Renewable Energy 149 (2020) 55e65

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

An optical-energy model for optimizing the geometrical layout of solar

photovoltaic arrays in a constrained field
Naveed ur Rehman a, b, *, Muhammad Uzair c, Usman Allauddin c
a
School of Engineering Trades and Technology, Southern Institute of Technology, Invercargill, 9840, New Zealand
b
School of Engineering, Computer and Mathematical Sciences, Auckland University of Technology, Auckland, 1010, New Zealand
c
Mechanical Engineering Department, NED University of Engineering & Technology, Karachi, 75270, Pakistan

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

Article history: The number of rows of photovoltaic (PV) modules in a field are limited by the area available for
Received 30 September 2019 installation. With the objective of achieving maximum solar energy collection by the modules, deter-
Received in revised form mining the geometric layout becomes an optimization problem that couples the tilt angle and inter-row
18 November 2019
spacing. The existing optimization methods are simplistic as they consider tilt angle and inter-row
Accepted 8 December 2019
spacing as constant values for all rows. This paper presents an optical-energy model for determining
the optimal geometrical layouts of PV arrays by considering different tilt angles and row spacings for
every row, given a field constraint. It is comprised of two sub-models: (i) an optical sub-model that
Keywords:
Inclination angle
determines mutual shading and (ii) an energy sub-model that evaluates the yearly solar energy collec-
Clearance distance tion. Optimization was performed by proposing random layouts for the model and using the one that
Modules layout yielded the maximum yearly collection. A situation in which the proposed model can be used as a
Optical irradiance powerful tool for optimization is explained via a case study conducted in Auckland, New Zealand. The
Solar optics effects of maintaining either a constant tilt angle or a safe row-spacing on year-round performance are
investigated.
© 2019 Elsevier Ltd. All rights reserved.

1. Introduction
Solar energy is the most abundantly available form of renewable
energy on earth [1]. It is sustainable, free and can be converted
directly into electricity using photovoltaic (PV) modules [2].
Depending upon the electric energy demand, more than a single PV
module may be required to fulfill the requirements [3]. For indus-
trial and commercial scale electricity production, several parallel
rows of these modules may be required, known as arrays, which are
installed facing north or south in fields in the southern or northern
hemispheres, respectively.
The arrays are inclined at an angle to maximise solar energy
collection during a given time frame. Several studies have been
conducted in the past to determine the optimum value of the tilt
angle. The methods generally employed include algorithms that
sweep through the angles from 0⁰ to 90⁰ [4e8]; the Maximization
Algorithm [9]; Genetic Algorithm [10,11]; Simulated Annealing
* Corresponding author. School of Engineering Trades and Technology, Southern
Institute of Technology, Invercargill, 9840, New Zealand.
E-mail address: naveed.urrehman@sit.ac.nz (N. Rehman).
[12]; Particle Swarm Optimization (PSO) [13]; Ant Direction Hybrid
Differential Evolution Algorithm [14]; Harmony Search Algorithm
[15]; and Artificial Neural Networks [16,17]. Frequently, the opti-
mum tilt angle for year-round performance is equal, or close to, the
latitude angle of the field [18,19]. However, the analyses in these
methods assume a single module with no obstructions in its sur-
roundings. Hence, they do not account for losses due to mutual
shading (shading by preceding rows). This phenomenon is a com-
mon problem in arrays and needs to be addressed; it creates
complications for module-level power electronic conversion sys-
tems, e.g. Maximum Power Point Trackers (MPPT) in efficiently
extracting power from the PV array [20].
A few studies have investigated the effect of mutual shading on
the design of solar collector fields. Bany and Appelbaum [21] and
Jones and Burkhart [22] found that the latitude angle of the field,
the modules’ dimensions, length of rows, tilt angle and inter-row
spacing are the key parameters contributing to the length and
shape of the shadows. These factors also limit the number of rows
that can be installed in any given field. However, the previous
studies assumed that the tilt angle and the inter-row spacing,
whether given or optimized, was constant for all rows.

https://doi.org/10.1016/j.renene.2019.12.040
0960-1481/© 2019 Elsevier Ltd. All rights reserved.
56 N. Rehman et al. / Renewable Energy 149 (2020) 55e65
Nomenclature
A Surface area of module (m2)
E Energy collected by a module (kJ)
Esite Year-round energy collected by all the modules in the
field (GJ)
G Normal solar beam irradiation (kJ/m2)
H Horizontal dimension of module (m)
L Length of field (m)
M Number of modules in each row
N Number of rows of modules in a field
n
All the rows before the row in observation
b
P Unit vector representing normal to the module
b gs
S Unit vector representing position of sun
V Vertical dimension of module (m)
W Width of field (m)
Y Coordinate of the intersection of shadow with the
module’s foot (m)
Interestingly, this assumption has continued in all subsequent
studies, where regardless of other variables, tilt angles and inter-
row spacings were held constant for all rows. For example,
Appelbaum and Bany [23] developed an approach for obtaining the
optimum number of rows by observing the changes in energy
collection after varying the inter-row spacing. However, the tilt
angle was used as an input to the model rather than a variable to be
optimized. Similarly, Weinstock and Appelbaum [24] investigated
maximising energy collection and sets of constraints coupling the
number of rows. Later, Sadineni et al. [25] evaluated minimum
inter-row spacing, which was again kept constant for all rows in an
inclined solar field with a fixed tilt angle, with the objective of
maximising solar energy collection. Copper et al. [26] presented a
vector-based method to calculate an optimized inter-row spacing
and system electrical size for any surface tilt and orientation, which
was validated by comparing the results with ray-tracing shadow
visualizations in the Ecotect software package. d’Alessandro et al.
[27] presented an automated tool for quickly evaluating the yield of
a PV plant accounting for mutual shading loses. Martín-Chivelet
Fig. 1. (a) A single module can fit in a constrained space when tilted at an optimum angle; (b) Two modules can fit in the same constrained space (one vertical and one horizontal)
without causing mutual shadowing issues.
b Lit condition of module
d Numberth of day of the year
j Length of the projection of module on ground (m)
l Distance of module from northern edge of field (m)
m Numberth of module in a row
n Numberth of module row
s Shadow length of module (m)
t Hour of the year (hr)
y Coordinate of the module’s foot (m)
as Altitude angle of sun (degrees)
aso Altitude angle of sun on the shortest day of year
during noon (degree)
b Tilt angle (degrees)
Azimuth angle of sun (degree)
D Row-spacing (m)
d Declination angle (degree)
q Incidence angle (degree)
4 Latitude angle of field (degree)
[28] proposed a method for maximising the yield based on the
optimizing the packing factor, defined as the ratio between the area
of PV array and the installation area. The optimum value of the tilt
angle was estimated using a correlation based on the latitude angle
of the site, while the inter-row spacing was minimized by setting
the shading losses as the primary criteria. Appelbaum [29]
concluded that the vertical bifacial modules facing east-west in a
field, separated by a fixed inter-row spacing, receive maximum
solar radiation. Horoufiany and Ghandehari [30] presented a
scheme for enhancing the power output of an array considering the
mutual shading conditions. However, it was assumed that all the
rows were facing nearly south, with the same tilt angle and inter-
row spacing. Alsadi and Nassar [31] developed an approach for
evaluating the solar irradiance received by a field having PV mod-
ules separated by constant spacing. The tilt angle was held constant
while simulations were performed to assess the influence of elec-
trical and geometrical design parameters on the performance and
profitability of the solar field. Sanchez-Carbajal and Rodrigo [32]
optimized inter-row spacing, normalized over the field’s length, to
 N. Rehman et al. / Renewable Energy 149 (2020) 55e65 57

Fig. 2. Schematic of the installation field.

maximise the technical and economic parameters of the PV plant,
keeping the tilt angle fixed at the site’s latitude. More recent
studies, such as one using the Sudoku algorithm [33] and another
that optimizes using PSO [34], also assume fixed tilt angles and
inter-row spacings for all rows.
This demonstrates that, although the geometric models avail-
able in the literature as discussed so far can perform a mutual
shading analysis, they can only predict a single value for the tilt
angle and/or the row spacing, which is constant for all rows. If the
tilt angle and the row-spacing are different for each row, the results
become more interesting. For example, if only a single module,
tilted at an optimum angle, can fit in a constrained field, then two
modules (one vertical and one horizontal) can also be fitted in the
same space without causing any mutual shadowing issues, as
illustrated in Fig. 1. The two cases can be compared to see which one
is receiving more energy within a given time period; e.g. in a year.
This opens a whole new domain for optimizing the geometric
layout of solar arrays. It will also affect the research and develop-
ment of modular conversion technologies such as efficient power
optimizers and microinverters, with shade-tolerant features
[35,36].
To evaluate the yearly solar energy collection of arrays with
different tilt angles and row spacings for each row, this study
developed a 3D geometric model, as shown in Section 2. The model

Fig. 3. Shadow of a row of modules on the ground toward the south.

58 N. Rehman et al. / Renewable Energy 149 (2020) 55e65

Fig. 4. A row of modules receiving shadow from the preceding row (a) At solar noon (b) Before solar noon.

is divided into two sub-models: (i) an optical model and (ii) an

energy model, as shown in Sections 2.1 and 2.2, respectively. The
optimization routine is then developed in Section 2.3. Section 3
presents the details of the validation (Section 3.1) and the appli-
cation of the model via a case study conducted in Auckland, New
Zealand (Section 3.2). Conclusions are presented in Section 4 of the
paper.

2. Model

2.1. Optical sub-model

Consider a finite horizontal surface as the installation field with

rows of modules, as shown in Fig. 2. The physical dimensions of the
field are L (length) W (width) and there are N rows of modules,
each having M modules. The dimensions of each module are V H
such that the side of length H is always on ground. Thus, these
modules can have a portrait orientation (V > H) or a landscape
Fig. 5. Illustration of incidence angle.
orientation (V < H). Each module is designated by its row number
(n) and module number in that row (m). The modules are non-
tracking, and are facing true north, as in the southern hemi- that the length (jn ) of the projection of module on the ground can
sphere, this faces the sun. The installation field is always assumed be calculated using Eq. (1):
to be free of shadows from nearby objects and there are no phys-
ical/opaque boundaries or fences.
jn ¼ V cosbn (1)
The tilt angle of the nth row of modules is represented by bn ,
which can range from 0 (horizontal) to 90 (vertical). This means The position of the sun can be defined by its altitude angle (as )
 N. Rehman et al. / Renewable Energy 149 (2020) 55e65 59

Fig. 6. Illustration of row-spacing between the modules.

and azimuth angle (gs ). The altitude angle is measured as a positive
value upward from ground and the azimuth angle is negative, 0 or
positive to the east, north and west, respectively. At any instant of
time in a day (when as > 0), the nth row will cast a shadow on the
ground. The length of that shadow towards the south (sn ),
measured from the southward edge of the ground projection of the
module, can be obtained by tracing the solar rays in a forward di-
rection (Fig. 3). This can mathematically be calculated using Eq. (2):
If the coordinates of the intersection lie between the coordinates
of the module’s base (ym;n;east  Yn
;n  ym;n;west ), the module is
said to be receiving shadow. Since the output of a photovoltaic
module is reduced significantly even when it is only partially in
shadow, the lit condition is therefore represented by a binary
number b (¼1 means producing output ¼ 0 means no output).
2.2. Energy sub-model

V sinbn The solar beam energy collected by the ðn; mÞ module during the
sn ¼ (2)
tanas t th hour of the year is represented by En;m;t and can be evaluated
using Eq. (5):
Hence, if the nth row is a distance ln away from the northward
edge of the field, at solar noon (gs ¼ 0 ) all of its modules will En;m;t ¼ GAbn;m;t cosqn;t (5)
receive at least some shadow from the preceding rows (n
), only if
ðln
þ jn
þ sn
Þ > ln , as shown in Fig. 4(a). However, if this condi- where G represents the hourly average normal beam irradiation
tion is satisfied during the time when sun is away from its noon that can be obtained from the solar databases [37,38], A is the area
position (e.g. in the morning or afternoon i.e. gs s0 ), the row may of the module ( ¼ V  H), b represents the lit condition of the
partially be in shadow. To determine whether the mth module in module (obtained from the optical sub-model) and qn;t is the
the nth row is in shadow, firstly the coordinates of the eastern incidence angle, which can be calculated from knowledge of the
(ym;n;east ) and western (ym;n;west ) corners of the module’s foot are vector dot-product between unit vectors representing normal to
obtained using Eq. (3): b n ) and the position of the sun (b
the module ( P S t ), as mathematically
    shown in Eq. (6) and graphically illustrated in Fig. 5:
ym;n;east ðm
1ÞH
¼ (3) b n :b
ym;n;west mH Cosqn;t ¼ P St (6)
Then, the coordinate of the point where the shadow of the east-
where,
most (or west-most) module of all the preceding rows (n
) in-
tersects the foot of the nth row is obtained by tracing the solar rays b n ¼ ðsinb Þbi þ ðcosb Þb
P n n k (7)
in the forward direction, such that (Fig. 4(b)):
 and
ðln
þ jn

ln Þtanjgs j; gs < 0 ðEastÞ
Yn
;n ¼ (4)
½W
ðln
þ jn

ln Þtanjgs j; gs > 0 ðWestÞ S t ¼ ðcosgst cosast Þbi þ ðsingst cosast Þbj þ ðsinast Þb
b k (8)

Table 1
Summary of the parameters used in the optimization routine.

Parameters Constant b Design Safe D Design

Objective Maximise Esite

Location’s constraints Latitude of field (4)
Declination angle on the relevant solstice day (d)
Hourly solar radiation database (e.g. from TMY file)
Geometrical constraints Field’s dimensions (W and L)
Modules’ dimensions (H and V )
Design subjective constraints Tilt angle of all rows is constant (b) Inter-row spacing is constant (Dn;safe )
Variables to optimize Number of total rows (N)
Inter-row spacing of every row (Dn ) Tilt angle of every row (bn )
60 N. Rehman et al. / Renewable Energy 149 (2020) 55e65

Fig. 7. Flow chart describing (a) the overall process of optimization; and (b) the proposed new random geometrical layout.

Table 2 The Esite is a very useful parameter as it describes the yearly solar
Scenarios set up to illustrate and validate the optical energy model. potential of a given constrained field, considering the mutual
Scenarios V (m) H (m) as gs shading when the modules’ rows are variably tilted and/or spaced
Scenario-I 2m 1m 30 0 (North)
in the field. The maximization of this parameter is discussed in the
Scenario-II 45 (East) next section.
Scenario-III 1m 2m 45 0 (North)
2.3. Optimization routine

Thus, the year-round collection of solar beam energy by the During the optimization, the year-round solar beam energy
modules in the field is represented by Esite and can be found using collections (Esite ) from the random layouts of modules were eval-
Eq. (9): uated, compared, and the one producing the maximum value was
chosen as the optimum layout. While proposing a new layout, the
XX
Esite ¼ En;m;t (9) rows of modules were kept within the physical boundaries of the
t ðm;nÞ field. Two types of designs were analysed: “Constant b”, which was
the group of layouts in which the tilt angle of each row was kept
 N. Rehman et al. / Renewable Energy 149 (2020) 55e65 61

Fig. 8. (a) Side-view of Scenario-I & II, (b) Top-view of Scenario-II and (c) Side-view of Scenario-III.
62 N. Rehman et al. / Renewable Energy 149 (2020) 55e65
Fig. 9. Lit condition and the angle of incidence evaluated for (a) Scenario-I (b)
Scenario-II and (c) Scenario-III.
same but the spacing between rows was varying. In contrast, “Safe
D” was the group of layouts in which each row could have a
different tilt angle, but each row was placed a safe distance from the
previous row. Here, the row spacing (Dn ) is the distance between
the foot of the nth row and the ground projection of the preceding
row (ln
1 þ jn
1 ), as shown in Fig. 6. The minimum clearance could
be zero, meaning that the next row could start just after the pre-
ceding row. In contrast, its maximum value could be infinite.
However, due to the limitations of the finite dimensions of the field,
the maximum value was considered as the length of the shadow of
the preceding row at solar noon on the shortest day of the year. In
this study, it was termed the safe row-spacing between two rows
(Dn;safe ). So, if the altitude angle of the sun during such a time is aso ,
Eq. (2) can be modified to yield the safe row-spacing:
Fig. 10. Cumulative mean yearly energy collection over the number of iterations
(portrait orientation in constant b design).
V sinbn
1
Dn;safe ¼ (10)
tanaso
whereas aso can be obtained by using [39]:
sinaso ¼ cos 4 cos d þ sin 4 sin d (11)
in which 4 is the latitude angle of the field and d is the declination
angle on the relevant solstice day (having the lowest position of the
sun at noon), given by:
 
284 þ d
d ¼ 23:45 sin 360 (12)
365
where d ¼ 172 (21 June) in the southern hemisphere and d ¼ 356
(22 December) in the northern hemisphere.
A summary of the parameters used in the optimization routine
is presented in Table 1. To propose a random layout, the following
algorithm was followed:
Step-1 The feet of the modules of the very first row were at the
northern edge of the field (i.e. l1 ¼ 0)
Step-2 Choose the tilt angle
In Constant b design: The predefined value of tilt angle
was chosen as it was constant for all rows.
In Safe D design: The tilt angle was chosen randomly
between 0 and 90 .
Step-3 If the distance between the northern edge of the field and
the southern edge of the projection of the row on the
ground was more than the total length of the field (i.e. ln þ
jn > L), this row was not considered as it crossed the
southern boundary of the field. The procedure at this point
was to stop adding any further rows of modules and pro-
pose this layout.
Step-4 Choose row-spacing for the next row
In Constant b design: The value for clearance was chosen
randomly between 0  Dn  Dn;safe .
In Safe D design: The value of clearance was Dn ¼ Dn;safe .
Step-5 If the distance between the northern edge of the field and
the end of the clearance was more than the total length of
the field (i.e. ln þ jn þ cn > L), the procedure was to stop
adding any further rows of modules and propose this
layout.
 N. Rehman et al. / Renewable Energy 149 (2020) 55e65
Fig. 11. Optimized layouts obtained by considering different designs and orientations.
Fig. 12. Yearly solar beam energy collected by the optimized layouts.
Step-6 Consider adding a new row (calculate the distance of the
new row from the northern edge of the field and then go to
Step-2).
Once the layout had been proposed, the optical-energy model
was executed to obtain its yearly performance. The result was
compared with the previous maximum value to see if this was a
new maximum value. The process was iterated several times before
determining the optimum layout. To check the adequacy of the
number of iterations, the cumulative averages of the results were
plotted against the number of iterations. If the plot became smooth
(converged), the number of iterations were deemed to be adequate.
63
The flow charts describing the overall optimization algorithm
and the method for proposing the new random geometrical layout
are provided in Fig. 7 (a) and (b), respectively.
3. Results and discussion
3.1. Validation
Three scenarios using different layouts under different sun po-
sitions were established to validate the optical-energy model
(Table 2). No optimization was performed at this step, so the tilt
angles and row spacings were chosen randomly. The difference
between Scenario-I and Scenario-II was the position of the sun; in
the former scenario, the sun was at noon while in the latter, the sun
was toward the east (morning). In these two scenarios, the modules
had a portrait orientation (V > H). In contrast, in Scenario-III, the
modules were in landscape orientation (V < H) and the sun was at
the noon position, which was similar to Scenario-I. The dimensions
of the installation field were 10mðLÞ  6mðWÞ. This gave 4 rows of
modules and 6 modules per row (total 24 modules) in Scenarios I
and II, and 10 rows of modules and 3 modules per row (total 30
modules) in Scenario III. The results from the optical sub-model for
these scenarios were compared with the results obtained from the
3D geometric models developed in Google SketchUp [40] and
showed good agreement.
The layouts of the modules in these scenarios and the length of
shadows are shown in Fig. 8. In Scenarios I and II, as the altitude angle
of the sun is same, the length of the shadows towards south will be the
same, as shown in Fig. 8(a). For Scenario I, all the rows except the first
one were found to be in the shadow and hence, would not be pro-
ducing any output. However, as the sun is towards the east in Scenario
II, some of the rows receiving shadows may be partially lit. Fig. 8(b)
shows that the first 3 modules towards the east of the field in the 3rd
64 N. Rehman et al. / Renewable Energy 149 (2020) 55e65
Fig. 13. Insights about the yearly solar beam energy collected by each module in the optimized layouts.
row are in the light. For Scenario III, the first 4 rows are illuminated
while the rest are in shadow, as shown in Fig. 8(c).
To determine the energy received by the modules, the energy
sub-model was applied to these scenarios. A constant value of G ¼
300kJ=m2 during the simulation hour was assumed, the area of
each module was 2 m2, the lit condition of each module was ob-
tained using the optical sub-model, and the angle of incidence for
each module was obtained using knowledge of vectors. The results
for the lit condition and the incidence angle are shown in Fig. 9. The
energy sub-model yielded the beam energies collected in Scenario-
I, Scenario-II and Scenario-III, which were found to be 2949 kJ,
3844 kJ and 6761 kJ, respectively.
3.2. Application
The model was then used as a powerful tool for optimizing the
layouts of modules, with the objective of maximising the yearly
solar beam energy collection. A computer program was developed
in an open-source environment for executing the optimization
routine and performing calculations as required in the optical en-
ergy model. As a case study, the simulations were performed
considering a 20mðLÞ  10mðWÞinstallation field in Auckland, New
Zealand. The module size was 2m  1m and modules were installed
in either portrait or landscape orientations. Both the Constant b and
Safe D designs were considered. For the former design, the tilt angle
of the modules was assumed equal to the latitude angle of the field
(b ¼ 4 ¼ 36:84 ), as recommended in several studies [39]. In
contrast, for the latter design, aso ¼ 29:7 was evaluated using Eq.
(11). The hourly sun positions and irradiation data were obtained
using TRNSYS software [38].
For each design and orientation, the optimization routine was
executed, iterating ~2.5 million random layouts. The adequacy of
the number of iterations was confirmed by visually inspecting the
smoothness of the cumulative mean of the results over the number
of iterations. As an example, Fig. 10 is shown, showing the
convergence of the cumulative mean of the simulations performed
for the case when the modules were installed in portrait orientation
using the Constant b design.
The final, optimized layouts for each group of designs and ori-
entations are shown in Fig. 11. The optimization produced a layout
consisting of 6 rows with 10 modules per row, totalling 60 modules
for portrait orientation in the Constant b design, which was the
smallest number of modules compared to the other layouts. For
both the layouts with horizontal orientations, the number of rows
and number of modules per row were found to be 13 and 5,
respectively, totalling 65 modules. The largest number of modules
was found for portrait orientation when the Safe D design was used,
involving 7 rows of modules with 10 modules per row, totalling 70
modules.
The yearly energy collected by all the modules in the optimized
layouts is compared in Fig. 12. Overall, the Safe D designs were
found to collect more energy compared to the Constant b designs.
The Safe D design with modules in portrait orientation was found to
receive 362 GJ of energy, which is 9% higher than the landscape
orientation for the same design (332 GJ) and 14% higher than the
portrait orientation in the Constant b design (317 GJ). The landscape
orientation in Constant b design received the smallest amount of
energy (263 GJ), which was 27% less than the maximum.
More insights about the yearly energy collection in the opti-
mized layouts are shown in Fig. 13. This is useful for determining
the critical rows or modules in the layouts that receive the largest
fraction of the solar energy. Here, the modules in the front row of
each layout are considered the benchmark, and are receiving 100%
of the energy, given their tilt angle. This is because they are never in
shadow at any time of the year. In the portrait orientation, the
modules at the periphery of both the Constant b and Safe D designs
are critical as they receive a higher fraction of the available energy
compared to the modules in the centre of the layout. Also, in both
the designs involving a landscape orientation, all the modules in a
single row seem to receive the same fraction of the energy. Hence,
the whole row is either critical or not. As an example, the 5th row
from the left in the landscape orientation of Constant b seems to be
receiving a negligible amount of energy compared with the first
row of modules.
4. Conclusion
This study developed an optical energy model for evaluating
yearly solar beam energy collection by solar photovoltaic modules.
The model was then used to optimize the layout of these modules
in an installation field with constrained physical dimensions. The
optical sub-model used geometrical relations, applying forward
ray-tracing to determine the shadows of rows on the ground and on
subsequent rows of modules. This helped to determine the lit
condition of each individual module, which is crucial. The energy
sub-model used the results from the optical sub-model, an hourly
solar radiation database, the incidence angle between the modules
and the sun’s position to evaluate the year-round solar beam
 N. Rehman et al. / Renewable Energy 149 (2020) 55e65
energy collection in each given layout. Three scenarios were
established using different layouts and the sun’s position for vali-
dation purposes. The results were found to be in good agreement
with measurements made in 3D models developed in Google
SketchUp.
For optimizing the layout of solar modules in constrained
installation fields, an optimization routine was also proposed. The
purpose of that routine was to generate random layouts, simulate
them to determine the yearly energy collection, and finally propose
the optimum layout for collecting the maximum amount of energy.
To demonstrate that it could be used as tool for optimizing layouts,
a 10 m  20 m field was considered in Auckland, New Zealand.
Installations of 2 m  1 m modules were considered in both portrait
and landscape orientations. The maximum amount of energy
(362 GJ) could be collected when the modules were installed in a
portrait orientation, allowing for a minimum safe clearance be-
tween them. The next best layout (collecting 332 GJ) was where the
modules were installed in landscape orientation while again
considering a minimum safe clearance between them. In both of
these layouts, the tilt angle of the rows of modules was variable. A
landscape orientation with a constant tilt angle was found to be the
worst layout, collecting only 263 GJ of energy over the year.
Some insights about the energy collected by each module in
these layouts were also achieved. This helped in determining the
critical modules in each field that seemed to be receiving the ma-
jority of the energy throughout the year. In the portrait orientation,
the modules at the periphery were found to be more critical than
those in the centre of the arrays. For the landscape orientations,
each module in an individual row received almost the same fraction
of the total incoming energy. However, different rows may receive
different fractions of the available energy throughout the year.
Future research directions include evaluating the electric yield,
performance, cost, and life of the system while taking into account
the specifications of the electrical components and mechanical
structure.
Author contributions section
Naveed ur Rehman: Conceived and designed the analysis,
Collected the data, Contributed data or analysis tools, Performed
the analysis, Wrote the paper. Muhammad Uzair: Collected the
data, Contributed data or analysis tools, Wrote the paper. Usman
Allauddin: Contributed data or analysis tools, Performed the
analysis, Wrote the paper.
Declaration of competing interest
There is no conflict of interest between other authors.
References
[1] A. Alam, M. Siddiqui, N. Rehman, Solar feed water heating feasibility for a
conventional steam power plant, J. Mech. Sci. Technol. 31 (7) (2017)
3573e3580.
[2] N. Rehman, M. Siddiqui, A novel methodology for determining sky blocking by
obstacles viewed virtually from any location on site, Energy Build. 128 (2016)
827e833.
[3] M. Yousuf, M. Siddiqui, N. Rehman, Solar energy potential estimation by
calculating sun illumination hours and sky view factor on building rooftops
using digital elevation model, J. Renew. Sustain. Energy 10 (1) (2018), 013703.
[4] M. Yakup, A. Malik, Optimum tilt angle and orientation for solar collector in
Brunei Darussalam, Renew. Energy 24 (2) (2001) 223e234.
[5] S. Armstrong, W. Hurley, A new methodology to optimise solar energy
extraction under cloudy conditions, Renew. Energy 35 (4) (2010) 780e787.
[6] M. Kacira, M. Simsek, Y. Babur, S. Demirkol, Determining optimum tilt angles
and orientations of photovoltaic panels in Sanliurfa, Turkey, Renew. Energy 29
(8) (2004) 1265e1275.
[7] M. Despotovic, V. Nedic, Comparison of optimum tilt angles of solar collectors
determined at yearly, seasonal and monthly levels, Energy Convers. Manag. 97
65
(2015) 121e131.
[8] M. Benghanem, Optimization of tilt angle for solar panel: case study for
Madinah, Saudi Arabia, Appl. Energy 88 (4) (2011) 1427e1433.
[9] E. Calabro, An algorithm to determine the optimum tilt angle of a solar panel
from global horizontal solar radiation, J. Renew. Energy (2013) 2013.
[10] P. Talebizadeh, M. Mehrabian, M. Abdolzadeh, Prediction of the optimum
slope and surface azimuth angles using the Genetic algorithm, Energy Build.
43 (11) (2011) 2998e3005.

[11] V. Congradac, M. Prica, M. Paspalj, D. Bojani 
c, D. Capko, Algorithm for blinds
control based on the optimization of blind tilt angle using a genetic algorithm
and fuzzy logic, Sol. Energy 86 (9) (2012) 2762e2770.
[12] Y. Chen, C. Lee, H. Wu, Calculation of the optimum installation angle for fixed
solar-cell panels based on the genetic algorithm and the simulated-annealing
method, IEEE Trans. Energy Convers. 20 (2) (2005) 467e473.
[13] Y. Chang, Optimal the tilt angles for photovoltaic modules using PSO method
with nonlinear time-varying evolution, Energy 35 (5) (2010) 1954e1963.
[14] Y. Chang, An ant direction hybrid differential evolution algorithm in deter-
mining the tilt angle for photovoltaic modules, Expert Syst. Appl. 37 (7) (2010)
5415e5422.
[15] M. Guo, H. Zang, S. Gao, T. Chen, J. Xiao, L. Cheng, Z. Wei, G. Sun, Optimal tilt
angle and orientation of photovoltaic modules using HS algorithm in different
climates of China, Appl. Sci. 7 (10) (2017) 1028.
[16] Y. Chang, Optimal design of discrete-value tilt angle of PV using sequential
neural-network approximation and orthogonal array, Expert Syst. Appl. 36 (3)
(2009) 6010e6018.
[17] E. Mehleri, P. Zervas, H. Sarimveis, J. Palyvos, N. Markatos, Determination of
the optimal tilt angle and orientation for solar photovoltaic arrays, Renew.
Energy 35 (11) (2010) 2468e2475.
[18] J. Kern, I. Harris, On the optimum tilt of a solar collector, Sol. Energy 17 (2)
(1975) 97e102.
[19] H. Gunerhan, A. Hepbasli, Determination of the optimum tilt angle of solar
collectors for building applications, Build. Environ. 42 (2) (2007) 779e783.
[20] S. Hosseini, S. Taheri, M. Farzaneh, H. Taheri, A high performance shade-
tolerant MPPT based on current-mode control, IEEE Trans. Power Electron.
34 (10) (2019) 10327e10340.
[21] J. Bany, J. Appelbaum, The effect of shading on the design of a field of solar
collectors, Sol. Cells 20 (3) (1987) 201e228.
[22] R. Jones Jr., J. Burkhart, Shading effects of collector rows tilted toward the
equator, Sol. Energy 26 (6) (1981) 563e565.
[23] J. Appelbaum, J. Bany, Shadow effect of adjacent solar collectors in large scale
systems, Sol. Energy 23 (6) (1979) 497e507.
[24] D. Weinstock, J. Appelbaum, Optimal solar field design of stationary collectors,
J. Sol. Energy Eng. 126 (3) (2004) 898e905.
[25] S. Sadineni, R. Boehm, R. Hurt, Spacing analysis of an inclined solar collector
field, in: ASME 2008 2nd International Conference on Energy Sustainability
Collocated with the Heat Transfer, Fluids Engineering, and 3rd Energy
Nanotechnology Conferences, 2009.
[26] J. Copper, A. Sproul, A. Bruce, A method to calculate array spacing and po-
tential system size of photovoltaic arrays in the urban environment using
vector analysis, Appl. Energy 161 (2016) 11e23.
[27] V. d’Alessandro, F. Di Napoli, P. Guerriero, S. Daliento, An automated high-
granularity tool for a fast evaluation of the yield of PV plants accounting for
shading effects, Renew. Energy 83 (2015) 294e304.
[28] N. Martín-Chivelet, Photovoltaic potential and land-use estimation method-
ology, Energy 94 (2016) 233e242.
[29] J. Appelbaum, Bifacial photovoltaic panels field, Renew. Energy 85 (2016)
338e343.
[30] M. Horoufiany, R. Ghandehari, Optimal fixed reconfiguration scheme for PV
arrays power enhancement under mutual shading conditions, IET Renew.
Power Gener. 11 (11) (2017) 1456e1463.
[31] S. Alsadi, Y. Nassar, Estimation of solar irradiance on solar fields: an analytical
approach and experimental results, IEEE Trans. Sustain. Energy 8 (4) (2017)
1601e1608.
[32] S. Sanchez-Carbajal, P. Rodrigo, Optimum array spacing in grid-connected
photovoltaic systems considering technical and economic factors, Int. J. Pho-
toenergy (2019) 2019.
[33] M. Horoufiany, R. Ghandehari, Optimization of the Sudoku based reconfigu-
ration technique for PV arrays power enhancement under mutual shading
conditions, Sol. Energy 159 (2018) 1037e1046.
[34] M. Shams, M. Kia, A. Heidari, D. Zhang, Optimal design of photovoltaic solar
systems considering shading effect and hourly radiation using a modified PSO
algorithm, Simulation (2019), 0037549719831362.
[35] D. Vinnikov, R. Kosenko, A. Chub, E. Liivik, Shade-tolerant photovoltaic
microinverter with time adaptive seamless P-V curve sweep MPPT, in: 19th
European Conference on Power Electronics and Applications, Warsaw, Poland,
2017.
[36] K. Niazi, Y. Yang, M. Nasir, D. Sera, Evaluation of interconnection configuration
schemes for PV modules with switched-inductor converters under partial
shading conditions, Energies 12 (14) (2019) 2802.
[37] NREL [Online]. Available, http://rredc.nrel.gov/solar/old_data/nsrdb/, 11 2016.
[38] Solar Energy Laboratory (SEL), TRNSYS 17 User Manuals, Solar Energy Labo-
ratories, Madison, 2012.
[39] J.A. Duffie, W.A. Beckman, Solar Engineering of Thermal Processes, 2013.
[40] Google, Google SketchUp - 3D modeling for Everyone [Online]. Available,
https://www.sketchup.com/. (Accessed 28 May 2018).