 exp −1
Vt a I0
Fig. 16. Photovoltaic array model circuit with a controlled current
source and a computational block that solves the I-V equation.
Fig. 15. Photovoltaic array model circuit with a controlled current
source, equivalent resistors and the equation of the model current
(Im ).

(Im ) and two resistors (Rs and Rp ). This circuit can be im-
plemented with any circuit simulator. The value of the model
current Im is calculated by the computational block that has
V , I, I0 and Ipv as inputs. I0 is obtained from (4) or (6) and
Ivp is obtained from (3). This computational block may be
implemented with any circuit simulator able to evaluate math
functions.
Fig. 16 shows another circuit model composed of only one
current source. The value of the current is obtained by numer-
ically solving the I-V equation. For every value of V a corre-
sponding I that satisfies the I-V equation (2) is obtained. The
solution of (2) can be implemented with a numerical method
in any circuit simulator that accepts embedded programming.
This is the simulation strategy proposed in [42].
Other authors have proposed circuits for simulating photo-
voltaic arrays that are based on simplified equations and/or
require lots of computational effort [12, 26, 27, 43]. In [12]
a circuit-based photovoltaic model is composed of a current
source driven by an intricate and inaccurate equation where
the parallel resistance is neglected. In [26] an intricate PSpice-
based simulation was presented, where the I-V equation is
numerically solved within the PSpice software. Although in-
teresting, the approach found in [26] is excessively elaborated
and concerns the simplified photovoltaic model without the
series resistance. In [27] a simple circuit-based photovoltaic
model is proposed where the parallel resistance is neglected.
In [43] a circuit-based model was proposed based on the piece-
wise approximation of the I-V curve. Although interesting
and relatively simple, this method [43] does not provide a so-
lution to find the parameters of the I-V equation and the circuit
model requires many components.
Figs. 17 and 18 show the photovoltaic model circuits im-
plemented with MATLAB/SIMULINK (using the SymPow-
erSystems blockset) and PSIM using the simulation strategy Fig. 17. Photovoltaic circuit model built with
of Fig. 15. Both circuit models work perfectly and may be MATLAB/SIMULINK.
used in the simulation of power electronics converters for pho-
tovoltaic systems. Figs. 19 and 20 show the I-V curves of the experimental remarkable points of the I-V curve of the
the Solarex MSX60 solar panel [41] simulated with the MAT- practical array. The method obtains the parameters of the I-
LAB/SIMULINK and PSIM circuits. V equation by using the following nominal information from
V. CONCLUSION the array datasheet: open-circuit voltage, short-circuit current,
maximum output power, voltage and current at the maximum
This paper has analyzed the development of a method for power point, current/temperature and voltage/temperature co-
the mathematical modeling of photovoltaic arrays. The objec- efficients. This paper has proposed an effective and straight-
tive of the method is to fit the mathematical I-V equation to forward method to fit the mathematical I-V curve to the three

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1250

 (V, I) remarkable points without the need to guess or to esti-
mate any other parameters except the diode constant a. This
paper has proposed a closed solution for the problem of finding
the parameters of the single-diode model equation of a prac-
tical photovoltaic array. Other authors have tried to propose
single-diode models and methods for estimating the model pa-
rameters, but these methods always require visually fitting the
mathematical curve to the I-V points and/or graphically ex-
tracting the slope of the I-V curve at a given point and/or suc-
cessively solving and adjusting the model in a trial and error
process. Some authors have proposed indirect methods to ad-
just the I-V curve through artificial intelligence [15, 44–46]
and interpolation techniques [25]. Although interesting, such
methods are not very practical and are unnecessarily compli-
cated and require more computational effort than it would be
expected for this problem. Moreover, frequently in these mod-
els Rs and Rp are neglected or treated as independent parame-
ters, which is not true if one wish to correctly adjust the model
so that the maximum power of the model is equal to the maxi-
mum power of the practical array.
An equation to express the dependence of the diode satura-
tion current I0 on the temperature was proposed and used in
the model. The results obtained in the modeling of two prac-
tical photovoltaic arrays have demonstrated that the equation
Fig. 18. Photovoltaic circuit model built with PSIM. is effective and permits to exactly adjust the I-V curve at the
open-circuit voltages at temperatures different of the nominal.
Moreover, the assumption Ipv ≈ Isc used in most of pre-
4
vious works on photovoltaic modeling was replaced in this
3.5 method by a relation between Ipv and Isc based on the series
and parallel resistances. The proposed iterative method for
3
solving the unknown parameters of the I-V equation allows to
2.5 determine the value of Ipv , which is different of Isc .
This paper has presented in details the equations that con-
2
stitute the single-diode photovoltaic I-V model and the algo-
1.5 rithm necessary to obtain the parameters of the equation. In or-
25°C,1000W/m2
der to show the practical use of the proposed modeling method
1
25°C,800W/m2 this paper has presented two circuit models that can be used to
2
0.5 75°C,1000W/m
2
simulate photovoltaic arrays with circuit simulators.
75°C,800W/m

0
This paper provides the reader with all necessary informa-
0 5 10 15 20
tion to easily develop a single-diode photovoltaic array model
for analyzing and simulating a photovoltaic array. Programs
Fig. 19. I-V curves of the model simulated with
and ready-to-use circuit models are available for download at:
MATLAB/SIMULINK.
http://sites.google.com/site/mvillalva/pvmodel.

APPENDIX - ASSOCIATION OF PV ARRAYS

In the previous sections this paper has dealt with the model-
ing and simulation of photovoltaic arrays that are single panels
or modules composed of several interconnected basic photo-
voltaic cells. Large arrays composed of several panels may be
modeled in the same way, provided that the equivalent param-
eters (short-circuit current, open-circuit voltage) are properly
inserted in the modeling process. As a result, the equivalent
parameters (resistances, currents, etc) of the association are
obtained. Generally experimental data are available only for
commercial low-power modules and this is the reason why this
paper has chosen to deal with small arrays. The term array has
Fig. 20. I-V curves of the model simulated with PSIM. been used throughout this paper to mean small commercial
photovoltaic modules or panels. This appendix shows how to

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1251

 I I
Rs Nser /Npar
+ -
(a)
Rs Nser
Im Rp Nser /Npar +
- V

...
...

Ipv Id Rp Nser V V I
     
Nser
V + Rs I
  Npar   Ipv
Ipv Npar − I0 Npar  
exp 
 − 1
 
Vt aNser I0
I
(b)
Rs /Npar
Id Npar Fig. 22. Model circuit of array composed of Nser x Npar modules.
... ... Rp /Npar V
Ipv Npar
where Ipv , I0 , Rs , Rp , and Vt are parameters of individual
modules. Please notice that Ns (the number of elementary
series cells of the individual photovoltaic module) used in the
I preceding sections is different of Nser .
The photovoltaic array can be simulated with the modified
(c)
Rs Nser /Npar I-V equation (12). The simulation circuit presented previ-
ously must be modified according to the number of associated
modules. The Nser and Npar indexes must be inserted in the
...

...

...

...

... ... Rp Nser /Npar V

Ipv Npar Id Npar model, respecting equation (12), as shown in Fig. 22.

REFERENCES
Fig. 21. (a) Array of Nser modules connected in series, (b) Array of [1] H. S. Rauschenbach. Solar cell array design handbook.
Npar modules connected in parallel, and (c) array composed of Nser
Van Nostrand Reinhold, 1980.
x Npar modules.
[2] J. A. Gow and C. D. Manning. Development of a photo-
voltaic array model for use in power-electronics simula-
model and simulate large arrays composed of several series or tion studies. Electric Power Applications, IEE Proceed-
parallel modules. Here, unlike in the rest of this paper, module ings, 146(2):193–200, 1999.
and array have distinct meanings. [3] J. A. Gow and C. D. Manning. Development of a model
Fig. 21a shows an array composed of several identic mod- for photovoltaic arrays suitable for use in simulation stud-
ules connected in series. The output voltage is increased pro- ies of solar energy conversion systems. In Proc. 6th Inter-
portionately to Nser and the currents remain unchanged. The national Conference on Power Electronics and Variable
equivalent series and parallel resistances are directly propor- Speed Drives, p. 69–74, 1996.
tional to the number of modules. [4] N. Pongratananukul and T. Kasparis. Tool for automated
Fig. 21b shows an array composed of several identic mod- simulation of solar arrays using general-purpose simula-
ules connected in parallel. The association provides increased tors. In Proc. IEEE Workshop on Computers in Power
output current and the output voltage is unchanged. The equiv- Electronics, p. 10–14, 2004.
alent series and parallel resistances are inversely proportional [5] S. Chowdhury, G. A. Taylor, S. P. Chowdhury, A. K.
to the number of parallel modules. Saha, and Y. H. Song. Modelling, simulation and per-
Fig. 21c shows a photovoltaic array composed of several formance analysis of a PV array in an embedded envi-
modules connected in series and parallel. In this case the ronment. In Proc. 42nd International Universities Power
equivalent resistances depend on the number of series and par- Engineering Conference, UPEC, p. 781–785, 2007.
allel connections. [6] J. Hyvarinen and J. Karila. New analysis method for
For any given array formed by Nser x Npar identic modules crystalline silicon cells. In Proc. 3rd World Conference
the following equivalent I-V equation is valid: on Photovoltaic Energy Conversion, v. 2, p. 1521–1524,
2003.
      [7] Kensuke Nishioka, Nobuhiro Sakitani, Yukiharu Uraoka,
Nser and Takashi Fuyuki. Analysis of multicrystalline sil-
V + Rs I
  Npar  
I = Ipv Npar − I0 Npar  
exp 
 − 1 −
 
icon solar cells by modified 3-diode equivalent circuit
Vt aNser model taking leakage current through periphery into con-
  sideration. Solar Energy Materials and Solar Cells,
Nser 91(13):1222–1227, 2007.
V + Rs I
Npar [8] C. Carrero, J. Amador, and S. Arnaltes. A single proce-
−  
Nser dure for helping PV designers to select silicon PV mod-
Rp
Npar ule and evaluate the loss resistances. Renewable Energy,
(12) 2007.

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1252

 [9] E. Koutroulis, K. Kalaitzakis, and V. Tzitzilonis. Devel- [23] Wang Yi-Bo, Wu Chun-Sheng, Liao Hua, and Xu Hong-
opment of a FPGA-based system for real-time simula- Hua. Steady-state model and power flow analysis of grid-
tion of photovoltaic modules. Microelectronics Journal, connected photovoltaic power system. In Proc. IEEE In-
2008. ternational Conference on Industrial Technology, ICIT,
[10] G. E. Ahmad, H. M. S. Hussein, and H. H. El-Ghetany. p. 1–6, 2008.
Theoretical analysis and experimental verification of PV [24] France Lasnier and Tony Gan Ang. Photovoltaic engi-
modules. Renewable Energy, 28(8):1159–1168, 2003. neering handbook. Adam Hilger, 1990.
[11] Geoff Walker. Evaluating MPPT converter topologies us- [25] K. Khouzam, C. Khoon Ly, C.and Koh, and Poo Yong
ing a matlab PV model. Journal of Electrical & Electron- Ng. Simulation and real-time modelling of space photo-
ics Engineering, Australia, 21(1), 2001. voltaic systems. In IEEE 1st World Conference on Pho-
[12] M. Veerachary. PSIM circuit-oriented simulator model tovoltaic Energy Conversion, Conference Record of the
for the nonlinear photovoltaic sources. IEEE Transac- 24th IEEE Photovoltaic Specialists Conference, v. 2, p.
tions on Aerospace and Electronic Systems, 42(2):735– 2038–2041, 1994.
740, April 2006. [26] M. C. Glass. Improved solar array power point model
[13] Ali Naci Celik and NasIr Acikgoz. Modelling and ex- with SPICE realization. In Proc. 31st Intersociety En-
perimental verification of the operating current of mono- ergy Conversion Engineering Conference, IECEC, v. 1,
crystalline photovoltaic modules using four- and five- p. 286–291, August 1996.
parameter models. Applied Energy, 84(1):1–15, January [27] I. H. Altas and A. M. Sharaf. A photovoltaic array
2007. simulation model for matlab-simulink GUI environment.
[14] Yeong-Chau Kuo, Tsorng-Juu Liang, and Jiann-Fuh In Proc. International Conference on Clean Electrical
Chen. Novel maximum-power-point-tracking controller Power ,ICCEP, p. 341–345, 2007.
for photovoltaic energy conversion system. IEEE Trans- [28] E. Matagne, R. Chenni, and R. El Bachtiri. A photo-
actions on Industrial Electronics, 48(3):594–601, June voltaic cell model based on nominal data only. In Proc.
2001. International Conference on Power Engineering, Energy
[15] M. T. Elhagry, A. A. T. Elkousy, M. B. Saleh, T. F. Elshat- and Electrical Drives, POWERENG, p. 562–565, 2007.
ter, and E. M. Abou-Elzahab. Fuzzy modeling of photo- [29] Yun Tiam Tan, D. S. Kirschen, and N. Jenkins. A
voltaic panel equivalent circuit. In Proc. 40th Midwest model of PV generation suitable for stability analysis.
Symposium on Circuits and Systems, v. 1, p. 60–63, Au- IEEE Transactions on Energy Conversion, 19(4):748–
gust 1997. 755, 2004.
[16] Shengyi Liu and R. A. Dougal. Dynamic multiphysics [30] A. Kajihara and A. T. Harakawa. Model of photovoltaic
model for solar array. IEEE Transactions on Energy Con- cell circuits under partial shading. In Proc. IEEE Inter-
version, 17(2):285–294, 2002. national Conference on Industrial Technology, ICIT, p.
[17] Weidong Xiao, W. G. Dunford, and A. Capel. A novel 866–870, 2005.
modeling method for photovoltaic cells. In Proc. IEEE [31] N. D. Benavides and P. L. Chapman. Modeling the ef-
35th Annual Power Electronics Specialists Conference, fect of voltage ripple on the power output of photovoltaic
PESC, v. 3, p. 1950–1956, 2004. modules. IEEE Transactions on Industrial Electronics,
[18] Y. Yusof, S. H. Sayuti, M. Abdul Latif, and M. Z. C. 55(7):2638–2643, 2008.
Wanik. Modeling and simulation of maximum power [32] W. De Soto, S. A. Klein, and W. A. Beckman. Improve-
point tracker for photovoltaic system. In Proc. National ment and validation of a model for photovoltaic array per-
Power and Energy Conference, PECon, p. 88–93, 2004. formance. Solar Energy, 80(1):78–88, January 2006.
[19] D. Sera, R. Teodorescu, and P. Rodriguez. PV panel [33] Q. Kou, S. A. Klein, and W. A. Beckman. A method for
model based on datasheet values. In Proc. IEEE Inter- estimating the long-term performance of direct-coupled
national Symposium on Industrial Electronics, ISIE, p. PV pumping systems. Solar Energy, 64(1-3):33–40,
2392–2396, 2007. September 1998.
[20] M. A. Vitorino, L. V. Hartmann, A. M. N. Lima, and [34] A. Driesse, S. Harrison, and P. Jain. Evaluating the ef-
M. B. R. Correa. Using the model of the solar cell for fectiveness of maximum power point tracking methods
determining the maximum power point of photovoltaic in photovoltaic power systems using array performance
systems. In Proc. European Conference on Power Elec- models. In Proc. IEEE Power Electronics Specialists
tronics and Applications, p. 1–10, 2007. Conference, PESC, p. 145–151, 2007.
[21] D. Dondi, D. Brunelli, L. Benini, P. Pavan, A. Bertac- [35] R. A. Messenger and J. Ventre. Photovoltaic systems en-
chini, and L. Larcher. Photovoltaic cell modeling for so- gineering. CRC Press, 2004.
lar energy powered sensor networks. In Proc. 2nd Inter- [36] F. Nakanishi, T. Ikegami, K. Ebihara, S. Kuriyama, and
national Workshop on Advances in Sensors and Interface, Y. Shiota. Modeling and operation of a 10 kW photo-
IWASI, p. 1–6, 2007. voltaic power generator using equivalent electric circuit
[22] H. Patel and V. Agarwal. MATLAB-based modeling to method. In Conference Record of the 28th IEEE Photo-
study the effects of partial shading on PV array char- voltaic Specialists Conference, p. 1703–1706, September
acteristics. IEEE Transactions on Energy Conversion, 2000.
23(1):302–310, 2008. [37] J. Crispim, M. Carreira, and R. Castro. Validation of

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1253

 photovoltaic electrical models against manufacturers data [43] R. C. Campbell. A circuit-based photovoltaic array model
and experimental results. In Proc. International Con- for power system studies. In Proc. 39th North American
ference on Power Engineering, Energy and Electrical Power Symposium, NAPS, p. 97–101, 2007.
Drives, POWERENG, p. 556–561, 2007. [44] Th. F. Elshatter, M. T. Elhagry, E. M. Abou-Elzahab,
[38] K.H. Hussein, I. Muta, T. Hoshino, and M. Osakada. and A. A. T. Elkousy. Fuzzy modeling of photovoltaic
Maximum photovoltaic power tracking: an algorithm for panel equivalent circuit. In Conference Record of the
rapidly changing atmospheric conditions. In Generation, 28th IEEE Photovoltaic Specialists Conference, p. 1656–
Transmission and Distribution, IEE Proceedings-, v. 142, 1659, 2000.
p. 59–64, January 1995. [45] M. Balzani and A. Reatti. Neural network based model of
[39] E. I. Ortiz-Rivera and F. Z. Peng. Analytical model for a PV array for the optimum performance of PV system. In
a photovoltaic module using the electrical characteristics Proc. PhD Research in Microelectronics and Electronics,
provided by the manufacturer data sheet. In Proc. IEEE v. 2, p. 123–126, 2005.
36th Power Electronics Specialists Conference, PESC, p. [46] H. Mekki, A. Mellit, H. Salhi, and B. Khaled. Model-
2087–2091, 2005. ing and simulation of photovoltaic panel based on artifi-
[40] KC200GT high efficiency multicrystal photovoltaic mod- cial neural networks and VHDL-language. In Proc. 14th
ule datasheet. IEEE International Conference on Electronics, Circuits
[41] Solarex MSX60 and MSX64 solar arrays datasheet, and Systems, ICECS, p. 58–61, 2007.
1998.
[42] M. M. Casaro and D. C. Martins. Modelo de arranjo fo-
tovoltaico destinado a análises em eletrônica de potência.
Revista Eletrônica de Potência, Sociedade Brasileira
de Eletrônica de Potência (SOBRAEP), 13(3):141–146,
2008.

978-1-4244-3370-4/09/$25.00 © 2009 IEEE 1254