ISSN 0015-4628, Fluid Dynamics, 2020, Vol. 55, no. 5, pp. 657–669. © Pleiades Publishing, Ltd., 2020.

Russian Text © The Author(s), 2020, published in Izvestiya RAN. Mekhanika Zhidkosti i Gaza, 2020, Vol. 55, No. 5, pp. 83–96.

Investigation of Variable Spanwise Waviness Wavelength Effect

on Wing Aerodynamic Performance
I. A. Tunioa,*,**, D. Kumarb,c, T. Hussaina, M. Jatoid, and Safiullahe
a Departmentof Mechanical Engineering, Mehran UET Jamshoro, Sindh, 76090 Pakistan
b Dept.
of Mechanical Engineering, MUET, Shaheed Zulfiqar Ali Bhutto Campus,
Khairpur Mir’s, Sindh, 66020 Pakistan
c
Centre for Smart Infrastructure & Digital Construction Department of Civil and Construction Engineering,
Faculty of Science, Engineering and Technology, Swinburne University of Technology,
Hawthon-3122, Melbourne, Australia
d
Department of Electrical Engineering Technology, The Benazir Bhutto Shaheed University of Technology& Skill
Development (BBSUTSD), Khairpur Mirs, Sindh, 66020 Pakistan
e Department of Mechanical System Engineering, Hiroshima University, Hiroshima, Japan

*e-mail: intizar.tunio@faculty.muet.edu.pk
**e-mail: intizar_tunio@hotmail.com
Received August 21, 2019; revised November 19, 2019; accepted December 17, 2019

Abstract—Tubercles on flippers of a humpback whale offer a hydrodynamic advantage. Previously,

multiple attempts were made to employ them on airplane wings to get an aerodynamic advantage. The
effects of the airfoil thickness, the wing planform, the Reynolds number Re, and the wavelength and
amplitude are actively investigated to understand their impact on the wavy wing aerodynamic perfor-
mance and the flow mechanism. However, the effect of varying wavelength along the span of the finite
wing and its flow physics is yet to be explored. This research work aims at investigating the effect of
varying waviness wavelength along the finite wingspan and its underlying flow mechanism in the pre-
stall and post-stall regimes. The wavy wing models are designed using the NACA0021 airfoil and the
simulation results are compared with the smooth leading edge (named as baseline model) at Reynolds
number Re = 1.2 × 105. The simulation results were validated against experimental results from the
literature. Two different wavy wing models, i.e., with increasing wavelength from root to tip of the wing
(λ0305h1) and decreasing wavelength from root to tip of the wing (λ0503h1) were simulated. From the
aerodynamic force behavior it is estimated that in the pre-stall regime the maximum reduction in the
lift-to-drag (L/D) ratio is 16.89% and 4.22% for λ0503h1 and λ0305h1, respectively. However, in the
post-stall regime the maximum increase in the L/D ratio is estimated as 2.97% and 19.18% for
λ0503h1 and λ0305h1, respectively, at 20o angle of attack. It is observed that the λ0503h1 model has
a lower L/D ratio in the post-stall regime due to the vortices produced on the wing surface. These vor-
tices create a flow recirculation zone over the wing which causes an increase in the pressure at the
upper surface of the wing. On the basis of the obtained results, it is concluded that the spanwise wav-
iness on aircraft wing is only beneficial in the post-stall regime. It is also concluded that an increase in
the waviness wavelength from the root to the tip provides higher aerodynamic advantage than for the
case with decreasing waviness wavelength toward the tip.

Keywords: passive flow control, aerodynamic performance, heterogeneous leading-edge waviness,

waviness wavelength, waviness amplitude, finite wing
DOI: 10.1134/S0015462820040102

The use of wings with a wavy leading edge as a way of obtaining improved aerodynamic performance
started gaining attention after the work of Fish and Battle [1], where the shape of humpback’s pectoral
flipper was analyzed from the standpoint of hydrodynamic performance. The flippers have tubercles on
leading edge, which can maintain lift at higher angles of attack (AoA) and delay stall along with efficient
maneuverability. In that study it was suggested that the function of the tubercles is similar to that of the
vortex generator on an aircraft wing. In the subsequent study [2] the three-dimensional panel method was
employed to compare the performance of smooth and wavy leading-edge wings at an incident angle of 10o.
The results showed an increment of 4.8% in the lift and a decrement of 10.9% in the drag in the pre-stall
regime, which further supported the work of [1]. Later, in [3] an experimental analysis on the idealized

657
658 TUNIO et al.

model of humpback whale flipper was performed and its results were compared with a smooth leading
edge model. The results revealed that the idealized humpback whale flipper model achieved an increase
of 6% in the maximum lift with 40% increase in the stall angle. Also, the study showed that the tubercles
on the wing leading edge cause reduction in the drag force on the wing for 11° to 18° AoA. In the subse-
quent study [4] experiments were performed to analyze and compare the aerodynamic performance of the
idealized model of humpback whale for full-span and half-span wings with and without leading-edge pro-
tuberances. The experimental results indicated that for the full-span model the lift force decreases and the
drag increases, whereas in the case of the half-span model the opposite behavior was noticed. To investi-
gate the 3D effect in [5] a series of experiments was performed, where it was found that the 3D effects are
significant in analyzing the hydrodynamic performance of the wing leading edge. Afterward, in [6] an
experimental analysis was performed to investigate the 3D effects of the tubercle geometry. The experi-
mental results indicated that the 3D effects are less significant than originally thought, however the tuber-
cle amplitude, the wavelength, and the Reynolds number can play a significant role.
The flow mechanisms responsible for the described aerodynamics were thought to be similar to that of
a vortex generator that increases the momentum within the boundary layer and delays flow separation in
the post-stall regime [3, 7–10]. However, in [11] it was suggested that the waviness wavelength and ampli-
tude are much higher than the boundary layer thickness; thus, tubercles cannot be regarded as vortex gen-
erators. It was proposed that nonuniform downwash causes restriction to flow separation behind tubercle
peaks, whereas the author argued that the chord length in the trough is lower, thus having greater stream-
wise pressure gradient resulting flow separation behind the trough. Studies concluded that the flow mech-
anism responsible for prescribed aerodynamic behavior still needs further explanation; however, in the
present study chord length is kept constant along the span; the only transformation was made.
An experimental analysis designed to analyze the effect of the tubercle wavelength and amplitude was
performed in [12, 13]. The NACA 634-021 airfoil was selected and the effects of wavelength and amplitude
were analyzed within the range from 0.25c to 0.5c and from 0.025c to 0.12c, respectively. The hydrody-
namic performance of smooth leading edge hydrofoil was compared with wavy leading edge hydrofoil.
The results indicated that the tubercles cause a decrease in the performance for the pre-stall regime but
lead to a 50% increase in the lift for the post-stall regime, as compared to the baseline model. In [14] the
waviness geometry effect was investigated at Re = 800 and 20°AoA using numerical simulation. The lead-
ing edge waviness wavelength and amplitude varied within a certain range. The simulation results showed
that the maximum drag reduction of 35% was observed, together with the reduction in the lift force, com-
pared to the baseline smooth-leading-edge model. Additionally, it was found that variations in the tuber-
cle wavelength and amplitude affect the recirculation zone, the wake topology, and the strength of wake
vortices. Their study showed that for amplitude-to-chord ratios A/c < 0.07 the vortex intensity is too low
to prevent flow separation. However, at A/c > 0.07 wake vortex shedding was weakened. The LES per-
formed by Kim and co-workers attempted to investigate the underlying flow mechanism responsible for
the aerodynamic advantage of an having a sinusoidal wavy leading edge. The numerical simulation was
performed in time-averaged and unsteady fluid domains at Reynolds number Re = 1.2 × 105 and 20° angle
of attack [15]. A recent study analyzed the tubercle geometry effect through direct numerical simulation
(DNS) at Reynolds number Re = 1000 for an infinite wing [16]. The simulation results revealed that at
this Re the wavy wing with a wavelength greater than 0.5c experience a reduction in the L/D ratio, whereas
an increase in the amplitude further reduces the L/D ratio. In addition, the study also found that the wavy
wing with a shorter tubercle wavelength showed a negligible effect. Studies suggested that the tubercle
parameters have a considerable effect on the wing aerodynamics, as well as on the governing flow mech-
anism. A similar assertion was made in a recent study [17].
Numerous studies showed that the flow behavior (i.e., recirculation zone, wake region, vortex
strength) is mainly governed by the tubercle geometry and the Reynolds number [14, 18]. To investigate
the Reynolds number effect on the flow mechanism DNS was performed in [19] at Re = 10000 and
50000, where it was concluded that the flow mechanism noticed at Re = 1000 in the previous study [16]
is still important at the higher Reynolds numbers. These results suggest that the effect of tubercle param-
eters is still important to get more insight into the flow mechanism.
However, most of the previous studies were conducted at variable chord lengths (i.e., having lower
chord lengths in the trough section) and the flow mechanism was discussed from precisely this standpoint [11].
Additionally, numerous studies investigated the effect of the tubercle wavelength and amplitude at differ-
ent AoAs for particular wavelength and amplitude values rather than for the case of variable (along the
span) waviness wavelength. However, in reality tubercles on humpback flippers are highly heterogeneous.
The primary objective of this research is to investigate the effect of variable waviness wavelength along
the span on the aerodynamic efficiency of a wing and the flow mechanics in the pre- and post-stall

FLUID DYNAMICS Vol. 55 No. 5 2020

 INVESTIGATION OF VARIABLE SPANWISE WAVINESS WAVELENGTH EFFECT 659

Table 1. Specification of the waviness for different analyzed cases

% of c increase/decrease
Models h/c ratio λ/c from root to tip along span in each
wavelength
Baseline model 0 – –
λ0503h1 0.1 0.5–0.3 0.05
λ0305h1 0.1 0.3–0.5 0.05

regimes. The study was performed with two wavy wing models, i.e., the λ0305h1 wing with increasing
wavelength from root to tip and the λ0503h1 wing decreasing wavelength from root to tip of the wing.
Moreover, the underlying flow mechanism responsible for the aerodynamic behavior is also described.

1. NUMERICAL SCHEME
1.1. Description of Physical Problem
A heterogeneous spanwise waviness is applied on a wing for which the NACA0021 airfoil is selected.
The humpback profile has approximately 21% of the maximum thickness at 40% of its chord [1]. The
tubercle model is produced by applying the coordinate transformation given by the equation

2 λ( )
X = c + ξ(Z ) = c − h cos 2π z . (1.1)

Figure 1a shows the geometry obtained by applying Eq. (1.1), while the coordinate transformation is
achieved at the trailing edge without affecting the chord length of the wing. The coordinate transformation
results in a wing having deformed leading and trailing edges equally, whereas in most of the previous stud-
ies only leading-edge treatment was considered. While dealing with the tubercle specification, it is import-
ant to consider the wavelength and amplitude within the range of humpback flaps. The previous studies
measured that for an adult humpback whale the peak-to-peak distance lies within the range from 0.25c to
0.5c, while the distance from the crest to the trough (amplitude) is within 0.05c to 0.1c [12, 20]. The crit-
ical analysis of humpback whale flippers revealed that the tubercles on the flipper are highly heteroge-
neous in nature. Additionally, the literature also reported that minor attention was given to variations in
the wavelength along the span. For this reason, in this study we consider an almost full range of the wave-
lengths and maximum amplitudes typical of an adult humpback whale, the more so that in certain studies
it was suggested that lower tubercle amplitudes have not shown a favorable effect on the wing perfor-
mance. A tubercle wing model is generated through varying wavelengths along the span. The wavelength
of tubercle model varying from 0.5c to 0.3c is reduced with the interval of 0.05c, whereas for the second
model the waviness wavelength increased from 0.3c to 0.5c with the interval of 0.05c. More specification
details are given in Table 1. All the wing models were developed in Pro-Engineer software, for which the
airfoil coordinates were obtained from the University of Illinois webpage. The models are named in such
a way that the four digits after λ represent the tenth multiple values of wavelength (λ/c) at the root and the
tenth multiple of (λ/c) at the wing tip, respectively, where h denotes the amplitude and the number with
h designates tenth amplitude for the corresponding variant.
Three dimensional (3D) solid models were imported in ANSYS software. The simulation was per-
formed at six different angles of attack which were achieved by rotating the wing model without distribut-
ing the free stream velocity components. The upstream and downstream far field boundaries were located
at seven chord lengths from the leading and trailing edges, respectively. Additionally, one lateral boundary
was attached to one end of the wing, while the second lateral boundary was at four times the chord length
from the root of the wing. To predict the realistic behavior the wing was supposed to be the cantilever fixed
with the fuselage at the root, while the tip was free. Three-dimensional flow effects were observed when
considering the finite wing. Tetrahedral mesh elements were used and the patch confirming algorithm was
implemented to get the fine mesh in high pressure gradient regions. Previous studies revealed that the wav-
iness wavelength is one of the important parameters that affect the wing performance and the correspond-
ing effect was explored by many researchers. However, the effect of variable waviness wavelength has got
minor attention. Additionally, the actual flipper of the humpback whale has heterogeneous protuberances
on its flippers. The nonperiodic mesh was generated and the same mesh was used for different wavelength
models. In order to capture the boundary layer effects the mesh near the solid wing was highly refined
through employing inflation layers. In the present study Y+ values are kept less than unity (Y+ < 1); in the

FLUID DYNAMICS Vol. 55 No. 5 2020

660 TUNIO et al.

Span (s) 

Chord (c)
(a)

(b)

Fig. 1. (a) The plan-form view of λ0503h1 wavy wing model and (b) zoomed view mesh model.

vicinity of the wing the corresponding height of the first element was 0.035 mm. The no-slip boundary
conditions were preassigned at the upper and lower surfaces of the airfoil. The meshed model of the wavy
wing with increasing wavelength toward the tip is shown in Fig. 1b. The transport equations were discret-
ized through finite volume method in the ANSYS Fluent 14.5 environment. The convergence criterion
was based on the coefficient of aerodynamic forces, momentum, continuity and turbulence model equa-
tions. The uniform velocity boundary condition with 0.8% turbulence intensity was applied at the inlet
boundary, the velocity vectors being normal to the inlet [6]. The zero pressure condition was employed at
the outlet. The governing fluid flow equations were solved using the SIMPLE algorithm, whereas the spa-
tial discretization was conducted by means of the second-order upwind scheme.

1.2. Computational Model

The research was carried out at low Reynolds numbers in view of the fact that the flow Mach number
is very low; therefore, the simulation was conducted under the incompressibility assumption. The flow of
air around the wing is modeled as three-dimensional, viscous, and steady. Hence, the governing partial

FLUID DYNAMICS Vol. 55 No. 5 2020

 INVESTIGATION OF VARIABLE SPANWISE WAVINESS WAVELENGTH EFFECT 661

differential equations, such as the continuity and RANS equations for the present case of steady, viscous
and three-dimensional incompressible flow take the form:
∂U i
= 0,
∂xi
∂ ρU U = − ∂P + ∂ μ  ∂U i + ∂U j  + ∂ (−ρu 'u ' ),
( i j)   
∂x j ∂xi ∂x j   ∂x j ∂xi  ∂x j
i j
(1.2)
i, j = 1,2, 3.
The last term on the right-hand side of the second equation −ρui'u j' represents Reynolds stress, which
account for turbulences present inflow. In Fluent TM a variety of turbulence models are available to model
the Reynolds stresses in accordance with to the physical conditions of the problem. In this study two dif-
ferent approaches are used to model the Reynolds stress term of RANS equations, namely, the k–ε real-
izable and k–ω SST models. The addition of cross-diffusion term in the equation of the k–ω SST model
ensures that the equations would be accurately solved both near the wall and in the far field region
∂ ( ku j )  ν  
= ∂  ν + t  ∂k  + P − ε
∂x j ∂x j  σ k  ∂x j 
and
∂ ( εu j )  ν  
= ∂  ν + t  ∂ε  + C1S ε − C2 ε
2
,
∂x j ∂x j  σε  ∂x j  k + νε
where
 η 
C1 = max 0.43, ,
 η + 5

η = S k,
ε
S = 2Sij Sij .
In these equations P represents the generation of the turbulent kinetic energy due to the mean velocity
gradients, σk and σε are the turbulent Prandtl numbers for k and ε, respectively. In the present work the
model constants are as follows: C2 = 1.9, σk = 1, and σε = 1.2.
∂ ( kui )  Γ ∂k 
= ∂  k  + P + Yk ,
∂x j ∂x j  ρ∂x j 
∂ ( ωui ) Γ 
= ∂  ω ∂ω  + Gω − Yω + Dω.
∂x j ∂x j  ρ ∂x j 
In these equations P represents the generation of the turbulent kinetic energy due to the mean velocity
gradients and Gω represents the generation of ω; Γk and Γω represent the effective diffusivities of k and ω,
respectively; Yk and Yω represent the dissipation of k and ω, respectively, due to turbulence, and Dω rep-
resents the cross-diffusion term.

1.3. Grid Independence Test (GIT)

The grid independency test (GIT) was performed, so that the aerodynamic force coefficients should
be independent of mesh refinement. There were many studies which investigated the effect of different
tubercle parameters but the gird independence test was performed only for a single model at a particular
angle of attack [21, 22]. Thus, the GIT at one incidence angle/configuration is sufficient to estimate the
acceptable level of accuracy. Hence, in the present study the GIT was performed for the baseline model,
i.e., having a smooth leading edge at a 4° angle of attack. To do this, six different computational meshes
were generated, and numerical analysis was performed for each mesh. The GIT results showed that grid
refinement has a significant impact on the aerodynamic force developed on the wing surface. It is found

FLUID DYNAMICS Vol. 55 No. 5 2020

662 TUNIO et al.

(a) (b)
Lift coefficient Drag coefficient
1.0 0.35
0.9 Experimental
0.30 (Hansen et al., 2010)
0.8 k−ε
0.7 0.25 k−ω SST
0.6 0.20
0.5
0.4 0.15
Experimental
0.3 (Hansen et al., 2010) 0.10
0.2 k−ε
0.1 k−ω SST
0.05

0 4 8 12 16 20 0 4 8 12 16 20
Angle of attack, deg Angle of attack, deg

Fig. 2. Comparison of CFD simulation results with experimental results [6]; (a) lift coefficient and (b) drag coefficient.

that a mesh with 3.8 million cells provides the same L/D ratio as that with 4 million cells mesh; therefore,
the further refinement of the mesh has no effect on the aerodynamic forces. Thus, the present study uses
the mesh with 3.8 million cells for all remaining cases.

1.4. Computational Model Validation

Wind tunnel experiments at Re = 1.2 × 105 were conducted at the University of Adelaide with the
NACA0021 airfoil [6]. The baseline model with a smooth leading edge was used for the validation of the
CFD turbulence model. The data in the form of the lift and drag coefficients were used. The present study
used two different turbulence models, namely, the k–ε realizable and k–ω SST models. The results
obtained through numerical simulation were compared and presented against the experimental data, as
shown in Figs. 2a and 2b. The comparison with respect to the predicted lift coefficient CL shows that the
results of the k–ε realizable turbulence model are in good agreement with the experimental data from α =
0° to α = 12° the deviation being within 4.5%. For an angle of attack greater than 12° the k–ε realizable
model overestimates CL with an about 60% error. The similar behavior can be noticed for the drag coeffi-
cient in Fig. 2b up to the 12° angle of attack, whereas the completely opposite trend is observed after the
12° angle of attack. The comparison of CL obtained through the k–ω SST model and the experimental
method at different AoAs showed that k–ω SST model underestimates CL up to 12° angle of attack with
maximum percentage error of 19.8%, whereas at higher angles of attacks in the post-stall regime the lift
coefficient predicted through the k–ω SST model showed good agreement with the experimental results
having error within 4.8%. A similar behavior was observed for the drag coefficient in Fig. 2b. On the basis
of the results obtained it is concluded that the k–ε realizable model predicts results close with the exper-
iment at angles of attack lower than 12° with a maximum variation of 4.5%, whereas for the higher angles
of attack the k–ω SST model results are better than those of the k–ε realizable model having a maximum
variation of 4.8%. Hence, this research work uses the k–ε realizable model up to 12° AoA and the k–ω
SST model for the higher angles of attack. Since the numerical simulation was performed at six different
angles of attacks with the step of 4°, the exact stall angle may be in between 12° and 16°, but, owing to the
discreteness of the variation in the angle of attack difference, the 12° AOA is taken for the stall angle.

2. RESULTS AND DISCUSSION

2.1. Aerodynamic Forces
The aerodynamic behavior of different wavy wing models was compared against the baseline model at
various angles of attack in the pre-stall and post-stall regimes. Figures 3a–3c compare the aerodynamic
performance of the baseline model against the wavy wing models with decreasing (λ0503h1) and increas-
ing (λ0305h1) wavelength from root to tip.
In Fig. 3a the drag coefficient for the above-mentioned cases is presented as a function of the angle of
attack. It can be seen that at zero AoA the drag coefficients for all the three cases are the same and it
increases gradually till the angle of attack is 12°. Once the angle of attack is slightly greater than 12°, the
drag coefficient rises rapidly till 20°. At initial angles of attack (i.e., 0° to 12°), the λ0305h1 shows the low-
est drag coefficient, while the two other cases demonstrate slightly better drag coefficient values. But as

FLUID DYNAMICS Vol. 55 No. 5 2020

 INVESTIGATION OF VARIABLE SPANWISE WAVINESS WAVELENGTH EFFECT 663

(a) (b)
Drag coefficient Lift coefficient
0.35 1.0
0.30 0.9 Baseline
Baseline 0.8 λ0503h1
0.25 λ0503h1 0.7 λ0305h1
0.20 λ0305h1 0.6
0.5
0.15 0.4
0.10 0.3
0.2
0.05 0.1
0 4 8 12 16 20 0 4 8 12 16
Angle of attack, deg Angle of attack, deg
L/D (c)
12
10
8
6 Baseline
λ0503h1
4 λ0305h1
2

0 4 8 12 16 20
Angle of attack, deg

Fig. 3. Comparison of (a) lift coefficient, (b) drag coefficient, and (c) L/D ratio of two-way wing models with a smooth
leading edge wing at a various angle of attack.

soon as the angle of attack is greater than 12°, the inverse pattern is noticed, i.e., the λ0305h1 model shows
the highest drag coefficient.
Figure 3b presents the comparison of the lift coefficients as functions of the AoA for the baseline,
λ0503h1, and λ0305h1 models. It is obvious that the lift coefficient increases in a regular pattern at the
initial stage of the plot, while it decreases irregularly at the later stage. At the initial stage, the baseline
model shows the highest value of the lift coefficient, whereas the other models perform identically. At the
later stage, a sudden decrease in the baseline model’s lift coefficient is observed, while the λ0305h1 model
shows the highest lift coefficient.
Figure 3c presents the comparison of the aerodynamic performance for the baseline, λ0503h1, and
λ0305h1 models at different angles of attack. It can be seen that the L/D ratio increases gradually with
increase in the angle of attack up to 8°, while it decreases once this value has been surpassed. Moreover,
the baseline model has the highest L/D ratio followed by the λ0305h1 and λ0503h1 models at lower angles
of attack. However, at higher angles of attack a different scenario is witnessed. Moving from 8° to 12°, the
baseline model has the highest L/D ratio, while the other models show approximately similar value. Like-
wise, the values of all the three models are found to be alike from 12° to 16°. After 16°, λ0305h1 again rises
at the top, while the remaining models exhibit nearly undistinguishable values.

2.2. Surface Pressure Distribution

A critical analysis of the static pressure distribution over the wing surface is the starting point to under-
stand the flow physics that govern the aerodynamic performance behavior. In Fig. 4 the pressure distribu-
tion over the baseline, λ0503h1, and λ0305h1 models is presented for 12°. The static pressure distribution
over the surface of the baseline model shown in Fig. 4a exhibits strong leading edge suction, whose mag-
nitude is maximum at the root, whereby it varies along the span and decreases toward the wing tip. While
observing the pressure distribution over a wavy wing it is noticed that leading-edge suction is maximum
behind the trough and the suction magnitude is maximum at the wing root and decreases along the span

FLUID DYNAMICS Vol. 55 No. 5 2020

664 TUNIO et al.

(а) (b)
 = 12  = 12

4.10e+02
3.47e+02
2.84e+02
2.21e+02
1.58e+02
9.50e+01
3.20e+01
−3.10e+01
−9.40e+01
−1.57e+02
−2.20e+02
−2.83e+02
−3.45e+02 Wingtip
−4.09e+02
−4.72e+02
−5.35e+02 (c)
−5.96e+02
−6.61e+02
−7.24e+02
 = 12
−7.87e+02
−8.50e+02

Wingtip

(d) (e)
4.10e+02
3.47e+02 Wing root  = 20  = 20
2.84e+02
2.21e+02
1.58e+02
9.50e+01
3.20e+01
−3.10e+01
−9.40e+01
−1.57e+02
−2.20e+02
−2.83e+02
−3.45e+02
−4.09e+02
−4.72e+02
−5.35e+02 Wingtip
−5.96e+02
−6.61e+02
−7.24e+02
−7.87e+02
−8.50e+02

Fig. 4. Static pressure distribution at 12° angle of attack over the (a) baseline wing, (b) wavy wing model λ0503h1, (c) wavy
wing model λ0305h1 and at 20° angle of attack, (d) baseline model, (e) wavy wing model λ0305h1.

for both wavy wing cases. A decrease in the tubercle wavelength has a minor effect on the wing pressure
distribution, as shown in Figs. 4b and 4c. Through comparing the baseline and the wavy wing pressure dis-
tributions, it is found that the baseline model shows strong leading-edge suction and it lasts very near to
the wingtip, whereas in the case of both wavy wings the leading-edge suction is only behind the wave
trough that causes a decrease in the leading-edge suction area. On the basis of pressure distribution com-
parison, it is hypothesized that a decrease in the leading-edge suction is the major cause in the reduction
of the wavy wing aerodynamic performance at a lower angle of attack. The careful analysis of the pressure
distribution along the wavy wing shows that employing tubercles along the wingspan causes a spanwise
pressure gradient, namely, a positive pressure gradient from the waviness trough to the peak and a negative
pressure gradient exist from the peak to the trough.
In Figs. 4d and 4e the pressure distributions over the baseline and wavy wing models are presented
at 20°, respectively. From the pressure distributions in the pre-stall and post-stall regimes it is concluded
that in pre-stall regime a suction peak occurs in the waviness trough, but it exists near the root and disap-

FLUID DYNAMICS Vol. 55 No. 5 2020

 INVESTIGATION OF VARIABLE SPANWISE WAVINESS WAVELENGTH EFFECT 665

(а) (b)
Velosity Velosity
streamlines Baseline model streamlines
4.189e+001
Waviness valley
4.189e+001

3.142e+001 3.142e+001

2.094e+001 2.094e+001

1.047e+001 1.047e+001

0 0
m/s
m/s

(c)
Velosity
streamlines At waviness peak
4.189e+001

3.142e+001

2.094e+001

1.047e+001

0
m/s

Fig. 5. Velocity vector showing flow separation at AOA = 20° for (a) baseline at z = 35 mm from root, (b) λ0503h1 model
at valley (z = 17.5 mm) from root, (c) λ0503h1 model at peak (z = 35 mm from root).

pears toward the tip, whereby in the post-stall regime the suction peak exists near the wingtip in both base-
line and wavy wing model cases. To better understand the aerodynamic behavior of both wings in the post-
stall regime, the careful examination of pressure distribution on the baseline and wavy wing model
revealed that the leading-edge suction area is greater in the case of the wavy wing model. Moreover, it is
clear that in the case of the smooth leading-edge wing the suction occurs only over a small area near the
tip, whereas the suction area and the suction pressure are much higher in the case of wavy wings. Thus,
the greater leading-edge suction pressure over the greater leading-edge area results in an increase in the
aerodynamic performance at higher angles of attack.
In Fig. 5 the velocity distributions over the baseline and wavy wings are presented. From Fig. 5a it can
be observed that the major wing portion is under the critical stall, where flow separation starts immedi-
ately after the leading edge. In Fig. 5b the velocity distribution over the wavy wing at waviness peak is pre-
sented; here it can be seen that flow separation is restricted behind the waviness peak, which is due to very
low leading-edge suction at the peak. Figure 5c presents the velocity distribution over the wavy wing at the
valley. From the velocity distribution, it is observed that the flow separates just behind the leading edge,
then reattaches to the wing surface, and then again starts to separate. Finally, it can be concluded that the
baseline wing is under the critical stall at 20° AOA, whereas the wavy wing experiences stall only at wing
valley. Flow separation is restricted at the waviness peak; these results are consistent with previous litera-
ture on the wavy wing [14, 23, 24].
The vorticity variation in the vicinity of the leading edge of the baseline and wavy wings is presented for
two different angles of attacks, α = 8° and α = 20°. In Fig. 6a the leading-edge vorticity is illustrated at
α = 8° for the baseline model. From this figure, it is observed that a very thin layer of negative vorticity is
attached to the upper surface of the wing. This thin layer of negative vorticity causes the clockwise rotation
of fluid particles; as a result of this rotation, the boundary layer momentum increases, due to the contin-
uous transfer of momentum from the outer fluid. This increase in momentum inside the boundary layer
enables the wing to achieve the higher aerodynamic performance at lower AoAs. In contrast to that, at

FLUID DYNAMICS Vol. 55 No. 5 2020

666 TUNIO et al.

(a) (b)

5.00e+01 1.50e+05 3.00e+05 4.50e+05 6.00e+05 7.50e+05 9.00e+05 1.00e+06 5.00e+01 1.50e+05 3.00e+05 4.50e+05 6.00e+05 7.50e+05 9.00e+05 1.00e+06

(c) (d)

5.00e+01 1.50e+05 3.00e+05 4.50e+05 6.00e+05 7.50e+05 9.00e+051.00e+06 5.00e+01 1.50e+05 3.00e+05 4.50e+05 6.00e+05 7.50e+05 9.00e+05 1.00e+06

Fig. 6. Leading edge vorticity distribution: (a) baseline at α = 8° at 35 mm from root, (b) baseline at α = 20° at 35 mm
from root, (c) λ0503h1 model at peak α = 20° at 35 mm from root, (d) λ0503h1 model at valley α = 20° at 35 mm from
root.

higher AoAs, as shown in Fig. 6b, negative regions of vorticity move away from the leading edge in an
upward direction and a positive vorticity layer is attached to the upper surface of the wing, causing fluid
particles to move in counter-clockwise direction. Due to the layer of positive vorticity, the boundary layer
momentum decreases which leads to flow separation. On the basis of the above-stated argument, it was
concluded that for the higher AoAs the aerodynamic performance of the aircraft wing decreases, owing
separation of the negative vorticity zone.
In Figs. 6c and 6d the vorticity distribution near the leading edge is presented for the spanwise wavy
wing at 20° angle of attack. As discussed above, the separation of the negative vorticity zone from the upper
surface of the wing causes flow separation. In contrast to that, from Figs. 6c and 6d, it was observed that
in the case of the wavy wing the negative vorticity region remains attached to the upper surface of the wing
along the chord at higher AoAs. This attached negative vorticity region causes clockwise rotation of fluid
particles, which in turn enhances the boundary layer momentum by transferring the momentum from the
the outer fluid, these results being in agreement with the findings of [1, 3, 12, 13, 25]. This increased
boundary layer momentum prevents flow separation and provides an aerodynamic advantage at a higher
angle of attack through increasing the stall angle.

2.3. Flow Mechanism

The streamline patterns for models λ0503h1 and λ0305h1 at 20° AOA are presented in Figs. 7a–7d.
The streamline behavior over the wing helps to understand the reduction in the aerodynamic performance
of wavy wings with decreasing wavelength from root to tip. In the case of decreasing wavelength (Figs. 7a
and 7b), it can be noticed that the vortices are produced at the top surface of the wing. These vortices are
produced due to the sharp wingtip of the wavy wing model with decreasing waviness wavelength toward
the tip. Due to the sharp wingtip, fluid particles at the lower surface of the wing move upward near the
trailing edge of the wing, which creates an adverse pressure gradient for fluid particles at the wing upper

FLUID DYNAMICS Vol. 55 No. 5 2020

 INVESTIGATION OF VARIABLE SPANWISE WAVINESS WAVELENGTH EFFECT 667

(a) Wingtip (b)

Wing
Recirculation zone root

(c) (d)
Wing Wingtip
root

Separation and recirculation zone near the wing Wing Smooth streamline pattern
root due tt very low waviness wavelength root over the entire wing
(i.e., sharp trailing edge)

Fig. 7. Streamline behavior over the wavy wing (a–b) with decreasing wavelength from root to tip, i.e., λ0503h1 at 20°
angle of attack (c–d) with increasing wavelength from root to tip, i.e., λ0305h1 at 20° angle of attack.

surface. Due to the adverse pressure gradient, flow recirculation zone is created and the pressure at the
wing upper surface increases. Additionally, it is observed that for the wavy wing model with decreasing
wavelength towards the tip (i.e., λ0503h1), vortices are produced over the entire upper surface of the wing.
These vortices can be produced, as the result of disturbances caused by the vortices near the wingtips. This
drop of the overall pressure difference between the upper and lower surfaces of the wing results in a
decrease in the L/D ratio. However, in the case of a wing with increasing waviness wavelength toward
wingtip vortices are produced only near the root of the wing (i.e., where the waviness wavelength is very
low), whereas only separate vortices of very low strength are observed over the rest of the wing surface, as
shown in Figs. 7c and 7d. Hence, the L/D ratio is higher.

SUMMARY
The effect of varying waviness wavelength along the wingspan and its underlying flow mechanism is
explored in the pre-stall and post-stall regimes. Two wing models, i.e., those with increasing wavelength
from root to tip (λ0305h1) and decreasing wavelength from root to tip (λ0503h1) were considered in com-

FLUID DYNAMICS Vol. 55 No. 5 2020

668 TUNIO et al.

parison with the baseline model. In the pre-stall regime a maximum reduction in the lift-to-drag (L/D)
ratio was 16.89% and 4.22% for models λ0503h1 and λ0305h1, respectively.
In contrast to that, in the post-stall regime a maximum increase in the L/D ratio of 2.97% and 19.18%
was noticed for λ0503h1 and λ0305h1, respectively, at 20° angle of attack. On the basis of obtained results,
it is suggested that the reduction in the waviness wavelength toward the wing tip has no favorable effect on
the aerodynamic performance of a wing. Moreover, an increase in the waviness wavelength provided ben-
efit in the post-stall regime and showed negligible loss of the L/D ratio in the pre-stall regime.
On the basis of the observed pressure distributions, it is established that a decrease in the leading-edge
suction area is the major cause of L/D ratio reduction in the pre-stall regime. In the post-stall regime the
limitedness of the flow separation zone behind the waviness peak and an increase in the boundary layer
momentum due to attached negative vorticity region were found to be the major reasons for improved
aerodynamic performance. Moreover, it is also shown that a decrease in the wavelength from root to tip
in the case of λ0503h1 causes vortices at the wing surface due to the sharp wingtip edges, which results in
the lower L/D ratio than in the case of λ0305h1. Also an increase in the pressure gradient at the wing sur-
face is witnessed, as the vortices create a flow recirculation zone. On the basis of obtained results it is also
hypothesized that there is a particular value of the waviness wavelength beyond which the further decrease
in the tubercle wavelength results in the trailing edge sharpening and provides favorable conditions for
flow separation.

NOMENCLATURE

X Transformation at required span position

x Chord length
Z Length in the spanwise direction
h Tubercle amplitude
λ Tubercle wavelength
Ui Velocity component
ρ Density
μ Dynamic viscosity
P Free-stream pressure

ACKNOWLEDGEMENTS
The authors are thankful to Mehran University of Engineering & Technology, Jamshoro and Hamdard Univer-
sity Karachi for providing all necessary facilities for the completion of this research work.

DECLARATION OF CONFLICTING INTERESTS

The Authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication
of this article.

REFERENCES
1. F. E. Fish and J. M. Battle, “Hydrodynamic design of the humpback whale flipper,” J. Morphology 225(1), 51–
60 (1995).
2. P. Watts and F. E. Fish, “The influence of passive, leading edge tubercles on wing performance,” in Proc. Twelfth
Intl. Symp. Unmanned Untethered Submers. Technol., 2001, Auton. Undersea Syst. Inst. Durham New Hamp-
shire.
3. D. Miklosovic, M. M. Murray, L. E. Howle, and F. E. Fish, “Leading-edge tubercles delay stall on humpback
whale (Megaptera novaeangliae) flippers,” Phys. Fluids 16(5), L39–L42 (2004).
4. D. S. Miklosovic, M. M. Murray, and L. E. Howle, “Experimental evaluation of sinusoidal leading edges,”
J. Aircraft 44(4), 1404–1408 (2007).
5. M. J. Stanway, Hydrodynamic Effects of Leading-Edge Tubercles on Control Surfaces and in Flapping Foil Propul-
sion (Massachusetts Institute of Technology, 2008).

FLUID DYNAMICS Vol. 55 No. 5 2020

 INVESTIGATION OF VARIABLE SPANWISE WAVINESS WAVELENGTH EFFECT 669

6. K. Hansen, R. Kelso, and B. Dally, “An investigation of three-dimensional effects on the performance of tuber-
cles at low Reynolds numbers,” in: 17th Australasian Fluid Mechanics Conference, Auckland, New Zealand. 2010.
7. F. Fish and G. Lauder, “Passive and active flow control by swimming fishes and mammals,” Annu. Rev. Fluid
Mech. 38, 193–224 (2006).
8. F.E. Fish, P.W. Weber, M.M. Murray, and L.E. Hawle, The Tubercles on Humpback Whales’ Flippers: Applica-
tion of Bio-Inspired Technology (Oxford University Press, 2011).
9. M. Zhang, G. Wang, and J. Xu, “Aerodynamic control of low-Reynolds-number airfoil with leading-edge pro-
tuberances,” AIAA J. 51(8), 1960-1971 (2013).
10. M. Zhang, G. Wang, and J. Xu, “Experimental study of flow separation control on a low-Re airfoil using lead-
ing-edge protuberance method,” Experiments Fluids 55(4), 1710 (2014).
11. E.A. Van Nierop, S. Alben, and M.P. Brenner, “How bumps on whale flippers delay stall: an aerodynamic mod-
el,"Phys. Rev. Letters 100(5), 054502 (2008).
12. H. Johari et al., “Effects of leading-edge protuberances on airfoil performance.,” AIAA J. 45(11), 2634–2642
(2007).
13. D.S. Custodio, The Effect of Humpback Whale-like Protuberances on Hydrofoil Performance (2007).
14. J. Favier, A. Pinelli, and U. Piomelli, “Control of the separated flow around an airfoil using a wavy leading edge
inspired by humpback whale flippers,” Comptes Rendus Mécanique 340(1–2), 107–114 (2012).
15. R. Pérez-Torró and J.W. Kim, “A large-eddy simulation on a deep-stalled aerofoil with a wavy leading edge,”
J. Fluid Mech. 813, 23–52 (2017).
16. D. Serson, J. Meneghini, and S. Sherwin, “Direct numerical simulations of the flow around wings with span-
wise waviness at a very low Reynolds number,” Computers Fluids 146, 117–124 (2017).
17. M. Zhao, M. Zhang, and J. Xu, “Numerical analysis of effects of leading-edge protuberances at low Reynolds
number,” EWEA (2015).
18. A. Skillen, A. Revel, J. Favier, A. Pinelli, and U. Piomelli, “Investigation of wing stall delay effect due to an un-
dulating leading edge: An LES study,” in: TSFP DIGITAL LIBRARY ONLINE (Begel House Inc., 2013).
19. D. Serson, J.R. Meneghini, and S.J. Sherwin, “Direct numerical simulations of the flow around wings with
spanwise waviness,” J. Fluid Mech. 826, 714–731 (2017).
20. H. Yoon, “Effect of the wavy leading edge on hydrodynamic characteristics for flow around low aspect ratio
wing,” Computers Fluids 49(1), 276–289 (2011).
21. M.L. Post, R. Decker, A.R. Sapell, and J.S. Hert, “Effect of bio-inspired sinusoidal leading-edges on wings,”
Aerospace Sci. Technol. 81, 128–140 (2018).
22. N. Rostamzadeh, R.M. Kelso, and B. Dally, “A numerical investigation into the effects of Reynolds number on
the flow mechanism induced by a tubercled leading edge,” Theor. Comput. Fluid Dyn. 31(1), 1–32 (2017).
23. A. Skillen, A. Revelli, U. Piomelli, and J. Favier, “Flow over a wing with leading-edge undulations,” AIAA J.
53(2), 464–472 (2014).
24. M. Zhao, M. Zhang, and J. Xu, “Numerical simulation of flow characteristics behind the aerodynamic perfor-
mances on an airfoil with leading edge protuberances,” Engineering Applications of Computational Fluid Me-
chanics 11(1), 193–209 (2017).
25. H.T. Pedro and M.H. Kobayashi, “Numerical study of stall delay on humpback whale flippers,” in: 46th AIAA
Aerospace Sciences Meeting and Exhibit (2008).

FLUID DYNAMICS Vol. 55 No. 5 2020