On Unstable Ship Motions Resulting From Strong Non-Linear Coupling

ARTICLE IN PRESS
Ocean Engineering 33 (2006) 1853–1883

www.elsevier.com/locate/oceaneng
On unstable ship motions resulting from strong

non-linear coupling
Marcelo A.S. Neves, Claudio A. Rodrı́guez
Department of Naval Architecture and Ocean Engineering, LabOceano/COPPE,
Federal University of Rio de Janeiro, C.P. 68.508, Rio de Janeiro, 21.945-970, Brazil
Received 4 April 2005; accepted 23 November 2005
Available online 27 March 2006
Abstract
In this paper, the modelling of strong parametric resonance in head seas is investigated.
Non-linear equations of ship motions in waves describing the couplings between heave, roll
and pitch are contemplated. A third-order mathematical model is introduced, aimed at
describing strong parametric excitation associated with cyclic changes of the ship restoring
characteristics. A derivative model is employed to describe the coupled restoring actions up to
third order. Non-linear coupling coefficients are analytically derived in terms of hull form
characteristics.
The main theoretical aspects of the new model are discussed. Numerical simulations
obtained from the derived third-order non-linear mathematical model are compared to
experimental results, corresponding to excessive motions of the model of a transom stern
fishing vessel in head seas. It is shown that this enhanced model gives very realistic results and
a much better comparison with the experiments than a second-order model.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Ship stability; Parametric resonance; Non-linear equations; Ship motions; Roll motion; Hill
equation
Corresponding author. Tel.: +55 21 2562 8715; fax: +55 21 2562 8715.
E-mail address: masn@peno.coppe.ufrj.br (M.A.S. Neves).
0029-8018/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.oceaneng.2005.11.009
ARTICLE IN PRESS
1854 M.A.S. Neves, C.A. Rodrı´guez / Ocean Engineering 33 (2006) 1853–1883
Nomenclature
r incremental volume for displaced hull

r density of water
f roll angular displacement
y pitch angular displacement
r0 volume at average hull position
r1 volume at instantaneous hull position
A0 waterplane area at average hull position
Aw wave amplitude
Fn Froude number
g acceleration of gravity
hi height of elemental prisms
Ixx0 transversal second moment of waterplane area
Iyy0 longitudinal second moment of waterplane area
Jxx transversal mass moment of inertia
Jyy longitudinal mass moment of inertia
k wave number
m ship mass
U ship speed of advance
xf0 longitudinal co-ordinate of centroid of waterplane
z heave displacement of the ship
l wavelength
z wave elevation
w wave incidence
oe encounter frequency
ow wave frequency
^ J;
I; ^ K^ unit vectors along axes of inertial frame
i;^ j;^ k^ unit vectors along axes fixed in the ship
x̄ longitudinal position of a transversal station
ȳ half-beam of a transversal station
z̄G vertical position of the ship’s centre of gravity
z̄B0 vertical position of hull volume centroid
1. Introduction
Any comprehensive investigation on the safety of intact vessels in waves must take
into consideration the possible occurrence of parametric resonance. In fact,
parametric rolling of ships has continuously received wide attention of researchers
and designers, since it is a relevant instabilizing mechanism, see Kerwin (1955),
Paulling and Rosenberg (1959), Blocki (1980), De Kat and Paulling (1989), Munif
and Umeda (2000). Much of such attention has been devoted to the particular
configuration of longitudinal regular waves, either with or without speed, bow or
ARTICLE IN PRESS
M.A.S. Neves, C.A. Rodrı´guez / Ocean Engineering 33 (2006) 1853–1883 1855
stern waves. Roll motion in such conditions has usually been modelled as an
uncoupled Mathieu type equation. Considering the well-known existence of the
Mathieu resonant frequencies, focus has been concentrated on the first region of
instability, the one defined by the proximity of encounter frequency to twice the roll
natural frequency.
For many years, more attention was given to parametric rolling in astern seas,
Oakley et al. (1974), De Kat and Paulling (1989), Umeda et al. (1995), Spyrou
(2000). Yet, more recently, some authors and accidents have called attention to the
problem of parametric excitation in head seas. France et al. (2003) reported on
strong roll amplification in the case of a large container ship, which suffered severe
damage to containers and structure due to excessive accelerations. Dallinga et al.
(1998), Luth and Dallinga (1999) reported on the development of head seas
parametric resonance in cruise vessels. Levadou and Palazzi (2003), recognizing the
potentially dangerous situation, attempted to evaluate the operational risks
associated with head seas parametric resonance. Another example of parametric
resonance in head seas has been recently reported by Palmquist and Nygren (2004).
A large experimental program with two similar models of fishing vessels on
parametric resonance in head seas at different speeds and loading conditions was
conducted some years ago, see Neves et al. (2002). The experiments showed that one
of the tested vessels, a transom stern hull, is very sensitive to parametric excitation in
the first region of instability. In particular, when tested in high waves and low
metacentric height, the transom stern hull tended to undergo very strong roll
amplifications at different speeds.
Employing Taylor series expansions up to second order, Neves and Valerio (2000),
Neves (2002) expressed restoring actions in the heave, roll and pitch modes in a
coupled way. Wave action was taken into consideration not only in the Froude-
Krilov plus diffraction first-order forcing functions, but also in second-order terms
resulting from volumetric changes of the submerged hull due to wave passage effects.
It can be shown that a system of second-order non-linear equations as defined
above may be reduced to a set of coupled Mathieu equations when the linear
variational equation is taken, Neves (2002). The set of coupled Mathieu equations
then describes the essential aspects of the stability of the dynamic system when small
perturbations are imposed on the basic linear periodic motions. Neves et al. (2003)
suggested that a Hill equation should be employed, instead of the Mathieu equation,
in order to describe the stability of parametrically excited motions.
For many ship designs the simulation models available are capable of reproducing
with confidence the roll amplifications resulting from parametric resonance. But,
unfortunately, there are some known cases in which the numerical models tend to
over-predict the resonant rolling motions observed in experiments, as pointed out by
Umeda et al. (2003). These strong amplifications are associated with specific values
of metacentric height, ship speed and stern shape (and, in general, other design
parameters). In these cases, the classical Mathieu type modelling, in which
parametric excitation is assessed considering terms up to the second order, tends
to predict excessive excitation. The Authors attempts to reproduce the strong
amplifications observed in the tests with the transom stern hull employing a
ARTICLE IN PRESS
second-order non-linear mathematical model also indicated excessive amplifications

in the simulations.
Bulian et al. (2003), Umeda et al. (2003), Matuziak (2003), are recent examples of
investigations in which some form of non-linearities have been heuristically
introduced as a possible remedy to the deficiencies encountered in matching
experimental results with large amplifications. Skomedal (1982), Shin et al. (2003)
report on numerical assessments of strong parametric amplifications, but with no
comparisons with experiments.
In this paper, the modelling of strong parametric resonance in head seas is
investigated. In an attempt to give a robust answer to this question and improve the
quality of the simulations, a new third-order mathematical model is introduced. The
model describes the heave, roll and pitch motions in a coupled way—up to third
order, thus establishing a mathematically congruent set of equations. Parametric
excitation is associated with cyclic changes in the ship restoring characteristics,
corresponding to variations in the submerged part of the hull due to the relative
body motions with respect to the wave motion along the hull. A holistic derivative
model is employed to describe the coupled restoring actions up to third order.
Given the extended complexities of the coupled non-linear systems, some effort is
devoted to interpreting the essential dynamic characteristics of the new mathematical
model. The appearance of super-harmonics and increased stiffness proportional to
wave amplitude squared due to third-order terms is highlighted. Non-linear coupling
coefficients are analytically derived in terms of hull form characteristics. In
particular, the longitudinal distributions of breadth and flare are derived as essential
constituents of the coefficients defining the complex dynamics. This aspect may be of
great interest and relevance in the ship design practice.
In the present investigation, numerical simulations obtained from the derived
third-order non-linear mathematical model are compared to experimental results
corresponding to excessive motions of a transom stern fishing vessel in head seas. It
is shown that this enhanced model more closely matches results from experiments
than a second-order model.
2. Equations of motion
Two right-handed co-ordinate systems are employed to describe the motions. An

inertial reference frame (C,x,y,z) is assumed to be fixed at the mean ship motion,
defined by the ship speed U. Regular waves are assumed to travel forming an angle w
with ship course. Another reference frame ðO; x̄; ȳ; z̄Þ is fixed at the ship having the xy
plane coinciding, for the ship at rest, with the undisturbed sea surface, z̄-axis passing
through the vertical that contains the centre of gravity. The two systems coincide
when excitations are absent. See Fig. 1.
Non-linear equations of motion considering the three restoring degrees of freedom
may be expressed in matrix form using a displacement vector:
T
sðtÞ ¼ zðtÞ fðtÞ yðtÞ
~
ARTICLE IN PRESS
Fig. 1. Co-ordinate axis and definition of the six degrees of motions.
defining the heave translational motion and the roll and pitch angular modes,
indicated in Fig. 1:
ðM ~~
~ þ AÞ _~
~ fÞ
s€ þ Bð ~res ð~
s_ þ C s; zÞ ¼ C _ zÞ.
~ext ðz; z; € (1)
~
Hull inertia M is a diagonal 3 3 matrix. Its elements are: m, the ship mass, Jxx,
Jyy the mass moments of inertia in the roll and pitch modes, respectively, taken with
reference to the chosen origin. Elements in matrix A~ represent hydrodynamic
generalized added masses. Following the reasoning of Abkowitz (1969), these
hydrodynamic reactions will be taken as linear. Damping terms Bð _ describe
~ fÞ
hydrodynamic reactions dependent on ship velocities, and may incorporate non-
linear terms in the roll equation. Vector C ~res ð~
s; zÞ describes non-linear restoring
forces and moments dependent on the relative motions between ship hull and wave
elevation z(t). On the right-hand side of Eq. (1), the generalized vector C _ zÞ
~ext ðz; z; €
represents wave external excitation, usually referred to in the literature as the
Froude-Krilov plus diffraction wave forcing terms, dependent on wave heading w,
encounter frequency oe, wave amplitude Aw and time t. In component form, the
heave force and roll and pitch moments will be defined as
_ zÞ
~ext ðz; z;
C € ¼ ½Z W ðtÞ K W ðtÞ M W ðtÞ T .
It is observed that for longitudinal waves, there is no roll external excitation:
K W ðtÞ ¼ 0.
3. Hydrodynamic coefficients associated with potential theory
The hydrodynamic inertia and damping matrices are expressed respectively, as

2 3 2 3
Z z€ 0 Z y€ Z z_ 0 Z y_
6 0 7 6 _ 0 7
A~ ¼ 4 0 K f€ 5; B~ ¼ 4 0 K f_ ðfÞ 5.
M z€ 0 M y€ M z_ 0 M y_
ARTICLE IN PRESS
_ all the other terms in

It is generally accepted that, with the exception of K f_ ðfÞ,
~
matrices A and B~ may be evaluated by means of potential theory, as described, for
instance, in Salvesen et al. (1970), Meyers et al. (1975). The uncoupled nature of
hydrodynamic reactions in the roll, on one side, and the heave and pitch modes on
the other, is evident from the structure of the A~ and B~ matrices.
4. Non-linear roll damping
Due to the occurrence of strong viscous effects, the roll damping moment can not
be satisfactorily computed by means of potential theory. The semi-empirical model
of roll damping moment due to Ikeda, see Himeno (1981), may be used to estimate
the main linear and non-linear viscous effects. In a modular way, the model assumes
that the total roll damping may be subdivided into five main components, each one
being computed separately:
_ f_ ¼ BW þ BF þ BE þ BL þ BBK ,
K f_ ðfÞ
where BW, BF, BE, BL and BBK correspond to wave, friction, eddy, lift and bilge keel
damping, respectively.
Damping due to the bilge keels may also be split into components:
BBK ¼ BBKN þ BBKH þ BBKW ,
where BBKN, BBKH and BBKW are contributions due to normal force, interaction
between hull and bilge keel and wave system generated by the bilge keels,
respectively.
Mathematically, the roll damping moment may be represented to third order as
3
K f_ ðfÞ _
_ f_ ¼ K _ f _
f þ Kf
_f_ f_ f
or, in a more convenient (and equivalent) form, see Bass and Haddara (1988):
K f_ ðfÞ _
_ f_ ¼ K _ f _ _
f þ Kf
_ jf_ j fjfj. (2)
Coefficients K f_ and K fj _ may be computed using expressions given in Himeno (1981).
_ fj
5. Non-linear restoring actions
We consider now the force and moments dependent on position. To third order,
positional terms due to the combined actions of ship motions in calm water and wave
elevation along the hull may be formally expressed in terms of a Taylor series
expansion about the average position. For this purpose, it is convenient to define a
generalized vector ~ q ¼ ½~ s; zT such that the positional actions are:

X4
q ~pos
C 1 X4 X 4
q 2~
C pos

1 X4 X 4 X 4
q 3~
C pos

~pos ¼
C qi þ qi qj þ qi qj qk :
i¼1
qq i 2 i¼1 j¼1
qq i qq j 6 i¼1 j¼1 k¼1
@q i qq j qq k
0 0 0
(3)
ARTICLE IN PRESS
Wave elevation, according to the Airy linear theory, see Newman (1977), may be
defined as
zðx; y; t; wÞ ¼ Aw cos½kx cosðwÞ þ ky sinðwÞ oe t,
where Aw is the wave amplitude, k the wave number, defined as k ¼ o2w =g ¼ 2p=l,
ow the wave frequency, l the wave length, g the acceleration of gravity, w the wave
heading (incidence), oe the encounter frequency, defined as oe ¼ ow kU cos w.
In longitudinal waves, head seas, the equation of wave surface elevation is
zðx; tÞ ¼ Aw cos½kx þ oe t. (4)
It is important to note in Eq. (3) that the first-, second- and third-order terms
independent of ~ s ¼ ½ z f y T may be more appropriately accommodated on the
right-hand side of Eq. (1). Following the formalism of Eq. (3):

q ~pos
C 1 q 2~
C pos

1 q 3~
C pos

3
~extðFKÞ ðzÞ ¼
C zþ z 2
þ z .
qz 2 qz2 6 qz3
0 0 0
These terms represent linear and non-linear Froude-Krilov force/moments

excitations, to be included in the forcing vector C _ zÞ
~ext ðz; z; € appearing on the right-
hand side of Eq. (1), in which are cast all the terms independent of the ship motions
vector ~s.
The vector C ~res , representing non-linear restoring actions, will be expressed as
~res ¼ ½ZðtÞ
C ~pos ðz; f; y; zÞ C
KðtÞ MðtÞ T ¼ C ~extðFKÞ ðzÞ. (5)
It is clear that the heave, roll and pitch motions will be completely coupled to
each other.
We start to introduce the derivative nomenclature to be used onwards. In the
Taylor expansions, as an example, below we show all the heave coefficients, where
the notation ‘‘j0’’ used to represent derivatives of forces and moments at average
equilibrium position was abandoned:
First order:
qZ qZ qZ
¼ Zz ; ¼ Zf ; ¼ Zy . (6)
qz qf qy
Second order:
q2 Z q2 Z q2 Z
¼ Z zz ; ¼ Z ff ; ¼ Zyy ,
qz2 qf2 qy2
q2 Z q2 Z q2 Z q2 Z
¼ ¼ Zzf ; ¼ ¼ Zzy ,
qzqf qfqz qzqy qyqz
q2 Z q2 Z
¼ ¼ Zfy . ð7Þ
qfqy qyqf
ARTICLE IN PRESS
Third order:
q3 Z q3 Z q3 Z
¼ Z zzz ; ¼ Z fff ; ¼ Zyyy ,
qz3 qf3 qy3
q3 Z q3 Z q3 Z
¼ ¼ ¼ Zzzf ,
qz2 qf qf qz2 qz qfqz
q3 Z q3 Z q3 Z
¼ ¼ ¼ Z zzy ,
qz2 qy qy qz2 qz qy qz
q3 Z q3 Z q3 Z
2 ¼ ¼ ¼ Z ffz ,
qf qz qz qf2 qf qz qf
q3 Z q3 Z q3 Z
2
¼ 2
¼ ¼ Z ffy ,
qf qy qy qf qf qy qf
q3 Z q3 Z q3 Z
¼ ¼ ¼ Z yyz ,
qy2 qz qz qy2 qy qz qy
q3 Z q3 Z q3 Z
2 ¼ ¼ ¼ Zyyf ,
qy qf qf qy2 qy qf qy
q3 Z q3 Z q3 Z q3 Z
¼ ¼ ¼
qz qf qy qf qz qy qy qz qf qf qy qz
q3 Z q3 Z
¼ ¼ ¼ Z zfy . ð8Þ
qy qf qz qz qy qf
6. Restoring coefficients in calm water
Considering the physics of the fluid-body interaction, see Paulling and Rosenberg
(1959), it is evident that the dependencies of the heave and pitch coefficients on roll
are even; and the coefficients in roll due to heave and pitch are odd; it follows that
some of the coefficients in the Taylor expansions are nil:
Zf ¼ Zzf ¼ Z fy ¼ Z fff ¼ Z zzf ¼ Z yyf ¼ Z zfy ¼ 0,
K z ¼ K y ¼ K zz ¼ K ff ¼ K yy ¼ K zy ¼ K zzz ¼ K yyy ¼ K zzy ¼ K ffz
¼ K ffy ¼ K yyz ¼ 0,
M f ¼ M zf ¼ M fy ¼ M fff ¼ M zzf ¼ M yyf ¼ M zfy ¼ 0. ð9Þ
A general derivation (up to the third order) of all restoring terms is given in
Appendix A. Tables 1a–1c shows all the calm water geometric derivatives given in
terms of relevant hull geometric characteristics.
7. Interactions between non-linear motions and incident wave
Wave effects due to incident waves of arbitrary direction along the hull are
modelled as a change of the hull average submerged shape defined by the
ARTICLE IN PRESS
instantaneous position of the wave. Contributions independent of the ship motions,

_ zÞ
~ext ðz; z;
contained in the C € forcing term, as is usually adopted, appear on the right-
hand side of Eq. (1), such that:
_ zÞ
~ext ðz; z;
C € ¼C
~extðFKÞ ðzÞ þ C _ zÞ
~extðDifÞ ðz; € ¼ ½Z W ðtÞ K W ðtÞ M W ðtÞ T .
Higher order diffraction force and moments are assumed not to be relevant; for
this reason, in the present mathematical model, diffraction forces/moments are
defined as being proportional to the first-order wave motion.
Non-linear effects due to body–wave interactions proportional to second and
third-order hull-wave relative displacements, due to their mathematical affinity with
the purely hydrostatic terms, are considered on the left-hand side of the equations of
motion, Eq. (5), being named wave passage effects (on the parametric excitation).
Employing similar nomenclature and procedures to those used for the establish-
ment of the hydrostatic terms, and considering that the coefficients corresponding to
body/incident-wave relative motion follow the same logics of Eq. (9), the following
interaction derivatives are found to be zero:
Z zf ¼ Zzzz ¼ Z zzz ¼ Zzzf ¼ Z zzf ¼ Z yyz ¼ Z zzy ¼ Z zzy ¼ Z zfy ¼ 0,
K zz ¼ K zy ¼ K zzz ¼ K zzz ¼ K zff ¼ K zyy ¼ K zzy ¼ K zzy ¼ 0,
M zf ¼ M zzz ¼ M zzz ¼ M zzf ¼ M zzf ¼ M yyz ¼ M zzy ¼ M zzy ¼ M zfy ¼ 0.
ð10Þ
Consequently, the non-zero derivatives are obtained as defined in Tables 2a and b
in terms of hull and incident wave characteristics. It may be observed that, up to
Table 1a
Hydrostatic restoring coefficients (calm water)—Linear
Heave Roll Pitch
Zz ¼ rgA0 Kz ¼ 0 M z ¼ rgA0 xf 0
Zf ¼ 0 K f ¼ rg½r 0 ðz̄B0 z̄G Þ þ I xx0 Mf ¼ 0
Zy ¼ rgA0 xf 0 Ky ¼ 0 M y ¼ rg½r 0 ðz̄B0 z̄G Þ þ I yy0
Table 1b
Hydrostatic restoring coefficients (calm water)—Second order
Heave Roll Pitch

R R
Zzz ¼ 2rg L qȳ
qz̄ dx
K zz ¼ 0 M zz ¼ 2rg L x̄ qȳ
qz̄ dx
R 2 qȳ
Zzf ¼ 0 K zf ¼ 2rg M zf ¼ 0
L ȳ qz̄ dx
R R
Zzy ¼ 2rg L x̄ qȳ dx K zy ¼ 0 M zy ¼ 2rg L x̄2 qȳqz̄ dx
R qz̄2 qȳ R 2 qȳ
Zff ¼ 2rg L ȳ qz̄ dx K ff ¼ 0 M ff ¼ 2rg L x̄ȳ qz̄ dx
R qȳ
Zfy ¼ 0 K fy ¼ 2rg xȳ2 dx M fy ¼ 0
L qz̄
R R
Zyy ¼ 2rg L x̄2 qȳqz̄ dx
K yy ¼ 0 M yy ¼ 2rg L x̄3 qȳ
qz̄ dx
1862
Table 1c
Hydrostatic restoring coefficients (calm water)—Third order
Heave
a a
Zzzz ¼ 0 Z zzf ¼ 0 Zzzy ¼ 0
2
R 2 Z fff ¼ 0 R
Zffz ¼ rg 4 L ȳ qȳ
qz̄ dx þ A 0 Zffy ¼ rg 4 L x̄ȳ qȳ
qz̄ dx þ A 0 x f 0
a a
Zyyz ¼ 0 Z yyf ¼ 0 Zyyy ¼ 0
Roll

K zzz ¼ 0 R 2 K zzy ¼ 0
K zzf ¼ rg 4 L ȳ qȳ
qz̄ dx þ A0

K ffz ¼ 0 R 3 qȳ2 K ffy ¼ 0
K fff ¼ rg 8 L ȳ qz̄ dx þ 2I xx0
2
K yyz ¼ 0 R K yyy ¼ 0
K yyf ¼ rg 4 L x̄2 ȳ qȳ
qz̄ dx þ I yy0
Pitch
a a
M zzz ¼ 0 M zzf ¼ 0 M zzy ¼ 0
2 2
R M fff ¼ 0 R
M ffz ¼ rg 4 L x̄ȳ qȳ
qz̄ dx þ A x
0 f0 M ffy ¼ rg 4 L x̄2 ȳ qȳ
qz̄ dx þ I yy0
ARTICLE IN PRESS
a a
M yyz ¼ 0 M yyf ¼ 0 M yyy ¼ 0
Heave-roll-pitch coupling
2
Zzfy ¼ 0 R M zfy ¼ 0
K zfy ¼ rg 4 L x̄ȳ qȳ
qz̄ dx þ A0 x f 0
a
These expressions were obtained analytically for the case of a ship with inclined wall side, corresponding to a good approximation in the case of ships of
M.A.S. Neves, C.A. Rodrı´guez / Ocean Engineering 33 (2006) 1853–1883
conventional forms, small displacements and smooth transversal curvatures ðq2 ȳ=qz̄2 ! 0Þ at the considered water-line.
ARTICLE IN PRESS
Table 2a
Derivatives due to wave passage—Second order
Heave Roll Pitch

R qȳ R
Zzz ðtÞ ¼ 2rg L qz̄ z dx K zz ðtÞ ¼ 0 M zz ðtÞ ¼ 2rg L x̄ qȳ
qz̄ z dx
R 2 qȳ
Zzf ðtÞ ¼ 0 K zf ðtÞ ¼ 2rg M zf ðtÞ ¼ 0
L ȳ qz̄ z dx
R qȳ R
Zzy ðtÞ ¼ 2rg L x̄ qz̄ z dx K zy ðtÞ ¼ 0 M zy ðtÞ ¼ 2rg L x̄2 qȳ
qz̄ z dx
second order, the coefficients in heave, roll and pitch are identical to those employed
in Neves and Valerio (2000), see Table 2a.
It is clear that the above coefficients are dependent on hull characteristics and on
wave amplitude, frequency and time. These coefficients may be expressed in terms of
their cosine and sine terms, such that their dependence on wave amplitude becomes
explicit. Thus, for example, in the case of roll motion:
K zf ðtÞ ¼ Aw K zfc cosðoe tÞ þ Aw K zfs sinðoe tÞ,

K zzf ðtÞ ¼ A2w K zzfc cosð2oe tÞ þ A2w K zzfs sinð2oe tÞ þ A2w K zzf0 ,
K zzf ðtÞ ¼ Aw K zzfc cosðoe tÞ þ Aw K zzfs sinðoe tÞ,
K zfy ðtÞ ¼ Aw K zfyc cosðoe tÞ þ Aw K zfys sinðoe tÞ, ð11Þ
where the newly defined coefficients, independent of wave amplitude and time, are as
given in Table 3.
These coefficients are dependent on wave frequency and hull geometry. Similar
derivations were presented by Paulling (1961), but with some simplifications in the
third-order terms. The present derivation, based on the instantaneous wave surface
intercepting the hull, aggregates the various integral terms involving the longitudinal
distribution of ȳðqȳ=qz̄Þ2 in Table 3.
8. Non-linear heave–roll–pitch equations of motions
Taking into account the developments of sessions 3 to 7, Eq. (1) may be expressed
in the form of the following non-linear set of coupled equations for the heave, roll
and pitch modes, valid for longitudinal waves:
1 1 1
ðm þ Z z€ Þ€z þ Zz_ z_ þ Z y€ y€ þ Zy_ y_ þ Zz z þ Zy y þ Z zz z2 þ Zff f2 þ Z yy y2
2 2 2
1 1 1 1 1
þ Z zy zy þ Z zzz z þ Z zzy z y þ Z ffz f z þ Z ffy f y þ Z yyz y2 z
3 2 2 2
6 2 2 2 2
1 3
þ Z yyy y þ Z zz ðtÞz þ Z zy ðtÞy þ Z zzz ðtÞz þ Z zzz ðtÞz2 þ Z zzy ðtÞy
6
þ Z zzy ðtÞzy þ Z ffz ðtÞf2 þ Z yyz ðtÞy2 ¼ Zw ðtÞ,
1864
Table 2b
Derivatives due to wave passage—Third order
Heave
Zzzz ðtÞ ¼ 0a Zzzf ðtÞ ¼ 0 Zzzy ðtÞ ¼ 0 a
Zzzz ðtÞ ¼ 0 a Zzzf ðtÞ ¼ 0 Zzzy ðtÞ ¼ 0 a

2

R Zyyz ðtÞ ¼ 0 a Zzfy ðtÞ ¼ 0
Zffz ðtÞ ¼ rg L 2ȳ qȳ
qz̄ þ ȳ z dx
Roll

2

K zzz ðtÞ ¼ 0 R K zzy ðtÞ ¼ 0
K zzf ðtÞ ¼ rg 2ȳ qȳ
L qz̄ þ ȳ z2 dx

2

K zzz ðtÞ ¼ 0 R K zzy ðtÞ ¼ 0
K zzf ðtÞ ¼ rg L 4ȳ qȳqz̄ þ 2ȳ z dx
2
K ffz ðtÞ ¼ 0 K yyz ðtÞ ¼ 0 R qȳ
K zfy ðtÞ ¼ rg L 4x̄ȳ qz̄ þ 2x̄ȳ z dx
Pitch
a
M zzz ðtÞ ¼ 0 M zzf ðtÞ ¼ 0 M zzy ðtÞ ¼ 0 a
a
M zzz ðtÞ ¼ 0 M zzf ðtÞ ¼ 0 M zzy ðtÞ ¼ 0 a
ARTICLE IN PRESS
2
R M yyz ðtÞ ¼ 0 a M zfy ðtÞ ¼ 0
M ffz ðtÞ ¼ rg L 2x̄ȳ qȳ
qz̄ þ x̄ȳ z dx
a
Same hypothesis are applied, as in the case of geometric derivatives in calm water.
M.A.S. Neves, C.A. Rodrı´guez / Ocean Engineering 33 (2006) 1853–1883
ARTICLE IN PRESS
Table 3
Coefficients of wave passage in the roll equation
R R
K zfc ¼ 2rg L ȳ2 qȳ cosðkxÞ dx K zfs ¼ 2rg L ȳ2 qȳ sinðkxÞ dx
qz̄ qz̄
R qȳ
2 R qȳ
2
K zzfc ¼ rg L ȳ qz̄ þ 12 ȳ cosð2kxÞ dx K zzfs ¼ rg L ȳ qz̄ þ 12 ȳ sinð2kxÞ dx

R 2
K zzf0 ¼ rg L ȳ qȳ
qz̄ þ 1
2 ȳ dx

R 2 R 2
K zzfc ¼ rg L 4ȳ qȳ qz̄ þ 2ȳ cosðkxÞ dx K zzfs ¼ rg L 4ȳ qȳ
qz̄ þ 2ȳ sinðkxÞ dx
R h @ȳ 2 i
R 2
K zfyc ¼ rg L 4x̄ȳ qȳ þ 2x̄ȳ cosðkxÞ dx K zfys ¼ rg L 4x̄ȳ @z̄ þ 2x̄ȳ sinðkxÞ dx
qz̄
1
ðJ xx þ K f€ Þf€ þ K f_ f_ þ K _ _ fj _ _
fjfj fj þ K f f þ K zf zf þ K fy fy þ K zzf z f
2
2
1 1
þ K fff f3 þ K yyf y2 f þ K zfy zfy þ K zf ðtÞf
6 2
þ K zzf ðtÞf þ K zzf ðtÞzf þ K zfy ðtÞfy ¼ 0;
1 1
ðJ yy þ M y€ Þy€ þ M y_ y_ þ M z€ z€ þ M z_ z_ þ M z z þ M y y þ M zz z2 þ M ff f2
2 2
1 1 1 1 1
þ M yy y2 þ M zy zy þ M zzz z3 þ M zzy z2 y þ M ffz f2 z þ M ffy f2 y
2 6 2 2 2
1 1
þ M yyz y2 z þ M yyy y3 þ M zz ðtÞz þ M zy ðtÞy þ M zzz ðtÞz
2 6
þ M zzz ðtÞz2 þ M zzy ðtÞy þ M zzy ðtÞzy þ M ffz ðtÞf2
þ M yyz ðtÞy2 ¼ M w ðtÞ. ð12Þ
Analytic expressions for the determination of all—linear and non-linear–deriva-
tives have been given in Tables 1–3.
It is pointed out that in general parametric resonance is modelled in the literature
considering uncoupled versions of the roll equation, with:
(a) vertical motions, z(t) and y(t), assumed to be purely harmonic,

(b) roll parametric excitation defined with non-linearities up to the second order,
and
(c) different levels of non-linearities in the calm water GZðfÞ curve,
as discussed in Blocki (1980), Neves et al. (1999), Spyrou (2000), Umeda et al. (2003),
Bulian et al. (2003).
In the nomenclature adopted in Eq. (12), such a simplified model of parametric
rolling would read as
_ þ K _ _ fj
ðJ xx þ K f€ Þf€ þ K f_ f _ _
fjfj fj þ K f f þ ½K zf z þ K fy y þ K zf ðtÞf
1
þ K fff f3 ¼ 0, ð13Þ
6
ARTICLE IN PRESS
where the three terms within brackets represent contributions to parametric

excitation (to second order) due to the heave and pitch motions and to volumetric
changes in the submerged hull produced by the wave passage, respectively. Clearly,
these contributions are proportional to the wave amplitude. Eq. (13) is sometimes
referred as Mathieu–Duffing equation.
9. Theoretical analysis of perturbed roll motions
Stability of motion of the non-linear set of equations may be assessed by means of

the variational system, Cesari (1971). In its linear form it may be derived under the
assumption that the non-linear motions may be decomposed as the sum of steady
oscillatory solutions plus some small perturbations:
zðtÞ ¼ z^ðtÞ þ xðtÞ ¼ Aw Z3 cosðoe t þ az Þ þ xðtÞ,

^ þ jðtÞ ¼ Aw Z cosðoe t þ af Þ þ jðtÞ,
fðtÞ ¼ fðtÞ 4
^ þ WðtÞ ¼ Aw Z cosðoe t þ ay Þ þ WðtÞ,
yðtÞ ¼ yðtÞ 5
^
where z^ðtÞ, fðtÞ ^
and yðtÞ correspond to the heave, roll and pitch well known
linear solutions, and Z3 ,Z4 ,Z5 are the corresponding transfer functions. Pertur-
bations in the heave, roll and pitch modes are defined as xðtÞ, jðtÞ, and WðtÞ,
respectively.
The linear variational equation of the roll motion associated with Eq. (12) is then
derived as
_ þ K f j þ ðK zf f^ þ K zzf z^f^ þ K zfy f
ðJ xx þ K f€ Þj€ þ K f_ j ^ yÞx
^

^ 1 ^ 2 1 2 1 ^ 2
^
þ K zf z^ þ K fy y þ K fff f þ K zzf z^ þ K yyf y þ K zfy z^y j
2 2 2
^ ^ ^ ^
þ ðK fy f þ K yyf fy þ K zfy z^fÞW þ K zf ðtÞj þ K zzf ðtÞfx ^ þ K zzf ðtÞ^zj
^ þ K zfy ðtÞfW
þ K zfy ðtÞyj ^ þ K zzf ðtÞj ¼ 0. ð14Þ
In the particular case of longitudinal waves, the roll linear solution is zero. That is,
^ 0. Hence:
f
_ þ K f j þ ½K zf z^ þ K fy y^ þ K zf ðtÞj
ðJ xx þ K f€ Þj€ þ K f_ j

1 1 2
þ K zzf z^2 þ K yyf y^ þ K zfy z^y^ j
2 2
þ ½K zzf ðtÞ^z þ K zfy ðtÞy^ þ K zzf ðtÞj ¼ 0, ð15Þ
in which it is possible to identify the distinct third-order contributions to parametric
excitation, from purely hydrostatic actions and wave interaction effects, respectively.
The three terms inside the third bracket, corresponding to third-order Froude-Krilov
moments, are given in Eq. (11) and Table 3.
ARTICLE IN PRESS
For the sake of comparison, it is noted here that the linear variational equation
derived from Eq. (13) is in the form of a Mathieu equation:
ðJ xx þ K f€ Þj€ þ K f_ j
_ þ K f j þ ½K zf z þ K fy y þ K zf ðtÞj ¼ 0. (16)
Returning to the examination of Eq. (15): using a very well known trigonometric
relationship, it is possible to relate squared oscillatory functions to the sum of a
constant term plus another oscillatory function with twice the frequency. For
instance (from Eq. (11)):
K zzf ðtÞ ¼ A2w K zzf0 þ A2w K zzfc cosð2oe tÞ þ A2w K zzfs sinð2oe tÞ.
The same reasoning may be extended to all terms inside the second bracket in Eq.
(15). Then it becomes apparent that each of these terms will introduce super-
harmonics proportional to wave amplitude squared, together with constant terms,
again proportional to the wave amplitude squared.
The occurrence of super-harmonics in the parametric excitation means that
the essential dynamics behind the non-linear motions described by Eq. (12) is not in
the form of coupled Mathieu equations; it is in fact a set of Hill equations. On the
other hand, the constant terms may be interpreted as additional stiffness
(proportional to wave amplitude squared) incorporated into the dynamical system
through third-order terms. Thus, it may be concluded that the extension of the
mathematical model to third-order brings new dynamic characteristics into the
resulting motions.
10. Numerical results for the heave and pitch motions
Numerical integration of Eq. (12) has been performed for a transom stern fishing
vessel, here denominated TS, which is very sensitive to parametric excitation. Hull
form and main characteristics of the vessel are given in Appendix B. First, the results
for the vertical motions are discussed.
In head seas, the heave and pitch motions are very important in favouring the
occurrence of roll parametric amplification. For small wave amplitudes, these
vertical motions are typically linear. Yet, for larger waves, these motions may
lose this characteristic, and asymmetries may be noticed and become remarkable. A
very complex set of coupling effects may now intervene in the resulting ship
dynamics.
Fig. 2 shows simulations of the heave and pitch motions for the considered ship in
head seas, F n ¼ 0:11, wave amplitude Aw ¼ 0:5 m, oe ¼ 2on4 . Results from the
third-order model are plotted against the linear response. It is observed that both
third-order heave and pitch motions are comparable to the linear displacements.
In Fig. 3 the same curves are obtained for the same speed and encounter
frequency, but for a higher wave, Aw ¼ 01:0 m. Now, some asymmetries are
observed in the responses of both modes, making it clear that the heave and pitch
motions are now far from the linear range. This result points out to the fact that the
exchange of energy between the vertical modes and the roll motion, which is an
ARTICLE IN PRESS
Heave - Ship TS: GM = 0.50 m, Fn = 0.11

We = 2Wn4, Aw = 0.50 m, ksi = 180 °
1.00
Heave (Linear)
Heave (3rd order)
0.50
Heave [m]
0.00
-0.50
-1.00
0 25 50 75 100
time [s]
Pitch - Ship TS: GM = 0.50 m, Fn = 0.11

We = 2Wn4, Aw = 0.50 m, ksi = 180 °
0.525
Pitch (Linear)
0.350
Pitch (3rd order)
Pitch [rad]
0.175
0.000
-0.175
-0.350
-0.525
0 25 50 75 100
time [s]
Fig. 2. Linear and non-linear responses in heave and pitch (Aw ¼ 0:50 m).
essential aspect of the dynamics of parametric resonance in head seas, must be

described as a set of coupled equations if all the complexities of excessive motions are
to be taken into consideration.
11. Results for the roll parametric amplifications
Figs. 4–9 show the roll motion obtained from numerical integration of the third-
order mathematical model described above. These simulations are given for
conditions that had been tested previously, Neves et al. (2002). In each of the
figures, the results for the third-order model are compared to the experimental
results and to numerical integration of the second-order counter part of the present
model.
In the numerical integrations the restoring curves have been approximated by
seventh-order polynomials. Fig. B2 in Appendix B shows the static restoring curves
(GZ curves) for two distinct metacentric heights, where it may be observed that the
polynomials give a fair description of the calm water restoring moment. It was
decided to retain the seventh-order modelling of the roll-roll restoring curve in the
numerical simulations, yet describing all the coupling effects of the parametric
excitation with terms up to the third order.
ARTICLE IN PRESS
Heave - Ship TS: GM = 0.50 m, Fn = 0.11

We = 2Wn4, Aw = 1.00 m, ksi = 180˚
1.00
Heave (Linear)
Heave (3rd order)
0.50
Heave [m]
0.00
-0.50
-1.00
0 25 50 75 100
time [s]
Pitch - Ship TS: GM = 0.50 m, Fn = 0.11

We = 2Wn4, Aw = 1.00 m, ksi = 180˚
0.525
Pitch (Linear)
0.350 Pitch (3rd order)
Pitch [rad]
0.175
0.000
-0.175
-0.350
-0.525
0 25 50 75 100
time [s]
Fig. 3. Linear and non-linear responses in heave and pitch, Aw ¼ 1:00 m.
Figs. 4 and 5 show the roll motion for GM ¼ 0:37 m, F n ¼ 0:15, oe ¼ 2on4 ¼
1:717 rd=s. In Fig. 4, for a low wave amplitude, Aw ¼ 0:45 m, there is excellent
agreement between the third-order model and the experimental results, whereas the
second-order model gives responses well in excess of the other curves. For the same
conditions, but for a higher wave amplitude, Fig. 5, the third-order model gives
results slightly lower than the experiments, but the comparison may be said to be
quite good. The second-order model is completely incapable of producing results
comparable to the time series obtained from the experiments.
Very similar interpretations are applicable to the cases shown in Figs. 6 and 7,
corresponding to GM ¼ 0:37 m, F n ¼ 0:20 and F n ¼ 0:30, oe ¼ 2on4 ¼ 1:717 rd=s.
In both conditions, agreement between the third-order model and experiments is
quite good, whereas the second-order model fails to reproduce the experiments.
Important to say, the condition given in Fig. 7 is one of quite excessive roll motions,
of the order of 38 degrees.
It should be observed that for the high GM case (GM ¼ 0:50 m,
oe ¼ 2on4 ¼ 1:968 rd=s), Figs. 8 and 9, possibly due to the higher GM, the roll
parametric amplification is not so intense. In particular, in the case of Fig. 9,
F n ¼ 0:15, Aw ¼ 0:60 m, the second and third-order models give good descriptions of
the time evolution of the roll motion.
ARTICLE IN PRESS
Ship TS: GM = 0.37 m, Fn =0.15

We = 2Wn4, Aw = 0.45 m, ksi=180˚
60
3rd.order
40 2nd.order
Roll Angle [deg]
Experimental
20
-20
-40
-60
0 25 50 75 100 125 150
time [s]
Fig. 4. Roll motion, GM ¼ 0:37 m, F n ¼ 0:15, Aw ¼ 0:45 m.
Ship TS: GM = 0.37 m, Fn =0.15

We = 2Wn4, Aw = 1.02 m, ksi = 180°
60
3rd. order
40 2nd. order
Roll Angle [deg]
Experimental
20
-20
-40
-60
0 25 50 75 100 125 150
time [s]
Ship TS: GM = 0.37 m, Fn =0.20

We = 2Wn4, Aw = 0.60 m, ksi = 180°
60
3rd. order
40 2nd. order
Roll Angle [deg]
Experimental
20
-20
-40
-60
0 25 50 75 100 125 150
time [s]

ARTICLE IN PRESS
Ship TS : GM = 0.37 m, Fn =0.30

We = 2Wn4, Aw = 0.78 m, ksi = 180°
60
3rd. order
40 2nd. order
Roll Angle [deg]
Experimental
20
-20
-40
-60
0 25 50 75 100 125 150
time [s]
Ship TS: GM = 0.50 m, Fn =0.11

We = 2Wn4, Aw = 0.63 m, ksi = 180°
60
3rd. order
40 2nd. order
Roll Angle [deg]
Experimental
20
-20
-40
-60
0 25 50 75 100 125 150
time [s]
Ship TS: GM = 0.50 m, Fn =0.15

We = 2Wn4, Aw = 0.60 m, ksi = 180°
60
3rd. order
40 2nd. order
Roll Angle [deg]
Experimental
20
-20
-40
-60
0 25 50 75 100 125 150
time [s]

ARTICLE IN PRESS
12. Conclusions
A derivative mathematical model was introduced, in which the heave, roll and
pitch motions and wave passage effects were described with coupling terms up to the
third order. Analytical derivation of all the coupling coefficients due to heave, roll
and pitch, and wave passage has been given. The derivation is structured such that
all coupling coefficients are expressed in terms of simple hull-wave characteristics.
This aspect should be recognized as relevant from the practical point of view of
understanding and correlating the complex non-linear responses to practical hull
configurations.
A qualitative analysis of the coupled non-linear system was performed. The
analysis points out to the possible appearance of super-harmonics and additional
stiffness due to the heave and pitch motions and wave passage. The essential
dynamic characteristics are those of a Hill equation with a hardening term
proportional to wave amplitude squared.
It is shown that, depending on the level of excitation, the heave and pitch motions
may display significant asymmetries. The appearance of these non-linear effects is
indicative of the importance of taking the ship motions in a coupled way, whenever
strong parametric excitation may occur.
Roll amplifications for low metacentric height tended to be strong. In these
cases the third-order numerical simulations based on the proposed model give
quite good comparisons with the experimental results. The increased stiffness,
proportional to wave amplitude squared, is thought to be responsible for
giving to the third-order model such a realistic description of the parametric
resonance. In the tested conditions with high metacentric height the roll
amplification is not excessive. Second and third-order models do not give very
distinct responses in these cases.
Acknowledgments
The present investigation is supported by CNPq within the STAB project (Non-
linear Stability of Ships). The authors also acknowledge financial support from
CAPES, FAPERJ and LabOceano.
Appendix A. Derivation of restoring coefficients
A.1. Restoring force
For the derivation of the hydrostatic heave force and restoring moments in roll
and pitch, a methodology analogous to the one presented in Neves (2002) is
employed. Vectors defined with respect to the coordinate systems of Fig. 1 of the
^ J;
main text are: for the inertial system, unit vectors are I; ^ K^ for the body fixed
^ ^ ^
system, unit vectors are i; j; k.
ARTICLE IN PRESS
G
φ
hi bi
O ∆i
Y
C dAi
W Bo
B1
E1
Fig. A1. Weight and hydrostatic forces in a generically displaced condition.
By definition, the restoring force is the resultant between weight and instantaneous
hydrostatic force:
F ~ þE
~H ¼ W ~1 ,
where the weight is

~ ¼ rgr 0 K.
W ^
The instantaneous hydrostatic force is

~1 ¼ rgr 1 K^ ¼ rgðr 0 þ r ÞK.
E ^
Thus:
~H ¼ rgr K.
F ^
In the above expressions, r is the density, r 0 is the submerged volume in the static
equilibrium of the ship, r 1 is the instantaneous submerged volume, and r is the
instantaneous incremental volume (variation of submerged volume) due to the
motions of the ship in heave, roll and pitch, see Fig. A1.
A.2. Restoring moments
By definition, the restoring moment is the resultant between the moments of

weight and instantaneous buoyancy:
~H ¼ M
M ~W þ M
~E
1
ARTICLE IN PRESS
where

~ W ¼ !
M ~ ¼ rg !
OG W OG r 0 K^ ,
! ~ ! h ! X ! i
~ E1 ¼ OB
M ^ ^
1 E 1 ¼ rg OB1 r 1 K ¼ rg OB 0 r 0 K þ Ob i dr i K^
It is clear that the restoring moment may be rewritten as

" #
! ! X !
~ ^
M H ¼ rg r 0 K OG OB 0 þ ^
Ob i dr i K .
i
Referring again to Fig. A1, the following expressions may be derived:

! ^
OG ¼ z̄G k,
! ^
OB0 ¼ z̄B0 k,
X X
r ¼ dr i ¼ hi dAi ,
i i
! hi ^
Ob i xAi I^ þ yAi J^ þ K,
2
k^ ¼ yI^ fJ^ þ K,
^
where z̄G is the vertical coordinate of the centre of gravity, z̄B0 is the vertical
coordinate of the centroid of the submerged volume, hi is the height of the elemental
prism, and (xAi, yAi) are the coordinates of the elemental prism in the instantaneous
plane of flotation, referred to system CXYZ.
With substitutions and regrouping, the following vector expression is obtained for
the restoring moment:
" #
X
M~ H ¼ Irg
^ r 0 ðz̄B0 z̄G Þf yAi hi dAi
i
" #
X
^
Jrg r 0 ðz̄B0 z̄G Þy þ xAi hi dAi .
i
In order to determine the restoring force and the moments, an analytical

expression will be derived to compute the height hi of each elemental prism. This
expression, it is clear, is a function of the relevant variables z, f, and y, that is, the
displacements in heave, roll and pitch (Neves, 2002).
A.3. Generalized hydrostatic actions
An approximation to obtain the height hi of each elemental prism (see Fig. A1)
caused by a generic displacement of the ship (z, f, y) may be obtained assuming that
ARTICLE IN PRESS
each hi is composed of a linear superposition of three contributions:

hi ¼ zr yAi f þ xAi y,
where (xAi, yAi), previously defined, is a point of a generic plane of flotation, and zr is
the relative vertical displacement relative to the instantaneous elevation of the sea
surface, defined by
zr ¼ z zðxAi ; yAi ; tÞ.
In this section we will consider the free surface as un-deformed (z ¼ 0), such that
zr ¼ z, but it is noted that wave passage effects may be obtained following the same
procedure described in this Appendix.
For zr ¼ z the volumetric variation will be
X X X
r ¼ dr i ¼ hi dAi ¼ ðz yAi f þ xAi yÞdAi
i i i
X X X
¼ z dAi f yAi dAi þ y xAi dAi ;
i i i
where the summations represent known geometric properties of the plane of

flotation:
X
dAi ¼ Aðz; f; yÞ area of instantaneous plane of flotation;
i
X
yAi dAi ¼ Ayf ðz; f; yÞ first static transversal moment of the area;
i
X
xAi dAi ¼ Axf ðz; f; yÞ first static longitudinal moment of the area;
i
such that:
r ¼ zAðz; f; yÞ fAyf ðz; f; yÞ þ yAxf ðz; f; yÞ
and the restoring force will be given by
~H ¼ rg½zA þ fAyf yAxf K.
F ^
For the restoring moment, again taking zr ¼ z and considering the following
known geometric properties of the plane of flotation:
X
y2Ai dAi ¼ I xx ðz; f; yÞ transversal moment of inertia of the area;
i
X
x2Ai dAi ¼ I yy ðz; f; yÞ longitudinal moment of inertia of the area;
i
X
xAi yAi dAi ¼ I xy ðz; f; yÞ product of inertia of the area;
i
ARTICLE IN PRESS
the following vector expression may be obtained for the restoring moment:
~ H ¼ rgf½Ayf z þ r 0 ðz̄B0 z̄G Þf þ I xx f I xy yI^
M
^
þ ½Axf z I xy f þ r 0 ðz̄B0 z̄G Þy þ I yy ygJ.
Finally, the restoring force and moments may be written as
Z H ¼ rgðAz þ Ayf f Axf yÞ,
K H ¼ rg½Ayf z þ r 0 ðz̄B0 z̄G Þf þ I xx f I xy y,
M H ¼ rg½Axf z I xy f þ r 0 ðz̄B0 z̄G Þy þ I yy y.
A.4. Hydrostatic actions up to third order
It is noted that the above expressions are functions of geometric properties (A,
Axf, Ayf, Ixx, etc.) of the instantaneous plane of flotation, which in turn depend on
the instantaneous displacements of the ship (z, f, y). These properties may be
approximated employing multi-variable Taylor series around the average position.
Aiming at describing the hydrostatic actions up to third order, we consider the
geometric functions expanded up to second order:

qA qA 1 q2 A 2 q2 A 1 q2 A 2 1 q2 A 2
A ¼ A0 þ z þ y þ z þ zy þ f þ y,
qz 0 qy 0 2 qz2 0 qz qy0 2 qf2 0 2 qy2 0

qAxf qAxf 1 q2 Axf 2 q2 Axf
Axf ¼ A0 xf 0 þ zþ yþ z þ zy
qz 0 qy 0 2 qz2 0 qz qy 0

1 q2 Axf 2 1 q2 Axf 2
þ f þ y ,
2 qf2 0 2 qy2 0

qAyf q2 Ayf q2 Ayf
Ayf ¼ fþ zf þ fy,
qf 0 qz qf qf qy
0 0

qI xx qI xx 1 q2 I xx 2 q2 I xx
I xx ¼ I xx0 þ z þ y þ z þ zy
qz 0 qy 0 2 qz2 0 qz qy 0

1 q2 I xx 2 1 q2 I xx 2
þ f þ y ,
2 qf2 0 2 qy2 0

qI yy qI yy 1 q2 I yy 2 q2 I yy
I yy ¼ I yy0 þ zþ yþ z þ zy
qz 0 qy 0 2 qz2 0 qz qy0

1 q2 I yy 2 1 q2 I yy 2
þ f þ y,
2 qf2 0
2 qy2 0
ARTICLE IN PRESS

qI xy q2 I xy q2 I xy
I xy ¼ f þ zf þ fy,
qf 0 qz qf0 qf qy0
where A0, A0xf0, Ixx0, Iyy0 are geometric properties of the average plane of flotation.
Substitution of the given expansions in the previous general expressions of the
restoring actions will result in third-order representations in terms of derivatives:
"
qA z2 qA qAyf f2 qAxf y2 q2 A z3

Z H ¼ rg A0 z A0 xf 0 y þ þ zy þ þ 2
qz 0 2 qy 0 qf 0 2 qy 0 2 qz 0 6
#
q2 A z2 y q2 A f2 z q Ayf f2 y q2 A y2 z q2 Axf y3
2
þ þ 2 þ þ 2 ,
qz qy0 2 qf 0 2 qf qy 2 qy 0 2 qy2 0 6
0
"
qI xx qI xx
K H ¼ rg r 0 ðz̄B0 z̄G Þf þ I xx0 f þ zf þ fy
qz 0 qy 0
#
q2 I xx z2 f q2 I xx y2 f q2 I xx q2 I xx f3
þ 2 þ þ zfy þ ,
qz 0 2 qy2 0 2 qz@y 0 qf2 0 6

qAxf z2 qI yy
M H ¼ rg A0 xf 0 z þ r 0 ðz̄B0 z̄G Þy þ I yy0 y þ zy
qz 0 2 qz 0

qI xy f2 qI yy y2 q2 Axf z3 q2 I yy z2 y q2 Axf f2 z
þ þ
qf 0 2 qy 0 2 qz2 0 6 qz2 0 2 qf2 0 2
#
q2 I yy f2 y q2 I yy zy2 q2 I yy y3
þ þ þ .
qf2 0 2 qz qy0 2 qy2 0 6
Considering that hydrostatic actions are functions with continuous partial

derivatives, second and third order mixed derivatives are equal, and the following
equalities are verified:

qA qAxf qAyf
qI xx
¼ ; ¼ ,
qy 0 qz 0 qf 0 qz 0

qI xx qI xy qI yy qAxf
¼ ; ¼ ,
qy 0 qf 0 qz 0 qy 0

q2 A q2 Ayf q2 I xx
¼ ¼ ,
qf2 0 qz qf
0
qz2 0

q2 A q2 Axf q2 I yy
¼ ¼ ,
qy2 0 qz qy 0 qz2 0

q2 A q2 Axf
¼ ,
qz qy 0 qz2 0
ARTICLE IN PRESS
Table A1a
Hydrostatic coefficients—Linear
Heave Roll Pitch
Zz ¼ rgA0 Kz ¼ 0 M z ¼ rgA0 xf 0
Zf ¼ 0 K f ¼ rg½r 0 ðz̄B0 z̄G Þ þ I xx0 Mf ¼ 0
Zy ¼ rgA0 xf 0 Ky ¼ 0 M y ¼ rg½r 0 ðz̄B0 z̄G Þ þ I yy0
Table A1b
Hydrostatic coefficients—Second order
Heave Roll Pitch

Zzz ¼ rgqA K zz ¼ 0 qAxf
qz 0 M zz ¼ rg qz 0

Zzf ¼ 0 K zf ¼ rg qIqzxx 0 M zf ¼ 0

Zzy ¼ rg qA K zy ¼ 0 qI yy
qy 0 M zy ¼ rg qz 0

qAyf K ff ¼ 0 qI xy
Zff ¼ rg qf 0 M ff ¼ rg qf 0

Zfy ¼ 0 K fy ¼ rg qIqyxx 0 M fy ¼ 0

qAx K yy ¼ 0 qI yy
Zyy ¼ rg qy f M yy ¼ rg qy 0
0

q2 I xx q2 I xy q2 Ayf q2 Axf
¼ ¼ ¼ ,
qz qy 0 qz qf0 qf qy qf2 0
0

q2 I xx q2 I xy q2 I yy
¼ ¼ ,
qy2 0 qf qy0 qf2 0

q2 I yy q2 Axf
¼ .
qz qy0 qy2 0
Comparing the above results for ZH, and KH, MH with corresponding terms in Eq.
(12) of the main text gives the expressions for the derivatives in terms of geometric
properties of the plane of flotation, as shown in Tables A1a–c.
A.5. Expressions for the geometric derivatives
It is possible and relevant to relate the derivatives given above to longitudinal

distributions of hull characteristics. Paulling and Rosenberg (1959) presented a
simple methodology for obtaining some of the geometric derivatives considering a
second-order model, in terms of semi-breadth and flare defined at the average
floating position. The same methodology was employed in Neves (2002) to obtain
ARTICLE IN PRESS
Table A1c
Hydrostatic coefficients—Third order
Heave Roll Pitch

2 K zzz ¼ 0 q2 Axf
Zzzz ¼ rg qqzA2 M zzz ¼ rg
0 qz2
0
Zzzf ¼ 0 2 M zzf ¼ 0
K zzf ¼ rg q qzI 2xx
0
q2 A K zzy ¼ 0 q2 I yy
Zzzy ¼ rg qz qy0 M zzy ¼ rg qz2 0

2 K ffz ¼ 0 q2 Axf
Zffz ¼ rg qqfA2 M ffz ¼ rg
0 qf2
0
Zfff ¼ 0 2 M fff ¼ 0
K fff ¼ rg qqfI xx
2
0
q2 Ay K ffy ¼ 0 q2 I yy
Zffy ¼ rg qf qyf M ffy ¼ rg qf2 0

0
2 K yyz ¼ 0 q2 I yy
Zyyz ¼ rg qqyA2 M yyz ¼ rg qz qy
0 0
Zyyf ¼ 0 2 M yyf ¼ 0
K yyf ¼ rg qqyI xx
2
0
q2 Axf K yyy ¼ 0 q2 I yy
Zyyy ¼ rg M yyy ¼ rg
qy2 qy2 0
0
xx
Zzfy ¼ 0 2 M zfy ¼ 0
K zfy ¼ rg qqzIqy
0
the complete set of second-order derivatives. Here the procedure is extended to the
present third-order model.
When deriving these expressions with respect to z, f and y, it will be observed that
the following rules of derivation are to be applied in the case of angular
displacements in roll and pitch:
q q qz q q qz
¼ ; ¼
qf qz qf qy qz qy
and from the general transformation matrix between a rotated and a fixed frame of
reference it can be deduced that in the case of roll:
qz
z ¼ ȳ sin f þ z̄ cos f; ‘ ¼ ȳ cos f z̄ sin f
qf
such that

qz q q
¼ ȳ ) ¼ ȳ
qf0 qf qz̄
and in the case of pitch:
qz
z ¼ x̄ sin y þ z̄ cos y; ‘ ¼ x̄ cos y z̄ sin y,
qy
ARTICLE IN PRESS

qz q q
¼ x̄ ) ¼ x̄
qy0 qy qz̄

and it is also observed that qz=qz̄0 ¼ 1.
Table A2
Second order geometric derivatives
R R
qA
¼ 2 L qȳ qA qA
¼ 2 L x̄ qȳ
qz 0 qz̄ dx qf0 ¼0 qy 0 qz̄ dx
R R
qAxf qȳ qAxf qAxf 2 qȳ
qz 0 ¼ 2 L x̄ qz̄ dx qf 0 ¼ 0 qy 0 ¼ 2 L x̄ qz̄ dx
R 2 qȳ
qAyf qAyf qAyf
qz 0 ¼ 0 qf 0 ¼ 2 L ȳ qz̄ dx qy 0 ¼ 0
R 2 qȳ R
qI xx qI xx qI xx 2 qȳ
qz 0 ¼ 2 L ȳ qz̄ dx qf 0 ¼ 0 qy 0 ¼ 2 L x̄ȳ qz̄ dx2
R 2 qȳ R 3 qȳ
qI yy qI yy qI yy
qz 0 ¼ 2 L x̄ qz̄ dx qf 0 ¼ 0 qy 0 ¼ 2 L x̄ qz̄ dx
R
qI xy qI xy 2 qȳ qI xy
qz ¼ 0 0 qf ¼ 2 L x̄ȳ qz̄ dx
0 qy ¼ 0
0
Table A3
Third order geometric derivatives
R 2
q2 A a q2 A q2 A

qz2 0
¼0 ¼4 ȳ qȳ dx þ A0
qy2 0
¼0a
qf2 0 L qz̄

q2 A q2 A a q2 A
qz qf0 ¼ 0 qz qy0 ¼ 0 qf qy0 ¼ 0
2
q2 Axf q2 Axf R q2 Axf
¼0a ¼ 4 x̄ȳ qȳ dx þ A0 xf 0 ¼0a
qz
2 qf 2 L qz̄ qy
2
0 0 0
q2 Axf 2
q Axf q2 Axf
a
qz qf ¼ 0 qz qy ¼ 0 qf qy ¼ 0
0 0 0
q2 Ayf q2 Ayf q2 Ayf
¼0 ¼0 ¼0
qz2 qf2 qy2
0 0 0
q2 Ayf R qȳ2 q2 Ayf q2 Ayf R 2
qȳ
qz qf ¼ 4 L ȳ qz̄ dx þ A0 qz qy ¼ 0 qf qy ¼ 4 L x̄ȳ qz̄ dx A0 xf 0
0 R 2 0
R 3 qȳ2 0
R 2
2 q2 I xx 2
q I xx
qz2 0
¼ 4 L ȳ qȳ
qz̄ dx þ A0 qf 02 ¼ 8 L ȳ qz̄ dx þ 2I xx0 q I xx
qy2 0
¼ 4 L x̄2 ȳ qȳ
qz̄ dx þ I yy0
R 2
q2 I xx q2 I xx qȳ q2 I xx
qz qf 0 ¼ 0 qz qy 0 ¼ 4 L x̄ȳ qz̄ dx A0 xf 0 qf qy 0 ¼ 0
R 2
q2 I yy q I yy
2 q2 I yy
qz2 0
¼0a ¼ 4 L x̄2 ȳ qȳ
qz̄ dx þ I yy0 2 ¼0a
qf2 0 qy 0

q I yy
2 q I yy
2
a q I yy
2
qz qf 0 ¼ 0 qz qy 0 ¼ 0 qf qy 0 ¼ 0

q2 I xy q2 I xy q2 I xy
qz2 0
¼0 qf2 0
¼0 qy2 0
¼0
R 2 R 2 qȳ2
q I xy
2
qȳ q I xy
2
q I xy
2
qz qf ¼ 4 L x̄ȳ qz̄ dx þ A0 xf 0 qz qy ¼ 0
0 qf qy ¼ 4 L x̄ ȳ qz̄ dx I yy0
0 0
a
These expressions were obtained analytically for the case of a ship with inclined wall side,
corresponding to a good approximation in the case of ships of conventional forms, small displacements
and smooth transversal curvatures ðq2 ȳ=qz̄2 ! 0Þ at the considered water line.
ARTICLE IN PRESS
With these results, the expressions for the non-vanishing second and third-order
hydrostatic derivatives can be derived, with the integrations of sectional geometrical
characteristics defined along the ship length. Tables A2 and A3 summarizes the
results.
Appendix B. Characteristics of fishing vessel
The hull form and main characteristics of the ship employed in the present
investigation—here denominated TS—are give in Fig. B1 and Table B1, respectively.
The ship was tested with two different metacentric heights, GM ¼ 0:37 m and
GM ¼ 0:50 m. The respective roll-roll restoring curves in calm water (GZ curves) are
given in Fig. B2.
Fig. B1. Hull form of transom stern fishing vessel.
Table B1
Ship main characteristics
Denomination Transom stern (TS)
Overall length [m] 25.91

Length between perpendiculars [m] 22.09
Breadth [m] 6.86
Depth [m] 3.35
Draft [m] 2.48
Displacement [ton] 170.3
Longitudinal radius of gyration [m] 5.52
ARTICLE IN PRESS
Static Stability Curve - Ship TS Static Stability Curve - Ship TS

(GM = 0.37 m) (GM = 0.50 m)
600 600
SSC-TS37 SSC-TS50
500 500 7th-order fit
7th-order fit
Rest. Moment [kN.m]

Rest. Moment [kN.m]
400 400
300 300
200 200
100 100
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
phi [deg] phi [deg]
Fig. B2. Restoring curves of fishing vessel TS for two different conditions (calm water).
References
Abkowitz, M.A., 1969. Stability and Motion Control of Ocean Vehicles. The MIT Press.
Bass, D.W., Haddara, M.R., 1988. Nonlinear models of ship roll damping. International Shipbuilding
Progress 35 (401), 5–24.
Blocki, W., 1980. Ship safety in connection with parametric resonance of the roll. International
Shipbuilding Progress 27 (306), 36–53.
Bulian, G., Francescutto, A., Lugni, C., 2003. On the non-linear modeling of parametric rolling in regular
and irregular waves. In: Proceedings of 8th International Conference on the Stability of Ships and
Ocean Vehicles (STAB’2003), Madrid, Spain, pp. 305–323.
Cesari, L., 1971. Asymptotic Behavior and Stability Problems in Ordinary Differential Equations, third
ed. Springer-Verlag, Berlin Heidelberg, Berlin, Germany.
Dallinga, R.P., Blok, J.J., Luth, H.R., 1998. Excessive rolling of cruise ships in head and following waves.
In: RINA International Conference on Ship Motions & Manoeuvrability. Royal Institute of Naval
Architects, London.
De Kat, J.O., Paulling, J.R., 1989. The simulation of ship motions and capsizing in severe seas.
Transactions of Society of Naval Architects and Marine Engineers 97.
France, W.N., Levadou, M., Treakle, T.W., Paulling, J.R., Michel, R.K., Moore, C., 2003. An
investigation of head-sea parametric rolling and its influence on container lashing systems. Marine
Technology 40 (1), 1–19.
Himeno, Y., 1981. Prediction of ship roll damping—state of the art. Dept. Naval Architecture and Marine
Engineering. The University of Michigan, Report no. 239.
Kerwin, J.E., 1955. Notes on rolling in longitudinal waves. International Shipbuilding Progress 2 (16),
597–614.
Levadou, M., Palazzi, L., 2003. Assessment of operational risks of parametric roll. In: Proceedings of
World Maritime Technology Conference (WMTC’2003), Society of Naval Architects and Marine
Engineers, San Francisco, United States.
Luth, H.R., Dallinga, R.P., 1999. Prediction of excessive rolling of cruise vessels in head and following
waves. In: Proceedings of PRAD’s Conference.
Matuziak, J., 2003. On the effects of wave amplitude, damping and initial conditions on the parametric
roll resonance. In: Proceedings of 8th International Conference on the Stability of Ships and Ocean
Vehicles (STAB’2003), Madrid, Spain, pp. 341–347.
ARTICLE IN PRESS
Meyers, W.G., Sheridan, D.J., Salvesen, N., 1975. Manual—NSRDC ship-motion and sea-load computer
program. Naval Ship Research and Development Center, Maryland, USA, Report no. 3376.
Munif, A., Umeda, N., 2000. Modeling extreme roll motions and capsizing of a moderate-speed ship in
astern waves. Journal of the Society of Naval Architects of Japan 187.
Neves, M.A.S., 2002. On the excitation of combination modes associated with parametric resonance in
waves. In: Proceedings of the Sixth International Ship Stability Workshop, Webb Institute, New York.
Neves, M.A.S., Valerio, L., 2000. Parametric resonance in waves of arbitrary heading. In: Proceedings of
the Seventh International Conference on Ship Stability and Ocean Vehicles (STAB’2000), Launceston,
Australia, pp. 680–687.
Neves, M.A.S., Pérez, N.A., Valerio, L., 1999. Stability of small fishing vessels in longitudinal waves.
Ocean Engineering 26 (12), 1389–1419.
Neves, M.A.S., Pérez, N., Lorca, O., 2002. Experimental analysis on parametric resonance for two fishing
vessels in head seas. In: Proceedings of the Sixth International Ship Stability Workshop, Webb
Institute, New York.
Neves, M.A.S., Pérez, N., Lorca, O., Rodrı́guez, C., 2003. Hull design considerations for improved
stability of fishing vessels in waves. In: Proceedings of the Eighth International Conference on the
Stability of Ships and Ocean Vehicles (STAB’2003), Madrid, Spain, pp. 291–304.
Newman, J.N., 1977. Marine Hydrodynamics. The MIT Press.
Oakley, O.H., Paulling, J.R., Wood, P.D., 1974. Ship motions and capsizing in astern seas. In: Proceedings
of the 10th Symposium on Naval Hydrodynamics, Cambridge, MA.
Palmquist, M., Nygren, C., 2004. Recordings of head-sea parametric rolling on a PCTC. Annex of Report
of Document SLF 47/6/6-IMO.
Paulling, J.R., 1961. The transverse stability of a ship in a longitudinal seaway. Journal of Ship Research 4
(4), 37–49.
Paulling, J.R., Rosenberg, R.M., 1959. On unstable ship motions resulting from non-linear coupling.
Journal of Ship Research 3 (1), 36–46.
Salvesen, N., Tuck, O.E., Faltinsen, O., 1970. Ship motions and sea loads. Transactions of SNAME 78,
250–287.
Shin, Y.S., Belenky, V.L., Lin, W.M., Weems, K.M., Engle, A.H., 2003. Nonlinear time domain
simulation technology for seakeeping and wave-load analysis for modern ship design. In: Proceedings
of the World Maritime Technology Conference (WMTC’2003), Society of Naval Architects and
Marine Engineers, San Francisco.
Skomedal, N., 1982. Parametric excitation of roll motion and its influence on stability. In: Proceedings of
the Second International Conference on Stability of Ships and Ocean Vehicles (STAB’82), Tokyo,
Japan, pp. 113–125.
Spyrou, K.J., 2000. Designing against parametric instability in following seas. Ocean Engineering 27 (6),
625–653.
Umeda, N., Hamamoto, M., et al., 1995. Model experiments of ship capsize in astern seas. Journal of
Society of Naval Architects of Japan 177, 207–217.
Umeda, N., Hashimoto, H., Vassalos, D., Urano, S., Okou, K., 2003. Non-linear dynamics on parametric
roll resonance with realistic numerical modeling. In: Proceedings of the Eighth International
Conference on the Stability of Ships and Ocean Vehicles (STAB’2003), Madrid, Spain, pp. 281–290.

On Unstable Ship Motions Resulting From Strong Non-Linear Coupling

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

On Unstable Ship Motions Resulting From Strong Non-Linear Coupling

Загружено:

Авторское право:

Доступные форматы

ARTICLE IN PRESS

Ocean Engineering 33 (2006) 1853–1883

On unstable ship motions resulting from strong

r incremental volume for displaced hull

second-order non-linear mathematical model also indicated excessive ampliﬁcations

Two right-handed co-ordinate systems are employed to describe the motions. An

Fig. 1. Co-ordinate axis and deﬁnition of the six degrees of motions.

3. Hydrodynamic coefﬁcients associated with potential theory

The hydrodynamic inertia and damping matrices are expressed respectively, as

_ all the other terms in

4. Non-linear roll damping

5. Non-linear restoring actions

zðx; y; t; wÞ ¼ Aw cos½kx cosðwÞ þ ky sinðwÞ oe t,

zðx; tÞ ¼ Aw cos½kx þ oe t. (4)

These terms represent linear and non-linear Froude-Krilov force/moments

6. Restoring coefﬁcients in calm water

7. Interactions between non-linear motions and incident wave

instantaneous position of the wave. Contributions independent of the ship motions,

Heave Roll Pitch

Heave Roll Pitch

Heave Roll Pitch

K zf ðtÞ ¼ Aw K zfc cosðoe tÞ þ Aw K zfs sinðoe tÞ,

8. Non-linear heave–roll–pitch equations of motions

(a) vertical motions, z(t) and y(t), assumed to be purely harmonic,

where the three terms within brackets represent contributions to parametric

9. Theoretical analysis of perturbed roll motions

Stability of motion of the non-linear set of equations may be assessed by means of

zðtÞ ¼ z^ðtÞ þ xðtÞ ¼ Aw Z3 cosðoe t þ az Þ þ xðtÞ,

10. Numerical results for the heave and pitch motions

Heave - Ship TS: GM = 0.50 m, Fn = 0.11

Pitch - Ship TS: GM = 0.50 m, Fn = 0.11

essential aspect of the dynamics of parametric resonance in head seas, must be

11. Results for the roll parametric ampliﬁcations

Heave - Ship TS: GM = 0.50 m, Fn = 0.11

Pitch - Ship TS: GM = 0.50 m, Fn = 0.11

Fig. 3. Linear and non-linear responses in heave and pitch, Aw ¼ 1:00 m.

Ship TS: GM = 0.37 m, Fn =0.15

Fig. 4. Roll motion, GM ¼ 0:37 m, F n ¼ 0:15, Aw ¼ 0:45 m.

Ship TS: GM = 0.37 m, Fn =0.15

Fig. 5. Roll motion, GM ¼ 0:37 m, F n ¼ 0:15, Aw ¼ 1:02 m.

Ship TS: GM = 0.37 m, Fn =0.20

Fig. 6. Roll motion, GM ¼ 0:37 m, F n ¼ 0:20, Aw ¼ 0:60 m.

Ship TS : GM = 0.37 m, Fn =0.30

Fig. 7. Roll motion, GM ¼ 0:37 m, F n ¼ 0:30, Aw ¼ 0:78 m.

Ship TS: GM = 0.50 m, Fn =0.11

Fig. 8. Roll motion, GM ¼ 0:50 m, F n ¼ 0:11, Aw ¼ 0:63 m.

Ship TS: GM = 0.50 m, Fn =0.15

Fig. 9. Roll motion, GM ¼ 0:50 m, F n ¼ 0:15, Aw ¼ 0:60 m.

Appendix A. Derivation of restoring coefﬁcients

A.1. Restoring force

Fig. A1. Weight and hydrostatic forces in a generically displaced condition.

where the weight is

The instantaneous hydrostatic force is

A.2. Restoring moments

By deﬁnition, the restoring moment is the resultant between the moments of

It is clear that the restoring moment may be rewritten as

Referring again to Fig. A1, the following expressions may be derived:

In order to determine the restoring force and the moments, an analytical

A.3. Generalized hydrostatic actions

each hi is composed of a linear superposition of three contributions:

where the summations represent known geometric properties of the plane of

K H ¼ rg½Ayf z þ r 0 ðz̄B0 z̄G Þf þ I xx f I xy y,

M H ¼ rg½Axf z I xy f þ r 0 ðz̄B0 z̄G Þy þ I yy y.