DEGREE PROJECT IN THE BUILT ENVIRONMENT,

SECOND CYCLE, 30 CREDITS

STOCKHOLM, SWEDEN 2017

Offshore Floating Platforms

Analysis of a solution for motion mitigation

ALBERTO RODRÍGUEZ MARIJUÁN

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ARCHITECTURE AND THE BUILT ENVIRONMENT
 Offshore Floating Platforms
Analysis of a solution for motion mitigation

Author: Alberto Rodrı́guez Marijuán

KTH Supervisor: Prof. Raid Karoumi, PhD.

External Supervisor: David Carrascosa, Saitec O.T.

[Link]. Thesis in Structural Engineering and Bridges AF223X

Spring 2017
KTH Royal Institute of Technology
TRITA-BKN. Master Thesis 526
ISSN 1103-4297
ISRN KTH/BKN/EX-526-SE
 Introduction

Offshore wind is one of the most promising renewable energy resource and
it is meant to be the baseline of the future 100% renewable energy-mix. The
only limitation of current bottom-fixed foundations is the depth that makes
offshore windfarms feasible only in shallow waters such as North Sea in Europe.
Upcoming Floating Offshore Wind Solutions removes that limitation enabling
Offshore Wind to be a Global Market.
Back in 2011, Saitec started looking into potential opportunities in the Off-
shore Wind Field being convinced about the company’s capabilities to provide
great technology solutions. The market analysis performed at the time con-
cluded with a clear picture of cost reduction potentials in the Offshore Wind
Market.
With the main focus on reducing the Levelized Cost Of Energy (LCOE),
Saitec’s team developed the SATH Floating Offshore Wind technology. SATH is
a disruptive solution in terms of costs and installation simplicity when compared
with the usual Oil & Gas style concepts. A big part of this success is due to
the different point of view brought by the Civil Engineering background of the
team.
Saitec is currently involved in several projects both pre-commercial and com-
mercial without overlooking the commitment to Research and Development ac-
tivities that enable to continuously improve the technology and services pro-
vided. Proof of this is the excellent study and analysis performed by Alberto
for his Master’s Thesis regarding the hydrodynamic and structural performance
of the SATH platform.

David Carrascosa Francis

External Supervisor
CTO in Saitec Offshore Technologies.
 Offshore Floating Platforms
Analysis of a solution for motion mitigation
Alberto Rodrı́guez Marijuán

Kungliga Tekniska Högskolan, Stockholm

arod@[Link]

Recent events regarding energy policies throughout the globe and advances
in technology are making offshore wind farms become a reality. Most offshore
wind farms are still, however, built close to land masses, and need to be rigidly
attached to the seabed in one way or another. In many countries, both public
and private entities are developing new concepts of floating platforms to over-
come the thirty to thirty-five-metre depth limit. Some of these new platforms
use and adapt previous Oil and Gas platform concepts, while others are built
up from scratch.
This Master Thesis covers a hydrodynamic and structural analysis of a new
concrete floating platform concept developed for medium to deep waters. This
work is based on data from experimental model-scale tests performed in a wave
tank and from numerical models using linear potential theory, limited here only
to regular wave trains. The study focused on the behavior of the heave plates
attached to the platform: test data was analyzed in order to find indicators of the
largest dynamic pressures on the plates when only motion data was available,
and the structural behavior of the plates was studied under different static
pressure distributions using a commercial Finite Element Method software.
The results from these analyses show that the normal accelerations of the
plates -assumed rigid- strongly correlate with the dynamic pressures measured;
and that the general structural behavior of the plate, in terms of deformations
and bending moments, is well captured when the hydrodynamic load distribu-
tion is simplified into a uniformly distributed load of the same magnitude.
The results obtained will help reduce the computational effort currently
needed in the design of these floating structures, especially at some stages,
when numerous scenarios, load cases and combinations need to be studied.

Keywords: Offshore Wind, floating platform, wave tank test, hydrodynamic

pressure, structural, hydrodynamic load, heave plate.
 Acknowledgments

I would like to thank Professor Raid Karoumi for his support

and close advice before and during the making of this Thesis.

To all my colleagues in Saitec who helped me all along, especially

to Araceli Martı́nez for all the support and for taking the time
and effort to answer to all questions I always had; and to
Alex Morató, for his continuous advice and guidance.

To Javier Garmendia, who encouraged me to come

to Bilbao to write this Thesis, and with whom
I resolved to go study in Stockholm, which will
always make me grateful and proud.

And to my parents, who helped me become who I am today,

thank you.
Table of Contents

1 Introduction 1
1.1 Aim and Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Content and Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 State of the Art 3

2.1 The Offshore Wind Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Bottom-Fixed Offshore Wind Turbines . . . . . . . . . . . . . . . . . . . . . . . . 4
2.3 Floating Platforms for Offshore Wind . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3.1 SATH platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3 Theory Background 10
3.1 Wave Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.1.1 Regular Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.1.2 Irregular Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 Floating Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2.1 Motions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2.2 Added Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2.3 Damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2.4 Restoring Spring Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2.5 Wave Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2.6 Wind Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2.7 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2.8 Natural Periods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

4 Method 23
4.1 Workflow summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2 Definition of motions and DOFs . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.3 Geometry and Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.3.1 Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.3.2 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.4 Hydrodynamic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.4.1 Numerical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.4.2 Wave Tank Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.5 Plate loads analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.6 Structural Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

I
 4.6.1 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.7 Load Case Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

5 Results 41
5.1 Damping, Natural Periods and RAOs . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.3 Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.4 Structural response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.4.1 Quality assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.4.2 Displacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.4.3 Plate forces and moments . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

6 Conclusions, Discussion and Future Studies 57

6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.2 Discussion and limitations of the analyses. . . . . . . . . . . . . . . . . . . . . . . 57
6.3 Recommendation for future studies. . . . . . . . . . . . . . . . . . . . . . . . . . 58

Appendices 62

Appendix A. Convergence Analysis 64

Appendix B. VBA scripts 67

Appendix C. Python scripts. 70

Appendix D. Definition of the Numerical Models. 86

Appendix E. Results of FE Model. 87

II
List of Abbreviations

• BC: Boundary Conditions.

• CAD: Computer Aided Design.

• CDM: Central Difference Method.

• COG: global center of gravity.

• CPU: Central Processing Unit, main core of a computer.

• DNV-GL: Det Norske Veritas and Germanischer Lloyd

• DOF: Degree of Freedom

• EC2: Eurocode 2.

• FE: Finite Element

• FFT: Fast Fourier Transform.

• GBF: Gravity-Based Foundations

• GM: metacentric height.

• GWEC: Global Wind Energy Council

• IHCantabria: Institute of Hydraulics in Cantabria.

• JONSWAP: Joint North Sea Wave Project

• LC: Load Case

• MDOF: Multiple Degree Of Freedom

• O&G: Oil and Gas

• P-M: Pierson-Moskowitz

• PMMA: poly(methyl methacrylate).

• R&D: Research and Development

• RAO: Response Amplitude Operator.

III
IV

• Rd: dynamic deformation response factor.

• SATH: Swinging Around Twin Hull

• SOT: Saitec Offshore Technologies

• SPM: Single Point Mooring

• TLP: Tension-Leg Platform

• VBA: Visual Basic for Applications.

List of Symbols

Matrices and Vectors

[A]: added mass matrix.

[B]: damping matrix.

[C]: restoring matrix.

[M ]: generalized mass matrix.

{η̈}: body acceleration vector, time derivative of the body speed.

{η̇}: body speed vector, time derivative of the body displacement.

{η}: body displacement vector.

{F }: external excitation forces vector.

Greek
δ: logarithmic decrement.

∆ω: constant difference between successive frequencies.

δz : vertical deformation of the plate.

i : random phase angle for ith component.

ηi , η̇i , η̈i : motion, velocity and acceleration for ith DOF.

λ: wavelength.

λA : area scale factor

λL : length scale factor

∇2 : Laplace operator.

ν: Poisson’s ratio.

V
VI

ω: circular frequency.

ωn : natural circular frequency.

φ: velocity potential function.

ρ: density.

σ, Y : intermediate variables for JONSWAP spectrum.

ζ: damping ratio.

ζa : wave amplitude.

Latin
apl : plate accelerations.

Ai : wave elevation for ith component.

Arotor : rotor swept area.

Bcr : critical damping facto.

CD : drag coefficient.

CM : inertia coefficient.

CT : thrust coefficient.

D: diameter.

E: Young’s Modulus.

Fxx , Fyy : plate axial force in the x and y directions.

FT : thrust force.

g: acceleration of gravity.

h: depth to sea floor.

Hs , H1/3 : significant wave height.

Ix , Iy , Iz : moments of inertia of the rigid body platform in the 3 axes.

k: wave number.

Mxx , Myy : plate bending moment causing bending stress perpendicular to the x and y
directions.

m0 : zeroth spectral moment.

S: spectrum function.
T : wave period.

T1 : statistical mean period.

Tn : natural period.

Tp : peak wave period.

u: particle velocity in x-direction.

v: particle velocity in y-direction.

vpl : plate velocities.

Vw : wind velocity.

w: particle velocity in z-direction.

zhub : height of the hub.

zref : reference height.

V : velocity vector.

VII
List of Figures

2.1 European Offshore Wind Atlas [4] . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.2 Cumulative installed capacity for offshore wind [5] . . . . . . . . . . . . . . . . . 5
2.3 Videnby wind farm. Image: Risø . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.4 Common bottom-fixed substructures for offshore wind turbines. [8] . . . . . . . . 6
2.5 Typical Floating Platform Concepts [11] . . . . . . . . . . . . . . . . . . . . . . . 6
2.6 Approximate location of the bimep site. Source: Google Earth. . . . . . . . . . . 8
2.7 Sketch of the geometry of SATH . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.1 Comparison of the JONSWAP and the Modified Two-Parameter P-M spectrum
(or Bretschneider spectrum) [15] . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 Hydrodynamic problem split in two subproblems [15] . . . . . . . . . . . . . . . . 14
3.3 Example of motion decay for various damping levels and no external forces. . . . 16
3.4 Simple SDOF oscillator with no damping and its motion with time. . . . . . . . 17
3.5 Righting moment and heeling moment curves [24] . . . . . . . . . . . . . . . . . . 19
3.6 Deformation response as a function of excitation force frequency . . . . . . . . . 20
3.7 Wave drift forces obtained from a time record [15] . . . . . . . . . . . . . . . . . 21

4.1 Summary flowchart of the method. . . . . . . . . . . . . . . . . . . . . . . . . . . 24

4.2 SATH geometry, with the coordinate system considered and the 6 DOFs for rigid
body motion, η1 through η6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.3 Actual geometry of the heave plates . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.4 Heave plates, as simplified in the modelled geometry in GeniE. . . . . . . . . . . 26
4.5 Half the simplified panel model defined in HydroD, (a) before and (b) after meshing. 28
4.6 Mesh discretization (a) before and (b) after manual refinement. . . . . . . . . . . 29
4.7 Solution for the scale model heave plates. . . . . . . . . . . . . . . . . . . . . . . 31
4.8 Scale model geometry, (a) theoretical and (b) built. . . . . . . . . . . . . . . . . . 32
4.9 Sketch of the platform in (a) upright position and (b) with a trim angle. . . . . . 34
4.10 FFT of the motion signal and its associated velocity and acceleration spectra,
unfiltered (left) and filtered (right) . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.11 Time series of the motion signal and its associated velocity and acceleration,
unfiltered (left) and filtered (right) . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.12 Example of normalized signals: (a) unfiltered and (b) filtered. . . . . . . . . . . . 35
4.13 Slab sections perpendicular to its in-plane directions. . . . . . . . . . . . . . . . . 37
4.14 Boundary conditions and plate-to-frame rigid links showing constrained DOFs. . 38

VIII
LIST OF FIGURES IX

4.15 Approximate shape of the pressures distribution input in the structural FE model. 40
4.16 Approximate shape of the pressures distribution obtained in the hydromechanical
models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

5.1 Estimation of damping by the traditional method using a free decay signal. Image
produced using a Python script developed by Alexandre Morató. . . . . . . . . . 41
5.2 Motion RAOs obtained in HydroD. . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.3 Motion RAOs obtained from the experimental data. . . . . . . . . . . . . . . . . 42
5.4 Pitch stability curve as computed numerically in Sesam. . . . . . . . . . . . . . . 43
5.5 Pressures on the plates with varying wave periods (experimental). . . . . . . . . 44
5.6 Frequency-wise velocities and accelerations on the bow plate (experimental). . . 45
5.7 Correlation plot of the net pressure on the heave plate with the plate (a) velocities
and (b) accelerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.8 Evaluation of velocities and accelerations for bottom pressure peaks. . . . . . . . 48
5.9 Slab sketch and virtual beam within the slab. . . . . . . . . . . . . . . . . . . . . 48
5.10 Equivalent indeterminate beam with spring. . . . . . . . . . . . . . . . . . . . . . 49
5.11 Cutting lines definition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.12 Vertical deflection contour plot for LC1. . . . . . . . . . . . . . . . . . . . . . . . 51
5.14 Bending moments Mxx along lineX for all load cases. . . . . . . . . . . . . . . . . 52
5.15 Bending moments Myy close to the support (lineY) and further away from it
(lineY2), for some comparable cases. . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.13 Plate bending moments (a) Mxx and (b) Myy in LC1. . . . . . . . . . . . . . . . 54
5.16 Plots comparing the response of the plate under uniform distributed load and
under the load distribution obtained in the hydromechanic model. Cases with
clamped BC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.17 Plots comparing the response of the plate under uniform distributed load and
under the load distribution obtained in the hydromechanic model. Cases with
pinned BC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

A.1 Natural periods obtained for different mesh densities. . . . . . . . . . . . . . . . . 64

A.2 RAO in pitch obtained for different mesh densities. . . . . . . . . . . . . . . . . . 65
A.3 RAO in roll obtained for different mesh densities. . . . . . . . . . . . . . . . . . . 66
A.4 RAO in heave obtained for different mesh densities. . . . . . . . . . . . . . . . . 66
E.1 Vertical deflection line cutting diagram along lineY2, for all load cases. . . . . . . 88
E.2 Bending moments Myy along lineY for all load cases. . . . . . . . . . . . . . . . . 89
E.3 Bending moments Myy along lineY2 for all load cases. . . . . . . . . . . . . . . . 90
List of Tables

4.1 Summary of the characterization of the mass model of SATH for HydroD. . . . . 30
4.2 Global characteristics of scale models. . . . . . . . . . . . . . . . . . . . . . . . . 32
4.3 Directional characteristics of the modelled slab. . . . . . . . . . . . . . . . . . . . 37
4.4 Materials defined in the structural FEM . . . . . . . . . . . . . . . . . . . . . . . 38
4.5 Load Case definition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

5.1 FE model Quality Assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

A1 Natural periods obtained for different mesh densities. . . . . . . . . . . . . . . . . 65

D1 Summary of equivalent tubular sections for the HydroD Morison model. . . . . . 86
D2 Plate thickness defined in the structural FEM. . . . . . . . . . . . . . . . . . . . 86

X
1. Introduction

In the last decades, many countries have been shifting their energy consumption away from fossil
fuels, including oil, natural gas and coal, toward renewable sources, especially solar and wind
energy. Many studies suggest that, even with current technology, a high degree of independence
of carbon-based fuels is feasible within the next 30 years in certain countries, such as Spain
[1], Germany or Sweden, thanks in part to a reduction in the total energy consumption and a
change in production, relying on renewable energy sources.
This vision is however not devoid of technical concerns, and requires policies that ease the
initial capital investment in the research and development phases of these new technologies.
Many countries, especially in northern Europe (Sweden has been the first country to agree on a
100% horizon on renewable energy generation [2]) and East Asia (China is an example of heavy
investment in new technology [3]), have been challenging traditional energy policy regarding
carbon-based fuels and renewable energy generation. Current policies now rely on estimated
energy pools in which wind energy is an important source; and within it, offshore wind, many
agree, has a huge potential for it solves many of the concerns with the traditional (onshore)
version of wind technology.

1.1 Aim and Motivation

The different design stages and aspects of the structures used in the offshore wind industry
often require many simulations, generating large amounts of data that need to be processed.
This Master’s Thesis aims at analyzing a floating platform for a wind turbine using several
modelling techniques and detailing the steps to transfer data between them, as well as a more
specific study on the behavior of an auxiliary part of this structure: the heave plates.
Studying the structural behavior of the heave plates might allow for certain simplifications
in what is now a tedious and computer intensive work. For instance, a correlation between
different variables that are usually not obtainable simultaneously (platform motions and local
hydrodynamic pressures) might help reduce the computer workload significantly, especially for
early stages of design.

1.2 Content and Scope

The Master’s Thesis presented in this document is structured as follows:

• State of the Art presents a brief overview of the offshore wind industry and current
solutions for both bottom-fixed and floating solutions.

• Theory background covers the basis behind most of the calculations in this document,
including wind and wave theory, ship stability and structural dynamics.

1
2 CHAPTER 1. INTRODUCTION

• In the Method chapter, the work flow followed is described in detail, and all assumptions,
models, limitations and desired outputs are discussed.

• Results presents all important outputs obtained through calculations and simulations run
in the making of this Thesis, as well as explanations of the specific assumptions and
limitations.

• Finally, the outcome is discussed and recommendations for further work given in the final
chapter: Conclusions, Discussion and Future Studies.

• Bibliography follows, as well as several Appendices, deemed necessary to provide further

insight on the work performed in this Master Thesis.

The work done and presented in this document covers an analysis of the hydrodynamic
behavior of a new concept of floating platform using state-of-the-art modelling and simulation
tools, and a linear-elastic structural analysis of a sensitive part of it: the bow and stern heave
plates, used to ensure a better response in heave and pitch motion.
This Master Thesis aims at developing a simple way of modelling the structural behavior of
a lightened voided concrete slab -as in the heave plate in a floating platform-, in order to easily
transfer complex sets of loads obtained in the hydrodynamic analyses. Besides, effort has been
put in analyzing experimental data, aimed at finding indicators of the highest hydrodynamic
pressures on the heave plates using only variables that are directly obtained in the hydrodynamic
analyses (platform motions and wave elevations), in order to reduce the computational resources
needed to establish an appropriate stress requirement for design.
The entire floating platform was studied as a rigid body with hydrodynamic models in
Sesam, with the mooring system simplified and its main restraining effects neglected. Thus,
results in surge, sway and yaw degrees of freedom -those most affected by the mooring system
in this configuration- are disregarded in this Master Thesis.
The results from these numerical analyses were compared with the results obtained in the
tank tests carried out as part of the design process by SOT, owner and developer of the SATH
concept solution, during a campaign during summer 2017. The experimental series obtained in
this test phase were used to find the sought correlation between the pressures (measured in the
tank test) and the accelerations of the plates.
The response of the platform was analyzed, its rigid-body eigenfrequencies identified, and
the hydrodynamic loads caused by waves on the heave plates were analyzed structurally using
the finite element method implemented in a commercial software.

A note from the author

Along this document and in the Appendices, the reader may find charts, tables and graphs
incomplete or missing one of their axes. These aspects were omitted because they may contain
sensitive and confidential information which belongs to Saitec Offshore Technologies, as owner
of the platform concept.
The author acknowledges the inconvenience, but considers that there is sufficient data so
that conclusions and results can be interpreted qualitatively, with no need to disclose such
information.
2. State of the Art

2.1 The Offshore Wind Industry

The concept of offshore wind originated as a way to minimize some of the issues traditional
onshore wind faces: the visual impact of the turbines relatively close to protected natural or
populated areas, the environmental impact of transportation of structures and materials and
the extensive use of land that could otherwise be used for other activities.
New technology behind offshore wind production tackles most of these problems by installing
the wind turbines in open water, namely seas and oceans, in or relatively close to the coastline
(although, in the case of floating offshore structures, this is not necessary). However, this in-
volves complex engineering processes that increase the costs of installation, design, maintenance
and decommissioning.
The reasoning behind offshore wind energy is that, thanks to a more stable and abundant
wind resource on the open seas, the increased initial capital costs can be balanced by a steadier
energy inflow and an efficient and low cost production can be achieved. Besides, the logistic
limitation when transporting the pieces of the turbine (blades, tower, nacelle and rotor) is
eliminated since this transport is no longer done through land with trailers, but with large
cargo ships instead.
Indeed, studies show that wind resource in offshore locations adds up to multiple times its
onshore analogous (figure 2.1), especially since improvements in floating structures allow for
ever greater depths of operation and the availability of usable space in open waters that seas
and oceans offer is immense.
Every year, the Global Wind Energy Council (GWEC) releases their market update on the
state of wind energy, taking into account installation of new wind farms, cumulative capacity
and R&D in technology.
The latest report by GWEC [5] shows yet a larger increase in total cumulative wind capacity
in the global picture if compared with the previous years, with China and the US as the two
largest wind production markets. As stated in their 2015 report, ”in 2015, increase in wind
generation was equal to almost half of global electricity growth” [6], which provides a measure
of the importance of the wind energy production these days.
More specifically for offshore wind, as can be seen in figure 2.2, the total installed offshore
wind capacity has grown steadily in the last decade, both in East Asia and in Europe (especially
UK and Germany). In 2015, offshore wind accounted for about a quarter of the total new
installed capacity, and investment in offshore wind doubled [6].
Regarding all engineering processes and technology, the offshore wind industry is a direct
descendant of the Oil and Gas (O&G) industry and, as such, recycles and reuses many of its
practices and standards. That is why in Europe, as of 2017, the most common standards and
rules are those given by Det Norske Veritas (DNV), initially thought for offshore drilling and
storage stations and adapted now for fixed and floating structures for wind energy production.

3
4 CHAPTER 2. STATE OF THE ART

Figure 2.1: European Offshore Wind Atlas [4]

2.2 Bottom-Fixed Offshore Wind Turbines

Most of the offshore wind energy production currently relies on bottom-fixed wind turbines:
this technology is considered mature in its own right. The first offshore wind farm was built in
Vindeby (Denmark) (figure 2.3), in 1991. Since then it has produced a total of 243 GWh (the
farm included eleven 0.45 MW turbines -current offshore wind turbines lie closer to 5 MW of
power-) [7] and is currently under decommissioning.
In Europe, almost all offshore turbines use bottom-fixed technology, being the monopile
solution the most common one, followed by gravity-based foundations and space frames (figure
2.4). The most common types of bottom-fixed foundations are [8]:

1. Monopile: usually a cylindrical steel tube that connects either directly or through a
transition piece the tower of the wind turbine with the seabed. The tube is driven into
the ground to provide the needed stiffness and resistance against met-ocean phenomena.
These are usually installed in shallow waters (< 20-30 m), since their cost increases
dramatically with depth.

2. Gravity-Based Foundations (GBF): relies on the weight of a large concrete mass (some-
times encased in steel), ballasted with sand, iron ore or crushed rock to give it the proper
stability. Optimally used in shallow depths (< 30 m).

3. Tripod: a relatively lighter version of the lattice or jacket structure. A steel frame transfers
the load from the central cylindrical shaft to the three steel piles acting as legs.
2.3. FLOATING PLATFORMS FOR OFFSHORE WIND 5

Figure 2.2: Cumulative installed capacity for offshore wind [5]

Figure 2.3: Videnby wind farm. Image: Risø

4. Jacket: a classical lattice solution for the supporting structure, with piles driven into the
seabed to provide stability.

5. Tripile: a three-legged jacket structure, similar to the tripod. The legs are connected to
the central shaft and driven into the seabed. These solutions based on the jacket or lattice
structure are used in so-called transitional waters, i.e. 20 to 60 m deep.

2.3 Floating Platforms for Offshore Wind

An important issue concerning offshore wind resource is that most of it is located at waters
relatively far from the coastline, corresponding with depths greater than thirty or forty meters.
In places where the seabed is so deep, the aforementioned bottom-fixed structures are not
feasible anymore, neither economically nor technically.
For these locations, in transitional to deep waters (> 30 m), the concept of floating plat-
forms for the turbine support was again adapted from the O&G industry. Floating platforms
6 CHAPTER 2. STATE OF THE ART

Figure 2.4: Common bottom-fixed substructures for offshore wind turbines. [8]

Figure 2.5: Typical Floating Platform Concepts [11]

basically consist of a main supporting structure stabilized and restrained by several means,
such as a mooring system (catenary, taut lines or vertical tendons are the main types; detailed
explanations of these can be found in the literature [8],[9],[10]). A preliminary classification is
indeed made for floating platforms as to how they achieve stability, as shown in figure 2.5:

1. Spar-buoy: or ballast-stabilized buoy, consists of a large draft cylinder-like structure that

is ballasted in its lowest part, increasing stability by moving the center of mass lower
than the center of buoyancy. They are usually restrained by catenary moorings. One of
the first large-scale offshore wind turbines, Hywind (by Statoil), that was installed in the
North Sea off the coast of Norway in September 2009 [12] uses this platform solution.

2. Tension-Leg-Platform (TLP): uses a combination of tensioned-cable moorings and positive

-excess- buoyancy to keep stability. These cables or tendons are fixed to the seabed by
suction caissons, which require a complex installation in large depths. This system is very
2.3. FLOATING PLATFORMS FOR OFFSHORE WIND 7

stable in the heave motion, and allows for a lighter structure that those of the semi-sub,
but it also increases the stresses on the mooring system and the operational risks if a
cable fails.

3. Semi-submersible (semi-sub or buoyancy-stabilized): achieve stability by distributed buoy-

ancy, and is kept in place by sets of catenary moorings anchored to the seabed by dead-
weights or drag anchors. They usually require a large mass of the superstructure to
maintain stability. Their low draft, however, makes them a very flexible solution in terms
of location and cost of installation and production.

The majority of these are medium-sized structures and are currently built out of steel, but
concrete is being thoughtfully considered thanks to advances in prestressing techniques, material
improvements and its inherent resistance to fatigue and durability under marine conditions.
Indeed, the platform analyzed in the case study later in this document is made of prestressed
concrete.

2.3.1 SATH platform

As previously stated, this Master’s Thesis covers an analysis of the solution adopted in the
SATH (Swinging Around Twin Hull) floating platform developed by Saitec Offshore Technolo-
gies (SOT) [13] against excessive pitch motions. The SATH structure is an example of a
semisubmersible concrete platform designed for offshore wind, with some advantages regard-
ing other current commercial or yet-non-commercial solutions, including a complete onshorized
construction, the use of high-durability prestressed concrete that minimizes maintenance and
repair costs, and a good scalability both in terms of turbine size and type and mass production
for large wind farms.
The first project in which SATH technology is to be involved is DemoSATH, which consists
on the installation of a full-scale prototype of the SATH platform with a 2 MW turbine. It is to
be installed in the bimep 1 site on the Cantabrian Sea, about 2 km off the north coast of Biscay
(figure 2.6). The construction and installation is scheduled to be finished in summer 2018 and
the prototype to be in operation by fall of that same year.

Geometry

The base geometry of the SATH concept is based on a catamaran ship, with a semi-submerged
twin hull to improve stability. This basic configuration was later adapted for the special condi-
tions of an offshore floating platform (figure 2.7), adding beams to connect the two hulls, the
support structure for the tower and the plates in the bow and stern part of the platform in
order to reduce pitch (and heave) motions (in dark blue, figure 2.7), which are further analyzed
later in this document. The hulls were stiffened with several diaphragms or bulkheads, as a
way of redistributing stresses and compartmentalizing the hulls.
However, perhaps the most relevant feature of SATH (the platform indeed owes part of its
name to it) is the Single-Point-Mooring (SPM ), which consists on a device that connects all
1 bimep is a open sea site run by the Basque Government, ”designed for testing and demonstrating prototype

devices for harnessing ocean energy in terms of their safety and economic and technical viability prior to their
full-scale commercial development.” [14]
8 CHAPTER 2. STATE OF THE ART

Figure 2.6: Approximate location of the bimep site. Source: Google Earth.

mooring lines and the main electric cable to the same point in space, longitudinally offset from
the platform bow (support structure for the SPM in bright green and SPM in yellow in 2.7).
The SPM allows for rotational motions and, since the mooring is attached to it, the platform
swings around the SPM when hit by waves or wind gusts in any directions other than directly
facing the platform, allowing for an improved energy production (the turbine can be aligned to
face the wind for more hours a year than only with the usual yaw control traditional onshore
turbines have) and reducing transverse loading on the main body of the platform.
One of the most challenging issues within the engineering design of such structure is to
perform a fully-coupled analysis, including all hydrodynamic motions from waves and currents in
the platform, plus the aero-servo-elastic analysis of the tower-nacelle-rotor system, considering
dynamic loading and responses from all members, and transfer of vibrations. In the case of a
floating platform of the aforementioned characteristics, the control system of the wind turbine
has to be carefully designed and tested in order to accommodate motions and accelerations of
the SPM-platform-turbine set while ensuring safety and the optimal energy production.
Results from simplified coupled analyses are presented later in this document. These, unless
stated otherwise, have not been run by the author of this Thesis but by colleagues in SOT and
their data used in later simulations.

Mooring System
The mooring system in the SATH concept is a combination of the classic catenary mooring and
a synthetic fibre taut mooring. This allows for an important reduction in the weight of line
suspended from the SPM connection compared to a complete steel chain catenary. Nevertheless,
some pieces of steel catenary are used at the ends of the lines, giving some advantages over
other systems. The piece of chain on top allows for a steeper angle at the SPM connection
compared with that of a simple taut line, allowing it to move without the concrete structure
blocking its way. The chains at the bottom provide enough weight to ensure that the mooring
2.3. FLOATING PLATFORMS FOR OFFSHORE WIND 9

(a) Render in perspective. (b) Top and side view.

Figure 2.7: Sketch of the geometry of SATH

forces only pull the anchor horizontally, yielding optimal anchor behavior.
Some moorings are designed with dead-weights in or along the lines, increasing the restoring
forces (also called station-keeping forces by some authors). Others use some extra weights on
the lines to ensure the point of contact with the seabed is set, and never gets too close to the
anchor, or a combination of some or all of these.
The mooring system attached to SATH will depend on the location to be installed. It could
consist of catenary lines (chains or wire rope) or taut synthetic lines combined with different
types of anchors, optimized for each soil material and installation depth.
A more detailed description of the mooring is deemed unimportant for the purpose of this
Thesis, since its effect has been simplified or neglected altogether in the analyses performed,
and thus need not be further explained here.
3. Theory Background

3.1 Wave Theory

Most of the theory presented in this section is based on the work done by O.M. Faltinsen in
Sea loads on ships and offshore structures [9], by J.M.J. Journée and W.W. Massie in Offshore
Hydromechanics [15], and the excellent review of these and other works presented by Eirik Wie
Furunes in his Master Thesis [16] for NTNU.
This section covers the theory and assumptions that the author deemed as necessary for the
easy understanding of the work presented in this Master Thesis.

3.1.1 Regular Waves

The basic assumption to some usual wave theories, including linear wave theory, also known
as Airy wave theory1 , is that seawater is considered incompressible and inviscid, and the
fluid motion is irrotational. Then, a velocity potential φ that describes the velocity vector
V(x, y, z, t)=(u, v, w) such that:

∂φ ∂φ ∂φ
∇φ ≡ i +j +k (3.1)
∂x ∂y ∂z
where i, j and k are unit vectors in the x, y and z directions, has to satisfy the Laplace
equation:

∂2φ ∂2φ ∂2φ

∇2 φ = + 2 + 2 =0 (3.2)
∂x2 ∂y ∂z
Equation 3.2 is valid, as mentioned earlier, for inviscid and incompressible fluid, and irrota-
tional flow. The concept of velocity potential has no physical meaning itself, but was introduced
because it was mathematically convenient. The relation between the potential function and the
fluid particle velocity can be expressed then as:

∂φ
=u (3.3a)
∂x
∂φ
=v (3.3b)
∂y
∂φ
=w (3.3c)
∂z
Where u is the velocity in the x-direction, v in the y-direction and w in the z-direction.
By formulating a kinematic2 and a dynamic3 boundary condition (their mathematical forms
1 First published by Sir George Biddell Airy in the XIX Century
2A fluid particle will follow the free surface ζa (x, t) at all times
3 The pressure at the surface is equal to the atmospheric pressure p , and constant
0

10
3.1. WAVE THEORY 11

are not presented here; the reader may refer to [9] for their complete derivation), when the
velocity potential φ oscillates harmonically in time with a circular frequency ω, we may state
the free surface condition as:

∂φ
− ω2 φ + g =0 on z = 0 (3.4)
∂z
Where g is the acceleration of gravity.
By assuming horizontal sea bottom and a free-surface of infinite extent, follows the derivation
of linear wave theory for propagating waves in finite depths. If we apply the sea bottom
condition:

∂φ
w= = 0 on z = −h (3.5)
∂z
Where z = 0 for the mean water level and h is the mean vertical depth to the sea bottom.
Equation 3.5 can be understood as that no flow can go through the seabed.
The derivation of linear propagating waves can be done for a finite or infinite depth, by
applying different boundary conditions. Using the sea bottom condition (3.5) and the free
surface condition (3.4), one can derive an expression for the velocity potential for finite water
depth:

g ζa cosh(k(z + h))
φ= cos(ω t − k x) for finite z (3.6)
ω cosh(k h)
On the other hand, by assuming no fluid disturbance for z → −∞, the velocity potential φ
for infinite water depth can be expressed as:

g ζa k z
φ= e cos(ω t − k x) for z → −∞ (3.7)
ω
Where, for both equations:

• ζa is the wave amplitude

• h is the water depth

• ω is the angular frequency of the wave ( 2π

T , being T the wave period)

• k is the wave number ( 2π

λ , with λ being the wavelength)

From the free-surface condition 3.4, one can derive the well-known dispersion relation:

ω2
= ktanh(kh) (3.8)
g

3.1.2 Irregular Waves

Since the linear wave theory is based upon linearization, it is possible to superimpose a set of
regular waves to simulate an irregular sea, whose wave elevation can be expressed as:
N
X
ζa (t) = Aj sin(ωj t − kj x + j ) (3.9)
j=1
12 CHAPTER 3. THEORY BACKGROUND

Figure 3.1: Comparison of the JONSWAP and the Modified Two-Parameter P-M spectrum (or
Bretschneider spectrum) [15]

Where Aj is the wave amplitude, ωj the circular wave frequency, kj the wave number and
j the random phase angle (uniformly distributed between 0 and 2π) of a wave component j.
Equation 3.9 represents a random sea surface. This implies that it is considered a stochastic
field that can be represented by a wave spectrum S(ωj ), where:

1 2
A = S(ωj ) ∆ω (3.10)
2 j
Where ∆ω is a constant difference between successive frequencies.
This wave spectrum is a representation of the sea state in the frequency domain, and can
be estimated from wave measurements. Some quite commonly used spectra are the Pierson-
Moskowitz (provides an accurate representation of fully-developed sea states), likely in its modi-
fied two-parameter form (also known as Bretschneider spectrum) and the Joint North Sea Wave
Project (JONSWAP) (used for sea states with limited fetch).
Figure 3.1 shows a comparison of how these two spectra represent sea states.
The JONSWAP spectrum may be written as:
2
H1/3
 
−944
S(ω) = 155 exp (3.3)Y (m2 s) (3.11)
T14 ω 4 T14 ω 4
With Y :
 2 !
0.191 ω T1 − 1
Y = exp − √ (3.12)
2σ
And σ:

σ = 0.07 if ω ≤ 5.24/T1
(3.13)
= 0.09 if ω > 5.24/T1
3.2. FLOATING STRUCTURE 13

Where H1/3 is the significant wave height, defined as the average height of the one third
largest waves. For narrow-banded spectra and Gaussian sea state elevations, it can be defined as
√ R∞
H1/3 ≈ 4 m0 , where m0 is the zeroth (0th ) spectral moment (being mk = 0 ω k S(ω)dω). T1 is
a mean wave period, defined as T1 = 2πm0 /m1 , and related to the peak period by Tp = 1.198 T1
[17].

3.2 Floating Structure

3.2.1 Motions
The equations describing the rigid body motions of a multiple degree of freedom (MDOF)
system can be expressed as:

([M ] + [A]) {η̈} + [B] {η̇} + [C] {η} = {F (t, η)} (3.14)
Where:
• [M ] represents the generalized mass matrix
• [A] is the added mass matrix
• [B] is the damping matrix
• [C] is the restoring matrix
• {η}, {η̇} and {η̈} are the displacement, velocity and acceleration vectors, respectively,
containing the 3 translatory and the 3 rotational DOFs (see section 4.2).
• {F (x, η)} is the external excitation force vector.
It is usually appropriate to describe the motions of a floating structure by analyzing its
response under regular waves, since linearity allows for a superposition of several regular wave
components to simulate an irregular sea state.
By assuming no initial transient effects, the linear dynamic motions and loads oscillate
harmonically with the same frequency as the incident waves. This hydrodynamic problem is
usually split in two subproblems (figure 3.2):
A. Wave excitation loads (restrained in waves): mainly diffraction and Froude-Krylov forces4 .
This problem represents a restrained structure with incident waves acting on it.
B. Oscillation in still water: there are no incident waves, and the structure is forced to
oscillate at the wave excitation frequency in any given mode. The loads acting on the
structure contain the added mass, damping and restoring terms.
If only linear response is considered, a common way of describing the motions of a floating
structure are the so-called transfer functions or Response Amplitude Operators (RAOs), that
describe the ratio of the vessel motion amplitude5 to incident wave amplitude ζηa as a function
of ω.
4 Named after William Froude and Alexei Krylov, these forces are produced by the unsteady pressure field

generated by undisturbed waves

5 There exist RAOs for velocities and accelerations, too, although they are not used in this analysis. The

interested reader may refer to [18] for more information on RAOs.

14 CHAPTER 3. THEORY BACKGROUND

Figure 3.2: Hydrodynamic problem split in two subproblems [15]

3.2.2 Added Mass

The concept of added mass is a very important one in the field of hydromechanics. It describes a
hydromechanical reaction force that is proportional to the acceleration of the body in the fluid,
and is caused by accelerations given to the water particles close to it (not, as it is sometimes
misunderstood, as a finite amount of water that oscillates rigidly connected to the body), that
cause hydrodynamic pressure variations. In such terms, the concept of added mass should be
F orce
understood as Acceleration .
The added mass is usually expressed in a dimensionless form, as the ratio of added mass to
displaced fluid mass. For a 6-DOF-structure, there is a total of 36 added mass coefficients:
 
A11 A12 A13 A14 A15 A16
A
 12 A22 A23 A24 A25 A26 

A A32 A33 A34 A35 A36 
 
[A] =  13 (3.15)
A14 A42 A43 A44 A45 A46 

 
A15 A52 A53 A54 A55 A56 
A16 A62 A63 A64 A65 A66

Where the sub-indexes correspond to the 6 DOFs of rigid body motion (see section 4.2), so
that A11 is the added mass coefficient for the surge-surge motion, A33 is for heave-heave, and
so on.
It can be shown, however, that for a structure with zero speed and with no current present,
the added mass matrix is symmetric, so only 21 of the coefficients in 3.15 are needed to com-
pletely define this phenomenon.
It is important to notice that the force given by [A]{η̈} does not dissipate any energy (by
integrating the work done separately by the added mass, the damping and the spring forces
separately in a purely harmonic oscillation such that η = ηa · sin(ωt), it can be shown that
damping is the only force that dissipates energy (see [15], p. 6-13. for the complete derivation),
so that, by integrating the work done by the added mass term over one oscillation period:
3.2. FLOATING STRUCTURE 15

Z T
1
{(m + a) η̈} {η̇ dt} = 0 (3.16)
T 0

This term is quite difficult to compute for complex geometries, like the one concerning this
Master Thesis, although it can be estimated for simpler shapes such as slender vertical cylinders
or spheres. In practice, they are most commonly computed numerically with specific software
or empirically through testing.

3.2.3 Damping
Most of the theory presented here is based on the book Dynamics of Structures: Theory and
Application to Earthquake Engineering [19], by Anil K. Chopra, and Offshore Hydromechanics,
by J.M.J. Journée and W.W. Massie [15].
As stated previously, damping is the force component causing energy dissipation and, thus,
is responsible for the decay of motion with time (figure 3.3). It corresponds to the term [B]{η̇}
in equation 3.14 and is a function of the velocity η̇.
Mathematically, we should distinguish between two types of damping: classical and non-
classical. Their mathematical forms are thoroughly explained in the literature (the reader may
refer to [19] and [20] for more information). The damping in SATH is currently being treated
as classical damping, meaning that the damping matrix [B] is diagonal (only the terms in the
main diagonal are non-zero):

B11

 B22 
[B] =  (3.17)
 
.. 
 . 
B66
Usually damping is a quite complex phenomenon to analyze theoretically, and thus is com-
monly studied empirically in tank tests or other lab setups. This is indeed the case in SATH,
where the damping matrix has been obtained through decay tests on a water tank in the Insti-
tute of Hydraulics in Cantabria (IHCantabria) in previous occasions, and is part of the Phase
I tests goals (see section 4.4.2).
Although in classic dynamics the damping is taken as linear with the velocity, a more
complete definition of a damping force may include a nonlinear term, such that, for a given
DOF i:

Bi η̇i = B1 η̇i + B2 η̇i |η̇i | (3.18)

Where B1 is the linear damping, B2 is the quadratic damping, and η̇ is the motion in the
given DOF (either translatory or rotational).
By splitting the damping into these two terms, the equation of motion can be rewritten for
any DOF and no external forces (free vibration) as:

(M + A)η̈ + B1 η̇ + B2 η̇|η̇| + Cη = 0 (3.19)

However, in order to operate with a more manageable equation, the total damping can be
linearized by stipulating that the energy dissipated by both the linear and non-linear terms be
16 CHAPTER 3. THEORY BACKGROUND

Figure 3.3: Example of motion decay for various damping levels and no external forces. ωn =
2.5 rad/s

equal to the energy dissipated by an equivalent linear damping term Beq (see [15], p. 6.15):

8
Beq = B1 + ω ηa B2 (3.20)
3π
Such that:

(M + A)η̈ + Beq η̇ + C η = 0 (3.21)

From equation 3.20 is clear that the equivalent damping is proportional to the amplitude of
motion ηa and its circular frequency ω.
Damping in structures is usually represented as the ratio ζ of damping B to critical damping
Bcr :

B
ζ= (3.22)
Bcr
Where:
√ 2k
Bcr = 2 k m = = 2 m ωn (3.23)
ωn
Being ωn the natural frequency of the structure, k its elastic spring coefficient and m its
mass (both dry and added mass accounted).
Critical damping refers to the minimum value of B that causes a structure in free decay to
return to its equilibrium position without any oscillation (see ζ = 1 in figure 3.3).
In order to compute a damping coefficient from a time-domain signal, the logarithmic decre-
3.2. FLOATING STRUCTURE 17

ment method can be used, as described in [19]. This method uses the oscillation peaks identified
in the time series and computes the logarithmic decrement δ as:

ui
δ = ln ' 2πζ (3.24)
ui+1
Where ui and ui+1 are the amplitude of oscillation at the peaks happening at time ti and
ti+1 , respectively (see figure 3.3).

3.2.4 Restoring Spring Forces

These forces can be taken as the equivalent to the stiffness in usual structures. They correspond
to the [C]{η} term (proportional to the relative displacement of the structure) in equation 3.14
and, unlike the damping forces, they do not dissipate any energy. A basic simplification for an
ideal single DOF system without damping is a weight hanging from a spring with stiffness k
and no external forces acting on it, then equation 3.14 can be simplified as:

m η̈ + k η = 0 (3.25)

And the motion of the weight follows a sinusoidal function η = ηa sin(ωn t), as shown in
figure 3.4.

Figure 3.4: Simple SDOF oscillator with no damping and its motion with time.

The mooring system is the main contributor to the restoring forces. There are several types
of mooring systems, and each makes use of different forces and effects. The mooring system
for TLPs, for instance, rely on the tension in the cables plus the extra buoyancy due to the
platform being pulled down to keep the station in place. Spread moorings, on the other hand,
rely on the weight of the cables or chains and a long piece of chain resting on the seafloor to
produce the necessary forces for station-keeping.
Since the mooring system is not to be included in the analyses performed in this Master
Thesis, neither is the mathematical basis for their calculations presented here. The interested
reader may refer to [9] and [18] for more information on mooring modelling.
In section 2.3.1, the specific mooring system of the SATH platform is defined and briefly
explained.
18 CHAPTER 3. THEORY BACKGROUND

3.2.5 Wave Forces

Wave forces are of great importance in the analysis of any floating or fixed offshore solution.
Various approaches can be taken in order to estimate the wave loads on structures, including
theories as 2D or 3D potential theory or Morison’s equation.
Potential theory has already been mentioned earlier in this chapter. Morison’s equation,
on the other hand, is a semi-empirical formulation that yields directly the forces acting on a
slender structure submerged in moving water. It combines two main sources of loads in such
structures: inertia loads (proportional to a coefficient CM ) and drag forces (proportional to
CD ). Equation 3.26 contains the expression for the forces on a strip of height dz of a vertical
cylinder of diameter D, when it is not fixed but moving with velocity η̇1 and acceleration η̈1 .
The reader may refer to [9] for a more thorough discussion on Morison’s equation.

1 πD2 πD2
dF = CD D dz (u − η̇1 )|u − η̇1 | + ρ CM dz ü − ρ (CM − 1) dz η̈1 (3.26)
2 4 4
The first right-hand term in 3.26 contains the drag forces and the second and third terms the
inertial forces. ρ is the water density, u and ü are the water particle velocity and acceleration
in the x-direction, respectively. The | bracket indicates absolute value, and the expression
(u − η̇1 )|u − η̇1 | is the mathematical way of indicating a squared variable, keeping the sign
unchanged. Integration over the total height z of the cylinder would then yield the total wave
excitation force on the body.

3.2.6 Wind Forces

The description and simulation of the complex aerodynamic effects on a wind turbine are out
of the scope of this Thesis. However, some manuals, such as the DNV Standards [21], propose
a simplified method for estimating the thrust force on the turbine rotor based on the one-
dimensional momentum theory, as:

1
FT = ρair CT Arotor Vw2 (3.27)
2
Where FT is the thrust force, ρair is the density of air, CT is the thrust coefficient (in [21],
1.1.11, figure 10-1), Arotor is the rotor swept area (Arotor = π R2 , where R is the rotor radius)
and Vw is the wind speed at the hub height.
The wind speed should, however, be defined at the center of the rotor, which means that
the measured values (taken at a reference height, zref ), are to be transformed into a specific
hub height, zhub . A simple and usual way of doing this transformation [22] is through a power
law, as in equation 3.28:
 α
zhub
Vw (zhub ) = Vw (zref ) (3.28)
zref
Where α, or wind shear exponent, is usually taken as 0.2 for onshore locations [23] and 0.14
for normal wind conditions in offshore wind farms [22].
For the purpose of this Thesis, the effect of wind has been modelled as a static thrust force
acting at the top of the tower, accounting for the wind speed at hub height, using the wind
3.2. FLOATING STRUCTURE 19

profile shown in equation 3.28.

3.2.7 Stability
Most of the content presented in this section is based on the DNV Standards for ship stability,
[24].
The stability of ships and ship-like structures depends on their geometry, displacement6 ∆,
and mass distribution within the structure. We may define the following key factors in any
stability analysis:

• Center of gravity (G) or center of mass, defined as the single point in space where, if a
force is applied to, causes only translations and no rotations in the body.

• Center of buoyancy (B), equivalent to the center of mass of the displaced volume of fluid.

• Metacenter (M), the point where the vertical of a heeled ship intersects the original vertical
of the upright ship (by definition, M is always on the same vertical as B).

• Metacentric height (GM) is the distance between G and M. It influences the natural
period in roll and pitch motions (GMT influences roll and GML has to do with pitch,
where the sub-indexes T and L stand for transverse and longitudinal, respectively).

In general terms, stability is achieved when any heeling moment caused by an external force
can be countered by a righting moment, given by the shape and mass distribution of the ship,
and buoyant forces.

Figure 3.5: Righting moment and heeling moment curves [24]

Generally, the heeling moment may be caused by wind acting on the exposed freeboard
of the ship or, in cases like a floating structure for offshore wind, by the wind thrust on the
6 In the naval industry, displacement of a ship or vessel refers to the mass of fluid it displaces, following

Archimedes’ principle.
20 CHAPTER 3. THEORY BACKGROUND

Figure 3.6: Deformation response as a function of excitation force frequency

turbine. Since this heeling moment is proportional to the area exposed, it decreases as the ship
heels (for simple sections, it is sometimes simplified as directly proportional to the square of
the cosine of the heel angle).
The righting moment is, however, given by the distance between the vertical lines passing
through the center of mass and the center of buoyancy, and so tends to increase for small heel
angles (see righting moment curve in figure 3.5).
The righting and heeling moment curves are usually given as in figure 3.5, and one of the
safety criteria stated in the DNV Standards for intact stability-as opposed to damage stability-
[24] is that, up to a heeling angle given by the second intercept or the down-flooding angle,
whichever is smaller, the area under the righting moment curve must be at least 40% larger
than the one under the heeling moment curve (that is, a factor of safety of 1.4).

3.2.8 Natural Periods

Natural or resonant periods are crucial in the design of any floating structure, since the ampli-
tudes of motion greatly increase at excitation frequencies close to or at the natural frequency
ωn (see figure 3.6).
The natural periods in surge, sway and yaw for a typical moored structure are of the order
of minutes. That means that they are long relative to the wave energy periods, but it should be
noted that some non-linear effects might excite the structure as such long periods. This may
happen for effects such as low-frequency wave drift, a sub-harmonic response of the floating
structure due to the energy carried by the slowly varying water surface, related to the long-
period waves associated to wave groups (figure 3.7).
3.2. FLOATING STRUCTURE 21

Figure 3.7: Wave drift forces obtained from a time record [15]

This low-frequency drift usually determines the response of the floating structure in surge
and sway, and thus is used to define and design the mooring system, which is not covered in
this work.
For the case of pitch, roll and heave, however, the natural periods are of the order of seconds
(although it varies with the type of floating system), which corresponds to some high-frequency
wave energy zone. The added complexity in sea-floating structures is that the excitation forces
can not usually be considered to follow a deterministic pattern, but instead should be regarded
as a stochastic phenomenon. It is common practice to avoid greater issues with platform
resonance in waves by keeping the natural periods in heave, pitch and roll above 20 s, knowing
that most of the wave spectral energy is concentrated below that threshold.
The natural period in heave for a semi-submersible platform can be estimated in closed-form
as:
  21
M + A33
Tn3 = 2 π (3.29)
ρ g Aw
Being, for pitch:
 2
 12
M r55 + A55
Tn5 = 2 π (3.30)
ρ g V GML
And roll:
 2
 12
M r44 + A44
Tn4 = 2 π (3.31)
ρ g V GMT
Where

• Tn the natural period, with the sub-indexes indicating DOF

• M being the corresponding component in the mass matrix for the DOF

• A the added mass coefficient for the DOF

• ρ the density of sea water (usually taken as 1.025 kg/m3 )

• r the radius of gyration through the corresponding axis passing through the center of
gravity.
22 CHAPTER 3. THEORY BACKGROUND

• GM are the longitudinal (GML ) and transverse (GMT ) metacentric heights, for pitch
and roll, respectively.

The response of the structure is greatly influenced by damping in the vicinity of the resonance
frequency (see figure 3.6). However, when the excitation-to-resonant frequency ratios are smaller
than 0.25 [25] (i.e. a slowly varying force), the problem is highly dependant on the stiffness of
the system, and the response may usually be regarded as pseudo-static. On the other hand,
when the excitation force has a very high frequency (ω/ωn > 1.5 in figure 3.6), the response is
mostly unaffected by damping, but highly influenced by the modal mass of the system.
4. Method

4.1 Workflow summary

The flowchart shown in figure 4.1 represents a brief summary of the flow among the different
models and software that have been used in the work involved in this Master Thesis.
All steps are explained in detail in this chapter.

4.2 Definition of motions and DOFs

The geometry of the platform can be considered, to a certain extent within the hydrodynamic
analysis, a rigid body and, thus, will experience a combination of motions in the degrees of
freedom (DOF) shown in figure 4.2. Below, they are listed with their respective names, the
way they will be referred to as in the rest of this document:

• Surge: δx = η1

• Sway: δy = η2

• Heave: δz = η3

• Roll: θx = η4

• Pitch: θy = η5

• Yaw: θz = η6

Where δi denotes translation and θi rotation, along and around the i axis, respectively.
Throughout this document, the notation ηi will be used for motions in their respective ith
DOF, along with their usual standard names, as indicated in the list above.
The reference axes x, y and z are shown in figure 4.2, and the reference origin (0, 0, 0) is set
to the point where all 3 symmetry planes of the cylindrical floaters intersect.
The geometry of the SATH platform, along with the mooring configuration and elasticity
and the SPM, implies that some DOFs are to be more problematic when hit by waves and
currents than others, and that some DOFs are sensitive to variations in certain configurations
-mooring, turbine, etc.- while others are not so. This will be covered later, when the platform
behavior is analyzed. Indeed, the mooring system, greatly simplified in most analyses in this
Master Thesis, mainly affects and restrains surge, sway and yaw (even though the case of yaw
is a special one in the SATH technology with SPM) and, thus, these 3 DOFs are not covered
in the hydrodynamic analyses.

23
24 CHAPTER 4. METHOD

Figure 4.1: Summary flowchart of the method. *Motion refers to displacements, velocities and
accelerations. **Load transfer contains pressures on the panels in the hydromechanic model.
4.3. GEOMETRY AND PROBLEM DEFINITION 25

Figure 4.2: SATH geometry, with the coordinate system considered and the 6 DOFs for rigid
body motion, η1 through η6

4.3 Geometry and Problem Definition

In this Master Thesis, the behavior of the heave plates in the SATH platform has been analyzed
under a set of metocean conditions (Hs , Tp and Vw ), and tank tests studied in order to identify
worst load conditions and the response of the plates using any of the ouputs from the simulation
(wave elevations in the stochastic sea state, motions and accelerations in any of the DOFs, etc.).
In this section, a brief description of the geometry is presented. The analysis procedure
follows, including hydrodynamic analyses of the whole platform with HydroD software and a
structural finite element (FE) model in MidasCivil to represent the mechanical behavior of the
heave plates.

4.3.1 Geometry
The geometry of the plates at the platform bow and stern consists of a base slab made of
concrete and a set of hollow cores to make the plate lighter (see figure 4.3). The main function
of the plates is to increase the moment of inertia of the platform, especially for pitch and heave,
where SATH is expected to concentrate most of its motions, and its added mass and modal
damping for several DOFs. They also lower the global center of gravity of the system, improving
stability. They are made of light-aggregate prestressed concrete and attached to the main body
by a prestressed concrete beam structure.
The voids in the concrete slab are expected to remain watertight (thus filled with air), which
implies a negligible apparent density. All hydrodynamic models covered in this document are
26 CHAPTER 4. METHOD

(a) Bow. (b) Stern.

Figure 4.3: Actual geometry of the heave plates

(a) Bow. (b) Stern.

Figure 4.4: Heave plates, as simplified in the modelled geometry in GeniE.

the result of assuming a solid slab (figure 4.4) with an equivalent density, obtained as in equation
4.1:
 
Vconcrete
ρeq = ρconcrete (4.1)
Vconcrete + Vvoids

4.3.2 Problem Definition

The problem to be solved in this Master Thesis involves studying the hydrodynamic and global
response of the heave plates in the SATH semisubmersible platform, and then developing a sim-
ple structural FE model in order to analyze their structural behavior under the loads generated
in the hydromechanic and coupled analyses, as well as in tests run in a laboratory set up.
The high number of load cases and long simulation times covered in the standards make
a pre-design process of such structures a very complex and resource-demanding one. This
highlights the need of having tools in hand in order to simplify or reduce the data needed to
process. In this Master Thesis, the tools used to analyze the behavior of the heave plates have
been used to explore possible relationships between the variables obtained in the hydrodynamic
analyses, such as the motions of the rigid body platform or the free surface elevation, the
frequency of the incident waves, etc. The peaks in the measured pressures time series were
identified, and the other variables (directly obtainable in a couple simulation) evaluated at
those same instants, with the aim of relating these outputs with the worst instant load on the
plates.
The geometry of the structure was produced using a commercial 3D modelling tool called
4.4. HYDRODYNAMIC ANALYSIS 27

Rhinoceros, and then exported into the analysis pre-processor included in the Sesam package
(DNV Software).
The plates were first analyzed hydro-dynamically assuming rigid-body-like behavior, as part
of a rigid platform. This linear hydrodynamic analysis was performed using the software pack-
age HydroD from Sesam, which calculates stability, uses frequency-domain analysis for eigenfre-
quencies and linear response computation, and a time-domain solver to compute local response
and local pressures. The output files generated in the frequency-domain analysis (including re-
sponse amplitude operators or RAOs, first and second order wave forces, added mass, damping
and hydrostatic restoring matrices) were then introduced in Sima in order to perform a coupled
analysis1 in a model with a simplified mooring system and a turbine acting only as a constant
thrust generating an overturning moment and static pitch angle (without a turbine controller).
Then, the output generated in the fully-coupled analysis (including the response of the
mooring-platform-turbine system and a set of time-dependant wave elevations) was reintroduced
in HydroD in order to complete the time-domain analysis, obtaining the global response of the
system and a set of local pressures on the modelled panels. These time-dependant pressures
(or load transfer ) were then imported into a structural FE model in MidasCivil to analyze
structural response of the heave plates under such loads.
For the coupled analysis, a hypothetical 3-blade upwind turbine of the following character-
istics was considered:

• Rated power: 2 MW.

• Rated speed, Vrated : 12 ms−1 .

• Cut in, Vin : 4 ms−1 .

• Cut out, Vout : 25 ms−1 .

• Rotor radius: 40 m.

• Tower height: 56 m.

This turbine basic definition is presented here for clarity, although a coupled analysis re-
quires a thorough definition of the geometry of the blades and tower, the control system of
the turbine, and that of the mooring system, if applicable. However, since the coupled system
has been simplified and the simulation is not part of the work performed by the author, these
specifications are not presented in this document, nor will they be further discussed in it.
The mooring system, in order for the models used in this Thesis to be later calibrated with
the tank tests in Phase I, has been simplified as a linear spring in the x direction, to mainly
restrain surge motion (although this configuration slightly affects the behavior in pitch and
other DOFs as well, this effect was neglected).

4.4 Hydrodynamic Analysis

A complete hydrodynamic analysis -frequency as well as time domain- was performed for the
whole geometry of the platform. The output of this analysis included stability, RAOs, added
1 As previously stated, the coupled analyses was run by Aitor Gallego using Sima, in the SOT office in Biscay.
28 CHAPTER 4. METHOD

(a) (b)

Figure 4.5: Half the simplified panel model defined in HydroD, (a) before and (b) after meshing.

mass, damping and hydrostatic restoring matrices, natural periods and motions of the rigid
body structure, although, for clarity, not all outputs are shown in this document.

4.4.1 Numerical Model

The software used for the numerical hydrodynamic analysis was HydroD, which combines dif-
ferent models, such as panel or Morison models and the FE method to solve for the required
output. The frequency-domain analysis performed in HydroD was run by the Wadam module,
which uses a combination of Morison’s equation and first and second order 3D potential theory
for the wave excitation calculations. For more information, the reader may refer to the user
manuals for GeniE [26], HydroD [27] and Wadam [28].
The geometry used in HydroD needed to be defined through the GeniE tool, distributed in
the same package. In this case, the guiding geometry was exported from Rhinoceros and the
simplified geometry (which did not include some structural details, since the global hydrody-
namic analysis did not require such degree of precision at this stage) defined and meshed in
GeniE.
The analysis in HydroD was done with specific assumptions for the different elements, such
as Morison theory for beam elements and potential theory for plates and shell elements. HydroD
allows for the models to be defined separately and the analysis set to use the two sub-models
in a so-called composite model.
All beam elements were simplified as circular tubes with equivalent diameters obtained as
in equation 4.2.
r
4A
deq = (4.2)
π
Where A is the cross-sectional area of the beams.
On the other hand, the shell model was prepared using the guiding geometry tool in GeniE,
importing guiding lines from the CAD software and making use of the symmetry of parts of
the structure (besides being completely symmetric in the x-z plane, some substructures are also
symmetric in the x-y and y-z planes), see figure 4.5 (a).
4.4. HYDRODYNAMIC ANALYSIS 29

A correct definition of the shells was ensured using the compartment tool, that generates
inner volumes of virtual fluid only in closed spaces. The geometry verification check was used
and passed satisfactorily, too.
The meshing process was performed using the advancing front quad mesher algorithm in
GeniE and then dividing the plates into simpler parts to ensure a correct discretization and
mesh topology. See mesh before and after manual correction in figure 4.6.

(a) (b)

Figure 4.6: Mesh discretization (a) before and (b) after manual refinement.

In order to ensure a correct convergence of results, some analyses were performed changing
the global mesh size and type. The complete results from this convergence analysis are presented
in Appendix A.
The analysis in HydroD was run with the following settings:

• Global characteristic mesh size: 0.4 m (length seed).

• Element types: 1st order (linear), 4-noded shell elements (internally referred to as F QU S)
and 3-noded shell elements (F T RS) (used where the meshing controls did not allow for
a 4-noded element), with reduced integration (1 Gauss point).

• Number of panel elements: 19796.

• Number of Morison elements: 458.

• Elapsed CPU time (frequency-domain): 7642 s, or around 2 h and 7 min.

This frequency-domain analysis was prepared using a set of environmental conditions, cor-
responding to a set of wave periods (the wave height used was a standard one in order to
non-dimensionalize the later response), and directions, as:

• Wave period set: from s to s, every 1 s.

• Wave direction set: from 0◦ -or propagation towards positive x- to 180◦ -or propagation
towards negative x-, every 15◦ (making use of symmetry to reduce CPU time).

The mass characteristics were implemented p in the model as a point mass and mass moments
of inertia -defined by the radii of gyration, ri = Ii /m- (see table 4.1) corresponding to the ones
previously computed in Rhinoceros. The reference point, or (0, 0, 0) was set to the intersection
of the 3 symmetry planes of the cylindrical floaters (see section 4.2) and the offsets of the global
center of gravity (COG) were referred to that point.
30 CHAPTER 4. METHOD

4.4.2 Wave Tank Tests

As part of the preliminary design process of the SATH platform, a campaign of tests was
scheduled in several phases during 2017. Each of these phases focused on one or several specific
targets that define the response of the floating platform.
The first phase in 2017 was run during July, inside the facilities of the IHCantabria. The
test model was a scale to the real size prototype (the scaling-down was made according to
Froude scaling), and focused on the hydrodynamic response of the platform under wave loads
with a stationary wind load. The mooring system was simplified as a set of springs for this test
phase, as in the numerical models described previously in this document.
The model was made out of steel, aluminum and polymers and designed to reliably represent
mass distribution and displaced volume, so as to correctly study the static and dynamic stability
and eigenfrequencies.
It is important to note that this model was, for all aspects regarding hydrodynamic behavior,
assumed to be completely rigid. This assumption indeed holds true for most parts, since the
floating structure will tend to move way more easily as a rigid body than to deform under wave
and wind loads. However, some parts of the structure cannot be regarded as rigid bodies. An
example of these is the tower, since its stiffness and dynamic behavior might affect the overall
motions of the platform, and thus should not strictly be assumed rigid.
The issues concerning the dynamic behavior of the tower were disregarded during this first
phase of tank tests. Thus, the tower was modelled in the first phase as a rigid element -a
stiffened hollow tube-, in order to account for the extra weight it adds to the global structure.
The wind load was included in the tests, modelled as a constant thrust acting on top of the
tower, causing a stationary pitch angle.
The Froude scaling laws are defined by a constant dimensionless monomial known as Froude’s
√
number (F = v/ gL), which gives the relation between inertial and gravity forces. By applying
a length scaling factor λL , such that λL = Lp /Lm , where Lp stands for prototype dimensions
and Lm for model dimensions (this notation will be used consistently throughout this section),
and the constant Froude’s monomial, all the other relevant units can be properly transformed
(the interested reader might refer to [29] for more information on Froude scaling).
There are, however, some important issues when building a model out of a different material
than that of the prototype, especially if the mass distribution is crucial (the model was made
of steel while the prototype is mainly prestressed concrete). Some parts had to be adapted in
different ways as to keep all important variables correctly modelled. Below, some key notes on
how this was achieved are presented.

• The outer dimensions were transformed directly using λL , as in:

Lp
Lm = (4.3)
λL

Table 4.1: Summary of the characterization of the mass model of SATH for HydroD.

[t] [m] [m] [m] [m] [m] [m]

Mass XCOG YCOG ZCOG rx ry rz
0
4.4. HYDRODYNAMIC ANALYSIS 31

(a) Designed plate. (b) Real plate, in the lab.

Figure 4.7: Solution for the scale model heave plates -only bow plate shown- (steel in grey and
PMMA attachments in yellow).

• The bar and beam elements had outer dimensions scaled down, too. Their thickness,
however, was chosen taking into account mass distribution and constructability in the
workshop.

• The outer dimensions of the cylindrical floaters were scaled down with λL , except for their
thickness, which was transformed as
 
Lp ρc
Lm = λρ (4.4)
λL ρs

where ρc and ρs are the densities of concrete and steel, respectively, and λρ is the
prototype-to-model density ratio, defined by the ratio of seawater (ρw ' 1.025 g/cm3 ) to
the freshwater used in the wave tanks (ρf w ' 1.0 g/cm3 ).

• The heave plates were treated somewhat differently, for they were crucial in both external
dimensions and mass distribution. The widths and lengths of the plates were transformed
with λL , in the same way as any other external dimension. To achieve a correct thickness
and mass, a mixed-material approach was taken: the plate (as it can be seen in figure
4.7) was built by covering a relatively thick plate made of a lighter material (in this case
poly(methyl methacrylate), or PMMA) on top and bottom with a thin steel plate, in
order to reduce the global mass of the plate while still allowing welded connections on it.

• Regarding the internal structural diaphragms -or bulkheads- inside the hulls, a compro-
mise solution was achieved. In such a scale model, the bulkheads (whose main function is
to stiffen and compartmentalize the hulls) are not strictly necessary. However, for the sake
of a correct mass and inertia representation, some thin round steel frames were welded to
the inside of the hulls.

A summary of the characteristics of the scale model is shown in table 4.2, where ID 1
corresponds to a theoretical -objective- model, directly scaled down (in concrete), and ID 2
to the scale model adapted following the considerations outlined previously (i.e. model made
of steel and mass and moments of inertia correctly represented). ID 3 describes the achieved
characteristics after construction and ballast calibration in the lab. They can be seen in figure
4.8 as well.
32 CHAPTER 4. METHOD

Table 4.2: Global characteristics of scale models.

[kg] [kg · m2 ]
ID Scale Model Mass Ix Iy Iz
1 Objective
2 Designed
Diff. to goal < 0.1% 2.66% 1.86% 0.27%
3 Built
Diff. to goal < 0.1% 0.45% 0.26% 2.45%

(a) (b)

Figure 4.8: Scale model geometry, (a) theoretical and (b) built.

As it is shown in table 4.2, the scale model was able to represent the mass and main moments
of inertia of the full scale prototype within a ±2.5% error range.

Once the model was built in the workshop, its mass distribution and inertias calibrated in
the lab and its watertightness ensured, the model was characterized in the wave tank in the
facilities of the University of Cantabria: decay tests were performed to find its eigenperiods and
modal damping values, drift and stability tests before the wave tests begun.

This model was also subjected to several sets of regular and irregular wave trains; the former
defined by pairs of regular Hw and Tm , the latter corresponding to JONSWAP-type spectra
defined by pairs of Hs and Tp , at various directions: aligned with the platform x axis (0◦ ), at
30◦ and 90◦ angles, to study its sensitivity to transverse wave excitation and load misalignment.
The wind was modelled using various levels of constant thrust generated by a drone on top of
the tower, causing a stationary pitch angle and a consequent change in geometry and shift in
eigenperiods, corresponding to a power production case in normal conditions.
4.5. PLATE LOADS ANALYSIS 33

4.5 Plate loads analysis

When studying some complex phenomena, such as the flow around the floaters, tower and heave
plates in this structure, one can take different approaches as to how to get the appropriate data
and the way to process it. During the making of this Thesis, a study was developed using some
raw data from the tank tests and some numerical simulations comparable to specific cases in
the tank tests, that were processed using Python scripts, shown in Appendix C. The main goal
of this study was to analyze the behavior of the heave plates in this semisubmersible floating
platform, in order to simplify a later structural design.
The raw data obtained in the coupled simulation consisted of:

• Global displacement series: in the x, y and z coordinates.

• Global rotation series: again, split in x, y and z.

• Free surface elevation, representing the waves generated in the simulation.

• The experimental data included two time series of the hydrodynamic pressures2 on the
center of the bow heave plate, on the top and bottom surfaces. The coupled simulation is,
however, virtually -computationally- incapable of generating a complete time-series of the
pressures for a long simulation. Instead, snapshots can be extracted for specific instants.

A frequency domain analysis was first addressed, by obtaining the global maximum value
for each time series in the plate pressures distribution obtained with the pressure cells, and
representing the magnitude of these pressures against the incident wave period. This analysis
allowed for a relatively general study of the loads on the heave plates for several wave heights
over a range of periods, identifying frequency regions where the plates were most loaded.
Also by using the time series obtained in the lab, a time domain study was later performed
and a qualitative analysis made, although the conclusions that could be drawn can not be
generalized to other structures or wave types due to the complexity of the phenomena involved
in the whole process.
In the time domain analysis, the loads on the plates were hypothesized to be caused by rigid-
body motions (including velocities and accelerations) in heave and pitch and by the incident
wave height and periods. In the frequency domain analysis, a relationship was sought between
the incident wave period, the plate velocity and acceleration and the value of the loads on the
plates.
If these hypotheses were true, it should be expected that the higher the pressures measured,
the higher the velocities or accelerations (or both).
The time series processed required certain transformations before they could be analyzed
and compared. For instance, all displacement and rotation time series were used to compute
the velocity and accelerations, by implementing a simple Python code to obtain the first and
second derivatives using the central difference method (CDM), as in:

ηi+1 − ηi−1
η̇i = + O(∆t2 ) (4.5)
2 ∆t
2 All measured values represent the variation of pressure from the static virtual zero, arbitrarily chosen as the

hydrostatic pressure on the plate with the upright platform with no waves or wind acting on it.
34 CHAPTER 4. METHOD

(a) (b)

Figure 4.9: Sketch of the platform in (a) upright position and (b) with a trim angle.

Figure 4.10: FFT of the motion signal and its associated velocity and acceleration spectra,
unfiltered (left) and filtered (right)

ηi+1 − 2ηi + ηi−1

η̈i = + O(∆t2 ) (4.6)
∆t2
Where O(∆t2 ) indicates that the truncation error introduced when using these CDM ap-
proximations is of the order of the square of the timestep in the data series.
For the experimental data, the signal needed to be filtered before numerically derived, since
the high-frequency noise caused unrealistic peaks and instabilities in the derived variables. In
figure 4.12, these instabilities can be easily observed in the time series. If the same data is plotted
in the frequency domain (figure 4.10) using a simple fast-Fourier-transform (FFT) routine in
Python (see Appendix E), one can easily observe the noise amplification for higher frequencies
that appears when using the numerical differentiation routine. A Low Pass Butterworth filter
was applied leaving only the first harmonic of the wave excitation, allowing for a cleaner and
simpler representation of the signal.
Following rigid body physics for motions and accelerations (see figure 4.9, the velocities and
4.5. PLATE LOADS ANALYSIS 35

Figure 4.11: Time series of the motion signal and its associated velocity and acceleration,
unfiltered (left) and filtered (right)

(a) (b)

Figure 4.12: Example of normalized signals: (a) unfiltered and (b) filtered.
36 CHAPTER 4. METHOD

accelerations in the heave plates vpl and apl normal to the plate surface, if the distance between
the center of rotation and the center of the heave plates is a constant distance r, were computed
as:
vpl = η̇1 sin(η5 ) + η̇3 cos(η5 ) − η̇5 r (4.7)

apl = η̈1 sin(η5 ) + η̈3 cos(η5 ) − η̈5 r (4.8)

Where ηi are the motions in the ith DOF, and the dots indicate time-derivatives, following the
notation used throughout this document. The expression in equation 4.8 can be directly derived
by a translation of the accelerations experienced in the COG of the platform to the plates and
accounting for the tangential component of the acceleration in the plates due to acceleration in
the pitch rotation ([30], (5)).
One study performed in the analysis of the time series consisted in identifying the set of
peaks in the pressure series and evaluating the other series at those same instants. This could
assess the phase correlation between the different parameters and could hint at whether some
of them are directly related to the pressures.

4.6 Structural Analysis

Once the static and hydrodynamic analyses have been performed and the responses of the
platform determined, the loads obtained as local pressures in HydroD and the tank tests can
be then transferred to a structural FE model. In order to do so, the software MidasCivil has
been used.

4.6.1 Finite Element Model

A simple FE model was developed with the purpose of studying the load transfer from the
hydrodynamic model and the effect of the load pressures on response of the heave plates.
Due to the way HydroD produces the load transfer functions (a set of elements defined by
nodes, being these defined by coordinates), the mesh defined in MidasCivil is restrained by that
of the the hydrodynamic model. This imposes a heavy limitation on the definition of the struc-
tural FE model, since, although the analyses in MidasCivil do not involve large computational
time, the time restriction lies on the time-domain load analysis in the hydrodynamic model.
The geometry, as well as all loads and boundary conditions (BC), is symmetric with the
xz plane. Thus, symmetry has been used to reduce the CPU time by modelling only half the
model, and imposing the corresponding BC, described below.
The relatively complex geometry of the hollow-core heave plates was simplified as a simple
directional slab. In order to obtain a simple model but still find a close behavior to the real
heave plate, the sectional properties of the slab were taken as uniform in its in-plane directions,
but different to each other (it is clear that, since the voids have a dominant direction, the
behavior of the slab should be different in these two directions, see figure 4.13).
The sectional inertia per unit width of the voided slab was computed for its two in-plane
directions (x and y) and compared to those of a solid thick slab. Then, these ratios were input
in the FE model as directional bending resistance corrections, as well as the area ratio of the
4.6. STRUCTURAL ANALYSIS 37

voided to solid slab for the directional axial stiffness. All corrections implemented are listed in
table 4.3.

(a) Y direction. (b) X direction.

Figure 4.13: Slab sections perpendicular to its in-plane directions.

Table 4.3: Directional characteristics of the modelled slab.

Area (m2 ) I (m4 /m) Area (m2 ) I (m4 /m)

Y Direction Y Direction X Direction X Direction
Solid slab 8.28 4.27E-02 17.68 4.27E-02
Voided slab 6.20 4.02E-02 10.84 3.85E-02
Stiffness affected Fyy Myy Fxx Mxx
Scale factor 0.749 0.943 0.613 0.903

In order to correctly implement all pressure loads, the thick slab should receive the two
distinct components of the pressure field from its top and bottom surfaces. However, since the
modelled slab is defined through its mid-surface and a thickness, these two pressure components
were simplified into a resultant force (or net pressure) acting on the plate element. This means
that any compressive strain within the plate itself is disregarded, since it could be considered
negligible compared with the deflection of the slab.
The slab elements were modelled as linear quadrilateral Mindlin elements3 and the diagonal
support beam as an Euler-Bernoulli beam, discretized in linear finite elements.
The BC applied are those shown in figure 4.14, including the xz symmetry on the middle edge
of the plate, where restrictions in δy , θx and θz have been imposed (standard 3D xz-symmetry
BC). In figure 4.14, the symmetry BC is colored in green and black showing constrained DOF
in green and independent DOF in black, for clarity.
There are two areas to pay special care to when modelling: the BC on the edge of the slab,
and the beam-to-plate connection. It was expected that the post-tensioned concrete connection
between the plate and the bottom part of the cylindrical floater (connection at the edge of the
slab) restrict most displacements and rotational motions, but the definition of the tendons was
3 Mindlin or thick plate theory assumes that ”a line that is straight and normal to the mid-surface prior to

deformation remains straight but not necessarily normal to the deformed mid-surface” [31], applies the same
principles as Timoshenko theory for beams.
38 CHAPTER 4. METHOD

(a) Plan view. (b) Perspective.

Figure 4.14: Boundary conditions and plate-to-frame rigid links showing constrained DOFs.

not yet complete and thus an accurate boundary could not be defined at the time the FE model
was prepared. However, two extreme cases that should theoretically represent upper and lower
bounds in the response of the plate were implemented and compared: a simple pin support
(restricts δx , δy , δz ), and a clamped support (all 6 DOFs are restricted).
The beam-to-plate connection was modelled, on the other hand, with a master-slave rigid
link between the nodes on the beam elements (master) and a set of nodes close to them in the
plate (as in [32]).
One material was used to define the FE model shown in table 4.4, as specified in EC2 for
C35/45.

Table 4.4: Materials defined in the structural FEM, from EN 1992-1-1:2004 [33].

Name E [kN/m2 ] ν [−] ρ [kN/m3 ]

C35/45 3.4E+07 0.2* 25
*[33] 3.1.3 (4)

4.7 Load Case Definition

Following the results from the experimental data, two regular wave cases were proposed to be
studied numerically to define the set of load cases. They were chosen in the vicinity of the
natural period in heave, which, according to what can been inferred from the experimental
results, should show highest pressures and net forces on the heave plates (see Chapter 5).
4.7. LOAD CASE DEFINITION 39

For the purpose of this Master Thesis, and since the models used to estimate loads use linear
potential theory, the wave sets chosen consisted of low to medium wave heights, corresponding
to some usual operational sea states, which fit the assumptions made in linear theory. A design
based on more severe sea states was considered out of the scope of this Master Thesis.
The output files from the time-domain analysis in HydroD contained snapshots of the pres-
sures acting on the panel elements for a discrete number of seconds, enough to record a complete
cycle once all transient effects were dissipated. On the other hand, the experimental data con-
sisted of about 500 waves, deemed sufficient for all transient effects to be dissipated and a
stationary response be recorded. This experimental data was then filtered and cut so as to
have a usable stationary signal, for both motions and pressures.
All the numerical data was analyzed and processed using algorithms implemented in VBA
in MS Excel and Python, in order to obtain a set of nodal coordinates, elements and values
of the pressure on the element. The VBA code used in Excel is shown in Appendix B. The
experimental data, containing pressures on the plates consist of one single data point for the top
surface in the bow plate and one for the bottom surface, was processed using Python scripts.
The loads obtained in the tank tests were processed in two different ways in order to define
the load cases.

• First, the pressure distribution was taken as constant throughout the entire surface of the
plate and equal to the value recorded by the pressure cells (uniform distribution, figure
4.15 (a)).

• Later, using data from both experimental and numerical analyses, it was assumed that
the pressure distribution was similar to the one obtained in the simulations (variable
distribution, figure 4.15 (b)), while the magnitude of the pressures was scaled so that the
magnitude of the pressure at the center of the plate equaled the one recorded in the tank
tests. This approach might better represent the load distribution on the plates, although
this distribution might change depending on the motions and phase lag of the different
phenomena, and thus the results from these combined load cases should be interpreted
with caution. It was noticed, however, that the numerical pressure distributions did not
deviate too much from the average (see figure 4.16), so an equivalent uniform distribution
should give a similar behavior. This fact will be mentioned later in section 5, when the
results are presented.

The load cases run are thus combinations of the different waves, pressures (either obtained
numerically or from the tank tests), pressure distribution on the plates (uniform, using the
single experimental data point, or variable, using the distribution taken from HydroD), and the
boundary conditions on the edge of the heave plate (either pinned or clamped connection).
The list of load cases is outlined in table 4.5.
40 CHAPTER 4. METHOD

(a) Uniform load (b) Non uniform load

Figure 4.15: Approximate shape of the pressures distribution input in the structural FE model.

(a) Front view (b) Side view

Figure 4.16: Approximate shape of the pressures distribution obtained in the hydromechanical
models.

Table 4.5: Load Case definition.

LoadCase H [m] T [s] Pressures Pressure distribution Plate BC

LC1 Low ∼Heave Numerical Variable Clamped
LC2 Low ∼Heave Experimental Uniform Clamped
LC3 Low ∼Heave Experimental Variable Clamped
LC4 Medium ∼Heave Numerical Variable Clamped
LC5 Medium ∼Heave Experimental Uniform Clamped
LC6 Medium ∼Heave Experimental Variable Clamped
LC7 Low ∼Heave Numerical Variable Pinned
LC8 Low ∼Heave Experimental Uniform Pinned
LC9 Low ∼Heave Experimental Variable Pinned
LC10 Medium ∼Heave Numerical Variable Pinned
LC11 Medium ∼Heave Experimental Uniform Pinned
LC12 Medium ∼Heave Experimental Variable Pinned
5. Results

5.1 Damping, Natural Periods and RAOs

The modal damping could be obtained through the experimental data recorded in several decay
tests. These consisted on forcing a displacement on the platform and letting it go: the varying
amplitude of the oscillations in the DOF studied were then recorded.
The decay tests showed that, especially for pitch and heave DOFs, the initial oscillations
do not decay linearly, indicating that there are important nonlinear effects taking place in the
first seconds of the data. An important nonlinear damping term would imply a positive effect
on the platform motions, that is smaller motions than predicted with only linear damping.
The approach taken to split the linear and nonlinear terms of the damping was based on the
traditional method, described in [34] and implemented in [35]. A sample of the work performed
on this matter is shown in figure 5.1.

Figure 5.1: Estimation of damping by the traditional method using a free decay signal. Image
produced using a Python script developed by Alexandre Morató.

The displacement transfer functions or RAOs for the DOFs analyzed are shown in figure
5.2. These were produced in the frequency-domain analysis in HydroD and clearly show the
amplified response of the structure close to each DOF natural period. The discretization in the
frequency domain was taken as 1 s, as stated previously. The author acknowledges that, for a
thorough hydrodynamic characterization of the floating platform, a finer discretization might
have been appropriate, and the RAOs studied for a more complete range, in order to study

41
42 CHAPTER 5. RESULTS

(a) Pitch (b) Heave

Figure 5.2: Motion RAOs obtained in HydroD.

(a) Pitch (b) Heave

Figure 5.3: Motion RAOs obtained from the experimental data.

possible long-wave excitation effects and mooring dynamics in long periods. However, this is
out of the scope of the study presented in this Master Thesis.

From the experimental data, these and other RAOs could be obtained, by using an average
of the peaks in the motion for the specific DOF (to avoid extreme unrealistic values present
in the measured time series), and plotting them against the incident wave periods. This way,
RAOs for velocities and accelerations in heave, pitch, and the bow heave plate were generated.

The RAOs for heave and pitch motions are shown in figure 5.3, compared with the ones
obtained in the numerical simulations. It is clear that the peaks at the resonance periods were
largely overestimated in HydroD, probably due to an important contribution of the heave plates
in added mass and nonlinear damping, especially in these two DOFs.

The heave RAO predicted by the model also suffers from the same overestimation in the
vicinity of the natural period in heave which, again, could be explained with an underestimation
of the damping for the heave DOF, provided by the large heave plates.
5.2. STABILITY 43

5.2 Stability
Regarding intact stability, the curves were obtained for pitch and roll motions in Sesam. Recall
that safety against capsizing is computed based on the ratio between the area under the righting
moment curve and that under the heeling moment curve (the heeling moment curve can be
simplified as MH ' Tmax hhub cos(α), where Tmax is the maximum turbine thrust, hhub is the
distance between the center of the hub and the center of buoyancy of the platform, and α is
the heel or trim angle, for roll and pitch respectively), and DNV proposes a minimum factor
of safety of 1.4.
The pitch stability curve obtained in Sesam is shown in 5.4, as well as the computed factor
of safety in pitch, with a value of 2.84.

Figure 5.4: Pitch stability curve as computed numerically in Sesam.

5.3 Loads
The loads were analyzed on two planes of the heave plates, corresponding to the top and bottom
surfaces, which coincide with where the panels were defined in the hydrodynamic model.
They were obtained from HydroD as a set of plate pressures on each of the panel elements
meshed, or from the pressure cells set in the tank tests (only one data point per surface), and
were post-processed and organized using MS Excel and Python scripts.
From the tank tests, the maximum loads were plotted for the bottom surface (as well as
the maximum differences between top and bottom pressures, corresponding to the maximum
net force acting on the plate) for different incident wave periods. These charts are shown in
figures 5.5 (a) and (b). They show certain patterns with the wave periods regardless of the
wave height, since all point sets show similar shifted patterns to greater pressures for each set
of larger waves.
Figure 5.5 (a) shows clearly that the region for which the pressures on the plates were larger
coincides with lower periods (corresponding to shorter waves), and these show a peak which
coincided with the approximate natural period of the platform in heave. This is coherent with
the previous assumptions, relating a larger amplitude of motion in heave (vertical translation)
with greater loads on the heave plates. In addition, it is clear that, for larger waves (in fig. 5.5,
44 CHAPTER 5. RESULTS

(a) Bottom surface pressure.

(b) Pressure differences.

Figure 5.5: Pressures on the plates with varying wave periods (experimental).
5.3. LOADS 45

(a) Plate velocities.

(b) Plate accelerations.

Figure 5.6: Frequency-wise velocities and accelerations on the bow plate (experimental).
46 CHAPTER 5. RESULTS

High), the load on the plate increased. This held true especially for shorter periods, since long
periods caused the platform to follow the surface of the wave, and thus the wave height became
less and less important. There is however a small increase in the pressures on both the top
and bottom surfaces for longer periods, which might have been caused by a slight resonance in
pitch motion, observed during the tests and predicted by the numerical models.
The peak in the plate pressures plot indicated that the heave plates slow down and restrain
the heave motion, thus smoothing out the peak in the heave motion RAO.
Focusing on the structural behavior of the heave plates as a whole, focus should be put
on the pressure difference instead of the surface pressures. The pressure difference trend is
plotted in 5.5 (b), and showed a clear drop in the net forces acting on the plates for increasing
wave periods (i.e. longer, flatter waves), with the highest loads appearing at the heave natural
period and below. Longer waves caused milder motions on the platform, whilst shorter and
thus steeper waves caused very complex flow patterns around the plates, which might explain
the higher pressure differences at these low periods.
One of the goals of the study performed on the experimental data was to investigate whether
there was a simple way of identifying uncommonly large peaks in the pressures on the plates
-or other loads acting on the floating platform- just by studying the time series that can be
directly obtained from Sima (motions at the COG of the platform and free surface elevations).
Values obtained and plotted in figures 5.5 and 5.6 indeed show that the hypotheses relating
the velocities and accelerations on the heave plates (obtained as in equations 4.7 and 4.8) with
the pressures hold: highest pressures occur around the natural period in heave (and below, for
the pressure difference); velocities and accelerations follow a very similar pattern (in figure 5.6).
It can be also inferred from the shape of the plots shown that heave motion is the dominant
phenomenon that determines the pressures on the plates.
The main result obtained through the investigation of the experimental data is shown in
figure 5.7 (b), which indicates a very strong linear correlation (r > 0.93) between the net
pressures and the accelerations on the heave plates. Figure 5.7 (b) shows a weaker correlation
with the velocities (r > 0.76). These results are coherent with the hypotheses proposed and lay
a foundation for future analysis of irregular wave time series, in order to easily identify parts
of the time series where highest loads should be expected, especially if the evaluation of the
complete pressure history is not directly obtainable.
If the complete time series were used instead of filtering and using only maximum values
for each series, it could be observed that, besides the fact that the highest accelerations in the
heave plates occur at around the same excitation frequencies as the maximum surface pressures
on the plates, the pressure oscillations are in phase with these plate accelerations. The phase
correlation, though, was shown to hold for some wave cases better than others: it appeared
more clearly for shorter periods than for longer, flatter waves. The case shown in figure 5.8 is
shown for its clarity and to illustrate the way the analyses were performed.
Figure 5.8 was thus obtained by identifying the pressure peaks (top graph, pressure), and
evaluating the plate velocities and accelerations (center and bottom graphs, respectively) at
the same instants the pressure peaks occurred. It could be observed that the peaks consistently
happened for maximum accelerations and zero (or close to 0) plate velocities, indicating a 0
and 90-degree phase shift of the pressures with the accelerations and velocities, respectively.
5.3. LOADS 47

(a)

(b)

Figure 5.7: Correlation plot of the net pressure on the heave plate with the plate (a) velocities
and (b) accelerations.
48 CHAPTER 5. RESULTS

Figure 5.8: Evaluation of velocities and accelerations for bottom pressure peaks. In red, the
positive peaks, in green, the negative ones.

5.4 Structural response

5.4.1 Quality assessment
The analysis of the structural behavior of the heave plates starts with a simple quality check.
The slab modelled with Finite Elements was compared to a beam structure -only the case with
the clamped boundary condition-, as if the strip between the connection and the support beam
could be considered a 1D beam with equivalent mechanical properties carrying all the load, as
shown in figure 5.9. For a more comprehensive explanation of the simplifications in order to
achieve the model shown in figure 5.10, see Appendix D.

(a) (b)

Figure 5.9: Slab sketch and virtual beam within the slab.
5.4. STRUCTURAL RESPONSE 49

Figure 5.10: Equivalent indeterminate beam with spring.

In the quality assessment, only cases defined with a uniform acting load have been analyzed
in the equivalent beam model. However, in order to have more data points from which to
analyze its validity, a new Check load case was defined, by applying a uniformly distributed
load of 12 kN/m2 on the slab. This and the other applicable load cases (LC2 and LC5) are
shown and compared in table 5.1. Some indicative results were chosen for the comparison, such
as the reaction force on the supporting beam (that is, the spring in the equivalent beam model),
R; the deflection at the spring-supported end, δ2 ; and the reaction moment at the clamped end,
M1 .

Table 5.1: FE model Quality Assessment. Subscripts b and F E stand for beam model and FE
model, respectively.

kN/m2 kN kN m m kN m kN m
LC q Rb RF E ErrR δ2b δ2F E Errδ2 M1b M1F E ErrM1
Check 12 514 489 -5% 7.0E-04 8.0E-04 12% 1565 1274 -23%
LC2 10.39 445 405 -10% 6.0E-04 6.6E-04 9% 1355 1032 -31%
LC5 19.53 837 761 -10% 1.1E-03 1.2E-03 8% 2548 1938 -31%

It can be seen in table 5.1 that the reaction force and displacement estimations lie within a
12% range between the two models, while the error in the estimation of the reaction moment
at the clamped end is larger. However, the slab should perform a more efficient redistribution
of stresses, the supporting beam restricts not only vertical deformations but rotations too, and
the connection between the supporting beam and the slab is not made in a single point, but
in several elements, covering a larger portion of the beam. In this regard, and considering the
obvious differences between the models, the behavior of the finite element model was deemed
satisfactory for the purpose of this analysis.
The analysis of the structural behavior of the slab is presented in the following sections as a
set of line cutting diagrams for 3 lines along the slab. These lines are identified as LineX, LineY
and LineY2, and are shown in figure 5.11, and were defined so as to analyze the distribution of
load within the plate in two areas in the y direction and one in the x direction while avoiding
any cuts too close to the boundary conditions or rigid links defined in the model.

5.4.2 Displacement
The vertical displacement contours and line plots clearly show the importance of the boundary
conditions imposed in the model. In any case, results show that the deflections of the plate itself
50 CHAPTER 5. RESULTS

Figure 5.11: Cutting lines definition.

5.4. STRUCTURAL RESPONSE 51

Figure 5.12: Vertical deflection contour plot for LC1.

are very small, and always of the order of millimeters, when loaded with the hydromechanic
loads.

5.4.3 Plate forces and moments

The nomenclature followed in this section corresponds to the usual one found in many textbooks
and manuals. Put simply, a bending moment Mxx corresponds to such moment in the plate
that causes bending stresses on the plane perpendicular to the x direction (in global model
coordinates).
The first output obtained regarding plate forces and moments was the bending moment Mxx
along lineX (figure 5.14), corresponding to the moments at the main load-carrying strip on the
plate. As it could be expected, the bending moment distribution for the clamped cases (load
cases 1 through 6) shows a similar pattern to that of a beam clamped on one end and with a
spring support on the other end, finding the largest negative bending moments at the support.
On the other hand, if the BC is set to a pin support, the largest bending moments are positive
and occur close to the middle of the plate.
Plate moments Myy plots along lines Y and Y2 show that the moment-carrying distribution
in the y direction is less affected by the boundary conditions on the edge of the plate (for
comparison, see figure 5.15). Some more graphs can be found in Appendix E, for the interested
reader, in order to give further detail into the plate behavior.
It can be noted that load cases LC1, LC4, LC7 and LC10 produce greater forces and
moments on the structure than their counterparts (the other load cases with the same wave set
applied to them). This comes from the fact that the value of the pressures input in the model
for these load cases comes from a linear hydrodynamic model, which only accounts for the
damping computed with potential theory, which is a fraction of the real damping the structure
has. A calibration of the linear models using the damping values observed in the tank tests
52 CHAPTER 5. RESULTS

Figure 5.14: Bending moments Mxx along lineX for all load cases.
5.4. STRUCTURAL RESPONSE 53

Figure 5.15: Bending moments Myy close to the support (lineY) and further away from it
(lineY2), for some comparable cases.
54 CHAPTER 5. RESULTS

(a) (b)

Figure 5.13: Plate bending moments (a) Mxx and (b) Myy in LC1.

should be performed in the future, but is considered out of the scope of this work.
There is an important fact to notice from the results presented in this section: there is only a
slight difference in the behavior of the plate when applying uniformly distributed loads or loads
that follow the distribution obtained from the hydrodynamic model (figures 5.16 and 5.17).
This might imply that, in early stages of design, it could be worth developing a structural FE
model that does not follow the same mesh discretization as the hydromechanical model, and
use only uniform loads. This might allow for finer mesh discretization in the structural model
and a more efficient use of computer resources.
5.4. STRUCTURAL RESPONSE 55

Figure 5.16: Plots comparing the response of the plate under uniform distributed load and
under the load distribution obtained in the hydromechanic model. Cases with clamped BC.
56 CHAPTER 5. RESULTS

Figure 5.17: Plots comparing the response of the plate under uniform distributed load and
under the load distribution obtained in the hydromechanic model. Cases with pinned BC.
6. Conclusions, Discussion and
Future Studies

6.1 Conclusions
In this Master’s Thesis, the author learned how to model a 3D geometry, to develop simple
tools for estimation of mass properties, and to understand how they affect the hydrodynamic
behavior. He also acquired a deeper understanding of the physical phenomena under hydro-
mechanic effects such as added mass and damping, and learned to apply it to a real case tank
test, using common concepts in the industry, such as RAOs, stability curves, frequency and
time-domain analyses, etc.
The plate load analysis resulted in a strong correlation found between the accelerations on
the heave plates and the pressures acting on them, which may allow for an easy identification
of worst load instants in more complex hydrodynamic analyses with irregular waves.
Regarding the structural analysis, the author faced some of the complexities when modelling
concrete structures, especially the effects of using different boundary conditions and how to
define them, and could identify the critical points in the heave plates structure and study their
behavior.
It has been shown that the structural behavior of the heave plates under a pressure distri-
bution as output from a linear hydromechanic model is similar to a uniform load distribution
of the same magnitude. In early stages of design, it could be worth developing a structural
FE model that does not follow the same mesh geometry and discretization as the hydrome-
chanical model, and could be loaded only with uniform loads. This might allow for finer mesh
discretization in the structural model and a more efficient use of computational resources.
In the case of heave plates similar to the ones analyzed in this Master Thesis, there might
be no need to obtain a high amount of element pressures, but only a few data points, and
check that the pressure distribution does not deviate significantly from an equivalent uniform
pressure of the same magnitude. It should also be noted the importance of a good estimation
of damping and added mass in the hydrodynamic models, to avoid a significant overestimation
of the loads acting on the structure.

6.2 Discussion and limitations of the analyses.

The main limitations concerning the work performed in this Master’s Thesis are related to the
linearity of the models used. This linearization is a fine and usable simplification for many
engineering problems, yet quite inadequate for certain purposes, as it has been shown regarding
the damping estimation, for instance. Tank tests show indeed the importance of calibration of
the models with linear and nonlinear damping, in order to reduce the estimated linear motions

57
58 CHAPTER 6. CONCLUSIONS, DISCUSSION AND FUTURE STUDIES

and loads on the structure.

All the data used throughout this Master’s Thesis corresponds to artificial regular wave
sets, with a constant wave height and period. This is indeed very useful for any effect of mode
coupling or nonlinear natural excitations to be identified, but is very limited regarding a real
design stage.
There are also some limitations on the FEM model, such as the effects of post-tensioning on
the plates. This implies an initial stress state in compression, so the loads are not likely to cause
the displacements analyzed here but even smaller deflections. This is indeed important if one
seeks to study fatigue and cracking effects. Besides, a linear analysis with a concrete structure
is limited if applied for design purposes. However, a linear structural model was indeed within
the scope of this Master Thesis, since the goal was not to design or optimize the structure but
to analyze its behavior under specific sets and shapes of loads.

6.3 Recommendation for future studies.

The main results from the plate load analysis are valid for the regular wave sets analyzed here,
but the conclusions drawn in this work should be further investigated if applied to real sea
states, defined by wave spectra such as JONSWAP.
The effect of using a fully coupled model with a working turbine and controller, complete
mooring system and stochastic wind models could give further insight into the consequences of
having a coupled system working together regarding the stress level on the heave plates.
Also, so far, the natural frequencies of the heave plates have been disregarded, since they are
considered to be far from the natural periods of the incident waves and the natural periods of
the platform itself as a rigid body, but another recommendation for further work into this topic
is to develop dynamic FE models, considering a complete geometry (disregarding symmetry),
in order to account for all symmetric and asymmetric vibration modes, and using time-varying
plate pressures on the plates (using the simplification of uniformly distributed load discussed
above, the amount of data to input into the model is reduced considerably).
The geometry of the heave plates could be defined as well using general solid-elements.
This could allow for a more detailed definition of the post-tensioning tendons, forces and time-
dependant losses. With those characteristics defined, one could perform a fatigue analysis, for
specific load cases, and an cumulative damage approach based on Miner’s Rule and rainflow
counting algorithms. As in almost any concrete structure, crack and crack propagation analysis
are crucial in the design.
Bibliography

[1] X. Garcı́a Casals, P. Linares Llamas, and F.J. Santos Pérez. Energy [R]evolution. A
sustainable world energy outlook 2015. Tech. rep. Greenpeace International, Global Wind
Energy Council, and SolarPowerEurope, 2015. url: http : / / www . greenpeace . org /
international / Global / international / publications / climate / 2015 / Energy -
[Link].
[2] Swedish Riksdag. Agreement on Swedish Energy Policy. http: // www. government. se/
articles/ 2016/ 06/ agreement-on-swedish-energy-policy/ .
[3] P. A. Lynn. Onshore and Offshore Wind Energy: An introduction. John Wiley & Sons,
Ltd., 2011. doi: 10.1002/9781119954613.
[4] I. Troen and E. Lundtang Petersen. European Wind Atlas. Tech. rep. Roskilde: Risø
National Laboratory., 1989.
[5] L. Fried et al. Global Wind Report. Annual Market Update. Tech. rep. Global Wind Energy
Council GWEC, 2016.
[6] L. Fried et al. Global Wind Report. Annual Market Update. Tech. rep. Global Wind Energy
Council GWEC, 2015.
[7] R. Falsing. “Verdens første havmøllepark rives ned [World’s first offshore wind farm being
demolished] ”. In: Ingeniøren (in Danish) (2017).
[8] L. Castro-Santos and V. Dı́az-Casás. Floating Offshore Wind Farms. Switzerland:
Springer International Publishing, 2016. doi: 10.1007/978-3-319-27972-5.
[9] O.M. Faltinsen. Sea Loads on Ships and Offshore Structures. Cambridge: Cambridge Uni-
versity Press, 1990.
[10] J. Jonkman. Dynamics Modelling and Loads Analysis of an Offshore Floating Wind Tur-
bine. Technical Report NREL/TP-500-41958. National Renewable Energy Laboratory,
2007.
[11] S. Butterfield et al. Engineering Challenges for Floating Offshore Wind Turbines. Confer-
ence Paper CP-500-38776. Available in [Link]
pdf Last retrieved 03-28-2017. National Renewable Energy Laboratory, 2007.
[12] J. Madslien. “Floating challenge for offshore wind turbine”. In: BBC News (Mar. 2009).
[13] http: // www. saitec-offshore. com/ . Last accessed 09-06-2017.
[14] http: // bimep. com/ en/ sobre-bimep/ . Last accessed 05-10-2017.

59
60 BIBLIOGRAPHY

[15] J.M.J. Journée and W.W. Massie. Offshore Hydromechanics. First Edition. Delft: Delft
University of Technology, 2001.
[16] E.W. Furunes and Norges Teknisk-Naturvitenskapelige Universitet. “Floating Wind Tur-
bines at Medium Water Depths”. MA thesis. Norges Teknisk-Naturvitenskapelige Uni-
versitet, Fakultet for Ingeniørvitenskap Og Teknologi, Institutt for Konstruksjonsteknikk,
2010.
[17] DNV-RP-H103 Modelling and Analysis of Marine Operations. Det Norske Veritas. 2011.
[18] S.K. Chakrabarty. Handbook of Offshore Engineering Vol. 1 and Vol. 2. Plainfield, Illinois,
USA: Elsevier. Offshore Structures Analysis, Inc., 2005.
[19] A. K. Chopra. Dynamics of structures: theory and applications to earthquake engineering
(4th ed.) Hoboken, NJ: Pearson, 2017.
[20] H.P. Gavin. “Classical Damping, Non-Classical Damping and Complex Modes”. CEE
541. Structural Dynamics, Department of Civil and Environmental Engineering, Duke
University. 2016.
[21] DNV-OS-J103. Design of Floating Wind Turbine Structures. Det Norske Veritas. 2013.
[22] IEC 61400-3:2009(E). Wind Turbines - Part 3: Design requirements for offshore wind
turbines. International Electrotechnical Commission. 2009.
[23] IEC 61400-1:2005(E). Wind Turbines - Part 1: Design requirements. International Elec-
trotechnical Commission. 2005.
[24] DNV-OS-C301. Stability and Watertight Integrity. Det Norske Veritas. 2011.
[25] R. Karoumi. “Structural Dynamics. Two and multiple degrees-of-freedom systems. Modal
analysis”. AF2011 Structural Dynamics, Department of Structural Engineering and
Bridges, KTH Royal Institute of Technology. 2016.
[26] Sesam User Manual. GeniE Vol. 1. Concept design and analysis of offshore structures.
GeniE Vol. 3. Modelling of Plate/Shell Structures. Det Norske Veritas. 2011.
[27] Sesam User Manual. HydroD. Wave load & stability analysis of fixed and floating struc-
tures. Det Norske Veritas. 2011.
[28] Sesam User Manual. Wadam. Wave Analysis by Diffraction and Morison Theory. Det
Norske Veritas. 2010.
[29] NTNU. “General Modelling and Scaling Laws”. TMR7 Experimental Methods in Marine
Hydrodynamics - lecture in week 34.
[30] J. Peraire and S. Widnall. “Lecture L25 - 3D Rigid Body Kinematics, Version 2.0. MIT
OpenCourseWare [Link]”. 16.07 Dynamics. 2009.
[31] C. Pacoste-Calmanovici. “Finite Element Modelling. Plate bending and shell elements.
Basic Assumptions.” AF2024 Finite Element Methods in Analysis and Design, Depart-
ment of Structural Engineering and Bridges, KTH Royal Institute of Technology. 2016.
[32] C. Pacoste-Calmanovici. “Finite Element Modelling. Closing lecture. Combination of dif-
ferent elements.” AF2024 Finite Element Methods in Analysis and Design, Department
of Structural Engineering and Bridges, KTH Royal Institute of Technology. 2016.
BIBLIOGRAPHY 61

[33] EN 1992-1-1:2004. Eurocode 2: Design of concrete structures - Part 1-1: General rules
and rules for buildings. European Committee for Standardization CEN. 2004.
[34] K.J. Rawson and Tupper E.C. Basic Ship Theory. Woburn MA: Butterworth-Heinemann,
2001.
[35] Edward M. Lewandowski. “Comparison of Some Analysis Methods for Ship Roll Decay
Data”. In: 12th International Ship Stability Workshop (June 2011).
Appendices

62
Appendix A

Convergence Analysis.
Frequency Domain Mesh in HydroD.
In order to study the convergence of results, 4 different analyses have been run in HydroD,
varying the mesh densities, generated from a seed length. The mesh seeds or reference lengths
used are 2 m for the coarser mesh, 1 m, 0.5 m and finally 0.4 m for the finest -this value is
limited by the maximum number of panels the software can analyze-. It should be noted that,
for a more thorough convergence analysis, the discretization of the frequency domain (number
of periods analyzed) could also be studied, as well as the frequency range in which the analysis
is run.
The convergence has been studied on several outputs: the rigid-body eigenfrequencies for
pitch, roll and heave, and their respective Response Amplitude Operator, or RAOs.

Convergence in eigenperiods
First, the values for the three eigenperiods obtained in this convergence study are shown below.
It can be seen that there is a convergence in the numerical value of the eigenperiod for all
three DOFs studied with increasing mesh density, although the absolute numerical difference
is quite small.
It can be easily seen, too, that the CPU time required to run analysis greatly increases with
mesh density, especially for the finer mesh seeds.

(a) Pitch (b) Roll (c) Heave

Figure A.1: Natural periods obtained for different mesh densities.

64
APPENDIX A 65

Table A1: Natural periods obtained for different mesh densities.

[m] [s] [s] % [s] % [s] %

Mesh Seed CPU time* TN 3 Error TN 3 TN 4 Error TN 4 TN 5 Error TN 5
2 43 0.86 0.30 1.17
1 87 0.37 0.15 0.43
0.5 3017 0.05 0.02 0.06
0.4 7642 − − −
* on an Intel® Core™ i5-3550 3.30GHz, running Windows-7 64bits

Convergence in RAOs
The RAOs also show a convergent behavior, especially in the areas away from the rigid-body
resonant period. This might be due to the low damping implemented in the model, which
causes a slight divergence in and around the eigenperiod, where the relative motions tend to ∞
or, at least, very high values.
These peak values obtained numerically were shown unrealistically conservative after the
wave tank tests, where the maximum motions of the platform under wave frequencies close
to resonance showed an amplitude smaller than that predicted by these models. This, again,
might be caused by the very low damping used as input in the early numerical models (before
calibration), and some other non-linear effects.
It should be noted that, for high periods -i.e low frequencies, long waves-, the RAO in heave
tends to unity, which means that the platform follows the boundary layer of the seawater and
is not hit by the wave, but moves with it (picture the usual example of a seagull on the water
following the surface of the wave).

Figure A.2: RAO in pitch obtained for different mesh densities.

66 APPENDIX A

Figure A.3: RAO in roll obtained for different mesh densities.

Figure A.4: RAO in heave obtained for different mesh densities.

Appendix B

VBA scripts.
Processing the nodal data.

VERSION 1 . 0 CLASS
BEGIN
MultiUse = −1 ’ True
END
A t t r i b u t e VB Name = ” Hoja1 ”
A t t r i b u t e VB GlobalNameSpace = F a l s e
A t t r i b u t e VB Creatable = F a l s e
A t t r i b u t e V B P r e d e c l a r e d I d = True
A t t r i b u t e VB Exposed = True

Sub a u x i l i a r ( )
Dim zdim
Dim z t o p
Dim zbtm
Dim contbtm
Dim c o n t t o p
Dim c o n t e q
Dim minval
contbtm = 2
conttop = 2
z t o p = −5.4
zbtm = −6.2
minval = 0 . 0 1

For i = 2 To 3700 ’ round t o z e r o a l l v e r y s m a l l v a l u e s

For j = 3 To 4
I f Abs ( C e l l s ( i , j ) . Value ) <= minval Then
C e l l s ( i , j ) . Value = 0
End I f
Next j
Next i

For i = 2 To 3700 ’ c l a s s i f y t h e nodes a s from top s u r f a c e

67
68 APPENDIX B

’ o r bottom s u r f a c e
I f C e l l s ( i , 5 ) . Value = z t o p Then ’ i f v a l u e o f z c o r r e s p o n d s
’ t o ztop , copy t h e c o o r d t o t h e i r new l o c a t i o n
For j = 2 To 5 ’ node ID , c o o r d x , y , z
C e l l s ( conttop , j + 7 ) . Value = C e l l s ( i , j ) . Value
I f j = 5 Then
conttop = conttop + 1
End I f
Next j
E l s e I f C e l l s ( i , 5 ) . Value = zbtm Then ’ i f v a l u e
’ o f z c o r r e s p o n d s t o zbtm ,
’ copy t h e c o o r d t o t h e i r new l o c a t i o n
For j = 2 To 5
C e l l s ( contbtm , j + 1 3 ) . Value = C e l l s ( i , j ) . Value
I f j = 5 Then
contbtm = contbtm + 1
End I f
Next j
Else
End I f
Next i

conttop = 2
contbtm = 2
conteq = 2

For i = 2 To 1850 ’ e q u i v a l e n c e o f nodes from top s u r f a c e

’ t o bottom s u r f a c e , compare x and y
’ v a l u e s , copy i f t r u e
For j = 2 To 1850
I f C e l l s ( i , 1 0 ) . Value = C e l l s ( j , 1 6 ) . Value Then
I f C e l l s ( i , 1 1 ) . Value = C e l l s ( j , 1 7 ) . Value Then
C e l l s ( conteq , 2 1 ) . Value = C e l l s ( i , 9 ) . Value
C e l l s ( conteq , 2 3 ) . Value = C e l l s ( j , 1 5 ) . Value
conteq = conteq + 1
Else
End I f
Else
End I f
Next j
Next i

End Sub
#######################
Processing elements data.
VERSION 1 . 0 CLASS
BEGIN
APPENDIX B 69

MultiUse = −1 ’ True
END
A t t r i b u t e VB Name = ” Hoja2 ”
A t t r i b u t e VB GlobalNameSpace = F a l s e
A t t r i b u t e VB Creatable = F a l s e
A t t r i b u t e V B P r e d e c l a r e d I d = True
A t t r i b u t e VB Exposed = True

Sub a u x i l i a r ( )

Dim c o n t
Dim t p e l e m e n t ( 1 ) As String
Dim b t e l e m e n t ( 1 ) As String
Dim c o i n ( 1 ) As String
cont = 3

For i = 3 To 1700
For j = 3 To 1700
t p e l e m e n t ( 0 ) = C e l l s ( i , 1 6 ) . Value
t p e l e m e n t ( 1 ) = C e l l s ( i , 1 7 ) . Value
b t e l e m e n t ( 0 ) = C e l l s ( j , 2 1 ) . Value
b t e l e m e n t ( 1 ) = C e l l s ( j , 2 2 ) . Value
I f C e l l s ( i , 1 6 ) . Value = C e l l s ( j , 2 1 ) . Value And
C e l l s ( i , 1 7 ) . Value = C e l l s ( j , 2 2 ) . Value Then
C e l l s ( cont , 2 3 ) . Value = C e l l s ( j , 1 8 ) . Value
cont = cont + 1
End I f

Next j
Next i

End Sub
Appendix C

Python scripts: plots.

All scripts in Python have been developed using Anaconda, an open source platform for distri-
bution of Python programming, and the spyder environment.

Time derivative function

# −∗− c o d i n g : u t f −8 −∗−
”””
C r e a t e d on Mon J u l 17 1 6 : 2 4 : 1 6 2017

@author : r o d a l b
”””
import numpy a s np

def d e r i v ( t , x ) :
’ r e t u r n s t h e 1 s t d e r i v a t i v e o f t h e s e r i e s , C e n t r a l D i f f Method ’
dt = t [1] − t [ 0 ]
dx= [ ]
f o r i in range ( len ( t ) ) :
i f i ==0:
dx = np . append ( dx , ( x [ i +1]−x [ i ] ) / dt )
else :
i f i != len ( t ) −1:
dx = np . append ( dx , ( x [ i +1]−x [ i − 1 ] ) / ( 2 ∗ dt ) )
i f i ==(len ( t ) −1):
dx = np . append ( dx , ( x [ i ]−x [ i −1])/ dt )
return dx

##########
Pressure analysis in the frequency domain.

# −∗− c o d i n g : u t f −8 −∗−
”””
C r e a t e d on Thu J u l 20 1 2 : 4 7 : 0 6 2017

@author : A l b e r t o R o d r i g u e z

70
APPENDIX C 71

”””

import numpy a s np # vectores y matrices

import matplotlib . pyplot as p l t # Para h a c e r p l o t s
import math # numero p i y o t r o s
import peakutils # by Lucas Hermann Negri , b a s e d on Matlab ’ s ’ f i n d p e a k s ’
# l u c a s h n e g r i @ g m a i l . com
from t i m e d e r i v a t i v e import d e r i v
from s c i p y import s i g n a l
from s c i p y . s t a t s import l i n r e g r e s s

plt . close ( ’ all ’ )

show x ticks = False

wavetype = ’ R e g u l a r ’ # ’ R e g u l a r ’ or ’ I r r e g u l a r ’

t e s t r u n s 2 9 T = np . g e n f r o m t x t ( ’ // s e r v i d o r v 4 / Offshore Wind TRAB /PS067 DemoSATH \

p r e l i m /01−WIP/PS067A ENGINEERING/WIP WTT/ \
Test Campaign 2017/ Data p r o c e s s i n g / A m p l i t u d e s \
A l b e r t o / T e s t r u n s A 2 9 T . t x t ’ , dtype = str ,
skip header = 0)

t e s t r u n s 5 T = np . g e n f r o m t x t ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / \

IHC/WindWaves 5T/ T e s t s c u t s i g n a l s r e g 5 T . t x t ’ ,
dtype = i n t )
r u n s c o l s = [ ’H ’ , ’T ’ , ’ c u t b e g i n n i n g ’ , ’ c u t e n d ’ ]
max top = [ ] ; max btm = [ ] ; m a x d i f f = [ ] ; max vel = [ ] ; max acc = [ ] ; max z = [ ] ; max dz = [ ] ;
max ddz = [ ] ; max ry = [ ] ; max dry = [ ] ; max ddry = [ ]

f o r i in range ( len ( t e s t r u n s 2 9 T ) ) : #l e n ( t e s t r u n s 2 9 T )
plt . close ( ’ all ’ )
H wave = i n t ( t e s t r u n s 2 9 T [ i , 0 ] ) #m
T wave = i n t ( t e s t r u n s 2 9 T [ i , 1 ] ) #s
# p r i n t ( ’ Run %s , H=%d T=%d ’%( i +1,H wave , T wave ) )
s = i
v e r s = ’ 00 ’
method = ’ 2ND ’
V wind = ’ 11 p5 ’
t h r u s t = ’ 29T ’
#V wind = ’4 p0 ’
# S e t t h r e s h o l d and minimum peak d i s t a n c e ( i n x−v e c t o r u n i t s )
t h r e s h o l d m a x = 0 . 1#t e s t r u n s 2 9 T [ i , 6 ]
m i n d i s t a n c e = 50
v e r s = ’%s ’%t e s t r u n s 2 9 T [ i , 1 3 ]
method = ’%s ’%t e s t r u n s 2 9 T [ i , 1 2 ]

i f wavetype == ’ I r r e g u l a r ’ :
f i l e n a m e = ’E : /WORKSPACE/ Python /IHC/ I r r e g u l a r W i n d W a v e s %s /H%s T%s / \
FIH17 −00177 C1 JS H%sp0 T%sp0 G3p3 D21p0 A0p0 WDI%s ABS %s ’ \
72 APPENDIX C

%( t h r u s t , H wave , T wave , H wave , T wave , V wind , v e r s )

i f wavetype == ’ R e g u l a r ’ :
f i l e n a m e = ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python /IHC/WindWaves\
%s /H%s T%s /FIH17 −00177 C1 %s H%sp0 T%sp0 D21p0 A0p0 WDI%s ABS %s ’ \
%( t h r u s t , H wave , T wave , method , H wave , T wave , V wind , v e r s )

# Cut t h e s i g n a l ( b e g g i n i n g and end ) i n t h e p r e s s u r e s and wave e l e v f i l e

c u t b e g i n n i n g = i n t ( t e s t r u n s 2 9 T [ i , 4 ] ) #minimum 12
cut end = int ( test runs29T [ i , 5 ] )
# p r i n t ( ’% d\ t%d ’%( c u t b e g i n n i n g , c u t e n d ) )

# Import p r e s s u r e s p s
d a t a p s = np . g e n f r o m t x t ( f i l e n a m e+ ’ PS . c s v ’ , s k i p h e a d e r=c u t b e g i n n i n g ,
s k i p f o o t e r=c u t e n d )
col ps = 2
l e g e n d p s = [ ’ time ’ , ’ ps btm ’ , ’ p s t o p ’ ]
u n i t s p s = [ ’ s ’ , ’ mca ’ , ’ mca ’ ]
# diff ps = []
# f o r i i n ra n g e ( l e n ( d a t a p s ) ) :
# d i f f p s . append ( d a t a p s [ i ,1] − d a t a p s [ i , 2 ] )

# Import motions m
data m = np . g e n f r o m t x t ( f i l e n a m e+ ’ QTM. t x t ’ , s k i p h e a d e r=c u t b e g i n n i n g ∗ 2 ,
s k i p f o o t e r=c u t e n d ∗ 2 )
c ol m = 3 # 0 time , 1 dX , 2 dY , 3 dZ , 4 rX , 5 rY , 6 rZ , 7 dXt , 8dYt , 9 dZt
legend m = [ ’ time ’ , ’ D i s p x ’ , ’ D i s p y ’ , ’ D i s p z ’ , ’ Rot x ’ , ’ Rot y ’ , ’ Rot z ’ ,
’ Disp xt ’ , ’ Disp yt ’ , ’ Disp zt ’ ]
u n i t s m = [ ’ s ’ , ’m ’ , ’m’ , ’m ’ , ’ deg ’ , ’ deg ’ , ’ deg ’ , ’m ’ , ’m ’ , ’m’ ]
# # D e r i v e v e l o c i t i e s and a c c e l e r a t i o n s from motion time s e r i e s
t = data m [ : , 0 ] ; x = data m [ : , 1 ] / 1 0 0 0 ; z=data m [ : , 3 ] / 1 0 0 0 ; ry = data m [ : , 5 ]

bps , aps = s i g n a l . b u t t e r ( 3 , t e s t r u n s 2 9 T [ i , 1 1 ] ∗ 2 )
ps btm = s i g n a l . f i l t f i l t ( bps , aps , d a t a p s [ : , 1 ] )
p s t o p = s i g n a l . f i l t f i l t ( bps , aps , d a t a p s [ : , 2 ] )
diff ps = []
f o r j in range ( len ( d a t a p s ) ) :
d i f f p s = np . append ( d i f f p s , ps btm [ j ]− p s t o p [ j ] )

maxima top = p e a k u t i l s . i n d e x e s ( p s t o p , t h r e s=t h r e s h o l d m a x ,

m i n d i s t=m i n d i s t a n c e )
maxima btm = p e a k u t i l s . i n d e x e s ( ps btm , t h r e s=t h r e s h o l d m a x ,
m i n d i s t=m i n d i s t a n c e )
m a x i m a d i f f = p e a k u t i l s . i n d e x e s ( d i f f p s , t h r e s=t h r e s h o l d m a x ,
m i n d i s t=m i n d i s t a n c e )

max top . append ( np . a v e r a g e ( p s t o p [ maxima top ] ) )

max btm . append ( np . a v e r a g e ( ps btm [ maxima btm ] ) )
APPENDIX C 73

m a x d i f f . append ( np . a v e r a g e ( d i f f p s [ m a x i m a d i f f ] ) )
print ( ’ Bottom Top ’ )
print ( ’ %.3 f %.3 f ’ %(np . a v e r a g e ( ps btm [ m a x i m a d i f f ] ) ∗ 3 4 3 . 3 5 ,
np . a v e r a g e ( p s t o p [ m a x i m a d i f f ] ) ∗ 3 4 3 . 3 5 ) )

b , a = signal . butter (3 , test runs29T [ i , 1 1 ] )

x f i l t = s i g n a l . f i l t f i l t (b , a , x)
x = x filt
z f i l t = s i g n a l . f i l t f i l t (b , a , z )
z = z filt
r y f i l t = s i g n a l . f i l t f i l t ( b , a , ry )
ry = r y f i l t

dx = d e r i v ( t , x )
ddx = d e r i v ( t , dx )
dz = d e r i v ( t , z )
ddz = d e r i v ( t , dz )
dry = d e r i v ( t , ry )
ddry = d e r i v ( t , dry )

# Change u n i t s t o r a d i a n s
ry = math . p i /180∗ ry
dry = math . p i /180∗ dry
ddry = math . p i /180∗ ddry
u n i t s m [ 4 : 7 ] = ’ rad ’

#D e f i n e d i s t a n c e from c e n t e r o f p l a t e t o COG, r
r = 16.75 # m
r = r /35 # i n model d i s t a n c e
acc pl = [ ]
for n in range ( len ( t ) ) :
a c c p l = np . append ( a c c p l , ddx [ n ] ∗ math . s i n ( ry [ n ] ) +
ddz [ n ] ∗ math . c o s ( ry [ n]) − ddry [ n ] ∗ r )
vel pl = [ ]
f o r n in range ( len ( t ) ) :
v e l p l = np . append ( v e l p l , dx [ n ] ∗ math . s i n ( ry [ n ] ) + dz [ n ] ∗ math . c o s ( ry [ n ] )
− dry [ n ] ∗ r )
maxima vel = p e a k u t i l s . i n d e x e s ( v e l p l −np . a v e r a g e ( v e l p l ) ,
t h r e s=t h r e s h o l d m a x , m i n d i s t=m i n d i s t a n c e )
maxima acc = p e a k u t i l s . i n d e x e s ( a c c p l −np . a v e r a g e ( a c c p l ) ,
t h r e s=t h r e s h o l d m a x , m i n d i s t=m i n d i s t a n c e )
maxima z = p e a k u t i l s . i n d e x e s ( z−np . a v e r a g e ( z ) ,
t h r e s=t h r e s h o l d m a x , m i n d i s t=m i n d i s t a n c e )
maxima dz = p e a k u t i l s . i n d e x e s ( dz−np . a v e r a g e ( dz ) ,
t h r e s=t h r e s h o l d m a x , m i n d i s t=m i n d i s t a n c e )
maxima ddz = p e a k u t i l s . i n d e x e s ( ddz−np . a v e r a g e ( ddz ) ,
t h r e s=t h r e s h o l d m a x , m i n d i s t=m i n d i s t a n c e )
maxima ry = p e a k u t i l s . i n d e x e s ( ry−np . a v e r a g e ( ry ) ,
t h r e s=t h r e s h o l d m a x , m i n d i s t=m i n d i s t a n c e )
74 APPENDIX C

maxima dry = p e a k u t i l s . i n d e x e s ( dry−np . a v e r a g e ( dry ) ,

t h r e s=t h r e s h o l d m a x , m i n d i s t=m i n d i s t a n c e )
maxima ddry = p e a k u t i l s . i n d e x e s ( ddry−np . a v e r a g e ( ddry ) ,
t h r e s=t h r e s h o l d m a x , m i n d i s t=m i n d i s t a n c e )
fig = plt . figure ()
plt . subplot (4 ,2 ,1)
p l t . p l o t ( t ∗math . s q r t ( 3 5 ) , ( v e l p l ) ∗ math . s q r t ( 3 5 ) ) ;
p l t . p l o t ( t [ maxima vel ] ∗ math . s q r t ( 3 5 ) , ( v e l p l [ maxima vel ] ) ∗ math . s q r t ( 3 5 ) ,
’ rx ’ ) ; p l t . t i t l e ( ’ v e l ’ )
plt . subplot (4 ,2 ,2)
p l t . p l o t ( t ∗math . s q r t ( 3 5 ) , a c c p l )
p l t . p l o t ( t [ maxima acc ] ∗ math . s q r t ( 3 5 ) , a c c p l [ maxima acc ] , ’ rx ’ )
p l t . t i t l e ( ’ acc ’ )
plt . subplot (4 ,2 ,3)
p l t . p l o t ( t ∗math . s q r t ( 3 5 ) , ( z ) ∗ 3 5 )
p l t . p l o t ( t [ maxima z ] ∗ math . s q r t ( 3 5 ) , ( z [ maxima z ] ) ∗ 3 5 , ’ rx ’ )
plt . t i t l e ( ’z ’ )
plt . subplot (4 ,2 ,4)
p l t . p l o t ( t ∗math . s q r t ( 3 5 ) , ( dz ) ∗ math . s q r t ( 3 5 ) )
p l t . p l o t ( t [ maxima dz ] ∗ math . s q r t ( 3 5 ) , ( dz [ maxima dz ] ) ∗ math . s q r t ( 3 5 ) , ’ rx ’ )
p l t . t i t l e ( ’ dz ’ )
plt . subplot (4 ,2 ,5)
p l t . p l o t ( t ∗math . s q r t ( 3 5 ) , ddz )
p l t . p l o t ( t [ maxima ddz ] ∗ math . s q r t ( 3 5 ) , ddz [ maxima ddz ] , ’ rx ’ )
p l t . t i t l e ( ’ ddz ’ )
plt . subplot (4 ,2 ,6)
p l t . p l o t ( t ∗math . s q r t ( 3 5 ) , ( ry ) ∗ 3 5 )
p l t . p l o t ( t [ maxima ry ] ∗ math . s q r t ( 3 5 ) , ( ry [ maxima ry ] ) ∗ 3 5 , ’ rx ’ )
p l t . t i t l e ( ’ ry ’ )
plt . subplot (4 ,2 ,7)
p l t . p l o t ( t ∗math . s q r t ( 3 5 ) , ( dry ) ∗ math . s q r t ( 3 5 ) )
p l t . p l o t ( t [ maxima dry ] ∗ math . s q r t ( 3 5 ) , ( dry [ maxima dry ] ) ∗ math . s q r t ( 3 5 ) , ’ rx ’ )
p l t . t i t l e ( ’ dry ’ )
plt . subplot (4 ,2 ,8)
p l t . p l o t ( t ∗math . s q r t ( 3 5 ) , ddry )
p l t . p l o t ( t [ maxima ddry ] ∗ math . s q r t ( 3 5 ) , ddry [ maxima ddry ] , ’ rx ’ )
p l t . t i t l e ( ’ ddry ’ )
f i g . t i g h t l a y o u t ( )#; f i g . s u b p l o t s a d j u s t ( t o p =0.88)
f i g . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
Check Peaks / Check H%s T%s . png ’ %(H wave , T wave ) )

max vel . append ( np . a v e r a g e ( v e l p l [ maxima vel ]) − np . a v e r a g e ( v e l p l ) )

max acc . append ( np . a v e r a g e ( a c c p l [ maxima acc ]) − np . a v e r a g e ( a c c p l ) )
max z . append ( np . a v e r a g e ( z [ maxima z ]) − np . a v e r a g e ( z ) )
max dz . append ( np . a v e r a g e ( dz [ maxima dz ]) − np . a v e r a g e ( dz ) )
max ddz . append ( np . a v e r a g e ( ddz [ maxima ddz ]) − np . a v e r a g e ( ddz ) )
max ry . append ( np . a v e r a g e ( ry [ maxima ry ]) − np . a v e r a g e ( ry ) )
max dry . append ( np . a v e r a g e ( dry [ maxima dry ]) − np . a v e r a g e ( dry ) )
max ddry . append ( np . a v e r a g e ( ddry [ maxima ddry ]) − np . a v e r a g e ( ddry ) )
APPENDIX C 75

print ( ’Run %s , H=%d T=%d . . . T:%d B:%d D:%d V:%d A:%d Z:%d RY:%d ’ \
%(s +1 ,H wave , T wave , len ( maxima top ) , len ( maxima btm ) ,
len ( m a x i m a d i f f ) , len ( maxima vel ) , len ( maxima acc ) ,
len ( maxima z ) , len ( maxima ry ) ) )

s f = 35 # l e n g t h s c a l e f a c t o r
m a x d i f f = [ m a x d i f f [ i ] ∗ s f ∗ 9 . 8 1 f o r i in range ( len ( m a x d i f f ) ) ]
max top = [ max top [ i ] ∗ s f ∗ 9 . 8 1 f o r i in range ( len ( max top ) ) ]
max btm = [ max btm [ i ] ∗ s f ∗ 9 . 8 1 f o r i in range ( len ( max btm ) ) ]
max vel = [ max vel [ i ] ∗ math . s q r t ( s f ) f o r i in range ( len ( max vel ) ) ]
max acc = [ max acc [ i ] f o r i in range ( len ( max acc ) ) ]
max z = [ max z [ i ] ∗ s f f o r i in range ( len ( max z ) ) ]
max dz = [ max dz [ i ] ∗ math . s q r t ( s f ) f o r i in range ( len ( max dz ) ) ]
max ddz = [ max ddz [ i ] f o r i in range ( len ( max ddz ) ) ]
max ry = [ max ry [ i ] ∗ 1 8 0 / math . p i f o r i in range ( len ( max z ) ) ]
max dry = [ max dry [ i ] ∗ 1 8 0 / math . p i ∗math . s q r t ( s f ) f o r i in range ( len ( max z ) ) ]
max ddry= [ max ddry [ i ] ∗ 1 8 0 / math . p i f o r i in range ( len ( max z ) ) ]

plt . figure ()
p l t . t i t l e ( ’ P r e s s u r e d i f f e r e n c e ( Bottom−Top ) ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , m a x d i f f [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , m a x d i f f [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , m a x d i f f [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
p l t . y l a b e l ( ’ $ \ D e l t a P$ $KN/mˆ2 $ ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / P r e s s u r e d i f f a v g ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
Pressures freq / Pressure diff avg ’ )

plt . figure ()
p l t . t i t l e ( ’ Top s u r f a c e p r e s s u r e ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max top [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max top [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max top [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
p l t . y l a b e l ( ’P $KN/mˆ2 $ ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / P r e s s u r e t o p a v g ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
Pressures freq / Pressure top avg ’ )
76 APPENDIX C

plt . figure ()
p l t . t i t l e ( ’ Bottom s u r f a c e p r e s s u r e ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max btm [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max btm [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max btm [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
p l t . y l a b e l ( ’P $KN/mˆ2 $ ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / P r e s s u r e b t m a v g ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
P r e s s u r e s f r e q / Pressure btm avg ’ )

plt . figure ()
plt . t i t l e ( ’ Plate V e l o c i t i e s ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max vel [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max vel [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max vel [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
p l t . y l a b e l ( ’V $m/ s $ ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / V e l a v g ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
Pressures freq / Vel plate avg ’ )

plt . figure ()
plt . t i t l e ( ’ Plate Accelerations ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max acc [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max acc [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max acc [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
p l t . y l a b e l ( ’A $m/ s ˆ2 $ ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / Acc avg ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
Pressures freq / Acc plate avg ’ )

max z [ 0 : 9 ] = [ max z [ i ] / 1 f o r i in range ( 0 , 9 ) ]

max z [ 9 : 1 7 ] = [ max z [ i ] / 2 f o r i in range ( 9 , 1 7 ) ]
max z [ 1 9 : 2 6 ] = [ max z [ i ] / 3 . 5 f o r i in range ( 1 9 , 2 6 ) ]
APPENDIX C 77

max ry [ 0 : 9 ] = [ max ry [ i ] / 1 f o r i in range ( 0 , 9 ) ]

max ry [ 9 : 1 7 ] = [ max ry [ i ] / 2 f o r i in range ( 9 , 1 7 ) ]
max ry [ 1 9 : 2 6 ] = [ max ry [ i ] / 3 . 5 f o r i in range ( 1 9 , 2 6 ) ]
max dz [ 0 : 9 ] = [ max dz [ i ] / 1 f o r i in range ( 0 , 9 ) ]
max dz [ 9 : 1 7 ] = [ max dz [ i ] / 2 f o r i in range ( 9 , 1 7 ) ]
max dz [ 1 9 : 2 6 ] = [ max dz [ i ] / 3 . 5 f o r i in range ( 1 9 , 2 6 ) ]
max ddz [ 0 : 9 ] = [ max ddz [ i ] / 1 f o r i in range ( 0 , 9 ) ]
max ddz [ 9 : 1 7 ] = [ max ddz [ i ] / 2 f o r i in range ( 9 , 1 7 ) ]
max ddz [ 1 9 : 2 6 ] = [ max ddz [ i ] / 3 . 5 f o r i in range ( 1 9 , 2 6 ) ]
max dry [ 0 : 9 ] = [ max dry [ i ] / 1 f o r i in range ( 0 , 9 ) ]
max dry [ 9 : 1 7 ] = [ max dry [ i ] / 2 f o r i in range ( 9 , 1 7 ) ]
max dry [ 1 9 : 2 6 ] = [ max dry [ i ] / 3 . 5 f o r i in range ( 1 9 , 2 6 ) ]
max ddry [ 0 : 9 ] = [ max ddry [ i ] / 1 f o r i in range ( 0 , 9 ) ]
max ddry [ 9 : 1 7 ] = [ max ddry [ i ] / 2 f o r i in range ( 9 , 1 7 ) ]
max ddry [ 1 9 : 2 6 ] = [ max ddry [ i ] / 3 . 5 f o r i in range ( 1 9 , 2 6 ) ]
rao z= np . p o l y f i t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max z [ 0 : 9 ] , 3 )
rao r y= np . p o l y f i t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max ry [ 0 : 9 ] , 3 )
rao z = np . po ly 1d ( r a o z )
rao r y = np . pol y1d ( r a o r y )

d a t a r a o p = np . g e n f r o m t x t ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / \
Workspace / Python /00 p l o t d a t a / RAO pitch2 . t x t ’ ,
s k i p h e a d e r =1, u s e c o l s = ( 0 , 1 ) )
d a t a r a o h = np . g e n f r o m t x t ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / \
Workspace / Python /00 p l o t d a t a /RAO heave2 . t x t ’ ,
s k i p h e a d e r =1, u s e c o l s = ( 0 , 1 ) )

plt . figure ()
p l t . t i t l e ( ’RAO heave ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max z [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max z [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max z [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
#p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , r a o z ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] ) , ’ r ’ , l i n e w i d t h =.5)
p l t . p l o t ( d a t a r a o h [ : , 0 ] , d a t a r a o h [ : , 1 ] , ’ dimgrey ’ , l i n e w i d t h = . 7 ,
l a b e l = ’ Simulation ’ )
p l t . yl im ( 0 , 3 . 5 )
p l t . y l a b e l ( ’ $ \ e t a 3 $ m/m ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q /RAO heave ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
P r e s s u r e s f r e q / RAO heave tests ’ )

plt . figure ()
p l t . t i t l e ( ’RAO p i t c h ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max ry [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
78 APPENDIX C

l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max ry [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max ry [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
#p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , r a o r y ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] ) , ’ r ’ , l i n e w i d t h =.6)
p l t . p l o t ( d a t a r a o p [ : , 0 ] , d a t a r a o p [ : , 1 ] , ’ dimgrey ’ , l i n e w i d t h = . 7 ,
l a b e l = ’ Simulation ’ )
p l t . ylim ( 0 , 6 )
p l t . y l a b e l ( ’ $ \ e t a 5 $ deg /m ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / RAO pitch ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
P r e s s u r e s f r e q / RAO pitch tests ’ )

plt . figure ()
p l t . t i t l e ( ’RAO v e l heave ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max dz [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max dz [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max dz [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
p l t . y l a b e l ( ’ $ \ dot \ e t a 3 $ (m/ s ) /m’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / RAO vel heave ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
P r e s s u r e s f r e q / RAO vel heave ’ )

plt . figure ()
p l t . t i t l e ( ’RAO a c c heave ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max ddz [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max ddz [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max ddz [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
p l t . y l a b e l ( ’ $ \ ddot \ e t a 3 $ $ (m/ s ˆ 2 ) /m$ ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / RAO acc heave ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
P r e s s u r e s f r e q / RAO acc heave ’ )

plt . figure ()
p l t . t i t l e ( ’RAO v e l p i t c h ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max dry [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
APPENDIX C 79

l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max dry [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max dry [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
p l t . y l a b e l ( ’ $ \ dot \ e t a 5 $ ( deg / s ) /m ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / R A O v e l p i t c h ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
P r e s s u r e s f r e q / RAO vel pitch ’ )

plt . figure ()
p l t . t i t l e ( ’RAO a c c p i t c h ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 0 : 9 , 1 ] , max ddry [ 0 : 9 ] , ’ bx ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Low ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 9 : 1 7 , 1 ] , max ddry [ 9 : 1 7 ] , ’ r ˆ ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = Medium ’ )
p l t . p l o t ( t e s t r u n s 2 9 T [ 1 9 : 2 6 , 1 ] , max ddry [ 1 9 : 2 6 ] , ’ gv ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’H = High ’ )
p l t . y l a b e l ( ’ $ \ ddot \ e t a 5 $ $ ( deg / s ˆ 2 ) /m$ ’ ) ; p l t . x l a b e l ( ’Wave p e r i o d T ( s ) ’ )
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / RAO acc pitch ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
P r e s s u r e s f r e q / RAO acc pitch ’ )

max acc norm = max acc /max( max acc )

max vel norm = max vel /max( max vel )
m a x d i f f n o r m = m a x d i f f /max( m a x d i f f )

s acc , i acc , r acc , , = l i n r e g r e s s ( max acc norm , m a x d i f f n o r m )

s vel , i vel , r vel , , = l i n r e g r e s s ( max vel norm , m a x d i f f n o r m )
reg acc = [ s a c c ∗ max acc norm [ i ]+ i a c c for i in range ( 2 7 ) ]
reg vel = [ s v e l ∗ max vel norm [ i ]+ i v e l for i in range ( 2 7 ) ]

f i g = p l t . f i g u r e ( ) ; ax = f i g . a d d s u b p l o t ( 1 1 1 )
p l t . t i t l e ( ’ C o r r e l a t i o n Net P r e s s u r e vs . P l a t e A c c e l e r a t i o n s ’ )
p l t . p l o t ( max acc norm , m a x d i f f n o r m , ’ b . ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’ Experimental ’ )
p l t . p l o t ( max acc norm , r e g a c c , ’ k ’ , l a b e l = ’ L i n e a r R e g r e s s i o n ’ ,
linewidth = 0.5)
p l t . t e x t ( 0 . 9 , 0 . 3 , ’ r = %.3 f ’%r a c c ,
v e r t i c a l a l i g n m e n t= ’ bottom ’ , h o r i z o n t a l a l i g n m e n t= ’ r i g h t ’ ,
t r a n s f o r m= ax . transAxes , f o n t s i z e =12 , bbox={ ’ f a c e c o l o r ’ : ’ dimgrey ’ ,
’ a l p h a ’ : 0 . 2 , ’ pad ’ : 6 } )
p l t . y l a b e l ( ’ $ \ D e l t a P$ n o r m a l i z e d ’ ) ; p l t . x l a b e l ( ’ $Acc { p l } $ n o r m a l i z e d ’ )
plt . legend ()
80 APPENDIX C

#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / RAO acc pitch ’ )

p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
Pressures freq / Corr Pdiff Acc ’ )

f i g = p l t . f i g u r e ( ) ; ax = f i g . a d d s u b p l o t ( 1 1 1 )
p l t . t i t l e ( ’ C o r r e l a t i o n Net P r e s s u r e vs . P l a t e V e l o c i t i e s ’ )
p l t . p l o t ( max vel norm , m a x d i f f n o r m , ’ b . ’ , f i l l s t y l e = ’ none ’ ,
l a b e l = ’ Experimental ’ )
p l t . p l o t ( max vel norm , r e g v e l , ’ k ’ , l a b e l = ’ L i n e a r R e g r e s s i o n ’ ,
linewidth = 0.5)
p l t . t e x t ( 0 . 9 , 0 . 3 , ’ r = %.3 f ’%r v e l ,
v e r t i c a l a l i g n m e n t= ’ bottom ’ , h o r i z o n t a l a l i g n m e n t= ’ r i g h t ’ ,
t r a n s f o r m= ax . transAxes , f o n t s i z e =12 , bbox={ ’ f a c e c o l o r ’ : ’ dimgrey ’ ,
’ a l p h a ’ : 0 . 2 , ’ pad ’ : 6 } )
p l t . y l a b e l ( ’ $ \ D e l t a P$ n o r m a l i z e d ’ ) ; p l t . x l a b e l ( ’ $ V e l { p l } $ n o r m a l i z e d ’ )
plt . legend ()
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / P r e s s u r e s f r e q / RAO acc pitch ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
Pressures freq / Corr Pdiff Vel ’ )
##########
Generate the input into MidasCivil.
# −∗− c o d i n g : u t f −8 −∗−
”””
C r e a t e d on Mon J u l 3 1 6 : 0 2 : 0 7 2017

@author : r o d a l b
”””

import numpy a s np
import m a t p l o t l i b . p y p l o t a s p l t # Para h a c e r p l o t s
import math

plt . close ( ’ all ’ )

e l e m e n t s = np . g e n f r o m t x t ( ’E : /WORKSPACE/ Python /HydroD/ e l e m e n t s . t x t ’ ,

s k i p h e a d e r =1, dtype = i n t ) # elem−node1−node2−node3−node4
nodes = np . g e n f r o m t x t ( ’E : /WORKSPACE/ Python /HydroD/ nodes . t x t ’ ,
s k i p h e a d e r =1) # node − x − y − z (m)
p r e s s u r e s = np . g e n f r o m t x t ( ’E : /WORKSPACE/ Python /HydroD/ p r e s s u r e s . t x t ’ ,
s k i p h e a d e r =0, u s e c o l s = ( 1 , 2 ) ) # elem − p n1 (N/m2)

LC name = ”ExampleLC”
Group name = ” D e f a u l t ”

# Check s i z e s o f i n p u t l i s t s
#p r i n t ( e l e m e n t s . s h a p e )
#p r i n t ( nodes . s h a p e )
#p r i n t ( p r e s s u r e s . s h a p e )
APPENDIX C 81

# I n p u t z v a l u e i n m o f t h e t o p and bottom p l a t e s , and s e t t o l e r a n c e

z t o p = −5.4
zbtm = −6.2
t o l = 1 e−8
# S p l i t nodes f o r t o p and bottom
n o d e s t o p = nodes [ nodes [ : , 3 ] = = ztop , 0 ]
nodes btm = nodes [ nodes [ : , 3 ] = = zbtm , 0 ]
#p r i n t ( n o d e s t o p . s h a p e ) ; p r i n t ( nod es btm . s h a p e )

# C r e a t e l i s t w i t h a v e r a g e z v a l u e s o f t h e 4 nodes f o r each e l e m e n t
zav = [ ]
f o r i in range ( len ( e l e m e n t s ) ) :
zav . append ( np . a v e r a g e ( nodes [ e l e m e n t s [ i , 1 : ] − 1 , 3 ] ) )

# I d e n t i f y p l a t e e l e m e n t s n o t on t h e s i d e s
p l a t e e l e m = [ e l e m e n t s [ i , 0 ] f o r i in range ( len ( e l e m e n t s ) ) i f
math . f a b s ( zav [ i ]− z t o p)<= t o l or math . f a b s ( zav [ i ]−zbtm)<= t o l ]

p p l a t e s = [ [ p r e s s u r e s [ i , 0 ] , p r e s s u r e s [ i , 1 ] / 1 0 0 0 ] f o r i in range ( len ( p r e s s u r e s ) )
i f p r e s s u r e s [ i , 0 ] in p l a t e e l e m ]

c o r r e s p = [ [ i n t ( p p l a t e s [ i ] [ 0 ] ) , i +1] f o r i in range ( len ( p p l a t e s ) ) ]

midas = [ [ e l e m e n t s [ c o r r e s p [ i ] [ 0 ] − 1 ] [ 1 : ] ] f o r i in range ( len ( c o r r e s p ) ) ]

with open ( ” e l e m p r e s s u r e s . t x t ” , ”w” ) a s t e x t f i l e :

f o r i in range ( len ( p l a t e e l e m ) ) :
i f i == 0 :
print ( ’%7s %9s %9s %9s %9s %9s %3s %3s %3s %5s %5s %3s %3s \
%3s %3s %3s %3s %3s %7s ’
%( ’ Element ’ , ’ Loadcase ’ , ’Cmd ’ , ’ Load Type ’ , ’ Edge/ Face ’ ,
’ D i r e c t i o n ’ , ’Vx ’ , ’Vy ’ , ’ Vz ’ , ’ P r o j ’ , ’PU ’ , ’ P1 ’ , ’ P2 ’ ,
’ P3 ’ , ’ P4 ’ , ’EPU ’ , ’EP1 ’ ,
’EP2 ’ , ’ Group ’ ) , f i l e= t e x t f i l e )
print ( ’%7d %9s %9s %9s %9s %9s %3d %3d %3d %5s %.3 f %3d %3d %3d \
%3d %3d %3d %3d %7s ’
%( c o r r e s p [ i ] [ 1 ] , LC name , ’ P r e s s u r e ’ , ’ P r e s s u r e ’ , ’ Face #1 ’ ,
’ L o c a l z ’ , 0 , 0 , 0 , ’No ’ , p p l a t e s [ i ] [ 1 ] , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
Group name ) , f i l e= t e x t f i l e )
# i f i ==0:

with open ( ” m i d a s i n p u t . t x t ” , ”w” ) a s t e x t f i l e :

print ( ”%7s %7s %10s %18s %9s %9s %6s %6s %6s %6s %6s %6s %6s %6s %6s ” \
%(” Element ” , ”Type” ,
” Wall type ” , ” Subtype ” , ” M a t e r i a l ” , ” P r o p e r t y ” , ” Angle ” , ”Node 1 ” ,
”Node 2 ” , ”Node 3 ” , ”Node 4 ” , ”Node 5 ” , ”Node 6 ” , ”Node 7 ” , ”Node 8 ” ) ,
f i l e= t e x t f i l e )
82 APPENDIX C

f o r i in range ( len ( midas ) ) :

print ( ”%7d %7s %10s %18s %9d %9d %6d %6d %6d %6d %6d %6d %6d %6d %6d”
% ( c o r r e s p [ i ] [ 1 ] , ”PLATE” , ” ” , ” Thick (w/ o D r i l l ) ” , 1 , 1 , 0 ,
midas [ i ] [ 0 ] [ 0 ] , midas [ i ] [ 0 ] [ 1 ] , midas [ i ] [ 0 ] [ 2 ] ,
midas [ i ] [ 0 ] [ 3 ] , 0 , 0 , 0 , 0 ) , f i l e= t e x t f i l e )
##########
Filter signals using Butterworth LowPass
# −∗− c o d i n g : u t f −8 −∗−
”””
C r e a t e d on Tue J u l 25 0 9 : 0 8 : 5 9 2017

@author : A l b e r t o
”””

import numpy a s np
from t i m e d e r i v a t i v e import d e r i v
import m a t p l o t l i b . p y p l o t a s p l t # Para h a c e r p l o t s
from s c i p y import s i g n a l

plt . close ( ’ all ’ )

c u t b e g i n n i n g = 30000 #30500 #minimum 12

c u t e n d = 30000 #30500

#d a t a = np . g e n f r o m t x t ( ’C: / Users / A l b e r t o / Dropbox / T e s i s S a i t e c / Python /IHC/ \

#WindWaves 29T/H4 T16/FIH17−00177 C1 2ND H4p0 T16p0 \
#D21p0 A0p0 WDI11p5 ABS 00 QTM . t x t ’ ,
# skip header = cut beginning , s k i p f o o t e r = cut end )
data = np . g e n f r o m t x t ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python /IHC/ \
WindWaves 29T/H2 T16/FIH17 −00177 C1 2ND H2p0 T16p0 \
D21p0 A0p0 WDI11p5 ABS 02 QTM . t x t ’ ,
skip header = cut beginning , s k i p f o o t e r = cut end )
t = data [ : , 0 ]
z = data [ : , 3 ] − np . a v e r a g e ( data [ : , 3 ] )

s t = i n t ( 1 / 5 ∗ len ( t ) )
f n = i n t ( 4 / 5 ∗ len ( t ) )

dz = d e r i v ( t , z )
ddz = d e r i v ( t , dz )

b , a = signal . butter (3 ,0.03)

z f = signal . f i l t f i l t (b , a , z )
dz f = deriv ( t , z f )
dz f = signal . f i l t f i l t (b , a , dz f )
ddz f = deriv ( t , dz f )
ddz f = signal . f i l t f i l t (b , a , ddz f )
APPENDIX C 83

plt . figure ()
p l t . p l o t ( t [ s t : f n ] , z [ s t : f n ] /max( np . abs ( z [ s t : f n ] ) ) , ’ g ’ , t [ s t : f n ] , dz [ s t : f n ] / \
max( np . abs ( dz [ s t : f n ] ) ) , ’ r ’ ,
t [ s t : f n ] , ddz [ s t : f n ] /max( np . abs ( ddz [ s t : f n ] ) ) , ’ b ’ )
p l t . x l a b e l ( ’ Time ( s ) ’ ) ;
#p l t . l e g e n d ( ( ’ D i s p l $\ e t a 3 $ ’ , ’ Vel $\ d o t \ e t a 3 $ ’ , ’ Acc $\ d d o t \ e t a 3 $ ’ ) , l o c =1)
p l t . l e g e n d ( ( ’Norm . D i s p l $ \ e t a 3 $ ’ , ’Norm . Vel $ \ dot \ e t a 3 $ ’ ,
’Norm . Acc$ \ ddot \ e t a 3 $ ’ ) ,
b b o x t o a n c h o r = ( 0 . , 1 . 0 2 , 1 . , . 1 0 2 ) , l o c =8, n c o l =3 ,
mode=” expand ” , b o r d e r a x e s p a d =0. , f o n t s i z e = 1 0 )
#p l t . t i t l e ( ’ U n f i l t e r e d s i g n a l ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / Images / \
S i g n a l s / U n f i l t e r e d . png ’ )

plt . figure ()
p l t . p l o t ( t [ s t : f n ] , z f [ s t : f n ] /max( np . abs ( z f [ s t : f n ] ) ) , ’ g ’ , t [ s t : f n ] ,
d z f [ s t : f n ] /max( np . abs ( d z f [ s t : f n ] ) ) , ’ r ’ ,
t [ s t : f n ] , d d z f [ s t : f n ] /max( np . abs ( d d z f [ s t : f n ] ) ) , ’ b ’ )
p l t . x l a b e l ( ’ Time ( s ) ’ ) ;
#p l t . l e g e n d ( ( ’ D i s p l $\ e t a 3 $ ’ , ’ Vel $\ d o t \ e t a 3 $ ’ , ’ Acc $\ d d o t \ e t a 3 $ ’ ) , l o c =1)
p l t . l e g e n d ( ( ’Norm . D i s p l $ \ e t a 3 $ ’ , ’Norm . Vel $ \ dot \ e t a 3 $ ’ ,
’Norm . Acc $ \ ddot \ e t a 3 $ ’ ) ,
b b o x t o a n c h o r = ( 0 . , 1 . 0 2 , 1 . , . 1 0 2 ) , l o c =8, n c o l =3 ,
mode=” expand ” , b o r d e r a x e s p a d =0. , f o n t s i z e = 1 0 )
#p l t . t i t l e ( ’ F i l t e r e d s i g n a l ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Python / \
Images / S i g n a l s / F i l t e r e d . png ’ )
##########
Plot the Response Amplitude Operator or RAOs.
”””
@author : A l b e r t o R o d r i g u e z
”””

import numpy a s np # vectores y matrices

import m a t p l o t l i b . p y p l o t a s p l t # Para h a c e r p l o t s
#i m p o r t math # numero p i y o t r o s
#from m a t p l o t l i b . o f f s e t b o x i m p o r t AnchoredText # Anclar t e x t o s en p l o t s
#from s c i p y . i n t e r p o l a t e i m p o r t s p l i n e # Interpolar

show x ticks = False

plt . close ( ’ all ’ )
#d a t a = np . g e n f r o m t x t ( ’E: / 5 FASTmodels/ Mod Bladed 2MW onshore /Bladed2MW . o u t ’ ,
#s k i p h e a d e r =8)
d a t a p = np . g e n f r o m t x t ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Workspace / \
Python /00 p l o t d a t a / RAO pitch2 . t x t ’ ,
s k i p h e a d e r =1)
#d a t a r = np . g e n f r o m t x t ( ’E: /WORKSPACE/ Python /00 p l o t d a t a / c o n v R A O r o l l . t x t ’ ,
84 APPENDIX C

# s k i p h e a d e r =1)
d a t a h = np . g e n f r o m t x t ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Workspace / \
Python /00 p l o t d a t a /RAO heave2 . t x t ’ ,
s k i p h e a d e r =1)

#f i g = p l t . f i g u r e ( f i g s i z e =(15 ,8))

x p = data p [ : , 0 ] ;
#x r = d a t a r [ : , 0 ] ;
x h = data h [ : , 0 ] ;
#x = np . c o l u m n s t a c k ( [ x p , x r , x h ] ) ;

#f i g , a x a r r = p l t . s u b p l o t s ( 3 , 1 )
#f o r j i n r an g e ( 1 , 3 ) :
# f o r i i n ra n g e ( 1 , 5 ) :
# axarr [ 0 ] . p l o t ( x [ : , j ] , data p [ : , i ] )
# axarr [ 1 ] . p l o t ( x [ : , j ] , data r [ : , i ] )
# axarr [ 2 ] . p l o t ( x [ : , j ] , data h [ : , i ] )
# a x a r r [ 0 ] . s e t t i t l e ( ’RAO f o r P i t c h ’ )
# a x a r r [ 1 ] . s e t t i t l e ( ’RAO f o r R o l l ’ )
# a x a r r [ 2 ] . s e t t i t l e ( ’RAO f o r Heave ’ )

#
##p l t . t i t l e ( ’RAO f o r P i t c h ’ )
##p l t . l e g e n d ( prop ={’ s i z e ’ : 8 } ) # l o c =3
##p l t . x l i m ( ( 0 , 2 ∗ math . p i ) )
#
#f i g . s u b p l o t s a d j u s t ( h s p a c e =3) # Es pa i v e r t i c a l e n t r e s u b p l o t s
#f i g . t i g h t l a y o u t ( ) # Reduir e l s marges d e l s p l o t s amb l a f i g u r a
#p l t . show ( )

fig2 = plt . figure ( f i g s i z e =(15 ,8));

plt . p l o t ( x p , d a t a p [ : , 1 ] , ’ b ’ , l a b e l = ’ Seed 0 . 4 m ’ ) #p l o t i n b l u e
plt . x l a b e l ( ’ Period , [ s ] ’ , f o n t s i z e = 1 5 )
plt . y l a b e l ( r ’ $ R e l a t i v e \ ; motion \ quad \ f r a c {\ e t a 5 }{\ z e t a a } \ quad [ deg /m] $ ’ ,
f o n t s i z e = 15)
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
#p l t . l e g e n d ( l o c =2) #prop ={’ s i z e ’ : 8 } ,
p l t . t i t l e ( ’ Pitch ’ , f o n t s i z e = 15)
p l t . g r i d ( b=True , which= ’ major ’ , c o l o r= ’ l i g h t g r e y ’ , l i n e s t y l e= ’−− ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Images / P l o t s / RAO pitch . png ’ ,
b b o x i n c h e s= ’ t i g h t ’ )
p l t . s a v e f i g ( ’E : /WORKSPACE/ Python / Images / RAO pitch . png ’ , b b o x i n c h e s= ’ t i g h t ’ )

#f i g 3 = p l t . f i g u r e ( f i g s i z e =(15 ,8));
#p l t . p l o t ( x [ : , 1 ] , d a t a r [ : , 4 ] , ’ b ’ , l a b e l = ’ Seed 0 . 4 m ’ ) #p l o t i n b l u e
#p l t . x l a b e l ( ’ Period , [ s ] ’ , f o n t s i z e = 15)
#p l t . y l a b e l ( r ’ $ R e l a t i v e \ ; motion \ quad \ f r a c {\ e t a 4 }{\ z e t a a } \ quad [ deg /m] $ ’ ,
# f o n t s i z e = 15)
APPENDIX C 85

#p l t . l e g e n d ( l o c =2) #prop ={’ s i z e ’ : 8 } ,

#p l t . t i t l e ( ’ R o l l ’ , f o n t s i z e = 15)
#p l t . g r i d ( b=True , which =’ major ’ , c o l o r =’ l i g h t g r e y ’ , l i n e s t y l e =’−−’)
#p l t . s a v e f i g ( ’C: / Users / r o d a l b / Dropbox / T e s i s S a i t e c / Images / P l o t s / RAO roll . png ’ ,
# b b o x i n c h e s =’ t i g h t ’ )
#p l t . s a v e f i g ( ’E: /WORKSPACE/ Python / Images / RAO roll . png ’ , b b o x i n c h e s =’ t i g h t ’ )

fig4 = plt . figure ( f i g s i z e =(15 ,8));

plt . p l o t ( x h , d a t a h [ : , 1 ] , ’ b ’ , l a b e l = ’ Seed 0 . 4 m ’ ) #p l o t i n b l u e
plt . x l a b e l ( ’ Period , [ s ] ’ , f o n t s i z e = 1 5 )
plt . y l a b e l ( r ’ $ R e l a t i v e \ ; motion \ quad \ f r a c {\ e t a 3 }{\ z e t a a } \ quad [m/m] $ ’ ,
f o n t s i z e = 15)
i f s h o w x t i c k s == F a l s e : p l t . x t i c k s ( [ ] )
#p l t . l e g e n d ( l o c =2) #prop ={’ s i z e ’ : 8 } ,
p l t . t i t l e ( ’ Heave ’ , f o n t s i z e = 1 5 )
p l t . g r i d ( b=True , which= ’ major ’ , c o l o r= ’ l i g h t g r e y ’ , l i n e s t y l e= ’−− ’ )
p l t . s a v e f i g ( ’C: / U s e r s / r o d a l b / Dropbox / T e s i s S a i t e c / Images / P l o t s /RAO heave . png ’ ,
b b o x i n c h e s= ’ t i g h t ’ )
p l t . s a v e f i g ( ’E : /WORKSPACE/ Python / Images /RAO heave . png ’ , b b o x i n c h e s= ’ t i g h t ’ )
p l t . show ( )

#p l t . p l o t ( x , y2 , ’ r ’ ) #p l o t x , y1 i n r e d
#p l t . p l o t ( x , y3 , ’ g ’ ) #p l o t x , y1 i n g r e e n
#p l t . p l o t ( x , y4 , ’ y ’ ) #p l o t x , y1 i n y e l l o w
Appendix D

Definition of the Numerical Models.

The hydrodynamic model defined in HydroD required a set of plate properties for the plate
elements and the different areas of the Morison elements. In table D1, the equivalent tube
diameters are listed.

Table D1: Summary of equivalent tubular sections for the HydroD Morison model.

[m]
ID Eq diameter Description
1 0.94 Upper SPM support
2 0.80 Lower SPM Support
3 0.79 Inclined heave plate support
4 1.14 Cross-beams, hz
5 1.12 Cross-beams, incl
6 0.90 Cross-beams, hz btm (stern)
7 1.14 Transverse tower support, top
8 0.79 Diagonal tower support, top
9 0.79 Inclined tower support
10 0.90 Transverse tower support, btm
11 0.79 Diagonal tower support, btm
12 1.08 Keel
13 0.80 Cross-keel

The structural FE model used in this Master Thesis was defined, as mentioned previously in
the document, in the software MidasCivil. In it, apart from the general geometry and material
characteristics (shown explicitly in 4.6), the plate thickness and its directional correction coef-
ficients needed to be defined as well. For clarity and brevity, a complete list of these is shown
in table D2.

Table D2: Plate thickness defined in the structural FEM.

[m]
ID Name Thickness
1 Thickness08 0.8

86
Appendix E

Results from structural FE model.

In this appendix, some extra results from the FE model are shown. They may help the reader
understand the complete behavior of the plate, even if the main conclusions and comments are
related with the results shown in the main body of the document.

87
88 APPENDIX E

Figure E.1: Vertical deflection line cutting diagram along lineY2, for all load cases.
APPENDIX E 89

Figure E.2: Bending moments Myy along lineY for all load cases.
90 APPENDIX E

Figure E.3: Bending moments Myy along lineY2 for all load cases.
TRITA -BKN. Master Thesis 526
ISSN 1103-4297
ISRN KTH/BKN/EX-526-SE

[Link]