Professional Documents
Culture Documents
A Novel MPPT Method For PV Systems With Irradiance Measurement
A Novel MPPT Method For PV Systems With Irradiance Measurement
com
ScienceDirect
Solar Energy 109 (2014) 95–104
www.elsevier.com/locate/solener
Received 3 April 2014; received in revised form 30 July 2014; accepted 7 August 2014
Available online 7 September 2014
Abstract
Maximum Power Point Tracking (MPPT) is needed in a photovoltaic system to ensure the operation in the Maximum Power Point
(MPP), maximizing the generated energy. Many MPPT techniques have been developed. Perturbation and observation, as well as incre-
mental conductance, are by far the most widely used because of their simplicity. However, they have some weak points and problems that
affect their efficiency, especially under rapid irradiance changes and partial shading. This paper proposes a novel MPPT method which
can be implemented in PV (photovoltaic) inverters and charge controllers. The most innovative feature is the presence of photodiodes as
irradiance sensors. Thanks to irradiance measurement, the proposed method can recognize problematic scenarios and execute specific
algorithms to prevent malfunction under these situations. The MPPT method has been theoretically analyzed; efficiency results are
shown in comparison with standard techniques, which verify its feasibility.
Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.solener.2014.08.017
0038-092X/Ó 2014 Elsevier Ltd. All rights reserved.
96 Á.-A. Bayod-Rújula, J.-A. Cebollero-Abián / Solar Energy 109 (2014) 95–104
characteristics of solar cells depend especially on irradiance (Mamarelis et al., 2014). With a lower voltage step this fluc-
and temperature. One of the objectives of the control cir- tuation decreases and the MPP can be located more accu-
cuit is to reach and keep the maximum power point, which rately. However, the tracking speed decreases and other
is known as Maximum Power Point Tracking. MPPT can problems like wrong tracking direction are more prone to
be implemented in a DC–DC converter, which is usually appear. Therefore, the voltage step size should be adjusted
included in PV inverters. There are also charge controllers to allow a balance between transient state and steady state.
with MPPT technology, for stand-alone systems. Direct methods do not work properly with low irradi-
Many MPPT techniques have been developed. Constant ance. In these conditions the P–V characteristic curve is
voltage, fractional open-circuit voltage (VOC), perturbation too flat, and little irradiance changes or even the ripple in
and observation, incremental conductance, neural net- the measured signals can confuse the algorithm.
works or fuzzy logic are some of the most used ones The measurement of some magnitudes, like short-circuit
(Esram and Chapman, 2007; Faranda et al., 2008; Reisi current or open-circuit voltage, requires the interruption of
et al., 2013; Bhatnagar and Nema, 2013; Eltawil and the power generation. A pilot cell can be used to avoid
Zhao, 2013; Salam et al., 2013). They can be classified into these interruptions, but it is an expensive solution (Esram
two groups: direct methods and indirect methods. Direct and Chapman, 2007). In addition, the accuracy may be
methods use the output power of the PV system as feed- low because each PV cell is slightly different due especially
back to search the MPP. Indirect methods measure other to the manufacturing processes.
magnitudes, such as the short-circuit current, the open- Several papers available in the literature propose opti-
circuit voltage or the temperature of the modules. mized MPPT techniques. Unfortunately, most of these
The most basic methods do not give very good MPPT techniques have a limited field of practical applicability
performance. More sophisticated methods can provide bet- because they require high computational power, expensive
ter results, but they are more complex and more expensive. hardware resources, high quality measurement sensors
The most important problems of these four MPPT meth- and/or high performance digital controllers. Consequently,
ods are malfunction with partial shading, wrong tracking in real commercial applications in which cost, size and
direction, power oscillations around MPP, bad working weight are the main constraints, many complex methods
with low irradiance and interruptions in the power presented in literature are not being used (Mamarelis
generation. et al., 2014).
Partial shading produces multi-peak output P–V curves. Moreover, in the literature very few MPPT methods
Many direct MPPT algorithms, such as perturbation and with irradiance measurement have been proposed. In
observation, are designed to reach the nearest local maxi- (Faranda et al., 2008) the temperature parametric equa-
mum, but they do not analyze and compare all peaks to tion method is explained; it is an indirect technique in
determine the true MPP (Min et al., 2011; Ishaque and which irradiance and temperature measures are used to
Salam, 2013). Some approaches to this problem can be estimate the MPP voltage. The pilot cell technique
found in the literature (Qi et al., 2014; Bouilouta et al., (Esram and Chapman, 2007) is an indirect method in
2013; Syafaruddin et al., 2012), but the majority of them which the irradiance is not directly measured. A small
require complex and expensive electronic devices to be pilot cell with the same characteristics as the panels is used
implemented. Economy and simplicity are a key factor in to estimate the open-circuit voltage or the short-circuit
small and/or distributed PV systems, so that complex current of the PV panels without stopping the power
methods are not suitable in practice. generation.
Wrong tracking direction problem only happens in In this paper, a new MPPT method is proposed. The
direct methods, such as perturbation and observation or most innovative feature of this proposed MPPT method
incremental conductance (Salas et al., 2006; Tey and is the presence of photodiodes as irradiance sensors. Their
Mekhile, 2014). Under changing irradiance conditions, information allows the algorithm to work properly regard-
the difference in the output power between one sample less the climatic conditions. A PV system has been modeled
and the previous one is due to two factors: the voltage per- and a computer application has been designed to simulate
turbation and the irradiance variation. When irradiance the behavior of MPPT techniques, using real measures of
varies rapidly, there is no evidence to determine whether irradiance and temperature. Four well-known MPPT
the main factor is the change of irradiance or the voltage methods have been tested:
perturbation. In some geographical areas, irradiance
changes are very common. Therefore, it is very important 1: Fixed voltage (indirect method)
to tackle this problem to get an optimal MPPT efficiency, 2: Fractional VOC (indirect method)
especially in cloudy days when solar resource is scarce. 3: Perturbation and observation (direct method)
Perturbation and observation operates periodically 4: Incremental conductance (direct method)
incrementing or decrementing the output voltage of the
PV array. This method can result in oscillations in the The simulation results have verified that the proposed
power output, due to the fact that the voltage is continu- method can provide better efficiency figures than most of
ously changing between three values in steady state the standard techniques.
Á.-A. Bayod-Rújula, J.-A. Cebollero-Abián / Solar Energy 109 (2014) 95–104 97
When the operating point is far from the MPP, a high volt-
age step is advisable in order to reach it quickly and avoid
the wrong tracking direction problem. However, after
reaching the MPP, a low voltage step size provides better
accuracy and reduces unwanted oscillations (Al-Diab and
Sourkounis, 2010).
The proposed method can choose the voltage step size
between two fixed values (Mellit et al., 2011; Yang and
Zhao, 2011). Some techniques have a completely variable
step size. However, the simulations results obtained in this
work have shown that two step values are enough; a high
one when the working point is far from the MPP and a
low one when the MPP has been reached. The voltage step
can take two values, 0.4 V and 0.2 V; it is chosen according
to DP/DV between the current sample and the previous Fig. 4. Flowchart of the proposed method.
one.
In order to determine whether choose the high or the The proposed method changes between two different
low voltage step, the slope of the P–V curve is evaluated. techniques, depending on the level of irradiance. A pertur-
In this curve, the closer to the MPP, the lower the absolute bation and observation based algorithm with variable volt-
value of the slope. The proposed method chooses the low age step size is used when the irradiance is above 200 W/m2
voltage step if the slope is smaller than a threshold value; and partial shading has not been detected. Fractional VOC
which means that the MPP is close. Otherwise, the high is used when the irradiance is below 200 W/m2 and partial
step size is used. shading has not been detected. Thanks to the combination
As mentioned, the slope of the P–V output also depends of techniques, the proposed method covers the whole irra-
on the irradiance over the surface of the panels; the curve diation range with excellent performance.
becomes flatter if that magnitude decreases. Therefore in The irradiance in each MPPT step is calculated as the
the proposed method the threshold value of the slope is arithmetic mean of the response of all photodiodes. In
not a fixed value, it is directly proportional to the practice, the threshold will not be exactly 200 W/m2, due
irradiance calculated through the arithmetic mean of the to the uncertainty in the response of the sensors. In
response of all photodiodes. addition, hysteresis has been set to avoid repetitive and
undesired changes of technique.
2.4. Combination of techniques The proportionality constant of the fractional VOC
method is kv = 0.8; this value has been chosen in order to
Fig. 4 contains the flowchart witch summarizes the maximize the efficiency of the MPPT method during the
proposed MPPT method. whole year. The open-circuit voltage is measured every
As mentioned, direct MPPT methods do not work 120 s while fractional VOC is being executed.
properly with very low irradiance. Indirect methods can Even with high irradiance, the proposed method
provide better efficiency under these circumstances. measures the open-circuit voltage before launching
Á.-A. Bayod-Rújula, J.-A. Cebollero-Abián / Solar Energy 109 (2014) 95–104 99
perturbation and observation. Thanks to this information, other parameters (Orioli and Gangi, 2012). The ideality
the initial operating point of this technique can be set close factors of the diodes a1 and a2 are chosen according to
to the MPP, decreasing the time taken to reach it. After fin- the characteristics of the PV panel. These constants express
ishing a partial shading situation, the measurement of the the degree of ideality of the diodes, which affect the curva-
open-circuit voltage is also executed before launching per- ture of the I–V curve. Some researchers have discussed
turbation and observation. ways to estimate the correct value of these constants, it is
usually assumed that a1 = 1 and 1 6 a2 6 2 (Tian et al.,
3. PV system model and MPPT simulation 2012).
The following expression represents the output current
Solar cells consist of a PN junction made of semiconduc- of a PV cell. This equation must be solved using numerical
tor material, usually silicon. Solar irradiance is directly iterative methods, because neither IPV nor VPV can be
converted into electricity through the photovoltaic effect isolated on one side of the equal sign. If VPV is known,
(Luque and Hegedus, 2003). The output I–V curves of IPV can be calculated, and vice versa.
PV modules depend mainly of the irradiance that reaches I PV ¼ I PVgen I D1 ðV PV ; I PV Þ I D2 ðV PV ; I PV Þ
the cells and the temperature of them.
V PV þ I PV Rs
A PV panel can be modeled from the physics of a PN ð1Þ
Rp
junction (Saloux et al., 2011; Celik and Acikgoz, 2007).
The simplest equivalent circuit of a PV module is composed The equation of the photogenerated current can be
by a current source and a diode in parallel. This model is mathematically expressed as follows. G and T are respec-
known as Ideal Model, Single-diode Model or No Resistor tively the irradiance and the cell temperature.
Model. One of its parameters is the diode ideality factor, G
which takes into account the diffusion current of PV cells. I PVgen ¼ ðI SC þ aðT T STC ÞÞ ð2Þ
GSTC
The Series Resistor Model, by far the most widely used
in simulation, includes a series resistor Rs; and the Two The equations of the diodes are shown. I01 and I02 are
Resistor Model, also called One Diode Model, includes the reverse saturation currents of diode 1 and diode 2.
both a series Rs and a parallel resistor Rp. The number of
V PV þ I PV Rs
parameters increases and the computational requirements I D1 ¼ I 01 exp 1 ð3Þ
a1 V T
are higher (Bennett et al., 2012). The series resistor repre-
sents internal losses due to the current flow; whereas the The thermal voltage of Ns cells connected in series is
parallel or shunt resistor represents some leakage current given by this expression, in which q is the electron charge
inside the PV cell. (q = 1.602176 1019 C) and k is the Boltzmann constant
In this work the Two Diode Model, a more advanced (k = 1.3806503 1023 J/K)
equivalent circuit, is used. This model, shown in Fig. 5, is Ns k T
similar to the Two Resistor Model but with two diodes VT ¼ ð4Þ
q
in parallel. The second diode represents the recombination
losses in the depletion region. This model has better accu- The reverse saturation current of a diode can be calcu-
racy, especially in the region close to VOC and when irradi- lated as follows. Eg is the band gap energy of the semicon-
ance or temperature is far from standard test conditions ductor and I0STC is the nominal saturation current at STC.
(STC) (Ishaque et al., 2011a,b). On the other hand, the 3
T STC q Eg 1 1
computational cost is significantly higher. I 0 ¼ I 0STC exp ð5Þ
T a k T STC T
The model uses these PV panel parameters, given for
STC (TSTC = 25 °C, GSTC = 1000 W/m2): VOC, ISC, VMPP, A more optimized equation has been taken, based on
IMPP, a, b, Ns, Rs, Rp, a1 y a2. (Ishaque et al., 2011a,b):
The first seven parameters are usually shown in the I SC þ aðT T STC Þ
datasheet of the PV panel. Rs and Rp can be estimated from I 01 ¼ ð6Þ
exp V OC þbðT T STC Þ
a1 V T
1
With the values of Rs and Rp, and the rest of the param- multiple peaks, as illustrated in Fig. 8; only one of them is
eters, the output curve of the PV panel is calculated using the global MPP (Petrone and Ramos-Paja, 2011). If the
Eq. (1) and successive. The Maximum Power Point of the current flowing through a PV cell is higher than its short-
panel is obtained from this curve; and it is compared to circuit current, the cell absorbs power and its temperature
the Maximum Power Point of the PV panel that provides rises. To avoid the rupture of the cell, they are provided
the manufacturer, this data is an input of the PV model with by-pass diodes.
that is used specifically to calculate Rs and Rp. The PV model has been interfaced with a MPPT con-
The difference between the two Maximum Power Points troller; both of them developed in MATLAB. The block
is evaluated. If the difference is less than a tolerance, in this diagram of the whole system is shown in Fig. 9. This sim-
case 0.01%, the iteration process finishes. If not, in the next ulation tool can emulate step by step the behavior of a wide
iteration Rs is increment in 0.001 X, and the previous step is range of MPPT methods; using irradiance, temperature
repeated again. The process finishes when the difference is and the parameters of the PV system as inputs. In each sim-
less than 0.01%. ulation run, the MPPT efficiency is calculated as:
RP ¼ h h
V þI R
MPPi MPPh s i i X
n
5 5
2 I-V (partially shaded array)
200 W/m , 25 ºC
2
4.5 500 W/m , 25 ºC 4.5
2
800 W/m , 25 ºC
4 4
3.5 3.5
3 3
I (A)
I (A)
2.5 2.5
2 2
1.5 1.5
1 1
0.5 0.5
0 0
0 5 10 15 20 25 0 10 20 30 40 50 60 70
V (V) V (V)
Fig. 6. I–V output curves for different levels of irradiance. Fig. 7. I–V characteristic of a partially shaded PV array.
Á.-A. Bayod-Rújula, J.-A. Cebollero-Abián / Solar Energy 109 (2014) 95–104 101
100 60
P-V (partially shaded array) Real tracking
Ideal tracking
90 2
200 W/m , 50 ºC
50 2
80 400 W/m , 50 ºC
2
600 W/m , 50 ºC
2
70 800 W/m , 50 ºC
40 2
1000 W/m , 50 ºC
60
P (W)
P (W)
50 30
40
20
30
20
10
10
0
0 10 20 30 40 50 60 70 0
0 2 4 6 8 10 12 14 16 18 20
V (V) V (V)
Fig. 8. P–V characteristic of a partially shaded PV array. Fig. 10. Real tracking and ideal tracking along a sunny day and a
partially cloudy day with the proposed method.
50
Real tracking
Ideal tracking
45
40
35
25
20
Table 1
Solar panel parameters. 15
Parameter Value 10
VOC 21.06 V
5
ISC 3.8 A
VMPP 17.1 V 0
0
IMPP 3.5 A Time
Cells 36
a 3 mA/°C Fig. 11. Real output power and maximum output power along a sunny
b 80 mV/°C day and a partially cloudy day with the proposed method.
a1 1
a2 1.2
Rs 0.326 O To verify and evaluate the behavior of the proposed
Rp 161.572 O method under problematic scenarios, each technique has
been tested in specific circumstances: in sunny days, in
cloudy days and in partly cloudy days. Table 2 shows the
ance and temperature data and with the same PV system, a results of these simulations. To calculate the MPPT effi-
five-panel string with the parameters given in Table 1. No ciency figures, Eq. (9) has been used. The PV simulation
partial shading has been established, because under this cir- tool have a diagnosis algorithm that determines the maxi-
cumstance the efficiency of each method strongly depends mum output power of the PV system in each simulation
on the shading pattern. step, according to the irradiance, the temperature and the
Table 2
Overall MPPT efficiencies.
Method All year (%) Sunny days (%) Cloudy days (%) Partly cloudy days (%)
Proposed method 98.62 99.62 98.23 97.37
Fixed voltage 96.25 97.03 95.95 95.21
Fractional VOC 97.31 97.67 96.25 97.34
Perturbation and observation 97.99 99.40 95.54 96.83
Incremental conductance 96.79 99.14 95.60 94.28
102 Á.-A. Bayod-Rújula, J.-A. Cebollero-Abián / Solar Energy 109 (2014) 95–104
P (W)
100
lated method. In addition, as can be seen, it overcomes
the conventional methods regardless of the climatic condi- 80
days.
40
From Figs. 10–15, the real tracking is represented in
continuous thick black line and the ideal tracking is repre- 20
50
100
P (W)
40 80
60
30
40
20
20
10
0
0 10 20 30 40 50 60 70
V (V)
0
0 5 10 15 20 25
V (V) Fig. 15. Real tracking and ideal tracking during the establishment of
partial shading with the proposed method.
Fig. 12. Real tracking and ideal tracking during a quick increment of
irradiance with standard perturbation and observation.
Salas, V., Olı́as, E., Barrado, A., Lázaro, A., 2006. Review of the under non-uniform irradiance conditions. Energy Convers. Manage.
maximum power point tracking algorithms for stand-alone photovol- 62, 131–140.
taic systems. Sol. Energy Mater. Sol. Cells 90 (11), 1555–1578. Tey, K.S., Mekhile, S., 2014. Modified incremental conductance MPPT
Saloux, E., Teyssedou, A., Sorin, M., 2011. Explicit model of photovoltaic algorithm to mitigate inaccurate responses under fast-changing solar
panels to determine voltages and currents at the maximum power irradiation level. Sol. Energy 101, 333–342.
point. Sol. Energy 85 (5), 713–722. Tian, H., Mancilla-David, F., Ellis, K., Muljadi, E., Jenkins, P., 2012. A
Solar Generation VI, 2011. <http://www.epia.org/uploads/tx_epiapubli- cell-to-module-to-array detailed model for photovoltaic panels. Sol.
cations/Solar_Generation_6__2011_Full_report_Final.pdf>. Energy 86 (9), 2695–2706.
Syafaruddin, Karatepe E., Hiyama, T., 2012. Performance enhancement Yang, Y., Zhao, F.P., 2011. Adaptive perturb and observe mppt technique
of photovoltaic array through string and central based MPPT system for grid- connected photovoltaic inverters. Proc. Eng. 23, 468–473.