You are on page 1of 20

Ocean Engineering 269 (2023) 113688

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Effect of leading-edge protrusion shapes for passive flow control measure


on wind turbine blades
Sujit Roy a, Biplab Das a, b, Agnimitra Biswas a, *
a
Department of Mechanical Engineering, National Institute of Technology Silchar, Silchar, Assam, 788010, India
b
School of the Built Environment, Centre for Sustainable Technologies, Ulster University, N Ireland, BT37 0QB, UK

A R T I C L E I N F O A B S T R A C T

Handling Editor: Prof. A.I. Incecik In the present study, protrusions in the leading edge of the NACA 4415 airfoil were incorporated as a passive flow
control measure on the wind turbine blades. Three airfoil models: unmodified leading-edge (ULE), spherical
Keywords: leading-edge protrusion (SLEP), and triangular leading-edge protrusion (TLEP), were investigated experimen­
NACA4415 airfoil tally. Thereafter, CFD investigations were carried out using ANSYS 14.0 simulation tool to observe the flow
Flow separation
characteristics around the airfoils. Further, LEP-based passive design was applied on a horizontal axis wind
Leading-edge protrusion
turbine (HAWT) to investigate its performance. An amplitude (A) of 1% of the chord length (C) was considered as
Lift coefficient
Subsonic wind tunnel the height of protrusion, while a distance of 0.25C was maintained between the protrusions to fabricate the
CFD experimental models. The study was performed at a low Reynolds number (Re) of 1.5 × 105 for a wide angle of
attack (α) of 0◦ –20◦ . The experimental results demonstrate that the SLEP model has a higher lift coefficient at α
≥ 18◦ , whereas the TLEP model performs poorly when compared with the ULE model. The instantaneous lift
coefficient plot indicates that the SLEP model generates a more stable force at a higher angle of attack. The
computational analysis reveals that a larger primary circulation (extended till 0.2C) is observed in the ULE model
at a post-stall angle of attack (α = 18◦ ), indicating early flow separation. Whereas, in the SLEP and TLEP models,
it is extended to 0.70C and 0.54C, respectively, indicating the best flow controlling measures achieved by using
the SLEP model. The investigations of HAWT rotors with LEP revealed that SLEP HAWT exhibit 8.2% more power
coefficient than ULE HAWT.

1. Introduction Fish and Battle, 1995). Therefore, researchers started to investigate the
flow characteristics around the protrusions. Leading-edge protrusions
The formation of separation bubbles and flow separation in the (LEPs) have helped prevent early flow separation by re-energizing the
suction surface of the airfoil at low Reynolds number (Re) flow de­ boundary layer and improving the turbulence intensity. LEPs generate
teriorates the aerodynamic performance of the wind turbine airfoil counter-rotating vortices, but how they interact with airflow over an
blade (Gopinathan and Rose, 2021). The occurrence of flow separation airfoil is remained unknown, especially for low-Re flows. Miklosovic
leads the airfoil to an early stall (lift coefficient starts decreasing from et al. (2004) postulated that an increase in the momentum of a boundary
this point onwards), affecting the overall lift generation. Researchers are layer, resulting in stall delay, may be achieved by using LEPs.
focusing on flow-control approaches to dodge or prolong the flow sep­ The researchers have tried to implement the LEPs in different airfoils
aration in the airfoil’s suction surface as a part of a new research trend. to investigate their effect on the airfoil’s aerodynamic behaviors. Han­
The separated flow causes a loss of lift and an increase in pressure drag sen et al. (2011) noticed that a leading-edge sinusoidal protrusion
(Zhou et al., 2017). Koca et al. (2018) examined the flow separation (LESSP) had enhanced the post-stall aerodynamic efficiency of the
process, the development of laminar separation bubbles (LSBs), vortex airfoil and had given a smoother stall than an unmodified airfoil.
shedding, and other fluid flow phenomena in a plain NACA 4412 airfoil Miklosovic et al. (2004) replaced a NACA 0020 wind turbine blade’s
operating at low Re. The protrusions on the leading edge of the hump­ leading-edge with a LESSP. They discovered that, except 9◦ –13◦ , the
back whale flippers had avoided the flow separation over the flippers, modified blade outperformed the ordinary blade. The maximum lift
leading to their agile motion and higher lift generation (Weihs, 1981; coefficient was increased by 6%, and the stall angle (angle of attack at

* Corresponding author.
E-mail addresses: sujitme12@gmail.com (S. Roy), bpd@mech.nits.ac.in, b.das@ulster.ac.uk (B. Das), agnibis@yahoo.co.in (A. Biswas).

https://doi.org/10.1016/j.oceaneng.2023.113688
Received 2 October 2022; Received in revised form 31 December 2022; Accepted 10 January 2023
Available online 19 January 2023
0029-8018/© 2023 Elsevier Ltd. All rights reserved.
S. Roy et al. Ocean Engineering 269 (2023) 113688

which stall phenomenon starts) was delayed by 40%. Johari et al. (2007) occurrence from the leading-edge stall to the trailing-edge stall. In the
redesigned the NACA 634–021 airfoil and observed lift coefficient in­ case of the rotating turbine model, LESSP resulted in an enhanced dy­
crease up to 50% in the post-stall zone. They also discovered that the namic lift of 16.3% at the azimuth angle of 78–110◦ and was able to
amplitude variation of a sinusoidal profile had a greater impact on suppress the dynamic stall at the azimuth angle of 90–150◦ , operating at
aerodynamic performance than the wavelength change. An airfoil with a a tip speed ratio of 2. Zhang et al. (2021) suggested using LESSP of
LESSP was changed by Zhang and Wu (2012) to study the effect of wind smaller amplitude, as larger amplitude and smaller wavelength could
velocity on the airfoil’s performance. With a redesigned airfoil and deteriorate the maximum lift coefficient. A constant wavelength of
higher inflow wind speed (15 m/s), they saw a maximum increase of 0.25C was used by Johari et al. (2007) and Kim et al. (2012) in their
24.8% in shaft torque while seeing no improvement in shaft torque at studies to observe the LESSP amplitude effect on aerodynamic perfor­
lower wind speeds (10 m/s). At a low-Re of 5 × 104, Zhang et al. (2013) mance. They had identified the positive impact of LESSP as it increased
tested the NACA 634–021 airfoil with and without LESSP. They found stall angle. They observed that the smaller amplitude (A = 0.018C)
that the redesigned airfoil had a lift coefficient 25% higher and a produced as much as a 9.6% enhanced lift to drag ratio. Mishra and De
lift-to-drag ratio of 39.2% higher in the post-stall zone. According to (2021) modified the NACA 0021 airfoil with LESSP of 0.01428C and
Huang et al. (2015), the modification of the SD 8000 wind turbine airfoil 0.05714C amplitude and observed that the modification could delay the
blade with a small wavelength and large amplitude delayed the stall and separation point, and the blade with the smallest amplitude produced
increased the post-stall lift to drag ratio at the cost of the maximum lift higher lift at the post-stall region. Joseph and Sathyabhama (2022) have
coefficient. A study by Bai et al. (2016) indicated that at a low wind noticed that incorporating LESSP reduced the maximum lift coefficient
speed of 6 m/s, a modified blade with the shortest wavelength and the by as much as 30% compared to the unmodified blade, at Re = 2.5 ×
smallest amplitude produced a 17.5% higher power coefficient than a 105. They also observed that LSB is completely removed in the modified
plain blade. Their findings showed that a shorter wavelength functioned blade. Fan et al. (2022) altered the leading edge of the airfoil model with
better at high tip speed ratios than a larger wavelength. Two sinusoidal tubercles of two different amplitudes and observed a higher
leading-edge adjustments, tested by Abate and Mavris (2018) on NREL lift generation with a smaller amplitude. Although, lift to drag ratio
wind turbine blades with LESSP, resulted in enhanced annual energy decreased but the modified airfoils received a smoother stall occurrence.
production (sinusoidal span 62% and 95% of total span). In addition, at Few researchers tried to modify the airfoil’s leading edge using
a higher wind speed of 20 m/s, the overall shaft torque was greater with spherical shape LEPs. Adding a spherical leading-edge protrusion (SLEP)
the redesigned blade. Aerodynamic performance was improved as a of 0.05C amplitude to the NACA 0012 improved the pressure coefficient
result of the leading-edge modification’s lower suction surface pressure. distribution around the airfoil, according to Gawad (2013). They found
Abate et al. (2019) also reported the same findings. A LESSP airfoil that stalling did not occur in the redesigned airfoil for a wide range of
boosted wind turbine power production at a greater wind velocity of 20 angles of attack (0◦ –25◦ ). NACA 4415 airfoil leading-edge modifications
m/s. When the wind speed dropped to 10 m/s, however, an entirely were studied in detail by Aftab and Ahmad (2017), with a range of
other pattern emerged. Using three different wavelengths and ampli­ amplitude from 0.015C to 0.0375C. A greater performance was found
tudes, Arunvinthan et al. (2020) altered the leading edge of a 634–021 for airfoils with smaller amplitudes for LESSP and SLEP models. Addi­
airfoil at Re = 2 × 105. They found that the pre-stall lift coefficient was tionally, after a 10◦ angle of attack, they found that the airfoil with SLEP
higher for the modification with the largest amplitude (0.5C) and had a higher lift to drag than the airfoil with an unmodified leading-edge
longest wavelength (0.12C) than the plain airfoil. The modified airfoil’s (ULE). Zadorozhna et al. (2021) suggested the use of SLEP in small wind
lift-to-drag ratio was lower than the unmodified airfoil near the stall turbines. They performed a computational study varying the spherical
angle due to the modified airfoil’s higher drag coefficient. The lift co­ protuberance amplitude in the range of 0.005C–0.03C. They found that
efficient of the smallest modification was also found to rise when tur­ the overall aerodynamic performance of smaller amplitude (A) is better
bulence intensity was decreased in the pre-stall zone. Sreejith and than all the studied airfoils. However, the largest amplitude showed the
Sathyabhama (2020) performed a computational and experimental highest lift to drag enhancement of 12.86% at α = 0◦ . They also observed
investigation of E216 airfoil using 2–8 mm LESSP and found a maximum that the skin friction drag reduced till α = 8◦ , as the favorable pressure
enhancement in lift coefficient of 4.51% with the highest amplitude at α distribution was achieved with the modified airfoils. The highest drag
= 6◦ . Gopinathan et al. (2020) investigated the flow physics over a swept reduction with A = 0.005C and 0.03C was observed to be 5.84% and
blade (NACA 0015 and NACA 4415) with LESSP of 0.05C and 0.12C 8.46% at α = 0◦ , respectively. Roy et al. (2022a) computationally
amplitude and observed that the flow separation occurred early behind analyzed the effect of SLEP amplitude on the lift and drag coefficient of
the trough section, whereas flow remained attached for a longer chord NACA 4415 airfoil and found that the smallest protrusion gave the
length behind peak section. The counter-rotating vortex generated along highest lift enhancement. However, a positive effect was observed in the
the centerline of the trough section has improved the aerodynamic post-stall region at α ≥ 18◦ . Moreover, it was also noticed that the SLEP
performance of modified blades. model could able to delay the occurrence of flow separation.
Airfoils with and without the LESSP of the NACA 4412 were studied
by Butt et al. (2021). Only in the post-stall region could they observed a 1.1. Inferences drawn and contribution of the present study
better aerodynamic performance due to leading-edge alteration. Yi-Nan
et al. (2021) substituted the source term in the momentum equation by Most studies have applied LESSP to enhance aerodynamic perfor­
adding an extra force generated by the LESSP to develop a novel mance and found a degraded aerodynamic performance in the pre-stall
modeling method. This method helped to reduce the computational cost region and enhanced performance in the post-stall region. The para­
for complex meshing near the LESSPs and eased the design. Their metric variation of the amplitude of spherical protrusion was mostly
modeled airfoil (without any LESSPs) predicted the aerodynamic per­ investigated as wavelength variation has a negligible impact on the
formance and fluid flow characteristics similar to that of an airfoil with airfoil performance improvement (Johari et al., 2007). The literature
LESSPs at different angles of attack. Yan et al. (2021) numerically also revealed that the LEPs with smaller amplitudes performed better
studied the effect of LESSP on the stationary and rotating H-Darrieus than larger ones (Roy et al., 2022a). Although few researchers investi­
turbine models considering NACA 0018 as the base airfoil. They gated the SLEP model’s effect numerically, no experimental in­
observed that the LESSP of smaller amplitude (A = 0.01C) provided the vestigations have been performed so far to obtain the aerodynamic
best aerodynamic performance in the studied range. In contrast, the performance insights with leading-edge spherical shape protrusions as
wavelength variation was insignificant in delaying the stall or causing well as with another protrusion shape. Further, the effects of flow con­
aerodynamic performance improvement. They also noticed that the trol measures by using different protrusion shapes were not investigated
incorporation of LESSP in the static model resulted a change in stall at low-Re flow conditions. Further, the stability of the performance of

2
S. Roy et al. Ocean Engineering 269 (2023) 113688

the airfoil blades with changing the angle of attack due to the dynamism fixed at the airfoil’s leading edge to modify the airfoil (see Fig. 1). The
in the operation of the wind turbine blades is still not investigated for gap between the LEPs (wavelength) was 0.25C, as the wavelength
low-Re flow conditions. In addition, how different protrusion shapes variation has an insignificant effect on the aerodynamic performance
impact the laminar to the turbulent transition of flow on the blades at improvement (Johari et al., 2007). The amplitude of the protrusion is
low-Re flow is still unanswered. The influence of SLEP on performance chosen based on the authors’ previous study on optimization of pro­
of HAWT has also not been investigated so far. trusion size for better aerodynamic performance (Roy et al., 2022a).
In the present study, an LEP with an amplitude of 0.01C and wave­ Whereas the wavelength is selected based on the studies reported by
length of 0.25C was considered to investigate the aerodynamic behav­ Johari et al. (2007), Sudhakar and Karthikeyan (2013), Zhang et al.
iors. Two different shapes of LEPs, i.e., SLEP and triangular leading-edge (2014), Wei et al. (2015), and Aftab and Ahmad (2017). The holes were
protrusion (TLEP), were considered in the present work to modify the created with the help of a 2 mm drill-bit in the leading edge to fix the
leading edge of the NACA 4415 airfoil. The results of the modified protrusions, and then with the use of fine-grade sandpaper, the spherical
models (SLEP and TLEP) were compared with the NACA 4415 model and triangular shapes were created. Two holes were provided at both
with an unmodified leading-edge (ULE). Further, CFD investigations ends of the models (at the center of pressure) to hold it in the supporting
were carried out to visualize the fluid flow characteristics around the stand inside the wind tunnel test section.
airfoils to obtain insights into the flow control features with leading-
edge protrusions in low-Re flow. The comparative study was per­ 2.2. Experimental set-up and procedure
formed at a constant chord-based low-Re of 1.5 × 105 for a wide range of
angles of attack (0◦ –20◦ ). It is revealed from the literature that the stall The experiments were conducted in a suction-type subsonic open-
angle of the unmodified NACA 4415 airfoil at Re = 1.5 × 105 is 12◦ circuit wind tunnel installed at the Fluid Machinery Laboratory, Na­
(Saliveros, 1988). Therefore, the flow characteristics were compared at tional Institute of Technology, India, as shown in Fig. 2(a). The tunnel
three different conditions of airfoils: (i) at α = 4◦ (pre-stall), (ii) at α = has a 140 × 140 cm square entrance to suck air from the atmosphere,
12◦ (stall), and (iii) at α = 18◦ (post-stall). The pressure, velocity, followed by a 20 cm settling chamber. A shorter convergent section and
streamline, and turbulent kinetic energy contours were analyzed prop­ a longer divergent section were maintained to avoid flow separation
erly to understand the flow characteristic around airfoil models with the inside the wind tunnel. A test section of 30 × 30 cm cross-section and
considered protrusion shapes. Additionally, the SLEP passive control 100 cm length was kept to perform the experiments. A hexagonal hon­
method was applied to the HAWT blade, and peformance of turbine eycomb aluminum mesh of 15 cm thickness was placed after the settling
model at different pitch angles and wind velocities was investigated chamber to keep the flow streamlined in the test section. A small
experimentally. To discourse on the novelty of the present investigation, opening at the top of the test section was present to insert the pitot tube
the following points are specified. (which can travel only in the y-direction) to measure the wind velocity
at a given position. A metallic supporting stand was fabricated to sup­
(i) Although the sinusoidal and spherical LEPs were compared port the airfoil models (Fig. 2(b)). A three-axis load cell of ±0.1 N ac­
computationally, no investigation was reported so far on the curacy was installed at the bottom of the test section to measure the lift
comparative analysis of different LEPs experimentally. Therefore, and drag force, as shown in Fig. 2(c). The models were placed at
an experimental campaign was conducted in the present study to different angles of attack from 0 to 20◦ with the help of a 360◦ protractor
discourse the aerodynamic performance insights of a wind tur­ arrangement. A hot-wire anemometer of accuracy ±0.1 m/s was used to
bine airfoil blade with different protrusion shapes under low-Re measure the turbulence intensity (TI) and flow uniformity inside the test
flow conditions. section. The turbulence intensity as well as flow non-uniformity in the
(ii) As the local angle of attack in the wind turbine blade changes test section were found to be ≤ 1%, which is within the acceptable limit
continuously with the wind velocity, a hysteresis study was (Bhuyan and Biswas, 2014; Sengupta et al., 2016). The experiments
conducted to find out the best model among the ULE, SLEP, and were carried out at a chord-based low-Re of 1.5 × 105 (TI = 0.44%),
TLEP in the dynamic angle of attack situations for low-Re number corresponding to the 21.9 m/s inflow wind velocity. As the wind tunnel
flow condition. blockage ratio (ratio of the structural blockage to flow and the tunnel
(iii) A detailed experimental and computational investigation at a cross-sectional area) is below 10%, the blockage correction is not
wide range of angles of attack in pre-stall, stall, and post-stall considered in the current investigation (Chen and Liou, 2011; Schreck
regions was carried out for the airfoil models. These types of et al., 2007).
combined studies on airfoil models with spherical and triangular Further, to observe the application of SLEP in the wind turbine blade,
LEPs were not performed so far. two small-scale horizontal axis wind turbine (HAWT) models (one with
(iv) The application of spherical LEPs in the horizontal axis wind ULE blade and another with SLEP blade) were fabricated and tested at a
turbine (HAWT) rotor model was not investigated so far. There­ low wind velocity (4 m/s - 7 m/s) condition. The velocity of the testing is
fore, the present study compared the performance of a three- limited by the range of velocity available at the exit of the wind tunnel.
bladed HAWT rotor with ULE and SLEP blades at low-velocity The Reynolds numbers at these tested velocities fall in the range of 0.25
conditions. × 105–0.5 × 105, which also fall under the low-Re condition. Un-tapered
and untwisted blades were constructed using wood to fabricate 3-bladed
2. Materials and methods small-scale HAWT models due to its ease of construction (Siram and
Sahoo, 2022). Un-tapered and untwisted HAWT models were also
2.1. Airfoil fabrication investigated by Butterfield et al. (1992), Corbus et al. (2006), Freere
et al. (2010), Hsiao et al. (2013). A fixed chord of 75 mm was maintained
The airfoils were fabricated using balsa wood, as it has a better from root to tip of the turbine blade with a blade span of 150 mm and
strength-to-weight ratio (Sengupta et al., 2022). NACA 4415 airfoil of rotor diameter of 400 mm. The span of the blade was restricted by the
100 mm chord length (C) was selected as the base airfoil to fabricate the exit dimension (60 × 60 cm) of the wind tunnel, whereas the chord
models. The span of the model was considered as 150 mm to maintain a length was chosen to maintain a similar chord-to-span ratio of 0.5, as
gap from the test-section wall. The base profile was attached to both used by Siram and Sahoo (2022). The amplitude of protrusion was kept
sides of a rectangular wooden block and chiseled to the desired shape. as 0.01C with a wavelength of 0.25C. The rope brake dynamometer
Sandpaper of different grades was used to make the surface of the arrangement was used to measure the torque generated by the HAWT
models smoother. A spherical protrusion (SLEP) and triangular protru­ model, as shown in Fig. 2(d). The power coefficients of HAWT models
sion (TLEP) of the amplitude of 1% of chord length (A = 0.01C) were were calculated using Eqs. (1)–(4), following the procedure mentioned

3
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 1. Different fabricated airfoil model (a) unmodified airfoil (Model-1) (b) airfoil with spherical leading-edge protrusion (Model-2) (c) airfoil with triangular
leading-edge protrusion (Model-3).

by Sengupta et al. (2022). As the dimension of the test section is smaller 2012).
in size, the HAWT models were placed at the exit of the wind tunnel, as
R = x1a1 . xa22 . x3a3 . ....... xann (6)
shown in Fig. 2(e). The model was kept at a distance of 1.5 m from the
wind tunnel exit to avoid the structural blockage effect. The blades were [( )2 ( ) ( ) ( ) ]1/2
fixed to the turbine hub with the help of nut and screw, so that different ωR ω1 ω2 2 ω3 2 ωn 2
= a1 + a2 + a3 + .......... + an (7)
pitch angles (angle between chord line and rotor plane) could be ob­ R x1 x2 x3 xn
tained. An anemometer was used to measure the wind velocity
approaching the HAWT model, and a digital contactless tachometer was where, a1 , a2 , a3 , …, an are the degree of the measure variables x1 , x2 ,
used to count the rotational speed of HAWT model. x3 , …, xn
( ) The fractional uncertainty for the lift coefficient (function of lift
Mechanical torqueT = (W − S) rshaft + 2rs g (1) force, L and inflow wind velocity, u) and drag coefficient (function of
drag force, D and inflow wind velocity, u) is found to be ±2.24%,
2T
Torque coefficientCt = (2) calculated using Eqs. (8) and (9). The uncertainties of the lift force and
ρAu2mean R
drag force are calculated from the accuracy of the three-axis load cell
(10 N ± 0.1 N), and the uncertainty of the inflow wind velocity is
Power coefficientCp = λ × Ct (3)
calculated from the accuracy of the pitot tube (30 m/s ± 1%).
ΩR [( ) ]
where, Tip speed ratio (TSR)λ = (4) ω CL ω L 2 ( ω u )2 1/2
u = + 2 (8)
CL L u
where, W is dead weight (kg), S is spring balance load (kg), rshaft is shaft [( ) ( ω )2 ]1/2
radius (m), rs is nylon string radius (m), Ω is angular velocity (rad/s), ρ is ω CD
=
ω 2 D
+ 2
u
(9)
air density (kg/m3), u is incoming average wind velocity (m/s), A is CD D u
rotor area (m2), and R is rotor radius. The uncertainties in calculating TSR and power coefficient were
calculated using Eq. (10) and Eq. (11) and found to be ±1.52% and
2.3. Uncertainty calculation ±1.717%, respectively. Uncertainty in torque measurement (ωT ) is
calculated from the least count of the spring balance (ωS = 1 kg/cm2)
There is always an unavoidable error in the experimental work due and expressed as Eq. (12) (Singh et al., 2015). The instrumental uncer­
to the instruments’ accuracy. An uncertainty analysis was carried out to tainty combining the uncertainties of anemometer and tachometer is
find the expected instrumental error in the lift and drag coefficient, calculated using Eq. (13) (Singh et al., 2015) and found to be ±1.414%.
using Eq. (5) (Kline and McClintock, 1953)
[( ) ( ω )2 ]1/2
[( )2 ]1/2 ωTSR ω 2
(10)
)2 ( )2 ( )2 ( Ω u
= + 2
∂R ∂R ∂R ∂R TSR Ω u
ωR = ω1 + ω2 + ω3 + ................ + ωn
∂x1 ∂x2 ∂x3 ∂xn
[( )2 (ω )2 ( ω )2 ]1/2
(5) ω Cp
=
ω TSR
+
T
+ 2
u
(11)
Cp TSR T u
where, ωR is the total uncertainty of the derived quantity, R (function of
(ω ) (ω )
x1 , x2 , x3 , …, xn ) and ω1 , ω2 , ω3 , … … …, ωn are the individual un­ T
=
S
(12)
certainty of the measured variables x1 , x2 , x3 , …, xn . T S
If the measured variables are related to the derived quantity in a [ ]1/2
product form, as shown in Eq. (6), the uncertainty can be expressed as ω Instrument = (ω anemometer )2 + (ω tachometer )2 (13)
fractional uncertainty, and Eq. (5) can be rewritten as Eq. (7) (Holman,

4
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 2. Experimental set up (a) subsonic wind tunnel with different major components (b) hexagonal mesh used at inlet (c) airfoil arrangement inside the test section
(d) Components and torque measuring arrangement of HAWT model (e) HAWT model placed at wind tunnel exit.

present computational study are shown in Fig. 3. The base airfoil was
2.4. Computational procedure selected as NACA 4415 with a chord length (C) of 100 mm, identical to
the chord length of the experimental model. The models were simulated
A comparative computational analysis was performed to capture the at a Re of 1.5 × 105 to match the experimental operating condition. As
flow physics around the airfoils. The steps followed to perform the

5
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 3. Flow chart of the present computational study.

the stall angle of unmodified NACA 4415 was found to be 12◦ (Saliveros, cross-section along the chord at mid-span (cut-plane is shown in Fig. 5
1988), a wide range of angles of attack from 0◦ to 20◦ were covered to (c)) of the airfoil blade is shown in Fig. 5 (d). The airfoil profile at the
properly investigate the impact of leading-edge modification in the cross-section (shown in Fig. 5(d)) was considered to perform the 2D
pre-stall (α = 4◦ ), stall (α = 12◦ ) and post-stall (α = 18◦ ) regions. The computational analysis to lessen the computation time and cost. Fig. 5
LEP’s amplitudes in both the modified models (SLEP and TLEP) were (e) displayed the study domain and the boundary conditions to formu­
chosen as 0.01C, similar to the fabricated models (Fig. 4). late the present study to an equivalent 2D computational study. The
upstream computational domain was considered semi-circular in shape
2.5. Computational domain and boundary conditions to replicate the physical phenomenon of inlet air accurately, and the
velocity inlet boundary condition was assigned. A rectangular down­
An undersized computational domain may include errors due to stream computational domain was maintained to capture the wake
boundary effect, whereas an oversized computational domain will generated behind the trailing edge, and a pressure outlet boundary
encompass higher computational cost. Therefore, the selection of a condition was assigned at the exit. The top and bottom walls were
proper computational domain is necessary to obtain a viable solution. assigned as a symmetry boundary condition. To diminish the boundary
The distances from the boundary walls were selected as per the sug­ effect, 15C and 6C spaces were kept from the airfoil leading-edge to
gestions of various researchers (Roy et al. (2022a); Asli et al. (2015); assign the outlet and inlet boundaries, respectively. At the solid-fluid
Skillen et al. (2015)). A 3D computational domain was initially interface, zero relative movement was specified.
considered to simulate the present airfoil blades. Fig. 5(a–d) showed the
airfoil blade with spherical leading-edge protrusion (SLEP). A

Fig. 4. Types of leading-edge of NACA 4415 airfoil (a) unmodified leading edge (ULE) (b) spherical leading-edge protrusion (SLEP) (c) triangular leading-edge
protrusion (TLEP).

6
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 5. (a) 3D computational domain (b) 3D airfoil model with spherical protrusion (c) cut-plane at half-span of SLEP model (d) cross-section of SLEP model (e) 2D
computational domain and boundary conditions.

2.6. Mesh generation and grid independency study was noticed. Until Grid-4, there was a noticeable improvement in the
coefficients. However, further mesh refining to Grid-5 resulted in a
Structured rectangular mesh near the airfoil boundary, fine un­ 0.74% improvement in the lift coefficient. It is also clear from Fig. 7 that
structured triangular mesh in the inner circle of 2C, and coarser un­ the slope of the coefficients decreased drastically after Grid-4, which
structured triangular mesh in the rest of the computational domain were indicated no significant improvement with further reduction in the
the three parts of discretization, as shown in Fig. 6(a). The element sizes element size. However, an additional computational cost was incurred
of the meshed computational domain varied within a range of 0.4 mm due to an increase in the number of elements. Thus, a grid with a min­
(minimum element size) and 10 mm (maximum element size). The first imum and maximum element size of 0.4 and 10 mm (i.e., Grid-4) was
layer was kept to a thickness of 0.01 mm near the airfoil boundary and selected as the best grid size for further computational investigations.
inflated with 12 layers at a growth rate of 1.5. The inflation zone is
visible in Fig. 6(c–f), where the leading edge of ULE (Fig. 6(c)), SLEP 2.7. Governing equation and model validation
(Fig. 6(d)), and TLEP (Fig. 6(e)) models were shown. The measurement
of non-dimensional Y+ confirms the stability of the generated mesh at The equations that governed the present simulation were (Eqs.(14)–
the airfoil boundary, and it was found to be substantially below the (21)): two-dimensional incompressible Reynolds-Averaged Navier-
expected limit (Y+≤ 1) by using the SST (Shear-Stress Transport) k-ω Stokes (RANS) equations and 4-equation Shear-Stress Transport (SST) k-
turbulence model. Fig. 6(b) demonstrates the variation of Y+ in all the ω transition turbulent model (Menter et al., 2006). The SST k-ω turbu­
three models along the chord length and found to be maximum in the lence model combines the benefits of the k-ε (used for free shear flow in
SLEP model with a value of 0.7, which is below the desired Y+ value as the far field) and k-ω models (applied near the wall because it is sus­
recommended in Versteeg and Malalasekera (2007). ceptible to the turbulence intensity). The SST k-ω transition model
Using the grid-independency test, one can determine the optimal combined with two additional equations: one for intermittency (γ) and
mesh element size for discretizing a computation’s domain of interest. another for transition onset momentum thickness Reynolds number
Using relatively coarser mesh results in erroneous findings, whilst using (Rẽθt), to describe the laminar-turbulent transition process. The inter­
finer mesh results in high processing costs. Table 1 shows the results of a mittency term is employed to activate the production term of the tur­
grid-independency test for the current computational model at Re = 1.5 bulent kinetic energy (TKE) downstream of the transition point in the
× 105. The comparison included five distinct grids, each with a different boundary layer, and the transition onset momentum-thickness Reynolds
minimum and maximum element sizes. When the grid size increased number term captures the non-local effect of the turbulence intensity.
from Grid-1 to Grid-2, a significant increase in lift and drag coefficients Chishty et al. (2011) used the Transitional SST Model to study the effect

7
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 6. Generated mesh (a) computational domain (b) Y+ values near the airfoil boundary (c) unmodified leading edge (d) spherical leading-edge protrusion (e)
triangular leading-edge protrusion (f) trailing edge of the airfoil.

Table 1
Grid independency study of NACA 4415 airfoil at α = 12◦ and Re = 1.5 × 105.
Grid Minimum element size (mm) Maximum element size (mm) Number of elements Lift Coefficient % increment Drag Coefficient % increment

Grid - 1 10.0 60 1015786 1.04 – 0.031 –


Grid - 2 4.0 40 1085799 1.16 11.54 0.038 22.58
Grid - 3 1.0 20 1127451 1.21 4.31 0.045 18.42
Grid - 4 0.4 10 1196084 1.243 2.71 0.0503 11.78
Grid - 5 0.3 5 1265297 1.252 0.74 0.0509 1.19

4-equations SST k-omega transition turbulence model in predicting


aerodynamic coefficients and flow separation process is also discussed
by Ge et al. (2019). A detailed investigation for selecting a suitable
turbulence model was reported in Roy et al. (2022b). The present
computational study was done with the help of ANSYS Fluent 14.0,
which is based on the finite volume method (FVM). The SIMPLE scheme
was used to link pressure and velocity terms in the momentum equation.
The second-order upwind scheme was used to separate the convective
terms. A small enough time step (Δt = 10− 4) was used to meet the CFL
criterion (Roy et al., 2022a). The governing equations are mentioned
below (Sreejith and Sathyabhama, 2020; Shah et al., 2015; ANSYS
Fluent 14.0, 2011):
∂ui
=0 (14)
∂xi
( ) ( )
∂(ρ ui ) ∂ ρ ui uj ∂p ∂ ∂ui
+ =− + μ − ρ u′ i u′ j (15)
∂t ∂xj ∂xi ∂xj ∂xj
Fig. 7. Grid independency study.
The symbols u andu stand for the fluctuating velocity and the mean

of dimples on low pressure turbine cascade blades and effectively pre­ velocity, respectively. t, ρ, μ, prefer to time, density, dynamic viscosity,
dicted the detachment and reattachment locations. Using the Transi­ and pressure, respectively, whereas i and j were used to refer cartesian
tional SST Model, Shah et al. (2015) numerically simulated and coordinates. The SST k- ω turbulent equations listed below were used to
investigated the process of the laminar separation bubble and the model the Reynolds stress term (ρ u′ i u′ j ) in the momentum equation:
laminar-turbulent transition over the airfoil UBD5494. As a result, the k (turbulent kinetic energy) equation:
same model is employed in the current work. The suitability of

8
S. Roy et al. Ocean Engineering 269 (2023) 113688

( )
∂(ρk) ∂(ρ k ui ) ∂ ∂k [ [( ) ] ]
+ = Γk + Gk − Yk (16) Rev
∂t ∂xi ∂xj ∂xj γ sep = 2 max − 1, 0 Freattach , 2 Fθt (20)
3.235Reθc
ω (specific dissipation rate) equation:
RT 4

∂(ρω) ∂(ρ ω ui ) ∂
(
∂ω
) where, Freattach = e− ( 20 ) and γeff = max(γ, γ sep )
+ = Γω + Gω − Yω + Dω (17) The transition model interacts with the SST turbulence model as
∂t ∂xi ∂xj ∂xj
below:
where, Gk, Gω, Yk, and Yω referred to the generation and dissipation of k ( )
∂(ρk) ∂(ρ k ui ) ∂ ∂k
and ω due to the mean velocity and turbulence, respectively. The + = Γk + G∗k − Yk∗ (21)
∂t ∂xi ∂xj ∂xj
respective effective diffusivity terms were denoted by Γk and Γω,
whereas, Dω represents the cross-diffusive term.
where, Yk∗ = min(max(γeff , 0.1) , 1.0) Yk and G∗k = γeff Gk
The intermittency term, γ is calculated solving transport equation:
The computational results of the unmodified NACA 4415 airfoil (ULE
( ) [( ) ]
∂(ργ) ∂ ρ uj γ ∂ μ ∂γ model) using the SST k-ω transition turbulence model were validated
+ = Pγ1 − Eγ1 + Pγ2 − Eγ2 + μ+ t (18) with the experimental results of Saliveros (1988) at Re = 1.5 × 105, as
∂t ∂xi ∂xj σγ ∂xj
shown in Fig. 8(a). The percentage inaccuracy was found to be within
where, P γ1, and E γ1 are the transition source terms. the acceptable range (less than 10%). A maximum 7.54% inaccuracy in
The local transition onset momentum thickness Reynolds number, lift coefficient was estimated in the computational result. It is possible to
Rẽθt, is determined by solving the following transport equation: acquire multiple angles of attack by varying the wind direction in
( ) computational analysis, but this cannot be achieved in an experimental
analysis since the wind direction is fixed in the experimental analysis.
∼ [ ]
∂ ρReθt ( ) ∼
∂ ρuj γ ∂ ∂Reθt
+ = Pθt + (μ + μt ) (19) Further, the surface roughness is also different in computational and
∂t ∂xi ∂xj ∂xj experimental studies. The discrepancy between experimental and
computational results might be due to these reasons. It can also be
where, Pθt is the blending function used to turn off the source term in the observed that the computational model has predicted less than the
boundary layer. experimental results at all the angles of attack, whereas it has projected a
The model without the correction predicts the turbulent reattach­ higher drag coefficient than the mentioned experimental drag coeffi­
ment location far downstream compared to experimental results (Men­ cient. The velocity contour of airfoil profile at the cut-plane (cut-plane
ter et al., 2006). This is because the turbulent kinetic energy (TKE), k, in shown in Fig. 5 (c)) of 3D airfoil blade is shown in Fig. 8(b), whereas,
the separating shear layer is smaller at lower free stream turbulence velocity contour of 2D SLEP model is shown in Fig. 8(c). Both figures (8
intensities. As a result, it takes longer for the TKE to grow to large (b) and 8(c)) are compared to establish the reliability of 2D model
enough values that would cause the boundary layer to reattach. To predicting flow analysis around the airfoil with leading-edge protrusions
correct this deficiency, a modification to the transition model is intro­ instead of 3D model. Although there is a small variation between the 2D
duced that allows the TKE to grow rapidly once the laminar boundary and 3D velocity distribution due to the 3D flow effect, the 2D airfoil can
layer separates. Separation-induced transition can be rewritten as, predict the aerodynamic performance and flow physics similar to
cross-section of 3D airfoil blade. Ge et al. (2019) also performed a 2D
simulation to study the airfoil withg leading-edge defects, and predicted

Fig. 8. (a) Validation of the computational model (b) velocity contour at cut-plane of 3D airofil blade (c) velocity contour of 2D SLEP model.

9
S. Roy et al. Ocean Engineering 269 (2023) 113688

the flow separation and aerodynamic performance quite well. Moreover, of hysteresis important for any experimental investigation. The differ­
the complex blade-fluid interaction, blade loading, wake distribution, ence in the prediction of airfoil’s aerodynamic forces during increasing
and vortices transportation phenomenon are not in the objectives of the and decreasing angles of attack leads to the introduction of the hyster­
present study. Therefore, the rest of the studies were carried out esis effect. Hysteresis occurred in the experimental process due to stor­
considering 2D computational domain to minimize the computational ing earlier measured values while measuring the next value. Fig. 9
time and expenses. depicts the occurrence of hysteresis in the tested airfoil models. The
angle of attack initially increases from 0◦ to 20◦ , then decreases back to
3. Results and discussion 0◦ at a small interval. The hysteresis loops have appeared clockwise in
the lift coefficient for all the studied models.
An experimental comparative study is performed for three different The formation of the hysteresis loop generally depends on the flow
types of leading-edge of the airfoil, viz. unmodified leading-edge (ULE), separation pattern. Due to early flow separation, the hysteresis loop in
leading-edge with spherical protrusion (SLEP), and leading-edge with the ULE model spread over the angle of attack (10–18)◦ , whereas the
triangular protrusion (TLEP). The lift and drag coefficients are calcu­ leading-edge modification helps to reduce the size of the hysteresis loop,
lated to compare their aerodynamic performances. In order to under­ i.e., (16–20)◦ in lift coefficient of SLEP model and (14–20)◦ in lift co­
stand the changes that occurred in the flow physics around the airfoil efficient of TLEP model. A reduction in maximum lift coefficient is
due to the modification of the leading edge, a 2D CFD analysis is carried observed in decreasing the angle of attack. For example, the maximum
out with identical models to the experimental models. The contours for lift coefficients in ULE, SLEP, and TLEP are 1.21, 1.11, and 0.78 when
pressure, velocity, streamlines, and turbulence kinetic energy around the angle of attack is increasing, whereas lift coefficients declined by
the airfoil are discussed at three different angles of attack, i.e., α = 4◦ 12.7%, 8.8%, and 13.2% at the respective angles of attack. Thus the
(pre-stall), α = 12◦ (stall), and α = 18◦ (post-stall). Following formulae SLEP modification is able to lessen the hysteresis losses in the aero­
(Eqs. (22) and (23)) are used to calculate non-dimensional aerodynamic dynamic coefficient, which increases the stability of the dynamic char­
coefficients (Roy et al., 2022a). acteristics of the airfoil (Zhang et al., 2020). As the angle of attack
increases, the flow gets separated by bursting the laminar separation
L, D
Lift, Drag coefficient, CL,D = 1 (22) bubble after the stall angle. While reducing the angle of attack, the
ρAu2
2 separated flow at the higher angle of attack cannot amend, leading to a
negative lift prediction. In contrast, the hysteresis loops have occurred
P − P∞
Pressure coefficient, CP = 1
(23) anti-clockwise direction, predicting the drag coefficient. It can be
ρAu2
2 observed that similar to the lift coefficient hysteresis loop, the spans of
the drag coefficient hysteresis loop are also decreased with the modified
3.1. Study of hysteresis effect leading-edge airfoil model.

Hysteresis is a low-Re phenomenon generally visible near the stall 3.2. Variation of aerodynamic coefficients
region that could affect the airfoil model’s performance (Biber and
Zumwalt, 1993). A continuous increment and decrement in the local The experimentally obtained lift and drag forces are presented in
angle of attack experienced by the wind turbine airfoils make the study terms of two non-dimensional coefficients: lift coefficient and drag

Fig. 9. Effect of hysteresis on the tested models (a) ULE (Model-1) (b) SLEP (Model-2) (c) TLEP (Model-3).

10
S. Roy et al. Ocean Engineering 269 (2023) 113688

coefficient. The instantaneous lift coefficients are measured at different indicating more fluctuation in the generated force. The computationally
angles of attack with the help of a three-axis load cell of 1000 sample predicted drag coefficients are compared with the experimental results
frequency. The fluctuations in the measured lift coefficient in the ULE, and found that less drag coefficient is generated computationally. A
SLEP, and TLEP at different angles of attack are shown in Fig. 10. The lift maximum absolute deviation of 10.07%, 8.39% and 9.25% were noticed
fluctuation in the ULE model is least at the pre-stall angle of attack, but it in ULE, SLEP and TLEP model, respectively. As the experimental models
increases with the increase in the angle of attack. This type of fluctuating are prepared using wood, the different surface roughness of the exper­
force is undesirable in the case of rotating power-generating devices like imental and computational models contributed to the errors. The error
wind turbines. In the SLEP model, this fluctuation is observed to be less present in the measuring instruments also increases the percentage error
than in the ULE and TLEP models at the post-stall angle of attack, in the measured lift and drag coefficients. Moreover, the present
indicating the positive use of SLEP at higher angle of attack. A similar computational analysis is performed in 2D, which adds more error to the
phenomenon was observed by Zhang et al. (2019). Thus the SLEP predicted forces than that obtained in the 3D investigation (Basumatary
modification has ameliorated the fluctuating characteristics of the airfoil et al., 2018; Sengupta et al., 2021). In comparison to Saliveros (1988),
at a post-stall angle of attack by reducing the fluctuation intensity. present wind tunnel experiment was able to predict the lift coefficient
Each experiment is conducted five times at a particular angle of and drag coefficient with a maximum absolute deviation of 15.2% and
attack to check the repeatability of the trials, and an average lift coef­ 21.6%, (see Figs. 11(e) and Fig. 12(e)). The difference in the prediction
ficient is calculated. A relative standard deviation (RSD) of the data set is may be: firstly, due to the error present in the measuring instruments,
calculated using Eq. (20) to observe the variation of the dataset from the which is addressed in uncertainty calculation and secondly, error
average value. Fig. 11(a–d) reveals the variation of lift coefficients and incurred in wind tunnel construction.
RSD with the angle of attack. The maximum lift coefficients are observed
Standard deviation (SD)
to be 1.21, 1.11, and 0.78 with ULE, SLEP, and TLEP models, respec­ Relative standard deviation (RSD) = × 100
Average (Cmean )
tively. The ULE model experiences a stall at α = 12◦ (stall angle),
whereas the SLEP and TLEP models are stalled at 18◦ and 16◦ , respec­ (20a)
tively. Thus, the stall occurrence is delayed at the expense of the ∑n √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅2
(Ci − Cmean )
maximum lift coefficient. It can be noticed that the ULE model has where, Standard deviation (SD) = i=1
n− 1 , and
∑n
outperformed the modified airfoils and remained higher till α = 16◦ . But C
Average (Cmean ) = n i=1 i
the SLEP model overcame the lift generation of the ULE model and
where, n is the times of experiment repeated at each angle of attack.
showed the highest lift coefficient among the studied airfoils at α ≥ 18◦ .
The TLEP model has remained the worst one among the studied models
at all angles of attack. TLEP model also offered the highest RSD, and a 3.3. Variation of pressure distribution
maximum value of 2.83% can be observed at α = 20◦ (Fig. 11(d)). It can
be noticed that initially, the RSD is least in the ULE model but became The pressure distributions around the airfoils are investigated by
higher than the SLEP model in the post-stall region (α ≥ 14◦ ). By using studying the pressure contour (qualitative) and pressure coefficient
the 2D CFD analysis of ULE, SLEP, and TLEP airfoils, identical to the (quantitative) plots. Fig. 13 represents the static pressure field distri­
experimental models, the computationally obtained lift coefficients are bution around the studied airfoils. The generation of the lift force is
compared with the experimental ones Fig. 11(e)). It is observed that positively influenced by the negative suction pressure and the larger
computational models have predicted higher lift coefficients with a pressure differential between the airfoil’s sides. It can be noticed that the
maximum absolute deviations of 9.66%, 13.76%, and 13.25% for ULE, point near the leading edge experiences the highest local pressure. This
SLEP, and TLEP models, respectively (Fig. 11(f)). point of maximum pressure is gradually shifted towards the bottom
It is observed from Fig. 12 that the drag coefficient increases with an surface of the airfoil as the incoming air strikes downwards with the
increase in the angle of attack, and the ULE model has the least drag increase in the angle of attack. The least local pressure also reduced as
coefficient among the studied models. This is due to the increase of the angle of attack increased, which gave a rise in the overall lift coef­
pressure drag as the angle of attack increases. The roughness induced in ficient till the stall angle. As the lift force obtained by integrating this
the SLEP and TLEP models exerts more friction, resulting in the rise of pressure difference from the leading edge to the trailing edge, there is a
friction drag. Therefore, both the modified models have a higher drag reduced lift generation in the ULE model, which has gone below the lift
coefficient than the ULE model. The maximum drag coefficients are generation in the SLEP model at α = 18◦ . Among the modified airfoils,
observed to be 0.12, 0.135, and 0.16 at α = 20◦ in the ULE, SLEP, and the SLEP model shows a lower negative pressure (− 277.8 Pa at α = 4◦ ,
TLEP models. The maximum RSD of 3.2% is observed in the TLEP model, − 415.9 Pa at α = 12◦ , and − 724.9 Pa α = 18◦ ), compare to the values of
TLEP model (− 266.4 Pa at α = 4◦ , − 409.2 Pa at α = 12◦ , and − 790 Pa α

Fig. 10. Instantaneous fluctuations of lift coefficient at different angles of attack.

11
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 11. Variation of lift coefficient at different angle of attack (a) ULE (Model-1) (b) SLEP (Model-2) (c) TLEP (Model-3) (d) relative standard deviation (e)
comparison of computational and experimental data (f) percentage error.

= 18◦ ) at all the angles of attack, making it preferable modification over e., 70%C and 54%C, respectively, at α = 18◦ . The area under the pres­
the TLEP model. The plots of the pressure coefficient will give better sure coefficient curve in the TLEP model has remained the least at all the
insight into the pressure distribution. chord lengths and angles of attack, which has produced the worst
The pressure coefficients on the plain and modified airfoils are aerodynamics performance of TLEP. Thus, in general, including LEP has
shown in Fig. 14. The curve is plotted with a reverse Y-axis to make the helped in preventing early flow separation. The separation phenomenon
plot analogous to the airfoil surfaces (the suction side on the top and the can be discussed better with the help of velocity field distribution.
pressure side on the bottom). A maximum pressure coefficient of 1 is
generated at the point of stagnation in the case of all the studied airfoils.
There is a disturbance observed in the pressure coefficient curve near the 3.4. Variation of velocity distribution
leading edge of the TLEP and SLEP model due to the small stagnation
zone created at the edges of the protrusions, although it is insignificant The velocity contours at α = 4◦ (pre-stall), 12◦ (stall), and 18◦ (post-
at the lower angle of attack (α = 4◦ ). The maximum pressure coefficient stall) for the ULE model and modified airfoils are shown in Fig. 15. A
cannot exceed this value as the maximum possible pressure in the flow zone of stagnation fluid is noticed near the leading edge, shifting to­
can be stagnation pressure. The value of the minimum pressure coeffi­ wards the pressure side of the airfoil as the angle of attack increases. A
cient increases with the increase in the angle of attack, but at a higher zero-velocity zone is created near the trailing edge, which is prominent
angle of attack, the occurrence of a separation point at a different chord at the higher angles of attack (α ≥ 12◦ ). The size of the zero-velocity
length negatively impacts the total lift generation. Among the airfoils, zone in the ULE model is much larger than the zones created in the
the pressure coefficient is minimum in the ULE at all angles of attack. SLEP and TLEP models. This is created due to the generation of the
The minimum pressure coefficient in ULE, SLEP, and TLEP models are primary circulation zone after the flow separation point. The flow sep­
observed to be − 3.84, − 2.79, and − 1.58, respectively, at α = 18◦ . The aration has occurred due to the excessive adverse pressure gradient at
pressure coefficient curve of the suction side became a horizontal flat some different chord lengths in the studied airfoil models. The circula­
line as the flow separation occurred. The presence of LSBs can be tion zone, created at different chord lengths starting from the trailing
identified by the plateau in the pressure coefficient line (Russell, 1979). edge, is named as the primary circulation zone. The extra circulation
The length of the LSBs in studied airfoils were marked in Fig. 14. It can zone is created due to the transition from laminar to turbulent near the
be noticed that the flow gets separated in the ULE model near 60%C and edges of LEPs, and after the trailing edge, is named as secondary cir­
20%C at α = 12◦ and 18◦ , respectively. In comparison, flow separation culation zone. These circulation zones can be observed in the plots of
has occurred at the SLEP and TLEP model at much later chord lengths, i. streamline distribution (Figs. 16–18). At the initial angle of attack, α =
4◦ , the spreads of the primary circulation zones are almost identical for

12
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 12. Variation of drag coefficient at different angle of attack (a) ULE (Model-1) (b) SLEP (Model-2) (c) TLEP (Model-3) (d) relative standard deviation (e)
comparison of computational and experimental data (f) percentage error.

Fig. 13. Variation of pressure contours at different angles of attack.

13
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 14. Variation of pressure coefficients at different angles of attack.

Fig. 15. Variation of velocity contours at different angle of attack.

all the airfoils, and with the rise in the angle of attack, it has progressed The zone generated within flow separation point (S) and reattach­
towards the leading edge. The SLEP and TLEP models have a secondary ment point (R) is known as laminar separation bubble (LSB) and the
circulation zone near the LEPs edges. However, this effect is insignificant nature of LSBs at the suction surface of all the airfoil models were
at the initial angle of attack, as observed in Fig. 16(b and c). It can be studied and shown in Figs. 16–18. The SLEP model experienced smaller
noticed that the size of the primary circulation zone in SLEP is weaker LSBs, whereas, relatively larger LSBs were present in ULE and TLEP
than that of TLEP at α = 12◦ (Fig. 17(b and c)) and α = 18◦ (Fig. 18(b and models. The LSBs were observed to be progressed towards the leading
c)). Due to this leading-edge circulation, the modified airfoils produced edge as the angle of attack increased. At higher angle of attack, the LSBs
a lesser minimum pressure coefficient at the leading edge, negatively busted to form larger flow separation zone in ULE and TLEP models,
impacting the lift generation, as observed in the previous section. which didn’t occur in the SLEP model. The spread of LSBs in different

14
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 16. Streamline distribution around airfoil with (a) unmodified leading-
edge (b) spherical leading-edge protrusion (c) triangular leading-edge protru­
sion at α = 4◦ .
Fig. 18. Streamline distribution around airfoil with (a) unmodified leading-
edge (b) spherical leading-edge protrusion (c) triangular leading-edge protru­
sion at α = 18◦ .

Table 2
Spread of the laminar separation bubble (LSB) along chord length.
α = 4◦ α = 12◦ α = 18◦
ULE model 0.45C–0.55C 0.25C–0.45C LSB brust
SLEP model 0.55C–0.7C 0.4C–0.55C 0.2C–0.35C
TLEP model 0.52C–0.72C 0.35C–0.55C LSB brust

gives an insight of the progression of the trailing-edge circulation zone


(primary circulation zone) and leading-edge circulation zone (secondary
circulation zone) in a clearway in addition to the velocity streamline
plots. The skin friction coefficient is investigated in this study along the
suction surface of the airfoils in order to get information about the
separation point. At the point where the skin friction coefficient is equal
to zero, the wall shear stress is also equal to zero, and as a result, the
velocity gradient is also equal to zero. The span of LSBs also can be
Fig. 17. Streamline distribution around airfoil with (a) unmodified leading-
identified from the skin friction coefficient. The skin friction coefficient
edge (b) spherical leading-edge protrusion (c) triangular leading-edge protru­ remained zero at the point of separation and reattachment. Fig. 19
sion at α = 12◦ . displays the skin friction coefficients of the tested airfoils at various
angles of attack before and after the stall. At an angle of attack of 4◦
airfoil models at varied angle of attack is shown in Table 2. At α = 18◦ , (Fig. 19(a)), it has been found that the effect of LEPs is negligible and
the ULE model produced a secondary circulation zone after the trailing flow stays connected for a similar chord length (around 88%C) in all the
edge in addition to a primary circulation zone (up to 0.2C), as shown in models (see inset view near to trailing edge). The secondary circulation
Fig. 18(a). The SLEP model has a comparatively shorter primary circu­ zones at the corner of LEPs also have an insignificant effect at α = 4◦ and
lation zone (up to 0.70C). The generation of the secondary circulation flow remained unseparated at the leading edge (see inset view near the
zone near the leading edge in the SLEP model degrades the lift genera­ leading edge). With the increase in the angle of attack, the primary
tion for most of the angles of attack, although flow remains attached for circulation zone progresses towards the leading edge and creates a larger
more distance of the chord length. Only at α ≥ 18◦ , the SLEP model can trailing edge separation bubble. Whereas the secondary circulation in­
overcome this negative effect and show a higher lift coefficient than the creases in size and creates a leading-edge separation in SLEP and TLEP
ULE model. On the other hand, the TLEP model cannot overcome the models. In ULE and TLEP model, the flow didn’t reattached once sepa­
initial loss in pressure coefficient near the leading edge due to the larger rated from the suction surface at higher angle of attack (see Fig. 19(c)),
secondary circulation zone, making it the worst performed one. which deteriorated the aerodynamic performance.
At α = 12◦ (Fig. 19(b)), it is noticed that the flow separation occurred
earlier in the ULE model (60%C) compared to SLEP (80%C) and TLEP
3.5. Variation of skin friction coefficient
(75%C) models. The flow is on the verge of separtion at the corner of
LEPs due to the presence of a secondary circulation zone in SLEP and
The flow is forced to become disassociated from the suction surface
TLEP models (see inset view near the leading edge). As the angle of
of the airfoils as a result of the low-velocity and circulation zone crea­
attack is set to 18◦ (Fig. 19(c)), the primary circulation zone reaches
tion. According to the reports in the published research, the leading-
more closer to the leading edge. Due to the secondary circulation zone
edge alteration contributes to delaying the flow separation. The inves­
near the trailing edge, the ULE model experienced a larger trailing-edge
tigation of the skin friction coefficient along the suction line of the airfoil

15
S. Roy et al. Ocean Engineering 269 (2023) 113688

Thus the SLEP model provides better aerodynamic performances than


the TLEP model. The re-energized boundary layer by LEPs helped flow to
remain connected for a greater chord-length, which boosts the perfor­
mance of the SLEP model at higher angles of attack.

3.6. Variation of turbulence kinetic energy distribution

Turbulent flows carry parcels with extremely transitory properties


known as eddies, which are part of the flow structure. The total amount
of energy contained in eddies in turbulent flows is known as turbulence
kinetic energy (TKE). The presence of high TKE values indicates the
presence of high turbulent energy extraction rates from mean flow. The
generation of turbulent kinetic energies in ULE, SLEP, and TLEP at
different angles of attack are shown in Fig. 20. The LEPs induced by the
flow become a cause of turbulence production, which is then transported
downstream and generates eddies of various sizes. Large eddies occurred
in the low-pressure zone, able to extract more kinetic energy from the
mean flow. It can be noticed that the ULE model has larger eddies with
more turbulent kinetic energy. The early transition at 20%C can be
observed in the ULE model, which indicates the early occurrence of flow
separation at a higher angle of attack (α = 18◦ ). However, it is almost
identical for all the airfoils at the initial angle of attack (α = 4◦ ). High
TKE indicates a transition at the edges of the LEPs, which can be seen in
TKE contours for the SLEP and TLEP models at α = 18◦ . But the flow gets
connected after the edges of LEPs and separated again at a chord length
later than the ULE model.

3.7. Application of LEP on small wind turbine model

The relative velocity component at the leading edge of the turbine


blade changes from root to tip, which alters the local angle of attack.
Moreover, the changes in incoming wind velocity also change the local
angle of attack at the turbine blade’s leading edge. As discussed in the
above sections, the SLEP model shows better aerodynamic performances
than the ULE model at higher angle of attack; it can work satisfactorily
over a wide range of angles of attack, delaying the stall occurrence.
Therefore, HAWT models were constructed using ULE and SLEP airfoil
models. The performance of HAWT model is discussed in terms of power
coefficient for different operating conditions. Power coefficient in­
dicates the turbine’s ability to convert the available power in the
incoming wind to the output power. As the blades used in the experi­
mental investigations are un-tapered and untwisted and not optimized,
there is a need to study the effect of pitch angle (β) on the turbine per­
formance to find the optimum pitch angle for the studied design and
operating conditions. As the current blades with β = 0◦ provide more
drag due to the larger frontal area, it is unable to rotate at low wind
velocity. Therefore, four different pitch angles in the range of (5◦ –20◦ )
were considered in the present investigation to place the turbine blade at
different orientations. The incoming wind velocity towards the HAWT
model varied in a low-velocity range available at built-in environmental
conditions; from 4 m/s to 7 m/s. The wind turbine rotor is placed after a
Fig. 19. Skin-friction coefficient along the suction line of the airfoil at (a) α = distance of 3m (1.5m inside the tunnel + 1.5m after the tunnel exit) from
4◦ (b) α = 12◦ (c) α = 18◦ . the fan. Therefore, the swirl induced by the fan will be weakened before
it reaches the turbine rotor. Similar methodology was followed by
separation bubble expanded up to 20%C and observed a reduced aero­ Sengupta et al. (2022) and Mazarbhuiya et al. (2020) in their experi­
dynamic performance than the SLEP model. On the other hand, the flow mental investigations. To define the wind flow field, wind velocities are
remained attached to the suction surface in the SLEP and TLEP models measured at different location across the height and width of the wind
until 70%C and 54%C, respectively, even though these models were turbine rotor. The wind velocity field at the upstream of the wind tur­
operating at a higher angle of attack. The secondary circulation zone bine rotor is shown in Fig. 21 to verify the flow uniformity. The root
present in the SLEP and TLEP models moves towards the trailing edge mean square (RMS) velocities at each velocity are represented by
with an increase in angle of attack and creates a larger leading-edge straight lines against each velocity.
separation bubble. The leading-edge separation bubble expands to Fig. 22 shows the effect of pitch angle and wind velocity on the
0.8%C in the TLEP model and experiences a complete flow separation. power coefficient for both the HAWT models. The power coefficient in
Whereas it expands up to 0.4%C in the SLEP model, but flow does not get all the cases followed a similar trend, as it increased up to a certain tip
separated completely (see inset view near the leading edge in Fig. 19(c)). speed ratio and then decreased. It can be observed that with the increase
in pitch angle, the maximum power coefficient shifted towards the lower

16
S. Roy et al. Ocean Engineering 269 (2023) 113688

Fig. 20. Distribution of turbulence kinetic energy at different angles of attack.

Fig. 21. Wind velocity distribution at upstream of wind turbine model (a) across the height of the rotor (h = variable height) (b) across the width of the rotor (w =
variable width).

Fig. 22. Effect of pitch angle and wind velocity on the power coefficient of HAWT models.

17
S. Roy et al. Ocean Engineering 269 (2023) 113688

TSR. It indicates that blade with higher pitch angle provides better • At α ≥ 18◦ , the SLEP model outperformed the ULE model. The
performance at lower rotational speed of turbine. On the other hand, the computational study revealed that due to a smaller primary circular
blade with lower pitch angle provides better performance at higher TSR. zone at the airfoil’s trailing edge, the redesigned airfoil with the SLEP
Therefore, the performance characteristics can be divided into two saw a 50% delay in stall onset. However, the drag coefficient of the
sections: (i) blade with higher pitch angle performs better at lower TSR SLEP model was 8.5% greater than the ULE model at α = 20◦ due to
and (ii) blade with lower pitch angle performs better at higher TSR. For higher friction drag.
all the cases the maximum power coefficient is achieved with β = 15◦ . • The computational analysis explained that the ULE model provided
Similar observations found in SLEP model with higher magnitude of the lowest suction pressure coefficient (− 3.84 at α = 18◦ ), but the
power coefficient. The power coefficient for both the ULE and SLEP total pressure differential between the airfoil surfaces decreased due
HAWT model increases due to the increase in torque when the wind to flow separation at 20% of the chord length, resulting in a loss in
velocity increases from 4 m/s to 5 m/s. But further increment in wind aerodynamic performance.
velocity deteriorates the power coefficient for all the studied cases. At • The computational study shows that the span of the primary circu­
higher wind velocity the blades rotate with higher rotational speed lation zone in the SLEP and TLEP models was reduced compared to
makes it similar to a solid disk resulted in a reduced torque generation. the ULE model. At α = 18◦ , the primary circulation zone in the ULE
The maximum power coefficients achieved in ULE and SLEP are model was up to 20%C, whereas the spans were upto 70%C and 54%
compared in Fig. 23. It is observed that the SLEP HAWT model achieved C in SLEP and TLEP models, respectively.
higher power coefficient than the ULE HAWT model under all the • The computational investigation identified that a larger secondary
operating conditions. The maximum power coefficient of 0.2257 is circulation zone was created after the trailing edge of the ULE model
achieved with SLEP HAWT model at TSR 1.98 under the operating at α ≥ 18◦ , whereas the SLEP and TLEP model experienced a smaller
condition of pitch angle 15◦ and wind velocity 5 m/s. In comparison, the secondary circulation zone at the LEP’s edges. Further, the secondary
ULE HAWT model shows the maximum power coefficient of 0.2072 at circulation zone was more prominent in the TLEP model compared to
TSR 1.92, which is 8.2% less than the SLEP HAWT model. Thus the SLEP model, which resulted in poor aerodynamic performance in
application of SLEP on the un-tapered and untwisted HAWT model TLEP model.
provides enhanced performance under the low wind velocity condition. • The experimental study on wind turbine rotor models revealed that
the turbine with higher blade pitch angle could perform better at
4. Conclusions lower TSR, whereas lower blade pitch angle achieved higher per­
formance at higher TSR. The SLEP HAWT model achieved higher
The effects of different LEPs (spherical and triangular) on the aero­ power coefficient than the ULE HAWT model under all the studied
dynamic performance of the NACA 4415 airfoil were experimentally conditions. The maximum power coefficient of 0.2257 was achieved
investigated. A constant amplitude of 0.01C was maintained as the by SLEP HAWT model at TSR 1.98 under the operating condition of
height of both the LEPs. The study was carried out at a low-Re = 1.5 × 15◦ pitch angle and wind velocity of 5 m/s, which is 8.2% more than
105 in the angle of attack ranges 0◦ –20◦ . The computational study was that achieved by ULE HAWT model.
performed to visualize the flow characteristics around the airfoil models
to obtain the necessary aerodynamic performance insights. The flow The pressure field around the airfoil was inspected computationally
physics were compared at three different angles of attack: at α = 4◦ (pre- in the present study due to the experimental set-up limitation. Pressure
stall), α = 12◦ (stall), and α = 18◦ (post-stall). The small-scale HAWT ports on the airfoil can be installed in future investigations to get the
with un-tapered and untwisted SLEP and ULE blades were experimen­ pressure field around the fabricated experimental airfoil models. The
tally investigated for different pitch angles and wind velocities. position of the protrusion can be altered and placed on the airfoil’s
Following major concluding remarks were drawn from the detailed suction and pressure surface. The present wind turbine experimental
experimental and computational analysis. investigation was limited to the application of SLEP on the un-tapered
and untwisted blades due to ease in construction, which will be
• The experimental study has revealed that the span of the hysteresis extended with the tapered and twisted blades.
loop was reduced by 50% in the SLEP model and 25% in the TLEP
model compared to the ULE model, which indicated the improved CRediT authorship contribution statement
dynamic stability of the SLEP model in low-Re flow.
• The time-dependent lift generation in the experimental trials indi­ Sujit Roy: Data curation, Validation, Formal analysis, Investigation,
cated that the SLEP model fluctuates less at a higher angle of attack Software, Writing – original draft. Biplab Das: Conceptualization,
than the ULE model. This ensured the stable force generation in the Writing – review & editing. Agnimitra Biswas: Visualization, Meth­
SLEP model at the higher angle of attack, which is a favorable con­ odology, Writing – review & editing.
dition for power generation.

Fig. 23. Effect of pitch angle on the maximum power coefficient at different wind velocity.

18
S. Roy et al. Ocean Engineering 269 (2023) 113688

Declaration of competing interest Acknowledgment

The authors declare that they have no known competing financial The authors sincerely thank the Department of Mechanical Engi­
interests or personal relationships that could have appeared to influence neering, NIT Silchar, India, for providing the computational and
the work reported in this paper. experimental facility.

Data availability

Data will be made available on request.

Nomenclatures

A Area of airfoil blade, m2


C Chord length, m
CD Drag coefficient
CL Lift coefficient
Cmeans Average lift and drag coefficient
CP Pressure coefficient
Cp Power coefficient
Ct Torque coefficient
D Drag force, N
d Rotor diameter, m
k Turbulent kinetic energy, m2/s2
L Lift force, N
P Static pressure, Pa
P∞ Atmospheric pressure, Pa
Re Reynolds number
t time, s
u Inflow wind velocity, m/s
u Fluctuating velocity, m/s

u Mean velocity, m/s

Greek Letters
α Angle of attack, deg
Ω Angular velocity, rad/s
ρ Density, kg/m3
Γ Effective diffusivity, m2/s
ω Specific dissipation rate, J/kg.s
λ Tip speed ratio
μ Viscosity, Pa.s
τw Wall shear stress, Pa

Abbreviations
HAWT Horizontal axis wind turbine
LEP Leading-edge protrusion
LESSP Leading-edge sinusoidal protuberance
RSD Relative standard deviation
SD Standard deviation
SLEP Spherical leading-edge protrusion
TKE Turbulence kinetic energy
TLEP Triangular leading-edge protrusion
TSR Tip speed ratio
ULE Unmodified leading edge

References Arunvinthan, S., Pillai, S.N., Cao, S., 2020. Aerodynamic characteristics of variously
modified leading-edge protuberanced (LEP) wind turbine blades under various
turbulent intensities. J. Wind Eng. Ind. Aerod. 202, 104188.
Abate, G., Mavris, D.N., 2018. Performance analysis of different positions of leading edge
Asli, M., Mashhadi Gholamali, B., Mesgarpour Tousi, A., 2015. Numerical analysis of
tubercles on a wind turbine blade. In: Wind Energy Symposium 2018. American
wind turbine airfoil aerodynamic performance with leading edge bump. Math. Probl
Institute of Aeronautics and Astronautics Inc, AIAA.
Eng. 2015, 493253.
Abate, G., Mavris, D.N., Sankar, L.N., 2019. Performance effects of leading edge tubercles
Bai, C.J., Wang, W.C., Chen, P.W., 2016. The effects of sinusoidal leading edge of turbine
on the NREL phase VI wind turbine blade. J. Energy Resour. Technol. 141 (5),
blades on the power coefficient of horizontal-axis wind turbine (HAWT). Int. J.
051206.
Green Energy 13 (12), 1193–1200.
Aftab, S.M.A., Ahmad, K.A., 2017. CFD study on NACA 4415 airfoil implementing
spherical and sinusoidal Tubercle Leading Edge. PLoS One 12 (11), 0188792.

19
S. Roy et al. Ocean Engineering 269 (2023) 113688

Basumatary, M., Biswas, A., Misra, R.D., 2018. CFD analysis of an innovative combined Roy, S., Biswas, A., Das, B., Reddy, B.V., 2022a. Flow control of a wind-turbine airfoil
lift and drag (CLD) based modified Savonius water turbine. Energy Convers. Manag. with a leading-edge spherical dimple. Int. J. Green Energy 1–19.
174, 72–87. Roy, S., Das, B., Biswas, A., 2022b. A comprehensive review of the application of bio-
Bhuyan, S., Biswas, A., 2014. Investigations on self-starting and performance inspired tubercles on the horizontal axis wind turbine blade. Int. J. Environ. Sci.
characteristics of simple H and hybrid H-Savonius vertical axis wind rotors. Energy Technol. 1–28.
Convers. Manag. 87, 859–867. Russell, J.M., 1979. Length and bursting of separation bubbles: a physical interpretation.
Biber, K., Zumwalt, G.W., 1993. Hysteresis effects on wind tunnel measurements of a In: Science and Technology of Low Speed Motorless Flight, vol. 2085. NASA
two-element airfoil. AIAA J. 31 (2), 326–330. Conference Publication.
Butt, U., Hussain, S., Schacht, S., Ritschel, U., 2021. Experimental Investigations of Flow Saliveros, E., 1988. The Aerodynamic Performance of the NACA-4415 Aerofoil Section at
over NACA Airfoils 0021 and 4412 of Wind Turbine Blades with and without Low Reynolds Numbers. Doctoral Dissertation, University of Glasgow, Glasgow, UK.
Tubercles. Wind Engineering, 0309524X211007178. Schreck, S.J., Sørensen, N.N., Robinson, M.C., 2007. Aerodynamic structures and
Butterfield, C.P., Scott, G., Musial, W., 1992. Comparison of wind tunnel airfoil processes in rotationally augmented flow fields. Wind Energy: An International
performance data with wind turbine blade data. J. Sol. Energy Eng. 114 (2), Journal for Progress and Applications in Wind Power Conversion Technology 10 (2),
119–124. 159–178.
Chen, T.Y., Liou, L.R., 2011. Blockage corrections in wind tunnel tests of small Sengupta, A.R., Biswas, A., Gupta, R., 2016. Studies of some high solidity symmetrical
horizontal-axis wind turbines. Exp. Therm. Fluid Sci. 35 (3), 565–569. and unsymmetrical blade H-Darrieus rotors with respect to starting characteristics,
Chishty, M.A., Parvez, K., Ahmed, S., Hamdani, H.R., Mushtaq, A., 2011. Transition dynamic performances and flow physics in low wind streams. Renew. Energy 93,
prediction in low pressure turbine (LPT) using gamma theta model and passive 536–547.
control of separation. ASME International Mechanical Engineering Congress and Sengupta, A.R., Biswas, A., Gupta, R., 2021. Aerodynamic analysis of cambered blade H-
Exposition 54877, 193–200. Darrieus rotor in low wind velocity using CFD. Wind Struct. 33 (6), 471–480.
Corbus, D., Hansen, A.C., Minnema, J., 2006. Effect of blade torsion on modeling results Sengupta, A.R., Kumar, Y., Biswas, A., Gupta, R., 2022. Performance investigation of
for the small wind research turbine (SWRT). J. Sol. Energy Eng. (4), 481–486. cavity shaped blade on H-Darrieus wind turbine in built environmental condition.
Fan, M., Dong, X., Li, Z., Sun, Z., Feng, L., 2022. Numerical and experimental study on Energy Sources, Part A Recovery, Util. Environ. Eff. 1–17.
flow separation control of airfoils with various leading-edge tubercles. Ocean Eng. Shah, H., Mathew, S., Lim, C.M., 2015. Numerical simulation of flow over an airfoil for
252, 111046. small wind turbines using the γ-Reθ model. International Journal of Energy and
Fish, F.E., Battle, J.M., 1995. Hydrodynamic design of the humpback whale flipper. Environmental Engineering 6 (4), 419–429.
J. Morphol. 51–60. Singh, M., Biswas, A., Misra, R., 2015. Investigation of self-starting and high rotor
Fluent, Ansys, 2011. Fluent 14.0 User’s Guide. ANSYS FLUENT Inc. solidity on the performance ofa three S1210 blade H-type Darrieus rotor. Renew.
Freere, P., Sacher, M., Derricott, J., Hanson, B., 2010. A low cost wind turbine and blade Energy 76, 381–387.
performance. Wind Eng. 34 (3), 289–302. Siram, O., Sahoo, N., 2022. Performance assessment of straight and linearly tapered
Gawad, A.A., 2013. Utilization of whale-inspired tubercles as a control technique to rotors through wind tunnel investigation for off-grid applications. Wind Eng. 46 (4),
improve airfoil performance. Transaction Series on Engineering Sciences and 1291–1310.
Technologies 2 (5), 212–218. Skillen, A., Revell, A., Pinelli, A., Piomelli, U., Favier, J., 2015. Flow over a wing with
Ge, M., Zhang, H., Wu, Y., Li, Y., 2019. Effects of leading edge defects on aerodynamic leading-edge undulations. AIAA J. 53 (2), 464–472.
performance of the S809 airfoil. Energy Convers. Manag. 195, 466–479. Sreejith, B.K., Sathyabhama, A., 2020. Experimental and numerical study of laminar
Gopinathan, V.T., Rose, J.B.R., 2021. Aerodynamics with state-of-the-art bioinspired separation bubble formation on low Reynolds number airfoil with leading-edge
technology: tubercles of humpback whale. Proc. IME G J. Aero. Eng. 235 (16), tubercles. J. Braz. Soc. Mech. Sci. Eng. 42, 1–15.
2359–2377. Sudhakar, S., Karthikeyan, N., 2013. Effect of leading edge tubercles on flow field over
Gopinathan, V.T., Rose, J.B.R., Surya, M., 2020. Investigation on the effect of leading NACA-4415 airfoil at low Reynolds number. In: 14th Asian Congress of Fluid
edge tubercles of sweptback wing at low Reynolds number. Mechanics & Industry 21 Mechanics. Hanoi and Halong, Vietnam.
(6), 621. Versteeg, H.K., Malalasekera, W., 2007. An Introduction to Computational Fluid
Hansen, K.L., Kelso, R.M., Dally, B.B., 2011. Performance variations of leading-edge Dynamics: the Finite Volume Method. Pearson Education.
tubercles for distinct airfoil profiles. AIAA J. 49 (1), 185–194. Wei, Z., New, T.H., Cui, Y.D., 2015. An experimental study on flow separation control of
Holman, J.P., 2012. Experimental Methods for Engineers, Eight Ed. McGraw-Hill, New hydrofoils with leading-edge tubercles at low Reynolds number. Ocean Eng. 108,
York. 336–349.
Hsiao, F.B., Bai, C.J., Chong, W.T., 2013. The performance test of three different Weihs, D., 1981. Effects of swimming path curvature on the energetics of fish motion.
horizontal axis wind turbine (HAWT) blade shapes using experimental and Fish. Bull. 79, 171–176.
numerical methods. Energies 6 (6), 2784–2803. Yan, Y., Avital, E., Williams, J., Cui, J., 2021. Aerodynamic performance improvements
Huang, G.Y., Shiah, Y.C., Bai, C.J., Chong, W.T., 2015. Experimental study of the of a vertical axis wind turbine by leading-edge protuberance. J. Wind Eng. Ind.
protuberance effect on the blade performance of a small horizontal axis wind Aerod. 211, 104535.
turbine. J. Wind Eng. Ind. Aerod. 147, 202–211. Yi-Nan, Z., Hui-Jing, C., Ming-Ming, Z., 2021. A calculation method for modeling the
Johari, H., Henoch, C., Custodio, D., Levshin, A., 2007. Effects of leading-edge flow characteristics of the wind turbine airfoil with leading-edge protuberances.
protuberances on airfoil performance. AIAA J. 45 (11), 2634–2642. J. Wind Eng. Ind. Aerod., 104613
Joseph, J., Sathyabhama, A., 2022. Leading edge tubercle on wind turbine blade to Zadorozhna, D.B., Benavides, O., Grajeda, J.S., Ramirez, S.F., de la Cruz May, L., 2021.
mitigate problems of stall, hysteresis, and laminar separation bubble. Energy A parametric study of the effect of leading edge spherical tubercle amplitudes on the
Convers. Manag. 255, 115337. aerodynamic performance of a 2D wind turbine airfoil at low Reynolds numbers
Kim, M.J., Yoon, H.S., Jung, J.H., Chun, H.H., Park, D.W., 2012. Hydrodynamic using computational fluid dynamics. Energy Rep. 7, 4184–4196.
characteristics for flow around wavy wings with different wave lengths. Int. J. Nav. Zhang, R.K., Wu, V.D.J.Z., 2012. Aerodynamic characteristics of wind turbine blades
Archit. Ocean Eng. 4, 447–459. with a sinusoidal leading edge. Wind Energy 15 (3), 407–424.
Kline, S.J., McClintock, F.A., 1953. Describing uncertainties in single-sample Zhang, M.M., Wang, G.F., Xu, J.Z., 2013. Aerodynamic control of low-Reynolds-number
experiments. Mech. Eng. 3. airfoil with leading-edge protuberances. AIAA J. 51 (8), 1960–1971.
Koca, K., Genc, M.S., Acikel, H.H., Cagdas, M., Bodur, T.M., 2018. Identification of flow Zhang, M.M., Wang, G.F., Xu, J.Z., 2014. Experimental study of flow separation control
phenomena over NACA 4412 wind turbine airfoil at low Reynolds numbers and role on a low-Re airfoil using leading-edge protuberance method. Exp. Fluid 55, 1710.
of laminar separation bubble on flow evolution. Energy 144, 750–764. Zhang, Y., Zhang, M., Cai, C., 2019. Flow control on wind turbine airfoil affected by the
Mazarbhuiya, H.M.S.M., Biswas, A., Sharma, K.K., 2020. Effect of blade attachments on surface roughness using leading-edge protuberance. J. Renew. Sustain. Energy 11
the performance of an asymmetric blade H-Darrieus turbine at low wind speed. (6), 063304.
Energy Sources, Part A Recovery, Util. Environ. Eff. 1–18. Zhang, Y.N., Zhang, M.M., Cai, C., Xu, J.Z., 2020. Aerodynamic load control on a
Menter, F.R., Langtry, R., Völker, S., 2006. Transition modelling for general purpose CFD dynamically pitching wind turbine airfoil using leading-edge protuberance method.
codes. Flow, Turbul. Combust. 77 (1), 277–303. Acta Mech. Sin. 36 (2), 275–289.
Miklosovic, D.S., Murray, M.M., Howle, L.E., Fish, F.E., 2004. Leading-edge tubercles Zhang, Y., Zhang, X., Yi, L.I., Chang, M., Jiakuan, X.U., 2021. Aerodynamic performance
delay stall on humpback whale (Megapteranovaeangliae) flippers. Phys. Fluids 16 of a low-Reynolds UAV with leading-edge protuberances inspired by humpback
(5), 39–42. whale flippers. Chin. J. Aeronaut. 34 (5), 415–424.
Mishra, A., De, A., 2021. Investigation of Passive Flow Control over an Airfoil Using Zhou, Y., Hou, L., Huang, D., 2017. The effects of Mach number on the flow separation
Leading Edge Tubercles arXiv preprint arXiv:2103.08854. control of airfoil with a small plate near the leading edge. Comput. Fluid 156,
274–282.

20

You might also like