5th International Conference on Modern Power Systems MPS 2013, 28-31 May 2013, Cluj-Napoca, România

Study on Temperature for Modeling of Photovoltaic

Solar Array using Experimental Test Data

Ioan Viorel BANU, Răzvan BENIUGĂ, Marcel ISTRATE

“Gheorghe Asachi” Technical University of Iasi
Iasi, Romania
[email protected], [email protected], [email protected]

Abstract—The paper presents a study on temperature for II. MODELING PV CELLS, PANELS AND ARRAYS
modeling of photovoltaic (PV) solar array using experimental
data in Matlab/Simulink program. Starting from the A. Modeling PV Cells
mathematical model is realized the PV source model for different This subsection contrasts the fundamental approaches that
temperatures, and finally are presented some generated fit can be used for modeling individual PV cells as well as strings
results of the three-dimensional current–voltage (I–V) surface of and arrays.
PV array from Curve Fitting Tool. This paper is useful in the
modeling of PV solar arrays. The PV cells are modeled using a particular type of
equivalent circuit to deploy the equation of mathematical
Keywords—photovoltaic array; temperature; modeling; model. Any PV model relies on diode behavior that gives to
simulation; PV cell its exponential characteristic [8]. In the literature, there
are several mathematical models that describe the operation
I. INTRODUCTION
and behavior of the photovoltaic generator. These models
Considering the actual energy and environmental context of differ in the procedure of calculation, accuracy and the number
the classical power resources around the world, the of parameters involved in the calculation of current–voltage
photovoltaic (PV) domain provide one of the most efficient characteristic [9].
ways of producing energy, with real perspectives in the future
[1, 2]. In 2011, one of the dominant features of the landscape of General equation of output current of solar cell according to
renewable energy was framing in technological costs. The Simulink documentation [10] has the form presented in (1):
prices of photovoltaic modules fell by almost 50%. These
changes have brought this important renewable energy  V NIVRs   VNIVRs  V  I  R
technology close to a competitive alternative to fossil fuels    
I I ph I s e t
 
1 I s 2  e 2 t  1 
  s  

such as coal and gas [3].     Rp

   
Many studies have been done on analysis of the effect of
temperature on the performance of PV modules [4]. Since the where: I ph is the solar-induced current: I ph  I ph 0  I r I r 0 ,
environmental conditions in that the PV cells are tested are where I r is the irradiance (light intensity) in W m 2 falling on
variable, being very difficult to have the same level of solar
radiation and temperature, is needed a substantial input for the the cell, I ph 0 is the measured solar-generated current for the
PV solar array [5]. Simulation is an effective and efficient way irradiance I r 0 ; I s and I s 2 are the saturation currents of the
to analyze the performance of PV systems [6]. Developing a
model of a PV array for simulation of the PV system is of first and second diode; N1 and N 2 are the quality factor
fundamental importance because the PV generator is the main (emission coefficient) of the first and second diode; Vt  kT q
component of a PV system [7]. is the thermal voltage, ( k is the Boltzmann constant, T is
The paper covered basically tree topics. The first section device temperature); Rs and R p are the series and parallel
contain general description regarding modeling components (shunt) resistance; V is the voltage at the terminals of PV cell.
such as PV solar cells, PV panels and combines into array and
about how the PV solar array can be modeled using The Solar Cell block from Simulink allows choosing one of
experimental data through Curve Fitting Toolbox™ from two models: a model with 8 parameters and a model with 5
Matlab®/Simulink®. In the second section are described the parameters if for (1) is applied the following simplifying
results and discussions for some generated fit of the current– assumptions: the impedance of the parallel resistor is infinite
voltage (I–V) surface of PV solar array source at different and the saturation current of the second diode is zero [10]. The
temperature levels from Curve Fitting Tool. Finally, in the last 5 parameters model provide short-circuit current and open-
section of paper are presented the conclusions drawn from this circuit voltage that the block converts to an equivalent circuit
study. model of the PV cell.

51
 5th International Conference on Modern Power Systems MPS 2013, 28-31 May 2013, Cluj-Napoca, România

The most used PV cell model is a 5 parameter model data obtained at a temperature of 25°C for different values of
(single–diode model), that provides a good compromise irradiance (400, 600, 800, 1000 and 1200 W/m2). Curve Fitting
between accuracy and simplicity [11], whose diagram is Tools allows very fast creation of mathematical curves for each
depicted in Fig. 1, formed from a series resistance Rs (models of these I–V curves, which represent the experimental behavior
bulk and contact resistances) connected in series with a parallel of the PV array [8].
combination of a current source, an exponential diodes and a Using a simple interpolation algorithm, like cubic
parallel resistance R p (models leakage currents, primarily due interpolation is generated a surface whose mathematical
to defects) [8, 10]. equation represents all points on the I–V curves and all points
between those curves. Afterwards is created a voltage vectors
and an irradiance vector and using that irradiance vector and
that voltage vector on that surface object is calculated what the
current values are for the PV array. Finally, is bringing up a
Look-up Table block from Simulink and automatically
populating that block with the array voltage values, irradiance
values that is already defined and the table data of the array
current that is calculated from the surface fit. This block
represents the PV solar array model and has two inputs: the
Fig. 1. The equivalent circuit of a solar cell [8]. irradiance and the voltage, and in the output is the current
value. This model can be use very quickly as a source for a PV
The solar-induced current I ph , diodes saturation currents solar array.
I s and I s 2 , the series resistance Rs and the parallel resistance III. RESULTS AND DISCUSSIONS
R p depend on temperature [10].
This section shows some results of generated fit of the
current–voltage surface of PV solar array operating at different
B. Modeling using experimental test data
temperatures. Since the I–V three-dimensional surface model
This subsection shows how to use experimental test data in of PV solar array is changed according to solar radiation, are
the form of typical curves, to generate a mathematical model created sources for PV solar array simulator at different
for a PV panel or array by import test data from Excel® files temperatures.
and some optimization and surface fitting capabilities in
Matlab. The data sets of I–V curves that are used in this study in
Curve Fitting Tool are obtained through simulation in
The experimental data of PV array can be introduced in Matlab/Simulink of a PV array with parameters given in Table
Matlab with tools that allow to do curve fitting in these data I, that is formed by 36 series solar cells from Simscape™
points to create a three-dimensional surface. Matlab allows the library with temperature parameters presented in Table II, for
capability to do directly mathematical representation to create a five levels of solar radiation: 400, 600, 800, 1000, 1200 W/m2,
new block in Simulink that represents directly the model of PV at following temperatures: –50, –25, –16, –15.9999, –15, 0, 25,
array [8] as is depicted in Fig. 2. 50, 70°C.
The sources for PV solar array are realized in Matlab with
help of the Curve Fitting Toolbox as was described in the
previous section. To generate a surface fit for this data was
used a custom equation with general model presented in (2),
calculated coefficients with 95 confidence bounds presented in
(3) and goodness of fit presented in (4). It was used a custom
equation because it is known the shape of this curves and it is
knows that the equation of solar generated current of PV solar
Fig. 2. Model using test equipment data [2, 8]. array have some kind of exponential nature plus a constant
term.
Using Curve Fitting Tool from Matlab is implemented a
fixed predictive model based on the I–V characteristic curve of TABLE I. PV ARRAY MODEL PARAMETERS
PV panel by creating a three-dimensional surface for these data
points. This surface is then used to generate the current array Table PV array parameters
and the voltage and irradiance vectors necessary to configure Head Parameter/ Value Description
the 2D Lookup Table block in Simulink. Curve Fitting Tools is 1 Isc=7.34 [A] Short-circuit current
a powerful algorithm that affords the achievement of
polynomial interpolation, or can run a custom equation, which 2 Voc=21.6 [V] Open-circuit voltage for entire PV Panel
in this case is very close to the one that is accomplished, 3 Rs=0 [Ω] Series resistance for entire PV Panel
because it is known the exponential form of I–V curves.
4 Rp=Inf [Ω] Parallel resistance for entire PV Panel
To create a PV solar array simulator is imported from Excel
five I–V experimental curves, corresponding to experimental

52
 5th International Conference on Modern Power Systems MPS 2013, 28-31 May 2013, Cluj-Napoca, România

TABLE II. SOLAR CELL TEMPERATURE BLOCK PARAMETERS

Table Temperature block parameters for a single solar cell

Head Parameter/ Value Description
1 TIPH1=0 [1/K] First order temperature coefficient for Iph
Temperature exponent for Is. The order
of the exponential increase in current
2 TXIS1=3 from the first diode as temperature
increases. The default value is 3. The
value must be greater than or equal to 0.
3 TRS1=0 Temperature exponent for Rs

4 TRP1=0 Temperature exponent for Rp

Extraction temperature parameter.
Temperature at which the parameters of
5 Tmeas=25 [°C] solar cell were measured. The value
must be greater than 0. The default Fig. 4. The surface fit of PV panel for 25°C.
value is 25°C.
Fixed circuit temperature. The
temperature at which is simulate the
6 TFIXED=25[°C]
solar cell. The default value is 25°C.
The value must be greater than 0.

 f(x, y) = a  y - b  exp((m  x) - 1) - (x + c)/d  

a = 0.007345 (0.007344, 0.007345)

b = 3.272e - 06 (3.254e - 06, 3.291e - 06)

 c = 0.4526 (-0.9534, 1.859)  
d = 1820 (1652, 1987)

m = 0.7233 (0.723, 0.7236)
Fig. 5. The surface fit of PV panel for –15.9999°C.
SSE : 0.00403
R - square : 1

   
Adjusted R - square : 1
RMSE : 0.002509

In Fig. 3–6 are presented the three-dimensional I–V

surfaces, generated in Curve Fitting Tool using custom
equation (2), for PV solar array operating at temperatures of
70, 25, –15.9999 and –25°C respectively.

Fig. 6. The surface fit of PV panel for –25°C.

The surface fit of PV array operating at –25°C generated

directly in Curve Fitting Tool using cubic interpolation for the
same data as for Fig. 6 is given in Fig. 7.
It is noticed that for implementation of the PV array source
using custom equation (2) in Curve Fitting Tool at
temperatures lower than –16°C the surface fit is wrong realized
as can be seen in Fig. 6, where are given the surface fit of PV
panel traced for –25°C. If instead (2) in Curve Fitting Tool is
Fig. 3. The surface fit of PV panel for 70°C. used a cubic interpolation, the surface fit of PV solar array at
operating temperature lower than –16°C is traced properly as
show in Fig. 7.

53
 5th International Conference on Modern Power Systems MPS 2013, 28-31 May 2013, Cluj-Napoca, România

System Identification Toolbox™ or Neural Network

Toolbox™ from Matlab, which are very good at creating
dynamic systems.
ACKNOWLEDGMENT
Răzvan Beniugă thanks to the project Cuantumdoc
“Doctoral Studies for European performances in research and
innovation” (ID-79407), financed by the European Social Fund
and the Romanian Government.
REFERENCES
[1] ABB, “Technical Application Papers No. 10 Photovoltaic plants,” 2010.
[2] I. V. Banu, M. Istrate, “Modeling of Maximum Power Point Tracking
Algorithm for Photovoltaic Systems,” Proceedings of the International
Conference and Exposition on Electrical and Power Engineering, 2012,
Fig. 7. The surface fit of PV panel for –25°C using cubic interpolation. pp. 953-954.
[3] UNEP, “Global Trends in Renewable Energy Investment 2012,” 2012.
IV. CONCLUSIONS [4] W. Zhou, H. Yang, Z. Fang, “A novel model for photovoltaic array
This paper presents a study on temperature in case of performance prediction,” Applied Energy 84, 2007, pp. 1187–1198.
modeling of PV solar array sources using experimental test [5] S. Zekai, “Solar Energy Fundamentals and Modelling Techniques,”
Springer, Ltd., 2008.
data (PV simulator). For implementation of the PV array
[6] J.A. Gow, C.D. Manning. “Development of a Photovoltaic Array Model
source at temperatures lower than –16°C with help of Curve for Use in Power-Electronics Simulation Studies,” IEE Proceedings of
Fitting Tool using custom equation (2) will never obtained Electric Power Applications, vol. 146, no. 2, 1999, pp. 193-200.
correct surface fit of PV solar array source, involving obtaining [7] G. Liu, Y. Ren, S.K. Nguang, “Simulated Photovoltaic Array Systems
completely wrong data at the output of PV solar array against under a Changing Environment for Temporal Performance,”
of case in that is used cubic interpolation. This involves that International Journal of Green Energy, vol. 9, no. 7, 2012, pp. 673-684.
when is desired modeling of PV solar array using experimental [8] *** www.mathworks.com, C. Osorio, “Recorded Webinar- Model-
test data at temperatures lower than –16°C to use another Based Design for Solar Power Systems,” 2009.
algorithm or custom equation model. [9] D. Rekioua, E. Matagne, “Optimization of Photovoltaic Power Systems,
Modelization, Simulation and Control,” Springer, 2012.
Because the I–V behavior is static, the proposed study use [10] *** www.mathworks.com, “Help - Solar Cell Blocks,” 2011.
Curve Fitting Toolbox to create a stable fit, but for dynamic [11] C. Carrero, J. Amador, S. Arnaltes, “A single procedure for helping PV
systems are used experimental data, and for this case is used a designers to select silicon PV module and evaluate the loss resistances,”
predictive models for PV solar array. Future directions of study Renewable Energy, vol. 32, issue 15, 2007, pp. 2579–2589.
could be creating predictive models for PV solar array using

54