KTH Royal Institute Of Technology

The Department of Civil and Architectural Engineering

Degree project in Structural Engineering and Bridges
Second cycle, 30 credits

CFD Analyses for Wind Load

Assessment
Wind-induced vibrations in the Bomarsunds Bridge hangers

Fatiha El Azrak

November 2023
Authors
Fatiha El Azrak
fatihaea@[Link]

Place for Project

KTH School of ABE
SE-100 44 Stockholm, Sweden

Examiner
Prof. Raid Karoumi
KTH Royal Institute of Technology

Prof. Cecilia Surace

Polytechnic Of Turin

Supervisor
Prof. John Leander
KTH Royal Institute of Technology

Eng. Marco Civera

Polytechnic Of Turin
 “To my father,

my first and greatest teacher,

whose guidance and strength

shaped my path.”
Abstract
The trend in recent years of building longer and slender bridge components has
introduced new challenges to ensure their stability and strength. This master
thesis focuses on the effects of wind-induced vibrations in the context of long,
slender arch bridge components, particularly the recently constructed Bomar-
sunds Bridge in Åland, Finland.

The primary goal of this study is to comprehensively analyze the dynamic wind
effects on the hangers due to the vortex shedding phenomenon, as the resulting
vibrations pose potential risks to its safety and structural integrity. The slender
hangers of the bridge, close to the centre of the span, have exhibited significant
vibrations, necessitating an in-depth investigation to understand the bridge’s re-
sponse to wind forces. Computational fluid dynamics (CFD) simulations were
performed using ANSYS Fluent to estimate more accurate aerodynamic quanti-
ties. Using CFD analysis, the behaviour of a given hanger section subjected to
wind flow can be described. In this way, it was possible to calculate the aerody-
namic coefficients that characterize that given section (i.e. Strouhal number (St ),
drag coefficient (CD ), etc). By integrating advanced computer simulations and
CFD analysis, the research addresses the complex challenges of investigating the
vortex-induced vibration (VIV) phenomena at different wind speeds.

The results showed an inconsistent trend for drag coefficients at varying wind
speeds and lower drag for geometries with rounded edges, with an average value
of drag coefficient of 1.60. The study highlighted the significant dependence of the
Strouhal number on wind speed, varying from 0.129 for a wind speed of 2.5 m/s to
0.063 for a wind speed of 30 m/s, challenging traditional geometry-based estima-
tions for this parameter. The drag frequency for each wind speed investigated is
twice as high as the lift frequency, showing that at wind speeds of 7.5 m/s a drag
frequency close to the fundamental transversal frequency of 6.6 Hz of the longest
hanger is reached. This leads to the conclusion that for this particular case study,
the headwind response is much more critical than the crosswind response. These
findings can be used to implement effective measures to mitigate wind-induced
vibrations in the studied hanger along the critical direction.

By analyzing the complex vortex shedding phenomenon, the study contributes

valuable insights into the field of wind engineering. This research plays a key role
in ongoing efforts to design robust, safe, and resilient bridge structures.
Keywords: Vortex shedding phenomenon, Computational Fluid Dynamics (CFD),
Aerodynamic coefficients, Drag, Strouhal number, Vortex-Induced Vibration (VIV),
Headwind - Crosswind response.

7
 Acknowledgments
I extend my heartfelt gratitude to Professor Raid Karoumi and Professor John
Leander for entrusting me with the opportunity to delve into this captivating case
study. Your invaluable guidance, unwavering support, and mentorship have been
instrumental during the completion of this thesis. Your depth of knowledge and
expertise has been a source of inspiration and amazement. I aspire to emulate
your dedication and passion. Thank you for rekindling my passion for knowledge
and research.

I am grateful to Professor Cecilia Surace and Doctor Marco Civera for their sup-
port, encouragement, and assistance. Thank you for embracing my proposal to
work on this case study, and exhibiting unwavering confidence in my skills and
abilities. Your advice and expertise proved to be indispensable in accomplish-
ing this research project. I extend my heartfelt appreciation to Marco for your
patience and promptness in offering valuable guidance. Your support has been
invaluable.

I am immensely grateful to Prof. Sasan Sadrizadeh for his invaluable guidance

throughout the fluid simulations. His mentorship has been instrumental in shap-
ing my understanding of this field. I extend my heartfelt appreciation to Nan Hu
for her invaluable advice and precise guidance in refining the model, ensuring its
accuracy and efficacy.

I would like to express my sincere gratitude to my family for their constant

encouragement, understanding, and support throughout this journey.

Stockholm, November 2023.

Fatiha El Azrak
Contents

1 Introduction 2
1.1 Scope of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Wind Load Analysis in Bridge Design 2

2.1 Wind Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.1.1 Turbulence and the Boundary Layer . . . . . . . . . . . . 2
2.1.2 Streamlined and Bluff Body . . . . . . . . . . . . . . . . . 3
2.1.3 Flow Patterns in Fluid Dynamics . . . . . . . . . . . . . . 4
2.2 Aerodynamic Quantities . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.1 Drag Coefficient: CD . . . . . . . . . . . . . . . . . . . . . 6
2.2.2 Lift Coefficient: CL . . . . . . . . . . . . . . . . . . . . . . 7
2.2.3 Moment Coefficient: CM . . . . . . . . . . . . . . . . . . . 8
2.2.4 Strouhal number: St . . . . . . . . . . . . . . . . . . . . . 8
2.2.5 Experimental Determination of Aerodynamic Coefficients . 9
2.3 Aeroelastic Wind Effects on Bridges . . . . . . . . . . . . . . . . . 13
2.3.1 Flutter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3.2 Galloping . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.3 Torsional Divergence . . . . . . . . . . . . . . . . . . . . . 15
2.3.4 Buffeting . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.5 Vortex Shedding Phenomenon . . . . . . . . . . . . . . . . 16
2.4 Wind Design Standards . . . . . . . . . . . . . . . . . . . . . . . 16
2.4.1 Wind Velocity . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.2 Vortex Shedding and Aerodynamic Instability . . . . . . . 18

3 Case study: Bomarsund Bridge 21

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

ii
CONTENTS

3.2 The Bridges Evolution . . . . . . . . . . . . . . . . . . . . . . . . 23

3.3 The Bridges Design . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.4 Bridge Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

4 Numerical Analysis of Hangers in the Bridge Structure 26

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.2.1 Geometric Modeling . . . . . . . . . . . . . . . . . . . . . 27
4.2.2 Material Properties . . . . . . . . . . . . . . . . . . . . . . 28
4.2.3 Mesh Modeling . . . . . . . . . . . . . . . . . . . . . . . . 28
4.2.4 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . 29
4.2.5 Analysis Setup . . . . . . . . . . . . . . . . . . . . . . . . 29
4.3 Results Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.4 Result Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

5 Ansys Fluent: 2D & 3D CFD analysis 36

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.2.1 Realizable k - ϵ . . . . . . . . . . . . . . . . . . . . . . . . 37
5.2.2 Wall treatment . . . . . . . . . . . . . . . . . . . . . . . . 37
5.2.3 Computational Wind Domain . . . . . . . . . . . . . . . . 39
5.2.4 Boundary Condition . . . . . . . . . . . . . . . . . . . . . 39
5.2.5 Mesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.2.6 Solver Method . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.3 Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.4 Transient Solution . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.4.1 Non-rounded edges cross-section . . . . . . . . . . . . . . . 51
5.4.2 Interaction between different sections . . . . . . . . . . . . 57
5.4.3 3D Transient Analysis of hanger H16 . . . . . . . . . . . . 60
5.4.4 Wind Load Calculation . . . . . . . . . . . . . . . . . . . . 63
5.5 Results Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6 Conclusions and further research 69

6.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.2 Further research . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

Bibliography 72

iii
 CONTENTS

A Abaqus Appendix 76
A.1 H16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
A.2 H15-16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
A.3 H14-16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
A.4 H14-17 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
A.5 H12-19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

B ANSYS Fluent Appendix 91

B.1 ANSYS Fluent 2D Simulation . . . . . . . . . . . . . . . . . . . . 91
B.1.1 2D Steady simulation . . . . . . . . . . . . . . . . . . . . . 92
B.1.2 2D Transient Simulations . . . . . . . . . . . . . . . . . . . 93
B.2 ANSYS Fluent 3D Simulations . . . . . . . . . . . . . . . . . . . 94

iv
List of Figures

2.1 Transition of the laminar boundary layer on a flat plate into a fully
turbulent boundary layer. Reproduced after Cengel et al. [2]. . . . 3
2.2 Fundamental differences between the flows of streamlined versus
bluff bodies. Reprinted from "Introduction to Aerospace Flight
Vehicles " by J. Gordon Leishman, [4]. . . . . . . . . . . . . . . . 4
2.3 Different flow states over a circular cylinder, dependence on the
Reynolds number. Reprinted from "Introduction to Aerospace
Flight Vehicles " by J. Gordon Leishman [4]. . . . . . . . . . . . . 5
2.4 Skidmore, Owings & Merrill LLP. (n.d.). Wind Tunnel Testing for
Supertall Buildings. . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.5 Strouhal number for rectangular cross-sections with sharp corners,
Eurocode 4, E1, [1]. . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3.1 New constructed Bomarsund Bridge, adapted from "Langerbalks-

bro över Bomarsund på Åland", Stålbyggnad . . . . . . . . . . . . 22
3.2 Goole maps location of the Bomarsund Bridge, with Google maps
photo of the old concrete arch bridge. . . . . . . . . . . . . . . . . 22
3.3 Beam profile for the arch element and 3D view of the beam. . . . 25
3.4 The tension member with its I-shaped cross-section and 3D view. 25
3.5 Hanger cross-section and 3D view. . . . . . . . . . . . . . . . . . . 25

4.1 Bomarsund Bridge: south side arch . . . . . . . . . . . . . . . . . 27

4.2 The five models were analyzed by using the software Abaqus. . . . 32

5.1 CFD rectangular section of the hanger . . . . . . . . . . . . . . . 40

5.2 Wind domain characteristics with relative boundary conditions. . 40
5.3 Mesh configuration of the wind domain for the steady simulation. 45
5.4 Wind flow velocity magnitude: contouring and pathlines. . . . . . 46

v
 LIST OF FIGURES

5.5 Wind flow contouring: static pressure (a) and turbulence intensity
(b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.6 Not rounded edges, wind speed pathlines varying from null speed
in blue, to the maximum speed reached during the simulation in red. 48
5.7 Edge r = 1 cm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.8 Edge r = 1.5 cm . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.9 CD and CM at different speeds of rectangular sections with rounded
and not rounded edges. . . . . . . . . . . . . . . . . . . . . . . . 50
5.10 Transient solution: aerodynamic coefficient at different speeds. . . 52
5.11 Transient solution: estimated peak frequency and Strouhal number 54
5.12 Time domain spectra and frequency domain spectra at wind speed
of 27 m/s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.13 Aerodynamic coefficients vs Reynolds . . . . . . . . . . . . . . . . 56
5.14 Transient solution: Velocity contour at different wind speeds. . . . 57
5.15 Transient solution: wind pathlines at different speeds. . . . . . . . 58
5.16 Transient solution: Drag coefficient of different models. . . . . . . 59
5.17 Transient solution multi-section model: lift signal and frequency
domain spectra for wind speed of 10 m/s. . . . . . . . . . . . . . . 60
5.18 Transient solution multi-section model: velocity magnitude con-
tour for a wind speed of 30 m/s, H15 - H16 - H17. . . . . . . . . . 61
5.19 Transient solution multi-section model: velocity pathlines of the
three hangers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.20 Transient analysis: 3D model for ANSYS simulation. . . . . . . . 62
5.21 Comparison between 2D model and 3D model results of drag co-
efficient, lift frequency and Strouhal number. . . . . . . . . . . . . 63
5.22 2.5 m/s Wind Velocity Contour. . . . . . . . . . . . . . . . . . . . 64
5.23 Velocity magnitude vectors for wind speed of 2.5 m/s. . . . . . . . 64

A.1 Mode shapes model H16 (continued) . . . . . . . . . . . . . . . . 77

A.1 Mode shapes model H16 (continued) . . . . . . . . . . . . . . . . 77
A.1 Mode shapes model H16 (continued) . . . . . . . . . . . . . . . . 77
A.1 Mode shapes model H16 (continued) . . . . . . . . . . . . . . . . 78
A.1 Mode shapes model H16 (continued) . . . . . . . . . . . . . . . . 78
A.2 Mode shapes model H15-16 (continued) . . . . . . . . . . . . . . . 79
A.2 Mode shapes model H15-16 (continued) . . . . . . . . . . . . . . . 80
A.2 Mode shapes model H15-16 (continued) . . . . . . . . . . . . . . . 80

vi
A.2 Mode shapes model H15-16 (continued) . . . . . . . . . . . . . . . 80
A.2 Mode shapes model H15-16 (continued) . . . . . . . . . . . . . . . 81
A.3 Mode shapes model H14-16 (continued) . . . . . . . . . . . . . . . 82
A.3 Mode shapes model H14-16 (continued) . . . . . . . . . . . . . . . 83
A.3 Mode shapes model H14-16 (continued) . . . . . . . . . . . . . . . 83
A.3 Mode shapes model H14-16 (continued) . . . . . . . . . . . . . . . 83
A.3 Mode shapes model H14-16- (continued) . . . . . . . . . . . . . . 84
A.4 Mode shapes model H14-17 (continued) . . . . . . . . . . . . . . . 85
A.4 Mode shapes model H14-17 (continued) . . . . . . . . . . . . . . . 86
A.4 Mode shapes model H14-17 (continued) . . . . . . . . . . . . . . . 86
A.4 Mode shapes model H14-17 (continued) . . . . . . . . . . . . . . . 86
A.4 Mode shapes model H14-17- (continued) . . . . . . . . . . . . . . 87
A.5 Mode shapes model H12-19 (continued) . . . . . . . . . . . . . . . 88
A.5 Mode shapes model H12-19 (continued) . . . . . . . . . . . . . . . 89
A.5 Mode shapes model H12-19 (continued) . . . . . . . . . . . . . . . 89
A.5 Mode shapes model H12-19 (continued) . . . . . . . . . . . . . . . 89
A.5 Mode shapes model H12-19 (continued) . . . . . . . . . . . . . . . 90

vii
List of Tables

4.1 Mode Shapes and Frequencies for Different Models (Part 1) . . . . 33

4.2 Mode Shapes and Frequencies for Different Models (Part 2) . . . . 34

5.1 Fluid Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

5.2 Rectangular cylinder: not rounded edges. . . . . . . . . . . . . . . 45
5.3 Aerodynamic coefficients for different model geometries with dif-
ferent radii. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.4 Aerodynamic coefficients for the steady solution with section r
= 1cm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.5 Transient solution multi-section model: data extracted for each
speed simulated . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.6 Parameters for Wind load Calculation According to Eurocode . . 65

A.1 Model H16: First ten Frequencies . . . . . . . . . . . . . . . . . . 76

A.2 Model H15-16: First ten Frequencies . . . . . . . . . . . . . . . . 79
A.3 Model H14-16: First ten Frequencies . . . . . . . . . . . . . . . . 82
A.4 Model H14-17: First ten Frequencies . . . . . . . . . . . . . . . . 85
A.5 Model H12-19: First ten Frequencies . . . . . . . . . . . . . . . . 88

B.1 2D ANSYS Fluent simulation settings . . . . . . . . . . . . . . . . 92

B.2 Transient solution: Aerodynamic coefficients for rectangular cross
section. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
B.3 Transient solution: Estimated peak frequencies and related ampli-
tudes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
B.4 Transient solution: Strouhal number . . . . . . . . . . . . . . . . 93
B.5 Transient solution data extracted for each speed simulated . . . . 94
B.6 Transient 3D analysis: simulation extracted data. . . . . . . . . . 94

viii
Chapter 1

Introduction

Bridges pose an enthralling challenge in civil engineering. Designers continu-

ously aspire to surpass records by constructing longer, lighter, and more slender
structures. Presently, bridges have evolved to embrace ever more intriguing and
innovative materials and shapes. As well as being a landmark of engineering ad-
vancement, they symbolize the progress of a nation. Bridges are constructed to
withstand progressively demanding loads and varied weather conditions. There-
fore, their design and maintenance are fundamental to their safety, integrity, and
efficiency.

The sophistication of bridge architecture poses challenges, particularly in con-

fronting natural forces. Wind loading emerges as a relevant obstacle for bridges,
triggering dynamic responses that may result in irreversible damage. Wind engi-
neering enhances knowledge of the impact of wind on bridges, unveiling intricate
phenomena like vortex shedding that could jeopardize their structural integrity.

In this thesis, the focus is on a particular newly built steel bridge, located on
an island between Finland and Sweden. Its hangers are affected by excessive
vibration at ordinary wind speeds for the area in which it is located. The goal
is to reveal how these hangers respond to the action of the wind, by using Com-
putational Fluid Dynamics (CFD) investigations. The process involves explor-
ing steady-state aerodynamics, observing transient responses at different wind
speeds, diving into the complexity of 3D interactions, and applying standards
such as Eurocode for wind load calculations.

2
1.1. SCOPE OF THE THESIS

1.1 Scope of the thesis

The objective of this research is to comprehend the interaction between wind
and hangers in a dynamic environment. The primary research questions focus
on the critical range of fundamental frequencies of the structure, considering var-
ious models with different numbers of hangers. This will help to identify their
interaction and its effect on the global response. Moreover, the research aims
to understand the dynamic responses of hangers in different time domains and
varying wind conditions with a logical progression of information. This study
within the field of aerodynamics aims to identify the effects of wind-induced
forces on significant aerodynamic parameters, specifically the aerodynamic coef-
ficient, Strouhal number, and shedding frequencies. The research also explores
the influence of geometric features, including rounded edges of the hangers.

A comparative study is a crucial aspect of the research as it investigates the

interactions between hangers in multi-section models and identifies the dispari-
ties in aerodynamic performance between two-dimensional and three-dimensional
models of hangers at different wind velocities. The research aligns with practical
applications, and the research questions cover the application of Eurocode stan-
dards for wind load calculations. The study examines the agreement between
these standards and simulated critical wind speeds for vortex shedding in bridge
hangers. It also explores the influence of incorporating shedding frequencies on
calculations of inertial wind force.

1.2 Limitations
While embarking on this exploration, it’s crucial to acknowledge the limitations.
The models, though intricate, have been simplified to achieve a balance between
complexity and computational feasibility, the study focuses only on the longest
hangers of the bridge. The digital domain has limitations and the results may not
be universally applicable to all bridge types and conditions: the dimensions and
the inclinations are of the chosen section of the Bomarsund Bridge. Additionally,
the field of CFD presents numerous intriguing possibilities that cannot all be
explored due to the restrictions of time and workload involved in a thesis research
study. Moreover, the computational tools used for simulation are powerful yet
demanding, requiring lengthy processing times and high computational loads. To
obtain satisfactory results in a reasonable time frame, simplification of the models

3
1.2. LIMITATIONS

created is often necessary.

This study aims to provide practical insights for the fields of bridge and wind
engineering. Through the exploration of wind-hanger dynamics, it seeks to in-
form improved design practices, enhance safety protocols, and contribute to the
development of bridges that not only stand tall but also sway elegantly with the
wind.

1
Chapter 2

Wind Load Analysis in Bridge

Design

2.1 Wind Flow

A flow over a body induces two important forces, a force in the direction of the
flow, called drag that tends to move the body into the same direction, and a
force normal to the first, that tends to move the stationary body in the opposite
direction, called lift.

2.1.1 Turbulence and the Boundary Layer

Turbulence and the boundary layer are fundamental aspects of fluid dynamics.
The boundary layer was introduced by Ludwig Prandtl, which revolutionized the
study of fluid mechanics, [2]. His solution consists of dividing the fluid flow into
two zones, an inviscid outer flow zone, and an inner flow zone, the boundary
layer. It is a region near a surface where the fluid velocity changes from zero to
a maximum. Understanding the boundary layer is critical because it’s where the
effect of the viscous layer can’t be neglected and many wind-induced effects on
structures originate. Inside the boundary layers, three different flow-type zones
can be distinguished, to define them it is important to introduce a fundamental
parameter in the aerodynamic field: the Reynolds number (Re) Eq. (2.1). This
is a dimensionless quantity used to define the ranges of flow regimes.

2
2.1. WIND FLOW

ρ·V ·L
Re = (2.1)
µ

Where Re is the Reynolds number, ρ is the density of the fluid, V is the velocity
of the fluid, L is the characteristic length, and µ is the dynamic viscosity of the
fluid.

Flow regimes are classifications of fluid flow, typically categorized as laminar or

turbulent flow with a transition face between these two, see Figure 2.1. Laminar
flow is characterized by smooth and orderly movement of fluid particles on parallel
streamlines with minimal mixing, [3]. This flow regime prevails at low Reynolds
(Re < 103 ). The turbulent regime is the most complex, there is no streamlined
or predicted direction for the fluid particles, and the flow is chaotic and irregular.
It occurs at high Reynolds. Between the two regimes, there is a third transitional
regime, where both regimes coexist, [2].

Figure 2.1: Transition of the laminar boundary layer on a flat plate into a fully
turbulent boundary layer. Reproduced after Cengel et al. [2].

2.1.2 Streamlined and Bluff Body

In the field of fluid dynamics, the shape and size of an object placed in a fluid flow
has a significant influence on the fluid dynamics. They can be categorized into
two macro groups, streamlined bodies and bluff bodies. The latter have distinc-
tive characteristics: square, non-aerodynamic, and irregular. Their unique shape
causes phenomena such as flow separation, turbulence, and pressure changes both
around and downstream of the object. A critical parameter in analyzing the be-
haviour of such bodies is the drag. It is the sum of the resistance of the profile:

3
 2.1. WIND FLOW

pressure drag and skin friction drag. The extent to which either of these compo-
nents dominates the total drag depends on the shape of the body. If the viscous
component, i.e. that due to skin friction, is the main contributor to the drag
forces, the body is classified as streamlined. On the other hand, if pressure is a
significant contributor to drag, the body falls into the bluff category, [4]. It is
important to understand the fluid dynamic difference between streamlined and
bluff bodies. Bluff bodies tend to induce flow separation, typically at the cor-
ners of the body, see Figure 2.2. This separation leads to the creation of a large
wake region characterized by low pressure, which causes high drag. Aerodynamic
bodies, on the other hand, keep the flows together, reducing flow separation and
consequently the drag [4].

Figure 2.2: Fundamental differences between the flows of streamlined versus bluff
bodies. Reprinted from "Introduction to Aerospace Flight Vehicles " by J. Gordon
Leishman, [4].

2.1.3 Flow Patterns in Fluid Dynamics

The study of flow patterns is essential to understanding the flow regime and the
shape of the body. At different Reynolds numbers, different flow patterns can be
identified, characterized by different types of turbulence, see Figure 2.3. What
occurs is a wake, a region of disturbed flow that trails behind an object placed
in a fluid medium. This wake often contains turbulence and eddies of varying
sizes. It is essential to recognize the characteristics of this wake, the frequency
of oscillation, and the associated drag. Furthermore, by observing the eddies
that form within the wake and observing their symmetries and asymmetries, it
is possible to identify the Reynolds range of the flow.

4
2.1. WIND FLOW

Figure 2.3: Different flow states over a circular cylinder, dependence on the
Reynolds number. Reprinted from "Introduction to Aerospace Flight Vehicles
" by J. Gordon Leishman [4].

At low Reynolds numbers, where viscous effects dominate, the flow is typically
described as a creeping or Stokes flow, [4], with no turbulence or vorticity. At
slightly higher, but still relatively low, Reynolds numbers, a stable pair of symmet-
rical eddies tends to form in the wake of the body, resulting in a relatively smooth

5
 2.2. AERODYNAMIC QUANTITIES

flow. As the Reynolds number increases, the behaviour of the wake changes. The
eddies alternately spread out in a Von Kármán vortex road pattern, creating an
unstable flow. Although this flow is unstable, no time-averaged unstable lift is
generated due to the asymmetry in the flow, [4]. At higher Reynolds numbers, the
flow patterns become more turbulent, and boundary layer separation can occur
further downstream. The result is a narrower wake, corresponding to less drag.
At very high Reynolds numbers, the inertial force overcomes viscosity resistance.
The flow patterns form a wake that closely resembles an inviscid flow. In these
cases, the object experiences no resistance, neither lift nor drag. This condition
is known as d’Alembert’s paradox, [4], which highlights the complex interaction
between viscous and inertial effects in flow dynamics.

2.2 Aerodynamic Quantities

The study of aerodynamic coefficients, including drag, lift, moment coefficients,
and Strouhal number, plays a vital role in the design of wind loads on the struc-
ture. These coefficients provide valuable insight into the forces and moments
experienced by the structure when subjected to fluid flow. In this section, the
theoretical background and experimental methods for the determination of drag,
lift, moment coefficients, and Strouhal are reviewed. Theoretical models and
equations are discussed to provide a basis for understanding these quantities. In
addition, experimental techniques such as wind tunnel testing and computational
fluid dynamics (CFD) are presented as a means of determining these coefficients
in practice.

2.2.1 Drag Coefficient: CD

The drag coefficient (CD ) is a dimensionless quantity that characterizes the resis-
tance encountered by an object subjected to a fluid flow, it is representative of the
resistance parallel to the free flow direction. It plays a crucial role in determining
the aerodynamic drag force acting on the object. The drag coefficient of a sta-
tionary body is influenced by various factors, including the shape of the object,
the Reynolds number, the surface roughness, and the characteristic dimension of
the body d. The last parameter characterizes the body’s size in a flow field. This
dimension varies depending on the shape of the body. Bluff bodies with rectan-
gular or square cross-sections have a characteristic dimension d which can be the

6
2.2. AERODYNAMIC QUANTITIES

width, height, or diagonal length depending on the orientation. For bodies with
circular cross-sections such as cylinders or pipes, the characteristic dimension is
typically the diameter. When dealing with complex geometries or irregular bluff
bodies, it may be necessary to determine the characteristic dimension by select-
ing a representative length or scale that effectively captures the overall size or
geometry of the body.

FD
CD = 1 (2.2)
2
ρU 2 A

It is determined by the Eq. (2.2), where ρ and U are the density and velocity of
the fluid, respectively, and A is the reference area; in the case of 2D analysis, the
area is given by the unit value multiplied by the characteristic dimension d.

2.2.2 Lift Coefficient: CL

The lift coefficient (CL ), Eq. (2.3), is another important parameter used to charac-
terize the aerodynamic behaviour of an object immersed in a fluid flow. It is more
relevant for three-dimensional simulations, since it is a force normal to the drag
resistance, in the opposite direction of gravity. In two-dimensional analysis, it is a
resistance acting perpendicular to the flow direction, in the cross-wind direction,
oscillating around a null value, and is responsible for generating a transversal
force on the object. It represents the ratio between the lift force experienced by
the object and the dynamic pressure of the fluid. The lift force and the resis-
tance force are nothing more than the two components, one parallel and the other
perpendicular to the flow of the total resistance of the solid subjected to a fluid
flow.

FL
CL = 1 (2.3)
2
ρU 2 A

Similar to the coefficient of resistance, the lifting coefficient is influenced by var-

ious factors, including the shape of the object, the Reynolds number, and the
roughness of the surface. The lift coefficient, Eq. (2.3), plays a crucial role in
determining the lift force (FL ) generated by an object, which can be any solid
surface. By studying the lift coefficient and its behaviour for cylinders of different
shapes and in variable flow conditions, it is possible to obtain information on the
aerodynamic forces acting on structures and improve design and performance in

7
 2.2. AERODYNAMIC QUANTITIES

many applications. It is a fundamental value for the determination of another

crucial parameter, the Strouhal number, which is described in Section 2.2.4.

2.2.3 Moment Coefficient: CM

The moment coefficient (CM ), Eq. (2.4), is a dimensionless parameter used to
describe the moment (M ) suffered by a body subjected to fluid flow. It provides
information on the distribution of moments along the length of the object, indi-
cating how this one responds to aerodynamic loads. Factors such as geometry,
flow conditions, and fluid properties influence the value of the coefficient. It rep-
resents the relationship between the moment experienced by the object and the
dynamic pressure of the fluid flow. A higher momentum coefficient indicates a
greater tendency of the section to rotate or twist under the influence of fluid flow
and therefore more vulnerable to torsional effects. This parameter can be used to
assess the structural stability, and integrity of a structural part during the design
phase.

M
CM = 1 (2.4)
2
ρU 2 d2

2.2.4 Strouhal number: St

The Strouhal number (St ), Eq. (2.5), is a dimensionless number used in fluid dy-
namics to characterize the flow-induced oscillations or vortex shedding behind an
object placed in a fluid flow. It is defined as the ratio of the shedding frequency
f times the characteristic length d and the flow velocity U . It helps in under-
standing the dynamic response and the energy transfer between the fluid and
the structure. The Strouhal number is named after the Czech physicist Vincenc
Strouhal, who first studied this phenomenon in the late 19th century.

fd
St = (2.5)
U

The Strouhal number is particularly relevant when studying vortex shedding be-
hind an object. When fluid flows around an obstructing body, such as a cylinder,
it can lead to the formation of eddies. If the shedding frequency (f ) of these
eddies is similar to the natural frequency of the body, this can lead to resonance,
causing vibrations and even structural damage. The shedding frequency is deter-
mined by the lift signal as explained in Section 2.2.2, so for its determination it

8
2.2. AERODYNAMIC QUANTITIES

is necessary to use computational fluid dynamic analysis.

In the context of the Eurocode and structural design, the Strouhal number is
used to evaluate the effects of wind-induced vibrations on structures. If the
shedding frequency is equal to the natural frequency of the structure, resonance
can occur, which is an important consideration in ensuring the safety and stability
of structures.

At high Strouhal numbers, the flow is dominated by oscillations. At low Strouhal

numbers, the fast-moving fluid sweeps away the oscillations. At intermediate
Strouhal numbers, approximately 0.2 - 0.3, the shedding of vortices downstream
takes place, [8]. For flows around bluff bodies, the Strouhal number usually ranges
from 0.1 to 0.3. However, it is important to note that these values can vary based
on factors such as the Reynolds number, which characterizes the flow regime, and
the specific dimensions and shape of the cross-section.

In practice, the Strouhal number is used to predict and mitigate wind-induced

vibrations in structures. By adjusting the shape, size, or orientation of objects
and structures, the shedding frequency can be influenced and the likelihood of
resonance reduced.

2.2.5 Experimental Determination of Aerodynamic Coeffi-

cients
The experimental determination of aerodynamic coefficients is a fundamental as-
pect of the field of aerodynamics. It involves performing physical experiments
to measure and quantify the aerodynamic forces acting on an object immersed
in a fluid flow, typically air or water with some fluid colourants, to detect the
wave that forms behind the object. In the most common case, the body is scaled
and fixed, for more advanced experiments a small motion is added to its support.
These experiments can be challenging due to factors such as controlled Reynolds
number, flow turbulence, and, in particular, model scaling issues. However, they
are essential as the aerodynamic coefficients provide crucial insight into the perfor-
mance, stability, and behaviour of the object in different flow regimes. A variety
of methods are used, ranging from wind tunnel testing to full-scale flight testing,
the latter being the most recent in the aerospace field, and the data obtained is
used to inform decisions relating to performance, stability, control, and overall
design optimization.

9
 2.2. AERODYNAMIC QUANTITIES

Figure 2.4: Skidmore, Owings & Merrill LLP. (n.d.). Wind Tunnel Testing for
Supertall Buildings.

Wind Tunnel testing

One of the best-known methods of aerodynamic testing is the wind tunnel. A

historic and classic method, it allows measurements to be taken on models of
various scales. Its advantages are undoubtedly its versatility and controllability.
In practice, a scale model is placed in the specially designed gallery. The tunnel
has a very simple structure, a long and narrow chamber with fans at the end
that allows the wind to be simulated under controlled conditions, see Figure 2.4.
The geometry of the tunnel is such that there is a regular and constant flow,
minimizing disturbances that could affect the validity of the measurements. Inside
the tunnel, there is usually support on which the test body is positioned, which
is usually fixed. This fixed body model simplifies the test procedure.

The tests are carried out at different speeds and also at different angles of at-
tack to reproduce the real regimes and flow conditions. As far as the speed is
concerned, the variation procedure is quite simple as it can be controlled by the
fans at the ends of the tunnel. A more advanced method is to use the variable
speed drive, which would allow more precise control of the wind parameters. For
the actual measurement of the aerodynamic forces and moments generated by
the flow, a range of instruments and sensors are used, with varying degrees of
sensitivity, accuracy, and precision. The classics are force balances, which make
it possible to measure the total force and moment exerted by the model. Another

10
2.2. AERODYNAMIC QUANTITIES

basic measurement parameter is the pressure distribution on the surface of the

model, which is monitored by pressure sensors. A third essential measurement
tool is strain gauges, which allow the deformation of the model to be measured
and the response of the model to the applied wind force to be studied. The
gauges presented here are the basis for a simplified test, of course, there are other
instruments used depending on the purpose of the simulation. It is important to
focus on the dimensions of these instruments, the scale models have dimensions
of the order of a centimetre, rarely a meter, so to influence the measurements as
little as possible, the instruments must be of millimetre size.

The next step is data collection, using automated data acquisition systems. These
collect data in real time, recording all the measurement parameters in such a way
as to provide a reading of the model’s response over time. The final step is
data analysis for the determination of the drag, moment and lift coefficients. To
verify the accuracy of the data obtained, the measured forces and moments are
calibrated against reference parameters such as dynamic pressure and the charac-
teristic dimensions of the object. In this calibration phase, statistical principles
are used to minimize errors and uncertainties in the measurements made. Finally,
the normalized force and moment values are compared with theoretical models or
reference data. This step in the experimental process is fundamental as it allows
the data obtained to be validated.

Wind tunnel testing has several advantages, including controllability and repeata-
bility. It also allows accurate observation of the vorticity effects of the flow and
how this interacts with the structure at the boundary layer level. This type of
test has important validity and reliability in the field of aerodynamics. Further-
more, the simplicity of this type of test makes it possible to effectively reproduce
different flow conditions on any type of object to be analyzed, of course always
within the limits of the tunnel scale.

The wind tunnel test is a standardized test governed by specific guidelines and
protocols. These cover every single step of the test, starting with the dimensions of
the tunnel, the mounting, and the scale of the object to be placed in the tunnel.
It is a very sensitive test and requires extreme accuracy. These standards are
often provided by organizations such as the American Society of Civil Engineers
(ASCE) or the International Organisation for Standardisation (ISO).

11
 2.2. AERODYNAMIC QUANTITIES

Computational Fluid Dynamics

A more modern branch is Computational Fluid Dynamics (CFD), a numerical

simulation technique used to analyze and predict the behaviour of fluid flows
around objects. It has become a powerful tool for determining the drag coefficients
of structures and for visualizing the dynamic effects of flows. Nowadays, there is
a wide range of software that can be used to simulate any type of structure under
any condition.

In very simplified terms, CFD analysis involves solving the equations governing
fluid flow, i.e. the Navier-Stokes equations, using numerical methods. A flow
domain of the simulated fluid is defined, the dimensions of which must be much
larger than those of the object being traversed to obtain an accurate simulation.
This domain is then divided into small computational cells using a meshing pro-
cess, and the equations for each cell are solved iteratively to obtain a solution
that represents the flow behaviour.

To perform a CFD analysis, a simulation setup is first established. This involves

a number of steps, the first of which is the definition of the geometry of the
structure and the surrounding domain, where the domain is sized according to
a basic parameter which is the characteristic length of the structure. This is
followed by the meshing step, which consists of selecting the type, order, and size
of the meshing unit. The boundary conditions and fluid properties must then
be defined. This gives the basic model on which to run simulations of the flow
of interest. The actual simulation parameters include the choice of turbulence
model, which captures the effects of turbulence in the flow, and the discretization
scheme, which approximates the governing equations. These parameters have
a significant impact on the accuracy and reliability of the results and must be
carefully selected based on the specific flow conditions and the desired level of
accuracy.

Once the CFD simulation has been performed, the post-processing phase begins,
which consists of extracting the desired data. Post-processing also includes visu-
alization of flow patterns, pressure distribution, velocity, turbulence, and other
flow variables to provide detailed information on the dynamics of the flow and
how it interacts with the structure. These visualizations and contours help to
identify aerodynamic concepts that are analytically very complex to determine,
such as flow separation, eddies, their distribution and symmetry, and other flow

12
2.3. AEROELASTIC WIND EFFECTS ON BRIDGES

phenomena that contribute to the generation of aerodynamic forces and moments.

It is important to note that CFD analysis is subject to some limitations and

uncertainties, its simplifications can lead to completely incorrect solutions. The
accuracy of the results depends on the quality and size of the mesh, especially
near the walls of the structure. The turbulence model chosen and the hypotheses
made during the simulation setup influence the final data, it is necessary to
have the settings adapted to the model to be simulated. There are guidelines
to follow, but each analysis has its peculiarities. Validation of CFD results with
experimental data is often performed to ensure their reliability and to identify
any discrepancies. However, CFD is a valuable and efficient tool for evaluating
the effects of flows on an object, especially when combined with wind tunnel tests
other experimental methods, or literature results.

2.3 Aeroelastic Wind Effects on Bridges

Aeroelasticity is a multidisciplinary field that brings together aerodynamics, struc-
tural dynamics, and mechanical engineering. In the field of forces, it is the science
that studies the interaction between the elastic, inertial, and aerodynamic force
fields acting on a structure. Aeroelasticity is a very complex field that can include
other factors such as thermal effects, which complicate the study even more. This
macro discipline can be divided into two areas: static aeroelasticity and dynamic
aeroelasticity, [9]. The former deals with the interaction between elastic and aero-
dynamic forces, while dynamic aeroelasticity adds inertial forces and non-linear
effects to the analysis.

Bridges are structures strongly affected by aeroelastic effects due to their slender
and elongated geometry and the fact that they are immersed in highly dynamic
flow fields. The most common aeroelastic effects on bridges are flutter, galloping,
divergence, and buffeting.

2.3.1 Flutter
Flutter is an aeroelastic phenomenon of dynamic instability resulting from the
interaction of elastic, inertial, and self-excited aerodynamic forces, [10] and is
one of the most critical phenomena for slender structures. What happens when
the critical wind speed is reached is that the structure begins to oscillate in a
divergent manner until it fails. These vibrations typically manifest as torsional

13
 2.3. AEROELASTIC WIND EFFECTS ON BRIDGES

or vertical oscillations. The critical speed of the corresponding wind is called

the flutter speed, which causes divergent oscillations in the structure, with the
same frequency and theoretically increasing amplitude tending towards an infi-
nite value, [11]. In practice, these oscillations are stabilized around a maximum
amplitude value. During the divergence process, the structure can suffer various
types of damage at key points, leading to failure after a few cycles of flutter, [10].
Flutter is a phenomenon that has been widely studied in the aerospace field and
became of critical importance after the catastrophic collapse of the Tacoma Nar-
rows Bridge in 1940. The ultimate goal of flutter verification is to determine the
minimum speed at which this phenomenon is triggered; lower or higher speeds
will not cause the structure to vibrate divergently.

2.3.2 Galloping

Galloping is a type of dynamic aeroelastic instability, often referred to as a vortex-

induced vibration (VIV) phenomenon, which particularly affects large and thin
baffle decks. It can lead to structural instability, fatigue damage, and poten-
tially catastrophic failure. This type of phenomenon generally occurs when wind-
generated eddies detach from the deck surface, creating alternating eddies typical
of the Kármán vortex phenomenon. The latter generally occurs at relatively nor-
mal wind speeds, [12], underlining the importance of studying this phenomenon
for structures such as bridges.

The motion consists of an oscillation perpendicular to the direction of flow, it

is a kind of self-sufficient phenomenon, i.e. once the oscillation phenomenon
starts, the lift increases, which further increases its amplitude. This effect, also
known as the feedback effect, is because all mechanical and aerodynamic damping
is completely eliminated, [10]. It is possible to distinguish two different types
of galloping on a slender structure. The so-called hard-galloping has a higher
amplitude, the swaying is more structured and intense and it is also predictable.
This type of galloping is quite dangerous for the structure, leading to the collapse
of the structure. Soft-galloping is more unpredictable, the swaying is more gentle,
and the oscillation is not constant, but it is still dangerous for the structure and
can lead to fatigue damage over time.

14
2.3. AEROELASTIC WIND EFFECTS ON BRIDGES

2.3.3 Torsional Divergence

Torsional Aeroelastic Divergence (ATD) is a dynamic phenomenon consisting of
a torsional movement out of the plane of the deck. The definition of ATD is very
complex as it is a set of deformations and rotations that the bridge deck undergoes
at a critical wind speed. At this critical wind speed, there is an unlimited growth
of the structural rotation, [17], this theoretical definition is unrealistic in the field
of bridge design and therefore it is more reasonable to define the critical divergence
of the wind speed as that at which the torsion of a body increases rapidly up to
the breaking point, [16]. What happens in practice is that the wind flow not only
generates aerodynamic forces but also a moment in the structure. If the wind
speed is equal to the critical speed, the moment generated by the flow is equal to
that associated with a torsional mode of the structure. The moment due to the
flow increases the torsional angle, inducing an out-of-plane torsional divergence,
there is also a second out-of-plane deformation due to a drag force. The effect of
the divergence is always of the combined type, lateral (vertical) instability, and
torsional divergence, [10]. The vertical movement causes a deterioration of the
torsional stiffness, this aspect is characteristic of the torsional divergence [15].
This aeroelastic effect is also very common in bridges with long, flexible, and
slender spans.

2.3.4 Buffeting
Buffeting is an aeroelastic phenomenon that can have significant effects on very
long-span bridges. Length and flexibility play a key role in amplifying the buffet-
ing effect. Unlike the other effects described so far, this one is characterized by
high-frequency oscillations that are always due to turbulence effects, in particular
the separation of these eddies that occurs when the flow meets the surface of the
body through which it passes. This effect is nothing more than the result of the
resonance that occurs for mode shapes at very high frequencies. All aeroelastic
phenomena are caused by turbulent winds interacting with the bridge structure.
The vibrations are very fast due to the high frequency of the bridge or parts of it.
The effect of high-intensity turbulence can lead to slamming, which can severely
damage the structure or key components of the structure, leading to complete
failure.

All these dynamic aeroelastic effects become significant when the bridge is slender,
thin, and highly flexible, even long spans are to be avoided, especially when the

15
 2.4. WIND DESIGN STANDARDS

wind conditions of the geographical position of the bridge vary over a very wide
range. Such structures are certainly more pleasing to the eye and require less
material, but if they have not been thoroughly checked from an aeroelastic point
of view, they may be subject to damage or, in the worst case, catastrophic failure.

2.3.5 Vortex Shedding Phenomenon

Vortex shedding causes wind-induced vibration, which is important to consider
when designing and assessing the safety of slender structures like long components
of bridges. This phenomenon occurs when wind interacts with a structure, caus-
ing vortices to alternate from one side to the other. These alternating vortices
generate fluctuating pressures that result in a fluctuating force perpendicular to
the wind direction, [13].

Vortex shedding has a significant impact on slender and lightly damped struc-
tures, like long hangers. When the shedding frequencies resonate with the natural
frequencies of the structure, it can cause substantial vibrations. To prevent this,
it is important to ensure that the structure is designed to withstand these vibra-
tions. The frequency of vortex shedding can be determined by using the Strouhal
number, which is a dimensionless number that depends on the structure’s cross-
sectional design and the flow velocity, see Section 2.2.4. Studies on aeroelastic-
ity have shown that the shape, size, and surface characteristics of the structure
deeply impact the interaction between wind and structures. These factors affect
the shedding frequency and the forces acting on the structure. Slender structures
with specific geometric shapes, square or rectangular geometries, may experience
critical vortex-induced vibrations. Therefore, tailored design considerations are
necessary, [14].

2.4 Wind Design Standards

The Wind Load Standards are guidelines that summarise the essential concepts
and checks to be carried out in the design of structures exposed to wind loads.
The aim is to guide the designer in the design of a structure that will ensure its
safety and structural integrity. The choice of the appropriate standard to follow
depends on the geographical location of the project, there are local, national,
and international regulations that provide guidelines based on the specific type
of structure to be designed. In Europe, the design of buildings and civil struc-

16
2.4. WIND DESIGN STANDARDS

tures is governed by the Eurocode, which is divided into different sections, each
with a specific purpose. In particular, Eurocode SS-EN 1991-1-4:2005 focuses
on the effects of wind on structures and is accompanied by national guidelines,
which must always comply with the main European standard but take into ac-
count the specificities of the geographic location. The Eurocode provides several
detailed procedures for calculating wind loads and their effects on different types
of structures, including bridges. Some of these are presented in this section.

2.4.1 Wind Velocity

Wind speed is the first parameter to be determined, a fundamental parameter in
the calculation of wind loads according to the Eurocode standards. The process
involves several stages, including the determination of the basic wind speed, and
the consideration of various coefficients depending on altitude and terrain cate-
gory, all of which should lead to the calculation of the mean wind speed. This is a
representative value of the wind speed to be taken into account when calculating
the wind load.

Basic Wind Velocity (Vb ) The basic wind velocity, denoted as Vb , represents
the 10-minute mean wind velocity at a height of 10 meters above ground level,
considering a geographic zone without obstacles.

Vb = cs · cr · Vb0 (2.6)

Where: cs is the season factor, cr is the wind directional factor, Vb0 is the funda-
mental value of the basic wind velocity. The values of cs and cr are provided in
the National Annex depending on the site-specific conditions, their conservative
value is equal to the unitary value.

Mean Wind Velocity (Vm ) The mean wind velocity (Vm ) is the 10-minute
mean wind velocity at the height of the structural design point. Eurocode pro-
vides a formula to calculate Vm :

Vm = cr (z) · c0 (z) · Vb0 (2.7)

wherecr (z) is the roughness factor, accounts for the variability of the mean wind
velocity at the site of the structure, and depends on the roughness length and

17
 2.4. WIND DESIGN STANDARDS

terrain factor, all coefficients given by the [Link] c0 is the orography

factor, given by the National Annex, its conservative value is the unitary value.
Each of these factors is determined based on site-specific conditions and the height
(z) of the structure.

2.4.2 Vortex Shedding and Aerodynamic Instability

Vortex shedding is one of the most important phenomena to be studied in struc-

tures exposed to wind. The concept of vortex shedding has been extensively
explained previously in Section 2.3.5, it is an aerodynamic effect of the turbu-
lence trail created behind the object crossed by the flow: the alternating eddies
are released from opposite sides of the structure when generating fluctuating
forces perpendicular to the wind direction, resulting in structural vibrations. In
cases where the vortex shedding frequency matches the natural frequency of the
structure, aerodynamic instability can occur as the structure goes into resonance.

Critical Wind Velocity (Vcr ) The critical wind velocity (Vcr ) is the wind
speed at which vortex shedding is most likely to induce aerodynamic instability.
It is a vital parameter in assessing the structural response to wind loads. Eurocode
defines Vcr based on factors like the structure’s geometry and natural frequency,
see also Eq. (2.5).

d · fi
Vcrit,i = (2.8)
St

Where d is the characteristic width of the cross-section, fi is the natural frequency

of the structure for the considering i flexural mode, and St is the Strouhal number,
given in the Eurocode based on the ratio between the two dimensions of the cross-
section, Figure 2.5.

18
2.4. WIND DESIGN STANDARDS

Figure 2.5: Strouhal number for rectangular cross-sections with sharp corners,
Eurocode 4, E1, [1].

The Eurocode sets a limit value above which the effects of vortex shedding must
be studied. If the critical wind speed, Eq. (2.8) for each mode of the structure
is less than or equal to 1.25 times the characteristic mean wind speed over 10
minutes, Eq. (2.7), vortex shedding cannot be neglected and must be analyzed.
The formula used by the Eurocode is given by Eq. (2.9).

Vcrit,i > 1.25 · Vm (2.9)

Vortex shedding load : the vibration caused by vortex shedding induces an

inertial action Fw (s) defined by Eq. (2.10), which is described by the Eurocode.
The guide gives precise instructions on the direction of the force to be applied
perpendicular to the wind direction, along the lift direction.

FW (s) = m(s) · (2 · π · fi,y )2 · ϕ(s) · yF,max (2.10)

Where m(s) represents the vibrating mass per unit length of the structure under
analysis, fi,y denotes the i-th frequency of the mode shape being investigated,
and ϕ(s) corresponds to the mode shape of the structure, normalized to a value
of 1 at the position of maximum displacement. The yF,max is the maximum
displacement related to ϕ(s). One of the approaches given by the Eurocode
consists of calculating the maximum displacement following the Eq. (2.11):

d · K · KW · Clat,0
yF,max = (2.11)
St2 · Sc

Where: d is the characteristic length of the section, K is the mode shape factor,

19
 2.4. WIND DESIGN STANDARDS

Kw is the effective correlation number, its evaluation depends on the mode shape
configuration. The Clat,0 is the lateral coefficient equal to 1.1 for a rectangular
section. St is the Strouhal number and Sc is the Scruton number given by the
Eq (2.12).

2 · δs · mi,e
Sc = (2.12)
ρ · d2
Where mi,e is the equivalent mass for the i-th mode shape. ρ is the air density and
d is the characteristic length. δs is the logarithmic decrement of structural damp-
ing. The approximate values of logarithmic decrement of structural damping in
the fundamental mode of different structural types are given by the Eurocode in
Table F.2, [1].

When all these parameters are defined, the inertial force applies along a specific
length of the beam structure at a position determined by the i-th mode shape
configuration. The Eurocode gives a specific indication of this length.

20
Chapter 3

Case study: Bomarsund Bridge

3.1 Introduction

The Bomarsund Bridge, Figure 3.1, is a newly constructed bridge, located in the
connection between Bomarsund and the island of Prästö, in Åland, Figure 3.2.
The island is located in a geographical position between Finland and Sweden, an
area affected by strong gusts of wind. It is a mainly steel arch structure designed
by the Swedish company WSP. Its construction was completed in the middle of
2023. This bridge has attracted a lot of attention, particularly for the interesting
structural and dynamic challenges it poses, such as high vibrations in the arch
hangers, especially the longer ones located in the mid-span of the bridge. These
excessive vibrations are of critical importance due to their potential to induce
structural fatigue problems, especially in the attachment part, and consequently
threaten the overall integrity of the bridge. Recognizing the importance of the
Bomarsund Bridge in the context of wind engineering, this case study examines
the correlation between the fluid dynamics and the high vibrations in the arch
hangers and the potential consequences of these vibrations on the overall struc-
tural health. This bridge is a very interesting case study, and its monitoring by
sensors and strain gauges could provide data of great importance for the design
of future more resilient and durable bridges in extreme wind conditions such as
those experienced by this bridge.

21
 3.1. INTRODUCTION

Figure 3.1: New constructed Bomarsund Bridge, adapted from "Langerbalksbro

över Bomarsund på Åland", Stålbyggnad

Figure 3.2: Goole maps location of the Bomarsund Bridge, with Google maps
photo of the old concrete arch bridge.

22
3.2. THE BRIDGES EVOLUTION

3.2 The Bridges Evolution

The history of the Bomarsund Bridge is a very interesting story, which has been
reconstructed in detail in the article: ’Langerbalksbro över Bomarsund på Åland’
in the magazine Stålbyggnad. The Bomarsund Bridge is an important part of
Åland’s history, a place of historical significance, and a key element in the re-
gion’s transport network. The original Bomarsund Bridge was a classic concrete
arch bridge with a steel-reinforced deck, a typical engineering design of the time.
However, it was decided to completely replace the bridge as it was not designed
to withstand today’s traffic loads, resulting in durability issues. It had extensive
damage at some key points in the structure. In addition, the previous bridge had
only one lane, a common feature of Åland bridges, which caused traffic problems,
especially during maintenance and resurfacing work. These were the main rea-
sons why the Åland government decided to replace the bridge. In selecting the
design for the new bridge, WSP’s structural engineers were tasked with finding
a solution that could overcome the potential challenges posed by the underlying
rock formations as well as the strong gusts of wind characteristic of the bridge’s
geographical location.

3.3 The Bridges Design

For the new design of the Bomarsund Bridge, WSP engineers chose a Langer beam
arch bridge, a massive and harmonious structure. The decision to use this type of
arch bridge was dictated by the conditions of the rocky subsoil: it was feared that
the rock under the bridge location would not be able to support a classic arch
bridge. The specifications of the new Bomarsund Bridge are impressive, with a
span of 70.4 meters and a total length of 80.8 meters. The height of the arch from
the roadway is just over 12 meters, the longest inclined hanger is over 11 meters
and the distance between the arches is 13 meters. The two arches of the bridge
are stabilized at the top by a K-shaped truss of square hollow steel members.
The bridge has been designed to accommodate different types of traffic, with two
lanes for vehicular traffic plus pedestrian and cycle paths, reaching a clear width
of 9.55 meters.

The designer of the bridge has optimized the design of the bridge to the maximum,
not only trying to find an efficient solution from a mechanical and aesthetic point
of view but also paying particular attention to adopting design solutions that, in

23
 3.4. BRIDGE GEOMETRY

addition to mechanical efficiency and aesthetic harmony, allow optimization in

the use of materials. His innovative solution was to use diagonal hangers made
of hollow square tubes instead of vertical ones, saving around 200 tonnes of steel
in the structure.

Special attention was paid to the realization of the structure, which was man-
ufactured by Nordec in Ylivieska using steel from SSAB in Raahe. The steel
components were delivered to the site in 17-meter units and fully painted, which
reduced the amount of work required at the site. Welding on site was necessary,
but was mainly imitated for the less complex parts to minimize the margin for
error and have more control over the construction of the structure. The Bomar-
sund Bridge is a testament to the fusion of history and innovation, combining the
historical heritage of the Bomarsund Fortress with modern engineering solutions
to create an efficient and harmonious structure that supports the needs of today’s
society.

3.4 Bridge Geometry

The geometry of the bridge is very complex, consisting of two arches thickened
by a grid of inclined hangers, connected horizontally in the upper part by a
bracing system of hollow square members. The deck is cast in concrete in static
interaction with the grid of steel beams.

The case study focuses more on the hangers, Figure 3.5, and the arch parts
connected to them. The arch is constructed with a hollow beam of a particular
shape, as shown in Figure 3.3, with stiffeners on the inside corresponding to the
attachment of the hangers. The elements of the arch cross-section are all welded.
The beam at the base of the arch is composed of I-shaped welded member with
plates welded to the top flange at the hanger connection, Figure 3.4.

24
3.4. BRIDGE GEOMETRY

Figure 3.3: Beam profile for the arch element and 3D view of the beam.

Figure 3.4: The tension member with its I-shaped cross-section and 3D view.

Figure 3.5: Hanger cross-section and 3D view.

25
Chapter 4

Numerical Analysis of Hangers in

the Bridge Structure

4.1 Introduction
Bridges are subject to various environmental loads, including wind forces and
the associated vortices they can generate. These aerodynamic flows can induce
vibrations and compromise structural integrity by causing fatigue damage for
example. This type of load can be more compromising than static loads, under-
standing and modelling these dynamic loads is of crucial importance to better
analyze the behaviour of the structure and achieve an accurate design. This
chapter will present finite element modelling and analysis of hangers using the
Abaqus software, to define natural frequencies and mode shapes to verify wind
load calculations and incorporate the effect of vortex shedding in accordance with
the Eurocode standards.

The objective of this chapter is to estimate natural frequencies and mode shapes,
which play a crucial role in the calculation of wind-induced forces. By accurately
capturing these dynamic characteristics, one can improve the reliability of wind
load evaluations and ensure a range of critical frequencies related to the structure
analyzed. To achieve the goal of the analysis, a progressive modelling approach
was adopted, starting first with a single hanger and gradually increasing complex-
ity by incorporating additional hangers. A total of five models were developed and
analyzed, each representing a different configuration of the hangers attached to
the hollow box arch and the I-shaped tension member. In the upcoming parts of

26
4.2. METHODOLOGY

this chapter, finer aspects of each model will be explored, detailing the modelling
techniques used and analyzing the results obtained.

4.2 Methodology
In this section, a detailed description of the progressive modelling approach is
provided by dictating the various models composed of Nielsen hangers, the hol-
low box arch, and the connecting tension member. In particular, the geometry,
dimensions, and material properties of the bridge components used to model the
structure on the finite element software Abaqus are presented. The analysis was
carried out by focusing on the arch on the south side, from this one a total of five
sub-models were analyzed. The first model focused on a single hanger, specifi-
cally H16, connected to the arch and the tension member. A frequency analysis
was conducted to determine the first ten natural frequencies and corresponding
mode shapes of the hanger. These results served as the basis for further inves-
tigation and validation against Eurocode requirements. Subsequent models were
then developed, gradually introducing additional hangers to simulate real-world
scenarios. Model 2 incorporated hangers H15 and H16, while model 3 included
hangers H14, H15 and H16. By gradually increasing the number of hangers, until
reaching the actual model of eight hangers, from H12 to H19, see Figure 4.1.

Figure 4.1: Bomarsund Bridge: south side arch .

4.2.1 Geometric Modeling

To accurately represent the hangers, the hollow box arch, and the tension member,
a geometric model was created consisting of shell elements. This approach allows
the structural behaviour to be captured while minimizing calculation costs. Shell
elements were chosen for their ability to effectively represent thin-walled struc-

27
 4.2. METHODOLOGY

tures, given their small thicknesses, and also to describe accurately intersecting
shells that exhibit non-linear behaviour. The geometric model was developed
based on the design drawings and specifications of the Bomarsund Bridge. Great
care was taken to accurately reproduce the dimensions, cross profiles, and con-
nection details of the hangers to the gussets, the arch with stiffening plates, and
the tension member. The model took into account the unique geometry and
interaction between the different components of the bridge structure.

4.2.2 Material Properties

To ensure reliable results from the frequency analysis performed on the various
models, it is essential to accurately represent the behaviour of the materials used
in the structure. The material properties have been carefully defined based on
the specific characteristics and specifications of the material used in the construc-
tion. For consistency and simplicity, the same set of material properties has been
assigned to all components. These properties include an elastic modulus (E) of
210 GPa, a Poisson’s ratio (ν) of 0.3, and a density of 7850 kg/m³. These prop-
erties reflect the specific characteristics of the materials used and are essential for
obtaining accurate and meaningful results from the analysis.

4.2.3 Mesh Modeling

To discretize the geometric finite element model, an appropriate meshing strategy
was used to accurately represent the complex geometry and capture the structural
behaviour of the hangers, the hollow box arch, and the tension member. Consid-
ering the unique features of the structure, such as curved elements, sharp edges,
and intersections between shells, a meshing strategy using quadratic-dominated
elements with a free technique and a median axis-based algorithm was employed.
Quadratic-dominated elements were selected for meshing because of their ability
to capture finer details of the structural response than linear elements. These
elements have higher-order shape characteristics, allowing a more accurate rep-
resentation of curved geometries. The free technique, also known as free mesh
or unstructured mesh, provides the flexibility to create irregular, non-uniform
meshes that closely match the geometry. It allows for better resolution of regions
with complex features, particularly in the junction area, including various gus-
sets, hangers, and stiffeners. The median axis-based algorithm helps maintain
mesh quality by ensuring proper element connectivity, reducing element distor-

28
4.2. METHODOLOGY

tion, and improving overall analysis accuracy. It is particularly beneficial when

dealing with complex geometries such as those present in the bridge structure
in question. Particular attention has been given to regions of interest, such as
attachment points, and local mesh refinement techniques have been applied to
these areas to ensure adequate mesh density and accurately capture the fine de-
tails of the dynamic response. Mesh sizes of 0.02 m and 0.025 m were chosen to
provide the best possible finish, taking into account the maximum number of
nodes allowed in Abaqus.

4.2.4 Boundary Conditions

For each of the five models analyzed, a specific section of the arch and tension
member was considered. To ensure a realistic representation of the structure,
fixed boundary conditions were applied to the section. These fixed boundary
conditions mimic the connections that would exist in the real bridge, providing
stability and constraining the movement of the modelled components.

It is important to note that for the last model, H12-19, an additional consideration
was made due to the considerable length of the tension member. As the main
interest was to study the frequencies of the hangers rather than the motion of
the beam itself, it was observed that the frequencies obtained from the analysis
were influenced by the motion of the tension member rather than the hangers.
To address this issue, the effect of the real existing crossbeams welded to the
top flange of the tension member, which constrained the out-of-plane motion of
the tension member, was simulated. Additional constraints were implemented to
prevent out-of-plane motion of the tension member and to focus the analysis on
accurately extracting the hanger frequencies.

By applying appropriate boundary conditions, including fixed boundary condi-

tions and additional constraints where necessary, the dynamic response of the
hanger bridge system could be accurately captured and analyzed.

4.2.5 Analysis Setup

This part of the study aimed to perform a frequency analysis, also known as
an eigenvalue analysis, to identify the dominant vibration modes and provide
information on the structural behaviour under dynamic loadings, such as wind
action. The frequency analysis was performed using the Abaqus software with

29
 4.3. RESULTS ANALYSIS

specific settings tailored to ensure accurate results. The analysis considered the
non-linear geometry of the structure, taking into account any geometric non-
linearity that could affect the dynamic response. The eigensolver chosen was
the Lanczos algorithm, a widely used and efficient method for solving eigenvalue
problems in frequency analysis.

A set of 10 eigenvalues were calculated to capture the relevant modes and provide
meaningful results. In addition, the eigenvectors representing the mode shapes of
the structure were normalized by displacement to ensure a consistent magnitude
of displacements and a better representation of the structural response. By using
these specific frequency analysis capabilities, the study aimed to accurately de-
termine the natural frequencies and mode shapes of the hangers, providing their
dynamic characteristics and their interaction with the overall bridge structure.

4.3 Results Analysis

The analysis focused on examining the frequencies obtained from frequency anal-
ysis for each of the five models. The natural frequencies provide valuable in-
sight into the dynamic behaviour of the hangers and their interaction with the
bridge structure. The frequencies were extracted and presented in a tabular for-
mat, highlighting the important characteristics of each model. For clarity in the
presentation of the results, only the first two frequencies and mode shapes are
reported in this chapter. For the complete results, please refer to Annex A.

Five different models, namely H16, H15-16, H14-16, H14-17, and H12-19, have
been analysed in the study of bridge hanger frequencies, see Figure 4.2. The
model H16 represents the single hanger (16th, see Figure 4.1), inclined at 10
degrees to the vertical axis. The arch and tension member are sectioned by two
vertical planes located 4.7 m apart. For the other models, they were added the
nearest hangers, taking into account their inclination, length, and distance from
the hanger H16. The final model is model H12-19, it is the most comprehensive
model, involving eight hangers.

To accurately simulate the behaviour of the models, fixed boundary conditions

were applied at the cut sections on the arch and the tension member. These fixed
boundary conditions restrict the movement of the structure at those points and
provide a realistic representation of the actual bridge system. Unique to H12-19
is the application of modified boundary conditions. It was observed for this model

30
4.3. RESULTS ANALYSIS

that the frequencies obtained from the analysis were influenced by the motion of
the tension member rather than the hangers alone since the length of the tension
member becomes of significant importance. To isolate the frequencies related to
the hangers and accurately capture their behaviour, additional constraints were
applied to simulate a cross beam that constrains the out-of-plane motion of the
tension member, thus preventing its contribution to the extracted frequencies.
This modification allows for a focused analysis of the hanger frequencies and
provides more accurate results.

31
 4.3. RESULTS ANALYSIS

(a) H16 (b) H15-16

(c) H14-16 (d) H14-17

(e) H12-19

Figure 4.2: The five models were analyzed by using the software Abaqus.

32
4.3. RESULTS ANALYSIS

Table 4.1: Mode Shapes and Frequencies for Different Models (Part 1)

Model 1st Mode Shape 2nd Mode Shape

(a) (b)
H16

Frequency: 6.63 Hz 14.05 Hz

out of plane in plane
(c) (d)
H15-16

Frequency: 6.44 Hz 7.47 Hz

out of plane out of plane
(e) (f)
H14-16

Frequency: 6.36 Hz 33 6.92 Hz

out of plane out of plane
 4.4. RESULT DISCUSSION

Table 4.2: Mode Shapes and Frequencies for Different Models (Part 2)

Model 1st Mode Shape 2nd Mode Shape

(a) (b)
H14-17

Frequency: 6.12 Hz 6.80 Hz

out of plane out of plane
(c) (d)
H12-19

Frequency: 6.32 Hz 6.78 Hz

out of plane out of plane

4.4 Result Discussion

The examination of different models of bridge hangers yields useful insights into
their dynamic behaviour and how the number of included hanger affect the results.
By analyzing the natural frequencies of these models, see Tables 4.1 - 4.2, intricate
relationships influenced by several factors such as the quantity and shape of the
hangers, become apparent.

In the case of a single hanger, Model H16, the first five modes of vibration pre-

34
4.4. RESULT DISCUSSION

dominantly exhibit motion within the hanger itself. Exhibiting a first natural
frequency of 6.6 Hz out of plane motion, and a second natural frequency of 14 Hz
in plane motion. As more hangers are introduced, Models H15-16, H14-16, and
H14-17, the dynamic response changes noticeably. By incorporating the second
hanger in model H15-16, the structure has a stiffer response with respect to the
previous one resulting in a first out-of-plane frequency slightly smaller, while the
first in-plane vibration corresponds to the third mode shape with a magnitude
around 14 Hz, see Figure A.2c.

Adding a third hanger, Model H14-16, transforms the system’s dynamic behaviour
further, where all three hangers interact and influence each other. This trend
persists with the inclusion of all the hangers used for the models. In multi-hanger
configurations, each hanger’s position affects the structure’s natural frequencies,
with variations in contribution across different modes. The first longitudinal
vibration as more hangers are incorporated into the model, slips to a higher
order mode shape, for the H14-16 model it occurs at the fourth mode shape,
see Figure A.3d, for the H14-17 model it occurs at the eighth mode shape, see
Figure A.4h, and for the model H12-19, it occurs at the tenth mode shape, see
Figure A.5j. For all the models the first longitudinal (in-plane) vibration of the
hangers occurs at a frequency around 14 Hz.

In conclusion, the analysis emphasizes that the quantity and positioning of hang-
ers considerably affect the fundamental frequencies of the structure. With an in-
crease in hangers, the interaction and coupling effects become more pronounced,
resulting in variations in natural frequencies and mode order. The key finding
of this study is the critical frequencies for the hangers, which are at 6.5 Hz for
the out-of-plane vibration and 14 Hz for the in-plane vibration. Such frequencies
can be employed for forthcoming analyses, aiding the evaluation of the bridge’s
dynamic behaviour.

35
Chapter 5

Ansys Fluent: 2D & 3D CFD

analysis

5.1 Introduction
ANSYS Fluent is a widely used computational fluid dynamics (CFD) software
that allows for the simulation and analysis of fluid flow and aerodynamic be-
haviour in various engineering applications. It provides a powerful tool for pre-
dicting and understanding the complex behaviours of fluids, such as air and water,
in different systems.

This chapter mainly summarises the 2D CFD analysis carried out using AN-
SYS Fluent to investigate the aerodynamic coefficients, shedding frequencies, and
Strouhal numbers of the models studied. The 3D simulations are carried out to
investigate the influence of the domain type on the results. The main objective
of the analysis was to evaluate the influence of the flow on the structures and
to understand their aerodynamic characteristics. Using ANSYS Fluent, it was
possible to simulate the airflow around the hanger section and study the flow
patterns, velocity distributions, and pressure profiles.

For the initial analysis, a wind domain was created within the simulation to
represent the airflow around a closed wall representative of the hollow rectan-
gular section. This allowed a basic understanding of the flow behaviour to be
established. The hanger sections were then gradually introduced into the wind
domain, taking into account their geometry, size, and positioning. This allowed
the interaction between the hanger sections and the airflow to be studied and

36
5.2. METHODOLOGY

the resulting aerodynamic effects, such as lift and drag, to be assessed. For each
model, the simulation setup was specified, including meshing techniques, solver
settings, and post-processing methods.

5.2 Methodology
The simulations are performed in two-dimensional (2D) space, which allows effi-
cient yet accurate representation of the primary aerodynamic characteristics while
minimizing computational complexity. Both steady-state and transient analyses
are performed to consider different aspects of wind-structure interactions. The
transient solution was required for this study to capture the shedding frequency
and calculate the Strouhal number. All the simulations have the same fluid
properties, as shown in Table 5.1. For the steady analysis, the solution was run
for 6000-12000 iterations, while for the transient solution, the convergence was
reached with a time step of 0.01s for a total of 60s of flow simulation.

Table 5.1: Fluid Properties

Fluid air -
Density 1.225 kg/m3
Viscosity 1.789 · 10−5 kg/(ms)

5.2.1 Realizable k - ϵ
To accurately model the turbulent flow patterns around the steel section, the
Realizable k-ϵ turbulence model was chosen, [19]. The term ’realizable’ means
that the model satisfies certain mathematical constraints on the Reynolds stresses.
Neither the standard k- ϵ model nor the Re-Normalisation Group (RNG) k- ϵ
model is realisable. The model used is known for its versatility in handling a
wide range of turbulent flows, making it well-suited to capture intricate flow
features such as streamlined curvature, vortices, and rotation.

5.2.2 Wall treatment

The k-ϵ turbulence model used for the analysis is mostly valid in the area in which
the turbulence is fully developed, and homogeneous, making it less appropriate

37
 5.2. METHODOLOGY

for near-wall flows where the boundary layer is not fully developed. In the near-
wall region, where the effects of viscosity and flow gradients are significant, more
advanced turbulence model with enhanced near-wall treatments should be used.
Many near-wall treatments can be applied, the standard ones are practical and
efficient from a computational point of view, but for this specific case study was
chosen to apply a different approach. These are based on assumptions that may
not be valid for the simulation model. For example, they assume a logarithmic
velocity profile near the wall and rely on the mesh near the wall being fine enough
to resolve the viscous sublayer. For this model, although was achieved a fairly
high mesh refinement, was decided to not use this approach as standard wall
functions may not give accurate results.

Enhanced Wall Treatment

ANSYS fluent allows you to use an advanced feature for the wall functions: En-
hanced Wall Treatment is an advanced modelling option that is particularly useful
when the standard wall functions are not sufficient. It is based on an improved
turbulence concept to provide more accurate predictions of velocity and other spe-
cific properties of the viscous substrate. These features have improved accuracy
in predicting wall shear stress, simulated heat transfer, and other parameters.
This is particularly important for the y + parameter as it reduces the specific
mesh quality requirements. In addition, this advanced feature is very flexible
and it is possible to select the areas in which it is to be applied, for example,
areas of complex geometry or points of particular turbulence. This adaptability
is essential as it allows the computational load to be reduced.

Parameter y + The y + value is a dimensionless parameter of significant im-

portance in the field of CFD, especially for the resolution of the fluid dynamics
equations in the boundary layer region. It represents the distance of the centroid
of the first grid cell from the wall of the simulated object, [7]. This is a distance
normalized to the thickness of the viscous boundary layer. In practice, the unit
value of this parameter means that the distance of the first cell of the lily is equal
to the thickness of the viscous boundary layer, [20]. Its value is calculated as
Eq. (5.1).
ρuτ y
y+ = (5.1)
µ

38
5.2. METHODOLOGY

Where ρ is the fluid density, uτ is the friction velocity, related to wall shear
stress, y is the distance from the wall (measured from the wall centroid to the
cell centroid), and µ is the dynamic viscosity of the fluid. If y + < 1, the first
grid cell is located in the laminar sublayer where viscous forces dominate. If y +
> 30, the first grid cell is in the boundary layer, where the flow is completely
turbulent. If 1< y + < 30, the first grid cell is within the viscous sub-layer but
not too close to the surface wall, in the transition zone of the boundary layer as
shown in Figure 2.1. A y + value that is too high or too low can lead to incorrect
predictions of the boundary layer, taking into account its value one can size the
mesh around the wall to obtain more accurate simulations.

5.2.3 Computational Wind Domain

For the 2D simulations, the flat section of the midpoint of the longest hanger of
the arch was considered. Due to its inclination, the section taken on the median
horizontal plane is not perfectly square, but has dimensions of d = 25.38x25 cm,
as shown in Figure 5.1, this characteristic dimension (d), it was used to calculate
the data of interest, such as wind domain size and aerodynamic coefficients. From
previous simulations found in the literature, the first domain was of a distance
from the inlet and far-field of 4 times d and a distance of 12 times d from the
outlet. Simulations showed that such domain was far too small, different sizes
were tried until the last one of 20 times d from the inlet and far-field and 40 times
d from the outlet was reached, see Figure 5.2. It should be noted that convergence
was obtained even with half the dimensions, but since the simulations had to be
carried out over a very wide range of speeds, it was preferable to adopt a larger
domain to have the same domain for all the 2D analyses.

5.2.4 Boundary Condition

In the computational domain, it is necessary to assign boundary conditions, which
involve assigning certain specifications to the fluid in accordance with these con-
ditions. There are three main boundary conditions: the inlet, the outlet, and
the wall. There is a fourth one assigned to the lateral sides of the domain, those
parallel to the fluid, usually the conditions related to the far field or symmetry are
assigned, but they are significant when they correspond to the walls of a cylinder
with a flow inside, for example, not important for the specific case study.

39
 5.2. METHODOLOGY

Figure 5.1: CFD rectangular section of the hanger

Figure 5.2: Wind domain characteristics with relative boundary conditions.

40
5.2. METHODOLOGY

Inlet

The inlet boundary condition governs the specifications of the fluid entering the
computational domain, such as the inlet velocity, the reference system, absolute
or relative to neighbouring cells, the velocity profile, and the pressure value.
Turbulence properties can be defined by the turbulence intensity and hydraulic
diameter, see Eq. (5.2), or by the turbulent viscosity ratio for the inlet conditions.

4Ainlet
DH = (5.2)
Pinlet

Where Ainlet is the area of a surface given by the length of the inlet multiplied
by 1 m of thickness, and Pinlet is its perimeter

Outlet

The outlet boundary condition defines the conditions at the exit point of the
computational domain, where the fluid leaves the domain. The settings for the
outlet are the reference frame for the backflow conditions, the gauge pressure, the
pressure profile multiplier, and the method of specifying the backflow direction.
The turbulence properties can be set to be the same as those specified for the
inlet.

Wall

The Wall boundary condition is applied to surfaces within the computational

domain to simulate interactions with solid walls. The wall boundary condition is
characterized by the motion of the wall, which is assumed to be stationary except
in advanced simulations. For this boundary, a number of shear conditions are
specified: the ’anti-slip’ condition, which ensures that the velocity of the fluid on
the wall is zero and that there is no relative motion between the fluid and the
wall.

For these analyses, the velocity of the inlet fluid was varying between a range
of 2.5 m/s and 30 m/s. The velocity profile is constant and absolute, always
orthogonal to both the inlet and outlet, while the pressure at the outlet is fixed
at zero with a unit multiplier. It was decided to set the turbulence specifications
via the turbulence intensity at 5% and the hydraulic diameter calculated by
considering a surface area equal to the length of the duct multiplied by one
meter of thickness. All the information about the simulation setting is collected

41
 5.2. METHODOLOGY

in Table B.1.

5.2.5 Mesh

The generation of meshes for CFD analysis is the most critical part, as a compro-
mise must be found between cell fineness and computational cost. The software
allows the automatic generation of a mesh, which can then be refined in the ar-
eas of interest, such as the vicinity of the case study section and the area where
the vortex wave is generated. For the multiple simulations carried out for the
case study, different mesh options were used: subdivision of a generic part of the
geometric model into a fixed number of parts or assignment of the subdivision
dimension. Where necessary, a dependency region was created, to obtain a denser
mesh. A very useful approach is to apply a growth rate that allows it to start
from the densest area of the mesh and increase the size of the cells in subsequent
layers by a given factor. For the area close to the section, where the mesh needs
to be denser, the inflation technique can be used, which allows the creation of
offsets of the section geometry with a given thickness.

In addition to the meshing approaches, it is essential to evaluate the mesh quality

to ensure reliable simulation results. There are several parameters to evaluate, the
most important being skewness, minimum orthogonal quality, and aspect ratio.
The skewness is the percentage difference between the optimal cell size and the
actual cell size, of which it is necessary to evaluate the maximum value that varies
between 0 and 1, the closer it is to zero the better, and the average skewness value
is better being around 0.33. The orthogonal quality is a vector calculation that
indicates the quality of the mesh. This quantity represents excellent mesh quality
when its value is close to one, the average value for the whole mesh should be
kept around 0.2. While the aspect ratio is the ratio of the base to the height of
the cell, a value less than 35 indicates a good mesh.

By combining these meshing approaches and keeping the quality parameters un-
der control, the mesh has been adapted to provide an accurate representation
of the flow field, particularly in areas subject to turbulence effects. This careful
mesh design contributes to the reliability and accuracy of the simulation results.

42
5.2. METHODOLOGY

5.2.6 Solver Method

ANSYS Fluent allows to use two solving approaches: one based on pressure and
one based on density. Both can be used for CFD analysis, but in some cases, one
formulation may work better than the other. These two approaches differ in the
way the continuity and momentum equations are solved and the set of equations
to be solved, depending on the type of analysis to be performed. In general, the
pressure-based approach is used for incompressible or slightly compressible fluids.
The second is used for compressible fluids and very high simulation speeds. For
this particular case study, the pressure-based approach was used.

Iterative Pressure-Velocity Algorithms

There are two solver algorithms for the pressure-based solver, one separate and
one coupled. In the former, the governing equations of motion and continuity are
solved individually, while in the latter, as the name suggests, they are solved in
a coupled manner. The latter has advantages in terms of convergence speed at
the cost of more memory.

There are five pressure-velocity coupling algorithms in the ANSYS software for
the pressure-based approach: SIMPLE, SIMPLEC, PISO, Coupled, FSM, and
NITA, the latter being a scheme used for transient flows using the non-iterative
time advancement scheme and generally this scheme works in conjunction with
the above algorithms, [18].

SIMPLE: this is the most widely used algorithm, both for transient and steady
solutions, and uses a relationship between velocity and pressure corrections to
enforce the conservation of mass and obtain the pressure field. SIMPLEC is a
version that allows faster convergence but has limitations, it is used for relatively
simple problems and laminar flows without additional turbulence models.

PISO: Pressure-Implicit with Splitting of Operators is best suited for transient

solutions, especially when a large time step is required. For steady-state anal-
ysis, it has no advantage over SIMPLE or SIMPLEC. The disadvantage of this
algorithm is the high computational cost.

FSM: The Fractional Step Method, is indicated when the NITA scheme is chosen
as it is slightly less expensive in terms of computation than the PISO scheme.
For this case study it was decided to use the SIMPLE algorithm as it is ideal for
both transient and steady-state solutions and is the best from a computational

43
 5.3. RESULT

point of view.

5.3 Result
This section presents the results of several simulations, each designed to meet
a specific purpose of this case study. In particular, there is a macro division
between steady-state and transient simulations, the difference between the two
is that the latter allows time-dependent data to be obtained and is essential to
capture the shedding frequency, which is of considerable importance for the re-
search. The steady-state simulations were carried out primarily to have a valid
model on which to run a transient simulation but also to assess how much the
aerodynamic values differ between the two types of simulation. To give an idea
of the time required for the simulations, on a computer of CPU 3.20 GHz and 4
processors working in parallel, the steady-state solutions, once optimized, vary in
the range of 4 - 12 hours, while the transient solutions can take days, especially
if animations of the simulation with contouring features are also extracted. The
steady-state simulations were used to answer another research question regarding
the geometry of the section and how this can influence the aerodynamic coeffi-
cients, the choice to use the steady-state solution was dictated by the advantage
in computational cost. Furthermore, once the relationship between the steady
and transient simulation data has been established, it is possible to apply it to
models of slightly different geometry.

Rectangular cross-section

The analysis was carried out on a single rectangular cross-section of 25.38 x 25 cm,
characterized by sharp, non-rounded edges. Several simulations were carried out,
covering a range of wind speeds from 2.5 m/s to a maximum of 10 m/s. Be-
fore finalizing the dimensions of the wind domain, a series of simulations were
systematically run at different speeds to assess boundary interactions.

The most accurate wind domain, as shown in Figure 5.2, was selected based on
observation of the simulation results and the interaction between the flow con-
dition around the centred section and the boundaries. The aim was to obtain
a wind domain large enough to allow a distinct flow zone around the focal sec-
tion but at a reasonable computational cost. In the vicinity of the section the
mesh was deliberately refined to ensure precision in data extraction, as shown

44
5.3. RESULT

in Figure 5.3. However, it is important to note that the steady-state simulation

Figure 5.3: Mesh configuration of the wind domain for the steady simulation.

used in this study does not capture shedding frequencies, making the derivation
of the Strouhal number unattainable. The main results extracted from these
simulations include the aerodynamic coefficients: drag CD , lift CL , and moment
CM .

The simulation started with several 500 iterations, but to reach the convergence
the number of iterations were increased to 6000 and for higher speeds doubled.
From the results obtained, collected in Table 5.2, it is possible to asses that for the
range of speeds simulated the drag coefficient CD increases slightly with higher
wind speeds, while the lift coefficient CL and moment coefficient CM exhibit minor
fluctuations.
Table 5.2: Rectangular cylinder: not rounded edges.

Wind speed [m/s] 2.5 5.0 7.5 10.0

n° of iterations 6000 12000 12000 12000
Drag Coeff. CD 1.379 1.397 1.398 1.402
Lift Coeff. CL 0.000 0.002 0.001 0.001
Moment Coeff. CM -5.345 -5.410 -5.414 -5.432
Turbulent Intensity 0.070 0.070 0.070 0.070

The drag coefficient, denoted by CD , represents the resistance of a cross-section

in the direction of the wind flow. The trend of increasing CD with increasing wind
speed is consistent with the fact that the drag force is proportional to the square
of the flow velocity. However, it should be noted that this relationship cannot be
extended to all speeds, as the drag force also depends on other factors. This issue
will be covered in the following sections. An intriguing outcome is achieved for
the lift coefficient CL , which gauges the force perpendicular to the flow. Since the
wind force acts without any angle of attack, it is unsurprising that the coefficient

45
 5.3. RESULT

fluctuates around zero. This implies that, on average, the hanger cross-section
undergoes negligible lift, suggesting that the structure does not create consider-
able cross-wind forces, a desirable trait for maintaining structure stability. The
moment coefficient CM denotes the hanger cross-section’s propensity for rotation
in response to fluid forces. Negative values signify the section’s orientation of
rotation. As with CD , CM increases with wind speed, denoting that the section
encounters amplified rotational forces at higher wind speeds.

An intriguing outcome of CFD simulations is the recognition of contouring and

pathlines near the section. This provides a simpler comprehension of the flow mo-
tion and its dynamics. Some examples are shown below in Figures 5.4 and 5.5.
These are extrcted from the simulation carried out with a wind speed of 5 m/s at
the inlet. The symmetrical contouring of pathlines and observation of two per-

Figure 5.4: Wind flow velocity magnitude: contouring and pathlines.

fectly symmetrical vortices behind the section indicate a steady flow simulation.
The vortex region is symmetric about the mid-axis of the wind domain, and its
length remains constant in time for each speed. However, an expected result is
that the vortex length increases with Reynolds, which corresponds to speed, [21].
Small vortices also form in proximity to the section’s side walls, and at the tail
end of the two primary vortices. These vortices generate dynamic forces on the
studied structure, and their positioning and dimensions can significantly impact
the section’s integrity. When a fluid strikes a body, two distinct pressure zones
are commonly observed. These regions in aerodynamics are referred to as the
"high-pressure region" and the "low-pressure region". The high-pressure zone
is located in the forward portion of the object, where the air slows down and
compresses after striking the front surface of the section. The second region is
situated at the rear of the section and relates to an area where the pressure
decreases due to the formation of vortices. These vortices entail a continuous
change in the wind direction. Identifying these two regions is highly important

46
5.3. RESULT

(a) (b)

Figure 5.5: Wind flow contouring: static pressure (a) and turbulence intensity
(b).

in the field of aerospace as they determine the lift and drag forces acting upon
the object. Furthermore, it is crucial to recognize them in structural engineering
to understand how aerodynamic forces act on the object.

High turbulence intensity regions form at the front of the section, indicating that
the air in these zones is turbulent and has significant fluctuations in velocity. This
phenomenon is commonly observed at the stagnation point of an object where
the flow comes to a halt and changes direction. The turbulent flow in this area
is the result of the interaction between the oncoming flow and the section wall.

Rectangular cross-section rounded edges

In this section, various simulations are conducted with diverse model geometries.
The wind domain dimensions and simulation settings remain constant and unal-
tered, with the change being implemented solely in the hanger section of concern.
Alterations to the mesh were necessary, but diligent effort was made to ensure
that quality parameters were comparable or higher than those in the previously
presented simulations. Firstly, three models with varying connecting radii but
identical speeds were assessed at a wind speed of 2.5 m/s. This analysis aims to
determine the impact of section geometry on the aerodynamic coefficients.

Table 5.3: Aerodynamic coefficients for different model geometries with different
radii.

Edge geometry CD CL CM
not- rounded 1.379 0.000 -5.345
r = 1 cm 0.889 0.059 -3.249
r = 1.5 cm 0.779 -0.034 -3.082

Table 5.3 displays the outcomes of three different models, one without rounded
edges, the second with an edge radius of 1 cm, and the third with a higher radius

47
 5.3. RESULT

of 1.5 cm. The presence of the radius in the wall section undoubtedly affects the
aerodynamic coefficient. Results show that non-rounded geometry experienced
the highest drag coefficient. At 1 cm and 1.5 cm radii, a notable decline in CD
was evident. Smoothing and increasing the size of the radii of the edges generally
decreases drag, following a common principle of aerodynamics. The reduction
in drag is linked to an improvement in the flow characteristics around the sec-
tion. The edge that was not rounded generated a significant negative moment,
demonstrating a strong inclination to rotate under the influence of fluid forces.
The negative moment magnitude decreased with rounded edges (r = 1 cm and r
= 1.5 cm), which implies a reduced tendency to rotate. Rounded shapes gener-
ally result in greater stability due to the reduction of moments that might cause
rotation.

Figure 5.6: Not rounded edges, wind speed pathlines varying from null speed in
blue, to the maximum speed reached during the simulation in red.

Figure 5.7: Edge r = 1 cm

48
5.3. RESULT

Figure 5.8: Edge r = 1.5 cm

After analyzing the pathlines shown in the Figures 5.6, 5.7, and 5.8, of the three
different sections, it becomes clear that the radius significantly influences the fluid
flow behaviour. In particular, the streamlines and vortices are notably impacted
by the presence of a radius. Sections with radii exhibit significantly shorter vor-
tex waves compared to the sections without rounded edges. This phenomenon
suggests that the vortices generated by rounded sections are more confined and
compact. The velocity of the fluid flow near the rounded section is significantly
lower. The presence of radii seems to reduce the turbulence effects with a con-
sequent increase in the velocity of the fluid flow. These observations lead to an
important conclusion: the resistance encountered by the wind flow is reduced
when a section’s edges are rounded. The introduction of rounded edges alters the
flow patterns, limiting the development of vortices and reducing the flow veloci-
ties in the vicinity of the section. In practical terms, this means that it is essential
to consider the actual radii of the section edges to achieve accurate and realis-
tic simulations. However, for a conservative approach, it may be advantageous
to ignore the radius in the simulation and consider higher aerodynamic quanti-
ties. This conservative approach ensures that the verification process accounts
for higher aerodynamic forces, thereby increasing the safety and reliability of the
structure.

This analysis was extended to evaluate the effect of edge radius, particularly at
1 cm radius, over the remaining three wind speeds. The aim was to investigate
whether the trends observed at 2.5 m/s persisted as wind speeds increased. The
results confirmed that the hypotheses formulated based on the 2.5 m/s simulations
were consistent and applicable to higher wind speeds.

49
 5.3. RESULT

Table 5.4: Aerodynamic coefficients for the steady solution with section r = 1cm.

Wind speed [m/s] 2.5 5.0 7.5 10.0

n° of iterations 6000 12000 12000 12000
Drag Coeff. CD 0.889 1.096 1.144 1.145
Lift Coeff. CL 0.059 0.019 0.000 -0.011
Moment Coeff. CM -3.249 -4.155 -4.390 -4.426
Turbulent Intensity 0.100 0.090 0.090 0.090

Table 5.4 provides a detailed breakdown of the effective aerodynamic coefficients

when considering the presence of a 1 cm edge radius.

Figure 5.9: CD and CM at different speeds of rectangular sections with rounded

and not rounded edges.

In summary, the analysis of the steady-state solutions showed consistent trends

over the range of wind speeds investigated. As the wind speed increased, the
aerodynamic coefficients showed an upward trend. In addition, these coefficients
decreased, i.e. resistance decreased, with the introduction of edge radii, see Fig-
ure 5.9. In the 2D simulations, two key aerodynamic coefficients emerged as
crucial: the drag coefficient, which represents the section’s resistance to wind
flow, and the moment coefficient, which indicates the resistance to rotation due
to wind forces.

50
5.4. TRANSIENT SOLUTION

5.4 Transient Solution

The time-dependent simulation creates a transient solution that allows the extrac-
tion of time-dependent data, capturing aerodynamic effects over different tempo-
ral scales. Shedding frequencies and related parameters, including the Strouhal
number, can be obtained through this process. This type of analysis provides a
detailed examination and comprehensive observation of the phenomenon of flow
regimes and turbulence effects.

The range of speeds studied is wider than that studied for the steady solution, as
the focus is shifted towards high speeds and the Reynolds number. All simulations
were run for 60 seconds with a time step size of 0.01 seconds. A sensitivity
analysis was conducted for the time step size by refining it for a few speeds, but
the variation in the results was insignificant. In this section, two different models
were employed. The first model utilized a single cross-section of the hanger H16,
which was identical to the one used for the transient solution, with no rounded
edges. The second model included three different sections to evaluate the impact
of multiple sections on the data of interest. For this final model, the wind domain
was increased while maintaining a minimum distance of 20d from the last lateral
wall.

5.4.1 Non-rounded edges cross-section

The study assessed various wind speeds ranging from 2.5 m/s to 30 m/s. The
decision to increase the maximum speed to 30 m/s was necessary, since the char-
acteristic mean wind speed observed at the bridge’s location, is 27 m/s. The
study’s findings will be presented in this section utilizing also the corresponding
Reynolds number, given that aerodynamic parameters are depicted as a function
of the Re number.

Aerodynamic coefficients, see Table B.2 were extracted for each simulation. The
estimated peak frequency of drag and lift, Table B.3, was then calculated using the
Fast Fourier Transform (FFT) of their respective signals, and related amplitudes
were subsequently extrapolated. This allowed for the calculation of the Strouhal
number, Table B.4 based on the estimated peak frequency of lift. Finally, the
magnitude of both drag and lift forces was evaluated, and all the data extracted
for this model are presented in Table B.5. The graphs in Figure 5.10 depict
aerodynamic coefficients for a rectangular cross-section that was derived from a

51
 5.4. TRANSIENT SOLUTION

Figure 5.10: Transient solution: aerodynamic coefficient at different speeds.

transient solution at varying wind speeds.

There is a tendency for the drag coefficient (CD ) to decrease as wind speed in-
creases, reaching a minimum at 30 m/s. However, until a speed of 10 m/s is
reached, the drag coefficient increases, following the same trend as observed in
the transient solution. Nevertheless, for the remaining speed range, the behaviour
aligns as expected with the results reported in the literature [23], [24]. This is be-
cause at higher wind speeds, flow around the structure becomes more streamlined,
thereby reducing drag.

For 2D simulations, the coefficient of lift (CL ) is of little significance as its value is
small and oscillates around zero. Therefore, it is not possible to define a specific
trend for the lift, which changes randomly in magnitude and sign due to the
oscillating value around zero. The values in Figure 5.10 are punctual values at
time t = 60 s, which represents the end of the simulated time. The peak value
in the lift at 5 m/s is not relevant because the mean value of the lift for the
simulation at this speed and all other simulations tends to be zero.

The moment coefficient gradually decreases at higher speeds, following the trend

52
5.4. TRANSIENT SOLUTION

of the drag coefficient. It appears that there has been a decrease in CM , indicating
a reduced susceptibility of the structure to rotation under wind influence.

The following results are not directly extracted from the simulation report but
are obtained by using the aerodynamic coefficient values calculated at each time
step to determine the frequency and the Stouhal number, see Eq. (2.5). The fre-
quency domain for both lift and drag was calculated for all transient simulations,
and subsequently, the Stouhal number was obtained after determining the peak
lift frequency. The initial 10 seconds of the data signal were removed as an ini-
tialization period to enhance the accuracy of the results. Then, the Fast Fourier
Transform (FFT) was utilized to convert the time-domain data into the frequency
domain. This transformation was crucial in comprehending the response of the
cross-section to the wind flow in the frequency domain. After obtaining the fre-
quency spectra, the peak frequencies were identified for both lift and drag, see
Figure 5.11. The trend of the frequencies consistently increases with higher wind
speeds, indicating a proportional response to the changing aerodynamic regime.
The amplitude variation of drag and lift forces is a compelling point of interest.
The drag coefficient’s magnitude is significantly greater than that of the lift coeffi-
cient, but the relation between their respective amplitudes is not consistent. The
lift experiences a much higher amplitude than the drag for all the speeds inves-
tigated, see Figure 5.12 as an example for the wind speed 27 m/s. Of note, wind
speeds exceeding 20 m/s result in a significant change in the dynamic response,
marked by a reduction in drag amplitude. This alteration may indicate a shift in
the main forces caused by the flow or a change in the general flow pattern around
the structure as the speed increases and the flow becomes more streamlined.

An intriguing observation concerns the drag frequency magnitude, which fluc-

tuates between 5.5 Hz and 8 Hz for wind speeds ranging from 5 m/s to 10m/s.
According to the Abaqus frequency analysis, the first natural frequency of the
structure is approximately 6 Hz, while the others vary within a range of up to
14 Hz. The initial natural frequency’s vibration is out of plane. As a result, it
can be concluded that the critical wind speed for the hangars ranges from 5 to
8 m/s. This wind speed range is an ordinary value for the geographic allocation
of the Bomarsund Bridge.

The calculation of the Strouhal number, a significant parameter that signifies the
shedding of vortices surrounding the structure, takes into account the lift peak
frequency. A decreasing trend of Strouhal numbers with increasing wind speed

53
 5.4. TRANSIENT SOLUTION

is observed, see Figure 5.11 indicating that a shift from regular to more irregular
vortex shedding patterns is occurring.

Figure 5.11: Transient solution: estimated peak frequency and Strouhal number

A characteristic phenomenon of vortex shedding was observed, wherein the peak

frequency of drag is twice that of lift in the frequency domain spectra. This
outcome can be elucidated by the vortex-shedding process and its interaction
with the composition. Two vortices of equal strength and opposite signs are shed
during each oscillation period in the wake of a structure, such as a rectangular
cross-section. It is important to note that the sign of the vortex has no impact on
the experienced drag force. This distinctive feature results in a period-doubling
effect, in contrast to the lift oscillation. An example of the frequency spectrum
in the simulation at a speed of 27 m/s can be observed in Figure 5.12.

54
5.4. TRANSIENT SOLUTION

Figure 5.12: Time domain spectra and frequency domain spectra at wind speed
of 27 m/s

55
 5.4. TRANSIENT SOLUTION

Figure 5.13: Aerodynamic coefficients vs Reynolds

In this particular study, the drag force is identified as the main force due to its
alignment with the flow direction. To ascertain the accuracy of the results, a
comparison was conducted with literature values, Figure 5.13, and two references
were consulted: Hoerner’s famous drag curve for circular cross-sections [23], and
a comprehensive investigation carried out by Yuce and Kareem in 2016 [24]. The
latter reference was of significance because it utilized ANSYS Fluent simulation
setup and investigated both square and circular sections. In their study, they
simulate a wider range of Reynolds numbers, but for the current study case, the
Reynolds range is correlated to the range of possible wind speed for the bridge.

Considering the Hoerner curve, the current simulation results consistently indi-
cated drag values above those of the reference. This variation is justified since
Hoerner’s curve relates to an ideal circular cylinder, whilst the study concerns a
bluff body with a rectangular cross-section.

Referring to the research of Yuce and Kareem, who established distinct curves
for circular and square cylinders, the actual rectangular cross-section results take
place in the middle of that range. This placement can be considered adequate,

56
5.4. TRANSIENT SOLUTION

the study case uses few simulations for a wide step of Reynolds, while the current
study uses narrow steps of Reynolds, describing better the curve in the considered
range, Figure 5.13.

Regarding the wind flow configuration, it is interesting to observe how it changes

in the investigated range of wind speeds. At low speeds behind the bluff body,
the Von Karman vortex street forms, allowing the identification of zones where
the flow changes direction and local eddies form. As wind speeds increase, the
vortex shedding tail becomes more compact and less turbulent, resulting in a
more streamlined flow. This explains the reduction in the magnitude of the drag
coefficient. In Figure 5.14, two contour images are presented as examples of
vortex shedding tail for low and high wind speeds: one at 7.5 m/s and the other
at 27 m/s.

(a) v = 7.5 m/s (b) v = 27 m/s

Figure 5.14: Transient solution: Velocity contour at different wind speeds.

It is also noteworthy to examine the form of the shedding vortex located at the
rear of the cross-section. For the steady solution, the two vortices were symmetric,
but with a transient analysis of the effect of turbulence over time, it is apparent
that this symmetry is lost. Additionally, lower wind speeds result in larger eddies
that are also distributed laterally to the section. At high velocities, the vortex
tail becomes more streamlined, and a second vortex forms above the first one.
Some examples are reported in Figure 5.15, for different speeds that cover the
whole range of wind speed simulated.

5.4.2 Interaction between different sections

The following analysis was conducted on a different model. To determine the
effect of interaction between hangers, the section of the nearest hangers on the
left and right of hanger H16 was added to the previous section. The section
spacing equals the actual distance between the hangers, measured at the middle

57
 5.4. TRANSIENT SOLUTION

(a) v = 2.5 m/s (b) v = 10 m/s (c) v = 27 m/s

Figure 5.15: Transient solution: wind pathlines at different speeds.

hanger’s height. The wind model is larger than the one used for the single section.
The distance between the lateral wall section and the far field is maintained at
three times d. To calculate the aerodynamic coefficient, the characteristic width
of influence is modified, with the new dnew = 0.7614 m being equal to the sum of
the widths of the three sections. It is worth noting that we used the lift frequency
of the multiple-section model for the Strouhal number but considered the width
of one cross-section when calculating its magnitude. This is because St is related
to a single vortex shedding wave. Before conducting transient simulations of the
three-section models, a steady simulation was carried out to assess mesh quality,
wind domain dimensions, and result convergence. All the remaining simulation
settings were maintained the same as the previous transient simulations.

Table 5.5: Transient solution multi-section model: data extracted for each speed
simulated

Wind speed [m/s] 2.5 10 20 25 30

n° time steps 6000 6000 6000 6000 6000
time steps size [s] 0.010 0.010 0.010 0.010 0.010
Drag Coeff CD 1.782 1.797 1.682 1.657 1.647
Lift Coef. CL -0.348 0.119 -0.038 -0.050 0.051
Moment Coeff. CM 26.81 17.73 19.59 19.55 17.44
Estimated Lift Peak Frequency [Hz] 1.333 4.233 6.432 6.949 7.3321
Estimated Strouhal Number 0.1353 0.1074 0.0816 0.0705 0.0620

The trend of decreasing drag coefficient with increasing wind speed is maintained
as studied for the single-section model. However, the drag coefficient for the
model comprising three cross-sections is significantly lower than that of the model
featuring only one cross-section, as depicted in Figure 5.16.

This analysis aimed to evaluate the interaction between multiple sections and
their impact on the aerodynamic coefficient and related parameters. By adding
two additional sections to the wind domain, the characteristic dimension of the
model increases almost three times that of the previous model. In terms of the

58
5.4. TRANSIENT SOLUTION

Figure 5.16: Transient solution: Drag coefficient of different models.

response of the entire body to the encountered wind flow, its resistance increases.
However, the three distinct vortex-shedding waves formed behind each section
interact with each other, influencing both the frequency of the system and the
amplitude.

In Figure 5.17, an example of the lift signal for a wind speed of 10 m/s is reported.
The signal’s amplitude drastically decreases around t = 10 s s and t = 40 s, which
is an effect of the interaction of the eddies formed behind each section.

An interesting observation is the velocity contour, displaying the distribution of

eddies within the wind domain. The central section displays the highest den-
sity of eddies, Figure 5.18, whereas their distribution varies across the distinct
sections, the orientation and the dimension of the principle vortex change. This
phenomenon is visible in Figure 5.19.

Focusing on the frequencies and amplitudes, both models demonstrate an increase

in the frequencies of drag and lift with increasing wind speeds. However, the sec-
ond model generally displays reduced frequencies in comparison with the first
model. The second model with several sections illustrates a more nuanced aero-
dynamic reaction. The findings imply that the structural intricacy introduced by
multiple sections affects aerodynamic behaviour considerably, accentuating the
significance of such details in realistic simulations.

59
 5.4. TRANSIENT SOLUTION

Figure 5.17: Transient solution multi-section model: lift signal and frequency
domain spectra for wind speed of 10 m/s.

5.4.3 3D Transient Analysis of hanger H16

This section introduces a distinct analysis using a 3D model, which is a departure
from the previous 2D space domain investigations. With the introduction of three-
dimensional geometry, significant changes occur in the mesh and the characteristic
reference dimensions.

In the model used, the inclination of the hanger of 10◦ was taken into account
and the final geometry is shown in Figure 5.20. The wind domain has an inlet
dimension of 10.25 m, while the far field dimensions extend to 15.25 m, and a
height of 11 m. The hanger is positioned centrally between the two far-fields and
5m from the inlet.

A steady-state solution was run on the model before initiating a transient simu-
lation. This step aimed to optimize the mesh size and wind domain volume to
find a compromise between the accuracy of the simulation results and the compu-
tational cost, which is much higher than the 2D simulation. The mesh followed
a similar approach to the 2D model, incorporating enhancements to maintain
a high-quality mesh. The meshing size was refined in proximity to the hanger
element, and the distribution of the boundary condition is the same as used for

60
5.4. TRANSIENT SOLUTION

Figure 5.18: Transient solution multi-section model: velocity magnitude contour

for a wind speed of 30 m/s, H15 - H16 - H17.

(a) Hanger H15. (b) Hanger H16 (c) Hanger H17

Figure 5.19: Transient solution multi-section model: velocity pathlines of the

three hangers.

the 2D models.

The simulations covered a range of wind speeds, mirroring the speeds analyzed
in the previous 2D rectangular cross-section model. This approach facilitates a
comparative assessment of the differences between the two models.

The simulations were run for a maximum of 60 seconds, with a time step size
of 0.01 s and a maximum of 10 iterations for each step. For each simulation the
aerodynamic coefficient, pressure, velocity magnitude, and turbulence parameters
were extracted, see Table B.6. The same calculation performed for the 2D models
was also performed for the 3D geometry. It was considered as a characteristic
area equal to the depth of the section of 0.25 m times the height of the hanger of
11.30 m. To obtain the Strouhal number, the depth of d = 0.2538 m was used as
the characteristic dimension since the aim was to define this value for the rect-
angular cross-section, rather than the 3D rectangular cylinder. Looking at the
graphs in Figure 5.21, a clear pattern in the drag coefficient with increasing wind
speed is evident when examining the results of the 3D model of the hanger. In

61
 5.4. TRANSIENT SOLUTION

Figure 5.20: Transient analysis: 3D model for ANSYS simulation.

contrast to the 2D model where there were significant variations of the aerody-
namic coefficient between different speeds, the 3D model depicts a more nuanced
distinction. In addition, the magnitude of the aerodynamic coefficients is gener-
ally lower in the 3D model, when compared to the 2D counterpart.

Besides, an especially captivating observation is apparent in the lift frequency

domain. The frequency generally increases over the wind speed range, but specific
values at various wind speeds exhibit unexpected disparities. Notably, at a wind
speed of 10 m/s, the lift frequency for the 3D model is significantly lower than the
corresponding value for the 2D model. However, at higher speeds, the frequency
magnitude slightly surpasses that of the 2D model. The same trend is observed
for the Strouhal number.

The analysis of drag frequency for the 3D model presents a significant limitation.
As previous simulations have assessed, the frequency of the drag ought to be
double that of the lift frequency to achieve a correlation well-established in the
literature, [27]- [28], which stems from the insensitivity of drag to the direction
of vortices present behind bluff bodies. However, accomplishing this correlation
in the ANSYS analysis of the 3D model requires a different mesh configuration,

62
5.4. TRANSIENT SOLUTION

Figure 5.21: Comparison between 2D model and 3D model results of drag coeffi-
cient, lift frequency and Strouhal number.

specifically the implementation of dynamic mesh. This change leads to a no-

table increase in computational costs and analysis complexity, necessitating more
powerful processors. Unfortunately, the standard computer used for this study
cannot support such simulations.

To illustrate the distribution of vortices along the hanger, three reference planes
at the base, middle, and top of the hanger, were created, an example is given
in Figure 5.22. The turbulence effects vary along the length, as demonstrated in
Figure 5.23.

5.4.4 Wind Load Calculation

The following section presents the findings for defining the wind load in accor-
dance with the Eurocode. The same approach outlined in Section 2.4 was used,
with parameters extracted and summarised in Table 5.6. To assess the results, the
critical wind speed was calculated and verification was conducted to determine if
vortex shedding phenomena should be taken into account. For this calculation,
the first two natural frequencies of the initial model implemented on Abaqus were

63
 5.4. TRANSIENT SOLUTION

Figure 5.22: 2.5 m/s Wind Velocity Contour.

(a) Top plane (b) Mid plane (c) Bottom plane

Figure 5.23: Velocity magnitude vectors for wind speed of 2.5 m/s.

used, see Table A.1. The first mode shape oscillates transversally to the direction
of the bridge at a frequency of 6.63 Hz, while the second natural frequency has
a magnitude of 14.06 Hz, relating to longitudinal vibrations. The critical wind
speeds for each i-th frequency, which were calculated using the equation Eq. (2.9),
are Vcrit,1 = 13.80 m/s and Vcrit,2 = 29.28 m/s, respectively. Both of these critical
velocities are below the safety threshold of Vcrit,min = 34 m/s above which vortex
shedding can be ignored.

After determining the critical wind speeds for each vibration mode shape, the
wind’s inertial force, taking into account the presence of shedding frequency, was
calculated using Eq. (2.10). The values of inertial wind force for the first and
second mode shapes are respectively: Fw,1 = 6 kN/m, Fw,2 = 28 kN/m. Instead
of using the Strouhal numbers given by the Eurocode, it is possible to use those
obtained from the transient simulation of the model with a single rectangular

64
5.5. RESULTS DISCUSSION

cross-section without rounded edges at a speed of 27 m/s (which corresponds to

the mean wind characteristic speed). Then the resulting wind forces for each
i-th natural frequency will be respectively: Fw,1 = 19 kN/m, Fw,2 = 87 kN/m. The
Strouhal number supplied by the Eurocode for a rectangular cross-section remains
constant regardless of wind speed, equating to St = 0.12. However, according to
the simulation conducted for this case study, the Strouhal number changes with
the wind speed and particularly decreases with wind speed. As a result, for
the simulation of a wind speed of 27 m/s, the Strouhal number was equal to
St = 0.068.

Table 5.6: Parameters for Wind load Calculation According to Eurocode

Parameter Value
Terrain Category Class I
Basic Wind Speed, vb0 23 m/s
Wind Directionality Factor, Cdir 1
Seasonal Factor, Cseason 1
Reference Wind Speed, vb 23 m/s
Reference Height, z 10 m
Roughness Length Factor, z0 0.01 m
Minimum Height for Wind Speed Profile, zmin 1m
Maximum Height for Wind Speed Profile, zmax 200 m
Terrain Factor (kr ) 0.17
Roughness Factor (cr ) 1.173
Orography Factor (c0 ) 1
Mean Wind Velocity (vm ) 27 m/s
Strouhal Number (St ) 0.12
Structural damping (δs ) 0.006
Scruton number (SC ) 11.575
Characteristic width (d) 0.25 m
Reynolds number (Re) 4.61E+05
Maximum displacement (ymax ) 200 mm

5.5 Results Discussion

The simulations provide an extensive comprehension of the aerodynamic per-
formance of the hanger section in changing wind scenarios. The differentiation
between steady-state and transient simulations offered beneficial observations,
enabling the detection of temporally dependent occurrences, such as shedding
frequencies.

However, the steady solution overestimates the magnitude of the aerodynamic co-

65
 5.5. RESULTS DISCUSSION

efficients and fails to provide any information about the shedding frequency and
time-dependent response of the cross-section. This type of simulation remains a
valid technique for obtaining an overview of the aerodynamic effects on the struc-
ture and for investigating different factors that impact the dynamic behaviour of
the structure. Nonetheless, the steady-state simulation unveiled trends in aero-
dynamic coefficients that pertained to drag, lift, and moment. These trends are
also applicable to the transient solution. Importantly, the effects of geometry,
particularly the presence of rounded edges, showed a decrease in drag coefficients
and an improvement in the coefficient values stability. These findings establish
the foundation for succeeding transient simulations.

Moving to transient solutions, by exploring a single rectangular cross-section at

varying wind speeds, the dynamic response of the structure was clarified. The
variability in aerodynamic coefficients, particularly drag, highlighted the struc-
ture’s sensitivity to wind speeds of different magnitudes. The examination of
shedding frequencies and the Strouhal number disclosed a transition from consis-
tent to inconsistent vortex shedding configurations as the wind speed increased.

The examination of different sections within the field of wind brings to light the
complexity of inter-sectional interactions. However, the results obtained from
the multisectional model do not differ significantly from those obtained from the
single rectangular cross-section. This model incurs a higher computational cost.
Therefore, for this case study, it suffices to consider only one cross-section to rep-
resent the dynamic behaviour of the entire structure. However, this modelling can
be considered for advanced purposes to deeply investigate interactions between
diverse structures in a single wind domain.

The findings of the 3D transient analysis of Hanger H16 demonstrate a nuanced

pattern in aerodynamic coefficients, primarily in the drag coefficient, as wind
speeds increase. The differences revealed in the 3D model, in comparison to its 2D
equivalent, emphasize the importance of taking into account three-dimensional
factors in aerodynamic simulations. However, the 2D analysis can be deemed
sufficiently accurate for this thesis. Additionally, the unanticipated fluctuations
in lift frequency, particularly at designated wind speeds, introduce an added layer
of complexity to the examination. The constraints faced, particularly in obtaining
the drag-frequency correlation, highlight the difficulties associated with moving
from 2D to 3D simulations. The requirement for dynamic mesh configurations,
while improving accuracy, results in considerable computational demands.

66
5.5. RESULTS DISCUSSION

The incorporation of Eurocode standards into wind load calculations has high-
lighted the significance of examining the impact of wind velocities on critical
factors required to determine the inertial force linked to the analysis of vortex
shedding. Evaluating the Strouhal numbers against Eurocode values underscored
the variability with wind velocity, which has a crucial impact on the likelihood of
resonant vibrations, and increment the magnitude of the inertial wind force that
takes into account the effect of vortex shedding.

Essentially, a complete view of the aerodynamic characteristics of the hanger

section is obtained by thoroughly evaluating steady-state, transient simulations,
and wind load calculations. This study provides important insights into the
impact of wind on structures, which are crucial for developing and evaluating
safety protocols in wind engineering design. The results offer valuable information
that can be used to enhance the understanding of the subject.

67
Chapter 6

Conclusions and further research

6.1 Conclusion
This study aims to provide a thorough investigation into the vortex-induced vi-
bration on bridge hangers and wind forces. The set objectives outlined in the
introduction have been achieved through a carefully structured sequence of chap-
ters.

The most important outcomes of this study can be summarized in the following
points:

• Fundamental frequencies of the hangers: the first fundamental fre-

quency in the transverse direction turns out to be 6.6 Hz for the single
hanger H16, adding other hangers an increase in stiffness is observed and
slightly lower frequencies are recorded. While for all models the first fun-
damental frequency in the longitudinal direction is approximately 14 Hz.

• Effect of the radius: Two geometries were analysed for the studied cross-
section, in one the corner radius was included, and in the other not. When
the radius is not taken into account, higher co-efficiencies are obtained than
when the radius is taken into account, which leads to the consideration of
higher aerodynamic forces.

• Steady versus transient solution: For CFD analysis, the steady state
solution has a lower computational cost compared to the transient solu-
tion. However, it is less accurate as it does not allow the extraction of
time-dependent data used to calculate the frequency and consequently the

69
 6.2. FURTHER RESEARCH

Strouhal number.

• Aerodynamic quantities: Aerodynamic quantities are usually associated

with the geometry of the body analysed, but with the CFD analyses on
ANSYS Fluent software, it was possible to determine a strong correlation
between flow velocity and aerodynamic quantities. It is not possible to asso-
ciate a precise correlation, as the type of regime varies as the velocity varies.
For medium to low wind speeds (up to 10 m/s), the tendency is to record
coefficients that increase with increasing velocity. At higher speeds, there
is a drop due to an intense flow narrowing the turbulence tail. While for
the Strouhal number was possible to define a trend, as the speed increases
its value decreases, in the speed range of 2.5 - 30 m/s it goes from a value of
0.13 to 0.06.

• Wind response and fundamentals vibration of the hanger: By

analysing the drag frequency (transverse direction) and lift frequency (lon-
gitudinal direction) and comparing them with the fundamental frequencies
of the hanger in longitudinal and transverse directions, the most critical
direction results to be the transverse one. The drag frequency reaches the
fundamental frequency in the transverse direction for a wind speed of 7.5 -
8 m/s. This means that the structure may undergo resonance for a wind
speed equal to the one mentioned previously.

The finding can be of practical value in the monitoring of Bomarsund Bridge and
future designs of bridges, enhancing their safety and resilience. However, this
study presents a few limitations: only one angle of wind attack was investigated,
and the interaction with the structure wasn’t investigated.

6.2 Further research

This study aimed to comprehend the response of hangers to wind forces and ana-
lyze parameters characterizing the vortex-induced vibration phenomenon. Future
extensions and improvements could involve:

• Exploration of Wind Flow Angles: A natural extension involves study-

ing the effect of different angles of attack in the wind flow.

• Wind-Structure Interaction Studies: Future studies could delve into

the mutual influence between the structure and the wind, providing a nu-

70
6.2. FURTHER RESEARCH

anced understanding of the dynamic interplay.

• Direct Assessment of Wind Load: Future research could focus on di-

rectly evaluating the wind load on the structure by incorporating detailed
analyses of aerodynamic forces. This step could contribute to a more precise
calculation of the forces acting on the hangers.

• Evaluation of Structural Fatigue: Beyond vibrations, assessing struc-

tural fatigue induced by wind loading is a critical aspect of wind engineering,
these data can be used to assess the fatigue of the studied hanger.

These potential extensions and improvements pave the way for a more compre-
hensive exploration of wind-structure interactions, with direct applications in the
study of structural fatigue within the realms of wind engineering and bridges.

71
Bibliography

[1] European Committee for Standardization (CEN), Eurocode EN 1991-1-4:

Wind Load, 2005, CEN.

[2] Yunus A. Çengel and John M. Cimbala, Fluid Mechanics, 2008, McGraw-
Hill.

[3] John D. Anderson, Fundamentals of Aerodynamics, 2010, McGraw-Hill Ed-

ucation.

[4] J. Gordon Leishman, Introduction to Aerospace Flight Vehicles, 2022,

[5] Paul A. Durbin and Gorazd Medic, Fluid Dynamics with a Computational
Perspective, Chapter ´Creeping Flow´, 2007, Cambridge University Press.

[6] Pijush K. Kundu and Ira M. Cohen and David R. Dowling and Grétar Tryg-
gvason, Fluid Mechanics, Sixth Edition, chapter 10, Boundary Layers and
Related Topics, 2015, Academic Press.

[7] Gersten, K., Radespiel, Rolf and Rossow, Cord-Christian and Brinkmann,
Benjamin Winfried, Hermann Schlichting and the Boundary-Layer Theory,
Hermann Schlichting – 100 Years,pp 3-17, 2009, Springer Berlin Heidelberg.

[8] Nikolaos D. Katopodes, Viscous Fluid Flow Chapter 5, Free-Surface Flow,

pp 384-385 2019, Butterworth-Heinemann.

[9] J. Wright = Encyclopedia of Vibration,FLUTTER, pp 553-565, 2001, Else-

vier, Oxford.

[10] Alberto Carpinteri, Advanced Structural Mechanics, 2017, CRC Press, Taylor
and Francis Group.

[11] T.J.A. Agar, Aerodynamic Flutter Analysis of Suspension Bridges by a Modal

Technique, 1989, Engineering Structures.

72
BIBLIOGRAPHY

[12] Lian, Jijian and Yan, Xiang and Liu, Fang and Zhang, Jun and Ren, Quan-
chao and Yang, Xu, Experimental Investigation on Soft Galloping and Hard
Galloping of Triangular Prisms 2017, Applied Sciences.

[13] Feng Fu, Design and Analysis of Tall and Complex Structures, ch. 2 Funda-
mentals of Tall Building Design, pp. 48-49, 2018 Butterworth-Heinemann.

[14] A.G. Straatman, R.J. Martinuzzi, A Study on the Suppression Of Vortex

Shedding From a Square Cylinder Near a Wall, Vol. Engineering Turbulence
Modelling and Experiments 5, pp. 403-411, 2002, Elsevier Science Ltd.

[15] Z.T. Zhang and Y.J. Ge and Y.X. Yang, Torsional stiffness degradation and
aerostatic divergence of suspension bridge decks, 2013, Journal of Fluids and
Structures.

[16] E.H. Dowell and R. Clark and others, A Modern Course in Aeroelasticity,
Fourth Revised and Enlarged Edition, 2004, Kluwer Academic Publishers

[17] Raymond L. Bisplinghoff and Holt Ashley and Robert L. Halfman, Aeroe-
lasticity, 1996 Addison-Wesley.

[18] Ansys Fluent 12.0 Theory Guide, Modeling Turbulence, Chapter 12-24, 2009
Ansys Inc.

[19] Ansys Fluent 12.0 Theory Guide, Turbulence, Chapter 4: 4.4.3 Realizable k-
ϵ Model, 2009 Ansys Inc.

[20] Salim, Salim M. and Cheah, S.C., Wall y+ strategy for dealing with wall-
bounded turbulent flows, Vol 2, 2009, Int. MultiConf. Eng. Comput. Sci.
(IMECS).

[21] Karlson, M.; Nita, B.G.; Vaidya, A., Numerical Computations of Vortex
Formation Length in Flow Past an Elliptical Cylinder. Fluids Vol. 5, 157,
2020.

[22] Prasenjit Dey and Ajoy Kr. Das, Numerical analysis of drag and lift reduction
of square cylinder, Engineering Science and Technology, an International
Journal, Vol. 18-4, pp. 758-768, 2015.

[23] Hoerner, Fluid Dyamic Drag, (1965).

73
 BIBLIOGRAPHY

[24] Mehmet Ishak Yuce, Dalshad Ahmed Kareem, A Numerical Analysis of Fluid
Flow Around Circular and Square Cylinders, American Water Works Asso-
ciation AWWA, Vol. 108-10, pp. E546-E554, 2016.

[25] L. Mao, Q. Liu, G. Wang, and S. Zhou, Lift force, drag force, and tension
response in vortex-induced vibration for marine risers under shear flow, Vi-
broengineering, Vol. 18-2, pp. 1187–1197, 2016.

[26] Williamson C. H. K., Roshko A., Vortex formation in the wake of an oscil-
lating cylinder, Journal of Fluids and Structures, Vol. 2, pp. 355-381, 1988.

[27] Dahl J.M., Hover F.S., Triantafyllou M.S. Two-degree-of-freedom vortex-

induced vibrations using a force-assisted apparatus Journal of Fluids and
Structures, Vol.22, pp. 807-818, 2006.

[28] Enhao Wang, Qing Xiao, Atilla Incecik, Three-dimensional numerical simu-
lation of two-degree-of-freedom VIV of a circular cylinder with varying nat-
ural frequency ratios at Re=500, Journal of Fluids and Structures, Vol. 73,
pp. 162-182, 2017.

74
Appendix A

Abaqus Appendix

This appendix presents the Abaqus models, the ten first mode shapes and their
corresponding frequencies.

A.1 H16
Table A.1: Model H16: First ten Frequencies

H16

MODE EIGENVALUE FREQUENCY

(RAD/s) (Hz)
1 1733.1 41.63 6.6257
2 7798.2 88.307 14.055
3 24185 155.52 24.751
4 56929 238.6 37.974
5 1.10E+05 330.92 52.667
6 1.20E+05 346.63 55.167
7 1.58E+05 396.88 63.166
8 1.87E+05 431.88 68.735
9 1.88E+05 433.47 68.99
10 2.08E+05 455.53 72.5

76
A.1. H16

(a) Mode 1 (b) Mode 2

Figure A.1: Mode shapes model H16 (continued)

(c) Mode 3 (d) Mode 4

Figure A.1: Mode shapes model H16 (continued)

(e) Mode 5 (f) Mode 6

Figure A.1: Mode shapes model H16 (continued)

77
 A.1. H16

(g) Mode 7 (h) Mode 8

Figure A.1: Mode shapes model H16 (continued)

(i) Mode 9 (j) Mode 10

Figure A.1: Mode shapes model H16 (continued)

78
A.2. H15-16

A.2 H15-16
Table A.2: Model H15-16: First ten Frequencies

H15-16
MODE EIGENVALUE FREQUENCY
(RAD/s) (Hz)
1 1635.2 40.438 6.4359
2 2203.2 46.938 7.4705
3 7769.7 88.146 14.029
4 7998 89.431 14.233
5 19775 140.62 22.381
6 27128 164.7 26.214
7 42362 205.82 32.757
8 56621 237.95 37.871
9 58436 241.73 38.473
10 62383 249.77 39.751

(a) Mode 1 (b) Mode 2

Figure A.2: Mode shapes model H15-16 (continued)

79
 A.2. H15-16

(c) Mode 3 (d) Mode 4

Figure A.2: Mode shapes model H15-16 (continued)

(e) Mode 5 (f) Mode 6

Figure A.2: Mode shapes model H15-16 (continued)

(g) Mode 7 (h) Mode 8

Figure A.2: Mode shapes model H15-16 (continued)

80
A.2. H15-16

(i) Mode 9 (j) Mode 10

Figure A.2: Mode shapes model H15-16 (continued)

81
 A.3. H14-16

A.3 H14-16
Table A.3: Model H14-16: First ten Frequencies

14-16
MODE EIGENVALUE FREQUENCY
(RAD/s) (Hz)
1 1598.40 39.98 6.36
2 1890.00 43.47 6.92
3 2382.50 48.81 7.77
4 7735.50 87.95 14.00
5 7780.70 88.21 14.04
6 7980.80 89.34 14.22
7 14997.00 122.46 19.49
8 24906.00 157.82 25.12
9 26087.00 161.51 25.71
10 27910.00 167.06 26.59

(a) Mode 1 (b) Mode 2

Figure A.3: Mode shapes model H14-16 (continued)

82
A.3. H14-16

(c) Mode 3 (d) Mode 4

Figure A.3: Mode shapes model H14-16 (continued)

(e) Mode 5 (f) Mode 6

Figure A.3: Mode shapes model H14-16 (continued)

(g) Mode 7 (h) Mode 8

Figure A.3: Mode shapes model H14-16 (continued)

83
 A.3. H14-16

(i) Mode 9 (j) Mode 10

Figure A.3: Mode shapes model H14-16- (continued)

84
A.4. H14-17

A.4 H14-17
Table A.4: Model H14-17: First ten Frequencies

H14-17
MODE EIGENVALUE FREQUENCY
(RAD/s) (Hz)
1 1476.60 38.43 6.12
2 1823.50 42.70 6.80
3 2032.40 45.08 7.18
4 2499.40 49.99 7.96
5 6921.40 83.20 13.24
6 7672.00 87.59 13.94
7 7714.60 87.83 13.98
8 7922.00 89.01 14.17
9 8059.90 89.78 14.29
10 14404.00 120.02 19.10

(a) Mode 1 (b) Mode 2

Figure A.4: Mode shapes model H14-17 (continued)

85
 A.4. H14-17

(c) Mode 3 (d) Mode 4

Figure A.4: Mode shapes model H14-17 (continued)

(e) Mode 5 (f) Mode 6

Figure A.4: Mode shapes model H14-17 (continued)

(g) Mode 7 (h) Mode 8

Figure A.4: Mode shapes model H14-17 (continued)

86
A.4. H14-17

(i) Mode 9 (j) Mode 10

Figure A.4: Mode shapes model H14-17- (continued)

87
 A.5. H12-19

A.5 H12-19
Table A.5: Model H12-19: First ten Frequencies

H12-19
MODE EIGENVALUE FREQUENCY
(RAD/s) (Hz)
1 1578.7 39.733 6.3236
2 1814.3 42.595 6.7791
3 1930.5 43.938 6.9929
4 2016 44.9 7.146
5 2.11E+03 45.957 7.3143
6 2.26E+03 47.566 7.5703
7 2.41E+03 49.066 7.8092
8 2.47E+03 49.666 7.9046
9 3.95E+03 62.869 10.006
10 7.47E+03 86.427 13.755

(a) Mode 1 (b) Mode 2

Figure A.5: Mode shapes model H12-19 (continued)

88
A.5. H12-19

(c) Mode 3 (d) Mode 4

Figure A.5: Mode shapes model H12-19 (continued)

(e) Mode 5 (f) Mode 6

Figure A.5: Mode shapes model H12-19 (continued)

(g) Mode 7 (h) Mode 8

Figure A.5: Mode shapes model H12-19 (continued)

89
 A.5. H12-19

(i) Mode 9 (j) Mode 10

Figure A.5: Mode shapes model H12-19 (continued)

90
Appendix B

ANSYS Fluent Appendix

B.1 ANSYS Fluent 2D Simulation

This appendix presents tables related to CFD analysis.

91
 B.1. ANSYS FLUENT 2D SIMULATION

B.1.1 2D Steady simulation

Table B.1: 2D ANSYS Fluent simulation settings

Model Settigns
Space 2D
Time Steady & Transient
Viscous Realizable k-epsilon turbulence model
Wall treatment Enhanced wall treatment
Iterations Steady solution 6000/12000
Iterations Transient solution 6000
Time Step Transient solution [s] 0.010
Reference Area [m^2] 0.254
Method Simple

Material properties -Air

Density [kg/m^3] 1.225 constant
Viscosity [kg/(ms)] 1.79E-05 constant

Boundary Conditions
Velocity Magnitude [m/s] 2,5 / . . . / 30
Inlet Turbulent Intensity [%] 5.000
Hydraulic Diameter [m] 1.820
Gauge Pressure [Pa] 0.000
Outlet Backflow Turbulent Intensity [%] 5.000
Backflow Turbulent Viscosity Ratio 10.000
Wall Motion Stationary
Wall
Shear Boundary Condition No Slip

92
B.1. ANSYS FLUENT 2D SIMULATION

B.1.2 2D Transient Simulations

Table B.2: Transient solution: Aerodynamic coefficients for rectangular cross

section.

v
2.5 5.0 7.5 10.0 15.0 20.0 25.0 27.0 30.0
[m/s]
Re 4.3e4 8.7e4 1.3e5 1.7e5 2.6e5 3.5e5 4.3e5 4.7e5 5.2e5
CD 1.682 1.713 1.718 1.689 1.607 1.519 1.439 1.419 1.401
CL -0.689 0.814 -0.247 -0.078 -0.327 -0.158 0.057 0.007 -0.063
CM -8.693 -3.912 -7.397 -6.733 -7.227 -6.341 -5.330 -5.419 -5.575

Table B.3: Transient solution: Estimated peak frequencies and related ampli-
tudes.

v [m/s] 2.5 5.0 7.5 10.0 15.0

Drag Estimated Peak Frequency [Hz] 2.600 4.880 6.800 8.380 11.000
Lift Estimated Peak Frequency [Hz] 1.280 2.440 3.400 4.200 5.499
Drag Amplitude 0.023 0.026 0.031 0.025 0.014
Lift Amplitude 0.338 0.798 0.891 0.709 0.568

v [m/s] 20.0 25.0 27.0 30.0

Drag Estimated Peak Frequency [Hz] 13.080 14.180 14.460 14.840
Lift Estimated Peak Frequency [Hz] 6.549 7.082 7.220 7.432
Drag Amplitude 0.004 0.001 0.001 0.000
Lift Amplitude 0.302 0.110 0.075 0.065

Table B.4: Transient solution: Strouhal number

v [m/s] 2.5 5.0 7.5 10.0 15.0 20.0 25.0 27.0 30.0
St 0.129 0.124 0.115 0.107 0.093 0.083 0.072 0.068 0.063

93
 Table B.5: Transient solution data extracted for each speed simulated

v [m/s] 2.5 5.0 7.5 10.0 15.0 20.0 25.0 27.0 30.0
n° time steps 6000 6000 6000 6000 6000 6000 6000 6000 6000
steps size [s] 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010 0.010
CD 1.682 1.713 1.718 1.689 1.607 1.519 1.439 1.419 1.401
CL -0.689 0.814 -0.247 -0.078 -0.327 -0.158 0.057 0.007 -0.063
CM -8.693 -3.912 -7.397 -6.733 -7.227 -6.341 -5.330 -5.419 -5.575
Drag f [Hz] 2.600 4.880 6.800 8.380 11.000 13.080 14.180 14.460 14.840
Lift f [Hz] 1.280 2.440 3.400 4.200 5.499 6.549 7.082 7.220 7.432
[Link] 0.023 0.026 0.031 0.025 0.014 0.004 0.001 0.001 0.000
[Link] 0.338 0.798 0.891 0.709 0.568 0.302 0.110 0.075 0.065
St 0.129 0.124 0.115 0.107 0.093 0.083 0.072 0.068 0.063
D. Force [N] 1.634 6.656 15.025 26.252 56.220 94.471 139.774 160.804 195.949
L. Force [N] 0.669 3.162 2.158 1.216 11.421 9.839 5.518 0.773 8.768

B.2 ANSYS Fluent 3D Simulations

Table B.6: Transient 3D analysis: simulation extracted data.

Wind speed [m/s] 2.5 10.0 20.0 27.0 30.0

Re 4.34E+04 1.74E+05 3.47E+05 4.69E+05 5.21E+05
n° time steps 6000 6000 6000 6000 6000
time steps size [s] 0.010 0.010 0.010 0.010 0.010
Drag Coeff. CD 1.307 1.235 1.234 1.234 1.233
Lift Coef. CL -0.008 -0.053 -0.053 -0.053 -0.053
Moment Coeff. CM 1.233 0.077 1.166085 0.011 1.165
Lift Estimated Peak Frequency [Hz] 1.340 2.960 5.680 7.420 8.120
Drag Amplitude 8.70E-03 1.03E-05 9.41E-06 1.05E-05 1.16E-05
Lift Amplitude 1.49E-02 5.44E-05 4.30E-05 4.37E-05 4.65E-05
Estimated Strouhal Number from Lift 0.134 0.074 0.071 0.069 0.068
Drag Force [kN] 1.41E-02 2.14E-01 8.54E-01 1.56E+00 1.92E+00
Lift Force [kN] 8.91E-05 9.25E-03 3.69E-02 6.73E-02 8.31E-02
TRITA-ABE-MBT-23632