Solar Energy 155 (2017) 647–653

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Explicit empirical model for photovoltaic devices. Experimental

validation
A. Massi Pavan a,e,⇑, S. Vergura b, A. Mellit c,d, V. Lughi a
a
Department of Engineering and Architecture, University of Trieste, Via A. Valerio, 6/A, 34127 Trieste, Italy
b
Department of Electrical and Information Engineering, Politecnico di Bari, 70125 Bari, Italy
c
Faculty of Sciences and Technologies, Renewable Energy Laboratory, Jijel University, Jijel 18000, Algeria
d
Abdus Salam International Centre for Theoretical Physics, Strada Costiera, 11, 34151 Trieste, Italy
e
Today with The School of Electrical and Electronic Engineering, The University of Manchester, Manchester, UK

a r t i c l e i n f o a b s t r a c t

Article history: A comparison between the experimental current-voltage (I-V) and power-voltage (P-V) characteristics of
Received 16 December 2016 PhotoVoltaic (PV) modules, and the prediction of an explicit empirical model has been carried out. The
Received in revised form 22 May 2017 model consists of an explicit expression for the current as a function of the voltage; the only inputs
Accepted 3 July 2017
are the parameters that are always directly available in the manufacturer’s datasheet. The comparison
Available online 8 July 2017
was carried out on four representative PV technologies, based on polycrystalline Si, Heterojunction with
Intrinsic Thin layer (HIT), Copper Indium Gallium Selenide (CIGS), and Cadmium Telluride (CdTe). The
Keywords:
comparison reveals that the model replicates the experimental I-V and P-V curves to a very good degree
CdTe
CIGS
of accuracy for the considered operating conditions and PV technologies. This validation sets a turning
HIT point in PV modelling, as it enables a reliable use of this accessible model.
I-V and P-V characteristics Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction and their electrical output in terms of current and power-voltage

characteristics are slightly different.
Nowadays, the main commercial Photovoltaic (PV) technology Effective use of PV modules requires reliable modelling meth-
is based on crystalline silicon. These solar devices represent the ods, aiming at predicting the behaviour of a PV system at condi-
first generation of photovoltaics and cover the 90% of the market. tions different from those characterized by the manufacturer’s
The remainder of the market is covered by thin film technologies datasheet. Such methods are helpful for monitoring the perfor-
mainly based on CdTe, CIGS, and amorphous silicon. These mance (Vergura et al., 2009; Cristaldi et al., 2012, 2015;
products, owning to the second generation of photovoltaics and d’Alessandro et al., 2015) and the losses in solar systems (Massi
characterized by a slightly lower efficiency than the devices from Pavan et al., 2013; Moballegh and Jiang, 2014; Massi Pavan et al.,
the first one, are today entering the market especially because of 2015; Spertino et al., 2015), for forecasting the produced power
their lower manufacturing cost and continuous increase in perfor- (Bouzerdoum et al., 2013; Bizzarri et al., 2013; Dolara et al.,
mances (Lineykin et al., 2014). In the future, a third generation of 2015; Dellino et al., 2015; Chicco et al., 2016), and for development
photovoltaics should commercially guarantee higher efficiencies and testing of maximum power point tracking algorithms
and lower costs. Dye Sensitized Solar Cells (DSSC), Organic PV (Manganiello et al., 2014; Boztepe et al., 2014;
(OPV), Intermediate electronic Band (IB) and Multiple Exciton Seyedmahmoudian et al., 2015). Reliable models are also needed
Generation (MEG) are only some examples of third generation for system fault diagnosis (Chine et al., 2014, 2016) and to study
devices that today are either not commercially available or have and evaluate the behaviour of defective PV cells. Description of
a very small market (Choubey et al., 2012). The first and second known defects in PV cells is reported in Breitenstein et al. (2004,
generation technologies are based on different physical 2001), Acciani et al. (2010), while in-depth investigations of the
mechanisms, come from a wide range of fabrication techniques, thermal effects of defects are proposed in Vergura et al. (2012,
2009a,b) where a finite element approach to model some classes
of defects commonly found in PV cells is presented.
Equivalent circuits, including a photocurrent source, one or
⇑ Corresponding author at: Department of Engineering and Architecture, Univer-
more resistors, and one or more nonlinear elements typically rep-
sity of Trieste, Via A. Valerio, 6/A, 34127 Trieste, Italy.
E-mail address: [email protected] (A. Massi Pavan).
resented by semiconductor diodes, are the most common topology

http://dx.doi.org/10.1016/j.solener.2017.07.002
0038-092X/Ó 2017 Elsevier Ltd. All rights reserved.
648 A. Massi Pavan et al. / Solar Energy 155 (2017) 647–653

for modelling crystalline Si PV devices (Duffie and Beckman, 1991). Validation of this model along the entire I-V and P-V character-
A widely used equivalent circuit is the ‘‘single-diode” model – istics is therefore of paramount importance for ensuring that this
often referred to as the ‘‘five-parameter” model (Fig. 1), as it may very accessible tool does have the necessary accuracy and reliabil-
be completely characterized by five parameters: shunt and series ity for professional and scientific purposes. In this work, we focus
resistances, diode ideality factor, photocurrent, and diode reverse on the validation of the model for the entire current-voltage (I-V)
saturation current. The single diode model ensures high accuracy and power-voltage (P-V) characteristics of four representative
through three characteristic points in the PV datasheet (open- commercial PV modules based on polycrystalline silicon, HIT tech-
circuit voltage, short-circuit current, and maximum power point), nology, CIGS, and CdTe. With reference to the solar irradiance, the
it guarantees that the maximum point generated by the mathe- validation was carried out in the range [900–1000 W/m2].
matical model coincides with the datasheet, and provides an excel- The paper is organized as follows: the next Section is on the
lent fit between to the experimental current-voltage (I-V) curve description of the model under validation. Section 3 presents the
(Mahmoud et al., 2013). test facility. Section 4 deals with results and discussion. Section 5
The five-parameters model is accurate enough for modelling presents the conclusions.
and simulation of crystalline Si PV modules, but the applicability
to other PV technologies (especially owning to the second genera- 2. The explicit empirical model
tion of photovoltaics) is found to be limited since the single-diode
equivalent circuit fails to describe the significantly different phys- The behaviour of a solar cell is commonly modelled with the
ical processes of converting radiant energy into electrical energy well-known five-parameter equivalent model represented in Fig. 1.
(Lineykin et al., 2014). For this reason, today many researchers The solar cells is modelled by an ideal current source in parallel
are focusing in the development of new models capable of describ- with a diode. The circuit is described by the following equation:
ing the behaviour of different technologies, such as for example
thin-films (Miceli et al., 2015).   V þ IRs
I ¼ IPh  Io  eðVþIRs Þ=nV t  1  ð1Þ
The explicit empirical model for general PV devices - that was Rsh
introduced in order to enable modelling based only on the parame-
where IPh (A) is the light generated current (i.e. the short circuit cur-
ters that are always listed in the datasheet of solar devices - over-
rent neglecting the parasitic resistances), Io (A) is the dark satura-
comes these drawbacks. It was initially introduced in Pavan et al.
tion current due to recombination, n is the ideality factor, Vt (V)
(2007) and then applied in Massi Pavan et al. (2014a), Barbini
is the thermal voltage, Rs (X) is a series resistance, and Rsh (X) is
et al. (2014) in a revised form for assessing the mismatch effect
a shunt resistance. The light generated current is directly propor-
due to the use of different classes of PV modules in large-scale solar
tional to the solar irradiance (Townsend et al., 1989):
parks. A revised form of the model was validated experimentally for
operation at Maximum Power Point (MPP) (Massi Pavan et al., G  
2014b), showing a very good prediction performance, better than IPh ¼  ILref þ a  ðT c  25Þ ð2Þ
1000
the ones obtained with the golden standard in PV modelling, i.e.
the five-parameters model. The explicit empirical model has been where G (W/m2) is the solar irradiance, 1000 W/m2 is the solar irra-
lately improved, introducing a correction factor that leads to a good diance at Standard Test Conditions (STC), ILref (A) is the short circuit
match with the experimental electrical characteristics also for oper- current at STC, a (A/°C) is the current-temperature coefficient at
ating points other than the MPP (Vergura and Massi Pavan, 2015). STC, Tc (°C) is the cell temperature and 25 °C is the STC cell
As mentioned, the model is based exclusively on the parameters temperature.
commonly found in the datasheets provided by the manufacturers, Combining Eqs. (1) and (2), we can write:
and is explicit – and therefore quite easy to implement in G     V þ IRs
computer-aided calculations. Explicit models are today increas- I¼  ILref þ a  ðT c  25Þ  Io  eðVþIRs Þ=nV t  1 
1000 Rsh
ingly being studied (Batzelis et al., 2014) due to these characteris-
ð3Þ
tics, and they represent a useful tool not only for scientists, but also
in all practical cases for PV plant designers, Operation and Mainte- The dark saturation current Io is a function of the cell tempera-
nance (O&M) personnel, and in general for PV professionals. In par- ture and can be written as (Kichou et al., 2016):
ticular, the model has a distinct advantage in terms of    
Ego Eg 3
computational complexity and time, both because its explicit form,  Tc
Io ¼ Ioref  e V to V t
 ð4Þ
and because the input parameters are readily available and do not 25
need to be computed in advance (see for example Sandrolini et al.,
where Ioref (A) and Vto (V) are the saturation current and the ther-
2010; Chatterjee et al., 2011; Lo Brano et al., 2010; Saleem and
mal voltage at STC, respectively, Eg (V) is the energy bandgap, while
Karmalkar, 2009; Bouzidi et al., 2007; Ortiz-Conde et al., 2003;
Ego (V) is the energy bandgap at T = 0 K. Combining Eqs. (3) and (4),
Amit and Kapoor, 2004; Ishaque et al., 2011; Vergura, 2016).
we can write:
   
G   Ego Eg
 Tc
3
I¼  ILref þ a  ðT c  25Þ þ Ioref  e V to V t

1000 25
 3 Ego Eg ðVþIRs Þ
Tc  þ V þ IRs
 Ioref   e V to V t nV t  ð5Þ
25 Rsh
The series resistance Rs is also a function of the operating con-
ditions being (Markvart and Castaner, 2006):

V OC
Rs ¼  rs ð6Þ
ISC
where Voc (V) and Isc (A) are the open circuit voltage and the short
Fig. 1. Solar cell equivalent circuit – five-parameters model. circuit current at arbitrary conditions of solar irradiance and cell
 A. Massi Pavan et al. / Solar Energy 155 (2017) 647–653 649

temperature respectively, and rs is the normalized solar cell’s and shunt resistances, the diode ideality constant, the diode
resistance. reverse saturation current, and the bandgap energy of the semi-
The open circuit voltage Voc depends on the cell temperature conductor (Villalva et al., 2009).
and can be written as (Luque and Hegedus, 2003): – It is explicit, which is a very desirable feature for simulation
applications, especially when the model is to be used repeatedly
V OC ¼ V OC;STC þ b  ðT c  25Þ ð7Þ
(as, for example, in the case of PV emulators (Massi Pavan et al.,
where VOC,STC (V) is the open circuit voltage at STC and b (V/°C) is 2009). Simulation times can be significantly reduced by avoid-
the voltage-temperature coefficient at STC. ing the numerical iterations required by implicit equation mod-
Finally, the short circuit current Isc depends on the solar irradi- els (Ortiz Conde et al., 2012).
ance and can be written as (Ishaque et al., 2011): – It can be used for any type of PV technology belonging to first
G and second generation photovoltaics as, for example, crystalline
ISC ¼  ISC;STC ð8Þ Si, CdTe, CIGS, etc.
1000
– It can be used for any type of PV device: solar cells, PV modules,
where ISC,STC (A) is the short circuit current at STC. PV strings and fields. The extension to the model to these latter
Substituting (7) and (8) in (6), then (6) in the third and fourth has been shown in Massi Pavan et al. (2014a) and comes from
term of (5), we obtain: the Kirchhoff’s laws and the induction principle.
G   – It can be used to calculate any operating conditions of the I-V
I¼  ILref þ a  ðT c  25Þ þ f 1 ðVÞ þ f 2 ðI; G; T c Þ and P-V characteristics (generic, at MPP, and in open and short
1000
þ expðV; I; G; T c Þ ð9Þ circuit configurations) for high levels of solar irradiance.

where f1 is a linear function of the voltage, f2 is a function of the cur-

rent and the operating conditions (thus similar to the first term of 3. Validation of the explicit empiric model
Eq. (5)), and the exponential depends on both the electrical vari-
ables and on the operating conditions. These three functions can The empirical model has been tested starting from the I-V and
be grouped in an empiric expression able to provide a correct cur- P-V characteristics of four PV modules mounted in our Laboratory
rent value for each voltage value in the range [0-VOC], whatever (Massi Pavan et al., 2014c). The four PV modules are representative
the environmental condition is: of different technologies: polycrystalline Si, CdTe, CIGS, and HIT. The
solar cells of this latter are made of a thin mono crystalline silicon
f 1 ðVÞ þ f 2 ðI; G; T c Þ þ expðV; I; G; T c Þ wafer surrounded by ultra-thin amorphous silicon layers. The differ-
ent parameters of the considered PV devices are reported in Table 1.
G em½Vbð25T c Þ  1
¼  ð10Þ The following instruments have been used to measure and log
1000 em  1
the different working conditions:
where m is an empiric exponential factor.
Combining (9) and (10), a more suitable empirical expression – An ISO9060 first class thermopile global radiometer type
where the current and the voltage are in a per unit representation C100RDPA153 from LSI Lastem S.r.l. measuring the global solar
(Vergura and Massi Pavan, 2015) is: irradiance (with a daily uncertainty less than 5%).
  – A contact probe type DLE124 produced by LSI Lastem S.r.l. (with
G em½V pu bð25T c Þ  1 an accuracy of ±0.15 °C).
Ipu ¼ IL;ref þ a  ðT c  25Þ  ð11Þ
1000 em  1 – Two data loggers type E-Log produced by LSI Lastem S.r.l.
where Ipu (p.u.) is the per unit current referred to ISC,STC, IL,ref (p.u.) is – A shunt type SHP300A60-Compact produced by Hobut Ltd. cal-
the per unit irradiance referred to 1000 W/m2, Vpu (p.u.) is the per ibrated with an accuracy better than 0.01%.
unit voltage referred to VOC,STC, a0 (1/°C) is the current-temperature
coefficient referred to ISC,STC (a0 = a/ISC,STC) and b0 (1/°C) is the Fig. 2 shows two of the considered PV modules, while in Fig. 3
voltage-temperature coefficient referred to VOC,STC (b0 = b/VOC,STC). the data logger and the shunt are visible.
The model represented by Eq. (11) has a wide applicability and
presents the following pros:

– It can be used considering only the electrical parameters which

can always be found in the solar cell/PV module datasheet. This
represents a clear advantage as the commonly used models
require parameters that cannot be found in the manufacturer’s
datasheets, such as the light-generated or PV current, the series Fig. 2. The CdTe and CIGS PV modules used in the test.

Table 1
Electrical characteristics at STC (except for NOCT).

Photovoltaic module Q.Pro HIP 215NHE5 UF-95 FS-272

Material technology Poly-Si HIT CIGS CdTe
Nominal power Pn (W) 230 215 95 72.5
Tolerance on the nominal power (W) +5/0% +10/5% +5/0% ±5.0%
Short circuit current ISC (A) 8.30 5.61 1.68 1.23
Open circuit voltage VOC (V) 36.6 51.6 78.0 88.7
Current at maximum power point Imp (A) 7.84 5.13 1.53 1.09
Voltage at maximum power point Vmp (V) 29.6 42.0 62.1 66.6
Current/temperature coefficient a (%/K) +0.04 +0.03 0.00 +0.04
Voltage/temperature coefficient b (%/K) 0.41 0.03 0.38 0.25
NOCT (°C) 47 48 51 45
650 A. Massi Pavan et al. / Solar Energy 155 (2017) 647–653

The I-V curves have been plotted using a variable resistive load.
Voltages and currents have been measured using the test facility
described in Massi Pavan et al. (2014c). The voltages are measured
with an accuracy better than 0.01%, while the currents have an
accuracy better than 0.1%.

4. Results

This section presents the comparison between the I-V and P-V
characteristics evaluated by the empirical model and those
obtained from the experimental measurements. The solar irradi-
ance is in the range [900–1000 W/m2].
Figs. from 4 to 7 depict the I-V and P-V characteristics for the Q.
Pro, HIP 215 NHE5, UF-95, and FS-272 modules respectively. For
each PV technology, a different value of ideality factor has been
used according to Yıldıran and Tacer (2016).
The analysis of the eight plots show a fully satisfactory corre-
spondence between the measured and the estimated curves. With
reference to the Sanyo HIP module, a slightly overestimated short
Fig. 3. Particular of the data acquisition system. circuit current and a slightly underestimated open circuit voltage
can be noticed. From a quantitative point of view, the results are

Fig. 4. Characteristics for the Q.Pro module (n = 1.3 Yıldıran and Tacer, 2016), Fig. 5. Characteristics for the HIP215NHE5 (n = 1.8 Yıldıran and Tacer, 2016),
G = 916 W/m2, Tc 57 °C: (a) I-V curve; (b) P-V curve. G = 976 W/m2, Tc 57 °C: (a) I-V curve; (b) P-V curve.
 A. Massi Pavan et al. / Solar Energy 155 (2017) 647–653 651

Fig. 6. Characteristics for the UF95 (n = 1.5 Yıldıran and Tacer, 2016),
G = 970 W/m2, Tc 58 °C: (a) I-V curve; (b) P-V curve. Fig. 7. Characteristics for the FS-272 (n = 1.5 Yıldıran and Tacer, 2016),
G = 936 W/m2, Tc 56 °C: (a) I-V curve; (b) P-V curve.

given in terms of statistical errors. The coefficient of determination

R2 and the root mean square deviation RMSD are listed in Table 2. Table 2
The obtained coefficient of determination, consistently larger than Statistical errors.
0.98 and mostly equal to 0.99, show the high performance of the Power Current Voltage
explicit empirical model for any technology. We have also com-
PV module RMSD (W) R2 RMSD (W) R2 RMSD (W) R2
puted the Absolute Percentage Errors (APE) for the main points
Q.PRO 5.92 0.99 0.24 0.99 2.43 0.98
of the I-V characteristics between the measured and estimated val-
HIP215NHE5 6.14 0.99 0.15 0.99 3.27 0.98
ues, respect to the measured ones (Table 3). The measures UF-95 1.64 0.99 0.03 0.99 1.72 0.98
(acquired with an accuracy of 0.01% for the voltage and 0.1% for FS-272 1.15 0.99 0.03 0.99 0.63 0.99
the current) do not affect the APE values.
Moreover, we have compared the accuracy of the proposed
model with the accuracy of the Rp-model and of the two-diode
models of Ishaque et al. (2011), considering the errors reported single-diode equation, used to derive the proposed model. A more
in the last two columns of Tables 5 and 7 (Poly-Si and thin film, accurate starting model, such as the double-diode model is
respectively) and Figs. 12–13 presented in the same work. Table 4 expected to perform better, but it would require two empirical
compares the errors of the proposed model with of the other two indexes, m1 and m2, to follow the characteristic curves of the
models reported in Ishaque et al. (2011). As it can be observed, two diodes, thus complicating the proposed model, which now is
the proposed model has a limited error, comparable with that of effective and fast, because it does not require any iterative
well-established models. approach, as it happens for other models. With reference to low
Summarizing, the proposed model performs well at the consid- levels of solar irradiance, and especially for the CdTe technology
ered high level of solar irradiance. In some cases a mismatch is (Nofuentes et al., 2014; Torres Ramírez et al., 2014), the empirical
observed. We speculate that this limitation is due to the model can be used but taking into account the influence of the
652 A. Massi Pavan et al. / Solar Energy 155 (2017) 647–653

Table 3
Absolute percentage errors (APE).

Voc Isc Vmp Imp PMPP FF

PV module : Q.PRO
Measured 32.70 8.21 22.30 6.96 155.3 0.57
Estimated 32.62 8.27 23.15 6.85 158.7 0.58
APE (%) 0.22 0.82 3.83 1.62 2.15 1.54
PV module : HIP215NHE5
Measured 47.30 5.55 36.30 4.93 179.1 0.69
Estimated 46.33 5.47 35.53 4.89 173.9 0.68
APE (%) 2.04 1.40 2.14 0.76 2.88 2.23
PV module : UF-95
Measured 71.50 1.65 56.60 1.36 76.8 0.65
Estimated 71.61 1.66 54.89 1.44 78.7 0.66
APE (%) 0.15 0.41 3.01 6.01 2.83 2.25
PV module: FS-272
Measured 81.70 1.21 59.9 0.90 54.1 0.54
Estimated 82.40 1.16 58.48 0.95 55.8 0.58
APE (%) 0.86 4.75 2.37 5.69 3.18 7.40

Table 4
Per cent error values for the main points of the I-V curve and of the P-V curve between the proposed model and both the Rp-model and the two-diode model of Ishaque et al.
(2011).

Technology PV module APEPmp APEVmp APEVoc

Poly-Si Q.PRO (proposed model) 2.15% 3.83% 0.22%
S36/KC200GT (Rp-model) (Amit and Kapoor, 2004) 0.219 2.055 0.2
S36/KC200GT (two-diode model) (Amit and Kapoor, 2004) 0.156 1.369 0.1
Thin film UF-95 (proposed model) 2.83% 3.01% 0.15%
FS-272 (proposed model) 3.18% 2.37% 0.86%
ST40/SQ150PC (Rp-model) (Amit and Kapoor, 2004) 0.912 1.418 1.2
ST40/ SQ150PC (two-diode model) (Amit and Kapoor, 2004) 0.853 0.709 0.9

angle of incidence and the spectral effect in the value of the solar Mr. Paolo Pruni, Mr. Diego Logar, and Mr. Adriano Zibai are
irradiance. acknowledged for their help in the construction of the test facility.
Mr. Maurizio Besenghi is acknowledged for his daily commit-
5. Conclusions ment that ensure the correct operations of the Laboratory at the
University of Trieste.
This paper presents the experimental validation of the explicit The second author would like to thank the International Centre
empirical model for general PV devices. for Theoretical Physics (ICTP), Trieste (Italy) for providing the
The model was tested at different operating points correspond- materials and the computer facilities for performing the present
ing to high values of solar irradiance. Moreover, the tests were per- work.
formed on four different representative PV technologies, i.e.
polycrystalline Si, HIT, CIGS, and CdTe - belonging to the first and References
the second generation of photovoltaic technologies, respectively.
The model, that had already been tested at maximum power Acciani, G., Falcone, O., Vergura, S., 2010. Typical Defects of PV-cells. IEEE-ISIE 2010,
point (Massi Pavan et al., 2014b), has shown to be able to predict July, 4–7, Bari, Italy, pp. 2745–2749.
Amit, J., Kapoor, A., 2004. Exact analytical solutions of the parameters of real solar
the key features of the actual I-V and P-V characteristics to a very cells using Lambert W-function. Sol. Energy Mater. Sol. Cells 81, 269–277.
good degree of approximation. The curves measured experimen- Barbini, N., Lughi, V., Mellit, A., Massi Pavan, A., Tessarolo, A., 2014. On the impact of
tally almost always lay on the predicted ones and the obtained photovoltaic module characterization on the prediction of PV plant
productivity. IEEE-EVER, 25–27/03/2014, Monaco, France, pp. 1–4.
results in terms of statistical errors quantitatively confirm the Batzelis, E.I., Routsolias, I.A., Papathanassiou, S.A., 2014. An explicit PV string model
excellent performance of the model. based on the Lambert W function and simplified MPP expressions for operation
Thus, the model represents a very useful and accessible tool not under partial shading. IEEE Trans. Sustainable Energy 5, 301–312.
Bizzarri, F., Buongiorno, M., Brambilla, A., Gruosso, G., Storti Gajani, G., 2013. Model
only for scientists but also in all practical cases for PV profession-
of photovoltaic power plants for performance analysis and production forecast.
als, such as PV plant designers and Operation and Maintenance IEEE Trans. Sustainable, Energy 4, 278–285.
(O&M) personnel. Bouzerdoum, M., Mellit, A., Massi Pavan, A., 2013. A hybrid model (SARIMA-SVM)
A future work will be focused on the assessment of a new for short-term forecasting of a small-scale grid-connected photovoltaic plant.
Sol. Energy 98, 226–235.
empirical model based on the idea of the two diodes equivalent Bouzidi, K., Chegaar, M., Bouhemadou, A., 2007. Solar cells parameters evaluation
circuit. considering the series and shunt resistance. Sol. Energy Mater. Sol. Cells 91,
1647–1651.
Boztepe, M., Guinjoan, F., Velasco-Quesada, G., Silvestre, S., Chouder, A., Karatepe,
Acknowledgements E., 2014. Global MPPT scheme for photovoltaic string inverters based on
restricted voltage window search algorithm. IEEE Trans. Ind. Electron. 61 (7),
3302–3312.
Dr. Vittorio Arcidiacono is acknowledged for his original work
Breitenstein, O., Langenkamp, M., Lang, O., Schirrmacher, A., 2001. Shunts due to
on the explicit empirical model. laser scribing of solar cell evaluated by highly sensitive lock-in thermography.
Q.Cells Italia S.r.l. and its former CEO Dr. Matthias Altieri are Sol. Energy Mater. Sol. Cells, 55–62.
kindly acknowledged for financial support of the test facility. Q. Breitenstein, O., Rakotoniaina, J.P., Al Rifai, M.H., Werner, M., 2004. Shunt type in
crystalline solar cells. Prog. Photovoltaics Res. Appl. 12, 529–538.
Cells SE with Dr. Aurora Tedesco is also acknowledged for provid- Chatterjee, A., Keyhani, A., Kapoor, D., 2011. Identification of photovoltaic source
ing the PV modules used in the study. models. IEEE Trans. Energy Convers. 26, 883–889.
 A. Massi Pavan et al. / Solar Energy 155 (2017) 647–653 653

Chicco, G., Cocina, V., Di Leo, P., Spertino, F., Massi Pavan, A., 2016. Error assesment Massi Pavan, A., Lughi, V., Mellit, A., Roitti, S., Sulligoi, G., Tessarolo, A., 2014. The
of solar irradiance forecasts and AC power from energy conversion model in photovoltaic laboratory at the University of Trieste. AEIT – IEEE Italy Section,
grid-connected photovoltaic systems. Energies 9. Trieste, September 2014.
Chine, W., Mellit, A., Massi Pavan, A., Kalogirou, S.A., 2014. Fault detection method Massi Pavan, A., Tessarolo, A., Barbini, N., Mellit, A., Lughi, V., 2015. The effect of
for grid-connected photovoltaic plants. Renewable Energy 66, 99–110. manufacturing mismatch on energy production for large-scale photovoltaic
Chine, W., Mellit, A., Lughi, V., Malek, A., Sulligoi, G., Massi, A., 2016. Pavan, « A novel plants. Sol. Energy 117, 282–289.
fault diagnosis technique for photovoltaic systems based on artificial neural Miceli, R., Orioli, A., di Gangi, A., 2015. A procedure to calculate the I-V
networks ». Renewable Energy 90, 501–512. characteristics of thin-film photovoltaic modules using an explicit rational
Choubey, P., Oudhia, A., Dewangan, R., 2012. A review: Solar cell current scenario form. Appl. Energy 155, 613–628.
and future trends. Recent Res. Sci. Technol. 4 (8), 99–101. Moballegh, S., Jiang, J., 2014. Modeling, prediction, and experimental validations of
Cristaldi, L., Faifer, M., Rossi, M., Ponci, F., 2012. A simple photovoltaic panel model: power peaks of PV arrays under partial shading conditions. IEEE Trans.
characterization procedure and evaluation of the role of environmental Sustainable Energy 5, 293–300.
measurements. IEEE Trans. Sustainable Energy 61, 2632–2641. Nofuentes, G., García Domingo, B., Muñoz, J.V., Chenlo, F., 2014. Analysis of the
Cristaldi, L., Faifer, M., Leone, G., Vergura, S., 2015. Reference strings for statistical dependence of the spectral factor of some PV technologies on the solar
monitoring of the energy performance of photovoltaic fields. In: IEEE-ICCEP spectrum distribution. Appl. Energy 113, 302–309.
2015 International Conference on Clean Electrical Power, 16–18/06/2015, Ortiz Conde, A., Lugo-Munoz, D., Garcia Sanchez, F.J., 2012. An explicit
Taormina, Italy, 2015. multiexponential model as an alternative to traditional solar cell models with
d’Alessandro, V., Di Napoli, F., Guerriero, P., Daliento, S., 2015. An automated high- series and shunt resistances. IEEE J. Photovoltaics 2, 261–268.
granularity tool for a fast evaluation of the yield of PV plants accounting for Ortiz-Conde, A., Garcia Sanchez, F.J., Muci, J., 2003. New method to extract the
shading effects. Renewable Energy 83, 294–304. model parameters of a solar cells from the explicit analytic solutions of their
Dellino, G., Laudadio, T., Mari, R., Mastronardi, N., Meloni, C., Vergura, S., 2015. illuminated I-V characteristics. Sol. Energy Mater. Sol. Cells 90, 352–361.
Energy production forecasting in a PV plant using transfer function models. Saleem, H., Karmalkar, S., 2009. An analytical method to extract the physical
IEEE-EEEIC 2015, 10-13/06/2015, Roma, Italy. parameters of a solar cell from four points on the illuminated J-V curve. IEEE
Dolara, A., Leva, S., Manzolini, G., 2015. Comparison of different physical models for Trans. Device Lett. 30, 349–352.
PV power output prediction. Sol. Energy 119, 83–99. Sandrolini, L., Artioli, M., Reggiani, U., 2010. Numerical method for the extraction of
Duffie, J.A., Beckman, W.A., 1991. Solar Engineering of Thermal Processes. John photovoltaic module double-diode model parameters through cluster analysis.
Wiley & Sons Inc., New York. Appl. Energy 87, 442–451.
Ishaque, K., Salam, Z., Taheri, H., 2011. Simple, fast and accurate two diode model Seyedmahmoudian, M., Rahmani, R., Mekhilef, S., Oo, A.M.T., Stojcevski, A., Soon, T.
for photovoltaic modules. Sol. Energy Mater. Sol. Cells 95, 586–594. K., Ghandhari, A.S., 2015. Simulation and hardware implementation of new
Kichou, S., Silvestre, S., Nofuentes, G., Torres-Ramírez, M., Chouder, A., Guasch, D., maximum power point tracking technique for partially shaded PV systems
2016. Characterization of degradation and evaluation of model parameters of using hybrid DEPSO method. IEEE Trans. Sustainable Energy 6, 850–861.
amorphous silicon photovoltaic modules under outdoor long term exposure. Spertino, F., Ciocia, A., Di Leo, P., Tommasini, R., Berardone, I., Corrado, M., Infuso, A.,
Energy 96, 231–241. Paggi, M., 2015. A power and energy procedure in operating photovoltaic
Lineykin, S., Averbukh, M., Kuperman, A., 2014. Issues in modeling amorphous systems to quantify the losses according to the causes. Sol. Energy 118, 313–
silicon photovoltaic modules by single-diode equivalent circuit. IEEE Trans. Ind. 326.
Electron., 6785–6793 Torres Ramírez, M., Nofuentes, G., Silva, J.P., Silvestre, S., Muñoz, J.V., 2014. Study on
Lo Brano, V., Orioli, A., Ciulla, G., Di Gangi, A., 2010. An improved five-parameters analytical modelling approaches to the performance of thin film PV modules in
model for photovoltaic modules. Sol. Energy Mater. Sol. Cells 94, 1358–1370. sunny inland climates. Energy 73, 731–740.
Luque, A., Hegedus, S., 2003. Handbook of photovoltaic science and engineering. Townsend, T.U., 1989. M.S. Thesis, Mechanical Engineering, U. of Wisconsin
Chichester, UK. Madison. A Method for Estimating the Long-Term Performance of Direct-
Mahmoud, Y., Xiao, W., Zeineldin, H.H., 2013. A parameterization approach for Coupled Photovoltaic Systems.
enhancing PV model accuracy. IEEE Trans. Ind. Electron. 60 (12), 5708–5716. Vergura, S., 2016. A complete and simplified datasheet-based model of PV cells in
Manganiello, P., Ricco, M., Petrone, G., Monmasson, E., Spagnuolo, G., 2014. variable environmental conditions for circuit simulation. Energies 9 (5), 326.
Optimization of perturbative PV MPPT methods through online system Vergura, S., Massi Pavan, A., 2015. On the photovoltaic explicit empirical model:
identification. IEEE Trans. Ind. Electron. 61 (12), 6812–6821. operations along the I-V curve. IEEE-ICCEP, 16–18/06/2015, Taormina Italy, pp.
Markvart, T., Castaner, L., 2006. Practical Handbook of Photovoltaics. Fundamentals 99–104.
and applications. Oxford, UK. Vergura, S., Acciani, G., Amoruso, V., Patrono, G., Vacca, F., 2009. Descriptive and
Massi Pavan, A., Castellan, S., Quaia, S., Roitti, S., Sulligoi, G., 2007. Power electronic inferential statistics for supervising and monitoring the operation of PV plants.
conditioning systems for industrial photovoltaic fields: centralized or string IEEE Trans. Ind. Electron. 56 (11), 4456–4464.
inverters?. IEEE-ICCEP, 21–23/05/2007, Capri, Italy, pp. 208–214. Vergura, S., Acciani, G., Falcone, O., 2009. Modeling defects of PV-cells by means of
Massi Pavan, A., Castellan, S., Sulligoi, G., 2009. An innovative photovoltaic field FEM. IEEE-ICCEP 2009, 9-11/06/2009, Capri, Italy, pp. 52–56.
simulator for hardware-in-the-loop test of power conditioning units. IEEE- Vergura, S., Acciani, G., Falcone, O., 2009. 3-D PV-cell model by means of FEM. IEEE-
ICCEP, 09–11/06/2009, Capri, Italy, pp. 41–45. ICCEP 2009, 9-11/06/2009, Capri, Italy, pp. 35–40.
Massi Pavan, A., Mellit, A., De Pieri, D., Kalogirou, S.A., 2013. A comparison between Vergura, S., Acciani, G., Falcone, O., 2012. A finite element approach to analyze the
BNN and regression polynomial methods for the evaluation of the effect of thermal effect of defects on silicon-based PV cells. IEEE Trans. Ind. Electron. 59
soiling in large scale photovoltaic plants. Appl. Energy 108, 392–401. (10), 3860–3867.
Massi Pavan, A., Mellit, A., De Pieri, D., Lughi, V., 2014a. A study on the mismatch Villalva, M.G., Gazoli, J.R., Filho, E.R., 2009. Comprehensive approach to modelling
effect due to the use of different photovoltaic modules classes in large-scale and simulation of photovoltaic arrays. IEEE Trans. Power Electron. 24, 1198–
solar parks. Prog. Photovoltaics: Res. Appl. 22, 332–345. 1208.
Massi Pavan, A., Mellit, A., Lughi, V., 2014b. Explicit empirical model for general Yıldıran, N., Tacer, E., 2016. Identification of photovoltaic cell single diode discrete
photovoltaic devices: experimental validation at maximum power point. Sol. model parameters based on datasheet values. Sol. Energy 127, 175–183.
Energy 101, 105–116.