D. OBREJA, R. NEBERGOJ, L. CRUDU, S. PACURARU (POPOIU) PARAMETRIC ROLLING ATUDC 629.5.017.22:629.5.015.

141
MAIN RESONANCE

Dan OBREJA1
Radoslav NABERGOJ2 Parametric Rolling at Main
Liviu CRUDU1
Sandita PACURARU Resonance
(POPOIU)1
Original scientific paper
The parametric resonance of the induced rolling is a typical dangerous situation for the trans-
versal stability of ships operating in longitudinal waves. This paper presents the results of experi-
mental tests regarding the seakeeping performances of a 2700 dwt cargo model at full loading, in
regular longitudinal waves. The induced roll motion, in the second instability domain, both at zero
and design speed was observed. The conditions for occurrence of the parametric resonance were
analysed and the instability domains of motion were identified. The theoretical analysis of the heave
and pitch motions in longitudinal waves, in frequency domain, was performed by using a computer
code based on the Frank-close fit method. A satisfactory correlation between the theoretical and
Authors’ addresses: experimental results was found for heave and pitch motions. In order to simulate the induced roll-
1
“Dunarea de Jos” University of ing a typical differential coupled equations system for heave, induced roll and pitch motions was
Galati, Romania used. The numerical solution was obtained using the Runge-Kutta method. The simulation results
Naval Architecture Faculty for the ship motions at zero speed, in following regular waves, are presented. A good agreement
Domneasca Str., no. 47, Zip code was obtained between the numerical and experimental results in time domain.
800008 Keywords: experimental tests, parametric rolling, time domain simulation
Tel/fax: (+40) 0236-495 400
E-mail: [email protected]
2
University of Trieste, Italy
Parametersko ljuljanje pri glavnoj rezonanciji
Department of Naval Architecture, Izvorni znanstveni rad
Ocean and Environmental Engineer-
Parametarska rezonancija ljuljanja tipična je opasnost za poprečni stabilitet broda na uzdužnim
ing valovima. Članak prikazuje rezultate pokusa ponašanja 2700 dwt modela teretnog broda pod
Via A. Valerio 10, Zip code 34127 punim opterećenjem na pravilnim uzdužnim valovima. Opažene su pojave induciranog ljuljanja u
Fax: (+39) 405583443 drugoj domeni nestabilnosti pri nultoj i projektnoj brzini. Analizirani su uvjeti za pojavu paramet-
Phone: (+39) 405583407 arske rezonancije te su određena područja nestabilnosti njihanja. Teorijska analiza u frekventnoj
E-mail: [email protected] domeni poniranja i posrtanja na uzdužnim valovima je provedena na računalu primjenom Frankove
metode prilagođavanja. Za poniranja i posrtanje teorijski i eksperimentalni rezultati su u suglasju.
Received (Primljeno): 2008-04-29 Za numeričko rješenje korišten je postupak Runge-Kutta. Prikazani su rezultati njihanja broda
Accepted (Prihvaćeno): 2008-06-08 na harmonijskim valovima pri nultoj brzini. U vremenskoj se domeni rezultati pokusa i numerički
Open for ciscussion (Otvoreno za rezultati dobro slažu.
raspravu): 2009-31-12 Ključne riječi: parametarsko ljuljanja, pokusi, simulacija u vremenskoj domeni

1 Introduction The main causes of the induced roll motion are considered
to be:
When the ship runs in longitudinal waves, the representative - the energetic “saturation” phenomena of heave and pitch
ship motions are the heave and pitch ones. Due to the ship-waves motion [2];
interaction the transversal metacentre position has a dynamic - the nonlinear coupling of the heave or pitch motions with the
modification in time domain, the excitation period being equal induced roll one [3].
to the incident wave one [1]. As an effect of lateral perturbations, Nayfeh considers that the energy put into the pitch and heave
an induced roll motion may occur. The amplitude of the roll mo- motions by the wave excitations may be partially transferred into
tion can increase, in parametric resonance conditions, according the roll motions by means of nonlinear coupling among these
to the relation modes; consequently roll motion can be indirectly excited. The
Te n ship will spontaneously develop large amplitude roll motions at
= (1) parametric resonance.
Tφ 2
The first instability domain (n=1) is the most dangerous one
for the ship’s transverse stability, but large amplitude roll mo-
where, Te is the incident wave period and Tφ the roll natural tions can occur in the main resonance domain, when n=2. The
period. seakeeping tests in longitudinal waves of a cargo ship model
The roll parametric resonance is known as a typical danger- [4] revealed the occurrence of parametric rolling, both at zero
ous situation for the transversal stability of ships operating in and design speed. The induced roll motion has occurred in a
longitudinal waves. narrow frequency domain of the incident waves. The body plan

340 59(2008)4, 340-347

PARAMETRIC ROLLING AT MAIN RESONANCE D. OBREJA, R. NEBERGOJ, L. CRUDU, S. PACURARU (POPOIU)

of the cargo ship is shown in Figure 1. The main particulars and is the restoring moment in longitudinal waves, φ is the roll angle
characteristics of the ship and the scaled model, at full loading and t is the time variable.
condition, are listed in Table 1. The dependency of the damping moment on the roll angle
may be synthetically written as

D(φ , φ ) = ∑ Dijφ iφ j , (i, j=0,1,2,…) (3)

i, j
where Dij represents the damping coefficients.
Within the limits of a cubic approximation, the restoring
moment may be written as

M r (φ , t ) = gm[GM (t )φ + k3φ 3 ] (4)

where, GM(t) is the time-dependent metacentric height, m is the

ship mass and g is the gravitational acceleration.
When the ship runs in longitudinal waves, the metacentric
height has a time-dependent harmonic variation expressed as
follows

GM (t ) = GM m − GM a cos ω t (5)

where, GMm and GMa are the mean value and the amplitude of
Figure 1 The body plan of the cargo ship the metacentric height and ω is the wave circular frequency.
Slika 1 Linije teretnog broda The excitation coefficient for the induced roll motion has the
expression
Table 1 Main particulars at full loading condition
Tablica 1 Glavne izmjere za potpuno nakrcani brod μ = 0.5GM a / GM m (6)

Full Model The instability domains are corresponding to the critical

Main characteristics
scale scale (1/30) frequency ones, where the ship may develop large amplitude
Length over all, Lmax 86.04 m 2.864 m roll motions.
A remarkable property of Mathieu’s equation is that the
Length between perpendiculars, L 79.84 m 2.661 m
solutions reach infinite values within instability domains. The
Breadth, B 14.5 m 0.483 m frontiers of both stability and instability domains are defined by
Volumetric displacement, — 4297.0 m3 0.159 m3
Mean draught, T 5.245 m 0.175 m
Longitudinal centre of gravity, Figure 2 The instability domains of induced roll motion
38.702 m 1.290 m Slika 2 Područje nestabilnosti induciranog ljuljanja
LCG
Vertical centre of gravity, KG 4.5 m 0.150 m
Metacentric height, GMT 1.6 m 0.053 m
Natural period of roll motion, Tφ 7.55 s 1.38 s
Roll radius of gyration, kxx 4.331 m 0.144 m
Pitch radius of gyration, kyy 18.703 m 0.623 m
Yaw radius of gyration, kzz 18.042 m 0.601 m
Ship speed, U 6.95 m/s 1.27 m/s
Froude number, Fn 0.25 0.25

2 Instability Domains of Induced Roll Motions

The uncoupled equation of the induced roll motion in
longitudinal regular waves may be written using the Mathieu
formulation [2]

( I 4 + A44 )φ + D(φ , φ ) + M r (φ , t ) = 0 (2)

where, I4 is the inertia moment

. of roll motion, A44 is the added
mass of roll motion, D(φ, φ) is the roll damping moment, Mr(φ, t)

59(2008)4, 340-347 341

D. OBREJA, R. NEBERGOJ, L. CRUDU, S. PACURARU (POPOIU) PARAMETRIC ROLLING AT MAIN RESONANCE

T or 2T periodical solutions of the equation. By imposing Math- values of the excitation coefficient. The data from Figure 2 are
ieu’s equation to have 2T period solutions, the frontiers of the obtained neglecting the damping moment contribution. If the
odd instability domain can be obtained. In the second instability influence of the damping term is considered, the area between
domain, Mathieu’s equation must have T period solution. the frontiers of the instability domains becomes smaller.
Figure 2 shows the diagram of the first three instability do- The excitation coefficient is not the only parameter that has an
mains of the induced roll motion for the cargo ship. The vertical influence on induced roll motions. Another necessary condition
axis is the 2ωφ/ ω ratio, where ωφ, is the roll natural circular fre- may be mathematically written as follows
quency and the horizontal axis is the excitation coefficient μ. One 1 4π
can observe that the first instability domain may lead to occur- Tφ = ⋅ , n ∈N*
n 2π g U
rence of the induced roll motion for small values of the excitation − 2π cos α i (7)
coefficient μ. The instability domains become larger for increased λ λ

Figure 3 The necessary conditions for induced roll motions in Figure 4 The necessary conditions for induced roll motions in
the first instability domain the second instability domain (main resonance)
Slika 3 Potrebni uvjeti za induciranje ljuljanja u prvom području Slika 4 Potrebni uvjeti za induciranje ljuljanja u drugom području
nestabilnosti nestabilnosti (glavna rezonancija)

342 59(2008)4, 340-347

PARAMETRIC ROLLING AT MAIN RESONANCE D. OBREJA, R. NEBERGOJ, L. CRUDU, S. PACURARU (POPOIU)

where, U is ship’s speed and αi is the heading angle. The graphi-

cal representation of the condition given by the equation (7) is
shown in Figures 3 and 4. On the vertical axis the L/λ ratio is
represented, where L is the ship length and λ is the length of the
regular wave, and on the horizontal axis the natural roll period
Tφ is represented. For the studied cargo ship, having the full scale
natural roll period Tφ = 7.55 s, the probability of a roll motion
occurrence in the second instability domain (main resonance) is
increased. For the case at zero speed, the first instability domain
corresponds to L/λ = 3.5 and the second instability domain to
L/λ = 0.896. Regular waves, with small wavelength, do not have
enough energy to generate induced roll motion. Consequently,
at zero speed, parametric roll resonance in the first instability
domain has low probability of occurrence. Moreover, due to the
limitations of the wave generator, the experimental tests for fm
= 1.43 Hz (corresponding to L/λ = 3.5) could not be carried out.
However, the induced roll motions at zero speed were measured,
in the second instability domain, at main resonance, for both head
and following waves respectively.
Figures 3 and 4 demonstrate that for both instability domains,
the induced roll motion in regular following waves, at design
speed, is not possible. When the ship runs at design speed in
head waves, the probability of occurrence of roll motion in the
second instability domain (L/λ = 0.443) is greater as compared
to the first instability domain (L/λ = 1.25), where the amplitudes
of motions of the ship model, on short waves, are very small. The
main roll resonance for the ship running at design speed, in head
waves, was observed and measured during the experimental tests.
Comparing with the parametric resonance in the first instability
domain, the induced roll motion at the main resonance for the
ship in longitudinal waves was not intensively investigated. The
literature offers little information on the presence of induced
roll motion in the second instability domain [5]. From this point
of view, a theoretical and experimental analysis of the main roll
resonance represents a necessary extension of the studies on
parametric roll resonance.

3 Theoretical and Experimental

Determination of Ship Motions in
Longitudinal Waves
The theoretical evaluation of ship motions in longitudinal
waves has been obtained by using a computer code based on the
Frank close-fit method [6]. Figures 5 and 6 show the transfer
functions for heave and pitch motions respectively, in regular Figure 5 Non-dimensional transfer functions of heave motion in
head waves, as a function of L/λ ratio. Figures 7 and 8 depict the regular head waves
Slika 5 Bezdimenzionalna prijenosna funkcija poniranja na
transfer functions for heave and pitch motions respectively, in pravilnim valovima u pramac
regular following waves, as a function of L/λ ratio. The non-di-
mensional formulations are obtained using z/ζw and θ/(kζw) ratios
where, z and θ are the amplitudes of heave and pitch motions, ζw values of non-dimensional transfer functions, φ/(kζw) depending
is the wave amplitude and k is the wave number. on L/λ ratio, for induced roll motions in longitudinal waves.
The comparison between the theoretical and experimental The maximum amplitude is obtained at the main resonance,
results, for the heave and pitch motions, reveals a good agree- both at zero speed (L/λ = 0.896) and design speed (L/λ = 0.443).
ment at zero speed. For the design speed the experimental transfer Figure 10 shows the experimental test arrangement in regular
function are smaller than the theoretical predicted ones. head waves.
For the ship running in longitudinal waves, the computer code The analysis of the experimental results given in Figure 9
does not calculate a solution of induced roll motion. However, the leads to the following observations:
experimental tests clearly demonstrate the existence of induced - the roll induced motions cover a small number of L/λ ratios,
rolling at the main resonance. Figure 9 shows the experimental being a typical phenomena of narrow frequency domain;

59(2008)4, 340-347 343

D. OBREJA, R. NEBERGOJ, L. CRUDU, S. PACURARU (POPOIU) PARAMETRIC ROLLING AT MAIN RESONANCE

- the model ship behaviour at zero speed in head waves, is

similar to the following waves case;
- the coupled induced roll, heave and pitch motions are obtained
in the second instability domain.
It is obvious that the development of the theoretical model
must consider the induced rolling at the main resonance as being
coupled with heave and pitch motions [7].

4 Numerical Modelling of Coupled Induced

Roll, Heave and Pitch Motions
Considering the ship moving in regular longitudinal waves,
heave (z), induced rolling (φ) and pitch (θ) motions were deter-
mined by numerical solving of the following coupled differential
equation system

( m + A33 )z + A34 ⋅ φ + A35 ⋅ θ = q1

z + ( I 4 + A44 )φ + A45 ⋅ θ = q2
A34 ⋅  (8)
z + A45 ⋅ φ + ( I 5 + A55 )θ = q3
A35 ⋅ 

q1 = F3a ⋅ cos ω t − ( B33 ⋅ z + C33 ⋅ z +

+ B ⋅ ϕ + B ⋅ θ + C ⋅ θ )
34 35 35

Figure 6 Non-dimensional transfer functions of pitch motion in q2 = F4 a ⋅ cos ω t − [ B34 ⋅ z + ( D01 +

regular head waves (9)
Slika 6 Bezdimenzionalna prijenosna funkcija posrtanja na pra- + D02 ⋅ ϕ )ϕ + g ⋅ m(GM +
vilnim valovima u pramac + K 3 ⋅ ϕ 2 )ϕ + B45 ⋅ θ ]
- the amplitude of the induced roll motion at design speed q3 = F5 a ⋅ cos ω t − ( B35 ⋅ z + C35 ⋅ z +
is smaller than the amplitude at zero speed, in longitudinal
waves; + B45 ⋅ ϕ + B55 ⋅ θ + C55 ⋅ θ )

Figure 7 Non-dimensional transfer functions of heave motion in regular following waves

Slika 7 Bezdimenzionalna prijenosna funkcija poniranja na pravilnim valovima u krmu

344 59(2008)4, 340-347

PARAMETRIC ROLLING AT MAIN RESONANCE D. OBREJA, R. NEBERGOJ, L. CRUDU, S. PACURARU (POPOIU)

Figure 9 Non-dimensional transfer functions of induced roll mo-

tion in regular longitudinal waves
Slika 9 Bezdimenzionalna prijenosna funkcija induciranog
ljuljanja na pravilnim uzdužnim valovima

Figure 8 Non-dimensional transfer functions of pitch motion in Figure 10 Experimental test in regular head waves
regular following waves Slika 10 Pokusi na pravilnim valovima u pramac
Slika 8 Bezdimenzionalna prijenosna funkcija posrtanja na
pravilnim valovima u krmu

Aij and Bij (i, j = 3, 4, 5) represent the added masses and the
damping coefficients, respectively, Cij are the restoring coef-
ficients and Fia (i=3, 4, 5) represent the amplitudes of excitation
forces and moments generated by the incident waves.
The evaluation of Aij components is based on theoretical and
experimental approach of the radiation problem [8], where the
coupled coefficients A34, D34, A45, B45 are considered as linear
functions of small heeling angle.
Solving the diffraction problem, the amplitudes of Fia force
were theoretically and experimentally determined [8]. Moreover,

59(2008)4, 340-347 345

D. OBREJA, R. NEBERGOJ, L. CRUDU, S. PACURARU (POPOIU) PARAMETRIC ROLLING AT MAIN RESONANCE

the amplitude of the excitation moment at induced rolling is a plitude slowly increases and becomes close to the experimental
linear function depending on the heeling angle measured value. The induced rolling amplitude depends on the
unit amplitude of roll excitation moment f4a, thus the excitation
F4 a = f4 aφ (10) coefficient of the induced roll motion can be defined as
f4 a
The damping coefficients of induced roll motion in longitu- μ= (11)
dinal waves, D01 and D02 were estimated based on experimental 2 ⋅ ω ⋅ ( I 4 + A44 )
2
φ

roll decay tests [4], analysed through energetic method proposed where, ωφ is the natural circular frequency.
by J.B. Roberts [9].
The hydrostatic and hydrodynamic values of the cargo ship 5 Conclusions
model are presented in Table 2.
The induced roll motion at parametric resonance represents a
Table 2 The hydrostatic and hydrodynamic values of the cargo complex physical process. The occurrence of induced roll motions
ship model
Tablica 2 Hidrostatički i hidrodinamički podaci za teretni brod Figure 11 Numerical and experimental results of the behaviour of
the ship model at zero speed, in regular following waves,
Ship model values at main resonance condition
Symbol Slika 11 Numerički i eksperimentalni rezultat ponašanja modela
(model scale 1:30)
broda pri nultoj brzini na pravilnim valovima u uvjetima
A33 170.97 kg glavne rezonancije
B33 682.3 kg/s
A55 64.582 kg⋅m2
B55 271.632 kg⋅m2/s
A44 0.71 kg⋅m2
D01 0.7848 kg⋅m2/s
D02 0.00592 kg⋅m2
A34 3.6325 ⋅˜ φÔ kg⋅m
B34 -13.853 ⋅˜ φÔ kg⋅m/s
A45 -0.2913 ⋅˜ φÔ kg⋅m2
B45 -2.142 ⋅˜ φÔ kg⋅m2/s
A35 -4.965 kg⋅m
B35 -30.082 kg⋅m/s
C33 10592.65 kg/s2
C44 85.122 kg⋅m2/s2
C55 4690.35 kg⋅m2/s2
C35 26.13 kg⋅m/s2
F3a 33.42 kg⋅m/s2
F4a 36.95 ⋅ φ kg⋅m2/s2
F5a 62.16 kg⋅m2/s2

The numerical solutions of the differential equation (8)

were found using the Gill version of the 4th order Runge-Kutta
method. The calculations were performed at zero speed, for the
case of induced roll motions in regular following waves, at the
corresponding values of L/λ = 0.896 and hw/λ=1/50, where hw is
the wave height. These conditions are specific for the parametric
resonance of the induced roll motions in the second instability
domain (main resonance). The initial conditions used to solve
the differential equations took into consideration the zero heeling
angle case. The experimental tests carried out under the above
mentioned conditions allowed measurement of heave, induced
roll and pitch motions, having the following amplitudes: z =
0.006 m, φ = 3.80 and θ = 1.80.
In Figure 11 the numerical and experimental results of the be-
haviour of the ship model at zero speed, in regular following waves
at the main resonance condition are exemplified. The correlation
between the numerical and experimental results is satisfactory.
As compared to heave and pitch motions, which are reaching
stabilised solutions for a small number of oscillations, the induced
roll motion has firstly a transitory phase and then the rolling am-

346 59(2008)4, 340-347

PARAMETRIC ROLLING AT MAIN RESONANCE D. OBREJA, R. NEBERGOJ, L. CRUDU, S. PACURARU (POPOIU)

in the instability domains is a result of multiple interdependence [2] NAYFEH, A.H., MOOK, D.T.: “Nonlinear Oscillation”, John
of physical parameters such as: the energy partially transferred Wiley & Sons, New York (1979).
to the roll mode of motion, the excitation coefficient depend- [3] NAYFEH, A.H., OH, I.G.: “Nonlinearly Coupled Pitch and
ing on the variation of the metacentric height on longitudinal Roll Motions in the Presence of Internal Resonance: Part I,
waves, the natural roll period, the ship speed, the wave length, Theory”, International Shipbuilding Progress, 42 (1995), p.
the heading angle, the encountering period, etc. Mention should 295-324.
be made that the diagrams given in Figures 3 and 4 suggest the [4] OBREJA, D.C., CRUDU, L.: “Experimental Tests for the In-
correlation between parametric resonance condition (1) and the vestigation of Roll Motions on Longitudinal and Beam Waves
relation between the encountering period and the incident wave for a Cargo Ship Model”, Technical Report, ICEPRONAV
period. Based on the above mentioned diagrams and instability Galati (2002).
domains analysis, the physical conditions of the induced rolling [5] BORODAY, I.K., MORENSCHILDT, V.A.: “Stability and
may be identified. Parametric Roll of Ships in Waves”, Proceedings of 3rd Inter-
The second instability domain was little analysed. This study national Conference on Stability and Ocean Vehicles, Gdansk
was performed by coupling the differential equations of induced (1986).
rolling with specific equations of heave and pitch motions. Ex- [6] FRANK, W.: “Oscillation of Cylinders in or Below the Free
cept for the restoring coefficients and the coupled hydrodynamic Surface of Deep Fluids”, Naval Ship Research and Develop-
coefficients, the remaining ones were experimentally measured ment Center, Washington, Report No. 2375 (1967).
during diffraction and radiation tests. The numerical solutions [7] HUA, J.: “A Study of the Parametrically Excited Roll Motion
were experimentally validated. of a RoRo-Ship in Following and Heading Waves”, Interna-
Numerical tests were performed in order to identify the physi- tional Shipbuilding Progress, 39 (1992), p. 345-366.
cal parameter having the main contribution on induced roll motion [8] OBREJA, D. C., NABERGOJ, R., CRUDU, L., PACURARU,
occurrence. It was demonstrated that the roll excitation moment S.: “Transverse Stability of a Cargo Ship at Parametric Roll-
had a decisive influence. In the considered case, a 10% reduction ing on Longitudinal Waves”, Proceeding of the ASME 27th
of roll excitation moment leads to the damping of induced roll International Conference on Offshore Mechanics and Arctic
motion. A 10% increment leads to a considerable amplification Engineering OMAE 2008, Estoril, Portugal (2008, accepted
of ship’s response. for publication).
The definition of the excitation coefficient, based on the [9] ROBERTS, J.B.: “Estimation of Nonlinear Ship Roll Damp-
experimental results of the diffraction problem, leads to a better ing for Free-Decay Data”, Journal of Ship Research, 29/2
understanding of the physical complex mechanism of the induced (1995), p. 127-138.
roll motions occurrence.

References
[1] KERWIN, J.E.: “Notes on Rolling in Longitudinal waves”,
International Shipbuilding Progress, 16 (Vol. 2, 1955), p.
597-614.

59(2008)4, 340-347 347