Solar Energy 178 (2019) 37–47

Contents lists available at ScienceDirect

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

An improved P&O algorithm integrated with artificial bee colony for T

photovoltaic systems under partial shading conditions
⁎
Deepthi Pilakkat , S. Kanthalakshmi
Dept. of Electrical and Electronics Engineering, PSG College of Technology, Coimbatore 641004, India

A R T I C LE I N FO A B S T R A C T

Keywords: For an efficient Photovoltaic (PV) system, tracking of true maximum power point (MPP) is essential. Therefore
Artificial bee colony algorithm (ABC) the maximum power point tracking (MPPT) controller is mandatory for harvesting maximum power from the
Maximum power point tracking (MPPT) solar panel. Perturb and Observe (P&O) MPPT is the simplest and most widely used low-cost MPPT method for
Partial shading condition (PSC) tracking MPP. The major drawback of P&O is steady state oscillations around MPP and tracking of local MPP
Perturb and observe (P&O) algorithm
(LMPP) instead of global MPP (GMPP) under partial shading conditions (PSC). Thus, this paper proposes a
Solar PV systems
modified P&O MPPT that can be used under PSC effectively, by integrating Artificial Bee Colony (ABC) algorithm
in the first stage and P&O algorithm in the second stage. In the proposed method GMPP is first tracked by calling
ABC algorithm followed by the P&O algorithm for LMPP. Thus the local search ability of P&O and global search
ability of ABC are effectively combined to produce optimum duty cycle for the boost converter in a fast and
efficient way. In this paper, the proposed ABC-PO algorithm is implemented in MATLAB/Simulink model and it
is compared with different MPPT algorithms such as P&O, Incremental conductance (INC) and ABC. The si-
mulation results clearly depicted that the proposed ABC-PO algorithm gives more than 99.5% efficiency under
PSC.

1. Introduction MPPT technique ever known. Conventional MPPT methods vary in

convergence speed, number of sensors used, and cost of the system
By the end of 2022 it is targeted by the Government of India that the (Esram and Chapman, 2007). It is to be noted that the conventional
total capacity of renewable energy to be 175 GW including the 100 GW MPPT methods are effective under uniform irradiance conditions.
solar power. Due to the growing energy needs and concern about en- Large numbers of PV panels are connected in series or parallel in
vironmental issues, global demands for the utilization of renewable order to get the required amount of current and voltage. In such systems
energies have shown much interest. Renewable energy sources like partial shading occurs due to shadow of trees, buildings or moving
geothermal, wind, solar, etc are attracting more consideration as al- clouds and PV modules will receive different solar insolation. Due to the
ternate energy sources. Among the renewable energy sources, the solar PSC the power vs voltage curve becomes more complex with the pre-
photovoltaic (PV) energy has been widely utilized in many applications sence of multiple peaks. Among these, only one peak represents true
due to its advantages such as direct electric power form, easy main- MPP called global MPP (GMPP) and other MPPs are referred as local
tenance, no noise, and pollution free. As the interest of this paper is MPPs (LMPP). The conventional methods discussed in the above
contracted to solar PV energy, it is detailed here after. paragraph are unable to track the true MPP and may fall into any of the
Increasing the efficiency of a solar panel has become a research LMPP. This will result in power loss and lead to have low efficiency of
interest to make the solar energy economical. In this point of view, PV system. So, by updating this algorithm to track the GMPP in the case
several MPPT techniques are developed to extract maximum power of shading phenomena will definitely improve the performance of the
from the PV system. Many conventional MPPT techniques have been system. This can be achieved by including an appropriate algorithm in
developed so far. Those techniques include Perturb and Observe (P&O) the first stage before applying the P&O (De Brito et al., 2013).
method (Kumar et al., 2018; Femia et al., 2005), the Incremental In recent years, the researchers have shown very much interest in
Conductance (INC) method (Mei et al., 2011), short circuit current observing the characteristics of many creatures in the world. Various
(Noguchi et al., 2002), open circuit voltage (Dorofte et al., 2005) etc. characteristics like foraging and mating behavior of insects, birds, an-
Among the various methods, P&O method is the simplest and easiest imals etc paved the way to develop many biologically inspired

⁎
Corresponding author.
E-mail addresses: [email protected] (D. Pilakkat), [email protected] (S. Kanthalakshmi).

https://doi.org/10.1016/j.solener.2018.12.008
Received 23 March 2018; Received in revised form 9 November 2018; Accepted 2 December 2018
0038-092X/ © 2018 Elsevier Ltd. All rights reserved.
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

optimization algorithms and to find solutions for the numerical opti-

mization problems. The swarm intelligence (SI) algorithm that is being
popularly used by researchers was also inspired from the social beha-
vior of species. Different SI algorithms like Particle Swarm Optimization
(PSO) (Ishaque and Salam, 2012), Artificial Bee Colony algorithm
(ABC) (Babar and Crăciunescu, 2014), Ant Colony Optimization (ACO)
(Sundareswaran et al., 2016), Firefly Algorithm (FA) (Sundareswaran
et al., 2014), have been proposed to find out the true MPP. The paper
Mohanty et al. (2014) compares the performance of different MPPT
techniques that are presently used in a solar PV system.
In this work, some modifications to the P&O algorithm has been
proposed to improve its performance under PSC. In the proposed
method, the biologically inspired ABC algorithm is integrated with the
conventional P&O algorithm in order to track the GMPP efficiently
under PSC. The global search ability of ABC is first utilized followed by
the local search ability of P&O to get a fast and efficient MPPT algo-
rithm. The most commonly used conventional MPPT algorithms, P&O
and INC is briefly explained in Sections 2 and 3 respectively. The newly
developing biologically inspired ABC algorithm is described in Section
4. The proposed ABC-PO algorithm is introduced in Section 5. Modeling
of PV panel under PSC is described in Section 6. Section 7 discusses the
simulation results and conclusion is given in Section 8.

2. Perturb & observe algorithm

The P&O method is performed on the basis of periodical perturba-

tions (i.e. incrementing or decrementing) of the panel voltage and
measure the difference between the PV output power with that of the
preceding perturbation cycle (Faranda et al., 2008). The first step is to
measure the initial PV voltage and current and to calculate the corre-
Fig. 1. Flowchart of P&O algorithm (Pilakkat and Kanthalakshmi, 2018).
sponding power. Considering a small perturbation in voltage (ΔV) or
duty cycle (ΔD) of the DC-DC converter in one direction, corresponding
power is calculated and is compared with the previous value of power. 4. Artificial bee colony algorithm
If the change in power (ΔP) is positive, then the perturbation is in the
correct direction; otherwise the direction should be reversed (ie, de- 4.1. Introduction to ABC
crementing ΔV). The flowchart shown in Fig. 1 illustrates the P&O al-
gorithm in detail. Artificial bee colony is a reasonably new stochastic algorithm for
One of the disadvantages caused by P&O method is its oscillations global optimization which was introduced by Karaboga in 2005. ABC
around MPP. By dropping the step size, these oscillations can bring algorithm is a biologically inspired approach motivated by the in-
down into some extend. Even though, small step size decelerates the telligent foraging activities of honey bee colonies (Karaboga and Akay,
MPP. 2011). The ABC consists of three fundamental groups. They are (A)
Another main drawback of P&O algorithm is that it fails to track employed bees, (B) onlooker bees and (C) scout bees. Fifty percent of
GMPP in a PV system under PSC and hence settles to LMPP. This brings the colony is possessed by employed bees and the other fifty percent
the consequence of reduced energy extraction and hence the efficiency composed of onlooker bees. ABC algorithm is as follows: Initial role is
of system gets decreased. So, this algorithm must be reorganized to played by the employed bees. At the first step, an employed bee goes to
follow the global MPP instead of local MPP, in the case of shading the previously known food source location to obtain nectar. Then it
phenomena. This can be achieved by adding an appropriate algorithm later on chooses a new food source location with the help of visual
in the first stage for tracking GMPP by scanning the PV curve. communication with its neighbors. If the new location has more nectar
In this paper, ABC algorithm is chosen as the first stage algorithm than that of the previous locations, the employed bee forgets the pre-
for tracking GMPP. vious location and memorizes only the new location. With the help of
waggle dance movements, the new location with more nectar is con-
veyed to the onlooker bees. Thus the onlooker bees waiting at the hive
3. Incremental conductance algorithm move towards the best food source position. Now the Employed bee
whose food source has been discarded will be converted as scout and
The idea behind INC algorithm is based on the slope of the P-V begins search of a fresh location.
curve. The slope will be zero at MPP; negative at right side of MPP and
positive at left side of MPP (Yetayew et al., 2016). The flow chart ex- 4.2. ABC for the optimization of MPPT
plaining the process of INC algorithm is given in Williams et al. (2016).
Eq. (1) describes the basic principle of INC algorithm. Many researchers have applied ABC algorithm for determining MPP
in solar PV systems. Bilal in 2013 used ABC algorithm for MPPT under
ΔI I different irradiance conditions and compared the results with P&O
⎧ At MPP , ΔV
= −V
⎪ ΔI I (Bilal, 2013). He concluded that, for lower irradiance levels ABC gives
At left side of MPP , >−
⎨ ΔV V comparable results to P&O whereas in higher irradiance, ABC outper-
⎪ At right side of MPP , ΔI I
<− forms P&O.
⎩ ΔV V (1)
A novel approach based on ABC has been proposed by

38
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

Sundareswaran et al. (2015) for tracking MPP under PSC in a PV (1) Initialization and evaluation phase:
system. The authors compared the results with PSO and Enhanced P&O
(EPO) methods and proved that ABC has better performance than these In this phase, consider the colony size to be S. Let one half of the
two methods. They clearly demonstrated the lesser oscillations in the colony represents the employed bees (Np) and the other half represents
output power and fast tracking characteristics of the proposed ABC the onlooker bees. All the employed bees are distributed at different
algorithm by experimental results. food source positions using the following equation:
For MPPT, in order to track maximum operating point for certain
x i = dmin + rand [0, 1](dmax − dmin ) (2)
irradiance and weather conditions, a DC-DC converter is inserted be-
tween PV module and load. The duty cycle of the switch of DC-DC where i = 1, 2, …, Np and dmin and dmax represent minimum and
converter is always adjusted in such a way as to operate the PV panel at maximum values of duty ratio of the DC–DC converter. dmin and dmax is
its MPP. While applying ABC algorithm for optimizing the duty ratio for taken as 0.1 and 0.9 respectively.
the converter, the output power of the PV system is considered as the Once initialization is performed, Maximum Cycle Number (MCN) is
nectar amount. The decision variable in this case is the duty ratio, set. For each cycle the search process repeats and evaluation of the
which is termed as food source position. The main challenge regarding quantity of nectar (output power) is done. Calculation of the output
ABC algorithm is to procure global optimum solution with minimum power is implemented using MATLAB/Simulink model and the corre-
computational time. The first step in this regard is to minimize the sponding mcode.
objective function, for that, the considered optimization problem has to
be first converted as the problem for finding out the best parameter (2) Identifying the new food source location:
solution. Then, the artificial bees arbitrarily come across a population
of primary decision variables and then periodically adjust them by Next stage is to track the food source position with maximum nectar
operating against more accurate solutions adopting greedy selection (GMPP) in the search space. This is performed in 2 phases as explained
procedure while leaving poor solutions. below.
A PV operated DC motor pump utilizing ABC algorithm for MPPT as
well as speed controller is proposed by Oshaba et al. (2015). In this (a) Employed bee phase:
paper the MPPT and speed control problem is defined as optimization
problem for ABC algorithm to find out the parameters of PI controllers. Each employed bee updates its new position in the vicinity using the
The reference speed of the DC series motor and MPP for DC-DC con- following equation:
verter is successfully tracked by ABC. The performance of the proposed
x i − new = x i + ∅i [x i − xk ] (3)
method is validated by simulation results.
The flowchart of ABC algorithm is given in Fig. 2 and the step by where k = (1, 2, …Np) which is randomly chosen indexes and is not
step procedure of ABC algorithm is explained below. equal to i. ∅i is an arbitrarily selected variable between [−1, 1]. Once
the bee explores a new food source location it checks the quantity of

Fig. 2. Flowchart of ABC algorithm.

39
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

Fig. 3. Flowchart of ABC-PO algorithm.

Table 1 Table 2
Parameters used for both ABC and ABC-PO algorithms. Symbols and descriptions in Pv module.
Components Specification Symbol Description Value Unit

Colony size 6 NP Number of parallel connected cells

Number of food sources 3 NS Number of series connected cells
Dimension (D) 1 q Electron charge 1.6 * 10−19 C
Maximum cycle number (MCN) 30 VPV Output voltage of PV module V
RS Series resistance of PV module Ω
A Ideality factor 1.5
k Boltzman constant 1.3805 * 10−23 J/K
Top Module operating temperature K
Tref Reference temperature k
Ki Short circuit current temperature coefficient A/K
Kv Open circuit voltage temperature coefficient V/K
G Irradiance W/m2
Eg Band gap energy 1.1 eV
ISC Short Circuit current A
VOC Open circuit voltage V

Table 3
Specifications of Vikram Solar Eldora-37 module.
Fig. 4. Equivalent circuit of a solar cell.
Description Specification

nectar at that point. If the quantity of nectar in the new position is more Maximum power (Pmax) 37 W
than that of the previous food position, then the employed bee con- Open circuit voltage (VOC) 21.8 V
tinues to remain in the new updated position. Otherwise employed bee Short circuit current (ISC) 2.4 A
Maximum power point voltage (Vmpp) 17 V
moves back to the old position.
Maximum power point current (Impp) 2.25 A
Module efficiency 10.10%
(b) Onlooker bee phase: Short circuit current temperature coefficient 0.04%/°C

The amount of nectar present at various food source locations is

conveyed by the employed bees to the onlooker bees with the help of explore a new location.
waggle dance movements. Through greedy search process, the onlooker The food source location with maximum nectar is determined by
bees choose the best food source location and proceed to exploit that comparing the probability factor associated with various food locations.
location. During the greedy search process they retain only the max- The probability is calculated with the help of Eq. (4).
imum food source location in its memory. Once an employed bee’s food
source is completely exploited, that bee turns into a scout and goes to

40
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

Fig. 5. Simulation model of PV system.

Fig. 7. Step variation in irradiance profile from (a) PSC1 to PSC2 (b) PSC2 to
PSC3.

Table 4
Shading patterns taken for the study.
Case Shading pattern LMPP (W) GMPP (W)

PSC1 [1000, 1000] 76.45 76.45

PSC2 [1000, 500] 36.65 42.88
PSC3 [300, 800] 26 29.79

even if there is any change in solar irradiance occurs. The real time
implementation of ABC algorithm in PV system is described in paper
(Oshaba et al., 2017).
Fig. 6. For different partial shading patterns, (a) I-V curves and (b) P-V curves.
5. Proposed ABC-PO algorithm

fiti
pi = The P&O algorithm, one of the most commonly employed MPPT
N
∑Np= 1 fitN (4) tracking scheme for PV power generation system miserably fails to re-
cognize global MPP under PSC. It even gets trapped in one of the local
where fiti is the fitness factor of ith location and is given by
MPPs. Therefore there arises the necessity for a more efficient GMPP
1 tracking algorithm. The proposed improved P&O algorithm combines
⎧ ; if Objval ≥ 0 ⎫
fiti = 1 + Objvali the advantages of both ABC and P&O algorithm overcoming their de-
⎨1 + abs (Objval ); Otherwise ⎬ merits in the best manner. As a result the overall aim of tracking GMPP
⎩ i ⎭ (5)
is achieved by the proposed algorithm very effectively.
The proposed method is simply integrating a checking algorithm
(c) Finishing phase:
onto the P&O algorithm. The checking algorithm used is ABC, which is
to identify the GMPP. Once the global MPP is detected, the operation of
The overall process terminates when there is no further improve-
P&O algorithm will takes place which will lead to higher accuracy by
ment in the power output of the PV system or when the MCN number is
operating at the previously detected GMPP. The overall process flow of
met. Then the DC-DC converter operates with the obtained optimal
the proposed algorithm is depicted using flowchart in Fig. 3.
duty cycle.
A simulation study was carried out to validate the performance of
On the other hand, the overall process will reinitiate whenever there
ABC based MPPT to track the global MPP under different shading
is a variation in power output due to solar irradiation changes. This
patterns. The parameters used for ABC and ABC-PO algorithms are
change in irradiation is characterized by the following inequality con-
given in Table 1.
dition.

Ppv − Ppv _old 6. Modeling of PV panel under partial shading conditions

≥ ΔPpv %
Ppv _old (6)
The mostly accepted demonstration to describe the PV module is a
The GMPP search will again execute if the above criterion is satisfied. current source in parallel with a diode, with a parallel and series res-
This ensures that the ABC algorithm is always able to detect the GMPP ister as shown in Fig. 4. The PV module can be modeled with the

41
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

P&O P&O
80 40

Power=27.04W
60
Power = 74.58W Power = 36.2W 30 Power = 36.2W
Power =74.58W Power=36.2W Power =36.2W
Power (W)

40 20 PSC2
PSC3
PSC1 PSC1
20 PSC2PSC2
10

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Fig. 8. Variation of output power when irradiance changes from (a) PSC1 to PSC2 (b) PSC2 to PSC3.

support of mathematical equations shown below. These equations dis- Module reverse saturation current Irs is defined by
cusses regarding the I-V characteristics of the PV cell and module.
ISC
The basic equation which describes the PV module output current Irs =
IPV is given by (7). [exp ( q·Voc
Ns·A·k·Top ) − 1] (10)

⎡ q (VPV + IPV Rs ) ⎫ ⎤ The symbols and its values used in the equations are described in
IPV = NP ·Iph − NP ·I0 ⎢exp ⎧ − 1⎥
⎨
⎩ NS ·A·k ·Top ⎬ ⎭ (7) Table 2.
⎣ ⎦
The open circuit voltage Voc of the PV module influenced by solar
The module photo currentIph , depends on both temperature and irra- insolation and temperature change can be written as (Tey and Mekhilef,
diance as shown in Eq. (8). 2014).
G Voc = Voc − m + Vt − N ln (G ) + KV ΔT (11)
Iph = [ISC + Ki (Top − Tref )]
1000 (8)
where Vt = N = NskTN◁q is thermal voltages of a PV module at
As the cell temperature varies, the module saturation current, Io also nominal temperature TN (in Kelvin).
varies and is described by Eq. (9).
3 7. Results and discussions
Top ⎤ ⎧ q·Eg ⎛ 1 1 ⎞⎫
Io = Irs ⎡ exp ⎜ − ⎟
⎢ Tref ⎥ ⎨ k ·A ⎝ Tref Top ⎠ ⎬
⎣ ⎦ ⎩ ⎭ (9) P&O, INC, ABC, and ABC-PO algorithms were implemented in

42
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

Fig. 10. Ramp variation in irradiance profile.

Table 6
Specifications of boost converter.
Components Specification
Fig. 9. Solar energy production profile.
Input voltage (Vin) 17–33 V
Output voltage (Vo) 35 V
MATLAB version 2015. For testing the algorithm, two series connected Output current (Io) 2A
Vikram Solar ELDORA-37 modules are modeled in MATLAB. The spe- Output capacitance (C) 90 μF
Load resistance (RL) 19 Ω
cification for this PV panel under STC (Standard Test Conditions) is
Inductance (L) 530 µH
given in Table 3. Fig. 5 shows the implementation of PV array and boost
converter in MATLAB/Simulink.
The resultant I-V curves for different irradiances during PSCs are Tracking of LMPP instead of GMPP makes the algorithm fail to serve the
shown in Fig. 6 (a). This is the summation of instantaneous currents overall intention of tracking true MPP which is the main drawback of P
from each branch. &O under PSC.
For more explanation, refer Sundareswaran et al. (2015). Similarly, Similarly, for the same irradiance profile INC algorithm is executed
Fig. 6(b) shows the P-V characteristics, which is calculated with the for 2 s. The power output of 75.08 W is obtained during PSC1 in 0.02 s.
help of resultant I-V curves (Alajmi et al., 2013). Three different When PSC1 changes to PSC2 at 1 s, 41.2 W power is obtained with the
shading conditions are considered for the studies which are namely help of INC algorithm. When compared with P&O algorithm INC gives
PSC1, PSC2 and PSC3. PSC1 is uniform irradiance condition at STC better response with higher power output. But the computational
(1000 W/m2 at 25 °C) and the remaining 2 are partially shaded condi- complexity and cost is higher in INC than P&O.
tions with bypass diode. The detailed description of the effect of bypass ABC algorithm when executed during PSC1 pattern gives a GMPP of
and blocking diodes on the PV characteristics under PSC is given in 76.12 W in 0.4 s. Slowly after 1 s, insolation gradually makes a transi-
paper (Patel and Agarwal, 2008). The variations of irradiance profile on tion to PSC2 and ABC algorithm successfully tracks GMPP of 42.3 W
both the PV panels are shown in Fig. 7. The theoretical LMPP and GMPP in1.25 s.
values obtained from the P-V curves for the considered PSCs are given Even though the ABC algorithm tracks maximum power than INC
in Table 4. and P&O algorithms, the settling time is greater than the other algo-
Panel output power when there is an irradiance transition from rithms discussed. The sampling time and the processing time of opti-
PSC1 to PSC2 is discussed in Fig. 8(a). In this figure, the two PSC mization delays the settling time and produce some oscillations around
patterns named as PSC1 and PSC2 exist from 0 to 1 s and 1 to 2 s, re- GMPP. It necessitates reduction in sampling time and an alternative
spectively. During PSC1 pattern, the tracking curve of P&O shows that method to reduce oscillations with the benefit of maximum power.
power output is tracked along 74.58 W in 0.05 s. But when the transi- Hence in this article, ABC algorithm having reduced sampling time with
tion from PSC1 to PSC2 occures, it is clear that P&O algorithm is a combination of P&O algorithm in order to diminish oscillations after
tracked at LMPP at 36.2 W and thus failing to track GMPP. When keenly reaching GMPP is proposed.
observed, there is presence of steady state oscillations around MPP. To facilitate the advantages of proposed ABC-PO algorithm, it is

Table 5
Performance of the different MPPT algorithms under partial shading conditions.
Pattern MPPT Output power obtained from Obtained output Energy produced per Efficiency (%) Units/month Costs/month (Rs/ Payback time
P-V curve (W) Power (W) day (Wh) (kWh/month) Month) (Months)

PSC1 P&O 76.45 74.58 745.8 97.55 22.374 55.935 48.3

INC 75.08 750.8 98.21 22.524 56.31 47.9
ABC 76.12 761.2 99.57 22.836 57.09 47.3
ABC-PO 76.4 764 99.93 22.92 57.3 47.1

PSC2 P&O 42.88 36.2 362 84.42 10.86 27.15 99.5

INC 41.2 412 96.08 12.36 30.9 87.4
ABC 42.3 423 98.64 12.69 31.725 85.1
ABC-PO 42.78 427.8 99.77 12.834 32.085 84.2

PSC3 P&O 29.79 27.04 270.4 90.76 8.112 20.28 133.1

INC 28.56 285.6 95.87 8.568 21.42 126.1
ABC 29.12 291.2 97.75 8.736 21.84 123.6
ABC-PO 29.67 296.7 99.59 8.901 22.252 121.3

43
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

(a) (b) (c)

Fig. 11. Variation in duty cycle, current, voltage and power output of DC-DC converter for ramp variation in irradiance profile (a) P&O, (b) INC, and (c) ABC
algorithm.

44
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

Fig. 12. (a) Variation in duty cycle, current, voltage and power output of DC-DC converter for ramp variation in irradiance profile using ABC-PO algorithm (b)
Zoomed version of the same.

Table 7
Performance comparison of mppt methods based on step variations in input.
MPPT Steady state error Dynamic performance Tracking speed (S) Tracking MPP under PSC Accuracy Efficiency (%)

P&O High Poor 0.05 Depends on PV curve Less 97.55

INC Moderate Moderate 0.02 Yes Moderate 98.21
ABC Less Good 0.4 Yes Moderate 99.57
ABC-PO Less Very Good 0.08 Yes High 99.93

performed under PSC1 and PSC2 conditions. Power output of 76.4 W in algorithms during irradiation changing from the pattern PSC2 to PSC3
0.08 s and 42.78 W in 1.05 s is obtained respectively in PSC1 and PSC2. is shown in Fig. 8(b). Here also ABC-PO outperforms all other algo-
A 30% drop in maximum overshoot is also achieved in the output rithms.
power plot of PV panel when working with ABC-PO MPPT algorithm. For an analysis in the aspect of energy production, solar irradiation
On focusing at the tracking plot, occurrence of delay time, distortions of Coimbatore city in Tamil Nadu, India is considered in this paper. The
and peak overshoot is minimal. real time data is captured from the PV power plant module installed at
Similarly the output plot obtained for all the four considered PSG College of Technology, Coimbatore.

45
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

The PV power production per day is shown in Fig. 9. From Fig. 9 it is makes use of ABC algorithm to track the optimum duty cycle. As a
observed that the minimum power required to meet the converter losses result, there is no big difference in output of ABC and ABC-PO algo-
is around 10% of the maximum power which is available from 7 AM to rithms. The simulation results proved that the optimum duty cycle is
5 PM. The significant power production duration of 10 h is considered very well tracked using ABC-PO algorithm with very less oscillations
for energy calculation in this study. and faster tracking characteristics than all the other mentioned algo-
Based on this calculation, energy produced by various shading rithms.
patterns is estimated. Each pattern is analyzed with the combination of A brief comparison of performance indexes of the considered MPPT
all MPPT algorithms. techniques are given in Table 7. It is noted from Table 7, that the ABC-
For further comparison of all the considered patterns, numerical PO offers better performance compared to all other algorithms analyzed
values of performances of the MPPT algorithms are listed in Table 5. in the aspects of steady state error, dynamic performance, accuracy,
Cost estimation per month based on Tamil Nadu Electricity Board efficiency and tracking MPP under PSC.
(TNEB) domestic tariff is also presented in Table 5. A Vikram poly-
crystalline 75 W solar panel costs around Rs. 2700/-. From Table 5 it is 8. Conclusion
estimated as payback time of panel using INC is 87.4 months in the case
of PSC2, whereas, ABC-PO returns it in 84.2 months. This means that In this paper a new improved P&O MPPT method has been proposed
around 3 months before itself ABC-PO achieved its payback time when for global maximum power point tracking in PV systems under partial
compared to INC. As far as the Vikram Solar panel is concerned, the shading conditions. It reduces the convergence probability at local MPP
manufacturer guarantees a life span of 27 years. during partial shading conditions and eliminates the steady state os-
Since a low power setup in domestic tariff is considered, charge per cillations. In the proposed method the conventional P&O algorithm and
unit is less which makes the difference in payback time small. Artificial Bee Colony algorithm are integrated together to track the
Application of ABC-PO in a high power rating/industrial usage will global MPP under varying shading patterns. The simulations were
result in a significant difference in the payback time. carried out on three different shading patterns using step input profile.
From Table 5, it is clear that for all the irradiation conditions the In all conditions, it is observed that the proposed ABC-PO algorithm
ABC-PO offers high efficiency with minimum payback time than that of outperforms the existing ABC, INC and P&O algorithms. It is found that
P&O, INC and ABC algorithms. Thus ABC-PO algorithm justifies to be in ABC-PO, a 30% decrease in maximum overshoot than the existing
an efficient tracking algorithm when compared to that of previously ABC algorithm. Moreover, ramp input is also given to study the varia-
discussed algorithms. tions in output for all the MPPTs. It is found that ABC and ABC-PO
In addition to the output power response for the step change in ir- algorithms are able to track the maximum power with minimum power
radiation, the ramp response is also presented in this paper. deviations compared to P&O and INC. The simulation result clearly
Fig. 10 shows the ramp profile used for irradiance variation in the shows highly accurate tracking characteristics of the ABC-PO algorithm
current study. From 0 to 2 s, irradiation increases linearly from 600 W/ with high efficiency, fast tracking and minimum payback time.
m2 to 1000 W/m2. From 2 to 4 s, irradiance decreases from 1000 W/m2
to 850 W/m2. References
Boost converter used in the suggested work is designed to operate at
35 V. The descriptions of the DC-DC boost converter are given in Alajmi, B.N., Ahmed, K.H., Finney, S.J., Williams, B.W., 2013. A maximum power point
Table 6. tracking technique for partially shaded photovoltaic systems in microgrids. IEEE
Trans. Ind. Electron. 60, 1596–1606. https://doi.org/10.1109/TIE.2011.2168796.
In Fig. 11, the duty ratio, voltage, current and power outputs of Babar, B., Crăciunescu, A., 2014. Comparison of artificial bee colony algorithm with other
various algorithms such as P&O, INC, and ABC using boost converter algorithms used for tracking of maximum power point of photovoltaic arrays. Int.
under ramp change in irradiance is shown. Conf. Renew. Energies Power Qual. 1, 1005–1008.
Bilal, B., 2013. Implementation of artificial bee colony algorithm on maximum power
In Fig. 11(a), it is observed that with the help of boost converter point tracking for PV modules. In: 2013 8Th Int. Symp. Adv. Top. Electr. Eng. 1–4.
using P&O algorithm, even if the insolation is changed continuously, https://doi.org/10.1109/ATEE.2013.6563495.
the output voltage is maintained around 34 V but with steady state De Brito, M.A.G., Galotto, L., Sampaio, L.P., De Azevedo Melo, G., Canesin, C.A., 2013.
Evaluation of the main MPPT techniques for photovoltaic applications. IEEE Trans.
oscillations. The output power increases from 44.52 W to 74.17 W as Ind. Electron. 60, 1156–1167. https://doi.org/10.1109/TIE.2012.2198036.
the irradiance increases from 600 W/m2 to 1000 W/m2. Then power Dorofte, C., Borup, U., Blaabjerg, F., 2005. A combined two-method MPPT control scheme
decreases to 63 W with decrease in irradiance from 1000 W/m2 to for grid-connected photovoltaic systems. In: 2005 Eur. Conf. Power Electron. Appl. 9,
10 pp.-P.10. https://doi.org/10.1109/EPE.2005.219714.
850 W/m2. The deviation in the power tracked by P&O is about 2.5 W.
Esram, T., Chapman, P.L., 2007. Comparison of photovoltaic array maximum power point
The poor tracking of MPP during continuous change in irradiation is the tracking techniques. IEEE Trans. Energy Convers. 22, 439–449. https://doi.org/10.
main drawback of P&O algorithm. 1109/TEC.2006.874230.
Similarly in Fig. 11(b), using INC algorithm, the boost converter Faranda, R., Leva, S., Maugeri, V., 2008. MPPT techniques for PV systems: energetic and
cost comparison. IEEE Power Energy Soc. 2008 Gen. Meet. Convers. Deliv. Electr.
tracked an output voltage around 34 V even if the irradiance profile Energy 21st Century, PES 1–6. https://doi.org/10.1109/PES.2008.4596156.
varies. Here the oscillations are lesser than that of P&O. In this case also Femia, N., Petrone, G., Spagnuolo, G., Vitelli, M., 2005. Optimization of perturb and
power deviation is around 1.5 W, which requires an improved con- observe maximum power point tracking method. IEEE Trans. Power Electron. 20,
963–973. https://doi.org/10.1109/TPEL.2005.850975.
troller to track the maximum power. Ishaque, K., Salam, Z., 2012. A deterministic particle swarm optimization maximum
Fig. 11(c) shows the output of boost converter using ABC algorithm. power point tracker for photovoltaic system under partial shading condition. IEEE
For the same irradiance variations, voltage is maintained at 34.13 V but Trans. Ind. Electron. 60https://doi.org/10.1109/TIE.2012.2200223. 1-1.
Karaboga, D., Akay, B., 2011. A modified Artificial Bee Colony (ABC) algorithm for
with fewer distortions. It can be seen from this figure, from 0 to 2 s, constrained optimization problems. Appl. Soft Comput. J. 11, 3021–3031. https://
output power increases from 45 W to 75.18 W. doi.org/10.1016/j.asoc.2010.12.001.
When the irradiance reduces to 850 W/m2, the power by ABC al- Kumar, N., Hussain, I., Singh, B., Panigrahi, B.K., 2018. Framework of maximum power
extraction from solar PV panel using self predictive perturb and observe algorithm.
gorithm reaches around 65 W. This reveals that ABC algorithm tracks
IEEE Trans. Sustain. Energy 9, 895–903. https://doi.org/10.1109/TSTE.2017.
efficiently compared to P&O and INC by reducing the power deviation 2764266.
less than 1 W. This is the main advantage of ABC algorithm. Mei, Q., Shan, M., Liu, L., Guerrero, J.M., 2011. A novel improved variable step-size
incremental-resistance MPPT method for PV systems. IEEE Trans. Ind. Electron. 58,
Finally, in Fig. 12(a) and (b), duty ratio, current, voltage and power
2427–2434. https://doi.org/10.1109/TIE.2010.2064275.
output of proposed ABC-PO is shown. From the zoomed view in Mohanty, P., Bhuvaneswari, G., Balasubramanian, R., Dhaliwal, N.K., 2014. MATLAB
Fig. 12(b) it is clear that ABC-PO produces better output where output based modeling to study the performance of different MPPT techniques used for solar
voltage is very well tracked at 34.58 V with negligible distortions. As PV system under various operating conditions. Renew. Sustain. Energy Rev. 38,
581–593. https://doi.org/10.1016/j.rser.2014.06.001.
the irradiance is continuously varying, ABC-PO algorithm repeatedly

46
D. Pilakkat, S. Kanthalakshmi Solar Energy 178 (2019) 37–47

Noguchi, T., Togashi, S., Nakamoto, R., 2002. Short-current pulse-based maximum- energy output from a PV system under partial shaded conditions through artificial
power-point tracking method for multiple photovoltaic-and-converter module bee colony. IEEE Trans. Sustain. Energy 6, 198–209. https://doi.org/10.1109/TSTE.
system. IEEE Trans. Ind. Electron. 49, 217–223. https://doi.org/10.1109/41.982265. 2014.2363521.
Oshaba, A.S., Ali, E.S., Abd Elazim, S.M., 2017. PI controller design using ABC algorithm Sundareswaran, K., Vigneshkumar, V., Sankar, P., Simon, S.P., Srinivasa Rao Nayak, P.,
for MPPT of PV system supplying DC motor pump load. Neural Comput. Appl. 28, Palani, S., 2016. Development of an improved P&O algorithm assisted through a
353–364. https://doi.org/10.1007/s00521-015-2067-9. colony of foraging ants for MPPT in PV system. IEEE Trans. Ind. Inform. 12, 187–200.
Oshaba, A.S., Ali, E.S., Elazim, S.M.A., 2015. Artificial bee colony algorithm based https://doi.org/10.1109/TII.2015.2502428.
maximum power point tracking in photovoltaic. System 10, 123–134. Tey, K.S., Mekhilef, S., 2014. Modified incremental conductance MPPT algorithm to
Patel, H., Agarwal, V., 2008. MATLAB-based modeling to study the effects of partial mitigate inaccurate responses under fast-changing solar irradiation level. Sol. Energy
shading on PV array characteristics. IEEE Trans. Energy Convers. 23, 302–310. 101, 333–342. https://doi.org/10.1016/j.solener.2014.01.003.
https://doi.org/10.1109/TEC.2007.914308. Williams, B.W., Helal, A.A., Elsaharty, M.A., Abdelsalam, A.K., Zakzouk, N.E., 2016.
Pilakkat, D., Kanthalakshmi, S., 2018. Drift free variable step size perturb and observe Improved performance low-cost incremental conductance PV MPPT technique. IET
MPPT algorithm for photovoltaic systems under rapidly increasing insolation. Renew. Power Gener. 10, 561–574. https://doi.org/10.1049/iet-rpg.2015.0203.
Electron. J. 22, 19–26. https://doi.org/10.7251/ELS1822019P. Yetayew, T.T., Jyothsna, T.R., Kusuma, G., 2016. Evaluation of Incremental conductance
Sundareswaran, K., Peddapati, S., Palani, S., 2014. MPPT of PV systems under partial and firefly algorithm for PV MPPT application under partial shade condition. In: 2016
shaded conditions through a colony of flashing fireflies. IEEE Trans. Energy Convers. IEEE 6th Int. Conf. Power Syst. ICPS 2016 1–6. https://doi.org/10.1109/ICPES.2016.
29, 463–472. https://doi.org/10.1109/TEC.2014.2298237. 7584089.
Sundareswaran, K., Sankar, P., Nayak, P.S.R., Simon, S.P., Palani, S., 2015. Enhanced

47