Acta Mechanica Sinica (2012) 28(6):1551–1556

DOI 10.1007/s10409-012-0158-8

RESEARCH PAPER

Numerical analysis on transitions and symmetry-breaking

in the wake of a flapping foil

Guo-Yi He · Qi Wang · Xing Zhang · Shu-Guang Zhang

Received: 24 May 2012 / Revised: 19 October 2012 / Accepted: 29 Ocotber 2012

©The Chinese Society of Theoretical and Applied Mechanics and Springer-Verlag Berlin Heidelberg 2012

Abstract Flying and marine animals often use flapping 1 Introduction

wings or tails to generate thrust. In this paper, we will use
the simplest flapping model with a sinusoidal pitching mo- Flapping motion is a common mode for animal flying
tion over a range of frequency and amplitude to investigate and swimming. Especially, vertical flapping including both
the mechanism of thrust generation. Previous work focuses pitching and plunging could generate a horizontal thrust by
on the Karman vortex street and the reversed Karman vor- the interaction of moving foils with surrounding flows. Since
tex street but the transition between two states remains un- the flapping motion is critically important to the design of
known. The present numerical simulation provides a com- micro-air-vehicles (MAVs), there exist various experimental
plete scenario of flow patterns from the Karman vortex street studies and numerical simulations for exploring the mech-
to reversed Karman vortex street via aligned vortices and the anism of thrust generation [1–6]. A simple model to un-
ultimate state is the deflected Karman vortex street, as the derstand the mechanism of flapping-generated-thrust is the
parameters of flapping motions change. The results are in Karman vortex streets (KVS) and the reversed Karman vor-
agreement with the previous experiment. We make further tex streets (RKVS). Both KVS and RKVS can be observed
discussion on the relationship of the observed states with in the wake behind flapping foils, where the former is as-
drag and thrust coefficients and explore the mechanism of sociated with drag and latter with thrust. In fact, as either
enhanced thrust generation using flapping motions. amplitudes or frequencies increase in pitching motions, the
transition from the KVS to the RKVS can be observed. The
Keywords Flapping motion · Immersed boundary method · present paper is devoted to the study of transition of flow pat-
Wake · Symmetry-breaking terns and the parameters in flapping motions associated with
thrust generation.
It was first observed by Knoller [7] and Betz [8] that
The project was supported by the Natural Science Foundation of a flapping wing could generate thrust. Karman and Burg-
Jiangxi Province (2010GZC0162). ers [9] offered the first theoretical explanation of drag or
thrust production based on the wake vortices, where the wake
G.-Y. He · Q. Wang of the flow past bluff bodies is modeled by an infinite row
The School of Aircraft Engineering, of alternating vortices, commonly known as KVS. Bohl and
Nanchang Hangkong University, Koochesfahani [10] experimentally studied the RKVS with
330063 Nanchang, China various reduced frequencies. Ellenrieder et al. [11] ex-
plained the Strouhal number dependency of the wake vortex
X. Zhang (¬)
structure behind a translating airfoil. Meanwhile, many re-
LNM, Institute of Mechanics,
searchers investigated the RKVS in different situations, such
Chinese Academy of Sciences,
as the wakes of fishes [12, 13] in both experiments [14–16]
100190 Beijing, China
and numerical simulations [16–21]. However, a clearly un-
e-mail: zhangx@lnm.imech.ac.cn
derstanding of the transition mechanism between KVS and
S.-G. Zhang RKVS in the wake behind flapping foils was not available.
The School of Transportation Science and Engineering, Recently, Godoy-Diana et al. [22] experimentally investi-
Beihang University, 100083 Beijing, China gated the transition mechanism from KVS to RKVS and dis-
1552 G.-Y. He, et al.

cussed the relation of the transition process with thrust gener- ∇ · u = 0, (3)
ation. In this paper, we will numerically study the transition
where u is the velocity, p is the pressure, and f is the exter-
of flow patterns in the wake of flapping foils and investigate
nal force. The boundary conditions are given as follows: At
their relation with thrust generations.
the inlet boundary, a uniform flow with velocity U = 1, and
We will use the immersed boundary method [23–27] to
the normal gradient of the pressure is set to be zero; At the
numerically simulate Godoy-Diana et al.’s experiment and
outlet, the velocities are extrapolated from the interior and
investigate the transition process from the KVS to the RKVS
the normal gradient of the pressure is also zero; At the lat-
and its relation with thrust generation. Moreover, the numer-
eral boundaries, the gradient of tangential velocity is zero,
ical simulation offers some details of flow patterns, which
the normal velocity is zero and the normal gradient of the
are not available in experiments, so that the complete scenar-
pressure is also zero. The non-slip boundary conditions are
ios on flow patterns can be observed. The rest of this paper
applied to the surfaces of the foil where the fluid velocity is
is organized as follows: Section 2 describes the mathemat-
the same as that of the foil.
ical models for the flows around the flapping foils and the
The computational region ranges from −3D to 19D in
numerical method. Section 3 presents the numerical simula-
the horizontal (or streamwise) direction (x direction), and
tion results on thrust or drag under different Strouhal num-
from −8D to 8D in the vertical direction with the origin at
bers and their relation to the flow patterns. Finally, we will
summarize the results of the present study in Sect. 4. the centre of the semicircle. The number of uniform grids is
240 × 120 with Δx = Δy = 0.025 in the rectangular region
of −D < x < 5D and −1.5D < y < 1.5D, the grids are then
2 Mathematical model and numerical method stretched towards the boundary with the size growth ratio of
1.04 (see Fig. 2). The time step is 0.002.
2.1 A simple model for flapping foil

The present numerical configuration is set up to simulate the

experiment by Godoy-Diana et al. [22]. The geometry of
flapping foil is shown in Fig. 1, the diameter of the semi-
circle D is 5 mm and its chord length c is 23 mm. The foil
oscillates in pitching with respective to the center of the
semicircle. The control parameters are the oncoming uni-
form velocity U, the flapping frequency f and the peak to
peak amplitude A. The main non-dimensional parameters
are the Reynolds number Re, the pitching amplitude AD and
the Strouhal number St, defined as Re = UD/v, AD = A/D
and St = f D/U. The pitching motion of foils is set as a
sinusoidal function of time Fig. 2 Grid used for the present computation

Af (t) = −AD sin(2π f t), (1)

where Af (t) is the flapping amplitude varying with time. 2.3 Numerical simulation

We use the immersed boundary method to simulate the flow

field around flapping foils, and force f in Eq. (2) will be cal-
culated using the volume-of-fluid (VOF) method [28], The
VOF method is an Eulerian scheme and uses the volume
fraction Φ: The coefficient Φ is unity in the regions occu-
pied entirely by solid phase and zero in regions occupied by
fluid phase, and 0 < Φ < 1 in the regions partially occupied
by either fluid and solid. Utilizing the VOF method, the so-
Fig. 1 Schematic views of the foil lution procedure in the immersed boundary method can be
summarized as follows:
(1) Solve the N–S equations to obtain the intermediate ve-
2.2 Governing equations
locity component ũu
The Navier–Stokes (N–S) equations for a 2-D incompress- ũu − u n 3 1
uu)n + ∇h (uu)n−1
= − ∇h (uu
uu
ible flow are written as Δt 2 2
∂uu 1 2 1
+ ∇ (uu
uu) = −∇p +
uu ∇ u + f, (2) ∇h pn +
−∇ ∇2 (uun + ũu). (4)
∂t Re 2Re h
Numerical analysis on transitions and symmetry-breaking in the wake of a flapping foil 1553

(2) Calculate the volume forces by VOF airfoils, we will calculate the drag coefficient defined as

f = Φ[uun+1 2
rig − ũ ]/Δt,
u (5) Cd = F1 ds, (10)
ρU 2 L
where u rig is the velocity of the foil.
where F1 is the force acting on the airfoil surface in the
(3) Correct the intermediate velocity using the volume streamwise direction.
forces We performed simulations of the case with Re = 255,
u ∗ − ũu St = 0.22 and AD = 1.07, using different time steps and grid
= f. (6) resolutions. For the time step of 0.002 and 0.001, the drag
Δt
coefficients equal to 0.552 and 0.563, respectively. For the
(4) Calculate the intermediate velocity u ∗∗ , compute the grid width of 0.025 and 0.01, the drag coefficients equal to
pressure and update the fluid velocity u n+1 0.552 and 0.549, respectively. The differences in the results
u ∗∗ − u ∗ obtained by using different time steps and mesh resolutions
= ∇h pn , (7) are sufficiently small.
Δt
∇h · u ∗∗ 3 Numerical results
∇2h pn+1 = , (8)
Δt
u n+1 − u ∗∗ The parameters in the present simulation are taken from the
= −∇h pn+1 . (9) experiment by Godoy-Diana et al. [22]: the Reynolds num-
Δt
ber Re = 255, the uniform oncoming flow U = 1 and fluid
Here, the projection method is used to decouple the solu- density ρ = 1. We will take different Strouhal numbers St
tion of velocity and that of pressure. In time advancing, and flapping amplitudes AD to study how the flow patterns
the Adams–Bashford and the Crank–Nicholson schemes are are dependent on these parameters and compare the results
used for the convective and the diffusive term, respectively. obtained with the previous experiments of Godoy-Diana et
The spatial discretization is based on the finite volume ap- al. [22].
proach. Figure 4 is plotted for a fixed St = 0.22 and different
The immersed boundary code has been developed by values of AD , equal to 0.36, 0.71, 1.07, 1.77 and 2.80, respec-
our research group. To validate the code, one testing case is tively. The left column is for vorticity contours, the middle
implemented. In this numerical example, we investigate the one for time averaged streamwise velocity contours and the
laminar flow induced by the harmonic in-line oscillation of right one for the time averaged streamwise velocity compo-
a circular cylinder. According to Ref. [28], the key param- nent in the near wake. The first row at AD = 0.36 is the typ-
eters are Re = Umax D/ν = 100, KC = Umax /( f D), where ical case of low-amplitude pitching which produces a KVS,
Umax is the maximum velocity of the cylinder motion; D de- and the mean flow is a typical wake profile which character-
notes the diameter of the cylinder and f is the frequency of izes drag. By increasing the amplitude AD to 0.71, the vor-
the oscillation. The translational motion is given by the har- tices of alternating signs are aligned with the symmetry line
monic oscillation of x(t) = −Ac sin(2π f t), where Ac denotes of the wake. When the amplitude increases to AD = 1.07,
the amplitude of the cylinder motion. Figure 3 shows the the rotation direction of vortices changes and the flows in
time history of the in-line force. From this figure, it is seen wakes transit from KVS to RKVS, and the mean flow is a
that the present numerical result is in agreement with that of typical jet profile which characterizes thrust. By further in-
Ref. [29]. creasing amplitude to 1.77, the vortex streets start to become
For studying the propulsive performance of flapping deflected and asymmetric to the center line. As the ampli-
tude increases to 2.8, a symmetry breaking is observed in the
RKVS. These plots clearly demonstrate that as the flapping
amplitude increases, the flow patterns transit from KVS to
RKVS and finally become deflected RKVS with symmetry-
breaking.
Similar scenarios can be found by fixing the ampli-
tude to AD = 1.07 and taking different Strouhal numbers,
St = 0.10, 0.22, 0.30, 0.40 (see Fig. 5). It is also observed
that the transition from RKVS to the deflected RKVS oc-
curs when the Strouhal number St or flapping frequency f is
higher than a critical value.
The bifurcation diagram for parameters St and AD is
plotted in Fig. 6, which indicates different regimes of flow
patterns: The KVS is located in the lower area, while the
Fig. 3 Time history of the in-line force RKVS is located in the upper-right area. The regime of
1554 G.-Y. He, et al.

aligned vortices is in between and the deflected RKVS is in The bifurcation diagram offers transition paths for different
the region of largest St and A/D. These parameter regimes Strouhal numbers and flapping amplitudes.
are qualitatively in agreement with the experimental results.

Fig. 4 Instantaneous vorticity contours (left column), time-averaged streamwise velocity contours (middle column) and time-averaged
streamwise velocity component in the near wake (right column) for fixed Re = 255 and St = 0.22, from top to bottom. a AD = 0.36;
b AD = 0.71; c AD = 1.07; d AD = 1.77 and e AD = 2.8

Figure 7 shows the dependency of the drag coefficient flapping amplitude could affect thrust generation. For larger
on the flapping amplitude AD for different Strouhal numbers. St, the increase in flapping amplitude could significantly im-
For the smallest St, there is no thrust generation even for prove the thrust generation.
large AD . As soon as St is larger than a threshold value, the
Numerical analysis on transitions and symmetry-breaking in the wake of a flapping foil 1555

Fig. 5 Instantaneous vorticity contours for fixed Re = 255 and AD = 1.07. a St = 0.10; b St = 0.22; c St = 0.30; d St = 0.40

flected RKVS). Godoy-Diana et al.’s experiment [22] first

showed such scenarios and the present numerical simulations
confirm the physical process with detailed observations. It is
observed from the bifurcation diagram that there exist inter-
mediate states and an ultimate state. The complete scenarios
from KVS to RKVS offer an essential understanding of the
thrust generation using flapping motions.

Fig. 6 The bifurcation diagram of St and AD for different flow

patterns. Real line: transition between Karman vortex streets and
reversed Karman vortex streets; Dotted line: transition between re-
versed Karman vortex streets and deflected reversed Karman vortex
streets
Fig. 7 The drag coefficients versus the flapping amplitudes AD for
different St. Cd0 denotes the drag coefficient of non-flapping foil at
zero angle of attack, Cd0 = 0.861
4 Conclusions

In summary, the present numerical simulation shows com-

plete transition scenarios of the wake behind a flapping foil: References
As either Strouhal number or flapping amplitude increases,
the flow patterns can transit from KVS to RKVS via the 1 Anderson, J.M., Streitlien, K., Barrett, D.S., et al.: Oscillating
aligned-vortices regime and finally become deflected RKVS foils of high propulsive efficiency. J. Fluid Mech. 360, 4–72
when the Strouhal number and flapping amplitude are rel- (1998)
atively large. The previous work has focused on KVS and 2 Buchholz, J.H.J., Smits, A.J.: On the evolution of the wake
RKVS. However, there is little work on the intermediate structure produced by a low-aspect-ratio pitching panel. J.
state (the aligned-vortices regime) and the ultimate state (de- Fluid Mech. 546, 43–443 (2006)
1556 G.-Y. He, et al.

3 Buchholz, J.H.J., Smits, A.J.: The wake structure and thrust AIAA J. 36, 1240–1246 (1998)
performance of a rigid low-aspect-ratio pitching panel. J. Fluid 17 Molina, J., Zhang, X., Angland, D.: On the unsteady motion
Mech. 603, 331–365 (2008) and stability of a heaving airfoil in ground effect. Acta Mech.
4 Couder, Y., Basdevant, C.: Experimental and numerical study Sin. 27,164–178 (2011)
of vortex couples in two dimensional flows. J. Fluid Mech. 18 Shyy, W., Liang, Y., Tang, J., et al.: Computational aerody-
173, 225–251 (1986) namics of low Reynolds number plunging, pitching and flexible
5 von Ellenrieder, K.D., Parker, K., Soria, J.: Flow structures be- wings. Acta Mech. Sin. 24, 351–373 (2008)
hind a heaving and pitching finite-span wing. J. Fluid Mech. 19 Dong, G.J., Lu, X.Y.: Numerical analysis on the propulsive per-
490, 129–138 (2003) formance and vortex shedding of fish-like traveling wavy plate.
6 von Ellenrieder, K.D., Pothos, S.: PIV measurements of the Int. J. Numer. Meth. Fluids 49, 1351–1373 (2005)
asymmetric wake of a two dimensional heaving hydrofoil. Exp. 20 Sun, M., Wang, J.K., Xiong, Y.: Dynamic flight stability of
Fluids 44, 733–745 (2008) hovering insects. Acta Mech. Sin. 23, 231–246 (2007)
7 Knoller, R.: Die Gesetze des Luftwiderstandes. Flug- und 21 Wu, J.H., Sun, M.: Unsteady aerodynamic forces of a flapping
Motor-technik (Wien). 3, 1–7 (1909) wing. J. Exp. Biol. 207, 1137–1150 (2004)
8 Betz, A.: Ein Beitrag zur Erkl¨arung des Segelfluges. 22 Godoy-Diana, R., Aider, J.L., Wesfreid, J.E.: Transitions in the
Zeitschrift fur Flugtechnik und Motorluftschiffahrt 3, 269–272 wake of a flapping foil. Physical Review E 77 (2008)
(1912)
23 Mittal, R., Iaccarino, G.: Immersed boundary methods. Annual
9 von Karman, T., Burgers, J.M.: General Aerodynamic Theory
Review of Fluid Mechanics 37, 239–261 (2005)
Perfect Fluids, Aerodynamic Theory. Durand, W. F. edn. Divi-
24 Kim, J., Kim, D., Choi, H.: An immersed-boundary finite-
sion E 2, Julius-Springer, Berlin (1943)
volume method for simulations of flow in complex geometries.
10 Bohl, D.G., Koochesfahani, M.M.: MTV measurements of the
J. Comp. Phys. 171, 132–150 (2001)
vertical field in the wake of an airfoil oscillating at high reduced
25 Su, S.W., Lai, M.C., Lin, C.A.: An immersed boundary tech-
frequency. J. Fluid Mech. 620, 63–88 (2009)
nique for simulating complex flows with rigid boundary. Com-
11 von Ellenrieder, K.D., Parker, K., Soria, J.: Fluid mechanics of
puters and Fluids 36, 313–324 (2007)
flapping wings. Experimental Thermal and Fluid Science 32,
1578–1589 (2008) 26 Yang, X.L., He, G.W., Zhang, X.: Large-eddy simulation of
12 Wolfgang, M.J., Anderson, J.M., Grosenbaugh, M.A., et al.: flows past a flapping airfoil using immersed boundary method.
Near-body flow dynamics in swimming fish. J. Exp. Biol. J. Science China Physics, Mechanics and Astronomy 53, 1101–
Exp. Biol. 202, 2303–2307 (1999) 1108 (2010)
13 Drucker, E.G., Lauder, G.V.: Locomotor function of the dorsal 27 Yang, X.L., Zhang, X., Li, Z.L., et al.: A smoothing tech-
fin in teleost fishes: Experimental analysis of wake forces in nique for discrete delta functions with application to immersed
sunfish. J. Exp. Biol. 204, 2943–2958 (2001) boundary method in moving boundary simulations. Journal of
14 Koochesfahani, M.M.: Vortical patterns in the wake of an os- Computational Physics 228, 7821–7836 (2009)
cillating airfoil. AIAA J. 27, 1200–1205 (1989) 28 Ravoux, J.F., Nadim, A., Haj-Hariri H.: An embedding method
15 Gopalkrishnan, R., Triantafyllou, M.S., Triantafyllou, G.S., et for bluff body flows: Interactions of two side-by-side cylinder
al.: Active vorticity control in a shear-flow using a flapping foil. wakes. Theoret. Comput. Fluid Dynamics 16, 433–466 (2003)
J. Fluid Mech. 274, 1–21 (1994) 29 Dutsch, H., Durst, F., Becker, S., et al.: Low-Reynolds-number
16 Jones, K.D., Dohring, C.M., Platzer, M.F.: An experimen- flow around an oscillating circular cylinder at low Keulegan–
tal and computational investigation of the Knoller–Betz effect. Carpenter numbers. J. Fluid Mech. 360, 249–271 (1998)