You are on page 1of 236

Numerical simulations of apping 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 apping 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 Magnicus 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 Magnicus, 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 Scientic Research (NWO),
NWO-ALW grant 814.02.019.
Voor mijn ouders
Summary
Both biological and engineering scientists have always been intrigued by the ight
of insects and birds. For a long time, the aerodynamic mechanism behind ap-
ping insect ight was a complete mystery, until several decades ago. Experiments
showed the presence of a vortex on top of the apping 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 apping wing Micro Air Vehicles, small
man-made yers to use in exploration and surveillance.
Several ow visualisation experiments and numerical simulations have been
performed to improve the understanding of apping wing aerodynamics in order
to design and optimise Micro Air Vehicles. However, the eects of wing kinemat-
ics on the ow 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 ow 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 apping wing in our numerical model, dierent 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 eciency, to solve the ow around apping wings.
The ow around apping wings, at the scale relevant to insect ight, is highly
unsteady and vortical, described by the unsteady incompressible Navier-Stokes
equations. Dierent dimensionless numbers are discussed, characterising the ow,
i.e. Strouhal and Reynolds numbers. Since the ow at the considered Reynolds
number, Re = O(100), is laminar, there is no need for additional turbulence
modelling, such that our simulations, assuming laminar ow, 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. Dierent mesh motion solvers are compared. Two existing methods
are assessed, solving the Laplace and a modied stress equation. Both methods
are very ecient 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 ight. 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 eciency, 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 eort considerably.
After the discussion of the governing equations, nite 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 eects of dier-
ent wing kinematic models, with increasing complexity, on hovering ight perfor-
mance. The results show that the sawtooth amplitude has a small eect on the
mean lift but the mean drag is aected signicantly. The second model simplica-
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 y model is
not aecting the mean lift to a large extent. The other realistic kinematic feature
is the deviation, which is found to have only a marginal eect on the mean lift and
mean drag in this two-dimensional study. However, the eective angle of attack is
altered such that the deviation leads to levelling of the force distribution.
Additionally, a numerical model for two-dimensional ow was used to investi-
gate the eect of foil kinematics on the vortex dynamics around an ellipsoid foil
subject to prescribed apping 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 apping 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
ow is periodic.
Furthermore, dierent results relevant to three-dimensional apping wing aero-
v
dynamics, are described. First, the ow around a dynamically scaled model wing
is solved for dierent 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 dierent 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 benecial for the stability of the leading-edge
vortex, leading to an increase in lift and eciency. 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 gure-of-eight pattern. As in two-dimensional
studies, the deviation may inuence the force distribution to a large extent, by
changing the eective angle of attack. Additionally, the three-dimensional ow is
compared with the two-dimensional studies performed on apping forward ight.
Finally, a preliminary investigation is performed to show the eect of wing
exing. Therefore, a pre-dened exing deformation is applied to a plunging airfoil
in two-dimensional forward ight and to a three-dimensional apping wing in
hovering ight. Concerning the exible airfoil in forward ight, 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 inu-
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 apping ight . . . . . . . . . . . . . . . . . . . . . . . . 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 ow simplications . . . . . . . . . 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 Diusion 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 ow solvers . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.10 Code validation and verication . . . . . . . . . . . . . . . . . . . . 34
2.10.1 2D vortex decay and convection . . . . . . . . . . . . . . . . 35
2.10.2 Validation using cylinder ows . . . . . . . . . . . . . . . . 39
2.11 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3 Mesh deformation techniques for apping ight 49
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.2 Dierent mesh deformation techniques . . . . . . . . . . . . . . . . 52
3.2.1 Laplace equation with variable diusivity . . . . . . . . . . 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 eciency . . . . . . . . . . . . . . . . . . 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 apping foils and wings 73
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.2 Governing equations for apping 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 exing . . . . . . . . . . . . . . . . 82
4.3.4 Numerical implementation of the wing kinematics . . . . . 83
4.4 Dynamical scaling of apping wings . . . . . . . . . . . . . . . . . 86
4.5 Computational domain and boundary conditions . . . . . . . . . . 88
4.6 Denition of force and performance coecients . . . . . . . . . . . 89
4.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5 Inuence of wing kinematics in two-dimensional hovering ight 93
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.1.1 Similarity and discrepancy between two- and
three-dimensional ows . . . . . . . . . . . . . . . . . . . . 94
5.1.2 Inuence 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 Dierent 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 apping foil 121
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
6.2 Flapping foil parametrisation . . . . . . . . . . . . . . . . . . . . . 122
6.3 Force coecients and performance . . . . . . . . . . . . . . . . . . 124
6.4 Numerical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.5 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . 126
6.5.1 Inuence of dimensionless wavelength . . . . . . . . . . . . 126
6.5.2 Inuence of dimensionless amplitude . . . . . . . . . . . . . 127
6.5.3 Inuence of angle of attack amplitude . . . . . . . . . . . . 129
6.5.4 Inuence of stroke plane angle . . . . . . . . . . . . . . . . 130
6.5.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
6.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
7 Vortical structures in three-dimensional apping ight 135
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
7.2 Three-dimensional apping 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 identication methods for ow visualisation . . . . . . . . . 145
7.5 Flapping wings at low Reynolds numbers . . . . . . . . . . . . . . 148
7.5.1 The angle of attack in apping ight . . . . . . . . . . . . . 149
7.5.2 Inuence of apping stroke curvature . . . . . . . . . . . . . 149
7.5.3 Inuence of Reynolds number . . . . . . . . . . . . . . . . . 152
7.5.4 Inuence of trapezoidal angle of attack . . . . . . . . . . . 158
7.5.5 Inuence of deviation . . . . . . . . . . . . . . . . . . . . . 160
7.6 Flapping wings in forward ight . . . . . . . . . . . . . . . . . . . . 163
7.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
8 Inuence of wing deformation by exing 171
8.1 Airfoil exing in two-dimensional forward apping ight . . . . . . 171
8.2 Wing exing in three-dimensional hovering ight . . . . . . . . . . 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 apping ight . . . . . . . . . . . . . . . . 181
9.3.1 Two-dimensional hovering . . . . . . . . . . . . . . . . . . . 181
9.3.2 Three-dimensional hovering . . . . . . . . . . . . . . . . . . 182
9.4 Conclusions on forward apping ight . . . . . . . . . . . . . . . . 184
9.4.1 Two-dimensional forward apping . . . . . . . . . . . . . . 184
9.4.2 Three-dimensional forward apping . . . . . . . . . . . . . 184
9.5 Preliminary conclusions on wing exing . . . . . . . . . . . . . . . 185
9.6 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
A Grid generation for apping 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-
tic observations during a voyage (1831-1836) around the world with his ship, the
Beagle. At the Gal apagos Archipelago, Darwin discovered slightly dierent bird
species living on the dierent 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 dierent island it adapted to the dierences in environ-
mental circumstances. This process has become known as natural selection, which
is also applicable to the early era of ight, millions of years ago. Birds are ancient
descendants of feathered dinosaurs (Templin, 2000) which developed the skill of
ight 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 o into
the thin air. Birds and insects are both apping their wings at dierent length
scales, leading to a dierent ow behaviour. The larger the animal, the lower the
need for apping wings, e.g. the Andean Condor only aps when it looses height
in the thermal winds, whereas a small insect, a fruit y (Weish-Fogh & Jensen,
1956) aps 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 yers to use in exploration and surveillance, see
2 Introduction
gure 1.1.
To optimise the ight performance of MAVs it is important to get a thorough
understanding of the complex ow generated by its wings, especially at smaller
length scales (< 5 cm). The apping wings induce complicated vortical struc-
tures which inuence the forces and performance characteristics in hovering and
forward ight. In order to study this kind of ows, researchers performed ow
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 ow 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 ow eld can not
be visualised in much detail, especially due to the reections and shadows of the
moving wings. On the other hand, when performing numerical simulations, using
Computational Fluid Dynamics (CFD), the forces and ow visualisations are a
direct result of the computations. Since it is interesting to solve for the forces
acting on a apping 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 ow around apping wings at low Reynolds num-
bers, O(100 1000). Section 1.2 briey provides background information on the
ow physics concerned, while section 1.3 deals with the dierent approaches for
analysing the ow, 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 apping ight
In order to illustrate the necessity and diculties with solving and visualising the
ow around apping foils and wings, dierent aspects of apping wing aerody-
namics are discussed. The vortex dynamics, the leading-edge vortex in particular,
is briey discussed, as well as the inuence of the wing kinematic modelling in
two- and three-dimensional problems.
Vortex generation in apping wing aerodynamics
Vortex generation in nature is fairly common in ows induced by aeroplanes, birds,
insects, but also by boats and trees. Large aeroplanes generate wingtip vortices,
see gure 1.2, which can cause damage to a following aeroplane which encounters
1.2 Physics of apping ight 3
(a) Wasp. (b) Entomopter. (c) Dely.
Figure 1.1 Dierent apping wing Micro Air Vehicle concepts. At lower Reynolds numbers,
apping MAV concepts can be used for hovering and low speed forward ight, which is especially
interesting for intelligence and exploration. (a) Flying insect scale robotic model, which is able to
perform a tethered take-o (Wood, 2008). (b) The U.S. patented Entomopter has four apping wings
powered by chemically-fuelled propulsion system (Michelson, 2008). (c) The Dely 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
ight. 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 ow patterns induced
by a hawkmoth. It was observed that the leading-edge vortex was stabilised by the radial ow 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 eect of vortex generation is ow 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, sh and insects,
e.g. gure 1.2 shows induced propulsive vortices generated by a water strider.
It is a common story that ies could not y 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 sucient to explain the high lift generation of insects. This mystery
persisted until the discovery of the unsteady vortical ow eld, gure 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 apping wing aerodynamics. (a) Two-dimensional il-
lustration of the wing kinematics and the resulting force vector generated by the apping airfoil at
Re = O(100) (Bos et al., 2008). (b) Three-dimensional leading-edge vortex generated by a apping
wing at Re = O(1000).
Leading-edge vortex
The potential benet of vortices attached to the wing was already discussed
by Maxworthy (1979) and Dickinson & Gotz (1993). It was Ellington et al. (1996)
who identied the presence of a leading-edge vortex (LEV) generated on top of the
apping 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 apping wing compared to two-dimensional
apping foil situations.
In apping foil aerodynamics the vortices are shed and form either a periodic
or chaotic wake pattern, depending on the kinematics, notably advance ratio and
dimensionless apping 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 ows, which is the case at low Reynolds numbers. It is thought that
the kinematics in two- and three-dimensional apping inuences the shear layer
direction and ow accelerations, which will undoubtedly inuence the development
of the leading-edge vortex (Lentink & Dickinson, 2009b).
In order to understand the physics of apping 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 y, operating at
Re = 110, from (Bos et al., 2008).
Inuence of insect wing kinematics on forces
The relevance of experiments and ow simulations of insect ight has been found
to depend on how reliably true insect wing kinematics are reproduced. Wang
1.2 Physics of apping ight 5
et al. (2004) and Sane & Dickinson (2001) showed that the kinematic modelling
signicantly inuences the mean force coecients and its distribution. Addition-
ally, Hover et al. (2004) showed that the angle of attack inuences the apping
foil propulsion eciency to a large extent. This illustrates the appreciable eects
which details of the wing kinematics, like parameter values and stroke patterns,
may have on ight performance. It further emphasises the need to critically assess
the inuence of kinematic model simplications.
In literature, dierent kinematic models have been employed to investigate the
aerodynamic features of insect ight. 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 apping 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 ow solutions which are strongly related
to the aerodynamic eciency. Lentink & Gerritsma (2003) varied airfoil shape
with amplitude and frequency xed at values representative to real fruit ies.
They concluded that the airfoil geometry choice is of minor inuence, 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 & Gotz, 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 ight. Both studies emphasised the
importance of angle of attack to inuence the propulsive eciency. Slightly more
complex fruit y kinematic models were used by Dickinson et al. (1999) and Sane
& Dickinson (2001) with their Roboy. Based on observation of true insect ight,
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 Roboy kinematic model. Using these models, the eect
of amplitude, deviation, angle of attack and the timing of the latter were explored.
The present thesis deals with dierent kinematic models from literature, both
the pure harmonic and the Roboy model, in order to investigate their inuence
on the aerodynamic performance (Bos et al., 2008). Furthermore, the results were
compared with more realistic fruit y kinematics obtained from the observation of
free ying fruit ies. Instead of performing a parameter study within the scope of
one kinematic model, the objective of the present study is to compare the eect
6 Introduction
(a) (b) (c)
Figure 1.4 Vortex wakes generated by cylinders and apping 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 simplications in kinematic modelling, which is of great importance to
both experiments and numerical simulations. Also, it may reveal the importance
of certain specic features of the stroke pattern, in relation to aerodynamic per-
formance.
The similarity between two- and three-dimensional ows
To limit both the parametric space as well as the computational eort, 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 ight and apping wing
propulsion, is the possibly restrictive applicability of two-dimensional results to
true insect ight. Additional important aspects are unsteady ow mechanisms,
wing exibility (uid structure interaction) and Reynolds number eects. In a
recent paper Wang et al. (2004) compared three-dimensional Roboy 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 nite-span wings at low Reynolds numbers. They observed
three-dimensional vortical structures around apping wings with low aspect ratio,
as was mentioned by Lighthill (1969).
Notwithstanding the possible discrepancy between two- and three-dimensional
ow, two-dimensional analysis has often been applied to obtain insight into the
aerodynamic eects of wing kinematics and geometry. Wang et al. (2004) con-
rmed that the similarities between two- and three-dimensional approaches are
sucient to warrant that a reasonable approximation of insect ight 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 & Gotz, 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 justication, the rst 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, dierent methods were used to solve and visualise the ow around
apping insect wings, from realistic fruit y measurements to three-dimensional
simulations using a representative model wing. In this section, dierent methods
will be briey addressed, from experimental methods to computational uid dy-
namics simulations.
Experimental investigations and quasi steady theory
Several experimental studies considered the ight performance of insects, and re-
vealed the complex nature of insect ight aerodynamics. The ow 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 gure 1.2. More recently, Srygley & Thomas
(2002) and Thomas et al. (2004) performed free ight and tethered experimental
visualisations using butteries and dragonies to show the complicated vortical
structures. This unsteady and vortical ow 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 y, Drosophila
Melanogaster, in particular).
Ellington (1984) indicated that the lift in insect ight is signicantly higher
than expected on the basis of quasi-steady aerodynamics, hence revealing that
important unsteady and vortical ow phenomena play a major role in insect ight.
In several studies (Dickinson & Gotz, 1993, Dickinson, 1994, Dickinson et al.,
1999) it was conrmed 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 amplied 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 ip between up and down stroke. The lift increasing eect of
the leading-edge vortex strongly depends on the kinematics of the apping wing
(Dickinson et al., 1999, Wang, 2000b, Sane & Dickinson, 2001, 2002, Lentink &
Dickinson, 2009a,b).
In order to understand insect ight 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 eects 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 conrms 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 ight. Dickinson et al. (1999)
investigated the ow around a apping Roboy model which moves in oil to meet
the same ow conditions as the real fruit y encounters (reproduction of Reynolds
number in particular).
Numerical simulations
Notwithstanding important advances in experimental techniques for non-intrusive
ow eld analysis, Particle Image Velocimetry in particular (Bomphrey et al., 2006,
Poelma et al., 2006), it remains dicult to capture all the relevant details of the
ow using only experimental techniques. An appealing approach, therefore, is to
supplement experiments with numerical ow simulations. A number of numerical
studies on full three-dimensional congurations have been reported, in relation to
specic insect geometries (moth: Liu & Kawachi (1998), fruit y: Ramamurti &
Sandberg (2002), Sun & Tang (2002), dragony: Young & Lai (2008), Isogai et al.
(2004)).
To perform numerical simulations around moving objects, such as apping
wings, one can use either immersed boundary methods (Peskin, 2002, Mittal &
Iaccarino, 2005), deforming mesh techniques (Boer de et al., 2007, Jasak, 2009),
see gure 1.5, or even complete re-meshing (Young & Lai, 2008, Zuo et al.,
2007). Although, the computational eort 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 xed 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 xed
boundaries (Mittal & Iaccarino, 2005). Nevertheless, when two wings touch, as in
the manoeuvre clap-and-ing (Weish-Fogh & Jensen, 1956), one will undoubtedly
need methods like overset, immersed boundary or re-meshing techniques.
Together with the unavailability of an accurate ow 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 eciency. 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 Dierent 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 apping foil and wing simulations.
Arbitrary Lagrangian-Eulerian formulation
The governing equations to solve the ow are generally discretised using the Eule-
rian description, where the uid is allowed to ow through the xed mesh. This is
in contrast to the Lagrangian formulation, where the mesh is xed to the uid or
material. If the material or uid deforms, the mesh deforms with it. This method
is commonly used to discretise the governing equations encountered in structure
mechanics. However, when the ow domain moves or deforms in time due to
a moving boundary, a xed 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 ow 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 uid ow through the mesh. At the time of writing, the
ALE method has become the standard implementation in most popular codes to
solve for the ow around moving boundaries while the mesh deforms accordingly.
1.4 Objectives and approach
Flapping ight 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 ecient 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 ow solver. Therefore, the overall goal of this research is to develop a reliable
mesh deformation technique, in terms of accuracy and eciency, to solve the ow
around apping wings. This method is used to study the complex vortical patterns
to identify optimal strategies in apping foil and wing aerodynamics. In order to
satisfy this aim, the following objectives are dened:
1. improve current mesh motion techniques and implementation, using an ac-
curate and ecient framework,
2. validate and verify the numerical solver with the implemented and improved
mesh motion technique,
3. solve for the ow around two-dimensional apping foils to study the wing-
wake interaction as well as the inuence of wing kinematics,
4. solve for the ow around three-dimensional apping wings to assess the im-
portance of parameters like apping 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 ow around apping 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 ow around a three-
dimensional apping 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 uid ow, a nite volume discretisation is
used, which is the subject of chapter 2. That chapter deals with the discretisation
of the dierent terms as well as a denition 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 veried. Within the


code, dierent 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 apping foil and wing aerodynamics. Before
proceeding to the numerical results of the ow around apping foils and wings, it
is important to discuss the numerical modelling for apping ight in chapter 4.
Chapters 5 and 6 deal with the numerical investigations of two-dimensional ow
around a apping foil in hovering and forward ight conditions, respectively. It is
1.4 Objectives and approach 11
found that the kinematic modelling has a large inuence on forces, performance
and wake patterns. These two-dimensional results provide good insight what to
expect of the three-dimensional ow around a apping wing, which is discussed
in chapter 7. Additionally, chapter 8 presents the preliminary results of a exing
wing in two- and three-dimensional apping ight. Complex vortical structures
induced by a model apping wing can be accurately solved and analysed. It will
be seen that accurately solving the ow around a apping 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 nite 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 nite 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 ight. For test problems involving vortex decay and
convection, it was found that the Van Leer ux limiter provides the most accurate
results, since the ow is dominated by the convection of vortices. Furthermore,
the ow around stationary and transversely oscillating cylinders showed that the
code of OpenFOAM solves the ow 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, dierent 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 Dierent 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 eld of CFD, three methods
are commonly used, the nite dierence, nite element and nite volume method.
Wesseling (2001) and Ferziger & Peric (2002) described these methods in more
detail. Traditionally, nite element methods are used for structural problems
whereas numerical simulations related to uid ow are mostly solved with nite
volume methods, as is the case in the current thesis. When using the nite 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 nite 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 ow solver uses this fact to solve the system in a more
ecient way. A drawback of a (block-) structured grid is that it is more dicult
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 eciency of the ow 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 dicult 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 dierent properties of the numerical solution method, consistency, con-
vergence, stability, conservation and boundedness, from (Ferziger & Peric, 2002).
Since it is often not possible to nd a numerical method which outperforms on
all aspects, the choice of numerical method is usually a trade-o. The following
properties are relevant concerning numerical simulations, especially when using a
nite volume approach, applied in general commercial and non-commercial CFD
solvers.
The rst important property is consistency. The discretisation should become
exact when the mesh resolution tends to innity, i.e. when the cell size approaches
zero. The dierence 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 dierential equations
as the mesh spacing tends to zero. Convergence is dicult to prove theoretically
for real engineering applications, so commonly the empirical approach is followed,
where the same computation is repeated on subsequently rened 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 amplied.
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 suciently 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 ux of a conserved quantity through
a specied system should be zero. Since the governing equations in nite volume
formulation are conservative, this property should be respected by the discretised
equations. One of the advantages of the nite 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 dierent CFD codes, the commercial ow 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 eld of computational
uid dynamics for decades. Two major drawbacks of a commercial solver are the
unavailability of the source code and the potential lack of sucient 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 uid ow 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 dierent terms, like diusion 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 ow 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
dened in order to test the inuence of dierent numerical settings, like discreti-
sation schemes, grid resolution and time-step size. The validation and verication
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 uid ow are a coupled set of non-linear par-
tial dierential equations (Anderson Jr., 1991, Panton, 2005). These equations
are derived from conservation of mass, momentum and energy within an innitesi-
mally small spatial control volume. For mass conservation, the following continuity
equation is obtained:

t
+

(u) = 0. (2.1)
Here, [kg/m
3
] is the density and u [m/s] the ow velocity vector. The nabla
operator is dened 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):
(u)
t
+

(uu) =

, (2.2)
where [N/m
2
] is the surface stress tensor, necessary in viscous uid ow. For
compressible ow calculations also the energy conservation equation is specied:
(e)
t
+

(eu) =

(u)

q +Q, (2.3)
where e [J/kg] is the total specic energy (including kinetic and potential energy),
q [W/s] is the heat ux vector and Q [Jm
3
/kg] equals the nett energy generation.
The full set of equations describing unsteady, compressible viscous ows, are called
the Navier-Stokes equations. The Navier-Stokes equations are non-linear, which
makes them dicult to solve; only for very simplied 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 uid, the stress tensor, , which is dened for a
Newtonian uid as
=
_
p +
2
3

u
_
I +
_
u +u
T
_
.
Here I represents the identity tensor, p [N/m
2
] is the pressure and [Ns/m
2
] is the
dynamic viscosity. To close the energy equation, the equation of state is specied,
such as the perfect gas law:
p = RT,
in which T [K] is the temperature and R [J/(molK)] the specic gas constant.
The constitutive relation for the total specic energy yields as follows:
e = e(p, T).
Additionally, the heat conduction is described using Fouriers law:
q = T,
with [W/(mK)] the heat conduction transport coecient.
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 ows, turbu-
lence research, multi-phase ows and a lot of other applications. However, these
equations can be dicult to solve and simplications can be made if applicable to
the concerning problem.
18 Finite volume discretisation
2.2.2 Incompressible laminar ow simplications
The current research deals with the ow around apping 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 ow 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
eects can be neglected. The incompressible Navier-Stokes equations are:

u = 0, (2.4)
u
t
+

(uu) =
p

+
2
u, (2.5)
with = / [m
2
/s] being the kinematic viscosity. For an incompressible ow,
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 dierent 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 are scaled with their reference values, as follows:
u

=
u
U
ref
, t

= t f
ref
, x

=
x
L
, p

=
p

ref
U
2
ref
,

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

= 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:

= 0, (2.7)
and
St
u

t
+

(u

) = p

+
1
Re

2
u

. (2.8)
In these equations, two main dimensionless numbers are identied 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

=
T
visc
T
conv
. (2.10)
These dimensionless numbers represent order estimates for time-scale ratios in the
ow. 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 uid ow is governed by non-linear partial dierential equa-
tions, which can only be solved analytically for extremely simplied model prob-
lems. The full Navier-Stokes equations, combined with the constitutive relations,
are applicable to all kind of ows, where the computational costs strongly depend
on the desired resolution and solution methods. When the ow is considered lam-
inar and incompressible, the governing equations are signicantly simplied, such
that the costs for solving may be reduced. However, these simplications need to
be justied by the concerned uid ow problem. Concerning apping wing physics
at lower Reynolds numbers (100 Re 1000) the ow inherently is incompress-
ible and laminar. Therefore, solving the unsteady incompressible laminar ow can
be seen as performing a Direct Numerical Simulation, at suciently 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 sucient 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 nite volume approach needs a domain sub-
division into a nite number of convex polyhedral control volumes without overlap,
completely lling 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 gure 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 elds, like surface normal gradients or face uxes, are dened 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 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 nite 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 can be seen as pressure or a velocity component.
Using a Taylor series approximation, the following expression is obtained:
(x) =
P
+ (x x
P
) ()
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 a similar expression can be found:
(t)
t
=
(t + t) (t)
t
+O(t). (2.12)
With this linear temporal behaviour of 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 renement.
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 briey 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 dened in gure 2.3(a) by the angle
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

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
nite volume cell quality. Cell non-orthogonality is dened 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 dened by the
vectors m and d. From (Jasak, 1996).
error of the diusion term. The second quality criterion is the cell skewness, see
gure 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 dened by:
=
|m|
|d|
,
where m and d are dened in gure 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 nite volume discretisation of the
incompressible momentum equation, which forms the basis for the incompressible
Navier-Stokes equations. This partial dierential equation has the following form:
u
t
+

(uu)

(u) =
p

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

(uu) : convection term,

(u) : diusion term.


Using the nite 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

(u)dV =
_
V
CV
p

dV. (2.14)
This equation is solved in both CFD codes used, Fluent

and OpenFOAM

. In
the remainder of this section, the dierent terms of equation (2.14) are elaborated
in more detail.
Before dealing with the discretisation of the dierent 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 diusion terms. For this, the scalar variable is used, which may represent
the dierent 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
(x)dV =
_
V
P
(
P
+ (x x
P
)()
P
)dV
=
P
_
V
P
dV + (
_
V
P
(x x
P
)dV ) (
P
)

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 denes 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 can be written,
using Gauss theorem, as:
_
V
P
dV =
_
S
CV
dS,
=

f
_
S
f
dS,

f
dS
f
.
f
.
2.5.1 Face interpolation schemes
Similar to the divergence term, the ux
f
is interpolated from the cell centre to
the face centres. A linear interpolation is performed using the following expression:

f
= f
x

P
+ (1 f
x
)
N
,
which is illustrated in gure 2.5. The linear interpolation factor f
x
is dened 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 dierencing.
In OpenFOAM

, Jasak et al. (1999) applied extra face interpolation schemes,


like upwind blending using a gamma coecient and dierencing using a ux 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 gure 2.4 (Sweby, 1984). The purpose
of ux 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 dierent ux 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

(
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

(
r
)
0 0.5 1 1.5 2 2.5 3
0
0.5
1
1.5
2
2.5
3
(b) Koren
r

(
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 Dierent ux splitting limiters. Flux splitting schemes are used to limit the gradient
of the solution in order to avoid spurious wiggles. The ux splitting scheme are a function of r, which
represents the ratio of successive gradients on the mesh. Dierent ux 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

(u)dV =

f
S
f
(u)
f
=

f
S
f
(u)
f

f
=

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

()dV =

f
S
f
()
f
=

f
(S )
f
.
Here, the terms (S )
f
and
f
need to be approximated using a proper method.
The face viscosity
f
is obtained by interpolation from cell centre to faces. The
other term (S )
f
is obtained on a non-orthogonal mesh by the following ex-
pression:
(S )
f
= |m|

N

P
|d|
+k ()
f
.
2.5 Discretisation of an incompressible momentum equation 25
ts
Flow direction
U P D

D
f
P d f m N
S
f

N
k
Figure 2.5 Variation of the ux . The
value of 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 gure 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 which needs to be discretised using new and old time values.
This time dierence is dened using prescribed time-step size t such that:

n+1
=
n
+ t,
where
n
and
n+1
are the scalar variable at the old and new time instances,
respectively. Two implicit time discretisation methods are considered, one rst-
order and and one second-order scheme. The rst-order discretisation is simply
the temporal dierence:

t
=

n+1

n
t
,
and the second-order discretisation, see (Wesseling, 2001, Hirsch, 1988), is given
by:

t
=
3
2

n+1

n
+
1
2

n1
t
.
This implicit scheme is referred to as the second-order backward dierencing
scheme, where
n1
is the old-old value of . Consequently, the corresponding
26 Finite volume discretisation
volume integrals obey the following relations:
_
C
V

t
dV =

n+1

n
t
V
P
,
_
C
V

t
dV =
3
2

n+1
2
n
+
1
2

n1
t
V
P
. (2.17)
Note, that these relations are only valid on xed meshes and constant time-steps.
According to (Wesseling, 2001, Hirsch, 1988), the explicit rst-order time integra-
tion method may be unstable if the Courant number is larger than 1, where the
Courant number is dened 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 rst- and second-order backward
scheme have been used. While the implicit methods are bounded and stable, a
pre-dened maximal Courant number Co
max
is used to vary the corresponding
time-step during the simulation. In that case, the coecients
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 dened 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, dening the solution gradient to be zero.
This condition is known as a Neumann-type condition, /n = a,
2. xed-value boundary condition, dening a specied value of the solution.
This is a Dirichlet-type condition, = b,
3. symmetry boundary condition, treats the conservation variables as if the
boundary was a mirror plane. This condition denes that the component
of the solution gradient normal to this plane should be xed 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 ux zero, using the Arbitrary Lagrangian Eulerian approach.
For external ow simulations, a distinction is made between the outer and the
inner boundaries, the latter corresponds to the moving wing or body. To minimise
the eects 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 xed boundaries, unless a free-stream is specied. In case of
forward apping ight, two domain boundaries are dened as inow and outow,
respectively. At the inow boundary the velocity is dened as xed-value (2) and
the pressure as zero-gradient (1). On the other hand, at the outow boundary,
the pressure has to be xed-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 xed-value (u = 0) is specied 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 ux
through the moving boundary.
2.7 Solution of the Navier-Stokes equations
Previously, the dierent terms to discretise the general momentum equation (2.13),
were described. This section briey 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

+
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 o-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 elds will be
used to calculate the matrix coecients 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 coecients 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 nd an expression for u
p
:
u
p
=
H(u)
a
p

1
a
p
p. (2.20)
The velocities on the face of a nite 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 uxes. 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 nal 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 ux 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 satised, the face uxes 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 ow solvers, Fluent

and OpenFOAM

used the PISO scheme for transient ows and the SIMPLE
scheme for steady ows. 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 eld.
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 eld can be obtained.
3. Explicit velocity correction stage. The last equation in this sequence
is (2.25), which determines the conservative uxes, which are consistent with
the new pressure eld. Since the approximated pressure eld, from stage 1,
is replaced by a better pressure eld, from stage 2, the velocity eld 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 ow
30 Finite volume discretisation
with additional turbulence modelling, the solution sequence can be summarised as
follows:
1. Initialisation of all elds, 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
uxes,
4. Iterate through the PISO loop until the pre-dened tolerance of the pressure-
velocity system is reached. The pressure and velocity elds are obtained for
the current time-step, as well as a new set of conservative uxes,
5. Using the new conservative uxes, all remaining equations of the system are
solved. If turbulence modelling is included, update the turbulent viscosity
at this stage,
6. Unless the nal time is reached, go back to step 2.
This procedure results in solution elds 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 modications 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 ow are generally discretised using the Eule-
rian description, where the uid is allowed to ow through the xed mesh (Ferziger
& Peric, 2002). This is in contrast to the Lagrangian formulation, where the mesh
is xed to the uid or material. If the material or uid deforms, the mesh deforms
with it. This method is commonly used to discretise the governing equations
encountered in structural mechanics. However, when the ow domain moves or
deforms in time due to a moving boundary, a xed 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 ow
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 uid ow 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 ow 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
eld can be derived on a moving mesh as

t
_
V
CV
dV +
_
S
CV
n(uu
s
)dS
_
S
CV

ndS =
_
V
CV
S

()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 dened 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 nite volume discretisation, the computational domain is split
into a nite number of polyhedral cells with varying shape and volume, since the
mesh is deforming. The cells do not overlap and completely ll 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 rst-order
Euler method or a second-order backward scheme (Tukovic & Jasak, 2007). For
eciency, 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:
3
n+1
P

n+1
P
V
n+1
P
4
n
P

n
P
V
n
P
+
n1
P

n1
P
V
n1
P
2t
+

f
( m
n+1
f

n+1
f

V
n+1
f
)
n+1
f
=

f
(

)
n+1
f
S
n+1
f
n
n+1
f

()
n+1
f
+s
n+1

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 n1 are, respectively, the new, old and
old-old values. The mass ux 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 uid mass ux m is obtained as part of the solution, satisfying mass
conservation. Furthermore, it is important to determine the volume face ux
such that it satises 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 ux 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 nite 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 dened 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
_

n+1
P

n+1
P
V
n+1
P

_
1 +
t
n+1
t
n
_

n
P

n
P
V
n
P
+
_
1 +
(t
n+1
)
2
t
n
(t
n+1
+ t
n
)
_

n1
P

n1
P
V
n1
P
+

f
( m
n+1
f

n+1
f

V
n+1
f
)
n+1
f
=

f
(

)
n+1
f
S
n+1
f
n
n+1
f

()
n+1
f
+s
n+1

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
n1
.
2.8 Swept volume calculation
The swept volume is dened 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
gure 2.7(a). The polyhedral face is decomposed into triangles, using its centroid,
2.9 Numerical ow solvers 33
which is illustrated in gure 2.7(b). The swept volume of a polyhedral face f
is equal to the sum of the swept volumes of the dierent decomposed triangles,
which need to be accurately calculated. As is illustrated in gure 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 gure, but the remaining
volume of the prism contains two more tetrahedrons. These two additional tetra-
hedrons can be constructed in two dierent unique ways, using either one of the
two diagonals in the right side face of the swept volume shown in gure 2.7(b),
see (Zuijlen van, 2006). Therefore, the total swept volume using the two dierent
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 dierent 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 ow
with mesh motion and proved to be suciently accurate. Therefore, this method
is implemented in OpenFOAM

.
2.9 Numerical ow solvers
Section 2.7 described the solution procedure to solve the Navier-Stokes equations,
necessary for uid ow. This section deals with a brief elaboration of the computer
software, i.e. the CFD codes, used throughout the present research. Two dierent
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 nite volume approach. For completeness, the main settings that we used,
are briey discussed. The spatial discretisation was second-order upwind and
the time discretisation was rst-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 ow 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 satised 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 dierential equations, e.g.
the Navier-Stokes equations, on a nite 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 dierencing, 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 specied. 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 dened 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 ow solvers Fluent

and OpenFOAM

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

has already been tested earlier specically for low Reynolds number ows, relevant
to apping insect ight in (Lentink & Gerritsma, 2003, Zuo et al., 2007, Bos et
al., 2008, Thaweewat et al., 2009). OpenFOAM

, however, has been tested exten-


sively for uid-structure interaction (Tukovic & Jasak, 2007), mesh motion (Jasak,
2009) and Large Eddy Simulations (Jasak, 1996, Juretic, 2004, Jasak et al., 2007),
but not for low Reynolds number ows. Therefore, this section only deals with
the validation and verication of the OpenFOAM

ow solver for test problems


relevant for low Reynolds number apping insect ight.
2.10 Code validation and verication 35
The open-source toolbox OpenFOAM

provides a framework of nite volume


based functions in order to build a specic application for solving partial dif-
ferential equations. Within the context of the present research, an application
is developed, which solves for the unsteady, incompressible ow using deforming
meshes. Besides the uid and ow properties it is necessary to specify dierent
interpolation schemes for the dierent terms of the governing equations. Since this
ow solver was used for low Reynolds number ows with vortex convection it is
necessary to choose the correct face interpolation scheme for the convection term.
Therefore, the face interpolation for diusion and source terms will be kept xed
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 dierencing, SuperBee splitting, Koren splitting and Van Leer
splitting, as was briey 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 ow solver settings, the accuracy of
the code is assessed using unsteady ows 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 apping ight, since these ows are dominated by unsteady
vortical structures. First a proper vortex denition is described, followed by the
two vortex simulations to discuss the results.
Vortex denition
When a vortex is used for code validation it is important that a well conned
denition is used, i.e. the vortex should have nite and monotonic velocity and
vorticity proles. The following denition is used from (Panton, 2005), but ana-
logue to (Zhou & Wei, 2003):
V

= t
m
f(), (2.27)
where V

represents the radial velocity, m is a compactness coecient and a sim-


ilarity parameter. Figure 2.8 shows the velocity prole for dierent values of m to
show the eect on the compactness of the vortex. In this gure some interesting
characteristics can be observed. With decreasing m, the curve prole 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 Dierent velocity proles of a similar vortex denition. The vortex denition,
from (Panton, 2005), provides dierent vortex proles 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 ows like in low Reynolds num-
ber vortex shedding problems (Panton, 2005). When the compactness coecient
m is xed to 1.5 the velocity prole of this Taylor vortex can be derived (Panton,
2005) and is given by:
V

= 2e
1
2
e
2
2
, (2.28)
where is a function of the similarity parameter :
=

2
_
(2)
.
Vortex decay
As a rst 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 nite volume grid of 100 x 100 mesh cells as is shown
in gure 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 eect 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 xed
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 ne enough. The Reynolds number
2.10 Code validation and verication 37
Figure 2.9 Initial solution of a Taylor vortex on an Cartesian mesh. Starting from this initial
Taylor vortex solution the ow diusion is solved, without the presence of a convection free-stream.
The velocity proles and total energy are monitored to identify the accuracy of the ow solver.
is xed to Re = 100 by setting the kinematic viscosity to = 0.01 and the velocity
vector to u = (1.0, 0.0, 0.0). The ow solver solves for 40 seconds such that the
temporal eect of the dierent face interpolation schemes on the shape and mag-
nitude of the velocity proles can be compared. Any excess or lack in numerical
diusion may become visible.
Results
Figure 2.10(a) and 2.10(b) show the initial and nal velocity variations in X- and
Y -direction for the dierent face interpolation schemes. Besides the evolution of
the velocity prole the total energy is shown in gure 2.10(c), which is a measure
for the diusion. 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 rst important observation, from gure 2.10, is that the total energy is
increasing for the SuperBee ux limiter. This scheme clearly introduced a large
amount of negative numerical diusivity which causes the Taylor vortex to grow,
which is not physical. The velocity prole as well as the total energy of both
Gamma and linear interpolation schemes are similar, without a signicant amount
of diusion. One major drawback of these methods is that the vortex looses sym-
metry after 5 s, gure 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 dierent face interpolation schemes, while (b) provides v(x). (c) shows the
decaying total energy due to numerical diusion.
diusive, see gure 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 gure 2.11. Similar to the vortex
decay problem, the discretisation of the face interpolation is varied and all other
discretisation terms are xed 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 ow solver solved the governing equations for 20 s, such that
the vortex was convected through the entire domain. The Reynolds number was
xed to Re = 100 by setting a kinematic viscosity to = 0.01 and an inlet velocity
2.10 Code validation and verication 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 suciently large grid resolution and time-step size, the eect of dierent face interpolation schemes
is compared.
to u = (1.0, 0.0, 0.0).
Results
First of all, gure 2.12(c) shows that the total energy in the entire domain is de-
creasing with time for all schemes. Extra diusion is clearly visible after t=14 s
when the vortex approaches the outlet boundary, smearing the vortex. Since
convection induces physical diusion, depending on the convection velocity, the
SuperBee scheme possibly still has negative numerical diusion, 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, gure 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 diusion 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 eect increases with time and therefore these two methods
are not appropriate to study the vortical wake patterns in insect ight. 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 ows
To validate the accuracy of the ow solver for unsteady and vortical ow, two
cylinder example problems are dened at suciently low Reynolds numbers. At
Reynolds numbers less than Re = 47 the ow exhibits steady behaviour (see
Williamson, 1998), which is not relevant for unsteady insect ight, 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 proles 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 prole 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 ow becomes turbulent and additional
turbulence modelling becomes necessary. Since the main objective of this research
is to solve for unsteady, vortical ow around apping wings, the characteristics of
that kind of ow needs to be present in the validation cases. The apping wing
simulations are performed in the laminar ow regime Re = O(100), with periodic
force histories. Therefore, in the range 100 Re 200, two validation cases
were selected, one concerns the ow 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 coe-
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 outow
2.10 Code validation and verication 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-
cient. (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 coecient (), pressure drag coecient () and total drag coecient () 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 ow solver. The ow 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 eects are minimal at this
domain size.
Flow around a stationary circular cylinder
The rst case, dealing with the ow 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,
gure 2.13(b) shows the results from an extensive study performed by Henderson
(1995) to identify a relation between drag coecient and Reynolds number. The
42 Finite volume discretisation
Figure 2.15 Vorticity visualisation of the Von Karman vortex street. The ow around a
stationary cylinder shows a periodic vortex street, of which the frequency depends on the Reynolds
number. Vorticity = u is used to identify the vortical structures, which are clearly visible at a
Reynolds number of Re = 150.
resulting drag coecient at Re = 150 is found to be C
D
= 1.333. The drag and
lift coecient are respectively dened as:
C
D
=
D
1
2
U
2
ref
, C
L
=
L
1
2
U
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 ow behaviour, gure 2.15 shows the instantaneous vorticity
( = 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 dierent mesh resolutions (25k, 50k and 100k) are shown
in gure 2.16. These limit cycles are determined by taking the periodic part of
the force histories as shown in gure 2.19. From these periodic forces, the time-
averaged drag coecient is determined and compared with literature in table 2.1.
From this table, it can be observed that the drag coecient of the coarsest case,
25k and Co
max
= 2.0, has the largest dierence with literature, 5.63%. When the
grid is rened and the time-step decreased, it is seen that the solution decreases
asymptotically. The drag on the nest 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-
cient. The dierences of the mean Strouhal number with literature ranges from
2.10 Code validation and verication 43
Drag coecient [-]
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 coecient [-]
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 ow around a stationary cylinder. The periodicity
of the ow 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 coecient, 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 ow around a stationary cylinder. Comparison of
the time-averaged drag coecient for dierent grids, 25k, 50k and 100k, and dierent 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 suciently small.
From table 2.1 and 2.2, it is clear that the ow solver produces results which
are suciently 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 ow solver is
developed in a numerically consistent way, the ow solution should converge to
an asymptotically value with increasing spatial and temporal resolution. In order
to illustrate if the solution converges, gure 2.17 shows the time-averaged drag
coecient 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 gures, is the extrapolated
values, using Richardsons 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 coecient as a function of spatial and temporal resolution
for the stationary cylinder. A ow solver is numerically consistent if the ow solution converges
with increasing grid resolution and decreasing time-step size. (a) shows the drag coecient with
increasing grid renement level for dierent time-steps. (b) shows the drag coecient with decreasing
time-step size for the dierent grid sizes. The Richardson extrapolated values are plotted at .
lated value is obtained using the following expression:

extrap
=
fine
+

fine

coarse
2
p
1
, (2.30)
where
extrap
is the extrapolated value,
fine
and
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 ow around a stationary cylinder. Com-
parison of the time-averaged Strouhal number for dierent grids, 25k, 50k and 100k, and dierent
time-steps corresponding to a maximal Courant number, Co
max
= 2.0, 1.0, 0.5 and 0.25.
2.10 Code validation and verication 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 coecient amplitude with increasing grid renement
level for dierent time-steps. (b) shows the time-averaged lift coecient amplitude with decreasing
time-step for the dierent 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 coecients 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 coecient,
gure 2.18 shows the convergence of the lift coecient amplitude in order to prove
the consistency of the ow solver. Concerning the extrapolated values of the drag
coecient, from table 2.1, it can be determined that the dierences varies from
2.7% to 3.5% compared to literature with decreasing mesh resolution. For all
meshes, the dierences in drag coecient 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 sucient 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 ow 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 dened as
y(t) = A
e
sin(2f
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 sucient to investigate the moving wing capabilities
of the numerical model. Previously conducted simulations on stationary cylinder
ow showed that a mesh of 50k provides a suciently 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 sucient, the following values are used to show that
the ow 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 ow solver applied to apping wings, the results of
an oscillating cylinder case were compared with literature. Guilmineau & Queutey
(2002) found a drag coecient of C
D
= 1.2. Figure 2.20(a) shows the limit cy-
cle results, using the present ow solver, with second-order temporal and spatial
discretisation. From this gure it is obvious that the ow solution converges with
decreasing time-step. This statement is conrmed if the time-averaged drag coef-
cient is plotted in gure 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
suciently accurate.
2.11 Conclusions
This chapter has presented the nite 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 ecient results, the mesh qual-
ity should be high in terms of non-orthogonality and skewness, both mesh quality
measures. The dierent terms of the governing equation were discretised using
second-order schemes and dierent ux splitting methods were described concern-
ing the face interpolation. In order to solve the ow on a computational grid, four
dierent types of boundary conditions were specied, xed-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 coecient [-]
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 ow around an oscillating cylinder. (a) shows the C
D
-C
L
limit
cycles for dierent time-steps on a 50k mesh. (b) shows the time-averaged drag coecient 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 dierent commercial and non-commercial CFD solvers. The present
research used the commercial CFD solver Fluent

and the open-source CFD code


OpenFOAM

, both were briey described. Fluent

has already been tested for


various ows in literature, OpenFOAM

too, but not for low Reynolds ows, rel-


evant for apping insect ight. Therefore, this chapter presented a validation of
OpenFOAM

for a number of relevant test cases. Vortex decay and convection


were used to study the inuence of the face interpolation scheme with dierent
ux limiters. It was found that the Van Leer ux limiter provides the most accu-
rate results, concerning vortex decay and convection. In addition, stationary and
transversely oscillating cylinder ows were used to successfully prove spatial and
temporal convergence. It is concluded that the open-source solver OpenFOAM

provides an accurate and ecient framework to investigate the ow around ap-


ping wings at low Reynolds numbers.
CHAPTER 3
Mesh deformation techniques for
apping ight
Submitted to Comput. Meth. Appl. Mech. Engrg. (January 2010).
In order to use mesh deformation techniques to investigate apping wing aerody-
namics it is necessary to maintain a high mesh quality for relevant wing kinematics.
Dierent mesh deformation techniques are compared in order to identify their ap-
plicability for cases with apping wings. The main diculty 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 dierent 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 apping wings and can handle exing boundaries. In order
to increase the eciency of this method, two techniques are applied, based on
boundary coarsening and smoothing of the radial basis function.
50 Mesh deformation techniques for apping ight
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 xed or deforming. In
engineering, there are numerous computational uid dynamics (CFD) problems
in which the ow solution involves geometrically deforming boundaries. Exam-
ples of such interaction problems are uid-structure-interaction cases like blood
ow through arteries or deforming ags. A simplied type of interaction is that
which concerns a one-way coupling, where the ow is being inuenced by a chang-
ing boundary shape. This may be caused by imposed external eects, like the
rigid body motion, e.g. a apping 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 rst method is
called the immersed boundary method (Peskin, 2002), which denes a moving
boundary on a stationary Cartesian background mesh. The disadvantages of the
immersed boundary method are the diculty 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
dierent 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 ow in an accurate and ecient 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 dierent aspects need to be formulated, quality,
eciency and robustness.
The quality of the resulting mesh is dened by the non-orthogonality and skew-
ness of the nite volume cells. Eciency 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 dened 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 ecient techniques available to deform the mesh, for
example Transnite Interpolation (Wang & Przekwas, 1994). They interpolated
the displacements of the boundary points along grid lines through the entire com-
putational mesh to nd the displacements of all interior mesh points. Using an
additional mapping (Wang, 2000b) the mesh quality can be improved signicantly
if the boundary is subjected to signicant 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 renement, the focus is put on mesh deformation techniques which
can be applied to unstructured meshes containing arbitrary polyhedral cells, used
in the nite 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 specic spring stiness.
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 dierential equa-
tion on the complete eld of internal mesh displacements for given boundary point
displacements. Concerning the governing partial dierential equations, the Laplace
and bi-harmonic operators (Lohner & Yang, 1996, Helenbrook, 2003) are often used
in combination with a constant or variable distance-based diusion coecient 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 nite element codes. Dwight (2004) modied this method to
incorporate rigid body rotation, which signicantly improves the mesh quality for
meshes subjected to large boundary translations and rotations.
Since these methods solve a partial dierential equation on the complete eld
of internal mesh points the existing iterative solvers can be used, already available
in existing CFD codes (Jasak & Tukovic, 2004, Jasak, 2009). Therefore, the par-
allel implementation of these methods is fairly straightforward. Depending on the
method, a variable diusivity eld needs to be dened, which acts as a stiness
of the system of equations, this inuences the eciency. 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 apping ight
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 uid-structure-interaction. (Boer de et al., 2007) used radial
basis function interpolation to couple two non-matching meshes at the interface of
a uid-structure interaction computation. An RBF interpolation function is used
to transfer the known boundary point displacements to the uid boundary mesh.
Since the application of RBFs to interpolate from and to the boundary mesh was
very accurate and ecient, 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 eciency. Only recent studies (Jakobsson & Amoignon,
2007, Rendall & Allen, 2008c,b) have been carried out to improve the eciency 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 diusivity and a modied 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 dierent mesh motion methods are described in section 3.2. In order to
assess the mesh quality, section 3.3 discusses two dierent criteria, based on non-
orthogonality and skewness, described in chapter 2. The resulting mesh quality
of the dierent 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 simplied moving block, two
more relevant test cases were considered, one using a three-dimensional apping
wing and the other involves a two-dimensional exing airfoil. Since the radial ba-
sis function mesh motion method is computationally expensive, section 3.5 deals
with two techniques to increase its eciency. Finally, the conclusions are drawn
in section 3.6.
3.2 Dierent 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 (uid) points. The displace-
ment of the boundary points can be considered to be given, either it is externally
dened, i.e. a prescribed rigid body motion, or it is part of the solution, which is
3.2 Dierent mesh deformation techniques 53
the case in uid-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 inuences 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
dierent methods, as will be shown in the next section.
3.2.1 Laplace equation with variable diusivity
One can think of a deforming computational domain as if it was a solid body under-
going internal stresses given by the Piola-Kirchho 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 eciently.
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 speci-
cation of a variable diusivity. This leads to the following denition of the Laplace
equation:

(x) = 0,
where x is the displacement eld and the diusion coecient, which decreases
with the radius r from the deforming boundary as follows:
(r) =
1
r
m
. (3.1)
The resulting mesh quality strongly depends on the chosen (r) function, which
depends on the distance from the moving boundary. This variable diusion coe-
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 (r) function like
equation (3.1). In addition to the freedom of choosing a diusion function, it is also
possible to dene (r) for every internal mesh cell for all time-steps independently.
This, however, appears to be very problem dependent and thus optimisation of
(r) seems not cost eective. To maintain robustness, in the current work we use a
quadratically, m = 2, decreasing diusion coecient, which was found to provide
ecient and a smooth mesh motion (Jasak & Tukovic, 2004). Additionally, one
could also have used an exponentially decreasing diusion coecient or a diusion
coecient related to the mesh deformation energy.
54 Mesh deformation techniques for apping ight
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

= f , (3.2)
where is the stress tensor and f the acting force vector. The stress tensor is
given in terms of the strain, which is given by the following constitutive relation:
= tr ()I + 2, (3.3)
where tr is the trace and and are Lame constants (Baruh, 1999), which are a
property of the elastic material. The constants can be related to Youngs modulus,
E, as
=
E
(1 +)(1 2)
, =
E
2(1 +)
,
where is Poissons ratio, meaning the material contraction ratio as it stretches.
The following equation:
=
1
2
(x +x
T
), (3.4)
denes 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:
=
1
2
(x +x
T
+x
T
x). (3.5)
Combining equations (3.5), (3.3) and (3.2), together with = E and = E, the
following solid body rotation stress equation is obtained:

(x) +((x x
T
)) tr(x) = 0, (3.6)
where is a similar diusion coecient 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-
ne the point motion using interpolation techniques, like the transnite 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 apping wing simulations.
3.2 Dierent mesh deformation techniques 55
3.2.3 Radial basis function interpolation
In the current work we use radial basis function interpolation to nd the dis-
placements of the internal uid 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

j
(||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 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 (Boer de et al., 2007).
A unique interpolant is given if the basis function is a conditionally positive denite
function. If the basis functions are conditionally positive denite 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 dened by
the coecients
j
which can be dened 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

j
p(x
b
j
) = 0,
which holds for all polynomials p with a degree less or equal than that of polynomial
q, the
j
values can be determined (Boer de et al., 2007).
The values for the coecients
j
and the linear polynomial can be obtained by
solving the system:
_
x
b
0
_
=
_

bb
Q
b
Q
T
b
0
_ _

_
, (3.8)
where is containing all coecients
j
, the four coecients of the linear poly-
nomial q,
bb
an n
b
n
b
matrix contains the evaluation of the basis function

b
i
b
j
= (||x
b
i
x
b
j
||) and can be seen as a connectivity matrix connecting all
boundary points with all internal uid 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 dicult 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 ecient way are discussed in section 3.5.
56 Mesh deformation techniques for apping ight
When the coecients in and are obtained they are used to calculate the
values for the displacements of all internal uid 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 coecients and , which are then used for step two, the evaluation
using equation (3.9).
In contrast to the Laplace and SBR Stress methods, no partial dierential
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 diusion coecient
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:
(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 inuenced 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 eciently using
common iterative techniques.
In table 3.1 various radial basis functions with compact support are shown
using the scaled variable = x/r. The rst 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 Dierent mesh deformation techniques 57
Ref. nr. RBF Name f()
1 CP C
0
(1 )
2
2 CP C
2
(1 )
4
(4 + 1)
3 CP C
4
(1 )
6
(
35
3

2
+ 6 + 1)
4 CP C
6
(1 )
8
(32
3
+ 25
2
+ 8 + 1)
5 CTPS C
0
(1 )
5
6 CTPS C
1
1 +
80
3

2
40
3
+ 15
4

8
3

5
+ 20
2
log()
7 CTPS C
2
a
1 30
2
10
3
+ 45
4
6
5
60
3
log()
8 CTPS C
2
b
1 20
2
+ 80
3
45
4
16
5
+ 60
4
log()
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 = 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 uid-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 apping ight
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 at 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 RBFs 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 nally 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 dierent mesh motion solvers.
Absolute and relative radial basis function interpolation
In principle there are two dierent 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 coecient arrays and are calculated and used to calculate the internal
point displacements at all time-steps. This method is very ecient 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 coecients are
not dened 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 dened 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 dierent 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 denition were introduced. In
order to compare the quality of the dierent 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 elds such that they can be used for post-processing.
The test cases used to compare the mesh quality of the dierent 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
nonortho
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 ecient 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 dierent 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
dierences in mesh quality obtained with the dierent mesh motion solvers. Addi-
tionally, the eect of dierent radial basis functions is discussed. The rst simple
case is a two-dimensional block which performs a combination of translation and
rotation. The initial computational mesh is shown in gure 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 gure. Mesh motion simulations are performed using
the dierent 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 apping wing, followed by an example of a two-dimensional
exible moving boundary in section 3.4.3.
3.4.1 Translation and rotation of a two-dimensional block
The rst 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 xed, such that the inuence of the moving boundary
points on all internal points could be studied independently. The resulting mesh
60 Mesh deformation techniques for apping ight
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 elds of
skewness and non-orthogonality. Using those elds, 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 diusivity coecient which decreases
proportionally to the distance from the moving boundary points. Following Jasak
& Tukovic (2004) and Jasak (2009), a quadratically decreasing diusivity coef-
cient, 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
xed. Finally, the newly implemented RBF mesh motion solver is used with ve
dierent 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 gure, 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 diusion coecient, is simply not robust enough to obtain high mesh
quality when the boundary rotates. As can be seen in gure 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.


gure 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
gure 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, gure 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 dierent mesh motion solvers, including dierent RBFs. 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 RBFs 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 eect of the support
62 Mesh deformation techniques for apping ight
(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 RBFs result in an maximal
skewness of respectively 32% and 45% compared to the Laplace motion solver,
while the dierence of the other RBFs 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, gure 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 prole, 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 simplied 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 dierent 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 dierent RBFs.
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 apping wing is shown in gure 3.6 at mid-stroke. Figure 3.6(a)
shows the non-orthogonality during the downstroke, while gure 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 gure 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 gures 3.4
and 3.5, showing the histograms of the non-orthogonality and skewness, respec-
64 Mesh deformation techniques for apping ight
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 gure 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
apping 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 eciency. These techniques are described in section 3.5. But rst, the
RBF mesh motion solver is tested by employing the exing of a two-dimensional
moving boundary.
3.4.3 Flexing of a two-dimensional block
Within the context of the present research, performing a uid-structure-interaction
simulation is too computationally demanding for a full three-dimensional apping
wing. Therefore, the eects of wing exibility was incorporated by dening the
wing exing 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 exing 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 exing, all dened with respect to the initial conguration. 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 apping 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 dened by:
x
t
= A
t
sin(2ft),
= A

sin(2ft),
where, A
t
and A

represent the translation and rotation amplitude vectors, f the


frequency and t the time. The exing of the boundary is dened by combining
two harmonic functions like:
x
f
= A
f
cos(2
x

d
c
2 c
f
) sin(2ft)

d
f
,
where A
f
is the exing amplitude vector, d
c
is the direction vector in-plane of the
exing surface, c
f
the length of the exing surface and d
f
represents the direction
vector of the exing. In this model problem of a moving two-dimensional block, the
amplitudes of both translation and rotation were xed to A
t
= (2.5, 2.5, 0.0) and
A

= (0.0, 0.0, 1.0), respectively. The exing was dened such that the main ex-
ing direction is perpendicular to the exible boundary surface, d
f
= (0.0, 1.0, 0.0)
with a exing amplitude of A
f
= 0.5 which is about 10% of the exible 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 12
block lengths, which is mainly caused by the xed 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 apping ight
(a) (b)
Figure 3.7 Deforming mesh around a exible block using RBF mesh motion. A exible
rectangular block which is translating, rotating and exing using the RBF mesh motion solver with
the globally supported thin plate spline (TPS). The exing is dened by simple harmonic functions.
(a) shows the mesh deformation at t = 0.25T and (b) at t = 0.75T.
3.5 Improving computational eciency
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 nd the coecient vectors and . The
upper diagonal block matrix
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 coecient vectors and . 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 eciency 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 RBFs. Computa-
tional costs for solving the system and evaluation the RBFs 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 apping 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 dierent
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 points, where is
the coarsening factor. More advanced coarsening techniques, based on greedy
algorithms are applied by Rendall & Allen (2008a).
Secondly, the eciency of the RBF method is improved, by the notion that
all outer boundary points are xed 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
dened (Jakobsson & Amoignon, 2007) as
( 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 ith 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 apping ight
between those radii this smoothing function is decreasing from 1.0 to 0.0. When
the RBF evaluation function (3.9) is multiplied by ( x) the corrected RBF eval-
uation is obtained by:
x
in
j
= s(x
in
j
)

( 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 xed 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
dierent 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 dierential 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 nal curve is caused by the choice
of the coarsening ratio to select the moving boundary points. While keeping the
coarsening ratio xed, 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 apping wing simulation (6 apping 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 ow
equations and the RBF mesh motion. Validation showed that a mesh resolution
of 800k provided accurate results for apping wing aerodynamics. Additionally,
gure 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 ecient, it must be noted that mesh coarsening and a smoothing function
are applied for acceleration.
3.5 Improving computational eciency 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 dierent 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 apping wing simulation. The times are subdivided by solving the
RBF mesh motion and the ow 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 dierent 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 suciently high quality.
Dierent 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 ecient to solve using existing iterative solver techniques.
Furthermore, a mesh motion problem leads in general to an ill-conditioned
system, which is dicult 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 dierences between points in
bb
. Despite the pre-conditioning, the
ill-conditioned system cannot be eciently 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 dierent way to improve the eciency 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 signicant gain in com-
puting time. Concerning the coarsening, complex greedy algorithms (Rendall &
70 Mesh deformation techniques for apping ight
Allen, 2008a) may be applied to select the necessary boundary points such that
the eciency is increased even further. Another method to increase the eciency
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 diculty, 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-
ecients,
i
and
i
, which are distributed over all partitions. Currently, this is
being implemented in OpenFOAM

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

, both based on solving a partial


dierential equation. The rst method solves the Laplace equation with a variable
diusion coecient, which is used to control the nal mesh quality. Secondly, the
linear stress equation was modied to include rigid body rotations in order to cope
with the severe mesh deformation present in apping wing simulations. As with
the Laplace equation, the solid body rotation stress (SBR Stress) mesh motion
uses the diusivity, acting as a stiness, to inuence the quality of the mesh. The
diusivity, in both cases, is dened to decrease quadratically with the distance
from the moving boundary.
Besides solving a partial dierential equation the motion of mesh points can
be dened 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
coecients. Using those coecients, 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 nite 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, dierent 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 apping wing with the possibility to
incorporate a exing moving boundary.
Since the RBF mesh motion technique encounters a dense system of equations,
dierent methods are implemented to increase its eciency. 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 justied to
neglect the outer (xed) 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-
ciency of the RBF mesh motion method.
CHAPTER 4
Physical and numerical modelling
of apping foils and wings
In this chapter, the physical and numerical modelling of apping wing and foils is
described. The relevant dimensionless numbers (Strouhal, Reynolds and Rossby
numbers) are identied after writing the Navier-Stokes equations in a rotating
reference frame. To systematically study the aerodynamics around apping wings,
a model planform and kinematic model is dened. The kinematic model, which
describes the wing motion, consists of a rigid body motion appended by a exing,
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 coecients are used in conjuncture
with the lift-to-drag ratio to assess the apping wing performance.
4.1 Introduction
This chapter deals with the physical and numerical modelling of apping wings, in
hovering as well as in forward ight conditions. Animals that y or swim, which are
equivalent from the uid-dynamic perspective (Triantafyllou et al., 1993, Taylor et
al., 2003) at certain scales, undergo signicant interactions with the environmental
uid in which they move. Therefore, for the mathematical analysis of swimming
or ying 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 ight.
74 Physical and numerical modelling of apping foils and wings
The Reynolds number
In order to improve our understanding of biological ows, like the ows around
apping wings or ns, 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 dened as the ratio between
the inertial and viscous forces present in a uid. It is a property of the ow, such
that it identies what kind of propulsive mechanism applies to the apping wing.
For example, when a apping wing operates at a very low Reynolds number, i.e.
Re = O(100), the forces in the ow 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 ow behaves turbulent or laminar, dening implicitly the
complexity of the mathematical model needed to solve the problem. A dierent
approach to dimensionless numbers is to dene 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 dened as the ratio between the convection time over the diusion time,
see chapter 2, which is easier to understand and to apply to a ow problem, such
as the Von K arm an vortex street behind a blu body.
The Strouhal number
Besides the Reynolds number, the Strouhal number plays an important role in
apping wing aerodynamics as well. The typical denition of the Strouhal num-
ber is the apping frequency times apping amplitude divided by a reference ow
velocity. The rst use of the Strouhal number was in the context of the natural
vortex shedding behind a stationary cylinder in a uniform ow. Williamson (1988)
found a universal relation between the Reynolds and Strouhal number based on the
observed vortex-shedding frequency in the laminar ow regime. For apping wings
or oscillating bodies, the Strouhal number can be dened based on the imposed
oscillation frequency and amplitude. For moving bodies and especially apping
wings, the Strouhal number can be very useful. For example, in forward apping
ight, the Strouhal number is proportional to the maximum value of the induced
angle of attack, provided that the wing aps in a stroke plane perpendicular to the
forward velocity (Lentink & Gerritsma, 2003, Taylor et al., 2003, Thaweewat et al.,
2009). This denition of the Strouhal number, based on the apping amplitude, is
closely related to the advance ratio, as dened by Ellington (1984), J = U/4
0
fR,
which is the ratio between the forward and apping distance, travelled by the wing.
Here
0
is half of the total apping angular amplitude, f the apping frequency
and R represents the distance from root to tip of the wing, i.e. the single wing
span. Another denition 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 apping and forward ight. In general
engineering applications, the Strouhal number is commonly used to characterise
4.2 Governing equations for apping wings 75
the vortex shedding, whereas the reduced frequency number is used in apping
wing problems. Additionally, the reciprocal of the Strouhal number is known as
the dimensionless wavelength

= U/f c which is often used in studies concerning


forward apping ight as it seems reasonably intuitive in that it corresponds to
the distance travelled over one apping period, relative to the mean chord.
Equations and other assumptions
As previously discussed, it is appropriate to use dimensionless numbers to study
the eect of apping 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 ow properties are varied
to study their inuence. In accordance with Lentink & Dickinson (2009a), spe-
cic attention is given to the appropriate denition of the governing dimensionless
numbers to investigate the ow around apping wings.
In view of simplicity, the present study deals with a model wing which is a
simplied representation of a ying insect wing operating at Reynolds numbers,
Re = 100, 500, and 1000, corresponding to the operating conditions of fruit ies,
house ies and bumblebees, respectively. Furthermore, only one apping wing is
considered, under hovering as well as forward ight conditions, which allows that
the induced vortical ow can be studied in more detail. This implies that no
interaction between two wings or with the body are included. Nevertheless, the
considered apping kinematics that result in large rotation rates put the current
numerical techniques to a signicant challenge.
In section 4.2, the governing equations are formulated for forward and hovering
apping ight. Secondly, the model wing selection and the denition of the kine-
matic model parameters are described in section 4.3, followed by the dynamical
scaling of apping ight 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 briey dealt with in 4.5. Section 4.6 describes the force and
performance denitions, followed by the conclusions of this chapter in section 4.7.
4.2 Governing equations for apping wings
Concerning apping ight in nature, like the operation of insects and sh, the
ow 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 dened by equation (2.4) and (2.5),
and re-stated here:

u = 0, (4.1)
u
t
+

(uu) =
1

p +
2
u, (4.2)
76 Physical and numerical modelling of apping foils and wings
These equations are derived by analysis of the forces on an innitely small uid 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 ight under
hovering conditions is considered, the initial velocity eld is zero as well as the
boundary conditions at the outer domain boundary. At the boundary of the ap-
ping wing, a no-slip condition (White, 1991) needs to be specied, which means
that the velocity, relative to the wing, has to be zero, in all directions. This is
accomplished by dening the mathematical velocity on the moving boundary to
be equal to the actual wing motion, which moves according to a specic kinematic
model, derived from realistic insect data (e.g. Fry et al., 2003).
Momentum analysis in rotating reference frame
A dierent 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 xed to the apping 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 ow 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 apping wing
will be such that the eective 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 dened as (Ginsberg, 1998,
Baruh, 1999)
u
XY Z
= u
xyz
+ (u
trans
+
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 ight, approximately.
wing
is the angular
velocity of the rotating reference frame, i.e. representing the apping 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
+
wing
r),
where u
WING
is the ow 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 apping 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 dened as
a
ang
=

r,
a
cen
= (r),
a
cor
= 2u
xyz
.
The main parameter in these three dierent accelerations is the angular velocity
[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 dierent acceleration terms,
needs to be related to the apping 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
+ (

r) + ((r)) + (2u) =
1

p +
2
u. (4.3)
This transformed equation describes the momentum balance for a uid 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 dierent
acceleration terms, in addition to the already described Reynolds and Strouhal
numbers. These new dimensionless numbers will become available if the dierent
78 Physical and numerical modelling of apping foils and wings
terms in equation (4.3) are scaled as
u

=
u
U
ref
, t

=
U
ref
c
t,

= c ,

, r

=
r
R
, p

=
p

ref
U
2
ref
,
leading to (dropping the stars for simplicity):
U
ref
f c

Du
Dt
+

Rc
U
2
ref
(

r)+

2
Rc
U
2
ref
((r))+
c
U
ref
(2u) = p+

U
ref
c

2
u,
where U
ref
[m/s] is the reference velocity, f [1/s] the apping frequency, which
is dened as f =
U
ref
c
, where c the average chord length. L [m] is the reference
length, [ rad/s] and

[rad/s] are the average rotational velocity and acceler-
ation, respectively. Furthermore,
ref
[kg/m
3
] is the reference density (constant
in incompressible ows). R [m] is a radius length. In addition to the Reynolds
and Strouhal numbers, other dimensional numbers can be identied, related to
the rotation of the reference frame, which is still attached to the wing, rewriting
gives:
1
St

Du
Dt
+
1
C
ang
(

r) +
1
C
cen
((r)) +
1
Ro
(2u) = p+
1
Re

2
u,
where the additional dimensionless number are respectively dened as
C
ang
=
U
2
ref

R c
, (4.4)
C
cen
=
U
2
ref

2
R c
, (4.5)
Ro =
U
ref
c
. (4.6)
By the rotation amplitude , these dimensionless numbers strongly depend on the
wing kinematics, especially the velocity and acceleration due to the wing rotation.
Therefore, the eect of dierent kinematics and ow features may be related to
these dimensionless numbers.
4.3 Wing shape and kinematic modelling
When investigating the three-dimensional aerodynamics around apping wings,
one may try to simulate the geometry and conditions of specic species, in order
to fully understand them, or to opt for a more generic approach. Previous inves-
tigations of specic insect species have been reported for e.g. a fruit y (Sane &
Dickinson, 2001, Birch & Dickinson, 2003), hawkmoth (Liu & Kawachi, 1998) or
4.3 Wing shape and kinematic modelling 79
a dragony (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 briey 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 exing 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 specic insect features, like the corrugated
wing planform (Luo & Sun, 2005) are of minor inuence on the resulting uid
behaviour. The three-dimensional elliptical planform is shown in gure 4.2 in
comparison to a more realistic representation of the fruit y 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 = /4 for c = 1.0, both dening the
planform. Figure 4.2 shows the wing planform and corresponding parameters for
a fruit y wing and a ellipsoidal model wing.
Radius of gyration
In order to dene a sound framework of comparison for dierent three-dimensional
and two-dimensional simulations, it is necessary to dene all reference parameters,
80 Physical and numerical modelling of apping 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 y 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 y (a) and ellipsoidal model wing (b). The planform is dened 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 dene 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 eect 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 apping, 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 dierent kinematic
models is concerned (Bos et al., 2008). The radius of gyration, R
g
is dened 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 ow induced by apping
wings with dierent aspect ratios and found that the radius of gyration provided
a reliable framework for force comparison when the apping 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 ow around apping 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
(t)
(t)
(t)
O
R
root
R
tip
start downstroke
start upstroke
mid-stroke
Stroke plane
Figure 4.3 Schematic illustration of the governing apping angles. Flapping wing motion
is governed by three angles, (t) corresponds to the stroke variation, (t) to the geometrical angle of
attack and deviation (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 ow around apping wings, the eect of dierent kinematic models on the
aerodynamic performance can be evaluated. Besides the use of an idealisation of
the insect wing planform, simplied kinematics was used, like harmonic motion,
to study dierent 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 dened by the variation of three independent
attitude angles, see gure 4.3. In this three-dimensional model the three degrees
of freedom of the wing motion are dened as the apping angle, (t), in the mean
stroke plane, the angle of attack, (t), with respect to the horizontal plane and
the deviation (t), which is measured in a plane normal to the stroke plane, as is
shown in gure 4.3. The deviation may be used to create a gure-of-eight pattern
which is present in realistic fruit y kinematics (Fry et al., 2003).
In literature, dierent kinematic models have been studied, from purely har-
monic motion to complex realistic fruit y kinematics. Using two-dimensional
numerical techniques, Bos et al. (2008) showed that the apping wing perfor-
mance may be inuenced by the specic 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 y kinematics (Fry et al., 2003). Figure 4.4(b)
shows that the realistic fruit y kinematics is characterised by an asymmetry in
apping angle, and angle of attack. Furthermore, the angle of attack shows a dip
(at t=0.1T), lowering the eective angle of attack. In addition, the shape of the
fruit y angle of attack clearly shows a plateau of constant value, which is ex-
82 Physical and numerical modelling of apping foils and wings
ploited by Sane & Dickinson (2001, 2002), Lehmann et al. (2005) using a Roboy.
Additionally, the realistic fruit y kinematics in characterised by the presence of
deviation which may result in a gure-of-eight pattern (Shyy et al., 2008b)
In order to illustrate the variation of the motion angles, the denition of the
harmonic model is described accordingly. The apping angle, (t), is described by
a cosine function. The geometric angle of attack, (t), is dened by a sine function
with respect to 90

and the deviation angle, (t), is given by a pure sine:


(t) = A

cos(2ft),
(t) =

2
A

sin(2ft), (4.11)
(t) = A

sin(2ft).
Here, A

is the apping amplitude, which is dened from stroke reversal to mid-


stroke. Remind that this amplitude is half times the value used in literature (e.g.

0
in Ellington, 1984). f is the apping frequency and A

represents the amplitude


of the angle of attack with respect to /2, which is the initial position under
hovering conditions. A

is the amplitude of the deviation angle which causes


the so-called gure-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).

(t) = A

(2f) sin(2ft),
(t) = A

(2f) cos(2ft), (4.12)

(t) = A

(2f) cos(2ft).
In general, this harmonic model is crude but fairly reasonable representation of
the apping motion of a fruit y. Also, simplied (harmonic) kinematics may
be interesting for Micro Air Vehicles (MAV) implementation. As will be shown in
chapter 5 (Bos et al., 2008), the apping wing performance may be signicantly in-
uenced by modications 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 dening a
non-zero harmonically variation of (t).
4.3.3 Modelling of active wing exing
In addition to the rigid body motion as shown in gure 4.3 it is possible to dene
an extra displacement concerning exing of the wing. Since a full uid structure
interaction (FSI) simulation is too expensive and beyond the scope of the current
research, a exing displacement of the wing surface is dened. The exing dis-
placement is dened with respect to the initial wing position and can be written
4.3 Wing shape and kinematic modelling 83
t/T [-]

]
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 [-]

]
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 y kinematics
Figure 4.4 Comparison of the harmonic and fruit y kinematic models. (a) shows the
least complex kinematic model, representing pure harmonic variations of the apping angle (t) (),
angle of attack (t) () and deviation (t) (). The variation of realistic fruit y kinematics (Fry
et al., 2003) is shown in (b), which is the most complex kinematics available. The realistic fruit y
kinematics is characterised by an asymmetric variation of apping angle, an extra bump and a degree
of trapezoidal in angle of attack and the presence of deviation.
as
x(t) = A
f
cos
_
2x
0

f
_
sin(2ft),
where the rst cosine function denes the wing shape and the sine function rep-
resents the time-variation. A
f
is the exing amplitude vector and x
0
the location
of the initial boundary points. The cosine shaped wing exing is dened by
f
,
which corresponds to the cosine ratio, i.e.
f
= 0.5 means that the shape is like a
half cosine function. Figure 4.5 shows a plunging airfoil incorporating exing for

f
= 0.25 (a) and
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 exing (deformation) component. The translational component is not used
in the current three-dimensional simulations, which is limited to hovering and for-
ward ow conditions with stationary position of the rotation origin. However, in
the two-dimensional simulations, the apping motion is dened 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-dened wing exing have been performed.
In general, the wing kinematics is calculated beforehand and applied to the
numerical ow 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 apping foils and wings
(a) Quarter cosine shape,
f
= 0.25 (b) Half cosine shape,
f
= 0.5
Figure 4.5 Illustration of a exing two-dimensional airfoil. The two-dimensional exing airfoil
is modelled by a time-varying cosine shape. This airfoil shape is either a quarter cosine,
f
= 0.25
(a) or a half cosine,
f
= 0.5 (b). Both gures (a) and (b) show the upstroke (left) and downstroke
(right).
tion and exing.
Translation
The displacement array of the boundary points, which is due to translation is
obtained from
x(t) = A
t
sin(2f
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 denition is used when the
wing needs to move according to an ordinary rigid body motion (Ginsberg, 1998).
When a apping motion is desired, a cosine function is used to dene 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 dierent components, due to a rotation around the X-, Y -
and Z-axis. According to (Ginsberg, 1998, Baruh, 1999), these three matrices are
dened as
R
X
(t) =
_
_
1 0 0
0 cos((t)) sin((t))
0 sin((t)) cos((t))
_
_
, (4.15)
R
Y
(t) =
_
_
cos((t)) 0 sin((t))
0 1 0
sin((t)) 0 cos((t))
_
_
, (4.16)
and
R
Z
(t) =
_
_
cos((t)) sin((t)) 0
sin((t)) cos((t)) 0
0 0 1
_
_
. (4.17)
Here, the rotation around the X-, Y - or Z-axis correspond to the deviation angle,
(t), apping angle, (t), and the angle of attack, (t), respectively. Using the
motion convention, shown in gure 4.3, it is clear that a hovering wing aps in the
X-Z plane, accelerating the uid 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 nd 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 apping wing is able to perform a
exing motion as well. This wing exing is dened as
x
ex
(t) = A
f
cos(2x
0
) sin(2f
f
t),
where A
f
is the exing amplitude vector, f
f
the exing frequency and x
0
repre-
sents the initial boundary points of the apping wing at t = 0. This denition
permits exing 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 exing, the following is obtained:
x
b
= x
trans
+ x
rot
+ x
ex
.
86 Physical and numerical modelling of apping foils and wings
This total displacement eld 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 apping wing is considered, it is desirable to start the numerical simulation at
maximal apping angle, in order to minimise the initial acceleration of the mesh.
Therefore, it was chosen to use a rigid body motion until the maximal apping
angle was reached, in general this occurs at T/4, where T is the apping 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 apping wings
In section 4.2 dierent dimensionless numbers, concerning rotational motion, were
identied in order to scale the Navier-Stokes equations, governing uid ow. To de-
ne these rotational dimensional numbers, C
ang
= U
2
ref
/

R c, C
cen
= U
2
ref
/
2
R c
and Ro = U
ref
/ c, it is important to identify the average rotation amplitude,
and its time-derivative

for the particular kinematics implemented:
=
1
T
_
T
0
|

|dt =
1
T
_
T
0
| A

(2f) sin(2ft)|dt = 4A

f, (4.20)

=
1
T
_
T
0
|

|dt =
1
T
_
T
0
| A

(2f)
2
cos(2ft)|dt = 8f
2
A

= 2f, (4.21)
where the apping velocity

was taken from equation (4.12). As shown, evaluation
gives = 4A

f and

= 2f. Besides the angular velocity , angular accelera-
tion

and average chord length c it is necessary to dene 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 nd U
ref
by multiplying the expression for
(4.20) by R
g
:
U
ref
= 4A

fR
g
. (4.23)
These relations hold if the wing is the only driving force behind the resulting ow
velocity, but in case of forward ight conditions, there is an additional free-stream
4.4 Dynamical scaling of apping wings 87

90

tan
1
(2Stsin)
2A

sin
Figure 4.6 Schematic illustration of the kinematic parameters in forward ight.
velocity, U

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

= 0) and forward ight


conditions:
U
ref
= U
R
g

_
U
2

+ (4A

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

4A

fR
g
=

4A

,
where the reduced frequency (Ellington, 1984),

, is given by

= U

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

is also known as the dimensionless wavelength, which is illustrated in gure 4.6.


The dimensionless amplitude at the radius of gyration R
g
is dened as
A

=
A

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

=
4A

fR
g
c

.
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 apping foils and wings
area by the wing root:
A
swept
= 2A

(R
2
tip
R
2
root
),
= 2A

(R
tip
R
root
) (R
tip
+R
root
),
= 2A

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 identied 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 eciency (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
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
innite; for a rotating wing, Ro is nite. If Ro is varied, the eect of dierent
rotation origins can be investigated, from nearly translating to strongly revolving.
However, the determination of Ro would be easier if dened 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 sh (Lentink, 2008), generating thrust
by moving the uid. If the wing planform is complex, like in real insects, it may be
dicult 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 apping wing simulations. Chapter 5, 6
and 7, which are dealing with respectively two-dimensional hovering, forward and
three-dimensional hovering ight, explain the specic 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
dierential equations are solved. It is necessary to use a suciently 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 ecient
4.6 Denition of force and performance coecients 89
X
Y

left

wing

right
(a) Two-dimensional
Z X
Y

wing

left

right

bottom

top

front

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 apping wing simulations uses the boxed
topology (b).
to use an O-type mesh, which is shown in gure 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,
wing
, and the cylinder-shaped outer
boundaries,
left
and
right
. The outer boundary is split into
left
and
right
in
order to be able to specify inow and outow boundary conditions, which are
necessary (Wesseling, 2001) when forward ight conditions are simulated. When
simulating hovering ight, the apping wing still induces a signicant amount of
downwash, leading to a small inow and outow 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 gure 4.7(b). Depending on the apping congurations,
the outer boundaries,
left
,
right
,
bottom
,
top
,
front
,
back
, are set to inow,
outow or symmetry planes.
4.6 Denition of force and performance
coecients
Besides ow eld 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 apping foils and wings
PSfrag
X
Y
Z
R
g
F
X
F
Y
F
Z

Figure 4.8 Forces on a general three-dimensional apping wing. The forces are dened in the
three-dimensional inertial reference frame. Depending on the type of motion, hovering or forward ight,
two- or three-dimensional, the lift and drag are constructed from the forces in X-, Y - or Z-direction.
is important to properly dene force and performance coecients. In this section,
the general determination of the force coecients is explained in combination with
performance characteristics.
Forces
In general three dimensions, a two-dimensional derivation is trivial, the denitions
of the forces, F
X
, F
Y
and F
Z
are shown in gure 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

u
n
dS,
where F
tot
[N] is the total force vector, S [m] the wing surface and dS [m] represents
an innitely small surface area element, p [N/m
2
] is the pressure and [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 this forms the wall
shear stress.
Two dierent force denitions can be dened, depending on the type of three-
dimensional motion, hovering or forward apping ight. In hovering conditions,
when the main apping direction is around the Y -axis, as shown in gure 4.3, the
lift force F
lift
is dened in the vertical direction and equal to F
Y
. The drag force
F
drag
, however, is dened in opposite direction of the apping wing motion. If
three-dimensional apping in hovering ight 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((t)) F
Z
(t) cos((t)),
where (t) is the apping 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 Denition of force and performance coecients 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 justied to neglect this spanwise force, also because it has little relevance to
performance.
In forward ight conditions, the same inertial reference frame is used as shown
in gure 4.3. The main apping direction is still around the Y -axis, but the
direction of the uniform ow 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 apping ight
the lift force is dened 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 dened as:
C
D
=
F
drag
q S
, C
L
=
F
lift
q S
,
where C
D
and C
L
are the drag and lift coecients and S the wing surface. The
mean dynamic pressure q is dened as:
q = 1/2U
2
ref
= 1/2
1
T
_
T
0
_
(U

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

, which is non-zero in for-


ward apping ight, the apping velocity U
ap
and the deviation velocity U
dev
,
perpendicular to the horizontal plane (in hovering ight).
Performance
The force coecients are the major parameters used to assess the inuence of the
dierent wing motion models. In addition, the ratio between time-averaged lift
coecient C
L
and time-averaged drag coecient 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 apping period, while for the drag the absolute
values are used for averaging. The drag is opposed to the apping motion, such
that the sign ips 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 coecients of the dierent kinematic models are matched,
92 Physical and numerical modelling of apping foils and wings
the lift-to-drag ratio is corrected for any dierences in lift. Therefore, a high lift-to-
drag ratio eectively 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
apping wings and two-dimensional apping airfoils. Two important dimensionless
numbers were identied, the Reynolds and Strouhal number. These numbers are
common in general uid ow, but for apping ight the denition has been slightly
changed. In order to analyse the ow around apping 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 apping wings at
the scale of insects, a model wing planform has been dened with an ellipsoidal
planform. Using that planform, the radius of gyration can be easily obtained,
which is used to dynamically scale the wing kinematics. Dierent apping wing
kinematic models are briey addressed, from realistic fruit y kinematics to a
fully harmonic model. In addition to the rigid body rotations, a deformation of
the wing is dened in order to study the eects of wing exing. This exing motion
is dened 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 apping foil as well as three-dimensional apping wing simulations.
Analysis of the vortical ow around the wings and foils is primarily performed
by plotting the force coecient. The lift is dened in vertical direction, while
the drag is opposed to the apping velocity. The corresponding coecients 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 apping
wing performance.
CHAPTER 5
A 2D investigation of the inuence
of wing kinematics in hovering
ight
J. Fluid Mech. (2008), vol. 594, pp. 341-368.
The inuence of dierent 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, simplied models are compared with averaged
representations of the hovering fruit y wing kinematics. With increasing com-
plexity, a harmonic model, a Roboy model and two more realistic fruit y 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
coecient 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 y
wing kinematics result in forces that dier signicantly from those resulting from
the simplied wing kinematic models. In addition, light is shed on the eect of
dierent specic characteristic features of the insect wing motion. The angle of
attack variation used by fruit ies increases aerodynamic performance, whereas
the deviation is most likely used for levelling the forces over the cycle.
94 Inuence of wing kinematics in two-dimensional hovering ight
5.1 Introduction
In order to investigate the full ow around a three-dimensional apping wing,
two-dimensional simulations are performed to get insight in the complicated ow
structures. This chapter deals with the evolution of the forces and the wake
originated by a apping foil in hovering conditions.
5.1.1 Similarity and discrepancy between two- and
three-dimensional ows
In a recent paper Wang et al. (2004) compared three-dimensional Roboy results
with two-dimensional numerical results. This showed that two-dimensional simu-
lations are useful to obtain a better understanding of the ow 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 nite-span wings
at low Reynolds numbers. They observed that the apping 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 ow, two-dimensional analysis has often been applied to obtain insight
into the aerodynamic eects of choices in kinematics, airfoil cross-section, Reynolds
numbers, etc. Wang et al. (2004) conrmed that the similarities between two- and
three-dimensional approaches are sucient to warrant that a reasonable approxi-
mation of insect ight 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
& Gotz, 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 justication, 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 Inuence of kinematic modelling
The relevance of (experimental or numerical) simulations of insect ight 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
signicantly inuences the mean force coecients and its distribution. Addition-
ally, Hover et al. (2004) showed that modelling the angle of attack inuences the
apping foil propulsion eciency to a large extent. This illustrates the appreciable
eects which details of the wing kinematics, like parameter values and stroke pat-
terns, may have on ight performance. It further emphasises the need to critically
assess the inuence of kinematic model simplications.
In literature, dierent kinematic models have been employed to investigate the
aerodynamic features of insect ight. 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 apping 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 ow solutions which are strongly related to the aero-
dynamic eciency. Lentink & Gerritsma (2003) varied airfoil shape with amplitude
and frequency xed at values representative to real fruit ies. They concluded that
the airfoil choice is of minor inuence, 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 & Gotz, 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 ight. Both studies emphasised the importance of angle of at-
tack modelling to inuence the propulsive eciency. Slightly more complex fruit
y kinematic models were used by Dickinson et al. (1999) and Sane & Dickinson
(2001) with their Roboy. Based on observation of true insect ight, 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 Roboy kinematic model. Using these models, the eect of ampli-
tude, deviation, angle of attack and the timing of the latter were explored.
In the present study, dierent models from literature were considered, both the
pure harmonic and the Roboy model, in order to investigate their inuence on the
aerodynamics. Furthermore, the results were compared with more realistic fruit
y kinematics obtained from the observation of free ying fruit ies (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 eect of the available
models as a whole. This leads to better insights in the consequences of simpli-
cations in kinematic modelling, which is of great importance to both experiments
and numerical simulations. Also, it can reveal the importance of certain specic
96 Inuence of wing kinematics in two-dimensional hovering ight
features of the stroke pattern, in relation to aerodynamic performance.
This study considers four dierent 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 ow. In brief, the rst 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 Roboy
at UC Berkeley (presently CalTech). The third model is a representation of the real
kinematics used by a hovering fruit y (Drosophila Melanogaster), based on data
measured by Fry et al. (2003). Finally, the fourth model is a slightly simplied
version of the latter, observed fruit y model. All these kinematic models are
dynamically scaled at a Reynolds number of Re = 110 which corresponds to
the ight conditions of the fruit y. In addition, these kinematic models are
constructed such that their mean quasi-steady lift coecients are comparable such
that our performance comparison is justied. This basis of comparison is veried
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
dierent kinematic models are treated in section 5.4 and concluding remarks are
given in 5.5.
5.2 Numerical simulation methods
The dierent kinematic models are implemented in the commercial ow 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 veried using harmonically
moving wings.
5.2.1 Flow solver and governing equations
To simulate the ow around moving wings with pre-dened motions the commer-
cial CFD solver Fluent

was used. The two-dimensional time-dependent Navier-


Stokes equations are solved using the nite volume method, assuming incompress-
ible ow which is justied since the Mach number of apping insect ight is typi-
cally O(10
3
) (see Brodsky, 1994). The mass and momentum equations are solved
in a xed 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 ow is assumed to
be laminar. Henderson (1995) and Williamson (1995) showed that for circular
cylinders, the transition from laminar to turbulent ow occurs at Re = 180 5,
which supports this assumption. Therefore, the transient incompressible laminar
5.2 Numerical simulation methods 97

1
x
y

1
x
y
0.4 0.42
0.56
0.58
Figure 5.1 O-type mesh topology with
boundary conditions on
1
,
2
and
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 ow around the moving airfoils, an O-type computational
domain is used, which is shown schematically in gure 5.1. The computational
domain is divided into two parts:
1
and
2
for the inner and outer mesh respec-
tively. The body surface
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
3
is located at 25L such that the inuence of the far eld 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 ight, such that a free-stream is absent, a symmetry boundary
condition was applied at
3
for numerical reasons. The inuence of this symmetry
condition has been investigated and found to be suciently 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 gure 5.2) a conformal mapping was applied (see
Wang, 2000b). The intermediate interface
2
divides the mesh into two separate
elds, corresponding respectively to the inner conformal mesh (
1
) and the outer
mesh (
2
). The complete inner mesh moves according to the wing kinematics,
while re-meshing takes place in the outer eld
2
. Since re-meshing occurs at
a distance of 25 to 30 body lengths away from the wing, the ow around the
wing is not aected by the mesh regeneration. The described computational setup
was thoroughly validated using the ow around stationary and moving circular
98 Inuence of wing kinematics in two-dimensional hovering ight
t +dt
t

ref

ref
+d
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 rst cell at the
wing surface varies between 2% and 50% of the wing thickness at the leading-edge
and in the middle of the prole 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 apping
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 inuence of the relative cell displacements. There-
fore, the motion of a reference cell was investigated, which is illustrated for the
rotational motion in gure 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 dened as

r
=

ref
and

y
=
y
y
ref
=
2f
e
A
e
Nt
y
ref
.
Here corresponds to the angular displacement of the reference cell, while
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
dierences in drag coecients remaining below 5%. The computational eorts 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 ight.
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 coecients between the present simulations and Wang
et al. (2004). Comparison of lift (a) and drag (b) coecients 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 coecients 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 gure 5.4(a) is dened
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 nds a larger lift and drag which is probably
the result of dierent numerical dissipation properties of both codes. The mean
lift and drag coecients are 0.84, 1.47 for our simulation, compared to 0.82, 1.44
obtained by Wang et al. (2004), which is a dierence of only 2% in lift and drag and
therefore, the computations were considered to be suciently accurate. Moreover,
within the context of comparing results of dierent stroke patterns, the present
numerical method is proved to be accurate.
Further details of the validation and verication 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 dene an equivalent two-dimensional geometry, while
100 Inuence of wing kinematics in two-dimensional hovering ight
Figure 5.5 Illustration of the main motion directions. (t) corresponds to the stroke variation,
(t) to the geometrical angle of attack and (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 denitions 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 inuence of dierent
kinematic wing motion models on the aerodynamic performance. The dierent
kinematic models are illustrated using the Roboy experimental set-up, shown
in gure 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 dened
as the angular displacement in the mean stroke plane, the angle of attack ,
with respect to the horizontal plane and the deviation from the horizontal plane
, as is shown in gure 5.6. The deviation causes a gure-of-eight pattern which
is present in real fruit y 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 dened at a representative
spanwise location such that the motion is conned 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 dierent 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

Figure 5.6 Force denition on the two-dimensional airfoil.


location (Lentink & Gerritsma, 2003) was used. Considering that the local velocity
of each cross-section varies during apping, 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
dened is the reference length, L
ref
, based on the mean chord length. A denition
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 =
R
g
c
, y =
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 dened 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 apping of the wings induces highly unsteady ow the relevant ow and
motion parameters have to be scaled dynamically. The period of the motion is
used to average the relevant ow velocity (Lentink & Gerritsma, 2003):
U =
1
T
_
T
0
_
u
2
+v
2
dt. (5.2)
102 Inuence of wing kinematics in two-dimensional hovering ight
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

=
fR
g
c


_
1
0
_
(

t
)
2
+ (

t
)
2
(5.3)
and
St =
fc
U
=
c
R
g

1
_
1
0
_
(

t
)
2
+ (

t
)
2
. (5.4)
Here f = 1/T is the frequency, and 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 (t) and deviation (t). The Strouhal number St is not to be varied
independently. We xed the Reynolds number to Re = 110.
5.3.3 Force and performance indicators
The denition of the drag and lift forces is shown in gure 5.6. The lift is equal
to the vertical force F
y
, while the drag is taken equal to the horizontal force
F
x
, dened 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 dened as
C
D
=
F
x
q c
, C
L
=
F
y
q c
,
where C
D
and C
L
are the drag and lift coecients. The mean dynamic pressure
q is dened as
q = 1/2U
2
= 1/2
1
T
_
T
0
_
_
x
t
_
2
+
_
y
t
_
2
_
dt,
where the integration is evaluated over one apping cycle. The force coecients
are the major parameters used to assess the inuence of the dierent wing motion
models. In addition, the ratio between time-averaged lift coecient, C
L
, and time-
averaged drag coecient, 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
coecients of the dierent kinematic models are matched, the lift-to-drag ratio is
corrected for any dierences in lift. Therefore, a high lift-to-drag ratio eectively
means low drag at equal lift.
5.3 Modelling insect wing kinematics 103
5.3.4 Dierent wing kinematic models
Since the main purpose of this study is to investigate the inuence of wing kinemat-
ics on the aerodynamic performance during hovering fruit y ight, four dierent
kinematic models, with dierent degree of complexity, have been analysed. Two
of these models, the pure harmonic motion and the Roboy experimental kine-
matics have appeared in literature. The third model represents the actual fruit y
kinematics as observed in experiments and the last one was a modication of the
latter, chosen to investigate the eect 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 coecient (Bos et al., 2008, appendix
A). Although according to Sane & Dickinson (2001) the mean drag is strongly
inuenced by the unsteady ow physics, which are not fully present in the quasi-
steady theory, the mean lift coecient is predicted well using this theory. Using
quasi-steady theory, dierent kinematic models were constructed, such that their
quasi-steady lift coecients 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 dierence 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 coecient of the numerical simulations are reasonably well matched for
all models, which provides an a-posteriori justication of our choices for the model
parameters.
The characteristic shapes of each model are described. Subsequently they are
used to investigate the inuence 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 y may benet from kinematic features which are absent in the
simpler models, and reveals the relevance of including these aspects in theoretical
models.
The rst 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 gure 5.7(a). The second
model takes the wing kinematics as used in the Roboy model (Dickinson et al.,
1999). In gure 5.7(b) it is shown that the ip 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 gure 5.7(c),
is derived from measurements on real fruit ies (Fry et al., 2003) and is therefore
considered as the most realistic fruit y kinematic model. This model does include
the deviation which results in a gure-of-eight pattern. Neither the displacement,
angle of attack nor deviation is symmetric during the apping period.
In order to investigate the fact that the observed fruit y kinematics lacks an
exact symmetry in the wing stroke pattern, a symmetrical model was constructed,
referred to as the symmetric fruit y model, displayed in gure 5.7(d). Within this
model the motion is identical for the downstroke and upstroke. Like the realistic
104 Inuence of wing kinematics in two-dimensional hovering ight
t/T [-]

]
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 [-]

]
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 [-]

]
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 [-]

]
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 dierent kinematic models. (a) Harmonic model. (b)
Roboy model. (c) fruit y model. (d) simplied fruit y model. : displacement angle , : angle
of attack , : deviation angle .
fruit y model this symmetric model includes a time-dependent deviation such
that the observer sees a gure-of-eight pattern of the wing. Neither of those
last two realistic kinematic models can be described by using simple analytical
functions without losing signicant detail.
When comparing the motion parameters, , and for each model it becomes
possible to identify certain important dierences. The Roboy initially has a
larger gradient in time of the angle of attack compared to the harmonic case, see
gure 5.7(a) and (b). During translation from about t = 0.1T to t = 0.4T the
angle of attack attens at a value of almost 40

. This trapezoidal shape of is


characteristic for the Roboy and may be inuencing the performance. Although
the Roboy model clearly shows similarities with the fruit y models the latter
has some typical additional features. The most obvious peculiarity of the realistic
fruit y models is the extra bump in angle of attack just after stroke reversal,
5.4 Results and Discussion 105
compared to the Roboy (gure 5.7(b) and (c)). It follows the same high angular
velocity, but instead of attening , the fruit y wing descends to the bump.
After the bump the angle of attack more or less matches the plateau found in
Roboy but starts to increase earlier. During stroke reversal the gradient of
matched the harmonic model closer than the Roboy with its high gradients.
The harmonic and Roboy models lack deviation, so no gure-of-eight is
present. The deviation of the fruit y model is asymmetric during the complete
cycle, but also during each half stroke (gure 5.7(c)). This is likely to inuence
the performance since the eective 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 y is averaged to derive the
simplied fruit y model, see gure 5.7(d). This last model is used to investigate
the inuence 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
Roboy 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 y models are characterised
by a bump in angle of attack and the presence of deviation. Results of two
comparative studies were presented. The rst is an overall comparison of the
complete kinematic models, which is described in section 5.4.1. In the second
study the eect of the characteristic features identied above, are considered in
more in detail. In order to assess the eect 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 modied 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 coecients are given for the four complete models,
the harmonic model, the Roboy model, the realistic fruit y model and the
simplied fruit y model. The mean drag, for each half-stroke, and lift coecients
are given, as well as the average lift-to-drag ratio, which characterises aerodynamic
performance.
The dierences of the obtained mean lift coecient are signicantly smaller
than the dierences in lift-to-drag ratios. Therefore, the conclusions on the per-
formance comparison are considered to be signicant.
The mean drag for the harmonic and Roboy models is substantially higher
106 Inuence of wing kinematics in two-dimensional hovering ight
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 coecient histories of the baseline kinematic models. : harmonic model,
: Roboy model, : realistic fruit y model, : simplied fruit y 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 coecient histories of the baseline kinematic models. : harmonic model;
: Roboy model, : realistic fruit y model, : simplied fruit y 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%
Roboy 1.417 8.0% 2.466 2.448 0.577 49%
realistic fruit y 1.540 baseline 1.387 1.335 1.132 baseline
simplied fruit y 1.454 5.6% 1.012 1.596 1.115 1.5%
Table 5.1 Time-averaged force coecients 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) Roboy model,
(c) realistic fruit y model, (d) symmetric fruit y model.
108 Inuence of wing kinematics in two-dimensional hovering ight
(a) harmonic model (b) realistic fruit y model
Figure 5.11 Vorticity contours around a apping airfoil. Vorticity contours are shown for
t = 0.1T (blue: clock-wise, corresponding to negative vorticity values).
compared to the fruit y models. This is also illustrated in gure 5.8 and 5.9 (lift
and drag histories) and gure 5.10 (force vectors). Figure 5.11 shows the vorticity
contours of the realistic fruit y model compared with the harmonic model. It can
be seen in gure 5.7(a) that the eective angle of attack is higher in the harmonic
case, compared to the realistic fruit y model, gure 5.7(c). Therefore, the mean
drag contribution of the leading-edge vortices (LEV) is higher. The decrease in
eective angle of attack in the realistic fruit y model is also enlarged by the
presence of the bump. This drag increasing eect is even larger in case of the
Roboy model due to the trapezoidal angle of attack. The sawtooth shaped
Roboy displacement could possibly play an important role as is discussed in the
next section. The dierent kinematic patterns are also illustrated in gure 5.10,
which shows the resultant force vectors during a full stroke for those baseline
kinematic models.
Furthermore, the mean drag coecient of the simplied fruit y 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 coecient
obtained with the realistic fruit y model.
When comparing the lift-to-drag ratios in table 5.1, it can be observed that
within the model assumptions, the fruit y models perform better than the less
complex models. Compared to the harmonic model, the realistic fruit y model
results in a signicant decrease in drag of 29% at comparable lift. The dierence
with the Roboy model is even larger, 49%. These performance increases are the
result of the lower drag coecients in both fruit y models due to certain benecial
5.4 Results and Discussion 109
kinematic features. The current results provide insight into the eects of certain
specic kinematic features. However, one has to be cautious when extrapolating
these results to real ying ies since in reality not every apping period displays
exactly the same kinematic prole. Next, the individual inuences of the dierent
interesting kinematic shapes are studied.
5.4.2 Kinematic features investigation
Inuence of sawtooth displacement used by the Roboy
The sawtooth shaped displacement of the Roboy is investigated in isolation
to assess its inuence on the force histories and the aerodynamic performance.
Therefore, the purely harmonic model was appended with the Roboy 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 gure 5.13. From gure 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 signicantly
larger for the sawtooth case, see gure 5.13(b). This also explains the larger
mean drag compared to the harmonic model which can be read from table 5.2.
In gures 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 gure 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 reected 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.
Inuence of trapezoidal angle of attack used by the Roboy
In combination with the sawtooth displacement, the Roboy uses a trapezoidal
shape for the angle of attack. In order to determine the eect 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 gure 5.12(a)
for the force vectors. The lift and drag coecients are plotted in gure 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 Inuence of wing kinematics in two-dimensional hovering ight

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 , (b)
harmonic model with trapezoidal , (c) harmonic model with extra bump , (d) harmonic model
with deviation .
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 coecients. Lift (a) and drag (b) histories to investigate the inuence
of the sawtooth displacement compared to the harmonic model. : harmonic ,,; : harmonic ,
and Roboy .
(a) harmonic model (b) harmonic model with sawtooth displacement
Figure 5.14 Vorticity contours around a apping airfoil. Vorticity contours are shown for
t=0.1T (blue: clock-wise, corresponding to negative vorticity values)
112 Inuence of wing kinematics in two-dimensional hovering ight
kinematic model C
L
C
Ddown
C
Dup
C
L
/C
Dave
harm. , and 1.483 (baseline) 1.848 1.839 0.804 (baseline)
harm. , + Roboy 1.366 (7.9%) 2.240 2.250 0.608 (24.3 %)
harm. , + Roboy 1.351 (8.9%) 2.302 2.733 0.537 (33.3 %)
harm. , + fruit y. 1.483 (0.0%) 1.221 1.969 0.930 (+15.6 %)
harm. , + fruit y. 1.323 (10.8%) 1.807 1.776 0.738 (8.2 %)
Table 5.2 Time-averaged force coecients to investigate the inuence 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 coecients. Lift and drag histories to study the inuence of the
trapezoidal angle of attack compared to harmonic model. : harmonic ,,; : harmonic , and
Roboy .
5.4 Results and Discussion 113
(a) harmonic model (b) harmonic model with trapezoidal
Figure 5.16 Vorticity contours around a apping 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 gure 5.15 it is clear that at the beginning of a stroke the lift peak of the
trapezoidal case is larger. Using gure 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 gure 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 gure 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 signicant performance decrease of 33.3% due to the trapezoidal angle of
attack variation, see table 5.2.
Inuence of extra bump in angle of attack used by the fruit y
The fruit y models are subject to an extra bump in angle of attack. To make
comparison plausible the symmetric bump variation used in the simplied fruit
y model is used to compare results with the harmonic model. Figure 5.12(c)
114 Inuence of wing kinematics in two-dimensional hovering ight
(a) harmonic model (b) harmonic model with trapezoidal
Figure 5.17 Vorticity contours around a apping 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 gure 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 signicantly. However, the drag during the downstroke
is very much aected. 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 gure 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
gure 5.19(a) and (b) are compared. The decrease in eective angle of attack as
a result of the bump is considerable compared to the harmonic case. The same
was found for the Roboy 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.
Inuence of wing deviation used by the fruit y
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 coecients. Lift (a) and drag (b) histories to study the inuence of
the extra bump in angle of attack. : harmonic ,,; : harmonic , and fruit y .
(a) harmonic model (b) harmonic model with extra bump in
Figure 5.19 Vorticity contours around a apping airfoil. Vorticity contours are shown for
t=0.1T (blue: clock-wise, corresponding to negative vorticity values)
116 Inuence of wing kinematics in two-dimensional hovering ight
(a) harmonic model (b) harmonic model with extra bump in
Figure 5.20 Vorticity contours around a apping airfoil. Vorticity contours are shown for
t=0.6T (blue: clock-wise, corresponding to negative vorticity values)
realistic and simplied fruit y model. This deviation causes a gure-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 eective angle of attack is highly aected. This motion
perpendicular to the stroke plane is illustrated in gure 5.12(d) which also shows
the force vectors.
Figure 5.21 shows the force coecients during one apping period with de-
viation added to the harmonic model. The mean lift and drag are not strongly
inuenced by the deviation, see table 5.2. The mean lift is decreased by 10.8% and
the mean drag is almost not aected by the presence of deviation, about 2%4%
dierence in both strokes. It is also revealed that the force distributions remain
symmetric.
The large inuence 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 ow dynamic mechanism for this is shown in the vorticity visualisations
of gure 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 inuence of the deviation is relatively large since the deviation increases the
eective 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 eective 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 coecients. Lift (a) and drag (b) histories to study the inuence of
the deviation compared to harmonic model. : harmonic ,,; : harmonic , and fruit y .
(a) harmonic model (b) harmonic model with deviation
Figure 5.22 Vorticity contours around a apping airfoil. Vorticity contours are shown for
t=0.1T (blue: clock-wise, corresponding to negative vorticity values).
118 Inuence of wing kinematics in two-dimensional hovering ight
(a) harmonic model (b) harmonic model with deviation
Figure 5.23 Vorticity contours around a apping 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 LEVs of comparable strength for both cases.
Summarising, the deviation is levelling the force distributions while the mean
lift and drag are almost unaected. This leads to the suggestion that a fruit y
may use the deviation to level the wing loading over a apping cycle. Three-
dimensional studies are needed to investigate to what extent this eect is also
present in real insect ight.
5.5 Conclusions
The eect of wing motion kinematics on the aerodynamic characteristics of hov-
ering insect ight was investigated by means of two-dimensional numerical ow
simulations. The results of the present two-dimensional study may provide useful
insights in the understanding of real three-dimensional insect ight (Wang et al.,
2004).
Four dierent kinematic models, with dierent complexity, have been anal-
ysed using two-dimensional time-dependent Navier-Stokes simulations. Two of
these models, pure harmonic motion and Roboy experimental kinematics have
appeared in literature. The third model represents the actual fruit y kinematics
as observed in experiments and the last one is a modication of the latter, chosen
to investigate the eect of symmetry. The most prominent aspects of the Roboy
kinematic model are the sawtooth displacement and the trapezoidal angle of
attack. The fruit y 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
coecient was matched.
It was found that the realistic fruit y wing kinematics result in signicantly
lower drag at similar lift compared with the simplied wing kinematic models
used in literature. The trend that the fruit y kinematics increases aerodynamic
performance agrees well with the predictions of the quasi-steady theory, but the
numerical ow simulations provide a more complete quantitative analysis of the
ow 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 dierence in performance in terms of mean lift-to-drag ratio
between the dierent kinematic models was signicant. The mean aerodynamic
drag at equal lift of the fruit y models is about 49% lower compared to the Roboy
model and about 29% lower with respect to the harmonic model. Therefore, the
eect 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 Roboy model
has a small eect on the mean lift but the mean drag is aected signicantly. 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-
plication used by the Roboy, 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 y model, is not
aecting 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 signicant 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
eect on the mean lift and mean drag. However, the eective 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 ight
have an appreciable eect on the accuracy of performance models of insect ight.
In particular they indicate that kinematic features, found in fruit y 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 apping
foil
AIAA paper 2009-791.
A two-dimensional numerical investigation is performed to study the vortical ow
around a apping foil that models an animal wing, n, or tail in forward motion.
The vortex dynamics and performance are studied to determine the inuence of
foil kinematics. The baseline kinematic model is prescribed by harmonic functions
which can be characterised by four variables, the dimensionless wavelength, the
dimensionless apping amplitude, the amplitude of geometric angle of attack, and
the stroke plane angle. The foil motion kinematics has a strong inuence 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 signicantly advantageous for the force enhancement during the
full stroke. Plots of eciency versus the independent variable show that, for
symmetric kinematics, the largest eciency is achieved at an intermediate value
of each variable within the parameter range considered, where periodic ow occurs.
122 Vortex wake interactions of a two-dimensional apping foil
6.1 Introduction
The ow around a apping wing, n, 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 specic 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 ow is the result of a match between the driving frequency
and the natural shedding frequency which is referred to synchronisation of the
ow (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 apping 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 apping foil displays a variation of the natural shedding
frequency as a function of angle of attack (Katz, 1981, Dickinson & Gotz, 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 apping
period. Several studies have shown that the wing benets 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 apping motion, plunging and pitching, at a Reynolds
number of Re = 150 which corresponds to the ight of a small insect, e.g. a fruit
y. Here only the near wake of the foil is studied. The motivation for this is
that performance of a apping foil is inuenced mainly by near wake dynamics.
The objective of the present study is to investigate the inuence of dierent foil
kinematics on the vortex-wake structure, force coecients, and performance.
6.2 Flapping foil parametrisation
The baseline kinematic model is based on harmonic motion, such as used by Wang
(2000b,a). The ow around a apping 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 inuence of dierent 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

),
the forward velocity of the foil U

[m/s], the chord length of the foil c [m] and


the stroke plane angle (

). The denition of the dimensionless parameters is


schematically illustrated in gure 6.1 and described in more detail below. The di-
mensionless wavelength

represents the number of chord lengths travelled during


one apping period:

=
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

:
St =
fA
U

=
A

.
It corresponds to the maximum induced angle of attack A

ind
at mid-stroke due
to the translation of the apping motion of the foil. The mean velocity U [m/s] is
obtained by averaging the velocity components over one apping period:
U =
1
T
_
T
0
_
(U

+U
ap
X
)
2
+ (U
ap
Y
)
2
dt .
Here T [s] is the period, U
ap
X
[m/s], and U
ap
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

,
where [m
2
/s] is the kinematic viscosity and changed for every computation to
match the Reynolds number. For the basic model the dimensionless wavelength

,
the dimensionless amplitude A

, the amplitude of geometric angle of attack A

,
and the stroke plane angle are chosen as independent variables. The Reynolds
number is kept constant at Re = 150.
In order to study the inuence of the kinematics, each parameter is varied from
the baseline model, dened by

= 6.8, A

= 1.5, A

= 15

, and = 90

. The
dimensionless wavelength was varied from

= 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

45

with a 15

increment. The inuence of the stroke plane angle is


124 Vortex wake interactions of a two-dimensional apping foil

90

(tan
1
)2Stsin()
2A

sin()
(a)
y
x
Y
X

2A
U

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

, the dimensionless amplitude


A

, the Strouhal number St, the angle amplitude A

, and the stroke plane angle . In this frame of


reference the observer is xed relative to the undisturbed air. The ight direction is from right to left.
(b) The two-dimensional relation between two inertial coordinate systems. The downstroke phase is
lled by dark blue and the upstroke by light blue.
investigated for two dierent angle amplitudes A

= 15

and 45

in combination
with 15

90

with a 15

increment. In the cases that the stroke plane angle


diers from 90

, the resulting apping motion is asymmetric.


6.3 Force coecients and performance
In this study two inertial coordinate systems are used, see gure 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 , 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 coecients are scaled using
the average dynamic pressure q [N/m
2
]:
q =
1
2
U
2
.
Using q, the force and moment coecients are dened as:
C
D
=
D
qc
, C
L
=
L
qc
, C
M
=
M
qc
2
.
Projecting the lift coecient C
L
and drag coecient C
D
onto the y- and x-axes
we obtain the foil lift coecient C
l
and the foil drag coecient C
d
respectively.
C
l
= C
L
sin +C
D
cos ,
6.4 Numerical model 125
C
d
= C
L
cos +C
D
sin .
Note that a negative drag coecient C
D
means thrust which is necessary in forward
locomotion whereas the foil drag coecient C
d
indicates the uid force that the
animal must overcome for translational motion of its wing, n, or tail and is
relevant to the required power of locomotion.
The comparative assessment of the aerodynamic performance of the dierent
kinematic models is based on the mechanical eciency of the foil motion. The
eciency [%] is the ratio between the eective propulsive power P
e
[Nm/s]
and the required power P
req
[Nm/s] which are given in (6.1), (6.2), and (6.3)
respectively:
=
P
e
P
req
100% , (6.1)
where
P
e
= D U

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

[m/s] the free-stream velocity, T [s] the apping


period, d [N] the foil drag, U
ap
[m/s] the translational velocity of the foil in the
stroke plane, M [Nm] the moment about the centre of rotation and
ap
[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 (, ) are
used following Bos et al. (2008). The constant and correspond to confocal el-
lipses and hyperbolas respectively. These elliptical coordinates can be transformed
to Cartesian coordinates via a conformal mapping:
x +iy = cosh( +i) .
The result of this conformal mapping can be seen in gure 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 inuence of far eld boundary condition can be neglected. A uniform grid in
(, ) 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 apping 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 uid at rest in which the initial velocity vector is
zero. The resulting wake patterns have been classied 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 apping cycle.
The moment when a LEV is shed from the wing is dened as the moment when
its core passes the trailing edge. The averaged aerodynamic force coecients in
table 6.1, 6.2, 6.3, and 6.4 are obtained using three simulation periods. Note that
the eciency is only calculated when the drag is negative, i.e. thrusting mode.
6.5.1 Inuence of dimensionless wavelength
The wake pattern and vortex behaviour are studied as a function of the dimension-
less wavelength in the range of 24

3. The numerical results are provided


in table 6.1 by decreasing dimensionless wavelength which is equivalent to an in-
crease in apping frequency at a constant ight velocity. Our results are similar to
the experimental results found by Lentink et al. (2008) using a soap-lm tunnel.
At high dimensionless wavelengths

= 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 eective angle of attack A

e
which leads to
positive drag. Thrusting modes are found for 12

because of generated LEVs


which pull the foil toward in forward direction. For dimensionless wavelengths
12

5.7 the LEVs are shed before stroke reversal. The amount of LEVs
6.5 Results and discussion 127

pattern C
L
C
Ldownstroke
C
Lupstroke
C
D

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 Inuence of dimensional wavelength. The numerical results are shown for 13 dierent
values for the dimensionless wavelength, A

= 1.5, A

= 15

, and = 90

.
and TEVs shed from the foil is decreasing with the dimensionless wavelength be-
cause the vortices have less time to develop and shed, see gure 6.3. Therefore,
the LEVs stay attached to the foil relatively longer at lower dimensionless wave-
lengths. The LEVs increase in size and strength due to increasing eective 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

= 3, so that the forces of this case are varying with relative


small changes from period to period. The asymmetry in the lift coecient 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 gure 6.4.
Therefore the foil does not fully benet from the attachment of LEVs.
6.5.2 Inuence of dimensionless amplitude
Six dimensionless amplitudes are chosen to investigate the inuence 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 eective 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 eective angle of attack is high enough to


128 Vortex wake interactions of a two-dimensional apping foil
(a) No vortices,

= 24, A

eff
= 7

(b) 2P+2S,

= 6.8, A

eff
= 39

(c) 2P+S,

= 6.0, A

eff
= 43

(d) 2P,

= 5.3, A

eff
= 46

(e) P+S,

= 4.5, A

eff
= 49

(f ) Aperiodic,

= 3, A

eff
= 53

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

. The ow is from left to right. All images are


taken at t = 0.35T. A

= 1.5, A

= 15

, and = 90

.
A

pattern C
L
C
Ldownstroke
C
Lupstroke
C
D

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 dierent dimensionless amplitudes.

= 6.8,
A

= 15

, and = 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 coecients to study the inuence of wavelength. Lift (a) and drag (b)
histories of dierent wake patterns for six dimensionless wavelengths. A

= 1.5, A

= 15

, and
= 90

.
A

pattern C
L
C
Ldownstroke
C
Lupstroke
C
D

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 dierent angle amplitudes.

= 6.8, A

= 1.5,
and = 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 coecients.
6.5.3 Inuence of angle of attack amplitude
Table 6.3 shows numerical results for dierent angle of attack amplitudes. The
plunging kinematic model, A

= 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 eective 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 dierence
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 apping foil
A

pattern C
L
C
Ldownstroke
C
Lupstroke
C
D

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 Inuence of the stroke plane angle. Numerical results of the kinematics for six dierent
stroke plane angles in combination with two dierent angle amplitudes.

= 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

= 45

, no signicant vortices are formed on the foil because the eective


angle of attack is low A

e
= 9

at mid-stroke. However, shear layers which are


generated by the foil, form themselves into a 2P pattern. At this low eective
angle of attack the foil produces lower lift and thrust.
6.5.4 Inuence of stroke plane angle
The stroke plane angle causes an asymmetry in the kinematics. Here the results
for the inuence of the stroke plane angle with two dierent angle amplitudes
A

= 15

and A

= 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 coecient
is increasing for decreasing stroke plane angle during the downstroke until the
ow 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 dierence in relative velocity between up
and downstroke also aects 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 coecient of symmetric kinematics.

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

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

diagram of our sinusoidal apping wing. The angle


amplitude and stroke plane angle are kept constant at A

= 15

and = 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 gure 6.5(a), where the time-averaged
lift coecient 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 gure 6.5(b). There is an im-
portant limitation in forward apping locomotion. To begin with, the (absolute)
eective 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

as:
St =
A

>
1
2
tan(A

geo
+A

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 inuence the stability and controllability of the MAVs. Fig-
ure 6.6 shows plots of eciency versus the independent motion parameters. In
symmetric kinematics there is an optimal value for each variable, see gure 6.6(a),
(b) and, (c). The peak eciency of 28.18% could conrm that the wing rotation
plays an important role in the unsteady aerodynamic force production.
132 Vortex wake interactions of a two-dimensional apping foil

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

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

2
P
2
P
+
2
S
0 10 20 30 40 50
0
10
20
30
(c)

= 15

= 45

2P
2
P
+
S
2P+2S
60 65 70 75 80 85 90 95 100
0
5
10
15
(d)
Figure 6.6 Inuence of apping kinematics on the eciency. Eciency 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 ow was used to investigate the eect of
foil kinematics on the vortex dynamics around an ellipsoid foil subjected to pre-
scribed apping 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 apping 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 ow 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 benecial for understanding the inuence
of wing kinematics on the performance characteristics.
CHAPTER 7
Vortical structures in
three-dimensional apping ight
Submitted to J. Fluid Mech. (January 2010).
Results are obtained by performing numerical simulations of the three-dimensional
ow around a apping wing. A parameter study is performed to investigate the
performance in apping ight and to get insight into the vortex dynamics and force
generation. Dierent aspects, relevant for three-dimensional apping wing aero-
dynamics, have been studied, namely the angle of attack, the Rossby number, the
Reynolds number and the stroke kinematics. First, the ow around a dynamically
scaled model wing is solved for dierent 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 dierent 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 gure-of-eight, a gure-of-O or a gure-of-U
pattern. Finally, the three-dimensional ow is compared with the two-dimensional
studies performed on apping forward ight.
7.1 Introduction
To understand the aerodynamic performance of apping wings at low Reynolds
numbers, relevant for insect ight, it is important to obtain insight into the vortex
136 Vortical structures in three-dimensional apping ight
dynamics and its inuence on force development. The most important feature
in apping 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 ight (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 ow
eld induced by the apping wings, several two-dimensional studies have been
performed (Dickinson & Gotz, 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 apping (Dickinson et al., 1999, Lehmann, 2004, Birch et al., 2004) model
wing, which rotates around its base. Those results indicate that three-dimensional
ow eects are essential for the LEV stability. Previously conducted research
addressed a possible analogy between the LEV on apping 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 ow, which could suggest
that a spanwise ow may play an important role concerning the LEV stability in
insect ight (Ellington et al., 1996, Van Den Berg & Ellington, 1997). Lentink
& Dickinson (2009b) discussed that the stability of the LEV growth specically
might be increased by the spanwise ow through the LEV core, driven by either
the dynamic pressure gradient on the wings surface, the centrifugal acceleration of
the boundary layer or the induced velocity eld of the spiral vortex lines (Ellington
et al., 1996). Additionally, the LEV stability may be strengthened by a reduction
of the eective angle of attack as a result of the tip vortex generation (Birch &
Dickinson, 2001, Shyy et al., 2008b). However, Birch & Dickinson (2001) showed
no signicant eect of the spanwise ow on the LEV strength and stability, using
plates at dierent spanwise locations to block the spanwise ow, 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 apping 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 ap around a base of which the location
can be varied. By varying the location of the centre of rotation, the inuence of
the revolving strength (Rossby number) and the eect 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 inuence on the apping 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 apping wing simulations 137
F
X
F
Y
F
Z
(t)
(t)
(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 denitions. Illustration of the wing
motion and force denitions. (t) corresponds to the stroke variation, (t) to the geometrical angle of
attack and (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 dierent 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. Dierent deviation patterns are investigated, following the shape of
gure-of-O, gure-of-U and gure-of-eight.
The apping 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 ecient, section 7.3 briey discusses the
validation and verication of the ow solver. The vortical ow needs to be visu-
alised in such a way that the resulting vortices are clearly visible. Dierent vortex
identication 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 apping wing simulations
In order to study the vortex dynamics and stability of the leading-edge vortex, the
ow is solved using Computational Fluid Dynamics (CFD), of which the details
are described in chapter 2. The three-dimensional apping wing is modelled in
order to provide a framework for comparison, which is still representative for true
insect ight, 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 ow phenomena of our interest. Additionally, the simulation
strategy is discussed in section 7.2.2.
138 Vortical structures in three-dimensional apping ight
7.2.1 Modelling and parameter selection
In general, most investigations concerning apping 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 y 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 inuence on the forces and the ow eld. In
addition, Luo & Sun (2005) showed that the airfoil corrugation, present in dragon-
y wings, did not inuence the force development signicantly. The length scales
of the corrugation are orders of magnitude smaller compared to the length scale
of the separated ow region or the leading-edge vortex, such that signicant eect
of corrugation on the ow can be neglected.
Planform selection
The single wing span is xed to b
s
= 2.0 and the chord at mid-span is c = 1.0.
The hinge around which the wing is able to ap is xed 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 Roboy was xed to 0.7 (Sane & Dickinson, 2002, Poelma
et al., 2006). Since the wing planform is chosen to be ellipsoidal, the wing sur-
face is dened by S = ab, where a = 0.5 and b = 1.0 are the semi-minor and
semi-major axes, respectively, such that S = /2. The average chord length of
this ellipsoidal planform is found to be c = S/b
s
= /4. So, the three geomet-
ric parameters important for the apping wing simulations are dened: b
s
, S and c.
Kinematic models
The apping wing motion is prescribed by three dierent motion angles dening
the deviation angle (t), apping angle (t) and the angle of attack (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

= 0 and A

= 20

. A combined devi-
ation and apping angle variation leads to a wing tip pattern. Depending on the
variation of the deviation angle, gure 7.2 shows the resulting gure-of-O, 7.2(a),
gure-of-eight, 7.2(b), or gure-of-U, 7.2(c).
Realistic fruit y kinematics (Fry et al., 2003, Lentink & Dickinson, 2009b,
Bos et al., 2008) resembles an harmonically varying deviation angle, a sawtooth
shaped apping 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 eect 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 eects of the Rossby
number, Reynolds number, angle of attack and stroke kinematics, i.e. trapezoidal
7.2 Three-dimensional apping wing simulations 139
Flapping angle, [

]
D
e
v
i
a
t
i
o
n
a
n
g
l
e
,

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

]
D
e
v
i
a
t
i
o
n
a
n
g
l
e
,

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

]
D
e
v
i
a
t
i
o
n
a
n
g
l
e
,

]
[
-
]
0 20 40 60 80 100 120 140 160 180
-80
-60
-40
-20
0
20
40
60
80
upstroke
downstroke
(c) gure-of-U
Figure 7.2 Dierent wing tip patterns. Dierent wing tip patterns as a result of the variation
in deviation with a combined apping motion. (a) gure-of-O. (b) gure-of-eight. (c) gure-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 = 15

to = 90

with increments of = 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 apping
period, such that T
rot
= 0.25T corresponds to fully harmonic angle of attack vari-
ation. The apping angle was chosen to vary harmonically to isolate the eects 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 xed: the dimensionless amplitude of the wings cross-section at
140 Vortical structures in three-dimensional apping ight
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
dened as:
Re
R
g
=
4A

fR
g
c

, (7.1)
where A

is the apping angle amplitude, f the apping frequency, c the average


chord length and the kinematic viscosity.
The kinematic viscosity is xed 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 apping angle amplitude in order to keep the average Reynolds number
and the displacement at R
g
comparable. Therefore, the apping 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

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

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 eect of three-dimensional wing kinematics in hov-
ering ight, which is the main subject of the present thesis, the hovering wing
kinematics is substituted into the expressions for the angular and centripetal co-
ecients, C
ang
and C
cen
, and the Rossby number Ro, equation (4.6) to nd the
following:
C
ang
=
2

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 dened by the wing geometry, whereas the
angular acceleration number C
ang
, depends on the wing kinematics.
Force and performance denitions
In order to determine the eect of the dierent motion and geometric parameters
on the forces and performance, proper denitions are necessary. Since the wing
rotates with a rotating reference frame, two dierent force denitions are possible,
7.2 Three-dimensional apping wing simulations 141
in the inertial and the rotating reference frame, which is shown in gure 7.1 in re-
lation to the motion angles. Because the present research concerns hovering ight,
the lift force is by denition 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() F
Z
cos(). (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 coecients, C
L
and C
D
, are obtained by division
using the average dynamic pressure, q = 0.5U
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 , the
Reynolds number Re and the wing kinematic model. In order to investigate the in-
uence of the wing kinematics, the shape of the angle of attack variation, reected
by T
rot
and the deviation amplitude, A

were varied. A systematic overview of all


cases is provided here.
Inuence 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 apping wing, under hovering conditions. The ow is
solved for dierent 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 sh operate at a Rossby number close to Ro = 3.0,
which seems to be a biologically convergent solution for animals moving in uids.
Inuence 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 = 15

to = 90

with increments of = 15

. This
provides insight in the force development as a function of angle of attack for a
fully revolving, i.e. apping, 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 eects on the behaviour of the leading-edge vortex.
142 Vortical structures in three-dimensional apping ight
Rotation origin, R
tip
2.5 3.0 4.0 5.0 6.0 7.0 12.0 102.0
A

]
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 apping wing. This matrix is used to
study the inuence 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 xed 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.
Inuence of the kinematic modelling
Besides the wing stroke curvature, Reynolds number and angle of attack, it remains
interesting to investigate the eect of two kinematic parameters, the deviation,
which may cause a gure-of-eight pattern and the trapezoidal shape (dened by
T
rot
) of the angle of attack variation. This trapezoidal angle of attack variation
is dened by
=
_

_
A

sin(2ft) 0 t < T
rot
,
A

T
rot
t <
1
2
T T
rot
,
A

cos(2ft)
1
2
T T
rot
t < T
rot
+
1
2
T,
A

T
rot
+
1
2
T t < T T
rot
,
A

cos(2ft) 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 dierent values of T
rot
, the angle of attack is plotted
in gure 7.3. Note that the geometric angle of attack is given by
geom
=

2
.
Table 7.2 shows dierent deviation amplitudes A

, in combination with a vary-


ing, T
rot
, which determines the amount of trapezoidal shape, as was already
illustrated in gure 7.2. The varying deviation angle amplitude, may cause dif-
ferent wing tip patterns, i.e. gure-of-O, gure-of-eight and the gure-of-U,
depending on the deviation frequency.
7.3 Flow solver accuracy 143
A

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 apping 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

. Two dierent angle of attack amplitudes are used, 45

and 60

, which is
shown to result in maximal lift coecients. The Reynolds number is xed to Re = 100, the grid
resolution to 800k and the time-step was chosen corresponding to Co
max
= 2.0.
t/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
50
60
70
80
90
100
110
120
130
Figure 7.3 Angle of attack variation with a trapezoidal shape. To investigate the inuence
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 ow solver, concerning highly unsteady
and vortical ows, numerical comparisons are performed. A verication 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 renement is uniform and the cells are clustered close to the ap-
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 verication study is performed using the ow around a three-dimensionally
apping wing. The kinematics is according to the simple harmonic model. The
apping angle amplitude was xed to A

= 63

(1.1 rad), the mid-stroke angle


of attack was given by A

= 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 apping ight
Mesh resolution
100k 200k 400k 800k 1600k
C
o
m
a
x
2.0 + +
1.0 + + + + +
0.5 + +
Table 7.3 Simulation matrix: verication. Verication matrix showing the cases used for
verication purposes, with varying mesh resolution (100k 1600k) and time-step. The time-step is
reected through the maximal Courant number, Co
max
, which varies from 2.0, 1.0 to 0.5. Two cases
are performed for two dierent Reynolds numbers, +: Re = 100, : Re = 1000. The kinematics of the
three-dimensional wing is simple harmonics with A

= 63

, A

= 45

and A

= 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 ow solver, the drag and lift coecients
are plotted in gure 7.4 for meshes from 100k to 1600k cells and Co
max
= 1.0. The
corresponding limit cycles are shown in gure 7.5. As can be clearly seen from
the gures, the ow is periodic and the force coecients (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 coecients are plotted with
increasing spatial and temporal resolution in gure 7.6. Figure 7.6(a) and 7.6(b)
shows a converging time-averaged lift and drag coecients with increasing grid
resolution, the value at is determined by Richardson extrapolation (Ferziger &
Peric, 2002), see chapter 2. Temporal convergence is shown in gure 7.6(c), where
the time-averaged lift and drag coecients are plotted with decreasing time-step.
In order to justify the choice for grid and temporal resolution for the three-
dimensional apping 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 dierences 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 suciently ne 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 dierences
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 suciently accurate force coecients,
but to capture the far wake vortex dynamics at least 800k cells are required.
7.4 Vortex identication methods for ow 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 verication: force coecients. Lift and drag coecients for the
verication 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

= 63

, A

= 45

and
A

= 0

.
N
lift
[%]
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 coecients for varying grid sizes. Error values of
lift and drag coecients [%] 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

= 63

, A

= 45

and A

= 0

.
Furthermore, at Co
max
= 2.0, the temporal errors are suciently small. Therefore,
a grid resolution of 800k in combination with Co
max
= 2.0 was used for all three-
dimensional apping wing simulations, described in this chapter.
7.4 Vortex identication methods for ow
visualisation
In order to study the vortex dynamics and the stability of the leading-edge vortex
in particular, a proper vortex identication criterion is essential. Dierent well-
known techniques to detect and visualise vortices are based on the velocity gradient
tensor, which requires a complete velocity eld. Two dierent vortex identication
criteria are discussed, namely the magnitude of vorticity || (Lu & Shen, 2008)
146 Vortical structures in three-dimensional apping ight
Drag coecient [-]
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 verication: limit cycle. The limit cycles, constructed from the
lift and drag coecients clearly shows periodic behaviour for the verication cases. The grid size was
xed 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

= 0 rad, A

= 1.1 rad and A

= 0.785 rad.
Co
max

lift
[%] 400k
drag
[%] 400k
lift
[%] 800k
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 coecients for varying temporal resolution. Error
values of lift and drag coecients [%] 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

= 63

,
A

= 45

and A

= 0

.
and the Q criterion (Hunt et al., 1988).
The rst criterion is based on the magnitude of vorticity ||, where = u.
If || reaches a user-dened threshold, that region is identied as a vortex. More
specically, 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
ows this may become a diculty. In two-dimensional ow, however, || 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 identied as a vortex. This second
invariant of u is written as
Q =
1
2
_
||
2
|S|
2
_
,
where the rate of strain tensor S is given by S =
1
2
(u +u
T
) and the vorticity
tensor by =
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 identication methods for ow 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
coecients for increasing spatial resolution and constant time-step, corresponding to Co
max
= 1.0.
The nal value, at , is obtained using Richardson extrapolation. The temporal convergence is
illustrated in (c) and (d), showing the average drag and lift coecients for decreasing time-step at
two specic grid sizes, 400k and 800k.
rate. Therefore, a region where Q > 0 indicates a clear swirling ow (as shown
by Chakraborty et al., 2005). It must be noted that Jeong & Hussain (1995)
found that Q > 0 is not a sucient condition to have a pressure minimum in the
vortex core of that specic region. In most cases, however, a pressure minimum
does occur. By neglecting these unsteady and viscous eects from the governing
Navier-Stokes equations the following relation can be obtained for the symmetric
tensor
2
+S
2
:

2
+S
2
=
1

(p),
where is the uid density and p the pressure. In order to identify which vortex
criterion should be used, gure 7.7 shows iso-surfaces around a apping wing.
The wing aps around a distance of 0.5 from the wing root and the apping
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 wings
148 Vortical structures in three-dimensional apping ight
(a) || (b) Q
Figure 7.7 Comparison of the near wake ow eld using dierent vortex identication
criteria. The spiralling leading-edge vortex is visualised using contour plots of the magnitude of
vorticity, || and Q. The arrow shows the apping direction of a downstroke and the ow is visualised
at mid-stroke. (a) shows the contour of the vorticity magnitude, || = 5.0 and (b) the Q = 2.0 value.
The colours represent values of helicity, h = (u )/(|u| ||), 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
identied using a carefully chosen threshold of the vortex identication criteria,
using the values || = 5.0 and Q = 2.0, normalised by their maximal values. The
colours show the helicity which is dened as h = (u )/(|u| ||), within the range
of 1.0 h 1.0. A positive helicity, h > 0, means that the direction vector of
vorticity ( = u) is aligned with the local ow velocity.
In gure 7.7(a) it can be observed that || 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 gure 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 oers sucient and adequate information about the local
ow eld, e.g. a rotation dominated region is identied 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 ap-
ping motion) and its inuence on the variation of forces, dierent geometric and
kinematic parameters are systematically varied. First, the inuence of the angle
of attack on the forces is briey discussed in section 7.5.1. In that section, the in-
uence of an increasing mid-stroke angle of attack is briey discussed on the force
development. In order to investigate the eect 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 dierent
Reynolds number as well. The inuence 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 ight simulations, section 7.6 deals with forward apping ight 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 apping ight
Previous studies showed that the angle of attack variation during the stroke in-
uences the forces considerably. This was conrmed by a recent two-dimensional
investigation (Bos et al., 2008), concerning hovering ight at fruit y conditions.
It was already mentioned that the angle of attack at mid-stroke is varied from
= 15

to = 90

. Note that = 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 coecient for varying angles. The Reynolds number is Re = 100 and Rossby
Ro = 3.2, corresponding to a apping wing with small radius of curvature. It is
seen that the maximal time-averaged lift coecient occurs at a mid-stroke angle
of attack of = 45

. However, the average lift-to-drag ratio obtains a maximal


value at an angle of attack of = 30

, from table 7.6.


Figure 7.8 shows a variation of the lift and drag coecients during a complete
apping cycle. It can be observed that the force variation is periodic and smooth.
The average drag is maximal and lift minimal for = 90

. Furthermore, it can be
seen in gure 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 apping
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 = 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 = 45

and = 60

, the drag is signicantly


lower for = 45

. In addition, it seems that the leading-edge vortex only grows


signicantly at higher ( 45

) angles of attack. The dierence between the


maximal ( = 45

) and minimal ( = 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 Inuence of apping 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 apping ight

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 coecients 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 apping wing with small stroke curvature.
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
= 15

= 30

= 45

= 60

= 75

= 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
[
-
]
= 15

= 30

= 45

= 60

= 75

= 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 coecients for a apping wing at Reynolds number
of Re = 100. The variation is shown for lift (a) and drag (b) coecients induced by a apping 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 coecients 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 xed to = 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 = 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 signicant, 18.7%. Figure 7.9 shows the force
histories concerning the nearly translating wing, i.e. Ro = 130. When compared
with gure 7.8 (which applies to the revolving wing Ro = 3.2) it is seen that
both lift and drag variations are signicantly 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 coecients during the apping cycle and the eect 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 apping 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, gure 7.13 shows the streamlines to illustrate
the leading-edge and tip vortices in more detail.
An additional observation, while comparing gures 7.8 and 7.9 is that the lift
drops signicantly (75% at least) during mid-stroke (t = 0.2T) for an angle of
attack of = 15

. Figure 7.12 shows the Q = 1.0 iso-surfaces of a translating


wing at Ro = 130 for = 15

and = 45

. It is clearly seen that the leading-


edge vortex for = 15

is not yet fully developed, which results in the lower


152 Vortical structures in three-dimensional apping ight
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
= 15

= 30

= 45

= 60

= 75

= 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
[
-
]
= 15

= 30

= 45

= 60

= 75

= 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 coecients for a translating wing at Ro = 130. Lift
(a) and drag (b) coecients induced by a apping wing at a Rossby number of Ro = 130, such that
the wing approximately translates. The Reynolds number remained xed at Re = 100. The angle of
attack amplitude is varied from : 15

, : 30

, : 45

, : 60

, : 75

and : 90

.
lift, compared to = 45

. It seems that the trend of the force development


with the angle of attack is similar for apping 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 apping 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 apping and translating wings
at = 15

, that the leading-edge vortex development is not signicant to increase


the lift, instead the lift decreases.
7.5.3 Inuence 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 y (Sane & Dickinson, 2001, Birch & Dickinson, 2003), hawk-
moth (Liu & Kawachi, 1998) and dragony (Isogai et al., 2004), respectively. The
time-averaged lift and drag coecients are plotted in gure 7.14(a) for dierent
angles of attack and dierent 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 gure 7.14(a) it can be deduced that the overall lift coecients are signif-
icantly higher for the apping (Ro = 3.2) compared to the translating (Ro = 130)
wing. The drag increases as well. At maximal lift, = 45

the dierence between


apping (Ro = 3.2) and translating (Ro = 130) in lift coecient 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 coecients for a apping wing with varying Rossby num-
bers. Lift (a) and drag (b) coecients induced by a apping wing around varying Rossby numbers,
Ro = 3.2 130. The amplitude of the angle of attack variation was xed such that at mid-stroke
= 45

. The average Reynolds number remained xed 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 dierent 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 xed to Re = 100. Colours indicate helicity.
154 Vortical structures in three-dimensional apping ight
(a) = 45

(b) = 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 = 15

and = 45

for a Rossby number of


Ro = 130. A time-frame is shown at mid-stroke, t = 0.25T. The average Reynolds number was xed
to Re = 100. Colours indicate helicity.
(a) Ro = 3.2 (b) Ro = 130
Figure 7.13 Streamline visualisation of the near wake for dierent 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 xed to Re = 100. Colours indicate helicity.
7.5 Flapping wings at low Reynolds numbers 155
Drag coecient [-]
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 dierent 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 dierence in lift increases (although slightly) with Re, apping is impor-
tant at lower Reynolds numbers, but becomes slightly more important at higher
Reynolds number. The dierence in drag is only signicant at larger angles of
attack. At = 45

the dierences in time-averaged drag coecient are negligi-


ble when the variation in Reynolds number is concerned, while at = 75

the
dierences in drag become signicant. However, considering a apping motion
(Ro = 3.2) with respect to translating (Ro = 130), an average dierence in drag
of 7.5% is obtained. While lift is enhanced signicantly, combined with a small
drag penalty, there is still a large gain in lift-to-drag, which is benecial in terms
of performance.
Another observation from gure 7.14(a) is that at large mid-stroke angles of
attack, the time-averaged lift and drag coecients show marginal variations with
Reynolds number, for a translating (Ro = 130) wing. On the other hand, the
Reynolds number has a larger eect on the lift and drag, while apping. This may
be explained by considering that the leading-edge vortex was found to be more
stable on a translating wing, compared to apping. Looking at gure 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 apping 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 apping ight
(a) Re = 100
(b) Re = 1000
Figure 7.15 Vortex visualisation of the near wake for dierent Reynolds numbers, Ro =
3.2. Iso-surface of Q = 1.0 are shown for two Reynolds numbers, Re = 100 and Re = 1000, for a
apping wing Ro = 3.2. A time-frame is shown at mid-stroke, t = 0.25T. Colours indicate helicity.
comparison are shown in gure 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
eects 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 signicant loss in lift. In or-
der to illustrate the irregular motion at larger Reynolds numbers, i.e. Re = 1000,
gure 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
apping motion, compared to translation. Figure 7.18 shows the Q iso-surfaces
for dierent Reynolds numbers Re = 100 and Re = 1000, for a translating wing.
In case of the apping wing Ro = 3.2, gure 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 apping 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 apping Ro = 3.2 and translating Ro = 130 ight. A Reynolds
number increase, leads to larger lift enhancement, but also to irregularities such
that the ow 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 dierent Reynolds numbers,
Ro = 3.2. Streamlines are shown for two Reynolds numbers, Re = 100 and Re = 1000, for a apping
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 dierent Reynolds numbers, Ro =
3.2. Iso-surface of Q = 1.0 are shown for two Reynolds numbers, Re = 100 and Re = 1000, for a
apping 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 apping ight
(a) Re = 100 (b) Re = 1000
Figure 7.18 Vortex visualisation of the near wake for dierent 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 Inuence of trapezoidal angle of attack
In order to study the inuence 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 dened by a rotation timing T
rot
, which is
varied from 0.25 to 0.1, the rst 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 xed.
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 signicant
increase in lift-to-drag ratio of 21.9%. Furthermore, it can be seen that the average
drag coecient, 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 gure 7.19 for varying
rotation duration during a full apping 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 eective 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 coecients 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 coecients for a apping wing with trapezoidal angle of attack. Lift
(b) and drag (c) coecients induced by a apping 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

= 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, eect
applies to the drag coecient.
At every time instance, the lift increase and drag decrease are only marginal,
but overall the results are signicant, 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 apping ight. In contrast to this, two-
dimensional studies (Bos et al., 2008) showed opposite eects, 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 apping ight
7.5.5 Inuence of deviation
The deviation amplitude A

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 gure-of-eight, which is present in
realistic fruit y kinematics (Fry et al., 2003). In (Bos et al., 2008) it was shown
that although deviation did not inuence the time-averaged values, the instanta-
neous lift and drag variations are signicantly aected. In order to investigate
the inuence of deviation the amplitude A

is varied from 0

to 20

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

from 0

to 20

and from 0

to 20

. The reference velocities are


adapted correspondingly.
First, the deviation amplitude is varied from A

= 0

to A

= 20

, according
to the gure-of-O wing tip pattern. Following gure 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 eective 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 coecients for
a wing following this gure-of-O motion, while the mid-stroke angle of attack is
xed to = 45

and the Reynolds number maintained at Re = 100. It is shown


that the time-averaged lift coecient decreases signicantly with 9.8%. However,
this is fully compensated by a decrease in drag with the same amount such that
the dierences 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 coecient. This
asymmetry in drag is the result of the asymmetrical variation in eective angle of
attack, as was previously discussed. Figure 7.20 show the lift and drag variations
during a complete apping stroke.
Secondly, the results are considered for a apping wing following the gure-
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 dierences in up- and downstroke drag are negligible for all deviation
amplitudes. This is in contrast to the observations, considering the gure-of-O.
The time-averaged drag coecient is constant for deviation amplitude variation,
while the lift coecient decreases with 9.4%, leading to a similar decrease in lift-
to-drag ratio. Because of symmetric (similar up- and downstroke) apping, a
decrease of lift-to-drag coecient is obtained, which is present, but not signicant
in comparison to the eect of varying Rossby and Reynolds numbers.
The third pattern is governed by the gure-of-eight wing tip motion. Al-
though, the previously described deviation patterns only cause marginal eects,
7.5 Flapping wings at low Reynolds numbers 161
A

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 coecients for varying deviation using a gure-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

= 0

to 20

, for a gure-of-O. The mid-stroke angle of attack was xed


at = 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

= 0
A

= 5
A

= 10
A

= 15
A

= 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

= 0
A

= 5
A

= 10
A

= 15
A

= 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 coecients for a apping wing with varying deviation using gure-of-
O. Lift (a) and drag (b) coecients induced by a apping wing, such that Ro = 3.2 and Re = 100.
The deviation amplitude is varied from A

= 0

to A

= 30

, using an angle of attack amplitude of


A

= 45

.
the gure-of-eight pattern may cause signicant changes in forces. Two types of
gure-of-eight patterns are used, from A

= 0

to 20

, this is called type 1 and


the second from A

= 0

to 20

, type 2. To achieve a gure-of-eight pattern,


the deviation frequency is twice the frequency of the other two wing tip patterns.
For both types of patterns, the eective angle of attack variation consists of three
parts during each half stoke, see gure 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 eective 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 ratios for both
types of gure-of-eight patterns. Since the variation of the eective angle of at-
tack is symmetric, the drag coecient is equal for up- and downstroke. The results
162 Vortical structures in three-dimensional apping ight
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

= 20

with
the baseline case A

= 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 signicant.
Figure 7.20 shows the lift and drag variations for this type 1 gure-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 eective angle of attack for a relative long period,
leading to a signicant loss of lift, as is seen in the gure. For A

= 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. = 15

, see section 7.5.1.


When considering the type 2 gure-of-eight pattern, table 7.11 shows no sig-
nicant decrease in drag. The lift, however, decreases considerably, although sig-
nicantly 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 dierences in drag, while increasing the deviation
amplitude. The maximal dierence in lift-to-drag ratio is therefore -12.0%. This
force balance is illustrated in gure 7.22, which shows the lift and drag during a
full stroke. At the beginning of a stroke, the eective angle of attack is decreased,
leading to lower lift and lower drag. During mid-stroke, the eective angle of
attack is increased, which is reected in the higher lift and drag.
Summarising, it was shown that the variation in lift and drag can be signi-
cantly inuenced by introducing deviation in the stroke pattern, i.e. gure-of-O,
gure-of-U and gure-of-eight. The gure-of-O patterns resulted in an asym-
metric force variation, due to asymmetric modulation of the eective angle of
attack. Lift and drag decrease with a similar amount, such that the lift-to-drag
ratio was only marginally aected. The time-averaged drag was not inuenced
by the symmetrical gure-of-U pattern. The average lift, however, did decrease,
such that a loss in lift-to-drag was observed. Two types of gure-of-eight patterns
were investigated, diering in the direction of motion. When the wing moved up-
A

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 coecients for varying deviation using a gure-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

= 0

to 20

, for a gure-of-U pattern. The mid-stroke angle of attack


was xed at = 45

and the average Reynolds number was maintained at Re = 100.


7.6 Flapping wings in forward ight 163
A

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 coecients for varying deviation using a gure-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

= 0

to 20

and A

= 0

to 20

, for a gure-of-eight pattern. The


mid-stroke angle of attack was xed at = 45

and the average Reynolds number was maintained at


Re = 100.
wards during mid-stroke, decreasing the eective 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 gure-of-U pattern.
The lift and drag are shown to be sensitive to dierent 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
ndings are very interesting for the development of Micro Air Vehicles (MAV) as
well.
7.6 Flapping wings in forward ight
In order to relate the results of the three-dimensional ow simulations to previously
conducted two-dimensional studies (Thaweewat et al., 2009), a three-dimensional
forward apping wing has been studied. Based on (Thaweewat et al., 2009), the
dimensionless wavelength is set to = 6.3 (maximal performance) in order to
justify a comparison. The mid-stroke angle of attack is varied from = 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 dierent cases. Additionally,
the average Reynolds number Re
R
g
is matched.
In table 7.12, the time-averaged force coecients are shown for three dier-
ent apping situations: a two-dimensional plunging airfoil, a three-dimensional
translating wing (Ro = 130) and a three-dimensional apping wing (revolving,
Ro = 3.2). For both apping and translating wings, the force coecients are com-
164 Vortical structures in three-dimensional apping ight
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
A

= 0
A

= 5
A

= 10
A

= 15
A

= 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

= 0
A

= 5
A

= 10
A

= 15
A

= 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 coecients for a apping wing with varying deviation using type 1 of
gure-of-eight. Lift (a) and drag (b) coecients induced by a apping wing, such that Ro = 3.2
and Re = 100. The deviation amplitude is varied from A

= 0

to A

= 20

, using an angle of attack


amplitude of A

= 45

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

c
i
e
n
t
[
-
]
A

= 0
A

= 5
A

= 10
A

= 15
A

= 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

= 0
A

= 5
A

= 10
A

= 15
A

= 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 coecients for a apping wing with varying deviation using type 2 of
gure-of-eight. Lift (a) and drag (b) coecients induced by a apping wing, such that Ro = 3.2
and Re = 100. The deviation amplitude is varied from A

= 0

to A

= 20

, using an angle of
attack amplitude of A

= 45

.
7.6 Flapping wings in forward ight 165
A

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 coecients in forward ight. Two- and three-dimensional
time-averaged force coecients for a apping wing in forward ight at Re = 150,

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

to 45

. A apping 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 coecient decreases with increasing an-
gle of attack amplitude. This is illustrated for the apping wing (Ro = 3.2)
in gure 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

= 30

, which is the case for both two-dimensional and


three-dimensional apping. 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-
cient. The two tip vortices result in a symmetric ow, 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 apping wing (Ro = 3.2) generates larger lift coecients
compared to the translating case, but lower compared to the two-dimensional
apping airfoil. The dierence with the translating wing becomes smaller with
increasing angle of attack amplitude. Figure 7.24 shows a comparison of the ow
elds for Ro = 130 and Ro = 3.2, without an angle of attack variation. For the
case without revolving (gure 7.24(a)), a smooth ring vortex is formed by the
166 Vortical structures in three-dimensional apping ight
t/T [-]
L
i
f
t
c
o
e

c
i
e
n
t
[
-
]
= 0

= 15

= 30

= 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
[
-
]
= 0

= 15

= 30

= 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 coecients for a three-dimensional wing in forward ight. Lift (a) and
drag (b) coecients for a three-dimensional apping wing in forward ight at Re = 150,

= 6.3 and
mid-stroke angle of attack of A

= 45

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

= 30

that the wake becomes more smooth, which is governed by a decrease in


eective 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 apping ight. 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, inuencing 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
apping wing aerodynamics.
7.7 Conclusions
This chapter deals with the results obtained by performing numerical simulations
of the ow around a three-dimensional apping wing. A numerical model has been
developed which solves the ow around a three-dimensional wing with complex
wing kinematics. The numerical code was veried by using a temporal and spatial
independence study.
Dierent aspects, relevant to three-dimensional apping 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 apping wing
in forward ight. Iso-surface of Q = 1.0 are shown for a three-dimensional apping wing in forward
ight at Re = 150,

= 6.3 and mid-stroke angle of attack of A

= 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 apping wing
in forward ight. Iso-surface of Q = 1.0 are shown for a three-dimensional apping wing in forward
ight at Re = 150,

= 6.3 and mid-stroke angle of attack of A

= 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 apping ight
have been studied. First, the ow around a dynamically scaled model wing is
solved for dierent 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 dierent 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 gure-of-eight
pattern. Finally, the three-dimensional ow is compared with the two-dimensional
studies performed on apping forward ight. 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 eect 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 = 45

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

. The apping 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 suciently high angles of attack = 30

. At = 15

the
leading-edge vortex is not strong enough, causing a sudden decrease in both lift
and drag during each half-stroke.
Secondly, the eects are studied of a varying stroke curvature, reected by the
Rossby number. Both time-averaged lift and drag decrease signicantly with in-
creasing Rossby number. At Ro = 130, a nearly translating wing, the lift decreases
with 24.7%, compared to the apping 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 signicantly reduced in size
and strength for the translating with at Ro = 130, compared to the apping wing.
In addition, the leading-edge vortex rolls-up to form two tip vortices instead of
one in case of the apping wings. This causes both lift and drag to be signicantly
lower at Ro = 130.
To study the eect of three dierent 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 dierences become
smaller at Re = 1000. The eect 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 apping wing at
Ro = 3.2. For both apping, 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 inuence of the angle of attack, Rossby and Reynolds numbers, the
7.7 Conclusions 169
eects 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 signicant 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 signicant 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
eects, however, lead to rm and stably attacked leading-edge vortex.
The second eect of kinematic modelling is reected by the presence of de-
viation. Three dierent wing tip patterns may be caused by deviation, such as
the gure-of-O, gure-of-U and the gure-of-eight. The gure-of-O wing tip
pattern leads to an asymmetric variation in eective angle of attack, leading to
dierences in drag for the up- and downstroke. Nevertheless, the eect of this
pattern on the time-averaged forces coecients is negligible. On the other hand,
the gure-of-U pattern does inuence the forces. The eective angle of attack is
signicantly decreased during mid-stroke. The time-averaged lift coecient is de-
creased by 9.4%. While drag is maintaining nearly constant, the lift-to-drag ratio
is decreased by about 9%. The gure-of-eight aects the forces most, however,
it depends on the starting direction. Two types are considered, the rst starting
the downstroke with an downward deviation motion, whereas the type 2 starts
upwards. The eective angle of attack is being inuenced in such a way that the
type 1 gure-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 gure-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 eects of Reynolds number, Rossby number and wing kine-
matics in hovering ight, a preliminary study is performed to compare two- and
three-dimensional forward apping ight. 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 dierent 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 dierence 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 signicant 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 apping aerodynamics.
CHAPTER 8
Inuence of wing deformation by
exing
A preliminary study is performed to investigate the eects of wing exing in ap-
ping wing aerodynamics. The eects of a cosine-shaped wing exing is analysed
in two-dimensional forward apping ight. For three-dimensional hovering ight,
the eects of wing exing in spanwise and chordwise directions are discussed. In
two-dimensional forward apping ight, the exing of the airfoil shows similar
eects as if wing rotation was applied, increasing its eciency. Furthermore, in
three-dimensional hovering, the exing reduces the strength of the leading-edge
vortex, compared to a rigid wing. This leads to an overall decrease in lift and
drag, this inuence is larger for chordwise compared to spanwise exing.
Section 8.1 deals with the inuence of exing for the two-dimensional plunging
airfoil in forward ight. Its inuence on the ow induced by a three-dimensional
wing in hovering ight is subject of study in section 8.2. The conclusions are
drawn in section 8.3.
8.1 Airfoil exing in two-dimensional forward
apping ight
In chapter 6, the inuence on the forces and vortex patterns of a variation in
dimensionless wavelength, apping amplitude, angle of attack and stroke plane
angle was investigated in two-dimensional forward apping ight. These results
are extended by using preliminary simulations of deformation airfoils, subjected to
a pre-dened exing. The boundary displacements due to the exing are dened
172 Inuence of wing deformation by exing
with respect to the initial boundary shape and varying in time such that the
maximal exing occurs at mid-stroke. The resulting airfoil shape is either a quarter
or half-cosine shape, corresponding to
f
= 0.25 or
f
= 0.5, respectively, as is
shown in gure 8.1. Besides the two dierent deformation mode shapes, the exing
amplitude is varied from A
f
= 0.1 to A
f
= 0.4, which is similar to maximal 40%
of the chord length. This exing was imposed on a baseline apping motion
(see chapter 6) with a dimensionless wavelength of

= 6.8 and a dimensionless


amplitude of A

= 1.5. The rotation amplitude is xed to A

= 0

and the average


Reynolds number was xed at Re = 150.
Table 8.1 shows the time-averaged lift, drag and lift-to-drag coecients for the
plunging airfoil. For both exing shapes the lift decreases with exing amplitude,
however, the time-averaged lift decreases faster for
f
= 0.25. Considering this
quarter-cosine shaped exing, a small exing amplitude of A
f
= 0.1 already results
in a signicant 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,
f
= 0.5, is smaller compared to all
f
= 0.25 cases, for similar
exing amplitudes, but increases more rapidly with increasing exing amplitude.
On the other hand, the half-cosine deformation (
f
= 0.5) is more ecient in
generation of thrust (negative drag). For example, the decrease in average lift for
A
f
= 0.4 and
f
= 0.5 is half the value obtained with
f
= 0.25, but instead, the
thrust is doubled. It appears that the half-cosine shaped exing increases thrust
signicantly, while the lift is less aected compared to the quarter-cosine shape.
Another observation from table 8.1 is the presence of asymmetry in the time-
averaged lift coecient for the quarter-cosine deection. Chapter 6 shows a similar
asymmetric ow 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
]

f
= 0.0

f
= 0.25

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 exing displacements
of a plunging airfoil are shown for dierent shape ratios. A shape ratio
f
= 0.25 corresponds to a
quarter-cosine and
f
= 0.5 to a half-cosine shape.
8.1 Airfoil exing in two-dimensional forward apping ight 173
A
f
C
Lave
C
Ldownstroke
C
Lupstroke
C
D
0 1.956 (baseline) 1.758 -2.154 0.058

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

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 coecients for a deforming plunging airfoil. The time-
averaged lift and drag are shown for a deforming two-dimensional airfoil, for two exing shapes,
f
=
0.25 and
f
= 0.5. The Reynolds number was xed to Re = 150, the dimensionless wavelength was
k = 6.8 and the rotation amplitude remained at A

= 0.
the force asymmetry is strongly related to the wake dynamics. Therefore, airfoil
exing using a quarter-cosine shape ( = 0.25) may induce a similar eect on the
wake dynamics as airfoil rotation. However, when applying the half-cosine airfoil
exing, the lift coecient 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 eect of the airfoil rotation, such that a symmetric force is
the result of a symmetric wake pattern, which is benecial in terms of eciency.
Figure 8.2 shows the drag coecients for both exing shapes. It is clear that the
thrust (negative drag) is more prominent for the half-cosine ( = 0.5) exing 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 ow physics, gure 8.3 shows vorticity contours for both exing shapes with
an amplitude of A
f
= 0.2 at t = 0.25T, which is midway during the downstroke.
Comparing gure 8.3(a) with 8.3(b), it is apparent that the leading-edge vortex is
of similar size and strength for both exing 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 ( = 0.5) exing mode, increasing its thrust.
Summarising, this preliminary two-dimensional investigation, to understand
the eect of wing exing, has led to some interesting results. The applied wing
exing inuences the ow in a similar way as was obtained by applying airfoil
rotation, indicating a similar mechanism in terms of the eects of angle of attack.
With the introduction of exing, the drag became negative to generate thrust in
forward ight, and the lift decreased signicantly. The lift and drag development
strongly depends on the shape of the wing exing. The half-cosine exing shape
174 Inuence of wing deformation by exing
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) = 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) = 0.5
Figure 8.2 Comparison of drag coecients for a exing airfoil. The drag coecients are shown
for the quarter-cosine (a) and half-cosine (b) shaped airfoil exing modes. The exing amplitude A
flex
is varied from 0.0 to 0.4, the latter corresponds to 40% of the chord length.
(a) = 0.25 (b) = 0.5
Figure 8.3 Vorticity contours. Vorticity contours around a plunging exing airfoil without ro-
tation at Re = 150 are shown during mid-stroke at t = 0.25T. The apping amplitude was xed to
A

= 1.5 and the dimensionless wavelength was set to

= 6.8. The exing amplitude was set to


A
f
= 0.2 using both exing shapes, while the rotation amplitude was kept to zero A

= 0

.
results in less lift than obtained using the quarter-cosine shape. Additionally, the
half-cosine exing 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 ow.
8.2 Wing exing in three-dimensional hovering ight 175
8.2 Wing exing in three-dimensional hovering
ight
In addition to the investigation of exing eects on an airfoil in forward apping
ight, 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 inuence of the Rossby number, Reynolds number and the kinematic
modelling, applied under rigid wing conditions, it may be interesting to study the
additional eects of wing deformation. The deformation of the wing is dened by a
user-dened exing function, leading to a realistic bending of the three-dimensional
wing (Combes & Daniel, 2003a,b, Shyy et al., 2008a).
In order to investigate the eects of three-dimensional wing exing on the
forces and performance, the exing is dened 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 exing 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 exing 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 deection at mid-stroke of 20% of the representative
exing length (wing span or chord, respectively). Figure 8.1 shows an illustration
of the exing 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 coecients for a deforming three-dimensional wing. Time-
averaged force coecients of a three-dimensional exing wing. The spanwise exing amplitude is varied
from A
f
= 0.0 to A
f
= 0.4, while the chordwise exing amplitude varies from A
f
= 0.0 to A
f
= 0.2.
.
176 Inuence of wing deformation by exing
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 coecients for a spanwise and chordwise deforming apping wing. Lift
(a) and drag (b) coecients induced by a apping 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 = 45

and the Reynolds number remained at Re = 100.


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

.
Table 8.2 shows the time-averaged force coecients for dierent exing amplitudes,
using spanwise or chordwise deformation. The most important result is that both
time-averaged lift and drag are only marginally inuenced by the spanwise exing,
so where the wing tip bends. The maximal dierence with the non-deforming case
(-3.6%) occurs for 40% wing tip deection, which is a signicant deformation.
The time-averaged drag remains nearly unaected as well, such that the maximal
dierence in lift-to-drag ratio is only -2.6%.
The chordwise deformation, on the other hand, inuences 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
exing, such that the maximal dierence in lift-to-drag ratio is -24.8%.
The force variation, during a full apping stroke, is shown in gure 8.4. It
is clear that the largest dierence in force coecients occurs during mid-stroke,
lowering the maximal lift coecient. It seems that the leading-edge vortex is ex-
pelled by the wing deformation, with a maximal eect at A
f
= 0.2 of chordwise
exing. When the lift decreases, the drag increases during mid-stroke. Addition-
ally, gure 8.5 shows the streamlines to visualise the leading-edge vortex during
mid-stroke (t = 0.25T) for both spanwise and chordwise exing. For comparison,
also the case using no exing is shown. Compared to the case without exing the
leading-edge vortex detaches earlier (less close to the wing tip) for both exing
cases. The spanwise exing has less inuence than the chordwise exing mode
shape. The inuence of spanwise exing is minor, which was also reected in the
8.3 Conclusions 177
time-averaged force coecients, which were only marginally lower. However, the
chordwise exing induced the leading-edge vortex to burst almost at the middle
of the wing, leading to signicantly less lift, see table 8.2.
8.3 Conclusions
In this chapter, the result have been described of a preliminary investigation to
understand the eects of wing exing. Therefore, a pre-dened exing deformation
has been applied to a plunging airfoil in two-dimensional forward ight and to a
three-dimensional apping wing in hovering ight. First of all, the two-dimensional
exing results in a similar eect on the ow and resulting forces as rotation of the
airfoil. With increasing exing amplitude, the larger eective angle of attack leads
to the generation of negative drag or thrust. In addition, the inuence of the exing
shape (quarter-cosine or half-cosine) of the airfoil was studied. A quarter-cosine
shaped exing results in signicantly higher thrust at less lift.
Secondly, a exing deformation has been applied to a three-dimensional ap-
ping wing in hovering ight. Two exing directions have been considered, span-
wise and chordwise exing. While keeping the exing amplitude constant, the
chordwise exing aects the ow the most. For a chordwise exing amplitude of
A
f
= 0.2 the time-averaged lift and drag coecients are signicantly decreased
and increased, respectively, such that the lift-to-drag decreases with 24.8%. The
spanwise exing does not show a comparable impact, only -2.6%.
178 Inuence of wing deformation by exing
(a) No exing
(b) Spanwise exing, A
f
= 0.2 (c) Chordwise exing, A
f
= 0.2
Figure 8.5 Streamlines on a exing wing in hovering ight. Iso-surface of Q = 1.0 are shown
for a three-dimensional apping wing in hovering ight at Re = 100 and mid-stroke angle of attack of
A

= 45

at t = 0.25T. (a) shows the wing without exing, while (b) shows the wing using spanwise
exing and (c) using chordwise exing, 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 ow around apping foils and wings. Dierent
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 eciently solve the ow around apping
foils and wings at low Reynolds numbers, for ow conditions corresponding to the
scale of ying insects. The improved and implemented mesh motion technique,
is used to solve the ow around apping wings in hovering and forward apping
ight. Additionally, a preliminary study of the inuence of wing exing has been
conducted as well.
The overall conclusions of this research are given in section 9.1. Secondly, the
conclusions on the assessment of dierent mesh motion techniques are drawn in
section 9.2. Section 9.3 presents the conclusions of a apping wing under hovering
conditions, for two-dimensional airfoil as well as for the three-dimensional wing.
Section 9.4 shows the results for forward apping ight, while section 9.5 describes
the conclusions of the preliminary study to understand the eects of wing exing.
Additional recommendation are made in Section 9.6.
9.1 Overall conclusions
The following overall conclusions are drawn:
1. The ow around a apping wing, at the scale relevant to insect ight, can
be solved accurately using advanced mesh motion techniques in existing ow
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 modied stress equation. How-
ever, mesh motion based on radial basis function interpolation is much more
computationally demanding. Dierent ways to improve its eciency are
discussed.
3. The wing kinematic pattern has a large inuence on the forces in two-
dimensional hovering and forward ight.
4. In forward ight, thrust generation is the result of the forward tilting of the
wing.
5. In two- and three-dimensional apping ight, the leading-edge vortex is a
very important lift enhancement mechanism.
6. The Rossby number, which describes the wing stroke curvature, has a large
inuence on the time-averaged lift force. A low Rossby number signicantly
increases lift as the result of an induced spiralling leading-edge vortex.
7. Wing exing may play an important role in insect ight 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 rst method solves the Laplace equation with
a variable diusion coecient, which is used to control the nal mesh quality.
Secondly, a modied 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 modied
stress equation, the mesh quality is not sucient for apping wing cases,
where the rotation angles are large. However, these methods are very ecient
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, dierent radial basis functions, concerning the RBF mesh motion,
are compared. The RBF mesh quality provides superior mesh quality over
the Laplace and modied 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 apping ight 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, dierent methods are implemented to increase its eciency. 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 apping ight
9.3.1 Two-dimensional hovering
The eects of wing motion kinematics on the aerodynamic characteristics of hov-
ering insect ight have been investigated by means of two-dimensional numerical
ow 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 ight.
Four dierent kinematic models, with dierent complexity, have been analysed.
Two of these models, pure harmonic motion and the Roboy experimental kine-
matics have been used extensively in literature. The most prominent aspects of
the Roboy kinematic model are the sawtooth displacement and the trapezoidal
angle of attack. The third model represents the actual fruit y kinematics as ob-
served in experiments and the last one was a modication of the latter, chosen
to investigate the eect of symmetry. The fruit y 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 coecient was matched. The following conclusions
are drawn about the two-dimensional hovering simulations:
1. It was found that the realistic fruit y wing kinematics result in signicantly
lower drag at similar lift compared with the simplied wing kinematic models
used in literature. The trend that the fruit y kinematics increases aerody-
namic performance agrees with the predictions of the quasi-steady theory,
but the numerical ow simulations provide a more complete quantitative
analysis of the ow behaviour.
2. It was shown that the dierence in performance in terms of mean lift-to-drag
ratio between the dierent kinematic models was signicant. The mean
aerodynamic drag at equal lift of the fruit y models is about 49% lower
compared to the Roboy model and about 29% lower with respect to the
harmonic model.
182 Conclusions and recommendations
3. The sawtooth amplitude used in the Roboy model has a small eect on
the mean lift but the mean drag is aected signicantly. 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 simplication used by the Roboy, 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 y model is not
aecting 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 signicant 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 eect on the mean lift and mean drag. However, the eective
angle of attack is altered such that the deviation leads to levelling of the force
distribution over the apping cycle.
9.3.2 Three-dimensional hovering
A numerical model has been developed for solving the ow around a three-dimen-
sional wing with complex wing kinematics. Dierent aspects, relevant to three-
dimensional apping wing aerodynamics, have been studied. First, the ow around
a dynamically scaled model wing is solved for dierent 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 dierent 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 gure-of-eight pattern. Finally, the three-dimensional ow is com-
pared with the two-dimensional studies performed on apping forward ight. The
following conclusions are drawn with respect to the three-dimensional simulations,
concerning hovering ight:
1. The eect of a variation in angle of attack is such that the maximal lift
occurs at a mid-stroke angle of = 45

. However, the performance, in


terms of maximal lift-to-drag ratio was found to be maximal at = 30

.
The apping 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 apping ight 183
strong at suciently high angles of attack > 30

. At = 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 signicantly with increasing Ross-
by number. At Ro = 130, a nearly translating wing, the lift decreases
with 24.7%, compared to the apping 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 signicantly
reduced in size and strength for the translating wing at Ro = 130, compared
to the apping wing. In addition, the leading-edge vortex rolls-up to form
two tip vortices instead of one in case of the apping wings. This causes
both lift and drag to be signicantly lower at Ro = 130.
3. It was seen that for increasing Reynolds number, both time-averaged lift
and drag increases, but the dierences become smaller. The eect 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 apping
wing at Ro = 3.2. For both apping (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 signicant 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 signicant 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
eects, however, lead to a rm and stably attached leading-edge vortex.
5. Three dierent wing tip patterns, caused by deviation, have been studied,
such as the gure-of-O, gure-of-U and the gure-of-eight. The gure-
of-O wing tip pattern leads to an asymmetric variation in eective angle of
attack, leading to dierences in drag for the up- and downstroke. Nonethe-
less, the eect of this pattern on the time-averaged forces coecients is
negligible. On the other hand, the gure-of-U pattern does inuence the
forces in a non-benecial way. The eective angle of attack is signicantly
decreased during mid-stroke. The time-averaged lift coecient 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 gure-of-eight aects the forces most, how-
ever, it depends on the starting direction. Two types are considered, the
rst starting the downstroke with an downward deviation motion, whereas
the type 2 starts upwards. The eective angle of attack is being inuenced
in such a way that the type 1 gure-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 gure-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 apping ight
9.4.1 Two-dimensional forward apping
A numerical model for two-dimensional ow has been used to investigate the ef-
fect of motion kinematics on the vortex dynamics around an ellipsoid foil subjected
to prescribed apping 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 apping
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 apping 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 ow is periodic.
9.4.2 Three-dimensional forward apping
In addition to the investigation of the eects of Reynolds number, Rossby number
and wing kinematics in hovering ight, a preliminary study is performed to com-
pare two- and three-dimensional forward ight. Both two- and three-dimensional
simulations are dynamically scaled using the radius of gyration to justify compar-
ison. The three-dimensional simulations, concerning forward ight 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 exing 185
1. For the case without rotation the three-dimensional simulations result in
14.5% less lift compared to its two-dimensional counterpart. This dierence
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 signicant 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 apping 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 dierent 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 dierence 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 signicant 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 apping aerodynamics.
9.5 Preliminary conclusions on wing exing
A exing deformation has been applied to a plunging airfoil in two-dimensional
forward ight and to a three-dimensional apping wing during hovering ight.
Concerning the exible airfoil in forward ight, a comparison is made with a
plunging airfoils, with additional rotation. The following conclusion are drawn:
1. The two-dimensional exing has a comparable eect on the ow and forces as
rotation of the airfoil. With increasing exing amplitude, the larger eective
angle of attack leads to the generation of negative drag or thrust. Besides,
the exing shape of the airfoil is important. A quarter-cosine shaped exing
results in signicantly higher thrust, while lift decreases.
2. The exing deformation was also applied to a three-dimensional apping
wing in hovering ight. Two exing directions were considered, spanwise and
chordwise exing. While keeping the exing amplitude constant, the chord-
wise exing aects the ow signicantly. For a chordwise exing amplitude
of A
f
= 0.2 the time-averaged lift and drag coecients are signicantly de-
creased and increased, respectively, such that the lift-to-drag decreases with
24.8%. The spanwise exing does not show a comparable inuence, only
-2.6%.
186 Conclusions and recommendations
9.6 Recommendations
The current thesis describes a comparison of dierent mesh deformation methods
and development of an improved method, based on radial basis function interpo-
lation. Additionally, these methods are used to solve the ow around two- and
three-dimensional apping airfoils and wings under hovering and forward ight
conditions. The following recommendation can be made for further research.
The eciency 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 signicantly reduced.
Additionally, the implementation of parallel iterative solver techniques may
increase its eciency 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 ow around multiple apping foils
or wings can be investigated. The current thesis describes the ow elds in-
duced by a single two-dimensional foil or three-dimensional wing. If multiple
wings are modelled, i.e. a dragony, the ow will be more complex and the
force development may be aected to certain extent.
Investigate the eects of transition and turbulence modelling on the ow
elds. The current research considered scales at the laminar ow regime. It
would be interesting to study the eects of turbulence on the force develop-
ment and vortex dynamics induced by a apping three-dimensional wing.
Investigate in more detail the vortex wake synchronisation in three-dimen-
sional forward ight. Concerning two-dimensional forward apping ight,
this thesis describes the vortex wake synchronisation of a apping foil for
dierent kinematic parameters. It would be interesting to study the vor-
tex wake synchronisation of a three-dimensional apping wing. The main
diculty will be to dene a proper framework to identify and quantify the
three-dimensional vortex structures to study the pattern formation.
The eects of wing deformation could be investigated in more detail using
more complex exing models. The current thesis describes a preliminary
investigation of the eects of pre-dened wing exing. Further research could
introduce more degrees of freedom in the exing model, possibly based on
real insects. This may lead to advanced wing shapes, which result in optimal
aerodynamic eciency, compared to rigid wings.
It would be interesting to use uid-structure interaction methods to study
the inuence of the ow 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 ow solver and the accompanying mesh motion method. The
main diculty will be the coupling of ow and structure, since the scaled
density of a fruit y wing is of similar order as the surrounding uid. 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-ing 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 apping wings
To solve the ow around apping wings using Computational Fluid Dynamics
(CFD), it is important to create high quality meshes. Three dierent tools are used
for that purpose. The rst, 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 dicult since dening 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 ow 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 apping 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
specied 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 simplied geometries.
with simplied 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 simplied cases. The computational
domain is specied by points, lines and blocks, see gure A.1. The connections be-
tween the points (p1p4) are dened by lines, which can be curved. Additionally,
the domain consists of blocks (b1 b8), specied by the lines, accordingly. The
relation between the points, lines and blocks need to be specied in a dictionary
le, within the OpenFOAM

case. The grading to dene the mesh resolution can


be set on all topology lines. For a uniform grading the mesh around a block is
shown in gure 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

ow 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 diculty.
A.3 Gambit
The commercial ow 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 gure 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 ows. 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 dierent 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 dierent compared to the other
grid generators, blockMesh and Gambit. In gure A.4, the inner red squared
block is snapped to the adjacent boundary
b
, and all other red squared regions
192 Grid generation for apping wings

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-dened mesh topology, such that the
mesh quality is optimised.
represent structured grid blocks. An illustration of the resulting grid is shown in
gure 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 apping wing, as illustrated in gure A.6.
A.5 Conclusions
Three dierent mesh generators have been described, blockMesh, Gambit and
GridPro

. The mesh generator, blockMesh, supplied with OpenFOAM

, is easy
to use for simplied problems. In the current thesis, blockMesh is used to gen-
erate the two-dimensional meshes, which were used to validate the ow 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 ows around stationary and plunging cylinders. Using the third
grid generator, GridPro

, a user-dened topology needs to be specied to optimise


the multiple blocks for the structured mesh generation. This topology procedure
is versatile, but dierent compared to the other packages. GridPro

is used for
the three-dimensional simulations around apping 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 dierent 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 dierent ow solvers have been used, the open-
source OpenFOAM

and the commercial package Fluent

. For both ow 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 nite volume based CFD solver. This section deals with the dif-
ferent ow 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 rst-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 ow, 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 ow 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 rst-order implicit time integration. The validation cases without mesh motion
uses second-order time integration. The multi-grid settings worked ne 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 dened
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 suciently small value. These values are convergence criterions 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
xed 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 apping wing
simulations a xed number of 10 iterations was used. Furthermore the complete
solution is written to hard-disk several times per apping 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, dierent ow 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 strengths of OpenFOAM

is the intuitive way of programming. Briey, two


dierent 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 species that the equations are solved for all time-steps until the specied
end time is reached, runTime.end(). Line 5-9 denes 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 diusion
198 Flow solver settings
terms, i.e. ddt and div are explicit functions to generate the matrix system for
a given velocity, pressure and ux eld. 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 dened, calculating the total energy for every
time instance available. Line 3-9 reads the velocity eld from a given solution, at
the specic time directory. The actual calculation is performed at line 17, where
fvc::domainIntegrate is an implicit function, calculating the sum over all nite-
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 elds 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 specic application, the uses should take a look in the
source-code of OpenFOAM

to nd similar pieces of code.


All ow solvers that are developed for this thesis, are based on icoDyMFoam,
slightly extended to use force output modications, or to make use of modied
mesh motion solvers. icoDyMFoam solves the unsteady incompressible laminar
Navier-Stokes equations for a Newtonian uid. Therefore, the dierent terms
of this equation need to be discretised accordingly, e.g. diusion, 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 dierent solvers were specied for the pressure equation, velocity equation
B.3 OpenFOAM

solver settings 199


Description Code Dierencing scheme
General interpolation - second-order linear
Temporal discretisation ddt(U) second-order backward
Gradient discretisation div(phi,U) second-order linear
Diusion discretisation laplacian(nu,U) second-order linear
Convection discretisation grad(p) Gamma, SuperBee, Koren,
Van Leer or linear
Table B.4 Dierencing methods for dierent terms in the transport equation. The tem-
poral, gradient, diusion 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 denite 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 dierent 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-eld aerodynamics in
insect hovering ight: an integrated computational study, The Journal of Exper-
imental Biology 211, 239257.
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), 13811388.
Beckert, A. & Wendland, H. (2001), Multivariate interpolation for uid-
structure-interaction problems using radial basis functions, Aerospace Science and
Technology 5(2), 125134.
Birch & Dickinson, M. H. (2003), The inuence of wing-wake interactions
on the production of aerodynamic forces in apping ight., The Journal of Ex-
perimental Biology 206, 22572272.
Birch, J. M. & Dickinson, M. H. (2001), Spanwise ow and the attachment
of the leading-edge vortex on insect wings, Nature 412, 729733.
Birch, J. M., Dickson, W. B. & Dickinson, M. H. (2004), Force production
and ow structure of the leading-edge vortex on apping wings at high and low
reynolds numbers., The Journal of Experimental Biology 207, 10631072.
Blom, F. J. (2000), Considerations on the spring analogy, International Jour-
nal for Numerical Methods in Fluids 32, 647669.
202 Bibliography
Blondeaux, P., Fornarelli, F. & Guglielmini, L. (2005a), Vortex struc-
tures generated by a nite-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 apping foils mimicking sh-
like locomotion, Physics of Fluids 17, 11360112.
Boer de, A., van der Schoot, M. S. & Bijl, H. (2007), Mesh deformation
based on radial basis function interpolation, Computers & Structures 85, 784795.
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, 546554.
Bos, F. M., Lentink, D., van Oudheusden, B. W. & Bijl, H. (2008), In-
uence of wing kinematics on aerodynamic performance in hovering insect ight,
Journal of Fluid Mechanics 594, 341368.
Bos, F. M., van Oudheusden, B. W. & Bijl, H. (2010a), Mesh deforma-
tion techniques for apping ight, Computer Methods in Applied Mechanics and
Engineering (Submitted).
Bos, F. M., van Oudheusden, B. W. & Bijl, H. (2010b), Three-dimensional
vortical structures in apping ight, 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, 138.
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 identication schemes, Journal of Fluid Mechanics
535, 189214.
Combes, S. A. & Daniel, T. L. (2003a), Flexural stiness in insect wings i
& ii, Journal of Experimental Biology 206, 29792997.
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, 29993006.
Bibliography 203
Degand, C. & Farhat, C. (2002), A three-dimensional torsional spring analogy
method for unstructured dynamic meshes, Computers & Structures 80, 305316.
Dickinson, M. H. (1994), The eects of wing rotation on unsteady aerody-
namic performance at low Reynolds number., The Journal of Experimental Biol-
ogy 192, 179206.
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, 100106.
Dickinson, M. H. & G otz, K. G. (1993), Unsteady aerodynamic performance
of model wings at low reynolds numbers, The Journal of Experimental Biology
174, 4564.
Dickinson, M. H., Lehmann, F.-O. & Sane, S. P. (1999), Wing rotation
and the aerodynamic basis of insect ight., Science 284, 19541960.
Dickson, W. B. & Dickinson, M. H. (2004), The eect of advance ratio on
the aerodynamics of revolving wings, Journal of Experimental Biology 207, 4269
4281.
Donea, J. (1982), An arbitrary lagrangian-eulerian nite element method for
transient uid-structure interactions, Computer Methods in Applied Mechanics
and Engineering pp. 689723.
Dong, H., Mittal, R., Bozkurttas, M. & Najjar, F. (2005), Wake structure
and performance of nite aspect-ratio apping 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 ight, Phil.
Trans. Roy. Soc. London. B 305, 1181.
Ellington, C. P., van den Berg, C., Willmott, A. P. & Thomas, A. L. R.
(1996), Leading-edge vortices in insect ight, Nature 384, 626630.
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, 231245.
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-ight maneuvers in Drosophila, Science 300, 295298.
204 Bibliography
Ginsberg, J. H. (1998), Advanced Engineering Mechanics, second edn, Cam-
bridge University Press.
Guglielmini, L. & Blondeaux, P. (2004), Propulsive eciency of oscillating
foils, European Journal of Mechanics B/Fluids 23, 255278.
Guilmineau, E. & Queutey, P. (2002), A numerical simulation of vortex
shedding from an oscillating circular cylinder, Journal of Fluids and Structures
16, 773794.
Helenbrook, B. T. (2003), Mesh deformation using the biharmonic operator,
International Journal for Numerical Methods in Engineering 56 (7), 10071021.
Henderson, R. D. (1995), Details of the drag curve near the onset of vortex
shedding, Physics of Fluids 7 (9), 21022104.
Hirsch, C. (1988), Numerical Computation of Internal and External Flows, Vol-
ume I: Fundamentals of Numerical Discretization, John Wiley & Sons.
Hover, F. S., Haugsdal, . & Triantafyllou, M. S. (2004), Eect of angle
of attack proles in apping foil propulsion, Journal of Fluids and Structures
19, 3747.
Hu, D. L., Chan, B. & Bush, J. W. M. (2003), The hydrodynamics of water
strider locomotion, Nature 424, 663666.
Hunt, J. C., Wray, A. A. & Moin, P. (1988), Eddies, stream, and convergence
zones in turbulent ows, in Center for Turbulence Research Report, pp. 193208.
Isogai, K., Fujishiro, Saitoh, T., Yamamoto, M., Yamasaki, M. & Mat-
subara, M. (2004), Unsteady three-dimensional viscous ow simulation of a
dragony hovering, AIAA Journal 42, 20532059.
Issa, R. I. (1986), Solution of the implicitly discretized uid ow equation by
operator splitting, Journal of Computational Physics 62, 4065.
Jakobsson, S. & Amoignon, O. (2007), Mesh deformation using radial ba-
sis functions for gradient-based aerodynamics shape optimization, Computers &
Fluids 36, 11191136.
Jasak, H. (1996), Error analysis and estimation in the nite volume method with
applications to uid ow, 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. & Tukovic, Z. (2004), Automatic mesh motion for the unstructured
nite volume method, Journal of Computational Physics (Submitted).
Jasak, H., Weller, H. G. & Gosman, A. D. (1999), High resolution nvd
dierence scheme for arbitrary unstructured meshes, International Journal for
Numerical Methods in Fluids 31, 431449.
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 identication of a vortex, Journal of
Fluid Mechanics 285, 6994.
Johnson, A. A. & Tezduyar, T. E. (1994), Mesh update strategies in par-
allel nite element computations of ow problems with moving boundaries and
interfaces, Computer Methods in Applied Mechanics and Engineering 119, 7394.
Juntasaro, V. & Marquis, A. J. (2004), Comparative study of ux-limiters
based on must dierencing scheme, International Journal of Computational Fluid
Dynamics 18 (7), 569576.
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 ow over
an airfoil, Journal of Fluid Mechanics 102, 315328.
Knupp, P. M. (2003), Algebraic mesh quality metrics for unstructured initial
meshes, Finite Elements in Analysis and Design 39, 217241.
Koren, B. (1993), A robust upwind discretisation method for advection, dif-
fusion and source terms, Numerical Methods for Advection-Diusion Problems
p. 117.
Kuzmin, D. & Turek, S. (2004), High-resolution fem-tvd schemes based on a
fully multidimensional ux limiter, Journal of Computational Physics 198, 131
158.
Lehmann, F.-O. (2004), The mechanisms of lift enhancements in insect ight,
Naturwissenschaften 91, 101122.
Lehmann, F.-O., Sane, S. P. & Dickinson, M. (2005), The aerodynamic
eect of wing-wing interaction in apping insect wings, The Journal of Experi-
mental Biology 208, 30753092.
Lentink, D. (2003), Inuence of airfoil shape on performance in insect ight,
Masters thesis, Delft University of Technology.
206 Bibliography
Lentink, D. (2008), Exploring the Biouiddynamics of Swimming and Flight,
PhD thesis, Wageningen University.
Lentink, D. & Dickinson, M. H. (2009a), Biouiddynamic scaling of ap-
ping, spinning and translating ns and wings, Journal of Experimental Biology
212, 26912704.
Lentink, D. & Dickinson, M. H. (2009b), Rotational accelerations stabilize
leading edge vortices on revolving y wings, Journal of Experimental Biology
212, 27052719.
Lentink, D. & Gerritsma, M. (2003), Inuence of airfoil shape on performance
in insect ight, 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 apping foil that models animal swimming
and ight, The Journal of Experimental Biology 211, 267273.
Lesoinne, M. & Farhat, C. (1996), Geometric conservation laws for ow prob-
lems with moving boundaries and deformable meshes, and their impact on aeroe-
lastic computations, Computer Methods in Applied Mechanics and Engineering
134, 7190.
Lewin, G. C. & Haj-Hariri, H. (2003), Modelling thrust generation of a
two-dimensional heaving airfoil in a viscous ow, Journal of Fluid Mechanics
492, 339362.
Lighthill, M. J. (1969), Hydromechanics of aquatic animal propulsion, Annual
Review of Fluid Mechanics 1, 413446.
Liu, H. & Kawachi, K. (1998), A numerical study of insect ight, Journal of
Computational Physics 146, 124156.
Lohner, R. & Yang, C. (1996), Improved ale mesh velocities for moving bod-
ies, Communications in Numerical Methods in Engineering 12 (10), 599608.
Luo, G. & Sun, M. (2005), The eects of corrugation and wing planform on
the aerodynamic force production of sweeping model insect wings, Acta Mech
Sinica 21, 531541.
Lu, Y. & Shen, G. X. (2008), Three-dimensional ow structures and evolution
of the leading-edge vortices on a apping wing, The Journal of Experimental
Biology 211, 12211230.
Maxworthy, T. (1979), Experiments on the weis-fogh mechanism of lift gener-
ation by insects in hovering ight. part 1. dynamics of the ing, Journal of Fluid
Mechanics 93, 4763.
Bibliography 207
Michelson, R. C. (2008), Test and evaluation of fully autonomous micro air
vehicles, ITEA Journal 29, 367374.
Mittal, R. & Iaccarino, G. (2005), Immersed boundary methods, Annual
Review of Fluid Mechanics 37, 239261.
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 ows, International
Journal of Heat and Mass Transfer 15, 1787.
Pedro, G., Suleman, A. & Djilali, N. (2003), A numerical study of the
propulsive eciency of a apping hydrofoil, International Journal for Numerical
Methods in Fluids 42, 493526.
Peskin, C. S. (2002), The immersed boundary method, Acta Numerica
pp. 479517.
Poelma, C., Dickson, W. B. & Dickinson, M. H. (2006), Time-resolved
reconstruction of the full velocity eld around a dynamically-scaled apping wing,
Experiments in Fluids 41 (2), 113.
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 ight, The Journal of
Experimental Biology 205, 15071518.
Rendall, T. C. S. & Allen, C. B. (2008a), Ecient 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
uid-structure coupling via ecient localised implementation, International Jour-
nal for Numerical Methods in Engineering 78 (10), 11881208.
Rendall, T. C. S. & Allen, C. B. (2008c), Unied uid-structure interpo-
lation and mesh motion using radial basis functions, International Journal for
Numerical Methods in Engineering 74 (10), 15191559.
Rhie, C. M. & Chow, W. L. (1983), Numerical study of the turbulent ow
past an airfoil with trailing-edge separation, AIAA Journal 21, 15251532.
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 ight force by a
apping wing: lift and drag production, The Journal of Experimental Biology
204, 26072626.
Sane, S. P. & Dickinson, M. H. (2002), The aerodynamic eects of wing
rotation and a revised quasi-steady model of apping ight, The Journal of Ex-
perimental Biology 205, 10871096.
Shyy, W., Lian, Y., Tang, J., Liu, H., Trizila, P. & Stanford, B. (2008a),
Computational aerodynamics of low reynolds number plunging, pitching and ex-
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 yers, 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), 5258.
Srygley, R. B. & Thomas, A. L. R. (2002), Unconventional lift-generating
mechanisms in free-ying butteries, Nature 420, 660664.
Stepniewski, W. Z. & Keys, C. N. (1984), Rotary-wing aerodynamics, Dover.
Sun, M. & Tang, J. (2002), Unsteady aerodynamic force generation by a
model fruit y wing in apping motion, The Journal of Experimental Biology
205, 5570.
Sweby, P. K. (1984), High resolution schemes using ux-limiters for hyperbolic
conservation laws, SIAM Journal of Numerical Analysis 21, 9951011.
Taylor, G. K., Nudds, R. L. & Thomas, A. L. R. (2003), Flying and
swimming animals cruise at a strouhal number tuned for high power eciency,
Nature 425, 707711.
Templin, R. J. (2000), The spectrum of animal ight: insect to pterosaurs,
Progress in Aerospace Sciences 36, 393436.
Tennekes, H. & Lumley, J. L. (1972), A rst 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
apping 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), Dragony ight: free-ight and tethered ow visualizations
reveal a diverse array of unsteady lift-generating mechanisms, controlled primarily
via angle of attack, The Journal of Experimental Biology 207, 42994323.
Triantafyllou, G. S., Triantafyllou, M. S. & Grosenbaugh, M. A. (1993),
Optimal thrust development in oscillating foils with application to sh propulsion,
Journal of Fluids and Structures 7, 205224.
Tukovic, Z. & Jasak, H. (2007), Updated lagrangian nite 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, 15471576.
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, 329340.
Van Leer, B. (1979), Towards the ultimate conservative dierence scheme v.
a second order sequel to godunovs method, Journal of Computational Physics
32, 101.
Wang, Z. J. (2000a), Two dimensional mechanism for insect hovering, Physical
Review Letters 85(10), 22162219.
Wang, Z. J. (2000b), Vortex shedding and frequency selection in apping ight,
Journal of Fluid Mechanics 410, 323341.
Wang, Z. J. (2005), Dissecting insect ight, Annual Review of Fluid Mechanics
37, 183210.
Wang, Z. J. (2008), Aerodynamic eciency of apping ight: analysis of a
two-stroke model, Journal of Experimental Biology 211, 234238.
Wang, Z. J., Birch, M. B. & Dickinson, M. H. (2004), Unsteady forces
and ows in low reynolds number hovering ight: two-dimensional computations
vs robotic wing experiments, The Journal of Experimental Biology 207, 461474.
Wang, Z. J. & Przekwas, A. J. (1994), Unsteady ow 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 ight. i.
basic principles of insect ight: a critical review, Phil. Trans. Roy. Soc. London
239 (667), 415458.
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 trager, Technical report.
Wendland, H. (1998), Error estimates for interpolation by compactly sup-
ported radial basis functions of minimal degree, Journal of Approximation Theory
93, 258272.
Wendland, H. (1999), On the smoothness of positive denite and radial func-
tions, Journal of Computational and Applied Mathematics 101, 177188.
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), Dening a universal and continuous strouhal-
reynolds number relationship for the laminar vortex shedding of a circular cylin-
der, Physics of Fluids 31(10), 27422744.
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, 10731085.
Williamson, C. H. K. & Roshko, A. (1988), Vortex formation in the wake
of an oscillating cylinder, Journal of Experimental Biology 2, 355381.
Willmott, A. P., Ellington, C. P. & Thomas, A. L. R. (1997), Flow
visualization and unsteady aerodynamics in the ight of the hawkmoth, Manduca
Sexta, Phil. Trans. Roy. Soc. London 305, 303316.
Wood, R. J. (2008), The rst takeo of a biologically inspired at-scale robotic
insect, IEEE Trans. on Robotics 24, 341347.
Young, J. & Lai, J. C. S. (2008), Simulation and parameter variation of
apping-wing motion based on dragony hovering, AIAA Journal 46, 918924.
Bibliography 211
Zhou, Y. C. & Wei, G. W. (2003), High resolution conjugate lters for the
simulation of ows, Journal of Computational Physics 189, 159179.
Zuijlen van, A. H. (2006), Fluid-Structure Interaction Simulations, Ecient
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 apping-wing insect hovering ight at unsteady ow, International Journal for
Numerical Methods in Fluids 53, 18011817.
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 appende vleugels te verklaren,
een raadsel. Tot een paar decennia terug. Experimenten lieten zien dat appende
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 appende vleugels te vergroten. Met deze kennis kun-
nen micro vliegtuigjes worden ontworpen en verbeterd. Het eect 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 appende 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 ecientie. Deze method is
tevens gebruikt om de stroming rond appende vleugels door te rekenen.
De stroming rond appende 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 commerciele 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 ecient, 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
gemplementeerd, 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 geevalu-
eerd. Er is geen connectiviteit van het rooster nodig, zodat deze methode een-
voudig kan worden toegepast op ongestructureerde roosters. Om de ecientie 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 aegt.
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 ap amplitude slechts een klein eect heeft op de gemiddelde
liftkracht, maar dat de weerstand aanzienlijk wordt benvloed. 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 ap periode. De extra bump van
de invalshoek, die aanwezig is bij de fruitvlieg vleugelbeweging, benvloedt de lift
niet beduidend. De laatste realistische vleugelbeweging karakteristiek, de deviatie,
heeft slechts een marginaal eect op de gemiddelde lift- en weerstandskrachten in
deze tweedimensionale studie. Desalniettemin verandert de eectieve 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 appende vlucht. Een numeriek model
215
voor een tweedimensionale stroming is gebruikt om het eect van de vleugelbe-
weging op de werveldynamica te onderzoeken voor een variatie van dimensieloze
golengte, 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 append
vleugelproel 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 appende 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
aegt, 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
ecientie 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 guur. 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
appende vlucht.
Het laatste onderwerp dat beschreven wordt in dit proefschrift, is het eect van
vleugel vervorming. Een vooraf gedenieerde vleugelvervorming is toegepast op
een tweedimensionaal translerend vleugelproel en een driedimensionale vleugel bij
condities voor stilhangende vlucht. Het vervormende vleugelproel in vooruit ap-
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 appende vleugel, representatief voor de vlucht van insecten en vogels, worden
benvloed 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 ap-
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 apping 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 scientic 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 Queens University of Belfast, Aero-
nautical department in the United Kingdom. He numerically investigated the
interaction between induced vortical ow and a turbulent boundary layer. He
obtained his Master of Science degree at the Aerodynamics department in 2005,
entitled Inuence of wing kinematics on performance in insect ight, 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 apping 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

You might also like