Mano Behaviour of Ships in Extreme Astern Seas

ARTICLE IN PRESS
Ocean Engineering 33 (2006) 2381–2434

www.elsevier.com/locate/oceaneng
Manoeuvring behaviour of ships in

extreme astern seas
Zafer Ayaza,, Dracos Vassalosa, Kostas J. Spyroub
a
Department of Naval Architecture and Marine Engineering, Universities of Glasgow and Strathclyde,
Henry Dyer Building, 100 Montrose Street, Glasgow, Scotland, G4 OLZ UK
b
School of Naval Architecture and Marine Engineering, National Technical University of Athens,
9 Iroon Polytechneiou, Zographou, Athens 15773, Greece
Received 14 February 2005; accepted 25 October 2005

Available online 9 March 2006
Abstract
In an attempt to contribute to efforts for a robust and effective numerical tool addressing
ship motion in astern seas, this paper presents the development of a coupled non-linear 6-DOF
model with frequency dependent coefficients, incorporating memory effects and random
waves. A new axes system that allows straightforward combination between seakeeping and
manoeuvring, whilst accounting for extreme motions, is proposed. Validation of the numerical
model with the results of benchmark tests commissioned by ITTC’s Specialist Group on
Stability demonstrated qualitative, yet not fully satisfactory agreement between numerical and
experimental results in line with other predictive tools. The numerical results indicate that the
inclusion of frequency coefficients definitely affects the accuracy of the predictions. In order to
enhance further the numerical model and obtain useful information on motion coupling,
extensive captive and free running model tests were carried out. Good agreement with the
experimental results was achieved. These studies provide convincing evidence of the capability
of the developed numerical model to predict the dangerous conditions that a ship could
encounter in extreme astern seas. As a result, it could offer new insights towards establishing
relationships linking ship behaviour to design, environmental and operational parameters.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Following/quartering seas; Extreme ship motions; Manoeuvring; Stability; Broaching; Surf-
riding; Parametric rolling
Corresponding author. Fax: +44 141 5484784.

E-mail address: zafer.ayaz@na-me.ac.uk (Z. Ayaz).
0029-8018/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.oceaneng.2005.10.023
ARTICLE IN PRESS
2382 Z. Ayaz et al. / Ocean Engineering 33 (2006) 2381–2434
Nomenclature
a amplitude of wave (m)

aH interaction factor between hull and rudder
ai, component wave amplitude (m)
B beam (m)
c phase velocity of wave (m/s)
Cb block coefficient
C j_ damping force (Nt)
df draught at fore (m)
d mean draught (m)
da draught at aft
D depth (m)
DP propeller diameter (m)
FN rudder normal forces (Nt)
Fn nominal Froude number
g gravitational acceleration (m/s2)
GM metacentric height (m)
H wave height (m)
Hs significant wave height (m)
Ixx roll moment of inertia (kg m2)
Iyy pitch moment of inertia (kg m2)
Izz yaw moment of inertia (kg m2)
Jp advance coefficient
k wave number
ki component wave number
K0 roll external moment (Nt m)
KG vertical position of centre of gravity from keel line (m)
KP proportional gain (s)
KR differential gain (s)
KT thrust coefficient
LBP length between perpendiculars (m)
LCG longitudinal position of centre of gravity from the amidships (m)
m ship mass (kg)
M0 pitch external moment (Nt m)
n propeller rate of rotation (rpm)
N0 yaw external moment (Nt m)
Nj normal Vector
p pressure (N/m2)
ps static wave pressure (N/m2)
pd dynamic wave pressure (N/m2)
P angular roll velocity (deg/s)
Q angular pitch velocity (deg/s)
R angular yaw velocity (deg/s)
ARTICLE IN PRESS
Z. Ayaz et al. / Ocean Engineering 33 (2006) 2381–2434 2383
RT(u) total resistance force (Nt)

S propeller slip ratio
Tp thrust deduction in forward motion and during a turn at the propeller
tr thrust deduction in forward motion and during a turn at the rudder
position
tD time constant (s)
Tj roll period (s)
Tz modal period (s)
U surge velocity (m/s)
Urw effective wind speed (m/s)
UR rudder inflow velocity (m/s), and angle
V sway velocity (m/s)
VR mean rudder inflow velocity (m/s)
Wp wake fraction in forward motion and during a turn at the propeller
wr wake fraction in forward motion and during a turn at the rudder
position
W heave velocity (m/s)
X0 surge external force (Nt)
xH longitudinal position of the point of action of the to hull interaction
force (m)
xR longitudinal position of the rudder’s centre of pressure (m)
Y0 sway external force (Nt)
Z0 heave external force (Nt)
zR vertical position of the rudder’s centre of pressure (m)
zy vertical coordinate of the centre of action of lateral force (m)
Greek symbols
w heading angle (deg)

wc autopilot course from the wave direction (deg)
d rudder Angle (deg)
dR actual rudder angle (deg)
f heel angle (deg)
l wave length (m)
y pitch angle (deg)
r density (kg/m3 )
si random phase angle (deg)
t Hanaoka parameter
o wave frequency (rad/s)
oe wave encounter frequency (rad/s)
owi component circular frequency (rad/s)
oG angular velocity (deg/s)
x position of ship on the wave (m)
c autopilot course (deg)
cR the desired heading angle (deg)
ARTICLE IN PRESS
z vertical position of ship (m)

z(t) irregular wave elevation (m)
za wave amplitude (m)
D displacement (kg)
FI potential associated with the incoming wave potential
FD potential of disturbed wave
Hydrodynamic coefficients
X u_ surge acceleration coefficient

Xvr surge velocity coefficient
Y v_ Y r_ sway acceleration coefficients
Yv, Yr linear sway velocity coefficients
Y vjvj ; Y rjrj higher order sway velocity coefficients
Zq_ ; Z w_ heave acceleration coefficients
Zq, Zw heave velocity coefficients
K p_ roll acceleration coefficient
M q_ ; M w_ pitch acceleration coefficients
Mq, Mw pitch velocity coefficients
Nv, Nr linear yaw velocity coefficients
N rjrj ; N vjvjr ; N rjrjv Higher order yaw velocity coefficients
1. Introduction
Ship motions in extreme astern seas have been of great scientific interest for the
last 40 years or so. The highly non-linear and unpredictable nature of the
phenomena has motivated several studies with different theoretical and experimental
approaches. Yet, the number of studies concerning coupled ship motion stability
could still be seen as limited compared to the vast number of studies in the other two
important areas of hydrodynamics to which these motions are assumed to be related:
seakeeping and manoeuvring. However, due to the unique character of phenomena
that involve two areas of hydrodynamics, study of these motions could be a catalyst
for achieving the long overdue marriage of seakeping and manoeuvring. Moreover,
the strong requirement from the maritime community and academia for establishing
adequate and robust stability and safety rules covering ship motions in extreme
astern seas has motivated further the development of reliable numerical codes for
performance-based assessment of ship stability in extreme seas. Notably, the stability
criteria issued by IMO have limited relevance to the as following sea environment
where a ship is most vulnerable.
The need for more theoretical and experimental investigations to study these
motions were emphasised by both the manoeuvring Committee (1999) and the
Specialist Committee on Stability (1999) of the International Towing Tank
Conference in 1999 (ITTC, 1999a,b). Based on this background, there has been an
ARTICLE IN PRESS
upsurge in efforts to establish reliable numerical models to simulate manoeuvring of

ships in extreme astern seas and to identify dangerous situations or other interesting
phenomena occurring during those motions.
The nature of dangerous conditions occurring in extreme astern seas has been
evaluated through extensive experimental and theoretical studies. In broad terms,
combination of direct steep waves and energy concentration in a narrow range of
following and quartering waves could lead to dangerous conditions which may
ultimately lead to capsizing: The semi-static mode so-called ‘‘pure loss of stability on
a wave crest’’ occurs when there is not enough buoyancy to keep the vessel upright
and capsize happens suddenly and abruptly. The dynamic mode known as low-cycle
resonance (parametric resonance) is caused by a Mathieu-type built-up of large roll
motion. This large roll motion arises when a number of factors are combined, such
as: the roll natural frequency is multiple of half the encounter frequency, roll
damping is low and wave length is comparable to ship length. The third mode
involves hydrodynamic lift forces and directional instability of the ship and is known
as broaching. Broaching is usually preceded by the so-called ‘‘surf-riding’’ behaviour
where the ship appears to be carried along by a single wave. In this unsafe condition
the yaw restoring force provided by the rudder could be inadequate to impede the
increasing oscillatory yawing motion and this leads to ‘‘loss of heading’’ or
‘‘broaching-to’’, which in turn could result in capsizing. In astern seas, coupling
between surge and lateral motions cannot be avoided therefore it is essential to take
both surf riding and broaching into account. However, broaching could also occur at
lower speeds with oscillatory build-up of yaw motion but the required waves are
steeper than in the previous broaching condition. Whilst this second mode might
seem to relate to horizontal plane motions only, capsize may also occur due to strong
coupling between roll and sway–yaw motions. Spyrou (1995,1996) has described in
detail the interesting dynamics of the various type of broaching. A fourth mode is
usually observed for high-speed craft having small bow height. It is referred-to as
‘‘bow-diving’’: with increasing nominal speed, in conditions similar to those leading
to surf riding, the bow submerges into the preceding wave slope because there the
ship achieves a condition of static equilibrium in the surge direction. Matsuda et al.
(2003) confirmed these phenomena with experimental results.
As mentioned earlier, the interesting dynamics of these phenomena has attracted
many researchers and a great amount of experience and knowledge have been gained
through experimental and theoretical studies. By being the most dynamic mode
amongst dangerous conditions, broaching-to has gained particular attention. Some
well-known early studies such as Davidson (1948), Grim (1963) investigated
directional stability and steering behaviour in following seas and developed non-
linear mathematical models. Meanwhile, Du Cane and Goodrich (1962) and Conolly
(1972) extensively investigated broaching and related aspects in a quantitative way.
Since the pioneering study of Davidson (1948), manoeuvring-type models are used in
order to study the motions of ships in astern seas. This is due to the fact that
hydrodynamic forces acting on a ship which travels in astern seas consist mainly of
lift components, wave-making components being negligibly small. In the past, a
number of researchers such as Abkowitz (1969) stated the differences on approaches
ARTICLE IN PRESS
between manoeuvring and seakeeping which are reflected in the evolution of two
different types of mathematical models. Manoeuvring usually deals with steady state
and transient (or zero and low frequency) motions in calm water, while seakeeping
deals only with higher frequency wave-induced motions. However, those manoeuvr-
ing models, which are based on calm water conditions, cannot describe dynamics
encountered in a wave environment. As a result, time domain models used to
describe motions in astern seas and capsizing take into account manoeuvring forces
and wave forces and they follow a modular approach.
Motora et al. (1982) carried out numerical simulations of surge, sway, yaw and
roll motions of a ship travelling in following waves by making use of the results of
captive model tests. Based on those investigations, it was concluded that ‘‘wave
exciting yaw moment which exceeds the course-keeping ability of the rudder results
in broaching-to phenomena’’. This finding was in contrast to previous opinions that
the extreme yaw motion comes from hydrodynamic moments as a result of the
unsymmetrical underwater volume following large heeling.
Following the aforementioned findings, Hamamoto and Akiyoshi (1988),
Hamamoto and Shirai (1989), Hamamoto and Kim (1992), and Hamamcoto and
Saito (1992) developed a six degrees of freedom mathematical model. The
calculation of incident wave forces was based on analytical approximations and
diffraction forces were not included. A quasi-steady theory under the assumption of
low encounter frequency was incorporated and hydrodynamic forces were evaluated
using practical formulae. The influence of wave height and length, relative position
of ship to wave, heading angle and loading condition (GM) were examined.
Capsizing due to parametric excitation and pure loss of stability were simulated.
Similarly, De Kat and Paulling (1989) presented a numerical model for simulating
large amplitude motions in severe waves and capsizing. Linear potential theory
was used for determining the wave induced pressures and the body radiation
and diffraction forces. Froude–Krylov forces were calculated up to instantaneous
free surface. Manoeuvring derivatives associated with viscous flow affecting the
sway force and yaw moment were incorporated. Low cycle resonance, loss of
static stability and broaching phenomenon were realized, the latter not by one wave
action but when subsequent steep quartering waves were striking the vessel.
Hamamoto et al. (1994) also used a similar mathematical approach to study
combined motions of sway, roll and a yaw in following seas. The mathematical
model was linearized and numerical simulations were carried out to show
unstable behaviour in limiting situations at zero frequency of encounter and in
overtaking waves of low frequency of encounter. In a similar way, Umeda and
Renilson (1992a,b) introduced a practical method for calculating the wave forces
on a ship running in quartering seas. It was claimed to be a limiting case of a
strip theory method where the effect of encounter frequency was ignored since it is
very low for this case. The experimental and numerical studies revealed
that Froude–Krylov forces alone cannot predict wave forces with sufficient
accuracy and by utilizing the Ohkusu (1986)’s study, it was found that the method
could be useful if broaching only is investigated for a transient period. Vassalos and
Maimun (1994), giving a brief review of individual mechanisms likely to contribute
ARTICLE IN PRESS
to broaching, confirmed, using numerical and experimental investigations, that

coupling between longitudinal and lateral motions are of paramount importance for
broaching as well as the ensuing extreme vessel behaviour. Bailey et al. (1998)
developed a unified mathematical model to study dynamical behaviour and
manoeuvring of a ship travelling in seaway, encapsulating theories of seakeeping
and manoeuvring. The relations between fluid actions defined in manoeuvring
analyses using a body fixed frame of reference and those for seakeeping analysis
using an equilibrium frame of reference were presented. Munif and Umeda (2000)
developed a fully non-linear six degrees of freedom numerical model in which
large angles were taken into account. It was concluded after extensive numerical
simulations that capsizing due to parametric resonance occurs only for
lower metacentric height and lower speed with the ship running with encounter
angles of 30–451 is more dangerous for capsizing than when the encounter angle is
between 0 and 151.
In the calculation of ship motions in astern seas since the frequency of encounter is
quite low, it is quite common to use ‘‘zero-frequency’’ constant hydrodynamic
coefficients as in calm-water manoeuvring calculations to use ‘‘zero-frequency’’
constant hydrodynamic coefficients. However, some studies such as Ankudinov
(1983) argued that wave effects associated with unsteady motion of the hull at the
free surface and vortices which are shed from the oscillating hull indicate that the
convolution terms (‘‘memory effects’’) do have considerable effects. In Ankudinov’s
(1983) 6 DOF non-linear model, heave and pitch were decoupled from surge as well
as from the lateral plane motions. The mathematical model was applied to
deterministic calculation of seakeeping and manoeuvring prediction in irregular seas.
McCreight (1986), using state space approximation, studied manoeuvring in waves
although the effect of frequency on large amplitude ship motions in astern seas was
not a focus of the study. Following similar approaches, Rhee et al. (1990), Lee (2000)
investigated the effect of frequency for simple manoeuvring motions in waves.
Hamamoto and Saito (1992) presented a practical method for the time-domain
description of ship motions in following waves. In the calculation of heave and pitch
motions they compared hydrodynamic forces described a fixed frequency with
convolution methods. However, the model is not exact even for the small amplitude
motion problem since the frequency of motion cannot be determined when non-
linear equations are involved in the equation.
Since the development of numerical tools in astern seas focused on motions in
regular waves, irregular seas studies mostly consisted of experimental results.
Takaishi (1982) showed an example of encountered waves measured in following
seas and stated that the ship travelling in irregular seas could be successively attacked
by large waves long enough time to undergo large rolling motion. De Kat and
Paulling (1989) found that in the following and quartering sea conditions, the wave
elevation observed amidships appears to be regular over certain periods of time in
spite of the randomness of the sea. Therefore, they suggested characterizing the
random wave process by superposition of only two regular wave systems which
would yield an equivalent wave system. Hamamoto et al. (1996) further evaluated
Takaishi’s s work incorporating GM fluctuation. It was found that the encountered
ARTICLE IN PRESS
wave spectrum band is very narrow at a frequency range that could coincide with the
ship roll natural frequency ensuing with GM fluctuations was considered.
In the light of the results of the benchmark studies commissioned in its first term
(1999), the ITTC Specialist Committee on Prediction of Extreme Ship Motions and
Capsizing (2002) has targeted in its second term the elements affecting the accuracy
of the numerical models. Amongst these, degrees of freedom and frequency effect or
so called ‘‘memory effect’’ were seen to be the most important. Concerning wave
irregularity, the report mentioned that the applicability of numerical models to
realistic seaways, that is, short-crested seaways, should be examined. It referred to
experimental results indicating that capsizing danger is least in short-crested
irregular waves, followed by long-crested and finally regular waves.
State-of-the-art time domain numerical codes now employ 6 DOF non-linear
numerical models incorporating the memory effects in regular and irregular waves
directly or by using hydrodynamic coefficients provided by experiments. Coupling
effects between modes of motion, despite not being yet completely understood, are
taken into account. Especially, the effect of vertical motions is not yet fully
investigated. Therefore, in the light of all these drawbacks, it is not surprising that
numerical models are still not fully reliable for drawing stability and safety
guidelines. However, they can prove very useful when combined with the model
experiments. Due to the practical limitations experimental studies have been used
less for irregular waves.
Deriving from this background, a research study has been undertaken to develop a
coupled non-linear 6-DOF model with frequency dependent coefficients, incorpor-
ating memory effects in random waves with a new axis system that allows
straightforward combination between seakeeping and manoeuvring models whilst
accounting for extreme motions. This paper presents details of this mathematical/
numerical model and its validation using experimental results from the ITTC
Benchmark tests. This is followed by analysis of the captive model tests carried out
to investigate the effect of encounter frequency and coupling effects as well as free
running model tests in irregular waves to investigate those effects in more realistic
environmental conditions. The findings of these studies are presented and discussed,
based on which conclusions are drawn and recommendations made.
2. Mathematical model
The numerical model incorporates non-linear six-degrees-of-freedom coupled

motion equations in the time-domain, with no restrictions on motion amplitude. In
order to provide a meaningful combination of seakeeping and manoeuvring, the
equations of motions are expressed in terms of a horizontal body axes system. The
horizontal body axes, which are closely related to, but not a special case of general
body axes, is quite a common system which has been used in many other studies of
ship manoeuvring that include roll (Chislett, 1990), in studies of manoeuvring
motion of ships in waves especially for studying capsizing, (Hamamoto and Kim,
ARTICLE IN PRESS
1993), and for studies of dynamic stability of ships in following and quartering seas,
(Hamamoto et al., 1996).
In deriving the basic equations of motion, three different coordinate systems are
used as shown in Fig. 1. The first is an earth fixed system, defined by 0-xZz The
second is a general body axes which is fixed in the ship with the origin G being
located at the centre of gravity of the ship defined by G-xyz. The third is the
horizontal body axes fixed in the ship with the origin at G and defined by G-x0 y0 z0 .
Notice that in contrast with conventional methods, no assumptions are made of a
small pitch angle and the motions are solved for large amplitudes. Thus results are
obtained from the non-linear equations. In light of the above facts, from Newton’s
second law of motions, equations of motion are obtained. Here, considering force
and moment components, the force and moment can individually be divided into
Fig. 1. Existing axes systems and horizontal body axis system (Hamamoto and Kim, 1992).
ARTICLE IN PRESS
components:

m V_ G þ oG V G ¼ XF ;
(1)
H_ G þ oG H G ¼ XM ;
where m is the mass of a ship, HG the angular momentum about the centre of gravity,
o the angular velocity, VG the linear velocity, XF the external force vector and XM
the moment vector. The first term of right-hand side of Eq. (1) is the reactive forces
acting on the hull and it will be divided into two parts: hydrostatic and
hydrodynamic forces. In order to describe the situation of the ship in the earth
fixed axes, it is normal to use a transformation of Eq. (1) in terms of Eulerian angles
f,y,c which are defined as the rotations about the body fixed axes (Fig. 1). Here, Eq.
(2) instead of Eq. (1), in which the only one rotation about the absolutely vertical
axis is considered, is used and this is called here as the Horizontal body axes, G-x0 y0 z0
in Fig. 1
ZZ

_ VG ¼
m V_ G þ kc pn ds;
s
ZZ
_ HG ¼
H_ G þ kc pðr nÞ ds; ð2Þ
s
V g ¼ iU þ jV þ kW H G ¼ iH x0 þ jH y0 þ kH z0 , (3)
where U, V and W are the linear velocities, Hx’, Hy’ and Hz’ the components of
moment vector with respect to the Horizontal body axes. The transformation
between body axes and horizontal body axes in terms of the Eulerian angles is as
follows:
2 03 2 32 3
x cos y sin j sin y cos j sin y x
6 07 6 0 cos y sin j 54 y 7
7 6
4y 5 ¼ 4 5. (4)
z0 sin y sin j cos y cos j cos y z
The angular velocity is described for the horizontal body axis system as
_ þ kC.
_ þ jY
o ¼ iF _ (5)
The transformation between the angular velocity of the horizontal system and the
body axis system is shown as
2 3 2 32 3
_
F cos y 0 0 j_
6 _ 7 6 76 7
4Y5 ¼ 4 0 1 0 54 y_ 5. (6)
C_ sin y 0 1 c_
ARTICLE IN PRESS
The components of the moment vector in terms of the angular velocity and moments
of inertia can be defined as
2 3 2 32 3
H x0 I x0 x0 I x0 y0 I x0 z0 _
F
6 H 0 7 6 I 0 0 I y0 y0 76 _ 7
I y0 z0 54 Y
4 y 5 ¼ 4 yx 5. (7)
H z0 I z0 x0 I z0 y0 I z0 z0 _
C
In (7), prime (0 ) indicates moments of inertia for the horizontal body axis system.
In order to derive the equations of motion some approximations are made. First,
because of symmetry and since the origin is located at the centre of gravity, it is
assumed that I yz ¼ 0, I xy ¼ 0, I xz ¼ 0 and in the horizontal system IyyffiIzz.
Furthermore, using Eqs. (4) and (7), the moments of inertia and product moments of
inertia are obtained as

I x0 x0 ¼ I xx cos2 y þ I zz cos2 f þ I yy sin2 f sin2 y ;
I y0 y0 ¼ I yy cos2 f þ I zz sin2 f ¼ I yy ¼ I zz ; (8)

I z0 z0 ¼ I xx sin2 y þ I zz cos2 f þ I yy sin2 f cos2 y ;
I x0 y0 ¼ 0;
I y0 z0 ¼ 0; (9)
2 2
I z0 x0 ¼ ðI xx I zz cos f I yy sin fÞ sin y cos y:
Considering force and moment components, the force and moment will
individually be divided into components. Further expansion of Newton’s equations
in Eq. (2) will yield final form of modified equation of motions in Horizontal Body
_ and R for C
_ Q for Y
Axis system. Here, substituting P for F, _ the moment equations
will be
mðU_ VRÞ ¼ X 0 ;
mðV_ þ URÞ ¼ Y 0 ; (10)
_ ¼ Z 0 þ mg;
mW

ðI yy I xx Þ sin 2 y QP þ 12R_ þ cos 2 y QR
þ ðI xx cos2 y þ I yy sin2 yÞ P_ I yy RQ ¼ K 0 ,

ðI yy I xx Þ sin 2 y 12R2 þ ðI xx cos2 y þ sin2 yÞRP þ I yy Q_ ¼ M 0 ,

ðI xx I zz Þ sin 2 y QR 1P_ cos 2yQP þ ðI xx sin2 y þ I zz cos2 yÞR_ ¼ N 0 ,ð11Þ
2
where X0 , Y0 , Z0 , K0 , M0 , N0 are surge, sway, heave, roll, pitch, yaw external forces and
moments. U, V, R are surge, sway, heave linear velocities, Q, P, R are roll, pitch, yaw
angular velocities in horizontal body axes system and Ixx, Iyy, Izz are roll, pitch, yaw
moments of inertias, respectively. It can be written in terms of variable as follows:
X : State Vector x 2 jR00 ;
(12)
X : ðza ; xG ; x0 ; y0 ; z0 ; U; V ; W ; P; Q; R; f; y; c; dÞT ;
ARTICLE IN PRESS
where x0 , y0 , z0 are kinematics, d denotes rudder angle xG and za represent horizontal

and vertical component of wave amplitude, respectively. In a more common way,
from general seakeeping models, equations of motions in Eqs. (10) and (11) can be
written as
ðM þ AÞX€ þ B ðX ÞX_ þ C ðX ÞX ¼ F ðzw ; X ; X_ ; X€ Þ, (13)
where M is inertia matrix, A is added inertia matrix, B is damping coefficient matrix,
C is restoring coefficient matrix, F is external force vector and zw is wave amplitude.
2.1. External forces
When the ship travels in following and quartering seas, several external factors
have an influence on the dynamic behaviour. However in this study, wave forces,
manoeuvring (hull) forces, rudder and propeller and wind forces will be taken into
account in the estimation of the external forces, since they represent the most
important components of the excitation. External forces are estimated using
conventional modular approach. The aforementioned external forces in modular
form can be given as follows:
X 0 ¼ X W þ X H þ X RD þ X P ;
Y 0 ¼ Y W þ Y H þ Y RD þ Y P ;
Z0 ¼ ZW þ ZH ;
(14)
K 0 ¼ K W þ K H þ K RD þ K P ;
M0 ¼ MW þ MH ;
N 0 ¼ N W þ N H þ N RD þ N P :
Here, W indicates wave forces and moments, H indicates hull (manoeuvring) forces
including the resistance force in surge motion and moments and radiation forces and
moments for vertical motions, RD indicates rudder forces and moments and P
indicates propeller forces and moments, respectively.
2.1.1. Wave forces

As well as being an important factor in identifying the safety of a vessel and
sometimes a major cause in dangerous situations and capsizing, waves can have a
significant effect on coursekeeping and manoeuvring of a ship in following and
quartering seas. A ship, attempting a steady course in rough seas, experiences wave-
induced oscillatory motions in all six degrees of freedom. According to classical
linear theory fluid is assumed to be inviscid and incompressible and the flow is
irrotational. Surface tension is neglected and infinite water depth is assumed. From
the classical seakeeping theory, wave forces will be considered in two parts; forces
induced by incident waves or so-called ‘‘Froude-Krylov’’ forces and forces induced
by diffracted waves.
2.1.1.1. Froude– Krylov including hydrostatic forces on a ship. Both experimental

and numerical studies confirmed that Froude–Krylov forces acting on surge
ARTICLE IN PRESS
direction in following and quartering seas can be significant, especially, for

investigating dangerous phenomenon such as surf-riding. Furthermore, the
introduction of the wave profile into the simulation of such motion was found to
be crucial. It is known very well that when motions in extreme following
and quartering seas are concerned, there is significant change in restoring moment
of vessel due to the wave, also the occurrence of large trim and sinkage
during the extreme seas condition would require certain coupling of vertical and
horizontal motions and this is done in the form of instantaneous wave
surface modelling. For instance, De Kat and Paulling (1989) pointed out that
linear wave theory applied to large amplitude waves in deep water has been found to
yield very adequate results as regards wave profile, water particle kinematics and
pressure and in case of the aforementioned important surge forces obtained from
integration of pressure gradient over the wetted volume. In steep waves however,
pressure gradient may be reduced below the one predicted by Froude–Krylov and
this generally leads to lower stability in real motions. Hence, the integration of
pressure over instantaneous wave surface is a very common method in the
investigation of ship motions in extreme astern seas conditions. Therefore, this is
followed in the present study as well.
The generalized Froude–Krylov force vector is given by integration of pressure in
the undisturbed wave system up to the instantaneous wetted surface
ZZ ZZ ZZ
pn dS ¼ pðH:S:Þn dS þ pðH:D:Þn dS: (15)
S S S
The former is static, the latter is dynamic wave pressure, and may be written as
ps ¼ rgB;
(16)
pd ¼ r qF
qt ;
I
where r is density and FI is the potential associated with the incoming wave
potential. Froude–Krylov forces and moment are written as
ZZ
F F :K ¼ pn dS
ZZS
M F :K ¼ pðr nÞ dS; ð17Þ
S
where n is normal vector and r n is vector fixed with respect to centre of gravity.
In order to evaluate the Froude-Krylov forces two different methods are used. First,
pressure is integrated using 3D panel method and the second is Gauss theorem,
(Hamamoto and Kim, 1992) while hull is modelled in 2D. The idea behind the first
method is to form the ship hull by panels and calculate the pressure on those panels.
The method first introduced to marine studies by Hess and Smith (1964), has gained
great importance in recent years thanks to high-speed computers. The hull is modelled
by plane quadrilateral surface elements. Using quadrilateral or triangular elements
cannot make a significant difference however Hess and Smith (1964) pointed out that
ARTICLE IN PRESS
the use of triangular elements eliminates the so-called ‘‘edge-effects’’ and the choice of
point where velocity and pressure are to be evaluated is not obvious for triangular
element. However, as it is mentioned earlier, one important requirement of the panel
method is to use geometrically equally distributed panels. Sometimes it can be difficult
to obtain such a hull mesh and for very curved forms, like fishing vessels, inaccuracy of
the results obtained from panels may rise due to unequal or sometimes overlapping
panels of the hull mesh. Therefore, Froude–Krylov forces calculations are carried out
using 2D Gauss theorem following the method used in Hamamoto and Kim (1992).
2.1.1.2. Diffraction forces. When a ship is running in the following waves, she
encounters the wave crests with very long period. Traditionally, the strip theory with
forward speed has been used for calculation of wave forces and loads in these waves.
However, due to now well-established fact of low encounter frequency in this
situation, strip theory could be inadequate to calculate those motions. Based on the
above background, this study follows the method proposed by Ohkusu (1986). In this
method effort is focussed on taking into account the effect of waves resulting from the
disturbance of the incident waves by the ship. Those waves are supposed to be of
higher order than the incident waves and their effect is naturally of higher order than
the Froude–Krylov forces. The terms include the effect between the disturbed incident
waves and the stationary waves generated when the ship runs on otherwise calm water.
In the light of the above statements using the dynamic pressure on the hull, general
wave force equations are obtained. For the purpose of the study, only the
disturbance equation is used which can be written as follows (Ohkusu, 1986):
Z
F DIF ¼ rU FD N j ds: (18)
GX
Here, j ¼ 1,2,3 denotes sway, heave, roll respectively FD indicates the disturbance
of waves, and Nj is the normal vector. For the pitch and yaw, values of the heave and
sway at each cross-section are multiplied by the distance between the cross section
and centre of gravity of the ship. However, as Umeda et al. (1995) indicate again, this
approximation is straightforward for a ship with transom stern while it is not for a
ship with other stern type. It is due to the fact that if the aft end has no breath and
draught, the strength of trailing vortices will be zero in the Ohkusu’s theory even
though in reality they exist. The strength of trailing vortices could be calculated
easily for a ship with transom stern because the aft end term will be too small and
therefore, Ohkusu’s assumption will be almost adequate. However, if a substantial
area exists behind the aft perpendicular, Ohkusu’s theory will ignore the trailing
vortices occurring in that part. In this work, the aft end was also geometrically
modelled. However, it is not a proper solution to the real problem of shedding of
trailing vortices from a hull surface.
2.2. Manoeuvring (hull) forces
As it is given previously, manoeuvring and seakeeping usually deal with differently

motivated situations in terms of steadiness and frequency, however when ship
ARTICLE IN PRESS
motions in astern seas are concerned, the hydrodynamic matters arise in this unique
phenomena and directional stability or more generally course-keeping problems
require combined approach of manoeuvring and seakeeping. A manoeuvring
mathematical model focusing on lift components can be used in this situation.
Having stated those facts, the main question here is what type of manoeuvring
mathematical model that employs hydrodynamic reaction forces can be used in the
numerical model? In the context of manoeuvring hydrodynamic reaction forces
consist of ideal fluid forces associated with potential flow theory, first-order viscous
forces in the presence of hydrodynamic lift forces and second-order or higher viscous
flow forces associated with cross-flow and separation effects (Spyrou, 1990). In terms
of the accuracy of a manoeuvring mathematical model, the most crucial element is
the accuracy of the hydrodynamic reaction forces.
Current numerical model follows the semi-empirical methodology where it is
assumed that the fluid forces are uniquely determined at any instant and independent
of any other details of the motion except for the geometrical properties of the ship
and physical properties of the fluid. The fluid forces can be expanded as Taylor series
in powers of displacements, velocity and accelerations and this leads directly to a set
of linear and non-linear terms in the equation of motion. The force coefficients are
determined either through analytical or experimental technique. Therefore, for the
calculation of the hull or manoeuvring forces, equations based on the MMG
(Japanese manoeuvring Group) model (1981) and Tasai (1961) formulations were
used. Surge, sway, roll and yaw, manoeuvring coefficients were obtained from the
model experiments in Japan for the vessels used in numerical tests in Sections 4 and
5. The heave and pitch radiation coefficients were found using Tasai’s (1961)
empirical formulae
u
X H ¼ X u_ u_ Y v_ vr Y r_ rr þ X vr vr RT ðuÞ,
juj
u
Y H ¼ Y v_ v_ þ Y r_ r_ þ Y v v þ Y r r þ Y vjvj vjvj þ Y rjrj rjrj,
juj
ZH ¼ Z w_ w_ þ Zw w þ Z q_ q_ þ Z q q,
K H ¼ K p_ f€ þ CðfÞ _ zy Y H ,
M H ¼ M q_ q_ þ M q q þ M w_ w_ þ M w w,
u
N H ¼ N r_ r_ þ N v_ v_ þ N r r þ N v v þ N rjrj rjrj þ N rjrjv rrv þ N vjvjr vvr, ð19Þ
j uj
where XH, YH, ZH, KH, NH, MH are surge, sway, heave, roll, pitch, yaw hull forces,
respectively, RT (u) is total resistance force, C ðj_ Þ is damping moment, zy is vertical
coordinate of the centre of action of lateral force. Others represent the acceleration
and velocity coefficients. As it can be seen above formulations include acceleration
terms, first-second- and third- order linear and non-linear velocity terms. The added
mass and moment terms are taken into left hand side of the equation in Eqs. (10) and
(11) for the solution matrix of Eq. (13). Here, acceleration and velocity terms are
estimated independent of frequency, however when frequency dependent added mass
and damping terms are incorporated to the motions with convolution integrals, the
ARTICLE IN PRESS
left-hand side of the Eqs. (10) and (11) will be linearized. This will be explained
in the next chapter. Furthermore, Eq. (19) does not include coupling between
vertical and horizontal motions. However hydrodynamic terms which result from
combined sinkage and rotation occur during heeling and are added to sway force
and yaw moment in Eq. (19) if experimental values are available. These terms are
represented in the first-order on the basis of linear sway and yaw velocity coefficients
as follows:

Y H ¼ Y j j þ Y vjjj vj þ Y rjjj rj;
(20)
N H ¼ N j j þ N vjjj vj þ N rjjj rj:
It should be noted that this model is only valid for small heeling angles and it does
not include the effect on the acceleration and higher order velocity coefficients. This
could seem to be contradicting with the purpose of this study which aims to develop
a numerical tool for the simulation of extreme motions. However, since wave effects
especially restoring terms will play dominant factors in such cases, the assumption of
small heeling angles in (20) does not expect to be causing significant error in the
simulation of the real phenomena in extreme astern seas.
2.3. Rudder forces
As an important element of the excitation for ship motions in following and

quartering seas, the forces on a rudder, when considered as a separate individual
element, can be obtained with the derivation of the lift and drag generated from the
rudder. Performance of rudder is greatly influenced by the interactions between
rudder to hull and rudder to propeller, due to the change of the lift.
Based on the above background, for calculation of rudder forces, Japanese MMG’
s (Japanese manoeuvring Group) model, Inoue et al. (1981), was adopted, including
the aforementioned interactions. The rudder forces and moments including rudder-
to-hull interaction are as follows:
X R ¼ F N sin d;
Y R ¼ ð1 þ aH ÞF N cos d;
(21)
N R ¼ ð1 þ aH ÞðxH =xR ÞÞxR F N cos d;
K R ¼ ð1 þ aH ÞzR F N cos d;
where XR, YR, NR, KR are surge, sway, yaw, roll rudder forces, respectively, FN,
rudder normal force, aH, rudder–to-hull interaction coefficient, xH, longitudinal
coordinate of the point of action of the rudder to hull interaction force, xR, zR,
longitudinal and vertical coordinates of the rudder’s centre of pressure.
2.4. Propulsive forces
Calculation of propulsive forces is modelled with respect to the thrust system used
in the ship whether it is propeller, water jet or the brand new azimuthing pod system.
ARTICLE IN PRESS
The calculation of propulsive forces therefore is dependent upon accurate

representation of wake, lifting and drag or more generally in the context of
propeller–hull–rudder interaction as it is mentioned in the previous section. In the
current model, propulsive forces are calculated using formulations in Inoue et al.
(1981) and Spyrou (1990) for the propeller system. All the dependents were obtained
from model tests results.
X P ¼ ð1 tp Þrn2 D4 K T ;
Y P ¼ rn2 D4 Y P ; (22)
2 5
N P ¼ rn D N P ;
where XP, YP, NP are surge, sway, yaw propulsive forces, respectively, tP, thrust
deduction at the propeller in forward motion, KT, thrust coefficient, D, propeller
diameter, n, propeller rate of rotation. Y Pn and N Pn are generally dependent upon u/
(nP) where P is propeller pitch.
2.5. Resistance forces
Resistance force is calculated from the model tests results. However, if the model
experiments are not available, it could be calculated with the methods such as
Holtrop and Mennen (1982) or if the certain regression coefficients exist from the
model data it could be calculated with below formulation
RT ðuÞ ¼ a0 u þ a1 ujuj þ a2 u3 , (23)
where a0, a1, a2 are regression coefficients.
2.6. Autopilot control
Although, automatic control systems can be regarded as parameter of ship

steering and control, as it is well known, the steering mechanism of the ships also
provides another external force. Following the previous research and experimental
set-up used in this study, the standard proportional–differential (PD) autopilot is
employed in order to keep the vessel on course
dR þ tr d_ R ¼ K R ðc cR Þ þ K P c_ (24)
dR is the actual rudder angle, cR is the desired heading angle, KR is yaw gain
constant, KP is a yaw rate gain constant and tr is the time constant in rudder
activation.
3. Impulse response functions (memory effects)
Following the research background on the effect of frequency dependent

terms, convolution terms are incorporated in the numerical model in
order to improve the prediction of the behaviour of the vessel at encounter
ARTICLE IN PRESS
frequencies which are not approximately zero. Impulse-response tests are required
for this, while for a purely theoretical prediction, at this stage, the frequency-
dependent coefficients are derived within the limits of potential theory. Based on the
theory of linear system, the hydrodynamic force due to an arbitrary motion can be
described with the convolution integral of the motion and the impulse response
function for hydrodynamic force, as far as linearity of hydrodynamic force is
assured. This would give rise to argument because the motions this study is dealing
with have most significantly high amplitudes and are of non-linear nature. The
suitability of the above approach for extreme motions will need, therefore, to be
explored.
When a body performs an irregular motion around its mean position, it is
appropriate to express the hydrodynamic force acting on the body in the time
domain. Following the work by Cummins (1962), the radiation force in the time
domain is written as
Z 1Z
F ij ¼ aij ð1ÞV_ j K ij ðtÞV j ðt tÞ dt,
0
i; j ¼ 1; 2; 3; 4; 5; 6, ð25Þ
where the first term is the infinite frequency added mass and the second
term is the impulse response function or the so-called ‘‘memory effect’’. It should
be noted that this formulation is based on linear theory. The retardation (Kernel)
function of Eq. (25) is the real part of Fourier transform of the frequency
domain damping function. In addition, the retardation can be described in terms of
damping as
Z
2 1
K ij ðtÞ ¼ Bij ðoÞ cos ot do, (26)
p 0
where Bij are damping coefficients . These equations are standard relations in linear
system theory. The impulse response function (Kij) will be solved from added mass
and damping data and the convolution integral (25) then evaluated for each term in
the equations of motion at each time step during the simulation.
3.1. Numerical solution of Kernel functions
To solve Kernel functions, use is made of Discrete Fourier transforms (DFT). The
DFT is particularly suitable for describing phenomena related to a discrete time
series. It can be developed from the Fourier transform of the continuing waveform,
samples of which are taken to form the time series. Hence, the retardation (Kernel)
function for any number of sample values is shown as
P
N1
K ij ðtÞ ¼ BðoÞ cosðotÞ do;
n (27)
0 n N 1;
ARTICLE IN PRESS
0.04
0.03
Impulse response function

0.02
0.01
0
0 2 4 6 8 10
-0.01
-0.02
-0.03
-0.04
Time (sec)
Fig. 2. Non-dimensional Kernel function of sway-roll coupled motion for Purse Seiner.
where B(o) is damping coefficient, do is frequency range. Note that the ot

expression can be written by means of general physical description as
2
ot ¼ p n tðnÞ, (28)
N
where t(N) indicates each time step.
The Kernel function is presented in Fig. 2 for a Purse Seiner fishing vessel which
will be explained in detail in the following sections. Damping values were obtained
by using 2D strip theory (Jasionowski, 2001) in 52 different frequencies with 0.07
range of frequency.
3.2. Implementation of convolution terms
Frequency dependent excitation forces (radiation) are calculated by using

convolution terms. The impulse response terms are calculated for radiation forces
in terms of excitation as it is given in Eq. (24) and its effect on numerical calculation
is shown in the next section.
For the calculation of radiation forces as mentioned above, 2D strip theory is
used. Although, strip theory has certain disadvantages for motions in astern seas
(Ayaz, 2003), its practicality and availability during the course of the study have
made usage of this numerical tool inevitable. Furthermore, as it is stated in the
previous section impulse response functions given in Eq. (25) are calculated based on
linear theory. Therefore, the radiation forces are calculated using a linear approach.
In the numerical tool (Jasionowski, 2001) the radiation forces are evaluated by a
strip theory formulation.
The radiation force terms computed in numerical model are given in a body fixed
axis system while the hydrodynamic coefficients are evaluated for stationary ship
attitude which follows the assumption that ship motions are the small amplitude.
However, as it is stated earlier, the ship amplitude can be very high for the motions
in extreme astern seas therefore those radiation forces are transformed into the
horizontal body axis system.
ARTICLE IN PRESS
With the inclusion of frequency dependent coefficients, Eq. (13) can be written as
follows:
Z 1Z
ðM þ AÞX€ þ B ðX ÞX_ þ C ðX ÞX þ K ij ðtÞV j ðt tÞ dt
0

¼ F zw ; X ; X_ ; X€ . ð29Þ
The Kernel functions are based on these coefficients for each station so that
radiation forces in the time domain are subject to the same limitations of the strip
theory. As it is well known surge is not considered with slender body theory as surge
forces are of higher order. However, radiation effects associated with surge could be
of importance when considering motions in following seas. In this study considering
that Froude–Krylov forces are most likely to be dominant forces in this case, the
radiation forces are omitted. However, surge added mass and damping coefficients
could be obtained for slender type of ships by means of a full three-dimensional
method. These coefficients are then used to generate Kernel functions related to
surge. In Eq. (29) in contrast to Eqs. (10) and (11), frequency dependent added mass
(at infinite frequency) and potential damping terms are calculated, again using the
numerical tool (Jasionowski, 2001) which includes the ‘‘end’’ terms as it is stated
before. Therefore, the damping coefficients in Eq. (26) might include the linear
manoeuvring components given in Eq. (19) inevitably resulting in overlapping of
these terms in the calculation with memory effects. Due to the nature of using linear
radiation forces by memory effects or ‘‘linear filter’’ the left-hand side of Eq. (29) is
linearized. The added mass term calculated at infinite frequency and damping terms
calculated for each encounter frequency are taken into the solution matrix. However
this results in the removal of non-linear coupled terms described in Eqs. (10) and
(11). Furthermore, those added mass and damping terms are stored for every 101
heading angles between 0 and 3601 and they need to be interpolated for particular
wave headings during the simulation of ship motions. Then, the response (Kernel)
functions are accounted for in predictions to motion continuously during the specific
encounter frequency therefore assuming unsteady characteristics of motion even
though ship might have steady motion when reaching certain speed in following and
quartering seas.
4. Verification of numerical model
In order to validate the numerical code, a time-domain simulation program

‘‘Simurg’’ has been developed. To validate of this numerical program two different
model ships were tested. A 23,720 tonnes displacement containership, investigated
using systematic model experiments at Osaka University in Japan (Hamamoto et al.,
1995) and a 712 tonnes displacement fishing vessel, investigated using systematic
model experiments at National Institute of Fisheries Engineering in Japan (Umeda
et al., 1994). The models are tested in different speeds, wave steepnesses, wave height
to ship length ratios and heading angles. The description of test methods,
ARTICLE IN PRESS
Fig. 3. Lines of the container vessel.
Table 1
Principal particulars of the full scale and the model scale container
Parameter Vessel Model (1/60)
LBP 150 m 2.5 m

B 27.2 m 0.453 m
D 13.5 m 0.225 m
df 8.5 m 0.142 m
da 8.5 m 0.142 m
Cb 0.667 0.667
D 23.720 t 110 kg
LCG 1.01 m 0.0168 m
KG 11.48 m 0.1913 m
GM 0.15 m 0.0025
Tj 43.3 s 5.59 s
Table 2
Control parameters of capsizing model runs
Nominal Froude number Fn 0.2 0.2 0.4

Autopilot course from the wave direction wc (degree) 0.0 45.0 30.0
Wave steepness H/l 1/25 1/25 1/25
Wave length to ship length ratio l/Lpp 1.5 1.5 1.5
Proportional gain KP 1.2 1.2 0.5
Differential gain KR (s) 6.84 6.84 2.85
instrumentation and all other details in the experiments are given by (Ayaz, 2003;
Hamamoto et al., 1995; Umeda et al., 1994). The full results of experiments and
comparison between numerical test with model tests according to the guidance given
by the ITTC for both vessels are also presented in (Ayaz, 2003; Ayaz et al., 2001;
Vassalos et al., 2003). Here, the effect of impulse response functions or so-called
‘‘memory effects’’ in the numerical model will be presented. For this purpose, two
aforementioned approaches in the numerical model (with and without memory
effect) have been tested against the experimental results for the container ship. The
first approach represented in Eqs. (12) and (13) was used in the numerical model for
6 DOF and the results presented in Figs. 3–6. The second approach represented in
Eq. (29) and with the methods explained in the previous section was employed in the
ARTICLE IN PRESS
Experiment
100
Roll angle (deg)

50
0
0 10 20 30
-50
-100
Time (sec)
Experiment
10
Pitch angle (deg)
0
0 10 20 30
-5
-10
Time (sec)
Experiment
10
5
Yaw angle (deg)
0
0 10 20 30
-5
-10
-15
Time (sec)
Experiment
15
Rudder angle (deg)
10
5
0
0 10 20 30
-5
-10
-15
Time (sec)
Fig. 4. Experimental simulation for the containership model in H/l¼ 1=25, l/Lpp ¼ 1.5, Fn ¼ 0.2,
wc ¼ 01.
ARTICLE IN PRESS
Num.1 Num.2
100
Roll angle (deg)

50
0
0 10 20 30
-50
-100
Time (sec)
Num.1 Num.2
10
Pitch angle (deg)
0
0 10 20 30
-5
-10
Time (sec)
Num.1 Num.2
10
Yaw angle (deg)
5
0
0 10 20 30
-5
-10
-15
Time (sec)
Num.1 Num.2
15
Rudder angle (deg)
10
5
0
-5 0 10 20 30
-10
-15
Time (sec)
Fig. 5. Numerical simulation for the containership model in H=l ¼ 1=25, l=Lpp ¼ 1.5, Fn ¼ 0.2, wc ¼ 01
(Num. 1—without memory effects, Num. 2—with memory effects).
numerical model for 6 DOF and the results are presented in Figs. 7–10. Lines and
particulars of containership and its model are given in Fig. 3 and Table 1. The GM
selected is not the design GM value, but one that only just satisfies the IMO
regulations. Environmental and control parameters are the same for each case and
the results for the runs considered are given in the same order (Table 2).
4.1. Discussion on results
Herein, the models are tested in different speeds, wave steepnesses, wave height to
ship length ratios and heading angles (Table 2). The graphs presented (Figs. 4–9),
ARTICLE IN PRESS
Experiment
50
Roll angle (deg)

0
0 20 40 60
-50
-100
Time (sec)
Experiment
10
Pitchl angle (deg)
0
0 20 40 60
-5
-10
Time (sec)
Experiment
60
Yaw angle (deg)
40
20
0
0 20 40 60
Time (sec)
Experiment
15
Rudder angle (deg)
10
5
0
0 20 40 60
-5
-10
-15
Time (sec)
Fig. 6. Experimental simulation for the containership model in H=l ¼ 1=25, l=Lpp ¼ 1.5, Fn ¼ 0.2,
wc ¼ 451.
ARTICLE IN PRESS
Num.1 Num.2
50
Roll angle (deg)

0
0 20 40 60
-50
-100
Time (sec)
Num.1 Num.2
10
Pitchl angle (deg)
0
0 20 40 60
-5
-10
Time (sec)
Num.1 Num.2
60
Yaw angle (deg)
40
20
0
0 20 40 60
Time (sec)
Num.1 Num.2
15
Rudder angle (deg)
10
5
0
-5 0 20 40 60
-10
-15
Time (sec)
show the comparison between the results of experiment and numerical method for
roll, pitch, yaw and rudder angles. In Figs. 4 and 5, the model was run in 0.2 Froude
number and 01 (pure following sea) autopilot course from the wave direction. The
second test (Figs. 6 and 7) was carried out in 0.2 Froude number and 451 autopilot
course from the wave direction. The third test (Figs. 8 and 9) was conducted for 301
autopilot course and 0.4 Froude numbers, respectively.
For numerical model runs, the 6 DOF model is used in aforementioned two
approaches, with and without memory effects. For the first approach, in the first run
(Fig. 4) the model experiences a wave crest at amidships for a while and this leads to
ARTICLE IN PRESS
Experiment
50
Roll angle (deg) 0

0 5 10 15
-50
-100
Time (sec)
Experiment
10
Pitch angle (deg)
0
0 5 10 15
-5
Time (sec)
Experiment
50
Yaw angle (deg)
40
30
20
10
0
0 5 10 15
Time (sec)
Experiment
15
10
Rudder angle (deg)
0
0 5 10 15
-5
-10
-15
Time (sec)
Fig. 8. Experimental simulation for the containership model in H=l ¼ 1=25, l=Lpp ¼ 1.5, Fn ¼ 0.2,
wc ¼ 451.
ARTICLE IN PRESS
Num.1 Num.2
50
Roll angle (deg)

0
0 5 10 15
-50
-100
Time (sec)
Num.1 Num.2
10
Pitch angle (deg)
0
0 5 10 15
-5
Time (sec)
Num.1 Num.2
50
Yaw angle(deg)
40
30
20
10
0
0 5 10 15
Time (sec)
Num.1 Num.2
15
Rudder angle (deg)
10
5
0
-5 0 5 10 15
-10
-15
Time (sec)
significantly large roll angle. After 15 s, the model cannot find enough buoyancy to
keep herself upright and it eventually capsizes. In the numerical model (Fig. 5), the
model reaches capsizing limit after a large jump in roll angle, instead of parametric
build up as it happened in the experiment.
On the other hand, the very large rudder deflection, first to port and then to the
starboard, is seen before capsizing. The numerical model seems to be having
problems in simulating rudder and yaw motions. The model proves to be too stiff.
Here ‘‘stiffness’’ means that deviations in yaw or rudder amplitudes are small or
the motion attitude is very rigid in contrast to large deviations in the experiments.
ARTICLE IN PRESS
Fig. 10. Body plan of the Purse Seiner.
This leads to smaller roll angles in the numerical model despite the fact that large
angles are fully taken into account in the numerical model. The numerical model
catches up satisfactorily the pitch angle, which is important in calculating the
instantaneous wave surface.
When the second approach was used in the same case (Fig. 5), the roll motion is
simulated with better accuracy in terms of amplitude. Problems are encountered
again, however, in simulating rudder motion and yaw motions. The model proves to
be too stiff again. However, this time yaw and rudder characteristics are highly
affected by frequency dependent terms and coupling between yaw and roll motion
leads to large roll periods and amplitude in the numerical model. The pitch angle is
almost identical to the first approach.
In the second run (Fig. 6) the model yet again experiences wave crest at
amidships and large roll angle after the autopilot course is set to 451, however
this time the model finds enough buoyancy to keep herself eventually
upright. The numerical model (Fig. 7) displays rudder and yaw motions rather
well, despite the same ‘‘stiffness’’ problem as in the first run still exists. Also, the roll
motion displays behaviour similar to sub-harmonic roll motion observed in the
experiment.
After the speed is further increased (Fig. 8), the model experiences very large roll
angle and ultimately capsizes. The numerical model (Fig. 9) satisfactorily displays
the roll and pitch motions and estimates the rudder and yaw angles with reasonable
accuracy.
When the second numerical approach is adopted (Fig. 9), this time, similar roll
and pitch characteristics to the first approach are found. However this numerical
model predicts the rudder and yaw behaviour with more reasonable accuracy.
An overview for the simulations of container ship indicates that there is a better
agreement with experiments for the second approach in which impulse response
functions are included in numerical model, than for the first approach in which
impulse response functions are not included, in terms of roll amplitude.
ARTICLE IN PRESS
The second approach estimates larger roll amplitude and harmonic motions which
are closer to experiments. It could be a result of better estimated damping forces
when the effect of encounter frequency is included. However, this difference could be
more realistically explained in terms of changes in the yaw and rudder motions. The
coupling between roll and yaw–rudder motions seems to create the aforementioned
larger roll amplitudes. The change in yaw and rudder characteristics indicates the
effect of frequency to the control systems. However, concerning these experiments,
the ITTC benchmark review committee (ITTC, 2002) urged that the exact
calculation of memory effects should be carried out from the start of the waves.
Therefore, this benchmark testing, which does not specify initial conditions of fluid
motions, is not appropriate for this purpose. Based on this finding, the effect of
frequency or so-called ‘‘memory effects’’ will be looked at in detail in the next
section.
5. Benchmark study of ITTC vessel
Following the current knowledge in the literature and the outcome of validation of
the new mathematical model which were presented in previous sections, extensive
captive and free running model experiments were carried out at National Research
Institute of Fisheries Engineering, Japan for the aforementioned 712 tonnes
Japanese fishing vessel in all 6 DOF and extreme random seas. The captive model
experiments were carried out while the model was fixed for all 6 DOF. Hence, it was
aimed to observe the manoeuvring and course-keeping behaviour of the vessel in
large vertical motions. The free running model experiments were also carried out in
random seas using both ITTC and JONSWAP spectra. The captive model
experiments were carried out for different speed, heading angle, sinkage, trim and
for some cases with different wave steepness, while the free running model
experiments were carried out in different significant wave heights, modal frequencies,
speeds and wave directions. Extremity of the conditions of captive model runs were
defined within limits and strength of model that was used in these experiments.
The experiments were carried out in the seakeeping and manoeuvring basin named
‘‘Marin Dynamics Basin’’ of the National Research Institutes of Fisheries
Engineering, Japan. The basin is 60 m long, 25 m wide and 3.2 m deep. It has an
X–Y towing carriage consisting of two carriages: a main carriage runs in the
longitudinal direction of the basin and a sub-carriage which runs in the transverse
direction of the basin. The maximum velocities of the main and sub carriages are 3
and 1.5 m/s, respectively. The basin is equipped with an 80-segments wave maker.
The carriages and wave maker are controlled by a digital feedback system.
5.1. Captive Model Tests
For these tests, a 2 m length (1/17.25 scale) model is used (Fig. 10). This vessel has
been tested as part of a series of the benchmark tests commissioned by the ITTC
Specialist Group on Ship Stability (ITTC, 1999b). Its principal particulars are given in
ARTICLE IN PRESS
Table 3
Principal particulars of fishing vessel and model
Parameter Vessel Model (1/17.5)
LBP 34.50 m 2m
B 7.60 m 0.441 m
D 3.07 m 0.178 m
df 2.50 m 0.145 m
da 2.80 m 0.162 m
Cb 0.597 0.597
D 425.18 t 81.08 Kg
LCG 1.31 m 0.076 m
KG 3.36 m 0.195 m
GM 1.0 m 0.058 m
Tj 7.4 s 1.6 s
G
ξ
χ
ξG/λ
y
O η
η
O
G
y
z
ζ
Fig. 11. Coordinate system of model basin (Umeda et al., 1995).
Table 3. The model is equipped with rudder but without propeller and bilge keels.
Being fixed in all 6 DOF, it was fitted to a turning table on the sub-carriage. Previously,
the model is weighted and balanced with a 6 components loadcell on the right
displacement. The 6 DOF forces and moments (surge, sway, heave forces, roll, pitch
and yaw moments) are measured by a dynamometer, which is placed on the loadcell.
ARTICLE IN PRESS
The centre of yaw moment is taken as longitudinal centre of buoyancy, L.C.B., and
that of the roll moment is distance of the vertical position of centre of gravity to still
water surface. The centres of heave and pitch are defined from the centre of X–Y axes
on the loadcell. A servo-needle wave probe was also fitted with the sub carriage. Based
on the coordinate systems shown in Fig. 11, the sign convention is as follows.
The positive surge force acts towards the bow, the positive sway force acts
starboard, the positive yaw moment induces a starboard turn, the positive roll
moment results in a downward moment inducing a starboard turn, and positive wave
elevation is downwards. The positive heave force is upwards and positive pitch
moment is aft downwards. The details of those procedure and measurement methods
are also explained in Ayaz (2003) and Umeda et al. (1994).
The experimental procedure is as follows. First, the wave maker generated a
regular wave train propagating in the longitudinal direction of the basin. Next,
combining movements of the two carriages, the model was towed with a certain
angle from the wave direction. Since the centre line of the model had been adjusted
to this towing direction, the model runs with a specified heading angle relative to the
waves without drift angle.
Test matrix for the aforementioned conditions is given in Table 4. Here the vessels
are tested for different combination of sinkage, trim and heel. The resistance force is
subtracted from the surge wave force and the additional roll moment created from
the vertical distance of the dynamometer to the water surface is also accounted for
the calculation of the wave roll moment. Results are also presented as function of the
relative position on the wave as it is explained above. The mathematical model in the
light of captive model tests can be written as follows:
Recalling from Eq. (13)
ðM þ AÞX€ þ B ðX ÞX_ þ C ðX ÞX ¼ F 00 ðzw Þ þ F 01 ðzw ; X Þ þ F 02 ðzw ; X_ Þ

þ F 03 ðzw ; X€ Þ. ð30Þ
Here, F00 and F01 are force components from captive model experiments, F02, F03 are
non-linear terms and BðX_ Þ, C(X) are obtained from forced tests in calm water or
from hydrodynamics calculations.
Table 4
Test matrix for l/L ¼ 1.5, nominal GM
H/l Fn w (deg) Sinkage (m) y (deg) f(deg)
0 45 60 0.2 0 0.2 1.43 0 1.43 10 0 10
1/25 0.2 O O O O
0.3 O O O O O O O
0.4 O O O O
0.3 O O O O O O O O O O O O
1/20 0.4 O O O O O O O O O O O
0.3 O O O O O O O
1/15 0.4 O O O O O O O
ARTICLE IN PRESS
0.12
0.1 Exp.
0.08 Num.
X′
0.06
0.04
0.02
0
0.04 0.05 0.06 0.07
H/λ
0.75
ξ/λ
0.5
Exp.
0.25
Num.
0
0.04 0.05 0.06 0.07
H/λ
Fig. 12. Wave induced surge force with respect to wave steepness (Fn ¼ 0.3, l/L ¼ 1.5, w ¼ 01, f ¼ 01,
y ¼ 01, sinkage ¼ 0 m).
Based on the above outcome, the challenge will be to identify the first two terms on
the right-hand side of Eq. (30) with respect to various parameters such as wave
steepness, heading angle and encounter frequency and as it is explained in the previous
section, relative position of the ship on the wave. Therefore, a number of figures were
plotted (Figs. 12–18) for this purpose and experimental results were compared against
numerical model explained in Sections 2 and 3. The presentation method of the results
is similar to Umeda et al. (1995), except, this time the experiments were carried out for
fully captive model with different heel, sinkage and trim. Following the above
statement, the presentation of the results was divided into two parts, first the case for
zero heel angle, sinkage and trim angle were plotted with respect to wave steepness,
heading angle and encounter frequency (Figs. 12–20) and in the second part the cases
for whole heading angle, sinkage and trim angle changes were taken into account as it
can be seen from the test matrix in Table 4 (Figs. 21–26). The effect of encounter
frequency was shown both in respect to heading angle and also t, Hanaoka parameter,
as described in Umeda et al. (1995). The Hanaoka parameter, t, is defined by
Uoe
t¼ . (31)
g
Here U is ship velocity, oe, is encounter frequency and g is gravitational acceleration.
The Hanaoka parameter governs wave patterns from an oscillating obstacle with
ARTICLE IN PRESS
0.15
0.1
Z′
0.05 Exp.
Num.
0
0.04 0.05 0.06 0.07
H/λ
0.75
ξ/λ
0.5
Exp.
0.25 Num.
0
0.04 0.05 0.06 0.07
H/λ
Fig. 13. Wave induced heave force with respect to wave steepness (Fn ¼ 0.3, l/L ¼ 1.5, w ¼ 01, f ¼ 01,
y ¼ 01, sinkage ¼ 0 m).
forward velocity. The relative position of centre of ship gravity to a wave trough
behind the ship is non-dimensionalized with respect to wave length. This non-
dimensional value, xG/l, of 0.0 indicates that the ship is situated on a wave trough.
Hydrodynamic forces and moments are non-dimensionalized as follows:
X 0 ¼ X W =ðrgzw BLÞ;
Y 0 ¼ Y W =ðrgzw BLÞ;
Z 0 ¼ Z W =ðrgzw BLÞ;
(32)
K 0 ¼ K W =ðrgzw B2 LÞ;
M 0 ¼ M W =ðrgzw BL2 Þ;
N 0 ¼ N W =ðrgzw BL2 Þ;
where prime superscript, ‘‘0 ’’, refers to non-dimensionalized quantities and zw is the
wave amplitude.
5.1.1. Discussion on captive model test results

Here, the results obtained will be of interest for the first two terms in Eq. (30). It is
stated previously that one of the aims of this study is to obtain a fully coupled
(6 DOF) numerical model for the accurate representation of the ship motions in
extreme astern seas.
ARTICLE IN PRESS
0.1
0.08 Exp.
Num.
0.06
M′
0.04
0.02
0
0.04 0.05 0.06 0.07
H/λ
0.75
ξ/λ
0.5
Exp.
0.25
Num.
0
0.04 0.05 0.06 0.07
H/λ
Fig. 14. Wave induced pitch moment with respect to wave steepness (Fn ¼ 0.3, l/L ¼ 1.5, w ¼ 01, f ¼ 01,
y ¼ 01, sinkage ¼ 0 m).
The requirement of obtaining those terms was accomplished through the wave
force values obtained in captive model tests. Since the aim was to carry out this study
in a full 6 DOF numerical model, the vertical motions were given special attention.
Therefore, the above results were demonstrated in two different conditions. The first
group is the zero heel, sinkage and trim (Figs. 12–20) and the second group with the
maximum roll, sinkage and trim values the model allowed (Figs. 21–26). Here, the
results are represented in terms of amplitude and phase angle. The results are given
in terms of wave steepness, encounter frequency and heading angle in order to assess
the fully captive results and the accuracy of numerical model for the same
conditions. In the first group, the effect of wave steepness, encounter frequency and
heading angle were investigated. In the second group the results were represented in
terms of wave forces and moments with respect to heading angle. The aforemen-
tioned Hanaoka parameter represents the effect of encounter frequency.
There are other major points that are required to be mentioned here. In the
calculation of wave force, while surge force includes Froude–Krylov forces only,
others are also included such as diffraction forces calculated from Ohkusu’s theory
as it is explained in Section 2.1.1.2.
Figs. 12–14 show the effect of wave steepness on the vertical wave forces and
moments for Fn ¼ 0.3 and zero sinkage, trim and heel angle condition. For the
experimental and numerical simulations in Fn ¼ 0.3 (Figs. 12 and 13) it could be seen
ARTICLE IN PRESS
0.1
Exp.
0.08
Num.
0.06 Calm
X′
0.04
0.02
0
0 15 30 45 60
χ (deg)
0.75
Exp.
ξ/λ
0.5
Num.
0.25
0
0 15 30 45 60
χ (deg)
Fig. 15. Wave induced surge force with respect to heading angle (Fn ¼ 0.3, l/L ¼ 1.5, H/l ¼ 1/20,
f ¼ 01, y ¼ 01, sinkage ¼ 0 m).
that non-dimensionalized surge and heave forces and pitch moment are almost
constant up to wave steepness, 1/15 which is highest wave steepness in the
calculations. In the numerical simulations, there is reasonably satisfactory agreement
in terms of amplitude and phase angle despite some deviations of pitch moment in
sway and yaw values for higher wave steepness (Fig. 14).
As a second criterion, the wave forces and moment were plotted with respect to the
wave heading (Figs. 15–17) for Fn ¼ 0.3, H=l ¼ 1=20 and zero sinkage, trim and
heel angles. Here, the calm water values for wave forces and moments are also added
to show the difference between wave motions and calm water values especially in
terms of vertical motions. As it can be seen from the figures there is linear increase
for wave force and moment values. The agreement is again reasonably satisfactory.
For the surge forces, although the experiments show increase in amplitude when
wave heading is getting closer to 601, in accordance with Froude–Krylov assumption
the numerical model shows linear decrease (Fig. 15). For other motions the biggest
deviation is seen in 601. Therefore, this could be attributed to the diffraction
modelling. Yet, the Hanaoka parameter which will be investigated later, indicates
that the encounter frequency does not significantly differ, therefore other factors
should be considered. One of these could be the calculation of the wave forces which
ARTICLE IN PRESS
0.5
Exp.
0.4
Num.
0.3 Calm
Z'
0.2
0.1
0
0 15 30 45 60
χ (deg)
1
0.75
ξ/λ
0.5
0.25 Exp.
Num.
0
0 15 30 45 60
χ (deg)
Fig. 16. Wave induced heave force with respect to heading angle (Fn ¼ 0.3, l/L ¼ 1.5, H/l ¼ 1/20,
f ¼ 01, y ¼ 01, sinkage ¼ 0 m).
considers change in wetted surface due to the position on the wave and the
additional roll moment created from the vertical distance of the dynamometer to the
water surface where the numerical model is believed to be more accurate for pure
following or head sea conditions. However, it seems that some inadequacies exist for
quartering seas cases where all rotations are highly important. Due to these problems
the amplitudes seem to be slightly lesser when wave heading is getting closer to beam
seas. The phase of motions show satisfactory agreement in this comparison as well.
Comparisons between the amplitude of vertical forces and moment and dynamic
forces and moments calculated in calm water indicate that there is a significant wave
effect which might justify the need to relying on 6 DOF mathematical models
(Figs. 16 and 17).
Although, the wave heading could be regarded as another indicator, here, wave
forces and moments with respect to the Hanaoka parameters were plotted in
H=l ¼ 1=15, w ¼ 01 to display the effect of encounter frequency (Figs. 18–20). It is
emphasized in Umeda et al. (1995) based on experiments that the effect of frequency
is very small and wave-induced sway forces, roll and yaw moments are almost
constant for to0.25. This conclusion can be confirmed to a certain extent for these
experiments as well (Ayaz, 2003). Then, the same conclusion could be extended for
the vertical motions (Figs. 19 and 20). The numerical simulation compared
reasonably well in terms of amplitude and phase angle. However, one interesting
point was seen in the numerical simulation of pitch moment (Fig. 20) for values very
near 0.25 of the Hanaoka parameter. Here, a singular behaviour exists that could be
ARTICLE IN PRESS
0.1
Exp.
0.08
Num.
0.06 Calm
M′
0.04
0.02
0
0 15 30 45 60
χ (deg)
0.75
ξ/λ
0.5
Exp.
0.25
Num.
0
0 15 30 45 60
χ (deg)
Fig. 17. Wave induced pitch moment with respect to heading angle (Fn ¼ 0.3, l/L ¼ 1.5, H/l ¼ 1/20,
f ¼ 01, y ¼ 01, sinkage ¼ 0 m).
explained by a 3D wave making theory. Such theory is not employed in this study,
however the effect of 3D correction terms given in Ohkusu’s theory (Ohkusu, 1986)
is believed to be the likely source of this behaviour in the numerical simulation.
In the second part of the experiments, results for the model runs with maximum
heel angle, trim angle and sinkage are shown. In the first part (Figs. 21–23) the
results for wave forces and moments with respect to wave heading in
H=l ¼ 1=20; Fn ¼ 0:3y ¼ 1:431 and sinkage ¼ 0.2 m are presented.
For the model runs with sinkage ¼ 0.2 m there is no significant difference with
the results in zero values in vertical force and moment amplitudes. Yet, this time the
numerical model overestimates the amplitudes for vertical motions in heading angles
45 and 601 (Figs. 22–23). However, this cannot be seen as significant difference.
Similar problems to the previous case persist in the calculation of wave forces and
moments in these motions as well. Furthermore, in the case of negative trim angle,
the model overestimates the amplitudes of vertical motions (Figs. 22 and 23) for
negative sinkage values. In case of positive trim angle they seem to be similar
(Figs. 24–26) although the phase angles of force and moment amplitudes change.
Similar discrepancies with zero sinkage and trim cases exist.
ARTICLE IN PRESS
0.1
Exp.
0.08
Num.
0.06
X′
0.04
0.02
0
0.1 0.15 0.2 0.25
τ
0.75
0.5
ξ/λ
Exp.
0.25
Num.
0
0.1 0.15 0.2 0.25
τ
Fig. 18. Wave induced surge force with respect to Hanaoka parameter (l/L ¼ 1.5, H/l ¼ 1/15, w ¼ 01,
f ¼ 01, y ¼ 01, sinkage ¼ 0 m).
Overall, the numerical model gives reasonably satisfactory agreement with

experimental results. Although, the effect of each individual wave force components
has not been shown, the differences occurring between numerical and experimental
results can be attributed to calculation of incident wave forces due to position on the
wave. The detailed results of captive model tests for all 6 DOF can be found in Ayaz
(2003).
5.2. Free running model tests
In the free running model tests, a 2.3 m model (1/15 scale) of a Purse Seiner fishing
vessel, which was previously mentioned in ITTC Benchmark testing (ITTC, 2002),
was used. The principal particulars of the model are given in Table 5. The model was
equipped with bilge keels, rudder and propeller. The model was propelled with
electric motors, with power supplied by batteries on board with feedback controls of
propeller revolution. Roll, pitch, and yaw angles were measured by fibreoptical
gyroscopes. The measured signals as well as the rudder angle and propeller
revolutions were recorded by on-board computers in digital form. The measured yaw
ARTICLE IN PRESS
0.3
Exp.
0.25
Num.
0.2
0.15
Z′
0.1
0.05
0
0.1 0.15 0.2 0.25
τ
1
Exp.
Num.
0.75
ξ/λ
0.5
0.25
0
0.1 0.15 0.2 0.25
τ
Fig. 19. Wave induced heave force with respect to Hanaoka parameter (l/L ¼ 1.5, H/l ¼ 1/15, w ¼ 01,
f ¼ 01, y ¼ 01, sinkage ¼ 0 m).
angle was also used for the autopilot control. More details on procedure and
measurement methods are also explained in Ayaz (2003) and Umeda et al. (1994).
5.2.1. Numerical method for random waves

In the numerical calculation of motions in random seas, the model developed by
Jasionowski (2001) and Letizia (1996) was used. The total incident wave excitation
and diffraction forces due to the action of confused seas can be written according to
horizontal body axes system as
X
N
F wave_exc ðtÞ ¼ ai ðF A 0 0
FKþDIF Þi cosðowi t ki ðxG þ x cos c y sin ci þ si Þ
i
þ ðF dFKþDIF Þi Þ, ð33Þ
where FA d
FK+DIF is the amplitude of the force in random seas and FFK+DIF is the
phase lag of the force. However, it should be noted that since the Froude–Krylov
forces are not linear, Eq. (33) is an approximate solution. This is the approach
ARTICLE IN PRESS
0.1
Exp.
0.08
Num.
0.06
M′ 0.04
0.02
0
0.1 0.15 0.2 0.25
τ
0.75
ξ/λ
0.5
Exp.
0.25 Num.
0
0.1 0.15 0.2 0.25
τ
Fig. 20. Wave induced pitch moment with respect to Hanaoka parameter (l/L ¼ 1.5, H/l ¼ 1/15, w ¼ 01,
f ¼ 01, y ¼ 01, sinkage ¼ 0 m).
followed in this study in accordance with the numerical tool employed for random
wave calculations (Jasionowski, 2001; Letizia, 1996). Similar to regular waves, speed
as well as heading of the vessel can change and this inevitably affects the forces.
Therefore, once the ship changes its mean speed and heading, Froude–Krylov and
diffraction forces are updated. The amplitude, ai, and phase angles, si, of each
harmonic component is usually found from a given wave energy spectrum and
through digital random number generators, a procedure widely discussed in
literature. As for the effect of encounter frequency in random waves, the same
method (Jasionowski, 2001; Letizia, 1996) has been employed for random waves
applying the procedure given in Section 3.1 and 3.2.
5.2.2. Discussion on free running model tests

From the free running tests point of view, the priority was to have a healthy wave
spectrum to lay foundations to the numerical models in the effort of determining the
boundaries of capsize. The model was run for different speeds and heading angles in
both ITTC and JONSWAP wave spectra.
ARTICLE IN PRESS
0.1
Exp.
0.08 Num.
Calm
0.06
X′
0.04
0.02
0
0 15 30 45 60
χ (deg)
0.75
ξ/λ
0.5
Exp.
0.25 Num.
0
0 15 30 45 60
χ (deg)
f ¼ 101, y ¼ 1.431, sinkage ¼ 0.2 m).
The model was run for three different heading angles (51, 451 and 601) in 3
different modal periods, Tp (4.7, 5.76, 6.48 at full scale) for three different significant
wave heights Hs (1.725, 2.5875, 3.45 m at full scale) at two different speeds (Fn ¼ 0.3
and 0.4) in both ITTC and JONSWAP spectra.
In order to simulate the motions of ships in random following and quartering seas
numerically, the aforementioned numerical tool (Jasionowski, 2001) was employed.
The wave history of motions is taken based on the inertial co-ordinate of the ship
with respect to point in which the waves are measured. Although the ship travels
along the model tank therefore the encounter wave history will be different with
time, the numerical model calculates the ship motions for the point wave history
measured in the experiments. In other words, the ship is always fixed to the one point
in which the wave history is measured.
The numerical model described in Section 5.2.1 was also used to carry out the
simulations for the above conditions and they are verified against some of the
experimental results. In order to test the accuracy of the numerical model two
different conditions of two spectra were chosen and the numerical simulations were
verified against the experimental results in those conditions for roll, pitch and yaw.
In Figs. 27–29, the simulations were carried out for Tp ¼ 5.76 (1.487 sec at model
scale), and H4 ¼ 2.588 m (0.1725 m at model scale) at two different speeds Fn ¼ 0.3
ARTICLE IN PRESS
0.5
Exp.
0.4
Num.
0.3 Calm
Z′ 0.2
0.1
0
0 15 30 45 60
χ (deg)
0.75
ξ/λ
0.5
Exp.
0.25
Num.
0
0 15 30 45 60
χ (deg)
f ¼ 101, y ¼ 1.431, sinkage ¼ 0.2 m).
and 0.4 for the ITTC spectrum. In Figs. 30–32 the model was simulated and
compared against experiments Tp ¼ 6.48 s (1.717 s at model scale), and Hs ¼ 3.45 m
(0.23 m at model scale) at two different speeds Fn ¼ 0.3 and 0.4 for the JONSWAP
spectrum. Both simulations were carried out in heading angle w, 51, 451, 601,
respectively. For the simulations, the initial conditions, propeller revolution and
autopilot parameters were modelled according to the experiment results.
In order to assess the overall results, in Figs. 33–36, single amplitude significant
roll and pitch angles for both spectra with respect to heading angle in Fn ¼ 0.3 and
0.4 were plotted for experiments and numerical results.
As it can be seen from the graphs time simulation is filtered for 100 s by applying
the method described in Section 5.2.1. Incident and diffraction wave excitation terms
and radiation terms are also expressed following the same methodology.
In general, these experiments have not been carried out to find boundaries of
capsizing for dangerous situations that occur in the extreme astern seas, it was
intended to verify the accuracy of the numerical model which could be used for such
purpose. Therefore, the amplitude of the experimental results and the accuracy of
ARTICLE IN PRESS
0.1
Exp.
0.08 Num.
Calm
0.06
M′
0.04
0.02
0
0 15 30 45 60
χ (deg)
0.75
ξ/λ
0.5
Exp.
0.25 Num.
0
0 15 30 45 60
χ (deg)
f ¼ 101, y ¼ 1.431, sinkage ¼ 0.2 m).
numerical model to predict those values will be regarded as an indicator for further
studies.
In Figs. 27–29, some of the simulations for the Tp ¼ 1.487 s (5.76 s at full scale),
and Hs ¼ 0.1725 m (2.588 m at full scale) at two different speeds Fn ¼ 0.3 and 0.4 for
the ITTC spectrum were shown at 51 initial heading angle. It could be seen in these
figures that best agreement is provided in simulation of yaw motion in contrast to
previous regular wave motions. In the numerical models, initial conditions and
values are taken from experiments using the actual rudder and autopilot parameters.
Since the motions do not have the same extreme amplitudes and transient behaviour
as in the regular wave motions, accuracy of the simulations for yaw motions can be
expected. Therefore, it can be assumed that coupling effects of those motions are
predicted with reasonably satisfactory agreement.
For the roll and pitch motions, the best indictor can be seen in the graphs
illustrated in Figs. 33 and 34. Here, the single amplitude significant roll and pitch
angles are plotted versus the wave heading. As for the roll motion, it is usually
considered that, since viscous effects play a major part, the amplitude is usually less
than for the numerical simulations which ignore viscous effects.
ARTICLE IN PRESS
0.1
Exp.
0.08 Num.
Calm
0.06
X′
0.04
0.02
0
0 15 30 45 60
χ (deg)
0.75
ξ/λ
0.5
Exp.
0.25 Num.
0
0 15 30 45 60
χ (deg)
f ¼ 101, y ¼ 1.431, sinkage ¼ 0.2 m).
However, when the ship motions in astern seas are concerned, the rapid change in
the restoring and numerical modelling of restoring could be a major factor in the
results as it was seen in Section 4. Here, too, the roll amplitude was found to be
larger than that in the experiment because of, presumably, the same reason. In
Fig. 33, it is seen that the roll amplitude is better predicted for the high-speed range,
as the change of restoring is less than at the slower speed. The accuracy of the
numerical model for the high-speed range has been seen throughout the simulations
in this study. The difference, however, in the experimental results are insignificant.
For the pitch motions, however, the accuracy is not as good as the roll motion. A
similar problem has been seen in the amplitude for the Purse Seiner trawler in regular
wave tests as well (Ayaz, 2003). Therefore, it could be attributed to geometrical form
of this particular vessel. The pitch motions for the container vessel Section 4, for
instance, had shown a good agreement. The agreement between numerical and
experimental results is similar to that of roll motion, although there is significant
difference in the amplitude for 451 wave heading in Fn ¼ 0.3 (Fig. 34).
In Figs. 30–32 the model was simulated and compared against experiments
Tp ¼ 1.717 s (6.48 s at full scale), and Hs ¼ 0.23 m (3.45 m at full scale) at two
ARTICLE IN PRESS
0.5
Exp.
0.4
Num.
0.3 Calm
Z′
0.2
0.1
0
0 15 30 45 60
χ (deg)
0.75
ξ/λ
0.5
Exp.
0.25
Num.
0
0 15 30 45 60
χ (deg)
f ¼ 101, y ¼ 1.431, sinkage ¼ 0.2 m).
different speeds Fn ¼ 0.3 and 0.4 for the JONSWAP spectrum. When the significant
wave height increases with longer modal period, the overall characteristics of the
motions do not show significant difference. The yaw and rudder motions yet again
were predicted with good agreement. Similar to the first case, single amplitude
significant roll and pitch angles were plotted with respect to wave heading in Figs. 35
and 36. Here, there is a better agreement for both motions especially in Fn ¼ 0.4.
However, the same difference in significant pitch angle amplitude for Fn ¼ 0.3 and
451 wave direction exists for this motion as well. The amplitude of the motions is
slightly less than the previous case despite the higher significant wave height. This
shows the important effect of the frequency in the motions.
Overall, the results indicate that motions with higher amplitudes occur closer to
pure following seas. This verifies the common knowledge in the motions of ships in
extreme astern seas. Concerning the effect of frequency, the numerical model shows
similarly good agreement for motions in 601 as in the following and quartering seas.
However, the difference between the motion attitudes is not significantly different.
The diffraction model based on the low encounter frequency may contribute to the
ARTICLE IN PRESS
0.1
Exp.
0.08
Num.
Calm
0.06
M′ 0.04
0.02
0
0 15 30 45 60
χ (deg)
0.75
ξ/λ
0.5
Exp.
0.25 Num.
0
0 15 30 45 60
χ (deg)
f ¼ 101, y ¼ 1.431, sinkage ¼ 0.2 m).
Table 5
Principal particulars of fishing vessel and model of free running tests
Parameter Vessel Model (1/15)
LBP 34.50 m 2.3 m

B 7.60 m 0.507 m
D 3.07 m 0.205 m
df 2.50 m 0.166 m
da 2.80 m 0.186 m
Cb 0.597 0.597
D 425.18 t 125.588 kg
LCG 1.31 m 0.087 m
KG 3.36 m 0.224 m
GM 1.0 m 0.0667 m
Tj 7.4 s 1.9 s
difference in wave headings near to beam quartering seas. Phase angle and other
parameters are obtained from the wave spectra and the simulations display good
agreement.
ARTICLE IN PRESS
Exp.
30
20
Roll Angle (deg) 10
0
0 20 40 60 80 100
-10
-20
-30
Time (sec)
Num.
30
20
Roll Angle (deg)
10
0
0 20 40 60 80 100
-10
-20
Time (sec)
Fig. 27. Roll angle, ITTC spectrum, Fn ¼ 0.3, w ¼ 51, Hs ¼ 0.1725 m, Tp ¼ 1.487 s.
Exp.
10
Pitch Angle (deg)
0
0 20 40 60 80 100
-5
-10
Time (sec)
Num.
10
Pitch Angle (deg)
0
0 20 40 60 80 100
-5
-10
Time (sec)
Fig. 28. Pitch angle, ITTC spectrum, Fn ¼ 0.3, w ¼ 51, Hs ¼ 0.1725 m, Tp ¼ 1.487 s.
ARTICLE IN PRESS
Exp.
60
40
Yaw Angle (deg)
20
0
0 20 40 60 80 100
-20
-40
-60
Time (sec)
Num.
60
40
Yaw Angle (deg)
20
0
0 20 40 60 80 100
-20
-40
-60
Time (sec)
Fig. 29. Yaw angle, ITTC spectrum, Fn ¼ 0.3, w ¼ 51, Hs ¼ 0.1725 m, Tp ¼ 1.487 s.
Exp.
30
20
Roll Angle (deg)
10
0
0 20 40 60 80 100
-10
-20
Time (sec)
Num.
30
20
Roll Angle (deg)
10
0
0 20 40 60 80 100
-10
-20
Time (sec)
Fig. 30. Roll angle, JONSWAP spectrum, Fn ¼ 0.4, w ¼ 601, Hs ¼ 0.235 m, Tp ¼ 1.717 s.
ARTICLE IN PRESS
Exp.
10
Pitch Angle (deg)

5
0
0 20 40 60 80 100
-5
-10
Time (sec)
Num.
10
Pitch Angle (deg)
0
0 20 40 60 80 100
-5
-10
Time (sec)
Fig. 31. Pitch angle, JONSWAP spectrum, Fn ¼ 0.4, w ¼ 601, Hs ¼ 0.235 m, Tp ¼ 1.717 s.
6. Conclusions
The main aim of this study was to develop a coupled non-linear 6-DOF model
with frequency dependent coefficients, incorporating memory effects in random
waves with a new axis system that allows straightforward combination between
seakeeping and manoeuvring models whilst accounting for extreme motions.
In order to achieve this goal a new mathematical model was developed which allowed
meaningful combination of manoeuvring and seakeeping motions in the prediction of
extreme ship motions. It is enhanced by incorporating convolution terms (so-called
‘‘memory effects’’) that improve the prediction of the behaviour of the vessel in non-zero
encounter frequencies. Furthermore, the numerical model is expressed for random
waves, thus it is suitable for simulating ship behaviour in a realistic environment.
Validation of the numerical model indicates that the numerical model could
provide reasonably adequate predictions of behaviour in extreme astern seas,
especially qualitatively, and it could identify the dangerous situations that may lead
to capsizing. The coupling between vertical motions and other motions should be
taken into account in an attempt to improve the accuracy of numerical calculations.
For certain initial conditions and wave conditions the frequency dependent terms
could affect the accuracy of the numerical model and there is a considerable effect of
the encounter frequency on ship controllability. However, these numerical
simulations showed that this effect cannot be justified as an ‘‘improvement’’.
ARTICLE IN PRESS
Exp.
90
60
Yaw Angle(deg) 30
0
0 20 40 60 80 100
-30
-60
-90
-120
Time (sec)
Num.
90
60
Yaw Angle (deg)
30
0
0 20 40 60 80 100
-30
-60
-90
-120
Time (sec)
Fig. 32. Yaw angle, JONSWAP spectrum, Fn ¼ 0.4, w ¼ 601, Hs ¼ 0.235 m, Tp ¼ 1.717 s.
Hs=0.1725 m; Tp=1.487 sec.

20
Sign. roll (deg), single ampl.
Exp. (Fn=0.3)
Exp. (Fn=0.4)
15 Num. (Fn=0.3)
Num. (Fn=0.4)
10
0
0 20 40 60 80
χ (deg)
Fig. 33. Single amplitude significant roll angle with respect to wave direction for Fn ¼ 0.3 and 0.4 in ITTC
spectrum.
In an attempt to enhance the knowledge regarding the effect of vertical motions

and to assess the reliability of the numerical model in a random waves environment,
captive and free running model experiments were conducted. The numerical model
provides good agreement for fully captive model and difference between wave and
ARTICLE IN PRESS
Hs=0.1725 m; Tp=1.487 sec.

10
Sign. pitch (deg), single ampl.

Exp. (Fn=0.3)
9
Exp. (Fn=0.4)
8
Num. (Fn=0.3)
7 Num. (Fn=0.4)
6
5
4
3
2
1
0
0 10 20 30 40 50 60 70
χ (deg)
Fig. 34. Single amplitude significant pitch angle with respect to wave direction for Fn ¼ 0.3 and 0.4 in
ITTC spectrum.
Hs=0.23 m; Tp=1.717 sec.

20
Exp. (Fn=0.3)
Sign. roll (deg), single ampl.
Exp. (Fn=0.4)
15 Num. (Fn=0.3)
Num. (Fn=0.4)
10
0
0 20 40 60 80
χ (deg)
Fig. 35. Single amplitude significant roll angle with respect to wave direction for Fn ¼ 0.3 and 0.4 in
JONSWAP spectrum.
calm water values indicate the importance of vertical forces and moments, yet
differences in terms of wave steepness, heading angle and encounter frequency are
not so significant when the maximum heeling, sinkage and trim values used in these
experiments are concerned.
The agreement of the numerical model in random astern waves is reasonably
satisfactory and it is believed that this model could be a useful tool for identifying
dangerous situations that the ship might face in extreme random seas conditions.
The pure following seas conditions was seen to be most dangerous situation based on
the motion amplitudes and quartering to beam seas are safer compared to aft to
quartering seas.
Further efforts should be concerned with both theoretical and experimental
studies. For the calculation of the frequency dependent coefficients, due to the
ARTICLE IN PRESS
Hs=0.23 m; Tp=1.717 sec.

10
Sign. pitch (deg), single ampl.

9 Exp. (Fn=0.3)
8 Exp. (Fn=0.4)
7 Num. (Fn=0.3)
6 Num. (Fn=0.4)
5
4
3
2
1
0
0 10 20 30 40 50 60 70
χ (deg)
Fig. 36. Single amplitude significant pitch angle with respect to wave direction for Fn ¼ 0.3 and 0.4 in
JONSWAP spectrum.
inadequacy of the strip theory approaches, a 3D including the aft effect could
improve the accuracy of the numerical model. Also, shedding of trailing vortices
from hull sources should be taken into account and this could be achieved by using
CFD flow solvers. Following the importance of hydrodynamic lift and viscous
effects in ship motions in astern seas, CFD tools can also be employed using
boundary layer theory and Navier–Stokes equations. Furthermore, the coupling
between vertical and horizontal motions in the manoeuvring models could be fully
taken into account. As a result of current developments into new hull forms and
propulsion systems, the effect of design parameters for different hull forms at the
very high speed range and directional stability of these vessels can be investigated.
The effect of control parameters on vessel motions is also important. Use of more
sophisticated autopilot models could be attempted. Studies focusing on the
aforementioned aspects are currently underway.
For experimental research, the focus should be on the effect of vertical motions in
more extreme conditions, with more extensive environmental and operations
parameters. The same approach could be applied to random wave tests, in order to
develop insights into the behaviour of ship in a realistic following/quartering seaway.
Acknowledgments
The authors would like to thank Dr. Naoya Umeda of Osaka University, Mr.
Akihiko Matsuda and members of the National Research Institute of Fisheries
Engineering of Japan (NRIFE). for providing the opportunity to carry out the
experiments and technical support during first author’s stay in Japan. Thanks are
also due to Mr. David Clelland of NA-ME, Universities of Strathclyde and Glasgow
for his help in the design of the experiments. Finally, the authors would like to
express their sincere gratitude to the Ministry of Education of Japan for their
collaboration in funding the experimental studies at NRIFE.
ARTICLE IN PRESS
References
Abkowitz, M.A., 1969. Stability and Motion Control of Ocean Vehicles. MIT Press, Cambridge, MA,
USA.
Ankudinov, V., 1983. Simulation analysis of ship motions. In: The Proceedings of International
Workshop on Ship and Platform Motions, Berkeley, CA, USA, October 26–28, pp. 384–403.
Ayaz, Z., 2003. Manoeuvring behaviour of ships in extreme astern seas. Ph.D. Thesis, University of
Strathclyde, Glasgow, UK.
Ayaz, Z., Spyrou, K.J., Vassalos, D., 2001. An improved numerical model for the study of controlled ship
motions in extreme following and quartering seas. In: The Proceedings of IFAC Conference on
Control Applications in Marine Systems. Glasgow, UK, July.
Bailey, P.A., Price, W.G., Temarel, P., 1998. A unified mathematical model describing the manoeuvring of
a ship travelling in a seaway. Transactions of RINA 140, 131–149.
Chislett, M.S., 1990. The addition of a hell-roll servo mechanism to the DMI horizontal planar motion
mechanism. In: Proceedings of International Conference on Marine Simulation and Ship Manoeuvr-
ability. Tokyo, Japan, pp. 349–359.
Conolly, J.E., 1972. Stability and control in waves: a survey of the problem. Proceedings of International
Symposium on Directional Stability and Control of Bodies Moving In Water. Journal of Mechanical
Science 14(7), (Supplementary Issue), 186–193.
Cummins, W.E., 1962. The impulse response function and ship motions. Schiffstechnik 9 (47), 101–109.
Davidson, K.S.M., 1948. The steering of ships in following seas. In: Proceedings of Sixth International
Conference for Applied Mechanics, vol. 2, pp. 554–568.
De Kat, J., Paulling, J.R., 1989. The simulation of ship motions and capsizing in severe seas. SNAME
Transactions 97, 139–168.
Du Cane, P., Goodrich, G.J., 1962. The following sea, broaching and surging. RINA Quarterly
Transactions 104 (2), 109–140.
Grim, O., 1963. Surging Motion and Broaching Tendencies in A Severe Irregular Sea. Davidson
Laboratory Report, R-929, USA.
Hamamoto, M., Akiyoshi, T., 1988. Study on ship motions and capsizing in following seas (1st Report —
Equation of Motion for Numerical Simulation). Journal of The Society of Naval Architects of Japan
163, 173–180.
Hamamoto, M., Kim, Y., 1992. Study on ship motions and capsizing in following waves (Final Report).
Journal of The Society of Naval Architects of Japan 170, 173–182.
Hamamoto, M., Saito, K., 1992. Time domain analysis of ship motions in following waves. In: 11th
Australasian Fluid Mechanics Conference,. University of Tasmania. Hobart, Australia, pp. 355–358.
Hamamoto, M., Kim, Y., 1993. A new coordinate system and the equations describing manoeuvring
motion of a ship in waves. Journal of the Society of Naval Architectures of Japan 173, 209–220.
Hamamoto, M., Shirai, T., 1989. Study on ship motions and capsizing in following seas (2nd Report—
Simulation of Capsizing). Journal of The Society of Naval Architects of Japan 165, 123–130.
Hamamoto, M., Fujino, M., Kim, Y.S., 1994. Dynamic stability of a ship in quartering seas. In:
Proceedings of International Conference on Stability on Ships in Ocean Vehicles. Florida, USA.
Hamamoto, M., et al., 1995. Model experiments of ships capsize in astern seas (Second Report). Journal of
the Society of Naval Architects of Japan 179, 77–87.
Hamamoto, M., Sera, W., Panjaitan, J.P., 1996. Analyses on low cycle resonance of ships in irregular
astern Seas. Journal of the Society of Naval Architects of Japan 178, 137–145.
Hess, J.L., Smith, A.M.O., 1964. Calculation of nonlifting potential flow about arbitrary three-
dimensional bodies. Journal of Ship Research 8, 22–44.
Holtrop, J., Mennen, G.G.J., 1982. An approximate power prediction method. International Shipbuilding
Progress 29 (335), 166–170.
Inoue, S., Hirano, M., Kijima, K., Takashina, J., 1981. A practical calculation method of ship
manoeuvring. International Shipbuilding Progress 28 (324), 207–222.
ITTC, 1999a. Report of the manoeuvring Committee. Proceedings of 22nd ITTC, vol 1, Seoul, Korea, pp.
71–118.
ARTICLE IN PRESS
ITTC, 1999b. Report of The Specialist Committee on Stability. Proceedings of 22nd ITTC, vol 2, Seoul,
Korea, pp. 399–431.
ITTC, 2002. Report of The Specialist Committee on Prediction of Extreme Ship Motions and Capsizing.
Proceedings of 23rd ITTC, vol 2, Italy, pp. 611–649.
Jasionowski, A., 2001. An integrated approach to damage stability survivability assessment. Ph.D. Thesis,
University of Strathclyde, Glasgow, UK.
Lee, S.-K., 2000. The calculation of zig-zag manoeuvre in regular waves with use of the impulse response
functions. Ocean Engineering 27 (1), 87–96.
Letizia, L., 1996. Damage survivability of passenger ships in a seaway. Ph.D. Thesis, University of
Strathclyde, Glasgow, UK.
Matsuda, A., Hashimoto, H., Umeda, N., 2003. Capsizing due to Bow diving. In: Proceedings of 8th
International Conference on the Stability of Ships and Ocean Vehicles. Madrid, Spain, pp. 563–570.
McCreight, W.R., 1986. Ship manoeuvring in waves. In: Proceedings of Sixteenth Symposium on Naval
Hydrodynamics. Berkeley, CA, USA.
Motora, S., Fujino, M., Fuwa, T., 1982. On the mechanism of broaching-to phenomena. In: Proceedings
of International Conference of Ships and Ocean Vehicles. Tokyo, Japan.
Munif, A., Umeda, N., 2000. Modelling extreme roll motions and capsizing of moderate-speed ship in
astern waves. Journal of Naval Architects of Japan 187, 51–58.
Ohkusu, M., 1986. Prediction of wave forces on a ship running in following waves with very low
encountered frequency. Journal of the Society of Naval Architects of Japan 159, 129–138.
Rhee, K.P., Kim, C.K., Lee, C.M., 1990. The manoeuvring motion of the marad type ship in waves. In:
International Conference on Marine Simulation and Ship Manoeuvrability. Tokyo, Japan, pp.
413–419.
Spyrou, K.J., 1990. A new approach for assessing ship manoeuvrability based on dynamical systems
theory. Ph.D. Thesis, University of Strathclyde, Glasgow, UK.
Spyrou, K., 1995. Surf-riding and oscillations of a ship in quartering seas. Journal of Marine Science and
Technology 1, 24–36.
Spyrou, K., 1996. Dynamic instability in quartering seas: the behaviour of a ship during broaching.
Journal of Ship Research 40 (1), 46–59.
Tasai, F., 1961. Damping force and added mass of ships in heaving pitching. Transactions of the West
Japan Society of Naval Architects 21, 109–132.
Takaishi, Y., 1982. Consideration on the dangerous situations leading to capsize of ships in waves. In:
Proceedings of Second International Conference on Stability on Ships in Ocean Vehicles. Tokyo,
Japan, pp. 243–253.
Umeda, N., Renilson, M., 1992a. Broaching—a dynamic analysis of yaw behaviour of a vessel in a
following sea. In: Manoeuvrability and Control of Marine Craft. Computational Mechanics
Publications, Southampton, UK, pp. 533–543.
Umeda, N., Renilson, M., 1992b. Wave forces on a ship running in quartering seas—a simplified
calculation method. In: 11th Australasian Fluid Mechanics Conference. University of Tasmania,
Hobart, p. Australia.
Umeda, N., et al., 1994. Model experiments of ships capsize in astern seas. Journal of the Society of Naval
Architects of Japan 177, 207–217.
Umeda, N., Yamakoshi, Y., Suzuki, S., 1995. Experimental study for wave forces on a ship running
quartering seas with very low encountering frequency. In: , Proceedings of The Sevastianov
Symposium, The International Symposium Ship Safety in a Seaway,, vol 1(14). Kaliningrad, Russia,
pp. 1–18.
Vassalos, D., Maimun, A., 1994. Broaching-to: thirty years on. In: Proceedings of International
Conference on Stability on Ships in Ocean Vehicles. Florida, USA.
Vassalos, D., Ayaz, Z., Umeda, N., Renilson, M., 2003. Benchmark testing of numerical prediction on the
capsizing of intact container vessels in astern seas. In: Proceedings of RINA Conference. Design &
Operation of Container Ships, London, UK.

Mano Behaviour of Ships in Extreme Astern Seas

Încărcat de

Informații document

Descriere originală:

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Mano Behaviour of Ships in Extreme Astern Seas

Încărcat de

Drepturi de autor:

Formate disponibile

ARTICLE IN PRESS

Ocean Engineering 33 (2006) 2381–2434

Manoeuvring behaviour of ships in

Received 14 February 2005; accepted 25 October 2005

Corresponding author. Fax: +44 141 5484784.

a amplitude of wave (m)

RT(u) total resistance force (Nt)

w heading angle (deg)

z vertical position of ship (m)

X u_ surge acceleration coefﬁcient

upsurge in efforts to establish reliable numerical models to simulate manoeuvring of

to broaching, conﬁrmed, using numerical and experimental investigations, that

The numerical model incorporates non-linear six-degrees-of-freedom coupled

where x0 , y0 , z0 are kinematics, d denotes rudder angle xG and za represent horizontal

2.1. External forces

2.1.1. Wave forces

2.1.1.1. Froude– Krylov including hydrostatic forces on a ship. Both experimental

direction in following and quartering seas can be signiﬁcant, especially, for

2.2. Manoeuvring (hull) forces

As it is given previously, manoeuvring and seakeeping usually deal with differently

2.3. Rudder forces

As an important element of the excitation for ship motions in following and

2.4. Propulsive forces

The calculation of propulsive forces therefore is dependent upon accurate

2.5. Resistance forces

2.6. Autopilot control

Although, automatic control systems can be regarded as parameter of ship

3. Impulse response functions (memory effects)

Following the research background on the effect of frequency dependent

3.1. Numerical solution of Kernel functions

Impulse response function

where B(o) is damping coefﬁcient, do is frequency range. Note that the ot

3.2. Implementation of convolution terms

Frequency dependent excitation forces (radiation) are calculated by using

4. Veriﬁcation of numerical model

In order to validate the numerical code, a time-domain simulation program

Fig. 3. Lines of the container vessel.

Parameter Vessel Model (1/60)

LBP 150 m 2.5 m

Nominal Froude number Fn 0.2 0.2 0.4

Roll angle (deg)

Roll angle (deg)

4.1. Discussion on results

Roll angle (deg)

Roll angle (deg)

Roll angle (deg) 0

Roll angle (deg)

Fig. 10. Body plan of the Purse Seiner.

5. Benchmark study of ITTC vessel

5.1. Captive Model Tests

Parameter Vessel Model (1/17.5)

Fig. 11. Coordinate system of model basin (Umeda et al., 1995).

ðM þ AÞX€ þ B ðX ÞX_ þ C ðX ÞX ¼ F 00 ðzw Þ þ F 01 ðzw ; X Þ þ F 02 ðzw ; X_ Þ

H/l Fn w (deg) Sinkage (m) y (deg) f(deg)

0 45 60 0.2 0 0.2 1.43 0 1.43 10 0 10

5.1.1. Discussion on captive model test results

Overall, the numerical model gives reasonably satisfactory agreement with

5.2. Free running model tests

5.2.1. Numerical method for random waves

5.2.2. Discussion on free running model tests

Parameter Vessel Model (1/15)

LBP 34.50 m 2.3 m

Roll Angle (deg) 10