Numerical simulations of flapping foil

and wing aerodynamics
Mesh deformation using radial basis functions
Copyright c 2009 by F.M. Bos
All rights reserved. No part of this material protected by this copyright notice
may be reproduced or utilized in any form or by any means, electronic or mechan-
ical, including photocopying, recording or by any other information storage and
retrieval system, without written permission from the copyright owner.
Printed by Ipskamp Drukkers B.V. in The Netherlands
ISBN: 978-90-9025173-8
An electronic version of this thesis is available at http://repository.tudelft.nl
Numerical simulations of flapping foil
and wing aerodynamics
Mesh deformation using radial basis functions
Proefschrift
ter verkrijging van de graad van doctor
aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus Prof. ir. K.C.A.M. Luyben,
voorzitter van het College voor Promoties,
in het openbaar te verdedigen op woensdag 24 februari 2010 om 10:00 uur
door
Frank Martijn BOS
ingenieur luchtvaart en ruimtevaart
geboren te Naaldwijk.
Dit proefschrift is goedgekeurd door de promotor:
Prof. dr. ir. drs. H. Bijl
Copromotor:
Dr. ir. B.W. van Oudheusden
Samenstelling promotiecommissie:
Rector Magnificus, voorzitter
Prof. dr. ir. drs. H. Bijl Technische Universiteit Delft, promotor
Dr. ir. B.W. van Oudheusden Technische Universiteit Delft, copromotor
Prof. dr. ir. P.G. Bakker Technische Universiteit Delft
Prof. dr. ir. B. Koren Universiteit Leiden
Centrum Wiskunde & Informatica
Prof. dr. H. Jasak Zagreb University
Prof. dr. F-O. Lehmann University of Ulm
Prof. dr. W. Shyy The University of Michigan
This research was supported by The Netherlands
Organisation for Scientific Research (NWO),
NWO-ALW grant 814.02.019.
Voor mijn ouders
Summary
Both biological and engineering scientists have always been intrigued by the flight
of insects and birds. For a long time, the aerodynamic mechanism behind flap-
ping insect flight was a complete mystery, until several decades ago. Experiments
showed the presence of a vortex on top of the flapping wings, generates forces
larger than obtained by using conventional aircraft aerodynamics. Flapping wings
produce both lifting and propulsive forces such that it becomes possible for insects
and smaller bird species, e.g. hummingbirds, to stay aloft and hover, but also to
perform extreme manoeuvres. Because of this versatility, insects and smaller birds
are an inspiration for the development of flapping wing Micro Air Vehicles, small
man-made flyer’s to use in exploration and surveillance.
Several flow visualisation experiments and numerical simulations have been
performed to improve the understanding of flapping wing aerodynamics in order
to design and optimise Micro Air Vehicles. However, the effects of wing kinemat-
ics on the flow and forces is still not fully understood. We performed two- and
three-dimensional numerical simulations in order to systematically vary relevant
parameters, related to the wing motion and flow physics. In order to capture the
boundary layer and the near wake, it is important to maintain a high mesh quality
near the moving wing, especially at large rotations. Therefore, an accurate mesh
motion technique is necessary, which is able to cope with large mesh deformations.
In order to incorporate a flapping wing in our numerical model, different mesh mo-
tion techniques are compared and improved. The overall goal of this part of the
research is to develop a reliable mesh deformation technique, in terms of accuracy
and efficiency, to solve the flow around flapping wings.
The flow around flapping wings, at the scale relevant to insect flight, is highly
unsteady and vortical, described by the unsteady incompressible Navier-Stokes
equations. Different dimensionless numbers are discussed, characterising the flow,
i.e. Strouhal and Reynolds numbers. Since the flow at the considered Reynolds
number, Re = O(100), is laminar, there is no need for additional turbulence
modelling, such that our simulations, assuming laminar flow, may be treated as a
iv Summary
Direct Numerical Simulation (DNS).
In order to solve the unsteady incompressible Navier-Stokes equations, the com-
mercial software Fluent

and the open-source code OpenFOAM

have been used
extensively. Different mesh motion solvers are compared. Two existing methods
are assessed, solving the Laplace and a modified stress equation. Both methods
are very efficient by using iterative solver techniques. However, these mesh motion
solvers are not able to maintain high mesh quality at large rotation angles, which
occur in insect flight. Therefore, a new mesh motion solver is implemented, which
is based on the interpolation of radial basis functions.
This mesh motion solver is a point based method, which means that the dis-
placement of all individual internal mesh points are evaluated, based on a given
boundary displacement, and updated accordingly. No mesh connectivity informa-
tion is necessary, so that it can be applied to unstructured polyhedral meshes.
To increase its efficiency, a coarsening is applied to the set of moving boundary
points, such that only selected control points are used. This decreases the size of
the system of equations and associated computational effort considerably.
After the discussion of the governing equations, finite volume discretisation in
OpenFOAM

and the assessment of the mesh motion solvers, the physical and
numerical modelling are described. The incompressible Navier-Stokes equations
are rewritten in the rotating reference frame in order to identify dimensionless
numbers related to the wing motion. The most important number is the Rossby
number, which represents the wing stroke path curvature.
First a two-dimensional study is performed to investigate the effects of differ-
ent wing kinematic models, with increasing complexity, on hovering flight perfor-
mance. The results show that the ‘sawtooth’ amplitude has a small effect on the
mean lift but the mean drag is affected significantly. The second model simplifica-
tion, the ‘trapezoidal’ angle of attack, caused the leading-edge vortex to separate
during the translational phase. This led to an increase in mean drag during each
half-stroke. The extra ‘bump’ in angle of attack as used by the fruit fly model is
not affecting the mean lift to a large extent. The other realistic kinematic feature
is the deviation, which is found to have only a marginal effect on the mean lift and
mean drag in this two-dimensional study. However, the effective angle of attack is
altered such that the deviation leads to levelling of the force distribution.
Additionally, a numerical model for two-dimensional flow was used to investi-
gate the effect of foil kinematics on the vortex dynamics around an ellipsoid foil
subject to prescribed flapping motion over a range of dimensionless wavelengths,
dimensionless amplitudes, angle of attack amplitudes, and stroke plane angles.
Both plunging and rotating motions are prescribed by simple harmonic functions
which are useful for exploring the parametric space despite the model simplicity.
Optimal propulsion using flapping foil exists for each variable which implies that
aerodynamics might select a range of preferable operating condition. The condi-
tions that give optimal propulsion lie in the synchronisation region in which the
flow is periodic.
Furthermore, different results relevant to three-dimensional flapping wing aero-
v
dynamics, are described. First, the flow around a dynamically scaled model wing
is solved for different angles of attack in order to study the force development
and vortex dynamics at small and large mid-stroke angle of attack. Secondly, the
Rossby number is varied at different Reynolds numbers. A varying Rossby number
represents a variation in stroke path curvature and thus angular acceleration. It is
shown that a low Rossby number is beneficial for the stability of the leading-edge
vortex, leading to an increase in lift and efficiency. Thirdly, the three-dimensional
wing kinematics is varied by changing the shape in angle of attack and by applying
a deviation, which may cause a ‘figure-of-eight’ pattern. As in two-dimensional
studies, the deviation may influence the force distribution to a large extent, by
changing the effective angle of attack. Additionally, the three-dimensional flow is
compared with the two-dimensional studies performed on flapping forward flight.
Finally, a preliminary investigation is performed to show the effect of wing
flexing. Therefore, a pre-defined flexing deformation is applied to a plunging airfoil
in two-dimensional forward flight and to a three-dimensional flapping wing in
hovering flight. Concerning the flexible airfoil in forward flight, a similar behaviour
was observed as for a rigid plunging airfoil, subjected to additional rotation.
The present simulations have led to important insight to understand the influ-
ence of wing kinematics and deformation on the aerodynamic performance. These
results may be important to design and optimise Micro Air Vehicles.
Contents
Summary iii
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Physics of flapping flight . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Experimental and numerical methods . . . . . . . . . . . . . . . . 7
1.4 Objectives and approach . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Finite volume discretisation 13
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 The Navier-Stokes equations . . . . . . . . . . . . . . . . . . . . . . 16
2.2.1 Constitutive relations . . . . . . . . . . . . . . . . . . . . . 17
2.2.2 Incompressible laminar flow simplifications . . . . . . . . . 18
2.2.3 Dimensionless numbers . . . . . . . . . . . . . . . . . . . . 18
2.3 Spatial and temporal discretisation . . . . . . . . . . . . . . . . . . 19
2.4 Measures of cell quality . . . . . . . . . . . . . . . . . . . . . . . . 20
2.5 Discretisation of an incompressible momentum equation . . . . . . 21
2.5.1 Face interpolation schemes . . . . . . . . . . . . . . . . . . 23
2.5.2 Convection term . . . . . . . . . . . . . . . . . . . . . . . . 23
2.5.3 Diffusion term . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.5.4 Temporal term . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.6 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.7 Solution of the Navier-Stokes equations . . . . . . . . . . . . . . . 27
2.7.1 Pressure equation and Pressure-Velocity coupling . . . . . . 28
2.7.2 Procedure for solving the Navier-Stokes equations . . . . . 29
2.7.3 Arbitrary Lagrangian Eulerian approach . . . . . . . . . . . 30
viii Contents
2.8 Swept volume calculation . . . . . . . . . . . . . . . . . . . . . . . 32
2.9 Numerical flow solvers . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.10 Code validation and verification . . . . . . . . . . . . . . . . . . . . 34
2.10.1 2D vortex decay and convection . . . . . . . . . . . . . . . . 35
2.10.2 Validation using cylinder flows . . . . . . . . . . . . . . . . 39
2.11 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3 Mesh deformation techniques for flapping flight 49
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.2 Different mesh deformation techniques . . . . . . . . . . . . . . . . 52
3.2.1 Laplace equation with variable diffusivity . . . . . . . . . . 53
3.2.2 Solid body rotation stress equation . . . . . . . . . . . . . . 54
3.2.3 Radial basis function interpolation . . . . . . . . . . . . . . 55
3.3 Mesh quality measures . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.4 Comparison of mesh motion solvers . . . . . . . . . . . . . . . . . . 59
3.4.1 Translation and rotation of a two-dimensional block . . . . 59
3.4.2 Flapping of a three-dimensional wing . . . . . . . . . . . . . 62
3.4.3 Flexing of a two-dimensional block . . . . . . . . . . . . . . 64
3.5 Improving computational efficiency . . . . . . . . . . . . . . . . . . 66
3.5.1 Boundary point coarsening and smoothing . . . . . . . . . . 67
3.5.2 Iterative techniques and parallel implementations . . . . . . 69
3.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4 Physical and numerical modelling of flapping foils and wings 73
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.2 Governing equations for flapping wings . . . . . . . . . . . . . . . . 75
4.3 Wing shape and kinematic modelling . . . . . . . . . . . . . . . . . 78
4.3.1 Wing shape and planform selection . . . . . . . . . . . . . . 79
4.3.2 Kinematic modelling . . . . . . . . . . . . . . . . . . . . . . 80
4.3.3 Modelling of active wing flexing . . . . . . . . . . . . . . . . 82
4.3.4 Numerical implementation of the wing kinematics . . . . . 83
4.4 Dynamical scaling of flapping wings . . . . . . . . . . . . . . . . . 86
4.5 Computational domain and boundary conditions . . . . . . . . . . 88
4.6 Definition of force and performance coefficients . . . . . . . . . . . 89
4.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5 Influence of wing kinematics in two-dimensional hovering flight 93
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.1.1 Similarity and discrepancy between two- and
three-dimensional flows . . . . . . . . . . . . . . . . . . . . 94
5.1.2 Influence of kinematic modelling . . . . . . . . . . . . . . . 94
5.2 Numerical simulation methods . . . . . . . . . . . . . . . . . . . . 96
5.2.1 Flow solver and governing equations . . . . . . . . . . . . . 96
5.2.2 Mesh generation and boundary conditions . . . . . . . . . . 97
Contents ix
5.2.3 Validation using harmonic wing kinematics . . . . . . . . . 99
5.3 Modelling insect wing kinematics . . . . . . . . . . . . . . . . . . . 99
5.3.1 Insect wing selection and model parameters . . . . . . . . . 100
5.3.2 Dynamical scaling of the wing model . . . . . . . . . . . . . 101
5.3.3 Force and performance indicators . . . . . . . . . . . . . . . 102
5.3.4 Different wing kinematic models . . . . . . . . . . . . . . . 103
5.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.4.1 Overall model comparison . . . . . . . . . . . . . . . . . . . 105
5.4.2 Kinematic features investigation . . . . . . . . . . . . . . . 109
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
6 Vortex wake interactions of a two-dimensional flapping foil 121
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
6.2 Flapping foil parametrisation . . . . . . . . . . . . . . . . . . . . . 122
6.3 Force coefficients and performance . . . . . . . . . . . . . . . . . . 124
6.4 Numerical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.5 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . 126
6.5.1 Influence of dimensionless wavelength . . . . . . . . . . . . 126
6.5.2 Influence of dimensionless amplitude . . . . . . . . . . . . . 127
6.5.3 Influence of angle of attack amplitude . . . . . . . . . . . . 129
6.5.4 Influence of stroke plane angle . . . . . . . . . . . . . . . . 130
6.5.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
6.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
7 Vortical structures in three-dimensional flapping flight 135
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
7.2 Three-dimensional flapping wing simulations . . . . . . . . . . . . 137
7.2.1 Modelling and parameter selection . . . . . . . . . . . . . . 138
7.2.2 Simulation strategy and test matrix selection . . . . . . . . 141
7.3 Flow solver accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . 143
7.4 Vortex identification methods for flow visualisation . . . . . . . . . 145
7.5 Flapping wings at low Reynolds numbers . . . . . . . . . . . . . . 148
7.5.1 The angle of attack in flapping flight . . . . . . . . . . . . . 149
7.5.2 Influence of flapping stroke curvature . . . . . . . . . . . . . 149
7.5.3 Influence of Reynolds number . . . . . . . . . . . . . . . . . 152
7.5.4 Influence of ‘trapezoidal’ angle of attack . . . . . . . . . . . 158
7.5.5 Influence of deviation . . . . . . . . . . . . . . . . . . . . . 160
7.6 Flapping wings in forward flight . . . . . . . . . . . . . . . . . . . . 163
7.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
8 Influence of wing deformation by flexing 171
8.1 Airfoil flexing in two-dimensional forward flapping flight . . . . . . 171
8.2 Wing flexing in three-dimensional hovering flight . . . . . . . . . . 175
8.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
x Contents
9 Conclusions and recommendations 179
9.1 Overall conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
9.2 Conclusions on mesh motion techniques . . . . . . . . . . . . . . . 180
9.3 Conclusions on hovering flapping flight . . . . . . . . . . . . . . . . 181
9.3.1 Two-dimensional hovering . . . . . . . . . . . . . . . . . . . 181
9.3.2 Three-dimensional hovering . . . . . . . . . . . . . . . . . . 182
9.4 Conclusions on forward flapping flight . . . . . . . . . . . . . . . . 184
9.4.1 Two-dimensional forward flapping . . . . . . . . . . . . . . 184
9.4.2 Three-dimensional forward flapping . . . . . . . . . . . . . 184
9.5 Preliminary conclusions on wing flexing . . . . . . . . . . . . . . . 185
9.6 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
A Grid generation for flapping wings 189
A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
A.2 BlockMesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
A.3 Gambit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
A.4 GridPro

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
A.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
B Flow solver settings 195
B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
B.2 Fluent

solver settings . . . . . . . . . . . . . . . . . . . . . . . . . 195
B.3 OpenFOAM

solver settings . . . . . . . . . . . . . . . . . . . . . 197
Bibliography 201
Samenvatting 213
Acknowledgements 217
Curriculum Vitae 219
CHAPTER 1
Introduction
1.1 Motivation
The year of writing, 2009, is known as the year of Charles Darwin (1809-1882),
since he was born 200 years ago. At the age of 50, he published his world famous
Origin of Species. That book describes the natural selection, inspired by his scien-
tific observations during a voyage (1831-1836) around the world with his ship, the
Beagle. At the Gal´ apagos Archipelago, Darwin discovered slightly different bird
species living on the different islands, whereas, he only knew one species on the
mainland of South-America. Apparently, the mainland bird species had travelled
to the islands and on every different island it adapted to the differences in environ-
mental circumstances. This process has become known as natural selection, which
is also applicable to the early era of flight, millions of years ago. Birds are ancient
descendants of feathered dinosaurs (Templin, 2000) which developed the skill of
flight in order to migrate over large distances and to catch prey. Long before the
origin of dinosaurs and birds, insects adapted to leave the ground to take off into
the thin air. Birds and insects are both flapping their wings at different length
scales, leading to a different flow behaviour. The larger the animal, the lower the
need for flapping wings, e.g. the Andean Condor only flaps when it looses height
in the thermal winds, whereas a small insect, a fruit fly (Weish-Fogh & Jensen,
1956) flaps with about 200 times per second.
Flapping wings produce both lifting and propulsive forces, such that it becomes
possible for insects and even smaller bird species, e.g. hummingbirds, to stay aloft
and hover, but also to perform extreme manoeuvres. Because of this versatility,
insects and smaller birds are a major inspiration of study to develop Micro Air
Vehicles (MAV), tiny man-made flyer’s to use in exploration and surveillance, see
2 Introduction
figure 1.1.
To optimise the flight performance of MAV’s it is important to get a thorough
understanding of the complex flow generated by its wings, especially at smaller
length scales (< 5 cm). The flapping wings induce complicated vortical struc-
tures which influence the forces and performance characteristics in hovering and
forward flight. In order to study this kind of flows, researchers performed flow
visualisations (Weish-Fogh & Jensen, 1956, Srygley & Thomas, 2002), detailed
experiments (Ellington et al., 1996, Sane & Dickinson, 2002, Poelma et al., 2006,
Lentink & Dickinson, 2009b) and numerical simulations (Wang et al., 2004, Sun
& Tang, 2002, Bos et al., 2008, Thaweewat et al., 2009). One limitation of doing
experiments is that they can be very expensive, in view of the need of precision
equipment and wind tunnel facilities. Secondly, the construction of models needs
to be very precise, which can be very costly as well. Additionally, when performing
wind tunnel experiments it is not straightforward to extract the force data, either
directly or indirectly, from the flow visualisation obtained by Particle Image Ve-
locimetry (PIV) (Poelma et al., 2006). Even when the most advanced techniques
are used, e.g. Digital Particle Image Velocimetry (DPIV), the flow field can not
be visualised in much detail, especially due to the reflections and shadows of the
moving wings. On the other hand, when performing numerical simulations, using
Computational Fluid Dynamics (CFD), the forces and flow visualisations are a
direct result of the computations. Since it is interesting to solve for the forces
acting on a flapping wing in combination with the vortical structures within the
near wake, performing CFD provides a suitable framework.
The present study deals with the development and improvement of computa-
tional techniques to solve the flow around flapping wings at low Reynolds num-
bers, O(100 − 1000). Section 1.2 briefly provides background information on the
flow physics concerned, while section 1.3 deals with the different approaches for
analysing the flow, experiments and numerical methods. Finally, section 1.4 de-
scribes the objectives and approach of the present study as well as the outline of
this thesis.
1.2 Physics of flapping flight
In order to illustrate the necessity and difficulties with solving and visualising the
flow around flapping foils and wings, different aspects of flapping wing aerody-
namics are discussed. The vortex dynamics, the leading-edge vortex in particular,
is briefly discussed, as well as the influence of the wing kinematic modelling in
two- and three-dimensional problems.
Vortex generation in flapping wing aerodynamics
Vortex generation in nature is fairly common in flows induced by aeroplanes, birds,
insects, but also by boats and trees. Large aeroplanes generate wingtip vortices,
see figure 1.2, which can cause damage to a following aeroplane which encounters
1.2 Physics of flapping flight 3
(a) Wasp. (b) Entomopter. (c) Delfly.
Figure 1.1 Different flapping wing Micro Air Vehicle concepts. At lower Reynolds numbers,
flapping MAV concepts can be used for hovering and low speed forward flight, which is especially
interesting for intelligence and exploration. (a) Flying insect scale robotic model, which is able to
perform a tethered take-off (Wood, 2008). (b) The U.S. patented Entomopter has four flapping wings
powered by chemically-fuelled propulsion system (Michelson, 2008). (c) The Delfly Micro is camera
equipped and is able to hover (designed and developed at Delft University of Technology).
(a) (b) (c)
Figure 1.2 Vortex induced force generation. (a) Wingtip vortex causes big disturbances in the
wake, limiting the time between two successive aeroplane approaches. (b) Vortex generation in insect
flight. A water strider generates vortices with its long legs to create the necessary propulsion (Hu et al.,
2003). (c) Willmott et al. (1997) performed smoke visualisation of the vortical flow patterns induced
by a hawkmoth. It was observed that the leading-edge vortex was stabilised by the radial flow moving
out towards the wing tip. Additionally, alternating vortex rings were seen in the wake, generated by
successive up and downstrokes.
this vortex. Another undesired effect of vortex generation is flow induced vibra-
tion of cables, bridges or struts in water. On the other hand, vortex generation
provides possibilities to generate forces, which is used by birds, fish and insects,
e.g. figure 1.2 shows induced propulsive vortices generated by a water strider.
It is a common story that flies could not fly according to conventional aircraft
theory as developed by Lanchester (1907) and Prandtl (1914-1918). Prandtl did
develop a relation between the tip vortices, circulation and lift generation, but
this was not sufficient to explain the high lift generation of insects. This mystery
persisted until the discovery of the unsteady vortical flow field, figure 1.2, and
especially the generation of the leading-edge vortex.
4 Introduction
¯
C
L
= 1.540
x
c
[−]
-3 -2 -1 0 1 2 3
(a) (b)
Figure 1.3 Forces and vortices in flapping wing aerodynamics. (a) Two-dimensional il-
lustration of the wing kinematics and the resulting force vector generated by the flapping airfoil at
Re = O(100) (Bos et al., 2008). (b) Three-dimensional leading-edge vortex generated by a flapping
wing at Re = O(1000).
Leading-edge vortex
The potential benefit of vortices attached to the wing was already discussed
by Maxworthy (1979) and Dickinson & G¨otz (1993). It was Ellington et al. (1996)
who identified the presence of a leading-edge vortex (LEV) generated on top of the
flapping wing, increasing the lift force to values much higher than predicted by con-
ventional wing theory. The stability of the helical three-dimensional leading-edge
vortex is still not yet fully understood and appears to heavily depend on the wing
kinematics and Reynolds number. It appears that the leading-edge vortex is more
stable around a three-dimensional flapping wing compared to two-dimensional
flapping foil situations.
In flapping foil aerodynamics the vortices are shed and form either a periodic
or chaotic wake pattern, depending on the kinematics, notably advance ratio and
dimensionless flapping amplitude (Thaweewat et al., 2009, Lentink et al., 2008).
The origin of the leading-edge vortex is the roll-up of shear layers, present in
highly viscous flows, which is the case at low Reynolds numbers. It is thought that
the kinematics in two- and three-dimensional flapping influences the shear layer
direction and flow accelerations, which will undoubtedly influence the development
of the leading-edge vortex (Lentink & Dickinson, 2009b).
In order to understand the physics of flapping wing aerodynamics, it is im-
portant to obtain insight in how an insect moves its wing. Figure 1.3(a) shows
a two-dimensional illustration of the wing kinematics of a fruit fly, operating at
Re = 110, from (Bos et al., 2008).
Influence of insect wing kinematics on forces
The relevance of experiments and flow simulations of insect flight has been found
to depend on how reliably true insect wing kinematics are reproduced. Wang
1.2 Physics of flapping flight 5
et al. (2004) and Sane & Dickinson (2001) showed that the kinematic modelling
significantly influences the mean force coefficients and its distribution. Addition-
ally, Hover et al. (2004) showed that the angle of attack influences the flapping
foil propulsion efficiency to a large extent. This illustrates the appreciable effects
which details of the wing kinematics, like parameter values and stroke patterns,
may have on flight performance. It further emphasises the need to critically assess
the influence of kinematic model simplifications.
In literature, different kinematic models have been employed to investigate the
aerodynamic features of insect flight. For example, Wang (2000a,b) and Lentink
& Gerritsma (2003) numerically investigated pure harmonic translational motion
with respectively small and large amplitudes. Wang (2000a,b) varied flapping am-
plitude and frequency and showed that at a certain parameter selection the lift
is clearly enhanced. Lewin & Haj-Hariri (2003) performed a similar numerical
study for heaving airfoils. Besides lift enhancement at certain reduced frequen-
cies, they found periodic and aperiodic flow solutions which are strongly related
to the aerodynamic efficiency. Lentink & Gerritsma (2003) varied airfoil shape
with amplitude and frequency fixed at values representative to real fruit flies.
They concluded that the airfoil geometry choice is of minor influence, but large
amplitudes lead to an increase of lift by a factor of 5 compared to static forces
generated by translating airfoils. It was also shown that wing stroke models with
only translational motion could not provide for realistic results, such that includ-
ing rotation is essential. In addition to the harmonic models with pure translation
(Dickinson & G¨otz, 1993), rotational parameters were investigated by Dickinson
(1994). They varied rotational parameters and showed that axis-of-rotation, rota-
tion speed and angle of attack during translation are of great importance for the
force development during each stroke. Harmonic wing kinematics, including wing
rotation, where used by Pedro et al. (2003) and Guglielmini & Blondeaux (2004)
in their numerical models to solve for forward flight. Both studies emphasised the
importance of angle of attack to influence the propulsive efficiency. Slightly more
complex fruit fly kinematic models were used by Dickinson et al. (1999) and Sane
& Dickinson (2001) with their Robofly. Based on observation of true insect flight,
the wing maintains a constant velocity and angle of attack during most of the
stroke, with a relatively strong linear and angular acceleration during stroke re-
versal. This results in the typical ‘sawtooth’ displacement and ‘trapezoidal’ angle
of attack pattern of the Robofly kinematic model. Using these models, the effect
of amplitude, deviation, angle of attack and the timing of the latter were explored.
The present thesis deals with different kinematic models from literature, both
the pure harmonic and the Robofly model, in order to investigate their influence
on the aerodynamic performance (Bos et al., 2008). Furthermore, the results were
compared with more realistic fruit fly kinematics obtained from the observation of
free flying fruit flies. Instead of performing a parameter study within the scope of
one kinematic model, the objective of the present study is to compare the effect
6 Introduction
(a) (b) (c)
Figure 1.4 Vortex wakes generated by cylinders and flapping wings. (a) Von K´ arm´ an vortex
street behind a stationary cylinder at Re = 150. (b) Periodic vortex wake behind a plunging airfoil at
Re = 110, one single and one vortex pair is generated each plunging period. (c) Chaotic vortex wake
behind a plunging airfoil at Re = 110, depending on the kinematics a chaotic wake pattern may occur
with unpredictable forces as the result.
of the available models as a whole. This leads to better insights into the conse-
quences of simplifications in kinematic modelling, which is of great importance to
both experiments and numerical simulations. Also, it may reveal the importance
of certain specific features of the stroke pattern, in relation to aerodynamic per-
formance.
The similarity between two- and three-dimensional flows
To limit both the parametric space as well as the computational effort, many stud-
ies have been performed as two-dimensional simulations (Thaweewat et al., 2009,
Bos et al., 2008, Wang et al., 2004, Lewin & Haj-Hariri, 2003). One of the major
(and partially unresolved) issues in modelling of insect flight and flapping wing
propulsion, is the possibly restrictive applicability of two-dimensional results to
true insect flight. Additional important aspects are unsteady flow mechanisms,
wing flexibility (fluid structure interaction) and Reynolds number effects. In a
recent paper Wang et al. (2004) compared three-dimensional Robofly results with
two-dimensional numerical results. Both Dong et al. (2005) and Blondeaux et
al. (2005b) concluded that two-dimensional studies over predict forces and per-
formances since the energy-loss, which is present in three dimensions, is not re-
solved. Dong et al. (2005), Blondeaux et al. (2005b) numerically investigated the
wake structure behind finite-span wings at low Reynolds numbers. They observed
three-dimensional vortical structures around flapping wings with low aspect ratio,
as was mentioned by Lighthill (1969).
Notwithstanding the possible discrepancy between two- and three-dimensional
flow, two-dimensional analysis has often been applied to obtain insight into the
aerodynamic effects of wing kinematics and geometry. Wang et al. (2004) con-
firmed that the similarities between two- and three-dimensional approaches are
sufficient to warrant that a reasonable approximation of insect flight can be ob-
tained using a two-dimensional approach. First, in case of advanced and symmet-
ric rotation the forces were found to be similar in the two-dimensional simulations
compared to the three-dimensional experiments. Secondly it was observed that
in both simulations and experiments the leading-edge vortex did not completely
separate for amplitude-to-chord ratios between 3-5 (Dickinson & G¨otz, 1993, Dick-
1.3 Experimental and numerical methods 7
inson, 1994). The current research deals with amplitudes that are in this range.
In view of the excessive computational expense required to perform accurate
three-dimensional simulations, and with the above justification, the first part of
this thesis makes extensive use of two-dimensional simulations. In the second
part, various three-dimensional simulations were performed using limited varia-
tions wing kinematics.
1.3 Experimental and numerical methods
In literature, different methods were used to solve and visualise the flow around
flapping insect wings, from realistic fruit fly measurements to three-dimensional
simulations using a representative model wing. In this section, different methods
will be briefly addressed, from experimental methods to computational fluid dy-
namics simulations.
Experimental investigations and quasi steady theory
Several experimental studies considered the flight performance of insects, and re-
vealed the complex nature of insect flight aerodynamics. The flow induced by the
motion of insect wings is highly unsteady and vortical, as visualised by Weish-
Fogh & Jensen (1956) using tethered locusts and by Willmott et al. (1997) using
a hawkmoth (Manduca Sexta), see figure 1.2. More recently, Srygley & Thomas
(2002) and Thomas et al. (2004) performed free flight and tethered experimental
visualisations using butterflies and dragonflies to show the complicated vortical
structures. This unsteady and vortical flow behaviour is a consequence of the
high relative frequencies, amplitudes and the very low Reynolds number involved
(Re < 1000 for a large number of insects and Re ≈ 110 for the fruit fly, Drosophila
Melanogaster, in particular).
Ellington (1984) indicated that the lift in insect flight is significantly higher
than expected on the basis of quasi-steady aerodynamics, hence revealing that
important unsteady and vortical flow phenomena play a major role in insect flight.
In several studies (Dickinson & G¨otz, 1993, Dickinson, 1994, Dickinson et al.,
1999) it was confirmed that important aspects, like delayed stall and wake capture
enhance the lift force beyond values predicted by quasi-steady theory. Ellington
et al. (1996) discovered that these lift increasing mechanisms are amplified by
the generation of a leading-edge vortex (LEV). It was shown that this leading-
edge vortex arises during the translational part of the wing motion rather than
during the rotational flip between up and down stroke. The lift increasing effect of
the leading-edge vortex strongly depends on the kinematics of the flapping wing
(Dickinson et al., 1999, Wang, 2000b, Sane & Dickinson, 2001, 2002, Lentink &
Dickinson, 2009a,b).
In order to understand insect flight performance Dickinson et al. (1999) and
Wang (2000b) applied the quasi-steady theory to compare with unsteady forces.
The quasi-steady approach was revised by Sane & Dickinson (2002) to include ro-
8 Introduction
tational effects but even then the results require further improvement. According
to Sane & Dickinson (2001) the mean lift is well predicted by quasi-steady theory,
but the mean drag is underestimated. This confirms the restricted applicability of
the quasi-steady theory due to lack of unsteady mechanisms like rotational lift and
wake capture. Several experimental studies have been performed with the aim of
characterising the unsteady aerodynamics of insect flight. Dickinson et al. (1999)
investigated the flow around a flapping Robofly model which moves in oil to meet
the same flow conditions as the real fruit fly encounters (reproduction of Reynolds
number in particular).
Numerical simulations
Notwithstanding important advances in experimental techniques for non-intrusive
flow field analysis, Particle Image Velocimetry in particular (Bomphrey et al., 2006,
Poelma et al., 2006), it remains difficult to capture all the relevant details of the
flow using only experimental techniques. An appealing approach, therefore, is to
supplement experiments with numerical flow simulations. A number of numerical
studies on full three-dimensional configurations have been reported, in relation to
specific insect geometries (moth: Liu & Kawachi (1998), fruit fly: Ramamurti &
Sandberg (2002), Sun & Tang (2002), dragonfly: Young & Lai (2008), Isogai et al.
(2004)).
To perform numerical simulations around moving objects, such as flapping
wings, one can use either immersed boundary methods (Peskin, 2002, Mittal &
Iaccarino, 2005), deforming mesh techniques (Boer de et al., 2007, Jasak, 2009),
see figure 1.5, or even complete re-meshing (Young & Lai, 2008, Zuo et al.,
2007). Although, the computational effort involved in three-dimensional stud-
ies is presently still extremely demanding, an integrated computational study was
performed by Aono et al. (2008) who developed a code to incorporate two wings
and a body using overset mesh techniques. In an immersed boundary method, the
moving boundary is projected on a fixed Cartesian background grid, which is not
allowed to deform. Besides interpolation issues, the conservation of mass and mo-
mentum in current immersed boundary methods is not obvious, even not for fixed
boundaries (Mittal & Iaccarino, 2005). Nevertheless, when two wings touch, as in
the manoeuvre clap-and-fling (Weish-Fogh & Jensen, 1956), one will undoubtedly
need methods like overset, immersed boundary or re-meshing techniques.
Together with the unavailability of an accurate flow solver with parallel sup-
port, it was chosen to assess and improve existing mesh motion techniques. The
commonly used mesh motion techniques result in high quality meshes as long as
the rotation of the moving boundaries is limited. In order to cope with high ro-
tation rates, mesh motion based on radial basis function (RBF) interpolation is
implemented in this thesis and improved in terms of accuracy and efficiency. This
modern mesh motion technique is incorporated in OpenFOAM
1
, which is an
open-source framework to solve the Navier-Stokes equations on three-dimensional
1
OpenFOAM

is a registered trade mark of OpenCFD

Limited, the producer of the
OpenFOAM

software.
1.4 Objectives and approach 9
(a) (b)
Figure 1.5 Different mesh motion solvers. Two illustrations of mesh motion solutions, (a) shows
a Laplacian mesh motion, while (b) shows the mesh motion obtained by using radial basis function
interpolation.
unstructured grids of polyhedral cells with full parallel support. This code is thor-
oughly tested and used for flapping foil and wing simulations.
Arbitrary Lagrangian-Eulerian formulation
The governing equations to solve the flow are generally discretised using the Eule-
rian description, where the fluid is allowed to flow through the fixed mesh. This is
in contrast to the Lagrangian formulation, where the mesh is fixed to the fluid or
material. If the material or fluid deforms, the mesh deforms with it. This method
is commonly used to discretise the governing equations encountered in structure
mechanics. However, when the flow domain moves or deforms in time due to
a moving boundary, a fixed mesh becomes inconvenient, because it requires the
explicit tracking of the domain boundary. Therefore, the Arbitrary Lagrangian
Eulerian (ALE) formulation is used to discretise the flow equations on moving and
deforming meshes (Donea, 1982). This method incorporates and combines both
Lagrangian and Eulerian frameworks. The Lagrangian contribution allows the
mesh to move and deform according to the boundary motion, whereas the Eule-
rian part takes care of the fluid flow through the mesh. At the time of writing, the
ALE method has become the standard implementation in most popular codes to
solve for the flow around moving boundaries while the mesh deforms accordingly.
1.4 Objectives and approach
Flapping flight aerodynamics is governed by many parameters, like advance ratio,
wing kinematics, Reynolds number, etc. In order to perform accurate numerical
simulations it is important to use an efficient code which is capable to solve for
10 Introduction
various conditions, using realistic wing kinematics. For large wing translations and
rotations the numerical grid needs to deform accordingly to maintain high accuracy
of the flow solver. Therefore, the overall goal of this research is to develop a reliable
mesh deformation technique, in terms of accuracy and efficiency, to solve the flow
around flapping wings. This method is used to study the complex vortical patterns
to identify optimal strategies in flapping foil and wing aerodynamics. In order to
satisfy this aim, the following objectives are defined:
1. improve current mesh motion techniques and implementation, using an ac-
curate and efficient framework,
2. validate and verify the numerical solver with the implemented and improved
mesh motion technique,
3. solve for the flow around two-dimensional flapping foils to study the wing-
wake interaction as well as the influence of wing kinematics,
4. solve for the flow around three-dimensional flapping wings to assess the im-
portance of parameters like flapping amplitude, frequency or Reynolds num-
ber,
5. study the three-dimensional structure of vortical patterns, especially the
leading-edge vortex.
Approach and outline
In order to solve for the flow around flapping foils and wings, an improved mesh
motion technique, based on radial basis function interpolation, is implemented in
the open-source framework OpenFOAM

. The mesh motion technique is used
by an incompressible unsteady CFD solver to solve for the flow around a three-
dimensional flapping wing on dense meshes in parallel.
To meet the objectives the current thesis is structured in the following chapters. In
order to solve the governing equations for fluid flow, a finite volume discretisation is
used, which is the subject of chapter 2. That chapter deals with the discretisation
of the different terms as well as a definition of mesh quality. Furthermore, the solu-
tion procedure is described together with a brief discussion about the open-source
framework OpenFOAM

, which is thoroughly validated and verified. Within the
code, different mesh deformation techniques are incorporated. These mesh defor-
mation techniques are described and assessed with respect to accuracy in chapter 3.
In addition to the already implemented mesh deformation techniques, a method
based on radial basis function interpolation is discussed. This mesh deformation
technique is implemented and used for flapping foil and wing aerodynamics. Before
proceeding to the numerical results of the flow around flapping foils and wings, it
is important to discuss the numerical modelling for flapping flight in chapter 4.
Chapters 5 and 6 deal with the numerical investigations of two-dimensional flow
around a flapping foil in hovering and forward flight conditions, respectively. It is
1.4 Objectives and approach 11
found that the kinematic modelling has a large influence on forces, performance
and wake patterns. These two-dimensional results provide good insight what to
expect of the three-dimensional flow around a flapping wing, which is discussed
in chapter 7. Additionally, chapter 8 presents the preliminary results of a flexing
wing in two- and three-dimensional flapping flight. Complex vortical structures
induced by a model flapping wing can be accurately solved and analysed. It will
be seen that accurately solving the flow around a flapping wing is not an easy
task when the wing performs complex rotational motion. The conclusions and
recommendations can be found in chapter 9.
CHAPTER 2
Finite volume discretisation
A second-order finite volume discretisation of the incompressible Navier-Stokes
equations on arbitrary polyhedral meshes is described. In addition to the mesh
quality measures non-orthogonality and skewness, the boundary conditions and the
solution procedure are presented. This finite volume approach is applicable to gen-
eral commercial and non-commercial CFD codes. The commercial code Fluent

and the open-source code OpenFOAM

have been used for the simulations de-
scribed in this thesis. Fluent

was already tested by Bos et al. (2008), such that
this chapter deals with the validation of OpenFOAM

using problems relevant for
low Reynolds number insect flight. For test problems involving vortex decay and
convection, it was found that the Van Leer flux limiter provides the most accurate
results, since the flow is dominated by the convection of vortices. Furthermore,
the flow around stationary and transversely oscillating cylinders showed that the
code of OpenFOAM solves the flow in detail. Spatial and temporal convergence
was proved as well.
2.1 Introduction
Important aspects concerning numerical simulations, are being described. A nu-
merical simulation needs to be performed in an accurate and concise way. There-
fore, different properties of a CFD simulation, like stability and convergence, are
addressed.
Important aspects of numerical simulations
Before describing the used methods in detail, four important aspects of numerical
14 Finite volume discretisation
(a) Structured (b) Unstructured
Figure 2.1 Different mesh generation methods. Meshes can be generated in an structured way
(a) and using unstructured methods (b).
simulations are discussed, the governing equations, the discretisation method, the
numerical grid, and the solution method to solve the system.
In order to solve the incompressible Navier-Stokes equations, a suitable method
to discretise these equations needs to be chosen. In the field of CFD, three methods
are commonly used, the finite difference, finite element and finite volume method.
Wesseling (2001) and Ferziger & Peric (2002) described these methods in more
detail. Traditionally, finite element methods are used for structural problems
whereas numerical simulations related to fluid flow are mostly solved with finite
volume methods, as is the case in the current thesis. When using the finite volume
method, the interpolation from cell centres to cell faces and how to approximate
the surface and volume integrals, needs to be described.
The third aspect, concerning a CFD simulation, is the generation of a numerical
grid, a division of the computational domain in a finite amount of cells. There are
three types of grids, structured, block-structured and unstructured grids (Ferziger
& Peric, 2002). Figure 2.1 shows an example of a structured and an unstructured
grid. When using (block-) structured grids, the cell ordering is fairly straight-
forward such that the flow solver uses this fact to solve the system in a more
efficient way. A drawback of a (block-) structured grid is that it is more difficult
to create around complex geometries (commonly encountered in engineering prob-
lems). This is the more important asset of unstructured grids. Besides the type
of grid, the cell shape can be varied from tetrahedral (three corners in two dimen-
sions), hexahedral (four corners) to polyhedral (arbitrary number of corners) cells.
However, for less complex geometries, a structured grid is favourable in terms of
accuracy and efficiency of the flow solver. Besides the spatial discretisation, the
time is discretised as well, which is necessary to perform unsteady simulations.
2.1 Introduction 15
Finally, the fourth aspect of a CFD simulation is the iterative solver. The dis-
crete system of equations needs to be solved up to a certain convergence criterion.
Depending on the governing equations, discretisation method and the choice of
grid, the system of discretised equations can be easy or difficult to solve, limiting
the iterative solver. When an appropriate iterative solver is used, a convergence
criterion needs to be applied for the inner (within the linear system) and outer
iterations (to couple the non-linear parts and perform non-orthogonal corrections).
Properties of numerical solution methods
In order to solve the governing equations in a satisfactory manner, it is important
to discuss different properties of the numerical solution method, consistency, con-
vergence, stability, conservation and boundedness, from (Ferziger & Peric, 2002).
Since it is often not possible to find a numerical method which outperforms on
all aspects, the choice of numerical method is usually a trade-off. The following
properties are relevant concerning numerical simulations, especially when using a
finite volume approach, applied in general commercial and non-commercial CFD
solvers.
The first important property is consistency. The discretisation should become
exact when the mesh resolution tends to infinity, i.e. when the cell size approaches
zero. The difference between the discretised and the exact solution is called the
truncation error. In order to check the consistency of the complete numerical
scheme, a grid and time-step convergence study has to be performed using in-
creasing grid resolution and decreasing time-step. The second important property
of a numerical scheme is convergence. The solution of the discretised system of
equations should tend to the exact solution of the governing differential equations
as the mesh spacing tends to zero. Convergence is difficult to prove theoretically
for real engineering applications, so commonly the empirical approach is followed,
where the same computation is repeated on subsequently refined meshes. When
the solution converges to a grid-independent solution, the solution process is said
to be converged. However, it may happen that the exact solution is not approx-
imated with decreasing time-step, when the method is not stable. Therefore,
stability is the third important aspect. When performing the iterations of the nu-
merical process it should be the case that the numerical errors are not amplified.
In that case the solution process is called stable. For general engineering prob-
lems the stability of the numerical process is strongly dependent on the time-step;
it should be sufficiently small, depending on the temporal discretisation scheme.
The fourth property of a numerical scheme is conservation. Considering a steady
problem, without sources or sinks, the mass flux of a conserved quantity through
a specified system should be zero. Since the governing equations in finite volume
formulation are conservative, this property should be respected by the discretised
equations. One of the advantages of the finite volume approach is that conserva-
tion is guaranteed for every small control volume and therefore, for the complete
computational domain as a whole. Finally, the last aspect is boundedness. Certain
variables in the governing equations contain physical bounds, like concentration or
16 Finite volume discretisation
density and all other non-negative variables. When the numerical process respects
these physical bounds, the method is called bounded.
In this thesis, two different CFD codes, the commercial flow solver Fluent

and the
non-commercial open-source code OpenFOAM

are used. Fluent

is a general-
purpose CFD solver, which has been an authority in the field of computational
fluid dynamics for decades. Two major drawbacks of a commercial solver are the
unavailability of the source code and the potential lack of sufficient support from
the company or the user community in code development. The used open-source
solver, OpenFOAM

, provides the source code and there is a big user community,
providing support for code development.
The remainder of this chapter deals with a description of the governing equa-
tions of fluid flow in section 2.2. To solve the governing equations, the spatial
and temporal discretisation methods are described in section 2.3, followed by a
discussion about the cell quality measures, like skewness and non-orthogonality
in 2.4. In section 2.5 a general transport equation will be discretised to show how
to deal with the different terms, like diffusion and convection. Additionally, a brief
discussion about the treatment of boundary conditions is provided in section 2.6.
When the discretised transport equation and corresponding boundary conditions
are fully explained, the solution procedure to solve the incompressible Navier-
Stokes equations will be dealt in section 2.7. These numerical solution procedures
are present in the used flow solvers, Fluent

and OpenFOAM

. Section 2.9 pro-
vides a brief description of the background and usage of both CFD codes. In order
to validate and verify the CFD codes for our problem, some small test problems are
defined in order to test the influence of different numerical settings, like discreti-
sation schemes, grid resolution and time-step size. The validation and verification
discussion is the subject of section 2.10. Finally, the major conclusions of this
chapter are summarised in section 2.11.
2.2 The Navier-Stokes equations
The governing equations for viscous fluid flow are a coupled set of non-linear par-
tial differential equations (Anderson Jr., 1991, Panton, 2005). These equations
are derived from conservation of mass, momentum and energy within an infinitesi-
mally small spatial control volume. For mass conservation, the following continuity
equation is obtained:
∂ρ
∂t
+∇

(ρu) = 0. (2.1)
Here, ρ [kg/m
3
] is the density and u [m/s] the flow velocity vector. The nabla ∇
operator is defined in three dimensions as
∇=
∂.
∂x
+
∂.
∂y
+
∂.
∂z
.
2.2 The Navier-Stokes equations 17
Secondly, for momentum conservation the following expression can be derived
(neglecting gravity and additional body forces):
∂(ρu)
∂t
+∇

(ρuu) = ∇

σ, (2.2)
where σ [N/m
2
] is the surface stress tensor, necessary in viscous fluid flow. For
compressible flow calculations also the energy conservation equation is specified:
∂(ρe)
∂t
+∇

(ρeu) = ∇

(σu) −∇

q +ρQ, (2.3)
where e [J/kg] is the total specific energy (including kinetic and potential energy),
q [W/s] is the heat flux vector and Q [J·m
3
/kg] equals the nett energy generation.
The full set of equations describing unsteady, compressible viscous flows, are called
the Navier-Stokes equations. The Navier-Stokes equations are non-linear, which
makes them difficult to solve; only for very simplified problems there exists an
analytical solution.
2.2.1 Constitutive relations
In order to close the system of equations (2.1), (2.2) and (2.3), constitutive relations
are needed. For a Newtonian fluid, the stress tensor, σ, which is defined for a
Newtonian fluid as
σ = −
_
p +
2
3
µ∇

u
_
I +µ
_
∇u +∇u
T
_
.
Here I represents the identity tensor, p [N/m
2
] is the pressure and µ [Ns/m
2
] is the
dynamic viscosity. To close the energy equation, the equation of state is specified,
such as the perfect gas law:
p = ρRT,
in which T [K] is the temperature and R [J/(mol·K)] the specific gas constant.
The constitutive relation for the total specific energy yields as follows:
e = e(p, T).
Additionally, the heat conduction is described using Fourier’s law:
q = λ∇T,
with λ [W/(m·K)] the heat conduction transport coefficient.
The governing equations (2.1), (2.2) and (2.3) in combination with additional tur-
bulence modelling can be used in a wide variety of engineering problems. Without
other restrictions, these equations are used for high and low speed flows, turbu-
lence research, multi-phase flows and a lot of other applications. However, these
equations can be difficult to solve and simplifications can be made if applicable to
the concerning problem.
18 Finite volume discretisation
2.2.2 Incompressible laminar flow simplifications
The current research deals with the flow around flapping wings at insect scale,
which is considered to be incompressible (Lentink, 2003, Bos et al., 2008) and
laminar (Williamson, 1995). Therefore, the incompressible laminar Navier-Stokes
equations are solved for Reynolds numbers ranging from Re = 100 to 1000.
A flow can be assumed to be incompressible, when the velocity is lower than 0.3
times the speed of sound (Lentink, 2003, Bos et al., 2008)) and thermal expansion
effects can be neglected. The incompressible Navier-Stokes equations are:


u = 0, (2.4)
∂u
∂t
+∇

(uu) = −
∇p
ρ
+ν∇
2
u, (2.5)
with ν = µ/ρ [m
2
/s] being the kinematic viscosity. For an incompressible flow,
this system of equations is closed such that there is no need to use the energy
equation and additional turbulence modelling.
2.2.3 Dimensionless numbers
In general, the relative relevance of the different terms in equations (2.4) and (2.5)
is revealed by making those equations dimensionless. Therefore, the main vari-
ables, u, t, x, p and ρ are scaled with their reference values, as follows:
u

=
u
U
ref
, t

= t · f
ref
, x

=
x
L
, p

=
p
ρ
ref
· U
2
ref
, ρ

=
ρ
ρ
ref
. (2.6)
The star (*) is used to indicate the dimensionless variables. In case of incom-
pressible flow, the density is constant, such that ρ

= 1. When substituting
equation (2.6) into equations (2.4) and (2.5) the following non-dimensional form
of the incompressible continuity and momentum equations is obtained:


u

= 0, (2.7)
and
St
∂u

∂t
+∇

(u

u

) = −∇p

+
1
Re

2
u

. (2.8)
In these equations, two main dimensionless numbers are identified as relevant
parameters, the Strouhal (St) and Reynolds number (Re):
St =
f
ref
L
ref
U
ref
=
T
conv
T
motion
, (2.9)
Re =
U
ref
L
ref
ν
=
T
visc
T
conv
. (2.10)
These dimensionless numbers represent order estimates for time-scale ratios in the
flow. In (2.9) and (2.10), these relevant time-scales are, respectively, the time-scale
2.3 Spatial and temporal discretisation 19
for convective transport, T
conv
, viscous transport, T
visc
and the relevant time-scale
of the body motion, T
motion
. In order for the dimensionless numbers to have a
proper physical meaning, the reference values need to be chosen appropriately.
It was seen that fluid flow is governed by non-linear partial differential equa-
tions, which can only be solved analytically for extremely simplified model prob-
lems. The full Navier-Stokes equations, combined with the constitutive relations,
are applicable to all kind of flows, where the computational costs strongly depend
on the desired resolution and solution methods. When the flow is considered lam-
inar and incompressible, the governing equations are significantly simplified, such
that the costs for solving may be reduced. However, these simplifications need to
be justified by the concerned fluid flow problem. Concerning flapping wing physics
at lower Reynolds numbers (100 ≤ Re ≤ 1000) the flow inherently is incompress-
ible and laminar. Therefore, solving the unsteady incompressible laminar flow can
be seen as performing a Direct Numerical Simulation, at sufficiently low Reynolds
numbers.
2.3 Spatial and temporal discretisation
This section deals with the spatial and temporal discretisation of the governing
mathematical equations. Since time can be interpreted as a parabolic coordi-
nate (Patankar & Spalding, 1972), it is sufficient to specify the initial time-step,
which is used to march linearly in time, starting at the initial solution. The
time-step may vary, dependent on the maximal Courant number, which will be
explained in section 2.5. Space, however, needs to be discretised throughout the
entire computational domain. The finite volume approach needs a domain sub-
division into a finite number of convex polyhedral control volumes without overlap,
completely filling the domain. OpenFOAM

uses a collocated variable arrange-
ment (Ferziger & Peric, 2002), which means that every control volume centre is
used to store the values of all variables, like pressure and velocity. Figure 2.2 shows
an arbitrary polyhedral control volume V
P
with centre P and neighbouring centre
N. The computational point x
P
is located at the centroid of the computational
cells, which is found from the following relation (Jasak, 1996):
_
V
P
(x −x
P
)dV = 0.
Every two cells, i.e. with centres P and N from figure 2.2, share an internal
face whose geometric centre is denoted by f and has an outward pointed normal
vector S
f
. Faces which are not shared are boundary faces, consequently. Derived
boundary fields, like surface normal gradients or face fluxes, are defined in the face
centre.
After the domain is discretised into a set of control volumes, face surfaces
and points, the governing equations need to be approximated over these cells.
Discretisation is performed assuming a linear variation of scalar variable φ across
20 Finite volume discretisation
x
y
z
V
P
r
P
P
f
S
f N
d
f
Figure 2.2 Discretisation of the computational domain using finite volume cells. An
arbitrary polyhedral control volume is constructed around a centre P and with volume V
P
. The vector
from the cell centre to the neighbouring cell centre N is d
f
. The faces of cell P are directed with the
unit normal vector S
f
and may have an arbitrary number of corners. From (Jasak, 1996).
a cell. This scalar variable φ can be seen as pressure or a velocity component.
Using a Taylor series approximation, the following expression is obtained:
φ(x) = φ
P
+ (x −x
P
) · (∇φ)
P
+O(|(x −x
P
)|
2
), (2.11)
where O(|(x − x
P
)|
2
) represents the second-order truncation error. For the tem-
poral variation of this scalar variable φ a similar expression can be found:
∂φ(t)
∂t
=
φ(t + ∆t) −φ(t)
∆t
+O(∆t). (2.12)
With this linear temporal behaviour of φ the truncation error is second-order
O(∆t
2
), similar to the spatial truncation O(∆x
2
). Both truncation errors can be
expanded using a full Taylor series expansion, which is not within the scope of the
present thesis, but can be found in (Wesseling, 2001, Ferziger & Peric, 2002, Jasak,
1996). Since this discretisation approach is able to cope with arbitrary polyhedral
cell volumes, this method can be used for complex unstructured three-dimensional
meshes, including local mesh refinement.
2.4 Measures of cell quality
Since the accuracy of the numerical solution heavily depends on the interpolation
from cell to face centre, one can imagine that the cell quality is very important.
We will briefly describe the cell quality based on non-orthogonality and skewness,
which will both be used to assess the performance of mesh motion solvers in
chapter 3.
First of all, cell non-orthogonality is defined in figure 2.3(a) by the angle α
N
between the face normal vector S
f
and the line connecting the two cell centres,
d. This angle needs to be as small as possible in order to minimise the truncation
2.5 Discretisation of an incompressible momentum equation 21
P N
f d
S
f
α
N
(a) Cell non-orthogonality.
P N f
i
S
f
f
m
d
(b) Cell skewness.
Figure 2.3 Quality measures using cell non-orthogonality and cell skewness. Two-
dimensional representation of cell non-orthogonality (a) and cell skewness (b) as a measure for the
finite volume cell quality. Cell non-orthogonality is defined as the angle between the face normal vector
S
f
and the direction vector between two cell centres P and N. The cell skewness is defined by the
vectors m and d. From (Jasak, 1996).
error of the diffusion term. The second quality criterion is the cell skewness, see
figure 2.3(b). When the line connecting the two neighbouring face centres does
not coincide with the connecting face centre, the cell is skewed. The degree of
skewness is defined by:
ψ =
|m|
|d|
,
where m and d are defined in figure 2.3(b). Assessing cell skewness is important,
since the interpolation from cell centre to face centre strongly depends on this
quality criterion as will later be seen in this chapter.
2.5 Discretisation of an incompressible momen-
tum equation
This section deals with the temporal and spatial finite volume discretisation of the
incompressible momentum equation, which forms the basis for the incompressible
Navier-Stokes equations. This partial differential equation has the following form:
∂u
∂t
+∇

(uu) −∇

(ν∇u) =
∇p
ρ
. (2.13)
Here, ρ [kg/m
3
] is the reference density, u [m/s] the transport velocity and ν =
µ/ρ [m
2
/s] is the kinematic viscosity. This expression contains a temporal, con-
22 Finite volume discretisation
vection and a diffusion term, given by:
∂u
∂t
: temporal term,


(uu) : convection term,


(ν∇u) : diffusion term.
Using the finite volume approach, the integral form of the incompressible Navier-
Stokes equations is obtained by integrating over a control volume, C
V
:
_
V
CV
∂u
∂t
dV +
_
V
CV


(uu)dV −
_
V
CV


(ν∇u)dV =
_
V
CV
∇p
ρ
dV. (2.14)
This equation is solved in both CFD codes used, Fluent

and OpenFOAM

. In
the remainder of this section, the different terms of equation (2.14) are elaborated
in more detail.
Before dealing with the discretisation of the different terms of equation (2.14)
it is important to discuss the evaluation of the volume, surface, divergence and
the gradient integrals, necessary to understand the evaluations of the convection
and diffusion terms. For this, the scalar variable φ is used, which may represent
the different velocity components. When substituting equation (2.11) into the
volume integral, a second-order approximation is obtained, such that the result is
a multiplication of the scalar value multiplied by the cell volume.
_
V
P
φ(x)dV =
_
V
P

P
+ (x −x
P
)(∇φ)
P
)dV
= φ
P
_
V
P
dV + (
_
V
P
(x −x
P
)dV ) · (∇φ
P
)
≈ φ
P
V
P
.
Similar, for the surface integral the following yields:
_
f
dS · a = S
f
· a
f
, (2.15)
where S
f
is the face surface area. The divergence and gradient terms are evaluated
using Gauss’ theorem (Panton, 2005, Anderson Jr., 1991), which defines a relation
between the volume and the surface integrals. Using Gauss’ theorem, the volume
integral of the divergence of a vector a can be written as the sum of all faces, like:
_
V
P


adV =
_
S
CV
dS

a,
=

f
_
S
f
dS

a,

f
dS
f

a
f
. (2.16)
2.5 Discretisation of an incompressible momentum equation 23
Here, C
V
is the control volume with surface normal vectors S
f
and a
f
is a vector
interpolated to the cell faces using a second-order linear interpolation method.
Discretisation of the gradient integral of a scalar variable φ can be written,
using Gauss’ theorem, as:
_
V
P
∇φdV =
_
S
CV
dSφ,
=

f
_
S
f
dSφ,

f
dS
f

f
.
2.5.1 Face interpolation schemes
Similar to the divergence term, the flux φ
f
is interpolated from the cell centre to
the face centres. A linear interpolation is performed using the following expression:
φ
f
= f
x
φ
P
+ (1 −f
x

N
,
which is illustrated in figure 2.5. The linear interpolation factor f
x
is defined as the
ratio of two distances, f
x
= |fD|/|PD|. So, both divergence and gradient volume
integrals can be reduced to a summation of the corresponding vector or scalar
variable over the cell faces. The standard face interpolation scheme is obtained by
central differencing.
In OpenFOAM

, Jasak et al. (1999) applied extra face interpolation schemes,
like upwind blending using a gamma coefficient and differencing using a flux split-
ting limiter such as SuperBee (Roe, 1986), the Koren limiter (Koren, 1993), or
Van Leer (Van Leer, 1979, Sweby, 1984). The SuperBee, Koren and Van Leer lim-
iter are shown using the Sweby diagram in figure 2.4 (Sweby, 1984). The purpose
of flux limiters is to limit the gradient of the solution in order to avoid spuri-
ous oscillations and to improve the stability of the scheme. Section 2.10 shows a
comparison of the results using different flux limiters on the solution of a model
problem of vortex decay and convection.
2.5.2 Convection term
When the volume integral of the convection term from equation (2.13) is consid-
ered, the following relation can be derived using equation (2.16) and a linearization
24 Finite volume discretisation
r
φ
(
r
)
0 0.5 1 1.5 2 2.5 3
0
0.5
1
1.5
2
2.5
3
(a) SuperBee (Roe)
r
φ
(
r
)
0 0.5 1 1.5 2 2.5 3
0
0.5
1
1.5
2
2.5
3
(b) Koren
r
φ
(
r
)
0 0.5 1 1.5 2 2.5 3
0
0.5
1
1.5
2
2.5
3
(c) Van Leer
Figure 2.4 Different flux splitting limiters. Flux splitting schemes are used to limit the gradient
of the solution in order to avoid spurious wiggles. The flux splitting scheme are a function of r, which
represents the ratio of successive gradients on the mesh. Different flux limiters are employed, (a)
SuperBee (Roe, 1986), (b) Koren (Koren, 1993) and (c) Van Leer (Van Leer, 1979, Sweby, 1984).
method (midpoint, least squares):
_
V
P


(uφ)dV =

f
S
f
· (uφ)
f
=

f
S
f
· (u)
f
φ
f
=

f

f
.
Here, F is the mass flux, given by F = S
f
· (u)
f
. The scalar variable φ needs to
be interpolated using a second-order interpolation method in combination with a
flux limiter, e.g. linear, Gamma, Van Leer.
2.5.3 Diffusion term
The volume integral of the diffusion term from equation (2.13) is discretised and
approximated using linearization as
_
V
P


(ν∇φ)dV =

f
S
f
· (ν∇φ)
f
=

f
ν
f
(S · ∇φ)
f
.
Here, the terms (S· ∇φ)
f
and ν
f
need to be approximated using a proper method.
The face viscosity ν
f
is obtained by interpolation from cell centre to faces. The
other term (S · ∇φ)
f
is obtained on a non-orthogonal mesh by the following ex-
pression:
(S · ∇φ)
f
= |m|
φ
N
−φ
P
|d|
+k · (∇φ)
f
.
2.5 Discretisation of an incompressible momentum equation 25
ts
Flow direction
U P D
φ
U
φ
P
φ
f
φ
D
f
P d f m N
S
f
α
N
k
Figure 2.5 Variation of the flux φ. The
value of φ at the face f is determined as a
function of upstream and downstream values.
Figure 2.6 Cell non-orthogonality
treatment. Illustration of the cell non-
orthogonality correction which is used
on meshes with large skewness and non-
orthogonality.
Here d is the vector between two adjacent cell centres and m is parallel to d with
magnitude of the surface normal vector S
f
. The decomposition of S
f
is shown
in figure 2.6 and derived in Jasak (1996), Juretic (2004) such that the following
relation holds:
S
f
= m+k,
where k is orthogonal to the surface normal vector S.
2.5.4 Temporal term
Since the unsteady Navier-Stokes equations are solved, a proper discretisation of
the temporal scheme is necessary. The time derivative represents the temporal
rate of change of φ which needs to be discretised using new and old time values.
This time difference is defined using prescribed time-step size ∆t such that:
φ
n+1
= φ
n
+ ∆t,
where φ
n
and φ
n+1
are the scalar variable φ at the old and new time instances,
respectively. Two implicit time discretisation methods are considered, one first-
order and and one second-order scheme. The first-order discretisation is simply
the temporal difference:
∂φ
∂t
=
φ
n+1
−φ
n
∆t
,
and the second-order discretisation, see (Wesseling, 2001, Hirsch, 1988), is given
by:
∂φ
∂t
=
3
2
φ
n+1
−φ
n
+
1
2
φ
n−1
∆t
.
This implicit scheme is referred to as the second-order backward differencing
scheme, where φ
n−1
is the old-old value of φ. Consequently, the corresponding
26 Finite volume discretisation
volume integrals obey the following relations:
_
C
V
∂φ
∂t
dV =
φ
n+1
−φ
n
∆t
V
P
,
_
C
V
∂φ
∂t
dV =
3
2
φ
n+1
−2φ
n
+
1
2
φ
n−1
∆t
V
P
. (2.17)
Note, that these relations are only valid on fixed meshes and constant time-steps.
According to (Wesseling, 2001, Hirsch, 1988), the explicit first-order time integra-
tion method may be unstable if the Courant number is larger than 1, where the
Courant number is defined as
Co =
u · ∆t
∆x
.
Implicit methods are in general more stable, compared to (semi-) explicit methods,
such that in the current research the implicit first- and second-order backward
scheme have been used. While the implicit methods are bounded and stable, a
pre-defined maximal Courant number Co
max
is used to vary the corresponding
time-step during the simulation. In that case, the coefficients
3
2
, 2 and
1
2
in (2.17)
should be elaborated to incorporate the ratio of the old and current time-steps.
In section 2.7.3 this will be discussed in more detail.
2.6 Boundary conditions
In order to solve the discretised governing equations, boundary conditions need
to be defined at the boundaries of the computational domain. There are four
boundary conditions (Hirsch, 1988, Wesseling, 2001), which are used to close the
system, namely:
1. zero-gradient boundary condition, defining the solution gradient to be zero.
This condition is known as a Neumann-type condition, ∂φ/∂n = a,
2. fixed-value boundary condition, defining a specified value of the solution.
This is a Dirichlet-type condition, φ = b,
3. symmetry boundary condition, treats the conservation variables as if the
boundary was a mirror plane. This condition defines that the component
of the solution gradient normal to this plane should be fixed to zero. The
parallel components are extrapolated from the interior cells,
4. moving-wall-velocity boundary condition is used on a moving boundary to
keep the flux zero, using the Arbitrary Lagrangian Eulerian approach.
For external flow simulations, a distinction is made between the outer and the
inner boundaries, the latter corresponds to the moving wing or body. To minimise
the effects of the outer boundaries it is desirable to specify a symmetry boundary
2.7 Solution of the Navier-Stokes equations 27
condition (3) at those fixed boundaries, unless a free-stream is specified. In case of
forward flapping flight, two domain boundaries are defined as inflow and outflow,
respectively. At the inflow boundary the velocity is defined as fixed-value (2) and
the pressure as zero-gradient (1). On the other hand, at the outflow boundary,
the pressure has to be fixed-value and the velocity zero-gradient (Hirsch, 1988,
Wesseling, 2001). On a stationary wall the no-slip condition needs to be guaran-
teed, therefore a fixed-value (u = 0) is specified for the velocity in combination
with a zero-gradient for the pressure. If the boundary of the wall moves, than
the proper boundary condition is the moving-wall-velocity (4) which introduces
an extra velocity in order to maintain the no-slip condition and ensures a zero flux
through the moving boundary.
2.7 Solution of the Navier-Stokes equations
Previously, the different terms to discretise the general momentum equation (2.13),
were described. This section briefly deals with the discretisation of the Navier-
Stokes equations and the solution procedure. The incompressible laminar Navier-
Stokes equations were given by (2.4) and (2.5):


u = 0,
∂u
∂t
+∇

(uu) = −
∇p
ρ
+ν∇
2
u.
There are two items, requiring special attention, namely the non-linear term
present in the momentum equation and the pressure-velocity coupling (Ferziger &
Peric, 2002). The non-linear term in these governing equations, ∇

(uu), can be
solved either by using a solver for non-linear systems or by Newton linearization.
Previously, it was seen that the convection term can be written as:
_
V
P


(uu)dV =

f
S
f
· (u)
f
(u)
f
=

f
F(u)
f
= a
p
u
p
+

N
a
N
u
N
,
where a
p
, a
N
and F are still depending on u. a
p
and a
N
represent the diagonal
and off-diagonal terms of the sparse system of equations, respectively. A complete
derivation can be found in (Jasak, 1996). Since F should satisfy the continuity
equation (2.4), both equations (2.4) and (2.5) should be solved together as if it
was a coupled system. In order to avoid the use of expensive solvers for non-linear
systems, this convection term is linearised such that existing velocity fields will be
used to calculate the matrix coefficients a
p
and a
N
.
28 Finite volume discretisation
2.7.1 Pressure equation and Pressure-Velocity coupling
Since the pressure depends on the velocity and vice-versa, a special treatment of
this inter-equation coupling is needed. In order to derive the pressure equation, a
semi-discrete formulation of the momentum equation is written as:
a
p
u
p
= H(u) −∇p. (2.18)
This equation is derived from the integral form of the momentum equation using
the previously described discretisation methods and divided by the volume. Fol-
lowing Rhie & Chow (1983) the pressure gradient in equation (2.18) is not yet
discretised. The H(u) term contains two parts, a convection and a source con-
tribution. The convection part includes the matrix coefficients for all neighbours
multiplied by their corresponding velocities. The source contribution consists of
all source terms, except the pressure term, including the transient term. Therefore
H(u) can be written as follows:
H(u) = −

N
a
N
u
N
+
u
0
∆t
.
Additionally, the discretised continuity equation (2.4) is given by:


u =

f
S
f

u
f
= 0, (2.19)
Now equation (2.18) is rewritten to find an expression for u
p
:
u
p
=
H(u)
a
p

1
a
p
∇p. (2.20)
The velocities on the face of a finite volume cell can be expressed as the interpolated
value on the face of equation (2.20):
u
f
=
_
H(u)
a
p
_
f

_
1
a
p
_
f
(∇p)
f
. (2.21)
This equation will be used to determine the face fluxes. If equation (2.21) is
substituted into equation (2.19), the following pressure equation can be obtained:


_
1
a
p
∇p
_
= ∇

_
H(u)
a
p
_
. (2.22)
The Laplacian operator is discretised using existing methods, which are previously
explained. Combining equations (2.18) to (2.22), the final form of the discretised
Navier-Stokes equations can be written as:
a
p
u
p
= H(u) −

f
S
f
(p)
f
, (2.23)
2.7 Solution of the Navier-Stokes equations 29
and

f
S
f

__
1
a
p
_
(∇p)
f
_
=

f
S
f

_
H(u)
a
p
_
f
, (2.24)
additionally, the face flux is calculated using:
F = S
f

u = S
f

_
_
H(u)
a
p
_
f

_
1
a
p
_
(∇p)
f
_
. (2.25)
When equation (2.22) is satisfied, the face fluxes are guaranteed to be conser-
vative (Ferziger & Peric, 2002). For the discretised form of the Navier-Stokes
equations (2.23) and (2.24) it can be observed that both equations are coupled
through the pressure and velocity, which requires special attention. Since a simul-
taneous approach would be too computationally demanding, this system is solved
in a segregated manner, which means that these equations are solved in sequence.
The inner-equation coupling is established using either PISO (Issa, 1986) or SIM-
PLE (Patankar & Spalding, 1972) based algorithms. Both flow solvers, Fluent

and OpenFOAM

used the PISO scheme for transient flows and the SIMPLE
scheme for steady flows. Since the PISO scheme was used, a brief description is
given. The PISO algorithm consists of the following steps:
1. Momentum predictor stage. The momentum equation (2.23) is solved
using the pressure gradient, known from the previous time-step, since the
actual pressure gradient is not yet calculated. Furthermore, equation (2.23)
provides an approximation of the new velocity field.
2. Pressure solution stage. Using the predicted velocity, from the previ-
ous stage, the H(u) term can be constructed such that the pressure equa-
tion (2.22) can be formed. Using this pressure equation a better approxima-
tion of the new pressure field can be obtained.
3. Explicit velocity correction stage. The last equation in this sequence
is (2.25), which determines the conservative fluxes, which are consistent with
the new pressure field. Since the approximated pressure field, from stage 1,
is replaced by a better pressure field, from stage 2, the velocity field has to be
corrected accordingly. This is performed using (2.20) in an explicit fashion.
For more detailed information, please consult (Jasak, 1996, Ferziger & Peric,
2002, Juretic, 2004).
2.7.2 Procedure for solving the Navier-Stokes equations
After dealing with the discretisation of the Navier-Stokes equations in combination
with the PISO algorithm, it has become possible to describe the solution procedure
to obtain the solution of the Navier-Stokes equations. In unsteady simulations all
other inter-equation couplings, besides the pressure-velocity equations, are lagged,
such that they are included in the PISO loop. For incompressible unsteady flow
30 Finite volume discretisation
with additional turbulence modelling, the solution sequence can be summarised as
follows:
1. Initialisation of all fields, including pressure and velocity, using the initial
condition,
2. Start the simulation to obtain the velocity and pressure values at the new
time-step,
3. Create and solve the momentum predictor equations using the obtained face
fluxes,
4. Iterate through the PISO loop until the pre-defined tolerance of the pressure-
velocity system is reached. The pressure and velocity fields are obtained for
the current time-step, as well as a new set of conservative fluxes,
5. Using the new conservative fluxes, all remaining equations of the system are
solved. If turbulence modelling is included, update the turbulent viscosity
at this stage,
6. Unless the final time is reached, go back to step 2.
This procedure results in solution fields for all solved variables, like pressure, veloc-
ity and possible turbulence variables. Since the present thesis deals with deforming
mesh problems, special attention is necessary to describe the modifications to the
discretised equations dealing with the Arbitrary Lagrangian Eulerian approach,
commonly used to satisfy conservation on deforming meshes.
2.7.3 Arbitrary Lagrangian Eulerian approach
The governing equations to solve the flow are generally discretised using the Eule-
rian description, where the fluid is allowed to flow through the fixed mesh (Ferziger
& Peric, 2002). This is in contrast to the Lagrangian formulation, where the mesh
is fixed to the fluid or material. If the material or fluid deforms, the mesh deforms
with it. This method is commonly used to discretise the governing equations
encountered in structural mechanics. However, when the flow domain moves or
deforms in time due to a moving boundary, a fixed mesh becomes inconvenient,
because it requires the explicit tracking of the domain boundary. Therefore, the
Arbitrary Lagrangian Eulerian (ALE) formulation is used to discretise the flow
equations on moving and deforming meshes (Donea, 1982). This method incorpo-
rates and combines both Lagrangian and Eulerian frameworks. The Lagrangian
contribution allows the mesh to move and deform according to the boundary mo-
tion, whereas the Eulerian part takes care of the fluid flow through the mesh. At
the time of writing, the ALE method has become the standard implementation in
most popular CFD codes to solve for the flow around moving boundaries while the
2.7 Solution of the Navier-Stokes equations 31
mesh deforms accordingly. In general, the momentum equation (2.14) for a scalar
field φ can be derived on a moving mesh as

∂t
_
V
CV
ρφdV +
_
S
CV
ρn·(u−u
s
)φdS−
_
S
CV
ρΓ
φ
n·∇φdS =
_
V
CV
S
φ
(φ)dV, (2.26)
where V
CV
is the arbitrary volume and u
s
the velocity of the moving surface.
The relationship between the rate of change of the volume V
CV
and the veloc-
ity u
s
of the boundary surface S is defined by the so-called Space Conservation
Law (SCL) (Ferziger & Peric, 2002) or Geometric Conservation Law (Lesoinne &
Farhat, 1996):

∂t
_
V
CV
dV −
_
S
CV
n · u
s
dS = 0.
Using the current finite volume discretisation, the computational domain is split
into a finite number of polyhedral cells with varying shape and volume, since the
mesh is deforming. The cells do not overlap and completely fill the domain (Jasak,
2009). In time, the temporal dimension is marched using a variable time-step,
corresponding to a maximal Courant number, using either an implicit first-order
Euler method or a second-order backward scheme (Tukovi´c & Jasak, 2007). For
efficiency, the second-order accurate three time levels backward scheme was used
throughout this research. If equation (2.26) is discretised in space and time the
following relation is obtained for a constant time-step:

n+1
P
φ
n+1
P
V
n+1
P
−4ρ
n
P
φ
n
P
V
n
P

n−1
P
φ
n−1
P
V
n−1
P
2∆t
+

f
( ˙ m
n+1
f
−ρ
n+1
f
˙
V
n+1
f

n+1
f
=

f
(ρΓ
φ
)
n+1
f
S
n+1
f
n
n+1
f

(∇φ)
n+1
f
+s
n+1
φ
V
n+1
P
,
where the subscript P denotes the cell values and f represents the values at the
face centres. The superscripts n+1, n and n−1 are, respectively, the new, old and
old-old values. The mass flux through the face is given by ˙ m
f
= n
f

u
f
S
f
and
the cell face volume change by
˙
V
f
= n
f

u
s
f
S
f
, where u
s
represents the cell face
velocity. The fluid mass flux ˙ m is obtained as part of the solution, satisfying mass
conservation. Furthermore, it is important to determine the volume face flux
such that it satisfies the Space Conservation Law. The temporal discretisation
scheme should be similar to the one used in the momentum equation, otherwise
inconsistency could introduce numerical errors. It is very important to determine
the volume face flux in a consistent way such that it equals the swept volume
calculation, see section 2.8.
Note, that the previous equation is derived for a constant time-step, i.e. the
maximal Courant number varies during the simulation. The current research uses
32 Finite volume discretisation
x
y
z
P
f
(a)
t
n
V
t
n+1
f
(b)
Figure 2.7 Finite volume cell decomposition to calculate swept volumes. The finite volume
cell decomposition is used to form a tetrahedral mesh used for point-based mesh motion solvers and to
calculate the swept volumes. (a) shows the decomposition of a polyhedral cell into tetrahedral volumes
and faces. The swept volume V of a decomposed face is shown in (b).
a defined maximal Courant number leading to a varying time-step. Therefore, the
following relation is derived for a non-constant time-step:
_
1 +
∆t
n+1
∆t
n+1
+ ∆t
n
_
ρ
n+1
P
φ
n+1
P
V
n+1
P

_
1 +
∆t
n+1
∆t
n
_
ρ
n
P
φ
n
P
V
n
P
+
_
1 +
(∆t
n+1
)
2
∆t
n
(∆t
n+1
+ ∆t
n
)
_
ρ
n−1
P
φ
n−1
P
V
n−1
P
+

f
( ˙ m
n+1
f
−ρ
n+1
f
˙
V
n+1
f

n+1
f
=

f
(ρΓ
φ
)
n+1
f
S
n+1
f
n
n+1
f

(∇φ)
n+1
f
+s
n+1
φ
V
n+1
P
,
where the new and old time-steps are respectively given by ∆t
n+1
= t
n+1
−t
n
and
∆t
n
= t
n
−t
n−1
.
2.8 Swept volume calculation
The swept volume is defined as the volume swept by a face of a polyhedral cell
between two subsequent time-steps, t
n
and t
n+1
. This calculation is necessary in
order to satisfy the Space Conservation Law, described in the previous section.
If a polyhedral face is swept from one time-step to the next, it may occur that
the volume becomes warped, such that a volume calculation will not be trivial.
Therefore, the polyhedral cells and faces are decomposed into tetrahedral cells, see
figure 2.7(a). The polyhedral face is decomposed into triangles, using its centroid,
2.9 Numerical flow solvers 33
which is illustrated in figure 2.7(b). The swept volume of a polyhedral face f
is equal to the sum of the swept volumes of the different decomposed triangles,
which need to be accurately calculated. As is illustrated in figure 2.7(b), the swept
volume of such a triangle is similar to a prism with a triangle-shaped bottom area.
Since this prism may be warped, the volume is calculated as the sum of three
tetrahedron volumes. One tetrahedron is shown in the figure, but the remaining
volume of the prism contains two more tetrahedrons. These two additional tetra-
hedrons can be constructed in two different unique ways, using either one of the
two diagonals in the right side face of the swept volume shown in figure 2.7(b),
see (Zuijlen van, 2006). Therefore, the total swept volume using the two different
unique tetrahedron decompositions is:
V
1
=
1
6
(V
p
1
+V
p
2
+V
p
3
),
and
V
2
=
1
6
(V
p
1
+V
p
4
+V
p
5
),
where the same base tetrahedron V
p
1
is used and V
p
2
, V
p
3
, V
p
4
and V
p
5
corre-
spond to different tetrahedron volumes. To obtain the total swept volume of the
decomposed triangle V
triangle
, the average of both V
1
and V
2
is taken as
V
triangle
=
1
2
(V
1
+V
2
).
This swept calculation is successfully validated on test cases using unsteady flow
with mesh motion and proved to be sufficiently accurate. Therefore, this method
is implemented in OpenFOAM

.
2.9 Numerical flow solvers
Section 2.7 described the solution procedure to solve the Navier-Stokes equations,
necessary for fluid flow. This section deals with a brief elaboration of the computer
software, i.e. the CFD codes, used throughout the present research. Two different
CFD codes are used, one commercial (Fluent

) and one open-source package
(OpenFOAM

).
Fluent

is a well-known, easy to use and proven CFD solver, which exploits
the finite volume approach. For completeness, the main settings that we used,
are briefly discussed. The spatial discretisation was second-order upwind and
the time discretisation was first-order implicit Euler (Hirsch, 1988), which is the
only method for which the dynamic mesh module is implemented by Fluent

.
The pressure-velocity coupling in incompressible flow simulations was obtained
using the iterative PISO scheme (Ferziger & Peric, 2002). The accuracy was set
to double-precision and the initial conditions were chosen to be uniform. The
boundary condition on the body was set to no-slip. The convergence criterion for
34 Finite volume discretisation
the iterative method was satisfied with mass and momentum residuals dropping
O(10
−5
) in magnitude.
The other code used in this research, OpenFOAM

, is a general object-oriented
toolbox, written in C++, which is used to solve partial differential equations, e.g.
the Navier-Stokes equations, on a finite volume mesh containing polyhedral cells,
making this code very versatile. One of the main assets of the code is that the
user writes the code in an intuitive way himself, without the need to dig deep in
the underlying code.
All terms are discretised using standard second-order central differencing, ex-
cept for the convection term. Section 2.10 will show that the best method to
discretise the convection term, for our low Reynolds number problems, turned out
to be the linear scheme with the Van Leer limiter (Van Leer, 1979). Concerning
the temporal discretisation scheme, the implicit second-order backward scheme is
used, in combination with a variable time-step corresponding a maximal Courant
number (Wesseling, 2001, Ferziger & Peric, 2002).
Additionally, the iterative solvers and their corresponding convergence criteria
need to be specified. The convergence criterion is based on the residual, which is
derived from the complete system of equations, which is written as
Ax = b,
such that the residual Res is defined by:
Res = b −Ax
The pressure equation is solved using a pre-conditioned conjugate Gradient (PCG)
iterative solver, while the pressure-velocity coupling equation employs its asym-
metric counterpart pre-conditioned bi-stab conjugate Gradient (PBiCG) solver.
The pre-conditioning method varies from incomplete Choleski to incomplete LU
decomposition (Wesseling, 2001, Jasak et al., 2007). Appendix B summarises the
used discretisation schemes and iterative solvers combined with the convergence
criteria for both flow solvers Fluent

and OpenFOAM

.
2.10 Code validation and verification
This section deals with the validation and verification of the CFD solvers that
were used throughout the current research. The commercial flow solver Fluent

has already been tested earlier specifically for low Reynolds number flows, relevant
to flapping insect flight in (Lentink & Gerritsma, 2003, Zuo et al., 2007, Bos et
al., 2008, Thaweewat et al., 2009). OpenFOAM

, however, has been tested exten-
sively for fluid-structure interaction (Tukovi´c & Jasak, 2007), mesh motion (Jasak,
2009) and Large Eddy Simulations (Jasak, 1996, Juretic, 2004, Jasak et al., 2007),
but not for low Reynolds number flows. Therefore, this section only deals with
the validation and verification of the OpenFOAM

flow solver for test problems
relevant for low Reynolds number flapping insect flight.
2.10 Code validation and verification 35
The open-source toolbox OpenFOAM

provides a framework of finite volume
based functions in order to build a specific application for solving partial dif-
ferential equations. Within the context of the present research, an application
is developed, which solves for the unsteady, incompressible flow using deforming
meshes. Besides the fluid and flow properties it is necessary to specify different
interpolation schemes for the different terms of the governing equations. Since this
flow solver was used for low Reynolds number flows with vortex convection it is
necessary to choose the correct face interpolation scheme for the convection term.
Therefore, the face interpolation for diffusion and source terms will be kept fixed
at second-order linear (central) interpolation. Concerning the convection term, on
the other hand, the most accurate scheme needs to be determined from linear dif-
ferencing, gamma differencing, SuperBee splitting, Koren splitting and Van Leer
splitting, as was briefly discussed in section 2.5.
In order to assess the accuracy of these interpolation schemes, two test cases
are investigated, the decay and convection of a Taylor vortex, described in sec-
tion 2.10.1. After the selection of the proper flow solver settings, the accuracy of
the code is assessed using unsteady flows around static and plunging cylinders in
section 2.10.2.
2.10.1 2D vortex decay and convection
To validate the code, two vortex cases are considered, one concerning a decaying
vortex, the other deals with the vortex convection. This problem is relevant for
low Reynolds number flapping flight, since these flows are dominated by unsteady
vortical structures. First a proper vortex definition is described, followed by the
two vortex simulations to discuss the results.
Vortex definition
When a vortex is used for code validation it is important that a well confined
definition is used, i.e. the vortex should have finite and monotonic velocity and
vorticity profiles. The following definition is used from (Panton, 2005), but ana-
logue to (Zhou & Wei, 2003):
V
θ
= t
−m
f(η), (2.27)
where V
θ
represents the radial velocity, m is a compactness coefficient and η a sim-
ilarity parameter. Figure 2.8 shows the velocity profile for different values of m to
show the effect on the compactness of the vortex. In this figure some interesting
characteristics can be observed. With decreasing m, the curve profile becomes
steeper until the asymptotic behaviour is lost at negative m. Conversely, the com-
pactness of the vortex is increased with increasing m, until m = 1.5 after which a
region with counter-rotation appears. When m is chosen to be 0.5 the well-known
Lamb-Oseen vortex is the result, which is not of the desired compactness, although
maximal angular momentum (Panton, 2005) is obtained. For validation purposes
a vortex corresponding to m = 1.5, a Taylor vortex (Panton, 2005, Zhou & Wei,
36 Finite volume discretisation
Figure 2.8 Different velocity profiles of a similar vortex definition. The vortex definition,
from (Panton, 2005), provides different vortex profiles for varying m, ranging from −0.5 to 2 in steps
of 0.5. This m values determines the compactness of the vortex, where m = 1.5 provides the most
compact vortex.
2003), provides a better representation of vortical flows like in low Reynolds num-
ber vortex shedding problems (Panton, 2005). When the compactness coefficient
m is fixed to 1.5 the velocity profile of this Taylor vortex can be derived (Panton,
2005) and is given by:
V
θ
= 2e
1
2
λe
−2λ
2
, (2.28)
where λ is a function of the similarity parameter η:
λ =
η
2
_
(2)
.
Vortex decay
As a first test problem, a Taylor vortex is considered, which decays in a two-
dimensional squared domain with dimensions (5 x 5). This squared domain is
discretised with a Cartesian finite volume grid of 100 x 100 mesh cells as is shown
in figure 2.9, which shows the vorticity of the vortex at t = 0. To solve the
incompressible Navier-Stokes equations, a standard OpenFOAM

solver is used,
icoFoam, without mesh motion, which is not necessary for this problem.
The temporal term is discretised using a second-order backward scheme. Fur-
thermore, all other terms, except for the convection term, are discretised using
a second-order linear interpolation. To study the effect of the face interpolation
of the convection term, the following schemes are used: Gamma (Jasak et al.,
1999), Koren limiter (Koren, 1993), SuperBee limiter (Roe, 1986), Van Leer lim-
iter (Van Leer, 1979) and standard linear interpolation. The mesh size was fixed
to (100 x 100) and the time-step is varying to meet a maximal Courant number of
Co
max
= 1.0, which are both considered to be fine enough. The Reynolds number
2.10 Code validation and verification 37
Figure 2.9 Initial solution of a Taylor vortex on an Cartesian mesh. Starting from this initial
Taylor vortex solution the flow diffusion is solved, without the presence of a convection free-stream.
The velocity profiles and total energy are monitored to identify the accuracy of the flow solver.
is fixed to Re = 100 by setting the kinematic viscosity to ν = 0.01 and the velocity
vector to u = (1.0, 0.0, 0.0). The flow solver solves for 40 seconds such that the
temporal effect of the different face interpolation schemes on the shape and mag-
nitude of the velocity profiles can be compared. Any excess or lack in numerical
diffusion may become visible.
Results
Figure 2.10(a) and 2.10(b) show the initial and final velocity variations in X- and
Y -direction for the different face interpolation schemes. Besides the evolution of
the velocity profile the total energy is shown in figure 2.10(c), which is a measure
for the diffusion. The total energy is calculated as
E
tot
=
N

i=1
0.5|u
i
|
2
,
where i represents the cell index, N the total number of cells and u
i
the velocity
in cell i.
The first important observation, from figure 2.10, is that the total energy is
increasing for the SuperBee flux limiter. This scheme clearly introduced a large
amount of negative numerical diffusivity which causes the Taylor vortex to grow,
which is not physical. The velocity profile as well as the total energy of both
Gamma and linear interpolation schemes are similar, without a significant amount
of diffusion. One major drawback of these methods is that the vortex looses sym-
metry after 5 s, figure 2.10(b). The Koren and Van Leer limiters are slightly more
38 Finite volume discretisation
U velocity [m/s]
Y
c
o
o
r
d
i
n
a
t
e
[
m
]
Initial solution
Gamma
Koren
SuperBee
van Leer
Linear
-1 -0.6 -0.2 0.2 0.6 1
3
3.5
4
4.5
5
5.5
6
(a)
X coordinate [m]
V
v
e
l
o
c
i
t
y
[
m
/
s
]
Initial solution
Gamma
Koren
SuperBee
van Leer
Linear
3 3.5 4 4.5 5 5.5 6
-1
-0.6
-0.2
0.2
0.6
1
(b)
Time [t]
T
o
t
a
l
e
n
e
r
g
y
[
m
2
/
s
2
]
Gamma
Koren
SuperBee
van Leer
Linear
0 2 4 6 8 10 12 14 16
0.94
0.96
0.98
1
1.02
1.04
1.06
1.08
1.1
1.12
(c)
Figure 2.10 Velocity and total energy variations of a decaying Taylor vortex. (a) shows
u(y), at t=20 s, for different face interpolation schemes, while (b) provides v(x). (c) shows the
decaying total energy due to numerical diffusion.
diffusive, see figure 2.10(c), but provide similar results. The Van Leer limiter is
called shape preserving and provides good results for both vortex decay and con-
vection as will be seen in the next section.
Vortex convection
The second validation case concerns a Taylor vortex, which is convected through
a channel with dimensions (20 x 5), shown in figure 2.11. Similar to the vortex
decay problem, the discretisation of the face interpolation is varied and all other
discretisation terms are fixed to second-order linear interpolation. The Cartesian
mesh resolution was set to (400 x 100) and the max Courant number was equal to
Co
max
= 1.0. The flow solver solved the governing equations for 20 s, such that
the vortex was convected through the entire domain. The Reynolds number was
fixed to Re = 100 by setting a kinematic viscosity to ν = 0.01 and an inlet velocity
2.10 Code validation and verification 39
Figure 2.11 Initial solution of a Taylor vortex on an Cartesian mesh, representing a
channel. This Taylor vortex solution is used as the initial solution for the convection validation case.
At sufficiently large grid resolution and time-step size, the effect of different face interpolation schemes
is compared.
to u = (1.0, 0.0, 0.0).
Results
First of all, figure 2.12(c) shows that the total energy in the entire domain is de-
creasing with time for all schemes. Extra diffusion is clearly visible after t=14 s
when the vortex approaches the outlet boundary, smearing the vortex. Since
convection induces physical diffusion, depending on the convection velocity, the
SuperBee scheme possibly still has negative numerical diffusion, as was seen in
the results for vortex decay. Furthermore, the total energy does not provide more
information about the accuracy of the scheme, since the integrated energy is close
for all schemes, except for the SuperBee scheme. Additionally, figure 2.12(a)
and 2.12(b) show, respectively, the velocity in X- and Y -direction through the
vortex core at t = 10 s. Besides the possibly negative numerical diffusion in the
SuperBee scheme (Juntasaro & Marquis, 2004), another observation can be made.
The second-order linear and Gamma schemes lead to an overshoot of the velocity
in Y -direction. This effect increases with time and therefore these two methods
are not appropriate to study the vortical wake patterns in insect flight. Again, the
Koren and Van Leer limiters are very close, without overshoots and with a proper
symmetry preservation of the vortex. When looking in real detail to these results,
the Van Leer limiter slightly outperforms the Koren limiter. The Van Leer limiter
leads to smoother, and more symmetrical vortices (Juntasaro & Marquis, 2004,
Kuzmin & Turek, 2004), such that this scheme was used throughout this study.
2.10.2 Validation using cylinder flows
To validate the accuracy of the flow solver for unsteady and vortical flow, two
cylinder example problems are defined at sufficiently low Reynolds numbers. At
Reynolds numbers less than Re = 47 the flow exhibits steady behaviour (see
Williamson, 1998), which is not relevant for unsteady insect flight, so a Reynolds
number Re > 47 needs to be chosen. On the other hand, when taking a Reynolds
40 Finite volume discretisation
U velocity [m/s]
Y
c
o
o
r
d
i
n
a
t
e
[
m
]
Gamma
Koren
SuperBee
van Leer
Linear
0.4 0.6 0.8 1 1.2 1.4 1.6
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
(a)
X coordinate [m]
V
v
e
l
o
c
i
t
y
[
m
/
s
]
Gamma
Koren
SuperBee
van Leer
Linear
12 13 14 15 16 17 18
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
(b)
Time [t]
T
o
t
a
l
e
n
e
r
g
y
[
m
2
/
s
2
]
Gamma
Koren
SuperBee
van Leer
Linear
0 2 4 6 8 10 12 14 16 18 20
50
50.1
50.2
50.3
50.4
50.5
50.6
50.7
50.8
50.9
51
(c)
Figure 2.12 Velocity profiles of a convected Taylor vortex. The velocity in X-direction u(y)
is shown in (a) at time t=10 s. (b) illustrates the velocity profile in Y -direction as a function of the
X-coordinate, v(x). (c) shows the total energy for the convected vortex.
number larger than about Re = 185, the flow becomes turbulent and additional
turbulence modelling becomes necessary. Since the main objective of this research
is to solve for unsteady, vortical flow around flapping wings, the characteristics of
that kind of flow needs to be present in the validation cases. The flapping wing
simulations are performed in the laminar flow regime Re = O(100), with periodic
force histories. Therefore, in the range 100 ≤ Re ≤ 200, two validation cases
were selected, one concerns the flow around a stationary cylinder at Re = 150 and
the other involves a transversely oscillating cylinder at Re = 185 (Guilmineau &
Queutey, 2002).
The main parameter selected for comparison is the time-averaged drag coeffi-
cient, which is well-documented in literature. Figure 2.14 shows the computational
domain used for both validation cases, the boundaries are located at 10D before,
above and below the cylinder, where D is the cylinder diameter. The outflow
2.10 Code validation and verification 41
Reynolds number [-]
S
t
r
o
u
h
a
l
n
u
m
b
e
r
[
-
]
40 60 80 100 120 140 160 180 200
0.12
0.13
0.14
0.15
0.16
0.17
0.18
0.19
0.2
(a)
Reynolds number [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
10
1
10
2
10
3
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
(b)
Figure 2.13 Relationships between Reynolds number, Strouhal number and drag coef-
ficient. (a) shows the relation between Strouhal number, which is related to the vortex shedding
frequency, and the Reynolds number (Williamson, 1998). The dot (•) shows a Strouhal number of
St = 0.183 for Re = 150 at which the stationary cylinder validation case is performed. (b) relates the
viscous drag coefficient (◦), pressure drag coefficient (•) and total drag coefficient (△) to the Reynolds
number (Henderson, 1995).
Figure 2.14 Computational grid around a cylinder. This grid, with sizes 25k, 50k and 100k is
used to validate the accuracy of the flow solver. The flow is from left to right and the inlet boundary
is located 10D upstream, the outlet 40D downstream and the upper and lower boundaries are located
10D from the cylinder surface, where D is the cylinder diameter.
boundary is located at a distance of 40D. Previous studies, e.g. Lentink & Ger-
ritsma (2003), Bos et al. (2008), showed that boundary effects are minimal at this
domain size.
Flow around a stationary circular cylinder
The first case, dealing with the flow around a static cylinder at Re = 150 is
inherently laminar and unsteady, resulting in a periodic vortex wake. Henderson
(1995) performed a spectral element numerical study which is used as the baseline
reference for this case. Figure 2.13(a) shows the relation between the Strouhal
and Reynolds number, according to (Williamson, 1998). A Strouhal number of
St = 0.183 is obtained for a stationary cylinder at Re = 150. Additionally,
figure 2.13(b) shows the results from an extensive study performed by Henderson
(1995) to identify a relation between drag coefficient and Reynolds number. The
42 Finite volume discretisation
Figure 2.15 Vorticity visualisation of the Von K´arm´an vortex street. The flow around a
stationary cylinder shows a periodic vortex street, of which the frequency depends on the Reynolds
number. Vorticity ω = ∇×u is used to identify the vortical structures, which are clearly visible at a
Reynolds number of Re = 150.
resulting drag coefficient at Re = 150 is found to be C
D
= 1.333. The drag and
lift coefficient are respectively defined as:
C
D
=
D
1
2
ρU
2
ref
, C
L
=
L
1
2
ρU
2
ref
. (2.29)
In order to investigate the temporal and spatial convergence of the solution, the
grid size is varied from 25k, 50k and 100k. The time-step is systematically
decreased according to a maximal Courant number corresponding to Co
max
=
2.0, 1.0, 0.5 and 0.25.
Results
To illustrate the flow behaviour, figure 2.15 shows the instantaneous vorticity
(ω = ∇×u) contours, which reveals the presence of the Von K´ arm´ an vortex street
behind the stationary cylinder. In practical applications such a vortex street ex-
ists behind struts in water, for example. The alternating vortex shedding pattern
leads to periodic force variations which can be visualised using C
L
-C
D
limit cycles.
The results, obtained for different mesh resolutions (25k, 50k and 100k) are shown
in figure 2.16. These limit cycles are determined by taking the periodic part of
the force histories as shown in figure 2.19. From these periodic forces, the time-
averaged drag coefficient is determined and compared with literature in table 2.1.
From this table, it can be observed that the drag coefficient of the coarsest case,
25k and Co
max
= 2.0, has the largest difference with literature, 5.63%. When the
grid is refined and the time-step decreased, it is seen that the solution decreases
asymptotically. The drag on the finest grid with the smallest time-step is about
2.7% larger compared to the value obtained by Henderson (1995). In addition,
the calculated time-averaged Strouhal number (shedding frequency) is shown in
table 2.2. From that table, it can be seen that the Strouhal number matches the
value from Henderson (1995) even more closely than the time-averaged drag coef-
ficient. The differences of the mean Strouhal number with literature ranges from
2.10 Code validation and verification 43
Drag coefficient [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
Grid Size=25k
Grid Size=50k
Grid Size=100k
1.34 1.36 1.38 1.4 1.42 1.44 1.46
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
(a) Co
max
= 2.0
Drag coefficient [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
Grid Size=25k
Grid Size=50k
Grid Size=100k
1.34 1.36 1.38 1.4 1.42 1.44 1.46
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
(b) Co
max
= 1.0
Figure 2.16 Lift-Drag limit cycles of the flow around a stationary cylinder. The periodicity
of the flow around a stationary circular cylinder at Re = 150 is illustrated by a C
L
-C
D
limit cycle.
Additionally, the grid convergence of the solution can be observed for Co
max
= 2.0, (a), and Co
max
=
1.0 (b).
Drag coefficient, C
D
Mesh size Co
max
2.0 1.0 0.5 0.25 Richardson
25k 1.408 1.393 1.385 1.381 1.379 (+3.5%)
50k 1.392 1.381 1.376 1.373 1.372 (+2.9%)
100k 1.385 1.377 1.372 1.370 1.369 (+2.7%)
Henderson (1995) 1.333
Table 2.1 Drag comparison for the flow around a stationary cylinder. Comparison of
the time-averaged drag coefficient for different grids, 25k, 50k and 100k, and different time-steps
corresponding to a maximal Courant number, Co
max
= 2.0, 1.0, 0.5 and 0.25.
3.28% to 1.64%, which is considered sufficiently small.
From table 2.1 and 2.2, it is clear that the flow solver produces results which
are sufficiently close to the values obtained from literature. Besides a comparison
solely with literature it is important to investigate the convergence of the solution
with increasing grid resolution and decreasing time-step size. If the flow solver is
developed in a numerically consistent way, the flow solution should converge to
an asymptotically value with increasing spatial and temporal resolution. In order
to illustrate if the solution converges, figure 2.17 shows the time-averaged drag
coefficient for increasing mesh resolution (for each time-step) and for decreasing
time-step (for each mesh). As can be seen, the solution decreases asymptotically,
which should be the case. The last value in these two figures, is the extrapolated
values, using Richardson’s extrapolation (Ferziger & Peric, 2002). The extrapo-
44 Finite volume discretisation
Spatial resolution
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
Co
max
= 2.0
Co
max
= 1.0
Co
max
= 0.5
Co
max
= 0.25
1 2 4

1.36
1.37
1.38
1.39
1.4
1.41
(a)
Temporal resolution
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
Grid Size=25k
Grid Size=50k
Grid Size=100k
1 2 4 8

1.36
1.37
1.38
1.39
1.4
1.41
(b)
Figure 2.17 Time-averaged drag coefficient as a function of spatial and temporal resolution
for the stationary cylinder. A flow solver is numerically consistent if the flow solution converges
with increasing grid resolution and decreasing time-step size. (a) shows the drag coefficient with
increasing grid refinement level for different time-steps. (b) shows the drag coefficient with decreasing
time-step size for the different grid sizes. The Richardson extrapolated values are plotted at ∞.
lated value is obtained using the following expression:
φ
extrap
= φ
fine
+
φ
fine
−φ
coarse
2
p
−1
, (2.30)
where φ
extrap
is the extrapolated value, φ
fine
and φ
coarse
are the two most accurate
solutions available. Theoretically, the order of the scheme p can be obtained
using (2.30) and should be 2 for a second-order discretisation scheme. This is true
for uniform Cartesian meshes (Ferziger & Peric, 2002) which is not the case for
the cylinder simulations. However, from table 2.1 and 2.2 it can be deduced that
for both second-order spatial and temporal schemes, the value of p lies between
1.5 and 2.
Strouhal number, St
Mesh size Co
max
2.0 1.0 0.5 0.25
25k 0.186 (+1.64%) 0.187 0.188 0.188 (+2.73%)
50k 0.187 (+2.19%) 0.188 0.188 0.189 (+3.28%)
100k 0.188 (+2.73%) 0.188 0.188 0.189 (+3.28%)
Henderson (1995) 0.183
Table 2.2 Strouhal number comparison for the flow around a stationary cylinder. Com-
parison of the time-averaged Strouhal number for different grids, 25k, 50k and 100k, and different
time-steps corresponding to a maximal Courant number, Co
max
= 2.0, 1.0, 0.5 and 0.25.
2.10 Code validation and verification 45
Spatial resolution
L
i
f
t
c
o
e

c
i
e
n
t
a
m
p
l
i
t
u
d
e
[
-
] Co
max
= 2.0
Co
max
= 1.0
Co
max
= 0.5
Co
max
= 0.25
1 2 4

0.54
0.56
0.58
0.6
0.62
0.64
(a)
Temporal resolution
L
i
f
t
c
o
e

c
i
e
n
t
a
m
p
l
i
t
u
d
e
[
-
]
Grid Size=25k
Grid Size=50k
Grid Size=100k
1 2 4 8

0.54
0.56
0.58
0.6
0.62
0.64
(b)
Figure 2.18 Time-averaged lift convergence with grid resolution and time-step size for the
stationary cylinder. (a) shows the average lift coefficient amplitude with increasing grid refinement
level for different time-steps. (b) shows the time-averaged lift coefficient amplitude with decreasing
time-step for the different grid sizes.
Time [s]
F
o
r
c
e
c
o
e

c
i
e
n
t
[
-
]
Drag
Lift
Co
max
= 2.0
Co
max
= 1.0
Co
max
= 0.5
Co
max
= 0.25
0 50 100 150
-0.5
0
0.5
1
1.5
Figure 2.19 Forces around a stationary cylinder, 25k. Lift and drag coefficients for a stationary
circular cylinder case at Re = 150, the grid size is 25k and the time-step was varied according to a
maximal Courant number of Co
max
= 2.0, 1.0, 0.5 and 0.25.
Additionally to the temporal and spatial convergence of the drag coefficient,
figure 2.18 shows the convergence of the lift coefficient amplitude in order to prove
the consistency of the flow solver. Concerning the extrapolated values of the drag
coefficient, from table 2.1, it can be determined that the differences varies from
2.7% to 3.5% compared to literature with decreasing mesh resolution. For all
meshes, the differences in drag coefficient compared to the extrapolated values,
are smaller than 1.0% for maximal Courant numbers Co
max
= 1.0, 0.5 and 0.25.
Therefore, it seems sufficient to consider the mesh of 50k and Co
max
= 1.0 to result
in an accurate solution, these settings were used throughout the present research.
Flow around a transversely oscillating circular cylinder
The last validation case concerns the flow around a transversely oscillating cylinder
46 Finite volume discretisation
at Re = 185 using a numerical study, performed by Guilmineau & Queutey (2002).
The oscillating direction is perpendicular to the free-stream direction. The cylinder
motion is defined as
y(t) = −A
e
sin(2πf
e
t), (2.31)
where the amplitude is set to A
e
= 0.2D, with D the cylinder diameter. The
frequency was set to f
e
= 0.154, corresponding to 0.8 times the natural shedding
frequency of a stationary cylinder at a Reynolds number Re = 185. An amplitude
of 0.2D is relatively small for insect aerodynamics, which employs amplitudes of
several chord lengths, but sufficient to investigate the moving wing capabilities
of the numerical model. Previously conducted simulations on stationary cylinder
flow showed that a mesh of 50k provides a sufficiently accurate solution, therefore
that mesh is also used for this test case. Although, a time-step corresponding to
Co
max
= 1.0 was found to be sufficient, the following values are used to show that
the flow solver converges to an asymptotic solution for an oscillating cylinder as
well, Co
max
= 2.0, 1.0, 0.5 and 0.25 were considered.
Results
To assess the accuracy of the flow solver applied to flapping wings, the results of
an oscillating cylinder case were compared with literature. Guilmineau & Queutey
(2002) found a drag coefficient of C
D
= 1.2. Figure 2.20(a) shows the limit cy-
cle results, using the present flow solver, with second-order temporal and spatial
discretisation. From this figure it is obvious that the flow solution converges with
decreasing time-step. This statement is confirmed if the time-averaged drag coef-
ficient is plotted in figure 2.20(b), for the 50k grid. The value for all time-steps
was within 2% compared to the extrapolated value, which is considered to be
sufficiently accurate.
2.11 Conclusions
This chapter has presented the finite volume discretisation of the incompressible
laminar Navier-Stokes equations. The discretisation concerns arbitrary polyhedral
meshes, such that this method can easily be applied to a wide variety of problems
with complex geometries. To obtain accurate and efficient results, the mesh qual-
ity should be high in terms of non-orthogonality and skewness, both mesh quality
measures. The different terms of the governing equation were discretised using
second-order schemes and different flux splitting methods were described concern-
ing the face interpolation. In order to solve the flow on a computational grid, four
different types of boundary conditions were specified, fixed-value (Dirichlet), zero-
gradient (Neumann), symmetry and moving-wall-velocity. Using those boundary
conditions, the discretised Navier-Stokes equations can be solved using a PISO
pressure velocity coupling in combination with an Arbitrary Lagrangian Eulerian
(ALE) approach if dynamic meshes are used.
2.11 Conclusions 47
Drag coefficient [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
Co = 2.0
Co = 1.0
Co = 0.5
Co = 0.25
1.2 1.25 1.3 1.35
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
(a)
Temporal resolution
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
1 2 4 8

1.255
1.26
1.265
1.27
1.275
1.28
(b)
Figure 2.20 Forces for the flow around an oscillating cylinder. (a) shows the C
D
-C
L
limit
cycles for different time-steps on a 50k mesh. (b) shows the time-averaged drag coefficient with
decreasing time-step for the 50k mesh. The Richardson extrapolated values are plotted at ∞.
In principle, the described discretisation method and solution procedure is ap-
plicable to different commercial and non-commercial CFD solvers. The present
research used the commercial CFD solver Fluent

and the open-source CFD code
OpenFOAM

, both were briefly described. Fluent

has already been tested for
various flows in literature, OpenFOAM

too, but not for low Reynolds flows, rel-
evant for flapping insect flight. Therefore, this chapter presented a validation of
OpenFOAM

for a number of relevant test cases. Vortex decay and convection
were used to study the influence of the face interpolation scheme with different
flux limiters. It was found that the Van Leer flux limiter provides the most accu-
rate results, concerning vortex decay and convection. In addition, stationary and
transversely oscillating cylinder flows were used to successfully prove spatial and
temporal convergence. It is concluded that the open-source solver OpenFOAM

provides an accurate and efficient framework to investigate the flow around flap-
ping wings at low Reynolds numbers.
CHAPTER 3
Mesh deformation techniques for
flapping flight
Submitted to Comput. Meth. Appl. Mech. Engrg. (January 2010).
In order to use mesh deformation techniques to investigate flapping wing aerody-
namics it is necessary to maintain a high mesh quality for relevant wing kinematics.
Different mesh deformation techniques are compared in order to identify their ap-
plicability for cases with flapping wings. The main difficulty is to maintain high
mesh quality when the wing exhibits large translations and rotations. In addition
to existing mesh motion methods, based on solving the Laplace and solid body
rotation stress equations, a mesh deformation routine based on the interpolation
of radial basis functions is introduced.
The radial basis function method can be used with different basis functions
with global or compact support. A globally supported basis function results in
the highest average mesh quality, but is computationally more expensive, com-
pared to a function with compact support. A test case, concerning a moving
two-dimensional block, is used to show that the radial basis function method pro-
vides superior mesh quality compared to the Laplace mesh motion solver. The
mesh quality, based on skewness and non-orthogonality, is found to be highest
when the thin plate spline is used as a basis function.
Additionally, it is shown that this method can be used for mesh deformation
for three-dimensional flapping wings and can handle flexing boundaries. In order
to increase the efficiency of this method, two techniques are applied, based on
boundary coarsening and smoothing of the radial basis function.
50 Mesh deformation techniques for flapping flight
3.1 Introduction
In order to solve the governing equations, using the discretisation methods from
chapter 2, a computational mesh is necessary. Using prescribed initial and bound-
ary conditions, the equations are iteratively solved on the computational domain.
The boundaries of the computational domain can be either fixed or deforming. In
engineering, there are numerous computational fluid dynamics (CFD) problems
in which the flow solution involves geometrically deforming boundaries. Exam-
ples of such interaction problems are fluid-structure-interaction cases like blood
flow through arteries or deforming flags. A simplified type of interaction is that
which concerns a one-way coupling, where the flow is being influenced by a chang-
ing boundary shape. This may be caused by imposed external effects, like the
rigid body motion, e.g. a flapping wing, or a prescribed body deformation, if this
happens to be known beforehand.
If the shape of the domain boundary is time-varying, it is important that the
internal mesh preserves its validity (no negative cell volumes) and quality (cell
orthogonality and skewness). In order to deal with moving objects, it is possible
to solve this mathematical case by changing the boundary conditions as if the
boundary was deforming or to deform the complete mesh. The first method is
called the immersed boundary method (Peskin, 2002), which defines a moving
boundary on a stationary Cartesian background mesh. The disadvantages of the
immersed boundary method are the difficulty to capture the boundary layer and
to meet the requirements for mass and momentum conservation. An appropriate
implementation of this method is not trivial. The second method to deal with a
deforming boundary, the one used in the current thesis, is the use of a mesh motion
solver which moves the internal mesh points. Current CFD solvers incorporate
different mesh motion techniques in order to change the location of the internal
mesh points according to the varying domain shape. Preservation of high mesh
quality is necessary to solve the flow in an accurate and efficient way. When using
a mesh motion solver, the computational mesh points are moved in order to keep
track of the changing location of boundary points. In order to assess the quality
of a mesh motion solver, three different aspects need to be formulated, quality,
efficiency and robustness.
The quality of the resulting mesh is defined by the non-orthogonality and skew-
ness of the finite volume cells. Efficiency is a measure of the used computation
time to calculate the displacements of the mesh points at the new time-step. Addi-
tionally, robustness is used to identify if a method is user-friendly. A robust mesh
motion solver is defined such that it needs little to no user-input. However, current
mesh motion techniques are not fully suitable to cope with the mesh deformation
around an object which moves with a large change in rotation. Therefore, existing
methods will be compared and an improved mesh motion solver is explored and
incorporated.
In literature, several mesh deformation methods have been presented using dif-
ferent approaches to calculate the motion of the computational mesh points. For
3.1 Introduction 51
structured meshes, there are efficient techniques available to deform the mesh, for
example Transfinite Interpolation (Wang & Przekwas, 1994). They interpolated
the displacements of the boundary points along grid lines through the entire com-
putational mesh to find the displacements of all interior mesh points. Using an
additional mapping (Wang, 2000b) the mesh quality can be improved significantly
if the boundary is subjected to significant rotation and deformation. These meth-
ods are perfectly suitable for structured but unsuitable for unstructured grids.
Since, unstructured meshes are used for complex geometries, possibly in combina-
tion with mesh refinement, the focus is put on mesh deformation techniques which
can be applied to unstructured meshes containing arbitrary polyhedral cells, used
in the finite volume code of OpenFOAM

(Weller et al., 1998, Jasak et al., 2004).
The most popular mesh deformation method, applicable to both structured
and unstructured meshes, is called the spring analogy (Batina, 1990) where the
point-to-point connection of every two neighbouring mesh points is represented
by a linear spring. However, this method proved to lack robustness, especially on
arbitrarily unstructured meshes, as was observed by (Blom, 2000), high resulting
mesh quality was only achieved by specifying a problem specific spring stiffness.
Additionally, Farhat et al. (1998), Degand & Farhat (2002) proposed a method to
incorporate torsional springs to improve the robustness of this method.
Other mesh deformation techniques involve solving a partial differential equa-
tion on the complete field of internal mesh displacements for given boundary point
displacements. Concerning the governing partial differential equations, the Laplace
and bi-harmonic operators (L¨ohner & Yang, 1996, Helenbrook, 2003) are often used
in combination with a constant or variable distance-based diffusion coefficient to
improve the mesh quality. Another choice of equations is made by Johnson &
Tezduyar (1994) who used the pseudo-solid equation, which assumes static equi-
librium for small deformations of a linear elastic solid (the mesh is treated as if it
was a solid). The latter method is often used in the Arbitrary Lagrangian Eule-
rian formulation of finite element codes. Dwight (2004) modified this method to
incorporate rigid body rotation, which significantly improves the mesh quality for
meshes subjected to large boundary translations and rotations.
Since these methods solve a partial differential equation on the complete field
of internal mesh points the existing iterative solvers can be used, already available
in existing CFD codes (Jasak & Tukovi´c, 2004, Jasak, 2009). Therefore, the par-
allel implementation of these methods is fairly straightforward. Depending on the
method, a variable diffusivity field needs to be defined, which acts as a stiffness
of the system of equations, this influences the efficiency. One major drawback
of these methods is that they all fail in maintaining high mesh quality when the
boundary points move with high rotation angles. Therefore, a new mesh defor-
mation method was developed and incorporated, based on the use of radial basis
function (RBF) interpolation to obtain the mesh point displacements. In (Boer de
et al., 2007, Bos et al., 2010a) it was shown that radial basis function interpolation
could improve mesh quality considerably.
Radial basis functions (RBF) are commonly used in literature to interpolate
52 Mesh deformation techniques for flapping flight
scattered data, because of their good approximation properties which is discussed
by Buhmann (2000). The application of radial basis functions is very wide, they
have been used in computer graphics, geophysics, error estimation, but also in cou-
pled simulations as in fluid-structure-interaction. (Boer de et al., 2007) used radial
basis function interpolation to couple two non-matching meshes at the interface of
a fluid-structure interaction computation. An RBF interpolation function is used
to transfer the known boundary point displacements to the fluid boundary mesh.
Since the application of RBF’s to interpolate from and to the boundary mesh was
very accurate and efficient, the idea was born to interpolate the boundary mesh
to all computational mesh points. A preliminary study was performed by Boer de
et al. (2007). Previously, radial basis functions were only applied to mesh motion
concerning the boundaries in multi-block meshes (Potsdam & Guruswamy, 2001).
They noted that applying this method to all mesh points would be too compu-
tationally expensive. Since mesh deformation using RBF interpolation results in
high quality meshes even with large body rotation angles, two techniques are im-
plemented to improve its efficiency. Only recent studies (Jakobsson & Amoignon,
2007, Rendall & Allen, 2008c,b) have been carried out to improve the efficiency of
mesh motion based on radial basis function interpolation.
In this chapter two existing mesh deformation techniques are compared, based on
the Laplace equation with variable diffusivity and a modified pseudo-solid equa-
tion, with radial basis function mesh motion. Both Laplace and pseudo-solid
mesh motion techniques are commonly used within the OpenFOAM

community.
These different mesh motion methods are described in section 3.2. In order to
assess the mesh quality, section 3.3 discusses two different criteria, based on non-
orthogonality and skewness, described in chapter 2. The resulting mesh quality
of the different mesh motion methods is studied using a two-dimensional test case
of a block which translates and rotates. The mesh quality is investigated using
a visualisation and histograms of the skewness and non-orthogonality criterion.
This is the subject of section 3.4. In addition to the simplified moving block, two
more relevant test cases were considered, one using a three-dimensional flapping
wing and the other involves a two-dimensional flexing airfoil. Since the radial ba-
sis function mesh motion method is computationally expensive, section 3.5 deals
with two techniques to increase its efficiency. Finally, the conclusions are drawn
in section 3.6.
3.2 Different mesh deformation techniques
When a moving mesh problem is considered, the shape of the computational do-
main is varying in time. Therefore, a distinction can be made between the motion
of the boundary points and the motion of the internal (fluid) points. The displace-
ment of the boundary points can be considered to be given, either it is externally
defined, i.e. a prescribed rigid body motion, or it is part of the solution, which is
3.2 Different mesh deformation techniques 53
the case in fluid-structure interaction problems. According to the given boundary
point motion, the internal points need to be moved in order to maintain mesh qual-
ity and validity. The internal point motion influences the solution only through
the discretisation errors (Ferziger & Peric, 2002), provided that the ALE formula-
tion is correctly implemented. The internal point motion can be calculated using
different methods, as will be shown in the next section.
3.2.1 Laplace equation with variable diffusivity
One can think of a deforming computational domain as if it was a solid body under-
going internal stresses given by the Piola-Kirchhoff stress-strain equation (Baruh,
1999). That equation is non-linear and thus expensive to solve using existing
numerical techniques. Therefore, other type of equations were used, namely the
Laplace equation and the solid body rotation stress (SBR Stress) equation, which
is a variant of the linear stress equation (Dwight, 2004). A mesh motion method
based on one of these equations is computationally cheap since the resulting matrix
system is sparse, such that existing iterative solvers can be used efficiently.
When the mesh motion is governed by the Laplace equation, the given bound-
ary point motion may be arbitrary and non-uniform. The nature of the Laplace
equation is that the point displacements will be largest close to the moving bound-
ary and small at large distance. Ideally, a user input is not desired, since it
decreases the robustness of the method. However, this method needs the specifi-
cation of a variable diffusivity. This leads to the following definition of the Laplace
equation:


(γ∇x) = 0,
where x is the displacement field and γ the diffusion coefficient, which decreases
with the radius r from the deforming boundary as follows:
γ(r) =
1
r
m
. (3.1)
The resulting mesh quality strongly depends on the chosen γ(r) function, which
depends on the distance from the moving boundary. This variable diffusion coeffi-
cient can be chosen such that a region next to the deforming or moving boundary
closely moves with the boundary. The resulting mesh contains less cell quality
deterioration next to the boundary. The current research uses a γ(r) function like
equation (3.1). In addition to the freedom of choosing a diffusion function, it is also
possible to define γ(r) for every internal mesh cell for all time-steps independently.
This, however, appears to be very problem dependent and thus optimisation of
γ(r) seems not cost effective. To maintain robustness, in the current work we use a
quadratically, m = 2, decreasing diffusion coefficient, which was found to provide
efficient and a smooth mesh motion (Jasak & Tukovi´c, 2004). Additionally, one
could also have used an exponentially decreasing diffusion coefficient or a diffusion
coefficient related to the mesh deformation energy.
54 Mesh deformation techniques for flapping flight
3.2.2 Solid body rotation stress equation
The second method to deform the mesh is based on the linear elasticity equation
and is called the solid body rotation stress (SBR Stress) equation (Dwight, 2004).
The equation of linear elasticity, valid for small displacements, may be written as


σ = f , (3.2)
where σ is the stress tensor and f the acting force vector. The stress tensor σ is
given in terms of the strain, which is given by the following constitutive relation:
σ = λtr (ǫ)I + 2µǫ, (3.3)
where tr is the trace and λ and µ are Lam´e constants (Baruh, 1999), which are a
property of the elastic material. The constants can be related to Young’s modulus,
E, as
λ =
νE
(1 +ν)(1 −2ν)
, µ =
E
2(1 +ν)
,
where ν is Poisson’s ratio, meaning the material contraction ratio as it stretches.
The following equation:
ǫ =
1
2
(∇x +∇x
T
), (3.4)
defines the relative change in length, where x is the position of an internal mesh
point, which is treated as if it was a linear solid. Although equation (3.4) does
not allow for rotations, there is nothing against changing this strain equation such
that rigid body rotations are allowed. In Dwight (2004) an extra term was added
to obtain the following strain relation:
ǫ =
1
2
(∇x +∇x
T
+∇x
T
· ∇x). (3.5)
Combining equations (3.5), (3.3) and (3.2), together with λ = −E and µ = E, the
following solid body rotation stress equation is obtained:


(γ∇x) +∇(γ(∇x −∇x
T
)) −λtr(∇x) = 0, (3.6)
where γ is a similar diffusion coefficient as in equation (3.1). Equation (3.6) allows
for rigid body motion and is still linear and therefore the computational costs are
of the same order as the costs necessary to solve the Laplace equation.
Solving the Laplace or the SBR Stress equation leads to a sparse system of equa-
tions, such that standard iterative techniques can be used, like the pre-conditioned
Conjugate Gradient (PCG) method. However, it is also possible to explicitly de-
fine the point motion using interpolation techniques, like the transfinite interpo-
lation (Wang & Przekwas, 1994) usually applied to the points of multi-blocks.
In section 3.4 it is shown that both previously described methods maintain high
mesh quality for problems with limited boundary rotation. In order to deal with
large rotations, a newly implemented mesh motion solver is based on radial basis
function interpolation, such that it can be used for flapping wing simulations.
3.2 Different mesh deformation techniques 55
3.2.3 Radial basis function interpolation
In the current work we use radial basis function interpolation to find the dis-
placements of the internal fluid points for given boundary displacements. The
interpolation function s(x) describing the displacement of all computational mesh
points, is approximated by a sum of basis functions:
s(x) =
N
b

j=1
γ
j
φ(||x −x
b
j
||) +q(x), (3.7)
where the known boundary value displacements are given by x
b
j
= [x
b
j
, y
b
j
, z
b
j
],
q is a polynomial, N
b
is the number of boundary points and φ is a given basis
function as a function of the Euclidean distance ||x||. The minimal degree of
polynomial q depends on the choice of the basis function φ (Boer de et al., 2007).
A unique interpolant is given if the basis function is a conditionally positive definite
function. If the basis functions are conditionally positive definite of order m ≤ 2, a
linear polynomial can be used (Beckert & Wendland, 2001). We only applied basis
functions that satisfy this criterion. A consequence of using a linear polynomial is
that rigid body translations are exactly recovered. The polynomial q is defined by
the coefficients γ
j
which can be defined by evaluating the interpolation function
s(x) in the known boundary points:
s(x
b
j
) = ∆x
b
j
.
Here ∆x
b
j
contains the known discrete values of the boundary point displacements.
Together with the additional requirements:
N
b

j=1
γ
j
p(x
b
j
) = 0,
which holds for all polynomials p with a degree less or equal than that of polynomial
q, the γ
j
values can be determined (Boer de et al., 2007).
The values for the coefficients γ
j
and the linear polynomial can be obtained by
solving the system:
_
∆x
b
0
_
=
_
Φ
bb
Q
b
Q
T
b
0
_ _
γ
β
_
, (3.8)
where γ is containing all coefficients γ
j
, β the four coefficients of the linear poly-
nomial q, Φ
bb
an n
b
× n
b
matrix contains the evaluation of the basis function
φ
b
i
b
j
= φ(||x
b
i
− x
b
j
||) and can be seen as a connectivity matrix connecting all
boundary points with all internal fluid points. Q
b
is an (n
b
× (d + 1)) matrix
with row j given by [ 1 x
b
j
]. In general, (3.8) leads to a dense matrix system,
which is difficult to solve using standard iterative techniques. Therefore, it needs
to be solved directly, by doing a LU decomposition. The possibilities of solving
the system in a more efficient way are discussed in section 3.5.
56 Mesh deformation techniques for flapping flight
When the coefficients in γ and β are obtained they are used to calculate the
values for the displacements of all internal fluid points ∆x
in
j
using the evaluation
function (3.7),
∆x
in
j
= s(x
in
j
). (3.9)
The result of (3.9) is transferred to the mesh motion solver to update all internal
points accordingly. This interpolation function is equal to the displacement of the
moving boundary or zero at the outer boundaries. Every internal mesh point is
moved based on its calculated displacement, such that no mesh connectivity is
necessary. The size of the system of this method (3.8) is ((N
b
+ 4) x (N
b
+ 4)),
which is considerably smaller than other techniques using the mesh connectivity.
The mesh connectivity techniques encounter systems of the order (N
in
x N
in
),
with N
in
the total number of mesh points, which is a dimension higher than the
total number of boundary points. Solving the system (3.8) gives the values of the
necessary coefficients γ and β, which are then used for step two, the evaluation
using equation (3.9).
In contrast to the Laplace and SBR Stress methods, no partial differential
equation needs to be solved and the evaluation of all internal boundary points is
straightforward to implement in parallel, since no mesh connectivity is needed.
Concerning robustness, this method is not using a variable diffusion coefficient
which has to be tuned by the user. Instead, the proper radial basis function needs
to be chosen to satisfy the need for robustness.
Radial basis functions with compact support
From literature, various radial basis functions are available, which are suitable for
data interpolation. Two types of radial basis functions can be distinguished: func-
tions with compact and functions with global support. Functions with compact
support have the following property:
φ(x/r) =
_
f(x/r) 0 ≤ x ≤ r,
0 x > r,
where f(x/r) ≥ 0 is scaled with a support radius r. When a support radius is used,
only the internal mesh points inside a circle (two-dimensional problem) or a sphere
(three-dimensional problem) with radius r around a centre x
j
are influenced by
the movement of the boundary points. When choosing r, it must be noted that
larger values for the support radius lead to more accurate mesh motion. On the
other hand, a very large support radius leads to a dense matrix system, while a
low support radius results in a sparse system which can be solved efficiently using
common iterative techniques.
In table 3.1 various radial basis functions with compact support are shown
using the scaled variable ξ = x/r. The first four are based on polynomials (Wend-
land, 1996). These polynomials are chosen in such a way that they have the
lowest degree of all polynomials that create a C
n
continuous basis function with
n ∈ {0, 2, 4, 6}. The last four are a series of functions based on the thin plate spline
3.2 Different mesh deformation techniques 57
Ref. nr. RBF Name f(ξ)
1 CP C
0
(1 −ξ)
2
2 CP C
2
(1 −ξ)
4
(4ξ + 1)
3 CP C
4
(1 −ξ)
6
(
35
3
ξ
2
+ 6ξ + 1)
4 CP C
6
(1 −ξ)
8
(32ξ
3
+ 25ξ
2
+ 8ξ + 1)
5 CTPS C
0
(1 −ξ)
5
6 CTPS C
1
1 +
80
3
ξ
2
−40ξ
3
+ 15ξ
4

8
3
ξ
5
+ 20ξ
2
log(ξ)
7 CTPS C
2
a
1 −30ξ
2
−10ξ
3
+ 45ξ
4
−6ξ
5
−60ξ
3
log(ξ)
8 CTPS C
2
b
1 −20ξ
2
+ 80ξ
3
−45ξ
4
−16ξ
5
+ 60ξ
4
log(ξ)
Table 3.1 Radial basis functions with compact support. Radial basis functions with compact
support are non-zero within the range of the support radius r. Note that ξ = x/r. Taken from
Wendland (1996).
which creates C
n
continuous basis functions with n ∈ {0, 1, 2} (Wendland, 1996).
There are two possible CTPS C
2
continuous functions which are distinguished by
subscript a and b.
Radial basis functions with global support
In contrast to the functions with compact support, functions with global support
are not equal to zero outside a certain radius, but cover the whole interpolation
space. Radial basis functions with global support generally lead to dense matrix
systems, which can be improved by multiplication with a smoothing function, as
will be discussed in section 3.5.
Table 3.2 shows six radial basis functions with global support which are com-
monly used in e.g. neural networks, computer graphics (Carr et al., 2003) and
for data transfer in fluid-structure interaction computations (Smith et al., 2000,
Boer de et al., 2007).
Ref. nr. RBF Name Abbrev. f(x)
9 Thin plate spline TPS x
2
log(x)
10 Multiquadratic Bi-harmonics MQB

a
2
+x
2
11 Inverse Multiquadratic Bi-harmonics IMQB
_
1
a
2
+x
2
12 Quadric Bi-harmonics QB 1 +x
2
13 Inverse Quadric Bi-harmonics IQB
1
1+x
2
14 Gaussian Gauss e
−x
2
Table 3.2 Radial basis functions with global support. Radial basis functions with global
support cover the whole interpolation space, i.e. the computational domain.
58 Mesh deformation techniques for flapping flight
The MQB and IMQB methods use a shape parameter a, which controls the
shape of the radial basis function. A large value of a gives a flat sheetlike function,
whereas a small value of a gives a narrow cone-like function. The value of a is
typically chosen in the range 10
−5
− 10
−3
. More information about RBF’s and
their error and convergence properties can be found in (Buhmann, 2000, Wend-
land, 1999, 1998). Boer de et al. (2007) compared the resulting mesh quality using
all radial basis functions from table 3.1 and 3.2. The best results were obtained
using the continuous polynomial C
2
, the functions based on a continuous thin
plate spline, C
1
, C
2
a
and C
2
b
and finally the globally supported thin plate spline.
In section 3.4 these six radial basis functions are applied to our test problem and
the best one is used throughout the current thesis. First, the mesh quality mea-
sures, skewness and non-orthogonality, are discussed to compare the mesh quality
for the different mesh motion solvers.
Absolute and relative radial basis function interpolation
In principle there are two different ways to implement this RBF mesh motion
method, the absolute and the relative implementation. The absolute method per-
forms a direct solve of the system (3.8) only once at the beginning of the simulation.
The coefficient arrays γ and β are calculated and used to calculate the internal
point displacements at all time-steps. This method is very efficient since the di-
rect matrix solve, which is more expensive than the evaluation, is only performed
initially. On the other hand, the mesh quality is limited since the coefficients are
not defined with respect to the previous time-step. Therefore, the relative method
is used when very large boundary displacements occur, like a 180

rotation.In this
method the inverse is calculated at every time-step and the motion is defined with
respect to the previous time-step. For reasonably small boundary displacements,
it is much cheaper to use the absolute method. When using the relative imple-
mentation it is important to use different techniques to decrease the number of
boundary points, resulting in lower computation costs (see section 3.5).
3.3 Mesh quality measures
In section 2.4 the skewness and non-orthogonality definition were introduced. In
order to compare the quality of the different meshes after mesh motion it is impor-
tant to elaborate on how to interpret those two mesh quality measures (Knupp,
2003). These mesh quality measures are based on the cell properties such as size,
orientation, shape and skewness. The skewness and non-orthogonality are written
to scalar fields such that they can be used for post-processing.
The test cases used to compare the mesh quality of the different mesh motion
solvers, contain a Cartesian grid around a square box, leading to optimal initial
mesh quality, this is shown in section 3.4. It is important that the ideal mesh
motion solver maintains high quality in terms of skewness and non-orthogonality
after mesh deformation.
3.4 Comparison of mesh motion solvers 59
In section 2.4 it is shown that mesh skewness should be within 0 and 1, and
that the mesh non-orthogonality, which is an angle, should be within 0

and 90

.
In both cases, a lower value means a higher mesh quality. Therefore, desirable
mesh quality bounds are:
0.0 ≤ f
skewness
≤ 1.0
0

≤ f
non−ortho
≤ 90

(3.10)
When assessing the mesh quality it is important to analyse the maximal and aver-
age values. The maximal value provides an indication if the numerical simulation
will be stable and converge at all. If the worst cell quality is too low, the simulation
will diverge within a couple of iterations. On the other hand, the average value
of the mesh quality measure will provide an indication of the average quality of
the mesh. The higher the average quality of the mesh, the more stable, accurate
and efficient the computation will be. In the next section, both the average and
minimal value of the skewness and non-orthogonality mesh quality measures are
used to compare the mesh motion solvers.
3.4 Comparison of mesh motion solvers
In section 3.2, three different kinds of mesh motion solvers are described, based on
solving the Laplace equation, solving the solid body rotation (SBR Stress) equa-
tion and based on interpolation using radial basis functions (RBF). This section
introduces with three numerical test cases which were performed to investigate the
differences in mesh quality obtained with the different mesh motion solvers. Addi-
tionally, the effect of different radial basis functions is discussed. The first simple
case is a two-dimensional block which performs a combination of translation and
rotation. The initial computational mesh is shown in figure 3.1. The domain size
of the test problem is limited to 25D x 25D and the size of the moving block
is 5D x 1D, the grid spacing corresponds to 1D in order to obtain a Cartesian
grid as can be seen in the figure. Mesh motion simulations are performed using
the different mesh motion solvers and a variation of the radial basis function. Af-
ter this simple model problem, section 3.4.2 deals with the mesh motion around
a three-dimensional flapping wing, followed by an example of a two-dimensional
flexible moving boundary in section 3.4.3.
3.4.1 Translation and rotation of a two-dimensional block
The first test case considers a combined motion of translation and rotation to com-
pare the mesh motion solvers under these conditions. The two-dimensional block
is initially centred and translates 2.5D in both X-and Y -direction. In addition,
the block is rotated around its translating centre with 57.3

(1.0 rad). The outer
boundary points are kept fixed, such that the influence of the moving boundary
points on all internal points could be studied independently. The resulting mesh
60 Mesh deformation techniques for flapping flight
Figure 3.1 Initial mesh around a moving block. The initial hexahedral mesh with optimal
quality around a moving block. The domain size is (25D x 25D) around a block with size (5D x 1D),
at every unit spacing, a grid point is places such that an optimal hexahedral mesh is obtained.
quality, after the combined translation and rotation, is assessed using the fields of
skewness and non-orthogonality. Using those fields, the maximal and average val-
ues are obtained as well as a complete visualisation of those mesh quality measures,
combined with mesh quality histograms.
Before proceeding to the results, some special settings, applicable to this test
case, need to be described. First, when using the Laplace and SBR Stress mesh
motion solver it is necessary to specify the diffusivity coefficient which decreases
proportionally to the distance from the moving boundary points. Following Jasak
& Tukovi´c (2004) and Jasak (2009), a quadratically decreasing diffusivity coef-
ficient, i.e. decreasing from the moving boundary, was chosen. Secondly, the
boundary conditions need to be set for the motion solver. In both cases, Laplace
and SBR Stress motion solvers, the boundary conditions on all outer boundary
points are set to Dirichlet type with value 0, so the outer boundary points are kept
fixed. Finally, the newly implemented RBF mesh motion solver is used with five
different functions, CP C
2
, CTPS C
1
, CTPS C
2
a
, CTPS C
2
b
and TPS, based on
the assessment performed by Boer de et al. (2007). Figure 3.2 shows the cell non-
orthogonality at maximal mesh deformation for the Laplace and the SBR Stress
motion solvers. As seen in the figure, with these standard methods the mesh qual-
ity near the moving boundary is low, especially near the leading and trailing edges.
The mesh motion method which solves the Laplace equation with a quadratically
decreasing diffusion coefficient, is simply not robust enough to obtain high mesh
quality when the boundary rotates. As can be seen in figure 3.2(a), the mesh
deformation is largest near the boundary, which is not desirable. In cases where
large rotation angles occur, it is best to apply a mesh motion solver, which leads
to the motion of all internal mesh points, coping with the boundary deformation.
The gain in mesh quality by using the SBR Stress method is marginal. From
3.4 Comparison of mesh motion solvers 61
(a) Laplace
(b) SBR Stress
Figure 3.2 Cell non-orthogonality of Laplace and SBR Stress mesh motion solvers. The
cell non-orthogonality is compared of the Laplace (a) and the SBR Stress (b) motion solvers, lower
(blue) is better. The non-orthogonality is visualised for a two-dimensional block with a combined
motion of translation and rotation. The boundary points translate over a distance of 2.5D in both X-
and Y -direction and rotate 57.3

(1.0 rad) around the translating centre.
figure 3.2(b) it can be seen that the cells with a high non-orthogonality occur at a
distance from the moving boundary. Still, the mesh quality near the body surface,
especially near the leading and trailing edges, needs improvement. Improvement
can be obtained by specifying a constraint to an inner mesh region, such that it
moves according to the body motion.
Figure 3.3(a) shows the resulted mesh obtained using RBF interpolation using
a thin plate spline function. When comparing the cell non-orthogonality with
figure 3.2(a) and 3.2(b) it may be seen that most of the cell deformation, with
the RBF mesh motion, occurs in the outer regions of the mesh and all cells are
dealing with the boundary displacement. Additionally, figure 3.3(b) shows the
resulting mesh, which is obtained using the relative implementation of the RBF
mesh motion method. The rotation of the boundary is very large, 180

, and the
mesh remains valid. This method is very robust but computationally expensive as
will be shown in section 3.5.
Table 3.3 shows a quantitative comparison of the resulting mesh quality ob-
tained with different mesh motion solvers, including different RBF’s. The max-
imal and averaged values of both skewness and non-orthogonality are compared.
The mesh quality calculated by the Laplace method is low, since the maximal
skewness and maximal non-orthogonality are large, respectively f
s
max
= 0.95 and
f
o
max
= 72.1. The results of the Laplace and SBR Stress mesh motion solver are
very similar, within 6.1%. Concerning the RBF method, four RBF’s with compact
support are used (CP C
2
, CTPS C
1
, CTPS C
2
a
and CTPS C
2
b
) and one with global
support, the thin plate spline TPS. In order to neglect the effect of the support
62 Mesh deformation techniques for flapping flight
(a) RBF absolute (b) RBF relative
Figure 3.3 Cell non-orthogonality of relative and absolute RBF mesh motion solvers.
The cell non-orthogonality is shown for the relative and absolute versions of the RBF mesh motion
solver. (a) shows the absolute and (b) the relative implementation, lower (blue) is better. The non-
orthogonality is visualised for a two-dimensional block with a combined motion of translation and
rotation. The boundary points translate a distance of 2.5D in both X- and Y -direction. The rotation
is 57.3

(1.0 rad) for (a) and 180

(3.14 rad) for (b).
radius, it is set to r = 75 which is about 3 times the domain size such that the
results in table 3.3 are independent of the support radius for r > 75. The RBF
method provides high mesh quality, maximal and averaged, for both C
2
and TPS
compared to the other functions. The C
2
and TPS RBF’s result in an maximal
skewness of respectively −32% and −45% compared to the Laplace motion solver,
while the difference of the other RBF’s is only about 10%. Similar results are
shown in the table for the average skewness and maximal orthogonality. Con-
cerning the average orthogonality, the C
2
RBF is outperformed by the TPS RBF,
which is the only mesh motion solver resulting in a lower value compared to the
Laplace method. Overall, the basis function thin plate spline provides the high-
est mesh quality for both skewness and non-orthogonality, such that this globally
supported function was used for the current investigations.
Finally, in addition to the non-orthogonality visualisations, figure 3.4 and 3.5
show the histograms of cell non-orthogonality and skewness for the Laplace and
RBF mesh motion solver, which are considered to result in the worst and best
mesh quality, respectively. For both non-orthogonality and skewness, the RBF
mesh motion results in a smooth profile, which emphasises the fact that all internal
cells are coping with the mesh motion.
3.4.2 Flapping of a three-dimensional wing
It was shown that high mesh quality was obtained for a simplified two-dimensional
test case, by using radial basis function interpolation. Especially, the globally sup-
3.4 Comparison of mesh motion solvers 63
Method (f
s
)
max
(f
s
)
ave
(f
o
)
max
(f
o
)
ave
Laplace 0.95 (baseline) 0.09 (-) 72.1 (-) 20.1 (-)
SBR Stress 0.96 (+0.7%) 0.08 (-4%) 75.1 (+4%) 21.3 (+6%)
CP C
2
0.65 (-32%) 0.065 (-28%) 59.6 (-17%) 23.4 (+16%)
CTPS C
1
0.86 (-10%) 0.105 (+17%) 74.5 (+3%) 31.9 (+59%)
CTPS C
2
a
0.81 (-15%) 0.103 (+14%) 73.8 (+2%) 31.4 (+56%)
CTPS C
2
b
0.88 (-7%) 0.11 (+22%) 76.0 (+5%) 32.8 (+63%)
TPS 0.52 (-45%) 0.051 (-41%) 52.5 (-27%) 19.0 (-6%)
Table 3.3 Comparison of mesh quality for different mesh motion solvers. The mean and
maximal values of the skewness f
s
and non-orthogonality f
o
are compared at maximal displacement
and rotation of the two-dimensional rectangular block. Results are shown for the Laplace, SBR Stress
and RBF mesh motion solver, the latter using different RBF’s.
Non-orthogonality
N
u
m
b
e
r
o
f
c
e
l
l
s
0 10 20 30 40 50 60 70 80 90
0
10
20
30
40
50
60
70
80
90
(a) Laplace
Non-orthogonality
N
u
m
b
e
r
o
f
c
e
l
l
s
0 10 20 30 40 50 60 70 80 90
0
10
20
30
40
50
60
70
80
90
(b) RBF
Figure 3.4 Cell non-orthogonality histograms for Laplace and RBF mesh motion solvers.
Mesh quality histograms show the variation in non-orthogonality for Laplace (a) and RBF mesh motion
solvers (b) using the TPS.
ported thin plate spline (TPS) provided high quality and robust mesh deformation.
To show the three-dimensional capabilities of the RBF mesh motion solver, the
mesh around a flapping wing is shown in figure 3.6 at mid-stroke. Figure 3.6(a)
shows the non-orthogonality during the downstroke, while figure 3.6(b) presents
the mesh quality halfway of the upstroke. The TPS was used as radial basis
function, without any user input, since it has global support. From the figure it is
clear that a large part of the near wake is deformed in order to deal with the three-
dimensional wing motion. Concerning the RBF mesh motion solver, the cells close
to the wing take a larger part of the deformation compared to the Laplace method.
This was already illustrated for the moving two-dimensional block in figures 3.4
and 3.5, showing the histograms of the non-orthogonality and skewness, respec-
64 Mesh deformation techniques for flapping flight
Skewness
N
u
m
b
e
r
o
f
c
e
l
l
s
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
50
100
150
200
250
300
350
400
450
500
(a) Laplace
Skewness
N
u
m
b
e
r
o
f
c
e
l
l
s
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
50
100
150
200
250
300
350
400
450
500
(b) RBF
Figure 3.5 Cell skewness histograms for Laplace and RBF mesh motion solvers. Mesh
quality histograms show the variation in skewness for Laplace (a) and RBF mesh motion solvers (b)
using the TPS.
tively. The Laplace method results in a small number of cells with large values
for both non-orthogonality and skewness, while the RBF mesh motion technique
leads a smoother decline of both quality measures in the histograms.
When comparing figure 3.6(a) and 3.6(b) it is seen that the mesh during both
the upstroke and downstroke is symmetric. This is caused by the fact that the
radial basis function interpolation determines the new internal mesh points with
respect to the initial mesh, such that the initial mesh is recovered after every
flapping period.
As discussed in subsection 3.2.3, the radial basis function mesh motion con-
tains a direct system solve and an evaluation to determine the displacement of all
internal mesh points. Therefore, the mesh deformation for three-dimensional cases
may become very expensive, such that it is necessary to implement techniques to
improve its efficiency. These techniques are described in section 3.5. But first, the
RBF mesh motion solver is tested by employing the flexing of a two-dimensional
moving boundary.
3.4.3 Flexing of a two-dimensional block
Within the context of the present research, performing a fluid-structure-interaction
simulation is too computationally demanding for a full three-dimensional flapping
wing. Therefore, the effects of wing flexibility was incorporated by defining the
wing flexing using harmonic functions, to mimic realistic insect wing deforma-
tion (Shyy et al., 2008a). In order to show that the RBF mesh motion method
is able to deal with a flexing boundary, the model problem of a moving block is
used. The motion of the two-dimensional block can be decomposed into transla-
tion, rotation and flexing, all defined with respect to the initial configuration. The
3.4 Comparison of mesh motion solvers 65
(a) t = 0.25T (b) t = 0.75T
Figure 3.6 Cell non-orthogonality for a three-dimensional wing. Cell non-orthogonality of
a mesh around a three-dimensional flapping model wing, obtained with the RBF mesh motion solver
with the globally supported thin plate spline (TPS), lower (blue) is better. (a) shows the mesh quality
at t = 0.25T and (b) at t = 0.75T, respectively during the downstroke and upstroke.
translation and rotation are defined by:
x
t
= A
t
· sin(2πft),
α = A
α
· sin(2πft),
where, A
t
and A
α
represent the translation and rotation amplitude vectors, f the
frequency and t the time. The flexing of the boundary is defined by combining
two harmonic functions like:
x
f
= A
f
· cos(2π
x

d
c
2 · c
f
) · sin(2πft)

d
f
,
where A
f
is the flexing amplitude vector, d
c
is the direction vector in-plane of the
flexing surface, c
f
the length of the flexing surface and d
f
represents the direction
vector of the flexing. In this model problem of a moving two-dimensional block, the
amplitudes of both translation and rotation were fixed to A
t
= (2.5, 2.5, 0.0) and
A
α
= (0.0, 0.0, 1.0), respectively. The flexing was defined such that the main flex-
ing direction is perpendicular to the flexible boundary surface, d
f
= (0.0, 1.0, 0.0)
with a flexing amplitude of A
f
= 0.5 which is about 10% of the flexible boundary
length, c
f
= 5.0. Figure 3.7 shows the resulting mesh deformation at t = 0.25T
and t = 0.75T where T = 1/f is the motion period. It can be seen that the whole
mesh is deformed by the RBF mesh motion solver, like was the case with a rigid
airfoil. Still, some high non-orthogonality can be observed within a region of 1−2
block lengths, which is mainly caused by the fixed points on the outer boundary
and the small computational domain. A larger domain will undoubtedly lead to
high quality meshes when using RBF mesh motion.
66 Mesh deformation techniques for flapping flight
(a) (b)
Figure 3.7 Deforming mesh around a flexible block using RBF mesh motion. A flexible
rectangular block which is translating, rotating and flexing using the RBF mesh motion solver with
the globally supported thin plate spline (TPS). The flexing is defined by simple harmonic functions.
(a) shows the mesh deformation at t = 0.25T and (b) at t = 0.75T.
3.5 Improving computational efficiency
When using direct methods, the computational costs of the new mesh motion solver
based on radial basis function interpolation, increases fast with an increased num-
ber of boundary and internal points. The method consists of two computationally
expensive steps:
1. Solving the system of equations (3.8) for given boundary points x
b
i
and cor-
responding displacements ∆x
b
i
to find the coefficient vectors γ and β. The
upper diagonal block matrix Φ
bb
is in general a dense symmetric matrix of
size (N
b
x N
b
). Therefore, standard direct solvers require O(N
3
b
) operations.
2. Evaluation of the radial basis function summation, equation (3.7), at all N
in
(of order O(N
in
)) internal mesh points, using the given boundary points x
b
i
and the in step 1 computed coefficient vectors γ and β. This evaluation
leads to a computational cost of order O(N
b
N
in
).
For large two- and three-dimensional meshes both the system solve and the eval-
uation procedures may become very computationally expensive, especially when
direct methods are used. In order to illustrate the increasing costs with increasing
number of mesh points, a two- and three-dimensional Cartesian grid is considered
with a uniform mesh distribution. The computational domain is square shaped
with an equal number of points on all edges of the boundary, equal to N
b
. Table 3.4
shows the total number of internal and boundary points as a function of N
b
, for
both the two- and three-dimensional example. Additionally, the total number of
operations for a direct solve and the RBF evaluation is given in the table, which
3.5 Improving computational efficiency 67
Internal points All boundary points Direct solve RBF evaluation
2D N
2
b
4N
b
64N
3
b
4N
3
b
3D N
3
b
6N
2
b
216N
6
b
6N
5
b
Table 3.4 Computational costs for solving the system and evaluate the RBF’s. Computa-
tional costs for solving the system and evaluation the RBF’s on all internal mesh points. Illustration of
a two-dimensional and three-dimensional uniform square shaped Cartesian mesh with an equal number
of boundary nodes N
b
in x- and y-direction.
shows that the computational costs for both a direct solve and evaluation scales
with N
3
b
for the two-dimensional case. Concerning the three-dimensional case, the
computational costs for the direct system solve are a factor N
b
larger compared
to the costs for the RBF evaluation. So when a complex three-dimensional case,
like a flapping wing, is considered, special treatment is necessary concerning the
system solve, e.g. by reducing the number of boundary nodes or using advanced
direct solver techniques.
3.5.1 Boundary point coarsening and smoothing
From table 3.4 it is seen that the total computation costs will decrease if a con-
straint is put on the number of mesh points. This can be achieved in two different
ways. First, a major part of the computational cost is spent at solving the system,
step 1, which is a factor N
b
more expensive than the evaluation. It seems rea-
sonable to reduce the number of moving boundary points by performing a sound
selection procedure. Especially when the body displacement follows a rigid body
motion, not all boundary points are necessary. Therefore, a coarsening technique
was incorporated which selects a boundary point for every ξ points, where ξ is
the coarsening factor. More advanced coarsening techniques, based on greedy
algorithms are applied by Rendall & Allen (2008a).
Secondly, the efficiency of the RBF method is improved, by the notion that
all outer boundary points are fixed in general. Therefore, the outer boundary
points can be neglected. This is achieved by specifying a smoothing function
such that the RBF contribution reduces to zero at the outer boundary, which is
defined (Jakobsson & Amoignon, 2007) as
ψ(¯ x) =
_
_
_
1, ¯ x ≤ 0,
1 − ¯ x
2
(3 −2¯ x), 0 ≤ ¯ x ≤ 1,
0, ¯ x ≥ 1,
(3.11)
where, ¯ x is given by:
¯ x =
||x
i
−R
inner
||
R
outer
−R
inner
, (3.12)
x
i
represents the coordinate of the i−th inner mesh point, to evaluate this smooth-
ing function at that particular location in space. R
inner
and R
outer
are two radii,
68 Mesh deformation techniques for flapping flight
between those radii this smoothing function is decreasing from 1.0 to 0.0. When
the RBF evaluation function (3.9) is multiplied by φ(¯ x) the corrected RBF eval-
uation is obtained by:
∆¯ x
in
j
= s(x
in
j
)

ψ(¯ x). (3.13)
Within R
inner
the contribution of the RBF evaluation remains unaltered, but out-
side R
outer
the value of the RBF becomes zero, such that all fixed outer boundary
points may be neglected. In principle, the inner radius is chosen to be multiple
boundary lengths (wing chords) and the outer radius is chosen as the distance
from the moving boundary to the outer boundary. Therefore, the system to be
solved, only contains the control points on the moving boundary, selected by the
coarsening function.
In addition, it is interesting to address the computing times concerning the
different mesh motion solvers. Figure 3.8(a) shows the computing times with in-
creasing grid resolution for the Laplace, SBR Stress and RBF mesh motion solvers.
Concerning the RBF mesh motion solvers, the three described variants are used,
the absolute implementation, relative implementation and the relative method in
combination with the previously dealt coarsening and smoothing techniques. It
is clear that the mesh motion solvers based on solving a partial differential equa-
tion are very fast, since standard iterative techniques can be used for these sparse
systems. If the absolute and relative RBF methods are compared, it is observed
that these methods require very large computing times, at least an order of mag-
nitude more. On the other hand, if the coarsening and smoothing techniques are
applied the computing times are of similar order compared with the fast Laplace
mesh motion. The non-linear behaviour of the final curve is caused by the choice
of the coarsening ratio to select the moving boundary points. While keeping the
coarsening ratio fixed, the mesh resolution is increased, the number of boundary
points used in the system solving is still growing non-linearly. For high resolution
meshes, the order of computing times can be decreased further by increasing the
coarsening ratio.
Figure 3.8(b) shows the computing times, per time-step, concerning a three-
dimensional flapping wing simulation (6 flapping periods). These three-dimension-
al simulations are performed for grid sizes of 100k, 200k, 400k, 800k and 1600k
cells and needed a total computing time of about 8, 32, 48, 103, 190 hours, respec-
tively. All simulations were performed on four CPU cores of an AMD Opteron

280 cluster. Figure 3.8(b) shows super-linear curves for both solving the flow
equations and the RBF mesh motion. Validation showed that a mesh resolution
of 800k provided accurate results for flapping wing aerodynamics. Additionally,
figure 3.8(b) shows that the computing time used for RBF mesh motion is less
than 10% of the computing time for the complete time-step. This is considered to
be very efficient, it must be noted that mesh coarsening and a smoothing function
are applied for acceleration.
3.5 Improving computational efficiency 69
Grid spacing
C
o
m
p
u
t
i
n
g
t
i
m
e
[
s
]
10
0
10
1
10
2
10
0
10
1
10
2
10
3
10
4
(a) Two-dimensional
Grid spacing
C
o
m
p
u
t
i
n
g
t
i
m
e
[
s
]
Mesh motion
Flow equation
Total timestep
0 1 2 4 8 16
0
10
20
30
40
50
60
70
(b) Three-dimensional
Figure 3.8 Computing times for different mesh motion solvers. (a) shows a comparison
of computing times for the mesh motion solvers based on the Laplace equation (◦), the SBR Stress
equation (), RBF absolute method (), RBF relative method (×) and RBF relative method in combi-
nation with coarsening and smoothing techniques (•). The computing times for the Laplace equation
(◦) and the SBR Stress equation () are nearly identical. (b) shows the computing times for one
time-step of a three-dimensional flapping wing simulation. The times are subdivided by solving the
RBF mesh motion and the flow equations. Meshes are used from 100k to 1600k.
3.5.2 Iterative techniques and parallel implementations
As was shown in the previous section, the two different phases of the radial ba-
sis function mesh motion, solving (I) and evaluation (II) can be very expensive.
However, it is not necessary to solve (I) and (II) exactly, since the internal point
motion can be arbitrary, as long as the resulting mesh is of sufficiently high quality.
Different pre-conditioning techniques are described in literature (Boer de et al.,
2007) to approximate the system of equations (3.8), leading to a system which is
more efficient to solve using existing iterative solver techniques.
Furthermore, a mesh motion problem leads in general to an ill-conditioned
system, which is difficult to solve directly and iteratively. The condition number in
case of the model problem of a moving square, section 3.4, was about O(10
10
). The
high condition numbers, which occur in this type of problems, are caused by the
boundary point locations. In general moving mesh applications, the combination of
cell clustering on the moving surface and the large distance to the outer boundary,
causes large differences between points in Φ
bb
. Despite the pre-conditioning, the
ill-conditioned system cannot be efficiently solved using iterative techniques. A
possible better choice would be to use parallel direct techniques, available in the
linear algebra packages SuperLU and ScaLAPACK, a parallel version of LAPACK.
A different way to improve the efficiency of the computation has already been
applied in section 3.4, which is about reducing the number of boundary points
by applying coarsening and smoothing techniques, with a significant gain in com-
puting time. Concerning the coarsening, complex greedy algorithms (Rendall &
70 Mesh deformation techniques for flapping flight
Allen, 2008a) may be applied to select the necessary boundary points such that
the efficiency is increased even further. Another method to increase the efficiency
to solve the system (3.8) is to decrease the condition number by only taking the
boundary points with a low mutual distance (Boer de et al., 2007), in addition,
re-ordering can be applied for further enhancement.
Finally, the speed of the evaluation (II) can be increased by various fast evalu-
ation algorithms (Boer de et al., 2007). Furthermore, the evaluation can be easily
implemented in parallel since it only involves a matrix-vector multiplication. The
major difficulty, concerning a parallel implementation, is that all processor parti-
tions need to know which control point belongs to itself and which to the other
partitions. Then every processor only performs the evaluation of the internal
points, of that particular partition, using all control points and corresponding co-
efficients, α
i
and β
i
, which are distributed over all partitions. Currently, this is
being implemented in OpenFOAM

.
3.6 Conclusions
In this chapter two different mesh motion techniques were described which are
commonly used within the code of OpenFOAM

, both based on solving a partial
differential equation. The first method solves the Laplace equation with a variable
diffusion coefficient, which is used to control the final mesh quality. Secondly, the
linear stress equation was modified to include rigid body rotations in order to cope
with the severe mesh deformation present in flapping wing simulations. As with
the Laplace equation, the solid body rotation stress (SBR Stress) mesh motion
uses the diffusivity, acting as a stiffness, to influence the quality of the mesh. The
diffusivity, in both cases, is defined to decrease quadratically with the distance
from the moving boundary.
Besides solving a partial differential equation the motion of mesh points can
be defined using interpolation techniques. A new mesh motion solver is incor-
porated in OpenFOAM

, which uses the interpolation of radial basis functions
(RBF). For given boundary point displacements the internal mesh displacements
are obtained by solving a system of equations to obtain an array of interpolation
coefficients. Using those coefficients, the internal point displacements are obtained
by evaluating the radial basis functions.
This new mesh motion technique does not need any information about the mesh
connectivity and can be applied to arbitrary unstructured meshes containing poly-
hedral cells, the way OpenFOAM

deals with the finite volume implementation.
The three mesh motion solvers are tested using a case of a two-dimensional rect-
angular block which moves through a Cartesian mesh. The cell non-orthogonality
and skewness are compared. Additionally, different radial basis functions, con-
cerning the RBF mesh motion, are compared. The RBF mesh quality provides
superior mesh quality over the Laplace and SBR Stress mesh motion solvers. Es-
pecially when using the thin plate spline (TPS) or the continuous polynomial C
2
3.6 Conclusions 71
as radial basis functions, the mesh quality is high in terms of low skewness and
non-orthogonality. The TPS has global support, whereas the C
2
basis function
has compact support. The RBF mesh motion was successfully tested on simple
test problems and for a three-dimensional flapping wing with the possibility to
incorporate a flexing moving boundary.
Since the RBF mesh motion technique encounters a dense system of equations,
different methods are implemented to increase its efficiency. First of all, a subset
of the moving boundary points was selected, because not all points are necessary
when the body performs a rigid body motion. So a coarsening algorithm selects
those control points. Secondly, a smoothing function is used to decrease the RBF
contribution to zero at the outer domain boundaries. Therefore, it is justified to
neglect the outer (fixed) boundary points, which reduces the system of equations
considerably. After this elaborate discussion it is concluded that, concerning the
three-dimensional wing simulations, the globally supported TPS should be used
in combination with the coarsening and smoothing techniques to increase the ef-
ficiency of the RBF mesh motion method.
CHAPTER 4
Physical and numerical modelling
of flapping foils and wings
In this chapter, the physical and numerical modelling of flapping wing and foils is
described. The relevant dimensionless numbers (Strouhal, Reynolds and Rossby
numbers) are identified after writing the Navier-Stokes equations in a rotating
reference frame. To systematically study the aerodynamics around flapping wings,
a model planform and kinematic model is defined. The kinematic model, which
describes the wing motion, consists of a rigid body motion appended by a flexing,
representing occasional wing deformation. Both geometry and wing kinematics are
dynamically scaled in order to design a sound framework for comparison, using
the radius of gyration. Additionally, the force coefficients are used in conjuncture
with the lift-to-drag ratio to assess the flapping wing performance.
4.1 Introduction
This chapter deals with the physical and numerical modelling of flapping wings, in
hovering as well as in forward flight conditions. Animals that fly or swim, which are
equivalent from the fluid-dynamic perspective (Triantafyllou et al., 1993, Taylor et
al., 2003) at certain scales, undergo significant interactions with the environmental
fluid in which they move. Therefore, for the mathematical analysis of swimming
or flying it is important to formulate the governing equations and accompanying
boundary conditions in an appropriate form. These equations are used to deduce
the dimensionless numbers relevant for insect flight.
74 Physical and numerical modelling of flapping foils and wings
The Reynolds number
In order to improve our understanding of biological flows, like the flows around
flapping wings or fins, it is of importance to make extensive use of dimensionless
numbers, like the Reynolds (Re) and Strouhal (St) number in particular (e.g. Pan-
ton, 2005, White, 1991). The Reynolds number is defined as the ratio between
the inertial and viscous forces present in a fluid. It is a property of the flow, such
that it identifies what kind of propulsive mechanism applies to the flapping wing.
For example, when a flapping wing operates at a very low Reynolds number, i.e.
Re = O(100), the forces in the flow are dominated by the viscous term, compared
to the inertial component, such that viscous phenomena, like shear layers and
vortex generation, will be more pronounced. Additionally, the Reynolds number
determines whether the flow behaves turbulent or laminar, defining implicitly the
complexity of the mathematical model needed to solve the problem. A different
approach to dimensionless numbers is to define them as the ratio of time or length
scales, instead of the ratio between two distinct forces (Tennekes & Lumley, 1972,
Lentink & Gerritsma, 2003, Bos et al., 2008). In that case the Reynolds number
can be defined as the ratio between the convection time over the diffusion time,
see chapter 2, which is easier to understand and to apply to a flow problem, such
as the Von K´ arm´ an vortex street behind a bluff body.
The Strouhal number
Besides the Reynolds number, the Strouhal number plays an important role in
flapping wing aerodynamics as well. The typical definition of the Strouhal num-
ber is the flapping frequency times flapping amplitude divided by a reference flow
velocity. The first use of the Strouhal number was in the context of the natural
vortex shedding behind a stationary cylinder in a uniform flow. Williamson (1988)
found a universal relation between the Reynolds and Strouhal number based on the
observed vortex-shedding frequency in the laminar flow regime. For flapping wings
or oscillating bodies, the Strouhal number can be defined based on the imposed
oscillation frequency and amplitude. For moving bodies and especially flapping
wings, the Strouhal number can be very useful. For example, in forward flapping
flight, the Strouhal number is proportional to the maximum value of the induced
angle of attack, provided that the wing flaps in a stroke plane perpendicular to the
forward velocity (Lentink & Gerritsma, 2003, Taylor et al., 2003, Thaweewat et al.,
2009). This definition of the Strouhal number, based on the flapping amplitude, is
closely related to the advance ratio, as defined by Ellington (1984), J = U/4Φ
0
fR,
which is the ratio between the forward and flapping distance, travelled by the wing.
Here Φ
0
is half of the total flapping angular amplitude, f the flapping frequency
and R represents the distance from root to tip of the wing, i.e. the single wing
span. Another definition was introduced by Dickinson (1994) and Wang (2000b)
using the average chord length as reference, St
c
= f c/U. This expression is very
similar to the reduced frequency (Shyy et al., 2008b), which is commonly used
to relate the two velocities due to either flapping and forward flight. In general
engineering applications, the Strouhal number is commonly used to characterise
4.2 Governing equations for flapping wings 75
the vortex shedding, whereas the reduced frequency number is used in flapping
wing problems. Additionally, the reciprocal of the Strouhal number is known as
the dimensionless wavelength λ

= U/f c which is often used in studies concerning
forward flapping flight as it seems reasonably intuitive in that it corresponds to
the distance travelled over one flapping period, relative to the mean chord.
Equations and other assumptions
As previously discussed, it is appropriate to use dimensionless numbers to study
the effect of flapping characteristics on the aerodynamic performance. In order
to perform a sound and valid comparison it is important to maintain constant
dimensionless numbers while kinematic parameters or flow properties are varied
to study their influence. In accordance with Lentink & Dickinson (2009a), spe-
cific attention is given to the appropriate definition of the governing dimensionless
numbers to investigate the flow around flapping wings.
In view of simplicity, the present study deals with a model wing which is a
simplified representation of a flying insect wing operating at Reynolds numbers,
Re = 100, 500, and 1000, corresponding to the operating conditions of fruit flies,
house flies and bumblebees, respectively. Furthermore, only one flapping wing is
considered, under hovering as well as forward flight conditions, which allows that
the induced vortical flow can be studied in more detail. This implies that no
interaction between two wings or with the body are included. Nevertheless, the
considered flapping kinematics that result in large rotation rates put the current
numerical techniques to a significant challenge.
In section 4.2, the governing equations are formulated for forward and hovering
flapping flight. Secondly, the model wing selection and the definition of the kine-
matic model parameters are described in section 4.3, followed by the dynamical
scaling of flapping flight in section 4.4. As a prelude to the numerical solution
of the governing equations, the mesh generation in combination with the bound-
ary conditions is briefly dealt with in 4.5. Section 4.6 describes the force and
performance definitions, followed by the conclusions of this chapter in section 4.7.
4.2 Governing equations for flapping wings
Concerning flapping flight in nature, like the operation of insects and fish, the
flow can be considered to be incompressible since the Mach number (a measure
for compressibility) is typically Ma = U/a = O(10
−3
) (Brodsky, 1994), where
U [m/s] is the reference velocity and a [m/s] the speed of sound. In section 2.2 the
incompressible Navier-Stokes equations were defined by equation (2.4) and (2.5),
and re-stated here:


u = 0, (4.1)
∂u
∂t
+∇

(uu) = −
1
ρ
∇p +ν∇
2
u, (4.2)
76 Physical and numerical modelling of flapping foils and wings
These equations are derived by analysis of the forces on an infinitely small fluid el-
ement in an inertial reference frame. Using Computational Fluid Dynamics (CFD)
techniques, these equations are solved on a discretised computational domain in
combination with appropriate initial and boundary conditions. When flight under
hovering conditions is considered, the initial velocity field is zero as well as the
boundary conditions at the outer domain boundary. At the boundary of the flap-
ping wing, a no-slip condition (White, 1991) needs to be specified, which means
that the velocity, relative to the wing, has to be zero, in all directions. This is
accomplished by defining the mathematical velocity on the moving boundary to
be equal to the actual wing motion, which moves according to a specific kinematic
model, derived from realistic insect data (e.g. Fry et al., 2003).
Momentum analysis in rotating reference frame
A different numerical approach to solve this problem is to transform the governing
equations and boundary conditions from the inertial reference frame (XY Z) to
a rotating reference frame (xyz), which is fixed to the flapping wing and moves
accordingly. In the present study, the reference frame approach is only used to
identify certain dimensionless numbers that are related to the rotation of the three-
dimensional wing, i.e. the actual flow computations are made with respect to the
inertial reference frame.
The rotating reference frame is attached with its origin to the joint around
which the wing rotates. The resulting boundary condition on the flapping wing
will be such that the effective velocity will be zero, since the reference frame
moves with the boundary. The corresponding velocity transformation, to make
the velocity at the moving boundary equal to zero, is defined as (Ginsberg, 1998,
Baruh, 1999)
u
XY Z
= u
xyz
+ (u
trans
+Ω
wing
×r),
where u
XY Z
corresponds to the velocity in the inertial reference frame, whereas
u
xyz
represents the velocity in the local rotating reference frame, r is the distance
from a rotating point to the origin. u
trans
is the translating velocity of the reference
frame itself and can be used to specify the translation of the insect body. In the
present study, the translation velocity of the rotating reference frame is assumed
to be zero, which the case in hovering flight, approximately. Ω
wing
is the angular
velocity of the rotating reference frame, i.e. representing the flapping motion of
the wing.
The boundary condition needs to take care of the rotation of the reference
frame, resulting in the following expression:
u
wing
= u
WING
−(u
trans
+Ω
wing
×r),
where u
WING
is the flow velocity at the wing in the inertial reference frame. This
relation results in a no-slip condition in the rotating reference frame. Additionally,
it is interesting to study the accelerations (Lentink, 2008) with respect to the
inertial frame (XY Z). The accelerations in the inertial and rotating reference
4.2 Governing equations for flapping wings 77
X
Y
Z
x
y
z
O
Figure 4.1 Illustration of the rotational reference frame. The rotational reference frame xyz is
moving with the wing and obtained by rotating the inertial reference frame XY Z by three orientation
angles.
frames are related using the following (Ginsberg, 1998, Baruh, 1999):
a
XY Z
= a
xyz
+ (a
ang
+a
cen
+a
cor
).
The angular, a
ang
[m/s
2
], centripetal, a
cen
[m/s
2
], and Coriolis a
cor
[m/s
2
] accel-
erations are respectively defined as
a
ang
=
˙
Ω×r,
a
cen
= Ω×(Ω×r),
a
cor
= 2Ω×u
xyz
.
The main parameter in these three different accelerations is the angular velocity
Ω [rad/s], which strongly depends on the wing kinematics. u
xyz
[m/s] is the
velocity in the rotating frame. In order to explore the different acceleration terms,
Ω needs to be related to the flapping motion. This will be elaborated in detail
in section 4.3. Now the expressions for velocity and accelerations in the rotating
reference frame are substituted into the Navier-Stokes equations (4.1) and (4.2)
such that the following transformed Navier-Stokes equations are obtained (for the
sake of simplicity, the subscripts are dropped):
Du
Dt
+ (
˙
Ω×r) + (Ω×(Ω×r)) + (2Ω×u) = −
1
ρ
∇p +ν∇
2
u. (4.3)
This transformed equation describes the momentum balance for a fluid particle
close to the wing (in the boundary layer) in the rotating reference frame. With this
approach it is possible to derive dimensionless numbers representing the different
acceleration terms, in addition to the already described Reynolds and Strouhal
numbers. These new dimensionless numbers will become available if the different
78 Physical and numerical modelling of flapping foils and wings
terms in equation (4.3) are scaled as
u

=
u
U
ref
, t

=
U
ref
c
t, ∇

= c · ∇,
˙


=
˙

˙

, Ω

=


, r

=
r
R
, p

=
p
ρ
ref
· U
2
ref
,
leading to (dropping the stars for simplicity):
U
ref
f · c
·
Du
Dt
+
˙
ΩRc
U
2
ref
·(
˙
Ω×r)+

2
Rc
U
2
ref
·(Ω×(Ω×r))+
Ωc
U
ref
·(2Ω×u) = −∇p+
ν
U
ref
c
·∇
2
u,
where U
ref
[m/s] is the reference velocity, f [1/s] the flapping frequency, which
is defined as f =
U
ref
c
, where c the average chord length. L [m] is the reference
length, Ω [ rad/s] and
˙
Ω [rad/s] are the average rotational velocity and acceler-
ation, respectively. Furthermore, ρ
ref
[kg/m
3
] is the reference density (constant
in incompressible flows). R [m] is a radius length. In addition to the Reynolds
and Strouhal numbers, other dimensional numbers can be identified, related to
the rotation of the reference frame, which is still attached to the wing, rewriting
gives:
1
St
·
Du
Dt
+
1
C
ang
· (
˙
Ω×r) +
1
C
cen
· (Ω×(Ω×r)) +
1
Ro
· (2Ω×u) = −∇p+
1
Re
· ∇
2
u,
where the additional dimensionless number are respectively defined as
C
ang
=
U
2
ref
˙
ΩR c
, (4.4)
C
cen
=
U
2
ref

2
R c
, (4.5)
Ro =
U
ref
Ω c
. (4.6)
By the rotation amplitude Ω, these dimensionless numbers strongly depend on the
wing kinematics, especially the velocity and acceleration due to the wing rotation.
Therefore, the effect of different kinematics and flow features may be related to
these dimensionless numbers.
4.3 Wing shape and kinematic modelling
When investigating the three-dimensional aerodynamics around flapping wings,
one may try to simulate the geometry and conditions of specific species, in order
to fully understand them, or to opt for a more generic approach. Previous inves-
tigations of specific insect species have been reported for e.g. a fruit fly (Sane &
Dickinson, 2001, Birch & Dickinson, 2003), hawkmoth (Liu & Kawachi, 1998) or
4.3 Wing shape and kinematic modelling 79
a dragonfly (Isogai et al., 2004). The present study follows the second approach
and considers a three-dimensional ellipsoidal model wing (Bos et al., 2008); a sim-
ilar ellipsoid was used by Wang et al. (2004) for their two-dimensional research.
Section 4.3.1 briefly describes the wing shape and morphology, while section 4.3.2
deals with the kinematic modelling. The general kinematic modelling consists
of a translation and a rotation component. In addition, the kinematic model
was appended with an active wing flexing component, which is described in sec-
tion 4.3.3. The numerical implementation of the wing kinematics is the subject of
section 4.3.4.
4.3.1 Wing shape and planform selection
The general model wing is described by an ellipsoid in the three-dimensional wing
reference frame:
_
x
a
_
2
+
_
y
b
_
2
+
_
z
c
_
2
= 1, (4.7)
where a, b and c are the semi-axes of the ellipsoidal wing. In order to obtain a wing
which has a single wing span, b
s
= 2.0, a maximal chord length c
max
= 1.0 and a
thickness of 10% of the chord, the semi-axes are chosen as a = 0.05, b = 0.5 and
c = 1.0 (for the two-dimensional airfoil, a similar elliptical cross-section is applied
for z = 0). Previous studies show that specific insect features, like the corrugated
wing planform (Luo & Sun, 2005) are of minor influence on the resulting fluid
behaviour. The three-dimensional elliptical planform is shown in figure 4.2 in
comparison to a more realistic representation of the fruit fly wing planform.
Since the planform is analytically given by equation (4.7), the average chord
length can be obtained by integration of the chord distribution c(r) along the wing
span from root to tip:
c =
1
R
_
R
0
c(r)dr, (4.8)
where R [m] is the radius of the wing tip and c(r) [m] the chord distribution along
the wing span. Assuming that the chord c(r) is represented by y(z) the following
relation is used to calculate the average chord:
c(r) =
¸
b
2
_
1 −
r
2
c
2
_
, (4.9)
which is obtained by rewriting equation (4.7). Evaluation of equation (4.8) and
(4.9) leads to an average chord length of c = π/4 for c = 1.0, both defining the
planform. Figure 4.2 shows the wing planform and corresponding parameters for
a fruit fly wing and a ellipsoidal model wing.
Radius of gyration
In order to define a sound framework of comparison for different three-dimensional
and two-dimensional simulations, it is necessary to define all reference parameters,
80 Physical and numerical modelling of flapping foils and wings
O
z
y
R
g
S
r
R
g
R
root
b
s
R
tip
Axis of rotation
c(r)
dr
(a) Fruit fly shape
O
z
y
R
g
S
r
R
g
R
root
b
s
R
tip
Axis of rotation
c(r)
dr
(b) Ellipsoidal shape
Figure 4.2 Model wing geometry and planform. Schematic illustration of the geometry and
planform of a fruit fly (a) and ellipsoidal model wing (b). The planform is defined by the chord
variation c(r), the single wing span b
s
and the planform surface area S. The radius of gyration
R
g
is used to define a sound framework for comparison between mutual three-dimensional and two-
dimensional simulations. In three-dimensional simulations the wing revolves around the origin O, of
which the location is varied to study the effect of the angular accelerations.
introduced in section 4.2, at a representative cross-sectional area of the wing. As
the local velocity of each cross-section varies during flapping, the spanwise refer-
ence location is chosen to be at the radius of gyration. According to Ellington
(1984), this is the location where the resulting lift acts. Besides the average chord
length c and the single wing span b
s
the radius of gyration is another impor-
tant geometric parameter, especially when the comparison of different kinematic
models is concerned (Bos et al., 2008). The radius of gyration, R
g
is defined as
the weighted second moment of inertia (Luo & Sun, 2005, Lentink, 2008) and is
calculated as
R
g
=
¸
1
S
_
R
0
r
2
c(r)dr. (4.10)
Here S is the wing planform, r the spanwise coordinate, R the distance from the
rotation origin to the wing tip and c(r) represents the chord distribution along
the wing. Additionally, Luo & Sun (2005) compared the flow induced by flapping
wings with different aspect ratios and found that the radius of gyration provided
a reliable framework for force comparison when the flapping velocity is varied.
4.3.2 Kinematic modelling
Besides the numerical interest in the development and improvement of mesh mo-
tion techniques, described in chapter 3, the purpose of the present research is also
to investigate the three-dimensional flow around flapping wings at low Reynolds
numbers, the scale at which insects operate. Previous two-dimensional studies
4.3 Wing shape and kinematic modelling 81
X
Y
Z
θ(t)
φ(t)
α(t)
O
R
root
R
tip
start downstroke
start upstroke
mid-stroke
Stroke plane
Figure 4.3 Schematic illustration of the governing flapping angles. Flapping wing motion
is governed by three angles, φ(t) corresponds to the stroke variation, α(t) to the geometrical angle of
attack and deviation θ(t), which is measured in a plane normal to the stroke plane.
showed the tight relationship between the aerodynamic forces or performance and
the kinematic model (Wang et al., 2004, Bos et al., 2008, Thaweewat et al., 2009).
Since an accurate three-dimensional numerical method has been developed to solve
for the flow around flapping wings, the effect of different kinematic models on the
aerodynamic performance can be evaluated. Besides the use of an idealisation of
the insect wing planform, simplified kinematics was used, like harmonic motion,
to study different parameters independently, e.g. the centre of rotation (which is
equivalent to the Rossby number Ro) or the angle of attack amplitude.
The kinematic wing motion is defined by the variation of three independent
attitude angles, see figure 4.3. In this three-dimensional model the three degrees
of freedom of the wing motion are defined as the flapping angle, φ(t), in the mean
stroke plane, the angle of attack, α(t), with respect to the horizontal plane and
the deviation θ(t), which is measured in a plane normal to the stroke plane, as is
shown in figure 4.3. The deviation may be used to create a ‘figure-of-eight’ pattern
which is present in realistic fruit fly kinematics (Fry et al., 2003).
In literature, different kinematic models have been studied, from purely har-
monic motion to complex realistic fruit fly kinematics. Using two-dimensional
numerical techniques, Bos et al. (2008) showed that the flapping wing perfor-
mance may be influenced by the specific features of a particular kinematic model.
Figure 4.4 shows the least and the most complex of the kinematics models, which
are considered. Figure 4.4 illustrates (a) the harmonic model (Wang et al., 2004)
compared to (b) the realistic fruit fly kinematics (Fry et al., 2003). Figure 4.4(b)
shows that the realistic fruit fly kinematics is characterised by an asymmetry in
flapping angle, and angle of attack. Furthermore, the angle of attack shows a dip
(at t=0.1T), lowering the effective angle of attack. In addition, the shape of the
fruit fly angle of attack clearly shows a plateau of constant value, which is ex-
82 Physical and numerical modelling of flapping foils and wings
ploited by Sane & Dickinson (2001, 2002), Lehmann et al. (2005) using a Robofly.
Additionally, the realistic fruit fly kinematics in characterised by the presence of
deviation which may result in a ‘figure-of-eight’ pattern (Shyy et al., 2008b)
In order to illustrate the variation of the motion angles, the definition of the
harmonic model is described accordingly. The flapping angle, φ(t), is described by
a cosine function. The geometric angle of attack, α(t), is defined by a sine function
with respect to 90

and the deviation angle, θ(t), is given by a pure sine:
φ(t) = A
φ
· cos(2πft),
α(t) =
π
2
−A
α
· sin(2πft), (4.11)
θ(t) = A
θ
· sin(2πft).
Here, A
φ
is the flapping amplitude, which is defined from stroke reversal to mid-
stroke. Remind that this amplitude is half times the value used in literature (e.g.
Φ
0
in Ellington, 1984). f is the flapping frequency and A
α
represents the amplitude
of the angle of attack with respect to π/2, which is the initial position under
hovering conditions. A
θ
is the amplitude of the deviation angle which causes
the so-called ‘figure-of-eight’ pattern and is sinusoidal shaped (Fry et al., 2003).
The corresponding angular velocities are found by taking the time-derivatives of
equation (4.11).
˙
φ(t) = −A
φ
· (2πf) · sin(2πft),
˙ α(t) = −A
α
· (2πf) · cos(2πft), (4.12)
˙
θ(t) = A
θ
· (2πf) · cos(2πft).
In general, this harmonic model is crude but fairly reasonable representation of
the flapping motion of a fruit fly. Also, simplified (harmonic) kinematics may
be interesting for Micro Air Vehicles (MAV) implementation. As will be shown in
chapter 5 (Bos et al., 2008), the flapping wing performance may be significantly in-
fluenced by modifications of the basic kinematics, such as the previously described
plateau in angle of attack, modelled by a ‘trapezoidal’ shape and the presence of
deviation (Birch & Dickinson, 2003, Fry et al., 2003, Lehmann et al., 2005, Lu
& Shen, 2008, Bos et al., 2008). The ‘trapezoidal’ shaped angle of attack is im-
plemented using a piece-wise continuous function and the deviation by defining a
non-zero harmonically variation of θ(t).
4.3.3 Modelling of active wing flexing
In addition to the rigid body motion as shown in figure 4.3 it is possible to define
an extra displacement concerning flexing of the wing. Since a full fluid structure
interaction (FSI) simulation is too expensive and beyond the scope of the current
research, a flexing displacement of the wing surface is defined. The flexing dis-
placement is defined with respect to the initial wing position and can be written
4.3 Wing shape and kinematic modelling 83
t/T [-]
φ
,
α
,
θ
[

]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-80
-60
-40
-20
0
20
40
60
80
100
(a) Harmonic kinematics
t/T [-]
φ
,
α
,
θ
[

]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-80
-60
-40
-20
0
20
40
60
80
100
(b) Fruit fly kinematics
Figure 4.4 Comparison of the harmonic and fruit fly kinematic models. (a) shows the
least complex kinematic model, representing pure harmonic variations of the flapping angle φ(t) (•),
angle of attack α(t) (◦) and deviation θ(t) (). The variation of realistic fruit fly kinematics (Fry
et al., 2003) is shown in (b), which is the most complex kinematics available. The realistic fruit fly
kinematics is characterised by an asymmetric variation of flapping angle, an extra ‘bump’ and a degree
of ‘trapezoidal’ in angle of attack and the presence of deviation.
as
x(t) = A
f
· cos
_
2πx
0
ǫ
f
_
· sin(2πft),
where the first cosine function defines the wing shape and the sine function rep-
resents the time-variation. A
f
is the flexing amplitude vector and x
0
the location
of the initial boundary points. The cosine shaped wing flexing is defined by ǫ
f
,
which corresponds to the cosine ratio, i.e. ǫ
f
= 0.5 means that the shape is like a
half cosine function. Figure 4.5 shows a plunging airfoil incorporating flexing for
ǫ
f
= 0.25 (a) and ǫ
f
= 0.5 (b).
4.3.4 Numerical implementation of the wing kinematics
In previous sections the physical kinematic modelling was described. The current
section deals with the numerical implementation of this particular wing kinematics.
In general, the wing kinematics can be decomposed into a translation, a rotation
and a flexing (deformation) component. The translational component is not used
in the current three-dimensional simulations, which is limited to hovering and for-
ward flow conditions with stationary position of the rotation origin. However, in
the two-dimensional simulations, the flapping motion is defined by a translation in
combination with one rotation angle, the angle of attack. In addition to the trans-
lation and rotation, a limited number of two- and three-dimensional simulations
using a pre-defined wing flexing have been performed.
In general, the wing kinematics is calculated beforehand and applied to the
numerical flow solver. At every time-step the location of the boundary points is
determined leading to three distinct displacement arrays, due to translation, rota-
84 Physical and numerical modelling of flapping foils and wings
(a) Quarter cosine shape,ǫ
f
= 0.25 (b) Half cosine shape, ǫ
f
= 0.5
Figure 4.5 Illustration of a flexing two-dimensional airfoil. The two-dimensional flexing airfoil
is modelled by a time-varying cosine shape. This airfoil shape is either a quarter cosine, ǫ
f
= 0.25
(a) or a half cosine, ǫ
f
= 0.5 (b). Both figures (a) and (b) show the upstroke (left) and downstroke
(right).
tion and flexing.
Translation
The displacement array of the boundary points, which is due to translation is
obtained from
x(t) = A
t
· sin(2πf
t
t),
where x(t) = (x(t), y(t), z(t)). A
t
and f
t
are respectively the translation ampli-
tude and frequency vectors. The sine function in this definition is used when the
wing needs to move according to an ordinary rigid body motion (Ginsberg, 1998).
When a flapping motion is desired, a cosine function is used to define the motion
in the inertial reference frame.
Rotation
The second boundary point displacement array, due to rotation, is calculated at
subsequent (old and new) time-steps with respect to the initial mesh:
x
old
= R
old
· x
0
, (4.13)
and
x
new
= R
new
· x
0
, (4.14)
such that:
∆x
rot
= x
new
−x
old
.
Here ∆x
rot
is the boundary displacement due to rotation. R
old
and R
new
are the
rotation transformation matrices at respectively the old and new time instances.
4.3 Wing shape and kinematic modelling 85
The initial boundary points are given by x
0
. These rotation transformation ma-
trices consist of three different components, due to a rotation around the X-, Y -
and Z-axis. According to (Ginsberg, 1998, Baruh, 1999), these three matrices are
defined as
R
X
(t) =
_
_
1 0 0
0 cos(θ(t)) −sin(θ(t))
0 sin(θ(t)) cos(θ(t))
_
_
, (4.15)
R
Y
(t) =
_
_
cos(φ(t)) 0 −sin(φ(t))
0 1 0
sin(φ(t)) 0 cos(φ(t))
_
_
, (4.16)
and
R
Z
(t) =
_
_
cos(α(t)) −sin(α(t)) 0
sin(α(t)) cos(α(t)) 0
0 0 1
_
_
. (4.17)
Here, the rotation around the X-, Y - or Z-axis correspond to the deviation angle,
θ(t), flapping angle, φ(t), and the angle of attack, α(t), respectively. Using the
motion convention, shown in figure 4.3, it is clear that a hovering wing flaps in the
X-Z plane, accelerating the fluid from top to bottom. After a general sequence of
matrix multiplications, the total rotation matrix is obtained by combining equa-
tions (4.15) to (4.17) as
R
rot
(t) = R
X
(t) · R
Y
(t) · R
Z
(t), (4.18)
which is substituted into equation (4.13) and (4.14) to find the displacement of
the boundary points at the old and new times, due to rotation:
∆x
rot
= [R
rot
(t
new
) −R
rot
(t
old
)](x
0
−r
0
), (4.19)
where r
0
is the direction vector of the initial rotation origin, which is varied sys-
tematically in chapter 7.
Flexing
In addition to the translation and rotation, the flapping wing is able to perform a
flexing motion as well. This wing flexing is defined as
x
flex
(t) = A
f
· cos(2πx
0
) · sin(2πf
f
t),
where A
f
is the flexing amplitude vector, f
f
the flexing frequency and x
0
repre-
sents the initial boundary points of the flapping wing at t = 0. This definition
permits flexing in each orientation of a three-dimensional wing, i.e. spanwise and
chordwise.
When combining the previous results for the displacements due to translation,
rotation and flexing, the following is obtained:
∆x
b
= ∆x
trans
+ ∆x
rot
+ ∆x
flex
.
86 Physical and numerical modelling of flapping foils and wings
This total displacement field for the boundary points can be applied to the mesh
motion solver, available in most commercial and non-commercial CFD codes.
Numerical initialisation
If a flapping wing is considered, it is desirable to start the numerical simulation at
maximal flapping angle, in order to minimise the initial acceleration of the mesh.
Therefore, it was chosen to use a rigid body motion until the maximal flapping
angle was reached, in general this occurs at T/4, where T is the flapping period.
This approach has two major advantages. First, the mesh deformation is symmet-
ric, yielding high mesh quality at the extreme wing positions. Secondly, the initial
velocity is as small as possible, yielding good conditions for numerical convergence.
4.4 Dynamical scaling of flapping wings
In section 4.2 different dimensionless numbers, concerning rotational motion, were
identified in order to scale the Navier-Stokes equations, governing fluid flow. To de-
fine these rotational dimensional numbers, C
ang
= U
2
ref
/
˙
ΩR c, C
cen
= U
2
ref
/Ω
2
R c
and Ro = U
ref
/Ω c, it is important to identify the average rotation amplitude, Ω
and its time-derivative
˙
Ω for the particular kinematics implemented:
Ω =
1
T
_
T
0
|
˙
φ|dt =
1
T
_
T
0
| −A
φ
· (2πf) · sin(2πft)|dt = 4A
φ
f, (4.20)
˙
Ω =
1
T
_
T
0
|
¨
φ|dt =
1
T
_
T
0
| −A
φ
· (2πf)
2
· cos(2πft)|dt = 8πf
2
A
φ
= 2πfΩ, (4.21)
where the flapping velocity
˙
φ was taken from equation (4.12). As shown, evaluation
gives Ω = 4A
φ
f and
˙
Ω = 2πfΩ. Besides the angular velocity Ω, angular accelera-
tion
˙
Ω and average chord length c it is necessary to define an appropriate reference
velocity. As previously described and in accordance with (Bos et al., 2008, Lentink,
2008, Luo & Sun, 2005), the reference cross-section of the three-dimensional wing
is positioned at the radius of gyration, R
g
, and the reference velocity is calculated
at that particular location. This time-averaged velocity follows from:
U
ref
=
1
T
_
T
0
|u
R
g
|dt =
1
T
_
T
0
_
u
2
(t) +v
2
(t) +w
2
(t)dt, (4.22)
where |u
R
g
(t)| is the absolute velocity at R
g
which can be decomposed into three
components u(t), v(t) and w(t) in respectively X-, Y - and Z-direction. Using
equation (4.22) it is straightforward to find U
ref
by multiplying the expression for
Ω (4.20) by R
g
:
U
ref
= 4A
φ
fR
g
. (4.23)
These relations hold if the wing is the only driving force behind the resulting flow
velocity, but in case of forward flight conditions, there is an additional free-stream
4.4 Dynamical scaling of flapping wings 87
λ

90

−β
A
α
tan
−1
(2Stsinβ)
2A

sinβ
Figure 4.6 Schematic illustration of the kinematic parameters in forward flight.
velocity, U

. Lentink & Gerritsma (2003) showed that the following relation can
be used as a good approximation for both hovering (U

= 0) and forward flight
conditions:
U
ref
= U
R
g

_
U
2

+ (4A
φ
fR
g
)
2
. (4.24)
In forward flight another important parameter is the advance ratio (Shyy et al.,
2008b), J, which describes the forward speed with respect to the flapping velocity
at a certain radius, R:
J =
U

4A
φ
fR
g
=
λ

4A

,
where the reduced frequency (Ellington, 1984), λ

, is given by λ

= U

/fc, with f
the flapping frequency. According to (Shyy et al., 2008b, Thaweewat et al., 2009),
λ

is also known as the dimensionless wavelength, which is illustrated in figure 4.6.
The dimensionless amplitude at the radius of gyration R
g
is defined as
A

=
A
φ
R
g
c
,
which is a measure for the dimensionless translation of the selected cross-sectional
area. In order to create an appropriate framework for comparison, the dimension-
less amplitude at R
g
, A

is kept constant for all relevant simulations. Additionally,
a constant A

leads to similar wing-wake interactions (Birch & Dickinson, 2003)
for the representing simulations. For completeness and consistency, two other im-
portant parameters are kept constant as well, the Reynolds number at R
g
, which
is an implicit result of keeping R
g
constant and the area swept by the wing, A
swept
.
The Reynolds number for hovering conditions and using R
g
can be written:
Re =
U
ref
c
ν
=
4A
φ
fR
g
c
ν
.
The area that is swept by the revolving wing (Usherwood & Ellington, 2002),
A
swept
, is obtained by subtracting the area swept by the wing tip from the swept
88 Physical and numerical modelling of flapping foils and wings
area by the wing root:
A
swept
= 2A
φ
· (R
2
tip
−R
2
root
),
= 2A
φ
· (R
tip
−R
root
) · (R
tip
+R
root
),
= 2A
φ
· b
s
· (R
tip
+R
root
),
where R
tip
and R
root
are the radii at respectively the wing tip and root. The
distance from R
tip
to R
root
is identified as the single wing span b
s
. From (Lentink,
2008), it is deduced that keeping the swept area constant is similar to maintaining
a constant Froude efficiency (Stepniewski & Keys, 1984, Lentink, 2008). Fur-
thermore, the Rossby number, Ro, which is inversely proportional to the Coriolis
acceleration, needs to be obtained by rewriting equation (4.6) as
Ro =
U
ref
Ω c
=
R
g
c
. (4.25)
When the reference cross-sectional area is located at the radius of gyration (Elling-
ton, 1984), the Rossby number is given by Ro = R
g
/ c, the radius of gyration
divided by the average chord length. For a translating wing, the value for Ro is
infinite; for a rotating wing, Ro is finite. If Ro is varied, the effect of different
rotation origins can be investigated, from nearly translating to strongly revolving.
However, the determination of Ro would be easier if defined as Ro = R
tip
/ c,
where R
tip
is the wing tip radius, since those values are readily available from lit-
erature, in general Ro = 3.0 for insect and fish (Lentink, 2008), generating thrust
by moving the fluid. If the wing planform is complex, like in real insects, it may be
difficult to obtain the radius of gyration. Additionally, for R
root
= 0, the Rossby
number is equivalent to the single wing aspect ratio AR
s
= R
tip
/ c, a geometric
wing characteristic.
4.5 Computational domain and boundary
conditions
In this section the general approach for mesh generation is described, that has been
applied for the two- and three-dimensional flapping wing simulations. Chapter 5, 6
and 7, which are dealing with respectively two-dimensional hovering, forward and
three-dimensional hovering flight, explain the specific mesh generation, domain
size and boundary conditions for these simulations in more detail.
When performing a Computational Fluid Dynamics (CFD) study, a compu-
tational domain is necessary to contain the mesh on which the governing partial
differential equations are solved. It is necessary to use a sufficiently large compu-
tational domain to minimise disturbances, with appropriate boundary conditions
in order to obtain large convergence rates and the correct solution. When gener-
ating a grid around a two-dimensional thin and ellipse-shaped airfoil, it is efficient
4.6 Definition of force and performance coefficients 89
X
Y
Γ
left
Γ
wing
Γ
right
(a) Two-dimensional
Z X
Y
Γ
wing
Γ
left
Γ
right
Γ
bottom
Γ
top
Γ
front
Γ
back
(b) Three-dimensional
Figure 4.7 Computational domain and boundary conditions. The two-dimensional domain
uses the O-type topology (a), while the three-dimensional flapping wing simulations uses the boxed
topology (b).
to use an O-type mesh, which is shown in figure 4.7(a). Using conformal map-
pings (Bos et al., 2008, Thaweewat et al., 2009), a high quality mesh is generated
between the ellipse-shaped wing boundary, Γ
wing
, and the cylinder-shaped outer
boundaries, Γ
left
and Γ
right
. The outer boundary is split into Γ
left
and Γ
right
in
order to be able to specify inflow and outflow boundary conditions, which are
necessary (Wesseling, 2001) when forward flight conditions are simulated. When
simulating hovering flight, the flapping wing still induces a significant amount of
downwash, leading to a small inflow and outflow at the boundaries of the compu-
tational domain. The meshes for the two-dimensional simulations were generated
using Gambit software, see appendix A.
Generation of a three-dimensional structured mesh around a wing is not an easy
task, which can be very cumbersome using manual procedures used in programs
like Gambit or Gridgen

. Therefore, an automated topology mesh generator is
used, GridPro

. Using GridPro

, it was possible to generate a high quality mesh
around a thin ellipsoidal wing in a three-dimensional box-shaped computational
domain, which is shown in figure 4.7(b). Depending on the flapping configurations,
the outer boundaries, Γ
left
, Γ
right
, Γ
bottom
, Γ
top
, Γ
front
, Γ
back
, are set to inflow,
outflow or symmetry planes.
4.6 Definition of force and performance
coefficients
Besides flow field analysis, using advanced visualisation techniques, the resulting
forces acting on the airfoils and wings are of primary importance to assess aero-
dynamic performance. In order to make a sound comparison of forces and perfor-
mance for mutual two- and three-dimensional simulations (chapters 5, 6 and 7) it
90 Physical and numerical modelling of flapping foils and wings
PSfrag
X
Y
Z
R
g
F
X
F
Y
F
Z
˙
φ
Figure 4.8 Forces on a general three-dimensional flapping wing. The forces are defined in the
three-dimensional inertial reference frame. Depending on the type of motion, hovering or forward flight,
two- or three-dimensional, the lift and drag are constructed from the forces in X-, Y - or Z-direction.
is important to properly define force and performance coefficients. In this section,
the general determination of the force coefficients is explained in combination with
performance characteristics.
Forces
In general three dimensions, a two-dimensional derivation is trivial, the definitions
of the forces, F
X
, F
Y
and F
Z
are shown in figure 4.8. The centre of the axes co-
incides with the origin of rotation of the three-dimensional wing. The total force
vector is integrated over the wing surface and contains a pressure and a viscous
contribution. These forces are calculated using the following expression:
F
tot
=
_
S
pdS −
_
S
µ
∂u
∂n
dS,
where F
tot
[N] is the total force vector, S [m] the wing surface and dS [m] represents
an infinitely small surface area element, p [N/m
2
] is the pressure and µ [Ns/m
2
]
the dynamic viscosity. The term ∂u/∂n is the gradient of the velocity vector with
respect to the normal vector to the wall, i.e. together with µ this forms the wall
shear stress.
Two different force definitions can be defined, depending on the type of three-
dimensional motion, hovering or forward flapping flight. In hovering conditions,
when the main flapping direction is around the Y -axis, as shown in figure 4.3, the
lift force F
lift
is defined in the vertical direction and equal to F
Y
. The drag force
F
drag
, however, is defined in opposite direction of the flapping wing motion. If
three-dimensional flapping in hovering flight without deviation is considered, the
following lift and drag variations are found:
F
lift
(t) = F
Y
(t),
F
drag
(t) = F
X
(t) · sin(φ(t)) −F
Z
(t) · cos(φ(t)),
where φ(t) is the flapping angle. If the motion includes a deviation velocity, such
that the wing is not moving in the horizontal plane, the drag force derivation
is more elaborate but similar. Besides the lift and drag, there is a force in the
4.6 Definition of force and performance coefficients 91
direction of the spanwise coordinate. This spanwise force is dominated by the
viscous wall shear stress and therefore small compared to the lift and drag forces.
It is justified to neglect this spanwise force, also because it has little relevance to
performance.
In forward flight conditions, the same inertial reference frame is used as shown
in figure 4.3. The main flapping direction is still around the Y -axis, but the
direction of the uniform flow is from top to bottom in direction of the negative
Y -axis. The complete system needs to be rotated around the Z-axis in order to get
a horizontal orientation of the free-stream. In the case of forward flapping flight
the lift force is defined in the positive X-direction and the drag in the negative
Y -direction, opposite to the free-stream velocity:
F
lift
(t) = F
X
(t),
F
drag
(t) = −F
Y
(t).
Commonly the forces are made dimensionless using the dynamic pressure based
on the average velocity. With the strong variation in velocity, however, it is deemed
more appropriate to scale the forces with the mean dynamic pressure itself (Bos
et al., 2008). Hence, the forces are defined as:
C
D
=
F
drag
q · S
, C
L
=
F
lift
q · S
,
where C
D
and C
L
are the drag and lift coefficients and S the wing surface. The
mean dynamic pressure q is defined as:
q = 1/2ρU
2
ref
= 1/2ρ ·
1
T
_
T
0
_
(U

+U
flap
(t))
2
+ (U
dev
(t))
2
dt,
where the integration is evaluated over one flapping cycle with period T [s]. The
reference velocity contains the free-stream velocity U

, which is non-zero in for-
ward flapping flight, the flapping velocity U
flap
and the deviation velocity U
dev
,
perpendicular to the horizontal plane (in hovering flight).
Performance
The force coefficients are the major parameters used to assess the influence of the
different wing motion models. In addition, the ratio between time-averaged lift
coefficient C
L
and time-averaged drag coefficient C
D
is used to characterise per-
formance. These force averages are obtained by integration of C
L
and C
D
. The
lift is averaged over the complete flapping period, while for the drag the absolute
values are used for averaging. The drag is opposed to the flapping motion, such
that the sign flips at stroke reversal.
The average lift-to-drag ratio, C
L
/C
Dave
is chosen as an indicator of aero-
dynamic performance, also known as the glide number in aerospace engineering.
When the average lift coefficients of the different kinematic models are matched,
92 Physical and numerical modelling of flapping foils and wings
the lift-to-drag ratio is corrected for any differences in lift. Therefore, a high lift-to-
drag ratio effectively means low drag at equal lift. Additionally, the power factor,
C
L
3/2
/C
D
, (see Ruijgrok, 1994), has been used to assess the required power for a
certain amount of lift (Wang, 2008).
4.7 Conclusions
This chapter dealt with the physical and numerical modelling of three-dimensional
flapping wings and two-dimensional flapping airfoils. Two important dimensionless
numbers were identified, the Reynolds and Strouhal number. These numbers are
common in general fluid flow, but for flapping flight the definition has been slightly
changed. In order to analyse the flow around flapping wing, the governing Navier-
Stokes equations are written in a rotating reference frame. This leads to an extra
important dimensionless number, related to the wing rotation, namely the Rossby
number. The Rossby number is a way to describe the radius of curvature and thus
the angular accelerations in dimensionless terms.
In order to systematically study the aerodynamics around flapping wings at
the scale of insects, a model wing planform has been defined with an ellipsoidal
planform. Using that planform, the radius of gyration can be easily obtained,
which is used to dynamically scale the wing kinematics. Different flapping wing
kinematic models are briefly addressed, from realistic fruit fly kinematics to a
fully harmonic model. In addition to the rigid body rotations, a deformation of
the wing is defined in order to study the effects of wing flexing. This flexing motion
is defined with respect to the initial wing position and has a time-varying cosine
shape.
Most importantly, to design a sound framework for comparison it is necessary
to dynamically scale the wing kinematics for all numerical simulations. This is
achieved by scaling the motion parameters such that the dimensionless amplitude,
the average Reynolds number and the area swept by the wing result in comparable
values. The radius of gyration is used as a reference cross-section for both two-
dimensional flapping foil as well as three-dimensional flapping wing simulations.
Analysis of the vortical flow around the wings and foils is primarily performed
by plotting the force coefficient. The lift is defined in vertical direction, while
the drag is opposed to the flapping velocity. The corresponding coefficients are
obtained by averaging the dynamic pressure, which has proved to be proper refer-
ence. In addition to the forces, the lift-to-drag ratio is used to assess the flapping
wing performance.
CHAPTER 5
A 2D investigation of the influence
of wing kinematics in hovering
flight
J. Fluid Mech. (2008), vol. 594, pp. 341-368.
The influence of different wing kinematic models on the aerodynamic performance
of a hovering insect is investigated by means of two-dimensional time-dependent
Navier-Stokes simulations. For this, simplified models are compared with averaged
representations of the hovering fruit fly wing kinematics. With increasing com-
plexity, a harmonic model, a Robofly model and two more realistic fruit fly models
are considered, all dynamically scaled to Re = 110. To facilitate the comparison,
the parameters of the models were selected such that their mean quasi-steady lift
coefficient were matched. Details of the vortex dynamics, as well as the resulting
lift and drag forces were studied. The simulation results reveal that the fruit fly
wing kinematics result in forces that differ significantly from those resulting from
the simplified wing kinematic models. In addition, light is shed on the effect of
different specific characteristic features of the insect wing motion. The angle of
attack variation used by fruit flies increases aerodynamic performance, whereas
the deviation is most likely used for levelling the forces over the cycle.
94 Influence of wing kinematics in two-dimensional hovering flight
5.1 Introduction
In order to investigate the full flow around a three-dimensional flapping wing,
two-dimensional simulations are performed to get insight in the complicated flow
structures. This chapter deals with the evolution of the forces and the wake
originated by a flapping foil in hovering conditions.
5.1.1 Similarity and discrepancy between two- and
three-dimensional flows
In a recent paper Wang et al. (2004) compared three-dimensional Robofly results
with two-dimensional numerical results. This showed that two-dimensional simu-
lations are useful to obtain a better understanding of the flow features, which can
then be investigated more thoroughly in three dimensions.
Both Dong et al. (2005) and Blondeaux et al. (2005b) concluded that two-di-
mensional studies overpredict forces and performances since the energy-loss, which
is present in three dimensions, is resolved. Dong et al. (2005) and Blondeaux et
al. (2005b) numerically investigated the wake structure behind finite-span wings
at low Reynolds numbers. They observed that the flapping wings with low aspect
ratio generates three-dimensional vortical structures as was mentioned by Lighthill
(1969).
Notwithstanding the possible discrepancy between two-dimensional and three-
dimensional flow, two-dimensional analysis has often been applied to obtain insight
into the aerodynamic effects of choices in kinematics, airfoil cross-section, Reynolds
numbers, etc. Wang et al. (2004) confirmed that the similarities between two- and
three-dimensional approaches are sufficient to warrant that a reasonable approxi-
mation of insect flight can be obtained using a two-dimensional approach. First, in
case of advanced and symmetric rotation the forces were found to be similar in the
two-dimensional simulations compared to the three-dimensional experiments. Sec-
ondly it was observed that in both simulations and experiments the leading-edge
vortex did not fully separate for amplitude-to-chord ratios between 3-5 (Dickinson
& G¨otz, 1993, Dickinson, 1994), a similar amplitude range was used in the present
research.
In view of the excessive computational expense required for accurate three-
dimensional simulations, and with the above justification, the present study was
restricted to two-dimensional simulations. In a two-dimensional simulation our
mesh resolution can be higher compared to a three-dimensional simulation, in
view of the limitation of computational resources.
5.1.2 Influence of kinematic modelling
The relevance of (experimental or numerical) simulations of insect flight has been
found to depend on how reliable true insect wing kinematics are reproduced. Wang
et al. (2004) and Sane & Dickinson (2001) showed that the kinematic modelling
5.1 Introduction 95
significantly influences the mean force coefficients and its distribution. Addition-
ally, Hover et al. (2004) showed that modelling the angle of attack influences the
flapping foil propulsion efficiency to a large extent. This illustrates the appreciable
effects which details of the wing kinematics, like parameter values and stroke pat-
terns, may have on flight performance. It further emphasises the need to critically
assess the influence of kinematic model simplifications.
In literature, different kinematic models have been employed to investigate the
aerodynamic features of insect flight. For example, Wang (2000a,b) and Lentink
& Gerritsma (2003) numerically investigated pure harmonic translational motion
with respectively small and large amplitudes. Wang (2000a,b) varied flapping am-
plitude and frequency and showed that at a certain parameter selection the lift is
clearly enhanced. Lewin & Haj-Hariri (2003) performed a similar numerical study
for heaving airfoils. Besides lift enhancement at certain reduced frequencies, they
found periodic and aperiodic flow solutions which are strongly related to the aero-
dynamic efficiency. Lentink & Gerritsma (2003) varied airfoil shape with amplitude
and frequency fixed at values representative to real fruit flies. They concluded that
the airfoil choice is of minor influence, but large amplitudes lead to an increase of
lift by a factor of 5 compared to static forces generated by translating airfoils. It
was also shown that wing stroke models with only translational motion could not
provide for realistic results, such that including rotation is essential. In addition
to the harmonic models with pure translation (Dickinson & G¨otz, 1993), rota-
tional parameters were investigated by Dickinson (1994). They varied rotational
parameters and showed that axis-of-rotation, rotation speed and angle of attack
during translation are of great importance of the force development during each
stroke. Harmonic wing kinematics, including wing rotation, were used by Pedro
et al. (2003) and Guglielmini & Blondeaux (2004) in their numerical models to
solve for forward flight. Both studies emphasised the importance of angle of at-
tack modelling to influence the propulsive efficiency. Slightly more complex fruit
fly kinematic models were used by Dickinson et al. (1999) and Sane & Dickinson
(2001) with their Robofly. Based on observation of true insect flight, the wing
maintains a constant velocity and angle of attack during most of the stroke, with
a relatively strong linear and angular acceleration during stroke reversal. This
results in the typical ‘sawtooth’ displacement and ‘trapezoidal’ angle of attack
pattern of the Robofly kinematic model. Using these models, the effect of ampli-
tude, deviation, angle of attack and the timing of the latter were explored.
In the present study, different models from literature were considered, both the
pure harmonic and the Robofly model, in order to investigate their influence on the
aerodynamics. Furthermore, the results were compared with more realistic fruit
fly kinematics obtained from the observation of free flying fruit flies (Fry et al.,
2003). Instead of performing a parameter study within the scope of one kinematic
model, the objective of the present study is to compare the effect of the available
models as a whole. This leads to better insights in the consequences of simplifi-
cations in kinematic modelling, which is of great importance to both experiments
and numerical simulations. Also, it can reveal the importance of certain specific
96 Influence of wing kinematics in two-dimensional hovering flight
features of the stroke pattern, in relation to aerodynamic performance.
This study considers four different wing kinematic models with varying degree
of complexity. These models are implemented in a general-purpose Computational
Fluid Dynamics (CFD) code, which solves the Navier-Stokes equations under the
assumption of incompressible flow. In brief, the first model describes the wing
motion using basic harmonics as derived by Wang (2000a). The second model
contains the kinematics implemented by Dickinson et al. (1999) for their Robofly
at UC Berkeley (presently CalTech). The third model is a representation of the real
kinematics used by a hovering fruit fly (Drosophila Melanogaster), based on data
measured by Fry et al. (2003). Finally, the fourth model is a slightly simplified
version of the latter, observed fruit fly model. All these kinematic models are
dynamically scaled at a Reynolds number of Re = 110 which corresponds to
the flight conditions of the fruit fly. In addition, these kinematic models are
constructed such that their mean quasi-steady lift coefficients are comparable such
that our performance comparison is justified. This basis of comparison is verified
a-posteriori from the force results of the actual simulations.
The outline of this chapter is as follows. In section 5.2 the numerical simulation
methods are described. In addition, the actual modelling of the insect parameters
is discussed in 5.3. The results of the numerical simulations obtained with the
different kinematic models are treated in section 5.4 and concluding remarks are
given in 5.5.
5.2 Numerical simulation methods
The different kinematic models are implemented in the commercial flow solver
Fluent

, which solves the governing incompressible Navier-Stokes equations on
a two-dimensional computational mesh. The resulting model has been validated
using stationary and moving circular cylinders and verified using harmonically
moving wings.
5.2.1 Flow solver and governing equations
To simulate the flow around moving wings with pre-defined motions the commer-
cial CFD solver Fluent

was used. The two-dimensional time-dependent Navier-
Stokes equations are solved using the finite volume method, assuming incompress-
ible flow which is justified since the Mach number of flapping insect flight is typi-
cally O(10
−3
) (see Brodsky, 1994). The mass and momentum equations are solved
in a fixed inertial reference frame incorporating a moving mesh following the Ar-
bitrary Lagrangian Eulerian (ALE) formulation (Ferziger & Peric, 2002).
At the considered Reynolds number, Re = O(100), the flow is assumed to
be laminar. Henderson (1995) and Williamson (1995) showed that for circular
cylinders, the transition from laminar to turbulent flow occurs at Re = 180 ± 5,
which supports this assumption. Therefore, the transient incompressible laminar
5.2 Numerical simulation methods 97
Γ
3
Γ
2
Γ
1
x
y

2

1
x
y
0.4 0.42
0.56
0.58
Figure 5.1 O-type mesh topology with
boundary conditions on Γ
1

2
and Γ
3
.
Figure 5.2 Body conformal moving
mesh around a 2% ellipsoid airfoil.
Navier-Stokes equations (2.1) and (2.2) are used. Additional solver settings can
be found in (Bos et al., 2008, appendix B).
5.2.2 Mesh generation and boundary conditions
In order to compute the flow around the moving airfoils, an O-type computational
domain is used, which is shown schematically in figure 5.1. The computational
domain is divided into two parts: Ω
1
and Ω
2
for the inner and outer mesh respec-
tively. The body surface Γ
1
is located in the centre of the computational domain.
It has the reference length L which corresponds to the wing chord length. The
outer boundary Γ
3
is located at 25L such that the influence of the far field bound-
ary condition is negligible (Lentink & Gerritsma, 2003). At the body surface a
no-slip boundary condition is applied. Since the moving wing simulations concern
hovering insect flight, such that a free-stream is absent, a symmetry boundary
condition was applied at Γ
3
for numerical reasons. The influence of this symmetry
condition has been investigated and found to be sufficiently small.
For the wing, which is modelled as an ellipse of 2% thickness, generation of a
high quality mesh is not as straightforward as for a cylinder. The geometric surface
gradient is high, especially at the leading and trailing edges. This complicates the
creation of a high quality mesh, i.e. high cell orthogonality. In order to create
this body conformal mesh (see figure 5.2) a conformal mapping was applied (see
Wang, 2000b). The intermediate interface Γ
2
divides the mesh into two separate
fields, corresponding respectively to the inner conformal mesh (Ω
1
) and the outer
mesh (Ω
2
). The complete inner mesh moves according to the wing kinematics,
while re-meshing takes place in the outer field Ω
2
. Since re-meshing occurs at
a distance of 25 to 30 body lengths away from the wing, the flow around the
wing is not affected by the mesh regeneration. The described computational setup
was thoroughly validated using the flow around stationary and moving circular
98 Influence of wing kinematics in two-dimensional hovering flight
t +dt
t
α
ref
α
ref
+dα
y
x
Figure 5.3 Relative cell displacement in rotation.
cylinders (Bos et al., 2008, appendix C).
The airfoil simulations were performed on a mesh of 50000 cells with 2000
time-steps within one motion period. At this mesh, the size of the first cell at the
wing surface varies between 2% and 50% of the wing thickness at the leading-edge
and in the middle of the profile respectively. The grid resolution near the wing, up
to 1 chord length, was 8800 (176x50) cells such that the leading and trailing edge
vortices where captured with at least 1000 cells. One run, simulating 18 flapping
periods needed approximately 10 days on one serial AMD Athlon 2500+ CPU.
In order to minimise the interpolation errors from one time-step to the next
it is important to analyse the influence of the relative cell displacements. There-
fore, the motion of a reference cell was investigated, which is illustrated for the
rotational motion in figure 5.3. From the relative displacements in rotational and
translational direction follow the constraints for the size of the time-step in or-
der to keep the interpolation errors within limits. The relative displacements in
rotational and translational direction are defined as
ǫ
r
=
△α
α
ref
and
ǫ
y
=
△y
y
ref
=
2f
e
A
e
N△t
y
ref
.
Here α corresponds to the angular displacement of the reference cell, while α
ref
is the original radial length of this cell. The linear displacement of this cell is y
and y
ref
is the original length of this cell. Furthermore, f
e
, A
e
and N correspond
respectively to, the frequency, amplitude and number of cells on the surface.
In (Bos et al., 2008, appendix C), it was shown that a relative displacement of
10% in both rotational and translational direction leads to accurate results with
differences in drag coefficients remaining below 5%. The computational efforts are
acceptable: 2000 time-steps within one excitation period. Additionally, Bos et al.
(2008) (appendix D) investigated the mesh and time-step independence for the
nominal solver settings using harmonic wing kinematics for hovering flight.
5.3 Modelling insect wing kinematics 99
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-1
-0.5
0
0.5
1
1.5
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-1
-0.5
0
0.5
1
1.5
(b)
Figure 5.4 Comparison of force coefficients between the present simulations and Wang
et al. (2004). Comparison of lift (a) and drag (b) coefficients using harmonic wing kinematics with
A = 2.8, Re = 75 for the present study (•) and obtained by Wang et al. (2004) (◦).
5.2.3 Validation using harmonic wing kinematics
The main numerical parameters, a mesh size of 50000 cells and 2000 time-steps
within one excitation period, are used to validate our results with those obtained
by Wang et al. (2004) for similar but not entirely identical conditions. A two-
dimensional case was selected, with a moving wing according to harmonic kinemat-
ics. The amplitude was 2.8 times the chord length, which corresponds to Re = 75.
Figure 5.4 shows the lift and drag coefficients for validation purposes. Our forces
are normalised with the maximum of the quasi-steady force, just as in (Wang et
al., 2004). In similarity to (Wang et al., 2004) the drag in figure 5.4(a) is defined
to be positive in the direction opposite to the horizontal motion.
Generally, our force distribution looks similar for both cases. Only just after
stroke reversal our computation finds a larger lift and drag which is probably
the result of different numerical dissipation properties of both codes. The mean
lift and drag coefficients are 0.84, 1.47 for our simulation, compared to 0.82, 1.44
obtained by Wang et al. (2004), which is a difference of only 2% in lift and drag and
therefore, the computations were considered to be sufficiently accurate. Moreover,
within the context of comparing results of different stroke patterns, the present
numerical method is proved to be accurate.
Further details of the validation and verification studies can be found in (Bos
et al., 2008, appendix C and D).
5.3 Modelling insect wing kinematics
In order to derive the two-dimensional kinematic models the three-dimensional
degrees of freedom need to be converted to their two-dimensional counter-parts.
A common procedure is to define an equivalent two-dimensional geometry, while
100 Influence of wing kinematics in two-dimensional hovering flight
Figure 5.5 Illustration of the main motion directions. φ(t) corresponds to the stroke variation,
α(t) to the geometrical angle of attack and θ(t) to the deviation from the horizontal stroke plane.
From (Sane & Dickinson, 2001).
maintaining the characteristic aspects of the wing motion. This two-dimensional
set-up is derived in section 5.3.1 in terms of wing selection and model parameters.
The dynamical scaling and the force definitions are described respectively in 5.3.2
and 5.3.3.
5.3.1 Insect wing selection and model parameters
The computational approach is applied to investigate the influence of different
kinematic wing motion models on the aerodynamic performance. The different
kinematic models are illustrated using the Robofly experimental set-up, shown
in figure 5.5 (see Sane & Dickinson, 2001, Dickinson et al., 1999). In this three-
dimensional model the three degrees of freedom of the wing motion are defined
as the angular displacement φ in the mean stroke plane, the angle of attack α,
with respect to the horizontal plane and the deviation from the horizontal plane
θ, as is shown in figure 5.6. The deviation causes a ‘figure-of-eight’ pattern which
is present in real fruit fly kinematics (see Fry et al., 2003). The two-dimensional
airfoil shape is chosen to be a 2% thick ellipsoid. Lentink & Gerritsma (2003) found
this airfoil an acceptable choice to model insect wings at low Reynolds numbers,
Re = O(100). The two-dimensional projection is to be defined at a representative
spanwise location such that the motion is confined to an arc around the wing root.
Birch & Dickinson (2003) found strongest vorticity at a spanwise location of 0.65R
from the wing root, where R is the wing span. Therefore, Wang et al. (2004) used
this distance to derive their two-dimensional model.
In the present study, a different argument for the selection of the projection
5.3 Modelling insect wing kinematics 101
x
ac
x
cg
c
F
y
F
r
F
x
M
α
Figure 5.6 Force definition on the two-dimensional airfoil.
location (Lentink & Gerritsma, 2003) was used. Considering that the local velocity
of each cross-section varies during flapping, the spanwise location was selected to
be at the radius of gyration where the mean lift acts (Ellington, 1984).
In view of providing completeness on the three-dimensional set-up, the used
values are for the wing surface S=0.0167 m
2
the wing tip radius R=0.254 m,
the location of centre of gravity x
cg
=0.0882 m, the location of the wing base
x
base
=0.0667 m and the moment of inertia I
cg
=40.42 · 10
−4
m
4
. For the radius
of gyration the following value was obtained R
g
= 0.6396 · R. When comparing
this distance to the value used by (Wang et al., 2004) the current cross-section
is less than 2% closer to the wing root. Apparently, the mean lift acts nearly at
the location where the vorticity is maximal. Another important parameter to be
defined is the reference length, L
ref
, based on the mean chord length. A definition
of the mean chord length based on the moment of inertia around the wing root was
proposed. This leads to a value for the mean chord length of c = 0.082 m. Finally,
the conversion from three-dimensional angles to non-dimensional displacements is
given by:
x =
φ · R
g
c
, y =
θ · R
g
c
, (5.1)
where R
g
is the radius of gyration. Both the displacement x and the deviation
y have been made dimensionless with the mean chord c. The centre of rotation
is defined in the aerodynamic centre which lies at the quarter chord point of the
mean chord.
5.3.2 Dynamical scaling of the wing model
Since the flapping of the wings induces highly unsteady flow the relevant flow and
motion parameters have to be scaled dynamically. The period of the motion is
used to average the relevant flow velocity (Lentink & Gerritsma, 2003):
U =
1
T
_
T
0
_
u
2
+v
2
dt. (5.2)
102 Influence of wing kinematics in two-dimensional hovering flight
Here T [s] is the period , u represents the non-dimensional velocity in the stroke
plane and v the non-dimensional deviation velocity. Both are given by u = ∂x/∂t
and v = ∂y/∂t, where t = t/T is the dimensionless time.
Substituting equation (5.1) into (5.2) and evaluating, the following relations
for the Reynolds and Strouhal numbers were derived:
Re =
Uc
ν
=
fR
g
c
ν
·
_
1
0
_
(
∂φ
∂t
)
2
+ (
∂θ
∂t
)
2
(5.3)
and
St =
fc
U
=
c
R
g
·
1
_
1
0
_
(
∂φ
∂t
)
2
+ (
∂θ
∂t
)
2
. (5.4)
Here f = 1/T is the frequency, φ and θ the three-dimensional kinematic an-
gles for the displacement and deviation. From (5.3) and (5.4) it can be observed
that the Reynolds number Re depends solely on the frequency f for a given dis-
placement φ(t) and deviation θ(t). The Strouhal number St is not to be varied
independently. We fixed the Reynolds number to Re = 110.
5.3.3 Force and performance indicators
The definition of the drag and lift forces is shown in figure 5.6. The lift is equal
to the vertical force F
y
, while the drag is taken equal to the horizontal force
F
x
, defined positive in the positive x-direction. Commonly the forces are made
dimensionless using the dynamic pressure based on the average velocity. With the
strong variation in velocity, however, it is deemed more appropriate to scale the
forces with the mean dynamic pressure itself. Hence, the forces are defined as
C
D
=
F
x
q · c
, C
L
=
F
y
q · c
,
where C
D
and C
L
are the drag and lift coefficients. The mean dynamic pressure
q is defined as
q = 1/2ρU
2
= 1/2ρ ·
1
T
_
T
0
_
_
∂x
∂t
_
2
+
_
∂y
∂t
_
2
_
dt,
where the integration is evaluated over one flapping cycle. The force coefficients
are the major parameters used to assess the influence of the different wing motion
models. In addition, the ratio between time-averaged lift coefficient, C
L
, and time-
averaged drag coefficient, C
D
, is used to characterise performance. These force
averages are obtained by integration of C
L
and C
D
. The lift is averaged over the
complete period, while for the drag the averages are per half stroke. The average
lift-to-drag ratio, C
L
/C
Dave
is chosen as an indicator of aerodynamic performance,
also known as the glide number in aerospace engineering. Since the average lift
coefficients of the different kinematic models are matched, the lift-to-drag ratio is
corrected for any differences in lift. Therefore, a high lift-to-drag ratio effectively
means low drag at equal lift.
5.3 Modelling insect wing kinematics 103
5.3.4 Different wing kinematic models
Since the main purpose of this study is to investigate the influence of wing kinemat-
ics on the aerodynamic performance during hovering fruit fly flight, four different
kinematic models, with different degree of complexity, have been analysed. Two
of these models, the pure harmonic motion and the Robofly experimental kine-
matics have appeared in literature. The third model represents the actual fruit fly
kinematics as observed in experiments and the last one was a modification of the
latter, chosen to investigate the effect of symmetry in the wing motion.
In order to facilitate the comparison the model parameters are chosen based
on matching the mean quasi-steady lift coefficient (Bos et al., 2008, appendix
A). Although according to Sane & Dickinson (2001) the mean drag is strongly
influenced by the unsteady flow physics, which are not fully present in the quasi-
steady theory, the mean lift coefficient is predicted well using this theory. Using
quasi-steady theory, different kinematic models were constructed, such that their
quasi-steady lift coefficients are matched within 1%. For the symmetric models
this force is equal to the resultant force. In view of the limitations of the quasi-
steady theory, the difference between predicted and simulated values is expected
to exceed this 1% tolerance. However, in section 5.4 it is shown that the computed
mean lift coefficient of the numerical simulations are reasonably well matched for
all models, which provides an a-posteriori justification of our choices for the model
parameters.
The characteristic shapes of each model are described. Subsequently they are
used to investigate the influence of the models on the force histories and the per-
formance in section 5.4. Analysing those aspects leads to a better understanding
of how the fruit fly may benefit from kinematic features which are absent in the
simpler models, and reveals the relevance of including these aspects in theoretical
models.
The first of the four models is described by pure sine and cosine functions and
will therefore be referred to as the harmonic model (see Wang et al., 2004). The
displacement, angle of attack and deviation, are shown in figure 5.7(a). The second
model takes the wing kinematics as used in the Robofly model (Dickinson et al.,
1999). In figure 5.7(b) it is shown that the flip from down to upstroke is postponed
to the end of the translational phase which results in the ‘sawtooth’ shape of the
displacement. Large accelerations at stroke reversal are the result. The deviation
is zero, just as in the harmonic model. The third model, shown in figure 5.7(c),
is derived from measurements on real fruit flies (Fry et al., 2003) and is therefore
considered as the most realistic fruit fly kinematic model. This model does include
the deviation which results in a ‘figure-of-eight’ pattern. Neither the displacement,
angle of attack nor deviation is symmetric during the flapping period.
In order to investigate the fact that the observed fruit fly kinematics lacks an
exact symmetry in the wing stroke pattern, a symmetrical model was constructed,
referred to as the symmetric fruit fly model, displayed in figure 5.7(d). Within this
model the motion is identical for the downstroke and upstroke. Like the realistic
104 Influence of wing kinematics in two-dimensional hovering flight
t/T [-]
φ
,
α
,
θ
[

]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-80
-60
-40
-20
0
20
40
60
80
100
(a)
t/T [-]
φ
,
α
,
θ
[

]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-80
-60
-40
-20
0
20
40
60
80
100
(b)
t/T [-]
φ
,
α
,
θ
[

]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-80
-60
-40
-20
0
20
40
60
80
100
(c)
t/T [-]
φ
,
α
,
θ
[

]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-80
-60
-40
-20
0
20
40
60
80
100
(d)
Figure 5.7 Kinematic angles of the different kinematic models. (a) Harmonic model. (b)
Robofly model. (c) fruit fly model. (d) simplified fruit fly model. •: displacement angle φ, ◦: angle
of attack α, : deviation angle θ.
fruit fly model this symmetric model includes a time-dependent deviation such
that the observer sees a ‘figure-of-eight’ pattern of the wing. Neither of those
last two realistic kinematic models can be described by using simple analytical
functions without losing significant detail.
When comparing the motion parameters, φ, α and θ for each model it becomes
possible to identify certain important differences. The Robofly initially has a
larger gradient in time of the angle of attack compared to the harmonic case, see
figure 5.7(a) and (b). During translation from about t = 0.1T to t = 0.4T the
angle of attack flattens at a value of almost 40

. This ‘trapezoidal’ shape of α is
characteristic for the Robofly and may be influencing the performance. Although
the Robofly model clearly shows similarities with the fruit fly models the latter
has some typical additional features. The most obvious peculiarity of the realistic
fruit fly models is the extra ‘bump’ in angle of attack just after stroke reversal,
5.4 Results and Discussion 105
compared to the Robofly (figure 5.7(b) and (c)). It follows the same high angular
velocity, but instead of flattening α, the fruit fly wing α descends to the ‘bump’.
After the ‘bump’ the angle of attack more or less matches the plateau found in
Robofly but starts to increase earlier. During stroke reversal the gradient of α
matched the harmonic model closer than the Robofly with its high gradients.
The harmonic and Robofly models lack deviation, so no ‘figure-of-eight’ is
present. The deviation of the fruit fly model is asymmetric during the complete
cycle, but also during each half stroke (figure 5.7(c)). This is likely to influence
the performance since the effective angle of attack is altered due to deviation.
It is also observed that the deviation is negative for a certain period during the
upstroke. Therefore, the deviation of the realistic fruit fly is averaged to derive the
simplified fruit fly model, see figure 5.7(d). This last model is used to investigate
the influence of deviation on the force histories and performance.
5.4 Results and Discussion
In the previous section it was observed that the most interesting aspects of the
Robofly kinematic model are the ‘sawtooth’ displacement and the ‘trapezoidal’
angle of attack. This implies that strong translational and rotational accelerations
occur at stroke reversal. The more realistic fruit fly models are characterised
by a ‘bump’ in angle of attack and the presence of deviation. Results of two
comparative studies were presented. The first is an overall comparison of the
complete kinematic models, which is described in section 5.4.1. In the second
study the effect of the characteristic features identified above, are considered in
more in detail. In order to assess the effect of these kinematic features in isolation,
the comparison is made using the simplest model, the harmonic model, as baseline.
Hereto this baseline model is subsequently modified by adding respectively the
‘sawtooth’ displacement,‘trapezoidal’ angle of attack, extra ‘bump’ in angle of
attack and the presence of deviation. The results of this comparison, in terms of
actual vortex dynamics, as well as the resulting lift and drag histories are studied
in 5.4.2.
5.4.1 Overall model comparison
In table 5.1 the mean force coefficients are given for the four complete models,
the harmonic model, the Robofly model, the realistic fruit fly model and the
simplified fruit fly model. The mean drag, for each half-stroke, and lift coefficients
are given, as well as the average lift-to-drag ratio, which characterises aerodynamic
performance.
The differences of the obtained mean lift coefficient are significantly smaller
than the differences in lift-to-drag ratios. Therefore, the conclusions on the per-
formance comparison are considered to be significant.
The mean drag for the harmonic and Robofly models is substantially higher
106 Influence of wing kinematics in two-dimensional hovering flight
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-1
0
1
2
3
4
5
(a)
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-1
0
1
2
3
4
5
(b)
Figure 5.8 Lift coefficient histories of the baseline kinematic models. •: harmonic model,
◦: Robofly model, : realistic fruit fly model, ▽: simplified fruit fly model.
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-8
-6
-4
-2
0
2
4
6
8
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-8
-6
-4
-2
0
2
4
6
8
(b)
Figure 5.9 Drag coefficient histories of the baseline kinematic models. •: harmonic model;
◦: Robofly model, : realistic fruit fly model, ▽: simplified fruit fly model.
kinematic model C
L
C
Ddown
−C
Dup
C
L
/C
Dave
harmonic 1.483 −3.7% 1.848 1.839 0.805 −29%
Robofly 1.417 −8.0% 2.466 2.448 0.577 −49%
realistic fruit fly 1.540 baseline 1.387 1.335 1.132 baseline
simplified fruit fly 1.454 −5.6% 1.012 1.596 1.115 −1.5%
Table 5.1 Time-averaged force coefficients using the complete baseline models.
5.4 Results and Discussion 107
¯
C
L
= 1.483
x
c
[−]
-3 -2 -1 0 1 2 3
(a)
¯
C
L
= 1.417
x
c
[−]
-3 -2 -1 0 1 2 3
(b)
¯
C
L
= 1.54
x
c
[−]
-3 -2 -1 0 1 2 3
(c)
¯
C
L
= 1.454
x
c
[−]
-3 -2 -1 0 1 2 3
(d)
Figure 5.10 Force vectors during each half-stroke. (a) harmonic model, (b) Robofly model,
(c) realistic fruit fly model, (d) symmetric fruit fly model.
108 Influence of wing kinematics in two-dimensional hovering flight
(a) harmonic model (b) realistic fruit fly model
Figure 5.11 Vorticity contours around a flapping airfoil. Vorticity contours are shown for
t = 0.1T (blue: clock-wise, corresponding to negative vorticity values).
compared to the fruit fly models. This is also illustrated in figure 5.8 and 5.9 (lift
and drag histories) and figure 5.10 (force vectors). Figure 5.11 shows the vorticity
contours of the realistic fruit fly model compared with the harmonic model. It can
be seen in figure 5.7(a) that the effective angle of attack is higher in the harmonic
case, compared to the realistic fruit fly model, figure 5.7(c). Therefore, the mean
drag contribution of the leading-edge vortices (LEV) is higher. The decrease in
effective angle of attack in the realistic fruit fly model is also enlarged by the
presence of the ‘bump’. This drag increasing effect is even larger in case of the
Robofly model due to the ‘trapezoidal’ angle of attack. The ‘sawtooth’ shaped
Robofly displacement could possibly play an important role as is discussed in the
next section. The different kinematic patterns are also illustrated in figure 5.10,
which shows the resultant force vectors during a full stroke for those baseline
kinematic models.
Furthermore, the mean drag coefficient of the simplified fruit fly is not sym-
metric, i.e. the drag during the upstroke is about 57% higher than during the
downstroke, which is attributed to the complex vortex dynamics. Nevertheless,
the average value during a complete stroke matches the mean drag coefficient
obtained with the realistic fruit fly model.
When comparing the lift-to-drag ratios in table 5.1, it can be observed that
within the model assumptions, the fruit fly models perform better than the less
complex models. Compared to the harmonic model, the realistic fruit fly model
results in a significant decrease in drag of 29% at comparable lift. The difference
with the Robofly model is even larger, 49%. These performance increases are the
result of the lower drag coefficients in both fruit fly models due to certain beneficial
5.4 Results and Discussion 109
kinematic features. The current results provide insight into the effects of certain
specific kinematic features. However, one has to be cautious when extrapolating
these results to real flying flies since in reality not every flapping period displays
exactly the same kinematic profile. Next, the individual influences of the different
interesting kinematic shapes are studied.
5.4.2 Kinematic features investigation
Influence of ‘sawtooth’ displacement used by the Robofly
The ‘sawtooth’ shaped displacement of the Robofly is investigated in isolation
to assess its influence on the force histories and the aerodynamic performance.
Therefore, the purely harmonic model was appended with the Robofly displace-
ment and the results were compared with the ones obtained using the original
harmonic model. Figure 5.12(a) shows the force vectors acting on the wing during
the up and downstroke. In addition, the force histories during one full stroke are
shown in figure 5.13. From figure 5.13 it is observed that compared to the har-
monic model the global force histories look similar. Two force peaks are observed
close to t = 0.1T and t = 0.4T, respectively, which are repeated since the motion
is symmetric. The lift peaks are almost equal but the drag peaks are significantly
larger for the ‘sawtooth’ case, see figure 5.13(b). This also explains the larger
mean drag compared to the harmonic model which can be read from table 5.2.
In figures 5.14(a) and (b) the vorticity contours are plotted at t = 0.1T for the
harmonic model and the one with the appended ‘sawtooth’ shaped displacement.
From figure 5.14 it can be seen that the LEV is stronger for the ‘sawtooth’ case
which explains the higher drag peak. The stronger LEV at the beginning of
the downstroke in the ‘sawtooth’ case is most likely caused by the higher velocity
gradient. This leads to a larger shear layer to form a stronger vortex. On the other
hand, at the end of the half-stroke the wing decelerates faster in the ‘sawtooth’
case which results in a lower strength in the LEV. Since the wing orientation is
almost vertical, at t = 0.1T, the drag peak is larger than the lift peak.
The larger mean drag is reflected in the integrated values in table 5.2. Due
to this larger drag during each stroke, the ‘sawtooth’ shaped displacement leads
to a lower lift-to-drag ratio, which shows a decrease of 24.3% with respect to the
harmonic case.
Influence of ‘trapezoidal’ angle of attack used by the Robofly
In combination with the ‘sawtooth’ displacement, the Robofly uses a ‘trapezoidal’
shape for the angle of attack. In order to determine the effect of this shape the
harmonic model is extended by this ‘trapezoidal’ angle of attack. The results are
compared with those obtained with the original harmonic model, see figure 5.12(a)
for the force vectors. The lift and drag coefficients are plotted in figure 5.15. A
surprising and unexpected observation is the asymmetry in the periodic force dis-
tribution for the ‘trapezoidal’ angle of attack notwithstanding the symmetry of
the kinematics. This leads to the non-zero mean horizontal force along a complete
110 Influence of wing kinematics in two-dimensional hovering flight
¯
C
L
= 1.366
x
c
[−]
-3 -2 -1 0 1 2 3
(a)
¯
C
L
= 1.351
x
c
[−]
-3 -2 -1 0 1 2 3
(b)
¯
C
L
= 1.483
x
c
[−]
-3 -2 -1 0 1 2 3
(c)
¯
C
L
= 1.323
x
c
[−]
-3 -2 -1 0 1 2 3
(d)
Figure 5.12 Force vectors during each half-stroke. (a) harmonic model with ‘sawtooth’ φ, (b)
harmonic model with ‘trapezoidal’ α, (c) harmonic model with ‘extra bump’ α, (d) harmonic model
with ‘deviation’ θ.
5.4 Results and Discussion 111
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(b)
Figure 5.13 Lift and drag coefficients. Lift (a) and drag (b) histories to investigate the influence
of the ‘sawtooth’ displacement compared to the harmonic model. •: harmonic φ,α,θ; ◦: harmonic α,
θ and Robofly φ.
(a) harmonic model (b) harmonic model with ‘sawtooth’ displacement
Figure 5.14 Vorticity contours around a flapping airfoil. Vorticity contours are shown for
t=0.1T (blue: clock-wise, corresponding to negative vorticity values)
112 Influence of wing kinematics in two-dimensional hovering flight
kinematic model C
L
C
Ddown
−C
Dup
C
L
/C
Dave
harm. φ, α and θ 1.483 (baseline) 1.848 1.839 0.804 (baseline)
harm. α, θ + Robofly φ 1.366 (−7.9%) 2.240 2.250 0.608 (−24.3 %)
harm. φ, θ + Robofly α 1.351 (−8.9%) 2.302 2.733 0.537 (−33.3 %)
harm. φ, θ + fruit fly. α 1.483 (0.0%) 1.221 1.969 0.930 (+15.6 %)
harm. φ, α + fruit fly. θ 1.323 (−10.8%) 1.807 1.776 0.738 (−8.2 %)
Table 5.2 Time-averaged force coefficients to investigate the influence of kinematic shapes.
Each characteristic shape is varied with respect to the harmonic motion model.
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(b)
Figure 5.15 Lift and drag coefficients. Lift and drag histories to study the influence of the
‘trapezoidal’ angle of attack compared to harmonic model. •: harmonic φ,α,θ; ◦: harmonic α, φ and
Robofly α.
5.4 Results and Discussion 113
(a) harmonic model (b) harmonic model with ‘trapezoidal’ α
Figure 5.16 Vorticity contours around a flapping airfoil. Vorticity contours are shown for
t=0.6T (blue: clock-wise, corresponding to negative vorticity values)
stroke cycle. Although this model is symmetric, the force distributions are not,
since the complex vortex dynamics are non-linear and asymmetric.
From figure 5.15 it is clear that at the beginning of a stroke the lift peak of the
‘trapezoidal’ case is larger. Using figure 5.16 this is illustrated at the beginning of
the upstroke using vorticity contours. The LEV is larger in case of the ‘trapezoidal’
angle of attack. This can be explained as follows. In the ‘trapezoidal’ case the
wing reaches the maximum angle of attack earlier in the stroke, see figure 5.12(b).
Therefore, the angle of attack is larger at the early start of a stroke compared to
the harmonic model. Since large angle of attacks cause high velocity gradients
over the leading-edge, larger vortices occur in the beginning of a stroke.
Another interesting result is the low second peak in the lift, at the end of each
stroke, compared to the harmonic model. Taking a closer look at figure 5.17(b),
one observes stronger and more pronounced vortices in the wake of the ‘trape-
zoidal’ case. This could indicate a larger amount of vortex shedding during the
period when the angle of attack is nearly constant. This results therefore in a
lower second peak since the LEV has decreased in size and strength. Altogether,
the mean lift is slightly decreased whereas the mean drag is increased. This leads
to a significant performance decrease of 33.3% due to the ‘trapezoidal’ angle of
attack variation, see table 5.2.
Influence of extra ‘bump’ in angle of attack used by the fruit fly
The fruit fly models are subject to an extra ‘bump’ in angle of attack. To make
comparison plausible the symmetric ‘bump’ variation used in the simplified fruit
fly model is used to compare results with the harmonic model. Figure 5.12(c)
114 Influence of wing kinematics in two-dimensional hovering flight
(a) harmonic model (b) harmonic model with ‘trapezoidal’ α
Figure 5.17 Vorticity contours around a flapping airfoil. Vorticity contours are shown for
t=0.4T (blue: clock-wise, corresponding to negative vorticity values)
shows the force vectors during up and down stroke. In figure 5.18 the lift and
drag forces are shown for the harmonic model with and without the symmetric
‘bump’ in angle of attack. From table 5.2 it is seen that using this feature the
mean lift does not change significantly. However, the drag during the downstroke
is very much affected. A decrease of at least 30% in mean drag is found, compared
to the harmonic case. It is also noted that this case results in asymmetric force
distributions as was the case when using the ‘trapezoidal’ angle of attack. On the
other hand the drag is slightly increased during the upstroke such that the mean
lift-to-drag ratio is still increased with more than 15.6%. From figure 5.18 it is
observed that the extra ‘bump’ generates an extra lift peak at the beginning of the
downstroke. The change in angle of attack due to the extra ’bump’ is shown when
figure 5.19(a) and (b) are compared. The decrease in effective angle of attack as
a result of the ‘bump’ is considerable compared to the harmonic case. The same
was found for the Robofly case. Therefore, for the case with the ‘bump’ in angle
of attack, the LEV provides nearly complete lift since the wing orientation is ap-
proximately horizontal. This is also the main reason for the lower drag during the
downstroke.
Figure 5.20 shows the vorticity at the beginning of the upstroke at the time
of the ‘bump’. The LEV is larger compared to the case with the extra ‘bump’ in
angle of attack. This causes the loss in lift just after stroke reversal in case of the
‘bump’ angle of attack compared to the harmonic model.
Influence of wing deviation used by the fruit fly
The last important characteristic of the kinematics is the deviation, present in the
5.4 Results and Discussion 115
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(b)
Figure 5.18 Lift and drag coefficients. Lift (a) and drag (b) histories to study the influence of
the extra ‘bump’ in angle of attack. •: harmonic φ,α,θ; ◦: harmonic α, φ and fruit fly α.
(a) harmonic model (b) harmonic model with ‘extra bump’ in α
Figure 5.19 Vorticity contours around a flapping airfoil. Vorticity contours are shown for
t=0.1T (blue: clock-wise, corresponding to negative vorticity values)
116 Influence of wing kinematics in two-dimensional hovering flight
(a) harmonic model (b) harmonic model with ‘extra bump’ in α
Figure 5.20 Vorticity contours around a flapping airfoil. Vorticity contours are shown for
t=0.6T (blue: clock-wise, corresponding to negative vorticity values)
realistic and simplified fruit fly model. This deviation causes a ‘figure-of-eight’
pattern, described by the wing tip instead of wing motion solely in the stroke
plane. Since deviation could introduce a large velocity component perpendicular
to the stroke plane, the effective angle of attack is highly affected. This motion
perpendicular to the stroke plane is illustrated in figure 5.12(d) which also shows
the force vectors.
Figure 5.21 shows the force coefficients during one flapping period with de-
viation added to the harmonic model. The mean lift and drag are not strongly
influenced by the deviation, see table 5.2. The mean lift is decreased by 10.8% and
the mean drag is almost not affected by the presence of deviation, about 2%−4%
difference in both strokes. It is also revealed that the force distributions remain
symmetric.
The large influence of the deviation on the variation of the lift force is observed
at the start (t = 0.1T and t = 0.6T) and end (t = 0.4T and t = 0.9T) of each
stroke. Just after stroke reversal a lift peak occurs, which is higher compared to
the harmonic case. However, at the end of each stroke the harmonic lift peak was
decreased by the deviation. It appears that the force distribution is levelled or
balanced by the deviation.
The flow dynamic mechanism for this is shown in the vorticity visualisations
of figure 5.22 which shows the vorticity at the beginning of the stroke. Compared
to the harmonic model, the deviation causes a slightly stronger LEV at t = 0.1T.
The influence of the deviation is relatively large since the deviation increases the
effective angle of attack considerably just after stroke reversal. At the end of a
stroke the wings move up again which leads to a decrease in effective angle of
5.4 Results and Discussion 117
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(b)
Figure 5.21 Lift and drag coefficients. Lift (a) and drag (b) histories to study the influence of
the deviation compared to harmonic model. •: harmonic φ,α,θ; ◦: harmonic α, φ and fruit fly θ.
(a) harmonic model (b) harmonic model with deviation
Figure 5.22 Vorticity contours around a flapping airfoil. Vorticity contours are shown for
t=0.1T (blue: clock-wise, corresponding to negative vorticity values).
118 Influence of wing kinematics in two-dimensional hovering flight
(a) harmonic model (b) harmonic model with deviation
Figure 5.23 Vorticity contours around a flapping airfoil. Vorticity contours are shown for
t=0.6T (blue: clock-wise, corresponding to negative vorticity values).
attack. Figure 5.23(a) and (b) show LEV’s of comparable strength for both cases.
Summarising, the deviation is levelling the force distributions while the mean
lift and drag are almost unaffected. This leads to the suggestion that a fruit fly
may use the deviation to level the wing loading over a flapping cycle. Three-
dimensional studies are needed to investigate to what extent this effect is also
present in real insect flight.
5.5 Conclusions
The effect of wing motion kinematics on the aerodynamic characteristics of hov-
ering insect flight was investigated by means of two-dimensional numerical flow
simulations. The results of the present two-dimensional study may provide useful
insights in the understanding of real three-dimensional insect flight (Wang et al.,
2004).
Four different kinematic models, with different complexity, have been anal-
ysed using two-dimensional time-dependent Navier-Stokes simulations. Two of
these models, pure harmonic motion and Robofly experimental kinematics have
appeared in literature. The third model represents the actual fruit fly kinematics
as observed in experiments and the last one is a modification of the latter, chosen
to investigate the effect of symmetry. The most prominent aspects of the Robofly
kinematic model are the ‘sawtooth’ displacement and the ‘trapezoidal’ angle of
attack. The fruit fly models are characterised by a ‘bump’ in angle of attack and
the presence of deviation. To facilitate the comparison all models are dynami-
cally scaled at Re = 110 and constructed such that their mean quasi-steady lift
5.5 Conclusions 119
coefficient was matched.
It was found that the realistic fruit fly wing kinematics result in significantly
lower drag at similar lift compared with the simplified wing kinematic models
used in literature. The trend that the fruit fly kinematics increases aerodynamic
performance agrees well with the predictions of the quasi-steady theory, but the
numerical flow simulations provide a more complete quantitative analysis of the
flow behaviour. To investigate which aspects of the kinematic shapes are the most
important, they were compared to the harmonic model.
First, an overall comparison of the complete kinematic models was given. It
was shown that the difference in performance in terms of mean lift-to-drag ratio
between the different kinematic models was significant. The mean aerodynamic
drag at equal lift of the fruit fly models is about 49% lower compared to the Robofly
model and about 29% lower with respect to the harmonic model. Therefore, the
effect of the characteristic features has been studied. Hereto the harmonic model
was extended by respectively the ’sawtooth’ displacement, ‘trapezoidal’ angle of
attack, extra ‘bump’ in angle of attack and the presence of deviation. The actual
vortex dynamics, as well as the resulting lift and drag histories were studied.
The results showed that the ‘sawtooth’ amplitude used in the Robofly model
has a small effect on the mean lift but the mean drag is affected significantly. Due
to the high acceleration during stroke reversal of the ‘sawtooth’ shaped amplitude,
the mean drag at comparable lift is increased by 24.3%. The second model sim-
plification used by the Robofly, the ‘trapezoidal’ angle of attack, caused the LEV
to separate during the translational phase. This led to an increase in mean drag
during each half-stroke. Also in this case large accelerations at stroke reversal lead
to a decrease in lift-to-drag ratio of 33.3%.
The extra ‘bump’ in angle of attack, as used by the fruit fly model, is not
affecting the mean lift to a large extent. During the beginning of the up and
downstroke the ‘bump’ decreases the angle of attack such that the wing orientation
is almost horizontal. This leads to a significant decrease in drag which improves
aerodynamic performance in the sense of lift-to-drag ratio by 15.6%. The other
realistic kinematic feature is the deviation, which is found to have only a marginal
effect on the mean lift and mean drag. However, the effective angle of attack is
altered such that the deviation leads to levelling of the force distribution.
The results from the present study show that special features of insect flight
have an appreciable effect on the accuracy of performance models of insect flight.
In particular they indicate that kinematic features, found in fruit fly kinematics,
like the extra ‘bump’ in angle of attack and deviation, may lead to drag reduction
or force levelling compared to harmonic kinematics.
CHAPTER 6
Vortex wake interactions of a
two-dimensional forward flapping
foil
AIAA paper 2009-791.
A two-dimensional numerical investigation is performed to study the vortical flow
around a flapping foil that models an animal wing, fin, or tail in forward motion.
The vortex dynamics and performance are studied to determine the influence of
foil kinematics. The baseline kinematic model is prescribed by harmonic functions
which can be characterised by four variables, the dimensionless wavelength, the
dimensionless flapping amplitude, the amplitude of geometric angle of attack, and
the stroke plane angle. The foil motion kinematics has a strong influence on the
vortex dynamics, in particular on the vortex-wake pattern behind the foil which
can be either periodic or aperiodic. Both symmetric and asymmetric solutions
are found. Evidence was found that the attachment of a leading-edge vortex
(LEV) is not significantly advantageous for the force enhancement during the
full stroke. Plots of efficiency versus the independent variable show that, for
symmetric kinematics, the largest efficiency is achieved at an intermediate value
of each variable within the parameter range considered, where periodic flow occurs.
122 Vortex wake interactions of a two-dimensional flapping foil
6.1 Introduction
The flow around a flapping wing, fin, or tail is highly unsteady and governed by the
dynamics of the generated vortices (Weish-Fogh & Jensen, 1956, Dickinson et al.,
2000). An experimental study (Lentink et al., 2008) showed that these shed vor-
tices interact with each other and organise themselves, similarly to an oscillating
cylinder as described by Williamson & Roshko (1988), into specific wake patterns
depending on the foil kinematics. The wake pattern can be either periodic or
aperiodic and directly determines the periodicity of the aerodynamic forces acting
on the foil. Periodic flow is the result of a match between the driving frequency
and the natural shedding frequency which is referred to synchronisation of the
flow (Williamson & Roshko, 1988, Lentink & Gerritsma, 2003). The wake will be
aperiodic if synchronisation of the vortex-wake does not occur. The synchronisa-
tion band organisation for the flapping foil may be very complex due to the large
extent and high dimension of the parametric space. In contrast to the cylinder,
vortex shedding from a flapping foil displays a variation of the natural shedding
frequency as a function of angle of attack (Katz, 1981, Dickinson & G¨otz, 1993).
A numerical study by Lentink & Gerritsma (2003) showed that symmetric foil
kinematics can result in either a symmetric or an asymmetric wake. In the case
of an asymmetric wake, the initial condition determines the orientation of the
wake and hence the orientation of the time-averaged lift over a complete flapping
period. Several studies have shown that the wing benefits from the attachment of
the LEV because of the low pressure core of the LEV acting on the wing during
the full stroke (Lentink & Dickinson, 2009b,a, Ellington et al., 1996, Dickinson,
1994). However, the propulsive performance of plunging foil kinematics without a
pitching motion is poor (Lentink & Gerritsma, 2003). Therefore, it was suggested
that foil rotation is an important source for production of thrust to increase the
aerodynamic performance.
In the present research, we studied the vortex structure generated in the wake of
an ellipsoid foil undergoing flapping motion, plunging and pitching, at a Reynolds
number of Re = 150 which corresponds to the flight of a small insect, e.g. a fruit
fly. Here only the near wake of the foil is studied. The motivation for this is
that performance of a flapping foil is influenced mainly by near wake dynamics.
The objective of the present study is to investigate the influence of different foil
kinematics on the vortex-wake structure, force coefficients, and performance.
6.2 Flapping foil parametrisation
The baseline kinematic model is based on harmonic motion, such as used by Wang
(2000b,a). The flow around a flapping foil and the foil kinematics can be char-
acterised by dimensionless parameters. The method used to make the govern-
ing equations dimensionless is the same as used by Lentink & Gerritsma (2003)
and Lentink et al. (2008). This approach enables us to perform a systematic inves-
6.2 Flapping foil parametrisation 123
tigation of the influence of different foil kinematic parameters on the vortex-wake
pattern. The main important parameters are the frequency f [1/s] of both the
translation and rotation, which are coupled with a phase shift of 90

, the ampli-
tude of translation A [m], the amplitude of the sinusoidal foil rotation A
α
(

),
the forward velocity of the foil U

[m/s], the chord length of the foil c [m] and
the stroke plane angle β (

). The definition of the dimensionless parameters is
schematically illustrated in figure 6.1 and described in more detail below. The di-
mensionless wavelength λ

represents the number of chord lengths travelled during
one flapping period:
λ

=
U

fc
.
The dimensionless amplitude A

represents the ratio of amplitude of the foil trans-
lation and the chord length of the foil:
A

=
A
c
.
The Strouhal number St is based on the stroke amplitude A, and is hence equal
to the ratio of the dimensionless amplitude A

and the dimensionless wavelength
λ

:
St =
fA
U

=
A

λ

.
It corresponds to the maximum induced angle of attack A
α
ind
at mid-stroke due
to the translation of the flapping motion of the foil. The mean velocity U [m/s] is
obtained by averaging the velocity components over one flapping period:
U =
1
T
_
T
0
_
(U

+U
flap
X
)
2
+ (U
flap
Y
)
2
dt .
Here T [s] is the period, U
flap
X
[m/s], and U
flap
Y
[m/s] the velocity of the foil kine-
matics in X and Y directions respectively. The time-averaged Reynolds number
Re becomes
Re =
Uc
ν
,
where ν [m
2
/s] is the kinematic viscosity and changed for every computation to
match the Reynolds number. For the basic model the dimensionless wavelength λ

,
the dimensionless amplitude A

, the amplitude of geometric angle of attack A
α
,
and the stroke plane angle β are chosen as independent variables. The Reynolds
number is kept constant at Re = 150.
In order to study the influence of the kinematics, each parameter is varied from
the baseline model, defined by λ

= 6.8, A

= 1.5, A
α
= 15

, and β = 90

. The
dimensionless wavelength was varied from λ

= 24, 20, 12, 10, 7.9, 6.8, 6.3, 6.0,
5.7, 5.3, 4.5, 4.0 to 3.0. The dimensionless amplitude is varied within the range of
0.5 ≤ A

≤ 3.0 with a 0.5 increment. The amplitude of angle of attack varies from
0

≤ A
α
≤ 45

with a 15

increment. The influence of the stroke plane angle is
124 Vortex wake interactions of a two-dimensional flapping foil
λ

90

−β
A
α
(tan
−1
)2Stsin(β)
2A

· sin(β)
(a)
y
x
Y
X
β
β
2A
U

(b)
Figure 6.1 Schematic illustration of the foil kinematics in forward flight. (a) Illustration
of the foil parameters in forward flight: the dimensionless wavelength λ

, the dimensionless amplitude
A

, the Strouhal number St, the angle amplitude A
α
, and the stroke plane angle β. In this frame of
reference the observer is fixed relative to the undisturbed air. The flight direction is from right to left.
(b) The two-dimensional relation between two inertial coordinate systems. The downstroke phase is
filled by dark blue and the upstroke by light blue.
investigated for two different angle amplitudes A
α
= 15

and 45

in combination
with 15

≤ β ≤ 90

with a 15

increment. In the cases that the stroke plane angle
differs from 90

, the resulting flapping motion is asymmetric.
6.3 Force coefficients and performance
In this study two inertial coordinate systems are used, see figure 6.1. The XY -
plane has the X-axis in the direction of the free-stream velocity and the Y -axis in
vertical direction. The xy-plane is tilted at an angle β, the stroke plane angle. The
lift force L is the component of the total aerodynamic force perpendicular to the
forward velocity of the foil and is positive when it is in the positive Y -direction.
The drag force D is the component of the total aerodynamic force parallel to
the forward velocity of the foil and is positive when directed in the positive X-
direction. In the present study, the force and moment coefficients are scaled using
the average dynamic pressure q [N/m
2
]:
q =
1
2
ρU
2
.
Using q, the force and moment coefficients are defined as:
C
D
=
D
qc
, C
L
=
L
qc
, C
M
=
M
qc
2
.
Projecting the lift coefficient C
L
and drag coefficient C
D
onto the y- and x-axes
we obtain the foil lift coefficient C
l
and the foil drag coefficient C
d
respectively.
C
l
= C
L
· sinβ +C
D
· cosβ ,
6.4 Numerical model 125
C
d
= −C
L
· cosβ +C
D
· sinβ .
Note that a negative drag coefficient C
D
means thrust which is necessary in forward
locomotion whereas the foil drag coefficient C
d
indicates the fluid force that the
animal must overcome for translational motion of its wing, fin, or tail and is
relevant to the required power of locomotion.
The comparative assessment of the aerodynamic performance of the different
kinematic models is based on the mechanical efficiency of the foil motion. The
efficiency η [%] is the ratio between the effective propulsive power P
eff
[Nm/s]
and the required power P
req
[Nm/s] which are given in (6.1), (6.2), and (6.3)
respectively:
η =
P
eff
P
req
· 100% , (6.1)
where
P
eff
= −D · U

, (6.2)
P
req
= −
1
T
_
T
0
d · U
flap
dt −
1
T
_
T
0
M · ω
flap
dt . (6.3)
Here −D [N] represents thrust, U

[m/s] the free-stream velocity, T [s] the flapping
period, d [N] the foil drag, U
flap
[m/s] the translational velocity of the foil in the
stroke plane, M [N·m] the moment about the centre of rotation and ω
flap
[rad/s]
represents rotational velocity of the foil. Note that we have neglected inertial cost
of mechanical work done by the foil.
6.4 Numerical model
In the present thesis, a 2% thick ellipsoid shape with unit chord length represents
the foil. In order to obtain a good quality mesh, elliptical coordinates (µ, θ) are
used following Bos et al. (2008). The constant µ and θ correspond to confocal el-
lipses and hyperbolas respectively. These elliptical coordinates can be transformed
to Cartesian coordinates via a conformal mapping:
x +iy = cosh(µ +iθ) .
The result of this conformal mapping can be seen in figure 6.2. The inner O-type
mesh of 50000 cells is surrounded by a ring of tetrahedral cells. The inner mesh
is able to move, whereas the outer mesh is re-meshed every time-step. The radius
of the inner computational domain is chosen to be 25 chord lengths, such that
the influence of far field boundary condition can be neglected. A uniform grid in
(µ, θ) is concentrated around the leading and trailing edges. This type of grid is
suitable for the problem since the vorticity is strongest near the edge of the foil.
Those two-dimensional simulations are performed on mesh resolutions of about
50000 cells, more information on this mapping can be found in (Bos et al., 2008,
Wang, 2000a).
126 Vortex wake interactions of a two-dimensional flapping foil
X
Y 0
0
0
0
0 0 0 0
10
10
20
20
30
30
-3
-3
-2
-2
-1
-1
(a) (b)
Figure 6.2 The body conformal moving mesh around a 2% ellipsoid foil. (a) The O-type
body conformal mesh with a grid size of 50000 cells is moving within a ring of tetrahedral cells. (b)
The close-up of the mesh at the foil surface shows that the grid is concentrated around the leading and
trailing edge.
6.5 Results and discussion
The simulations start with the fluid at rest in which the initial velocity vector is
zero. The resulting wake patterns have been classified using a symbolic code of
letters and numbers developed by Williamson & Roshko (1988) that describes the
combination of pairs (P) and single (S) vortices shed during each flapping cycle.
The moment when a LEV is shed from the wing is defined as the moment when
its core passes the trailing edge. The averaged aerodynamic force coefficients in
table 6.1, 6.2, 6.3, and 6.4 are obtained using three simulation periods. Note that
the efficiency is only calculated when the drag is negative, i.e. thrusting mode.
6.5.1 Influence of dimensionless wavelength
The wake pattern and vortex behaviour are studied as a function of the dimension-
less wavelength in the range of 24 ≥ λ

≥ 3. The numerical results are provided
in table 6.1 by decreasing dimensionless wavelength which is equivalent to an in-
crease in flapping frequency at a constant flight velocity. Our results are similar to
the experimental results found by Lentink et al. (2008) using a soap-film tunnel.
At high dimensionless wavelengths λ

= 24 and 20 the numerical results give no
strong vortices shedding from the foil in relation to the foil oscillation. Thus the
lift and drag are a function of the effective angle of attack A
α
eff
which leads to
positive drag. Thrusting modes are found for 12 ≥ λ

because of generated LEV’s
which pull the foil toward in forward direction. For dimensionless wavelengths
12 ≥ λ

≥ 5.7 the LEV’s are shed before stroke reversal. The amount of LEV’s
6.5 Results and discussion 127
λ

pattern C
L
C
Ldownstroke
C
Lupstroke
C
D
η
24.0 no vortices 0.002 0.324 -0.320 0.218 -
20.0 no vortices 0.002 0.481 -0.478 0.185 -
12.0 2P+2S -0.001 0.925 -0.927 -0.028 4.71
10.0 2P+2S -0.003 1.095 -1.100 -0.092 10.66
7.9 2P+2S 0.009 1.495 -1.440 -0.186 13.05
6.8 2P+2S 0.005 1.633 -1.624 -0.252 13.47
6.3 2P+2S -0.004 1.704 -1.712 -0.289 13.48
6.0 2P+S -0.071 1.577 -1.719 -0.287 13.00
5.7 2P+S -0.103 1.580 -1.786 -0.302 12.70
5.3 2P 0.004 1.874 -1.833 -0.342 12.16
4.5 P+S 0.241 2.254 -1.772 -0.418 11.35
4.0 P+S 0.502 2.937 -1.934 -0.568 11.43
3.0 Aperiodic -0.494 2.034 -3.021 -0.617 9.06
Table 6.1 Influence of dimensional wavelength. The numerical results are shown for 13 different
values for the dimensionless wavelength, A

= 1.5, A
α
= 15

, and β = 90

.
and TEV’s shed from the foil is decreasing with the dimensionless wavelength be-
cause the vortices have less time to develop and shed, see figure 6.3. Therefore,
the LEV’s stay attached to the foil relatively longer at lower dimensionless wave-
lengths. The LEV’s increase in size and strength due to increasing effective angle
of attack. As a result of this, the foil produces higher lift and thrust during each
half-stroke for decreasing dimensionless wavelength. A further decrease in dimen-
sionless wavelength results in stronger vortex-wake interactions which lead to an
aperiodic wake at λ

= 3, so that the forces of this case are varying with relative
small changes from period to period. The asymmetry in the lift coefficient is a
result of the asymmetry in wake pattern. It is observed for cases when vortices are
formed on the foil that the lift changes its direction before stroke reversal. This
means that the foil cannot produce lift enhancement just before the end of each
half-stroke whether the LEV is shed before or after stroke reversal, see figure 6.4.
Therefore the foil does not fully benefit from the attachment of LEV’s.
6.5.2 Influence of dimensionless amplitude
Six dimensionless amplitudes are chosen to investigate the influence of this pa-
rameter. In table 6.2 the numerical results are given for the six cases. At low
dimensionless amplitude A

= 0.5, no vortices are formed on the foil due to the
low effective angle of attack. Nevertheless, shear layers from the foil organise
themselves into a 2S pattern. As a result, the force distributions have a sinusoidal
shape because the foil cannot produce force enhancement. For medium dimen-
sionless amplitude A

= 1.0 and 1.5 the effective angle of attack is high enough to
128 Vortex wake interactions of a two-dimensional flapping foil
(a) No vortices, λ

= 24, A
α
eff
= 7

(b) 2P+2S, λ

= 6.8, A
α
eff
= 39

(c) 2P+S, λ

= 6.0, A
α
eff
= 43

(d) 2P, λ

= 5.3, A
α
eff
= 46

(e) P+S, λ

= 4.5, A
α
eff
= 49

(f ) Aperiodic, λ

= 3, A
α
eff
= 53

Figure 6.3 Vorticity contours for decreasing wavelength. Vorticity contours of various wake
patterns for decreasing dimensionless wavelength λ

. The flow is from left to right. All images are
taken at t = 0.35T. A

= 1.5, A
α
= 15

, and β = 90

.
A

pattern C
L
C
Ldownstroke
C
Lupstroke
C
D
η
0.5 2S 0.002 0.376 -0.373 0.198 -
1.0 2P+2S 0.002 1.151 -1.147 -0.109 12.70
1.5 2P+2S 0.005 1.633 -1.624 -0.252 13.47
2.0 Aperiodic -0.038 1.561 -1.638 -0.217 9.94
2.5 Aperiodic -0.086 1.598 -1.770 -0.274 8.71
3.0 Aperiodic -0.064 1.830 -1.959 -0.302 7.13
Table 6.2 Numerical results of the kinematics for six different dimensionless amplitudes. λ

= 6.8,
A
α
= 15

, and β = 90

.
6.5 Results and discussion 129
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
k = 24, No vortices
k = 6.8, 2P+2S
k = 6.0, 2P+S
k = 5.3, 2P
k = 4.5, P+S
k = 3.0, Aperiodic
0 0.2 0.4 0.6 0.8 1
-8
-6
-4
-2
0
2
4
6
8
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
k = 24, No vortices
k = 6.8, 2P+2S
k = 6.0, 2P+S
k = 5.3, 2P
k = 4.5, P+S
k = 3.0, Aperiodic
0 0.2 0.4 0.6 0.8 1
-2
-1.5
-1
-0.5
0
0.5
1
1.5
(b)
Figure 6.4 Force coefficients to study the influence of wavelength. Lift (a) and drag (b)
histories of different wake patterns for six dimensionless wavelengths. A

= 1.5, A
α
= 15

, and
β = 90

.
A
α
pattern C
L
C
Ldownstroke
C
Lupstroke
C
D
η
0 2P+S 0.216 2.282 -1.850 0.050 -
15 2P+2S 0.005 1.633 -1.624 -0.252 13.47
30 2P 0.001 1.097 -1.094 -0.348 28.18
45 2P 0.004 0.588 -0.579 -0.089 14.58
Table 6.3 Numerical results of the kinematics for four different angle amplitudes. λ

= 6.8, A

= 1.5,
and β = 90

.
form a LEV which leads to lift enhancement and thrust. For high dimensionless
amplitude A

≥ 2.0, vortices with a diameter larger than chord length are formed.
Some of these vortices are hit by the foil during stroke reversal. Strong foil-vortex
interactions lead to an aperiodic wake pattern causing aperiodic force coefficients.
6.5.3 Influence of angle of attack amplitude
Table 6.3 shows numerical results for different angle of attack amplitudes. The
plunging kinematic model, A
α
= 0

, results in an asymmetric 2P+S pattern. The
LEV in the upstroke is weaker than those generated in the downstroke, which gives
positive mean lift over a period. No thrust is generated for this setting. Once the
foil is allowed to rotate, non-zero angle of attack amplitude, the effective angle of
attack is lower. This results in decreasing lift in each half-stroke for increasing
angle amplitude. However, the reverse trend is found for thrust. The foil rotation
leads to thrust generation due to the frontal surface area for the pressure difference
acting toward in forward direction (Lentink & Gerritsma, 2003).
A peak performance of 28.18% is obtained which is considerably larger com-
130 Vortex wake interactions of a two-dimensional flapping foil
β A
α
pattern C
L
C
Ldownstroke
C
Lupstroke
C
D
η
90 15 2P+2S 0.005 1.633 -1.624 -0.252 13.47
75 15 2P+S 0.518 2.124 -1.087 -0.103 5.42
60 15 2P 1.092 3.053 -0.868 0.303 -
45 15 Aperiodic 1.282 3.183 -0.619 0.521 -
30 15 Aperiodic 1.632 3.082 0.182 2.402 -
15 15 Aperiodic 1.160 2.264 0.056 2.464 -
90 45 2P 0.004 0.588 -0.579 -0.089 14.58
75 45 2P+S 0.600 1.262 -0.062 -0.002 0.36
60 45 3P+S 0.988 1.892 0.084 0.242 -
45 45 P+3S 1.344 2.624 0.063 0.626 -
30 45 P+2S 1.500 2.844 0.156 1.120 -
15 45 Aperiodic 1.768 3.240 0.296 2.045 -
Table 6.4 Influence of the stroke plane angle. Numerical results of the kinematics for six different
stroke plane angles in combination with two different angle amplitudes. λ

= 6.8 and A

= 1.5
pared to other cases. This is because the thrust component of the resulting aero-
dynamic force is high compared to the normal component, due to the foil rotation.
At A
α
= 45

, no significant vortices are formed on the foil because the effective
angle of attack is low A
α
eff
= 9

at mid-stroke. However, shear layers which are
generated by the foil, form themselves into a 2P pattern. At this low effective
angle of attack the foil produces lower lift and thrust.
6.5.4 Influence of stroke plane angle
The stroke plane angle causes an asymmetry in the kinematics. Here the results
for the influence of the stroke plane angle with two different angle amplitudes
A
α
= 15

and A
α
= 45

are shown in table 6.4. From the baseline kinematics the
stroke plane angle is tilted backward by 15

.
A similar trend is found for both angle amplitudes that the lift coefficient
is increasing for decreasing stroke plane angle during the downstroke until the
flow becomes aperiodic. Also the negative lift in the upstroke is decreasing in
magnitude. This is because during the downstroke of asymmetric kinematics the
foil undergoes a greater relative velocity. Therefore, the averaged lift is mainly
generated during the downstroke. The difference in relative velocity between up
and downstroke also affects the drag contribution in a similar way.
6.5.5 Discussion
In the symmetric kinematics, non-zero average lift exists only as a result of an
asymmetry in wake pattern. The orientation of the mean lift depends on initial
6.5 Results and discussion 131
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Aperiodic P+S 2P 2P+S 2P+2SNo vortices
C
L
(a) Mean lift coefficient of symmetric kinematics.
λ

A

Present study
Diptera
No vortices No vortices
Aperiodic
P+S
2P 2P+S
2P+2S
0 5 10 15 20 25
0
0.5
1
1.5
2
2.5
3
3.5
4
(b) Vortex-wake synchronisation A

− λ

dia-
gram.
Figure 6.5 Influence of the kinematics on the vortex wake pattern and force generation.
(a) The mean lift coefficient over a complete period of symmetric kinematics as a function of wake
pattern. (b) Vortex-wake synchronisation A

−λ

diagram of our sinusoidal flapping wing. The angle
amplitude and stroke plane angle are kept constant at A
α
= 15

and β = 90

. The dash line represents
our theoretical estimate of the boundary governed by equation (6.4). We have added the operating
conditions of insects belonging to the order Diptera.
conditions. The results are summarised in figure 6.5(a), where the time-averaged
lift coefficient is plotted against wake pattern. The grey bands indicate a sym-
metric wake pattern in which the nearly zero mean lift is obtained. The vortex
synchronisation diagram for all models is shown in figure 6.5(b). There is an im-
portant limitation in forward flapping locomotion. To begin with, the (absolute)
effective angle of attack should be high enough to form a LEV in order to generate
force enhancement. This approximately restricts the values of St, i.e. the ratio of
A

and λ

as:
St =
A

λ

>
1

· tan(A
α
geo
+A
α
stall
) , (6.4)
which is illustrated by the dashed line. Besides, it is thought that the results could
also shed light on the Micro Air Vehicle (MAV) design. When the wing operates
outside the synchronisation region, the wake and consequently the forces become
aperiodic which will influence the stability and controllability of the MAV’s. Fig-
ure 6.6 shows plots of efficiency versus the independent motion parameters. In
symmetric kinematics there is an optimal value for each variable, see figure 6.6(a),
(b) and, (c). The peak efficiency of 28.18% could confirm that the wing rotation
plays an important role in the unsteady aerodynamic force production.
132 Vortex wake interactions of a two-dimensional flapping foil
λ

η
A
p
e
r
i
o
d
i
c
P
+
S
2
P
2
P
+
S
2
P
+
2
S
2 4 6 8 10 12 14
0
5
10
15
(a)
A

η
A
p
e
r
i
o
d
i
c
2
P
+
2
S
0 0.5 1 1.5 2 2.5 3 3.5 4
0
5
10
15
(b)
A
α
η
2
P
2
P
+
2
S
0 10 20 30 40 50
0
10
20
30
(c)
β
η
A
α
= 15

A
α
= 45

2P
2
P
+
S
2P+2S
60 65 70 75 80 85 90 95 100
0
5
10
15
(d)
Figure 6.6 Influence of flapping kinematics on the efficiency. Efficiency as a function of
the independence variables, (a) dimensionless wavelength, (b) dimensionless amplitude, (c) angle
amplitude and (d) stroke plane angle.
6.6 Conclusions
A numerical model for two-dimensional flow was used to investigate the effect of
foil kinematics on the vortex dynamics around an ellipsoid foil subjected to pre-
scribed flapping motion over a range of dimensionless wavelengths, dimensionless
amplitudes, angle of attack amplitudes, and stroke plane angles at the Reynolds
number of 150. Both plunging and rotating motions are prescribed by simple har-
monic functions which are useful for exploring the parametric space despite the
model simplicity.
The resulting wake patterns behind the foil are categorised using the concept
of Williamson & Roshko (1988). Although such an attempt at classifying vortex
patterns can lead to confusion due to the shedding, tearing, or merging of tiny
vortices, it is suitable for straightening out the shedding vortices in our simulations.
The results are in satisfactory agreement with the comparable experiments.
6.6 Conclusions 133
Optimal propulsion using flapping foil exists for each variable, which implies
that aerodynamics might select a range of preferable operating conditions. The
conditions that give optimal propulsion lie in the synchronisation region in which
the flow is periodic. Since the computational costs are high and the parameters
cannot be varied continuously, the synchronisation band was not investigated com-
pletely. However, the present study is beneficial for understanding the influence
of wing kinematics on the performance characteristics.
CHAPTER 7
Vortical structures in
three-dimensional flapping flight
Submitted to J. Fluid Mech. (January 2010).
Results are obtained by performing numerical simulations of the three-dimensional
flow around a flapping wing. A parameter study is performed to investigate the
performance in flapping flight and to get insight into the vortex dynamics and force
generation. Different aspects, relevant for three-dimensional flapping wing aero-
dynamics, have been studied, namely the angle of attack, the Rossby number, the
Reynolds number and the stroke kinematics. First, the flow around a dynamically
scaled model wing is solved for different angles of attack in order to study the force
development and vortex dynamics at small and large mid-stroke angles of attack.
Secondly, the Rossby number is varied at different Reynolds numbers. A varying
Rossby number represents a variation in the radius of the stroke path and thus
the magnitude of the angular acceleration. Thirdly, the three-dimensional wing
kinematics is varied by changing the shape in angle of attack and by applying a
deviation, which may result in a ‘figure-of-eight’, a ‘figure-of-O’ or a ‘figure-of-U’
pattern. Finally, the three-dimensional flow is compared with the two-dimensional
studies performed on flapping forward flight.
7.1 Introduction
To understand the aerodynamic performance of flapping wings at low Reynolds
numbers, relevant for insect flight, it is important to obtain insight into the vortex
136 Vortical structures in three-dimensional flapping flight
dynamics and its influence on force development. The most important feature
in flapping wing aerodynamics has been established to be the generation of a
stable leading-edge vortex (LEV) on top of the wing, which is responsible for the
unexpectedly large force augmentation in hovering insect flight (Maxworthy, 1979,
Ellington et al., 1996, Dickinson et al., 1999, Srygley & Thomas, 2002, Lentink
& Dickinson, 2009b). In order to gain insight into the three-dimensional flow
field induced by the flapping wings, several two-dimensional studies have been
performed (Dickinson & G¨otz, 1993, Dickinson, 1994, Wang, 2005, Bos et al., 2008).
It was shown that the leading-edge vortex generated by a two-dimensional moving
foil is shed after several travelled chord lengths, while a three-dimensional LEV
remains stably attached to a three-dimensional revolving (Usherwood & Ellington,
2002) or flapping (Dickinson et al., 1999, Lehmann, 2004, Birch et al., 2004) model
wing, which rotates around its base. Those results indicate that three-dimensional
flow effects are essential for the LEV stability. Previously conducted research
addressed a possible analogy between the LEV on flapping wings and the LEV
generated by swept and delta wings (Ellington et al., 1996, Van Den Berg &
Ellington, 1997). The spiral leading-edge vortex generated by a translating swept
or delta wing is stabilised by the induced spanwise flow, which could suggest
that a spanwise flow may play an important role concerning the LEV stability in
insect flight (Ellington et al., 1996, Van Den Berg & Ellington, 1997). Lentink
& Dickinson (2009b) discussed that the stability of the LEV growth specifically
might be increased by the spanwise flow through the LEV core, driven by either
the dynamic pressure gradient on the wing’s surface, the centrifugal acceleration of
the boundary layer or the induced velocity field of the spiral vortex lines (Ellington
et al., 1996). Additionally, the LEV stability may be strengthened by a reduction
of the effective angle of attack as a result of the tip vortex generation (Birch &
Dickinson, 2001, Shyy et al., 2008b). However, Birch & Dickinson (2001) showed
no significant effect of the spanwise flow on the LEV strength and stability, using
plates at different spanwise locations to block the spanwise flow, but they did not
completely explain the LEV stability in their experiments.
In order to investigate the vortex dynamics and the stability of the leading-edge
vortex in particular, an accurate simulation method is developed to perform a CFD
simulation of a three-dimensional flapping wing. Based on the discussion about
LEV stabilisation due to wing revolving (Usherwood & Ellington, 2002, Birch et
al., 2004, Lentink & Dickinson, 2009b,a, Bos et al., 2010b), a three-dimensional
wing was modelled which was able to flap around a base of which the location
can be varied. By varying the location of the centre of rotation, the influence of
the revolving strength (Rossby number) and the effect of the tip vortices can be
studied. In addition, the kinematics is varied from simple harmonics by adding a
deviation and ‘trapezoidal’ shaped angle of attack. Recent two-dimensional sim-
ulations (Bos et al., 2008) suggested that the wing kinematics may also have a
large influence on the flapping performance in three-dimensional hovering. Addi-
tionally, (Lentink, 2008) showed interesting results concerning the stability of the
three-dimensional leading-edge vortex depending on the Rossby number (equiv-
7.2 Three-dimensional flapping wing simulations 137
F
X
F
Y
F
Z
θ(t)
φ(t)
α(t)
O
R
root
R
tip
start downstroke
start upstroke
mid-stroke
Stroke plane
F
drag
F
span
F
normal
Figure 7.1 Illustration of the wing motion and force definitions. Illustration of the wing
motion and force definitions. φ(t) corresponds to the stroke variation, α(t) to the geometrical angle of
attack and θ(t) to the deviation from the horizontal stroke plane.
alent to the stroke path curvature) and the Reynolds number. Therefore, the
Rossby number is systematically varied for different Reynolds numbers and mid-
stroke angles of attack. In agreement with (Bos et al., 2008), the kinematic model
is extended with a ‘trapezoidal’ shaped angle of attack and a non-zero deviation
is applied. Different deviation patterns are investigated, following the shape of
‘figure-of-O’, ‘figure-of-U’ and ‘figure-of-eight’.
The flapping wing modelling is described in section 7.2, which also addresses
wing geometry, kinematic models and the simulation strategy. In order to show
that the CFD method is accurate and efficient, section 7.3 briefly discusses the
validation and verification of the flow solver. The vortical flow needs to be visu-
alised in such a way that the resulting vortices are clearly visible. Different vortex
identification methods are described in 7.4. Furthermore, the results are discussed
in 7.5 and 7.6, while the conclusions are summarised in 7.7.
7.2 Three-dimensional flapping wing simulations
In order to study the vortex dynamics and stability of the leading-edge vortex, the
flow is solved using Computational Fluid Dynamics (CFD), of which the details
are described in chapter 2. The three-dimensional flapping wing is modelled in
order to provide a framework for comparison, which is still representative for true
insect flight, this is the subject of section 7.2.1. In view of limiting computing
resources, a selection of geometric and kinematic parameters is made to systemi-
cally investigate the flow phenomena of our interest. Additionally, the simulation
strategy is discussed in section 7.2.2.
138 Vortical structures in three-dimensional flapping flight
7.2.1 Modelling and parameter selection
In general, most investigations concerning flapping wing aerodynamics make use
of the modelling convention as previously described by Sane & Dickinson (2002)
and Dickson & Dickinson (2004) as applied in the experiments with a dynamically
scaled robotic fruit fly wing. The current research uses a model wing with an
ellipsoidal shaped planform with 10% thickness, since Lentink & Gerritsma (2003)
showed that airfoil shape was of minor influence on the forces and the flow field. In
addition, Luo & Sun (2005) showed that the airfoil corrugation, present in dragon-
fly wings, did not influence the force development significantly. The length scales
of the corrugation are orders of magnitude smaller compared to the length scale
of the separated flow region or the leading-edge vortex, such that significant effect
of corrugation on the flow can be neglected.
Planform selection
The single wing span is fixed to b
s
= 2.0 and the chord at mid-span is c = 1.0.
The hinge around which the wing is able to flap is fixed to a distance of 0.5 from
the wing root, such that the wing tip radius becomes R
tip
= 2.5, while the rota-
tional distance of the Robofly was fixed to 0.7 (Sane & Dickinson, 2002, Poelma
et al., 2006). Since the wing planform is chosen to be ellipsoidal, the wing sur-
face is defined by S = πab, where a = 0.5 and b = 1.0 are the semi-minor and
semi-major axes, respectively, such that S = π/2. The average chord length of
this ellipsoidal planform is found to be c = S/b
s
= π/4. So, the three geomet-
ric parameters important for the flapping wing simulations are defined: b
s
, S and c.
Kinematic models
The flapping wing motion is prescribed by three different motion angles defining
the deviation angle θ(t), flapping angle φ(t) and the angle of attack α(t). The de-
viation angle is the angle with respect to the horizontal stroke plane, as described
in chapter 4 and (Bos et al., 2008). During the stroke, the deviation is varied
harmonically with an amplitude between A
θ
= 0 and A
θ
= 20

. A combined devi-
ation and flapping angle variation leads to a wing tip pattern. Depending on the
variation of the deviation angle, figure 7.2 shows the resulting ‘figure-of-O’, 7.2(a),
‘figure-of-eight’, 7.2(b), or ‘figure-of-U’, 7.2(c).
Realistic fruit fly kinematics (Fry et al., 2003, Lentink & Dickinson, 2009b,
Bos et al., 2008) resembles an harmonically varying deviation angle, a ‘sawtooth’
shaped flapping angle and a ‘trapezoidal’ shaped angle of attack (Sane & Dick-
inson, 2002, Dickson & Dickinson, 2004) with an incidental ‘bump’, shortly after
stroke reversal (Bos et al., 2008). A two-dimensional investigation (Bos et al.,
2008) showed that the effect of the ‘trapezoidal’ angle of attack was most promi-
nent.
From the discussion in section 7.1 it can be concluded that there is need for a
detailed three-dimensional numerical study to investigate the effects of the Rossby
number, Reynolds number, angle of attack and stroke kinematics, i.e. ‘trapezoidal’
7.2 Three-dimensional flapping wing simulations 139
Flapping angle, φ [

]
D
e
v
i
a
t
i
o
n
a
n
g
l
e
,
θ
[

]
[
-
]
0 20 40 60 80 100 120 140 160 180
-80
-60
-40
-20
0
20
40
60
80
upstroke
downstroke
(a) ‘figure-of-O’
Flapping angle, φ [

]
D
e
v
i
a
t
i
o
n
a
n
g
l
e
,
θ
[

]
[
-
]
0 20 40 60 80 100 120 140 160 180
-80
-60
-40
-20
0
20
40
60
80
upstroke
downstroke
(b) ‘figure-of-eight’
Flapping angle, φ [

]
D
e
v
i
a
t
i
o
n
a
n
g
l
e
,
θ
[

]
[
-
]
0 20 40 60 80 100 120 140 160 180
-80
-60
-40
-20
0
20
40
60
80
upstroke
downstroke
(c) ‘figure-of-U’
Figure 7.2 Different wing tip patterns. Different wing tip patterns as a result of the variation
in deviation with a combined flapping motion. (a) ‘figure-of-O’. (b) ‘figure-of-eight’. (c) ‘figure-of-U’.
shape and deviation. The current study varied the Rossby number from Ro = 3.2,
which is relevant for vortex induced propulsion in nature (Lentink & Dickinson,
2009a,b), to a nearly translating wing, Ro = 130. In addition to the variation of
Reynolds number from Re = 100, 500 and 1000, the (geometric) angle of attack is
varied from α = 15

to α = 90

with increments of α = 15

. The current research
varies the amount of the ‘trapezoidal’ shape by varying the speed of rotation just
after stroke reversal from T
rot
= 0.10T to T
rot
= 0.25T, where T is the flapping
period, such that T
rot
= 0.25T corresponds to fully harmonic angle of attack vari-
ation. The flapping angle was chosen to vary harmonically to isolate the effects of
the deviation, the ‘trapezoidal’ angle of attack, the Rossby and Reynolds numbers.
Framework for comparison
In order to design a frame of comparison it is important to keep the following
three parameters fixed: the dimensionless amplitude of the wing’s cross-section at
140 Vortical structures in three-dimensional flapping flight
the radius of gyration A

R
g
, the Reynolds number at the radius of gyration Re
R
g
,
and the area swept by the wing A
swept
. Using R
g
=
_
1
S
_
R
0
r
2
c(r)dr, the radius
of gyration is determined from the rotation origin to the tip r = 0 to R
tip
. For
the baseline case, where R
tip
= 2.5 the radius of gyration becomes R
g
= 1.58.
Using (4.23), the average Reynolds number, based on the radius of gyration is
defined as:
Re
R
g
=
4A
φ
fR
g
c
ν
, (7.1)
where A
φ
is the flapping angle amplitude, f the flapping frequency, c the average
chord length and ν the kinematic viscosity.
The kinematic viscosity is fixed for three selected values, Re
R
g
= 100, 500 and
1000, provided that the wing kinematics and geometry are given. If the distance
of the rotation origin is varied, the wing tip radius changes, which is compensated
by the flapping angle amplitude in order to keep the average Reynolds number
and the displacement at R
g
comparable. Therefore, the flapping angle amplitude
is determined from (7.1), for every rotation radius.
The result of this scaling is a comparable average Reynolds number, Re
R
g
=
100, average velocity U
R
g
and displacement of the cross-section at the radius of
gyration, A

R
g
= A
φ
R
g
/c. On the other hand, maximal values, occurring at the
wing tip are still varying like Re
R
, U
R
and A

R
= A
φ
R/c. Concerning the baseline
case, with R
tip
= 2.5 the resulting amplitude of the cross-section at R
g
becomes
A

R
g
≈ 2.2, which is of similar order as used in the two-dimensional analysis in (Bos
et al., 2008).
In order to investigate the effect of three-dimensional wing kinematics in hov-
ering flight, which is the main subject of the present thesis, the hovering wing
kinematics is substituted into the expressions for the angular and centripetal co-
efficients, C
ang
and C
cen
, and the Rossby number Ro, equation (4.6) to find the
following:
C
ang
=
2
π
A
φ
R
g
c
= A

R
g
, (7.2)
C
cen
=
R
tip
c
= AR
s
, (7.3)
Ro =
R
tip
c
= AR
s
. (7.4)
Here, AR
s
is the single wing aspect ratio. It remains clear that both the centripetal
C
cen
, and the Rossby number Ro, are defined by the wing geometry, whereas the
angular acceleration number C
ang
, depends on the wing kinematics.
Force and performance definitions
In order to determine the effect of the different motion and geometric parameters
on the forces and performance, proper definitions are necessary. Since the wing
rotates with a rotating reference frame, two different force definitions are possible,
7.2 Three-dimensional flapping wing simulations 141
in the inertial and the rotating reference frame, which is shown in figure 7.1 in re-
lation to the motion angles. Because the present research concerns hovering flight,
the lift force is by definition vertical and thus equal to F
Y
. On the other hand,
the drag force is opposite to the motion direction, derived by a decomposition of
F
X
, F
Y
and F
Z
in the rotating reference frame. Therefore, the three-dimensional
lift and drag are given by:
F
lift
= F
Y
, (7.5)
and
F
drag
= F
X
· sin(φ) −F
Z
· cos(φ). (7.6)
The force in spanwise direction is not used throughout our analysis, since that
force is small compared to the lift and drag. As discussed previously (Bos et al.,
2008) in chapter 4, the force coefficients, C
L
and C
D
, are obtained by division
using the average dynamic pressure, q = 0.5ρU
2
ref
. Two performance indicators
are used, the lift-to-drag ratio, also known as the glide factor, C
L
/C
D
and the
power factor, C
3/2
L
/C
D
(see Ruijgrok, 1994).
7.2.2 Simulation strategy and test matrix selection
The following variables are varied throughout the current research, the Rossby
number Ro (due to varying rotation origin), the angle of attack amplitude α, the
Reynolds number Re and the wing kinematic model. In order to investigate the in-
fluence of the wing kinematics, the shape of the angle of attack variation, reflected
by T
rot
and the deviation amplitude, A
θ
were varied. A systematic overview of all
cases is provided here.
Influence of wing stroke curvature
The stroke curvature is varied in order to investigate if there is a possible relation
between the angular acceleration, centripetal acceleration or the Rossby number
and the forces acting on the flapping wing, under hovering conditions. The flow is
solved for different wing tip radii, equivalent to the Rossby number.
The wing tip radius is varied from fully revolving at R
tip
= 2.5 to a nearly
translating wing at R
tip
= 102. This range in stroke path curvature corresponds
to a changing Rossby number from Ro = 3.2 to 130. Lentink & Dickinson (2009b)
found that most insects and fish operate at a Rossby number close to Ro = 3.0,
which seems to be a biologically convergent solution for animals moving in fluids.
Influence of Reynolds number and angle of attack
For two selected Rossby numbers Ro = 3.2 and Ro = 130, the mid-stroke angle
of attack is varied from α = 15

to α = 90

with increments of α = 15

. This
provides insight in the force development as a function of angle of attack for a
fully revolving, i.e. flapping, and translating wing. Additionally, the Reynolds
number is varied from Re = 100, Re = 500 and Re = 1000, including a variation
in angle of attack, to study its effects on the behaviour of the leading-edge vortex.
142 Vortical structures in three-dimensional flapping flight
Rotation origin, R
tip
2.5 3.0 4.0 5.0 6.0 7.0 12.0 102.0
A
α
[

]
90 + • ◦ + • ◦
75 + • ◦ + • ◦
60 + • ◦ + + + + + + + • ◦
45 + • ◦ + + + + + + + • ◦
30 + • ◦ + • ◦
15 + • ◦ + • ◦
Table 7.1 Simulation matrix: wing stroke curvature origin, angle of attack and Reynolds
number. Simulation matrix to vary the rotation origin of the flapping wing. This matrix is used to
study the influence of the stroke curvature on the structure of the leading-edge vortex and corresponding
forces. The angle of attack at mid-stroke is varied from 90

to 15

together with the Reynolds number,
+: Re = 100, •: Re = 500, ◦: Re = 1000. The grid resolution was fixed to 800k and the time-step
was chosen corresponding to Co
max
= 2.0.
Table 7.1 shows an overview of the variation of the wing stroke curvature, angle
of attack and Reynolds number.
Influence of the kinematic modelling
Besides the wing stroke curvature, Reynolds number and angle of attack, it remains
interesting to investigate the effect of two kinematic parameters, the deviation,
which may cause a ‘figure-of-eight’ pattern and the ‘trapezoidal’ shape (defined by
T
rot
) of the angle of attack variation. This ‘trapezoidal’ angle of attack variation
is defined by
α =
_
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
_
A
α
sin(2πft) 0 ≤ t < T
rot
,
A
α
T
rot
≤ t <
1
2
T −T
rot
,
A
α
cos(2πft)
1
2
T −T
rot
≤ t < T
rot
+
1
2
T,
−A
α
T
rot
+
1
2
T ≤ t < T −T
rot
,
−A
α
cos(2πft) T −T
rot
≤ t < T.
(7.7)
Here T
rot
is the rotation duration, such that T
rot
= 0.25 recovers a fully harmonic
angle of attack variation. For different values of T
rot
, the angle of attack is plotted
in figure 7.3. Note that the geometric angle of attack is given by α
geom
=
π
2
−α.
Table 7.2 shows different deviation amplitudes A
θ
, in combination with a vary-
ing, T
rot
, which determines the amount of ‘trapezoidal’ shape, as was already
illustrated in figure 7.2. The varying deviation angle amplitude, may cause dif-
ferent wing tip patterns, i.e. ‘figure-of-O’, ‘figure-of-eight’ and the ‘figure-of-U’,
depending on the deviation frequency.
7.3 Flow solver accuracy 143
A
θ
0 5 10 15 20
T
r
o
t
0.25 + + + + +
0.20 +
0.15 +
0.10 + + + + +
Table 7.2 Simulation matrix: kinematic modelling. Simulation matrix to vary the rotation
duration and the deviation of the flapping stroke. The rotation duration is varied from T
rot
= 0.25
to T
rot
= 0.10 in order to get a ‘trapezoidal’ shaped angle of attack. The deviation is varied by the
deviation amplitude A
θ
. Two different angle of attack amplitudes are used, 45

and 60

, which is
shown to result in maximal lift coefficients. The Reynolds number is fixed to Re = 100, the grid
resolution to 800k and the time-step was chosen corresponding to Co
max
= 2.0.
t/T [-]
α
[
-
]
T
rot
= 0.10
T
rot
= 0.15
T
rot
= 0.20
T
rot
= 0.25
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
50
60
70
80
90
100
110
120
130
Figure 7.3 Angle of attack variation with a ‘trapezoidal’ shape. To investigate the influence
of a ‘trapezoidal’ shape in angle of attack, the amount of this shape is systematically varied by T
rot
.
7.3 Flow solver accuracy
In order to test the accuracy of the used flow solver, concerning highly unsteady
and vortical flows, numerical comparisons are performed. A verification is per-
formed by varying the grid resolution (grid independence study) and the time-
step size, by decreasing the maximum Courant number. The meshes for these
three-dimensional simulations are constructed with GridPro

using a structured
approach. Grid refinement is uniform and the cells are clustered close to the flap-
ping wing boundary. More detailed information on grid generation can be found
in appendix A.
In order to show that the numerical solution is grid and time-step indepen-
dent, a verification study is performed using the flow around a three-dimensionally
flapping wing. The kinematics is according to the simple harmonic model. The
flapping angle amplitude was fixed to A
φ
= 63

(1.1 rad), the mid-stroke angle
of attack was given by A
α
= 45

and the wing tip radius corresponds to fully
revolving, the Rossby number was Ro = 3.2 and the average Reynolds number
Re
R
g
= 100. Table 7.3 gives an overview of the performed simulations by varying
144 Vortical structures in three-dimensional flapping flight
Mesh resolution
100k 200k 400k 800k 1600k
C
o
m
a
x
2.0 + +
1.0 + + +• +• +
0.5 + +
Table 7.3 Simulation matrix: verification. Verification matrix showing the cases used for
verification purposes, with varying mesh resolution (100k − 1600k) and time-step. The time-step is
reflected through the maximal Courant number, Co
max
, which varies from 2.0, 1.0 to 0.5. Two cases
are performed for two different Reynolds numbers, +: Re = 100, •: Re = 1000. The kinematics of the
three-dimensional wing is simple harmonics with A
φ
= 63

, A
α
= 45

and A
θ
= 0

.
grid resolution and maximal Courant number. The smaller Co
max
, the smaller
the time-step. The grid resolution was varied from 100k to 1600k cells and the
Co
max
from 2.0 to 0.5. The spatial grid independence study was performed for a
maximal Courant number of Co
max
= 1.0 and the temporal convergence for two
grid sizes of 400k and 800k cells.
In order to assess the accuracy of the flow solver, the drag and lift coefficients
are plotted in figure 7.4 for meshes from 100k to 1600k cells and Co
max
= 1.0. The
corresponding limit cycles are shown in figure 7.5. As can be clearly seen from
the figures, the flow is periodic and the force coefficients (lift and drag) appear
to be close for the grid resolutions considered. In order to assess spatial and
temporal convergence both time-averaged lift and drag coefficients are plotted with
increasing spatial and temporal resolution in figure 7.6. Figure 7.6(a) and 7.6(b)
shows a converging time-averaged lift and drag coefficients with increasing grid
resolution, the value at ∞ is determined by Richardson extrapolation (Ferziger &
Peric, 2002), see chapter 2. Temporal convergence is shown in figure 7.6(c), where
the time-averaged lift and drag coefficients are plotted with decreasing time-step.
In order to justify the choice for grid and temporal resolution for the three-
dimensional flapping wing simulations, table 7.4 and 7.5 show the spatial and
temporal errors in average lift and drag with the Richardson extrapolated values.
Table 7.4 shows that even the differences in lift and drag for 100k mesh cells and
Co
max
= 1.0 are reasonably small, i.e. less than 4%. This may be explained by
the fact that the forces are mainly dependent on the near wake, with several chord
lengths from the wing. Apparently, all grid resolutions considered, from 100k to
1600k, are sufficiently fine to capture the near wake, on which the forces depend.
However, in order to visualise the vortices in the far wake (Bos et al., 2008) at least
400k, but preferably 800k mesh cells are desired. Table 7.5 shows the errors in lift
and drag with respect to the Richardson extrapolated values for both 400k and
800k with decreasing time-step (Co
max
= 2.0, 1.0 and 0.5). Again, the differences
are small, even for the largest time-step, corresponding to Co
max
= 2.0, the error
is less than 0.2%.
Summarising, all generated grids provide sufficiently accurate force coefficients,
but to capture the far wake vortex dynamics at least 800k cells are required.
7.4 Vortex identification methods for flow visualisation 145
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
100k
200k
400k
800k
1600k
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
1.5
2
2.5
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
100k
200k
400k
800k
1600k
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-3
-2
-1
0
1
2
3
(b)
Figure 7.4 Three-dimensional verification: force coefficients. Lift and drag coefficients for the
verification cases with varying grid size, 100k −1600k. The time-step is taken such that Co
max
= 1.0.
The kinematics of the three-dimensional wing is simple harmonics with A
φ
= 63

, A
α
= 45

and
A
θ
= 0

.
N ǫ
lift
[%] ǫ
drag
[%]
100k -3.66 -3.19
200k -2.52 -2.32
400k -1.27 -1.23
800k -0.49 -0.50
1600k -0.12 -0.13
Table 7.4 Error values of lift and drag coefficients for varying grid sizes. Error values of
lift and drag coefficients [%] with respect to the extrapolated values for varying grid sizes, ranging
from 100k to 1600k. The time-step was determined by a max Courant number of Co
max
= 1.0. The
kinematics of the three-dimensional wing is simple harmonics with A
φ
= 63

, A
α
= 45

and A
θ
= 0

.
Furthermore, at Co
max
= 2.0, the temporal errors are sufficiently small. Therefore,
a grid resolution of 800k in combination with Co
max
= 2.0 was used for all three-
dimensional flapping wing simulations, described in this chapter.
7.4 Vortex identification methods for flow
visualisation
In order to study the vortex dynamics and the stability of the leading-edge vortex
in particular, a proper vortex identification criterion is essential. Different well-
known techniques to detect and visualise vortices are based on the velocity gradient
tensor, which requires a complete velocity field. Two different vortex identification
criteria are discussed, namely the magnitude of vorticity |ω| (Lu & Shen, 2008)
146 Vortical structures in three-dimensional flapping flight
Drag coefficient [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
-3 -2 -1 0 1 2 3
-0.5
0
0.5
1
1.5
2
2.5
3
Figure 7.5 Three-dimensional verification: limit cycle. The limit cycles, constructed from the
lift and drag coefficients clearly shows periodic behaviour for the verification cases. The grid size was
fixed to 800k. The time-step is taken such that Co
max
= 1.0. The kinematics of the three-dimensional
wing is simple harmonics with A
θ
= 0 rad, A
φ
= 1.1 rad and A
α
= 0.785 rad.
Co
max
ǫ
lift
[%] 400k ǫ
drag
[%] 400k ǫ
lift
[%] 800k ǫ
drag
[%] 800k
2.0 0.082 0.183 0.067 0.159
1.0 0.053 0.097 0.043 0.087
0.5 0.013 0.024 0.011 0.022
Table 7.5 Error values of lift and drag coefficients for varying temporal resolution. Error
values of lift and drag coefficients [%] with respect to the extrapolated values for varying temporal
resolutions, corresponding to Co
max
= 2.0 to Co
max
= 0.5. Shown are the errors for two grid sizes,
400k and 800k. The kinematics of the three-dimensional wing is simple harmonics with A
φ
= 63

,
A
α
= 45

and A
θ
= 0

.
and the Q criterion (Hunt et al., 1988).
The first criterion is based on the magnitude of vorticity |ω|, where ω = ∇×u.
If |ω| reaches a user-defined threshold, that region is identified as a vortex. More
specifically, within this particular region, there is a concentration of vorticity. Since
shear layers and curved streamlines also are a source of vorticity, this criterion may
lead to undesired contours of e.g. shear layers. Especially in three-dimensional
flows this may become a difficulty. In two-dimensional flow, however, |ω| is the
common vortex visualisation method (Bos et al., 2008).
The Q criterion (Hunt et al., 1988) is the second invariant of the local velocity
gradient tensor ∇u. For Q > 0 the region is identified as a vortex. This second
invariant of ∇u is written as
Q =
1
2
_
|Ω|
2
−|S|
2
_
,
where the rate of strain tensor S is given by S =
1
2
(∇u +∇u
T
) and the vorticity
tensor by Ω =
1
2
(∇u − ∇u
T
). Hence, a positive value of Q > 0 is a measure
for any excess of rotation rate (in terms of vorticity) with respect to the strain
7.4 Vortex identification methods for flow visualisation 147
Spatial resolution [-]
A
v
e
r
a
g
e
l
i
f
t
c
o
e

c
i
e
n
t
[
-
]
12 4 8 16

1.18
1.19
1.2
1.21
1.22
1.23
1.24
(a)
Spatial resolution [-]
A
v
e
r
a
g
e
d
r
a
g
c
o
e

c
i
e
n
t
[
-
]
12 4 8 16

1.97
1.98
1.99
2
2.01
2.02
2.03
2.04
2.05
(b)
Temporal resolution [-]
A
v
e
r
a
g
e
d
r
a
g
c
o
e

c
i
e
n
t
[
-
]
400k
800k
12 4 8 16

2.015
2.02
2.025
2.03
2.035
(c)
Temporal resolution [-]
A
v
e
r
a
g
e
l
i
f
t
c
o
e

c
i
e
n
t
[
-
]
400k
800k
12 4 8 16

1.21
1.212
1.214
1.216
1.218
1.22
1.222
1.224
1.226
1.228
1.23
(d)
Figure 7.6 Spatial and temporal convergence. (a) and (b) are showing the average lift and drag
coefficients for increasing spatial resolution and constant time-step, corresponding to Co
max
= 1.0.
The final value, at ∞, is obtained using Richardson extrapolation. The temporal convergence is
illustrated in (c) and (d), showing the average drag and lift coefficients for decreasing time-step at
two specific grid sizes, 400k and 800k.
rate. Therefore, a region where Q > 0 indicates a clear swirling flow (as shown
by Chakraborty et al., 2005). It must be noted that Jeong & Hussain (1995)
found that Q > 0 is not a sufficient condition to have a pressure minimum in the
vortex core of that specific region. In most cases, however, a pressure minimum
does occur. By neglecting these unsteady and viscous effects from the governing
Navier-Stokes equations the following relation can be obtained for the symmetric
tensor Ω
2
+S
2
:

2
+S
2
= −
1
ρ
∇(∇p),
where ρ is the fluid density and p the pressure. In order to identify which vortex
criterion should be used, figure 7.7 shows iso-surfaces around a flapping wing.
The wing flaps around a distance of 0.5 from the wing root and the flapping
angles are varying harmonically. The iso-surface is visualised at t = 0.25T, during
the downstroke. At t = 0.25T the leading-edge vortex is formed on the wing’s
148 Vortical structures in three-dimensional flapping flight
(a) |ω| (b) Q
Figure 7.7 Comparison of the near wake flow field using different vortex identification
criteria. The spiralling leading-edge vortex is visualised using contour plots of the magnitude of
vorticity, |ω| and Q. The arrow shows the flapping direction of a downstroke and the flow is visualised
at mid-stroke. (a) shows the contour of the vorticity magnitude, |ω| = 5.0 and (b) the Q = 2.0 value.
The colours represent values of helicity, h = (u · ω)/(|u| |ω|), within the range of −1.0 ≤ h ≤ 1.0.
upper surface and rolls up into a tip vortex, the vortices from the previous stroke
should be visible as well. The leading-edge vortex, rolling up into a tip vortex, is
identified using a carefully chosen threshold of the vortex identification criteria,
using the values |ω| = 5.0 and Q = 2.0, normalised by their maximal values. The
colours show the helicity which is defined as h = (u· ω)/(|u| |ω|), within the range
of −1.0 ≤ h ≤ 1.0. A positive helicity, h > 0, means that the direction vector of
vorticity (ω = ∇×u) is aligned with the local flow velocity.
In figure 7.7(a) it can be observed that |ω| shows not only the vortical struc-
tures, but also the shear layers near the wing and between the vortices. This leads
to a thicker iso-surface, such that detail of the vortical structures is lost. The Q cri-
terion shows more detail, in figure 7.7(b), a smooth leading-edge vortex is shown,
rolling up into a tip vortex. Some of the previously shed vortices are still present.
Since the Q criterion offers sufficient and adequate information about the local
flow field, e.g. a rotation dominated region is identified by Q < 0, this criterion is
used throughout the remainder of the present research, for all three-dimensional
simulations.
7.5 Flapping wings at low Reynolds numbers
In order to provide insight into the vortex dynamics (for a purely harmonic flap-
ping motion) and its influence on the variation of forces, different geometric and
kinematic parameters are systematically varied. First, the influence of the angle
of attack on the forces is briefly discussed in section 7.5.1. In that section, the in-
fluence of an increasing mid-stroke angle of attack is briefly discussed on the force
development. In order to investigate the effect of the Rossby number, the radius
7.5 Flapping wings at low Reynolds numbers 149
of curvature is subsequently varied in section 7.5.2, this is performed for different
Reynolds number as well. The influence of the Reynolds number on the forces
and leading-edge vortex stability is assessed in 7.5.3. Additionally, the kinematic
model is varied by considering a ‘trapezoidal’ shape adaptation and the addition
of deviation in section 7.5.4 and 7.5.5, respectively. In addition to the variety of
hovering flight simulations, section 7.6 deals with forward flapping flight with sim-
ilar conditions as the two-dimensional simulations performed by Bos et al. (2008),
Thaweewat et al. (2009).
7.5.1 The angle of attack in flapping flight
Previous studies showed that the angle of attack variation during the stroke in-
fluences the forces considerably. This was confirmed by a recent two-dimensional
investigation (Bos et al., 2008), concerning hovering flight at fruit fly conditions.
It was already mentioned that the angle of attack at mid-stroke is varied from
α = 15

to α = 90

. Note that α = 90

implies that the wing keeps a constant up-
right position during the entire stroke. Table 7.6 shows the time-averaged lift and
drag coefficient for varying angles. The Reynolds number is Re = 100 and Rossby
Ro = 3.2, corresponding to a flapping wing with small radius of curvature. It is
seen that the maximal time-averaged lift coefficient occurs at a mid-stroke angle
of attack of α = 45

. However, the average lift-to-drag ratio obtains a maximal
value at an angle of attack of α = 30

, from table 7.6.
Figure 7.8 shows a variation of the lift and drag coefficients during a complete
flapping cycle. It can be observed that the force variation is periodic and smooth.
The average drag is maximal and lift minimal for α = 90

. Furthermore, it can be
seen in figure 7.8 that the maximal force values occur at approximately halfway
through the down and upstroke, t = 0.25T and t = 0.75T, where T is the flapping
period.
Furthermore, it must be noted that the lift is nearly always non-negative,
which means that during the hovering conditions, lift is being generated during
the complete stroke. The drag for α = 15

shows two minor peaks within each half-
stroke, due to shedding of the leading-edge vortex at low mid-stroke angle of attack.
While the lift is nearly identical for α = 45

and α = 60

, the drag is significantly
lower for α = 45

. In addition, it seems that the leading-edge vortex only grows
significantly at higher (α ≥ 45

) angles of attack. The difference between the
maximal (α = 45

) and minimal (α = 15

) lift is 57%. The next section will
discuss the drops in more detail, since these periods of lower lift appear to occur
at low angles of attack, independent of Rossby and Reynolds number.
7.5.2 Influence of flapping stroke curvature
In order to investigate the forces and development of the leading-edge vortex, the
radius of curvature is increased to decrease the angular acceleration consequently.
The Rossby number was increased from Ro = 3.2 to Ro = 130, typical for revolving
150 Vortical structures in three-dimensional flapping flight
α
geom
C
L
C
Ddownstroke
C
Dupstroke
C
L
/C
Dave
90 0.000 3.546 3.543 0.000
75 0.703 3.325 3.329 0.178
60 1.127 2.750 2.739 0.436
45 1.224 2.034 2.028 0.667
30 0.977 1.339 1.333 0.722
15 0.526 0.963 0.957 0.397
Table 7.6 Force coefficients for Re = 100 and Ro = 3.2. Time-averaged lift C
L
, drag C
D
and
lift-to-drag C
L
/C
Dave
are shown as a function of the mid-stroke geometrical angle of attack for given
Re = 100 and Ro = 3.2, so a flapping wing with small stroke curvature.
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
α = 15

α = 30

α = 45

α = 60

α = 75

α = 90

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
4
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
α = 15

α = 30

α = 45

α = 60

α = 75

α = 90

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(b)
Figure 7.8 Variation of lift and drag coefficients for a flapping wing at Reynolds number
of Re = 100. The variation is shown for lift (a) and drag (b) coefficients induced by a flapping wing at
Ro = 3.2 and a Reynolds number of Re = 100. The mid-stroke angle of attack is varied from ♦ : 15

,
•: 30

, ▽ : 45

, : 60

, △: 75

and : 90

.
7.5 Flapping wings at low Reynolds numbers 151
Ro C
L
C
Ddownstroke
C
Dupstroke
C
L
/C
Dave
3.2 1.224 (baseline) 2.034 2.028 0.603 (baseline)
3.8 1.175 (−3.9%) 1.983 2.981 0.593 (−1.6%)
5.1 1.105 (−9.7%) 1.933 1.935 0.572 (−5.2%)
6.4 1.058 (−13.6%) 1.916 1.915 0.552 (−8.4%)
7.6 1.023 (−16.4%) 1.908 1.904 0.537 (−11.0%)
8.9 0.997 (−18.5%) 1.903 1.899 0.525 (−13.0%)
15.3 0.943 (−22.9%) 1.892 1.886 0.500 (−17.2%)
130 0.922 (−24.7%) 1.883 1.877 0.490 (−18.7%)
Table 7.7 Time-averaged force coefficients for varying Rossby numbers at Re = 100. The
variation of average lift (C
L
), drag (C
D
), lift-to-drag ratio (C
L
/C
D
) are shown for Rossby numbers
from Ro = 3.2 to Ro = 130. The mid-stroke angle of attack was fixed to α = 45

and the Reynolds
number to Re = 100.
and translating wings, respectively. Table 7.7 shows the time-averaged values for
lift, drag and lift-to-drag ratio, with increasing Rossby number for α = 45

. It is
clear that both lift and drag are decreasing with increasing Rossby number, i.e.
decreasing curvature of the stroke path. At Ro = 130 the wing nearly performs
a two-dimensional motion leading to a decrease in lift of 24.7% in comparison to
the baseline case with Ro = 3.2. The decrease in drag is small, such that the
decrease in lift-to-drag ratio is still significant, 18.7%. Figure 7.9 shows the force
histories concerning the nearly translating wing, i.e. Ro = 130. When compared
with figure 7.8 (which applies to the revolving wing Ro = 3.2) it is seen that
both lift and drag variations are significantly lower when the Rossby number is
large. This is due to the loss in energy by the tip vortices which was also studied
by (Blondeaux et al., 2005a). Figure 7.10 shows the variation of the lift and
drag coefficients during the flapping cycle and the effect of Rossby number as it
increases. It is clear that the major loss in lift for high Rossby numbers, occurs at
mid-stroke t=0.25T and t=0.75T. The loss in drag, just after stroke reversal and
during mid-stroke are of similar magnitude.
Figure 7.11 shows the iso-surfaces of Q = 1.0 at t = 0.25T, which is at the
midst of the downstroke. It can be clearly observed that the fully flapping wing
shows a pronounced leading-edge vortex which spirals towards the tip to form a
tip vortex. However, the translating wing shows a leading-edge vortex which stays
symmetric with respect to the wing centre plane, feeding two wing tip vortices, at
the root and the tip. Additionally, figure 7.13 shows the streamlines to illustrate
the leading-edge and tip vortices in more detail.
An additional observation, while comparing figures 7.8 and 7.9 is that the lift
drops significantly (75% at least) during mid-stroke (t = 0.2T) for an angle of
attack of α = 15

. Figure 7.12 shows the Q = 1.0 iso-surfaces of a translating
wing at Ro = 130 for α = 15

and α = 45

. It is clearly seen that the leading-
edge vortex for α = 15

is not yet fully developed, which results in the lower
152 Vortical structures in three-dimensional flapping flight
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
α = 15

α = 30

α = 45

α = 60

α = 75

α = 90

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
4
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
α = 15

α = 30

α = 45

α = 60

α = 75

α = 90

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-6
-4
-2
0
2
4
6
(b)
Figure 7.9 Variation of lift and drag coefficients for a translating wing at Ro = 130. Lift
(a) and drag (b) coefficients induced by a flapping wing at a Rossby number of Ro = 130, such that
the wing approximately translates. The Reynolds number remained fixed at Re = 100. The angle of
attack amplitude is varied from ♦ : 15

, •: 30

, ▽ : 45

, : 60

, △: 75

and : 90

.
lift, compared to α = 45

. It seems that the trend of the force development
with the angle of attack is similar for flapping and translating wings, as long as
the scaling is appropriate, such that the dimensionless amplitude A

R
g
, average
Reynolds number Re
R
g
and swept area A
swept
are comparable.
Summarising, it can be stated that a flapping wing motion is of crucial impor-
tance for lift generation at a small penalty of drag, compared to wing translation.
Additionally, the leading-edge vortex is important for the gain in lift. This leading-
edge vortex is larger and more stable at angles of attack larger than about 30

.
At smaller angles of attack, it was shown for both flapping and translating wings
at α = 15

, that the leading-edge vortex development is not significant to increase
the lift, instead the lift decreases.
7.5.3 Influence of Reynolds number
In addition to the angle of attack and stroke curvature, a selection of Reynolds
numbers is used, Re = 100, Re = 500 and Re = 1000, relevant for insect aerody-
namics of a fruit fly (Sane & Dickinson, 2001, Birch & Dickinson, 2003), hawk-
moth (Liu & Kawachi, 1998) and dragonfly (Isogai et al., 2004), respectively. The
time-averaged lift and drag coefficients are plotted in figure 7.14(a) for different
angles of attack and different Reynolds numbers. In addition, the results of a
variation in Rossby number are shown for Re = 100, 500 and 1000. Figure 7.14(b)
shows the power factor C
L
3/2
/C
D
as a function of lift-to-drag ratio C
L
/C
D
.
From figure 7.14(a) it can be deduced that the overall lift coefficients are signif-
icantly higher for the flapping (Ro = 3.2) compared to the translating (Ro = 130)
wing. The drag increases as well. At maximal lift, α = 45

the difference between
flapping (Ro = 3.2) and translating (Ro = 130) in lift coefficient is 32.8%, 33.9%
7.5 Flapping wings at low Reynolds numbers 153
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
Ro = 3.2
Ro = 3.8
Ro = 6.4
Ro = 8.9
Ro = 130
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-0.5
0
0.5
1
1.5
2
2.5
3
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
Ro = 3.2
Ro = 3.8
Ro = 6.4
Ro = 8.9
Ro = 130
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-4
-3
-2
-1
0
1
2
3
4
(b)
Figure 7.10 Variation of force coefficients for a flapping wing with varying Rossby num-
bers. Lift (a) and drag (b) coefficients induced by a flapping wing around varying Rossby numbers,
Ro = 3.2 − 130. The amplitude of the angle of attack variation was fixed such that at mid-stroke
α = 45

. The average Reynolds number remained fixed at Re = 100. The Rossby number is varied
from ◦ : 3.2, •: 3.8, ▽ : 6.4, : 8.9, △: 130.
(a) Ro = 3.2
(b) Ro = 130
Figure 7.11 Vortex visualisation of the near wake for different Rossby numbers. Iso-
surfaces of Q = 1.0 are shown for Rossby numbers Ro = 3.2 and Ro = 130. A time-frame is shown at
mid-stroke, t = 0.25T. The average Reynolds number was fixed to Re = 100. Colours indicate helicity.
154 Vortical structures in three-dimensional flapping flight
(a) α = 45

(b) α = 15

Figure 7.12 Vortex visualisation of the near wake for a translating wing at low and high
angle of attack. Iso-surfaces of Q = 1.0 are shown for α = 15

and α = 45

for a Rossby number of
Ro = 130. A time-frame is shown at mid-stroke, t = 0.25T. The average Reynolds number was fixed
to Re = 100. Colours indicate helicity.
(a) Ro = 3.2 (b) Ro = 130
Figure 7.13 Streamline visualisation of the near wake for different Rossby numbers.
Streamlines are shown for Rossby numbers Ro = 3.2 and Ro = 130. A time-frame is shown at mid-
stroke, t = 0.25T. The average Reynolds number was fixed to Re = 100. Colours indicate helicity.
7.5 Flapping wings at low Reynolds numbers 155
Drag coefficient [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
Re=100, Ro=3.2
Re=100, Ro=130
Re=500, Ro=3.2
Re=500, Ro=130
Re=1000, Ro=3.2
Re=1000, Ro=130
0.5 1 1.5 2 2.5 3 3.5 4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
(a)
Glide ratio [-]
P
o
w
e
r
f
a
c
t
o
r
[
-
]
Re=100, Ro=3.2
Re=100, Ro=130
Re=500, Ro=3.2
Re=500, Ro=130
Re=1000, Ro=3.2
Re=1000, Ro=130
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
1
(b)
Figure 7.14 Force and performance polars. (a) shows the force polars as a function of angle
of attack amplitude. (b) illustrates the power factor C
L
3/2
/C
D
as a function of the lift-to-drag ratio
C
L
/C
D
. In both (a) and (b), the results are shown for three different Reynolds numbers, Re = 100,
500 and 1000, and two Rossby numbers, Ro = 3.2 and Ro = 130. The angle of attack varies from 90

to 15

, from right to left in (a) and from left to right in (b).
and 35.8% for increasing Reynolds numbers Re = 100, 500 and 1000, respectively.
Since the difference in lift increases (although slightly) with Re, flapping is impor-
tant at lower Reynolds numbers, but becomes slightly more important at higher
Reynolds number. The difference in drag is only significant at larger angles of
attack. At α = 45

the differences in time-averaged drag coefficient are negligi-
ble when the variation in Reynolds number is concerned, while at α = 75

the
differences in drag become significant. However, considering a flapping motion
(Ro = 3.2) with respect to translating (Ro = 130), an average difference in drag
of 7.5% is obtained. While lift is enhanced significantly, combined with a small
drag penalty, there is still a large gain in lift-to-drag, which is beneficial in terms
of performance.
Another observation from figure 7.14(a) is that at large mid-stroke angles of
attack, the time-averaged lift and drag coefficients show marginal variations with
Reynolds number, for a translating (Ro = 130) wing. On the other hand, the
Reynolds number has a larger effect on the lift and drag, while flapping. This may
be explained by considering that the leading-edge vortex was found to be more
stable on a translating wing, compared to flapping. Looking at figure 7.14(a),
the variations in lift and drag with Reynolds number are larger for lower Rossby
numbers. So, the structure of the leading-edge vortex strongly depends on the
Reynolds number in cases of large angular accelerations.
Figure 7.15 shows the iso-surface of Q = 1.0 to visualise the leading-edge vortex
on a flapping wing (Ro = 3.2) during the downstroke for both Reynolds numbers
Re = 100 and Re = 1000. As was already discussed, the leading-edge vortex be-
comes slightly unstable with increasing Reynolds numbers, which is visualised by
irregularities in the iso-surfaces. In addition, the streamlines for the corresponding
156 Vortical structures in three-dimensional flapping flight
(a) Re = 100
(b) Re = 1000
Figure 7.15 Vortex visualisation of the near wake for different Reynolds numbers, Ro =
3.2. Iso-surface of Q = 1.0 are shown for two Reynolds numbers, Re = 100 and Re = 1000, for a
flapping wing Ro = 3.2. A time-frame is shown at mid-stroke, t = 0.25T. Colours indicate helicity.
comparison are shown in figure 7.16. Besides the irregularities, the leading-edge
vortex clearly detaches at a smaller distance from the wing root for Re = 1000.
However, although the leading-edge vortex may be less stable, the lift increasing
effects of the leading-edge vortex are larger for higher Reynolds numbers. Addi-
tionally, for higher Reynolds numbers, the leading-edge vortex may burst as was
discussed by Lentink & Dickinson (2009b), without a significant loss in lift. In or-
der to illustrate the irregular motion at larger Reynolds numbers, i.e. Re = 1000,
figure 7.17 shows the vortical motion just before stroke reversal from down to
upstroke.
It was noted that the leading-edge vortex may play a more important role for
flapping motion, compared to translation. Figure 7.18 shows the Q iso-surfaces
for different Reynolds numbers Re = 100 and Re = 1000, for a translating wing.
In case of the flapping wing Ro = 3.2, figure 7.15 shows some irregularities of the
leading-edge vortex, with increasing Reynolds numbers. For a translating wing
Ro = 130, these irregularities are less pronounced. While the leading-edge vortex
detaches at a smaller distance from the wing tip, on a flapping wing for increasing
Reynolds numbers, this is not the case for a translating wing.
Therefore, it seems plausible that the generation of a leading-edge vortex is
important for both flapping Ro = 3.2 and translating Ro = 130 flight. A Reynolds
number increase, leads to larger lift enhancement, but also to irregularities such
that the flow at low Ro is more sensitive to changes in Reynolds number.
7.5 Flapping wings at low Reynolds numbers 157
(a) Re = 100 (b) Re = 1000
Figure 7.16 Streamline visualisation of the near wake for different Reynolds numbers,
Ro = 3.2. Streamlines are shown for two Reynolds numbers, Re = 100 and Re = 1000, for a flapping
wing Ro = 3.2. A time-frame is shown at mid-stroke, t = 0.25T. Colours indicate helicity.
(a) Re = 100
(b) Re = 1000
Figure 7.17 Vortex visualisation of the near wake for different Reynolds numbers, Ro =
3.2. Iso-surface of Q = 1.0 are shown for two Reynolds numbers, Re = 100 and Re = 1000, for a
flapping wing Ro = 3.2. A time-frame is shown at the end of the downstroke, t = 0.5T. Colours
indicate helicity.
158 Vortical structures in three-dimensional flapping flight
(a) Re = 100 (b) Re = 1000
Figure 7.18 Vortex visualisation of the near wake for different Reynolds numbers, Ro =
130. Iso-surface of Q = 1.0 are shown for two Reynolds numbers, Re = 100 and Re = 1000, for a
translating wing Ro = 130. A time-frame is shown at mid-stroke, t = 0.25T. Colours indicate helicity.
7.5.4 Influence of ‘trapezoidal’ angle of attack
In order to study the influence of the ‘trapezoidal’ angle of attack and later com-
pared to an earlier two-dimensional study (Bos et al., 2008), the shape of the
angle of attack is varied. Various experimental and numerical studies have been
conducted, using a ‘trapezoidal’ shaped angle of attack variation. As previously
described, the ‘trapezoidal’ shape is defined by a rotation timing T
rot
, which is
varied from 0.25 to 0.1, the first representing a harmonic variation, whereas the
latter corresponds to a strong rotation during stroke reversal. Figure 7.3 in sec-
tion 7.2.2 shows the angle of attack as a function of the rotation duration T
rot
.
While varying the rotation duration, the Reynolds number and mid-stroke angle
of attack remained fixed.
Table 7.8 shows the time-average lift, drag and lift-to-drag values. The most
important observation is that with decreasing rotation duration, i.e. increasing
angular acceleration during stroke reversal, a gain in average lift is obtained of
10.8%. The average drag decreases with a similar amount, leading to a significant
increase in lift-to-drag ratio of 21.9%. Furthermore, it can be seen that the average
drag coefficient, generated in both up- and downstroke, are within 1.0%, so drag
generation is symmetric during a complete stroke.
The time variation of both lift and drag is shown in figure 7.19 for varying
rotation duration during a full flapping stroke. At low rotation duration T
rot
=
0.10, the angular acceleration just after stroke reversal is large, which leads to an
increase in lift, accompanied by a decrease in drag, which may be caused by a fast
decrease in effective angle of attack. Since the wing rotates relatively quickly after
reversal, it reaches its mid-stroke angle of attack early in the stroke, compared
7.5 Flapping wings at low Reynolds numbers 159
T
rot
C
L
C
Ddownstroke
C
Dupstroke
C
L
/C
Dave
0.25 1.127 (baseline) 2.750 2.740 0.411 (baseline)
0.20 1.177 (+4.5%) 2.629 2.625 0.448 (+9.2%)
0.15 1.215 (+7.8%) 2.539 2.536 0.479 (+16.6%)
0.10 1.248 (+10.8%) 2.496 2.493 0.500 (+21.9%)
Table 7.8 Time-averaged force coefficients for varying ‘trapezoidal’ shape of angle of
attack. The time-averaged lift C
L
, drag C
D
and lift-to-drag ratio C
L
/C
D
are shown for a rotation
duration, varying from T
rot
= 0.25 to T
rot
= 0.10. This corresponds to respectively harmonic variation
to a ‘trapezoidal’ shape. The average Reynolds number was maintained at Re = 100.
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
T
rot
= 0.10
T
rot
= 0.15
T
rot
= 0.20
T
rot
= 0.25
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-0.5
0
0.5
1
1.5
2
2.5
3
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
T
rot
= 0.10
T
rot
= 0.15
T
rot
= 0.20
T
rot
= 0.25
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-5
-4
-3
-2
-1
0
1
2
3
4
5
(b)
Figure 7.19 Force coefficients for a flapping wing with ‘trapezoidal’ angle of attack. Lift
(b) and drag (c) coefficients induced by a flapping wing, such that Ro = 3.2 and Re = 100. The
rotation duration is varied from T
rot
= 0.25 to T
rot
= 0.10, using an angle of attack amplitude of
A
α
= 60

.
with the harmonic case T
rot
= 0.25. This causes a long period of lift enhancement,
which leads to the integrated gain in lift of 10.8%. A similar, but opposite, effect
applies to the drag coefficient.
At every time instance, the lift increase and drag decrease are only marginal,
but overall the results are significant, 21.9% lift-to-drag enhancement. Therefore,
it can be stated that a ‘trapezoidal’ shape in angle of attack considerably increases
performance in three-dimensional hovering flapping flight. In contrast to this, two-
dimensional studies (Bos et al., 2008) showed opposite effects, due to a premature
vortex shedding of the leading-edge vortex during the long period of high angle
of attack. The explanation for this discrepancy is that in the three-dimensional
simulations the leading-edge vortex was found to remain more stable than in two-
dimensional investigations. Therefore, it can be concluded that the leading-edge
vortex stability should be studied with a three-dimensional approach.
160 Vortical structures in three-dimensional flapping flight
7.5.5 Influence of deviation
The deviation amplitude A
θ
is used to tilt the two-dimensional airfoil or three-
dimensional wing out of the horizontal stroke plane. It may be used to generate
certain wing tip patterns, e,g. the well-known ‘figure-of-eight’, which is present in
realistic fruit fly kinematics (Fry et al., 2003). In (Bos et al., 2008) it was shown
that although deviation did not influence the time-averaged values, the instanta-
neous lift and drag variations are significantly affected. In order to investigate
the influence of deviation the amplitude A
θ
is varied from 0

to 20

. In addition,
three different stroke patterns are studied, by varying the deviation frequency,
resulting in patterns that can be characterised as ‘figure-of-O’, ‘figure-of-U’ and
‘figure-of-eight’, which are shown in figure 7.2. Considering the ‘figure-of-eight’
patterns, two different deviation directions are studied, corresponding to a vari-
ation of A
θ
from 0

to 20

and from 0

to −20

. The reference velocities are
adapted correspondingly.
First, the deviation amplitude is varied from A
θ
= 0

to A
θ
= 20

, according
to the ‘figure-of-O’ wing tip pattern. Following figure 7.2, the wing moves consec-
utive down and up during the downstroke and up and down during the upstroke.
Since a downward motion increases the effective angle of attack, which is therefore
subjected to an increase, decrease, decrease and another increase during the four
consecutive half-strokes. Table 7.9 shows the time-averaged force coefficients for
a wing following this ‘figure-of-O’ motion, while the mid-stroke angle of attack is
fixed to α = 45

and the Reynolds number maintained at Re = 100. It is shown
that the time-averaged lift coefficient decreases significantly with 9.8%. However,
this is fully compensated by a decrease in drag with the same amount such that
the differences in average lift-to-drag ratio are negligible. The average lift-to-drag
was obtained by using the average drag over up- and downstroke. This was nec-
essary, because of the asymmetry appearing in the average drag coefficient. This
asymmetry in drag is the result of the asymmetrical variation in effective angle of
attack, as was previously discussed. Figure 7.20 show the lift and drag variations
during a complete flapping stroke.
Secondly, the results are considered for a flapping wing following the ‘figure-
of-U’ pattern, which is similar to the one used in (Bos et al., 2008). When using
this kinematic pattern, the wing moves down and up during every half-stroke, i.e.
the upstroke is identical to the downstroke. In table 7.10 is can easily be seen
that the differences in up- and downstroke drag are negligible for all deviation
amplitudes. This is in contrast to the observations, considering the ‘figure-of-O’.
The time-averaged drag coefficient is constant for deviation amplitude variation,
while the lift coefficient decreases with 9.4%, leading to a similar decrease in lift-
to-drag ratio. Because of symmetric (similar up- and downstroke) flapping, a
decrease of lift-to-drag coefficient is obtained, which is present, but not significant
in comparison to the effect of varying Rossby and Reynolds numbers.
The third pattern is governed by the ‘figure-of-eight’ wing tip motion. Al-
though, the previously described deviation patterns only cause marginal effects,
7.5 Flapping wings at low Reynolds numbers 161
A
θ
C
L
C
Ddownstroke
C
Dupstroke
C
L
/C
Dave
0 1.224 (baseline) 2.030 2.030 0.603 (baseline)
5 1.204 (-1.6%) 2.067 1.934 0.602 (-0.15%)
10 1.180 (-3.6%) 2.072 1.850 0.601 (-0.23%)
15 1.151 (-5.9%) 2.056 1.767 0.602 (-0.12%)
20 1.104 (-9.8%) 1.996 1.676 0.601 (-0.28%)
Table 7.9 Time-averaged force coefficients for varying deviation using a ‘figure-of-O’
pattern. The time-averaged lift C
L
, drag C
D
and lift-to-drag ratio C
L
/C
D
are shown for a deviation
amplitude, varying from A
θ
= 0

to 20

, for a ‘figure-of-O’. The mid-stroke angle of attack was fixed
at α = 45

and the average Reynolds number was maintained at Re = 100.
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
20
A
θ
= 0
A
θ
= 5
A
θ
= 10
A
θ
= 15
A
θ
= 20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-0.5
0
0.5
1
1.5
2
2.5
3
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
A
θ
= 0
A
θ
= 5
A
θ
= 10
A
θ
= 15
A
θ
= 20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-3
-2
-1
0
1
2
3
4
(b)
Figure 7.20 Force coefficients for a flapping wing with varying deviation using ‘figure-of-
O’. Lift (a) and drag (b) coefficients induced by a flapping wing, such that Ro = 3.2 and Re = 100.
The deviation amplitude is varied from A
θ
= 0

to A
θ
= 30

, using an angle of attack amplitude of
A
α
= 45

.
the ‘figure-of-eight’ pattern may cause significant changes in forces. Two types of
‘figure-of-eight’ patterns are used, from A
θ
= 0

to 20

, this is called type 1 and
the second from A
θ
= 0

to −20

, type 2. To achieve a ‘figure-of-eight’ pattern,
the deviation frequency is twice the frequency of the other two wing tip patterns.
For both types of patterns, the effective angle of attack variation consists of three
parts during each half stoke, see figure 7.2. Both up- and downstroke follow exactly
the same, thus symmetrical, motion. The type 1 patterns starts each half-stroke
with a downward motion, than it goes up until it has to go done just before stroke
reversal. This wing tip motion leads to a consecutive increase, decrease and in-
crease in effective angle of attack, where the period of decrease is twice the period
of increase. The type 2 pattern follows precisely the inverse motion.
Table 7.11 shows the time-averaged lift, drag and lift-to-drag ratio’s for both
types of ‘figure-of-eight’ patterns. Since the variation of the effective angle of at-
tack is symmetric, the drag coefficient is equal for up- and downstroke. The results
162 Vortical structures in three-dimensional flapping flight
of the type 1 motion pattern are remarkable. There is a considerable decrease in
both time-averaged lift (52%) and drag (44%) when comparing A
θ
= 20

with
the baseline case A
θ
= 0

. Since the drag decrease is of similar magnitude as the
lift decrease, the loss in lift-to-drag is limited to 15.5%, which is still significant.
Figure 7.20 shows the lift and drag variations for this type 1 ‘figure-of-eight’ pat-
tern, where the wing tip moves down, up and down, consecutively during each
half-stroke. It can be observed that the short period of downward motion at the
beginning of each stroke increases lift. On the other hand, the large period of
upward motion, decreases the effective angle of attack for a relative long period,
leading to a significant loss of lift, as is seen in the figure. For A
θ
= 20

the lift
even shows a clear minimum at t = 0.2T, which was also present in cases without
deviation but at small angles of attack, e.g. α = 15

, see section 7.5.1.
When considering the type 2 ‘figure-of-eight’ pattern, table 7.11 shows no sig-
nificant decrease in drag. The lift, however, decreases considerably, although sig-
nificantly smaller compared to the type 1 kinematic pattern, 53% versus 12.6%.
The motion of the type 2 deviation is apparently equally distributed, resulting
in only 12.6% less lift and no differences in drag, while increasing the deviation
amplitude. The maximal difference in lift-to-drag ratio is therefore -12.0%. This
force balance is illustrated in figure 7.22, which shows the lift and drag during a
full stroke. At the beginning of a stroke, the effective angle of attack is decreased,
leading to lower lift and lower drag. During mid-stroke, the effective angle of
attack is increased, which is reflected in the higher lift and drag.
Summarising, it was shown that the variation in lift and drag can be signifi-
cantly influenced by introducing deviation in the stroke pattern, i.e. ‘figure-of-O’,
‘figure-of-U’ and ‘figure-of-eight’. The ‘figure-of-O’ patterns resulted in an asym-
metric force variation, due to asymmetric modulation of the effective angle of
attack. Lift and drag decrease with a similar amount, such that the lift-to-drag
ratio was only marginally affected. The time-averaged drag was not influenced
by the symmetrical ‘figure-of-U’ pattern. The average lift, however, did decrease,
such that a loss in lift-to-drag was observed. Two types of ‘figure-of-eight’ patterns
were investigated, differing in the direction of motion. When the wing moved up-
A
θ
C
L
C
Ddownstroke
C
Dupstroke
C
L
/C
Dave
0 1.224 (baseline) 2.030 2.030 0.603 (baseline)
5 1.187 (-3.0%) 2.003 1.994 0.593 (-1.7%)
10 1.167 (-4.6%) 2.012 2.001 0.580 (-3.8%)
15 1.137 (-7.1%) 2.014 2.002 0.565 (-6.3%)
20 1.108 (-9.4%) 2.021 2.007 0.548 (-9.0%)
Table 7.10 Time-averaged force coefficients for varying deviation using a ‘figure-of-U’
pattern. The time-averaged lift C
L
, drag C
D
and lift-to-drag ratio C
L
/C
D
are shown for a deviation
amplitude, varying from A
θ
= 0

to 20

, for a ‘figure-of-U’ pattern. The mid-stroke angle of attack
was fixed at α = 45

and the average Reynolds number was maintained at Re = 100.
7.6 Flapping wings in forward flight 163
A
θ
C
L
C
Ddownstroke
C
Dupstroke
C
L
/C
Dave
0 1.224 (baseline) 2.030 2.030 0.603 (baseline)
5 1.107 (-9.5%) 1.859 1.852 0.596 (-1.2%)
10 0.934 (-23.7%) 1.621 1.616 0.576 (-4.5%)
15 0.747 (-39.0%) 1.370 1.365 0.545 (-9.5%)
20 0.575 (-53.0%) 1.129 1.125 0.510 (-15.5%)
-5 1.220 (-0.3%) 2.067 2.064 0.600 (-0.5%)
-10 1.205 (-1.5%) 2.057 2.047 0.586 (-2.8%)
-15 1.151 (-6.0%) 2.056 2.040 0.560 (-7.1%)
-20 1.070 (-12.6%) 2.017 2.003 0.530 (-12.0%)
Table 7.11 Time-averaged force coefficients for varying deviation using a ‘figure-of-eight’
pattern. The time-averaged lift C
L
, drag C
D
and lift-to-drag ratio C
L
/C
D
are shown for a deviation
amplitude, varying from A
θ
= 0

to 20

and A
θ
= 0

to −20

, for a ‘figure-of-eight’ pattern. The
mid-stroke angle of attack was fixed at α = 45

and the average Reynolds number was maintained at
Re = 100.
wards during mid-stroke, decreasing the effective angle of attack for a long period,
the performance was limited in terms of lift and lift-to-drag ratio. On the other
hand, if the wing moved downward during each mid-stroke, the performance was
similar to the ‘figure-of-U’ pattern.
The lift and drag are shown to be sensitive to different stroke patterns, such
that the forces and performance can be easily modulated. Therefore, insects could
use stroke plane deviation in extreme hovering or manoeuvring conditions. These
findings are very interesting for the development of Micro Air Vehicles (MAV) as
well.
7.6 Flapping wings in forward flight
In order to relate the results of the three-dimensional flow simulations to previously
conducted two-dimensional studies (Thaweewat et al., 2009), a three-dimensional
forward flapping wing has been studied. Based on (Thaweewat et al., 2009), the
dimensionless wavelength is set to λ = 6.3 (maximal performance) in order to
justify a comparison. The mid-stroke angle of attack is varied from α = 0

to 45

,
while maintaining a Reynolds number of Re = 150. The three-dimensional wing
motion is harmonic and scaled such that the dimensionless amplitude A

R
g
, based
on the radius of gyration R
g
is comparable between different cases. Additionally,
the average Reynolds number Re
R
g
is matched.
In table 7.12, the time-averaged force coefficients are shown for three differ-
ent flapping situations: a two-dimensional plunging airfoil, a three-dimensional
translating wing (Ro = 130) and a three-dimensional flapping wing (revolving,
Ro = 3.2). For both flapping and translating wings, the force coefficients are com-
164 Vortical structures in three-dimensional flapping flight
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
A
θ
= 0
A
θ
= 5
A
θ
= 10
A
θ
= 15
A
θ
= 20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-0.5
0
0.5
1
1.5
2
2.5
3
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
A
θ
= 0
A
θ
= 5
A
θ
= 10
A
θ
= 15
A
θ
= 20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-3
-2
-1
0
1
2
3
(b)
Figure 7.21 Force coefficients for a flapping wing with varying deviation using type 1 of
‘figure-of-eight’. Lift (a) and drag (b) coefficients induced by a flapping wing, such that Ro = 3.2
and Re = 100. The deviation amplitude is varied from A
θ
= 0

to A
θ
= 20

, using an angle of attack
amplitude of A
α
= 45

.
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
A
θ
= 0
A
θ
= −5
A
θ
= −10
A
θ
= −15
A
θ
= −20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
A
θ
= 0
A
θ
= −5
A
θ
= −10
A
θ
= −15
A
θ
= −20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-4
-3
-2
-1
0
1
2
3
4
(b)
Figure 7.22 Force coefficients for a flapping wing with varying deviation using type 2 of
‘figure-of-eight’. Lift (a) and drag (b) coefficients induced by a flapping wing, such that Ro = 3.2
and Re = 100. The deviation amplitude is varied from A
θ
= 0

to A
θ
= −20

, using an angle of
attack amplitude of A
α
= 45

.
7.6 Flapping wings in forward flight 165
A
α
C
Lave
C
D
C
L
/C
Dave
2
D
0 2.376 (baseline) 0.0676 35.148
15 1.969 (baseline) -0.401 -4.920
30 1.403 (baseline) -0.580 -2.419
45 0.834 (baseline) -0.429 -1.947
R
o
=
1
3
0
0 1.710 (-28.0%) 0.028 61.571
15 1.414 (-28.2%) -0.231 -6.119
30 1.016 (-27.6%) -0.316 -3.211
45 0.644 (-22.8%) -0.175 -3.673
R
o
=
3
.
2 0 1.904 (-20.9%) 0.068 27.964
15 1.544 (-21.6%) -0.270 -5.719
30 1.061 (-24.4%) -0.366 -2.897
45 0.604 (-27.6%) -0.214 -2.824
Table 7.12 Time-averaged force coefficients in forward flight. Two- and three-dimensional
time-averaged force coefficients for a flapping wing in forward flight at Re = 150, λ

= 6.3. The
mid-stroke angle of attack is varied from 0

to 45

. A flapping wing (Ro = 3.2) and a translating wing
(Ro = 130) are compared with a two-dimensional plunging airfoil.
pared with the two-dimensional results. Note that negative drag means thrust.
It is clear that the time-averaged lift coefficient decreases with increasing an-
gle of attack amplitude. This is illustrated for the flapping wing (Ro = 3.2)
in figure 7.23. Maximal lift occurs without wing rotation, but the resulting
thrust is minimal. The drag varies with angle of attack such that it reaches a
thrust optimum for A
α
= 30

, which is the case for both two-dimensional and
three-dimensional flapping. Concerning the three-dimensional translating wing
(Ro = 130), the lift decreases with 28% compared to the two-dimensional plung-
ing airfoil. This is the case for mid-stroke values of the angle of attack. The
generation of tip vortices causes a loss of energy, which results in a lower lift coef-
ficient. The two tip vortices result in a symmetric flow, such that for all angles of
attack, a relative equal amount of energy is lost, leading to a similar decrease in
lift, as is shown in table 7.12. Therefore, it can be stated that a three-dimensional
wing, performing a two-dimensional motion, leads to similar force results as if the
study was completely two-dimensional.
The three-dimensional flapping wing (Ro = 3.2) generates larger lift coefficients
compared to the translating case, but lower compared to the two-dimensional
flapping airfoil. The difference with the translating wing becomes smaller with
increasing angle of attack amplitude. Figure 7.24 shows a comparison of the flow
fields for Ro = 130 and Ro = 3.2, without an angle of attack variation. For the
case without revolving (figure 7.24(a)), a smooth ring vortex is formed by the
166 Vortical structures in three-dimensional flapping flight
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
α = 0

α = 15

α = 30

α = 45

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-5
-4
-3
-2
-1
0
1
2
3
4
5
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
α = 0

α = 15

α = 30

α = 45

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
(b)
Figure 7.23 Force coefficients for a three-dimensional wing in forward flight. Lift (a) and
drag (b) coefficients for a three-dimensional flapping wing in forward flight at Re = 150, λ

= 6.3 and
mid-stroke angle of attack of A
α
= 45

.
two counter-rotating tip vortices. This vortex ring is shed and convected into the
wake. The flapping wing (figure 7.24(b)), however, generates a more complex flow
field, induced by a spiralling leading-edge and tip vortex, even though the angle
of attack is zero. When the angle of attack is increased, figure 7.25 shows for
A
α
= 30

that the wake becomes more smooth, which is governed by a decrease in
effective angle of attack. The revolving wing (Ro = 3.2) induces a stable leading-
edge vortex, leading to larger lift and thrust, compared to a translating (Ro = 130)
three-dimensional wing.
Summarising, the conclusion can be drawn that a leading-edge vortex is very
important in three-dimensional flapping flight. While the leading-edge vortex de-
taches and convects in two-dimensional plunging, it remains stably attached on
top of a three-dimensional wing, increasing both lift and thrust. However, the
induced vortices are strongly dependent on the Rossby number, influencing the
forces accordingly. Although, the two-dimensional forces are higher, compared to
the three-dimensional cases, the force variation is similar. Therefore, performing
a two-dimensional analysis may be representative to investigate three-dimensional
flapping wing aerodynamics.
7.7 Conclusions
This chapter deals with the results obtained by performing numerical simulations
of the flow around a three-dimensional flapping wing. A numerical model has been
developed which solves the flow around a three-dimensional wing with complex
wing kinematics. The numerical code was verified by using a temporal and spatial
independence study.
Different aspects, relevant to three-dimensional flapping wing aerodynamics,
7.7 Conclusions 167
(a) Ro = 130 (b) Ro = 3.2
Figure 7.24 Vortex visualisation of the near wake of a three-dimensional flapping wing
in forward flight. Iso-surface of Q = 1.0 are shown for a three-dimensional flapping wing in forward
flight at Re = 150, λ

= 6.3 and mid-stroke angle of attack of A
α
= 0

. (a) shows the wing for
Ro = 130, while (b) shows the iso-surfaces for Ro = 3.2. Colours indicate helicity.
(a) Ro = 130 (b) Ro = 3.2
Figure 7.25 Vortex visualisation of the near wake of a three-dimensional flapping wing
in forward flight. Iso-surface of Q = 1.0 are shown for a three-dimensional flapping wing in forward
flight at Re = 150, λ

= 6.3 and mid-stroke angle of attack of A
α
= 30

. (a) shows the wing for
Ro = 130, while (b) shows the iso-surfaces for Ro = 3.2. Colours indicate helicity.
168 Vortical structures in three-dimensional flapping flight
have been studied. First, the flow around a dynamically scaled model wing is
solved for different angles of attack in order to study the force development and
vortex dynamics at small and large mid-stroke angle of attack. Secondly, the
Rossby number is varied at different Reynolds numbers. A varying Rossby num-
ber represents a variation in radius of stroke path and thus angular acceleration.
Thirdly, the three-dimensional wing kinematics is varied by changing the shape in
angle of attack and by applying a deviation, which may cause a ‘figure-of-eight’
pattern. Finally, the three-dimensional flow is compared with the two-dimensional
studies performed on flapping forward flight. All numerical simulations were dy-
namically scaled, using the radius of gyration. The radius of gyration was used
in order to design a proper framework for comparison for the whole parameter
space investigated, two-dimensional as well as three-dimensional. The considered
parameters are subsequently investigated leading to the following results.
The effect of a variation in angle of attack in this three-dimensional study is
such that the maximal lift occurs at a mid-stroke angle of α = 45

. However,
the performance, in terms of maximal lift-to-drag ratio was found to be maximal
at α = 30

. The flapping motion induces a leading-edge vortex, which causes a
peak in lift halfway during each up- and downstroke. This leading-edge vortex
appears to be strong at sufficiently high angles of attack α = 30

. At α = 15

the
leading-edge vortex is not strong enough, causing a sudden decrease in both lift
and drag during each half-stroke.
Secondly, the effects are studied of a varying stroke curvature, reflected by the
Rossby number. Both time-averaged lift and drag decrease significantly with in-
creasing Rossby number. At Ro = 130, a nearly translating wing, the lift decreases
with 24.7%, compared to the flapping wing with Ro = 3.2. The major decrease
in lift and drag occurs during mid-stroke, between t = 0.25T and t = 0.5T. Flow
visualisations showed that the leading-edge vortex is significantly reduced in size
and strength for the translating with at Ro = 130, compared to the flapping wing.
In addition, the leading-edge vortex rolls-up to form two tip vortices instead of
one in case of the flapping wings. This causes both lift and drag to be significantly
lower at Ro = 130.
To study the effect of three different Reynolds numbers (Re = 100, Re = 500
and Re = 1000), the force polars are constructed which also shows the relation with
the angle of attack and Rossby numbers. It was seen that with increasing Reynolds
number, both time-averaged lift and drag increases, but the differences become
smaller at Re = 1000. The effect of a changing Reynolds number is negligible
for both lift and drag at high mid-stroke angles of attack translation. This is
caused by the larger importance of the leading-edge vortex for a flapping wing at
Ro = 3.2. For both flapping, Ro = 3.2, and translation Ro = 130, irregularities in
the leading-edge vortex occur, when the Reynolds number increases. For higher
Reynolds numbers the vortex separates earlier, but lift and drag still increases.
When the Reynolds number increases even further, it probably reaches a limit,
beyond which the leading-edge vortex bursts, causing a loss in performance.
Besides the influence of the angle of attack, Rossby and Reynolds numbers, the
7.7 Conclusions 169
effects of the wing kinematics has been investigated. First, the shape of the angle
of attack variation is varied along various ‘trapezoidal’ shapes. The ‘trapezoidal’
shape causes a fast pitch-up motion at the beginning of each up- and downstroke.
Therefore, the mid-stroke angle of attack is reached earlier and maintained for a
longer period, compared to cases with a simple harmonic angle of attack variation.
This leads to a significant increase in time-averaged lift of 10.8%. The lift increase
is accompanied by a decrease in drag of a similar amount, such that the lift-to-
drag ratio shows a significant gain of 21.9%. Two-dimensional studies showed
an opposite result which was caused by the fact that the leading-edge vortex
separated, while translating at a constant angle of attack. Three-dimensional
effects, however, lead to firm and stably attacked leading-edge vortex.
The second effect of kinematic modelling is reflected by the presence of de-
viation. Three different wing tip patterns may be caused by deviation, such as
the ‘figure-of-O’, ‘figure-of-U’ and the ‘figure-of-eight’. The ‘figure-of-O’ wing tip
pattern leads to an asymmetric variation in effective angle of attack, leading to
differences in drag for the up- and downstroke. Nevertheless, the effect of this
pattern on the time-averaged forces coefficients is negligible. On the other hand,
the ‘figure-of-U’ pattern does influence the forces. The effective angle of attack is
significantly decreased during mid-stroke. The time-averaged lift coefficient is de-
creased by 9.4%. While drag is maintaining nearly constant, the lift-to-drag ratio
is decreased by about 9%. The ‘figure-of-eight’ affects the forces most, however,
it depends on the starting direction. Two types are considered, the first starting
the downstroke with an downward deviation motion, whereas the type 2 starts
upwards. The effective angle of attack is being influenced in such a way that the
type 1 ‘figure-of-eight’ decreases both lift and drag considerably. Lift decreases up
to 53%. Since drag decreases fast as well, the loss in lift-to-drag ratio is limited to
15.5%. The type 2 ‘figure-of-eight’ motion, which starts with an upward motion
at the beginning of the stroke, the drag is maintained constant. However, lift
decreases 12.6% such that the lift-to-drag ratio decreases with about 12.0%.
In addition to the effects of Reynolds number, Rossby number and wing kine-
matics in hovering flight, a preliminary study is performed to compare two- and
three-dimensional forward flapping flight. Both two- and three-dimensional simu-
lations are dynamically scaled using the radius of gyration to justify comparison.
For a dimensionless wavelength of k = 6.3 the lift and drag forces are compared
for different rotation amplitudes at a Reynolds number of Re = 150. For the
translating wing (Ro = 130), the three-dimensional simulations result in 28% less
lift compared to the two-dimensional case. This difference is mainly caused by
the loss in lift due to the generation of a tip vortex which is only present in the
three-dimensional simulations. However, an increase in Rossby number resulted
in a significant gain in lift. In combination with a higher thrust this observation
leads to the conclusion that a stable leading-edge vortex (induced by the revolving
motion) plays an important role in three-dimensional flapping aerodynamics.
CHAPTER 8
Influence of wing deformation by
flexing
A preliminary study is performed to investigate the effects of wing flexing in flap-
ping wing aerodynamics. The effects of a cosine-shaped wing flexing is analysed
in two-dimensional forward flapping flight. For three-dimensional hovering flight,
the effects of wing flexing in spanwise and chordwise directions are discussed. In
two-dimensional forward flapping flight, the flexing of the airfoil shows similar
effects as if wing rotation was applied, increasing its efficiency. Furthermore, in
three-dimensional hovering, the flexing reduces the strength of the leading-edge
vortex, compared to a rigid wing. This leads to an overall decrease in lift and
drag, this influence is larger for chordwise compared to spanwise flexing.
Section 8.1 deals with the influence of flexing for the two-dimensional plunging
airfoil in forward flight. Its influence on the flow induced by a three-dimensional
wing in hovering flight is subject of study in section 8.2. The conclusions are
drawn in section 8.3.
8.1 Airfoil flexing in two-dimensional forward
flapping flight
In chapter 6, the influence on the forces and vortex patterns of a variation in
dimensionless wavelength, flapping amplitude, angle of attack and stroke plane
angle was investigated in two-dimensional forward flapping flight. These results
are extended by using preliminary simulations of deformation airfoils, subjected to
a pre-defined flexing. The boundary displacements due to the flexing are defined
172 Influence of wing deformation by flexing
with respect to the initial boundary shape and varying in time such that the
maximal flexing occurs at mid-stroke. The resulting airfoil shape is either a quarter
or half-cosine shape, corresponding to ǫ
f
= 0.25 or ǫ
f
= 0.5, respectively, as is
shown in figure 8.1. Besides the two different deformation mode shapes, the flexing
amplitude is varied from A
f
= 0.1 to A
f
= 0.4, which is similar to maximal 40%
of the chord length. This flexing was imposed on a baseline flapping motion
(see chapter 6) with a dimensionless wavelength of λ

= 6.8 and a dimensionless
amplitude of A

= 1.5. The rotation amplitude is fixed to A
α
= 0

and the average
Reynolds number was fixed at Re = 150.
Table 8.1 shows the time-averaged lift, drag and lift-to-drag coefficients for the
plunging airfoil. For both flexing shapes the lift decreases with flexing amplitude,
however, the time-averaged lift decreases faster for ǫ
f
= 0.25. Considering this
quarter-cosine shaped flexing, a small flexing amplitude of A
f
= 0.1 already results
in a significant loss in lift of 14%, which gradually increases up to 49.5% for
A
f
= 0.8, which is a very large deformation. The decrease in lift for the half-
cosine shape, ǫ
f
= 0.5, is smaller compared to all ǫ
f
= 0.25 cases, for similar
flexing amplitudes, but increases more rapidly with increasing flexing amplitude.
On the other hand, the half-cosine deformation (ǫ
f
= 0.5) is more efficient in
generation of thrust (negative drag). For example, the decrease in average lift for
A
f
= 0.4 and ǫ
f
= 0.5 is half the value obtained with ǫ
f
= 0.25, but instead, the
thrust is doubled. It appears that the half-cosine shaped flexing increases thrust
significantly, while the lift is less affected compared to the quarter-cosine shape.
Another observation from table 8.1 is the presence of asymmetry in the time-
averaged lift coefficient for the quarter-cosine deflection. Chapter 6 shows a similar
asymmetric flow behaviour concerning the airfoil rotation. It was shown that
X-coordinate [m]
Y
d
i
s
p
l
a
c
e
m
e
n
t
[
m
]
ǫ
f
= 0.0
ǫ
f
= 0.25
ǫ
f
= 0.5
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Figure 8.1 Flexing displacements to represent wing deformation. The flexing displacements
of a plunging airfoil are shown for different shape ratio’s. A shape ratio ǫ
f
= 0.25 corresponds to a
quarter-cosine and ǫ
f
= 0.5 to a half-cosine shape.
8.1 Airfoil flexing in two-dimensional forward flapping flight 173
A
f
C
Lave
C
Ldownstroke
C
Lupstroke
C
D
0 1.956 (baseline) 1.758 -2.154 0.058
ǫ
f
=
0
.
2
5
0.1 1.683 (-14.0%) 1.880 -1.485 -0.044
0.2 1.510 (-22.8%) 1.393 -1.628 -0.109
0.4 1.492 (-23.7%) 1.492 -1.492 -0.263
0.8 0.988 (-49.5%) 0.988 -0.987 -0.229
ǫ
f
=
0
.
5
0.1 1.867 (-4.6%) 1.865 -1.868 -0.088
0.2 1.810 (-7.5%) 1.810 -1.810 -0.215
0.4 1.732 (-11.5%) 1.632 -1.628 -0.465
0.8 1.021 (-47.8%) 1.021 -1.021 -0.332
Table 8.1 Time-averaged force coefficients for a deforming plunging airfoil. The time-
averaged lift and drag are shown for a deforming two-dimensional airfoil, for two flexing shapes, ǫ
f
=
0.25 and ǫ
f
= 0.5. The Reynolds number was fixed to Re = 150, the dimensionless wavelength was
k = 6.8 and the rotation amplitude remained at A
α
= 0.
the force asymmetry is strongly related to the wake dynamics. Therefore, airfoil
flexing using a quarter-cosine shape (ǫ = 0.25) may induce a similar effect on the
wake dynamics as airfoil rotation. However, when applying the half-cosine airfoil
flexing, the lift coefficient is symmetric over a complete stroke. While the quarter-
cosine increases the airfoil camber, the half-cosine shape redirects the trailing-edge
towards the uniform free-stream. This redirection of the trailing-edge compensates
for the destabilising effect of the airfoil rotation, such that a symmetric force is
the result of a symmetric wake pattern, which is beneficial in terms of efficiency.
Figure 8.2 shows the drag coefficients for both flexing shapes. It is clear that the
thrust (negative drag) is more prominent for the half-cosine (ǫ = 0.5) flexing mode.
The main regions of thrust enhancement are during mid-stroke, from t = 0.1T to
t = 0.4T and t = 0.6T to t = 0.9T. In order to obtain a better understanding of
the flow physics, figure 8.3 shows vorticity contours for both flexing shapes with
an amplitude of A
f
= 0.2 at t = 0.25T, which is midway during the downstroke.
Comparing figure 8.3(a) with 8.3(b), it is apparent that the leading-edge vortex is
of similar size and strength for both flexing shapes. However, the trailing edge of
the half-cosine shaped airfoil appears to enhance the speed of the vortex shedding.
Therefore, at similar time instance, the shed vortices are farther away downstream
for the half-cosine (ǫ = 0.5) flexing mode, increasing its thrust.
Summarising, this preliminary two-dimensional investigation, to understand
the effect of wing flexing, has led to some interesting results. The applied wing
flexing influences the flow in a similar way as was obtained by applying airfoil
rotation, indicating a similar mechanism in terms of the effects of angle of attack.
With the introduction of flexing, the drag became negative to generate thrust in
forward flight, and the lift decreased significantly. The lift and drag development
strongly depends on the shape of the wing flexing. The half-cosine flexing shape
174 Influence of wing deformation by flexing
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
A
flex
= 0.0
A
flex
= 0.1
A
flex
= 0.2
A
flex
= 0.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-2
-1.5
-1
-0.5
0
0.5
(a) ǫ = 0.25
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
A
flex
= 0.0
A
flex
= 0.1
A
flex
= 0.2
A
flex
= 0.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-2
-1.5
-1
-0.5
0
0.5
(b) ǫ = 0.5
Figure 8.2 Comparison of drag coefficients for a flexing airfoil. The drag coefficients are shown
for the quarter-cosine (a) and half-cosine (b) shaped airfoil flexing modes. The flexing amplitude A
flex
is varied from 0.0 to 0.4, the latter corresponds to 40% of the chord length.
(a) ǫ = 0.25 (b) ǫ = 0.5
Figure 8.3 Vorticity contours. Vorticity contours around a plunging flexing airfoil without ro-
tation at Re = 150 are shown during mid-stroke at t = 0.25T. The flapping amplitude was fixed to
A

= 1.5 and the dimensionless wavelength was set to λ

= 6.8. The flexing amplitude was set to
A
f
= 0.2 using both flexing shapes, while the rotation amplitude was kept to zero A
α
= 0

.
results in less lift than obtained using the quarter-cosine shape. Additionally, the
half-cosine flexing mode also generates the largest thrust. This behaviour was
related to the fact that the half-cosine shape caused the generated vortices to
convect faster, since the trailing-edge became aligned with the flow.
8.2 Wing flexing in three-dimensional hovering flight 175
8.2 Wing flexing in three-dimensional hovering
flight
In addition to the investigation of flexing effects on an airfoil in forward flapping
flight, it was chosen to investigate a three-dimensional wing as well. Hovering
conditions are assumed, since it was extensively studied in the previous chapter.
Besides the influence of the Rossby number, Reynolds number and the kinematic
modelling, applied under rigid wing conditions, it may be interesting to study the
additional effects of wing deformation. The deformation of the wing is defined by a
user-defined flexing function, leading to a realistic bending of the three-dimensional
wing (Combes & Daniel, 2003a,b, Shyy et al., 2008a).
In order to investigate the effects of three-dimensional wing flexing on the
forces and performance, the flexing is defined according to the model described
in section 4.3.3. The deformation model is hence similar to the one used for
the two-dimensional airfoil in the previous section. Additionally, the flexing was
independently applied to the spanwise and the chordwise directions. The three-
dimensional wing is deformed such that its shape prescribes a quarter-cosine func-
tion, which is maximal at the wing tip or the trailing edge, respectively. This
cosine shaped deformation varies harmonically in time during the stroke such that
largest change in deformation occurs during stroke reversal, which is considered
to be realistic. The spanwise and chordwise directions are investigated separately.
The flexing amplitude A
f
is varied from A
f
= 0.0 to A
f
= 0.4 in spanwise di-
rection, while in chordwise direction the amplitude was varied from A
f
= 0.0 to
A
f
= 0.2, due to limitations of the mesh motion solver. An amplitude of A
f
= 0.2
corresponds to a maximal deflection at mid-stroke of 20% of the representative
flexing length (wing span or chord, respectively). Figure 8.1 shows an illustration
of the flexing model.
A
f
C
L
C
Ddownstroke
C
Dupstroke
C
L
/C
Dave
S
p
a
n
0 1.234 (baseline) 2.034 2.028 0.608 (baseline)
0.1 1.212 (-1.74%) 2.017 2.011 0.602 (-0.9%)
0.2 1.213 (-1.73%) 2.029 2.023 0.599 (-1.5%)
0.3 1.206 (-2.3%) 2.034 2.028 0.594 (-2.3%)
0.4 1.190 (-3.6%) 2.031 1.992 0.592 (-2.6%)
C
h
o
r
d
0.1 1.137 (-7.9%) 2.129 2.126 0.534 (-12.1%)
0.2 1.036 (-16.0%) 2.267 2.266 0.457 (-24.8%)
Table 8.2 Time-averaged force coefficients for a deforming three-dimensional wing. Time-
averaged force coefficients of a three-dimensional flexing wing. The spanwise flexing amplitude is varied
from A
f
= 0.0 to A
f
= 0.4, while the chordwise flexing amplitude varies from A
f
= 0.0 to A
f
= 0.2.
.
176 Influence of wing deformation by flexing
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
A
f
= 0.0
A
f
= 0.2
A
f
= 0.4
A
f
= 0.1
A
f
= 0.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
1.5
2
2.5
(a)
t/T [-]
D
r
a
g
c
o
e

c
i
e
n
t
[
-
]
A
f
= 0.0
A
f
= 0.2
A
f
= 0.4
A
f
= 0.1
A
f
= 0.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-3
-2
-1
0
1
2
3
(b)
Figure 8.4 Force coefficients for a spanwise and chordwise deforming flapping wing. Lift
(a) and drag (b) coefficients induced by a flapping wing, which deforms in spanwise (A
f
= 0.0 − 0.4)
and chordwise (A
f
= 0.0 − 0.2) direction. The latter are shown by and △. The angle of attack at
mid-stroke remained at α = 45

and the Reynolds number remained at Re = 100.
.
While varying the flexing amplitude, the average Reynolds number is kept
constant at Re = 100 and the mid-stroke angle of attack is chosen to be α = 45

.
Table 8.2 shows the time-averaged force coefficients for different flexing amplitudes,
using spanwise or chordwise deformation. The most important result is that both
time-averaged lift and drag are only marginally influenced by the spanwise flexing,
so where the wing tip bends. The maximal difference with the non-deforming case
(-3.6%) occurs for 40% wing tip deflection, which is a significant deformation.
The time-averaged drag remains nearly unaffected as well, such that the maximal
difference in lift-to-drag ratio is only -2.6%.
The chordwise deformation, on the other hand, influences the time-averaged
forces to a large extent. As is easily seen, the time-averaged lift decrease with
7.9% for only a small (A
f
= 0.1) chord deformation and 16% for A
f
= 0.2. The
drag increases for the chordwise deforming wing, with respect to the case without
flexing, such that the maximal difference in lift-to-drag ratio is -24.8%.
The force variation, during a full flapping stroke, is shown in figure 8.4. It
is clear that the largest difference in force coefficients occurs during mid-stroke,
lowering the maximal lift coefficient. It seems that the leading-edge vortex is ex-
pelled by the wing deformation, with a maximal effect at A
f
= 0.2 of chordwise
flexing. When the lift decreases, the drag increases during mid-stroke. Addition-
ally, figure 8.5 shows the streamlines to visualise the leading-edge vortex during
mid-stroke (t = 0.25T) for both spanwise and chordwise flexing. For comparison,
also the case using no flexing is shown. Compared to the case without flexing the
leading-edge vortex detaches earlier (less close to the wing tip) for both flexing
cases. The spanwise flexing has less influence than the chordwise flexing mode
shape. The influence of spanwise flexing is minor, which was also reflected in the
8.3 Conclusions 177
time-averaged force coefficients, which were only marginally lower. However, the
chordwise flexing induced the leading-edge vortex to burst almost at the middle
of the wing, leading to significantly less lift, see table 8.2.
8.3 Conclusions
In this chapter, the result have been described of a preliminary investigation to
understand the effects of wing flexing. Therefore, a pre-defined flexing deformation
has been applied to a plunging airfoil in two-dimensional forward flight and to a
three-dimensional flapping wing in hovering flight. First of all, the two-dimensional
flexing results in a similar effect on the flow and resulting forces as rotation of the
airfoil. With increasing flexing amplitude, the larger effective angle of attack leads
to the generation of negative drag or thrust. In addition, the influence of the flexing
shape (quarter-cosine or half-cosine) of the airfoil was studied. A quarter-cosine
shaped flexing results in significantly higher thrust at less lift.
Secondly, a flexing deformation has been applied to a three-dimensional flap-
ping wing in hovering flight. Two flexing directions have been considered, span-
wise and chordwise flexing. While keeping the flexing amplitude constant, the
chordwise flexing affects the flow the most. For a chordwise flexing amplitude of
A
f
= 0.2 the time-averaged lift and drag coefficients are significantly decreased
and increased, respectively, such that the lift-to-drag decreases with 24.8%. The
spanwise flexing does not show a comparable impact, only -2.6%.
178 Influence of wing deformation by flexing
(a) No flexing
(b) Spanwise flexing, A
f
= 0.2 (c) Chordwise flexing, A
f
= 0.2
Figure 8.5 Streamlines on a flexing wing in hovering flight. Iso-surface of Q = 1.0 are shown
for a three-dimensional flapping wing in hovering flight at Re = 100 and mid-stroke angle of attack of
A
α
= 45

at t = 0.25T. (a) shows the wing without flexing, while (b) shows the wing using spanwise
flexing and (c) using chordwise flexing, both with A
f
= 0.2. Colours indicate helicity.
CHAPTER 9
Conclusions and recommendations
In the present study we investigate mesh motion techniques to be able to per-
form parametric variation of the flow around flapping foils and wings. Different
existing mesh motion methods have been compared using cell quality measures.
To improve the mesh quality a mesh motion technique has been implemented,
based on the interpolation of radial basis functions on the mesh interior. Using
that technique, it has become possible to efficiently solve the flow around flapping
foils and wings at low Reynolds numbers, for flow conditions corresponding to the
scale of flying insects. The improved and implemented mesh motion technique,
is used to solve the flow around flapping wings in hovering and forward flapping
flight. Additionally, a preliminary study of the influence of wing flexing has been
conducted as well.
The overall conclusions of this research are given in section 9.1. Secondly, the
conclusions on the assessment of different mesh motion techniques are drawn in
section 9.2. Section 9.3 presents the conclusions of a flapping wing under hovering
conditions, for two-dimensional airfoil as well as for the three-dimensional wing.
Section 9.4 shows the results for forward flapping flight, while section 9.5 describes
the conclusions of the preliminary study to understand the effects of wing flexing.
Additional recommendation are made in Section 9.6.
9.1 Overall conclusions
The following overall conclusions are drawn:
1. The flow around a flapping wing, at the scale relevant to insect flight, can
be solved accurately using advanced mesh motion techniques in existing flow
180 Conclusions and recommendations
solvers.
2. Radial basis function mesh motion leads to improved mesh quality, compared
to methods based on solving the Laplace or a modified stress equation. How-
ever, mesh motion based on radial basis function interpolation is much more
computationally demanding. Different ways to improve its efficiency are
discussed.
3. The wing kinematic pattern has a large influence on the forces in two-
dimensional hovering and forward flight.
4. In forward flight, thrust generation is the result of the forward tilting of the
wing.
5. In two- and three-dimensional flapping flight, the leading-edge vortex is a
very important lift enhancement mechanism.
6. The Rossby number, which describes the wing stroke curvature, has a large
influence on the time-averaged lift force. A low Rossby number significantly
increases lift as the result of an induced spiralling leading-edge vortex.
7. Wing flexing may play an important role in insect flight to modulate the
forces and to generate thrust.
9.2 Conclusions on mesh motion techniques
Two existing mesh motion techniques have been described which are commonly
used in existing CFD codes. The first method solves the Laplace equation with
a variable diffusion coefficient, which is used to control the final mesh quality.
Secondly, a modified stress equation was used as the basis for mesh motion. Ad-
ditionally, a mesh motion solver is implemented, which uses the interpolation of
radial basis functions (RBF). The following conclusions are drawn:
1. For both existing mesh motion solvers, based on the Laplace and a modified
stress equation, the mesh quality is not sufficient for flapping wing cases,
where the rotation angles are large. However, these methods are very efficient
since existing iterative solvers can be used to solve a sparse system.
2. The mesh motion solver based on the radial basis function (RBF) interpola-
tion does not need any information about the mesh connectivity and can be
applied to arbitrary unstructured meshes containing polyhedral cells. Addi-
tionally, different radial basis functions, concerning the RBF mesh motion,
are compared. The RBF mesh quality provides superior mesh quality over
the Laplace and modified stress equation mesh motion. Especially, when
using the thin plate spline (TPS) or the continuous polynomial C
2
as radial
9.3 Conclusions on hovering flapping flight 181
basis functions, the mesh quality is high in terms of low skewness and non-
orthogonality. The TPS has global support, whereas the C
2
basis function
has compact support.
3. Since the RBF mesh motion technique encounters a dense system of equa-
tions, different methods are implemented to increase its efficiency. First of
all, a subset of the moving boundary points was selected, because not all
points are necessary if the body performs a rigid body motion. Therefore, a
coarsening algorithm selects the desired control points. Secondly, a smooth-
ing function is used to decrease the RBF contribution to zero at the outer
domain boundaries.
9.3 Conclusions on hovering flapping flight
9.3.1 Two-dimensional hovering
The effects of wing motion kinematics on the aerodynamic characteristics of hov-
ering insect flight have been investigated by means of two-dimensional numerical
flow simulations. The results of the present two-dimensional study has provided
useful insights, which may be relevant also for the understanding of realistic three-
dimensional insect flight.
Four different kinematic models, with different complexity, have been analysed.
Two of these models, pure harmonic motion and the Robofly experimental kine-
matics have been used extensively in literature. The most prominent aspects of
the Robofly kinematic model are the ‘sawtooth’ displacement and the ‘trapezoidal’
angle of attack. The third model represents the actual fruit fly kinematics as ob-
served in experiments and the last one was a modification of the latter, chosen
to investigate the effect of symmetry. The fruit fly models are characterised by a
‘bump’ in angle of attack and the presence of deviation. To facilitate the compar-
ison these models were dynamically scaled at Re = 110 and constructed such that
their mean quasi-steady lift coefficient was matched. The following conclusions
are drawn about the two-dimensional hovering simulations:
1. It was found that the realistic fruit fly wing kinematics result in significantly
lower drag at similar lift compared with the simplified wing kinematic models
used in literature. The trend that the fruit fly kinematics increases aerody-
namic performance agrees with the predictions of the quasi-steady theory,
but the numerical flow simulations provide a more complete quantitative
analysis of the flow behaviour.
2. It was shown that the difference in performance in terms of mean lift-to-drag
ratio between the different kinematic models was significant. The mean
aerodynamic drag at equal lift of the fruit fly models is about 49% lower
compared to the Robofly model and about 29% lower with respect to the
harmonic model.
182 Conclusions and recommendations
3. The ‘sawtooth’ amplitude used in the Robofly model has a small effect on
the mean lift but the mean drag is affected significantly. Due to the high
acceleration during stroke reversal of the ‘sawtooth’ shaped amplitude, the
mean drag at comparable lift is increased by 24%.
4. The second model simplification used by the Robofly, the ‘trapezoidal’ angle
of attack, caused the LEV to separate during the translational phase. This
led to an increase in mean drag during each half-stroke. Also in this case
large accelerations at stroke reversal lead to a decrease in lift-to-drag ratio
of 33%.
5. The extra ‘bump’ in angle of attack as used by the fruit fly model is not
affecting the mean lift to a large extent. During the beginning of the up
and downstroke the ‘bump’ decreases the angle of attack such that the wing
orientation is almost horizontal. This leads to a significant decrease in drag
which improves aerodynamic performance in the sense of lift-to-drag ratio
by 15.6%.
6. The other realistic kinematic feature is the deviation, which is found to have
only a marginal effect on the mean lift and mean drag. However, the effective
angle of attack is altered such that the deviation leads to levelling of the force
distribution over the flapping cycle.
9.3.2 Three-dimensional hovering
A numerical model has been developed for solving the flow around a three-dimen-
sional wing with complex wing kinematics. Different aspects, relevant to three-
dimensional flapping wing aerodynamics, have been studied. First, the flow around
a dynamically scaled model wing is solved for different angles of attack in order to
study the force development and vortex dynamics at small and large mid-stroke
angle of attack. Secondly, the Rossby number is varied at different Reynolds num-
bers. A varying Rossby number represents a variation in stroke path curvature
and thus angular acceleration. Thirdly, the three-dimensional wing kinematics is
varied by changing the shape in angle of attack and by applying a deviation, which
may cause a ‘figure-of-eight’ pattern. Finally, the three-dimensional flow is com-
pared with the two-dimensional studies performed on flapping forward flight. The
following conclusions are drawn with respect to the three-dimensional simulations,
concerning hovering flight:
1. The effect of a variation in angle of attack is such that the maximal lift
occurs at a mid-stroke angle of α = 45

. However, the performance, in
terms of maximal lift-to-drag ratio was found to be maximal at α = 30

.
The flapping motion induces a leading-edge vortex which caused a peak in
lift halfway during each up- and downstroke. This leading-edge vortex is
9.3 Conclusions on hovering flapping flight 183
strong at sufficiently high angles of attack α > 30

. At α = 15

the leading-
edge vortex is less strong, causing a sudden decrease in both lift and drag
during each half-stroke.
2. Both time-averaged lift and drag decrease significantly with increasing Ross-
by number. At Ro = 130, a nearly translating wing, the lift decreases
with 24.7%, compared to the flapping wing with Ro = 3.2. The major
decrease in lift and drag occurs during mid-stroke, between t = 0.25T and t =
0.5T. Flow visualisations showed that the leading-edge vortex is significantly
reduced in size and strength for the translating wing at Ro = 130, compared
to the flapping wing. In addition, the leading-edge vortex rolls-up to form
two tip vortices instead of one in case of the flapping wings. This causes
both lift and drag to be significantly lower at Ro = 130.
3. It was seen that for increasing Reynolds number, both time-averaged lift
and drag increases, but the differences become smaller. The effect of a
changing Reynolds number is negligible for both lift and drag at high mid-
stroke angles of attack, when the wing is nearly translation at Re = 130. This
is caused by the larger importance of the leading-edge vortex for a flapping
wing at Ro = 3.2. For both flapping (Ro = 3.2) and translation (Ro = 130),
irregularities in the leading-edge vortex occur, when the Reynolds number
increases. For higher Reynolds numbers the vortex separates earlier, but lift
and drag still increases. When the Reynolds number increases even further,
a limit is reached, beyond which the leading-edge vortex bursts, causing a
loss in performance.
4. The ‘trapezoidal’ shape in angle of attack causes a fast pitch-up motion at
the beginning of each up- and downstroke. Therefore, the mid-stroke angle
of attack is reached earlier and maintained for a longer period, compared
to cases with a simple harmonic angle of attack variation. This leads to
a significant increase in time-averaged lift of 10.8%. The lift increase is
accompanied by a decrease in drag of a similar amount, such that the lift-to-
drag ratio shows a significant gain of 21.9%. Two-dimensional studies showed
an opposite result which was caused by the fact that the leading-edge vortex
separated, while translating at a constant angle of attack. Three-dimensional
effects, however, lead to a firm and stably attached leading-edge vortex.
5. Three different wing tip patterns, caused by deviation, have been studied,
such as the ‘figure-of-O’, ‘figure-of-U’ and the ‘figure-of-eight’. The ‘figure-
of-O’ wing tip pattern leads to an asymmetric variation in effective angle of
attack, leading to differences in drag for the up- and downstroke. Nonethe-
less, the effect of this pattern on the time-averaged forces coefficients is
negligible. On the other hand, the ‘figure-of-U’ pattern does influence the
forces in a non-beneficial way. The effective angle of attack is significantly
decreased during mid-stroke. The time-averaged lift coefficient is decreased
184 Conclusions and recommendations
by 9.4%. While drag is maintaining nearly constant, the lift-to-drag ratio is
decreased by about 9%. The ‘figure-of-eight’ affects the forces most, how-
ever, it depends on the starting direction. Two types are considered, the
first starting the downstroke with an downward deviation motion, whereas
the type 2 starts upwards. The effective angle of attack is being influenced
in such a way that the type 1 ‘figure-of-eight’ decreases both lift and drag
considerably. Lift decreases up to 53%. Since drag decreases fast as well,
the loss in lift-to-drag ratio is limited to 15.5%. The type 2 ‘figure-of-eight’
motion, which starts with an upward motion at the beginning of the stroke,
the drag is maintained constant. However, lift decreases 12.6% such that the
lift-to-drag ratio decreases with about 12.0%.
9.4 Conclusions on forward flapping flight
9.4.1 Two-dimensional forward flapping
A numerical model for two-dimensional flow has been used to investigate the ef-
fect of motion kinematics on the vortex dynamics around an ellipsoid foil subjected
to prescribed flapping motion over a range of dimensionless wavelengths, dimen-
sionless amplitudes, angle of attack amplitudes, and stroke plane angles at the
Reynolds number of 150. Both plunging and rotating motions are prescribed by
simple harmonic functions which are useful for exploring the parametric space
despite the model simplicity. Concerning the two-dimensional forward flapping
simulations, the following overall conclusions are drawn:
1. The resulting wake patterns behind the foil are categorised. Although such
an attempt at classifying the observed vortex patterns can lead to a degree
of uncertainty in determining the exact wake pattern due to the shedding,
tearing, or merging of big and small vortices.
2. Optimal propulsion using flapping foil exists for each variable which implies
that aerodynamics might select a range of preferable operating condition.
The conditions that give optimal propulsion lie in the synchronisation region
in which the flow is periodic.
9.4.2 Three-dimensional forward flapping
In addition to the investigation of the effects of Reynolds number, Rossby number
and wing kinematics in hovering flight, a preliminary study is performed to com-
pare two- and three-dimensional forward flight. Both two- and three-dimensional
simulations are dynamically scaled using the radius of gyration to justify compar-
ison. The three-dimensional simulations, concerning forward flight are performed
for a dimensionless wavelength of k = 6.3 and two Rossby numbers, Ro = 130
and Ro = 3.2 at a Reynolds number of Re = 150. The following conclusions are
drawn:
9.5 Preliminary conclusions on wing flexing 185
1. For the case without rotation the three-dimensional simulations result in
14.5% less lift compared to its two-dimensional counterpart. This difference
is mainly caused by the loss in lift due to the generation of a tip vortex which
is only present in the three-dimensional simulations.
2. An increase in rotation angle resulted in significant higher lift, up to 70.5%,
which is the opposite as in the two-dimensional plunging wing. Together
with the higher thrust, this leads to the conclusion that a stable leading-edge
vortex plays a more important role in three-dimensional flapping compared
to two-dimensional plunging, in the situation that the rotation angle is non-
zero.
3. For a dimensionless wavelength of k = 6.3 the lift and drag forces are com-
pared for different rotation amplitudes at a Reynolds number of Re = 150.
For the translating wing (Ro = 130), the three-dimensional simulations re-
sult in 28% less lift compared to the two-dimensional case. This difference is
mainly caused by the loss in lift due to the generation of a tip vortex which
is only present in the three-dimensional simulations. However, an increase in
Rossby number resulted in a significant gain in lift. In combination with a
higher thrust this observation leads to the conclusion that a stable leading-
edge vortex (induced by the revolving motion) plays an important role in
three-dimensional flapping aerodynamics.
9.5 Preliminary conclusions on wing flexing
A flexing deformation has been applied to a plunging airfoil in two-dimensional
forward flight and to a three-dimensional flapping wing during hovering flight.
Concerning the flexible airfoil in forward flight, a comparison is made with a
plunging airfoils, with additional rotation. The following conclusion are drawn:
1. The two-dimensional flexing has a comparable effect on the flow and forces as
rotation of the airfoil. With increasing flexing amplitude, the larger effective
angle of attack leads to the generation of negative drag or thrust. Besides,
the flexing shape of the airfoil is important. A quarter-cosine shaped flexing
results in significantly higher thrust, while lift decreases.
2. The flexing deformation was also applied to a three-dimensional flapping
wing in hovering flight. Two flexing directions were considered, spanwise and
chordwise flexing. While keeping the flexing amplitude constant, the chord-
wise flexing affects the flow significantly. For a chordwise flexing amplitude
of A
f
= 0.2 the time-averaged lift and drag coefficients are significantly de-
creased and increased, respectively, such that the lift-to-drag decreases with
24.8%. The spanwise flexing does not show a comparable influence, only
-2.6%.
186 Conclusions and recommendations
9.6 Recommendations
The current thesis describes a comparison of different mesh deformation methods
and development of an improved method, based on radial basis function interpo-
lation. Additionally, these methods are used to solve the flow around two- and
three-dimensional flapping airfoils and wings under hovering and forward flight
conditions. The following recommendation can be made for further research.
• The efficiency of the radial basis function mesh motion can be further im-
proved by methods which selects only the necessary boundary points. The
size of the matrix, which need to be solved, can be significantly reduced.
Additionally, the implementation of parallel iterative solver techniques may
increase its efficiency even further.
• The current implementation of the radial basis function mesh motion in
OpenFOAM

should be generally implemented in parallel. The current
method is only able to address up to 4 processors, depending on the mesh
partitioning.
• Using the RBF mesh motion solver, the flow around multiple flapping foils
or wings can be investigated. The current thesis describes the flow fields in-
duced by a single two-dimensional foil or three-dimensional wing. If multiple
wings are modelled, i.e. a dragonfly, the flow will be more complex and the
force development may be affected to certain extent.
• Investigate the effects of transition and turbulence modelling on the flow
fields. The current research considered scales at the laminar flow regime. It
would be interesting to study the effects of turbulence on the force develop-
ment and vortex dynamics induced by a flapping three-dimensional wing.
• Investigate in more detail the vortex wake synchronisation in three-dimen-
sional forward flight. Concerning two-dimensional forward flapping flight,
this thesis describes the vortex wake synchronisation of a flapping foil for
different kinematic parameters. It would be interesting to study the vor-
tex wake synchronisation of a three-dimensional flapping wing. The main
difficulty will be to define a proper framework to identify and quantify the
three-dimensional vortex structures to study the pattern formation.
• The effects of wing deformation could be investigated in more detail using
more complex flexing models. The current thesis describes a preliminary
investigation of the effects of pre-defined wing flexing. Further research could
introduce more degrees of freedom in the flexing model, possibly based on
real insects. This may lead to advanced wing shapes, which result in optimal
aerodynamic efficiency, compared to rigid wings.
• It would be interesting to use fluid-structure interaction methods to study
the influence of the flow on the wing shapes. The wing shape is deformed
9.6 Recommendations 187
by inertial and aerodynamic forces. It will be interesting to couple these
forces with the flow solver and the accompanying mesh motion method. The
main difficulty will be the coupling of flow and structure, since the scaled
density of a fruit fly wing is of similar order as the surrounding fluid. Such
a strongly coupled problem is very sensitive and a converged solution is not
an easy objective.
• When advanced wing motions are desired, the possibility to implement im-
mersed boundary methods needs to be explored. Using immersed boundary
methods it will become possible to model the clap-and-fling motion, when
two wings touches each other. Another application of the immersed bound-
ary method will be the use of multiple wings and bodies with extreme body
motion to simulate advanced manoeuvring.
APPENDIX A
Grid generation for flapping wings
To solve the flow around flapping wings using Computational Fluid Dynamics
(CFD), it is important to create high quality meshes. Three different tools are used
for that purpose. The first, blockMesh, is a mesh generation utility supplied with
OpenFOAM

. This utility creates a parametric mesh with grading and curved
edges. A practical use of blockMesh is limited to simple domains and geometries.
The second mesh generation software is Gambit, supplied with Fluent

, and is ca-
pable of generating high quality meshes around complex geometries. Nevertheless,
using Gambit for structured meshes can be difficult since defining an appropriate
grading can be cumbersome. The third grid generation tool is the commercial
GridPro

package. GridPro

generates a structured grid with grading using a
block strategy around a complex geometry. The user generates a topology and the
grid solver creates a structured block such that the resulting mesh quality is high.
A.1 Introduction
This appendix deals with the mesh generation in order to obtain a high quality
initial mesh to solve for the flow around a moving airfoil or wing. There are two
types of mesh generation, unstructured and structured. In principle, OpenFOAM

and Fluent

are capable of solving the discretised equations in an unstructured
way, so there are no problems using one of those types. Since the meshes need to
deform the initial mesh needs to be of very high quality, which can be more easily
obtained using a structured approach. Besides, the iterative solvers converge faster
on structured meshes. However, mesh generation around complex bodies is much
easier using unstructured meshing techniques, but since the present study deals
190 Grid generation for flapping wings
b1 b2 b3
b4
b5 b6 b7
b8
p1 p2
p3 p4
X
Y
Figure A.1 The topology used by the
blockMesh utility. A blockMesh topology con-
sists of points, lines and block. All need to be
specified manually, which can be cumbersome.
Figure A.2 A mesh generated by the
blockMesh utility. A mesh generated by
blockMesh, which can be obtained very fast,
but is limited to simplified geometries.
with simplified model airfoils and wings, it is decided to choose structured meshing
strategies. The meshes used in the present thesis were generated using one of the
earlier mentioned meshing tools: blockMesh, Gambit or GridPro

. These three
tools are described in sections A.2, A.2 and A.2, respectively.
A.2 BlockMesh
The open-source mesh generator, supplied with OpenFOAM

, is blockMesh. It is
an easy to use and robust utility, applicable for simplified cases. The computational
domain is specified by points, lines and blocks, see figure A.1. The connections be-
tween the points (p1−p4) are defined by lines, which can be curved. Additionally,
the domain consists of blocks (b1 − b8), specified by the lines, accordingly. The
relation between the points, lines and blocks need to be specified in a dictionary
file, within the OpenFOAM

case. The grading to define the mesh resolution can
be set on all topology lines. For a uniform grading the mesh around a block is
shown in figure A.2.
Since the manual creation of topology can be cumbersome, this mesh gener-
ation utility is limited to simple computational domains. Therefore, the current
thesis used blockMesh to generate the meshes around a two-dimensional block
for testing the mesh deformation methods (chapter 3). Additionally, the meshes
are generated, concerning the two-dimensional channel cases, used to test the
OpenFOAM

flow solver with vortex decay and convection (chapter 2).
A.3 Gambit 191
Figure A.3 Mesh around a circular cylinder, generated by Gambit. Meshes around complex
geometries is possible using Gambit, but clustering of cells is the main difficulty.
A.3 Gambit
The commercial flow solver Fluent

also provides the mesh generator Gambit.
Gambit is able to generate structured and unstructured meshes, the latter con-
taining only tetrahedral cells. It was already explained that the current thesis
makes only use of structured meshes, Gambit was used to generate these meshes
using hexahedral cells. Within Gambit, a strong graphical user interface is avail-
able, such that meshes around complex CAD designed geometries are possible.
However, in order to generate a structured mesh with appropriate cell clustering
near the body at regions with large geometric gradients, it is invincible to result
with high mesh resolution throughout the computational domain, see figure A.3.
It is shown that a grid is generated using multiple blocks, generated manually,
in order to maintain high mesh quality. These multiple blocks, lead to an excess
of cells in regions where these are not necessary. This can be solved by gener-
ating more blocks, which is not straightforward. Gambit is used to generate the
two-dimensional meshes, which are used to test both Fluent

and OpenFOAM

using stationary and plunging cylinder flows. The next grid generator, GridPro

,
is capable to generate the multiple blocks automatically, which will improve the
mesh quality considerably, especially for three-dimensional cases.
A.4 GridPro

GridPro

is an advanced commercial multi-block structured mesh generator. The
usage is different compared to other mesh generators, because the multiple blocks
are automatically generated to optimise mesh quality. Figure A.4 shows the topol-
ogy, which needs to be generated, this approach is different compared to the other
grid generators, blockMesh and Gambit. In figure A.4, the inner red squared
block is snapped to the adjacent boundary Γ
b
, and all other red squared regions
192 Grid generation for flapping wings
Γ
b
X
Y
Figure A.4 GridPro

topology used to
generate the (block-) structured mesh.
The red blocks indicate extra topology, ac-
cording to which GridPro

generated the mul-
tiple blocks. These blocks are shaped such
that the mesh quality is optimised.
Figure A.5 Grid around a circular
cylinder, generated by GridPro

. The
grid is generated by GridPro

according to
the user-defined mesh topology, such that the
mesh quality is optimised.
represent structured grid blocks. An illustration of the resulting grid is shown in
figure A.5. Besides the high quality meshes, GridPro

is easy to use for complex
geometries. Additionally, for three-dimensional cases, it is very important to put
the multiple blocks in an optimal way such that the cell clustering is only present
in the regions of interest. Therefore, the current thesis uses GridPro

to generate
the three-dimensional meshes around a flapping wing, as illustrated in figure A.6.
A.5 Conclusions
Three different mesh generators have been described, blockMesh, Gambit and
GridPro

. The mesh generator, blockMesh, supplied with OpenFOAM

, is easy
to use for simplified problems. In the current thesis, blockMesh is used to gen-
erate the two-dimensional meshes, which were used to validate the flow solvers.
The second grid generator, Gambit, has more capabilities and can be used to
generated meshes around complex geometries. This mesh generator is used for
two-dimensional flows around stationary and plunging cylinders. Using the third
grid generator, GridPro

, a user-defined topology needs to be specified to optimise
the multiple blocks for the structured mesh generation. This topology procedure
is versatile, but different compared to the other packages. GridPro

is used for
the three-dimensional simulations around flapping wings.
A.5 Conclusions 193
(a) 3D GridPro

mesh
(b) 3D GridPro

mesh, close-up at t = 0T (c) 3D GridPro

mesh, close-up at t = 0.5T
Figure A.6 Three-dimensional grid around a wing, generated by GridPro

. (a) shows the
full computational domain which is used for mesh generation using GridPro

. A close-up of the mesh
near the three-dimensional wing is shown for two different time instances in (b) and (c), respectively
t = 0T and t = 0.5T.
APPENDIX B
Flow solver settings
B.1 Introduction
Within the current thesis, two different flow solvers have been used, the open-
source OpenFOAM

and the commercial package Fluent

. For both flow solvers,
the settings are described in this appendix. Section B.2 deals with Fluent

,
whereas section B.3 describes the settings used in OpenFOAM

.
B.2 Fluent

solver settings
Fluent

is a finite volume based CFD solver. This section deals with the dif-
ferent flow solver settings, that are necessary to reproduce the results from this
thesis. Fluent

is used for the two-dimensional hovering simulations described in
chapter 5.
Solver Segregated
Space 2D/3D
Time first-order implicit
Velocity formulation Absolute
Gradient option Cell-based
Viscous model laminar
Accuracy double precision
Table B.1 Solver settings
196 Flow solver settings
Pressure second-order
Pressure-Velocity coupling PISO
Equations second-order Upwind
Table B.2 Discretisation settings
Smoothing: Spring constant 0.1
Boundary node relaxation 0.3
Convergence tolerance 0.01
Number of iterations 20
Re-meshing: min. cell volume 1.73e-7 - 3e-7 (grid dependent)
max. cell volume 0.488 - 1.39 (grid dependent)
max. cell skewness 0.4
Table B.3 Dynamic mesh settings
In order to solve the flow, the user needs to specify which models are used, see
table B.1. The laminar viscous model is used which is quite misleading. No turbu-
lence model is used so in fact a Direct Numerical Simulation (DNS) is performed
using the Navier-Stokes equations. For the highest accuracy the ‘double precision’
version is used. According to (Lentink, 2003, Bos et al., 2008) this high accuracy
is necessary since insect flow might be very sensitive to initial conditions, under
certain circumstances.
The moving wings were studied using the ‘dynamic mesh’ module. Up to now
Fluent

is only capable of using this ‘dynamic mesh’ module in combination with
the first-order implicit time integration. The validation cases without mesh motion
uses second-order time integration. The multi-grid settings worked fine at default
settings. Table B.2 shows the discretisation settings. For space discretisation a
second-order upwind scheme is used together with standard PISO scheme for the
pressure-velocity coupling.
The ‘dynamic mesh’ parameters are shown in table B.3. The mesh is moving
using two methods, smoothing and re-meshing. Re-meshing means a complete
examination of the mesh and adaptation of the nodes where needed. Smoothing
on the other hand holds the nodes together in such a way that they do not move
arbitrarily in any direction, but stay together in a way. Re-meshing is defined
by the maximal and minimal cell volumes. These values are bases on the grid
in that case. The maximal skewness is needed in order to keep the mesh quality
within acceptable range. A value of 0.4 turned out to give satisfying results. The
smoothing performs an iteration to smooth the mesh when it is updated. The
smoothness is given by a spring constant which holds the nodes together and a
boundary node relaxation which gives some freedom to boundary nodes to move.
Concerning unsteady cases, the solution is varying in time until the residuals
B.3 OpenFOAM

solver settings 197
reach a sufficiently small value. These values are convergence criterion’s and can
be changed by the user. In this study the solution was considered converged as
the residuals reached a value of 1 · 10
−4
for every component. Per time-step a
fixed number of 20 iterations was needed to converge in case of using the PISO
pressure-velocity coupling in case of the validation study. During the flapping wing
simulations a fixed number of 10 iterations was used. Furthermore the complete
solution is written to hard-disk several times per flapping period; the lift and drag
histories every time-step.
B.3 OpenFOAM

solver settings
Since Fluent

could not cope with the extreme three-dimensional mesh deforma-
tion, OpenFOAM

was explored and provided good results. OpenFOAM

(Open
Field Operation And Manipulation) is a C++ toolbox for the customisation and
extension of numerical solvers for continuum mechanics problems, including Com-
putational Fluid Dynamics (CFD). It is produced by OpenCFD

Ltd. and is
freely available and open source, licensed under the GNU General Public Licence.
Since the source-code is fully accessible, different flow solvers and utilities are
developed. For example, the unsteady Navier-Stokes equations were solved, using
mesh motion techniques, with icoDyMFoam or icoDyMFoamRBF, the latter using ra-
dial basis function interpolation for mesh deformation. Utilities like setTaylorVortices
or meshQuality write respectively the velocities for a Taylor vortex initial con-
dition or the quality of the mesh (skewness and non-orthogonality). One of the
main strength’s of OpenFOAM

is the intuitive way of programming. Briefly, two
different illustrations are given, from the solver icoFoam and utility totalEnergy.
The following piece of code is taken from icoDyMFoam, which solves the unsteady
incompressible Navier-Stokes equations:
1 for (runTime++; !runTime.end(); runTime++)
2 {
3 Info<< "Time = " << runTime.timeName() << nl << endl;
4
5 fvVectorMatrix UEqn
6 (
7 fvm::ddt(U)
8 + fvm::div(phi, U)
9 - fvm::laplacian(nu, U)
10 );
11
12 solve(UEqn == -fvc::grad(p));
13
14 runTime.write();
15 }
Line 1 specifies that the equations are solved for all time-steps until the specified
end time is reached, runTime.end(). Line 5-9 defines the implicit part of the
governing equation, which is the unsteady incompressible Navier-Stokes equation.
The implicit discretisation is performed by the fvm class such that the total system
of equations is properly constructed from the temporal, convection and diffusion
198 Flow solver settings
terms, i.e. ddt and div are explicit functions to generate the matrix system for
a given velocity, pressure and flux field. The explicit fvc is used, on line 12,
to equate the system with the source terms and the complete system is solved
using a chosen iterative solver. This piece of code needs to be placed within a
pressure-velocity coupling loop.
The next code example concerns the utility totalEnergy, which calculates the
total energy, integrated over the complete domain:
1 void Foam::calc(const argList& args, const Time& runTime, const fvMesh& mesh)
2 {
3 IOobject Uheader
4 (
5 "U",
6 runTime.timeName(),
7 mesh,
8 IOobject::MUST_READ
9 );
10
11 if (Uheader.headerOk())
12 {
13 Info<< " Reading U" << endl;
14 volVectorField U(Uheader, mesh);
15
16 Info << " Calculating totalEnergy" << endl;
17 dimensionedScalar totalEnergy = fvc::domainIntegrate(0.5*magSqr(U));
18 Info << " Total energy = " << totalEnergy.value() << endl;
19 }
20 else
21 {
22 Info << " No U" << endl;
23 }
24 }
Here, from line 1-24, the time-loop is defined, calculating the total energy for every
time instance available. Line 3-9 reads the velocity field from a given solution, at
the specific time directory. The actual calculation is performed at line 17, where
fvc::domainIntegrate is an implicit function, calculating the sum over all finite-
volume cells in the domain of 0.5*magSqr(U), which is equivalent to 0.5|u|
2
. The
statement Info is a templated function, which is able to return strings, scalar
values and tensor fields back to the screen. These two illustrations are only two
examples. OpenFOAM

comes with a wide variety of solvers, utilities and tutorial
cases. If there is need for a specific application, the uses should take a look in the
source-code of OpenFOAM

to find similar pieces of code.
All flow solvers that are developed for this thesis, are based on icoDyMFoam,
slightly extended to use force output modifications, or to make use of modified
mesh motion solvers. icoDyMFoam solves the unsteady incompressible laminar
Navier-Stokes equations for a Newtonian fluid. Therefore, the different terms
of this equation need to be discretised accordingly, e.g. diffusion, convection.
Table B.4 shows the schemes that were used throughout this thesis. The convection
scheme was varied for validation purposes, but the Van Leer scheme was used for
the majority of numerical simulations.
In order to solve discretised governing equations, an iterative solver is used.
Three different solvers were specified for the pressure equation, velocity equation
B.3 OpenFOAM

solver settings 199
Description Code Differencing scheme
General interpolation - second-order linear
Temporal discretisation ddt(U) second-order backward
Gradient discretisation div(phi,U) second-order linear
Diffusion discretisation laplacian(nu,U) second-order linear
Convection discretisation grad(p) Gamma, SuperBee, Koren,
Van Leer or linear
Table B.4 Differencing methods for different terms in the transport equation. The tem-
poral, gradient, diffusion and convection term, present in the general transport equations needs to be
discretised properly. To minimise temporal errors all chosen schemes are of second-order, see (Weller
et al., 1998, Jasak, 1996, Jasak et al., 2004).
and mesh motion, respectively. Table B.5 shows the chosen solvers, combined with
the convergence criterion. For solving the equations for pressure and velocity, the
PISO coupling. Every time-step PISO evaluates an initial u and p, performs
multiple corrections (commonly twice in this thesis) until a convergence criterion
is met. An iterative method like PISO can be accelerated by Krylov subspace
methods (Saad, 2003). This means that the matrix A of the system Ay = b to be
solved is split as: A = M − N. M is used to pre-condition the problem, which
means that Ay = b is replaced by its pre-conditioned counter-part:
M
−1
Ay = M
−1
b. (B.1)
In this thesis, Incomplete Cholesky decomposition is used to pre-condition the
system as follows:
L
−1
AL
−T
˜ y = L
−1
b, y = L
−T
˜ y , (B.2)
where LL
T
is an Incomplete Cholesky decomposition of A and L
−T
= (L
T
)
−1
.
When Incomplete Cholesky (IC) decomposition is applied in combination with
Conjugate Gradient (CG), this is another Krylov subspace method for linear sys-
tems Ay = b with a symmetric self-adjoint positive definite matrix A. Combining
Term iterative solver convergence criteria
pressure, p PCG with DIC precond 10
−6
velocity, u PBiCG with DILU precond 10
−5
mesh motion PCG with DIC precond 10
−8
Table B.5 Iterative solvers for the different equations. The pressure and mesh motion equa-
tions are solved using the pre-conditioned conjugate gradient (PCG) solver with an diagonal incomplete
Choleski (DIC) pre-conditioner. The pressure-velocity coupling equation is solved using the asymmetric
solver pre-conditioned Bi-Stab conjugate gradient (PBiCG), with a diagonal incomplete LU decompo-
sition (DILU) pre-conditioner.
200 Flow solver settings
IC and CG results in a new method called ICCG. If the matrix A is not self-adjoint,
one can apply BiConjugate Gradient method (BCG). Using this in combination
with IC results in BiCG. For every variable calculated the linear solver methods
and their solution tolerances are listed in table B.5. A more elaborate description
of iterative solvers is beyond the scope of this thesis. For more information, please
consult (Wesseling, 2001, Hirsch, 1988, Ferziger & Peric, 2002, Jasak et al., 2007).
Bibliography
Anderson Jr., J. D. (1991), Fundamentals of Aerodynamics, second edn,
McGraw-Hill Inc.
Aono, H., Liang, F. & Liu, H. (2008), ‘Near- and far-field aerodynamics in
insect hovering flight: an integrated computational study’, The Journal of Exper-
imental Biology 211, 239–257.
Baruh, H. (1999), Analytical Dynamics, McGraw-Hill Inc.
Batina, J. T. (1990), ‘Unsteady euler airfoil solutions using unstructured dy-
namic meshes’, AIAA Journal 28 (8), 1381–1388.
Beckert, A. & Wendland, H. (2001), ‘Multivariate interpolation for fluid-
structure-interaction problems using radial basis functions’, Aerospace Science and
Technology 5(2), 125–134.
Birch & Dickinson, M. H. (2003), ‘The influence of wing-wake interactions
on the production of aerodynamic forces in flapping flight.’, The Journal of Ex-
perimental Biology 206, 2257–2272.
Birch, J. M. & Dickinson, M. H. (2001), ‘Spanwise flow and the attachment
of the leading-edge vortex on insect wings’, Nature 412, 729–733.
Birch, J. M., Dickson, W. B. & Dickinson, M. H. (2004), ‘Force production
and flow structure of the leading-edge vortex on flapping wings at high and low
reynolds numbers.’, The Journal of Experimental Biology 207, 1063–1072.
Blom, F. J. (2000), ‘Considerations on the spring analogy’, International Jour-
nal for Numerical Methods in Fluids 32, 647–669.
202 Bibliography
Blondeaux, P., Fornarelli, F. & Guglielmini, L. (2005a), Vortex struc-
tures generated by a finite-span oscillating foil, in ‘43rd AIAA Aerospace Sciences
Meeting and Exhibit, Reno’, 2005-84.
Blondeaux, P., Fornarelli, F., Guglielmini, L., Triantafyllou, M. S. &
Verzicco, R. (2005b), ‘Numerical experiments on flapping foils mimicking fish-
like locomotion’, Physics of Fluids 17, 113601–12.
Boer de, A., van der Schoot, M. S. & Bijl, H. (2007), ‘Mesh deformation
based on radial basis function interpolation’, Computers & Structures 85, 784–795.
Bomphrey, R. J., Lawson, N. J., K., T. G. & Thomas, A. L. R. (2006),
‘Application of digital particle image velocimetry to insect aerodynamics: mea-
surement of the leading-edge vortex and near wake of a hawkmoth’, Experiments
in Fluids 40, 546–554.
Bos, F. M., Lentink, D., van Oudheusden, B. W. & Bijl, H. (2008), ‘In-
fluence of wing kinematics on aerodynamic performance in hovering insect flight’,
Journal of Fluid Mechanics 594, 341–368.
Bos, F. M., van Oudheusden, B. W. & Bijl, H. (2010a), ‘Mesh deforma-
tion techniques for flapping flight’, Computer Methods in Applied Mechanics and
Engineering (Submitted).
Bos, F. M., van Oudheusden, B. W. & Bijl, H. (2010b), ‘Three-dimensional
vortical structures in flapping flight’, Journal of Fluid Mechanics (Submitted).
Brodsky, A. K. (1994), The Evolution of Insect Flight, Oxford University Press
Inc.
Buhmann, M. D. (2000), ‘Radial basis functions’, Acta Numerica 9, 1–38.
Carr, J. C., Beatson, R. K., McCallum, B. C., Fright, W. R., McLennan,
T. J. & Mitchell, T. J. (2003), Smooth surface reconstruction from noisy
range data, First International Conference on Computer Graphics and Interactive
Techniques.
Chakraborty, P., Balachandar, S. & Adrian, R. J. (2005), ‘On the rela-
tionships between local vortex identification schemes’, Journal of Fluid Mechanics
535, 189–214.
Combes, S. A. & Daniel, T. L. (2003a), ‘Flexural stiffness in insect wings i
& ii’, Journal of Experimental Biology 206, 2979–2997.
Combes, S. A. & Daniel, T. L. (2003b), ‘Into thin air: contributions of
aerodynamic and inertial-elastic forces to wing bending in the hawkmoth Manduca
sexta’, Journal of Experimental Biology 206, 2999–3006.
Bibliography 203
Degand, C. & Farhat, C. (2002), ‘A three-dimensional torsional spring analogy
method for unstructured dynamic meshes’, Computers & Structures 80, 305–316.
Dickinson, M. H. (1994), ‘The effects of wing rotation on unsteady aerody-
namic performance at low Reynolds number.’, The Journal of Experimental Biol-
ogy 192, 179–206.
Dickinson, M. H., Farley, C. T., Full, R. J., Koehl, M. A. R., Kram,
R. & Lehman, S. (2000), ‘How animals move: An integrated view’, Science
288, 100–106.
Dickinson, M. H. & G¨ otz, K. G. (1993), ‘Unsteady aerodynamic performance
of model wings at low reynolds numbers’, The Journal of Experimental Biology
174, 45–64.
Dickinson, M. H., Lehmann, F.-O. & Sane, S. P. (1999), ‘Wing rotation
and the aerodynamic basis of insect flight.’, Science 284, 1954–1960.
Dickson, W. B. & Dickinson, M. H. (2004), ‘The effect of advance ratio on
the aerodynamics of revolving wings’, Journal of Experimental Biology 207, 4269–
4281.
Donea, J. (1982), ‘An arbitrary lagrangian-eulerian finite element method for
transient fluid-structure interactions’, Computer Methods in Applied Mechanics
and Engineering pp. 689–723.
Dong, H., Mittal, R., Bozkurttas, M. & Najjar, F. (2005), Wake structure
and performance of finite aspect-ratio flapping foils, in ‘43rd AIAA Aerospace
Sciences Meeting and Exhibit, Reno’, 2005-81.
Dwight, R. P. (2004), Robust mesh deformation using the linear elasticity equa-
tions, in H. Deconinck & E. Dick, eds, ‘Computational Fluid Dynamics 2006’,
Springer.
Ellington, C. P. (1984), ‘The aerodynamics of hovering insect flight’, Phil.
Trans. Roy. Soc. London. B 305, 1–181.
Ellington, C. P., van den Berg, C., Willmott, A. P. & Thomas, A. L. R.
(1996), ‘Leading-edge vortices in insect flight’, Nature 384, 626–630.
Farhat, C., Degand, C., Koobus, B. & Lesoinne, M. (1998), ‘Torsional
springs for two-dimensional dynamic unstructured meshes’, Computer Methods in
Applied Mechanics and Engineering 163, 231–245.
Ferziger, J. H. & Peric, M. (2002), Computational Methods for Fluid Dynam-
ics, third edn, Springer-Verlag Berlin.
Fry, S. N., Sayaman, R. & Dickinson, M. H. (2003), ‘The aerodynamics of
free-flight maneuvers in Drosophila’, Science 300, 295–298.
204 Bibliography
Ginsberg, J. H. (1998), Advanced Engineering Mechanics, second edn, Cam-
bridge University Press.
Guglielmini, L. & Blondeaux, P. (2004), ‘Propulsive efficiency of oscillating
foils’, European Journal of Mechanics B/Fluids 23, 255–278.
Guilmineau, E. & Queutey, P. (2002), ‘A numerical simulation of vortex
shedding from an oscillating circular cylinder’, Journal of Fluids and Structures
16, 773–794.
Helenbrook, B. T. (2003), ‘Mesh deformation using the biharmonic operator’,
International Journal for Numerical Methods in Engineering 56 (7), 1007–1021.
Henderson, R. D. (1995), ‘Details of the drag curve near the onset of vortex
shedding’, Physics of Fluids 7 (9), 2102–2104.
Hirsch, C. (1988), Numerical Computation of Internal and External Flows, Vol-
ume I: Fundamentals of Numerical Discretization, John Wiley & Sons.
Hover, F. S., Haugsdal, Ø. & Triantafyllou, M. S. (2004), ‘Effect of angle
of attack profiles in flapping foil propulsion’, Journal of Fluids and Structures
19, 37–47.
Hu, D. L., Chan, B. & Bush, J. W. M. (2003), ‘The hydrodynamics of water
strider locomotion’, Nature 424, 663–666.
Hunt, J. C., Wray, A. A. & Moin, P. (1988), Eddies, stream, and convergence
zones in turbulent flows, in ‘Center for Turbulence Research Report’, pp. 193–208.
Isogai, K., Fujishiro, Saitoh, T., Yamamoto, M., Yamasaki, M. & Mat-
subara, M. (2004), ‘Unsteady three-dimensional viscous flow simulation of a
dragonfly hovering’, AIAA Journal 42, 2053–2059.
Issa, R. I. (1986), ‘Solution of the implicitly discretized fluid flow equation by
operator splitting’, Journal of Computational Physics 62, 40–65.
Jakobsson, S. & Amoignon, O. (2007), ‘Mesh deformation using radial ba-
sis functions for gradient-based aerodynamics shape optimization’, Computers &
Fluids 36, 1119–1136.
Jasak, H. (1996), Error analysis and estimation in the finite volume method with
applications to fluid flow, PhD thesis, Imperial College, University of London.
Jasak, H. (2009), Dynamic mesh handling in openfoam, in ‘47th AIAA
Aerospace Sciences Meeting, Orlando’, 2009-341.
Jasak, H., Jemcov, A. & Maruszewski, J. P. (2007), Pre-conditioned linear
solvers for large eddy simulations, in ‘CFD 2007 Conference’, CFD Society of
Canada, Toronto.
Bibliography 205
Jasak, H. & Tukovi´c, Z. (2004), ‘Automatic mesh motion for the unstructured
finite volume method’, Journal of Computational Physics (Submitted).
Jasak, H., Weller, H. G. & Gosman, A. D. (1999), ‘High resolution nvd
difference scheme for arbitrary unstructured meshes’, International Journal for
Numerical Methods in Fluids 31, 431–449.
Jasak, H., Weller, H. G. & Nordin, N. (2004), ‘In-cylinder CFD simulation
using a c++ object-oriented toolkit’, SAE Technical paper 2004-01-0110.
Jeong, J. & Hussain, F. (1995), ‘On the identification of a vortex’, Journal of
Fluid Mechanics 285, 69–94.
Johnson, A. A. & Tezduyar, T. E. (1994), ‘Mesh update strategies in par-
allel finite element computations of flow problems with moving boundaries and
interfaces’, Computer Methods in Applied Mechanics and Engineering 119, 73–94.
Juntasaro, V. & Marquis, A. J. (2004), ‘Comparative study of flux-limiters
based on must differencing scheme’, International Journal of Computational Fluid
Dynamics 18 (7), 569–576.
Juretic, F. (2004), Error analysis in Finite Volume CFD, PhD thesis, Imperial
College London.
Katz, J. (1981), ‘A discrete vortex method for non-steady separated flow over
an airfoil’, Journal of Fluid Mechanics 102, 315–328.
Knupp, P. M. (2003), ‘Algebraic mesh quality metrics for unstructured initial
meshes’, Finite Elements in Analysis and Design 39, 217–241.
Koren, B. (1993), ‘A robust upwind discretisation method for advection, dif-
fusion and source terms’, Numerical Methods for Advection-Diffusion Problems
p. 117.
Kuzmin, D. & Turek, S. (2004), ‘High-resolution fem-tvd schemes based on a
fully multidimensional flux limiter’, Journal of Computational Physics 198, 131–
158.
Lehmann, F.-O. (2004), ‘The mechanisms of lift enhancements in insect flight’,
Naturwissenschaften 91, 101–122.
Lehmann, F.-O., Sane, S. P. & Dickinson, M. (2005), ‘The aerodynamic
effect of wing-wing interaction in flapping insect wings’, The Journal of Experi-
mental Biology 208, 3075–3092.
Lentink, D. (2003), Influence of airfoil shape on performance in insect flight,
Master’s thesis, Delft University of Technology.
206 Bibliography
Lentink, D. (2008), Exploring the Biofluiddynamics of Swimming and Flight,
PhD thesis, Wageningen University.
Lentink, D. & Dickinson, M. H. (2009a), ‘Biofluiddynamic scaling of flap-
ping, spinning and translating fins and wings’, Journal of Experimental Biology
212, 2691–2704.
Lentink, D. & Dickinson, M. H. (2009b), ‘Rotational accelerations stabilize
leading edge vortices on revolving fly wings’, Journal of Experimental Biology
212, 2705–2719.
Lentink, D. & Gerritsma, M. (2003), Influence of airfoil shape on performance
in insect flight, in ‘43rd AIAA Fluid Dynamics Conference and Exhibit, Orlando’,
2003-3447.
Lentink, D., Muijres, F. T., Donker-Duyvis, F. J. & van Leeuwen, J. L.
(2008), ‘Vortex-wake interactions of a flapping foil that models animal swimming
and flight’, The Journal of Experimental Biology 211, 267–273.
Lesoinne, M. & Farhat, C. (1996), ‘Geometric conservation laws for flow prob-
lems with moving boundaries and deformable meshes, and their impact on aeroe-
lastic computations’, Computer Methods in Applied Mechanics and Engineering
134, 71–90.
Lewin, G. C. & Haj-Hariri, H. (2003), ‘Modelling thrust generation of a
two-dimensional heaving airfoil in a viscous flow’, Journal of Fluid Mechanics
492, 339–362.
Lighthill, M. J. (1969), ‘Hydromechanics of aquatic animal propulsion’, Annual
Review of Fluid Mechanics 1, 413–446.
Liu, H. & Kawachi, K. (1998), ‘A numerical study of insect flight’, Journal of
Computational Physics 146, 124–156.
L¨ohner, R. & Yang, C. (1996), ‘Improved ale mesh velocities for moving bod-
ies’, Communications in Numerical Methods in Engineering 12 (10), 599–608.
Luo, G. & Sun, M. (2005), ‘The effects of corrugation and wing planform on
the aerodynamic force production of sweeping model insect wings’, Acta Mech
Sinica 21, 531–541.
Lu, Y. & Shen, G. X. (2008), ‘Three-dimensional flow structures and evolution
of the leading-edge vortices on a flapping wing’, The Journal of Experimental
Biology 211, 1221–1230.
Maxworthy, T. (1979), ‘Experiments on the weis-fogh mechanism of lift gener-
ation by insects in hovering flight. part 1. dynamics of the ‘fling”, Journal of Fluid
Mechanics 93, 47–63.
Bibliography 207
Michelson, R. C. (2008), ‘Test and evaluation of fully autonomous micro air
vehicles’, ITEA Journal 29, 367–374.
Mittal, R. & Iaccarino, G. (2005), ‘Immersed boundary methods’, Annual
Review of Fluid Mechanics 37, 239–261.
Panton, R. L. (2005), Incompressible Flow, John Wiley & Sons.
Patankar, S. V. & Spalding, D. B. (1972), ‘A calculation procedure for heat,
mass and momentum transfer in three-dimensional parabolic flows’, International
Journal of Heat and Mass Transfer 15, 1787.
Pedro, G., Suleman, A. & Djilali, N. (2003), ‘A numerical study of the
propulsive efficiency of a flapping hydrofoil’, International Journal for Numerical
Methods in Fluids 42, 493–526.
Peskin, C. S. (2002), ‘The immersed boundary method’, Acta Numerica
pp. 479–517.
Poelma, C., Dickson, W. B. & Dickinson, M. H. (2006), ‘Time-resolved
reconstruction of the full velocity field around a dynamically-scaled flapping wing’,
Experiments in Fluids 41 (2), 1–13.
Potsdam, M. A. & Guruswamy, G. P. (2001), A parallel multiblock mesh
movement scheme for complex aeroelastic applications, Technical report, AIAA-
2001-0716.
Ramamurti, R. & Sandberg, W. C. (2002), ‘A three-dimensional compu-
tational study of the aerodynamic mechanisms of insect flight’, The Journal of
Experimental Biology 205, 1507–1518.
Rendall, T. C. S. & Allen, C. B. (2008a), Efficient mesh motion tech-
niques using radial basis functions with data reduction algorithms, in ‘46th AIAA
Aerospace Sciences Meeting and Exhibit’.
Rendall, T. C. S. & Allen, C. B. (2008b), ‘Improved radial basis function
fluid-structure coupling via efficient localised implementation’, International Jour-
nal for Numerical Methods in Engineering 78 (10), 1188–1208.
Rendall, T. C. S. & Allen, C. B. (2008c), ‘Unified fluid-structure interpo-
lation and mesh motion using radial basis functions’, International Journal for
Numerical Methods in Engineering 74 (10), 1519–1559.
Rhie, C. M. & Chow, W. L. (1983), ‘Numerical study of the turbulent flow
past an airfoil with trailing-edge separation’, AIAA Journal 21, 1525–1532.
Roe, P. L. (1986), ‘Characteristic-based schemes for the euler equations’, Annual
Review of Fluid Mechanics 18, 337.
208 Bibliography
Ruijgrok, G. J. J. (1994), Elements of airplane performance, Delft University
Press.
Saad, Y. (2003), Iterative Methods for Sparse Linear Systems, Society for In-
dustrial Mathematics.
Sane, S. P. & Dickinson, M. H. (2001), ‘The control of flight force by a
flapping wing: lift and drag production’, The Journal of Experimental Biology
204, 2607–2626.
Sane, S. P. & Dickinson, M. H. (2002), ‘The aerodynamic effects of wing
rotation and a revised quasi-steady model of flapping flight’, The Journal of Ex-
perimental Biology 205, 1087–1096.
Shyy, W., Lian, Y., Tang, J., Liu, H., Trizila, P. & Stanford, B. (2008a),
Computational aerodynamics of low reynolds number plunging, pitching and flex-
ible wings for mav applications, in ‘46th AIAA Aerospace Sciences Meeting and
Exhibit’.
Shyy, W., Lian, Y., Tang, J., Viieru, D. & Liu, H. (2008b), Aerodynamics
of low Reynolds number flyers, Cambridge University Press.
Smith, M. J., Cesnik, C. E. S. & Hodges, D. H. (2000), ‘Evaluation of
some data transfer algorithms for noncontiguous meshes’, Journal of Aerospace
Engineering 13(2), 52–58.
Srygley, R. B. & Thomas, A. L. R. (2002), ‘Unconventional lift-generating
mechanisms in free-flying butterflies’, Nature 420, 660–664.
Stepniewski, W. Z. & Keys, C. N. (1984), Rotary-wing aerodynamics, Dover.
Sun, M. & Tang, J. (2002), ‘Unsteady aerodynamic force generation by a
model fruit fly wing in flapping motion’, The Journal of Experimental Biology
205, 55–70.
Sweby, P. K. (1984), ‘High resolution schemes using flux-limiters for hyperbolic
conservation laws’, SIAM Journal of Numerical Analysis 21, 995–1011.
Taylor, G. K., Nudds, R. L. & Thomas, A. L. R. (2003), ‘Flying and
swimming animals cruise at a strouhal number tuned for high power efficiency’,
Nature 425, 707–711.
Templin, R. J. (2000), ‘The spectrum of animal flight: insect to pterosaurs’,
Progress in Aerospace Sciences 36, 393–436.
Tennekes, H. & Lumley, J. L. (1972), A first course in turbulence, The MIT
Press.
Bibliography 209
Thaweewat, N., Bos, F. M., van Oudheusden, B. W. & Bijl, H. (2009),
Numerical study of vortex-wake interactions and performance of a two-dimensional
flapping foil, in ‘47th AIAA Aerospace Sciences Meeting, Orlando’, 2009-791.
Thomas, A. L. R., Taylor, G. K., Srygley, R. B., Nudds, R. L. & Bom-
phrey, R. J. (2004), ‘Dragonfly flight: free-flight and tethered flow visualizations
reveal a diverse array of unsteady lift-generating mechanisms, controlled primarily
via angle of attack’, The Journal of Experimental Biology 207, 4299–4323.
Triantafyllou, G. S., Triantafyllou, M. S. & Grosenbaugh, M. A. (1993),
‘Optimal thrust development in oscillating foils with application to fish propulsion’,
Journal of Fluids and Structures 7, 205–224.
Tukovi´c, Z. & Jasak, H. (2007), ‘Updated lagrangian finite volume solver for
large deformation dynamic response of elastic body’, Transactions of FAMENA
31, 1.
Usherwood, J. R. & Ellington, C. P. (2002), ‘The aerodynamics of revolving
wings I-II’, Journal of Experimental Biology 205, 1547–1576.
Van Den Berg, C. & Ellington, C. P. (1997), ‘The three-dimensional leading-
edge vortex of a ‘hovering’ model hawkmoth’, Phil. Trans. Roy. Soc. Lond. B Biol.
Sci. 352, 329–340.
Van Leer, B. (1979), ‘Towards the ultimate conservative difference scheme v.
a second order sequel to godunov’s method’, Journal of Computational Physics
32, 101.
Wang, Z. J. (2000a), ‘Two dimensional mechanism for insect hovering’, Physical
Review Letters 85(10), 2216–2219.
Wang, Z. J. (2000b), ‘Vortex shedding and frequency selection in flapping flight’,
Journal of Fluid Mechanics 410, 323–341.
Wang, Z. J. (2005), ‘Dissecting insect flight’, Annual Review of Fluid Mechanics
37, 183–210.
Wang, Z. J. (2008), ‘Aerodynamic efficiency of flapping flight: analysis of a
two-stroke model’, Journal of Experimental Biology 211, 234–238.
Wang, Z. J., Birch, M. B. & Dickinson, M. H. (2004), ‘Unsteady forces
and flows in low reynolds number hovering flight: two-dimensional computations
vs robotic wing experiments’, The Journal of Experimental Biology 207, 461–474.
Wang, Z. J. & Przekwas, A. J. (1994), Unsteady flow computation using
moving grid with mesh enrichment, Technical report, AIAA-94-0285.
210 Bibliography
Weish-Fogh, T. & Jensen, M. (1956), ‘Biology and physics of locust flight. i.
basic principles of insect flight: a critical review’, Phil. Trans. Roy. Soc. London
239 (667), 415–458.
Weller, H. G., Tabor, G., Jasak, H. & Fureby, C. (1998), ‘A tensorial ap-
proach to computational continuum mechanics using object-oriented techniques’,
Computers in Physics.
Wendland, H. (1996), Konstruktion und untersuchung radialer basisfunktionen
mit kompaktem tr¨ager, Technical report.
Wendland, H. (1998), ‘Error estimates for interpolation by compactly sup-
ported radial basis functions of minimal degree’, Journal of Approximation Theory
93, 258–272.
Wendland, H. (1999), ‘On the smoothness of positive definite and radial func-
tions’, Journal of Computational and Applied Mathematics 101, 177–188.
Wesseling, P. (2001), Principles of Computational Fluid Dynamics, Springer-
Verlag Berlin.
White, F. M. (1991), Viscous Fluid Flow, second edn, McGraw-Hill Inc.
Williamson, C. H. K. (1988), ‘Defining a universal and continuous strouhal-
reynolds number relationship for the laminar vortex shedding of a circular cylin-
der’, Physics of Fluids 31(10), 2742–2744.
Williamson, C. H. K. (1995), Fluid Vortices, Vol. 30 of Fluid Mechanics and
Its Applications, Kluwer Academic Publishers, chapter Vortex dynamics in the
wake of a cylinder.
Williamson, C. H. K. (1998), ‘A series in 1/

Re to represent the strouhal-
reynolds number relationship of the cylinder wake’, Journal of Fluids and Struc-
tures 12, 1073–1085.
Williamson, C. H. K. & Roshko, A. (1988), ‘Vortex formation in the wake
of an oscillating cylinder’, Journal of Experimental Biology 2, 355–381.
Willmott, A. P., Ellington, C. P. & Thomas, A. L. R. (1997), ‘Flow
visualization and unsteady aerodynamics in the flight of the hawkmoth, Manduca
Sexta’, Phil. Trans. Roy. Soc. London 305, 303–316.
Wood, R. J. (2008), ‘The first takeoff of a biologically inspired at-scale robotic
insect’, IEEE Trans. on Robotics 24, 341–347.
Young, J. & Lai, J. C. S. (2008), ‘Simulation and parameter variation of
flapping-wing motion based on dragonfly hovering’, AIAA Journal 46, 918–924.
Bibliography 211
Zhou, Y. C. & Wei, G. W. (2003), ‘High resolution conjugate filters for the
simulation of flows’, Journal of Computational Physics 189, 159–179.
Zuijlen van, A. H. (2006), Fluid-Structure Interaction Simulations, Efficient
Higher Order Time Integration of Partitioned Systems, PhD thesis, Delft Univer-
sity of Technology.
Zuo, D., Peng, S., Chen, W. & Zhang, W. (2007), ‘Numerical simulation
of flapping-wing insect hovering flight at unsteady flow’, International Journal for
Numerical Methods in Fluids 53, 1801–1817.
Samenvatting
Wetenschappers in de biologie en techniek zijn altijd al gefascineerd geweest door
de vlucht van insecten en vogels. Gedurende een lange tijd bleef het aerodynamis-
che mechanisme om de vlucht van insecten met flappende vleugels te verklaren,
een raadsel. Tot een paar decennia terug. Experimenten lieten zien dat flappende
vleugels een voorrand wervel veroorzaken, die aanzienlijke krachten veroorzaakt.
Het werd gevonden dat deze opgewekte krachten groter zijn dan bepaald met con-
ventionele vliegtuig aerodynamica. Flappende vleugels produceren een lift en een
vooruit stuwende kracht, zodat insecten en zelfs kleine vogels, zoals de kolibrie,
stil kunnen blijven hangen en extreme manoeuvres uit kunnen voeren. Dankzij
deze veelzijdigheid vormen insecten en kleine vogels een krachtige bron van inspi-
ratie voor de ontwikkeling van micro vliegtuigjes, kleine door de mens gemaakte
apparaatjes, inzetbaar voor ontdekkings- en verkenningsmissies.
Verscheidene experimentele en numerieke visualisaties zijn uitgevoerd om de
kennis van de stroming rond flappende vleugels te vergroten. Met deze kennis kun-
nen micro vliegtuigjes worden ontworpen en verbeterd. Het effect van de vleugel-
beweging op de stroming en de krachten wordt nog steeds niet volledig begrepen.
Wij hebben twee- en driedimensionale computersimulaties uitgevoerd, waarbij be-
langrijke parameters voor de vleugelbeweging systematisch zijn gevarieerd. Om
de grenslaag en het zog goed in beeld te brengen, is het belangrijk om de kwaliteit
van het rekenrooster dicht bij de vleugel te behouden, in het bijzonder als de ro-
tatiehoeken groot zijn. Daarom is het belangrijk om een nauwkeurige methode
te gebruiken om het rekenrooster te vervormen, die ook geschikt is voor grote
rotaties. Om in staat te zijn de stroming rond een flappende vleugel uit te reke-
nen, is het noodzakelijk om de techniek voor roostervervorming te optimaliseren.
Een belangrijk doel van dit proefschrift beschrijft een betrouwbare techniek voor
roostervervorming, in termen van nauwkeurigheid en effici¨entie. Deze method is
tevens gebruikt om de stroming rond flappende vleugels door te rekenen.
De stroming rond flappende vleugels, op een schaal die relevant is voor de vlucht
van insecten, is sterk in-stationair, viskeus en wordt beschreven door de onsamen-
214 Samenvatting
drukbare Navier-Stokes vergelijkingen. Verschillende dimensieloze getallen die het
stromingsgedrag karakteriseren, zijn beschreven, zoals het Strouhal en Reynolds
getal. Omdat de stroming bij het beschouwde Reynolds getal van Re = O(100)
zich laminair gedraagt, is er geen noodzaak om turbulentie te modelleren. Zo-
doende mogen onze simulaties voor de laminaire stroming als een Directe Nu-
merieke Simulatie (DNS) worden beschouwd.
Om de onsamendrukbare Navier-Stokes vergelijkingen op te lossen, is inten-
sief gebruik gemaakt van het commerci¨ele pakket Fluent

en de open-bron code
OpenFOAM

. Verschillende technieken voor roostervervorming zijn vergeleken.
Twee van deze methoden zijn gebaseerd op de Laplace vergelijking en een aangepaste
spanningsvergelijking. Beide methoden zijn erg effici¨ent, omdat bestaande iter-
atieve technieken kunnen worden gebruikt. Echter, de roosterkwaliteit is niet vol-
doende voor gevallen met grote vleugelrotaties, wat het geval is bij het simuleren
van de vlucht van insecten. Zodoende is een nieuwe rooster deformatie techniek
ge¨ımplementeerd, gebaseerd op de interpolatie van radiale basis functies.
Deze techniek om rekenroosters te vervormen is gebaseerd op puntverplaatsin-
gen, zodat de beweging van alle individuele interne roosterpunten wordt ge¨evalu-
eerd. Er is geen connectiviteit van het rooster nodig, zodat deze methode een-
voudig kan worden toegepast op ongestructureerde roosters. Om de effici¨entie van
deze methode te verhogen, wordt een vergroving van de bewegende randpunten
toegepast. Dit verkleint het stelsel van vergelijkingen aanzienlijk, waardoor de
snelheid van de radiale basis functie methode, behoorlijk wordt verhoogd.
Na een discussie van de stromingsvergelijkingen, de eindige volume discretisatie
in OpenFOAM

en de vergelijking van verschillende technieken voor roosterver-
vorming, beschrijven we de fysische en numerieke modellen. De onsamendrukbare
Navier-Stokes vergelijkingen zijn herschreven in een roterend referentie assens-
telsel, zodat dimensieloze getallen die gerelateerd zijn aan de vleugel beweging
zijn afgeleid. Een belangrijk getal is het Rossby getal, dat een representatie is van
de kromming van het pad dat de vleugel aflegt.
Ten eerste is een tweedimensionale studie uitgevoerd om de invloed van ver-
schillende vleugelbeweging modellen op de prestaties te onderzoeken. De con-
dities voor stilhangende vlucht zijn hiervoor gebruikt. De resultaten laten zien
dat de ‘zaagtand’ flap amplitude slechts een klein effect heeft op de gemiddelde
liftkracht, maar dat de weerstand aanzienlijk wordt be¨ınvloed. De tweede model
vereenvoudiging, de ‘trapezium’ vorm van de invalshoek, leidt tot de loslating van
de voorrand wervel tijdens de translatie fase. Dit leidt tot een verhoging van
de gemiddelde weerstand tijdens elke halve flap periode. De extra ‘bump’ van
de invalshoek, die aanwezig is bij de fruitvlieg vleugelbeweging, be¨ınvloedt de lift
niet beduidend. De laatste realistische vleugelbeweging karakteristiek, de deviatie,
heeft slechts een marginaal effect op de gemiddelde lift- en weerstandskrachten in
deze tweedimensionale studie. Desalniettemin verandert de effectieve invalshoek
dusdanig dat de deviatie leidt tot een gelijkmatiger verdeling van de krachten.
Naast de tweedimensionale stroming voor stilhangende vlucht is een vergeli-
jkbare studie uitgevoerd voor voorwaarts flappende vlucht. Een numeriek model
215
voor een tweedimensionale stroming is gebruikt om het effect van de vleugelbe-
weging op de werveldynamica te onderzoeken voor een variatie van dimensieloze
golflengte, amplitude van de invalshoek en de hoek van het vlak waarin de vleugel
beweegt. Zowel translerende als roterende bewegingen zijn beschreven met simpele
harmonische functies, welke nuttig zijn om de parameter ruimte te onderzoeken,
ondanks de model vereenvoudigingen. Optimale voortstuwing met een flappend
vleugelprofiel bestaat voor elke variabele, zodat de aerodynamica waarschijnlijk
een range van wenselijke operationele condities selecteert. De condities voor op-
timale voortstuwing liggen in een synchronisatie regio waarin een periodieke stro-
ming bestaat.
Voorts zijn verschillende resultaten beschreven die relevant zijn voor de dried-
imensionale stroming rond een flappende vleugel. Ten eerste is de stroming rond
een dynamisch geschaalde vleugel numeriek berekend voor verschillende inval-
shoeken om de ontwikkeling van de krachten en wervel dynamica te onderzoeken.
Daarnaast is het Rossby getal gevarieerd bij verschillende Reynolds getallen. Een
kleiner Rossby getal betekent een sterkere kromming van het pad dat de vleugel
aflegt, zodat de hoekversnelling ook hoger is. We laten zien dat een laag Rossby
getal gunstig is voor de stabiliteit van de voorrandwervel, zodat de liftkracht en
effici¨entie worden vergroot. Ten derde is de driedimensionale vleugelbeweging
gevarieerd door de vorm van de invalshoek te veranderen. Ook is een deviatie
toegepast, wat kan leiden tot een acht-vormige figuur. De deviatie kan leiden tot
een geleidelijke verdeling van de krachten. Tenslotte is de driedimensionale stro-
ming vergeleken met de tweedimensionale stroming voor een vleugel in voorwaarts
flappende vlucht.
Het laatste onderwerp dat beschreven wordt in dit proefschrift, is het effect van
vleugel vervorming. Een vooraf gedefinieerde vleugelvervorming is toegepast op
een tweedimensionaal translerend vleugelprofiel en een driedimensionale vleugel bij
condities voor stilhangende vlucht. Het vervormende vleugelprofiel in vooruit flap-
pende vlucht leidt tot vergelijkbare resultaten als bij een starre vleugel, uitgebreid
met rotatie.
De huidige simulaties hebben geleid tot meer inzicht in hoe de prestaties van
een flappende vleugel, representatief voor de vlucht van insecten en vogels, worden
be¨ınvloed door de vleugelbeweging en vervorming. Dit inzicht kan belangrijk zijn
voor het ontwerp en optimalisatie van micro vliegtuigjes.
Acknowledgements
This doctoral thesis presents the research that I have performed at the Aerody-
namics Group of the Faculty of Aerospace Engineering at the Delft University of
Technology. After obtaining my MSc degree (2005) in numerical studies of flap-
ping foils, Professor Hester Bijl provided me the opportunity to investigate the
subject even further. I really enjoyed diving deep into the numerical techniques of
mesh motion, but also to apply these methods to real physical problems related
to flapping wing aerodynamics. Thank you, Hester, for giving me the freedom
to shape my own research framework, providing me the opportunity to attend
international conferences and initiate several MSc projects.
Besides Hester Bijl, I would like to thank Bas van Oudheusden for his profes-
sional insights and dedication. You have been a pleasant supervisor. I would also
thank all (former) PhD colleagues for the pleasant and inspiring working environ-
ment. In particular, I would like to thank Alex Loeven en Peter Lucas for being
pleasant room mates and friends, having many scientific and general discussions.
Sander van Zuijlen provided lots of support concerning code development, thank
you for that.
I would like to thank the OpenFOAM community, for many stimulating dis-
cussions on the workshops, conferences and on-line. In particular, I would like to
thank Professor Hrvoje Jasak and Henry Weller for many interesting and enlight-
ening discussions. Hrvoje, thanks for your support and patience concerning my
programming skills. Additionally, I thank Dubravko Matijasevic for early imple-
mentation of mesh motion based on radial basis function interpolation.
Finally, I say thanks to all my friends from Naaldwijk, Delft and beyond, for
the many joyful times to relax, drink beer or whisky and discuss many irrelevant
things. Special recognition goes to my parents, brother and family-in-law for
their unconditional support, love and fun. Most importantly, I thank Marieke for
entering my life and having a lot of fun, together and with our beautiful son.
Frank Bos
Naaldwijk, January 2010
Curriculum Vitae
Frank Bos was born on March 17, 1980 in Naaldwijk, The Netherlands. He at-
tended secondary school at the Interconfessionele Scholengemeenschap, Het West-
land, in Naaldwijk from 1992 until he graduated the Atheneum in 1998. In 1998
he started his study at the Aerospace Engineering faculty of the Delft University
of Technology. He completed his Propedeuse year in 1999.
In order to obtain his Master of Science degree, he performed an internship
with a duration of 5 months in 2002 at the Queen’s University of Belfast, Aero-
nautical department in the United Kingdom. He numerically investigated the
interaction between induced vortical flow and a turbulent boundary layer. He
obtained his Master of Science degree at the Aerodynamics department in 2005,
entitled “Influence of wing kinematics on performance in insect flight, a numerical
investigation”, supervised by dr.ir. David Lentink, dr.ir. Bas van Oudheusden and
Prof.dr.ir.drs. Hester Bijl. While studying, he performed several jobs at the Delft
University of Technology, related to project management and supervising students.
In August 2005 he started his Ph.D. project in the Computational Aerodynam-
ics group supervised by Prof.dr.ir.drs. Hester Bijl. The results of this research on
the numerical simulations of flapping foils and wings, are presented in this thesis.
He implemented mesh deformation, based on radial basis functions, in the open-
source CFD code OpenFOAM

. In February 2010 he successfully defended this
thesis with accompanying propositions. He presented his work in several publica-
tions and conference presentations. Additionally, he initiated MSc projects and
supervised several students.
Email: frank.m.bos@gmail.com
Linked-in: http://www.linkedin.com/in/fmbos

Copyright c 2009 by F.M. Bos All rights reserved. No part of this material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording or by any other information storage and retrieval system, without written permission from the copyright owner. Printed by Ipskamp Drukkers B.V. in The Netherlands ISBN: 978-90-9025173-8 An electronic version of this thesis is available at http://repository.tudelft.nl

Numerical simulations of flapping foil and wing aerodynamics
Mesh deformation using radial basis functions

Proefschrift

ter verkrijging van de graad van doctor aan de Technische Universiteit Delft, op gezag van de Rector Magnificus Prof. ir. K.C.A.M. Luyben, voorzitter van het College voor Promoties, in het openbaar te verdedigen op woensdag 24 februari 2010 om 10:00 uur

door

Frank Martijn BOS
ingenieur luchtvaart en ruimtevaart geboren te Naaldwijk.

Dit proefschrift is goedgekeurd door de promotor: Prof. dr. ir. drs. H. Bijl

Copromotor: Dr. ir. B.W. van Oudheusden

Samenstelling promotiecommissie: Rector Magnificus, Prof. dr. ir. drs. H. Bijl Dr. ir. B.W. van Oudheusden Prof. dr. ir. P.G. Bakker Prof. dr. ir. B. Koren Prof. dr. H. Jasak Prof. dr. F-O. Lehmann Prof. dr. W. Shyy voorzitter Technische Universiteit Delft, promotor Technische Universiteit Delft, copromotor Technische Universiteit Delft Universiteit Leiden Centrum Wiskunde & Informatica Zagreb University University of Ulm The University of Michigan

This research was supported by The Netherlands Organisation for Scientific Research (NWO), NWO-ALW grant 814.02.019.

Voor mijn ouders .

.

hummingbirds. Because of this versatility. generates forces larger than obtained by using conventional aircraft aerodynamics. Several flow visualisation experiments and numerical simulations have been performed to improve the understanding of flapping wing aerodynamics in order to design and optimise Micro Air Vehicles. at the scale relevant to insect flight. the effects of wing kinematics on the flow and forces is still not fully understood. For a long time. which is able to cope with large mesh deformations. e.Summary Both biological and engineering scientists have always been intrigued by the flight of insects and birds. described by the unsteady incompressible Navier-Stokes equations. Experiments showed the presence of a vortex on top of the flapping wings. Different dimensionless numbers are discussed. The flow around flapping wings. i. especially at large rotations. but also to perform extreme manoeuvres. in terms of accuracy and efficiency. In order to capture the boundary layer and the near wake.e. it is important to maintain a high mesh quality near the moving wing. different mesh motion techniques are compared and improved. We performed two.and three-dimensional numerical simulations in order to systematically vary relevant parameters. small man-made flyer’s to use in exploration and surveillance. The overall goal of this part of the research is to develop a reliable mesh deformation technique. the aerodynamic mechanism behind flapping insect flight was a complete mystery. Re = O(100). Strouhal and Reynolds numbers. an accurate mesh motion technique is necessary. Flapping wings produce both lifting and propulsive forces such that it becomes possible for insects and smaller bird species. there is no need for additional turbulence modelling. may be treated as a . Therefore. to stay aloft and hover. insects and smaller birds are an inspiration for the development of flapping wing Micro Air Vehicles. Since the flow at the considered Reynolds number. In order to incorporate a flapping wing in our numerical model. characterising the flow. is highly unsteady and vortical. to solve the flow around flapping wings. until several decades ago. such that our simulations. However. is laminar. related to the wing motion and flow physics.g. assuming laminar flow.

the ‘trapezoidal’ angle of attack. Therefore. on hovering flight performance. which occur in insect flight. Two existing methods are assessed. with increasing complexity. which is found to have only a marginal effect on the mean lift and mean drag in this two-dimensional study. Additionally. solving the Laplace and a modified stress equation. a coarsening is applied to the set of moving boundary points. finite volume discretisation in OpenFOAM and the assessment of the mesh motion solvers. the commercial software Fluent and the open-source code OpenFOAM have been used extensively. The second model simplification. This decreases the size of the system of equations and associated computational effort considerably. which means that the displacement of all individual internal mesh points are evaluated. No mesh connectivity information is necessary. so that it can be applied to unstructured polyhedral meshes. a new mesh motion solver is implemented.iv Summary Direct Numerical Simulation (DNS). which is based on the interpolation of radial basis functions. and stroke plane angles. However. a numerical model for two-dimensional flow was used to investigate the effect of foil kinematics on the vortex dynamics around an ellipsoid foil subject to prescribed flapping motion over a range of dimensionless wavelengths. The most important number is the Rossby number. Furthermore. The other realistic kinematic feature is the deviation. which represents the wing stroke path curvature. caused the leading-edge vortex to separate during the translational phase. and updated accordingly. This led to an increase in mean drag during each half-stroke. In order to solve the unsteady incompressible Navier-Stokes equations. the physical and numerical modelling are described. However. The results show that the ‘sawtooth’ amplitude has a small effect on the mean lift but the mean drag is affected significantly. To increase its efficiency. This mesh motion solver is a point based method. First a two-dimensional study is performed to investigate the effects of different wing kinematic models. angle of attack amplitudes. the effective angle of attack is altered such that the deviation leads to levelling of the force distribution. Different mesh motion solvers are compared. Both plunging and rotating motions are prescribed by simple harmonic functions which are useful for exploring the parametric space despite the model simplicity. dimensionless amplitudes. Both methods are very efficient by using iterative solver techniques. The extra ‘bump’ in angle of attack as used by the fruit fly model is not affecting the mean lift to a large extent. The incompressible Navier-Stokes equations are rewritten in the rotating reference frame in order to identify dimensionless numbers related to the wing motion. such that only selected control points are used. based on a given boundary displacement. Optimal propulsion using flapping foil exists for each variable which implies that aerodynamics might select a range of preferable operating condition. After the discussion of the governing equations. these mesh motion solvers are not able to maintain high mesh quality at large rotation angles. The conditions that give optimal propulsion lie in the synchronisation region in which the flow is periodic. different results relevant to three-dimensional flapping wing aero- .

These results may be important to design and optimise Micro Air Vehicles. Concerning the flexible airfoil in forward flight. the flow around a dynamically scaled model wing is solved for different angles of attack in order to study the force development and vortex dynamics at small and large mid-stroke angle of attack. As in two-dimensional studies. The present simulations have led to important insight to understand the influence of wing kinematics and deformation on the aerodynamic performance. subjected to additional rotation. A varying Rossby number represents a variation in stroke path curvature and thus angular acceleration. First. which may cause a ‘figure-of-eight’ pattern. the three-dimensional wing kinematics is varied by changing the shape in angle of attack and by applying a deviation. the three-dimensional flow is compared with the two-dimensional studies performed on flapping forward flight. Additionally. are described. by changing the effective angle of attack. Therefore. leading to an increase in lift and efficiency. the deviation may influence the force distribution to a large extent. Finally.v dynamics. It is shown that a low Rossby number is beneficial for the stability of the leading-edge vortex. a pre-defined flexing deformation is applied to a plunging airfoil in two-dimensional forward flight and to a three-dimensional flapping wing in hovering flight. . the Rossby number is varied at different Reynolds numbers. Thirdly. Secondly. a preliminary investigation is performed to show the effect of wing flexing. a similar behaviour was observed as for a rigid plunging airfoil.

.

. . 2. . 2. .5. .5.2. . . .1 Face interpolation schemes . . . . . . . . . . methods . . . . . . . . . . . . . .7.4 Objectives and approach . . . . . . . . 2. . . . . . . . . 2. . . . . 2. .5 Discretisation of an incompressible momentum equation . . . .4 Measures of cell quality .6 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Diffusion term . . . . 2. . . . . . . . . .2 Convection term . . . . . . . . . . . . .4 Temporal term . . . . . . . .7. . . . . . . .1 Introduction . . . . . . . .2 The Navier-Stokes equations . .5.Contents Summary 1 Introduction 1. . . . . . . . . . . . . . . . . . . . . . . 2. . . . . . . . 2. . .2.1 Pressure equation and Pressure-Velocity coupling . . . . . . . 1. . . . . . . . . . . . . . . . . .2. . . . . . . . . . . . . .2 Physics of flapping flight . . . .2 Incompressible laminar flow simplifications . . . . . 2. . . . . . .3 Experimental and numerical 1. . . 2. . . . . . . . . . . 2. . . . . . . . . . . . . .7. . . . . . . . . . . . . . . 1. . . . . . .5. . . . . . 2. . 2. .3 Dimensionless numbers . . . . . . . .1 Constitutive relations . .7 Solution of the Navier-Stokes equations . . . . . . . . . . . . . 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Arbitrary Lagrangian Eulerian approach . . .3 Spatial and temporal discretisation . iii 1 1 2 7 9 13 13 16 17 18 18 19 20 21 23 23 24 25 26 27 28 29 30 2 Finite volume discretisation 2. . . . . .2 Procedure for solving the Navier-Stokes equations 2. . .

. . .10. . . . . . . . . . 3. 3.11 Conclusions . . . . . . .6 Conclusions . . . . . . . . .5 Improving computational efficiency . . . . 3. . . . . . . . .2. .4. . . . . . .3. . . 3. . . . . . . . . . . . . . . . . . .7 Conclusions . . . . . . . . .5. . . 4.1.1 Introduction .4 Dynamical scaling of flapping wings . . . .2. . . . . . 2. 2. . . . . .1 Boundary point coarsening and smoothing . . . . .2 Kinematic modelling . . . . . 4. . . . 5. . . . .2 Influence of kinematic modelling . 3. . . . . . . . . . . . . . . . . . . . . . .1 Laplace equation with variable diffusivity . . 4. . . . . . . .4. . . . . . . . .1 Wing shape and planform selection . . . 3. . . 4. . 4. . . . . . . . 3. . . . . . . . . 2. 4. . . . . . . . . . . . . . . . 3. . .4. . 4. . 5. . . .9 Numerical flow solvers . 3. . . . . . .2. . . .4 Comparison of mesh motion solvers . . . . . . . . . . . . . . . .4 Numerical implementation of the wing kinematics . . . . .2 Solid body rotation stress equation . . . . . . .2 Validation using cylinder flows . . . . . .2 Iterative techniques and parallel implementations . 2. . . . . . .2 Numerical simulation methods . . .1 2D vortex decay and convection . . . . . . . . . .2 Different mesh deformation techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . .1. . . . . . . . . Contents . . . . . . . .3 Flexing of a two-dimensional block . .2. . . . . . . . . . . . . . . .3 Radial basis function interpolation .viii 2. . . . . . . . . . . . . . . . . . . . 5 Influence of wing kinematics in two-dimensional hovering flight 5. . . . . 2. . . . . . . . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. .3 Modelling of active wing flexing . . . . . 5. . . . . .1 Flow solver and governing equations . . . . . . . . . .10 Code validation and verification . . . . . . . . . . . 4. . .10. . . . 32 33 34 35 39 46 49 50 52 53 54 55 58 59 59 62 64 66 67 69 70 73 73 75 78 79 80 82 83 86 88 89 92 93 94 94 94 96 96 97 3 Mesh deformation techniques for flapping flight 3. . . . . . . . . . .3. . . . . . . .3 Mesh quality measures . . . . .5. 4.2 Flapping of a three-dimensional wing . . . . . . . . . . . . . .and three-dimensional flows . . . . . . . . . . 3. . . . . 5. . . . . . . . . . . . . . . . . . .5 Computational domain and boundary conditions . . . . . . . . . .1 Similarity and discrepancy between two. . . . . . . . . .6 Definition of force and performance coefficients . . . . . . .2 Governing equations for flapping wings . . . . . . . .3 Wing shape and kinematic modelling . . . . . . . . . . . . . . . . . .2. . . . . . . . . . . . . . .2 Mesh generation and boundary conditions . . .1 Translation and rotation of a two-dimensional block 3.8 Swept volume calculation . . . . . . . . . . . 4 Physical and numerical modelling of flapping foils and wings 4. . . . . . . . . 3. . . .3. . . . . . . . . . . . . . . . . . . . . . 3. . . . . . . . . .3. . . . . . . . 4.

. . . 6. . .5 Flapping wings at low Reynolds numbers . . . 8. . . . . . . . . . . . . . . . .3 Force coefficients and performance . . . . . . .4 Vortex identification methods for flow visualisation . . . .Contents 5. . . . . . . 5. . . . . . . . .2. . .3 Flow solver accuracy . . . . . . . . .3.3 Conclusions . . . . . . . . . . . . . 6. . . . . . . . . . . . . . . 5. .3. . . 6. . . . . . . . . . . . .5. . . . . . .6 Flapping wings in forward flight .5 Discussion . . . .2 Wing flexing in three-dimensional hovering flight . .3. . . . . . . . . 7.5. . . . 8 Influence of wing deformation by flexing 8. 7. . . . . . . . . . . . . . . .2 Dynamical scaling of the wing model . . . . . 7. . . . . . . . . . . . . . 6. . . . . . . . . . . . . . . . 6. 5.5 Results and discussion . . . . . . . . . . . .2 Flapping foil parametrisation . . 7. . . . . . . . . .1 Airfoil flexing in two-dimensional forward flapping flight . . . . . . .2. . . . . .2 Influence of dimensionless amplitude .2 Simulation strategy and test matrix selection .5. . . . . . . . . . 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5. . . . . . . .1 Overall model comparison . . . . . 7. . .1 The angle of attack in flapping flight . . . . . .3 Influence of angle of attack amplitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Numerical model . . . . . . . . . . . . . 6.4 Influence of ‘trapezoidal’ angle of attack . . . . 6. . . . . . .4. . . . . . . 7. . .4 5. . . . . . . . . . .5. . . . . . . . . . . . . . . .5. . . . . .5. . . . 8. . . . 6. . .2. . . . .3 Influence of Reynolds number . . . . . . Conclusions . . . . . . . .5 Influence of deviation . . . . . . . . . . . . . . . 7.1 Introduction . . . . . . . . . . . .2 Influence of flapping stroke curvature . . .5. . . . . . . .1 Influence of dimensionless wavelength . . . . . 5.1 Modelling and parameter selection . . . . . . . . . . . . . . . . . . . . . . . . ix 99 99 100 101 102 103 105 105 109 118 121 122 122 124 125 126 126 127 129 130 130 132 135 135 137 138 141 143 145 148 149 149 152 158 160 163 166 171 171 175 177 5. . .2 Three-dimensional flapping wing simulations . . . . . . . . . . . . . . .3 Validation using harmonic wing kinematics Modelling insect wing kinematics . . . . . . . . . . . . . . . . 7. 5. . . . .7 Conclusions . . . . . . .4 Influence of stroke plane angle . . . . . . . . . . . . . . . . 7. . .6 Conclusions . . .3. . . . . . . . . . . . . . . flapping foil . . . . . . . . . 6. . .1 Insect wing selection and model parameters 5. 7. . . 6. . . . . . . . . . . . . . . . . .4 Different wing kinematic models . .5 6 Vortex wake interactions of a two-dimensional 6. . . . . .3 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . . . . . . . Results and Discussion . . . . . 7. . . . . .3 Force and performance indicators . . . . .4.5. .2 Kinematic features investigation . . . . . . . . . . . . . . . . . . . . . . . . . 7. . 7 Vortical structures in three-dimensional flapping flight 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5. . . .

. . . . . . . . . . . A. . . . .5 Preliminary conclusions on wing flexing . . . . . . .3 Conclusions on hovering flapping flight . . . . . . . Bibliography Samenvatting Acknowledgements Curriculum Vitae . . . . . . . . . . .2 BlockMesh . . . . . . . . . . . . . . . . . . . . . . . . . . A Grid generation for flapping wings A. . . . . . . . . . . . . . . . . . . A. . . . . . . . . . . . . . . . . . . . .2 Fluent solver settings . . . . 9. . . . . . . . . . . . . . 9.x 9 Conclusions and recommendations 9. . . . . . . . . . 9. . . . . . . . . .4 GridPro .4 Conclusions on forward flapping flight . . . . . Contents 179 179 180 181 181 182 184 184 184 185 186 189 189 190 191 191 192 195 195 195 197 201 213 217 219 . . . . . . . . .4. . .3 OpenFOAM solver settings . . . . . . .1 Introduction . . . . . . . . . . . . . B Flow solver settings B. . . . . . . . . . . . . . . . 9. . . . . . . . . 9. . . . . . . .2 Three-dimensional hovering . . . . . . . . . . .6 Recommendations .4. . . . . . . . . . . . . . . B. . . . . . . . . . . . . .1 Introduction . . . . . . . . .1 Two-dimensional forward flapping . . . . . . . . . . . . . . .1 Overall conclusions . . . . . . . . A. . . . . . . . . . . . . . . 9. . . . . .3 Gambit .3. . . . .3. . . . . . . . . . . . . . . . . B. . . . . . . . . . . . . . . . . . . . . .2 Conclusions on mesh motion techniques .2 Three-dimensional forward flapping 9. . . 9. . . . . . . . . . . . . .1 Two-dimensional hovering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. . . . . . . . . . . . . 9. .5 Conclusions . . . . . .

1956) flaps with about 200 times per second. insects adapted to leave the ground to take off into the thin air. since he was born 200 years ago. but also to perform extreme manoeuvres. e. the mainland bird species had travelled to the islands and on every different island it adapted to the differences in environmental circumstances. whereas a small insect.g. see . hummingbirds. 2000) which developed the skill of flight in order to migrate over large distances and to catch prey. such that it becomes possible for insects and even smaller bird species. Because of this versatility. inspired by his scientific observations during a voyage (1831-1836) around the world with his ship. This process has become known as natural selection. e. the Beagle. leading to a different flow behaviour. Flapping wings produce both lifting and propulsive forces.1 Motivation The year of writing.CHAPTER 1 Introduction 1. he published his world famous Origin of Species. At the Gal´pagos Archipelago. That book describes the natural selection. Birds and insects are both flapping their wings at different length scales. Birds are ancient descendants of feathered dinosaurs (Templin. to stay aloft and hover. tiny man-made flyer’s to use in exploration and surveillance. Darwin discovered slightly different bird a species living on the different islands. The larger the animal. insects and smaller birds are a major inspiration of study to develop Micro Air Vehicles (MAV).g. millions of years ago. which is also applicable to the early era of flight. At the age of 50. whereas. Long before the origin of dinosaurs and birds. the lower the need for flapping wings. he only knew one species on the mainland of South-America. a fruit fly (Weish-Fogh & Jensen. is known as the year of Charles Darwin (1809-1882). the Andean Condor only flaps when it looses height in the thermal winds. Apparently. 2009.

. birds. 2009). see figure 1. when performing wind tunnel experiments it is not straightforward to extract the force data.2. the leading-edge vortex in particular. the construction of models needs to be very precise. different aspects of flapping wing aerodynamics are discussed. detailed experiments (Ellington et al.and three-dimensional problems. 2002). In order to study this kind of flows. either directly or indirectly. Additionally.. insects. is briefly discussed.. which can be very costly as well.g. 2002. Large aeroplanes generate wingtip vortices. using Computational Fluid Dynamics (CFD). which can cause damage to a following aeroplane which encounters .1.3 deals with the different approaches for analysing the flow. Srygley & Thomas. 2008. Finally. To optimise the flight performance of MAV’s it is important to get a thorough understanding of the complex flow generated by its wings. Secondly.... Poelma et al. the flow field can not be visualised in much detail. as well as the influence of the wing kinematic modelling in two. from the flow visualisation obtained by Particle Image Velocimetry (PIV) (Poelma et al. O(100 − 1000). Lentink & Dickinson.2 Introduction figure 1. experiments and numerical methods. Even when the most advanced techniques are used. performing CFD provides a suitable framework. On the other hand. The vortex dynamics. Sun & Tang. in view of the need of precision equipment and wind tunnel facilities.2 briefly provides background information on the flow physics concerned. Section 1. 2002.2 Physics of flapping flight In order to illustrate the necessity and difficulties with solving and visualising the flow around flapping foils and wings. Digital Particle Image Velocimetry (DPIV). The present study deals with the development and improvement of computational techniques to solve the flow around flapping wings at low Reynolds numbers. 1. Vortex generation in flapping wing aerodynamics Vortex generation in nature is fairly common in flows induced by aeroplanes. the forces and flow visualisations are a direct result of the computations.4 describes the objectives and approach of the present study as well as the outline of this thesis. The flapping wings induce complicated vortical structures which influence the forces and performance characteristics in hovering and forward flight. One limitation of doing experiments is that they can be very expensive. 2009b) and numerical simulations (Wang et al. Sane & Dickinson. 1956. 2004. section 1. especially at smaller length scales (< 5 cm). but also by boats and trees. 1996. when performing numerical simulations. Since it is interesting to solve for the forces acting on a flapping wing in combination with the vortical structures within the near wake. 2006. especially due to the reflections and shadows of the moving wings. researchers performed flow visualisations (Weish-Fogh & Jensen. Bos et al. e. while section 1. 2006). Thaweewat et al.

generated by successive up and downstrokes. but this was not sufficient to explain the high lift generation of insects.2 Vortex induced force generation. (a) Wingtip vortex causes big disturbances in the wake. A water strider generates vortices with its long legs to create the necessary propulsion (Hu et al. (b) The U. patented Entomopter has four flapping wings powered by chemically-fuelled propulsion system (Michelson. 2008). fish and insects.S. (a) (b) (c) Figure 1. circulation and lift generation. It is a common story that flies could not fly according to conventional aircraft theory as developed by Lanchester (1907) and Prandtl (1914-1918). This mystery persisted until the discovery of the unsteady vortical flow field. 2003). (b) Entomopter. figure 1. alternating vortex rings were seen in the wake. Prandtl did develop a relation between the tip vortices. and especially the generation of the leading-edge vortex.g. which is able to perform a tethered take-off (Wood. e. bridges or struts in water. which is used by birds.2 Physics of flapping flight 3 (a) Wasp. (c) Delfly. limiting the time between two successive aeroplane approaches. (1997) performed smoke visualisation of the vortical flow patterns induced by a hawkmoth.2. figure 1. (b) Vortex generation in insect flight. At lower Reynolds numbers.. 2008). Another undesired effect of vortex generation is flow induced vibration of cables. which is especially interesting for intelligence and exploration. It was observed that the leading-edge vortex was stabilised by the radial flow moving out towards the wing tip. vortex generation provides possibilities to generate forces. (a) Flying insect scale robotic model.1.2 shows induced propulsive vortices generated by a water strider. Figure 1. (c) The Delfly Micro is camera equipped and is able to hover (designed and developed at Delft University of Technology). flapping MAV concepts can be used for hovering and low speed forward flight. this vortex. On the other hand. (c) Willmott et al. Additionally.1 Different flapping wing Micro Air Vehicle concepts. .

4 Introduction ¯ CL = 1. Figure 1. 2009b). Lentink et al. operating at Re = 110.and three-dimensional flapping influences the shear layer direction and flow accelerations. It was Ellington et al. depending on the kinematics. (1996) o who identified the presence of a leading-edge vortex (LEV) generated on top of the flapping wing. notably advance ratio and dimensionless flapping amplitude (Thaweewat et al. In flapping foil aerodynamics the vortices are shed and form either a periodic or chaotic wake pattern. Leading-edge vortex The potential benefit of vortices attached to the wing was already discussed by Maxworthy (1979) and Dickinson & G¨tz (1993). 2008).540 -3 -2 -1 (a) x c [−] 0 1 2 3 (b) Figure 1. In order to understand the physics of flapping wing aerodynamics.. (b) Three-dimensional leading-edge vortex generated by a flapping wing at Re = O(1000). (a) Two-dimensional illustration of the wing kinematics and the resulting force vector generated by the flapping airfoil at Re = O(100) (Bos et al. Wang . from (Bos et al..3(a) shows a two-dimensional illustration of the wing kinematics of a fruit fly..3 Forces and vortices in flapping wing aerodynamics. it is important to obtain insight in how an insect moves its wing. 2008). It is thought that the kinematics in two.. Influence of insect wing kinematics on forces The relevance of experiments and flow simulations of insect flight has been found to depend on how reliably true insect wing kinematics are reproduced. 2009. increasing the lift force to values much higher than predicted by conventional wing theory. The origin of the leading-edge vortex is the roll-up of shear layers. 2008). which will undoubtedly influence the development of the leading-edge vortex (Lentink & Dickinson. present in highly viscous flows. It appears that the leading-edge vortex is more stable around a three-dimensional flapping wing compared to two-dimensional flapping foil situations. The stability of the helical three-dimensional leading-edge vortex is still not yet fully understood and appears to heavily depend on the wing kinematics and Reynolds number. which is the case at low Reynolds numbers.

b) and Lentink & Gerritsma (2003) numerically investigated pure harmonic translational motion with respectively small and large amplitudes. rotation speed and angle of attack during translation are of great importance for the force development during each stroke. the wing maintains a constant velocity and angle of attack during most of the stroke. 1993).2 Physics of flapping flight 5 et al. may have on flight performance. Slightly more complex fruit fly kinematic models were used by Dickinson et al. (2004) showed that the angle of attack influences the flapping foil propulsion efficiency to a large extent. such that including rotation is essential.1. Besides lift enhancement at certain reduced frequencies. deviation. (1999) and Sane & Dickinson (2001) with their Robofly. angle of attack and the timing of the latter were explored. the results were compared with more realistic fruit fly kinematics obtained from the observation of free flying fruit flies. (2004) and Sane & Dickinson (2001) showed that the kinematic modelling significantly influences the mean force coefficients and its distribution. including wing rotation. Furthermore. different kinematic models have been employed to investigate the aerodynamic features of insect flight. Both studies emphasised the importance of angle of attack to influence the propulsive efficiency. In literature. This results in the typical ‘sawtooth’ displacement and ‘trapezoidal’ angle of attack pattern of the Robofly kinematic model. both the pure harmonic and the Robofly model. Instead of performing a parameter study within the scope of one kinematic model. Using these models. They varied rotational parameters and showed that axis-of-rotation. Additionally. Wang (2000a. Hover et al. In addition to the harmonic models with pure translation (Dickinson & G¨tz. Harmonic wing kinematics. rotational parameters were investigated by Dickinson o (1994). they found periodic and aperiodic flow solutions which are strongly related to the aerodynamic efficiency.. the effect of amplitude. Based on observation of true insect flight. where used by Pedro et al. Lentink & Gerritsma (2003) varied airfoil shape with amplitude and frequency fixed at values representative to real fruit flies. like parameter values and stroke patterns. The present thesis deals with different kinematic models from literature. It further emphasises the need to critically assess the influence of kinematic model simplifications. with a relatively strong linear and angular acceleration during stroke reversal. Lewin & Haj-Hariri (2003) performed a similar numerical study for heaving airfoils. This illustrates the appreciable effects which details of the wing kinematics. For example. (2003) and Guglielmini & Blondeaux (2004) in their numerical models to solve for forward flight. It was also shown that wing stroke models with only translational motion could not provide for realistic results. the objective of the present study is to compare the effect . but large amplitudes lead to an increase of lift by a factor of 5 compared to static forces generated by translating airfoils. They concluded that the airfoil geometry choice is of minor influence. in order to investigate their influence on the aerodynamic performance (Bos et al. 2008). Wang (2000a.b) varied flapping amplitude and frequency and showed that at a certain parameter selection the lift is clearly enhanced.

in case of advanced and symmetric rotation the forces were found to be similar in the two-dimensional simulations compared to the three-dimensional experiments.and three-dimensional approaches are sufficient to warrant that a reasonable approximation of insect flight can be obtained using a two-dimensional approach. Bos et al. One of the major (and partially unresolved) issues in modelling of insect flight and flapping wing propulsion. which is present in three dimensions. In a recent paper Wang et al. as was mentioned by Lighthill (1969). wing flexibility (fluid structure interaction) and Reynolds number effects. in relation to aerodynamic performance. of the available models as a whole. Additional important aspects are unsteady flow mechanisms.6 Introduction (a) (b) (c) Figure 1. 2009. (2005). (2004) confirmed that the similarities between two. First. depending on the kinematics a chaotic wake pattern may occur with unpredictable forces as the result.and three-dimensional flows To limit both the parametric space as well as the computational effort. The similarity between two. (2005b) concluded that two-dimensional studies over predict forces and performances since the energy-loss. is the possibly restrictive applicability of two-dimensional results to true insect flight.. 2008. Also. (2005b) numerically investigated the wake structure behind finite-span wings at low Reynolds numbers. Blondeaux et al. They observed three-dimensional vortical structures around flapping wings with low aspect ratio. (2004) compared three-dimensional Robofly results with two-dimensional numerical results. Dicko . Wang et al. is not resolved.and three-dimensional flow. 2004.. (a) Von K´rm´n vortex a a street behind a stationary cylinder at Re = 150. which is of great importance to both experiments and numerical simulations. one single and one vortex pair is generated each plunging period. This leads to better insights into the consequences of simplifications in kinematic modelling. many studies have been performed as two-dimensional simulations (Thaweewat et al.4 Vortex wakes generated by cylinders and flapping wings. Secondly it was observed that in both simulations and experiments the leading-edge vortex did not completely separate for amplitude-to-chord ratios between 3-5 (Dickinson & G¨tz. (b) Periodic vortex wake behind a plunging airfoil at Re = 110. (2005) and Blondeaux et al. two-dimensional analysis has often been applied to obtain insight into the aerodynamic effects of wing kinematics and geometry. 2003). Both Dong et al. 1993. Lewin & Haj-Hariri.. (c) Chaotic vortex wake behind a plunging airfoil at Re = 110. it may reveal the importance of certain specific features of the stroke pattern. Wang et al. Notwithstanding the possible discrepancy between two. Dong et al.

like delayed stall and wake capture enhance the lift force beyond values predicted by quasi-steady theory. and revealed the complex nature of insect flight aerodynamics. Drosophila Melanogaster.2. amplitudes and the very low Reynolds number involved (Re < 1000 for a large number of insects and Re ≈ 110 for the fruit fly. (1996) discovered that these lift increasing mechanisms are amplified by the generation of a leading-edge vortex (LEV). (1999) and Wang (2000b) applied the quasi-steady theory to compare with unsteady forces.b). the first part of this thesis makes extensive use of two-dimensional simulations. 1993. In the second part. hence revealing that important unsteady and vortical flow phenomena play a major role in insect flight. 1994. 1. The current research deals with amplitudes that are in this range. In this section. different methods were used to solve and visualise the flow around flapping insect wings. 1999. and with the above justification.3 Experimental and numerical methods In literature. from experimental methods to computational fluid dynamics simulations. as visualised by WeishFogh & Jensen (1956) using tethered locusts and by Willmott et al. The lift increasing effect of the leading-edge vortex strongly depends on the kinematics of the flapping wing (Dickinson et al. Lentink & Dickinson. from realistic fruit fly measurements to three-dimensional simulations using a representative model wing.3 Experimental and numerical methods 7 inson. 2001. 2000b. 2009a. In order to understand insect flight performance Dickinson et al. It was shown that this leadingedge vortex arises during the translational part of the wing motion rather than during the rotational flip between up and down stroke. in particular). Dickinson et al. 1994). various three-dimensional simulations were performed using limited variations wing kinematics. In several studies (Dickinson & G¨tz.. Experimental investigations and quasi steady theory Several experimental studies considered the flight performance of insects. Ellington (1984) indicated that the lift in insect flight is significantly higher than expected on the basis of quasi-steady aerodynamics. (2004) performed free flight and tethered experimental visualisations using butterflies and dragonflies to show the complicated vortical structures. (1997) using a hawkmoth (Manduca Sexta). Srygley & Thomas (2002) and Thomas et al. 2002. different methods will be briefly addressed. Sane & Dickinson.1. Dickinson. In view of the excessive computational expense required to perform accurate three-dimensional simulations. o 1999) it was confirmed that important aspects. The flow induced by the motion of insect wings is highly unsteady and vortical. The quasi-steady approach was revised by Sane & Dickinson (2002) to include ro- .. see figure 1. Ellington et al. More recently. This unsteady and vortical flow behaviour is a consequence of the high relative frequencies. Wang.

. 2007. or even complete re-meshing (Young & Lai. Several experimental studies have been performed with the aim of characterising the unsteady aerodynamics of insect flight. 2005). such as flapping wings. In order to cope with high rotation rates. dragonfly: Young & Lai (2008). Mittal & Iaccarino. Poelma et al. as in the manoeuvre clap-and-fling (Weish-Fogh & Jensen. it remains difficult to capture all the relevant details of the flow using only experimental techniques. An appealing approach. therefore. one can use either immersed boundary methods (Peskin. when two wings touch. 2006). one will undoubtedly need methods like overset. Numerical simulations Notwithstanding important advances in experimental techniques for non-intrusive flow field analysis. This confirms the restricted applicability of the quasi-steady theory due to lack of unsteady mechanisms like rotational lift and wake capture. Particle Image Velocimetry in particular (Bomphrey et al. the computational effort involved in three-dimensional studies is presently still extremely demanding. 2006. To perform numerical simulations around moving objects. In an immersed boundary method. Jasak.. in relation to specific insect geometries (moth: Liu & Kawachi (1998). the producer of the . 2009). which is an open-source framework to solve the Navier-Stokes equations on three-dimensional 1 OpenFOAM is a registered trade mark of OpenCFD OpenFOAM software. 2002. Nevertheless.8 Introduction tational effects but even then the results require further improvement. (2004)). immersed boundary or re-meshing techniques. Limited. an integrated computational study was performed by Aono et al. even not for fixed boundaries (Mittal & Iaccarino. 2008. Dickinson et al. Sun & Tang (2002). deforming mesh techniques (Boer de et al. 2005). 1956). The commonly used mesh motion techniques result in high quality meshes as long as the rotation of the moving boundaries is limited. Isogai et al.. A number of numerical studies on full three-dimensional configurations have been reported. (2008) who developed a code to incorporate two wings and a body using overset mesh techniques. is to supplement experiments with numerical flow simulations. mesh motion based on radial basis function (RBF) interpolation is implemented in this thesis and improved in terms of accuracy and efficiency.5. but the mean drag is underestimated. the moving boundary is projected on a fixed Cartesian background grid. it was chosen to assess and improve existing mesh motion techniques. Zuo et al. which is not allowed to deform.. fruit fly: Ramamurti & Sandberg (2002). Although. see figure 1. This modern mesh motion technique is incorporated in OpenFOAM 1 . the conservation of mass and momentum in current immersed boundary methods is not obvious. Together with the unavailability of an accurate flow solver with parallel support. 2007). (1999) investigated the flow around a flapping Robofly model which moves in oil to meet the same flow conditions as the real fruit fly encounters (reproduction of Reynolds number in particular). According to Sane & Dickinson (2001) the mean lift is well predicted by quasi-steady theory. Besides interpolation issues.

whereas the Eulerian part takes care of the fluid flow through the mesh. while (b) shows the mesh motion obtained by using radial basis function interpolation. where the mesh is fixed to the fluid or material. 1.4 Objectives and approach Flapping flight aerodynamics is governed by many parameters. Reynolds number. Two illustrations of mesh motion solutions.4 Objectives and approach 9 (a) (b) Figure 1. a fixed mesh becomes inconvenient. The Lagrangian contribution allows the mesh to move and deform according to the boundary motion. when the flow domain moves or deforms in time due to a moving boundary.1. However. Arbitrary Lagrangian-Eulerian formulation The governing equations to solve the flow are generally discretised using the Eulerian description. If the material or fluid deforms. (a) shows a Laplacian mesh motion. This method is commonly used to discretise the governing equations encountered in structure mechanics. 1982). the mesh deforms with it. like advance ratio. This code is thoroughly tested and used for flapping foil and wing simulations. etc. In order to perform accurate numerical simulations it is important to use an efficient code which is capable to solve for . At the time of writing. unstructured grids of polyhedral cells with full parallel support. where the fluid is allowed to flow through the fixed mesh.5 Different mesh motion solvers. because it requires the explicit tracking of the domain boundary. This is in contrast to the Lagrangian formulation. Therefore. This method incorporates and combines both Lagrangian and Eulerian frameworks. the ALE method has become the standard implementation in most popular codes to solve for the flow around moving boundaries while the mesh deforms accordingly. wing kinematics. the Arbitrary Lagrangian Eulerian (ALE) formulation is used to discretise the flow equations on moving and deforming meshes (Donea.

These mesh deformation techniques are described and assessed with respect to accuracy in chapter 3. validate and verify the numerical solver with the implemented and improved mesh motion technique. Before proceeding to the numerical results of the flow around flapping foils and wings. To meet the objectives the current thesis is structured in the following chapters. which is the subject of chapter 2. an improved mesh motion technique. Chapters 5 and 6 deal with the numerical investigations of two-dimensional flow around a flapping foil in hovering and forward flight conditions. improve current mesh motion techniques and implementation. For large wing translations and rotations the numerical grid needs to deform accordingly to maintain high accuracy of the flow solver. 3. This method is used to study the complex vortical patterns to identify optimal strategies in flapping foil and wing aerodynamics. It is . using realistic wing kinematics. in terms of accuracy and efficiency. the overall goal of this research is to develop a reliable mesh deformation technique. the solution procedure is described together with a brief discussion about the open-source framework OpenFOAM . This mesh deformation technique is implemented and used for flapping foil and wing aerodynamics. study the three-dimensional structure of vortical patterns. In order to satisfy this aim. Approach and outline In order to solve for the flow around flapping foils and wings.10 Introduction various conditions. using an accurate and efficient framework. the following objectives are defined: 1. In order to solve the governing equations for fluid flow. solve for the flow around two-dimensional flapping foils to study the wingwake interaction as well as the influence of wing kinematics. respectively. 2. to solve the flow around flapping wings. That chapter deals with the discretisation of the different terms as well as a definition of mesh quality. which is thoroughly validated and verified. based on radial basis function interpolation. Within the code. 4. The mesh motion technique is used by an incompressible unsteady CFD solver to solve for the flow around a threedimensional flapping wing on dense meshes in parallel. a finite volume discretisation is used. different mesh deformation techniques are incorporated. 5. solve for the flow around three-dimensional flapping wings to assess the importance of parameters like flapping amplitude. it is important to discuss the numerical modelling for flapping flight in chapter 4. Therefore. In addition to the already implemented mesh deformation techniques. a method based on radial basis function interpolation is discussed. is implemented in the open-source framework OpenFOAM . especially the leading-edge vortex. Furthermore. frequency or Reynolds number.

1. performance and wake patterns. Additionally.and three-dimensional flapping flight. It will be seen that accurately solving the flow around a flapping wing is not an easy task when the wing performs complex rotational motion. which is discussed in chapter 7. chapter 8 presents the preliminary results of a flexing wing in two. . Complex vortical structures induced by a model flapping wing can be accurately solved and analysed. These two-dimensional results provide good insight what to expect of the three-dimensional flow around a flapping wing. The conclusions and recommendations can be found in chapter 9.4 Objectives and approach 11 found that the kinematic modelling has a large influence on forces.

.

four important aspects of numerical . since the flow is dominated by the convection of vortices. Therefore. the boundary conditions and the solution procedure are presented. Fluent was already tested by Bos et al. Important aspects of numerical simulations Before describing the used methods in detail.CHAPTER 2 Finite volume discretisation A second-order finite volume discretisation of the incompressible Navier-Stokes equations on arbitrary polyhedral meshes is described. (2008). it was found that the Van Leer flux limiter provides the most accurate results. 2. This finite volume approach is applicable to general commercial and non-commercial CFD codes.1 Introduction Important aspects concerning numerical simulations. Furthermore. the flow around stationary and transversely oscillating cylinders showed that the code of OpenFOAM solves the flow in detail. A numerical simulation needs to be performed in an accurate and concise way. For test problems involving vortex decay and convection. Spatial and temporal convergence was proved as well. In addition to the mesh quality measures non-orthogonality and skewness. like stability and convergence. are being described. The commercial code Fluent and the open-source code OpenFOAM have been used for the simulations described in this thesis. different properties of a CFD simulation. such that this chapter deals with the validation of OpenFOAM using problems relevant for low Reynolds number insect flight. are addressed.

1 shows an example of a structured and an unstructured grid. the numerical grid. Wesseling (2001) and Ferziger & Peric (2002) described these methods in more detail. the interpolation from cell centres to cell faces and how to approximate the surface and volume integrals. the time is discretised as well. the cell ordering is fairly straightforward such that the flow solver uses this fact to solve the system in a more efficient way. hexahedral (four corners) to polyhedral (arbitrary number of corners) cells. is the generation of a numerical grid. When using (block-) structured grids. When using the finite volume method. Figure 2. the discretisation method. which is necessary to perform unsteady simulations. the governing equations. 2002). Besides the type of grid.14 Finite volume discretisation (a) Structured (b) Unstructured Figure 2. and the solution method to solve the system. concerning a CFD simulation. structured. Meshes can be generated in an structured way (a) and using unstructured methods (b). the cell shape can be varied from tetrahedral (three corners in two dimensions). Traditionally. simulations are discussed. However. needs to be described. as is the case in the current thesis. In the field of CFD. There are three types of grids. Besides the spatial discretisation. block-structured and unstructured grids (Ferziger & Peric. This is the more important asset of unstructured grids. The third aspect. In order to solve the incompressible Navier-Stokes equations. a structured grid is favourable in terms of accuracy and efficiency of the flow solver. . for less complex geometries. finite element and finite volume method.1 Different mesh generation methods. a division of the computational domain in a finite amount of cells. a suitable method to discretise these equations needs to be chosen. the finite difference. finite element methods are used for structural problems whereas numerical simulations related to fluid flow are mostly solved with finite volume methods. A drawback of a (block-) structured grid is that it is more difficult to create around complex geometries (commonly encountered in engineering problems). three methods are commonly used.

When an appropriate iterative solver is used.1 Introduction 15 Finally. When performing the iterations of the numerical process it should be the case that the numerical errors are not amplified. One of the advantages of the finite volume approach is that conservation is guaranteed for every small control volume and therefore. it should be sufficiently small. discretisation method and the choice of grid. especially when using a finite volume approach. the last aspect is boundedness. the system of discretised equations can be easy or difficult to solve. stability is the third important aspect. When the solution converges to a grid-independent solution. However. The second important property of a numerical scheme is convergence.e. it may happen that the exact solution is not approximated with decreasing time-step. from (Ferziger & Peric. In that case the solution process is called stable. conservation and boundedness. it is important to discuss different properties of the numerical solution method. the choice of numerical method is usually a trade-off. i. Since the governing equations in finite volume formulation are conservative. In order to check the consistency of the complete numerical scheme. when the cell size approaches zero. Considering a steady problem. where the same computation is repeated on subsequently refined meshes. convergence. consistency. this property should be respected by the discretised equations. Depending on the governing equations. The solution of the discretised system of equations should tend to the exact solution of the governing differential equations as the mesh spacing tends to zero. 2002). the mass flux of a conserved quantity through a specified system should be zero. Finally. The fourth property of a numerical scheme is conservation. The difference between the discretised and the exact solution is called the truncation error. like concentration or . so commonly the empirical approach is followed. The following properties are relevant concerning numerical simulations. The discrete system of equations needs to be solved up to a certain convergence criterion. For general engineering problems the stability of the numerical process is strongly dependent on the time-step. when the method is not stable. The first important property is consistency. Properties of numerical solution methods In order to solve the governing equations in a satisfactory manner. applied in general commercial and non-commercial CFD solvers. Certain variables in the governing equations contain physical bounds. Therefore. the fourth aspect of a CFD simulation is the iterative solver. The discretisation should become exact when the mesh resolution tends to infinity. the solution process is said to be converged. Convergence is difficult to prove theoretically for real engineering applications.2. a grid and time-step convergence study has to be performed using increasing grid resolution and decreasing time-step. a convergence criterion needs to be applied for the inner (within the linear system) and outer iterations (to couple the non-linear parts and perform non-orthogonal corrections). Since it is often not possible to find a numerical method which outperforms on all aspects. depending on the temporal discretisation scheme. limiting the iterative solver. stability. for the complete computational domain as a whole. without sources or sinks.

5 a general transport equation will be discretised to show how to deal with the different terms. When the numerical process respects these physical bounds. The used open-source solver. For mass conservation.7. the spatial and temporal discretisation methods are described in section 2. (2.3.. These equations are derived from conservation of mass. 2005). 1991. Additionally. The validation and verification discussion is the subject of section 2. like diffusion and convection. In this thesis. momentum and energy within an infinitesimally small spatial control volume. Finally. the solution procedure to solve the incompressible NavierStokes equations will be dealt in section 2. OpenFOAM . The nabla ∇ operator is defined in three dimensions as ∇= ∂. To solve the governing equations. which has been an authority in the field of computational fluid dynamics for decades. ∂. like discretisation schemes. Fluent is a generalpurpose CFD solver. providing support for code development. provides the source code and there is a big user community. the commercial flow solver Fluent and the non-commercial open-source code OpenFOAM are used. some small test problems are defined in order to test the influence of different numerical settings. Fluent and OpenFOAM . 2. a brief discussion about the treatment of boundary conditions is provided in section 2.2 The Navier-Stokes equations The governing equations for viscous fluid flow are a coupled set of non-linear partial differential equations (Anderson Jr. ∂x ∂y ∂z . Section 2. Two major drawbacks of a commercial solver are the unavailability of the source code and the potential lack of sufficient support from the company or the user community in code development. In order to validate and verify the CFD codes for our problem.10. ∂. two different CFD codes. Panton.16 Finite volume discretisation density and all other non-negative variables. the method is called bounded. grid resolution and time-step size.9 provides a brief description of the background and usage of both CFD codes.1) ∂t Here. These numerical solution procedures are present in the used flow solvers. ρ [kg/m3 ] is the density and u [m/s] the flow velocity vector.11. the major conclusions of this chapter are summarised in section 2. + + . the following continuity equation is obtained: ∂ρ + ∇ • (ρu) = 0. In section 2. like skewness and non-orthogonality in 2. When the discretised transport equation and corresponding boundary conditions are fully explained. followed by a discussion about the cell quality measures.4. The remainder of this chapter deals with a description of the governing equations of fluid flow in section 2.6.2.

with λ [W/(m·K)] the heat conduction transport coefficient. compressible viscous flows. For a Newtonian fluid. the stress tensor.1). the heat conduction is described using Fourier’s law: q = λ∇T. ∂t (2. The governing equations (2. T ).2 The Navier-Stokes equations 17 Secondly. only for very simplified problems there exists an analytical solution. Here I represents the identity tensor. for momentum conservation the following expression can be derived (neglecting gravity and additional body forces): ∂(ρu) + ∇ • (ρuu) = ∇ • σ. Without other restrictions.2) where σ [N/m2 ] is the surface stress tensor. the equation of state is specified. turbulence research. σ.1 Constitutive relations In order to close the system of equations (2. these equations can be difficult to solve and simplifications can be made if applicable to the concerning problem. q [W/s] is the heat flux vector and Q [J·m3 /kg] equals the nett energy generation. For compressible flow calculations also the energy conservation equation is specified: ∂(ρe) + ∇ • (ρeu) = ∇ • (σu) − ∇ • q + ρQ. p [N/m2 ] is the pressure and µ [Ns/m2 ] is the dynamic viscosity.2.2) and (2. these equations are used for high and low speed flows. Additionally. (2. The Navier-Stokes equations are non-linear.2. which makes them difficult to solve. 2. necessary in viscous fluid flow. such as the perfect gas law: p = ρRT. However.3) in combination with additional turbulence modelling can be used in a wide variety of engineering problems. To close the energy equation.3) where e [J/kg] is the total specific energy (including kinetic and potential energy). .2) and (2. in which T [K] is the temperature and R [J/(mol·K)] the specific gas constant. (2. constitutive relations are needed. The constitutive relation for the total specific energy yields as follows: e = e(p.1). which is defined for a Newtonian fluid as 2 σ = − p + 3 µ∇ • u I + µ ∇u + ∇uT . are called the Navier-Stokes equations. The full set of equations describing unsteady. multi-phase flows and a lot of other applications. ∂t (2.3).

. 2003. this system of equations is closed such that there is no need to use the energy equation and additional turbulence modelling. A flow can be assumed to be incompressible.2. 2. (2.5) the following non-dimensional form of the incompressible continuity and momentum equations is obtained: ∇ • u∗ = 0. 2003. ∇p ∂u + ∇ • (uu) = − + ν∇2 u. Therefore. these relevant time-scales are. (2.. x.9) Tvisc Uref Lref = . as follows: u∗ = u . the incompressible laminar Navier-Stokes equations are solved for Reynolds numbers ranging from Re = 100 to 1000.10) ν Tconv These dimensionless numbers represent order estimates for time-scale ratios in the flow. the relative relevance of the different terms in equations (2. In case of incompressible flow. which is considered to be incompressible (Lentink. two main dimensionless numbers are identified as relevant parameters. Therefore.4) (2.5) is revealed by making those equations dimensionless. the Strouhal (St) and Reynolds number (Re): St St = fref Lref Tconv = .3 Dimensionless numbers In general. When substituting equation (2. u. ∂t ρ (2.10). ρref (2. t. x∗ = x . Uref Tmotion (2. such that ρ∗ = 1. 2008)) and thermal expansion effects can be neglected. L p∗ = p 2 . ρref · Uref ρ∗ = ρ . the time-scale Re = . Bos et al. respectively.2. In (2. the density is constant.9) and (2.6) into equations (2.3 times the speed of sound (Lentink. p and ρ are scaled with their reference values.4) and (2.18 Finite volume discretisation 2. The incompressible Navier-Stokes equations are: ∇ • u = 0. Uref t∗ = t · fref .5) with ν = µ/ρ [m2 /s] being the kinematic viscosity. the main variables.6) The star (*) is used to indicate the dimensionless variables.7) 1 2 ∗ ∂u∗ + ∇ • (u∗ u∗ ) = −∇p∗ + ∇ u . Bos et al.8) ∂t Re In these equations.4) and (2. 2008) and laminar (Williamson. 1995).2 Incompressible laminar flow simplifications The current research deals with the flow around flapping wings at insect scale. when the velocity is lower than 0. and (2. For an incompressible flow.

Tvisc and the relevant time-scale of the body motion. combined with the constitutive relations. Discretisation is performed assuming a linear variation of scalar variable φ across . consequently. face surfaces and points. the reference values need to be chosen appropriately. the governing equations need to be approximated over these cells.2. Tmotion .2. completely filling the domain. i. which will be explained in section 2.3 Spatial and temporal discretisation 19 for convective transport. OpenFOAM uses a collocated variable arrangement (Ferziger & Peric. with centres P and N from figure 2. the governing equations are significantly simplified. Derived boundary fields. solving the unsteady incompressible laminar flow can be seen as performing a Direct Numerical Simulation.2 shows an arbitrary polyhedral control volume VP with centre P and neighbouring centre N .5. The full Navier-Stokes equations. 1996): (x − xP )dV = 0. Since time can be interpreted as a parabolic coordinate (Patankar & Spalding.3 Spatial and temporal discretisation This section deals with the spatial and temporal discretisation of the governing mathematical equations. are applicable to all kind of flows. starting at the initial solution. which can only be solved analytically for extremely simplified model problems. dependent on the maximal Courant number. are defined in the face centre.e. however. Faces which are not shared are boundary faces. Tconv . viscous transport. 1972). In order for the dimensionless numbers to have a proper physical meaning. VP Every two cells. these simplifications need to be justified by the concerned fluid flow problem. at sufficiently low Reynolds numbers. The time-step may vary. 2002). Therefore. After the domain is discretised into a set of control volumes. needs to be discretised throughout the entire computational domain. However. such that the costs for solving may be reduced. like pressure and velocity. It was seen that fluid flow is governed by non-linear partial differential equations. share an internal face whose geometric centre is denoted by f and has an outward pointed normal vector Sf . which is used to march linearly in time. The computational point xP is located at the centroid of the computational cells. like surface normal gradients or face fluxes. Space. it is sufficient to specify the initial time-step. When the flow is considered laminar and incompressible. Concerning flapping wing physics at lower Reynolds numbers (100 ≤ Re ≤ 1000) the flow inherently is incompressible and laminar. which means that every control volume centre is used to store the values of all variables. The finite volume approach needs a domain subdivision into a finite number of convex polyhedral control volumes without overlap. where the computational costs strongly depend on the desired resolution and solution methods. which is found from the following relation (Jasak. 2. Figure 2.

11) where O(|(x − xP )|2 ) represents the second-order truncation error. 1996). a cell.2 Discretisation of the computational domain using finite volume cells. d. The vector from the cell centre to the neighbouring cell centre N is df .4 Measures of cell quality Since the accuracy of the numerical solution heavily depends on the interpolation from cell to face centre. The faces of cell P are directed with the unit normal vector Sf and may have an arbitrary number of corners. An arbitrary polyhedral control volume is constructed around a centre P and with volume VP . 2. We will briefly describe the cell quality based on non-orthogonality and skewness. which is not within the scope of the present thesis.3(a) by the angle αN between the face normal vector Sf and the line connecting the two cell centres. Both truncation errors can be expanded using a full Taylor series expansion. including local mesh refinement. ∂t ∆t (2. 1996). similar to the spatial truncation O(∆x2 ). Using a Taylor series approximation.12) With this linear temporal behaviour of φ the truncation error is second-order O(∆t2 ). From (Jasak. Ferziger & Peric. the following expression is obtained: φ(x) = φP + (x − xP ) · (∇φ)P + O(|(x − xP )|2 ). this method can be used for complex unstructured three-dimensional meshes. which will both be used to assess the performance of mesh motion solvers in chapter 3. one can imagine that the cell quality is very important. (2. Since this discretisation approach is able to cope with arbitrary polyhedral cell volumes. This angle needs to be as small as possible in order to minimise the truncation . This scalar variable φ can be seen as pressure or a velocity component. For the temporal variation of this scalar variable φ a similar expression can be found: ∂φ(t) φ(t + ∆t) − φ(t) = + O(∆t). Jasak. cell non-orthogonality is defined in figure 2. 2002. but can be found in (Wesseling. First of all. 2001.20 Sf f Finite volume discretisation N df P rP z y VP x Figure 2.

Figure 2. From (Jasak.13) Here. The second quality criterion is the cell skewness.3(b). u [m/s] the transport velocity and ν = µ/ρ [m2 /s] is the kinematic viscosity. error of the diffusion term. When the line connecting the two neighbouring face centres does not coincide with the connecting face centre. The cell skewness is defined by the vectors m and d. This expression contains a temporal.2. |d| where m and d are defined in figure 2. which forms the basis for the incompressible Navier-Stokes equations. con- . ρ [kg/m3 ] is the reference density. see figure 2. Cell non-orthogonality is defined as the angle between the face normal vector Sf and the direction vector between two cell centres P and N . ∂t ρ (2. 2.5 Discretisation of an incompressible momentum equation 21 Sf f m N Sf αN P f d N d fi P (a) Cell non-orthogonality. This partial differential equation has the following form: ∇p ∂u + ∇ • (uu) − ∇ • (ν∇u) = . the cell is skewed. Twodimensional representation of cell non-orthogonality (a) and cell skewness (b) as a measure for the finite volume cell quality. Assessing cell skewness is important.3 Quality measures using cell non-orthogonality and cell skewness. 1996). (b) Cell skewness.3(b). since the interpolation from cell centre to face centre strongly depends on this quality criterion as will later be seen in this chapter.5 Discretisation of an incompressible momentum equation This section deals with the temporal and spatial finite volume discretisation of the incompressible momentum equation. The degree of skewness is defined by: ψ= |m| .

the integral form of the incompressible NavierStokes equations is obtained by integrating over a control volume. For this. (x − xP )dV ) · (∇φP ) Similar. dS • a. Before dealing with the discretisation of the different terms of equation (2. given by: ∂u ∂t Finite volume discretisation ∇ • (uu) ∇ • (ν∇u) : temporal term.14) This equation is solved in both CFD codes used. (2. In the remainder of this section. like: ∇ • adV VP = SCV dS • a.. a second-order approximation is obtained.14) are elaborated in more detail. Using Gauss’ theorem. which may represent the different velocity components.16) . When substituting equation (2. the different terms of equation (2. The divergence and gradient terms are evaluated using Gauss’ theorem (Panton.14) it is important to discuss the evaluation of the volume. ρ (2. f Sf = ≈ dSf • af .22 vection and a diffusion term. f (2. the scalar variable φ is used. surface. Anderson Jr. necessary to understand the evaluations of the convection and diffusion terms. CV : VCV ∂u dV + ∂t VCV ∇ • (uu)dV − ∇ • (ν∇u)dV = VCV VCV ∇p dV. : diffusion term. divergence and the gradient integrals. Fluent and OpenFOAM . the volume integral of the divergence of a vector a can be written as the sum of all faces. : convection term. such that the result is a multiplication of the scalar value multiplied by the cell volume. 1991).11) into the volume integral. Using the finite volume approach. which defines a relation between the volume and the surface integrals. φ(x)dV VP = VP (φP + (x − xP )(∇φ)P )dV dV + ( VP VP = φP ≈ φP V P . 2005.15) where Sf is the face surface area. for the surface integral the following yields: f dS · a = Sf · af .

2. 1993).4 (Sweby.16) and a linearization . dSφ. A linear interpolation is performed using the following expression: φf = fx φP + (1 − fx )φN . The SuperBee. as: ∇φdV VP = SCV dSφ. Discretisation of the gradient integral of a scalar variable φ can be written. the flux φf is interpolated from the cell centre to the face centres. (1999) applied extra face interpolation schemes. f Sf = ≈ dSf .1 Face interpolation schemes Similar to the divergence term.5 Discretisation of an incompressible momentum equation 23 Here. or Van Leer (Van Leer. Koren and Van Leer limiter are shown using the Sweby diagram in figure 2.2. CV is the control volume with surface normal vectors Sf and af is a vector interpolated to the cell faces using a second-order linear interpolation method. In OpenFOAM . f 2.2 Convection term When the volume integral of the convection term from equation (2. Jasak et al.5. using Gauss’ theorem. fx = |f D|/|P D|. Section 2. the following relation can be derived using equation (2. 1986). So. the Koren limiter (Koren. The purpose of flux limiters is to limit the gradient of the solution in order to avoid spurious oscillations and to improve the stability of the scheme. 1984). both divergence and gradient volume integrals can be reduced to a summation of the corresponding vector or scalar variable over the cell faces.13) is considered. 1984).10 shows a comparison of the results using different flux limiters on the solution of a model problem of vortex decay and convection. Sweby. The linear interpolation factor fx is defined as the ratio of two distances. 1979. The standard face interpolation scheme is obtained by central differencing. which is illustrated in figure 2. like upwind blending using a gamma coefficient and differencing using a flux splitting limiter such as SuperBee (Roe.5.φf .5.

Gamma.13) is discretised and approximated using linearization as ∇ • (ν∇φ)dV VP = f Sf · (ν∇φ)f νf (S · ∇φ)f . (S · ∇φ)f = |m| |d| . method (midpoint.g.5 0 0 0.5 r 2 2. Van Leer.5 r 2 2. which represents the ratio of successive gradients on the mesh. e.5 1 0.5.5 1 0.4 Different flux splitting limiters. given by F = Sf · (u)f . The other term (S · ∇φ)f is obtained on a non-orthogonal mesh by the following expression: φN − φP + k · (∇φ)f . 1984). 2. 1986). 1979.5 3 φ(r) 3 2.5 1 1.3 Diffusion term The volume integral of the diffusion term from equation (2.5 1 1.5 3 Finite volume discretisation 3 2. The scalar variable φ needs to be interpolated using a second-order interpolation method in combination with a flux limiter. Different flux limiters are employed. 1993) and (c) Van Leer (Van Leer. the terms (S · ∇φ)f and νf need to be approximated using a proper method.5 0 0 0. F is the mass flux. Sweby.24 3 2. (b) Koren (Koren. The face viscosity νf is obtained by interpolation from cell centre to faces. The flux splitting scheme are a function of r. (a) SuperBee (Roe.5 2 1. least squares): ∇ • (uφ)dV VP = f Sf · (uφ)f Sf · (u)f φf F φf .5 2 φ(r) 1. linear.5 1 1. = f = f Here.5 0 0 0.5 1 0. = f Here. Flux splitting schemes are used to limit the gradient of the solution in order to avoid spurious wiggles.5 2 φ(r) 1.5 r 2 2.5 3 (a) SuperBee (Roe) (b) Koren (c) Van Leer Figure 2.

ts

2.5 Discretisation of an incompressible momentum equation
φD

25

φf Sf φP φU U P Flow direction f D P d f αN m N k

Figure 2.5 Variation of the flux φ. The value of φ at the face f is determined as a function of upstream and downstream values.

Figure 2.6 Cell non-orthogonality treatment. Illustration of the cell nonorthogonality correction which is used on meshes with large skewness and nonorthogonality.

Here d is the vector between two adjacent cell centres and m is parallel to d with magnitude of the surface normal vector Sf . The decomposition of Sf is shown in figure 2.6 and derived in Jasak (1996), Juretic (2004) such that the following relation holds: Sf = m + k, where k is orthogonal to the surface normal vector S.

2.5.4

Temporal term

Since the unsteady Navier-Stokes equations are solved, a proper discretisation of the temporal scheme is necessary. The time derivative represents the temporal rate of change of φ which needs to be discretised using new and old time values. This time difference is defined using prescribed time-step size ∆t such that: φn+1 = φn + ∆t, where φn and φn+1 are the scalar variable φ at the old and new time instances, respectively. Two implicit time discretisation methods are considered, one firstorder and and one second-order scheme. The first-order discretisation is simply the temporal difference: ∂φ φn+1 − φn = , ∂t ∆t and the second-order discretisation, see (Wesseling, 2001, Hirsch, 1988), is given by: 1 3 n+1 φ − φn + 2 φn−1 ∂φ = 2 . ∂t ∆t This implicit scheme is referred to as the second-order backward differencing scheme, where φn−1 is the old-old value of φ. Consequently, the corresponding

26 volume integrals obey the following relations:

Finite volume discretisation

CV

φn+1 − φn ∂φ dV = VP , ∂t ∆t
3 n+1 2φ 1 − 2φn + 2 φn−1 VP . ∆t

CV

∂φ dV = ∂t

(2.17)

Note, that these relations are only valid on fixed meshes and constant time-steps. According to (Wesseling, 2001, Hirsch, 1988), the explicit first-order time integration method may be unstable if the Courant number is larger than 1, where the Courant number is defined as u · ∆t Co = . ∆x Implicit methods are in general more stable, compared to (semi-) explicit methods, such that in the current research the implicit first- and second-order backward scheme have been used. While the implicit methods are bounded and stable, a pre-defined maximal Courant number Comax is used to vary the corresponding 3 time-step during the simulation. In that case, the coefficients 2 , 2 and 1 in (2.17) 2 should be elaborated to incorporate the ratio of the old and current time-steps. In section 2.7.3 this will be discussed in more detail.

2.6

Boundary conditions

In order to solve the discretised governing equations, boundary conditions need to be defined at the boundaries of the computational domain. There are four boundary conditions (Hirsch, 1988, Wesseling, 2001), which are used to close the system, namely: 1. zero-gradient boundary condition, defining the solution gradient to be zero. This condition is known as a Neumann-type condition, ∂φ/∂n = a, 2. fixed-value boundary condition, defining a specified value of the solution. This is a Dirichlet-type condition, φ = b, 3. symmetry boundary condition, treats the conservation variables as if the boundary was a mirror plane. This condition defines that the component of the solution gradient normal to this plane should be fixed to zero. The parallel components are extrapolated from the interior cells, 4. moving-wall-velocity boundary condition is used on a moving boundary to keep the flux zero, using the Arbitrary Lagrangian Eulerian approach. For external flow simulations, a distinction is made between the outer and the inner boundaries, the latter corresponds to the moving wing or body. To minimise the effects of the outer boundaries it is desirable to specify a symmetry boundary

2.7 Solution of the Navier-Stokes equations

27

condition (3) at those fixed boundaries, unless a free-stream is specified. In case of forward flapping flight, two domain boundaries are defined as inflow and outflow, respectively. At the inflow boundary the velocity is defined as fixed-value (2) and the pressure as zero-gradient (1). On the other hand, at the outflow boundary, the pressure has to be fixed-value and the velocity zero-gradient (Hirsch, 1988, Wesseling, 2001). On a stationary wall the no-slip condition needs to be guaranteed, therefore a fixed-value (u = 0) is specified for the velocity in combination with a zero-gradient for the pressure. If the boundary of the wall moves, than the proper boundary condition is the moving-wall-velocity (4) which introduces an extra velocity in order to maintain the no-slip condition and ensures a zero flux through the moving boundary.

2.7

Solution of the Navier-Stokes equations

Previously, the different terms to discretise the general momentum equation (2.13), were described. This section briefly deals with the discretisation of the NavierStokes equations and the solution procedure. The incompressible laminar NavierStokes equations were given by (2.4) and (2.5): ∇ • u = 0, ∂u ∇p + ∇ • (uu) = − + ν∇2 u. ∂t ρ There are two items, requiring special attention, namely the non-linear term present in the momentum equation and the pressure-velocity coupling (Ferziger & Peric, 2002). The non-linear term in these governing equations, ∇ • (uu), can be solved either by using a solver for non-linear systems or by Newton linearization. Previously, it was seen that the convection term can be written as: ∇ • (uu)dV
VP

=
f

Sf · (u)f (u)f F(u)f

=
f

= ap up +
N

aN uN ,

where ap , aN and F are still depending on u. ap and aN represent the diagonal and off-diagonal terms of the sparse system of equations, respectively. A complete derivation can be found in (Jasak, 1996). Since F should satisfy the continuity equation (2.4), both equations (2.4) and (2.5) should be solved together as if it was a coupled system. In order to avoid the use of expensive solvers for non-linear systems, this convection term is linearised such that existing velocity fields will be used to calculate the matrix coefficients ap and aN .

28

Finite volume discretisation

2.7.1

Pressure equation and Pressure-Velocity coupling

Since the pressure depends on the velocity and vice-versa, a special treatment of this inter-equation coupling is needed. In order to derive the pressure equation, a semi-discrete formulation of the momentum equation is written as: ap up = H(u) − ∇p. (2.18)

This equation is derived from the integral form of the momentum equation using the previously described discretisation methods and divided by the volume. Following Rhie & Chow (1983) the pressure gradient in equation (2.18) is not yet discretised. The H(u) term contains two parts, a convection and a source contribution. The convection part includes the matrix coefficients for all neighbours multiplied by their corresponding velocities. The source contribution consists of all source terms, except the pressure term, including the transient term. Therefore H(u) can be written as follows: H(u) = − aN uN +
N

u0 . ∆t

Additionally, the discretised continuity equation (2.4) is given by: ∇•u =
f

Sf • uf = 0,

(2.19)

Now equation (2.18) is rewritten to find an expression for up : up = H(u) 1 − ∇p. ap ap (2.20)

The velocities on the face of a finite volume cell can be expressed as the interpolated value on the face of equation (2.20): uf = H(u) ap − 1 ap (∇p)f .
f

(2.21)

f

This equation will be used to determine the face fluxes. If equation (2.21) is substituted into equation (2.19), the following pressure equation can be obtained: ∇•
1 ap ∇p

= ∇•

H(u) ap

.

(2.22)

The Laplacian operator is discretised using existing methods, which are previously explained. Combining equations (2.18) to (2.22), the final form of the discretised Navier-Stokes equations can be written as: ap up = H(u) − Sf (p)f ,
f

(2.23)

a brief description is given. are lagged. 2. from the previous stage.22) can be formed. the face fluxes are guaranteed to be conservative (Ferziger & Peric.22) is satisfied. Using the predicted velocity. since the actual pressure gradient is not yet calculated. which means that these equations are solved in sequence.7. 2.24) it can be observed that both equations are coupled through the pressure and velocity. Pressure solution stage. The last equation in this sequence is (2.20) in an explicit fashion.23) and (2. 1972) based algorithms.2. known from the previous time-step. The PISO algorithm consists of the following steps: 1. In unsteady simulations all other inter-equation couplings. 1986) or SIMPLE (Patankar & Spalding. For incompressible unsteady flow . The momentum equation (2. please consult (Jasak. f (2. 2002). Ferziger & Peric. Juretic. (2. Explicit velocity correction stage. which are consistent with the new pressure field. Using this pressure equation a better approximation of the new pressure field can be obtained. For the discretised form of the Navier-Stokes equations (2.23) is solved using the pressure gradient. it has become possible to describe the solution procedure to obtain the solution of the Navier-Stokes equations. from stage 2. which determines the conservative fluxes. is replaced by a better pressure field. 2004). Fluent and OpenFOAM used the PISO scheme for transient flows and the SIMPLE scheme for steady flows. 3. besides the pressure-velocity equations. equation (2.24) additionally.7 Solution of the Navier-Stokes equations and Sf • f 29 1 ap (∇p)f = f Sf • H(u) ap . the H(u) term can be constructed such that the pressure equation (2. Since the PISO scheme was used. the velocity field has to be corrected accordingly.25) f When equation (2. The inner-equation coupling is established using either PISO (Issa.25). which requires special attention. this system is solved in a segregated manner. the face flux is calculated using: F = Sf • u = Sf • H(u) ap − 1 ap (∇p)f . Momentum predictor stage. Furthermore. 2002. from stage 1.23) provides an approximation of the new velocity field. This is performed using (2. such that they are included in the PISO loop. 1996. Since the approximated pressure field.2 Procedure for solving the Navier-Stokes equations After dealing with the discretisation of the Navier-Stokes equations in combination with the PISO algorithm. Both flow solvers. Since a simultaneous approach would be too computationally demanding. For more detailed information.

The Lagrangian contribution allows the mesh to move and deform according to the boundary motion. like pressure. commonly used to satisfy conservation on deforming meshes. the mesh deforms with it. 2. because it requires the explicit tracking of the domain boundary. all remaining equations of the system are solved. go back to step 2. the solution sequence can be summarised as follows: 1. as well as a new set of conservative fluxes. 3. when the flow domain moves or deforms in time due to a moving boundary. a fixed mesh becomes inconvenient. special attention is necessary to describe the modifications to the discretised equations dealing with the Arbitrary Lagrangian Eulerian approach. However. 1982). Using the new conservative fluxes. 5. At the time of writing. update the turbulent viscosity at this stage. If the material or fluid deforms. Initialisation of all fields. This procedure results in solution fields for all solved variables. whereas the Eulerian part takes care of the fluid flow through the mesh. where the mesh is fixed to the fluid or material. This method incorporates and combines both Lagrangian and Eulerian frameworks. 6.7. the ALE method has become the standard implementation in most popular CFD codes to solve for the flow around moving boundaries while the . the Arbitrary Lagrangian Eulerian (ALE) formulation is used to discretise the flow equations on moving and deforming meshes (Donea. Start the simulation to obtain the velocity and pressure values at the new time-step. velocity and possible turbulence variables. If turbulence modelling is included. Since the present thesis deals with deforming mesh problems. This method is commonly used to discretise the governing equations encountered in structural mechanics. including pressure and velocity. where the fluid is allowed to flow through the fixed mesh (Ferziger & Peric. 2002). This is in contrast to the Lagrangian formulation. using the initial condition. 4. The pressure and velocity fields are obtained for the current time-step. 2. Unless the final time is reached.3 Arbitrary Lagrangian Eulerian approach The governing equations to solve the flow are generally discretised using the Eulerian description. Therefore. Iterate through the PISO loop until the pre-defined tolerance of the pressurevelocity system is reached. Create and solve the momentum predictor equations using the obtained face fluxes.30 Finite volume discretisation with additional turbulence modelling.

e. otherwise inconsistency could introduce numerical errors.7 Solution of the Navier-Stokes equations 31 mesh deforms accordingly. the second-order accurate three time levels backward scheme was used throughout this research. In time. For c efficiency. The current research uses . The fluid mass flux m is obtained as part of the solution. dV − ∂t VCV SCV Using the current finite volume discretisation. the momentum equation (2.26) where VCV is the arbitrary volume and us the velocity of the moving surface. the maximal Courant number varies during the simulation. If equation (2. 1996): ∂ n · us dS = 0. using either an implicit first-order Euler method or a second-order backward scheme (Tukovi´ & Jasak. Note.2. The mass flux through the face is given by mf = nf • uf Sf and ˙ ˙ the cell face volume change by Vf = nf • usf Sf . (2.8. the temporal dimension is marched using a variable time-step. that the previous equation is derived for a constant time-step. the computational domain is split into a finite number of polyhedral cells with varying shape and volume. where us represents the cell face velocity. satisfying mass ˙ conservation. φ f f f f + = where the subscript P denotes the cell values and f represents the values at the face centres. old and old-old values. 2007). corresponding to a maximal Courant number.26) is discretised in space and time the following relation is obtained for a constant time-step: n+1 n−1 n−1 n−1 n 3ρn+1 φn+1 VP − 4ρn φn VP + ρP φP VP P P P P 2∆t ˙ (mn+1 − ρn+1 Vfn+1 )φn+1 ˙f f f f n+1 n+1 (ρΓφ )n+1 Sf nn+1 • (∇φ)n+1 + sn+1 VP . the new. n and n−1 are. The superscripts n+1. see section 2. It is very important to determine the volume face flux in a consistent way such that it equals the swept volume calculation. it is important to determine the volume face flux such that it satisfies the Space Conservation Law. 2002) or Geometric Conservation Law (Lesoinne & Farhat. 2009).14) for a scalar field φ can be derived on a moving mesh as ∂ ∂t ρφdV + VCV SCV ρn·(u−us )φdS− SCV ρΓφ n·∇φdS = VCV Sφ (φ)dV. since the mesh is deforming. The temporal discretisation scheme should be similar to the one used in the momentum equation. i. In general. respectively. The cells do not overlap and completely fill the domain (Jasak. Furthermore. The relationship between the rate of change of the volume VCV and the velocity us of the boundary surface S is defined by the so-called Space Conservation Law (SCL) (Ferziger & Peric.

the polyhedral cells and faces are decomposed into tetrahedral cells. described in the previous section. tn and tn+1 . it may occur that the volume becomes warped.32 Finite volume discretisation tn+1 f P z y f V x tn (a) (b) Figure 2. The swept volume V of a decomposed face is shown in (b). (a) shows the decomposition of a polyhedral cell into tetrahedral volumes and faces. The finite volume cell decomposition is used to form a tetrahedral mesh used for point-based mesh motion solvers and to calculate the swept volumes. If a polyhedral face is swept from one time-step to the next.7(a). 2. Therefore.7 Finite volume cell decomposition to calculate swept volumes. The polyhedral face is decomposed into triangles. . using its centroid.8 Swept volume calculation The swept volume is defined as the volume swept by a face of a polyhedral cell between two subsequent time-steps. Therefore. such that a volume calculation will not be trivial. a defined maximal Courant number leading to a varying time-step. φ f f f f where the new and old time-steps are respectively given by ∆tn+1 = tn+1 − tn and ∆tn = tn − tn−1 . the following relation is derived for a non-constant time-step: 1+ ∆tn+1 ∆tn+1 + ∆tn n+1 ρn+1 φn+1 VP − 1 + P P ∆tn+1 ∆tn n ρn φn VP P P + + = 1+ (∆tn+1 )2 ∆tn (∆tn+1 + ∆tn ) f n−1 n−1 n−1 ρP φP VP ˙ (mn+1 − ρn+1 Vfn+1 )φn+1 ˙f f f n+1 n+1 (ρΓφ )n+1 Sf nn+1 • (∇φ)n+1 + sn+1 VP . This calculation is necessary in order to satisfy the Space Conservation Law. see figure 2.

Vp4 and Vp5 correspond to different tetrahedron volumes. i. Therefore. For completeness.9 Numerical flow solvers 33 which is illustrated in figure 2.2. The pressure-velocity coupling in incompressible flow simulations was obtained using the iterative PISO scheme (Ferziger & Peric. To obtain the total swept volume of the decomposed triangle Vtriangle . using either one of the two diagonals in the right side face of the swept volume shown in figure 2. Since this prism may be warped.e. One tetrahedron is shown in the figure. Fluent is a well-known. The accuracy was set to double-precision and the initial conditions were chosen to be uniform. The convergence criterion for . 6 1 (Vp1 + Vp4 + Vp5 ). necessary for fluid flow. the volume is calculated as the sum of three tetrahedron volumes. but the remaining volume of the prism contains two more tetrahedrons. are briefly discussed. which exploits the finite volume approach. The spatial discretisation was second-order upwind and the time discretisation was first-order implicit Euler (Hirsch. 2.7(b).7(b). the CFD codes. These two additional tetrahedrons can be constructed in two different unique ways. 1988). The swept volume of a polyhedral face f is equal to the sum of the swept volumes of the different decomposed triangles. this method is implemented in OpenFOAM . used throughout the present research. 2 This swept calculation is successfully validated on test cases using unsteady flow with mesh motion and proved to be sufficiently accurate. Two different CFD codes are used. This section deals with a brief elaboration of the computer software. the average of both V1 and V2 is taken as V2 = Vtriangle = 1 (V1 + V2 ). 2002). 2006).7 described the solution procedure to solve the Navier-Stokes equations. easy to use and proven CFD solver. The boundary condition on the body was set to no-slip.7(b). one commercial (Fluent ) and one open-source package (OpenFOAM ). which is the only method for which the dynamic mesh module is implemented by Fluent . the main settings that we used. As is illustrated in figure 2. Therefore. see (Zuijlen van. 6 where the same base tetrahedron Vp1 is used and Vp2 . which need to be accurately calculated.9 Numerical flow solvers Section 2. the swept volume of such a triangle is similar to a prism with a triangle-shaped bottom area. the total swept volume using the two different unique tetrahedron decompositions is: V1 = and 1 (Vp1 + Vp2 + Vp3 ). Vp3 .

One of the main assets of the code is that the user writes the code in an intuitive way himself. c 2009) and Large Eddy Simulations (Jasak. relevant to flapping insect flight in (Lentink & Gerritsma. 2002). 2003. The commercial flow solver Fluent has already been tested earlier specifically for low Reynolds number flows. making this code very versatile. but not for low Reynolds number flows. All terms are discretised using standard second-order central differencing. Bos et al. Zuo et al. . which is used to solve partial differential equations. mesh motion (Jasak.10 Code validation and verification This section deals with the validation and verification of the CFD solvers that were used throughout the current research. Jasak et al. 2007). the iterative solvers and their corresponding convergence criteria need to be specified. while the pressure-velocity coupling equation employs its asymmetric counterpart pre-conditioned bi-stab conjugate Gradient (PBiCG) solver. which is derived from the complete system of equations. has been tested extensively for fluid-structure interaction (Tukovi´ & Jasak. The convergence criterion is based on the residual. Additionally. the Navier-Stokes equations. such that the residual Res is defined by: Res = b − Ax The pressure equation is solved using a pre-conditioned conjugate Gradient (PCG) iterative solver.. 1979). Therefore. however. turned out to be the linear scheme with the Van Leer limiter (Van Leer. The other code used in this research. 1996. OpenFOAM . OpenFOAM .10 will show that the best method to discretise the convection term.. Thaweewat et al. 2004. in combination with a variable time-step corresponding a maximal Courant number (Wesseling. 2007). 2008. 2007). e.. Ferziger & Peric. except for the convection term. 2001. written in C++.. on a finite volume mesh containing polyhedral cells. which is written as Ax = b. 2009). is a general object-oriented toolbox. for our low Reynolds number problems. without the need to dig deep in the underlying code. Appendix B summarises the used discretisation schemes and iterative solvers combined with the convergence criteria for both flow solvers Fluent and OpenFOAM . Concerning the temporal discretisation scheme. Jasak et al. The pre-conditioning method varies from incomplete Choleski to incomplete LU decomposition (Wesseling.. the implicit second-order backward scheme is used. 2007.34 Finite volume discretisation the iterative method was satisfied with mass and momentum residuals dropping O(10−5 ) in magnitude. Juretic. Section 2. this section only deals with the validation and verification of the OpenFOAM flow solver for test problems relevant for low Reynolds number flapping insect flight.g. 2001. 2.

2. 2003): Vθ = t−m f (η). two vortex cases are considered. This problem is relevant for low Reynolds number flapping flight.27) where Vθ represents the radial velocity. followed by the two vortex simulations to discuss the results. which solves for the unsteady. described in section 2.8 shows the velocity profile for different values of m to show the effect on the compactness of the vortex. the most accurate scheme needs to be determined from linear differencing. Within the context of the present research. 2005) is obtained. on the other hand. Since this flow solver was used for low Reynolds number flows with vortex convection it is necessary to choose the correct face interpolation scheme for the convection term. Vortex definition When a vortex is used for code validation it is important that a well confined definition is used. the compactness of the vortex is increased with increasing m. With decreasing m. the curve profile becomes steeper until the asymptotic behaviour is lost at negative m.1. 2005. two test cases are investigated. Concerning the convection term. Figure 2. SuperBee splitting.5.10 Code validation and verification 35 The open-source toolbox OpenFOAM provides a framework of finite volume based functions in order to build a specific application for solving partial differential equations. the accuracy of the code is assessed using unsteady flows around static and plunging cylinders in section 2. For validation purposes a vortex corresponding to m = 1. the vortex should have finite and monotonic velocity and vorticity profiles. . Conversely. one concerning a decaying vortex.e. the other deals with the vortex convection. Therefore. m is a compactness coefficient and η a similarity parameter. Koren splitting and Van Leer splitting. although maximal angular momentum (Panton.5 after which a region with counter-rotation appears.5 the well-known Lamb-Oseen vortex is the result. since these flows are dominated by unsteady vortical structures.2. incompressible flow using deforming meshes. gamma differencing. In this figure some interesting characteristics can be observed.1 2D vortex decay and convection To validate the code. the face interpolation for diffusion and source terms will be kept fixed at second-order linear (central) interpolation. the decay and convection of a Taylor vortex. i. When m is chosen to be 0.10. until m = 1. (2. After the selection of the proper flow solver settings. In order to assess the accuracy of these interpolation schemes. which is not of the desired compactness.10.10. as was briefly discussed in section 2. First a proper vortex definition is described. an application is developed. Zhou & Wei.5. but analogue to (Zhou & Wei. Besides the fluid and flow properties it is necessary to specify different interpolation schemes for the different terms of the governing equations. a Taylor vortex (Panton.2. The following definition is used from (Panton. 2005).

The mesh size was fixed to (100 x 100) and the time-step is varying to meet a maximal Courant number of Comax = 1.5.36 Finite volume discretisation Figure 2.. The temporal term is discretised using a second-order backward scheme. 1986). To study the effect of the face interpolation of the convection term. which are both considered to be fine enough. Koren limiter (Koren.28) Vθ = 2e 2 λe−2λ . 2005).9. where λ is a function of the similarity parameter η: λ= η 2 (2) . Vortex decay As a first test problem. The Reynolds number . provides a better representation of vortical flows like in low Reynolds number vortex shedding problems (Panton. a standard OpenFOAM solver is used. from (Panton. icoFoam. a Taylor vortex is considered. 1999). ranging from −0. This squared domain is discretised with a Cartesian finite volume grid of 100 x 100 mesh cells as is shown in figure 2. which decays in a twodimensional squared domain with dimensions (5 x 5). 1979) and standard linear interpolation. which is not necessary for this problem. 2003). The vortex definition. except for the convection term. Furthermore. This m values determines the compactness of the vortex. without mesh motion.0.5 to 2 in steps of 0.8 Different velocity profiles of a similar vortex definition. 2005) and is given by: 2 1 (2. SuperBee limiter (Roe.5 provides the most compact vortex. Van Leer limiter (Van Leer. To solve the incompressible Navier-Stokes equations. all other terms. When the compactness coefficient m is fixed to 1. provides different vortex profiles for varying m. where m = 1. which shows the vorticity of the vortex at t = 0. 1993). are discretised using a second-order linear interpolation.5 the velocity profile of this Taylor vortex can be derived (Panton. the following schemes are used: Gamma (Jasak et al. 2005).

Any excess or lack in numerical diffusion may become visible.0). One major drawback of these methods is that the vortex looses symmetry after 5 s. The Koren and Van Leer limiters are slightly more .10(c). without the presence of a convection free-stream. The velocity profiles and total energy are monitored to identify the accuracy of the flow solver.and Y -direction for the different face interpolation schemes. 0.01 and the velocity vector to u = (1. 0. Besides the evolution of the velocity profile the total energy is shown in figure 2. The total energy is calculated as N Etot = i=1 0.10 Code validation and verification 37 Figure 2. is that the total energy is increasing for the SuperBee flux limiter. where i represents the cell index. Starting from this initial Taylor vortex solution the flow diffusion is solved.10(b) show the initial and final velocity variations in X.2. figure 2. The first important observation. The velocity profile as well as the total energy of both Gamma and linear interpolation schemes are similar.10. N the total number of cells and ui the velocity in cell i.0. which is a measure for the diffusion.5|ui |2 . from figure 2. without a significant amount of diffusion. This scheme clearly introduced a large amount of negative numerical diffusivity which causes the Taylor vortex to grow.10(b).9 Initial solution of a Taylor vortex on an Cartesian mesh. Results Figure 2.0. The flow solver solves for 40 seconds such that the temporal effect of the different face interpolation schemes on the shape and magnitude of the velocity profiles can be compared.10(a) and 2. which is not physical. is fixed to Re = 100 by setting the kinematic viscosity to ν = 0.

94 0 2 4 6 Gamma Koren SuperBee van Leer Linear (b) 10 8 Time [t] 12 14 16 (c) Figure 2.6 U velocity [m/s] 1 3 Initial solution Gamma Koren SuperBee van Leer Linear 1 V velocity [m/s] 0.10(c).5 4 3.02 1 0.5 Y coordinate [m] 5 4. such that the vortex was convected through the entire domain.96 0.38 6 5. shown in figure 2.5 -1 3 -1 -0.6 -0.0. see figure 2. The Reynolds number was fixed to Re = 100 by setting a kinematic viscosity to ν = 0. the discretisation of the face interpolation is varied and all other discretisation terms are fixed to second-order linear interpolation.08 1.5 5 4 X coordinate [m] 5. while (b) provides v(x). The Cartesian mesh resolution was set to (400 x 100) and the max Courant number was equal to Comax = 1. Similar to the vortex decay problem.01 and an inlet velocity . for different face interpolation schemes.2 0.10 Velocity and total energy variations of a decaying Taylor vortex.2 0.12 1. The Van Leer limiter is called shape preserving and provides good results for both vortex decay and convection as will be seen in the next section.04 1.5 6 (a) 1.2 -0.11. but provide similar results. Vortex convection The second validation case concerns a Taylor vortex. The flow solver solved the governing equations for 20 s. at t=20 s.1 Total energy [m2 /s2 ] 1. (a) shows u(y). diffusive.98 0.6 Finite volume discretisation Initial solution Gamma Koren SuperBee van Leer Linear 3.2 -0.6 0. which is convected through a channel with dimensions (20 x 5).5 4.06 1. (c) shows the decaying total energy due to numerical diffusion.

2004). so a Reynolds number Re > 47 needs to be chosen. This Taylor vortex solution is used as the initial solution for the convection validation case. smearing the vortex.0. and more symmetrical vortices (Juntasaro & Marquis.and Y -direction through the vortex core at t = 10 s.10 Code validation and verification 39 Figure 2. the Van Leer limiter slightly outperforms the Koren limiter. to u = (1. 2004. Furthermore. depending on the convection velocity.12(b) show. Additionally. This effect increases with time and therefore these two methods are not appropriate to study the vortical wake patterns in insect flight. the Koren and Van Leer limiters are very close. Extra diffusion is clearly visible after t=14 s when the vortex approaches the outlet boundary. 2004). When looking in real detail to these results. Results First of all. since the integrated energy is close for all schemes. which is not relevant for unsteady insect flight. 1998).11 Initial solution of a Taylor vortex on an Cartesian mesh. At Reynolds numbers less than Re = 47 the flow exhibits steady behaviour (see Williamson. representing a channel. respectively. The second-order linear and Gamma schemes lead to an overshoot of the velocity in Y -direction. such that this scheme was used throughout this study.12(a) and 2. the SuperBee scheme possibly still has negative numerical diffusion. 0. the effect of different face interpolation schemes is compared.2.12(c) shows that the total energy in the entire domain is decreasing with time for all schemes. when taking a Reynolds . another observation can be made. the velocity in X. Again. as was seen in the results for vortex decay. The Van Leer limiter leads to smoother. On the other hand. two cylinder example problems are defined at sufficiently low Reynolds numbers. except for the SuperBee scheme. without overshoots and with a proper symmetry preservation of the vortex. Kuzmin & Turek. 0.0. Besides the possibly negative numerical diffusion in the SuperBee scheme (Juntasaro & Marquis.10.0). At sufficiently large grid resolution and time-step size. figure 2. the total energy does not provide more information about the accuracy of the scheme. 2.2 Validation using cylinder flows To validate the accuracy of the flow solver for unsteady and vortical flow. figure 2. Since convection induces physical diffusion.

The velocity in X-direction u(y) is shown in (a) at time t=10 s. with periodic force histories. Figure 2. vortical flow around flapping wings.2 50.6 0. the flow becomes turbulent and additional turbulence modelling becomes necessary. (c) shows the total energy for the convected vortex. The main parameter selected for comparison is the time-averaged drag coefficient. in the range 100 ≤ Re ≤ 200. where D is the cylinder diameter.12 Velocity profiles of a convected Taylor vortex. one concerns the flow around a stationary cylinder at Re = 150 and the other involves a transversely oscillating cylinder at Re = 185 (Guilmineau & Queutey.2 -0.9 Total energy [m2 /s2 ] 50.40 5 4.5 4 Y coordinate [m] 3.1 50 0 2 4 6 (b) Gamma Koren SuperBee van Leer Linear 8 10 12 14 16 18 20 Time [t] (c) Figure 2.4 0.2 1 U velocity [m/s] 1.5 0 0. 2002). above and below the cylinder.6 Gamma Koren SuperBee van Leer Linear 0.3 -0. number larger than about Re = 185. Therefore.4 -0. v(x).14 shows the computational domain used for both validation cases.2 0. Since the main objective of this research is to solve for unsteady.5 3 2.4 50. (b) illustrates the velocity profile in Y -direction as a function of the X-coordinate.5 50.5 0.5 2 1.3 V velocity [m/s] 0.4 0. the boundaries are located at 10D before.4 1. The flapping wing simulations are performed in the laminar flow regime Re = O(100).5 12 Finite volume discretisation Gamma Koren SuperBee van Leer Linear 13 14 16 15 X coordinate [m] 17 18 (a) 51 50.6 50.8 1. The outflow . two validation cases were selected.7 50.3 50.8 50.1 0 -0. the characteristics of that kind of flow needs to be present in the validation cases.5 1 0.1 -0. which is well-documented in literature.

13(a) shows the relation between the Strouhal and Reynolds number.183 is obtained for a stationary cylinder at Re = 150. A Strouhal number of St = 0. according to (Williamson. Bos et al. dealing with the flow around a static cylinder at Re = 150 is inherently laminar and unsteady. (b) relates the viscous drag coefficient (◦). e. Flow around a stationary circular cylinder The first case. where D is the cylinder diameter. boundary is located at a distance of 40D.18 0.4 1.2 0. Henderson (1995) performed a spectral element numerical study which is used as the baseline reference for this case. figure 2.2 0 101 102 Reynolds number [-] 41 103 (a) (b) Figure 2. The . resulting in a periodic vortex wake.13(b) shows the results from an extensive study performed by Henderson (1995) to identify a relation between drag coefficient and Reynolds number. The flow is from left to right and the inlet boundary is located 10D upstream.g.17 0. Figure 2. which is related to the vortex shedding frequency.10 Code validation and verification 0. the outlet 40D downstream and the upper and lower boundaries are located 10D from the cylinder surface.183 for Re = 150 at which the stationary cylinder validation case is performed.8 0.6 0. Previous studies.16 0. 1998).14 0.2 1 0.19 Strouhal number [-] Drag coefficient [-] 0. Lentink & Gerritsma (2003). Figure 2.12 40 60 80 100 120 140 160 180 200 Reynolds number [-] 2 1. (2008).6 1.13 Relationships between Reynolds number. The dot (•) shows a Strouhal number of St = 0. pressure drag coefficient (•) and total drag coefficient (△) to the Reynolds number (Henderson. with sizes 25k.2. showed that boundary effects are minimal at this domain size.15 0. This grid. Strouhal number and drag coefficient.4 0.8 1. (a) shows the relation between Strouhal number. 50k and 100k is used to validate the accuracy of the flow solver. Additionally.13 0. 1995). and the Reynolds number (Williamson.14 Computational grid around a cylinder. 1998).

25k and Comax = 2. the timeaveraged drag coefficient is determined and compared with literature in table 2. it can be seen that the Strouhal number matches the value from Henderson (1995) even more closely than the time-averaged drag coefficient. The drag and lift coefficient are respectively defined as: CD = D 1 2 2 ρUref . it can be observed that the drag coefficient of the coarsest case. When the grid is refined and the time-step decreased. The time-step is systematically decreased according to a maximal Courant number corresponding to Comax = 2.1. which reveals the presence of the Von K´rm´n vortex street a a behind the stationary cylinder. From these periodic forces. the grid size is varied from 25k.15 Vorticity visualisation of the Von K´rm´n vortex street. 1. Vorticity ω = ∇ × u is used to identify the vortical structures. it is seen that the solution decreases asymptotically.2. which are clearly visible at a Reynolds number of Re = 150. The alternating vortex shedding pattern leads to periodic force variations which can be visualised using CL -CD limit cycles.19. the calculated time-averaged Strouhal number (shedding frequency) is shown in table 2. 50k and 100k. The drag on the finest grid with the smallest time-step is about 2.16. These limit cycles are determined by taking the periodic part of the force histories as shown in figure 2. of which the frequency depends on the Reynolds number.63%. Results To illustrate the flow behaviour.0.42 Finite volume discretisation Figure 2.0. obtained for different mesh resolutions (25k.25.333. figure 2. 5. has the largest difference with literature. CL = L 1 2 2 ρUref . In addition. for example. From that table. From this table.15 shows the instantaneous vorticity (ω = ∇×u) contours.0. The differences of the mean Strouhal number with literature ranges from . 0. The results. (2.5 and 0. The flow around a a a stationary cylinder shows a periodic vortex street.7% larger compared to the value obtained by Henderson (1995). 50k and 100k) are shown in figure 2. In practical applications such a vortex street exists behind struts in water.29) In order to investigate the temporal and spatial convergence of the solution. resulting drag coefficient at Re = 150 is found to be CD = 1.

The periodicity of the flow around a stationary circular cylinder at Re = 150 is illustrated by a CL -CD limit cycle. which is considered sufficiently small. The extrapo- .0 25k 50k 100k Henderson (1995) 1. As can be seen. From table 2.6 1. is the extrapolated values.6 0.46 Grid Size=25k Grid Size=50k Grid Size=100k (a) Comax = 2.385 1. figure 2.369 (+2.2.4 1.64%.385 1. If the flow solver is developed in a numerically consistent way.372 (+2.34 1.28% to 1. 3.42 1.370 1. 25k. 2002).0 (b) Comax = 1. (a).4 -0.4 1.34 1. Comparison of the time-averaged drag coefficient for different grids.25. and Comax = 1.2.5 and 0. 1. Drag coefficient.2 -0. 50k and 100k.1 Drag comparison for the flow around a stationary cylinder.42 1.2 0 -0. In order to illustrate if the solution converges. which should be the case.44 1.0.36 1. Besides a comparison solely with literature it is important to investigate the convergence of the solution with increasing grid resolution and decreasing time-step size.1 and 2.9%) 1.4 -0.5%) 1.408 1. Additionally.46 Grid Size=25k Grid Size=50k Grid Size=100k 0.2 0 -0. the solution decreases asymptotically.372 1.376 1. it is clear that the flow solver produces results which are sufficiently close to the values obtained from literature. and different time-steps corresponding to a maximal Courant number.38 Drag coefficient [-] 1.392 1.38 Drag coefficient [-] 1.0 (b).377 Comax 0.4 0.2 -0.6 1.10 Code validation and verification 43 0.0 Figure 2. using Richardson’s extrapolation (Ferziger & Peric. the flow solution should converge to an asymptotically value with increasing spatial and temporal resolution. The last value in these two figures. Comax = 2.373 1.7%) Table 2.4 0. 0. the grid convergence of the solution can be observed for Comax = 2.25 1. CD Mesh size 2.6 Lift coefficient [-] Lift coefficient [-] 0.36 1.393 1.0.0.5 0.44 1.333 Richardson 1.0 1.16 Lift-Drag limit cycles of the flow around a stationary cylinder.17 shows the time-averaged drag coefficient for increasing mesh resolution (for each time-step) and for decreasing time-step (for each mesh).381 1.379 (+3.381 1.

25 1.187 0.0 1.44 Finite volume discretisation 1.39 1.5 and 2.4 1. The Richardson extrapolated values are plotted at ∞. 0.188 0.188 (+2.187 (+2. 1.64%) 0.30) and should be 2 for a second-order discretisation scheme. (b) shows the drag coefficient with decreasing time-step size for the different grid sizes. the value of p lies between 1.188 0. St Mesh size 2.37 1.0 0.30) where φextrap is the extrapolated value.37 Grid Size=25k Grid Size=50k Grid Size=100k 4 Spatial resolution ∞ 1. lated value is obtained using the following expression: φextrap = φf ine + φf ine − φcoarse . the order of the scheme p can be obtained using (2.2 Strouhal number comparison for the flow around a stationary cylinder.2 it can be deduced that for both second-order spatial and temporal schemes.0 25k 50k 100k Henderson (1995) 0. Strouhal number.28%) Table 2.36 1 2 4 8 Temporal resolution ∞ (a) (b) Figure 2.4 1. Comparison of the time-averaged Strouhal number for different grids. .39 1.186 (+1. However.38 1.25.73%) Comax 1.41 1.0. 25k. (a) shows the drag coefficient with increasing grid refinement level for different time-steps.38 1. Theoretically.188 0.17 Time-averaged drag coefficient as a function of spatial and temporal resolution for the stationary cylinder.0 0. and different time-steps corresponding to a maximal Courant number.188 (+2. 2002) which is not the case for the cylinder simulations. φf ine and φcoarse are the two most accurate solutions available.189 (+3.1 and 2. Comax = 2. 2p − 1 (2.73%) 0.188 0.41 Drag coefficient [-] 1.19%) 0.5 0.36 1 2 Drag coefficient [-] Comax Comax Comax Comax = = = = 2. 50k and 100k.189 (+3. A flow solver is numerically consistent if the flow solution converges with increasing grid resolution and decreasing time-step size. This is true for uniform Cartesian meshes (Ferziger & Peric.28%) 0. from table 2.188 0.5 0.5 and 0.25 0.183 0.0.

5 0 -0. For all meshes.58 0.18 Time-averaged lift convergence with grid resolution and time-step size for the stationary cylinder.0. 1.56 0. figure 2.7% to 3.0 1. are smaller than 1.2.25 0.5 Force coefficient [-] Drag 1 Lift 0. it can be determined that the differences varies from 2.5 and 0.62 0.0 0.5 0.54 1 2 4 Spatial resolution ∞ Comax Comax Comax Comax = = = = 2.0 to result in an accurate solution.0% for maximal Courant numbers Comax = 1.5 0. the grid size is 25k and the time-step was varied according to a maximal Courant number of Comax = 2.5% compared to literature with decreasing mesh resolution.58 0.64 0.25. Flow around a transversely oscillating circular cylinder The last validation case concerns the flow around a transversely oscillating cylinder .1.54 1 2 4 8 Temporal resolution Grid Size=25k Grid Size=50k Grid Size=100k 45 ∞ (a) (b) Figure 2. 25k. it seems sufficient to consider the mesh of 50k and Comax = 1.62 0. (a) shows the average lift coefficient amplitude with increasing grid refinement level for different time-steps. (b) shows the time-averaged lift coefficient amplitude with decreasing time-step for the different grid sizes.64 Lift coefficient amplitude [-] Lift coefficient amplitude [-] 0.25. 0.0 1.0 0.25 Additionally to the temporal and spatial convergence of the drag coefficient.0. 1.56 0. 0. Concerning the extrapolated values of the drag coefficient.6 0.10 Code validation and verification 0. Lift and drag coefficients for a stationary circular cylinder case at Re = 150. these settings were used throughout the present research.5 and 0.6 0. Comax Comax Comax Comax = = = = 2. the differences in drag coefficient compared to the extrapolated values.5 0 50 100 150 Time [s] Figure 2. from table 2.18 shows the convergence of the lift coefficient amplitude in order to prove the consistency of the flow solver. Therefore.0.19 Forces around a stationary cylinder.

Figure 2.20(b).0. the mesh quality should be high in terms of non-orthogonality and skewness. The value for all time-steps was within 2% compared to the extrapolated value. zerogradient (Neumann).2. (2. The frequency was set to fe = 0. 2. From this figure it is obvious that the flow solution converges with decreasing time-step. 0. which is considered to be sufficiently accurate. which employs amplitudes of several chord lengths. An amplitude of 0. both mesh quality measures. This statement is confirmed if the time-averaged drag coefficient is plotted in figure 2. Although. The discretisation concerns arbitrary polyhedral meshes. four different types of boundary conditions were specified. with D the cylinder diameter. Using those boundary conditions.8 times the natural shedding frequency of a stationary cylinder at a Reynolds number Re = 185. using the present flow solver. performed by Guilmineau & Queutey (2002). Comax = 2. Results To assess the accuracy of the flow solver applied to flapping wings.20(a) shows the limit cycle results. the discretised Navier-Stokes equations can be solved using a PISO pressure velocity coupling in combination with an Arbitrary Lagrangian Eulerian (ALE) approach if dynamic meshes are used. for the 50k grid. fixed-value (Dirichlet). Guilmineau & Queutey (2002) found a drag coefficient of CD = 1. . symmetry and moving-wall-velocity. corresponding to 0. 1. such that this method can easily be applied to a wide variety of problems with complex geometries. the following values are used to show that the flow solver converges to an asymptotic solution for an oscillating cylinder as well.2D. The oscillating direction is perpendicular to the free-stream direction.11 Conclusions This chapter has presented the finite volume discretisation of the incompressible laminar Navier-Stokes equations. therefore that mesh is also used for this test case.5 and 0. The cylinder motion is defined as y(t) = −Ae sin(2πfe t).25 were considered. with second-order temporal and spatial discretisation.31) where the amplitude is set to Ae = 0. the results of an oscillating cylinder case were compared with literature. To obtain accurate and efficient results.0 was found to be sufficient.46 Finite volume discretisation at Re = 185 using a numerical study.154. but sufficient to investigate the moving wing capabilities of the numerical model. The different terms of the governing equation were discretised using second-order schemes and different flux splitting methods were described concerning the face interpolation. a time-step corresponding to Comax = 1.0. Previously conducted simulations on stationary cylinder flow showed that a mesh of 50k provides a sufficiently accurate solution. In order to solve the flow on a computational grid.2D is relatively small for insect aerodynamics.

Fluent has already been tested for various flows in literature.2.3 1. Vortex decay and convection were used to study the influence of the face interpolation scheme with different flux limiters.0 0.1 1. stationary and transversely oscillating cylinder flows were used to successfully prove spatial and temporal convergence. this chapter presented a validation of OpenFOAM for a number of relevant test cases.5 0.0 1.26 -0.275 1. In principle.265 1. concerning vortex decay and convection. relevant for flapping insect flight.25 Drag coefficient [-] 1.20 Forces for the flow around an oscillating cylinder. (a) shows the CD -CL limit cycles for different time-steps on a 50k mesh.11 Conclusions 0.15 -0. OpenFOAM too. but not for low Reynolds flows. The Richardson extrapolated values are plotted at ∞.27 47 -0. It was found that the Van Leer flux limiter provides the most accurate results. In addition.28 1.2 1.05 -0. (b) shows the time-averaged drag coefficient with decreasing time-step for the 50k mesh. The present research used the commercial CFD solver Fluent and the open-source CFD code OpenFOAM . Therefore.2 0.25 Drag coefficient [-] 1.2 1.35 1. .05 0 Co Co Co Co = = = = 2.255 1 2 4 8 Temporal resolution ∞ (a) (b) Figure 2. the described discretisation method and solution procedure is applicable to different commercial and non-commercial CFD solvers.15 Lift coefficient [-] 0. both were briefly described.1 0. It is concluded that the open-source solver OpenFOAM provides an accurate and efficient framework to investigate the flow around flapping wings at low Reynolds numbers.

.

a mesh deformation routine based on the interpolation of radial basis functions is introduced. two techniques are applied. In order to increase the efficiency of this method. . A test case. Mech. based on solving the Laplace and solid body rotation stress equations. based on boundary coarsening and smoothing of the radial basis function. Engrg. In order to use mesh deformation techniques to investigate flapping wing aerodynamics it is necessary to maintain a high mesh quality for relevant wing kinematics. The main difficulty is to maintain high mesh quality when the wing exhibits large translations and rotations. compared to a function with compact support. Different mesh deformation techniques are compared in order to identify their applicability for cases with flapping wings. (January 2010). Additionally. it is shown that this method can be used for mesh deformation for three-dimensional flapping wings and can handle flexing boundaries. The mesh quality. Appl. The radial basis function method can be used with different basis functions with global or compact support. concerning a moving two-dimensional block. is found to be highest when the thin plate spline is used as a basis function. is used to show that the radial basis function method provides superior mesh quality compared to the Laplace mesh motion solver. Meth. In addition to existing mesh motion methods. based on skewness and non-orthogonality. but is computationally more expensive.CHAPTER 3 Mesh deformation techniques for flapping flight Submitted to Comput. A globally supported basis function results in the highest average mesh quality.

The disadvantages of the immersed boundary method are the difficulty to capture the boundary layer and to meet the requirements for mass and momentum conservation. it is possible to solve this mathematical case by changing the boundary conditions as if the boundary was deforming or to deform the complete mesh. which defines a moving boundary on a stationary Cartesian background mesh. like the rigid body motion. 2002). In literature. using the discretisation methods from chapter 2. it is important that the internal mesh preserves its validity (no negative cell volumes) and quality (cell orthogonality and skewness). existing methods will be compared and an improved mesh motion solver is explored and incorporated. However. In order to assess the quality of a mesh motion solver. a computational mesh is necessary.1 Introduction In order to solve the governing equations. the equations are iteratively solved on the computational domain. For . Preservation of high mesh quality is necessary to solve the flow in an accurate and efficient way. e. The second method to deal with a deforming boundary. Therefore. If the shape of the domain boundary is time-varying. if this happens to be known beforehand. current mesh motion techniques are not fully suitable to cope with the mesh deformation around an object which moves with a large change in rotation. the one used in the current thesis. Examples of such interaction problems are fluid-structure-interaction cases like blood flow through arteries or deforming flags. where the flow is being influenced by a changing boundary shape. The quality of the resulting mesh is defined by the non-orthogonality and skewness of the finite volume cells. Using prescribed initial and boundary conditions. This may be caused by imposed external effects. In order to deal with moving objects. The first method is called the immersed boundary method (Peskin. A robust mesh motion solver is defined such that it needs little to no user-input. An appropriate implementation of this method is not trivial. Additionally. In engineering. there are numerous computational fluid dynamics (CFD) problems in which the flow solution involves geometrically deforming boundaries. the computational mesh points are moved in order to keep track of the changing location of boundary points. robustness is used to identify if a method is user-friendly. efficiency and robustness. is the use of a mesh motion solver which moves the internal mesh points. Efficiency is a measure of the used computation time to calculate the displacements of the mesh points at the new time-step. quality.50 Mesh deformation techniques for flapping flight 3. three different aspects need to be formulated. Current CFD solvers incorporate different mesh motion techniques in order to change the location of the internal mesh points according to the varying domain shape. several mesh deformation methods have been presented using different approaches to calculate the motion of the computational mesh points. or a prescribed body deformation. When using a mesh motion solver. a flapping wing. A simplified type of interaction is that which concerns a one-way coupling. The boundaries of the computational domain can be either fixed or deforming.g.

which assumes static equilibrium for small deformations of a linear elastic solid (the mesh is treated as if it was a solid). is called the spring analogy (Batina. which acts as a stiffness of the system of equations. Helenbrook. which significantly improves the mesh quality for meshes subjected to large boundary translations and rotations. high resulting mesh quality was only achieved by specifying a problem specific spring stiffness. Another choice of equations is made by Johnson & Tezduyar (1994) who used the pseudo-solid equation. for example Transfinite Interpolation (Wang & Przekwas. Depending on the method. Jasak et al. the parc allel implementation of these methods is fairly straightforward. Concerning the governing partial differential equations. The latter method is often used in the Arbitrary Lagrangian Eulerian formulation of finite element codes. Farhat et al.1 Introduction 51 structured meshes. Radial basis functions (RBF) are commonly used in literature to interpolate . possibly in combination with mesh refinement. Therefore. Therefore. The most popular mesh deformation method. 1990) where the point-to-point connection of every two neighbouring mesh points is represented by a linear spring. 2004).. (1998). 2007. used in the finite volume code of OpenFOAM (Weller et al. 2000).. However. Other mesh deformation techniques involve solving a partial differential equation on the complete field of internal mesh displacements for given boundary point displacements. Using an additional mapping (Wang. the focus is put on mesh deformation techniques which can be applied to unstructured meshes containing arbitrary polyhedral cells. Degand & Farhat (2002) proposed a method to incorporate torsional springs to improve the robustness of this method. 2004. this method proved to lack robustness. Dwight (2004) modified this method to incorporate rigid body rotation. One major drawback of these methods is that they all fail in maintaining high mesh quality when the boundary points move with high rotation angles. In (Boer de et al. They interpolated the displacements of the boundary points along grid lines through the entire computational mesh to find the displacements of all interior mesh points. Since.3. as was observed by (Blom. this influences the efficiency. applicable to both structured and unstructured meshes. based on the use of radial basis function (RBF) interpolation to obtain the mesh point displacements.. 2003) are often used o in combination with a constant or variable distance-based diffusion coefficient to improve the mesh quality. especially on arbitrarily unstructured meshes. already available in existing CFD codes (Jasak & Tukovi´. Since these methods solve a partial differential equation on the complete field of internal mesh points the existing iterative solvers can be used. These methods are perfectly suitable for structured but unsuitable for unstructured grids. 1998. the Laplace and bi-harmonic operators (L¨hner & Yang. 2000b) the mesh quality can be improved significantly if the boundary is subjected to significant rotation and deformation. a new mesh deformation method was developed and incorporated. Additionally. Jasak.. Bos et al. a variable diffusivity field needs to be defined. there are efficient techniques available to deform the mesh. 2010a) it was shown that radial basis function interpolation could improve mesh quality considerably. 2009). unstructured meshes are used for complex geometries. 1994). 1996.

section 3. 2007. the idea was born to interpolate the boundary mesh to all computational mesh points.2 Different mesh deformation techniques When a moving mesh problem is considered. described in chapter 2. radial basis functions were only applied to mesh motion concerning the boundaries in multi-block meshes (Potsdam & Guruswamy. A preliminary study was performed by Boer de et al. geophysics. error estimation. a distinction can be made between the motion of the boundary points and the motion of the internal (fluid) points. they have been used in computer graphics. the shape of the computational domain is varying in time. Both Laplace and pseudo-solid mesh motion techniques are commonly used within the OpenFOAM community. two techniques are implemented to improve its efficiency. or it is part of the solution. This is the subject of section 3. which is . two more relevant test cases were considered.b) have been carried out to improve the efficiency of mesh motion based on radial basis function interpolation. In addition to the simplified moving block. The displacement of the boundary points can be considered to be given.6. They noted that applying this method to all mesh points would be too computationally expensive. The application of radial basis functions is very wide. Since the application of RBF’s to interpolate from and to the boundary mesh was very accurate and efficient.2. Rendall & Allen. Therefore. 2001). (2007). either it is externally defined. Since mesh deformation using RBF interpolation results in high quality meshes even with large body rotation angles. but also in coupled simulations as in fluid-structure-interaction. 2007) used radial basis function interpolation to couple two non-matching meshes at the interface of a fluid-structure interaction computation. a prescribed rigid body motion. based on the Laplace equation with variable diffusivity and a modified pseudo-solid equation. In order to assess the mesh quality. An RBF interpolation function is used to transfer the known boundary point displacements to the fluid boundary mesh. Since the radial basis function mesh motion method is computationally expensive.e. the conclusions are drawn in section 3. In this chapter two existing mesh deformation techniques are compared. 3. Only recent studies (Jakobsson & Amoignon. (Boer de et al.52 Mesh deformation techniques for flapping flight scattered data.4. section 3. The resulting mesh quality of the different mesh motion methods is studied using a two-dimensional test case of a block which translates and rotates. because of their good approximation properties which is discussed by Buhmann (2000). Finally. Previously. i. The mesh quality is investigated using a visualisation and histograms of the skewness and non-orthogonality criterion.3 discusses two different criteria. with radial basis function mesh motion.5 deals with two techniques to increase its efficiency. one using a three-dimensional flapping wing and the other involves a two-dimensional flexing airfoil.. 2008c. based on nonorthogonality and skewness. These different mesh motion methods are described in section 3.

2002). The internal point motion influences the solution only through the discretisation errors (Ferziger & Peric. one c could also have used an exponentially decreasing diffusion coefficient or a diffusion coefficient related to the mesh deformation energy. This variable diffusion coefficient can be chosen such that a region next to the deforming or moving boundary closely moves with the boundary. other type of equations were used.2. since it decreases the robustness of the method. To maintain robustness. m = 2. 3.1) The resulting mesh quality strongly depends on the chosen γ(r) function. which depends on the distance from the moving boundary. The internal point motion can be calculated using different methods. . decreasing diffusion coefficient. However. provided that the ALE formulation is correctly implemented. rm (3. appears to be very problem dependent and thus optimisation of γ(r) seems not cost effective. The current research uses a γ(r) function like equation (3. the internal points need to be moved in order to maintain mesh quality and validity. which is a variant of the linear stress equation (Dwight. as will be shown in the next section. When the mesh motion is governed by the Laplace equation. it is also possible to define γ(r) for every internal mesh cell for all time-steps independently. 2004).3. Ideally. such that existing iterative solvers can be used efficiently. the given boundary point motion may be arbitrary and non-uniform. This. a user input is not desired. The resulting mesh contains less cell quality deterioration next to the boundary. In addition to the freedom of choosing a diffusion function. 2004). namely the Laplace equation and the solid body rotation stress (SBR Stress) equation. where x is the displacement field and γ the diffusion coefficient. in the current work we use a quadratically. however. The nature of the Laplace equation is that the point displacements will be largest close to the moving boundary and small at large distance.2 Different mesh deformation techniques 53 the case in fluid-structure interaction problems.1 Laplace equation with variable diffusivity One can think of a deforming computational domain as if it was a solid body undergoing internal stresses given by the Piola-Kirchhoff stress-strain equation (Baruh. Additionally. which decreases with the radius r from the deforming boundary as follows: γ(r) = 1 .1). This leads to the following definition of the Laplace equation: ∇ • (γ∇x) = 0. Therefore. According to the given boundary point motion. 1999). A mesh motion method based on one of these equations is computationally cheap since the resulting matrix system is sparse. this method needs the specification of a variable diffusivity. That equation is non-linear and thus expensive to solve using existing numerical techniques. which was found to provide efficient and a smooth mesh motion (Jasak & Tukovi´.

The constants can be related to Young’s modulus. which is treated as if it was a linear solid. E. (3. the following solid body rotation stress equation is obtained: ∇ • (γ∇x) + ∇(γ(∇x − ∇xT )) − λ tr(∇x) = 0. In order to deal with large rotations.2.2 Solid body rotation stress equation The second method to deform the mesh is based on the linear elasticity equation and is called the solid body rotation stress (SBR Stress) equation (Dwight. meaning the material contraction ratio as it stretches. which are a e property of the elastic material.54 Mesh deformation techniques for flapping flight 3. 1994) usually applied to the points of multi-blocks.4) ǫ = (∇x + ∇xT ). may be written as ∇ • σ = f. In Dwight (2004) an extra term was added to obtain the following strain relation: ǫ= 1 (∇x + ∇xT + ∇xT · ∇x). µ= . a newly implemented mesh motion solver is based on radial basis function interpolation. such that it can be used for flapping wing simulations.4) does not allow for rotations. it is also possible to explicitly define the point motion using interpolation techniques. Although equation (3. The equation of linear elasticity. where x is the position of an internal mesh point. 2 (3. (3. 1999). 2004). there is nothing against changing this strain equation such that rigid body rotations are allowed.3) where tr is the trace and λ and µ are Lam´ constants (Baruh.4 it is shown that both previously described methods maintain high mesh quality for problems with limited boundary rotation. which is given by the following constitutive relation: σ = λ tr (ǫ)I + 2µǫ. The stress tensor σ is given in terms of the strain.6) allows for rigid body motion and is still linear and therefore the computational costs are of the same order as the costs necessary to solve the Laplace equation.5) Combining equations (3.2).5). such that standard iterative techniques can be used. 2 defines the relative change in length. However. like the transfinite interpolation (Wang & Przekwas. In section 3.6) where γ is a similar diffusion coefficient as in equation (3.2) where σ is the stress tensor and f the acting force vector.1). Equation (3. as E νE . Solving the Laplace or the SBR Stress equation leads to a sparse system of equations. λ= (1 + ν)(1 − 2ν) 2(1 + ν) where ν is Poisson’s ratio. like the pre-conditioned Conjugate Gradient (PCG) method. . valid for small displacements. together with λ = −E and µ = E. (3.3) and (3. The following equation: 1 (3. (3.

which is difficult to solve using standard iterative techniques. The values for the coefficients γj and the linear polynomial can be obtained by solving the system: Φbb Qb γ ∆xb = (3.3 Radial basis function interpolation In the current work we use radial basis function interpolation to find the displacements of the internal fluid points for given boundary displacements.8) leads to a dense matrix system. j=1 which holds for all polynomials p with a degree less or equal than that of polynomial q. The interpolation function s(x) describing the displacement of all computational mesh points. a linear polynomial can be used (Beckert & Wendland.2. Here ∆xbj contains the known discrete values of the boundary point displacements. The possibilities of solving the system in a more efficient way are discussed in section 3. Together with the additional requirements: Nb γj p(xbj ) = 0.. the γj values can be determined (Boer de et al. The polynomial q is defined by the coefficients γj which can be defined by evaluating the interpolation function s(x) in the known boundary points: s(xbj ) = ∆xbj . Nb is the number of boundary points and φ is a given basis function as a function of the Euclidean distance ||x||. 2007). If the basis functions are conditionally positive definite of order m ≤ 2.2 Different mesh deformation techniques 55 3. is approximated by a sum of basis functions: Nb s(x) = j=1 γj φ(||x − xbj ||) + q(x). 2007).5.8) . The minimal degree of polynomial q depends on the choice of the basis function φ (Boer de et al. In general. A unique interpolant is given if the basis function is a conditionally positive definite function.3. Φbb an nb × nb matrix contains the evaluation of the basis function φbi bj = φ(||xbi − xbj ||) and can be seen as a connectivity matrix connecting all boundary points with all internal fluid points. 2001). . (3. by doing a LU decomposition. ybj .. it needs to be solved directly. zbj ]. (3. We only applied basis functions that satisfy this criterion.7) where the known boundary value displacements are given by xbj = [xbj . Therefore. β the four coefficients of the linear polynomial q. Qb is an (nb × (d + 1)) matrix with row j given by [ 1 xbj ]. A consequence of using a linear polynomial is that rigid body translations are exactly recovered. 0 β QT 0 b where γ is containing all coefficients γj . q is a polynomial.

Instead. the proper radial basis function needs to be chosen to satisfy the need for robustness. no partial differential equation needs to be solved and the evaluation of all internal boundary points is straightforward to implement in parallel. Radial basis functions with compact support From literature. such that no mesh connectivity is necessary. In contrast to the Laplace and SBR Stress methods. Functions with compact support have the following property: φ(x/r) = f (x/r) 0 0 ≤ x ≤ r. a very large support radius leads to a dense matrix system. 4. which are then used for step two. These polynomials are chosen in such a way that they have the lowest degree of all polynomials that create a C n continuous basis function with n ∈ {0. various radial basis functions are available. (3. it must be noted that larger values for the support radius lead to more accurate mesh motion. which is considerably smaller than other techniques using the mesh connectivity. only the internal mesh points inside a circle (two-dimensional problem) or a sphere (three-dimensional problem) with radius r around a centre xj are influenced by the movement of the boundary points. 6}. Two types of radial basis functions can be distinguished: functions with compact and functions with global support. with Nin the total number of mesh points. where f (x/r) ≥ 0 is scaled with a support radius r. since no mesh connectivity is needed. which are suitable for data interpolation.8) gives the values of the necessary coefficients γ and β. When choosing r. 1996). When a support radius is used.1 various radial basis functions with compact support are shown using the scaled variable ξ = x/r. x > r.9) is transferred to the mesh motion solver to update all internal points accordingly. the evaluation using equation (3.8) is ((Nb + 4) x (Nb + 4)). Solving the system (3. 2. while a low support radius results in a sparse system which can be solved efficiently using common iterative techniques. Every internal mesh point is moved based on its calculated displacement. The result of (3.7). The last four are a series of functions based on the thin plate spline .9) ∆xinj = s(xinj ). which is a dimension higher than the total number of boundary points.9). In table 3. this method is not using a variable diffusion coefficient which has to be tuned by the user.56 Mesh deformation techniques for flapping flight When the coefficients in γ and β are obtained they are used to calculate the values for the displacements of all internal fluid points ∆xinj using the evaluation function (3. On the other hand. The size of the system of this method (3. This interpolation function is equal to the displacement of the moving boundary or zero at the outer boundaries. Concerning robustness. The first four are based on polynomials (Wendland. The mesh connectivity techniques encounter systems of the order (Nin x Nin ).

. 2007). the computational domain. i. .e. There are two possible CTPS C 2 continuous functions which are distinguished by subscript a and b. TPS MQB IMQB QB IQB Gauss f (x) x2 √ log(x) a2 + x2 1+x e 1 1+x2 −x2 1 a2 +x2 2 Table 3. which creates C n continuous basis functions with n ∈ {0. Table 3. nr. 1 2 3 4 5 6 7 8 RBF Name CP C 0 CP C 2 CP C 4 CP C 6 CTPS C 0 CTPS C 1 2 CTPS Ca 2 CTPS Cb f (ξ) (1 − ξ)2 (1 − ξ)4 (4ξ + 1) (1 − ξ)6 ( 35 ξ 2 + 6ξ + 1) 3 (1 − ξ)8 (32ξ 3 + 25ξ 2 + 8ξ + 1) (1 − ξ)5 8 80 2 3 1 + 3 ξ − 40ξ + 15ξ 4 − 3 ξ 5 + 20ξ 2 log(ξ) 2 3 4 1 − 30ξ − 10ξ + 45ξ − 6ξ 5 − 60ξ 3 log(ξ) 1 − 20ξ 2 + 80ξ 3 − 45ξ 4 − 16ξ 5 + 60ξ 4 log(ξ) Table 3.g. functions with global support are not equal to zero outside a certain radius. as will be discussed in section 3. Boer de et al. Taken from Wendland (1996). 1996).5. which can be improved by multiplication with a smoothing function. Radial basis functions with global support cover the whole interpolation space.2 Radial basis functions with global support. 9 10 11 12 13 14 RBF Name Thin plate spline Multiquadratic Bi-harmonics Inverse Multiquadratic Bi-harmonics Quadric Bi-harmonics Inverse Quadric Bi-harmonics Gaussian Abbrev. 2} (Wendland. neural networks. 1.. Note that ξ = x/r. Ref. nr. computer graphics (Carr et al.. Radial basis functions with compact support are non-zero within the range of the support radius r.2 shows six radial basis functions with global support which are commonly used in e. Radial basis functions with global support In contrast to the functions with compact support.3. 2003) and for data transfer in fluid-structure interaction computations (Smith et al.2 Different mesh deformation techniques 57 Ref.1 Radial basis functions with compact support. but cover the whole interpolation space. 2000. Radial basis functions with global support generally lead to dense matrix systems.

4 the skewness and non-orthogonality definition were introduced. It is important that the ideal mesh motion solver maintains high quality in terms of skewness and non-orthogonality after mesh deformation. The skewness and non-orthogonality are written to scalar fields such that they can be used for post-processing. 1998). (2007) compared the resulting mesh quality using all radial basis functions from table 3. Therefore. 2003). the absolute and the relative implementation. These mesh quality measures are based on the cell properties such as size. resulting in lower computation costs (see section 3. . leading to optimal initial mesh quality.58 Mesh deformation techniques for flapping flight The MQB and IMQB methods use a shape parameter a. 1999. This method is very efficient since the direct matrix solve.3 Mesh quality measures In section 2. When using the relative implementation it is important to use different techniques to decrease the number of boundary points. this is shown in section 3. like a 180◦ rotation. Wendland.In this method the inverse is calculated at every time-step and the motion is defined with respect to the previous time-step.1 and 3. Absolute and relative radial basis function interpolation In principle there are two different ways to implement this RBF mesh motion method. the functions based on a continuous thin 2 2 plate spline. the mesh quality measures. the relative method is used when very large boundary displacements occur. contain a Cartesian grid around a square box. skewness and non-orthogonality. shape and skewness. The coefficient arrays γ and β are calculated and used to calculate the internal point displacements at all time-steps. Ca and Cb and finally the globally supported thin plate spline. is only performed initially. More information about RBF’s and their error and convergence properties can be found in (Buhmann. orientation. A large value of a gives a flat sheetlike function. which controls the shape of the radial basis function. The value of a is typically chosen in the range 10−5 − 10−3 . The best results were obtained using the continuous polynomial C 2 . which is more expensive than the evaluation. C 1 . 2000. the mesh quality is limited since the coefficients are not defined with respect to the previous time-step. 3.4 these six radial basis functions are applied to our test problem and the best one is used throughout the current thesis. On the other hand.5). it is much cheaper to use the absolute method. First. The absolute method performs a direct solve of the system (3. In order to compare the quality of the different meshes after mesh motion it is important to elaborate on how to interpret those two mesh quality measures (Knupp. In section 3. are discussed to compare the mesh quality for the different mesh motion solvers. Boer de et al. whereas a small value of a gives a narrow cone-like function.2. The test cases used to compare the mesh quality of the different mesh motion solvers.4. For reasonably small boundary displacements.8) only once at the beginning of the simulation.

the more stable. the effect of different radial basis functions is discussed. In both cases.4. On the other hand. 3.4. In addition. three different kinds of mesh motion solvers are described. In the next section. the block is rotated around its translating centre with 57. The outer boundary points are kept fixed. The domain size of the test problem is limited to 25D x 25D and the size of the moving block is 5D x 1D. solving the solid body rotation (SBR Stress) equation and based on interpolation using radial basis functions (RBF).2 deals with the mesh motion around a three-dimensional flapping wing. The first simple case is a two-dimensional block which performs a combination of translation and rotation.1. followed by an example of a two-dimensional flexible moving boundary in section 3. the simulation will diverge within a couple of iterations.0 rad). The maximal value provides an indication if the numerical simulation will be stable and converge at all. If the worst cell quality is too low.3. 3.2. Therefore. both the average and minimal value of the skewness and non-orthogonality mesh quality measures are used to compare the mesh motion solvers. The higher the average quality of the mesh. After this simple model problem. The initial computational mesh is shown in figure 3. a lower value means a higher mesh quality. and that the mesh non-orthogonality. The resulting mesh .4.5D in both X-and Y -direction. should be within 0◦ and 90◦ .0 ≤ 90◦ (3.0 0◦ ≤ fskewness ≤ fnon−ortho ≤ 1.3. based on solving the Laplace equation. section 3. This section introduces with three numerical test cases which were performed to investigate the differences in mesh quality obtained with the different mesh motion solvers.10) When assessing the mesh quality it is important to analyse the maximal and average values. accurate and efficient the computation will be.4 it is shown that mesh skewness should be within 0 and 1. the grid spacing corresponds to 1D in order to obtain a Cartesian grid as can be seen in the figure.4 Comparison of mesh motion solvers In section 3. The two-dimensional block is initially centred and translates 2.1 Translation and rotation of a two-dimensional block The first test case considers a combined motion of translation and rotation to compare the mesh motion solvers under these conditions. Additionally. the average value of the mesh quality measure will provide an indication of the average quality of the mesh.3◦ (1.4 Comparison of mesh motion solvers 59 In section 2. Mesh motion simulations are performed using the different mesh motion solvers and a variation of the radial basis function. such that the influence of the moving boundary points on all internal points could be studied independently. which is an angle. desirable mesh quality bounds are: 0.

which leads to the motion of all internal mesh points. decreasing from the moving boundary. the newly implemented RBF mesh motion solver is used with five 2 2 different functions. The domain size is (25D x 25D) around a block with size (5D x 1D). applicable to this test case. it is best to apply a mesh motion solver. quality. As can be seen in figure 3.60 Mesh deformation techniques for flapping flight Figure 3. From . In cases where large rotation angles occur. Before proceeding to the results. Using those fields. the maximal and average values are obtained as well as a complete visualisation of those mesh quality measures. with these standard methods the mesh quality near the moving boundary is low. coping with the boundary deformation. need to be described. Finally.e. a quadratically decreasing diffusivity coefc ficient. the mesh deformation is largest near the boundary. after the combined translation and rotation. (2007). especially near the leading and trailing edges. CP C 2 .2(a). the boundary conditions need to be set for the motion solver. Figure 3. Following Jasak & Tukovi´ (2004) and Jasak (2009). at every unit spacing. In both cases. The mesh motion method which solves the Laplace equation with a quadratically decreasing diffusion coefficient. i. CTPS Cb and TPS. so the outer boundary points are kept fixed.1 Initial mesh around a moving block. CTPS Ca . based on the assessment performed by Boer de et al. The initial hexahedral mesh with optimal quality around a moving block. a grid point is places such that an optimal hexahedral mesh is obtained. combined with mesh quality histograms. some special settings. is assessed using the fields of skewness and non-orthogonality. As seen in the figure. which is not desirable. is simply not robust enough to obtain high mesh quality when the boundary rotates. CTPS C 1 . Secondly. First. when using the Laplace and SBR Stress mesh motion solver it is necessary to specify the diffusivity coefficient which decreases proportionally to the distance from the moving boundary points. was chosen. The gain in mesh quality by using the SBR Stress method is marginal.2 shows the cell nonorthogonality at maximal mesh deformation for the Laplace and the SBR Stress motion solvers. the boundary conditions on all outer boundary points are set to Dirichlet type with value 0. Laplace and SBR Stress motion solvers.

The cell non-orthogonality is compared of the Laplace (a) and the SBR Stress (b) motion solvers.2(a) and 3.5D in both Xand Y -direction and rotate 57. The rotation of the boundary is very large. since the maximal skewness and maximal non-orthogonality are large.2(b) it can be seen that the cells with a high non-orthogonality occur at a distance from the moving boundary. figure 3. with the RBF mesh motion. occurs in the outer regions of the mesh and all cells are dealing with the boundary displacement. Still. The results of the Laplace and SBR Stress mesh motion solver are very similar. The boundary points translate over a distance of 2. such that it moves according to the body motion. Table 3. figure 3. including different RBF’s. CTPS Ca and CTPS Cb ) and one with global support.3(b) shows the resulting mesh. four RBF’s with compact 2 2 support are used (CP C 2 . the mesh quality near the body surface. This method is very robust but computationally expensive as will be shown in section 3. CTPS C 1 . respectively fsmax = 0. within 6. and the mesh remains valid.1.4 Comparison of mesh motion solvers 61 (a) Laplace (b) SBR Stress Figure 3. which is obtained using the relative implementation of the RBF mesh motion method. Figure 3. In order to neglect the effect of the support .1%. The non-orthogonality is visualised for a two-dimensional block with a combined motion of translation and rotation.3(a) shows the resulted mesh obtained using RBF interpolation using a thin plate spline function.3 shows a quantitative comparison of the resulting mesh quality obtained with different mesh motion solvers. When comparing the cell non-orthogonality with figure 3. needs improvement.2(b) it may be seen that most of the cell deformation.95 and fomax = 72.2 Cell non-orthogonality of Laplace and SBR Stress mesh motion solvers. Improvement can be obtained by specifying a constraint to an inner mesh region. the thin plate spline TPS. The maximal and averaged values of both skewness and non-orthogonality are compared.5.0 rad) around the translating centre.3. 180◦ . especially near the leading and trailing edges. Additionally. Concerning the RBF method.3◦ (1. The mesh quality calculated by the Laplace method is low. lower (blue) is better.

the basis function thin plate spline provides the highest mesh quality for both skewness and non-orthogonality. it is set to r = 75 which is about 3 times the domain size such that the results in table 3.0 rad) for (a) and 180◦ (3. in addition to the non-orthogonality visualisations.3 are independent of the support radius for r > 75. The nonorthogonality is visualised for a two-dimensional block with a combined motion of translation and rotation. Overall. The rotation is 57. The C 2 and TPS RBF’s result in an maximal skewness of respectively −32% and −45% compared to the Laplace motion solver. the RBF mesh motion results in a smooth profile. respectively.4 and 3.3 Cell non-orthogonality of relative and absolute RBF mesh motion solvers. Concerning the average orthogonality. Especially. the globally sup- . while the difference of the other RBF’s is only about 10%. which emphasises the fact that all internal cells are coping with the mesh motion.14 rad) for (b).62 Mesh deformation techniques for flapping flight (a) RBF absolute (b) RBF relative Figure 3.3◦ (1. maximal and averaged.5D in both X. which are considered to result in the worst and best mesh quality. The cell non-orthogonality is shown for the relative and absolute versions of the RBF mesh motion solver. lower (blue) is better. the C 2 RBF is outperformed by the TPS RBF. which is the only mesh motion solver resulting in a lower value compared to the Laplace method. The boundary points translate a distance of 2. 3. For both non-orthogonality and skewness.4.5 show the histograms of cell non-orthogonality and skewness for the Laplace and RBF mesh motion solver. The RBF method provides high mesh quality. radius.2 Flapping of a three-dimensional wing It was shown that high mesh quality was obtained for a simplified two-dimensional test case. by using radial basis function interpolation. figure 3. (a) shows the absolute and (b) the relative implementation. for both C 2 and TPS compared to the other functions. Finally.and Y -direction. Similar results are shown in the table for the average skewness and maximal orthogonality. such that this globally supported function was used for the current investigations.

3.11 0.5 (-17%) (+3%) (+2%) (+5%) (-27%) (fo )ave 20.4 Comparison of mesh motion solvers 63 Method Laplace SBR Stress CP C 2 CTPS C 1 2 CTPS Ca 2 CTPS Cb TPS (fs )max 0.103 0.08 (fs )ave (-) (-4%) (-28%) (+17%) (+14%) (+22%) (-41%) (fo )max 72.5.86 0.4 and 3. while figure 3.4 31.0 (+16%) (+59%) (+56%) (+63%) (-6%) 0. Results are shown for the Laplace. respec- . since it has global support.6(b) presents the mesh quality halfway of the upstroke.065 0.6 at mid-stroke.88 0.0 52. SBR Stress and RBF mesh motion solver.81 0. the latter using different RBF’s. Figure 3.52 (-32%) (-10%) (-15%) (-7%) (-45%) 0. This was already illustrated for the moving two-dimensional block in figures 3.96 (+0. without any user input.95 (baseline) 0.1 (-) 21. the mesh around a flapping wing is shown in figure 3.5 73.051 Table 3.3 Comparison of mesh quality for different mesh motion solvers. The mean and maximal values of the skewness fs and non-orthogonality fo are compared at maximal displacement and rotation of the two-dimensional rectangular block.4 32.9 31.1 (-) 75. ported thin plate spline (TPS) provided high quality and robust mesh deformation.65 0. 90 80 70 Number of cells 60 50 40 30 20 10 0 0 10 20 30 40 50 60 70 Non-orthogonality 80 90 Number of cells 90 80 70 60 50 40 30 20 10 0 0 10 20 30 40 50 60 70 Non-orthogonality 80 90 (a) Laplace (b) RBF Figure 3.4 Cell non-orthogonality histograms for Laplace and RBF mesh motion solvers.1 (+4%) 59. Concerning the RBF mesh motion solver. To show the three-dimensional capabilities of the RBF mesh motion solver.7%) 0.6 74.8 19. showing the histograms of the non-orthogonality and skewness.8 76. The TPS was used as radial basis function. Mesh quality histograms show the variation in non-orthogonality for Laplace (a) and RBF mesh motion solvers (b) using the TPS.09 0.6(a) shows the non-orthogonality during the downstroke. the cells close to the wing take a larger part of the deformation compared to the Laplace method.3 (+6%) 23. From the figure it is clear that a large part of the near wake is deformed in order to deal with the threedimensional wing motion.105 0.

The .9 Skewness 350 300 250 200 150 100 50 1 0 0 0. the RBF mesh motion solver is tested by employing the flexing of a two-dimensional moving boundary.5 Cell skewness histograms for Laplace and RBF mesh motion solvers.4 0. tively. the effects of wing flexibility was incorporated by defining the wing flexing using harmonic functions.6(b) it is seen that the mesh during both the upstroke and downstroke is symmetric.. performing a fluid-structure-interaction simulation is too computationally demanding for a full three-dimensional flapping wing.1 0.9 Skewness 1 (a) Laplace (b) RBF Figure 3. This is caused by the fact that the radial basis function interpolation determines the new internal mesh points with respect to the initial mesh. The motion of the two-dimensional block can be decomposed into translation. These techniques are described in section 3. the model problem of a moving block is used.6 0. When comparing figure 3. to mimic realistic insect wing deformation (Shyy et al. But first.8 0.7 0.2. As discussed in subsection 3.3 0.6 0.3 Flexing of a two-dimensional block Within the context of the present research.2 0. 2008a). such that the initial mesh is recovered after every flapping period. Therefore. the mesh deformation for three-dimensional cases may become very expensive.4 0. Therefore. such that it is necessary to implement techniques to improve its efficiency.3.2 0.6(a) and 3.8 0.1 0. 3.7 0.5 0. rotation and flexing.4. Mesh quality histograms show the variation in skewness for Laplace (a) and RBF mesh motion solvers (b) using the TPS. while the RBF mesh motion technique leads a smoother decline of both quality measures in the histograms.3 0. all defined with respect to the initial configuration.5. The Laplace method results in a small number of cells with large values for both non-orthogonality and skewness. In order to show that the RBF mesh motion method is able to deal with a flexing boundary. the radial basis function mesh motion contains a direct system solve and an evaluation to determine the displacement of all internal mesh points.64 500 450 400 Number of cells 350 300 250 200 150 100 50 0 0 Mesh deformation techniques for flapping flight 500 450 400 Number of cells 0.5 0.

25T (b) t = 0.0. 0. A larger domain will undoubtedly lead to high quality meshes when using RBF mesh motion. 2 · cf α = Aα · sin(2πf t). lower (blue) is better.5 which is about 10% of the flexible boundary length. which is mainly caused by the fixed points on the outer boundary and the small computational domain.75T . df = (0. (a) shows the mesh quality at t = 0.0) with a flexing amplitude of Af = 0.0.25T and (b) at t = 0. like was the case with a rigid airfoil. where Af is the flexing amplitude vector. 2.7 shows the resulting mesh deformation at t = 0. f the frequency and t the time. respectively during the downstroke and upstroke. where. 0.0.0. At and Aα represent the translation and rotation amplitude vectors. The flexing was defined such that the main flexing direction is perpendicular to the flexible boundary surface. the amplitudes of both translation and rotation were fixed to At = (2.0) and Aα = (0.0).25T and t = 0.3. dc is the direction vector in-plane of the flexing surface.75T Figure 3. cf the length of the flexing surface and df represents the direction vector of the flexing.0. some high non-orthogonality can be observed within a region of 1 − 2 block lengths. Still. translation and rotation are defined by: xt = At · sin(2πf t).5. . 0. Figure 3. respectively. 1. 1. obtained with the RBF mesh motion solver with the globally supported thin plate spline (TPS). The flexing of the boundary is defined by combining two harmonic functions like: xf = Af · cos(2π x • dc ) · sin(2πf t) • df . It can be seen that the whole mesh is deformed by the RBF mesh motion solver. Cell non-orthogonality of a mesh around a three-dimensional flapping model wing.4 Comparison of mesh motion solvers 65 (a) t = 0.6 Cell non-orthogonality for a three-dimensional wing. In this model problem of a moving two-dimensional block.75T where T = 1/f is the motion period. cf = 5.5.

(a) shows the mesh deformation at t = 0. Table 3. equation (3. 2. the total number of operations for a direct solve and the RBF evaluation is given in the table.and three-dimensional meshes both the system solve and the evaluation procedures may become very computationally expensive. 3.66 Mesh deformation techniques for flapping flight (a) (b) Figure 3.4 shows the total number of internal and boundary points as a function of Nb . especially when direct methods are used. equal to Nb . A flexible rectangular block which is translating.5 Improving computational efficiency When using direct methods. The computational domain is square shaped with an equal number of points on all edges of the boundary.25T and (b) at t = 0. For large two. using the given boundary points xbi and the in step 1 computed coefficient vectors γ and β. standard direct solvers require O(Nb ) operations.8) for given boundary points xbi and corresponding displacements ∆xbi to find the coefficient vectors γ and β. rotating and flexing using the RBF mesh motion solver with the globally supported thin plate spline (TPS). Therefore. for both the two. In order to illustrate the increasing costs with increasing number of mesh points. Additionally.7 Deforming mesh around a flexible block using RBF mesh motion. This evaluation leads to a computational cost of order O(Nb Nin ).and three-dimensional Cartesian grid is considered with a uniform mesh distribution. a two.and three-dimensional example. The flexing is defined by simple harmonic functions. The upper diagonal block matrix Φbb is in general a dense symmetric matrix of 3 size (Nb x Nb ).75T . which . The method consists of two computationally expensive steps: 1. at all Nin (of order O(Nin )) internal mesh points. Solving the system of equations (3.7). Evaluation of the radial basis function summation. increases fast with an increased number of boundary and internal points. the computational costs of the new mesh motion solver based on radial basis function interpolation.

special treatment is necessary concerning the system solve. This can be achieved in two different ways.12) Router − Rinner xi represents the coordinate of the i−th inner mesh point. x= From table 3. based on greedy algorithms are applied by Rendall & Allen (2008a). It seems reasonable to reduce the number of moving boundary points by performing a sound selection procedure. the outer boundary points can be neglected. which is defined (Jakobsson & Amoignon. (3. Concerning the three-dimensional case. More advanced coarsening techniques. 3. step 1. 0 ≤ x ≤ 1. So when a complex three-dimensional case. like a flapping wing. 2007) as  x ≤ 0. by reducing the number of boundary nodes or using advanced direct solver techniques. Rinner and Router are two radii. the efficiency of the RBF method is improved.1 Boundary point coarsening and smoothing ||xi − Rinner || .  1. shows that the computational costs for both a direct solve and evaluation scales 3 with Nb for the two-dimensional case. Computational costs for solving the system and evaluation the RBF’s on all internal mesh points. x is given by: . to evaluate this smoothing function at that particular location in space. by the notion that all outer boundary points are fixed in general.11) ψ(x) =  0. is considered.3. Therefore. where. 1 − x2 (3 − 2x). Especially when the body displacement follows a rigid body motion.g. not all boundary points are necessary.5 Improving computational efficiency 67 Internal points 2D 3D 2 Nb 3 Nb All boundary points 4Nb 2 6Nb Direct solve 3 64Nb 6 216Nb RBF evaluation 3 4Nb 5 6Nb Table 3. Therefore. which is a factor Nb more expensive than the evaluation. This is achieved by specifying a smoothing function such that the RBF contribution reduces to zero at the outer boundary. where ξ is the coarsening factor.and y-direction. First. a coarsening technique was incorporated which selects a boundary point for every ξ points.4 Computational costs for solving the system and evaluate the RBF’s. (3. a major part of the computational cost is spent at solving the system.4 it is seen that the total computation costs will decrease if a constraint is put on the number of mesh points. Illustration of a two-dimensional and three-dimensional uniform square shaped Cartesian mesh with an equal number of boundary nodes Nb in x. the computational costs for the direct system solve are a factor Nb larger compared to the costs for the RBF evaluation. Secondly.5. e. x ≥ 1.

Concerning the RBF mesh motion solvers. per time-step. Additionally. the mesh resolution is increased.8(a) shows the computing times with increasing grid resolution for the Laplace. the system to be solved. Figure 3. the three described variants are used. If the absolute and relative RBF methods are compared. It is clear that the mesh motion solvers based on solving a partial differential equation are very fast. Figure 3. the number of boundary points used in the system solving is still growing non-linearly. but outside Router the value of the RBF becomes zero. 190 hours. since standard iterative techniques can be used for these sparse systems. only contains the control points on the moving boundary. if the coarsening and smoothing techniques are applied the computing times are of similar order compared with the fast Laplace mesh motion. These three-dimensional simulations are performed for grid sizes of 100k. such that all fixed outer boundary points may be neglected. Figure 3.8(b) shows super-linear curves for both solving the flow equations and the RBF mesh motion. 103. selected by the coarsening function. the absolute implementation.68 Mesh deformation techniques for flapping flight between those radii this smoothing function is decreasing from 1. Therefore. it is interesting to address the computing times concerning the different mesh motion solvers.0 to 0.0. concerning a threedimensional flapping wing simulation (6 flapping periods). 200k.9) is multiplied by φ(x) the corrected RBF evaluation is obtained by: ∆xinj = s(xinj ) • ψ(x). relative implementation and the relative method in combination with the previously dealt coarsening and smoothing techniques. the inner radius is chosen to be multiple boundary lengths (wing chords) and the outer radius is chosen as the distance from the moving boundary to the outer boundary. figure 3. This is considered to be very efficient. While keeping the coarsening ratio fixed.13) Within Rinner the contribution of the RBF evaluation remains unaltered. 400k. it must be noted that mesh coarsening and a smoothing function are applied for acceleration. Validation showed that a mesh resolution of 800k provided accurate results for flapping wing aerodynamics. All simulations were performed on four CPU cores of an AMD Opteron 280 cluster. When the RBF evaluation function (3. For high resolution meshes. SBR Stress and RBF mesh motion solvers. 48.8(b) shows that the computing time used for RBF mesh motion is less than 10% of the computing time for the complete time-step. the order of computing times can be decreased further by increasing the coarsening ratio. respectively. 32. In principle. . In addition. at least an order of magnitude more. On the other hand.8(b) shows the computing times. it is observed that these methods require very large computing times. The non-linear behaviour of the final curve is caused by the choice of the coarsening ratio to select the moving boundary points. (3. 800k and 1600k cells and needed a total computing time of about 8.

the SBR Stress equation ( ). the two different phases of the radial basis function mesh motion. RBF relative method (×) and RBF relative method in combination with coarsening and smoothing techniques (•). Concerning the coarsening. Meshes are used from 100k to 1600k. which is about reducing the number of boundary points by applying coarsening and smoothing techniques.4. A possible better choice would be to use parallel direct techniques. a mesh motion problem leads in general to an ill-conditioned system.4.8 Computing times for different mesh motion solvers. section 3. leading to a system which is more efficient to solve using existing iterative solver techniques. the ill-conditioned system cannot be efficiently solved using iterative techniques.. which occur in this type of problems. (b) shows the computing times for one time-step of a three-dimensional flapping wing simulation. 3. which is difficult to solve directly and iteratively. solving (I) and evaluation (II) can be very expensive. RBF absolute method ( ).5 Improving computational efficiency 104 70 60 Computing time [s] 103 Computing time [s] 50 40 30 20 10 10 0 10 0 69 Mesh motion Flow equation Total timestep 102 101 101 Grid spacing 102 0 0 1 2 4 8 Grid spacing 16 (a) Two-dimensional (b) Three-dimensional Figure 3. are caused by the boundary point locations. 2007) to approximate the system of equations (3. Different pre-conditioning techniques are described in literature (Boer de et al. However. as long as the resulting mesh is of sufficiently high quality.5. the combination of cell clustering on the moving surface and the large distance to the outer boundary. causes large differences between points in Φbb . was about O(1010 ). Despite the pre-conditioning. In general moving mesh applications. it is not necessary to solve (I) and (II) exactly. (a) shows a comparison of computing times for the mesh motion solvers based on the Laplace equation (◦).3. with a significant gain in computing time. Furthermore.2 Iterative techniques and parallel implementations As was shown in the previous section. a parallel version of LAPACK. A different way to improve the efficiency of the computation has already been applied in section 3. complex greedy algorithms (Rendall & . since the internal point motion can be arbitrary.8). The condition number in case of the model problem of a moving square. The high condition numbers. The times are subdivided by solving the RBF mesh motion and the flow equations. The computing times for the Laplace equation (◦) and the SBR Stress equation ( ) are nearly identical. available in the linear algebra packages SuperLU and ScaLAPACK.

. Furthermore. The diffusivity. the way OpenFOAM deals with the finite volume implementation. The first method solves the Laplace equation with a variable diffusion coefficient. in both cases. 2007). using all control points and corresponding coefficients. re-ordering can be applied for further enhancement. the evaluation can be easily implemented in parallel since it only involves a matrix-vector multiplication. αi and βi . both based on solving a partial differential equation. Then every processor only performs the evaluation of the internal points. the solid body rotation stress (SBR Stress) mesh motion uses the diffusivity. concerning a parallel implementation. the internal point displacements are obtained by evaluating the radial basis functions. The major difficulty. Currently. are compared. A new mesh motion solver is incorporated in OpenFOAM . Especially when using the thin plate spline (TPS) or the continuous polynomial C 2 . this is being implemented in OpenFOAM . 2008a) may be applied to select the necessary boundary points such that the efficiency is increased even further. concerning the RBF mesh motion. to influence the quality of the mesh. is defined to decrease quadratically with the distance from the moving boundary.70 Mesh deformation techniques for flapping flight Allen. which uses the interpolation of radial basis functions (RBF).8) is to decrease the condition number by only taking the boundary points with a low mutual distance (Boer de et al. The cell non-orthogonality and skewness are compared. For given boundary point displacements the internal mesh displacements are obtained by solving a system of equations to obtain an array of interpolation coefficients. Secondly. is that all processor partitions need to know which control point belongs to itself and which to the other partitions. acting as a stiffness. Besides solving a partial differential equation the motion of mesh points can be defined using interpolation techniques. Finally. As with the Laplace equation.. which are distributed over all partitions. 2007). Using those coefficients. 3.6 Conclusions In this chapter two different mesh motion techniques were described which are commonly used within the code of OpenFOAM . of that particular partition. This new mesh motion technique does not need any information about the mesh connectivity and can be applied to arbitrary unstructured meshes containing polyhedral cells. Additionally. Another method to increase the efficiency to solve the system (3. the linear stress equation was modified to include rigid body rotations in order to cope with the severe mesh deformation present in flapping wing simulations. in addition. The RBF mesh quality provides superior mesh quality over the Laplace and SBR Stress mesh motion solvers. which is used to control the final mesh quality. the speed of the evaluation (II) can be increased by various fast evaluation algorithms (Boer de et al. The three mesh motion solvers are tested using a case of a two-dimensional rectangular block which moves through a Cartesian mesh. different radial basis functions.

After this elaborate discussion it is concluded that. the mesh quality is high in terms of low skewness and non-orthogonality. The TPS has global support. which reduces the system of equations considerably. The RBF mesh motion was successfully tested on simple test problems and for a three-dimensional flapping wing with the possibility to incorporate a flexing moving boundary. whereas the C 2 basis function has compact support. Secondly. So a coarsening algorithm selects those control points. different methods are implemented to increase its efficiency. Therefore. concerning the three-dimensional wing simulations. . First of all.3. Since the RBF mesh motion technique encounters a dense system of equations.6 Conclusions 71 as radial basis functions. a subset of the moving boundary points was selected. it is justified to neglect the outer (fixed) boundary points. a smoothing function is used to decrease the RBF contribution to zero at the outer domain boundaries. because not all points are necessary when the body performs a rigid body motion. the globally supported TPS should be used in combination with the coarsening and smoothing techniques to increase the efficiency of the RBF mesh motion method.

.

1993. undergo significant interactions with the environmental fluid in which they move. Additionally. Animals that fly or swim. Both geometry and wing kinematics are dynamically scaled in order to design a sound framework for comparison. 4. the force coefficients are used in conjuncture with the lift-to-drag ratio to assess the flapping wing performance. Reynolds and Rossby numbers) are identified after writing the Navier-Stokes equations in a rotating reference frame.. Taylor et al. 2003) at certain scales. the physical and numerical modelling of flapping wing and foils is described. representing occasional wing deformation. . To systematically study the aerodynamics around flapping wings.. which are equivalent from the fluid-dynamic perspective (Triantafyllou et al. These equations are used to deduce the dimensionless numbers relevant for insect flight. using the radius of gyration.1 Introduction This chapter deals with the physical and numerical modelling of flapping wings. which describes the wing motion. The kinematic model. for the mathematical analysis of swimming or flying it is important to formulate the governing equations and accompanying boundary conditions in an appropriate form. Therefore. a model planform and kinematic model is defined. The relevant dimensionless numbers (Strouhal. in hovering as well as in forward flight conditions. consists of a rigid body motion appended by a flexing.CHAPTER 4 Physical and numerical modelling of flapping foils and wings In this chapter.

The Reynolds number is defined as the ratio between the inertial and viscous forces present in a fluid. 2005. Here Φ0 is half of the total flapping angular amplitude.. The first use of the Strouhal number was in the context of the natural vortex shedding behind a stationary cylinder in a uniform flow. A different approach to dimensionless numbers is to define them as the ratio of time or length scales. J = U/4Φ0 f R. This definition of the Strouhal number. Additionally. White.. the Strouhal number is commonly used to characterise . 2003. travelled by the wing. see chapter 2.74 Physical and numerical modelling of flapping foils and wings The Reynolds number In order to improve our understanding of biological flows. 2008b). 2003. like the flows around flapping wings or fins. This expression is very similar to the reduced frequency (Shyy et al. the Strouhal number plays an important role in flapping wing aerodynamics as well. i. In general engineering applications. Lentink & Gerritsma. Williamson (1988) found a universal relation between the Reynolds and Strouhal number based on the observed vortex-shedding frequency in the laminar flow regime. the Strouhal number can be very useful. The typical definition of the Strouhal number is the flapping frequency times flapping amplitude divided by a reference flow velocity. Bos et al. For example. provided that the wing flaps in a stroke plane perpendicular to the forward velocity (Lentink & Gerritsma.e. defining implicitly the complexity of the mathematical model needed to solve the problem. which is easier to understand and to apply to a flow problem. like shear layers and vortex generation. For moving bodies and especially flapping wings. which is commonly used to relate the two velocities due to either flapping and forward flight. the Strouhal number is proportional to the maximum value of the induced angle of attack.e. compared to the inertial component. the single wing span. such that it identifies what kind of propulsive mechanism applies to the flapping wing. the Reynolds number determines whether the flow behaves turbulent or laminar. instead of the ratio between two distinct forces (Tennekes & Lumley. Stc = f c/U .g. f the flapping frequency and R represents the distance from root to tip of the wing. like the Reynolds (Re) and Strouhal (St) number in particular (e. in forward flapping flight. will be more pronounced. as defined by Ellington (1984).. such that viscous phenomena. 2008). Panton. based on the flapping amplitude. It is a property of the flow. For flapping wings or oscillating bodies. 2003. the Strouhal number can be defined based on the imposed oscillation frequency and amplitude. 1972. is closely related to the advance ratio. such as the Von K´rm´n vortex street behind a bluff body. the forces in the flow are dominated by the viscous term. Another definition was introduced by Dickinson (1994) and Wang (2000b) using the average chord length as reference. Re = O(100). it is of importance to make extensive use of dimensionless numbers. when a flapping wing operates at a very low Reynolds number.. Thaweewat et al. i. which is the ratio between the forward and flapping distance. For example. In that case the Reynolds number can be defined as the ratio between the convection time over the diffusion time. a a The Strouhal number Besides the Reynolds number. Taylor et al. 2009). 1991).

only one flapping wing is considered. Furthermore. 4. whereas the reduced frequency number is used in flapping wing problems. Additionally. relative to the mean chord. like the operation of insects and fish.2. followed by the conclusions of this chapter in section 4. the flow can be considered to be incompressible since the Mach number (a measure for compressibility) is typically M a = U/a = O(10−3 ) (Brodsky. In view of simplicity. Section 4. and re-stated here: ∇ • u = 0. corresponding to the operating conditions of fruit flies. In section 2. and 1000. ∂t ρ (4. specific attention is given to the appropriate definition of the governing dimensionless numbers to investigate the flow around flapping wings.2 Governing equations for flapping wings Concerning flapping flight in nature. In section 4. the governing equations are formulated for forward and hovering flapping flight.4) and (2.1) 1 ∂u + ∇ • (uu) = − ∇p + ν∇2 u.4.5). Equations and other assumptions As previously discussed. where U [m/s] is the reference velocity and a [m/s] the speed of sound. This implies that no interaction between two wings or with the body are included. Secondly. In accordance with Lentink & Dickinson (2009a).2) . 500. the reciprocal of the Strouhal number is known as the dimensionless wavelength λ∗ = U/f c which is often used in studies concerning forward flapping flight as it seems reasonably intuitive in that it corresponds to the distance travelled over one flapping period. followed by the dynamical scaling of flapping flight in section 4. the present study deals with a model wing which is a simplified representation of a flying insect wing operating at Reynolds numbers.3.6 describes the force and performance definitions. which allows that the induced vortical flow can be studied in more detail. under hovering as well as forward flight conditions. In order to perform a sound and valid comparison it is important to maintain constant dimensionless numbers while kinematic parameters or flow properties are varied to study their influence. house flies and bumblebees. Nevertheless.2 the incompressible Navier-Stokes equations were defined by equation (2. the considered flapping kinematics that result in large rotation rates put the current numerical techniques to a significant challenge. the model wing selection and the definition of the kinematic model parameters are described in section 4. (4. 1994).2 Governing equations for flapping wings 75 the vortex shedding. it is appropriate to use dimensionless numbers to study the effect of flapping characteristics on the aerodynamic performance. Re = 100.4. As a prelude to the numerical solution of the governing equations. respectively.7.5. the mesh generation in combination with the boundary conditions is briefly dealt with in 4.

Using Computational Fluid Dynamics (CFD) techniques.76 Physical and numerical modelling of flapping foils and wings These equations are derived by analysis of the forces on an infinitely small fluid element in an inertial reference frame. has to be zero. approximately. the reference frame approach is only used to identify certain dimensionless numbers that are related to the rotation of the threedimensional wing. which means that the velocity. Ωwing is the angular velocity of the rotating reference frame. 1998. This relation results in a no-slip condition in the rotating reference frame. When flight under hovering conditions is considered. is defined as (Ginsberg. The corresponding velocity transformation. In the present study. 2008) with respect to the inertial frame (XY Z). in all directions. The resulting boundary condition on the flapping wing will be such that the effective velocity will be zero. a no-slip condition (White. the actual flow computations are made with respect to the inertial reference frame. resulting in the following expression: uwing = uWING − (utrans + Ωwing × r). where uWING is the flow velocity at the wing in the inertial reference frame.e. derived from realistic insect data (e. 1991) needs to be specified. representing the flapping motion of the wing. since the reference frame moves with the boundary.. where uXY Z corresponds to the velocity in the inertial reference frame. these equations are solved on a discretised computational domain in combination with appropriate initial and boundary conditions. the initial velocity field is zero as well as the boundary conditions at the outer domain boundary. which the case in hovering flight. 2003). i. In the present study.g. utrans is the translating velocity of the reference frame itself and can be used to specify the translation of the insect body. This is accomplished by defining the mathematical velocity on the moving boundary to be equal to the actual wing motion. The boundary condition needs to take care of the rotation of the reference frame. 1999) uXY Z = uxyz + (utrans + Ωwing × r). The rotating reference frame is attached with its origin to the joint around which the wing rotates. to make the velocity at the moving boundary equal to zero. Fry et al. the translation velocity of the rotating reference frame is assumed to be zero. which moves according to a specific kinematic model. At the boundary of the flapping wing. which is fixed to the flapping wing and moves accordingly. whereas uxyz represents the velocity in the local rotating reference frame. i. relative to the wing. Additionally. The accelerations in the inertial and rotating reference . Momentum analysis in rotating reference frame A different numerical approach to solve this problem is to transform the governing equations and boundary conditions from the inertial reference frame (XY Z) to a rotating reference frame (xyz).e. r is the distance from a rotating point to the origin. Baruh. it is interesting to study the accelerations (Lentink.

Baruh.2) such that the following transformed Navier-Stokes equations are obtained (for the sake of simplicity. Ω needs to be related to the flapping motion. Dt ρ (4. 1999): aXY Z = axyz + (aang + acen + acor ).2 Governing equations for flapping wings Y y x O X 77 Z z Figure 4. This will be elaborated in detail in section 4. Now the expressions for velocity and accelerations in the rotating reference frame are substituted into the Navier-Stokes equations (4. The rotational reference frame xyz is moving with the wing and obtained by rotating the inertial reference frame XY Z by three orientation angles. 1998. in addition to the already described Reynolds and Strouhal numbers. These new dimensionless numbers will become available if the different . uxyz [m/s] is the velocity in the rotating frame. In order to explore the different acceleration terms. and Coriolis acor [m/s2 ] accelerations are respectively defined as aang acen acor ˙ = Ω × r. the subscripts are dropped): 1 Du ˙ + (Ω × r) + (Ω × (Ω × r)) + (2Ω × u) = − ∇p + ν∇2 u. which strongly depends on the wing kinematics.3. aang [m/s2 ].1 Illustration of the rotational reference frame. centripetal. acen [m/s2 ]. The angular.1) and (4.4.3) This transformed equation describes the momentum balance for a fluid particle close to the wing (in the boundary layer) in the rotating reference frame. = Ω × (Ω × r). The main parameter in these three different accelerations is the angular velocity Ω [rad/s]. With this approach it is possible to derive dimensionless numbers representing the different acceleration terms. frames are related using the following (Ginsberg. = 2Ω × uxyz .

2R c Ω (4. other dimensional numbers can be identified. ∇∗ = c · ∇. Furthermore. Ωc By the rotation amplitude Ω. 4. 2001. ˙ ΩR c 2 Uref . r∗ = .3 Wing shape and kinematic modelling When investigating the three-dimensional aerodynamics around flapping wings. related to the rotation of the reference frame. rewriting gives: 1 Du 1 1 1 1 ˙ · + · (Ω × r) + · (Ω × (Ω × r)) + · (2Ω × u) = −∇p + · ∇2 u. ρref [kg/m3 ] is the reference density (constant in incompressible flows). In addition to the Reynolds and Strouhal numbers. 1998) or . especially the velocity and acceleration due to the wing rotation. these dimensionless numbers strongly depend on the wing kinematics. St Dt Cang Ccen Ro Re where the additional dimensionless number are respectively defined as Cang = 2 Uref . 2003). or to opt for a more generic approach. the effect of different kinematics and flow features may be related to these dimensionless numbers. t∗ = t. in order to fully understand them. R [m] is a radius length. one may try to simulate the geometry and conditions of specific species.4) Ccen = (4. which ref is defined as f = Uc . where c the average chord length.5) (4. a fruit fly (Sane & Dickinson. L [m] is the reference ˙ length. Ω∗ = . respectively.g.3) are scaled as u∗ = ˙ Uref r p Ω u Ω ˙ . which is still attached to the wing. Birch & Dickinson. f · c Dt Uref Uref Uref Uref c where Uref [m/s] is the reference velocity. Ω∗ = . hawkmoth (Liu & Kawachi. Previous investigations of specific insect species have been reported for e. Therefore. f [1/s] the flapping frequency.78 Physical and numerical modelling of flapping foils and wings terms in equation (4. p∗ = 2 . ˙ Uref c Ω R ρref · Uref Ω leading to (dropping the stars for simplicity): ˙ Uref Du ΩRc ˙ Ω2 Rc Ωc ν · + 2 ·(Ω×r)+ 2 ·(Ω×(Ω×r))+ ·(2Ω×u) = −∇p+ ·∇2 u. Ω [ rad /s] and Ω [rad/s] are the average rotational velocity and acceleration.6) Ro = Uref .

7).9) which is obtained by rewriting equation (4.4.5 and c = 1. Radius of gyration In order to define a sound framework of comparison for different three-dimensional and two-dimensional simulations.8) and (4. Section 4. the average chord length can be obtained by integration of the chord distribution c(r) along the wing span from root to tip: 1 R c= c(r)dr.2 shows the wing planform and corresponding parameters for a fruit fly wing and a ellipsoidal model wing.4.1 Wing shape and planform selection The general model wing is described by an ellipsoid in the three-dimensional wing reference frame: x 2 y 2 z 2 + + = 1.2 deals with the kinematic modelling. (4.7). . b and c are the semi-axes of the ellipsoidal wing. In order to obtain a wing which has a single wing span.7) a b c where a.3. c2 (4.3 Wing shape and kinematic modelling 79 a dragonfly (Isogai et al.9) leads to an average chord length of c = π/4 for c = 1.05.0. the kinematic model was appended with an active wing flexing component.. The present study follows the second approach and considers a three-dimensional ellipsoidal model wing (Bos et al.0 and a thickness of 10% of the chord. a similar ellipsoid was used by Wang et al. the semi-axes are chosen as a = 0.. Since the planform is analytically given by equation (4. The numerical implementation of the wing kinematics is the subject of section 4. Assuming that the chord c(r) is represented by y(z) the following relation is used to calculate the average chord: c(r) = b2 1 − r2 .1 briefly describes the wing shape and morphology. while section 4. 4.3. a similar elliptical cross-section is applied for z = 0). Previous studies show that specific insect features. bs = 2. which is described in section 4. it is necessary to define all reference parameters.3.3. In addition. like the corrugated wing planform (Luo & Sun. Figure 4. 2005) are of minor influence on the resulting fluid behaviour. 2008). The three-dimensional elliptical planform is shown in figure 4.3.3.8) R 0 where R [m] is the radius of the wing tip and c(r) [m] the chord distribution along the wing span.0.0 (for the two-dimensional airfoil. both defining the planform.2 in comparison to a more realistic representation of the fruit fly wing planform. Evaluation of equation (4. The general kinematic modelling consists of a translation and a rotation component. (4. (2004) for their two-dimensional research. a maximal chord length cmax = 1. b = 0. 2004).

.80 Axis of rotation Physical and numerical modelling of flapping foils and wings Axis of rotation Rtip Rroot y z c(r) Rg O S dr dr r Rg r Rg (b) Ellipsoidal shape S bs Rroot y z c(r) Rg Rtip bs O (a) Fruit fly shape Figure 4. Luo & Sun (2005) compared the flow induced by flapping wings with different aspect ratios and found that the radius of gyration provided a reliable framework for force comparison when the flapping velocity is varied. The planform is defined by the chord variation c(r). especially when the comparison of different kinematic models is concerned (Bos et al. Lentink. Previous two-dimensional studies . (4. 4. 2008). 2008) and is calculated as 1 R 2 r c(r)dr. this is the location where the resulting lift acts.2 Model wing geometry and planform. the single wing span bs and the planform surface area S. of which the location is varied to study the effect of the angular accelerations. 2005. The radius of gyration. the purpose of the present research is also to investigate the three-dimensional flow around flapping wings at low Reynolds numbers. Rg is defined as the weighted second moment of inertia (Luo & Sun. R the distance from the rotation origin to the wing tip and c(r) represents the chord distribution along the wing. As the local velocity of each cross-section varies during flapping. In three-dimensional simulations the wing revolves around the origin O.3.2. r the spanwise coordinate.2 Kinematic modelling Besides the numerical interest in the development and improvement of mesh motion techniques. at a representative cross-sectional area of the wing. described in chapter 3. Additionally. Schematic illustration of the geometry and planform of a fruit fly (a) and ellipsoidal model wing (b).10) Rg = S 0 Here S is the wing planform. Besides the average chord length c and the single wing span bs the radius of gyration is another important geometric parameter. the spanwise reference location is chosen to be at the radius of gyration. According to Ellington (1984). introduced in section 4. The radius of gyration Rg is used to define a sound framework for comparison between mutual three-dimensional and twodimensional simulations. the scale at which insects operate.

Figure 4. 2003). In addition. from purely harmonic motion to complex realistic fruit fly kinematics. The kinematic wing motion is defined by the variation of three independent attitude angles. Furthermore.. as is shown in figure 4.3 Schematic illustration of the governing flapping angles. In this three-dimensional model the three degrees of freedom of the wing motion are defined as the flapping angle.. like harmonic motion. φ(t) corresponds to the stroke variation. the centre of rotation (which is equivalent to the Rossby number Ro) or the angle of attack amplitude.. which is measured in a plane normal to the stroke plane. the shape of the fruit fly angle of attack clearly shows a plateau of constant value.. showed the tight relationship between the aerodynamic forces or performance and the kinematic model (Wang et al. which are considered. Figure 4. Using two-dimensional numerical techniques. 2009). 2008. The deviation may be used to create a ‘figure-of-eight’ pattern which is present in realistic fruit fly kinematics (Fry et al. 2003). Figure 4. e. different kinematic models have been studied. the angle of attack. in the mean stroke plane.1T).. Since an accurate three-dimensional numerical method has been developed to solve for the flow around flapping wings.. with respect to the horizontal plane and the deviation θ(t). which is ex- . 2004) compared to (b) the realistic fruit fly kinematics (Fry et al. α(t) to the geometrical angle of attack and deviation θ(t). 2004. the angle of attack shows a dip (at t=0.4(b) shows that the realistic fruit fly kinematics is characterised by an asymmetry in flapping angle. simplified kinematics was used.4. (2008) showed that the flapping wing performance may be influenced by the specific features of a particular kinematic model. Besides the use of an idealisation of the insect wing planform.4 shows the least and the most complex of the kinematics models.3 Wing shape and kinematic modelling Y 81 α(t) Rtip Stroke plane O θ(t) φ(t) start upstroke mid-stroke Rroot X start downstroke Z Figure 4.3. to study different parameters independently. α(t). In literature. Thaweewat et al. see figure 4.3.g. which is measured in a plane normal to the stroke plane. lowering the effective angle of attack. φ(t). the effect of different kinematic models on the aerodynamic performance can be evaluated. Bos et al. Bos et al. Flapping wing motion is governed by three angles. and angle of attack.4 illustrates (a) the harmonic model (Wang et al.

the definition of the harmonic model is described accordingly. is defined by a sine function with respect to 90◦ and the deviation angle. the realistic fruit fly kinematics in characterised by the presence of deviation which may result in a ‘figure-of-eight’ pattern (Shyy et al. 4. Remind that this amplitude is half times the value used in literature (e.82 Physical and numerical modelling of flapping foils and wings ploited by Sane & Dickinson (2001.. 2003).. (4. Bos et al. Fry et al. The ‘trapezoidal’ shaped angle of attack is implemented using a piece-wise continuous function and the deviation by defining a non-zero harmonically variation of θ(t). (4. 2008). 2005. such as the previously described plateau in angle of attack. 2002). which is defined from stroke reversal to midstroke. α(t). simplified (harmonic) kinematics may be interesting for Micro Air Vehicles (MAV) implementation. Φ0 in Ellington. Additionally. 2008b) In order to illustrate the variation of the motion angles.3. 2003. Lehmann et al.12) In general. 2003. The corresponding angular velocities are found by taking the time-derivatives of equation (4. 2008. Aθ is the amplitude of the deviation angle which causes the so-called ‘figure-of-eight’ pattern and is sinusoidal shaped (Fry et al. is given by a pure sine: φ(t) = Aφ · cos(2πf t).11). The geometric angle of attack. φ(t).3 Modelling of active wing flexing In addition to the rigid body motion as shown in figure 4. ˙ ˙ θ(t) = Aθ · (2πf ) · cos(2πf t).g. The flapping angle.. 2 θ(t) = Aθ · sin(2πf t). ˙ φ(t) = −Aφ · (2πf ) · sin(2πf t)... is described by a cosine function. θ(t).. 1984). Lehmann et al. which is the initial position under hovering conditions. Also. a flexing displacement of the wing surface is defined. Lu & Shen. (2005) using a Robofly. Aφ is the flapping amplitude. The flexing displacement is defined with respect to the initial wing position and can be written .11) Here. 2008). As will be shown in chapter 5 (Bos et al. f is the flapping frequency and Aα represents the amplitude of the angle of attack with respect to π/2. modelled by a ‘trapezoidal’ shape and the presence of deviation (Birch & Dickinson. Since a full fluid structure interaction (FSI) simulation is too expensive and beyond the scope of the current research.3 it is possible to define an extra displacement concerning flexing of the wing. α(t) = −Aα · (2πf ) · cos(2πf t). π α(t) = − Aα · sin(2πf t). the flapping wing performance may be significantly influenced by modifications of the basic kinematics. this harmonic model is crude but fairly reasonable representation of the flapping motion of a fruit fly.

rota- . At every time-step the location of the boundary points is determined leading to three distinct displacement arrays.1 0.9 1 t/T [-] φ. as x(t) = Af · cos 2πx0 ǫf · sin(2πf t). The cosine shaped wing flexing is defined by ǫf . representing pure harmonic variations of the flapping angle φ(t) (•). the angle of attack. an extra ‘bump’ and a degree of ‘trapezoidal’ in angle of attack and the presence of deviation. (a) shows the least complex kinematic model. which corresponds to the cosine ratio. The realistic fruit fly kinematics is characterised by an asymmetric variation of flapping angle. which is the most complex kinematics available.5 0. where the first cosine function defines the wing shape and the sine function represents the time-variation.6 0.3 Wing shape and kinematic modelling 100 80 60 40 φ. the wing kinematics is calculated beforehand and applied to the numerical flow solver.5 (b). 4.8 0.3 0. ǫf = 0. In general.8 0.e. due to translation.5 shows a plunging airfoil incorporating flexing for ǫf = 0.2 0. α.4 Numerical implementation of the wing kinematics In previous sections the physical kinematic modelling was described. The translational component is not used in the current three-dimensional simulations. in the two-dimensional simulations.7 0. the wing kinematics can be decomposed into a translation. The current section deals with the numerical implementation of this particular wing kinematics.7 0. In addition to the translation and rotation.5 means that the shape is like a half cosine function. a limited number of two.4. a rotation and a flexing (deformation) component. which is limited to hovering and forward flow conditions with stationary position of the rotation origin.1 0. However. The variation of realistic fruit fly kinematics (Fry et al.4 0.5 0. the flapping motion is defined by a translation in combination with one rotation angle.4 Comparison of the harmonic and fruit fly kinematic models.. Af is the flexing amplitude vector and x0 the location of the initial boundary points. angle of attack α(t) (◦) and deviation θ(t) ( ). θ[◦ ] 20 0 -20 -40 -60 -80 0 0.9 1 t/T [-] 83 (a) Harmonic kinematics (b) Fruit fly kinematics Figure 4.3 0. 2003) is shown in (b). Figure 4.3. i. In general.2 0.and three-dimensional simulations using a pre-defined wing flexing have been performed. α. θ[◦ ] 100 80 60 40 20 0 -20 -40 -60 -80 0 0.25 (a) and ǫf = 0.4 0.6 0.

z(t)).5 Illustration of a flexing two-dimensional airfoil. a cosine function is used to define the motion in the inertial reference frame. Translation The displacement array of the boundary points.25 (b) Half cosine shape. Rold and Rnew are the rotation transformation matrices at respectively the old and new time instances.5 Figure 4. Here ∆xrot is the boundary displacement due to rotation. y(t). Both figures (a) and (b) show the upstroke (left) and downstroke (right). such that: ∆xrot = xnew − xold .ǫf = 0. The two-dimensional flexing airfoil is modelled by a time-varying cosine shape. tion and flexing. The sine function in this definition is used when the wing needs to move according to an ordinary rigid body motion (Ginsberg.25 (a) or a half cosine. ǫf = 0. ǫf = 0. (4. At and ft are respectively the translation amplitude and frequency vectors.14) (4. is calculated at subsequent (old and new) time-steps with respect to the initial mesh: xold = Rold · x0 . Rotation The second boundary point displacement array.84 Physical and numerical modelling of flapping foils and wings (a) Quarter cosine shape. where x(t) = (x(t). and xnew = Rnew · x0 . This airfoil shape is either a quarter cosine. ǫf = 0. 1998). due to rotation.13) .5 (b). When a flapping motion is desired. which is due to translation is obtained from x(t) = At · sin(2πft t).

3 Wing shape and kinematic modelling 85 and The initial boundary points are given by x0 . ff the flexing frequency and x0 represents the initial boundary points of the flapping wing at t = 0.18) which is substituted into equation (4. 1998.16) RY (t) =  sin(φ(t)) 0 cos(φ(t)) cos(α(t)) RZ (t) =  sin(α(t)) 0  −sin(α(t)) cos(α(t)) 0  0 0 . This wing flexing is defined as xflex (t) = Af · cos(2πx0 ) · sin(2πff t). According to (Ginsberg. Y . the flapping wing is able to perform a flexing motion as well. due to rotation: ∆xrot = [Rrot (tnew ) − Rrot (told )](x0 − r0 ). After a general sequence of matrix multiplications. Y and Z-axis. spanwise and chordwise. accelerating the fluid from top to bottom. the total rotation matrix is obtained by combining equations (4. respectively. which is varied systematically in chapter 7. these three matrices are defined as   1 0 0 (4. where Af is the flexing amplitude vector. This definition permits flexing in each orientation of a three-dimensional wing. rotation and flexing.14) to find the displacement of the boundary points at the old and new times.e. Flexing In addition to the translation and rotation. (4. the following is obtained: ∆xb = ∆xtrans + ∆xrot + ∆xflex . These rotation transformation matrices consist of three different components.or Z-axis correspond to the deviation angle. due to a rotation around the X-.17) Here. θ(t).4. 1 (4.3. it is clear that a hovering wing flaps in the X-Z plane. α(t). (4. 1999). Using the motion convention.13) and (4. Baruh. 0 1 0 (4. shown in figure 4.17) as Rrot (t) = RX (t) · RY (t) · RZ (t). i. 0 sin(θ(t)) cos(θ(t))   cos(φ(t)) 0 −sin(φ(t)) . the rotation around the X-.15) to (4. and the angle of attack. When combining the previous results for the displacements due to translation.15) RX (t) =  0 cos(θ(t)) −sin(θ(t))  . φ(t). . flapping angle.19) where r0 is the direction vector of the initial rotation origin.

the reference cross-section of the three-dimensional wing is positioned at the radius of gyration. 2008. where T is the flapping period. angular accelera˙ tion Ω and average chord length c it is necessary to define an appropriate reference velocity. Besides the angular velocity Ω. evaluation ˙ gives Ω = 4Aφ f and Ω = 2πf Ω.2 different dimensionless numbers. First.22) it is straightforward to find Uref by multiplying the expression for Ω (4. This approach has two major advantages.4 Dynamical scaling of flapping wings In section 4. it is desirable to start the numerical simulation at maximal flapping angle. but in case of forward flight conditions. were identified in order to scale the Navier-Stokes equations. Rg .22) where |uRg (t)| is the absolute velocity at Rg which can be decomposed into three components u(t). yielding good conditions for numerical convergence. Cang = Uref /ΩR c. yielding high mesh quality at the extreme wing positions.and Z-direction. in order to minimise the initial acceleration of the mesh. 0 (4. (4. 2008. Lentink. Therefore.. As previously described and in accordance with (Bos et al. there is an additional free-stream . (4. 2005). 4. governing fluid flow. To de2 2 ˙ fine these rotational dimensional numbers. As shown. Ccen = Uref /Ω2 R c and Ro = Uref /Ω c. This time-averaged velocity follows from: Uref = 1 T T 0 |uRg |dt = 1 T T u2 (t) + v 2 (t) + w2 (t)dt. Ω ˙ and its time-derivative Ω for the particular kinematics implemented: Ω= 1 ˙ Ω= T 1 T T 0 T 0 ˙ |φ|dt = 1 T 1 T T T 0 | − Aφ · (2πf ) · sin(2πf t)|dt = 4Aφ f. the initial velocity is as small as possible. it was chosen to use a rigid body motion until the maximal flapping angle was reached.86 Physical and numerical modelling of flapping foils and wings This total displacement field for the boundary points can be applied to the mesh motion solver.20) ¨ |φ|dt = 0 | − Aφ · (2πf )2 · cos(2πf t)|dt = 8πf 2 Aφ = 2πf Ω. Luo & Sun. v(t) and w(t) in respectively X-. and the reference velocity is calculated at that particular location.23) These relations hold if the wing is the only driving force behind the resulting flow velocity. in general this occurs at T /4. the mesh deformation is symmetric. concerning rotational motion. it is important to identify the average rotation amplitude. available in most commercial and non-commercial CFD codes. Numerical initialisation If a flapping wing is considered. (4.20) by Rg : Uref = 4Aφ f Rg .21) ˙ where the flapping velocity φ was taken from equation (4. Secondly. Y .12). Using equation (4.

The Reynolds number for hovering conditions and using Rg can be written: Re = Uref c 4Aφ f Rg c = . Aswept . 2008b. Additionally. which is illustrated in figure 4.. 2008b). Aswept .4. the Reynolds number at Rg .. J. ν ν The area that is swept by the revolving wing (Usherwood & Ellington. (4. the dimensionless amplitude at Rg . Thaweewat et al. 2002). with f the flapping frequency. U∞ . For completeness and consistency..6. which describes the forward speed with respect to the flapping velocity at a certain radius. two other important parameters are kept constant as well. R: λ∗ U∞ . 1984). A∗ is kept constant for all relevant simulations. 2009).6 Schematic illustration of the kinematic parameters in forward flight. In order to create an appropriate framework for comparison. is given by λ∗ = U∞ /f c. c which is a measure for the dimensionless translation of the selected cross-sectional area. The dimensionless amplitude at the radius of gyration Rg is defined as A∗ = Aφ Rg .24) In forward flight another important parameter is the advance ratio (Shyy et al. = J= 4Aφ f Rg 4A∗ where the reduced frequency (Ellington. is obtained by subtracting the area swept by the wing tip from the swept . According to (Shyy et al.4 Dynamical scaling of flapping wings 90◦ − β 87 Aα 2A∗ sinβ tan−1 (2Stsinβ) λ∗ Figure 4. 2003) for the representing simulations. velocity. a constant A∗ leads to similar wing-wake interactions (Birch & Dickinson. Lentink & Gerritsma (2003) showed that the following relation can be used as a good approximation for both hovering (U∞ = 0) and forward flight conditions: 2 Uref = U Rg ≈ U∞ + (4Aφ f Rg )2 . which is an implicit result of keeping Rg constant and the area swept by the wing. λ∗ is also known as the dimensionless wavelength. λ∗ .

it is deduced that keeping the swept area constant is similar to maintaining a constant Froude efficiency (Stepniewski & Keys. Additionally.6) as Ro = Rg Uref = . If Ro is varied. 2Aφ · (Rtip − Rroot ) · (Rtip + Rroot ).and three-dimensional flapping wing simulations. generating thrust by moving the fluid. For a translating wing. which is inversely proportional to the Coriolis acceleration. Ωc c (4. It is necessary to use a sufficiently large computational domain to minimise disturbances. 2008). the value for Ro is infinite. However. since those values are readily available from literature. the determination of Ro would be easier if defined as Ro = Rtip / c. where Rtip is the wing tip radius. from nearly translating to strongly revolving. the Rossby number is given by Ro = Rg / c. the Rossby number is equivalent to the single wing aspect ratio ARs = Rtip / c. it is efficient . 1984. From (Lentink. When generating a grid around a two-dimensional thin and ellipse-shaped airfoil. a computational domain is necessary to contain the mesh on which the governing partial differential equations are solved.0 for insect and fish (Lentink. 2Aφ · bs · (Rtip + Rroot ).5 Computational domain and boundary conditions In this section the general approach for mesh generation is described. domain size and boundary conditions for these simulations in more detail. 6 and 7. explain the specific mesh generation. Furthermore. for a rotating wing.88 Physical and numerical modelling of flapping foils and wings area by the wing root: Aswept = = = 2 2 2Aφ · (Rtip − Rroot ). where Rtip and Rroot are the radii at respectively the wing tip and root. Lentink. that has been applied for the two. the Rossby number. 2008). 2008). the effect of different rotation origins can be investigated. a geometric wing characteristic. which are dealing with respectively two-dimensional hovering. the radius of gyration divided by the average chord length. needs to be obtained by rewriting equation (4. in general Ro = 3. 4. Ro is finite. 1984). like in real insects. it may be difficult to obtain the radius of gyration. for Rroot = 0. When performing a Computational Fluid Dynamics (CFD) study. Ro. If the wing planform is complex. forward and three-dimensional hovering flight. with appropriate boundary conditions in order to obtain large convergence rates and the correct solution. The distance from Rtip to Rroot is identified as the single wing span bs . Chapter 5.25) When the reference cross-sectional area is located at the radius of gyration (Ellington.

to use an O-type mesh. which are necessary (Wesseling. Thaweewat et al. leading to a small inflow and outflow at the boundaries of the computational domain. 2008. Depending on the flapping configurations. which is shown in figure 4. The meshes for the two-dimensional simulations were generated using Gambit software. GridPro . Γleft and Γright .4. Using conformal mappings (Bos et al. Γfront .. The two-dimensional domain uses the O-type topology (a). it was possible to generate a high quality mesh around a thin ellipsoidal wing in a three-dimensional box-shaped computational domain.6 Definition of force and performance coefficients 89 Γtop Γleft Γwing Γleft Γwing Γback Γright Γright Y Y X Γfront Γbottom Z X (a) Two-dimensional (b) Three-dimensional Figure 4. and the cylinder-shaped outer boundaries. Γright . Γbottom .6 Definition of force and performance coefficients Besides flow field analysis. while the three-dimensional flapping wing simulations uses the boxed topology (b). Generation of a three-dimensional structured mesh around a wing is not an easy task. 6 and 7) it .and three-dimensional simulations (chapters 5. Γleft .7(b). the outer boundaries. When simulating hovering flight. 4. Γtop . 2001) when forward flight conditions are simulated. using advanced visualisation techniques.7 Computational domain and boundary conditions. In order to make a sound comparison of forces and performance for mutual two. Γback . 2009). Using GridPro . outflow or symmetry planes. Γwing . the flapping wing still induces a significant amount of downwash.. Therefore. are set to inflow. an automated topology mesh generator is used. a high quality mesh is generated between the ellipse-shaped wing boundary. the resulting forces acting on the airfoils and wings are of primary importance to assess aerodynamic performance. The outer boundary is split into Γleft and Γright in order to be able to specify inflow and outflow boundary conditions. which is shown in figure 4. see appendix A. which can be very cumbersome using manual procedures used in programs like Gambit or Gridgen .7(a).

S [m] the wing surface and dS [m] represents an infinitely small surface area element. the lift and drag are constructed from the forces in X-. In hovering conditions. Y . hovering or forward flight. together with µ this forms the wall shear stress. If three-dimensional flapping in hovering flight without deviation is considered. such that the wing is not moving in the horizontal plane. hovering or forward flapping flight. as shown in figure 4. The centre of the axes coincides with the origin of rotation of the three-dimensional wing. is important to properly define force and performance coefficients. i. ∂n where Ftot [N] is the total force vector. the general determination of the force coefficients is explained in combination with performance characteristics. the lift force Flift is defined in the vertical direction and equal to FY . where φ(t) is the flapping angle. If the motion includes a deviation velocity. the drag force derivation is more elaborate but similar. when the main flapping direction is around the Y -axis.8. a two-dimensional derivation is trivial.e. The total force vector is integrated over the wing surface and contains a pressure and a viscous contribution.90 PSfrag Physical and numerical modelling of flapping foils and wings FY Z Y ˙ φ X FZ FX Rg Figure 4.or Z-direction.8 Forces on a general three-dimensional flapping wing.or three-dimensional. depending on the type of threedimensional motion. Two different force definitions can be defined. FY and FZ are shown in figure 4. however.3. The term ∂u/∂n is the gradient of the velocity vector with respect to the normal vector to the wall. the definitions of the forces. p [N/m2 ] is the pressure and µ [Ns/m2 ] the dynamic viscosity. is defined in opposite direction of the flapping wing motion. Depending on the type of motion. Fdrag (t) = FX (t) · sin(φ(t)) − FZ (t) · cos(φ(t)). The drag force Fdrag . In this section. The forces are defined in the three-dimensional inertial reference frame. Besides the lift and drag. Forces In general three dimensions. there is a force in the . two. FX . These forces are calculated using the following expression: Ftot = S pdS − µ S ∂u dS. the following lift and drag variations are found: Flift (t) = FY (t).

the ratio between time-averaged lift coefficient CL and time-averaged drag coefficient CD is used to characterise performance. The lift is averaged over the complete flapping period. opposite to the free-stream velocity: Flift (t) = FX (t). Commonly the forces are made dimensionless using the dynamic pressure based on the average velocity. the forces are defined as: CD = Fdrag . The average lift-to-drag ratio. the same inertial reference frame is used as shown in figure 4. however. q·S where CD and CL are the drag and lift coefficients and S the wing surface. This spanwise force is dominated by the viscous wall shear stress and therefore small compared to the lift and drag forces.4. In the case of forward flapping flight the lift force is defined in the positive X-direction and the drag in the negative Y -direction. perpendicular to the horizontal plane (in hovering flight). 2008). the flapping velocity Uflap and the deviation velocity Udev . The reference velocity contains the free-stream velocity U∞ . while for the drag the absolute values are used for averaging. CL /CD ave is chosen as an indicator of aerodynamic performance..3.6 Definition of force and performance coefficients 91 direction of the spanwise coordinate. such that the sign flips at stroke reversal. Fdrag (t) = −FY (t). The drag is opposed to the flapping motion. It is justified to neglect this spanwise force. Hence. 0 where the integration is evaluated over one flapping cycle with period T [s]. also known as the glide number in aerospace engineering. also because it has little relevance to performance. which is non-zero in forward flapping flight. but the direction of the uniform flow is from top to bottom in direction of the negative Y -axis. These force averages are obtained by integration of CL and CD . In forward flight conditions. The main flapping direction is still around the Y -axis. With the strong variation in velocity. q·S CL = Flift . Performance The force coefficients are the major parameters used to assess the influence of the different wing motion models. In addition. When the average lift coefficients of the different kinematic models are matched. it is deemed more appropriate to scale the forces with the mean dynamic pressure itself (Bos et al. The mean dynamic pressure q is defined as: 2 q = 1/2ρUref = 1/2ρ · 1 T T (U∞ + Uflap (t))2 + (Udev (t))2 dt. . The complete system needs to be rotated around the Z-axis in order to get a horizontal orientation of the free-stream.

The radius of gyration is used as a reference cross-section for both twodimensional flapping foil as well as three-dimensional flapping wing simulations. The corresponding coefficients are obtained by averaging the dynamic pressure. 1994). the radius of gyration can be easily obtained. 4. Different flapping wing kinematic models are briefly addressed. Most importantly. Two important dimensionless numbers were identified. 3/2 CL /CD . the governing NavierStokes equations are written in a rotating reference frame. The Rossby number is a way to describe the radius of curvature and thus the angular accelerations in dimensionless terms. from realistic fruit fly kinematics to a fully harmonic model. In order to analyse the flow around flapping wing. In addition to the forces. These numbers are common in general fluid flow.92 Physical and numerical modelling of flapping foils and wings the lift-to-drag ratio is corrected for any differences in lift. (see Ruijgrok. In order to systematically study the aerodynamics around flapping wings at the scale of insects. Additionally. a high lift-todrag ratio effectively means low drag at equal lift. a deformation of the wing is defined in order to study the effects of wing flexing. to design a sound framework for comparison it is necessary to dynamically scale the wing kinematics for all numerical simulations. the Reynolds and Strouhal number. has been used to assess the required power for a certain amount of lift (Wang. Using that planform. namely the Rossby number. but for flapping flight the definition has been slightly changed. a model wing planform has been defined with an ellipsoidal planform. the average Reynolds number and the area swept by the wing result in comparable values. In addition to the rigid body rotations.7 Conclusions This chapter dealt with the physical and numerical modelling of three-dimensional flapping wings and two-dimensional flapping airfoils. which has proved to be proper reference. Therefore. which is used to dynamically scale the wing kinematics. while the drag is opposed to the flapping velocity. the power factor. . 2008). the lift-to-drag ratio is used to assess the flapping wing performance. This flexing motion is defined with respect to the initial wing position and has a time-varying cosine shape. related to the wing rotation. Analysis of the vortical flow around the wings and foils is primarily performed by plotting the force coefficient. This is achieved by scaling the motion parameters such that the dimensionless amplitude. This leads to an extra important dimensionless number. The lift is defined in vertical direction.

whereas the deviation is most likely used for levelling the forces over the cycle. 594. With increasing complexity. simplified models are compared with averaged representations of the hovering fruit fly wing kinematics. To facilitate the comparison. as well as the resulting lift and drag forces were studied. For this. a Robofly model and two more realistic fruit fly models are considered. all dynamically scaled to Re = 110. In addition. The influence of different wing kinematic models on the aerodynamic performance of a hovering insect is investigated by means of two-dimensional time-dependent Navier-Stokes simulations. (2008). Details of the vortex dynamics. light is shed on the effect of different specific characteristic features of the insect wing motion. vol.CHAPTER 5 A 2D investigation of the influence of wing kinematics in hovering flight J. The simulation results reveal that the fruit fly wing kinematics result in forces that differ significantly from those resulting from the simplified wing kinematic models. pp. the parameters of the models were selected such that their mean quasi-steady lift coefficient were matched. The angle of attack variation used by fruit flies increases aerodynamic performance. . 341-368. a harmonic model. Fluid Mech.

Dickinson.2 Influence of kinematic modelling The relevance of (experimental or numerical) simulations of insect flight has been found to depend on how reliable true insect wing kinematics are reproduced. Notwithstanding the possible discrepancy between two-dimensional and threedimensional flow. and with the above justification. Dong et al. In view of the excessive computational expense required for accurate threedimensional simulations. a similar amplitude range was used in the present o research. (2005) and Blondeaux et al. 1994). (2005) and Blondeaux et al. (2004) confirmed that the similarities between two. This chapter deals with the evolution of the forces and the wake originated by a flapping foil in hovering conditions. Both Dong et al. is resolved. (2005b) concluded that two-dimensional studies overpredict forces and performances since the energy-loss. In a two-dimensional simulation our mesh resolution can be higher compared to a three-dimensional simulation. in view of the limitation of computational resources.1 Similarity and discrepancy between two. 5. the present study was restricted to two-dimensional simulations. Reynolds numbers. 5.1. (2005b) numerically investigated the wake structure behind finite-span wings at low Reynolds numbers.and three-dimensional approaches are sufficient to warrant that a reasonable approximation of insect flight can be obtained using a two-dimensional approach. which is present in three dimensions. First. Wang et al. Wang et al. which can then be investigated more thoroughly in three dimensions. two-dimensional analysis has often been applied to obtain insight into the aerodynamic effects of choices in kinematics.94 Influence of wing kinematics in two-dimensional hovering flight 5. in case of advanced and symmetric rotation the forces were found to be similar in the two-dimensional simulations compared to the three-dimensional experiments. airfoil cross-section. Secondly it was observed that in both simulations and experiments the leading-edge vortex did not fully separate for amplitude-to-chord ratios between 3-5 (Dickinson & G¨tz. 1993. (2004) compared three-dimensional Robofly results with two-dimensional numerical results. two-dimensional simulations are performed to get insight in the complicated flow structures. They observed that the flapping wings with low aspect ratio generates three-dimensional vortical structures as was mentioned by Lighthill (1969). etc.1 Introduction In order to investigate the full flow around a three-dimensional flapping wing.1. (2004) and Sane & Dickinson (2001) showed that the kinematic modelling . This showed that two-dimensional simulations are useful to obtain a better understanding of the flow features.and three-dimensional flows In a recent paper Wang et al.

It further emphasises the need to critically assess the influence of kinematic model simplifications. Also.b) and Lentink & Gerritsma (2003) numerically investigated pure harmonic translational motion with respectively small and large amplitudes. Hover et al. were used by Pedro et al. rotao tional parameters were investigated by Dickinson (1994).. (1999) and Sane & Dickinson (2001) with their Robofly. different models from literature were considered. may have on flight performance. such that including rotation is essential. the results were compared with more realistic fruit fly kinematics obtained from the observation of free flying fruit flies (Fry et al. It was also shown that wing stroke models with only translational motion could not provide for realistic results. Slightly more complex fruit fly kinematic models were used by Dickinson et al. Wang (2000a. it can reveal the importance of certain specific .5. the objective of the present study is to compare the effect of the available models as a whole. Using these models. in order to investigate their influence on the aerodynamics. (2004) showed that modelling the angle of attack influences the flapping foil propulsion efficiency to a large extent. with a relatively strong linear and angular acceleration during stroke reversal. This results in the typical ‘sawtooth’ displacement and ‘trapezoidal’ angle of attack pattern of the Robofly kinematic model. which is of great importance to both experiments and numerical simulations. like parameter values and stroke patterns. the effect of amplitude. both the pure harmonic and the Robofly model. Based on observation of true insect flight. This illustrates the appreciable effects which details of the wing kinematics. Besides lift enhancement at certain reduced frequencies. Wang (2000a.1 Introduction 95 significantly influences the mean force coefficients and its distribution. Lewin & Haj-Hariri (2003) performed a similar numerical study for heaving airfoils. In addition to the harmonic models with pure translation (Dickinson & G¨tz. they found periodic and aperiodic flow solutions which are strongly related to the aerodynamic efficiency. including wing rotation. (2003) and Guglielmini & Blondeaux (2004) in their numerical models to solve for forward flight. Harmonic wing kinematics. They varied rotational parameters and showed that axis-of-rotation. different kinematic models have been employed to investigate the aerodynamic features of insect flight. Both studies emphasised the importance of angle of attack modelling to influence the propulsive efficiency.b) varied flapping amplitude and frequency and showed that at a certain parameter selection the lift is clearly enhanced. For example. 1993). They concluded that the airfoil choice is of minor influence. rotation speed and angle of attack during translation are of great importance of the force development during each stroke. This leads to better insights in the consequences of simplifications in kinematic modelling. deviation. angle of attack and the timing of the latter were explored. but large amplitudes lead to an increase of lift by a factor of 5 compared to static forces generated by translating airfoils. In literature. 2003). In the present study. Lentink & Gerritsma (2003) varied airfoil shape with amplitude and frequency fixed at values representative to real fruit flies. Furthermore. the wing maintains a constant velocity and angle of attack during most of the stroke. Additionally. Instead of performing a parameter study within the scope of one kinematic model.

5. In brief. which solves the governing incompressible Navier-Stokes equations on a two-dimensional computational mesh. the fourth model is a slightly simplified version of the latter. Therefore. Henderson (1995) and Williamson (1995) showed that for circular cylinders. 1994). The second model contains the kinematics implemented by Dickinson et al. which solves the Navier-Stokes equations under the assumption of incompressible flow. (2003). At the considered Reynolds number. (1999) for their Robofly at UC Berkeley (presently CalTech).1 Flow solver and governing equations To simulate the flow around moving wings with pre-defined motions the commercial CFD solver Fluent was used. the actual modelling of the insect parameters is discussed in 5. these kinematic models are constructed such that their mean quasi-steady lift coefficients are comparable such that our performance comparison is justified. which supports this assumption. Finally. This basis of comparison is verified a-posteriori from the force results of the actual simulations. In section 5.2 Numerical simulation methods The different kinematic models are implemented in the commercial flow solver Fluent . The resulting model has been validated using stationary and moving circular cylinders and verified using harmonically moving wings. 5.3. observed fruit fly model. In addition. The results of the numerical simulations obtained with the different kinematic models are treated in section 5. Re = O(100). The two-dimensional time-dependent NavierStokes equations are solved using the finite volume method. in relation to aerodynamic performance.4 and concluding remarks are given in 5. This study considers four different wing kinematic models with varying degree of complexity. based on data measured by Fry et al.2.96 Influence of wing kinematics in two-dimensional hovering flight features of the stroke pattern. The mass and momentum equations are solved in a fixed inertial reference frame incorporating a moving mesh following the Arbitrary Lagrangian Eulerian (ALE) formulation (Ferziger & Peric. All these kinematic models are dynamically scaled at a Reynolds number of Re = 110 which corresponds to the flight conditions of the fruit fly. These models are implemented in a general-purpose Computational Fluid Dynamics (CFD) code. the transient incompressible laminar . the first model describes the wing motion using basic harmonics as derived by Wang (2000a). 2002). The third model is a representation of the real kinematics used by a hovering fruit fly (Drosophila Melanogaster ). In addition.2 the numerical simulation methods are described. the transition from laminar to turbulent flow occurs at Re = 180 ± 5. The outline of this chapter is as follows. assuming incompressible flow which is justified since the Mach number of flapping insect flight is typically O(10−3 ) (see Brodsky. the flow is assumed to be laminar.5.

1 O-type mesh topology with boundary conditions on Γ1 . while re-meshing takes place in the outer field Ω2 . 2000b). a symmetry boundary condition was applied at Γ3 for numerical reasons. This complicates the creation of a high quality mesh..2.2) are used. which is shown schematically in figure 5. At the body surface a no-slip boundary condition is applied.56 x 0.e. an O-type computational domain is used. The intermediate interface Γ2 divides the mesh into two separate fields. Additional solver settings can be found in (Bos et al. i. In order to create this body conformal mesh (see figure 5.Γ2 and Γ3 .58 Ω1 Ω2 Γ1 Γ2 Γ3 y 0.42 Figure 5.2 Mesh generation and boundary conditions In order to compute the flow around the moving airfoils.5.2 Numerical simulation methods 97 y 0. The geometric surface gradient is high. The computational domain is divided into two parts: Ω1 and Ω2 for the inner and outer mesh respectively. The body surface Γ1 is located in the centre of the computational domain. generation of a high quality mesh is not as straightforward as for a cylinder. The complete inner mesh moves according to the wing kinematics. For the wing. corresponding respectively to the inner conformal mesh (Ω1 ) and the outer mesh (Ω2 ). high cell orthogonality. The influence of this symmetry condition has been investigated and found to be sufficiently small. Figure 5. 2008.1) and (2. It has the reference length L which corresponds to the wing chord length. such that a free-stream is absent. 2003). Navier-Stokes equations (2. The described computational setup was thoroughly validated using the flow around stationary and moving circular . the flow around the wing is not affected by the mesh regeneration. especially at the leading and trailing edges.2 Body conformal moving mesh around a 2% ellipsoid airfoil.4 x 0.2) a conformal mapping was applied (see Wang. appendix B).1. which is modelled as an ellipse of 2% thickness. Since the moving wing simulations concern hovering insect flight. Since re-meshing occurs at a distance of 25 to 30 body lengths away from the wing. The outer boundary Γ3 is located at 25L such that the influence of the far field boundary condition is negligible (Lentink & Gerritsma. 5.

The grid resolution near the wing. The linear displacement of this cell is y and yref is the original length of this cell. simulating 18 flapping periods needed approximately 10 days on one serial AMD Athlon 2500+ CPU. In order to minimise the interpolation errors from one time-step to the next it is important to analyse the influence of the relative cell displacements. it was shown that a relative displacement of 10% in both rotational and translational direction leads to accurate results with differences in drag coefficients remaining below 5%. Therefore. From the relative displacements in rotational and translational direction follow the constraints for the size of the time-step in order to keep the interpolation errors within limits. up to 1 chord length. The relative displacements in rotational and translational direction are defined as ǫr = and ǫy = △α αref △y 2fe Ae N △t = . amplitude and number of cells on the surface. yref yref Here α corresponds to the angular displacement of the reference cell. (2008) (appendix D) investigated the mesh and time-step independence for the nominal solver settings using harmonic wing kinematics for hovering flight.. Ae and N correspond respectively to. Additionally. . Bos et al. At this mesh.98 Influence of wing kinematics in two-dimensional hovering flight t + dt y x Figure 5. In (Bos et al. fe . The airfoil simulations were performed on a mesh of 50000 cells with 2000 time-steps within one motion period. appendix C). the frequency. One run. Furthermore. 2008. The computational efforts are acceptable: 2000 time-steps within one excitation period.. which is illustrated for the rotational motion in figure 5. cylinders (Bos et al. while αref is the original radial length of this cell. appendix C). the motion of a reference cell was investigated.3 αref + dα αref t Relative cell displacement in rotation. the size of the first cell at the wing surface varies between 2% and 50% of the wing thickness at the leading-edge and in the middle of the profile respectively. 2008.3. was 8800 (176x50) cells such that the leading and trailing edge vortices where captured with at least 1000 cells.

5.3 0. 5. 2004).82. Further details of the validation and verification studies can be found in (Bos et al. our force distribution looks similar for both cases.5 1 0. Our forces are normalised with the maximum of the quasi-steady force. The amplitude was 2. Moreover. A common procedure is to define an equivalent two-dimensional geometry.7 0. (2004) (◦).8 times the chord length.3 Validation using harmonic wing kinematics The main numerical parameters.8 0. while . just as in (Wang et al.3 0. which corresponds to Re = 75. Generally.. 2004) the drag in figure 5.7 0. Comparison of lift (a) and drag (b) coefficients using harmonic wing kinematics with A = 2.8 0. The mean lift and drag coefficients are 0.5 1 0.2. Only just after stroke reversal our computation finds a larger lift and drag which is probably the result of different numerical dissipation properties of both codes.5. (2004).9 1 t/T [-] Drag coefficient [-] Lift coefficient [-] 0.1 0.3 Modelling insect wing kinematics 1. In similarity to (Wang et al. 1. (2004) for similar but not entirely identical conditions. are used to validate our results with those obtained by Wang et al.84.4 Comparison of force coefficients between the present simulations and Wang et al.4 0.5 0 -0.4 0.4(a) is defined to be positive in the direction opposite to the horizontal motion.4 shows the lift and drag coefficients for validation purposes.5 0 -0. with a moving wing according to harmonic kinematics. a mesh size of 50000 cells and 2000 time-steps within one excitation period. 1. within the context of comparing results of different stroke patterns. Re = 75 for the present study (•) and obtained by Wang et al. compared to 0.5 -1 0 (b) 99 0. appendix C and D).44 obtained by Wang et al.2 0. which is a difference of only 2% in lift and drag and therefore.3 Modelling insect wing kinematics In order to derive the two-dimensional kinematic models the three-dimensional degrees of freedom need to be converted to their two-dimensional counter-parts. the present numerical method is proved to be accurate.6 0.1 0..47 for our simulation. Figure 5. 2008.8.6 0.9 1 t/T [-] Figure 5..5 -1 0 (a) 1. (2004).5 0. A twodimensional case was selected.5 0.2 0. the computations were considered to be sufficiently accurate.

1 in terms of wing selection and model parameters.3. Re = O(100). In the present study. Wang et al. From (Sane & Dickinson.3.3.3. a different argument for the selection of the projection . This two-dimensional set-up is derived in section 5.5 (see Sane & Dickinson. α(t) to the geometrical angle of attack and θ(t) to the deviation from the horizontal stroke plane. shown in figure 5. with respect to the horizontal plane and the deviation from the horizontal plane θ.3. where R is the wing span. Lentink & Gerritsma (2003) found this airfoil an acceptable choice to model insect wings at low Reynolds numbers. as is shown in figure 5. 2001).. 5. 2001.100 Influence of wing kinematics in two-dimensional hovering flight Figure 5. Dickinson et al.5 Illustration of the main motion directions. the angle of attack α. maintaining the characteristic aspects of the wing motion.6.. 2003).2 and 5. 1999). In this threedimensional model the three degrees of freedom of the wing motion are defined as the angular displacement φ in the mean stroke plane. φ(t) corresponds to the stroke variation.1 Insect wing selection and model parameters The computational approach is applied to investigate the influence of different kinematic wing motion models on the aerodynamic performance. The deviation causes a ‘figure-of-eight’ pattern which is present in real fruit fly kinematics (see Fry et al. The dynamical scaling and the force definitions are described respectively in 5. The two-dimensional projection is to be defined at a representative spanwise location such that the motion is confined to an arc around the wing root. (2004) used this distance to derive their two-dimensional model. The two-dimensional airfoil shape is chosen to be a 2% thick ellipsoid. Birch & Dickinson (2003) found strongest vorticity at a spanwise location of 0. Therefore.65R from the wing root. The different kinematic models are illustrated using the Robofly experimental set-up.

3.. based on the mean chord length. Apparently.42 · 10−4 m4 . the conversion from three-dimensional angles to non-dimensional displacements is given by: φ · Rg θ · Rg x= . 0 (5.6 Force definition on the two-dimensional airfoil. the spanwise location was selected to be at the radius of gyration where the mean lift acts (Ellington. The centre of rotation is defined in the aerodynamic centre which lies at the quarter chord point of the mean chord. For the radius of gyration the following value was obtained Rg = 0.0167 m2 the wing tip radius R=0.0882 m.1) c c where Rg is the radius of gyration. Finally. This leads to a value for the mean chord length of c = 0. 1984). 5. 2003): U= 1 T T u2 + v 2 dt.0667 m and the moment of inertia Icg =40. Lref .6396 · R. In view of providing completeness on the three-dimensional set-up. Both the displacement x and the deviation y have been made dimensionless with the mean chord c. 2003) was used. A definition of the mean chord length based on the moment of inertia around the wing root was proposed.082 m.254 m. 2004) the current cross-section is less than 2% closer to the wing root. the location of the wing base xbase =0.3 Modelling insect wing kinematics Fy Fr 101 Fx M xac α xcg c Figure 5.2 Dynamical scaling of the wing model Since the flapping of the wings induces highly unsteady flow the relevant flow and motion parameters have to be scaled dynamically. When comparing this distance to the value used by (Wang et al. y= . The period of the motion is used to average the relevant flow velocity (Lentink & Gerritsma. Another important parameter to be defined is the reference length. the used values are for the wing surface S=0.5. the location of centre of gravity xcg =0. location (Lentink & Gerritsma. (5. Considering that the local velocity of each cross-section varies during flapping. the mean lift acts nearly at the location where the vorticity is maximal.2) .

CL . (5. while the drag is taken equal to the horizontal force Fx . In addition. Both are given by u = ∂x/∂t and v = ∂y/∂t.102 Influence of wing kinematics in two-dimensional hovering flight Here T [s] is the period . Since the average lift coefficients of the different kinematic models are matched. From (5. where the integration is evaluated over one flapping cycle. The mean dynamic pressure q is defined as q = 1/2ρU 2 = 1/2ρ · 1 T T 0 The definition of the drag and lift forces is shown in figure 5. the ratio between time-averaged lift coefficient. it is deemed more appropriate to scale the forces with the mean dynamic pressure itself. The lift is equal to the vertical force Fy . 5. With the strong variation in velocity. φ and θ the three-dimensional kinematic angles for the displacement and deviation. u represents the non-dimensional velocity in the stroke plane and v the non-dimensional deviation velocity. q·c q·c ∂x ∂t 2 + ∂y ∂t 2 dt. The average lift-to-drag ratio. a high lift-to-drag ratio effectively means low drag at equal lift. CL = . Substituting equation (5. however. These force averages are obtained by integration of CL and CD .3) and (5. The force coefficients are the major parameters used to assess the influence of the different wing motion models. the lift-to-drag ratio is corrected for any differences in lift. The lift is averaged over the complete period. while for the drag the averages are per half stroke. Therefore. where t = t/T is the dimensionless time. the forces are defined as Fx Fy CD = .1) into (5. Hence. also known as the glide number in aerospace engineering. CD . We fixed the Reynolds number to Re = 110.6. is used to characterise performance. defined positive in the positive x-direction. .4) it can be observed that the Reynolds number Re depends solely on the frequency f for a given displacement φ(t) and deviation θ(t). CL /CD ave is chosen as an indicator of aerodynamic performance.3. and timeaveraged drag coefficient. The Strouhal number St is not to be varied independently.3) 1 0 ( ∂φ )2 + ( ∂θ )2 ∂t ∂t (5.3 Force and performance indicators where CD and CL are the drag and lift coefficients. Commonly the forces are made dimensionless using the dynamic pressure based on the average velocity.4) Here f = 1/T is the frequency.2) and evaluating. the following relations for the Reynolds and Strouhal numbers were derived: Re = and St = Uc f Rg c = · ν ν fc c = · Rg U 1 ( 0 ∂θ ∂φ 2 ) + ( )2 ∂t ∂t 1 .

In order to investigate the fact that the observed fruit fly kinematics lacks an exact symmetry in the wing stroke pattern. a symmetrical model was constructed. referred to as the symmetric fruit fly model. Two of these models. appendix A). the mean lift coefficient is predicted well using this theory..7(d). are shown in figure 5. 2003) and is therefore considered as the most realistic fruit fly kinematic model.5. 2008. Neither the displacement. four different kinematic models. In view of the limitations of the quasisteady theory.3 Modelling insect wing kinematics 103 5.4 it is shown that the computed mean lift coefficient of the numerical simulations are reasonably well matched for all models. Within this model the motion is identical for the downstroke and upstroke. angle of attack nor deviation is symmetric during the flapping period. different kinematic models were constructed. just as in the harmonic model. 1999). angle of attack and deviation. The first of the four models is described by pure sine and cosine functions and will therefore be referred to as the harmonic model (see Wang et al. Analysing those aspects leads to a better understanding of how the fruit fly may benefit from kinematic features which are absent in the simpler models.4. which are not fully present in the quasisteady theory. The second model takes the wing kinematics as used in the Robofly model (Dickinson et al. in section 5. is derived from measurements on real fruit flies (Fry et al. Like the realistic . This model does include the deviation which results in a ‘figure-of-eight’ pattern.. which provides an a-posteriori justification of our choices for the model parameters. shown in figure 5.7(b) it is shown that the flip from down to upstroke is postponed to the end of the translational phase which results in the ‘sawtooth’ shape of the displacement. with different degree of complexity. 2004). displayed in figure 5. and reveals the relevance of including these aspects in theoretical models. have been analysed. Although according to Sane & Dickinson (2001) the mean drag is strongly influenced by the unsteady flow physics.7(a). The third model represents the actual fruit fly kinematics as observed in experiments and the last one was a modification of the latter. However..7(c).3. The deviation is zero. The third model. The displacement. In order to facilitate the comparison the model parameters are chosen based on matching the mean quasi-steady lift coefficient (Bos et al. Large accelerations at stroke reversal are the result. The characteristic shapes of each model are described. Subsequently they are used to investigate the influence of the models on the force histories and the performance in section 5. In figure 5.. Using quasi-steady theory. such that their quasi-steady lift coefficients are matched within 1%. chosen to investigate the effect of symmetry in the wing motion. the pure harmonic motion and the Robofly experimental kinematics have appeared in literature. For the symmetric models this force is equal to the resultant force. the difference between predicted and simulated values is expected to exceed this 1% tolerance.4 Different wing kinematic models Since the main purpose of this study is to investigate the influence of wing kinematics on the aerodynamic performance during hovering fruit fly flight.

5 0.1T to t = 0. θ[◦ ] 100 80 60 40 20 0 -20 -40 -60 -80 0 (d) 0.7 Robofly model.9 1 t/T [-] Kinematic angles of the different kinematic models.1 0.6 0.8 0.7 0.9 1 t/T [-] φ. fruit fly model this symmetric model includes a time-dependent deviation such that the observer sees a ‘figure-of-eight’ pattern of the wing. (d) simplified fruit fly model. (c) fruit fly model. This ‘trapezoidal’ shape of α is characteristic for the Robofly and may be influencing the performance. The most obvious peculiarity of the realistic fruit fly models is the extra ‘bump’ in angle of attack just after stroke reversal.2 0.4 0.7 0. •: displacement angle φ.3 0.2 0.9 1 t/T [-] 20 0 -20 -40 -60 -80 0 (b) 0.6 0.8 0. α and θ for each model it becomes possible to identify certain important differences.2 0. Neither of those last two realistic kinematic models can be described by using simple analytical functions without losing significant detail.4 0. φ. α.4T the angle of attack flattens at a value of almost 40◦ .1 0. The Robofly initially has a larger gradient in time of the angle of attack compared to the harmonic case. : deviation angle θ. ◦: angle of attack α. .6 0.3 0.1 0.5 0.6 0.104 100 80 60 40 φ.5 0. θ[◦ ] 20 0 -20 -40 -60 -80 0 (c) 0. During translation from about t = 0.4 0.1 0. (b) Figure 5. θ[◦ ] 0. α. Although the Robofly model clearly shows similarities with the fruit fly models the latter has some typical additional features.7 0.3 0.5 0. When comparing the motion parameters.7 0.8 0.3 0.4 0.7(a) and (b).8 0. θ[◦ ] 20 0 -20 -40 -60 -80 0 (a) Influence of wing kinematics in two-dimensional hovering flight 100 80 60 40 φ. see figure 5. α. α.9 1 t/T [-] 100 80 60 40 φ. (a) Harmonic model.2 0.

In order to assess the effect of these kinematic features in isolation. The first is an overall comparison of the complete kinematic models. The differences of the obtained mean lift coefficient are significantly smaller than the differences in lift-to-drag ratios. as baseline. This last model is used to investigate the influence of deviation on the force histories and performance. This implies that strong translational and rotational accelerations occur at stroke reversal. the harmonic model. the Robofly model. the realistic fruit fly model and the simplified fruit fly model. It is also observed that the deviation is negative for a certain period during the upstroke. in terms of actual vortex dynamics.7(d). 5. The harmonic and Robofly models lack deviation. the comparison is made using the simplest model.4. which characterises aerodynamic performance. In the second study the effect of the characteristic features identified above. but instead of flattening α. the fruit fly wing α descends to the ‘bump’.4 Results and Discussion In the previous section it was observed that the most interesting aspects of the Robofly kinematic model are the ‘sawtooth’ displacement and the ‘trapezoidal’ angle of attack. as well as the resulting lift and drag histories are studied in 5. The mean drag. are considered in more in detail.1. During stroke reversal the gradient of α matched the harmonic model closer than the Robofly with its high gradients. Therefore. It follows the same high angular velocity. After the ‘bump’ the angle of attack more or less matches the plateau found in Robofly but starts to increase earlier.7(c)). the deviation of the realistic fruit fly is averaged to derive the simplified fruit fly model. as well as the average lift-to-drag ratio.4. extra ‘bump’ in angle of attack and the presence of deviation. Therefore.1 the mean force coefficients are given for the four complete models. The deviation of the fruit fly model is asymmetric during the complete cycle. The more realistic fruit fly models are characterised by a ‘bump’ in angle of attack and the presence of deviation. Hereto this baseline model is subsequently modified by adding respectively the ‘sawtooth’ displacement. and lift coefficients are given.4 Results and Discussion 105 compared to the Robofly (figure 5. for each half-stroke.7(b) and (c)). Results of two comparative studies were presented. which is described in section 5.1 Overall model comparison In table 5. the conclusions on the performance comparison are considered to be significant. but also during each half stroke (figure 5. see figure 5. The results of this comparison.4.‘trapezoidal’ angle of attack. This is likely to influence the performance since the effective angle of attack is altered due to deviation. so no ‘figure-of-eight’ is present. The mean drag for the harmonic and Robofly models is substantially higher . the harmonic model. 5.5.2.

8 0.1 0.8 0. .466 1.448 1.6 0. : realistic fruit fly model. kinematic model harmonic Robofly realistic fruit fly simplified fruit fly Table 5.387 1.5 0.577 1.7 0.1 0.9 1 t/T [-] Figure 5.848 2.417 1.9 1 t/T [-] Drag coefficient [-] 8 6 4 2 0 -2 -4 -6 -8 0 (b) 0.6 0.1 0.106 Influence of wing kinematics in two-dimensional hovering flight 5 4 Lift coefficient [-] 3 2 1 0 -1 0 (a) 0. ◦: Robofly model.5 0. 8 6 Drag coefficient [-] 4 2 0 -2 -4 -6 -8 0 (a) 0.3 0.540 1.6 0.4 0.4 0.5 0.596 CL /CD ave 0. ◦: Robofly model.4 0. : realistic fruit fly model. •: harmonic model.8 0.2 0.4 0.2 0. ▽: simplified fruit fly model.805 0.7% −8.1 CL 1.3 0.9 1 t/T [-] Lift coefficient [-] 5 4 3 2 1 0 -1 0 (b) 0. ▽: simplified fruit fly model.839 2.1 0.8 Lift coefficient histories of the baseline kinematic models.8 0.7 0.5% Time-averaged force coefficients using the complete baseline models.012 −CD up 1.483 1.6 0.115 −29% −49% baseline −1.9 1 t/T [-] Figure 5.7 0.335 1.5 0.3 0.0% baseline −5.454 −3.7 0.3 0. •: harmonic model.2 0.9 Drag coefficient histories of the baseline kinematic models.132 1.6% CD down 1.2 0.

454 1 2 3 (d) x c [−] 0 -3 -2 -1 x c [−] 0 1 2 3 Figure 5. (d) symmetric fruit fly model. (a) harmonic model. (c) realistic fruit fly model.5.10 Force vectors during each half-stroke.417 1 2 3 (b) x c [−] 0 -3 -2 -1 x c [−] 0 1 2 3 ¯ CL = 1.483 -3 -2 -1 (a) ¯ CL = 1.54 -3 -2 -1 (c) ¯ CL = 1. (b) Robofly model.4 Results and Discussion 107 ¯ CL = 1. .

49%. Therefore. compared to the realistic fruit fly model. the fruit fly models perform better than the less complex models. it can be observed that within the model assumptions. It can be seen in figure 5. figure 5.11 Vorticity contours around a flapping airfoil.108 Influence of wing kinematics in two-dimensional hovering flight (a) harmonic model (b) realistic fruit fly model Figure 5. the mean drag coefficient of the simplified fruit fly is not symmetric.10 (force vectors).1. the mean drag contribution of the leading-edge vortices (LEV) is higher. compared to the fruit fly models.10.11 shows the vorticity contours of the realistic fruit fly model compared with the harmonic model. These performance increases are the result of the lower drag coefficients in both fruit fly models due to certain beneficial . The different kinematic patterns are also illustrated in figure 5. the drag during the upstroke is about 57% higher than during the downstroke. corresponding to negative vorticity values). When comparing the lift-to-drag ratios in table 5. Vorticity contours are shown for t = 0. the realistic fruit fly model results in a significant decrease in drag of 29% at comparable lift.1T (blue: clock-wise. Furthermore. The ‘sawtooth’ shaped Robofly displacement could possibly play an important role as is discussed in the next section. Compared to the harmonic model. which shows the resultant force vectors during a full stroke for those baseline kinematic models. which is attributed to the complex vortex dynamics.7(c). i. This drag increasing effect is even larger in case of the Robofly model due to the ‘trapezoidal’ angle of attack. the average value during a complete stroke matches the mean drag coefficient obtained with the realistic fruit fly model.e. Figure 5. Nevertheless. The decrease in effective angle of attack in the realistic fruit fly model is also enlarged by the presence of the ‘bump’. This is also illustrated in figure 5.8 and 5.7(a) that the effective angle of attack is higher in the harmonic case.9 (lift and drag histories) and figure 5. The difference with the Robofly model is even larger.

13(b). This leads to a larger shear layer to form a stronger vortex.1T for the harmonic model and the one with the appended ‘sawtooth’ shaped displacement. see figure 5.1T and t = 0. The current results provide insight into the effects of certain specific kinematic features. In order to determine the effect of this shape the harmonic model is extended by this ‘trapezoidal’ angle of attack. the purely harmonic model was appended with the Robofly displacement and the results were compared with the ones obtained using the original harmonic model. From figure 5.13. This leads to the non-zero mean horizontal force along a complete . The lift peaks are almost equal but the drag peaks are significantly larger for the ‘sawtooth’ case. A surprising and unexpected observation is the asymmetry in the periodic force distribution for the ‘trapezoidal’ angle of attack notwithstanding the symmetry of the kinematics. one has to be cautious when extrapolating these results to real flying flies since in reality not every flapping period displays exactly the same kinematic profile. the Robofly uses a ‘trapezoidal’ shape for the angle of attack.5. Next. 5. the force histories during one full stroke are shown in figure 5. Figure 5. at t = 0.1T . The stronger LEV at the beginning of the downstroke in the ‘sawtooth’ case is most likely caused by the higher velocity gradient. The results are compared with those obtained with the original harmonic model. the drag peak is larger than the lift peak. In addition. The lift and drag coefficients are plotted in figure 5. On the other hand.4. respectively.3% with respect to the harmonic case. see figure 5.14(a) and (b) the vorticity contours are plotted at t = 0.2 Kinematic features investigation Influence of ‘sawtooth’ displacement used by the Robofly The ‘sawtooth’ shaped displacement of the Robofly is investigated in isolation to assess its influence on the force histories and the aerodynamic performance. which are repeated since the motion is symmetric. the individual influences of the different interesting kinematic shapes are studied.13 it is observed that compared to the harmonic model the global force histories look similar. The larger mean drag is reflected in the integrated values in table 5.4 Results and Discussion 109 kinematic features.15. at the end of the half-stroke the wing decelerates faster in the ‘sawtooth’ case which results in a lower strength in the LEV. Influence of ‘trapezoidal’ angle of attack used by the Robofly In combination with the ‘sawtooth’ displacement.2. Due to this larger drag during each stroke.2. In figures 5. This also explains the larger mean drag compared to the harmonic model which can be read from table 5. Since the wing orientation is almost vertical. From figure 5. Therefore.14 it can be seen that the LEV is stronger for the ‘sawtooth’ case which explains the higher drag peak.12(a) shows the force vectors acting on the wing during the up and downstroke.12(a) for the force vectors.4T . However. which shows a decrease of 24. Two force peaks are observed close to t = 0. the ‘sawtooth’ shaped displacement leads to a lower lift-to-drag ratio.

(c) harmonic model with ‘extra bump’ α. . (d) harmonic model with ‘deviation’ θ. (b) harmonic model with ‘trapezoidal’ α.366 -3 -2 -1 (a) ¯ CL = 1.110 Influence of wing kinematics in two-dimensional hovering flight ¯ CL = 1. (a) harmonic model with ‘sawtooth’ φ.12 Force vectors during each half-stroke.323 1 2 3 (d) x c [−] 0 -3 -2 -1 x c [−] 0 1 2 3 Figure 5.351 1 2 3 (b) x c [−] 0 -3 -2 -1 x c [−] 0 1 2 3 ¯ CL = 1.483 -3 -2 -1 (c) ¯ CL = 1.

8 0.6 0. Lift (a) and drag (b) histories to investigate the influence of the ‘sawtooth’ displacement compared to the harmonic model.3 0.7 0.3 0.2 0.1 0. corresponding to negative vorticity values) . Vorticity contours are shown for t=0.4 0. ◦: harmonic α.1 0. •: harmonic φ.θ.14 Vorticity contours around a flapping airfoil.6 0.8 0.2 0.5. θ and Robofly φ.5 0.4 0.4 Results and Discussion 111 6 Drag coefficient [-] 0.α.9 1 t/T [-] (b) Lift coefficient [-] 4 2 0 -2 -4 -6 0 (a) 6 4 2 0 -2 -4 -6 0 0.7 0. (a) harmonic model (b) harmonic model with ‘sawtooth’ displacement Figure 5.9 1 t/T [-] Figure 5.1T (blue: clock-wise.5 0.13 Lift and drag coefficients.

8%) CD down −CD up 1.3 %) (+15.1 0. θ + Robofly φ φ. harm. .930 0. harm. α and θ α.7 0.323 CL (baseline) (−7.5 0.733 1.112 Influence of wing kinematics in two-dimensional hovering flight kinematic model harm. φ and Robofly α.807 1.8 0.15 Lift and drag coefficients.221 1.9 1 t/T [-] (b) Lift coefficient [-] 4 2 0 -2 -4 -6 0 (a) 6 4 2 0 -2 -4 -6 0 0. •: harmonic φ.6 %) (−8.7 0.848 2.9%) (−8.3 0.3 0.2 Time-averaged force coefficients to investigate the influence of kinematic shapes.9 1 t/T [-] Figure 5. θ + fruit fly.4 0.969 1.608 0. harm. α φ. θ 1.4 0. φ. 6 Drag coefficient [-] 0.302 1.5 0.α.738 (baseline) (−24.0%) (−10.θ.483 1. ◦: harmonic α. α + fruit fly.6 0. Each characteristic shape is varied with respect to the harmonic motion model.8 0.537 0.2 0.2 %) Table 5.240 2. Lift and drag histories to study the influence of the ‘trapezoidal’ angle of attack compared to harmonic model.483 1.250 2. θ + Robofly α φ.3 %) (−33. harm.6 0.2 0.804 0.9%) (0.366 1.351 1.1 0.776 CL /CD ave 0.839 2.

see figure 5. Since large angle of attacks cause high velocity gradients over the leading-edge. one observes stronger and more pronounced vortices in the wake of the ‘trapezoidal’ case. Using figure 5.2. From figure 5.15 it is clear that at the beginning of a stroke the lift peak of the ‘trapezoidal’ case is larger. This leads to a significant performance decrease of 33.12(c) . at the end of each stroke. the angle of attack is larger at the early start of a stroke compared to the harmonic model. compared to the harmonic model. the mean lift is slightly decreased whereas the mean drag is increased.12(b).16 Vorticity contours around a flapping airfoil. This can be explained as follows. To make comparison plausible the symmetric ‘bump’ variation used in the simplified fruit fly model is used to compare results with the harmonic model. Taking a closer look at figure 5. In the ‘trapezoidal’ case the wing reaches the maximum angle of attack earlier in the stroke.16 this is illustrated at the beginning of the upstroke using vorticity contours.4 Results and Discussion 113 (a) harmonic model (b) harmonic model with ‘trapezoidal’ α Figure 5. Another interesting result is the low second peak in the lift. This results therefore in a lower second peak since the LEV has decreased in size and strength. Vorticity contours are shown for t=0. The LEV is larger in case of the ‘trapezoidal’ angle of attack.17(b). see table 5. since the complex vortex dynamics are non-linear and asymmetric. corresponding to negative vorticity values) stroke cycle. Altogether. Influence of extra ‘bump’ in angle of attack used by the fruit fly The fruit fly models are subject to an extra ‘bump’ in angle of attack. larger vortices occur in the beginning of a stroke. This could indicate a larger amount of vortex shedding during the period when the angle of attack is nearly constant. Figure 5.3% due to the ‘trapezoidal’ angle of attack variation. the force distributions are not.5. Although this model is symmetric. Therefore.6T (blue: clock-wise.

This is also the main reason for the lower drag during the downstroke. It is also noted that this case results in asymmetric force distributions as was the case when using the ‘trapezoidal’ angle of attack.20 shows the vorticity at the beginning of the upstroke at the time of the ‘bump’. From figure 5. Vorticity contours are shown for t=0. The decrease in effective angle of attack as a result of the ‘bump’ is considerable compared to the harmonic case. From table 5.18 it is observed that the extra ‘bump’ generates an extra lift peak at the beginning of the downstroke. This causes the loss in lift just after stroke reversal in case of the ‘bump’ angle of attack compared to the harmonic model. On the other hand the drag is slightly increased during the upstroke such that the mean lift-to-drag ratio is still increased with more than 15. The change in angle of attack due to the extra ’bump’ is shown when figure 5. However.2 it is seen that using this feature the mean lift does not change significantly. the drag during the downstroke is very much affected. compared to the harmonic case. present in the . In figure 5. the LEV provides nearly complete lift since the wing orientation is approximately horizontal. A decrease of at least 30% in mean drag is found. Figure 5.19(a) and (b) are compared.6%.17 Vorticity contours around a flapping airfoil.18 the lift and drag forces are shown for the harmonic model with and without the symmetric ‘bump’ in angle of attack. corresponding to negative vorticity values) shows the force vectors during up and down stroke. The LEV is larger compared to the case with the extra ‘bump’ in angle of attack. Influence of wing deviation used by the fruit fly The last important characteristic of the kinematics is the deviation.114 Influence of wing kinematics in two-dimensional hovering flight (a) harmonic model (b) harmonic model with ‘trapezoidal’ α Figure 5.4T (blue: clock-wise. The same was found for the Robofly case. for the case with the ‘bump’ in angle of attack. Therefore.

4 0.2 0.θ. corresponding to negative vorticity values) . ◦: harmonic α.5. •: harmonic φ.19 Vorticity contours around a flapping airfoil.7 0. φ and fruit fly α.1T (blue: clock-wise.5 0.6 0.2 0.1 0.4 0.8 0.5 0.6 0. Vorticity contours are shown for t=0.3 0. (a) harmonic model (b) harmonic model with ‘extra bump’ in α Figure 5.3 0.α. Lift (a) and drag (b) histories to study the influence of the extra ‘bump’ in angle of attack.9 1 t/T [-] (b) Lift coefficient [-] 4 2 0 -2 -4 -6 0 (a) 6 4 2 0 -2 -4 -6 0 0.9 1 t/T [-] Figure 5.8 0.1 0.4 Results and Discussion 115 6 Drag coefficient [-] 0.7 0.18 Lift and drag coefficients.

8% and the mean drag is almost not affected by the presence of deviation.1T .2. The influence of the deviation is relatively large since the deviation increases the effective angle of attack considerably just after stroke reversal.20 Vorticity contours around a flapping airfoil. The mean lift is decreased by 10. Compared to the harmonic model. It is also revealed that the force distributions remain symmetric. The flow dynamic mechanism for this is shown in the vorticity visualisations of figure 5. about 2% − 4% difference in both strokes. Vorticity contours are shown for t=0. which is higher compared to the harmonic case. At the end of a stroke the wings move up again which leads to a decrease in effective angle of . the deviation causes a slightly stronger LEV at t = 0.116 Influence of wing kinematics in two-dimensional hovering flight (a) harmonic model (b) harmonic model with ‘extra bump’ in α Figure 5. This motion perpendicular to the stroke plane is illustrated in figure 5. Since deviation could introduce a large velocity component perpendicular to the stroke plane. see table 5.6T (blue: clock-wise. the effective angle of attack is highly affected. However.6T ) and end (t = 0.12(d) which also shows the force vectors.1T and t = 0. described by the wing tip instead of wing motion solely in the stroke plane.21 shows the force coefficients during one flapping period with deviation added to the harmonic model. The mean lift and drag are not strongly influenced by the deviation.22 which shows the vorticity at the beginning of the stroke. Figure 5. Just after stroke reversal a lift peak occurs. corresponding to negative vorticity values) realistic and simplified fruit fly model. This deviation causes a ‘figure-of-eight’ pattern. The large influence of the deviation on the variation of the lift force is observed at the start (t = 0.9T ) of each stroke. at the end of each stroke the harmonic lift peak was decreased by the deviation. It appears that the force distribution is levelled or balanced by the deviation.4T and t = 0.

2 0.22 Vorticity contours around a flapping airfoil.4 0.2 0.9 1 t/T [-] Figure 5.21 Lift and drag coefficients.5 0. ◦: harmonic α. Lift (a) and drag (b) histories to study the influence of the deviation compared to harmonic model.6 0.9 1 t/T [-] (b) Lift coefficient [-] 4 2 0 -2 -4 -6 0 (a) 6 4 2 0 -2 -4 -6 0 0. •: harmonic φ.5 0.4 Results and Discussion 117 6 Drag coefficient [-] 0. Vorticity contours are shown for t=0.5. corresponding to negative vorticity values).7 0.α.1 0.1T (blue: clock-wise. φ and fruit fly θ.6 0.4 0.1 0. .8 0.7 0.8 0. (a) harmonic model (b) harmonic model with deviation Figure 5.3 0.3 0.θ.

. with different complexity. This leads to the suggestion that a fruit fly may use the deviation to level the wing loading over a flapping cycle. the deviation is levelling the force distributions while the mean lift and drag are almost unaffected. attack. The third model represents the actual fruit fly kinematics as observed in experiments and the last one is a modification of the latter. 2004). Two of these models. corresponding to negative vorticity values).118 Influence of wing kinematics in two-dimensional hovering flight (a) harmonic model (b) harmonic model with deviation Figure 5.6T (blue: clock-wise.23(a) and (b) show LEV’s of comparable strength for both cases.5 Conclusions The effect of wing motion kinematics on the aerodynamic characteristics of hovering insect flight was investigated by means of two-dimensional numerical flow simulations. The fruit fly models are characterised by a ‘bump’ in angle of attack and the presence of deviation. chosen to investigate the effect of symmetry.23 Vorticity contours around a flapping airfoil. Figure 5. The most prominent aspects of the Robofly kinematic model are the ‘sawtooth’ displacement and the ‘trapezoidal’ angle of attack. Summarising. The results of the present two-dimensional study may provide useful insights in the understanding of real three-dimensional insect flight (Wang et al. Threedimensional studies are needed to investigate to what extent this effect is also present in real insect flight. To facilitate the comparison all models are dynamically scaled at Re = 110 and constructed such that their mean quasi-steady lift . 5. pure harmonic motion and Robofly experimental kinematics have appeared in literature. have been analysed using two-dimensional time-dependent Navier-Stokes simulations. Vorticity contours are shown for t=0. Four different kinematic models.

may lead to drag reduction or force levelling compared to harmonic kinematics. First. The trend that the fruit fly kinematics increases aerodynamic performance agrees well with the predictions of the quasi-steady theory. caused the LEV to separate during the translational phase. extra ‘bump’ in angle of attack and the presence of deviation. To investigate which aspects of the kinematic shapes are the most important.6%. It was shown that the difference in performance in terms of mean lift-to-drag ratio between the different kinematic models was significant. This leads to a significant decrease in drag which improves aerodynamic performance in the sense of lift-to-drag ratio by 15. Hereto the harmonic model was extended by respectively the ’sawtooth’ displacement.3%. they were compared to the harmonic model. In particular they indicate that kinematic features. During the beginning of the up and downstroke the ‘bump’ decreases the angle of attack such that the wing orientation is almost horizontal. the ‘trapezoidal’ angle of attack. It was found that the realistic fruit fly wing kinematics result in significantly lower drag at similar lift compared with the simplified wing kinematic models used in literature. The other realistic kinematic feature is the deviation. The extra ‘bump’ in angle of attack. the effective angle of attack is altered such that the deviation leads to levelling of the force distribution. . found in fruit fly kinematics.3%. ‘trapezoidal’ angle of attack. like the extra ‘bump’ in angle of attack and deviation. as used by the fruit fly model. Also in this case large accelerations at stroke reversal lead to a decrease in lift-to-drag ratio of 33. the effect of the characteristic features has been studied. as well as the resulting lift and drag histories were studied. the mean drag at comparable lift is increased by 24. The results from the present study show that special features of insect flight have an appreciable effect on the accuracy of performance models of insect flight. Due to the high acceleration during stroke reversal of the ‘sawtooth’ shaped amplitude. The actual vortex dynamics. However.5 Conclusions 119 coefficient was matched. This led to an increase in mean drag during each half-stroke. an overall comparison of the complete kinematic models was given. is not affecting the mean lift to a large extent. The second model simplification used by the Robofly.5. which is found to have only a marginal effect on the mean lift and mean drag. Therefore. but the numerical flow simulations provide a more complete quantitative analysis of the flow behaviour. The results showed that the ‘sawtooth’ amplitude used in the Robofly model has a small effect on the mean lift but the mean drag is affected significantly. The mean aerodynamic drag at equal lift of the fruit fly models is about 49% lower compared to the Robofly model and about 29% lower with respect to the harmonic model.

CHAPTER

6

Vortex wake interactions of a two-dimensional forward flapping foil
AIAA paper 2009-791.

A two-dimensional numerical investigation is performed to study the vortical flow around a flapping foil that models an animal wing, fin, or tail in forward motion. The vortex dynamics and performance are studied to determine the influence of foil kinematics. The baseline kinematic model is prescribed by harmonic functions which can be characterised by four variables, the dimensionless wavelength, the dimensionless flapping amplitude, the amplitude of geometric angle of attack, and the stroke plane angle. The foil motion kinematics has a strong influence on the vortex dynamics, in particular on the vortex-wake pattern behind the foil which can be either periodic or aperiodic. Both symmetric and asymmetric solutions are found. Evidence was found that the attachment of a leading-edge vortex (LEV) is not significantly advantageous for the force enhancement during the full stroke. Plots of efficiency versus the independent variable show that, for symmetric kinematics, the largest efficiency is achieved at an intermediate value of each variable within the parameter range considered, where periodic flow occurs.

122

Vortex wake interactions of a two-dimensional flapping foil

6.1

Introduction

The flow around a flapping wing, fin, or tail is highly unsteady and governed by the dynamics of the generated vortices (Weish-Fogh & Jensen, 1956, Dickinson et al., 2000). An experimental study (Lentink et al., 2008) showed that these shed vortices interact with each other and organise themselves, similarly to an oscillating cylinder as described by Williamson & Roshko (1988), into specific wake patterns depending on the foil kinematics. The wake pattern can be either periodic or aperiodic and directly determines the periodicity of the aerodynamic forces acting on the foil. Periodic flow is the result of a match between the driving frequency and the natural shedding frequency which is referred to synchronisation of the flow (Williamson & Roshko, 1988, Lentink & Gerritsma, 2003). The wake will be aperiodic if synchronisation of the vortex-wake does not occur. The synchronisation band organisation for the flapping foil may be very complex due to the large extent and high dimension of the parametric space. In contrast to the cylinder, vortex shedding from a flapping foil displays a variation of the natural shedding frequency as a function of angle of attack (Katz, 1981, Dickinson & G¨tz, 1993). o A numerical study by Lentink & Gerritsma (2003) showed that symmetric foil kinematics can result in either a symmetric or an asymmetric wake. In the case of an asymmetric wake, the initial condition determines the orientation of the wake and hence the orientation of the time-averaged lift over a complete flapping period. Several studies have shown that the wing benefits from the attachment of the LEV because of the low pressure core of the LEV acting on the wing during the full stroke (Lentink & Dickinson, 2009b,a, Ellington et al., 1996, Dickinson, 1994). However, the propulsive performance of plunging foil kinematics without a pitching motion is poor (Lentink & Gerritsma, 2003). Therefore, it was suggested that foil rotation is an important source for production of thrust to increase the aerodynamic performance. In the present research, we studied the vortex structure generated in the wake of an ellipsoid foil undergoing flapping motion, plunging and pitching, at a Reynolds number of Re = 150 which corresponds to the flight of a small insect, e.g. a fruit fly. Here only the near wake of the foil is studied. The motivation for this is that performance of a flapping foil is influenced mainly by near wake dynamics. The objective of the present study is to investigate the influence of different foil kinematics on the vortex-wake structure, force coefficients, and performance.

6.2

Flapping foil parametrisation

The baseline kinematic model is based on harmonic motion, such as used by Wang (2000b,a). The flow around a flapping foil and the foil kinematics can be characterised by dimensionless parameters. The method used to make the governing equations dimensionless is the same as used by Lentink & Gerritsma (2003) and Lentink et al. (2008). This approach enables us to perform a systematic inves-

6.2 Flapping foil parametrisation

123

tigation of the influence of different foil kinematic parameters on the vortex-wake pattern. The main important parameters are the frequency f [1/s] of both the translation and rotation, which are coupled with a phase shift of 90◦ , the amplitude of translation A [m], the amplitude of the sinusoidal foil rotation Aα (◦ ), the forward velocity of the foil U∞ [m/s], the chord length of the foil c [m] and the stroke plane angle β (◦ ). The definition of the dimensionless parameters is schematically illustrated in figure 6.1 and described in more detail below. The dimensionless wavelength λ∗ represents the number of chord lengths travelled during one flapping period: U∞ . λ∗ = fc The dimensionless amplitude A∗ represents the ratio of amplitude of the foil translation and the chord length of the foil: A∗ = A . c

The Strouhal number St is based on the stroke amplitude A, and is hence equal to the ratio of the dimensionless amplitude A∗ and the dimensionless wavelength λ∗ : A∗ fA = ∗ . St = U∞ λ It corresponds to the maximum induced angle of attack Aαind at mid-stroke due to the translation of the flapping motion of the foil. The mean velocity U [m/s] is obtained by averaging the velocity components over one flapping period: U= 1 T
T 0

(U∞ + UflapX )2 + (UflapY )2 dt .

Here T [s] is the period, UflapX [m/s], and UflapY [m/s] the velocity of the foil kinematics in X and Y directions respectively. The time-averaged Reynolds number Re becomes Uc , Re = ν where ν [m2 /s] is the kinematic viscosity and changed for every computation to match the Reynolds number. For the basic model the dimensionless wavelength λ∗ , the dimensionless amplitude A∗ , the amplitude of geometric angle of attack Aα , and the stroke plane angle β are chosen as independent variables. The Reynolds number is kept constant at Re = 150. In order to study the influence of the kinematics, each parameter is varied from the baseline model, defined by λ∗ = 6.8, A∗ = 1.5, Aα = 15◦ , and β = 90◦ . The dimensionless wavelength was varied from λ∗ = 24, 20, 12, 10, 7.9, 6.8, 6.3, 6.0, 5.7, 5.3, 4.5, 4.0 to 3.0. The dimensionless amplitude is varied within the range of 0.5 ≤ A∗ ≤ 3.0 with a 0.5 increment. The amplitude of angle of attack varies from 0◦ ≤ Aα ≤ 45◦ with a 15◦ increment. The influence of the stroke plane angle is

the angle amplitude Aα . see figure 6. 2 Using q. qc CM = M . . In the cases that the stroke plane angle differs from 90◦ . Cl = CL · sinβ + CD · cosβ . qc2 Projecting the lift coefficient CL and drag coefficient CD onto the y. (b) The two-dimensional relation between two inertial coordinate systems. The xy-plane is tilted at an angle β.1.and x-axes we obtain the foil lift coefficient Cl and the foil drag coefficient Cd respectively. The downstroke phase is filled by dark blue and the upstroke by light blue. the resulting flapping motion is asymmetric.3 Force coefficients and performance In this study two inertial coordinate systems are used.124 Vortex wake interactions of a two-dimensional flapping foil Y β 90 − β ◦ Aα 2A∗ · sin(β) U∞ 2A (tan−1 )2Stsin(β) y λ∗ x β X (a) (b) Figure 6. investigated for two different angle amplitudes Aα = 15◦ and 45◦ in combination with 15◦ ≤ β ≤ 90◦ with a 15◦ increment. (a) Illustration of the foil parameters in forward flight: the dimensionless wavelength λ∗ . The XY plane has the X-axis in the direction of the free-stream velocity and the Y -axis in vertical direction. The drag force D is the component of the total aerodynamic force parallel to the forward velocity of the foil and is positive when directed in the positive Xdirection. 6. the stroke plane angle.1 Schematic illustration of the foil kinematics in forward flight. In this frame of reference the observer is fixed relative to the undisturbed air. the force and moment coefficients are scaled using the average dynamic pressure q [N/m2 ]: 1 2 ρU . In the present study. and the stroke plane angle β. The flight direction is from right to left. the force and moment coefficients are defined as: q= CD = D . the dimensionless amplitude ∗ A . The lift force L is the component of the total aerodynamic force perpendicular to the forward velocity of the foil and is positive when it is in the positive Y -direction. the Strouhal number St. qc CL = L .

d · Uflap dt − 1 T 0 (6. and (6..2. The comparative assessment of the aerodynamic performance of the different kinematic models is based on the mechanical efficiency of the foil motion. 2008. The result of this conformal mapping can be seen in figure 6. (6. d [N] the foil drag. The inner O-type mesh of 50000 cells is surrounded by a ring of tetrahedral cells. T [s] the flapping period. Wang. U∞ [m/s] the free-stream velocity. (6. 6. θ) is concentrated around the leading and trailing edges.1). The constant µ and θ correspond to confocal ellipses and hyperbolas respectively.3) Here −D [N] represents thrust. Uflap [m/s] the translational velocity of the foil in the stroke plane. more information on this mapping can be found in (Bos et al.1) η= Preq where 1 T T 0 Peff = −D · U∞ . whereas the outer mesh is re-meshed every time-step. The inner mesh is able to move. Those two-dimensional simulations are performed on mesh resolutions of about 50000 cells. In order to obtain a good quality mesh.2) T Preq = − M · ωflap dt .4 Numerical model Cd = −CL · cosβ + CD · sinβ . The radius of the inner computational domain is chosen to be 25 chord lengths. a 2% thick ellipsoid shape with unit chord length represents the foil. A uniform grid in (µ.6. This type of grid is suitable for the problem since the vorticity is strongest near the edge of the foil. or tail and is relevant to the required power of locomotion. 125 Note that a negative drag coefficient CD means thrust which is necessary in forward locomotion whereas the foil drag coefficient Cd indicates the fluid force that the animal must overcome for translational motion of its wing. These elliptical coordinates can be transformed to Cartesian coordinates via a conformal mapping: x + iy = cosh(µ + iθ) . The efficiency η [%] is the ratio between the effective propulsive power Peff [Nm/s] and the required power Preq [Nm/s] which are given in (6. elliptical coordinates (µ. M [N·m] the moment about the centre of rotation and ωflap [rad/s] represents rotational velocity of the foil. such that the influence of far field boundary condition can be neglected.3) respectively: Peff · 100% . Note that we have neglected inertial cost of mechanical work done by the foil. fin.4 Numerical model In the present thesis. (2008).2). (6. . θ) are used following Bos et al. 2000a).

7 the LEV’s are shed before stroke reversal. Thrusting modes are found for 12 ≥ λ∗ because of generated LEV’s which pull the foil toward in forward direction. thrusting mode. The resulting wake patterns have been classified using a symbolic code of letters and numbers developed by Williamson & Roshko (1988) that describes the combination of pairs (P) and single (S) vortices shed during each flapping cycle. and 6.1 by decreasing dimensionless wavelength which is equivalent to an increase in flapping frequency at a constant flight velocity. (a) The O-type body conformal mesh with a grid size of 50000 cells is moving within a ring of tetrahedral cells. 6. 6. For dimensionless wavelengths 12 ≥ λ∗ ≥ 5. 6. 6.5.e.2 The body conformal moving mesh around a 2% ellipsoid foil.2. The numerical results are provided in table 6.1 Influence of dimensionless wavelength The wake pattern and vortex behaviour are studied as a function of the dimensionless wavelength in the range of 24 ≥ λ∗ ≥ 3. The moment when a LEV is shed from the wing is defined as the moment when its core passes the trailing edge. i. (b) The close-up of the mesh at the foil surface shows that the grid is concentrated around the leading and trailing edge. Our results are similar to the experimental results found by Lentink et al. The averaged aerodynamic force coefficients in table 6.4 are obtained using three simulation periods.5 Results and discussion The simulations start with the fluid at rest in which the initial velocity vector is zero. Thus the lift and drag are a function of the effective angle of attack Aαeff which leads to positive drag.1. The amount of LEV’s . At high dimensionless wavelengths λ∗ = 24 and 20 the numerical results give no strong vortices shedding from the foil in relation to the foil oscillation.126 Vortex wake interactions of a two-dimensional flapping foil 30 20 10 Y 0 -10 -20 -30 0 -3 -20 -10 X 0 10 20 30 (a) (b) Figure 6. Note that the efficiency is only calculated when the drag is negative. (2008) using a soap-film tunnel.3.

772 -1.5.440 -1.580 1.577 1.002 0. The asymmetry in the lift coefficient is a result of the asymmetry in wake pattern.0 127 pattern no vortices no vortices 2P+2S 2P+2S 2P+2S 2P+2S 2P+2S 2P+S 2P+S 2P P+S P+S Aperiodic CL 0.478 -0.617 η 4.0 and 1.004 0. and TEV’s shed from the foil is decreasing with the dimensionless wavelength because the vortices have less time to develop and shed.320 -0.186 -0. 6.8 6.218 0.034 CLupstroke -0.2 Influence of dimensionless amplitude Six dimensionless amplitudes are chosen to investigate the influence of this parameter.934 -3. Therefore the foil does not fully benefit from the attachment of LEV’s.0 7.633 1. shear layers from the foil organise themselves into a 2S pattern. the LEV’s stay attached to the foil relatively longer at lower dimensionless wavelengths.502 -0.66 13.9 6.095 1. The LEV’s increase in size and strength due to increasing effective angle of attack.47 13. Therefore.494 CLdownstroke 0.43 9.1 Influence of dimensional wavelength.927 -1.005 -0.418 -0. Nevertheless.624 -1. see figure 6.71 10.719 -1. For medium dimensionless amplitude A∗ = 1.7 5.16 11.5 Results and discussion λ∗ 24.712 -1.925 1. so that the forces of this case are varying with relative small changes from period to period. As a result of this.5 the effective angle of attack is high enough to .2 the numerical results are given for the six cases.001 -0.495 1.021 CD 0.0 5.35 11.6. no vortices are formed on the foil due to the low effective angle of attack.009 0.937 2. and β = 90◦ .071 -0.70 12.833 -1.4.48 13. At low dimensionless amplitude A∗ = 0.481 0.028 -0.287 -0.568 -0.704 1. In table 6. The numerical results are shown for 13 different values for the dimensionless wavelength. see figure 6. This means that the foil cannot produce lift enhancement just before the end of each half-stroke whether the LEV is shed before or after stroke reversal.0 20.002 -0.0 3. Aα = 15◦ .092 -0.254 2.252 -0.3.302 -0.786 -1.241 0.3 4.342 -0. A∗ = 1.003 0.185 -0.06 Table 6.00 12. As a result.0 10.5.3 6.05 13.100 -1. the foil produces higher lift and thrust during each half-stroke for decreasing dimensionless wavelength. the force distributions have a sinusoidal shape because the foil cannot produce force enhancement.0 12.004 -0.324 0.103 0.5 4.874 2. It is observed for cases when vortices are formed on the foil that the lift changes its direction before stroke reversal. A further decrease in dimensionless wavelength results in stronger vortex-wake interactions which lead to an aperiodic wake at λ∗ = 3.289 -0.5.

70 13.064 CL downstroke 0.252 -0.830 CL upstroke -0. λ∗ = 4. Aαeff = 49◦ (f ) Aperiodic.198 -0.373 -1. λ∗ = 6.151 1.5 3. Aαeff = 39◦ (c) 2P+S.71 7. λ∗ = 3. Aαeff = 43◦ (d) 2P.0.561 1.8. Aα = 15◦ . λ∗ = 24.5. Vorticity contours of various wake patterns for decreasing dimensionless wavelength λ∗ .147 -1.638 -1.0 2.8.5 1. All images are taken at t = 0.0 pattern 2S 2P+2S 2P+2S Aperiodic Aperiodic Aperiodic CL 0.13 Table 6. The flow is from left to right.302 η 12. Aαeff = 46◦ (e) P+S.0 1. Aαeff = 53◦ Figure 6. and β = 90◦ .086 -0. and β = 90◦ . λ∗ = 6.005 -0.376 1.3. Aα = 15◦ . λ∗ = 5.47 9.5.770 -1. Aαeff = 7◦ (b) 2P+2S.217 -0.002 0.128 Vortex wake interactions of a two-dimensional flapping foil (a) No vortices.274 -0.35T . .2 Numerical results of the kinematics for six different dimensionless amplitudes.3 Vorticity contours for decreasing wavelength. A∗ 0. A∗ = 1. λ∗ = 6.038 -0.002 0.624 -1.5 2.109 -0.598 1.94 8.959 CD 0.633 1.

58 Table 6.282 1.004 CL downstroke 2.47 28.5.5 0 -0.3 Numerical results of the kinematics for four different angle amplitudes.3 shows numerical results for different angle of attack amplitudes.3. Aα 0 15 30 45 pattern 2P+S 2P+2S 2P 2P CL 0. P+S = 3.4 Force coefficients to study the influence of wavelength.2 Drag coefficient [-] k k k k k k = 24. Aperiodic 0.348 -0.0. A∗ = 1. Aα = 0◦ . A peak performance of 28.5. the reverse trend is found for thrust. Strong foil-vortex interactions lead to an aperiodic wake pattern causing aperiodic force coefficients. vortices with a diameter larger than chord length are formed.0. the effective angle of attack is lower. results in an asymmetric 2P+S pattern.850 -1. The foil rotation leads to thrust generation due to the frontal surface area for the pressure difference acting toward in forward direction (Lentink & Gerritsma.050 -0.624 -1. and ◦ β = 90 . However. This results in decreasing lift in each half-stroke for increasing angle amplitude.094 -0.2 0.18% is obtained which is considerably larger com- .6 0.3 Influence of angle of attack amplitude Table 6.8. 2P+2S = 6.5. 2P+2S = 6. and β = 90◦ .579 CD 0.5 -1 -1.8 1 -2 0 0. Aperiodic 1. No thrust is generated for this setting.18 14.0.5.0.005 0.001 0.5 Results and discussion 129 8 6 Lift coefficient [-] 4 2 0 -2 -4 -6 -8 0 0.588 CL upstroke -1. form a LEV which leads to lift enhancement and thrust.252 -0. The plunging kinematic model.4 t/T [-] 0.5 k k k k k k = 24. non-zero angle of attack amplitude.5.3. Once the foil is allowed to rotate. 2003).633 1. For high dimensionless amplitude A∗ ≥ 2. Lift (a) and drag (b) histories of different wake patterns for six dimensionless wavelengths. P+S = 3. λ∗ = 6. A∗ = 1.089 η 13.8. 2P+S = 5. Some of these vortices are hit by the foil during stroke reversal.6 t/T [-] 0.4 0.6. 2P+S = 5. No vortices = 6.097 0.8 1 (a) (b) Figure 6.8. 2P = 4. 2P = 4.0. Aα = 15◦ .216 0. which gives positive mean lift over a period. No vortices = 6. The LEV in the upstroke is weaker than those generated in the downstroke.5 1 0. 6.

240 CLupstroke -1. At Aα = 45◦ . At this low effective angle of attack the foil produces lower lift and thrust.252 -0. The orientation of the mean lift depends on initial .264 0. due to the foil rotation.084 0. shear layers which are generated by the foil.053 3. This is because during the downstroke of asymmetric kinematics the foil undergoes a greater relative velocity.344 1.082 2.600 0.36 - Table 6.056 -0. Therefore.004 0. Numerical results of the kinematics for six different stroke plane angles in combination with two different angle amplitudes.262 1.4.296 CD -0.988 1.518 1.632 1.045 η 13.242 0.063 0. This is because the thrust component of the resulting aerodynamic force is high compared to the normal component. non-zero average lift exists only as a result of an asymmetry in wake pattern.624 2.120 2.5.626 1.768 CLdownstroke 1.844 3. λ∗ = 6.4 Influence of the stroke plane angle. 6.892 2.156 0.182 0.500 1.5 pared to other cases.8 and A∗ = 1.303 0. From the baseline kinematics the stroke plane angle is tilted backward by 15◦ . The difference in relative velocity between up and downstroke also affects the drag contribution in a similar way.160 0.183 3.58 0.619 0.087 -0.002 0.868 -0.062 0.47 5.089 -0.282 1. Also the negative lift in the upstroke is decreasing in magnitude. A similar trend is found for both angle amplitudes that the lift coefficient is increasing for decreasing stroke plane angle during the downstroke until the flow becomes aperiodic.103 0.464 -0. 6.579 -0.4 Influence of stroke plane angle The stroke plane angle causes an asymmetry in the kinematics. However.402 2. no significant vortices are formed on the foil because the effective angle of attack is low Aαeff = 9◦ at mid-stroke.624 -1.42 14.633 2.005 0.5 Discussion In the symmetric kinematics.5.130 Vortex wake interactions of a two-dimensional flapping foil β 90 75 60 45 30 15 90 75 60 45 30 15 Aα 15 15 15 15 15 15 45 45 45 45 45 45 pattern 2P+2S 2P+S 2P Aperiodic Aperiodic Aperiodic 2P 2P+S 3P+S P+3S P+2S Aperiodic CL 0.521 2.124 3. form themselves into a 2P pattern. Here the results for the influence of the stroke plane angle with two different angle amplitudes Aα = 15◦ and Aα = 45◦ are shown in table 6.588 1.092 1. the averaged lift is mainly generated during the downstroke.

∗ λ 2π (6. Figure 6. . The peak efficiency of 28.4 0.6 0.4). the ratio of A∗ and λ∗ as: St = A∗ 1 > · tan(Aαgeo + Aαstall ) .6 shows plots of efficiency versus the independent motion parameters.2 -0.5 1 P+S 2P+2S 2P 2P+S No vortices -0. There is an important limitation in forward flapping locomotion. conditions. it is thought that the results could also shed light on the Micro Air Vehicle (MAV) design. Besides.6. where the time-averaged lift coefficient is plotted against wake pattern.5 Results and discussion 131 0. The grey bands indicate a symmetric wake pattern in which the nearly zero mean lift is obtained. To begin with. The vortex synchronisation diagram for all models is shown in figure 6. (b) Vortex-wake synchronisation A∗ − λ∗ diagram.5(a).5(b).2 4 3. the (absolute) effective angle of attack should be high enough to form a LEV in order to generate force enhancement.5 Aperiodic P+S 2P 2P+S 2P+2SNo vortices 0 0 5 10 λ∗ 15 20 25 (a) Mean lift coefficient of symmetric kinematics.5 3 2. This approximately restricts the values of St. In symmetric kinematics there is an optimal value for each variable. see figure 6. (a) The mean lift coefficient over a complete period of symmetric kinematics as a function of wake pattern. The results are summarised in figure 6.4) which is illustrated by the dashed line. The angle amplitude and stroke plane angle are kept constant at Aα = 15◦ and β = 90◦ . i. the wake and consequently the forces become aperiodic which will influence the stability and controllability of the MAV’s. We have added the operating conditions of insects belonging to the order Diptera.6 0.6(a).e.5 Aperiodic Present study Diptera CL 0 A∗ 2 1. (b) Vortex-wake synchronisation A∗ − λ∗ diagram of our sinusoidal flapping wing. Figure 6.18% could confirm that the wing rotation plays an important role in the unsteady aerodynamic force production.5 Influence of the kinematics on the vortex wake pattern and force generation. (c). The dash line represents our theoretical estimate of the boundary governed by equation (6.4 -0. When the wing operates outside the synchronisation region. (b) and.

(b) dimensionless amplitude.5 4 (a) 30 (b) 15 Aα = 15◦ Aα = 45◦ 2P+S 2P 2P+2S 20 10 η 2P+2S 10 η 5 2P 0 0 10 20 30 40 50 0 60 Aα 65 70 75 80 β 85 90 95 100 (c) (d) Figure 6. The results are in satisfactory agreement with the comparable experiments. 6. and stroke plane angles at the Reynolds number of 150. Although such an attempt at classifying vortex patterns can lead to confusion due to the shedding. (c) angle amplitude and (d) stroke plane angle.5 Aperiodic 3 P+S 3.5 1 1. . or merging of tiny vortices. Both plunging and rotating motions are prescribed by simple harmonic functions which are useful for exploring the parametric space despite the model simplicity. dimensionless amplitudes. it is suitable for straightening out the shedding vortices in our simulations.132 Vortex wake interactions of a two-dimensional flapping foil 15 15 10 10 η Aperiodic η 2P+2S 0 2 4 6 λ 8 ∗ 10 12 14 0 2P+2S 5 5 2P 2P+S 0 0.6 Influence of flapping kinematics on the efficiency.6 Conclusions A numerical model for two-dimensional flow was used to investigate the effect of foil kinematics on the vortex dynamics around an ellipsoid foil subjected to prescribed flapping motion over a range of dimensionless wavelengths. angle of attack amplitudes. The resulting wake patterns behind the foil are categorised using the concept of Williamson & Roshko (1988). tearing. (a) dimensionless wavelength.5 A∗ 2 2. Efficiency as a function of the independence variables.

However. which implies that aerodynamics might select a range of preferable operating conditions. the synchronisation band was not investigated completely. .6. Since the computational costs are high and the parameters cannot be varied continuously. the present study is beneficial for understanding the influence of wing kinematics on the performance characteristics.6 Conclusions 133 Optimal propulsion using flapping foil exists for each variable. The conditions that give optimal propulsion lie in the synchronisation region in which the flow is periodic.

.

Finally.1 Introduction To understand the aerodynamic performance of flapping wings at low Reynolds numbers. the Reynolds number and the stroke kinematics. A varying Rossby number represents a variation in the radius of the stroke path and thus the magnitude of the angular acceleration. namely the angle of attack. (January 2010). Fluid Mech. Results are obtained by performing numerical simulations of the three-dimensional flow around a flapping wing. have been studied. the three-dimensional flow is compared with the two-dimensional studies performed on flapping forward flight. relevant for three-dimensional flapping wing aerodynamics. 7. Secondly. the Rossby number. the Rossby number is varied at different Reynolds numbers. a ‘figure-of-O’ or a ‘figure-of-U’ pattern. the three-dimensional wing kinematics is varied by changing the shape in angle of attack and by applying a deviation. Different aspects. relevant for insect flight. it is important to obtain insight into the vortex .CHAPTER 7 Vortical structures in three-dimensional flapping flight Submitted to J. First. which may result in a ‘figure-of-eight’. Thirdly. the flow around a dynamically scaled model wing is solved for different angles of attack in order to study the force development and vortex dynamics at small and large mid-stroke angles of attack. A parameter study is performed to investigate the performance in flapping flight and to get insight into the vortex dynamics and force generation.

Dickinson et al. By varying the location of the centre of rotation. Bos et al.. Ellington et al. However. 2002) or flapping (Dickinson et al. Dickinson. (Lentink. 1993. 1997). 2002. 2004.136 Vortical structures in three-dimensional flapping flight dynamics and its influence on force development. using plates at different spanwise locations to block the spanwise flow. Lehmann.. which could suggest that a spanwise flow may play an important role concerning the LEV stability in insect flight (Ellington et al. In addition. Shyy et al. 2008) showed interesting results concerning the stability of the three-dimensional leading-edge vortex depending on the Rossby number (equiv- . In order to gain insight into the three-dimensional flow field induced by the flapping wings. Lentink & Dickinson. 1994. which rotates around its base. 2002. Wang. Lentink & Dickinson. Recent two-dimensional simulations (Bos et al. Lentink & Dickinson (2009b) discussed that the stability of the LEV growth specifically might be increased by the spanwise flow through the LEV core.a. 1996. a three-dimensional wing was modelled which was able to flap around a base of which the location can be varied. 2001. Van Den Berg & Ellington. driven by either the dynamic pressure gradient on the wing’s surface.. the influence of the revolving strength (Rossby number) and the effect of the tip vortices can be studied. several two-dimensional studies have been performed (Dickinson & G¨tz. 2008).. In order to investigate the vortex dynamics and the stability of the leading-edge vortex in particular. The most important feature in flapping wing aerodynamics has been established to be the generation of a stable leading-edge vortex (LEV) on top of the wing. Additionally. 2004) model wing.. The spiral leading-edge vortex generated by a translating swept or delta wing is stabilised by the induced spanwise flow. while a three-dimensional LEV remains stably attached to a three-dimensional revolving (Usherwood & Ellington. 1996). 1996. but they did not completely explain the LEV stability in their experiments.. o It was shown that the leading-edge vortex generated by a two-dimensional moving foil is shed after several travelled chord lengths.. the LEV stability may be strengthened by a reduction of the effective angle of attack as a result of the tip vortex generation (Birch & Dickinson. Van Den Berg & Ellington. Birch & Dickinson (2001) showed no significant effect of the spanwise flow on the LEV strength and stability. 2004.. 1997). 1999. Previously conducted research addressed a possible analogy between the LEV on flapping wings and the LEV generated by swept and delta wings (Ellington et al. 2008) suggested that the wing kinematics may also have a large influence on the flapping performance in three-dimensional hovering. Additionally. an accurate simulation method is developed to perform a CFD simulation of a three-dimensional flapping wing.. which is responsible for the unexpectedly large force augmentation in hovering insect flight (Maxworthy. 1999. 1996. 2009b). the kinematics is varied from simple harmonics by adding a deviation and ‘trapezoidal’ shaped angle of attack. Birch et al. 2008b). Bos et al... 1979. 2009b. Birch et al. Those results indicate that three-dimensional flow effects are essential for the LEV stability. Srygley & Thomas. Based on the discussion about LEV stabilisation due to wing revolving (Usherwood & Ellington. 2005. the centrifugal acceleration of the boundary layer or the induced velocity field of the spiral vortex lines (Ellington et al.. 2010b).

kinematic models and the simulation strategy.1 Illustration of the wing motion and force definitions. Different vortex identification methods are described in 7.4. φ(t) corresponds to the stroke variation. while the conclusions are summarised in 7.2. Different deviation patterns are investigated. this is the subject of section 7. the Rossby number is systematically varied for different Reynolds numbers and midstroke angles of attack. a selection of geometric and kinematic parameters is made to systemically investigate the flow phenomena of our interest. alent to the stroke path curvature) and the Reynolds number. the results are discussed in 7. which is still representative for true insect flight. α(t) to the geometrical angle of attack and θ(t) to the deviation from the horizontal stroke plane.2. In order to show that the CFD method is accurate and efficient.6. Therefore. The vortical flow needs to be visualised in such a way that the resulting vortices are clearly visible. . 2008).3 briefly discusses the validation and verification of the flow solver. of which the details are described in chapter 2. following the shape of ‘figure-of-O’. The three-dimensional flapping wing is modelled in order to provide a framework for comparison. 7.7.1.5 and 7. ‘figure-of-U’ and ‘figure-of-eight’.. Additionally.7. the flow is solved using Computational Fluid Dynamics (CFD). section 7.2. In view of limiting computing resources.2 Three-dimensional flapping wing simulations 137 FY Fnormal α(t) Rtip Fspan Stroke plane O Rroot FX θ(t) φ(t) start downstroke Fdrag start upstroke mid-stroke FZ Figure 7. Furthermore.2 Three-dimensional flapping wing simulations In order to study the vortex dynamics and stability of the leading-edge vortex. In agreement with (Bos et al.2. the simulation strategy is discussed in section 7. The flapping wing modelling is described in section 7. Illustration of the wing motion and force definitions. the kinematic model is extended with a ‘trapezoidal’ shaped angle of attack and a non-zero deviation is applied. which also addresses wing geometry.

2006)... respectively. a ‘sawtooth’ shaped flapping angle and a ‘trapezoidal’ shaped angle of attack (Sane & Dickinson. Bos et al. Planform selection The single wing span is fixed to bs = 2. 2008) resembles an harmonically varying deviation angle. the three geometric parameters important for the flapping wing simulations are defined: bs . 7. 2002. 2008).. such that significant effect of corrugation on the flow can be neglected. 2002. 2008) showed that the effect of the ‘trapezoidal’ angle of attack was most prominent. i. The average chord length of this ellipsoidal planform is found to be c = S/bs = π/4. did not influence the force development significantly. ‘trapezoidal’ .138 Vortical structures in three-dimensional flapping flight 7. the deviation is varied harmonically with an amplitude between Aθ = 0 and Aθ = 20◦ . angle of attack and stroke kinematics.. ‘figure-of-eight’.5. Reynolds number. where a = 0. 7. Dickson & Dickinson. In addition. 2003. since Lentink & Gerritsma (2003) showed that airfoil shape was of minor influence on the forces and the flow field. present in dragonfly wings. shortly after stroke reversal (Bos et al. while the rotational distance of the Robofly was fixed to 0. Lentink & Dickinson.2.0. Poelma et al. The deviation angle is the angle with respect to the horizontal stroke plane. or ‘figure-of-U’.5 from the wing root. such that the wing tip radius becomes Rtip = 2.7 (Sane & Dickinson. During the stroke. flapping angle φ(t) and the angle of attack α(t). figure 7. From the discussion in section 7. The current research uses a model wing with an ellipsoidal shaped planform with 10% thickness. S and c.2 shows the resulting ‘figure-of-O’.1 it can be concluded that there is need for a detailed three-dimensional numerical study to investigate the effects of the Rossby number.5 and b = 1.. Kinematic models The flapping wing motion is prescribed by three different motion angles defining the deviation angle θ(t).2(a). Realistic fruit fly kinematics (Fry et al. 2008). Since the wing planform is chosen to be ellipsoidal. 7.0 and the chord at mid-span is c = 1.1 Modelling and parameter selection In general. 2004) with an incidental ‘bump’. Depending on the variation of the deviation angle.e. Luo & Sun (2005) showed that the airfoil corrugation.2(b). the wing surface is defined by S = πab. as described in chapter 4 and (Bos et al. So. 2009b. A two-dimensional investigation (Bos et al. A combined deviation and flapping angle variation leads to a wing tip pattern. most investigations concerning flapping wing aerodynamics make use of the modelling convention as previously described by Sane & Dickinson (2002) and Dickson & Dickinson (2004) as applied in the experiments with a dynamically scaled robotic fruit fly wing.0 are the semi-minor and semi-major axes. The length scales of the corrugation are orders of magnitude smaller compared to the length scale of the separated flow region or the leading-edge vortex. The hinge around which the wing is able to flap is fixed to a distance of 0. such that S = π/2..2(c).

θ [◦ ] [-] 80 upstroke 80 60 40 20 0 -20 -40 -60 -80 0 downstroke upstroke downstroke 20 40 60 80 100 120 140 160 180 Flapping angle.2 Three-dimensional flapping wing simulations 139 Deviation angle.25T . The current research varies the amount of the ‘trapezoidal’ shape by varying the speed of rotation just after stroke reversal from Trot = 0. the (geometric) angle of attack is varied from α = 15◦ to α = 90◦ with increments of α = 15◦ . The current study varied the Rossby number from Ro = 3. (a) ‘figure-of-O’. θ [◦ ] [-] 60 40 20 0 -20 -40 -60 -80 0 Deviation angle. φ [◦ ] (a) ‘figure-of-O’ (b) ‘figure-of-eight’ Deviation angle.7. (b) ‘figure-of-eight’.25T corresponds to fully harmonic angle of attack variation. Different wing tip patterns as a result of the variation in deviation with a combined flapping motion. φ [◦ ] upstroke (c) ‘figure-of-U’ Figure 7. 2009a. the Rossby and Reynolds numbers. In addition to the variation of Reynolds number from Re = 100. which is relevant for vortex induced propulsion in nature (Lentink & Dickinson.10T to Trot = 0. 500 and 1000. the ‘trapezoidal’ angle of attack. to a nearly translating wing.2.2 Different wing tip patterns. θ [◦ ] [-] 80 60 40 20 0 -20 -40 -60 -80 0 downstroke 20 40 60 80 100 120 140 160 180 Flapping angle. shape and deviation. such that Trot = 0.b). φ [◦ ] 20 40 60 80 100 120 140 160 180 Flapping angle. (c) ‘figure-of-U’. Ro = 130. Framework for comparison In order to design a frame of comparison it is important to keep the following three parameters fixed: the dimensionless amplitude of the wing’s cross-section at . The flapping angle was chosen to vary harmonically to isolate the effects of the deviation. where T is the flapping period.

Since the wing rotates with a rotating reference frame.5 the resulting amplitude of the cross-section at Rg becomes A∗ g ≈ 2. which is compensated by the flapping angle amplitude in order to keep the average Reynolds number and the displacement at Rg comparable.1) ν where Aφ is the flapping angle amplitude. whereas the angular acceleration number Cang . Force and performance definitions In order to determine the effect of the different motion and geometric parameters on the forces and performance.4) Here. are defined by the wing geometry. proper definitions are necessary.2) (7. f the flapping frequency. . For the baseline case. maximal values.140 Vortical structures in three-dimensional flapping flight the radius of gyration A∗ g .1). Concerning the baseline R case.2. the wing tip radius changes. average velocity U Rg and displacement of the cross-section at the radius of gyration. (7. two different force definitions are possible. the flapping angle amplitude is determined from (7. R 1 and the area swept by the wing Aswept . The result of this scaling is a comparable average Reynolds number. the average Reynolds number. U R and A∗ = Aφ R/c. which is the main subject of the present thesis.58. If the distance of the rotation origin is varied. R π c Rtip = ARs . and the Rossby number Ro. Cang and Ccen . 2008). Using Rg = S 0 r2 c(r)dr. A∗ g = Aφ Rg /c. for every rotation radius. which is of similar order as used in the two-dimensional analysis in (Bos R et al. based on the radius of gyration is defined as: 4Aφ f Rg c ReRg = . 500 and 1000. where Rtip = 2. the radius of gyration is determined from the rotation origin to the tip r = 0 to Rtip . and the Rossby number Ro.5 the radius of gyration becomes Rg = 1. the hovering wing kinematics is substituted into the expressions for the angular and centripetal coefficients.. c Rtip = ARs . c (7. ARs is the single wing aspect ratio. the Reynolds number at the radius of gyration ReRg . The kinematic viscosity is fixed for three selected values. c the average chord length and ν the kinematic viscosity. equation (4.3) (7. In order to investigate the effect of three-dimensional wing kinematics in hovering flight. It remains clear that both the centripetal Ccen . ReRg = 100. occurring at the R wing tip are still varying like ReR . ReRg = 100. with Rtip = 2. Using (4. On the other hand. provided that the wing kinematics and geometry are given.23).6) to find the following: R Cang Ccen = = Ro = 2 Aφ Rg = A∗ g . depends on the wing kinematics. Therefore.

The flow is solved for different wing tip radii. In order to investigate the influence of the wing kinematics. Because the present research concerns hovering flight. 2008) in chapter 4. the Rossby number Ro (due to varying rotation origin). 1994). (7.2. Additionally. This range in stroke path curvature corresponds to a changing Rossby number from Ro = 3. to study its effects on the behaviour of the leading-edge vortex. the lift force is by definition vertical and thus equal to FY .5) and Fdrag = FX · sin(φ) − FZ · cos(φ).1 in relation to the motion angles. CL /CD (see Ruijgrok. the force coefficients.e. the angle of attack amplitude α. including a variation in angle of attack.5 to a nearly translating wing at Rtip = 102. Lentink & Dickinson (2009b) found that most insects and fish operate at a Rossby number close to Ro = 3.0. Re = 500 and Re = 1000. (7. and translating wing. CL /CD and the 3/2 power factor.7.6) The force in spanwise direction is not used throughout our analysis. As discussed previously (Bos et al. derived by a decomposition of FX .2 to 130. the Reynolds number Re and the wing kinematic model. which seems to be a biologically convergent solution for animals moving in fluids. the drag force is opposite to the motion direction. Influence of Reynolds number and angle of attack For two selected Rossby numbers Ro = 3. q = 0. also known as the glide factor. which is shown in figure 7. the mid-stroke angle of attack is varied from α = 15◦ to α = 90◦ with increments of α = 15◦ . under hovering conditions. A systematic overview of all cases is provided here.2 and Ro = 130. CL and CD . i. are obtained by division 2 using the average dynamic pressure. Influence of wing stroke curvature The stroke curvature is varied in order to investigate if there is a possible relation between the angular acceleration. FY and FZ in the rotating reference frame. . centripetal acceleration or the Rossby number and the forces acting on the flapping wing. the three-dimensional lift and drag are given by: Flift = FY . On the other hand. the Reynolds number is varied from Re = 100. Two performance indicators are used. Aθ were varied.2 Three-dimensional flapping wing simulations 141 in the inertial and the rotating reference frame. 7. reflected by Trot and the deviation amplitude.. the shape of the angle of attack variation.5ρUref . equivalent to the Rossby number. The wing tip radius is varied from fully revolving at Rtip = 2. since that force is small compared to the lift and drag. flapping. the lift-to-drag ratio.2 Simulation strategy and test matrix selection The following variables are varied throughout the current research. This provides insight in the force development as a function of angle of attack for a fully revolving. Therefore.

25 recovers a fully harmonic angle of attack variation.1 Simulation matrix: wing stroke curvature origin. i. Trot .1 shows an overview of the variation of the wing stroke curvature.e. the deviation. Simulation matrix to vary the rotation origin of the flapping wing. ‘figure-of-O’. angle of attack and Reynolds number. depending on the deviation frequency. The varying deviation angle amplitude.2 shows different deviation amplitudes Aθ . Note that the geometric angle of attack is given by αgeom = π − α.0. 2 (7. which determines the amount of ‘trapezoidal’ shape. the angle of attack is plotted in figure 7.0 5. as was already illustrated in figure 7. Aα [◦ ] 1 2T Trot ≤ t < 1 T − Trot . 2 Trot + 1 T ≤ t < T − Trot . •: Re = 500. The angle of attack at mid-stroke is varied from 90◦ to 15◦ together with the Reynolds number.3. − Trot ≤ t < Trot + 1 T.2. +: Re = 100. 2 T − Trot ≤ t < T. For different values of Trot . This ‘trapezoidal’ angle of attack variation is defined by   Aα sin(2πf t)      Aα   Aα cos(2πf t) α=     −Aα     −Aα cos(2πf t) 0 ≤ t < Trot . which may cause a ‘figure-of-eight’ pattern and the ‘trapezoidal’ shape (defined by Trot ) of the angle of attack variation. ‘figure-of-eight’ and the ‘figure-of-U’.7) Here Trot is the rotation duration. .0 12. such that Trot = 0. in combination with a varying. angle of attack and Reynolds number. The grid resolution was fixed to 800k and the time-step was chosen corresponding to Comax = 2.0 7. 2 Table 7. it remains interesting to investigate the effect of two kinematic parameters.0 6.0 Rotation origin. ◦: Re = 1000. This matrix is used to study the influence of the stroke curvature on the structure of the leading-edge vortex and corresponding forces. Reynolds number and angle of attack. may cause different wing tip patterns. Table 7.0 + + + + + + + + + + + + 102. Influence of the kinematic modelling Besides the wing stroke curvature. Rtip 4.5 +•◦ +•◦ +•◦ +•◦ +•◦ +•◦ 3.142 Vortical structures in three-dimensional flapping flight 90 75 60 45 30 15 2.0 +•◦ +•◦ +•◦ +•◦ +•◦ +•◦ Table 7.

130 120 110 100 α [-] 90 80 70 60 50 0 0.3 Angle of attack variation with a ‘trapezoidal’ shape.9 1 t/T [-] Trot Trot Trot Trot = = = = 0.6 0.7 0.2 and the average Reynolds number ReRg = 100. The flapping angle amplitude was fixed to Aφ = 63◦ (1. The Reynolds number is fixed to Re = 100.3 Flow solver accuracy In order to test the accuracy of the used flow solver. Table 7.8 0. a verification study is performed using the flow around a three-dimensionally flapping wing. The rotation duration is varied from Trot = 0. The deviation is varied by the deviation amplitude Aθ .20 0.3 gives an overview of the performed simulations by varying .2 0. 45◦ and 60◦ .4 0. In order to show that the numerical solution is grid and time-step independent. the grid resolution to 800k and the time-step was chosen corresponding to Comax = 2. The meshes for these three-dimensional simulations are constructed with GridPro using a structured approach.25 Figure 7. the mid-stroke angle of attack was given by Aα = 45◦ and the wing tip radius corresponds to fully revolving. The kinematics is according to the simple harmonic model.2 Simulation matrix: kinematic modelling.15 0. More detailed information on grid generation can be found in appendix A.0.20 0. by decreasing the maximum Courant number. which is shown to result in maximal lift coefficients. Simulation matrix to vary the rotation duration and the deviation of the flapping stroke. 7. concerning highly unsteady and vortical flows.3 0.15 0.5 0. numerical comparisons are performed. A verification is performed by varying the grid resolution (grid independence study) and the timestep size. the Rossby number was Ro = 3.7.3 Flow solver accuracy 143 0. Grid refinement is uniform and the cells are clustered close to the flapping wing boundary. the amount of this shape is systematically varied by Trot .25 to Trot = 0. To investigate the influence of a ‘trapezoidal’ shape in angle of attack.25 0. Two different angle of attack amplitudes are used.10 in order to get a ‘trapezoidal’ shaped angle of attack.1 rad).10 0 + + + + 5 + Aθ 10 15 + + 20 + Trot + + + + Table 7.1 0.10 0.

2002). but preferably 800k mesh cells are desired.0 and 0. less than 4%. Summarising.0. where the time-averaged lift and drag coefficients are plotted with decreasing time-step. which varies from 2. i.5 Comax + Mesh resolution 200k 400k 800k + + + +• +• + + 1600k + Table 7.5). the smaller the time-step.5.2%. The smaller Comax .0 0.0 to 0.6(c). This may be explained by the fact that the forces are mainly dependent on the near wake. The spatial grid independence study was performed for a maximal Courant number of Comax = 1. •: Re = 1000. Table 7. 1. the value at ∞ is determined by Richardson extrapolation (Ferziger & Peric.0.0 1.6(b) shows a converging time-averaged lift and drag coefficients with increasing grid resolution. the flow is periodic and the force coefficients (lift and drag) appear to be close for the grid resolutions considered. the drag and lift coefficients are plotted in figure 7.5 show the spatial and temporal errors in average lift and drag with the Richardson extrapolated values. In order to assess the accuracy of the flow solver.0 and the temporal convergence for two grid sizes of 400k and 800k cells.4 and 7.0 to 0.4 shows that even the differences in lift and drag for 100k mesh cells and Comax = 1. the error is less than 0. with several chord lengths from the wing.0. on which the forces depend. Again. Comax . the differences are small. table 7. However.144 Vortical structures in three-dimensional flapping flight 100k 2.0 are reasonably small. are sufficiently fine to capture the near wake. even for the largest time-step.0. from 100k to 1600k. all generated grids provide sufficiently accurate force coefficients.3 Simulation matrix: verification. The corresponding limit cycles are shown in figure 7. but to capture the far wake vortex dynamics at least 800k cells are required. with varying mesh resolution (100k − 1600k) and time-step. The grid resolution was varied from 100k to 1600k cells and the Comax from 2.5.e. Two cases are performed for two different Reynolds numbers. all grid resolutions considered. .4 for meshes from 100k to 1600k cells and Comax = 1. corresponding to Comax = 2. The kinematics of the three-dimensional wing is simple harmonics with Aφ = 63◦ . Aα = 45◦ and Aθ = 0◦ . The time-step is reflected through the maximal Courant number. 2008) at least 400k. Temporal convergence is shown in figure 7. Apparently. As can be clearly seen from the figures. in order to visualise the vortices in the far wake (Bos et al.6(a) and 7.6. grid resolution and maximal Courant number. Table 7. Figure 7. +: Re = 100. In order to assess spatial and temporal convergence both time-averaged lift and drag coefficients are plotted with increasing spatial and temporal resolution in figure 7.. see chapter 2. In order to justify the choice for grid and temporal resolution for the threedimensional flapping wing simulations. Verification matrix showing the cases used for verification purposes. 1.5 shows the errors in lift and drag with respect to the Richardson extrapolated values for both 400k and 800k with decreasing time-step (Comax = 2.5.

4 Three-dimensional verification: force coefficients. Furthermore. which requires a complete velocity field. 100k − 1600k.2 0.1 0.66 -2. Lift and drag coefficients for the verification cases with varying grid size.13 Table 7. Aα = 45◦ and Aθ = 0◦ . ranging from 100k to 1600k.4 0. the temporal errors are sufficiently small.49 -0. The time-step is taken such that Comax = 1.5 0. N 100k 200k 400k 800k 1600k ǫlift [%] -3.3 0.9 t/T [-] 1 -3 0 100k 200k 400k 800k 1600k 2 Drag coefficient [-] 1 0 -1 -2 145 100k 200k 400k 800k 1600k 0.5 1 0.4 Error values of lift and drag coefficients for varying grid sizes.7 0. Different wellknown techniques to detect and visualise vortices are based on the velocity gradient tensor. The time-step was determined by a max Courant number of Comax = 1.7. at Comax = 2. a proper vortex identification criterion is essential.5 0. described in this chapter.8 0.23 -0.4 Vortex identification methods for flow visualisation In order to study the vortex dynamics and the stability of the leading-edge vortex in particular. namely the magnitude of vorticity |ω| (Lu & Shen.6 0.52 -1.0.5 Lift coefficient [-] 2 1.8 0. 2008) . The kinematics of the three-dimensional wing is simple harmonics with Aφ = 63◦ .50 -0. a grid resolution of 800k in combination with Comax = 2.27 -0.19 -2.7 0.3 0.0.9 t/T [-] 1 (a) (b) Figure 7.4 0.4 Vortex identification methods for flow visualisation 3 2. 7.12 ǫdrag [%] -3.2 0. Therefore.0 was used for all threedimensional flapping wing simulations. Aα = 45◦ and A θ = 0◦ .32 -1. Two different vortex identification criteria are discussed.0.5 0 0 0.6 0.1 0. Error values of lift and drag coefficients [%] with respect to the extrapolated values for varying grid sizes. The kinematics of the three-dimensional wing is simple harmonics with Aφ = 63◦ .

shear layers.082 0. 400k and 800k. a positive value of Q > 0 is a measure for any excess of rotation rate (in terms of vorticity) with respect to the strain . Aφ = 1. 1988).5 1 0.067 0. 1988) is the second invariant of the local velocity gradient tensor ∇u.. More specifically. where ω = ∇ × u. The Q criterion (Hunt et al.0 0. 2 where the rate of strain tensor S is given by S = 1 (∇u + ∇uT ) and the vorticity 2 1 tensor by Ω = 2 (∇u − ∇uT ).183 0.0 to Comax = 0.146 3 2. The kinematics of the three-dimensional wing is simple harmonics with Aθ = 0 rad. this criterion may lead to undesired contours of e. For Q > 0 the region is identified as a vortex.. corresponding to Comax = 2.5 ǫlift [%] 400k 0. The time-step is taken such that Comax = 1. |ω| is the common vortex visualisation method (Bos et al.043 0.5 Three-dimensional verification: limit cycle. there is a concentration of vorticity.0 1. within this particular region. Shown are the errors for two grid sizes.097 0.1 rad and Aα = 0. however.785 rad. If |ω| reaches a user-defined threshold. Since shear layers and curved streamlines also are a source of vorticity.053 0.011 ǫdrag [%] 800k 0. constructed from the lift and drag coefficients clearly shows periodic behaviour for the verification cases.024 ǫlift [%] 800k 0.5 Error values of lift and drag coefficients for varying temporal resolution. The grid size was fixed to 800k. Comax 2. 2008).5 -3 -2 -1 1 0 Drag coefficient [-] 2 3 Figure 7. that region is identified as a vortex.159 0.g. In two-dimensional flow. and the Q criterion (Hunt et al. Aα = 45◦ and Aθ = 0◦ .087 0. This second invariant of ∇u is written as Q= 1 |Ω|2 − |S|2 . Hence..5 0 Vortical structures in three-dimensional flapping flight -0.5. The limit cycles.5 Lift coefficient [-] 2 1. Especially in three-dimensional flows this may become a difficulty. The first criterion is based on the magnitude of vorticity |ω|.013 ǫdrag [%] 400k 0.022 Table 7.0. The kinematics of the three-dimensional wing is simple harmonics with Aφ = 63◦ . Error values of lift and drag coefficients [%] with respect to the extrapolated values for varying temporal resolutions.

7.4 Vortex identification methods for flow visualisation 1. ρ where ρ is the fluid density and p the pressure.03 2.224 1.. The temporal convergence is illustrated in (c) and (d). The final value. Therefore.97 (a) Average drag coefficient [-] 2.7 shows iso-surfaces around a flapping wing.25T .02 2. corresponding to Comax = 1. The iso-surface is visualised at t = 0. rate.04 2.22 1.015 12 4 8 16 Temporal resolution [-] ∞ Average lift coefficient [-] 400k 800k (b) 1. a pressure minimum does occur.03 147 Average drag coefficient [-] ∞ 2.216 1.218 1.0.19 12 4 16 8 Spatial resolution [-] 1.02 2.05 2.01 2 1. a region where Q > 0 indicates a clear swirling flow (as shown by Chakraborty et al.23 1.214 1.22 1. during the downstroke.2 2.98 12 4 8 16 Spatial resolution [-] ∞ 1. showing the average drag and lift coefficients for decreasing time-step at two specific grid sizes.23 1. 2005). figure 7. By neglecting these unsteady and viscous effects from the governing Navier-Stokes equations the following relation can be obtained for the symmetric tensor Ω2 + S2 : 1 Ω2 + S2 = − ∇(∇p). It must be noted that Jeong & Hussain (1995) found that Q > 0 is not a sufficient condition to have a pressure minimum in the vortex core of that specific region. 400k and 800k.035 2. (a) and (b) are showing the average lift and drag coefficients for increasing spatial resolution and constant time-step.222 1. The wing flaps around a distance of 0.25T the leading-edge vortex is formed on the wing’s . At t = 0.025 2.18 1.226 1.212 1.99 1.228 1.5 from the wing root and the flapping angles are varying harmonically.21 1.24 Average lift coefficient [-] 1.21 400k 800k 12 4 8 16 Temporal resolution [-] ∞ (c) (d) Figure 7. In order to identify which vortex criterion should be used. however. In most cases.6 Spatial and temporal convergence. at ∞. is obtained using Richardson extrapolation.

The leading-edge vortex. The Q criterion shows more detail. Some of the previously shed vortices are still present. In that section. within the range of −1. the vortices from the previous stroke should be visible as well. e. normalised by their maximal values.g. The spiralling leading-edge vortex is visualised using contour plots of the magnitude of vorticity. within the range of −1.1. |ω| and Q. a smooth leading-edge vortex is shown. (a) shows the contour of the vorticity magnitude. in figure 7. In figure 7. The colours represent values of helicity. is identified using a carefully chosen threshold of the vortex identification criteria.0. rolling up into a tip vortex. Since the Q criterion offers sufficient and adequate information about the local flow field. This leads to a thicker iso-surface.7(a) it can be observed that |ω| shows not only the vortical structures. rolling up into a tip vortex. different geometric and kinematic parameters are systematically varied. The arrow shows the flapping direction of a downstroke and the flow is visualised at mid-stroke. but also the shear layers near the wing and between the vortices. In order to investigate the effect of the Rossby number. a rotation dominated region is identified by Q < 0. h = (u · ω)/(|u| |ω|).148 Vortical structures in three-dimensional flapping flight (a) |ω| (b) Q Figure 7.0 value. h > 0. The colours show the helicity which is defined as h = (u · ω)/(|u| |ω|). |ω| = 5.0.0 and Q = 2. upper surface and rolls up into a tip vortex.0. the radius . using the values |ω| = 5.5 Flapping wings at low Reynolds numbers In order to provide insight into the vortex dynamics (for a purely harmonic flapping motion) and its influence on the variation of forces.0 and (b) the Q = 2.7(b). First. 7.5. this criterion is used throughout the remainder of the present research. A positive helicity.7 Comparison of the near wake flow field using different vortex identification criteria. the influence of an increasing mid-stroke angle of attack is briefly discussed on the force development.0 ≤ h ≤ 1.0 ≤ h ≤ 1. means that the direction vector of vorticity (ω = ∇ × u) is aligned with the local flow velocity. for all three-dimensional simulations. the influence of the angle of attack on the forces is briefly discussed in section 7. such that detail of the vortical structures is lost.

. Figure 7. In addition. due to shedding of the leading-edge vortex at low mid-stroke angle of attack.6 deals with forward flapping flight with similar conditions as the two-dimensional simulations performed by Bos et al. this is performed for different Reynolds number as well. The average drag is maximal and lift minimal for α = 90◦ .8 that the maximal force values occur at approximately halfway through the down and upstroke. the average lift-to-drag ratio obtains a maximal value at an angle of attack of α = 30◦ . since these periods of lower lift appear to occur at low angles of attack.5.6 shows the time-averaged lift and drag coefficient for varying angles.4 and 7.5. Note that α = 90◦ implies that the wing keeps a constant upright position during the entire stroke. typical for revolving .2. The Rossby number was increased from Ro = 3.2 to Ro = 130. concerning hovering flight at fruit fly conditions.5.1 The angle of attack in flapping flight Previous studies showed that the angle of attack variation during the stroke influences the forces considerably.5. the radius of curvature is increased to decrease the angular acceleration consequently. corresponding to a flapping wing with small radius of curvature. the kinematic model is varied by considering a ‘trapezoidal’ shape adaptation and the addition of deviation in section 7.6.75T . The drag for α = 15◦ shows two minor peaks within each halfstroke. (2009). t = 0. Thaweewat et al.2 Influence of flapping stroke curvature In order to investigate the forces and development of the leading-edge vortex. (2008). from table 7. where T is the flapping period. The Reynolds number is Re = 100 and Rossby Ro = 3.8 shows a variation of the lift and drag coefficients during a complete flapping cycle. Additionally. respectively. section 7.7. it must be noted that the lift is nearly always non-negative. This was confirmed by a recent two-dimensional investigation (Bos et al. 2008).5 Flapping wings at low Reynolds numbers 149 of curvature is subsequently varied in section 7.5. While the lift is nearly identical for α = 45◦ and α = 60◦ . Furthermore. It is seen that the maximal time-averaged lift coefficient occurs at a mid-stroke angle of attack of α = 45◦ . The influence of the Reynolds number on the forces and leading-edge vortex stability is assessed in 7.2. lift is being generated during the complete stroke. Furthermore.3. 7. The next section will discuss the drops in more detail. which means that during the hovering conditions. It can be observed that the force variation is periodic and smooth. Table 7. independent of Rossby and Reynolds number. The difference between the maximal (α = 45◦ ) and minimal (α = 15◦ ) lift is 57%. It was already mentioned that the angle of attack at mid-stroke is varied from α = 15◦ to α = 90◦ . 7. However.5. In addition to the variety of hovering flight simulations. the drag is significantly lower for α = 45◦ .5. it can be seen in figure 7.25T and t = 0. it seems that the leading-edge vortex only grows significantly at higher (α ≥ 45◦ ) angles of attack.

436 0.977 0.2 and a Reynolds number of Re = 100.034 1. : 60◦ .000 0. The variation is shown for lift (a) and drag (b) coefficients induced by a flapping wing at Ro = 3.7 0.028 1.9 1 t/T [-] (a) (b) Figure 7. •: 30◦ . ▽ : 45◦ . 4 3.333 0.224 0.6 Force coefficients for Re = 100 and Ro = 3.5 0.9 1 t/T [-] α α α α α α = = = = = = 15◦ 30◦ 45◦ 60◦ 75◦ 90◦ 6 Drag coefficient [-] 4 2 0 -2 -4 -6 0 α α α α α α = = = = = = 15◦ 30◦ 45◦ 60◦ 75◦ 90◦ 0.329 2. .5 0 -0. drag CD and lift-to-drag CL /CD ave are shown as a function of the mid-stroke geometrical angle of attack for given Re = 100 and Ro = 3.2 0. △: 75◦ and : 90◦ .2 0.6 0.750 2.1 0.3 0.5 0.6 0.3 0.667 0.4 0.2.5 2 1.127 1.8 0.2. The mid-stroke angle of attack is varied from ♦ : 15◦ .397 Table 7.722 0.339 0.5 Lift coefficient [-] 3 2.957 CL /CD ave 0. so a flapping wing with small stroke curvature.8 Variation of lift and drag coefficients for a flapping wing at Reynolds number of Re = 100.178 0.4 0.150 Vortical structures in three-dimensional flapping flight αgeom 90 75 60 45 30 15 CL 0.5 1 0.7 0.5 0 0.000 0.526 CD downstroke 3.546 3.325 2. Time-averaged lift CL .739 2.963 CD upstroke 3.8 0.1 0.543 3.703 1.

However. The mid-stroke angle of attack was fixed to α = 45◦ and the Reynolds number to Re = 100.e.883 CD upstroke 2. occurs at mid-stroke t=0.058 1. It is clear that both lift and drag are decreasing with increasing Rossby number.2T ) for an angle of attack of α = 15◦ . At Ro = 130 the wing nearly performs a two-dimensional motion leading to a decrease in lift of 24.8 and 7. feeding two wing tip vortices. This is due to the loss in energy by the tip vortices which was also studied by (Blondeaux et al.997 0. Ro = 130. When compared with figure 7..11 shows the iso-surfaces of Q = 1. Figure 7.572 0.3 130 1.2 3.922 CL (baseline) (−3.75T.8 (which applies to the revolving wing Ro = 3.023 0.e.1 6. decreasing curvature of the stroke path. The decrease in drag is small.933 1. the translating wing shows a leading-edge vortex which stays symmetric with respect to the wing centre plane.899 1.904 1.8 5.9%) (−9.9 is that the lift drops significantly (75% at least) during mid-stroke (t = 0.886 1.877 CL /CD ave 0. Figure 7.5 Flapping wings at low Reynolds numbers 151 Ro 3.981 1.2 to Ro = 130. The loss in drag. i. with increasing Rossby number for α = 45◦ . just after stroke reversal and during mid-stroke are of similar magnitude.537 0. An additional observation.892 1.908 1.25T and t=0.4%) (−18.2%) (−8.4%) (−11. 18.903 1.9%) (−24. It is clear that the major loss in lift for high Rossby numbers.552 0.4 7.224 1. The variation of average lift (CL ). Additionally.13 shows the streamlines to illustrate the leading-edge and tip vortices in more detail. Table 7. Figure 7.7% in comparison to the baseline case with Ro = 3. drag (CD ).0 at t = 0. drag and lift-to-drag ratio.7.5%) (−22.7%) Table 7. figure 7.490 (baseline) (−1.7 shows the time-averaged values for lift.2%) (−18.0%) (−13. which is at the midst of the downstroke.12 shows the Q = 1. It can be clearly observed that the fully flapping wing shows a pronounced leading-edge vortex which spirals towards the tip to form a tip vortex.175 1.7%) (−13. Figure 7. which results in the lower .2.028 2. such that the decrease in lift-to-drag ratio is still significant.593 0.2) it is seen that both lift and drag variations are significantly lower when the Rossby number is large.105 1. at the root and the tip.935 1.500 0. lift-to-drag ratio (CL /CD ) are shown for Rossby numbers from Ro = 3.6 8.603 0.943 0.916 1.6%) (−16. respectively.9 15.915 1. while comparing figures 7. 2005a).6%) (−5.983 1.0 iso-surfaces of a translating wing at Ro = 130 for α = 15◦ and α = 45◦ .25T .034 1.9 shows the force histories concerning the nearly translating wing.10 shows the variation of the lift and drag coefficients during the flapping cycle and the effect of Rossby number as it increases.525 0.7%) CD downstroke 2. and translating wings. i.7 Time-averaged force coefficients for varying Rossby numbers at Re = 100.7%.0%) (−17. It is clearly seen that the leadingedge vortex for α = 15◦ is not yet fully developed.

14(a) for different angles of attack and different Reynolds numbers.1 0.8 0.6 0.14(b) shows the power factor CL /CD as a function of lift-to-drag ratio CL /CD . it was shown for both flapping and translating wings at α = 15◦ . Re = 100. The drag increases as well. This leadingedge vortex is larger and more stable at angles of attack larger than about 30◦ . lift. compared to α = 45◦ . 7.5 0. Figure 7. α = 45◦ the difference between flapping (Ro = 3.2) and translating (Ro = 130) in lift coefficient is 32. The angle of attack amplitude is varied from ♦ : 15◦ . 2004).6 0. In addition. as long as the scaling is appropriate. such that the dimensionless amplitude A∗ g . average R Reynolds number ReRg and swept area Aswept are comparable.. it can be stated that a flapping wing motion is of crucial importance for lift generation at a small penalty of drag. It seems that the trend of the force development with the angle of attack is similar for flapping and translating wings. Summarising.14(a) it can be deduced that the overall lift coefficients are significantly higher for the flapping (Ro = 3. The Reynolds number remained fixed at Re = 100. •: 30◦ .5 0. The time-averaged lift and drag coefficients are plotted in figure 7.4 0. respectively. Birch & Dickinson. Re = 500 and Re = 1000. such that the wing approximately translates.2 0. 500 and 1000. that the leading-edge vortex development is not significant to increase the lift.3 0. △: 75◦ and : 90◦ .5. compared to wing translation. a selection of Reynolds numbers is used.2 0.5 0 -0. 33.5 0 Vortical structures in three-dimensional flapping flight Drag coefficient [-] α α α α α α = = = = = = 15◦ 30◦ 45◦ 60◦ 75◦ 90◦ 6 4 2 0 -2 -4 -6 0 α α α α α α = = = = = = 15◦ 30◦ 45◦ 60◦ 75◦ 90◦ 0. At maximal lift. the leading-edge vortex is important for the gain in lift. Lift (a) and drag (b) coefficients induced by a flapping wing at a Rossby number of Ro = 130. Additionally. ▽ : 45◦ . 2003).7 0.9 1 t/T [-] (a) (b) Figure 7.5 1 0.8%.2) compared to the translating (Ro = 130) wing.9 1 t/T [-] 0. From figure 7.3 0. : 60◦ . the results of a variation in Rossby number are shown for Re = 100.8 0. instead the lift decreases. relevant for insect aerodynamics of a fruit fly (Sane & Dickinson. 2001.5 2 1.4 0.3 Influence of Reynolds number In addition to the angle of attack and stroke curvature. 1998) and dragonfly (Isogai et al.9% 3/2 .5 Lift coefficient [-] 3 2.9 Variation of lift and drag coefficients for a translating wing at Ro = 130.1 0. hawkmoth (Liu & Kawachi. At smaller angles of attack.152 4 3.7 0.

9.4.2 (b) Ro = 130 Figure 7.2 0. A time-frame is shown at mid-stroke.2 3.6 0.4 8.25T .9 130 0. Lift (a) and drag (b) coefficients induced by a flapping wing around varying Rossby numbers.5 1 0. •: 3.8. ▽ : 6.3 0.8 6. Ro = 3.2.8 0. The amplitude of the angle of attack variation was fixed such that at mid-stroke α = 45◦ .7 0.5 0.5 0 -0. .5 0.4 0.9 130 4 3 Drag coefficient [-] 2 1 0 -1 -2 -3 -4 0 Ro Ro Ro Ro Ro = = = = = 3.2 0. : 8. Colours indicate helicity.5 Flapping wings at low Reynolds numbers 153 3 2.4 0. t = 0.2 3.9 1 t/T [-] Ro Ro Ro Ro Ro = = = = = 3.4 8.1 0. The average Reynolds number remained fixed at Re = 100.2 and Ro = 130.5 Lift coefficient [-] 2 1.9 1 t/T [-] (a) (b) Figure 7.5 0 0.8 0.8 6.0 are shown for Rossby numbers Ro = 3.2 − 130.1 0.3 0. (a) Ro = 3.6 0.7. △: 130. The Rossby number is varied from ◦ : 3. Isosurfaces of Q = 1.10 Variation of force coefficients for a flapping wing with varying Rossby numbers.11 Vortex visualisation of the near wake for different Rossby numbers. The average Reynolds number was fixed to Re = 100.7 0.

Colours indicate helicity. . A time-frame is shown at mid-stroke.154 Vortical structures in three-dimensional flapping flight (a) α = 45◦ (b) α = 15◦ Figure 7. Streamlines are shown for Rossby numbers Ro = 3.13 Streamline visualisation of the near wake for different Rossby numbers. A time-frame is shown at midstroke.2 and Ro = 130.12 Vortex visualisation of the near wake for a translating wing at low and high angle of attack. The average Reynolds number was fixed to Re = 100. (a) Ro = 3.25T .2 (b) Ro = 130 Figure 7.0 are shown for α = 15◦ and α = 45◦ for a Rossby number of Ro = 130. The average Reynolds number was fixed to Re = 100. t = 0. t = 0. Iso-surfaces of Q = 1.25T . Colours indicate helicity.

14(a) is that at large mid-stroke angles of attack.6 0. As was already discussed.5 4 0 Re=100.6 0. Re = 100. The angle of attack varies from 90◦ to 15◦ .7 0. In both (a) and (b). Ro=130 Re=500. On the other hand. for a translating (Ro = 130) wing.6 1. an average difference in drag of 7. considering a flapping motion (Ro = 3.8 0. 500 and 1000.2 Re=500. flapping is important at lower Reynolds numbers. Ro=3.8 0. Figure 7.2 0 -0.3 0. Looking at figure 7. Ro=3. the structure of the leading-edge vortex strongly depends on the Reynolds number in cases of large angular accelerations. the time-averaged lift and drag coefficients show marginal variations with Reynolds number.7.9 1 Glide ratio [-] Drag coefficient [-] (a) Figure 7. while flapping.2 Re=500.6 0. Ro=3. (b) illustrates the power factor CL /CD as a function of the lift-to-drag ratio CL /CD . and two Rossby numbers. At α = 45◦ the differences in time-averaged drag coefficient are negligible when the variation in Reynolds number is concerned.2 3 3.5 Re=100.2 0. while at α = 75◦ the differences in drag become significant. Another observation from figure 7.2 Re=1000.4 0.5 0.5 0 0. the results are shown for three different Reynolds numbers. and 35.5 Flapping wings at low Reynolds numbers 1. from right to left in (a) and from left to right in (b).4 0. which is beneficial in terms of performance.8% for increasing Reynolds numbers Re = 100.8 0.4 Lift coefficient [-] Power factor [-] 1. The difference in drag is only significant at larger angles of attack. which is visualised by irregularities in the iso-surfaces. Ro=130 Re=1000. Since the difference in lift increases (although slightly) with Re. Ro=130 Re=1000.1 0. However. Ro = 3.15 shows the iso-surface of Q = 1. Ro=3. Ro=130 1 1.14 (b) Force and performance polars. While lift is enhanced significantly. respectively.5 2 2.2 1 0.2 Re=100. Ro=3.14(a).2 and Ro = 130.2) with respect to translating (Ro = 130). the variations in lift and drag with Reynolds number are larger for lower Rossby numbers.2 0. the leading-edge vortex becomes slightly unstable with increasing Reynolds numbers. there is still a large gain in lift-to-drag.2 Re=1000. the streamlines for the corresponding .4 0.0 to visualise the leading-edge vortex on a flapping wing (Ro = 3. (a) shows the force polars as a function of angle 3/2 of attack amplitude. Ro=3. but becomes slightly more important at higher Reynolds number. combined with a small drag penalty. 500 and 1000. In addition.5% is obtained. Ro=130 155 1 0.2 Re=100. So. This may be explained by considering that the leading-edge vortex was found to be more stable on a translating wing. compared to flapping.2) during the downstroke for both Reynolds numbers Re = 100 and Re = 1000. Ro=130 Re=500. the Reynolds number has a larger effect on the lift and drag.

Figure 7. .e. but also to irregularities such that the flow at low Ro is more sensitive to changes in Reynolds number. Iso-surface of Q = 1. although the leading-edge vortex may be less stable.16.2. Additionally. However.2 and translating Ro = 130 flight. this is not the case for a translating wing. for a flapping wing Ro = 3.156 Vortical structures in three-dimensional flapping flight (a) Re = 100 (b) Re = 1000 Figure 7. figure 7. compared to translation. While the leading-edge vortex detaches at a smaller distance from the wing tip. In order to illustrate the irregular motion at larger Reynolds numbers. it seems plausible that the generation of a leading-edge vortex is important for both flapping Ro = 3. with increasing Reynolds numbers.2. Therefore. Colours indicate helicity. Besides the irregularities.2.15 Vortex visualisation of the near wake for different Reynolds numbers. It was noted that the leading-edge vortex may play a more important role for flapping motion. comparison are shown in figure 7. without a significant loss in lift. For a translating wing Ro = 130. the leading-edge vortex may burst as was discussed by Lentink & Dickinson (2009b).15 shows some irregularities of the leading-edge vortex. In case of the flapping wing Ro = 3. on a flapping wing for increasing Reynolds numbers. for a translating wing. for higher Reynolds numbers.0 are shown for two Reynolds numbers. leads to larger lift enhancement. t = 0.18 shows the Q iso-surfaces for different Reynolds numbers Re = 100 and Re = 1000. A time-frame is shown at mid-stroke. these irregularities are less pronounced. A Reynolds number increase. Ro = 3.17 shows the vortical motion just before stroke reversal from down to upstroke. i.25T . Re = 100 and Re = 1000. figure 7. the lift increasing effects of the leading-edge vortex are larger for higher Reynolds numbers. Re = 1000. the leading-edge vortex clearly detaches at a smaller distance from the wing root for Re = 1000.

Re = 100 and Re = 1000. t = 0. .5 Flapping wings at low Reynolds numbers 157 (a) Re = 100 (b) Re = 1000 Figure 7. (a) Re = 100 (b) Re = 1000 Figure 7. A time-frame is shown at the end of the downstroke. Ro = 3. Re = 100 and Re = 1000.2. Ro = 3.2. A time-frame is shown at mid-stroke. Colours indicate helicity.25T . Iso-surface of Q = 1.2. for a flapping wing Ro = 3.17 Vortex visualisation of the near wake for different Reynolds numbers.2. Streamlines are shown for two Reynolds numbers.0 are shown for two Reynolds numbers. for a flapping wing Ro = 3. t = 0.7. Colours indicate helicity.5T .16 Streamline visualisation of the near wake for different Reynolds numbers.

whereas the latter corresponds to a strong rotation during stroke reversal.4 Influence of ‘trapezoidal’ angle of attack In order to study the influence of the ‘trapezoidal’ angle of attack and later compared to an earlier two-dimensional study (Bos et al. 7. generated in both up. the Reynolds number and mid-stroke angle of attack remained fixed. it reaches its mid-stroke angle of attack early in the stroke. for a translating wing Ro = 130. are within 1. Iso-surface of Q = 1.8 shows the time-average lift. accompanied by a decrease in drag.158 Vortical structures in three-dimensional flapping flight (a) Re = 100 (b) Re = 1000 Figure 7.19 for varying rotation duration during a full flapping stroke.5.0%. drag and lift-to-drag values..10. which is varied from 0. At low rotation duration Trot = 0. the angular acceleration just after stroke reversal is large. using a ‘trapezoidal’ shaped angle of attack variation.and downstroke. Ro = 130. Colours indicate helicity. The time variation of both lift and drag is shown in figure 7. so drag generation is symmetric during a complete stroke. 2008). The most important observation is that with decreasing rotation duration. Figure 7. The average drag decreases with a similar amount.8%. Furthermore. compared .2 shows the angle of attack as a function of the rotation duration Trot . increasing angular acceleration during stroke reversal.0 are shown for two Reynolds numbers. it can be seen that the average drag coefficient. Table 7. the shape of the angle of attack is varied.25 to 0. Various experimental and numerical studies have been conducted.1. While varying the rotation duration.e. a gain in average lift is obtained of 10. the ‘trapezoidal’ shape is defined by a rotation timing Trot . i.25T . Since the wing rotates relatively quickly after reversal.9%. leading to a significant increase in lift-to-drag ratio of 21. t = 0. A time-frame is shown at mid-stroke. the first representing a harmonic variation. As previously described. which may be caused by a fast decrease in effective angle of attack.3 in section 7. which leads to an increase in lift.2. Re = 100 and Re = 1000.18 Vortex visualisation of the near wake for different Reynolds numbers.

The explanation for this discrepancy is that in the three-dimensional simulations the leading-edge vortex was found to remain more stable than in twodimensional investigations.5 Lift coefficient [-] 2 1.6 0.5 1 0.493 CL /CD ave 0. 2008) showed opposite effects. At every time instance. but opposite.536 2.9 1 t/T [-] Trot Trot Trot Trot = = = = 0.740 2.25 0.8%) (+10. Lift (b) and drag (c) coefficients induced by a flapping wing.7. The rotation duration is varied from Trot = 0. Therefore. The average Reynolds number was maintained at Re = 100.15 0. using an angle of attack amplitude of Aα = 60◦ . 3 2.8 0.8%) CD downstroke 2. which leads to the integrated gain in lift of 10.2%) (+16. This corresponds to respectively harmonic variation to a ‘trapezoidal’ shape.496 CD upstroke 2.3 0.127 1. The time-averaged lift CL . effect applies to the drag coefficient. A similar.9 1 t/T [-] Trot Trot Trot Trot = = = = 0.19 Force coefficients for a flapping wing with ‘trapezoidal’ angle of attack.2 0.6 0.10 0.25 to Trot = 0.5 Flapping wings at low Reynolds numbers 159 Trot 0. drag CD and lift-to-drag ratio CL /CD are shown for a rotation duration.4 0.2 0.20 0.9% lift-to-drag enhancement.6%) (+21.20 0.215 1.5 0. the lift increase and drag decrease are only marginal. .8 0.9%) Time-averaged force coefficients for varying ‘trapezoidal’ shape of angle of Table 7. Therefore.5%) (+7.500 (baseline) (+9.411 0.15 0.5 0 -0.448 0.10 1.5 0 0.1 0.8 attack.629 2.10.10 0. 21.625 2.25 (a) (b) Figure 7.248 CL (baseline) (+4.15 0.8%.7 0.4 0.539 2. In contrast to this.7 0.25. such that Ro = 3.10.20 0. varying from Trot = 0. twodimensional studies (Bos et al.1 0.479 0.3 0.25 to Trot = 0.25 5 4 Drag coefficient [-] 3 2 1 0 -1 -2 -3 -4 -5 0 0. it can be concluded that the leading-edge vortex stability should be studied with a three-dimensional approach. with the harmonic case Trot = 0.2 and Re = 100. it can be stated that a ‘trapezoidal’ shape in angle of attack considerably increases performance in three-dimensional hovering flapping flight. but overall the results are significant. due to a premature vortex shedding of the leading-edge vortex during the long period of high angle of attack.750 2.. This causes a long period of lift enhancement.177 1.5 0.

160

Vortical structures in three-dimensional flapping flight

7.5.5

Influence of deviation

The deviation amplitude Aθ is used to tilt the two-dimensional airfoil or threedimensional wing out of the horizontal stroke plane. It may be used to generate certain wing tip patterns, e,g. the well-known ‘figure-of-eight’, which is present in realistic fruit fly kinematics (Fry et al., 2003). In (Bos et al., 2008) it was shown that although deviation did not influence the time-averaged values, the instantaneous lift and drag variations are significantly affected. In order to investigate the influence of deviation the amplitude Aθ is varied from 0◦ to 20◦ . In addition, three different stroke patterns are studied, by varying the deviation frequency, resulting in patterns that can be characterised as ‘figure-of-O’, ‘figure-of-U’ and ‘figure-of-eight’, which are shown in figure 7.2. Considering the ‘figure-of-eight’ patterns, two different deviation directions are studied, corresponding to a variation of Aθ from 0◦ to 20◦ and from 0◦ to −20◦ . The reference velocities are adapted correspondingly. First, the deviation amplitude is varied from Aθ = 0◦ to Aθ = 20◦ , according to the ‘figure-of-O’ wing tip pattern. Following figure 7.2, the wing moves consecutive down and up during the downstroke and up and down during the upstroke. Since a downward motion increases the effective angle of attack, which is therefore subjected to an increase, decrease, decrease and another increase during the four consecutive half-strokes. Table 7.9 shows the time-averaged force coefficients for a wing following this ‘figure-of-O’ motion, while the mid-stroke angle of attack is fixed to α = 45◦ and the Reynolds number maintained at Re = 100. It is shown that the time-averaged lift coefficient decreases significantly with 9.8%. However, this is fully compensated by a decrease in drag with the same amount such that the differences in average lift-to-drag ratio are negligible. The average lift-to-drag was obtained by using the average drag over up- and downstroke. This was necessary, because of the asymmetry appearing in the average drag coefficient. This asymmetry in drag is the result of the asymmetrical variation in effective angle of attack, as was previously discussed. Figure 7.20 show the lift and drag variations during a complete flapping stroke. Secondly, the results are considered for a flapping wing following the ‘figureof-U’ pattern, which is similar to the one used in (Bos et al., 2008). When using this kinematic pattern, the wing moves down and up during every half-stroke, i.e. the upstroke is identical to the downstroke. In table 7.10 is can easily be seen that the differences in up- and downstroke drag are negligible for all deviation amplitudes. This is in contrast to the observations, considering the ‘figure-of-O’. The time-averaged drag coefficient is constant for deviation amplitude variation, while the lift coefficient decreases with 9.4%, leading to a similar decrease in liftto-drag ratio. Because of symmetric (similar up- and downstroke) flapping, a decrease of lift-to-drag coefficient is obtained, which is present, but not significant in comparison to the effect of varying Rossby and Reynolds numbers. The third pattern is governed by the ‘figure-of-eight’ wing tip motion. Although, the previously described deviation patterns only cause marginal effects,

20

7.5 Flapping wings at low Reynolds numbers

161

Aθ 0 5 10 15 20 1.224 1.204 1.180 1.151 1.104

CL (baseline) (-1.6%) (-3.6%) (-5.9%) (-9.8%)

CD downstroke 2.030 2.067 2.072 2.056 1.996

CD upstroke 2.030 1.934 1.850 1.767 1.676

CL /CD ave 0.603 0.602 0.601 0.602 0.601 (baseline) (-0.15%) (-0.23%) (-0.12%) (-0.28%)

Table 7.9 Time-averaged force coefficients for varying deviation using a ‘figure-of-O’ pattern. The time-averaged lift CL , drag CD and lift-to-drag ratio CL /CD are shown for a deviation amplitude, varying from Aθ = 0◦ to 20◦ , for a ‘figure-of-O’. The mid-stroke angle of attack was fixed at α = 45◦ and the average Reynolds number was maintained at Re = 100. 3 2.5 Lift coefficient [-] 2 1.5 1 0.5 0 -0.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 t/T [-] Aθ Aθ Aθ Aθ Aθ = = = = = 0 5 10 15 20 4 3 Drag coefficient [-] 2 1 0 -1 -2 -3 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 t/T [-] Aθ Aθ Aθ Aθ Aθ = = = = = 0 5 10 15 20

(a)

(b)

Figure 7.20 Force coefficients for a flapping wing with varying deviation using ‘figure-ofO’. Lift (a) and drag (b) coefficients induced by a flapping wing, such that Ro = 3.2 and Re = 100. The deviation amplitude is varied from Aθ = 0◦ to Aθ = 30◦ , using an angle of attack amplitude of Aα = 45◦ .

the ‘figure-of-eight’ pattern may cause significant changes in forces. Two types of ‘figure-of-eight’ patterns are used, from Aθ = 0◦ to 20◦ , this is called type 1 and the second from Aθ = 0◦ to −20◦ , type 2. To achieve a ‘figure-of-eight’ pattern, the deviation frequency is twice the frequency of the other two wing tip patterns. For both types of patterns, the effective angle of attack variation consists of three parts during each half stoke, see figure 7.2. Both up- and downstroke follow exactly the same, thus symmetrical, motion. The type 1 patterns starts each half-stroke with a downward motion, than it goes up until it has to go done just before stroke reversal. This wing tip motion leads to a consecutive increase, decrease and increase in effective angle of attack, where the period of decrease is twice the period of increase. The type 2 pattern follows precisely the inverse motion. Table 7.11 shows the time-averaged lift, drag and lift-to-drag ratio’s for both types of ‘figure-of-eight’ patterns. Since the variation of the effective angle of attack is symmetric, the drag coefficient is equal for up- and downstroke. The results

162

Vortical structures in three-dimensional flapping flight

of the type 1 motion pattern are remarkable. There is a considerable decrease in both time-averaged lift (52%) and drag (44%) when comparing Aθ = 20◦ with the baseline case Aθ = 0◦ . Since the drag decrease is of similar magnitude as the lift decrease, the loss in lift-to-drag is limited to 15.5%, which is still significant. Figure 7.20 shows the lift and drag variations for this type 1 ‘figure-of-eight’ pattern, where the wing tip moves down, up and down, consecutively during each half-stroke. It can be observed that the short period of downward motion at the beginning of each stroke increases lift. On the other hand, the large period of upward motion, decreases the effective angle of attack for a relative long period, leading to a significant loss of lift, as is seen in the figure. For Aθ = 20◦ the lift even shows a clear minimum at t = 0.2T , which was also present in cases without deviation but at small angles of attack, e.g. α = 15◦ , see section 7.5.1. When considering the type 2 ‘figure-of-eight’ pattern, table 7.11 shows no significant decrease in drag. The lift, however, decreases considerably, although significantly smaller compared to the type 1 kinematic pattern, 53% versus 12.6%. The motion of the type 2 deviation is apparently equally distributed, resulting in only 12.6% less lift and no differences in drag, while increasing the deviation amplitude. The maximal difference in lift-to-drag ratio is therefore -12.0%. This force balance is illustrated in figure 7.22, which shows the lift and drag during a full stroke. At the beginning of a stroke, the effective angle of attack is decreased, leading to lower lift and lower drag. During mid-stroke, the effective angle of attack is increased, which is reflected in the higher lift and drag. Summarising, it was shown that the variation in lift and drag can be significantly influenced by introducing deviation in the stroke pattern, i.e. ‘figure-of-O’, ‘figure-of-U’ and ‘figure-of-eight’. The ‘figure-of-O’ patterns resulted in an asymmetric force variation, due to asymmetric modulation of the effective angle of attack. Lift and drag decrease with a similar amount, such that the lift-to-drag ratio was only marginally affected. The time-averaged drag was not influenced by the symmetrical ‘figure-of-U’ pattern. The average lift, however, did decrease, such that a loss in lift-to-drag was observed. Two types of ‘figure-of-eight’ patterns were investigated, differing in the direction of motion. When the wing moved up-

Aθ 0 5 10 15 20 1.224 1.187 1.167 1.137 1.108

CL (baseline) (-3.0%) (-4.6%) (-7.1%) (-9.4%)

CD downstroke 2.030 2.003 2.012 2.014 2.021

CD upstroke 2.030 1.994 2.001 2.002 2.007

CL /CD ave 0.603 0.593 0.580 0.565 0.548 (baseline) (-1.7%) (-3.8%) (-6.3%) (-9.0%)

Table 7.10 Time-averaged force coefficients for varying deviation using a ‘figure-of-U’ pattern. The time-averaged lift CL , drag CD and lift-to-drag ratio CL /CD are shown for a deviation amplitude, varying from Aθ = 0◦ to 20◦ , for a ‘figure-of-U’ pattern. The mid-stroke angle of attack was fixed at α = 45◦ and the average Reynolds number was maintained at Re = 100.

7.6 Flapping wings in forward flight

163

Aθ 0 5 10 15 20 -5 -10 -15 -20 1.224 1.107 0.934 0.747 0.575 1.220 1.205 1.151 1.070

CL (baseline) (-9.5%) (-23.7%) (-39.0%) (-53.0%) (-0.3%) (-1.5%) (-6.0%) (-12.6%)

CD downstroke 2.030 1.859 1.621 1.370 1.129 2.067 2.057 2.056 2.017

CD upstroke 2.030 1.852 1.616 1.365 1.125 2.064 2.047 2.040 2.003

CL /CD ave 0.603 0.596 0.576 0.545 0.510 0.600 0.586 0.560 0.530 (baseline) (-1.2%) (-4.5%) (-9.5%) (-15.5%) (-0.5%) (-2.8%) (-7.1%) (-12.0%)

Table 7.11 Time-averaged force coefficients for varying deviation using a ‘figure-of-eight’ pattern. The time-averaged lift CL , drag CD and lift-to-drag ratio CL /CD are shown for a deviation amplitude, varying from Aθ = 0◦ to 20◦ and Aθ = 0◦ to −20◦ , for a ‘figure-of-eight’ pattern. The mid-stroke angle of attack was fixed at α = 45◦ and the average Reynolds number was maintained at Re = 100.

wards during mid-stroke, decreasing the effective angle of attack for a long period, the performance was limited in terms of lift and lift-to-drag ratio. On the other hand, if the wing moved downward during each mid-stroke, the performance was similar to the ‘figure-of-U’ pattern. The lift and drag are shown to be sensitive to different stroke patterns, such that the forces and performance can be easily modulated. Therefore, insects could use stroke plane deviation in extreme hovering or manoeuvring conditions. These findings are very interesting for the development of Micro Air Vehicles (MAV) as well.

7.6

Flapping wings in forward flight

In order to relate the results of the three-dimensional flow simulations to previously conducted two-dimensional studies (Thaweewat et al., 2009), a three-dimensional forward flapping wing has been studied. Based on (Thaweewat et al., 2009), the dimensionless wavelength is set to λ = 6.3 (maximal performance) in order to justify a comparison. The mid-stroke angle of attack is varied from α = 0◦ to 45◦ , while maintaining a Reynolds number of Re = 150. The three-dimensional wing ∗ motion is harmonic and scaled such that the dimensionless amplitude ARg , based on the radius of gyration Rg is comparable between different cases. Additionally, the average Reynolds number ReRg is matched. In table 7.12, the time-averaged force coefficients are shown for three different flapping situations: a two-dimensional plunging airfoil, a three-dimensional translating wing (Ro = 130) and a three-dimensional flapping wing (revolving, Ro = 3.2). For both flapping and translating wings, the force coefficients are com-

5 1 0.164 Vortical structures in three-dimensional flapping flight 3 2.6 0. such that Ro = 3.5 0 0. 3.7 0.21 Force coefficients for a flapping wing with varying deviation using type 1 of ‘figure-of-eight’. The deviation amplitude is varied from Aθ = 0◦ to Aθ = 20◦ .2 and Re = 100. The deviation amplitude is varied from Aθ = 0◦ to Aθ = −20◦ .4 0. such that Ro = 3.22 Force coefficients for a flapping wing with varying deviation using type 2 of ‘figure-of-eight’.6 0. Lift (a) and drag (b) coefficients induced by a flapping wing.1 0.4 0.7 0.9 1 t/T [-] Aθ Aθ Aθ Aθ Aθ = = = = = 0 5 10 15 20 (a) (b) Figure 7.5 0.9 1 t/T [-] Aθ Aθ Aθ Aθ Aθ = = = = = 0 −5 −10 −15 −20 (a) (b) Figure 7. .5 Lift coefficient [-] 2 1.5 3 Lift coefficient [-] 2.8 0.8 0.5 0.3 0.1 0.5 2 1. using an angle of attack amplitude of Aα = 45◦ .2 and Re = 100.7 0.4 0.5 0.7 0.8 0.5 0 -0.3 0.5 0.8 0.9 1 t/T [-] Aθ Aθ Aθ Aθ Aθ = = = = = 0 5 10 15 20 3 2 Drag coefficient [-] 1 0 -1 -2 -3 0 0.3 0.2 0.2 0.2 0.5 1 0.6 0.9 1 t/T [-] Aθ Aθ Aθ Aθ Aθ = = = = = 0 −5 −10 −15 −20 4 3 Drag coefficient [-] 2 1 0 -1 -2 -3 -4 0 0.1 0.2 0.6 0. Lift (a) and drag (b) coefficients induced by a flapping wing.5 0 -0.1 0. using an angle of attack amplitude of Aα = 45◦ .3 0.4 0.5 0 0.

028 -0. a smooth ring vortex is formed by the Ro = 3.580 -0. The three-dimensional flapping wing (Ro = 3.0%) (-28. Therefore. leads to similar force results as if the study was completely two-dimensional. Maximal lift occurs without wing rotation.231 -0.429 0.214 CL /CD ave 35.12. a relative equal amount of energy is lost.9%) (-21.2. it can be stated that a three-dimensional wing.2 Ro = 130 2D . as is shown in table 7.12 Time-averaged force coefficients in forward flight.964 -5. Two.969 1.8%) (-20.148 -4.403 0. The two tip vortices result in a symmetric flow. The difference with the translating wing becomes smaller with increasing angle of attack amplitude.316 -0.401 -0. λ∗ = 6.571 -6. Concerning the three-dimensional translating wing (Ro = 130). which results in a lower lift coefficient.710 1.119 -3.23. Figure 7. the lift decreases with 28% compared to the two-dimensional plunging airfoil. For the case without revolving (figure 7. It is clear that the time-averaged lift coefficient decreases with increasing angle of attack amplitude.366 -0. The drag varies with angle of attack such that it reaches a thrust optimum for Aα = 30◦ .2) generates larger lift coefficients compared to the translating case. Note that negative drag means thrust. leading to a similar decrease in lift.904 1. The mid-stroke angle of attack is varied from 0◦ to 45◦ .376 1.24 shows a comparison of the flow fields for Ro = 130 and Ro = 3. but lower compared to the two-dimensional flapping airfoil.414 1.834 1.824 Table 7.24(a)).673 27. A flapping wing (Ro = 3.3. which is the case for both two-dimensional and three-dimensional flapping.0676 -0.544 1. but the resulting thrust is minimal.897 -2.604 CLave (baseline) (baseline) (baseline) (baseline) (-28.016 0.270 -0.7. This is the case for mid-stroke values of the angle of attack.2%) (-27. performing a two-dimensional motion.419 -1.920 -2.175 0. The generation of tip vortices causes a loss of energy. pared with the two-dimensional results.2) in figure 7.4%) (-27.719 -2.211 -3. without an angle of attack variation.and three-dimensional time-averaged force coefficients for a flapping wing in forward flight at Re = 150. This is il