You are on page 1of 10

Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 122 (2015) 640–649
www.elsevier.com/locate/solener

A novel maximum power point tracking strategy based on


optimal voltage control for photovoltaic systems under
variable environmental conditions
Jian Zhao a, Xuesong Zhou b,⇑, Youjie Ma b, Wei Liu b
a
College of Electrical Engineering and Automation, Tianjin University, Tianjin 300071, China
b
School of Electrical Engineering, Tianjin University of Technology, Tianjin 300384, China

Received 30 June 2015; received in revised form 22 September 2015; accepted 28 September 2015
Available online 10 November 2015

Communicated by: Associate Editor Elias Stefanakos

Abstract

From the engineering cybernetics perspective, determining the output voltage of a solar cell corresponding to the maximum power
point (MPP) under variable temperature and irradiation conditions is essential for tracking the maximum power point. However, con-
ventional methods could not effectively achieve this goal, and approaches for utilizing this information do not exist, including the output
voltage and duty cycle at MPP, in the control process. This paper describes a novel method for maximum power point tracking (MPPT)
control in parameter optimization corresponding to variable environmental conditions. The results show a definable nonlinear relation
between variable environmental parameters and the output voltage of a solar cell at MPP both theoretically and experimentally. Addi-
tionally, this research gives the mathematical expression for it. Then, MPPT control rules are created based on the findings from a non-
linear relation that can respond at high speeds to variations in irradiation and the temperature of a solar cell. In view of this relationship,
we found that the output voltage at MPP is robust to irradiance in a certain range, and this directs the choices of irradiance in exper-
iments. Therefore, MATLABÒ/SimulinkÒ numerical simulations are presented to illustrate that this control algorithm can implement the
MPP of a PV generation system efficiently even in the low solar irradiation condition.
Ó 2015 Elsevier Ltd. All rights reserved.

Keywords: PV generation system; MPPT; Variable environmental condition; Optimal voltage control

Abbreviations: PV, photovoltaic; MPPT, maximum power point tracking; P&O, perturbation and observation; HC, hill climbing; INC, incremental
conductance; NN, neural network; MP, maximum power; DC, direct current; MPP, maximum power point; P, output power of solar arrays (in watt); I,
output current of solar arrays (in ampere); V, output voltage of solar arrays (in volt); Vpmax, Voltage at MPP is calculated by dV dP
¼ 0 (in volt); VOVC,
operating voltage at MPP by using OVC method (in volt); V pmax , operating voltage at MPP by using INC method (in volt); Ipmax, output current at MPP
by calculation (in ampere); Pmax, output power at MPP by calculation (in watt); Pomax, output power at MPP by using OVC method (in watt); P max ,
output power at MPP by using INC method (in watt); Rs, series resistor (in ohms); Rsh, parallel resistor (in ohms); I0, reverse saturation current of diode
(in ampere); Iph, is the photocurrent (in ampere); n, a dimensionless junction material factor; q, electron charge (1.602  1019 in coulombs); k, Boltz-
mann’s constant (1.38  1023 in joules per kelvin); T, solar cells’ temperature (in degrees Celsius); Isc, short circuit current at standard condition (in
ampere); Im, current at MPP under standard condition (in ampere); Voc, open circuit voltage (in volt); Vm, voltage at MPP under standard condition (in
volt); S, solar irradiance (in watt per square meters); I 0sc , short circuit current under nonstandard condition (in ampere); V 0oc , open circuit voltage under
nonstandard condition (in volt); I 0m , open circuit voltage under nonstandard condition (in ampere); V 0m , voltage at MPP under nonstandard condition (in
volt); Tref, solar cells’ temperature under standard condition (in degrees Celsius); Sref, solar irradiance under standard condition (in watt per square meter).
⇑ Corresponding author. Tel./fax: +86 022 60214081.
E-mail address: sjteam@163.com (X. Zhou).

http://dx.doi.org/10.1016/j.solener.2015.09.040
0038-092X/Ó 2015 Elsevier Ltd. All rights reserved.
J. Zhao et al. / Solar Energy 122 (2015) 640–649 641

1. Introduction rather than as the input–output voltage and current of the


DC–DC converter.
Studies on sustainable energy sources are actively pro- Vpmax of the solar cell depends on the cell’s environmen-
moted to mitigate the petroleum crisis and the increasing tal conditions, including the temperature of the solar cell
demand for energy, coupled with environmental issues. and the irradiance of sunlight (Bratcu et al., 2011; Liu
Solar photovoltaic (PV) generation has a very powerful et al., 2013). Thus, it is necessary to adopt the environmen-
potential because of the absence of fuel cost, the scalability tal parameters directly as a compute variable for Vpmax. In
in power and the simplicity and small amount of mainte- fact, there are many exciting MPPT methods for consider-
nance required by the PV system. The amount of solar ing the environmental parameters directly (Mutoh et al.,
energy reaching the Earth’s surface in a year is a thousand 2006; Li, 2014). However, these methods do not study
times the energy consumed in this period of time. So, solar the direct relationship between the environmental condi-
energy is a primary factor in all other processes of energy tion and the output voltage of the solar cell at MPP
production on earth (De Brito et al., 2013). (Vpmax), which is the control goal. Therefore, the primary
Because the amount of electric power generated by solar task of the proposed MPPT method in this paper is build-
arrays is always changing mainly with the solar cell temper- ing the direct relationship between the environmental
ature and irradiance, an efficient maximum power point parameters and Vpmax for variable conditions. This rela-
tracking (MPPT) design of a PV system is necessary (Yu tionship plays a key role in the theoretical study.
et al., 2002). Thus, many MPPT methods have been devel- First, this paper builds the nonlinear relation between
oped and implemented. They include the perturbation and variable parameters and the output voltage of the solar cell
observation (P&O) algorithm (Wasynczuk, 1983; Teulings at MPP. Then, the MPPT control strategy based on the
et al., 1993; Femia et al., 2005; Yang et al., 2010; relation with variable environmental parameters is pro-
Zegaoui et al., 2011; Petrone et al., 2011), which is known posed. The feasibility, availability and advantages of this
as the hill climbing (HC) method, the incremental conduc- MPPT method are analyzed by several MATLAB simula-
tance (INC) algorithms (Xiao and Dunford, 2004; Salas tion experiments for validating this method.
and Barrado, 2006; Liu et al., 2008; Zegaoui et al., 2011;
Zhou et al., 2010; Kish et al., 2012), the neural network
2. Principle of proposed MPPT control method
(NN) method (Chaouachi et al., 2010), and the fuzzy logic
method (Messai et al., 2011; Rajesh and Mabel, 2014).
2.1. PV panel model
These methods execute MPPT by three ways: searching
for the maximum power (MP) while checking the sign of
The photovoltaic panel ideal equivalent circuit for a PV
the differential coefficient (dP/dI) for the power (P) with
module is addressed in Mutoh et al. (2006).
respect to the current (I) in the P–I characteristics; search-
In the PV cell model, the expression for its output cur-
ing for MP while checking the sign of the differential coef-
rent I related to its output voltage V is given by Mutoh
ficient (dP/dV) for the power (P) with respect to the voltage
et al. (2006):
(V) in the P–V characteristics; and searching for MP while
   
checking the sign of the incremental conductance (dI/dV). q
I ¼ I ph  I 0 exp ðV þ IRs Þ  1
These methods have a common feature in that they search nkðT þ 273:15Þ
MP by increasing or decreasing the cycle coefficient of the V þ IRs
DC–DC converter with a fixed step approach to the cycle  ð1Þ
Rsh
coefficient at the maximum power point (MPP) by judging
the plus-minus of the differential coefficients. Then, the out- where I0 is the reverse saturation current (in amperes) of
put voltage of the solar cell approach to the output voltage the diode; Iph is the photocurrent (in amperes); n is a
of the solar cell at MPP–Vpmax. This type of control para- dimensionless junction material factor; q is the electron
digm could produce problems including a larger scale charge (1.602  1019 in coulombs); k is Boltzmann’s con-
delay, the inaccuracy of more detection circuits and sen- stant (1.38  1023 in joules per kelvin); and T is the solar
sors, and power oscillation under low irradiation condi- cell temperature (in degrees Celsius). An ideal PV cell has a
tions (Mutoh et al., 2006). very small equivalent series resistance Rs and a very large
From an engineering cybernetics perspective, the ulti- equivalent parallel resistance Rsh in the general engineering
mate goal of MPPT is reaching Vpmax rapidly and accu- application (Mutoh et al., 2006). Therefore, these two
rately by controlling the output voltage of the solar cell. internal resistances Rs and Rsh can generally be neglected
Namely, the reference input sign of MPPT is Vpmax. How- and Eq. (1) is further simplified to
ever, because these conventional methods of MPPT could
not gain the calculation algorithms of Vpmax in variable I ¼ I ph  I 0 fexpðAV Þ  1g ð2Þ
environment conditions, they have no way to utilize any
where
direct information of Vpmax. This is unreasonable for
MPPT control designs. In this paper, the proposed control q
A¼ ð3Þ
strategy adopts Vpmax directly as the input signal for MPPT nkT
642 J. Zhao et al. / Solar Energy 122 (2015) 640–649

The following formulas are present at two operating where A and Voc represent the variable environmental
points in the short and open circuits (Mutoh et al., 2006): parameter whose value is related to the parameters of solar
arrays, S and T. Therefore, Eq. (14) shows the relationship
I sc ¼ I ph ð4Þ
between output voltage V at MPP and environment param-
I 0 ¼ I sc expðAV oc Þ ð5Þ eters S and T. Namely, the solar generation system will
I m ¼ I sc f1  exp½AðV m  V oc Þg ð6Þ operate at MPP when the output voltage V equals to
  1
½Lambert W ðeAV oc þ1 þ eÞ  1. In addition, we defined V
1 Im A
A¼ ln 1  ð7Þ using Vpmax in this paper. It should be noted that Vpmax
V m  V oc I sc
can be easily calculated at the standard condition because
where Isc is the short-circuit current; Voc is the open-circuit the values of parameters (Isc, Voc, Im, Vm) are all given by
voltage; Im is the maximum power point current; and Vm is the PV array manufacturer. By contrast, at the nonstan-
the maximum power point voltage. All of the parameters dard condition, a new method should be studied for calcu-
above are measured by the PV array manufacturer under lating those parameters.
standard conditions of solar irradiance (S in watt per We define the parameters Isc, Voc, Im, and Vm with I 0sc ,
square meters) and temperature (T), which are 1000 V oc , I 0m , and V 0m at the nonstandard condition. Vpmax can
0

W/m2 and 25 °C, respectively. A represents a variable envi- be calculated based on Eq. (14) when parameters I 0sc , V 0oc ,
ronmental parameter whose value is related to the variable I 0m , and V 0m are given. On the other hand, because parame-
parameters of the solar cell, T and S, but here it is a con- ters I 0sc , V 0oc , I 0m , and V 0m are related to S and T, we must
stant for the standard condition. When using (4), (5) and determine the accurate mathematical relationship among
(7), the output current I and power P can be expressed them.
by a function of Isc, A, Voc and output voltage V, Under the condition of engineering accuracy, the
respectively. computational method of parameters I 0sc , V 0oc , I 0m , and V 0m
 
I ¼ I sc 1  eAV oc ðeAV  1Þ ð8Þ are given by the following equations (Rauschenbach,
 AV oc AV
 1980):
P ¼ I  V ¼ VI sc 1  e ðe  1Þ ð9Þ
S
All parameters are constants except for P and V when I 0sc ¼ I sc  ð1 þ aDT Þ ð15Þ
both S and T are definite in Eq. (9). Therefore, Eq. (9) S ref
shows the theoretical relationship between output power V 0oc ¼ V oc  ð1  cDT Þ lnðe þ bDSÞ ð16Þ
P and voltage V. To obtain the output voltage of solar S
arrays at MPP, the following works focus on the theoreti- I 0m ¼ I m  ð1 þ aDT Þ ð17Þ
S ref
cal analysis of the relationship between Vpmax and the envi-
V 0m ¼ V m  ð1  cDT Þ lnðe þ bDSÞ ð18Þ
ronment parameters.

2.2. Relationship between Vpmax and environment parameters where Tref is the temperature of the solar cell under the
standard condition; Sref is the solar irradiance under the
From the P&O algorithm, the slope of the P–V charac- standard condition; DT = T  Tref; DS ¼ S ref
S
 1; the typi-
teristic is equal to zero at MPP and is expressed by the fol- cal values of a, b, c are 0.0025/°C, 0.5, and 0.00288/°C,
lowing equation: respectively. The four parameters can be easily calculated
dP by using Eqs. (15)–(18) with standard parameters, which
¼0 ð10Þ are provided by the PV array manufacturer, and environ-
dV
ment parameters T and S, which are measured by sensors.
Substituting (9) into (10) gives For the record, the output characteristics of a solar cell
dP    determined by the formulas above only satisfy the request
¼ I sc 1  eAV oc eAV  1 þ VI sc eAV oc eAV ¼ 0 ð11Þ of engineering and the accuracy within 6% by using an
dV
experiment approach for the range of parameters for the
Reorganizing Eq. (11) gives: solar battery component, which are allowed by most PV
eAV AV oc ð1 þ AV Þ ¼ 1 þ eAV oc ð12Þ array manufacturers. However, the accuracy (6%) cannot
meet the requirements of MPPT control. Namely, V 0m ,
Multiplying both sides of the equal sign by e1þAV oc and which is calculated by Eq. (18), is not the real output volt-
reorganizing (12) gives: age at MPP and the formula dV dP
does not equal to zero. In
ð1 þ AV ÞeAV þ1 ¼ eð1 þ eAV oc Þ ð13Þ fact, determining the output voltage of solar arrays satisfy-
ing the formula dVdP
¼ 0 (Vpmax) is a must for maximum
Use the Lambert W function to obtain the solution for
power point track control. Therefore, we correct V 0m with
Eqs. (13) and (10):
Eq. (14) and obtain the mathematical expression of Vpmax
1  under variable environment conditions, as expressed by
V ¼ Lambert W ðeAV oc þ1 þ eÞ  1 ð14Þ
A the following equations:
J. Zhao et al. / Solar Energy 122 (2015) 640–649 643

1 A0 V 0oc þ1 of T and S are in the range of 0–70 and 0–1000 based on the
V pmax ¼ 0 ½Lambert W ðe þ eÞ  1 ð19Þ
A   real condition, respectively. The color of the curve in the
1 I 0m figures from red to blue represents the gradual change in
A0 ¼ 0 0 ln 1  0 ð20Þ
V m  V oc I sc voltage from high to low. It can be seen from Fig. 1a that
the different environmental conditions can correspond to a
Then, the corrective output current and power of the common operating voltage, and the operating voltage lim-
solar cell array at MPP–Ipmax and Pmax are given by the fol- its within narrow bounds of 42.637–69.12 correspond to a
lowing equations: wide range of environment conditions. This characteristic
 0 0 0  of optimal voltage puts forward strict accuracy require-
I pmax ¼ I 0sc 1  eA V oc ðeA V pmax  1Þ ð21Þ
ments for temperature and irradiance sensors. To study
P max ¼ V pmax  I pmax ð22Þ
the relationship between the voltage and environmental
By using the MATLABÒ/SimulinkÒ platform, the parameters more clearly, we project Fig. 1a into the
Vpmax–S–T surface characteristic based on Eq. (19) can Vpmax–S plane and Vpmax–T plane, respectively. Fig. 1b
be illustrated in Fig. 1a. Fig. 1b and c describes the perfor- shows that the Vpmax–S surface consists of a group of
mance of Vpmax–S and Vpmax–T, which are projected into approximate parabola from high to low corresponding to
the Vpmax–S plane and Vpmax–T plane in Fig. 1a, the temperature of the solar cell from low to high. Mean-
respectively. In the MATLABÒ model of the solar cell, while, Vpmax is relatively low and is sensitive to the change
all constant values that take from the data sheet of of variable S in the range of 500 to 800. In this range, it can
XINYU/L-1260 are used to perform the simulation. Addi- gain better results for MPPT control even though it has a
tionally, Isc, Voc, Im and Vm at standard conditions are low accuracy sensor of irradiance. Namely, the optimal
25.44 A, 66 V, 23.25 A and 54.2 V, respectively. The limits voltage of the solar array (Vpmax) is robust to S in a certain
range. This feature of Vpmax is beneficial to MPPT control.
Fig. 1c shows that the Vpmax–T surface consists of a group
of approximate oblique lines that have approximately the
same slope. So, this is necessary for high accuracy require-
ments for the temperature sensor in MPPT control.
For further research, the Pmax–S–T surface characteris-
tic based on Eq. (22) can be illustrated in Fig. 2a. Pmax is
the maximum power corresponding to variable environ-
ment conditions. Fig. 2b and c describes the performance
of Pmax–S and Pmax–T, where Fig. 2a is projected into
the Pmax–S plane and Pmax–T plane, respectively. It can
be seen from Fig. 2a that a simple relationship holds
between Pmax and S–T compared with Fig. 1a. Fig. 2b
shows that the approximation of a linear holds between
Pmax and S in which Pmax increases with an increase in S.
Pmax changes considerably and corresponds to the change
of variable S in the range of 0–1000, as it is from 500 to
800. However, Vpmax is nearly unchanged in the bounds
from 500 to 800. This further verifies that the optimal oper-
ation of the solar array is robust to S in a certain range.
Therefore, the operation of the solar array should be
avoided in the range for simulations and experiments of
MPPT. Meanwhile, Fig. 2c shows that T has a negligible
impact on Pmax but that it has a great influence on Vpmax
based on Fig. 1c. Thus, the main role of T is to influence
the optimal voltage Vpmax, and it can prove that the high
accuracy requirements of the temperature sensor in MPPT
control is even more necessary.
To summarize, Eq. (19) shows the relationship between
Vpmax and T and S under variable environment conditions.
Vpmax is the final reference input signal of MPPT. There-
fore, a functional relation between the control signal and
Fig. 1. (a) Vpmax–S–T characteristic surface of PV system under variable
environmental conditions; (b) Vpmax–S characteristic curves under variable the environmental parameters (T and S) can be built by
environmental conditions; and (c) Vpmax–T characteristics curves under Eq. (19). According to these relationships, an MPPT con-
variable environmental conditions. trol strategy will be proposed.
644 J. Zhao et al. / Solar Energy 122 (2015) 640–649

Fig. 3. Flowchart for the proposed MPPT control method.

(2) When the error between V and Vpmax is medium (i.e.,


the operating point is near MPP), the change in the
duty cycle is small to avoid the oscillation of nearby
MPP.
(3) When the error between V and Vpmax is almost equal
to zero (i.e., the operating point is reached and
remains near MPP), the duty cycle is not changed.
Fig. 2. (a) Pmax–S–T characteristic surface of PV cell under variable
environmental conditions; (b) Pmax–S characteristic curves under variable So, this MPPT control method can be called the ‘‘the
environmental conditions; and (c) Pmax–T characteristic curves under MPPT method based on optimal voltage control (OVC)”
variable environmental conditions. (Guo et al., 2013).

3. Proposed MPPT control method 4. Analysis and discussion of the simulation results

From the engineering cybernetics perspective, the ulti- To investigate the accuracy and performance of the pro-
mate goal of MPPT is to control the output voltage of posed method, a PV system includes one solar panel, a
the solar cell to reach Vpmax rapidly and accurately. Specif- DC–DC boost converter, a load and a control system that
ically, the reference input sign of MPPT is Vpmax. In this are considered and simulated on the MATLABÒ/Simu-
research work, the proposed control strategy adopts Vpmax linkÒ platform. The Simulink model of the PV system with
directly as the input signal for MPPT based on Eq. (19). the optimal voltage control is shown in Fig. 4. Fig. 5
The control procedures cited above are summarized in describes the subsystem model of the optimal voltage con-
the flowchart shown in Fig. 3. The inputs are T, S and troller, which includes the calculation block for Vpmax and
V, and these values can be measured by sensors. The out- the control logic block. In the MATLABÒ model of the
put is the error between V and Vpmax, which is calculated solar cell, Isc, Voc, Im and Vm are the same as Section 2.2
by Eq. (19) with T, S. Obviously, the error is the control at standard conditions. The inductance and the capacitor
signal for MPPT, which determines the optimal duty cycle of the Boost DC/DC converter are the ideal components;
of the DC–DC converter. the snubber resistance, internal resistance and forward
The whole proposed method includes only two subsets: voltage of the diode are 500 X, 0.001 X and 0.8 V, respec-
Calculate and Compare. In the control process, Vpmax is tively; IGBT with the 0.001 X internal resistance and
constant when T and S are determinate. Consequently, 105 X snubber resistance is chosen as the switch; the load
we can draw the following three conclusions: resistance is 500 X. Other parameters such as R1, L, C
and C1 are 1 X, 0.01 H, 2 mF and 2 mF, respectively. Real
(1) When the error between V and Vpmax is large (i.e., the environment data collected at Tianjin, China for a sunny,
operating point is far from MPP), the change in the cloudy and rainy day from 5 a.m. to 6 p.m. are used to ana-
duty cycle is large and reaches MPP rapidly. lyze the performance of the proposed MPPT strategy.
J. Zhao et al. / Solar Energy 122 (2015) 640–649 645

Fig. 4. Simulink model of PV system with proposed MPPT control.

Fig. 5. Subsystem model of optimal voltage control.

According to the control strategy proposed that controls in low irradiation are larger than in high irradiation. In
the PV system that operates at MPP, the real-time value of contrast, the errors between VOVC and Vpmax are almost
Vpmax should be calculated and the simulation results are a uniform distribution and are not affected by the environ-
shown in Table 1. The columns with and without super- ment parameters T and S. By comparing POmax with the
script ‘‘*” represent the data of the MPPT controller using P max column, we can see that the values of POmax are
the INC algorithm and the proposed strategy, respectively; slightly larger than P max and that the deviations in low irra-
VOVC is the true output voltage of the PV array operating diation are larger than in high irradiation. The deviations
at MPP by using the OVC method; the dP/dV column rep- shows similar characteristics with the errors of
resents the degree of deviation between the actual power at V pmax  V pmax . Therefore, a conclusion can be drawn that
Vpmax and the theoretical power at MPP, with plus or this MPPT method based on optimal voltage control is fea-
minus corresponding to the left or right of the actual power sible, where the tracking performance is not affected by
relative to the theoretical power. Focusing on the dP/dV environment parameters.
column, we can find that all values approach zero infinitely, To compare the performance of the proposed MPPT
which means that Vpmax equals to the theoretical voltage of method with the INC algorithm, the following three cases
MPP practically. It can be seen from the VOVC, Vpmax and are simulated and analyzed:
V pmax rows that the voltage values VOVC and V pmax are
approximately equal to the corresponding value Vpmax Case 1: Normal operating conditions.
when the PV generation system  operates
 at MPP and the Case 2: Sustained and varied low solar irradiance
average error between VOVC V pmax and Vpmax is 0.13 conditions.
Case 3: Fast variation of the cell temperature and the
(0.4111). In addition, the errors between V pmax and Vpmax solar irradiance in global.

Table 1
Experimental results under various (S, T).
(S, T) (W/m2, °C) VOVC (V) Vpmax (V) V pmax (V) dP/dV POmax (W) P max (W)
(200, 10) 47.26 47.3297 48.47 4.4409e15 212.807 211.6
(300, 10) 48.36 48.5308 49.78 4.4409e15 327.3114 325.7
(400, 10) 49.96 49.7069 50.91 3.5527e15 446.9907 445.1
(800, 30) 51.126 51.1882 51.21 3.5527e15 968.4473 968.4
(900, 30) 52.17 52.1942 52.21 5.6843e14 1111 1110.5
(1000, 30) 53.13 53.1815 53.23 1.4211e14 1258.2 1257.7
(800, 40) 48.68 49.6924 49.68 3.5527e15 963.3621 963.3
(900, 40) 50.67 50.6690 50.68 1.0658e14 1105.1 1105
(1000, 40) 51.59 51.6275 51.65 3.5527e15 1251.1 1251
646 J. Zhao et al. / Solar Energy 122 (2015) 640–649

4.1. Normal operating conditions

To assess the features of proposed method, the perfor-


mances obtained by using the OVC method will be com-
pared with the corresponding performances obtained by
using the INC algorithm. Of course, to perform a meaning-
ful comparison, both techniques must be properly opti-
mized. Fig. 6 shows the experimental results by adopting
the OVC method and the INC algorithm.
In Fig. 6a, the numerical results obtained by adopting Fig. 7. Time-varying irradiation profile.
the INC algorithm are shown. The curves represent the
time-domain behavior of the power extracted by the PV
module under stationary environmental conditions (S =
(800, 1000) W/m2, T = 25 °C) during the turn on transient
of the boost converter until the steady-state is settled. The
curves shown in Fig. 6b refer to the corresponding quanti-
ties obtained by adopting the OVC method.
It can be seen from Fig. 6a that the INC and the OVC
algorithms can lead, at the steady-state, to an extracted
power nearly equal to PMPP = (1260, 970.3) W, corre-
sponding to the irradiation parameter S = (800, 1000)
W/m2. From Fig. 6a, we can see that the output power
of solar arrays using the INC algorithm at the steady-
state has an oscillation within narrow bounds, which
slowly become greater with a decrease in the value of S.
Conversely, the corresponding behaviors using the OVC
method remain non-vibrational throughout the entire con-
trol operation. Thus, the output power when using the
OVC method has a better steady-state performance than

Fig. 8. (a) Time-domain behavior of the powers produced by the PV


sources, under time-varying conditions (INC algorithm); (b) correspond-
ing quantities obtained by adopting OVC method; (c) comparison of the
time-domain behavior of the output power.

the INC algorithm in normal operating conditions. It


should be noted that the non-vibration state is entirely
unrelated to the value of S because the OVC method calcu-
lates the operating voltage at MPP accurately. Meanwhile,
Fig. 6. (a) Time-domain behavior of the powers produced by the
PV sources, under stationary environmental conditions (S = (800, 1000)
by comparing the settling time in Fig. 6a and b, it is possi-
W/m2, T = 25 °C, IC); and (b) corresponding quantities obtained by ble to show that a higher speed is obtained by using the
adopting OVC method. OVC method.
J. Zhao et al. / Solar Energy 122 (2015) 640–649 647

Table 2
Calculated results in Case 2.
Fig. 8 0 6 t 6 1 (s) Fig. 8 0.55 6 t 6 0.8 (s)
EOVC (W s) 398.99 83.722
EINC (W s) 390.17 81.354
g 2.2605% 2.9107%

Fig. 9. Fast variations of (a) solar cell temperature and (b) solar
irradiance.

Fig. 10. The MPPT trajectories for INC and OVC algorithms under (a)
4.2. Sustained and varied low solar irradiance conditions fast variations of solar cell temperature and solar irradiance, (b) fast
variations of solar irradiance and (c) fast variations of solar cell
This investigation is implemented to assess and compare temperature.
the performance of INC and OVC algorithms under sus-
method is larger than the value of the output power using
tained and varied low solar irradiance conditions. As an
the INC algorithm when S changes from 400 to 300.
example, the irradiation profile shown in Fig. 7 has been
From Fig. 8c, the simulation results show that the PV
considered. Fig. 8a and b shows the numerical results
system could output more power with the proposed algo-
obtained by adopting the INC and OVC methods, respec-
rithm than by using the incremental conductance (INC)
tively. The curves in Fig. 8 represent the time-domain
algorithm. Based on the mathematical definition of work,
behavior of the power extracted by the PV module. From
the value of work equals to the value of the area bounded
Fig. 8a, we can see that the output power of solar arrays
by the power curve and time axes. By using the
using the INC algorithm at steady-state has a high fre-
MATLABÒ/SimulinkÒ platform, the area bounded by
quency oscillation with a large amplitude at the low solar
the power curve and time axes could be calculated with
irradiance condition. Conversely, the corresponding behav-
0.0001 precision. So, the increment rate of output energy
iors using the OVC method remain non-vibrational
(g) in Case 2 using the proposed method could be analyzed
throughout the entire control operation from Fig. 8b.
based on the following equation:
Fig. 8c, where a comparison of the time-domain behavior
of the power produced by the PV modules under time- g ¼ ðEOVC  EINC Þ=EINC ð23Þ
varying environmental conditions is reported, clearly
demonstrates that the value of the red curve is nearly equal where g is the increment rate of output energy; EOVC is the
to the peak value of the black curve. It is evident that even output energy of the PV system by using the proposed
if the solar irradiance varies under a low value, the PV sys- algorithm (W s), i.e., the area bounded by the power curve
tem could operate at MPP steadily and with a non- and time axes using the proposed algorithm; EINC is the
vibration by using the OVC method. It should also be output energy of the PV system using the INC algorithm
noted that the value of the output power using the OVC (W s), i.e., the area bounded by the power curve and time
648 J. Zhao et al. / Solar Energy 122 (2015) 640–649

Table 3
Calculated results in Case 3.
Fig. 10a 0 6 t 6 1 (s) Fig. 10b 0 6 t 6 1 (s) Fig. 10c 0 6 t 6 1 (s) Fig. 10a 0.8 6 t 6 1 (s) Fig. 10b 0.8 6 t 6 1 (s)
EOVC (W s) 775.01 788.44 1228.9 88.908 88.923
EINC (W s) 761.39 775.56 1214.1 86.629 86.647
g 1.7888% 1.6607% 1.2190% 2.6307% 2.6267%

axes using the INC algorithm. The calculated results are of the solar energy is output energy when using the pro-
shown in Table 2. posed method more than the INC algorithm in global.
Focusing on the g row of Table 2, we find that approx- From Figs. 6, 8 and 10, the simulations results show that
imately 2.2605% of the solar energy is output energy when the proposed OVC method has adequate accuracy both
using the proposed method more than the INC algorithm under stationary and time-varying environmental condi-
in global. When time is in the range of 0.55–0.8 s (then irra- tions. In addition, the setting time of the output power is
diance keeps 300 W/m2), 2.9107% of the solar energy is less than 0.04 s. At steady-state, the output power curves
output energy when using the proposed method more than are non-vibrating under all experimental conditions when
the INC algorithm. the proposed method is used. This means that the OVC
method is instrumental in eliminating the power oscillation
4.3. Fast variation of the cell temperature and the solar under the low solar irradiance condition.
irradiance in global
5. Conclusions
To investigate and verify the performance and accuracy
of the proposed OVC method under a rapidly changing From the engineering cybernetics perspective, a novel
solar cell temperature and the solar irradiance, a step method (OVC) has been presented and discussed in this
change in global is applied to the solar cell temperature paper. This method is based on the optimal operating volt-
and the solar irradiance, as represented in Fig. 9. The out- age of solar arrays, which is calculated and corrected accu-
put power trajectories of the solar array for INC and OVC rately by determining the relationship between the optimal
algorithms are also shown in Fig. 10a. In Fig. 10b, varia- operating voltage and the external environmental parame-
tions in cell temperature are assumed to be constant at ters. Its main advantages are the simplicity of implementa-
25 °C, and solar irradiance changes similar to Fig. 9b. This tion, high MPPT accuracy, and non-vibration in global
figure shows similar results to Fig. 9a. In Fig. 10c, varia- both under stationary and time-varying environmental
tions in solar irradiance are assumed to be constant at conditions.
1000 W/m2, and the cell temperature changes similar to A numerical analysis has been performed by using the
Fig. 9a. Fig. 10a–c shows that the OVC algorithm can find INC algorithm as a benchmark reference MPPT technique.
MPP rapidly and accurately regardless of whether we The numerical results confirm the validity of the proposed
change the solar cell temperature, the solar irradiance, or method because the performances of OVC are always bet-
both the temperature and the irradiance. It should be noted ter with respect to the corresponding performances obtain-
that the output power curves when using the OVC method able by adopting the INC algorithm.
have no oscillation globally in these conditions. Con- In addition, the proposed method only needs three sen-
versely, the output power of solar arrays when using the sors for implementing less than the need of the INC algo-
INC algorithm at steady-state has an oscillation in global. rithm, which could decrease the total measurement error,
It is evident that the OVC method has better accuracy and and the OVC method is unrelated to the topology of the
time response in comparison with the INC algorithm. In DC/DC converter. It is beneficial for the modularity and
addition, Fig. 10c shows that the output power of the solar the ‘‘plug and play” of the MPPT controller.
arrays has a small relationship with cell temperature. This
result concurs with the result shown in Fig. 2c. Acknowledgements
Based on Eq. (23), the increment rate of output energy
(g) in Case 3 when using the proposed method could be This work was supported by National Natural Science
analyzed and the calculated results are shown in Table 3. Foundation of China (NO. 50877053) and Natural Science
Focusing on the g row of Table 3, g are 1.7888%, Foundation of Tianjin of China (NO. 09JCYBJC07100).
1.6607% and 1.2190% corresponding to Fig. 10a–c in glo-
bal, respectively. When time is in the range of 0.55–0.8 s References
(then irradiance keeps 400 W/m2), g are 2.6307% and
Bratcu, A.I., Munteanu, I., Bacha, S., Picault, D., Raison, B., 2011.
2.6267% corresponding to Fig. 10a and b, respectively. Cascaded DC–DC converter photovoltaic systems: power optimiza-
From Tables 2 and 3, the calculated results show that tion issues. IEEE Trans. Ind. Electron. 58 (2), 403–411.
approximately 2.5% of the solar energy is output energy Chaouachi, A., Kamel, R.M., Nagasaka, K., 2010. A novel multi-model
when using the proposed method more than the INC algo- neuro-fuzzy-based MPPT for three-phase grid-connected photovoltaic
system. Sol. Energy 84, 2219–2229.
rithm in low irradiance. Additionally, approximately 1.5%
J. Zhao et al. / Solar Energy 122 (2015) 640–649 649

De Brito, M.A.G., Galotto, L., Sampaio, L.P., de Azevedo e Melo, G., Rauschenbach, H.S., 1980. Solar Cell Array Design Handbook: The
Canesin, C.A., 2013. Evaluation of the main MPPT techniques for Principles and Technology of Photovoltaic Energy Conversion, first
photovoltaic applications. IEEE Trans. Ind. Electron. 60 (3), 1156– ed. Springer.
1167. Rajesh, R., Mabel, M.C., 2014. Efficiency analysis of a multi-fuzzy logic
Femia, N., Petrone, G., Spagnuolo, G., Vitelli, M., 2005. Optimization of controller for the determination of operating points in a PV system.
perturb and observe maximum power point tracking method. IEEE Sol. Energy 99, 77–87.
Trans. Power Electron. 20 (4), 963–973. Salas, V., Barrado, A., 2006. Review of the maximum power point
Guo, Q., Sun, H., Zhang, M., 2013. Optimal voltage control of PJM smart tracking algorithms for stand-alone photovoltaic systems. Sol. Energy
transmission grid: study, implementation, and evaluation. IEEE Trans. Mater. Sol. Cells 90, 1555–1578.
Smart Grid 4 (3), 1665–1674. Teulings, W.J.A., Marpinard, J.-C., Capel, A., O’Sullivan, D., 1993. A
Kish, G.J., Lee, J.J., Lehn, P.W., 2012. Modeling and control of new maximum power point tracking system. In: 24th Annual IEEE
photovoltaic panels utilizing the incremental conductance method Power Electronics Specialists Conference, Seattle, WA, June 20–24,
for maximum power point tracking. IET Renew. Power Gen. 6, 259– pp. 833–838.
266. Wasynczuk, O., 1983. Dynamic behavior of a class of photovoltaic power
Liu, F., Duan, S., Liu, F., Liu, B., Kang, Y., 2008. A variable step size systems. IEEE Trans. Power Appar. Syst. PSA-102 (9), 3031–3037.
INC MPPT method for PV systems. IEEE Trans. Ind. Electron. 55, Xiao, W., Dunford, W.G., 2004. A modified adaptive hill climbing MPPT
2622–2628. method for photovoltaic power systems. In: Power Electronics
Liu, Y.-H., Liu, C.-L., Huang, J.-W., Chen, J.-H., 2013. Neural-network Specialists Conference, June 20–25, pp. 1957–1963.
based maximum power point tracking methods for photovoltaic Yu, G.J., Jung, Y.S., Choi, J.Y., Choy, I., Song, J.-H., Kim, G.S., 2002. A
systems operating under fast changing environments. Sol. Energy 89, novel two-mode MPPT control algorithm based on comparative study
42–53. of existing algorithms. In: Conference Record of the Twenty-Ninth
Li, S., 2014. A MPPT control strategy with variable weather parameter IEEE Photovoltaic Specialists Conference, May 19–24, pp. 1531–1534.
and no DC/DC converter for photovoltaic systems. Sol. Energy 108, Yang, B., Li, W., Zhao, Y., He, X., 2010. Design and analysis of a grid
117–125. connected photovoltaic power system. IEEE Trans. Power Electron.
Mutoh, N., Ohno, M., Inoue, T., 2006. A method for MPPT control while 25 (4), 992–1000.
searching for parameters corresponding to weather conditions for PV Zhou, X., Song, D., Ma, Y., Chen, D., 2010. The simulation and design
generation systems. IEEE Trans. Ind. Electron. 53, 1055–1065. for MPPT of PV system based on incremental conductance method.
Messai, A., Mellit, A., Guessoum, A., Kalogirou, S.A., 2011. Maximum In: 2010 WASE International Conference on Information Engineering
power point tracking using a GA optimized fuzzy logic controller and (ICIE), August 14–15, Beidaihe, Hebei, pp. 314–317.
its FPGA implementation. Sol. Energy 85, 265–277. Zegaoui, A., Aillerie, M., Petit, P., Sawicki, J.P., Charles, J.P., Belarbi, A.
Petrone, G., Spagnuolo, G., Vitelli, M., 2011. A multivariable perturb- W., 2011. Dynamic behavior of PV generator trackers under irradi-
and-observe maximum power point tracking technique applied to a ation and temperature changes. Sol. Energy 85, 2953–2964.
single-stage photovoltaic inverter. IEEE Trans. Ind. Electron. 58 (1),
76–84.

You might also like