machines
Article
Multi-Objective Optimization and Optimal Airfoil Blade
Selection for a Small Horizontal-Axis Wind Turbine (HAWT)
for Application in Regions with Various Wind Potential
Vahid Akbari 1 , Mohammad Naghashzadegan 2, *, Ramin Kouhikamali 2 , Farhad Afsharpanah 3
and Wahiba Yaïci 4
Citation: Akbari, V.; Naghashzadegan,
M.; Kouhikamali, R.; Afsharpanah, F.;
Yaïci, W. Multi-Objective Optimization
and Optimal Airfoil Blade Selection
for a Small Horizontal-Axis Wind
Turbine (HAWT) for Application in
Regions with Various Wind Potential.
Machines 2022, 10, 687. https://
doi.org/10.3390/machines10080687
Academic Editors: Davide Astolfi,
Antonio J. Marques Cardoso and
Francesco Castellani
Received: 23 June 2022
Accepted: 12 August 2022
Published: 13 August 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Machines 2022, 10, 687. https://doi.org/10.3390/machines10080687
1 Department of Mechanical Engineering, University Campus 2, University of Guilan, Rasht 41996, Iran
2 Department of Mechanical Engineering, University of Guilan, Rasht 41996, Iran
3 Department of Mechanical Engineering, Babol Noshirvani University of Technology, Babol 47148, Iran
4 CanmetENERGY Research Centre, Natural Resources Canada, Ottawa, ON K1A 1M1, Canada
* Correspondence: naghash@guilan.ac.ir
Abstract: The type of airfoil with small wind turbine blades should be selected based on the wind
potential of the area in which the turbine is used. In this study, 10 low Reynolds number airfoils,
namely, BW-3, E387, FX 63-137, S822, S834, SD7062, SG6040, SG6043, SG6051, and USNPS4, were
selected and their performance was evaluated in a 1 kW wind turbine in terms of the power coefficient
and also the startup time, by performing a multi-objective optimization study. The blade element
momentum technique was utilized to perform the calculations of the power coefficient and startup
time and the differential evolution algorithm was employed to carry out the optimization. The results
reveal that the type of airfoil used in the turbine blade, aside from the aerodynamic performance,
completely affects the turbine startup performance. The SG6043 airfoil has the highest power
coefficient and the BW-3 airfoil presents the shortest startup time. The high lift-to-drag ratio of the
SG6043 airfoil and the low inertia of the turbine blades fitted with the BW-3 airfoil make them suitable
for operation in windy regions and areas with low wind speeds, respectively.
Keywords: multi-objective optimization; small wind turbine; renewable energy; turbine blade; airfoil;
wind power; power coefficient; artificial neural network (ANN); startup behavior; wind potential
1. Introduction
In today’s developed world, energy is a cardinal factor for the sustainable development
and economic growth of a country [1], and, in recent years, continuous access to inexpensive
and reliable energy has become an essential right for humanity [2]. However, the rise
in energy demand and dependence on fossil fuels has led to energy crises, declining
resources, and environmental pollution [3]. Although optimizing energy systems [4,5] and
employing energy storage devices [6] are considered by scholars, the use of renewable
energy sources [7] is one of the most effective solutions to cope with these difficulties.
Among renewable energy sources, wind energy is the most common and sustainable type
of energy and, currently, the application of this free energy source is being considered more
than ever [8]. Harnessing wind energy and converting it to electrical power is carried out
with the aid of wind turbines.
The shape of a wind turbine blade has a substantial role in its performance and should
be determined according to the design goal or goals. Compared to the experimental and
simulation studies conducted on wind turbine blades, optimization algorithms present
more feasible and cost-effective solutions for acquiring blade sections with suitable per-
formance in different scenarios. The enhancements obtained with these algorithms result
from their searching strategy. Contrary to the other techniques, these algorithms do not
depend on searching and testing the entire possibilities. These algorithms make an effort to
https://www.mdpi.com/journal/machines
Machines 2022, 10, 687 2 of 22

acquire the optimal solutions by trying the model performance at each trial by the means
of an objective function.
Recently, nature-inspired metaheuristic strategies for solving different optimization
problems, especially in computer science and engineering, have rapidly increased.
Soni et al. [9] compared the performance of 12 nature-inspired optimization algorithms, in-
cluding Bat, Lion, Particle swarm, Water wave, Elephant herding, Optics inspired, Cuckoo
search, Flower, Genetic, Differential evolution (DE), Harmony, and Simulated annealing
algorithms. They discussed the speed, accuracy, performance, convergence, efficiency, and
complexity of these algorithms. Their findings revealed that compared to the other selected
algorithms, the differential evolution algorithm presents fast speed and performance, as
well as the best accuracy.
Tušar and Filipič [10] compared the performance of genetic algorithms and differential
evolution in solving multi-objective optimization problems. Their results showed that the
DE algorithm provides better and more stable solutions than genetic algorithms. Similar
results were observed in the research of Lilla et al. [11].
According to the International Electrotechnical Commission (IEC) 61400-2 [12], a small
horizontal-axis wind turbine has a swept area of less than 200 m2 , which corresponds to a
blade radius and a power output of less than 8 m and around 50 kW, respectively. In a fur-
ther subdivision, small wind turbines (SWTs) are divided into three groups, namely micro,
mid, and mini, with output power less than 1, 5, and 50 kW, respectively [13]. Although
large wind turbines have gained significant attention from researchers [14], SWTs have
not received the research interest they deserve. These turbines can be employed to supply
electricity for domestic and agricultural purposes, and their application is economically
viable [15]. Unlike large wind turbines which are used in windy areas, these turbines can
also be installed in places where there is no potential for continuous high-speed wind [16].
An important difference between large and SWTs is the lack of pitch controllers in
SWTs for reducing costs [17]. Thus, in these turbines, optimally positioning the blade
sections in the wind direction is not possible. This point is significantly important when
the SWTs are beginning to rotate from a stationary state, since throughout the startup
process, the high angles of attack along the blade, result in the reduction in the lift force
and aerodynamic torque, and, consequently, the blade rotates slowly and power generation
is delayed [18].
The time required for the turbine to accelerate from a stationary state to a tip speed
ratio of around one is called the startup time (Ts ) [19]. In addition to achieving the maximum
power coefficient (Cp ), which is the main goal in designing wind turbine blades, reducing
this time has also been one of the interesting topics for researchers in recent years. In
this regard, the optimization of SWT blades has been performed to maximize the Cp and
minimize Ts [19]. The results showed that most of the turbine power is obtained from
the blade tip part while the root section can facilitate the startup process. By properly
distributing the chord length and twist angle in these two parts of the blade, a significant
improvement was observed in the Cp and Ts of these turbines. Pourrajabian et al. [20]
compared the performance of solid and hollow blades in terms of Cp and Ts in an SWT. It
was observed that compared to the solid blades, the hollow blades exhibit a higher Cp and
lower Ts . Rahgozar et al. [21] studied the influence of linear and non-linear distributions of
the twist angle and chord length on the Cp and Ts of a small horizontal-axis wind turbine.
The results showed that, while the use of non-linear distribution can increase the Cp in
windy areas, with linear distribution the blades have a better Ts in areas with low wind
speeds. Minimizing the aerodynamic noise [22], mass, and cost [23] are other goals of
designing and optimizing SWT blades that have been studied by researchers.
The shape of the airfoil determines the drag and lift coefficients (Cd and Cl ) and plays
a crucial role in the aerodynamic torque generated by the wind turbine blade and its
output power. In this regard, researchers have always attempted to use airfoils with high
aerodynamic efficiency in blade design [24].
 Machines 2022, 10, 687 3 of 22

In general, in large turbines, several airfoils are used in different parts of the blades,
each of which has its own task. Thin airfoils with better aerodynamic performance are used
at the tip of the blade to provide more power, while thick airfoils are employed at the root
of large blades to resist structural stresses [25]. Unlike large blades, to reduce construction
costs as much as possible, the airfoil type of the SWT blades does not change [26]. The
operating Reynolds number (Re) range of SWTs is less than 500,000 [27]. Therefore, there
are not a variety of airfoils to use in these turbines. In this range of Re, the flow on the
upper surface of the airfoil is often laminar, so the formation of a laminar separation bubble
is possible. To solve this problem and to reduce the adverse pressure gradient on the
upper surface of the airfoil, airfoils thinner than the traditional ones are recommended
for operation in SWT blades [28]. Giguere and Selig [29] designed a group of SG series
airfoils (SG6040–SG6043) specifically for use in SWT blades. Studies have shown that
increasing the leading edge nose radius, as well as cusping of the trailing edge, improves
the aerodynamic performance of thin airfoils, and, in this regard, some airfoils have been
proposed specifically for use in SWTs [30].
Given the importance of the airfoil type used in the blade for the aerodynamic per-
formance of a wind turbine, its proper selection, in terms of the startup process, needs
further study. The previous works that were performed in this field are mostly focused
on the distribution of chord length and twist angle along the blade, and the effect of the
airfoil type used in the blade has not been investigated from the startup viewpoint, which
is discussed in the present study. For this purpose, the performance of 10 different airfoils
that are specifically designed for SWT blades is examined in terms of the Cp and Ts .
The remainder of this study is organized as follows. The selected airfoils are introduced
in Section 2. Section 3 presents the numerical techniques that are employed, as well as
their validations. The results of the study which include the performance evaluation of the
airfoils in windy areas and areas with low wind speed are presented in Section 4. Finally,
the conclusions are discussed in Section 5.

2. The Selected Airfoils

Despite the variety of airfoils available for use in large wind turbine blades, there
are not many airfoils designed for SWTs. By examining previous research studies and
also considering the operating Re of the turbine blade investigated in the present study,
10 airfoils that are used in the industry of SWTs, including, BW-3 [31], E387 [30], FX 63-
137 [30], S822 [31], S834 [30], SD7062 [31], SG6040 [31], SG6043 [31], SG6051 [32], and
USNPS4 [31] were selected for examination. Their aerodynamic coefficients, which are
presented in Figure A1 of Appendix A, were tabulated and coded for various Re and angle
Machines 2022, 10, 687 of attack values, and they were employed in the numerical method. The geometry of4the of 23
selected airfoils has been shown in Figure 1. In this figure, x/c indicates the distance along
the chord line, while y/c shows the airfoil thickness normalized with the chord length.
0.15 0.15

0.10 0.10

0.05 0.05
y/c

y/c

0.00 0.00

−0.05 −0.05

−0.10 −0.10
BW−3 (5%) E387 (9.1%)
−0.15 −0.15
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
x/c x/c
0.15 0.15

0.10 0.10

0.05 0.05
y/c
y/c

0.00 0.00

−0.05 −0.05

−0.10 −0.10
FX 63−137 (13.7%) S822 (16%)
−0.15 −0.15
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
x/c x/c
0.15 0.15

0.10 1. Cont.
Figure 0.10

0.05 0.05
y/c

y/c

0.00 0.00

−0.05 −0.05

−0.10 −0.10
S834 (16%) SD7062 (14%)
−0.15 −0.15
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
x/c x/c
 −0.15 −0.15
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
x/c x/c
0.15 0.15

0.10 0.10

0.05 0.05

y/c
y/c
0.00 0.00

Machines 2022, 10, 687 −0.05 −0.05 4 of 22

−0.10 −0.10
FX 63−137 (13.7%) S822 (16%)
−0.15 −0.15
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
x/c x/c
0.15 0.15

0.10 0.10

0.05 0.05

y/c

y/c
0.00 0.00

−0.05 −0.05

−0.10 −0.10
S834 (16%) SD7062 (14%)
−0.15 −0.15
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
x/c x/c
0.15 0.15

0.10 0.10

0.05 0.05

y/c
y/c

0.00 0.00

−0.05 −0.05

−0.10 −0.10
SG6040 (16%) SG6043 (10%)
−0.15 −0.15
0.0 0.2 0.4 0.0 0.2 0.4 0.6 0.8 1.0
x/c 0.6 0.8 1.0
x/c
0.15 0.15

0.10 0.10

0.05 0.05
y/c

y/c
0.00 0.00

−0.05 −0.05

−0.10 −0.10
SG6051 (12%) USNPS4 (11.9%)
−0.15 −0.15
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4
x/c x/c 0.6 0.8 1.0

Figure 1.
Figure The shape
1. The shape of
of the
the selected
selected airfoils.
airfoils.
3. Numerical Procedure
3. Numerical Procedure
3.1. Multi-Objective Optimization
3.1. Multi-Objective Optimization
A multi-objective optimization was used to determine how to optimally distribute the
twistAangle
multi-objective optimization
and chord length along the was used
blade fortoeachdetermine how toairfoils
of the selected optimally distribute
to maximize
the
the twist
Cp andangle and chord
minimize length
Ts . For along theinblade
this purpose, for each
the present of the
study, theselected airfoils
differential to max-
evolution
(DE) algorithm
imize the Cp andwhich
minimize was first introduced
Ts. For by Storn
this purpose, et al.
in the [33] instudy,
present 1997 was
the employed.
differentialDE evo-
is a population-based method that, among the recently presented
lution (DE) algorithm which was first introduced by Storn et al. [33] in 1997 was em- metaheuristic search
algorithms
ployed. (MSAs),
DE is has become one
a population-based of thethat,
method mostamongpopularthe ones for solving
recently presented optimization
metaheuris-
problems
tic [34]. Presenting
search algorithms better
(MSAs), hasglobal
become convergence
one of the and mostrobustness,
popular ones simpler and more
for solving opti-
straightforward
mization problemsimplementation,
[34]. Presentingasbetterwell global
as having good performance
convergence in solving
and robustness, multi-
simpler and
objective optimization problems, are the main factors that make
more straightforward implementation, as well as having good performance in solving this algorithm superior to
other MSAs [35].optimization
multi-objective In this algorithm, the initial
problems, population
are the main factors(blades)
thatismake
generated completely
this algorithm su-
randomly, based on the upper and lower limits of the design variables. Then, using
perior to other MSAs [35]. In this algorithm, the initial population (blades) is generated
mutation and crossover operators, children (new blades) are generated. Subsequently, with
completely randomly, based on the upper and lower limits of the design variables. Then,
the help of the selection operator, each child is compared with the parent which generated it.
using mutation and crossover operators, children (new blades) are generated. Subse-
If the child has a better objective evaluation, it replaces the parent, and the next generation
quently, with the help of the selection operator, each child is compared with the parent
is generated. This process is repeated in the same way until the termination criteria are met.
which
In the generated it. Ifthe
present study, thetermination
child has a better objective
criterion was the evaluation,
number ofitgenerations.
replaces the Although
parent, and
the
raising this number (500 generations in this study) increases the computation the
next generation is generated. This process is repeated in the same way until termi-
time, it
nation
guarantees to reach the global solution. It is worth mentioning that to ensure that the of
criteria are met. In the present study, the termination criterion was the number
generations.
obtained answerAlthough raisingone,
is a global thisthe
number
values(500of thegenerations
objective in this study)
function were increases
monitored. the
Since its values remained unchanged for 100 consecutive generations, it was made sure
that the answer is global and the solution can be terminated. Employing the weighted
coefficient method [36], to maximize Cp and minimize Ts , the following objective function
was considered [20]:
 
CP min( Ts )
Maximize : n + (1 − n ) (1)
max(CP ) Ts

using which, the objective function tries to simultaneously maximize the Cp (the first goal)
and minimize the Ts (the second goal). The weighting factor, n (0 < n ≤ 1) determines
the influence of these two goals on the objective function. Values of max(CP ) and min( Ts )
represent the blades with the highest Cp and also the fastest Ts , respectively, which are
 𝐶 min(𝑇 )
𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒: 𝑛 + (1 − 𝑛) (1)
max(𝐶 ) 𝑇
using which, the objective function tries to simultaneously maximize the Cp (the first goal)
and minimize the Ts (the second goal). The weighting factor, n (0 𝑛 1) determines the
Machines 2022, 10, 687 5 of 22
influence of these two goals on the objective function. Values of max(𝐶 ) and min(𝑇 )
represent the blades with the highest Cp and also the fastest Ts, respectively, which are
updated in each generation. In other words, in each generation, the blades are compared
updated
and those in with
each generation.
the best Cp In other
and words, become
Ts values in each generation,
candidates theandblades
form are compared
max(𝐶 ) and
min(𝑇 ) for that generation. When the solution proceeds to the next generation, (aCnew
and those with the best C p and T s values become candidates and form max P ) andset
min ( Ts ) for) that
of max(𝐶 andgeneration.
min(𝑇 ) areWhen the solution
specified. proceedsthat
It is noteworthy to the nextvalues
these generation,
remainaconstant
new set
of
after CP ) and minlarge
maxa(sufficiently ( Ts ) are specified.
number It is noteworthy
of generations. This that
way,these
it is values remain
made sure thatconstant
the DE
after a sufficiently large number of
algorithm approaches the optimal solution. generations. This way, it is made sure that the DE
algorithm approaches the optimal solution.
3.2. Calculating Design Goals
3.2. Calculating Design Goals
To calculate the 𝐶 , the first objective of the design, the blade element momentum
To calculate the C p , the first objective of the design, the blade element momentum
(BEM) technique was
(BEM) technique was employed.
employed. This
This theory,
theory, which
which is is the
the most
most widely
widely used
used engineering
engineering
model for simulating wind turbine blades and propellers
model for simulating wind turbine blades and propellers [37], was proposed in[37], was proposed in 1935
1935 byby
Glauret [38]. BEM is the combination of the blade element method
Glauret [38]. BEM is the combination of the blade element method and momentum theory and momentum theory
in which
in which thethe blade
blade isis divided
divided into
into several
several independent
independent elements,
elements, each
each element
element having
having aa
specific twist
specific twist angle
angle and
and chord
chord length.
length. Figure
Figure 22 shows
shows the the velocities
velocities on on aa blade
blade element,
element,
where 𝑈 is the wind velocity, and 𝑈 is the total velocity, which is obtained
where U is the wind velocity, and UT is𝑇the total velocity, which is obtained according to the according
to theof
value value of the velocity
the velocity at thesection
at the blade blade section 1 ) and
(U1 ) and(𝑈also thealso the angular
angular velocity velocity of the
of the turbine
turbine
(ω). The(𝜔). Theshown
angles angles shown in this are
in this figure figure
theare the of
angle angle
attackof attack (α), angle
(α), twist twist angle (θ),flow
(θ), and and
flow angle
angle (𝜙). Based
(φ). Based on theon the figure,
figure, φ can be𝜙 can be calculated
calculated from the from the following
following equation:
equation:
 (1 − 𝑎)𝑈
𝜙 tan
= 𝑡𝑎𝑛
−1 ( 1 − a )U (2)
(1(1++a0 )𝑎rω
)𝑟𝜔
φ= (2)

where a and 𝑎 represent the axial and rotational induction factors, respectively, whose
values aare
where and a0 represent
calculated in anthe axial and
iterative rotational
process [19]. induction factors, respectively, whose
values are calculated in an iterative process [19].

Figure 2. Velocities on an element of the blade.

Figure 3 shows the axial (dFx ) and tangential (dFy ) forces on a blade element. Using
the continuity equation along with the momentum equations for each element, the local
values of axial force and aerodynamic torque (dQ) are calculated from:

1
dFx = ρUT 2 cdr (Cl cosφ + Cd sinφ) (3)
2
1
dQ = r dFy = ρUT 2 crdr (Cl sinφ − Cd cosφ) (4)
2
In the above equations, ρ is the air density, r is the radial distance from the rotor hub,
c is the chord length, and Cl and Cd are the lift and drag coefficients for the blade element,
 1
𝑑𝐹 = 𝜌𝑈 𝑐𝑑𝑟 (𝐶 𝑐𝑜𝑠 𝜙 + 𝐶 𝑠𝑖𝑛 𝜙) (3)
2
1
𝑑𝑄 = 𝑟 𝑑𝐹 = 𝜌𝑈 𝑐𝑟𝑑𝑟 (𝐶 𝑠𝑖𝑛 𝜙 −𝐶 𝑐𝑜𝑠 𝜙) (4)
2
Machines 2022, 10, 687 6 of 22
In the above equations, 𝜌 is the air density, 𝑟 is the radial distance from the rotor
hub, 𝑐 is the chord length, and 𝐶𝑙 and 𝐶𝑑 are the lift and drag coefficients for the blade
element, respectively. By summing the torque generated by each element, the total aero-
respectively. By summing the torque generated by each element, the total aerodynamic
dynamic torque Q is calculated.
torque Q is calculated.

Figure
Figure 3.
3. Axial
Axial and
and tangential
tangential forces
forces on
on aa blade
blade element.
element.

Finally, the C𝐶p𝑝 ofofthe

theturbine
turbineisiscalculated
calculatedfrom
from[39]:
[39]:
𝑄𝜔
Qω
C p𝐶== 0.5𝜌𝑆𝑈3 (5)
(5)
0.5ρSU
in which, S is the swept area of the blades.
in which, S is the swept area of the blades.
To take the
To take the tip
tip losses
losses into
into account,
account, Prandtl’s
Prandtl’s model 𝐹 ) is
model ((F) is employed.
employed. The
The purpose
purpose of
of
this work is to determine the effect of blade tip losses along the blade. This parameter,
this work is to determine the effect of blade tip losses along the blade. This parameter, which
which is a function
is a function of the of the number
number of blades
of blades (𝑁),angle
(N), flow flow angle
(φ), and(ϕ),also
andthe
also the radial
radial coor-
coordinate
dinate along
along the the(r),
blade blade (𝑟), is defined
is defined as follows
as follows [40]: [40]:
2 𝑐𝑜𝑠 (𝑒  )
𝐹 =2cos−1 e− f (6)
𝜋
F= (6)
𝑁(𝑅π − 𝑟)
𝑓= (7)
N (2𝑟
R 𝑠𝑖𝑛
− r )𝜙
f = (7)
By using the BEM theory and using analytical 2rsinφ correlations for aerodynamic coeffi-
cientsBy
atusing
high angles
the BEMof theory
attack,and
it is using
possible to determine
analytical the 𝑇𝑠forfrom
correlations the stationary
aerodynamic state.
coefficients
Wood [41] showed that at high α values, from an aerodynamic viewpoint, airfoils
at high angles of attack, it is possible to determine the Ts from the stationary state. Wood [41] behave
like a two-dimensional
showed that at high α flat plate,from
values, and an
to calculate
aerodynamictheir viewpoint,
lift and drag coefficients,
airfoils behavethe fol-a
like
lowing equations can
two-dimensional be used
flat plate, with
and acceptable
to calculate accuracy
their and
lift and regardless
drag of the
coefficients, airfoil
the type:
following
equations can be used with acceptable accuracy and regardless of the airfoil type:

Cl = sin2α (8)

Cd = 2 sin2 α (9)
By implementing Equations (8) and (9) in Equation (4) and normalizing c and r with
the blade tip radius (R), and also all speeds (rω, UT , and U1 ) with the wind velocity (U)
the startup torque QS is calculated as follows:

Z1
1/2
QS = NρU 2 R3 (1 + λr 2 ) cr sinθ (cosθ − λr sinθ )dr (10)
rh
Machines 2022, 10, 687 7 of 22

where, rh is the hub radius, and λr = rω/U represents the local tip speed ratio. From the
assumption that no power is generated throughout the startup time and the aerodynamic
torque is merely used to rotate the blades from the stationary state, the variation in the tip
speed ratio can be computed using the following ordinary differential equation [20]:

dλ R ( Q s − Qr )
= (11)
dt JU

in which, λ = Rω/U is the tip speed ratio, Qr shows the generator resistive torque, and
J indicates the total rotational inertia, which is the sum of the blade inertia (Jb ) and the
generator inertia (JG ). By considering a specific value for the tip speed ratio (λ), this
equation is solved and the time to reach λ from the stationary state (λ = 0) to the specific
value (which is considered in advance) is calculated. The value of λ = 1 is considered the
ending point of the startup process and the above equation is solved using this method. It
is necessary to explain that the parameters such as R, Qr , and also U, are constant values in
the present study.
By normalizing c and r by R, the moment of inertia of the blade is determined from [19]:
Z Z 
1
Z
2
Jb = Nρb AR5 cr
( ) dr + c 4
cos 2
θdr + A 2
c 4
sin 2
θ dr (12)
12

in which, ρb is the density of the blade, and A is the surface area of the airfoil used in the
blade, assuming that the unit is chord length.
It should be noted that the startup performance of a 5 kW wind turbine measured
by Sessrego and Wood [42] has confirmed the accuracy of Equation (11). This reference
indicates that the startup stage involves two important steps: (1) the long idling step, in
which the blade has a slow rotational acceleration and continues until the angles of attack
along the blade fall to less than 20◦ , and (2) the rapid acceleration step, in which the blade
reaches an operational angular velocity. Therefore, to minimize the startup time, the initial
phase should be minimized. Integrating Equation (11) up to λ = 1 properly captures this
step. To calculate Equation (11), the Adams–Moulton method was used and the trapezoidal
integration method was employed to calculate the blade inertia of Equation (12).

3.3. Adjusting the Input Parameters

Machines 2022, 10, 687 The flowchart of the used optimization algorithm is illustrated in Figure
8 of 4.
23 A code
that was developed in MATLAB software was employed for the optimization process and
calculation of the design objectives.

Start

Calculating the objective Generating the initial

function: Eq. (1) population

DE cycle Mutation

Crossover

Selection

Calculating the Stop? Yes The optimal blade

objective function geometry

No End

Figure Themulti-objective
Figure 4.4.The multi-objective optimization
optimization process.
process.

Based on the values suggested in the literature [19,43], the blade was divided into 15
elements. Table 1 shows the minimum and maximum values of the design variables ac-
cording to the limitations and construction problems for the twist angle/chord length dis-
tribution. Additionally, Table 2 lists the input parameters considered in the DE algorithm.
Machines 2022, 10, 687 8 of 22

Based on the values suggested in the literature [19,43], the blade was divided into
15 elements. Table 1 shows the minimum and maximum values of the design variables
according to the limitations and construction problems for the twist angle/chord length
distribution. Additionally, Table 2 lists the input parameters considered in the DE algorithm.

Table 1. The specified limits for the design parameters [20].

Parameter Minimum Maximum

Twist, θ(◦ ) −5 25
Chord, c/R 0.01 0.2

Table 2. The considered input parameters in the DE algorithm.

Parameters Values/Settings
Population 2000
Number of generations 500
Mutation strategy DE/rand/1
Mutation weighting factor 0.8
Crossover operator Uniform
Crossover constant 0.1

The selected small horizontal-axis wind turbine was introduced and numerically
evaluated by Pourrajabian et al. [44]. The technical specifications of this turbine, which
include the output power (P), number of blades (N), blade radius (R), hub radius (rh ),
angular velocity (ω), rated tip speed ratio (λrated ), rated wind speed (Urated ), generator
inertia (JG ), generator resistive torque (Qr ), blade density (ρb ), startup wind speed (Us ),
and the type of airfoil used in the blade are given in Table 3.

Table 3. The technical specifications of the considered base turbine [44].

Parameters Values and Units Parameters Values and Units

P 1000 (w) Urated 10 (m/s)
0.01 kgm2


N 3 JG
R 1.21 (m) Qr 0.5 (Nm)
rh 0.125 (m) ρb 550 (kg/m3 )
ω 450 (rpm) Us 5 (m/s)
λrated 5.71 Airfoil type SG6043

3.4. Validation of the BEM Code

To evaluate the accuracy of the developed BEM code, the results of the present
study have been validated with the experimental results from the wind tunnel test of
a small wind turbine with a radius of 1.5 m. This two-bladed turbine was examined by
Anderson et al. [45], and then further investigated by Wood [19] and Spera [37]. The ge-
ometry of this turbine, which uses NACA4412 airfoil as the blade cross-section, is shown
in Figure 5.
The aerodynamic coefficients of the NACA4412 airfoil, including the lift and drag
coefficients at various angles of attack and different Reynolds numbers, are also depicted
in Figure 6.
As mentioned in Section 2, all aerodynamic coefficients of the airfoils used in the
present study are coded, and, based on the Re number and the angle of attack of each blade
element, the corresponding values of Cl and Cd are computed and called during the BEM
calculations and optimization steps.
Figure 7 shows the changes of Cp at different λ values. It can be observed that
the results obtained from the developed numerical code are well-compatible with the
experimental results with a maximum error of 6.47%. It is noteworthy that the mean
 𝜔 450 (rpm) 𝑈 5 (m⁄s)
𝜆 25 5.71 Airfoil type SG6043
Twist (deg) Chord (cm)
20

Twist (deg), Chord (cm)

Machines 2022, 10, 687
3.4. Validation of the BEM Code
To evaluate the accuracy of the developed BEM code, the results of the present study
have 15been validated with the experimental results from the wind tunnel test of a small 9 of 22
wind turbine with a radius of 1.5 m. This two-bladed turbine was examined by Anderson
et al. 10
[45], and then further investigated by Wood [19] and Spera [37]. The geometry of
this turbine, which uses NACA4412 airfoil as the blade cross-section, is shown in Figure
squared error (MSE) of the obtained numerical values was found to be 0.000228562. More
5.
detailed5 data for this comparison are given in Table A1 in Appendix B.

25 0
Twist (deg) Chord (cm)
20

Twist (deg), Chord (cm)

−5
15 0 20 40 60 80 100 120 140 160
Radius (cm)
10

Figure
5 5. The twist angle and chord length distributions of the turbine in the work of And
al. [45].
0

The20 aerodynamic
0 40 60
−5
80
coefficients of the NACA4412 airfoil, including the lift an
100 120 140 160
coefficients at various angles
Radius (cm) of attack and different Reynolds numbers, are also d
in Figure
Figure
Figure5.5.The
6.twistangle
Thetwist angleand
andchord
chordlength
lengthdistributions ofof
distributions thethe
turbine in in
turbine thethe
work of of
work Anderson et et al. [45].
Anderson
al. [45].
1.4
The aerodynamic coefficients of the NACA4412 airfoil, including the lift and drag
1.2
coefficients
1.0
at various angles of attack and different Reynolds numbers, are also depicted
in Figure 6.
0.8

0.6
Cl

1.4

1.20.4

1.0
0.2 Re = 42,000 Re = 83,000
0.8
Re = 160,000 Re = 330,000
0.6
0.0 Re = 640,000
Cl

−0.2
0.4

0.2
−4 −2 0 2 4 6 8 10 12 14
Re = 42,000 Re = 83,000

0.0
Re = 160,000 α (deg)
Re = 330,000
Re = 640,000

−0.2
0.07−4 −2 0 2 4 6 8 10 12 14
α (deg)
Machines 2022, 10, 687 0.06 10 of 23
0.07
Re = 42,000 Re = 83,000
0.05
0.06 Re = 160,000 Re = 330,000
Re = 640,000
Re = 42,000 Re = 83,000
0.05
Cd

0.04 Re = 160,000 Re = 330,000

As mentioned
Re = 640,000 in Section 2, all aerodynamic coefficients of the airfoils used in the
Cd

0.04
present
0.03 study are coded, and, based on the Re number and the angle of attack of each
0.03
blade element, the corresponding values of Cl and Cd are computed and called during the
0.02
0.02
BEM calculations and optimization steps.
0.01 Figure 7 shows the changes of Cp at different λ values. It can be observed that the
0.01
results−4obtained
−2 0 from
2 4 the
6 developed
8 10 12 14numerical code are well-compatible with the experi-
−4 −2 0 2 4 6 8 10 12 14
mental results withα (deg) a maximum error of 6.47%. It is noteworthy that the mean squared
α (deg)
error (MSE)
Figure 6. Theofchanges
the obtainedof the numerical values
Cl (top) and Cd was foundvalues
(bottom) to be 0.000228562. More detailed
of the NACA4412 airfoil [19,26].
data for this comparison are given in Table A1 in Appendix B.
Figure
Figure 6.6.The
The changes
changes ofCthe
of the Cl and
l (top) (top)
Cd and Cd (bottom)
(bottom) values
values of the of theairfoil
NACA4412 NACA4412
[19,26]. airfoil [19,2

0.5

0.4

0.3
Cp

Present study
0.2
Anderson et al.

0.1
5 6 7 8 9 10 11 12 13 14
λ
Figure 7.7.
Figure Changes in the
Changes inCthe
p at different λ values for the present study and the research of Anderson
Cp at different λ values for the present study and the research of Anderson et al. [45].
et al. [45].

3.5. Validation of the Optimization Code

The ideal equations proposed by Burton et al. [46] were employed to evaluate the
performance of the optimization code. By ignoring the tip losses and drag, the following
correlations determine the twist angle and the chord length along the blade to achieve the
 Machines 2022, 10, 687 10 of 22

3.5. Validation of the Optimization Code

The ideal equations proposed by Burton et al. [46] were employed to evaluate the
performance of the optimization code. By ignoring the tip losses and drag, the following
correlations determine the twist angle and the chord length along the blade to achieve the
maximum power coefficient:
2
tan φ = (13)
3λr + λ2r
16π
cCl = r h i2 (14)
4/9 + λr + 2
9Nλ (9λr )
Machines 2022, 10, 687 11 of 23
Machines 2022, 10, 687
Considering the BW-3 airfoil for the 1 kW turbine blade of the present study, Figure 8
11 of 23
shows the distribution of the twist angle (θ) and chord length (c) obtained from the opti-
mization algorithm (n = 1) and the ideal Equations of (13) and (14).
25
20 25 n=1 Eq. (13)
Twist (deg)

15 20 n=1 Eq. (13)

Twist (deg)

10 15
5 10
0 5

0.10 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.1 0.2 0.3 0.4 Radius,
0.5 r/R0.6 0.7 0.8 0.9 1.0

0.25 Radius, r/R

0.25 n=1 Eq. (14)
0.20
Chord, c/R

0.20 n=1 Eq. (14)

Chord, c/R

0.15
0.15
0.10
0.10
0.05
0.05
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.1 0.2 0.3 0.4 0.5
Radius, r/R0.6 0.7 0.8 0.9 1.0
Radius, r/R
Figure 8. The twist angle (top) and chord length (bottom) distributions of the blade fitted with the
Figure
Figure
BW-3 8.8.The
airfoil. Thetwist
twistangle
angle(top) and
(top) chord
and length
chord (bottom)
length distributions
(bottom) of the
distributions blade
of the fitted
blade withwith
fitted the the
BW-3
BW-3airfoil.
airfoil.
As can be observed, there is a good agreement between the distributions along the
blade. As Ascan canbe
Therefore, beobserved,
itobserved,
is ensured there
thereis isathe
that agood
good
finalagreement
agreement
answer ofbetween
between
the thethe
distributions
optimization code isalong
distributions the the
along
optimal.
blade. Therefore, it is ensured that the final answer of the optimization code is optimal.
Ofblade.
course,Therefore,
there are itsome is ensured that thein
discrepancies final
theanswer of the optimization
distributions considered bycode is optimal. Of
the optimiza-
Of course, there are some discrepanciesininthe thedistributions
distributions considered
considered by the optimiza-
tion algorithm in the root section, the reason for which is revealed in detail inoptimization
course, there are some discrepancies by the the next
tion algorithm in root
the root section, the reason for which is revealed in detailthe in next
the next
algorithm
section. in the
It should be noted section, the
that, to use reason for which
the ideal Equations (13) andin(detail
is revealed 14), it inis necessary section.
to
section.
It should It should be noted that, to use the ideal Equations ( 13 ) and ( 14) , it is necessary to
specify the αbe andnoted
Cl valuesthat, to use airfoil
of the the ideal Equations
at the maximum (13) and (14), ratio,
lift-to-drag it is necessary to specify
which, accord-
specify the
α and C α and C l values of the airfoil at the maximum lift-to-drag ratio, which, accord-
ingtheto Figure values
9,l are 5.39°ofand the1.034,
airfoilrespectively,
at the maximum for thelift-to-drag
BW-3 airfoil ratio,
at the which, according to
Re of 200,000.
ing to Figure
Figure 9, are 9, are◦ 5.39°
5.39 and 1.034,
and 1.034, respectively,
respectively, forBW-3
for the the BW-3 airfoil
airfoil at the
at the Re of Re200,000.
of 200,000.

80
80
70
70
60
60
Lift-to-drag ratio
Lift-to-drag ratio

50
50
40
40
30
30
Re = 60,000
20 Re = 60,000
20 Re = 100,000
Re = 100,000
Re = 200,000
10 Re = 200,000
10 Re = 300,000
Re = 300,000
Re = 400,000
0 0 Re = 400,000

−4−4 −2−2 00 22 44 6 6 88 10
10 12
12
αα(deg)
(deg)
1.61.6
Figure 9. Cont.
1.41.4
1.21.2
1.01.0
0.80.8
Cl

ReRe
==60,000
Cl

0.60.6 60,000
 40

Lift-to-dra
30
Re = 60,000
20 Re = 100,000
Re = 200,000
10 Re = 300,000
Machines 2022, 10, 687 Re = 400,000 11 of 22
0

−4 −2 0 2 4 6 8 10 12
α (deg)
1.6
1.4
1.2
1.0
0.8

Cl
Re = 60,000
0.6
Re = 100,000
0.4 Re = 200,000
Re = 300,000
0.2 Re = 400,000

0.0
−0.2
−4 −2 0 2 4 6 8 10 12
α (deg)

Figure 9. Lift-to-drag ratio (top) and Cl (bottom) for the BW-3 airfoil [31].
Figure 9. Lift-to-drag ratio (top) and Cl (bottom) for the BW-3 airfoil [31].
4. Discussion of Results
Similar to the specified ranges for the design variables (mentioned in Table 1) and also
the value of Tables 2 and 3, n is an input parameter for running the optimization. In multi-
objective optimization problems, there is no single answer. In this study, the answers are
determined by changing the value of n in the objective function (Equation (1)), which leads
to the Pareto front. Here, Pareto front members are the blades whose at least one component
of the objective function has a larger Cp or 1/Ts value than the other blades. Generally,
larger values are considered for n, because the output power is a more important goal
than the startup time [19]. Imposing low weighting factors (n) causes poor aerodynamic
performance of the blade and makes it practically unusable. Selecting the final blade from
the Pareto front depends on the wind potential. Although n = 1 is used in windy areas to
achieve the maximum power coefficient, n = 0.8 is recommended in areas with low wind
speeds where the startup performance of turbines is more important [20,21].
By considering the input parameters mentioned in Section 3.3, as well as the expla-
nations provided above, the optimization process was used to evaluate the performance
of the selected airfoils, the results, which include Cp and Ts , are summarized in Table 4. It
is necessary to explain that the details of the optimization results were also obtained for
n = 0.6, and the tabulated results can be found and compared in Table A2 of Appendix C.

Table 4. Optimal values of Cp and Ts for the selected airfoils.

Windy Areas (n = 1) Low Wind Areas (n = 0.8)

Airfoil
Cp Ts (s) Cp Ts (s)
BW-3 0.496 2.78 0.486 1.82
E387 0.502 4.87 0.485 3.10
FX 63-137 0.499 8.44 0.489 3.26
S822 0.495 13.25 0.481 7.02
S834 0.498 11.54 0.483 6.87
SD7062 0.497 6.15 0.488 4.49
SG6040 0.495 15.67 0.480 6.02
SG6043 0.506 5.72 0.496 2.89
SG6051 0.504 9.8 0.490 5.13
USNPS4 0.503 5.36 0.493 4.22

4.1. Investigating the Performance of Airfoils in Windy Areas

In areas with high wind speeds (n = 1) where Ts is removed from the objective function,
the SG6043 airfoil yields the highest Cp while the S822 and SG6040 airfoils present the
lowest Cp (see Table 4). The lift-to-drag ratio of airfoils is an important parameter in the
efficiency of wind turbine blades. Figure 10 shows the maximum lift-to-drag ratio of the
selected airfoils over a wide range of Re numbers.
Machines 2022, 10, 687 13 of 23
Machines 2022,10,
Machines2022, 10,687
687 13 of 22
12 of 23

140
140 BW−3 E387
BW−3 E387
FX 63−137 S822

Maximum lift−to−drag ratio

FX 63−137 S822

Maximum lift−to−drag ratio

120 S834 SD7062
120 S834 SD7062
SG6040 SG6043
SG6040
SG6051 SG6043
USNPS4
100 SG6051 USNPS4
100

80
80
60
60
40
40
100,000 200,000 300,000 400,000 500,000
100,000 200,000 Re 300,000 400,000 500,000
Re
Figure 10. The maximum lift-to-drag ratio for the selected airfoils [30,31,47].
Figure10.
Figure 10.The
Themaximum
maximumlift-to-drag
lift-to-dragratio
ratiofor
forthe
theselected
selectedairfoils
airfoils[30,31,47].
[30,31,47].
It can be observed that, compared to the other airfoils, this value is higher for the
SG6043 ItItcan
canbebeobserved
airfoil. Itobserved
should bethat,
that, compared
noted that the to
compared tothe
type ofother
the other airfoils,
airfoilairfoils, this
completelythisvalue
value isishigher
influences higher for
forthe
the distri-the
SG6043
SG6043 airfoil.
airfoil. ItItshould
should be noted
be noted that the
that thetype
typeof airfoil
of completely
airfoil completely
bution of twist and chord along the blade. Indeed, the optimization algorithm considers influences
influences the distribu-
the distri-
tion
the of twist
bution
best andand
of twist
geometry chord
for thealong
chord
blade the
along blade.
theon
based Indeed,
blade.
the the optimization
Indeed,
aerodynamic the coefficients
optimizationalgorithm
and considers
algorithm
geometric char-the
considers
best geometry
the best geometry
acteristics for the blade based
for theFigure
of each airfoil. on
blade based the
11 shows aerodynamic
on thetheaerodynamic coefficients
distribution of and
coefficients
twist andand geometric
chordgeometriccharac-
of the char-
se-
teristics of
acteristics
lected each
offor
airfoils eachairfoil. Figure
n =airfoil.
1. 11 shows the distribution of twist and chord
Figure 11 shows the distribution of twist and chord of the se-of the selected
airfoils for n = 1.
lected airfoils for n = 1.

25
BW−3 E387
25 FXBW−3
63−137 S822
20 E387
S834
FX 63−137 SD7062
S822
20
Twist (deg)

15 SG6040
S834 SG6043
SD7062
SG6051 USNPS4
Twist (deg)

1015 SG6040 SG6043

SG6051 USNPS4
510

05

−5 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
−5
0.1 0.2 0.3 0.4 Radius,
0.5 r/R0.6 0.7 0.8 0.9 1.0
Radius, r/R
0.21
BW−3 E387
0.21
0.18 FX 63−137 S822
BW−3
S834 E387
SD7062
0.18 FX 63−137
SG6040 S822
SG6043
Chord, c/R

0.15
S834
SG6051 SD7062
USNPS4
SG6040 SG6043
Chord, c/R

0.15
0.12
SG6051 USNPS4
0.12
0.09

0.09
0.06

0.06
0.03
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.03
Radius, r/R
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure Radius, r/R

Figure11.
11.The
Thedistributions
distributionsofoftwist
twist(top)
(top)and
andchord
chord(bottom) ofof
(bottom) the selected
the airfoils
selected (n (n
airfoils = 1).
= 1).
Figure 11. The distributions of twist (top) and chord (bottom) of the selected airfoils (n = 1).
Byfocusing
By focusingon onthis
thisfigure,
figure,ititcan
canbebeseen
seenthat
thatthe
thevalues
values ofof the
the chord
chord length
length forfor
thethe
blade that was fitted with the FX 63-137 airfoil are smaller than the values
blade that was fitted with the FX 63-137 airfoil are smaller than the values considered for considered
By focusing on this figure, it can be seen that the values of the chord length for the
for other
the the other selected
selected airfoils.
airfoils. DueDue to this,
to this, despite
despite thethe
high lift-lift-to-drag
high to-drag ratio
ratio
of of this
this airfoil
airfoil
blade that was fitted with the FX 63-137 airfoil are smaller than the values considered for
(Figure 10), the Cp is not so high. Because the smaller chord length leads to a lower Re and,
the other selected airfoils. Due to this, despite the high lift-to-drag ratio of this airfoil
Machines 2022, 10, 687 14 of 23
Machines 2022, 10, 687 14 of 23

Machines 2022, 10, 687 13 of 22

(Figure 10), the Cp is not so high. Because the smaller chord length leads to a lower Re and,
hence, reduces the lift-to-drag ratio. The opposite is true for the USNPS4
(Figure 10), the Cp is not so high. Because the smaller chord length leads to a lower Re and, airfoil. According
to Figure
hence, 10, it can
reduces thebe observedratio.
lift-to-drag that this
The airfoil
oppositedoes not have
is true a high
for the USNPS4 lift-to-drag ratio, but
airfoil. According
hence,
the reduces
larger
to Figure 10, itthe
chord can lift-to-drag
length valuesratio.
be observed theThe
of that thisopposite
blade is true
fitteddoes
airfoil with for
this
not the aUSNPS4
airfoil
have increase
high airfoil.
the ReAccording
lift-to-drag and, con-
ratio, but
to Figure
sequently, 10, it
raise can
the be
C . observed that this airfoil does not have a
the larger chord length values of the blade fitted with this airfoil increase the Re and, con-
p high lift-to-drag ratio,
but the larger
It should
sequently, chord
raisebe the
noted length
Cp. thatvalues of the blade
for all selected fitted
airfoils, with
the twistthis airfoil
angles increase
and chord the Re and,
lengths are
consequently,
not smooth at raise
the the
root C p .
elements (as seen in Figure 8). This is due to
It should be noted that for all selected airfoils, the twist angles and chord lengths are the low contribution
of
not It
these should
smooth atbethe
elements notedthethat
inroot totalfor all(as
selected
aerodynamic
elements seen in airfoils,
torque
Figure the
8). twist
and, hence,
This isangles
the
dueCto pand
ofthechord
the lengths
turbine,
low are
which
contribution
not smooth
makes
of these the at the root
optimization
elements elements
algorithm
in the total (as seen in Figure
focus ontorque
aerodynamic 8).
the evolutionThis
and, hence,is due
of the to the
the design low
Cp of the contribution
variables
turbine, in the
which
of these
middle elements
and tip in the
elements total
and aerodynamic
creates a torque
smoother and, hence,
distribution
makes the optimization algorithm focus on the evolution of the design variables in the the
in C
these
p of the turbine,
elements. To which
take a
makes
closer the
look optimization
at this point, algorithm
Figure 12 focus on
illustrates the
the evolution
torque of the
generation
middle and tip elements and creates a smoother distribution in these elements. To take a design variables
distribution in
along the
the
middle
blade and tip
closerthroughout
look elements
at this the power
point, and creates
generation
Figure a smoother
for thethe
12 illustrates distribution
SG6043
torqueand in these
S822
generation elements. To
airfoils.
distribution takethe
along a
closer look at this point, Figure 12 illustrates the torque generation distribution along the
blade throughout the power generation for the SG6043 and S822 airfoils.
blade throughout the power generation for the SG6043 and S822 airfoils.
0.010
0.010
torque

0.008
torque

0.008
0.006
entent

SG6043
0.006 S822
SG6043
elem

0.004
S822
elem

0.004
B lade

0.002
B lade

0.002
0.000
0.0000.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.1 0.2 0.3 0.4 Radius,
0.5 r/R
0.6 0.7 0.8 0.9 1.0
Radius, r/R
Figure 12. The torque generated by the blade elements during power generation for n = 1.
Figure 12.The
Figure12. Thetorque
torquegenerated
generatedby
bythe
theblade
bladeelements
elementsduring
duringpower
powergeneration forn n==1.1.
generationfor
As can be seen, regardless of the airfoil type, a significant portion of the aerodynamic
torqueAs
As iscan
can be
beseen,
generatedseen,byregardless
the middle
regardless ofofthe
andairfoil
the type,
type,aasignificant
tip elements
airfoil of the blade,
significant portion ofofthe
and the
portion theaerodynamic
share of the root
aerodynamic
torque
elements is generated
is considerably by the middle
smaller. It and
should tip elements
be noted of
that
torque is generated by the middle and tip elements of the blade, and the share of the
theblade,
slight and the
decrease share
in the ofaero-
the the
root
root elements
dynamic torque is considerably
at the tip of smaller.
the blade isIt should
because be
of noted
consideringthat
elements is considerably smaller. It should be noted that the slight decrease in the aero- the
the slight
blade decrease
tip losses in
in the
the
aerodynamic
BEM torque
calculations.
dynamic at the tip of the blade is because of considering the
torque at the tip of the blade is because of considering the blade tip losses in the blade tip losses in
the BEM calculations.
BEM calculations.
4.2. Investigating the Performance of Airfoils in Areas with Low Wind Speed
4.2. Investigating the Performance of Airfoils in Areas with Low Wind Speed
4.2. Investigating
In areas withthe low Performance
wind speeds of Airfoils
(n = 0.8),in small
Areas with
windLow Wind need
turbines Speedto react quickly to
In areas with low wind speeds (n = 0.8), small wind turbines need to react quickly
the wind and start generating power (n in the shortest possible time. Inneed
this regard, Worasin-
to the wind and start generating power in the shortest possible time.to In
In areas with low wind speeds = 0.8), small wind turbines react quickly
this regard,to
chai et al.
the wind and [18] have shown that reducing startup time increases energy capture.
Worasinchai et start generating
al. [18] have shown power thatinreducing
the shortest possible
startup timetime. In this
increases regard,
energy Worasin-
capture.
According
chaiAccording
et al. [18] haveto Table
shown 4, the
that startup
reducing performance
startup timeof increases
the BW-3energy airfoil capture.
is better than the
to Table 4, the startup performance of the BW-3 airfoil is better than the
other According
selected airfoils.
to Table Thus, thestartup
4, the
the turbineperformance
that uses thisofairfoil
thehas has the
BW-3 shortest Ts. The
thanrea-
other selected airfoils. Thus, turbine that uses this airfoil the airfoil
shortest is Tbetter
s . The reason
the
son is the
other low inertia
selected airfoils.ofThus,
the blades
the fitted with
turbine that this
uses airfoil
this (see Figure
airfoil has the 13).
shortest Ts. The rea-
is the low inertia of the blades fitted with this airfoil (see Figure 13).
son is the low inertia of the blades fitted with this airfoil (see Figure 13).
2.0

2.0
1.5
Jb (kgm 2
)

1.5
1.0
Jb (kgm 2
)

1.0
0.5

0.5
0.0
BW−3 E387 FX 63−137 S822 S834 SD7062 SG6040 SG6043 SG6051 USNPS4
0.0
BW−3 E387 FX 63−137 S822 AirfoilSD7062
S834 SG6040 SG6043 SG6051 USNPS4
Airfoil
Figure 13.
Figure 13. The moment of inertia
inertia (J
(𝐽b ) values
values of
of the
the optimized
optimized blades
blades (n
(n == 0.8).
Figure 13. The moment of inertia (𝐽 ) values of the optimized blades (n = 0.8).
The S822, S834, and SG6040 airfoils have the highest Ts and their application in areas
with low wind speeds is not recommended at all. The common aspect of these three airfoils
Machines 2022, 10, 687 15 of 23
Machines 2022, 10, 687 15 of 23

Machines 2022, 10, 687 The S822, S834, and SG6040 airfoils have the highest Ts and their application in14areas of 22
The S822, S834, and SG6040 airfoils have the highest Ts and their application in areas
with low wind speeds is not recommended at all. The common aspect of these three air-
with low wind speeds is not recommended at all. The common aspect of these three air-
foils is their high surface area compared to the other selected airfoils, which is an effective
foils is their high surface area compared to the other selected airfoils, which is an effective
factor
is theirin thesurface
high blade moment of inertia,
area compared aside
to the from
other the chord
selected length
airfoils, (Figure
which is an14).
effective factor
factor in the blade moment of inertia, aside from the chord length (Figure 14).
in the blade moment of inertia, aside from the chord length (Figure 14).
0.12
0.12
0.10
0.10
0.08
0.08

2 2)
(m(m)
0.06
0.06
AA 0.04
0.04
0.02
0.02
0.00
0.00 BW−3 E387 FX 63−137 S822 S834 SD7062 SG6040 SG6043 SG6051 USNPS4
BW−3 E387 FX 63−137 S822 S834 SD7062 SG6040 SG6043 SG6051 USNPS4
Airfoil
Airfoil
Figure 14. The surface area (A) of the selected airfoils.
Figure14.
Figure 14.The
Thesurface
surfacearea
area(A)
(A)of
ofthe
theselected
selectedairfoils.
airfoils.

But it is also important to note that a thinner airfoil does not always present a better
Butititisisalso
But alsoimportant
importantto tonote
notethat
thataathinner
thinnerairfoil
airfoildoesdoesnot notalways
alwayspresent
presentaabetter
better
performance at low wind speeds. For example, the E387 airfoil is thinner than the SG6043
performanceatatlow
performance lowwind
windspeeds.
speeds.For Forexample,
example,the theE387
E387airfoil
airfoilisisthinner
thinnerthanthanthe theSG6043
SG6043
airfoil and has a smaller surface area (see Figure 14), but it has a longer Ts (Table 4 for n =
airfoil and
airfoil and has a smaller
smaller surface
surfacearea area(see
(seeFigure
Figure14),
14),but butit it
hashasa longer
a longer Ts T(Table
s (Table4 for n=
4 for
0.8). The reason is the higher moment of inertia and the lower startup torque (the startup
n0.8). TheThe
= 0.8). reason is the
reason is higher
the highermoment momentof inertia and the
of inertia and lower startupstartup
the lower torque torque
(the startup
(the
torque is examined at the end of the current section) of the blades fitted with this airfoil
startup
torque torque is examined
is examined at the endat theof end of the current
the current section) section)
of the of the blades
blades fittedthis
fitted with withairfoil
this
compared
airfoil to the to
compared blades fitted with
the blades the SG6043 airfoil.airfoil.
compared to the blades fitted fitted
with thewith the SG6043
SG6043 airfoil.
Figure 15
Figure 15 shows the the optimal distribution
distribution of twist
twist and chord chord for the the selected airfoils
airfoils
Figure 15shows shows theoptimal
optimal distribution of of twistand and chordfor for theselected
selected airfoils
for n = 0.8. By comparing this figure with Figure 11, it can be observed that regardless of
fornn== 0.8.
for 0.8. ByBy comparing
comparing this this figure
figure with
with Figure
Figure 11,
11, ititcan
canbe beobserved
observedthat thatregardless
regardlessof of
the
the airfoil type, raising the twist and chord values in the root part of the blade decreases
theairfoil
airfoiltype,
type,raising
raising thethetwist
twistand
andchord
chordvalues
valuesin inthetheroot
rootpart
partofofthetheblade
bladedecreases
decreases
the TTs.. It
the should be noted that
that during the the startup process,
process, the angles angles of attack
attack along the the
the Tss. It
It should
should be be noted
noted that during
during the startup
startup process, the the angles of of attackalong
along the
blade are
blade are high, which
which decrease by by raising thethe twist angle,
angle, and this this increases the the startup
blade arehigh, high, whichdecrease
decrease byraising raising thetwist
twist angle, and and thisincreases
increases thestartup
startup
torque.
torque. However, since the blade root elements have a smaller share in power generation,
torque.However,
However,since sincethe
theblade
bladeroot
rootelements
elementshavehaveaa smaller
smaller share
share in in power
power generation,
generation,
the optimization
the optimization algorithm applies applies these increments
increments (twist and and chord) only only in this
this area so
so
the optimizationalgorithm
algorithm appliesthese these increments(twist (twist andchord) chord) onlyin in thisarea
area so
that the C does not decline much.
that theCCpp does
thatthe does not
not decline
decline much.
p
much.

25
25 BW−3 E87
BW−3 E87
20 FX63−137 S822
20 FX63−137 S822
S834 SD7062
S834 SD7062
(deg)

15 SG6040 SG6043
Twist(deg)

15 SG6040 SG6043
SG6051 USNPS4
10 SG6051 USNPS4
10
Twist

5
5
0
0
−5
−50.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Radius, r/R
Radius, r/R
0.21
0.21 BW−3 E387
BW−3 E387
0.18 FX 63−137 S822
0.18 FX 63−137 S822
S834 SD7062
S834 SD7062
c/R

0.15 SG6040 SG6043

Chord,c/R

0.15 SG6040 SG6043

SG6051 USNPS4
0.12 SG6051 USNPS4
Chord,

0.12
0.09
0.09
0.06
0.06
0.03
0.030.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Radius, r/R
Radius, r/R
Figure 15. The distributions of twist (top) and chord (bottom) of the selected airfoils (n = 0.8).
 Machines 2022, 10, 687 16 of 23

Machines 2022, 10, 687 15 of 22

Figure 15. The distributions of twist (top) and chord (bottom) of the selected airfoils (n = 0.8).

Figure 16 shows the variations of startup torque during the startup time of the se-
Figure 16 shows the variations of startup torque during the startup time of the selected
lected airfoils for n = 0.8. Regardless of the airfoil type, as the blade begins to rotate, the
airfoils for n = 0.8. Regardless of the airfoil type, as the blade begins to rotate, the startup
startup torque initially decreases, and then it starts to increase. When the blade is station-
torque initially decreases, and then it starts to increase. When the blade is stationary, the
ary, the drag force is perpendicular to the direction of rotation and does not affect the
drag force is perpendicular to the direction of rotation and does not affect the startup torque.
startup torque. As the blade begins to rotate, the drag force reduces the startup torque,
As the blade begins to rotate, the drag force reduces the startup torque, but then the startup
but then the startup torque starts to increase.
torque starts to increase.

0 1 2 3 4 5 6 7 8 9 10 11
1.92 USNPS4
1.90
1.89
1.83 SD7062
1.82
1.81
1.76 FX 63−137
1.74
1.73
1.74 SG6043
1.73
1.71

1.72 E387

1.70
𝐐𝐬 (Nm)

1.68
1.71 S822
1.69
1.68
1.71 SG6051
1.69
1.68
1.70
SG6040
1.68
1.66

1.68 S834
1.67
1.66
1.65 BW−3
1.64
1.63
0 1 2 3 4 5 6 7 8 9 10 11
𝐓𝐬 (s)

Figure 16. Variations of the startup torque for

Figure 16. for the
the selected
selected airfoils
airfoils(n
(n==0.8).
0.8).

Figure
Figure 1616 shows that the
shows that the blade
blade that
that uses
uses the
the USNPS4
USNPS4airfoil
airfoilproduces
producesthe thehighest
highest
startup
startup torque,
torque, but
but having
having aa high
high moment
moment of of inertia
inertia raises itsTTs.s .The
raisesits Thelowest
loweststartup
startuptorque
torque
is
is produced in the blade with the BW-3 airfoil, but, as mentioned, its low moment of inertia
produced in the blade with the BW-3 airfoil, but, as mentioned, its low moment of iner-
is
tiathe advantageous
is the advantageous aspect
aspect ofof
this
thisairfoil
airfoilininareas
areaswith
with low
low wind speed.ItItisisnoteworthy
wind speed. noteworthy
that
that the
the startup
startup torque is aa function
torque is function ofof two
two parameters,
parameters,including
includingthe thetwist
twistangle
angleand
andthethe
chord length (see Equation (10)). The difference observed in the values
chord length (see Equation (10)). The difference observed in the values of startup torque of startup torque
for the selected airfoils is due to the different twist angle and chord length values that the
optimization algorithm has considered for each airfoil. Finally, Figure 17 summarizes the
performance of the selected airfoils in terms of Cp and Ts . The arrangement of these airfoils
Machines 2022, 10, 687 17 of 23

for the selected airfoils is due to the different twist angle and chord length values that the
Machines 2022, 10, 687 optimization algorithm has considered for each airfoil. Finally, Figure 17 summarizes 16 the
of 22

performance of the selected airfoils in terms of Cp and Ts. The arrangement of these airfoils
is based on exhibiting the highest Cp in windy areas and having the shortest Ts in areas
with low wind
is based speeds. the highest Cp in windy areas and having the shortest Ts in areas
on exhibiting
with low wind speeds.

Performance of
the selected
airfoils

Highest Cp in Lowest Ts in areas

windy areas with low wind
(n = 1) speed (n = 0.8)

1. SG6043 1. BW−3
2. SG6051 2. SG6043
3. USNPS4 3. E387
4. E387 4. FX 63−137
5. FX 63−137 5. USNPS4
6. S834 6. SD7062
7. SD7062 7. SG6051
8. BW−3 8. SG6040
9. S822/SG6040 9. S834
10. S822

Figure17.
Figure Comparisonofofthe
17.Comparison theperformance
performanceof
ofthe
theselected
selectedairfoils.
airfoils.

5. Conclusions
5. Conclusions
In the present study, the effect of the airfoil type was investigated on the power
In the present study, the effect of the airfoil type was investigated on the power co-
coefficient, as well as the startup time of a 1 kW small wind turbine by developing a
efficient, as well as the startup time of a 1 kW small wind turbine by developing a numer-
numerical code. For this purpose, 10 airfoils, namely, BW-3, E387, FX 63-137, S822, S834,
ical code. For this purpose, 10 airfoils, namely, BW-3, E387, FX 63-137, S822, S834, SD7062,
SD7062, SG6040, SG6043, SG6051, and USNPS4, were selected and the differential evolution
SG6040, SG6043, SG6051, and USNPS4, were selected and the differential evolution (DE)
(DE) optimization technique was employed to determine the best geometric shape of the
optimization
blades. The technique
twist anglewas andemployed
chord length to determine the bestasgeometric
were considered shape ofwhile
design variables the
blades. The twist
maximizing angle and
the power chord length
coefficient were considered
and minimizing as design
the startup time werevariables while
defined max-
as design
imizing the power coefficient and minimizing the startup time were
goals. The blade element moment (BEM) method was used to calculate the power coefficient defined as design
goals.
and theThe blade time.
startup elementThemoment
BEM code (BEM)
and method was usedtools
the optimization to calculate the validated.
were both power coeffi-
The
cient and the startup time.
main results can be summarized as: The BEM code and the optimization tools were both validated.
The main results can be summarized as:
• Regardless of the type of airfoil, using ideal equations to determine the twist angle
• Regardless of the type of airfoil, using ideal equations to determine the twist angle
and chord length to maximize the power coefficient, raises the turbine startup time;
• and chord length to maximize the power coefficient, raises the turbine startup time;
The SG6043 airfoil has the highest power coefficient while the S822 and SG6040 airfoils
• The
have SG6043 airfoilpower
the lowest has the highest power
coefficients. The coefficient
reason for while the S822 and
the superiority SG6040 isair-
of SG6043 its
foils have the lowest power coefficients. The reason for the
high lift-to-drag ratio. It is highly recommended to use this airfoil in windy superiority of SG6043 is
areas
its high lift-to-drag ratio. It is highly recommended to use
where the purpose of designing small wind turbines is to achieve the maximum this airfoil in windy areas
where
power the purpose of designing small wind turbines is to achieve the maximum
coefficient;
• power
Among coefficient;
the optimal blades for achieving the maximum power coefficient, the blades
• Among
with thethe FXoptimal
63-137 andblades for achieving
USNPS4 airfoils the
havemaximum
the shortestpowerand coefficient,
longest chord the lengths,
blades
with the FX 63-137 and USNPS4 airfoils have the shortest and
respectively. This has caused the power coefficient not to be as high as expected, longest chord lengths,
respectively.
despite the high Thislift-to-drag
has causedratiothe power coefficient
of the FX not to be as high as expected, de-
63-137 airfoil;
• spite the high lift-to-drag ratio of the FX 63-137 airfoil;
Regardless of the airfoil type, raising the twist angle and chord length in the root
• Regardless
section reduces of the airfoil
the type,
turbine raising
startup the twist angle and chord length in the root
time;
• section
From the reduces
startuptheviewpoint,
turbine startup time; airfoil has the best performance among the
the BW-3
selected airfoils. This is due to the low inertia of the blades fitted with this airfoil.
Therefore, in areas with low wind speeds where having a low startup time is greatly
important, the use of this airfoil is highly recommended;
• The S822, S834, and SG6040 airfoils have the highest startup time. The common aspect
of these three airfoils is their high surface area;
Machines 2022, 10, 687 17 of 22

• Although the thinness of the airfoil is an advantage for reducing the blade moment of
inertia and hence obtaining a better performance of the turbine at low wind speeds,
the blade fitted with a thinner airfoil does not necessarily have a lower startup time
than the blade with a thicker airfoil. This is because the airfoil type completely affects
the distribution of twist angle and chord length. This is accompanied by fundamental
changes in the startup torque and moment of inertia, both of which play an influential
role in the startup process of the turbine;
• Regardless of the airfoil type, when the blade begins to rotate, the startup torque first
decreases slightly and then starts to increase;
• The highest startup torque is produced by the blade fitted with the USNPS4 airfoil
and the lowest startup torque is produced by the blade fitted with the BW-3 airfoil.

Author Contributions: Conceptualization, V.A.; methodology, V.A., M.N. and R.K.; software, V.A.;
validation, V.A.; formal analysis, R.K.; investigation, V.A.; resources, F.A.; data curation, V.A.; writing—
original draft preparation, V.A. and F.A.; writing—review and editing, F.A. and W.Y.; visualization,
F.A.; supervision, M.N., R.K. and W.Y.; project administration, M.N.; funding acquisition, W.Y. All
authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The data presented in this study are available on reasonable request
from the corresponding author.
Conflicts of Interest: The authors declare no conflict of interest.

Nomenclature

A Airfoil cross-sectional area [m2 ]

a Axial induction factor
a0 Rotational induction factor
Cd Drag coefficient
Cl Lift coefficient
CP Power coefficient
c Blade chord [m]
F Prandtl tip loss factor
Fx Axial force [kg·m·s−2 ]
Fy Total tangential force [kg·m·s−2 ]
f Term in Prandtl tip loss factor
J Rotational inertia [kg·m2 ]
N Number of blades
n Weighting factor
P Power [kg·m2· s−3 ]
Q Torque [kg·m2 ·s−2 ]
Qr Resistive torque [kg·m2 ·s−2 ]
Qs Startup torque [kg·m2 ·s−2 ]
Qs0 Startup torque at t = 0 [kg·m2 ·s−2 ]
R Blade tip radius [m]
Re Reynolds number
r Radial coordinate along blade [m]
t Time [s]
S The swept area of the blades [m2 ]
Ts Startup time [s]
U Wind velocity [m·s−1 ]
Urated Wind velocity for rated power [m·s−1 ]
UT Total velocity at blade element [m·s−1 ]
 Machines 2022, 10, 687 19 of 23
Machines 2022, 10, 687 18 of 22

𝑈 Wind velocity for rated power [m·s−1]

Greek
𝑈 Symbols Total velocity at blade element [m·s−1]
α Greek Symbols Angle of attack
θ 𝛼 Angleangle
Blade twist of attack
λ 𝜃 Blade
Tip speed twist angle
ratio
λ Tip speed ratio
λrated Tip speed ratio for rated power
λ Tip speed ratio for rated power
λr λ Local tip speed ratio
Local tip speed ratio
ρ 𝜌 DensityDensity −3 ]
[kg·m[kg·m −3]

φ 𝜙 Blade inflow angleangle

Blade inflow
ω 𝜔 Angular velocity
Angular [s−1[s
velocity ] −1]
Subscripts
Subscripts
1 1 The upwind
The upwind face
face of of the
the rotor
rotor
b Blade
b Blade
G Generator
G h Generator
Hub
h s Hub Startup
s Abbreviations Startup
BEM
Abbreviations Blade element momentum
BEMDE Differential
Blade element evolution algorithm
momentum
HAWT Horizontal-axis wind turbine
DE Differential evolution algorithm
NACA U.S. National Advisory Committee on Aeronautics
HAWT Horizontal-axis wind turbine
NACA U.S. National Advisory Committee on Aeronautics
Appendix A
Figure A1 illustrates the changes in Cl and Cd values for different airfoil blades which
Appendix A
were investigated throughout this research. These values were fed into the developed nu-
Figure
merical A1asillustrates
code the
a database forchanges in Clnumerical
the current and Cd values for different
computations. It isairfoil blades
necessary which
to ex-
were investigated throughout this research. These values were fed into the
plain that, in addition to the relevant references, these coefficients are also available on developed
numerical
the website code as a database
maintained for theMichael
by Professor currentSelig’s
numerical
groupcomputations.
at the University It of
is Illinois
necessary at to
explain that, in addition to the(accessed
www.ae.illinoise.edu/m-seliq/ relevant onreferences,
5 Augustthese
2022).coefficients are also available
Numerous researchers in this on
the website
field maintained
have employed theby Professor
data acquiredMichael Selig’s
from this group
website at thestudy
for their University
[48–51],ofandIllinois
the at
www.ae.illinoise.edu/m-seliq/
use of this website as a valid source (accessed on 5 August
for obtaining 2022).
lift and drag Numerous
coefficient researchers
values was rec- in
this field have
ommended in employed
a referencethe
bookdata
[19].acquired from this website for their study [48–51], and
the use of this website as a valid source for obtaining lift and drag coefficient values was
recommended in a reference book [19].
1.6
2.0
1.2
1.2
1.5

0.8
0.8 Re = 100,000
1.0
Re = 200,000 Re = 100,000
Cl

Re = 60,000
Cl

Re = 300,000 Re = 150,000
Cl

Re = 100,000 0.4 Re = 350,000 Re = 200,000

0.4 Re = 200,000 Re = 460,000
0.5 Re = 350,000
Re = 300,000 Re = 500,000 Re = 500,000
Re = 400,000
0.0 0.0
0.0

−0.4 −0.5
−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16
−4 −2 0 2 4 6 8 10 12
α (deg) −6 −4 −2 0 2 4 6 8 10 12 α(deg)
0.08 α(deg) 0.15

Re = 100,000
Re = 60,000 0.08 0.12 Re = 150,000
Re = 100,000 Re = 100,000
0.06 Re = 200,000
Re = 200,000
Re = 200,000 Re = 350,000
Re = 300,000 Re = 300,000 Re = 500,000
0.09
0.06 Re = 350,000
Cd

Re = 400,000
Re = 460,000
Cd

0.04 Re = 500,000 0.06

Cd

0.04
0.03
0.02
0.02
0.00
−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16

−4 −2 0 2 4 6 8 10 12 α(deg)
0.00
α (deg) −6 −4 −2 0 2 4 6 8 10 12
α(deg)

(a) (b) (c)

Figure A1. Cont.

 Machines 2022, 10, 687 20 of 23
Machines 2022, 10, 687 19 of 22

1.5
1.5

1.0 Re = 100,000 1.6

Re = 200,000 1.0
Re = 300,000
Re = 400,000 1.2
0.5 0.5
Re = 100,000
Cl

Cl
Re = 150,000 0.8

Cl
Re = 100,000
0.0 0.0 Re = 200,000 Re = 200,000
Re = 350,000 Re = 300,000
Re = 500,000 0.4 Re = 400,000
−0.5
−0.5
0.0
−1.0
−1.0
−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 −12−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 −6 −4 −2 0 2 4 6 8 10 12 14 16

α(deg) α(deg) α(deg)

0.08 0.06
0.05
Re = 100,000
0.07 Re = 100,000
Re = 150,000 0.05
Re = 200,000 Re = 200,000
Re = 300,000
0.06
0.04 Re = 350,000 Re = 100,000
Re = 400,000 Re = 500,000 0.04 Re = 200,000
0.05 Re = 300,000
Re = 400,000

Cd
Cd
0.04
Cd

0.03 0.03
0.03
0.02
0.02 0.02

0.01 0.01
0.01
0.00 −6 −4 −2 0 2 4 6 8 10 12 14 16
−12−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16
−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 α(deg)
α(deg)
α(deg)

(d) (e) (f)

1.5 1.2

2.0
1.0 0.8
1.5

0.4

Cl
0.5 Re = 100,000
Re = 100,000
Cl

1.0 Re = 150,000
Re = 150,000
Re = 200,000
Cl

Re = 200,000 Re = 100,000
Re = 300,000 Re = 150,000 0.0 Re = 300,000
0.0 0.5
Re = 200,000
Re = 400,000 Re = 400,000
Re = 300,000
Re = 500,000 Re = 500,000
Re = 400,000
0.0 Re = 500,000
−0.5 −0.4

−0.5
−8 −6 −4 −2 0 2 4 6 8 10 12 14 −6 −4 −2 0 2 4 6 8 10 12
−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16
α(deg) α(deg) α(deg)
0.07 0.10
0.05
Re = 100,000 Re = 100,000
0.06 0.08 Re = 100,000
Re = 150,000
Re = 150,000 Re = 150,000
Re = 200,000 Re = 200,000 Re = 200,000
0.05 Re = 300,000 Re = 300,000 0.04 Re = 300,000
0.06 Re = 400,000 Re = 400,000
Re = 400,000
Re = 500,000
Cd

Re = 500,000 Re = 500,000
0.04
Cd

Cd

0.04 0.03
0.03
0.02
0.02
0.02
0.00
−8 −6 −4 −2 0 2 4 6 8 10 12 14
0.01 0.01
α(deg)
−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16
−6 −4 −2 0 2 4 6 8 10 12
α(deg)
α(deg)

(g) (h) (i)

1.6

1.2

0.8
Cl

Re = 100,000
Re = 200,000
Re = 300,000
0.4
Re = 400,000

0.0

−6 −4 −2 0 2 4 6 8 10 12 14 16
α(deg)

0.06
Re = 100,000
Re = 200,000
0.05 Re = 300,000
Re = 400,000
0.04
Cd

0.03

0.02

0.01

−6 −4 −2 0 2 4 6 8 10 12 14 16
α(deg)

(j)

Figure A1. Changes of the Cl and Cd values for (a) BW-3 [31], (b) E387 [30], (c) FX 63-137 [30],
(d) S822 [31], (e) S834 [30], (f) SD7062 [31], (g) SG6040 [31], (h) SG6043 [31], (i) SG6051 [31], and
(j) USNPS4 [31].
Machines 2022, 10, 687 20 of 22

Appendix B
Table A1 presents the Cp values from the current study and the experimental research
of Anderson et al. [45]. This table also presents the deviation between these two sets
of results.

Table A1. Detailed changes of Cp at different λ values for the current research and the study of
Anderson et al. [45].

Cp
λ
Current Numerical Code Experimental Data [45] Absolute Error Squared Error Error (%)
5.45 0.25960 0.26498 0.00538 2.89444 × 10−5 2.03
6.17 0.31963 0.34177 0.02214 0.00049018 6.47
6.93 0.36843 0.39325 0.02482 0.000616032 6.31
7.759 0.41506 0.41983 0.00477 2.27529 × 10−5 1.13
8.34 0.43715 0.42827 −0.00888 7.88544 × 10−5 −2.07
8.97 0.45241 0.44008 −0.01233 0.000152029 −2.80
9.329 0.45768 0.43207 −0.02561 0.000655872 −5.92
9.73 0.46076 0.44346 −0.0173 0.00029929 −3.90
10.16 0.46081 0.45443 −0.00638 4.07044 × 10−5 −1.40
10.48 0.45829 0.44726 −0.01103 0.000121661 −2.46
10.918 0.45114 0.43966 −0.01148 0.00013179 −2.61
11.62 0.43138 0.42152 −0.00986 9.72196 × 10−5 −2.33
11.89 0.42146 0.40000 −0.02146 0.000460532 −5.36
13.02 0.36678 0.36878 0.002 4 × 10−6 0.54

Appendix C
Table A2 shows the optimal values of Cp , Ts , Jb , and QS0 for n = 1, 0.8, and 0.6 weighting
factors. The values of n = 1 and n = 0.8 are two accepted values in the literature for the design
of small wind turbine blades to operate in windy areas and areas with low wind speed,
respectively. In this regard, the work of Pourrajabian et al. [20] and Rahgozar et al. [21]
can be referred to. Considering small values for n results in impractical blades whose
aerodynamic performance is poor. The table below summarizes the optimization results
of the current research with n = 0.6, along with n = 1 and n = 0.8. As can be noticed from
the table, regardless of the airfoil type, considering n = 0.6 drastically reduces the power
coefficient, which is indeed the main goal of wind turbine blade design.

Table A2. The optimal values of Cp , Ts , Jb , and QS0 for selected airfoils for n = 1, 0.8, and 0.6.

n=1 n = 0.8 n = 0.6

A
Airfoil Jb QS0 Jb QS0 Jb QS0
(m2 ) Cp Ts (s) Cp Ts (s) Cp Ts (s)
(kgm2 ) (Nm) (kgm2 ) (Nm) (kgm2 ) (Nm)
BW-3 0.0364 0.496 2.78 0.438 1.165 0.486 1.82 0.494 1.651 0.435 1.33 0.364 1.673
E387 0.0573 0.502 4.87 0.786 1.213 0.485 3.10 0.889 1.722 0.439 2.27 0.660 1.753
FX 63-137 0.0831 0.499 8.44 0.584 0.795 0.489 3.26 0.964 1.757 0.449 2.59 0.760 1.749
S822 0.1087 0.495 13.25 1.600 1.006 0.481 7.02 2.000 1.711 0.416 4.71 1.337 1.725
S834 0.1042 0.498 11.54 1.872 1.193 0.483 6.87 1.921 1.682 0.411 4.37 1.137 1.641
SD7062 0.0883 0.497 6.15 1.394 1.435 0.488 4.49 1.421 1.834 0.438 3.31 1.003 1.794
SG6040 0.1042 0.495 15.67 1.105 0.817 0.480 6.02 1.678 1.695 0.429 4.31 1.233 1.726
SG6043 0.0685 0.506 5.72 0.580 0.972 0.496 2.89 0.842 1.743 0.459 2.33 0.687 1.759
SG6051 0.0839 0.504 9.80 1.253 1.118 0.490 5.13 1.456 1.708 0.428 3.45 0.950 1.689
USNPS4 0.0884 0.503 5.36 1.586 1.731 0.493 4.22 1.423 1.919 0.443 3.14 0.987 1.836
Machines 2022, 10, 687 21 of 22

References
1. Afsharpanah, F.; Pakzad, K.; Mousavi Ajarostaghi, S.S.; Poncet, S.; Sedighi, K. Accelerating the Charging Process in a Shell and
Dual Coil Ice Storage Unit Equipped with Connecting Plates. Int. J. Energy Res. 2022, 46, 7460–7478. [CrossRef]
2. Shyu, C.-W. A Framework for ‘Right to Energy’ to Meet UN SDG7: Policy Implications to Meet Basic Human Energy Needs,
Eradicate Energy Poverty, Enhance Energy Justice, and Uphold Energy Democracy. Energy Res. Soc. Sci. 2021, 79, 102199.
[CrossRef]
3. Karthikeyan, N.; Kalidasa Murugavel, K.; Arun Kumar, S.; Rajakumar, S. Review of Aerodynamic Developments on Small
Horizontal Axis Wind Turbine Blade. Renew. Sustain. Energy Rev. 2015, 42, 801–822. [CrossRef]
4. Afsharpanah, F.; Mousavi Ajarostaghi, S.S.; Arıcı, M. Parametric Study of Phase Change Time Reduction in a Shell-and-Tube Ice
Storage System with Anchor-Type Fin Design. Int. Commun. Heat Mass Transf. 2022, 137, 106281. [CrossRef]
5. Afsharpanah, F.; Cheraghian, G.; Hamedani, F.A.; Shokri, E.; Ajarostaghi, S.S.M. Utilization of Carbon-Based Nanomaterials and
Plate-Fin Networks in a Cold PCM Container with Application in Air Conditioning of Buildings. Nanomaterials 2022, 12, 1927.
[CrossRef] [PubMed]
6. Afsharpanah, F.; Pakzad, K.; Mousavi Ajarostaghi, S.S.; Arıcı, M. Assessment of the Charging Performance in a Cold Thermal
Energy Storage Container with Two Rows of Serpentine Tubes and Extended Surfaces. J. Energy Storage 2022, 51, 104464.
[CrossRef]
7. Lattieff, F.A.; Atiya, M.A.; Mahdi, J.M.; Majdi, H.S.; Talebizadehsardari, P.; Yaïci, W. Performance Analysis of a Solar Cooling
System with Equal and Unequal Adsorption/Desorption Operating Time. Energies 2021, 14, 6749. [CrossRef]
8. Zhao, Z.; Wang, D.; Wang, T.; Shen, W.; Liu, H.; Chen, M. A Review: Approaches for Aerodynamic Performance Improvement of
Lift-Type Vertical Axis Wind Turbine. Sustain. Energy Technol. Assessments 2022, 49, 101789. [CrossRef]
9. Soni, V.; Sharma, A.; Singh, V. A Critical Review on Nature Inspired Optimization Algorithms. In IOP Conference Series Materials
Science and Engineering; IOP Publishing: Bristol, UK, 2021; Volume 1099, p. 012055. [CrossRef]
10. Tušar, T.; Filipič, B. Differential Evolution versus Genetic Algorithms in Multiobjective Optimization. In Proceedings of the
International Conference on Evolutionary Multi-Criterion Optimization, Matsushima, Japan, 5–8 March 2007; Volume 4403,
pp. 257–271. [CrossRef]
11. Lilla, A.D.; Khan, M.A.; Barendse, P. Comparison of Differential Evolution and Genetic Algorithm in the Design of Permanent
Magnet Generators. In Proceedings of the IEEE International Conference on Industrial Technology, Cape Town, South Africa,
25–28 February 2013; pp. 266–271. [CrossRef]
12. IEC 61400.2; Wind Turbines-Design Requirements for Small Wind Turbines. British Standards European Norm: London, UK,
2013; pp. 61400–61402.
13. Clausen, P.D.; Wood, D.H. Recent Advances in Small Wind Turbine Technology. Wind Eng. 2000, 24, 189–201. [CrossRef]
14. Astolfi, D. A Study of the Impact of Pitch Misalignment on Wind Turbine Performance. Machines 2019, 7, 8. [CrossRef]
15. Mostafaeipour, A. Economic Evaluation of Small Wind Turbine Utilization in Kerman, Iran. Energy Convers. Manag. 2013,
73, 214–225. [CrossRef]
16. Castellani, F.; Astolfi, D.; Peppoloni, M.; Natili, F.; Buttà, D.; Hirschl, A. Experimental Vibration Analysis of a Small Scale Vertical
Wind Energy System for Residential Use. Machines 2019, 7, 35. [CrossRef]
17. Eltayesh, A.; Castellani, F.; Burlando, M.; Bassily Hanna, M.; Huzayyin, A.S.; El-Batsh, H.M.; Becchetti, M. Experimental and
Numerical Investigation of the Effect of Blade Number on the Aerodynamic Performance of a Small-Scale Horizontal Axis Wind
Turbine. Alex. Eng. J. 2021, 60, 3931–3944. [CrossRef]
18. Worasinchai, S.; Ingram, G.L.; Dominy, R.G. Effects of Wind Turbine Starting Capability on Energy Yield. J. Eng. Gas Turbines
Power 2012, 134, 042603. [CrossRef]
19. Wood, D. Small Wind Turbines: Analysis, Design, and Application; Green Energy and Technology; Springer: London, UK, 2011;
Volume 38, ISBN 9781849961745.
20. Pourrajabian, A.; Dehghan, M.; Javed, A.; Wood, D. Choosing an Appropriate Timber for a Small Wind Turbine Blade: A
Comparative Study. Renew. Sustain. Energy Rev. 2019, 100, 1–8. [CrossRef]
21. Rahgozar, S.; Pourrajabian, A.; Kazmi, S.A.A.; Kazmi, S.M.R. Performance Analysis of a Small Horizontal Axis Wind Turbine
under the Use of Linear/Nonlinear Distributions for the Chord and Twist Angle. Energy Sustain. Dev. 2020, 58, 42–49. [CrossRef]
22. Clifton-Smith, M.J. Aerodynamic Noise Reduction for Small Wind Turbine Rotors. Wind Eng. 2010, 34, 403–420. [CrossRef]
23. Andrew Ning, S.; Damiani, R.; Moriarty, P.J. Objectives and Constraints for Wind Turbine Optimization. J. Sol. Energy Eng. 2014,
136, 041010. [CrossRef]
24. Natarajan, K.; Suthakar, T. Insight Aerodynamic Analysis on Small-Scale Wind Turbines Airfoils for Low Reynolds Number
Applications. Environ. Prog. Sustain. Energy 2022, 41, e13807. [CrossRef]
25. Bai, C.-J.; Wang, W.-C. Review of Computational and Experimental Approaches to Analysis of Aerodynamic Performance in
Horizontal-Axis Wind Turbines (HAWTs). Renew. Sustain. Energy Rev. 2016, 63, 506–519. [CrossRef]
26. Pourrajabian, A.; Dehghan, M.; Rahgozar, S. Genetic Algorithms for the Design and Optimization of Horizontal Axis Wind
Turbine (HAWT) Blades: A Continuous Approach or a Binary One? Sustain. Energy Technol. Assess. 2021, 44, 101022. [CrossRef]
27. Elizondo, J.; Martínez, J.; Probst, O. Experimental Study of a Small Wind Turbine for Low-and Medium-wind Regimes. Int. J.
Energy Res. 2009, 33, 309–326. [CrossRef]
Machines 2022, 10, 687 22 of 22

28. Singh, R.K.; Ahmed, M.R. Blade Design and Performance Testing of a Small Wind Turbine Rotor for Low Wind Speed Applications.
Renew. Energy 2013, 50, 812–819. [CrossRef]
29. Giguere, P.; Selig, M.S. New Airfoils for Small Horizontal Axis Wind Turbines. J. Sol. Energy Eng. 1998, 120, 108–114. [CrossRef]
30. Selig, M.S.; McGranahan, B.D. Wind Tunnel Aerodynamic Tests of Six Airfoils for Use on Small Wind Turbines. J. Sol. Energy Eng.
2004, 126, 986–1001. [CrossRef]
31. Lyon, C.A.; Broeren, A.P.; Giguere, P.; Gopalarathnam, A.; Selig, M.S. Summary of Low-Speed Airfoil Data, Vol. 3.; Soartech: Virginia
Beach, VA, USA, 1997; ISBN 0-9646747-3-4.
32. Selig, M.; McGranahan, B.; Broughton, B. UIUC Low-Speed Airfoil Tests; University of Illinois at Urbana-Champaign: Champaign,
IL, USA, 2001.
33. Storn, R.; Price, K. Differential Evolution—A Simple and Efficient Heuristic for Global Optimization over Continuous Spaces.
J. Glob. Optim. 1997, 11, 341–359. [CrossRef]
34. Ahmad, M.F.; Isa, N.A.M.; Lim, W.H.; Ang, K.M. Differential Evolution: A Recent Review Based on State-of-the-Art Works. Alex.
Eng. J. 2022, 61, 3831–3872. [CrossRef]
35. Stokes, Z.; Mandal, A.; Wong, W.K. Using Differential Evolution to Design Optimal Experiments. Chemom. Intell. Lab. Syst. 2020,
199, 103955. [CrossRef]
36. Jureczko, M.; Mrówka, M. Multiobjective Optimization of Composite Wind Turbine Blade. Materials 2022, 15, 4649. [CrossRef]
37. Spera, D.A. Wind Turbine Technology; ASME Publishing: New York, NY, USA, 1994.
38. Glauert, H. Airplane Propellers. In Aerodynamic Theory; Springer: Berlin/Heidelberg, Germany, 1935; pp. 169–360.
39. Hau, E. Wind Turbines: Fundamentals, Technologies, Application, Economics; Springer Science & Business Media: Berlin, Germany,
2013; Volume 9783642271, ISBN 9783642271519.
40. Castellani, F.; Astolfi, D.; Natili, F.; Mari, F. The Yawing Behavior of Horizontal-Axis Wind Turbines: A Numerical and
Experimental Analysis. Machines 2019, 7, 15. [CrossRef]
41. Wood, D.H. A Blade Element Estimation of the Cut-in Wind Speed of a Small Turbine. Wind Eng. 2001, 25, 125–130. [CrossRef]
42. Sessarego, M.; Wood, D. Multi-Dimensional Optimization of Small Wind Turbine Blades. Renew. Wind. Water Sol. 2015, 2, 9.
[CrossRef]
43. Clifton-Smith, M.J.; Wood, D.H. Further Dual Purpose Evolutionary Optimization of Small Wind Turbine Blades. In The Journal of
Physics: Conference Series; IOP Publishing: Bristol, UK, 2007; Volume 75, p. 12017.
44. Pourrajabian, A.; Ebrahimi, R.; Mirzaei, M. Applying Micro Scales of Horizontal Axis Wind Turbines for Operation in Low Wind
Speed Regions. Energy Convers. Manag. 2014, 87, 119–127. [CrossRef]
45. Anderson, M.B.; Milborrow, D.J.; Ross, J.N. Performance and Wake Measurements on a 3 M Diameter Horizontal Axis Wind
Turbine. Comparison of Theory, Wind Tunnel and Field Test Data. In Proceedings of the Papers Presented at the International
Symposium on Wind Energy Systems, Cambridge, UK, 21–24 September 1982; Volume 2, pp. 113–135.
46. Burton, T.; Jenkins, N.; Sharpe, D.; Bossanyi, E. Wind Energy Handbook; John Wiley & Sons: Hoboken, NJ, USA, 2011;
ISBN 111999392X.
47. Selig, M. UIUC Applied Aerodynamics Group, Department of Aerospace Engineering. Available online: https://m-selig.ae.
illinois.edu/ (accessed on 5 August 2022).
48. Wang, L.; Xu, J.; Luo, W.; Luo, Z.; Xie, J.; Yuan, J.; Tan, A.C.C. A Deep Learning-Based Optimization Framework of
Two-Dimensional Hydrofoils for Tidal Turbine Rotor Design. Energy 2022, 253, 124130. [CrossRef]
49. Gupta, P.; Tyagi, P.; Singh, R.K. Analysis of Generative Adversarial Networks for Data-Driven Inverse Airfoil Design. Lect. Notes
Netw. Syst. 2022, 350, 251–261. [CrossRef]
50. Duru, C.; Alemdar, H.; Baran, O.U. A Deep Learning Approach for the Transonic Flow Field Predictions around Airfoils. Comput.
Fluids 2022, 236, 105312. [CrossRef]
51. Wang, S.; Sun, G.; Chen, W.; Zhong, Y. Database Self-Expansion Based on Artificial Neural Network: An Approach in Aircraft
Design. Aerosp. Sci. Technol. 2018, 72, 77–83. [CrossRef]