Adam Freeman - 071229 - Thesis PDF

Bachelor of Engineering Thesis
Experimental Testing of the Effects of Stiffness

Variation on Vortex Induced Vibrations of Subsea
Risers
Adam Freeman
26th October 2011
Supervisor: Mr Rowan Frost
Thesis submitted in partial fulfilment of the requirements for the degree

of
Bachelor of Engineering (JEE 418)
National Centre for Maritime Engineering and Hydrodynamics

DECLARATION
This project report contains no material which has been accepted for a degree or
diploma by AMC, University of Tasmania or any other institution, except by way of
background information and duly acknowledged in the report, and to the best of my
knowledge and belief, no material previously published or written by another person
except where due acknowledgement is made in the text of the report.
Signed:
Date:
STATEMENT 1
This project report is the result of my own investigation, except where otherwise
stated. Other sources are acknowledged in the text giving explicit references. A list of
references is appended.
Signed:
Dated:
STATEMENT 2
I hereby give consent for my project report to be available for photocopying, inter-
library loan, electronic access to AMC and UTAS staff and students via the UTAS
Library, and for the title and summary to be made available to outside organisations.
Signed:
Dated:
ii
Experimental Testing of the Effects of Stiffness Variation on Vortex Induced Vibrations
ACKNOWLEDGEMENTS
 My family for their continued support during this project and throughout my degree
 My supervisor, Mr Rowan Frost for your guidance and technical assistance
 My 4th year mates
 AMC staff who gave guidance at various stages throughout the project, in particular:
o Dr Shinsuke Matsubara for your support and Matlab prowess;
o Dr Shuhong Chai for your late words of wisdom;
o Kirk Meyers, Michael Underhill and Rowan Cairns for your assistance in
conducting decay tests
 Alan Faulkner for your assistance at the Flume Tank
Adam Freeman Page i`

TABLE OF CONTENTS
Declaration ................................................................................................................................ii
Acknowledgements ..................................................................................................................... i
Table of Contents .......................................................................................................................ii
List of Figures ........................................................................................................................... iv
List of Tables ...........................................................................................................................vii
Nomenclature ......................................................................................................................... viii
Axis System .............................................................................................................................. ix
Units ........................................................................................................................................... x
Terms, Abbreviations and Definitions ...................................................................................... xi
Abstract ....................................................................................................................................xii
1. Introduction ............................................................................................................. 1
1.1 Background ...................................................................................................... 1
1.2 Aim of Research Project .................................................................................. 2
2. Theory...................................................................................................................... 3
2.1 Natural Frequency ............................................................................................ 3
2.2 Reynolds Number ............................................................................................. 4
2.3 Strouhal Number .............................................................................................. 6
2.4 Vortex Shedding Frequency ............................................................................. 7
2.5 Added Mass ...................................................................................................... 8
2.6 Reduced Velocity ............................................................................................. 9
2.7 Free Span Response ....................................................................................... 11
2.8 Bending Stiffness ........................................................................................... 11
2.9 System Tensioning ......................................................................................... 12
3. Literature Review .................................................................................................. 13
4. Experimental Testing ............................................................................................. 19
4.1 Testing Facilities ............................................................................................ 19
4.2 Testing Apparatus .......................................................................................... 19
5. Experimental Progression ...................................................................................... 23
5.1 Testing Regime .............................................................................................. 24
5.2 Model Scaling ................................................................................................ 25
5.3 Model Selection.............................................................................................. 25
5.4 Member Natural Frequency ........................................................................... 26
5.5 Applicable Codes ........................................................................................... 28
6. Experimental Models ............................................................................................. 29
6.1 Instrumentation............................................................................................... 29
6.2 End Connections ............................................................................................ 31
Adam Freeman Page ii`
6.3 Member Tensioning ....................................................................................... 32

6.4 Testing Procedure ........................................................................................... 33
7. Results and Discussion .......................................................................................... 35
7.1 Fast Fourier Transform................................................................................... 35
7.2 Natural Frequencies........................................................................................ 37
7.3 Tension ........................................................................................................... 41
7.4 Thin-Walled PVC Member ............................................................................ 44
7.5 Thick-Walled PVC Member .......................................................................... 47
7.6 Aluminium Member ....................................................................................... 52
7.7 Model Comparison ......................................................................................... 57
7.8 Full Scale Implications ................................................................................... 64
8. Conclusions ........................................................................................................... 66
9. Future Recommendations ...................................................................................... 67
References ................................................................................................................................ 68
APPENDIX A .......................................................................................................................... 70
APPENDIX B .......................................................................................................................... 71
Adam Freeman Page iii`

LIST OF FIGURES
Figure 1-1 Velocity distribution near the point of separation (Espeyrac & Pascaud, 2002) ..... 1
Figure 2-1 Simplified representation of vibration model (Ojeda, 2011) ................................... 3
Figure 2-2 Fluid flow regimes across a smooth cylinder (Laidlaw, 2010) ................................ 6
Figure 2-3 Strouhal number as a function of Reynolds number (Techet, 2005) ....................... 7
Figure 2-4 Vortex shedding patterns of a circular cylinder (Sarpkaya, 2004)........................... 8
Figure 3-1 Von Karman Vortex Street (Espeyrac L., Pascaud S., 2002) ................................ 14
Figure 3-2 Effects of varying riser tension due to self-weight (Chen et al., 2011) ................. 15
Figure 3-3 Modal shape of riser with varying stiffness increasing from left to right (Chen et
al., 2011) .................................................................................................................................. 16
Figure 3-4 RMS of displacement for tension T = 1.91 x 107 N (Chen et al., 2011) ................ 16
Figure 3-5 Numerical results of RMS hydrodynamic force of CF direction (Tang et al., 2011)
.................................................................................................................................................. 17
Figure 3-6 Numerical results of RMS hydrodynamic force of IL direction (Tang et al., 2011)
.................................................................................................................................................. 18
Figure 4-1 Testing apparatus design basis (Michael, 2011) .................................................... 20
Figure 4-2 Experimental model top supporting structure ........................................................ 21
Figure 4-3 Base supporting structure front section .................................................................. 22
Figure 5-1 Testing configuration used to conduct decay analyses .......................................... 28
Figure 6-1 Accelerometer axes and ports used for DAQ system connection .......................... 30
Figure 6-2 Accelerometer moulds used to create waterproof casing ....................................... 31
Figure 6-3 Flume tank DAQ system ........................................................................................ 32
Figure 6-4 Tensioning mechanism used during experimental testing ..................................... 33
Figure 7-1 Accelerometer orientation and voltage output (Dimension Engineering, 2012) ... 35
Figure 7-2 FFT plot of thick wall PVC tested at T = 0kg and v = 0.8m/s ............................... 36
Figure 7-3 Decay test for thin wall PVC member with 490.5 N applied tension .................... 37
Figure 7-4 Displacements each middle accelerometer for v = 1.21m/s and T = 981 N .......... 41
Figure 7-5 Effects of tension on the aluminium model at constant v = 0.8m/s ....................... 42
Figure 7-6 Middle accelerometer displacement at T = 0kg and v = 1.21m/s .......................... 44
Figure 7-7 Cross flow response for each accelerometer at T = 0 N ........................................ 45
Figure 7-8 Cross flow and in-line flow response of the thin-walled PVC member at T = 490.5
N............................................................................................................................................... 46
Figure 7-9 Effects of Reynolds number on non-dimensional amplitude at T = 490.5 N ........ 47
Adam Freeman Page iv`

Figure 7-10 Middle accelerometer displacement at v = 1.7m/s and T = 0 N .......................... 48

Figure 7-11 Cross flow and in-line flow response of the thick-walled PVC member at T = 0N
.................................................................................................................................................. 49
Figure 7-12 Cross flow response at each accelerometer for T = 0 N ...................................... 50
Figure 7-13 Cross flow and in-line response at for T = 490.5 N ............................................. 51
Figure 7-14 Effects of Reynolds number on non-dimensional amplitude at T = 490.5 N ...... 52
Figure 7-15 Middle accelerometer displacement at v = 1.21m/s and T = 490.5 N ................. 53
Figure 7-16 FFT plot of aluminium member at T = 490.5 N .................................................. 54
Figure 7-17 Cross flow and in-line flow response of the aluminium member at T = 490.5 N 54
Figure 7-18 Cross flow and in-line flow response of the aluminium member at T = 981 N ... 55
Figure 7-19 Effects of Reynolds number on non-dimensional amplitude at T = 490.5 N ...... 56
Figure 7-20 Cross flow response for each model at v = 1.7 m/s and T = 490.5 N .................. 58
Figure 7-21 Comparison of cross flow and in-line flow for each member for v = 1.7 and T =
490.5 N..................................................................................................................................... 59
Figure 7-22 Cross flow response for each model at v = 1.7 m/s and T = 981 N ..................... 60
Figure 7-23 Comparison of cross flow and in-line flow for each member at v = 1.7 m/s and T
= 981 N .................................................................................................................................... 61
Figure 7-24 Cross flow comparison for each bending stiffness at v = 0.4 m/s and T = 490.5 N
.................................................................................................................................................. 62
.................................................................................................................................................. 63
.................................................................................................................................................. 64
Figure 9-1 Effects of tension on the thin wall PVC model at constant v = 0.8 m/s................. 71
Figure 9-2 Effects of tension on the thick wall PVC model at constant v = 0.8 m/s ............... 71
Figure 9-3 Effects of tension on the aluminium model at constant v = 0.8 m/s ...................... 72
Figure 9-10 Effects of reduced velocity on the thin wall PVC model at T = 0 N ................... 75
Figure 9-11 Effects of reduced velocity on the thin wall PVC model at T = 490.5 N ............ 76
Adam Freeman Page v`
Figure 9-12 Effects of reduced velocity on the thin wall PVC model at T = 981 N ............... 76
Figure 9-13 Effects of reduced velocity on the thick wall PVC model at T = 0 N ................. 77
Figure 9-14 Effects of reduced velocity on the thick wall PVC model at T = 490.5 N .......... 77
Figure 9-15 Effects of reduced velocity on the thick wall model at T = 981 N ...................... 78
Figure 9-16 Effects of reduced velocity on the aluminium model at T = 0 N ......................... 78
Figure 9-17 Effects of reduced velocity on the aluminium model at T = 490.5 N .................. 79
Figure 9-18 Effects of reduced velocity on the aluminium model at T = 981N ...................... 79
Adam Freeman Page vi`

LIST OF TABLES
Table 2-1 VIV response to reduced velocity (Chai, 2012) ........................................................ 9
Table 2-2 .................................................................................................................................... 9
Table 2-3 Basic non-dimensional VIV parameters (Chakrabarti, 2005) ................................. 10
Table 2-4 Free span characteristics as a function of L/D (DNV RP-F105, 2006) ................... 11
Table 4-1 AMC Flume Tank specifications (AMC, 2012) ...................................................... 19
Table 5-1 Effects of increased material property on natural frequency................................... 23
Table 5-2 Vortex shedding frequencies for proposed experimental models ........................... 24
Table 5-3 Model testing time estimate..................................................................................... 24
Table 5-4 Froude scaling factors for model design (Chakrabarti, 2005) ................................. 25
Table 5-5 Properties of three materials used for testing .......................................................... 26
Table 5-6 Relevant project codes ............................................................................................. 28
Table 6-1 Comparison between two models of accelerometer considered for testing ............ 29
Table 6-2 Colour configuration used to identify vibration axes .............................................. 30
Table 7-1 Natural frequencies of each model from the dry decay test compared to calculated
values ....................................................................................................................................... 38
Table 7-2 Wetted natural frequencies of each member used for testing.................................. 39
Table 7-3 Natural and response frequencies for each experimental data set ........................... 39
Table 7-4 Adjusted vortex shedding frequencies..................................................................... 40
Table 7-5 Member dry natural frequency variations due to tension ........................................ 42
Table 7-6 Response characteristics of each model at v = 1.7 m/s and T = 490.5 N ................ 59
Table 7-7 Response characteristics of each model at v = 1.7 m/s and T = 981 N ................... 61
Adam Freeman Page vii`

NOMENCLATURE
Symbol Description Units
a* Structural aspect ratio Dimensionless
A* Vibration amplitude ratio Dimensionless
ζ Damping ratio Dimensionless
D Cylinder diameter m
E Young’s Modulus GPa
fn Natural frequency Hz
fs Vortex shedding frequency Hz
g Gravitational acceleration m/s2
I Moment of inertia m4
λ Scaling factor Dimensionless
k Spring stiffness N/m
L Length m
m’ Added mass per unit length kg
ρ Density kg/m3
P Paired modal response Dimensionless
Re Reynolds number Dimensionless
S Single mode response Dimensionless
St Strouhal number Dimensionless
t Wall thickness mm
T Tension N
µ Dynamic viscosity Ns/m2
Uc Free stream fluid velocity m/s
V Volts Volts
υ Kinematic viscosity m2/s
v Flow velocity m/s
VR Reduced velocity Dimensionless
Adam Freeman Page viii`

AXIS SYSTEM
The following axis system has been used throughout the project to describe accelerometer
response.
 X axis – Perpendicular in plane to fluid flow
 Z axis – Parallel to fluid flow
 Y axis – Vertical
Fluid flow
Z
Z
Profile View Plan View
Adam Freeman Page ix`

UNITS
Hz Hertz
kg Kilograms
kW Kilowatts
m Meters
N Newtons
Pa Pascals
s Seconds
V Volts
Adam Freeman Page x`

TERMS, ABBREVIATIONS AND DEFINITIONS

3D Three dimensional
AMC Australian Maritime College
BNC Bayonet Neill-Concelman
CF Cross-flow
CFD Computational Fluid Dynamics
CWC Circulating Water Channel
DAQ Data Acquisition
DE Dimension Engineering
DNV Det Norske Veritas
DOF Degree(s) of Freedom
EI Flexural rigidity
FFT Fast Fourier Transform
IL In-line
OS Offshore standards
PVC Polyvinyl Chloride
PVC 1 Thin-walled PVC testing member
PVC 2 Thick-walled PVC testing member
RP Recommended practices
RPM Revolutions per Minute
VIV Vortex-induced vibrations
Adam Freeman Page xi`

ABSTRACT
Expanding material technologies are continually increasing the variety and availability of
riser materials and properties in the offshore oil and gas industry. Flexible pipe use has seen a
particularly large growth rate in recent years due to its adaptability to various subsea
applications including umbilicals, flowlines and jumpers. Varied flexural rigidity causes
changes in natural frequency and riser response to vortex shedding. Predicting and
eliminating vortex induced vibration (VIV) response of these risers is critical to maintaining
the structural integrity of the system, particularly as exploration expands in ultra-deep water.
Experimental testing was conducted in the AMC Flume Tank facility at Beauty Point on three
models of the varying stiffness with the same outer diameter. Stiffness variation was achieved
with the use of two PVC models of differing wall thickness and a thin-walled aluminium
member. A decay test was performed on each model to determine the natural frequencies of
each system. Models were subject to varied tension to replicate the effects of self-weight in
deep water. Flow velocities were varied in regular intervals and response accelerations
measured at three equidistant locations on each model. Response in both cross flow and in-
line flow was recorded and analysed.
Lower stiffness properties typically demonstrated the largest response amplitude as a result of
the decreased natural frequency. This trend was particularly prominent at increased fluid
velocity where cross flow amplitude dominates response as a result of alternating pressure
distribution. Tensioning effects on the natural frequency range was considerable, placing
members outside the range of potential lock-in. Effects of the large variation in bending
stiffness were compared, with instable response characteristics resulting from PVC models at
lower reduced velocities. It was shown that non-dimensional amplitude significantly
decreased as a result of increased bending stiffness, which was particularly prominent for
conditions of high system tension.
Adam Freeman Page xii`

1. INTRODUCTION
1.1 BACKGROUND
The ever-evolving search for hydrocarbons requires new technologies for production in ultra-
deep water. The large riser span required for these depths changes material properties of the
riser and amplifies the effects of environmental loadings. It is the cause of much research into
understanding and reducing these forces.
Vortex-induced vibration (VIV) is the most important dynamic response of a deep water riser
(Beardman, 2009). It is a phenomenon caused by a fluctuating pressure distribution as fluid
passes over a cylindrical object. A boundary layer created by the fluid viscosity separates
from the body due to the bluff oscillatory forces perpendicular to fluid flow. Vortices formed
after separation lead to an alternating pressure distribution, causing the body to vibrate.
Shedding of vortices create a lifting forces on the body, leading to transverse vibration. These
vortices can be categorised as singular or paired eddies, or a combination of both (Figure
1-1).
Figure 1-1 Velocity distribution near the point of separation (Espeyrac & Pascaud, 2002)
The existence of VIV has been known for centuries. The ancient Greeks used the
phenomenon in the design of the Aeolian harp to transfer wind energy into musical sounds
(Beardman, 2009). It has been the focus of significant research in the offshore industry since
the nineteenth century. As flexible risers are becoming more common in deeper waters,
research into VIV is an increasing priority for the offshore industry. Associated increases in
riser length with water depth bring new challenges to VIV behaviour of risers. Varying riser
tension due to increased self-weight and changing current profiles are typical examples of
problems that add to the challenges of deep water exploration.
Adam Freeman Page 1`

Early investigation into VIV began with the first research paper of George Stokes (1842). He
suggested the possibility of an incompressible, inviscid fluid deviating from computed flows
and raised the prospect of continual perpetual eddying. Osbourne Reynolds (1883) later
conducted experimental investigations into the transition between laminar and turbulent flow
in the case of vertical pipes. This resulted in the development of a ratio between inertia and
viscosity later known as Reynolds number.
At certain excitation frequencies, periodically alternating fluid separation initiates vortex

shedding. When the frequency of the excitation force approaches one of the natural
frequencies of the cylinder, oscillation amplitude significantly increases. This is due to vortex
shedding and VIV. Resonance between the vortex shedding frequency and natural frequency
of the system is known as “lock-in”. It has the ability to detrimentally affect the operational
lifespan of subsea risers.
Effects of VIV at high Reynolds numbers are still not completely understood. Much of what
the industry knows about the phenomenon is the result of observational and empirical
analysis. There is a large scope for future research into increasing the current knowledge on
VIV behaviour. This research project is being conducted in the hope of doing so.
1.2 AIM OF RESEARCH PROJECT

The aim of this Research Project is to investigate the effects of stiffness variation on VIV of
subsea risers. Three experimental models of varying bending stiffness will be tested in pre-
determined flow velocities and tensions to achieve a state of lock-in and subsequent
correlation of results.
Previous studies have been carried out in relation to the experimental setup of the AMC
Flume Tank facility to enable subsequent theses to be carried out without extensive research
into the design of experimental testing apparatus. This information will be used as an
experimental design basis for testing and modified as required. Experimental results will be
used to determine correlations between risers of varying stiffness under a variety of testing
conditions. Results will be made non-dimensional where possible to enable use for future
applications. This will develop our knowledge for the prediction of full scale modal
frequencies of flexible risers.

2. THEORY
2.1 NATURAL FREQUENCY
VIV of a bluff body occurs when the boundary layer created around the surface separates
from the body and creates a pressure differential on the body. This pressure can be generated
in line with the body or perpendicular to the flow. As the pressure fluctuates the member is
forced into periodical oscillations which are dependent on the structural properties of that
body. A significant increase in response amplitude occurs when the natural frequency of the
bluff body coincides with the vortex shedding frequency. This phenomenon is known as
‘lock-in’. The natural frequency of a simplified system is governed by mass and stiffness and
can be idealised by a one degree of freedom oscillatory system (Figure 2-1).
Figure 2-1 Simplified representation of vibration model (Ojeda, 2011)
The natural frequency of the testing member can be determined by idealizing the system as a
linear-elastic beam (Palmer & King, 2006). Equation (1) can be used to determine the pipe
natural frequency, taking into account boundary constraints and pipe length.
C F
fn  (1)
L2 m
Where: fn = Natural frequency of the system (Hz)

C = End constraint constant
L = Pipe length (m)
F = Flexural rigidity (Nm2)
m = Mass per unit length (kg/m)

Vibration of a system can take two forms:

1) Free vibrations - This occurs when the structure responds to an initial excitation.
Oscillatory motion occurs at one or more of the member’s natural frequencies.
2) Forced vibrations – This occurs when the structure responds to cyclic external
excitations. Resonance occurs when the natural frequency of the member coincides
with the excitation frequency.
The natural frequency of a system is the oscillatory motion caused by free vibration. The first
resonant frequency is known as the ‘fundamental’ frequency. Higher modal frequencies are
encountered at various increments above the fundamental frequency. Knowledge of the
natural frequency of any floating system is critical to avoid the occurrence of lock-in.
Alteration of the system natural frequency of a subsea riser can be achieved by altering the
system parameters listed in equation (1). Due to the exponential relationship of riser length
with natural frequency coupled with the ability to alter the free span, changing the effective
length of the riser is the most effective method of avoiding the lock-in regime.
Chakrabarti (2005) outlines the important hydrodynamic quantities influencing VIV as:
 Reynolds number,
 Shedding frequencies and their interactions,
 Added mass (or mass ratio) and damping,
 Lift coefficient, and
 Correlation of force components.
2.2 REYNOLDS NUMBER

Flow past a bluff body can be categorised by the non-dimensional ratio of inertial to viscous
forces, known as the Reynolds number. Flow is categorised as either laminar or turbulent. As
the fluid flows past the object viscous stresses cause laminar flow at the surface of the object
which has the same velocity as the object itself. A boundary layer is created, resulting in a
velocity profile. Shear stresses result at adjacent layers fluid particles move between layers as
turbulence causes an interchange of momentum. These adjacent layers in turn undergo
reduced velocity until the edge of boundary layer at 99% free stream velocity, where there are
no further effects of the shear stresses and the boundary layer is negligent. As the distance

from the body increases, so does the Reynolds number as turbulence is reached at a certain
point.
Drag occurring at the surface of the body is due to skin friction which is dependent on the
roughness of the surface. Skin friction typically dominates drag at lower Reynolds numbers.
Form, or pressure drag, is the dominant friction force at higher Reynolds numbers where the
shape of the object influences the inertial drag.
Dynamic similitude is achieved by matching the Reynolds number of the experimental and
full scale models. Reynolds number is calculated using equation (2).
VD VD
Re   , (2)
 
Where: Re = Reynolds number

ρ = Density of free stream fluid (kg/m3)
V = Steady velocity of the upstream fluid (m/s)
D = Cylinder diameter (m)
µ = Dynamic viscosity of free stream fluid (Ns/m2)
υ = Kinematic viscosity of the fluid (m2/s)
As the diameter of the models and kinematic viscosity of the fluid will be constant, the
Reynolds will be directly proportional to the fluid velocity. To achieve the highest likelihood
of lock in a range of 300 < Re <3 x 105 is required for a fully turbulent vortex street
(Lienhard, 1966). Values higher than this range will result in turbulence, dissipating the
vortices until the turbulent vortex street is re-established at 3.5 x 106 < Re (Figure 2-2).

Figure 2-2 Fluid flow regimes across a smooth cylinder (Laidlaw, 2010)
2.3 STROUHAL NUMBER

Strouhal number is a dimensionless factor used to describe oscillating flow mechanisms. As
high Reynolds number viscous forces become insignificant. Vortex shedding frequencies
causing lock-in with the bluff body can be represented in terms of free stream velocity and
diameter.
fsD
St  , (3)
Uc
Where: St = Strouhal number

fs = Vortex shedding frequency (Hz)
Uc = Free stream fluid velocity (m/s)
Cylinders undergoing vortex shedding exhibit hydrodynamic instability, or Strouhal

Instability, which can be identified by a phenomenon known as Von Karman Vortex Street.

Strouhal number as a function of Reynolds is shown in Figure 2-3.
Figure 2-3 Strouhal number as a function of Reynolds number (Techet, 2005)
Strouhal number is shown to be approximately 0.2 for Reynolds numbers in the Von Karman
Vortex Street range i.e. 40 < Re < 105. This approximation has been used in calculations to
determine vortex shedding frequencies of testing members.
2.4 VORTEX SHEDDING FREQUENCY

The vortex shedding frequency of a structure is the excitation frequency at which lock-in
occurs. Flow velocity is varied inducing resonance when matched with the structure’s natural
frequency. The vortex-shedding frequency for a fixed, rigid cylinder in uniform flow can be
calculated by rearranging equation(3).
Uc
f s  St , (4)
D
Where: fs = Vortex shedding frequency of cylinder

St = Strouhal number
Uc = Free stream fluid velocity (m/s)
A value of 0.2 can be used as an approximation for Strouhal Number in the region 2.5 x 102 <
Re < 2.5 x 105 (Chakrabarti, 2005) which can be seen in Figure 2-3. If the natural frequency
of the member matches the calculated natural frequency resonance is likely to occur.

Vortices can be described in terms of singular and/or paired modes. Factors effecting vortex
shedding behaviour are fluid velocity, member natural frequency, cylinder diameter, and
response amplitude. Figure 2-4 can be used to categorise expected vortex response for a
variety of flow regimes.
Figure 2-4 Vortex shedding patterns of a circular cylinder (Sarpkaya, 2004)
The 2P mode relates to the splitting of a region of vorticity in each half cycle, causing two
counter-rotating pairs to be shed per cycle. The 2S mode is categorised by the shedding of a
single vortex in each half cycle (Sarpkaya, 2004). P + S modes occur due to asymmetric wake
modes, with paired vortices on one side of the wake and single on the other (Blackburn &
Henderson, 1996). 2S, 2P and P + S modes are the principle modes near the fundamental
lock-in region (Williamson & Govardhan, 2004).
2.5 ADDED MASS

Hydrodynamic added mass is the effective increase in mass of a body undergoing
acceleration or deceleration. The increase is caused by the displacement of surrounding fluid
creating a pressure on the body during acceleration. This pressure causes the fluid force to
oppose the motion. The effect of added mass on the body can be calculated using equation (5)
m '   c 'V (5)
Where: m’ = Added mass

ρ = Fluid density (kg/m3)
c’ = Added mass coefficient
V = CSA of model (m2)
The effective mass of the riser is the sum of the mass of the body and the added mass.
Additional force placed on the riser from the fluid changes the natural frequency of the
system and must be accounted for during any calculations relating to the wetted natural
frequency of the riser.
2.6 REDUCED VELOCITY

Variation in flow velocity causes various behavioural responses of the bluff body. Vortex
shedding response characteristics of the body can be categorised by the reduced velocity
region in which it lies. Reduced velocity can be calculated using equation(6) (DNV-RP-F105,
2006).
Uc
VR  , (6)
fn D
Where: VR = Reduced velocity

Uc = Mean fluid velocity (m/s)
fn = Natural frequency of cylinder (Hz)
A simplified relationship between VIV response and reduced velocity can be seen in Table
2-1.
Table 2-1 VIV response to reduced velocity (Chai, 2012)
Table 2-2
Range Response Characteristic

1.0<VR<2.2 Symmetric vortex shedding, causing in-line oscillations
2.2<VR<3.5 Alternative vortex shedding, causing in-line oscillations
4.8<VR<12 Alternative vortex shedding, causing cross-flow oscillations
Reduced velocity is a non-dimensional parameter which allows models of varying system

properties to be compared and evaluated. This will be a key component in the experimental
analysis to account for the variation in natural frequencies and flow velocity regime. Reduced

velocities at for each analysis will be calculated and response behaviour compared to Table
2-1.
Additional fundamental non-dimensional parameters effecting VIV are given in Table 2-3.
Vibration amplitude ratio will also be a key parameter used in the experimental analysis. This
allows non-dimensional analysis of oscillatory response and accounts for variations in model
diameter.
Table 2-3 Basic non-dimensional VIV parameters (Chakrabarti, 2005)
Structural aspect ratio l

a* 
D
Vibration amplitude ratio A
A* 
D
Mass ratio (total includes hydrodynamic
added mass) total mass m
m*  
structure length  D 2
Damping ratio c
 
cc

2.7 FREE SPAN RESPONSE

VIV characteristics relating to free span of a cylinder can be categorised in terms of the
cylinder length to diameter ratio. This can be used to predict the presence of VIV and the
dominant behavioural characteristics. An overview of typical free span characteristics is
given below in Table 2-4.
Table 2-4 Free span characteristics as a function of L/D (DNV RP-F105, 2006)
L/D Response Description

L/D < 30 Very little dynamic amplification.
Normally not required to perform comprehensive design check.
Insignificant dynamic response from environmental loads expected
and unlikely to experience VIV.
30 < L/D < 100 Response dominated by combined beam and cable behaviour.
Typical span length for operating conditions. Natural frequencies
sensitive to boundary conditions (and effective axial force).
100< L/D < 200 Response dominated by combined beam and cable behaviour.
Relevant for free spans at uneven seabed in temporary conditions.
Natural frequencies sensitive to boundary conditions, effective axial
force (including initial deflection, geometric stiffness) and pipe
“feed in”.
L/D > 200 Response dominated by cable behaviour.
Relevant for small pipe diameters in temporary conditions. Natural
frequencies governed deflected shape and effective axial force.
2.8 BENDING STIFFNESS

Bending stiffness is a key criterion in determining model response. Bending stiffness, also
known as flexural rigidity in the case of beams, is a measure of the deflection of the object in
relation to the applied bending moment. Bending stiffness of the testing member can be
obtained using equation(7).
F   R3tE (7)
Where: F = Flexural rigidity (Nm2)

R = Mean cylinder radius (mm)

t = Wall thickness (mm)

E = Young’s Modulus (GPa)
Bending stiffness values required to accurately scale the model are given in Error!
Reference source not found. and Error! Reference source not found.. Material selection
will be based on the Young’s Modulus and wall thickness of the model to achieve the
required bending stiffness. Comparison of deflections at various locations on each model will
enable resonance behaviour and modal shape to be determined and related to the bending
stiffness characteristics of the riser.
2.9 SYSTEM TENSIONING

The natural frequency of the structure will increase in accordance with system tension. The
increased range in natural frequency achieved by subjecting each model to a range of tensions
will increase the likelihood of lock-in. An adjustment to equation(1) can be made to include
the effects of tension variation on the member (Beards, 1996).
T
ft  fn  (8)
4 A L2
Where: ft = Natural frequency with applied tension (Hz)

T = Applied tension (N)
A = Member cross sectional area (m2)
Tension will be applied to the member via a pulley system to act as a simple and accurate
method of ensuring the same conditions between testing members. The system used in the
Flume Tank also enables the model to undergo unrestricted translation. The same tensions
used in experimental testing will be applied to each model during a decay test to replicate
testing properties of the riser and calculate the natural frequency of the system.

3. LITERATURE REVIEW
Despite intensive research into VIV, behavioural response predictions of cylindrical systems
at high Reynolds numbers remains elusive. Sarpakaya (2004) lists the ultimate objectives of
current VIV research as “the understanding, prediction, and prevention of VIVs (preferably
without drag penalty), partly through the direct numerical simulations of fundamental as well
as industrially significant fluid–structure interactions.” While certain key parameters
influencing vortex shedding behaviour of structures are easily identifiable, Sarpakaya also
states that “the robust parameters do not serve as unique identification cards for the
integrated effects of the individually non-quantifiable parameters.”
Progress in the understanding of numerical and experimental effects of VIV is generally

restricted to the low-Reynolds number regime. Sarpakaya explains this as “VIV is not a small
perturbation superimposed on a mean steady motion. It is an inherently nonlinear, self-
governed or self-regulated, multi-degree-of-freedom phenomenon.” Much of what is
currently understood about the high-Reynolds number regime is based on a combination of
empirical and descriptive knowledge.
Czech physicist Vincent Strouhal conducted the first recognised experimental analysis
involving long slender members in the nineteenth century to explain the singing of overhead
power lines. He concluded that transverses oscillations occurred when the vortex shedding
frequency coincided with the natural frequency of the wire (Bearman, 2009). This
preliminary analysis leads to the development of the Strouhal number.
Theodore von Karman (1911) further analysed the alternating double-row of vortices behind
a bluff in a fluid stream (Figure 3-1), now known as Von Karman Vortex Street (Casillas,
2010).

Figure 3-1 Von Karman Vortex Street (Espeyrac L., Pascaud S., 2002)
Extensive research conducted by Morse and Williamson (2009) focused on variation of

cylinder amplitude and frequency at two constant Reynolds numbers, Re = 4000 and 12000,
in a flume tank. The rigid cylinder was controlled to oscillate transversely to free stream. The
transverse force acting on the cylinder subject to forced sinusoidal motions was measured
while data was recorded on in-phase and out-of-phase response to the excitation force.
Oscillatory motions were sustained for the in-phase component while the out-of phase
motions act as an inertia force affecting the oscillation of the cylinder. Previous debates
regarding similarities between flows generated by freely oscillating and forced oscillated
cylinders questioned the flow similarities between the two for flow history and harmonics.
The influence of possible nonlinearities in structural quantities such as stiffness and damping
has previously been thought to have a potential affect. Morse and Williamson have
demonstrated that under carefully controlled conditions there is a very close correspondence
between the two (Bearman, 2009).
The associated increase in riser length with increasing water depth alters the structural
configuration of the riser. “Axial tension is no longer regarded as a constant along the
overall length of riser due to considerable structure weight.” (Chen, Li, Zheng and Tan,
2011). Structural inconsistencies across the riser length lead to variations in vortex shedding
behaviour at these areas. Coupled with the variation in current velocity profile this can lead to
significant discontinuities in behavioural response of the riser.
Chen et al. also investigated the effects of self-weight on natural frequency of the riser.
Figure 3-2 compares the effects on a 1000m long riser of outer and inner diameters 0.500m
and 0.445m respectively, top tension 6.24 x 66 N and a top tension factor T/W is 1.57 where
W is the riser weight.

Figure 3-2 Effects of varying riser tension due to self-weight (Chen et al., 2011)
They describe the relationship between modal wave length and response amplitude as “for
lower modes the amplitude is larger at the axial position where the tension is smaller,
whereas for higher modes the response amplitude is larger at the axial position where the
bending stiffness is lower.” Variation in riser tension along the model length will be difficult
to achieve during experimental testing. This deepwater analysis will give a good comparison
for data on the effect of tension variation on oscillation amplitude.
Chen, Li, Zheng and Tan also go on to discuss the effects of stiffness distribution on VIV.
This will be another point of analysis for model testing. “The distribution of the amplitude
along the riser length is approximately inversely proportional to the distribution of the
stiffness EI(x) that becomes more evident for higher mode, e.g. mode 43” (the last plot in
Figure 3-3). Models tested will be fabricated from homogenous materials, however as with
tension distribution this will be a good source for data comparison and analysis.
“For the case of a beam with axially varying bending stiffness EI(x), amplitude peak would
move along the beam length from the position with larger stiffness EI(x) to the position with
smaller EI(x) as the mode number increases.” (Refer to Figure 3-3).

Figure 3-3 Modal shape of riser with varying stiffness increasing from left to right (Chen et al., 2011)
Displacement response for the model riser subject to tension T = 1.91 x 107 N with increasing
stiffness from the right to the left of the riser can be seen in Figure 3-4. It exhibits an increase
in amplitude and reduction in frequency in the lower section of the riser as a result of the
stiffness variation. As Chen et al. explain, “This amplitude distribution is similar with that of
the mode shape.”
Figure 3-4 RMS of displacement for tension T = 1.91 x 107 N (Chen et al., 2011)

This statement is confirmed by Tang, Lu, Teng, Park, Song and Zhang (2011) who noted a
“similar periodic characteristic between hydrodynamic force and displacement” during their
experimental analysis of a long flexible riser in uniform flow. Three different pre-tensions
were tested of 600 N, 700 N and 800 N which remained uniform across the length of the riser
during testing. It was observed that the odd-time harmonic frequencies (i.e. fn,1 and fn,3) are
associated with the cross-flow (CF) motions, while the even-time (i.e. fn,2 and fn,4) are
associated with the in-line (IL) direction.
Numerical analysis conducted by Tang et al. presented the effects of higher order frequency
on hydrodynamic force. Figure 3-5 shows the results of these forces at fn,1 and fn,3 i.e. CF
dominated frequencies at various sections of the riser length (z/L).
Figure 3-5 Numerical results of RMS hydrodynamic force of CF direction (Tang et al., 2011)
The largest value of hydrodynamic CF clearly occurs at fn,3 which has a value of
approximately 14 times the hydrodynamic force occurring at fn,1. This value demonstrates the
importance of ensuring that the structure will not be subject to higher modal frequencies.
Testing was also conducted on the in-line flow dominated frequencies (fn,2 and fn,4) where a
similar trend is observed (Figure 3-6).

Figure 3-6 Numerical results of RMS hydrodynamic force of IL direction (Tang et al., 2011)
Analysis of the results indicated that the damping force is much smaller in comparison to the
other two hydrodynamic force components of reduced mass and stiffness. Tang et al.
summate the damping effects as “The contribution of the damping force to the total
hydrodynamic force can be neglected without any significant loss of accuracy.”
Researched proposed for this project will lead on from these literature reviews. Focus will be
on developing a relationship between the harmonic effects of various stiffness and tensions of
homogenous risers. These articles will be reviewed after experimental testing for relevant
data comparison.

4. EXPERIMENTAL TESTING
4.1 TESTING FACILITIES
Testing will be conducted in the AMC Flume tank located in Beauty Point, Tasmania. The
tank was constructed to test behavioural properties of structures in a variety of currents. It has
the ability to simulate drag and lift forces due to seabed interaction using the floor conveyor
belt which runs the full length of the test section. Tank specifications are outlined in Table
4-1.
Table 4-1 AMC Flume Tank specifications (AMC, 2012)
Length 17m
Width 5m
Depth 2.5m
Water holding capacity 700,000L
Flow velocity 0 - 1.7 m/s
Conveyor belt velocity 0 - 1.7 m/s
Electric motor power 4 x 90.0kW
Propeller diameter 1.2m
The largest section of the tank features an 11m wide viewing window for close observation
during testing. Testing can also be viewed from overhead from the observational carriage
where a monitor displays a real-time profile view of the experiment. An interface located on
the carriage enables flow velocities and DAQ system to be controlled and monitored from the
same location.
4.2 TESTING APPARATUS

The experimental apparatus used will be based on previous studies regarding optimising
experimental setup for VIV research by Michael (2011). The apparatus has been designed to
provide stability for the model while allowing access to accelerometer data cables. Testing
setup is already constructed and commissioned within the Circulating Water Channel (CWC).
Supporting apparatus consists of a top supporting frame and bottom supporting base, as
shown in Figure 4-1.

Figure 4-1 Testing apparatus design basis (Michael, 2011)
Some modifications will be applied to the testing apparatus to improve testing capabilities.
Hydraulic tensioning of the riser will be replaced by a simplified method of weights attached
to a pulley system. This will achieve a broader range of testing tensions, which was
previously limited to 500N based on the hydraulic cylinder capacity. Maximum tension
applied to the pulley system will be limited by the load cell to approximately 1000N. This
system will simulate the self-weight of the riser to some degree. The DAQ system located at
the tank has also been replaced.
4.2.1 TOP SUPPORTING APPARATUS

The top of the riser will be supported by a stainless steel frame attached to the Flume Tank
observational carriage. The carriage is able make longitudinal micro adjustments to achieve
vertical orientation of the model. Load cells will be placed at the top of each model to gauge
tensioning forces before and during testing. The universal joints used to attach the riser to the
supporting frame are installed to create minimal movement restrictions and an accurate
depiction of uninhibited riser response. Natural frequency is reduced by the pin connection,
increasing the likelihood of lock-in for the stiffer models. The supporting frame is shown in
Figure 4-2.

Figure 4-2 Experimental model top supporting structure
The top support structure also consists of an additional aluminium carriage that is able to be
placed on the tracks of the flume tank during large load cases. Under recommendation from
previous studies, the carriage will be implemented during testing to reduce and potentially
eliminate structural vibration of the aluminium frame during testing at higher frequencies.
Previous testing conducted without the addition of the aluminium carriage recorded vibration
amplitude of 12mm (Michael, 2011).
4.2.1 BASE SUPPORTING APPARATUS

The base supporting frame was designed by Michael (2011) to create a stable gravity base
platform for the testing member while minimising flow interference. The frame consists of a
reinforced aluminium plate with an integrated front section for member connection and
additional steel weights applied on each side of the streamline scoop. Lifting lugs are located
at the extremities of the raised section to allow installation and removal of the base. The
combined weight of the base plate and front section is approximately 250kg, providing a
sufficiently stable platform for the model. The front section as of the base plate as designed
can be seen below in Figure 4-3.

Figure 4-3 Base supporting structure front section
CFD was conducted on the base support before construction and results indicate minimal
flow interference on the model. Any turbulence created by the structure does not appear to
form within a close proximity of the riser, while flow disturbance created by the clump
weights is on the outer regions of the flow and does not appear to interfere with the
downstream flow.

5. EXPERIMENTAL PROGRESSION
Models will be fabricated from three different materials of varying stiffness at a constant
diameter to gauge the effects of stiffness scaling. Each model will be subject to varying
tensions and flow velocities to achieve the highest potential for lock-in.
Selection of experimental model is crucial to achieve an accurate representation of the full

scale riser for analysis. Testing members will be selected based on:
1) Accessibility of materials
2) Material properties
3) Member dimensions
4) Member geometry (internal and external)
Material properties and their effects on the member’s natural frequency are given in Table
5-1.
Table 5-1 Effects of increased material property on natural frequency
Increased Material Effect on Natural

Property Frequency
Mass Decrease
Stiffness Increase
Diameter Decrease
Length Decrease
Wall thickness Increase
End constraint Increase
Young’s Modulus Increase
Tension Increase
Three accelerometers will be installed in each testing member to indicate riser response at
equidistant locations. Each will be made waterproof by resin casing and wired to the Flume
Tank DAQ system. The use of multiple accelerometers will allow modal response
fluctuations to be identified.
The natural frequencies of each model will be determined in air by a decay test and compared
against numerical values. This will involve the testing member being subject to free vibration
while natural oscillation periods are recorded. Effects of added mass will then be added to the

model and wetted natural frequency calculated. Discrepancies calculated from the predicted
and dry natural frequencies will be used to estimate wetted natural frequencies.
Strouhal number can be used to determine vortex shedding frequencies of testing members at
varying flow velocity for each considered model diameter. These values have been calculated
using an approximated Strouhal number of 0.2, as discussed in Section 2.3.
Table 5-2 Vortex shedding frequencies for proposed experimental models
Flow Velocity Model Diameter (mm)

(m/s) 120 100 80 60 40
0.4 0.67 0.80 1.00 1.33 2.00
0.8 1.33 1.60 2.00 2.67 4.00
1.2 2.00 2.40 3.00 4.00 6.00
1.6 2.67 3.20 4.00 5.33 8.00
5.1 TESTING REGIME

Testing of each member will occur at flow velocities given Table 5-2. Riser tensions will be
applied based on calculations indicating the natural frequency coinciding with vortex
shedding frequencies also listed in Table 5-2. Testing models will consist of three different
stiffness variations at a constant diameter.
An overall model testing time estimate is given in Table 5-3.
Table 5-3 Model testing time estimate
Task Duration
Setup 8 hours
Testing 36 runs @ 20 mins each
Repeat testing (contingency) 15 runs @ 20 mins each
Pack up 8 hours
Total 33 hours
Note: Run time estimates have included flow acceleration.

5.2 MODEL SCALING

Any scaling required will be carried out using Froude scaling. This will achieve structural
and hydrodynamic similitude with the full scale riser and assist in structural parameter
calculation for full scale applications. Froude scaling factors used for design are listed below
in Table 5-4.
Table 5-4 Froude scaling factors for model design (Chakrabarti, 2005)
Variable Symbol Scale Factor

All linear dimensions D λ
Fluid or structure velocity u λ1/2
Fluid or structure acceleration ü 1
Time or period t λ1/2
Structure mass m λ3
Structure moment of inertia I λ5
Section moment of inertia I λ4
Structure displacement volume V λ3
Structure restoring moment C λ4
Force F λ3
Moment M λ4
Stress σ λ
Spring constant K λ2
Pressure p λ
Gravity g 1
Fluid density ρ 1
Fluid kinematic velocity ν 1
Reynolds number Re λ3/2
5.3 MODEL SELECTION

Selection of testing members was based on achieving lock-in by matching natural frequencies
of the model with the vortex shedding frequency. The three materials used for experimental
testing with associated properties are listed below in Table 5-5 Properties of three materials
used for testing

Table 5-5 Properties of three materials used for testing
Outer Wall Weight in Submerged Bending

Material
Diameter Thickness Air Weight Stiffness
PVC #1 42.66 mm 2 mm 0.50 kg/m 3.10 kg/m 1,426.96 Nm2
PVC #2 39.75 mm 4 mm 0.88 kg/m 2.91 kg/m 1,939.85 Nm2
Aluminium 39.79 mm 1.6 mm 0.66 kg/m 2.95 kg/m 19,598.23 Nm2
Materials were purchased from various hardware stores and steel manufactures in the
Launceston vicinity. It was decided to use the two previous models tested by Michael (2011)
and construct a third model of differing stiffness. Thick walled PVC was chosen as the most
appropriate stiffness in between the two previous models tested. This was the only available
off-the-shelf material available that would be suitable for testing. 43mm aluminium round bar
was purchased for fabrication of connection pieces. The outer diameter was selected to be
slightly larger than the model diameter to enable machining of connection pieces to the same
diameter as the testing model.
5.4 MEMBER NATURAL FREQUENCY

Models have been selected as close to the design diameter of 40mm as possible. Table 2-4
can be used to determine likely free span characteristics as a function of L/D. For the 2m
models used for analysis free span response can be categorised from Section 2.7.
2.0
L/D 
0.04
 50
Free span ratio indicates the riser will be susceptible to VIV and sensitive to boundary
conditions. This category represents the typical free span ration of most operating conditions
as given in Table 2-4. Calculation of the natural frequency of each member was conducted
for comparison with experimental data using equation (1). Results are given in Appendix A.
A free decay test was conducted on all the models to determine the member natural frequency
and associated reduced velocity required to achieve lock-in. Test would ideally have been
conducted in the Flume Tank to replicate model testing environment, however practically this

was not possible due to tank depth and accessibility. The test was therefore conducted in the
AMC Towing Tank. The effects of added mass have been added to the results to account for
the displaced fluid during testing.
The towing tank DAQ system and power supply were used for data collection and processing
via Labview. BNC connections were used to record data in the X axis for each model at the
three tensions used for experimental analysis. Vertical alignment in the X and Z directions
was achieved using a spirit level.
Decay testing apparatus was designed to enable tensioning of the system while replicating
experimental testing conditions as closely as possible. The model was tensioned from the
bottom section using experimental weight increments. The lower section was secured to a
tensioning system which was attached to a universal joint. An adjustable clamp was placed
on the tensioning rod, negating any horizontal translation caused by the hanging weights as
the pipe underwent vibration. This configuration restricted horizontal translation at the both
ends of the model while allowing tensioning of the system. The 90kg mass of the bottom
supporting structure helped to restrict horizontal motion. End connections resembled
experimental testing conditions by causing universal joints to behave as a pin connection.
The top section was also restricted from horizontal translation above the universal joint to
replicate a pin connection. A plate was constructed with a welded connection piece matching
the inside diameter of the universal joints. The plate was then clamped to the roof beam and
the top universal joint secured to the connection piece with a 6mm stainless steel pin. The
setup used to conduct the decay test is shown below in Figure 5-1.

Figure 5-1 Testing configuration used to conduct decay analyses
5.5 APPLICABLE CODES

Design and analysis of models will be conducted in accordance with codes listed in Table
5-6.
Table 5-6 Relevant project codes
Code Title Year

DNV-OS-F101 Submarine Pipeline Systems 2010
DNV-OS-F201 Dynamic Risers 2010
DNV-RP-D101 Structural Analysis of Piping Systems 2008
DNV-RP-F105 Free Spanning Pipelines 2006
DNV-RP-F204 Riser Fatigue 2010

6. EXPERIMENTAL MODELS
Model diameter was identified as a key design criterion to enable the highest chance of lock-
in. The smallest diameter of 40mm was chosen as the most suitable model for testing based
on its low natural frequency and increased possibility of lock-in. An additional PVC model of
increased wall thickness was constructed to increase the natural frequency of the model.
Some modifications were made to the existing models to allow the required interface between
the data cables and DAQ system. The thick walled PVC model was constructed using the
details given below.
6.1 INSTRUMENTATION
ADXL330 and DE-ACCM3D accelerometers were both considered for model testing.
Comparisons between models were made based on functionality, price and availability. A
comparison of specifications between the two models is given below in Table 6-1.
Table 6-1 Comparison between two models of accelerometer considered for testing
ADXL330 DE-ACCM3D Required

|Size 4mm x 4mm x 1.5mm 21mm x 10mm x 8mm Fit in 35.7mm ID pipe
Axis 3D 3D 2D
Bandwidth 550 Hz 500 Hz > 200 Hz
Accelerations ±3g ±3g > 1g
Water Proofing No No Ideal
Availability 7 days 7 days ASAP
DE-ACCM3D accelerometers from Dimension Engineering were selected for use in the
experimental models. Inclusive of the criteria given in Table 6-1, this model was readily
available been used previously in VIV testing and proven sufficient. Motion is detectable on
a 3 dimensional axis system, however only 2 dimensional data is required for cross-flow and
in-line flow response analysis. Accelerometer properties are given in Figure 6-1.

Figure 6-1 Accelerometer axes and ports used for DAQ system connection
After accelerometer procurement wiring was required for DAQ connection. Four core cables
were soldered to the X, Z, positive and negative terminals. Associated colours and wiring
configuration is shown below in Table 6-2.
Table 6-2 Colour configuration used to identify vibration axes
Accelerometer Terminal Colour

X axis Green
Z axis White
Positive Red
Negative Yellow
Once the accelerometers were wired and shrink wrapped waterproofing was required. This
was achieved using epoxy resin casings. Moulds for each accelerometer were constructed
using offcuts from the testing model. A small length of five core cable was placed vertically
on the inside of the mould to allow cables to run vertically on the inside of the model once
the casings were installed. Moulds and cable spacers were then taped internally so that the
resin was able to detach once set. Wired accelerometers were then positioned vertically inside
the mould and coated with electrical Grade 2 part joining epoxy resin. This particular resin
was selected to reduce curing temperature and potential circuitry damage. The accelerometer
moulds can be seen below in Figure 6-2.

Figure 6-2 Accelerometer moulds used to create waterproof casing
A single cable with B & C connections for input into the DAQ system was then soldered to
the 4 core cable. This allowed a single input into the DAQ system for each X and Z data
output from the 4 cord cable. Wired accelerometers were then screwed into the model at the
centre and equidistant spacing of 0.5m above and below the centre of the model. Additional
wiring was required for the existing models to link the 5 core cables to the DAQ system using
the same circuitry configuration as the thick walled PVC model. Calibration was then
conducted on the load cell using intervals of known weights. Output voltage was recorded
and graphed to validate tension readings and subsequent data analysis.
6.2 END CONNECTIONS

End connection pieces were machined from 43mm solid aluminium round bar. The top
connection piece enabled the model to be pinned to the universal connection joint and attach
to the pulley tensioning system. The pin also reduced the natural frequency of the member to
increase the chance of lock-in and enable fast interchange with the other models.
The bottom connection also consisted of an aluminium connection piece attached to a

universal joint. Connection with the base plate was achieved with a threaded connection piece
also machined from the solid aluminium round bar which was able to be screwed into a
threaded connection on the base plate. This enables models to be easily removed and installed
from the tank carriage. Universal joints at each end if the model enables it to behave as pin a
connection member, reducing the natural frequency of the system.

Once installed into testing position the model was connected to the data acquisition (DAQ)
system. DAQ system used for testing was the NI PXIe-4330 with integrated signal
conditioning, shown in Figure 6-3.
Figure 6-3 Flume tank DAQ system
6.3 MEMBER TENSIONING

Based on recommendations from previous testing conducted at the AMC Flume Tank a new
system was implemented to place each member in tension. A pulley system was constructed
using the carriage crane to suspend a specially fabricated supporting frame attached to a
bicycle wheel. A pulley rope was connected to each member via a load cell and over the
bicycle wheel where it shackled to the weights. Change of system tension during
experimental modelling was achieved by lowering the frame using the carriage crane and
interchanging weights. This system enabled tensioning of the system to be easily altered in
known increments during testing and negated tensioning restrictions encountered during
previous testing. The system tensioning mechanism used for testing is shown in Figure 6-4
below.

Figure 6-4 Tensioning mechanism used during experimental testing
6.4 TESTING PROCEDURE

Correct testing procedure is crucial to ensure valid output data. Information recorded from
each run must be correctly documented and filed to enable easy identification of run numbers
and results during post-processing. The following procedure outlines the steps taken in model
testing and data recording.
1. Setup and install model as outlined in Section 5. Ensure riser orientation is vertical
using laser calibration.
2. Secure model to the testing top and bottom supporting structure and attach tensioning
system.
3. Plug all cables into DAQ system.
4. Tension system with initial weight and check load cell voltage output.
5. Turn on power to the DAQ system and check each accelerometer is producing a
steady output voltage.
6. Setup run sheet and recording files.
7. Record zeros when carriage is steady.
8. Convert flow velocity to RPM and input into DAQ system.

9. Record 30 seconds of data once steady flow velocity is reached.

10. Monitor accelerometer output in real-time to ensure each accelerometer is producing
an output voltage.
11. Document results in run sheet including time of run, material, tension, average flow
velocity and any observations made during the recording interval.
12. Increase flow velocity by inputting the RPM of the next highest flow velocity into the
Flume Tank control interface.
13. Once all four flow velocities have been recorded stop the electric motors and increase
system tension.
14. Repeat steps 7 – 13 for the next highest system tension until all data has been
recorded.
15. Remove testing member and replace with the next model.
16. Repeat procedure from step 2 until all required data has been recorded and verified.

7. RESULTS AND DISCUSSION

Testing was conducted on each member for three different tensions and four different flow
velocities. 30 seconds of data was recorded for each run at a sampling rate of 200Hz. Data
has been post-processed and converted to meaningful data using Matlab. Fast fourier
transforms (FFT’s) were performed on each run to determine fundamental and resonant
frequencies at each flow velocity. This data was then compared to theoretical values to
determine any correlation, as well as data validation. Raw data was smoothed using a high
pass filter, eliminating any effect of noise on results. Smoothed voltages were then converted
to accelerations in the X and Z axes using an equation supplied in the accelerometer
specifications by adjusting for the 3.3V supply, sensitivity of 333mV/g and voltage output in
each axis as shown below in Figure 7-1.
Figure 7-1 Accelerometer orientation and voltage output (Dimension Engineering, 2012)
Acceleration values were then integrated twice to convert each data set to displacement
values. Non-dimensional model response amplitude can be compared at each load condition
to determine instances of lock-in. This data was then used to compare response of each
accelerometer on individual models and to identify any correlations in stiffness between
behaviour. Modal shape is also able to be indicated by accelerometer amplitude at each
section of the model.
7.1 FAST FOURIER TRANSFORM

Fast fourier transforms have been used to analyse data and extract key information from the
large data set. Analysis was initially conducted in Microsoft Excel and later in Matlab due to
the restricted maximum number sample size of 4096 in Excel. Fast fourier transforms are an
algorithm used to convert data from the time domain to the frequency domain. Dominant

frequencies and associated maximum displacements are able to be identified from large data
sets. If the system frequency response is close to that of the natural frequency of the system,
non-linear displacements should confirm the instance of lock-in. An example of several
modal responses determined by FFT can be seen below in Figure 7-2.
FFT Plot Thick-Wall PVC

0.35
X axis
Z axis
0.3
0.25
0.2
0.15
0.1
0.05
0
0 2 4 6 8 10 12 14 16 18
Frequency (Hz)
Figure 7-2 FFT plot of thick wall PVC tested at T = 0kg and v = 0.8m/s
Results from the FFT show several modes of frequency response. As discussed by Tang et al.
(2011), odd frequencies are representative of dominant response in the x axis, while even
frequencies correspond with z axis response. These modal frequencies will induce increased
oscillatory response if matched with the member natural frequency and can account for large
amplitude response of the model at the top and bottom accelerometers.
FFT’s and displacement analyses were performed on each run. Correlations of results
between various tensions and stiffness have been investigated at each flow velocity. Because
of the extensive set of results only a select sample is included in the body of results. Full
results are given in Table 7-3.

Displacements of each testing model have been made non-dimensional using the vibration
amplitude ratio and reduced velocity. This enables results to be compared with data from
various diameter models.
7.2 NATURAL FREQUENCIES

As mentioned in Section 5.4 a dry decay test was conducted on all experimental models at
each of the three tensions used for testing. FFT’s and displacement plots were produced for
each decay test to present the data in the frequency domain, indicating the natural frequency
of each tensioned model. This enabled an estimation of submerged natural frequencies for
each condition to be determined based on the added mass. Each model was struck firmly in
the centre with a mallet causing the pipe to undergo free oscillation. Results from the thin
walled PVC decay test under 490.5N of tension can be seen in Figure 7-3.
Decay Test
1.8
Volts
1.3
0.8
0.3
13.8 14.3 14.8 15.3 15.8 16.3
Time (s)
Figure 7-3 Decay test for thin wall PVC member with 490.5 N applied tension
In order to calculate wetted natural frequencies a spreadsheet was first set up to calculate the
dry natural frequencies of each member using formulae given in Section 2. These values were
then compared to data obtained in the dry decay test, given below in Table 7-1.

Table 7-1 Natural frequencies of each model from the dry decay test compared to calculated values
0N 490.5 N 981 N
Member
Calculated Actual Calculated Actual Calculated Actual
PVC Thin 6.7 Hz 6.7 Hz 11.3 Hz 14.0 Hz 13.2 Hz 14.2 Hz
PVC Thick 5.9 Hz 6.5 Hz 10.9 Hz 14.8 Hz 12.9 Hz 13.9 Hz
Aluminium 21.5 Hz 22.4 Hz 26.5 Hz 22.2 Hz 28.6 Hz 26.4 Hz
Added mass can be calculated and combined with the linear mass of the system to account for
fluid displacement by the riser during testing. As discussed in Section 2.5, added mass can be
calculated using Eq(5). Added mass coefficient has been taken as 1 for all cylindrical
members, while diameter varies slightly in accordance with Table 5-5. An example
calculation of added mass of the aluminium model is given below in equation (9)
m '   c 'V

 1000 1  0.0382 (9)
4
 1.13kg / m
The added mass calculated for each model can then be added to the linear dry mass and the
internal fluid column to determine the effective submerged linear mass of the system.
Empirically determined natural frequency from the dry decay test can then be used to
estimate the natural frequency of the system in fresh water by including the added mass into
equation (1).
C F
fn  (10)
L2 m  m'
Calculated values for dry natural frequencies were then expressed as a fraction of the results
from the decay test. The wetted natural frequency was then calculated and multiplied by the
error fraction of the decay test results to give an estimated value based on dry frequency
discrepancies. This was done as an error approximation as the same structural parameters
were used in the calculation of wetted natural frequency. Results are given below in Table
7-2.

Table 7-2 Wetted natural frequencies of each member used for testing
Member 0N 490.5 N 981 N

PVC Thin Wall 2.69 Hz 9.04 Hz 9.90 Hz
PVC Thick Wall 3.56 Hz 11.17 Hz 11.05 Hz
Aluminium 10.59 Hz 12.69 Hz 15.90 Hz
As mentioned in section Section 1, the vortex shedding frequency must match the natural
frequency of the testing member to achieve lock-in. To asses model response it is important
to identify which flow velocities are more likely to enter the lock-in regime and which are
less likely to be affected. Calculated natural frequencies have been compared against vortex
shedding frequencies generated using FFT analysis and given below in Table 7-3.
Table 7-3 Natural and response frequencies for each experimental data set
0N 490.5 N 981 N
Member Velocity Natural Response Natural Response Natural Response
0.4 m/s 2.69 Hz 3.67 Hz 9.04 Hz 5.03 Hz 9.90 Hz 3.65 Hz
Thin Wall
0.8 m/s 2.07 Hz 2.60Hz 6.37 Hz

PVC
1.2m/s 2.50 Hz 2.90 Hz 3.42 Hz

1.7 m/s 4.20 Hz 4.20 Hz 4.47 Hz
0.4 m/s 3.56 Hz 1.63 Hz 11.17 Hz 6.37 Hz 11.05 Hz 6.52 Hz
Thick Wall
0.8 m/s 2.27 Hz 6.17 Hz 7.43 Hz

PVC
1.2m/s 2.77 Hz 6.97 Hz 7.50 Hz

1.7 m/s 4.30 Hz 9.10 Hz 8.90 Hz
0.4 m/s 10.59 Hz 11.30 Hz 12.69 Hz 11.42 Hz 15.90 Hz 12.37 Hz
Aluminium
0.8 m/s 11.87 Hz 12.23 Hz 12.24 Hz

1.2m/s 11.83 Hz 12.30 Hz 12.67 Hz
1.7 m/s 12.47 Hz 12.90 Hz 13.20 Hz
Vortex shedding frequencies have been adjusted from Section 2.4 to account for slight
discrepancies in testing model diameter. Results for each flow condition are given below in

Table 7-4.

Table 7-4 Adjusted vortex shedding frequencies
Member 0.4 m/s 0.8 m/s 1.2 m/s 1.7 m/s

PVC 1 1.88 Hz 3.75 Hz 5.63 Hz 7.97 Hz
PVC 2 2.01 Hz 4.03 Hz 6.04 Hz 8.55 Hz
Aluminium 2.01 Hz 4.03 Hz 6.04 Hz 8.55 Hz
For lock-in to take place the shedding frequency must match or approach the natural
frequency of the member, as mentioned in Section 2.3. This will develop a coupling between
the structural motion and the wake dynamics. The shedding frequency in the wake of the
body is shown to generally increase with flow velocity and should be proportional with fluid
flow (Chakrabarti, 2005). Table 7-3 shows a general increase in response frequencies with an
increase in flow velocity. A non-linear trend can be seen in the thin walled PVC which can be
attributed to instability caused by low flexural rigidity. Data indicates natural frequencies of
the two PVC members are within vortex shedding range in the zero tension condition. These
members will be outside the range of lock-in when under tension. The natural frequency of
the aluminium member places it out of lock-in range for all testing conditions due to the flow
velocity limitations of the Flume Tank.
Analyses were conducted in Matlab using raw accelerometer data. Voltages were converted
to accelerations based on the voltage output in the zero position and divided by the
sensitivity. Displacements at each accelerometer were determined by integrating the
accelerations twice. This information was then plotted and to indicate vibration response at
each accelerometer. An example of cross flow displacements experienced at the middle
accelerometer for the same tension and flow velocity is given below in Figure 7-4.

Displacement at Middle Accelerometers
Alum
0.1 PVC 1
PVC 2
0.05
Displacement (m)
-0.05
-0.1
7 7.2 7.4 7.6 7.8 8

Time (s)
Figure 7-4 Displacements each middle accelerometer for v = 1.21m/s and T = 981 N
Figure 7-4 shows the standard model reactions when subject to identical experimental
conditions. Individual analysis of each member will be conducted initially to determine key
behavioural response of each model focusing on areas of increased amplitude response.
Results will then be compared between members to investigate and identify the effects of
VIV in response to varying flexural rigidity. Particular attention has been focused on cross
flow behaviour as this generally displays the largest response amplitude and potential for
structural detriment.
7.3 TENSION
Before analysis can take place the effect of tension variation on model response must be
identified in order to determine the effect in conjunction with flow velocity. Large variations
in member natural frequencies were recorded during the dry decay test. The increase due to
natural frequency on each of the models during the dry decay test is summarised below in

Table 7-5.

Table 7-5 Member dry natural frequency variations due to tension
Model Increase % Increase

PVC Thin Wall 7.5 Hz 211.9 %
PVC Thick Wall 7.4 Hz 213.8 %
Aluminium 4.0 Hz 117.9 %
Determining the linearity of tension effects on member natural frequency will aid in assessing
response behaviour and lock-in regimes. As mentioned in Section 2.8 an increase in system
tension results in an increase in member natural frequency. To identify the effects of tension
on the cross flow and in-line flow response the flow velocity can be held constant so that any
effects on vibration amplitude ratio are due solely to tension variation. Aluminium has been
used for tension comparison in Figure 7-5 due do its reasonably constant response
mechanisms.
Effects of Tension on Vibration Amplitude

0.4
CF
IF
0.35
0.3
0.25
A/D
0.2
0.15
0.1
0.05
0
0 200 400 600 800 1000
Tension (N)
Figure 7-5 Effects of tension on the aluminium model at constant v = 0.8m/s
As can be seen in Figure 7-5 tension has a substantial effect on model response amplitude. A
reduced velocity range of 1.91 to 1.27 resulted from the variation in system stiffness. The
largest differential in cross flow and in-line flow occurs at the lowest tension with a cross

flow over three times the amplitude of the in-line response. As expected cross flow reduces as
the model is. This results in a reduction of over 32 times and 9 times the zero-tension cross
flow and in-line flow respectively. The increase in in-line flow at 4905 N tension can be
attributed to the reduced velocity of 1.59, placing it in the in-line oscillation regime as
outlined in Table 2-1 Further results of tension effects can be seen in Appendix B. On
analysis of these results it can be seen that no set trend is observed due to tension effects.
This will make it difficult to relate any results directly to the effects of tension. Reference can
be made to tensioning effect results in Appendix B to account for any unexpected trends in
data.

7.4 THIN-WALLED PVC MEMBER

Low flexural rigidity due to decreased Young’s Modulus of the thin-walled PVC member
(PVC 1) leads to an increased likelihood of large deflections during vortex shedding.
Significant displacements were recorded for the higher reduced velocities. This corresponds
with the increase in amplitude and wavelength in sections of lower rigidity as stated by Chen
et al (2011). Displacement response of the member for cross flow and in-line flow are
therefore expected to be most significant at low tension. Model response at zero tension and
1.21m/s flow velocity can be seen in Figure 7-6.
Displacement at Middle Accelerometer

0.05
X axis
0.04 Z axis
0.03
0.02
Displacement (m)
0.01
-0.01
-0.02
-0.03
-0.04
-0.05
5 5.5 6 6.5 7
Time (s)
Figure 7-6 Middle accelerometer displacement at T = 0kg and v = 1.21m/s
From Figure 7-6 it can be seen that there is unstable vortex shedding occurring in the X axis.
This is most likely due to 2P modal response. Large reduced velocity of 10.52 results from
the low flexural rigidity and zero tension condition. This results in structural instability
causing non-linear response in the model, as can be seen in Table 7-3 with the
disproportionate increase in member natural frequency with flow velocity.
Cross flow vibration amplitude at the middle accelerometer for the range of reduced
velocities is able to give a good indication of model response behaviour under the various
flow regimes. Figure 7-7 shows the response of each model at the condition of zero tension.

Vibration Amplitude Ratio vs. Reduced Velocity

1.4
CF
IF
1.2
0.8
A/D
0.6
0.4
0.2
0
2 4 6 8 10 12 14 16
Reduced Velocity
Figure 7-7 Cross flow response for each accelerometer at T = 0 N
It can be seen from Figure 7-7 that the highest amplitude occurs in-line with the flow at a
reduced velocity of 15.08. There is a spike of vibration amplitude ratio up to 1.06 at the
reduced velocity of 10.52, which can be accounted for by natural frequencies shown in Table
7-3. Amplitudes in the cross flow direction are shown to reduce after the peak at 10.52, while
in-line flow continues to rise linearly until a reduced velocity of 15.08. This is most likely
due to the significant reduced velocity which is out of the margins of cross flow
characteristics given in Table 2-1. The large reduced velocity and low stiffness property of
the material is causing instability of the member, leading to non-linear results.
Tensioning of 450.9 N can be used to compare results from a testing regime that does not
experience lock-in. Table 7-3 indicates natural frequency of the member is well outside the
range of expected lock-in due to the applied tension. Oscillatory response can be represented
by vibration amplitude response in the cross flow and in-line flow directions. A plot of non-
dimensionalised amplitude at each reduced velocity is given in Figure 7-8.


3.5
CF
IF
3
2.5
2
A/D
1.5
0.5
0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Reduced Velocity
Figure 7-8 Cross flow and in-line flow response of the thin-walled PVC member at T = 490.5 N
Looking at Figure 7-8 it can be observed that there is a more distinct trend in amplitude
response as reduced velocity increases. Cross-flow response can be seen to have a significant
increase as reduced velocity enters the cross flow dominant regime of 4.8 - 12. Cross flow
response dominates the highest reduced velocity of 4.50 with amplitude of almost 1.8 larger
than the in-line response. This correlates with data given in Table 2-1. Model response
amplitude increases as the fluctuating pressure distribution increases, forcing the model to the
area of low pressure. Significant separation is occurring due to oscillatory forces
perpendicular to fluid flow, resulting in vibration amplitude of almost three times the model
diameter.
The second highest reduced velocity of 3.14 produced a more even in-line and cross flow
response at the middle accelerometer. This is consistent with reduced velocity behaviour
presented in Table 2-1, as alternative vortex shedding is causing in-line oscillations between
reduced velocities of 2.2 to 3.5.

Non-dimensional Displacement vs. Reynolds Number

3
2.5
2
A/D
1.5
0.5
0
1 2 3 4 5 6 7 8
Re 4
x 10
Figure 7-9 Effects of Reynolds number on non-dimensional amplitude at T = 490.5 N
Figure 7-9 shows an exponential increase in cross flow response with increased Reynolds
number. Each resulting Reynolds number indicates flow in the fully turbulent vortex street
regime. Model response to Reynolds number is expected to follow a linear trend when there
is no instance of lock-in. The exponential behaviour exhibited is most likely due to instability
of the member due to low Young’s Modulus.
7.5 THICK-WALLED PVC MEMBER

The thick-walled PVC member (PVC 2) was constructed from 4mm wall thickness PVC,
which was the only suitable material available to represent a bending stiffness in between
regular PVC and aluminium. The higher bending stiffness compared to the thin walled PVC
is expected to lead to lower amplitude oscillations and modal shape (Chen et al., 2011).
Table 7-3 indicates the zero tension experimental testing condition spans the range of
potential lock-in. The calculated value for member natural frequency at this tension is 3.56
Hz, only 0.47 Hz less than the vortex shedding frequency at 0.8 m/s. However the response
frequency of 2.27 Hz determined from FFT analysis indicates that no lock-in will take place.

The displacement resulting in the X and Z axes for the flow velocity of 1.7 m/s at zero
tension can be seen below in Figure 7-10.
Displacement at Middle Accelerometer

0.06
X axis
Z axis
0.04
0.02
Displacement (m)
-0.02
-0.04
-0.06
8 8.5 9 9.5 10
Time (s)
Figure 7-10 Middle accelerometer displacement at v = 1.7m/s and T = 0 N
Cross flow is shown to exhibit a higher response to the increased flow velocity than in-line
flow. A 1:1 relationship for the two regimes exists, demonstrating uniform oscillatory
behaviour. To gauge the response over the full range of reduced velocities tested cross flow
has been compared against in-line flow response in Figure 7-11.


1.8
CF
1.6 IF
1.4
1.2
1
A/D
0.8
0.6
0.4
0.2
0
2 4 6 8 10 12 14
Reduced Velocity
Figure 7-11 Cross flow and in-line flow response of the thick-walled PVC member at T = 0N
Results show a dominance of cross flow response across the full range of flow velocities
tested. There is no significant spike in vibration amplitude at any of the reduced velocities,
confirming that no lock-in is present. A linear trend can be observed for both cross flow and
in-line behaviour as reduced velocity is increased, confirming the absence of lock-in. This
suggests that response of the thick-walled PVC member is more reliable than that of the thin-
walled member.
Cross flow non-dimensional amplitudes at each accelerometer can be seen in Figure 7-12 and
Figure 7-13.

Vibration Amplitude at Each Accelerometer

1.8
Bot
1.6 Mid
Top
1.4
1.2
1
A/D
0.8
0.6
0.4
0.2
0
2 4 6 8 10 12 14
Reduced Velocity
Figure 7-12 Cross flow response at each accelerometer for T = 0 N
It can be seen from Figure 7-12 that the largest cross flow vibration amplitude for each
reduced velocity occurs at the middle accelerometer. Significant amplitude was also recorded
at the bottom accelerometer for each flow regime. This could be the result of the modal shape
encountered during testing or most likely due to incorrect accelerometer alignment. Extensive
attempts at accelerometer alignment were attempted in Matlab using the recommended tilt
adjustment method from the accelerometer manufacturer. This method produced unrealistic
results and so smoothed and de-trended accelerometer data was used for analysis.


3
CF
IF
2.5
2
A/D
1.5
0.5
0
0.5 1 1.5 2 2.5 3 3.5 4
Reduced Velocity
Figure 7-13 Cross flow and in-line response at for T = 490.5 N
Response of the thick-walled PVC model shows minimal oscillation in both the X and Z axes
for reduced velocities below 2.7. A significant increase in cross flow vibration amplitude was
observed during the highest reduced velocity as it enters the cross flow oscillation regime of
3.89 as outlined in Table 2-1. Figure 7-13 shows a large the largest in-line response also
occurs at the same reduced velocity as cross-flow. This non-linear trend may also be caused
by member instability as reduced velocity increases due to low Young’s Modulus.
Submerged natural frequency of the member at 490.5 N of tension has been calculated to be
11.17 Hz, which is outside the vortex shedding frequency of 8.55 Hz.
Results from of an FFT of this run show a third harmonic occurring at 10.4 Hz which is
within 6.9% of the natural frequency. This could be responsible for the increased amplitude
ratio at seen in Figure 7-13. As also confirmed in numerical analysis conducted by Tang et al.
(2011), the largest value of hydrodynamic cross-flow force clearly occurs at fn,3.
To assess the effects of Reynolds number on the non-dimensional cross flow response,
vibration amplitude ratios at all four Reynolds numbers have been given in Figure 7-14.


3
2.5
2
A/D
1.5
0.5
0
1 2 3 4 5 6 7
Re 4
x 10
Figure 7-14 shows the increased cross flow response amplitude corresponding with a
Reynolds number of Re = 6.84 x 104. This corresponds to the critical Reynolds regime where
VIV is a key structural concern. Amplitude increases with reduced velocity in an exponential
fashion, as was also the case for the thin-walled PVC member in Figure 7-9. This is most
likely a result of the low member stiffness, or potentially by third harmonic response
frequency.
7.6 ALUMINIUM MEMBER

Decreased vibration amplitude is expected from the aluminium member due to the much
larger bending stiffness in comparison to the PVC models. Significantly increased vortex
response frequencies were recorded across the range of testing conditions. Natural
frequencies in Table 7-3 indicate vortex shedding frequencies calculated in

Table 7-4 are outside the potential range for lock-in. This should result in linear trends of
amplitude ratio with increases in reduced velocity. Cross flow response at flow velocity of
1.2 m/s for this tension can be seen in Figure 7-15.
-3
x 10 Displacement at Middle Accelerometer
4
X axis
Z axis
3
2
Displacement (m)
-1
-2
-3
-4
5 5.2 5.4 5.6 5.8 6
Time (s)
Figure 7-15 Middle accelerometer displacement at v = 1.21m/s and T = 490.5 N
Looking at Figure 7-15, distinct response amplitude in the X axis can be observed.
Fluctuations in Z axis response about the position of zero displacement indicates that
displacement is occurring on both the local and global axes as due to pressure caused by form
drag in line with the flow. The larger cross flow exhibited at the reduced velocity of 2.39 is
2.4 times greater than the highest in-line flow vibration amplitude recorded. This differential
then increases to a magnitude of 5.56 as the flow velocity is increased to 1.7 m/s with a
reduced velocity of 3.46 (refer to Figure 7-17).
The same response frequency in X and Z axes can be observed in Figure 7-15, indicating that
the same vortex response frequency is occurring for each. This can be confirmed by the FFT
plot in Figure 7-16 indicating a single mode for each axis.

FFT Aluminium at T = 490.5 N

0.7
X axis
Z axis
0.6
0.5
0.4
0.3
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18
Frequency (Hz)
Figure 7-16 FFT plot of aluminium member at T = 490.5 N
A comparison of cross flow and in-line flow response can be seen in Figure 7-17.

1
CF
0.9 IF
0.8
0.7
0.6
A/D
0.5
0.4
0.3
0.2
0.1
0
0.5 1 1.5 2 2.5 3 3.5
Reduced Velocity
Figure 7-17 Cross flow and in-line flow response of the aluminium member at T = 490.5 N

Figure 7-17 shows a linear increase in cross flow response of the member. Table 7-3 indicates
a natural frequency of 12.90 Hz, which is well outside the range of vortex shedding
frequency. This is confirmed by the linear response to reduced velocity. The differential
between cross flow and in-line flow is much more significant in this region as the reduced
velocity enters the range of cross flow oscillation region given in Table 2-1.
Experimental data from the highest tensioned testing condition displays the largest response
frequency of the testing members at 13.20 Hz. Cross flow and in-line flow amplitude
response at this tension can be seen below in Figure 7-18.

1.4
CF
IF
1.2
0.8
A/D
0.6
0.4
0.2
0
0.5 1 1.5 2 2.5 3
Reduced Velocity
Figure 7-18 Cross flow and in-line flow response of the aluminium member at T = 981 N
Figure 7-18 shows a similar trend to Figure 7-17 of cross-flow domination across the range of
tested reduced velocities. It can be noted that the in-line vibration amplitude of the highest
reduced velocity has slightly decreased with a proportional increase in cross flow. The lower
reduced velocities display no response to the flow. This is due to the high tension acting on
the model increasing the natural frequency.
Effects of Reynolds number on the cross flow response of the model can be seen in Figure
7-19.


1
0.9
0.8
0.7
0.6
A/D
0.5
0.4
0.3
0.2
0.1
0
1 2 3 4 5 6 7
Re 4
x 10
Unlike the effects of Reynolds number determined from the other two models, results for the
aluminium member show a linear relationship in cross flow vibration amplitude with an
increase in Reynolds number. This confirms that no lock-in is occurring at any of the test
regimes. The exponential response of the two PVC members can be attributed the low
Young’s modulus of both members, causing instability. This is particularly the case at
increased reduced velocities where the apparent exponential response occurs. All Reynolds
numbers in the regimes tested are in the fully turbulent vortex street range outlined in Figure
2-2.

7.7 MODEL COMPARISON

Significant differential in material bending stiffness properties has been shown to cause large
variations in natural frequency and response amplitude behaviour of the materials. The
aluminium member possessed the much larger bending stiffness property of approximately
ten times the bending stiffness of the closest PVC model. Ideally there would be a suitable
material with a mid-range bending stiffness; however this was not achievable with off-the-
shelf pipe materials. Because of this large differential it was difficult correlate results that are
not affected by lock-in. As the testing members were shown to behave erratically when
subjected to varied tension it was decided to hold tension constant and vary reduced velocity
for member comparison.
To enable comparison of each model response at the various flow velocities it is important to
make riser response non-dimensional. Vibration amplitude is a key parameter in comparing
risers, taking into account the 2.91mm differential range in outside diameters resulting from
pipe availability. Maximum amplitude experienced by each model has been used to calculate
the amplitude ratio, as the extreme vibration of the system represents the worst loading on the
system and connections. By taking onto account the reduced velocity of the system a true
non-dimensional comparison can be made allowing data to be used for full scale models.
Correlation of results was attempted by scaling model stiffness and relating reduced velocity
to vibration amplitude response of each model. Because of the large variation in bending
stiffness and instability of models with low Young’s Modulus this proved difficult to achieve.
The range of vortex shedding frequencies resulting from testing also made comparison
difficult as lock-in was regularly observed. Evaluation of tensioning effects can however be
concluded from Table 7-3. Vortex shedding frequency has been shown to increase in a
somewhat proportional manner for the two stiffer models. As tension is applied to each
model the resulting natural frequency creates an increasing differential between vortex
shedding frequencies and member natural frequency. An initial indication of the effects of
bending stiffness on cross flow vibration response of each accelerometer at constant flow
velocity and tension is shown below in Figure 7-20. This gives a good indication of the scale
of bending stiffness affecting the response amplitude of each member.

Vibration Amplitude Ratio vs Bending Stiffness

3
Bot
Mid
Top
2.5
2
A/D
1.5
0.5
0 0.5 1 1.5 2
2 4
Bending Stiffness (Nm ) x 10
Figure 7-20 Cross flow response for each model at v = 1.7 m/s and T = 490.5 N
Figure 7-20 shows the largest vibration amplitude at middle accelerometers for each model.
The decrease in amplitude with the increased bending stiffness gives a good representation of
what is expected from model response when lock-in is avoided. Amplitudes recorded at the
top and bottom accelerometers show discrepancies for each model. Table 7-3 shows all
natural frequencies falling outside the range of reduced velocities.. A trend in cross flow
vibration amplitude reduction can be identified with cross flow and in-line flow factors being
reduced by a magnitude of 3.26 and 10.56 respectively as a result of bending stiffness
increase (refer to Table 7-6). This large decrease occurs in conjunction with faster pressure
fluctuations as the vortex shedding response increases with bending stiffness. As the response
frequency increases so too does the vibration velocity. Oscillation amplitude is therefore
smaller in members of increased natural frequency as vortices are being shed more rapidly.
This results in a change of pressure distribution and direction of oscillation towards the area
of low pressure.
Comparison of cross flow and in-line flow response of each model has been conducted to
identify oscillatory behaviour at each flow regime. This form of analysis also gives an
indication of the modal response of each riser and the vortex shedding behaviour under each

testing condition. Effects of bending stiffness on modal response can be achieved by a

comparison of the oscillatory response of all risers at each accelerometer for a given flow
velocity.

3
CF
IF
2.5
2
A/D
1.5
0.5
0
3.4 3.6 3.8 4 4.2 4.4 4.6 4.8
Reduced Velocity
Figure 7-21 Comparison of cross flow and in-line flow for each member for v = 1.7 and T = 490.5 N
As expected there is a significant increase in cross flow response as reduced velocity

increases. Cross flow vibration amplitude increases by the factor of 3.26 over the range of
reduced velocity of only 1.04, indicating the impact of reduced velocity on oscillatory
response. Reduced velocity and vibration amplitude response for each cross flow and in-line
flow is given below in Table 7-6.
Table 7-6 Response characteristics of each model at v = 1.7 m/s and T = 490.5 N
Model VR A* (CF) A* (IF)

PVC 1 4.50 2.97 1.69
PVC 2 3.89 2.82 1.68
Aluminium 3.46 0.91 0.16
The same approach has been undertaken to identify response characteristics for the highest
flow velocity of 1.7 m/s at the same tension of 981 N. Cross flow response at these conditions
can be seen below in Figure 7-22.


3
Bot
Mid
Top
2.5
2
A/D
1.5
0.5
0 0.5 1 1.5 2
2 4
Figure 7-22 Cross flow response for each model at v = 1.7 m/s and T = 981 N
As with Figure 7-20 there is a distinct trend in vibration amplitude reduction with increasing
values of bending stiffness. The linearity of response at middle accelerometers suggests each
model avoided lock-in. Table 7-3 shows the natural frequencies of each member at these
testing conditions are well outside the vortex shedding frequencies as a result of the applied
tension. This indicates the importance of tensioning risers sufficiently to place the natural
frequency outside the range of vortex shedding frequencies, thus avoiding lock-in. Tension
may be applied to full scale risers via top tensioning or self-weight in deep water
environments. Altering the natural frequency of risers is particularly important in deep water
applications where large free span increases susceptibility and response to VIV.
Model response in the cross flow and in-line directions for the same testing conditions is
shown below in Figure 7-23.


3
CF
IF
2.5
2
A/D
1.5
0.5
0
2.6 2.8 3 3.2 3.4 3.6 3.8 4 4.2
Reduced Velocity
Figure 7-23 Comparison of cross flow and in-line flow for each member at v = 1.7 m/s and T = 981 N
A close response can be seen from the two PVC members can be seen in Figure 7-23.
Increased cross flow response is proportional to the decrease in-line response over the
reduced velocity differential of 0.15. This indicates linear behaviour for both models, which
was not the case for lower tension and fluid velocity conditions. Cross flow and in-line
behavioural response at each reduced velocity is given in Table 7-7.
Table 7-7 Response characteristics of each model at v = 1.7 m/s and T = 981 N
Model VR A* (CF) A* (IF)

PVC 1 4.11 2.61 0.97
PVC 2 3.96 2.53 1.66
Aluminium 2.73 1.01 0.34

For a better understanding of cross flow response due to flow velocity, tension has been held
constant at 490.5 N and flow velocity varied. Vibration amplitude at each model stiffness be
examined in more detail in Figure 7-24 to Figure 7-26Error! Reference source not found..

0.045
Bot
0.04 Mid
Top
0.035
0.03
0.025
A/D
0.02
0.015
0.01
0.005
0
0 0.5 1 1.5 2
2 4
Figure 7-24 compares the effects of bending stiffness on cross flow oscillation at each
accelerometer location for lowest flow velocity of 1.4 m/s. As expected there is increased
amplitude response for the lowest bending stiffness member, however there is no correlation
of response behaviour with the increased bending stiffness of the other two models. Bottom
accelerometers on all three members produced the highest amplitude, however it is important
to note the very small scale of vibration amplitude. Similar response behaviour can be seen
for the thick-walled PVC and aluminium members, suggesting that neither member was
significantly effected by fluid flow.
Figure 7-25 compares vibration response under at the flow velocity of 0.8 m/s.


0.4
Bot
Mid
0.35 Top
0.3
0.25
A/D
0.2
0.15
0.1
0.05
0 0.5 1 1.5 2
2 4
Middle accelerometer vibration amplitude of the aluminium model is shown to be over two
times larger than at the thin-walled PVC member. This result is unexpected and can be
attributed to instability of the PVC models at low flow velocity. As previously mentioned,
vortex shedding frequency of the aluminium member is well outside the range of vortex
shedding frequency as confirmed by Figure 7-17. This discounts lock-in as the cause for the
increased amplitude ratio response. The aluminium member has typically shown steady
response behaviour, which is verified for this run in Figure 7-17. Results discrepancies can
therefore be attributed to the two PVC members, which have been shown to give unreliable
results when subject to low flow velocities. This is a result of non-linear pressure distribution
between bending stiffness values at low response frequencies.
Figure 7-26 shows the cross flow vibration amplitude ratio for the flow velocity of 1.2 m/s at
the same tension.


1.4
Bot
1.3 Mid
Top
1.2
1.1
1
A/D
0.9
0.8
0.7
0.6
0.5
0.4
0 0.5 1 1.5 2
2 4
As with Figure 7-25, Figure 7-26 shows unexpected response from the PVC members. No
lock-in response if expected from any of the models under these testing conditions as outlined
in each member’s individual analysis. Response of each member is shown to be most
significant at the middle accelerometer, indicating an odd number of modes. FFT analysis
conducted on the aluminium member confirms the odd modal shape by indicating a single
mode in the X and Z axes. The same analysis conducted on the thin-walled and thick-walled
PVC members also confirmed odd modal shapes of three and five modes respectively.
7.8 FULL SCALE IMPLICATIONS

Non-linear vibration amplitude response of members not affected by lock-in has proven
difficult to predict under the varied tensions, flow velocities and stiffness. This type of
response unpredictability caused difficulty in data correlation between members. It was
initially intended to scale member stiffness and compare non-dimensional results to assess the
effects of bending stiffness scaling on the experimental models. Comparisons were then to be
made with commercially available risers using non-dimensional results for a Mid-Water Arch
system at a known water depth. Due to the unstable responses of the PVC members it was

decided that any results from stiffness scaling analysis would be erroneous and would not
contribute to the study of VIV.
There are, however, several conclusions that can be made from the model analyses in relation
to full-scale risers. As no lock-in was expected or noted for either PVC member a linear
increase in response amplitude was expected as Reynolds number was increased. All flow
Reynolds numbers calculated during testing placed the flow regime in the fully turbulent
vortex street. This indicates that vortex shedding behaviour should not change as a result of
the variation in fluid velocities tested. The exponential response of PVC members to
increases in Reynolds number highlights the susceptibility of non-rigid pipelines to VIV due
to current velocity, even in instances of no lock-in. As knowledge of VIV increases it is
commonplace for checks to be carried out on instances of resonance. Non-dimensional
response amplitude of the thin-walled PVC member of almost three times the model diameter
indicates that highly detrimental response can occur even when there is no lock-in present.
Testing has also shown the significant effects of tension on avoiding lock-in. Increases in
natural frequency over 200% enable both PVC members to avoid vortex shedding frequency
and any possibility of lock-in. This highlights the effects of tensioning and self-weight on
offshore risers for VIV ramifications. However, as mentioned above significant vibrations
can still result under tension.
Increases in reduced velocity have also been show to result in varying response behaviour of
the models. Increases in reduced velocity above values of approximately 3.0 have shown to
have notable impacts on the in-line and cross flow response of the models, particularly in the
higher stiffness aluminium model. As reduced velocity is increased in the critical range the
cross flow response continues to increase proportionally, while the in-line response decreases
in the same proportion. This behaviour is consistent with reduced velocity theory presented in
Section 2.6 and can be applied to full scale model response predictions.

8. CONCLUSIONS
This experimental investigation has presented the effects of stiffness on model response to
VIV. Variations in system tension and flow velocity have proven to have a large effect on
model response behaviour over the flow velocities and tensions tested. The following
conclusions can be drawn from the investigation:
 Response amplitude tended to significantly decrease in models with higher stiffness
properties while oscillation frequency increased. Higher stiffness placed put the
aluminium member outside the range of potential lock-in.
 Model in-line and cross-flow response has been shown to be largely effected under
small variations flow regime. Reduced velocities exceeding 3.0 tended to
proportionally decrease in-line response and increase cross flow response
 Post-processing code written using Matlab has proven effective in analysing a large
quantity of data efficiently with minimal user input. A large variety of comparisons
can be made by results generated from one set of data.
 The filtering function used in Matlab proved to be successful in eliminating noise
pollution. Steady curves were produced, resulting in an effective idealisation of model
response.
 PVC models with low Young’s Modulus have been shown to produce more stable
results at higher flow velocities and tensions.
 Lock-in range has been substantially reduced by member tensioning. No lock-in was
encountered during testing due primarily to natural frequency increases resulting from
tensioning.
 Natural frequency was able be determined from the decay test and calculated for the
submerged condition using added mass. These results were consistent with calculated
values, confirming their validity.

9. FUTURE RECOMMENDATIONS
The following list of recommendations is intended to assist future studies on VIV at the AMC
Flume Tank.
 Experimental analysis was originally meant to be conducted in tandem with a
numerical simulation. This would be ideal in future studies to relate testing results to
commercially available software.
 Check accelerometer orientation during testing. This proved to be difficult to correct
and future studies will benefit from ensuring accurate alignment for modal analysis.
 Conduct testing with fixed connections on PVC models to increase natural frequency.
 In depth analysis of VIV suppression methods. PVC members were shown to exhibit
very large response behaviour but unpredictable at low tensions.
 More in-depth analysis on relationship between in-line and cross flow. Investigate
amplitude ratios at various reduced velocities.
 Setup system to conduct decay test in the same environment as model testing to
achieve completely accurate natural frequency values.
 Configure a system to determine the exact response of the Flume Tank carriage under
vibration.

REFERENCES
Beardman, P. (2009). Understanding and predicting vortex-induced vibrations. Journal of
Fluid Mechanics, 1.
Blackburm, H., & Henderson, R. (1996). Lock-in Behaviour in Simulated Vortex-Induced
Vibrations. Expeimental Thermal and Fluid Science, 4.
Chakrabarti, S. K. (2002). Advanced Series on Ocean Engineering. The Theory and Practice
of Hydrodynamics and Vibration, 392.
Chakrabarti, S. K. (2005). Handbook of Offshore Engineering. Illinois: Offshore Structure
Analysis, Inc.
Chen W., L. M. (2011). Dynamic characteristics and VIV of deepwater riser with axially
varying structural properties. Beijing: Elsevier Ltd.
Darrigol, O. (2002). Stability and Instability in Nineteenth-Century Fluid Mechanics. Revue
d'histoire des mathematiques.
DNV. (2006). RP-F105. Free Spanning Pipelines.
DNV. (2006). RP-F204. Riser Fatigue.
DNV. (2010). RP-C205. Environmental Conditions and Environmental Loads.
Espeyrac L., P. S. (2002). MFV. Retrieved 04 18, 2012, from
http://hmf.enseeiht.fr/travaux/CD0102/travaux/optmfn/gpfmho/01-
02/grp1/phy_know.htm
Laidlaw, A. (2010). Pipeline Vortex Induced Vibration Under Free Span Conditions.
Bachelor of Engineering Thesis, 17, 30.
Lie H., K. K. (2006). Modal analysis of measurements from a large-scale VIV model test of a
riser in linearly sheared flow. Trondheim: Elsevier Ltd.
Lienhard, J. H. (1966). Synopsis of Lift, Drag, and Vortex Frequency Data for Rigid Circular
Cylinders. Bulletin 300.

Michael, J. B. (2011). Experimental Design for testing Vortex Induced Vibration on
Cylindrical Members. Bachelor of Engineering Thesis, 5, 11-15, 35.
Ojeda, R. (2011). Vibration. Free vibration and damping, 3-6, 24, 41.
Palmer, A., & King, R. A. (2006). Subsea Pipeline Engineering. Oklahoma USA: PenWell.
Ranmuthugala, D. (2003). Fluid Mechanics. Launceston: Australian Maritime College.
Sarpkaya, T. (2001). A critical review of the intrinsic nature of vortex-induced vibrations.
Journal of Fluids and Structures, 2,3.
Techet, A. H. (2005). Vortex Induced Vibrations, 16.

APPENDIX A
Microsoft Excel spreadsheet used for natural frequency calculations.
Input Output
Cylinder OD = 42.66 mm Pipe mass (empty) = 0.50 kg/m
Wall thickness = 2.00 mm Pipe mass (full) = 1.67 kg/m
Young's modulus = 3378550753.00 Pa Added Mass = 1.43 kg/m
Pipe density = 1530.00 kg/m3 Total pipe mass = 3.10 kg/m
End constraint = 0.50 Flexural rigidity = 1426.96 Nm2
Pipe length = 2.00 m Natural frequency (dry) = 6.69 Hz
Fluid density = 1000.00 kg/m3 Natural frequency (wet) = 2.68 Hz
Applied tension (kg) = 100.00 kg Tensioned natural frequency (dry) = 13.24 Hz
Applied tension (N) = 981.00 N Tensioned natural frequency (wet) = 9.23 Hz

APPENDIX B
0.7
CF
IF
0.6
0.5
0.4
A/D
0.3
0.2
0.1
0
0 200 400 600 800 1000
Tension (N)
Figure 9-1 Effects of tension on the thin wall PVC model at constant v = 0.8 m/s

0.7
CF
IF
0.6
0.5
0.4
A/D
0.3
0.2
0.1
0
0 200 400 600 800 1000
Tension (N)
Figure 9-2 Effects of tension on the thick wall PVC model at constant v = 0.8 m/s


0.4
CF
0.35 IF
0.3
0.25
A/D
0.2
0.15
0.1
0.05
0
0 200 400 600 800 1000
Tension (N)
Figure 9-3 Effects of tension on the aluminium model at constant v = 0.8 m/s

1
CF
IF
0.9
0.8
0.7
A/D
0.6
0.5
0.4
0 200 400 600 800 1000

Tension (N)


1.3
CF
1.2 IF
1.1
0.9
A/D
0.8
0.7
0.6
0.5
0.4
0 200 400 600 800 1000
Tension (N)

1
CF
0.9 IF
0.8
0.7
A/D
0.6
0.5
0.4
0.3
0.2
0 200 400 600 800 1000
Tension (N)


3
CF
IF
2.5
2
A/D
1.5
0.5
0
0 200 400 600 800 1000
Tension (N)

3
CF
2.8 IF
2.6
2.4
2.2
A/D
1.8
1.6
1.4
0 200 400 600 800 1000

Tension (N)


2
CF
1.8 IF
1.6
1.4
1.2
A/D
0.8
0.6
0.4
0.2
0
0 200 400 600 800 1000
Tension (N)

1.6
Bot
1.4 Mid
Top
1.2
1
A/D
0.8
0.6
0.4
0.2
0
2 4 6 8 10 12 14 16
Reduced Velocity
Figure 9-10 Effects of reduced velocity on the thin wall PVC model at T = 0 N


3
Bot
Mid
2.5 Top
2
A/D
1.5
0.5
0
0.5 1 1.5 2 2.5 3 3.5 4 4.5
Reduced Velocity
Figure 9-11 Effects of reduced velocity on the thin wall PVC model at T = 490.5 N

3
Bot
Mid
2.5 Top
2
A/D
1.5
0.5
0
0.5 1 1.5 2 2.5 3 3.5 4 4.5
Reduced Velocity
Figure 9-12 Effects of reduced velocity on the thin wall PVC model at T = 981 N


1.8
Bot
1.6 Mid
Top
1.4
1.2
1
A/D
0.8
0.6
0.4
0.2
0
2 4 6 8 10 12 14
Reduced Velocity
Figure 9-13 Effects of reduced velocity on the thick wall PVC model at T = 0 N

3
Bot
Mid
2.5 Top
2
A/D
1.5
0.5
0
0.5 1 1.5 2 2.5 3 3.5 4
Reduced Velocity
Figure 9-14 Effects of reduced velocity on the thick wall PVC model at T = 490.5 N


3
Bot
Mid
2.5 Top
2
A/D
1.5
0.5
0
0.5 1 1.5 2 2.5 3 3.5 4
Reduced Velocity
Figure 9-15 Effects of reduced velocity on the thick wall model at T = 981 N

2
Bot
1.8 Mid
Top
1.6
1.4
1.2
A/D
0.8
0.6
0.4
0.2
0
0.5 1 1.5 2 2.5 3 3.5 4 4.5
Reduced Velocity
Figure 9-16 Effects of reduced velocity on the aluminium model at T = 0 N


1
Bot
0.9 Mid
Top
0.8
0.7
0.6
A/D
0.5
0.4
0.3
0.2
0.1
0
0.5 1 1.5 2 2.5 3 3.5
Reduced Velocity
Figure 9-17 Effects of reduced velocity on the aluminium model at T = 490.5 N

1.4
Bot
1.2 Mid
Top
0.8
A/D
0.6
0.4
0.2
0
0.5 1 1.5 2 2.5 3
Reduced Velocity
Figure 9-18 Effects of reduced velocity on the aluminium model at T = 981N

Adam Freeman - 071229 - Thesis PDF

Uploaded by

Copyright:

Available Formats

Adam Freeman - 071229 - Thesis PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Adam Freeman - 071229 - Thesis PDF

Uploaded by

Copyright:

Available Formats

Bachelor of Engineering Thesis

Experimental Testing of the Effects of Stiffness

26th October 2011

Supervisor: Mr Rowan Frost

Thesis submitted in partial fulfilment of the requirements for the degree

Bachelor of Engineering (JEE 418)

National Centre for Maritime Engineering and Hydrodynamics

Adam Freeman Page i`

6.3 Member Tensioning ....................................................................................... 32

Adam Freeman Page iii`

Adam Freeman Page iv`

Figure 7-10 Middle accelerometer displacement at v = 1.7m/s and T = 0 N .......................... 48

Adam Freeman Page vi`

Adam Freeman Page vii`

Adam Freeman Page viii`

Profile View Plan View

Adam Freeman Page ix`

Adam Freeman Page x`

TERMS, ABBREVIATIONS AND DEFINITIONS

Adam Freeman Page xi`

Adam Freeman Page xii`

Adam Freeman Page 1`

At certain excitation frequencies, periodically alternating fluid separation initiates vortex

1.2 AIM OF RESEARCH PROJECT

Adam Freeman Page 2`

Figure 2-1 Simplified representation of vibration model (Ojeda, 2011)

Where: fn = Natural frequency of the system (Hz)

Adam Freeman Page 3`

Vibration of a system can take two forms:

2.2 REYNOLDS NUMBER

Adam Freeman Page 4`

Where: Re = Reynolds number

Adam Freeman Page 5`

2.3 STROUHAL NUMBER

Where: St = Strouhal number

Cylinders undergoing vortex shedding exhibit hydrodynamic instability, or Strouhal

Adam Freeman Page 6`

Strouhal number as a function of Reynolds is shown in Figure 2-3.

Figure 2-3 Strouhal number as a function of Reynolds number (Techet, 2005)

2.4 VORTEX SHEDDING FREQUENCY

Where: fs = Vortex shedding frequency of cylinder

Adam Freeman Page 7`

Figure 2-4 Vortex shedding patterns of a circular cylinder (Sarpkaya, 2004)

2.5 ADDED MASS

m '   c 'V (5)

Where: m’ = Added mass

V = CSA of model (m2)

2.6 REDUCED VELOCITY

Where: VR = Reduced velocity

Range Response Characteristic

Reduced velocity is a non-dimensional parameter which allows models of varying system

Adam Freeman Page 9`

Structural aspect ratio l

Adam Freeman Page 10`

2.7 FREE SPAN RESPONSE

L/D Response Description

2.8 BENDING STIFFNESS

Where: F = Flexural rigidity (Nm2)

Adam Freeman Page 11`