Ocean Engineering 253 (2022) 111090

Contents lists available at ScienceDirect

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

An assessment of mooring systems’ forces of ships berthed at dolphins

Arina van L.B. de Carvalho *, Eduardo M.B. Campello, Guilherme R. Franzini, Kalil J. Skaf
Department of Structural and Geotechnical Engineering, Polytechnic School, University of São Paulo, P.O. Box 61548, 05424-970, São Paulo, SP, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: In the current practice of structural design for berth structures, the mooring forces (tension in lines and
Mooring lines compression in fenders) are typically determined through simplified linear static analyses, and then magnified by
Marine fenders dynamic amplification coefficients. It is known, however, that the mooring’s stiffness presents nonlinearities
Mooring forces
(both physical and geometrical) that may at times become relevant, raising the issue of whether linear analyses
Nonlinear static analysis
Linear dynamic analysis
are really adequate. Also, the environmental loads induce oscillatory behavior whose peak values may not always
be satisfactorily estimated through dynamic amplification coefficients, which are far from being well defined in
the technical literature. This work presents an investigation on the forces developed on mooring systems of
vessels berthed at dolphins, and the consequences of the current design practice’s simplifications. Different hi­
erarchical models are defined (linear static, nonlinear static and linear dynamic in the frequency domain) for a
case study of a Floating Storage and Regasification Unit (FSRU) permanently berthed at dolphins at the Bar­
carena’s Liquefied Natural Gas (LNG) terminal in Pará, Brazil. The vessel’s displacements and their peak values
are determined through stochastic and statistical theories. The corresponding forces on the mooring system are
computed and the dynamic amplification coefficients therefrom obtained. A thorough discussion on the results is
provided. The methodology presented may be a useful tool to ascertain the dynamic amplification coefficients
required in the simplified methodologies from the international standards.

1. Introduction to mention the nonlinear behavior) of the mooring elements. Plus, they
estimate the dynamic response (e.g., peak values due to the oscillatory
The operational safety of a port terminal is ensured by the proper environmental loads) through the use of dynamic amplification co­
restriction of movements of berthed vessels during loading and efficients. Albeit the limitations, they are very convenient for design and
unloading operations. The main forces acting on a vessel at berth come analysis and examples of their usage can be found in the works of Das
from environmental sources, which are random and have an oscillatory et al. (2015), Comin and De Souza (2017), Gaythwaite (2014) and
nature. The movements’ restrictions on the horizontal plane are given Bianco (2015), among others.
through mooring lines and fenders, which comprise the vessel’s mooring It is known, however, that the mooring stiffness presents non­
system. The lines typically consist of steel or synthetic wire ropes that linearities (both physical, such as the fenders’ softening behavior due to
fasten the ship to the berth’s structure. The fenders, in turn, are their elastomeric nature, and geometrical, such as the occasional loss of
buckling-column-type units that act to reduce the reaction forces from contact between fenders and the ship) that may at times become rele­
the berth structure onto the ship’s hull, and to restrict the vessel’s vant, raising the issue of whether linear analyses are really adequate. In
movements (especially in the longitudinal or surge direction) by friction addition, peak values from the dynamic response may not always be
forces. The current practice of structural design for berth structures is satisfactorily estimated through dynamic amplification coefficients,
based on the relevant international standards, such as the widely used which are far from being well defined in the technical literature (see, e.
Spanish standard Recomendaciones para Obras Marítimas 2.0-11 (ROM g., Barros and Mazzilli (2018) and Barros (2018)). This work aims to
2.0-11, 2012), which prescribe the computation of the mooring forces investigate the forces developed on mooring systems of berthed ships
(lines’ tension and fenders’ compression) through simplified methods. through the use of more elaborate (hierarchically superior) models,
These methods typically consist of a linear static analysis that decouples namely, a nonlinear static model and a linear dynamics model, and
the vessel’s movements and does not consider the relative stiffness (not compare the results against those obtained with the simplified methods.

* Corresponding author.
E-mail addresses: [email protected] (A.L.B. de Carvalho), [email protected] (E.M.B. Campello).

https://doi.org/10.1016/j.oceaneng.2022.111090
Received 18 July 2021; Received in revised form 4 February 2022; Accepted 11 March 2022
Available online 29 April 2022
0029-8018/© 2022 Elsevier Ltd. All rights reserved.
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

The models are applied to a case study of a natural gas regasification Debabrata (2015) and De Wilde et al. (2009). There is a rich literature on
terminal at the Port of Barcarena (Pará), Brazil, consisting of a Floating mooring systems of berthed vessels and berth structures, including both
Storage Regasification Unit (FSRU) permanently berthed at dolphins theoretical and practical aspects, along with methods for their forces’
and subjected to environmental wind and current loads. For the computation, analysis and design. We will not address these aspects
nonlinear static model, the nonlinearities considered are both physical here, nor will comment on their myriad applications, but instead refer
(stemming from the fenders’ elastomeric stiffness and the interested reader to some of the established literature on the field,
buckling-column-type behavior) and geometrical (caused by the slack­ and the various references therein, such as Gaythwaite (2004) and
ing of the mooring lines and the occasional loss of contact between the (2014), Watai et al. (2018), Yue et al. (2020), Rosa-Santos et al. (2014),
vessel and fenders). The model describes the vessel’s motion in the Faltinsen (1990) and Molin (2002).
horizontal plane through its in-plane degrees of freedom (surge, sway The paper is organized as follows. Section 2 presents a brief summary
and yaw), leading to a system of nonlinear equations that is iteratively of the simplified (linear static) method from the Spanish standard ROM
solved through a Newton-Raphson scheme. For the linear dynamics 2.0-11 (2012). Section 3 brings our nonlinear static model. Section 4
model, in turn, the problem is formulated in the frequency domain presents the linear dynamics model, including a detailed description of
taking into account the random and oscillatory nature of the environ­ the wind’s representation. Section 5 describes our case study, including
mental loads. To this aim, firstly the system’s stiffness is linearized the results obtained with the three models and a thorough discussion.
around the equilibrium configuration obtained with the nonlinear static Finally, section 6 showcases our conclusions and final considerations.
model under mean wind and current loads. Next, the hydrodynamic Throughout the text, italic letters (a, b, …, A, B, …) represent scalar
coefficients of added mass and damping are computed. Then, the dy­ quantities, while boldface-italic letters (a, b, …, A, B, …) represent vec­
namic displacements and their peak values are determined through tors and matrices.
stochastic and statistical theories, in which we resort to a special power
spectral density to represent the wind load. Finally, the mooring forces 2. Simplified method from the Spanish Standard
at peak displacements are computed and the dynamic amplification
coefficients therefrom obtained. Results are compared to those obtained The simplified, linear static method from the Spanish standard ROM
with the nonlinear static model and the simplified methodology from the 2.0-11 (2012) assumes an a priori known mooring configuration (i.e.,
Spanish standard ROM 2.0-11 (2012), and the dynamic amplification known tensioned lines and compressed fenders) and small displace­
coefficients required in these latter are verified. ments and rotations, such that the ship’s equilibrium position may be
Besides ascertaining the validity of the simplified methods, this work taken the same as the initial (undisplaced) one. Also, it is based on a
also brings some original features into the models devised. First, the simplification of the environmental actions’ oscillatory nature, whereby
fenders’ nonlinear elastomeric response is taken into account, in a balance is established between the static component (i.e., mean value)
contrast to other approaches from the literature such as Barros and of the active environmental actions and the reaction forces caused by the
Mazzilli (2018) and Barros (2018). Second, for the linear dynamic mooring lines’ stretching and the fenders’ compression. The mooring
model, in the cases where the system’s stiffness is not locally linear (as system is responsible for restriction of the ship’s horizontal movements,
required for the analysis in the frequency domain), we propose a simple i.e., the vessel’s longitudinal and transversal translations and rotation
iterative procedure to compute the stiffness at the maximum dynamic around the vertical axis passing on its center of gravity (surge, sway and
displacements configuration. This may occur, e.g., when the maximum yaw, respectively). A three degrees-of-freedom (DOFs) model follows
dynamic displacements slackens one or more lines or causes a loss of therefrom. The method assumes that the forces’ distribution between
contact between the vessel and any of the fenders. Third, the frequency cables and fenders depends only on the adopted arrangement’s geom­
domain model follows the so-called direct approach. This implies that etry and the elements’ constitutive properties. It follows that the vessel’s
the results are exact (within the assumptions made) and, since the movements along the three DOFs end up entirely uncoupled. The system
damping is non-proportional, no approximations are necessary, in is in static equilibrium when the applied environmental actions, given
contrast to the usual normal mode approach, widely followed by other by the mean values of the longitudinal and transversal forces and the
authors (see, e.g., Barros and Mazzilli (2018), and references therein). moment around the vessel’s vertical axis (R|L , R|T , R|M , respectively),
Another distinct feature is that the maximum dynamic displacements are balance the lines’ and fenders’ restoring forces and moments (the latter
estimated based on extreme-value distribution statistical theory through computed w.r.t. the vessel’s center of gravity). The equilibrium equa­
the so-called peak factors, differently from what is observed in other tions can be written as (see Fig. 1):
works such as Feikema and Wichers (1991) and Li et al. (2019). Finally, ∑ ∑
the methodology herein presented assembles in a didactic and system­ ΔQm,i|L + ΔQf ,j|L = R|L , (1)
i j
atical way several aspects that are somewhat scattered or fragmented in
the literature (especially for the linear dynamics model), hopefully ∑ ∑
ΔQm,i|T + ΔQf ,j|T = R|T , (2)
helping to guide newcomers to the field and analysts interested in the i j
design of port structures. This work emphasizes the importance of the
∑ ∑
adoption of a proper dynamic amplification coefficient for static models. ΔQm,i|T am,i + ΔQf ,j|T bf ,j = R|M , (3)
In the absence of specific information, we suggest how to estimate this i j
coefficient in a rigorous and safe manner through a linear dynamic
analysis in the frequency domain. Further details on the presented where ΔQm,i|L and ΔQm,i|T are the ith line’s longitudinal and transversal
model can be found in the first author’s dissertation at Carvalho (2021). tension (increase or decrease) components, ΔQf,j|L and ΔQf,j|T are the jth
We remark that a dynamic model in the time domain would also be fender’s longitudinal (friction) and transversal reaction (increase or
possible to be devised. This approach, however, brings additional dif­ decrease) components, and am,i and bf,j are the ith mooring chock’s and jth
ficulties such as the definition of the time-series of the environmental fenders’ horizontal coordinates as measured from the ship’s center of
loads, the need for a continuous update of the system’s stiffness, added gravity (see Fig. 1). The above system of equations is statically inde­
mass and damping (which vary with the ship’s response) throughout the terminate, with multiple unknowns (due to the number of lines and
solution, and the numerical time-integration of the equations of motion fenders) but only three balance equations. To overcome the indetermi­
(which is not a trivial task for long-term dynamics computations). This is nacy, and facilitate the design practice, the Spanish standard introduces
the next level in the modeling hierarchy and is left outside the scope of some simplifications, allowing one to estimate the mooring forces in a
the present work. Examples of recent contributions on this regard can be rather straightforward way. In the case of a vessel laterally moored at
found in Ruggeri et al. (2018), Watai et al. (2015), Shivaji Ganesan and

2
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

Fig. 1. (a) Static equilibrium, (b) mooring lines’ tension and (c) fenders’ compression according to ROM’s definitions. Adapted from Recomendaciones para Obras
Marítimas (2012).

dolphins with the resulting forces acting to move it towards the struc­ components, which induce the ship’s longitudinal and transverse dis­
ture, the so-called Method 1 (using the standard’s terminology) is placements as well as rotation w.r.t. the vertical axis, (ucg , vcg , θ).
applied. When the resulting forces move the vessel away from the berth Defining the coordinates of a material point on the ship’s hull at the
structure (i.e., towards the sea), the so-called Method 2 (again, using the reference configuration by (xR , yR , zR ), and assuming small displace­
standard’s terminology) is recommended instead. Both are described in ments and rotations such that cos θ ≅ 1 and sin θ ≅ θ, it is possible to
the Spanish standard ROM 2.0-11 (2012). This approach provides the describe the point’s displacements on the horizontal plane as follows
lines’ tension and fenders’ compression in terms of the environmental ( )
actions’ mean values. To account for their oscillatory nature, the u = ucg − yR − ycg,R θ, (4)
Spanish standard recommends the use of dynamic amplification co­ ( )
efficients (γ D ) in order to estimate peak values. Such coefficients depend v = vcg + xR − xcg,R θ, (5)
on a number of factors, such as the vessel’s characteristics, the envi­
where u and v are the point’s longitudinal and transversal displace­
ronmental actions themselves, the mooring system’s geometry, the berth
ments, respectively, and θ its rotation, which is the same as the center of
structure, etc. A closed-form expression, however, is not available, and
gravity’s rotation. The point’s coordinates (x, y) at the current configu­
in the absence of more specific information, the use of γ D = 2.0 is rec­
ration are then obtained by
ommended (supposedly on the safe side). In this work, we adopt γD =
( )
1.5, which follows from the authors’ design practice experience. This x = xR + ucg − yR − ycg,R θ , (6)
value will be ascertained by the dynamic analysis in section 5.
( )
y = yR + vcg + xR − xcg,R θ , (7)
3. Nonlinear static model
Each mooring line i is represented by a linear spring whose direction
The nonlinear static model follows a similar framework as that of is given by the vector pointing from the line’s chock on the ship to the
Barros and Mazzilli (2018) and Barros (2018), but with some distinct respective bollard on the dolphin, the latter being assumed fixed during
features. Accordingly, the ship is represented as a two-dimensional rigid the vessel’s movement. The line’s geometry (both direction and elon­
body defined by the projection of the hull’s contour line on the hori­ gation) is constantly updated according to the displacements of the
zontal plane, whereas the mooring lines and fenders are described as chock, using Eqs. (6) and (7). Assuming a linear elastic constitutive
oriented (linear and nonlinear) springs. The position of the ship’s center behavior, and designating by ℓ0,i and ℓm,i the line’s initial and final
of gravity in the reference configuration is given by coordinates (xcg,R , lengths, respectively, from Hooke’s law it follows that the tension tm,i in
ycg,R , zcg,R ), as show in Fig. 2. Similarly to the linear static model, the line i is:
environmental actions are considered through their static (mean value)

Fig. 2. Displacements of the vessel’s center of gravity and of a generic material point on the ship’s hull (left, magnified for better visualization) and the mooring
lines’ vector representation (right).

3
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

Ei Ai ( ) ( ) the cross-product between the vector-distance from the vessel’s center of

tm,i = ℓm,i − ℓ0,i = km,i ℓm,i − ℓ0,i , (8)
ℓ0,i gravity to the fender’s contact point with the ship hull and the fender’s
reaction vector.
where km,i is the line’s stiffness. If the line slackens, such that ℓm,i < ℓ0,i , To avoid initial slackening of the lines and at the same time to pre-
its tension is enforced to zero. We define the line’s tension vector by compress the fenders in order to reduce the vessels’ motions, a pre-
{ } tension of 5–10% of the lines’ minimum breaking load (MBL) is usu­
tm,i = tm,i vm,i = tm,i vmx,i , vmy,i , vmz,i , (9)
ally applied (see e.g. Gaythwaite (2004)). This is achieved here in a
preliminary analysis before application of the environmental loads,
with vm,i as the line’s unit vector and vmx,i , vmy,i and vmz,i its components
allowing for an initial (pre-stressed) equilibrium configuration. Finally,
in the longitudinal, transversal and vertical directions, respectively. If
the balance of forces and moments related to movements in the hori­
the line is tensioned, it induces a moment mmz,i on the ship around the
zontal plane is achieved by enforcing equilibrium w.r.t. the vessel’s
vertical axis passing on its center of gravity, which is obtained from the
center of gravity. For a mooring system with m cables, n fenders and
cross-product between the vector-distance from the vessel’s center of
applied environmental forces (mean static values) given by R|L (longi­
gravity to the line’s chock and the line’s tension vector.
According to the Oil Companies International Forum (OCIMF, 2008), tudinal), R|T (transversal) and R|M (moment), respectively, we have:
for very high stiffness lines, like steel or HMPE (High-Modulus Poly­ ( ) ∑m

ethylene) lines, it is usual to adopt a more flexible material at the tails as Rx ucg , vcg , θ = tm,i vmx,i + R|L = 0, (13)
to improve load distribution between the lines. In this case, each set (line
i=1

plus tail) works like two springs in series. Naming kline and ktail , ( ) ∑m ∑
n
respectively, the high-strength and tail flexible material’s stiffnesses, the Ry ucg , vcg , θ = tm,i vmy,i + rf ,j + R|T = 0, (14)
total or equivalent stiffness km of the set can be obtained by i=1 j=1

1 1 1 ) ∑m ∑
n
= + . (10) (
Rθ ucg , vcg , θ = mmz,i + mfz,j + R|M = 0 . (15)
km ktail kline
i=1 j=1
The mooring fenders are fixed equipment at the dolphin’s structures
Expressions (13) to (15) comprise a system of nonlinear equations
and are taken with fixed coordinates. From the ship’s hull contour line in
with three unknowns, namely, the vessel’s longitudinal and transversal
the current configuration, it is possible to compute the fenders’ de­
displacements and in-plane rotation w.r.t. its center of gravity, ucg , vcg
formations (δ) and their occasional loss of contact with the vessel. The
and θ. Its solution is pursued numerically through a Newton-Raphson
nonlinear response from the fenders’ elastomeric material and buckling-
scheme, following an incremental procedure whereby the external
column-type behavior are considered through the formulation proposed
loading is applied gradually from zero up to its nominal value. Here,
by Antolloni et al. (2017). Accordingly, for the type of fenders to be used
starting from the initial configuration of the system (i.e., zero dis­
in the case study of section 5, a Trelleborg’s (Trelleborg, 2020) SCK Cell
placements and rotation), we apply the loads in increments of 1/20 (this
type with maximum reaction rmax , the reaction rf (in kN) as a function of
is an empirical increment size and may be adjusted according to the
the strain (δf ) is given by
problem at hand). For the convergence criteria, we control the
[ ( )( )]
50 normalized residual and the maximum displacement difference between
rf = 2.6δ + 27248 0.37 − δf 1− √ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( ̅ ) rmax , two consecutive iterations. Both must be less than ε = 10− 6 .
2500 + 4δf 0.74 − δf
(11) 4. Linear dynamic model
Fig. 3 illustrates the reaction versus deformation curve as given by
Eq. (11). We remark that the friction forces between the vessel and the For the linear dynamic model, we start from the motion equation for
fenders’ panels are neglected here. This simplification tends to increase a free-floating vessel with three degrees-of-freedom (surge, sway and
the lines’ tension, since the lines turn to be the only elements responsible yaw) in the horizontal plane, which may be written as
for restriction of longitudinal movements. The reaction on fender j is (m + a(ω))ẍ + b(ω)ẋ + cx = F(t). (16)
represented by the following reaction vector, which is aligned with the
global direction y: Here, ẍ, ẋ, and x are the acceleration, velocity and displacement
{ } vectors (relative to the three degrees of freedom), respectively; m and
rf ,j = 0, rf ,j , 0 , (12) a(ω) are the mass and hydrodynamic added mass matrices, respectively
(wherein ω is the vector of the system’s oscillation frequencies); b(ω) is
The reaction on fender j induces a moment mfz,j on the ship around
the damping matrix; c is the system’s restoring forces stiffness matrix;
the vertical axis passing on its center of gravity, which is obtained from
and F(t) is the time-varying external forces vector. Each of these
matrices will be detailed in the next subsections, along with the external
forces vector. We note that, as our mooring system does not present
dissipative devices, the damping here is given only from hydrodynamic
phenomena. For this reason, both the added mass and damping matrices
may be called hydrodynamic coefficients. Both depend on the oscillation
frequency (ω) of the respective degree of freedom, and are different for
each one. The restorative forces stiffness (c), in turn, is related to the
mooring lines and fenders, which, as said before, have both geometrical
and physical nonlinearities. Finally, the external forces F(t) stem from
the environmental actions (we will not consider loads from the port’s
operation).
Solution to equation (16) is formulated in the frequency domain,
which requires its linearization around the equilibrium configuration
about which the system oscillates, so that classical methods of structural
Fig. 3. Reaction versus Deformation curve (in terms of rmax ) of the SCK- dynamics may be applied. To this aim, first we compute the equilibrium
Cell Fender.

4
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

configuration through the nonlinear static model from the previous ⎡ ⎤

cxx cxy cxθ
section, using the mean (static) values of the environmental actions.
c = ⎣ cyx cyy cyθ ⎦. (21)
Then, we verify whether the mooring’s response is (at least approxi­
cθx cθy cθθ
mately) locally linear around this configuration. In case it is, we proceed
with the dynamic solution; otherwise, we follow a simple iterative
procedure whereby we compute the system’s stiffness at the maximum 4.4. External forces
dynamic displacements configuration, and only then the dynamic so­
lution is carried out. Details will be shown in Section 5. We consider external forces stemming only from the environmental
actions, as said before. Loads due to the port’s operation are not
4.1. Mass and added mass matrices considered. In this setting, we consider only the current and wind loads.
Waves are neglected on the assumption that the harbor is sufficiently
For the degrees of freedom of surge and sway, the inertial forces protected, which is the case for our case study in Section 5. The current is
correspond to the vessel’s displacement (Δ) in the considered load sit­ taken as a stationary load given as a function of its mean speed (assumed
uation. For the yaw, it is the vessel’s moment of inertia around the constant in time). Its total force FC on the ship is
vertical axis (IZZ ), for which the ship’s mass distribution in the consid­ ρwater
(22)
2
Fc = V c CDc Ac f (αc ),
ered load situation must be taken into account. Accordingly, the mass 2
matrix is given by
⎡ ⎤ where ρwater is the water’s specific mass, V c is the current’s mean speed,
Δ 0 0 CDc is the ship’s drag coefficient, Ac is the ship’s cross-sectional area w.r.
m = ⎣0 Δ 0 ⎦ . (17) t. the current and f(αc ) is the sine/cosine of the current’s angle of inci­
0 0 Izz
dence (αc ) w.r.t. the ship’s submerged surface.
The added mass matrix is computed here through the software The wind, in turn, is taken as an oscillatory (and random) load,
Wamit® (Wamit, 2020). The program uses the Boundary Integral derived from the wind speed.1 Accordingly, the wind speed is divided
Equation Method (BIEM), also known as the panel method. Regarding into a constant component, called mean wind, and another with oscil­
the horizontal plane (surge, sway and yaw – indices 1, 2 and 6 below), latory behavior, named turbulence or wind’s fluctuation. The latter is a
matrix a(ω) may be written as random, stationary, ergodic process with zero mean value, which can be
⎡ ⎤ conveniently described through the wind’s speed power spectral density
a11 (ω) 0 0 SVw (ω). Here, we use the Ochi-Shin’s spectrum, as this spectrum con­
a(ω) = ⎣ 0 a22 (ω) a26 (ω) ⎦ . (18) centrates most of its energy in a frequency range that contains the
0 a62 (ω) a66 (ω)
typical natural frequencies of berthed vessels, in contrast to other
spectra (see, e.g., Feikema and Wichers (1991)). As presented by Fei­
4.2. Damping matrix kema and Wichers (1991), the Ochi-Shin’s spectrum is defined by the
following expressions:
The damping matrix is also obtained through the software Wamit® 2 /
(Wamit, 2020). The software considers the water as an ideal fluid and SVw (ω) = CV w Fg ω,
employs the potential flow theory. To incorporate the contribution of 0 ≤ ω ≤ 0, 001885V w →Fg = 583x,
the water’s viscosity on the damping, for the analyzed degrees of
420x0,7
freedom the current practice indicates that it is sufficient to assume that 0, 001885V w ≤ ω ≤ 0, 0628V w →Fg = [ ]11,5 ,
the viscous damping is of the order of 5% of the potential damping. 1 + x0,35
Accordingly, the damping matrix b(ω) may be written as 838x (23)
0, 0628V w ≤ ω→Fg = [ ]11,5 ,
⎡ ⎤ 1 + x0,35
b11 (ω) 0 0
⎣
b(ω) = 1, 05⋅ 0 b22 (ω) b26 (ω) ⎦, (19) 1, 592ω
0 b62 (ω) b66 (ω) x= ,
Vw

where indices 1, 2 and 6 designate the surge, sway and yaw DOFs, C = (750 + 69V w )10− 6 ,
respectively.
where V w is the mean wind’s speed and ω the oscillation’s frequency in
rad/s. Fig. 5 at subsection 5.2 shows the Ochi-Shin’s spectrum for our
4.3. Linearized stiffness matrix case study. The time series of the wind’s fluctuation component v(t) can
be written through a Fourier’s Transform, as shown in Eq. (24). Through
We assume that the dynamic problem corresponds to small oscilla­ standard arguments, the latter equation can be simplified and expressed
tions around a static equilibrium configuration. This latter, in turn, is in terms of the wind’s speed power spectral density, as indicated in Eq.
given by the displacements induced by the dynamic loading’s mean (25) (wherein ϕ ∈ [0, 2π] is a random phase angle and ω and Δω are the
(static) components. Therefore, the system’s stiffness may be computed oscillation’s frequency and its discretization interval, respectively). The
by applying small disturbances to the static equilibrium configuration wind’s total velocity Vw (t), then, is the sum of the mean and the fluc­
obtained with the nonlinear model of the previous section. Accordingly, tuation components, as given by Eq. (26):
we impose small displacements (δ) on each degree of freedom i at this
∫∞
configuration, and obtain the stiffness coefficients (cij ) by computing the 1
v(t) = V(ω)eiωt dω, (24)
ratio of the forces’ variation in the direction of interest (δFi ) to the 2π
imposed displacement (δj ), as follows: − ∞

δFi
cij = . (20)
δj
1
We remark that defining the wind load up to a fine scale or flow resolution
Regarding the degrees of the freedom of the horizontal plane (surge, around the ship, such as through a full aerodynamic analysis wherein local
sway and yaw – indices x, y and θ), the linearized stiffness matrix c is effects like vortices or even VIV and galloping/flutter aeroelastic phenomena
written as: may be captured, is not in the scope of the present work.

5
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

∑
N √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
v(t) ≅ 2 SVw (ω)Δωcos(ωt + ϕk ) , (25)
k=1 The wind’s action along the degrees of freedom of surge, sway and
yaw are the longitudinal and transversal oscillatory forces and moment
Vw (t) = V w + v(t). (26) w.r.t. the vertical axis, respectively. Analogously as done for the wind’s
The wind’s total force Fw (t) on the ship is total force, they can be written as a function of the respective force
component’s power spectral density along each degree of freedom,
ρair
Fw (t) = (V w + v(t))2 CDw Aw f (αw ), (27) SFwx (ω), SFwy (ω) and SMw (ω), which may be written as
2
( )2
2f wx
ρair 2 ρair V w
2 SFwx (ω) = SVw (ω), (33)
Fw (t) = V w CDw Aw f (αw ) + 2 v(t) CDw Aw f (αw ) Vw
2 2 Vw
2 ( )2
ρair V w 2f wy
+ 2
v(t)2 CDw Aw f (αw ), SFwy (ω) = SVw (ω), (34)
2 Vw Vw
( )2
where ρair is the air’s specific mass, Aw is the ship’s cross-sectional area 2mw
SMw (ω) = SVw (ω). (35)
w.r.t. the wind, CDw is the ship’s aerodynamic drag coefficient and f(αw ) Vw
is the sine/cosine of the wind’s angle of incidence (αw ) w.r.t. the ship’s
surface. The three terms on the right-hand side of Eq. (27)2 can be Here, f wx , f wy and mw are the mean wind’s forces and moment along
rewritten as follows each DOF, which may be straightforwardly computed as established by
the ROM 2.0-11 (2012)’s procedure, as mentioned in Sections 2 and 3
ρair
(28) (therein, these forces were named R|L , R|T and R|M , respectively,
2
fw = V w CDw Aw f (αw ),
2
following the Spanish Standard terminology).
2
ρair V w fw
fw (t) = 2 v(t)CDw Aw f (αw ) = 2 v(t), (29) 4.5. Analysis in the frequency domain
2 Vw Vw

2 The fluctuating displacements along the DOFs of interest may be

ρair V w fw
(30) computed from the wind forces’ power spectra defined above. Under the
′
fw (t) = 2
v(t)2 CDw Aw f (αw ) = 2 2
v(t)2 = f w,2 + fw,2 (t).
2 Vw Vw premises of forces being a random, stationary, ergodic and Gaussian
It follows that the wind’s total force may also be divided into a process and the hypothesis that the mooring system is locally linear,
constant component, named mean force and composed of the sum of Eq. Wirsching et al. (2006) propose the so-called direct method. It provides
(28) and the first term on the right side of Eq. (30) (f w + f w,2 ), and an an exact and direct stationary solution, but may not be very practical in
problems with too many degrees of freedom due to its high computa­
oscillatory component, named fluctuating force and composed by the
tional demand. Here, however, since our model considers only the three
sum of Eq. (29) and the second term on the right side of Eq. (30) (fw (t) +
DOFs of the horizontal plane, such approach is suitable. In the case of
fw,2 (t)). The terms f w,2 and fw,2 (t) indicated in Eq. (30) are, respectively,
random forces acting on multi-degrees-of-freedom systems, the
mean and fluctuating quadratic force components that are neglected in displacement along each coordinate is influenced not only by the force
this work due to their smaller magnitude in comparison to the compo­ applied along the respective coordinate but also by all other forces
nents presented in Eqs. (28) and (29). Further details on these quadratic applied to the system. According to Wirsching et al. (2006), the
terms through the development of Eq. (30) can be found in Feikema and displacement spectral density matrix Su (ω) is determined from the force
Wichers (1991). spectral density matrix SF (ω) through the following expression:
The oscillatory component can be expressed as a function of the
force’s power spectral density SFw (ω), as Su (ω) = H(ω)SF (ω)H* (ω)T , (36)
∑ where H(ω) is called the frequency response function matrix or transfer
N √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
fw (t) = 2SFw (ωk )Δωcos(ωk t + φk ), (31)
k=1 function matrix and H* (ω)T is its conjugate transpose matrix. For a
generic multi-degree-of-freedom system, it is given by (see Wirsching
where φ ∈ [0, 2π] is a random phase angle and ω and Δω are the oscil­ et al. (2006))
lation’s frequency and its discretization interval, respectively. The ( )− 1
force’s power spectral density can be expressed in terms of the wind H(ω) = − ω2 M + iωB + C , (37)
speed’s power spectral density by replacing Eqs. (25) and (31) into Eq.
(29), leading to: with M, B and C being the system’s mass, damping and stiffness
matrices. The diagonal elements of Su (ω) are the displacement’s power

∑N √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
f ∑ N √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
2 SFw (ωk )Δωcos(ωk t + φk ) = 2 w 2SVw (ωk )Δωcos(ωk t + ϕk ),
V w k=1
(32)
k=1
( )2
f
SFw (ω) = 2 w SVw (ω).
Vw

6
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

spectral density functions of each analyzed degree of freedom and are,

vcg,max = vcg + gy σy , (45)
for this reason, real-valued. The off-diagonal terms, in turn, are the
displacements’ cross-spectral density functions, being, hence, complex.
θmax = θ + gθ σθ , (46)
Applying the above concepts to our 3-DOF model, the frequency
response function matrix H(ω) may be written as:
where ucg , vcg and θ are the mean displacements (i.e., those due to the
( )− 1
H(ω) = − ω2 [m + a(ω)] + iωb(ω) + c . (38) mean wind and current loads) and gx , gy and gθ their peak factors. These
latter, according to Davenport (1964), can be estimated by (omitting the
The wind’s fluctuating component is defined through the force index for conciseness):
spectral density matrix SFw (ω), which is given by:
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ 0.57722
⎡
SFwx (ω) 0 0
⎤ g= 2ln(υT) + √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ , (47)
2ln(υT)
⎣
SFw (ω) = 0 SFwy (ω) 0 ⎦, (39)
0 0 SMw (ω)
where T is the sample period and υ is the zero-up-crossing frequency in
Hz, that is, the frequency with which the random process crosses the
with SFwx (ω), SFwy (ω) and SMw (ω) as shown in Eq. (33) through (35).
abscissa axis (i.e., assumes a zero value) with a positive slope. For the
As described in the previous section, the wind’s fluctuation is
sample period, we take T = 10 minutes, since this is the period over
expressed as a sum of different frequencies’ components (ωk ). Never­
which the World Meteorological Organization WMO (2010) defines the
theless, the ship’s added mass and damping depend on the oscillation’s
average wind speed. For the zero-up-crossing frequency, in turn, we
frequency. Therefore, each frequency ωk is analyzed separately. The
consider the following expression, which is an estimate based on the
procedure consists in computing the respective added mass a(ωk ) and
coordinate displacement’s power spectral density S(ω) in rad/s:
damping matrix b(ωk ) for each frequency ωk . The stiffness matrix c is
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅
obtained for the vessel’s static nonlinear equilibrium position, consid­ ∫∞
ω2 S(ω) dω
1
ering mean wind and current loadings. Following that, it is possible to υ= 0∫
∞ (48)
2π S(ω) dω
obtain the respective frequency response function matrix H(ωk ). Then, 0

with the respective force spectral density matrix SFw (ωk ), the displace­ Once the maximum displacements are obtained, the maximum lines’
ment spectral density matrix Su (ωk ) is computed at the frequency ωk . Eq. tension and fenders’ compression may be computed through the
(40) through Eq. (42) summarize the procedure: nonlinear static model described in section 3 (the displacements are
( )− 1 imposed and the forces follow from the iterative solution). One should
H(ωk ) = − ωk 2 [m + a(ωk ) ] + iωk b(ωk ) + c , (40)
note that, as equations 44–46 are derived in the frequency domain, the
⎡ ⎤ phases between the displacements are lost. On the safe side, we combine
SFwx (ωk ) 0 0
the maximum displacements in order to obtain the maximum mooring
SFw (ωk ) = ⎣ 0 SFwy (ωk ) 0 ⎦, (41)
forces.
0 0 SMw (ωk )

Sxx (ωk )
⎡
Sxy (ωk ) Sxθ (ωk )
⎤ 5. Case study
* T
Su (ωk ) = H(ωk )SFw (ωk )H (ωk ) = ⎣ Syx (ωk ) Syy (ωk ) Syθ (ωk ) ⎦. (42)
Sθx (ωk ) Sθy (ωk ) Sθθ (ωk ) The methodologies presented in sections 2, 3 and 4 were imple­
mented in a Mathematica’s routine and verified in simple examples with
The Su (ωk ) diagonal terms correspond to the surge Sxx (ωk ), sway known solution, as well as against existing results from the literature (we
Syy (ωk ) and yaw Sθθ (ωk ) spectral density functions. Through the above used the works from Barros (2018) and Barros and Mazzilli (2018) for
methodology, the frequency range of interest may be analyzed, and as a this purpose). In this section, they are applied to a case study of a Liq­
result one has the displacement’s spectra graphs for each DOF. In this uefied Natural Gas (LNG) terminal located in the Port of Barcarena,
work, we consider a range of 0.01–10 rad/s, as it corresponds to the Brazil. Our goal is to ascertain the validity of the simplified method from
frequencies within which the Ochi-Shin’s wind spectrum is most rele­ the Spanish standard and, in particular, the adopted value for the dy­
vant (in terms of energy), and discretize it in intervals of Δω = 0.01 rad/ namic amplification coefficient. This later, as said previously, is taken
s. here as γ D = 1.5, following from the authors’ design practice experience.
It is known that, for a random, stationary, ergodic and Gaussian
process, the mean-square value is equal to the power spectral density’s
integral. It is also known that, if the mean value of the random process is 5.1. Case description
equal to zero, the mean-square value is equal to the variance. As a result,
the displacement’s standard deviation may be estimated from the The case consists of a Liquefied Natural Gas Carrier (LNGC) vessel
discrete form of the displacement power spectral density for surge, sway permanently berthed at dolphins and responsible for storing and re-
and yaw (σ x , σ y and σ θ , respectively), according to: gasifying the LNG from incoming loaded ships. The Floating Storage
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ and Regasification Unit (FSRU) has representative characteristics taken
√ N
√∑
√ N
√∑
√ N
√∑ from ROM 2.0-11 (2012), considering a membrane-type vessel with a
√ √ SYY (ωk )Δω ; σ θ = √
σx = SXX (ωk )Δω ; σ y = Sθθ (ωk )Δω . capacity of 140,000 m3, as shown in Table 1. The terminal’s structure is
i=1 i=1 i=1
composed of four breasting dolphins, six mooring dolphins located 40 m
(43)
Finally, we resort to Davenport (1964)’s suggested expressions to Table 1
estimate the maximum instantaneous value of a random and stationary FSRU’s representative characteristics.
function that occurs in samples of certain duration, such as the Maximum Displacement 105.000 t Beam (B) 42.6 m
displacement due to the wind’s fluctuating component. Accordingly, the (ΔMAX)
maximum surge, sway and yaw displacements (ucg,max , vcg,max and θmax ) Lenght Overall (LOA) 279 m Depth (Ds) 26.5 m
Lenght Between 266 m Maximum Draft 12.3 m
can obtained as: Perpendiculars (LBP) (Dmax)
KG (a) (Loaded) 16.3 m
ucg,max = ucg + gx σx , (44)
a
Vertical distance between the vessel’s center of gravity to the bottom of its
keel.

7
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

Fig. 4. Case study’s layout.

Fig. 5. Ochi-Shin’s spectrum for V w = 27.2 m/s (left) and wind’s transversal force spectrum (right) for αW = αC = 90◦ .

Table 2 Table 4
Terminal’s structure’s dimensions. Fenders’ coordinates.
Dolphins’ level +8,50 m Seabed level − 15,00 m Fender D1 D2 D3 D4
Bollards’ level +8,90 m Highest Astronomical Tide +3,70 m
Xf (m) − 55,0 − 35,0 35,0 55,0
Fenders’ level +5,80 m Lowest Astronomical Tide − 0,30 m
Yf (m) 0,0 0,0 0,0 0,0
Zf (m) 5,8 5,8 5,8 5,8

backwards (towards land) and two mooring dolphins aligned to the

main platform. The coordinate axis’ origin O is located on the structure’s described by Eq. (11). The lines’ and fenders’ coordinates are shown in
symmetry axis, which is aligned with the vessels’ center of gravity. The Table 3 and Table 4.
layout is illustrated in Fig. 4 and its geometrical characteristics can be The environmental loads are estimated according to the OCIMF
found in Table 2. (2008)’s recommendations, with the incidence angles about the ship’s
The mooring lines’ arrangement was adopted following the recom­ longitudinal axis as shown at Fig. 4. Accordingly, for the current, we
mendations of ROM 2.0-11 (2012) for a simplified symmetrical config­ consider a constant (stationary) speed of 1.54 m/s (3 knots) for inci­
uration with vertical angles lesser than 25◦ and the minimum number of dence angles (αC) of 0◦ or 180◦ ; 1.03 m/s (2 knots) for incidence angles
cables, with two stern/bow lines, two breast lines, and two springs. of 10◦ or 170◦ ; and 0.5 m/s (0.75 knots) at a 90◦ angle. This load is
Although (see, e.g., Gaythwaite (2004)) large ships usually use 16 to 18 considered acting simultaneously with the wind load. For this latter, in
lines, we adopted the minimum number of cables as possible, even if this turn, in the static models we follow OCIMF (2008)’s recommendation of
results in larger lines’ diameters, with the purpose of investigating the a 30-s gust speed of 31 m/s (60 knots) at any incidence angle (αW) (we
above-described models in a simpler setting. The mooring lines consider angles of 0◦ up to 360◦ , at increments of 10◦ ). However, in the
considered here are made of synthetic fiber of HMPE (High-Modulus dynamic model, the wind’s speed power spectral density considers the
Polyethylene) with a MBL of 7078 kN. As this is a material with high wind’s mean speed, which, according to the WMO (2010), is defined
elasticity modulus, a more flexible material (synthetic fiber of polyolefin over a period of 10 min. Hence, as to maintain consistency between the
and polyester) is considered for the tails, with a MBL of 9284 kN. For the analyses, the 30-s wind gust speed of 31 m/s is converted into a mean
nonlinear static and the linear dynamic models, a pre-tensioning of 10% wind speed of 10 min through the formulation recommended by Dert
of the lines’ MBL is enforced in a preliminary analysis (before applica­ Norske Veritas (DNV, 2010). This results in a mean speed of 27.2 m/s for
tion of the environmental loads). The fenders used are the Trelleborg’s the dynamic model. The same incidence angles (αW) as in the static
(Trelleborg, 2020) SCK Cell Fender type SCK 2250 – E 2.1, with models are considered. We recall that waves are neglected, since the port
maximum reaction of 3514.7 kN and elastomer’s height of 2.25m, is fully protected. The study is carried out considering forces acting to­
whose stiffness has a nonlinear behavior with varying compression as wards the sea (incidence angles from 0◦ to 180◦ ) or the land (incidence

Table 3
Lines’ coordinates at the bollard (index B) and the chock (index C).
Line Type XB (m) YB (m) ZB (m) XC (m) YC (m) ZC (m) α (◦ ) β (◦ )

L1 Stern − 195,0 − 37,5 8,9 − 139,5 23,8 17,9 47,8 6,2

L2 Aft Breast − 140,0 − 37,5 8,9 − 139,5 23,8 17,9 89,5 8,4
L3 Aft Spring − 35,0 − 5,0 8,9 − 85,0 2,5 17,9 − 8,5 10,1
L4 Fwd Spring 35,0 − 5,0 8,9 85,0 2,5 17,9 8,5 10,1
L5 Fwd Breast 140,0 − 37,5 8,9 139,5 23,8 17,9 − 89,5 8,4
L6 Bow 195,0 − 37,5 8,9 139,5 23,8 17,9 − 47,8 6,2

8
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

Table 5 (33)–(35), are shown in Table 8. The Ochi-Shin’s wind’s speed power
Static models – Environmental forces (V w = 31 m/s) – γD = 1.5 spectrum density for V w = 27.2 m/s, along with the resulting wind’s
Case αW αC f sx f sy ms transversal force power spectrum density for the case with αW = αC =
[◦ ] [◦ ] [kN] [kN] [kNm] 90◦ , are shown in Fig. 5 (the power spectra for the longitudinal force and
moment w.r.t. the vertical axis, as well as for the case with αW = 240◦
Maximum lines’ tension 90 90 0 9227 − 243382
Maximum fenders’ 240 190 − 1296 − 5432 386807 and αC = 270◦ , are not shown for conciseness).
compression (linear model) For the maximum lines’ tension (αW = αC = 90◦ ) and fenders’
Maximum fenders’ 240 270 544 − 8407 348913 compression (αW = 240◦ and αC = 270◦ ) cases, the surge, sway and yaw
compression (nonlinear power spectral densities that follow from eq. (42) are show in Fig. 6 and
model)
Fig. 7, respectively. Finally, using equations (44)–(46), the maximum
displacements under the excitation of the wind’s fluctuation are ob­
angles from 180◦ to 360◦ ), at increments of 30◦ (one analysis for each tained. Table 9 shows the results for both cases (mean u, standard de­
direction). The static (mean) forces, for both current and wind, are viation σ , peak factor g, dynamic displacement ud = gσ and maximum
calculated according to equations (22) and (28) and the ROM 2.0-11 displacement umax ). By introducing these displacements into the
(2012)’s specifications, respectively. The oscillatory forces (wind’s nonlinear static model, the mooring lines’ tension and fenders’
fluctuation), in turn, as described in Section 4.4. In the next subsection, compression follow. The results are depicted in Tables 9 and 10. The
we present the results obtained for the two most critical incidence an­ resulting dynamic amplification coefficients for the case are also shown.
gles, i.e., the ones that imply the maximum line’s tension (found to be at It is noteworthy that, for the case with αW = αC = 90◦ , as the external
αW = αC = 90◦ ) and maximum fender’s compression (found to be at αW forces move the vessel towards the sea (away from the dolphins), the
= 240 and αC = 190◦ , or at αW = 240 and αC = 270◦ , depending on the fenders end up entirely uncompressed, even for the displacements at the
model). mean equilibrium.
When observing the lines’ tension in Table 10, it is seen that, for the
case with αW = 240◦ and αC = 270◦ , the aft breast line (L2) slackens at
5.2. Results and discussion the maximum dynamic displacements. This fact is a strong indication of
a system’s stiffness change. To verify whether the system’s stiffness is
In regard to the static models, after carrying out the analyses for the linear around the equilibrium configuration, for each case we compute
several incidence angles of the environmental forces mentioned above, the “peak” stiffness matrix (cp,90 and cp,240 ), i.e., the stiffness at the
we found that the critical situation for maximum lines’ tension occurs maximum displacements position, and compare it with the “mean”
for αW = αC = 90◦ in both linear and nonlinear models. The maximum (cm,90 and cm,240 ) matrix, i.e., the one computed at the mean equilibrium
fenders’ compression, in turn, occurs for αW = 240 and αC = 190◦ in the displacement. The results are shown in Eq. (49) and Eq. (50). The Δc
linear model and αW = 240 and αC = 270◦ in the nonlinear model. matrix indicates the percentage difference between the two:
Table 5 and Table 6 below show the results from both models in these
critical situations, with the values of the environmental forces on Table 5
and the maximum mooring lines’ tension, fenders’ compression and the

⎡ ⎤ ⎡ ⎤ ⎡ ⎤
9430 − 10 − 135903 9445 − 19 − 131457 +0.2 ⋅ ⋅
cm,90 = ⎣ − 10 9585 − 1840 ⎦, cp,90 = ⎣ − 19 9627 − 3278 ⎦, Δc[%] = ⎣ ⋅ +0.4 ⋅ ⎦. (49)
− 135902 − 1638 1.9128Δ108 − 131477 − 2863 1.9205Δ108 ⋅ ⋅ +0.4

vessel’s center of gravity (cg) equilibrium displacements (these latter

only for the nonlinear model) on Table 6. We recall that the linear model
follows the ROM 2.0-11 (2012)’s procedure and the dynamic amplifi­
cation coefficient adopted is γD = 1.5. Table 7
For the linear dynamic model, first we compute the equilibrium’s Environmental forces (wind plus current, with V w = 27.2 m/s) and equilibrium’s
displacements for mean wind and current loads through the nonlinear displacements obtained through the nonlinear static model with no dynamic
static model, considering the wind’s mean speed of 27.2 m/s and no amplification (γD = 1.0).
dynamic amplification (γD = 1.0), as presented in Table 7. We then αW αC f sx f sy ms ucg [m] vcg [m] θ [10− 3

compute the linearized stiffness matrix through eqs. (20) and (21) (in [◦ ] [◦ ] [kN] [kN] [kNm] rad]
the end, as shown soon below, we will check whether the system’s 90 90 0.0 5315 − 124914 − 0.010 0.279 − 0.660
stiffness is really locally linear). The wind’s mean forces (i.e., without 240 270 279 − 4894 179077 0.040 − 0.098 0.697
the current), needed for computation of the wind’s force spectra in Eqs.

Table 6
Static models – Maximum mooring lines’ tension, fenders’ compression and the vessel’s cg’s displacements (these latter only for the nonlinear model) – γD = 1.5
Case αW [◦ ] αC [◦ ] Analysis Element Max Force [kN] ucg [m] vcg [m] θ [10− 3
rad]

Lines’ 90 90 Linear L1/L2 3178 – – –

Tension 90 90 Nonlinear L2 3513 − 0.018 0.686 − 1.278
Fenders’ 240 190 Linear D1 3226 – – –
Compression 240 270 Nonlinear D1 3145 0.073 − 0.238 2.272

9
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

Table 8
Wind’s mean forces (V w =27.2 m/s) – γD = 1.0
αW [◦ ] αC [◦ ] f wx [kN] f wy [kN] mw [kNm]

90 90 0 2798 − 124914
240 270 − 279 − 2377 179077

Fig. 6. Surge (left), sway (center) and yaw (left) power spectral densities for αW = αC = 90◦ .

Fig. 7. Surge (left), sway (center) and yaw (left) power spectral densities for αW = 240◦ and αC = 270◦ .

Table 9
Dynamic model – Vessel’s cg’s displacements for the cases of maximum lines’ tension and fenders’ compression.
Vessel’s cg Displacement αW = αC = 90◦ αW = 240◦ and αC = 270◦
u σ g ud umax u σ g ud umax

ucg [m] − 0.010 ±0.014 2.80 ±0.039 0.028 0.040 ±0.026 2.79 ±0.073 0.113
vcg [m] 0.279 ±0.121 2.60 ±0.315 0.594 − 0.098 ±0.020 2.88 ±0.057 − 0.155
3
θ [10− rad] − 0.660 ±0.266 2.87 ±0.763 − 1.423 0.697 ±0.270 2.93 ±0.791 1.488

Table 10
Dynamic model – Mooring lines’ tension.
Displacement αW = αC = 90◦ αW = 240◦ and αC = 270◦
L1 L2 L3 L4 L5 L6 L1 L2 L3 L4 L5 L6

Mean (u) 1429 1912 975 744 1314 1096 391 73 428 884 703 633
Max. (umax ) 2339 3282 1114 961 1991 1444 192 0 42 1211 875 608
γD 1.64 1.72 1.14 1.29 1.51 1.32 0.49 0 0.10 1.37 1.24 0.96

Fig. 8. Surge (left), sway (center) and yaw (left) power spectral densities for αW = 240◦ and αC = 270◦ at the first (i=1) and last (i=9) iterations.
10
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

Table 11 corresponding ones from Tables 9–11, i.e., before the procedure,
Dynamic model – Fenders’ compression. evidencing that the change in the system’s stiffness must always be
Displacement αW = αC = 90◦ αW = 240◦ and αC = 270◦ checked, and measures taken if needed.
In general, in respect to the maximum lines’ tension, we observe that
D1 D2 D3 D4 D1 D2 D3 D4
the three models predict a maximum tension in the same line (L2), as
Mean (u) 0 0 0 0 1981 1871 1443 1309 shown in Tables 6 and 10. For the static models, this tension is seen to be
Max. (umax ) 0 0 0 0 2634 2466 1708 1439
γD 0 0 0 0 1.33 1.32 1.18 1.10
roughly 10% greater in the nonlinear model than in the linear one (3513
kN against 3178 kN). This difference is due to the Spanish standard’s
simplifications, which, besides the linearity hypothesis, do not consider
the lines’ different stiffness caused by their different lengths. Such
Table 12
simplifications cause uneven forces distribution between the lines that
Dynamic model – Vessel’s cg’s displacements after the iterative procedure
(converged results) for αW = 240◦ and αC = 270◦ .
may be rather significant. Regarding the dynamic model, Table 10
shows that the lines’ tension has a maximum dynamic amplification of
Vessel’s cg Displacement u g ud umax
σ
γ D = 1.72. This is greater than that adopted in the static models. How­
ucg [m] 0.040 ±0.081 2.69 ±0.218 0.258 ever, the corresponding maximum lines’ tension is 3282 kN, which is
vcg [m] − 0.098 ±0.036 2.77 ±0.100 − 0.198 only 3% greater than in the linear and 7% less than in the nonlinear
θ [10-3 rad] 0.697 ±0.523 2.78 ±1.454 2.150 model. From the design point of view, this discrepancy is very small. At
any rate, we recall that, as the dynamic analysis was performed in the
frequency domain, the phases between the displacements are lost, and
hence it was assumed that the maximum displacements occur all at the
⎡ ⎤
9412 − 12 − 134784 same time and in the most unfavorable combination possible. This tends
cm,240 ⎣
= 12 44104 162586 ⎦, cp,240 to overestimate the dynamic amplification. Overall, it can be observed
− 141274 163683 2.6488Δ108 that, for the case of maximum mooring lines’ tension, the dynamic
⎡ ⎤
9416 − 49 − 129760 amplification coefficient of γD = 1.5 proved to be quite adequate for this
⎣
= − 5 36228 771204 ⎦, Δc[%] case study. We note that this is smaller than the ROM 2.0-11 (2012)’s
− 137576 767486 1.9013Δ108 general recommendation (γD = 2.0). It is important to remark, however,
⎡ ⎤
− 0, 1 ⋅ ⋅ that another mooring arrangement could lead to maximum line tensions
= ⎣ ⋅ − 17, 9 ⋅ ⎦. (50) at different lines in each model and, consequently, to more discrepant
⋅ ⋅ − 28, 2 results between them. This could lead eventually to a more diverging
dynamic amplification coefficient, which could only be ascertained
For the case of maximum lines’ tension (αW = αC = 90◦ ), shown in Eq.
through a dynamic methodology such as the one proposed in Section 4.
(49), it can be noted an only marginal increase at the diagonal terms,
Regarding the maximum fenders’ compression, we observe that the
which refer to the displacements of interest, due to the lines’ elongation.
results from the three models are also fairly close, as seen in Tables 6 and
The coupled terms, however, vary a bit widelier. The problem is recal­
13. The linear model presented a discrepancy of less than 3% w.r.t. the
culated using the final stiffness to verify the changes in the maximum
nonlinear model (3226 kN against 3145 kN), and the dynamic model a
displacements and mooring forces. A difference of less than 1% is
discrepancy of 8% and 5% less than the static linear and nonlinear ones,
observed. Therefore, for this case, the hypothesis that the system stiff­
respectively (2993 kN against 3226 kN and 3145 kN). Again, from the
ness is locally linear is valid. For the case of maximum fenders
design point of view, these differences are very small. In terms of dy­
compression (αW = 240◦ and αC = 270◦ ), shown in Eq. (50), it can be
namic amplification, a coefficient of γD = 1.51 was observed, which is
noted that the diagonal terms undergo significant changes in the sway
virtually identical to the one adopted in the static models (γ D = 1.5). For
and yaw DOFs, and the hypothesis of a locally linear stiffness is no
the lines’ tension in the case of maximum fenders’ compression
longer valid. This occurs due to the vessel’s approach to the dolphins,
(Table 13), it is noted that, despite a dynamic amplification of γ D = 2.02
which reduces the lines’ tension and even slackens them. To circumvent
is found, the corresponding maximum tension is much smaller than the
this issue, an iterative procedure is carried out, whereby we recalculate
one observed in the case of maximum lines’ tension (Table 10 and 1789
the stiffness matrix at the maximum dynamic displacements and
kN against 3282 kN, almost 50% smaller). Since the goal here is to es­
recompute the maximum displacements. This operation is repeated until
timate the maximum mooring forces, this coefficient can be disregarded
the difference in the sway displacements between two successive itera­
in this case. Another important observation is that the hypothesis of a
tions is less than 0.1 mm (10− 4 m) (in this situation, it was observed that
locally linear stiffness is not satisfied due to breast and spring lines’
the components’ discrepancy in terms of stiffness, tension and
slackening, and the procedure adopted to circumvent this issue (taking
compression was less than 1%). This approach is on the safe side as it is
the stiffness at the maximum dynamic displacements iteratively) leads
assumed that the system’s stiffness at the maximum displacements is
to greater displacements and thereby greater fenders’ compression,
smaller due to the lines’ slackening. The procedure converges after 9
being on the safe side. Besides, as explained before, the consideration of
iterations. Fig. 8 presents the surge, sway and yaw power spectral
the simultaneity of maximum displacements due to the loss of phases in
densities at the first (i = 1) and last (i = 9) iterations, and Table 12 and
analysis in the frequency domain is also on the safe side. Altogether, for
Table 13 show the final (i.e., converged) results of the vessel cg’s dis­
the case of maximum fender’s compression, the dynamic amplification
placements and mooring lines’ tension and fenders’ compression,
coefficient of γ D = 1.5 also proved to be quite adequate for this case
respectively. Note the difference between these results and the
study, with the advantage of being smaller than the one from the ROM

Table 13
Dynamic model – Mooring lines’ tension and fenders’ compression after the iterative procedure (converged results) for αW = 240◦ and αC = 270◦ .
Displacement Lines’ tension [kN] Fenders’ compression [kN]

L1 L2 L3 L4 L5 L6 D1 D2 D3 D4

Mean (u) 391 73 427,6 884 703 633 1981 1871 1443 1309
Max. (umax ) 187 0 0 1789 1034 446 2993 2815 1875 1502
γD 0.48 0 0 2.02 1.47 0.70 1.51 1.50 1.30 1.15

11
A.L.B. de Carvalho et al. Ocean Engineering 253 (2022) 111090

2.0-11 (2012)’s recommendation. Coast. Ocean Eng. 143 (1), 04016014 https://doi.org/10.1061/(ASCE)WW.1943-
5460.0000360.
Barros, P.W.F., 2018. Dinâmica de Navios Atracados em Bacias Portuárias. Master’s
6. Conclusions Dissertation, University of São Paulo, São Paulo.
Barros, P.W.F., Mazzilli, C.E.N., 2018. The nonlinear dynamic behaviour in an alongside
A simplified linear static methodology of a ship berthed at dolphins is berth mooring arrangement. J. Braz. Soc. Mech. Sci. Eng. 40, 355.
Bianco, L.P.O., 2015. Critérios de projeto em obras portuárias: uma comparação entre
a quick and practical way to estimate the maximum mooring lines’ normas brasileira e estrangeiras. Master’s Dissertation, Federal University of Rio de
tension and fenders’ compression on a port’s structure. A nonlinear Janeiro, Rio de Janeiro.
static model, on the other hand, indicates that the system’s geometrical Carvalho, A.V.L.B., 2021. Contribuições às Análises Estática e Dinâmica do Sistema de
Amarração de Navios Atracados em Dolphins. Master’s Dissertation, University of
and physical nonlinearities, together with the lines’ different (and São Paulo, São Paulo.
varying with the vessel’s displacements) stiffnesses, cause uneven forces Comin, C., De Souza, R.M., June 2017. Port structures – the distribution of forces on
distribution between the lines that may be rather significant. The linear infrastructure due to mooring and berthing of vessels. Ibracon. Struct. Mater.J.
626–652.
dynamic model, in turn, enables the computation of the dynamic Das, S.N., Kulkarni, S., Kudale, M.D., 2015. Design of safe mooring arrangement for large
amplification coefficient. For the case studied here, the three method­ oil tankers. Procedia Eng. 116, 528–534.
ologies proved to predict mooring forces that are fairly close to each Davenport, A.G., 1964. Note on the distribution of the largest value of a random function
with application to gust loading. Proc. Inst. Civ. Eng. 28, 187–196.
other, with a discrepancy of not more than 10% for the maximum lines’ De Wilde, J., Van Dijk, A., Van Den Berg, J., Dekker, J., 2009. Direct time domain
tension and 8% for the maximum fenders’ compression. This conclusion, downtime assessment for LNG operations using computer cluster. In: Proceedings of
however, was achieved through an a priori selected dynamic amplifi­ the Nineteenth (2009) International Offshore and Polar Engineering Conference,.
The International Society of Offshore and Polar Engineers (ISOPE), Osaka.
cation coefficient for the static models. The validity of such coefficient
Dert Norske Veritas (Dnv), 2010. Recommended Practice Det Norske Veritas Dnv-Rp-
can only be verified through a dynamic model, such as the one presented C205 - Environmental Conditions and Environmental Loads. Oslo.
in Section 4 of this work. In this sense, the adopted value of γ D = 1.5 was Faltinsen, O.M., 1990. Sea Loads on Ships and Offshore Structures. Cambridge University
found to be quite adequate for the case studied – with the advantage of Press, Cambridge.
Feikema, G.J., Wichers, J.E.W., 1991. The effect of wind spectra on the low frequency
being smaller than the one from ROM 2.0-11 (2012)’s recommenda­ motions of a moored tanker in survival conditions. Offshore Technology Conference,
tions. This has been observed even on the conservative assumption that Houston.
the maximum displacements in the dynamic model occur all at the same Gaythwaite, J.W., 2004. Design of Marine Facilities for the Berthing, Mooring, and
Repair of Vessels. ASCE Press, Virginia, p. 564.
time and in the most unfavorable combination possible. Besides, for the Gaythwaite, J.W., 2014. Mooring of Ships to Piers and Wharves. American Society of
cases when the dynamic displacements affect the system’s stiffness to a Civil Engineers., Reston, Virginia.
significant extent, since the iterative procedure proposed increases the Li, C., Zheng, S., Peng, A., Zhou, S., Xiao, Y., Hu, G., Wang, H., 2019. Equivalent static
wave/wind loads for the structural design of semi-submersible platform of floating
response obtained, the dynamic amplification also falls on the conser­ offshore wind turbine. In: Proceedings of the Twenty-Ninth. International Ocean and
vative side. We believe that models such as the one proposed in Section Polar Engineering Conference, Honolulu, Hawaii, USA, 2019.
4, which in turn requires the model from Section 3, may be a useful tool Molin, B., 2002. Hydrodynamique des structures offshore. Editions Technip.
Oil Companies International Marine Forum (OCIMF), Mooring Equipment Guidelines,
for engineers and analysts to ascertain the mooring system’s forces and third ed., 2008. Witherby Seamanship Int (MEG3), Livingston.
adopted dynamic amplification coefficients required in the simplified Recomendaciones para Obras Marítimas, 2012. ROM 2.0-11: Recomendaciones para el
(linear static) methodologies from the international standards. Finally, proyecto y ejecución en Obras de Atraque y Amarre. Gobierno de España, Puertos del
Estado.
we note that a more rigorous, yet elaborate model in the time domain is
Rosa-Santos, P., Taveira-Pinto, F., Veloso-Gomes, F., 2014. Experimental evaluation of
also possible. Such an approach would need to account, step by step, for the tension mooring effect on the response of moored ships. Coast. Eng. 85, 60–71.
the influence of the nonlinearities in the system’s stiffness (displace­ Ruggeri, F., Watai, R.A., Sampaio, C.M.P., Simos, A.N., 2018. On the development of a
ments, contacts and mooring forces). The results would be closer to the higher order time-domain Rankine panel method for linear and weakly non-linear
seakeeping computations. J. Braz. Soc. Mech. Sci. Eng. 40–70.
real system’s behavior – but the solution is significantly more laborious Shivaji Ganesan, T., Debabrata, S., 2015. Direct time domain analysis of floating
for daily design and analysis of port structures. This is not in the scope of structures with linear and nonlinear mooring stiffness in a 3D numerical wave tank.
the present work. Appl. Ocean Res. 51, 153–170.
Trelleborg marine and infrastructure, "Fender. Syst. Product. Brochure.," [Online], 16-
22. Available: https://www.trelleborg.com/en/marine-and-infrastructure/produc
Declaration of competing interest ts-solutions-and-services/marine/marine-fenders/rubber-fenders. [Accessed 01 05
2020].
Wamit Incorporated and Massachusetts Institute of Technology, "WAMIT® User Manual -
The authors declare that they have no known competing financial Version 7.4.," [Online]. Available: https://www.wamit.com/manual.html. [Accessed
interests or personal relationships that could have appeared to influence 30 05 2020].
the work reported in this paper. Watai, R.A., Dinoi, P., Ruggeri, F., Souto-Iglesias, A., Simos, A.N., 2015. Rankine time-
domain method with application to side-by-side gapflow modeling. Appl. Ocean Res.
50, 69–90.
Acknowledgements Watai, R.A., Ruggeri, F., Tannuri, E.A., Santos, N.F., Tavares, B.R., dos Santos, J.M.G.,
2018. An analysis methodology for the passing ship problem considering real-time
simulations and moored ship dynamics: application to the Port of Santos, in Brazil.
The authors would like to thank EGT Engenharia and Technomar,
Appl. Ocean Res. 80, 148–165.
who motivated this work and provided part of the data that was used in Wirsching, P.H., Paez, T.H., Ortiz, K., 2006. Random Vibrations: Theory and Practice.
the examples. Second and third author acknowledge financial support Dover Publications, New York.
by CNPq (Conselho Nacional de Desenvolvimento Científico e Tec­ World Meteorological Organization WMO, 2010. Guidelines for Converting between
Various Wind Averaging Periods in Tropical Cyclone Conditions. Geneva,
nológico), Brazil, under the grants 307368/2018-1 and 305945/2020-3. Switzerland.
J. Yue, W. Kang, W. Mao, P. Chen and X. Wang, "Prediction of dynamic responses of
References FSRU-LNGC side-by-side," Ocean. Eng. 195, 2020.

Antolloni, G., Carbonari, S., Gara, F., Lorenzoni, C., Mancinelli, A., 2017. Simple physical
models to simulate the behavior of buckling-type marine fenders. J. Waterw. Port,

12