Wake Signature and Strouhal Number Dependence of Finite-Span Flapping Wings

Corresponding author: Joel E. Guerrero E-mail: [email protected]

Journal of Bionic Engineering 7 Suppl. (2010) S109–S122

Wake Signature and Strouhal Number Dependence of Finite-Span Flapping Wings

Joel E. Guerrero

University of Genoa, Via Montallegro 1, Genoa, 16145, Italy

Abstract A numerical study was conducted in order to investigate the unsteady aerodynamics of finite-span flapping rigid wings. The

unsteady laminar incompressible Navier-Stokes equations were solved on moving overlapping structured grids using a sec-ond-order accurate in space and time finite-difference scheme. Specifically, finite-span rigid wings undergoing pure heaving and root-flapping motions were studied. From the results presented, it is found that root-flapping wings produce wake structures similar to those of heaving wings, but with the difference that the latter wing kinematics generates larger vortices and forces than root-flapping wings; aside from this, similar wake regimes occur at comparable values of the Strouhal number. The numerical simulations were performed at a Reynolds number of Re = 250 and at different values of Strouhal number and reduced frequency.

Keywords: rolling wings, overlapping grids, wake signature, vortex rings, horseshoe shaped vortex, thrust production Copyright © 2010, Jilin University. Published by Elsevier Limited and Science Press. All rights reserved. doi: 10.1016/S1672-6529(09)60224-9

1 Introduction

Biologists, zoologists and engineers are actively investigating birds, bats, insects, fishes and cetaceans, as they represent illuminating examples of unsteady aero-dynamics, high manoeuvrability, endurance, and large aero/hydro-dynamics efficiency. The main reason of studying natural flyers and swimmers relies on the use of Mother Nature as the inspiration model for improving existing applications or developing new technologies by just mimicking nature evolutionary-optimization proc-ess (biomimetics). Such applications may include drag and noise reduction by using feather-like structures[1] or the development of new propulsion/lift generation sys-tems for Micro-Air-Vehicles (MAVs), Nano-Air-Ve-hicles (NAVs) and Autonomous-Underwater-Vehicles (AUVs) inspired by flapping wings or oscillating fins[2–5].

When studying flapping flight, the term “flapping” is often used to refer to the complex motion of birds, bats and insects’ wings. Flapping flight involves up and down movement of the wings about the body joint (rolling). During the downstroke (or power-stroke), the wings move downward and forward. During the up-

stroke (or recovery-stroke), the wings move upward and are drawn in towards the body to reduce drag. The wings also change their angle of attack depending on the stroke and also systematically change shape. Flapping flight is basically rowing in the air with the added complication that lift must be generated. In many experimental and numerical flapping wings studies[6–10] this very complex flapping motion has been simplified to pure heaving or coupled heaving-and-pitching motions, wherein the rolling motion of the wings is replaced by a heaving motion. These studies have given us useful but limited insight into the unsteady aerodynamics of flapping flight, as they have not taken into account the root-flapping motion characteristic of birds, bats and insects.

Another important aspect of flying using flapping wings or swimming using oscillating fins is the ability of generating thrust with relatively high propulsive effi-ciency (defined as the ratio of aerodynamic power out-put to mechanical power input); early attempts at building fish-inspired mechanisms achieved disap-pointingly low propulsive efficiencies[11]. It was only through a deeper understanding of the vorticity and wake produced by swimming animals that significant progress was achieved[12].

Journal of Bionic Engineering (2010) Vol.7 Suppl. S110

At this point, an immediate question that comes to our mind when studying natural flyers and swimmers is the range of flapping parameters that may be chosen to optimize the aero/hydro-dynamic performance. Dar-winian evolution is not guaranteed to find a solution that is globally optimized among the range of available pa-rameters. However, a thorough examination of Mother Nature’s technique is a logical starting point in defining guiding principles. In this field of study, several re-searchers[13–15] have found that flying and swimming animals cruise at Strouhal number (St) corresponding to a regime of vortex growth and shedding in which the propulsion efficiency peaks. St is a dimensionless pa-rameter defined as

,fStU×= (1)

where f is the flapping frequency, the peak to peak amplitude of the flapping stroke and U is the forward velocity. Biologists, zoologists and engineers, have ob-served that most flying and swimming animals oscillate their wings or tails in a narrow range corresponding to 0.2 < St < 0.4, where propulsive efficiency is high[13–15].

In the current numerical study, we aim to perform a comprehensive analysis of the wake signature of fi-nite-span rigid wings undergoing root-flapping motion, a motion mode which seems not to have been suffi-ciently studied. In addition, we also check whether this motion mode generates a similar vortex growth and shedding regimes to those of pure heaving motion at similar St values. The unsteady laminar incompressible 3D Navier-Stokes equations are numerically approxi-mated and all unsteady, viscous and spanwise effects are solved. The simulations are conducted for St regimes between 0.15 and 0.5 and for two different reduced frequency values, one corresponding to high frequency and the other one to low frequency. This is done in order to study Leading Edge Vortex (LEV) shedding de-pendency. The reduced frequency k, which is another dimensionless parameter that characterizes the unsteady aerodynamics of flapping wings or oscillating fins, is a measure of the residence time of a particle (or a vortex) convecting over the wing (or fin) chord, compared to the period of motion and is defined as

,f ckU

× ×= (2)

with c the wing cross-section chord. The remainder of this paper is organized as follows.

We begin by briefly describing the numerical method and gridding methodology. Then, we present a descrip-tion of the computational domain and a short discussion of the results obtained from a grid dependence study. This is followed by a detailed discussion of the results. Finally, we present the conclusions and perspectives.

2 Solution method

In this section, we briefly outline the solution methodology used to solve the governing equations on moving overlapping structured grids. The complete description of the numerical method and gridding methodology can be found in Refs. [16] and [17], re-spectively. In primitive variables (u, v, w, p), the Initial Boundary Value Problem (IBVP) for the laminar in-compressible Navier-Stokes equations is

2

0

pt

∂ −∇+ ⋅ ∇ = + ∇∂∇ ⋅ =

u u u u

u for ∈x , t > 0, (3)

with the following boundary conditions and initial con-ditions

( )

0

, for , 0

( ,0) ( ) for , 0

B p t

t

= ∈∂ >

= ∈ =

u g x

Q x q x x. (4)

In this IBVP, x = (x, y, z) (for = 3, where is the number of space dimensions) is the vector containing the Cartesian coordinates in physical space , is a bounded domain in ∈R , ∂ is the boundary of the domain , t is the physical time, u = (u, v, w) is the vector containing the velocity field in , p is the pres-sure, v is the kinematic viscosity, is the density, B is a boundary operator, g is the boundary data and q0 is the initial data. An alternative formulation to the system of Eqs. (3) and (4), called the velocity-pressure formulation is

2

2

0x y z

pt

p + u + v + w

∂ −∇+ ⋅∇ = + ∇∂∇ ∇ ⋅ ∇ ⋅ ∇ ⋅ =

u u u u

u u u for ∈x , t > 0,

(5)

with the following boundary and initial conditions

Guerrero: Wake Signature and Strouhal Number Dependence of Finite-Span Flapping Wings S111

( )

0

, for , 00 for , 0

( ,0) ( ) for , 0

B p t t

x t

= ∈∂ >∇ ⋅ = ∈∂ >

= ∈ =

u g xu x

Q x q x

(6)

The system of Eqs. (5) and (6) is equivalent to the original formulation, Eqs. (3) and (4)[16,18,19], and is the form of the equations that will be discretized. Hence, we look for an approximate numerical solution of Eq. (5) in a given domain , with prescribed boundary conditions and given initial conditions (Eq. (6)).

Eqs. (5) and (6) are solved in logically rectangular grids in the transformed computational space = ( , , , ) (refer to Refs. [10], [16], [17], [20] and [21] for a

detailed derivation), using second-order centred fi-nite-difference approximations on structured overlap-ping grids.

The overlapping structured grids method consists in generating a set of body-fitted conforming structured components grids g that completely cover the domain

that is being modelled in physical space = (x, y, z, t) and overlap where they meet (see Fig. 1). The gov-erning equations are then solved separately in computa-tional space = ( , , , ) on each component grid

g and domain connectivity is obtained through proper interpolation in the overlapping areas.

The presence of moving bodies changes the relative position of the overlapping grids continuously during the flow simulation. As the component grid (around a moving body) traverses through the computational do-main, overlapping connectivity information, such as interpolation stencils and unused points regions, is re-computed. The automation of the hole cutting and in-terpolation stencils computation makes the present

methodology a powerful tool for the simulation of flows with one or multiple moving bodies, since the grids do not have to be regenerated as the solution evolves. In general, the motion of the component grids g and/or boundaries may be an user-defined time dependent function, may obey the Newton-Euler equations for the case of rigid body motion or may correspond to the boundary nodes displacement in response to the stresses exerted by the fluid pressure for the case of fluid-structure interaction problems. When a component grid changes position during a moving grid computation, the overlapping grid generator is called at each time step in order to update the interpolation stencils and unused points regions. The component grids themselves do not have to be recomputed unless they deform in shape.

As the problem of flapping wings is implicitly a moving boundaries problem, we have to solve the gov-erning equations in a frame that moves with the grid. For moving overlapping grids, Eq. (5) is expressed in a ref-erence frame moving with the component grid as fol-lows,

( ) 2

2

0x y z

p+ +t

p + u + v + w

∂ −∇− ⋅ ∇ = ∇∂∇ ∇ ⋅ ∇ ⋅ ∇ ⋅ =

u u G u u

u u u for ∈x , t > 0,

(7)

where G is the rate of change of the position of a given set of grid points g

px in the physical space (grid velocity). It is important to mention that the new gov-erning equations expressed in the moving reference frame must be accompanied by proper boundary condi-tions. For a moving body with a corresponding moving no-slip wall, only one constraint may be applied and this corresponds to the velocity on the wall, such as

wall, wall, wall wall( | ) ( | ), where | ( ).g g gp t p t p t= ∈∂u x G x x (8)

Finally, the discretized equations are integrated in time using a semi-implicit multi-step method, which uses a Crank-Nicolson scheme for the viscous terms and a second-order Adams-Bashforth/Adams-Moulton pre-dictor-corrector approach for the convective terms. By using this time-stepping scheme, the velocity is advance in time by using a second-order Adams-Bahforth pre-dictor step, followed by a second-order Adams- Moulton corrector step. The solution method previously described,

Fig. 1 Simple overlapping grid system in physical space .


yields a second-order accurate in space and time nu-merical scheme on moving overlapping structured grids.

To assemble the overlapping grid system and solve the laminar incompressible Navier-Stokes equa-tions in their velocity-pressure formulation, the Over-ture* framework is used. The large sparse non-linear system of equations arising from the discretization of the laminar incompressible Navier-Stokes equations is solved using the PETSc# library, which was interfaced with Overture. The system of non-linear equations is then solved using a Newton-Krylov iterative method, in combination with a suitable preconditioner.

3 Computational domain, grid dependence study and wake structures visualization

The overlapping grid system layout used to conduct this parametric study is shown in Fig. 2. In this figure, the Background Grid (BG) extends 4.0 × c away from the wing Leading Edge (LE), 10.0 × c away from the wing Trailing Edge (TE), 2.0 × c away from the Left and Right Wing-Tips (LH-WT and RH-WT respectively) and 4.0 × c + h away from the point of maximum thickness of the upper and lower surfaces (where h is the flapping amplitude). These dimensions are the dimen-sions for the overlapping grid system layout when the wing is in the mean position of the flapping cycle.

The wing aspect ratio is defined as AR = S2/A, where S is the wingspan (wingtip-to-wingtip distance) and A is the wing area; for a rectangular wing it is simply AR = S/c. In all the cases to be presented, a rectangular wing with aspect ratio equal to AR = 2 is used. Finally, the rectangular wing has an elliptical cross-section with a corresponding major axis a = 0.25 and minor axis b = 0.025; hence, the wing chord c is equal to c = 0.5.

The initial conditions used for all the flapping wing simulations are those of a fully converged solution of the corresponding fixed wing. In Fig. 2, the left boundary of the BG of the computational domain layout in the x-y plane corresponds to an inflow boundary condition [(u) = (1.0, 0.0, 0.0), ˆ p∂n = 0] and the top, bottom and right boundaries of the BG are outflow boundaries (ba-sically vanishing pressure gradient and velocity ex-trapolated from the interior points). All the boundaries of the BG of the computational domain layout in the z-y * https://computation.llnl.gov/casc/Overture/ # http://www-unix.mcs.anl.gov/petsc/petsc-as/

plane correspond to outflow condition. On the moving wing surface we impose a no-slip boundary condition ( , ,x y z=u G G G ). The rest of the boundaries are inter-polation boundaries, where we use a non-conservative Lagrange interpolation scheme to interpolate the solu-tion between the overlapping component grids g . The Reynolds number (Re = U × c/ ) is equal to 250 for all the numerical experiments.

It is a very well known fact that the effect of mesh size is an important factor to consider when assessing the quality of a numerical solution. Thus, we conduct a grid dependence study in order to determine the most suitable grid in terms of computing time and solution accuracy from a quantitative point of view. To conduct this study, we use the Grid Convergence Index (GCI) method as described by Roache[22]. Here, a pure heaving wing is considered with different grid sizes, layouts and clus-tering. Several simulations are run at a St = 0.3 (k = 1.570 795). In our calculations, unsteadiness is observed to disappear typically after 4 cycles of wing flapping motion and further calculations show negligible non-periodicity. In Table 1, the parameters for the three overlapping grids system used for the GCI study are presented. In order to better resolve the wing wake, we use inverse hyperbolic tangent grid stretching in the x, y and z directions of the BG; we also use inverse hyper-bolic tangent grid stretching towards the LE, TE, LH-WT and RH-WT, since we expect vortices to be generated in these areas. Table 1 show the grid dimen-sions, grid spacing ratio (from the finer grid to the coarser grid) and position of the first node normal to the wing surface. The Wing Grid (WG) is used as the ref-erence grid to conduct the GCI study and a grid spacing refinement ratio r = 2 is employed. Each simulation is also checked for acceptable iterative convergence. Table 1 Grid dimensions used for the grid dependence study. Here, BG stands for background grid, WG-CS for wing grid center-section, WG-TS for wing grid tip-section (where is made up of two component WG-TS), GSR for grid refinement spacing ratio and 1st NW for position of the first node normal to the wall

1 2 3

BG 161 × 121 × 101 161 × 121 × 101 161 × 121 × 101

WG-CS 221 × 121 × 41 201 × 101 × 31 161 × 81 × 31

WG-TS 81 × 61 × 41 61 × 51 × 31 51 × 41 × 31

GSR 1 2 4

1st NW 0.001 × 2c 0.002 × 2c 0.004 × 2c


In unsteady aerodynamics, in order to quantify the flow characteristics, we compute the lift and drag coef-ficients (cl and cd respectively), defined as

l 2,1

2

LcU Aρ

= d 2.1

2

DcU Aρ

= (9)

Since in the study of flapping wings propulsion the main task is thrust production, it is more convenient to think in terms of thrust force T instead of drag force D. The thrust force is equal in magnitude but opposite in direction to the drag force (T = D), hence, the thrust coefficient ct is equal to

t d 2.1

2

Tc cU Aρ

= − = (10)

In Eqs. (9) and (10), the lift force L and the thrust force T are computed by integrating the viscous and pressure forces over the wing surface. The cl, cd, ct can be aver-aged in time as follows

l l

d d

t t

1 ( )d

1 ( )d ,

1 ( )d

t

t

t

t

t

t

c c t t

c c t t

c c t t

+

+

+

=

=

=

(11)

where is the period of flapping motion and is equal to = 1/f.

In Table 2, the observed values of the average drag coefficient dc and rms

lc for the grid dependence study are presented; these values are used to compute the ob-served order of convergence pobs, the value of the ob-served quantity at zero grid spacing spacing = 0, the fine grid convergence index GCI12 and GCI23, and the con-

stancy of GCI23 = r p× GCI12. Based on the GCI index values found (refer to Table 3), we can say that dc is estimated to be 0.057 708 with an error band of 0.397 203% for grid 1 and within an error band of 1.641 617% for grid 2 ; whereas rms

lc is estimated to be 0.908 738 with an error band of 0.402 503% for grid 1 and within an error band of 1.567 203% for grid 2 . In Table 3, it can be also evidenced that the value of (GCI23/GCI12) × (1/rp) for both observed quantities is approximately 1, which indicates that the results for grids 1 and 2 are well within the asymptotic range of convergence[22].

Additionally to the quantitative GCI study, we also conduct an extra grid dependence study but from the qualitative point of view (wake structures resolution). The results for grids 1 and 2 presented in Table 1 are illustrated in Fig. 3. In this figure, we use the iso-surfaces of | |-criterion (vorticity magnitude) and the iso-surfaces of Q-criterion to capture the vortices and their corresponding cores. It can be seen from this figure

3.0

×c

+ h

3.0

×c

+ h

Fig. 2 (a) Computational domain layout in the x-y plane, (b) computational domain layout in the z-y plane.

Table 2 Observed values of dc and the root mean square of c1 ( rms

lc ) for the grid dependence study

Grid g GSR dc rmslc

1 1 0.057 892 0.905 822

2 2 0.058 476 0.897 486

3 4 0.060 914 0.865 326

Table 3 GCI study results

Outcome dc rmslc

pobs 2.061 650 1.947 838

spacing=0 0.057 708 0.908 738

GCI12 (%) 0.397 203 0.402 503

GCI23 (%) 1.641 617 1.567 203 (GCI23/GCI12 ) ×(1/r p) 0.990 012 1.009 249


that the | |-criterion, although capable of capturing the general vortical structures, has the disadvantage of also showing the shear layers near the wing surface and between the vortices. On the other hand, the Q-criterion[23,24] shows the vortical structures details more clearly as it avoids the shear layers close to the wing, and provides nearly identical structures to those obtained with the | |-criterion. Hence, we choose the Q-criterion as the main criterion for wake topology characterization. As it can be seen in Fig. 3, there is no discernible difference between the solutions, indicating that the grids are adequate for wake structures resolu-tion.

Summarizing the quantitative and qualitative re-sults previously presented, it can be said that the solu-tions obtained by using grid 1 and 2 are grid inde-pendent. Taking into account the computational re-sources available, CPU time restrictions and solution accuracy, 2 with a corresponding first node normal to the wall located at a distance equal to 0.001 × 2c from the wing surface is used as the base grid to perform all further computations. In the case of a smaller or bigger computational domain (refer to Fig. 2), the grid dimen-sions are scaled in order to keep the same grid spacing as for this domain. In Fig. 4, we illustrate a typical over-lapping grid system used in the wake topology study discussed in the next section.

Fig. 4 Typical overlapping grid system used in the wake topology study. (a) side view; (b) front view; (c) bottom view; (d) detailed perspective view. Notice that the wing grid is made up of three component grids, namely, wing grid center-section 1, left wing grid tip-section 2 and right wing grid tip-section 3. 0 is the background grid.

Fig. 3 Vortex topology at the beginning of the upstroke (t = 5.0). In light gray we show the iso-surfaces of | |-criterion and in dark gray the iso-surfaces of Q-criterion. Heaving parameters: Re = 250, St = 0.3, k = 1.570 795. (a) Corresponds to 1 and (b) to

2 .


4 Simulation results

Hereafter we carry out a comprehensive parametric study in order to assess the wake signature of heaving and root-flapping rigid finite-span wings. We begin by studying a pure heaving wing case and then we compare the results obtained to those of root-flapping wings in order to study whether this motion generates similar wake structures to those of pure heaving wings at similar St and k values. For the pure heaving case, the wing heaves in the y-direction according to

y(t) = h × sin(2 × × f × t), (12)

where y(t) is the heaving motion, h the heaving amplitude, f the heaving frequency and t is the time. In Table 4, we present the kinematics parameters governing this nu-merical experiment, we also present the quantitative results, where tc is the average thrust coefficient and lc is the maximum lift coefficient (measured during the downstroke). By inspecting these results, we observe that, as we increase St and k, the values of tc and lc also in-crease. We can also notice that for values of St < 0.30 we are in the drag producing regime; for values of 0.30 < St < 0.35 we produce little or no drag (or thrust), whereas for values of St > 0.35 we are in the thrust producing

regime. We also observe two different behaviours of the aerodynamic forces for high and low reduced frequencies k. Hence, it seems that for flapping wings, the flapping frequency plays an important role in the vortex genera-tion and shedding and, henceforth, on the aerodynamic forces. These observations of the frequency dependence are similar to those presented by Wang[25], but here we extend them to three-dimensional cases.

In Figs. 5 and 6, the vortex topology for a heaving case in the thrust producing regime is illustrated (3DH-11 in Table 4). These figures show that the downstream wake of this heaving wing consists of two sets of complex shaped Vortex Rings (VR) which con-vect at oblique angles about the centerline of the motion. Thus, the flow induced by each vortex ring along its axis is expected to have a net streamwise component. This net streamwise momentum excess in the wake is con-nected with the thrust production of flapping wings. The process by which the vortex rings are formed can be explained by examining the vortex formation and shed-ding in the vicinity of the wing surface. During the heaving motion and close to the wing, four vortices are formed; namely, Leading Edge Vortex (LEV), Trailing Edge Vortex (TEV) and two Wing-Tip Vortices (WTV), that is to say, left and right WTV (LH-WTV and RH-WTV respectively). These four vortices are all connected and form a closed Vortex Loop (VL); as this vortex loop is convected downstream, it fully discon-nects from the wing, forming a ring. It is also of interest the fact that each vortex loop has two sets of Thin Con-trails (TC) that connect the VL to the VR generated in the previous stroke; these structures are segments of the wing-tip vortices and, as the vortex loops are convected downstream they become weaker and ultimately disap-pear, as for vortex ring VR2 (see Fig. 5). During a complete heaving cycle, two VRs are formed, one at the end of the upstroke and the other one at the end of the downstroke.

In Fig. 7, we present the wake topology for a drag producing case (3DH-3 in Table 4). It is clear from this figure that the wake topology is very different from the one of the thrust producing case. In this case, as the vortex loops are convected downstream, they do not convert into vortex rings; instead, they keep their original shape and are diffused. It can also be observed that the wake height is very compact, opposite to that of

Table 4 Simulation results for the pure heaving parametric study (positive tc indicate thrust production whereas negative tc indicates drag production)

Case No. St k tc lc

3DH-1

3DH-2

3DH-3

3DH-4

3DH-5

3DH-6

3DH-7

3DH-8

3DH-9

3DH-10

3DH-11

3DH-12

3DH-13

3DH-14

3DH-15

3DH-16

0.15

0.15

0.20

0.20

0.25

0.25

0.30

0.30

0.35

0.35

0.40

0.40

0.45

0.45

0.50

0.50

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

0.122 675

0.111 949

0.106 101

0.087 877

0.086 136

0.068 118

0.058 294

0.052 475

0.022 962

0.036 191

0.061 370

0.008 296

0.143 680

0.010 296

0.199 366

0.024 617

0.560 613

0.422 249

0.784 628

0.597 156

1.040 502

0.798 169

1.238 780

1.024 220

1.651 670

1.257 870

2.030 980

1.539 740

2.417 032

1.891 345

2.944 020

2.150 710


the thrust production case. Also, the vortex loops are inclined in the direction opposite to the wing travel di-rection, resulting in a momentum surfeit linked to drag production.

We now proceed to simulate a wing rolling about its travelling axis (root-flapping motion). In this case, the wing is hinged at one wing-tip and is rolling about the travelling axis (where the travelling axis and the hinged point are collinear) as per equation

φ(t) = φ × sin(2 × × f × t), (13)

where φ(t) is the rolling motion, φ the dorsoventral stroke angle, f the rolling frequency and t is the time. The

Strouhal number for this case is based on φ as proposed by Taylor[15] and is computed as follows

sin( / 2) .S fStUφ× ×= (14)

In Table 5, we present the kinematics parameters governing this case, together with results for tc and lc . As it can be seen, for values of St less than 0.30 we are in the drag production regime, for values approximately between to 0.30 < St < 0.35 little or no drag (or thrust) is produced, whereas for values higher than St = 0.35 we are in the thrust production regime. From these results, it is clear that this behaviour is similar to that of heaving

Fig. 5 Vortex wake topology at the beginning of the upstroke for case 3DH-11 (t = 5.0). Heaving parameters: Re = 250, St = 0.4, k = 1.570 795 (thrust producing wake). An animation is available at the author’s website, http://www.dicat.unige.it/guerrero/flappingsimulationsnew.html

Fig. 6 Vortex wake topology at the beginning of the upstroke for case 3DH-11 (t = 5.0). Heaving parameters: Re = 250, St = 0.4, k = 1.570 795 (thrust producing wake). An animation is available at the author’s website. http://www.dicat.unige.it/guerrero/flappingsimulationsnew.html

Fig. 7 Vortex wake topology at the beginning of the upstroke for case 3DH-3 (t = 5.0). Heaving parameters: Re = 250, St = 0.2, k = 1.570 795 (drag producing wake). An animation is available at the author’s website. http://www.dicat.unige.it/guerrero/flappingsimulationsnew.html


wings. Comparing these results with the results for pure heaving wings, we find that the latter motions generate larger vortices and forces than root-flapping motion, presumably because the average velocity is higher across the wingspan, but otherwise the same wake re-gimes occurs at similar St.

In Figs. 8 and 9, the three-dimensional wake structures for a thrust producing case (3DR-11 in Table 5) and a drag producing case (3DR-1 in Table 5) are pre-sented. The salient feature that needs to be pointed out from these figures is the absence of any link at all or the presence of a very weak link between the root-tip vortex (hinged extreme) and the LEV and TEV vortices. We also observe that for the case 3DR-1, there is no con-nection between both wing-tip vortices and LEV and TEV vortices; this is due to the fact that the vortices generated at this low St have very low strength. Aside from these differences, we can observe that the down-stream wake of the rolling wing have some common features with the heaving wing wake, such as the for-mation of sets of well shaped vortex rings, the convec-tion of these vortex rings at oblique angles and the gen-eration of two VRs during a complete rolling cycle.

Fig. 8 Vortex wake topology at the middle of the upstroke for the rolling wing case 3DR-11 (t = 5.0). Flapping pa-rameters: Re = 250, St = 0.4, k = 1.570 795. The black dotted line represents the rolling axis (thrust producing wake). An animation is available at the author’s website. http://www.dicat.unige.it/guerrero/flappingsimulationsnew.html

Table 5 Simulation results for the rolling wing parametric study (positive tc indicate thrust production whereas negative tc indicates drag production)

Case No. St k tc lc

3DH-1

3DH-2

3DH-3

3DH-4

3DH-5

3DH-6

3DH-7

3DH-8

3DH-9

3DH-10

3DH-11

3DH-12

3DH-13

3DH-14

3DH-15

3DH-16

0.15

0.15

0.20

0.20

0.25

0.25

0.30

0.30

0.35

0.35

0.40

0.40

0.45

0.45

0.50

0.50

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

1.570 795

0.785 397

0.097 292

0.127 856

0.080 279

0.106 010

0.073 560

0.101 956

0.009 014

0.057 741

0.008 048

0.009 846

0.010 457

0.009 964

0.018 115

–

0.034 010

–

0.549 903

0.518 200

0.770 973

0.708 133

1.197 320

1.011 560

1.577 630

1.495 240

1.865 820

1.785 210

2.584 290

2.528 980

3.115 787

–

3.433 700

–


Fig. 9 Vortex wake topology at the middle of the upstroke for the rolling case 3DR-1 (t = 5.0). Flapping parameters: Re = 250, St = 0.15, k = 1.570 795. The black dotted line represents the rolling axis (drag producing wake). An animation is available at the author’s website, http://www.dicat.unige.it/guerrero/flappingsimulationsnew.html

The vortex rings formation process for rolling

wings is a little bit different from the one for heaving wings; here, the VL close to the rolling wing does not form a fully closed loop; instead, an open VL or Horseshoe Shaped Vortex (HSV) is observed (Fig. 8), which wraps the wing and comprises a LEV, a wing-tip vortex associated to the free wing-tip (WTV-F) and a TEV. In Fig. 10, a sequence corresponding to case 3DR-9 during upstroke motion is illustrated. In this sequence and as the wing accelerates during the upstroke motion, the LEV and TEV grow in size and in strength, stretching from the free wing-tip towards the wing base

(hinged wing-tip), presumably as a result of the onset of the WTV-F during the beginning of the upstroke. Sub-sequently as the wing approaches to its horizontal posi-tion, the TEV begins to detach from the trailing edge, forming the Starting Vortex (STV), which remains connected to the WTV-F, forming a very complex 3D structure around the free wing-tip. After the wing goes beyond its horizontal position and decelerates, the HSV fully detaches from the wing surface, while being con-vected in the streamwise direction. As the HSV is being shed, the open extremes of the HSV connect due to its induced rotation, forming in this way a well shaped VR

Fig. 10 Vortex shedding and formation sequence during upstroke motion. The Q-criterion iso-surfaces are coloured by pressure levels, where red corresponds to high pressure values and blue to low pressure values. The wing surface is coloured by pressure gradient magnitude levels. The sequence is from (a) to (e), where (a) 0.75 × , (b) 0.875 × , (c) 1.0 × , (d) 1.125 × , (e) 1.25 × . Flapping parameters: St = 0.35, k = 1.570 795. In figures A to E, the black dotted line represents the rolling axis. HL is the reference horizontal line (horizontal position).


far from the wing surface. As the wing reaches its top-most position and initiates its downstroke motion, a similar vortex shedding and formation process is ob-served; hence, two sets of VRs are formed during a full rolling cycle. As for the heaving cases, the flow induced by each vortex ring along its core is expected to have a net streamwise component which, for this case, is con-nected to thrust production (momentum excess).

It is also observed that during the first part of the upstroke motion, the LEV initially remains attached to the wing surface, with no evidence of breaking down (separation) or shedding. As the wing accelerates to its horizontal position, the LEV becomes a large, conical spiral vortex with a strong axial flow at the core (Fig. 11). This leads to a steadily increasing negative pressure region on the lower surface and hence augmentation of the forces acting on the wing. The occurrence of this axial flow is thought to be closely related to the spanwise pressure gradient resulting from the different velocities along the wing due to the rolling motion. The existence of this axial flow at the core of the LEV delays the vortex shedding and likely stabilizes the LEV, independently of the chosen reduced frequency (Fig. 11). Shortly after the wing reaches its horizontal position, the LEV breaks down at approximately 80% to 90% of the wing span

(measured from the wing base), as shown in Fig. 11. This break down is due to the reverse pressure gradient which destabilizes the LEV at the position where the spanwise flow disappears. A similar phenomenology is observed during the downstroke.

Additionally, in Fig. 12 we compare the wake structure and LEV shedding for cases 3DR-7 and 3DR-8 (same St and different values of k). As it can be seen in the sequences shown in this figure, the LEV vortex shedding process for both cases is very similar. In both cases, the LEVs are generated and shed approximately at the same rate for similar flapping periods. All these qualitative observations are also confirmed by the force measurement results shown in Table 5. From these re-sults, we can observe that the computed forces are al-most the same, independently of the value of k used. From these qualitative and quantitative results it is clear that the frequency selection does not introduce a fre-quency dependence into the results, opposite to heaving wings, where the aerodynamic performance is highly influenced by the LEV separation and convection, which provides a mechanism of optimal selection of heaving frequency (in the sense of maximum propulsive efficiency), as discussed by Wang[25] and Young and Lai[26].

Fig. 11 Vortex shedding comparison for two cases rolling at different reduced frequency values. The iso-surfaces are plotted by using Q-criterion. The wing surface is coloured by pressure gradient magnitude levels. Top row, flapping parameters: St = 0.35, k = 1.570 795. The sequence is from (a) to (d), where (a) 0.75 × , (b) 0.875 × , (c) 1.0 × , (d) 1.125 × (upstroke motion - bottom view). Bottom row, flapping parameters: St = 0.35, k = 0.785 397. The sequence is from (e) to (h), where (e) 0.75 × , (f) 0.875 × , (g) 1.0 × , (h) 1.125 × (upstroke motion - bottom view). The white dotted arrow represents the core and direction of the pressure gradient on the wing surface. The black dotted vertical line represents the rolling axis.


Fig. 12 Vortex shedding comparison for two cases rolling at different reduced frequency values. The Q-criterion iso-surfaces are coloured by pressure levels, where red corresponds to high pressure values and blue to low pressure values. In the sequence from (a) to (c), the flapping parameters are: St = 0.30, k = 1.570 795, where (a) 0.75 × , (b) 0.875 × , (c) 1.0 × (upstroke motion). In the sequence from (d) to (f), the flapping parameters are: St = 0.30, k = 0.785 397, where (d) 0.75 × , (e) 0.875 × , (f) 1.0 × (up-stroke motion). The black dotted line represents the rolling axis.

5 Conclusions and perspectives

In this paper, we have studied the unsteady aero-dynamics of finite-span flapping rigid wings. The laminar incompressible Navier-Stokes equations have been solved in their velocity-pressure formulation using a second-order accurate in space and time fi-nite-difference scheme on overlapping structured mov-ing grids. Both pure heaving and root-flapping motion have been studied. All the simulations have been conducted at different St values (0.15 < St < 0.5) and two reduced frequency values (k = 0.785 397 and k = 1.570 795).

The simulations show that the wake of thrust pro-ducing rigid finite-span heaving wings is formed by two sets of interconnected vortex loops that slowly convert into vortex rings as they are convected downstream. It is also observed that the vortex rings are inclined with respect to the free-stream; the angle of inclination of the vortex rings is in the direction of their travel and in the streamwise direction for thrust producing configurations; conversely, for drag producing configurations the angle of inclination is opposite to the direction of the stream-wise flow. The presence of thin contrails that link the

vortex loops is of interest; these structures are segments of the wing-tip vortices and as the vortex loops are convected downstream, they become weaker and ulti-mately disappear. In general, the observed structures are qualitatively similar to those observed in the experi-ments by Parker et al.[27] and the numerical simulations of Dong et al.[6] and Blondeaux et al.[7].

From the force measurement study for pure heaving wings, two different behaviours are observed for the average thrust coefficient tc and maximum lift coeffi-cient lc , for high and low reduced frequencies k. Hence, it seems that for heaving wings, the oscillating frequency also plays an important role in the vortex generation and shedding and, henceforth, on the aerodynamic forces.

From the qualitative and quantitative study of the root-flapping motion characteristic of flying animals (which as far as the author is aware still remains virtually unexplored) it is found that, indeed, root-flapping mo-tion produces wake structures similar to those of fi-nite-span heaving wings, but with the difference that the latter motion generates larger vortices and forces than root-flapping motion; apart from this, similar wake regimes (in terms of thrust producing or drag producing


wake) occur at similar St values. It is also observed that the vortex rings are formed from an open vortex loop or HSV, which as it is convected downstream connects to form a well shaped vortex ring. As for the heaving wing case, two sets of complex shaped vortex rings are formed during each flapping cycle. The vortex rings are also convected at oblique angles about the centerline of the motion and with a lateral inclination angle. As for heaving wings, the oblique inclination of the vortex rings is a direct indication of momentum excess or sur-feit (thrust or drag production respectively). The pres-ence of only one thin contrail that links the HSV with the vortex rings, is also worthy of note; this thin contrail is generated in the free wing-tip.

Conversely to the heaving wings, it is observed that for rolling wings the frequency selection does not in-troduce a dependence in the results. In fact, it is found that for different reduced frequency values the LEV shows a similar shedding process. The stabilization of the LEV is thought to be related to the axial flow that traverses the core of the LEV and delays the vortex break down and shedding. The LEV break down is found shortly after the wing reaches its horizontal posi-tion and approximately at 80-90% of the wingspan (measured from the wing base). This LEV break down is due to the reverse pressure gradient, which destabilizes the LEV and cause rapid separation. These observations clearly suggest that an additional parameter, such as pitching angle or wing geometry, should be taken into account in order to avoid LEV break down. Nevertheless, prolonging or delaying the LEV break down does not necessarily translate into increased global aerodynamic efficiency.

Finally, for the limited range of St and k values and simplified wing geometry and flapping kinematics cov-ered in this study, all the qualitative and quantitative results presented are in close agreement with the ex-perimental observations of Nudds et al.[14], Parker et al.[27], Rohr and Fish[12], Taylor et al.[15] and Trian-tafyllou et al.[13]; this supports the hypothesis that “fly-ing and swimming animals cruise at a St tuned for high power efficiency”[15].

The results presented in this paper are limited to laminar flow; nevertheless, they provide an excellent insight into the wake signature of the unsteady aerody-namics of flapping wings. The extension of the current

study to the turbulent case and the dynamics of separa-tion bubbles in the turbulent regime are envisaged. Also, the use of more realistic flapping kinematics and wing geometries, and the use of flexible wings and feathered surfaces will be the subject of future investigations. Lastly, the selection of an “optimal” wing shape and kinematics in terms of the maximum global aerody-namic efficiency is an open issue that we hope to deal with in the future.

Acknowledgements

This research is financially supported by the Marie Curie action EST, project FLUBIO (MEST-CT-2005- 020228). The simulations were performed on the na-tional supercomputers NEC SX-9 and NEC Nehalem Cluster at the High Performance Computing Center in Stuttgart (HLRS) under grant number WTAP3DW/ 12832.

References [1] Favier J, Dauptain A, Basso D, Bottaro A. Passive separa-

tion control using a self-adaptive hairy coating. Journal of Fluid Mechanics, 2009, 627, 451–483.

[2] Jones K D, Platzer M F. Experimental investigation of the aerodynamic characteristics of flapping-wing micro air ve-hicles. The 41st Aerospace Sciences Meeting and Exhibit, 2003, Washington, DC, 2003, AIAA-2003-0418.

[3] Singh S N, Simha A, Mittal R. Bionic AUV maneuvering by pectoral fins: Inverse control design based on CFD param-eterization. IEEE Journal of Oceanic Engineering, 2004, 29, 777–785.

[4] McMichael J M, Francis M S. Micro Air Vehicles - Toward a new Dimension in Flight, Technical Report, DARPA, USA, 1997.

[5] Michelson R, Naqvi M A. Extraterrestrial Flight (Ento-mopter-based Mars Surveyor), Lecture Series, Von Karman Institute for Fluid Dynamics, Brussels, Belgium, 2003.

[6] Dong H, Mittal R, Najjar F M. Wake topology and hydro-dynamic performance of low-aspect-ratio flapping foils. Journal of Fluid Mechanics, 2006, 566, 309–343.

[7] Blondeaux P, Fornarelli F, Guglielmini L, Triantafyllou M S, Verzicco R. Numerical experiments on flapping foils mim-icking fish-like locomotion. Physics of Fluids, 2005, 17, 113601-1–113601-12.

[8] Ramamurti R, Sandberg W C. A three-dimensional compu-tational study of the aerodynamic mechanisms of insect flight. Journal of Experimental Biology, 2002, 205, 1507–1518.


[9] Parker K, von Ellenrieder K, Soria J. Flow structures behind a heaving and pitching finite- span wing. Journal of Fluid Mechanics, 2003, 490, 129–138.

[10] Guerrero J. Numerical Simulation of the Unsteady Aerody-namics of Flapping Flight. PhD thesis, University of Genoa, Italy, 2009.

[11] Triantafyllou G S, Triantafyllou M S. An efficient swim-ming machine. Scientific American, 1995, 272, 40–48.

[12] Rohr J J, Fish F E. Strouhal number and optimization of swimming by odontocete cetaceans. Journal of Experimen-tal Biology, 2004, 207, 1633–1642.

[13] Triantafyllou M S, Triantafyllou G S, Gopalkrishnan R. Wake mechanics for thrust generation in oscillating foils. Physics of Fluids, 1991, 3, 2835–2837.

[14] Nudds R L, Taylor G K, Thomas A R. Tuning of Strouhal number for high propulsive efficiency accurately predicts how wingbeat frequency and stroke amplitude relate and scale with size and flight speed in birds. Proceedings of the Royal Society: Biological Sciences, 2004, 271, 2071–2076.

[15] Taylor G K, Nudds R L, Thomas A R. Flying and swimming animals cruise at a Strouhal number tuned for high power efficiency. Nature, 2003, 425, 707–711.

[16] Henshaw W D. A fourth-order accurate method for the incompressible Navier-Stokes equations on overlapping grids. Journal of Computational Physics, 1994, 113, 13–25.

[17] Chesshire G, Henshaw W. Composite overlapping meshes for the solution of partial differential equations. Journal of Computational Physics, 1990, 90, 1–64.

[18] Petersson N. Stability of pressure boundary conditions for stokes and Navier-Stokes equations. Journal of Computa-

tional Physics, 2001, 172, 40–70. [19] Sani R L, Shen J, Pironneau O, Gresho P M. Pressure

boundary condition for the time-dependent incompressible Navier-Stokes equations. International Journal for Nu-merical Methods in Fluids, 2006, 50, 673–682.

[20] Henshaw W D. Ogen: An Overlapping Grid Generator for Overture, Technical Report, Lawrence Livermore National Laboratory, USA, UCRL-MA-132237, 1998.

[21] Vinokur M. Conservation equations of gas-dynamics in curvilinear coordinate systems. Journal of Computational Physics, 1974, 14, 105–125.

[22] Roache P J. Verification and Validation in Computational Science and Engineering, Hermosa Publishers, Albuquerque, New Mexico, USA, 1998.

[23] Jeong J, Hussain F. On the identification of a vortex. Journal of Fluids Mechanics, 1995, 285, 69–94.

[24] Haimes R, Kenwright D. On the velocity gradient tensor and fluid feature extraction. The 14th Computational Fluid Dy-namics Conference, Norfolk, Virginia, USA, 1999, AIAA-1999-3288.

[25] Wang Z J. Vortex shedding and frequency selection in flapping flight. Journal of Fluid Mechanics, 2000, 410, 323–341.

[26] Young J, Lai J. Oscillation frequency and amplitude effects on the wake of a plunging airfoil. AIAA Journal, 2004, 42, 2042–2052.

[27] Parker K, Soria J, von Ellenrieder K. Thrust measurements from a finite-span flapping wing. AIAA Journal, 2007, 45, 58–70.

Wake Signature and Strouhal Number Dependence of Finite-Span Flapping Wings

Documents