Ocean Engineering 120 (2016) 238–245

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Coupled vessel-dry-granular-cargo roll dynamics in regular beam seas

Christos C. Spandonidis, Kostas J. Spyrou n
School of Naval Architecture and Marine Engineering National Technical University of Athens, Greece

art ic l e i nf o a b s t r a c t

Article history: The behaviour of granular cargos transported by sea has not been sufficiently investigated yet from a
Received 1 October 2015 theoretical perspective. International regulations controlling the loading and stowage of bulky cargos,
Received in revised form despite some updating and improvement, remain empirical, lacking a solid theoretical foundation. In the
7 April 2016
paper is investigated the inception of roll instability for a vessel containing a rectangular hold partially
Accepted 21 April 2016
Available online 29 April 2016
filled with dry granular material, by applying a simulation approach based on coupled micro-scale
modelling of particles with vessel motion. The method of molecular dynamics is employed, combined
Keywords: with a rudimentary model of vessel motion in beam seas. The key assumptions behind the modelling of
Granular material cargo's movement are that, the granular material is dry, all particles are spherical and adjacent particles
Molecular dynamics
interact through non-linear elastic and frictional forces. Characteristic simulation results, including
Cargo shift
comparisons against cases of rigid cargo, are presented, pointing out cases where the agility of the cargo
Ship capsize
may have a positive or a negative effect on vessel's stability.
& 2016 Elsevier Ltd. All rights reserved.

1. Introduction
For a number of years, the problem of cargo shift is identified as
one of the most critical ship safety issues (INTERCARGO, 2012). The
recent improvement of international regulations concerning the
loading and stowage of ship cargos in bulky form was the result of,
almost entirely, empirical considerations, a fact implying the lack
of a fundamental theoretical framework for the problem (IMO,
2012). Indeed, approaches using micro-scale modelling of cargo
particles' motion and considering their interaction with the mov-
ing ship under wind/wave excitations are up-to-now practically
non-existent. A probable reason is because such a task is very
demanding, calling for an interdisciplinary approach for over-
coming the often fragmented nature of scientific efforts. As type of
problem, cargo shift could be classified along with sloshing; with
clear methodological analogies becoming prevalent when, for the
latter, a smooth particles hydrodynamics (SPH) modelling ap-
proach is applied. As well known, coupled ship motions affected
by liquid sloshing have been extensively studied in the past, since
high impacting pressures on tank walls constitute a perennial ship
operation hazard. Faltinsen and Timokha (2009) provided an in
depth presentation of analytical and numerical models for the
coupled fluid motion in a tank. Moreover, in Monaghan (2005) was
included a review of commonly applied SPH techniques. On the
other hand, cargo shift is a stick-slip phenomenon and, as a matter
n
Corresponding author.
E-mail address: k.spyrou@central.ntua.gr (K.J. Spyrou).
of fact, its effect on a vessel's motion at sea is very different from
the effect of sloshing. Concerning granular cargos behaviour, the
authors could not identify comprehensive research efforts of si-
milar standing. A category of interesting works have dealt, for
example, with the dynamics of sloshing cargos in silo vehicles (e.g.
Fleissner et al., 2009). However, in these no link to the ship motion
problem has been presumed.
A key element in the coupled vessel-granular cargo problem
formulation is the feedback from the ensemble of discontinuously
moving particles to vessel's body. Abrupt particle movement
(avalanching), leading to cargo shift can appear when a vessel
becomes subjected to large wave loads incurring on her significant
transverse acceleration. Such sudden movement of the cargo
constitutes the main contributing factor for a ship to acquire large
roll bias, a dangerous state wherefrom capsize is imminent. In
general, in terms of dynamics, the particles behave like oscillating
masses interacting with each other and also, in unison, with the
containing hold. Actually, significant quantities of energy can be
exchanged between the particles and the carrying body in a rather
intermittent manner. Such effects of interaction take the problem
out of the familiar domain of rigid body dynamics where ship
stability studies are customarily performed.
As part of research initiated at NTUA in 2011 for filling this gap,
the motions of granular materials when their hold is subjected to
prescribed oscillations have been already considered. In particular,
the mechanism of formation of the angle of repose (and associated
phenomena) has been studied (Spandonidis and Spyrou, 2012,
2013). A step further is taken in the current paper, studying how

http://dx.doi.org/10.1016/j.oceaneng.2016.04.023
0029-8018/& 2016 Elsevier Ltd. All rights reserved.
 C.C. Spandonidis, K.J. Spyrou / Ocean Engineering 120 (2016) 238–245 239

the ship and the granular material behave, as a system, under the
effect of wave excitation on ship's hull. In the longer term, the aim
is, to understand how the coupling between material and vessel
works, and subsequently, to create a suitable tool for assessing the
severity of this coupling effect, taking into consideration design
and operational parameters. Although a full ship motion model
would be indispensable for a comprehensive investigation of ship-
cargo dynamics in waves, at present we have restricted our at-
tention only to regular beam sea waves in deep water, applied on a
prism with rectangular cross section restrained to move only in
roll (1 DOF).

2. Aspects of behaviour of vibrated granular material

Granular flows find many industrial applications despite that

they often have a complex rheological character. Typically, a
granular material flow is comprised of a particulate solid in an
interstitial fluid (usually air) subjected to a shearing force. In the
Fig. 1. The three different coordinate systems.
most common granular material vibration experiments, the con-
sidered materials are sand, grain or glass spheres (usually 0.5–
3 mm diameter with standard deviation 5%). They are placed in a
rectangular or cylindrical container mounted on a rigid base which
is subjected to multiaxial, sinusoidal oscillations. Even these sim-
ple systems can exhibit surprisingly complex behaviour that has
yet to be fully explained. For example, the particle bed can behave
as a “cloud” of particles with little or no structure; and in other
cases the particle bed moves as a coherent mass. Depending on the
frequency and amplitude of acceleration, vibrated granular mate-
rials can give rise to various phenomena such as compaction,
convective flow, size segregation and “arching”. Particle re-
arrangements induced by vibrations lead to lower shear strength
and higher propensity to flow. In the full fluidisation regime, there
are no permanent contacts between particles and the system be-
haves as a dissipative gas. When particle accelerations remain
below the gravitational acceleration, the system keeps its static
nature and the vibrational energy propagates through a rather
compact network of inter-particle contacts. On the other hand,
vibrations at high frequency and low amplitude lead to slow Fig. 2. Displacement of mass centre as function of vessel inclination (the illustra-
tion is based on the yKz coordinate system).
(logarithmic) decay of the pore space as a function of time. Several
theories trying to explain this behaviour have been proposed (e.g.
Laroche et al., 1989; Gallas et al., 1992; Corwin et al., 2005); yet bottom left corner; but this one follows the motion of the hold in
none seem to be universally accepted. Another phenomenon that such a manner that its z-axis is always perpendicular to the bot-
is observed for deep beds is the formation of surface waves. The tom. System aPb is assumed to be at the centre of rotation of the
waves travel from the lowest point of the heap up the slope to the
vessel (which in the current study coincides with the middle point
peak but do not interfere with the continuous avalanche of par-
of the ideal flat free surface). This coordinate system is useful for
ticles associated with the convection pattern. The waves increase
monitoring vessel's rotation. All three systems of coordinates are
in length and decrease in height as they travel up the slope and
eventually disappear at the peak. The onset of the travelling sur- shown in Fig. 1.
face waves depends on the vibration- amplitude-to-particle-dia- In what follows, only the dynamics on the xy plane is under
meter ratio. Standing surface waves form at half the excitation investigation due to the 2D nature of the developed granular
frequency for certain ranges of the excitation amplitude and fre- material algorithm. Thus, the barge is assumed to be long enough
quency, following a mechanism similar to the Faraday instability (2 m) compared to her depth (0.3 m) and beam (0.34 m). These
(Douady and Fauve, 1988; Miles and Henderson, 1990). dimensions should facilitate direct comparison with the results of
experiments in the future.
Two different problems need to be tackled simultaneously: the
3. Equations of motion wave induced ship motion and the motion of the particles inside
the hold. For the latter, the position ri of each individual particle
The current analysis is performed in model scale and a small can be predicted, involving solution of Newton's equation of mo-
rectangular hold is considered. It is placed in a box barge that is
tion. The force acting on each particle i due to the interaction with
allowed to be rotated in roll around a specific axis; i.e. the roll
all its neighbouring particles is given by:
centre is fixed (see point P in Fig. 1). Each particle's motion is
monitored from an earth-fixed coordinate system, whose origin is N
placed at the initial location of hold's bottom left corner (ZOY ). Fi= ∑ Fij
j = 1, j ≠ i (1)
This coordinates system remains always horizontal (i.e. it does not
follow vessel's rotation). Another system, yKz , is also placed at the where the force between two interacting particles Fij is given by:
240 C.C. Spandonidis, K.J. Spyrou / Ocean Engineering 120 (2016) 238–245

42o
25o

Fig. 3. Views of cargo in inclined hold for 25° inclination (left); and 42° inclination (right). (For interpretation of the references to color in this figure, the reader is referred to
the web version of this article.)

t=0s t = 90 s

e
Fig. 4. State of cargo when rotation begins (left); and on return to initial position after half a period of vessel oscillation (right). The tilting rate is 1°/s.

⎧ Fnor +Ftan, ξ >0
Fij=⎨
⎩ 0, ξ≤0
and ξ is the penetration distance between two particles (with radii
Ri and Rj ) given by:
ξ=max ( 0,Ri+Rj− rij )
Normally to the line of contact of particles during collision have
been considered nonlinear elastic and frictional forces ( Fnor ); while
transversely it was assumed acting a purely frictional force (Ftan ):
Fnor = − k n ξ 3/2−γn ξ1/2ξ ̇
( )
Ftan= − min γs us , μFn ·sign ( us )
γn and γs are normal and shear damping constants respectively;
k n is non-linear stiffness coefficient, us is the shear velocity com-
ponent and μ stands for the dynamic friction coefficient (for de-
tails see Spandonidis and Spyrou (2012)).
Collision forces are computed for each pair of particles, de-
pending on their proximity, at every time step. The forces that act
on the side plates of the hold are derived in a similar manner, by
considering a fictitious array of spherical particles in contact with
each other and fixed on the walls. Our modelling follows in gen-
eral the “molecular dynamics” approach. Detailed description of
our method including the adopted interaction forces and an in-
depth comparison with other similar models is found in Spando-
nidis and Spyrou (2013).
(2)
Ship motion is predicted by solving Newton's equation at every
time step. The hull is considered receiving loads due to inertia,
hydrodynamic damping, restoring (with consideration of time
varying centre of mass of cargo) and the impacting moment due to
(3) possible particles' impact with hold's side walls. Vessel's roll mo-
tion can then be described as follows:
. . .
(I +δΙ ) φ¨+Β1φ+Β2 φ φ +MR=mg GM Ak·ω2· cos (ωt )+MI (6)
φ is the absolute roll angle, ω is the wave - vessel encounter
(4) frequency, GM is the metacentric height, I and δΙ are respectively
the roll moment of inertia and the “added” moment of inertia, Β1
(5) and Β2 are the linear and the quadratic roll damping coefficients. δI
and Β1 are frequency dependent while Β2 was assumed as fre-
quency independent (Delorme et al., 2006). MR is the restoring
moment calculated as the sum of three moments ( MB due to
buoyancy, MS due to ship's weight and MC due to cargo's weight)
calculated around the fixed axis of rotation:
MR=MB+MS +MC (7)
Moreover, the impact moment MI is calculated by:
K
MI = ∑ Mi
i=1 (8)
where Mi is the moment due to the contact force Fi, calculated on
 C.C. Spandonidis, K.J. Spyrou / Ocean Engineering 120 (2016) 238–245
Fig. 5. Comparison of response of solid and granular cargo. The wave amplitude is fixed at 0.01 m and the frequency is varied as follows (from top to bottom): (a) 3.5 rad/s,
(b) 4.4 rad/s, (c) 4.7 rad/s and (d) 6.5 rad/s.
Fig. 6. Phase plots of vessel response (model scale) when carrying solid (dashed
line) and granular (continuous line) cargo, for wave amplitude 0.01 m and fre-
quency 4.4 rad/s.
the basis of Eq. (1), acting on a particle i of the wall. This force is
produced by interaction with neighbouring cargo particles having,
with respect to particle i, nonzero relative velocity. K is the total
number of hold's wall particles.
In order to verify our results, reliability tests were performed,
as proposed by Haile (1997). Different operating systems (Win-
dows 8, Windows XP, Linux RedHat 5.1), hardware configuration
and computational software (Microsoft Visual C þ þ, Mathematica,
Matlab) were used, in order to calculate the systematic error. Also,
several repetitions were performed for each numerical experiment
in order to check conservation principles and monitor the statis-
tical error. The algorithm can provide accurate calculations with a
statistical and systematic error whose deviation does not exceed
2%. Validation of the coupled motion results could not be done at
this stage due to lack of suitable experimental measurements and
241
we relied on separate validation of the two sub-algorithms. For the
validation of the molecular dynamics algorithm we refer to
Spandonidis and Spyrou (2013). Concerning the ship motion
equation, results for “empty hold” were compared against similar
experimental results obtained by Rognebakke and Faltinsen
(2003) and by Murashige and Aihara (1998), indicating that the
algorithm performs very well, with a standard deviation less than
1%.
4. Case study
We assumed the hold partly filled with cellulose acetate par-
ticles of diameter 3 mm. On the yz plane appear 2000 particles.
Particle parameters (friction, stiffness, etc.) coincide with those
experimentally measured by Foerster et al. (1994). The draught at
the initial position (with the material inside) is 0.11 m. The ma-
terial height-to-width-ratio is 0.36 which corresponds clearly to a
“finite” material depth case. The vessel was free to respond only in
the roll direction, under an external excitation owed to waves
having fixed amplitude and frequency. The particles were con-
sidered as smooth spheres with diameters that could be either
fixed or varied within a predefined range (standard deviation 1%).
They can be translated (horizontally-vertically) and rotated. At this
specific stage we assumed the particles as completely dry objects
and the weight of the cargo equalled that of the vessel itself (i.e.
the weight of the carrier was considered negligible). Several tests
were performed before beginning the actual numerical experi-
ments, in order to identify the parameters that are critical for
behaviour. At first step, the granular material was left to balance in
calm water. As the particles do not always acquire exactly the
same location at every numerical experiment initialisation, the
draught of the barge and its initial roll angle were calculated for
242 C.C. Spandonidis, K.J. Spyrou / Ocean Engineering 120 (2016) 238–245

Fig. 7. Mean roll amplitude (up) and absolute maximum roll angle (down), for Fig. 8. Comparison of responses of solid and granular cargo: (up) wave amplitude
0.01 m excitation amplitude. Their difference is indicative of asymmetric vessel 0.005 m and frequency 4.5 rad/s; (down) wave amplitude 0.015 m and frequency
response. 4.3 rad/s.

several slightly different initial free surface configurations. The

eigenperiod of the loaded barge was designed to lie within
acceptable (for a realistic model-scale scenario) limits.

4.1. Identification of angle of repose

The tilting table method described in the International Mar-

itime Solid Bulk Cargoes Code (IMSBC) was applied in numerical
context (IMO, 2012). On this basis, we determined the critical in-
clinations for the realisation of cargo shift. In Fig. 2 is depicted the
horizontal displacement of the centre of mass versus the tilting
angle, for various tilting rates. When a slow rate is applied (below
1 rad/s) two critical angles are identified, at about 25° and 42°. The
former one is the angle where heap formation occurs. However, as
the true angle of repose should be considered the second one (for
further evidence on this see Spandonidis and Spyrou (2013)). On
the other hand, when a larger tilting ratio is applied, the flow of
Fig. 9. Mean roll amplitude for 0.017 m incident wave amplitude.
cargo realised at the first critical angle becomes more profound.
In Fig. 3 is shown cargo's state inside the hold at specific time
instants when critical inclinations are reached. Red colour is used the initial cargo surface configuration (and hence as preference of
for indicating that a particle has kinetic energy close to zero while the cargo for retaining its shifted position) when hold's rotation is
the blue colour is used when it has clearly nonzero kinetic energy. reversed. More specifically, in the slow tilting scenario, soon after
In this way, the actual percentage of material mass that is involved the angle of repose (42°) was exceeded, the tilting rate was re-
in cargo shift can be directly illustrated. versed until the hold was returned to its initial horizontal position
4.2. Hysteretic behaviour (0°). In Fig. 4 appear time-shots indicating the final granular ma-
terial configuration, displaying its very weak tendency for re-
Taking a step further, we investigated the nonlinear phenom- turning to its initial state as the hold is brought back to the upright
enon of hysteresis, manifested by the lack of reversibility towards position.
 C.C. Spandonidis, K.J. Spyrou / Ocean Engineering 120 (2016) 238–245 243

4.3. Numerical experiments for coupled motions of ship and cargo

Two kinds of additional numerical experiments were per-

formed, for the purpose of finding out the effect of shifting cargo
on vessel's motion. The first was basically an artificial reference
case and the vessel was loaded with what will be identified
hereafter as solid cargo; that is, cargo having weight and mass
centre identical to those of the granular cargo, but lacking the
freedom to move independently. The second series involved tests
with actual granular cargo where cargo shift can be realized, af-
fecting back vessel's motion and so forth. The wave frequency and
amplitude regions of interest were 2.3–6.5 rad/s and 0.005–
0.025 m, respectively.
In Fig. 5 are shown some examples of comparison of behaviour,
for four different frequency values, with wave amplitude fixed at
0.01 m. The key findings are the following:
Fig. 10. Comparison of responses for the solid and granular cargo cases, when the
wave amplitude is 0.0017 m and the frequency is 4.5 rad/s. Capsize is realized in the
case of the granular cargo. a) For wave frequency values below 3.5 and above 6 rad/s, bar-
ge's response is of low importance [Fig. 5a and d]. After a short
transient, a peak roll angle no more than 5° was reached. Since

t = 7.1 s t = 7.25 s t = 7.9 s

t = 8.4 s t = 8.6 s t = 9.25 s

t = 10.05 s t = 10.15 s t = 10.75 s

Fig. 11. State of the cargo inside the hold during the three last roll cycles (the 6th, the 7th and the 8th) before capsize. The creation of a substantial roll bias due to cargo's
shift to port is noticed.
244 C.C. Spandonidis, K.J. Spyrou / Ocean Engineering 120 (2016) 238–245
the granular material did not perform any substantial motion,
barge's behaviour did not depend on whether the cargo was
movable.
b) For wave frequency values between 3.5 and 6 rad/s, the barge
entered a resonance region. For the case of solid cargo (see
Fig. 5b), the peak of roll amplitude appeared at frequency
4.4 rad/s and a maximum roll angle of approximately 36° was
reached. With the cargo in granular form, the system showed a
tendency to move less. The peak roll angle reached was less than
30°. The lower response is because, whilst the granular material
reaches the first critical angle of heap formation, however, the
phase of cargo's movement opposes that of the vessel, and thus
the overall motion is reduced. The calculation of moments
indicated that, the moment due to cargo's mass centre, and not
the impact moment, is the primary cause of this effect.
c) Further observation of Fig. 5b led to additional findings. Firstly,
the motion of the vessel with the granular cargo has become
asymmetrical. Local movement of particles around the one
end of the free surface, led to a lower absolute roll peak when
the hold rotated clockwise. Secondly, for certain values of
wave frequency, the steady-state is characterised by more than
one frequency. As confirmed from Fig. 5b and c, where a 50 s
simulation result is shown, this phenomenon is not transient
but it represents the true long-term response. The last
observation is better depicted in Fig. 6, where the vessel's
response phase portrait for wave amplitude 0.01 m and
frequency 4.4 rad/s is presented.
d) Inside the resonant region and after a critical value of 4.7 rad/s,
the two systems produce almost identical mean roll responses.
Despite though their macroscopically similar behaviour, the
initial movement of the granular cargo during the transient
stage creates larger roll. For the case of 4.7 rad/s (Fig. 5c), both
systems oscillate with mean roll amplitude of 30°. But for the
granular cargo, an almost 2° dynamic roll bias is incurred.
These findings are summarised in Fig. 7.
For wave amplitudes between 0.005 and 0.015 m, our simula-
tions yielded that the behaviour remains qualitatively similar to
that observed for 0.01 m wave amplitude.
In Fig. 8 appear comparisons at the two ends of the investigated
region of wave amplitude [0.005 m (up) and 0.015 m (down)-wave
frequency, respectively, 4.5 and 4.3 rad/s]. Notably, for wave am-
plitude 0.015 m, the granular cargo case is associated with smaller
response but with a “clear” second frequency existing in the steady
state due to material's displacement.
Further increase of the excitation leads however to a sub-
stantially different behaviour (Fig. 9): for wave amplitude 0.017 m
and frequency 4.5°/s, the material inside the hold moves in such a
way that the barge cannot return towards the upright position and
capsize is finally realized.
To investigate deeper this behaviour, we have looked into the
first seconds of this motion. In Fig. 10 are illustrated, in com-
parative manner, the first 20 s of the motion, for the two systems.
At about the 10th second, the barge with the granular cargo cap-
sized. Whilst she initially behaved in a similar to the 0.01 m wave
amplitude case (that is, smaller mean roll angle and quasi-steady
state response), soon a large peak value was invoked (almost 45°).
This is above the material's second critical angle associated with
its angle of repose. In the ensuing cycle the roll angle grew even
further and capsize finally occurred. Visual inspection of the cargo
inside the hold indicated that, due to high acceleration, significant
amount of material was displaced to the one side, remaining there
during the two roll cycles before the capsize (see the simulation
snapshots of Fig. 11). For wave amplitudes higher than 0.017 m, a
qualitatively similar behaviour was observed.
5. Concluding remarks
A step towards a systematic numerical investigation of the
coupled granular-material-and-vessel- motion in regular beam
seas was described. The vessel was allowed to move only in the
roll direction, macroscopically performing thus a restricted 1 DOF
motion. Whilst the granular material was considered here as dry,
cases of wet cargo can also be examined by directly expanding on
the described microscopic approach.
Several values of wave amplitude and frequency were applied
on a scaled container, partly filled with dry spherical particles, in
order to capture the response of the system. Special attention was
paid to the comparison of the results against corresponding cases
where the cargo behaved like a rigid body, insensitive to vessel's
oscillations.
The results indicated that, in the case of granular cargo and for
wave amplitudes below a certain limit, vessel motion is generally
milder. This means that, for low wave amplitudes a vessel carrying
granular cargo is even more stable, even though it appears having
some difficulty to reach steady state. For higher wave amplitudes,
as a result of significant cargo movement, the barge exhibits larger
response. For wave amplitude 0.017 m and frequency 4.5°/s, the
material inside the hold moves in such a way that the barge cannot
return towards the upright position and capsize is finally realized.
Further consideration of this behaviour revealed that, it is related
with a shift of the resonance peak value.
The experimental reproduction of the presented results will of
course add extra confidence on the relevance of the above find-
ings. Other steps that we take in the direction of corroborating and
establishing patterns of behaviour are: (a) a new series of nu-
merical experiments for different material-height-to-hold-width
ratios, various wave steepnesses, more than one frequency wave
packets, different material parameters and different vessel shapes.
(b) Dynamical analysis tools are being considered for capturing
more rigorously vessel dynamics as influenced by the granular
material's movement. (c) Last but not least, an improved mathe-
matical model, featuring the coupled roll, sway and heave motions
in beam seas is currently under verification/validation.
Acknowledgements
The present work was supported by the Greek General Secre-
tariat for Research and Technology (Ministry of Education and
Religious Affairs) – project category Excellence (“Aristeia”) – 1,
project title “HOMSHIP” with contract Reference number GSRT-
252.
References
Corwin, E.I., Jaeger, H.M., Nagel, S.R., 2005. Structural signature of jamming in
granular media. Nature 435, 1075–1078.
Delorme L., Bulian GMc Cue., L., Souto-Iglesias A., 2006. Coupling between sloshing
and ship roll motion: Comparison between first order potential theory and SPH.
In: Proceedings of the 26th Symposium on Naval Hydrodynamics, Italian Ship
Model Basin (INSEAN), US Office of Naval Research (ONR), pp. 17–22.
Douady, S., Fauve, S., 1988. Pattern selection in Faraday instability. Eur. Lett. 6,
221–226.
Faltinsen, O.M., Timokha, A.N., 2009. Sloshing. Cambridge University Press, New
York.
Fleissner, F., Alessandro, V., Schiehlen Wand, Eberhard, P., 2009. Sloshing cargo in
silo vehicles. J. Mech. Sci. Technol. 23, 968–973.
Foerster, S.F., Louge, M.Y., Chang, H., Allia, K., 1994. Measurements of the collision
properties of small spheres. Phys. Fluids 6, l108.
Gallas, J.A.C., Herrmann, H.J., Sokolowski, S., 1992. Convection cells in vibrating
granular media. Phys. Rev. Lett. 69, 1371–1374.
Haile, J.M., 1997. Molecular Dynamics Simulation: Elementary Methods. Wiley -
Interscience, New York, ISBN-13: 978-0471184393.
International Association of Dry Cargo Shipowners (INTERCARGO), 2012.
 C.C. Spandonidis, K.J. Spyrou / Ocean Engineering 120 (2016) 238–245
Benchmarking Bulk Carriers 2011–2012. INTERCARGO, London.
International Maritime Organization (IMO), 2012. Solid Bulk Cargoes Code [IMSB
Code 268(85)]. IMO Publishing, London, ISBN: 978-92-801-1535-2.
Laroche, C., Douady, S., Fauve, S., 1989. Convective flow of granular masses under
vertical vibrations. J. Phys. France 50, 699–706.
Miles, J., Henderson, D., 1990. Parametrically forced surface waves. Annu. Rev. Fluid
Mech. 22, 143–165.
Monaghan, J.J., 2005. Smoothed particle hydrodynamics. Rep. Progress. Phys. 68,
1703–1759.
245
Murashige, S., Aihara, K., 1998. Coexistence of periodic roll motion and chaotic one
in a forced flooded ship. Int. J. Bifurc. Chaos 8, 619–626.
Rognebakke, O.F., Faltinsen, O.M., 2003. Coupling of sloshing and ship motions. J.
Ship Res. 47 (3), 208–221.
Spandonidis C., Spyrou J., 2012. Use of granular material dynamics simulation for
the study of cargo shift of ships. In: Proceedings of the 11th International
Conference on the Stability of Ships and Ocean Vehicles, Athens, pp. 497–507.
Spandonidis, C., Spyrou, K.J., 2013. Μicro-scale modelling of excited granular ship
cargos: a numerical approach. Ocean Eng. 74, 22–36.