Original Article

Proc IMechE Part A:

J Power and Energy
Chaotic baker map-based array 2019, Vol. 233(5) 559–575
! IMechE 2019

reconfiguration in solar photo-voltaic Article reuse guidelines:

sagepub.com/journals-permissions

systems under shading conditions DOI: 10.1177/0957650919857436

journals.sagepub.com/home/pia

Venkata Madhava Ram Tatabhatla, Anshul Agarwal and

Tirupathiraju Kanumuri

Abstract
The location of shaded and unshaded panels affect power generation of solar photo-voltaic array, which not only
causes the disturbance in characteristic curves but also achieves multiple maximum peaks. In order to overcome
these issues, panel reconfiguration is proposed based on chaotic baker map technique, which employs the concept of
image processing. In this approach, the static position of a panel is changed without altering the electrical circuit
to minimize the power loss and enhance the maximum power. The proposed method is tested with different sizes of
solar photo-voltaic array considering a wide range of shading spectrum on both software and hardware platforms.
The proposed approach augments the output power and minimizes the mismatch power loss by reducing the mismatch
in row currents with both simulations and prototype results. The results of the proposed approach are compared
with hitherto known configurations under various shading conditions, which support the effectiveness of the
proposed approach.

Keywords
Solar photo-voltaic array, power generation, partial shading, chaotic baker map, shade dispersion, solar power

Date received: 1 April 2019; accepted: 22 May 2019

nearby buildings or huge erected structures.

Introduction
Moreover, poor maintenance may also result in accre-
The continuously increasing demand for energy tion of dirt. This will also result in unsteady reception
across the world can be provided by the plentiful of irradiation. The shaded panels and unshaded
availability of solar energy. Power generated by panels have different electrical characteristics. As the
solar photo-voltaic (SPV) technology has been met shade intensity increases, the mismatch increases. This
with general admiration, appreciation and positive in turn minimizes the energy harvest. In such condi-
responses among populations; governments are offer- tions, increasing the output will be a leading issue to
ing subsidies and the reduction in price per watt due be solved as the decrement of the output parameters
to the vast improvement in technology. Besides these not only depends on shading area but also on the
gigantic technological improvements, there are some panel reconfiguration strategy (array configuration),
choke points like lesser level of efficiency in conver- shade concentration and arrangement of shaded and
sion, reliance on insolation levels and changing in unshaded panels in the SPV array.2,3 Therefore, desir-
environmental circumstances, which deteriorate the able techniques are to be developed for reducing the
performance metrics.1 One of the preponderant rea- effect of partial shading.
sons that influence the performance metrics is the loss
caused by change in characteristics of PV panels. The
frequent roots for mismatch comprises miserable sol- Department of Electrical and Electronics Engineering, National Institute
dering bonds, manufacturing defects, aged panels, of Technology Delhi, New Delhi, India
and shading. Amidst the above, the shading condi-
Corresponding author:
tions have a prejudicial effect on the entire system.
Tirupathiraju Kanumuri, Department of Electrical and Electronics
Shading is inevitable in many cases though the Engineering, National Institute of Technology Delhi, Narela, New Delhi,
panels are placed and affixed properly. This is because India.
of the space limitations and is mostly occurred due to Email: ktraju@nitdelhi.ac.in
560 Proc IMechE Part A: J Power and Energy 233(5)

The generated photo-current is decreased when an A controllable switching matrix implements an elec-
SPV array undergoes shading and the decrement is trical array re-configuration (EAR) scheme that
proportionate to its shading factor ‘‘S’’. If the allows to reconnect the modules electrically.19,20
shaded panel is in a series string, it will inflict restric- Nevertheless, the usage of this method has got its
tion on the string current. If the shaded panel is own disadvantages, i.e., high complexity and cost,
strained to conduct an additional amount of current, and is suitable for smaller systems only. To increase
reverse bias happens and thermal stress is increased the approachable power in an array under standard
across the shaded panel. Apart from this, hot-spots temperature environmental conditions (STC), an
evolve when there is a nonuniform distribution of array re-configuration algorithm is reported in
current, which can be nullified by connecting the Tabanjat et al.21 and Parlak.22 Yet, in order to carry
bypass diode affixed anti-parallel to the panel.4 on the algorithm, the equipment of switches and sen-
Comprehension of bypass diodes also encloses vari- sors are mandatory. To reduce the mismatch losses, the
ous elevations in PV characteristic curves and various interconnections of PV modules are reformed using an
steps in IV characteristic curves. The global maximum auto-adjustable reconfiguration process established on
power point tracking (GMPPT) is demanded by the fuzzy logic estimator was protruded,23,24 each re-con-
creation of various elevations in characteristic curves figuration technique involves sensors and switches to
because the established tracking algorithms may not reconfigure the array and to nullify the losses, which
distinguish between local peaks (LP) to global peaks incurs more cost and complexity to the system.
(GP). Several multi-stage evolutionary algorithms are
presented to track GP.5–8 Furthermore, the technique
Motivation and contribution
in which panels are connected has an important effect
on output under partial shading scenarios.9 Several In order to achieve the maximum energy harvest
re-configuration techniques were proposed to con- under partial shading conditions, array reconfigur-
dense the shading effect and can be classified as ation is necessary. The static configurations are
static and dynamic techniques.10 futile because of zero shade dispersion. So, to tackle
To increase the output power, a pre-set inter- the panels under partial shading conditions, a rando-
connection estimate is employed by static re- mized reconfiguration is needed. Therefore, this work
configurations and it can be named as one-time proposes a novel randomized configuration inspired
arrangement. As the panel location is fixed for every by the ‘‘chaotic baker map’’ (CBM). The CBM is the
shading condition, these do not modify the array most used technique for image encryption, which
dynamically. The benchmark configuration, total moves the position of pixels within the image.
cross tied (TCT), is utilized to lessen the mismatch Motivated by this, the CBM is applied for the recon-
loss with various calculations depending upon the figuration of the SPV array by considering the panel
knowledge of isolation levels and shading scen- in SPV array as pixels of the image.
arios.11–13 Nevertheless, the advancement of this The CBM configuration involves shade dispersion
scheme is difficult, particularly for asymmetric shad- by relocating the shaded and unshaded panels in the
ing circumstances as the calculations will be carried SPV array by spreading the dense shade over the
based upon the shading area.14 A one-time arrange- entire SPV array without altering its electrical cir-
ment in physical relocation of panels attached in the cuitry. To demonstrate the proposed work, CBM
TCT configuration has been displayed in Rani et al.15 reconfiguration technique is tested on 8  8 SPV
The modules are physically relocated on the Sudoku array for different types of shading conditions. The
puzzle pattern keeping the electrical circuit same for a proposed configuration yields maximum output
diversity of shades. This configuration spreads the power by reducing the mismatch power loss, thereby
shade throughout the array leading to increase in improving the output power. In order to generalize
the PV power. The consequence of the partial shading the proposed approach, this configuration also
on the SPV array was reported in detail in Satpathy tested on 4  4 and 6  6 SPV array.
and Sharma.16 Furthermore, the mismatch and line Further, the organization of the paper is as follows:
losses are reduced by re-numbering the panel organ- in the subsequent section, fundamental system param-
ization with mathematical formulation under various eters used for this study are modeled. The proposed
shading circumstances in Potnuru et al.17 In order to shade dispersion reconfiguration with CBM configur-
lessen the computational time and improving power, a ation is introduced followed by theoretical method-
novel method is presented in Mahmoud and El- ology through a case study. The qualitative and
Saadany,18 which relocates some panels keeping the quantitative analysis of the proposed approach for
remaining panels fixed to maintain optimality. wide-range of shading spectrum is explained in test
To improve the output power, dynamic re-config- and validation section followed by the generalization
uration strategies include electrical components like of the proposed approach. The experimental valid-
switches and electronic equipment such as controllers ation of the proposed approach is explained in the
and sensors to reconfigure the array electrically by next section and finally, the paper is concluded with
altering the electrical circuit of an array. a conclusion remark.
Tatabhatla et al. 561

Modeling of fundamental parameters The photo-electric current is in terms of short cir-

cuit current, which is given by
Modeling of the solar cell
 
Modeling of PV module is an important phase in G ðRs þ Rsh Þ
Iph ¼ Isc ð1 þ ðT  To ÞÞ ð2Þ
calculating the efficiency of the array. Several tech- Go Rsh
niques in modeling of solar cell are reported: such
as, single-diode, two-diode and three-diode. The where Isc represents the short circuit current, G and Go
single-diode model is used in this work because of are the incident irradiation and standard irradiation
its simplicity. An equivalent circuit comprising respectively, To and T are standard temperature and
of ideal, practical and simplified models are shown temperature of the cell at incident irradiation G, and 
in Figure 1. The following equation gives the relation is the temperature coefficient.
between voltage and current of the SPV cell
      Array interconnection configurations
V þ IRse V þ IRse
I ¼ Iph  Io exp 1 
A Rsh The array interconnection configurations tested and
ð1Þ compared with the proposed work is shown in
Figure 2 and are explained clearly below:
where Iph represents the photo-electric current, Io the
reverse saturation current, Rsh the shunt resistance . The series connected panels are nonresilient to
and Rse the series resistance, V the voltage across shade whereas the parallel connected panels are
solar cell, I the current through solar cell, the diode resilient. So, the number of panels are intercon-
ideality factor is denoted by A and expressed math- nected in a matrix arrangement with rows in
ematically as A ¼ nkT=q where, n is the number of series and columns in parallel known as a series–
cells connected in series, q the elementary charge, k parallel (SP) configuration with zero redundancy as
the Boltzmann constant, T the temperature of the shown in Figure 2(a).
solar cell at STC. . Bridge-link (BL) configuration is adapted from
Wheatstone bridge along with bridge rectifier. It
has minimum redundancy as shown in Figure 2(b).
. In honey comb (HC) configuration, the connection
of panels are motivated by the honeycomb archi-
tecture (hexagon shape) as shown in Figure 2(c).
. The total cross tied (TCT) is formed by establishing
the interconnections among the rows and columns
of SP configuration as shown in Figure 2(d).

Disadvantages of conventional configurations under

partial shading conditions
Despite the fact that the TCT interconnection possesses
greater potential to obtain the maximum power, it
Figure 1. Equivalent circuit for single-diode model of the comes up against few hindrances like nil dispersion of
SPV cell. shade and advent of several peaks. As the electrical

(a) (b) (c) (d)

Figure 2. Different interconnection configurations for an 8  8 SPV array: (a) series–parallel (SP); (b) bridge-link (BL); (c) honeycomb
(HC); (d) total cross tied (TCT).
562 Proc IMechE Part A: J Power and Energy 233(5)

connections are rigid, the row currents of the panels

under shading conditions fluctuate over a wide range.
Hence, the panels possessing a lower level of row cur-
rent are violated to insulate them so that they are not
damaged. Such violated row currents generate disturb-
ances in P-V and I-V characteristic curves that in turn
causes the maximization of mismatch losses in the
system. Therefore, shade dispersion by rearrangement
is required to decrease the panel violations thereby
increasing the energy yield of SPV array. A pioneering
image processing-based reconfiguration, CBM, is pro-
posed in this work, which scatters the shade on the
complete array thus raising the output power and mini-
mizing the power losses.
Figure 3. Chaotic baker map: (a) two-dimensional baker
map; (b) generalized baker map.
Methodology of the proposed approach
In the image processing, an image is taken as a matrix The generalized baker map extends every rectangle
consisting of pixels as its elements. To improve security, horizontally with the factor 1=vi . Simultaneously, the
image encryption is performed by changing the pixel rectangle is shrunk vertically by the factor of vi .
positions of an image.25 These pixel positions are chan- Ultimately, each one of the rectangles is piled upon
ged on the basis of a secret key so that only end users one another as in Figure 3(b) by Bernoulli’s shifts.28
can decode it.26 The CBM is the most commonly used The baker map randomization is performed
tool in the image processing to encrypt an image. The through the following steps and the generalized
main advantage of using CBM among other chaotic baker map is shown in Figure 3(b):
maps is that these baker maps will naturally create opti-
mized secret key among the large set of possible keys. . The matrix is divided into m rectangles and each
rectangle having N elements with width vi where,
N ¼ v1 þ v2 þ    þ vi
Chaotic baker map
. In each selected rectangle, all the elements are
In this method, the map performs an action on the arranged in a row-wise manner starting from
unit square (N  N) and the mathematical representa- bottom left corner followed by upward elements.
tion of baker map ‘‘B’’ is delineated by the formula . The selection of the rectangles is from right to
underneath27 left and the upper rectangles occupy rows ahead
than the lower rectangles in the modified

ða=2, 2bÞ 04b 5 1=2 configuration.
Bða, bÞ ¼ ð3Þ
ða=2 þ 1=2, 2bÞ 1=24b 5 1
This work employs a representative array of size
The left vertical column ½0, 1=2Þ  ½0, 1Þ is extended 8  8, with secret key (2, 4, 2). Initially, in the first
horizontally and shrinked vertically to form a rect- stage, the 8  8 array is divided into three major rect-
angle ½0, 1Þ  ½0, 1=2Þ. The similar mapping is done angles each with two, four and two columns respect-
on the right vertical column from ½1=2, 1Þ  ½0, 1Þ to ively as per secret key. Hence, primary rectangles are
½0, 1Þ  ½1=2, 1Þ as shown in Figure 3(a). formed with array sizes, 8  2, 8  4, 8  2, respect-
As an alternative of division into two identical rect- ively. Subsequently, at the second stage, these three
angles, it is divided on the basis of the CBM. The baker rectangles would be further partitioned into various
map is represented by Bðv1 , v2 , . . . , vm Þ, where sub-blocks containing ‘‘eight’’ elements, as the
the vector ½v1 , v2 , . . . , vm  denotes the secret key. The number of rows and columns considered for this
secret key contains m elements, which divide the work are ‘‘eight’’. Following this, the first rectangle
entire pixels in an image into m rectangles. N denotes is divided into number of sub-blocks having eight
the pixel in a particular row. The selection of secret key elements each. The first rectangle can be divided
is done in such a way that each integer vi divides N and into ‘‘8  1’’, or ‘‘4  2’’ sub-block but, ‘‘8  1’’ does
Vi ¼ v1 þ v2 þ    þ vi with Vi 4l 5 Vi þ vi where the not satisfy the primary condition of secret key
new location of the pixel (l, s) is given by (dividing the columns as per secret key). So, ‘‘4  2’’
  1 is the only sub-block, which is satisfying all the con-
0
N
 Vi Þ þ s mod N , ditions. So, this primary rectangle of ‘‘8  2’’ is
vi ðl
Bðv1 , v2 , . . . , vm Þðl, sÞ ¼ @    vi A divided into two sub-blocks containing two columns
vi N
N s  s mod vi þ Vi and four rows (of array size ‘‘4  2’’) each which are
ð4Þ indicated in violet and red colors respectively.
Similarly, the second rectangle is divided into four
Tatabhatla et al. 563

(a) (b) (c)

Figure 4. Randomization through chaotic baker map of array size 8  8: (a) dividing the matrix into rectangles using secret key
(2, 4, 2); (b) sample unwrapping of elements in sub-blocks to rows; (c) final randomized array.

(a) (b) (c)

Figure 5. Evolution of chaotic baker map reconfiguration: (a) unconnected panels in TCT configuration; (b) configured CBM without
connections; (c) structural arrangement of CBM reconfigured panels.

sub-blocks containing four columns and two rows

Chaotic baker map configuration of SPV array
each (each sub block of size ‘‘2  4’’), which are indi- The proposed work is based on the analogy between
cated in green-, blue-, orange-, and pink-colored sub- solar panel in the SPV array and picture element
blocks. It is clearly witnessed that the second rectangle (pixel) in a digital image. A popular image encryption
is divided in such a way that the secret key condition technique, called chaotic baker map is employed to
is satisfied as well as each sub-block contains eight achieve efficient reconfiguration of panels without
elements. Similarly, the third rectangle is divided affecting their electrical connections. The technique
into two sub-blocks containing two columns and is used here to enhance the output power of SPV
four rows each (of array size ‘‘4  2’’), which are indi- array under partial shading conditions. This decreases
cated in grey and yellow colors, respectively. the mismatch of row currents, reduces the power loss
Figure 4(a) clearly depicts the division of array into and hence improves the performance parameters. In
rectangles and with secret key. the proposed reconfiguration, the panels are con-
Later on, all the second stage sub-blocks are nected based on the CBM method with the electrical
unwrapped to form row matrices satisfying the circuit being the same. This will not cause any change
below conditions successively. The sub-blocks are in its electrical properties and hence reduces the inten-
scanned from top to bottom starting from right to sity of shading by dispersing the shadow over the SPV
left. Now, each block is converted into a row matrix array. The evolution for the proposed CBM reconfig-
by scanning bottom to top and left to right as shown uration technique is shown in Figure 5. Figure 5(a)
in Figure 4(b). The same procedure is followed for depicts unconnected TCT panels, which are physically
remaining sub-blocks (yellow, green, blue, orange, relocated using the proposed approach as depicted in
pink, violet, red sub-blocks, respectively) and the Figure 5(b). The electrical connections of the pro-
resulted row matrices are stacked together as row 2 posed configuration are same as that of TCT only,
to 8 to form the randomized array of size 8  8 as whereas the electrical connections are not altered as
shown in Figure 4(c). displayed in Figure 5(c). As a result, the proposed
564 Proc IMechE Part A: J Power and Energy 233(5)

image processing based panel reconfiguration first four panels in row-1 are shaded while the remain-
approach leads in maximum energy harvest. ing remain unshaded, which is clearly depicted in
Figure 6(a). The first row current is given by
Theoretical methodology of the
X
8
proposed approach (a case study) IR1 ¼ S1y Im ¼ S11 Im þ S12 Im þ S13 Im þ S14 Im
y¼1
In order to evaluate the process of shade dispersion
through array reconfiguration in the CBM method, a þ S15 Im þ S16 Im þ S17 Im þ S18 Im ð8Þ
shading of 6  4 (24 panels) case is considered on an
8  8 SPV array. The shading for TCT and shade dis- for, S11 ¼ G11 =Go ¼ S; . . . ; S14 ¼ G14 =Go ¼ S ð 5 1Þ
persion through CBM is depicted in Figure 6.
The current generated by a single panel at any IR1 ¼ 4SIm þ 4Im ¼ ð4S þ 4ÞIm ð9Þ
irradiation G is given as I ¼ SIm , S ¼ G=Go , where
Im represents the current at standard irradiation of Row-1 to Row-6 have the same irradiation level
G0 ¼ 1000 W=m2 and S is the shading factor. The
voltage of the SPV array ðVa Þ is given as IR2 ¼ IR3 ¼ IR4 ¼ IR5 ¼ IR6 ¼ 4SIm þ 4Im
ð10Þ
¼ ð4S þ 4ÞIm
X
a
Va ¼ Vxy ð5Þ Row-7 and Row-8 receive the same uniform irradi-
x¼1
ation level
where Vxy refers to the voltage of the panels at yth row.
The voltage of the 8  8 SPV array by neglecting voltage IR7 ¼ IR8 ¼ 8Im ð11Þ
drop across the diodes is Va ¼ 8Vm . With the Kirchoff’s
current law, the current across each node is given by As the cross-ties of the series–parallel configuration
are tied to form the TCT configuration, due to the series
X
b
Ia ¼ Ixy  Iðxþ1Þ y ¼ 0, for x ¼ 1, . . . , ðx  1Þ connection of eight panels, the current is limited to
y¼1 ð4S þ 4ÞIm as shown in Figure 6(a). Therefore, the
ð6Þ power generated through TCT configuration is given as

The power across the SPV array without effect of Pa,TCT ¼ Va Ia ¼ ðð4S þ 4ÞIm Þð8Vm Þ
shading and under STC conditions for an 8  8 SPV ð12Þ
¼ ð32ðS þ 1ÞVm Im Þ
array is given as

Pa ¼ Va Ia ¼ ð8Vm Þð8Im Þ ¼ 64Vm Im ð7Þ

Generation of power in proposed CBM
configuration
Generation of power in TCT configuration
The four panels in each of six rows (6  4 shading
The power generated by the TCT configured SPV conditions) shaded in the TCT configuration is dis-
array with 6  4 shading conditions in which the persed in all the eight rows of three panels (8  3) in

(a) (b)

Figure 6. Shading of 8  8 SPV array in (a) TCT configuration and (b) proposed CBM configuration.
Tatabhatla et al. 565

the SPV array by using the CBM reconfiguration

technique as shown in Figure 6(b). This technique dis-
Shading spectrum
tributes dense shade on to the complete SPV array, The proposed method is validated under various
thereby enhancing the power and reducing the mis- shading patterns that vary in size and shape for
match of row currents. The row current for the first array size 8  8. These cases are the most occurring
row is given by shading patterns. The shadings considered are
namely, diagonal shading, L-shaped shading, corner
IR1 ¼ 3SIm þ 5Im ¼ ð3S þ 5ÞIm ð13Þ shading, center shading, arbitrary shading, and
Z-shaped shading, which are represented in Figure 7
From Figure 6(b), it is clearly observed that three for 8  8 SPV array, respectively. These types of shad-
shaded panels and five unshaded panels are equally ing conditions affect the SPV array enormously when
distributed in all the rows. The current generated is the panels are shaded and operated under STC. The
given as unshaded panels receive an irradiation level of
900 W=m2 , while the shading panels receives
IR2 ¼ IR3 ¼ IR4 ¼ IR5 ¼ IR6 ¼ IR7 ¼ IR8 300 W=m2 . The SPV array under normal conditions
ð14Þ (at 900 W=m2 ) generates a maximum power of
¼ ð3S þ 5ÞIm
6292 W.
The current produced in the CBM configuration is
limited to ð3S þ 5ÞIm , which is relatively higher than
the TCT configuration because of shade dispersion.
Comparative analysis
Therefore, the power generated through CBM config- The results for the different shading conditions are
uration is given as quantitatively analyzed between SP, BL, TCT, HC,
and proposed CBM configurations, and are presented
Pa,CBM ¼ Va Ia ¼ ðð3S þ 5ÞIm Þð8Vm Þ as follows.
ð15Þ
¼ ð24S þ 40ÞVm Im
Diagonal shading. In this shading condition, 10 panels
receive different irradiation levels as shown in
Figure 7(a). The maximum power for the SP that
Mismatch power loss
can be generated is 4350 W at 180.62 V and 24.08 A,
The power generated by the shaded panels will always which is improved by BL with 4465 W at 205.97 V and
be lesser than the power generated by unshaded 21.68 A. The power generated by HC is 56 W more
panels due to the current limitation of the series con- than the BL configuration and is obtained to be
nected panels. The mismatch power loss ðPML Þ for the 4521 W at 207.26 V and 21.81 A. TCT being the best
TCT configuration is given by among all these four configurations possess the output
of 4712 W at 211.58 V and 22.27 A.
PML,TCT ¼ Pa  Pa,TCT ¼ ð64Vm Im Þ  ð32ðS þ 1ÞVm Im Þ Due to the distribution of shade on the entire
ð16Þ array, the global peak for CBM is achieved at
5428 W at 201.78 V and 26.90 A, which is 700 W
PML,TCT ¼ 32ð1  SÞVm Im ð17Þ more than the TCT configuration. From the charac-
teristic curves, it is observed that the maximum power
The mismatch power loss ðPML Þ for the CBM con- occurs in the CBP arrangement of the SPV array,
figuration is given by which is shown in Figure 8(a) and (b). The results
for generated power and power loss for this type of
PML,CBM ¼ Pa  Pa,CBM ¼ ð64Vm Im Þ  ðð24S þ 40ÞVm Im Þ shading for different configurations are tabulated in
ð18Þ Table 1.

PML,CBM ¼ 24ð1  SÞVm Im ð19Þ L-shaped shading. In L-shaped shading conditions, all
the panels are shaded in seventh and eighth rows com-
It is observed that PML,CBM 5 PML,TCT from equa- pletely and in every other row, two panels are shaded
tions (17) and (19). Therefore, by the CBM configured thus forming L-shaped pattern as displayed in
SPV array the power can be enhanced and the mis- Figure 7(b). The PV curve has more disturbances as
match losses can be minimized irrespective of the the array is subjected to shading. The global peak of
incident irradiation. the SP, BL, TCT, and HC are approximately the same
and is achieved at 132.77 V, 26.55 A and a global max-
imum power of 3525 W, 3499 W, 3523 W, and 3520 W,
Test and validation
respectively. The P–V and I–V characteristic curves
The conventional and proposed configurations are for this type of shading condition are shown in
tested for a spectrum of shading under MATLAB/ Figure 9(a) and (b), respectively. In the proposed con-
Simulink environment. figuration, as the shade is dispersed throughout the
566 Proc IMechE Part A: J Power and Energy 233(5)

(a) (b) (c)

(d) (e) (f)

Figure 7. Shading spectrum for 8  8 SPV array: (a) diagonal shading; (b) L-shaped shading; (c) corner shading; (d) center shading;
(e) arbitrary shading; (f) Z-shaped shading.

(a) (b)

Figure 8. (a) P–V characteristic and (b) I–V characteristic curves for different configurations during diagonal shading conditions for
an 8  8 SPV array.

array, i.e. the entire array, the shade is diluted and the The results for these shading are tabulated in
power generation is enhanced. The global peak for the Table 2. The output power is maximum (4266 W)
proposed CBM configuration is achieved at 195.94 V for the CBM configuration with minimum power
and 21.77 A, with a power of 4266 W. The power pro- loss of 32.19%, when compared with the other SP,
duced is 17%, greater than the benchmark TCT BL, TCT, and HC configurations. The results reveal
configuration. that the proposed configuration is outperforming the
Tatabhatla et al. 567

existing configurations in terms of high energy yield CBM, the power loss is reduced by 10% than the
and reduced power loss with L-shaped shading. conventional configurations and a loss of 22.2% is
obtained.
Corner shading. In this shading condition, shading is
occurred at one corner of the SPV array. The Center shading. With the number of shaded panels
bottom left 4  4 panels receive different irradiation being the same in both corner shading and center
levels as shown in Figure 7(c). While in that of the shading, there is no disturbance in the P–V and I–V
L-shaped shading, the maximum power between SP, characteristic curves for center shading conditions as
TCT, BL, and HC configurations are not having a in that of corner shading conditions as shown in
large variation. However, in the proposed configur-
ation, a power enhancement of 400 W is observed
when the TCT configured array is changed to the
CBM configured array. The power enhancement is Table 2. Generated output power (W) and power loss for
different configurations under L-shaped shading conditions.
achieved because of the shade dispersion. The global
peak for CBM is achieved at 195.94 V and 21.77 A, Topology Power (W) Power Loss (W) Power loss (%)
with a power of 4895 W. The P–V and I–V character-
istic curves are shown in Figure 10(a) and (b). It is SP 3525 2767 43.97
observed that the maximum power is achieved for the BL 3499 2793 44.38
CBM arrangement of the SPV array. TCT 3523 2769 44.00
Table 3 gives the summary of the output power and HC 3520 2772 44.05
power loss for different configurations. The power Proposed 4266 2026 32.19
loss in SP, BL, TCT, and HC configurations are
SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC:
varied in the range of 29–32%. Nevertheless, in honeycomb.

Table 1. Generated output power (W) and power loss for Table 3. Generated output power (W) and power loss for
different configurations under diagonal shading conditions. different configurations under corner shading conditions.

Topology Power (W) Power Loss (W) Power loss (%) Topology Power (W) Power Loss (W) Power loss (%)

SP 4350 1942 30.86 SP 4239 2053 32.62

BL 4465 1580 25.11 BL 4323 1969 31.29
TCT 4712 1827 29.03 TCT 4406 1886 29.97
HC 4521 1771 28.14 HC 4322 1970 31.30
Proposed 5428 864 13.73 Proposed 4895 1397 22.20
SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC: SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC:
honeycomb. honeycomb.

(a) (b)

Figure 9. (a) P–V characteristic and (b) I–V characteristic curves for different configurations during L-shaped shading conditions for
an 8  8 SPV array.
568 Proc IMechE Part A: J Power and Energy 233(5)

(a) (b)

Figure 10. (a) P–V characteristic and (b) I–V characteristic curves for different configurations during corner shading conditions for
an 8  8 SPV array.

(a) (b)

Figure 11. (a) P–V characteristic and (b) I–V characteristic curves for different configurations during center shading conditions for
an 8  8 SPV array.

Figure 11(a) and (b). This proves that not only the power losses for center type of shading with different
shaded panels but also the location of shaded and configurations are listed in Table 4.
unshaded panels have some adverse effects on the It is revealed from Table 4 that the power loss of
output of the SPV array. CBM is comparatively lower than other configur-
SP configuration has the least generated power, ations. The output power generated by CBM
which is 4239 W at 195.34 V and 21.70 A. The (5188 W) is high with the minimum percentage of
series–parallel configuration is improved by BL with power loss (17.54%) when compared with the SP
4351 W at 197.90 V and 21.98 A. The power generated (32.62%), BL (30.84%), TCT (29.97%), and HC
by HC is approximately near to the configuration and (30.64%).
is obtained to be 4364 W at 199.14 V and 22.12 A.
TCT being the benchmark configuration is the best Arbitrary shading. In this shading condition, maximum
among all these four configurations that possess panels as much as 24 are shaded in the arbitrary
output of 4406 W at 198.18 V and 22.02 A. Due to manner as shown in Figure 7(e). For the SP config-
the distribution of shade on the entire array, the global ured array, the 2858 W of power is generated at
peak for CBM is achieved at 5188 W at 197.25 V and 136.31 V and 20.97 A. For the same voltage and cur-
26.30 A, which is 782 W more than the TCT configur- rent, BL and HC configured array also produced
ation. From the characteristic curves, it is observed 2978 W and 3034 W, respectively. However, in the
that the maximum power occurs in the CBM arrange- TCT configuration, due to the rigidness of the con-
ment of the SPV array, which are shown in Figure nections, the power is improved exponentially and
11(a) and (b). The results for generated power and 4336 W of power is achieved at 197.56 V and
Tatabhatla et al. 569

Table 4. Generated output power (W) and power loss for Table 5. Generated output power (W) and power loss for
different configurations under center shading conditions. different configurations under arbitrary shading conditions.

Topology Power (W) Power Loss (W) Power loss (%) Topology Power (W) Power Loss (W) Power loss (%)

SP 4239 2053 32.62 SP 2858 3434 54.57

BL 4351 1941 30.84 BL 2978 3314 52.67
TCT 4406 1886 29.97 TCT 4336 1956 31.08
HC 4364 1928 30.64 HC 3034 3258 51.78
Proposed 5188 1104 17.54 Proposed 4648 1644 26.12
SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC: SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC:
honeycomb. honeycomb.

(a) (b)

Figure 12. (a) P–V characteristic and (b) I–V characteristic curves for different configurations during arbitrary shading conditions for
an 8  8 SPV array.

(a) (b)

Figure 13. (a) P–V characteristic and (b) I–V characteristic curves for different configurations during Z-shaped shading conditions
for an 8  8 SPV array.

21.95 A. CBM being the proposed method, distributes tabulated in Table 5. SP, BL, and HC connected
the shade on to the entire array thereby enhancing the SPV array has the highest power loss of 54.57%,
output power to 4648 W at a nominal voltage of 52.67%, and 51.78% respectively, which is reduced
192.83 V and nominal current of 24.10 A. The charac- to 31.08% in the TCT arrangement. The proposed
teristic curves depicted in Figure 12 prove that the CBM arrangement outperforms all the conventional
proposed approach is the best possible way to methods by delivering more output power thereby
obtain the maximum power. reducing the power loss to 26.12%. Hence, this pro-
The results for generated power and power loss for posed configuration is more effective under these
this type of shading for different configurations are shading conditions.
570 Proc IMechE Part A: J Power and Energy 233(5)

Table 6. Generated output power (W) and power loss for Z-shaped shading. In this shading condition, all the
different configurations under Z-shaped shading conditions. panels in top and bottom rows are shaded along
with anti-diagonal panels forming a ‘‘Z’’, named as
Topology Power (W) Power Loss (W) Power loss (%)
Z-shaped shading as shown in Figure 7(f). The max-
SP 3234 3058 48.60 imum power generated by the SP configuration is
BL 3853 2439 38.76 3234 W. The power generated by the BL and HC con-
TCT 3862 2430 38.62 figurations is the same because of its limited intercon-
HC 3853 4521 38.76 nections with 3853 W, which is improved by 620 W.
Proposed 4727 1565 24.87 TCT being the best among all conventional configur-
ations delivers the output of 3862 W.
SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC: Due to the distribution of shade on the entire
honeycomb.
array, the global peak for CBM is achieved at

Figure 14. Comparison of output power for a 8  8 SPV array during various shading conditions.

Figure 15. Comparison of output power for a 4  4 SPV array during various shading conditions.
Tatabhatla et al. 571

4727 W, which is 1493 W more than the SP configur-

ation, which is 874 W more than BL and HC config-
Power loss for 4  4 SPV array
urations and which is 865 W more than the TCT The output power generated for different configur-
configuration. From the characteristic curves, it is ations during shading conditions such as, diagonal
observed that the maximum power occurs in CBM, shading, L-shaped shading, corner shading, center
which is clearly depicted in Figure 13(a) and (b). The shading, arbitrary shading, and Z-shaped shading is
results for generated power and power loss under this presented in Figure 15. It is found that the proposed
shade condition for different configurations are tabu- CBM configuration delivers the maximum output
lated in Table 6. The summary of the output power power for all the shading conditions, when compared
(W) with all the shading conditions for an 8  8 SPV to SP, BL, TCT, and HC configurations. The power
array is represented pictorially in Figure 14. loss for the CBM is minimum for all the shading con-
ditions considered as shown in Table 7. The CBM
Generalized realization of the proposed configuration involves shade dispersion technique
that distributes the effect of concentrated shade
approach
throughout the array, so that the power loss is mini-
In order to generalize the proposed configuration, the mized to increase the energy harvest under partial
SPV array is tested for different array sizes, such as shading conditions.
4  4 and 6  6. The shading conditions were the same
as given in section-V, where unshaded panels are con-
sidered at 900 W/m2 and shaded are considered at
Power loss for 6  6 SPV array
300 W/m2 and the discussions are as follows. The same shading spectrum as shown in Figure 7 is
considered to analyze the power loss of the 6  6 SPV
array. The output power achieved for SP, BL, TCT,
HC, and CBM configurations is presented in
Figure 16.
Table 7. Power loss (W) by 4  4 SPV array for different The proposed configuration exhibits that the max-
configurations under various shading conditions.
imum output power is the maximum for the proposed
Shading SP BL TCT HC CBM configuration when compared to all the hitherto
known configurations. The power loss for all the con-
Diagonal 513 472 472 513 276
figurations during various shading conditions are
L-Shaped 859 867 867 880 767 reported in Table 8. It can be witnessed from
Corner 513 472 486 505 276 Table 8 that the power loss is minimized in the
Center 513 472 486 505 276 CBM configuration under various shading conditions
Arbitrary 1022 756 980 979 547 considered, which in turn enhances the maximum
SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC: honey-
power. This is achieved because of the dilution of
comb; CBM: chaotic baker map. the concentrated shade.

Figure 16. Comparison of output power for a 6  6 SPV array during various shading conditions.
572 Proc IMechE Part A: J Power and Energy 233(5)

Experimental validation of the proposed irradiation is measured by using the solar power
meter. Sun, being the source of the prototype, pro-
approach duces constant irradiance at a particular instant of
In order to validate the efficacy of the proposed time. The shading environment for irradiance vari-
approach, a lab prototype in field conditions is experi- ation is created by using opaque blocks to stop the
mented for different configurations under a wide incident light. Different opaque, transparent, and
range of shading spectrum as shown in Figure 17. translucent blocks of different colors are used for dif-
The experiments are carried for 6  6 (36 panels) ferent level of irradiations. For different shading scen-
and 4  4 (16 panels) SPV arrays using the solar arios, the obtained results are recorded and analyzed.
panels of equal rating (1.5 W at STC), a rheostat The maximum power obtained for 4  4 and 6  6
(500!, 5A), digital multimeter, and wires for connec- SPV arrays, at STC, are obtained to be 24 W and
tions. The detailed specifications of the panel used for 54 W, respectively.
this work is tabulated in Table 9. The level of The experimental output power generated for SP,
BL, TCT, HC, and CBM configurations for a 6  6
array is shown pictorially in Figure 18. The output

Table 8. Power loss (W) by 6  6 SPV array for different

configurations under various shading conditions.
Table 9. Specifications of a panel used for practical prototype.
Shading SP BL TCT HC CBM
Parameters Specifications
Diagonal 1146 926 1013 1007 551
L-Shaped 1989 1976 1982 1984 1231 Panel power (P) 1.5 W
Corner 1563 1527 1541 1556 1163 Open circuit voltage (Voc) 7.15 V
Center 760 593 635 593 337 Short circuit current (Isc) 0.27 A
Arbitrary 1379 1056 1352 1343 826 Voltage at MPP 6V
SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC: honey-
Current at MPP 0.25 A
comb; CBM: chaotic baker map. MPP: maximum power point.

Figure 17. Field prototype for various configurations under different shading scenarios to validate the proposed approach.
Tatabhatla et al. 573

Figure 18. Experimental output power for a 6  6 SPV array with different configurations during various shading conditions.

Figure 19. Experimental output power for a 4  4 SPV array with different configurations during various shading conditions.

power generated by the CBM reconfiguration method shading conditions, and 41.48% in Z-shaped shading
yields the best output under all shading conditions.
conditions. The mismatch power loss for 6  6 array The output power produced by 4  4 SPV array for
is tabulated in Table 10. It is witnessed from the various configurations under different shading scen-
table that, under field test shading conditions, arios are plotted in Figure 19. The proposed CBM
the proposed method yields better results than the method generates the maximum power with minimal
conventional configurations. The mismatch power mismatch power loss under a wide range of shading
loss is reduced to 16.36% in diagonal shading spectrum. The mismatch power loss (W and %) for
conditions, 36% in L-shaped shading conditions, 4  4 array is tabulated in Table 11. It is observed
33.81% in center shading conditions, 10.24% in from the table that under diagonal, L-shaped,
corner shading conditions, 24.43% in arbitrary corner, center, arbitrary, and Z-shaped shading
574 Proc IMechE Part A: J Power and Energy 233(5)

Table 10. Experimental mismatch power loss (W and %) by 6  6 SPV array for different configurations under various shading
conditions.

SP BL TCT HC CBM

Shading W % W % W % W % W %

Diagonal 17.65 32.69 15.25 28.24 13.96 25.85 15.2 28.15 8.94 16.56
L-Shaped 31.91 59.09 32.63 60.43 32.62 60.41 32.63 60.43 19.44 36
Corner 24.22 44.85 23.84 44.15 23.58 43.67 24.09 44.61 18.26 33.81
Center 11.9 22.04 9.73 18.02 8.99 16.65 10.39 19.24 5.53 10.24
Arbitrary 21.92 40.59 21.76 40.3 16.3 30.19 21.57 39.94 13.19 24.43
Z-Shaped 34.9 64.63 29.34 54.33 29.34 54.33 32.99 61.09 22.4 41.48
SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC: honeycomb; CBM: chaotic baker map.

Table 11. Experimental mismatch power loss (W and %) by 4  4 SPV array for different configurations under various shading
conditions.

SP BL TCT HC CBM
Shading
W % W % W % W % W %

Diagonal 11.17 46.54 10.84 45.17 10.59 44.13 10.41 43.38 7.98 33.25
L-Shaped 13.57 56.54 14.13 58.88 14.11 58.79 13.98 58.25 11.86 49.42
Corner 7.93 33.04 7.2 30 7.45 31.04 7.76 32.33 4.46 18.58
Center 7.92 33.01 7.2 30 7.2 30 7.93 33.04 4.46 18.58
Arbitrary 15.7 65.42 11.61 48.38 15.22 63.42 15.23 63.46 8.74 36.42
Z-Shaped 15.99 66.63 15.71 65.46 15.81 65.88 16.03 66.79 11.86 49.42
SP: series–parallel; BL: bridge-link; TCT: total cross tied; HC: honeycomb; CBM: chaotic baker map.

conditions, the power loss is Reduced to 7.98 W, curtailed to 15.56%, 34.78%, 32.86%, 9.52%,
11.86 W, 4.46 W, 4.46 W, 8.74 W, 11.86 W, 23.33%, and 41.28%, respectively. The proposed
respectively. method is tested under field conditions and the results
prove the efficacy of the proposed system. The advan-
tage of the proposed configuration is that it is free
Conclusion
from different types of electrical or electronic equip-
In this work, an image processing-based CBM array ment such as switches or sensors and can be imple-
reconfiguration technique was proposed to mitigate mented easily and economically even on large scale
the shading losses. Following the physical relocation, during partial shading conditions. The reported
the electrical connections in the proposed reconfigur- results reveals high efficacy of the proposed configur-
ation remains unchanged. Six shading scenarios such ation, where it outperforms several state-of-the-art
as diagonal, L-shaped, corner, center, arbitrary, and configurations.
Z-shaped shadings were considered to validate the
efficacy of the proposed work. In order to generalize Declaration of Conflicting Interests
the proposed approach, different array sizes were con-
The author(s) declared no potential conflicts of interest with
sidered. The proposed method reported reduced respect to the research, authorship, and/or publication of
power loss and enhanced energy harvest in compari- this article.
son with hitherto known configurations with different
shading conditions. It was observed from the results
Funding
that, for diagonal, L-shaped, corner, center, arbitrary,
and Z-shaped shading conditions, the proposed The author(s) received no financial support for the research,
approach minimized the power loss to 13.73%, authorship, and/or publication of this article.
32.19%, 22.20%, 17.54%, 26.12%, and 24.87% for
8  8 array. Similarly, for 4  4 array, the proposed ORCID iD
approach reduced the power loss to 17.54%, 48.76%, Anshul Agarwal https://orcid.org/0000-0003-0143-564X
17.54%, 17.54%, 34.77%, and 48.76% for all shading Tirupathiraju Kanumuri https://orcid.org/0000-0002-
conditions and in case of 6  6 array, the power loss is 0441-7642
Tatabhatla et al. 575

References 14. Sahu HS, Nayak SK and Mishra S. Maximizing the

1. Villalva MG, Gazoli JR and Ruppert Filho E. power generation of a partially shaded PV array.
Modeling and circuit-based simulation of photovoltaic IEEE J Emerg Select Topics Power Electron 2016; 4:
arrays. In: Brazilian power electronics conference, 626–637.
COBEP’09, 2009, pp.1244–1254. New York: IEEE. 15. Rani BI, Ilango GS and Nagamani C. Enhanced power
2. Pareek S and Dahiya R. Enhanced power generation generation from PV array under partial shading condi-
of partial shaded photovoltaic fields by forecasting tions by shade dispersion using Su Do Ku configur-
the interconnection of modules. Energy 2016; 95: ation. IEEE Trans Sustain Energy 2013; 4: 594–601.
561–572. 16. Satpathy PR and Sharma R. Power loss reduction in
3. Malathy S and Ramaprabha R. Comprehensive ana- partially shaded PV arrays by a static SDP technique.
lysis on the role of array size and configuration on Energy 2018; 156: 569–585.
energy yield of photovoltaic systems under shaded con- 17. Potnuru SR, Pattabiraman D, Ganesan SI, et al.
ditions. Renew Sustain Energy Rev 2015; 49: 672–679. Positioning of PV panels for reduction in line losses
4. Silvestre S, Boronat A and Chouder A. Study of bypass and mismatch losses in PV array. Renew Energy 2015;
diodes configuration on PV modules. Appl Energy 2009; 78: 264–275.
86: 1632–1640. 18. Mahmoud Y and El-Saadany EF. Enhanced reconfig-
5. Mao M, Duan Q, Duan P, et al. Comprehensive uration method for reducing mismatch losses in PV sys-
improvement of artificial fish swarm algorithm for tems. IEEE J Photovolt 2017; 7: 1746–1754.
global MPPT in PV system under partial shading con- 19. La Manna D, Vigni VL, Sanseverino ER, et al.
ditions. Trans Inst Meas Control 2018; 40. Reconfigurable electrical interconnection strategies for
6. Liu L, Liu C, Wang J, et al. Simulation and hardware photovoltaic arrays: a review. Renew Sustain Energy
implementation of a hill-climbing modified fuzzy-logic Rev 2014; 33: 412–426.
for maximum power point tracking with direct control 20. Babu TS, Ram JP, Dragičević T, et al. Particle swarm
method using boost converter. J Vib Control 2015; 21: optimization based solar PV array reconfiguration of
335–342. the maximum power extraction under partial shading
7. Duan Q, Mao M, Duan P, et al. An intelligent algo- conditions. IEEE Trans Sustain Energy 2018; 9: 74–85.
rithm for maximum power point tracking in photovol- 21. Tabanjat A, Becherif M and Hissel D. Reconfiguration
taic system under partial shading conditions. Trans Inst solution for shaded PV panels using switching control.
Meas Control 2017; 39: 244–256. Renew Energy 2015; 82: 4–13.
8. Li H and Liu LQ. A novel particle swarm optimization 22. Parlak KŞ. PV array reconfiguration method under par-
maximum power point tracking control method based tial shading conditions. Int J Electr Power Energy Syst
on the direct current voltage superposition principle 2014; 63: 713–721.
under partially shaded conditions. J Vib Control 2014; 23. Badran O, Mamlook R and Abdulhadi E. Toward
20: 1356–1360. clean environment: evaluation of solar electric power
9. Wang YJ and Hsu PC. An investigation on partial technologies using fuzzy logic. Clean Technol Environ
shading of PV modules with different connection con- Policy 2012; 14: 357–367.
figurations of PV cells. Energy 2011; 36: 3069–3078. 24. Guruambeth R and Ramabadran R. Fuzzy logic con-
10. Josephine R, Suja S and Karunambika G. Combination troller for partial shaded photovoltaic array fed modu-
of fixed configuration and reconfiguration method for lar multilevel converter. IET Power Electron 2016; 9:
maximum power extraction from PV arrays. Adv Nat 1694–1702.
Appl Sci 2014; 8: 67–73. 25. Naeem EA, Elnaby MMA, El-sayed HS, et al. Wavelet
11. Satpathy PR, Sharma R and Jena S. A shade dispersion fusion for encrypting images with a few details. Comput
interconnection scheme for partially shaded modules in Electr Eng 2016; 54: 450–470.
a solar PV array network. Energy 2017; 139: 350–365. 26. Abitha K and Bharathan PK. Secure communication
12. Samikannu SM, Namani R and Subramaniam SK. based on Rubik’s cube algorithm and chaotic baker
Power enhancement of partially shaded PV arrays map. Procedia Technol 2016; 24: 782–789.
through shade dispersion using magic square configur- 27. Fridrich J. Symmetric ciphers based on two-dimen-
ation. J Renew Sustain Energy 2016; 8: 063503. sional chaotic maps. Int J Bifurc Chaos 1998; 8:
13. Rakesh N and Madhavaram TV. Performance enhance- 1259–1284.
ment of partially shaded solar PV array using novel 28. Naeem EA, Elnaby MMA, Soliman NF, et al. Efficient
shade dispersion technique. Front Energy 2016; 10: implementation of chaotic image encryption in trans-
227–239. form domains. J Syst Softw 2014; 97: 118–127.