~ rl
~
l~
ARGONAUTICS
MARINE ENGINEERING
naval architects
.
 marine surveyors .
heavy lift transportation consultants .
SEAKEEPING:
Ship Behaviour in
Rough Weather
A. R. J. M. LLOYD,
B.Sc.,Ph.D.
Distributors:
Australia and New Zealand:
JACARANDA WILEY LIMITED
GPO Box 859, Brisbane, Queensland 4001, Australia
Canada:
JOHN WILEY & SONS CANADA LIMITED
22 Worcester Road, Rexdale, Ontario, Canada
Europe and Africa:
JOHN WILEY & SONS LIMITED
Baffins Lane, Chichester, West Sussex, England
North and South America and the rest of the world:
Halsted Press: a division of
JOHN WILEY & SONS
605 Third Avenue, New York, NY 10158, USA
SouthEast Asia
JOHN WILEY & SONS (SEA) PTE LIMITED
37 J alan Pemimpin # 0504
Block B, Union Industrial Building, Singapore 2057
Indian Subcontinent
WILEY EASTERN LIMITED
4835/24 Ansari Road
Daryaganj, New Delhi 110002, India
COPYRIGHT NOTICE
All Ri!lhts Reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechamcal, photocopying, recording or otherwise,
without the permission ofElhs Horwood Limited, Market Cross House, Cooper Street, Chichester, West
Sussex, England.
Table of contents
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
A note on units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1 Seakeeping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
___/
Fluid dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.1
2.2
2.3
2.4
2.5
2.6
2. 7
2.8
2.9
2.10
2.11
3
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Euler's equations of motion for an inviscid fluid . 1. . . . . . . . . . . . .
Equation of continuity . . . . . . . . . . . . . . . . .
The velocity potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Integration of Euler's equations of motion: Bernoulli's equation ...
Laplace's equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The stream function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Some simple flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conformal transformations . . . . . . . . . . . . . . . . . . . . . . . . . . .
Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Lifting surface characteristics . . . . . . . . . . . . . . . . . . . . . . . . . .
28
29
33
35
38
40
41
43
52
58
59
Regular waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The potential function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Pressure contours and the surface profile . . . . . . . . . . . . . . . . . .
Wave slope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Regular wave characteristics . . . . . . . . . . . . . . . . . . . . . . . . . .
Particle orbits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Pressure fluctuations under a wave . . . . . . . . . . . . . . . . . . . . . .
Energy of a regular wave ......... ,, . . . . . . . . . . . . . . . . . .
64
65
66
72
73
74
86
86
Table of contents
3.9
4 Ocean waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.1
Wave generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.2 Statistical analysis of time histories of irregular waves . . . . . . . . . . 94
4.3
Fourier analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.4 The wave energy spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.5 Spectral moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.6 Idealised wave energy spectra . . . . . . . . . . . . . . . . . . . . . . . . 106
4.7 Wave slope spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.8 Wave spreading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5 Ocean wave statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Visual observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3 Wave atlases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
121
121
121
124
132
132
133
139
141
............................
.. .. .. ... .. .. ... .. .. . .. . .. . .
.. .. .. .. ... .. ... . .. .. .. . .. . .
. ... .. .. ... .. .. .. .. .. . .. . .. .
............................
151
151
151
154
158
9 Strip theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2 Strip motions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.3 Hydrodynamic coefficients . . . . . . . . . . . . . . . . . . . . . . . . . .
9.4 Excitations in regular waves . . . . . . . . . . . . . . . . . . . . . . . . .
170
170
171
173
181
10 Hydrostatic coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.2 Vertical plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.3 Lateral plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
191
191
191
192
11
196
196
196
206
211
Table of contents
11.5
12
7
218
Roll damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.1
Sources of roll damping . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.2
Nonlinear roll damping: equivalent linearisation . . . . . . . . . . . .
12.3
Eddy roll damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.4
Skin friction roll damping . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.5
Appendage roll damping . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.6
Total roll damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
223
223
223
225
228
13
Ship
13.1
13.2
13.3
13.4
13.5
13.6
13.7
13.8
13.9
13.10
13.11
13.12
234
234
234
235
240
243.
247
252
253
257
257
259
262
14
263
263
264
266
269
271
272
275
15
16
Model testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.1
Reasons for model seakeepi))g experiments . . . . . . . . . . . . . . .
16.2
Model experiment scaling . . . . . . . . . . . . . . . . . . . . . . . . . . .
Open water model experiments . . . . . . . . . . . . . . . . . . . . . . .
16.3
16.4
Laboratory test facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.5
Wave makers and.beaches . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.6
Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.7
Model materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2~
233
286
286
286
297
297
300
303
307
Table of contents
8
16.8
16.9
16.10
16.11
308
312
318
323
17 Probability formulae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.2
Probability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.3
HistQgrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4 The probability density function . . . . . . . . . . . . . . . . . . . . . . .
17.5
The Gaussian probability density function . . . . . . . . . . . . . . . .
17.6 The Rayleigh probability density function . . . . . . . . . . . . . . . . .
17.7 Significant wave height and related statistics . . . . . . . . . . . . . . .
17.8 Joint probabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
327
327
327
327
331
333
337
339
342
18 Roll stabilisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.1
Motion reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.2
Bilge keels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.3
Active roll stabiliser fins . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.4
Passive tanks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
343
343
344
349
377
398
398
398
400
401
403
406
20
409
409
410
413
421
422
424
21
425
425
426
429
433
22
437
437
439
440
449
Table of contents
22.5
23
Operational effectiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23.2
Sea area and season . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23.3
Ship speed and course . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23.4
Calculation of operational effectiveness . . . . . . . . . . . . . . . . . .
455
455
455
456
457
24
460
460
461
461
468
473
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
Numerical values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
For
Sonya, Abigail and Tobin
who never complained in four and a half years
Acknowledgements
I have been engaged in research on seakeeping since 1968. During that time I have
been helped by many colleagues both within the Admiralty Research Establishment
and outside. In particular I would like to pay tribute to the outstanding contributions
of Mrs P.R. Loader, Dr R.N. Andrew and Mr W. B. Marshfield. Without their
constant support, advice and inspiration over the years this book would probably
never have been written.
Gosport
January 1988
A. R. J. M. Lloyd
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
Notation
(a)
ROMAN SYMBOLS
Symbol
A
Meaning
area;
parameter in Bretschneider wave
energy spectrum formula;
Fourier coefficient;
parameter in SM formula
generalised virtual mass or inertia
coefficient: ith force or moment
due to jth unit acceleration
crosssection area of superstructure and hull above waterline
Units
metres 2
metres 2/second 4
metres
kN/(metre/second 2 )
or
kN metres/(metre/second 2 )
metres 2 or
kN/(radian/second 2 )
or
kN metres/(radian/second 2 )
aspect ratio;
inertia;
radius of cylinder;
resistance augment
mapping coefficients
stabiliser fin servo coefficients
a;(i=l,6)
aij (i, j
= 1,6)
tonnes
metres
r;
 , S{!conds,
seconds 2
kN/(metre/second 2 )
or
I,
'
kN metres/(metre/second 2 )
kN /(metre/second 2 ) or
kN metres/(metre/second 2 )
or
kN/(radian/second 2 )
or
a;,(i=1,6)
kN metres ( radian/second 2 )
kN/(radian/second 2 )
or
kN metres/(radian/second 2 )
kN metres/(metre/second 2 )
or
kN metres/(radian/second 2 )
Notation
14
b; (i = 1,6)
b;j
(i,j = 1,6)
kN metres/(radian/second 2 )
seconds 4
metres
metres
metres
metres
kN seconds/metre
kN/(metre/second)
or
kN metres/(metre/second)
kN/(metre/second)
or
kN metres/(metre/second)
or
kN/(radian/second)
or
b;~
b~;
(i = 1,6)
(i = 1,6)
kn metres/(radian/second)
,seconds,
seconds2
kN/(radian/second)
or
kN metres/(radian/second)
kN metres/(metre/second)
or
kN metres/(radian/second)
kN metres/(radian/second)
kN/metre 2
Notation
C;
C;j
(i = 1,6)
(i, j
= 1,6)
wave celerity;
stiffness coefficient;
fin chord
ith force or moment due to unit
wave depression
generalised stiffness coefficient;
ith force or moment due to jth
unit displacement
15
metres/second
kN/metre
metres
kN/metre
or
kN metres/metre
kN/metre
or
kN metres/metre
or
kN/radian
or
E
F
F; (i = 1,6)
kN metres/radian
kN metres/radian
kN metres/radian
kN metres/radian
kN
metres
metres
metres
metres
joules/metre width
kN
metres
kN
or
kN metre
Froude number
ith force or moment due to waves
on restrained ship
kN
or
kN metres
metres  1 , radians 
metres
metres
etc.
Notation
16
I
I;;
(i = 4,6)
gain margin
rollrighting lever
acceleration due to gravity
wave height;
beam/draught ratio;
feedback path complex gain
characteristic wave height
significant wave height
distance from pivot point to
centre of gravity;
height
mass moment of inertia
ith mass moment of inertia
crossproducts of inertia
second moment of area of waterplane about transverse axis
second moment of area of waterplane about longitudinal axis
y( 1)
KG
Ko
Ku
KT
Kl, K2, K3
KG
k
k; (i = 4,6)
L
LFE
M
MI (i = 2,4)
metres
metres/second 2
metres
metres
metres
metres
metres
tonne metres 2
tonne metres 2
tonne metres 2
metres 4
metres 4
,seconds, seconds 2
metres
metres 1
metres
kN
metres
metres/second 2
metres 3/second
tonne metres/second
per metre
or
M~;
(i= 2,4)
Notation
Mil
MSI
17
minutes 1
per cent
metres 2/second
tonnes
metres 2
or
variance of velocity
radians 2
metres 2/second 2
or
variance of acceleration
radians 2/second 2
metres 2/second 4
or
radians 2/second4
metres 2/second"
or
radians2 /second"
N
n
p
number of observations;
number of motion cycles;
number of observations per hour;
propeller revolutions per second
number of ways in which r positive answers can be achieved from
N questionnaires
dimension normal to passive tank
axis
pressure;
probability;
proportion of time;
power
proportion of questionnaires
returning a positive response to a
particular question
coefficients to weight contribution of 2mth multi pole
kx 81 cos 11;
propeller torque
passive tank parameter
total velocity;
passive tank coefficient of
resistance;
minimum number of votes in a
questionnaire to establish a
majority opinion with a 95% confidence level
hours 1
seconds 1
metres
kN/metre 2
kW
kN metres
tonne metres
metres/second
18
S(ro)
Notation
scale ratio Ls/Lm;
resistance
Reynolds number
radial coordinate;
number of positive questionnaire
answers
nth ship response
relative local vertical motion
between ship and sea surface
energy spectral ordinate
kN
metres
various
metres
metres 2/(radian/second)
or
SM
SA;(i=1,3)
s8 ; (i= 1,3)
S;
(i = 1,3)
u,v
v
v
radians2/(radian/second)
various
metres
metres
metres/second 2
metres/second 2
metres
seconds
kN
seconds
seconds
seconds
metres/second
metres/second
metres/second
or knots
metres/second
metres 3
metres/second
volts
kN metres
Notation
w
w(z)
WT
X,Y
x,y
xi (i = 1,6)
XoJ
Xoo
Z 1, Z2
(b)
width
complex potential function
<Pz + i'l'z in z plane
Taylor wake fraction
force per unit mass in x and y
directions
cartesian coordinate system
ith displacement of centre of gravity relative to moving origin 0
amplitude of Jth peak or trough
initial value of x in decaying
oscillation
eddymaking rolldamping
functions
complex variable x + iy;
displacement of fluid surface
from stabiliser tank datum level
19
metres
metres 2/second
metres/second 2
metres
metres or radians
metres
metres
metres
metres
GREEK SYMBOLS
Symbol
(X
r"
y
Yi (i = 1,6)
v
Lix,~y
Meaning
angle of incidence;
wave slope;
slope of pressure contour
slope of beach;
fin depression angle
deadrise angle at keel
counting functional for nth ship
response
parameter in JONSWAP wave
energy spectrum formula;
inclination of hull section at
waterline
ith peak wave force or moment
leads maximum wave depression
at 0 by Yi radians
volume of displacement
deviations of particle from datum
position beneath a wave
boundarylayer thickness
ith peak (positive) motion displ(!cement leads maximum wave
depression at 0 by oi radians
Units
radians
radians
radians
radians
radians
radians
radians
radians
metres:>
metres
metres
radians
radians
20
J.tw
v
p
(J
CJo
Notation
bandwith parameter;
phase
ith peak (positive) force or
moment leads maximum positive
motion by Ei radians
phase margin
passive tank motion phase: tank
angle leads roll motion by 11
radians
passive tank moment phase: tank
moment leads roll motion by 12
radians
complex variable x 82 + ix 83 ;
depression of water surface below
mean level
significant wave amplitude
wave amplitude exceeded with
probability 1/n
propeller efficiency;
decay coefficient
angular coordinate
tuning factor co/coc
wave length
primary wave direction
fractional loss in metacentric
height due to passive stabiliser
tank
viscosity
secondary wave direction
mass density
section area coefficient
rms displacement
radians
radians
degrees
radians
radians
metres
metres
metres
metres
radians
metres
radians
tonnes/(metre second)
radians
tonnes/metre 3
metres
or
rms velocity
radians
metres/second
or
rms acceieration
radians/second
metres/second 2
or
<j), <P'
X
time constant;
stabiliser tank angle;
shear stress
velocity potential
angle of inclination of flow to y
axis
stream function
radians/second 2
seconds
radians
kN/metre 2
metres 2/second
radians
metres 2/second
Notation
force potential
frequency;
wave frequency
Q
(J)
(c)
21
metres 2/second 2
radians/second
radians/second
Unless otherwise shown in the table of notation the following suffixes are used:
Symbol
A
a
area
B
BK
BL
b
c
crit
D
D
d
ds
E
e
FP
FS
f
GH
H
h
I
in
J
k
L
LA
M
m
max
n
obs
out
p
r
Meaning
appendage; apparent; air
aft; irregular amplitude or height; added; augmented
sea area
axes fixed in hull; Bretschneider
bilge keel
boundary layer
bilge
calm water; fin controller
critical value or criterion
demanded angle; drag
mean draught
duct; decay
deck submergence
Earthfixed axes; eddy
encounter; effective
fin; skin friction; frame
measured from forward perpendicular
fin servo
i
forward
group; forward path
open loop
time history or run time; feedback path
histogram bin
interference
f
suffix used to indicate degree of freedom or direction of motion
input
JONSWAP
keel; kinetic
lift; experiment tank length; beach
lateral acceleration
measured from midships
measured value; model value
maximum
nth observation or frequency
visually observed
output
propeller; wave probe; parent; peaks; port; pressure P; potential
halfseparation of wires, struts or potentiometer strings; reservoir; root
I'
22
r3
s
s3
season
slam
stag
stall
TH
t
u
v
w
z
z
s
s
(X
X
0
113
Notation
relative motion
ship value; starboard
absolute vertical motion
season
slamming
stagnation
stall
experiment tank
wave period and height
passive tank; tip
ship speed
voltage
waves; wavemaking; wind; waterplane
zero crossing
value in complex z plane
value in complex plane
wave slope
wave amplitude
wave direction
regular amplitude; modal value
significant value
undamped natural frequency or period
average; complex conjugate
local value; nondimensional value
oscillating part
A note on units
Systeme Internationale (SI) units are used throughout this book. Tonnes are
adopted as the unit of mass with kilonewtons (kN) as the corresponding unit of force
which will accelerate a unit mass at one metre/second 2 . Metres and seconds are used
as the units of length and time. Following common practice, ship speeds are given in
knots or nautical miles per hour.
Appropriate units are quoted for all equations since this serves to remind the
reader of the physical reality expressed by the equations. The reader may, however,
substitute his own units providing these are based on a rational system in which a unit
force is defined as one which imparts unit acceleration to a unit mass. Thus, for
example, a pressure equation quoted here in kilonewtons per square metre is equally
valid in pounds (force) per square foot (when the unit of mass is the slug) or poundals
per square foot (if the mass unit is the pound).
i
I'
>
1
Seakeeping
There are three things which are too wonderful for me, yea, four which I
know not:
the way of an eagle in the air;
the way of a serpent on a rock;
the way of a ship in the midst of the sea;
and the way of a man with a maid.
Proverbs Chapter 30 verses 1819
In the days of sail ships were very much more dependent on the weather than they
are today. Squarerigged sailing vessels could not sail directly into the wind and were
strictly limited in their ability to go where the master wanted tin severe conditions it
was necessary to shorten sail and even to ride out a storm under bare poles. Many a
ship was lost because she was driven ashore under such circumstances.
Economic pressures often demanded that the ship's ma~ter spread as much
canvas as he dared in order to make the best speed. This is nowhere more graphically
illustrated than in the stories of the clipper 'races' from China to Europe in the
nineteenth century. The first ship home with the newly harvested tea crop co~ld
demand a premium price for her cargo. Speed was of the essen,ce and these ships
sprouted all sorts of additional sails to make the most of every breath of wind
available.
A heavily laden overcanvassed ship must have been an unpleasant home for the
sailors and passengers in rough weather. With the lee gunwhale submerged, the
decks continually awash, deckhouses damp and cold, life must have been miserable.
Yet even in these circumstances the crew would be expected to continue to navigate
and steer the ship and to go aloft in order to shorten sail or spread additional canvas
as the master demanded.
However, the real problems of seakeeping only came to be recognised with the
demise of sail and the advenLof steam as the prime motive power. Now, for the first
time, ships could steam directly into the wind and s~a with a consequent increase of
26
Seakeeping
[Ch. 1
pitch and heave motions. The damaging effects of shipping heavy seas over the bow
began to be experienced. The punishing effects of high speed in rough weather were
not fully understood and at least one ship (HMS Cobra in 1901) is believed to have
been lost after her hull broke in two after slamming in rough weather.
At the same time the steadying effects of tall masts and a good spread of canvas
were lost and the new steam ships were found to roll heavily. It is ironic that this
beneficial effect of sails has only recently been rediscovered with the emerging
technology of windassisted propulsion for lowpowered merchant ships.
At about this time William Froude, an eminent Victorian engineer, proposed to
build what would become the world's first model towing tank at Torquay in Great
Britain. He had recently developed scaling laws for predicting the resistance of ships
from tests on models and he intended to use the tank for the required scale model
experiments. The British Admiralty accepted Froude's proposal on the condition
that he also used the tank to inyestigate ways of reducing the rolling motion of ships.
In due course towing tanks were built in many different countries. These were
often fitted with wave makers which allowed the behaviour of model ships in waves to
be studied at leisure and provided, for the first time, a technique for refining a fullscale design to ensure adequate performance in rough weather. These model
experiments were usually confined to tests in regular head or following waves with
occasional tests at zero speed in beam waves. Tests at other headings or in more
realistic irregular waves were impossible because of the long narrow shape of the
towing tanks and the simplicity of the wave makers.
These early model experiments allowed some limited developments in the study
of seakeeping but they could not be used to predict the actual performance of ships at
sea because no technique for relating the behaviour of the model in the regular waves
of the laboratory to the behaviour of the ship in the chaotic environment of the real
ocean was available. This situation prevailed for sixty years or more and the study of
seakeeping remained in effective limbo until the publication of a landmark paper by
StDenis and Pierson in 1953. This showed, for the first time, how this problem could
be solved using the techniques of spectral analysis borrowed from the field of
electromagnetic communications.
At about the same time theoretical methods of predicting the behaviour of ships
in regular waves were being developed. The breakthrough came with Ursell's
(1949a,b) theory for predicting the characteristics of the flow around a circular
cylinder oscillating in a free surface. Classical transformation techniques allowed
these results to be applied to a wide range of shapes of shiplike crosssection and the
fundamentals of modern ship motion theory were born.
These developments some forty years ago provided the basic tools requiied to
develop routine techniques for the prediction of ship motions in something
approaching the real irregular wave environment of the ocean. It was now possible
for the first time to predict the rough weather performance of a ship at the design
stage and to allow seakeeping to take its rightful place in the design process.
Since that time seakeeping has remained an active field of research, but
developments have been in the nature of progressive refinements rather than
spectacular advances. Techniques for designing roll stabilisers, criteria determination, prediction of longterm motion statistics and operational effectiveness have
Ch. 1]
27
Seakeeping
all been added to the naval architect's armoury of weapons: seakeeping performance
prediction should now be a routine in any ship design office.
Unfortunately these developments have not been accompanied by much readily
obtainable literature on the subject outside the specialist papers and publications of
the learned societies and research institutes. The time therefore seems ripe for the
publication of a standard text on seakeeping covering all essential aspects in some
detail.
Although the underlying physical principles of seakeeping theory are not generally difficult to understand, the intimate details are mathematically complicated. It
follows that calculations of ship motions and related phenomena require access to
suitable computer programs and computers. No real progress can be made without
them. Fortunately many suitable programs are available in educational, research and
design establishments as well as computer bureaux throughout the world. The PAT86 suite of seakeeping computer programs (available at the Admiralty Research
Establishment at Haslar in the United Kingdom) was used for the examples of ship
motion calculations presented in this book.
No book can hope to cover such a complex subject completely. Indeed such an
undertaking would be inappropriate for all but the most specialised readers. This
book is therefore intended for the practising (and practical) naval architect and the
student. It is hoped that others on the fringes of the profession will also find the book
useful.
I'
2
Fluid dynamics
2.1
INTRODUCTION
Many aspects of the behaviour of ships in rough weather depend on the general laws
of fluid flow. The study of the waves on the sea surface, the resulting responses to
them and detailed considerations of the flow around appendages such as roll
stabiliser fins are all based on the classical equations of fluid dynamics which were
first expounded by the great mathematicians of the eighteenth century. An extensive
knowledge of fluid dynamics is not required for an understanding of seakeeping
theory: nevertheless, a basic knowledge of certain aspects of the subject is needed
and this chapter is intended to give the necessary grounding. Readers requiring a
more detailed treatment of the subject are referred to O'Neill and Charlton (1986).
At the time when the basic axioms were first established the results had little
practical application, and remained only of academic interest until the advent of
flight and powered ships in the late nineteenth century. Even then the practical
application of this rigorous early work was often hampered by the impossibility of
solving the resulting equations for any but the simplest of cases. It was usually
necessary to pretend that the fluid was 'ideal'; that is, it had no viscosity or surface
tension and was incompressible (i.e. the density remains constant at all times). These
rather limiting assumptions resulted in predictions of fluid flow which were sometimes at variance with the observations of experiments.
Fortunately for the student of seakeeping, the negiect of viscosity, surface
tension and compressibility in these equations allows good theoretical predictions of
many kinds of fluid flow which are important in the determination of the behaviour of
ships in rough weather. This is not to say that water does not have viscosity, surface
tension or compressibility; rather that these qualities apparently have little effect on
the fluid flows concerned. So relatively simple solutions to many relevant problems
are possible using the classical equations of fluid dynamics described in the following
sections.
Fluid flows are, in general, threedimensional and those associated with the
Sec. 2.2]
29
behaviour of ships in rough weather are no exception. However, we shall see that it is
often possible to simplify matters and treat certain limited aspects of these flows as
twodimensional. In this case the flow is always in a plane parallel to the xy plane as
shown in Fig. 2.1. There are no variations of velocity, pressure, density or any other
r:
property of fluid in the direction normal to the xy plane. For convenience the fluid is
assuJ11ed to have a constant depth d metrest.
I'
2.2
Euler developed the basic equations of motion for fluid particles by considering the
forces on a small rectangular block of fluid. For present purposes his derivation may
be simplified by considering only twodimensional flow: in this case the block has
sides of length ox, oy and d metres and has its centre at a point (x, y) in the xy plane
as shown in Fig. 2.2. ox and oy are at first supposed to be small but finite.
In. general the properties of the fluid which are of interest (pressure, density,
velocity, acceleration etc) will vary throughout the xy plane but will have specific
values at the point (x, y). Suppose that the pressure at the point (x, y) is P kN/metre 2 .
A possible variation of pressure in the x direction is shown in Fig. 2.3. Obviously this
t The use of the word 'depth' does not necessarily imply that
t~e
[Ch.2
Fluid dynamics
30
0~~
r1
~
.........(?+
PaP ,)x)ov~
ax
(x,y)
ap
ax
i'lx)bv
2
'
Fig. 2.2 Euler's equations of motion: forces acting on a particle in the x direction.
<)X
<)X
I
Sec. 2.2]
31
will result in different pressures on the end faces of the block and give rise to a
'pressure force' in the x direction. Other possible forces might be externally applied
(such as gravity) or due to friction as adjacent fluid overtakes or is overtaken by the
block under consideration. If the fluid is assumed inviscid (that is, it has no viscosity)
there can be no frictional forces and the only forces which can exist are therefore due
to pressure differences or are applied externally.
If the dimension ox is small the variation of pressure in the x direction may be
approximated by the straight line of slope
oP
ox
kN/metre 2/metre
as shown in Fig. 2.3. Then the pressures at the centres of each end face of the block
shown in Fig. 2.2 are approximately
P oP ox
ox 2
kN metre 2
P+ oP ox
ax 2
kN/metre 2
ox) oy d( pap
ox 2
(p + oPox ox) oy
2
= 
oP ox oy d
ox
kN
I'
The volume of the block is ox oy d metres 3 . If the mass density of the fluid is p
tonnes/metre 3 the mass of fluid in the block is p ox oy d tonnes. So if X is the
externally applied force per unit mass the resulting force in the x direction is
Xp
ox oy d
kN
kN
Fluid dynamics
32
[Ch. 2
These forces result in an acceleration it of the fluid block in the x direction. From
Newton's second law
force
mass x acceleration
so
aP
ox oy d+Xp ox oy d
ax
pox oy d it kN
(2.1)
1 aP
X p ax
it metres/second 2
(2.2)
The fluid block will have some velocity which will have components in the x andy
directions. We shall designate these as u and v respectively. These component
velocities will vary with time and with the position (x, y) ofthe particle. For example
u = u(x, y, t)
metres/second
du
au
au
au
dx + dy + dt metres/second
ax
ay
at
(2.3)
au dx au dy au
du
u =  =   +  + metres/second 2
dt
ax dt ay dt at
(2.4)
but
u
dx
dt'
dy
dt
metres/second
(2.5)
so that
au
au au
+ v  + metres/second 2
ax
ay at
u
(2.6)
Equation of continuity
Sec. 2.3]
33
I oP
xP ox
ou + v ou +ou
ox oy ot
u
metres/second 2
(2.7a)
Y~ oP
P ay
ov
ax
au ov
oy ar
u  + v  + metres/second 2
(2.7b)
These equations, as we have seen, are approximate for finite dimensions of the
fluid block but become exact if the block dimensions are made infinitesimal. They are
valid for inviscid flow but the fluid may be compress;ble. In the general form stated
they are insoluble.
2.3
EQUATION OF CONTINUITY
If the velocity components of the fluid at the centre of the framework are again u
and v metres/second and the mass density at that point is p tonnes/ metre 3 , the mass
of fluici flowing in the x direction through the centre of the frame is approximately
pu oy d tonnes/second. Again this result is only generally exact if the volume is
infinitesimal so that the value of pu at its centre is the same as the average value ofpz:t
over the area oy d. The product pu will in general vary over the entire flow regime and
have different values at the two end faces of the framework. Following the approach
used for pressure variations in the derivation of Euler's equations, the mass flow in
the x direction through the face nearest the origin is approximately
ox) us:y d
tonnes/second
a(pu) ox) uy
s: d
( pu+ax 2
tonnes/second
(pu
a(pu)
ox
Hence the net mass flow into the framework through these two faces is
34
Fluid dynamics
[Ch.2
0~~x~
 CJ(pu) bx by d
ax
tonnes/second
The same approach is used to find the net mass flow into the framework through the
other pair of faces:
 CJ(pu) bx by d
CJy
tonnes/second
 (
a~xu) + a~v))
bx by d
tonnes/second
Sec. 2.4]
35
changes. In this case the rate of increase of the mass of fluid inside the framework is
(op/cU) ox oy d tonnes/second and this must balance the net mass flow through the
faces into the framework. Hence
op
ot
ox oy d = 
(o(pu)
ox
+ o(pv))
oy
ox oy d
tonnes/second
op
ot
= 
ox
oy
(2.8)
which is valid for both real and ideal fluids. If the density is constant (i.e. the fluid is
incompressible) this reduces to
ou ov
ox uy
 + ; =
seconds 1
(2.9)
2.4
The flow of an ideal fluid can be described in terms of the velocity potential. This is a
function <l>(x, y, t) which has some value everywhere in a fluid flow and varies such
that the velocity components are given by its partial derivatives:
u =
v 
o<l>
ox
o<l>
oy
metres/second
(2.10a)
metres/second
(2.10b)
I'
[Ch. 2
Fluid dynamics
36
_ _ _ _ _ _o
~e
til
til
Is
Section at AA
0~~x
Fig. 2.5 Velocity potential in a twodimensional fluid flow.
2.6. If the velocity potential at A (r, 8) is <P and the velocity potential at
B (r +or, 8 + o8) is <J> + o<j>, it follows that
<P + o<J>
o<J>
o<J>
<P+ox+ oy
ax
oy
o<J>
o<J>
<P +or+oe
ae
or
metres 2/second
or
ax
Oy
Or
ae
Now
ox
oy
so
r oe cos e+ or sin
e metres
or cos e r oe sin e metres
Sec. 2.4]
37
br Cos II
bx
(i
metres 2/second
I'
o<J>
or Dr+ oe De
metres 2/second
Separating radial and tangential components we obtain for the radial velocity
u,
= u sine+ v cos e
o<J>
or
z d
metres /secon
(2.1la)
38
Fluid dynamics
[Ch.2
2.5
= U COS 8  V sin 8
1 aq,
= ;. ae metres2/second
positive anticlockwise
(2.11b)
An analogous force potential n may also be defined for the externally applied forces
so that
X  ClQ
ax
metres/second 2
= an
metres/second 2
ay
(2.12b)
au
ay
av
ax
a
axay
2
<j>
=  =
second 1
(2.13)
Substituting the 'x' equations (2.10a), (2.12a) and (2.13) into Euler's equation of
motion for the x direction (equation (2. 7a)) we obtain
an 1 aP
au av a2 <j>
=
u
ax + v ax+ axat
ax pax
metres/second
(2.14)
Now
(v2
v av = ~
ax ax
metres/second 2
(2.15a)
metres/second 2
(2.15b)
2
)
~
(u + v + aq, n + ~)
=
ax 2 2 at
p
2
metres/second 2
(2.16)
39
if pis assumed to be constant (i.e. the fluid is incompressible). Equation (2.16) may
now be integrated to give
qz + o<J> ~~
n + _P  Fx (y, t)
me t res z;secon dz
2
at
p
(2.17a)
o<J>
qz
+Q+2
at
p
(2.17b)
where Fy is not a function of y. Since the lefthand sides of equations (2.17a) and
(2.17b) are the same, it is clear that the functions Fx and Fv must be identical and have
the same values at all positions and that they are therefore functions only of time.
Hence the integrated form of Euler's equations of motion for an in viscid incompressible fluid is
qz
o<J>
at
+Q+
F(t)
metres 2/second 2
(2.18)
Now equation (2.10) shows that the velocities are functions only of the derivatives of the velocity potential and not of the potential function itself. It follows that
any arbitrary constant may be subtracted from the potential function without
affecting the velocities. The quantity
J~ F(t) dt
metres 2/second
I'
has the same value everywhere in the fluid at some arbitrary timet and may therefore
be regarded as a constant. We may define a new velocity potential
<P'
<P J~ F(t)
dt
metres 2/~econd
at
at
40
Fluid dynamics
[Ch. 2
qz aq,'
P
+Q+
at
0 metres 2/second 2
qz aq,
P
+Q+
at
0 metres 2/second 2
(2.19)
If the flow is steady (i.e. no variations with time) equation (2.18) reduces to
q2
 Q
+ = F
p
metres 2/second 2
(2.20)
where F is now an arbitrary constant. If there is no external applied force the force
potential Q is constant and the equation reduces to the well known form
P+
pi
= Pstag
kN/metre 2
(2.21)
where Pstag is another arbitrary constant. If the fluid is brought to rest at some point
so that q = 0, the pressure will be
P
and
2.6
Pstag
= PStag kN/metre 2
LAPLACE'S EQUATION
Substituting equations (2.10) for the velocity potential in the continuity equation for
an incompressible fluid (equation (2.9)) leads to the Laplace equation
0 seconds 1
(2.22)
which must be satisfied at every point in an ideal fluid flow. A potential function
which satisfies Laplace's equation will therefore describe some ideal flow of an
inviscid incompressible fluid.
Sec. 2.7]
2. 7
41
Consider a small rectangular element of sides 8x and 8y and depth d metres in the
twodimensional flow field shown in Fig. 2. 7. The volume flow through the element is
'I'=
(u dy v dx)
u 8y d
metres 3/second
I(!
v 8x d metres 3/second
in the positive y direction. So the total volume flow across the line OS 1R is
d (u dy v dx)
I'
metres 3/second
\v
metres 2/second
(2.23)
is the volume flow across the line OS 1 R per unit depth of fluid.
Consider now a second line OS 2 R joining the origin to the point R. Provided that
42
Fluid dynamics
[Ch. 2
no fluid is created or destroyed in the space between the two lines (i.e. the equation
of continuity (2. 9) is satisfied) the value of \jl will be unchanged. \jl is therefore a
function only of the position of Rand is independent of the path of integration. \jl is
called the stream function.
Differentiating equation (2.23) gives
d\jl
= u dy v dx
d\11
metres 2/second
but
't'
a\jl dx + a\jl dy
ax
ay
metres 2/second
u =
~;
metres/second
a\jl
v 
metres/second
ax
(2.24a)
(2.24b)
Referring to equations (2.10) we see that the stream function is related to the velocity
potential by
a\jl
a<j>
ay = ax
metres/second
a\jl
a<j>
ax
ay
metres/second
(2.25a)
(2.25b)
Lines on which \jl is constant are called streamlines, and the fluid flows along these
lines when the flow is steady. Streamlines intersect lines of constant potential at right
angles.
In the polar coordinates defined in Fig. 2.6 the stream function is related to the
tangential and radial velocities by
u0
a\jl
a,
ur
1 a\jl
 ~ ae
metres/second
metres/second
(2.26a)
(2.26b)
Sec. 2.8]
2.8
43
USin X
~
,..
'
UCosx
u
v
= U sin x
= U cos x
metres/second
metres/second
, ii
and the stream function at any point R is, from equation (2:23)
\jl
Uy sin
x Ux cos x
metres 2/second
(2.27)
I'
u  U sin
x = ~!
U cos X
acp
ay
metres/second
metres/second
44
Fluid dynamics
[Ch.2
ing that it represents a valid flow of an ideal fluid. Fig. 2.9 shows streamlines and
equipotential contours for a uniform stream derived from equations (2.27) and
(2.28).
2.8.2 Sources and sinks
Fig. 2.10 illustrates a point source at the origin in the xy plane. The source may be
visualised as a narrow tube with porous walls extending over the depth of the fluid.
Fluid is created within the tube at a rate md metres 3/second and the source strength is
said to be m metres 2/second. At some radial distance r metres the fluid recently
created by the source is flowing outwards at a rate md metres 3 /second across a
cylindrical boundary of circumference 2rrr metres and depth d metres. The radial
velocity is therefore
ur
md
2rrrd
2rrr
metres/second
where
= u, sine =
mx
2rrr 2
my
a<jl
~;
ax
metres/second
a'I'
ax
a<P
ay
metres/second
Integrating, we obtain the stream function for a point source at the origin:
'I'
(~)
m tan 1
2rr
x
m (rr/2  e)
2rr
metres 2/second
(2.29)
m
2
rr loge r
metres 2/second
(2.30)
Streamlines and equipotential contours for a point source are shown in Fig. 2.11.
As expected, the streamlines radiate outwards from the source and the equipotential
contours are concentric circles centred on the source.
A sink is simply a source with a negative strength. Fluid is therefore drawn into
the source at a rate md metres 3/second and the stream and potential functions are
Sec. 2.8]
45
y
I'
m(rr/2 e)
2rr
metres 2/second
(2.31)
(2.32)
Note that the equation of continuity is violated at the source (or sink) but is valid
elsewhere in the flow.
46
Fluid dynamics
[Ch. 2
Fig. 2.11 Streamlines and equipotential contours for a point source at the origin in the xy
plane.
21T
m
_ 1 ( Y)
tan 1 Y) tan
(x +S
21T
X S
metres 2/second
47
Sec. 2.8]
Now
tan 1
tan  1 a  tan 1 b
ab)
( 1 +ab
metres 2/second
If we now choose to move the source and sink towards the origin and at the same time
to increase their strengths in such a way that the product M = 2ms metre 3 /second
remains constant, we may write for the stream function
\jl
My
M cos
metres 2/second
(2.33)
M sin 9
'
metres 2/second
(2.34)
The combination of a source and a sink at the origin is called a doublet or a dipole.
Fig. 2.13 shows the streamlines and equipotential contours associated with a dipqle
aligned with the x axis. These are all circles centred on they and x axes respectively.
The centres of the circles are at (0, M/27T\jl) and (M/21T(j>, 0).
In a similar way it can be shown that the stream and potential functions for a
dipole aligned with the y axis are given by
\jl
<1>
Mx
27T y(xz + yz)
My
27T y(xz + yz)
=
"
M sin 9
metres 2/second
27Tr
M cos
21Tr
metres 2/second
(2.35)
(2.36)
48
Fluid dynamics
.....
[Ch. 2
'
'
X
,/
Fig. 2.13 
2.8.4 Multipoles
Sources, sinks and dipoles are often termed singularities because their stream and
potential functions adopt infinite values at the origin. The potential functions satisfy
Laplace's equation everywhere except at the singularity itself and these functions
therefore represent valid flows of an ideal fluid.
These particular singularities may be visualised in physical terms. As an example
we have already seen that a source may be imagined as a porous tube from which
fluid flows in all directions. However, this is not a necessary requirement for a valid
potential function. A wide range of singularities exist which represent no simple
physically understandable flow. Yet they still represent a valid flow of an ideal fluid
provided that the potential function satisfies Laplace's equation (2.22). Multipoles
fall into this category. The stream and potehtial functions for multipoles aligned with
the x axis are defined as
\jl
M cos (me)
"'"..,.m
metres 2/second
m = 1,2,3, ...
(2.37)
metres 2/second
(2.38)
~,,
M sin (me)
21Trm
<l>
1,2,3, ...
and the corresponding functions for multipoles aligned with the y axis are
\jl
M sin (me)
21Trm
metres 2/second
1,2,3, ...
(2.39)
Sec. 2.8]
49
<!>
M cos (m9)
2rrrm
metres 2/second
1,2,3, ...
(2.40)
r  
'
',
IL
_____ j
m=3
m=1 (DIPOLE)
,
: j '
I__ _
________ _j
m=4
m=2 (QUADRUPOLE)
'
2.8.5
The stream and potential functions for a uniform stream of velocity U metres/second
parallel to the x axis are, from equations (2.27) and (2.28),
'l' = Uy
metres 2/second
[Ch.2
Fluid dynamics
50
<j>
= Ux metres2/second
\jl
U
My
y  21T(x2 + y2)
<1>
Ux+
Mx
21T (xz + yz)
metres 2/second
(2.41)
metres 2/second
(2.42)
..
<!>INCREASING
I
I
:JJ
)>
(fl
G)
=: :
..,
Now we have seen that the lines on which the stream function is a constant are
streamlines. Since in steady flow the fluid flows along these lines, any one of them
~ould be replaced by a solid boundary of the same shape without altering the
characteristics of the rest of the flow. If we choose the particular streamline
Sec. 2.8]
"' =
51
metres2/second
metres
(2.43)
or
M
metres 2
21TU
x 2+y 2
= 
(2.44)
Equation (2.43) is the x axis and equation (2.44) is the equation of a circle of radius
~ C:u)
(2.45)
metres
and we can see that equations (2.41) and (2.42) represent the flow around a circular
cylinder in a uniform stream. For simplicity only streamlines and equipotential
contours outside the circle are shown in the illustration of the flow in Fig. 2.15. The
velocities are obtained by differentiating equation (2.41) or (2.42):
o\jl
oy
o<j>
ox
M ( x
metres/second
(2.46)
'll
0\j/
':\x
"
o<j>
ay
Mxy
( 2 2) 2 metres/second
1T x + y
,
I
(2.47)
At the points Az( r, 0) and Bz(r, 0) the velocities are zero, and these points at
the front and rear of the cylinder are called stagnation points. As explained in syctipn
2.5 the pressure is then equal to the stagnation pressure of the fluid.
'
At Cz(O,  r) and Dz(O, r) on the top and bottom of the cylinder the velocities are
= 2U metres/second
v = 0 metres/second
So the velocity at these points is twice the freestream velocity and, as would be
expected, parallel to the x axis. The pressure at these points is a minimum given by
equation(2.21):
Fluid dynamics
52
2.9
[Ch.2
CONFORMAL TRANSFORMATIONS
The solution for the potential flow around a circular cylinder in a uniform stream is
but one example of the many ways in which complicated flows can be synthesised by
adding the stream and potential functions for elementary flows. Many such solutions
can be built up by suitable combinations of sources, sinks and uniform streams.
A powerful additional technique allows a wide range of further flow patterns to
be derived from these basic synthesised solutions. The method involves mapping the
streamlines and equipotential contours of a known flow into the streamlines and
equipotential contours of the required flow using some suitable mapping function.
In order to exploit the method we redefine the xy plane of our known flow
solutions (for example, the flow around the cylinder in the uniform stream) as the
complex plane z such that
z
= x + iy
metres
w(z) = Uz sin
x to the y axis
x iUz cos x
metres 2/second
(2.48)
w(z) =
m
2
loge
1T
lzl
metres 2/second
(2.49)
w(z) = 
m
1T loge
2
lzl
metres 2 /second
(2.50)
w(z)
M
21TZ
metres 2/second
(2.51)
Conformal transformations
Sec. 2.9]
53
w(z)
The velocities
function:
uz
and
Vz
metres 2/second
(2.52)
o<Pz + 1. 0\jfz
dw(z)
dz
ox
ox
metres/second
or
dw(z)
dz
a (<Pz + i'Jfz ) =  1
. o<J>z +0\jfz
ay ay
 i
Uz  ivz
~y
metres/second
metres/second
(2.53)
s = Xsz + ix
83
'!'I
metres
 
.~~~~
~=f(z)
Z Plane
__. _...
..:::
~Plane
z=F(~)
ixs3
iy
Fig. 2.16 Mapping a shape in the z plane into another shape in the L, plane.
Fluid dynamics
54
[Ch. 2
= f(z)
metres
F(~)
metres
relating all points in the z plane to corresponding points in the~ plane. We require
the function to map all points on the surface of the body in our known solution in the
z plane (for example, the circular cylinder) onto the surface of the body about which
the flow is required in the ~ plane. Finding a suitable function to map a simple
geometric shape in the z plane into an arbitrary shape in the ~ plane may be a
formidable problem. However, a wide range of shapes can be produced and it is
often possible to achieve a reasonable approximation to a desired shape using
reiativeiy simpie mapping functions.
Let us suppose that a suitable mapping function has been found. Then for any
point in the ~ plane we may calculate the numerical value of the complex potential
w(z) at the corresponding point in the z plane. Let us call this W(~) where
W(Xsz + ixs3)
W(~)
w(z)
~uX 82
(.+.
. ) _
'+'z + 1\jfz 
I~ux 83
(,~.
'+'z
. )
1\j/2
d
metres1secon
aq,z
~
metres/second
(2.54)
uXB3
Comparing equations (2.54) with equations (2.25) it is evident that the complex
potential w(z) calculated in the z plane must also represent some valid fluid flow in
the~ plane.
Consider the value of the complex potential on any streamline (including the
surface of the body) in the z plane. Since the stream function is constant everywhere
along a streamline the complex potential must be of the form
w(z)
Since w(z) has the same value at corresponding points in the~ plane it is clear that the
Conformal transformations
Sec. 2.9]
55
streamlines in the z plane must map into streamlines in the plane. By a similar
argument equipotential contours in the z plane must also map into equipotential
contours in the plane.
The velocities in the plane are obtained by differentiating the complex potential
with respect to
s
s:
dz dw(z)
dz
dw(z)

ds
ds
ur, 
ivr,
metres/second
(2.55)
metres
(2.56)
where the coefficients a0 , a1 and a3 are real. Choosing different values for these
coefficients allows the flow around the circular cylinder in the z plane to be mapped
into the flow around a wide variety of shapes in the plane. We shall meet this
mapping function again in Chapter 11 where we shall see that it is used to obtain the
solution to the flow around shapes of shiplike crosssection. The coefficients are
then required to have particular values to ensure that the mapped shape in the plane
approximates to the required hull crosssection. However, f9 r the time being let us
choose the values
'
s = z + alz
z = ia e;e metres
z=
ia e;e metres
I,
Fluid dynamics
56
lzl = a
[Ch.2
metres
where
~ (2~U)
metres
(equation (2.45))
8 metres
at
a+
metres.
a '
Fig. 2.17 shows the streamlines and equipotential contours obtained from this
mapping function for the particular case of
4a
3 '
2a
3
metres
M)
2
d ( Vz+2  
a 1 dz
21Tz
57
Conformal transformations
Sec. 2.9]
<1>
=
Increasing
metres/second
'
(2.57)
1!
The front and rear stagnation points Az and Bz on the circular cylinder map into the
corresponding points As and Bs on the ellipitical cylinder (Fig. 2.17). Putting z = a
in equation (2.57) yields us ivs = 0 metres/second, so that these points are ,also
stagnation points in the ~ plane. Similarly the points Cz and Dz on the circular
cylinder, which experience the maximum velocity 2U metres/second in the z plane,
map into the corresponding points Cs and Ds on the elliptical cylinder. Putting
z = ia metres in equation (2.57) and equating real and imaginary parts gives
u =
v
3U
metres/second
0 metres/second
for the velocities at Cs and Ds. The velocities at any point in the~ plane may be found
by this method.
Fluid dynamics
58
2.10
[Ch. 2
VISCOSITY
Fig. 2.18 Growth and separation of boundary layer (boundary layer thickness exaggerated).
front stagnation point the boundary layer will be laminar with a smooth well ordered
structure and the velocity profile shown in Fig. 2.19. (x andy are here taken as
0.8
0.2
0.4
0.6
0.8
1.0
.!!_
u'
Sec. 2.11]
59
parallel and normal to the local body surface.) At the transition point eddies will
begin to grow in the boundary layer and the velocity distribution changes to the
turbulent profile also shown in Fig. 2.19.
The shear stress applied by one layer of fluid moving over another is given by
't
du
f..tw dy
kN/metre 2
(2.58)
where f..tw is the coefficient of viscosity. At the surface this appears as a frictional force
on the body. Clearly the force applied to the body by the turbulent boundary layer is
much greater than that applied by the laminar boundary layer.
Outside the boundary layer the velocity gradient is very small and the viscous
forces are negligible. So potential flow methods can be applied provided that the
boundary layer is relatively thin. This is generally true over the forward portion of
the body where there is a favourable (falling) pressure gradient. This helps to
minimise the growth of the boundary layer and keeps the flow firmly attached to the
body surface. However, the pressure gradient beyond the maximum diameter of the
body is adverse (rising) which tends to slow the flow and leads to a rapid thickening of
the boundary layer. At some stage the velocity gradient at the surface may become
zero as shown in Fig. 2.18. Beyond this point flow reversal occurs and the boundary
layer is said to separate. Largescale eddies which are not predicted by potential flow
methods will then occur. Predictions of forces based on potential flow may then be in
error.
Flow separation may occur whenever there is an adverse pressure gradient on
bodies with tapering tails. The more rapid the taper, the more likely is separation. In
particular, separation is virtualiy guaranteed at any discontiJ,1~ity or sharp corner on
the body surface.
'
2.11
Ship hulls are usually fitted with appendages such as rudders, propyller shaft brackets
and roll stabiliser fins. These can influence the behaviour of the ship in rough weather
and we shall require a method of estimating the forces developed by them. These
forces may be estimated using potential flow methods based on the techniques
described above, but results of adequate accuracy can be obtained using the simple
empirical formulae given below.
Consider the typical lifting surface appendage shown mounted on the hull in Fig.
2.20. The geometry of the surface is conveniently defined by the root and tip chords
cr and ct and the outreach b. The mean chord is
metres
Fluid dynamics
60
[Ch. 2
c,
c,
(2.59)
2b
4b
(2.60)
When the surface is at an angle of incidence ex to the incident flow it will generate a
lift force L and a drag force D. These forces are respectively normal and parallel to
the direction of the incident flow.
For a given angle of incidence and planform shape the lift and drag are found to
be proportional to the square of the forward speed and the planform area. So the lift
and drag may be expressed in nondimensional terms as
Sec. 2.11]
61
(2.61)
D
!pU 2A
(2.62)
Stall
c
Q)
'()
~
Q)
0
.::=
:.::;
Incidence angle a
For small angles of incidence the lift coefficient increases more or less linear'ly
with the angle of incidence and we may write the lift coefficient as
(2.63)
where dCdd~ is the lift curve slope.
The slope of the curve diminishes 'as the angle of incidence is increased and
maximum lift occurs at the stall angle ~stall The lift curve slope increases with the
aspect ratio, but surfaces of high aspect ratio stall earlier and more abruptly than
those of low aspect ratio. TQ.e lift characteristics of symmetrical sections are only
weakly dependent on the section shape.
Fluid dynamics
62
[Ch. 2
Whicker and Fehlner (1958) tested a variety of lifting surfaces of low aspect ratio
such as are typically employed on ships and derived an empirical formula for the lift
curve slope as a function of aspect ratio:
l.81Ta
d'
1
1. 8 + y'(a 2 + 4) ra tans
(2.64)
Fig. 2.22 shows this formula plotted for rectangular lifting surfaces together with
Aspect ration a
Fig. 2.22 Lift curve slope. (After Whicker and Fehlner (1958).)
illustrations of the surface planforms associated with various aspect ratios. Clearly
the lift curve slope increases dramatically with increasing aspect ratio: in other words
long slender lifting surfaces (like a glider~s wings) are much more effective than short
stubby surfaces.
Whicker and Fehlner also reported the stall angles found for their lifting surfaces.
These are given approximately by
1.05 0.445a + 0.075a 2
0.39
radians
radians
for a>3.0
for a <3.0
(2.65a)
(2.65b)
If the angle of incidence approaches 90 degrees the lift force (normal to the flow
Sec. 2.11]
63
direction) becomes zero. The drag force is then very large and acts normal to the
plane of the surface. For this case Hoerner (1965) gives
C0
1.17
(2.66)
'II
r
I,
3
Regular waves
3.1
INTRODUCTION
The waves which influence the behaviour of ships at sea are generally irregular and
more or less random in nature. No two waves have exactly the same height and they
travel across the surface at different speeds and in different directions. Techniques
for coping with the chaotic nature of the real sea surface are described in Chapter 4,
but it is first necessary to discuss the characteristics of ideal regular waves.
Such waves never occur in the real ocean environment although they can be
produced in laboratory towing tanks and form the basis of many seakeeping model
experiments. Of equal importance is the fact that the theory of irregular waves is
based on the assumption that they can be represented by 'superposing' or adding
together a suitable assembly of regular waves. So it is clear that the characteristics of
regular waves have a profound influence on the behaviour of ships in rough weather
even though they are never actually encountered at sea: an understanding of their
nature is one of the vital tools in the study of seakeeping.
Fig. 3.1 shows a train of regular waves advancing across the surface of a body of
water of constant depth d. The waves are twodimensional: that is, they advance in
the x direction and the crests are perpendicular to the x axis. The crests may be
considered as extending to infinity on either side of the x axis; alternatively the waves
may be imagined to be advancing down a long narrow tank bounded by vertical walls
parallel to the x axis;
The saiient characteristics of the waves are:
~
~0
'A
c
the instantaneous depression of the water surface below the mean level
the wave amplitude or vertical distance from the mean level (y=O) to a crest or
a trough; ~ 0 is always positive
the wave height: twice the wave amplitude
the wave length: the horizontal distance (in thex direction) between one crest
(or trough) and the next
the wave celerity: the velocity of an individual crest in the x direction
Sec. 3.2]
65
cxo
RIA.
the wave period: the time interval between successive crests (or troughs)
passing a fixed point
the instantaneous wave slope: the gradient of the surface profile (in radians)
the maximum wave slope (in radians)
the wave steepness.
'ri
These waves progress across the surface in a regular orderly fashion. Each wave
crest advances at the same steady velocity c so that the waves never overtake each
other and the wavelength A. and period Tremain constant. The shape of each wave
remains the same and the whole wave train appears to advance like a rigid corrugated
sheet. Fortunately for the student of seakeeping, the characteristics and deta~led 1
structure of regular waves are very well predicted by the technique of classical fluid'
mechanics outlined in Chapter 2. In common with the treatment of many other
amenable flows it is necessary to assume that the water is incompressible and inviscid
in order to obtain a workable solution. This does ,not imply that water is actually
incompressible or inviscid: merely that the values of compressibility and viscosity are
such that they have little discernible influence on the characteristics of regular waves.
We shall also assume that the effects of surface tension are negligible. This restricts
the validity of the solutions to wavelengths greater than about 0.1 metres.
Regular waves
66
(Ch. 3
to satisfy Laplace's equation (2.22) could be formulated and each would describe
some flow of an ideal in viscid incompressible fluid. The choice of a potential function
to describe some particular flow is a matter of considerable mathematical skill and
insight coupled, no doubt, with a good deal of trial and error. Lamb (1932) showed
that the potential function
.~. _g(, 0 cosh(k(d y)]
(k _ )
ro
cosh( kd)
cos x rot
'+' 
(3.1)
=0
metres/second
( ~<P)
y
=0
metres/second
y=d
q
2
+ a<j) n + ~ = 0
at
metres 2 /second 2
(3.2)
must appiy everywhere and can be used to find the surface profile associated with the
velocity potential given by equation (3.1). The only force applied externally to any
fluid particle is gravity. Hence, from equation (2.12),
X= an = 0 metres/second 2
ax
an = g
Y =
ay
metres/second2
67
Sec. 3.3]
Hence
Q=
gy metres 2/second 2
q2 aq,
P
 +   gy +  = 0 metres 2/second 2
2
at
(3.3)
P= pgyP kN/metre 2
and a constant pressure contour is a horizontal straight line. Under regular waves this
contour is distorted as shown in Fig. 3.2. The depth of a point on this contour is
y
= yP + Sp metres
Surface profile
I
Constant pressure contour
~P~ogy,
y
I'
Since the pressure everywhere along the contour and the depth yP are both
constant, the quantity
Regular waves
68
J~ (~gyP) dt
[Ch.3
metres /second
will be a constant on the contour at any given timet. It may be added to the potential
without affecting the velocities in any way (since they are functions of the potential
gradients and not of the potential itself). So we may define a new velocity potential
<j>'
= <j> +
J~ (~gyP) dt
metres 2/second
so that
or, since
Sp is small,
(3.4)
Substituting the expression for the velocity potential (equation (3 .1)) yields the
equation for the constant pressure contour at depth metres:
yP
Sp so
cosh[k(d Yp)] .
_
cosh(kd)
sm (kx rot)
metres
(3.5)
69
Sec. 3.3]
This and many other expressions which follow can be simplified by using the
following approximations:
(a) for water depth greater than about half the wave length
cosh[k(d Yp)] = sinh[k(d Yp)] = cosh[k(d Yp)] = sinh[k(d Yp)] =ex ( _ k )
cosh(kd)
cosh(kd)
sinh(kd)
sinh(kd)
p
Yp
(3.6)
tanh(kd) = 1.0
(b) for water depth less than about 0.03 times the wavelength
cosh[k(d Yp)] =
10
cosh(kd)
'
(3.7)
sinh[k(d Yp)] _ d Yp
sinh(kd)
d
tanh( kd) = kd
So the constant pressure contour in deep water is given by
Sp
metres
'tl
(3.8)
'I
metres
(3.9)!,
I
70
Regular waves
[Ch. 3
K/1\A ld\f\o,
~p
I
I
' l'
VVlI
:
I
~0.25
".../
/" /'.I
v
v lI
I
Vvl
I0.5
I
r., 0.75
'11.0
Depth 10m
Depth 2m
Depth 100m
Fig. 3.3 Constant pressure contours under a 100 metre long wave in various depths of water.
(3.10)
metres
so
which is the equation of a regular wave of small amplitude metres advancing across
the fluid surface. This result is independent of water depth. So the chosen velocity
potential function does indeed represent the fluid flow associated with a regular
wave.
Equation (3.10) is illustrated in Fig. 3.4. Consider first the wave shape in the
geographical or spatial sense. This is tantamount to fixing time at some instant t as,
for example, when taking a photograph. If for simplicity we choose t = 0, equation
(3.10) is reduced to
s=so sin(kx)
metres
= 2rr metres 1
A
(3.11)
71
Sec. 3.3]
t=t2
x,
),
kx,
6)
so
t, t2
! '
and the term rot1/k (metres) can be recognised as a phase 'lag' which governs the
location of the wave along the x axis. As time passes, the lag increases and the wave
advances steadily away from the origin with velocity
(l)
c =k
metres/second
(3.12)
72
Regular waves
[Ch. 3
resulting sine wave of amplitude /,; 0 metres and period T seconds is also illustrated in
Fig. 3.4 and the frequency ffi is related to the period by
ffi
2
;
radians/second
(3.13)
k~ 1 ) J
(3.14)
metres
and the term kx 11ffi (seconds) may now be recognised as a phase 'lag' which governs
the temporal location of the sine wave on the taxis.
3.4
WAVE SLOPE
It is sometimes convenient to quantify the effects of the waves in terms of their slope
rather than their elevation or depression below the mean level. The slope of the
pressure contours may be obtained by differentiating equations (3.5), (3.8) and (3.9)
as appropriate.
For any depth yP
~a:Yr
dx  k/,; 0
cosh(dyp)
_
cosh(kd) cos(kx ffit)
radians
(3.15)
cos(kx ffit)
radians
(3.16)
= k/,; 0
cos(kx  ffit)
radians
(3.17)
The wave slope at the surface is obtained by setting yP = 0 metres to give, for any
water depth,
radians
(3.18)
= k/,; 0
radians
(3.19)
So the wave slope varies sinusoidally in both time and space in much the same way as
73
Sec. 3.5]
the surface elevation. Fig. 3.5 shows the surface profile and the corresponding wave
slope at some timet= t1 The wave slope is a maximum when the surface elevation is
zero and vice versa.
(J)t,
~\
1\
~0
\ "__/
'
I,
and this is compared with the corresponding surface elevation time history (equation
(3.14)) in Fig. 3.6. The wave slope lags the surface depression by a quarter of a wave
period.
3.5
74
Regular waves
[Ch. 3
kxl
"'
.. t
(see equation (3.4)). Substituting equation (3.1) for the velocity potential we obtain
the relationship between the wave frequency and the wave number for regular waves
of small amplitude 1;,0 in water of any (constant) depth d:
co= y[gk tanh (kd)]
radians/second
(3.20)
c=
~ (f tanh (kd))
metres/second
(3.21)
Combining equations (3.5)(3.21) in various ways leads to a multitude of relationships between wave frequency, period, number, length and celerity, and these are
listed for easy reference in Table 3.1. The deep water approximations are usually
considered adequate for most practical purposes.
Particle orbits
Sec. 3.6]
75
Figs 3.7 and 3.8 show how the wave period and celerity vary with wavelength for
various water depths. As might have been expected, long waves have very low
frequencies and vice versa. As if to compensate for this the celerity increases with
wave length: for example, the celerity of a 1000 metre wave is almost 40 metres/
second (over 75 knots) in deep water compared with only about 4 metres/second
(about 8 knots) for a 10 metre wave. This dependency of celerity on wavelength
distinguishes surface waves from some other forms of wave motion (notably
electromagnetic radiation) and we shall see that it is responsible for some peculiar
properties of waves on the sea surface.
In shallow water the wave celerity is reduced and the 1000 metre wave's deep
water celerity is almost halved in water of 50 metres depth. In very shallow water the
celerity becomes independent of wavelength and depends only on the water depth:
c=\l(gd)
o<!>
.
u = ax =  u0 sm (kx rot)
a<j>
V = oy =  Vo
COS
metres/second
(3.22)
'ilI
(3.23)
metres/second
(3.24)
I'
v0 = 0
1.
(3.25)
These relationships may again be simplified for the two special cases of deep and
shallow water. For deep water
!fo = Vo = ro'C,o exp (  ky)
metres/second
(3.26)
(3.27)
Regular waves
76
[Ch.3
o
c
0
(.)
Q)
"'c
();
.~
o
~
>
(.)
c
Q)
:::l
0"
Q)
>
ctl
0
1
8 10
20
40
60
200
1000
400
40
c=
~[~tanh(2~d)J
'0
c
0
(.)
Q)
"'~
a;
E
~
20m
Q)
10m
5m
8 10
20
40
60
100
200
Quantity
In terms
of
Any depth
Deep water
(d > O.SA)
Shallow water
(d<0.03A)
C/l
~
!"">
(.;J
(J)
21T
T
21T
T
y'(gk)
k y'(gd)
~e~g)
21T y'(gd)
A
21T
T
(J)
..,"'=
21T
21T
Ill
21T
(J)
(J)
(J)
21T
y'[gk tanh(kd)]
21T
y'(gk)
21T
k y'(gd)
~e;A)
A
y'(gd)
r.;
;
..,
s:
"'
27T'lt
A
~ g tanh e~d)
2c
1Tg
.l
.l
Quantity
In terms
of
Any depth
Deep water
(d > 0.5A)
(!)2
(!)
Shallow water
(d<0.03A)
(!)
V(gd)
41T2
gT2
21T
T\f(gd)
21T
A
21T
A
00
21T
A
(JQ
=
S'
..,
g
c
c2
~
D:l
~
..,
(!)
21Tg
0)2
gT2
21T
21T
k
21T
k
21TC 2
g
21rV(gd)
(!)
T\l(gd)
21T
k
Q
(jJ
Quantity
In terms
of
Any depth
Deep water
(d>0.5A)
Shallow water
(d<0.03A)
Vl
(b
(')
V)
(l)
(l)
gT
21T
~(~tanh (kd))
~(f)
~[~~tanh (2 ~d) J
~(~~)
2
mso
y(gd)
41Tzso
gTz
21Tso
Ty(gd)
ffi
(l)
e.
y(gd)
<Xo
.,..,

So
g
kso
kso
kso
21Tso
A
21Tso
A
21Tso
A
gso
a.
tD
..,
Q
8:
.....
"'
.I
\0
Quantity
Uo
In terms
of
Deep water
(d > 0.5A)
Any depth
Shallow water
(d < 0.03A)
00
0
(J)
21Tso
(  41T Y)
T
exp gf2
IJQ
e..,
2
so (21Tg) 112 cosh ( 1T( dA y))
s0
~ (21Tg)
exp (21Ty)
A
A
(J)
<
"'
so8
c
So~(~)
soro exp (  :Zy)
2
Ill
Vo
Ill
21Tso
(  41T Y)
T
exp gf2
soro (d y)
n::r
21Tso (d y)
Td
t.N
Quantity
Vo
Any depth
In terms
of
Shallow water
(d<0.03A)
Deep water
(d>0.5A)
So V(gk) exp(
ky)
(l)
!"">
v:>
2
so y(2rrg) cosh( rr(dA y))
A
2:so
~ (~) (d _y)
y A smh
.,""=
Col
e.
a.
gy)
.,
so exp( 2
c
c
~
.....
"'
Xo
so exp( :
0)
y)

:rg
So exp ( 4rr y)
s cosh[k(d y)]
0
sinh(kd)
so exp(
ky)
: ~(~)
~; ~(~)
So
kd
00
......
Quantity
Any depth
In terms
of
Xo
Deep water
(d>O.SA.)
Shallow water
(d<0.03A.)
00
N
21Ty)
soexp (A.
A.
. h(27Td)
SID

so"
27T
A.
s 0 1~xp (
7gy)
2
Yo
co
so e:xp( : y)
47T2y)
soexp (T

~
~
IJQ
.,=e.
~
so sinh[k(d y)]
sinh (kd)
k
s
=
~
"'
so exp( ky)
. h(27T (dy))
A
0 SID
A.
So
21Ty)
e~xp (A
. h(27Td)
SID

nP"
A.
so exp( Jy)
.I itUIII:: .J .I
Quantity
In terms
of
~ ~UllllllUt:U)
Any depth
Deep water
(d>0.5A.)
Shallow water
(d<0.03A.)
C/l
(b
!'1
v.l
Yo
Ua
(!)
So (d y)
d
2ffi
gT
41T
'"I
=.
k
A.
+A sinh(~~d)
4wd
12.
~(:k)
'"I
s:.....
"'
vg"'
81T
~(g:)
00
v.l
Regular waves
84
[Ch.3
Fig. 3.9 illustrates these formulae for shallow and deep water as well as for the
general case. The vertical velocity is always zero at the sea bed (since the bottom is
assumed to be impervious) and the motion is therefore purely horizontal at this
point. The horizontal velocity is generally greater than the vertical velocity except in
very deep water where the two amplitudes are everywhere the same. In shallow
water the horizontal velocity amplitude is constant but it generally increases with
height above the sea bed and assumes an exponential variation in very deep water.
Following our physical interpretation of the characteristics of the wave profile we
shall now adopt the same techniques for examining the geometry of the structure of
the flow at a particular location by fixing x andy and allowing time to proceed. Since
the wave amplitude is assumed to be small the velocity amplitudes given by equations
(3 .24) and (3 .25) must also be small and it follows that a particle of water oscillating
about some point (x, y) will never stray very far from that point. The path of the
particle can therefore be calculated approximately by assuming that it is always
subject to the velocities calculated for the point (x, y). With this assumption the
particle's trajectory is obtained by integrating equations (3.22) and (3.23) to give
Ax=  x 0 cos (kx rot)
~y
metres
(3.28)
metres
(3.29)
where Ax and ~yare the deviations of the particle from its datum position (x, y) and
the amplitudes of its displacements are
x = u0 = gks 0 cosh[k(d y)] =
0
Yo
ro
ro 2
s cosh[k(d y)]
0
cosh(kd)
sinh(kd)
s sinh[k(d y)]
0
sinh(kd)
metres
(3.30)
metres
(3.31)
Again the deep and shallow water approximations may be applied. In deep water
metres
(3.32)
Xo =
~~'
Yo= so (d y)
metres
(3.33)
Equations (3.28) and (3.29) represent an elliptical orbit with major (horizontal) axis
2x0 and minor (vertical) axis 2y 0 . Individual particles of water under a wave will
therefore follow elliptical paths as shown in Fig. 3.10. At the surface (y = 0) the
minor axis is always twice the wave amplitude but the orbits become flattened as the
85
Particle orbits
Sec. 3.6]
0.5
I
1.0
I
1.5
I
2.0
2.5
0.5
1.0
1.5
0.5
1Vo
0.25
Depth
100m
0.75
1.0
Depth 2m
y!d
0.5
Uo
Uo
Depth
2m
1.0
Depth 10m
Depth 100m
ffiEB~
cp
y
'tl
'I
~
.
'
_____...___________ 4=
__L __________ l
Bottom
111111111
ill
I.
sea bed is approached. At the bottom the particles merely oscillate to and fro with no
vertical displacement as the wave passes overhead. The major (horizontal) axes of
the orbits decrease as the water depth increases until the orbits become circular when
the water is very deep. In this case the orbit radius decreases very rapidly (exponentially) with depth below the surface: at halfthe sea depth the radius is only about 4%
of its surface value. Fig. 3.11 shows the relationship between the profile of a deep
water wave and the circular orbit of a particle at the surface.
86
Regular waves
[Ch. 3
f\~~Zt
Fig. 3.11 Orbit of a particle at the wave surface.
3. 7
The pressure at any point under a regular wave may be found from Bernoulli's
equation (3.3). If we assume that the velocity is small we obtain
P=pgy+P
so the pressure at any depth y metres oscillates around the steady hydrostatic
pressure pgy kN/metre 2 The fluctuating part of the pressure is
cosh[k(d y)] .
P =  pg(,0 cosh(kd) sm (kx rot)
kN/metre
(3.34)
(3.35)
(3.36)
Fig. 3.12 shows the variation of the pressure amplitude beneath a 100 metre long
wave for three different depths. In very shallow water the pressure amplitude is
constant and everywhere equal to the hydrostatic pressure associated with the
surface wave amplitude. In deeper water the pressure amplitude decreases with
depth and becomes negligible at the bottom in very deep water.
3.8
The energy associated with a train of regular waves includes contributions from both
potential and kinetic energy. Consider a small length of the regular wave shown in
Fig. 3.13. The surface depression(, is given by equation (3.10) and the mass of water
over the length is approximately  p(, per unit width of the wave. The centre of
gravity of this mass is approximately  S/2 metres above the undisturbed surface
level
ox
ox
ox
Sec. 3.8]
87
10
yld
Depth
2m
Depth
20m
''11.0
Mass/unit
width=Q~
bx
ox
and its potential energy relative to the undisturbed (calm water) state is lpgl,? per
unit width of the wave. If we now allow
to become infinitesimally small we may
integrate to obtain the total potential energy summed over a single wave length:
ox
Regular waves
88
EP =
J"!pg~2
0
pg~ii"
d.x =
pg~6
J" sin
2
[Ch.3
(3.37)
Consider now a small element of fluid beneath a wave as shown in Fig. 3.14. The
...
I.
I
~~r_.~~x
~~M
l
d
l:,J~u
.;.,<.~.....,       .
YI~~xv.~v
Bottom
mass of the element per unit width of the wave is p ox oy and it has a total velocity q
given by
(3.38)
So the kinetic energy of the particle is
!pq 2
ox oy
Sec. 3.9]
89
the total kinetic energy of the fluid in one wave length between the surface and the
bottom:
Ek = lp
J:(Jr~ q dy) dx
2
joules/metre
(3.39)
So the potential and kinetic energies are equal and the total energy in one wavelength
is
~2A.
E = pg 0 joules/metre
2
(3.40)
which leads to the remarkable result that the average energr per square metre of sea
surface is independent of the wave frequency and depedds only on the wave
amplitude:
E = pg~6 joules/metre2
2
3.9
(3.41)
The energy associated with a sequence of regular waves is transmitted along the
direction of their propagation. The rate of energy transmission can be found by
considering the energy flux across the plane AA in Fig. 3.15. We begin by calculating
the rate at which the fluid on the left of~ small element of height oy is doing work on
the fluid on the right of the element.
Since 'the element is small the pressure and velocity acting on its face may be
regarded as constant (at a given time) and the force exerted by the fluid on the left is
P oy kN per metre width of the element. The work done by the fluid on the left i:;
uP oy joules per second per metre width. If we now allow oy to become infinitesimal
the total rate of transmission of energy across the plane AA is obtained by integrating
90
[Ch. 3
Regular waves
A
X
u
y
Bottom
AI.
..
'._,. ':.::.~.5:>
:::'~..<.:~~<~
..,
.
=~~.
.~,..
. ' . .:::..;~~
..... .
over the depth of the fluid. Neglecting the small contribution due to the portion of
fluid above the undisturbed surface level (y = 0), the rate of transmission of energy is
E = Jd
uP dy
(3.42)
()
Using equations (3.22) and (3.35) we find that the energy is transmitted at a rate
_ pg
2
s_o_
k___
sinL(kx,_ J, rot) {d Y dy
1
W l:U:SH ~KU)
J ()
joules/(metre second)
. _ pg(,~c (
2kd ) . 2
E 2 1 +sinh (2kd) sm (kx rot)
pg2L,okd2
.
ro cosh (kd) sm(kx rot)
2
joules/(metre second)
Sec. 3.9]
91
The rate of transmission of energy evidently fluctuates with time but we are
concerned with its mean value. Over a long period of time (or an integral number of
wave periods) the mean value of sin 2 (kx rot) is! and the mean value of sin (kx rot)
is zero. So the mean rate of energy transmission is
'
pgs6c ( 1
2kd )
+ sinh(2kd)
E =4
joules/(metre second)
(3.43)
Now the total energy is given by equation (3.41) and this energy is transmitted at a
mean velocity given by
;.,t 0
c(
E
2kd )
= E = 2 1 + sinh ( 2kd)
metres/second
0~~_.~~~~~_._.~
0.01
0.02
0.04 0.06 0.1
0.2
d/i,
(3.44a)
92
Regular waves
2kd
[Ch.3
=0
sinh (2kd)
and
u0 =  metres/second
2
(3.44b)
In shallow water
2kd
sinh (2kd)
1.0
and
u0 = c metres/second
(3.44c)
For deep water we may interpret this result by considering the progression of a
group of regular waves down a laboratory tank. If the energy associated with each
wave length is E joules/metre the amplitude of the waves is, from equation 3.40,
so=~(:~)
metres
(3.45)
total energy per square metre within the group is kept constant.
At the leading edge of the group the first wave will be propagating into calm
water. So this orderly exchange of energy from wave to wave is interrupted and after
one wave period the energy of the leading wave is halved. The wave amplitude is
reduced and this process continues as the leading edge of the wave train propagates
down the tank at the wave celerity.
The leading edge of the group proper (defined as the position of the first wave of
full amplitude given by equation (3.45)) propagates down the tank at u0 metres/
second and this velocity is called the group velocity. Individual waves within the
group propagate at the wave celerity c, which in deep water is twice the group
velocity (see equation (3 .44b)).
4
Ocean waves
4.1
WAVEGENERATION
c>
0
~
(b) Ripples
~
(c) Small waves
94
Ocean waves
[Ch. 4
If the wind continues to blow for long enough and sufficient length of water or 'fetch'
is available, the ripples will advance and grow in length and height until they can
more properly be called waves. At the same time, the wind generates new ripples on
the surface of the growing waves and these ripples will eventually grow into waves
themselves. The process is of course continuous and the observed waves at any
particular place and time will consist of a mixture of wavelengths and heights
superimposed on each other.
The individual wave components apparently still behave in the same way as they
would in ideal conditions, uncontaminated by waves of other lengths. Thus the fast
moving long waves continually overtake the slow moving short waves and the shape
of the surface is changing all the time as the waves progress through each other.
Clearly the waves are absorbing energy from the wind. This energy absorption is
countered by two principal decay mechanisms: wave breaking and viscosity. If the
wind continues to blow at constant velocity for long enough and sufficient fetch is
available, the rate at which energy is absorbed by the waves will eventually be exactly
balanced by the rate of energy dissipation and a steady state 'fully developed' wave
system will be achieved. Such wave systems are rare because the required steady
conditions do not often presist for long enough and the fetch may be limited by the
local geography.
If the wind ceases to blow, the wave system it has created will gradually decay.
Since wave breaking is a relatively powerful decay mechanism, the short steep
waves, which are more likely to break, decay first, leaving the longer waves to be
dissipated by the relatively weak forces of viscosity. This decay process may last
several days, during which these fast moving long waves may travel several thousand
kilometeres and be recognised at some distant location as a swell. Swells are
generally of long period and comparatively regular. Locally generated wave systems
may therefore be contaminated by swells generated elsewhere. These swells will of
course bear no relationship to the local wind.
4.2
Whatever the complexities of local geography and the vagaries of the wind, an
observer at sea will see a confused (and confusing) pattern of ever changing wave
crests and troughs travelling in different directions. For many years this apparent
chaos (and the resulting unpredictable nature of ship motions) provided an insur_.......,. .... ...,..f.....,.\....1...,. ....1..n+.n...1....
UlUUUlaUH;;;
VU;)la\,J~
"LJ"\.'1.'1.70'1.7Q'I"
.L..I.VVY\,.IV\,.;.1'
~n
rOI"'OT"li
J.H .l\.1\,.;\.;UL
ua.r:arco
J""U.l.:>
Sa (metres)
Sec. 4.2]
95
TH
1.~~~~
T,
~(metres)
*Negative amplitude
'rl'I
Fig. 4.2 Typical wave record: analysis of peaks and troughs.
trough above the mean level gives a negative amplitude; otherwise amplitudes are always positive)
the vertical distance from a trough to a succeediog
peak or vice versa (always positive)
the time between two successive peaks
the time between two successive upward or downward zero crossings.
These individual measurements are unique to the particular part of the record
chosen for analysis and are of little use for characterising the whole time history. So it
is customary to describe the general ch;uacteristics of the complete time history in
terms of the mean values of these quantities:
~a
Ha
TP
96
Ocean waves
Tz
[Ch. 4
~113
H 113
ments of Sa (metres)
Ha (metres).
They are related as follows:
H 113
In addition to the statistical measures associated with peaks, troughs and zero
crossings there exists another class of measurements used to quantify the characteristics of an irregular wave record. Here the time history is sampled at discrete (short)
intervals of time to obtain successive measurements of the surface depression ~n
relative to some arbitrary datum as shown in Fig. 4.3. For a typical irregular wave
'
record an appropriate time interval would be 0.5 or 1.0 second.
Timet (seconds)
l;, (metres)
97
Fourier analysis
Sec. 4.3]
N
L~
(4.1)
metres
n=l
L (sn ~)
n=l
metres 2
cr0
(4.2)
= ym0 metres
(4.3)
Very long records of, say, several hours should also be avoided. This is because it
is quite likely that real changes in the wave statistics would occur in this time due to
changes in the wind speed or the arrival of swells from distant 1~torms.
"
4.3
FOURIER ANALYSIS
The coptinuous process of wave generation (and the typical form of an irregular wave
record) suggests that any given time history of length, say, T H seconds might
reasonably be represented by the Fourier series
I
s(t)
(4.4)
n=l
(n
= 1, 2, 3, ... , oo)
radians/second
(4.5)
Ocean waves
98
[Ch. 4
An
JTH
0 s(t) cos ((J)n t) dt
metres
(4.6)
Bn
= T2
metres
(4.7)
TH
~+
metres
(4.8)
n=l
21T
TH
radians/second
(4.11)
Sec. 4.4]
99
surprising if this were the case). The synthesised time history will repeat itself at
intervals ofTH whereas the original (real) time history will never repeat itself.
4.4
""0
c
0
QJ
.~
""0
;;:
(/)
~
Q)
I.
of all the wave components within that range of frequencies. It follows that the total
area enclosed by the spectrum is proportional to the total energy per square metre of
the complete wave system.
If we set
Ocean waves
100
ffia
[Ch. 4.
8ro
(J) radians/second
n
2
and
(J)b
8ro
(J)n
+2
radians/second
there will be only one component frequency ron in the range IDa to rob. Actually a real
irregular wave would also contain components at other frequencies within this range,
but the Fourier analysis technique does not identify them explicitly. Instead their
effects are amalgamated in the single frequency identified by the Fourier analysis.
The wave amplitude spectral density ordinate corresponding to this frequency ron is
given by
~~0 metres2J(radian/second)
28ro
(4.12)
A wave energy spectrum corresponding to any irregular wave time history can be
derived in this way and a typical example is shown in Fig. 4.5(a). The spectrum is
discontinuous and consists of a series of rectangles of width oro. The area of each
rectangle is proportional to the energy attributed to that frequency band and
represented by the corresponding single sine wave component.
If the wave energy spectrum is known it is possible to reverse this process and
generate a corresponding time history by adding a large number of component sine
waves according to equation (4.8). In this case the sine wave amplitudes are obtained
by rearranging equation (4.12) to give
(4.13)
It is also necessary to specify the phase angles en and these should be chosen at
random. An infinite number of choices is possible and each will give a different time
history. Nevertheless, all the time histories will have the same wave energy
spectrum. One possible choice would be the phase angles corresponding to the
original wave time history from which the spectrum was derived. Only in this case
will the synthesised time history be identical to the original time history.
In principle an infinite number of sine wave components are required but
acceptable results can be obtained with a limited number. The form of the wave
energy spectrum can be used as a guide to choosing an appropriate range of
Spectral moments
Sec. 4.5]
101
3
(b)
=cc:
0
(.)
Q)
"'
'0
lfffllllllllllffl
N(/)
Q)
E
8
;;;
0
Wave frequency''' (radians/secon9)
Fig. 4.5 Wave energy spectra: (a) typical spectrum from Fourier analysis of irregular wave
time history; (b) typical line spectrum corresponding to time history synthesised by summing
sine waves.
SPECTRAL MOMENTS
(4.14)
Ocean waves
102
[Ch. 4
if the time history has a zero mean and the number of observations is very large. A
time history represented by equation (4.8) therefore has a variance
mo =
T1
H
n=l
metres 2
(4.15)
Since the frequencies are chosen in accordance with equation (4.5) this reduces to
2: s~o
n=I
metres 2
(4.16)
2: s~ (())) oro
mo
n=I
(4.17)
So the variance of the irregular wave time history is equal to the area under the wave
amplitude energy density spectrum.
The time history given by equation (4.8) can be differentiated to obtain the
vertical velocity and acceleration of the sea surface:
~(t)
~(t)
metres/second
(4.18)
metres/second 2
(4.19)
n=l
n=I
sno<D~ cos
(ront+en)
These can be regarded as irregular wave time histories in their own right and can be
analysed to obtain statistics of velocity and acceleration in exactly the same way
as for surface elevation. The amplitudes of the component sine waves are now
Sno<Dn metres/second and Sno<D~ metres/second 2 respectively.
These velocity and acceleration time histories can be analysed to produce
corresponding velocity and acceleration energy spectra. By analogy with equation
(4.12) the spectral density ordinates are
Spectral moments
Sec. 4.5]
S~(ro)
4r2
Wn'onO
S~(ro)
28ro
103
ro~S t; (ro)
(4.20)
ro~St; (ro)
metres2fsecond 4f(radian/second)
(4.21)
So the velocity and acceleration spectral densities can be obtained by multiplying the
amplitude spectral density by appropriate powers of the frequency.
By analogy with equation (4.17) it is clear that the area under the velocity and
acceleration spectra must be equivalent to the variances of velocity and acceleration
respectively. The variance of velocity is
J: s~
m2
(ro)dro
(4.22)
and the variance of acceleration is
J:
',)
'I
metres /second
(4.23)
m 2 and m 4 are called spectral moments since they can be considered as moments of
area of the amplitude spectrum about the vertical axis. In general
(4.24)
and n may take any positive integer value (n = 0,1,2, ... ).
The average frequency can be found by determining the centre of area of the
spectrum from
(!)
ml
radians/second
(4.25)
Ocean waves
104
T = 21Tmo
seconds
[Ch.4
(4.26)
It can also be shown (Ochi and Bolton (1973)) that the mean period of the peaks is
TP =
21T
~ (::)
seconds
(4.27)
Tz =
27T
~ (::)
seconds
(4.28)
Strictly speaking equations (4.27) and (4.28) are valid only if the surface depression
measured at equal intervals of time is normally distributed (see Chapter 17). In
practice this assumption is invariably true for real ocean waves.
Fig. 4.6 shows two irregular wave time histories, and sketches of the corresponding wave energy spectra are shown in Fig. 4.7. The 'narrow band' time history of Fig.
4.6(a) could loosely be described as a sine wave of varying amplitude, and the origin
of the terminology is clear from the appearance of the spectrum: the wave energy is
concentrated in a narrow band of frequencies and little or no energy is present at
other frequencies. One property of this form of time history is that any peak is almost
invariably followed in orderly succession by a downward zero crossing, a trough, an
upward zero crossing and another peak. Peaks below the datum level are very rare
and it follows that the average period of the peaks is almost the same as the average
zerocrossing period.
The 'wide band' time history contains energy over a wider band of frequencies as
shown in Fig. 4.7(b). In this case there are many peaks and troughs which are not
immediately followed by zero crossings and the average period of the peaks is very
much less than the average zerocrossing period. There are many peaks below the
datum level and many troughs above the datum level.
The ratio between the average period of the peaks and the average zerocrossing
period can be regarded as a measure of the 'narrow handedness' of the time history
and its wave energy spectrum. The 'bandwidth parameter' is defined by
(4.29)
and values of E lie in the range 0 to 1: E = 0 corresponds to a very narrow banded
spectrum and E = 1 corresponds to a very wide banded spectrum.
It has been shown (Cartright and LonguetHiggins (1956)) that the significant
Sec. 4.5]
Spectral moments
105
t (seconds)
L,
(metres)
~=.It
(seconds)
(metres)
T,
'1'1
'I
Fig. 4.6 (a) Narrow and (b) wide band time histories.
wave amplitude can be related to the area under the wave energy spectrum. In terms
of significant wave height the relationship is
,. 1
H 113
= 4.00
~ (1;) Vm
metres
(4.30)
so that
metres
1f s
(4.31)
4.00ym 0 metres
if s
(4.32)
2.83 Vm 0
and
Ocean waves
106
[Ch. 4
(a)
E=O
1'1
I\
OJ
._
(radians/seconds)
(b)
c){
t> (radians/second)
Thus, as expected from the general appearance of Fig. 4.6(b), the significant wave
height of a wide banded time history is relatively small.
,
It is often convenient to assume that e = 0 for real wave systems, and equation
(4.32) is assumed to apply so that the significant wave height can readily be estimated
by integrating the wave energy spectrum. In fact e is usually of the order of 0.5 and
this practice results in an overestimate of the significant wave height.
4.6
In general the wave energy spectrum derived from an analysis of an irregular wave
record obtained at a particular place and time in the ocean will be a unique result that
will never be repeated. Although it may be a useful guide to likely wave conditions,
its use for ship design purposes is strictly limited and it is customary to rely instead on
families of idealised wave spectra. Current practice is to use different formulae for
open ocean and coastal (limited fetch) conditions.
Sec. 4.6]
107
SB~;(ro)
= : 5 exp
~~)
metres2f(radian/second)
(4.33)
where
= 691
f4
lil
1'
172.75
metres 2/second4
(4.34)
second 4
(4.35)
The 'two parameters' are the characteristic wave height H 1 and the average period T
(equation (4 .26)). We shall see that the characteristic wave height is often assumed to
be the same as the significant wave height.
The spectral moments of the Bretschneider spectrum are
00
m0 =
exp (
= =
4B
~~) dro
0.0625Hl metres 2
ooA
(4.36)
'I
2.916
'fl
ro exp
(B)
(1)4
= ~r(O)
where r is the gamma function.
Hence, from equation (4.36),
t International Towing Tank Conference.
lil
fZ
metres 2/second 2
(4.37)
dro
oo _ metres 2/second 4
(4.38)
Ocean waves
108
H1
4.00 ym0
[Ch.4
metres
(4.39)
and the characteristic wave height may be related to the area under the spectrum.
The mean zerocrossing period (equation (4.28)) is
21T
(4.40)
TP = 21T ~ (::) =
0 seconds
(4.41)
1.0
(4.42)
and the Bretschneider spectrum is therefore extremely broad banded. These results
imply that a true realisation of a time history corresponding to a Bretschneider
spectrum (including all frequencies up to infinity) would have countless tiny ripples
of infinitesimal period superimposed on the more visually obvious largescale wave
structure as shown in Fig. 4.6(b). These ripples are responsible for reducing the
significant wave height from the narrow banded value (4.0 vmo) to the wide banded
value (2.83 vmo)
While this result is mathematically correct, the infinitesimal ripples which reduce
the significant wave height have no discernible effect on the largescale visual
appearance of the waves or the ship motions which are caused by them. In practice a
time history realisation of a Bretschneider spectrum covering a finite, but,adequate,
range of frequencies of practical importance would not include these tiny ripples, and
a practical analysis of the record to find the significant wave height would include
only visually obvious peaks and troughs. Such a synthesised time history would
invariably have a narrow banded appearance with TP = Tz so that the significant
wave height derived from the record would be much more neariy given by equation
(4.32) than by equation (4.31).
Comparing equations (4.32) and (4.39) it is seen that
(4.43)
and indeed the characteristic wave height for the Bretschneider spectrum is often
loosely referred to as the significant wave height.
The modal period T 0 of the spectrum corresponds to the frequency ro 0 of the
Sec. 4.6]
109
peak, which may be obtained by differentiating equation (4.33) and setting the result
to zero. It is found that
4.849
radians/second
(4.44)
21T
roo
l.296T
= 1.41Tz seconds
(4.45)
= 0.01846H{T
= 0.01425H[T0 metres 2/(radian/second)
(4.46)
Equations (4.40) and (4.45) may now be used to define the constants A and B
more fully:
li[
R[
172.75 T 4
487.3 Tti
li[
123.8 'f4
metres2/second 4
(4.47)
' d
1949
691
Tti
'f4
495
'I
="4 secondsTz
(4.48)
0.658CS 8 ~;(ro)
metres:?f(radianlsecond)
(4.49)
where S8 ~;( ro) is the Bretschneider wave spectral density ordinate (equation (4.33)).
The factor C is given by

Ocean waves
110
[Ch.4
5P~~r~~
20 sec
4
"0
c
0
(J
Q)
(/)
15 sec
.~
"0
(/)
Q)
U./
.s
"'c
~
0
t5Q)
a.
(/)
Q)
>
s"'
Frequency w (radians/second)
Fig. 4.8 Bretschneider wave energy spectra; characteristic wave height 4 metres.
exp
[=!
(roTo 1)
2y
21T
2
c = 3.3
where
2
]
(4.50)
Sec. 4.6]
0.07
for
0.09
for ro
ac
(J)
>
111
21T
To
(4.51)
<21T
To
10
(.)
Modal frequency
Ql
(/)
cco
1:l
~
N
(/)
~
Q)
E
.,
(/)
u;
c
Ql
1:l
~
(.)
Ql
c.
(/)
Ql
>
co
2.0
Frequency w (radians/second)
T0 =10sec
:::::
1:l
c
0
(.)
3l
~JONSWAP
~
1:l
(/)
~
Q)
...
c}{
0
, Frequency tJ (radians/second)
Fig. 4.10 JONSWAP and Bretschneider spectra; significant wave height 4 metres.
I,
Ocean waves
112
[Ch.4
shows a comparison between the two spectra for a characteristic wave height of 4
metres and a modal period of 10 seconds. The effect of the additional factors in the
JONSWAP formula is to increase the height of the peak of the spectrum. There is,
however, a corresponding reduction in the spectral ordinates, on either side of the
peak and the areas enclosed by each spectrum are the same since the characteristic
wave heights are the same.
4. 7
When considering the effects of waves on the angular motions of ships (pitch, roll,
yaw, etc.) it is often convenient to express the energy of the wave system in terms of a
wave slope spectrum rather than the conventional wave amplitude spectrum already
discussed.
We have seen in Chapter 3 that the slope of the surface of a regular sine wave also
varies sinusoidally (equation (3.18)) and that the wave slope amplitude is
cx0 =
k~ 0
(equation (3.19))
radians
metres 1
so that the wave slope amplitude of the nth component sine wave becomes
radians
(4.52)
The time history of the slope of an irregular wave is also an irregular time history
and can be represented by the sum of an infinite number of sine waves in a manner
analogous to equation (4.8). The amplitudes of these wave slope components are
given by equation (4.52). The wave slope irregular time history has its q.wn energy
spectrum and the wave slope spectral ordinates are given by
(1)4
= 1
S~(ro)
g
radians 2/(radians/second)
(4.53)
Ss"'( ro) =
A
2
rog
exp
(B)
4
ro
radians 2/(radianlsecond)
(4.54)
Sec. 4.8]
Wave spreading
S1 cx(ro)
113
(4.55)
Fig. 4.11 shows examples of these wave slope spectra. The JONSW AP spectrum
T0 =10 sec
............... JONSWAP
"D
c
0
(J
Q)
VJ
c
"'
"D
~
'tl
'I
is very sharply peaked but the most striking comparison with the shapes of the
corresponding wave amplitude spectra (Fig. 4.10) is the much greater comparative
importance of high wave frequencies. This corresponds with practical observatiorls:
short highfrequency waves are often very steep even though their amplitudes are
very small.
All the relationships derived for wave amplitude spectra (equations
(4.12)(4.30)) have analogous relationships for wave slope spectra. Thus, for
example, the variance of wave slope can be obtained by integrating the wave slope
spectrum.
r
4.8
WAVE SPREADING
In ideal conditionsin the open ocean the waves might all be expected to travel in the
same direction. However, these 'long crested' waves in: which the infinitely long wave
crests remain straight and parallel are never experienced outside the artificial
Ocean waves
114
[Ch.4
= ~~i0
2 oro ov
(4.56)
where ~njo is' now the amplitude of the component sine wave appropriate to the nth
frequency and the jth direction. For ship design purposes it is assumed that the
directional wave spectral ordinates are related to the ordinates of the equivalent total
wave energy spectrum S~;(ro) by
(4.57)
where D is a constant and m is a positive integer. Since the total wave energy is
assumed to be distributed over the range of directions from  Vmax to Vmax it follows
that
Wave spreading
Sec. 4.8]
115
Limit of
speading
Secondary
wave
direction
metres ?(radian/second)
Putting
v'
we obtain
;ax frr/2
2v
rr
12
(4.58)
116
Ocean waves
[Ch.4
SJ<n,v)
(I)
(I)
and
1T
4vm~
f" ,
(4.59)
()
1
'1 ..
..Vmax
1T
4vmax
1
Vmax
In general
form
form
form
(4.60)
Wave spreading
Sec. 4.8]
1.3.5.7 .... .m
1r
2.4.6.8 .... .(m 1) 4vmax
form odd
2.4.6.8 .... .m
1r
1.3.5.7 .... .(m 1) 2Ymax
form even
117
Fig. 4.14 shows the spreading function D cosm (v') for various values of the
2.0
,r"T"",
20
1.5
10
1.0
Cn
0
Cl
0.5
Relative Heading V
Ocean waves
118
[Ch. 4
Equations ( 4.57) and (4.60) are of little direct use in practical computations of
ship motions in short crested seas. These calculations (see Chapter 14) require the
spread wave spectrum to be represented by a discrete contribution from each of a
finite number of secondary wave directions within the range of the spreading. Each
contribution is essentially a scaled down version of the total wave energy spectrum as
shown in Fig. 4.15. If the secondary wave directions are spaced at intervals of ()v the
appropriate wave energy spectrum at each secondary direction is given by
I.L)) bv
(4.62)
Sec. 4.8]
Wave spreading
0.083SJw)
90
Fig. 4.15 Representation of directional spectrum at discrete heading intervals of 15; cosine
squared spreading over 90.
119
120
Ocean waves
[Ch. 4
Table 4.1 Weighting factors for calculations of ship motions in short crested seas;
ov = 15
m
V/1
(degrees)
Vmax
10
20
0.063
0.125
0.125
0.125
0.125
0.125
0.125
0.125
0.063
0
O.Q75
0.139
0.181
0.196
0.181
0.139
O.Q75
0
0
0.037
0.125
0.213
0.250
0.213
0.125
0.037
0
0
0.003
0.065
0.248
0.368
0.248
0.065
0.003
0
0
0
0.016
0.230
0.508
0.230
0.016
0
0
0
0
0.001
0.146
0.710
0.146
0.001
0
0
0.042
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.042
0
0.034
0.066
0.093
0.113
0.127
0.131
0.127
0.113
0.093
0.066
0.034
0
0
0.011
0.042
0.083
0.125
0.156
0.167
0.156
0.125
0.083
0.042
0.011
0
0
0
0.008
0.043
0.119
0.206
0.245
0.206
0.119
0.043
0.008
0
0
0
0
0
0.011
0.080
0.239
0.338
0.239
0.080
0.011
0
0
0
0
0
0
0
0.027
0.236
0.473
0.236
0.027
0
0
0
0
0.031
0.063
0.063
0.063
0.063
0.063
0.063
0.063
0.063
0.063
0.063
0.063
0.063
0.063
0.063
0.063
0.031
0
0.019
0.038
0.054
0.069
0.081
0.091
0.096
O.Q98
0.096
0.091
0.081
0.069
0.054
0.038
0.019
0
0
0.005
0.018
0.039
0.063
0.086
0.107
0.120
0.125
0.120
0.107
0.086
0.063
0.039
O.D18
0.005
0
0
0
0.002
0.010
0.033
0.073
0.124
0.167
0.184
0.167
0.124
0.073
0.033
0.010
0.002
0
0
0
0
0
0.001
0.008
0.040
0.115
0.209
0.254
0.209
0.115
0.040
0.008
0.001
0
0
0
0
0
0
0
0
0.009
0.073
0.241
0.355
0.241
0.073
0.009
0
0
0
0
0
= 90o
90
75
60
45
30
15
0
15
30
45
60
75
90
Vmax
= 60o
60
45
30
15
0
15
30
45
60
Vmax
= 120o
120
105
90
75
60
45
30
15
0
15
30
45
60
75
90
105
120
5
Ocean wave statistics
5.1
INTRODUCTION
Chapter 4 described how an idealised wave energy spectrum may be defined in terms
of the significant wave height and various measures of the average wave period. This
allows representative spectra to be constructed for any point in the ocean provided
that these quantities are known. Of course many different combinations of significant wave height and average period may occur at any particular point. For practical
ship design purposes we need to choose appropriate values for the sea areas and
seasons in which the ship is expected to operate.
This chapter reviews the available sources of wave data.
'I!
122
[Ch. 5
0
1
2
3
4
5
6
7
8
9
Significant wave
height (metres)
Mean
Range
0
00.1
0.10.5
0.51.25
1.252.5
2.54.0
4.06.0
6.09.0
9.014.0
Over 14.0
0
0.05
0.3
0.875
1.875
3.25
5.0
7.5
11.5
Over 14.0
Description
Calm (glassy)
Calm (rippled)
Smooth (wavelets)
Slight
Moderate
Rough
Very rough
High
Very high
Phenomenal
(5.1)
Tz = 0. 73
'fobs
seconds
(5.2)
T0 = 1.12
'fobs
seconds
(5.3)
(5.4)
Tz = 0 82 ('!'obs)0 96
(5.5)
seconds
T.0 = 1 16 ('!'obs )0 96
seconds
(5.6)
These relationships are illustrated in Figs 5.1 and 5.2. It may be concluded that
observers' estimates of average wave height correspond reasonably closely to the
significant wave height. Since the true mean wave height must by definition be less
Sec. 5.2]
Visual observations
123
(ij
16
""0
c
0
(.)
QJ
.!!!.
"'
It:
""0
""0
(.)
Q)
.:::
QJ
c.
OJ
""0
0
u;
2
~
<J)
c.
co
(.)
""0
:2:
QJ
co
QJ
_,
124
[Ch. 5
than the significant wave height, this implies that observers ignore the smaller waves
when making their estimates.
Average visual estimates of wave period apparently agree quite well with the
modal period, but Hogben and Lumb found that individual estimates were often
widely scattered and could not be regarded as reliable.
5.3
WAVEATLASES
180
150
120
90
60
30
30
Go
90
120
150
160
150
120
90
60
30
30
60
90
120
150
180
Fig. 5.3 Sea areas used by Hogben, Dacunha and Olliver (1986). (Reproduced by permission
of British Maritime Technology Ltd.)
The reliability of the raw visual observations of wave height was enhanced by
correlating them with simultaneous observations of wind speed. This allowed
unrealistic estimates of wave height to be eliminated from the data base. The
unreliable visual estimates of the wave period were not used at all. Instead, wave
period statistics were constructed from correlations with measured data.
Table 5.2 shows a typical set of data from Area 9 (west of the British Isles) in
winter. The data are subdivided into different wave directions and are presented in
the form of scatter diagrams, giving the joint frequency of occurrence (in parts per
thousand) of particular combinations of significant wave height and zerocrossing
Table 5.2 Wave height and period statistics. (After Hogben, Dacunha and Olliver (1986).) Reproduced by permission of British
Maritime Technology Ltd.)
(/)
~
DECEMBER TO FEBRUARY
AREAS
!'
Ut
NORTH
NORTH WEST
PERCENT AGE OF OBS 
..!.
11
287.
41!
14
91 0
' ''
'' "' ""
" u
' "n u
u
n"
" '
I
78
67
56
,_,
,_,
,_,
17
17
I
12
01
45
<4
IS
87
56
S9
711
I
I
I
I
I
I
I
I
101 I
'' ..! ,
G
TOTAL
2) 11) 241
27!t 198
97
37
11
PERCENTAGE OF OBS
4 1000
''
'
131
G 121
,,_,
:;;
II
...
:I:
TOTAL
NORTH EAST
131
'
'' ''
I
, "
,."
'
'
u
'  ""
'
""
"'
'
"',.
'
5
10
\4
.
7
\4
...
I
I
I
I
2
191
220
I
I
''
121
111
...
~
!19
7!1
67
56
. . ..
. .
7
20
75
01
45
1213
TOTAL
1112
>13
<4
20
7
7
80
50
76
~ ~ ____!~ __!}_
B7
89
56
10
20
~ ;:!
~ ~=;
Ill
t01t
910
I
I
I
I
~ 1~=~ I)
7B
I
I
I
''
''
''
"
""
7
I
I
76
"'
197
250
208
"
1011
1213
TOTAL
910
1112
>13
n~
:!
'
~
~ ,_, ' t
  ~
 ~
 ''
1~1
121
111
101 I
~
~
~
56
~
~
~
~~
<4
  
  I
4.
58~
26
II
' "%"'
I
I
I
75
~
78
3 1000
. .
.
I
I
'' "
u"
,_, '"'
10
. "18
I
I
I
115
187
I
I
'"
'".,
 I
tn
I
I
11
I?
1"1.
910
1112
ZERO CROSSINC PEqJOO (s)
74
TOTAl
'"
ALL DIRECTIONS
WEST
!:
,
131
121
.'
I
0
ll9
...
78
67
56
45
J4
23
12
01
...
"'
"'
'
i '~=~
111
I
I
I
''5
' '
' " """
" " "
' "
_____ _
,_ "
" "
7
\4
52
72
45
<4
18
87
56
I
I
I
I
I
IS
89
''
u"
n
7
I
I
I
il
...
'
131
121
111
101 I
0
,_,
201
, 11
''
48
t4
45
87
56
I
I
I
78
67
56
..
18
,_,
,_,
,_,
12
01
I
45
<4
7
17
87
..
07
70
18
I
89
7!1
"
""
"'
'"
I
I
I
I
'
'
'
'
,_,
'' "' ""
l
,_,
'" "" nM ""
'
,, ~' ,_, "
72
199
12
01
"'
1011
1213
TOTAL
910
1112
>13
'
'
u"
7
8
I
I
I
I
I
2
IU
7
I
...
1011
12IJ
TOTAL
56
78
910
1112
>13
ZERO CROSSlNC PERIOD (s)
,,
~
~
~
1)1
121
1 ,_,
17
''
'
47
15
2
5
10
56
IB
10
27
70
81
12
01
45
119
<4
67
5~
I
B9
7B
17
17
I
I
I
"
""
,
011
fll
114
187
I
I
'
' ''
.
.. .. ".' '''
I
78
67
I
I
'
''
' " "
,.' " "
' " " ,." " "'
'
I
45
~
~
tD
' '
'' " ''' '
"" "" ''
"u "'
'"
70
1213
78
910
1112
Zf.:RO CROSSINC Pf.:IUOO (s)
41~
101 I
91
..
~ ,_,
,_,
,_,
16
3 1000
TOTAl
>13
SOUTH EAST
TOTAL
IJ
29
"
131
121
78
67
56
..
45
7!1
'I
,_,
\1 101
,_, 0
118
45
<.4
1000
' '
' ''
'
I
I
I
101 I
91 0
TOTAL
SOUTH
zes zJo tn
>I I
~ 131
121
I
I
I
16~
PERCENTAGE OF OBS ,. 6.
6 1000
52.75
12
01
SOUTH WEST
12
IJ
1011
1213
TOTAL
910
1112
>13
I
I
'' "
' ''
' H" "R "" "" ''
' q" "n " " " '
'' u" "' " "' '
I
TOTAL
20
,_,
"
85
,_, '
07
"'
"'
,_,
,_,
181
'
,,
7B
50
76
110
27
TOTAL
5 1000
I
I
I
10
16
I
78
IS
' ''
''
" "' '''
" '
..
.. .. .. ""'
I
I
82
EAST
50
17
I
I
I
''
'
il
~
~
""
235
81
"'
'"
"
160
1011
1213
TOTAL
910
1112
>13
,,
1)1
31
!1!1
9. 40~
34
11
4 1000
:I : : : : : : : : : : :
:;=:u1011
910
B9
78
67
56
45
34
2l
12
01
<4
1
2
45
1
4
14
11
1
2
56
TABULATED PROBABILITIES ARE IN PARTS PE~ THOUSAND OF THE POPULATION IN EACH TABLE
1
4
1)
35
61
18
B7
1
3
7
19
44
1
1
2
5
1
1
2
4
7
12
1!1
21
1&
5
1
1
1
2
4
5
7
7
4
1
1
1
2
2
2
2
1
3
5I
9
17
7
IJ
1
JJ
18
HI
1
S2
34 31
1 115
82 46
11)6
114 84 48
275
!ll 52
19
238
11
3
1
45
!19
1011
1213
TOTAl
78
i10
U12
>13
......
N
Ut
126
[Ch. 5
Wave atlases
Sec. 5.3]
127
Q)
c:
"'
""0
Q)
Q)
Q)
Area 18:
Sea of Japan
.~
:.0
"'0
..Q
a':
Area 32 : Gulf of
Mexico
0.001
12
limited data on visibility, cloud cover, precipitation, relative humidity, air and water
temperatures, sea level pressure and ice. Table 5.3 shows an example of a scatter
diagram for wind speed and significant wave height for the entire North Atlantic.
Lee, Bales and Sowby (1985) have also published a similar atlas for the Pacific
Ocean.
The hindcast technique a'.(oids the problems of accuracy and fair weather bias
associated with visually observed wave data but depends, of course, on the accuracy
and reliability of the mathematical model used to predict the wave conditions.
128
[Ch. 5
Fig. 5.5 Sea areas in the North Atlantic. (After Bales, Lee and Voelker (1981)).
Table 5.3  Annual wind speed and wave height statistics for the North Atlantic;
probabilities in parts per thousand. (After Bales, Lee and Voelker (1981))
Fl 113 (m)
>24
2024
1620
1416
1214
1012
910
89
78
67
56
45
34
23
12
01
+
+
+
+
+
1
2
5
17
41
+
+
+
+
+
1
2
4
12
34
68
11
+
+
+
1
1
4
10
26
69
92
+
+
+
1
2
5
12
28
66
23
24
+
+
+
+
1
3
7
16
37
67
16
+
+
+
1
2
7
14
30
43
10
+
+
+
+
+
1
2
5
11
21
15
2
+
+
41
+
+
1
2
5
8
8
3
1
+
+
48
55
+
+
+
2
2
2
1
+
+
+
+
1
1
+
+
+
+
+
Total
+
+
+
+
+
+
+
1
4
6
11
18
32
52
80
125
187
259
225
Wave atlases
Sec. 5.3]
129
measurements of wave spectra must provide the most reliable data of all. However,
measuring wave data over a protracted period (years) is an expensive and complicated undertaking and few attempts at systematic data collection have been made.
Probably the most comprehensive is that organised by the US National Oceanographic and Atmospheric Administration (NOAA) and publishd by Gilhousen et al.
(1983). Fig. 5.6 shows the locations of the wavemeasuring buoys around the United
Fig. 5.6 Locations of NOAA wave bouys. (After Gilhousen et al. (1983)).
, II
States. All of the buoys were deployed for at least three years and some have been in
continuous operation for as long as nine years. The buoys recorded information on
air and sea temperature and atmospheric pressure as well as wihd and waves. The
waves were sampled every three hours and a wave spectrum derived from the
recorded time history. The significant wave height and mean zerocrossing period
were derived using equations (4.32) and (4.28) and scatter diagrams similar to those1
shown in Table 5.2 prepared.
,
Many other measurements of wave conditions have of course been made for
specific purposes at various locations throughoutc the world. Typically these are
relatively shortterm studies intended to provide data on the local environment for
use in research or specific projects such as ship seakeeping trials or the design of
offshore or harbour installations. Much of the data have been acquired by commercial organisations who regard them as proprietary information not available to the
general public. However, in 1982 the United Kingdo~ Marine Information and
Advisory Service (MIAS) published a catalogue listing the data sources open to
general use. Over 1350 entries were catalogued and their locations are shown in Fig.
5. 7. The majority of the measur~ments have been made in the coastal waters around
the British Isles and in the North Sea (see Fig. 5.8), but a.significant quantity of data
are also available for North American and Australian\vaters.
130
[Ch.5
Fig. 5.7 Availability of measured wave data. (From Marine Information Advisory Service
Catalogue of Wave Data (1982).)
10W
131
Wave atlases
Sec. 5.3]
0E
I
5W
60N
5E
60N
...
J$..:'
10W
5W
REGIONAL MAPS
0E
5E
Fig. 5.8 Sites of measured wave data around the British Isles. (Reproduced by permission of
Institute of Oceanographic Sciences.)
, II
6
The springmass system
6.1
INTRODUCTION
. J.
Mass a
Force F
Stiffness c
Displacement
x,...
x= 0 metres
Harmonic response
Sec. 6.2]
133
A varying force F is applied at the free end and we require to find the resulting
(varying) displacement x of the mass relative to its undisturbed equilibrium position.
Each of the three components of the system absorbs a proportion of the applied force
so that, at any instant of time,
spring force + dashpot force + mass force
=F
kN
We assume that the spring has no mass, contributes no damping and obeys
Hooke's law so that the spring force is directly proportional to the displacement x.
Similarly we assume that the dashpot has no mass and no stiffness and that the
dashpot force is directly proportional to the velocity x. Finally we assume that the
mass contributes only inertia to the system so that the mass force is directly
proportional to the acceleration .X. So at any instant of time
ai+bx+cx=F kN
(6.1)
where a is the mass in tonnes (or kN/(metre/second 2 )), i.e. the force required, to
accelerate the mass at 1 metre/second 2 to the right; b is the dashpot damping in
kN/(metre/second), i.e. the force required to extend the dashpot at a rate of 1 metre/
second; and cis the spring stiffness in kN/metre, i.e. the force required to extend the
spring by 1 metre.
This kind of system is known as a 'secondorder linear system', 'Secondorder'
implies that equation (6.1) contains terms up to the second derivative (.X) but nothing
of higher order. 'Linear' means that each of the component forces is directly
proportional to the appropriate derivative of x. In other words there are no terms
involving powers like x 2 , x 3 , etc.
' II
Real systems may not be truly linear: for example, it is possible to have a stiffness
which increases progressively as the spring is extended as shOWijl in Fig. 6.2. Another
common form of nonlinearity is the hard limit caused by mechanical stops which
limit the spring's extension. Nonlinear damping is also possible and, in particular,
the dashpot may be fitted with a relief valve to limit the damping force at ~Ofiile
predetermined velocity.
These inconvenient properties complicate the behaviour of springmass systems
and it is usual to assume that real systems are linear where this can reasonably be
justified. Equation (6.1) can then be used to examine the behaviour of the system.
Fortunately the linear assumption can be justified for many problems of ship motions
(with certain well defined and specific exceptions) and the results we shall obtain are
directly relevant to the study of ship behaviour in rough weather.
6.2
HARMONIC RESPONSE
Let us suppose that the force F varies in a sinusoidal manner with amplitude F0 kN
and frequency co radians/second:
F = F0 sin(co t)
kN
(6.2)
134
[Ch.6
Nonlinear
Linear
Hard
limit
Progressive
increase in
stiffness
Displacement x (metres)
.:.!
LL.
Slope=b
(kN/metre/second)
Velocity
'Relief valve'
limit
x (metres/second)
(b) Damping
Fig. 6.2 Linear and nonlinear characteristics.
We might expect that the resulting motion of the mass would also be sinusoidal and it
is indeed found that
(6.3)
is a solution of equations (6.1) and (6.2). x 0 is the motion amplitude in metres and sis
a phase angle in radians. In other words a sinusoidally varying force applied to a
linear damped springmass system will result in a sinusoidally varying displacement
at the same frequency. In practice the phase angle e is found to be negative so that the
displacement sine wave lags the force sine wave as shown in Fig. 6.3. The maximum
(positive) displacementx0 occurs s/ro seconds after the maximum (positive) force F0
Differentiating equation (6.3) gives the velocity and acceleration of the mass:
Sec. 6.2]
135
Harmonic response
(metres)
t (seconds)
T=2rr
(0
Fig. 6.3 Sinusoidal response to a sinusoidal force acting on a linear damped springmass
system.
x= x ro cos (rot + c)
0
.X =
metres/second
(6.4)
(6.5)
'd
x0
( 
x0
( 
After some reduction these two equations yield for the motion amplitude
(6.6)
and for the phase
136
tan e =
 bro
c aro 2
[Ch.6
(6.7)
Let us assume for the time being that there is no damping so that b = 0. Then the
motion amplitude becomes infinite at the undamped natural frequency ro. given by
~ ( ~)
ro* =
radians/second
(6.8)
(6.9)
and
(6.10)
(6.11)
(6.12)
Fig. 6.4 shows the amplitude and phase responses for a secondorder linear
damped springmass system according to equations (6.9) and (6.10) for various
values of the decay coefficient 11 At zero frequency the applied force is steady and
the damping and inertia have no effect because there is no velocity or acceleration.
The displacement is governed only by the spring stiffness:
Xo
F0 c
metres/kN
At A= 1.0 the force due to the spring stiffness exactly balances the force due to the
inertia of the mass. The amplitude response is then
Harmonic response
Sec. 6.2]
137
Q)
CJ)
c
co
Q)
(/)
co
0:
150
200~~~
(X)
Fo
=1
_Q
A=l.O
metres/kN~
(6.13)
2T}C
or 11211 times the zerofrequency response. When there is no damping the amplitude
becomes infinite at the undamped natural frequency, as we have already seen.
However, for finite damping the maximum amplituge occurs at a lower frequency,
called the damped natural frequency, given by
138
[Ch.6
(6.14)
or
radians/second
(6.15)
(Xo)
Fo
max
1
2TJC y'(1 TJ 2 )
metres/kN
(6.16)
When 11 is small the damped and undamped natural frequencies are almost the same,
as shown in Table 6.1. For larger values of TJ the differences become more
TJ
0
0.05
0.10
0.25
0.50
0.707
Ao
1.0
0.997
0.990
0.935
0.707
0.0
(Xo)
Fo
A=LO
(Xo)
Fo
max
00
00
10/c
5/c
2/c
1/c
0.707/c
10.01/c
5.03/c
2.07/c
1.15/c
1.00/c
appreciable until the maximum response occurs at zero frequency when 11 exceeds
0.707. The system is said to be critically damped when TJ = 0.707.
_
At higher frequencies the amplitude response falls towards zero regardless of the
decay coefficient or the spring stiffness. Physically this corresponds to the situation
where the oscillation is so rapid that the system has insufficient time to respond
appreciably.
Fig. 6.4 also shows the phase response of the system. At very low frequencies the
phase is nearly zero and the displacement x is almost in phase with the applied force
F. In other words the system responds more or less instantaneously to the slowly
varying force. As the frequency is increased, the displacement begins to lag behind
the force and the phase 10 becomes negative. As might have been expected, the lag
increases with damping, showing that a well damped system responds sluggishly to
the applied force. The phase is always  90 at the undamped natural frequency
regardless of the damping. At higher frequencies the lag increases still further and
tends to 180 at infinite frequency.
Free decay
Sec. 6.3]
6.3
139
FREE DECAY
Let us now suppose that the springmass system is deflected to some initial
displacement x 00 and then released. We require to examine the subsequent motion.
Since there is no applied force after the system is released, F= 0 and equation (6.1)
becomes
ai+bx+cx=O
(6.17)
kN
We might expect that the resulting oscillation would resemble a 'sine' wave with a
continually decreasing amplitude. In fact the response
= x 00 exp
~ t) cos (rod t)
(6.18)
metres
c
kN/metre
(6.19)
2a
1
't =  =  seconds
b 11ro*
II
(6.20)
140
[Ch.6
~ [\ [\ [\ [\ (j [\_, "
V VlJVV V\
~Ac,? .,~o e
x~4I(\+VtA~\J~c=>.~. . . ,.,. .=  " '= ~ ~~
 n=0.1
l "
4v~~==~ 1]=0.25
L\'./~, 1]=0.5
lL~===,1
1]=0.707
intervals of rr/roct
time
Jrr
tJ = 
:!;
seconds
(j)*
(see Fig. 6.6). The corresponding peak amplitude is (from equation (6.18))
XoJ
= x00 exp
lrr)
(cro*
metres
= X 00 exp
( lrrTJ)
metres
Now the ratio between the Jth and the (J + l)th amplitude is
XoJ
Xo(J+l)
exp ( lrrTJ)
exp [ (J + 1)rrTJ] = exp(rrTJ)
Sec. 6.4]
141
1
(  XQJ )
l1 =lo
:rr 9e Xo{J+1)
Xoo
J:r
1
( XOJ)
11 =log
rr
(6.22)
Xa(l+l)
'II
The decay coefficient can therefore be estimated from the decaying oscillation by
determining the ratio between any pair of successive amplitude~. When the damping
is very small and the oscillation decays very slowly, several estimates of the decay
coefficient can be obtained from a single record. The method is not really practical
when 11 is much greater than about 0.2 and is in any case strictly valid only for ~m'\ll
values of 11.
ai+bx=F kN
(6.23)
and we again require to find the response of the system to a sinusoidally varying force
F= F0 sin (rot)
A solution is
kN'
142
[Ch. 6
Mass a
Force F
Displacement x
x = x 0 sin (rot + s)
metres
F0
metres/kN
(6.24)
1
tans=,
.
0)
(6.25)
aro
ro=b
(6.26)
Fig. 6.8 shows the amplitude and phase response of the zerostiffness system
according to equations (6.24) and (6.25). The responses are quite different from
those of the springmass system. The absence of stiffness means that there is no
natural frequency and the amplitude rises steadily as the frequency approaches zero.
At zero frequency the steady force F0 is resisted only by the damping force bx (since
the acceleration, after an initial transient, is zero) and the mass moves at a steady
velocity given by
x. = Fo metresIsecond
Sec. 6.4]
143
5a
b2r~rr,
w'
or.,
Iii
Q)
~
Ol
Q)
:5'.
w
Q)
"'
(0
.I::
c._
200~~~
0
2
w'
7
Heading and encounter frequency
7.1
HEADING
The ship's heading is defined with reference to the direction of propagation of the
waves. The convention chosen is shown in Figs 7.1 and 7 .2. The ship is assumed to be
Encounter frequency
Sec. 7.2]
145
Following
[l=0
Starboard 1+1
beam
[t=90
Head
[t=180
= 90
J.1 =
J.1
oo
270
= 180
corresponds to following waves with the waves and the ship travelling in
the same direction
corresponds to beam waves with the waves approaching the ship from the
starboard side
' II
corresponds to beam waves with the waves approaching the ship from the
port side
corresponds to head waves with the waves travelling in the opposite
direction to the ship.
1
Quartering waves are defined as heading angles between oo and 90 (or 270 and
360). Bow seas are defined as heading angles between 90 and :180 (or 180 and
270).
7.2
ENCOUNTER FREQUENCY
The characteristics of regular waves were discussed in Chapter 3 and it was shown
that the wave frequency ro, with which a train of regular waves would pass a fixed
point in the ocean, is one of the most important wave parameters. Now although this
wave frequency has some direct influence on ship motions, they are also critically
dependent on the frequency wi}h which a moving ship ";ould encounter these regular
waves.
146
[Ch. 7
Fig. 7.1 shows a ship heading at an angle JL relative to the direction of propagation
of a train of regular waves. The component velocity of the ship in the direction of
wave propagation is
U cos JL
metres/second
and the waves will overtake the ship with a relative velocity
c  U cos JL metres/second
Since the wave crests are A metres apart, a crest will meet the ship once every Te
seconds, where the encounter period is given by
A
Te =     seconds
c U cos JL
(7.1)
2rr
::;: (c U cos p,)
radians/second
(7.2)
ro kU cos JL
ro 2 U
ro    cos JL radians/second
g
(7.3)
(l)e max
g
4U cos JL
'='
radians/second
(7.4)
(7.5)
Encounter frequency
Sec. 7.2]
147
ro=
='cos J.L
= 4roemax
radians/second
, II
(7.6)
Since ro = g/c in deep water this corresponds to the condition when the componeht
velocity is equal to the wave celerity:
U cos J.L
c metres/second
The encounter frequency is negative for higher values of ro. These highfrequency
waves advance only slowly and the negative encounter frequency means that the ship
is overtaking the waves. More precisely.~ a negative encounter frequency means that
the waves are being encountered on their trailing faces, while positive encounter
frequencies mean that the waves are being encountered on their leading faces. This is
of course selfevident on headings forward of the beam where the encounter
frequency is always positive. In following and quartering waves a positive encounter
frequency means that the waves are overtaking the ship.
148
[Ch. 7
In following and quartering waves a given (absolute) value of encounter frequency may be experienced in three different wave systems (if Iroe I < roe max) as
shown in Fig. 7. 3. Two of these wave systems will give positive encounter frequencies
and the third will give a negative encounter frequency. If lroel > ffiemax the given
encounter frequency will arise in only one wave system and roe will be negative.
The corresponding wave frequencies can be obtained by rearranging equation
(7 .3) to give
ro =
2U
radians/second
(7.7)
The physical interpretation of this phenomenon is best illustrated with a numerical example. Consider a ship steaming at 20 knots (U = 10.3 metres/second) in
regular following waves. Suppose that the ship encounters the wave system at a
frequency
lroel = 0.2
radians/second
It is required to find the wave systems which could be responsible. Possible results,
obtained from equation (7.7), are given in Table 7.1 together with corresponding
celerities and wave lengths.
Table 7.1 Regular wave systems giving lroel = 0.2 radians/second; ship speed 20
knots; following waves
Wave
no.
(J)e
(J)
(rad/sec)
(rad/sec)
1
2
3
(4)
0.2
0.2
0.2
( 0.2)
0.285
0.667
1.12
( 0.17)
c
(m/sec)
A.
(m)
34.4
14.7
8.8
759
139
' 49
Wave no. 1 has crests about 0.75 kilometres apart, but its celerity is very high and
it overtakes the ship with a relative velocity of nearly 47 knots. So the high celerity
compensates for the distant crests and results in the required encounter frequency.
Wave no. 2 is much shorter and slower and overtakes the ship with a relative velocity
of only about 8.5 knots. However, the closer crests compensate for the lower relative
velocity and the wave again gives the required encounter frequency. Wave no. 3 is
very short and the celerity is only 8.8 metres/second. So the ship overtakes this wave
system with a relative velocity of about 3 knots, giving the required encounter
frequency. Again the very low relative velocity compensates for the short wave
length. Wave no. 4 is a trivial result: negative wave frequencies have no physical
meaning.
Sec. 7.2]
Encounter frequency
149
The relationship between encounter frequency and wave length may be further
examined by rearranging equation (7.3) to give
U cos JJ.
2
2~ ~ ~8) roe J
[
metres/second
(7.8)
and this is plotted in Fig. 7 .4. The diagram may be used to find the wavelength
1000
Head
~seas
Beam
seas
Following
seas +
150
[Ch. 7
corresponding to any given encounter frequency for a particular speed and heading.
As already demonstrated, a particular encounter frequency is experienced at only
one wavelength (or wave frequency) in head or bow waves; but in following and
quartering waves up to three different wave systems will yield the same absolute
encounter frequency. Fig. 7.4 also demonstrates another peculiar property of regular
waves. In following and quartering waves a wide range of wavelengths may produce
virtually the same encounter frequency. For example, a ship steaming at 20 knots in
quartering waves (JL = 45) has a component velocity of about 7.0 metres/second. In
this condition all the wavelengths from about 50 metres to about 400 metres yield an
encounter frequency close to about 0.3 radians/second. We shall see in Chapter 13
that this phenomenon can have profound implications for roll motions in quartering
seas.
8
Basic equations for ship motions
in regular waves
8.1
INTRODUCTION
Ships do not, in the normal course of events, experience regular waves at sea. So the
study of ship motions in regular waves appears at first sight to be an academic
exercise of no practical significance. Yet it is an essential first step in the calculation of
ship motions in a realistic irregular seaway; moreover; an appreciation of regular
wave motions will give the reader an insight which will prove invaluable in
understanding the general nature of the motions of ships in rough weather.
I
II
152
[Ch. 8
combination of the time histories of three lineart and three angular displacements.
These six displacements are defined using the righthanded axis system shown in Fig.
8.2.
The axis system Exy has its origin fixed at E at the mean water level and regular
waves propagate along the Ex axis. A second axis system ExE 1 xEzY also has its origin
atE but is rotated through the heading angle JJ so that ExE 1 coincides with the mean
track of the ship.
A point 0, lying at the mean water level, moves along ExEl at the mean speed of
the ship, U metres/second. (This speed is approximately the same as the ship would
achieve at the same power in calm water. In head waves the speed will be slightly
reduced and in following waves it may be increased.) The mean position of the ship's
centre of gravity G 0 lies vertically above 0 and is taken as the origin of a third axis
system G 0 x 1 x 2 x3 . At any instant of time the position of the ship's centre of gravity G
relative to the moving origin G0 is defined by three linear displacements:
surge x 1 metres: positive forward
sway x 2 metres: positive to starboard
heave x 3 metres: positive down.
The attitude of the ship is defined by three angular rotations about the axes G 0 x 1 ,
G 0 x 2 and G 0 x3 :
t 'Linear' here means a displacement along an axis as opposed to a rotation about an axis. There is no
necessary implication that the motion responses are linear in the sense that they are directly proportional
to a force.
Sec. 8.2]
153
~0
metres
where the timet may be measured from an arbitrary dat~m. Transforming to the axis
system aligned with the ship's track we find that the wave depression at any point
(xEl, XEz) is
~ = ~0 Sin (k.xEl COS/) kXEz Sin/) rot)
c
metreS
154
[Ch. 8
kU
1T
cos JL
seconds
xE 1
XEz
= x2
x 1
+U
(t
kU ;os
J.L)
metres
metres
y = x3  OG0
metres
8.3
xB 2 , xB 3 ) relative to the centre of gravity of the ship. If the ship has linear
accelerations i 1 , i 2 and i 3 metres/second 2 and angular accelerations i 4 , i 5 and i 6
radians/second2 the mass Dm will have linear accelerations
(xBu
metres/second 2
forward
Sec. 8.3]
i~ = i
i~
2 
xB3 i 4 + xB 1 i 6 metres/second 2
= i 3 + XBz i 4 
xB 1
is metres/second
155
to starboard
downwards
oFl
om
i~
kN
forward
oF2
om
.X~
kN
to starboard
oF3
om
.X~
kN
downwards
oF4
kN metres
OFs
kN metres
oF6
kN metres
The forces and moments required to sustain the linear and angular accelerations
of the whole ship are obtained by allowing om to approach zero and integrating over
the volume of the ship. Bearing in mind that, by definition of the centre of gravity,
JxBz dm
JxB3 dm
we obtain
mi;
= F;
kN
(i
I44 i 4  I 4s is  I46 i 6
1, 3)
= F4
 Is 4 i 4 +Iss is  Is 6 i 6
 I 64 i 4  I6s is + I66 i 6
II
kN metres
= Fs
= F6
kN metres
kN metres
where F; (i = 1, 3) are the surge, sway and heave forces and F; (i = 4, 6) are the roll,
pitch and yaw moments reHuired to sustain the accelerations of the ship. m is the total
mass in tonnes and I 44 , Iss and I 66 are the mass moments of inertia of the ship defined
by
I44
Jcx~z +x~3) dm
tonne metre 2
about the
Iss
Jcx~l +x~3 ) dm
tonne metre 2
xB 1
axis
[Ch.8
156
/45
/46
]56
/54
xBJ XBz dm
tonne metre 2
/64
JxBl xB3 dm
tonne metre 2
165
JXBz XB3 dm
tonne metre 2
For conventional ships the product moments of inertia are usually small and are
invariably neglected. The equations of motion then reduce to
mi;
I;;i;
= F;
= F;
kN
(i
kN metres
1, 3)
(i
4, 6)
(8.2)
F; = F; {1;;, ~'
1, 6)}
kN or kN metres
(8.3)
If the wave amplitude is small compared with the wave and ship lengths motions will
also be small and we may use a Taylor series expansion to obtain a linear
approximation to equations (8.3):
kN or kN metres
(i
1, 6)
(8.4)
Sec. 8.3]
157
where the coefficients a;, a;j, etc are functions of the wavelength (or wave frequency),
ship speed and hull form; a;, b; and C; are also functions of the heading angle.
Substituting equations (8.4) into the six equations (8.2) we obtain six general
linearised equations for smallamplitude motions in regular waves:
6
1, 6)
(i
kN or kN metres
(8.5)
j=l
where
(} = 1, 6; i = 1, 6; j
* i)
and
Aij
= m+a;j
A;j
I;j+ aij
1 3 i = 1 3 ,. = i)
' '
' '
(} = 4 6 i = 4 6 ,. = i)
' '
' '
(} =
and the exciting forces and moments due to the waves are
Fw;
kN or kN metres
(i
= 1, 6)
(8.6)
It is customary to relate all the regular wave ship motions, t9 the wave depression
experienced at the moving origin 0 in Fig. 8.2. In practice tl\e waves here will be
distorted by the presence of the hull (indeed 0 will often be within the hull so that the
surface depression cannot be defined at that point). So, to be precise, the motions are
related to the waves which would have been observed at 0 in the absence of the hull.
Settingx 1 =x2 = 0 metres in equation (8.1) gives for the surface depression at 0
s = so sin (roet)
(8.7)
metres
and the velocity and acceleration of the sea surface perceived by an observer on the
ship at 0 are
~
and
~=
(8.8)
metres/second
metres/second 2
(8.9)
Fwi
kN or kN metres
(i
1, 6)
(8.10)
[Ch. 8
158
1, 6)
(8.11)
(8.12)
tan Y;
L (Aij ij +
b;jxj
+ c;jx)
(i
1, 6)
j=l
(8.13)
Solutions to these equations have the form
X;
X;o
metres or radians
(i
= 1, 6)
(8.14)
The theoretical methods outlined in Chapter 9 are usually used to determine the
coefficients in the equations of motion, but these quantities may also be determined
by experiment. It is instructive to examine the techniques involved since this throws
some light on the physical meanings and characteristics of each coefficient.
Fullscale experiments are impractical but experiments with models are a viable
alternative. The coefficients can be measured in forced oscillation experiments in
which the forces and moments applied by the waves are replaced by forces imd
moments applied by some suitable mechanism while the model is towed in calm
water. Fig. 8.5 shows a typical experiment rig in which the model is mounted on two
vertical struts spaced equally about the centre of gravity. If the struts are oscillated in
unison so that the strut motions s3 a and s3 t (positive down) are the same,
Sec. 8.4]
159
Wave depression
at 0
X;
o;
/w.
Motion
Time
Fig. 8.4 Time histories of wave depression, exciting force and motipn in regular waves.
(8.15)
and the model executes a sinusoidal heave motion. All other motions are restra'ine'd
and the forces necessary to impose the heave oscillation are measured by transducers
at the ends of the struts and recorded on suitable apparatus. The aft transducer is
fitted with a swinging link and this ensures that the longitudinal force is measured in
its entirety by the forward transducer.
The longitudinal force consists of a steady force required to tow the model at the
constant speed U metres/second and an additional oscillatory component due to the
oscillatory motions of the model. We may write for the oscillatory part
where F 10 is the force ampljtude and the heave motionJeads the force by e3 radians.
For an arbitrary shaped hull form the transducers ~ill also experience vertical and
160
[Ch.8
lateral forces and roll moments. If the motions are small the total forces and
moments imposed on the model may be obtained by appropriate addition and
subtraction of the forces measured by the individual transducers:
Fz = Fza + Fzf = F20 sin (roet)
kN
F3 = F3a + F3f
kN
F4a + F4f
kN metres
F4
(F3a F3f)
Fs
F6
Xr
 (Fza Fzf)
kN metres
kN metres
The motions of the model are related to these forces and mOI~ents by six
equations analogous to equations (8.13):
6
kN or kN metres
j=l
(i
1, 6)
(8.16)
Sec. 8.4]
161
(i
1, 6)
(8.17)
and these may be recognised as the equations of motion of six secondorder linear
damped springmass systems with sinusoidal excitation (see equation (6.1)). The
motion response of each system is given by equation (8.15) and the amplitude and
phase response are, from equations (6.6) and (6.7),
1
(i
= 1, 6)
(8.19)
Combining these equations yields for the inphase and quadrature components of the
Fa
___!.._
(i
X;o
1,6)
(8.20)
and
 F;o sin e;
X;o
kN/metre or kN metres/metre
(i
= 1, 6)
!I
(8.21)
The components of the six applied forces and moments which are in phase with the
heave motion are therefore associated with the stiffness and inertia coefficients,
while the quadrature components are associated with damping.
The coefficients which are of most interest in the heave oscillation experiments
are a 33 , b33 and c33 which relate the heave motion to the applied heave force. Fig. 8.6
shows the physical mechanisms responsible for these coefficients. At zero frequency
the model has no heave velocity or acceleration and the heave forceJis related only to
the heave displacement through the coefficient c33 This arises because a steady
downward heave displacement produces an additional displaced volume and a
steady upward restoring force due to buoyancy. A typical relationship between the
heave displacement and the force is shown in Fig. 8.7(a). Provided that the heave
displacement is small, this may be approximated by a straight line whose slope is c33 .
c33 is specifically defined as the gradient oithe curve as it passes through the origin.
At higher frequencies the inph!!.se component of the applied heave force
includes a contribution from heave inertia A 33 This is made up of contributions from
the so called 'added mass' a 33 as well as the real mass m ofthe ship. The former arises
because the accelerating hull causes changes in the fluid velocities adjacent to its
surface as shown in Fig. 8.6(a). The additional force re_quired to accelerate this water
162
[Ch. 8
1J
Fluid
accelerates
'~*

Waves radiate
outwards
(b) Heave velocity .,.__
as well as the hull is included in the inertia coefficient and the ship behaves as though
it has an increased mass. A 33 is sometimes called the 'heave virtual mass'.
Fig. 8.8 shows the results of some heave oscillation experiments by Smith (1967)
on a model of the Dutch Friesland Class destroyer. The added mass a 33 is about the
same as the mass of the ship over much of the frequency range and rises to even
higher values at low frequencies.
The heave damping b 33 arises because the oscillating ship generates waves which
racliate outward and dissipate energy as shown in Fig. 8.6(b ). Energy is also
dissipated by friction but these effects are very small. Smith also measured b 33 and
some results are shown in Fig. 8.8.
The pitch moment F50 sin (roet) measured in the heave oscillation experiments
will yield estimates of the coefficients a53 , b 53 and c53 These describe the influence of
Sec. 8.4]
163
(a)
Keel emerges
Heave displacementx3
~
/
Q)Q)
CE
(b)
~0
~E
"''Q)O
mu
_J ...
Slope=O
Q)
.E
Vertical plane motion
(c)
Lateral plane
moti~nd
heave on pitch in the equations of motion (8.13) and they occur because local inertia,
damping and stiffness forces everywhere along the hull exert pitching moments
about the centre of gravity. If the ship has fore and aft symmetry like a canoe, the
moments arising from the forces on the forward half of the ship will almost exactly
balance those arising from the after half of the ship, and these 'coupling' coefficients
will be very smallt; but for more orthodoxforms residual moments will remain which
may not be negligible.
For arbitrary shaped hull forms forces and moments in the other four degrees of
freedom (surge, sway, roll and yaw) will also be required to sustain a pure heave
~
164
[Ch.8
Frequency w. (radians/second)
2.5
Frequency
(1) 0
(radians/second)
Fig. 8.8 Heave added mass and damping; Friesland Class destroyer. (After Smith (1967).)
oscillation. However, most practical ship forms have port/starboard symmetry and
all lateral plane excitations associated with motions in the vertical plane are zero. In
other words the relationships between the lateral plane forces and moments and
vertical plane motions have the form shown in Fig. 8.7(b) and all the associated
coefficients are zero.
The apparatus shown in Fig. 8.5 may also be used to induce a pitch oscillation by
oscillating the struts in opposition. Provided that the motions are small the pitch is
given by
Sec. 8.4]
Xs
+ Bs)
165
radians
where
x 50
===
2s3 a =
Xr
2s3 f radians
Xr
The analysis proceeds along lines exactly similar to those used in the heave oscillation
experiment and yields estimates of the terms A 55 , b55 , c55 , a53 , b53 and c53 Again all
the lateral plane forces and moments and associated coefficients are zero if the hull
has port/starboard symmetry.
The pitch oscillations cause local vertical motions everywhere along the hull so
that each section of the hull experiences local inertia, damping and stiffness forces
analogous to those experienced by the whole model in the heave oscillation
experiment. These forces exert moments about the centre of gravity and are
responsible for the coefficients a55 , b55 and c55 The local forces distributed over the
forward part of the hull oppose those on the after part of the hull so that the residual
heave forces associated with the coefficients a53 , b53 and c53 are usually small. Indeed
they would be zero on a hull with fore and aft symmetry at zero speed.
The pitch virtual inertia coefficient A 55 includes contributions from the so called
added mass moment of inertia as well as the true mass moment of inertia of the ship's
structure. This is analogous to the heave added mas~,already discussed. The true
mass moment of inertia may be expressed as
155
(8.22)
'1'/
where the longitudinal radius of gyration about a transverse axis through the centre
1
of gravity is usually in the range
0.2Ls < k 5 < 0.25L5
metres
166
Frequency
We
[Ch. 8
(radians/second)
Frequency w. (radians/second)
Fig. 8.9 Pitch added moment of inertia and damping; Friesland Class destroyer.
'
(After Smith (1967).)
In most cases the coefficients are found to be nearly independent of the motion
amplitudes used in the experiment and this justifies the assumption of linearity in the
derivation of the equations of motion. Coefficients associated with roll motion,
however, are an exception to this general rule as shown in Fig. 8.11. Roll added mass
moment of inertia decreases and roll damping increases with roll amplitude. So the
assumption of linearity may not be justified in this case.
In general, vertical plane forces and moments will always be required to sustain
motions in the lateral plane even for ships with port/starboard symmetry. However,
the relationship between the vertical plane excitation and the lateral plane motion
for ships with lateral symmetry will have the symmetrical Ushaped form shown in
167
Sec. 8.4]
2.0
<le V(B/2g)
2.0 '
VleY(B/2g)
Fig. 8.7(c). In other words the vertical plane excitation will have the same magnitude
and direction regardless of the direction of the lateral plane motion. Since we are
concerned only with small motions and our linearisation requires the coefficients to
be determined from the slope at the origin, all such coefficients are zero.
A third category of coefficients is always zero regardless of the ship's shape.
These are all stiffness coefficients associated with 'the ship's geographical location
with respect to the origin G 0 No forces or moments are required to sustain surge and
sway displacements x 1 and x2 so that
cil
c;2
(i
1, 6)
Table 8.1lists 60 coefficients (out of a total of 108) which are zero for a ship with
port/starboard symmetry. A further 12 coefficients are usually negligible and are
invariably neglected.

168
[Ch. 8
100 20 10
54 3
(1)
Y(B/2g)
Fig. 8.11 Roll added mass moment of inertia and damping. (After Vugts (1968).)
This results in six much simpler equations for smallamplitude motions of a ship
with lateral symmetry:
(8.23)
sway:
(m + a 22 )
Xz
(8.24)
a 42 .X 2
(8.25)
(8.26)
Sec. 8.4]
169
Motionsj
Motionsj
1
2
3
4
5
6
1
2
3
4
5
6
02
03
s
s
s
s
03
02
02
02
02
03
02
03
02
02
01
01
02
03
03
03
03
03
03
01
01
01
01
03
01
01
02
03
02
02
01
01
03
02
03
01
01
02
02
03
Surge
Sway
Heave
Roll
Pitch
Yaw
Surge
Sway
Heave
Roll
Pitch
Yaw
Stiffness c;1
pitch:
=
yaw:
+CssXs
kN metres
(8.27)
(8.28)
The vertical plane motion equations (heave and pitch) are coupled as we have
already seen. In other words the heave equation includes terms dependent on pitch
so that heave is influenced by pitch and vice versa. However, the surge equation is
uncoupled and independent of the other motions. The lateral plai).e motions are also
coupled so that these motions are affected by each other. There is, however, no
coupling between the vertical plane motions and the lateral plane motions. So heave,
pitch and surge are not affected by events in the lateral plane; neither are sway, roll
or yaw motions affected 'by heave, pitch or surge. t This allows the vertical and lateral
plane motions to be considered separately.
9
Strip theory
9.1
INTRODUCTION
Sec. 9.2]
Strip motions
171
Each strip has associated local hydrodynamic properties such as added mass,
damping and stiffness which contribute to the coefficients for'tije complete hull in the
equations of motion. Similarly the wave excitations experienced by the hull are
composed of contributions from all the strips.
.
Strip theory assumes that these local hydrodynamic propefties are the same as
would be experienced if the strip were part of an infinitely long cylinder of the same
crosssectional shape as shown in Fig. 9 .1. In other words threedimensional effects,
such as mutual interference between the strips, flow leakage around the ends of the
ship and effects due to changes in the shape of the strip over the length oxB 1 , are
ignored.
172
Strip theory
x~ 0
x4a
= x3  xB 1 sin x5
== x3  x81x5 metres
= x4
radians
[Ch.9
to starboard
(9.1)
downwards
(9.2)
to starboard
(9.3)
Consider an observer stationed at some fixed point alongside the ExEI axis in Fig.
8.2. The oscillating ship passes him at a steady velocity U metres/second. At some
instant of time a certain strip is opposite the observer and his perception of its lateral
velocity is given by the total differential of equation (9 .1):
Now the distance x81 from the strip to the approaching centre of gravity is
diminishing at the rate
x81 = 
U metres/second
(9.4)
i4o =
x4
radians/second
(9.7)
.X4o =
x4
radians/second 2
(9.8)
An origin 0 in the waterplane, shown in Fig. 9.2, is chosen for calculations of the
Hydrodynamic coefficients
Sec. 9.3)
r
'
I
173
1
x~f:G_.....,...,.._.,..
~
x;G
:1
I Waterline
local hydrodynamic properties of each strip. The velocities and acclerations of this
point are then
x~
= x~ 0 
OG x~
x~
OG x~
x~ 0
metres/second
metres/second
(9.9)
2
(9.10)
x3
'I
X3a
'I
metres/second
(9.11)
I
1/
x3a
metres/second 2
(9.12)
x~ = X~a
radians/second
(9.13)
x~ = X~a
radians/second 2
(9.14)
x3
Strip theory
174
[Ch. 9
(9.15)
(9.16)
(9.17)
where all the motions and excitations are referred to an origin 0 in the waterplane.
The coefficients a22 , b22 , etc. are all local values (per metre length of strip) analogous
to the corresponding coefficients for the complete hull (with due allowance for the
change of origin). These local coefficients are functions of the size and shape of the
strip in question and may be determined by the methods described in Chapter 11.
9.3.2 Coefficients in the heave and pitch equations
Consider a ship undergoing simultaneous forced heave and pitch motions in calm
water. The momentum of the surrounding water in the plane of the strip is
M~
a~ 3 x~
where the vertical velocity is given by equation (9 .11). The force required to oppose
the hydrodynamic reactions and sustain the motions of the strip is composed of the
rate of change of this momentum together with contributions from the damping and
stiffness:
As successive strips pass the observer stationed alongside the ExEl axis he will
perceive a changing local added mass a~ 3 So the rate of change of momentum is
'I
a33
= dxd
a33 XB1
Bl
=  Udxd
Bl
(a~ 3 )
(see equation (9.4)). Using equations (9.11) and (9.12) the downward vertical force
on the strip becomes
Hydrodynamic coefficients
Sec. 9.3]
(b '
33
175
uda~3)(
.,
uXs )
dxst x3Xs1Xs+
(9.18)
The total heave force and pitch moment required to balance the hydrodynamic
reactions and sustain the heave and pitch motions of the ship are obtained by
allowing ox81 to approach zero and integrating over the length of the hull:
(9.19)
= 
F5
Jx
81
dF3
(9.20)
kN metres
Comparing these equations with equations (8.25) and (8.27) enables us to obtain
expressions for each coefficient in the equations of motion for pitch and heave. For
conventional ships with zero crosssection area at the forward perpendicular the local
added mass is always zero at the stem. It follows that
I
da~3
dx
dx81
,
a 33 a
II
tonnes/metre
81
where a~ 3 a is the local added mass for v.ertical motions at the stern in tonnes/metre
and x 81 a is the distance from the centre of gravity to the stern (negative).
The coefficients then become
(9.21a)
176
[Ch. 9
Strip theory
b33
Jc~3dxB1
C33
a3s
kN/(metre/second)
kN/metre
(9.21c)
kN/(radian/second 2)
(9.21d)
b3s
(9.21b)
2U
kN/(radian/second)
(9.21e)
C3s
lj
as3
 JxB1a33dxB1
bss
kN metres/(metre/second2)
(9.21f)
(9.22a)
JxB c~ 3 dxB 1
kN metres/metre
(9.22c)
kN metres/(radian/second2)
(9.22d)
ass
kN/radian
kN metres/(radian/second)
(9.22e)
kN metres/radian
(9.22f)
177
Hydrodynamic coefficients
Sec. 9.3]
(a~ 2 x~
II
(9.23)
The angular momentum of the water in the plane of the strip includes a contribution
due to the lateral velocity x~:
M~ = a~ 4 x~ + a~2 x~
The roll moment about the axis through G required to balance the hydrodynamic
reactions and sustain the motions includes contributions from the rate of change of
this angular momentum as well as contributions associated with the roll damping and
the moment required to sustain the lateral velocity. In addition there are contribu
Strip theory
178
[Ch. 9
tions associated with the roll stiffness c~ and the moment due to the lateral force oF2
acting through 0:
oF4
= [ a~4 x4 + ( b~
u::)
x4 +
c~4x4
(9.24)
ox
81
to
Fz
fdF2 kN
(9.25)
F4
(9.26)
F6
dF4 kN metres
Xa1 dF2
kN metres
(9.27)
Comparing these equations with equations (8.24), (8.26) and (8.28) allows us to
obtain expressions for the lateral plane hydrodynamic coefficients. Again we note
that for conventional ships with zero crosssection area at the forward perpendicular
daiz
dx
dx81
,
a22 a tonnes/metre
(9.28a)
81
(9.28b)
179
Hydrodynamic coefficients
Sec. 9.3]
(9.28c)
where a~2 a is the local added mass for horizontal motions in tonnes/metre at the
stern.
Similar expressions are valid for the coefficients a~4 and a~2 Equating terms in
equations (9.25)(9.27) and (8.24), (8.26) and (8.28) we obtain
azz
Ja~2
hzz
Jb~2
dxB 1
dxB 1 
(9.29a)
tonnes
Ua~2 a
(9.29b)
kN/(metre/second)
(9.29c)
Ua~ 4a
(9.29d)
kN/(radian/second)
I
az 6
b26
XBr
b~2
1/
kN/(radian/second 2 )
dxB 1 
Ua~2a 2UJa~2
dxB 1
(9.29e)
kN/(radian/second) (9.29f)
(9.29g)
Ja~z
dxB 1 
OG
a~2
dxBl.
kN metres/(metre/second2 )
(9.30a)
(9.30b)
[Ch. 9
Strip theory
180
a44 =
Ia~4
+ OG2
b44 =
Ia~z
OG
dxBl 
Ja~2
Ib~4
dxB 1
dxBl
Ib~4
 OG
dxBl 
a46
Ic~4
I a~z
dxBl
XB1
dxBl
kN metres/(radian/second2 )
dxBl
Ib~z
Ua~2a
dxBl
+ OG2
+ OG Ua~4a
c44
Ia~4
OG
dxB 1 
OG
OG
(9.30d)
kN metres/( radian/second)
kN metres/radian
dxBl
(9.30c)
xB 1 a2z dxB 1
(9 .30e)
kN metres/(radian/second2 )
(9.30f)
I
Ia~2
U a~2a uIb~z
 OG
+ 20G U
c46
= 2
dxB 1
+ OG XBla Ua~za
kN metres/(radian/second)
dxBl
Ib~z
+ OG u
 OG U2a~2a kN metres/radian
(9.30g)
dxBl
(9.30h)
(9.31a)
(9.31b)
Sec. 9.4]
181
kN metres/(radianlsecond2 )
(9.31c)
b64
b66
Jx~ 1
(9.31d)
a2 2 dxB 1 kN metres/(radianlsecond2 )
(9.31e)
JX~1bzzdxB1 UxBlazza
(9.31f)
(9.31g)
!I
The linearisation of the equations of motion in Chapter 8 allows the wave excitations
to be considered independent of any ship motions and to be expressed as functions of
the wave amplitude alone. In other words the wave excitations are assumed to be the
same as the ship would experience if it were rigidly restrained and allowed no
motions at all.
It was shown in Chapter 8 (equation (8.1)) that the wave depression at any point
(x 1 , x2 ) related to the moving origin 0 could be written as
~
metres
Since the ship is allowed no motions, the centre of gravity remains above 0 and
we may write
~
x1
= xB 1
metres
The wave depression varies across each strip but we ~ssume the ship to be slender
(that is, the waterline beam of all strips is much lesslhan the wavelength) and this
allows us to calculate the wave depression with sufficient accuracy by setting
182
Strip theory
x2
[Ch. 9
= 0 metres
sin(roet Q)
metres
(9.32)
= A
B
metres
(9.33)
where A is the crosssectional area of the strip in metres 2 and B is the waterline beam
of the strip in metres.
Referring to equation (3.36) we find that the pressure fluctuation at the mean
draught Dis
P=
and, following a similar procedure to that described above, this can be written as
P= 
 pg exp( kD)I;,
kN/metre 2
(9.34)
Similarly the vertical velocity of the water at the mean draught is, from equations
(3.23) and (3.26),
(9.35)
v=
1;,
metres/second 2
(9.36)
The horizontal velocity at the mean draught is, from equations (3.22) and (3.26),
Sec. 9.4]
metres/second
183
co exp( kD)
s sin fL
metres/second
to starboard
(9.37)
cov sin fL
metres/second 2
(9.38)
The slope of the pressure contour at depth Dis, from equation (3.16),
ks
cx15 = 
cov
g
radians
CXDz
. fL
= cov sm
g
ra d"1ans
(9.39)
The rate of change of the athwartships component of this slope (equivalent to the
average angular velocity of the water) is
' 1/
ixm
= 
ssin fL
radians/second
(9.40)
= 
kcov sin fL
radians/second 2
(9.41)
In addition there are contributions arising from the rate of change of the vertical
momentum of the water surrounding the strip and a force associated with the vertical
velocity of the water. The total vertical force on the strip is then
Strip theory
184
[Ch. 9
= 
U dx
Bl
Obtaining P, v and
each strip is
8Fw 3
where
P2 = ro ( b~ 3  U
da' )
dx::
tonnes/(metre/second2 )
The total heave force and pitch moment are obtained by allowing
approach zero and integrating over the length of the hull:
ox
81
to
(9.43)
Sec. 9.4]
185
tan y 3
__ JRzdxs1
(9.45)
__ JxalRzd.xal
tan y 5
(9.46)
where
9.4.3
R1
R2
1/
Fig. 9.3 shows the hydrostatic force experienced by a restrained strip in waves. The
inclined water surface causes a lateral shift of the centre of buoyancy from B to B 1
and the buoyancy force vector is assumed to act normal to the pressure contour at the
mean draught D. The buoyancy force is
pgA Ox81
kN
kN
to starboard
In addition there are contributions dueto the rate of change of horizontal momentum of the water surrounding the strip and forces associated with the lateral and
rotational velocities of the water. The total horizontal force on the strip is then
186
[Ch.9
Strip theory
M'
T
Pressure
contour at
depth D
Fig. 9.3 Lateral plane hydrostatic force and moment on a restrained strip.
M'w 2
a~ 2 u 2
+ a~4 &D2
D ( I )
Dt Mwz
da~z
= azzUz U dx
Uz + az4CXvz
I
,_
81
da~ 4 cxDZ
._
U dx
Bl
Obtaining u 2 , u2 , etc., from equations (9 .37)(9 .41) we find that the total
horizontal force on the restrained strip is
Sec. 9.4]
187
(9.47)
where
The total sway force and yaw moment are obtained by allowing oxB 1 to approach
zero and integrating over the length of the hull:
(9.48)
(9.50)
(9.51)
188
Strip theory
[Ch. 9
where
R3
R4
B2
B2
oA =  ex sin p, =  ex
metres 2
B
3
metres
Hence
and
metres
The buoyancy force exerts a roll moment about 0:
CiA=A
b 1 b2
Sec. 9.4]
pg (
B3
_
)
 OBA
12
<X02
189
.
kN metres per metre length of strip
The total roll moment exerted on the strip about 0 also includes contributions from
the rate of change of the angular momentum of the water surrounding the strip as
well as contributions associated with the rate of rotation and the horizontal velocity
of the water. The equations of motion (8.23)(8.28) require the roll moment to be
related to an axis through the centre of gravity. The lateral force Fw 2 acting through 0
exerts a moment about G and this is included to give
DFw4 =
(9.52)
..
a44!XDz
uda~4,
IXDz + a4zUz
dx
Bl
uda~2
dx
Uz
Bl
kN metres
(9.53)
where
The total roll excitation on the strip is obtained by allowing oxB 1 to approach zero
and integrating over the length of the hull. We obtain
190
Strip theory
[Ch.9
(9.55)
where
R 5 = (P5 cos Q
R6
10
Hydrostatic coefficients
10.1
INTRODUCTION
The hydrostatic or stiffness coefficients cij in the strip theory formulae (equations
(9.21), (9.22) and (9.30) are associated with changes in buoyancy forces due to steady
displacements of each strip. This chapter outlines the calculation of these buoyancy
fores to allow determination of the stiffness coefficients.
pgBx3 oxB1 kN
So the downward heave force required to sustain the steady downward heave
displacement is
(10.1)
192
Hydrostatic coefficients
[Ch. 10
(10.3)
The second moment of area of the waterplane about a transverse axis under the
centre of gravity is
(10.4)
We may now simplify the expressions for the vertical plane stiffness coefficients given
for the complete hull in equations (9.21) and (9.22);
c33 = pgAw kN/metre
(10.5a)
(10.5b)
(10.5c)
(10.5d)
Sec. 10.3]
193
Lateral plane
!I
where Vis the volume of each wedge in metres 3 and Vis the volume of displacement
of the hull in metres 3
1
The buoyancy force vector now acts through the new centre of buoyancy B 1 and
interesects the plane of symmetry of the ship at the metacentre M. For small angles of
roll the distance
metres
194
Hydrostatic coefficients
2B
3
[Ch. 10
metres apart
It follows that
where the second moment of area of the waterplane about the longitudinal axis is
(10.6)
We now have
BM =IT
metres
(10.7)
kN
kN metres/radian
(10.8)
GM 5 is called the 'solid' metacentric height and is an important measure of the ship's
stability. If GM5 is negative the ship will be unstable, and liable to capsize.
This calculation assumes that there are no free surfaces within the ship. Fig. 10.2
shows the effect of a free surface such as might be found, for example, in a partially
filled fuel or ballast tank. As the ship rolls, the fluid in the tank collects in its low side
and causes a lateral shift in the position of the centre of gravity from G to G 1 . The
righting lever GZ is thereby reduced to G 1Z and this may be col)veniently expressed
Sec. 10.3]
195
Lateral plane
!I
11
Local hydrodynamic properties
11.1
INTRODUCTION
Progress with strip theory requires the evaluation of the local added mass and
damping coefficients ajj and bjj in equations (9.21), (9.22), (9.30) and (9.31).
Methods of solution for shiplike forms have been developed by Ursell (1949a,b),
Tasai (1959, 1960), Grim (1959), Porter (1960) and others. The techniques used are
involved and laborious, requiring devious and intricate methods of solution for even
the simplest of cases. Their application is quite impracticable without the aid of a
digital computer.
A full explanation is inappropriate for this book. The presentation which follows
is necessarily superficial but should serve to give a taste of the methods used. The
reader who requires a more detailed knowledge is referred to an excellent paper by
de Jong (1973).
11.2
U rsell made the first step towards solving the general problem of calculating the twodimensional potential flow around a cylinder of arbitrary shape floating in a free
surface in 1949. He derived a solution for a heaving circular cylinder as shown in Fig.
11.1.
The water depth was assumed to be infinite and the heaving motion of the
cylinder was taken to be
y =Yo cos( rot+ e)
metres
where
(11.1)
Sec. 11.2]
197
y=yo cos(OJt+E)
X
~
Waves radiate
away from the
cylinder
1/
\jl
=!:[ ( + ~1
'l'c
+ ( 'l's + ~ q2m).jlzm)
sin( rot)]
metresYsecond
(11.2)
198
<P=gso[(<Pc+
7TCO
m= 1
+ ( <Ps +
Pzm<Pzm)
[Ch. 11
cos(cot)
1; qzm <Pzm)
sin(cot)]
metres2 /second
(11.3)
where so is the amplitude of the radiated wave at an infinite distance from the
cylinder. The terms
gso['l'c cos(cot)+ 'Vs sin(cot)]
7TCO
<Ps sin(cot)]
1T
(11.4a)
<Pc =
1T
(11.4b)
'l's =
7r
(11.4c)
(11.4d)
represent an infinite number of pulsating multipoles aligned with the y axis (see
equations (2.39) and (2.40)). The individual components are
Sec. 11.2]
_ zm(sin(2m9)
o/zm a
..h
,zm
k sin[(2m 1)9])
+ (2m l)r 2m1
(11.5a)
k cos[(2m 1)9])
+ (2m l)r 2m1
(11.5b)_
zm(cos(2m9)
'!'2m a
199
,zm
Ur
1 0\j/
= ;;
metres/second
09
Ur
Now
d: cos 9 = 
Yoffi
(11.6)
Since the stream function \jl is defined as zero when 9 = 0, the constant of integration
c(t) is zero.
After some manipulation of equations (11.2) and (11.6) Ursell showed that
(11.7)
where
~ ~( 1)m1 2kap2m
 ( )
A o/c 8=7T/2 + .L..J 1
m=l
(11.8a)
200
m=l
(
kaqzm
1)ml 2"=1:
m
[Ch. 11
(11.8b)
and
Pzmfzm
(11.9a)
L:
qzmfzm
(11.9b)
m=l
m=l
where
fzm =
 {sin(2m9) +
(11.9c)
and 'l'c and 'l's are evaluated on the surface of the cylinder (r =a).
Ursell obtained an approximate solution for Pzm and q2m by limiting the infinite
series on the righthand side of equations (11.8) and (11.9) to only six terms. The
equations were then formulated at ten values of e from oo to 90 at intervals of 10.
This gave ten simultaneous equations for p and another ten equations for q. Each
equation had only six unknowns (p 2 , p 4 , p 6 , p 12 and q2 , q4 , q6 , q 12) and the
best approximate solutions were found by a least squares method.
This approach ensured that the stream and potential functions behaved correctly
at the surface ofthe cylinder. We may examine the flow at an infinite distance from
the cylinder by allowing r and x to approach infinity in equation (11.3). The
contributions of the multipoles now become negligible and equation (11.3) reduces
to
<1>
metres 2/second
(J)
Comparing this with equation (3.1) we see that the potential represents a regular
wave of amplitude 1; 0 metres and frequency ro radians/second in water of infinite
depth. So the potential function has the expected behaviour on the free surface at an
infinite distance from the cylinder.
The added mass and damping coefficients of the oscillating cylinder are obtained
by considering the pressure fluctuations on its surface. From Bernoulli's equation
(2.18) the pressure is
Sec. 11.2]
P = p(n a<j> q
at
201
kN/metre 2
P = p ( gy 
~~)
kN/metre2
an
ay = g metres/second2
at
kN/metre 2
Consider now a small element of length os on the surface of the cylinder (Fig.
11.2). The force due to the fluctuating pressure acting on the element is
!I
metres
202
[Ch. 11
and the total vertical force applied to the cylinder by the fluid is
TT/2
rr/2
So the total force required to oppose these pressures and sustain the oscillation of the
cylinder is
F=
rrt2
rr/2
P cos e ds = pa
f"'2
rr/2
a<j>
at
cos e de
(11.10)
where
Mo=
No=
"
f
f
6
12
1
( 1)m q2m
1Tkaq2
o (<J>s)r=acos9d9+ ];1
4m21
+4
(11.11a)
rr/2
6
( 1)m1 P2m
1Tkap2
o (<J>s)r=acos9d9+ ];1
4m21
+4
(11.11b)
y=
Sec. 11.2]
1Ta
metres 2/second
203
(11.13)
For the time being we may express the force on the cylinder (equation (11.10)) as
(11.14)
where
F = 2 pagt;oMo kN/metre
c
1T
F
s
=  2 pagt;oNo
1T
(11.15a)
kN/metre
(11.15b)
and the velocity and acceleration (equations (11.12) and (11.13)) are
y = Vc cos(rot) + v s sin(rot)
metres/second
(11.16)
metres/second 2
(11.17)
where
 gt;oA.
1taro
v
s
=  gt;oB
1taro
a =  gt;oB
c
1Ta
=  gt;oA.
1Ta
metres/second
!I
(11.18a)
metres/second
(11.18b)
metres/second 2
(11.18c)
metres/second 2
(11.18d)
These equations are illustrated in a vector diagram in Fig. 11.3. By using similar
triangles it can be shown that the amplitude of the component of the force in phase
with the acceleration is
(11.19)
204
[Ch. 11
sin(wt)
Fig. 11.3 Vector diagram showing exciting force, acceleration and velocity for a heaving
cylinder.
2pag~0 (A.No
+ B.Mo) [A. sm(rot)B.
.
A~+B~
cos(rot)] kN/metre(11.20)
Fa=1T
The amplitude of the component of the force in phase with the velocity is
(11.21)
so that the force in phase with the velocity is
Fv=
A~+B~
(A.No + B.Mo)
A~+ B~
Sec. 11.2]
205
(11.24)
 pg~6A.  1Tpg ~6
E    o;z joules/metre length of wave crest
2
2
(11.25)
This energy is radiated away from the cylinder at the group velocity u0 given by
equation (3.44). In deep water
c
u0 = Z metres/second
and the energy passing a fixed point in one motion (or wave) period is E/2
joules/metre length of the wave crest. But since waves are generated on both sides of
the cylinder the total energy radiated in one motion cycle is E joules/metre and this is
equal to the work done in oscillating the cylinder through one motion cycle:
2~2
1T
E = pg2
(0
J2rr/ro Fy dt
joules/metre
II
(11.26)
\jl=g~o
[A. cos(rot)+B. sin(rot)] metres~/second
1T(O
(11.29)
206
[Ch. 11
metres 2/second
(11.30)
After some manipulation of these two equations the ratio of the wave amplitude at
infinity to the cylinder motion amplitude is found to be
~0
Yo=
11.3
rrka
(11.31)
V(A: + B:)
LEWIS FORMS
Ursell's solution for the flow around a heaving circular cylinder was a useful step
towards creating a viable theoretical method of predicting ship motions in waves.
However, ship hull crosssections are not usually circular and we require a technique
for predicting the flow and resulting hydrodynamic coefficients for cylinders of
arbitrary, or at least shiplike, crosssection.
The techniques of conformal transformatiQn described in Chapter 2 provide a
method of transforming a circle in the z plane
z=x+iy=reie metres
(11.32)
+ ixB 3 metres
(11.33)
metres
(11.34)
will map any point on the circle in the z plane into a corresponding point on any
symmetrical (about the xB 3 axis) shape in the ~ plane provided that appropriate
values of the coefficients a0 , aI> a3 , a5 etc. are chosen. In practice it is usual to set the
radius of the cylinder at
a=r=l.O metre
and to truncate the transformation series to only three terms:
Lewis forms
Sec. 11.3]
207
(11.35)
We shall see that this still allows a wide variety of shiplike crosssections to be
generated from the unit circle. These forms will not in general be exact replicas of any
given hull crosssection, but the match will usually be sufficiently close to allow
adequate estimates of the hydrodynamic coefficients for ship motion calculation.
The resulting family of forms are known as Lewis forms, after F. M. Lewis who first
proposed their use in 1929.
Substituting equations (11.32) and (11.33) into equation (11.35) and separating
real and imaginary parts we obtain a pair of parametric equations in describing the
shape of the Lewis form in the I; plane:
XBz
= a0 [(1 + a 1)
xB 3
sin
a0 [(1 a 1) cos
e a 3 sin(39)]
metres
(11.36a)
e+ a 3 cos(39)]
metres
(11.36b)
= 0 corresponds to the bottom of the unit circle and the keel of the Lewis form.
Substituting = 0 in equations (11.36) we obtain
XBz
=0
(11.37a)
metres
(11.37b)
,I I.1
(11.38a)
xB 3
=0
metres
(11.38b)
where B is the waterline beam of the Lewis form in metres. The beam/draught ratio
of the Lewis form is
H = B = 2(1 + a 1 + a3 )
D
1a1 +a~
The crosssectional area of the Lewis form is
(11.39)
208
[Ch. 11
B/2
=2
metres 2
x 83 dx 82
(11.40)
(11.41)
(11.42)
(11.43a)
a3=
3 C+ y(9 2C)
(11.43b)
where
1T
a0 is simply a scale factor governing the overall size of the Lewis form.
Lewis forms may therefore be defined in terms of their beam/draught ratio Hand
their section area coefficient
Fig.11.4 shows a range of Lewis forms for various
values of Hand t Evidently a wide variety of conventional hull crosssections may
be represented reasonably well by choosing Lewis forms having the same beam/
draught ratio and section area coefficient. There is no limit to the permissible beam/
draught ratio but only a limited range of section area coefficients are possible.
Clearly the formula for a 3 (equation (11.43b)) becomes invalid when
cr.
cr.
C>i
t
Sec. 11.3]
Lewis forms
209
0
H=1
cr<
(11.44)
Lewis forms having section area coefficients greater than this cannot exist. In
practice section area coefficients approaching this limit ltave rather angular shapes of
the type shown in Fig. 11.5(a). They are not repr~~entative of conventional hull
210
[Ch. 1l
Xsz
~~
H=2
o=1.15
forms: more to the point is the fact that such section shapes would probably
experience flow separation around the sharp bilges, and the potential ,flow techniques to be employed for predicting the hydrodynamic coefficients would not be
expected to give very reliable results. To avoid forms of this nature it is usual to
suggest that the Lewis forms should lie completely within the circumscribing
rectangle so that
If the section area coefficient is too small the Lewis form will adopt physically
impossible shapes with negative values ofxs2 andxs 3 as shown in Fig. 11.5(b). We
therefore also insist that
Sec. 11.4]
211
o:::::;e:::::;1T
2
37T
37T
(4H) :::::; cr:::::; (24+H)
64
256
for H:::::;2
These limits are shown, together with the ideal upper limit given by the inequality
(11.44), in Fig. 11.6.
The special case
H=2
1T
cr='
4
!I
= ao sin 9,
Xs3
= ao COS 9
metres
11.4
splane.
11.4.1 Introduction
The determination of the hydrodynamicyroperties of Lewis forms oscillating in the
free surface proceeds along lines broadly similar to those pioneered by Ursell for the
heaving circular cylinder. Heave, sway and roll motions are considered in turn: in
each case the stream and potential functions are again given by equations (11.2) and
(11.3) although the components 'l'c, <l>c, 'l's and <l>s and the infinite series of multi poles
are defined differently. In principle it is again necess_'!ry to determine the weighting
coefficients Pzm and q2 m for the infinite number of multipoles. In practice an
212
[Ch. 11
No Lewis forms
1.2
1.0
.....
c
Q)
c::;
~
Q)
0.8
(.)
co
~
co
0.6
."B
Q)
(/)
0.4
0.2
0
Beam/draught ratio H
(11.45)
pBzro1Tz
(11.46)
Sec. 11.4]
213
where
P2m ( 1)m 1
k:Q~ 1
metres 2/second
(11.47)
~
kBQ 1
B.='l's(O, 7T/2)+ ~ q2m(1)m 1 Q
metres 2/second
2 2
(11.48)
J
J
Q3
1 [
1rkBQ 5
cl>c(0,9)Q d9+Q
[p2m(1)m 1 Q 4 ] +
Q8 2
2
2 m=1
metres 2/second
(11.50)
Mo=
No=
00
Q3
1 [
1rkBQ 5
1
cl>s(0,9)Q d9+Q
[q2m(1)m Q4]+gQ
2
2 m=1
2
metres 2/second
(11.49)
rr/2
rr/2
00
1T
cl>c =
1T
exp( kx 83 ) cos(kx82 )
metres2/second
(11.51)
I
oo
oo
J
0
metres 2/second
exp(  vlxs21)
.
[ v sm( VXs3) + k cos( vx 83 )] dv
2 k2
v
II
(11.52)
2
metres /second
(11.53)
exp( vlxs21)
~.
.
.
[ v cos(vx 83 ) + k sm( vx 83 )] dv
2 k2
v +
2
metres /second
(11.54)
The weighting coefficients Pzm and q 2 m are obtained by approximate solution for
a finite number of unknowns of the simultaneous equations
214
Q6
N
'l'c(1, e) Q 'l'c(1, 7T/2) = ]; P2m/2m
2
metres 2/second
(11.55)
(11.56)
'l's(1,
[Ch. 11
metres /second
where
(11.57)
_ _1_ _ _a_1__ ~
2m+ 1 2m+ 3
(11.58)
Q 1  2m 1
(11.59)
(11.60)
(11.61)
(11.62)
(11.63)
Q
7
2m+1
_ _1_ _ _a_1__ ~
Qs 2m  1 2m+ 1 2m + 3
2m+3
(11.64)
(11.65)
<l>c(r, e), <j>.(r, e), etc.' in these formulae imply values calculated at (r, e) in the circle
plane.
11.4.3 Swaying Lewis form
The added mass and damping coefficients for sway motions of a Lewis form are
, pB 2 NoPo + Moqo
a22 = tonnes/metre
2
PZ+
qz
(11.66)
215
Sec. 11.4]
2
b' = proB MoPo Noqo
22
2
pfi + q{j
(11.67)
, B 3 PoXR + qoYR
2
a42kN metres/(metre/second ) per metre
2
pfi + q{j
(11.68)
3
b' _B PoYR qoXR kN metres/(metre/second) per metre
42 2
pfi + q{j
(11.69)
where
Q9
3a37f"P2
<Ps(O, 8) Q d8~Q :
Q9
3a37f"P2
kB( 1)m 1
<Pc(O, 8) Q d8~Q P2m
Q
Qw(11.71)
2 22
2
2 m=1
rr/2
Mo=
No=
rr/2
kB( 1)m1
L
q2m
Q
Qw(11.70)
2 22
m=1
oo
00
'rl
= Jrrt2,!.. (0 8) Qn d8 + rrkB(alP2 a3p4) + ~ q2m( 1)m+1Q
y
R
't's
'
Q2
2
16Q2
2
LJ
m=1
Q2
2
12
(11.73)
<Pc(r, 8) and <Ps(r, 8) in these formulae imply evaluation at the polar coordinates (r, 8)
in the circle plane. In cartesian coordinates the stream and potential functions are
'l'c = 7r exp( kx 83 ) cos(kx83 )
<Pc = 
7r
metres 2/second
(11.74)
(11.75)
exp( vlxs21)
.
k2 + v~ . [v cos(vx83) k sm(vx83 )]
Xs2
2
:
dv   k( 2
2 ) metres /second _"
Xs2+Xs3
(11.76)
216
[Ch. 11
exp( =t vlxB21)
.
k2
[k cos(vxB3) + v sm(vxB 3)] dv
2
+v
o
XB2
2
d
+ k( 2
2 ) metres /secon
XB2 +xB3
oo
(11.77)
The weighting coefficients p 2 m and q2 m are found by approximate solution for a finite
number of unknowns of the simultaneous equations
N
L P2mf2m
(11.78)
m=l
q2mf2m
(11.79)
m=l
where
(11.80)
kB
f 2m = cos(2m + 1)9] + 2 Q
Q14 +
kB ( 1)m+l
Q
2 2
Q1s
(11.81)
'V2m ( r, 9 
_ 3a cos[(2m + 4)9]
3(2m + 4)r2m+4
. )
11 82
(11.84)
(11.85)
Sec. 11.4]
2a 1 (1 + a3)
8a3
Q 12 =(2m+ 1) 2  4 +(2m+ 1) 2 16
217
(11.86)
(11.87)
(11.88)
(11.89)
11.4.4 Rolling Lewis form
The added mass and damping coefficients for the rolling motions of a Lewis form are
(11.90)
pB 3 Moqo + NoPo
PZ + q&
=8
Po
qo
rl
L P2mf2m
(11.94)
m=O
2: q2mf2m
(11.95)
m=O
where
(11.96)
218
(m=FO)
[Ch. 11
(11.97)
11.5
Accurate calculation of the hydrodynamic properties of cylinders of shiplike crosssection is clearly of paramount importance in the prediction of ship motions in waves.
It is therefore somewhat surprising to find that relatively few experiments to verify
these calculations have been carried out.
Vugts published the results of the classic experiments in this field in 1968. He
tested a number of cylinders in the towing tank at the Delft Shipbuilding Laboratory.
Fig. 11.7 shows, in simplified form, the arrangement he used. The 4.2 metres long
cylinders were mounted across the tank and an oscillation mechanism was used to
force heave, sway and roll motions in turn. The waves generated by tile cylinder
motions radiated away and were absorbed by beaches at each end of the tank, some
70 metres from the cylinder. The forces and moments necessary to sustain the
cylinder motions were measured and used to determine the added mass, damping
and crosscoupling coefficients for each motion.
Details of the cylinders are given in Fig. 11.8 and some ofthe results are shown in
Figs 11.911.11. Cylinders A, Band Call had
H=2,
7T
cr=4
Cylinder A was circular while cylinders B and Chad shiplike crosssections. Fig.
11.9 shows that the added mass and damping coefficients were virtually identical,
Sec. 11.5]
219
Cylinder
11:
Shape
Circle
=v
Ship
~
Ship
=v
Rectangle
Rectangle
Rectangle
1.155
0.5
11:
11:
Triangle
~
I
!=.
.v
.,
'I
confirming that these are essentially functions only of beam/draught ratio and section
area coefficient.
Fig. 11.10 shows how the heave added mass and damping increase with beam/
draught ratio; the opposite trend is shown for the sway coefficients in Fig. 11.11.
220
[Ch. 11
L:ewis form
theory  
2
w V(BI2g)
Oi
N
til
>
"'"'
::9.
"(
QJ
0.5
0
w Y(B/2g)
Fig. 11.9 Hydrodynamic coefficients for three heaving cylinders; H=2, cr=rr/4. (From
Vugts (1968).)
Sec. 11.5]
0
2.5
Oi
N
co
>
1.5
<:(
~
rl
:S
0.5
w V(BI2g)
Fig. 11.10 
221
222
[Ch. 11
0
<u V(B~2g)
0.5
w y(B/2g)
Fig. 11.11 
12
Roll damping
Roll damping
224
...

[Ch. 12

......
Wave making
0
Eddies
~J)
~
"j
0
~
,;_
Skin friction
~
~
...,...
')
0
Appendage
forces
In order to circumvent these unwelcome problems we may calculate an equivalent linear damping coefficient which allows for the effects of the nonlinearities but
is used in a linear way. This allows us to continue to use the linear equations of
motion and the spectral techniques for irregular wave calculations. The equivalent
linear damping coefficient is chosen so that the calculated energy dissipated by this
term in the equation of motion is the same as that which is actually dissipated by the
nonlinear effects. In general this means that the equivalent linear damping coefficient depends on the roll motion being experienced and a new value of the damping
must be calculated for every situation.
Since the predominant rolling motions experienced at sea occur at the natural roll
Sec. 12.3]
225
1T
ro.4
b44
xl0
kN metres
(12.2)
E
1T
2
ro.4 X4o
kN metres/(radian/second)
(12.3)
Tanaka (1960) conducted a series of model experiments to determine the eddyshedding roll damping characteristics of a number of different h~ll section shapes as
shown in Fig. 12.2. Schmitke (1978) used these results to develop expressions for the
eddymaking damping coefficients for each type of hull section. He postulated that
the force due to the eddy shedding acts at the relevant sharp corner at a radius rb
metres from the centre of gravity. If the roll velocity is i 4 radians/second the local
flow velocity in the plane of the hull section will be rbi4 metres/second. The local
force resisting the roll motion is expressed in the form
(12.4)
where s and dxa 1 are the girth and length of the hull section and CE is a drag
coefficient depending on the hull form. ~Now for sinusoidal rolling motion (equation(12.1)) the force F exerts a moment about G given by
(12.5)
and the energy dissipated by this moment in one rollcycle is
RoD damping
226
[Ch. 12
G~
UN
Full
Triangular
Round bilge
X40
d.x4
= 3 P r6 xlo
ro~4
s CE dxB 1
kN metres
(12.6)
Using equation (12.3) we see that the equivalent linearised local damping coefficient
for eddy making is
b~E
4pro.4
3
= 3;;:x40 rb s CE dxB 1
kN metres/(radian/second)
(12.7)
The total equivalent linearised rolldamping coefficient for the complete hull is
obtained by integrating along the length of the hull:
kN metres/(radian/second)
(12.8)
Sec. 12.3]
227
It remains to determine the drag coefficient CE which varies along the hull.
Schmitke (1978) gave the following empirical formula:
(12.9)
where Z 1 and Z 2 are given as functions of BjKG, y (the inclination of the hull section
at the waterline) and reiD in Fig. 12.3.; re is the effective radius at the keel given by
z,
4
I
!I
0.20
Fig. 12.3 Z 1 and Z 2 for UN sections. (After Tanaka (1960), Schmitke (1978).)
Roll damping
228
re =
~s
4.122.69
~~ + 0.823 ( ~~YJ
[Ch. 12
metres
KG
forB< 2.1
s
KG
for>2.1
Bs
(12.10)
and
14.146.7 x 40 + 61.7xl0
(12.11)
0.4380.449 (BJKG)
+ 0.236 (B
(12.12)
/KG)2
The water flowing past the ship's hull exerts frictional forces on the hull surface. It is
usual to express the force acting on a small element of the hull surface in terms of a
nondimensional local skin friction drag coefficient defined as
frictional force on element
CF = ~~~~! p x (local velocity)Z x area of element
(12.14)
Consider a girthwise elementos of length oxB 1 metres as shown in Fig. 12.4. Let
the element be positioned at (xB 2, xB 3), a distance r metres from the centre of gravity.
If the roll velocity is 4 radians/second the velocity at the element will be r 4
metres/second and the component velocity tangential to the surface of the hull will be
r x4sin(el + e2) metres/second. eland e2are the polar location of the element and the
slope of the hull surface so that:
Sec. 12.4]
229
Xs3
sin el
sin e2
Xsz
0Xs3
cos e2
OS '
cos el
(12.15)
OXsz
OS
(12.16)
kN
II
(12.17)
oF4
ox:4
kN metres
(12.19)
Roll damping
230
[Ch. 12
Then the equivalent linearised rolldamping coefficient for skin friction is obtained
from equation (12.3) by allowing 3xB 1 and to approach zero and integrating along
the hull and around the girth:
os
b44F
4 Cp
= "3;;:p ffi4 X40
JL, Is (
O
O r
XB3
2
dxB 2
dxB 3 )
~ XBz ~
ds
dxBl
(12.20)
kN metres/(radian/second)
It remains to determine the local skin friction coefficient Cp. Schmitke (1978)
suggested that the Schoenherr formula for the average skin friction coefficient for
'smooth turbulent' flow used in calculations of ship resistance is appropriate:
Cp
(12.21)
where the Reynolds number is based on the forward speed and length of the ship:
_ P U Ls
RN
(12.22)
f.lw
This is clearly inappropriate if the forward speed is zero, and Kato's (1958)
formula may then be used:
Cp
(12.23)
where RN is now a Reynolds number based on the average rolling velocity and the
average distance rfrom the centre of gravity:
(12.24)
=
1T
metres
(12.25)
(12.26)
Sec. 12.5]
12.5
231
(a)
U=O
F=D
(b) U>O
(12.27)
where C0 is the nondimensional drag coefficient.
The drag force yields a roll damping moment
(12.28)
and the energy dissipated in one roll cycle is
X40
F4 dx 4
kN metre;;
(12.29)
Roll damping
232
[Ch. 12
and equation (12.3) then gives the equivalent linearised damping coefficient due to
appendage drag forces at zero speed as
b44AD
= 4Cn
17' p X4o Oh4 ""'
L.,;
3
3
AA r A
)
kN metres/ ( radtan!second
(12.31)
where the summation is over all appendages. A suitable value for the drag coefficient
is given by equation (2.66).
12.5.2 Lift forces on appendages
If the forward speed is not zero the rolling motion induces an angle of attack on each
appendage as shown in Fig. 12.5b. The angle of attack is
ex
tan
1('A:4)
radians
(12.32)
radians
(12.33)
= U metres/second
The appendage develops lift and drag forces which are respectively normal and
parallel to the local velocity vector as shown in Fig. 12.5. The total force normal to
the ship's longitudinal axis is
L cos ex + D sin ex kN
=L
kN
(12.34)
if ex is small
Hence if the induced angle of attack ex is small the total roll moment applied to the
(12.35)
and the rolldamping coefficient attributable to the lift forces developed on the
appendages is
Sec. 12.6]
b44AL
=i
pU
2: dd~L AA d.
kN metres/(radian/second)
233
(12.36)
where the summation is for all appendages. It should be noted that this damping
coefficient is independent of roll angle and no special linearisation techniques are
necessary.
12.6 TOTAL ROLL DAMPING
The total roll damping is obtained by adding the contributions from the individual
roll damping sources discussed above:
b44
b44w
for U = 0
(12.37a)
b44w
for U > 0
(12.37b)
or
b44
13
Ship motions in regular waves
13.1 INTRODUCTION
The strip theory outlined in previous chapters may be used to estimate the motions a
ship would experience in regular sinusoidal waves of small amplitude. For conventional ships at moderate speeds these estimates are usually found to be of adequate
accuracy for everyday engineering purposes. As an example this chapter gives the
results of a specimen set of calculations of the motions of a frigate of length 125
metres and explains ,the physical reasons for their nature. Ship motions are of course
functions of hull shape and size, and the results given here should not be used to give
numerical estimates of the motions of other hull forms. Nevertheless, the same
general characteristics will be found to apply to the motions of all conventional
monohull ships.
metres
(13.1)
o;)
metres or radians
(i = 1, 6)
(13.2)
The motion amplitudes x ;o and the phases oi are functions of the speed U, heading
p., and encounter frequency roe. The amplitudes are assumed to be proportional to the
wave amplitude 1,;0 and it is usual to express them in nondimensional form: linear
motion amplitudes x 10 , x 20 and x 30 are nondimensionalised by dividing by the wave
Sec. 13.3]
235
~(~::)
= ~(~::)
ro.3 =
radians/second
ffis
radians/second
II
(13.3)
(13.4)
0
G(__ ____J/
Fig. 13.1 Maximum heave and pitch excitations in very long waves.
236
[Ch. 13
The surge equation is independent of all the other equations and has no stiffness
term c 11x 1 Surge motions would therefore be expected to be approximately
analogous to those of a damped system with no stiffness and there is no natural surge
frequency.
In very long waves the encounter frequency roe is very low and dynamic effects
associated with added mass and damping are virtually negligible. So the excitations
and reactions experienced by the ship are almost wholly attributable to the buoyancy
changes as the waves pass the hull. Maximum pitch moment occurs at the wave nodes
and maximum heave force occurs at the wave crests and troughs as shown in Fig.
13.1.
These large excitations in very long waves result in the large motion amplitudes
illustrated in Fig. 13.2. For moderate ship speeds the wave celerity is very much
greater than the ship speed and the vessel may be regarded as virtually stationary as
Sec. 13.3]
237
the wave passes by. The ship will behave more or less like a particle of water at the
surface, following a circular orbit of radius ~ 0 given by equation (3.32). So maximum
heave (equal to ~ 0) will occur at the wave crests and troughs and maximum surge
(also equal to ~ 0 ) will be experienced at the wave nodes. The ship surges towards
the approaching crests and recedes after the crest has passed by.
Viewed from the vessel's deck the ship will appear to be crawling like a tiny ant
over a succession of very long shallow hills. The ship will always be aligned with the
wave surface so that maximum pitch (equal to the wave slope amplitude k~ 0) will
occur at the wave nodes.
In shorter waves the buoyancy forces alternate along the ship's hull as shown in
Fig. 13.3. This, together with the growing importance of dynamic effects at the
higher encounter frequencies, results in a general reduction in excitation in shorter
waves. Smith measured the total excitations experienced by a restrained model of the
Friesland Class destroyer in regular head waves in 1967. He used apparatus similar to
that shown in Fig. 8.5 and some of his results are shown in Figs. 13.4 and 13.5. These
show that the ship only experiences significant excitations when the waves are longer
than about threequarters of the ship length.
Typical calculated heave and pitch transfer functions are shown for the 125 metre
frigate in head waves in Figs 13.6 and 13.7. As expected, all responses approach unity
at zero encounter frequency, corresponding to the long wave case discussed above.
The responses are generally reduced at higher frequencies because of the
substantial reductions in excitation experienced in these shorter waves. However, as
the speed is increased, the range of wave lengths having significant levels of
Fig. 13.3 Buoyancy forces on a restrained ship in regUlar waves of constant amplitude.
238
[Ch. 13
'
\.
\
\
1.6
Encounter frequency we (radians/second)
1oor,
0=~
"'
Q)
Q)
.;;:. 100
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Fig. 13.4 Heave exciting force; Friesland class destroyer in head seas. (After Smith (1966).)
excitation is encountered over a wider range of frequencies. This range of frequencies may eventually include the natural frequencies of heave and pitch given by
equations (13.3) and {13.4). The responses may then exhibit resonant peaks as
shown at 30 knots in Fig. 13.6. Pitch and heave motions are, however, invariably
heavily damped and the resonant peaks are never very pronounced.
Figs 13.6 and 13.7 also give the heave and pitch phases. In very long waves the
heave phase
Sec. 13.4]
239
l.orx~TTT4~3r~2~~1T.5~1T~~~r0.,.5
'
'
Encounter frequency
we (radians/second)
(/)
Q)
      ~~
g' 90
"0
1.6
0
Encounter frequency
We
(radians/second)
Fig. 13.5 Pitch exciting moment; Friesland class destroyer in head seas. (After Smith (1966).)
indicates that the heave motion is synchronised with the wave motion and that
maximum heave (down) occurs in the wave troughs.
The pitch phase is
at zero encounter frequency .This indicates that maxi.mum positive (bow up) pitch
motion occurs one quarter of an encounter period after the wave trough has passed
the ship's centre of gravity (see Fig. 13.2).

240
[Ch. 13
0.5
30 knots
2.5
Encounter frequency
O>e
(radians/second)
2oor~
200
13.4
In following waves the motions are again confined to surge, heave and pitch. Figs
13.8 and 13.9 show calculated heave and pitch transfer functions in regular following
waves. These again approach unity when the waves are very long and the encounter
frequency approaches zero. Only a limited range of (positive) frequencies can be
encountered at any given speed for the reasons discussed in Chapter 7. The transfer
functions consequently adopt the shapes shown with two possible motion responses
(corresponding to different wave lengths) at any one positive encounter frequency.
Sec. 13.4]
I'
I'
1.5
... I"'
. 0.
r'
0.5
30 knots
iJLs
k ....
knots
241
120 knots
1 0 knots
JJ
:": 1.0
:il
'<
0.5
2.5
0
Encounter frequency'''" (radians/second)
2.5
0
Encounter frequency
'''e (radians/second)r fl
over most of the range of encounter frequencies for which the response is significant,
indicating that the heave motion is again nearly synchronised with the wave motion.
The pitch phase is now approximately
.
242
[Ch. 13
.IJ
.;;
c
0
.E
c
.2
(/)
"'
(.J1
~
Q)
>
"'
Q)
0
0.6
0.4
Encounter frequency
Ole
0.6
(radians/second)
0
Ui'
Q)
~
Ol
Q)
'"
/.:
Q)
(/)
"'c.
.!:
Q)
>
"'
Q)
0.6
0
Encounter frequency
Ole
0.6
(radians/second)
over most of the significant range of encounter frequencies. Maximum positive (bow
up) pitch motion now leads the maximum wave depression at the centre of gravity by
approximately onequarter of the encounter period.
Sec. 13.5]
243
.0'
'if
c
0
'
g
c
:::J
....
Q)
( /)
tn
~
,...
..c
.B
0::
15
Ui
~
100
Cl
Q)
~
Q)
(/)
200
co
..c
c.
..c
(.)
J:
300
0.6
0.6
13.5
In oblique waves the ship motions are no longer confined to the vertical plane. Roll,
sway and yaw motions are also present. However, the vertical plane equations of
motion (8.23), (8.25) and (8.27)) for a symmetrical shiP are independent ofthose for
the lateral plane ((8.24), (8.26) and (8.28)). So the lateral plane motions in oblique
244
[Ch. 13
waves of small amplitude will have no effect on the vertical plane motions and these
may therefore be considered in isolation.
In very long oblique waves the ship again appears to be crawling over a succession
of long shallow hills as shown in Fig. 13.10. At the crests and troughs the heave
motion will again equal the wave amplitude, exactly as in head and following waves.
The 'effective wave length' measured along the ship's track is
A sec IL metres
and the corresponding effective wave slope amplitude is therefore reduced to
21T/,;o
cxw = A
sec IL
= k/,;0 cos IL
radians
(13.5)
In these long oblique waves the ship will again always align itself with the wave
surface, and the maximum pitch, equal to k/,; 0 cos fL, will occur at the wave nodes.
~0
f
Fig. 13.10 Heave and pitch motions in very long oblique seas.
Figs 13.11 and 13.12 show typical oblique wave transfer functions for the frigate
at 20 knots. On headings forward of the beam (90 < IL < 180) the responses are
broadly similar in general form to the head wave responses already discussed. Only
one response is possible at any given encounter frequency and the motions generally
decrease with increasing encounter frequency. The heave responses increase as the
heading approaches 90 and the wave excitation becomes synchronised along the
entire length of the hull. The pitch response decreases as the heading approaches 90
and would be zero in beam waves if the hull had fore/aft symmetry.
Sec. 13.5]
245
I~= :x
.:;
c
0
~
....::::>
....Q;
Cll
~
Q)
>
"'
::r:
Q)
Encounter frequency
We
(radians/second)
(i)
Q)
Cl
Q)
"0
;:;
.c
Q)
Cll
"'c.
.<=
Q)
>
"'
Q)
::r:
0.5
2.5
0
Encounter frequency
We
(radians/second)
Fig. 13.11 Heave transfer functions for a frigate in oblique waves; speed 20 knots.
On headings abaft the beam (0 < p, < 90) the responses adopt the general form
of those already described fo:r following waves. The ,range of possible encounter
frequencies is reduced, depending on the heading, and more than one response is
possible at any given encounter frequency.

246
[Ch. 13
0
0.5
2.5
Encounter frequency we (radians/second)
(i)
Q)
C)
Q)
:9.
"'
tO
Q)
"'"'0.
120
..r::;
..r::;
....0
a:
0.5
2.5
Fig. 13.12 Pitch transfer functions for a frigate in oblique seas; speed 20 knots.
in very long waves, indicating that heave is synchronised with wave depression at all
headings. Pitch phase is
Sec. 13.6]
247
85 =
85 =
and
wave frequency or nondimensional wave length. Fig. 13.13 shows the frigate's
oblique wave transfer functions plotted in these forms. They have the singular
advantage that the responses are now all singlevalued and the complications of
multivalued responses at a given encounter frequency are avoided.
1.5
1.5
1.0
0
~
.;(
0.5
0
Wave frequency"' (radians/second)
2.0
Wave frequency w (radians/second)
i.IL,
'telL,
Fig. 13.13 Alternative presentations of oblique wave transfer functions for a 125 metre
frigate at 20 knots.
In beam waves pitch motions are, as we have already seen, usually very small. Yaw is
usually also negligible and the ship motions are essentially confined to heave, sway
and roll. Fig. 13.14 illustrates these motions in veryJong waves. The ship again
follows the circular orbit of a particle of water at th~ surface. The heave and sway
motions are therefore equal to the wave amplitude 1;0 : maximum heave motion
248
[Ch. 13
occurs at the wave crests and troughs (as already seen) and maximum sway occurs at
the wave nodes. The ship sways towards the approaching wave crest and recedes
after the crest has passed by.
If there are no internal free surface effects to reduce the effective metacentric
height (see Chapter 10) the ship's deck will always be aligned with the wave surface.
Maximum roll, equal to the wave slope amplitude k~ 0 , will occur at the wave
nodes. Fig. 13.15 shows the sway transfer function in beam waves. The sway
equation (8.25) has no stiffness term c22x2 so there is no sway resonance. Sway
amplitudes decrease with increasing encounter frequency and the phase remains
essentially constant with
Sec. 13.6]
249
2.0
w=w. (radians/second)
0
en
Q)
Cl
Q)
:s
.s
r
150
0
2.0 I
w=w. (radians/second)
Fig. 13.15 Sway transfer function for a frigate at 20 knots in beam waves.
Ol 4 =
~ ( ~:)
radians/second
(13.6)
250
[Ch. 13
v=we (radians/seconds:
w=we (radians/second)
Fig. 13.16 Roll transfer functions for a frigate in beam waves; GMF = GM 5 .
The damping increases with forward speed. This gives a general reduction in the
peak roll response and a slight reduction in the frequency at which the peak roll
response occurs.
At zero frequency the roll phase is
Sec. 13.6]
251
indicating that positive roll (to starboard) leads the maximum wave depression by
onequarter of a period as shown in Figs. 13.14 and 13.17. At the natural roll
frequency the roll phase is
and the maximum roll is then synchronised with wave crests and troughs as shown in
Fig. 13.17. At very high frequencies
and the ship then rolls in opposition to the wave slope. Roll motions are then,
however, quite small.
c<:J
c<:J
The responses shown in Fig. 13.16 were calculated with no allowance for free
surface effects on the metacenttic height. If these effect~ are significant the reduction
in the roll stiffness reduces the natural roll frequency and increases the roll response
at low frequencies as shown in Fig. 13.18.
c
252
[Ch. 13
<>=<>e (radians/second)
'''="' (radians/second)
Fig. 13.18 Roll transfer functions for a frigate in beam waves; effect of metacentric height;
IL =goo.
Sec. 13.8]
Absolute motions
253
,5\
..
~
)(
c
0
.;=
u
::J
(/)
~
0
0:::
Fig. 13.19 Roll transfer functions for a frigate at 20 knots in oblique waves.
encountered at frequencies close to the natural roll frequencYrK see Fig. 7.4) and the
roll response is significantly increased. For this reason roll motion is often a
maximum in quartering seas, particularly at high speed.
Figs 13.20 and 13.21 show the sway and yaw transfer functio'ns in oblique waves.
In very long waves (ro = 0) the sway amplitudes approach the athwartships component of the wave orbit radius ~0 sin p,. On headings forward of the beam (90 < JL <
180) both sway and yaw amplitudes decrease rapidly with wave frequency. Maximum responses occur in quartering seas and rise to very high values when the
encounter frequency approaches zero. Strip theory predictions are likely to be
inaccurate in these circumstances. In practice the ship would be steered by a
helmsman or an autopilot and this would effectively limit these largemotion
amplitudes.
The six motions considered so far completely define the possible movements of a ship
in a seaway. However, seakeeping studies often call for assessments of the motions
experienced at some particular; point on the ship, such ~s the bridge or the flight deck
of a warship. These can be calculated from a knowledge of the six motions we have
already defined with respect to the centre of gravity~"
254
[Ch. 13
~t=60
we=O
~
X
c
0
"
c
.2
;:
>"'
0
The angular motions are the same everywhere in the ship but the local linear
motions depend on the location within the ship. Let us consider a location defined by
the coordinates (x 81 , x 82 , x 83 ) with respect to the centre of gravity. The longitudinal
displacement of this point includes contributions from the surge of the whole ship as
Sec. 13.8]
255
Absolute motions
well as the products of the lever arms and the pitch and yaw motions. If the angular
motions are small the longitudinal displacement relative to the moving origin 0 is
positive forward
(13.7)
Similarly the lateral and vertical displacements of the point (xB 1, xB 2, xB 3 ) are
positive to starboard
(13.8)
positive down
(13.9)
5 ;)
metres
(i = 1,3)
(13.10)
metres
(13.11a)
Szo
= (P + Pi0) 112
metres
(13.1lb)
S3o
(13.11c)
Ps
tan 651 = p
(13.12a)
Pw
tan 652 = p
(13.12b)
(13.12c)
where
(13.13a)
(13.13b)
xB~ 40
xB~ 40
cos 64 + xB 1 x 60 cos 6 6
metres
(13.13c)
(13.13d)
(13.13e)
P12 =
x 30
(13.13f)
256
[Ch. 13
The form of the absolute motion transfer function depends of course on the
position considered in the ship. Fig. 13.22 shows some typical head wave absolute
vertical motion transfer functions for a point on the bridge of the frigate. In very long
waves (ro = 0) the transfer functions approach unity as the ship contours the waves.
At very high frequencies the motions become negligible; but at intermediate
frequencies the motion phases are such that the contributions from pitch and heave
are synchronised and large absolute motions, considerably greater than the wave
amplitude, are the result.
0
2.5
"'c0
.E
c
.2
~Ul
c
~
c
0
g
E
.g"'
Q;
>
Ql
:;
0
Ul
.Q
<1:
Fig. 13.22 Absolute motion transfer functions at the bridge of a frigate in head waves.
Fig. 13.23 Relative motion transfer functions for the bow of a frigate in head waves.
Sec. 13.10]
257
(13.14)
where s 3 is the absolute vertical motion given by equation (13.9) and~ is the local
wave depression given by equation (9.32). Substituting equations (9.32) and (13.9)
we find that the relative motion is sinusoidal with
(13.15)
where the relative motion amplitude is
(13.16)
and the phase is given by
(13.17)
In practice the presence of the hull causes a considerable distortion of the waves
close to the ship and equation (13.14) is only likely to be reliable at the forward
perpendicular. Further aft the equation may underestimate thd felative motion by as
much as 50%. Techniques for estimating this distortion or 'swellup' are still the
subject of research and no method has yet won universal agreement. Nevertheless,
equation (13.14) is still used to estimate relative motion, sometimes with empirical
corrections for the swellup as described in Chapter 20. Fig. 13.23 shows some typical
calculations for a point on the forefoot of the frigate in head waves. In very long
waves the relative motions are zero because the ship contours the waves. In short
waves the ship is essentially stationary so that the wave motion is the only sizable
contribution to the relative motion and the transfer function approaches unity.
At some intermediate frequency the absolute, motion phase is such that the
upward absolute motion is synchronised with the wave depression at the particular
location chosen for the calculation. The relative motion is then a maximum and
sharply peaked resonances can occur at high speed.
[Ch. 13
258
Xo
= Xo (J)e,
..
Xo
= Xo
2
(J)e
So the velocity and acceleration transfer functions for any motion can simply be
obtained by multiplying the displacement amplitude responses by the encounter
frequency and the square of the encounter frequency. Fig. 13.24 shows examples of
this procedure for heave in head and following waves.
2.0
Wave frequency oJ (radians/second)
Fig. 13.24 Heave displacement, velocity and acceleration transfer functions for a frigate at 20 knots.
Sec. 13.11]
13.11
259
Many seakeeping studies call for estimates of the motions experienced by passengers
and crew and inanimate objects within the ship. These motions are important in
determining seasickness and the ability of the crew to work effectively as well as
estimating the likelihood of unsecured objects sliding across the deck or toppling
over. These problems are discussed in more detail in Chapter 21, but for the time
being we may observe that they are essentially functions of the apparent accelerations experienced by the people and objects within the ship.
Consider, for example, an object of mass m tonnes on the deck of a ship as shown
in Fig. 13.25. If the object is at (xB 1, xB 2,xB3) the absolute lateral and vertical motion
I //1
I
s3.
Ss3;
I
I
Fig. 13.25 Apparent forces on an object on the deck.
displacements relative to the mean track of the ship are given by equations (13.10).
Following the procedure outlined in Section 13.10 the lateral and vertical accelerations, relative to the Earth are given ~y
positive to starboard
(13.18)
positive downwards
(13.19)
260
[Ch. 13
= Sz cos x 4 + s 3 sin x 4
sB 3
= s 3 cos x4 
s2
metres/second2
in the plane of the deck to starboard
sin x4 metres/second2
downwards normal to the deck
These accelerations of the ship will tend to lift the object off the deck and slide it
towards the port side. The object is also subject to the components of gravity
resolved in directions normal and parallel to the deck. The total apparent force
experienced by the object in the plane of the deck is
msB 2  mg sin x4 kN
to port
When the ship is upright the gravity force on the object is mg kN downwards, but this
is reduced to mg cos x 4 kN normal to the deck when the ship is rolled. So the total
apparent force normal to the deck is
upwards
s AZ
sA3
= sB 2  g sin x 4
= s2 cos x4 + s3 sin x4 =
sB 3 
g (1 cos x 4) =
s3
g sin x 4
cos x4 
metres/second 2
s2
to starboard
= Sz 
gx4 metres/second 2
sA3
=s3
metres/second2
to starboard
(13.20)
So the apparent acceleration normal to the deck is the same as the absolute vertical
acceleration. The apparent acceleration in the plane of the deck is often called the
Sec. 13.11]
261
lateral force estimator (LFE). Substituting equations (13.2) and (13.18) in equation
(13.20) we find that the LFE in regular waves is sinusoidal with
metres/second 2
positive to starboard
~
Q)
E
Q;
Cl.
'bc:
0
()
Q)
"'
0;
~
Ql
.S
0
J.J>
6N
..:
'"'
Q)
"0
::J
.'!:
c.
E
"'
Q)
>
"'
~
_;
"0
1.0
c.
E
w
"'
LL
...J
2.0
Wav.e frequency w
(radians/sec~nd)
Fig. 13.26 LFE transfer functions for a frigate at 20 knots;xB 1 = 40 m, xB2 = 7 m, xB3= 10m.
262
[Ch. 13
where
P13 =  s20
P14 =
ro~
 Szo ro~
metres/second2
(13.21a)
metres/second2
(13.21b)
The LFE varies throughout the ship and the LFE transfer function will depend on
the location chosen for the calculation. Fig. 13.26 shows an example for a particular
location on the frigate at 20 knots.
13.12 NONLINEARITIES
The foregoing discussion of the characteristics of motion transfer functions presupposes that the motion responses are linear: that is, the motion amplitude at any
particular speed, heading and wave frequency is directly proportional to the wave
amplitude. Indeed this assumption is implicit in the definition of the transfer
function.
The equations of motion (8.23)(8.28) will only yield linear motion responses if
all the coefficients aii' bii and cii are independent of the motion amplitude. In
principle this ~equires negligible changes of underwater hull form as the ship
proceeds through the waves: in other words the relative motion amplitudes must be
small.
If the motion amplitudes are large the underwater shape may change considerably. Consider, for example, the relative motion at the bow of a ship in head waves
(Fig. 13.23). If the motion exceeds the local draught of the ship the keel will emerge
from the water during the upward part of the motion cycle. The local excitation and
all the hydrodynamic reactions will temporarily disappear until the keel reenters
and they resume finite values. This might be expected to have a dramatic effect on the
motion responses in these conditions but it is found that the nonlinearities associated with this kind of phenomenon are surprisingly weak. Perhaps this is because
both sides of the equations of motion are affected and the effects are only ~ssociated
with the extremes of the motion cycles. Extremely large motions are in any case only
of limited interest because they cannot be readily tolerated by the crew.
When the roll damping is very small, large roll motions may develop in quite
moderate waves and nonlinearities in the righting lever curve may become important. In other words the effective roll stiffness c44 will depend on the roll angle. The
roll damping itself will usually be dependent to some extent on the roll amplitude and
this will be a further source of nonlinear behaviour. These effects are not usually
14
Ship motions in irregular waves
14.1
For many years the assessment of seakeeping performance at the design stage
progressed no further than comparisons of ship motions in regular waves. The
shortcomings of this approach were widely recognised but further progress had to
await the development of new techniques first proposed by St Denis and Pierson
(1953). These methods were based on ideas developed in the electronics and
communications fields and it is no exaggeration to suggest that their introduction,
together with the development of strip theory, form the two main foundations of the
modern theory of seakeeping.
St Denis and Pierson suggested that the ship could be treated in much the same
way as the 'black box' electronic filter shown in Fig. 14.1. The input signal received
!d
Input
Waves

Filter
Output _
Ship
Motion _
264
[Ch. 14
by the filter contains a number of different frequency components and these are
amplified or attenuated to produce some modified output signal according to the
characteristics of the filter. For example, a socalled 'lowpass' filter will attenuate
the highfrequency components of the input signal and allow the lowfrequency
components to pass more or less unscathed.
The analogy suggests that the ship can also be regarded as a filter, not of electrical
signals, but of the waves. In other words we can think of the ship as a black box which
receives the waves as input and generates ship motions as output. Of course there are
a number of different ship motion outputs so we should really regard the ship as a
collection of filters, each with its own individual characteristics.
Let us consider the case of heave motion in head waves. Figure 13.6 shows typical
heave transfer functions for p, = 180 and we may regard these as defining the
characteristics of the 'heave filter' of the black box ship. We can see that this is
essentially a lowpass filter: at low frequencies the wave motions are translated into
corresponding heave motions without attenuation or phase shift. As the frequency
rises, the heave motions are reduced and at very high frequencies the input is
completely attenuated so that there are no resulting heave motions.
These ideas can be formalised and quantified by means of the so called 'spectral
calculation'. This is mathematically valid and rigorous provided that the ship motions
are directly proportional to the wave amplitude at any given speed, heading and
frequency. This is nearly always true, as we have seen, and the spectral calculation is
widely used in seakeeping calculations.
The first step is to determine the wave energy spectrum as described in Chapter 4.
For the time being we shall assume that the waves are long crested. The spectrum
may be measured but it is more usual to employ one of the idealised wave energy
spectrum formulae (equations (4.33) or (4.49)). These formulae give the wave
energy spectrum for a fixed point in the ocean: we require to transform this to the
reference frame of an observer on the moving ship.
We have already seen that waves are encountered by the ship at the encounter
frequency defined in equation (7.3). So the frequencies with which the waves are
encountered are increased in head waves and decreased in following waves. It
follows that the wave energy spectrum must be shifted along the frequency axis to
cover a different range of frequencies when observed from a moving ship.
Fig. 14.2 illustrates the result obtained in head waves: every wave frequency is
transformed into a corresponding encounter frequency according to equation (7 .3).
The frequency interval oro centred on the wave frequency ro transforms into a
corresponding encounter frequency interval oroe. The relationship between the two
intervals is obtained by differentiating equation (7.3):
droe = _ 2roU cos p,
1
dro
g
or
Sec. 14.2]
265
bw
Areas equal
Corresponding
Encounter spectrum
Fig. 14.2 Transforming the wave energy spectrum into the encounter spectrum.
<>roe = ( 1
2
: U cos IL )oro
radians/second
(14.1)
Now we have seen in Chapter 4 that the area under the wave energy spectrum
bounded by the frequency interval oro is proportional to the energy contained within
that band of frequencies. Transforming the spectrum to the moving frame of
reference of the ship does not change this energy and it follows that the area within
the wave frequency range oro must be exactly reproduced as an equal area within the
corresponding encounter frequencyrange <>roe. Hence the ordinates of the wave
spectrum and its counterpart in the encounter frequency domain must be related by
(14.2)
266
[Ch. 14
dro
Sr,(ro)d
roe
S (ro)
g
metres 2/(radian/second)
r,
g 2roU cos tt
(14.3)
14.3
The next step is to generate a motion energy spectrum by filtering the encountered
wave energy spectrum with the appropriate motion transfer function. This is
achieved by multiplying each spectral ordinate of the encountered spectrum by the
square of the motion transfer function at the corresponding encounter frequency.
This approach is valid and appropriate for any ship motion when the transfer
function is normalised by dividing by the wave amplitude (e.g. surge, heave, absolute
motion, relative motion, etc.). A typical calculation for heave motion at 20 knots in
head waves is shown on the righthand side of Fig. 14.3. The heave motion energy
spectrum ordinates (Fig. 14.3(f)) are given by
(14.4)
Sec. 14.3]
Wave frequency
domain
Encounter Frequency
domain
6
(a)
267
(b)
()
Q)
!!'_
""0
1
2
J;
0
0.5
1.0
1.5
1.5
1.5
(c)
0.5
1.0
1.5
(d)
1.0
,'j.
'a
.;;:
.;;:
0.5
0.5
1.0
1.5
1.5
6
(f)
(e)
()
Q)
!!'_
""0
J
0
<J (radians/second)
'" (radians/second)
described above. For pitch motion the pitch energy spectrum (Fig. 14.4(f)) is given
by
(14.5)
and similar expressions apply for other angular motions.
Chapter 4 derived various formulae (equations (4.3), (4.17) and (4.22)(4.28))
for the statistical quantities associated with wave energy spectra and these may be
268
Wave frequency
domain
[Ch. 14
Encounter frequency
domain
o.ooar........"""
(a) Wave slope spectrum
::::0.006
T0 =12.4 Sec
Hy =5.5m
3
(.)
Q)
~
'0
0.004
1.5
1.5
1.5
(c)
(d)
1.0
0
JJ
JJ
""',
><
)(
0.5
1.5
(e)
2.0
(f)
Q)
0.004
i;
2.0
''' (radians/second)
<>~e
(radians. second)
applied with equal validity to ship motion energy spectra in the encounter frequency
domain. In particular the rms motion displacement is given by
cr0
= \/m0
metres or radians
(14.6)
Sec. 14.4]
269
where m 0 is the motion variance given by the area under the motion energy
spectrum:
(14.7)
Similarly the rms velocities and accelerations are given by
<Jz
cr4 =
ym4
(14.8)
metres/second2 or radians/second2
(14.9)
(14.10)
(14.11)
and in general the nth moment of area of the motion energy spectrum is
.
The associated average periods of the motions are theri given by equations
( 4.26)(4.28).
14.4 ALTERNATIVE METHOD OF CALCULATING MOTION STATISTICS
The motion energy spectrum in the encounter frequency domain may be transformed back into the wave frequency domain as shown in Figs 14.3(e) and 14.4(e).
The ordinates in the wave frequency domain are calculated by rearranging the
transformation formula given by equation (14.3). With appropriate change of
notation this becomes
(14.13)
or radians2 /( radian/second)
270
[Ch. 14
The area under the curve in the wave frequency domain is of course the same as the
area under the motion energy spectrum in the encounter frequency domain. So the
motion variance may also be obtained by integrating in the wave frequency domain:
m0
J:
(14.14)
The ordinates of this transformed motion energy spectrum could have been
obtained much more simply from a direct calculation in the wave frequency domain
with none of the complication of transforming from one frequency domain to
another.
For motion transfer functions normalised by dividing by wave amplitude:
Sx(ro) =
(14.15)
and for motion transfer functions normalised by dividing by wave slope amplitude:
Sx(ro)
S"'(ro)
(:;J
radians2/(radianlsecond)
(14.16)
metres 2/(radianlsecond)
or radians 2/(radianlsecond)
and using equations (14.12) and (7.3) the spectral moments are
oo (
ro 
metres2/second n
or radians 2/secondn
(14.17)
This may then be used with equations (14.8) and (14.9) to calculate the required
motion statistics.
Sec. 14.5] Effect of matching the wave spectrum and the transfer function
271
S (ro)
x
g
g  2ro U cos fJ
metres 2/(radian/second)
or radians 2 /(radian/second)
14.5
(14.18)
High transfer function ordinates occurring at frequencies with a good deal of wave
energy will give large contributions to the motion energy spectrum. It follows that the
rms motion depends on the extent to which the motion transfer function 'matches'
the wave spectrum. Fig. 14.5 shows as an example the effect of varying the modal
10
10
Q)
:a
T0 =9.7 sec
18
'J1
0.5
1.5
"0
~
Fig. 14.5 Effect of matching.the wave energy ~ectrum with the transfer function, Relative
motions at 20 knots, H 113 = 5.5fnetres.
272
[Ch. 14
period of the wave energy spectrum on the rms relative motion for the forefoot of a
frigate in head waves.
Fig. 14.6 shows these and other motions plotted as a function of modal period.
en
E:
Q)
Head waves
E
.....
c
QJ
E
QJ
(,)
"'
c.
.~
'0
Heave
(/)
Relative motion
at bow
25
0
Modal period T0 (seconds)
4
en
QJ
E:
Cl
QJ
.....
<.:
QJ
E
QJ
(,)
"'c.
.~
'0
head waves
(/)
25
0
Modal period T0 (seconds)
Fig. 14.6 Effect of modal period on ship motions of a frigate at 20 knots; H113 = 5.5 metres.
The motions which are most sensitive to modal period are those, like relative motion
and roll, which have distinct transfer function peaks.
14.6
The procedures outlined above may be used to calculate motions in irregular long
crested waves. These are rare, as we have seen, and it is sometimes necessary to
extend these techniques to cope with more realistic short crested waves.
When the waves are short crested the 'total' wave energy spectrum S~;(ro) is
Sec. 14.6]
273
obtained by integrating the spread wave spectrum S~(ro,v) over the range of
directions from  Vmax to + Vmax (see equation (4.58)). Now it was shown in Chapter
4 that the continuous short crested wave spectrum S~(ro,v) could be represented by a
finite number of reduced long crested spectra distributed around the predominant
wave direction (see Fig. 4.15). Each long crested spectrum is given by
W S~(ro)
where W is a weighting factor depending on the secondary wave direction v  p, given
in Table 4.1.
Table 14.1 Specimen calculation of roll motion in short crested waves
v /J(deg)
(predominant)
90
75
60
45
30
15
0
15
30
45
60
75
90
mo
(deg)
45
30
15
0
15
30
45
60
75
90
105
120
135
0.000
0.011
0.042
0.083
0.125
0.156
0)67
0.156
0.125
0.083
0.042
0.011
0.000
rl
omo
(deg2)
(deg 2)
36.24
12.11
2.92
0.00
2.92
12.11
36.24
73.27
18.23
8.64
4.34
2.72
1.42
0.00
0.13
0.12
0.00
0.37
1.89
6.05
11.43
2.28
0.72
0.20
0.03
0.00
= 45.
It follows that the contribution to the motion variance from each secondary wave
direction is
...
and the total motion variance in short crested wave2 is obtained by summing the
contributions from all the reduced long crested wave spectra.
274
[Ch. 14
Table 14.1 gives the results of a specimen calculation of the rolling motion of a
frigate at 20 knots in quartering waves (JL = 45). In long crested waves the rms roll at
this heading is 6.03 for the particular wave spectrum used in the calculation. Cosine
squared wave spreading reduces this to only 4.75. Fig. 14.7 shows the effect of wave
...c
Q)
E_

Q)rn
t.lQ)
_..,.
CO'Q.Q)
E
c
.!!!
Heave
rn
E
...
180
0
Heading 11 (degrees)
10
(i)
Q)
Q)
,.
I \
e
Q)
...
::!:!.
cQ)
E
Q)
Roll
I
I
\
t.l
co
c.
rn
'0
rn
E
0
Heading 11 (degrees)
...c
Q)
EQ)Vl
t.l Q)
coQ)
C..Cl
...
'
VlQ)
':
rn
E
...
II
\
\
"C
c_
180
Heading 11 (degrees)
Fig. 14.7 Effect of wave spreading on ship motions; ship speed 20 knots, H113 = 5.5 metres,
T0 = 12.4 seconds.
spreading on heave and pitch as well as on roll. In general, wave spreading smooths
out the more extreme variations of the motion. The effects are small for heave but
quite dramatic for roll. Wave spreading results in significant roll motions in following
Sec. 14.7]
275
waves and, to a lesser extent, in head waves. It also reduces the roll motions at the
worst heading by a considerable amount. In the same way spreading increases
the pitch motions in beam waves, but the effects at other headings are less
pronounced.
14.7
The procedures outlined above rely on the assumption that the motion responses are
linear with respect to the wave amplitude. We have seen that this is usually the case,
but roll motions may be an exception to this general rule. In this case a slightly more
involved procedure, illustrated in Fig. 14.8, is required.
I
I
I
Initial estimate of average roll
amplitude and period
I
Calculate equivalent roll damping ... Calculate average roll amplitude
I
Regular wave calculation
I
irregular wave calculation
I
RMS roll amplitude
average zerocrossing period
'
Difference acceptable
!
r
Final result:
select new calcualtion
conditions
r
It is first necessary to estimate or guess the rms roll expected in the particular
combination of speed, heading, significant wave height and modal period for which
the calculation is being performed. The averag~" roll amplitude may then be
276
[Ch. 14
estimated using equation (17.32). This done, the total equivalent linearised rolldamping coefficient may be calculated for the chosen roll amplitude using the
methods outlined in Chapter 12. The average frequency may be taken as the natural
roll frequency of the ship. The calculation then proceeds through the usual stages of
determining the motion transfer functions and combining these with the appropriate
wave spectra to obtain rms values. Wave spreading should be taken into account.
The rms roll results are compared with the initial guess: if the differences are large
(as will usually be the case) the calculation is repeated using the new rms value and an
improved estimate of the roll damping.
This procedure is repeated until the rms roll angle reaches an asymptotic value
and the calculation is terminated. The whole calculation must then be repeated for
every speed, heading, significant wave height and modal period required.
15
Seakeeping trials
15.1
FULLSCALE TRIALS
Fullscale seakeeping trials, in which the motions, deck wetness and other seakeeping phenomena of interest are monitored in a measured wave environment, may
seem to be an attractive method of assessing and comparing the performance of ships
in rough weather. The waves would, of course, be irregular and it would be necessary
to record them and the motions simultaneously and tg analyse the results using the
spectral analysis techniques discussed in Chapter 4. The rms motions (or other
seakeeping responses such as deck wetness frequency) could then be plotted as a
function of significant wave height and compared with those obtained from trials in
'
i II
other ships.
However, we have seen that the rough weather behaviour of a ship is a function
not only of the significant wave height, but also of the modal wave period, the shape
of the wave spectrum and the degree of wave spreading. So results obtained and
compared in this simple way are likely to be scattered and possibly misleading if the
wave conditions at the time of the trials were in any way dissimilar.
The only way that these problems can be overcome is by running two or more
ships side by side in simultaneous trials in nominally identical wave conditions. Trials
of this nature have been reported by Bledsoe, Bussemaker and Cummins (1960) and
by Andrew and Lloyd (1981). Fig. 15.1 shows the rms pitch motions and the mean
zerocrossing periods measured on two frigates in the latter trials in severe head
waves (significant wave heights 67 metres). These results form an objective
comparison between the motions experienced by two particular ships in a particular
rough weather environment: any pecUliarities or changes in the wave conditions
during the time of the trial must have been experienced in equal measure by both
ships (since they were only about 400 metres apart) so any differences in the motions
may be attributed to differences in the design of the two ships.
Trials with a single ship are infrequent _because of the expense involved, and
comparative trials involving two ships are even more_,9f a rarity. Such trials certainly
278
[Ch. 15
Seakeeping trials
3
U)
Q)
Q)
...
C)
Q)
~
.s::::
...
0
c.
rn
E
22
U)
o
c
0
Q)
!!!.
h'
o
0
;::
Q)
0.
C)
c
rn
Trial
u;
eu
6
Q;
Leander
6.
Tribal
Theory

c
co
Q)
~
22
Speed (knots)
Fig. 15.1 Comparative seakeeping trial results. (After Andrew and Lloyd (1981)).
Sec. 15.3]
279
(15.1)
with similar expressions for other linear motion transfer functions. In the same way
equation (14.5) may be rearranged to give
(15.2)
with similar expressions for other angular motion transfer functions.
t Registered Trade Mark. Manufactured by Datawell bv, Zomerlhstraat 4, 2012 LM Haarlem, The
Netherlands.
280
Seakeeping trials
[Ch. 15
gyros of the type used in aircraft navigation systems. In warships it is often possible to
use the ship's own weapon system gyros but on other ships the trials team will usually
have to supply its own transducers.
Sec. 15.4]
281
il
St
+ XBzX6 xB3is
metres/second 2
(15.3)
1
+ XB3i4 XBt i6
Xz
i3
= 53 
Sz
xB 2i 4
+ xB 1i 5
11
(15.4)
metres/second 2
(15.5)
metres/second
Seakeeping trials
282
[Ch. 15
15.5
We have seen in Chapter 4 that at least 100pairs of peaks and troughs are required in
an irregular time history in order to ensure a reasonably reliable estimate of the rms
motion. Each trial run must be of sufficient duration to achieve this minimum
standard. The actual length required may be estimated from strip theory calculations
of the motions in the wave spectrum expected during the trial. The mean period of
the peaks for each motion may then be calculated from equation (4.27). Fig. 15.4(a)
shows the results obtained for a trial planned in a frigate.
As expected, the mean periods are longest in following waves where the
encounter frequencies are low. The run time required to achieve 100 motion cycles is
given by
100TP
60
minutes
(15.6)
where TP is the mean period of the peaks for the chosen motion.
The required run time is given by the maximum value of T H for all the motions
and this is shown in Fig. 15 .4(b).
In practice longer runs than this absolute minimum are advisable. A certain
amount of additional time should be allowed for the ship to settle onto its new course
and speed at the beginning of each run: more importantly it should be realised that
every precaution should be taken to ensure that data of adequate quality are
collected. The opportunity of conducting a seakeeping trial occurs so rarely that it
283
Sec. 15.5]
Ill 111!1111
!I!
ill
JW+Leander
, I,
10
15
25
20
30
100
en
o
c
90
(.)
Q)
en
80
co
70
Q)
+'
c
en
en
Q)
c
+'
Q)
$:
.::.:.
(.)
Q)
o
Q)
Cl
co
.._
Q)
>
60
50
"
40
30
20
10
<(
12
14
16
18
20
Speed (knots)
Fig. 15.3 Measurements of deck wetness. (After Andrew and Lloyd (1981)). Reproduced by
permission of the Royal Institution of Na':~l Architects.
284
Seakeeping trials
en
"0
[Ch. 15
25
(a)
r:::
(.)
Q)
.!E.
"'
"
Ctl
Q)
c.
r:::
0
'_;j
E
Q)
0
"0
.Q
Qj
c.
r:::
Ctl
Q)
:;E
180
0
Heading (degrees)
60
(b)
en
Q)
"5
r:::
.E
Q)
'_;j
r:::
::J
a:
180
0
Heading (degrees)
Fig. 15.4 Minimum run times for 100 motion peaks; frigate at 20 knots, T0
= 12.4 seconds.
Sec. 15.5]
fl
285
Predominant
V wave direction
Ol
c:
.2
"C
::c
LL
co
Q)
B Wave buoy
20 km
10 miles
planned on this basis in the frigate at 20 knots. The wave buoy is launched at the
beginning of the head sea run and the course sequence is chosen to minimise the
distance from the ship to the buoy.
c
16
Model testing
Sec. 16.2]
287
of the motions which would have been experienced by the ship at full scale? To
answer this question we employ the techniques of dimensional analysis. These are
discussed in detail by Massey (1986) and we shall not give a general treatment here.
Suffice it to say that the technique allows the proper identification of the correct
model test conditions in terms of non dimensional groups of the quantities which are
relevant.
Let us consider as an example the heave motions of a ship and its model. For the
time being we may assume that we have no detailed knowledge of the physical
processes involved: even so we might surmise that the heave displacement amplitude
will be a function of the wave amplitude and frequency, the speed and heading and
the size, shape and inertias of the hull. In addition, the heave amplitude would be
expected to depend on the physical properties of the water (density and viscosity)
and the acceleration due to gravity. We might therefore write a general mathematical
expression relating these eleven quantities as:
(16.1)
where f 1 is some as yet undetermined function. [xB] represents a sufficient number of
coordinates to define the shape of the hull and [I] represents the moments of inertia
of the hull.
Massey (1986) shows how an expression of this form can be rearranged and
written in terms of a lesser number of nondimensional parameters. Many different
formulations are possible and equally valid but it is COJ;lvenient to consider the form:
X3o
fz
{So
L'
00
/(L) U
[xB] [I] pUL}
\j g 'y(gL)' p., L' pL 5 ' P.w
(16.2)
(16.3)
[Ch. 16
Model testing
288
Mass
(tonnes)
Length
(metres)
Time
(seconds
0
0
0
0
0
0
0
1
1
1
0
1
1
0
1
0
1
1
2
3
1
1
0
0
1
1
0
0
0
0
0
1
2
X3o
~0
(0
u
fJ
[xB]
[I]
p
f.Jw
_ pUL
RN 
(16.4)
f.Jw
dv
dv ds
dt
ds dt
dv
=vds
which is proportional to U 2/L. So the inertia force ( mass x acceleration) experienced by the particle will be proportional to pL 2 U 2 The gravity force on the particle
is simply (mass x g) which is proportional to pgL 3 So the ratio of the inertia and
gravity forces on the particle is
Sec. 16.2]
p,
289
/( iner~ia force)
\j gravity force
The surface area of the fluid particle is proportional to L 2 and the viscous shear
stress is proportional to #1w times the velocity gradient. The velocity gradient is
proportional to U/ L and the viscous force on the particle is therefore proportional to
11wUL. So the ratio ofthe inertia and viscous forces on the particle is
pU2 L
11wUL
pUL
11w
R=
(16.5)
where Lm and Ls are the lengths of the model and the ship.
Then for the dimensionless coordinates [x8 ]/L to be identical for model and ship
we require each corresponding dimension to be related by
Xsm
Xss
Lm
Ls
so that
~
Xss
Xsm
metres
(16.6)
In other words the model must be geometrically similar to the ship in all respects. The
underwater hull shape should be accurately reproduced and it is convenient to model
290
Model testing
[Ch. 16
the hull up to the weather deck. It is not usually necessary to represent the
superstructure as this has little effect on ship motions except possibly in very severe
conditions.
The requirement to represent the underwater hull form accurately demands that
the model's waterline be correctly located. This requires that the model's mass and
trim be correctly scaled. The model's mass is
(16.7)
where the integration is performed over the length of the hull. Now all the model's
dimensions must be scaled according to equation (16.6). So the model mass may be
written as
Pm ms
tonnes
Ps R3
(16.8a)
ms
R 3 tonnes
(16.8b)
Im
= Pm Is
Ps Rs
I
tonne metres 2
.
(16.9a)
(16.9b)
Sec. 16.2]
291
fifth power of the dimension ratio if the water densities are the same at model and full
scale. Again it is necessary to allow for differing water densities by using equation
(16.9a).
Taking the pitch moment of inertia as an example we may write
lssm
lsss
where k 5 m and k 58 are the pitch radii of gyration at model and ship scale respectively.
Substituting equations (16.8a) and (16.9a) we find that
kss
metres
(16.10)
and similar expressions may be derived for the radii of gyration appropriate to the
other angular motions. So the radii of gyration are, like other model dimensions,
reduced in proportion to the dimension ratio.
16.2.4 Wave scaling laws
In order to make the nondimensional wave amplitude '(, 0 / L the same for both model
and ship we require
= Sos
R
metres
(16.11)
1
rl
so that the wave amplitude, like other model distances, must be reduced in
proportion to the dimension ratio.
The model wave frequency is determined by the identity
and
rom
00 8
y'R radians/secondt
(16.12)
Model wave frequencies must therefore be increased in proportion to the square root
of the dimension ratio and are higher than the corresponding frequencies in ship
scale.
Since the wave length is given by
t It is assumed that model and fullscale values of g are the same.
292
Model testing
211'g
roz
metres
= As
R
metres
[Ch. 16
it follows that
(16.13)
and we see, as expected, that the model wave lengths are also reduced in proportion
to the the dimension ratio.
The wave period is
T
= 211' seconds
(f)
and it follows from equation (16.12) that model wave periods are reduced in
proportion to the square root of the dimension ratio:
Ts
yR
seconds
(16.14)
k =
211'
A
and the model and fullscale wave numbers are therefore related by
(16.15)
The model heading must, of course, be the same for both model and ship:
IJm
Ps
radians or degrees
or
(16.16)
Sec. 16.2]
us
y'R
metres/second
293
(16.17)
so that the model speed must be reduced in proportion to the square root of the
dimension ratio.
For Reynolds number identity we require
Ps Us Ls
f.Lws
Ps f.Lwm Us R
Pm f.Lws
UsR
metres/second
(16.18a)
(16.18b)
metres/second
Fortunately viscous forces do not play a very important role in ship motion
dynamics (except perhaps in the determination of roll damping). If this is the case,
the requirement to scale viscous forces in the correct proportion to inertia forces may
be waived and it is no longer essential to match model and full scale Reynolds
numbers. In any case this matching is not usually a practical proposition since it
would demand impossibly high model test speeds: for example, a ship speed of 30
knots at a model dimension ratio of 30 would demand a model test speed (equation
(16.18b)) of 900 knots! In contrast, Froude number identity requires reduced model
test speeds and this example would yield a much more practical model speed of about
5.5 knots (2.8 metres/second).
Neglect of Reynolds number can yield misleading results in certain specific
circumstances. If the Reynolds number is too iow (as is usual in ship model
experiments) the transition point will be too far aft and too much of the boundary
layer will be laminar. Flow separation is then more likely and the skin friction will be
too low. This may have some effect on the behaviour of the model. It is usual practice
to stimulate turbulence by roughening the model surface at the estimated transition
"
.,
t In principle it would be possible to achieve proper Reynolds number scaling if the model could be tested
in a fluid with decreased viscosity and/or increased density. No suitable fluid has yet been proposed.
294
Model testing
[Ch. 16
point to compensate for the neglect of Reynolds number scaling. This precaution
can, however, only be regarded as a palliative measure and it is always advisable to
adopt the largest practical model scale to reduce Reynolds number scaling problems
to a minimum.
16.2.6 Functional form of motion responses in regular waves
For a correctly scaled model equation (16.2) reduces to
(16.19)
where f 3 is an unknown function of the four listed nondimensional quantities.
In practice the nondimensional motion amplitude is often found to be essentially
independent of the nondimensional wave amplitude '(, 0 / L provided that the wave
amplitude is moderate. In other words the motion amplitude is linearly dependent on
the wave amplitude, as assumed in strip theory. If this is the case we may write
equation (16.19) in the simple.r form:
(16.20a)
In passing, it should be noted that the nondimensional frequency could be
expressed as a function of fhe nondimensional wave length or the nondimensional
encounter frequency. The functional relationships could then be written as
(16.20b)
or
(16.20c)
Sec. 16.2]
295
For other seakeeping responses it will generally be necessary to include additional quantities to give the denominator the same dimensions as the numerator. For
example, angular displacement amplitudes are usually nondimensionalised by
dividing by the wave slope amplitude to give quantities like
Scaling laws for seakeeping responses may easily be derived from these relationships. For example, if the nondimensional heave velocity is expressed as .X30 /ro1;; 0 we
may infer that in a properly scaled model experiment
showing that the heave velocity, like the forward speed, is reduced in proportion to
the square root of the dimension ratio. Similarly we may show that accelerations at
model and ship scale are identical:
and that model encounter frequencies are, like wave frequencies, increased in
proportion to the square root of the dim~nsion ratio. One important consequence of
this result is that the model's response to the waves appears, to the untutored eye, to
be too lively. A more realistic appearance can be obtained by recording the model's
motions on film or video tape and playing back at a reduced speed. The playback
speed should be reduced in prDportion to the square root of the dimension ratio.
Table 16.2 gives a comprehensive list of scaling f~ctors for model test conditions
and responses.
Model testing
296
[Ch. 16
Examples
Mass
Length
Ship mass
Ship length; all dimensions; surge, sway, heave, absolute and
relative displacements; wave amplitudes and wave lengths
Time
Velocity
lt\/R
1/"yR
Acceleration
Surge, sway, heave, absolute and relative accelerations; acceleration due to gravity
Angle
Angular
velocity
Angular
acceleration
Pressure and
stress
Frequency
Force
Moment
lm/(I,R3)
lm/(I,R4)
(16.22)
In other words the nondimensional rms heave is a function of the nondimensional significant wave height and modal period, the Froude number and the
heading. It is easily seen that the model significant wave height must therefore be
Sec. 16.4]
297
scaled in proportion to the dimension ratio, and the modal period must be reduced in
proportion to the square root of the dimension ratio. For moderate significant wave
heights the functional dependence on the nondimensional significant wave height is
often weak and the rms heave motion for a given modal wave period, Froude number
and heading is then directly proportional to the significant wave height.
298
Model testing
[Ch. 16
vertical rod which slides in linear bearings, allowing the model freedom to heave. No
surge motion is allowed.
A limited freedom to surge can be obtained using the sprung arrangement shown
in Fig. 16.2(c). True surge motions can only be obtained ifthe model is selfpropelled
and the arrangements shown in Figs 16.2(d) and 16.2(e) are possible.
It is of course necessary to use very high quality bearings in these rigs to reduce
frictional effects on the measured motions to an absolute minimum. The weight of
the rig is supported by the model and due allowance for this must be made when
ballasting and trimming. Frictional effects may be virtually eliminated by dispensing
with any guidance arrangements as shown in Fig. 16.2(e).
In recent years the seakeeping basin, specifically designed for seakeeping model
tests, has been introduced. These tanks are usually of the order of 50 metres square
and are fitted with wave makers and a beach at opposite ends. An ideal arrangement
is shown in Fig. 16.3. A main carriage spans the tank and runs on rails in much the
same way as on the traditional towing tank. A subcarriage is mounted on the main
carriage so that it may be positioned at any point over the water surface. During a
selfpropelled experiment the subcarriage's position may be maintained over the
model by an automatic control system. Alternatively for towed experiments the
subcarriage may be driven across the tank at some predetermined heading to the
waves.
For selfpropelled experiments the model is connected to the subcarriage by an
umbilical cable as shown in Fig. 16.2(e). This cable supplies the model with electrical
power for its propulsion and instrumentation and also serves to feed the model's
response signals back to the carriage for recording and analysis. The cable is
299
Sec. 16.4]
t
~r
II
II
h..._____
~
(c) Free to surge, heave and
pitch (unpowered model
restrained by springs)
supposed to be sufficiently light and flexible to preclude any interference with the
model's motions.
This ideal arrangement is of course very expensive and many tanks have no
carriage, relying instead on free running selfcontained models as shown in Fig.
16.2(f). This means that the model mus~ carry its own batteries for power supply and
that its responses must be recorded on board or telemetred ashore. In either case the
additional weight makes the achievement of proper mass and inertia scaling more
difficult and large models may be necessary to allow adequate freedom of ballast
adjustment.
300
[Ch. 16
Model testing
Beach
M'
/
am carnage
",~)
,~" 4
Subcaiiage
Wave maker
301
Sec. 16.5]
(c) Doubleangle
wedge
,,
I)
II
II
If
ff
(f) Pneumatic
Drive signal
Wave maker
servo
Wave maker
Wave maker
motion
'
Waves
302
Model testing
[Ch. 16
'Aid
Beach slope
c
Q)
c::;
i:Q)
0
"c0
u
Q)
;;::
Q)
cr
1.8
2.0
"'\/(dig)
Fig. 16.7 Typical beach reflection characteristics. (After Hsiung et al. (1983).)
~
plotted as a function of the nondimensional wave frequency ro y(d/g) (where dis the
depth of the tank). The results show that the beach is most effective at high
Sec. 16.6]
Instrumentation
303
frequencies (i.e. in short waves) and the best results are obtained when the beach
slope is very small. Such a shallow beach may well occupy a significant proportion of
the length of the tank and practical considerations may place a limit on the beach
slope which can actually be used.
Over the range of wavelengths of interest to most model experiments the
reflection coefficient is usually of the order of 0.050.10. These reflections will mix
with the incident waves and spoil their characteristics. It is therefore important to
ensure that measurements are taken before these unwanted reflections reach the
model. This is discussed in more detail in Section 16.9.
16.6 INSTRUMENTATION
Where a carriage is available this provides a convenient datum for the measurement
of ship motions. Some examples of commonly used techniques for pitch and heave
are shown in Fig. 16.8 and adaptations of these systems are used for the other
2x,
Model testing
304
x3
Xs
St+Sa
2
Sa sf
[Ch. 16
metres
(16.23)
radians
(16.24)
Xr
where St and sa are the absolute motions measured forward and aft and 2xr is the
longitudinal separation of the two measurement locations.
Another technique is illustrated in Fig. 16.8(b). Here the heave is measured
directly by monitoring the motion of the heave post and the pitch is obtained by
coupling a potentiometer to the pitch pivot pin in the model.
These arrangements are suitable for the towing and guidance arrangements
shown in Figs, 16.2(b)16.2(d). Where there is no physical connection {apart from
the umbilical cable) between the model and the carriage, systems like that shown in
Fig. 16.8(c) have found favour. Here the pitch is measured by a gyro of the type used
in aircraft navigation systems and the heave is monitored using a sonic transducer.
The transducer emits a shortduration pulse of highfrequency sound (above the limit
of human perception) and this is reflected from a horizontal board mounted under
the carriage. The time required for the sound to travel from the transducer to the
board and back again is monitored and is proportional to the distance from the
transducer to the board.
Where there is no carriage, heave and other linear ship motions cannot be
measured directly because no convenient datum level is available for use as a
reference. It is then necessary to resort to accelerometers as in full scale ship trials
and this is illustrated in Fig. 16.8(d).
Two kinds of wave transducers are shown in Fig. 16.9. Both involve a pair of
metallic surface piercing elements. Electronic circuits are used to monitor the
resistance of the water between the two elements and this is a function of the depth of
immersion. Alternatively the elements may be regarded as the plates of a capacitor
using the water as the dielectric medium. The capacitance is then monitored to
provide an analogue of the depth of immersion.
These wave probes have some disadvantages. Surface tension effects may cause
the water level experienced on the probe surface to be slightly different to the true
level away from the immediate vicinity of the probe. The errors due to this effect are
not, however, very serious unless the probes are used to measure very small waves.
Much more significant effects are experienced if these surfacepiercing probes are
used on a moving carriage to measure the waves encountered by the model. The
probe inevitably causes some surfaye disturbance and this is likely to introduce errors
in the measured wave profile dueio the probe's own 'bow' wave. Speeds in excess of
12 metres/second may introduce noticeable errors.
More sophisticated wave probes which avoid contact with the water surface have
been developed. One type uses transducers to detect an ultrasonic pulse reflected
from the water surface. Optical systems using lasers are also being considered and
the 'servo needle' is being developed in Japan. This uses a servocontrolled probe
which is continually adjusted so that it is just in contact with the water surface.
Instrumentation
Sec. 16.6]
(a) Foils
305
It is advisable to position the wave probe perhaps qne metre ahead and one metre
to one side of the model to avoid measuring any surface disturbance caused by the
model. Care should be taken to ensure that any surface disturbance due to the probe
does not interfere with the model.
Fig. 16.10 shows some of the types of instrumentation ti~d to measure relative
bow motion. Most of the transducers are developments of those used to monitor
waves. The simplest form, using a pair of foils ahead of the m<;>del, is shown in Fig.
16.10(a). This gives a general indication of the relative motion ahead of the bow but
is positioned so that the disturbance due to the proximity of the hull may not be
measured.
This problem is overcome with the arrangement of flush mounted tapes shown in
Fig. 16.10(b). Here an aluminium foil tape is fixed directly to the surface ofthe model
using doublesided adhesive tape. Tl;J.e aluminium is insulated from the water by a
layer of 'Teflon' tape. The aluminium and the w~ter form the plates of a capacitor,
the Teflon tape being the dielectric. The electrical circuit of the capacitor is
completed through an uninsulated aluminium tape on the surface of the hull, and this
allows the capacitance, and hence the relative motion, to be monitored by suitable
circuits. Insulated tapes may be located at a number of stations, allowing the
measurement of the longitudinal variation of relative motion. A single return tape
will suffice for several measuremennocations.
Measurements with this arrangement will certainly include the effects of the
disturbance due to the hull Q,ut may suff<:<r from the surface tension effects experienced by wave probes. The rig shown in Fig. 16.10(c) has been proposed as an
Model testing
306
[Ch. 16
Aluminium foil tape
Teflon tape
alternative. Here resistance wires are stretched taut from keel to deck. The wires do
not touch the hull surface so they should not be subject to the unwelcome effects of
surface tension: at the same time they should be sufficiently close to the hull to give
measurements which include the local wave disturbance effects due to the hull.
Experiments to investigate deck wetness and slamming are also of some importance and suitable transducers to monitor deck wetness are shown in Fig. 16.11. In
Sec. 16.7]
Model materials
307
Fig. 16.11(a) a vertical plate incorporating pressure sensitive cells is mounted on the
forecastle and may be used to measure both the impact pressures and the frequency
of deck wetness. The cells are covered with a thin flexible diaphragm and are
connected to pressure transducers mounted on the carriage by lengths of flexible
tubing. An alternative technique using resistance wire probes is shown in Fig.
16.1l(b). Here the probes measure the depth of water on the deck but can again bt;
used to monitor the frequency of deck wetness.
In slamming experiments pressure transducers are usually mounted flush with the
model keel and under the bow flare to measure impact pressures directly.
William Froude developed the technique of building models from paraffin wax and
this is used in some establishments to this day. This somewhat unlikely material has
many advantages. It is easily worked and models can be modified at any time using
simple hand tools. After a model's useful life has expired it can be melted down and
the wax reused. However, wax models are not very robust: indeed they will gradually
distort if left unsupported over a prolonged period. This can be avoided by keeping
them submerged if they are required for future experiments. This is, however, hardly
convenient since all the internal equipment must be removed. A harder wearing
material is usually chosen for seakeeping model experiments.
Wood is a favourite material, being easily worked and durable. The usual
technique is to cut out a series of boards to the shapes of the hull waterlines and to
assemble these in the so called 'bread and butter' construction. The excess material is
removed by hand using templates for guidance in the final stages. The inside of the
model is carved away to allow room for the required equipment and to reduce
' il
weight.
It is always necessary to allow a considerable wall thickness to ensure adequate
strength in a wooden model and the usable internal space is often restricted. Wooden
models are often quite heavy and only a small amount of additional ballast is needed
to bring them up to the required mass. There is therefore little freedom to position
the ballast to obtain the proper inertias and centre of gravity. Wood also has the
disadvantage that it always seems to absorb moisture from the water, no matter how
well it is waterproofed. So dimensional stability is difficult to maintain as the wood
swells and subsequently contracts as>it dries out.
For this reason alternative modern materials are often used, particularly for free
running models which must carry their own batteries for power supply. A favourite
material is glass reinforced plastic (GRP). The shell of the hull need then be only a
few millimetres thick and a very light yet strong and stiff model can be produced. It is
first neccessary to build a plug (male) mould to the exact finished shape of the hull
and this is conveniently done in paraffin wax. The plug is coated with a release agent
and the female mould is built up on the plug with layers of glass cloth impregnated
with resin. When the resin has cured, the two moulds are separated and the plug is
discarded. The inside of the female mould is then coated with the release agent and
layers of resinimpregnated cloth are built up to form the finished GRP hull. After
curing, the hull and the mould are separated and the hull is ready for fitting out. If
308
Model testing
[Ch. 16
required the mould can be used again to reproduce any number of exactly identical
models.
Expanded polystyrene foam has also been used in some laboratories. This is very
light and stiff and has many of the advantages of glass reinforced plastic.
16.8
The first step in trimming and ballasting a model is to weigh the hull, complete with
all internal fittings such as instrumentation, batteries, propulsion motors, etc. A
weight to represent the weight of any towing or restraint apparatus to be supported
by the model should be included. The additional ballast required to bring the model's
mass up to the required value may then be calculated and the necessary ballast
weights stowed in the hull. The model may then be placed in the water and the
positions of the ballast weights adjusted until the required trim is obtained. The
model is usually required to have no heel angle and this can be checked by a spirit
level sited on some suitable datum surface. The longitudinal trim is best determined
by simple adjustable trim gauges of the type illstrated in Fig. 16.12. These allow the
freeboard to be determined at specified locations forward and aft. If the model's trim
is correct it follows that the longitudinal position of the centre of gravity must be
correctly located. It remains to determine the vertical location of the centre of gravity
and to adjust it if necessary. This is done by means of the inclining experiment. A
small measured heeling moment is applied to the model and the resulting heel angle
is measured. The moment is most easily applied by moving a known weight a
measured lateral distance. This enables the solid metacentric height GM5 to be
determined (see equation (10.8)). The position of the metacentre M will be known
from the ship's hydrostatic diagrams and this allows an estimate of the vertical
location of the centre of gravity. The VCG can be adjusted by appropriate vertical
movement of ballast weights.
Sec. 16.8]
309
(a) Pitch
(b) Roll
Fig .. 16.13 Compound pendulum rigs to determine pitch and roll radii of gyration.
lrJ
technique for measuring the pitch moment of inertia Iss The model is suspended in a
light frame so that the centre of gravity ish metres below a pivot point. The entire rig
is then oscillated by hand and the natural period of oscillation T. determined by
measuring the time required for, say, ten complete oscillations. The total moment of
inertia of the complete rig is, by the parallel axis theorem,
tonne metres 2
(16.25)
g kN~metres/radian
where
mp is the mass of the suppot.ting frame in tonnes
lp is the mass moment of inertia of the supportingJrame in tonne metres2
(16.26)
Model testing
310
[Ch. 16
Fig. 16.14 Bifilar suspension rig for estimating yaw radius of gyration.
hp is the distance from the centre of gravity of the supporting frame to the pivot
point in metres.
'
Now, from equations (6.8) and (6.21), the oscillation frequency is approximately
ro. =
~ = ~ ( ~)
radians/second
(16.27)
(16.28)
An exactly similar procedure is used for finding the roll radius of gyration as
shown in Fig. 16.13(b). In both cases it is desirable to keep the mass and inertia of the
supporting frame as small as possible to minimise errors in the estimation of the hull's
characteristics. The inertias of the frame may be found by measuring its natural
period of oscillation without the hull attached.
A somewhat simpler procedure for finding the yaw radius of gyration without
using a supporting frame is shown in Fig. 16.14. The model is suspended on two wires
Sec. 16.8]
311
from a suitable overhead beam. Typically the wires will be five or six metres in
length. The model is oscillated in yaw, taking care to avoid roll or sway motions. The
natural period of oscillation is recorded as before. The stiffness of the system is
calculated as follows.
Suppose that the model is yawed through a small angle x 6 radians as shown. Then
the wires will swing through a small angle
radians
Now the tension in each of the two wires must be half the model weight:
mmg kN
2
and the horizontal component of these forces tending to swing the model back to its
equilibrium position is approximately
kN
kN metres
mmgx;
h
(16.29)
kN metres/radian
The moment of inertia of the wires supporting the model is negligible and tpe natural
frequency of the system is, from equations (6.8) a_nd (6.21),
ro. =
~ = ~ ( m: k~) = ~ ( ~) Z:
radians/ second
(16.30)
(16.31)
Model testing
312
[Ch. 16
The bifilar suspension method is widely used to estimate the yaw radius of
gyration because of its simplicity and convenience, requiring no more apparatus than
a stopwatch and a pair of wires suspended from hooks in a suitable overhead beam.
In principle the same method could also be used to estimate the pitch radius of
gyration with the model turned on its side. This is, not usually practical, however,
because much of the internal equipment is not sufficiently well secured. Instead it is
often assumed that the pitch radius of gyration is the same as the yaw radius of
gyration.
The model's radii of gyration may be adjusted to the required values by moving
the internal ballast weights. The radii of gyration may be reduced by moving ballast
towards the middle of the model and vice versa. Care should be taken to ensure that
any adjustment to ballast on one side (or end) of the model is exactly balanced by a
corresponding adjustment at the other. Otherwise the location of the model's centre
of gravity will be changed. It is good practice to check the centre of gravity position
after swinging the model to ensure that all is well in this respect.
16.9
metres
Now the waves are overtaking the model with a relative velocity
c  U cos
IL
metres/second
and a wave trough recorded at the probe would have been alongside the model's
centre of gravity at a time
Sec. 16.9]
313
c U cos
p,
seconds
earlier. So the phase lead measured with reference to the waves recorded at the wave
probe should be reduced by an amount
op
(l)e
tp
p,
radians
(16.32a)
[Ch. 16
Model testing
314
21T (
)
T
x 1P cos 11 x 2 P sm 11
radians
(16.32b)
(16.32c)
radians
(16.32d)
radians
energy and that these reflections will eventually spoil the characteristics of the waves
generated by the wave maker. Model experiments run in these contaminated waves
will give misleading results. This problem can be avoided by careful attention to the
timing of the experiment run in relation to the time at which the wave maker is
started.
Consider the model in the experiment tank shown in Fig. 16.16. Now it was
Beach
1
rReflected wave
disturbance
UG
Wave front
tyu
/7
XT
(/
L.....
Wave maker
Fig. 16.16 Wave fronts and model location.
Sec. 16.9)
315
shown in Chapter 3 that the main body of waves having the proper wave amplitude
propagates down the tank at the group velocity. This is preceded by a secondary
wave front of reduced amplitude travelling at the wave celerity. The experiment run
(i.e. that part of the model's run in which measurements of its behaviour are
recorded) must obviously be confined to the part of the tank which contains waves of
the proper amplitude. In other words the model's responses should only be observed
and recorded while the model is behind the advancing wave frontWW. At the same
time it is necessary to avoid taking measurements of the model's responses after it has
encountered the initial wave disturbance RR reflected from the beach.
This may be analysed with the aid of the distance/time diagrams shown in Fig.
16.17. In these diagrams xT is the distance in metres from the wave maker and tis the
(/)
Q)
t (seconds)
t (seconds)
time in seconds. The wave maker is started at t = 0 seconds and the model progresses
'down' the tank at a component velocity U cos IL metres/second.
The path of the initial wave .disturbance is represented by the line 0 A with slope c
metres/second. This initial disturbance will be reflected from the beach at time
Model testing
316
tL
= LT
c
seconds
[Ch. 16
(16.33)
(where LT is the effective length of the tank) and will arrive back at the wave maker
at time
2tL
seconds
(16.34)
The path of the 'proper' wave front is represented by the line OB with slope ua
metres/second. This reflects from the beach at time
t
2tL
seconds
(16.35)
(since the group velocity is half the wave celerity in deep water).
When the model is at a location represented by a point above the line OB it will
not be experiencing the proper wave amplitudes of the main body of waves. Similarly
a model located at a position represented by a point above the line AC will be
experiencing the unwanted reflected wave disturbance. It follows that the conditions
outlined above for valid model test observations in regular waves are only experienced when the model's location is defined by points within the triangle ODC.
Let us first suppose that the model is to be tested in following or quartering waves
so that
U cos f.t
> 0 metres/second
Ignoring the short distances needed to accelerate the model up to the required test
speed and to bring it to rest at the end of the run, the model's progress down the tank
can be represented as a line with a positive slope equal to U cos f.t metres/second. The
line may be positioned anywhere in the diagram depending on the time chosen to
start the experiment run, but the optimum locations, depending on whether U cos f.t
is greater or less than the group velocity, are as defined by the broken lines in Fig.
16.17(a).
If U cos f.t > ua the optimum track (giving the longest possible run) is shown by
the line ED and the run 'length' is given by
(16.36)
t = 2LT (2
3
c
u cos f.t
seconds
(16.37)
Sec. 16.9]
317
u cos f.L
3U cos f.L
seconds
(16.38)
If U cos f.L < u 0 the optimum track is shown by the line OF and the run 'length' is
given by
2U cos f.L
u cos f.L + c
(16.39)
and this is achieved by starting the model as soon as the wave maker is started at time
t = 0 seconds. The run time is then given by
TH
2LT
u cos f.L + c seconds
(16.40)
The case of head and bow wave experiments is shown in Fig. 16.17(b). U cos f.L is
now negative and the model's track is represented by lines with a negative slope: If
the model speed and heading are such that U cos f.L < . c, the optimum model track
is shown by the line DG and the run 'length' is given by
2
3
(16.41)
This is achieved by starting the model run at a point LT/3 metres from the beach at
time
4tL
(16.42)
seconds
3U cos f.L
seconds
(16.43)
For the case 0 > U cos f.L >  c the optimum m.odel track is shown by the line HC
and the run 'length' is now giwen by
2U cos f.L
(16.44)
318
[Ch. 16
Model testing
metres
seconds
(16.45)
2LT
= ::::=c 2U cos /.L
seconds
(16.46)
These results are summarised in Fig. 16.18. They show that even in the most
1.0
Beach
End of run
Start of run
f
::::'f 0.5
)(
Start of run
Wave maker
3
2
1
(U cos !l)lc
Follqwing and
quartering waves
favourable conditions no more than two thirds of the tank length is usable. In
practice slightly greater lengths of the tank may be used if the beach is very effective.
16.10
Tests in irregular waves call for a slightly more involved procedure. It is first
necessary to scale the required wave energy spectral ordinates and frequencies using
the scaling laws listed in Table 16.2. A typical result for a Bretschneider spectrum
Sec. 16.10]
319
with significant wave height of 5.5 metres and modal period of 12.4 seconds for a
dimension ratio of 36 is shown in Fig. 16.19(c). We require to drive the wave maker
150
(a)
u
en
=o
QJ
::'
~
0
2.
50
>
C/)
8
0.030
(b)
~
~
QJ
E
Wave maker
transfer function
1~1
0.03
(c)
u
en
=o
QJ
Wave energy
spectrum
0.02
::'
0.01
JJ
0
Model wave frequency w (radians/second)
320
Model testing
[Ch. 16
with an irregular electrical signal which will produce a train of irregular waves having
this wave spectrum. The spectrum required for the input signal may be deduced from
a knowledge of the wave maker transfer function. Using the spectral calculation
procedures described in Chapter 14 we may write the wave energy spectrum as
Sr, (ro)
Su(ro)
(~:r
metred(radian/second)
(16.47)
Sr,(ro)
volts 2/(radian/second)
(16.48)
1;,0 )
( Vo
A suitable time history having the spectrum given by equation (16.48) must then
be constructed using the wave synthesis techniques described in Chapter 4. Driving
the wave maker with this signal would then be expected to produce a wave time
history with the desired wave energy spectrum. In practice this simple technique may
not give results of adequate accuracy. This is believed to be because the wave maker
response suffers from poorly understood interactions between the many frequencies
present in the irregular waves being generated. These interactions are absent when
the wave maker is used to generate a singlefrequency regular wave. So the regular
wave transfer function can only be regarded as a first approximation to that required
to quantify the response in irregular waves. Moreover, the required transfer function
apparently depends on the particular time history being generated and not just on the
spectrum characteristics. So a different transfer function is required for every new
time history.
These difficulties can be overcome by empirical adjustments t<;> the wave maker
drive signal spectrum. Where the measured wave spectral ordinates are too low the
drive signal spectral ordinates should be increased and vice versa. It is usually
possible to achieve a good match to the desired wave spectrum with two or three
adjustments of this kind.
Each component of the system of irregular waves will propagate down the tank at
its own group velocity, preceded by an advance party of reduced amplitude waves of
that frequency moving at the appropriate wave celerity. So the lowest frequency
component, which has the highest group velocity and wave celerity, will overtake the
rest of the waves and arrive at the end of the tank well before the other frequency
components. Clearly the 'complete' wave spectrum will not be experienced at a given
location in the tank until the highest frequency component has arrived at that point.
By this time the lowest frequency waves may well have been reflected from the beach
and already be spoiling the waves propagating down the tank in the proper direction.
This can be avoided by introducing the frequency components to the wave maker
321
Sec. 16.10]
drive signal (and hence to the generated waves) in descending order. The highest
frequency is introduced first and subsequent components are included at times
specified to ensure that all frequency components arrive simultaneously at some
specified point in the tank.
Suppose that the required wave spectrum contains N frequency components
The wave component with the highest frequency roN will propagate down the tank
with the group velocity
uaN
2roN
(16.49)
metres/second
(16.50)
seconds
arriv~~~t
(16.51)
Obviously the pointxT at which the waves are required to coalesce, should be chosen
to maximise the length of the tank available for the experiment. Reference to Fig.
16.18 shows that for a given speed and heading the most critical conditions occur
when the wave celerity is highest. ln other words the available test length is small
when the wave frequency is lo.w. So the lowest frequency in the wave spectrum will
dictate the location of the coalescence point and this will also determine the
permissible run time. This is illustrated in the following worked example.
Model testing
322
[Ch. 16
Consider a model to be tested in head waves in a ship tank with a usable length
LT = 100 metres. The model's dimension ratio is R = 36 and the model test speed is to
represent a ship speed of 20 knots. The wave spectrum (at full scale) includes
frequencies in the range 0.31.6 radians/second and this is to be represented by
discrete frequency components at intervals of 0.1 radians/second. t
It is required to find the optimum run time and start position for the test and the
frequency component time lags for the wave maker drive signal.
It is convenient to work in ship scale: the tank length becomes
LT
3600 metres
U cos 11
= 
10.3 metres/second
The lowest frequency component (ro 1 = 0.3 radians/second) will have the greatest
celerity:
32.7 metres/second
Now 0 > U cos 11 >  c 1 and equations (16.44)(16.46) apply. So the start
position is given by
Xr
metres
or 38.7 metres
at ship scale
at model scale
Sec. 16.11]
323
c 1  2U cos J.t
135
seconds
or 22.5
at ship scale
seconds
at model scalet
The wave maker drive signal lags are, from equation (16.51):
Frequency
(radians/second)
(ship scale)
Lag
(seconds)
(model scale)
0.3
0.4
0.5
0.6
0.7
10.3
9.5
8.7
7.9
7.1
6.3
5.5
4.7
3.9
3.2
2.4
1.6
0.8
0.9
1.0
1.1
1.2
1.3
1.4
o.8
1.5
1.6
16.11
rl
0.0
An oscillating model acts like a wave maker and radiates waves on either side.
Indeed this is the mechanism responsible for dissipating energy and providing
motion damping. The model will of course oscill;te at the encounter frequency and
the generated waves will radiate away from the model at the celerity appropriate to
this frequency. The waves will eventually reach the tank walls and be reflected back
towards the model, as shown in Fig. 16.20.
~
t Such a short run time will be insuffiCient to ensure stable motion statistics. A run time sufficient to
encounter at least one hundred wave crests is usually regarded as necessary to obtain consistent estimates
of rms motions in irregular waves. Even longer runs will be necessary to obtain consistent results for rarely
occurring events like deck wetness an9 slamming. The required run ~me can be achieved by scarfing short
runs together. The 'mean' rms niotions can be calculated by averaging the motion variances obtained in
each short run.
Model testing
324
[Ch. 16
If the model speed is very low these reflected waves will return to the centre of the
tank before the model has moved away, as shown in Fig. 16.20(a). In this case the
model's motions will be influenced by these reflected waves and misleading results
will be obtained. If the model speed is high enough the reflections will reach the
centre of the tank after the model has passsed by and no interference will occur.
The critical velocity at which tank wall interference occurs may be calculated as
follows.
Suppose that the model is running in head or following waves down the centre of a
long narrow tank. The encounter frequency is then given by
ro 2
ro 
u cos Jt
g
radians/second
(16.52)
metres/second
(16.53)
(16.54)
to travel from the model to the tank wall and back to the tank centreline. Tank wall
interference will occur if the model moves less than its own length in this time. The
critical model speed at which interference begins is
Lm metres/second
(16.55)
Ucrit
00
metres/second (16.56)
Sec. 16.11]
325
(following waves)
(16.57b)
Figure 16.21(a) illustrates the relationship giv~n by equation (16.57a) for head
waves. In this case the Froude number must be greater than FN crit to avoid tank wall
interference: the critical Froude number increases with the length of the model and is
very large for low frequencies (long waves).
Fig. 16.21(b) shows the critical Froude number for model tests in following waves
obtained from equation (16.57b). In this case the Froude number must lie within a
finite range to avoid interference and the range decreases as the model length
increases. When the model length is onequarter of the width of the tank there is only
one Froude number for each w.ave frequency that will give results which do not suffer
from interference. Models of greater length will_,_always experience tank wall
326
[Ch. 16
Model testing
0.3
FN crit
10
,,,V(Lmlg)
i.!Lm
0.5
0.4
Tank wall
interference
0.3
FN crit
interference whatever the Froude number or wave frequency. It follows that the
maximum permissible model length for tests in following waves is onequarter of the
width of the tank.
17
Probability formulae
17.1
INTRODUCTION
In Chapters 4 and 14 it was shown that the irregular time histories of both waves and
ship motions could be characterised in terms of energy spectra and various statistical
quantities like mean values, periods, rms values and so on. Seakeeping studies,
however, often demand a more intimate knowledge of the characteristics of waves
and motions. In particular the likelihood of a particular event occurring (such as a
particular motion level being exceeded) is often of interest. Wave and ship motion
information. The discussion
time histories can be analysed to provide this sort
which follows is written in terms of wave analysis but applies equally to all ship
motions.
of
17.2
PROBABILITY ANALYSIS
Two methods of analysis of irregular time histories are comm0nly used. In the first
the time history is analysed by reading discrete values of the record at fixed intervals
of time (say every second) as shown in Fig. 4.3. This method can be used to find the
probability or the proportion of time that the wave depression exceeds a particular
level.
Some of the measurements obtained in this way may by chance be peaks, troughs
or zero crossings but they are not given any special significance. An alternative
method of analysis is concerned only with these ,salient points in the record and is
commonly used in many aspects of seakeeping work. Typically the analysis consists
of measuring successive wave amplitudes and periods as defined in Fig. 4.2. This
technique is used to extract information on the probability of peaks and troughs
exceeding a given level.
17.3 HISTOGRAMS
Whichever kind of analysis is. used the results will consist of an apparently random
sequence of measurements: these can be sorted _according to their values into
[Ch. 17
Probability formulae
328
discrete ranges or histogram 'bins'. Tables 17.1 and 17.2 show typical examples for a
30 minute wave time history using 0.5 metre bins. Table 17.1 gives results based on an
analysis of the record at discrete time intervals of 1.0 second: Table 17.2 gives the
corresponding analysis based on wave amplitudes measured from the mean surface
depression. No distinction is made between peaks and troughs.
In Table 17.1 it is shown, for example, that 37 measurements of the wave
depression were found in the range 4.55.0 metres below the arbitrary datum level
chosen for the analysis. In Table 17 .2, 23 measurements of wave amplitude (peaks
and troughs) were found in the range 3.54.0 metres. The corresponding histograms
are shown in Figs 17.1 and 17.2.
Histogram
bins
Number of
observations
in each bin
Probability
of
occurrence
(metres)
Nh
Probability
density
function
ordinate
f (metres 1 )
0
1
3
11
23
36
0.000
0.001
0.002
0.006
0.013
0.020
0.039
0.068
0.100
0.117
0.133
0.139
0.117
0.096
0.067
0.042
0.021
0.011
0.005
0.003
0.002
0.001
0.000
0.000
0.001
0.003
0.012
0.026
0.040
0.079
0.136
0.200
0.233
0.267
0.278
0.233
0.191
0.133
0.083
0.041
0.022
0.010
0.006
0.004
0.001
0.000
3.5 to
3.0to
2.5 to
2.0 to
1.5 to
1.0 to
0.5 to
O.Oto
0.5 to
1.0 to
1.5 to
2.0 to
2.5 to
3.0 to
3.5 to
4.0 to
4.5 to
5.0 to
5.5 to
6.0to
6.5 to
7.0to
7.5 to
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
8.0
71
122
180
210
240
250
210
172
120
75
37
20
9
5
4
1
0
329
Histograms
Sec. 17.3]
(metres)
Number of
measurements
of wave
amplitude
Nh
Probability
of
occurrence
169
171
234
265
204
145
101
23
12
0
Total: 1324
0.128
0.129
0.177
0.200
0.154
0.110
0.076
0.017
0.009
0.000
0.0 to 0.5
0.5 to 1.0
1.0 to 1.5
1.5 to 2.0
2.0 to 2.5
2.5 to 3.0
3.0 to 3.5
3.5 to 4.0
4.0 to 4.5
4.5 to 5.0
Mean wave amplitude ~a
TH
Probability
density
function
(metres l)
0.255
0.258
0.353
0.400
0.308
0.219
0.153
0.035
0.018
0.000
= 1. 7 metres
1800 seconds.
300
Mean surface depresion
Arbitary
datum
;i
250
r
200
Q)
r
c
<J)
Q)
E 150
::J
<J)
Q)
Q)
_Q
r
50
1
r
4
co
....E 100
,w,
1
1
r
E
::J
z
r
~ rJ
r
_Q
.c
(J
co
r
rrf
2
1~
h
0
Fig. 17.1 Typical histogram of measurements at regut'ar time intervals (Table 17.1).
330
Probability formulae
300
Mean amplitude
).._
~
..c
()
co
Significant single
amplitude
ii
Q)
[Ch. 17
200
1
c"'
Q)
::J
"'co
Q)
0 100
1
!
Q)
..0
E
::J
z
G:,
The total number of measurements (whether they be amplitudes or measurements obtained at fixed time intervals) is obtained by summing the observations in all
the histogram bins:
(17.1)
s to s is
1
(17.2)
s is
1
Sec. 17.4]
331
(17.3)
1;,
IS =
=
(17.4)
1
rl
The histogram has a major disadvantage: the ordinates (the numbers of observations in each bin) depend on the record length and the width,ofthe histogram bins
and this complicates comparisons between different results. The probability density
function (PDF) is a form of histogram which eliminates this dependency on record
length and bin width. The PDF is defined such that the area enclosed by the PDF
curve over a bin is equivalent to the probabilty of the measurement falling within that
bin.
Iff is the probability density function ordinate and wh is the width of the bin
p = Nh
N
(17.5)
(17.6)
Probability formulae
332
[Ch. 17
The probabilities defined in equations (17.2) and (17.3) can now be written
(17.7)
(17.8)
The total area under the PDF is equivalent to the probability that an individual
measurement will lie within the range of all the measurements:
wh
Lf = 1.0
(17.9)
(17.10)
(17.11)
(17.12)
and the total area under the continuous PDF curve is unity:
(17.13)
Area = Probability
of depression
lying in range
3.03.5 metres
~"'
Q)
E
~
333
Sec. 17.5]
0.2
"'B
c
.....::J
u;
c
Q)
"'0
.
:c
0.1
co
.0
0
a:
o~~~~~~~~._~
4
2
0
Surface
2
depression~
(metres)
measu~ements
(17.14)
where m 0 and ~ are the variance and mean value of the time history as defined in
equations (4.1) and (4.2). This convenient result,means that the PDF for a regularly
sampled wave record can be estimated if the mean and variance are known.
As an example, Fig. 17.3 shows the Gaussian PDF obtained for the wave time
history analysed in Table 17 .1.
In practice it is usual to arrange fOJ the mean value ~to be zero by making the
arbitrary datum for the measurements the same as the mean value and equation
(17.14) becomes

(17.15)
[Ch. 17
Probability formulae
334
Corresponding
Rayleigh PDF
c
0
.;::;
u
Area = probability
of amplitude
lying in range
3.03.5 metres
.2
.
(f)
Q)
"0
.
.a
co
.a
0
a:
2
Wave amplitude
sa (metres)
The probability that an individual measurement will lie within the range /,; 1 to /,;2 is
now given by
(17.16)
where the error function is defined as
erf(x)
and is tabulated in Table 17.3. The error function has the properties
(17.17)
Sec. 17.5]
335
erf(x)
x erf(x)
x erf(x)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
0.000
0.020
0.040
0.060
0.079
0.099
0.118
0.137
0.155
0.174
0.192
0.209
0.226
0.242
0.258
0.273
0.288
0.302
0.316
0.329
erf( x)
erf( oo)
= y'(1211')
1.00
1.05
1.10
1.15
1.20
1.25
1.30
1.35
1.40
1.45
1.50
1.55
1.60
1.65
1.70
1.75
1.80
1.85
1.90
1.95
0.341
0.353
0.364
0.375
0.385
0.394
0.403
0.412
0.419
0.427
0.433
0.439
0.445
0.451
0.455
0.460
0.464
0.468
0.471
0.474
Jx exp ( 2 2) dz
2
x erf(x)
2.00
2.05
2.10
2.15
2.20
2.25
2.30
2.35
2.40
2.45
2.50
2.55
2.60
2.65
2.70
2.75
2.80
2.85
2.90
2.95
0.477
0.480
0.482
0.484
0.486
0.488
0.489
0.491
0.492
0.493
0.494
0.495
0.495
0.496
0.497
0.497
0.497
0.498
0.498
0.498
erf(x)
erf( oo)
3.00
3.05
3.10
3.15
3.20
3.25
3.30
3.35
3.40
3.45
3.50
3.55
3.60
3.65
3.70
3.75
3.80
3.85
3.90
3.95
0.499
0.499
0.499
0.499
0.499
0.499
0.500
0.500
0.500
0.500
0.500
0.500
0.500
0.500
0.500
0.500
0.500
0.500
0.500
0.500
(17.18)
0.5
x erf(x)
(17.19)
Note that equation (17 .20) refers to positive values and relates only to one side of the
Gaussian probability density function. The probability of an individual measurement
lying outside the range 1 is given by
Probability formulae
336
12
[Ch. 17
erf(~J
(17.21)
The probability that an individual measurement will lie within the total range of all
measurements is
P( oo
(17.22)
These properties are illustrated in Fig. 17.5. Table 17.4 lists some of the salient
erf
~1
no
Probability of
0
'2
JJ
0.5
Probability of
between
~1 and ~2
~lying
.....
~
'Ui
QJ
"'0
Probability of
.0
~exceeding
C1l
~2
.0
[l_
(1?)
properties ofthe Gaussian PDF based on equations (17.20) and (17.21). From these
results it can be seen that about 95% of all measurements obtained from a regularly
sampled wave record will lie within twice the rms value.
Sec. 17.6]
Table 17.4 
337
t;lcr0
Probability of
exceeding t;
Probability of
exceeding t;
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.500
0.308
0.159
0.067
0.023
0.006
0.001
1.000
0.616
0.318
0.134
0.046
0.012
0.002
The mean value of all the upward or downward observations in the record can be
obtained from
{"" YI'AY
1~1
00
= 2j (jdt; = 0.798Vm0
0
oo
(17.23)
J fdt;
0
_k_ exp ( m0
t;~)
2m0
(17.24)
where m0 is the variance of the time history defined in equation (4.2). This
convenient result allows the amplitude PDF to be estimated if the variance of the
wave depression is known. The Rayleigh PDF for the wave time history analysed in
Table 17.2 is shown in Fig. 17 .4.
The probability that an individual measurement of amplitude t;a will lie within the
range t;a 1 to t;a2 is given by

Probability formulae
338
[Ch. 17
1;;~2)
1;;~1)
exp  exp 2m0
2m0
(17.25)
and the probability that the amplitude will exceed a given levell;;a 1 is
(17.26)
This function is plotted in Fig. 17.6. Again, the probability that an individual
amplitude will lie within the total range of all the measurements is unity:
P( 
oo
< Sa <
oo )
_..!__
mo
exp (
oo
oo
I;;~) dl;;a
2mo
1. 0
(17.27)
Equation (17 .26) can be rewritten to give the amplitude which has a given
probability of being exceeded:
or
(17.28)
Table 17.5 gives some sample results from equations (17.26) and (17.28). From
these results it can be seen that the probability of an individual amplitude exceeding
Sec. 17.7]
339
0.
E
ro
Cll
"0
Q)
Q)
(.)
X
Q)
.
:0
ro
.0
Cl...
rl
about three times the rms value is very small: only about one peak (or trough) in
every hundred would be expected to exceed this level.
17.7
It was shown in Chapter 5 that the significant wave height, defined as the mean of the
highest third of the heights recorded in a wave tim~ history, was closely related to the
average wave height estimated visually by an experienced observer. In the same way
it might be expected that the experienced sailor's estimates of 'average' ship motions
might be similar to their significant amplitudes. Interest is therefore often centred on
these quantities.
In more general terms an expressionis required for the mean value of the highest
(1/n)th of all observations of amplitudes (where n = 3 for significant values). If the
probability density function is known the required amplitude ~ltn is given by the
moment of area of the shaded portion shown in Fig, 17.7. The shaded area is of
course equal to lin.
340
[Ch. 17
Probability formulae
Table 17.5 
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Probability of
exceedance
p
1.000
0.882
0.606
0.325
0.135
0.044
0.011
0.002
0.0003
1.000
0.500
0.333
0.100
0.010
0.001
0.0001
0.00
1.18
1.48
2.15
3.03
3.72
4.29
Sln
.....
c
.E
c
.....::J
.
(/)
Q)
""0
.
:0
Area=
probability of
amplitude
exceeding ~Y"
co
..0
e
a...
Amplitude::,
If the PDF is given by the Rayleigh formula (equation (17 .23)) the amplitude 1; 11n
(which is exceeded with a probability 1/n or once in n amplitudes) is given by
equation (17.28):
Sec. 17. 7]
341
(17.29)
_
r
J"" safdsa
::.._...::;ll.;..:.n_ _
':>lin 
J fdsa
" safdsa
J
P(sa >
co
lin
lin
Slln)
nf"" I,;Jdsa
(17.30)
lin
If the PDF is given by the Rayleigh formula (equation 17.24) equation (17.30)
becomes
~lin
co
1,;2
( _ 1,;2)
_!!_exp  a dl,;a
2mo
lin mo
rl
so
(17.31)
Selected results are given in Table 17 .6. Of particular interest are the results for
1 andn = 3:
Putting n = 1 gives the mean value of all amplitudes
~a
Putting n
~113
1.25cro
(17.32)
2.00cr0
(17.33)
Probability formulae
342
[Ch. 17
n
1
2
1.25
1.77
2.00
10
100
2.54
3.34
3.72
4.29
1000
10000
2~113
= 4.00cr0
(17.34)
These results are widely assumed to apply to all ship motions and wave records. It
should be remembered however that they are strictly only true if the Rayleigh
formula (equation (17.24)) holds.
17.8 JOINT PROBABILITIES
exp (
2m0 ;
2m 0i
x1)
= exp (  xA  ..:::1.}_
2m0 ; 2m 0i
where m 0 ; and moi are the variances of the respective motions.
(17.35)
18
Roll stabilisation
18.1
MOTION REDUCTION
Roll stabilisation
344
[Ch. 18
Bilge keels are the simplest form of roll stabilisation device. These are long narrow
keels mounted at the turn of the bilge as shown in Fig. 18.1. If active roll stabiliserfins
~c::::~
_____
c ______
bf~BK_._P~J~
~I
Sec. 18.2]
Bilge keels
345
are also required the bilge keels may be segmented to accommodate them as shown
in Fig. 18.2.
Fig. 18.2 Bilge keels and roll stabiliser fins on a destroyer. (MoD Photo.)
Bilge keels are very effective roll stabilisation devices which work well at all
speeds. They have the significant advantage that they have no moving parts and
require no maintenance beyond that normally given to the hull surface. Their only
disadvantage is that they increase the resistance of the ship, but the effects can be
minimised by carefully aligning the keels with the :flow streamlines around the bilges.
This is usually done using some kind of flow visualisation technique on a model
during the design stage. Correct alignment can only be achieved at one speed (the
cruising speed is usually chosen) but the resistance penalty at other speeds is usually
small.
Bilge keels work by generating drag forces which oppose the rolling motion of the
ship. The mechanism is similar to that shown for appendages at zero speed in Fig.
12.5. The roll damping moment for a single bilge keel is given by equation (12.28)
and the equivalent linearised toll damping coefficient is given by equation (12.31). It
remains to determine suitable values of the drag coefficient for bilge keels, C0 . Cox
and Lloyd (1977) cited experimental data published by Martin (1958) and by
[Ch. 18
Roll stabilisation
346
Ridjanovic (1962).This is presented (in slightly different form) in Fig. 18.3. The
normal force coefficient is given as a function of the equivalent aspect ratio
(18.1)
and the nondimensional bilge keel radius parameter
(18.2)
The results may be expressed in the form
C0
= 0.849
(f,
J)
(18.3)
0.025
0.05
...c
0.10
Q)
'()
0.15
Q)
(.)
0.20
Ol
0.30
0.40
0.14
0
Equivalent aspect ration a 8 K
Sec. 18.2]
Bilge keels
347
where
/=14.66J
K
1(_2_)
'V
(18.4)
aBK
K=
(18.5)
y'(2aBK)
(14.66 J)(0.109 0.208aBK)
(18.6)
Fig. 18.3 shows the benefits of increasing the bilge keel aspect ratio for a given bilge
keel area. In other words a short wide bilge keel is much more effective than a long
narrow bilge keel.
Figs 18.4 and 18.5 show examples of the calculated effects of bilge keels on roll
motion for a frigate in a moderately severe seaway. The heaviest rolling occurs in
quartering seas and a pair of 30 metre 2 bilge keels offer a substantial reduction in
motion. Fig. 18.4 demonstrates the superior performance of bilge keels of high
aspect ratio and Fig. 18.5 shows that the bilge keels are most effective at low speeds.
e
en
E
FT, 3 =5.5 m,
T0 = 12.4 sec
'
Heading (degrees)
Fig. 18.4 Effect of bilge keel aspect ratio on roll motion of a frigate at 20 knots.
348
Roll stabilisation
[Ch. 18
Cl
Q)
::9.
e
"'
E
....
20
25
30
Speed (knots)
Fig. 18.5 Effect of bilge keels on rolling at worst heading.
Fig. 18.6. Low aspect ratio nonretractable fins fitted to a warship. (MoD Photo.)
Sec. 18.3]
349
Fig. 18.7 
High aspect ratio retractable fin fitted tQ,.a merchant ship. (Reproduced by
permission of Sperry Marine Inc.)
[Ch. 18
Roll stabilisation
350
[J
Sec. 18.3]
351
4.04
(L)aF=l 1.8 + y(4 +a~)
L
(18.7)
Equation (18.7) is plotted in Fig. 18.9 and it can be seen that the gains associated
with increasing the chord if the outreach is limited are minimal if the aspect ratio is
reduced below about 1.0. Thus practical nonretractible fins are limited to an aspect
ratio of about 1.0. More lift can therefore be achieved only by installing more pairs of
fins.
Limited
outreach
o __o__
Increasing
area
3.0
Aspect ration aF
(18.8)
exerting a roll moment
2L
rF
kN metres
352
[Ch. 18
Roll stabilisation
View looking
forward
.Lsmf1
L cos fl
L
2L Sin
fl
Fig. 18.10 Sway force and yaw moment caused by stabiliser fins.
The two lift forces have vertical components which cancel so that there is no resultant
vertical force on the ship. However, their horizontal components add and yield a
sway force
2 L sin
kN
to starboard
If the fins are mounted xBlF metres forward of the centre of gravity, this horizontal
force will exert a yaw moment
2L
xBlF
sin
kN metres
to starboard
Sec. 18.3]
353
The effects of the stabiliser fins on the motions of the ship in waves may be
computed by including these additional terms in the equations of motion
(8.24)(8.29). Only the lateral plane equations are affected and these become
kN
(18.9)
(18.10)
yaw:
+ ~ EF L
xBlF sin~=
Fw 6 o sin (wet+ y6 )
kN metres
(18.11)
In these equations the summations refer to the number of fins fitted to the ship and
is an effectiveness factor defined as
EF
(18.12)
EF is generally less than 1.0 because of various hydrodynamic effects which are
discussed in section 18.3.4.
18.3.4.1 General
Lloyd (1975, 1977) investigated the effectiveness of roll stabiliser fins by measuring
the lift developed by model stabiliser fins and bilge keels in a variety of configurations
on a ground board. His experiments were conducted in the Circulating Water
Channel at the Admiralty Research Laboratory at Haslar in the United Kingdom.
His apparatus is sketched in Fig. 18.11. He identified three major causes ofloss in fin
performance:
(a) hull boundary layer
(b) finfin interference
(c) finbilge keel interference.
Roll stabilisation
354
[Ch. 18
0 0.
1.0
Sec. 18.3]
355
E
BL
=1 0_ 0
21
bF
(18.13)
The boundary layer thickness on the hull may be estimated using the equation
8 = 0.377
Xpp (RN)
o.z
(18.14)
_ p U Xpp
R N
(18.15)
J.Lw
18.3.4.3
Finlin interference
Roll stabiliser fins, like all lifting surfaces, work by developing a pressure difference
between their upper and lower surfaces. The water is tempted to roll round the tip of
the fin from the highpressure to the lowpressure surface and a vortex is formed. Fig.
18.13 shows the vortex generated by a fin at a fixed (nose up) angle of incidence ex.
This vortex is shed from close to the tip of the fin and trails away along the side of the
hull imparting a swirling motion to the water close to the hull. This causes a
'downwash' in the region between the vortex and the hull surface and an 'upwash' in
the region outboard of the vortex. Clearly the sense of the ~lfirling motion and the
Lift
356
Roll stabilisation
[Ch. 18
resultant flow directions depend on the direction of the lift developed by the fin. In
Fig. 18.13 an upward lift on a port side fin produces a clockwise motion (looking
forward). A nosedown fin deflection would produce an anticlockwise vortex
rotation. These directions are reversed for fins on the starboard side of the ship.
An oscillating fin produces a vortex of continually varying strength and direction
which is convected away along the side of the ship. In fact the vortex is a record or
'memory' of the lift developed by the fin. Fig. 18.14 illustrates the flow behind an
oscillating fin and it can be seen that there are alternate regions of downwash and
upwash in the wake ofthe fin, depending on the lift developed in the immediate past.
A second stabiliser fin mounted immediately behind the first fin will experience a
downwash over most of its outreach. This will generally decrease its angle of
incidence and reduce the lift developed. If the fin is mounted further aft in a region of
upwash, the lift developed will be increased.
Fig. 18.15 shows the finfin interference factor E 1F measured by Lloyd (1977) for
a pair of oscillating fins. The results are plotted as a function of a nondimensional
frequency parameter and the longitudinal separation of the fins. At zero frequency
(a) Unfavourable
Increased
lift
(b) Favourable
Sec. 18.3]
357
(I)~F =0.20
0.16
0.12
0.08
0.04
0
30
the interference is quite dramatic even for well spaced fins. For example, a fin spaced
20 outreaches behind the first fin will develop only about 50% of the nominal lift. The
interference effects become less important as the frequency and the separation are
increased,' until at very high frequencies and separations the second fin is in a region
of upwash and the interference becomes beneficial.
18.3.4.4 Finbilge keel interference
A bilge keel mounted abaft a stabiliser fin will also experience downwash and will
develop a lift which opposes the fin lift. Lloyd (1975) measured this opposing lift for
the case of zero frequency and his results are presented in the form of an effectiveness
factor
E IBK
in Fig. 18.16.
fin lift
(18.16)
[Ch. 18
Roll stabilisation
358
20
The detrimental effects of an aft mounted bilge keel are mitigated if the bilge keel
is followed by a second stabiliser fin. The bilge keel has a straightening effect on the
flow and removes some of the down wash due to the trailing vortex from the upstream
fin. This enhances the lift generated by the second fin as shown in Fig. 18.17. This
effect can be approximated by setting
EIBK
= 1.0
(18.17)
Sec. 18.3]
20
30
359
40
50
Fig. 18.17 Finfin interference factors. Effect of bilge keel at zero frequency.
1
rl
Such a configuration will only suffer from relatively insignificant hull boundary layer
losses. It is probably not worthwhile taking the trouble to locate the fin as far forward
as possibl~ to minimise these losses as this will make it difficult to accomodate a bilge
keel of adequate size forward of the fin.
If sufficient stabilisation cannot be obtained from a single pair of fins (bearing in
mind the limitations which may be imposed on the fin outreach) it will be necessary to
adopt a multiple fin configuration. In this case, the separation should ideally be
chosen to take advantage ofthe favourable interference effects shown in Fig. 18.15.
The fin configuration should be optimised to achieve the best performance at the
natural roll frequency (where most of the roll motion occurs) and at the ship's
cruising speed. In principle we require ..the second fin to be at a distance xFF metres
abaft the first fin such that the tim~ required for the vortex to convect from the
upstream fin is equal to half a roll period:
Xpp
T.4
1T
=
seconds
360
Roll stabilisation
[Ch. 18
so that
7TU
xFF= metres
(J).4
(18.19)
Unfortunately this gives fin separations which are often impractical. For example,
taking a cruising speed of 10.3 metres/second (20 knots) and a natural roll frequency
of 0.5 radians/second (natural period 12.6 seconds) we find an optimum separation
xFF = 63 metres
which may be difficult to achieve on all but the largest ships.
Bilge keels should not be located abaft the after fin.
In order to obtain the maximum possible roll moment the stabiliser fins should be
mounted at the turn of the bilge so that the roll lever arm rp is maximised (see Fig.
18.8). It is also advantageous to mount the fin stock normal to the hull surface. This
simplifies the mechanical arrangements and minimises the gap between the hull
surface and the fin root when the fin is at an angle of incidence. Such gaps are a
potential source of leakage between the high and lowpressure sides of the fin and
will result in a considerable loss of effectiveness. Fixed fins should also be aligned to
maximise the fin outreach within the enclosing rectangle of the ship.
These requirements often result in large values of the fin depression angle ~ (45
degrees or more). Reference to the modified equations of motion (18.9)(18.11)
shows that this will result in sway and yaw motions whenever the fins move in their
efforts to control the roll motion.
Suppose that the fins are set to some fixed angle of incidence to give steady lift
forces L kN to generate a stabilising roll moment to port as shown in Fig. 18.18. We
have already seen that this will result in a sway force and a yaw moment to starboard.
The ship will respond in exactly the same way as it responds to motions of the rudders
and will begin to turn to starboard. A centrifugal force to port then acts through the
centre of gravity, opposed by inboard hydrodynamic forces acting to starboard below
the waterline. These forces form a couple which tends to roll the ship to port,
enhancing the port roll moment directly generated by the fins. The total roll moment
generated by forward mounted fins is therefore increased by this swayyaw effect.
If the fins are mounted abaft the centre of gravity x 81p is negative. The ship then
turns in the opposite direction in response to the fins (to port for the example given
above). The roll moment caused by the turning motion of the ship then opposes the
roll moment directly generated by the fins and their effectiveness is reduced. In
extreme cases with near vertical fins mounted well aft the roll moment due to the
turning motion may actually exceed the roll moment directly generated by the fins
and the total moment will then be in the 'wrong' direction.
These effects are reduced when the fins are oscillating at higher frequencies and
they are not usually very significant at the natural roll frequency where most of the
rolling motion occurs. However, we shall see in section 18.3.6 that extreme aft fin
Sec. 18.3]
361
Outboard .,.,..,__ _
...,
1      "
centrifugal
force
Inboard
hydrodynamic
force
(a)
Couple enhances
roll moment
generated
by fins
Fins mounted
forward
IJo
View looking forward
Outboard
centrifugal _ _,..      force
Couple opposes
roll moment
generated by
fins
aft
Inboard
hydrodynamic 01111114t \         '  force
View looking forward
locations with large angles of depression may result in motion amplification at very
low frequencies. For this reason these locations should be avoided if possible.
Beneficial sway and yaw effects can be maximised by mounting the fins well
forward with a large angle of depression. However, this makes it difficult to
accomodate a bilge keel forward of the fins and is rarely attempted.
In practice fins are usually mounted somewhere near the middle of the ship and
the swayyaw effects may thep degrade their performance at low frequencies. The
degradation can be minimised by keeping the depression angle ~ as small as possible
but this will also tend to reduce the roll lever arm r.p':
362
Roll stabilisation
[Ch. 18
18.3.6.1 Introduction
Fig. 18.19 shows a block diagram representation of a ship stabilised with active fins.
Each component or block in the diagram may considered as a 'black box' having an
input and an output which are related by the block's transfer function. For example,
the ship block accepts an input in the form of a roll moment Fw4 from the waves and
generates a roll motion output x 4 Similarly the stabiliser fin controller generates a
demanded fin angle <Xn in response to the roll motion of the ship. The fin servo
mechanism responds and drives the fins to an achieved fin angle ex and the fins convert
this into a stabilising roll moment Fp4 This is subtracted from the roll moment
generated by the waves, thus reducing the roll motion of the ship.
Fw4
Ship
x4
Roll
F
FF4
Fins
Gyro
Measured
roll
E
a
Actual
fin
angle
Fin
servo
uo
D
Controller
Demanded
fin an g le
Fig. 18.19 Block diagram for a ship with roll stabiliser fins.
The transfer functions of the ship, the fin servo and the fin are essentially fixed for
a given design. The fin controller transfer function is, however, adjustable and must
be set up in such a way as to ensure that the fins develop roll moments which generally
Sec. 18.3]
363
where
CXo
X4m
Ko
Ku
K1
Kz
K3
b 1 , b2 and b3
Ko
Kl
Kz
K3
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
2.50
0
1
2
3
4
5
6
7
8
0
5
10
15
20
25
30
35
40
45
50
0
1
2
3
4
5
6
7
8
9
10
rl
9
10
bl = 1.00
b 2 =0.5
b3 =0.05
Ku = 0.0 for U < 5 knots
= 1.0 for 5 < U < 15 knots= 225/ U 2 for U > 15 knots
Roll stabilisation
364
[Ch. 18
overloading the fin stocks at very high speed. In modern systems Ku is varied
automatically with
cc
uz
as shown in Fig. 18.20. At very low speeds this would give very large gains, resulting
in excessive stabiliser fin activity and frequent demands for fin angles greater than the
maximum available. The fin servo mechanisms would be continually driving the fins
up against the mechanical stops, which are usually set to limit their travel to 25 or 30
degrees, leading to rapid wear and possible damage to the mechanical components.
So the speeddependent gain is usually limited to some finite value at speeds less
than, say, half the cruising speed. When the ship is hove to the fins are completely
Ku= Constant
,
I
I
I
c
co
O'l
....c
Kuu U2
Q)
"0
Manually
~switched
I
Q)
c.
Q)
"0
o
Q)
Q)
c.
(Jl
Ku=O
u,
Speed
ineffective and the speeddependent gain is then set to zero to avoid needless wear on
the system.
In less sophisticated systems the ship's crew are required to switch manually from
low to highspeed settings depending on the ship's speed.
Sec. 18.3]
365
and setting
Ec 
tan
_ 1 (
K2 ro.4
K _K
1
)
2
3 ro.4
tan
_ 1 (
b2 ro.4
b _b 2
1
3 ro.4
rad1ans
(18.22)
(18.23)
where the required phase angle is
(18.24)
Roll stabilisation
366
[Ch. 18
from which it can be seen that the required phase must be achieved by choosing
appropriate ratios between the sensitivities.
Practical application of this technique requires the phases E8 and Eps to be
determined. These may be found in a forced rolling trial at sea as shown in Fig. 18.21.
The ship is run in calm water so that the roll moment from the waves is zero or at least
negligible. If any waves are present their effects should be minimised by running in
head or following seas. Rudder motions will influence the rolling motions of the ship
so the helmsman should keep the wheel amidships and the autopilot, if the ship has
one, should be switched off.
Sec. 18.3]
367
The stabiliser controller is isolated from the system by breaking the circuit at the
point Din Fig. 18.19. The fin servos are instead driven by a sinusoidal demand signal
equivalent to, say, 15 degrees fin amplitude at some selected frequency. The ship
will then roll at the same frequency and the roll response is measured at the point C in
Fig. 18.19. The actual fin angle is monitored at point E. Es is the phase between the
signals monitored at C and E.
Fig. 18.22 shows the expected form of the results. These depend on the location
and depression of the fins. For fins mounted somewhere near the middle of the ship
with a moderate angle of depression the phase at zero frequency is zero, showing that
the ship rolls in the expected sense: a fin incidence giving a steady roll moment to port
results in a steady port heel.
0.8
.01 0
~
)(
Q)
"'0
~
0.
E 0.4
"'c
!;
=e1
0
a:
"'I
~I
z
0
I
1.0
1.5
: il
200
(j)
7
7
Q)
~
O'l
Q)
2.
Q)
(/)
"'0.
..c
100
Frequency (radians/second)

Roll stabilisation
368
[Ch. 18
If the fins are mounted well aft with a large depression angle the phase at zero
frequency becomes 180 for the reasons already explained in section 18.3.5. The roll
response to the fins can be quite large but is then in the opposite sense to that
expected.
In either case the ship phase Es can be measured at the natural roll frequency as
shown.
The fin servo phase Eps is measured between the signals monitored at points D
and E in Fig. 18.19. Typical results are given in Fig. 18.23.
1.5
1.0
"
t$
I=>
0 {)
0.5
I Cii~
I :::J~
r
....
Ql
:::J
0"
Ql
"' ~
JZ"""
II
0
0.5
1.0
1.5
0.5
1.0
1.5
0
Ui
Ql
~
Ol
Ql
~100
Ql
(/)
"'
..c
a..
200
Frequency (radians/second)
<X= <Xo
sin (ole()
radians
(18.26)
in equation (18.8). The equations will then give sinusoidal motion responses(in sway
and yaw as well as roll) and the phase E5 may be determined.
Sec. 18.3]
369
Worked example
Find appropriate controller coefficients to match the responses given for forward
mounted fins in Figs 18.22 and 18.23 using the controller coefficients listed in Table
18.1.
Natural roll frequency:
~>s
=  70,
~>Fs
=  10
~: 70 + 10 +tan
l(
0.5X0.45
) 0
l.O _ O.OS x 0.4S 2  93
(Kl
Kz =  42.4
K3
K3
0.2)
K1
Kz
K3
0
0
0
0
0
0
1
1
1
1
1
10
15
25
35
40
50
40
30
25
15
10
1
2
3
4
5
6
10
9
8
7
6
All of these settings will give a satisfactory perfq;rmance since they all ensure that
the moment applied by the stabilisers exactly opposes the wave moment at the
370
Roll stabilisation
[Ch. 18
natural roll frequency. In practice it may be found that some of the settings are
marginally preferable to others but the benefits to be gained are usually small. For
example, setting K 1 to zero will ensure that the fins do not waste energy by
attempting to correct a steady list or try to hold the ship upright in a turn. The steady
roll moment available at zero frequency is so small (see Fig. 18.22) that efforts in this
area are probably doomed to failure. The only appreciable effect will be an increase
in the ship's fuel consumption due to the small increase in resistance. Similarly, high
values of K 3 may lead to excessive stabiliser fin activity at high frequencies and
increased wear, noise and vibration with no noticeable reduction in roll motion.
ex
al
cx0
a1 + a2 s + a3 s2
(18.27)
Putting
al
CXno
(18.28)
a2 ro
e2
al a3 roe
(18.29)
The coefficients a1 , a2 and a3 are chosen to match these equations to the measured fin
servo responses.
Sec. 18.3]
371
oscillation in ten. Sea state 7 is to be interpreted using the WMO sea state code and
the most probable modal wave period for annual conditions in the North Atlantic.
Cosine squared wave spreading is to be assumed.'
The stabilised roll motion is computed using equations (18. 9)(18.11) with the fin
incidence a: now given by equations (18.20) and (18.27). Typical results of such a
calculation, taking account of the speed dependent gain, are shown in Fig. 18.24. As
Overall gain KG
expected the roll motion decreases with increasing overall gain at the expense of
increased fin motion. Using the gains available in the control system specified in
Table 18.1 we find that the roll target is achieved with
K0
=1.25
Roll stabilisation
372
[Ch. 18
demands will also lead to cavitation which, in extreme cases, may damage the fins
and will certainly generate noise. The latter may be of particular importance in
warships. It should also be noted that the assumptions of linearity inherent in
equation (18.8) describing the fin lift characteristics will lead to an overestimate of
the stabiliser performance if the fin motion (and hence the lift) is actually limited by
mechanical constraints. In any case the equation will overestimate the lift at large
angles of incidence (see Fig. 2.21).
The specification given above requires the probability of the fin motion amplitude
exceeding CXmax ( = 25) to be no more than 0.1. Equation (17 .28) may be used
to calculate the corresponding maximum allowable rms fin motion. Using Table 17.5
we find that
CXmax
O'o
= 2.15
The rms fin motion required to meet the roll target in Fig. 18.24 is 12.4. In this
case the stabiliser capacity will need to be enhanced by increasing the fin area, using a
more effective aspect ratio, improving the fin/bilge keel layout to avoid interference
or increasing the number of fins.
(18.30)
where IGI is the amplitude response or gain and EGis the phase response.
Suppose that the system is excited with an input vin and responds with an output
Vout Then the output of the feedback block will be Hv out and the total input to the
ship will be vin  Hvout. So the input and output of the complete closed loop system
are related by
(18.31)
Sec. 18.3]
373
G
Vaut
H Vout
...
...
G
G
V in
and the transfer function or 'closed loop gain' of the complt:1tff system is given by
Vout
Vin
G
1+G H
=
(18.32)
GHis the transfer function the system would have if the feedback loop were left open
and is termed the open loop gain. The stability of the closed loop system may be
examined using the Nyquist diagram illustrated in Fig. 18.26. In this diagram the
open loop gain is plotted as a vector of length IGHI and argument BaH where
The location of the end of the vector varies with frequency, moving around the
diagram as the frequency increases. At :lero frequency, in conventional systems with
fins in the middle part of the ship, the phase is zero and the open loop gain vector lies
along the positive real axis. As the frequency increases some phase advance is
introduced by the ship phase r~sponse (see Fig. 18.22) and by the controller. In a well
designed system the phase is, as we have already seen, arranged to be zero at the
natural roll frequency co. 4 and the gain vector again lies along the positive real axis.
374
Roll stabilisation
[Ch. 18
c:co
c
Ol
co
The gain should be a maximum at this point. At higher frequencies the phase
becomes negative and, in a stable system, the gain steadily diminishes until it
becomes zero at infinite frequency and the locus curve approaches the origin.
If the ship were unstabilised the response to the excitation vin would be G vin as
shown in Fig. 18.25(b). So the roll stabilisers will only reduce the rolling motion at
some particular frequency if
I1+GHI > 1
Now 11 + G HI is the distance from the point ( 1, 0) to the appropriate point on the
open loop gain locus (see Fig. 18.26). So we can see that the stabilisers will only
reduce the rolling motion at frequencies for which the open loop gain locus lies
outside a circle of unit radius centred at the point ( 1, 0). The roll motion will be
amplified for all frequencies lying within this unit circle. Fig. 18.27(a) shows that this
will always occur at high frequencies for conventional fin locations. For aft mounted
fins with large angles of depression the phase reversal at zero frequency results in the
locus of the open loop gain vector beginning somewhere on the negative real axis as
shown in Fig. 18.27(b). So these installations will always give motion amplification at
low frequencies unless the gain is made zero by setting
Sec. 18.3]
R
(a) Amplification at
high frequencies
375
c::.......___,;
.,
'!
(c) Unstable system
Equation (18.32) shows that the system will become unstable (i.e. the amplitude
response will become infinite) when the open loop gain
GH= 1
(18.33)
and this occurs if the gain yector locus passes thro,ugh the point ( 1, 0) on the
negative real axis as shown in Fig. 18.27(c). The phase is then  180 and the roll
moment due to the fins then enhances the roll moment due to the waves at some
Roll stabilisation
376
[Ch. 18
particular frequency. Any excitation at this frequency will then cause very large fin
motions which will enhance the initial excitation and increase the fin motions still
further. In practice, of course, the reponse will be limited by the mechanical stops,
fin rate limits on the fin servos and fin stall but large undesirable fin oscillations may
still occur.
Clearly we must ensure that the open loop gain locus never passes through the
point ( 1, 0). It is also desirable to avoid approaching it too closely because this will
result in motion amplification even though the motions will be stable. Two commonly used criteria for defining adequate safety margins are the gain and phase
margins defined in Fig. 18.28. The gain margin is defined as
G
m
~>oH
(18.34)
=  180
and the minimum acceptable value of G m is generally taken to be 2 which implies that
the open loop gain GH must not exceed 0.5 when the phase is 180.
The phase margin is defined as
~>m
11 + G
HI = 1
(18.35)
and the minimum acceptable phase margin is 30; 60is regarded as very good
practice.
If the system stability is unsatisfactory it can be improved by reducing the overall
gain K 0 or by choosing different values of the sensitivities K 1 , K 2 and K 3 (but still
satisfying equation (18.25)).
Passive tanks
Sec. 18.4]
377
Fig. 18.29 shows the performance of a typical active fin roll stabiliser system with
10
...,
CD
Cl
CD
::s
Cl
c:
'0
"'
..c:
CD
I
I
I
~
....
....
Cl)
....s:
\.
"'
Stabilised
'
Cl)
u,
0
U2
10
25
Speed (knots)
Fig. 18.29 Effect of speed on roll stabiliser fin performance for a frigate at worst heading.
speed dependent gain. At very low speeds the fins are comple~~ly ineffective because
the overall gain is set to zero (see Fig. 18.20). At speed U1 the fins are switched on
and a substantial roll reduction is achieved. As the speed is increased to U2 the fins
become progressively more effective and the stabilised roll motion decreases. At
speeds above U2 the gain is reduced and the stabilised roll motion becomes nearly
independent of speed.
.....:]
00
.........
...,..
I~ I
(a)
t=O
~I ==(b)
t=
rr
(1h4
"'
(c)t=~
2UJ.4
~"'
~
"
g
>'
00
Passive tanks
Sec. 18.4]
(c)
Simple Utube
tank
379
rl
Utube tanks have also been fitted in a number of ships. In this case the free
surface is confined to the two arms of the Utube which are connected by a horizontal
duct. The tops of the vertical arms may be open to the atmosphere or they may be
connected by a horizontal air duct. In this case a throttle valve may be included to
exert some control over the motions of the fluid. Some designs incorporate a throttle
valve or a pump in the bottom duct.
Passive tanks work well at low speeds but they are not usually as effective as a well
designed active fin system at high speed. For this reason they are often specified for
ships like survey vessels or weather ships which must spend the majority of their time
hove to.
Tanks have the advantage that they~have no moving parts (except perhaps for a
pump or controlled throttle valve) and require little maintenance. They also avoid
the small resistance penalty associated with fins and bilge keels. They take up a
considerable volume of the ship's hull but it may be possible to use the fresh water
supply or some of the fuel oiL.as the working fluid so this loss of volume may not be
serious. The optimum tank position high in the ship ~ften makes access along the ship
difficult.
Roll stabilisation
380
[Ch. 18
A major disadvantage is that the free surface always reduces the metacentric
height so that roll stability will be reduced. As a consequence all tanks amplify roll
motions at low encounter frequencies. In certain circumstances this amplification
may become a serious problem and it may be necessary to immobilise the tank by
draining it or filling it completely. This will invariably take a considerable time and
passive tanks are therefore not suitable for ships which are required to change course
frequently (e.g. warships).
18.4.2 Theory for a Utube passive tank
In spite of the apparent simplicity of the flume tank, no adequate theory for
predicting its performance has been developed. However, Stigter (1966) has developed a theory for Utube passive tanks and a modified version of this is described
below.
18.4.2.1
Fig. 18.32 shows a simple Utube passive tank. The tank is assumed to consist of two
Port
reservo ir r
h,
Datum
fluid
level
2
Star board
rese rvoir
;
T~;;___ 
h,
rd
'
vtl
I . w,
1d
Duct
Yd
ilo
I
 f.
...
1
Wr
Passive tanks
Sec. 18.4]
381
reservoir) is v metres/second. Three additional axes are defined: y ct has its origin at 0
and runs parallel to the duct, positive to port;yrp andyrs have their origins on the duct
centreline and run parallel to the reservoir walls as shown.
n is the width of the tank perpendicular to the y axis. Note that n is a variable
which has different values hct on the duct and wr on the two reservoirs. It is assumed
that there is no flow in the 'n' direction and the motions of a unit mass (1 tonne) of
fluid in the tank will be governed by a simplified version of Euler's equation (2. 7b):
(18.36)
where Y is the external force per unit mass and p 1 is the mass density of the fluid in the
tank.
Now the duct and the reservoirs are assumed to be of constant cross section so we
may write
av =0
ay
everywhere except at the junctions between the duct and the reservoirs. Neglecting
these corner effects equation (18.36) reduces to
av y  1ap

at
Pt
(18.37a)
ay
dv _ y 1 dP
 dt
(18.37b)
Pt dy
If the difference in the height of the fluid level in the two reservoirs is z metres the
velocity in each reservoir will be
d
vr = dt
(z)2 = 2wi
metres/second
where 't, which is assumed to be small, is the tank 'angle' defined in Fig. 18.32,
w = wd + wr
metrt:s
(18.38)
382
Roll stabilisation
[Ch. 18
(a)
(b)
Fig. 18.33 External forces applied to unit mass in (a) Utube duct and (b) Utube reservoirs.
Passive tanks
Sec. 18.4]
w. v.
n
w. w t
2n
v =   = 
383
metres/second
(18.39)
The external force per unit mass Y is made up of contributions due to the
accelerations applied to the tank and the frictional forces arising from the losses in
any throttle valve, wall friction etc. Figs 18.33(a) and 18.33(b) show these contributions. They are:
(a) The component of the acceleration due to gravity along the y direction
 g cos
<j> 1
metres/second 2
COS X4
= S2
metres/second 2
(18.40a)
=s2 sin x 4 = 0
since both s2 and x 4 are assumed to be small
YLA
(18.40b)
qv
kN/tonne
n
where q is a coefficient of resistance to be estimated or determined by
experiment.
Equation (18.37b) then becomes
Roll stabilisation
384
,!.,
..
[Ch. 18
= Pt dy
metres/second
(18.41)
We now integrate this equation with respect toy to obtain an equation giving the
motion of the fluid in the tank (in terms of the angle 't) as a function of the pressure
difference at the surface in the two reservoirs. Strictly the integration should proceed
from the surface level in the starboard reservoir (negative y) to the surface level in
the port reservoir (positive y). However, the continually varying fluid levels
introduce complications and we therefore obtain an approximate solution by
integrating between the datum levels in each reservoir. We also assume that the
lateral acceleration s2 does not vary appreciably along y. We obtain
Pt Wr W Il 't
2
Pt q Wr W I2 i
+
2
= Ps PP
+ Pt g
Pt I4 i4
Pt s2 Is
kN/metre 2
(18.42)
where
I1 
tank
2
dyd JO +
dYrs Jh'dy
W
dy Jw/ +
2..!1?_
n   w/2 hd
h, Wr
0 Wr  hd
+2hr
Wr
(18.43)
(18.44)
I3 = J
cos
cf>l
tank
I4 =
2
dy = x4 Jw/ dyd  Jo dyrs + Jh, dyrp = w x4 metres
 w/2
h,
0
(18.45)
w/2
tank
w/2
rd dyd +
o
J
2 dYrs +
h
Jh, w
'
metres
(18.46)
2
dy =  J
Is=  J
duct
where
tank
dyd =  Jw/
duct
w/2
dyd =  w
(18.47)
implies integration along the y axis from the datum level in the
Passive tanks
Sec. 18.4]
385
is confined to
duct
the duct. The angles <\> 1 and <\> 2 are defined for the duct and the reservoirs in Figs
18.33(a) and 18.33(b).
The hydrostatic pressures at the datum levels in the two reservoirs are
w
Ps= PP= p1 g2t
kN/metre 2
(18.48)
metres3
= 0
kN metres
(18.49)
=  Q1
kN metres/(radianlsecond2 )
a~ 4 = Q1 (rct +h.)
c~ 4
= Q1g
a~6 = 
kN metres/(radianlsecond 2 )
kN metres/radian
Q1x 81
kN metres/(radian!second 2 )
kN: ffi"etres/(radianlsecond 2 )
b~~ = Q
q w. (
;J +:~)
kN metres/(radian!second)
(18.50)
(18.51)
(18.52)
(18.53)
(18.54)
(18.55)
Roll stabilisation
386
[Ch. 18
(18.56)
with
Qt = Pt wr2w
Xt
tonne metres
(18.57)
The tank angle 't may be regarded as an additional degree of freedom in the equations
of motion for the ship (8.26)(8.31). Its effects are taken into account by including
additional terms of the form
(i = 1,6)
in these equations.
Many of these coefficients are zero. In particular, the tank has no effect on the
surge, heave or pitch motions so that
(i = 1; i = 3; i = 5)
Simple physical arguments also demonstrate that no sway force or yaw moments can
be caused by a steady tank angle and that the rate of change of tank angle can have
little appreciable effect. So
(i = 2; i = 6)
In addition it is assumed that the rate of change of tank angle has a negligible
influence on the roll moment so that
The lateral plane equations of motion for a ship stabilised with a passive tank are
then:
sway:
roll:
(roe
t+y 2) kN
(18.58)
+ C46 x 6 
[a4r
(roe
t+ Y4)
kN metres (18.59)
(where the expression inside the square brackets is the tank stabilising moment)
Sec. 18.4]
387
Passive tanks
yaw:
(18.60)
The vertical plane equations remain as for the unstabilised ship (equations (8.23),
(8.25) and (8.27)).
We now derive the tank acceleration coefficients a 2 ~, a 4~ and a 6~. These may be
considered as the sway force, roll and yaw moments required to sustain a tank angle
acceleration of 't = 1 radian/second 2 .
Consider the tank shown in Fig. 18.34. If the tank angle acceleration is 't
(.
in,
tv,
m,
.v,
rd
,._
,.,""
md
 l i d ....
.....
r" . . . _ . .
radians/second 2 the fluid accelerations in the reservoirs and the duct are, by equation
(18.39),
metres/second 2
and the masses of the fluid in the reservOirs and the duct are
The lateral force which must be applied to the tank to sustain these clockwise
(positive) accelerations is
388
Roll stabilisation
[Ch. 18
kN to starboard
so that
(18.61)
If the tank is located x81 metres forward of the centre of gravity the yaw moment
= x81
a 2 ~ 't
kN metres to starboard
so that
(18.62)
The roll moment required to maintain the acceleration of the fluid in the reservoirs is
kN metres to port
and the roll moment required to sustain the acceleration of the fluid in the duct is
and we obtain
(18.63)
Finally we obtain the coefficient c 4~ by considering the moment required to
sustain a steady positive tank angle t. The weight of fluid above the datum level in the
port reservoir in Fig. 18.35 is
kN
Sec. 18.4]
Passive tanks
w
389
w
Ptgw,wx,
~
G
w,
w,
and a similar weight is displaced from the starboard reservoir. So the applied
moment is
kN metres to port
and
e4~
18.4.3
= Q1 g =en= e~ 4 kN metres/radian
(18.64)
The motion of the fluid in the tank is governed by equation (18.49) which may be
rewritten in the form
(18.65)
This has the same form as the equation governing the behaviour of a secondorder
linear damped springmass system (equation (6.1)) with the right hand side providing the excitation to the tank from the ship. The natural frequency of the tank is, by
equations (6.8), (18.54) and (18.56),
ro 
*t
)
+ 2hrhd
~(anen)  ~(WrW2ghd
radians/second
(18.66)
[Ch. 18
Roll stabilisation
390
The nondimensional tank damping or decay coefficient is, from equation (6.12),
(18.67)
The tank decay coefficient may be determined with a simple free decay experiment
on a fixed model of the tank. The model should be of fairly large scale and is
conveniently made in acrylic sheet or some other transparent material so that the
oscillations of the fluid may easily be observed. The tank fluid should be displaced
towards one side of the tank and then released. The subsequent decay of the tank
angle oscillations should be recorded and the decay coefficient estimated from
equation (6.22). The dimensional tank damping coefficient is then given by
bn =
~tQt ~[ g (~ + ~J]
Wr
= 2 ~~
Q 1 g kN metres/(radian/second)
co.t
18.4.4
(18.68)
Roll
movement
due
to waves
Ship
Roll
Roll m oment j
due to tank
Tank
Fig. 18.36 Block diagram for a ship with a passive stabiliser tank.
Passive tanks
Sec. 18.4]
391
radians
kN metres
Substituting these expressions in equations (18.49) and (18.59) we obtain the tank
392
Roll stabilisation
[Ch. 18
20
cCll
"0
15
()l
....E:
Q)
10
01
!}!~
0
2.0
w
Q)
Cll
Cll
Q)
en
Cll
..c
a...
This shows that the stabilising moment is a near maximum at the tank natural
frequency Cth 1 and that the phase at this frequency is
Evidently the optimum tank performance will be assured if we arrange for the
tank and ship roll natural frequencies to be the same. This ensures that the stabilising
moment is a maximum and leads the roll motion by 90 at the natural roll frequency.
18.4.4.2
Equation (18.66) gives the tank natural frequency as a function of the tank
dimensions hct, w, wr and the depth of the fluid hr. As an example of the use of this
Passive tanks
Sec. 18.4]
393
equation Fig. 18.38 shows the effect of changing tank dimensions and fluid depth on
the tank natural frequency. These graphs are for an initial tank design having the
characteristics given above.
The tank natural frequency decreases with the widths w and wrand increases with
the duct depth hct. However, the natural frequency is quite insensitive to the depth of
fluid hr in the tank. It follows that there is little scope for adjusting the natural
frequency after the tank has been designed and fitted to the ship.
(18.71)
!\
=1
2 ~~
kN metres/radian
I fl
(18.72)
Now we have seen that the natural frequency is determined by the major tank
dimensions w, wrand hct. So the required peak stabilising moment is best achieved by
choosing appropriate values of the remaining parameters p1, x 0 ~~ and 'ct The
stabilising moment increases with the fluid density p1 and the tank length x1
Fig. 18.39 shows the variation of peak moment per metre length of tank with the
vertical location of the duct relative to the ship's centre of gravity for the specimen
tank characteristics given above. This shows thanhe tank becomes more effective if
it is located high in the ship (i.e. rct is small).
1.0
1.0
u
=a
Q)
(J)
Q)
"C
[Ch. 18
Roll stabilisation
394
0.5
0.5
3
0
40
w, (metres)
w (metres)
u
=a
1.0
1.0
u
=a
Q)
(J)
Q)
(J)
0.5
~~]
'
h,
3
J
I
4
I
2
0.5
0.5
1.0
2.0
hd (metres)
h,(metres)
c
Q)
E
0
E
20
0
rd (metres)
where F4 is some steady applied roll moment. Now the tank angle is, from equation
(18.49),
=  x4
radians
Sec. 18.4]
Passive tanks
395
m g GMs (1  JJ1)
X4
= F4
kN metres
(18.74)
and it is usually found that a satisfactory degree of stabilisation can be achieved if m 1
is ofthe order of 15% ofthe ship mass.
18.4.4. 7 Tank damping
The tank damping may be adjusted by installing an obstruction or a throttle valve in
the duct as shown in Fig. 18.31. Fig. 18.40 shows the effect of increasing the tank
Unstabilised
10
Speed 0 knots
Beam waves
Q)
"0
.~
c.
E
"'c.
Q)
Stabilised, ~=0.2,
high tank damping
"iii
Q)
>
~"'
a:
\j '\
2
Stabilised,
~=0.2,
1.2
Frequency (radians/second)
_,,_
Fig. 18.40 Typical roll transfer functions showing effect of passive tank damping.
Roll stabilisation
396
[Ch. 18
damping on the roll transfer function for a ship in beam waves at zero speed. With
low tank damping the roll response peak at the natural roll frequency is effectively
eliminated, but this is at the expense of resonant peaks at higher and lower
frequencies. These indicate that the tank will amplify the motions at these frequencies, possibly leading to an overall amplification ofthe roll motion, depending on the
shape of the wave energy spectrum. They may be eliminated, or at least reduced, by
increasing the tank damping. The motion at low frequencies is still amplified but this
is a characteristic common to all passive tanks since the loss in stability ensures that
the 'stabilised' roll motion at zero frequency always exceeds the unstabilised roll
motion.
15
0 knots
Cl

Q)
Unstabilised roll
::!:!.
~
Cl
c:
"'
e
rn
200
15
Cl
Q)
::!:!.
20 knots
~
Cl
c:
co
e
rn
200
Probability
0.0 of exceeding
1
0.1
150
200
Heading (degrees)
Tmax
Sec. 18.4]
Passive tanks
397
and a tank mass of 1.87% of the ship mass. At zero speed the tank gives a useful
reduction in roll motion at all headings. This is because the inherent roll damping due
to the hull, bilge keels and other appendages at low speed is small and the damping
provided by the tank makes a substantial additional contribution. At 20 knots the
hydrodynamic damping is much higher and the contribution provided by the tank is
relatively insignificant. So the tank is unable to achieve a worthwhile reduction in roll
motion.
Fig. 18.41 also shows the penalty of the loss of stability at high speed in following
seas. The encounter frequencies are then very low and the tank amplifies the roll
motion.
Also shown in Fig. 18.41 is the rms tank motion for each speed. Equation (18.71)
gives the maximum permissible tank angle as
The rms tank angles corresponding to various probabilities of exceeding this level
may be estimated from Table 17.5:
Probability of
exceeding 24.2
0.1
0.01
0.001
11.3
7.4
6.0
and these are plotted in Fig. 18.41. Evidently the tank motion is sufficient to reach
the tops of the reservoirs and the duct about once in every 100 oscillations on the
worst heading in this particular sea condition. This would be regarded as satisfactory
in practice. A more frequent rate of exceedance would invalidate the calculation
which takes no account of any such limits in the tank's stabilising capacity. This could
be rectified by increasing the height of the reservoirs and the depth h. of the working
fluid.
19
Added resistance and involuntary
speed loss in waves
19.1
INTRODUCTION
The speed a ship can achieve in calm water is governed by its resistance, propeller
efficiency and the power of its engines. In rough weather the resistance may be
changed by the action of the waves and the wind and the resulting change in the load
on the propeller usually reduces the propeller efficiency. The speed the ship can
achieve for a given engine power is usually reduced by these effects. This 'involuntary' speed loss does not often amount to more than two or three knots but may still
result in substantial financial losses for merchant ships.
19.2
A ship towed in regular waves will have a fluctuating resistance as illustrated in Fig.
19.1. In head waves the mean value of the resistance will be greater than the calm
Resistance
Resistance in waves
Raw
Calm water resistance
Rc~L
Time
Sec. 19.2]
399
water resistance and the difference may be attributed to the effects of the waves.
The simple theory for this added resistance presented here is based on that
proposed by Gerritsma and Beukelman (1971) and has its origins in the strip theory
described in Chapter 9. We shall confine our attention to long crested head waves
which is generally accepted as the most severe case. Only vertical plane motions then
occur.
Consider the relative motion of a strip located x 81 metres forward of the centre of
gravity. At the water surface this is given by equation (13.14) as:
Since we are here concerned with the average relative vertical motion experienced
over the draught of the ship, it is appropriate to take the wave elevation at the local
mean draught D (equation (9.31)) rather than at the surface. In this case the relative
motion becomes
r3
= s3 
exp( kD)
metres
(19.1)
The force required to sustain this motion is, by analogy with equation (9.16),
(19.2)
and the work done by the strip in one complete cycle is then
(19.3)
The total work done by the whole~ship in one encounter period is obtained by
allowing ox 81 to approach zero and integrating equation (19.3) over the hull length:
6
Ls (
E 
1Tffie
J
0
d a33
U dx
1
b33
81
2
r 30
.,_
dx81
kN metres
(19.4)
400
[Ch. 19
where Rc is the calm water resistance and Raw is the added resistance due to the
waves, the additional work required to drive the ship through one wave length is
E = RawA
kN metres
(19.5)
Now the relative motion amplitude r 30 is proportional to the wave amplitude ~ 0
and it therefore follows that the added resistance in regular waves must be
proportional to the square of the wave amplitude. A suitable nondimensional added
resistance 'transfer function' applicable to all wave amplitudes must therefore
include the wave amplitude squared in the denominator. Fig. 19.2 shows a widely
accepted non dimensional form for plotting the added resistance in regular waves.
These results were obtained by Gerritsma and Beukelman (1971) and show an
encouraging comparison between their predictions and measurements on a model of
a fast cargo ship in regular head waves. The added resistance response peak occurs
when the relative motions are a maximum. In very long waves (low wave frequencies) the relative motions are very small (see Fig. 13.23) and the added resistance
tends to zero. In very short waves the relative motion approaches the wave amplitude
as the absolute motions become negligible. The added resistance is then due to wave
diffraction and reflection and approaches some small but finite value.
19.3
Consider the narrow band of frequencies centred on some encounter frequency roe in
the encountered wave energy spectrum shown in Fig. 14.2.
If we replace the wave components in this small range of frequencies by a single
sine wave, the amplitude of the sine wave must be, by analogy with equation (4.13))
Sec. 19.4]
401
Experiment
Theory
FN=0.25
L5 =152.5 m
0.8
1.0
Fig. 19.2Typical added resistance response for a fast cargo ship in regular head waves. (After
Gerritsma and Beukelman (1971).)
where
= Raw
s6
kN/metre2
(19.6)
is a dimensional added resistance response function. The total added resistance due
to all the wave components in the encountered wave spectrum is obtained by
allowing <>roe to approach zero and integrating to give
(19.7)
Fig. 19.3 shows the results of some typical calculations of the resistance in
irregular waves for a frigate. The resistance rises rapidly with significant wave height,
and the greatest increase relative to the calm water resistance occurs at low speed.
402
../
[Ch. 19
/
/
/
Q)
()
.""'
Vl
600
o'"'
~'+S'
'].:
Q)
cc
12
0
Significant wave height (metres)
(19.8)
where
D c is the drag force in kN
PA is the density of air in tonnes/metre 3
U is the speed of the ship in metres/second
As is the maximum crosssection area of the superstructure and above water part of
the hull in metres2 .
The drag coefficient C0 may be determined from wind tunnel tests of a waterline
model of the ship.
Waves are generally accompanied by wind and this increases the aerodynamic
drag to
Sec. 19.5]
Propeller characteristics
403
PROPELLER CHARACTERISTICS
The speed attained by the ship for a given resistance depends on the hydrodynamic
characteristics of the propeller. Fig. 19.4 shows a typical set of these characteristics
y
/
Rough water
selfpropulsion
0.05
Ko
0.2
0.1
I)
0.04
0.8
0.03
0.6
0.02
0.4
O.ol
I
w
I
Jc
I
0.6
1.0
Advance coefficient J
Fig. 19.4 Typical propeller characteristics.
404
[Ch. 19
obtained from cavitation tunnel tests on a model propeller. The diagram shows, in
nondimensional form, the thrust developed and the torque absorbed by the
propeller as a function of the advance coefficient. The advance coefficient is a
measure of the 'slip' of the propeller and is defined by
=~
(19.10)
Nd
where
UP is the mean velocity of the flow through the propeller disc in metres/second
N is the number of propeller revolutions/second
dis the propeller diameter in metres.
The velocity through the propeller disc is somewhat less than the forward speed
of the ship because of the effects of the boundary layer on the hull. The two velocities
are related by the Taylor wake fraction which is defined as
WT
uu...~::P
= __
(19.11)
from which we find that the mean velocity through the propeller disc is
up =
U(1wT)
metres/second
(19.12)
The Taylor wake fraction is usually of the order of 10% and may be measured in
suitable model experiments.
Combining equations (19 .10) and (19 .12) we see that the advance coefficient may
also be written as
(19.13)
KT
Ko
T
pNzd4
Q
pNzd5
(19.14)
(19.15)
Propeller characteristics
Sec. 19.5]
, =
power delivered
power absorbed
TU(1wT)
= '=
2TTNQ
405
(19.16)
Rc(1 + a) kN
(19.17)
Nc =
U(1 wT)
fed
revolutions/second
(19.20)
406
[Ch. 19
(19.21)
The characteristics of the propeller are determined for the steady flow conditions
experienced in calm water. In rough weather the waves and the motions of the ship
will cause considerable fluctuations in the flow around the propeller but it is generally
assumed that the operating characteristics of the propeller, the Taylor wake fraction
and the resistance augment will remain unchanged from their calm water values. In
this case the selfpropulsion advance coefficient and the power required to drive the
ship at a given speed in waves may be determined in exactly the same way as for calm
water. The resistance is now given by
(19.22)
(19.24)
The self propulsion advance coefficient Jw for rough water is determined by the
intersection of equation (19.23) with the KT versus] curve as shown in Fig. 19.4. The
effect of the added resistance is to reduce the selfpropulsion advance coefficient so
that the propeller revolutions required to maintain a given speed become:
N
w
1
U(  wT)
fwd
revolutions/second
(19.25)
The propeller efficiency is reduced and the power required to maintain this speed is
(19.26)
Sec. 19.6]
Speed loss
407
that the engine delivers constant power at a given throttle setting regardless of the
load.
The power required to drive the ship/propeller combination at a given speed
in a specified wave system may be calculated using the methods described above. Fig.
19.5 shows the results of such a calculation for the ship whose resistance characteristics are given in Fig. 19.3, fitted with the propeller of Fig. 19.4. As expected the
power rises steeply with forward speed and significant wave height.
It is convenient to cross plot these results in the manner shown in Fig. 19.6. This
shows the power as a function of speed for a given significant wave height. It is then a
simple matter to read off the speed that can be achieved at a given power level for a
number of significant wave heights and plot the results as shown in Fig. 19.7.
The speed loss is quite small at high power. At lower power levels much more
dramatic losses occur and in extreme cases the speed may be reduced to zero.
0
Significant wave height (metres)
[Ch. 19
408
10m
20
Q)
5
0
c...
10
10
15
20
25
30
Speed (knots)
Fig. 19.6 Power in head waves for a frigate.
20
(/)
0c
:::.
'0
15
Q)
Q)
0.
(f)
20
Slamming, deck wetness and
propeller emergence
20.1
INTRODUCTION
The relative motions between the ship and the water surface are generally largest at
the ends of the ship. In high waves the motions may be so large that the forefoot and
propeller are exposed and the deck submerged. This occurs most frequently at high
speed in head waves although it is not unknown in <;~ther conditions.
The reentry of the keel after emergence may result in a substantial impact or
'slam' as the ship's bottom strikes the water surface. Ships with heavily flared bows
may also experience similar, but less severe, impacts under the bow flare even when
there is no keel emergence. These slam impacts may be se'Vtjfe enough to cause local
structural damage to the ship's plating. In extreme cases the loading may be sufficient
to distort the ship's hull permamently and some ships are believed to have broken up
following slamming. Even moderate slamming will cause the hull to vibrate at its
natural frequency (generally of the order of a few cycles per second) and the resulting
fatigue loading will reduce the life of the hull. The vibration following a slam is called
'whipping' and often provides the captain with his first indication that a slam has
occurred.
Deck wetness may occur anywhere along the length of the ship, particularly
where the freeboard is low. However, the most severe deck wetness generally
occurs, like slamming, at the bow at high speed in head waves. In these conditions the
forward speed of the ship accentuates the effects of the water shipped onto the
foredeck, and damage to deck fittings and cargo may occur. Any crew or passengers
on deck may be injured or washed overboard. In extreme cases the ship might even
capsize and sink due to the weight of water taken on board.
Propeller racing will begin to occur when the upper tips of the blades emerge from
the sea surface. The sudden reduction and subsequent increase of torque loading as
the propeller becomes fully Ubmerged again may damage the engine and propeller
410
[Ch. 20
Clearly these phenomena are undesirable and the prudent captain will try to
avoid them if possible. Since they generally become more severe at high speed they
impose an effective limit on the ship speed in rough weather, especially in head
waves. This aspect is discussed in Chapter 22.
7
*
L~....__,.___~
"f
Xs1M
D,
Designwaterlineat
zero speed
Dpe
Actual running
waterline
r,
which depends on the speed. This gives a steady relative motion r3 even when the
ship is running in calm water. If we assume that this datum relative motion is
unchanged when the ship is in waves, we may regard it as a change to the draught,
freeboard, etc. Thus the effective draught to the keel becomes
Dke
= D + r3 metres
(20.1)
(20.2)
Probability of occurrence
Sec. 20.2]
411
and the effective depth of the tips of the upper propeller blades is
(20.3)
where f 3 is of course determined at the appropriate location on the ship.
The notional relative motion in waves is obtained by subtracting the wave
depression from the absolute motion (see equation 13.14). As the hull dips into the
water the increasing submerged volume causes a local 'swellup' of the water surface.
The effect disappears as the hull rises. This enhances the relative motion over and
above the notional value. We define a swell up coefficient as
C
s
(20.4)
and it is found that C5 is a function of hull form, location on the hull, speed and wave
length. At the time of writing no universally accepted method of calculating C 5 has
been developed, but it has been measured in model experiments by several authors.
Probably the most comprehensive set of experiment data was published by Blok and
Huisman (1985). A selection of their results for a small frigate in head waves of
length equal to the ship is shown in Fig. 20.2.
Referring to equation (17.26), the probability . of the local relative motion
exceeding the effective draught (i.e. the probability of keel emergence) is
p ke = exp ( 
2 Ct mo)
Dke
(20.5)
where m0 is the variance of the notional relative motion at the appropriate location
on the ship. A similar expression may be written for the probability of deck
submergence
1 p2 )
(20.6)
p pe  exp
(!~)
2 Ct n:Jo
The average period of the peaks may be calculated from equation 4.27 as
(20.7)
412
[Ch. 20
<.J
c
Q)
u
~
Q)
0
(.)
0.
::J
a:;
(/)
0.5
(20.8)
where m2 and m 4 are the variances of the notional relative motion velocity and
acceleration at the appropriate location on the ship. The average number of keel
emergences, deck submergences and propeller emergences per hour are then
N
3600
per hour
(20.9)
per hour
(20.10)
(20.11)
ke
Pke
N ds 3600
T p ds
p
pe
Sec. 20.3]
Slamming
413
20.3 SLAMMING
20.3.1 Introduction
Fig. 20.3 shows a frigate at high speed in rough weather. The relative motion is
sufficient to expose a considerable length of the keel and a slam is clearly imminent.
The ship is shrouded in spray from a deck wetting which has just occurred and much
of the water that was shipped is pouring down the sides to return to the sea beneath
the keel. Some water may also be drawn up under the keel as it emerges from the sea
surface. This can be seen in Figs 20.4(a) and 20.4(b) which show a model ofthe ship
being tested in a towing tank.
The subsequent slam will be into a mixture of air and water which probably helps
to cushion the impact. Slamming impact loads are also affected by the local hull
section shape, the relative velocity at impact, the relative angle between the keel and
the water surface, the local flexibility of the ship's bottom plating and the overall
flexibility of the ship's structure. A complete prediction of slamming phenomena is a
complex task which is beyond the scoge of any existing theory.
Fullscale measurements of slamming pressures at sea are rare because of the
practical difficulties involved, but Fig. 20.5 shows a slamming pressure time history
recorded by Sellars (1972) on the keel of the US Coast Guard vessel Unimak during a
severe slam. The pressure Fises very quickly after the initial impact: indeed it is
suspected that the measured rise is probably alway~limited by the relatively sluggish
response of the pressure transducer.
414
[Ch. 20
(a)
(b)
Fig. 20.4 (a) Model slamming experiment. (b) View from ahead. (MoD Photo.)
415
Slamming
Sec. 20.3]
1500
(/)
Q)
:::.
~
::J
(/)
(/)
Q)
a:
0.1
0.2
0.3
Time (seconds)
characteristics of these sections may therefore be examined by dropping twodimensional wedge sections into wate&.
Dimensional analysis (Chapter 16) suggests that the peak impact pressure
developed on the wedge during a siam will be given in the form
P = CP i p rj
kNftnetre2
(20.12)
416
[Ch. 20
Symbol
Body
Curved
wedge
20 wedge
D.
0
Source
Cone
Hagiwara and
Yuhara (1974)
Chuang (1970)
Chuang and Milne
(1971)
100
'
o".....
'
5
Slamming
Sec. 20.3]
417
result derived by Wagner (1932) in connection with impact loads on seaplane floats:
(20.13)
Slamming pressures estimated using these results are likely to be too high because
twodimensional drop tests take no account of the effect of the air/water mixture
likely to be below the keel just before impact; nor do they allow for the effects of the
relative 'pitch' angle between the keel and the water surface. All these effects are
likely to decrease the impact pressure. Nevertheless, the results do give a general
indication of the effect of deadrise angle and confirm the basic physics of the
phenomena involved.
20.3.3 Model experiments in waves
The alternative approach of measuring slamming pressures on scale models in waves
has been pursued by a number of workers in the field. However, scale effects are
likely to be important and it is always difficult to measure the relative velocity at
impact.
Ochi (1964), in a classic paper, described measurements of slamming pressures
on a model of a merchant ship in waves. Although he was unable to measure the
impact velocities directly, he confirmed the relationship given in equation (20 .12) but
found that no appreciable slamming impacts occurred if the relative velocity was less
than a certain value. According to Ochi this critical velocity was about 3.7 metres/
second for a 161 metre ship. Assuming that the critical velocity obeys the Froude
r rl
scaling law' it may be expressed as
5)
metres/second
(20.14)
Pslam
D~e
rlcrit )
= exp (  2C?mo 2C?mz
(20.15)
(see equation (17.35)). m 0 and m 2 in this formula are the variances of the notional
418
[Ch. 20
_ 3600 Pslam
T
slam
per hour
(20.16)
metres
(20.17)
t=O
3 r3
Impact
Keel emerges
(20.18)
Or3 = sin  l
(20.19)
so that
( 
Dke)
radians
r3o
Slamming
Sec. 20.3]
r = ror r
3
30
419
metres/second
(20.20)
rio )
P=exp ( 1 2 C82 m0
(20.22)
and this peak relative motion will occur once inN oscillations, where
(20.23)
where THis an arbitrary sample period. Hence the relative motion amplitude which
will be exceeded once in TH seconds is
r 30 = Cs cr0
~[
2 loge ( T
~:J J
metres
(20.24)
ro r3
yf(mmz
4
radians/second
(20.25)
Using these equations we may now estimate the peak slamming pressures which
are likely to be exceeded once in T H seconds. Fig. 20.9 shows the results of a
specimen calculation for a frigate in head seas with T Hset at 900 seconds. Fig. 20.9(a)
shows the deadrise angle at the keel and the corresponding slamming pressure
coefficient CP given by equation (20.13). The rms relative motion and velocity for the
chosen speed and wave conditions calculated by strip theory are shown in Fig.
20.9(b). Fig. 20.9(c) shows the derived relative motion amplitude and impact
velocity given by equations (20~24) and (20.21 ). These represent the worst conditions
likely to occur in a period of900 seconds. Note that the impact velocity is a maximum
at the forward end of the ship and falls to zero at some location where keel emergence
420
[Ch. 20
60
(a)
01
QJ
~
QJ
Ol
:::J
<J)
40
<Jl+'
QJC
~QJ
c..u
co
0);,;:::
QJ
C"'"
 QJ
co
EO
Eo
co
(jj
'0
QJ
,;
,....._..Relative velocity
/
~
Critical velocity
400 (d)
200
Fig. 20.9 Slamming pressure calculation for a frigate in head waves; TH = 900 seconds.
is unlikely. Some of the predicted impact velocities are less than the critical velocity
calculated according to equation (20 .14). It is assumed that no slamming will occur at
these stations.
Sec. 20.4]
Deck wetness
421
Finally Fig. 20.9(d) shows the slamming pressure calculated according to equation (20.12). The pressure is set to zero where the impact velocity is less than the
critical velocity.
422
20.5
[Ch.20
FREEBOARD EXCEEDANCE
Freeboard exceedance
Sec. 20.5]
423
/
Actual relative motion I
E.
"
...... \
0
~
Q)
>
.;::;
"'
~
II)
0.5
10
E
"2
"'0
..0
Sheer line
Q)
Q)
U:
OWL
0.1
Q)
c 0.4
._co
()
oo
>Q)
~
:=x
;!:::::
..OQJ
"'o 0.2
..o~
Oro
~
C....o
Q)
.....~
0
Xs1M/L,
Fig. 20.12 
ship. Equation 20.6 then gives the probability of freeboard exceedance which is
typically a maximum some way abaft th~ stem.
While this result is typical it will generally overestimate the actual frequency of
deck wetness since most of the freeboard exceedances which occur at this location
will not result in water coming onto the foredeck. Fig. 20.11 shows this pheneomenon very well. Lloyd, Salsich and Zseleczky (1986), in an extensive series of model
experiments, found that the observed frequency of geck wetness was most closely
424
[Ch. 20
correlated with freeboard exceedances at the stem head. In other words a deck
wetting is almost always the result of a freeboard exceedance at the stem head and
freeboard exceedances elsewhere do not result in wetness unless they are also
accompanied by an exceedance at the stem head. It follows that the frequency of
freeboard exceedance at the stem head will probably give a reasonably accurate
estimate of the true deck wetness frequency at least at high speed in head waves.
20.6 EFFECT OF BOW SHAPE
It seems obvious that deck wetness frequency and severity must be affected by the
above water form of the bow. It is therefore suprising to find that there is no universal
agreement on the effects of features like flare, stem rake or knuckles. The available
objective experimental evidence is confused and contradictory and remains the
subject of active research.
High freeboard is the only characteristic which is universally agreed to have a
beneficial effect on deck wetness.
21
Effects of ship motions on
passengers and crew
21.1
INTRODUCTION
Ship motions have two undesirable effects on the people within the ship. They cause
motion sickness and also make it more difficult to move in a controlled and coherent
manner so that the performance of everyday tasks is impaired.
The balance organs located in the inner ear qm detect changes of both the
magnitude and direction of the apparent gravitational acceleration as well as angular
accelerations. Excessive stimulation of these organs will, in most individuals, result
in motion sickness. The condition will to some extent be alleviated if the accelerations are confirmed by visual cues from the eyes. Thus~~ ride on a fairground
switchback railway can be enjoyable and exciting even if large accelerations are
experienced. The same accelerations experienced by a blil;1dfolded rider would
almost certainly result in quite distressing motion sickness. In the same way it is
possible to stimulate motion sickness without any motions being present at all. This
can be done, for some individuals, by showing them a film of a violent fairground
ride.
On board ship it follows that motion sickness is most likely to occur if passengers
or crew are confined below decks so that they cannot see the horizon. Bittner and
Guignard (1985) also showed that motion sickness can be exacerbated by facing
diagonally across the ship. Fore and aft or athwartships seating is to be preferred.
Other factors which may promote seasickness are anxiety, fatigue, hunger, smells
(particularly cooking and fuel oil), greasy food, reading and carbonated or alcoholic
drinks. Nieuwenhuijsen (1958) found that women and young children in the liner SS
Masdam were more susceptible to sea sickness than men. Elderly people were
generally less affected than people of middle age.
Fortunately the symptoms of seasickness usually disappear after a few days. Fig.
21.1 shows this effect in grapj;lical form, based on a study of seasickness in the Royal
Navy. Nevertheless, seasickness remains a deterrent to travel by sea for many people
426
[Ch. 21
30
CD
CD
""0
c;
c
20 I
10 I
Ul
Ul
CD
"'
u
u;
c
.'2
Day of cruise
Fig. 21.1 Motion sickness incidence: effect ')f acclimatisation. (After Walters (1964).)
21.2
Sec. 21.2]
427
windows so that the subjects could not receive any visual motion cues and their only
task was to monitor their state of nausea by pressing buttons on a control panel. The
experiments lasted for up to two hours or until the subjects vomited.
O'Hanlon and McCauley found that the 'motion sickness incidence' (the percentage of subjects who vomited within two hours) could be expressed in the form
(21.1)
where
ls3 l is the vertical acceleration averaged over half a motion cycle and
J.LMSI
(21.2)
Equation (21.1) may be evaluated with the help of Table 17.3 and is plotted in
Fig. 21.2. This shows how MSI increases with acceleration and is most severe at a
100rT,~
QJ
QJ
"'0
"(3
"'"'
c
QJ
"
;;;
c
0
."
2:
2
<rle
(radians/second)
Fig. 21.2 Motion sickness incidence. (After O'Hanlon and McCauley (1974).)
428
[Ch. 21
requires us to make assumptions about the relationship between the random motions
of the ship and the sinusoidal motions of the simulator. Consider an irregular vertical
acceleration time history measured on board a ship. If we assume that the accelerations are distributed according to the Gaussian probability density function then
the average modulus of the acceleration is given by equation (17.23):
(21.3)
where m 4 is the variance of the vertical acceleration. The average period of the
acceleration peaks is, by inference from equation (4.27),
TP =
2rr
~ (::)
seconds
where
is the variance of the second derivative of the acceleration (see equation (4.24)).
These integrals must be evaluated using numerical methods. In practice it may be
difficult to achieve a satisfactory result because the high powers of me will ensure that
the integrals will not converge until the frequency is very high. The final result will
therefore depend on the accurate estimation of the very small motion spectral
ordinates at these very high frequencies. In practice it is better to assume that the
average period of the acceleration peaks is the same as the average period of the
displacement peaks
(21.4)
so the average frequency of the motion may be taken as
~e
UJ

'J/(mmz4)
radians/second
(21.5)
Subjective motion
Sec. 21.3]
429
MSI
JMSIWdxBlM
JWdxBlM
~
(21.6)
where the integrals are evaluated over the length of the ship. Fig. 21.3 shows a simple
weighting function giving equal weight to all the occupants of the ship over the length
of the passenger accommodation. For this example the weighted average MSI is
1
MSI = 7%
rl
An experienced, well motivated and acclimatised crew will not suffer unduly from
seasickness but may still find that vertical ship motions will inhibit their ability to
work effectively. No satisfactory method of properly estimating these effects has yet
been developed but one technique which has been used is based on some experiments by Shoenberger (1975).
He subjected eight experienced US Air Force pilots to vertical sinusoidal motions
in a simple oscillating chair capable of amplitudes up to 1.5 metres. The pilots
wen~ blindfolded to remove any visual motion cues. After some preliminary
experiments they were subjected to ~a 'standard' reference motion of 0.6g at
1.0 Hz. This motion was assigned value of 10 on an arbitrary 'subjective motion'
(SM) scale. The frequency and amplitude of the motions were then changed and each
subject was asked to assess the severity of the new motion in relation to the standard.
Thus a motion which felt twice as severe was assigned an SM of 20 and one which felt
half as severe was assigned a value of 5.
The subjects were generally able to make their assessments within a minute and
430
[Ch.21
1.5
c
0
.;o
"'Q;
a;_
UN
""
"'"'
"'
"'uE
t:>"'
"'
0
.~S:
~~
.s::o
"'~
"
s:"'"'
::~ i : : : : : : i : ~
0
25
"'
Ill
~~
20
t5;
"
"'c
CCV
.;::oc
u
oc
:2'
20
~~
.s::~
.2'UJ
~:2
0
0.5
Stern
0
Xs1M/Ls
0.5
Bow
Fig. 21.3 MSI calculation for a passenger ferry at 10 knots in head waves;
H113 : 5.5 m, T0 = 12.4 sec.
none suffered from motion sickness during the experiments. Shoenberger obtained
remarkably consistent results which could be expressed in the form
(21.7)
Subjective motion
Sec. 21.3]
431
SM WdxBlM
SM
= .:....___ _ __
(21.10)
where the integrals are evaluated over tne length of the ship. For the example shown
the weighted mean SM is
SM
= 14
432
[Ch. 21
80r,,,
A
10
P>e
Experiment data
20
30
(radians/second)
30~~~~
Intolerable
20
c
.9

Hazardous
0
E
  
Q)
>
Severe : necessary
to 'hang on' all
the time
(.)
Q)
:.0
::J
(/)

10
Sec. 21.4]
433
c:
0
:;::;
.,
E!
a;_
<.><>
.,.,
(,)N
~
~E
".,
>
(I)
0~_.~~~~._~~~~~
40~~~~,~~~~~~
.,
.2: c:
o
<>..,_
0
JE
en
:2:
en
o
20
2l
..c:
a;
10
5:
0.5
Stern
Xs1M/Ls
Bow
Fig. 21.6 Subjective motion calculation for a frigate at 20 knots in head waves;
H 113 = 5.5 m, T0 = 12.4 sec.
434
[Ch. 21
as an occasion when a crewman would have to stop working at his current task and
hold on to some convenient anchorage to prevent loss of balance. They proposed a
relationship between LFE and the frequency of Mil experienced on the flight deck of
a destroyer. This is shown in somewhat modified form in Fig. 21.7.
"'"'
c
Extremely
hazardous
U;
c
0
g_
E
Q)
c
a
Severe
Q)
()
:::J
a
c
Serious
c0
Fig. 21.7 Motioninduced interruptions; natural roll frequency 0.59 rad/sec. (After Baitis,
Woolaver and Beck (1985).)
Baitis's estimates were for a ship with a natural roll frequency of about 0.59
radians/second. Since roll makes a dominant contribution to LFE it follows that the
number of motioninduced interruptions per minute will be increased (for the same
rms LFE) if the natural roll frequency is increased. An approximate correction for
ships with different natural roll frequencies may be applied by using
Mil
(1).4
Milo.s9 0.
59
per minute
(21.11)
Figs 21.8 and 21.9 show how LFE and Mil vary in beam waves with location in the
ship: evidently the lowest levels of Mil are found close to the keel at midships.
Sec. 21.4)
"'E
Q)
~~
Q)
...
om
~E
~
.2l
.!!!
Ul
E
...
'tl
Q)
::J
'tl
c:
~
0
.;::;
0
~
0~_.~_.
0.5
Stern
__.__.__
._~~~~
0.5
Bow
Fig. 21.8  Effect of longitudinal location on lateral force estimator and motioninduced
interruptions in a frigate in beam waves at 15 knots; H113 = 5.5 m, T0 = 12.4 sec.
435
436
[Ch. 21
...
.8
"'
~
Q)_
~N(,)
...
Q)
~~
E
m._
...
Q)
Ul
"'
c
I
I
::J
2c
""0
g_
4
Q)
(J
::J
""0
c0
'iS
2:
20
Height above keel (metres)
Fig. 21.9 Effect of height on lateral force estimator and motioninduced interruptions in a
frigate in beam waves at 15 knots; H 113 = 5.5 m, T0 = 12.4 sec.
22
Seakeeping criteria and voluntary
speed loss in rough weather
22.1
INTRODUCTION
The methods described so far in this book have been aimed at predicting the
responses experienced by a ship in rough weather. We now need to consider whether
these responses will be acceptable in practice. This requires us to determine limiting
values or 'criteria' for each of the responses we predict.
Ships are required to carry out many different tasks and activities at sea and
criteria for acceptable responses depend on the task in hand. For example, motions
which are acceptable in a warship hunting a submarine woul~ not be tolerated by the
passengers on a cruise liner. Deck wetness which might be acceptable in a frigate
closed down for a highspeed dash in rough weather would. not be condoned for
operations which require men to work on the exposed upper deck.
It is first essential to identify the responses which actually limit performance of
the task. Consider, for example, a helicopter landing on the deck of a small frigate.
This is undoubtedly inhibited if the ship motions are large and we might expect the
following problems to arise in severe conditions:
(a) A large relative velocity between the flight deck and the helicopter might
damage the undercarriage or even the deck itself.
(b) A steady roll angle will "ause the wheels on one side of the aircraft to touch down
before those on the other with consequent greater risk of sliding across the deck.
(c) A high lateral velocity might topple the helicopter on touchdown or even cause it
to land with a wheel over the edge of the deck.
(d) Deck wetness and spray reduces visibility and deck friction making sliding more
likely: spray may also be ingested and damage the aircraft's engines.
For this particular example relative wind conditions will also play an important
438
[Ch. 22
role. Most helicopters can be satisfactorily controlled over only a limited range of
wind speeds and directions.
Criteria should be expressed in terms of these responses (and any others deemed
to be important) and not in terms of responses which have no direct relevance to the
activity being considered. For example, a heave motion criterion would be inappropriate for helicopters (unless the flight deck is over the centre of gravity) because
heave has no direct influence on the helicopter performance.
Having determined the motions which are believed to influence performance, the
next step is to estimate numerical values of the criteria. The ideal method would be to
conduct controlled trials in a variety of sea states to measure performance at the task
as a function of measured ship motions and to determine the conditions in which the
task becomes essentially impossible. Unfortunately this is generally impractical for
the following reasons:
(a) Sea state cannot be varied in a controlled manner and trials covering the
required range of conditions would be very protracted and expensive.
(b) Tasks involving human intervention (as most tasks do) would be heavily
influenced by all the psychological and physiological factors to which humans are
susceptible (motivation, morale, fatigue, acclimatisation etc) and it would be
very difficult to obtain consistent results unless a very large number of experiments with different personnel was conducted. These experiments would
inevitably be conducted in a variety of sea states making a proper analysis very
difficult.
Trials in ship motion simulators have the advantage that the same sequence of
ship motions may be reproduced over and over again so that controlled experiments
on a large number of subjects can be undertaken. However, most simulators are
incapable of reproducing the very large ship motions which are required to inhibit
performance and they cannot usually reproduce all six degrees of freedom (and other
more intangible aspects of life at sea) simultaneously. Furthermore, practical
experiments in most simulators are limited to nonstrenuous tasks rather than tasks
requiring great physical effort and manual dexterity.
The only practical method of determining criteria for most tasks is to observe the
apparent performance of actual ships' crews in the ordinary everyday ocean
environment. This can be done in a number of different ways. One of the most
successful is to collect performance data using questionnaires and this is discussed in
Section 22.3 below. An alternative is to monitor ship motions over a long period of
time (months or years) and to correlate the measured motions with the activities of
the ship recorded in the log book. This will eventually build up an envelope of
motions which were actually tolerated while specific tasks were being performed in
the ship.
We may summarise the requirements discussed above in terms ofthree 'rules' for
the determination of criteria:
(1) Criteria must be related to a particular task or mission.
(2) The responses chosen for criteria assessment should be of actual concern to the
mission being considered.
Sec. 22.2]
Equipment criteria
439
(22.1)
where cr0 is the rms fin motion. The time the fin is being eroded is then given by
seconds
(22.2)
where THis the total time spent in the cavitation prone conditions of the calculation.
On the other hand we r.ight be more interested in the frequency with which the
fin emits bursts of cavitation noise. We are then concerned with the probability that
the peak fin incidence in each motion cycle will exceed cxcrit This may be obtained
from the Rayleigh probability density function by using equation 17.26:
1 ( cx~:t
P( ex > CXcrit) = exp [ " 2
)2]
(22.3)
440
[Ch. 22
The mean period of the fin motion peaks is, from equation (4.27),
2rr~ (::)
seconds
(22.4)
where m2 and m 4 are the variances of the fin motion velocity and acceleration. The
critical incidence is then exceeded
TP
(22.5)
t and N in these formulae may be regarded as criteria for acceptable fin motions.
Numerical values of these criteria are best established by monitoring the actual
performance of ships at sea as required by Rule (3).
22.3 QUESTIONNAIRES
22.3.1
Introduction
Questionnaires provide one of the few practical methods of obtaining data on actual
performance at specific tasks at sea. Part of a typical questionnaire to obtain criteria
relating to damage to deck cargo on container ships is shown in Table 22.1. The
compiler of the questionnaire should always bear in mind that the recipient will
probably not be very interested in the business of criteria determination and the
questionnaire should therefore be made as brief and self explanatory as possible.
Otherwise it is likely to be consigned to the waste paper basket rather than be
properly completed. The questionnaire should therefore open with the minimum
number of questions designed to establish the identity of the recipient and his ship
and follow with a brief summary of the scenario postulated. A statement promising
confidentiality should always be included since some recipients (and their
employers) may consider any confession of damage experience as an admission of
poor seamanship. Questions soliciting opinions on design changes often yield fruitful
results and have the added advantage of boosting the ego of the recipient, thus
encouraging a timely response.
Ship operators will rarely be able to provide reliable direct estimates of limiting
ship motions or other criteria because they generally have no measurement systems
available to monitor the rough weather behaviour of their ship. Sailors are, however,
often reasonably well schooled in estimating wave conditions and much more fruitful
results will be obtained by asking for estimates of the worst sea state in which a
particular task can be completed. Subsequent calculations based on strip theory or
441
Questionnaires
Sec. 22.3]
Sea state
Sea state
(1)
(2)
(3)
(4)
(5)
(6)
Thank you for your assistance. Please now return this questionnaire to:
model tests may then be used to estimate the corresponding motions (or other rough
weather phenomena) in suitable idealised sea spectra. If this technique is used, the
speed and heading must, of course, bespecified in the questionnaire.
22.3.2 Analysis of numerical data
22.3.2.1. Mean value and staqdard deviation
,
Numerical data such as sea state estimates may be analysed to determine mean
values and standard deviations using the formulae given below. If the questionnaires
442
[Ch. 22
yield N estimates of a numerical quantity x (say the limiting sea state for a particular
activity), the mean value is
X=
(22.6)
cro
f(L[(xx)Z])
N 1
(22.7)
\j
where
tcrit
(22.8)
where a A> cr8 so that v'F > 1.
If y'F is greater than the critical value given in Fig. 22.2 there will be a 95%
probability that the two standard deviations are significantly different. If y'F is less
Sec. 22.3]
Questionnaires
443
50
Number in sample N
Fig. 22.1 Critical values of Student's t function for 95% confidence level.
5
Ns=3
4
\ Fcc;t
rl
5
2
10
20
X
444
[Ch. 22
than y Fcrit the samples may be combined to give a single estimate of standard
deviation for both samples:
cro
(22.9)
'V
If the standard deviations are not significantly different and have been pooled in
this way we may apply a further test to establish whether the two mean values are
significantly different. If the test function
(22.10)
exceeds the critical value tcrit given in Fig. 22.1 the two means are significantly
different at the 95% confidence level (i.e. there is a significant difference in the
performance of the two ship classes). N is taken as the total number of returns from
both classes for this test. In this formula Sect is the standard error of the differences
defined by
(22.11)
Example
A questionnaire issued to two ship classes yields the following data for limiting sea
state for a particular activity:
Class A
Class B
Mean value
Standard
deviation
Number in
sample
5.7
4.9
1.2
1.4
7
10
What confidence levels can be attributed to each individual result? Are the results
significantly different?
Using Student's ttest and Fig. 22.1 there is a 95% confidence that the true mean
value for each sample lies within the following ranges:
Class A:
5.7 1.08;
Class B:
4.90.98;
Using the F test we find that the ratio of the standard deviations is
Questionnaires
Sec. 22.3]
445
= 2.0
so that the two standard deviations are not significantly different. Equation (22.9)
then gives the combined standard deviation as
cr0 = 1.32
and the standard error of the differences is, from equation (22.11),
sed
= 0.65
1.23
and there is a total of 17 samples from the combined questionnaires. Fig. 22.1 gives
the critical value of t as
lcrit
2.04
and the results obtained from the two classes are seen to be not significantly different
at the 95% confidence level. The two samples may therefore be combined to yield a
mean value
X=
5.7x7+4.9x10
17
5.2
and Student's t test may again be applied to find the confidence limits for this
combined result. We find a 95% confidence that the true mean value will lie within
the range
5.2 1.32 X 0.53
5.2 0.7;
It can be seen that combining the samples, where this can be justified, results in an
improved estimate of the mean value.
446
[Ch. 22
Outcome number
Return
number
1
2
3
4
10 11 12 13 14 15 16
N
N
N
N
y
N
N
N
N
y
N
N
y
y
N
N
N
N
y
N
y
N
y
N
N
y
y
N
y
y
y
N
N
N
N
y
y
N
N
y
N
y
N
y
y
y
N
y
N
N
y
y
y
N
y
y
N
y
y
y
y
y
y
y
Suppose that the questionnaire yielded a majority vote (i.e. three or more of the
recipients voted 'yes'). This could be achieved in the five different ways marked with
an asterisk. The probability that this result occurred by chance is therefore 5/16 =
0.3125 and the probability that it was not a chance result is 10.3125 = 0.6875. So
the confidence level that the majority of a hypothetical much larger population of
ships' masters would vote positively is 68.75%.
Standard practice in statistical analysis is to demand a confidence level of at least
95% and we can see that this cannot be achieved with a sample of only four returns.
Even if all four recipients respond with a 'yes' vote there is a 1/16 chance that the
result is by chance and the confidence level that this would indicate a majority
opinion in the world at large is only 11/16 = 93.75%.
We may extend this approach to an arbitrary number of questionnaire returns as
follows. We require to find the minimum number of votes q for a sample size N which
indicate a majority opinion with a 95% confidence level. Consider a questionnaire
sample of N returns with r 'yes' answers to a particular question. The proportion or
vote for the opinion expressed is
Questionnaires
Sec. 22.3]
447
(22.12)
Nc,
N!
r!(Nr)!
(22.13)
(22.14)
If the answers to the questions were truly random all pvssible combinations of
answers would be equally likely. Hence the probability of q or more 'yes' answers
would be
P=
(22.15)
and the probability or confidence level that this is not a chance result is
:II
C
1P
(22.16)
Table 22.2 gives the minimum number of positive responses q required for a 95%
confidence level as a function of the sample size N, calculated using the equations
derived above. From this it can be seen that the absolute minimum number of returns
required is five and they must all vote positively before this can be accepted as a valid
indication of a majority opinion. The required proportion of positive votes falls as
the sample size is increased. Note that the same analysis applies with equal validity to
negative opinions (i.e. not ticking a box).
22.3.4 Increasing the sample size
Evidently there are considerable benifits to be obtained if the sample size is made as
large as possible. Unfortunately it is usually only the largest navies which have more
than, say, half a dozen ships in a class and we have seen that this will generally yield
results of only marginal reliability. For some aspects of seakeeping it may be possible
to pool the results from more than one class of ship, as described above, but this of
course precludes the possibfity of distinguishing any effects of design differences
between the classes.
448
[Ch. 22
Table 22.2 Minimum number of ticks for majority vote at 95% confidence level
Sample size
N
Number of
returns q
ticking a
given box
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
5
6
7
7
8
9
9
10
10
11
12
12
13
13
19
20
21
22
23
24
25
26
27
28
29
30
14
15
15
16
16
17
18
18
19
19
20
20
Sample size
N
Number of
returns q
ticking a
given box
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
21
22
22
23
23
24
24
25
26
26
27
27
28
28
29
30
30
31
31
32
32
52
53
54
55
56
57
58
59
60
33
33
34
35
35
36
36
37
37
Sec. 22.4]
22.4
449
One 'mission' which has received considerable attention is the ability to maintain
speed in severe head seas. This is often regarded as a general indication of the
seakeeping qualities of a ship since excessive motions, slamming, deck wetness, etc.
force the captain to reduce speed to avoid damaging his ship and its contents and
injuring his crew and passengers.
A typical questionnaire designed to obtain data on speed loss in rough
weather is shown in Table 22.3. Some results for two classes of frigates are shown in
Table 22.3 
Name of captain
Name of ship
How long have you served in this ship?
Ship speed is limited in rough weather by two factors:
(a) In moderate sea states the action of the wind and waves causes the ship to slow down even if
full power is maintained.
(b) In more severe sea states the captain may decide to reduce power or change course in order to
alleviate slamming, deck wetness, propeller emergence, ship motions, etc.
Imagine that you are required to make a highspeed passage in rough weather. All ship
equipment should be fully operational at the end of the pasage. In the table below please indicate
the maximum speed that you could maintain in your ship in the given sea states. Please also
indicate with a tick the sea state in which you would first reduce power.
Sea state
(please state)
Fig. 22.3. Estimates of mean speed and standard deviation have been calculated
using equations (22.1) and (22.2). Asexpected the ships suffer a dramatic speed loss
as the sea state worsens. The standard deviation increases in high sea states,
reflecting the difficulty of estimating speed loss and perhaps the lack of extreme
rough weather experience of t'he commanding officers'~
Application of Student's t test gives the confidence limits shown as shaded areas
450
[Ch. 22
Not significantly
different
(f)
0c
15
""0
Sea state
for power
reduction
(j)
(j)
c.
(/)
10
8
Sea state
in Fig. 22.3. The high standard deviation and the small number of ships in Class B
widen the confidence limits considerably so that the estimates of mean speed for this
class are much less reliable than those for class A.
Application of the F test described above shows that the standard deviations are
not significantly different and they may be pooled to give a common value using
equation (22.9). The test function tis greater than the critical value (tcrit = 2.03 for N
= 37; see Fig. 22.1) for all but the highest sea states. So the two results are
Sec. 22.5]
451
significantly different at the 95% confidence level and we may be confident that the
performance of class A is better than that of class B at least in moderate sea states.
22.5
The questionnaire shown in Table 22.3 asks for the captain's reasons for reducing
speed. For most conventional ships it is usually found that slamming is the primary
cause of speed reduction with either deck wetness or ship motions given as the
secondary factor. Propeller emergence is usually only important for merchant ships
in ballast. For the frigates described above the order of importance found from the
questionnaire was:
(1) slamming
(2) ship motions
(3) deck wetness.
Fig. 22.4 shows predictions of slamming frequency for ship class A in head seas as
0..
u..
.t=
"'
"'
..Q
25
.J
N
ci
co....
:::J
..r::
c;;
E
"'
(/)
10
12
a function of speed and significant wave height (the most probable modal periods for
the North Atlantic have been assumed). Now the questionnaire gives the maximum
permissible speeds in given wave conditions and the results have been plotted in Fig.
22.4 as a locus of acceptable combinations of speed and wave height. This allows us to
452
[Ch. 22
estimate the maximum tolerable slamming frequency: in this case the captains of ship
class A apparently tolerate about 6080 slams per hour, and }Ve may take this as a
.E0
E
Q)
u>
Q)
:.0
::J
c"'
"'
E
Q)
""0
2l
.<:::
"'
0
Significant wave height (metres)
We see that the captains of these frigates apparently tolerate about 80100 deck
submergences per hour at the forward perpendicular (equation (20.10)) and a
weighted mean subjective motion of about 1112 (equation (21.10)).
However, deck wetness and ship motions are not in this case the limiting factors
which force the captain to reduce speed. So these results cannot be taken as criteria
for acceptable levels. All we can say is that the captain would tolerate more frequent
deck submergence and higher levels of motions if slamming were not so severe.
These results may help in the determination of criteria but they cannot be used in the
definitive way which was possible for slamming. We may therefore guess that the
Sec. 22.5]
453
400
c..
LL
co
~
:::l
0
~
U;
Q)
()
Q)
2l
Q)
..0
:::l
V)
"'
()
Q)
0
Significant wave height (metres)
maximum tolerable deck submergence frequency in this ship is about 120 per hour
and that the maximum tolerable subjective motion is about 15.
Fig. 22.7 shows the speed loss from the questionnaire compared with speed loss
"0
Q)
Q)
c.
(/)
14
Significant wave height (metres)
454
[Ch. 22
curves derived from Figs 22.422.6 for these estimated criteria. As expected, the
slam limited speeds fit the questionnaire data tolerably well, at least in the middle
range of sea states, while the other curves are higher. Note that all three limits are too
high in high sea states. This may arise because the ship and its crew must then tolerate
severe levels of all three phenomena simultaneously and this will be expected to
reduce their tolerance to each individual factor. In other words severe motions and
deck wetness will reduce the crew's tolerance of slamming and limit the maximum
possible speed.
Table 22.4 lists a selection of seakeeping criteria for speed in rough weather
derived by various authors from questionnaires, trials and intuition.
Ship
type
Slamming
Wetness
Kehoe
(1973)
Warship
60/hour
at 0.15L
60/hour
atFP
Ochi and
Motter
(1974)
Merchant
Probability
0.03
Probability
0.07
Shipbuilding
Research
Association
of Japan
(1975)
Merchant
Probability
0.01
Probability
0.02
Lloyd and
Andrew
(1977)
Warship
Lloyd and
Andrew
(1977)
Merchant
Aertssen
(1963, 1966,
1968, 1972)
Merchant
Andrew and
Lloyd
(1981)
Warship
Comstock
Warship
Yamamoto
(1984)
Walden and
Grundmann
(1985)
Propeller
emergence
Vertical
acceleration
Probability
0.1
36/hour
SM = 15
120/hour
Probability
0.03 or 0.04
Probability
0.25
90/hour
SM = 12
20/hour
30/hour
0.2g rms
at bridge
Merchant
Probability
0.02
Probability
0.02 at FP
Probability
of
exceeding
0.4g at
bridge = 0.05
Warship
Probability
0.03
Probability
O.D7
eta/.
(1982)
23
Operational effectiveness
23.1
INTRODUCTION
23.2
[Ch. 23
Operational effectiveness
456
...
Q)
Ol
;::
E
E
....
...
c
Q)
.=::!
...
c
E
E
.=::!
::J
<t:
Ol
Q)
c.
Q)
c.
'IJ
::J
(J)
<t:
(J)
::J
(J)
.10
.10
.10
.05
.03 .03
.03
.20
.20 .20
.10
.20
.20 .20
.10
.04
.04
.04
.07
.04 .04
.04
.07
10
.11
.11
.11
.11
.07
.07
.07
.10
11
.0
.07
.07
.07
.10
.07
.04
.04 .04
.07
.07
.07
.07
.10
.04
.07
::J
(1J
~
(1J
(1J
Q)
(J)
15 .25
.25 .25
16 .44 .44
17
.25
.25 .25
0
23
.07 .07
.07
.07
.04 .04
24
.13 .13
.13
.13
.03
.03 .03
.05
25
.0
.03
.03
.03
.05
.25
.25
.25
.25
.25
.25
.25
.25
Transatlantic
cargo ship
fseason
fseason
Warship
Fig. 23.1 Typical conditional frequency distributions of sea area, !area> for given seasons in
the North Atlantic.
spend in each area along the route. Sea areas are defined according to Hogben,
Dacunha and Olliver (1986) (see Fig. 5.3).
The warship is required to operate throughout the North Atlantic with most of its
time spent in the northern areas 1, 3 and 4. In winter its operational area is moved
south.
457
Sec. 23.4]
Ship course
30
NE
SE
SW
NW
1.0
1.0
0.5
0.5
25
en
0c
=="C
20
15
Q)
Q)
10
c.
(/)
c.
5
.s::
C/)
0
fcourse
Ship course
N
NE
SE
SW
NW
.01
.01
.01
.01
.01
.01
.01
.01
en
.10
.10
.10
.10
.10
.10
.10
.10
.50
.50
.50
.50 .50
.30
c. 10
c.
.07
.s::
5
C/)
.02
0
.125
.30
.30
.30 .30
.30
.30
.30
.07
.07
.07
.07
.07
.07
.07
.02
.02
.02
.02
.02
.02
.02
30
25
0 20
15
=="C
Q)
Q)
(/)
course
(b) Warship
Fig. 23.2 Typical condition frequency distributions of speed, f u, for given ship courses.
The warship is again required to be much more versatile. All courses are equally
likely and a wide range of speeds is demanded. Nevertheless, the economical cruising
speed is frequently used. Very high and very low speeds are rare.
458
Operational effectiveness
[Ch. 23
Criteria
/season /area
fx fTH
/course
fu
(23.1)
where
/season
/area
fx
/course
Sec. 23.4]
fv
459
is the conditional frequency distribution of ship speed for a given ship course
The ship can successfully execute its mission if all the responses are less than the
critical levels for the mission being considered. The proportion of time for which this
is the case (the operational effectiveness) is obtained from the weighted sum of all the
possible values of P. Mathematically this may be written as
(23.2)
where :Eseason :Earea :Ex, :ETH :Ecourse and :E 0 imply summation over all seasons, areas,
wave directions, zero crossing periods and significant wave heights, ship courses and
speeds respectively. r n is a counting functional defined by
for rn < rn,ril
for rn > rn";'
where rn is the nth response of interest (e.g. roll, subjective magnitude etc.) and rn .
'"'
is its critical value.
Table 23.1 shows an example of a calculation of operational effectiveness taken
Unstabilised
Pitch only
Roll only
Pitch and roll
0.976
0.650
0.649
0.976
0.851 '
0.851
from Andrew, Loader and Penn (1984). This shows the operational efffectivness of a
frigate with and without roll stabilisers in the North Atlantic. For this example it is
assumed that the frigate's mission will be impossible if the rms pitch exceeds 2.0
degrees and the rms roll exceeds 3.0 degrees.
Considering first the pitch motions in isolation we see that the ship is able to
achieve its mission almost all the time. Roll stabilisers, of course, have no influence
on this result. Roll motions have a much greater effect on the ship's ability to achieve
its mission and the unstabilised ship is effective for little more than half the time.
These effects are considerably alleviated by the stabilisers.
24
The effect of hull size and form on seakeeping
24.1
INTRODUCTION
The methods outlined in this book allow the designer to quantify and assess the
seakeeping qualities of a new design before the ship is built. If the predicted
performance is inadequate the designer will need to change the size and/or shape of
the hull to effect the necessary improvements.
The designer therefore requires some guidance on the performance improvements which are likely to result from changes to hull form and size. The methods
described earlier may be used to provide information on these trends and this chapter
gives the results of some specimen calculations. These are necessarily specific to a
particular basic hull form and should not be used to give numerical estimates of the
changes in performance of other hull forms. However, the trends described will be
found to be generally applicable and the results may be used to suggest suitable
changes in size and shape to a wide range of hulls. The actual performance of a
particular design should always be estimated from strip theory calculations or model
tests.
We will only consider the effect of hull form and size on motions in the vertical
plane and confine our attention to head waves. This heading generally gives the worst
vertical plane motions and it is found that a form which has low motions in head
waves nearly always has a satisfactory performance at other headings. The most
important motions to be considered in this context are the absolute vertical
accelerations and the relative motions at the bow. The former are an indication of the
severity of the motions experienced by the crew and passengers and we choose to
calculate the acceleration at x81 MILs = 0.15, a typical location for the bridge of a
warship or the passenger accommodation on a ferry. In general it is found that a ship
which has satisfactory motions at this location will also have acceptable motions at
other locations occupied by passengers or crew.
We shall calculate the relative motion at x81 M/L 5 = 0.3, a typical location for a
severe slam, and use this result to estimate the probability of keel emergence at this
station.
Sec. 24.3)
461
Lateral plane motions are also influenced by changes in hull form and size.
However, it is not generally considered worthwhile to optimise hull form to achieve
small motions in the lateral plane. These are more effectively reduced by installing
suitable roll reduction devices such as bilge keels, active fins or passive tanks and, of
course, the rudder.
24.2
There are an infinite number of ways in which the size and shape of a given hull may
be changed and we must choose suitable constraints on the changes we shall
consider. We first need to select a parent hull form to provide a basis for our
calculations. We choose a typical frigate having the following characteristics:
draught/length ratio
DPILP
0.034
Cwfp
= 2 Awtpi(Bp Lp) =
0. 700
We shall examine the effects of changing the size and shape of the frigate hull
form as follows:
(a) changing the size (length) of the hull while keeping the shape constant
(b) Changing the shape of the hull while keeping the length constant.
24.3
The effect of hull size may be studied by calculating the responses of a series of
geometrically similar ships all having the same hull shape ~t differing lengths as
shown in Fig. 24.1. Changing the length ofthe hull while keeping the shape constant
results in proportional changes to all the linear dimensions (~earn, draught, freeboard, etc.) and the displacement varies as the cube of the length. So these hulls may
be regarded as scale models or geosims of the parent.
Transfer functions for heave and pitch of the parent form in head waves are
shown in Fig. 24.2. These are given in nondimensional form as functions of the nondimensional wave frequency roy (L.fg), wave length A./ L, and Froude number. In this
form these transfer functions apply to all the derived hull forms because they all have
the same geometrical shape. All the motion responses are essentially unity in waves
which are much longer than the ship and more or less negligible in waves shorter than
a critical length which is about threequarters of the ship length. In other words ships
tend to contour very long waves but do not respond to very short waves, as already
discussed in Chapter 13.
Transfer functions for a given ship length and speed may be derived from these
results and examples for 20 knots are shown in Figs 24.3 and 24.4. These figures show
the calculation of the rms motions in a typical long crested Bretschneider wave
energy spectrum using the s\}.ort (wave frequency domain) method described in
Chapter 14 (Section 14.4).
Consider first an infinitely long ship. All the waves in the seaway are shorter than
462
[Ch. 24
[=::::>
~L~/7
_c____________
~
L?
t::::::::_ _ _ _ _ _
_~/
c
~~~~/?
Fig. 24.1 Geometrically similar ships.
the critical wave length, the heave and pitch transfer functions are zero over the
entire range of frequencies and the ship does not respond to the wave input at all.
As the ship length is reduced, some of the waves in the seaway begin to exceed the
critical length and the transfer functions adopt appreciable values in the range of
important wave frequencies. So the heave and pitch responses are increased, as
shown in Figs 24.3(c) and 24.4(c).
A very short ship with Ls = 0 metres finds that all the waves in the seaway are
longer than the ship and the transfer functions are unity over the entire range of
frequencies. The ship heave response 'output' Sx,(ro) is then the same as the wave
energy spectrum 'input' S;(ro). So the rms heave is the same as the rms wave
depression. Using equation (4.32)
metres
In the same way the pitch response output for a very small ship is the same as the
wave slope energy spectrum input. In other words a very small ship contours all the
waves in the seaway and suffers large heave and pitch motions.
Fig. 24.5 shows the rms heave and pitch motions obtained from estimating the
area under the response curves shown in Fig. 24.3(c) and 24.4(c). These clearly
demonstrate that small ships suffer from increased absolute motions in a given
seaway.
Fig. 24.6 shows a similar calculation for the vertical acceleration at xB 1M/Ls =
0.15.
Sec. 24.3]
463
1.5
1.0
0
x."'
0.5
6
(IJ
Y(L 5 /g)
1.5
1.0
0
JJ
><
0.5
0
(I)
V(Lsfg)
Fig. 24.2 Nondimensional heave and pitch transfer functions for p~rent hull form in head
waves.
Ss/co) = ( cot:30
Sc,
(co)
(metres/second2_)2f(radian/second)
Fig. 24.3 Effect of shp size on heave motion in head waves at 20 knots;
H 113 = 5.5 m; T0 = 12.4 sec.
have seen that the infinitely long ship does not respond to the waves and it follows
that the relative motion must then be the same as the wave depressiont (see equation
(13.14)). This is confirmed in Fig. 24.7 where we see that the infinitely long ship has a
relative motion transfer function which is unity over the entire range of frequencies.
t Apart from any distortion caused by swellup effects.
Sec. 24.3]
0
1.5
(b)
1.0
0
JJo
"'
)(
0.5
:X:
0.008
Q)
::0"'
f:'
%
f:'
J
:X:
0
Wave frequency'w (radians/second)
Fig. 24.4 Effect of sniP size on pitch motion in head waves at 20 knots;
H 113 = 5.5 m, T0 = 12.4 sec._,.
465
466
[Ch. 24
(i)
(i)
Q)
a;
E_
"'
~
Q)
Q)
..c
>
"'
B
a.
Q)
..c
Ul
Ul
Fig. 24.5 Effect of ship size on rms heave and pitch in head waves at 20 knots;
R 113 = 5.5 m, T0 = 12.4 sec.
So the response is the same as the wave energy spectrum and the rms relative motion
is
0.25 H 113
metres
As the length is reduced, the ship begins to contour the longer waves and the
transfer function encompasses a smaller range of frequencies. However, peaks
appear in the transfer function and the response is amplified if these peaks coincide
with the peak of the wave energy spectrum. For very small ships (Ls = 0 metres)
contouring all the waves the transfer function is everywhere zero and there are no
relative motions.
These trends are summarised in Fig. 24.8.
The probability of keel emergence calculated using equation (20.5) is shown in
Fig. 24.9. For the seaway considered here keel emergence is common for ships in the
range 70200 metres and is most common for ships about 120 metres in length.
Smaller ships are less susceptible because of their small relative motions. These ships
will, however, suffer from high vertical accelerations which would make life on board
intolerable at the speed considered in this example.
The greater draught of the larger ships ensures that keel emergence is unlikely for
Sec. 24.3]
6
467
(a)
uQl
(/)
=o
1:
~J
(b)
We/
/
JJ
(/)
=o
"'
L5 =0 m
0
80
uQl
(c)
uQl
(/)
=o
~
N
'1;
Ql
(/)
E
:::;
3
M
VJ~
2.0
these vessels in spite of their appreciable relative motions. Increasing the size of the
ship also gives a dramatic reduction in vertical acceleration. Large ships are generally
more comfortable than small Qnes in rough weather.
468
c:;
[Ch. 24
L5 =150m
Ql
1/)
=c
~
l
3
~
Om
0
2.0
Wave frequency w (radians/second)
Fig. 24.7 Effect of ship size on relative bow motion at 20 knots; H 113 = 5.5 m, T0 = 12.4 sec.
Sec. 24.4]
469
""0
Cii
~Q>
"
Q>
(/)
.s
(i;
"EQ>
Relative motion
~
1i)
.s
Q>
...,0
""'
0.
(/)
'0
"'Q;
Qi
(/)
"
"'"
(/)
0
0
500
Ship length (metres)
Fig. 24.8 
Effect of ship size O!!_ absolute and relative motions at 20 knots in head waves;
H 113 = 5.5 m, T0 = 12.4 sec.
o.2or~~~.....;..;.~...=;;.....;........,.,
Q)
{.l
Q)
2'
Q)
E
Q)
Q)
.~
:.a
"'
.0
0
r.t
500
Ship l_~ngth (metres)
Fig. 24.9 Effect of shig_ size on keel emergence at 20 knots in head waves;
H 113 = 5.5 m, T0 = 12.4 sec.
470
[Ch. 24
!.....=:=::::::==============zL
_I
L
 c=:::::::;,/'
c::::yL
Relative motion
II
.......
.
Vertical
acceleration
T~
= 3.5, FN = 0.3.
Sec. 24.4]
471
nondimensional form which makes them more easily applicable to ships of different
sizes. They are presented for a Froude number of 0.3 and a nondimensional modal
period defined as
T~ = T0 Y(g!L,) = 3.5
which corresponds approximately to a modal period of 12.4 seconds for a ship length
of 125 metres. Similar trends are found for other modal periods and ship lengths.
Reducing the draught increases the added mass and damping coefficients, as
shown in Fig. 11.10, and this has the generally beneficial effect of reducing both the
absolute and relative motions. Fig. 24.12 shows the corresponding effects on the
0.15
472
[Ch.24
c____________________
_______

0.6
0.4
 0.4
0.6
  Relative
motion
l:f
'o
0
0.2
Absolute 0.2
acceleration
0.1
0.2
tl')
M
l:f
:r"
0
0
0.3
8 5 /Ls
Fig. 24.14 Effect of beam! length ratio on rms motions in head waves;
0.1
0.2
T~ =
3.5, FN = 0.3.
0.3
Summary
Sec. 24.5]
473
L
_______
:>
Fig. 24.16 Forward waterplane area coefficient variations.
DWLI
I
I
I
I
I
I
Fig. 24.17 Effect on section shape of increasing the forward waterplane area coefficient.
the bow and gives favourable changes to the hydrodynamic coefficients in this region,
again at the expense of increased wave excitation. If the displacement is kept
constant it also leads to more favourable section shapes with higher deadrise angles
as shown in Fig. 24.17.
A large forward waterplane is clearly beneficial, reducing both absolute and
relative motions and the probability of keel emergence. The more favourable
deadrise angles will alleviate slamming when it does occur.
24.5
SUMMARY
In summary these results show that a large ship will generally be more comfortable
than a small one. Increasing ship size will almost always result in improved
seakeeping performance.
474
[Ch.24
Fig. 24.18 Effect on forward waterplane area coefficient on motions in head waves;
T,; = 3.5, FN = 0.3.
Fig. 24.19 Effect of forward waterplane area coefficient on probability of keel emergence;
T;, = 3.5, FN = 0.3.
If the ship length is already determined, low levels of vertical acceleration can be
achieved with a shallow draught/wide beam hull form. This may, however, suffer
from frequent keel emergence and slamming. Immunity from slamming can best be
achieved by increasing the draught at the penalty of increased vertical accelerations.
It is for the designer to decide on the best compromise for these conflicting
requirements. A large forward waterplane area coefficient is always beneficial.
Bibliography
The following abbreviations are used in the bibliographical details of the items listed.
AM
ASEM
ATTC
BMT
DMVW
DTMB
HMSO
HSV
!MechE
lOS
ISP
JSNAJ
NACA
NEJ
NSRC
NSRDC
ONR
PRS
QJMAM
SRAJ
SRI
STAR
TRINA
TSNAME
USDC
USNI
UCEP
476
Bibliography
Bibliography
477
Hagiwara, K. & Yuhara, T. (1974) Study of wave impact load on ship bow. Japan
Shipbuilding and Marine Engineering 8 No.4.
Hoerner, S. F. (1965) Fluid Dynamic Drag. Published by the author, 148 Busteed
Drive, Midland Park, New Jersey 07432.
Hogben, N., Dacunha, N. M. C. & Olliver, G. F. (1986) Global Wave Statistics.
BMT.
Hogben, N. & Lumb, F. E. (1967) Ocean Wave Statistics. HMSO.
Hsiung, C. C., Friis, D., Milne, W., Peters, G. R. & Weber, H. W. (1983) New
towing facilities at Memorial University of Newfoundland. 20th ATTC.
Kato, H. (1958) On the frictional resistance to the roll of ships. JSNAJ 102.
Kehoe, J. W. (1973) Destroyer Seakeeping Ours and Theirs. USNI.
Lamb, H. (1932) Hydrodynamics. Cambridge University Press.
Lee, W. T., Bales, S. L. & Sowby, S. E. (1985) Standardized wind and wave
environments for North Pacific Ocean areas. DTNSRDC Report SPD091902.
Lewis, F. M. (1929) Theinertiaofthewatersurroundinga vibrating ship. TSNAME.
Lloyd, A. R. J. M. (1975) Roll stabiliser fins: a design procedure. TRINA 117.
Lloyd, A. R. J. M. (1977) Roll stabiliser fins: inkrference at nonzero frequencies.
TRINA 119.
Lloyd, A. R. J. M. & Andrew, R.N. (1977) Criteria for ship speed in rough weather.
18thATTC.
Lloyd, A. R. J. M., Salsich, J. 0. & Zseleczky, J. J. (1986) The effect of bow shape
on deck wetness in head seas. TRINA 128.
Mack, C. (1966) Essentials of Statistics for Scientists and Technologists. Heinemann
Educational Books, London.
Marine Information and Advisory Service (1982) Catalogue of Wave Data. MIAS,
lOS.
Massey, B. S. (1986) Measures in Science and Engineering, their expression,
relation and interpretation. Ellis Horwood, Chichester!. 11
Nieuwenhuijsen, J. H. (1958) Experimental Investigations on Sea sickness. Drukkerij Schotanus and Jens, Utrecht.
Nordenstrom, N. (1969) Methods for predicting long term distributions of wave
loads and probability of failure for ships, AP11. DNV Report No. 6922S.
Ochi, M. K. (1964) Extreme behaviour of ships in rough seas slamming and
shipping of green water. TSNAME 72.
Ochi, M. K. & Bolton, W. E. (1973) Statistics for prediction of ship performance in a
.
seaway. Parts 13. ISP Nos 222, 224, 229.
Ochi, M. K. & Motter, L. E. (1974) Predictions of extreme ship responses in rough
seas of the North Atlantic. DMVW.
O'Hanlon, J. F. & McCauley, M. E. (1974) Motion sickness incidence as a function
of the frequency and acceleration of vertical sinusoidal motion. AM.
O'Neill, M. E. & Chorlton, F. (1986) Ideal and Incompressible Fluid Dynamics.
Ellis Horwood, Chichester.
Pierson, W. J., Tick, L. J. & Baer, I. (1966) Computer based procedures for
preparing global forecasts and wind field analyses capable of using wave data
obtained by spacecraft. ~th ONR, Washington !)C.
Porter, W. (1960) Pressure distribution, added mass and damping coefficients for
cylinders oscillating in a free surface. UCEP Series 8216.
478
Bibliography
Glossary
Figs G.l and G.2 give a diagrammatic illustration of most of the terms defined.
~
::J
(.)
'6
"'
u:i
u
c.
Q;
c.
;;:"
~ c
ro
Q)
OQ)
u..C.
Q;
c.
~
.... L/2     
L/2
Superstructure
Freeboard
Bow
Draught
Forefoot
Keel
defin_~tions:
480
Glossary
Sinkage
Running
trim
Bow
Draught
Fetch
Forecastle
Forefoot
Forward perpendicular (FP)
Freeboard
Ideal fluid
Length
Sheer line
Sinkage
Stem
Stemhead
Stern
Superstructure
Swell
Glossary
Transom stern
Trim
Wallsided
Waterline
Weather deck
481
Numerical values
1.0 tonnes/metre 3
1.025 tonnes/metre 3
0.001225 tonnes/metre 3
1.14 x 10 6 kN seconds/metre2
1.808 x 10s kN seconds/metre2
9.81 metres/second 2
0.515 metres/second
Index
Added resistance, 398
irregular head waves, 400
regular head waves, 398
wind, 401
Aertssen, 454
Andrew, 11
Andrew, Loader and Penn, 457
Andrew and Lloyd, 277, 282, 454
appendage,231
aspect ratio, 60
Baitis, Woolaver and Beck, 433
Bales, Lee and Voelker, 126
bilge keels, 344
Bittner and Guignard, 425
Bledsoe, Bussemaker and Cummins, 277
Blok and Huisman, 411
BM, 193
bow shape, 424
boundarylayer,58,293
Cartright and LonguetHiggins, 104
chord, 59
Chuang, 416
Chuang and Milne, 416
circular cylinder
heaving in free surface, 196
in uniform stream, 49
Cobra, HMS, 26
coefficient
hydrodynamic, 173
in heave and pitch equations, 174
in lateral plane equations, 177
hydrostatic: 191
lateral plane, 192
vertical plane, 191
in equations of motion, 158
lift, 61
Comstock et al., 454
confidence levels, 442
box ticking, 446
Student's t test, 442
F test, 442
conformal transformation, 52
484
twodimensional, 29, 64
turbulent, 59, 293
uniform stream, 43
force
exciting, 157
Fourier analysis, 97
freeboard
effective, 410
exceedance, 422
free decay, 139
frequency
encounter, 145
model, 291
natural, 137,235,251, 389
of deck submergence, 412, 452
of keel emergence, 412
of propeller emergence, 412
of slamming, 418,451
wave, 73
Friesland class destroyer, 162, 165, 237
FroudeKriloff hypothesis, 170
Froude number, 287,292
Froude, William, 26, 307
fluid
ideal, 28
inviscid, 29
Gerritsma and Beukelman, 170,399
Gilhousen et al., 129
GM, 194
green seas, 421
Grim, 196
GZ, 194
Hagiwara and Yuhara, 416
harmonic response, 133
heading, 144, 292
helicopters, 437
histograms, 327
Hoerner, 63
Hogben and Lumb, 122, 124
Hogben, Dacunha and Olliver, 124,458
hull shape, 468
hull size, 461
incidence, 60
Kato, 230
Kehoe, 454
Lamb, 66
Lee, Bales and Sowby, 127
Lewis forms, 206
heaving, 212
hydrodynamic properties, 211
permissible forms, 208
rolling, 217
swaying, 214
lift, 60
coefficient, 61
curve slope, 61
linear system, 133
Lloyd, 353, 356
Index
Lloyd and Andrew, 454
Lloyd, Salsich and Zseleczky, 423
Loader, II
local hydrodynamic properties: 196
measurement, 218
Mack, 442
mapping function, 53, 55
Marshfield, II
Martin, 345
mass
added, 161, 165, 204, 212,214,217,218
moment of inertia, 155
virtual, 162
Massey, 287
metacentric height, 194
MIAS, 129
model experiments
ballasting, 308
beaches,297,300
bifilar suspension rig, 310
carriage, 297
compound pendulum rig, 309
forced oscillation, 158, 165
irregular waves, 296,318
model materials, 307
model restraint systems, 299
instrumentation, 303
open water, 297
regular waves, 294, 312
scaling Jaws, 286, 289
seakeeping basin, 298
slamming, 414, 417
tank wall interference, 323
trimming, 308
towing tank, 297
wave makers, 300, 319
moment of inertia
added mass, 165
mass, 155, 165, 290, 309
product, 156
virtual, 165
multipole, 48
Nieuwenhuijsen, 425
NOAA, 129
non linearity, 133, 166, 223, 262, 275
Nordenstrom, 122
Ochi, 417
Ochi and Bolton, 104
Ochi and Motter, 454
O'Hanlon and McCauley, 426
O'Neill and Charlton, 28
operational effectiveness, 455
outreach, 59
PAT86 seakeeping computer program, 27
Pierson, Tick and Baer, 126
Porter, 196
potential
complex, 52
Index
velocity, 35
Probability: 327
density function: 329
Gaussian, 333
Normal, 333
Rayleigh, 337
of deck submergence, 411
of keel emergence, 411, 466, 471,472
of occurrence, 410
of propeller emergence, 411
of slamming, 417
joint, 342
propeller
advance coefficient, 404
characteristics, 403
efficiency, 405
racing, 409
self propulsion, 405
thrust coefficient, 404
torque coefficient, 404
questionnaires, 440
radius of gyration, 165,291,310
Reynolds number, 230,287, 293
Ridjanovic, 346
StDenis and Pierson, 26, 263
Salvesen, Tuck and Faltinsen, 170
scatter diagrams, 124
Schmitke, 170, 225,227,230
sea state code, 121
second order, 133
Sellars, 413
Shipbuilding Research Association of Japan, 454
ship motions
absolute, 250
accelerations, 257, 269,295,462,470, 471,472
acclimatisation, 425
axes, 151
beam waves, 243
contouring waves, 462
coupling, 163,169,235
definitions, 151
effect of hull shape, 468
effect of hull size, 461
energy spectrum, 266
following waves, 240
head waves, 235
heave, 152,235,236,244,462
irregular waves, 263, 296
lateral force estimator, 259, 433
matching wave spectrum and transfer function,
271
measurements, 279
motion induced interruptions, 433
motion sickness incidence, 426
oblique waves, 243, 252
phase shift due to wave probe location, 313
pitch, 153, 235, 236
485
notional, 411
resonance, 248
rms values, 268
roll, 153,243,247,248
short crested waves, 272
subjective motion, 429, 452
surge, 152,235
spectral moments, 270
sway, 152,243,243,249
transfer functions, 234, 312
velocities, 257, 269
wave reflection, 314
yaw, 153,243,249
Shoenberger, 429
sink, 44,52
skin friction, 228, 293
slamming, 281,409,413,451
critical velocity, 417
drop tests, 414
frequency, 418
pressure, 418
probability, 417
Smith, 162, 165, 237
source, 44, 52
speed loss
involuntary, 398, 406
voluntary, 437,449
criteria for, 451
springmass system, 132
Stabilisation, 343
active fins, 349
boundary layer losses, 354
constraints on design, 350
control systems: 362
gain margin, 376
overall gain, 370
phase margiil,~B70
sensitivity setti'ngs, 365
speed dependant gain, 363
stability, 372
transfer function, 362
design recommendations, 358
equations of motion for fin stabilised ship, 350
finbilge keel interference, 357
finfin interference, 355
fin servo, 370
forced rolling trial, 366
hydrodynamic losses, 353
overall effectiveness, 358
performance, 377
swayyaw effect, 360
bilge keels, 344
passive tanks, 377
design, 390
dimensions, 392
equations of motion for tank fluid, 380
equations of motion for tank stabilised ship,
386
fluid depth, 393
flume tank, 377
Joss of metacentric stability, 393
mass of working fluid, 395
486
performance, 397
stabilising moment, 393
tank damping, 389, 395
tank natural frequency, 389
U tube tanks, 379
stagnation
point, 51, 57, 58
pressure, 40
stall angle, 61
standard deviation, 97
Stigter, 380
stream function, 41
strip theory, 170
superposition, 64
swellup, 257, 411
system with no stiffness, 141
Takaishi, Matsumoto and Ohmatsu, 126
Tanaka, 228
Tasai, 170, 196
Taylor wake fraction, 404
transformation, 206
transition point, 59, 293
trials, 277
courses, 282
forced rolling, 366
run lengths, 282
Unimak USCG, 413
units, 23
Ursell, 26, 196
variance, 97, 101,269,270
viscosity, 58
coefficient of, 59
Vugts, 165
Wagner, 416
Walden and Grundmann, 454
Walters, 426
waves
amplitude, 64, 94
significant single, 96
bandwidth, 104, 108
breaking, 94
buoys, 129, 279
celerity, 64, 71, 73
characteristics, 73
energy, 86
fully developed, 94
frequency, 73
average, 103
encounter, 145
generation, 93
group velocity, 89
height, 64, 95
characteristic, 107, 108
Index
significant, 96, 105, 339
length, 64, 291
long crested, 113
measurement, 279
number, 70, 292
ocean, 93
Atlantic, 126, 127
atlases, 124
Baltic Sea, 126
Black Sea, 126
fair weather bias, 126
Gulf of Mexico, 126
hindcasting, 126
measured data, 129
Mediterranean Sea, 126
Pacific, 126, 127
North Sea, 126
scatter diagrams, 124
statistics, 121
visual observations, 121, 124
orbits, 74
period, 65, 95
average, 103, 107
mean period of peaks, 104, 108
mean zero crossing period, 104, 108
modal, 108
model, 292
phase velocity, 71
potential function, 65
pressure, 66, 86
reflection, 314
regular, 64
short crested, 114
slope, 65, 72
spectral
moments, 101, 103, 107
ordinate, 100
spectrum, 99
Bretschneider, 107, 112
directional, 114
encountered, 264
idealised, 106
ITTC (two parameter), 107
JONSWAP, 109, 112
slope spectrum, 112
spreading, 113
steepness, 65
synthesis, 98
swell, 94
surface profile, 66
Whicker and Fehlner, 62
whipping, 409
wind, 93, 127,401
WMO, 121
Yamamoto, 454