Waves, Motions and Manoeuvring

LECTURE NOTES
for
SJO 745 Wave loads and Seakeeping

based on

Wave Induced Loads and Ship Motions 2010

by Lars Bergdahl
Fartygs Rörelser i Vågor och vid Manövrering 1988
by Gilbert Dyne
SEAKEEPING Ship Behaviour in Rough Weather 1989
by A.R.J.M. Lloyd
Principles of Naval Architecture Vol III 1989
SNAME, Edward V. Lewis Editor

Edited by Carl-Erik Janson

Department of Shipping and Marine technology

CHALMERS UNIVERSITY OF TECHNOLOGY
Göteborg, Sweden 2015-03-18
 CONTENTS

1 INTRODUCTION 1
1.1 Ship Motions in Waves . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Calculation Chain . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 On Wave-Induced Forces . . . . . . . . . . . . . . . . . . . . . . 5

2 POTENTIAL FLOW FOR WAVES 7

2.1 The Equation of Continuity . . . . . . . . . . . . . . . . . . . . . 7
2.2 Irrotational Flow . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 The Velocity Potential and Laplace Differential Equation . . . . . 10
2.4 Bernoullis Equation for Potential Flow . . . . . . . . . . . . . . . 11
2.4.1 Navier-Stokes’ Equations in two dimensions . . . . . . . 11
2.4.2 Deduction of the Bernoulli Equation . . . . . . . . . . . . 12

3 REGULAR WAVES 15
3.1 The Velocity Potential . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Linear Airy Wave Theory . . . . . . . . . . . . . . . . . . . . . . 16
3.3.1 Solution for a progressive wave . . . . . . . . . . . . . . 17
3.4 The Application of the Velocity Potential . . . . . . . . . . . . . 20
3.4.1 Real-valued expressions . . . . . . . . . . . . . . . . . . 21
3.5 Velocity Potential in Deep Water . . . . . . . . . . . . . . . . . . 21
3.5.1 Real-valued expressions . . . . . . . . . . . . . . . . . . 22
3.5.2 Particle paths . . . . . . . . . . . . . . . . . . . . . . . . 23
3.6 Wave Properties in Shallow Water . . . . . . . . . . . . . . . . . 23
3.6.1 Real-valued expressions . . . . . . . . . . . . . . . . . . 24
3.6.2 Summary of Airy waves . . . . . . . . . . . . . . . . . . 24
3.7 “Oblique” Waves . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.8 Wave Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

iii
CONTENTS

3.9 Wave Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.10 Wave Celerity, Group Velocity and Particle Velocities . . . . . . . 29
3.10.1 Dispersion and celerity . . . . . . . . . . . . . . . . . . . 29
3.10.2 Group velocity . . . . . . . . . . . . . . . . . . . . . . . 30
3.10.3 Water particle motion and dynamic pressure . . . . . . . . 32
3.11 Finite Amplitude Waves and Higher-Order Waves . . . . . . . . . 32
3.11.1 Stokes’ second order wave in deep water . . . . . . . . . 33
3.11.2 Finite-height shallow water waves . . . . . . . . . . . . . 33
3.11.3 Solitary waves . . . . . . . . . . . . . . . . . . . . . . . 34
3.12 Propagation and Transformation . . . . . . . . . . . . . . . . . . 37
3.12.1 Dispersion and transformation to swell . . . . . . . . . . 37
3.12.2 Shoaling . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.12.3 Refraction . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.12.4 Diffraction and reflection . . . . . . . . . . . . . . . . . . 40
3.12.5 Wave breaking . . . . . . . . . . . . . . . . . . . . . . . 41
3.12.6 Bottom friction dissipation . . . . . . . . . . . . . . . . . 43

4 WIND WAVES 45
4.1 Characteristics of Wind Waves . . . . . . . . . . . . . . . . . . . 46
4.2 Fourier Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2.1 Fourier series . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2.2 Parsevals equation, orthogonality . . . . . . . . . . . . . 50
4.3 Amplitude Spectra and Phase Spectra . . . . . . . . . . . . . . . 51
4.3.1 Wave record . . . . . . . . . . . . . . . . . . . . . . . . 51
4.3.2 Amplitude, variance and energy spectra . . . . . . . . . . 51
4.3.3 Standard spectra . . . . . . . . . . . . . . . . . . . . . . 53
4.3.4 Moments and wave characteristics . . . . . . . . . . . . . 58
4.4 Synthesised Waves . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.4.1 Directional sea . . . . . . . . . . . . . . . . . . . . . . . 64

5 WAVE STATISTICS AND PROBABILITY 67

5.1 The Concept of Risk . . . . . . . . . . . . . . . . . . . . . . . . 68
5.1.1 Design event at given risk . . . . . . . . . . . . . . . . . 68
5.2 Design Approaches . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.2.1 Design with safety factors . . . . . . . . . . . . . . . . . 70
5.2.2 Safety format with load and material coefficients . . . . . 70
5.3 Short Term Statistics . . . . . . . . . . . . . . . . . . . . . . . . 73
5.3.1 The distribution of one harmonic component . . . . . . . 73
5.3.2 The distribution of the water level in an irregular wave . . 75
5.4 Extremes in a Sea State . . . . . . . . . . . . . . . . . . . . . . . 76
5.4.1 Extreme wave height . . . . . . . . . . . . . . . . . . . . 77
5.5 Long Term Wave Statistics . . . . . . . . . . . . . . . . . . . . . 78
5.5.1 Wave statistics . . . . . . . . . . . . . . . . . . . . . . . 78
5.5.2 Probability distributions of significant wave heights . . . . 79

iv
 CONTENTS

5.5.3 Probability distributions of individual wave heights . . . . 82

5.5.4 Joint probability of wave heights, winds and currents . . . 83
5.5.5 Design wave for a ship’s hull girder . . . . . . . . . . . . 85

6 EQUATIONS OF MOTION FOR A SHIP 87

6.1 COORDINATE SYSTEMS AND DEFINITIONS . . . . . . . . . 87
6.2 SHIP MOTION IN AN EARTH FIXED
COORDINATE SYSTEM . . . . . . . . . . . . . . . . . . . . . 90
6.3 SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM . 91
6.3.1 Translation . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.3.2 Velocity contribution from yaw . . . . . . . . . . . . . . 92
6.3.3 Velocity contribution from pitch . . . . . . . . . . . . . . 93
6.3.4 Velocity contribution from roll . . . . . . . . . . . . . . . 95
6.3.5 Components of acceleration due to yaw . . . . . . . . . . 95
6.3.6 Components of acceleration due to pitch . . . . . . . . . . 97
6.3.7 Components of acceleration due to roll . . . . . . . . . . 98
6.3.8 Accelerations in the x̄, ȳ and z̄-directions . . . . . . . . . . 99
6.3.9 Equations of motion: Forces in the x̄, ȳ and z̄-directions . . 99
6.3.10 Equations of motion: Moments about the x̄, ȳ and z̄-directions101
6.3.11 Equations of motion for arbitrary position of the coordi-
nate system . . . . . . . . . . . . . . . . . . . . . . . . . 105
6.3.12 Equations of motion when the origin of the coordinate
system is at center of gravity . . . . . . . . . . . . . . . . 106
6.3.13 Equations of motion for starboard-port symmetry . . . . . 107
6.3.14 Equations of motion for starboard-port and fore-aft sym-
metry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
6.3.15 Equations of motion for starboard-port symmetry and ori-
gin at center of gravity . . . . . . . . . . . . . . . . . . . 108
6.3.16 Equations of motion for starboard-port, fore-aft symmetry
and origin at center of gravity . . . . . . . . . . . . . . . 109

7 MOTION RESPONSE TO LOADING 111

7.1 Free Vibration of a Floating Ship in Heave . . . . . . . . . . . . . 112
7.2 Response to Constant Loads . . . . . . . . . . . . . . . . . . . . 113
7.3 Response to Harmonic Loads . . . . . . . . . . . . . . . . . . . . 114
7.3.1 Amplitude response function, response amplitude opera-
tor and transfer function . . . . . . . . . . . . . . . . . . 115
7.3.2 Amplification factor . . . . . . . . . . . . . . . . . . . . 116
7.3.3 Complex notation . . . . . . . . . . . . . . . . . . . . . . 116
7.3.4 The transfer function . . . . . . . . . . . . . . . . . . . . 118
7.3.5 Response of a harmonic load on an initially fixed body . . 118
7.4 System with Six Degrees of Freedom . . . . . . . . . . . . . . . 120
7.4.1 Solution in the frequency domain . . . . . . . . . . . . . 120

v
CONTENTS

8 HEADING AND
ENCOUNTER FREQUENCY 123
8.1 HEADING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
8.2 ENCOUNTER FREQUENCY . . . . . . . . . . . . . . . . . . . 123

9 THE SEAKEEPING PROBLEM 129

10 SOLUTION METHODS -
STRIP THEORY 133
10.1 Solution methods in general . . . . . . . . . . . . . . . . . . . . 133
10.2 Strip theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
10.3 Lewis forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
10.4 Experimental methods . . . . . . . . . . . . . . . . . . . . . . . 137

11 SHIP IN REGULAR WAVES 143

11.1 The Two-Dimensional Heave Problem . . . . . . . . . . . . . . . 143
11.1.1 Forces on the two-dimensional ship heaving in calm water 143
11.1.2 Added mass . . . . . . . . . . . . . . . . . . . . . . . . . 144
11.1.3 Radiation damping . . . . . . . . . . . . . . . . . . . . . 145
11.1.4 Two-dimensional added-mass and radiation coefficients
in heave . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
11.1.5 Wave-excited heave force . . . . . . . . . . . . . . . . . 151
11.1.6 Heave motion of the two-dimensional ship-section in reg-
ular beam waves . . . . . . . . . . . . . . . . . . . . . . 153
11.1.7 The sway and roll problem . . . . . . . . . . . . . . . . . 156
11.2 The Uncoupled Three-Dimensional Heave Problem Using Strip
Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
11.2.1 Strip method for a ship . . . . . . . . . . . . . . . . . . . 156
11.2.2 Forces on the three-dimensional ship heaving in calm water 157
11.2.3 Wave-excited force in heave . . . . . . . . . . . . . . . . 158
11.3 The Pitch Problem Using Strip Theory . . . . . . . . . . . . . . . 160
11.3.1 Forces on the three-dimensional ship pitching in calm water160
11.3.2 Wave-excited moment in pitch . . . . . . . . . . . . . . . 163
11.4 The Roll Problem . . . . . . . . . . . . . . . . . . . . . . . . . . 165
11.4.1 Moments on the two-dimensional body rolling in calm water165
11.4.2 Roll damping . . . . . . . . . . . . . . . . . . . . . . . . 168
11.4.3 Added viscous roll damping . . . . . . . . . . . . . . . . 170
11.4.4 Wave-excited roll moment . . . . . . . . . . . . . . . . . 170
11.4.5 Roll motion . . . . . . . . . . . . . . . . . . . . . . . . . 172
11.5 Forward-Speed Effect . . . . . . . . . . . . . . . . . . . . . . . . 172
11.6 Resonance Frequencies . . . . . . . . . . . . . . . . . . . . . . . 175
11.7 Derived Responses . . . . . . . . . . . . . . . . . . . . . . . . . 176
11.7.1 The motion at a point . . . . . . . . . . . . . . . . . . . . 176
11.7.2 Acceleration . . . . . . . . . . . . . . . . . . . . . . . . 177

vi
 11.7.3 Green water . . . . . . . . . . . . . . . . . . . . . . . . . 177
11.7.4 Propeller emergence and risk for slamming . . . . . . . . 179
11.7.5 Slamming . . . . . . . . . . . . . . . . . . . . . . . . . . 180
11.8 Coupled Linear Pitch and Heave Motion at Forward Speed . . . . 183
11.8.1 The coupling between heave and pitch . . . . . . . . . . . 184
11.9 Coupled Pitch and Heave Motion at Zero Speed Including Non-
Linear Viscous Damping . . . . . . . . . . . . . . . . . . . . . . 185
11.9.1 The wave . . . . . . . . . . . . . . . . . . . . . . . . . . 185
11.9.2 Heave . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
11.9.3 Pitch . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
11.9.4 Equations of motion . . . . . . . . . . . . . . . . . . . . 186
11.10Equivalent Linearised Drag Damping . . . . . . . . . . . . . . . 188

12 SHIP IN IRREGULAR WAVES 191

12.1 The Heave Motion in Irregular Waves . . . . . . . . . . . . . . . 191
12.1.1 Heave transfer function . . . . . . . . . . . . . . . . . . . 191
12.1.2 Heave motion in the time domain . . . . . . . . . . . . . 193
12.1.3 Forming the heave motion spectrum . . . . . . . . . . . . 193
12.2 The Pitch Response . . . . . . . . . . . . . . . . . . . . . . . . . 196
12.3 The Response of Vertical Motion at Station x . . . . . . . . . . . 197

13 DERIVED RESPONSES 201

13.1 The Motion of a Point on a Body . . . . . . . . . . . . . . . . . . 201
13.2 Green Water on Deck . . . . . . . . . . . . . . . . . . . . . . . . 202
13.2.1 Probability of green water on deck . . . . . . . . . . . . . 203
13.3 Propeller emergence . . . . . . . . . . . . . . . . . . . . . . . . 203
13.3.1 Probability of propeller emergence . . . . . . . . . . . . . 203
13.4 Slamming, Whipping, Springing . . . . . . . . . . . . . . . . . . 204
13.4.1 Probability of Slamming . . . . . . . . . . . . . . . . . . 204
13.5 Passenger and Crew . . . . . . . . . . . . . . . . . . . . . . . . . 206
13.5.1 Seasickness . . . . . . . . . . . . . . . . . . . . . . . . . 206
13.5.2 Human performance . . . . . . . . . . . . . . . . . . . . 207
13.5.3 Recommended allowed limits of motions for safe working
conditions . . . . . . . . . . . . . . . . . . . . . . . . . . 208
13.6 Motions of moored ships in harbours . . . . . . . . . . . . . . . . 208
13.6.1 Recommended allowed limits of motions for safe moor-
ing conditions . . . . . . . . . . . . . . . . . . . . . . . . 209

REFERENCES 213

vii
 1
INTRODUCTION

Traditionally the sea-keeping and manoeuvring properties of ships have been con-
sidered of second importance to the fuel consumption for cargo ships and forward
speed for faster passenger ships. Knowledge of vertical motions and accelera-
tions are however important for estimating loads on cargo and equipment as well
as the possibility for the crew to work safely. The accelerations may also cause
seasickness to crew and passengers. The relative motion between the ship and the
water surface is important, especially in severe sea states, to investigate the risk of
“green water” on deck, propeller emergence and slamming. In the offshore indus-
try, on the other hand, the sea-keeping properties are the most important proper-
ties, as these properties are decisive for the usefulness of floating platforms. The
“downtime” when one cannot drill, have oil-exporting lines (risers) connected to
the sea floor, moor service or living quarters platforms close to fixed or other
floating production platforms, exchange cargo with supply ships nor even stay on
station must be as short as possible. Therefore, over the past decades, considerable
advances have been achieved in the theoretical prediction of motions of floating
platforms and ships, and in computer algorithms. The manoeuvring capabilities
is of great importance for safe operation of a ship in harbors, restricted waters and
at open sea. The manoeuvring theory is the basis both for tools used to predict a
ships manoeuvrability during the design phase, and for ship simulators used for
education and training of bridge officers. The development of computer process-
ing capacity has made methods for simulation of sea-keeping and manoeuvring
available for practical use..
A ship is moving i 6 degrees of freedom (6 DOF), surge, sway, heave, roll,
pitch and yaw. It is common to treat heave, roll and pitch together and call this the
sea-keeping problem. The surge, sway and yaw are all assumed to take place on
the undisturbed free-surface level. This is often called the manoeuvring problem.
This split is motivated by the small coupling between the degrees of freedom in
the two problems.

1
1. INTRODUCTION

This compendium will describe both the sea-keeping and the manoeuvring
problem. First the environment and the equations of motion will be described
followed by the sea-keeping problem and last the manoeuvring problem.
The sea-keeping part starts from the general equations of motion. Assump-
tions and approximations that can be justified for the sea-keeping problem are
introduced to the equations. A description of wave loads and the ships response
due to the loads are explained. Here the strip theory is introduced as a tool to
solve the sea-keeping problem. Full scale sea-keeping trials and model testing are
briefly explained. Other topics covered are derived responses such as slamming,
green water on deck, freeboard exceedance and propeller emergence. Added re-
sistance in waves and involuntary speed loss are discussed as well as voluntary
speed reduction. The effect on passengers and crew due to ship motions is also
discussed and finally design aspects with respect to sea-keeping is covered.
Also the manoeuvring part starts from the general equations of motion. A
final set of linear equations for the manoeuvring problem will be derived based
on assumptions and approximations relevant for the maneouvring problem. The
experimental setup to determine the stability derivatives in the final equations are
explained. Stability theory will be introduced and ways to analyze and improve
the course stability are discussed. Full scale tests used to determine the ships
manoeuvring properties are described together with the stop manoeuvre. The
influence due to restricted water is described and finally some practical aspects
are covered together with a description of rudders and other steering devices.
The ship motion problem including both sea-keeping and manoeuvring is a
three-dimensional time dependent non-linear problem with 6 DOF. It is neces-
sary to introduce assumptions and approximations in order to solve the problem
for practical purposes. In this compendium the focus will be on frequency do-
main methods based on two-dimensional strip theory for the sea-keeping prob-
lem. This approach was initially put forward by Korvin-Kroukovsky and Jacobs
[27] and was later developed by other authors, is still used extensively and has
been proven by both model tests and full scale trials to predict ship motions in a
seaway with acceptable accuracy. It is especially used for conceptual design and
risk assessment procedures [23].
A natural seaway is always irregular, and an irregular wave system can ap-
proximately be thought of as a superposed sum of an infinite number of regular
sinusoidal waves, each of which is characterized by frequency, amplitude, direc-
tion of propagation and a random phase angle. Under the assumption that the
motion response to the waves is linear, the superposition method can be utilized
also for the sea loads and the ship motions. This assumption is valid for moder-
ately steep waves. Thus once the response to regular waves has been assessed,
the response of the ship in an irregular seaway can be determined. The irregular
seaway or sea state is often represented by a spectrum and by multiplication of
this, for each frequency, with e.g. the linear response ratio of motion in that fre-
quency a response spectrum of the motion can be produced. Thereafter statistical
methods can be utilized to assess characteristics of responses in each sea state or

2
 1.1. Ship Motions in Waves

in all anticipated sea states during e.g. 30 years.

For large or steep waves and large relative motions non-linear time-domain or
non-linear frequency-domain methods [24] must be used. This class of methods
are only briefly discussed in the compendium and are out of scope of this course.
However, some essential non-linear effects like viscous quadratic damping (very
important in roll motion of ships and heave motion of offshore platforms) and
slamming will be discussed. Methods to handle the three-dimensional non-linear
time dependent problem are under development.
The two-dimensional strip theory cannot be used for offshore structures. Three-
dimensional approaches has been developed for the analysis of large fixed or float-
ing offshore structures with complex shapes[13].
A ship manoeuvre involves both time and position. The manoeuvring prob-
lem must therefore be solved in the time domain. A set of linear equation is used
to solve the manoeuvring problem. The coefficients for the linear terms are often
determined from experiments, but there is now an increasing competition from nu-
merical techniques based on solutions of the RANS-equations and in some cases
also from LES techniques.
The ship motion problem is very complex and there are several aspects that are
not discussed such as current forces, wind-drift forces and wind loads on moored
ships or structures.

1.1 S HIP M OTIONS IN WAVES

A ship with steady forward speed in irregular short-crested sea will oscillate in six
degrees of freedom. In the simplified case of steady speed in meeting or following
regular waves the ship will heave (vertical motion), pitch (tilting motion) and
surge (bow-aft motion) around its mean forward advancing position. In very long
waves its motion will just follow the sea surface motion but for shorter waves
near the vertical heave and pitch resonances of respective motion the motion will
be strongly amplified and out of phase with sea surface motion. For somewhat
shorter waves the motions will be opposed to the wave motion but less amplified,
so when the crest of the wave passes the ship the ship will be at its lowest position,
and when the wave slopes forward the ship will slope backwards with obvious
consequences for risk of green water and propeller emergence. See Fig 1.1.

1.2 C ALCULATION C HAIN

The chain of calculations to assess the sea keeping properties of a ship is outlined
in Figure 1.2.

1. Gathering of wave data for the route where the ship will operate. Weather
data may be taken from archived observations, satellite observations or

3
1. INTRODUCTION

Figure 1.1: Maximum pitching and heaving of SS Hamburg in wave conditions

that caused a speed reduction from 16 to 12 knots. (From Kempf
and Hoppe, 1926)

4
 1.3. On Wave-Induced Forces

Figure 1.2: Calculation chain

be “hindcasted” by wave generation models from historical meteorologi-

cal data. New measurements may be started to check the results from the
wave-generation models.

2. Settling design-weather conditions.

3. Choosing and applying some adequate wave theory.

4. Applying an adequate method for the hydrodynamic forces and reactions of

the ship or structure.

5. Calculate the response motions.

6. And finally derive the load effects i.e. sectional forces and moments, ten-
sions, risk for propeller emergence, slamming and green water. For moored
ships and structures also the mooring-line tensions are derived.

1.3 O N WAVE -I NDUCED F ORCES

One may say that there are two fundamentally different ways to calculate wave-
induced forces on structures in the sea. In one method one considers the structure
as a whole and assesses the total wave force from empirical or computed co-
efficients applied on water velocities and accelerations in the undistorted wave

5
1. INTRODUCTION

motion. In the other method the pressure distribution around the surface of the
structure is computed with due consideration to the water motion distorted by the
structure itself, and subsequently integrated around the structure.
In both cases some mathematical model for describing the wave properties is
necessary. For instance, by making the simplified assumption that the wave mo-
tion can be regarded as potential flow, velocities, accelerations and water motion
can be computed in any point under a gravity surface wave by a scalar quantity,
the velocity potential. In Chapter 2 some basics of potential flow theory will be
given and in Chapter 3 it will be used to derive kinematics of linear waves.

6
 2
POTENTIAL FLOW FOR WAVES

Many flow problems are elegantly solved by help of potential flow theory. Free-
surface waves is one of them. It is then assumed that the fluid is incompressible
and the flow irrotational. Irrotational flow is a flow where any selected fluid
packed does not rotate around its centre. It may, however be strongly deformed.

2.1 T HE E QUATION OF C ONTINUITY

Look at an infinitesimal control volume of fluid with the density, ρ, in a flow with
a co-coordinate system (x , y , z) and corresponding velocity components (u , v , w).
From Fig. 2.1 it is evident that to first order the resulting inflow of mass can be
written

 
∂ (ρu) ∂ (ρv) ∂ (ρw)
− + + ∆x ∆y ∆z . (2.1)
∂x ∂y ∂z

! "
∂(ρ u)
ρ u ∆y∆z ρu + ∆x ∆y∆z
∂x
∆z
z ∆y
y ∆x
x

Figure 2.1: The continuity of flow of an infinitesimal control volume.

7
2. POTENTIAL FLOW FOR WAVES

But this mass inflow must equal the increase of mass in the infinitesimal control
volume, ∆x ∆y ∆z, i.e.
 
∂ (ρu) ∂ (ρv) ∂ (ρw) ∂ρ
− + + ∆x ∆y ∆z = ∆x ∆y ∆z , (2.2)
∂x ∂y ∂z ∂t
which also can be expanded to
 
∂ρ ∂ρ ∂ρ ∂ρ ∂u ∂v ∂w
+u +v +w +ρ + + = 0, (2.3)
∂t ∂x ∂y ∂z ∂x ∂y ∂z
or written  
dρ ∂u ∂v ∂w
+ρ + + = 0. (2.4)
dt ∂x ∂y ∂z
For a homogeneous incompressible fluid each of the derivatives of ρ is zero,
and if the fluid is incompressible it is easily understood that the total derivative,
dρ/dt = 0, i.e. that the density of a chosen control volume of fluid is not changed
during its motion. Observe, on the other hand, that various parts of the fluid may
have different density due to e.g. varying salinity or temperature. The fact that
dρ/dt = 0 leads to the continuity conditions for both three-dimensional and two-
dimensional flow:
∂u ∂v ∂w
3-D + + = 0, (2.5a)
∂x ∂y ∂z
∂u ∂v
2-D + =0 (2.5b)
∂x ∂y

2.2 I RROTATIONAL F LOW

The rotation of a fluid element around its centre can be expressed by the spatial
gradients of the local fluid velocities in the x-, y- and z-directions. For the two-
dimensional case sketched in Fig. 2.2 we find that the gradients generally would
deform the element, because the section a – a rotates counter clockwise and the
section b - b rotates clockwise with the angular velocities

v + ∆x ∂∂ xv − v ∂ v
= , (2.6)
∆x ∂x
u + ∆y ∂∂ uy − u ∂u
− =− . (2.7)
∆y ∂y
The rotation of the element per unit of time, ωz , around the z-axis is defined as the
mean of the two angular velocities above, i.e.
 
1 ∂v ∂u
ωz = − . (2.8)
2 ∂x ∂y

8
 2.2. Irrotational Flow

Figure 2.2: Rotation and deformation of an infinitely small fluid element.

The rotation around the x- and y-axes, ωx and ωy , are similarly

 
1 ∂w ∂v
ωx = − , (2.9)
2 ∂y ∂z
 
1 ∂u ∂w
ωy = − . (2.10)
2 ∂z ∂x

The flow is said to be irrotational or free from circulation if

ωx = ωy = ωz = 0 , (2.11)

i.e. it does not rotate around its centre.

The Eqs. (2.8) to (2.11) then give

∂w ∂v
= , (2.12)
∂y ∂z
∂u ∂w
= , (2.13)
∂z ∂x
∂v ∂u
= . (2.14)
∂x ∂y

Note that even in irrotational flow the fluid element can be strongly deformed,
e.g. be deformed from a cube to a diamond. For examples of irrotational and
rotational flows see Figs. 2.3 and 2.4.

9
2. POTENTIAL FLOW FOR WAVES

(a) (b)

Figure 2.3: Parallel flow. (a) irrotational and (b) rotational.

(a) (b)

Figure 2.4: Curved flow. (a) irrotational and (b) rotational.

2.3 T HE V ELOCITY P OTENTIAL AND L APLACE D IFFERENTIAL E QUA -

TION

Under which conditions does a continuous and differentiable function, φ (x, z,t),
exist such that
∂φ ∂φ ∂φ
u= , v= , w= . (2.15)
∂x ∂y ∂z
Equation (2.15) means that the spatial gradient of φ in each point shall give the
velocity vector of the flow, U = ∇φ = grad φ . The function, φ , is therefore named
the velocity potential. Observe that the sign in Eq. (2.15) as well could be set
to minus as it is a pure definition, and so is also the convention in many civil
engineering textbooks and some other scientific literature, which one has to note
when studying different sources. A rational motive for choosing a negative sign is
that a minus sign corresponds to the analogy that the gravity force acts downhill
when going uphill i.e. in opposite direction to the slope. Here we will follow
the less intuitive convention according to (2.15) as this is most common in naval
architecture.
If the flow is irrotational then in e.g. in the x − z plane according to Eq. (2.13)
∂ u/∂ z = ∂ w/∂ x and thus that
∂u ∂ 2φ ∂w ∂ 2φ
= = = , (2.16)
∂z ∂z∂x ∂x ∂x∂z
i.e. that the mixed second derivatives are equal, which in turn proves that φ is
continuous and differentiable with respect to x and z. The same is true in the

10
 2.4. Bernoullis Equation for Potential Flow

other two orthogonal planes. The potential function φ exists thus if the flow is
irrotational and irrotational flow is therefore also called potential flow.
Further for an incompressible fluid the condition of continuity Eq. (2.5) is true
which with (2.15) substituted gives Laplace differential equation:

∂ 2φ ∂ 2φ ∂ 2φ
3D + + 2 = 0, (2.17a)
∂ x2 ∂ y2 ∂z
2
∂ φ ∂ φ2
2D + = 0, (2.17b)
∂ x2 ∂ y2

or symbolically ∆φ = 0 or ∇2 φ = 0.
Irrotational flow for an incompressible fluid can thus be described by the dif-
ferential equations (2.17), which are linear differential equations of second order.
A surface water wave motion with “small” amplitude in relation to its wavelength
and the water depth can with good precision be described as potential flow. The
deviation from the true physical wave motion for higher waves depends on ap-
proximations of the boundary condition and on viscous and rotational effects.
Due to the linearity of the differential equations (2.17) a wave motion decomposed
into many harmonic wave components with different amplitudes, frequencies and
phase lags, first the flow for each component can be solved and then linearly be
added to a total solution which gives all velocities and accelerations anywhere in
the fluid.

2.4 B ERNOULLIS E QUATION FOR P OTENTIAL F LOW

Velocities and accelerations in potential flow can thus be obtained directly from
the velocity potential, φ , by taking the derivatives with respect to space and time.
To calculate pressures and wave elevations it is, however, necessary to use an ad-
ditional condition, namely the Bernoulli Equation for incompressible, irrotational
flow. Most of us have met it before in the context of one-dimensional pipe flow,
but the version here is somewhat different and valid in three dimensions. In the
next paragraph we will derive it for two dimensions from Navier-Stokes’ Equa-
tions for an incompressible fluid.

2.4.1 Navier-Stokes’ Equations in two dimensions

Navier-Stokes’ Equations for two-dimensional flow, see e.g. [42, 8], can be writ-
ten:
 
∂u ∂u ∂u ∂ h 1 ∂ p µ ∂ 2u ∂ 2u
+u +w = −g − + + , (2.18a)
∂t ∂x ∂z ∂ x ρ ∂ x ρ ∂ x2 ∂ z2
 
∂w ∂w ∂w ∂ h 1 ∂ p µ ∂ 2w ∂ 2w
+u +w = −g − + + 2 , (2.18b)
∂t ∂x ∂z ∂ z ρ ∂ z ρ ∂ x2 ∂z

11
2. POTENTIAL FLOW FOR WAVES

where t is time, x and z are the space coordinates, u and w denote the velocities in
the x- and z-directions, respectively. The earth acceleration is denoted by g, while
h is height, a coordinate in negative g-direction i.e. h is positive upwards. Finally,
ρ is the density of the fluid and µ the dynamic viscosity of the fluid.

2.4.2 Deduction of the Bernoulli Equation

Substituting the conditions for irrotational flow Eq. (2.13), ∂ u/∂ z = ∂ w/∂ x, into
Eq. (2.18) gives
 
∂u ∂u ∂w ∂ h 1 ∂ p µ ∂ 2u ∂ 2w
+u +w = −g − + + , (2.19a)
∂t ∂x ∂x ∂ x ρ ∂ x ρ ∂ x2 ∂ z ∂ x
 
∂w ∂u ∂w ∂ h 1 ∂ p µ ∂ 2u ∂ 2w
+u +w = −g − + + . (2.19b)
∂t ∂z ∂z ∂ z ρ ∂ z ρ ∂ x ∂ z ∂ z2
As both u and w are continuous and differentiable the order of differentiation can
be changed and ∂ 2 /(∂ x ∂ z) = ∂ 2 /(∂ z ∂ x). Introduce also the fact that u∂ u/∂ x =
∂ (0.5u2 )/∂ x etc. and then:
   
∂ u ∂ u2 w2 p µ ∂ ∂u ∂w
+ + + + gh = + , (2.20a)
∂t ∂ x 2 2 ρ ρ ∂x ∂x ∂z
   
∂ w ∂ u2 w2 p µ ∂ ∂u ∂w
+ + + + gh = + . (2.20b)
∂t ∂ z 2 2 ρ ρ ∂z ∂x ∂z
But the continuity condition (2.5) is ∂ u/∂ x + ∂ w/∂ z = 0, and therefore the right
hand sides of the Eqs. (2.20) are identically zero for a continuous, incompressible
fluid.
The velocity potential, φ , according to the definition (2.15) gives the acceler-
ations expressed as:
 
∂u ∂ ∂φ ∂ 2φ ∂ 2φ
= = = , (2.21a)
∂t ∂t ∂ x ∂ x ∂t ∂t ∂ x
 
∂w ∂ ∂φ ∂ 2φ ∂ 2φ
= = = . (2.21b)
∂t ∂t ∂ z ∂ z ∂t ∂t ∂ z
Substitution of these expressions into Eq. (2.20) and integration with respect
to x and z then gives:
∂ φ u2 w2 p
+ + + + gh = f1 (t, z) , (2.22a)
∂t 2 2 ρ
∂φ u 2 w2 p
+ + + + gh = f2 (t, z) . (2.22b)
∂t 2 2 ρ
The Eqs. (2.22) must be satisfied simultaneously in all points (x , z) in the flow,
and therefore
f1 (t, z) = f2 (t, z) = f (t) , (2.23)

12
 2.4. Bernoullis Equation for Potential Flow

and the result is one single equation, the Bernoulli Equation, for an incompressible
fluid in irrotational two-dimensional flow:
∂ φ u2 w2 p
+ + + + gh = f (t) . (2.24)
∂t 2 2 ρ

The time-dependant part of f (t) can be included in ∂ φ /∂t, and then

∂ φ u2 w2 p
+ + + + gh = C , (2.25)
∂t 2 2 ρ
where C is a constant.
For three-dimensional flow the Bernoulli Equation is written:

∂φ 1 2
p
+ u + v2 + w2 + + gh = C (2.26)
∂t 2 ρ

For stationary flow ∂ φ /∂t = 0, and then we recognise the Bernoulli Equation
familiar from stationary one-dimensional pipe flow.

13
 3
REGULAR WAVES

3.1 T HE V ELOCITY P OTENTIAL

If the motion in a water mass due to a surface gravity wave can be approximated by
potential flow, we can derive the properties of the wave motion from the Laplace
Differential Equation as stated in Chapter 2,

∆φ = 0 , (3.1)

where φ is the velocity potential. From φ we can then derive i.e. velocities,
accelerations and pressures everywhere in the water mass.
The Laplace Differential Equation is widely applicable for field problems i.e.
treating heat, sound, electromagnetism and structural mechanics. In civil engi-
neering we meet it for description of ground water flow and for diffusion of heat
and chemical matters in structures. Note, however, that potential flow is free from
losses caused by viscosity, which ground water flow is not.
In this chapter we will treat “small-amplitude” wave theory – also called Airy
wave theory, first order wave theory or linear wave theory – which well describes
waves with the wave amplitude much smaller than the wavelength and the water
depth. For steep waves or finite amplitude waves in shallow water, higher-order
wave theories and non-linear wave theories for shallow water must be used. See
e.g. [53, 10, 29, 45].
The small-amplitude wave theory is, in spite of the underlying simplified as-
sumptions, very useful for many applications. It functions well for wave steep-
ness1 up to H/λ = 0.03 and furthermore, as it is linear, one can superpose solu-
tions for different frequencies and with varying direction of propagation and thus
calculate motion in irregular sea states.
1 Note the difference between the wave steepness H/λ = 2a/λ and the wave slope, which latter
is the slope of the water surface ∂ ζ /∂ x and sometimes the maximum slope which is ak = 2πa/λ
for a sinusoidal wave.

15
3. REGULAR WAVES

Figure 3.1: A progressive gravity wave, T = 10 s and d = 10 m.

3.2 B OUNDARY C ONDITIONS

The waveform of a plane harmonic wave progressing in the x-direction as illus-

trated in Fig. 3.1 can be written as a cosine function:
ζ (x,t) = a cos(k x − ω t) , (3.2)
The most obvious and also the simplest boundary condition is that the velocity
perpendicular to the bottom must be zero. Thus the Bottom Boundary Condition
∂φ
= w |z=−d = 0 . (3.3)
∂z
The Free Surface Kinematic Boundary Condition states that a particle on the sur-
face will stay at the surface:
∂ζ ∂ζ ∂ζ ∂φ ∂ζ ∂φ
+u −w = + − = 0. (3.4)
∂t ∂x ∂t ∂x ∂x ∂z
The Free Surface Dynamic Boundary Condition derives from the Bernoulli Equa-
tion on the assumption that the pressure is constant on the free surface. Then,
especially, if the atmospheric pressure is set to the reference pressure 0:
 2  2 !
∂φ 1 2
∂ φ 1 ∂φ ∂φ
+ u + w2 + gζ = + + + gζ = 0 . (3.5)
∂t 2 ∂t 2 ∂x ∂z

The velocity potential, φ , of the wave, Eq. (3.2) must satisfy the Laplace
Equation (3.1) and the boundary conditions (3.3) to (3.5). The solution of this
problem is not easy because the free-surface boundary conditions are non-linear
following the moving free surface.

3.3 L INEAR A IRY WAVE T HEORY

The simplest, but still very useful, wave theory is the Airy [1], linear or small
amplitude wave theory, which is based on the assumption that the wave amplitude,
a, is small compared to the wavelength, λ .

16
 3.3. Linear Airy Wave Theory

The linear wave theory is also called the first-order theory because one can ne-
glect terms that are above first order when expanding the solution in a perturbation
series. The solution of φ and the wave profile are then assumed to be expanded in
power series of a non-dimensional perturbation parameter, ε, in terms of the wave
slope at the zero down crossing of the wave:
2πa
ε= = ka , (3.6)
λ
in which λ is the wavelength, a the wave amplitude and k the wave number. Then
we can write for the potential
∞
φ= ∑ ε n φn , (3.7)
n=1

and for the wave profile

∞
ζ= ∑ ε n ζn , (3.8)
n=1
with index n indicating the nth order term.
Inserting the expansions (3.7) and (3.8) into the free-surface boundary condi-
tions retaining only first order terms gives:
∂ ζ1 ∂ φ1
− = 0, (3.9)
∂t ∂z
∂ φ1
+ gζ1 = 0 . (3.10)
∂t
One can show that these, to first order, can be applied at the mean water level,
see e.g. [28] (for second order Stokes theory see e.g. [4]).
The wave profile is given at z = 0 by (3.10) as
1 ∂ φ1
ζ1 = − . (3.11)
g ∂t
The two linearised free-surface boundary conditions can be combined to one
by eliminating ζ1 :
∂ 2 φ1 ∂ φ1
+ g = 0. (3.12)
∂t 2 ∂z

3.3.1 Solution for a progressive wave

The solution for a progressive wave is easiest accomplished by using complex
notation:
ζc = a eı (k x−ω t) = a eı k x e−ı ω t = a [cos(k x − ω t) + ı sin(k x − ω t)] , (3.13)
so that the real progressive wave Eq. (3.2) is given by
Re(ζc ) = Re (cos(k x − ω t) + ı sin(k x − ω t)) = a cos(k x − ω t) . (3.14)

17
3. REGULAR WAVES

We can from Eq. (3.13) see that the waveform, which is a function of both
time and space, can be separated into a product of two functions, each a function
of only one independent variable. Assuming the solution can be written as a
product of three single-variable functions, the solution could be written

φ = X(x) Z(z) T (t) . (3.15)

Then the Laplace Equation gives

∂ 2 φ ∂ 2 φ  00 00

+ = X (x) Z(z) + X(x) Z (z) T (t) = 0 , (3.16)
∂ x2 ∂ z2
which can be separated into two equations for T (t) not being identically zero:
(
X 00 Z 00 X 00 + k12 X = 0 ,
= − = −k12 ⇒ (3.17)
X Z Z 00 − k12 Z = 0 .

The solution to these two second-order differential equations in x and z have the
forms (
X(x) = Aeı k1 x + Be−ı k1 x ,
(3.18)
Z(z) = Cek1 z + De−k1 z ,
where the sign of Eq. (3.17) has been chosen to get a harmonic solution in the
x-direction. Thus
  
φ = Aeı k1 x + Be−ı k1 x Cek1 z + De−k1 z T (t) , (3.19)

in which A, B, C and D are constants to be determined from the boundary condi-

tions.
The time function of the complex waveform (3.13) is e−ı ω t and therefore
T (t) = e−ı ω t . Further, the waveform (3.13) progresses to the right as eı(k x−ω t) ,
and then B must be zero. Additionally A can arbitrarily be set to 1 by including
it into C and D and then X(x) = eı k1 x . We then also realise that k1 must be the
same constant as the wave number k otherwise the celerity - the propagation speed
of the waveform, ω/k - would be wrong. We note that there is of course also a
solution for waves progressing to the left for A = 0 and B = 1.
Now the problem is reduced to
  
ı k1 x −ı k1 x k1 z −k1 z
φ = Ae + Be Ce + De T (t)
 
= A eı k x Cek z + De−k z e−ı ω t
 
= Cek z + De−k z eı(k x−ω t)
 
kz −k z
= Ce + De X(x) T (t) . (3.20)

18
 3.3. Linear Airy Wave Theory

Two constants remain to be determined from the bottom and free surface
boundary conditions. The bottom boundary condition gives

∂φ  
−k d kd
= k Ce − De X(x) T (t) = 0 . (3.21)
∂z z=−d

As X(x) and T (t) are not identically zero this can be solved for instance for C:

Ce−k d − Dek d = 0 , (3.22)

2k d
C = De , (3.23)

which gives
 
Z(z) = D ek d ek(z+d) + e−k(z+d) = D1 ek d cosh(k(z + d)) (3.24)

with one unsolved constant D1 = 2D.

The linearised kinematic free surface boundary condition (3.9) and the wave
form Eq. (3.13) then gives D1 :
∂ ζc ∂φ
= −ı a ω eı k x = = eı k x D1 ek d k(sinh(k d)) e−ı ω t , (3.25)
∂t ∂z
and
−ı a ω
D1 = , (3.26)
ek d ksinh(k d)
which substituted into (3.20) gives one version of the complex velocity potential:
−ı a ω
φ = eı k x ek d cosh(z + d)e−ı ω t
ek d k
sinh(k d)
a ω cosh(k(z + d)) ı(k x−ω t)
= −ı e . (3.27)
k sinh(k d)
The combined free-surface boundary condition, (3.12),

∂ 2φ ∂φ
2
+g = 0,
∂t ∂z
will finally yield a functional relation between k and ω. Starting with

X(x) Z(z) T̈ (t) + g X(x) Z 0 (z) T (t) = 0 , (3.28)

dividing by X(x), and substituting Z(z) (3.24) and T (t) = e−ı ω t gives


D1 ek d cosh(k(z + d)) −ω 2 e−ı ω t + g k D1 ek d sinh(k(z + d)) e−ı ω t = 0 . (3.29)

Finally
ω2
= tanh(k d) (3.30)
gk

19
3. REGULAR WAVES

which implicitly gives the wave celerity, ω/k = f (k), and is called the dispersion
relation, because it shows that the wave celerity depends on the wave length,
which fact in turn makes groups of waves disperse.
The final complex velocity potential for plane progressive waves in finite water
depth is then after having substituted the constants and the dispersion relation:
a ω cosh(k(z + d)) ı (k x−ω t) a g cosh(k(z + d)) ı (k x−ω t)
φc = −ı e = −ı e (3.31)
k sinh(k d) ω cosh(k d)
and its corresponding real valued expression is
 
a g cosh(k(z + d)) ı (k x−ω t)
φ = Re −ı e
ω cosh(k d)
 
a g cosh(k(z + d))
= Re −ı (cos(k x − ω t) + ı sin(k x − ω t))
ω cosh(k d)
a g cosh(k(z + d))
= sin(k x − ω t) (3.32)
ω cosh(k d)

3.4 T HE A PPLICATION OF THE V ELOCITY P OTENTIAL

The velocity potentials (3.31) or (3.32) for a progressive plane harmonic wave are
very useful, because from one single scalar equation we can derive all particle
velocities and pressures in the water mass beneath the still-water level.
According to the definition (2.15) the complex particle velocities will be
∂ φc a g k cosh(k(z + d)) ı (k x−ω t)
uc = = e , (3.33a)
∂x ω cosh(k d)
∂ φc a g k sinh(k(z + d)) ı (k x−ω t)
wc = = −ı e , (3.33b)
∂z ω cosh(k d)
and the complex particle accelerations becomes, after taking the time derivatives,
∂ 2 φc cosh(k(z + d)) ı (k x−ω t)
u̇c = = −ı a g k e , (3.34a)
∂t ∂ x cosh(k d)
∂ 2 φc sinh(k(z + d)) ı (k x−ω t)
ẇc = = −a g k e , (3.34b)
∂t ∂ z cosh(k d)
The pressure above the atmospheric reference pressure can be calculated using
the Bernoulli equation ( 2.26)
 
∂ φc 1 2 2


pc = − ρ + ρ uc + wc + ρ g z . (3.35)
∂t 2
Here the first and second term are the dynamic fluctuating pressure due to the
waves and the third term the hydrostatic pressure. To first order the velocity
squared term can be neglected. Thus
 
∂ φc cosh(k(z + d)) ı (k x−ω t)
pc = −ρ +gz = ρ ag e −ρ gz. (3.36)
∂t cosh(k d)

20
 3.5. Velocity Potential in Deep Water

The wave form can be retained from the linearised free-surface boundary con-
dition (3.10):
1 ∂ φc
ζc = − = a eı(k x−ω t) . (3.37)
g ∂t

3.4.1 Real-valued expressions

The real valued particle velocities are from Eq. (3.33)
∂φ a g k cosh(k(z + d))
u= = cos(k x − ω t) , (3.38a)
∂x ω cosh(kd)
∂φ a g k sinh(k(z + d))
w= = sin(k x − ω t) , (3.38b)
∂z ω cosh(kd)
and the real valued particle accelerations
∂ 2φ cosh(k(z + d))
u̇ = = agk sin(k x − ω t) , (3.39a)
∂t ∂ x cosh(kd)
∂ 2φ sinh(k(z + d))
ẇ = = −a g k cos(k x − ω t) , (3.39b)
∂t ∂ z cosh(kd)

The real valued pressure above the atmospheric reference pressure to first or-
der is
cosh(k(z + d))
p = ρg a cos(k x − ω t) − ρg z . (3.40)
cosh(kd)
Furthermore the particle paths in the wave can be calculated from the Eq. (3.38)
by integration with respect to time. The resulting form is elliptic with the largest
axis horizontal. Close to the bottom the vertical axis is very short and the water
is oscillating horizontally only. In deep water the particle paths are circular, see
Fig. 3.2.

3.5 V ELOCITY P OTENTIAL IN D EEP WATER

By rearranging the velocity potential (3.31) by hyperbolic trigonometry and then

letting d → ∞ bringing with it tanh(k d) → 1 and ω 2 = gk, expressions for waves
in deep water in relation to the wavelength are yielded
ag cosh(k(z + d)) ı (k x−ω t)
φc = −ı e
ω cosh(kd)
 
ag sinh(kz) sinh(kd)
= −ı cosh(kz) + eı (k x−ω t)
ω cosh(kd)
ag
= −ı (cosh(kz) + sinh(kz) tanh(kd)) eı (k x−ω t)
ω
a ω kz ı (k x−ω t)
= −ı e e , (3.41)
k

21
3. REGULAR WAVES

Figure 3.2: Sketch of fluid-particle motion beneath a wave as predicted by linear

wave theory. From [49].

which thus gives a simple exponential function of z.

The complex velocities in deep water are thus

∂ φc
uc = = a ω ekz eı (k x−ω t) , (3.42a)
∂x
∂ φc
wc = = −ı a ω ekz eı (k x−ω t) , (3.42b)
∂z
and the complex particle accelerations in deep water,

∂ 2 φc
u̇c = = −ı a ω 2 ekz eı (k x−ω t) , (3.43a)
∂t ∂ x
∂ 2 φc
ẇc = = −a ω 2 ekz eı (k x−ω t) . (3.43b)
∂t ∂ z

The complex pressure read

pc = ρg a ekz eı (k x−ω t) − ρg z . (3.44)

3.5.1 Real-valued expressions

Equation (11.37) corresponding real-valued expression is
a ω kz
φ= e sin(k x − ω t) . (3.45)
k
The real valued particle velocities are from Eq. (3.42)

u = a ω ekz cos(k x − ω t) , (3.46a)

kz
w = a ω e sin(k x − ω t) , (3.46b)

22
 3.6. Wave Properties in Shallow Water

and the real valued particle accelerations from Eq. (3.43)

u̇ = a ω 2 ekz sin(k x − ω t) , (3.47a)

ẇ = −a ω 2 ekz cos(k x − ω t) . (3.47b)

The real valued pressure to first order from Eq. (3.44) is

p = ρg a ekz cos(k x − ω t) − ρg z . (3.48)

3.5.2 Particle paths

As told above the particle paths in a wave at any depth can be calculated from the
Eq. (3.38) by integration with respect to time. We will do this for the deep-water
wave to show the principle. First the horizontal position of a particle as a function
of time is
Z Z t
ξ= udt = a ω ekz cos(k x − ω τ)dτ = −a ekz sin(k x − ω t) + ξ0 , (3.49)

and the vertical position

Z Z t
ζ= wdt = a ω ekz sin(k x − ω τ)dτ = −a ekz cos(k x − ω t) + ζ0 , (3.50)

which using the trigonometric unity gives

(ξ − ξ0 )2 + (ζ − ζ0 )2 = a2 e2kz = r2 (z) , (3.51)

where r(z) is recognized as the radius of the particle path. Thus in deep water the
particles move in circular orbits with radii that decrease exponentially with depth.

Exercise 3.1
How large is the particle radius at the level z = −λ /2?

3.6 WAVE P ROPERTIES IN S HALLOW WATER

In shallow water the general expression for the velocity potential must be used,
but the derived velocities, accelerations and pressure can be simplified using the
fact that when d → 0 so tanh(kd) → kd.
The complex velocities in shallow water are thus
a ω ı(k x−ω t)
uc = e , (3.52a)
kd
z + d ı(k x−ω t)
wc = −ı aω e , (3.52b)
d

23
3. REGULAR WAVES

and the complex particle accelerations in shallow water,

a ω 2 ı(k x−ω t)
u̇c = −ı e , (3.53a)
kd
z + d ı(k x−ω t)
ẇc = −a2 ω e , (3.53b)
d
The complex pressure
pc = ρg a eı(k x−ω t) − ρg z . (3.54)

3.6.1 Real-valued expressions

The real valued particle velocities are
aω
uc = cos(k x − ω t) , (3.55a)
kd
z+d
wc = aω sin(k x − ω t) , (3.55b)
d
and the real valued particle accelerations are
a ω2
u̇c = sin(k x − ω t) , (3.56a)
kd
z+d
ẇc = −aω 2 cos(k x − ω t) . (3.56b)
d
The real valued pressure is
p = ρg a cos(k x − ω t) − ρg z . (3.57)

3.6.2 Summary of Airy waves

In Table 3.1 we present a summary of the expressions of the wave characteristics
of an Airy wave.

3.7 “O BLIQUE ” WAVES

For the discussion of motions of ships in obliquely approaching waves, the two-
dimensional linear plane wave above can be considered to be a three-dimensional
wave train with straight, infinitely long wave crests. If the considered wave is
travelling at the angle, µ, anticlockwise to the x-axis it can be described by the
equation:
ζ (x, y,t) = a cos (k(x cos(µ) + y sin(µ)) − ω t) .
In fact, all arguments (k x−ω t) can be exchanged by(k(x cos(µ) + y sin(µ)) − ω t)
in all equations. In some literature the notations kx = k x cos(µ) and ky = k y sin(µ)
are used giving
ζ (x, y,t) = a cos(kx x + ky y − ω t) .

24
 3.7. “Oblique” Waves

Table 3.1: Summary of linear (Airy) wave theory. Wave characteristics. Here
L = λ . From [? ]. !"#$$$%&'&$$%%#()*+,#--.
$#/01#%2#(34*156#'.

!"#$%&'"()"*%+ ,+$##-.(/$%"0 10$23&%&-2$#(/$%"0 )""*(/$%"0

! 5 5 ! 5 ! 5
4 4 4 8
" 67 67 " 6 " 6
! ! ! !
#! 4 4 #! 4 #! 8
57 57 6 6

59(/$'"(*0-:&#" ,$;"(<3(8( $ 6!% 6!& $ 4(,$;"(<3

" ! =-3 " ! =-3 #
6 " ' 6
69(/$'"(="#"0&%> " " )' 6!! " )'
( ! ! )! ( ! ! %$2+ ( ! (7 ! !
' ' 6! " ' 6!
?9(/$'"#"2@%+ " ! ' )! ! ( ' )' 6 6!! )' 6
" ! %$2+ " ! "7 ! ! (7 '
6! " 6!
A9(B0-C*('"#-=&%> () ! ( ! )! 5 A!!D" 5 )'
() ! * ( ! 5 # ( () ! ( !
6 3&2+ EA!!D"F 6 A!
G9(/$%"0(*$0%&=#"
((((('"#-=&%>
(((((E$F(H-0&I-2%$# $ ) $ )' =-3+ J6!E,#!FD"K 6!,
+ ! =-3 # + ! =-3 # !$ "
6 " =-3+ E6!!D"F + ! - =-3 #
6 ! '
(((((ELF(M"0%&=$# $! , $ )' 3&2+ J6!E,#!FD"K 6!,
. ! 5# 3&2 # . ! 3&2 # !$ "
' ! 6 " =-3+ E6!!D"F . ! - 3&2 #
'
N9(/$%"0(*$0%&=#"
$=="#"0$%&-23
(((((E$F(H-0&I-2%$# $! ) )!$ =-3+ J6!E,#!FD"K 6 6!,
/% ! 3&2 # /% ! 3&2 # ! "
" =-3+ E6!!D"F /% ! 6 $ - 3&2 #
' ! '
(((((ELF(M"0%&=$# ! 6
, )!$ 3&2+ J6!E,#!FD"K 6 6!,
/, ! "6 $ 5# =-3 # /, ! " =-3 # ! "
" =-3+ E6!!D"F /, ! "6 $ - =-3 #
' ! '
O9(/$%"0(*$0%&=#"
((((P&3*#$=";"2%3
(((((E$F(H-0&I-2%$# $' ) $ =-3+ J6!E,#!FD"K 6!,
$ ! " 3&2 # $ ! " 3&2 # $ "
6 3&2+ E6!!D"F $ ! " - 3&2 #
A! ! 6
(((((ELF(M"0%&=$# $ , $ 3&2+ J6!E,#!FD"K 6!,
% ! 5# =-3 # % ! =-3 # $ "
6 ! 6 3&2+ E6!!D"F % ! - =-3 #
6
Q9(,CL3C0:$="( 0 ! & ) E"",F =-3+ J6!E,#!FD"K 6!,
((((*0"33C0" 0 ! &)" "&), "
=-3+ E6!!D"F 0 ! &)" - "&),

7850+6#--&$&9:###;0<<*+=#>?#@816*+#(A8+=.#B*C6#,46>+=#&#B*C6#D4*+*D,6+8E,8DE

$33C;*%&-2(;$>(L"(RC"3%&-2$L#"9((<(%+&0P(P&;"23&-2#"33(*$0$;"%"0S(.+&=+(;$>(L"(C3"P(%-(0"*#$="("&%+"0(%+"
.$'"(3%""*2"33(-0(0"#$%&'"(.$%"0(P"*%+S(;$>(L"(P":&2"P($3(%+"(0$%&-(-:(.$'"(3%""*2"33(%-(0"#$%&'"(.$%"0(P"*%+9
1+C3S

$D" $
! ETTU5UNNF
!D" !

!"#$%&!"'$&($)*"+,)- 25
../0/10
3. REGULAR WAVES

Figure 3.3: The mass of the lifted water between the two dashed verticals is
ρgζ dx and this mass is lifted the height ζ /2.

3.8 WAVE E NERGY

The local mean energy in a regular progressive wave can be estimated by integrat-
ing the potential and kinetic energy in the wave over a wavelength.
The potential energy of the wave at a time instant, e.g. t = 0, can be calculated
as the deformation work needed to give the form of the wave, see Fig. 3.3 . Thus
the mean energy over one wavelength is
Z λ Z λ
1 ζ ρg 1
Ep = ρ gζ dx = ζ 2 dx = ρ g a2 . (3.58)
λ 0 2 2λ 0 4

The kinetic energy is the total kinetic energy contained in all the water from the
free water surface to the bottom of the sea. To first order we can only integrate
from the mean water surface, as the theory is not valid above the mean water
surface, and would in fact give large errors for finite waves. For the wave in deep
water it is especially easy to make the integration as the water particle moves with
constant speed V along their circular paths with the radius decreasing with the
level of submergence, z,
2π
r = r0 ek z = r0 e λ z , (3.59)
where r0 = a is the radius at the mean water level, z = 0, equal to the wave ampli-
tude. The velocity is thus

2πr(z) p 2
V (z) = = u + w2 , (3.60)
T
and the kinetic energy per unit volume is

1
ρ (V (z))2 . (3.61)
2
It is interesting to note that the kinetic energy is constant over the horizontal planes
at each level, while the potential energy varies with time.

26
 3.9. Wave Power

Finally integrating from the depth to the mean water level

Z 0  
1 2π k z 2
Ek = ρ ae dz
2 −∞ T
 2
1 2π 1 h 2kz i0
= ρ a2 e
2 T 2k −∞
1 1
= ρω 2 a2
2 2k
1
= ρga2 . (3.62)
4
The last equality comes from the dispersion relation in deep water gk = ω 2 . The
total wave energy follows as
1
E = E p + Ek = ρ g a 2 , (3.63)
2
which has the unit J/m2 . Equation (3.63) shows that the energy content averaged
over a horizontal area in a small-amplitude, harmonic wave is proportional to the
wave amplitude squared.

3.9 WAVE P OWER

The energy transport or wave power per unit width of a plane wave can be calcu-
lated by estimating the work done in the propagation direction on the water mass
to the left.
In deep water the work done at the level z in the vertical x = 0 m during the
time dt is
dW = p dx dz , (3.64)
and per time unit of time
dW dx
= p dz = pu dz . (3.65)
dt dt
Integrated along the vertical the wave power is
Z 0  
P(t) = ρg aekz cos(−ω t) − z a ω ekz cos(−ω t)dz
−∞
Z 0 Z 0
= ρga2 ωe2kz cos2 (ω t)dz + ρg(−z)a ωekz cos(ω t)dz
−∞ −∞
Z 0 Z 0
= ρg a2 ω cos2 (ω t) e2kz dz − ρg a ω cos(ω t) zekz dz
−∞ −∞
Z 0 Z 0
1
= ρg a2 ω (1 + cos(2ω t)) e2kz
dz − ρg a ω cos(ω t) zekz dz .
2 −∞ −∞
(3.66)

27
 ρ (V ( z ) ) . …(5.61)
2

It is interesting to note that the kinetic energy is constant over the

horizontal planes at each level.

Finally integrating from the depth to the mean water level

3. REGULAR WAVES
2 2
, 2π kz ) 1 & 2π #
0
1 1 2 kz 0
2 −-∞*+ T
Figure 5.4 Pressure distribution Ek = ρ ae ' dz = ρ $ ! a 2 e Ι =
in a deep-w ater w ave ( 2 % "
T 2 k −∞
…(5.61)
1 1 1
= ρω a
2 2
= ρga 2

2 2k 4

The last equality comes from the dispersion relation in deep water
gk = !2. The total wave energy follows as

1
E = E p + Ek = ρga 2 ,
2

which has the unit J/m2, shows that the energy content averaged
over a horizontal area in a small-amplitude, harmonic wave is
proportional to the wave amplitude squared.

5.9 Wave Power

The energy transport or wave power per unit width of a plane
wave can be calculated by estimating the work done in the
propagation direction on the water mass to the left.

In deep water the work done at the level z in the vertical x = 0 m

during the time dt is

dW = p dx dz …(5.62)

and per time unit of time

Figure 3.4: Pressure distribution in a deep-water wave.

The first term on the right hand side in the last line in Eq. (3.66) thus oscillates
with twice the frequency of the wave, but is always larger than zero, while the
second term has the same frequency as the wave but gives no net transport as its
time average is zero.

The net or mean power energy transport is thus

Z 0 Z
1 1 T
P(t) = ρ g a2 ω (1 + cos(2ω t)) dt e2k z dz
T 2 0 −∞
1 1 1 1ω C
= ρ g a2 ω = ρ g a2 = E = Cg E , (3.67)
2 2k 2 2k 2

from which it is seen that in deep water the wave energy is transported at half the
wave celerity C. This transport velocity, Cg , is also called the group velocity due
to the fact that a group of waves must propagate with this velocity. Otherwise it
would leave its energy behind and disappear. See also the next paragraph.

28
 3.10. Wave Celerity, Group Velocity and Particle Velocities

Figure 3.5: The water level as a function of time for a composed wave train
at three points along its route of propagation. The uppermost trace
is “upwind” and the lower ones are increasingly farther downwind.
The arrows indicate how the shorter waves arrive later (solid arrow)
downwind than the longer waves (dashed arrow). This wavetrain
was produced by a moving ship.

3.10 WAVE C ELERITY, G ROUP V ELOCITY AND PARTICLE V ELOCITIES

3.10.1 Dispersion and celerity

An important feature of surface gravity waves is that the speed of propagation
of the waveform, i.e. the phase speed or the celerity depends on the wave pe-
riod. This has the effect that a group of waves containing components of various
frequencies will be spread out or dispersed as it propagates.
To illustrate the effect of dispersion a wave train registered at three different
points along its route of propagation is shown in Fig. 3.5. The composed wave
train in the uppermost time trace separates into longer and shorter waves, the
longer component waves arriving earlier than the shorter component waves further
along the route of propagation.
The celerity is by definition:

λ ω
C= = , (3.68)
T k
where ω and k are interdependent through an implicit dispersion relation, Equa-
tion (3.30), that was derived from the velocity potential

ω 2 = g k tanh(k d) , (3.69)

which gives the celerity

r
ω g g
C= = tanh(k d) = tanh(k d) . (3.70)
k k ω

29
3. REGULAR WAVES

Table 3.2: Some important relations for linear waves at various relative depths.

Deep water Intermediate water Shallow water

(k d > π) (π/10 < k d < π) (k d > π/10)

g g √
Celerity C= C= tanh(k d) C= gd
ω ω

gT 2 gT 2 p
Wavelength λ= λ= tanh(k d) λ =T gd
2π 2π
 
C g C 2k d √
Group velocity Cg = = Cg = 1+ Cg = C = gd
2 2ω 2 sinh(2k d)

Amplitude of
vertical particle aω aω aω
velocity at the
water surface

Amplitude of p
horizontal particle aω a ω cosh(k d) a g/d
velocity at the
water surface

This expression cannot be solved explicitly for k or C, but must be solved by

iteration in k or using approximations (e.g. a Padé approximation).
In the limits for deep water (d  λ ) and shallow water (d  λ ), however,
explicit expressions for the celerity and wavelength are attained. See Tables 3.1
and 3.2.

3.10.2 Group velocity

The energy of surface gravity waves is generally propagating slower than the
waveform, which means that, except in extremely shallow water, a group of waves
will propagate slower than the individual waves. Thus the rearmost wave will
progress through the wave group to finally die out at the front of the wave group
while new waves are continuously developing in the rear.
This phenomenon is illustrated for a deep-water wave in Fig. 3.6, where a
wave group is snapshot at equally spaced time intervals along its route of propa-
gation. The crest of the rearmost wave in the earliest snapshot is marked with a
circle and traced in the consecutive snapshots. It can be seen how it is progress-
ing through the group and how its amplitude first grows and then attenuates to nil

30
 Wave-Induced Loads and Ship Motions
02 December
3.10. 2008 Group Velocity and Particle Velocities
Wave Celerity,
Page 51

Water level

Distance

Figure 3.6: Snapshot

Figure 5.6 Snapshot ofof aadeep-water
deep-water wave
wave group
group atatequally
equallyspaced
spacedintervals
intervals
along its route of propagation. The rearmost wave in the
along its route of propagation. The rearmost wave in the earliest earliest
snapshot is marked
snapshot marked with
with aacircle
circleand
andtraced
tracedininthe
theconsecutive
consecutive snap-
shots.
snapshots.

An alternative to the calculation of the group velocity from the energy transport is to
superpose two progressive waves with almost the same frequencies ω and ω’ and
when it leaves
consequently thethe
almost group
sameinwave
the front.
numbers k and k’. See Figure 5.7.
The speed of propagation of the energy and the wave group is called the group
ζ ( xvelocity
, t ) = a(cos(and ωt )half
kx − is + cos(k ′x −
the ω ′t )) =celerity in deep water and equal to the wave celerity
wave
,2 very
in & 1 shallow # water. & 1 Equations#/ 2 for& the1 group# velocity & 1 are given #/ ) in Table 3.2,
= a +0cos2 $ (kx − ωt ) ! − sin 2 $ (kx − ωt ) !- + 0cos2 $ (k ′x − ω ′t ) ! − sin 2 $ (k ′x − ω ′t ) !- ( =
*1
and %
the 2 group "velocity % 2 in deep ". 1water % was
2 derived " from% 2 the energy". ' ...(5.68)
transport in
Eq. &(3.67).
1 1 # &1 1 #
= 2a cos$ (k ′x − ω ′t ) + (kx − ωt ) ! cos$ (k ′x − ω ′t ) − (kx − ωt ) ! =
% 2 alternative
An 2 to the"calculation
%2 of the 2 group" velocity from the energy trans-
port& is to superpose two# progressive
1 & 1
= 2a cos$ (kx − ωt ) + (!kx − !"t ) ! cos$ (!kx − !"t ) !
#
waves with almost the same frequencies ω
and %ω ∗ and consequently
2 " almost
%2 the same " wave numbers k and k∗ . See Fig. 3.7.

whereζ (x,t) k′ −
!k = = k and x − ω t) + cos(k∗ x − ω ∗t))
a (cos(k
!" = ω ′ −ω .    
12 2 1
= a cos (k x − ω t) − sin (k x − ω t)
As δk and δω are assumed 2 to be small we can write
2
   
2 1 ∗ ∗ 2 1 ∗ ∗
+ cos (k x − ω t) − sin (k x − ω t)
&1 #
(kx − ωt ) cos$ (!kx − !"t ) ! 
2
ζ ( x, t ) = 2a cos 2 …(5.69)
 
∗% "
1 ∗ 2 1 1 ∗ ∗ 1
= 2a cos (k x − ω t) + (k x − ω t) cos (k x − ω t) − (k x − ω t)
2 2 2 2
The “carrier wave” a cos  (kx − ωt ) will thus be modulated
 by the function 
1 1
= 2a cos (k x − ω t) + (δ k x − δ ω t) cos (δ k x − δ ω t) , (3.71)
2 2
&1 #
cos$ (!kx − !"t ) !
where δ%k = k − k and
2 ∗ " δ ω = ω ∗ − ω. As δ k and δ ω are assumed to be small we

31
3. REGULAR WAVES

Figure 3.7: Snapshot of a wave composed of two sinusoidal waves of almost the
same wavelength. The longer-period wave is the modulated wave
and the shorter-period wave is the “carrier” wave.

can write  
1
ζ (x,t) = 2a cos(k x − ω t) cos (δ k x − δ ω t) . (3.72)
2
The “carrier wave” a cos(k x − ω t) will thus be modulated by the function
cos(0.5(δ k x − δ ω t) which has the phase velocity δ ω/δ k.
Each group of waves that is contained between the zero-crossings of the mod-
ulating function is thus moving with this velocity, and therefore it is denoted the
group velocity.
In the limit when δ k → 0 the group velocity will be given by the partial deriva-
tive of k in Equation (5.30) with respect to ω
δk ∂k
lim = = Cg . (3.73)
δ k→0 δ ω ∂ω

3.10.3 Water particle motion and dynamic pressure

The water particle velocities were derived from the velocity potential above. The
vertical velocity amplitude of a regular linear wave is always 2πa/T or aω. In
deep water the horizontal velocity amplitude also equals that value, while in shal-
lower water the horizontal velocity amplitude is larger. See Table 3.2. Note also
that the particle velocity is generally different from the wave celerity and the group
velocity, except for very steep, breaking wave crests, where the particle velocity
can approach or even supersede the celerity, thus making the particles leave the
wave surface.

3.11 F INITE A MPLITUDE WAVES AND H IGHER -O RDER WAVES

Recall that the linear wave theory assumes that the wave height is small compared
to the wavelength and water depth. In natural wind waves the steepness H/λ sel-

32
 3.11. Finite Amplitude Waves and Higher-Order Waves

dom exceeds 0.05 to 0.08 in deep water, so the small amplitude theory is often rea-
sonably valid. In some applications it is, however, necessary to use non-linear or
finite-amplitude wave theory. Physically the difference between linear and finite-
amplitude theories is that finite amplitude-theories consider the influence of the
wave itself on its properties. Therefore the phase speed, wavelength, water sur-
face elevation and other properties are functions of the actual wave height.
There are a number of different finite-amplitude wave theories. For deep to
intermediate-depth water (d/λ < 1/8) the most commonly adopted is the theory
by Stokes (1847, see e.g. [53]). For shallower water Cnoidal Wave Theory (Ko-
rteweg de Vries, 1895, see e.g. [53]) or Stream Function Theory (see e.g. [10]) is
more applicable. Williams [54] has produced tables of progressive gravity waves
covering the full range of wavelength from solitary to infinite-depth waves, and
up to the waves of limiting heights with sharp crests.

3.11.1 Stokes’ second order wave in deep water

Linear wave theory predicts sinusoidal waves with equal crest height and trough
depth. However, steep real waves in deep water have peaked high crests and flat
shallow troughs. For applications such as determining deck elevation of offshore
structures this is important, and usually Stokes’ 2nd order theory is used for this
problem, while for calculating wave forces on fixed structures from extremely
large long period waves rather Stokes’ 3rd or 5th order theory is applied. Also,
linear wave theory predicts no net mass transport, as the water particles move in
closed orbits, while finite amplitude theories predict a small net transport in the
direction of wave propagation.
In deep water the surface elevation of a second order Stokes’ wave with the
wave height H = 2a can be written
πa2
ζ (x,t) = a cos(k x − ω t) + cos (2(k x − ω t)) . (3.74)
2λ
Its elevation as a function of the space co-ordinate x is shown in Fig. 3.8.

3.11.2 Finite-height shallow water waves

For real shallow water waves of finite height the flaws of linear theory become still
more apparent. E.g. the mass transport is significant, waves may break, the crests
get steeper fronts than backs and the mean water level is elevated above the still
water level. In the limit, the whole wave may be above the still water level and
the whole water mass of the wave is transported forward for each individual wave
passage. In this limit solitary wave theory can be used for non-breaking waves
while for waves with steep fronts other depth-integrated theories like the shallow
water equation may be more useful describing bores and shocks.
In Figure 3.9 dimensionless wave profiles of 40 cases [9] (as published by
CERC [49]), are shown as an illustration of the difference to the linear, sinusoidal

33
3. REGULAR WAVES

Figure 3.8: An example of a second order Stokes wave as a function of the hori-
zontal co-ordinate x. The crest and trough elevations are also given.

wave profile of the Airy theory. In the figure the parameter, d/gT 2 , is a shallow-
water parameter (∼ water-depth to deep-water wavelength) and the parameter,
H/Hb , is a kind of steepness parameter (ratio between actual wave height and the
breaking wave height for the considered water depth and wavelength). In the left
low corner the profile of a moderately steep wave in intermediate water is shown,
and it can be seen that this profile is reasonably sinusoidal. As a contrast, in the
right high corner a maximally steep wave in very shallow water shows a profile
far from sinusoidal.
In Fig. 3.10 a graph over areas of best fit for wave theories according to
Le Méhauté, as published by [49], is shown. The horizontal axis is a water-depth
to deep-water wavelength parameter, d/gT 2 , and the vertical axis a wave height
to deep-water wavelength parameter, H/gT 2 .

3.11.3 Solitary waves

The theories above consider oscillatory or almost oscillatory waves, i.e. the water
moves forward and backwards. The linear wave is truly oscillatory, as the water
particles will return to their starting position for each cycle. In finite-amplitude
waves, however, the water particles are translated a small net amount forward for
each wave passage. When the water particles move only in the direction of wave
propagation, the wave is called a wave of translation. The solitary wave is such
a wave. A true solitary wave is entirely above the still water level, shows up as
a smooth hump of water with no waves behind or in front of it and translates at
constant celerity without loosing wave-height. It requires finite-depth and it is
two-dimensional. It is difficult to form a solitary wave in practice although it is
rather easy to form an approximate one with a tail of small dispersive waves [53].

34
 3.11. Finite Amplitude Waves and Higher-Order Waves

Figure 3.9: Dimensionless wave profiles for 40 cases of steep periodic waves in
shallow to intermediate water depth (the numbers on each plot rep-
resent the value of H/gT 2 for each case). Here η = ζ λ . From [49].

35
3. REGULAR WAVES

Figure 3.10: Regions of validity for various wave theories. Here L = and Lo is
the deep water wave length. From [49].

36
 3.12. Propagation and Transformation

Figure 3.11: Comparison of measured and theoretical wave profiles . From [53].

For regular waves running up a beach it is recommended to rather use the

cnoidal wave theory or the stream function wave theory. The cnoidal wave theory
approaches the solitary wave theory for shallow water waves and the linear or
Stokes theory for deep-water waves [49]. For a comparison of waveforms see Fig.
3.11.

3.12 P ROPAGATION AND T RANSFORMATION

3.12.1 Dispersion and transformation to swell

Weather systems usually move at a much slower speed than the wind velocity
within them. As the celerity of the waves in a well developed sea state are ap-
proximately the same as the wind velocity, and the group velocity is less than that
- in deep water only half the celerity - the result will be that the waves run out of
the windy area where they are generated. After leaving the generation area they
are no longer acted upon by the wind, and rather soon internal friction (viscosity),
parasitic capillary waves and air resistance will dissipate the sharp wave crests
and the shortest components of the spectrum. Such a free wave system is referred
to as swell. Also, due to dispersion the longer faster waves with longer periods
will arrive at distant points long before the shorter waves. The total effect is that
at distant points the swell can become almost monochromatic with a wave period
slowly decreasing with time, as decreasingly shorter waves continue to arrive. In
the low atoll islands of Polynesia, traditionally, the first arriving swell served as
an alarming forewarning of approaching hurricanes [53, 26].

37
3. REGULAR WAVES

Figure 3.12: Variation of the shoaling coefficient, Ks = H/H0 , with the non-
dimensional water depth, d/L0 . L0 is the deep-water wavelength.

3.12.2 Shoaling

Consider the two-dimensional problem of regular waves at normal incidence to

a shoaling beach. Neglecting dissipation, e.g. bottom friction, the time average
energy transport, P̄, must be constant through all vertical sections, and as the
energy is transported by the group velocity which varies with the depth, d, the
following relation must hold from deep water to finite water depth

P̄ = Cg0 E0 = Cg E , (3.75)

where the index 0 denote ‘deep water’, Cg is the group velocity and E is the wave
energy.
With the energy at any depth proportional to amplitude squared – in deep
water E0 = 0.5ρ g a20 and in finite-depth water E = 0.5ρ g a2 – the ratio between
the wave amplitude or wave height in finite depth water to that in deep water can
be solved. This ratio is called the shoaling coefficient,
s
a H Cg0
Ks = = = . (3.76)
a0 H0 Cg

As shown in Fig. 3.12, going from deep water into shallower water, the shoal-
ing coefficient first decreases slightly below one before increasing rapidly. At the
same time the wavelength becomes shorter due to decreasing celerity, and finally
the waves may break. See the paragraph on wave breaking below.

38
 3.12. Propagation and Transformation

Figure 3.13: Oblique waves traversing a uniformly sloped shelf. From [44].

3.12.3 Refraction

As pointed out above, the wave celerity decreases with water depth. Then, if a
long-crested wave approaches a uniformly sloped shelf at an oblique angle the
wave slows down and the wave crest will bend to become more parallel to the
depth contours, see Fig. 3.13. This phenomenon is equivalent to the refraction of
light in optics.
Going into more detail to see how the refraction affects the wave height one
can follow wave rays or wave rays normal to the wave crests and parallel to the
wave celerity in each point. Being parallel in deep water they will either spread
or approach each other approaching land, depending on the bottom topography.
Further, assuming that no energy will be transported across the wave rays but be
contained between two adjacent wave rays, the effect will be decreasing wave
height if they spread and increasing wave height if they approach each other.
Letting them be evenly distributed in deep water at a distance b0 , the distance
will become b somewhere closer to land. The result will be a change pin wave
height by a factor additional to the shoaling effect. The factor, Kr = b0 /b, is
called the refraction coefficient. The total effect of shoaling and refraction will be
s r
a H Cg0 b0
= = = Ks Kr . (3.77)
a0 H0 Cg b

39
 a H go bo
= = = Ks Kr . …(5.74)
ao H o Cg b

On a3.straight, shoaling coast the wave rays will spread and thus the refraction
REGULAR WAVES
coefficient will become lower than one: Kr < 1. See Figure 5.14.

Figure 3.14: Orthogonal spacing over a uniformly sloped beach. From [44].
Figure 5.14 Orthogonal spacing over a uniformly sloped beach.
(From Silvesterxxvi, 1974)

On On a straight,
a coastline withshoaling coast and
headlands the wave
bays rays will spread
the result and thusand
of refraction the shoaling
refraction will be
coefficient
increased wavewillaction
become
onlower r < decreased
than one: Kand
the headlands 1, see Fig. 3.14.
action in the bays. See Figure
5.15. On a coastline with headlands and bays the result of refraction and shoaling
will be increased wave action on the headlands and decreased action in the bays.
See Fig. 3.15.

3.12.4 Diffraction and reflection

In water deeper than half the wavelength there is no refraction, but the waves will
spread around steep rock peninsulas, piers and structures with steep walls due to
diffraction, when energy spreads along the wave crests into “shadow” areas.
While waves approaching a gradually shoaling coastline will be absorbed by
bottom friction and wave breaking, waves hitting steep rocks, piers or other struc-
tures will be reflected. In the examples shown in Figure 3.16 e.g. waves will be
reflected on the “wave ward” side of the structures, so that a complicated wave
pattern composed of reflected, incident and diffracted reflected waves will set up
on the “wave ward” side. An example is shown in Fig. 3.17. If thus the bottom is
steep or structures dominate an area, where wave propagation shall be modelled,
models must be used that can take diffraction and reflection into account, while
simpler models may be used for mildly sloping bottom topographies.

40
 3.12. Propagation and Transformation

Figure 3.15: Refraction along an irregular shoreline. From [49].

3.12.5 Wave breaking

The “white capping” or wave breaking involves dissipation of energy. Regular
waves will theoretically break when the wave steepness, H/λ , approaches
 
Hb 2πd
= γ tanh , (3.78)
λ λ

with γ = 0.142 [33]. In Fig 3.10 this limit is denoted H0 /L0 = 0.14. In practice
however γ = 0.12. In deep water this means that the steepness will normally not
exceed 0.12. See, e.g., Silvester [44] for further details. Although steeper waves
than given by Eq. (3.78) have been registered now and then, it is not until lately
it has been recognised that these rogue or freak waves may be more frequent than
anticipated, and thus are responsible for many losses of ships and some unfore-
seen damage to offshore drilling platforms. Such waves are three-dimensional in
character and just now (2004) subject to much research2 . In an irregular sea the
steepness breaking will transfer some energy to longer wave components [44] but
largely involves dissipation of energy of shorter wave components. See Young
[55] for further details.
If the wave does not break before it has entered the sloping beach the slope
itself has an influence on the breaking process. One criterion by Collins, as cited
by [44], takes this into account:
Hb
= 0.72 + 5.6S , (3.79)
d
2 http://www.esa.int/esaCP/SEMOKQL26WD index 0.html

41
3. REGULAR WAVES

Figure 3.16: Diffraction of waves: (A) wave diffraction behind a semi-infinite

breakwater, (B) diffraction through a breakwater gap and (C)
diffraction behind an island or offshore breakwater. From [44].

42
 3.12. Propagation and Transformation

Figure 3.17: Sketch of wave crests of regular waves incident, reflected and
diffracted against a semi-infinite breakwater. From [44].

where S is the bottom slope. The breaker height is thus increased at breaking.
Often however a simplified expression is used only containing the water depth:

Hb
= 0.78 . (3.80)
d
In Fig. 3.10 this limit is denoted H/d = 0.78. Different types of breakers are
shown in Fig. 3.18. The breaker type depends on deep-water steepness, beach
slope and wave period. See e.g. [49, 44].
For irregular waves there are some different approaches to depth limited break-
ing. Battjes and Jansen (as cited by [55]) e.g. look at individual waves assuming
them to be Rayleigh distributed, and let all waves with heights above the limiting
criterion be dissipated. Young [55] limits the total energy of the spectrum by a
criterion containing the average wave celerity.

3.12.6 Bottom friction dissipation

The forth term listed by Young [55] is the bottom friction dissipation, which for
rigid impermeable bottoms depends on the sheer just above the bottom. There-
fore it is negligible in water deep in relation to the wavelength and increases for
shallower waters. It is, however, important for longer waves in rather deep water

43
3. REGULAR WAVES

Figure 3.18: Successive profiles of breakers of various types. From [44].

e.g. for tidal waves, the “Bernoulli wave”, or shoaling secondary ship generated
waves. Silvester [44] gives an account for regular waves, and Young [55] for ir-
regular waves. The degree of dissipation is governed by the bottom roughness,
which depends on grain diameter of the bottom material, the geometry of ripples
or dunes etc. Permeable and soft bottoms increase the dissipation.
The lost wave energy is partly dissipated into heat and partly used for erosion,
ripple and dune formation, and net transport of bottom materials.

44
 4
WIND WAVES

In this chapter some properties of real wind waves are described, it is shown how
they can be looked upon as a linear combination, superposition, of regular waves,
and how realistic wind waves can be synthesised or simulated. Last some basic
wave statistics are given. For broader information, see e.g. the classical book by
Kinsman [26] or a more recent book by Dean and Dalrymple [10].
Waves at sea show a constantly changing, never repeated pattern. They grow
under the action of the wind, and during the growth phase the wave-height, wave
period and wavelength are due to the wind force (wind speed), the duration of the
wind and the length, fetch, of which the wind can act on the waves. On the high
seas the possible wave height is thus limited by the strength, diameter and motion
of the low-pressures, in lakes and landlocked seas of the wind speed and distance
to the upwind shore. At Fully Arisen Sea (FAS) the celerity of dominating waves
approach the wind speed and as a result the wind cannot transfer more energy
to the waves. The exact mechanisms for the generation of waves from a smooth
wave-surface are still not completely explained, but there are hypothetical models
describing the energy transfer from the wind to the waves, e.g. Jeffreys’ and
Phillips-Miles’ models (See Massel [32]). Based on these, empirical functions
were developed during the first half of the 20th century notably by Sverdrup,
Munk och Bredtschneider (See SPM 1983). Later more sophisticated models
have been developed, that in some countries are run on a daily basis to give wave
prognoses. An orientation over recent models is given by Young [55].
Under the progression towards coasts and beaches the waves will be affected
by the bottom so that their height, wavelength and direction of propagation are
changed due to variation in depth, bottom friction and currents still under the
influence of the wind. These effects, refraction, reflection, diffraction and wave-
breaking are equivalent to those of regular waves and were explained in Chapter
3.
If the waves progress out of the low-pressure where they are generated by

45
 Wave-Induced Loads and Ship Motions
4. WIND WAVES 02 December 2008
Page 114

0.5
+
H a Still water
Level (m)

0 level
a-
0.5

T trough
1
60 80 100 120 140
Time (s)
Figure 7.1 Some fundamental definitions of wave properties.
Figure 4.1: Some fundamental definitions of wave properties.
In space, the wavelength, λ, corresponding to the wave period is defined as the
horizontal distance between two consecutive up-crossings in the direction of wave
advance,
the strongandwind
the wave heightofand
into areas amplitude
little withinthe
or low wind, a wavelength are defined
shortest waves in
will gradually
analogy
be dissipated and the longer waves will due to their higher celerity overtake and
to those in time. For an irregular wave as in Figure 7.1 the wave heights the
amplitudes in space are normally not the same as those in time, due to the dispersive
shorter waves, so as a result, at a place far away from the low-pressure, they will
properties of the wave.
be almost regular but with a frequency gradually increasing with time. Such waves
Inare called
this swell. we will not delve further on the generation of the wind waves,
compendium
An example
but direct our interest of atowards
time trace of a waveofisdifferent
the description shown in seaFig. 4.1.
states andInstatistics
the figure
for
also some basic, fundamental definitions are illustrated. The
calculating of design waves. Wave generation is described thoroughly by e.g. wave height, H, is
the difference
Kinsman (1965). in level between a wave crest and the following wave trough, the
positive amplitude, a+ , is the crest height above the still-water level, the negative
7.1
amplitude, a− , is the trough
Characteristics of Wind
depthWaves
below the still-water level, and the wave period,
T , is the time lapse between the up-crossings of the still-water level (in some
treatises the down-crossings are used instead of the up-crossings). Note that an
up-crossing in time corresponds to a down-crossing in space.
In space, the wavelength, λ , corresponding to the wave period is defined as
the horizontal distance between two consecutive up-crossings in the direction of
wave advance, and the wave height and amplitude within a wavelength are defined
in analogy to those in time. For an irregular wave as in Fig. 4.1 the wave heights
and amplitudes in space are normally not the same as those in time, due to the
dispersive properties of the wave.
In this compendium we will not delve further on the generation of the wind
waves, but direct our interest towards the description of different sea states and
Figure
statistics 7.2 Point record
for calculating of awaves.
of design wave elevation. The wave
Wave generation is periods defined
described as zero
thoroughly
by e.g. Kinsman [26]. up-crossing periods.

In Figure 7.2 an example of a time trace of a wave elevation in a point is shown. The
wave
4.1 periods are there definedOF
C HARACTERISTICS as W
theIND
times between the zero up-crossings. Before the
WAVES
advance of computers such traces were in the form of paper graphs and were
evaluated byan
In Fig. 4.2 hand. The result
example of such
of a time traceanofevaluation was a list
a wave elevation inof wave is
a point periods, i, and
shown.TThe
connected waveare
wave periods heights, Hi. From
there defined asthat
the atimes
seriesbetween
of characteristics of the wave can
the zero up-crossings. be
Before
defined and is still used, although nowadays the methods of evaluation
the advance of computers such traces were in the form of paper graphs and were are different,
which will be described later. One used to say that the analysed wave record should
evaluated by hand. The result of such an evaluation was a list of wave periods, Ti ,
contain at least 200 waves for the analysis to be meaningful.
and connected wave heights, Hi . From that a series of characteristics of the wave
Mean zero up-crossing period, often called only zero up-crossing period:
46
 4.1. Characteristics of Wind Waves

Figure 4.2: Point record of a wave elevation. The wave periods defined as zero
up-crossing periods.

can be defined and is still used, although nowadays the methods of evaluation are
different, which will be described later. One used to say that the analysed wave
record should contain at least 200 waves for the analysis to be meaningful.
The mean zero up-crossing period, often called only zero up-crossing period,
read:
1 Nz τ
Tz = ∑ Tzi = , (4.1)
Nz i=1 Nz
where Nz is the number of waves in the record, Tzi the individual zero-crossing
periods of Fig. 4.2 and τ the length of the record. This definition leaves shorter
small waves riding on the long waves uncounted.
For certain purposes we also need to define periods between local time max-
ima or crests, the crest periods, (see Fig. 4.3) from which the mean crest period is
defined as
1 Nc τ
Tc = ∑ Tci = , (4.2)
Nc i=1 Nc
where Nc is the number of waves in the record and Tci the individual crest periods
of Fig. 4.3.
In the example case above, Tz = τ/5, Tc = τ/7 and generally Nc ≥ Nz and
Tc ≤ Tz .
The wave heights are normally referred to the zero crossing definition, and the
zero-crossing wave heights and number of waves are used from here and onwards

Figure 4.3: The same point record of a wave elevation as in Fig. 4.2. The wave
periods defined as wave crest periods.

47
4. WIND WAVES

without index. There are many possibilities to characterise the wave height. The
most common are defined below.

• Mean wave height:

1 N
H̄ = ∑ Hi . (4.3)
N i=1

• Root mean square wave height or the sample variance:

2 1 N 2
Hrms = ∑ Hi . (4.4)
N i=1

• Significant wave height the mean of the highest 1/3 of the N wave heights:

H1/3 = Hs . (4.5)

It can be noted that using down-crossings or up-crossings

√ gives somewhat dif-
ferent values on H1/3 . It can be shown that H1/3 ≈ 2Hrms . Experience shows
also that Hs is approximately equal to the wave height, Hv that is visually es-
timated by an experienced observer. In Fig. 4.4 a confirmation of this from a
Swedish investigation [52] is given. Different light-house keepers may however
have different calibration factors.
For visual observations from ships Nordenström [38] gave the following rela-
tions between visually observed and measured wave height and wave period after
analysis of data from Hogben and Lumb.

H1/3 = 1.68Hv0.75 , (4.6a)

Tz = 0.82Tv0.96 . (4.6b)

A measure to assess the irregularity or broadbandedness of the wave could be

to count the total number of crest maxima and the number of zero-crossing max-
ima. The ratio between these two numbers is Tc /Tz , and can be used to estimate a
“spectral-width parameter”, ε:
s  2
Tc
ε = 1− . (4.7)
Tz

Assessment of wave characteristics directly from time traces is not very repro-
ducible, mostly because the number of waves and consequently Tz are very much
depending on how small undulations of the time trace that is taken into account.
Due to modern computer processing and algorithms developed for Fast Fourier
Transformation (FFT) this latter Fourier technology is used in stead, which makes
possible objective, reproducible filtering.

48
 4.2. Fourier Analysis

Figure 4.4: Comparisons between by light-house keepers visually observed

wave heights and by pressure wave probes measured significant
wave heights. From [52].

4.2 F OURIER A NALYSIS

4.2.1 Fourier series

From analysis we know that every piecewise continuous function can on a finite
interval be approximated by a sum of sine and cosine functions. A point registra-
tion of the wave elevation, ζ (t) = f (t), can thus be written on (0 < t < T ):

N     
1 2π 2π
f (t) ≈ a0 + ∑ ai cos i t + bi sin i t . (4.8)
2 i=1 T T

It is then implicitly assumed that the point registration is repeated for the registra-
tion period, T . See Fig. 4.21 for a simulated point registration of a wave elevation.
The coefficients ai and bi can be calculated by
Z  
2 T +t0 2π
ai = f (t) cos i t dt , i = 0, 1, . . . , N , (4.9)
T t0 T
Z T +t0  
2 2π
bi = f (t) sin i t dt , i = 1, 2, . . . , N . (4.10)
T t0 T

49
4. WIND WAVES

Alternatively, Eq. (4.8) can be written

N    
1 2π
f (t) ≈ a0 + ∑ ci cos i t + εi , (4.11)
2 i=1 T

where
q
ci = a2i + b2i , (4.12)

and
 
bi
εi = arctan − . (4.13)
ai

The Fourier series (4.8) or (4.11) approximates f (t) well if the number of com-
ponents in the series is sufficient. Observe also that the longest wave that can be
detected by a record with the length, T , is a wave with the period, T . The mea-
surement must thus be long enough in relation to the wave periods contained in
the seastate to give any relevant information. Usually 100 to 200 components are
sufficient to approximate a smooth function such as a train of non-breaking waves.

4.2.2 Parsevals equation, orthogonality

In Chapter 3 the energy in a harmonic water gravity wave was shown to be pro-
portional to the square of its amplitude, 0.5ρga2 . For a Fourier series with zero
mean, (a0 = 0), the sum of the square of the component amplitudes equals the
variance of the function itself

1 Z T +t0
1 2 1 N 2 1 2 1 N 2
a0 + ∑ ci = a0 + ∑ ai + b2i = ( f (t))2 dt = Var ( f (t)) .
4 2 i=1 4 2 i=1 T t0
(4.14)
This equation is a form of Parseval’s equation.
The consequence for water waves, where we only take the elevation in relation
to the mean water level into account, is then that the waves are not only geomet-
rically additive but that also the sum of the energy of the components equals the
energy of the composed sea state, at least over one period of analysis, T . Also, all
wave-induced forces and resulting motions are orthogonal and can be calculated
for each component frequency independently of the motions at other frequencies.
Then the resultant irregular motions are given by superposition of the component
motions with due respect to the random phases of the wave components.
The derivation of Fourier series and Parseval’s equation is based on the or-

50
 4.3. Amplitude Spectra and Phase Spectra

Figure 4.5: A 3 hour record of a wave in the Norwegian sea.

thogonality of the trigonometric functions:

Z T +t0   Z T +t0  
2π 2π
1 cos t dt = 1 sin t dt = 0 , (4.15)
t0 T t0 T

Z T +t0      0 , if i 6= j ,
2π 2π
sin i t sin j t dt = T (4.16)
t0 T T  , if i = j ,
2
Z T +t0      0 , if i 6= j ,
2π 2π
cos i t cos j t dt = T (4.17)
t0 T T  , if i = j ,
    2
Z T +t0
2π 2π
sin i t cos j t dt = 0 , (4.18)
t0 T T

where i and j are positive integers. For a more exhaustive treatise on Fourier
analysis see e.g. Hildebrand [18], or other textbooks in calculus.

4.3 A MPLITUDE S PECTRA AND P HASE S PECTRA

The Fourier series, resulting from the analysis of the wave records can be illus-
trated graphically as amplitude and phase spectra. To illustrate this we will in this
section analyse a sample wave record.

4.3.1 Wave record

In Fig. 4.5 a 3 hour record of a wave in the Norwegian sea is shown. In Fig. 4.6 a
blow up of the 3 hour record is shown to see the shape of the waves.

4.3.2 Amplitude, variance and energy spectra

We will now show the result of an FFT analysis of the wave record. The wave
record contains 19 051 discrete values at the time interval 0.556 s. The used FFT
demands the number of values to be 2n where n is an integer. The nearest number

51
4. WIND WAVES

Figure 4.6: A blow up of the 3 hour record in Fig. 4.5.

Figure 4.7: Component amplitudes as a function of angular frequency or ampli-

tude spectrum.

is 214 = 16 384 which gives the useful time-length to TR = 2.5744 h. The maxi-
mum resolution of the discrete variance spectrum is then ∆ f = 1/TR , where TR is
the length of the record, and the resulting amplitude, variance or energy spectra
will in this case contain 8 192 discrete amplitudes. In Fig. 4.7 the resulting com-
ponent amplitudes as a function of angular frequency or the amplitude spectrum
is shown and in Fig. 4.8 the corresponding component phases. The graphs reveal
that the amplitudes, although having stochastic magnitudes, show some kind of
pattern with the dominant amplitudes around 0.7 rad/s, period around 8 s, but that
the phases seem to be completely random.
If the original time record should be reconstructed both the amplitudes and
phases must be saved. However, mostly only the component amplitude spectra
are saved, as any particular wave record is seen as one realisation of many pos-
sible. Even a few wavelengths downwind or upwind the shape of the elevation
graph would have been different due to the different celerity, dispersion, of the
component waves.
Traditionally the information is saved as a variance spectrum or “wave-energy
spectrum”. Recall that the variance of each component is 0.5a2i and the energy
0.5ρg a2i . Such a discrete variance spectrum of the sample wave record is shown
in Fig. 4.9. To make the further discussion more clear the variance spectrum is

52
 4.3. Amplitude Spectra and Phase Spectra

Figure 4.8: Component phases as a function of angular frequency.

blown up in Fig. 4.10.

To make the appearance of the discrete variance spectrum look more “deter-
ministic” and be able to compare it to other spectra or standard spectra it can be
smoothed. The sample record here was, after some trials, smoothed by taking a
centred arithmetic mean of 201 discrete variances. The result of this smoothing is
shown in Fig. 4.11.
For the comparison between spectra, deducted from records of various record
lengths TR , sampling intervals, ∆t, etc., the discrete variance spectrum has to be
transformed to a continuous density, spectrum. This is done by dividing the dis-
crete variances by the frequency division ∆ω:

a2i
Si = . (4.19)
2∆ω
The result of this action is shown in Fig. 4.12, and we now have a spectrum
that resembles the standard, empirical spectra used in the ship industry. Such stan-
dard spectra are the result of assembling and taking the mean of many measured
spectra. We will try to fit some standard spectrum to the smoothed measured
spectrum. But to do that we will describe the standard spectra, and how wave
characteristics are derived from spectra by using spectral moments.

4.3.3 Standard spectra

A fundamental spectrum is the Pierson-Moscowitz (PM) spectrum, which should
describe wave spectra for fully developed sea, or fully arisen sea (FAS), when a
constant wind blowing infinitely long cannot increase the energy in the waves, but
the energy transfer is balanced by dissipation. This spectrum is a one-parameter
spectrum completely described by the wind speed:
ω0 4
SPM (ω) = α g2 ω −5 e−0.74( ω ) , (4.20)

where α = 0.0081 is Phillip’s constant, g the earth acceleration, ω0 = g/U19.5 and

U19.5 is the wind speed at the height 19.5 m above still water level. See Fig. 4.13
for examples of PM spectra for some wind speeds.

53
4. WIND WAVES

Figure 4.9: Discrete variance as a function of angular frequency.

Figure 4.10: The discrete variance spectrum of Fig. 4.9, blown up between 0
and 2 rad/s, as a function of angular frequency.

Figure 4.11: The smoothed discrete variance spectrum, blown up between 0 and
2 rad/s, as a function of angular frequency.

54
 4.3. Amplitude Spectra and Phase Spectra

Figure 4.12: The smoothed density variance spectrum or “wave energy spec-
trum” of the sample record.

For this spectrum T01 = 1.086T02 and Tm = 1.408T02 = 1.14T0 . The periods
T01 and T02 are estimates of the zero-upcrossing period, Tz , and will be defined
in the next section. Tm is the modal period or the period for the spectral peak.
T0 = 2π/ω0 .
A variant of the one-parameter PM-spectrum is the International Towing Tank
Conference (ITTC) spectrum:

− 3.11 −4
2 ω
SIT TC (ω) = α g2 ω −5 e Hs , (4.21)

In this spectrum the significant wave height, Hs , is used instead of the wind speed
or mean period.
Mostly the sea state is, however, not fully developed as the wind speed and di-
rection change, the fetch is too short, or the duration is not long enough, especially
for strong winds and high waves. Then two-parameter spectra for developing seas
can be used, e.g. some where the wave height and frequency are the parame-
ters. This was originally proposed by Bredtschneider and offers more flexibility,
because the energy of the spectra can be placed at arbitrary locations on the fre-
quency axis:
1.25 2 ωm4 −1.25( ωm )4
SB (ω) = H e ω , (4.22)
4 s ω5

where ωm is the modal angular frequency (maximum of spectrum).

Such spectra belonging to the PM-family are also, somewhat incorrectly, re-
ferred to as PM-spectra. Two other spectra are the International Ship Structures
Congress (ISSC) spectra, see Eq. 4.23. The periods T01 a T02 are estimated by
moments of measured spectra and are both estimates of the zero-crossing period.

55
4. WIND WAVES

Figure 4.13: PM spectra for some wind speeds.

56
 4.3. Amplitude Spectra and Phase Spectra

Spectral moments and the periods T01 and T02 will be defined in the next section.
 2
 −4
T01
 Hs  −0.44
SISSCa (ω) = 0.11   2  ω −5 e
ω
2π
, (4.23a)
T01
2π
 2
 −4
T02
1  Hs  −5 −( π1 ) 2π ω
SISSCb (ω) =   2  ω e . (4.23b)
4π T02
2π

An almost identical spectrum1 as (4.23a) was recommended by ITTC for

“open sea spectral formulation” [21].
Two-parameter spectra still give too little freedom to reproduce realistic spec-
tra of developing sea. In 1973 Hasselman et al. [17] published the JONSWAP
spectrum, which was one outcome from the Joint North Sea WAve Project:
2
e
− 12 ( ω−ω
σ ωm )
m
SJONSWAP (ω) = SPM (ω) γ
2
2 −5 −1.25( ωωm )
4 − 12
e ( ω−ω
σ ωm )
m
=αg ω e γ , (4.24)
2
− 12 ( ω−ω
σ ωm )
m
where γ e is the peak enhancement factor, ω the angular frequency, ωm
the modal angular frequency (maximum of spectrum) and σ is a spectral width
parameter related to the “standard deviation” of the peak enhancement factor. The
spectral width parameter is
(
σa , if ω < ωm ,
σ=
σb , if ω > ωm .

Recommended values are, when the fetch, F, and the wind speed, U10 , at
10 m height is used: γ = 3.30 , σa = 0.07 , σb = 0.09 , α = 0.076F0−0.22 , ωm =
7π(g/U10 )F0−0.33 . Here F0 = gF/U10 2.

Another formulation of the spectrum (4.24) is recommended by ITTC for

“long crested limited fetch sea spectral formulation” [21]. The peak enhancement
factor is identical2 , but the SPM (ω) factor is different:
!
  (0.191 ω T01 −1)2
−
Hs2 − T 4944ω 4 e 2ω
SJ (ω) = 155 4 5 e 01 γ . (4.25)
T01 ω

The expression (4.25) gives identical results to (4.24) provided T01 and Hs are the
same.
1 Notations are different but numerical difference is in the order of 1 %.
2 Using T1 instead of ωm it looks different.

57
4. WIND WAVES

The JONSWAP spectrum is in common use for design of drilling platforms

in the offshore industry because it offers more flexibility with its five parameters,
and can produce more realistic spectra. The parameters are then chosen from
wave statistics combined with systematic parameter fitting.
For the JONSWAP spectrum T01 = 1.073T02 = 0.834Tm = 0.924T−1 .
Note in Fig. 4.14 the different characteristics of the two spectra with the sharp
peak of the JONSWAP spectrum. In Fig. 4.14 the JONSWAP spectrum has a
larger variance and thus significant wave height. In Fig. 4.15 the JONSWAP spec-
trum and PM spectrum have the same variance and significant wave height but
different modal periods.
In Fig. 4.16 fits of standard spectra are shown to the measured smoothed spec-
trum (S j ) of Fig. 4.12 using calculated T01 for the ISSCa spectrum, T02 for the
ISSCb spectrum, Hs for the ITTC spectrum and a manual “best” fit of the five
parameters of the JONSWAP spectrum. It is obvious that only the JONSWAP
spectrum has enough parameters to make a good fit possible.
Sometimes a sea state is a result of superposition of local wind waves and
swell from distant storms. Then the spectrum must have double peaks. Such
spectra are the six parameter spectrum of Ochi and Hubble or the spectrum by
Thorsethaugen. This will not be referred here.

4.3.4 Moments and wave characteristics

The wave characteristics like the significant wave height, mean period etc. are
nowadays estimated indirectly from moments of the spectra and not from the time
records. This yields more reproducible results. The spectral moments are defined
as Z ∞
mn = ω n S(ω) dω , (4.26)
0
where n is an integer. A limitation with this is, however, that the fourth moment of
the PM-family spectra including the JONSWAP spectrum is infinite, and for mea-
sured spectra depends on the resolution or measuring intervals. St Denis (1980,
See PNA, 1989) proposes that the integration for m4 is taken up to 5ωm .
The 0th moment gives the variance, σ 2 , significant wave height, Hs , and root
mean square wave height, Hrms , as

Var(ζ (t)) = m0 = σ 2 , (4.27)

√ √
Hs ≈ Hm0 = 4 m0 = 4σ = 2Hrms (4.28)
√ Hs
σ = m0 ≈ .
4
The zero-crossing period, Tz , is for a broad spectrum best estimated as
m0
Tz ≈ T01 = 2π , (4.29)
m1

58
 4.3. Amplitude Spectra and Phase Spectra

Figure 4.14: A JONSWAP spectrum and PM spectrum with the same modal
frequency, ωm = 0.878ω0 = 0.431 rad/s.

Figure 4.15: A JONSWAP spectrum and PM spectrum with the same significant
wave height Hs .

59
4. WIND WAVES

Figure 4.16: Fit of ISSCa, ISSCb, ITTC and JONSWAP spectra to the measured
smooth spectrum (S j) shown in Fig. 4.12 using calculated T01 , T 02
and Hs respectively for the first three and a manual fit of the five
parameters of the JONSWAP spectrum.

and for a more common narrower spectrum to

r
m0
Tz ≈ T02 = 2π . (4.30)
m2
Sometimes, in the context of the significant wave, the notion of significant
wave-period is used. This is a little shorter than T01 .
The crest period, Tc , is estimated as
r
m2
Tc ≈ T24 = 2π . (4.31)
m4
The spectral width or broadbandedness is estimated by (Longuet-Higgins [31])
 2  2  2 
2 Tc T24 m2
ε = 1− ≈ 1− = 1− . (4.32)
Tz T02 m0 m4
Another proposal was made by Mollison [34]:
s
m0 m−2
ε0 = −1.
m2−1
Mean period of component waves
Z ∞ Z ∞
2π
T S(ω)dω S(ω)dω
ω
T−1,0 = Z0 ∞ = 0Z
∞
S(ω)dω S(ω)dω
0 0Z
∞
2π ω −1 S(ω)dω
m−1
= Z0 ∞ = 2π . (4.33)
S(ω)dω m0
0

60
 4.4. Synthesised Waves

Average wave length between zero up-crossings

r
g g m0
hλ i = Tz Tc ≈ T02 T24 = 2πg . (4.34)
2π 2π m4
Variance of the slope of the wave surface
*  +
∂ζ 2 m4
= 2. (4.35)
∂x g

In Fig. 4.17 the same wave record as in Fig. 4.5 is shown but now lines for plus
and minus half the calculated significant wave height have been added in order to
give a feeling for the connexion between the significant wave height and the wave
record. In Fig. 4.18 T01 , T02 and T24 to show that these are of the expected order
of magnitude as the individual zero crossing periods, Tzi , and wave crest periods
Tci . However, as would be expected it is almost impossible to find an individual
period exactly equal to the corresponding mean.

4.4 S YNTHESISED WAVES

The observation of a wavy lake or ocean surface reveals that the surface normally
is pretty confused with significantly varying wave heights, periods and wave-
lengths, as well as waves progressing in various directions. By using a the spectral
Fourier model described in the previous section, i.e. assuming that the wavy sur-
face is built up by an addition (superposition) of linear waves one can describe
records of such wavy surfaces and reproduce or synthesise them.
The water-surface elevation in a point x is then approximated by a sum

ζ (t) = ∑ ai cos (ki x − ωi t + εi ) , (4.36)

i

where ai , ωi , ki and εi are the amplitude, circular frequency, wave number and
phase of the ith component, respectively. A fundamental prerequisite for the
waves to be modelled in such a way is that they should be part of a stationary
process whose statistical properties may not change abruptly with time.
Further, the sum above is usually considered to be a Fourier series with or-
thogonal components on a finite time interval, e.g. TR = 20 minutes. This has
the consequence that the elevation is repeated exactly every time interval. The
repeating time, TR , implies the frequency division, ∆ f = 1/TR . Now, choosing
amplitudes of the components from some established standard spectrum and as-
signing random phases distributed evenly between 0 and 2π radians (Fig. 4.20), a
synthetic irregular wave can be formed. In Figs. 4.19 to 4.21 the steps leading to
such a wave is illustrated. We have used a PM spectrum.
The amplitudes are normally chosen as
p
ai = 2S(ωi )∆ω , (4.37)

61
4. WIND WAVES

Figure 4.17: The sample wave elevation with +/- half the significant wave Hs ≈
Hm0 height for comparison.

Figure 4.18: The sample wave elevation with characteristic periods estimated
from its spectrum.

62
 Wave-Induced Loads and Ship Motions 4.4. Synthesised Waves
02 December 2008
Page 132

ai

0 5 10 15 20 25 30 35
i

Figure Fig.
4.19:7.19 Thirty-two
Thirty-two component
component amplitudes
amplitudes chosen
chosen by help
by help of aofPM
a PM
spec-
trum.spectrum

εi

0
0 10 20 30 40
i

Fig. 7.20 Thirty-two random component phase angles between 0 and 2π

radians

The assigned amplitudes and phase angles inserted into equation (7.35) will then
produce the wave “record” shown in Figure 7.21, where the randomness seem to be
OK within each time interval, TR. The simulated synthetic wave is shown to repeat
itself, which is a consequence of using components evenly distributed over the
frequency range with the division ∆f. In fact the repetition time TR = 1/∆f, and vice
Figureif 4.20:
versa TR is the length ofrandom
Thirty-two a record or observation
component phasethe highest
angles frequency
between 0 andresolution
2π ra-
would be ∆f =dians.
1/TR. (See Bendat and Piersol, 1986)

t 64 t 128

which will produce a discrete spectrum of the same shape as the original PM
Elevation

spectrum, Fig. 4.19. The choice can be criticised because it does not produce
a realistic amplitude spectrum as the one in Fig. 4.7, and consequently the ran-
domness of the resulting synthesised record will be too small. Instead random
amplitudes from a Gaussian distribution withTime mean, ai , and standard deviation
2 1/2
(0.5ai ) should be used according to Tucker et. al [48].
Fig. 7.21
The assigned Simulatedand
amplitudes water surface
phase elevation.
angles Note
inserted theEq.
into repeated pattern.
(4.36) will then
produce the wave “record” shown in Fig. 4.21, where the randomness seem to
be OK within each time interval, TR . The simulated synthetic wave is shown to
repeat itself, which is a consequence of using components evenly distributed over
the frequency range with the division ∆ω. In fact the repetition time TR = 1/∆ f ,

63
 The assigned amplitudes and phase angles inserted into equation (7.35) will then
produce the wave “record” shown in Figure 7.21, where the randomness seem to be
OK within each time interval, TR. The simulated synthetic wave is shown to repeat
itself, which is a consequence of using components evenly distributed over the
4. WIND range with the division ∆f. In fact the repetition time TR = 1/∆f, and vice
WAVES
frequency
versa if TR is the length of a record or observation the highest frequency resolution
would be ∆f = 1/TR. (See Bendat and Piersol, 1986)

t 64 t 128

Elevation
Time

Fig. 7.21 Simulated water surface elevation. Note the repeated pattern.
Figure 4.21: Simulated water surface elevation. Note the repeated pattern.

and vice versa if TR is the length of a record or observation the highest frequency
resolution would be ∆ω = 2π∆ f = 2π/TR . (See Bendat and Piersol, 1986)

4.4.1 Directional sea

The spectral model can be extended to include wave components propagating in
different directions by
N K 

ζ (x, y,t) = ∑ ∑ ai j cos ki x cos(θ j ) + y sin(θ j ) − ωi t + εi j , (4.38)
i=1 j=1

with
K
ai = ∑ ai j . (4.39)
j=1

where θ j is the angle between the x-axis and the direction of propagation of the
component.
The variance of the directional waves are distributed according to some spread-
ing function, D(θ ), fulfilling the demand that the integral around the horizon
Z 2π
D(θ )dθ = 1 , (4.40)
0

so that the directional spectrum becomes

S(ω , θ ) = S(ω) D(θ ) . (4.41)

Usually the spread is restricted to +/- 90◦ around the main direction of propaga-
tion.
The simplest spreading functions are independent of the wave frequency, and
a commonly used one is the cosine square distribution:
(
2 2 π
D(θ ) = π cos (θ ) , if |θ | < 2 , (4.42)
0, otherwise .

64
 4.4. Synthesised Waves

Figure 4.22: A 3D wire plot of a cos2 PM spectrum.

The cosine square distribution is illustrated in Fig. 4.22.

Real waves exhibit a more complicated pattern with longer, long-period waves
having less directionality than shorter, short period waves. One directionality
function taking care of that is the SWOP distribution (Pierson et al., 1960):
(
1 π
D(θ ) = π (1 + (0.5 + 0.82R) cos(2θ ) + 0.32R cos(4θ )) , if |θ | < 2 ,
0, otherwise ,
(4.43)
where  4
ωU19.5
− 21
R=e g
,
and in which U19.5 is the wind speed at 19.5 m height.
Often the directionality of the sea is ignored, especially for severe sea states
as the directionality then is smaller.

65
 5
WAVE STATISTICS AND PROBABILITY

Up till now we have only discussed how to describe sea states and the statistical
properties of the water surface elevation in these. This is often called short-term
statistics. However, to be able to design structures and ships, we need know the
probability of the appearance of sea states of different wave heights and periods
or the probability of extreme individual waves during the expected life time of
the structures. This is often called long-term statistics. Founded on the long-term
statistics, the probability of extreme loads on fixed stiff structures, and extreme re-
sponses of flexible structures or floating bodies. For a flexible or floating structure
it is not certain that the “worst sea state” or the “worst wave” will give the largest
stress or motion. The response probabilities are also of interest for estimation of
fatigue, downtime of floating offshore production or drilling platforms, frequency
of green water, slamming, accelerations etc. for ships.
A basic term used in design codes and guidelines is the return period or recur-
rence interval. It is an estimate of the mean interval of time between events like
an earthquake, flood or river discharge flow of a certain intensity or size. It is a
statistical measurement denoting the average recurrence interval over an extended
period of time, and is usually required for risk analysis (i.e. whether a project
should be allowed to go forward in a zone of a certain risk) and also to dimension
structures so that they are capable of withstanding an event of a certain return
period (with its associated intensity of a design quantity).
In the guidelines concerning loads and load effects on load-bearing structures
in the petroleum activity in Norwegian waters [39] it is stated that one should
use a design sea state given by a JONSWAP spectrum with one-hundred-year-
100 yr 100 yr
return significant wave height, Hm0 , combined with the peak period, Tp ,
or other spectra with larger probability and other shape if such spectra give larger
load effects. There is also an option to use a one-hundred-year design wave,
100 yr 100 yr 100 yr 100 yr
Hmax = 1.9Hm0 combined with the period Tmax = 0.92Tp . For float-
ing structures, especially, the wave period should be varied in order to investigate

67
5. WAVE STATISTICS AND PROBABILITY

the responses for shorter waves with maximum steepness. The factor 1.9 above is
founded on the assumption that the duration of the sea state is taken as 3 h. One
can note that the 100 year maximum wave height derived in this way is a little
smaller than if the maximum probable wave height is taken as the most probable
largest of all individual waves. Compare results in Table 5.10 and Table 5.12 for
the North Atlantic.
In Section 5.1 we discuss the concept of risk. Section 5.2 is a short account of
design with safety factors versus with load and material coefficients. In Section
5.3follows short-term statistics of waves and in Section 5.5 long-term statistics.

5.1 T HE C ONCEPT OF R ISK

Before we look at the limit states for design we will acquaint ourselves with the
concept of risk. A hazard is a source of danger but does not contain any likelihood
or actual impact it will have on people, environment or economics. Risk combines
both likelihood and impact, and a risk analysis tries to answer the questions:

• What can happen?

• How likely is it to happen?

• Given that it occurs, what are the consequences?

Risk can then be expressed as a combination of the probability, E, of the event

and its consequence. Some authors simply give the risk, R, as the product of the
probability, E, and some measure of the consequence, C, i.e. R = CE. The conse-
quence can be e.g. the number of killed or wounded people due to the undesired
event or the economic cost i.e. lost income, repair cost etc. In the latter case R can
be named the risk cost. See Graham [14] for a more thorough discussion.

5.1.1 Design event at given risk

Denote the probability that an event with the return period, TR , shall occur or
be exceeded a certain year by Q. An estimate of this probability is Q = 1/TR .
The probability for not exceeding is then P = 1 − Q in average per year, and the
probability for the event not to be exceeded any year during a lifetime, L, of the
design object is the product of the probabilities for each year i.e.:

P(no year) = (1 − Q)L , (5.1)

and consequently the probability, E, that it would be exceeded at least once during
the L years:
 
L 1 L
E = P(at least once) = 1 − (1 − Q) = 1 − 1 − . (5.2)
TR

68
 5.1. The Concept of Risk

Table 5.1: The probability, E, of an event with the return period TR (years) to be
exceeded at least once during the lifetime L (years).

TR LR
1 5 10 50 100 500 1000 5000 10000
1 1 1 1 1 1 1 1 1 1
5 0.2 0.672 0.893 1 1 1 1 1 1
10 0.1 0.41 0.651 0.995 1 1 1 1 1
50 0.02 0.096 0.183 0.636 0.867 1 1 1 1
100 0.01 0.049 0.096 0.395 0.634 0.993 1 1 1
500 0.002 0.01 0.02 0.095 0.181 0.632 0.865 1 1
1000 0.001 0.005 0.01 0.049 0.095 0.394 0.632 0.993 1
5000 2E-4 0.001 0.002 0.01 0.02 0.095 0.181 0.632 0.865
10000 1E-4 5E-4 0.001 0.005 0.01 0.049 0.095 0.393 0.632

This function is shown in Table 5.1. For L = TR it asymptotically approaches

1−e−1 = 0.632 when TR increases. The probability, E, for exceeding the hundred-
year wave is thus 63 % in 100 years.

Example 5.1
In many rules it is stated that the weighted probability per year of a
combination of design events should be 10−4 . This means a return period
of 10 000 years. One can then ask: How large is the probability, E, that it
would occur during a lifetime of 50 years?

For TR = 10 000 years and L = 50 years the table gives E = 0.005, or a

0.5 % risk of occurance.

One can invert the problem and ask: If you accept a probability, E, during a
lifetime, which return period should be used? This can be solved from Eq. (5.2):
1
TR = . (5.3)
1 − (1 − E)1/L

Example 5.2
A probability, E, of 1 % is accepted for a design event to be exceeded in
the lifetime 50 years. Which return period should be used?

For E = 0.01 = 10−2 and L = 50 Table 5.1 gives TR = 5000 years. Alter-
natively, Eq. (5.3) yields 4975 years.

69
5. WAVE STATISTICS AND PROBABILITY

Observe that the real probability of damage certainly is different, as one uses
safety factors or load and material coefficients at the design to safeguard the limi-
tation in knowledge of statistical distributions of events, loads, load effects, mate-
rial properties as well as the quality of construction work. Finally it is actually so
that damage often appears due to unforeseen types of events and human mistakes.

5.2 D ESIGN A PPROACHES

There are typically two approaches for designs. One is to use safety factors i.e. to
calculate the load effect, e.g. a stress, for given design loads and compare the load
effect with a given criteria, e.g. the breaking strength, divided by a safety factor
S > 1 depending of construction loads etc. The other is to use a safety format
with load and material coefficients and assign coefficients γ f > 1 to each load,
and material coefficients γm < 1 to each component. Then check that the total
load effect calculated with the load coefficients is smaller than the design capacity
calculated with the material coefficients.

5.2.1 Design with safety factors

In many rules e.g. from DnV [11], the British Health and Safety Executive [20],
the Norwegian Marine Directorate [37], design with safety factors are applied.
Then the safety factor or ratio, S, between permissible load effect, R, and cal-
culated load effect, F, exceed certain minimum values for specified load cases.
Examples of load effects are motions, clearance distance or tension.
As an example of safety factors in Table 5.2 the safety factors for the forces in
mooring lines according to DnV’s POSMOOR rules for floating units not placed
close to other installations.

5.2.2 Safety format with load and material coefficients

The general requirement to achieve sufficient safety is that the design load effect,
Sd , not exceed the design capacity, Rd :

Sd ≤ Rd . (5.4)

The equality, Sd = Rd , defines a limit state.

The design load is normally formed by the product of a characteristic load, Fk ,
and a load coefficient, γ f . The load effect, S, is a function of the design load, so
that adding the load effects (in a linear system) you get the design load effect as:

Sd = ∑ S(γ f Fk ) . (5.5)

The characteristic loads shall be established for each load category, Table 5.3, and
limit state, Table 5.4.

70
 5.2. Design Approaches

Table 5.2: Safety factors According to DnV’s mooring rules for mobile offshore
units (POSMOOR [11]).

Operation condition Quasi-static Dynamic

analysis analysis
Intact system 1.80 1.50
Survival Transient motion after single-line failure 1.10 1.00
Temporary mooring after single-line failure 1.25 1.10
Intact system 2.70 2.30
Operation Transient motion after single-line failure 1.40 1.20
Temporary mooring after single-line failure 1.80 1.50
The extreme weather condition to be considered is the most unfavourable of:
1. 10 min average wind speed and sea state corresponding to a 100-year return period
combined with a 10 year-return-period current;
2. Current velocity and sea state with 100-year return period combined with a 10 min
average wind speed with a 10-year-return period.

Table 5.3: Load categories.

Legend Load category Examples of typical loads

P Permanent load
L Variable functional loads payload, ballast, mooring, etc.
D Deformation loads pretension, temperature, subsidence, etc.
E Environmental loads wind, waves, currents, tides,
varying water levels, ice and snow, etc.
A Accidental loads collision by ships, breakage of mooring line,
flooding of flotation body or hull compartment,
fire, etc.

Table 5.4: Limit states.

Legend Limit state

ULS Ultimate Limit State
FLS Fatigue Limit State
ALS Progressive or Accidental Limit State
SLS Serviceability Limit State

71
5. WAVE STATISTICS AND PROBABILITY

Table 5.5: Load coefficients, γ f , and their combinations in the ultimate limit
state, ULS, According to DnV’s tentative rules for fish farms.

The most unfavourable load combination Load categories

of (a) and (b) below shall be applied P L D E
(a) 1.3 1.3 1.0 0.7
Combination of design loads
(b) 1.0 1.0 1.0 1.3∗
∗ For unmanned floaters this coefficient can be taken as 1.15.

For permanent loads the coefficient shall be taken as 1.0 if this

is less favourable. The coefficients should be the same for the
whole structure.

Table 5.6: Material coefficients γm .

Limit state Steel struct. Concrete struct. Chain Wire rope Synthetic rope
Concrete Reinforced
ULS 1.15 1.40 1.25 1.50 1.30 3.00 (2.50)∗
FLS 1.00 1.20 1.10
ALS 1.00 1.20 1.10 1.30 1.15 2.50 (1.30)∗
SLS 1.00 1.00 1.00
∗ Coefficients within brackets apply if there is satisfying documentation of the rope.

Load effects can be:

• Motion;

• Acceleration;

• Stress;

• Deformation.

An example of is given in Table 5.5.

We also note that:

• In the accidental limit state, ALS, all load coefficients are set to 1.0.

• The environmental loads for ULS and ALS shall have a yearly probability
of 0.01 or less.

• The fatigue limit state, FLS, shall be investigated. All load coefficients are
set to 1.0.

Example of material coefficients are given in Table 5.6.

72
 5.3. Short Term Statistics

5.3 S HORT T ERM S TATISTICS

The water elevation in an irregular wave as a function of time constitutes a stochas-

tic process. We will show that its variation around the still water elevation is a
Gaussian process with zero mean, as a result of the addition of many harmonic
components ζi (t) with uniformly distributed phase angles εi :

ζi (t) = ai cos (ωi t + εi ) . (5.6)

The means of these component processes are all zero;

Z T Z T
1 1
µi = ζi (t)dt = ai cos(ω1 t + εi )dt = 0 , (5.7)
T 0 T 0

and the variances around the means are

Z T Z T
1 1 1
Var(ζi (t)) = σi2 = 2
(ζi (t)) dt = (ai cos(ω1 t + εi ))2 dt = a2i . (5.8)
T 0 T 0 2

5.3.1 The distribution of one harmonic component

If the phase angle εi of ζi (t) is randomly distributed on the interval [0 , 2π) then,
according to a theorem in statistics, its frequency function is
dεi
p(ζi (t)) = p(εi ) . (5.9)
dζi
If further εi is uniformly distributed on the interval then
1
p(εi ) = . (5.10)
2π
Solving εi from Eq. (5.6)
 
ζi
εi = arccos − ωi t , (5.11)
ai
and thus
dεi 1
=q , (5.12)
dζi 2
a i − ζi2

which, however, only is defined on the interval [0 , π) and therefore p(εi ) has to
be doubled on this interval. Finally substituting Eqs. (5.10) and (5.12) into (5.9)
the frequency distribution of ζi is
1
p(ζi ) = q , −ai ≤ ζi ≤ ai . (5.13)
π a2i − ζi2

This frequency function is shown in Fig. 5.1 and its distribution function in Fig.
5.2. In the same figures also the corresponding functions of a Gaussian process

73
 5. WAVE STATISTICS AND PROBABILITY

Wave-Induced Loads and Ship Motions

02 December 2008
Page 136

1 Wave-Induced Loads and Ship Motions

02 December 2008
Page 136
( )
p ζi

(
dnorm ζi , 0⋅ m , σ ) 0.5
1

( )
p ζi
0
(
dnorm ζi , 0⋅ m , σ ) 0.52 1 0
ζi
1 2

Figure 7.23 Comparison between the frequency distribution of the level of a

Figure 5.1: Comparison between the frequency distribution of the level of a har-
0 harmonic function and the frequency distribution of a Gaussian
2 1 frequency distribution
monic function and the 0 of a1Gaussian process
2
process with the same standard deviation
with the same standard deviation. ζi
Figure 7.23 Comparison between the frequency distribution of the level of a
1 harmonic function and the frequency distribution of a Gaussian
process with the same standard deviation
( )
P ζi
0.5 1
(
pnorm ζi , 0⋅ m , σ )
P ( ζi)
0.5
(
pnorm ζi , 0⋅ m , σ ) 0
2 1 0 1 2
ζi
Figure 7.24 0
Comparison between the accumulative distribution of the level
2of a harmonic
1 function and
0 the accumulative
1 distribution
2 of a
Gaussian process with the
ζi same standard deviation
Figure 7.24 Comparison between the accumulative distribution of the level
The mean
Figureof5.2:
theComparison
frequency-distribution
of abetween
harmonic function
thefunction andisdistribution
accumulative of the level
the accumulative of a
distribution of a
harmonic function and the accumulative distribution of a Gaussian
Gaussian process with the same standard deviation
ai process with the same standard deviation.
µ i =of!the
The mean
ζ i pfrequency-distribution
(ζ i )dζ i = 0
function is
…(7.50)
− ai

ai

µ i = ! ζ i p(ζ i )dζ i = 0
and its variance …(7.50)
− ai
ai
1 2
i = ! ζ i p (ζ i ) dζ i =
and itsσvariance
2 2
ai …(7.51)
− ai
2
i a
1 2
!−a ζarei pthe
74
which, ofσcourse
2
= 2
(ζ i )same
dζ i =as for
ai the time function itself. …(7.51)
i
i
2
The distribution of the water level in an irregular wave
which,
The wateroflevel
course are the same
variation around asthe
for mean
the time
can,function itself.
as stated before, be written as a sum of
cosine functions with varying frequencies, ωi, and random phase angles, εi.
The distribution of the water level in an irregular wave
 5.3. Short Term Statistics

with the same mean and standard deviation are shown. One can note that the
frequency function is far from similar to the Gaussian frequency function, while
the distribution functions are more similar.
The mean of the frequency-distribution function is
Z ai
µi = ζi p(ζi )dζi = 0 , (5.14)
−ai

and its variance Z ai

1
σi2 = ζi2 p(ζi )dζi = a2i , (5.15)
−ai 2
which, of course, are the same as for the time function itself.

5.3.2 The distribution of the water level in an irregular wave

The water level variation around the mean can, as stated before, be written as a
sum of cosine functions with varying frequencies, ωi , and random phase angles,
εi :
ζ (t) = ∑ ζi (t) = ∑ ai cos (ωi t + εi ) . (5.16)
i i

Each component is a stochastic variable with mean, µi , and variance, σi2 . Then,
according to the central limit theorem the sum of many components, ∑i ζi (t), ap-
proaches a Gaussian process with the mean

µ = ∑ µi = 0 , (5.17)
i

and the variance

1
σ 2 = ∑ σi2 = ∑ a2i = m0 , (5.18)
i i 2

where m0 is the area under the wave spectrum or 0th moment and σ the standard
deviation.
Thus the water level in an irregular wave is Gaussian distributed with the fre-
quency function
2 2
1 −ζ 1 −ζ
p(ζ ) = √ e 2σ 2 = √ e 2m0 . (5.19)
σ 2π 2πm0

This is also the case for all derived responses.

Using this knowledge of the stochastic Gaussian process in a sea state, e.g.
the fraction of time during which the water level is above a certain level can be
estimated, the fraction of time green water is flowing in over the deck or the
fraction of time the ship bottom is above the water surface.

75
5. WAVE STATISTICS AND PROBABILITY

Example 5.3
How large part of the time will the surface level be above 1 m around a
platform leg in a sea state with the significant wave height, Hs = 3 m?

Assuming the duration of the sea state is T and the sum of all times the
water level is above ζ = 1 m is t, the fraction of time is
Z ∞ 2 Z ∞ ζ2
t 1 −ζ2 1 − 2m
= P(ζ > 1 m ) = √ e 2σ dζ = √ e0 dζ .
T 1 σ 2π 2πm0 1
(5.20)
√
As Hs = 4 m0 and σ = Hs /4, Eq. (5.20) will give 0.0912 e.g. by help of
the standard normal distribution with (1 m)/σ = 1.33 as found in table in
Standard Mathematical Tables:

1 − F(1.33) = P(ζ > 1 m ) = 0.0912 ,

i.e. the surface level will be above 1 m 9 % of the time.

5.4 E XTREMES IN A S EA S TATE

For a reasonably narrow-banded Gaussian sea state the probability density for the
level of wave crests or local maxima, ζmax , is approximately Rayleigh distributed
with the frequency function
2
ζmax − ζ2m
max
p (ζmax ) = e 0 (5.21)
m0

which has the variance of the sea surface elevation or 0th moment of the spectrum
as its sole parameter. This expression can be used for maxima of all other derived
processes too.
The probability density for the wave heights, H, is also approximately Rayleigh
distributed with the frequency function

H2
H − 8m
p (H) = e 0. (5.22)
4m0

Its distribution function or probability that the wave height H < Hq is then
Z Hq 2 H2
H H
− 8m − 8mq
P(H < Hq ) = e 0 dH = 1 − e 0 . (5.23)
0 4m0

The probability Q that the wave height should be higher than Hq is then

H2
− 8mq
Q(H > Hq ) = 1 − P(H < Hq ) = e 0 . (5.24)

76
 5.4. Extremes in a Sea State

From this follows that the wave Hq that is exceeded with the probability Q is
s  
1
Hq = 8m0 ln . (5.25)
Q

Example 5.4
How many waves will be greater than 5 m during 6 hours in a sea state
with the significant wave height, Hs = 3 m and mean wave period 10 s?

The number of waves during 6 hours is

3600
N=6 = 2160 .
10
The relative number of waves exceeding Hq = 5 m of these is (using that
16m0 = Hs )
Hq2 52
− − 2
Q(H > Hq ) = e Hs2 /2 =e 32 /2 = e−2(5/3) = 0.00387

The number of waves nq larger than 5 m is finally

nq = QN = 8.35 ≈ 8

Of course one cannot expect that the number of waves larger than 5 m
should be just 8 in all such sea states, but the number should be seen as
an expected number in such sea states. Many sea states should give the
arithmetic mean around 8.4.

According to the given example an approximate value of Q is:

nq
Q= (5.26)
N
and consequently s  
N
Hq = 8m0 ln . (5.27)
nq
is the wave height that is exceeded nq times of N waves.

5.4.1 Extreme wave height

The expected value of the wave height that is exceeded once of N waves is then
p
Hmax = 8m0 ln(N) , (5.28)

because this probability is approximately QN = 1/N.

77
5. WAVE STATISTICS AND PROBABILITY

Table 5.7: Maximum wave height.

√
Q Hq / m0 Hq /Hrms Hq /Hs
10 % 4.29 1.52 1.07
1% 6.07 2.15 1.52
0.1 % 7.43 2.62 1.86
0.01 % 8.58 3.03 2.15

The wave height that is exceeded by 10, 1, 0.1 and 0.01 % of the number of
waves in a sea state can also be assessed by Eq. (5.25) and are sometimes given
in literature as references. See Table 5.7.
The maximum wave, Hmax , in a sea state is often given as the most probable
maximum wave height of a thousand waves, that is, corresponding to 0.1 % above.
In a sea state with the mean period 10 s this corresponds to a duration of 10 000 s
or 2.8 hours. In a certain sea state the largest wave can of course be higher or
lower but the mean of many measurements or simulations should be 1.86Hs .
The same methods for estimating periods, variances and extremes, that have
been used for the waves in this chapter can be applied to all derived motions,
accelerations, loads, etc in Chapter 13.

5.5 L ONG T ERM WAVE S TATISTICS

5.5.1 Wave statistics

In Chapter 4 we treated the statistical distribution of wave heights within a sea
state and found that these wave heights were Rayleigh-distributed with the fre-
quency distribution
H − 8m H2
p(H) = e 0 (5.29)
4m0
if the water level constitutes a stationary Gaussian process.
In the context of this chapter it would be practical to exchange the parameter
√
m0 for Hs using Hs = 4 m0 so that

4H −2( HH )2
p(H) = e s . (5.30)
Hs2

The significant wave height varies in its turn between the recorded sea states rep-
resented by point measurements of water elevations. Often the measurements are
performed at regular intervals and recorded and analysed for a set period of time,
during which it is assumed that the sea state is stationary. Often measurements
are done for 20 minutes every three hours, but as computer capacity increases it
may be done more often, although the period of analysis should not contain large

78
 5.5. Long Term Wave Statistics

changes in significant wave height or mean period invalidating the stationary as-
sumption.

5.5.2 Probability distributions of significant wave heights

The statistical distribution of the significant wave height can often be approxi-
mated by a two-parameter Weibull distribution
γ
P(Hs > x) = e−( Hc ) ,
x
(5.31)
where Hc and γ are parameters that must be derived from measured significant
wave heights, Hs .
As an example in Table 5.8 a joint frequency table of significant wave height
and zero-crossing period or “scatter diagram” for the North Atlantic is given. The
table is taken from the publication by Hogben et al. [19], the most comprehen-
sive publication of wave data covering 104 ocean areas , denoted Marsden areas.
Jensen [22] discusses the limitation and use of this data.
In order to fit the Weibull distribution, Eq. (5.31), to the measurements of the
North Atlantic in Table 5.8 one can plot the cumulated sums from the table in
a Weibull paper, which is constructed such that a perfect distribution plots as a
straight line1 , and then fit a line by linear regression. The wave height classes are
divided in ∆Hs = 1 m intervals. The sample probability of Hs being smaller than
upper class limits is plotted as
∑ij=1 n j
Pi = P(Hs < i∆Hs ) = , or P(Hs > i∆Hs ) = 1 − Pi , (5.32)
N +1
where n j is the number of observations Class j out of the total number N. N is
here arbitrarily set to 1000 in conformity with Table 5.9.
In Fig. 5.3 the sample probabilities, 1−Pi , Eq. (5.32) and the fitted distribution
Eq. (5.31) with γ = 1.74 and Hc = 3.94 m are plotted.
Now the fitted distribution can be used directly to assess the probability of sig-
nificant wave heights. If the probability is P(Hs > x) that an observed significant
wave height is larger than or equal to x then the inverse is called the return period:
1
Rp = ,
P(Hs > x)
which should be interpreted as the mean number of observations needed to ob-
serve one value larger or equal to x. If we assume that the observations are equidis-
tant in time, τ, then we can express the return period as a time span needed for
such an observation as
τ
R= . (5.33)
P(Hs > x)
1 One can write Eq. (5.31) as ln(P(Hs )) = −(Hs /Hc )γ , then ln(− ln(P(Hs ))) = γ(ln(Hs ) −
ln(Hc )) which is a straight line of the form y = kx + l in Fig. 5.3 with slope k = γ and gives
x = ln(Hc ) for y = ln(− ln(P(Hs ))) = 0.

79
 Table 5.8: Scatter diagram from the North Atlantic, Area 16, All directions: adapted from Hogben et al. [19]. The sum column and row
column denoted Hogben are taken directly from Hogbens table, while the other sum column and row column are results of
summing the figures in the table.
Hs (m) Tz (s)
<4 4–5 5 –6 6–7 7–8 8–9 9–10 10–11 11-12 12–13 > 13 Hogben Row sum Cumulated sum
<1 0 2 13 22 14 5 1 0 0 0 0 57 57 57
1–2 0 0 11 53 78 51 18 4 1 0 0 218 216 273
2–3 0 0 4 31 77 80 44 15 4 1 0 255 256 529
3–4 0 0 1 13 45 64 47 21 6 2 0 197 199 728
4–5 0 0 0 4 21 37 34 18 7 2 0 124 123 851
5–6 0 0 0 1 9 19 20 13 6 2 0 70 70 921
5. WAVE STATISTICS AND PROBABILITY

6–7 0 0 0 1 3 9 11 8 4 1 0 37 37 958
7–8 0 0 0 0 1 4 6 5 3 1 0 20 20 978
8–9 0 0 0 0 1 2 3 3 2 1 0 11 12 990
9–10 0 0 0 0 0 1 2 1 1 0 0 6 5 955
10–11 0 0 0 0 0 0 1 1 1 0 0 3 3 998
11–12 0 0 0 0 0 0 0 1 0 0 0 2 1 999
12–13 0 0 0 0 0 0 0 1∗ 0 0 0 1 1 1000
13–14 0 0 0 0 0 0 0 0 0 0 0 1 0 1000
> 14 0 0 0 0 0 0 0 0 0 0 0 1 0 1000
Total 1003 1000
Hogben 0 3 28 124 249 271 186 90 34 10 3 998
Column sum 0 2 29 125 249 272 187 91 35 10 1000

80
∗ 1 ‰ observations are added for 12–13 m at 10–11 s to make the table sum 1000 ‰.
 5.5. Long Term Wave Statistics
Wave-Induced Loads and Ship Motions
02 December 2008
Page 149

4
( )
ln Hc

ln ( − ln ( 1 −P i) )
0
( ( ( )))
ln − ln P Hs
i

4
0 0.5 1 1.5 2 2.5 3
( )
ln Hs
i
Figure 8.1 Sample probability distribution Pi and fitted Weibull distribution.
Figure 5.3: Sample probability distribution Pi and fitted Weibull distribution.
γ
Table 8.8 Sample probabilities of P( H s < x ) = 1 − P( H s > x ) = 1 − e − ( x / H ) c

constructed from Table 8.7 and Equation 8.8.

Wave height Class No i Permille of Permille of Probability of Hs
classes observations observations being smaller
Hs(m)\ in each class below upper than upper x γ
class
P(Hs < x) = 1limit
Table 5.9: Sample probabilities of(‰) − P(Hofseach 1 − e−( Hc )
> x) = limit
class (‰)
constructed from Table 5.8 and Eq. (5.32). Pi = P ( H s < i∆H s )

<1
Significant Class No i 1Permille of 57 Permille of 57 Probability 0.057
1-2 height
wave 2observations 216observations of Hs being 0.273
273
2-3 (m)
classes 3in each class 256
below upper 529smaller than up- 0.529
3-4 4 199 728 0.728
limit of each per class limit
4-5 5 123 851 0.851
6
class
70
P(H < i∆Hs ) 0.921
921 s
5-6
<6-7
1 1 757 57
37 0.057
958 0.958
7-8
1–2 2 8216 20
273 978
0.273 0.978
8-9 9 12 990 0.990
2–3
9-10 3 10256 529
5 0.529
995 0.995
10-11
3–4 4 11199 3
728 998
0.728 0.998
11-12 12 1 999 0.999
4–5 5 123 851 0.851
12-13 13 1 1000 1
5–6
Column sum 6 70 921
1000 0.921
6–7 7 37 958 0.958
Now the fitted distribution can be used directly to assess the probability of significant
7–8
wave heights. If8the probability is20P ( H > x ) that978 0.978 wave height
an observed significant
s
8–9
is larger than or9equal to x then the
12 inverse is called
990the return period
0.990
9–10 10 5 995 0.995
1
10–11 R p = 11 , 3 998 0.998
P( H s > x )
11–12 12 1 999 0.999
12–13 13 1 1000 1

81
5. WAVE STATISTICS AND PROBABILITY

Table 5.10: Significant wave heights and maximum wave heights in the North
Atlantic for various return periods.

R(years) HsR R
Hmax
10 16.84 31.84
20 17.37 32.84
50 18.06 34.14
100 18.57 35.10
500 19.71 37.25
1000 20.18 38.25

Wave observations are usually based on 20 minutes time series of measured

waves. The observations are often not performed every 20 minutes but rather
every 3 hrs or 6 hrs, but are assumed to be representative also for all periods
between the observations, and then we can set τ = 20 min. By solving Hs from
Eq. (5.31) we can now calculate e.g. the 10-year, 20-year, 50-year, 100-year,
500-year or 1000-year significant wave height. The statistics given here give the
results in Table 5.10.
If we further assume that the expected “maximum” individual wave height for
various return periods are given by sea states with the same return period and if
these are set as the three-hour extreme wave height Equation (5.26)
r
p 1
Hmax = 8m0 ln(N) = Hs ln(N) .
2
In Table 5.8 one can see that the most common mean period is Tz = 8.5 s then
20 yrs
three hours contain N = 1 271 waves and thus Hmax = 1.89Hs and e.g. Hmax =
100 yrs
32.8 m and Hmax = 35.1 m. (Note that the weighted average of all Tz in Table
5.8 happen to be 8.42 s. To be used below)

5.5.3 Probability distributions of individual wave heights

The method to estimate the maximum individual wave height used in the previous
paragraph is used by e.g. Norwegian authorities in the context of petroleum ex-
ploration. Traditionally however one used to estimate it by finding the distribution
for all individual wave heights. E.g. a numerical distribution of individual waves
can be formed by adding the Rayleigh distributions for all observations in Table
5.9.  2
M x
−2 Hs j
P(H < x) = 1 − ∑ e n j /N , (5.34)
j=1

where P(H < x) is the long term sample probability that a wave height does not
exceed x. See Jensen (2001) or Faltinsen (1990). The method described below

82
 5.5. Long Term Wave Statistics

was proposed by Nordenstrm (1973).

Under the assumption that the individual wave heights in a sea state obey
the Rayleigh distribution Equation (5.21) the long term distribution of individual
wave heights can be written as a Weibull distribution given by
 D
x
− 1/d
PL (H > x) = e CHc , (5.35)

where Hc is one of the parameters of the Weibull distribution of the significant

wave heights and the coefficients C and D are functions of the other parameter
the slope, γ. The coefficient, d, depends on whether measured Hs data or visual
wave heights Hv are used. For visual data d = 4/3 and for significant wave heights
4/3
d = 1. Compare Equation (4.6), which gives Hv = 0.501Hs .
For the given data for the North Atlantic γ = 1.74 and Hc = 3.94 m were
found above and d = 1 because significant wave data is used. Then C = 0.476 and
D = 1.009 are found by interpolation in Table 5.11. Further the return period of
individual waves exceeding x is written
τ
R= , (5.36)
PL (H > x)

where τ is the average wave period during the return period, R, or equivalently
the number of waves are R/τ. The mean wave period derived from Table 5.8 is
Tz = 8.42 s which can be set to τ. Then the following maximum expected indi-
vidual wave height for some return periods will be given as in Table 5.12. These
maximum wave heights are a little higher than those derived from the significant
R-year return wave heights as would be expected.
Jensen (2001) derives the 20 year maximum wave to 36 m from the same data
as here. He uses a one sided normal distribution for the distribution of significant
wave heights which then gives a Gumbel distribution of individual wave heights.

5.5.4 Joint probability of wave heights, winds and currents

In Norwegian guidelines for the design of offshore platforms the extreme weather
condition to be considered is the most unfavourable combination of

• wind speed and sea state corresponding to a 100-year return period com-
bined with a 10 year-return-period current;

• current velocity and sea state with 100-year return period combined with a
wind speed with a 10-year-return period.

If these conditions are supposed to be independent the combined return period

would be 10−5 per year, which obviously is incorrect as the wind, waves and
maybe also the current are not independent. The target probability is stated as
10−4 per year.

83
5. WAVE STATISTICS AND PROBABILITY

Table 5.11: The parameters C and D of the long term distribution of individual
wave heights Equation (5.29).

γd C D
d = 4/3 d=1
∞ 1.189 0.707 2.000
10.00 1.056 0.628 1.780
8.00 1.029 0.612 1.712
6.00 0.992 0.590 1.614
4.00 0.930 0.553 1.444

3.33 0.901 0.536 1.354

2.86 0.876 0.521 1.276
2.50 0.855 0.508 1.208
2.22 0.837 0.497 1.144
2.00 0.820 0.488 1.086

1.82 0.807 0.480 1.034

1.67 0.794 0.472 0.988
1.54 0.783 0.465 0.944
1.43 0.772 0.459 0.904
1.33 0.762 0.453 0.868

1.25 0.754 0.448 0.834

1.18 0.746 0.444 0.802
1.11 0.739 0.439 0.744
1.05 0.732 0.435 0.746
1.00 0.726 0.432 0.722

0.67 0.689 0.410 0.538

0.50 0.666 0.396 0.428
0.40 0.656 0.390 0.356

84
 5.5. Long Term Wave Statistics

Table 5.12: Maximum wave heights in the North Atlantic for various return pe-
riods.

R(years) R
Hmax
10 31.89
20 33.14
50 34.80
100 36.06
500 38.97
1000 40.23

A recent treatise of the joint probability of wind and waves in the Northern
North Sea is given by Johannessen et al. (2002). In this the marginal distribu-
tion of the wind speed is taken as a two-parameter Weibull distribution. In each
wind-speed class the significant wave heights are found to follow a two-parameter
Weibull distribuition and the peak periods a log-normal distribution.

5.5.5 Design wave for a ship’s hull girder

For the design of a ship’s hull girder the most critical wave event may be taken as
a head or bow, regular, plane wave having a wave length, λ , equal to the length,
L pp , of the ship. A ship in such wave will alternatively have the wave crests at
the bow and stern and the wave trough amidships (sag), or troughs at the bow and
stern and a wave crest amidships (hog). The wave should have a reasonable wave
amplitude, a, and steepness, 2a/λ as is described in the following.
The most probable maximum wave amplitude during the period τ = 3 h is
given by s  
Hs τ
a= 2 ln ≈ Hs , (5.37)
4 Tz
where the last approximation is somewhat conservative. The maximum possible
wave steepness is 0.142 giving the maximum wave amplitude as
L pp
a< = 0.071L pp . (5.38)
14
According to Table 5.10 the 20 year significant wave height can be set to 17.37
m in the North Atlantic. Then for a 200 m long ship the ship-girder design wave
will have the amplitude a = Hs = 17.37 m or a = 200/14 m = 14.2 m, whichever
is smallest. So in this case a =
p14.2 m, i.e. the wave height 28.6 m. The corre-
sponding wave period is T = 2πL pp /g = 11.3 s, in deep water without current
that is. Compare this with the information given in the scatter diagram Table 5.8.
For a more elaborate discussion, see Jensen [22] or Tromans et al (1991) for an
alternative form of the design wave.

85
 6
EQUATIONS OF MOTION FOR A SHIP

6.1 COORDINATE SYSTEMS AND DEFINITIONS

Three coordinate systems are normally used when the equations of motion for a
ship are defined. Unfortunately there is no universally accepted coordinate sys-
tem convention. For the seakeeping problem most authors and organizations use
x in the forward direction, y to port and z upwards, while in the manoeuvring
problem x is forward, y is to starboard and z is downwards. There are however
exceptions from this ”rule”. These Lecture Notes deal with both the seakeeping
problem and the manoeuvring problem and it is not obvious how to define the
coordinate systems. It was decided to follow the ”rule” described above and use
separate coordinate system definitions for the two problems. This is probably a
combination the students will meet if they deal with seakeeping and manoeuvring
in their professional life as Naval Architects. The three coordinate systems shown
in Fig. 6.1 follow the convention for the manouevring problem and they are here
used for the derivation of the equations of motion for a ship.
The three coordinate systems are:

1. an earth fixed coordinate system (x0 , y0 , z0 ) with the origin at any desirable
location. The z0 axis is positive downwards and the x0 , y0 plane is normally
located at the calm water level. The x0 axis is pointing in the forward di-
rection of the ship. This coordinate system is used as the reference system
when deriving the equations of motion for a ship. It is also used to define
the incident wave system.

2. a coordinate system that moves with a constant speed U in the positive x0

direction. This coordinate system is normally referred to as the inertial
coordinate system. The x, y plane is also aligned with the calm water level
and z is positive downwards.The x-axis is coincident with the x0-axis.

87
6. EQUATIONS OF MOTION FOR A SHIP

Figure 6.1: Definition of the earth-fixed, the inertial and the body-fixed coordi-
nate systems

3. a third coordinate system(x̄, ȳ, z̄) is body fixed and the motions of the ship is
determined by the orientation of the body fixed coordinate system relative
to the inertial coordinate system.

The transformation from the earth fixed to the inertial coordinate system is
given by:

x0 = x +U · t (6.1)
y0 = y (6.2)
z0 = z (6.3)
Three translations
• η1 - surge is a translation in the x̄-direction, positive forward

• η2 - sway is a translation in the ȳ-direction, positive to starboard

• η3 - heave is a translation in the z̄-direction, positive downwards

and three rotations

• η4 - roll is a rotation about the x̄-axis, positive when starboard side goes in
to the water

• η5 - pitch is a rotation about the ȳ-axis, positive when the bow goes out of
the water

88
 6.1. COORDINATE SYSTEMS AND DEFINITIONS

Figure 6.2: Definition of η4

• η6 - yaw is a rotation about the z̄-axis, positive when the bow turns to star-
board

are used to define the motions of the ship, see Figs. 6.2 , 6.3 and 6.4. It must
be noted that the rotations about the x̄, ȳ and z̄-axes are not commutative. The final

Figure 6.3: Definition of η5

89
6. EQUATIONS OF MOTION FOR A SHIP

Figure 6.4: Definition of η6

position depends on the order in which the rotations are carried out. However, if
the rotations are infinitesimal, it can be shown that they do satisfy the commutative
law of addition. Thus the angular velocity vector can be expressed as the time rate
of change of infinitesimal rotations.
A convention is therefore used for the order of the rotations in order to obtain
the angular position of the ship with respect to the earth fixed coordinate system.
The convention is as follows:

1. Yaw about the z̄-axis

2. Pitch about the ȳ-axis

3. Roll about the z̄-axis

The translation velocities are denoted η̇1 in the x̄-direction, η̇2 in the ȳ-direction
and η̇3 in the z̄-direction. The angular velocities are denoted η̇4 about the x̄-axis,
η̇5 about the ȳ-axis and η̇6 about the z̄-axis.

6.2 SHIP MOTION IN AN EARTH FIXED

COORDINATE SYSTEM

In an earth fixed coordinate system the equations of motion for the center of grav-
ity (xG0 , yG0 , zG0 ) of a rigid body are

mη̈10 = F10 (6.4)

90
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

Figure 6.5: Equations of motion in the earth-fixed coordinate system

mη̈20 = F20 (6.5)

mη̈30 = F30 (6.6)
and the rotations about axes of the earth fixed coordinate system are described
by the equations:

Ix0 η̈40 = F40 (6.7)

Iy0 η̈50 = F50 (6.8)
Iz0 η̈60 = F60 (6.9)
The latter equations include the moments and the moments of inertia about the
axes of the earth fixed coordinate system, see Fig. 6.5.

6.3 SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

The ship motion relative to the earth fixed coordinate system is normally large
which means that the forces, moments, moments of inertia and the products of
inertia vary with time. They are therefore difficult to evaluate. A body fixed
coordinate system where the moments of inertia and the products of inertia are
constant is therefore introduced to avoid these difficulties.

The price we have to pay is that the equations of motions must also be trans-
ferred to the body fixed coordinate system which make them more complicated.
After solving the equations of motion in the body fixed coordinate system we

91
6. EQUATIONS OF MOTION FOR A SHIP

Figure 6.6: Equations of motion in the earth-fixed coordinate system

get information about the motion of the origin of this system and about the rota-
tion about the axes of the body fixed coordinate system. The origin of the body
fixed coordinate system is normally located in the center plane at Lpp/2 at the
undisturbed free surface level. The position of the center of gravity is denoted
(xG , yG , zG ).

6.3.1 Translation
First assume that the ship performs translations in surge, sway and heave only
while the rotations are all zero. A small mass element dm at (x̄, ȳ, z̄) will then
have the same velocities η̇1 , η̇2 , η̇3 in the x̄, ȳ, z̄ directions as the origin of the body
fixed coordinate system, see Fig. 6.6. The angle between the axes of the (x̄, ȳ, z̄)
and the (x, y, z) systems do not change with time.

6.3.2 Velocity contribution from yaw

Assume that an angular velocity η̇6 is added to the translations, see Fig. 6.7. The
origin of the body fixed coordinate system is moving with the velocity η̇1 , η̇2 , η̇3
but a velocity component

• rη̇6

is added to the velocity of the small mass element dm. The added velocity is
perpendicular to the position vector of the mass element dm. The components of

92
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

Figure 6.7: Influence Of Yaw

the velocity due to the angular velocity are then added to the translation velocities
η̇1 , η̇2 , η̇3 in the body fixed coordinate system

• η̇1 − ȳ · η̇6 in the x̄-direction

• η̇2 + x̄ · η̇6 in the ȳ-direction

• η̇3 in the z̄-direction (not changed)

6.3.3 Velocity contribution from pitch

Now assume that an angular velocity η̇5 is added to the translations and yaw, see
Fig. 6.8.

• rη̇5

The added velocity is perpendicular to the position vector of the mass element
dm. The components of the velocity due to the angular velocity η̇5 are then added
to the translation velocities and the velocity contributions from yaw in the body
fixed coordinate system.

• η̇1 − ȳ · η̇6 + z̄ · η̇5 in the x̄-direction

• η̇2 + x̄ · η̇6 in the ȳ-direction

• η̇3 − x̄ · η̇5 in the z̄-direction

93
6. EQUATIONS OF MOTION FOR A SHIP

Figure 6.8: Influence Of Pitch

Figure 6.9: Influence Of Roll

94
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

6.3.4 Velocity contribution from roll

Now assume that an angular velocity η̇4 is added to translation, yaw and pitch see
Fig. 6.9.

• rη̇4

The added velocity is perpendicular to the position vector of the mass element
dm. The components of the velocity due to the angular velocity η̇4 are then added
to the translation velocities and to the velocity contributions from yaw and pitch in
the body fixed coordinate system. We can now write the total velocity components
for motion in six degrees of freedom in the x̄, ȳ and z̄ directions respectively:

η̇1T OT = η̇1 − ȳ · η̇6 + z̄ · η̇5 (6.10)

η̇2T OT = η̇2 + x̄ · η̇6 − z̄ · η̇4 (6.11)

η̇3T OT = η̇3 − x̄ · η̇5 + ȳ · η̇4 (6.12)

6.3.5 Components of acceleration due to yaw

In order to formulate the equations of motion we need the components of the ac-
celerations. First investigate the time derivative of the velocity components in the
x, y-plane, see Fig. 6.10.

At the time t we have the velocity components η̇1T OT and η̇2T OT . At t +

dt η̇1T OT has increased to η̇1T OT + d η̇1T OT and η̇2T OT has increased to η̇2T OT
+d η̇2T OT while the body fixed coordinate system has turned dη6 with respect
to the inertial coordinate system during dt. It is assumed that dη6 is small,
cos dη6 =1, sin dη6 =dη6 , d η̇1T OT is small compared to η̇1T OT and that d η̇2T OT
is small compared to η̇2T OT .

The increment of the velocity components in the x̄-direction is:

(η̇1T OT +d η̇1T OT ) cos dη6 − η̇1T OT −(η̇2T OT +d η̇2T OT ) sin dη6 = d η̇1T OT − η̇2T OT dη6
(6.13)

The x̄-component of the acceleration yields

d η̇1T OT − η̇2T OT dη6
lim = η̈1T OT − η̇2T OT η̇6 (6.14)
dt→0 dt

95
6. EQUATIONS OF MOTION FOR A SHIP

Figure 6.10: Acceleration in the x-y plane

Figure 6.11: Acceleration in the x-z plane

96
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

Figure 6.12: Acceleration in the y-z plane

In the ȳ-direction the increment is

(η̇2T OT +d η̇2T OT ) cos dη6 − η̇2T OT +(η̇1T OT +d η̇1T OT ) sin dη6 = d η̇2T OT + η̇1T OT dη6
(6.15)

and the ȳ component of the acceleration is

d η̇2T OT + η̇1T OT dη6

lim = η̈2T OT + η̇1T OT η̇6 (6.16)
dt→0 dt

6.3.6 Components of acceleration due to pitch

In the xz-plane the velocity components and their increment during a time step dt
are shown in Fig. 6.11.The ship and the body fixed coordinate system has turned
the pitch angle dη5 during dt.

At the time t we have the velocity components η̇1T OT and η̇3T OT . At t +

dt η̇1T OT has increased to η̇1T OT +d η̇1T OT and η̇3T OT has increased to η̇3T OT
+d η̇3T OT while the body fixed coordinate system has turned dη5 with respect
to the inertial coordinate system during dt. It is assumed that dη5 is small,
cos dη5 =1, sin dη5 =dη5 , d η̇1T OT is small compared to η̇1T OT and that d η̇3T OT
is small compared to η̇3T OT .

97
6. EQUATIONS OF MOTION FOR A SHIP

The increment of the velocity components in the x̄-direction is:

(η̇1T OT +d η̇1T OT ) cos dη5 − η̇1T OT +(η̇3T OT +d η̇3T OT ) sin dη6 = d η̇1T OT + η̇3T OT dη5
(6.17)

The x̄-component of the acceleration yields

d η̇1T OT + η̇3T OT dη5
lim = η̈1T OT + η̇3T OT η̇5 (6.18)
dt→0 dt

In the w̄-direction the increment is

(η̇3T OT +d η̇3T OT ) cos dη5 − η̇3T OT −(η̇1T OT +d η̇1T OT ) sin dη5 = d η̇3T OT − η̇1T OT dη5
(6.19)

and the w̄ component of the acceleration is

d η̇3T OT − η̇1T OT dη5
lim = η̈3T OT + η̇1T OT η̇5 (6.20)
dt→0 dt

6.3.7 Components of acceleration due to roll

In the yz-plane the velocity components and their increment during a time step dt
are shown in Fig. 6.12.The ship and the body fixed coordinate system has turned
the pitch angle dη4 during dt.

At the time t we have the velocity components η̇2T OT and η̇3T OT . At t +

dt η̇2T OT has increased to η̇2T OT +d η̇2T OT and η̇3T OT has increased to η̇3T OT
+d η̇3T OT while the body fixed coordinate system has turned dη4 with respect
to the inertial coordinate system during dt. It is assumed that dη4 is small,
cos dη4 =1, sin dη4 =dη4 , d η̇2T OT is small compared to η̇2T OT and that d η̇3T OT
is small compared to η̇3T OT .

The increment of the velocity components in the ȳ-direction is:

(η̇2T OT +d η̇2T OT ) cos dη4 − η̇2T OT −(η̇3T OT +d η̇3T OT ) sin dη6 = d η̇2T OT − η̇3T OT dη4
(6.21)

The ȳ-component of the acceleration yields

98
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

d η̇2T OT − η̇3T OT dη4

lim = η̈2T OT − η̇3T OT η̇4 (6.22)
dt→0 dt

In the w̄-direction the increment is

(η̇3T OT +d η̇3T OT ) cos dη4 − η̇3T OT +(η̇2T OT +d η̇2T OT ) sin dη4 = d η̇3T OT + η̇2T OT dη4
(6.23)

and the w̄ component of the acceleration is

d η̇3T OT + η̇2T OT dη4
lim = η̈3T OT + η̇2T OT η̇4 (6.24)
dt→0 dt

6.3.8 Accelerations in the x̄, ȳ and z̄-directions

The contributions to the accelerations due to yaw, pitch and roll can now be writ-
ten. Remember that the contributions η̇1 T OT , η̇2 T OT and η̇3 T OT must only be
counted once.

η̈1 = η̈1T OT − η̇2T OT η̇6 + η̇3T OT η̇5 =

(η̈1 − ȳη̈6 + z̄η̈5 ) − η̇6 (η̇2 + x̄η̇6 − z̄η̇4 ) + η̇5 (η̇3 − x̄η̇5 + ȳη̇4 ) (6.25)

η̈2 = η̈2T OT + η̇1T OT η̇6 − η̇3T OT η̇4 =

(η̈2 + x̄η̈6 − z̄η̈4 ) + η̇6 (η̇1 − ȳη̇6 + z̄η̇5 ) − η̇4 (η̇3 − x̄η̇5 + ȳη̇4 ) (6.26)

η̈3 = η̈3T OT − η̇1T OT η̇5 + η̇2T OT η̇4 =

(η̈3 − x̄η̈5 + ȳη̈4 ) − η̇5 (η̇1 − ȳη̇6 + z̄η̇5 ) + η̇4 (η̇2 + x̄η̇6 − z̄η̇4 ) (6.27)

6.3.9 Equations of motion: Forces in the x̄, ȳ and z̄-directions

The force in a direction is equal to mass times acceleration in the same direction
according to Newtons second law. Applying this law in the directions of the body
fixed coordinate system yields

dF1 = dmη̈1 (6.28)

99
6. EQUATIONS OF MOTION FOR A SHIP

dF2 = dmη̈2 (6.29)

dF3 = dmη̈3 (6.30)
Integration over the length of the ship give:

Z
F1 = η̈1 dm =
Z Z LZ

η̈1 dm − η̈6 ȳdm + η̈5 z̄dm −

Z L Z L ZL
η̇6 η̇2 dm − η̇62 x̄dm + η̇6 η̇4 z̄dm +
LZ LZ LZ

η̇4 η̇3 dm − η̇52 x̄dm + η̇5 η̇4 ȳdm (6.31)

L L L

Z
F2 = η̈2 dm =
Z Z LZ

η̈2 dm + η̈6 x̄dm − η̈4 z̄dm +

Z L Z L ZL
η̇6 η̇1 dm − η̇62 ȳdm + η̇5 η̇4 z̄dm −
LZ L Z LZ

η̇4 η̇3 dm + η̇4 η̇5 x̄dm − η̇42 ȳdm (6.32)

L L L

Z
F3 = η̈3 dm =
Z Z LZ

η̈3 dm − η̈5 x̄dm + η̈4 ȳdm −

Z L ZL ZL
η̇5 η̇1 dm + η̇5 η̇6 ȳdm − η̇52 z̄dm +
LZ LZ LZ

η̇4 η̇2 dm + η̇4 η̇6 x̄dm − η̇42 z̄dm (6.33)

L L L

Introducing the mass and the center of gravity

Z
dm = m (6.34)
L
Z
x̄dm = mx̄G (6.35)
L
Z
ȳdm = mȳG (6.36)
L

100
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

Z
z̄dm = mz̄G (6.37)
L

to the equations ( ), ( ) and ( ) yields:

F1 = m[(η̈1 − η̇6 η̇2 + η̇5 η̇3 ) −

2 2
x̄G (η̇6 + η̇5 ) − ȳG (η̈6 − η̇5 η̇4 ) + z̄G (η̈5 + η̇6 η̇4 )] (6.38)

F2 = m[(η̈2 + η̇6 η̇1 − η̇4 η̇3 ) +

x̄G (η̈6 + η̇4 η̇5 ) − ȳG (η̇62 + η̇42 ) − z̄G (η̈4 − η̇6 η̇5 )] (6.39)

F3 = m[(η̈3 − η̇5 η̇1 + η̇4 η̇2 ) −

x̄G (η̈5 − η̇4 η̇6 ) + ȳG (η̈4 + η̇5 η̇6 ) − z̄G (η̇52 + η̇42 )] (6.40)

6.3.10 Equations of motion: Moments about the x̄, ȳ and z̄-directions

The contributions to the moments from a small mass element dm about the z̄,
ȳ and x̄-axes are shown in Fig. 6.13, Fig. 6.14 and Fig. 6.15 respectively. The
contributions are

dF4 = dm(ȳη̈3 − z̄η̈2 ) (6.41)

dF5 = dm(z̄η̈1 − x̄η̈3 ) (6.42)

dF6 = dm(x̄η̈2 − ȳη̈1 ) (6.43)

Integration over the length of the ship yields:

101
6. EQUATIONS OF MOTION FOR A SHIP

Figure 6.13: Contribution to moment in the x-y plane

Figure 6.14: Contribution to moment in the x-z plane

102
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

Figure 6.15: Contribution to moment in the y-z plane

Z
F4 = (ȳη̈3 − z̄η̈2 )dm =
Z Z L Z
η̈3 ȳdm − η̈5 x̄ȳdm + η̈4 ȳ2 dm −
Z L ZL ZL
2
η̇5 η̇1 ȳdm + η̇5 η̇6 ȳ dm − η̇52 ȳz̄dm +
ZL ZL ZL
η̇4 η̇2 ȳdm + η̇4 η̇6 x̄ȳdm − η̇42 ȳz̄dm −
L Z ZL ZL
η̈2 z̄dm − η̈6 x̄z̄dm + η̈4 z̄2 dm −
Z L Z L ZL
η̇6 η̇1 z̄dm + η̇62 ȳz̄dm − η̇6 η̇5 z̄2 dm +
LZ L Z LZ

η̇4 η̇3 z̄dm − η̇4 η̇5 x̄z̄dm + η̇42 ȳz̄dm (6.44)

L L L

103
6. EQUATIONS OF MOTION FOR A SHIP

Z
F5 = (z̄η̈1 − x̄η̈3 )dm =
Z Z L Z
η̈1 z̄dm − η̈6 ȳz̄dm + η̈5 z̄2 dm −
Z L Z L ZL
η̇6 η̇2 z̄dm − η̇62 x̄z̄dm + η̇6 η̇4 z̄2 dm +
ZL ZL ZL
η̇5 η̇3 z̄dm − η̇52 x̄z̄dm + η̇5 η̇4 ȳz̄dm −
LZ LZ ZL
η̈3 x̄dm + η̈5 x̄2 dm − η̈4 x̄ȳdm +
Z L ZL ZL
η̇5 η̇1 x̄dm − η̇5 η̇6 x̄ȳdm + η̇52 x̄z̄dm −
LZ LZ LZ

η̇4 η̇2 x̄dm − η̇4 η̇6 x̄2 dm + η̇42 x̄z̄dm (6.45)

L L L

Z
F6 = (x̄η̈2 − ȳη̈1 )dm =
Z Z L Z
2
η̈2 x̄dm + η̈6 x̄ dm − η̈4 x̄z̄dm +
Z L Z L ZL
η̇6 η̇1 x̄dm − η̇62 x̄ȳdm + η̇6 η̇4 x̄z̄dm −
ZL LZ ZL
η̇4 η̇3 x̄dm + η̇5 η̇4 x̄2 dm − η̇42 x̄ȳdm −
LZ ZL ZL
η̈1 ȳdm + η̈6 ȳ2 dm − η̈5 ȳz̄dm +
Z L Z L ZL
η̇6 η̇2 ȳdm + η̇62 x̄ȳdm − η̇6 η̇4 ȳz̄dm −
LZ LZ LZ

η̇5 η̇3 ȳdm + η̇52 x̄ȳdm − η̇5 η̇4 ȳ2 dm (6.46)

L L L

Introduce the moments of inertia and the products of inertia

Z
(z̄2 + ȳ2 )dm = Ix̄ (6.47)
L
Z
x̄ȳdm = Ix̄ȳ (6.48)
L
Z
ȳz̄dm = Iȳz̄ (6.49)
L
Z
x̄z̄dm = Ix̄z̄ (6.50)
L

104
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

Z Z Z Z Z Z
ȳ2 dm − z̄2 dm = ȳ2 dm + x̄2 dm − z̄2 dm − x̄2 dm =
L L L L L L
(6.51)
Z Z
(x̄2 + ȳ2 )dm − (x̄2 + z̄2 )dm = Iz̄ − Iȳ (6.52)
L L

F4 = Ix̄ η̈4 + (Iz̄ − Iȳ )η̇6 η̇5 − Ix̄ȳ (η̈5 − η̇6 η̇4 ) − Iȳz̄ (η̇52 − η̇62 ) − Ix̄z̄ (η̈6 + η̇4 η̇5 ) +
(6.53)
m[ȳG (η̈3 − η̇5 η̇1 + η̇4 η̇2 ) − z̄G (η̈2 + η̇6 η̇1 − η̇4 η̇3 )]
(6.54)

F5 = Iȳ η̈5 + (Ix̄ − Iz̄ )η̇6 η̇4 − Iȳz̄ (η̈6 − η̇5 η̇4 ) − Ix̄z̄ (η̇62 − η̇42 ) − Ix̄ȳ (η̈4 + η̇5 η̇6 ) +
(6.55)
m[z̄G (η̈1 − η̇6 η̇2 + η̇5 η̇3 ) − x̄G (η̈3 − η̇5 η̇1 + η̇4 η̇2 )]
(6.56)

F6 = Iz̄ η̈6 + (Iȳ − Ix̄ )η̇5 η̇4 − Ix̄z̄ (η̈4 − η̇6 η̇5 ) − Ix̄ȳ (η̇42 − η̇52 ) − Iȳz̄ (η̈5 + η̇6 η̇4 ) +
(6.57)
m[x̄G (η̈2 + η̇5 η̇1 − η̇4 η̇3 ) − ȳG (η̈1 − η̇6 η̇2 + η̇5 η̇3 )]
(6.58)

6.3.11 Equations of motion for arbitrary position of the coordinate system

The most general formulation of the equations of motion for a ship is:

F1 = m[(η̈1 − η̇6 η̇2 + η̇5 η̇3 ) −

(6.59)
x̄G (η̇62 + η̇52 ) − ȳG (η̈6 − η̇5 η̇4 ) + z̄G (η̈5 + η̇6 η̇4 )] (6.60)

F2 = m[(η̈2 + η̇6 η̇1 − η̇4 η̇3 ) +

(6.61)
x̄G (η̈6 + η̇4 η̇5 ) − ȳG (η̇62 + η̇42 ) − z̄G (η̈4 − η̇6 η̇5 )] (6.62)

105
6. EQUATIONS OF MOTION FOR A SHIP

F3 = m[(η̈3 − η̇5 η̇1 + η̇4 η̇2 ) −

(6.63)
x̄G (η̈5 − η̇4 η̇6 ) + ȳG (η̈4 + η̇5 η̇6 ) − z̄G (η̇52 + η̇42 )] (6.64)

F4 = Ix̄ η̈4 + (Iz̄ − Iȳ )η̇6 η̇5 − Ix̄ȳ (η̈5 − η̇6 η̇4 ) − Iȳz̄ (η̇52 − η̇62 ) − Ix̄z̄ (η̈6 + η̇4 η̇5 ) +
(6.65)
m[ȳG (η̈3 − η̇5 η̇1 + η̇4 η̇2 ) − z̄G (η̈2 + η̇6 η̇1 − η̇4 η̇3 )]
(6.66)

F5 = Iȳ η̈5 + (Ix̄ − Iz̄ )η̇6 η̇4 − Iȳz̄ (η̈6 − η̇5 η̇4 ) − Ix̄z̄ (η̇62 − η̇42 ) − Ix̄ȳ (η̈4 + η̇5 η̇6 ) +
(6.67)
m[z̄G (η̈1 − η̇6 η̇2 + η̇5 η̇3 ) − x̄G (η̈3 − η̇5 η̇1 + η̇4 η̇2 )]
(6.68)

F6 = Iz̄ η̈6 + (Iȳ − Ix̄ )η̇5 η̇4 − Ix̄z̄ (η̈4 − η̇6 η̇5 ) − Ix̄ȳ (η̇42 − η̇52 ) − Iȳz̄ (η̈5 + η̇6 η̇4 ) +
(6.69)
m[x̄G (η̈2 + η̇6 η̇1 − η̇4 η̇3 ) − ȳG (η̈1 − η̇6 η̇2 + η̇5 η̇3 )]
(6.70)

6.3.12 Equations of motion when the origin of the coordinate system is at center
of gravity
Assume that the origin of the coordinate system is located at the center of gravity
of the ship. All terms containing x̄G , ȳG and z̄G are then zero.

F1 = m(η̈1 − η̇6 η̇2 + η̇5 η̇3 ) (6.71)

F2 = m(η̈2 + η̇6 η̇1 − η̇4 η̇3 ) (6.72)

F3 = m(η̈3 − η̇5 η̇1 + η̇4 η̇2 ) (6.73)

F4 = Ix̄ η̈4 + (Iz̄ − Iȳ )η̇6 η̇5 − Ix̄ȳ (η̈5 − η̇6 η̇4 ) − Iȳz̄ (η̇52 − η̇62 ) − Ix̄z̄ (η̈6 + η̇4 η̇5 )]
(6.74)

106
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

F5 = Iȳ η̈5 + (Ix̄ − Iz̄ )η̇6 η̇4 − Iȳz̄ (η̈6 − η̇5 η̇4 ) − Ix̄z̄ (η̇62 − η̇42 ) − Ix̄ȳ (η̈4 + η̇5 η̇6 )]
(6.75)

F6 = Iz̄ η̈6 + (Iȳ − Ix̄ )η̇5 η̇4 − Ix̄z̄ (η̈4 − η̇6 η̇5 ) − Ix̄ȳ (η̇42 − η̇52 ) − Iȳz̄ (η̈5 + η̇6 η̇4 )]
(6.76)

6.3.13 Equations of motion for starboard-port symmetry

Assume that the ship has starboard-port symmetry. ȳG , Ix̄ȳ and Iȳz̄ are then zero
and the equations of motion can be written:

F1 = m[(η̈1 − η̇6 η̇2 + η̇5 η̇3 ) − x̄G (η̇62 + η̇52 ) + z̄G (η̈5 + η̇6 η̇4 )] (6.77)

F2 = m[(η̈2 + η̇6 η̇1 − η̇4 η̇3 ) + x̄G (η̈6 + η̇4 η̇5 ) − z̄G (η̈4 − η̇6 η̇5 )] (6.78)

F3 = m[(η̈3 − η̇5 η̇1 + η̇4 η̇2 ) − x̄G (η̈5 − η̇4 η̇6 ) − z̄G (η̇52 + η̇42 )] (6.79)

F4 = Ix̄ η̈4 + (Iz̄ − Iȳ )η̇6 η̇5 − Ix̄z̄ (η̈6 + η̇4 η̇5 ) +
(6.80)
m[−z̄G (η̈2 + η̇6 η̇1 − η̇4 η̇3 )] (6.81)

F5 = Iȳ η̈5 + (Ix̄ − Iz̄ )η̇6 η̇4 − Ix̄z̄ (η̇62 − η̇42 ) +

(6.82)
m[z̄G (η̈1 − η̇6 η̇2 + η̇5 η̇3 ) − x̄G (η̈3 − η̇5 η̇1 + η̇4 η̇2 )] (6.83)

F6 = Iz̄ η̈6 + (Iȳ − Ix̄ )η̇5 η̇4 − Ix̄z̄ (η̈4 − η̇6 η̇5 ) +
(6.84)
m[x̄G (η̈2 + η̇6 η̇1 − η̇4 η̇3 )] (6.85)

107
6. EQUATIONS OF MOTION FOR A SHIP

6.3.14 Equations of motion for starboard-port and fore-aft symmetry

Assume that the ship has starboard-port symmetry. ȳG , Ix̄ȳ , Iȳz̄ and Ix̄z̄ are then zero
and the equations of motion can be written:

F1 = m[(η̈1 − η̇6 η̇2 + η̇5 η̇3 ) − x̄G (η̇62 + η̇52 ) + z̄G (η̈5 + η̇6 η̇4 )] (6.86)

F2 = m[(η̈2 + η̇6 η̇1 − η̇4 η̇3 ) + x̄G (η̈6 + η̇4 η̇5 ) − z̄G (η̈4 − η̇6 η̇5 )] (6.87)

F3 = m[(η̈3 − η̇5 η̇1 + η̇4 η̇2 ) − x̄G (η̈5 − η̇4 η̇6 ) − z̄G (η̇52 + η̇42 )] (6.88)

F4 = Ix̄ η̈4 + (Iz̄ − Iȳ )η̇6 η̇5 +

(6.89)
m[−z̄G (η̈2 + η̇6 η̇1 − η̇4 η̇3 )] (6.90)

F5 = Iȳ η̈5 + (Ix̄ − Iz̄ )η̇6 η̇4 +

(6.91)
m[z̄G (η̈1 − η̇6 η̇2 + η̇5 η̇3 ) − x̄G (η̈3 − η̇5 η̇1 + η̇4 η̇2 )] (6.92)

F6 = Iz̄ η̈6 + (Iȳ − Ix̄ )η̇5 η̇4 +

(6.93)
m[x̄G (η̈2 + η̇6 η̇1 − η̇4 η̇3 )] (6.94)

6.3.15 Equations of motion for starboard-port symmetry and origin at center of

gravity
If the ship has starboard-port symmetry and the origin of the coordinate system is
located at center of gravity the equations are:

F1 = m(η̈1 − η̇6 η̇2 + η̇5 η̇3 ) (6.95)

F2 = m(η̈2 + η̇6 η̇1 − η̇4 η̇3 ) (6.96)

F3 = m(η̈3 − η̇5 η̇1 + η̇4 η̇2 ) (6.97)

108
 6.3. SHIP MOTION IN A BODY FIXED COORDINATE SYSTEM

F4 = Ix̄ η̈4 + (Iz̄ − Iȳ )η̇6 η̇5 − Ix̄z̄ (η̈6 + η̇4 η̇5 ) (6.98)

F5 = Iȳ η̈5 + (Ix̄ − Iz̄ )η̇6 η̇4 − Ix̄z̄ (η̇62 − η̇42 ) (6.99)

F6 = Iz̄ η̈6 + (Iȳ − Ix̄ )η̇5 η̇4 − Ix̄z̄ (η̈4 − η̇6 η̇5 ) (6.100)

6.3.16 Equations of motion for starboard-port, fore-aft symmetry and origin at

center of gravity
If fore-aft symmetry is assumed also Ix̄z̄ is zero and the equations are reduced to:

F1 = m(η̈1 − η̇6 η̇2 + η̇5 η̇3 ) (6.101)

F2 = m(η̈2 + η̇6 η̇1 − η̇4 η̇3 ) (6.102)

F3 = m(η̈3 − η̇5 η̇1 + η̇4 η̇2 ) (6.103)

F4 = Ix̄ η̈4 + (Iz̄ − Iȳ )η̇6 η̇5 (6.104)

F5 = Iȳ η̈5 + (Ix̄ − Iz̄ )η̇6 η̇4 (6.105)

F6 = Iz̄ η̈6 + (Iȳ − Ix̄ )η̇5 η̇4 (6.106)

These equations are often used in the analysis of ship motion, but it is then
important to be aware of the assumptions included and check that the assumptions
are not violated.

109
 7
MOTION RESPONSE TO LOADING

The loads on a platform or ship can be constant in time, transient – i.e. of short
duration – or harmonic. Irregular or random loads from e.g. sea waves can to a
first, linear, approximation be treated as a superposition of harmonic loads. The
responses are fundamentally different for the three types of loads. To clearly
illustrate this we will in this chapter use a simple, one-degree-of-freedom system
as shown in Figure 7.1. The equation of motion for this system can be written

(m + a) ẍ + bẋ + cx = F(t) . (7.1)

For bodies in water the mass inertia is increased by an “added mass”, a, or

hydrodynamic mass. This is a result of the fact that to accelerate the body it
is also necessary to accelerate the water surrounding the body. For submerged
bodies close to the water surface the added mass can be negative, but for deeply
submerged bodies it is always positive.
The content in this chapter is meant as a repetition of fundamental mechanics
of vibration, more thoroughly given in some fundamental textbooks e.g. books
by Craig [6], Roberts and Spanos [41] or Thompson [47]. Some application on
six-degree-of-freedom floating systems is given in the end of the chapter.

Figure 7.1: A mechanical system with one degree of freedom, mass, m, added
mass, a, damping coefficient, b, and spring stiffness, c.

111
7. MOTION RESPONSE TO LOADING

7.1 F REE V IBRATION OF A F LOATING S HIP IN H EAVE

Before the discussion of response to different types of loading we will repeat a

little about the free vibrations of the single-degree-of-freedom (SDOF) system.
The equation of motion for a ship in heave can be written

(m + a) ẍ + bẋ + cx = 0 , (7.2)

which follows directly from Eq. 7.1 setting F(t) = 0.

Assuming a solution of the form

x = Ceκt , (7.3)

we get the characteristic equation

κ 2 + 2ξ ωN κ + ωN2 = 0 , (7.4)
p
where ωN = c/(m + a) is the “natural”
p angular frequency, that is, the undamped
angular frequency and ξ = b/(2 c(m + a)) is the damping factor.
The roots of Eq. (7.4) are
q
κ1,2 = −ξ ωN ± ωN ξ 2 − 1 . (7.5)

These roots are complex, zero or real depending on the value of ξ . The damping
factor can thus be used to distinguish between three cases:

• underdamped (0 ≤ ξ < 1)

• critically damped (ξ = 1)

• overdamped (ξ > 1)

See Fig. 7.2 for the motion of a body released from the position x(0) = 1 m at
t = 0 s. The underdamped case displays an attenuating vertical oscillation, while
the other cases display motions monotonously approaching the vertical equilib-
rium position. A floating ship in heave, roll and pitch would normally display
underdamped characteristics with a damping factor of the order of 10−1 in heave
and pitch and of the order of 10−2 in roll. In the horizontal degrees of freedom
there are no stiffnesses so the horizontal motions do not exhibit resonant charac-
teristics.
The damping factor is often called the damping ratio, as it is equal to the ratio
between
p the current damping coefficient b and the critical damping coefficient
2 c(m + a).

112
 7.2. Response to Constant Loads

Figure 7.2: Response of a damped SDOF system with various damping ra-
tios. Dashed line: overdamped case (ξ = 1.5), solid line: critically
damped case (ξ = 1) and dotted line: underdamped case (ξ = 0.1).

7.2 R ESPONSE TO C ONSTANT L OADS

A constant or static load Fo (see Fig. 7.3) acting on the system in Fig. 7.1 gives
as response a displacement to a static equilibrium position, x = xo , because the
equation of motion, Eq. (7.1), gives cx = Fo ⇒ x = Fo /c = xo if Fo is constant, as
ẍ and ẋ must be identically zero.
In dynamic problems one usually calculates the equilibrium position for con-
stant, static loads first, and thereafter, new co-ordinates are defined from this static
equilibrium position. Static loads can for instance be the weight of a ship in calm
water balanced by the equally constant buoyancy to yield a specific draught, or
wind and current forces acting on a moored platform giving a constant offset bal-
anced by the mooring arrangement.

Figure 7.3: Static displacement or response xo to a constant load Fo .

113
 Wave-Induced Loads and Ship Motions
7. MOTION RESPONSE TO LOADING
02 December 2008
Page 14

Kraft, förskjutning
Force F ( t )

x ( t ) 4 2 0 2 4
Displacement

t
Time
Tid

Figure 7.4: Figure 3.4 harmonic

The exciting The exciting harmonic
load F(t) and theload F(t) andresponse,
stationary the stationary
x(t),
Response,
for a linear system. x(t), for a linear system.

We can solve Equation (3.1) for the given harmonic load, Equation (3.6), simply by
substituting the particular solution Equation (3.7) into it
7.3 R ESPONSE TO H ARMONIC L OADS
(m + a ) !x! + bx! + cx = Fo cos (ωt)
A harmonic load
x = x" cos(ωt − ε )F(t) = Fo cos(ωt) , (7.6)
x! = −ωx" sin(ωt − ε )
where Fo is the load amplitude, ω = 2π/T is the angular frequency and T is the
!!
x = −ω 2
x" cos(ωt − ε )
time period. A harmonic load, as from regular waves for instance, gives a response
of the same harmonic type:
The substitution gives
x(t) = x̂ cos(ωt − ε) , (7.7)
(c − (m + a)ω 2 ) xˆ cos(ωt − ε ) − bωxˆ sin(ωt − ε ) = Fo cos(ωt )
in which x̂ is the amplitude of displacement and ε denotes the phase lag between
Using the trigonometric expressions for sine and cosine of angle differences then
the force and displacement. The motion x(t) is the stationary response to the
yields:
harmonic load and is the particular solution to Eq. (7.1) with the right hand side
F(t) given by Eq. (7.6), see 2Fig. 7.4.
(c − (m + a )ω ) xˆ (cos(ωt ) cos(ε ) + sin(ωt ) sin(ε )) −
We can solve Eq. (7.1) for the given harmonic load, Eq. (7.6), simply by
− bωxˆ (sin(ωt ) cos(ε ) − (cos(ωt ) sin(ε )) = F cos(ωt )
substituting the particular solution, Eq. (7.7), into it o
Identification and(m
separation
+ a)ẍ + bofẋ +
terms
cx =with sin(ωt) ,and cos(ωt) gives
F cos(ωt) o

(c − (m + a )ω 2 ) xˆ sin(ε ) − bωxˆ cos(ε ) = 0 …(3.8)

in which
and
x = x̂ cos(ωt − ε) ,
(c − (m + a)ω 2ẋ) xˆ=cos(
−ωε )x̂−sin(ω t −εε)
bωxˆ sin( ) =, Fo …(3.9)
ẍ = −ω 2 x̂ cos(ω t − ε) .
From Equation (3.8) follows directly
The substitution gives
bω
tan(ε ) =2
…(3.10)
(m + at)ω
c − (m + a) ω (cx̂−cos(ω −2ε)) − b ω x̂ sin(ω t − ε) = Fo cos(ω t) .

114
 7.3. Response to Harmonic Loads

Using the trigonometric expressions for sine and cosine of angle differences then
yields:


c − (m + a) ω 2 x̂ (cos(ω t) cos(ε) + sin(ω t) sin(ε))
−b ω x̂ (sin(ω t) cos(ε) − cos(ω t) sin(ε)) = Fo cos(ω t) .
Identification and separation of terms with sin(ω t) and cos(ω t) gives


c − (m + a)ω 2 x̂ sin(ε) − b ω x̂ cos(ε) = 0 , (7.8)
and

c − (m + a)ω 2 x̂ cos(ε) + b ω x̂ sin(ε) = Fo . (7.9)
From Eq. (7.8) follows directly
bω
tan(ε) = , (7.10)
c − (m + a)ω 2
i.e.  
bω
ε = arctan , (7.11)
c − (m + a)ω 2
where also the correct quadrant must be decided.
Squaring and adding Eqs. (7.8) and (7.9) gives after using the trigonometric
unity  
2


2 2
x̂ c − (m + a)ω + b ω = Fo2 ,
2 2
(7.12)
and finally, as the amplitude, x̂, by definition is positive,
Fo
x̂ = q . (7.13)
2 2 2 2
(c − (m + a)ω ) + b ω

7.3.1 Amplitude response function, response amplitude operator and transfer

function
The ratio x̂/Fo is an amplitude response function and gives the ratio between the
amplitude of the harmonic response to the amplitude of the harmonic disturbance
that excited it. It is often denoted T (ω) and has in this example the dimension
length/force (m/N),
x̂ 1
T (ω) = =q . (7.14)
Fo 2 2 2 2
(c − (m + a)ω ) + b ω
It is sometimes also called transfer function, but the name transfer function should
also encompass the phase lag. See further under complex notation. In sea keeping
the special name response amplitude operator (RAO) is often used and denotes
the ratio between the response amplitude of any studied variable to the wave am-
plitude. In some literature the RAO is defined as the square of the quantity used
here. The word operator is used because the RAO is used to operate on the fre-
quency spectrum of a sea state to produce a spectrum of motion, moment, stress
etc.

115
7. MOTION RESPONSE TO LOADING

7.3.2 Amplification factor

The ratio between the amplitude of the calculated harmonic response, x̂, and the
displacement, xo , which the mass would have got under a constant load the same
as the load amplitude, Fo , is known as the amplification factor and is always di-
mensionless:
x̂ x̂ x̂ c c
Y (ω) = = = =q . (7.15)
xo (Fo /c) Fo 2
(c − (m + a)ω 2 ) + b2 ω 2

This amplification factor is illustrated in Fig. 7.5 together with the corresponding
phase lag ε. In the figure the frequency axis is nondimensionalised by the natural
frequency, N = ωN /(2π), which by definition is the frequency of the undamped
eigenvibration. Thus the abscissa, frequency axis, in Fig. 7.5 is scaled in the
nondimensionalised frequency, which reads
f ω
Ω= = . (7.16)
N ωN
Equation (7.15) can now be written
c
Y (Ω) = q
2
(c − (m + a)(ωN Ω)2 ) + b2 (ωN Ω)2
1
=q , (7.17)
2 2 b2 2
(1 − Ω ) + c2 (ωN Ω)

and finally introducing also the damping ratio defined in section 7.1
1
Y (Ω , ξ ) = p . (7.18)
(1 − Ω2 )2 + 4(ξ Ω)2

7.3.3 Complex notation

It is often very convenient to use complex notation when treating harmonic oscil-
latory motion. In the equation of motion, Eq. (7.1), the real valued driving force,
F(t), is then substituted by
Fc (t) = Fo eı ω t , (7.19)
where thus the real driving force is


F(t) = Re Fo eı ω t = Fo cos(ω t) , (7.20)

because
eı ω t = cos(ω t) + ı sin(ω t) . (7.21)
Observe that when analysing performed work, power etc. the real valued quan-
tity, F(t), must be used. Sticking to complex notation then the sum of, Fc (t), and

116
 7.3. Response to Harmonic Loads

(a) (b)

Figure 7.5: (a) Amplification factor, Y (Ω , ξ ), and (b) phase lag, ε(Ω , ξ ), for
a system with one degree of freedom. Ω is the non-dimensional
frequency and ξ is the damping ratio.

its complex conjugate can be used, which sum equals 2F(t). Here we will use
Re(Fc (t)) when necessary.
The complex motion, xc (t), is then set to

xc (t) = x̂c (t) eı ω t = x̂ e−ı ε eı ω t = x̂ eı(ω t−ε) , (7.22)

with the complex “amplitude” containing the phase angle x̂c = x̂ e−ı ε and where
as before the real motion is
x(t) = Re (xc (t)) = x̂ cos(ω t − ε) . (7.23)
The complex velocity and acceleration are
ẋc (t) = ı ω x̂c eı ω t , (7.24)
ẍc (t) = −ω 2 x̂c eı ω t . (7.25)
Recall that the equation of motion, Eq. (7.1), is
(m + a) ẍ + bẋ + cx = F(t) . (7.26)
Substituting Eqs. (7.19) and (7.23) – (7.25) into Eq. (7.26) yields
−(m + a)ω 2 x̂c eı ω t + ı b ω x̂c eı ω t + c x̂c eı ω t = Fo eı ω t . (7.27)
Here we can divide both sides with eı ω t and extract x̂c :


x̂c c − (m + a)ω 2 + ı b ω = Fo , (7.28)
yielding
Fo
x̂c (ω) = . (7.29)
c − (m + a)ω 2 + ı b ω

117
7. MOTION RESPONSE TO LOADING

The complex amplification factor will now become

c
Yc (ω) = 2
= Y (ω) e−ı ε , (7.30)
c − (m + a)ω + ı b ω

which includes both the amplification factor, Y (ω), and the phase lag, ε(ω). The
amplification factor is given by the modulus of Yc (ω), i.e. Y (ω) = |Yc (ω)|, and
the phase lag by its argument, i.e. ε(ω) = arg (Yc (ω)).

Exercise 7.1
Confirm that this gives the same result as Eqs. (7.15) and (7.11).

7.3.4 The transfer function

The complex motion, xc (t), can now, with the use of Eqs. (7.22) and (7.29), be
written:
Fo eı ω t
xc (t) = x̂c eı ω t = , (7.31)
c − (m + a)ω 2 + ı b ω
but Fc (t) = Fo eı ω t and thus

1
xc (t) = Fc (t) = Tc (ω)Fc (t) , (7.32)
c − (m + a)ω 2 + ı b ω

i.e. the motion is for any regular frequency of excitation given by a complex
multiplication of a transfer function, Tc (ω), and the driving force, Fc (t).
A comparison with Eq. (7.14) reveals that

Tc (ω) = T (ω) e−ı ε , (7.33)

and thus the amplitude response function equals the modulus of the complex trans-
fer function and the phase lag equals ε(ω) = arg(Tc ), i.e. arctan(Im(Tc )/Re(Tc ))
with additional information of into which quadrant Tc (ω) points.

7.3.5 Response of a harmonic load on an initially fixed body

Note that the response to harmonic loads described above is valid only under the
condition that the excitation and the response have been going on a long time,
strictly speaking infinitely long time. If the body is let loose at time zero the
response of the body will be composed of the transient response to the release
and gradually approach the steady oscillatory motion as time goes by. Equiva-
lently, the force can start abruptly at time zero, which will give exactly the same
response. In Fig. 7.6 below the transient motion caused by a cosine excitation
Fo cos(ω t) started at time zero is illustrated. To make the difference even more
clear a comparison between the steady-state oscillation and the transient response
is shown in Fig. 7.7.

118
 7.3. Response to Harmonic Loads

Figure 7.6: The transient motion response of a one-degree-of-freedom system

caused by a cosine excitation Fo cos(ω t) started at time zero.

Figure 7.7: Comparison in the time domain between the transient motion of
Fig. 7.6 and steady-state oscillation for the same load amplitude.

119
7. MOTION RESPONSE TO LOADING

The implication of the difference is that - strictly speaking - harmonic, frequen-

cy-domain responses must be applied with some caution for sea states of short
duration. E.g. in tank tests for assessing “response amplitude operators” the tran-
sient in the test must be taken away by filtering the measuring signal. For ships
this is most important for roll, and for moored ships or platforms it is also im-
portant in the horizontal degrees of freedom, for which resonance of the moored
system may influence the response. For transient loads, it is thus important to be
able to solve the problem in the time domain also. This can be done by direct
time-integration methods as is illustrated for the six degree of freedom system in
the end of this chapter, (Eq. 7.34), or by convolution techniques described in the
next paragraph. Also for problems where non-linear damping or non-linear moor-
ing characteristics are important time-domain solutions are necessary as will be
shown later in Paragraph 11.9

7.4 S YSTEM WITH S IX D EGREES OF F REEDOM

The motion in six degrees of freedom of a ship at zero speed can be described
by the following six coupled equations of motion, which looks the same as Eq.
(9.13):
(M + A)η̈ + Bη̇ + Cη = F . (7.34)
Here M, A, B and C are 6 × 6 matrices, η a 6 × 1 vector with the six motions and
F a 6 × 1 vector with the three forces and the three moments. See Fig. 9.1. The
forces and moments, F, are functions of time which implies that the motions, η,
also are so.
If the coefficients, the elements of the matrices, in this equation all are con-
stants, it can be solved directly in the time domain for arbitrary loads by methods
equivalent to those in Paragraph 3.3. For large motions the elements of damping
matrix, B, is however depending on the quadratic drag forces and the velocity am-
plitudes relative to the water motion. For floating bodies the matrices A and B are
functions of the frequency of the motion and Eq. (7.34) can only be solved for
one harmonic motion at a time.

7.4.1 Solution in the frequency domain

As told above the matrices A and B are functions of the frequency of the motion
for floating bodies and one cannot make a time integration as simply as above, but
one must use a special convolution technology [7, 50, 2]. This is sometimes used
when non-linear terms as fender forces, drag forces, power take-off or mooring
reactions are included in the equation of motion. For motion of “small” amplitude
in an irregular sea state, however, a technique, where the wave is assumed to be
composed of a series of harmonic functions is mostly used. By multiplying each
harmonic, sine function, with the transfer function for the response of interest for
each frequency a series of harmonic response function are produced, the sum of

120
 7.4. System with Six Degrees of Freedom

which makes the total response. This method requires that A and B are functions
of only the frequency of the wave and not the wave amplitude.
In the equation of motion the motion, η(t), is simplest written as a complex
position vector, η c (t), with the elements (i = 1 , 2 , . . . , 6):

ηic = η̂ic eı ω t = η̂i e−ı εi eı ω t , (7.35)

and the force vector

Fic = F̂ic eı ω t = F̂i e−ı αi eı ω t , (7.36)
with individual phase lags αi for the six forces and moments. Compare with
Eq. (7.19).
If Eqs. (7.35) and (7.36) is substituted into the equation of motion (7.34) one
can divide both sides of the equation system by eı ω t and thereby eliminate the
time dependence except the phase lag between the motions and the forces:


−ω 2 (M + A) − ı ω B + C η c = F c . (7.37)

Equation (7.37) can be solved by help of some standard computer program

that can handle complex matrices; alternatively the equation system can be split
into two real-valued systems, one for the real part and the other for the imaginary
part.
Symbolically the solution can be written
 −1
η c = −ω 2 (M + A) − ı ω B + C Fc . (7.38)

The decomposition of the irregular load into harmonic components, solution

of the equation of motion for each component as above and superposition of the
resulting harmonic motion components makes it possible to fast predict statistical
properties of the studied motion without knowing its course in detail.

121
 8
HEADING AND
ENCOUNTER FREQUENCY

8.1 HEADING

The heading is defined with respect to the propagation direction of the waves as
shown in Fig. 8.1a. The angle µ between the intended track of the ship and the
direction of wave propagation is denoted the heading angle or heading.

Fig. 8.1b shows some common definitions used when defining a ships heading.

• µ = 0◦ Following waves with the waves and the ship traveling in the same
direction

• µ = 90◦ Beam waves with the waves coming from the starboard side.

• µ = 180◦ Head waves with the waves traveling in the opposite direction of
the ship.

• µ = 270◦ Beam waves with waves coming from the port side.

8.2 ENCOUNTER FREQUENCY

Encounter frequency is one of the most important parameters for a ships behavior
in waves. Wave disturbances at or close to a ships resonance frequency can cause
large motion amplitudes even at moderate sea states. It is therefore important to
know how the frequency experienced by the ship ωe is related to the ships speed
and heading.

123
8. HEADING AND
ENCOUNTER FREQUENCY

(a) (b)
Figure 8.1: (a) Definition of the heading angle µ and (b) names of common
heading directions.

The component of the ships velocity in the direction of the propagating wave
is

U cos(µ) (8.1)
The waves will overtake the ship with the relative velocity

c −U cos(µ) . (8.2)
A ship progressing through a following wave with the speed, U, at the heading,
µ, in relation to the direction of wave propagation, will be passed by a wave crest
every encounter period
λ
Te = . (8.3)
c −U cos(µ)
The angular frequency of encounter is thus
2π 2π
ωe = = (C −U cos(µ)) , (8.4)
Te λ
2π g
Since T = λc ;ω = T ;ω = c for deep water Eq. (11.73) can be written

ω2U
ωe = ω − cos(µ) . (8.5)
g

124
 8.2. ENCOUNTER FREQUENCY

In forward seas (90◦ < µ < 270◦ ) cos(µ) is always negative and the encounter
frequency is greater than the wave frequency. In beam sea (µ = 90◦ or 270◦ )
cos(µ) = 0 and the encounter frequency is equal to the wave frequency and is
not influenced by the by the ship speed.

An important but much more complicated situation occurs for (0◦ < µ < 90◦ )
or (270◦ < µ < 360◦ ). cos(µ) is then always positive and the encounter frequency
now has a maximum value
g
ωe max = (8.6)
4U cos(µ)

which occurs when the wave frequency is

ω = 2ωe max . (8.7)

The encounter frequency is zero when

ω = 4ωe max . (8.8)

In deep water this corresponds to

U cos(µ) = c (8.9)

since ω = g/c .

The encounter frequency is negative for higher values of ω which means that
the ship is overtaking the waves. Positive encounter frequency means that the
waves are overtaking the ship.
In following and quartering waves a given (absolute) value of encounter fre-
quency may be experienced in three different wave systems if | ωe |< ωe max as
shown in Fig. 8.2. Two of these wave systems will give positive encounter fre-
quencies and the third will give a negative encounter frequency. If | ωe |> ωe max
there will be one negative encounter frequency only.
The physical interpretation of this is that long waves having a high celerity
will give the same encounter frequency as shorter waves with lower celerity when
| ωe |< ωe max . Very short waves having a celerity lower than the ship speed will
be overtaken by the ship resulting in a negative encounter frequency.
In head or bow waves there will be only one encounter frequency for each
wavelength while following and quartering waves may allow for up to three dif-
ferent wave systems to give the same encounter frequency. The relation between
ship speed, heading, wavelength and encounter frequency is shown in Fig. 8.3.
For regular waves a wide range of wavelengths (100-1000m) will give nearly the
same encounter frequency for 0.25 < ωe < 0.15. For this encounter frequency
range a large number of wave components may contribute to excitations of ship
motions at the same frequency.

125
8. HEADING AND
ENCOUNTER FREQUENCY

Figure 8.2: Encounter frequency and heading

126
 8.2. ENCOUNTER FREQUENCY

Figure 8.3: Relation between encounter frequency, ship speed and wave length.

127
 9
THE SEAKEEPING PROBLEM

For hydrodynamic purposes a floating body can mostly be regarded as rigid but
moving. It then exhibits six motional degrees of freedom. See Fig. 9.1, where
also the co-ordinate system used for the seakeeping problem is shown. Such a
space-fixed, right-handed co-ordinate system is usually oriented with respect to
the position of the body in rest. Its origin is either placed in the centre of gravity
of the body or in the still water surface vertically above the centre of buoyancy as
is chosen here. Usually the z-axis is vertical and points upwards, the x-axis points
horizontally forward, as in Fig. 9.1.
The six oscillating motions of a floating body have established names in En-

Figure 9.1: Motional degrees of freedom for the Seakeeping problem.

129
9. THE SEAKEEPING PROBLEM

Table 9.1: Names of the motional degrees of freedom.

Legend English Swedish Norwegian Dutch

η1 surge — jage schrikken
η2 sway (svaja) svaie verzetten
η3 heave häva hive dompen
η4 roll rulla rulle slingeren
η5 pitch stampa stampe stampen
η6 yaw (gira) (gire) (gieren)

glish but lack this in some other languages, see Fig. 9.1 and Table 9.1.
Symbolically the equation of motion of a floating body can be written

M η̈ = F , (9.1)

where M is a mass matrix, η = [η1 , η2 , η3 , η4 , η5 , η6 ]T is the vector of positions

in the six degrees of freedom, η̈ = d 2 η/dt 2 denotes the body acceleration and
F = [F1 , F2 , F3 , F4 , F5 , F6 ]T = [Fx , Fy , Fz , Mx , My , Mz ]T is the vector of forces and
moments acting on the body.

In Chapter 6 the equations of motion for a rigid ship were derived for dif-
ferent levels of approximation regarding symmetry and position of the coordinate
system.
NOTE that the equations in Chapter 6 are derived for the manoeuvring prob-
lem with the z-axis pointing downwards. But the rigid body equations will be the
same also for z upwards.

The most general equations of motion Eq. 6.60 to Eq. 6.70 are nonlinear since
they contain products of the unknown degrees of freedom. It is therefore difficult
to solve the equations. A first step towards a practical method to solve the sea-
keeping problem is to linearize the equations. It is then assumed that the motions
are small and that products of the unknowns can be neglected. The linearized
equations are:

F1 = m[η̈1 + z̄G η̈5 ] (9.2)

F2 = m[(η̈2 − z̄G η̈4 + x̄G η̈6 ] (9.3)

F3 = m[η̈3 − x̄G η̈5 ] (9.4)

F4 = Ix̄ η̈4 − Ix̄z̄ η̈6 − m[z̄G η̈2 ] (9.5)

130
 Figure 9.2: Superposition of wave excitation (left); added mass, radiation damp-
ing and restoring loads (middle); to the total hydrodynamic problem
(right). From Faltinsen [12].

F5 = Iȳ η̈5 + m[z̄G η̈1 − x̄G η̈3 ] (9.6)

F6 = Iz̄ η̈6 − Ix̄z̄ η̈4 + m[x̄G η̈2 ] (9.7)

In Chapter 1 ship motions were divided into a ”Seakeeping problem” includ-

ing heave, roll and pitch and a ”Manoeuvring problem” including surge, sway and
yaw. The seakeeping problem is descibed by equations 9.4, 9.5 and 9.6. If the
coupling with surge, sway and yaw in the manouvring problem is assumed to be
small the equations can be written:

F3 = m[η̈3 − x̄G η̈5 ] (9.8)

F4 = Ix̄ η̈4 (9.9)

F5 = Iȳ η̈5 − m[x̄G η̈3 ] (9.10)

It can be noted that heave and pitch are coupled through Equations 9.8 and
9.10 while the equation 9.9 for roll is uncoupled.

The seakeeping problem for ship motion in waves can be solved by introduc-
ing:

1. the forces on the body when this is forced to oscillate with an arbitrary
amplitude in calm water,

2. the forces on the body when this is fixed in the incident waves,

3. the mooring forces on the body, and last

131
9. THE SEAKEEPING PROBLEM

4. the dynamic equilibrium or equation of motion when the difference between

the forces in points 1, 2 and 3 is balanced in each time instant by the inertia
force of the accelerating body.

The force vector F in Equation (9.1) can be split into the exciting forces, the
reaction forces from the water and from moorings if any. Neglecting, for the time
being, other exciting forces than the wave excited forces a convenient split of the
forces are:
F = F e + F r + F rs , (9.11)
in which F e contains the wave-excited forces on the fixed structure, F r denotes
the hydrodynamic reaction forces from the water on the moving body in the ab-
sence of the waves and F rs are the reaction forces from the mooring system.

The hydrodynamic reaction forces F r , i.e. the hydrodynamic properties of the

body is in the linear approximation characterised by three properties, namely:

A hydrodynamic mass or added mass

B hydrodynamic damping or radiation damping coefficients
C hydrostatic stiffness

The hydrodynamic reaction force can then be written:

F r = −Aη̈ − Bη̇ − Cη . (9.12)

Substituting Eq. (9.11) and Eq. (9.12) into Eq. (9.1) the following simple expres-
sion results:
(M + A) η̈ + +Bη̇ − Cη = F e + F rs . (9.13)

M, A, B and C in Eq. (9.13) are all 6 × 6 matrices with the elements Mi j , Ai j , Bi j

and Ci j (i, j = 1 , 2 , . . . , 6).

NOTE that the direction of the z-axis is now upwards. The direction of the z-
axis will have an influence for the sign of the wave excitation, hydrodynamic and
mooring forces. This sign convention will be used for the seakeeping problem.

Linear reaction forces from e.g. the mooring system can be included in the
coefficient matrices A, B and C above, while non-linear reaction forces must be
included in the right-hand side of the equation of motion. Additionally, for bodies
floating in the water surface or bodies being submerged but close to the water
surface, the hydrodynamic properties A and B are functions of the frequency of
the motion, which is caused by the generation of waves around the body. Then
Eq. (9.13) can be easily solved only for cases when the excitation is a harmonic
function.

132
 10
SOLUTION METHODS -
STRIP THEORY

10.1 S OLUTION METHODS IN GENERAL

The Seakeeping problem defined in Chapter 9 can be solved in several ways.

The most advanced methods are based on numerical methods for viscous flow
including free surface capability and a 6DOF model for the ship motions. These
methods show the ”State of the Art” for the research in seakeeping, but they are
not yet at a level where they can be used for practical design work. Computation
time is one reason and the capability to compute the complex water flow close to
the hull in waves is another. Fully 3D potential methods are also available for the
seakeeping problem. Here the viscosity is neglected and the details of the flow can
not be resolved. The main effects for heave and pitch are captured by potential
flow methods, but the viscous contributions are important for the roll damping.
The viscous effects on roll damping must therefore be introduced though a roll
damping model.

10.2 S TRIP THEORY

The methods used for practical design work today are based on the potential flow
assumption, but in addition it is assumed that the ship can be represented by a
number of 2D strips, see Fig. 10.1. The 3D effects are thus assumed to be small
and are therefore neglected. This assumption implies that:

• The ship is slender

• The hull is rigid

• Moderate speed, no planing

133
10. SOLUTION METHODS -
STRIP THEORY

Figure 10.1: The hull is represented by 2D strips

• The motions are small

• The ship hull sections are wall sided

• Deep water

With this assumption the coefficients for added mass, damping and restoring
force can be determined by simple 2D potential flow solutions for each strip. Al-
ternatively the coefficients can be computed based on analytical solutions of each
strip after coordinate transformation to a circular 2D cylinder, see Section 10.3.
The coefficients can also be determined by experiments for 2D cylinder having
the shape of the hull segments. A summation over the strips will then give the
properties for the hull.

Experience shows that the strip methods in many cases can be used for practi-
cal applications even if the assumptions are violated. The strip methods can also
be extended to take the nonlinear effects of the wavy intersection between the hull
and the free surface into account when the pressure forces are computed. These
methods are referred to as ”Nonlinear Strip Methods”.

10.3 L EWIS FORMS

A widely used method to determine the sectional added mass and damping prop-
erties for heave, sway and roll is via Lewis transforms. This method is based
on analytical solutions of added mass and damping for ship-like sectional forms

134
 10.3. Lewis forms

Figure 10.2: Typical Lewis forms

known as Lewis forms. The Lewis forms are controlled by two parameters: the
Beam/Draught ratio H and the sectional area coefficient σ . Typical Lewis forms
are shown in Fig. 10.2 for 0.5 < H < 6 and 0.5 < σ < 0.9.
The analytical expressions for the Lewis forms are:

B 2(1 + a1 + a3 )
H= = (10.1)
D 1 − a1 + a3
 
A π 1 − a21 − 3a23
σ= = (10.2)
BD 4 1 − a21 + 2a3 + a23
where
 
H −2
a1 = (1 + a3 ) (10.3)
H +2
√
3 −C + 9 − 2C
a3 = (10.4)
C

135
10. SOLUTION METHODS -
STRIP THEORY

Figure 10.3: Examples of invalid Lewis forms

  
4σ 4σ H −2 2
C = 3+ + 1− (10.5)
π π H +2

In addition there is a relation between the Beam/Draught ratio H and the sec-
tional area coefficient σ that must be satisfied:

π
σ< (H 2 + 20H + 4) (10.6)
64H
Examples of invalid Lewis forms are shown in Fig. 10.3
The sectional 2D added mass and damping properties for heave, sway, and roll
are available through analytical expressions for valid forms. For details see [30].
Comparison between results obtained from Lewis forms and by experiments are
shown in Figs. 10.7, 10.8 and 10.9.
It should be noted that the Lewis form is a two-parameter approximation of
the real sectional shape. A comparison between sections of a real body plan and
the corresponding Lewis forms is shown in Fig. 10.4

136
 10.4. Experimental methods

Figure 10.4: Comparison of real sectional shape and Lewis form representation

10.4 E XPERIMENTAL METHODS

Sectional properties of damping and added mass can also be determined by ex-
periments. Some of the experiments concerns sectional forms that can be exactly
represented by Lewis forms, while other experiments include details of the shape
that cannot be captured by Lewis forms. Experiments were carried out by Vugts
at Delft Shipbuliding Laboratory [51] for a number of sectional forms. The ex-
perimental set up is indicated in Fig. 10.5 and the forms are shown in Fig. 10.6.
Note that the differences between cylinders A, B and C cannot be captured
by Lewis forms since they have the same Beam/Draught ration and sectional area
coefficient. Results for the cylinders are compared to Lewis form theory in Figs.
10.7, 10.8 and 10.9.

137
10. SOLUTION METHODS -
STRIP THEORY

Figure 10.5: Vugts 2D experiments

Figure 10.6: Vugts cylinders

138
 10.4. Experimental methods

Figure 10.7: Hydrodynamic coefficients for the three heaving cylinders A, B

and C

139
10. SOLUTION METHODS -
STRIP THEORY

Figure 10.8: Hydrodynamic coefficients for heave, rectangular and triangular

cylinders

140
 10.4. Experimental methods

Figure 10.9: Hydrodynamic coefficients for sway, rectangular cylinders

141
 11
SHIP IN REGULAR WAVES

We will first study the problem of wave-induced motions of a two-dimensional

floating body in regular sea. With two-dimensional body is meant an infinitely
long ship, whose cross-sectional shape is constant in the x-direction, and where
all body motions (heave, sway and roll) and the water flow are parallel to the
y − z-plane. This problem is fundamental in the much used strip theory, in which
two-dimensional slices of mildly varying sectional ship shapes are added two form
oblong approximately three-dimensional ships, as will be demonstrated later.

11.1 T HE T WO -D IMENSIONAL H EAVE P ROBLEM

11.1.1 Forces on the two-dimensional ship heaving in calm water

Assume that the ship is forced to oscillate in heave in calm water with the ampli-
tude, η̂3 , and the angular frequency, ω:

η3 = η̂3 cos(ω t) . (11.1)

The arising forces are due to the vertical position, η3 , Eq. (11.1), vertical velocity

η̇3 = −ω η̂3 sin(ω t) , (11.2)

and vertical acceleration

η̈3 = −ω 2 η̂3 cos(ω t) , (11.3)

relative the still water.

The vertical reaction force acting on the body from the water can be written

f3 = −a33 η̈3 − b33 η̇3 − c33 η , (11.4)

where the coefficients a33 , b33 and c33 must be assessed.

143
11. SHIP IN REGULAR WAVES

Figure 11.1: A standing, or evanescent, wave around a heaving ships cross-

section.

The vertical restoring force, −c33 η3 , is due to the change of displacement. If

the beam (width, breadth) of the body is B then the stiffness, c33 , must be

c33 = ρ g B . (11.5)

The dynamic forces due to the acceleration and velocity of the body are associated
with the forced oscillatory motion of the ambient water. For example the accelera-
tion of a floating body is associated with a local, evanescent wave (standing wave,
clapotis) when the water is forced to shift back and forth between the bottom and
the sides. See Fig. 11.1.

11.1.2 Added mass

As the force a33 η̈3 is in phase1 with the force needed to accelerate the body, the
coefficient a33 is often called the “added mass” and sometimes the hydrodynamic
mass.
For bodies floating in the water surface or positioned close to the water surface
the added mass is a function of the frequency of oscillation due to the evanescent
waves and degree of resonance. Far from the free surface the added mass is con-
stant and depends only on the shape of the body and its vicinity to other bodies
or fixed boundaries. Also for floating bodies, in the limit, as ω → ∞, no waves
are formed, and the added mass becomes independent on the frequency. See Fig.
11.2b for an example of added-mass coefficients for a rectangular cross section in
water shallow compared to the draft.

1 Or sometimes in antiphase for slightly submerged bodies, that is, the added mass a33 is nega-
tive.

144
 11.1. The Two-Dimensional Heave Problem

11.1.3 Radiation damping

The heave motion also causes waves that radiate out from the body, and the force,
needed to maintain this oscillation is in phase with the velocity. The radiated
waves transport energy away from the body and therefore introduces hydrody-
namic or radiation damping. Therefore the coefficient, b33 , is called the damping
coefficient or radiation damping coefficient, when stressing the phenomenon of
the radiated waves. For bodies floating in the water surface or positioned close to
the water surface also the radiation damping is a function of the frequency of os-
cillation due to the generation of waves. For such bodies, in the limit, as ω → ∞,
no waves can be formed, and the radiation damping is nil as it is for deeply sub-
merged bodies, that cannot produce surface gravity waves. The radiation damping
is actually nil both for ω → 0 and ω → ∞. There is a maximum somewhere in
between where the body has a maximal ability to radiate energy or reciprocally
absorb wave energy. See Fig. 11.2c for an example of radiation-damping coeffi-
cients for a rectangular cross section in water shallow compared to the draft.
As stated above any body forced to oscillate in the water surface will create
waves that propagate away from the body thus transporting out or radiating energy.
For the two-dimensional case this will be illustrated below. See Fig. 11.3.
The damping force in heave for a two-dimensional body can be written

frd3 = −b33 η̇3 . (11.6)

During a period, T , of oscillation, the exciting force − frd3 necessary to maintain

a harmonic motion at the amplitude η̂3 will exert the mean power
Z T Z T

1 1 η̂ 2
(− frd3 η̇3 ) dt = b33 η̇32 dt = ω 2 b33 3 . (11.7)
T 0 T 0 2
The body will radiate waves equally to starboard and port side with the ampli-
tude, a, and the radiated power –in deep water – to one side is according to Equa-
tion (5.64)
1 ω
ρ g a2 . (11.8)
4 k
Setting twice this radiated power equal to the exerted power - in order to take both
starboard and portside wave into account - gives

η̂32 1 ω
ω 2 b33 = 2 ρ g a2 , (11.9)
2 4 k
from which b33 can be solved
 2  2
a 1 a g2
b33 = ρ g =ρ , (11.10)
η̂3 kω η̂3 ω3
which is valid for all frequencies and shows that the damping is always positive,
which is not necessarily true for the added mass. Remember that according to the

145
11. SHIP IN REGULAR WAVES

(a)

(b)

Figure 11.2: A two-dimensional rectangular hull in shallow water. (a) Nota-

tions: 2T = B, d/T = 2, d/B = 1, ∀ = T B = displacement, k =
wave number and b = B/2; (b) added mass coefficients. From [25].

146
 11.1. The Two-Dimensional Heave Problem

(c)

Figure 11.2: A two-dimensional rectangular hull in shallow water (continued).

(c) radiation-damping coefficients. From [25].

147
 33 $ ηˆ ! kω $ ηˆ ! ω 3
% 3" % 3"

which is valid for all frequencies and shows that the damping is always positive,
which is not necessarily true for the added mass. Remember that according to the used
linear assumptions and the potential theory, the given coefficients are only valid for
11. SHIP IN REGULAR WAVES
small-amplitude motion and viscous non-linear damping is not taken into account
either. In reality the total damping is always larger than the radiation damping.

ζ ( x, t )

ζ( x, 0.25⋅ s )

ζ( x, 0.5⋅ s)

Figure 11.3: Radiated

x
2D waves.

Figure 6.5 Radiated 2D waves

used linear assumptions and the potential theory, the given coefficients are only
Two-dimensional added-mass and radiation coefficients in heave
valid for small-amplitude motion and viscous non-linear damping is not taken
In order to calculate the added mass and the radiation-damping coefficient potential
into account either. In reality the total damping is always larger than the radiation
problems described by Laplace equation are solved. For heave this is done by adding
damping.
the solutions for the incident undisturbed wave (Chapter 5), the solution for the
scattered (reflected and transmitted) wave from the fixed body and the radiated wave
11.1.4fromTwo-dimensional
the body heavingadded-mass
in calm water.
and The solution
radiation should satisfy
coefficients the boundary
in heave
conditions on the water surface, the sea bottom and on the wetted part, hull, of the
body.
In order to The first two
calculate conditions
the added masswere discussed
and the in Chapter 5.coefficient
radiation-damping The last condition
poten-
demandsdescribed
tial problems that the water velocity
by Laplace perpendicular
Equation to theFor
are solved. hullheave
mustthis
be the samebyas the
is done
addingvelocity of the hull,
the solutions so that
for the no water
incident passes the
undisturbed hull,the
wave, thatsolution
is, the total relative
for the scat-velocity
teredperpendicular
(reflected andtotransmitted)
the hull must be nil.
wave from Also
the the radiation
fixed body andcondition must wave
the radiated be satisfied
from the body heaving in calm water. The solution should satisfy the boundary and
which for the two-dimensional problem means that the reflected, transmitted
radiated
conditions onwaves mustsurface,
the water propagate theoutwards
sea bottom from theonbody
and the and
wettedbe bound.
part, hull, of
the body. The first two conditions were discussed in Chapter 3. The last condi-
The solution
tion demands thatcanthe be attained
water by simple
velocity Rankin sources
perpendicular distributed
to the hull must beover all the
the same
boundaries of the problem, or by smarter Greens’ functions or Havelock sources
as the velocity of the hull, so that no water passes the hull, that is, the total rel-
fulfilling by themselves the free-boundary and radiation conditions. In this chapter we
ative velocity perpendicular to the hull must be nil. Also the radiation condition
will not describe this further but accept the results calculated by Grim (1959) for
must various
be satisfied which for
Lewis-form the two-dimensional
cross sections.xxix problem means that the reflected,
transmitted and radiated waves must propagate outwards from the body and be
bound.The added mass can then be assessed by reading the coefficient C from the diagrams
The solution
in Figure can be attained by simple Rankine sources distributed over all the
6.7,
boundaries of the problem, or by smarter Greens’ functions or Havelock sources
fulfilling by themselves the free-boundary and radiation conditions. In this chapter
we will not describe this further but accept the results calculated by Grim [16] for
various Lewis-form cross sections.
The added mass can then be assessed by reading the coefficient C from the
diagrams in Fig. 11.4,
ρ π B2 C
a33 = , (11.11)
8
where B as before is the beam of the ship section and C is a function of

• the beam to draught ratio B/T ,

• the section-area coefficient β = S/(BT ) and

• a dimensionless wave number expressed as ωe2 B/(2 g).

148
 11.1. The Two-Dimensional Heave Problem

Figure 11.4: Coefficient, C, of added mass for two-dimensional floating ship-

shaped Lewis-form sections. From [5].

149
11. SHIP IN REGULAR WAVES

Figure 11.5: Coefficient, A, of radiation damping for two-dimensional floating

ship-shaped Lewis-form sections. From [5].

150
 11.1. The Two-Dimensional Heave Problem

In Fig. 11.4 B and β are indexed n to stress that they are valid for a section No
n along the ship. The angular frequency ωe is the encounter, angular frequency
experienced by the ship. At zero speed the encounter frequency is identical to the
wave frequency.
Similarly the radiation-damping coefficient can be assessed by reading the
coefficient A from the diagrams in Fig. 11.5. Here A is the ratio between the
amplitude of the radiated waves and the amplitude of the driving heave motion.
Compare Equation (6.10).
ρ g2 A2
b33 = . (11.12)
ω3
A is a function of the same parameters as the added mass, namely B/T , β , and
ωe2 B/(2 g).
Jensen et al. [23] simply used for a ship at speed V and heading, θ , the constant
added mass, a33 = ρ g B T , with astonishingly good result, and a somewhat more
complicated closed-form expression for b33 :

ρ g2 A2
b33 = 3 3 ,
ω α
where    
ωe2 B −
ωe2 T 1 2
A = 2 sin e g = 2 sin k B α e−k T α .
2
2g 2

√given byωe = ω −kV cos(θ ) ≡ α ω, the param-

Furher, the encounter frequency is
defined as α = 1 − Fn k L cos(θ ), in which Fn is the Froude number,
eter α is √
Fn = V / k L, and V the forward speed.

11.1.5 Wave-excited heave force

To set up the wave excited heave force we assume the body to be fixed in a regular
beam wave progressing from portside to starboard

ζ = a cos(k y + ω t) , (11.13)

and ask ourselves which forces that will act on the body. See Fig. 11.6. This so-
called wave-diffraction or scattering problem can be solved in a similar manner as
used for solving the added mass and the radiation-damping coefficient. The differ-
ence is now that the boundary condition on the hull states that the water velocities
perpendicular to the hull surface should be nil, that is, the water particle velocities
of the scattered wave must compensate the normal water particle velocities of the
incident wave. This will cause a part of the wave to be reflected and the other part
to be transmitted passed the hull.
To make it simple here we will restrict our treatise to using the small-body
assumption, which demands that the cross-section width, beam, should be less
than a fourth to fifth of a wavelength for producing reasonable results. Then the

151
11. SHIP IN REGULAR WAVES

Figure 11.6: A ship cross-section in beam waves.

variation of incident wave properties across the beam can be neglected2 , which is
equivalent of setting y = 0 in Eq. (11.13).
Sticking to the small-body assumption the wave excited force acting on the
cross-section can be expressed by help of the same coefficients a33 , b33 and c33 as
above.
Assume temporarily the cross-section is rectangular with a flat, horizontal bot-
tom and the draught T , and further that the wave is not distorted by the presence
of the ship, which by help of the more complete theory can be shown to be ap-
proximately valid. Then the pressure from the incident wave on the ship’s bottom
is  
−k T
p = ρ g T +ae cos(ω t) , (11.14)

where the first term is balanced by the displacement of the ship and can be dis-
missed in this context, and the wave excited force due to relative displacement
will be

B ρ g a e−k T cos(ω t) = c33 a e−k T cos(ω t) = c33 ζ e−k T , (11.15)

as
c33 = B ρ g . (11.16)
That is, the wave-excited force depends on the relative motion between the wave
and the fixed ship, in the same way as the reaction force on the ship in still water
depended on the motion of the ship. This force is calculated using the undisturbed
2 Some authors use the term long-wave approximation, but this notation is in wave contexts

mostly used to denote wave-theory approximations for waves with a longer wavelength than 10 to
20 water depths.

152
 11.1. The Two-Dimensional Heave Problem

pressure in the incident wave on the body surface and is often named the Froude-
Krylov force. It can with small errors be used under the small-body assumption,
wile for shorter wave in relation to the body diameter the exciting pressure must
be calculated from the disturbed potential flow and the Bernoulli equation taking
the presence of the body into account. See e.g. discussion in Korvin-Krokowsky
and Jacobs [27] in connexion with the ship motion problem.
When the wave passes the fixed ship the relative velocity will be

ζ˙ = −aω e−k T sin(ω t) , (11.17)

and the relative acceleration

ζ¨ = −aω 2 e−k T cos(ω t) , (11.18)

The total exciting force can therefore be written

 
f3e = a33 ζ¨ + b33 ζ˙ + c33 ζ e−k T . (11.19)

If the ship’s cross-section is not full the draught, T , can as an approximation be

exchanged by the cross-sectional mean draught T̄ = S/B.

11.1.6 Heave motion of the two-dimensional ship-section in regular beam waves

If now the two-dimensional ship-section is let free to heave without rolling, the
total relative motion at the ships bottom between the ships bottom and the incident
wave is η3 − ζ e−k T and the sum of the wave excited force Eq. (11.19) and the
reaction force Eq. (11.4) on the moving ship must be balanced by its inertia force
m η̈3 :
 
f3e + f3 = −a33 η̈3 − b33 η̇3 − c33 η3 + a33 ζ¨ + b33 ζ˙ + c33 ζ e−k T = m η̈3 .
(11.20)
Separating this equation so that terms containing the dependent unknown variable
η3 is gathered on the left-hand side gives
 
¨ ˙
(m + a33 ) η̈3 + b33 η̇3 + c33 η3 = a33 ζ + b33 ζ + c33 ζ e−k T , (11.21)

which can be solved by the same technology as was demonstrated in Chapter 7.

The difference from Chapter 7 is now that the “driving force” on the right-hand
side displays a phase lag, α, in relation to the wave.
 
f3e = a33 ζ¨ + b33 ζ˙ + c33 ζ e−k T
q
2
= (c33 − ω 2 a33 ) + (ω b33 )2 e−k T ζˆ cos(ω t − α) . (11.22)

Again the easiest way to solve this problem is to use complex notation. Compare
Eq. 11.27.

153
11. SHIP IN REGULAR WAVES

Let, here, the complex wave progress from portside to starboard, i.e. in the
negative y-direction, compare Eq. 11.13.

ζc = a eı (k y+ω t) = a eı k y eı ω t = a (cos(k y + ω t) + ı sin(k y + ω t)) , (11.23)

so that the real progressive wave is given by

Re (ζc ) = Re (a (cos(k y + ω t) + ı sin(k y + ω t))) = a cos(k y + ω t) , (11.24)

the vertical displacement, velocity and acceleration of the water at the depth
z = −T are

ζc |z=−T = a e−k T eı ω t , (11.25a)

ζ˙c |z=−T = ı a ω e−k T eı ω t , (11.25b)
ζ¨c |z=−T = −a ω 2 e−k T eı ω t . (11.25c)

The dynamic pressure is, again dismissing the statically balanced mean pres-
sure ρ g T in Eq. (3.44)

pc |z=−T = ρ g a e−k T eı ω t . (11.26)

Let similarly the heave motion be

η3c = η̂3c eı ω t , (11.27)

where now also the amplitude is complex containing the information of the phase
angle between the wave and the heave motion. The vertical velocity and acceler-
ation are then

η̇3c = ı ω η̂3c eı ω t , (11.28a)

2 ıω t
η̈3c = −ω η̂3c e . (11.28b)

Substitute this into the equation of motion Eq. (11.20) and solve for the complex
amplitude η̂3c :
c33 − a33 ω 2 + ı ω b33
η̂3c = a e−k T . (11.29)
c33 − (m + a33 )ω 2 + ı ω b33
The motion amplitude is finally given by the modulus, η̂3 = |η̂3c |, the phase angle
between wave motion and heave by the argument, arg (η̂3c ), and the complex
transfer function from wave motion to heave motion by Tc = η̂3c /a. The factor
e−k T is often called the Smith effect and shows that the deeper the draught the less
are the excitation forces. In Fig. 11.7 the response amplitude operator RAO3 =
|η̂3c | /a and the wave-excited force normalised by the displacement force ρ S are
shown as functions of angular frequency. In Fig. 11.8 the phase angles between
wave and heave motion; and between wave and wave-excited force as functions
of angular frequency are shown.

154
 11.1. The Two-Dimensional Heave Problem

Figure 11.7: The heave-response amplitude operator, RAO3 , (continuous line)

and 10 times the wave-excited force divided by the displacement
force (dashed curve) as functions of normalised angular frequency,
Ω = ω/ωN , for a two-dimensional ship.

Figure 11.8: The phase angles between wave and heave motion (continuous
line) and between wave and wave-excited force (dashed line) as
functions of normalised angular frequency, Ω = ω/ωN , for a two-
dimensional ship.

155
11. SHIP IN REGULAR WAVES

In Fig. 11.7 it is seen that for long waves ω → 0 the amplitude of the heave
motion is the same as the amplitude of thepwave, then grows to a maximum just
below the natural angular frequency, ωN = c33 /(m + a33 ), and finally attenuates
to nil for higher frequencies.
In Fig. 11.8 it is seen that for long waves ω → 0 the wave-excited pforce is
in phase with the wave, then as the wave frequency increases to ω = c33 /a33
the force is 90◦ before the wave and for higher frequencies the force becomes in
opposition to the waveform. For all frequencies the maximum of the wave-excited
force appears before the maximum of the wave elevation, 0 < εwF < 180◦ . The
heave motion, on the other hand, appears for all frequencies after both the wave
motion and the wave-excited force, εwm < 0◦ < εwF .
Note, however, that strictly speaking the small-body assumption in this ex-
ample is violated for angular frequencies above approximately 1 rad/s. We have
furthermore assumed that the added mass and the radiation-damping coefficient
are constant. Yet, the features of the response are valid. A more thorough calcu-
lation using correct potential forces and coefficients derived from potential theory
would give the same result in principle, although with somewhat different graphs,
especially for angular frequencies above 1 rad/s.

11.1.7 The sway and roll problem

The sway problem can for a two-dimensional ship be treated similarly but as con-
cerns the roll problem the dynamic equilibrium must be solved for the ship as
an entity as the roll stiffness is due to the mass distribution of the entire ship.
Therefore we will save the roll problem till later.

11.2 T HE U NCOUPLED T HREE -D IMENSIONAL H EAVE P ROBLEM U SING

S TRIP T HEORY

11.2.1 Strip method for a ship

The simple method used above for the two-dimensional ship can be used also
for real three-dimensional ships, in an approximate variant of the so called strip
theory, where the ship is divided into slices - or strips of the hull surface - and
the two-dimensional flow problem is assumed to hold for each slice or strip (see
Fig. 11.9). The problem is then reduced to integrating the forces or hydrodynamic
characteristics along the ship knowing the added-mass and radiation-damping co-
efficients for each strip. The ship has to be reasonably slender, at least, L/B > 5
to neglect the end effects. The approximation is also better for ships with pointed
ends as the three-dimensional end effects are less pronounced for such ships.

156
 11.2. The Uncoupled Three-Dimensional Heave Problem Using Strip Theory

Figure 11.9: In the strip theory the ship is divided into slices or strips of the hull
surface. For each strip the two-dimensional hydrodynamic prob-
lem is then solved.

11.2.2 Forces on the three-dimensional ship heaving in calm water

Assume that the ship is forced to oscillate without pitching in calm water with the
amplitude, η̂3 , and the angular frequency, ω:

η3 = η̂3 cos(ω t) . (11.30)

The vertical reaction force acting on the body from the water can after integrating
the two-dimensional reaction force, Eq. 11.4, along the ship be written

F3 = −A33 η̈3 − B33 η̇3 −C33 η3 , (11.31)

where now the coefficients A33 , B33 and C33 are the integrated or summed quanti-
ties along the ship
Z
A33 = a33 (x)dx = ∑ a33n ∆xn , (11.32)
L n
Z
B33 = b33 (x)dx = ∑ b33n ∆xn , (11.33)
L n
Z
C33 = c33 (x)dx = ∑ c33n ∆xn = ∑ ρ g Bn ∆xn . (11.34)
L n n

As an example in this compendium we use a box-like ship, which due to the

bluff ends is not ideal for the strip-theory approximation, but makes the mathe-
matics simple. Note again that the coefficients a33 and b33 and thus A33 and B33
are functions of the oscillation frequency. For simplified estimates a characteris-
tic frequency of the exciting wave can be used and, actually, for each sea state the
resulting sea-keeping properties will be astonishingly realistic.
For the box-like ship thus

A33 = a33 L , (11.35a)

B33 = b33 L , (11.35b)
C33 = c33 L = ρ g B L = ρ g AW L , (11.35c)

where B is the beam and AW L is the water-line area.

157
11. SHIP IN REGULAR WAVES

11.2.3 Wave-excited force in heave

The heave wave-excited force on the fixed ship in waves can be written
Z L/2 Z L/2 Z L/2
F3e = p(x, −T ) B dx + ẇ(x, −T ) a33 dx + w(x, −T ) b33 dx ,
−L/2 −L/2 −L/2
(11.36)

where T is the draught; p(x, −T ) the undisturbed pressure at the box bottom
z = −T ; ẇ(x, −t) the water acceleration at the bottom z = −T ; w(x, −T ) the water
velocity at the bottom z = −T ; a33 the vertical two-dimensional added mass; b33
the vertical two-dimensional radiation damping coefficient; B the beam (breadth)
of the ship section and, finally, L the length of the ship.
Also in this case we have utilised the “small-body” assumption that the beam
of the ship should be small in relation to the wavelength, i.e. B < λ /4. For
following or meeting waves this is not a restrictive assumption as the apparent
wavelength for theses two conditions is infinitely long.
Again we will, for simplicity use complex notation. Also, because it is more
useful and not very complicated we will directly formulate the wave-excited force
for waves oblique to the ship in the following example. The potential for such
waves in complex notation is
a g k z ı k (x cos(θ )+y sin(θ )) −ı ω t
φc = ı e e e . (11.37)
ω
For the ship heading along the x-axis i.e. y = 0 it reduces to:
a g k z ı k x cos(θ ) −ı ω t
φc = ı e e e .
ω
Further, at y = 0 we also have the water vertical motion at the bottom z = −T :

ζc (x, −T ) = a e−k T eı k x cos(θ ) e−ı ω t ,

and the pressure at the bottom

pc (x, −T ) = ρ g a e−k T eı k x cos(θ ) e−ı ω t ,

vertical velocity at the bottom

wc (x, −T ) = −ı a ω e−k T eı k x cos(θ ) e−ı ω t ,

and vertical acceleration at bottom

ẇc (x, −T ) = −a ω 2 e−k T eı k x cos(θ ) e−ı ω t .

Substituting these expressions into Eq. (11.36) gives

Z L/2

F3e = ρgaB − a ω 2 a33 − ı a ωb33 e−k T eı k x cos(θ ) e−ı ω t dx , (11.38)
−L/2

158
 11.2. The Uncoupled Three-Dimensional Heave Problem Using Strip Theory

or

Z L/2
F3e = a e−k T e−ı ω t ρgB − ω 2 a33 − ı ωb33 eı k x cos(θ ) dx . (11.39)
−L/2

The integral is solved separately as

Z L/2  
1
eı k x cos(θ ) dx = eı k cos(θ )L/2 − e−ı k cos(θ )L/2
−L/2 ı k cos(θ )
2ı sin(k cos(θ )L/2)
=
ı k cos(θ )
sin(k cos(θ )L/2)
=2 , (11.40)
k cos(θ )
which is also valid for beam sea θ = π/2, because for small arguments cos(θ ) → 0,
sin(k cos(θ )L/2) → k cos(θ )L/2 and
sin(k cos(θ )L/2) k cos(θ )L/2
2 =2 = L. (11.41)
k cos(θ ) k cos(θ )
That is, in beam sea the wave-excited force on the 3-D ship will equal the two-
dimensional wave-excited force multiplied by the length of the ship.
If, however the waves progress along the longitudinal direction of the ship,
following or meeting sea, for very long waves the force is in principle the same
as for beam sea, but when the wavelength decreases the wave-excited force will
decrease so that when the wavelength is close to the length of the ship it will be at
a minimum. And for still smaller wavelengths, higher frequencies, the force will
approach nil. See Fig. 11.10.
Now set the ship motion to η3c = η̂3c e−ı ω t , which gives the vertical velocity
and acceleration of the ship

η̇3c = −ı ω η̂3c e−ı ω t ,

η̈3c = −ω 2 η̂3c e−ı ω t .

Using Eq. 11.31 with the complex motion above the equation of motion finally
gives


C33 − (m + A33 )ω 2 − ı ω B33 η̂3c e−ı ω t =

sin(k cos(θ )L/2)
a e−kT ρgB − ω 2 a33 − ıωb33 e−ı ω t 2 , (11.42)
k cos(θ )
which can be solved for the complex motion to form a nice closed-form expression

ρgB − ω 2 a33 − ı ω b33 sin(k cos(θ )L/2)

η̂3c = a e−k T 2
2 . (11.43)
ρgBL − (ρT BL + a33 L) ω − ı ω b33 L k cos(θ )

159
11. SHIP IN REGULAR WAVES

Figure 11.10: Heave wave-excited force as a function of ship length over wave-
length, L/λ .

Exercise 6.1
Inspecting Eq. (11.43) one can see that the amplitude response function
|η̂3c |/a has one minimum around ρgB = ω 2 a
33 , minima in deep water
for sin(k cos(θ )L/2) = sin (ω 2 /g) cos(θ )L/2 , and a maximum around
ρgBL = (ρT BL + a33 L)ω 2 .

Please explain why.

The natural angular frequency is obtained by psetting the driving force and
damping to nil in Eq. (11.42), which gives ωN = C33 /(m + A33 ).
In Fig. 11.11 the heave motion as a function of time at zero speed in head
waves are shown, in Fig. 11.12 the amplitude response and in Fig. 11.13 the phase
lag between heave motion and wave are shown as functions of angular frequency.

11.3 T HE P ITCH P ROBLEM U SING S TRIP T HEORY

11.3.1 Forces on the three-dimensional ship pitching in calm water

Assume now that the ship is forced to oscillate in pitch in still water without
heaving with the amplitude, η5 , and the angular frequency, ω:
η5 = η̂5 cos(ω t) . (11.44)
The vertical reaction moment acting on the body from the water can by integrating
the two-dimensional reaction force, Eq. 11.4, along the ship be written
F5 = −A55 η̈5 − B55 η̇5 −C55 η5 . (11.45)

160
 11.3. The Pitch Problem Using Strip Theory

Figure 11.11: Heave motion and water level as a function of time in head waves
with the wave amplitude a = 1 m.

where now the coefficients A55 , B55 and C55 are the integrated quantities along the
ship:

Z L/2
U2
A55 = a33 (x) x2 dx −A33 , (11.46)
−L/2 ωe2
Z L/2
U2
B55 = b33 (x) x2 dx + 2 B33 , (11.47)
−L/2 ωe
Z L/2
C55 = ρgV (zB − zG ) + ρg B(x) x2 dx = ρgV GM L . (11.48)
−L/2

Here the first terms in the added mass and radiation-damping coefficient are caused
by the motion in still water as in Eq. 11.4 or 11.31 taking into account that the
vertical motion due to pitch at each cross section is the pitch motion multiplied
by the lever, x, and that the reaction moment also is the sectional reaction force
multiplied by the lever, x. From this fact comes the x2 in the integrations. The
second terms, the forward-speed terms, are caused by the forward-speed potential
and will not be explained further here, see Salvesen et al. [43]. U is the speed of
the ship and ωe is the angular frequency of encounter.
For the box-like ship, again the coefficients a33 and b33 are functions of the os-
cillation frequency but independent of x, and the beam is B(x) = B. For simplified
estimates a characteristic frequency of the exciting wave can be used and, actu-
ally, for each sea state the resulting sea-keeping properties will be astonishingly
realistic.

161
11. SHIP IN REGULAR WAVES

Figure 11.12: Amplitude response function at the headings: 0, π/4 and π/2 rad.

Figure 11.13: Phase lag between heave motion and wave at the headings: 0, π/4
and π/2 rad.

162
 11.3. The Pitch Problem Using Strip Theory

For the box-like ship at zero speed thus

L3
A55 = a33 , (11.49a)
12
L3
B55 = b33 , (11.49b)
12
L3 L3
C55 = ρgV (zB − zG ) + ρgB ≈ ρgB . (11.49c)
12 12

11.3.2 Wave-excited moment in pitch

The pitch wave-excited moment on the box-like fixed ship in waves can be written
Z L/2 Z L/2
F5e = − pc (x, −T )B x dx − ẇc (x, −T ) a33 x dx
−L/2 −L/2
Z L/2
− wc (x, −T ) b33 x dx , (11.50)
−L/2

where T is the draught; pc (x, −T ) the undisturbed pressure at the box bottom
z = −T ; ẇc (x, −T ) the water acceleration at the bottom z = −T ; wc (x, −T ) the
water velocity at the bottom z = −T ; a33 the vertical two-dimensional added
mass; b33 the vertical two-dimensional radiation damping coefficient; B the beam
(breadth) of the ship section and L the length of the ship.
Again we will, for simplicity use complex notation. Also, because it is more
useful and not very complicated we will directly formulate the wave-excited mo-
ment for waves oblique to the ship in the following example. The potential and
derived properties for such waves in complex notation were given above in con-
nection with the heave problem Eq. (11.37). Substituting these expressions into
Eq. (11.50) gives
Z L/2

F5e = − ρ g a B − a ω 2 a33 − ı a ω b33 x e−kT eıkx cos(θ ) e−ı ω t dx , (11.51)
−L/2

or

Z L/2
−kT −ı ω t
2
F5e = −a ρ g B − ω a33 − ı ω b33 e e eıkx cos(θ ) x dx . (11.52)
−L/2

The integral can be solved separately [23]:

Z L/2     
ıkx cos(θ ) 2ı L L L
e x dx = sin k cos(θ ) − k cos(θ ) cos k cos(θ ) .
−L/2 (k cos(θ ))2 2 2 2
(11.53)

which approaches zero in the limit as θ → π/2, and thus confirms that in beam
regular waves the pitch is zero.

163
11. SHIP IN REGULAR WAVES

Figure 11.14: The amplitude of the pitch wave excited forces at zero speed as
a function of ship length divided by wave length in deep water,
L/λ .

Exercise 6.2
Look at Fig. 11.14. Why are the forces zero for meeting waves at 1.5,
2.5, 3.5 etc.?

Now set the pitch motion to η5c = η̂5c e−ı ω t , which gives the pitch angular
velocity and acceleration of the ship to

η̇5c = −ı ω η̂5c e−ı ω t ,

η̇5c = −ω 2 η̂5c e−ı ω t

Using Eq. (11.45) with the complex motion above, the equation of motion
finally gives



C55 − (I5 + A55 )ω 2 − ı ω B55 η̂5c e−ı ω t = −a ρgB − ω 2 a33 − ı ω b33 ×
    
−k T −ı ω t 2ı L L L
e e sin k cos(θ ) − k cos(θ ) cos k cos(θ ) ,
(k cos(θ ))2 2 2 2
(11.54)

which can be solved for the complex pitch motion amplitude in a closed-form

164
 11.4. The Roll Problem

expression
 
−kT ρgB − ω 2 a33 − ıωb33
η̂5c = − a e ×
C55 − (I5 + A55 )ω 2 − ı ω B55
    
2ı L L L
sin k cos(θ ) − k cos(θ ) cos k cos(θ ) . (11.55)
(k cos(θ ))2 2 2 2

Exercise 6.3
Inspecting Eq. (11.55) one can see that the amplitude response func-
tion |η̂5c |/a has one minimum around ρgB 2
L L

= ω a33 , minima for
sin(k cos(θ )L/2) = k cos(θ ) 2 cos k cos(θ ) 2 , and a maximum around
C55 = (I5 + A55 )ω 2 .

Please explain why.

The natural angular frequency is obtained by setting

p the driving moment and
damping to nil in Eq. (11.54), which gives ωN5 = C55 /(I5 + A55 ).
The amplitude response function and the corresponding phase lag as functions
of angular wave frequency are shown in Figs. 11.15 and 11.16.
As an example, in Fig. 11.17 the pitch motion and wave slope at the origin
of a ship as functions of time are shown for a swell with a period around 43 s.
For this very long wave the ship will follow the slope of the wave surface almost
exactly, but due to the definitions of slope and pitch angle, the time functions are
180 degrees out of phase.

11.4 T HE ROLL P ROBLEM

To solve the roll-motion problem for a two-dimensional ship, we can – as before –

split the hydrodynamic problem into two problems: the problem to assess the
moment on the ship caused by the ship rolling in calm water and the problem to
assess the moment acting on the fixed ship by the progressing waves.
As most ships are symmetric with respect to their longitudinal axes, the cou-
plings to the heave and pitch motions are negligible, while there are - usually
weak - couplings to the sway and yaw motions. Here only uncoupled roll motion
is treated.

11.4.1 Moments on the two-dimensional body rolling in calm water

Assume that the body is forced to oscillate in roll in calm water with the amplitude,
η4c , and the angular frequency, ω;

η4c = η̂4c e−ı ω t . (11.56)

165
11. SHIP IN REGULAR WAVES

Figure 11.15: Pitch amplitude response functions at the headings: 0, π/4 and
π/2 rad.

Figure 11.16: Phase lags between pitch motion and wave at the headings: 0,
π/4 and π/2 rad.

166
 11.4. The Roll Problem

Figure 11.17: Pitch motion and wave slope at the origin of a ship as functions
of time in a swell with the period around 43 s. Observe that the
ship follows the slope of the wave-surface exactly for this swell,
but due to the definitions of slope and pitch angle, the functions
are 180 deg out of phase.

The roll angular velocity is then

η̇4c = −ı ω η̂4c e−ı ω t . (11.57)

and the roll angular acceleration

η̈4c = −ω 2 η̂4c e−ı ω t . (11.58)

relative to the horizontal still-water surface.

The roll reaction moment acting on the ship from the water can be written

F4rc = −A44 η̈4c − B44 η̇4c −C44 η4c , (11.59)

where the coefficients A44 , B44 and C44 must be assessed.

The linear static restoring moment C44 η4 is due to the change of the shape of
the displacement and the vertical position of the static centre of buoyancy in rela-
tion to the centre of gravity of the ship. See the course in “stability and weight”.
The restoring moment can then be written in terms of the transverse metacentre
height, GM T , as

B3
C44 = ρgV (zB − zG ) + ρgL = ρgV GM T . (11.60)
12
The dynamic moments due to the roll angular acceleration and roll angular
velocity of the ship are - as for the other degrees of freedom - associated with
the forced oscillatory motion of the ambient water. The roll acceleration of the

167
11. SHIP IN REGULAR WAVES

Figure 11.18: Radiated 2D waves caused by roll motion.

ship is thus associated with a local, evanescent wave in which the water is forced
to shift back and forth between the bottom and the sides alternately to starboard
and portside. The effect is an “added moment of inertia”, which can be calculated
by help of two-dimensional potential theory in combination with strip theory or
directly from three-dimensional potential theory for short ships or floating objects
with complicated shapes like offshore drilling platforms. As, for a ship, the order
of magnitude of the added moment of inertia is only 10 to 20 % of the mass
moment of inertia of the ship itself, it can suffice here to use an approximate
value. Thus
A44 = 0.15 I4 . (11.61a)
For bodies floating in the water surface or positioned close to the water surface
this added moment is a function of the frequency of oscillation. Far from the free
surface the added moment of inertia is constant and depends only on the shape
of the body and its vicinity to other bodies or fixed boundaries. Also for floating
bodies, in the limit, as ω → ∞, no waves are formed, and the added moment
becomes independent of the frequency and is half that of the body mirrored in the
water surface and submerged deeply below the free surface.
To have an approximate value of the roll moment of inertia one can use the
fact that usually the radius of inertia in roll i4 is of the order of 0.4B, where B is
the beam of the ship, and thus

I4 = ρCB B L T i24 = ρ V i24 ≈ 0.16ρV B2 , (11.61b)

where CB = V /(BLT ) is the block coefficient and V the displacement.

11.4.2 Roll damping

As the heave motion causes waves that radiate out from the body, so does the roll
motion, and the moment, B44 η̇4 , needed to maintain this radiation is in phase with
the angular roll velocity. Only now the waves are anti-symmetric with respect to
the axis of the ship. See Fig. 11.18. Again these radiated waves transport energy
away from the body and therefore introduces hydrodynamic or radiation damping.
However, in roll motion the damping is not dominated by this radiation damping

168
 11.4. The Roll Problem

but depends to a substantial degree on turbulent skin friction, turbulence caused

by bilge keels and appendages. Therefore the total damping cannot be calculated
by potential theory only, but also other damping contributions must be taken into
account. Traditionally one depends on experiments with obvious problems of
scaling, but recently possibilities have been created to use turbulent, viscous flow
computations.
In summary the roll-damping moment is caused by

1. radiated waves

2. turbulent skin friction between the hull and the water

3. appendage or bilge-keel vortex shedding and

4. moorings for moored ships or moored offshore platforms.

To make it further complicated the contributions from items 2 to 4 are func-

tions on velocities squared so they depend on the amplitude of roll.
For bodies floating in the water surface or positioned close to the water surface
also the radiation roll damping is a function of the frequency of oscillation due to
the generation of waves. For floating bodies, in the limit, as ω → ∞, no waves can
be formed, and the radiation damping is nil as it is for deeply submerged bodies,
that cannot produce surface gravity waves. The radiation damping is actually nil
both for ω → 0 and ω → ∞. There is a maximum somewhere in between where the
body has a maximal ability to radiate energy or reciprocally absorb wave energy.
Jensen et al. [23] could not find simple closed-form solutions for the sectional
radiation damping, b44 , not even for rectangular shapes. Therefore they deter-
mined the coefficient by a 2-D potential method, the Frank close-fit method, for a
range of B/T ratios both for triangular sections and for rectangular sections. They
then fitted parametric curves, linear in B/T , through the results,
s  
b44 B B b( B )(ωe s/rad)−1.3
(ωe s/rad)d ( T ) ,
B
=a e T (11.62)
ρ Ac B2 2g T

where Ac is the cross-sectional area of the submerged part of the section. For
triangular cross-sections with 3 ≤ B/T ≤ 6 the functions a( · ), b( · ) and d( · )
became:
 
B B
a = 0.256 − 0.286 , (11.63a)
T T
 
B B
b = −0.11 − 2.55 , (11.63b)
T T
 
B B
d = 0.033 − 1.419 . (11.63c)
T T

169
11. SHIP IN REGULAR WAVES

Table 11.1: Added damping ratio ξ for a Panamax container vessel in different
headings. From [23].

Heading∗ (degrees) Added damping ratio ξ Comments

25 0 Response too low
45 3 OK
65 20 OK
205 20 Responses overpredicted
225 20 Responses overpredicted
245 20 Responses overpredicted
∗ 180 degrees corresponds to head sea.

For ships with fuller lines, e.g. container ships and tankers, the same procedure
was used for rectangular cross-sections with 1 ≤ B/T ≤ 3 to obtain:
 
B B
a = −3.94 + 13.69 , (11.64a)
T T
 
B B
b = −2.12 − 1.89 , (11.64b)
T T
 
B B
d = 1.16 − 7.97 . (11.64c)
T T

11.4.3 Added viscous roll damping

For assessing the total roll damping one must rely on model tests. One way is
then to add a fraction of critical damping to calibrate calculated heave amplitude
operators against model tests for a similar ship.
The critical damping in roll is, compare Chapter 3.1,
p
Bcrit
44 = 2 C44 (I44 + A44 ) . (11.65)
Jensen et al. [23] proposed to set the total damping to
Btot crit
44 = B44 + ξ B44 . (11.66)
For a Panamax container vessel progressing at a speed equivalent to Fn = 0.245,
Jensen made calculations agree with experiments with values on ξ according to
Table 11.1.

11.4.4 Wave-excited roll moment

The sectional wave-excited moment can approximately be expressed in terms of
the hydrodynamic damping, b44 (Newman (1978) as cited by [23]):
r
ρ g2
f4 = a b44 eı k x cos(θ ) e−ı ω t sin(θ ) . (11.67)
ω

170
 11.4. The Roll Problem

This moment is in phase with the cross-beam slope velocity of the wave because
here we have neglected the force in phase with the slope and with the slope accel-
eration, which are less important when the magnitude of the roll motion shall be
assessed.
Integrating along our box-like ship yields the moment
Z L/2 r 2
ρg
F4 = a b44 eı k x cos(θ ) e−ı ω t sin(θ ) dx
−L/2 ω
r  
ρ g2 2 sin(θ ) −ı ω t L
= b44 e sin k cos(θ ) . (11.68)
ω k cos(θ ) 2

In this equation ω can as an approximation be exchanged with the encounter fre-

quency, ωe , but maintaining k = f (ω) thus making the expression approximately
valid for a ship at speed.
Because θ → π/2 gives sin(k cos(θ )L/2) → k cos(θ )L/2 Eq. (11.68) is also
valid in the limit for beam sea as
r
ρ g2 π
F4 = b44 L sin(θ ) e−ı ω t , for θ close to ± . (11.69)
ω 2
See Fig. 11.19.

Figure 11.19: The roll-excited wave-moment amplitude ratio F̂4 /a as a function

of angular wave frequency for zero speed forward, U = 0 m/s,
and some chosen headings.

171
11. SHIP IN REGULAR WAVES

As the wave excited roll moment should be added to the reaction moment from
the water due to the roll of the ship in still water, Eq. (11.59), and be balanced by
the roll moment of inertia the following equation of roll motion is yielded:
(I4 + A44 ) η̈4c + B44 η̇4c +C44 η4c = F4 . (11.70)

11.4.5 Roll motion

The ship must roll with the encounter frequency, ωe . Using complex notation and
solving for the amplitude gives the motion amplitude as
F4
η̂e = , (11.71)
C44 − ωe2 (A44 + I4 ) − ı ω Btot
44 (ωe , ξ )
which is a function of encountered angular wave frequency, ωe , ship’s heading,θ ,
and ship speed, U, defined as positive in the propagation direction of the waves.
In Figs. 11.20 to 11.23, the roll response amplitude operator η̂4 /a is illustrated for
various combinations of these variables.
The natural angular frequency is obtained by setting
p the driving moment and
damping to nil in Eq. (11.71), which gives ωN4 = C44 / (I4 + A44 ).

11.5 F ORWARD -S PEED E FFECT

Although the wave frequency variation in an irregular sea state influence the ship
motion at zero speed, the effect of the frequency change due to ship moving
through the waves is much more pronounced. A ship progressing through a fol-
lowing wave with the speed, U, at the heading, θ , in relation to the direction of
wave propagation, will be passed by a wave crest every encounter period
λ
Te = . (11.72)
c −U cos(θ )
Negative encounter period means that the ship moves faster than the wave. This
can only happen for waves abaft the beam, −90◦ < θ < 90◦ . In head waves,
meeting (encountered) waves, 90◦ < θ < 270◦ , the encountered period is always
positive and shorter than the wave frequency.
The angular frequency of encounter is thus
2π 2π
ωe = = (c −U cos(θ )) , (11.73)
Te λ
or
λ
= c −U cos(θ ) ,
Te
For deep water Eq. (11.73) can be written
ω2U
ωe = ω − cos(θ ) . (11.74)
g
See Fig. 11.24.

172
 11.5. Forward-Speed Effect

Figure 11.20: The dimensionless roll response slope-amplitude operator

η̂4 /(ka) (deg/deg) as a function of angular wave frequency for
beam sea, θ = 90◦ , zero speed forward, U = 0 m/s and damping
ratios ξ = 0, 0.05 and 0.2.

Figure 11.21: A blow up of Fig. 11.20 around the roll resonance frequency.

173
11. SHIP IN REGULAR WAVES

Figure 11.22: The dimensionless roll response slope-amplitude operator

η̂4 /(ka) (deg/deg) as a function of angular wave frequency for
zero speed, damping ratio, ξ = 0.2, and different headings. Ob-
serve that for 0 deg heading the wave slope projected on the ship
is zero and thus is the roll amplitude.

Figure 11.23: The dimensionless roll response slope-amplitude operator

η̂4 /(ka) (deg/deg) as a function of angular wave frequency for
damping ratio, ξ = 0.2, heading 20 deg, and different speeds for-
ward.
174
 11.6. Resonance Frequencies

(a) (b)

Figure 11.24: Encounter wave frequency. (a) ωe as a function of wave frequency

at zero heading (following waves) and forward speed 4 m/s and
(b) ωe as a function of forward speed at 180 deg heading (meeting
waves) and angular wave frequency 0.5 rad/s.

11.6 R ESONANCE F REQUENCIES

The heave, pitch and roll motions are resonant motions, while the horizontal mo-
tions sway, surge and yaw are non-resonant because there are no stiffnesses in the
latter modes of motion for unmoored ships.

The eigenfrequencies, undamped resonance frequencies or natural frequencies

can be written

s s
C33 ρgAw
in heave ωN3 = = ,
ρV + A33 ρV + A33
r
C44
in roll ωN4 = ,
I4 + A44
s r
C55 ρgIw5
in pitch ωN5 = ≈ .
I5 + A55 I5 + A55

The last approximation can be used because the hydrostatic stiffness dominates in
pitch. Aw is the water-plane area and Iw5 = BL3 /12 is the area moment in pitch of
the water-plane area.

175
11. SHIP IN REGULAR WAVES

For the box-like ship one can approximately write

s s
C33 ρgB
in heave ωN3 = = ,
ρV + A33 ρBT + a33
r s
C44 ρgLB (B2 + T (zB − zG ))
in roll ωN4 = = ,
I4 + A44 I4 + a44 L
s r s
ρg 3
C55 ρgIw5 12 BL
in pitch ωN5 = ≈ = .
I5 + A55 I5 + A55 I5 + a55 L

Typical resonance periods are in heave 10 s, in roll 8 – 12 s and in pitch 10 –

20 s. For anchored ships the resonance periods in surge may be > 200 s, in sway
> 100 s and in yaw > 100 s and are highly dependent on the mooring system.

11.7 D ERIVED R ESPONSES

Hitherto we have assessed the global motion of the ship in six degrees of freedom
referred to the origin of the chosen co-ordinate system. For applications we have
to be able to describe the translational motion in three degrees of freedom at any
point of the ship. The aim can be to investigate the freeboard, the risk for pro-
peller emergence out of the water, the risk for slamming and also accelerations
for assessing cargo fastenings, comfort and seasickness. These local motions are
called derived responses.

11.7.1 The motion at a point

The motion of a point rT = [x , y , z] on a body can for small rotations ΩT =
[η4 , η5 , η6 ] (less than 10 degrees or 0.2 rad) be written:
 
η1
 
s =  η2  + Ω × r , (11.75)
η3

which explicitly in the chosen co-ordinate system is

 
η1 + zη5 − yη6
 
s =  η2 − zη4 + xη6  . (11.76)
η3 + yη4 − xη5

As we in the strip theory cannot predict the surge motion with any degree of
success the first row of Eq. 11.76 describing the horizontal motion in the surge
direction is of no interest in this chapter. Furthermore as we in this treatise have

176
 11.7. Derived Responses

not calculated the yaw motion, although it is fully feasible, we cannot use the
second row for the crossbeam horizontal motion. The most important motion is,
however, the vertical motion described by the third row. Thus

s3 = η3 + yη4 − xη5 . (11.77)

Here we have retained all the vertical motions heave, pitch and roll [η3 , η4 , η5 ]
contributing to the vertical motion of a point. It should be warned that as the
used method for calculating the roll motion is not phase correct, it should not be
included in Eq. (11.77) if it lessens the vertical motion. For statistical estimates
in irregular waves see Chapter 9.
The vertical velocity and acceleration of the same point are likewise

ṡ3 = η̇3 + yη̇4 − xη̇5 . (11.78)

and
s̈3 = η̈3 + yη̈4 − xη̈5 . (11.79)
The vertical motions heave, pitch and roll [η3 , η4 , η5 ] are most conveniently in-
troduced into Eqs. (11.77) to (11.79) in their complex time-domain form. The
real motion is then assessed by taking the real part of the derived motion s3 .
An example of a calculation of the vertical bow motion of the example ship
is shown in Fig. 11.25 with y = 0 m, or no roll motion. The ship is a box 100 m
long, 20 m wide and with 10 m draught. In Figs. 11.26 and 11.27 the response
amplitude operator and phase lag is shown.

11.7.2 Acceleration
It is important to assess the accelerations of a ship because it is difficult to work
if the accelerations are too large, say g/3; accelerations at certain frequencies
also causes seasickness; cargo may get loose or fastenings must be attached and
designed. In the passenger and cruising trade the comfort is important, in the off-
shore industry the focus is to avoid downtime of operations like drilling or pump-
ing oil and gas, while in cargo traffic the safety of the goods is most important but
also speed reductions should be avoided. In Figs. 11.28 to 11.30 the acceleration,
Eq. (11.79), at the bow of the example ship is shown.

11.7.3 Green water

If the wave elevation comes above the instantaneous position of the railing or
above the freeboard at any point of the ship this will result in water on deck, so
called green water. This can be tested by the following expression for regular
plane waves

sFB (t) = η3 (t) + yη4 (t) − xη5 (t) + zFB − ζ (x cos(θ ) , y sin(θ ) ,t) > 0 , (11.80)

177
11. SHIP IN REGULAR WAVES

Figure 11.25: Vertical bow motion and wave elevation near pitch and heave res-
onance.

Figure 11.26: The response amplitude operator for the vertical bow motion.

Figure 11.27: The phase lag of the vertical bow motion.

178
 11.7. Derived Responses

Figure 11.28: Vertical bow acceleration and wave elevation near pitch and heave
resonance.

Figure 11.29: The response amplitude operator for the vertical bow accelera-
tion.

where sFB is the instantaneous freeboard; zFB the static freeboard at station (x , y)
and actual trim and ζ (x cos(θ ) , y sin(θ ) ,t) the instantaneous wave elevation at
station (x , y). See Figs. 11.31 and 11.32.

11.7.4 Propeller emergence and risk for slamming

The risk for the propeller to emerge out of the water and the risk that the bottom
will rise above the water and experience slamming at re-entry must be assessed.

179
11. SHIP IN REGULAR WAVES

Figure 11.30: The phase lag of the vertical bow acceleration.

This can be tested by a similar expression as was used for assessing the freeboard.

sBE (t) = −η3 (t) − yη4 (t) + xη5 (t) + T + ζ (x cos(θ ) , y sin(θ ) ,t) > 0 , (11.81)

where sBE is the instantaneous submergence of the ship’s bottom or propeller; T

the draught at station (x , y) and actual trim and ζ (x cos(θ ) , y sin(θ ) ,t) the instan-
taneous wave elevation at station (x , y). See Figs. 11.33 and 11.34.

11.7.5 Slamming
Slamming will appear when the ship’s bottom has risen out of the wave and hits
back at re-entry. The slamming pressure depends on the relative velocity squared,
between the ships bottom and the water surface, the angle between them, their
irregularity and content of air bubbles. The slamming is complicated and it is
referred to e.g. Faltinsen [12] for a deeper description. The slamming pressure can
very roughly be approximated by ps = C ρ Urel 2 , where C is a constant. Then the

starting point of assessment is to calculate the relative velocity. If we approximate

this by the relative vertical velocity we can just take the time derivative of Eq.
(11.81) with y = 0:

∂
Urel (t) = (−η3 (t) − yη4 (t) + xη5 (t) + T + ζ (x cos(θ ) , y sin(θ ) ,t))
∂t
= −η̇3 (t) + xη̇5 (t) + ζ˙ (x cos(θ ) , y sin(θ ) ,t) . (11.82)

180
 11.7. Derived Responses

Figure 11.31: Elevation of railing aft and wave elevation at the same place.

Figure 11.32: Instantaneous freeboard at bow and aft.

181
11. SHIP IN REGULAR WAVES

Figure 11.33: The elevation of the propeller centre and the wave surface as a
function of time. In this case the propeller is submerged all the
time.

Figure 11.34: The submergence of the propeller centre as a function of time.

The difference between the graphs in Fig. 11.33.

182
 11.8. Coupled Linear Pitch and Heave Motion at Forward Speed

This velocity should be evaluated for all moments when Eq. (11.81) is zero

−η3 (t) − yη4 (t) + xη5 (t) + T + ζ (x cos(θ ) , y sin(θ ) ,t) = 0 , (11.83)

and Urel is positive. See further for statistical simplifictions in irregular waves in
Paragraph 9.6.

11.8 C OUPLED L INEAR P ITCH AND H EAVE M OTION AT F ORWARD S PEED

Assume that, in calm water, the ship is forced to oscillate and pitch with the am-
plitudes, η̂3 and η̂5 , at the angular frequency, ω. The vertical reaction force and
pitch reaction moment acting on the body from the water can be written, after
integrating the two-dimensional reaction forces along the ship,

F3 = −A33 η̈3 − B33 η̇3 −C33 η3 − A35 η̈5 − B35 η̇5 −C35 η5 , (11.84)

and

F5 = −A53 η̈3 − B53 η̇3 −C53 η3 − A55 η̈5 − B55 η̇5 −C55 η5 , (11.85)

or in matrix form
F r = −A η̈ − B η̇ − C η , (11.86)

where A, B and C are 2 × 2 matrices and η a 2D vector. To assess the coupled

motion due to fore-aft asymmetry of a real ship now the coupling coefficients
between heave and pitch are needed:
Z L/2
U
A35 = − x a33 (x) dx +
B33 , (11.87)
−L/2 ωe2
Z L/2
U
A53 = − x a33 (x) dx − 2 B33 , (11.88)
−L/2 ωe
Z L/2
B35 = − x b33 (x) dx −U A33 , (11.89)
−L/2
Z L/2
B53 = − x b33 (x) dx +U A33 , (11.90)
−L/2

and Z
C35 = C53 = ρg x ds . (11.91)
AW P

Additionally, we have the diagonal elements as defined in Eqs. (11.32) –

183
11. SHIP IN REGULAR WAVES

(11.34) and (11.46) – (11.48):

Z L/2
A33 = a33 (x)dx = ∑ a33n ∆xn ,
−L/2 n
Z L/2
B33 = b33 (x)dx = ∑ b33n ∆xn ,
−L/2 n
Z L/2
C33 = c33 (x)dx = ∑ c33n ∆xn = ∑ ρ g Bn ∆xn .
−L/2 n n
Z L/2
U2
A55 = a33 (x) x2 dx − A33 ,
−L/2 ωe2
Z L/2
U2
B55 = b33 (x) dx + 2 B33 ,
−L/2 ωe
Z L/2
C55 = ρgV (zB − zG ) + ρg B(x) dx = ρgV GM L .
−L/2

11.8.1 The coupling between heave and pitch

At zero speed A35 = A53 , B35 = B53 , and C35 = C53 and the equation system is
symmetric. For the chosen co-ordinate system and the box-like ship used in the
examples with its centre of gravity amidships they are zero. Further, for a mer-
chant ship the coupling between pitch and surge can be ignored, but for an off-
shore platform it must be taken into account, while the coupling between heave
and pitch may be ignored instead.
The 2 × 2 mass matrix, M, is diagonal i.e. M35 = M53 = 0, because the origin
of the coordinate system is placed above the centre of gravity.
The driving force and moment is somewhat modified due to the forward speed
and we can include the dependence on station x of the ship but specialise for
following sea (for meeting sea change −ı ω t to +ı ω t):
Z L/2
−ı ω t
F3e = a e eı k x e−kT (x) [c33 (x) − ω (ωe a33 (x) − ı b33 (x))] dx , (11.92)
−L/2
Z L/2 
−ı ω t ı k x −kT (x)
F5e = −a e e e x [c33 (x) − ω (ωe a33 (x) − ı b33 (x))]
−L/2

U
− ω (ωe a33 (x) − ı b33 (x)) dx . (11.93)
ı ωe
Compare Eqs. (11.38) and (11.51).
With F = [F3e , F5e ]T then the equation of motion is
M η̈ = F r + F = −A η̈ − B η̇ − C η + F , (11.94)
or
(M + A) η̈ + B η̇ + C η = F , (11.95)
which is recognised as Equation (3.41), and can be solved in the time or frequency
domain by methods advised in Section 3.4 of Chapter 3.

184
 11.9. Coupled Pitch and Heave Motion at Zero Speed Including Non-Linear Viscous
Damping

Figure 11.35: Definition of the chosen coordinate system for a ship. The z-axis
is drawn vertically through the centre of buoyancy, CB, and the
centre of gravity, CG. (From Faltinsen).

11.9 C OUPLED P ITCH AND H EAVE M OTION AT Z ERO S PEED I NCLUD -

ING N ON -L INEAR V ISCOUS DAMPING

For the box-like ship used in the examples we will investigate the effect of viscous,
drag damping, which is quadratic and therefore hinder us from solution in the
frequency domain, the use of complex numbers and linear superposition. We will
therefore formulate the problem in the time domain, using the notations we used
before and the coordinate system presented in Fig. 11.35.

11.9.1 The wave

Elevation or vertical displacement of a sheet of water particles at the level z = −T :
ζ (t , x , −T ) = a e−kT cos(k x − ω t) . (11.96)
Vertical particle velocity
ζ˙ (t , x , −T ) = a ω e−kT sin(k x − ω t) = w . (11.97)
Vertical particle acceleration
ζ¨ (t , x , −T ) = −a ω 2 e−kT cos(k x − ω t) = ẇ . (11.98)

11.9.2 Heave
Displacement
η3 (t) = η̂3 cos(−ω t − ε3 ) . (11.99)
Velocity
η̇3 (t) = η̂3 ω sin(−ω t − ε3 ) . (11.100)
Acceleration
η̈3 (t) = −η̂3 ω 2 cos(−ω t − ε3 ) . (11.101)

185
11. SHIP IN REGULAR WAVES

11.9.3 Pitch
Displacement
η5 (t) = η̂5 cos(−ω t − ε5 ) . (11.102)
Angular velocity
η̇5 (t) = η̂5 ω sin(−ω t − ε5 ) . (11.103)
Acceleration
η̈5 (t) = −η̂5 ω 2 cos(−ω t − ε5 ) . (11.104)

11.9.4 Equations of motion

Fore/aft symmetrical box

Heave
Z L/2   Z L/2  
m η̈3 = ¨
a33 ζ − η̈3 + xη̈5 dx + b33 ζ˙ − η̇3 + xη̇5 dx
−L/2 −L/2
| {z } | {z }
added mass force linear damping force
Z L/2  
1
+ ρCD B ζ˙ − η̇3 + xη̇5 ζ˙ − η̇3 + xη̇5 dx
−L/2 2
| {z }
viscous drag force
Z L/2 Z L/2
+ ρgBζ dx + ρgB (−η3 + xη5 ) dx . (11.105)
−L/2 −L/2
| {z } | {z }
Froude-Krylov force buoyancy force
Assemble exciting forces on the right hand side and linear reaction terms on the
left hand side yields
Z L/2  
(m + A33 ) η̈3 + B33 η̇3 +C33 η3 = ¨ ˙
a33 ζ + b33 ζ + ρgBζ dx
−L/2
Z L/2  
1
+ ρCD B ζ˙ − η̇3 + xη̇5 ζ˙ − η̇3 + xη̇5 dx .
−L/2 2
(11.106)
We have then used that
Z L/2
A33 = a33 dx , (11.107a)
−L/2
Z L/2
B33 = b33 dx , (11.107b)
−L/2
Z L/2
C33 = ρgB dx , (11.107c)
−L/2

186
 11.9. Coupled Pitch and Heave Motion at Zero Speed Including Non-Linear Viscous
Damping

and e.g. that

Z L/2 Z L/2
x η̇5 dx = η̇5 x dx = 0 , (11.108)
−L/2 −L/2

i.e. all linear coupling terms vanish.

We are left with the last non-linear coupling term, which cannot be separated
into dependent and non-dependent terms because of the product with its modulus,
and therefore must be evaluated in the time domain, which makes the use of the
convenient complex variables impossible. Compare e.g. ζ |ζ | and ζc |ζc |:

ζc = a eı(k x−ω t) ,
ζ = Re (ζc ) = a cos(k x − ω t) ,
ζc |ζc | = a2 eı (k x−ω t) = a2 (cos(k x − ω t) + ı sin(k x − ω t)) ,
Re (ζc |ζc |) = a2 cos(k x − ω t) ,

which is not the same as

ζ |ζ | = a2 | cos(k x − ω t)| cos(k x − ω t) .

Pitch

Z L/2   Z L/2  
I5 η̈5 = − ¨
a33 ζ − η̈3 + xη̈5 x dx − b33 ζ¨ − η̈3 + xη̈5 x dx
−L/2 −L/2
| {z } | {z }
added mass moment linear damping moment
Z L/2  
1 ˙ ˙
− ρCD B ζ − η̇3 + xη̇5 ζ − η̇3 + xη̇5 x dx
−L/2 2
| {z }
viscous drag moment
Z L/2 Z L/2
− ρgBζ x dx − ρgB (−η3 + xη5 ) x dx . (11.109)
−L/2 −L/2
| {z } | {z }
Froude-Krylov moment buoyancy moment

Assemble exciting moments on the right hand side and linear reaction terms on
the left hand side.
Z L/2  
(I5 + A55 ) η̈5 + B55 η̇5 +C55 η5 = − a33 ζ¨ + b33 ζ˙ + ρgBζ x dx
−L/2
Z L/2  
1
− ρCD B ζ˙ − η̇3 + xη̇5 ˙
ζ − η̇3 + xη̇5 x dx .
−L/2 2
(11.110)

187
11. SHIP IN REGULAR WAVES

We have then used that

Z L/2
A55 = a33 x2 dx , (11.111a)
−L/2
Z L/2
B55 = b55 x2 dx , (11.111b)
−L/2
Z L/2 Z L/2
2
C55 = ρgB x dx − (zG − zB ) ρ gV ≈ ρ g B x2 dx , (11.111c)
−L/2 −L/2

and e.g. that

Z L/2 Z L/2
x η̇3 dx = η̇3 x dx = 0 , (11.112)
−L/2 −L/2

i.e. again all linear coupling terms vanish and we are left with the last non-linear
coupling term.

11.10 E QUIVALENT L INEARISED D RAG DAMPING

The non-linear coupled drag term constitutes a problem, when making assessment
of sea-keeping properties in the frequency domain.
In roll, the linear radiation damping is very small and the viscous non-linear
drag damping dominates and must therefore be assessed to get realistic motion.
Jensen increased the linear damping coefficient, Equation (6.66), by comparing
the calculated motion with seakeeping model tests. In this section we will show
that such an equivalent linear drag-damping coefficient depends on the amplitude
of motion.
Neglecting the coupling between roll and sway we can symbolically write the
drag damping moment in beam regular sea as
 
˙ ˙
FD4 = K ζy − η̇4 ζy − η̇4 , (11.113)

where K can be set to ρ CD B2 L/2 and ζ˙y = ∂ 2 ζ /∂t∂ y is the angular velocity of
the wave slope in the y direction i.e. in the starboard-portside direction.
When the non-linear roll damping is important usually ζ˙y  η̇4 and then

FD4 ≈ K|η̇4 |η̇4 , (11.114)

which is simpler but still non-linear.

To assess the equivalent linear coefficient we can compare the dissipated en-
ergy over one wave cycle with an equivalent linear expression, assuming harmonic
motion with the roll angular velocity amplitude η̂˙ 4 ,

η̇4 = η̂˙ 4 sin(ω t) . (11.115)

188
 11.10. Equivalent Linearised Drag Damping

The dissipated energy is

Z T Z T
K|η̇4 | η̇42 dt = Be44 η̇42 dt . (11.116)
0 0

The L.H.S. becomes:

Z T /2 Z T
 Z T /2

4
K 3
η̇4 dt + η̇43 dt = 2K η̇43 dt = K η̂˙ 43 , (11.117)
0 T /2 0 3

and the R.H.S.: Z T

π
Be44 η̇42 dt = Be44 η̂˙ 42 . (11.118)
0 2
Finally, as L.H.S. = R.H.S. we obtain:
4 ˙3 π
K η̂4 = Be44 η̂˙ 42 , (11.119)
3 2
or
8 ˙
Be44 =K η̂4 . (11.120)
3π
That is, the equivalent damping coefficient, Be44 , depends on the amplitude of the
roll motion.
In a harmonic wave motion e.g. roll in beam swell iteration on the roll am-
plitude must be performed, guessing the roll amplitude, calculating it from the
equation of motion and then repeat with this new value until it converges. In an
irregular sea a guess of the significant roll amplitude is made, then this is calcu-
lated from the equation of motion and then fed as input for consecutive rounds of
iteration.

189
 12
SHIP IN IRREGULAR WAVES

In Chapter 11 we studied the forces and motions of a ship in a regular sinusoidal

wave, in Chapter 4 the modelling of wind waves without and with directional
spread and in Chapter 5 the statistics of waves and sea states. For many years
the seakeeping performance was only assessed for regular design waves. But, as
was first proposed by St Denis and Pierson [46], the ship can be looked upon as a
mechanical filter, which filters the irregular sea into a ship motion, very much like
the electronic filter in a radio receiver filters radio waves, see Fig. 12.1.

12.1 T HE H EAVE M OTION IN I RREGULAR WAVES

12.1.1 Heave transfer function

In Chapter 6 the heave motion at zero speed in a plane regular wave with the
amplitude, a, and the propagation direction, θ , was deducted. For a box-like ship

Figure 12.1: The ship as a mechanical filter.

191
12. SHIP IN IRREGULAR WAVES

the wave excited heave motion was then represented by a complex heave motion:

ρg B − ω 2 a33 − ı ω b33
η3c = η̂3c eı ω t =a e−k T ×
ρgBL − ω 2 (ρT BL + a33 L) − ı ω b33 L
sin (k(cos(θ )L/2) −ı ω t
2 e . (12.1)
k cos(θ )

Here the modulus |η̂3c | of the complex amplitude represents the heave ampli-
tude, the argument, arg (η̂3c ), the phase lag to the wave motion and the real part,
Re (η3c ), the real heave motion.
By dividing the complex amplitude by the wave amplitude, a, we get a com-
plex transfer function between the wave motion and the heave motion as

η̂3c ρg B − ω 2 a33 − ı ω b33

T3c (ω , θ ) = =e−k T ×
a ρgBL − ω 2 (ρT BL + a33 L) − ı ω b33 L
sin (k(cos(θ )L/2)
2 . (12.2)
k cos(θ )

Consider the complex wave function

ζc (ω ,t , ε) = a eı(−ω t+ε) = a (cos(−ω t + ε) + ı sin(−ω t + ε)) . (12.3)

Now forming the product of the transfer function (12.2) and the wave function
(12.3) at x = 0 m, we get the complex heave motion in the time domain as

η3c (ω , θ ,t , ε) = T3c (ω ,t)ζc (ω ,t , ε) . (12.4)

The wave function (12.3) is the same as Equation (5.13), but complemented with
a phase angle, ε.
The transfer function (12.2) transfers both the amplitude, |T3c (ω , θ )|, and the
phase lag, arg(T3c (ω , θ )), to the wave motion. The real heave motion can thus be
written

Re (η3c (ω , θ ,t , ε)) = Re (T3c (ω , θ )ζc (ω ,t , ε))

= a |T3c (ω , θ )| cos (−ω t + ε − arg(T3c (ω , θ ))) . (12.5)

The amplification factor, |T3c (ω ,t)|, or frequency response function is the heave
amplitude divided by the wave amplitude (m/m) and is often called the heave-
response amplitude operator or RAO1 and has it equivalents in other degrees of
freedom as well as for derived responses.
1 In some literature the square of |T (ω , θ )| is called RAO because the square is used in the
3c
multiplication by the wave spectrum to form the motion spectrum.

192
 12.1. The Heave Motion in Irregular Waves

12.1.2 Heave motion in the time domain

An irregular plane with the direction of propagation, θ , in relation to a ship head-
ing along the x-axis projected on the ship, can be represented by a complex sum
written as
∞
ζc (t) = ∑ ai eı (ki x cos(θ )−ωi t+εi ) , (12.6a)
i=1
where εi are the random phase angles, ai the amplitudes, ki the wave numbers and
ωi the angular frequencies of the component waves.
In the origin of the co-ordinate system x = 0 m, so the elevation there is
∞
ζc (t) = ∑ ai eı (−ωi t+εi ) . (12.6b)
i=1

From Eqs. ( 12.4) and (12.6) now follows by superposition that the heave motion
in the time domain can be written
∞
η3c (t) = ∑ ai T3c (ωi , θ ) eı(−ω t+εi ) . (12.7)
i=1

This time-domain heave is shown in Fig. 12.2 in three wave directions for a
box-like ship with dimensions length × beam × draught = L × B × T = 100 ×
20 × 10 m3 . The above simulation was done in the frequency domain starting out
with a wave spectrum, multiplying with a transfer function to produce a response
spectrum and from this response spectrum simulating a time trace of response by
an inverse fast-Fourier transformation (IFFT). By preserving the random phase
angles, εi , of the component waves and “adding” the phase lags of the response
components by the complex transform multiplication the time trace of response
shows the correct time response – to first order that is.

12.1.3 Forming the heave motion spectrum

The discrete heave spectrum can be formed from the time domain simulation by
1
S3i (θ ) = |ai T3c (ωi , θ )|2 . (12.8)
2
The continuous heave spectrum is formed analogously by the simple operation

S3 (ω , θ ) = |T3c (ω , θ )|2 S(ω) . (12.9)

In Fig. 12.3 the heave spectrum for following (or meeting) waves in two dif-
ferent sea states from Table 5.8 are shown using the ISSC wind-sea spectrum Eq.
(4.25). It is seen how the sea state with shorter mean period does not excite the
ship much within the frequency range where the transfer function is appreciable.
In Fig. 12.4 the response spectra for three headings are shown. There it is clearly
seen that beam sea is unfavourable for the heave motion.

193
12. SHIP IN IRREGULAR WAVES

(a)

(b)

(c)

Figure 12.2: Heave motion in 300 seconds of simulated plane waves with Hs = 2
m and T02 = 5 s using an ISSCb spectrum: (a) following waves, (b)
quartering waves and (c) beam waves.

194
 12.1. The Heave Motion in Irregular Waves

Figure 12.3: Heave-response spectra of the box-like ship in the same heading
(θ = 0◦ ) at two different sea states.

Figure 12.4: Heave-response spectra of the box-like ship in three headings at

two different sea states.

195
12. SHIP IN IRREGULAR WAVES

Table 12.1: Some characteristics of the two wave spectra and their heave re-
sponses.

Hs (m) T02 (s) θ (◦ ) 4σr (m) T02r (s) 4σr /Hs

2 5 0 0.62 9.4 0.32
45 0.66 9.1 0.33
90 1.20 8.8 0.60

3 8 0 1.94 11.2 0.64

45 2.76 10.5 0.92
90 3.96 10.0 1.32

The characteristics of the responses as zero-crossing period, crest period, sig-

nificant response and standard deviation can now be calculated from the moments
of the response spectra, because the responses or derived responses are also - like
the wave motion - normal-distributed Gaussian processes as they were produced
by a linear filter or transfer function. The most interesting characteristics are the
mean zero-upcrossing period evaluated as T02 and the standard deviation, σ . The
significant double amplitude Hs = 4. See Chapter 4 . Also the short term statistics
and maxima or peaks can be assessed as was outlined for the waves in paragraphs
5.3 and 5.4. Long-term statistics can be assessed by operating the RAOs on the
wave statistics of Table 5.8 for example. A choice of type spectrum must of course
be made.
In Table 12.1 one can see that the zero-crossing periods of the responses are
longer than the zero-crossing periods of both exciting sea states; that the signifi-
cant double-amplitude motions are smaller than the significant wave heights in all
headings except in beam sea. In the larger beam sea state the heave response of
the ship is amplified by a factor 1.32 (32 % larger) compared to the heave motion
of the sea. The natural frequency of the box like ship heaving in calm water is 9
s. What is the deep-water wave length corresponding to T02 ?

12.2 T HE P ITCH R ESPONSE

The pitch response can be treated exactly as the heave response with the complex
transfer function
η̂5c (ω , θ )
T5c (ω , θ ) = , (12.10)
a
which has the unit rad/m. The complex pitch amplitude η̂5c (ω , θ ) is given by
Equation (11.55). One can also choose to form a wave-slope spectrum from the
wave amplitude spectrum first and use a dimensionless transfer function
η̂5c (ω , θ )
T5c,slope (ω , θ ) = , (12.11)
k(ω)a

196
 12.3. The Response of Vertical Motion at Station x

This is, however, unnecessarily complicated.

Thus the continuous pitch spectrum is simplest formed by the operation
S5 (ω , θ ) = |T5c (ω , θ )|2 S(ω) . (12.12)
Some results of pitch-response spectra for the same two sea states as for the
heave response are shown in Figs. 12.5 and 12.6.
Time domain pitch can be simulated as was done with heave. If the same
arbitrary phase angles are used for the wave, heave and pitch the pitch, heave and
wave will be contemporary.

12.3 T HE R ESPONSE OF V ERTICAL M OTION AT S TATION x

To be able to calculate “green water”, propeller or bottom emergence, slamming

and acceleration we need the response of vertical motion at any station x along the
ship. The 3D motion, velocity and acceleration at a point (x, y) are described by
Equations (11.77) to (11.79):
s3 = η3 + yη4 − xη5 ,
ṡ3 = η̇3 + yη̇4 − xη̇5 ,
s̈3 = η̈3 + yη̈4 − xη̈5 .
Neglecting roll motion, which is difficult to assess and small in head sea (θ =
π), the expressions for the vertical motion, velocity and acceleration at a station
(x) along the ship are reduced to the following equations in complex form:
s3c = η3c − xη5c , (12.13)
ṡ3c = η̇3c − xη̇5c , (12.14)
s̈3c = η̈3c − xη̈5c . (12.15)
The vertical motion, velocity and accelerations responses can be treated as the
heave and pitch responses with the complex transfer functions
η̂3c (ω , θ ) η̂5c (ω , θ )
Ts3c (ω , θ , x) = −x
a a
= T3c (ω , θ ) − xT5c (ω , θ ) , (12.16)
 
η̂3c (ω , θ ) η̂5c (ω , θ )
Ts3velc (ω , θ , x) = −ı ω −x
a a
= −ı ω (T3c (ω , θ ) − xT5c (ω , θ )) , (12.17)
 
η̂3c (ω , θ ) η̂5c (ω , θ )
Ts3accc (ω , θ , x) = −ω 2 −x
a a
= −ω 2 (T3c (ω , θ ) − xT5c (ω , θ )) , (12.18)
which have the units m/m, m/(sm) and m/(s2 m), respectively.
Examples of vertical motion spectra, velocity spectra and acceleration spectra
formed by transformations equivalent to Eq. (12.12) are given in Figs. 12.7 to 9.9
below.

197
12. SHIP IN IRREGULAR WAVES

Figure 12.5: Pitch-response spectra of the box-like ship in the same heading
(θ = 0◦ ) at two different sea states.

Figure 12.6: Pitch-response spectra of the box-like ship in three headings at two
different sea states.

198
 12.3. The Response of Vertical Motion at Station x

Figure 12.7: Vertical motion response of the box-like ship in three stations.

Figure 12.8: Vertical velocity response of the box-like ship at bow in following
waves.

Figure 12.9: Vertical acceleration response of the box-like ship at bow in fol-
lowing waves.

199
 13
DERIVED RESPONSES

13.1 T HE M OTION OF A P OINT ON A B ODY

The motion of a point rT = [x , y , z] on a body can for small rotations ΩT =

[η4 , η5 , η6 ] (less than 10 degrees or 0.2 rad) be written:

 
η1
 
s =  η2  + Ω × r , (13.1)
η3

which explicitly in the chosen co-ordinate system is

 
η1 + zη5 − yη6
 
s =  η2 − zη4 + xη6  . (13.2)
η3 + yη4 − xη5

The velocities and accelerations of the same point is likewise

   
η̇1 η̈1
   
ṡ =  η̇2  + Ω̇ × r , s̈ =  η̈2  + Ω̈ × r . (13.3)
η̇3 η̈3

For large angles of rotation these simple expressions are not valid but more
complicated expressions must be made like those in the manoeuvring compendium.

201
13. DERIVED RESPONSES

Exercise 2.1
Assume that a ship is moving in a closed elliptical orbit in the x − z plane
without pitching, i.e. η2 = η4 = η5 = η6 = 0. The orbit is described by
" # " #
η1 η̂1 sin(ωt)
η= = .
η3 ηˆ3 cos(ωt)

The constant buoyancy force is balanced by the weight of the body. The
sum of all varying forces acting on the platform must then be given by the
following simple equation of motion:
" # " #
m 0 d 2 η1
F= .
0 m dt 2 η3

Assume that the displacement of the ship is 2 500 tonnes (1 tonne =

1000 kg), the heave amplitude 2.5 m, the surge amplitude 5 m and the
wave period 17 s.

Determine the exciting force as a function of time.

13.2 G REEN WATER ON D ECK

As was told in Chapter 11 , if the wave elevation comes above the instantaneous
position of the railing or above the freeboard at any point of the ship this will result
in water on deck, so called green water. To treat this problem in the frequency
domain we form the transfer function for the relative vertical motion between the
ship and water elevation for any heading and at any station, x, but only along the
longitudinal axis of the ship, y = 0:

ŝ3relc (ω , θ , x) = η̂3c (ω , θ ) − xη̂5c (ω , θ ) − ζˆc (x , θ ) , (13.4)

Where ζˆc (x , θ ) = a eı (k x cos(θ )+ε) is the complex amplitude of wave elevation at

station x.
The transfer function becomes

η̂3c (ω , θ ) − x η̂5c (ω , θ ) − ζˆc (x , θ )

Ts3relc (ω , θ , x) = , (13.5)
a

and the response spectrum

Ss3rel (ω , θ , x) = |Ts3relc (ω , θ , x)|2 S(ω) . (13.6)

Examples of response spectra for the vertical relative motion are shown in Fig. 13.1

202
 13.3. Propeller emergence

Figure 13.1: Response spectra of relative motion at bow, amidships and aft.

13.2.1 Probability of green water on deck

Green water will appear on deck if the relative motion is larger than the free-
board, Fb (x). The process of relative motion will be normal distributed because
its constituent processes, heave and pitch, are normal distributed. As then the
relative motion is a normal distributed process with zero mean and standard de-
viation, σs3rel (x), the maxima or minima of the relative motion will be Rayleigh
distributed. Therefore the probability of an individual minimum of relative motion
being deeper than the freeboard or the probability of green water will be
 2
Fb (x)
− 12 σs3rel (x)
Pgw (x) = P (s3rel (x) > Fb (x)) = 1 − P (s3rel (x) < Fb (x)) ≈ e . (13.7)
The number of green-water occurrences within a time period, t, can be assessed
by
t
Ngw = Pgw (x) . (13.8)
T02s3rel
For Hs = 8 m, T02 = 10 s, Fb = 10 m, x = −50 m and zero heading, the standard
deviation is σs3rel = 3.27 m and the response mean upcrossing period T02s3rel =
9.4 s. This gives the number of green water occurrences per hour Ngw /t = 3.54/hr
or equivalently one occurrence every 17 minutes. One can also ask how will the
No of occurrences change with freeboard. This is illustrated in Fig. 13.2.

13.3 P ROPELLER EMERGENCE

13.3.1 Probability of propeller emergence

The risk for the propeller to emerge out of the water and the risk that the bottom
will rise above the water and experience slamming at re-entry can be assessed in

203
13. DERIVED RESPONSES

Figure 13.2: No of green water occurrences per hour as a function of static free-
board.

complete analogy to what was done for green water. If the draught is T or the
propeller is T below the mean-water surface.
 2
T (x)
− 12 σs3rel (x)
PT (x) = P (s3rel (x) > T (x)) = 1 − P (s3rel (x) < T (x)) ≈ e . (13.9)

The number of emergencies within a time period, t, can then be assessed by

t
NT = PT (x) . (13.10)
T02s3rel
Thus if T = Fb = 10 m, and all other conditions also are the same as above, then
it must follow that NT = Ngw .

13.4 S LAMMING , W HIPPING , S PRINGING

13.4.1 Probability of Slamming

As told in Paragraph 6.8 slamming will occur when the ship’s bottom has risen
out of the wave and hits back at re-entry. For slamming to happen the relative
velocity at re-entry must exceed a threshold velocity, v0 . The first probability is
given by Eq. 13.9, the second probability must be assessed by the relative velocity
response. To treat this problem we form the transfer function for the relative
vertical velocity between the ship and the water elevation at any station, x:
 
s3relvelc (ω , θ , x) = −ı ω η̂3c (ω , θ ) − xη̂5c (ω , θ ) − ζˆc (x , θ ) , (13.11)

where, as before, ζˆc (x , θ ) = a eı (k x cos(θ )+ε) is the complex amplitude of wave

elevation at station x.

204
 13.4. Slamming, Whipping, Springing

Figure 13.3: Response spectrum of relative vertical velocity at bow.

The transfer function becomes

!
η̂3c (ω , θ ) − x η̂5c (ω , θ ) − ζˆc (x , θ )
Ts3relvelc (ω , θ , x) = −ı ω , (13.12)
a

and the response spectrum

Ss3relvel (ω , θ , x) = |Ts3relvelc (ω , θ , x)|2 S(ω) . (13.13)

Examples of response spectra for the vertical relative velocity are shown in Fig. 13.3.
As the relative vertical position and the vertical relative velocity can be shown
to be statistically uncorrelated processes [22] the probability of slamming is
 2  2
T (x) v0
− 12 σs3rel (x)
+ σs3relvel (x)
Pslam (x) = [P (s3rel (x) > T (x))] [P (s3relvel (x) > v0 )] ≈ e .
(13.14)
The number of slamming occurrences within a time period, t, can be assessed by
t
Nslam = Pslam (x) . (13.15)
T02s3rel

For Hs = 8 m, T02 = 10 s, T = 10 m, threshold velocity v0 = 2 m/s, x = 50 m

and zero heading, the standard deviation is s3relvel = 2.17 m/s and the response
mean upcrossing period T02s3rel = 9.4 s. This gives the number of slamming oc-
currences per hour Nslam /t = 2.31/hr or equivalently one occurrence every 26 min-
utes. One can also ask how the No of occurrences will change with draught. To do
that correctly the motion and velocity transfer functions should be reassessed for
each draught, which has not been done here. Approximately it has been illustrated
in Fig. 13.4 for draughts +/- 20 % of the given draught.

205
13. DERIVED RESPONSES

Figure 13.4: No of bottom emergences and slammings per hour as a function of

static draught. Threshold velocity arbitrary set to v0 = 2 m/s.

13.5 PASSENGER AND C REW

It is difficult to work if the accelerations onboard are too large. This is impor-
tant for the safety on ships and in the offshore industry, where also downtime of
operations should be avoided. In the passenger and cruising trade the comfort is
important. In freighters cargo may get loose so fastenings must be attached and
designed. In tankers the sloshing in incompletely filled tank compartments may
cause structural problems.

13.5.1 Seasickness
Vertical acceleration is probably the prime reason for seasickness (motion sick-
ness, kinetosis). For instance in Fig. 13.5 a motion-sickness index (MSI) is given
as the share of exposed persons throwing up within a given time at combinations
of mean frequency and standard deviation1 of acceleration.
As an application we can from the acceleration spectrum above calculate the
spectral moments and estimate the mean period of acceleration to T02acc = 9.7 s
and the standard deviation to 0.6 m/s2 . Then assessing seasickness in Fig. 13.5,
we find that between 10 and 20 % of passengers will through up within 2 hours
under such conditions.
1 RMS in Figure 9.9 and Table 9.2 is defined for the acceleration itself. As the mean is zero it

is identical to the standard deviation. The RMS value used in other contexts in this compendium
is defined for the amplitudes or double amplitudes of the quantities e.g. Hrms .

206
 13.5. Passenger and Crew

Figure 13.5: Motion-sickness index, MSI, is the share of exposed persons

throwing up within the given time at combinations of RMS (stan-
dard deviation) acceleration and mean frequency (Mandel 1979).

13.5.2 Human performance

For the performance of the crew the standard deviation 0.1 g for vertical acceler-
ation at the bridge can be adopted, a criterion denoted by “Intellectual work” by
a Nordforsk study [36]. In cabins and restaurants the criterion of “Transit pas-
sengers” is stricter and given as 0.05 g for vertical acceleration. Also criteria for
transversal (horizontal) acceleration are given. See Table 13.1.

In addition to the acceleration criteria, crew performance and passenger com-

fort is governed by the ability to move around the ship. This criterion can be
formulate in terms of ‘motion induced interruptions” (MMI) [15]. MMI gives an
indication of the number of events per minute in which a standing person will look
for support to maintain balance. According to [15] the allowed MMI is 1/min.
Above this limit crew performance is substantially degraded. For the passenger
areas the allowed MMI is 0.5/min. For practical application of MMI it is referred
to [15].

207
13. DERIVED RESPONSES

Table 13.1: Examples of human performance criteria.

MII Standard deviation Significant double amplitude

vertical transversal vertical transversal

acceleration acceleration acceleration acceleration
[1/min] [g] [g] [m/s2 ] [m/s2 ]
Navigation 1.0 0.10 0.05 4.0 2.0
bridge
Cabins and 0.5 0.05 0.04 2.0 1.6
restaurants

13.5.3 Recommended allowed limits of motions for safe working conditions

Recommended allowed limits of motions for safe working conditions at loading
and unloading of various ship types, from pleasure boats through fishing vessels
to large bulk carriers are presented by PIANC [40], see Table 13.2, and by the
Nordic Council [35]. The working group of the Nordic Council discusses its rec-
ommendations thus: ‘The vessel movements accepted during loading/unloading
operations are to some extent determined by local traditions. Generally, this ac-
ceptance does not consider possible reduced cost effectiveness. In relation to har-
bour layouts, however, it is necessary to take such factors into consideration.”
The choice of motion criteria depend on:
• loading/unloading methods, gear and personnel;
• vessel type, type of goods;
• characteristics of the vessel motions.

13.6 M OTIONS OF MOORED SHIPS IN HARBOURS

To calculate wave excited first order motions for moored ships in the open sea
or moored offshore platforms is a straightforward matter nowadays - with the
exception of roll motions of ships. To calculate motions for a ship at berth in a
harbour introduces the complications of the wave penetration into and possible
resonance of the harbour to the waves, the non-linear mooring arrangements, but
also the necessity to take the proximity of the quay walls, shore slopes and sea
floor into account, when assessing the hydrodynamic properties of the ship. This
is all automatically “included” when making physical model tests with regular or
pseudo random waves, albeit with model scale deficiencies. Field measurements
are also useful, but cannot be used for testing changes of harbour and mooring
layout. It is however gradually becoming possible to make “complete” numerical
modelling of ships moored in harbours [3].

208
 13.6. Motions of moored ships in harbours

Table 13.2: Criteria for safe working conditions [40].

13.6.1 Recommended allowed limits of motions for safe mooring conditions

See Tables 13.3 to 13.5 for criterias regarding limits of motions for mooring con-
ditions.

209
13. DERIVED RESPONSES

Table 13.3: Criteria for vessel movements for safe mooring conditions at berth.
The movements are peak-peak values. For the berth to be accept-
able, the frequency of these movements should be less than 3 h/year.
From Nordic Council [35].

Table 13.4: Recommended velocity criteria for Safe mooring conditions of var-
ious ships [40].

Ship Size Surge Sway Heave Yaw Pitch Roll

DWT m/s m/s m/s ◦ /s ◦ /s ◦ /s

1 000 0.6 0.6 – 2.0 – 2.0

2 000 0.4 0.4 – 1.5 – 1.5
8 000 0.3 0.3 – 1.0 – 1.0

210
 13.6. Motions of moored ships in harbours

Table 13.5: Wave criteria for small craft and pleasure boats. The acceptable
frequency of occurrence is one to a few times per year [40].

Ship Length Beam/Quartering Seas Head Seas

Period Height Hs Period Height Hs
(m) (s) (m) (s) (m)
< 2.0 0.2 < 2.5 0.2
4–10 2.0 – 4.0 0.1 2.5 – 4.0 0.15
> 4.0 0.15 > 4.0 0.20

< 3.0 0.25 < 3.5 0.30

10–16 3.0 – 5.0 0.15 3.5 – 5.5 0.20
> 5.0 0.20 > 5.5 0.30

< 4.0 0.30 < 4.5 0.30

20 4.0 – 6.0 0.15 4.5 –7 .0 0.25
> 6.0 0.25 > 7.0 0.30

211
 REFERENCES

[1] B.G. Airy. Encycl. Metrop., chapter Tides and waves. London, 1845.

[2] L. Bergdahl and M. Johansson. Time simulation of the motion of a tension leg
platform. In BOSS ’88, Trondheim, June 1988.

[3] H. Bingham. A hybrid Boussinesq-panel method for prediction the motion of a

moored ship. Coastal Engineering, 40:21–38, 2000.

[4] S.K. Chakrabarti. Hydrodynamics of Offshore Structures. Springer-Verlag, 1987.

[5] J.P. Comstock, editor. Principles of Naval Architecture. The Society of Naval Ar-
chitecture and Marine Engineers, 1967.

[6] R.R. Craig. Structural Dynamics. John Wiley and Sons, New York, 1981.

[7] W.E. Cummins. The impulse response function and ship motions. Schiffstechnik,
9:101–109, 1962.

[8] J.W. Daily and D.R.F. Harleman. Fluid Dynamics. Addison-Wesley Pub. Co., Read-
ing, Mass., 1966.

[9] R.G. Dean. Evaluation and development of water wave theories for engineering
application. presentation of research results. Special report I, US Army, Corps of
Engineers, Coastal Engineering Research Center, 1974.

[10] R.G. Dean and R.A. Dalrymple. Water Wave Mechanics for Engineers and Scien-
tists. World Scientific, 1991.

[11] DnV. Position mooring (POSMOOR), rules for the classification of mobile offshore
units; part 6, chapter 2. Technical report, Det Norske Veritas, 1989.

[12] O.M. Faltinsen. Sealoads on Ships and Offshore Structures. Cambridge University
Press, 1990.

213
REFERENCES

[13] O.M. Faltinsen and F. Michelsen. Motions of large structures in waves at zero
Froude number. In R.E.D. Bishop and W.G. Price, editors, International Symposium
of Dynamics of Marine Vehicles and Structures in Waves, pages 91–106, London,
1974. Mechanical Engineering Publications Ltd.

[14] L. P. Graham. Safety Analysis of Swedish Dams: Risk Analysis for the Assessment
and Management of Dam Safety. Licentiate Thesis, Division of Hydraulic Engi-
neering, Department of Civil and Environmental Engineering, Royal Institute of
Technology, 1995.

[15] R. Graham. Motion-induced interruptions as ship operability criteria. Naval Ship

Engineers Journal, pages 65–71, 1990.

[16] O. Grim. Oscillation of buoyant two dimensional bodies and the calculation
of the hydrodynamic forms. Technical Report 1171, Hamburgische Schiffbau-
Versuchsanstalb, 1959.

[17] K. Hasselmann and et al. Measurements of wind-wave growth and swell decay
during the joint Nort sea wave project (JONSWAP), volume Reihe A(8◦ ) 12 of
Ergänzungsheft zur Deutschen Hydrographischen Zeitschrift. Deutsches Hydro-
graphisches Institut, Hamburg, 1973.

[18] F.B. Hildebrand. Advanced Calculus for Applications. Prentice-Hall, Englewood

Cliffs, N.J., 1976.

[19] N. Hogben, N.M.C. Dacunha, and G.F. Olliver. Global Wave Statistics. British
Maritime Technology, Unwin Brothers Ltd., UK, 1986.

[20] HSE. Offshore installations: Guidance on design, construction and certification,

section 32, station keeping, draft. Technical report, Health and Safety Executive
(UK), 1991.

[21] ITTC. Recommended procedures and guidelines – testing and extrapolation meth-
ods for responses, sea keeping, sea keeping experiments. Technical report, Interna-
tional Towing Tank Conference, 2005.

[22] J.J. Jensen. Load and Global Response of Ships, volume 4 of Ocean Engineering
Book Series. Elsevier, 2001.

[23] J.J. Jensen, A.E. Mansour, and A.S. Olsen. Estimation of ship motions using closed-
form expressions. Ocean Engineering, 31(61–85), 2004.

[24] J.J. Jensen and T.P. Pedersen. Wave-induced bending moments in ships a quadratic
theory. Transctions of The Royal Institute of Naval Architects, 1978.

[25] M. Johansson. Barrier-Type Breakwaters: Transmission, Reflection and Forces.

PhD thesis, Department of Hydraulics, Chalmers University of Technology, 1989.

[26] B. Kinsman. Wind Waves. Prentice Hall, Englewood Cliffs, N.J., USA, 1965.

214
 REFERENCES

[27] B.V. Korvin-Kroukovsky and W.R. Jacobs. Pitching and heaving motions of a ship
in regular waves. Transactions SNAME, 1957.

[28] Krogstad and Arntsen. Linear wave theory. www.ntnu.no, 2002.

[29] B. LeMéhauté. An Introduction to Hydrodynamics and Water Waves. Springer-

Verlag, New York, 1976.

[30] A.R.J.M. Lloyd. SEAKEEPING: Ship Behaviour in Rough Weather. ELLIS HOR-
WOOD LIMITED, Chichester, England, 1989.

[31] M.S. Longuet-Higgins. Statistical properties of wave groups in a random sea state.
Philosophical Transactions of the Royal Society of London Series A, 312:219, 1984.

[32] S.R. Massel. Ocean Surface Waves: Their Physics and Prediction, volume 11 of
Advanced Series on Ocean Engineering. World Scientific, Singapore, 1996.

[33] R. Miche. Mouvements ondulatoires des mers en profondeur constante on de-

croisante. Annales des Ponts et Chaussees, 1944.

[34] D. Mollison. The prediction of device performance. In B. Count, editor, Power from
Sea Waves, pages 135–172. Academic Press, London, 1980.

[35] NC. Ship motions in harbours (in Norwegian. Technical report, Official Committee
of the Nordic Council on Transport Questions, 1986.

[36] NCP. Assessment of ship performance in a seaway. Technical report, The Nordic
Cooperative Project, 1987.

[37] NMD. Regulations of 10 January 1994 for mobile units with technical installations
for production on board (in Norwegian). Technical report, Norwegian Maritime
Directorate, 1994.

[38] N. Nordenström. A method to predict long-term distributions of waves and wave-

induced motions and loads on ships and offshore structures. Technical report, DnV
Publication No. 81, 1973.

[39] NPD. Veijledning om laster og lastvirkninger til forskrift om baerende konstruk-

sjoner i petroleumsvirksomheten. Technical report, Norwegian Petroleum Direc-
torate, 1992.

[40] PIANC. Criteria for movements of moored ships in harbours, a practical guide. Ptc
ii, report of working group no. 24, supplement to bulletin no. 88, PIANC, 1995.

[41] J.B. Roberts and P.D. Spanos. Random Vibration and Statistical Linearization. John
Wiley and Sons, Chichester, 1990.

[42] H. Rouse, editor. Engineering Hydraulics. Wiley, New York, 5th edition, 1950.

[43] N. Salvesen, E.O. Tuck, and Faltinsen O.M. Ship motions and sea loads. Trans.
SNAME, 78:250–287, 1970.

215
REFERENCES

[44] R. Silvester. Coastal Engineering 1. Elsevier, 1974.

[45] L. Skjelbreia and J. Hendrickson. Fifth order gravity wave theory. In Proc. 7th
Coastal Engr. Conf., volume 1, pages 184–196, 1961.

[46] M. St. Denis and W. J. Pierson. On the motion of ships in confused seas. Transac-
tions SNAME, 61:280–357, 1953.

[47] W.T. Thompson. Theory of Vibration with Applications. Prentice-Hall Inc., Engle-
wood Cliffs, 1972.

[48] M.J. Tucker, P.G. Cahallenor, and D.J.T. Carter. Numerical simulation of a ran-
dom sea: a common error and its effect upon wave group statistics. Applied Ocean
Research, 6(2), 1984.

[49] U.S. Army Coastal Engineering Research Center. Shore Protection Manual, 1984.

[50] G. van Oortmerssen. The motions of a moored ship in waves. Technical Report 510,
Neth. Ship Model Basin, 1976.

[51] J.H. Vugts. The hydrodynamic coefficients for heaving, swaying and rolling cylin-
ders in a free surface. NSRC report 112S, 2002.

[52] G. Wahl. Wave statistics from swedish coastal waters. In R.E.D. Bishop and W.G.
Price, editors, The Dynamics of Marine Vehicles and Structures in Waves, volume 4,
pages 33–40. Office of Naval Research, and the Royal Institution of Naval Archi-
tects, 1974.

[53] R. Wiegel. Oceanographical Engineering. Prentice Hall, Englewood Cliffs, N.J.,

USA, 1964.

[54] J.M. William. Tables of Progressive Gravity Waves. Pitman Publishing Ltd, 1985.

[55] I.R. Young. Wind Generated Ocean Waves, volume 2. Elsevier Ocean Engineering
Series, 1999.

216